|Home | About | Journals | Submit | Contact Us | Français|
A constructed lactate dehydrogenase (LDH)-negative mutant of Enterococcus faecalis V583 grows at the same rate as the wild type but ferments glucose to ethanol, formate, and acetoin. Microarray analysis showed that LDH deficiency had profound transcriptional effects: 43 genes in the mutant were found to be upregulated, and 45 were found to be downregulated. Most of the upregulated genes encode enzymes of energy metabolism or transport. By two-dimensional (2D) gel analysis, 45 differentially expressed proteins were identified. A comparison of transcriptomic and proteomic data suggested that for several proteins the level of expression is regulated beyond the level of transcription. Pyruvate catabolic genes, including the truncated ldh gene, showed highly increased transcription in the mutant. These genes, along with a number of other differentially expressed genes, are preceded by sequences with homology to binding sites for the global redox-sensing repressor, Rex, of Staphylococcus aureus. The data indicate that the genes are transcriptionally regulated by the NADH/NAD ratio and that this ratio plays an important role in the regulatory network controlling energy metabolism in E. faecalis.
Lactic acid bacteria (LAB) are widely used for production of lactic acid in fermented food. During the fermentation process, pyruvate is converted to lactate in addition to a number of minor metabolites, such as acetic acid, acetaldehyde, ethanol, acetoin, and acetate. However, under certain conditions, these bacteria shift from homolactic to heterolactic (or mixed-acid) fermentation, with formate, acetate, acetoin, ethanol, and CO2 as end products. In Lactococcus lactis, mixed-acid fermentation has been shown to take place at low grow rates under microaerobic conditions (11), under true carbon-limited conditions, and while growing at low pH on carbon sources other than glucose (15, 20).
Mixed-acid fermentation was also seen after removing the lactate dehydrogenase (LDH) activity in Enterococcus faecalis V583 (12). This bacterium has two ldh genes, but ldh-1 is the main contributor to lactate production. A mutant with deletions in both ldh genes (the Δldh1.2 mutant) was constructed and shown to direct its carbon flow from pyruvate away from lactate toward formate, acetoin, and alcohol production (12). Alternative carbon fluxes in different knockout mutants have also been reported for Lactococcus lactis (22).
The mechanism of the shift from homolactic to mixed-acid fermentation is still not fully understood. During transformation of pyruvate to lactate, LDH regenerates NAD+ from NADH formed during glycolysis. When pyruvate is converted to acetyl-coenzyme A (acetyl-CoA) by either pyruvate formate lyase (PFL) or pyruvate dehydrogenase (PDH), reduction of acetyl-CoA to ethanol regenerates NAD+ from NADH and is an alternative to lactate formation in redox balancing. The carbon flux is biochemically regulated (4, 5). Fructose-1,6-bisphosphate is an allosteric activator of lactate production, and dihydroxyacetone phosphate and d-glyceraldehyde-3-phosphate are strong inhibitors of the pyruvate formate lyase in Lactococcus lactis (4).
However, less is known about the regulation of the synthesis of glycolytic enzymes, especially in E. faecalis. In L. lactis, enzyme levels are regulated in response to growth conditions, and correlations between metabolic and transcriptomic or proteomic data have been established (3, 5). Combining the three approaches in one study provides more information and an improved understanding of the shift in LAB from homolactic to mixed-acid metabolism. Given that lactate production is extremely important for all LAB, including the emerging pathogen E. faecalis, we compared the Δldh1.2 mutant and its wild type by metabolic, transcriptomic, and proteomic analyses. Lactate dehydrogenase deficiency affects a large number of genes, and our data provide new insight into the regulation of energy metabolism in E. faecalis.
Enterococcus faecalis V583 and a mutant lacking lactate dehydrogenase (the Δldh1.2 mutant) (12) were used throughout this study. The bacteria were grown in a chemically defined medium (CDM-LAB) containing 1.1% glucose, 0.1% sodium acetate, 0.06% citrate, 19 amino acids, and growth factors at 37°C (12, 16). For all analyses, the cells were grown anaerobically in tightly capped, filled 50-ml screw-cap tubes with a starting pH of 7.4 to an optical density at 600 nm (OD600) of 0.6. The cells were then harvested by centrifugation at 4°C for 10 min at 6,000 × g, and pellets were either flash frozen in liquid nitrogen or treated according to the protein extraction protocol (see below). Supernatants were frozen at −20°C until metabolite analyses. All experiments were run in triplicate.
After removal of bacterial cells by centrifugation (10 min, 6,000 × g), metabolites in the cultures were analyzed by high-performance liquid chromatography (HPLC) (17). Ethanol and acetoin were analyzed by headspace gas chromatography (14). Lactate and glucose were also measured by using Megazyme enzymatic kits (Wicklow, Ireland).
Flash-frozen pellets were stored at −80°C until RNA isolation. Total RNA was isolated by use of FastPrep (Bio101/Savant) and an RNeasy minikit (Qiagen) as previously described (33). The RNA concentration was determined with a NanoDrop spectrophotometer (NanoDrop Technologies), and the quality was tested by using an RNA 600 Nano LabChip kit and a Bioanalyzer 2100 instrument (Agilent Technologies). cDNA synthesis, labeling, and hybridization were performed as described previously (18). The microarray used was described by Solheim et al. (32). It contained 3,219 70-mer probes representing 3,219 open reading frames (ORFs) of the genome of E. faecalis V583. Three replicate hybridizations with mRNAs were obtained with three separate growth experiments. The Cy3 and Cy5 dyes (Amersham) used during cDNA synthesis were swapped in two of the three replicate hybridizations. Hybridized arrays were scanned with a Tecan LS scanner (Tecan). Fluorescence intensities and spot morphologies were analyzed using GenePix Pro 6.0 (Molecular Devices), and spots were excluded based on slide or morphology abnormalities.
Analysis of microarray data was done by the LIMMA package (www.bioconductor.org) in the R computing environment (www.r-project.org). Preprocessing and normalization were done according to the methods of Smyth and Speed (29). A linear mixed model (27) was used in tests for differential gene expression. A mixed-model approach was used to describe variation between arrays as previously described (33). Empirical Bayes smoothing of gene-wise variances was conducted according to the method of Smyth et al. (28).
To verify the microarray results, the following genes were selected for analysis by real-time quantitative reverse transcription-PCR (qRT-PCR): EF0900 (adhE; bifunctional acetaldehyde-CoA/alcohol dehydrogenase gene), EF1612 (pflA; pyruvate formate lyase activating enzyme gene), EF0082 (major facilitator family transporter gene), EF1964 (gap-2; glyceraldehyde-3-phosphate dehydrogenase gene), and EF0255 (ldh; l-lactate dehydrogenase gene). 23S rRNA was used to normalize the data (Table (Table1).1). Real-time quantitative PCR (qPCR) was performed using a Rotor-Gene 6000 centrifugal amplification system (Corbett Research) and a 20-μl final reaction volume containing 2.5 μl 100×-diluted cDNA, 7.5 μM (each) forward and reverse primers (Sigma), and 12.5 μl Higher Power SYBR green PCR master mix (Roche). The transformation to cDNA was performed as described above. The PCR included an initial denaturation cycle at 95°C for 10 s, followed by 40 cycles of denaturation at 95°C for 10 s, annealing for 15 s, and elongation at 72°C for 30 s. Relative gene expression was calculated by the ΔCT method, using the 23S rRNA gene as the endogenous reference gene.
Proteins from bacterial cultures were isolated by alkaline lysis at 4°C. In brief, 50 ml of bacterial culture was centrifuged at 6,000 × g at 4°C. Bacterial pellets were suspended in 0.5 ml 0.9% (wt/vol) NaCl, washed three times, and resuspended in 400 μl of rehydration buffer containing 8 M urea, 2 M thiourea, 0.5% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), 0.1% IPG buffer, 10 mM dithiothreitol (DTT), and a trace of bromophenol blue. Cells were then broken by use of FastPrep (Bio101/Savant) at 6 m/s three times for 45 s each at 4°C, with 60-s pauses between. Unbroken cells were removed by centrifugation at 8,000 × g for 10 min at 4°C. The samples were stored at −80°C until further analysis. The total protein concentration for each sample was measured using the colorimetric assay RC DC protein assay reagent (Bio-Rad), using bovine serum albumin (BSA) as a standard.
Protein separation, gel analysis, trypsin treatment, and extraction of proteins of interest were performed as described previously (1). The gels were scanned and analyzed by Delta2D software (Decodon, Greifswald, Germany) and by a pixel-based analysis of multiple images for the identification of proteome patterns of two-dimensional (2D) gel electrophoresis images (6). Extracted peptides were desalted with C18 Stage tips (24). The peptides were eluted with 1 μl 70% (vol/vol) acetonitrile (ACN), and then 0.5 μl of each sample was mixed with 0.5 μl of the matrix mixed with 15 mg/ml alpha-cyano-4-hydroxycinnamic acid and applied to a matrix-assisted laser desorption ionization (MALDI) target plate (Bruker Daltonics, Billerica, MA). Peptide mass fingerprinting (PMF) and tandem mass spectrometry (MS/MS) were performed on Ultra Flex MALDI-tandem time of flight (MALDI-TOF/TOF) (Bruker Daltonics) instruments. The mass range for MALDI-TOF/MS was 800 to 4,000 Da, with a mass accuracy of 50 ppm. Protein identification was carried out using Mascot (Matrix Science Inc., Boston, MA) software and searches under “other Firmicutes” in the NCBI database.
The microarray data obtained in this study have been deposited in the ArrayExpress database (http://www.ebi.ac.uk/arrayexpress/) under accession number E-MTAB-472.
E. faecalis V583 and its lactate dehydrogenase-negative mutant (the Δldh1.2 mutant) were grown under anaerobic conditions at 37°C to an OD600 of 0.6. As shown in Table Table2,2, lactic acid was the major metabolic end product in the wild type, while the mutant produced increased amounts of acetoin, formate, and ethanol and some pyruvate, but no lactic acid. Neither strain produced acetate as a metabolic end product.
The differences in expression profiles of the wild type and the mutant were assessed by the expression ratio between each gene in the mutant and the respective wild-type gene. The results presented in Table Table33 and in Table S1 in the supplemental material are the means for three independent biological replicates. Altogether, 88 genes were found to be expressed differentially (>2-fold); 43 were upregulated, and 45 were downregulated. Many of the genes affected were genes engaged in energy, pyrimidine, and citrate metabolism and in transport functions, but a number of genes of unknown function were also affected.
Enterococcus faecalis has four main routes of pyruvate catabolism. In addition to lactate formation, these lead to the production of acetoin, formate plus acetyl-CoA, and CO2. The acetyl-CoA formed can be reduced to ethanol to maintain redox balance. As shown in Table Table3,3, the genes for all of these pathways (EF0900 [bifunctional acetaldehyde-CoA/alcohol dehydrogenase gene], EF1213 [acetolactate synthetase gene], EF1353 and EF1354 [pyruvate dehydrogenase complex genes], EF1612 [pyruvate formate lyase activating enzyme gene], and EF1613 [formate acetyltransferase gene]) were upregulated in the mutant. The gene encoding the main lactate dehydrogenase, ldh-1 (EF0255), was truncated in the mutant, but the sequence recognized by the hybridizing probe was present and showed about 10-fold enhanced transcription (Table (Table33).
Interestingly, most of the genes involved in pyrimidine biosynthesis (EF1712 to EF1720) were significantly upregulated in the Δldh1.2 mutant (Table (Table3),3), but the transcriptional data for the EF1714, EF1715, and EF1716 genes were more doubtful due to poor P values. Also, the EF0677 gene, encoding phosphoglucomutase, which converts glucose-6-phosphate to glucose-1-phosphate, was significantly upregulated in the Δldh1.2 mutant. This enzyme is also important in the production of uracil-glucose, since glucose-1-phosphate is used as a substrate in UDP-glucose production. The EF1721 (pyrR) gene encodes a bifunctional pyrimidine regulatory protein that exerts the uracil phosphoribosyltransferase catalysis that is crucial for UDP-glucose production, and it might be upregulated (log2 value = 2.1), but a poor P value (0.27) precluded it from being included among the upregulated genes. Also, EF1720, the uracil permease gene, is probably upregulated (log2 value = 1.9), though the P value (0.34) kept it from being considered upregulated. In summary, our transcription results suggest that the Δldh1.2 mutant triggers an increased production of UDP-glucose that could be used both in cell wall biosynthesis and in polysaccharide production. However, indications of increased polysaccharide production, such as altered colony appearance or culture viscosity, were not observed.
Unlike the wild type, the mutant consumed some of the citrate present in the growth medium, causing increased acetoin production. In line with this, most of the genes for citrate metabolism (EF3315 to EF3327) were found to be upregulated, indicating that both cit operons are affected by the ldh deletion.
Table S1 in the supplemental material summarizes the genes downregulated in the mutant. A majority of the genes are hypothetical (17 of 35 chromosomal genes), and eight are located on plasmid pTEF2. Several of the downregulated genes encode cell envelope-associated proteins. The gene showing the strongest reduction of transcription encodes a major facilitator family transporter (EF0082) (log2 value = −3.7). In a gene cluster involved in the biosynthesis of aromatic amino acids (EF1561 to EF1568), four genes were found to be downregulated significantly (EF1562, EF1564, EF1565, and EF1566).
Among the three plasmids of the V583 strain, only pTEF2 carries genes that were significantly affected in transcription in the deletion mutant. Of the 62 genes annotated for pTEF2, 14 genes were transcriptionally affected. The plasmid-carried genes EFB0038 and EFB0042 to EFB0046 were among the most affected and were upregulated up to a log2 value of 6.3, while the adjacent gene clusters EFB0048 to EFB0051 and EFB0053 to EFB0056 were strongly downregulated (up to a log2 value of 2.8).
To verify the quality of the microarray results, the relative amounts of mRNAs of five genes were analyzed by qPCR. As shown in Table S2 in the supplemental material, the qPCR results were in agreement with the data obtained by the microarrays.
The proteomes of the two strains were compared by 2D gel electrophoresis. About 400 gel spots were distinguished. Differentially expressed proteins were isolated and identified by MALDI-TOF/TOF-MS analysis. Altogether, 45 differentially expressed proteins (P < 0.05) were identified (Table (Table4),4), of which 24 were upregulated and 21 were downregulated. LDH (EF0255) was absent in the mutant, while the cell division protein DivIVA (EF1002) (23) was not found on the gel of the wild type. Among other proteins identified was a bifunctional acetaldehyde-CoA/alcohol dehydrogenase (EF0900). This protein was present in equal amounts in both strains. By sorting the identified proteins according to metabolic function, we found that most of the differences in expression were among proteins engaged in energy metabolism (nine proteins), followed by seven proteins related to fatty acid metabolism, phospholipid metabolism, and amino acid biosynthesis. A single protein (EF3293) involved in purine metabolism was expressed less in the mutant.
Most of the genes encoding the differentially expressed proteins were not represented by statistically significant data in the transcriptomic data. However, the expression of four proteins correlated well with the transcriptomic data, including the pyruvate dehydrogenase complex E1 component beta subunit (EF1354), pyruvate formate lyase activating protein (EF1612), and two hypothetical proteins (EF3313 and EF1617). Discrepancies between proteomic and transcriptomic data were also seen. The transcription of the bifunctional acetaldehyde-CoA/alcohol dehydrogenase (EF0900) gene was upregulated >10-fold in the mutant, but the protein was present in equal amounts in the two strains. The ldh mutant also appeared to contain reduced levels of a plasmid-encoded single-strand-binding protein (EFB0043), though its transcription was highly upregulated compared to that in the wild type.
Moreover, the mutant contained more glyceraldehyde-3-phosphate dehydrogenase (EF1964) protein and triosephosphate isomerase (EF1962) protein than the wild type did. The transcriptomic data for the corresponding genes were of unsatisfactory quality, but the RT-PCR showed that EF1964 was not differentially expressed. Altogether, these results indicate that there are important regulations at the translational level as well.
The increased production of pyruvate and ethanol suggests an elevated NADH/NAD ratio in the mutant (30). The global gene regulator Rex is known to respond to this ratio by differential binding to Rex operators (7). We therefore examined the E. faecalis V583 genome sequence for putative Rex boxes and compared them to our transcriptomic and proteomic data. We used the consensus palindromic sequence (TGTGANNNNNNTCACA) established for Staphylococcus aureus (7) for the genome-wide search. By allowing for two mismatches, we found the sequence in 151 intergenic regions and upstream of open reading frames annotated as genes (data not shown). Putative Rex boxes were found upstream of 22 genes/operons showing differential expression in our transcriptome or proteome analyses (Table (Table5),5), among which 16 were positively regulated and 6 were negatively regulated.
The biochemical regulation of carbon flow in energy metabolism of LAB has been well investigated, but only a few studies have been carried out using the new transcriptomic and proteomic technologies. In this study, we also demonstrated regulation of central carbon metabolism at the level of biosynthesis of the proteins involved.
An E. faecalis mutant lacking ldh metabolizes sugar by pathways that are used very little, if at all, by the wild type, and this is accompanied by increased transcription of genes engaged in these pathways.
Our metabolite data show that pyruvate was converted to acetyl-CoA by PFL and further reduced to ethanol. This generated excess NADH, which had to be reoxidized for redox balance. This could have been done by acetate production from acetyl-CoA formed by either PFL or PDH. The process involving only PFL for pyruvate metabolism would produce more ATP per glucose molecule consumed than does normal lactic acid fermentation. However, excess NADH was used for acetoin production, in a process that produces the same amount of ATP as the wild type. In the mutant, PDH was upregulated, but pyruvate oxidation did not take place. Ward et al. showed that PDH can be active in E. faecalis under anaerobic conditions (34), but the activity is reduced at a high NADH/NAD ratio (31). Snoep et al. showed that E. faecalis producing ethanol has an elevated NADH/NAD ratio, and this might explain why PDH was not active in the mutant (30).
The NADH/NAD ratio also regulates the activity of the transcription factor Rex, and putative Rex boxes were found upstream of a number of the differentially expressed genes and operons in S. aureus (7, 19). In Bacillus subtilis, Rex regulates genes encoding proteins of the respiratory chain (25), and in S. aureus, Rex controls transcription involved in the transition from aerobic to anaerobic growth (19). Pagels et al. found 461 putative Rex binding sites in the S. aureus genome by using their Rex box consensus sequence and allowing for two mismatches. However, they demonstrated that Rex could bind to some, but not all, of these sites, indicating that additional sequence features are required for Rex-mediated regulation (19). Thus, it is likely that our sequence search overestimates the number of Rex boxes in the E. faecalis genome. However, our data suggest that Rex also acts as a repressor under anaerobic conditions. All of the genes involved in the four different pathways of energy metabolism of pyruvate appear to be regulated by Rex and were upregulated in the mutant. Interestingly, the genes encoding the enzymes for NAD regeneration during anaerobic growth, ldh-1 and adhE, showed the strongest upregulation and are both preceded by two Rex boxes. The ldh of S. aureus is also preceded by two Rex boxes (19). A putative Rex box was also found upstream of ldh-2 (EF0641), the second ldh gene in E. faecalis, but the biological significance of ldh-2 is apparently very low in E. faecalis V583 compared to that of ldh-1 (12).
Rex has been recognized as a repressor, and its DNA binding can be influenced strongly by NADH. NADH causes derepression of genes by binding to the Rex repressor in a complex that diminishes its ability to bind the Rex box (10, 19). This appears to happen in E. faecalis operons as well, leading to increased transcription in the mutants. However, we also found putative Rex boxes upstream of genes downregulated in the mutant. This suggests that Rex may activate transcription in the wild type. To our knowledge, a role of Rex in activation of transcription has not previously been reported.
In E. faecalis V583, the EF2638 and EF2933 genes both encode putative Rex proteins (21). For EF2933, we found enhanced transcription in the mutant (1.9-fold; P = 0.04), and a Rex box upstream of the gene suggests that the gene is autoregulated in E. faecalis V583. In Streptomyces coelicolor, the Rex gene is also preceded by a Rex binding site, and the protein has been shown to repress its own transcription (2).
Enzymes encoded by the central glycolytic operon (EF1962 and EF1964) were transcriptionally expressed at higher rates in the mutant, and this was also confirmed by the proteomic analysis.
The proteome data revealed several differentially expressed proteins that were not verified by the transcriptome analysis. In most cases, this could probably be attributed to noise/poor statistics for the microarray data or just to a low level of transcription but highly efficient translation, including high stability of the transcripts. However, no changes were found by qPCR analysis of EF1964 transcripts. The operon encompassing EF1962 to EF1965 is probably transcriptionally regulated by the cell's energy status via the CggR regulator (18). The data presented here indicate additional posttranscriptional or translational regulation. Discrepancies between transcriptomic and proteomic data were also noticed for EFB0043 and EF0900, again suggesting regulation beyond the level of transcription. Despite an unaltered protein level, the metabolic data clearly reflect increased activity of the adhE gene (EF0900) in the mutant. In a study of Lactococcus lactis, it was concluded that translational regulation had a major influence compared to transcriptional regulation of glycolytic enzymes (5).
The genes for another energy-yielding process, citrate metabolism, also appear to be regulated transcriptionally by Rex. Pyrimidine synthesis genes were also upregulated in the mutant. It has been demonstrated in L. lactis that the expression of these genes is affected by energy sources and by a disrupted regulation of arginine metabolism (9, 13).
Notably, EF0082 was the most downregulated gene in this study, and a similar result has been found for other mutants, including bacteriocin-resistant mutants (8, 18). The gene encodes a major facilitator family transporter, and its transcription has been suggested to be regulated by Ers (8) and the carbon catabolite protein through an upstream catabolite-responsive element (cre) (18). A number of the other differentially expressed genes in our mutant appear to be under catabolite control (18). The major glucose phosphotransferase system (PTS), the mannose-PTS, also appears to be dually regulated. In addition to the sigma54 promoter preceding EF0019, a Rex box found in front of EF0020 and elevated levels of the EF0020 protein suggest that the PTS is regulated by the NADH/NAD ratio. Moreover, LDH appears to be regulated by Rex but is also catabolically activated through cre regulation mediated by CcpA (18). These and many of the other proteins described here appear to be regulated by a network involving global regulators and energy and redox sensing aimed at maintaining homeostasis. Central in this regulatory network are the global regulators Rex and CcpA. Their interdependence is illustrated by the presence of a Rex box upstream of ccpA (data not shown), indicating that CcpA transcription is also sensitive to NAD/NADH.
The present study evokes the complexity of the central energy metabolism of LAB and suggests revised and complex regulations for how these bacteria cope with their changing access to energy sources. Many new aspects and questions related to the regulation of central energy metabolism have been raised, and a substantial amount of work is needed to scrutinize and confirm the various regulatory pathways that govern these pathways.
This work was supported by the SysMO-LAB project, which is financed by the Research Council of Norway.
We thank Morten Skaugen, Kari R. Olsen, and Linda H. Godager for technical assistance.
Published ahead of print on 4 February 2011.
†Supplemental material for this article may be found at http://aem.asm.org/.