Mutations in PMM2, a phosphomannomutase gene on chromosome 16p13, in carbohydrate-deficient glycoprotein type I syndrome (Jaeken syndrome).

Carbohydrate-deficient glycoprotein syndrome type 1 (CDG1 or Jaeken syndrome) is the prototype of a class of genetic multisystem disorders characterized by defective glycosylation of glycoconjugates. It is mostly a severe disorder which presents neonatally. There is a severe encephalopathy with axial hypotonia, abnormal eye movements and pronounced psychomotor retardation, as well as a peripheral neuropathy, cerebellar hypoplasia and retinitis pigmentosa. The patients show a peculiar distribution of subcutaneous fat, nipple retraction and hypogonadism. There is a 20% lethality in the first years of life due to severe infections, liver insufficiency or cardiomyopathy. CDG1 shows an autosomal recessive mode of inheritance and has been mapped to chromosome 16p. Most patients show a deficiency of phosphomannomutase (PMM)8, an enzyme necessary for the synthesis of GDP-mannose. We have cloned the PMM1 gene, which is on chromosome 22q13 (ref.9). We now report the identification of a second human PMM gene, PMM2, which is located on 16p13 and which encodes a protein with 66% identity to PMM1. We found eleven different missense mutations in PMM2 in 16 CDG1 patients from different geographical origins and with a documented phosphomannomutase deficiency. Our results give conclusive support to the biochemical finding that the phosphomannomutase deficiency is the basis for CDG1.

Effects of fructosamine-3-kinase deficiency on function and survival of mouse pancreatic islets after prolonged culture in high glucose or ribose concentrations.

Due to their high glucose permeability, insulin-secreting pancreatic beta-cells likely undergo strong intracellular protein glycation at high glucose concentrations. They may, however, be partly protected from the glucotoxic alterations of their survival and function by fructosamine-3-kinase (FN3K), a ubiquitous enzyme that initiates deglycation of intracellular proteins. To test that hypothesis, we cultured pancreatic islets from Fn3k-knockout (Fn3k(-/-)) mice and their wild-type (WT) littermates for 1-3 wk in the presence of 10 or 30 mmol/l glucose (G10 or G30, respectively) and measured protein glycation, apoptosis, preproinsulin gene expression, and Ca(2+) and insulin secretory responses to acute glucose stimulation. The more potent glycating agent d-ribose (25 mmol/l) was used as positive control for protein glycation. In WT islets, a 1-wk culture in G30 significantly increased the amount of soluble intracellular protein-bound fructose-epsilon-lysines and the glucose sensitivity of beta-cells for changes in Ca(2+) and insulin secretion, whereas it decreased the islet insulin content. After 3 wk, culture in G30 also strongly decreased beta-cell glucose responsiveness and preproinsulin mRNA levels, whereas it increased islet cell apoptosis. Although protein-bound fructose-epsilon-lysines were more abundant in Fn3k(-/-) vs. WT islets, islet cell survival and function and their glucotoxic alterations were almost identical in both types of islets, except for a lower level of apoptosis in Fn3k(-/-) islets cultured for 3 wk in G30. In comparison, d-ribose (1 wk) similarly decreased preproinsulin expression and beta-cell glucose responsiveness in both types of islets, whereas it increased apoptosis to a larger extent in Fn3k(-/-) vs. WT islets. We conclude that, despite its ability to reduce the glycation of intracellular islet proteins, FN3K is neither required for the maintenance of beta-cell survival and function under control conditions nor involved in protection against beta-cell glucotoxicity. The latter, therefore, occurs independently from the associated increase in the level of intracellular fructose-epsilon-lysines.

L: -2-Hydroxyglutaric aciduria, a disorder of metabolite repair.

The neurometabolic disorder L: -2-hydroxyglutaric aciduria is caused by mutations in a gene present on chromosome 14q22.1 and encoding L: -2-hydroxyglutarate dehydrogenase. This FAD-linked mitochondrial enzyme catalyses the irreversible conversion of L: -2-hydroxyglutarate to alpha-ketoglutarate. The formation of L: -2-hydroxyglutarate results from a side-activity of mitochondrial L: -malate dehydrogenase, the enzyme that interconverts oxaloacetate and L: -malate, but which also catalyses, very slowly, the NADH-dependent conversion of alpha-ketoglutarate to L: -2-hydroxyglutarate. L: -2-Hydroxyglutarate has no known physiological function in eukaryotes and most prokaryotes. Its accumulation is toxic to the mammalian brain, causing a leukoencephalopathy and increasing the susceptibility to develop tumours. L: -2-Hydroxyglutaric aciduria appears to be the first disease of 'metabolite repair'.

L-2-hydroxyglutaric aciduria, a defect of metabolite repair.

L-2-hydroxyglutaric aciduria is a metabolic disorder in which L-2-hydroxyglutarate accumulates as a result of a deficiency in FAD-linked L-2-hydroxyglutarate dehydrogenase, a mitochondrial enzyme converting L-2-hydroxyglutarate to alpha-ketoglutarate. The origin of the L-2-hydroxyglutarate, which accumulates in this disorder, is presently unknown. The oxidation-reduction potential of the 2-hydroxyglutarate/alpha-ketoglutarate couple is such that L-2-hydroxyglutarate could potentially be produced through the reduction of alpha-ketoglutarate by a NAD- or NADP-linked oxidoreductase. In fractions of rat liver cytosolic extracts that had been chromatographed on an anion exchanger we detected an enzyme reducing alpha-ketoglutarate in the presence of NADH. This enzyme co-purified with cytosolic L-malate dehydrogenase (cMDH) upon further chromatography on Blue Sepharose. Mitochondrial fractions also contained an NADH-linked, 'alpha-ketoglutarate reductase', which similarly co-purified with mitochondrial L-malate dehydrogenase (mMDH). Purified mMDH catalysed the reduction of alpha-ketoglutarate to L-2-hydroxyglutarate with a catalytic efficiency that was about 10(7)-fold lower than that observed with oxaloacetate. For the cytosolic enzyme, this ratio amounted to 10(8), indicating that this enzyme is more specific. Both cMDH and mMDH are highly active in tissues and alpha-ketoglutarate is much more abundant than oxaloacetate and more concentrated in mitochondria than in the cytosol. As a result of this, the weak activity of mMDH on alpha-ketoglutarate is sufficient to account for the amount of L-2-hydroxyglutarate that is excreted by patients deficient in FAD-linked L-2-hydroxyglutarate dehydrogenase. The latter enzyme appears, therefore, to be responsible for a 'metabolite repair' phenomenon and to belong to the expanding class of 'house-cleaning' enzymes.

The gene mutated in l-2-hydroxyglutaric aciduria encodes l-2-hydroxyglutarate dehydrogenase.

The biochemical defect in L-2-hydroxyglutaric aciduria is still unknown, but the mutated gene has recently been identified on chromosome 14q22. Transfection of human embryonic kidney (HEK) cells with a cDNA encoding the product of the human gene led to a>15-fold increase in L-2-hydroxyglutarate dehydrogenase activity. The overexpressed enzyme had similar biochemical characteristics (including sensitivity to FAD and association with membranes) as the rat liver enzyme. Western blot analysis indicated that it is processed through the removal of a N-terminal approximately 4 kDa fragment, in agreement with a mitochondrial localization. Transfection experiments indicated that the mutations (K81E, E176D, Delta-exon9) found in patients with L-2-hydroxyglutaric aciduria suppressed L-2-hydroxyglutarate dehydrogenase activity. Western blot analysis showed that the three mutated proteins were expressed to various degrees in HEK cells, but were abnormally processed. Taken together, these data indicate that L-2-hydroxyglutaric aciduria is due to a deficiency in L-2-hydroxyglutarate dehydrogenase.

Variability in erythrocyte fructosamine 3-kinase activity in humans correlates with polymorphisms in the FN3K gene and impacts on haemoglobin glycation at specific sites.

BACKGROUND: Part of the fructosamines that are bound to intracellular proteins are repaired by fructosamine 3-kinase (FN3K). Because subject-to-subject variations in erythrocyte FN3K activity could affect the level of glycated haemoglobin independently of differences in blood glucose level, we explored if such variability existed, if it was genetically determined by the FN3K locus on 17q25 and if the FN3K activity correlated inversely with the level of glycated haemoglobin. RESULTS: The mean erythrocyte FN3K activity did not differ between normoglycaemic subjects (n = 26) and type 1 diabetic patients (n = 31), but there was a wide interindividual variability in both groups (from about 1 to 4 mU/g haemoglobin). This variability was stable with time and associated (P < 0.0001) with two single nucleotide polymorphisms in the promoter region and exon 6 of the FN3K gene. There was no significant correlation between FN3K activity and the levels of HbA1c, total glycated haemoglobin (GHb) and haemoglobin fructoselysine residues, either in the normoglycaemic or diabetic group. However, detailed analysis of the glycation level at various sites in haemoglobin indicated that the glycation level of Lys-B-144 was about twice as high in normoglycaemic subjects with the lowest FN3K activities as compared to those with the highest FN3K activities. CONCLUSION: Interindividual variability of FN3K activity is substantial and impacts on the glycation level at specific sites of haemoglobin, but does not detectably affect the level of HbA1c or GHb. As FN3K opposes one of the chemical effects of hyperglycaemia, it would be of interest to test whether hypoactivity of this enzyme favours the development of diabetic complications.

Study of the regulatory properties of glucokinase by site-directed mutagenesis: conversion of glucokinase to an enzyme with high affinity for glucose.

To identify the amino acids involved in the specific regulatory properties of glucokinase, and particularly its low affinity for glucose, mutants of the human islet enzyme have been prepared, in which glucokinase-specific residues have been replaced. Two mutations increased the affinity for glucose by twofold (K296M) and sixfold (Y214A), the latter also decreasing the Hill coefficient from 1.75 to 1.2 with minimal change in the affinity for ATP. Combining these two mutations with N166R resulted in a 50-fold decrease in the half-saturating substrate concentration (S0.5) value, which became then comparable to the Km of hexokinase II. The location of N166, Y214, and K296 in the three-dimensional structure of glucokinase suggests that these mutations act by favoring closure of the catalytic cleft. As a rule, mutations changed the affinity for glucose and for the competitive inhibitor mannoheptulose (MH) in parallel, whereas they barely affected the affinity for N-acetylglucosamine (NAG). These and other results suggest that NAG and MH bind to the same site but to different conformations of glucokinase. A small reduction in the affinity for the regulatory protein was observed with mutations of residues on the smaller domain and in the hinge region, confirming the bipartite nature of the binding site for the regulatory protein. The K296M mutant was found to have a threefold decreased affinity for palmitoyl CoA; this effect was additive to that previously observed for the E279Q mutant, indicating that the binding site for long-chain acyl CoAs is located on the upper face of the larger domain.

The putative glucose 6-phosphate translocase gene is mutated in essentially all cases of glycogen storage disease type I non-a.

The purpose of this work was to test the hypothesis that mutations in the putative glucose 6-phosphate translocase gene would account for most of the cases of GSD I that are not explained by mutations in the phosphohydrolase gene, ie that are not type Ia. Twenty-three additional families diagnosed as having GSD I non-a (GSDIb, Ic or Id) have now been analysed. The 9exons of the gene were amplified by PCR and mutations searched both by SSCP and heteroduplex analysis. Except for one family in which only one mutation was found, all patients had two allelic mutations in the gene encoding the putative glucose 6-phosphate translocase. Sixteen of the mutations are new and they are all predicted to lead to non-functional proteins. All investigated patients had some degree of neutropenia or neutrophil dysfunction and the clinical phenotype of the four new patients who had been diagnosed as GSD Ic and the one diagnosed as GSD Id was no different from the GSD Ib patients. Since these patients, and the four type Ic patients from two families previously studied, shared several mutations with GSD Ib patients, we conclude that their basic defect is in the putative glucose 6-phosphate translocase and that they should be reclassified as GSD Ib. Isolated defects in microsomal Pi transporter or in microsomal glucose transporter must be very rare or have phenotypes that are not recognised as GSD I, so that in practice there are only two subtypes of GSD I (GSD Ia and GSD Ib).

Structure of the gene mutated in glycogen storage disease type Ib.

We report the structure of the human gene encoding the putative glucose 6-phosphate translocase that is mutated in glycogen storage disease type Ib. Northern blots showed that the encoded 2.4 kb mRNA is mainly expressed in liver and in kidney, but is also present, although in barely detectable amounts, in leucocytes. The gene contains nine exons, one of which (exon 7) is not present in human liver or leucocyte RNA. RT-PCR analysis of mouse RNA indicates that exon 7, which is 63 bp long compared with 66 bp in man, is not expressed in liver and kidney but well in heart and brain. 5'-RACE and RNase protection assays performed on RNAs from human liver, kidney and leucocytes indicated the presence of two main regions of transcription start at approximately -200 and -100 bp with respect to the initiator ATG.

Human L-3-phosphoserine phosphatase: sequence, expression and evidence for a phosphoenzyme intermediate.

We report the sequence of the cDNA encoding human L-3-phosphoserine phosphatase. The encoded polypeptide contains 225 residues and shows 30% sequence identity with the Escherichia coli enzyme. The human protein was expressed in a bacterial expression system and purified. Similar to known L-3-phosphoserine phosphatases, it catalyzed the Mg2(+)-dependent hydrolysis of L-phosphoserine and an exchange reaction between L-serine and L-phosphoserine. In addition we found that the enzyme was phosphorylated upon incubation with L-[32P]phosphoserine, which indicates that the reaction mechanism proceeds via the formation of a phosphoryl-enzyme intermediate. The sensitivity of the phosphoryl-enzyme to alkali and to hydroxylamine suggests that an aspartyl- or a glutamyl-phosphate was formed. The nucleotide sequence of the cDNA described in this article has been deposited in the EMBL data base under accession number Y10275.

Sequence of a putative glucose 6-phosphate translocase, mutated in glycogen storage disease type Ib.

We report the sequence of a human cDNA that encodes a 46 kDa transmembrane protein homologous to bacterial transporters for phosphate esters. This protein presents at its carboxy terminus the consensus motif for retention in the endoplasmic reticulum. Northern blots of rat tissues indicate that the corresponding mRNA is mostly expressed in liver and kidney. In two patients with glycogen storage disease type Ib, mutations were observed that either replaced a conserved Gly to Cys or introduced a premature stop codon. The encoded protein is therefore most likely the glucose 6-phosphate translocase that is functionally associated with glucose-6-phosphatase.

Identification of the cDNA encoding human 6-phosphogluconolactonase, the enzyme catalyzing the second step of the pentose phosphate pathway(1).

We report the sequence of a human cDNA encoding a protein homologous to devB (a bacterial gene often found in proximity to the gene encoding glucose-6-phosphate dehydrogenase in bacterial genomes) and to the C-terminal part of human hexose-6-phosphate dehydrogenase. The protein was expressed in Escherichia coli, purified and shown to be 6-phosphogluconolactonase, the enzyme catalyzing the second step of the pentose phosphate pathway. Sequence analysis indicates that bacterial devB proteins, the C-terminal part of hexose-6-phosphate dehydrogenase and yeast Sol1-4 proteins are most likely also 6-phosphogluconolactonases and that these proteins are related to glucosamine-6-phosphate isomerases.

Glucokinase regulatory protein is essential for the proper subcellular localisation of liver glucokinase.

Glucokinase (GK), a key enzyme in the glucose homeostatic responses of the liver, changes its intracellular localisation depending on the metabolic status of the cell. Rat liver GK and Xenopus laevis GK, fused to the green fluorescent protein (GFP), concentrated in the nucleus of cultured rat hepatocytes at low glucose and translocated to the cytoplasm at high glucose. Three mutant forms of Xenopus GK with reduced affinity for GK regulatory protein (GKRP) did not concentrate in the hepatocyte nuclei, even at low glucose. In COS-1 and HeLa cells, a blue fluorescent protein (BFP)-tagged version of rat liver GK was only able to accumulate in the nucleus when it was co-expressed with GKRP-GFP. At low glucose, both proteins concentrated in the nuclear compartment and at high glucose, BFP-GK translocated to the cytosol while GKRP-GFP remained in the nucleus. These findings indicate that the presence of and binding to GKRP are necessary and sufficient for the proper intracellular localisation of GK and directly involve GKRP in the control of the GK subcellular distribution.

New evidences for a regulation of deoxycytidine kinase activity by reversible phosphorylation.

Recent studies indicate that deoxycytidine kinase (dCK), which activates various nucleoside analogues used in antileukemic therapy, can be regulated by post-translational modification, most probably through reversible phosphorylation. To further unravel its regulation, dCK was overexpressed in HEK-293 cells as a His-tag fusion protein. Western blot analysis showed that purified overexpressed dCK appears as doublet protein bands. The slower band disappeared after treatment with protein phosphatase lambda (PP lambda) in parallel with a decrease of dCK activity, providing additional arguments in favor of both phosphorylated and unphosphorylated forms of dCK.

Pathway and regulation of erythritol formation in Leuconostoc oenos.

It was recently observed that Leuconostoc oenos GM, a wine lactic acid bacterium, produced erythritol anaerobically from glucose but not from fructose or ribose and that this production was almost absent in the presence of O2. In this study, the pathway of formation of erythritol from glucose in L. oenos was shown to involve the isomerization of glucose 6-phosphate to fructose 6-phosphate by a phosphoglucose isomerase, the cleavage of fructose 6-phosphate by a phosphoketolase, the reduction of erythrose 4-phosphate by an erythritol 4-phosphate dehydrogenase and, finally, the hydrolysis of erythritol 4-phosphate to erythritol by a phosphatase. Fructose 6-phosphate phosphoketolase was copurified with xylulose 5-phosphate phosphoketolase, and the activity of the latter was competitively inhibited by fructose 6-phosphate, with a Ki of 26 mM, corresponding to the Km of fructose 6-phosphate phosphoketolase (22 mM). These results suggest that the two phosphoketolase activities are borne by a single enzyme. Extracts of L. oenos were also found to contain NAD(P)H oxidase, which must be largely responsible for the reoxidation of NADPH and NADH in cells incubated in the presence of O2. In cells incubated with glucose, the concentrations of glucose 6-phosphate and of fructose 6-phosphate were higher in the absence of O2 than in its presence, explaining the stimulation by anaerobiosis of erythritol production. The increase in the hexose 6-phosphate concentration is presumably the result of a functional inhibition of glucose 6-phosphate dehydrogenase because of a reduction in the availability of NADP.

1,3-Propanediol:NAD+ oxidoreductases of Lactobacillus brevis and Lactobacillus buchneri.

In the cofermentation of glycerol with a sugar by Lactobacillus brevis and Lactobacillus buchneri, a 1,3-propanediol:NAD+ oxidoreductase provides an additional method of NADH disposal. The enzyme has been purified from both L. brevis B22 and L. buchneri B190 and found to have properties very similar to those reported for the enzyme from Klebsiella pneumoniae. The enzymes required Mn2+ and are probably octamers with a molecular mass of 350 kDa. Although not absolutely specific for 1,3-propanediol when tested as dehydrogenases, the enzymes have less than 10% activity with glycerol, ethanol, and 1,2-propanediol. These properties contrast sharply with those of a protein isolated from another Lactobacillus species (L. reuteri) that ferments glycerol with glucose and previously designated a 1,3-propanediol dehydrogenase.

Investigation on the mechanism by which fructose, hexitols and other compounds regulate the translocation of glucokinase in rat hepatocytes.

In isolated hepatocytes in suspension, the effect of sorbitol but not that of fructose to increase the concentration of fructose 1-phosphate and to stimulate glucokinase was abolished by 2-hydroxymethyl-4-(4-N,N-dimethylamino-1-piperazino)-pyrimidine (SDI 158), an inhibitor of sorbitol dehydrogenase. In hepatocytes in primary culture, fructose was metabolized at approximately one-quarter of the rate of sorbitol, and was therefore much less potent than the polyol in increasing the concentration of fructose 1-phosphate and the translocation of glucokinase. In cultures, sorbitol, commercial mannitol, fructose, D-glyceraldehyde or high concentrations of glucose caused fructose 1-phosphate formation and glucokinase translocation in parallel. Commercial mannitol was contaminated by approx. 1% sorbitol, which accounted for its effects. The effects of sorbitol, fructose and elevated concentrations of glucose were partly inhibited by ethanol, glycerol and glucosamine. Mannoheptulose increased translocation without affecting fructose 1-phosphate concentration. Kinetic studies performed with recombinant human beta-cell glucokinase indicated that this sugar, in contrast with N-acetylglucosamine, binds to glucokinase competitively with the regulatory protein. All these observations indicate that translocation is promoted by agents that favour the dissociation of the glucokinase-regulatory-protein complex either by binding to the regulatory protein (fructose I-phosphate) or to glucokinase (glucose, mannoheptulose). They support the hypothesis that the regulatory protein of glucokinase acts as an anchor for this enzyme that slows down its release from digitonin-permeabilized cells.

How many forms of glycogen storage disease type I?

UNLABELLED: Glucose-6-phosphatase is a multicomponent enzymatic system of the endoplasmic reticulum, which catalyses the terminal steps of gluconeogenesis and glycogenolysis by converting glucose-6-phosphate to glucose and inorganic phosphate. Glycogen storage diseases type I (GSD I) are a group of metabolic disorders arising from a defect in a component of this enzymatic system, i.e. the glucose-6-phosphate hydrolase (GSD Ia), the glucose-6-phosphate translocase (GSD Ib) and possibly also the translocases for inorganic phosphate (GSD Ic) or glucose (GSD Id). The genes encoding the glucose-6-phosphate hydrolase and the glucose-6-phosphate translocase have both been cloned and assigned to human chromosomes 17q21 and 11q23, respectively. Investigation of patients with GSD I shows that those with GSD Ia are mutated in the glucose-6-phosphate hydrolase gene, whereas those diagnosed as GSD Ib, GSD Ic or GSD Id are mutated in the glucose-6-phosphate translocase gene, and are therefore GSD Ib patients, in agreement with the fact that they all have neutropenia or neutrophil dysfunction. This suggests that the biochemical assays used to differentiate GSD Ic and GSD Id from GSD Ib are not reliable. CONCLUSION: In practice therefore appears to be only two types of GSD I (Ia and Ib), which can be differentiated by (1) measurement of glucose-6-phosphatase activity in fresh and detergent-treated homogenates and (2) by mutation search in the genes encoding the glucose-6-phosphate hydrolase and the glucose-6-phosphate translocase.

Cloning and expression of a Xenopus liver cDNA encoding a fructose-phosphate-insensitive regulatory protein of glucokinase.

Xenopus liver contains a protein inhibitor of glucokinase that, in contrast to the mammalian regulatory protein of glucokinase, is insensitive to fructose 6-phosphate and fructose 1-phosphate [Vandercammen A. & Van Schaftingen, E. (1993) Biochem. J. 294, 551-556]. The purpose of this work was to compare the primary structure and other properties of this Xenopus protein with those of its rat liver counterpart. A Xenopus laevis liver cDNA library was screened using the cDNA encoding the rat liver regulatory protein as a probe. The cloned cDNA was 2534 bp long and encoded a 619-amino-acid protein with a molecular mass of 68695 Da and 57% identity with the rat liver regulatory protein. This identity was only about 30% in an internal region (amino acids 349-381) and in the 70 carboxy terminal-residues. The Xenopus cDNA was expressed in Escherichia coli and the recombinant regulatory protein was purified to near homogeneity and found to have the same size, reactivity to antibodies and effects on the kinetics of glucokinase as the protein purified from Xenopus liver. In contrast to the rat liver regulatory protein, both recombinant and native Xenopus regulatory proteins were insensitive to fructose 6-phosphate, fructose 1-phosphate and to physiological concentrations of Pi, and they inhibited Xenopus glucokinase with greater affinity than rat glucokinase. These results allow one to conclude that the fructose-phosphate-insensitive protein of lower vertebrates is homologous to the fructose-6-phosphate-sensitive and fructose-1-phosphate-sensitive protein found in mammals.

Application of C Nuclear Magnetic Resonance To Elucidate the Unexpected Biosynthesis of Erythritol by Leuconostoc oenos.

Natural-abundance C nuclear magnetic resonance (C-NMR) revealed the production of erythritol and glycerol by nongrowing cells of Leuconostoc oenos metabolizing glucose. The ratio of erythritol to glycerol was strongly influenced by the aeration conditions of the medium. The elucidation of the metabolic pathway responsible for erythritol production was achieved by C-NMR and H-NMR spectroscopy using specifically C-labelled d-glucose. The H-NMR spectrum of the cell supernatant resulting from the metabolism of [2-C]glucose showed that only 75% of the glucose supplied was metabolized heterofermentatively and that the remaining 25% was channelled to the production of erythritol. The synthesis of this polyol resulted from the reduction of the C-4 moiety of the intermediate fructose 6-phosphate. Oxygen has an inhibitory effect on the production of erythritol by L. oenos. Preaeration of a suspension of nongrowing cells of L. oenos resulted in 30% less erythritol and in 70% more glycerol formed during the anaerobic metabolism of glucose. The anaerobic production of erythritol from glucose was also found in growing cultures of L. oenos, although to a smaller extent.