The anaerobic chytridiomycete fungus Piromyces sp. E2 produces ethanol via pyruvate:formate lyase and an alcohol dehydrogenase E

Authors

  • Brigitte Boxma,

    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
    • §

      These authors contributed equally to this work.

  • Frank Voncken,

    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
    • Present addresses: ZMBH, Im Neuenheimer Feld 282, Postfach 106249. Heidelberg, D-69120, Germany.

    • §

      These authors contributed equally to this work.

  • Sander Jannink,

    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
  • Theo Van Alen,

    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
  • Anna Akhmanova,

    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
    • Department of Cell Biology and Genetics, Erasmus University, PO Box 1738, NL-3000 DR Rotterdam, The Netherlands.

  • Susanne W. H. Van Weelden,

    1. Department of Biochemistry and Cell Biology, Faculty of Veterinary Medicine, Utrecht University, PO Box 80176, NL-3508 TD Utrecht, The Netherlands.
    Search for more papers by this author
  • Jaap J. Van Hellemond,

    1. Department of Biochemistry and Cell Biology, Faculty of Veterinary Medicine, Utrecht University, PO Box 80176, NL-3508 TD Utrecht, The Netherlands.
    Search for more papers by this author
  • Guenola Ricard,

    1. Nijmegen Centre for Molecular Life Sciences (NCMLS) and Centre for Molecular and Biomolecular Informatics, Toernooiveld 1, NL- 6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
  • Martijn Huynen,

    1. Nijmegen Centre for Molecular Life Sciences (NCMLS) and Centre for Molecular and Biomolecular Informatics, Toernooiveld 1, NL- 6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author
  • Aloysius G. M. Tielens,

    1. Department of Biochemistry and Cell Biology, Faculty of Veterinary Medicine, Utrecht University, PO Box 80176, NL-3508 TD Utrecht, The Netherlands.
    Search for more papers by this author
  • Johannes H. P. Hackstein

    Corresponding author
    1. Department of Evolutionary Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld 1, NL-6525 ED Nijmegen, The Netherlands.
    Search for more papers by this author

E-mail hack@sci.kun.nl; Tel. (+31) 24 365 2935; Fax (+31) 24 355 3450.

Summary

Anaerobic chytridiomycete fungi possess hydrogenosomes, which generate hydrogen and ATP, but also acetate and formate as end-products of a prokaryotic-type mixed-acid fermentation. Notably, the anaerobic chytrids Piromyces and Neocallimastix use pyruvate:formate lyase (PFL) for the catabolism of pyruvate, which is in marked contrast to the hydrogenosomal metabolism of the anaerobic parabasalian flagellates Trichomonas vaginalis and Tritrichomonas foetus, because these organisms decarboxylate pyruvate with the aid of pyruvate:ferredoxin oxidoreductase (PFO). Here, we show that the chytrids Piromyces sp. E2 and Neocallimastix sp. L2 also possess an alcohol dehydrogenase E (ADHE) that makes them unique among hydrogenosome-bearing anaerobes. We demonstrate that Piromyces sp. E2 routes the final steps of its carbohydrate catabolism via PFL and ADHE: in axenic culture under standard conditions and in the presence of 0.3% fructose, 35% of the carbohydrates were degraded in the cytosol to the end-products ethanol, formate, lactate and succinate, whereas 65% were degraded via the hydrogenosomes to acetate and formate. These observations require a refinement of the previously published metabolic schemes. In particular, the importance of the hydrogenase in this type of hydrogenosome has to be revisited.

Introduction

Anaerobic chytridiomycete fungi are important symbionts in the gastrointestinal tract of many herbivorous mammals (Trinci et al., 1994). They contribute substantially to the degradation of plant polymers that form a major constituent of the diets of both ruminants and hindgut fermenters. These anaerobic eukaryotic microorganisms lack mitochondria; instead, they possess ATP-generating organelles called ‘hydrogenosomes’ (Yarlett et al., 1986; Müller, 1993). Besides ATP, these fungal hydrogenosomes produce hydrogen, CO2, acetate and formate (Marvin-Sikkema et al., 1990; 1993; 1994; Akhmanova et al., 1999).

Hydrogenosomes have been regarded as alternative versions of mitochondria that evolved from a common endosymbiont as adaptations to life in anaerobic environments (Martin and Müller, 1998; Tielens et al., 2002). Hydrogenosomes occur in a number of phylogenetically distinct groups such as, for example, parabasalian flagellates, anaerobic ciliates and anaerobic fungi (Roger, 1999). There is growing evidence that hydrogenosomes evolved repeatedly (reviewed by Hackstein et al., 1999; 2001; Voncken, 2001; Voncken et al., 2002a,b; Embley et al., 2003) and, notably, that they differ with respect to their ultrastructure and physiology. A major difference between trichomonad and chytrid hydrogenosomes concerns pyruvate catabolism: whereas trichomonads use pyruvate:ferredoxin oxidoreductase (PFO) for the decarboxylation of pyruvate to acetyl-CoA (Lindmark and Müller, 1973; Müller, 1993; 1998), the major enzymatic activity involved in the pyruvate metabolism of the anaerobic chytridiomycete fungi Neocallimastix sp. L2 and Piromyces sp. E2 is exerted by pyruvate:formate lyase (PFL) (Marvin-Sikkema et al., 1993; Akhmanova et al., 1999). The presence of PFO is still elusive in anaerobic chytrids. A low enzymatic activity attributed to PFO has been measured in Neocallimastix sp. L2 and N. patriciarum (Yarlett et al., 1986; Marvin-Sikkema et al., 1993), whereas the absence of any detectable PFO activity has been reported in another species, N. frontalis (O’Fallon et al., 1991). Notably, the observation of putative PFO activity has not been substantiated by the purification of the enzyme or the identification of a PFO gene from chytrids until now. On the other hand, it has been shown that cytoplasmic and hydrogenosomal variants of pyruvate:formate lyase (PFL) in Neocallimastix sp. L2 and Piromyces sp. E2 are encoded by a multigene family (Akhmanova et al., 1999). This is rather unusual for eukaryotes, as PFL activity is characteristic of certain facultative anaerobic Enterobacteria and Firmicutes. These bacteria perform mixed-acid fermentations under anaerobic conditions just as do certain anaerobic chytridiomycete fungi (Marvin-Sikkema et al., 1990; 1992; 1993; Yarlett, 1994; Julliand et al., 1998).

As a rule, bacterial mixed-acid fermentations do not produce ethanol from pyruvate via pyruvate decarboxylase and alcohol dehydrogenase as in the alcoholic fermentation of yeast, but by the successive action of PFL and alcohol dehydrogenase E (ADHE). The latter enzyme combines aldehyde dehydrogenase (ALDH) and alcohol dehydrogenase (ADH) activities using acetyl-CoA as substrate (Kessler et al., 1991; Arnau et al., 1998; Fontaine et al., 2002). Until now, ADHE has been found exclusively in eubacteria, with the remarkable exception of certain eukaryotic ‘type I’ anaerobes such as Giardia, Spironucleus, Entamoeba and Mastigamoeba (Bruchhaus and Tannich, 1994; Sánchez, 1998; Dan and Wang, 2000; Field et al., 2000; Andersson et al., 2003). However, these eukaryotic anaerobes do not perform a bacterial-type mixed-acid fermentation, they do not exhibit PFL activity and, notably, they lack energy-generating organelles such as mitochondria or hydrogenosomes (Müller, 1998).

Here, we show that Piromyces sp. E2, in contrast to the type II anaerobes Trichomonas and Tritrichomonas and type I anaerobes such as Giardia and Entamoeba (and most probably Spironucleus and Mastigamoeba too), exhibits a bacterial-type mixed-acid fermentation. We provide evidence that Piromyces sp. E2 uses PFL in the degradation of carbohydrates, and that it possesses an ADHE, which depends on acetyl-CoA for the production of ethanol. We discuss the origin(s) of eukaryotic ADHEs and their role in the anaerobic metabolism of eukaryotes.

Results

Bacterial-type mixed-acid fermentation in Piromyces sp. E2

Analysis of the fermentation products formed during axenic culture of Piromyces sp. E2 on a defined culture medium in the presence of different concentrations of fructose (0.1–0.5% w/v) showed that hydrogen, formate, acetate, lactate, succinate and ethanol were the major end-products of the metabolism (Table 1). The molar ratios of formate to acetate plus ethanol equalled 1.0 over the whole range of fructose concentrations demonstrating that, for each molecule of acetate or ethanol formed, one molecule of formate was produced. This stoichiometry can only be explained when pyruvate is degraded exclusively via PFL and not via PFO. Degradation of pyruvate via PFO would result in the generation of one molecule of hydrogen per molecule of acetyl-CoA formed and, hence, also per molecule of its metabolic end-products, i.e. acetate and ethanol. As the amount of hydrogen formed was much lower than that of acetate and ethanol (Table 1), a significant role for PFO in the degradation of pyruvate by Piromyces can be excluded.

Table 1.  Fructose metabolism of Piromyces sp. E2.
Fructose
concentration
(%)
Excreted end-products (µmol)Fructose
degraded
to end-products
(µmol)
Fructose
consumed
(µmol)
Carbon
recovery
(%)
Dry weight
biomass
at end of
incubation (g)
LactateEthanolFormateAcetateHydrogenSuccinate
  1. Axenic cultures of Piromyces sp. E2 were grown anaerobically at 39°C in medium M2, supplemented with varying concentrations of fructose. Aliquots of the culture medium and samples of the gas phase were analysed after 48 h of culture (see Experimental procedures).

0.18.2 ± 0.2 29 ± 2185 ± 2148 ± 4 91 ± 9 3.4 ± 0.1 94 ± 2104901.05
0.2 24 ± 4 66 ± 15341 ± 16258 ± 40 97 ± 1110.8 ± 0.4179 ± 22209862.83
0.3 55 ± 6116 ± 28475 ± 16343 ± 66 96 ± 7  21 ± 2268 ± 36315854.37
0.4 88 ± 18168 ± 29542 ± 14355 ± 66105 ± 6  39 ± 1325 ± 37407804.70
0.5 75 ± 12171 ± 29560 ± 5357 ± 38102 ± 2  57 ± 1330 ± 25462716.20

Table 1 shows that growth of Piromyces sp. E2 in the presence of increasing concentrations of fructose was accompanied by changes in the fermentation pattern. Although the amounts of hydrogen formed in the incubations during growth at increasing concentrations of fructose remained more or less constant, the amount of lactate, ethanol, formate, acetate and succinate increased (Table 1). These observations suggest that increasing amounts of a fermentable carbon source cause a relative shift from hydrogenosomal carbon metabolism to the cytosolic one.

In order to confirm the fermentative origin of succinate and to exclude a contribution by oxidative Krebs cycle activity, Piromyces sp. E2 was cultured in the presence of [6-14C]-glucose. Table 2 shows that the results obtained in these experiments are in agreement with those generated earlier using unlabelled fructose (Table 1). [6-14C]-glucose was catabolized into the labelled fermentation products acetate, ethanol, lactate and succinate in approximately the same molar ratios that resulted from the fermentation of unlabelled fructose. Two minor, unidentified, labelled fermentation products were detected, which accounted for less than 3% of the total excreted end-products. The production of significant amounts of 14C-labelled CO2 during the incubation with labelled glucose could be excluded. Consequently, it can be concluded that Piromyces sp. E2 does not oxidize carbohydrates into CO2 through Krebs cycle activity (cf. Tielens et al., 1992).

Table 2.  Labelled end-products from [6-14C]- and [U-14C]-glucose degradation by Piromyces sp. E2.
End-productsSubstrate (µmol·h−1·mg−1 protein)
[6-14C]-glucose[U-14C]-glucose
  • Cells were incubated for 48 h at 39°C in anaerobic medium M2 supplemented with [6-14C]- or [U-14C]-glucose. Excreted labelled end-products are shown as the mean of three (with standard deviation) and two independent experiments. Two other minor unidentified excreted end-products were each < 3% of the total excreted labelled end-products and are not shown.

  • ND, not detectable.

  • a

    . Not detectable because of massive production of labelled formate, which elutes close to succinate.

  • b

    . Less than 0.5 µmol·h−1·mg−1 protein.

Acetate17.3 ± 0.718.1
Lactate 3.4 ± 0.5 3.5
Succinate 1.5 ± 0.2NDa
Ethanol 7.8 ± 2.2 5.3
FormateND25.2
CO2 0.3 ± 0.1NDb

Incubations in the presence of uniformly labelled [U-14C]-glucose led to the formation of large amounts of 14C-labelled formate in addition to 14C-labelled acetate, lactate and ethanol (Table 2). However, also after growth on [U-14C]-glucose, there was no indication that significant amounts of labelled CO2 had been formed during the incubation. This observation excludes any significant PFO or PDH activity during the growth of Piromyces sp. E2 because these enzymes would generate one molecule of labelled carbon dioxide per molecule of pyruvate degraded. Furthermore, the amount of formate that had been formed during the incubation approximately equalled the amount of acetate plus ethanol, reinforcing the fact that PFL and not PFO is responsible for the formation of acetyl-CoA. Also, these observations exclude any significant carbon dioxide formation via the pentosephosphate pathway (Tables 1 and 2).

Piromyces sp. E2 and Neocallimastix sp. L2 exhibit ADHE activity

Given that ethanol appears to be generated from acetyl-CoA and not from pyruvate (see above), it is likely that ADHE instead of ADH is the enzyme responsible for this reaction. Therefore, ADH and ALDH activities were assayed in cell-free extracts of Piromyces sp. E2 under anaerobic and aerobic conditions. Under both conditions, a reduction in NAD+ with ethanol or acetaldehyde, respectively, could be measured, which was linear in time. As reduction of NAD+ with acetaldehyde as a substrate did not occur in the absence of CoA-SH, it is likely that ALDH activity is responsible for the CoA-SH-dependent oxidation of acetaldehyde. Moreover, there was no evidence for the presence of an NADP+-dependent ALDH activity. ADH and ALDH exhibited specific activities of 2.1 µmol·min−1·mg−1 protein and  0.47 µmol·min−1·mg−1 protein respectively.

Cell-free extracts of Piromyces sp. E2 and Neocallimastix sp. L2 were incubated in the presence of 14C-labelled acetyl-CoA in order to confirm the presence of ADHE activity directly. In both cases, the incubation resulted in the production of 14C-labelled ethanol. Control incubations with purified ADH from yeast (Boehringer Mannheim) did not produce significant amounts of labelled ethanol. The identity of ethanol produced by Piromyces sp. E2 and Neocallimastix sp. L2 was confirmed by enzymatic analysis (see Experimental procedures). Under the conditions studied, the production of ethanol from 14C-labelled acetyl-CoA and NADH occurred with a specific activity of 0.04 µmol·min−1·mg−1 protein in Piromyces sp. E2, and 0.08 µmol·min−1·mg−1 protein in Neocallimastix sp. L2.

Molecular cloning of adhE

A partial cDNA clone with substantial sequence similarity to ADHE has been recovered from a Piromyces cDNA library by random sequencing. Using polymerase chain reaction (PCR) and the rapid amplification of cDNA ends (RACE) procedure, we succeeded in isolating a 3024 bp cDNA (E2adhE). DNA sequence analysis revealed the presence of a complete open reading frame (ORF) of 2655 bp, predicting a protein with a molecular mass of 97 kDa. The non-coding sequences upstream and downstream of the ORF are very AT rich as in other genes of anaerobic chytrids (see, for example, Brondijk et al., 1996; Akhmanova et al., 1998; 1999; Voncken et al., 2002a,b and references therein).

Southern blotting indicated the presence of only one adhE gene in the genome of Piromyces sp. E2. As the coding region of the adhE gene has only one EcoRI and one ClaI restriction site, which is located upstream of the probe, the Southern blot revealed the presence of only one labelled EcoRI and ClaI fragment. The adhE gene also contains four HindIII restriction sites: the Southern blot showed the expected fragments of 344 bp and 438 bp as well as a fragment substantially larger than 860 bp (not shown).

Northern blotting reveals that adhE is expressed not only when Piromyces sp. E2 is grown on fructose, but also under growth on lactose and cellobiose (Fig. 1). The length of this RNA (3600 nt) exceeds the length of the isolated cDNA (3024 bp). This difference results from the difficulties in cloning the extremely AT-rich non-coding regions in full length (see above).

Figure 1.

Northern blot of the E2adhE of Piromyces sp. E2. Poly (A)+ RNA of Piromyces sp. E2 grown on lactose (A), cellobiose (B) and fructose (C) was probed with the cDNA fragment of E2adhE. The length of the transcript is ≈ 3600 nt.

Subcellular localization of E2ADHE

The expression and subcellular localization of ADHE in Piromyces sp. E2 and Neocallimastix sp. L2 were investigated by Western blotting using an antiserum raised against ADHE from Escherichia coli. The anti-ADHE serum cross-reacted predominantly with a protein of ≈ 100 kDa, which was found exclusively in the cytosolic fraction of both species (Fig. 2). The apparent size of the cross-reacting protein is compatible with the predicted molecular mass of 97 kDa for ADHE. Minor cross-reacting bands had a size below 60 kDa. Potentially they light up degradation products of ADHE as none of the hydrogenosomal fractions contained any cross-reacting material.

Figure 2.

Western blot analysis and subcellular localization of ADHE of Neocallimastix sp. L2 (A) and Piromyces sp. E2 (B). Homogenates of the mycelia were subjected to cellular fractionation according to the method of Voncken et al. (2002a). CY, cytosol; HS, hydrogenosomal matrix proteins, obtained after lysis by freeze–thawing; HC, hydrogenosomal membrane proteins obtained by extraction with 0.1 M carbonate, pH 11.5; HM, integral membrane proteins after two extractions with carbonate. The heterologous ADHE antiserum cross-reacts predominantly with a cytosolic protein of ≈ 100 kDa, which is in good agreement with the expected value of 97 kDa.
C. Protein markers: molecular mass in kDa.

Phylogenetic analysis

The deduced amino acid sequence (E2ADHE) exhibits high similarity (40–51% identity) to the adhE sequences of several eubacteria, but also to the adhE genes of Giardia intestinalis (47%; U93353.1), Spironucleus barkhanus (50%; AY132367.1), Entamoeba histolytica (42%; Q24803) and Mastigamoeba balamuthi (47%; AY113188.1) (Fig. 3). Like the adhE genes of G. intestinalis, E. histolytica and M. balamuthi, the adhE of Piromyces lacks an N-terminal extension that could function as a targeting signal to the hydrogenosomes (van der Giezen et al., 1997; Akhmanova et al., 1998; 1999; Voncken et al., 2002b). Phylogenetic analysis of the entire adhE gene using mrbayes (Huelsenbeck and Ronquist, 2001) indicates that the G. intestinalis and Piromyces sp. E2 sequences are monophyletic, while these sequences tend to be paraphyletic in the neighbour-joining phylogenies. The adhE of M. balamuthi is closely related to the G. intestinalis and Piromyces sequences but is not monophyletic with the latter (Fig. 4). The adhE of E. histolytica clusters in a different branch from the other eukaryotes, comprising certain Proteobacteria (Pasteurella) and various Firmicutes.

Figure 3.

Figure 3.

Alignment of the deduced amino acid sequence E2ADHE from Piromyces sp. E2. The linker between the ALDH and ADH domains is shaded. NBS, nucleotide binding site; IBS, iron binding site; P.sp. E2, Piromyces sp. E2 (AY282576); M.bal., Mastigamoeba balamuthi (AAM51642); E.hist., Entamoeba histolytica (S53319); G.int., Giardia intestinalis (AAC47539); E.coli, Escherichia coli (AAG56096); V.chol, Vibrio cholerae (D82127); S.aur., Staphylococcus aureus (H89775); C.ace., Clostridium acetobutylicum (AF321779); L.lact., Lactococcus lactis (CAA04467).

Figure 3.

Figure 3.

Alignment of the deduced amino acid sequence E2ADHE from Piromyces sp. E2. The linker between the ALDH and ADH domains is shaded. NBS, nucleotide binding site; IBS, iron binding site; P.sp. E2, Piromyces sp. E2 (AY282576); M.bal., Mastigamoeba balamuthi (AAM51642); E.hist., Entamoeba histolytica (S53319); G.int., Giardia intestinalis (AAC47539); E.coli, Escherichia coli (AAG56096); V.chol, Vibrio cholerae (D82127); S.aur., Staphylococcus aureus (H89775); C.ace., Clostridium acetobutylicum (AF321779); L.lact., Lactococcus lactis (CAA04467).

Figure 4.

Phylogenetic analysis. Phylogeny of the ADHEs calculated with mrbayes. mrbayes and neighbor-joining bootstrap values are indicated on the branches. The evolution of ADHE appears to have involved at least one horizontal transfer event to the eukaryotes from the Firmicutes (see also the ADHE in Entamoeba, etc.; Field et al., 2000), besides horizontal transfers to the proteobacteria and one cyanobacterium. The ADHE from Spironucleus barkhanus (AY132367.1) has been omitted from the tree because it is incomplete; the available sequence is monophyletic with the one from Giardia intestinalis.
Asterisks indicate the same values as the ADHE tree was unrooted. The position of the root is based on an ADH/ADHE tree that included ADH1 from C. saccharobutylicum. The identifiers of the sequences: T. elongatus BAC07780, C. thermocellum ZP_00060060, V. cholerae NP_231667, C. acetobutylicum Q9ANR5, C. perfringens Q8XHF4, E. coli P17547, S. typhimurium AAB07 × 1597, S. enterica serovar typhi Q8Z7F1, Y. pestis Q8ZEJ0, M. balamuthi AAM51642, C. acetobutylicum ADHE_CLOAB P33744, G. intestinalis O09435, L. innocua Q8Y6Q0, P. multocida Q9CKZ7, S. agalactiae AAM98961, E. histolytica EHNADDAD_1, S. pyogenes AAM78643, L. gasseri ZP_00046353, S. pneumoniae NP_359429, L. mesenteroides ZP_00063848, L. lactis O86282, O. oeni ZP_00069413, Piromyces sp.E2 AY282576.

In all likelihood, the fusion of an aldh with an adh that gave rise to adhE occurred only once as all the adhEs form a monophyletic cluster within either adh/adhE- or aldh/adhE-based trees (not shown). Of all adhs and aldhs, the adh1 from Clostridium saccharobutylicum (P13604) is the most similar to the adhEs, and its location in the adh/adhE tree is indicated at the root in Fig. 4.

Discussion

Mixed-acid fermentation in anaerobic chytrids

We have shown here that the anaerobic chytridiomycete fungus Piromyces sp. E2 performs a bacterial-type mixed-acid fermentation during axenic growth on a defined culture medium (Tables 1 and 2). Similar to the mixed-acid fermentation in certain Enterobacteria and Firmicutes, formate, acetate, lactate, ethanol and succinate are the prevalent end-products of its carbon metabolism. Not only this type of fermentation, but also the fact that PFL (Tables 1 and 2; Akhmanova et al., 1999) and ADHE (Figs 1–5) are the key enzymes in the anaerobic carbon metabolism make Piromyces sp. E2 unique among all eukaryotes studied so far. We have also shown that the anaerobic chytrid Neocallimastix sp. L2 possesses the enzymatic activity to convert acetyl-CoA into ethanol, i.e. ADHE activity, and a protein that cross-reacts with an ADHE antiserum (Fig. 2). As Neocallimastix sp. L2 also possesses gene(s) with substantial sequence similarity to PFL (Akhmanova et al., 1999), it is likely that mixed-acid fermentation with the involvement of PFL and ADHE is a widespread property of anaerobic chytrids. The presence of significant amounts of formate among the fermentation products of quite a number of anaerobic chytrids strongly supports this assumption (e.g. Julliand et al., 1998).

Figure 5.

Energy metabolism of Piromyces sp. E2.
A. A scheme of the metabolic pathways involved in the production of the major end-products.
B. The relative fluxes (cf. Tables 1 and 3) through the pathways indicated in (A). The thickness of the arrows is proportional to the calculated fluxes in the presence of 0.3% fructose.
The numbers in (A) indicate the following enzymes: (1) hexokinase, glucose-6-phosphate isomerase, phosphofructokinase 1, aldolase and triose phosphate isomerase; (2) glyceraldehyde 3-phosphosphate dehydrogenase; (3) phosphoglycerate kinase, phosphoglycerate mutase and enolase; (4) phosphoenolpyruvate carboxykinase; (5) malate dehydrogenase; (6) fumarase; (7) fumarate reductase; (8) pyruvate kinase; (9) lactate dehydrogenase; (10) cytosolic pyruvate:formate lyase; (11) alcohol dehydrogenase E; (12) pyruvate import into hydrogenosomes; (13) malic enzyme; (14) hydrogenase; (15) hydrogenosomal pyruvate:formate lyase; (16) acetate:succinate CoA-transferase; (17) succinyl-CoA synthetase; (18) ADP/ATP carrier. AcCoA, acetyl-CoA; EtOH, ethanol; FUM, fumarate; G3P, glyceraldehyde-3-phosphate; MAL, malate; OXAC, oxaloacetate; PEP, phosphoenolpyruvate; PYR, pyruvate; SUCC, succinate.

Our results show that the metabolism of anaerobic chytrids is clearly different from that of the type II (hydrogenosome-bearing) anaerobes Trichomonas vaginalis and its relatives Tritrichomonas foetus and Monocercomonas sp. (Lindmark and Müller, 1973; Müller, 1993; 1998; Hackstein et al., 1999). T. vaginalis does possess PFO in its hydrogenosomes, but it expresses neither ADHE or PFL and does not perform a (bacterial-type) mixed-acid fermentation either. In contrast, several type I anaerobes (eukaryotes lacking hydrogenosomes or mitochondria), such as Giardia or Entamoeba, exhibit ADHE activity, which is essential for their survival under anaerobic conditions (Espinosa et al., 2001). These type I anaerobes possess PFO- or PFO-like enzymes (Horner et al., 1999), but seem to lack PFL, and they do not perform a mixed-acid fermentation (Bruchhaus and Tannich, 1994; Yang et al., 1994; Rosenthal et al., 1997; Müller, 1998; Field et al., 2000). The above-mentioned type I anaerobes might be derived from ancestral type II (or even mitochondriate organisms) by a loss of their compartmentalized carbon metabolism (Martin and Müller, 1998; Andersson et al., 2003). Therefore, a common denominator for the presence or absence of certain fermentations, PFO, PFL or ADHE, has remained elusive until now. One might speculate whether the extremely high concentration of CO2 in the rumen and the gastrointestinal tract of herbivorous animals hampers the function of PFO by product inhibition. The use of PFL might offer a possibility of dealing with high environmental CO2 concentrations. Moreover, PFL does not require any additional electron sinks, in marked contrast to PFO, which requires ferredoxin/hydrogenase as electron acceptors.

Giardia and Mastigamoeba depend on ADHE during anaerobic growth, although they possess ‘classical’ ADHs in addition. It has been speculated whether ADHE precludes the formation of free molecules of acetaldehyde, which seems to be especially toxic under anaerobic conditions (Espinosa et al., 2001). However, it remains unclear whether this hypothesis is sufficient to explain the evolution of ADHE. The adaptation of chytrids to strictly anaerobic rumen or gut environments might have favoured the acquisition of an ADHE by lateral gene transfer for other reasons. Notably, the presence of a PFL (which might either be a eukaryotic relic or also be acquired by lateral gene transfer) might depend on the presence of an ADHE. In E. coli, where PFL is only expressed under anaerobic growth conditions, ADHE can act as a PFL inactivase, thereby protecting PFL against irreversible damage by oxygen (Kessler et al., 1991; 1992; Sawers and Watson, 1998). The regulation of PFL activity by ADHE might extend beyond the above-mentioned inactivase role and provide clues for an understanding of: (i) the frequently observed co-appearance of PFL and ADHE; and (ii) the evolution of a compartmentalized carbon metabolism (see below). However, this co-appearance does not appear universal, as a number of species, such as for example the archaea Methanobacterium thermoautotrophicum and Archaeoglobus fulgidus, possess pfl but lack adhe genes (http://dove.embl-heidelberg.de/STRING).

The phylogenetic origin of ADHE

ADHE is a fusion protein of an (N-terminal) NAD- and CoA-dependent acetaldehyde dehydrogenase with a (C-terminal) NAD-dependent alcohol dehydrogenase (Kessler et al., 1991; Arnau et al., 1998; Fontaine et al., 2002). The fusion might have occurred among the Firmicutes because species of that taxon appear throughout the tree. Notably, the eukaryotic ADHEs appear at two different positions in the ADHE tree, which otherwise only comprises bacterial sequences. This indicates that there has been at least one horizontal gene transfer of ADHE from the bacteria to the eukaryotes, to Entamoeba histolytica (Field et al., 2000). The phylogenetic clustering of the other eukaryotic ADHEs suggests either an ancient eukaryotic protein with subsequent losses in aerobic eukaryotes or multiple gene transfers between the anaerobic eukaryotes (Fig. 4; cf. Nixon et al., 2002; Andersson et al., 2003 for a phylogenetic analysis of class III ADHs). Given that the hydrogenosomes of anaerobic chytrids arose from fungal mitochondria (Voncken, 2001; van der Giezen et al., 2002; 2003; Voncken et al., 2002a), the evolution of a mixed-acid fermentation together with acquisition of PFL and ADHE would favour the assumption of a concerted lateral gene transfer of both enzymes. Nevertheless, the available data also do not allow the identification of the evolutionary origins of PFL with certainty (cf. Akhmanova et al., 1999).

The role of the hydrogenosomes in the energy metabolism of Piromyces sp. E2

The observation that PFL and ADHE are the key enzymes in the degradation of carbohydrates by anaerobic chytrids requires certain refinements of the scheme for carbon catabolism in these organisms (Fig. 5A; cf. Marvin-Sikkema et al., 1994; Akhmanova et al., 1999; Hackstein et al., 1999; Voncken, 2001). On the basis of the results displayed in Tables 1 and 2, it becomes feasible to calculate the relative fluxes of glucose (and fructose) through the indicated catabolic pathways (Fig. 5B; Table 3)

Table 3.  Relative flux of carbohydrate degradation through various pathways.a
Fructose
conc.
(% w/v)
CytosolicHydrogenosomal
PathwayGlycolysisMalate
production
Succinate
production
Pyruvate
formation
Lactate
production
Ethanol and
formate
production
Pyruvate
import
Hydrogen
production
Acetate and
formate
production
Enzymes
involvedb
2, 34, 56, 78910, 111213, 1415–17
  • a

    . Calculated fluxes through the various pathways of carbohydrate degradation as shown in Fig. 5. Fluxes are expressed relative to the glycolytic flux through glyceraldehyde 3-phosphate (G3P), which was set at 100 for each growth condition. See Table 1 for the original data.

  • b

    . See Fig. 5A.

0.1 10049251 415324779
0.2 10029371 718462773
0.3 100224781021471865
0.4 100226781326391655
0.5 100248771125401555

Glycolysis leads to the formation of phosphoenolpyruvate (PEP) as a key intermediate (Fig. 5A, enzymes 1–3). In anaerobic chytrids, this PEP is converted into oxaloacetate as well as into pyruvate (enzymes 4 and 8). As ADHE (enzyme 11) is only found in the cytosol, the amount of formate produced by the cytosolic PFL (enzyme 10) equals the amount of ethanol produced. This implies that the amount of formate generated by PFL inside the hydrogenosomes (enzyme 15) equals the total amount of formate minus the amount of formate produced in the cytosol, which equals the total amount of ethanol formed. This amount of hydrogenosomal formate equals the amount of acetate produced and represents the metabolic flux through the hydrogenosomes. Because the production of hydrogenosomal NADH, and hence the formation of hydrogen (enzyme 14), depends exclusively on malate oxidation, the amount of hydrogen produced is equal to the flux through the hydrogenosomal malic enzyme (ME, enzyme 13) activity. The flux through ME plus the flux from malate to succinate is equal to the flux through phosphoenolpyruvate carboxykinase (PEPCK, enzyme 4), which converts PEP into oxaloacetate.

Besides the malate that is imported into the hydrogenosome and degraded via PFL after its conversion to pyruvate, cytosolically produced pyruvate must be an important substrate for further degradation by the hydrogenosomal PFL (enzyme 15). This flux of pyruvate into the hydrogenosomes (enzyme 12) can be calculated: it is the difference between the above calculated flux through the hydrogenosomal PFL (enzyme 15) and the flux through ME (enzyme 13), which equals the import of malate into the hydrogenosomes. This implies that the total flux through pyruvate kinase (PK, enzyme 8) can be calculated as the total of the production of ethanol and lactate, plus the above calculated import of pyruvate into the hydrogenosomes. Summation of the fluxes through PEPCK (enzyme 4) and PK (enzyme 8) results in the calculated total flux through glycolysis. In Fig. 5B, the thickness of the arrows is proportional to the flux through the various reactions, i.e. the contribution of each metabolic pathway can be envisaged directly [in this case, in the presence of 0.3% (16.7 mM) fructose].

On the basis of this metabolic scheme and the calculated fluxes (Fig. 5; Table 3), it has to be concluded that the major role of these hydrogenosomes in carbon catabolism is a compartmentalization of the final steps, and not the removal of reduction equivalents by the generation of hydrogen. Notably, the major flow through the chytrid hydrogenosome involves pyruvate, which is split by PFL without the generation of reduction equivalents. Thus, these steps do not contribute directly to hydrogen formation by its [Fe] hydrogenase (Davidson et al., 2002; Voncken et al., 2002b). The latter reaction seems to represent only a minor pathway of the anaerobic energy metabolism in this type of hydrogenosome and might indicate a role in controlling or fine tuning of the intrahydrogenosomal environment. Thus, the major role of the chytrid hydrogenosomes seems to be the generation of ATP by substrate-level phosphorylation. The presence of PFL in the absence of hydrogenosomal ADHE most probably directs all organellar pyruvate into substrate-level ATP formation. A possible presence of ADHE inside the hydrogenosomes would compromise this function of the hydrogenosome as an energy-generating organelle. In the cytoplasm, however, ADHE might allow regulation of PFL activity, thus saving pyruvate (and its metabolites) for anapleurotic pathways. This hypothesis is supported by the observation that several ‘mitochondrial’ enzymes involved in anabolic reactions, e.g. malate dehydrogenase, aconitase, isocitrate dehydrogenase and acetohydroacid reductoisomerase, have been retargeted to the cytoplasm in Piromyces sp. E2 (Akhmanova et al., 1998; Hackstein et al., 1999). Consequently, compartmentalization of the energy metabolism seems to enhance the possibilities for regulation of the metabolic pathways of this organism.

Experimental procedures

Organisms and growth conditions

Axenic cultures of Piromyces sp. E2 and Neocallimastix sp. L2 were grown anaerobically at 39°C in medium M2 (20 ml; Teunissen et al., 1991), supplemented with varying concentrations of fructose (0.1–0.5% w/v; 5.5–27.8 mM). After 48 h of culture, liquid and gas samples were taken for further analysis.

Analysis of metabolic products

Samples from the headspace and the culture medium were analysed by gas chromatography for acetate, lactate, ethanol and hydrogen according to the method of Teunissen et al. (1991). Formate, succinate and fructose concentrations were determined with the aid of test combinations purchased from Boehringer Mannheim. Incubations with 14C-labelled glucose were performed at 39°C in sealed incubation bottles with 20 ml of medium M2 containing Piromyces sp. E2. The cultures were grown for 24 h before the addition of labelled glucose. Incubations were then performed for another 48 h and contained either 10 µCi of [U-14C]-glucose or 10 µCi of [6-14C]-glucose (2.07 GBq mmol−1), both from Amersham. Incubations were terminated by chilling (0°C) and the addition of 300 µl of 6 M HCl to lower the pH from 7.2 to 2.0. Analysis of excreted end-products was performed as described earlier (Tielens et al., 1992).

Isolation of genomic DNA, total RNA and messenger RNA

Genomic DNA and total RNA was prepared as described earlier (Voncken et al., 2002a,b). Poly(A)+ RNA was isolated with the RNeasy kit (Qiagen). Adaptor-ligated cDNA was prepared according to the Clontech SMART RACE cDNA amplification kit.

Isolation of the ADHE-encoding cDNA

The Piromyces sp. E2 cDNA library in the vector λ ZAPII was constructed as described earlier (Akhmanova et al., 1998). During random screening (Akhmanova et al., 1998; 1999; Hackstein et al., 1999), a sequence with high similarity to adhE genes was found. A full-length cDNA was recovered by PCR on adapter-ligated cDNA. Sequencing was performed with the ABI Prism model 310 automatic sequencer, using a dRhodamine terminator cycle sequencing ready reaction DNA sequencing kit (Perkin-Elmer Applied Biosystems).

Southern and Northern and blotting

Southern blotting was performed according to the method of Sambrook et al. (1989). The Southern blot was kindly provided by Dr H. Harhangi. Total RNA was separated on 1.2% agarose–formaldehyde gels, 15 µg per lane. A partial adhE cDNA clone (amino acids 409–877) was used to prepare a labelled DNA probe by PCR with [α-32P]-dATP using universal M13 forward and reverse primers.

Localization of the adhE gene product

Hydrogenosomal (Hyd) and cytosolic (Cyt) fractions were prepared anaerobically from Piromyces sp. E2 homogenate by differential centrifugation, according to the protocol described by Voncken et al. (2002a). Equal amounts of protein from all fractions were separated on an SDS-polyacrylamide gel, blotted to a polyvinylidene difluoride (PVDF) membrane (Bio-Rad) and probed with the E. coli ADHE antiserum kindly provided by Drs D. Kessler and J. Knappe (University of Heidelberg).

Enzyme assays

The assay mixture to determine ADH (EC1.1.1.1) activity in the direction of ethanol oxidation contained 0.4 M ethanol and 1 mM NAD+ in a 50 mM glycine–NaOH buffer, pH 9.0. To determine ALDH (CoA-acetylating; EC1.2.1.10) activity in the direction of acetaldehyde oxidation, the assay mixture contained 2 mM acetaldehyde, 0.2 mM CoA-SH and 0.8 mM NAD+ in a 50 mM glycine–NaOH buffer, pH 9.0. The enzymatic reactions, performed at 39°C, were initiated by adding cell-free extract (CFE). Reduction of NAD+ was monitored spectrophotometrically at 340 nm in a HP 8453 UV-visible spectrophotometer, and the quantity of NADH was calculated from e340 = 6220 M−1 cm−1 (Sánchez, 1998; Dan and Wang, 2000). The enzyme assays were performed both aerobically and anaerobically (95% N2/5% H2).

To measure ADHE activity in the direction of ethanol formation, a radioactive end-point assay was performed. Cell-free extracts of Piromyces sp. E2 and Neocallimastix sp. L2 were incubated in the presence of 2 mM [1-14C]-acetyl-CoA (0.4 MBq), 50 mM Tris-HCl (pH 7.4) and 5 mM NADH. The reaction was performed at 20°C and terminated after 10 min by the addition of ice-cold trichloroacetic acid (10% final concentration, w/v). Subsequently, the assay mixtures were stored on ice. The assay mixture was centrifuged (2 min at 10000 g), and the resulting supernatant was analysed. Ethanol and acetyl-CoA were separated by anion-exchange chromatography, using a Bio-Rad AG 1-X8, 100–200 mesh column (60 × 1.1 cm), and 2.5 ml fractions were collected for liquid scintillation counting (van Oordt et al., 1989). Ethanol was identified as the labelled end-product by comparison of its retention volume on this column. The identity of ethanol was further confirmed by enzymatic analysis: after incubation of the products of the 14C-labelled ADHE assay with yeast ADH, the labelled end-product of the ADHE assay, was converted into (labelled) acetate.

Phylogenetic analysis

Sequence alignments were created with clustalx (Thompson et al., 1997) from a representative set of sequences of ADHEs (maximum level of identity 95%). Unequivocally aligned positions were selected with gblocks (Castresana, 2000). A total of 630 positions were selected in 21 blocks (representing 64% of the initially 981 positions) and used for phylogenetic analysis. Phylogenetic trees were calculated with mrbayes (Huelsenbeck and Ronquist, 2001) using four gamma-distributed rate categories plus invariant positions and the JTT model of protein sequence evolution (Jones et al., 1992). Alternatively, distance-based phylogenies were calculated using neighbour joining (Saitou and Nei, 1987) as implemented in puzzle (Strimmer and von Haeseler, 1996), again based on the JTT amino acid substitution matrix with one invariable and four variable gamma rate categories. Most of the tree partitions that had the highest frequency in the mrbayes analysis (200 000 iterations sampled every 100 iterations after saturation of the likelihood value) were consistently predicted by neighbour joining (100 bootstraps generated by seqboot; Felsenstein, 1996), the consensus tree being generated with consense (Felsenstein, 1996). The main exception is that mrbayes predicts the Giardia and Piromyces sequences to be monophyletic, whereas these sequences tend to be paraphyletic in the neighbour-joining phylogenies. The position of the root is based on the position of ADH1 from Clostridium saccharobutylicum (P13604) in an ADH/ADHE tree that was calculated in the same way as the ADHE tree.

Acknowledgements

We are very grateful to Drs D. Kessler and J. Knappe, Heidelberg, for providing the antiserum against E. coli ADHE. We are indebted to Drs Harry Harhangi and Matthe Wagenmakers, Nijmegen, for support with the Southern blotting. The suggestions made by three anonymous reviewers are gratefully acknowledged.

Ancillary