(Received for publication, August 21, 1996, and in revised form, November 19, 1996)
From the Institut für Biochemie der Pflanzen, Heinrich-Heine-Universität Düsseldorf, Universitätsstraße 1, D-40225 Düsseldorf, Federal Republic of Germany
The archaeal ATPase of the halophile
Haloferax volcanii synthesizes ATP at the expense of a
proton gradient, as shown by sensitivity to the uncoupler carboxyl
cyanide p-trifluoromethoxyphenylhydrazone, to the ionophore
nigericin, and to the proton channel-modifying reagent
N,N-dicyclohexylcarbodiimide. The conditions
for an optimally active ATP synthase have been determined. We were able
to purify the enzyme complex and to identify the larger subunits with
antisera raised against synthetic peptides. To identify additional
subunits of this enzyme complex, we cloned and sequenced a gene cluster encoding five hydrophilic subunits of the A1 part of the
proton-translocating archaeal ATP synthase. Initiation, termination,
and ribosome-binding sequences as well as the result of a single
transcript suggest that the ATPase genes are organized in an operon.
The calculated molecular masses of the deduced gene products are 22.0 kDa (subunit D), 38.7 kDa (subunit C), 11.6 kDa (subunit E), 52.0 kDa
(subunit B), and 64.5 kDa (subunit A). The described operon contains
genes in the order D, C, E, B, and A; it contains no gene for the
hydrophobic, so-called proteolipid (subunit c, the proton-conducting
subunit of the A0 part). This subunit has been isolated and
purified; its molecular mass as deduced by SDS-polyacrylamide gel
electrophoresis is 9.7 kDa.
The archaeon (archaebacterium) Haloferax volcanii belongs to the more moderate halophilic archaea, growing best in a medium containing 2 mol/liter NaCl and 0.25 mol/liter MgCl2 at pH 7 (1). The proton-conducting ATPase of this archaeon is a halophilic enzyme that shows maximum activity at 1.75 mol/liter NaCl. It was characterized recently with respect to its hydrolytic function (2). The membrane-bound ATPases from archaea share properties with both bacterial (formerly eubacterial) and eucaryal (formerly eukaryotic) ATPases. Like F0F1-ATPases, they are able to synthesize ATP from ADP and phosphate coupled to a proton flow. Genetically derived amino acid sequences of the large subunits of archaeal ATPases (2-6), immunological cross-reactivity (7, 8), and inhibition of the enzymes by specific inhibitors (1, 9-11) deliver the main arguments for the conclusion that the archaeal ATPases resemble eucaryal vacuolar ATPases (V-ATPases)1 more closely than bacterial F-ATPases.
The primary structures of subunits A and B of H. volcanii
are closely related to those of Halobacterium salinarium
(halobium) (6) and indeed more related to the class of
the eucaryal V-ATPases. On the other hand, the function of eucaryal
V-ATPases is to generate a H+ at the
expense of ATP hydrolysis; they are not able to synthesize ATP in a
reversed reaction. More detailed investigations on the intermediate
position of different archaeal ATPases are required.
Up to now, almost nothing was known about the small subunits of
halophilic ATPases. F-ATPases have one copy each of three different
small subunits, ,
, and
, which are involved in proton conduction and/or regulation (for review, see Refs. 12 and 13). V-ATPases, as far as we know, have three to four different small subunits, C, D, E, and F. We know almost nothing about the function of
these subunits (for review, see Ref. 14). This paper reports on the
proton-dependent ATP synthesis of the H. volcanii ATPase, the purification and immunodetection of subunits,
and the nucleotide sequence of the ATPase operon encoding the
A1 part of the enzyme that contains three small subunits as
in F-ATPases.
H. volcanii WR 340 was grown in a medium containing (w/v each) 16.2% NaCl, 3% MgSO4, 1% KCl, 0.08% CaCl2, 40 mmol/liter Tris-Cl, pH 7.2, 0.43% peptone, and 0.26% yeast extract according to Ref. 15.
Preparation of VesiclesH. volcanii cells were harvested by centrifugation at 10 °C in a Beckman centrifuge (Model J2-21) at 5000 × g for 30 min. The pellets were pooled, resuspended, and washed with a medium containing 1.75 mol/liter NaCl and 50 mmol/liter Tris-HCl, pH 7. After centrifugation, the cell concentration was adjusted to a value corresponding to 60 A520, followed by sonication on ice for 3 × 30 s (Branson sonifier with microtip). By this method, up to 80% inside-out vesicles were made as shown by the membrane-bound menadione-dependent NADH dehydrogenase reaction. After centrifugation (4 °C, 8000 × g, 30 min), the supernatants were collected and transferred to a Beckman ultracentrifuge (Model L8-70M) to remove the remaining unruptured cells, followed by ultracentrifugation for 60 min at 140,000 × g and 4 °C. The resulting pellet, containing the vesicles, was collected in a small volume of 1.75 mol/liter NaCl and 50 mmol/liter Tris-HCl, pH 7, and again centrifuged at 140,000 × g for 5 min. Aliquots of this suspension were used for measurements of ATP synthesis.
ATP SynthesisAcid/base-dependent ATP synthesis (by alkalization of the outer medium) was measured by the addition of vesicles prepared at pH 7 (see above) in a medium comprising 20 mmol/liter MgCl2, 5 mmol/liter Pi, 5 mmol/liter ADP, 1.75 mmol/liter NaCl, and 50 mmol/liter Tris-Cl, pH 9. The final volume was 700 µl, and the temperature was adjusted to 40 °C, if not indicated otherwise. The reaction was started by the addition of 100 µl of prepared vesicles (see above) and stopped after the indicated times by adding 100-µl portions to 100 µl of perchloric acid (final concentration of 0.5 mol/liter). Immediate neutralization of the samples was followed by centrifugation and determination of the ATP concentration by the luciferin/luciferase reaction (Pharmacia Luminometer). ATP was determined in a medium containing 0.1 mol/liter Tris/acetic acid, pH 7.75, 2 mmol/liter EDTA, 5 mmol/liter MgSO4, 10 mmol/liter K2SO4, and 50 µl of luciferase reagent (ATP bioluminescence kit, CLS II, Boehringer, Mannheim, Germany). The reaction was started by the addition of 50 µl of neutralized supernatant (see above) to 500 µl of reaction medium.
Isolation of ProteolipidThe isolation procedure was
performed according to Denda et al. (16). Membranes of
H. volcanii (~400 mg of protein) were suspended in 60 ml
of 50 mM Tricine, pH 7.5, followed by ultracentrifugation (140,000 × g, 55 min), and were washed twice. The
precipitated membranes were resuspended in 60 ml of diethyl ether and
precipitated by centrifugation (7000 × g, 10 min). We
repeated the washing procedure twice. After resuspension of the pellet
in 60 ml of chloroform/methanol (2:1, plus 5 mmol/liter
dithiothreitol), the suspension was stirred overnight at 4 °C. The
addition of 20 ml of water and centrifugation at 7000 × g for 15 min resulted in two phases. The lower organic phase
was concentrated to a final volume of 2 ml and mixed with 8 ml of
ice-cold (20 °C) diethyl ether. We collected the resulting pellet
by centrifugation and washed it with ice-cold acetone. This final
pellet was dried under vacuum, followed by resuspension with ~100
µl of chloroform/methanol (2:1), and stored at
20 °C. Protein
determinations were performed according to Bensadoun and Weinstein
(17).
Genomic DNA from H. volcanii WR 340 was prepared as described previously (2). RNA preparation was carried out according to Chomczynski and Sacchi (18).
Northern and Southern hybridization experiments were performed
according to standard protocols (19). -32P-Labeled gene
probes were obtained by using the random-primed labeling kit from
Boehringer. Restriction fragments were subcloned in pBluescript
SK
and pUC19, respectively, using Escherichia
coli XL-1 Blue as host. Sequencing was carried out by employing
the dideoxy chain termination method (20) using the T7 sequencing kit
(Pharmacia, Freiburg, Germany). Regions forming secondary structures
during sequencing reactions or during electrophoresis were additionally sequenced with the TaQence kit (U. S. Biochemical Corp.) using Taq polymerase, both with and without 7-deaza-dGTP.
Furthermore, the T7 sequencing kit was used together with 7-deaza-dGTP
and 7-deaza-dATP.
The protein distribution of crude membranes, isolated protein, and isolated subunit c of the ATPase fractions was studied on SDS-polyacrylamide gels (15%) (21) after silver staining (22). The molecular mass of the proteolipid was estimated from the positions of standard proteins. Polyclonal antisera against synthetic peptides of subunits A and B of the Haloferax ATPase were raised in rabbits and used as probes in Western blots to confirm that the purified protein represents the A1A0-ATPase from H. volcanii. The oligopeptides used were DEKWEFEPTVSEGDE (amino acids 123-137 of subunit A) and EDDYEDDAESVEAER (the final 15 amino acids of subunit B) (2). Western blot analysis was carried out as described earlier (2).
Characterization of the ATP Synthase in Membrane Vesicles
Archaea are able to convert a transmembrane potential into ATP using an ATP synthase that is in some respects similar to F- and V-type ATPases from other organisms. H. volcanii contains an A0A1-ATPase that has been characterized with respect to the ATP hydrolyzing capacity. Conditions for the ATPase function are high NaCl concentrations, a medium pH of 9, and the presence of divalent cations. The optimal temperature for ATP hydrolysis is 60 °C (1, 2). ATP synthesis performed by membrane vesicles of H. volcanii should be due to this enzyme. To confirm this, it was essential to analyze the optimal conditions for the ATP synthesis reaction.
ATP synthesis in washed membranes of H. volcanii was optimal
at pH 9. Lower pH values decreased the activity very strongly, whereas
at pH 10, 75% of the optimal activity was left. The optimal temperature for ATP synthesis is 60 °C, whereas at 25 °C, only 10% of the optimal activity was obtained (30% at 70 °C). In
contrast to the ATP hydrolysis reaction, where Mn2+ and
Mg2+ could be used almost equally, the ATP synthesis
reaction strictly depends on Mg2+. Mn2+ can
replace Mg2+, but with a dramatic loss of activity (Fig.
1). Like ATP hydrolysis, ATP synthesis could be
performed in 1.75 mol/liter KCl instead of NaCl. Using Mg2+
and ADP as substrates, we determined a Michaelis-Menten constant of
0.67 mmol/liter (ADP) and a Vmax of 18.2 nmol of
ATP/mg of protein/min; using Mn2+ instead of
Mg2+ diminished the Vmax to 7.1 nmol
of ATP/mg of protein/min (Fig. 1).
The ATP synthase activity was inhibited neither by the F-ATPase
inhibitors 7-chloro-4-nitrobenzo-2-oxa-1,3-diazole (up to 1 mmol/liter)
and phlorizin (up to 2 mmol/liter) nor by the V-ATPase inhibitor
bafilomycin (0.1 mmol/liter). N-Ethylmaleimide, a potent V-ATPase inhibitor, caused only a 5-10% loss of activity if the vesicles were preincubated for 2 h and a concentration of 10 mmol/liter was employed. These results are in accordance with our
previous analysis of ATP hydrolysis by the Haloferax ATPase
(2). Incubation of the membranes overnight at 4 °C in the presence
of 1 mmol/liter N,N-dicyclohexylcarbodiimide
resulted in a residual activity of 18% of the ATP synthesis rate
measured with vesicles that were stored under the same conditions
without N,N
-dicyclohexylcarbodiimide. Lower
concentrations of N,N
-dicyclohexylcarbodiimide
and short incubation times had only negligible effects.
Carboxyl cyanide p-trifluoromethoxyphenylhydrazone, an
uncoupler of photophosphorylation in photosynthesis, is able to
diminish ATP synthesis very effectively if added in relatively high
concentrations (final concentration of 200 µmol/liter carboxyl
cyanide p-trifluoromethoxyphenylhydrazone in the experiment
shown in Fig. 2). In contrast, the ionophore nigericin
is effective at concentrations that are 2 orders of magnitude lower
(50% inhibition at a concentration of 4 µmol/liter (data not
shown)).
Immunodetection of the A1 Part of the ATP Synthase
H. volcanii ATP synthase was removed from the membranes by passing the membranes through a French press at 20,000 p.s.i. (23) and was purified by fractionated precipitation using polyethylene glycol. Most of the ATP hydrolysis activity was found in the 25% polyethylene glycol fraction. The respective pellet was applied to a Superose 6 HR 10/30 column (Pharmacia) and eluted with 1.75 mol/liter NaCl plus 50 mmol/liter Tricine, pH 7.5, and 200 mmol/liter MgCl2. After this procedure, a purification factor of 90 and a corresponding yield of 32% were achieved. The specific activity (ATP hydrolysis) of the purified enzyme is 31.4 µmol of Pi/mg of protein/min. When the eluate of the Superose column was subjected to SDS-PAGE, polypeptides with apparent molecular masses of 70, 63, 55, 43, and 37 kDa were observed in addition to smaller peptides. Employing further purification methods, e.g. ammonium sulfate-mediated HPLC (24), resulted in additional small bands and in a dramatic loss of activity (to ~10% of the original activity).
Examining the immunological reactivity of the membranes with antisera
against oligopeptides that were constructed according to the deduced
amino acid sequences of the large subunits A and B (2), we found that
in the membrane fraction, one 70-kDa peptide (subunit A) and one 55-kDa
peptide (subunit B) were labeled (Fig. 3). However, when
fractions of the Superose gel filtration product were studied employing
the antisera, a diminished 70-kDa band and an additionally labeled
51-kDa peptide (subunit A) were obtained (Fig. 3). Obviously, a
fragment of 18.3 kDa was split from the 70-kDa protein employing gel
filtration. The gel filtration product of subunit B showed
immunological reactivity with the 55- and 70-kDa peptides (Fig. 3). At
least two additional bands were labeled when the product of a
subsequent HPLC column (ammonium sulfate-mediated chromatography) was
studied with the help of Western blots (data not shown). Despite the
high sodium chloride concentration maintained during every step of
purification, degradation of the halophilic ATPase seems to be
unavoidable.
Structure of the atp Gene Cluster
To obtain more data concerning the structure of the archaeal
ATPases, we cloned and sequenced the atpA and
atpB genes of the archaeon H. volcanii (2). When
sequencing atpA and atpB, we found that at least
two more open reading frames exist upstream of atpA, whereas
no reading frame can be found downstream of atpB. Moreover,
Northern blot experiments using polymerase chain reaction products of
subunits A and B (2) as probes resulted in a single labeled mRNA
transcript of ~5 kilobase pairs in both cases, supporting the idea of
an A1-encoding operon (Fig. 4). We cloned
two genomic DNA fragments, both overlapping with the previously
described fragment containing atpA and atpB, and
sequenced these fragments completely in both directions. Fig.
5 shows the restriction map for this region.
Three additional reading frames were found upstream of atpA.
The genes were designated atpD, atpC, and
atpE according to their size and similarity to known ATPase
subunits. Upstream of atpD, a 5-untranslated region of 550 base pairs exists. The noncoding regions between all reading frames are
extremely short: atpA is only four nucleotides apart from
atpB; there is a spacer of 6 base pairs between
atpA and atpE; and the start and stop codons of
the reading frames in front of atpE even overlap by 4 base pairs each. The reading frames atpD, atpC, and
atpE were translated into amino acid sequences of 194, 348, and 106 residues, respectively (Fig. 6). The calculated
molecular masses of the ATPase subunits are 22.03 (subunit D), 38.76, (subunit C), and 11.58 (subunit E) kDa. There are three reasons to
assume that the A1-ATPase genes are organized in a single
operon. (i) A single transcript of ~5 kilobase pairs was detected by
Northern hybridization; (ii) there is a typical promoter region (Fig.
6, boxed) together with a large 5
-untranslated region; and
(iii) we found a pyrimidine-rich transcription termination sequence
(Fig. 6, underlined).
Properties of the Respective Peptides and Amino Acid Sequence Identities
All five putative H. volcanii ATPase genes encode very hydrophilic proteins, with ~20% of their amino acids being aspartic acid and glutamic acid (Table I). This high amount of negatively charged amino acids is well known for halophilic proteins, which lose their native conformation and folding when studied under low salt conditions. Table I shows the amounts of hydrophobic, neutral, and hydrophilic residues of subunits C, D, and E as well as the respective calculated isoelectrical points. Subunit C is the most hydrophobic subunit of the A1 part of the ATPase, with an isoelectrical point of 4.13.
|
The first reading frame of the H. volcanii
operon (atpD) was translated into an amino acid sequence of
194 residues. The calculated molecular mass is 22.03 kDa. Data base
searches revealed that the putative gene product of atpD
resembles subunit E of the bovine kidney vacuolar ATPase (25).
Alignment of the deduced amino acid sequence with different V-ATPase E
subunits is shown in Fig. 7. Comparing subunit E of
Saccharomyces cerevisiae (26) and subunit E of the bovine
kidney V-ATPase (25) with the atpD gene product of H. volcanii resulted in 23% identical amino acids. Moreover, we
found 16.5% identity to subunit E of the V-ATPase from
Enterococcus hirae (formerly Streptococcus
faecalis) (32). In the latter eubacterium, a
H+-translocating F-ATPase resides besides a
Na+-translocating V-ATPase that plays an important role at
high pH, where the proton potential is not generated. Comparing the
translated amino acid sequence for atpD with F-ATPase
-subunits, which have similar molecular masses (spinach chloroplast
ATPase
-subunit, 20.5 kDa (27); and Odontella sinensis
-subunit, 21.1 kDa (28)), resulted in very low similarities of 12 and 13%, respectively. We found no significant similarity comparing
the atpD product with the
-subunit of the
Sulfolobus ATPase. Although the overall sequence homology
among the compared proteins is very poor, the N- and C-terminal regions
revealed some degree of sequence conservation. Cross-link experiments
have shown that the N-terminal region is involved in the connection of
the
-subunit to subunit I of the membrane part of chloroplast ATPase
(29). The
-subunit (subunit D) is supposed to be involved in proton
conduction from the membrane-integral part to the catalytic
subunits.
atpC
The second reading frame, overlapping with the first one
by four nucleotides, was translated into 348 amino acids, resulting in
a polypeptide with a calculated molecular mass of 38.7 kDa. The
calculated molecular mass of the atpC gene product
corresponds to the size of the partially purified enzyme of 37 kDa as
estimated by SDS-PAGE (Table II). The size of subunit C
is comparable to various C subunits of V-ATPases, e.g.
subunit C of E. hirae (Fig. 7). The similarity between
subunit C of V-ATPases from different sources and/or the -subunit of
F-ATPases is generally not very strong. Only 17% identical residues
could be found comparing H. volcanii subunit C with
Enterococcus subunit C (Fig. 7).
|
The third reading frame, translated into a sequence of
106 amino acids with a molecular mass 11.58 kDa, was compared with the
14-kDa subunit (subunit F) of the V-ATPase from the hawk moth Manduca sexta. The atpE gene product of H. volcanii showed 23% identical amino acids when compared with this
only known subunit F of eucaryal V-ATPases. An amount of 16% identical
amino acids was found when comparing subunit E with E. hirae
subunit G (Fig. 7), but only 12% was found when comparing with
-subunits of F-ATPases. Despite the poor similarity, the amino
terminus of the sequence shows rather high identity to M. sexta subunit F and E. hirae subunit G: 10 of 22 amino
acids are identical or conservative exchanges.
The Proton-conducting Membrane Part of the Haloferax ATPase
Since the H. volcanii ATPase operon does not contain a
gene coding for the membrane-spanning proton channel or subunit c, we
isolated this peptide from the membranes according to Denda et
al. (16) and Kakinuma et al. (30). After purification, a single protein was obtained with a molecular mass of 9.7 kDa as
determined by SDS-PAGE. To identify this protein, it was blotted onto
polyvinylidene difluoride membrane and transferred to a gas-phase sequencer for automatic amino acid sequence analysis. The first 10 amino acids (Fig. 8) indicate that the peptide is
related to the proteolipids (c subunits) of E. coli and
spinach chloroplast ATPase, whereas there is almost no similarity to
the N terminus of the respective peptide of Sulfolobus
acidocaldarius (30).
So far there is only one complete archaeal ATPase operon described
in the literature: the acidothermophile archaeon S. acidocaldarius contains five genes encoding hydrophilic
polypeptides (atpD, atpA, atpB,
atpG, and atpE) that precede one gene
(atpP) coding for a very hydrophobic subunit (c or
"proteolipid") (31). The authors describe the small subunits to
have no significant similarity either to F-ATPase or to eucaryal
V-ATPase subunits, whereas the large - and
-subunits and the
channel-building subunit c do have homologous counterparts in other
ATPases.
The H. volcanii ATPase operon is the first reported
halophilic archaeal ATPase gene cluster. In Fig. 9, the
arrangement of the Haloferax genes is compared with the
respective operon of the archaeon S. acidocaldarius (31).
Furthermore, the gene arrangement coding for the V-ATPase
(Na+-ATPase) from E. hirae (32) as well as the
F-ATPase gene clusters from E. coli (33) and
Rhodospirillum rubrum (34) are shown. The most surprising
finding was that the Haloferax operon is quite different
from the operon in the acidothermophilic archaeon S. acidocaldarius, whereas complete correspondence could be found when comparing the Haloferax operon with the eubacterial
E. hirae Na+-ATPase gene cluster. Thus, our
finding indicates a conserved operon structure for V- and A-ATPases. In
a very recent paper, Wilms et al. (47) reported on the
organization of a gene cluster encoding the archaeal ATP synthase from
Methanosarcina mazei Gö1. This gene cluster combines
the arrangement of that of H. volcanii with the one of
S. acidocaldarius since it contains genes homologous to the
Sulfolobus - and
-subunits. In
Enterococcus, the proteolipid-encoding gene is
atpK, positioned in front of atpE, which encodes
the peptide homologous to H. volcanii subunit D. The operon
of the S. acidocaldarius ATPase ends with the gene for the
hydrophobic subunit c that builds up the proton-conducting,
membrane-spanning part of the ATPase. In the photosynthetic bacterium
R. rubrum, the F1- and F0-subunits are encoded by different operons, which are transcribed independently (34). This may also be true for H. volcanii since we could
not find any coding region for the hydrophobic membrane part or for a
related membrane-spanning protein.
The molecular mass of the so-called proteolipid in the F0 part of F-ATPases is ~8 kDa, whereas V-ATPases possess proteolipids with molecular masses of ~16 kDa, consisting of four membrane-spanning helices. A data bank search revealed that the first half and the last half of the protein are both similar to the F-ATPase proteolipid (35). Thus, both proteolipids are proposed to have originated from a common ancestral gene that underwent gene duplication (35). From the structures described above and the apparent similarities of F-, A-, and V-ATPases, the structure of the Haloferax ATP synthase has been predicted on the basis of the well known structure of F-ATPases (36).
Since the universal phylogenetic tree has been restructured by Woese et al. (37), the halophilic bacteria belong to the Euryarchaeota together with the methanogens and their relatives. The second archaebacterial kingdom, named Crenarchaeota, contains the thermoacidophiles. Together, they form the domain Archaea. This new order was mainly based upon 16 S RNA sequence comparisons (38, 39). The presence of a number of eucaryal features in archaea, including the relationship between the large subunits of the A-ATPases and eucaryal V-ATPases, has led to a closer grouping of archaea and eucarya than of archaea and eubacteria (37). Whereas the origin of the F-ATPases is rooted in eubacteria, which evolved into the cell organelles mitochondria and chloroplasts, according to Nelson and Taiz (40) and Gogarten et al. (41), the origin of V-ATPases is proposed to be related to archaebacteria. Nelson (42) proposed that evolution of the V-ATPases to become exclusively proton pumps is related to the duplication of their gene encoding the proteolipid.
Comparing the operon structure of the E. hirae V-ATPase gene
with the Haloferax ATPase operon, we were able to confirm
the proposed relationship concerning the A1 and
V1 parts of archaeal and vacuolar ATPases. Yet, the
membrane-embedded proteolipid seems to be more closely related to the
respective protein of F-ATPases, as shown by its size and N-terminal
sequence. It has been proposed that the gene clusters for the catalytic
portion of the ATPases initially have evolved separately (for review,
see Ref. 42). As an example for separated operons, the gene clusters of
R. rubrum are shown in Fig. 9. The split between the operons
is between the - and
-subunits. The H. volcanii gene
cluster ends with atpD, encoding the
-homolog subunit D. The same position can be observed for the gene encoding the
-subunit
of S. acidocaldarius, although the arrangement of the genes
is different from that of Haloferax. Work to obtain more
information about the second ATPase gene cluster that codes for the
A0 part of the Haloferax ATPase is in
progress.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X79516[GenBank].
Protein sequencing was performed by D. Kienzle (Forschungszentrum Jülich). We thank Dr. Jörg Soppa (Max-Planck-Institute Martinsried) for the generous gift of H. volcanii strain WR 340 and for help. We thank Dr. David Mozley and Martina Kroth for critical reading of the manuscript and for linguistic corrections.