Báo cáo khoa học: A novel isoform of pantothenate synthetase in the Archaea potx

A novel isoform of pantothenate synthetase in
the Archaea
Silvia Ronconi, Rafal Jonczyk and Ulrich Genschel
Lehrstuhl fu
¨
r Genetik, Technische Universita
¨
tMu
¨
nchen, Freising, Germany
Pantothenate is the essential precursor to CoA, which
is of central importance for all parts of metabolism.
This is shown by the fact that more than 400 enzyme-
catalyzed reactions are known to involve CoA (KEGG
database [1]). Many more enzymes utilize acylated
forms of CoA or require the CoA-derived phospho-
pantetheine as a prosthetic group. Typically, plants,
fungi and microorganisms are able to synthesize panto-
thenate de novo, whereas animals rely on pantothenate
in their diet.
Pantothenate synthetase (PS) catalyzes the last step in
the biosynthesis of pantothenic acid, also known as vita-
min B
5
. The enzyme (EC 6.3.2.1) has been extensively
studied in Escherichia coli [2,3], Mycobacterium tubercu-
losis [4,5], and Arabidopsis thaliana [6], and is highly
conserved in the Bacteria and Eukaryota. Bacterial PS
(Eqn 1) generates pantothenate from pantoate and
b-alanine. It is an AMP-forming synthetase that
proceeds via an acyl-adenylate intermediate and belongs

to the HIGH superfamily of nucleotidyltransferases [3].
Keywords
archaeal metabolism; CoA biosynthesis;
evolution of metabolism;
Methanosarcina mazei; pantothenate
synthetase
Correspondence
U. Genschel, Lehrstuhl fu
¨
r Genetik,
Technische Universita
¨
tMu
¨
nchen, Am
Hochanger 8, 85350 Freising, Germany
Fax: +49 8161 715636
Tel: +49 8161 715644
E-mail:
(Received 7 February 2008, revised 17
March 2008, accepted 19 March 2008)
doi:10.1111/j.1742-4658.2008.06416.x
The linear biosynthetic pathway leading from a-ketoisovalerate to panto-
thenate (vitamin B
5
) and on to CoA comprises eight steps in the Bacteria
and Eukaryota. Genes for up to six steps of this pathway can be identified
by sequence homology in individual archaeal genomes. However, there are
no archaeal homologs to known isoforms of pantothenate synthetase (PS)
or pantothenate kinase. Using comparative genomics, we previously identi-

fied two conserved archaeal protein families as the best candidates for the
missing steps. Here we report the characterization of the predicted PS gene
from Methanosarcina mazei, which encodes a hypothetical protein
(MM2281) with no obvious homologs outside its own family. When
expressed in Escherichia coli, MM2281 partially complemented an auxo-
trophic mutant without PS activity. Purified recombinant MM2281 showed
no PS activity on its own, but the enzyme enabled substantial synthesis of
[
14
C]4¢-phosphopantothenate from [
14
C]b-alanine, pantoate and ATP when
coupled with E. coli pantothenate kinase. ADP, but not AMP, was
detected as a coproduct of the coupled reaction. MM2281 also transferred
the
14
C-label from [
14
C]b-alanine to pantothenate in the presence of panto-
ate and ADP, presumably through isotope exchange. No exchange took
place when pantoate was removed or ADP replaced with AMP. Our results
indicate that MM2281 represents a novel type of PS that forms ADP and
is strongly inhibited by its product pantothenate. These properties differ
substantially from those of bacterial PS, and may explain why PS genes, in
contrast to other pantothenate biosynthetic genes, were not exchanged
horizontally between the Bacteria and Archaea.
Abbreviations
COG, clusters of orthologous groups; KPHMT, ketopantoate hydroxymethyltransferase; KPR, ketopantoate reductase; PANK, pantothenate
kinase; PS, pantothenate synthetase.
2754 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS

D-pantoate þ b-alanine þ ATP ! D-pantothenate þ AMP þ PP
i
ð1Þ
CoA biosynthesis is best understood in the Bacteria,
where eight steps lead from a-ketoisovalerate to CoA
(Fig. 1) [7,8]. The eukaryotic CoA pathway has been
studied to various degrees in fungi, plants, and animals
[9], and consists of both highly conserved bacterial-type
enzymes and divergent isoforms. Much less is known
about CoA biosynthesis in the Archaea. We previously
used comparative genomics to reconstruct the universal
CoA biosynthetic pathway in the Bacteria, Eukaryota,
and Archaea [10]. Archaeal genes for the ultimate four
steps can be identified by homology in all archaeal
genomes, and experimental confirmation of this assign-
ment is available for three of these steps [11,12]. In
addition, bacterial-type genes for the first two steps are
obvious in a number of the nonmethanogenic Archaea.
However, homologs to bacterial PS or any of the three
established isoforms of pantothenate kinase (PANK)
[13] are generally missing from archaeal genomes. We
approached this problem by using a nonhomology
search strategy based on conserved chromosomal prox-
imity, a method that exploits the tendency of function-
ally related genes to cluster along the chromosome [14].
Using the archaeal CoA biosynthetic genes with homol-
ogy to bacterial or eukaryotic genes on the CoA path-
way as a starting point, this identified the clusters of
orthologous groups (COG)1701 and COG1829 protein
families as the best candidates for the PS and PANK

steps in archaeal CoA biosynthesis [10]. (COG protein
family identifiers cited in this report are defined in the
COG database [15]).
Here we report the characterization of the predicted
PS gene from Methanosarcina mazei. We conclude that
the COG1701 family represents the archaeal isoform
of PS, which utilizes the same substrates as bacterial
PS, but forms ADP instead of AMP and has distinct
kinetic properties. This supports the view that the
intermediates of pantothenate biosynthesis are univer-
sally conserved, whereas the corresponding enzymes
were recruited independently in the Bacteria and
Archaea.
Results
Prediction of conserved archaeal protein families
for PS and PANK
Genomic context and phylogenetic pattern analysis
previously identified the COG1701 and COG1829 pro-
tein families as the best candidates for the missing
steps leading from pantoate to 4¢-phosphopantothenate
in archaeal CoA biosynthesis [10]. Meanwhile, many
more archaeal genomes have been completed, and the
comparative genomics search for the missing steps was
repeated by using the STRING tool [16]. This analysis
revealed additional, previously undetected, links
between established archaeal CoA genes and the
COG1701 and COG1829 families, confirming that the
latter are strong candidates for archaeal PS and
PANK (Fig. 1).
Fig. 1. The CoA biosynthetic pathway and its reconstruction in the

Archaea. The linear pathway leading from a-ketoisovalerate to CoA
comprises eight steps in the Bacteria and Eukaryota. It proceeds
via pantoate, pantothenate (vitamin B
5
), and 4¢-phosphopantothe-
nate. The remaining intermediates, as well as the branch for pro-
duction of b-alanine, are left out for clarity. For six of these steps,
homologs can be established in the Archaea, and the corresponding
COG families are shown in color to indicate the average level of
sequence identity to the respective E. coli or human CoA biosyn-
thetic enzymes [10]. Nonhomologous functional links to the archa-
eal homologs were obtained from the STRING database [16], as
described in Experimental procedures. Taken together, the links to
COG1701 and COG1829 clearly support these protein families as
the best candidates for the missing steps in archaeal CoA biosyn-
thesis. The functional assignments for COG1701 (archaeal PS) and
COG1829 (archaeal PANK) are explained in the main text. PPCS,
phosphopantothenoylcysteine synthetase; PPCDC, phospho-
pantothenoylcysteine decarboxylase; PPAT, phosphopantetheine
adenylyltransferase; DPCK, dephospho-Co A kinase; P-pantothe-
nate, 4¢-phosphopantothenate.
S. Ronconi et al. Pantothenate synthetase from Methanosarcina mazei
FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS 2755
Consistent with the assumption that both protein
families represent archaeal isoforms of CoA biosyn-
thetic enzymes, their members are found in nearly all
archaeal genomes, but not outside the archaeal
domain. Furthermore, COG1701 and COG1829 share
a strictly conserved phylogenetic profile and frequently
occur in tandem in potential operons (e.g. in Me. maz-

ei; Fig. 2). A straightforward general function predic-
tion is possible for the COG1829 family, which
belongs to a superfamily of small molecule kinases
(GHMP kinases [17]) and is therefore proposed to rep-
resent archaeal PANK. This leaves COG1701, an
orphan family with no obvious links to other protein
families, as the best candidate for archaeal PS. Using
the hhpred prediction server [18], COG1701 was
found to be a distant homolog of acetohydroxyacid
synthase, which ligates two molecules of pyruvate to
yield acetolactate. Specifically, there is approximately
20% sequence identity between COG1701 proteins and
the b-domain of acetohydroxyacid synthases. This
domain has no specific catalytic function but is
thought to be important for the structural integrity of
acetohydroxyacid synthase [19].
Functional complementation of an E. coli
panC mutant
In the genome of Me. mazei, the predicted ORF for
PS (MM2281) is situated in a potential operon
together with the predicted PANK gene and the dfp
gene (Fig. 2), and this cluster is therefore expected to
cover the CoA biosynthetic steps leading from panto-
ate to 4¢-phosphopantetheine. The MM2281 ORF was
cloned by PCR and tested for its ability to comple-
ment the E. coli panC mutant strain AT1371 in liquid
minimal medium (Fig. 3). The panC gene, which
encodes PS in E. coli, and the empty pBluescript KS
vector served as positive and negative controls in this
experiment, respectively. All transformants grew well

in cultures supplemented with pantothenate (not
shown). The cultures generally showed long lag peri-
ods, during which the cells recovered from the starving
procedure in pantothenate-free medium (see Experi-
mental procedures). In the absence of supplements,
AT1371 cells harboring MM2281 showed a shorter lag
period and faster growth than the negative control,
but did not grow as well as the positive control
(Fig. 3A). The same pattern was observed in minimal
medium containing 1 mm pantoate, a substrate of PS,
except that the growth of cells containing MM2281 or
the negative control was stimulated (Fig. 3B). In our
hands, the E. coli panC mutant carrying empty vector
(negative control) showed minor growth in minimal
Fig. 2. Potential operon for CoA biosynthesis in Me. mazei. The
predicted genes for PS and PANK, as well as the dfp gene encod-
ing the bifunctional enzyme PPCS ⁄ PPCDC (MM2281 through
MM2283), occur in a cluster, which corresponds to the steps lead-
ing from pantoate to 4¢-phosphopantetheine. PPCS, phospho-
pantothenoylcysteine synthetase; PPCDC, phosphopantothenoyl
cysteine decarboxylase.
Fig. 3. Functional complementation of an E. coli pantothenate
auxotrophic mutant. The pantothenate-requiring E. coli mutant car-
rying the Me. mazei MM2281 gene (d), the E. coli panC gene (h)
or empty vector (s) was grown in liquid culture in the absence of
pantothenate, as described in Experimental procedures. The mini-
mal medium contained no supplements (A) or an additional 1 m
M
pantoate (B). All transformants grew equally well when the medium
was supplemented with 1 m

M pantothenate (not shown).
Pantothenate synthetase from Methanosarcina mazei S. Ronconi et al.
2756 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS
medium. This might be caused by an endogenous non-
specific activity able to produce pantothenate or by the
emergence of revertants. Nevertheless, regardless of the
actual reason for this behavior, the observation that
MM2281-carrying cells recovered more quickly from
pantothenate starvation and grew faster than the nega-
tive control in two independent experiments indicates
that expression of MM2281 partially complements the
auxotrophic phenotype of E. coli AT1371.
PS activity of recombinant MM2281
The MM2281 protein was overproduced as an N-ter-
minal His-tag fusion protein in E. coli and had a sub-
unit molecular mass in good agreement with its
predicted size (30 kDa as judged by SDS ⁄ PAGE). The
native molecular mass of MM2281 estimated by gel
filtration was 57 000 Da, indicating that the enzyme
is apparently a dimer in solution.
Purified recombinant MM2281 was checked for its
ability to synthesize pantothenate from pantoate,
b-alanine and ATP by using a sensitive isotopic assay
procedure. However, we were not able to demonstrate
PS activity of MM2281 alone. Even after incubation
for 3 h, the amount of [
14
C]b-alanine converted into
[
14

C]pantothenate was below the lower limit of detec-
tion (1% conversion). This means that PS activity of
MM2281 was absent or below 0.6 nmolÆmin
)1
Æmg
)1
in
our assay.
In an attempt to confirm pantothenate as a product
of MM2281, we coupled the reaction with excess
E. coli PANK, which converts pantothenate to 4¢-
phosphopantothenate. MM2281, individual helper
enzymes or combinations of these were assayed for
their ability to convert [
14
C]b-alanine into [
14
C]panto-
thenate or [
14
C]4¢-phosphopantothenate under stan-
dard conditions. The
14
C-labeled reaction products
were separated by TLC, revealed by phosphoimaging
(Fig. 4), and quantified to calculate specific PS activi-
ties (Table 1). Whereas MM2281 alone had no signifi-
cant pantothenate-synthesizing activity in this assay, it
was obvious that E. coli PS efficiently converted
[

14
C]b-alanine into [
14
C]pantothenate (Fig. 4, lanes 2
and 3). E. coli PANK alone did not act on [
14
C]b-ala-
nine but was conducive to quantitative formation of
[
14
C]4¢-phosphopantothenate when coupled with E. coli
PS (Fig. 4, lanes 4 and 5). As the products of E. coli
PS and E. coli PANK are firmly established, the reac-
tions with these enzymes provide chromatography
standards for pantothenate and 4¢-phosphopantothe-
nate and also confirm that the E. coli PANK prepara-
tion used here was not contaminated with detectable
PS activity. Therefore, the [
14
C]4¢-phosphopanto-
Fig. 4. Synthesis of pantothenate or 4¢-phosphopantothenate
through MM2281 (Me. mazei PS) or helper enzymes. Standard
enzyme assays were carried out as described in Experimental
procedures, containing no enzyme (control), individual enzymes,
or enzyme combinations, as indicated. The figure shows the
14
C-labeled products after a reaction time of 3 h. Separation was
achieved by TLC. The enzyme abbreviations are as follows: EcPS,
E. coli PS; EcPANK, E. coli PANK; PyrK, rabbit pyruvate kinase;
bAla, b-alanine; PA, pantothenate; PPA, 4¢-phosphopantothenate.

Table 1. Formation of pantothenate or 4¢-phosphopantothenate
through MM2281 and helper enzymes. MM2281 and helper
enzymes were tested individually or in combinations for their ability
to form [
14
C]pantothenate or [
14
C]4¢-phosphopantothenate from
[
14
C]b-alanine, pantoate and ATP under the conditions of the stan-
dard assay (Fig. 4). ND, not detectable; PyrK, rabbit pyruvate
kinase.
Enzyme(s)
Specific pantothenate synthetase
activity (nmolÆmin
)1
Æmg
)1
)
[
14
C]
Pantothenate
[
14
C]
4¢-Phosphopantothenate
MM2281 ND ND
E. coli PS > 900

a
ND
E. coli PANK ND ND
E. coli PS + E. coli PANK ND > 900
a
MM2281 + E. coli PANK ND 94 ± 11
b
MM2281 + PyrK ND ND
MM2281 + PyrK +
E. coli PANK
ND 140 ± 22
b
a
With respect to E. coli PS. This value is a lower estimate because
the reaction was complete within the first interval.
b
With respect
to MM2281.
S. Ronconi et al. Pantothenate synthetase from Methanosarcina mazei
FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS 2757
thenate produced by the combined action of MM2281
and E. coli PANK clearly demonstrates the capacity of
MM2281 to synthesize pantothenate from pantoate
and b-alanine (Fig. 4, lane 6). The rate of 4¢-phospho-
pantothenate synthesis in this assay corresponds to a
PS activity of 94 nmolÆmin
)1
Æmg
)1
with respect to

MM2281, indicating that the E. coli PANK-mediated
removal of pantothenate accelerated the synthesis of
pantothenate through MM2281 at least 100-fold.
Principally, the above behavior can be explained by
assuming either that MM2281 is potently inhibited by
pantothenate or that equilibrium is reached after only
a small fraction of substrates has reacted. In both
cases, the reaction is expected to accelerate when pan-
tothenate is removed, because this should abolish inhi-
bition or displace the equilibrium. It should be
mentioned that active removal of pantothenate has no
significant effect on the reaction rate of bacterial PS,
essentially excluding the possibility that the MM2281
preparation was contaminated with bacterial PS. This
is because pantothenate was shown to be a very weak
product inhibitor of My. tuberculosis PS [4] and the
equilibrium of the reaction catalyzed by bacterial PS
lies far on the product side. The latter statement can
be derived by considering the equilibrium constant of
the reaction catalyzed by bacterial PS (Eqn 1). The
equilibrium constant for Eqn (1) has not been deter-
mined experimentally, but can be deduced from the
equilibrium constants for the hydrolysis of pantothe-
nate into pantoate and b-alanine (K¢ = 42 at pH 8.1
and 25 °C [20]) and the phosphorolysis of ATP into
AMP and PP
i
(K¢ =3· 10
9
at pH 8 and 25 °C [21]).

Combining the above constants gives the overall equi-
librium constant for Eqn (1) at approximately pH 8
and 25 °C(K¢
Eqn (1)
= 7.2 · 10
7
). The large value
means that the reaction in Eqn (1) will go to comple-
tion under physiological conditions, including the
enzyme assay used in this study (pH 8.0, 37 °C).
Given the large effect of removing pantothenate on
MM2281, we also tested the effect of removing the
possible coproduct ADP. ADP was removed by pyru-
vate kinase, which generates ATP from ADP in the
presence of excess phosphoenolpyruvate. This system
did not detectably accelerate pantothenate synthesis by
MM2281 alone but, interestingly, increased the rate of
4¢-phosphopantothenate formation through MM2281
and E. coli PANK approximately 1.5-fold (Fig. 4,
lanes 7 and 8).
With a view to directly observing possible adenosine
nucleotide coproducts of MM2281-catalyzed pantothe-
nate synthesis, standard assays were analyzed by using
a TLC system that provides separation of ATP, ADP,
and AMP (Fig. 5). Whereas MM2281 alone had no
discernible hydrolytic activity towards ATP, the cou-
pled reaction of MM2281 and E. coli PANK generated
a substantial amount of ADP. AMP was not detected
as a coproduct, suggesting that MM2281 is not an
AMP-forming PS according to Eqn (1). By compari-

son, stoichiometric coupling of E. coli PANK, which
produces ADP, with a bacterial, AMP-forming PS
would lead to the accumulation of equimolar amounts
of ADP and AMP. The observations that synthesis of
phosphopantothenate through MM2281 and E. coli
PANK was accelerated by removing ADP and not
accompanied by production of AMP gave rise to the
hypothesis that MM2281 is an ADP-forming synthe-
tase according to Eqn (2):
D-pantoate þ b-alanine þ ATP ! D-pantothenate þ ADP þ P
i
ð2Þ
The equilibrium constant for Eqn (2) can be calcu-
lated in the same way as that for Eqn (1) (see above).
Using the equilibrium constant for phosphorolysis of
ATP into ADP and P
i
(K¢ = 1.6 · 10
7
at pH 8 and
25 °C [21]), the overall equilibrium constant for
Eqn (2) at approximately pH 8 and 25 °C becomes
Fig. 5. Detection of adenosine nucleotides produced by MM2281
(Me. mazei PS) and E. coli PANK. Standard assays containing
MM2281 alone (lane 4) or together with E. coli PANK (lane 5) were
carried out as described in Experimental procedures. Reaction mix-
tures were separated by TLC, and nucleotides were visualized under
UV light. Authentic standards of ATP, ADP and AMP were cochro-
matographed on the same plate (lanes 1–3). EcPANK, E. coli PANK.
Pantothenate synthetase from Methanosarcina mazei S. Ronconi et al.

2758 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS
K¢
Eqn (2)
= 3.9 · 10
5
. Although K¢
Eqn (2)
is smaller
than K¢
Eqn (1)
, the reaction shown in Eqn (2) will still
essentially go to completion in our enzyme assay or
under physiological conditions.
MM2281-catalyzed pantothenate–b-alanine
isotope exchange
The role of ATP and ADP in the MM2281-catalyzed
de novo synthesis of pantothenate could not be investi-
gated independently, because the assay for this forward
activity required the presence of E. coli PANK, which
utilizes ATP and generates ADP. In order to circum-
vent this problem, we assayed MM2281 alone for its
ability to catalyze an isotope exchange between [
14
C]b-
alanine and pantothenate (Table 2). The cosubstrate
dependence of this exchange activity then allowed con-
clusions about the role of adenosine nucleotides and
the mechanism of MM2281. Generally, isotope
exchange between a given substrate–product pair
occurs in the presence of all cosubstrates and coprod-

ucts (complete system) or, if the enzyme catalyzes a
partial reaction, in the presence of a subset of reac-
tants. Each cosubstrate may either be indispensable for
the exchange reaction to occur or merely affect the
exchange rate [22].
The MM2281-catalyzed incorporation of
14
C-label
from [
14
C]b-alanine into pantothenate was investigated
in the presence of full sets or subsets of the reactants
in Eqn (1) or Eqn (2), respectively (Table 2, Experi-
ment I). In the presence of the full set of reactants
(complete system), the assay based on Eqn (2) revealed
a five-fold higher rate than that based on Eqn (1).
Removing pantoate reduced the incorporation of
14
C-label into pantothenate to negligible levels in both
the Eqns (1,2) systems. In contrast, removing ATP
abolished the accumulation of [
14
C]pantothenate only
in the Eqn (1) system, whereas the Eqn (2) system
retained approximately 50% exchange activity. This
means that the pantothenate–b-alanine exchange in the
Eqn (2) system has no absolute requirement for ATP,
and this result was confirmed by a second set of iso-
tope exchange assays (Table 2, Experiment II). When
inorganic phosphate (P

i
) was removed from the
Eqn (2) system in addition to ATP, there was no sig-
nificant further reduction in exchange activity, showing
that both ATP and P
i
are dispensable. However, when
both ATP and ADP were removed, the resulting
exchange activity was negligible. In summary,
MM2281 catalyzed significant transfer of
14
C-label
from b-alanine to pantothenate in presence of pantoate
and ADP. When pantoate was removed, the resulting
pantothenate–b-alanine exchange was negligible. Also,
the exchange reaction occurred only in the presence of
adenosine nucleotide, and ADP but not AMP could
satisfy this requirement.
Again, this behavior shows that the MM2281 prepa-
rations were not contaminated with E. coli PS, because
bacterial PS requires only AMP to catalyze the panto-
thenate–b-alanine isotope exchange [4,6]. The data in
Table 2 also show that, apart from pantoate, both
ATP and ADP have a strong effect on the rate of the
MM2281-catalyzed pantothenate–b-alanine exchange
reaction. The simplest explanation for this behavior is
that pantoate, ATP and ADP are all substrates or
products of MM2281, which is consistent with the
notion that the enzyme is a synthetase that drives
pantothenate formation by hydrolysis of ATP.

MM2281 alone showed no detectable net synthesis
of pantothenate in the forward assay (see above;
Fig. 4), and the forward rate would be expected to be
even lower in the presence of products. We assume,
therefore, that the transfer of
14
C-label from b-alanine
to pantothenate was due to isotope exchange. Maximal
exchange activity occurred in the complete system of
Eqn (2), pointing to Eqn (2) as the basic reaction for
MM2281. Moreover, our data suggest that MM2281
is able to catalyze an ADP-dependent, but not an
AMP-dependent, pantothenate–b-alanine exchange
Table 2. MM2281-catalyzed isotope exchange between [
14
C]b-ala-
nine and pantothenate. MM2281 was assayed for its ability to
transfer
14
C-label from [
14
C]b-alanine to pantothenate in the pres-
ence or absence of cosubstrates. The cosubstrates in the complete
system are ATP, AMP, PP
i
, and pantoate (Eqn 1), or ATP, ADP, P
i
,
and pantoate (Eqn 2). Different preparations of MM2281 were used
in two independent experiments (Experiments I and II). Missing val-

ues indicate that the cosubstrate combination indicated was not
tested. ND, not detectable.
Reactants
Initial exchange rate
a
(%)
Experiment I Experiment II
Eqn (2)
Complete system 100 100
Minus pantoate 2
Minus ATP 46 50
Minus ATP, minus pantoate 3
Minus ATP, minus P
i
45
Minus ATP, minus ADP 3
Eqn (1)
Complete system 20
Minus pantoate 1
Minus ATP ND
Minus ATP, minus pantoate ND
a
Normalized to the value in the complete system of Eqn (2), which
was equal to 2.5 and 1.5 · 10
)3
Æmin
)1
in Experiments I and II,
respectively.
S. Ronconi et al. Pantothenate synthetase from Methanosarcina mazei

FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS 2759
reaction in the presence of pantoate. This is difficult to
reconcile with Eqn (1), and further supports Eqn (2) as
the overall reaction catalyzed by MM2281. The resid-
ual exchange activity in the complete system of
Eqn (1) may be due to contamination of the commer-
cial ATP preparation with ADP.
We also considered the theoretical possibility that
MM2281 is a hydrolase, which facilitates the equilib-
rium between pantoate, b-alanine and pantothenate
according to Eqn (3):
D-pantoate þ b-alanine $ D-pantothenate ð3Þ
Although there is no simple argument to rule out
Eqn (3) as the basic reaction of MM2281, explaining
the overall behavior of MM2281 in this way is very
difficult and also generates a conflict with the reported
value for K¢
Eqn (3)
. Most importantly, Eqn (3) implies
that ATP and ADP are not substrates of MM2281.
The observed effects of ADP and ATP on the isotope
exchange rate can then be explained by assuming that
ADP and ATP effectively promote a switch from inac-
tive to active enzyme. However, this is at odds with
the observation that enzymatic removal of ADP accel-
erated MM2281 in the forward direction (see above).
Second, considering the equilibrium constant of
Eqn (3) (K¢
Eqn (3)
=1⁄ 42 at pH 8.1 and 25 °C [21]),

this reaction clearly favors hydrolysis of pantothenate.
Thus, based on Eqn (3) and K¢
Eqn (3)
, the majority of
the pantothenate in the isotope exchange assay used
here would be converted to pantoate and b-alanine.
Using K¢
Eqn (3)
and the initial concentrations of panto-
ate, b-alanine and pantothenate in the assay, the
maximum fraction of
14
C-label associated with panto-
thenate at equilibrium would be 35%. However, we
observed that [
14
C]pantothenate accumulated up to
60% of the total
14
C-label during the assay. This cor-
responds to an equilibrium constant of ‡ 1 ⁄ 15, which
is much larger than the reported value for K¢
Eqn (3)
.
Discussion
Experimental confirmation of computationally
predicted archaeal PS
Metabolic reconstruction of the CoA biosynthetic
pathway in representative organisms previously
revealed that the Archaea lack known genes for the

conversion of pantoate into 4¢-phosphopantothenate.
The protein families COG1701 and COG1829 were
then identified as the best candidates for the missing
steps by comparative analysis of the 16 completely
sequenced archaeal genomes available at the time [10].
The STRING database, which currently integrates 26
archaeal genomes, revealed additional functional asso-
ciations that support a role for COG1701 and
COG1829 in archaeal CoA biosynthesis (Fig. 1). On
the basis of distant homology relationships, we tenta-
tively assigned the PS and PANK functions to the
COG1701 and COG1829 protein families, respectively.
Three lines of experimental evidence support the
computational prediction of archaeal PS. First, the
cloned COG1701 member from Me. mazei (MM2281)
partially complemented the auxotrophic phenotype of
an E. coli mutant lacking PS activity (Fig. 3). Second,
the recombinant proteins MM2281 and E. coli PANK
together facilitated the synthesis of 4¢-phosphopantoth-
enate from pantoate, b-alanine, and ATP (Fig. 4).
Arguably, the enzyme preparations were not contami-
nated with bacterial PS, allowing the conclusion that
pantothenate synthesis in the coupled assays was due
to MM2281. Third, MM2281 catalyzed the transfer of
14
C-label from b-alanine to pantothenate, presumably
by isotope exchange, in a cosubstrate-dependent man-
ner (Table 2). Given that function is typically con-
served within orthologous groups [15], demonstration
of PS activity for one member of the group provides

strong support for the prediction that COG1701 repre-
sents the archaeal PS protein family.
Properties of MM2281 (Me. mazei PS)
Our data suggest that MM2281 is an ADP-forming
pantothenate synthetase (Eqn 2) that is subject to
strong product inhibition by pantothenate. The behav-
ior of MM2281 in the isotope exchange experiments
clearly suggests that MM2281 is an adenosine nucleo-
tide-dependent pantothenate synthetase and not a
reversible pantothenate hydrolase. On the basis of the
large values of the equilibrium constants for Eqns (1,2)
(see above), the equilibria of both reactions can be
assumed to lie on the side of pantothenate formation.
In other words, regardless of the type of synthetase
reaction, coupling of pantothenate synthesis from pan-
toate and b-alanine to the hydrolysis of ATP will drive
the equilibrium to the product side. Therefore, the
strong acceleration of MM2281-catalyzed pantothenate
synthesis by the removal of pantothenate is very prob-
ably not due to a shift in the equilibrium of the reac-
tion, leaving potent inhibition of MM2281 by
pantothenate as the best explanation. We propose that
MM2281 is an ADP-forming synthetase according to
Eqn (2), because this is consistent with the observation
that the synthesis of 4¢-phosphopantothenate through
MM2281 and E. coli PANK was accompanied by the
accumulation of ADP but not AMP (Fig. 5) and accel-
erated by removing ADP. Furthermore, the isotope
Pantothenate synthetase from Methanosarcina mazei S. Ronconi et al.
2760 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS

exchange data can readily be accounted for by Eqn (2)
but not by Eqn (1).
The highest PS activity of MM2281 observed in this
study was 140 nmolÆmin
)1
Æmg
)1
, which is equivalent to
a turnover of 0.07 s
)1
. This is significantly below typi-
cal values for k
cat
, which range from about 0.8 s
)1
to
2 · 10
5
s
)1
[22]. More specifically, the MM2281 activ-
ity reported here was more than 15-fold lower than
that of E. coli PS under optimal conditions (calculated
from data in [2]) and nearly 50-fold lower than the k
cat
reported for My. tuberculosis PS [4]. Given the com-
paratively low activity of MM2281, it is possible that
the enzyme requires an activator or cofactor that was
absent from the standard assay. One attractive candi-
date for this role is the predicted PANK in Me. mazei

(MM2282; Fig. 2), which may be more effective than
E. coli PANK in accelerating MM2281. Moreover,
conserved phylogenetic profiles and chromosomal
proximity indicate a strong functional link between
archaeal PS (COG1701) and archaeal PANK
(COG1829) (Fig. 1). Also, lack of an interacting pro-
tein required for optimal activity could explain why
expression of MM2281 achieved only partial comple-
mentation of the E. coli panC mutant (Fig. 3). How-
ever, our attempts to express and purify MM2282 did
not meet with success (data not shown), so this
hypothesis could not be tested.
The observation that MM2281 facilitated the panto-
thenate–b-alanine exchange in the absence of ATP and
P
i
may be taken to indicate that MM2281 is a Ping
Pong enzyme able to catalyze a partial reaction. How-
ever, the isotope exchange data in Table 2 show clearly
that the kinetic mechanism of MM2281 is different
from the Ping Pong system of bacterial PS. The latter
consists of two half-reactions, which proceed via an
enzyme-bound pantoyl adenylate intermediate [2–5].
As a result, the pantothenate–b-alanine exchange reac-
tion of bacterial PS is independent of pantoate and
has an absolute requirement for only AMP. By com-
parison, the cosubstrate dependence of the MM2281-
catalyzed exchange reaction differs on several counts
(see above) and is inconsistent with the Ping Pong
system of bacterial PS. This shows that archaeal

PS evolved a distinct mechanism to synthesize
pantothenate.
Evolution of phosphopantothenate biosynthesis
We hypothesize that the entire upstream portion of
CoA biosynthesis, leading from common precursors to
phosphopantothenate, evolved independently in the
Bacteria and Archaea. This view was initially based on
the finding that many archaeal genomes contain no
homologs to any of the corresponding bacterial
enzymes and on the prediction of distinct archaeal
forms of PS and PANK [10]. The experimental
evidence in this study provides strong support for the
predicted identity of archaeal PS (COG1701). This, in
turn, also supports the prediction of archaeal PANK
(COG1829), because there are strong nonhomologous
links between the two families (Fig. 1).
The linear pathway from a-ketoisovalerate to phos-
phopantothenate comprises four steps in the Bacteria
(Fig. 1). So far, unrelated archaeal genes on this path-
way have been computationally predicted for the third
and fourth steps (i.e. PS and PANK), but not for the
first and second steps [i.e. ketopantoate hydroxymeth-
yltransferase (KPHMT) and ketopantoate reductase
(KPR)]. Interestingly, the phyletic distribution of PS
genes indicates that they were not subject to horizontal
transfer, in either direction, between the archaeal and
bacterial domains. Archaeal PS and PANK isoforms
show strict co-occurrence and are present in most of
the Archaea, except in the Thermoplasmata class of
the Euryarchaeota and in Nanoarchaeum equitans.

Also, they are absent from the Bacteria and
Eukaryota. All of the Archaea that have archaeal-type
PS and PANK universally lack homologs to bacterial
or eukaryotic PS and PANK isoforms. In fact, bacte-
rial PS is entirely absent from the archaeal domain,
and archaeal homologs to bacterial PANK are limited
to the Thermoplasmata class.
A different situation is encountered for the first two
CoA biosynthetic steps. In the Bacteria, these steps
are catalyzed by KPHMT and KPR, which convert
a-ketoisovalerate into pantoate. A subset of the non-
methanogenic Archaea acquired these enzymes, pre-
sumably by horizontal gene transfer, from
thermophilic bacteria [10]. Individual archaeal
genomes encode either both bacterial-type KPHMT
and bacterial-type KPR or either one or none of them.
This pattern suggests that some archaeal species
produce pantoate by combining an archaeal KPHMT
isoform with bacterial-type KPR or by combining an
archaeal KPR isoform with bacterial-type KPHMT.
In other words, the distribution of bacterial-type
KPHMT and KPR genes supports the view that the
majority of the Archaea contain so far unidentified
genes that encode unrelated isoforms of KPHMT and
KPR. Moreover, the observed distribution may well
be the result of nonorthologous gene displacement
[23], where the archaeal isoforms of KPHMT and
KPR were individually replaced by their bacterial
counterparts in certain archaeal species. Verification of
this hypothesis awaits, of course, identification of the

archaeal genes for pantoate synthesis.
S. Ronconi et al. Pantothenate synthetase from Methanosarcina mazei
FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS 2761
Given that horizontal gene transfer occurred exten-
sively between the thermophilic Bacteria and Archaea
[24,25], the question arises of why KPHMT and KPR
genes were transferred, whereas PS genes were not.
The archaeal PS (MM2281) characterized in this study
produces pantothenate from the same precursors as
bacterial PS but has clearly distinct kinetic properties.
The most striking difference is the apparent inhibition
of MM2281 by pantothenate, raising the possibility
that this step has a role in regulating archaeal CoA
biosynthesis. In contrast, the most important control
point in bacterial CoA biosynthesis is PANK [9], and
no regulatory function is known for PS. It is attractive
to speculate, therefore, that horizontal gene transfer
of PS, and possibly PANK, was suppressed by incom-
patible regulatory properties.
Experimental procedures
Materials
E. coli strain AT1371 [panC4, D(gpt-proA)62, lacY1,
tsx-29, glnV44(AS), galK2(Oc), LAM-, Rac-0, hisG4(Oc),
rfbD1, xylA5, mtl-1, argE3(Oc), thi-1] [26] was obtained
from the E. coli Genetic Stock Center, Yale University.
[3-
14
C]b-alanine (55 mCiÆmmol
)1
) was from American

Radiolabeled Chemicals ⁄ Biotrend Chemikalien (Cologne,
Germany). Rabbit pyruvate kinase and all other reagents
were from Sigma-Aldrich (Munich, Germany) unless indi-
cated otherwise. d-Pantoate was prepared from d-pantoyl
lactone as described elsewhere [27]. Genomic DNA from
the Me. mazei strain Goe1 (DSM 3647) was a gift from
K. Pflu
¨
ger, Universita
¨
tMu
¨
nchen.
Cloning of the Me. mazei and E. coli genes for PS
The Me. mazei ORF MM2281 (GenBank accession number
AE008384) was PCR-amplified from genomic DNA by
using Pfu polymerase (Stratagene, Amsterdam, the Nether-
lands) and the primers dGCGCGCATATGACcGATATtC
CGCACGAtCACCCGcGcTACGAATCC and dGCGCGC
TCGAGTtAGTAgCCgGTTTCCGCGGCCATGGT. The
start and stop codons are in bold. Lower-case letters desig-
nate silent nucleotide changes that were introduced to
reduce the number of rare codons for expression in E. coli.
The amplified ORF was subcloned via NdeI and XhoI
restriction sites in the primers into the pET28-a vector
(Novagen ⁄ Merck Chemicals, Darmstadt, Germany). The
resulting plasmid, pET–MM2281, contains the MM2281
ORF in translational fusion with the vector-encoded
N-terminal His-tag, leading to the expression of NH
2

-
MGSSHHHHHHSSGLVPRGSH-MM2281.
For functional complementation of the E. coli panC
mutant (AT1371), the MM2281 ORF was reamplified from
pET-MM2281 using the primers dGCGCG
AGAAGGAG
ATATACCATGACCGATATTCCGCACGATCACCCGC
GC and dGCGCGCTCGAGTTAGTAGCCGGTTTCCG
CGGCCATGGT. The ribosome-binding site in the for-
ward primer is underlined, and the start and stop codons
are in bold. The PCR product was inserted into the
pGEM-T vector (Promega, Mannheim, Germany), and a
clone carrying the MM2281 ORF in the correct orientation
for expression under the lac promoter was selected and
named pGEM–MM2281. The E. coli panC ORF was
amplified from genomic DNA of E. coli strain XL1Blue
(Stratagene) using Pfu polymerase and the primers
dCGCGCCTCG
AGGAGGAGTCACGTTATGTTAATTA
TCGAAACC and dGCGCGTCTAGATTACGCCAGCTC
GACCATTTT. The PCR product was inserted into pBlue-
script KS (Stratagene) via the restriction sites XhoI and
XbaI. The resulting plasmid (pBKS–panC) harbors the
panC gene under the control of the lac promoter and served
as a positive control in the functional complementation
experiment. Automated DNA sequencing of the inserts
in pET–MM2281, pGEM–MM2281 and pBSK–panC con-
firmed the desired sequences.
Functional complementation of E. coli
AT1371 (panC

–
)
The plasmids pGEM–MM2281 and pBKS–panC (positive
control) and the empty pBluescript KS– vector (negative
control) were introduced into the pantothenate-auxotrophic
E. coli strain AT1371. Single colonies of the transformants
were grown overnight at 37 °C in 5 mL of liquid dYT
medium (1.6% tryptone, 1% yeast extract, 0.5% NaCl)
containing 100 lgÆlL
)1
ampicillin. The E. coli cells were
pelleted and washed twice in 5 mL of GB1 buffer [100 mm
potassium phosphate, pH 7.0, 2 gÆL
)1
(NH
4
)
2
SO
4
]. The
pelleted cells were resuspended in GB1 buffer, adjusted to
an D
600 nm
of 0.3, and incubated at 25 °C for 1 h. The
starved cells were then used to inoculate [0.5% (v ⁄ v)] the
experimental cultures (4 gÆL
)1
glucose, 0.25 gÆL
)1

MgSO
4
.10H
2
O, 0.25 mgÆL
)1
FeSO
4
.7H
2
O, 5 mgÆL
)1
thia-
mine, 68 mgÆL
)1
adenine, 127 mgÆL
)1
l-arginine, 16 mgÆL
)1
l-histidine, 230 mgÆL
)1
l-proline, and 100 mgÆL
)1
ampicil-
lin in GB1 buffer), which were grown at 37 °C with
shaking. For each transformant, three cultures were started
that contained an additional 1 mm pantothenate, 1 mm
pantoate, or no further supplements, respectively. D
600 nm
was determined over an incubation time of 24 h.

Overexpression and purification of MM2281 and
helper enzymes
The MM2281 protein was expressed in E. coli BL21(DE3)
carrying the pET–MM2281 plasmid described above and
purified on Ni–nitrilotriacetic acid agarose (Qiagen, Hilden,
Pantothenate synthetase from Methanosarcina mazei S. Ronconi et al.
2762 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS
Germany) following the manufacturer’s standard protocol.
After affinity chromatography, MM2281 was loaded onto a
MonoQ anion exchange column equilibrated in 50 mm
Tris ⁄ HCl (pH 8.8). In a linear 0–1 m KCl gradient,
MM2281 eluted at approximately 350 mm KCl. The
enzyme preparation was then dialyzed exhaustively against
50 mm Tris ⁄ SO
4
(pH 8.0) and 5 mm dithiothreitol, frozen
in liquid N
2
, and stored in aliquots at )70 °C. The native
molecular mass of MM2281 was estimated by gel filtration
chromatography as previously described [6]. E. coli PS [6]
and E. coli PANK [28] were overexpressed and purified as
previously described. Protein concentrations were deter-
mined using the Bradford protein assay kit (Bio-Rad,
Munich, Germany) with BSA as standard.
Enzyme assays
The standard assay for PS activity contained 20 mm potas-
sium d-pantoate, 1 mm b-alanine, 0.08 mm [3-
14
C]b-alanine

(55 mCiÆmmol
)1
), 5 mm ATP, 10 mm MgSO
4
, 7.5 mm
K
2
SO
4
,5mm dithiothreitol, 50 mm Tris ⁄ SO
4
(pH 8.0) and
2.5 lg of MM2281 in a final volume of 25 lL. The reaction
was initiated by the addition of substrates, and incubated
at 37 °C, and 5 lL aliquots were removed at 30, 90 and
180 min time points. Separation of reaction products
(10 nCi aliquots) by TLC, quantitation of
14
C-label above
nonenzymatic activity and estimation of initial rates was as
previously described [6]. The detection of
14
C-label was lin-
ear between 0.1 and 20 nCi, covering a range of 1–100% of
the amount analyzed per time point. The assay was carried
out in the absence or presence of E. coli PANK (2.5 lg) or
pyruvate kinase from rabbit (2 units). Phosphoenolpyruvate
(2 mm) was included in the assay when pyruvate kinase was
present. Control reactions in the absence of MM2281 con-
tained either or both of the helper enzymes E. coli PS

(1 lg) and E. coli PANK (2.5 lg) or no enzymes.
In order to detect possible adenosine nucleotide products
of MM2281, standard assays containing MM2281 alone or
together with E. coli PANK were carried out as described
above, except that [3-
14
C]b-alanine was omitted. The reac-
tion was quenched after 180 min, and the products were
cochromatographed with authentic ATP, ADP and AMP
standards (Sigma). Adenosine nucleotides were separated
on silica plates using dioxane ⁄ NH
3
(25%) ⁄ H
2
O(6:1:4)
as a mobile phase and detected under UV light (254 nm).
Isotope exchange assay
The pantothenate–b-alanine isotope exchange was assayed
at 25 °C, and the standard reaction contained 1 mm b-ala-
nine, 0.07 mm [3-
14
C] b-alanine (55 mCiÆmmol
)1
), 5 mm
pantothenate, 5 mm ADP, 5 mm sodium phosphate, 5 mm
ATP, 20 mm potassium d-pantoate, 10 mm MgSO
4
,
7.5 mm K
2

SO
4
,5mm dithiothreitol, 50 mm Tris ⁄ SO4
(pH 8.0), and 1.1 lgÆlL
)1
MM2281. Individual reactions
contained all of the above components [Eqn (2), complete
system] or lacked one or more of the reactants ATP, ADP,
pantoate, or sodium phosphate. A second set of exchange
reactions was carried out with AMP and sodium pyrophos-
phate replacing ADP and sodium phosphate in the above
scheme [Eqn (1), complete system]. Aliquots were removed
from the reactions at 7, 24 and 48 h after initiation by the
addition of MM2281. Quantitation of
14
C-label associated
with pantothenate and b-alanine and estimation of initial
exchange velocities was as previously described [6].
Nonhomologous functional links
In order to verify the prediction of COG1701 and
COG1829 as the missing steps in archaeal CoA biosynthesis
[10], the STRING database [16] was searched for nonho-
mologous functional links using the criteria ‘Neighborhood’
(conserved chromosomal proximity) and ‘Co-occurrence’
(conserved phylogenetic profile). To this end, the archaeal
members of the protein families COG0413, COG1893,
COG0452, COG1019, and COG0237, which are implied in
archaeal CoA biosynthesis through homology (Fig. 1), were
used as protein queries. Functional links to COG1701 or
COG1829 were classified as strong links (high or highest

confidence in STRING) or weak links (low or medium con-
fidence in STRING).
Acknowledgements
We would like to thank Katharina Pflu
¨
ger, Universita
¨
t
Mu
¨
nchen, for genomic DNA from Me. mazei, and
Ishac Nazi, McMaster University, for the E. coli
PANK expression plasmid pPANK. We also thank
Erich Glawischnig for critically reading this manu-
script. S. Ronconi was funded by graduate scholar-
ships from the German Academic Exchange Service
(DAAD) and Technische Universita
¨
tMu
¨
nchen
(Frauenbu
¨
ro). Work on pantothenate and CoA
biosynthesis in this laboratory was funded by the
Deutsche Forschungsgemeinschaft.
References
1 Kanehisa M, Goto S, Hattori M, Aoki-Kinoshita KF,
Itoh M, Kawashima S, Katayama T, Araki M & Hirak-
awa M (2006) From genomics to chemical genomics:

new developments in KEGG. Nucleic Acids Res 34,
D354–D357.
2 Miyatake K, Nakano Y & Kitaoka S (1979) Pantothe-
nate synthetase from Escherichia coli [D-pantoate: beta-
alanine ligase (AMP-forming), EC 6.3.2.1]. Methods
Enzymol 62, 215–219.
3 von Delft F, Lewendon A, Dhanaraj V, Blundell TL,
Abell C & Smith AG (2001) The crystal structure of
S. Ronconi et al. Pantothenate synthetase from Methanosarcina mazei
FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS 2763
E. coli pantothenate synthetase confirms it as a member
of the cytidylyltransferase superfamily. Structure 9,
439–450.
4 Zheng R & Blanchard JS (2001) Steady-state and
pre-steady-state kinetic analysis of Mycobacterium
tuberculosis pantothenate synthetase. Biochemistry 40,
12904–12912.
5 Wang S & Eisenberg D (2006) Crystal structure of the
pantothenate synthetase from Mycobacterium tuberculo-
sis, snapshots of the enzyme in action. Biochemistry 45,
1554–1561.
6 Jonczyk R & Genschel U (2006) Molecular adaptation
and allostery in plant pantothenate synthetases. J Biol
Chem 281, 37435–37446.
7 Jackowsky S (1996) Biosynthesis of pantothenate and
coenzyme A. In Escherichia coli and Salmonella
typhimurium: Cellular and Molecular Biology (Neidhardt
FC, ed), pp. 687–694. American Society for Microbiol-
ogy, Washington, DC.
8 Begley TP, Kinsland C & Strauss E (2001) The biosynthe-

sis of coenzyme A in bacteria. Vitam Horm 61, 157–171.
9 Leonardi R, Zhang YM, Rock CO & Jackowski S
(2005) Coenzyme A: back in action. Prog Lipid Res 44,
125–153.
10 Genschel U (2004) Coenzyme A biosynthesis: recon-
struction of the pathway in archaea and an evolutionary
scenario based on comparative genomics. Mol Biol Evol
21, 1242–1251.
11 Armengaud J, Fernandez B, Chaumont V,
Rollin-Genetet F, Finet S, Marchetti C, Myllykallio
H, Vidaud C, Pellequer JL, Gribaldo S et al. (2003)
Identification, purification, and characterization of a
eukaryotic-like phosphopantetheine adenylyltransferase
(coenzyme A biosynthetic pathway) in the hyper-
thermophilic archaeon Pyrococcus abyssi. J Biol Chem
278, 31078–31087.
12 Kupke T & Schwarz W (2006) 4¢-Phosphopantetheine
biosynthesis in Archaea. J Biol Chem 281, 5435–5444.
13 Brand LA & Strauss E (2005) Characterization of a
new pantothenate kinase isoform from Helicobacter
pylori. J Biol Chem 280, 20185–20188.
14 Overbeek R, Fonstein M, D’Souza M, Pusch GD &
Maltsev N (1999) The use of gene clusters to infer func-
tional coupling. Proc Natl Acad Sci USA 96, 2896–2901.
15 Tatusov RL, Koonin EV & Lipman DJ (1997) A
genomic perspective on protein families. Science 278,
631–637.
16 von Mering C, Jensen LJ, Kuhn M, Chaffron S,
Doerks T, Kru
¨

ger B, Snel B & Bork P (2007)
STRING 7 – recent developments in the integration
and prediction of protein interactions. Nucleic Acids
Res 35, D358–D362.
17 Bork P, Sander C & Valencia A (1993) Convergent
evolution of similar enzymatic function on different
protein folds: the hexokinase, ribokinase, and galacto-
kinase families of sugar kinases. Protein Sci
2, 31–40.
18 So
¨
ding J (2005) Protein homology detection by HMM–
HMM comparison. Bioinformatics 21, 951–960.
19 Pang SS, Duggleby RG & Guddat LW (2002) Crystal
structure of yeast acetohydroxyacid synthase: a target
for herbicidal inhibitors. J Mol Biol 317, 249–262.
20 Airas RK (1988) Pantothenases from pseudomonads
produce either pantoyl lactone or pantoic acid. Biochem
J 250, 447–451.
21 Alberty RA (1993) Levels of thermodynamic treatment
of biochemical reaction systems. Biophys J 65, 1243–
1254.
22 Segel IH (1975) Enzyme Kinetics, pp. 79–80. John Wiley
& Sons, New York, NY.
23 Koonin EV, Mushegian AR & Bork P (1996) Non-
orthologous gene displacement. Trends Genet 12, 334–
336.
24 Aravind L, Tatusov RL, Wolf YI, Walker DR & Koo-
nin EV (1998) Evidence for massive gene exchange
between archaeal and bacterial hyperthermophiles.

Trends Genet 14, 442–444.
25 Nelson KE, Clayton RA, Gill SR, Gwinn ML, Dodson
RJ, Haft DH, Hickey EK, Peterson JD, Nelson WC,
Ketchum KA et al. (1999) Evidence for lateral gene
transfer between Archaea and bacteria from
genome sequence of Thermotoga maritima. Nature 399,
323–329.
26 Cronan JE Jr, Littel KJ & Jackowski S (1982) Genetic
and biochemical analyses of pantothenate biosynthesis
in Escherichia coli and Salmonella typhimurium. J Bacte-
riol 149, 916–922.
27 King HL Jr, Dyar RE & Wilken DR (1974) Ketopan-
toyl lactone and ketopantoic acid reductases. Character-
ization of the reactions and purification of two forms of
ketopantoyl lactone reductase. J Biol Chem 249, 4689–
4695.
28 Nazi I, Koteva KP & Wright GD (2004) One-pot
chemoenzymatic preparation of coenzyme A analogues.
Anal Biochem 324, 100–105.
Pantothenate synthetase from Methanosarcina mazei S. Ronconi et al.
2764 FEBS Journal 275 (2008) 2754–2764 ª 2008 The Authors Journal compilation ª 2008 FEBS