Mechanistic aspects and redox properties of
hyperthermophilic
L-proline dehydrogenase from
Pyrococcus furiosus related to dimethylglycine
dehydrogenase⁄ oxidase
Phillip J. Monaghan, David Leys and Nigel S. Scrutton
Manchester Interdisciplinary Biocentre and Faculty of Life Sciences, University of Manchester, UK
The membrane-associated flavoprotein PutA in enteric
bacteria is a proline catabolic enzyme that catalyzes
the oxidation of proline to glutamate in a two-step
reaction to form glutamate (Fig. 1). The protein is also
a transcriptional repressor of the proline utilization
(put) genes [1–3]. Cytoplasmic PutA represses tran-
scription from its own gene and also from the
Na
+
⁄ proline transporter PutP [4–6]. Proline catabolism
enables enteric bacteria to use l-proline as a source of
carbon, nitrogen and electrons, and the reaction is
initiated in the FAD-binding domain by two-electron
oxidation of l-proline to form D
1
-pyrroline-5-carboxy-
late (P5C) [5,6]. Following oxidation of l-proline, the
two-electron reduced FAD cofactor passes electrons to
an acceptor in the electron transfer chain. The interme-
diate P5C is hydrolyzed to glutamate 5-semialdehyde,
which is then oxidized to glutamate by the P5C
dehydrogenase domain, with NAD
+
acting as electron

m
as a function
of pH was observed. The pH dependence of k
cat
is sigmoidal, described by
a single macroscopic pK
a
of 7.7 ± 0.1, tentatively attributed to ionization
of l-proline in the Michaelis complex. The preliminary crystal structure of
the enzyme revealed active site residues conserved in related amine dehy-
drogenases and potentially implicated in catalysis. Studies with H225A,
H225Q and Y251F mutants ruled out participation of these residues in a
carbanion-type mechanism. The midpoint potential of enzyme-bound FAD
has a linear temperature dependence () 3.1 ± 0.05 mVÆC°
)1
), and extra-
polation to physiologic growth temperature for P. furiosus (100 °C) yields
a value of ) 407 ± 5 mV for the two-electron reduction of enzyme-bound
FAD. These studies provide the first detailed account of the kinetic ⁄ redox
properties of this hyperthermophilic l -proline dehydrogenase. Implications
for its mechanism of action are discussed.
Abbreviations
DMGO, dimethylglycine oxidase; P5C, D
1
-pyrroline-5-carboxylate; PRODH, L-proline dehydrogenase; TAPSO, 3-{[tris(hydroxymethyl)-
methyl]amino}-2-hydroxypropane sulfonic acid; TMADH, trimethylamine dehydrogenase.
2070 FEBS Journal 274 (2007) 2070–2087 ª 2007 The Authors Journal compilation ª 2007 FEBS
acceptor (Fig. 1). The structure of a truncated form of
PutA comprising the FAD-containing proline dehy-
drogenase domain has been elucidated, and reveals a

gi_18977617 (a subunit) and gi_18977618 (b subunit),
in the protein extraction, description and analysis tool
(pedant) database] in the genome of P. furiosus
DSM 3638 that show sequence homology to the a and
b subunits of the flavoprotein amine oxidoreductase,
tetrameric sarcosine oxidase [13]. The translated amino
acid sequence of gi_18977618 (b subunit) also aligns
with another member of the amine oxidoreductase
family, dimethylglycine oxidase (DMGO) from
Arthrobacter globiformis [13], for which the crystal
structure has been determined to 1.6 A
˚
resolution [14].
This alignment indicates the conservation of three resi-
dues (His225, Tyr251 and Gly262 in the gi_18977618
translation), known to reside in the active site of
DMGO, that have been implicated in the catalytic
mechanism of dimethylglycine oxidation [14]. The crys-
tal structure of a related enzyme from Pyrococcus hori-
koshii OT-3 has been elucidated [15], and its basic
solution properties have been analyzed [16], but, to
date, detailed mechanistic studies of the activity of this
or related enzymes have not been reported. With this
aim in mind, we present an analysis of recombinant
protein expressed from the two ORFs found in
P. furiosus DSM 3638 that share sequence similarity
with the gene encoding DMGO. We show that these
genes encode a new member of the hyperthermo-
philic class of PRODHs that couples the oxidation of
l-proline to the reduction of a noncovalently bound

where it forms the fourth b-strand of the Rossman
fold and the connecting loop. In flavoproteins, the
ATG motif has a defined function, in that it is always
present at the junction with the substrate-binding
domain, and not within a domain, as in NADPH-
binding proteins [18]. The b subunit also contains an
ADP-binding motif, and shares 27% sequence identity
with the b subunit of tetrameric sarcosine oxidase from
Corynebacterium sp. P-1, and 26% sequence identity
with the N-terminal half of human dimethylglycine
dehydrogenase. Of particular note is the finding that
the b subunit of PRODH shows sequence conservation
with active site residues His225, Tyr259 and Gly270 of
DMGO from A. globiformis and mouse lung dimethyl-
glycine dehydrogenase (Fig. 2), residues that are pre-
sent in a number of sarcosine dehydrogenase-like
proteins. Given these sequence similarities, we conjec-
tured that the protein encoded by the two identified
A
B
Fig. 2. Multiple sequence alignments of amino acid sequences of the a and b subunits of PRODH. (A) Multiple sequence alignment for the
a subunit, showing 18% sequence identity with the N-terminal region of the a subunit of tetrameric sarcosine oxidase from both Corynebac-
terium sp. P-1 [20] and Arthrobacter sp. 1-IN [13]. The 11 residues that comprise the ADP-binding motif are highlighted in bold, and shaded
where residues are conserved. All 11 residues in PRODHa obey the physicochemical requirements established by Wierenga et al. [17]. The
conserved GG doublet and ATG motif are also shaded. (B) Multiple sequence alignment deduced for the b subunit of PRODH, showing 24%
sequence identity with DMGO from A. globiformis [13] and 26% sequence identity with the cDNA translation product of dimethylglycine
dehydrogenase from M. musculus lung tissue [21]. The N-terminal ADP-binding motif is highlighted in bold, and shaded where residues are
conserved. Again, all 11 residues satisfy the consensus sequence, with the exception of the glutamate residue at position 1, although this
hydrophilic residue has the correct physicochemical requirements for this position. DMGO active site residues His225, Tyr259 and Gly270,
identified from the crystal structure, align with conserved residues in both dimethylglycine dehydrogenase and PRODHb, and are highlighted

tion (approximately $ 13%) of the a subunit was trun-
cated. The sequence MKVQRQ was obtained for the
truncated a subunit N-terminus, indicating that trunca-
tion in the a subunit is located 83 amino acids from
the initiating methionine in the full-length a subunit.
Enzyme expressed from plasmid pPRODH2 lacked a
truncated a subunit, consistent with removal of the
internal ribosome-binding site by mutagenesis. The a
and b subunits were coexpressed in a molar ratio of
$ 1 : 1 from pPRODH2, as judged by SDS⁄ PAGE
peak area image scanning. Analysis of purified enzyme
by MALDI-TOF MS gave a molecular mass of
42 437.5 Da for the b subunit, comparable to the pre-
dicted molecular mass of 42 481.2 Da from the gene
sequence. Electrospray mass data for the b subunit
gave a molecular mass of 42 474.0 Da. Mass data for
the larger a subunit could not be obtained using the
MALDI-TOF or electrospray methods. Purification of
H225A, H225Q and Y251F mutant enzymes was as
described for wild-type PRODH. Mutagenesis of these
active site residues had no effect on recombinant pro-
tein yield, and all mutants were purified in FAD-
bound form. Despite the close proximity of both
His225 and Tyr251 to the isoalloxazine ring moiety of
FAD, no major perturbations in the absorption prop-
erties of the enzyme-bound flavin were evident as a
consequence of mutagenesis.
Holoenzyme cofactor content
Our preliminary crystallographic analysis of the
enzyme indicates a heterooctomer (ab)

the preliminary X-ray crystal structure of PRODH
that one molecule of ATP cofactor is bound in the
ADP-binding motif of the a subunit (Fig. 4A). This
cofactor has no obvious function from a mechanistic
perspective, but may play a stabilizing role under the
harsh physiologic conditions that P. furiosus is subject
to. The ADP-binding motif in the b subunit binds one
molecule of noncovalent FAD (Fig. 4B). FMN is
located at the interface of the a and b subunits
(Fig. 4A).
Reductive titration of PRODH highlighted a single
isosbestic point at 340 nm, with no evidence for a
semiquinone species obtained during titration with
sodium dithionite (Fig. 3B). FAD-bound PRODH for
mechanistic studies was confirmed by MALDI-TOF
MS after heat treatment to remove cofactor. A
mass ⁄ charge peak at 787 corresponded to the posi-
tively charged quasimolecular ion ([M +H]
+
)of
FAD. Partial hydrolysis of FAD during heat treatment
was revealed by mass ⁄ charge peaks at 348 and 458,
assigned to AMP and FMN ([M +H]
+
) hydrolysis
products, respectively. ATP cofactor was not detected
(Fig. 5), although the preliminary crystal structure of
PRODH indicates its presence in the enzyme.
Reduction with amine substrates
Alignment of the a and b subunit sequences with well-

identified substrates for PRODH is that they are
all secondary a-amino acids (Fig. 6). The ability of
P. furiosus PRODH to oxidize multiple amine com-
pounds is in stark contrast to the catalytic properties
reported for dye-linked proline dehydrogenase 1 of
Pyrococcus horikoshii OT-3, which has been shown to
act exclusively on l-proline, with l-pipecolic acid and
sarcosine being inert as substrates [16]. A spectral fea-
ture is apparent at $ 550 nm upon kinetic reduction
of PRODH with each of the three identified amine
substrates. This signal may represent a minor transient
population of a charge-transfer species during the cat-
alytic reaction. Addition of sodium sulfite (50 mm)to
purified enzyme did not perturb the flavin absorption
spectrum, indicating that a flavin–N5–sulfite adduct
does not form. This suggests that the enzyme is not a
flavoprotein oxidase, as reactivity with sulfite is a
characteristic of this class of flavoenzyme [22].
Steady-state turnover analysis with
L-proline
and
L-pipecolic acid
In developing a suitable and continuous turnover
assay for wild-type enzyme at an elevated temperature
H225β
Y251β
FAD
FAD
FMN
ATP

determined under aerobic conditions. Analysis of hyper-
bolic plots of initial velocity as a function of substrate
concentration yielded apparent K
m
values for the wild-
type enzyme of 30.8 ± 1.1 mm and 212.3 ± 17.0 mm
for l-proline and l-pipecolic acid, respectively. The cor-
responding apparent k
cat
values were 18.1 ± 0.2 s
)1
and
0.4 ± 0.02 s
)1
for l-proline and l-pipecolic acid,
respectively, and the calculated specificity constants
(k
cat
⁄ K
m
) were 0.59 ± 0.03 s
)1
Æmm
)1
(l-proline) and
0.002 ± 0.0002 s
)1
mm
)1
(l-pipecolic acid). We infer

adduct m ⁄ z peaks at 348 and 370, respectively. (D) Released cofactor of heat-denatured PRODH showing
the [M +H]
+
ion of FAD cofactor with an m ⁄ z peak of 787 identical to that of the authentic FAD standard. The m ⁄ z peak at 825 repre-
sents the K
+
adduct of released FAD cofactor. The m ⁄ z peaks of 348 and 458 represent the [M +H]
+
ion of both AMP and FMN,
respectively, which result from partial hydrolysis of FAD cofactor during protein heat treatment. The m ⁄ z peak of 496 represents the K
+
adduct of the FAD cofactor heat hydrolysis product, FMN. K
+
ions are present from the purification buffers. Conditions: samples of
authentic FAD, FMN and AMP were prepared as 1 mgÆmL
)1
stock solutions in double deionized H
2
O, and filtered using a 0.22 lm Acro-
disc; PRODH was exchanged into double deionized H
2
O.
P. J. Monaghan et al.
L-proline dehydrogenase from P. furiosus
FEBS Journal 274 (2007) 2070–2087 ª 2007 The Authors Journal compilation ª 2007 FEBS 2075
anaerobic assays. Additionally, we were unable to
show electron transfer from l-proline-reduced PRODH
to P. furiosus ferredoxin under anaerobic turnover
conditions, either in the presence or in the absence of
exogenous FMN.

¼ 83.4 ± 2.9 kJÆmol
)1
, DS
à
¼ 27.2 ± 1.0 JÆmol
)1
Æ
K
)1
,andDG
à
¼ 73.3 kJÆmol
)1
(at 373 K). Incubation of
PRODH at elevated temperatures prior to activity assay
showed that the enzyme is extremely stable, with no loss
of activity being evident up to 100 °C. Above this tem-
perature, thermal denaturation of PRODH is apparent,
with complete loss of activity after 10 min of incubation
in glycerol buffer at temperatures ‡ 115 °C (data not
shown). Thus, PRODH from P. furiosus is the most ther-
mostable PRODH described to date.
A
B
C
D
E
Fig. 6. Absorption changes as a function of
reaction time accompanying reduction of
wild-type PRODH with sarcosine,

temperature.
Temperature (°C) K
m
(mM) k
cat
(s
)1
) k
cat
⁄ K
m
(s
)1
ÆmM
)1
)
40 5.6 ± 0.4 0.4 ± 0.005 0.08 ± 0.006
45 5.6 ± 0.2 0.8 ± 0.005 0.1 ± 0.005
50 5.3 ± 0.4 1.3 ± 0.02 0.2 ± 0.02
55 8.0 ± 0.4 2.5 ± 0.02 0.3 ± 0.02
60 9.9 ± 0.4 4.0 ± 0.04 0.4 ± 0.02
65 11.6 ± 1.0 5.9 ± 0.1 0.5 ± 0.05
70 19.9 ± 2.3 10.4 ± 0.4 0.5 ± 0.08
75 25.9 ± 1.3 14.8 ± 0.2 0.6 ± 0.04
80 30.8 ± 1.1 18.01 ± 0.2 0.6 ± 0.03
L-proline dehydrogenase from P. furiosus P. J. Monaghan et al.
2076 FEBS Journal 274 (2007) 2070–2087 ª 2007 The Authors Journal compilation ª 2007 FEBS
For mechanistic analyses, steady-state turnover
assays were performed aerobically with l-proline at
60 °C over the pH range 5.5–10.0, to identify

H225Q mutant forms (see below). The pH depend-
ence of k
cat
exhibited a simple sigmoid behavior that,
when analyzed by fitting to Eqn (3) (see Experimen-
tal procedures) (Fig. 8B), produced a macroscopic
pK
a
value of 7.7 ± 0.1. The pK
a
value for the
protonation of free proline is 10.6, but this might be
lowered on binding to enzyme in the Michaelis
complex by ) 2.9 pH units. A precedent for stabiliza-
tion of the free base form of amine substrates at
physiologic pH values is available from studies with
trimethylamine dehydrogenase (TMADH) [25], and is
consistent with mechanistic proposals that require
the unprotonated amine substrate species to react
with the enzyme-bound flavin [26].
B
A
Fig. 7. Temperature dependence and Eyring analysis of initial velo-
city data for wild-type PRODH reacting with
L-proline. (A) Three-
dimensional plot showing initial velocity (y-axis) versus time (x-axis)
versus temperature (z-axis). Reactions were performed in the pres-
ence of saturating
L-proline (200 mM) over the temperature range
40–90 °C. The dimension of time demonstrates any potential loss

)1
at 100 °C.
Table 2. Steady-state kinetic parameters determined for the reac-
tion of PRODH with
L-proline at different pH values and at constant
ionic strength. Assays were performed at 60 °C.
pH k
cat
(s
)1
) K
m
(mM) k
cat
⁄ K
m
(s
)1
ÆmM
)1
)
5.5 1.10 ± 0.06 9.96 ± 2.20 0.11 ± 0.03
6.0 1.67 ± 0.03 5.76 ± 0.51 0.29 ± 0.03
6.5 2.40 ± 0.02 1.08 ± 0.08 2.22 ± 0.18
7.0 4.08 ± 0.04 1.27 ± 0.11 3.21 ± 0.31
7.5 6.83 ± 0.14 1.81 ± 0.32 3.76 ± 0.74
8.0 11.68 ± 0.25 5.82 ± 0.65 2.01 ± 0.27
8.5 16.36 ± 0.46 19.54 ± 1.73 0.84 ± 0.10
9.0 16.36 ± 0.25 42.35 ± 1.56 0.39 ± 0.02
9.5 16.54 ± 0.41 87.21 ± 4.19 0.19 ± 0.01

max
1 þ
K
m
½S
þ
½S
K
i
ð2Þ
where v is the initial velocity, V
max
is the maximum
value of the initial velocity, [S] is the substrate concen-
tration, K
m
is the substrate concentration at half the
maximal velocity, and K
i
is the equilibrium constant
for inhibitor binding. Marked inhibition has also been
reported in studies of mutant forms of the flavoprotein
morphinone reductase under conditions of high sub-
strate concentration [27,28]. pH dependence studies
revealed that the H225A mutant enzyme was unstable
and precipitated from solution below pH 7.0. This
consequently compromised the accuracy of data analy-
sis in the acid solution pH region. The H225Q mutant
was somewhat more stable, displaying activity down to
solution pH 6.0. The pH dependence of k

pH for the PRODH-catalyzed oxidation of
L-proline. (A) pH depen-
dence of k
cat
⁄ K
m
following ionizations in the free enzyme and sub-
strate. Fitting of data to Eqn (4) gave two pK
a
values of 7.0 ± 0.2
and 7.6 ± 0.2. (B) pH dependence of k
cat
following the pK
a
of the
enzyme–substrate complex. Fitting of data to Eqn (3) showed a
simple sigmoid relationship, giving a pK
a
of 7.7 ± 0.1. This value is
tentatively assigned to deprotonation of the substrate
L-proline.
Conditions: three-component buffer system comprising 0.052,
0.052 and 0.1
M Mes, TAPSO, and diethanolamine, respectively;
60 °C.
Table 3. Steady-state kinetic parameters determined for the H225A,
H225Q and Y251F mutant PRODH forms. Conditions: buffer
composed of Mes, TAPSO and diethanolamine at final concentra-
tions of 0.052, 0.052 and 0.1
M, respectively, pH 7.5, at an assay

(4) (see Experimental procedures) gave macroscopic
pK
a
values of 6.8 ± 0.1 and 9.9 ± 0.1 (H225A),
6.8 ± 0.1 and 9.4 ± 0.2 (H225Q), and 6.0 ± 0.1 and
7.4 ± 0.1 (Y251F) (supplementary Fig. S2). The initial
idea that the pK
a
of 7.0 ± 0.1 determined for the
wild-type enzyme might represent ionization of the
conserved His225 active site residue in the free enzyme
has been rejected, as mutant pH dependence data
reveal that this ionization is not lost, but is apparent
from the acid limb of the bell-shaped fits in the
k
cat
⁄ K
m
plots for both H225A and H225Q mutant
forms. This analysis has revealed that His225 and
Tyr251 are not active site base residues, and are not
essential for catalysis.
The results obtained suggest that PRODH stabilizes
the deprotonated form of l-proline substrate in the
Michaelis complex, analogously to the substrate activa-
tion mechanisms observed in TMADH [25] and mono-
meric sarcosine oxidase [29]. Given this finding and the
absence of an active site residue that acts as base dur-
ing oxidation of l -proline, the data suggest that
PRODH-catalyzed amine oxidation may occur by

of P5C (supplementary Fig. S3). Additionally, we also
analyzed the product of o-aminobenzaldehyde reaction
with P5C using electrospray MS. In this case, a single
peak with a mass ⁄ charge ratio of 217 was observed,
corresponding to the positive ion of the P5C–o-amino-
benzaldehyde condensation product (Fig. 9).
Reduction potential of the enzyme-bound FAD
at physiologic temperature
The midpoint potential (E
m
) of FAD–PRODH was
determined by potentiometric redox titration with
sodium dithionite at ambient temperature and pH 7.0.
During the course of reductive titration, the oxidized
flavin was reduced directly to the dihydroflavin form,
without a visible population of a flavin semiquinone
species, indicating that the potential of the oxidized ⁄
semiquinone flavin couple is much lower than that of
the semiquinone ⁄ hydroquinone couple (Fig. 10A).
Data were fitted to the two-electron Nernst function
(Eqn 5) (see Experimental procedures) by least-squares
regression analysis, and gave a midpoint two-electron
potential value of ) 192 ± 3 mV and a corresponding
unrestricted RT/nF value of 28.9 ± 0.4 mV (where R
is gas constant, T is temperature, n is number of elec-
trons and F is Faraday constant), consistent with the
expected value (29.5 mV) for two-electron reduction of
the enzyme-bound FAD (Fig. 10B). The temperature
dependence of the two-electron midpoint potential was
measured within the range 7.5–31 °C (the limits

Experimental procedures) by least-squares regression
analysis, and gave midpoint potential values of
) 169 ± 3 mV (H225A), ) 155 ± 3 mV (H225Q),
and ) 157 ± 3 mV (Y251F) (supplementary Fig. S4),
P. J. Monaghan et al. L-proline dehydrogenase from P. furiosus
FEBS Journal 274 (2007) 2070–2087 ª 2007 The Authors Journal compilation ª 2007 FEBS 2079
which compares with ) 174 ± 3 mV for the wild-type
enzyme at the same temperature (25 °C).
Discussion
The mechanism of substrate oxidation by flavoprotein
amine dehydrogenase ⁄ oxidases remains contentious.
In recent years, however, substrate oxidation by
members of this enzyme class has been shown to
occur by quantum mechanical tunneling [30,32]. We
and others have demonstrated that analysis of the
temperature dependence of kinetic isotope effects sug-
gests that H-transfer by quantum tunneling occurs
during substrate C–H bond cleavage, and that the
temperature effects are consistent with contemporary
environmentally coupled tunneling models [33]. A
major restriction in the use of variable temperature to
study tunneling reactions is the limited temperature
range available in experimental studies. For this rea-
son, in this study we have targeted a hyperthermo-
philic amine dehydrogenase, with a view to extending
in future work the temperature window available for
detailed physicochemical analysis of the enzyme
chemistry. Thus, we have cloned and expressed
two ORFs from P. furiosus that show some sequence
similarity with bacterial amine-specific flavoprotein

kinetic parameters. The enzyme is not an oxidase, and
does not use nicotinamide cofactors as electron accep-
tors. It also does not show any sequence homology
with the PRODH (PutA protein) of E. coli [7]. The
enzyme therefore represents a new member of a
recently identified class of PRODHs of hyperthermo-
philic origin.
We have shown that the PRODH active site is
situated in the b subunit and that these residues, as
suggested from structural and kinetic studies of
A. globiformis DMGO [14], play a role in the mechan-
ism of amine substrate oxidation but do not partici-
pate as catalytic base residues. The presence of iron
coordination in the Cys-clustered domain of the a sub-
unit could not be confirmed from the structure
(Fig. 4A). The anaerobic environment that P. furiosus
inhabits led to the possibility that any Fe–S cluster in
the PRODH a subunit may be aerobically labile, so
wild-type enzyme was expressed and purified under
strict anaerobic conditions (see Experimental pro-
cedures), and anaerobic assays were repeated with
P. furiosus ferredoxin and NADP
+
to measure elec-
tron transfer to putative physiologic acceptors via the
Fe–S cluster. No evidence of an Fe–S cluster was
apparent from the UV-visible absorption spectrum of
anaerobic PRODH, and no activity towards ferredoxin
and ⁄ or NADP
+

0.5
0.6
0.7
0.8
0.9
Abs (450 nm)
(Potential versus NHE) (mV)
A
B
C
Fig. 10. Redox potentiometric titration of wild-type monoflavinyl-
ated PRODH with sodium dithionite, and temperature dependence
of the midpoint potential. (A) Spectral changes accompanying
reductive titration of PRODH with sodium dithionite at 25 °C. (B)
Plot of absorbance at 450 nm versus the observed potential (cor-
rected against the normal hydrogen electrode). The data are fitted
to the Nernst equation (Eqn 5) for a two-electron reduction pro-
cess, giving a midpoint reduction potential, E
m
,of) 192 ± 3 mV,
and an RTF value of 28.9 ± 0.4 mV. (C) Plot of E
m
versus tempera-
ture, illustrating the linear dependence in the range 7.5–31 °C; gra-
dient, ) 3.1 ± 0.05 mVÆC°
)1
). The plot extrapolates to an operational
midpoint potential at physiologic temperature (100 °C for P. furio-
sus)of) 407 ± 5 mV. Conditions: 100 m
M potassium phosphate

to $ pH 6.5) for substrate ionization on formation of
the enzyme–substrate complex [25]. Our data for
P. furiosus PRODH suggest that a similar shift in pK
a
of ) 2.9 pH units (from 10.6 to 7.7) occurs on binding
of l-proline to the enzyme, although unequivocal
demonstration of this must await detailed stopped-flow
studies with protiated and deuterated substrates. A sig-
nificant change in the pK
a
value for the ionization of
the alkaline limb of the k
cat
⁄ K
m
plot is seen following
mutagenesis (Fig. 8 and supplementary Fig. S2). At
this stage, we are unable to tentatively assign this ion-
ization to a functional group, but its location is most
likely in the active site. Mutagenesis could affect the
pK
a
of this neighboring group substantially. At this
stage, we cannot unequivocally rule out changes in the
rate-limiting step, or contributions from other ionizable
groups, as the origin of the observed dependence of k
cat
on solution pH. Mutagenesis studies of TMADH have
indicated a role for residues His172 and Tyr60 in per-
turbing the pK

tion description and analysis tool (pedant) database] were
identified in the genome of P. furiousus DSM 3638 that
encode a putative flavoenzyme amine dehydrogenase ⁄ oxid-
ase. These genes were amplified by PCR using the oligonu-
cleotides 5¢-GTG AGA AAC TTG AGG CCA CTA GAC
TTA ACG G-3¢ and 5¢-TCA ACC CAT TTG AAG AGC
AAC AGT TCT TAA TTC TCC C-3¢. The PCR product
was purified by agarose gel electrophoresis, and used as
template DNA in a second PCR reaction to incorporate
flanking restriction sites (5¢ XbaI and 3¢ BamHI) for direc-
tional cloning, and a 5¢-ribosome-binding site to allow
expression of the cloned DNA. Primers used during this
PCR reaction were: 5¢-GGG GGG TCT AGA AAG GAG
ATA AAG AGA TGA GAA ACT TGA G-3¢,and5¢-GGG
GGG GGA TCC TCA ACC CAT TTG A AG AGC A-3 ¢.
The PCR product was digested with XbaI and BamHI,
ligated with vector pET11d, previously made end-compat-
ible with the same restriction endonucleases, and trans-
formed into Novablue cells (Novagen, Merck Chemicals
Ltd., Nottingham, UK). Restriction analysis confirmed pos-
itive recombinants, and DNA sequencing confirmed the
correct sequence of the recombinant clone, designated
pPRODH1.
A silent mutation was incorporated into pPRODH1 to
eradicate an internal ribosome-binding site responsible for
translation initiation within the gene, forming a truncated
population of the a subunit in E. coli. The following prim-
ers were used in the mutagenesis protocol marketed by
Stratagene (Amsterdam, the Netherlands) (QuikChange):
5¢-GGT GTC GAT GCT AGG AAA ACA AA

previously [19]. Mutant forms were purified using the purifi-
cation protocol described for the wild-type enzyme. The pro-
tein concentration was determined by the method of
Bradford [35].
Anaerobic expression and partial purification
of wild-type PRODH
Anaerobic 2xYT media containing 20 mm sodium fumarate
was supplemented with ampicillin (50 lg ÆmL
)1
) and chlo-
ramphenicol (34 lgÆmL
)1
), and inoculated with E. coli
Rosetta(DE3)pLysS cells transformed with plasmid
pPRODH2. Cells were grown at 37 °C until the attenuance
at 600 nm reached $ 0.8. Cells were induced with isopropyl
thio-b-d-galactoside at a final concentration of 1 mm, and
grown for a further 48 h at 37 °C. Cells were harvested
under anaerobic conditions, resuspended in 50 mm anaero-
bic Mops buffer (pH 7.9), and sealed in an airtight centri-
fuge tube. Cells were lysed anaerobically by three cycles of
freeze–thaw treatment, and cell debris was removed by
centrifugation (12 100 g for 50 min using a Beckman
Avanti J-25 centrifuge, rotor type JA 25.50) following
DNA hydrolysis. The supernatant was transferred to an
airtight tube inside a glovebox, and the solution was subjec-
ted to a heat denaturation step at 80 °C for 1 h, with dena-
tured protein being removed by centrifugation (27 200 g for
30 min using a Beckman Avanti J-25 centrifuge, rotor type
JA 25.50). The supernatant was loaded onto an anaerobic

in a Belle Technology (Portesham, UK) glovebox under a
positive pressure atmosphere of nitrogen, with residual oxy-
gen levels being maintained at < 0.05 p.p.m. with a BASF
(Cheadle, UK) R3-11 oxygen-scavenging catalyst.
Identification of flavin cofactor in PRODH
Following release from the enzyme, the chemical identity of
the flavin cofactor in PRODH was determined by MALDI-
TOF MS. Cofactor was released from the enzyme by heat
denaturation, and precipitated protein was removed by cen-
trifugation (17 000 g for 10 min at 4 °C in the dark using a
Microcentrifuge Fisherbrand accuSpin Micro 17R). Cofac-
tor was mass analyzed in double deionized H
2
O alongside
authentic FAD, FMN, ATP, ADP and AMP treated in the
same way. Analysis was performed using a Bruker Biflex
mass spectrometer Bruker Daltonics Ltd., Coventry, UK,
calibrated with a combination of peptides of known mass,
and the matrix was dihydroxybenzoic acid.
Optical titrations with reducing substrates
and steady-state kinetic analysis
PRODH (800 lL; 19.4 lm in 100 mm potassium phosphate
buffer, pH 7.5) was incubated in a quartz cuvette at 80 °C
for 10 min to allow temperature equilibration. Anaerobic
stocks (2 m) of dimethylglycine, sarcosine, glycine, glycine
betaine, l-proline, d-proline, l-pipecolic acid and sodium
sulfite were mixed with enzyme to a final concentration of
20 mm. Enzyme reduction was monitored by time-depend-
ent spectral acquisition in the region 300–600 nm.
Steady-state kinetic measurements with identified sub-

state data, enzyme activity was measured in a continuous
assay using the ferricenium ion as an artificial electron
acceptor. Substrate concentrations were maintained at 10
times K
m
. Reaction temperatures were recorded by direct
measurement inside the cuvette. The temperature was
recorded during temperature equilibration prior to the
reaction at both the top and the bottom of the cuvette to
detect any temperature gradient in the assay mixture.
Temperatures were also recorded immediately after assay
completion, and reactions were repeated if temperature
fluctuations exceeded 0.1 C°. Assays were initiated by
addition of microliter volumes of enzyme to ensure that
there was no significant effect on overall reaction tempera-
ture. Initial velocity data were plotted as a complete data-
set of rate versus time versus temperature, analogously to
the studies performed on thermophilic enzymes by Peter-
son et al. [38], who described an equilibrium model to
determine the thermal parameter T
eq
that represents a
submillisecond timescale-reversible temperature-dependent
equilibrium between active enzyme and inactive (or less
active) forms. The effect of a decrease in enzyme activity
above the optimal temperature occurring due to a shift in
T
eq
is up to two orders of magnitude greater than the
contribution of thermal denaturation. It is important to

relevant pK
a
values.
k
cat
¼
EH Â 10
ðÀpHÞ
þ E Â 10
ðÀpK
a
Þ
10
ðÀpHÞ
þ 10
ðÀpK
a
Þ
ð3Þ
k
cat
K
m
¼
T
max
1 þ 10
ðpK
a1
ÀpHÞ

at ) 20 ° C. The extinction coefficient used for quantifying
the complex was e
443
¼ 2.71 mm
)1
Æcm
)1
. Baseline controls
were performed with each assay component in the absence
of enzyme. The o-aminobenzaldehyde–P5C chromophore
was analyzed using electrospray MS. Ten-microliter samples
were injected via a Rheodyne valve into a mobile phase of
methanol flowing at 0.2 mLÆmin
)1
into the electrospray
source. The source temperature was maintained at 80 °C,
and the needle voltage was $ 3.0 kV. The sample cone was
operated at 20 V, and nitrogen was used as the desolvation
and sheath gas at 600 and 100 LÆh
)1
, respectively. The spec-
trometer was calibrated with a solution of sodium iodide.
Direct analysis of enzyme reaction product was also per-
formed using MALDI-TOF MS. Following enzyme turn-
over, a 1 lL sample of reaction mixture was plated for
analysis on a Bruker Biflex mass spectrometer in positive
ion mode, using dihydroxybenzoic acid as matrix. A
1mgÆmL
)1
standard solution of authentic l-proline was

obs
¼
ða þ b10
ðE
12
ÀEÞ=29:5
Þ
1 þ 10
ðE
12
ÀEÞ=29:5
ð5Þ
In Eqn (5), A
obs
is the absorbance value at the peak for oxi-
dized flavin at the electrode potential E, and a and b are the
absorbance values of the fully oxidized and reduced enzyme
at this wavelength, respectively. E
12
is the midpoint potential
for the concerted two-electron reduction of the flavin. Data
manipulation and analysis were performed using origin
software package version 6.0 (Microcal, OriginLab Corpora-
tion, Northampton, MA, USA). All redox potentials are
given relative to the standard hydrogen electrode.
Structure determination
Crystals were obtained as described previously [19].
Molecular replacement using the PRODH from Pyrococcus
horikoshii OT-3 as a model (75% and 90% identical to
P. furiosus PRODH a and b subunits, respectively) was per-

Mr Nahid Hasan and Professor W. Hagen of Delft
University, The Netherlands for supplying a sample of
P. furiosus 4Fe)4S ferredoxin. P. J. Monaghan thanks
the BBSRC for the award of a studentship. D. Leys
is a Royal Society University Research Fellow and
an EMBO Young Investigator. This work was funded
by the UK Biotechnology and Biological Sciences
Research Council and The Royal Society.
References
1 Brown ED & Wood JM (1992) Redesigned purification
yields a fully functional PutA protein dimer from
Escherichia coli. J Biol Chem 267, 13086–13092.
2 Menzel R & Roth J (1981) Purification of the putA
gene product. A bifunctional membrane-bound protein
from Salmonella typhimurium responsible for the two-
step oxidation of proline to glutamate. J Biol Chem 256,
9755–9761.
3 Scarpulla RC & Soffer RL (1978) Membrane-bound
proline dehydrogenase from Escherichia coli. Solubiliza-
tion, purification, and characterization. J Biol Chem
253, 5997–6001.
4 Maloy SR & Roth JR (1983) Regulation of proline uti-
lization in Salmonella typhimurium : characterization of
put::Mu d(Ap, lac) operon fusions. J Bacteriol 154,
561–568.
5 Menzel R & Roth J (1981) Regulation of genes for pro-
line utilization in Salmonella typhimurium: autogenous
repression by the putA gene product. J Mol Biol 148,
21–44.
6 Wood JM (1981) Genetics of l-proline utilization in

Chem 277, 12861–12867.
13 Meskys R, Harris RJ, Casaite V, Basran J & Scrutton
NS (2001) Organization of the genes involved in
dimethylglycine and sarcosine degradation in Arthrobac-
ter spp.: implications for glycine betaine catabolism. Eur
J Biochem 268, 3390–3398.
14 Leys D, Basran J & Scrutton NS (2003) Channelling
and formation of ‘active’ formaldehyde in dimethyl-
glycine oxidase. EMBO J 22, 4038–4048.
15 Tsuge H, Kawakami R, Sakuraba H, Ago H, Miyano
M, Aki K, Katunuma N & Ohshima T (2005) Crystal
structure of a novel FAD-, FMN-, and ATP-containing
l-proline dehydrogenase complex from Pyrococcus hori-
koshii. J Biol Chem 280, 31045–31049.
16 Kawakami R, Sakuraba H, Tsuge H, Goda S,
Katunuma N & Ohshima T (2005) A second novel
dye-linked l-proline dehydrogenase complex is present
in the hyperthermophilic archaeon Pyrococcus horikoshii
OT-3. FEBS J 272, 4044–4054.
17 Wierenga R, Terpstra P & Hol W (1986) Prediction of
the occurrence of the ADP-binding aaa-fold in proteins,
using an amino acid sequence fingerprint. J Mol Biol
187, 101–107.
18 Vallon O (2000) New sequence motifs in flavoproteins:
evidence for common ancestry and tools to predict
structure. Proteins: Structure, Function Genet 38 ,
95–114.
19 Monaghan PJ, Leys D & Scrutton NS (2005) Crystalli-
zation and preliminary X-ray diffraction analysis of a
flavoenzyme amine dehydrogenase ⁄ oxidase from

Scrutton NS (2005) Reaction of morphinone reductase
with 2-cyclohexen-1-one and 1-nitrocyclohexene: proton
donation, ligand binding, and the role of residues histi-
dine 186 and asparagine 189. J Biol Chem 280, 10695–
10709.
28 Messiha HL, Bruce NC, Sattelle BM, Sutcliffe MJ,
Munro AW & Scrutton NS (2005) Role of active site
residues and solvent in proton transfer and the modula-
tion of flavin reduction potential in bacterial morphi-
none reductase. J Biol Chem 280, 27103–27110.
29 Zhao G & Jorns MS (2005) Ionization of zwitterionic
amine substrates bound to monomeric sarcosine oxi-
dase. Biochemistry 44, 16866–16874.
30 Basran J, Sutcliffe MJ & Scrutton NS (2001) Deuterium
isotope effects during carbon–hydrogen bond cleavage
by trimethylamine dehydrogenase. Implications for
mechanism and vibrationally assisted hydrogen tunnel-
ing in wild-type and mutant enzymes. J Biol Chem 276,
24581–24587.
31 Wood JM (1987) Membrane association of proline
dehydrogenase in Escherichia coli is redox dependent.
Proc Natl Acad Sci USA 84, 373–377.
32 Harris RJ, Meskys R, Sutcliffe MJ & Scrutton NS
(2000) Kinetic studies of the mechanism of C–H
bond breakage by the heterotetrameric sarcosine oxid-
ase of Arthrobacter sp-1-IN. Biochemistry 39, 1189–
1198.
33 Sutcliffe MJ, Masgrau L, Roujeinikova A, Johannissen
LO, Hothi P, Basran J, Ranaghan KE, Mulholland AJ,
Leys D & Scrutton NS (2006) Hydrogen tunnelling in

Enzymol 87, 405–426.
40 Dutton PL (1978) Redox potentiometry: determination
of midpoint potentials of oxidation–reduction compo-
nents of biological electron-transfer systems. Methods
Enzymol 54, 411–435.
41 Daff SN, Chapman SK, Turner KL, Holt RA,
Govindaraj S, Poulos TL & Munro AW (1997) Redox
control of the catalytic cycle of flavocytochrome P-450
BM3. Biochemistry 36, 13816–13823.
42 Munro AW, Noble MA, Robledo L, Daff SN &
Chapman SK (2001) Determination of the redox
properties of human NADPH-cytochrome P450
reductase. Biochemistry 40, 1956–1963.
43 Navaza J (1994) AMoRe: an automated package for
molecular replacement. Acta Crystallogr A50, 157–163.
44 Murshudov GN, Vagin AA & Dodson EJ (1997)
Refinement of macromolecular structures by the maxi-
mum-likelihood method. Acta Crystallogr D53, 240–255.
45 Cleveland WS (1993) Visualising Data. Hobart Press,
Summit, NJ.
46 Farrant RD, Walker V, Mills GA, Mellor JM & Lang-
ley GJ (2001) Pyridoxal phosphate de-activation by
pyrroline-5-carboxylic acid. Increased risk of vitamin B6
deficiency and seizures in hyperprolinemia type II.
J Biol Chem 276, 15107–15116.
Supplementary material
The following supplementary material is available
online:
Fig. S1. Plot of initial velocity versus l -proline concen-
tration for the Y251F PRODH enzyme.