High resolution structure and catalysis of O-acetylserine
sulfhydrylase isozyme B from Escherichia coli
Georg Zocher, Ulrich Wiesand and Georg E. Schulz
Institut fu
¨
r Organische Chemie und Biochemie, Albert-Ludwigs-Universita
¨
t, Freiburg im Breisgau, Germany
In bacteria, archaea and plants, the biosynthesis of
l-cysteine involves l-serine and inorganic sulfur com-
pounds [1–5]. In higher animals, however, l-cysteine
is derived from l-methionine [1]. The bacterial path-
way starts with a transferase that uses acetyl-CoA to
modify serine. The resulting O-acetylserine (OAS) is
then converted to cysteine by a sulfhydrylase (OASS,
EC 2.5.1.47), which in general uses hydrogen sulfide.
In a number of bacteria, the second step of synthesis
is performed by the two isozymes A and B, named
CysK and CysM, respectively. CysK uses mostly
hydrogen sulfide, which is produced in a reduction
pathway that begins with sulfate and requires dioxy-
gen. In contrast, CysM has a characteristic main
chain variation around position 210 that opens the
active center for larger thiol-carrying compounds, in
particular for thiosulfate [2,6]. The reaction with thio-
sulfate results in S-sulfo-cysteine, which can be easily
converted to cysteine and sulfate. Consequently, the
use of thiosulfate is of particular importance in an
anaerobic environment, because it does not require
dioxygen for the reduction of sulfate to hydrogen
sulfide. The isozyme CysM is of technical interest

˚
resolution by X-ray diffraction analysis.
The C1-carboxylate of citrate was bound at the carboxylate position of
O-acetylserine, whereas the C6-carboxylate adopted two conformations.
The activity of the enzyme and of several active center mutants was deter-
mined using an assay based on O-acetylserine and thio-nitrobenzoate
(TNB). The unnatural substrate TNB was modeled into the reported struc-
ture. The substrate model and the observed mutant activities may facilitate
future protein engineering attempts designed to broaden the substrate spec-
trum of the enzyme. A comparison of the reported structure with previ-
ously published CysM structures revealed large conformational changes.
One of the crystal forms contained two dimers, each of which comprised
one subunit in a closed and one in an open conformation. Although the
homodimer asymmetry was most probably caused by crystal packing, it
indicates that the enzyme can adopt such a state in solution, which may be
relevant for the catalytic reaction.
Abbreviations
CysK, O-acetylserine sulfhydrylase (EC 2.5.1.47) isozyme A; CysM, O-acetylserine sulfhydrylase (EC 2.5.1.47) isozyme B from Escherichia
coli; CysM(K268A), surface mutant K268A of CysM; CysM(RKE), triple surface mutant E57R-Y148K-R184E of CysM; CysM(salmo),
isozyme B from Salmonella typhimurium; DTNB, S,S¢-bis(5-thio-2-nitrobenzoate); TNB, thio-nitrobenzoate; OAS, O-acetylserine; OASS,
O-acetylserine sulfhydrylase; PLP, pyridoxal 5¢-phosphate.
5382 FEBS Journal 274 (2007) 5382–5389 ª 2007 The Authors Journal compilation ª 2007 FEBS
Five structures of CysK-type enzymes from bacteria
[10–14], archaea [15] and plants [16,17], and two struc-
tures of bacterial CysM [6,18], have been published.
The differences between the isozymes CysK and CysM
have been described [6,18]. In this article, we present
the structure of CysM complexed with the substrate
analog citrate at high resolution, together with enzy-
matic activity data of several mutants. Moreover, we

˚
resolution in crystal form III
(Table 1). Crystal form III was grown essentially under
the same conditions as form I, except for the absence
of ammonium sulfate. The surface mutation K268A
was at the rim of a packing contact and was not
required for crystallization, but was essential for the
superior packing order and for reproducible crystal
growth.
The structure of crystal form III was determined by
the molecular replacement method. In contrast with
the other crystal forms, form III contained only one
subunit per asymmetric unit and a lower solvent
content, both of which are typical prerequisites for
high resolution X-ray diffraction (Table 1). Although
crystal forms I and III were grown from the same
citrate buffer, only form III showed a citrate molecule
bound to the active center. Apparently, the high ionic
strength of ammonium sulfate prevented citrate
binding in form I. The structure of CysM in crystal
form III is shown in Fig. 1. Citrate was bound in
two conformations with occupancies of 60% and
40%, as revealed by the electron density depicted in
Fig. 2. Binding in multiple conformations indicates
low affinity, which, in turn, agrees with our observa-
tion that citrate does not inhibit the enzyme (see
below).
In order to identify established structures of related
enzymes, we searched the Protein Data Bank for
sequence homologs and detected 11 entries with

and c ¼ 209.8 A
˚
containing one CysM
subunit per asymmetric unit and 55% solvent.
Data collection
Resolution (A
˚
) 63–1.33 (1.37–1.33)
Unique reflections 83156 (6220)
Completeness (%) 98.6 (87.9)
Multiplicity 7.4 (7.9)
R
sym-I
(%) 5.9 (35)
Average I ⁄ r
I
21.4 (3.5)
Refinement
Number of atoms, protein
(residues 1–294)
2290
Number of atoms, glycerol ⁄ citrate 12 ⁄ 26
Number of water molecules 329
R
cryst
⁄ R
free
(2% test set) 0.158 ⁄ 0.172
Average isotropic B-factors (A
˚

to 25 mm citrate, but observed no change. Therefore,
citrate is not an inhibitor. This agrees with the two
observed citrate conformations, because multiple bind-
ing is usually weak.
The observed kinetic parameters of wild-type CysM
from E. coli are in general agreement with those of the
homolog CysM(salmo) [18,19]. Of the active center
mutants produced, the deletion of a methyl group near
Fig. 1. Stereo ribbon plot of the high resolu-
tion structure of the CysM dimer, including
the molecular twofold axis (black), which is
crystallographic. The position of the surface
mutation K268A is shown as a yellow
sphere 25 A
˚
away from the active center.
The cofactor PLP covalently linked to Lys41,
the bound citrate molecule in its major con-
formation and the mutated residues Thr68,
Gln140 and Arg210 in the active center are
depicted as ball-and-stick models. The sub-
units have different colors. The mobile loops
defined in Fig. 5 are labeled using gray
spheres. The active center pocket opening
is indicated by a yellow stick.
Fig. 2. Detailed stereoview of the active
center of CysM. The covalently bound PLP
and the associated citrate are shown in
orange. Citrate was bound with 100% occu-
pancy. The minor conformation of citrate is

(TNB)
(%)
Temperature
dependence
a
Wild-type 24 0.7 100
b
(41)
c
2.4
T68S 11 0.6 55 (26) 2.1
R210A – – 2 (0.8) 2.5
Q140A – – 0.4 (0.1) 4
T68A – – 0.1 (0.01) 10
Q140E – – Inactive –
a
The temperature dependence is defined here as k
cat
⁄ K
M
(TNB)
measured at 37 °C relative to the value measured at 25 °C.
b
The
absolute k
cat
⁄ K
M
(TNB) value at 37 °C was 3.5 · 10
4

decrease to merely 2% catalytic efficiency was
observed with mutant R210A. Even stronger decreases
were caused by the removal of a carboxamide in
mutant Q140A, and by the deletion of a hydroxyl
group in mutant T68A. The enzyme was inactive when
a carboxylate was introduced at position 140 (Q140E).
The moderate activity reduction of T68S and the
strong effects of mutations Q140A, T68A and Q140E
agree well with the data derived for the corresponding
mutants of the CysK-type enzyme from Arabidopsis
thaliana [16].
In a second series of experiments, we determined the
k
cat
⁄ K
M
(TNB) values at 25 °C. The results were similar
to those at 37 °C, except for a 2.3-fold decrease for the
wild-type and for mutants T68S and R210A (Table 2).
The 2.3-fold decrease relates well to the decrease in k
cat
expected from the ‘rule-of-thumb’ factor of two for a
10 K temperature drop [20], showing that the activa-
tion energy of the catalyzed reaction lies in the usual
range and does not change for T68S and R210A. In
contrast, mutants Q140A and T68A showed much
higher temperature dependence factors, corresponding
to an appreciable increase in the activation energy [20].
We conclude that Q140A and T68A, which are close
to PLP, directly affect the reaction. In contrast, the

In order to model the reaction geometry, we used
the established external aldimine structure of a related
CysK structure [11] and transferred it to CysM, where
it could be accommodated without steric collision
(Fig. 3). The expected reaction geometry at the exter-
nal aldimine intermediate [11] defines the thiol position
of TNB to a small region above the plane of the acry-
late double bond. As a result of this constraint and of
the spacious active center pocket of CysM, TNB was
placed rather easily. In our model, the carboxylate of
TNB is fixed by Arg210 and the nitro group points to
the solvent (Fig. 3). The thiolate is located above the
Fig. 3. Stereoview of the reaction geometry based on the structure of CysM(K268A). The observed internal aldimine with Lys41 is given in a
transparent mode (gray). The external aldimine structure has been transferred from a CysK-type enzyme [11]. It is shown together with a
manually placed model of the bound substrate TNB, the carboxylate of which is fastened to Arg210. The thiolate of TNB is approximately at
the same position as the attacking sulfur of thiosulfate in a previous model [6], which is well suited for the nucleophilic attack (red dotted
line) on the amino acrylate double bond (green spheres). Hydrogen bonds are given as black broken lines. All van der Waals distances
between TNB and its environment are above 3.0 A
˚
. The two shortest contacts are marked by green dotted lines.
G. Zocher et al. Structure of the O-acetylserine sulfhydrylase CysM
FEBS Journal 274 (2007) 5382–5389 ª 2007 The Authors Journal compilation ª 2007 FEBS 5385
acrylate plane forming an S–Cb–Ca angle of about
90°. This is an ideal position for attacking the double
bond. In summary, our negative experience with muta-
tions close to PLP suggests that this region should not
be touched when trying to produce novel l-amino
acids [7–9]. Rather, such engineering attempts should
follow the TNB model, which suggests residues
Met119, Phe141, Thr175, Pro207 and Arg210 as the

B-factor distributions that report the polypeptide chain
mobility. As the B-factor level is strongly dependent
on the quality of the crystal order, the B-factor distri-
butions have been normalized by referring them to the
average B-factors of the respective chains. They are
displayed in Fig. 5. The distribution of CysM(K268A)
shows nine characteristic mobility peaks. Of these, the
loops at peak positions 94, 116, 132, 202 and 215 form
the lips of the mouth of the active center pocket
(Fig. 1) and are therefore important for catalysis. The
other peaks correspond to loops at the surface that are
usually mobile (Fig. 1). Interestingly, loop 69 is close
to PLP and not mobile (Fig. 5), although it partici-
pates in the induced fit (Fig. 4).
The mobility distributions of CysM(K268A), wild-
type CysM and CysM(salmo), and those of subunits B
and D of CysM(RKE), resemble each other closely
(Fig. 5). However, a most surprising deviation of the
B-factor distribution occurs in subunits A and C of
CysM(RKE) [6]. The CysM(RKE) crystal contains
dimers A–B and C–D, providing four independent sub-
unit structures. Dimer A–B is asymmetric with respect
to mobility and also with respect to structure. The
B-factor distribution of subunit A is exceptional, as it
shows almost no mobility peak. In contrast, the
respective distribution of subunit B shows the common
mobility peaks, including those of the active center lips
(Fig. 5). The same asymmetry is observed with sub-
units C and D of the other dimer. As the three muta-
tions of CysM(RKE) are all at the surface distant

Mutagenesis and activity assay
The mutants were produced with the QuikChange method
(Stratagene, Heidelberg, Germany), verified by DNA
sequencing (SeqLab, Go
¨
ttingen and GATC, Konstanz,
Germany) and expressed and purified as described previ-
ously [6]. They were stored at )20 °Cina12mgÆmL
)1
solution containing 10 mm Tris ⁄ HCl pH 8.0. For the
assay, we incubated 950 lL of buffer A (100 mm Hepes
pH 7.0, 10 mm OAS, 10–1000 lm TNB) at 37 °C (or
25 °C) for 3 min, and started the reaction by adding
50 lL of a solution containing 0.5–80 lg of the enzyme.
The enzyme solution was always freshly prepared from
stored protein so that the exposure time to 37 °C (or
25 °C) was minimized. This was important for the low
activity mutants at positions 68 and 140 near PLP. TNB
was always freshly prepared in 50 mm Hepes pH 7.0
by adding 2 mm dithiothreitol and 0.5 mm S,S¢-bis(5-thio-
2-nitrobenzoate) (DTNB) to yield 1 mm TNB. The
absorption of TNB was monitored at 412 nm using e
412
¼
13 600 m
)1
Æcm
)1
[19], as well as at 500 nm using e
500

yleneglycol) 3000]. Crystals of CysM(K268A) grew within
about 10 days at 20 °C to sizes of up to 1000 lm ·
400 lm · 400 lm. The crystals were transferred in four
steps to 28% (v ⁄ v) glycerol in reservoir buffer and flash-
frozen in a 100 K nitrogen gas stream.
Fig. 5. Relative B-factor distributions of CysM subunits in four dif-
ferent crystal forms. The B-factors were referred to the respective
subunit averages in order to eliminate differences arising from crys-
tal packing quality variations. All distributions were smoothed by
sliding a three-residue-averaging window along the chain. The top
diagram K268A refers to the reported high resolution structure with
labels at nine high mobility peaks (see Figs 1 and 4). Distribution
WT is an average of the four closely related subunit chains of the
wild-type structure [6]. The distribution of CysM(salmo) is from sub-
unit A, which is virtually the same as those of the other seven sub-
units [18]. The two distributions at the bottom are from dimers
A–B and C–D of CysM(RKE) [6] which, however, were split into an
average of the closely related subunits B and D and the equally
well-related subunits A and C.
G. Zocher et al. Structure of the O-acetylserine sulfhydrylase CysM
FEBS Journal 274 (2007) 5382–5389 ª 2007 The Authors Journal compilation ª 2007 FEBS 5387
The X-ray data were collected at the Swiss Light Source
(Villigen, Switzerland) (Table 1) and processed with pro-
grams xds and xscale [22]. Using phaser [23] and the
wild-type CysM structure [6], the phases were established
by molecular replacement. To avoid model bias, the CysM
structure and the water structure were completely rebuilt
using arp ⁄ warp [24]. The structure was manually com-
pleted using coot [25] and then refined with refmac5 [26].
Finally, we performed a translation libration screw refine-

4 Mino K & Ishikawa K (2003) Characterization of a
novel thermostable O-acetylserine sulfhydrylase from
Aeropyrum pernix K1. J Bacteriol 185, 2277–2284.
5 Wirtz M, Droux M & Hell R (2004) O-acetylserine
(thiol) lyase: an enigmatic enzyme of plant cysteine bio-
synthesis revisited in Arabidopsis thaliana. J Exp Bot 55,
1785–1798.
6 Claus MT, Zocher GE, Maier THP & Schulz GE (2005)
Structure of the O-acetylserine sulfhydrylase isoenzyme
CysM from Escherichia coli. Biochemistry 44, 8620–
8626.
7 Maier THP (2003) Semisynthetic production of unnatu-
ral 1-a-amino acids by metabolic engineering of the cys-
teine biosynthetic pathway. Nat Biotechnol 21, 422–427.
8 Watkins KJ (2001) Peptides: a boom in the making.
Chem Eng News January 8, 11–15.
9 Kaldor SW, Kalish VJ, Davies IIJF, Shetty BV,
Fritz JE, Appelt K, Burgess JA, Campanale KM, Chir-
gadze NY, Clawson DK et al. (1997) Viracept (Nelfina-
vir Mesylate, AG1343): a potent, orally bioavailable
inhibitor of HIV-1 protease. J Med Chem 40, 3979–3985.
10 Burkhard P, Rao GSJ, Hohenester E, Schnackerz KD,
Cook PF & Jansonius JN (1998) Three-dimensional
structure of O-acetylserine sulfhydrylase from Salmo-
nella typhimurium. J Mol Biol 283, 121–133.
11 Burkhard P, Tai CH, Ristroph CM, Cook PF & Janso-
nius JN (1999) Ligand binding induces a large confor-
mational change in O-acetylserine sulfhydrylase from
Salmonella typhimurium. J Mol Biol 291 , 941–953.
12 Burkhard P, Tai CH, Jansonius JN & Cook PF (2000)

18 Chattopadhyay A, Meier M, Ivaninskii S & Burkhard P
(2007) Structure, mechanism, and conformational
dynamics of O-acetylserine sulfhydrylase from Salmo-
nella typhimurium: comparison of A and B isozymes.
Biochemistry 46, 8315–8330.
19 Tai CH, Nalabolu SR, Jacobson TM, Minter DE &
Cook PF (1993) Kinetic mechanisms of the A and B
isozymes of O-acetylserine sulfhydrylase from
Structure of the O-acetylserine sulfhydrylase CysM G. Zocher et al.
5388 FEBS Journal 274 (2007) 5382–5389 ª 2007 The Authors Journal compilation ª 2007 FEBS
Salmonella typhimurium LT-2 using the natural and
alternative reactants. Biochemistry 32, 6433–6442.
20 Price NC & Stevens L (1984) Fundamentals of Enzymol-
ogy, p. 136. Oxford University Press, Oxford.
21 Brzovic PS, Choi WE, Borchardt D, Kaarsholm NC &
Dunn MF (1994) Structural asymmetry and half-site
reactivity in the T to R allosteric transition of the insu-
lin hexamer. Biochemistry 33, 13 057–13 069.
22 Kabsch W (1993) Automatic processing of rotation dif-
fraction data from crystals of initially unknown symme-
try and cell constants. J Appl Crystallogr 26, 795–800.
23 Storoni LC, McCoy AJ & Read RJ (2004) Likelihood-
enhanced fast rotation functions. Acta Crystallogr D60,
432–438.
24 Perrakis A, Morris RM & Lamzin VS (1999) Auto-
mated protein model building combined with iterative
structure refinement. Nat Struct Biol 6, 458–463.
25 Emsley P & Cowtan K (2004) COOT: model-building
tools for molecular graphics. Acta Crystallogr D60,
2126–2132.