Content uploaded by Rowena Green Matthews
Author content
All content in this area was uploaded by Rowena Green Matthews
Content may be subject to copyright.
Content uploaded by Rowena Green Matthews
Author content
All content in this area was uploaded by Rowena Green Matthews
Content may be subject to copyright.
letters
nature structural biology • advance online publication 1
Domain alternation
switches B
12
-dependent
methionine synthase to the
activation conformation
Vahe Bandarian, Katherine A. Pattridge,
Brett W. Lennon, Donald P. Huddler,
Rowena G. Matthews and Martha L. Ludwig
Biophysics Research Division and Department of Biological Chemistry,
University of Michigan, Ann Arbor, Michigan 48109-1055, USA.
Published online: 3 December 2001, DOI: 10.1038/nsb738
B
12
-dependent methionine synthase (MetH) from
Escherichia coli
is a large modular protein that uses bound
cobalamin as an intermediate methyl carrier. Major domain
rearrangements have been postulated to explain how
cobalamin reacts with three different substrates: homo-
cysteine, methyltetrahydrofolate and S-adenosylmethionine
(AdoMet). Here we describe the 3.0 Å structure of a 65 kDa
C-terminal fragment of MetH that spans the cobalamin- and
AdoMet-binding domains, arranged in a conformation suit-
able for the methyl transfer from AdoMet to cobalamin that
occurs during activation. In the conversion to the activation
conformation, a helical domain that capped the cofactor
moves 26 Å and rotates by 63°, allowing formation of a new
interface between cobalamin and the AdoMet-binding (acti-
vation) domain. Interactions with the MetH activation
domain drive the cobalamin away from its binding domain in
a way that requires dissociation of the axial cobalt ligand and,
thereby, provide a mechanism for control of the distribution
of enzyme conformations.
In the catalytic cycle, the cobalamin cofactor of B
12
-dependent
methionine synthase (MetH) is alternately methylated by
methyltetrahydrofolate (CH
3
H
4
folate) to form methylcobalamin
and demethylated by homocysteine (Hcy) to form cob(I)alamin
1
(Fig. 1). Occasional oxidation of the cob(I)alamin intermediate
produces an inactive cob(II)alamin enzyme, which is reactivated
by a reductive methylation that uses S-adenosylmethionine
(AdoMet) as the methyl donor
2
and flavodoxin or a flavodoxin-
like domain as an electron donor
3–5
. Thus, methionine synthase
supports three distinct methyl transfer reactions, each involving
the cobalamin cofactor and a substrate bound to its own func-
tional unit.
The modular organization of MetH has allowed dissection of
this large enzyme into four functional units
6–8
. The first two
modules bind homocysteine and methyltetrahydrofolate, the
third module incorporates the B
12
cofactor and the C-terminal
module binds AdoMet, which is essential for reactivation of the
cob(II)alamin form of the enzyme
7
. Structures of the cobalamin-
binding and AdoMet-binding modules have been deter-
mined
9,10
. In the cobalamin-binding fragment of MetH, the imi-
dazole side chain of His 759 replaces the dimethylbenzimidazole
(DMB) ligand of the cofactor. His 759 is linked by hydrogen
bonds to Asp 757 and Ser 810. These three residues constitute a
ligand triad that enhances the efficiency of methyl transfer dur-
ing primary turnover
11
and affects access of the three substrate-
bearing modules to the cobalamin
12
.
In the structure of the cobalamin-binding module, the upper
face of the cobalamin is shielded by a four-helix bundle domain,
the ‘cap’. Our evidence suggests that a significant fraction of the
intact enzyme exists in the ‘cap-on’ state in the absence of sub-
strates
13
. We have assumed that MetH adopts a series of distinct
conformations to allow substrates to be presented to cobal-
amin
14–17
. Formation of the species that supports methylation of
cobalamin by AdoMet is especially intriguing because it must be
controlled. There is strong discrimination against utilization of
the methyl group from AdoMet for conversion of homocysteine
to methionine
2,17
. In vivo, this selectivity avoids futile cycles that
would lead to ATP hydrolysis rather than net synthesis of
methionine.
Analyses of structures of methionine synthase trapped in
defined conformational states can provide detailed descriptions
of the conformations that support each of the methyl transfers,
along with clues to the features that control the equilibria
between the different states. Here we describe the structure of
the 65 kDa C-terminal fragment (residues 649–1227) spanning
the cobalamin- and AdoMet-binding domains. The two
domains are arranged in a conformation that brings the AdoMet
binding site close to cobalamin.
The activation reaction
Recent investigations of the in vitro reactivation of
cob(II)alamin MetH by flavodoxin hydroquinone
18
have estab-
lished that only a small fraction of the enzyme is initially in a
conformation suitable for rapid reduction by flavodoxin
(Fig. 1). For the remainder of the enzyme, the first and slowest
step in reactivation involves dissociation of the His 759 ligand,
which is detected by characteristic changes in the
cob(II)alamin spectrum. The unligated (‘base-off’) species
binds flavodoxin and is reduced, cob(I)alamin is then
methylated and, finally, methylcob(III)alamin is religated to
Fig. 1 Methionine synthesis and reactivation of methionine synthase. In
the reactivation cycle (bold), an electron donor and a methyl group from
AdoMet
2
convert the inactive cob(II)alamin form of the enzyme to
methylcob(III)alamin. In Escherichia coli, flavodoxin serves as the elec-
tron donor for this priming reaction
3
. Reductive activation can proceed
even at a high potential when driven by coupling with the favorable ∆G
of methyl transfer from AdoMet
4
. The first step in reductive reactivation
is exceedingly slow, proceeding at ∼0.4 s
–1
in the presence of reduced
flavodoxin, relative to 27 s
–1
for turnover in catalysis.
© 2001 Nature Publishing Group http://structbio.nature.com
© 2001 Nature Publishing Group http://structbio.nature.com
letters
2 nature structural biology • advance online publication
His 759. These kinetic analyses find that formation of the
cob(I)alamin enzyme faces a kinetic barrier at the step involv-
ing dissociation of the axial His 759 ligand to cobalt
18
We
believe that this slow step not only entails dissociation of His
759, as signaled by spectral changes, but is also associated with
a large protein rearrangement that generates a structure com-
petent to bind flavodoxin and catalyze methyl group transfer
from AdoMet to the reduced cobalamin.
The activation conformation
To obtain a structure displaying the interactions of cobalamin
with the activation domain, we exploited earlier findings that the
mutation H759G, which deletes the lower ligand to produce a
base-off species, favors the activation-competent conforma-
tion
12,19
. The C-terminal fragment of MetH that includes both
the cobalamin binding and the activation domains has recently
been expressed and purified
11
. The wild type fragment supports
methionine synthesis in trans when combined with the N-termi-
nal half of MetH
11
, whereas the mutant H759G fragment is inac-
tive in catalysis but active in reductive methylation by AdoMet
(data not shown). Thus, the H759G mutant fragment was an
attractive candidate for structure analysis of the activation con-
formation, and we were able to crystallize it in the cob(II)alamin
form.
The structure of the C-terminal fragment bearing the H759G
mutation was determined to 3.0 Å resolution by molecular
replacement (see Methods; Table 1). The helical cap, the
B
12
-binding α/β domain and the AdoMet-binding domain
constitute the C-terminal half of MetH (Fig. 2a). We refer to this
arrangement as the activation complex because the cobalamin
adjoins the AdoMet binding site. Comparison of this activation
complex with the structure of the cobalamin binding fragment
(Fig. 2b) reveals a dramatic movement of the four-helix bundle
domain, which caps the cofactor in the isolated fragment. In the
conversion to the activation complex, each of the three domains
of the C-terminal fragment behaves as a rigid unit with little
internal deformation. Residues that connect the B
12
and cap
domains undergo local conformation changes that allow the cap
to adopt its new position.
Cobalamin movements and interactions
The significant movement of the cobalamin, relative to its posi-
tion in the isolated B
12
-binding fragment, was unexpected.
Orthogonal views (Fig. 3a) show the changes in the positions of
the corrin ring and its side chains. From the DMB group
upward, the cofactor is displaced by alterations in torsion angles.
The corrin ring lifts, tilts ∼18° and slides (Fig. 3b,c). Four hydro-
gen bonds that connect the corrin side chain amides and the
residues of the His-bearing loop in the cap-on state are not
formed in the activation conformer (Fig. 3b). Seven new hydro-
gen bonds to the activation domain secure the displaced corrin
and help draw it away from its binding domain. These latter
interactions displace most of the hydrogen bonds to the B
12
domain that position the corrin in the cap-on conformation.
More important, severe steric clashes are relieved by displace-
ment of the corrin. For example, if the corrin were to remain in
the position it occupies when the cap is on, C12 in pyrrole ring C
would be ∼1.0 Å from the carbonyl oxygen of Gly 1174. The B
ring propionamide and acetamide would similarly overlap the
side chains of Ala 1170 and Met 1171 (Fig. 3c). Residues
1170–1174 act as a wedge that pries the corrin away from the
cobalamin-binding domain and the His 759 loop, which is not
significantly perturbed when the corrin is displaced. The combi-
nation of changes (Fig. 3) increases the distance between Cα of
residue 759 and cobalt by 2.3 Å. Thus, the observed displace-
ment of corrin would force the dissociation of the lower histi-
dine ligand from cobalt.
Binding of S-adenosylmethionine
The structure refined at 3.0 Å resolution (Fig. 4a; Table 1) was
obtained from crystals grown in the absence of added AdoMet,
and water peaks are observed in the expected substrate-binding
site adjoining Tyr 1139. Because cocrystallizations with AdoMet
were unsuccessful, we added the substrate to preformed crystals
to determine how AdoMet is bound in the activation complex.
After the addition of AdoMet, there is difference density in the
maps determined to 3.8 Å resolution (Fig. 4b); the loss of dif-
fraction induced by soaking limits the resolution that can be
attained. A model of AdoMet, bound in the same way as in the
a b
Fig. 2 A stereo ribbon drawing of the C-terminal (651–1227) fragment of methionine synthase. a, The four-helix bundle (yellow) that caps the
corrin ring in the resting state (b) has been displaced, and the bowl-like interior of the C-terminal activation domain (blue) encloses the corrin ring
and interacts with the B
12
domain (red). Cobalamin is shown in gold. The area buried in the interface between the activation domain and the B
12
domain is ∼1,400 Å
2
; cobalamin contacts contribute 600 Å
2
of this 1,400 Å
2
. The activation and B
12
domain partners are connected by 10 hydrogen
bonds (seven to cobalamin). In comparison, the area of the cap–B
12
domain interface in the structure of the isolated cobalamin-binding fragment
9
is
1,160 Å
2
, with six interdomain hydrogen bonds, two of which connect the cap to the cofactor. The formation of a crosslink
24
to Lys 959 indicates that
flavodoxin binds in the hollow at the front face of the structure. This panel and the remaining drawings of Figs 2, 3 were prepared using RIBBONS
25
.
b, Movement of the four-helix cap that accompanies association of the B
12
domain with the AdoMet (activation) domain. The position of the cap
domain in the isolated (649–896) fragment
9
is shown in silver, and its location in the activation conformation is shown in yellow. Parameters describ-
ing the cap motion, a displacement of 26.2 Å and a rotation of 62.7°, were determined by treating the cap domain as a rigid body. The domain
movements are analyzed in more detail on the website (www.umich.edu/
∼
biophys/faculty/ludwig). In the activation complex, the cap is held in place
primarily by its interactions with the activation domain. The relatively small interface between the cap and the B
12
domain suggests that the cap
domain might move to yet another position when catalytic units react with the B
12
cofactor.
© 2001 Nature Publishing Group http://structbio.nature.com
© 2001 Nature Publishing Group http://structbio.nature.com
letters
nature structural biology • advance online publication 3
Fig. 3 Displacement of the cobalamin and changes in its interactions
with the protein. a, Side views of the bound cobalamin from vantage
points separated by 90° about the vertical axis. The gray model repre-
sents the cofactor in the cobalamin binding fragment, where it is capped
by the four-helix bundle and His 759 is ligated to methylcobalamin. The
gold model represents the cofactor in the activation complex.
Translation of the corrin macrocycle to the right and its movement away
from the B
12
domain are evident in the right-hand panel. Close-up views
of corrin–protein interactions, comparing b, the cap-on B
12
-binding frag-
ment and c, the activation complex show changes in hydrogen-bonding
that occur on conversion to the activation complex. For (b, c), the van-
tage point is approximately the same as in (a). The reference for compar-
ison of the structures is the loop Gly 756-Asp-Val-His-Asp-Ile, which
includes the His 759 ligand to cobalt in wild type enzyme; this loop is
positioned identically in both views (with some side chains omitted).
Hydrogen bonds between corrin side chain amides and this loop occur at
residues 756, 758 and 760 in the cap-on state but are not formed in the
activation conformation. Corrin side chains move to generate new pat-
terns of hydrogen bonding in the activation complex. d, A stereo dia-
gram showing the critical overlaps that drive displacement of the corrin
ring. The cobalamin (silver) is positioned as in the structure of the cap-on
form; the cobalamin (gold) and the peptide Ala 1170-Met-Trp-Pro-Gly-
Ala-Ser are from the activation complex. In the activation conformation,
the corrin ring (gold) is wedged toward the viewer and to the right to
avoid clashes at Ala 1170, Met 1171 and Gly 1174. The cap-on and activa-
tion structures were aligned by matching atoms in the B
12
domains. The
pyrrole rings and selected carbon atoms of the corrin skeleton have been
labeled.
isolated activation domain
10
, can be fit to the difference density.
However, the methyl group is ∼6 Å from the cobalt, a distance
too long for transfer, and Tyr 1139 protrudes into the path
between the S-methyl group and cobalt. Crystal cracking and
loss of resolution in diffraction patterns is consistent with the
idea that productive binding of AdoMet induces additional
structural changes that we cannot observe in this crystal form.
We suspect that the Tyr 1139 ring must swing away for methyl
transfer to proceed but cannot be moved aside without major
rearrangement of the peptide backbone. Additional structural
changes may be associated with the reduction to cob(I)alamin
that precedes methyl transfer
18
.
Implications
The structure of the C-terminal fragment of methionine synthase
provides direct evidence for the major domain rearrangement
postulated to occur in the switch to the activation complex. In
contrast, the rather large movement of corrin, which also accom-
panies formation of the activation complex, was not anticipated.
Inspection of the structure suggests that the corrin displacement,
which must be accompanied by dissociation of the axial ligand, is
driven by steric hindrance and stabilized by attractive interac-
tions between the cofactor and the activation domain. The wedge
that forces dissociation of the imidazole ligand provides a molec-
ular mechanism for the switch between conformations. The same
displacements are expected to occur in the wild type enzyme.
Although we have not determined the relative positions of the
activation and B
12
domains in the cap-on conformation of the
C-terminal fragment, the known structures of the activation
complex and the B
12
fragment can be used to model the inter-
conversion of the cap-on and activation conformations. The
movement of the cap domain, with its large translation compo-
nent, cannot be characterized in the same way as other docu-
mented domain motions
20
but may provide a prototype for
situations in which tethered domains undergo large translations
with respect to one another (see www.umich.edu/
∼
biophys/
faculty/ludwig).
The intact enzyme must be distributed among several states,
including conformations in which the folate or Hcy binding
modules are positioned for methyl transfer to or from the cobal-
amin, as well as the activation-competent and cap-on states that
have been compared here. We expect these distributions to be
linked to the oxidation, as well as the ligation, state of the cobalt
and to be influenced by the presence of substrates or flavodoxin.
For instance, in the absence of flavodoxin, the activation confor-
mation is disfavored in the cob(II)alamin form of the holoen-
zyme by ∼1.8 kcal mol
–1
(refs 12,18). Binding of flavodoxin shifts
the equilibrium in favor of the base-off activation form. In con-
trast, the methylcobalamin form of MetH retains the spectral
signature of base-on cobalamin in the presence of flavodoxin
(K
d
>70 µM)
19
. Because too much energy is required to dissoci-
ate the axial ligand from methylcobalamin, the activation con-
formation is not populated in methylated MetH and flavodoxin
does not bind. Thus, cofactor chemistry and mobility both
prove to be determinants that affect the ensemble of conforma-
tions adopted by MetH. The present structure of the C-terminal
fragment affords new insights into how the oxidation state of the
cobalamin shifts the equilibrium between the activation and
other conformations and allows access to AdoMet only during
reductive activation of cob(II)alamin enzyme.
a
b c
d
© 2001 Nature Publishing Group http://structbio.nature.com
© 2001 Nature Publishing Group http://structbio.nature.com
letters
4 nature structural biology • advance online publication
Methods
Protein expression and purification. Expression and purification
of the C-terminal fragment of MetH have been described
11
. The
mutant H759G fragment is isolated as the cob(II)alamin form.
Crystallization and structure determination. Crystals were
grown at 22 °C in hanging drops from solutions initially 12 mg ml
–1
in protein and mixed with equal volumes of a reservoir containing
0.1 M cacodylate buffer, pH 6.5, 0.2 M ammonium sulfate and
30% (w/v) PEG 8000. The space group is P4
3
2
1
2, with one molecule
per asymmetric unit (Table 1). Intensities were measured on an
R-AXIS IV system (MSC) at 140 K after flash cooling crystals in a cryo-
solvent in which 17% (w/v) meso-erythritol had been added to the
reservoir solution. Data reduction was carried out with the HKL
suite
21
. A clear molecular replacement solution was found using the
activation domain as a search model
22
. Maps phased with this par-
tial model revealed cobalamin and part of the B
12
domain. The
remainder of the density emerged when recognizable elements
were added and subjected to rigid body refinement. The full atom-
ic model was refined in CNS
23
to R-factor = 0.218 and R
free
= 0.277 in
rounds of computation that included initial minimization, SA tor-
sional refinement, Powell minimization and restrained B-factor
adjustment. Restraints for the cobalamin cofactor were from
Drennan
9
. For studies of AdoMet binding, crystals were soaked
overnight in holding solution containing 2.7 mM AdoMet.
Coordinates. The coordinates and structure factors for the struc-
tures of the activation complex have been deposited in the Protein
Data Bank (accession codes 1K7Y for the fragment and 1K98 for the
AdoMet complex).
Acknowledgments
Supported by NIH grants to M.L.L. and to R.G.M., and NRSA postdoctoral
fellowships to V.B. and B.W.L. are acknowledged.
Correspondence should be addressed to M.L.L. email: mlludwig@umich.edu
Received 23 August, 2001; accepted 1 November, 2001.
1. Matthews, R.G. In Chemistry and biochemistry of B
12
(ed. Banerjee, R.) 681–706
(Wiley, New York; 1999).
2. Taylor, R.T. & Weissbach, H. Arch. Biochem. Biophys. 129, 728–744 (1969).
3. Fujii, K. & Huennekens, F.M. J. Biol. Chem. 249, 6745–6753 (1974).
4. Banerjee, R.V., Harder, S.R., Ragsdale, S.W. & Matthews, R.G. Biochemistry 29,
1129–1135 (l990).
5. LeClerc, D. et al. Proc. Nat. Acad. Sci. USA 95, 3059–3064 (1998).
6. Banerjee, R.V., Johnston, N.L., Sobeski, J.K., Datta, P. & Matthews, R.G. J. Biol.
Chem. 264, 13888–13895 (1989).
7. Drummond, J.T., Huang, S., Blumenthal, R.M. & Matthews, R.G. Biochemistry 32,
9290–9295 (1993).
8. Goulding, C.W., Postigo, D. & Matthews, R.G. Biochemistry 36, 8082–8091 (1997).
9. Drennan, C.L., Huang, S., Drummond, J.T., Matthews, R.G. & Ludwig, M.L.
Science 266, 1669–1674 (1994).
10. Dixon, M., Huang, S., Matthews, R.G. & Ludwig, M.L. Structure 4, 1263–1275
(1996).
11. Bandarian, V. & Matthews, R.G. Biochemistry 40, 5056–5064 (2001).
12. Jarrett, J.T. et al. Biochemistry 35, 2464–2475 (1996).
13. Jarrett, J.T. et al. Bioorg. Med. Chem. 4, 1237–1246 (1996).
14. Ludwig, M.L. & Matthews, R.G. Annu. Rev. Biochem. 66, 269–313 (1997).
15. Drennan, C.L. et al. In Vitamin B
12
and B
12
proteins (eds Krautler, B., Arigoni, D. &
Golding, B.T.) 133–156 (Wiley-VCH, Weinheim; 1998).
16. Matthews, R.G. In Enzymatic mechanisms (eds Frey, P.A. & Northrup, D.B.)
155–161 (IOS Press, Amsterdam; 1999).
17. Jarrett, J.T., Huang, S. & Matthews, R.G. Biochemistry 37, 5372–5382 (1998).
18. Jarrett, J.T., Hoover, D.M., Ludwig, M.L. & Matthews, R.G. Biochemistry 37,
12649–12658 (1998).
19. Hoover, D.M. et al. Biochemistry 36,127–138 (1997).
20. Hayward, S. Proteins Struct. Funct. Genet. 36, 425–435 (1999).
21. Otwinowski, Z. & Minor, W. Methods Enzymol. 276, 307–326 (1997).
22. Tong, L. & Rossmann, M.G. Methods Enzymol. 276, 594–610 (1997).
23. Brünger, A.T. et al. Acta Crystallogr. D 54, 905–921 (1998).
24. Hall, D.A., Jordan-Starck, T.C., Loo, R.O., Ludwig, M.L. & Matthews, R.G.
Biochemistry 39, 10711–10719 (2000).
25. Carson, M. Methods Enzymol. 277, 493–505 (1997).
26. McRee, D.E. J. Struct. Biol. 125, 156–165 (1999).
27. Merritt, E.A. & Bacon, D.J. Methods Enzymol. 277, 505–524 (1997).
a
b
Table 1 Crystal data and refinement statistics
Fragment (649–1227) Fragment + AdoMet
Space group P4
3
2
1
2P4
3
2
1
2
Unit cell (Å)
a = b 109.72 108.50
c 148.59 146.47
Completeness (%)
1
97.4 (95.4) 85.1 (62.8)
R
sym
1
0.080 (0.396) 0.062 (0.444)
Resolution (Å) 27.0–3.0 35.0–3.75
Measured reflections 80,208 34,361
Unique reflections 18,245 8,057
Refinement statistics
Resolution limits (Å) 15.0–3.0 15.0–3.75
R
cryst
0.218 0.308
R
free
0.277 0.363
Protein atoms 4,566 4,566
Heteroatoms
Cofactor 91 91
SO
4
50 5
Solvent 42 0
R.m.s. deviations
Bond lengths (Å) 0.013 0.020
Bond angles (˚) 1.49 1.97
1
Numbers in parentheses are for the outermost bins of data.
Fig. 4 Stereo views of electron density in the vicinity of the
cobalamin and the AdoMet binding site. a, The green contours
(at 1 σ) represent electron density from a 3.0 Å (2|F
o
| – |F
c
|) map
computed at the conclusion of refinement. Conventional atom
coloring has been used to display the protein and cobalamin;
cobalt is in cyan. The phosphate group of the cofactor can be
glimpsed at the bottom edge of the drawing. Helix residues at
the upper right are 1,091–1,097; the loop below the corrin
includes residues 806–808, and Tyr 1139 lies above the corrin.
This figure was prepared with XtalView
26
and rendered using
RASTER3D
27
. b, The electron density corresponding to bound
AdoMet. The green contours (at +3 σ) are from a 3.8 Å-differ-
ence Fourier map computed after rigid body and torsional
refinement of the protein and cofactor versus data from a crys-
tal soaked in AdoMet. The difference densities shown here are
the strongest connected positive features in the map. The
AdoMet model from the structure of the isolated activation
domain
10
has been positioned in the density without further
refinement of its conformation. Density adjoining AdoMet has
been attributed to a bound sulfate ion that interacts with both
AdoMet and corrin side chains.
© 2001 Nature Publishing Group http://structbio.nature.com
© 2001 Nature Publishing Group http://structbio.nature.com