The solution structure of reduced dimeric copper zinc superoxide dismutase The structural effects of dimerization Lucia Banci, Ivano Bertini, Fiorenza Cramaro, Rebecca Del Conte and Mari
Trang 1The solution structure of reduced dimeric copper zinc superoxide dismutase
The structural effects of dimerization
Lucia Banci, Ivano Bertini, Fiorenza Cramaro, Rebecca Del Conte and Maria Silvia Viezzoli
Department of Chemistry and Centro Risonanze Magnetiche, University of Florence, Italy
The solution structure of homodimeric Cu2Zn2superoxide
dismutase (SOD) of 306 aminoacids was determined on a
13C,15N and 70%2H labeled sample Two-thousand
eight-hundred and five meaningful NOEs were used, of which 96
intersubunit, and 115 dihedral angles provided a family of 30
conformers with an rmsd from the average of 0.78 ± 0.11
and 1.15 ± 0.09 A˚ for the backbone and heavy atoms,
respectively When the rmsd is calculated for each subunit,
the values drop to 0.65 ± 0.09 and 1.08 ± 0.11 A˚ for the
backbone and heavy atoms, respectively
The two subunits are identical on the NMR time scale, at
variance with the X-ray structures that show structural
ferences between the two subunits as well as between
dif-ferent molecules in the unit cell The elements of secondary
structure, i.e eight b sheets, are the same as in the X-ray
structures and are well defined The odd loops (I, III and V)
are well resolved as well as loop II located at the subunit
interface On the contrary, loops IV and VI show some
disorder The residues of the active cavity are well defined
whereas within the various subunits of the X-ray structure
some are disordered or display different orientation in
dif-ferent X-ray structure determinations The copper(I) ion and its ligands are well defined This structure thus represents a well defined model in solution relevant for structure–func-tion analysis of the protein The comparison between the solution structure of monomeric mutants and the present structure shows that the subunit–subunit interactions in-crease the order in loop II This has the consequences of inducing the structural and dynamic properties that are optimal for the enzymatic function of the wild-type enzyme The regions 37–43 and 89–95, constituting loops III and V and the initial part of the b barrel and showing several mutations in familial amyotrophis lateral sclerosis (FALS)-related proteins have a quite extensive network of H-bonds that may account for their low mobility Finally, the con-formation of the key Arg143 residue is compared to that in the other dimeric and monomeric structures as well as in the recently reported structure of the CCS–superoxide dismu-tase (SOD) complex
Keywords: superoxide dismutase; solution structure; dimeric protein; NMR; FALS
Cu2Zn2SOD is a well known homodimeric enzyme of
32 000 Da that catalyzes the dismutation of the superoxide
radical to hydrogen peroxide and oxygen through a two step
reaction [1–5]:
Cu2þþ Oÿ2 ! Cuþþ O2
Cuþþ Oÿ2 ! Cu2þÿO2ÿ2
!2Hþ Cu2þþ H2O2
The active site of each subunit contains both a zinc and a copper ion, the latter being the site of the reaction Copper occurs in the oxidized and in the reduced state, both of which are necessary for the function The X-ray structure of the oxidized form has been available since 1982 for the bovine enzyme [6,7] and several other structures have become available [8–18] Reduced state structures are also available although the picture is less clear-cut around the copper-binding site [19–21] Certainties on the protona-tion of His63, which bridges Cu and Zn in the oxidized form but is protonated in the reduced form, come from1H NMR studies [22–25] Eventually, monomeric forms were obtained through site-specific mutagenesis and the NMR solution structure [26,27] as well as the crystal structure [28]
of the reduced form were reported Also the backbone mobility of the monomeric state was investigated and compared with that of the dimeric species and it was concluded that, as far as motions in the ps to ns timescale are concerned, the region consisting of residues 131–142, which forms one side of the active site channel, is less mobile
in the monomeric mutant than in the dimeric wild-type protein; structural fluctuations in this region have been suggested to play a role in assisting the superoxide anion in sliding towards the active site [29,30] Moreover, the regions consisting of residues 47–59, 76–86 and 151–153, which are
Correspondence to I Bertini, Department of Chemistry and Centro
Risonanze Magnetiche, University of Florence, Via Luigi Sacconi 6,
50019 Sesto Fiorentino, Italy.
Fax: + 39 055 4574271, Tel.: + 39 055 4574272,
E-mail: bertini@cerm.unifi.it
Abbreviations: SOD, superoxide dismutase; Q133M2SOD, F50E/
G51E/E133Q monomeric mutant superoxide dismutase; FALS,
familial amyotrophis lateral sclerosis; M4SOD, F50E/G51E/V148K/
I151K monomeric mutant superoxide dismutase; CCS, yeast copper
chaperone for superoxide dismutase; TPPI, time proportional phase
increments.
Note: The PDB ID code for the solution structure of homodimeric
Cu 2 Zn 2 superoxide dismutase is 1L3N.
(Received 5 October 2001, revised 4 February 2002, accepted 16
February 2002)
Trang 2located at the subunit–subunit interface, were found to be
more rigid in the dimer This behavior was rationalized by
considering that the presence of the second subunit
produces residue–residue interactions, thus reducing their
motions [30] Also on the ms to ls time range, the subunit–
subunit interface displays increased mobility in the
mono-meric state with respect to the dimono-meric one In particular,
conformational equilibria were observed for residues
around Cys57 and Cys146 The former residue forms an
H-bond with the guanidinium group of Arg143 [31], which
is located in the active site channel pointing towards the
copper ion and whose side-chain orientation is optimized
for correctly orienting the incoming superoxide anion for
the electron transfer process The equilibrium between
multiple conformations for this group and a different
average structural orientation does not allow Arg143 to
assume the optimal orientation for the enzymatic reaction
[26,27,30] This could account for the reduced enzymatic
rates of the artificial monomeric species
The X-ray structures of the dimeric wild-type form of this
protein show different structural details in the active site,
both between the two subunits of the same molecule and
among crystallographically independent subunits This
holds also for a number of loops [32] The question of
why SOD is a dimer and whether there is a cooperativity or
anticooperativity between the two subunits in the
physio-logical picture has never been completely solved
In this context we decided to solve the solution structure
of the reduced dimeric protein (by using a classical NMR
approach) in order to compare the solution structures of the
monomeric and dimeric species as well as the solution and
the crystal structures The aim is to classify the effects of
dimerization on the structural details
M A T E R I A L S A N D M E T H O D S
Sample preparation
Dimeric human SOD was expressed in Escherichia coli
TOPP1 strain (Stratagene) The 15N and 15N, 13C, 2H
labeled proteins were obtained by growing the cells in
minimal medium (M9) as previously reported [30] The
samples were isolated and purified according to previously
published protocols [33] The triple labeled dimeric SOD
contained about 70%2H Reduction of the copper ion was
achieved by addition of sodium isoascorbate to a final
concentration of about 4–6 mM, in 20 mMphosphate buffer
at pH 5.0 under anaerobic conditions The NMR samples
had a concentration of about 2 mMin dimeric protein and
contained 10% D2O for the lock signal
NMR experiments
The NMR experiments were recorded on Bruker Avance
800, 700 and 600 spectrometers operating at 18.7, 16.4 and
14.1 T, respectively
The assignment of the backbone is already available [30]
For the assignment of side chains H(C)CH-TOCSY, at
600 MHz, and (H)CCH-TOCSY, at 800 MHz [34], were
performed, using 1024 (1H)· 112 (13C)· 256 (1H) data
points and spectral windows of 9258 Hz (1H)· 11 184 Hz
(13C)· 9258 Hz (1H) and 1024 (1H)· 128 (13C)· 280
(13C) data points with spectral windows of 12 019 Hz
(1H)· 16 667 Hz (13C)· 16 667 Hz (13C), respectively A
15N-NOESY-HSQC and a 13C-NOESY-HSQC [35] were collected at 800 MHz to obtain dipolar connectivities; a HNHA [36], at 700 MHz, and HNHB [37], at 800 MHz, experiments, were performed to determine the 3JHNHa coupling constants and additional constraints for the v1 torsion angles as well as stereospecific assignments for the
Hb protons The15N-NOESY-HSQC was recorded with spectral windows of 9569 Hz (1H)· 2989 Hz (15N)·
9569 Hz (1H) for 2048 (1H)· 88 (15N)· 296 (1H) data points The 13C-NOESY-HSQC was acquired with 1024 (1H)· 112 (13C)· 256 (1H) data points with spectral windows of 9615 Hz (1H)· 19 230 Hz (13C)· 9615 Hz (1H) For both experiments the mixing time was 130 ms The three-dimensional HNHA experiment was carried out using spectral windows of 9124 Hz (1H)· 9124 Hz (1H)· 3125 Hz (15N) for 1024 (1H)· 128 (1H)· 32 (15N) data points, the three-dimensional HNHB and the two-dimensional reference experiments were carried out using spectral windows of 11 160 Hz (1H)· 3244 Hz (15N)·
11 160 Hz (1H) for 1024 (1H)· 48 (15N)· 128 (1H) and
11 160 Hz (1H)· 3244 Hz (15N) for 1024 (1H)· 256 (15N) data points, respectively
These experiments were collected at 296 K and they were performed using pulsed field gradients along the z-axis Watergate two-dimensional NOESY experiments [38] at
296 K and at 286 K were registered at 800 MHz to identify connectivities involving histidines of the active site In both experiments 2048 (1H)· 1024 (1H) data points were acquired with spectral windows 9124 Hz (1H)· 9124 Hz (1H); mixing times of 130 and 60 ms for the experiments acquired at 296 and 286 K, respectively, were used
In order to detect the amide hydrogen–deuterium exchange a series of1H-15N HSQC spectra on the sample prepared dissolving the lyophilized protein in D2O solution were collected at 800 MHz, at 296 K 1024 (1H)· 256 (15N) data points were acquired with spectral windows 12 019 Hz (1H)· 4065 Hz (15N) for each spectrum Each 1H-15N HSQC spectrum was acquired in 20 min every hour over a 24-h period After 4 days from the dissolution in D2O one experiment was acquired to detect the remaining amide protons
Quadrature detection in the indirect dimensions was performed and water suppression was achieved through the WATERGATE sequence [39]
Data were processed with the standard Bruker software packages (XWINNMR) Data analysis and assignment was performed using the programXEASY(ETH, Zurich, Swit-zerland) [40]
Structure calculation NOE cross-peaks in three-dimensional 15N and 13 C-NOESY-HSQC spectra and in two-dimensional NOESY spectra were integrated and converted into upper distance limits for interproton distances with the programCALIBA [41] The calibration curves for this conversion were adjusted iteratively as the structure calculations proceeded NOE cross peaks due to couplings between the two subunits were converted into upper distance limits using specific calibra-tion curves In the case of protons belonging to the ligand histidines an independent calibration has been used for each histidine Stereospecific assignments of diastereotopic
Trang 3protons have been obtained using the programGLOMSA[41]
and by the analysis of the HNHB experiment
Backbone dihedral angle restraints / were derived from
3JHNHa coupling constants by means of the appropriate
Karplus relationship For3JHNHavalues larger than 7 Hz
the / angle ranges between)155° and )80° while for values
lower than 4.5 Hz it ranges between)70° and )30° [36]
Backbone dihedral angle w for residue (i) 1) was
deter-mined from the ratio of the intensity of the daN(i) 1,i) and
dNa(i,i) NOE found on the15N plane of residue (i) in the
15N-NOESY-HSQC Ratio values of the residue (i) 1)
larger than 1 are characteristic of b sheets, with w values
ranging between 60° and 175°, while values smaller than 1
indicate a right handed a helix, with w values between)90°
and)10° [42] v1torsion angle constraints were derived by
the intensity ratios between the volume integral, dNb(i,i), in
the three-dimensional HNHB and the volume integral,
dNH(i,i), in the two-dimensional reference spectra, as
previously reported [37]
Structure calculations were performed using the program
DYANA [43] Fourteen-hundred random conformers were
annealed using the above constrains in 18 000 steps for the
initial calculations on a single subunit and in 22 000 steps
for the dimeric form The dimeric nature of the protein was
taken into account by connecting the amino acid sequence
of the two subunits with a chain of linkers composed of
atoms with a null Van der Waals radius The metal ions
were included in the calculations by adding special linkers
(pseudoresidues) in the amino-acid sequence following the
same procedure already used for the monomeric forms [26]
The linkers only define the metal–nitrogen distances, leaving
the conformation of the histidines completely free; the bond
angles at the copper and zinc ions are not imposed but can
freely change in the structural calculations, being only
determined by the experimental intrahistidine NOEs The
presence of the disulfide bridge between Cys57 and Cys146
was checked through SDS/PAGE and through the analysis
of the 13C shifts of the Cb of the cysteines In the
calculations three upper and three lower distance limits
were used to enforce the disulfide bond Cys57–Cys146
[between the Sc moieties of the two Cys, 2.0 (lower) and 2.1
(upper) A˚, and between the Cb of one Cys with the Sc of the
other, 3.0 (lower) and 3.1 (upper) A˚] [44]
The programCORMA[45], which is based on relaxation
matrix calculations, was used to back calculate the NOESY
cross-peaks from the calculated structure to assess the
quality of the structure
The final family was made up of 30 structures with the
lowest target function Restrained energy minimization in
vacuum (REM calculations) was applied to each member of
the family using the programAMBER5.0 [46] The setup of
the program and the parameters for the metal ions are as
previously reported [21] The value of NOE and torsion
angle potentials have been applied with force constants of
50 kcalÆmol)1ÆA˚)2(NOE), 32 kcalÆmol)1Ærad)2 (/,Y) and
2 kcalÆmol)1Ærad)2(v1)
The programMOLMOL[47] was used for identification of
hydrogen bonds (within the distance of 2.6 A˚ between
donor and acceptor, the N-H-O angle larger than 140° and
occurrence in at least 50% of the conformers)
The quality of the structure has been estimated by
Ramachandran plots obtained using the program
- [48]
R E S U L T A N D D I S C U S S I O N
Resonance assignment Native dimeric SOD is composed of two identical subunits which produce degenerate resonances Although this was already shown for the active site resonances from the investigation on the Cu2Co2SOD derivative [49], it is a relevant result as X-ray data [32] often indicate different conformations for the two subunits [10] Thus, averaging occurs on the NMR time scale, i.e faster than seconds The proton and15N resonances are not well dispersed and experience extensive overlap due to the specific folding
of the protein, characterized by extensive b sheet structure However, most of the 1H-15N cross peak degeneracies present in15N-HSQC spectrum were resolved in at least one
of the HNCA, HN(CO)CA, HNCO and HN(CA)CO-TROSY-type spectra already performed for the backbone assignment, reported by us [30]
The assignment of the resonances of the side chains was performed through the analysis of three-dimensional H(C)CH-TOCSY and (H)CCH-TOCSY spectra together with15N-NOESY-HSQC and13C-NOESY-HSQC spectra
In this way about 92% of the total proton resonances were assigned All the backbone proton and nitrogen resonances, with the single exception of Phe64, were assigned All the nitrogen side chain resonances of Asn and Gln, with the exception of Gln153 (the last one in each subunit), were assigned Ninety-nine percent of the backbone 13C reso-nances were assigned and about 86% of the13C side-chain resonances All the ring protons of the histidines of the active site and of His43 were assigned through the two-dimensional NOESY map The histidine coordination mode was determined through 1H-15N heteronuclear experiments, by detecting the2J15N-1H coupling between the imidazole nitrogen and nonexchangeable imidazole protons
Structure constraints and calculations
In three-dimensional15N- and13C-NOESY-HSQC spectra and in two-dimensional 15N-NOESY spectra, 3566 NOE cross peaks were assigned and converted into distance constraints Forty-nine dihedral / angles constraints were obtained from the analysis of the HNHA spectrum,
52 dihedral w angles were obtained from the 15 N-NOESY-HSQC spectrum and 14 v1 torsion angles from the HNHB spectrum A total of 45 proton pairs were stereospecifically assigned with the program GLOMSA: 16 protons belonging to bCH2, 11 to aCH, 1 to cCH2, 1 to dCH3and 1 to cCH3; and 15 bCH2were assigned by the analysis of the HNHB experiment In each subunit the metal ions were included by allowing copper to bind to Ne2 of His48 and His120 and to Nd1of His46, and zinc to Nd1 of His63, His71, and His80 and to Od1 of Asp83 Lower and upper distance limits of 1.8 and 2.3 A˚, respectively, were imposed between the metal ions and the donor atoms Finally, 3276 upper distance limits were generated of which 2853 were due to meaningful NOEs All the available information on the system, the linewidth and the number of signals, lead to a dimeric species with twofold symmetry The NOEs and 115 dihedral angles were initially used for the structural calculations of the
Trang 4monomeric species During these initial structure
calcula-tions, the presence of a small but significant number of
NOEs (96) inconsistent with couplings with protons of
residues of the same subunit, were identified and assigned
to connectivities between protons belonging to two
differ-ent subunits They were introduced in the calculations at a
later stage after refinement of the monomeric structure
Only one NOE has a contribution from both inter and
intra subunit; no severe violation with respect to the
calibration was observed For the calculations of the
dimeric structure, the intra subunit NOEs and dihedral
angle constrains were duplicated for each subunit and the
inter subunit NOEs were included The number of
constraints, divided in classes, are listed in Table 1 The
number of experimental NOEs per residue per subunit is
reported in Fig 1
From the final calculations, a family of 30 conformers,
with the lowest target function of 5.02 A˚2 (average value
5.99 A˚2), was obtained with an average violation per residue
of 0.016 A˚ Each conformer of this family was refined
further through REM calculations The rmsd to the mean
structure of the family is 0.79 ± 0.11 and 1.35 ± 0.09 A˚
for the backbone and the heavy atoms, respectively (rmsd
calculated over the fragment 3–151 for the holo protein)
After the refinement, the rmsd values are 0.78 ± 0.11 and
1.15 ± 0.09 A˚ for the backbone and the heavy atoms,
respectively If the rmsd is evaluated for each subunit of the
protein the values drop to 0.65 ± 0.09 and 0.66 ± 0.10 A˚,
respectively, for the backbone of the two subunits The
difference between rmsd of monomeric and dimeric species
is due to the indetermination of the reciprocal orientation of the two subunits The average total penalty for the REM family of the dimeric protein is of 1.42 ± 0.07 A˚2for the distance constrains; while for the average structure the value
is 1.36 A˚2 The rmsd values per residue, with the respect to the average structure, are shown in Fig 2
General shape of the protein and comparison with X-ray structures
A tube representation of the family of structures (back-bone and metal ions only) is shown in Fig 3 The family
of conformers was analyzed withPROCHECK-NMRand the results of the analysis are reported in Table 1 The secondary structure elements are eight antiparallel
b strands and a short five-residue a helix, which, connected
by loop regions, produce the typical SOD Greek key fold The secondary structure part of the protein is well defined The average rmsd values for the segments involved in the
b barrel are 0.50 ± 0.08 A˚ and 0.85 ± 0.06 A˚ for the backbone and all heavy atoms, respectively, which indi-cates that the b strands are characterized by lower disorder than the loops connecting them If a single subunit is considered, the b barrel rmsd values are 0.38 ± 0.06 A˚ and 0.77 ± 0.06 A˚, for the backbone and all heavy atoms, respectively These values indicate that the
b strands in each subunit are well defined The a helix within the family of conformers has an average rmsd to
Table 1 Restraint violations and structural and energetic statistics for the solution structure of reduced human SOD.
RSM violations per experimental distance constraint (A˚) b REM a (30 structures) <REM> a (mean) Intraresidue (723) 0.0251 ± 0.0013 0.0245
Sequential (1546) 0.0124 ± 0.0010 0.0119
Medium range (924) c 0.0149 ± 0.0009 0.0140
Long range (2513) 0.0115 ± 0.0005 0.0109
RSM violations per experimental dihedral angle constraints (deg)b
Average number of violations per structure lower than 0.3 A˚
Average no of NOE violations larger than 0.3 A˚ 0 0
Structural analysisd
% of residues in most favourable regions 71.6 73.6
% of residues in allowed regions 25.6 24.4
% of residues in generously allowed regions 2.3 2.1
% of residues in disallowed regions 0.6 0
a REM indicates the energy minimized family of 30 structures, <REM> is the energy minimized mean structure obtained from the coordinates of the individual REM structures.bThe number of experimental constraints for each class is reported in parentheses.cMedium range distance constraints are those between residues (i,i + 2) (i,i + 3) (i,i + 4) and (i,i + 5).dAs it results from the Ramachandran plot analysis.
Trang 5the mean structure of 0.39 ± 0.01 A˚ and 0.91 ± 0.21 A˚,
for the backbone and all heavy atoms, respectively These
values drop to 0.16 ± 0.08 A˚ and 0.65 ± 0.26 A˚ when a
single subunit is considered The comparison of the
present structure with the X-ray structures of the human
oxidized protein (1SOS) [11] and its G37R mutant [50]
show that the protein has the same folding in solution and
in solid state
The loops connecting the secondary structure elements can be divided in two groups: the loops I, III and V are quite well defined, while loops II, IV and VI are more disordered The odd loops are located on the opposite side of the barrel with respect to region involved in the subunit–subunit interface The even loops are in part located at the subunit– subunit interaction The first part of loop IV (49–62) shows (Fig 4) a much lower backbone rmsd in the present
Fig 1 Number of intraresidue (white), sequential (light grey), medium-range (grey) and long-range (black) intra subunit NOEs per residue (bottom) and number of inter subunit NOEs per residue (top) in human reduced native SOD.
Fig 2 Average rmsd values of backbone (j) and heavy atom (h) on two subunits per residue with respect to the average structure of human reduced native SOD (bottom) Backbone (j) and heavy atom (h) rmsd values of a single subunit per residue with respect to the average structure of human reduced native SOD (top).
Trang 6structure than in the monomeric Q133M2SOD structure.
This can be related to the occurrence of interactions with the
other subunit that minimize the exposure to solvent of
residues at the interface and stabilizes a single conformation
for them Indeed, in the segment 50–59 of Q133M2SOD,
five backbone HN signals were not assigned, probably due
to line-broadening as a consequence of proton exchange
with solvent due to their surface location This is consistent
with the analysis of the amide hydrogen–deuterium
ex-change behaviour previously reported [51] The other loops
are still disordered even in the dimeric form Therefore, this
behaviour suggests that this is a feature typical of this part of
the protein independent of its quaternary structure The
change in conformation of loop 50–59 upon dimerization is
reflected also on the location of Cys57, which can or cannot
perform a H-bond with the side chain of Arg143 depending
on its conformation (see below)
FALS mutations are spread over the entire molecule but
a higher density of mutations are clustered in a few regions
of the protein: at the interface between the two subunits (mainly in loop IV and b 8), in the odd loops and at the corresponding end of the b barrel, and in the even loops [52–56] Some of the residues involved in FALS mutations are conserved in SOD structures from different species (Fig 5) [11] The FALS mutations located in the first region are thought [11] to significantly destabilize the subunit– subunit contacts This is in agreement with the NMR data
Fig 5 Close-up of one subunit of human reduced SOD showing FALS mutations The FALS mutations located in odd loops are shown in gray, those in b strands are in black and located in the region close to the subunit–subunit interaction are coloured black and the residue labels are underlined.
Fig 3 Tube representation of the family of 30 structures of human reduced native SOD obtained with DYANA calculations and refined with REM
calculations Elements of secondary structure are highlighted (gray, b structure; black, a structure) The drawing has been produced with MOLMOL
[47].
Fig 4 Comparison of rmsd values to the average structure for the
backbone between dimeric SOD (s) and E133QM2SOD (.).
Trang 7on the solution structures [27,57] and on mobility studies
[30] of monomeric variants and human dimeric SOD, where
it has been shown that the absence of interactions with the
other subunit has sizable effects on enzymatic stability and
activity
Hydrogen–deuterium exchange
A total of 104 amide protons out of 147 were still present in
the1H-15N HSQC spectrum acquired 6 h after the
disso-lution of the lyophilized sample in D2O Fifty-one residues
are located in regions having a defined secondary structure
as they are involved in an extensive H-bond networks which
stabilize the b barrel structure typical of this protein Few
exceptions are observed in one of the b sheets (b6) where
amide protons belonging to three residues (Asp96, Ser98,
Glu100) out of six, exchange within 40 min and those
belonging to Asp101 within 1 h 40 min Also the a helix
shows exchanging amide protons in the time range between
40 min to 12 h
After 4 days 85 peaks, mostly belonging to the b barrel
and to loops III and VII were still present
Metal sites
In Fig 6 the active site is shown and compared with that of
oxidized human SOD All the metal ligands are well defined
in a single conformation For all the ligands, the rmsd value
calculated for all heavy atoms is smaller than 0.8 A˚, a value
that is similar to that obtained for secondary structure
elements The ligand conformation is also very close to that
observed in all the structures available for eukaryotic SOD,
either based on X-ray or NMR analysis in solution, dimeric
or monomeric The only exceptions are His63 and the
copper ion His63 experiences a larger variability among
the various structures and its orientation is dependent on the
copper oxidation state In oxidized SOD, His63 is
coordi-nated to the Cu ion through its Ne2, the distance between
copper and Ne2 being about 2.1 A˚ in the human structures
(1SOS) [11] and about 2.7 A˚ in a mutant (G37R) [50] In the
reduced state, the bond between Cu and His63 is broken, producing an increase in distance between the two In the case of the reduced dimeric yeast enzyme this distance increases to 3.2 A˚ [20] In the present structure the reduced copper is clearly tricoordinated, as expected from the data
on the monomer Indeed, upon copper reduction His63 becomes protonated at the Ne2 position, the bound proton resonating at 12.3 p.p.m and the distance between copper and Ne2 being 3.3 A˚ In the present structure the major structural changes induced by copper reduction is the movement of the copper ion which moves away from His63, experiencing a displacement of about 1.7 A˚ with respect to the oxidized enzyme So, the increased Cu–His63 distance, in the reduced state, is due to a movement of copper more than to a change in conforma-tion of His63 The posiconforma-tion of the other metal ligands, in the present structure, is very close to that found in the reduced dimeric yeast isoenzyme [20], whereas the copper ion positions in the two structures differ by about 1.0 A˚ It should be noted, however, that in the reduced yeast structure Cu and Ne2 of His63 are at a distance shorter than the sum of their van der Waals radii
The Zn ion does not experience significant movement from its site compared to the other structures In reduced monomeric human mutants (Q133M2SOD and M4SOD) the Zn ion moves farther from the copper ion The distance between the metals in the present structure is 7 A˚, this is similar to that in the dimeric yeast isoenzyme (6.7 A˚), while
in the human oxidized structure it ranges between 6.1 A˚ to 6.3 A˚
About the active site channel The active site channel is located between the electrostatic loop VII (120–144), implicated in assisting and increasing the affinity for the active site of substrate, and loop IV (49– 82) A network of H-bonds between the side chains of some residues belonging to loop VII plays a crucial role in increasing the diffusion rates of the superoxide radical inside the cavity [33] Comparing X-ray structures (1SOS, G37R and 1JCV) with the present one, it can be observed that the orientation of the a helix is the same in all the structures and the backbone remains almost unaltered In contrast the side chains experience different conformations: Glu132 shows a different orientation in each of the structures and in each of the subunits in the crystal cells, whereas no meaningful comparison can be carried out for Glu133, which shows disorder in the side chain Ser134 and Thr135 are quite ordered in the present structure, but they have a different orientation of the hydroxyl group with respect to the X-ray structures Side chain of Lys136 has different orientations in each of the X-ray structures The present one is closer to that in 1SOS and G37R structures Thr137 shows no significant changes in the orientation of the side chain although a movement towards Arg143 is observed, which slightly decreases the width of the active site channel
Thr58 and Glu133, with Glu132, define the opening of the active cavity The width is about 13 A˚ (distance between Thr58 Cc and Glu133 Oe), which is decreased by 1 A˚ with respect to 1SOS and 2 A˚ with respect to G37R Arg143 with Thr137 form a ÔbottleneckÕ for the active site, which excludes sterically large nonphysiological anions In
Fig 6 Active site of the family conformers of the reduced human dimer
(blue) and of the oxidized human dimer (red).
Trang 8the present structure also these residues are slightly closer
than the X-ray structure
Arg143 is important in orienting the superoxide anion
towards the Cu ion Comparing the present structure with
1SOS, G37R and 1JCW, the side chain of Arg143, in most
of the conformers shows no significant changes in the
orientation, while in the case of the monomers
(Q133M2SOD and M4SOD) the Arg143 side chain has a
different orientation (Fig 7) Cys57, with residues 58 and
61, was proposed to stabilize the orientation of Arg143
[26,27], as a result of hydrogen bonds between the side chain
of Arg143 (protons of Ng1 and Ng2 groups) and backbone
carbonyls of Cys57, Thr58 and Gly61 The H-bonds
involving Cys57, Thr58 and Gly61 are present in several
conformers The latter three residues are defined by 23, 24,
20 NOEs, respectively Furthermore, Cys57 forms a
disul-fide bond with Cys146, which is defined by 36 NOEs Side
chains of Arg143 and Cys57 are defined by 26 and seven
NOEs, respectively The Cu–Ng1 and Cu–Ng2 average
distances of Arg143 are 7.2 and 7.3 A˚ from copper, while in
the oxidized human protein the distances are 5.8 and 7.0 A˚
This is consistent with the already discussed movement of
the copper ion upon reduction
Cys57 seems to play a fundamental role in the process of
copper transfer from the copper chaperone for SOD (CCS)
and SOD itself as shown by the recently solved structure of
the CSS–SOD complex [58] In the latter structure, Cys229
of CCS forms a disulfide bond with Cys57 of SOD [58],
which therefore is not interacting any longer with Arg143
The guanidinium group of the latter residue in the complex
is very far away from the site where copper should be
introduced and is pointing towards the chaperone The
conformation of Arg143 is extremely sensitive to the
position of Cys57 [26,27] In the monomeric species, where
Cys57 experiences conformational equilibria, still
maintain-ing the disulfide bond, Arg143 is further from copper than
in the wild-type protein but closer than in the copper-free
SOD in the complex In the present solution structure of
wild-type SOD where the Cys57 is quite rigid, Arg143
assumes the optimal conformation respect to the copper
Therefore it seems that Arg143 is experiencing a movement that leads it to assume the correct conformation when SOD
is passing from the complex with CCS (where SOD is in a monomeric state) to the single monomeric protein, to the final dimeric structure
Relevant H-bonds
A network of H-bonds in dimeric human oxidized SOD [8] was proposed to play an important role in building the Greek key structure and in designing the metal binding site and the active cavity of the system The analysis of the H-bonds in the present structure has been carried out with MOLMOLprogram [47] on the final structure Except a few cases discussed later the hydrogens involved in H-bonds do not exchange in D2O Some of the H-bonds present in the human oxidized X-ray structures are observed also in the solution structure The H-bonds among ring hydrogens of His43 and backbone carbonyls of Thr39 and the Cu-ligand His120 are present in almost all the conformers of the family However the He2 of His43 do exchange in D2O indicating solvent exposure of such H-bond H-bonds involving ring hydrogens of His43 are important in linking the loop III to the b barrel and the active site The presence
of these H-bonds is consistent with the NMR observation of two HN ring protons signals for His43 (which is not involved in metal binding) at pH 5.0 In the present structure, the side chain Od of Asp124 forms a H-bond with the ring hydrogens He2 of His71 and of His46 Asp124 constitutes a long-range bridge between the copper site and the zinc site [8] Mutations of residue 124, which have been found in FALS proteins, affect mainly the zinc site and its affinity for the zinc ion [59] These mutations might produce zinc deficient species that have been shown to gain peroxynitrite producing activity, a possible cause of the FALS disease [60–63] A conserved H-bond between backbone HN of His71 and CO of Thr135, important in stabilizing the active site channel, is present in several conformers Thr135 belongs to the six residue helix involved
in the recognition and in the electrostatic guidance of the superoxide anion The amino acid site chains of Glu132, Glu133, Lys136 and Thr137 are involved in a hydrogen bonding network [33] In the present structure this H-bond network is maintained
For the FALS mutations located in the region constituted
by odd loops and one end of b barrel (Fig 5), as for example G37R, the absence of some H-bonds in the b hairpin region (loop V) is supposed to be responsible of the misfunction of the enzyme [50] In the present structure all the odd loops are well defined and this is consistent with the presence of a network of H-bonds that stabilizes this part of the protein This region is centered on Leu38, called the ÔplugÕ of one end of the b barrel [11], which fills a cavity formed by an array of apolar aminoacids present in different b strands (Ile35, the ring face of His43 and Leu144) and loop I (Val14) Thus producing a packed arrangement, crucial for correct enzymatic function and protein stability [64] Conserved H-bonds observed in this crucial part of the protein, observed in the present structure and identified with the programMOLMOL, are summarized
in Table 2 The H-bond connecting loop III and loop V, containing b hairpin (HN of Leu38 and CO of Gly93) and the H-bond between HN of Gly93 and CO of Asp90 (loop
Fig 7 View of the active channel of human reduced native SOD
Ori-entation of the side chain of Arg143 is reported for the reduced human
dimer (blue), for the oxidized human dimer (red), for the Q133M2SOD
(cyan) and for the reduced enzyme derived from yeast (yellow) The Cu
ion is shown as a sphere and the a helix is in orange.
Trang 9V) and between HN of Asp92 and side chain carboxylic
group of Asp90 are well conserved in all conformers of the
present family and in the X-ray structures (1SOS) even if
some difference in the stability of H-bond could be present
Indeed the amide proton of Asp92 disappears in D2O after
about 2 h, whereas the others are still present four days after
the dissolution in D2O In the FALS mutant G37R the
H-bond between HN of Asp92 and the carboxylic group of
Asp90 is present in only one of the two subunits [50] The
loss of this hydrogen bond in the G37R mutant [50] was
proposed to allow an increase flexibility in the b hairpin,
with respect to the wild-type protein; the latter, in fact, is
characterized by the absence of motions in the ps-ns
timescale in this region [30] Gly41, Gly37 and Gly93 seem
necessary to support main chain conformations and the
packing interaction in the hydrophobic plug [64] Gly41 is
involved in H-bond with Ala89 that, in its turn, is close to
the b-hairpin which is further stabilized by the H-bond
between Asp90 and Val94 The presence of extensive
H-bond networks seem to play a fundamental role in
stabilizing the secondary and tertiary structure of the
protein
C O N C L U S I O N S
The solution structure of dimeric human reduced Cu2Zn2
SOD was determined to a satisfactory degree of resolution
The two monomers are identical on the NMR time scale
The elements of secondary structure are the same as in the
X-ray structures and well resolved as well as the three odd
loops and the first part of loop IV, at the inter–subunit
interface The even loops, have a relatively high rmsd A
similar behavior is observed in a recently reported X-ray
structure of bovine SOD [32] The structure is also similar to
the solution structure of the monomeric mutants with the
exception of the significantly better definition of the first
part (49–63) of loop IV, which is disordered in the
monomers and experiences significant local mobility The
active channel is formed by the electrostatic loop VII, where
charged residues important in catalysis lie, and loop IV,
where Cys57 is located Upon dimerization, loop IV looses
the conformational exchange equilibria, occurring in the
ms-ls time range, and assumes a conformation which favors the formation of the hydrogen bond network The optimal conformation of the side chain of Arg143 is ensured by the formation of H-bonds between its terminal guanidinium group and the backbone oxygen atoms of Cys57 and Gly61 (loop IV) The latter network contributes to determine the optimal orientation of the strategic residue Arg143 and reduces its mobility in the subnanosecond time scale In contrast, the increased mobility, in the subnanosecond time scale, of the electrostatic loop VII (a helix) could assist O2
in sliding inside the active cavity, where it reaches the correct position helped by the interaction with the correctly oriented Arg143 The optimal orientation of Arg143 is found also in
a wild-type bacterial SOD [65], which is naturally mono-meric and where Cys57 (human numeration) is still H-bonded to Arg143
The copper site in the present dimeric structure is in a position similar to that of the monomeric mutants Because X-rays, when irradiating the crystals, may change the oxidation state or the solid state and may induce subtle structural changes, the present characterization of the reduced active site represents a further reliable picture of the reduced protein Upon reduction, copper moves inside the active cavity This is consistent with the earlier proposal [20,28,66] that the superoxide ion hardly reaches copper (I) but rather interacts with the e2 proton of His63 and is activated by this interaction for the transfer of one electron from copper (I) Finally, the strong H-bond network involving odd loops and one end of the b barrel (Table 2)
is observed in solution It may be relevant that some FALS mutants disrupt this network, giving them the capability of catalyzing other toxic reactions
In conclusion, the present structure of the dimeric wild-type SOD, although at lower resolution with respect to the X-ray structures, provides a clear refined picture of the relevant residues in solution and allows a thorough under-standing of the effects of establishing a quaternary structure
A C K N O W L E D G E M E N T S
This work was supported by the European Community (Contract number HPRI-CT-1999-00009 and QLG2-CT-1999-01003), by Italian CNR (Progetto Finalizzato Biotecnologie 99.00286.PF49 and 99.00950.CT03) and by MIUR-ex 40%.
R E F E R E N C E S
1 Fridovich, I (1974) Superoxide dismutase Adv Enzymol 41, 35–97.
2 Fridovich, I (1986) Superoxide dismutase Adv Enzymol Relat Areas Mol Biol 58, 61–97.
3 Valentine, J.S & Pantoliano, M.W (1981) Protein–metal ion in-teractions in cuprozinc protein (superoxide dismutase) In Copper Proteins 8 (Spiro, T.G., ed.), pp 291–291 Wiley, New York.
4 Halliwell, B & Gutteridge, J.M (1989) Free Radicals in Biology and Medicine pp 22–408 Clarendon Press, Oxford.
5 Fee, J.A & Gaber, B.P (1972) Anion binding to bovine ery-throcyte superoxide dismutase Evidence for multiple binding sites with qualitatively different properties J Biol Chem 247, 60–65.
6 Tainer, J.A., Getzoff, E.D., Richardson, J.S & Richardson, D.C (1983) Structure and mechanism of copper, zinc superoxide dis-mutase Nature 306, 284–287.
7 Tainer, J.A., Getzoff, E.D., Beem, K.M., Richardson, J.S & Richardson, D.C (1982) Determination and analysis of 2A˚
Table 2 H-bonds, present in the solution structure of human dimeric
reduced SOD, involving residues located in odd loops regions and one end
of b barrel and experiencing mutations in FALS mutants.
Gln15 (loop I) HN CO Lys36 (b 3)
Lys36 (b 3) HN CO Gln15 (loop I)
Leu38 (loop III) HN CO Gly93 (b hairpin)
Gly41 (loop III) HN CO Ala89 (loop V)
His43 (b 4) He2 CO Thr39 (loop III)
His43 (b 4) Hd1 CO His120 (loop VII)
His43 (b 4) HN CO Val87 (b 5)
Val87 (b 5) HN CO His43 (b 4)
Ala89 (loop V) HN CO Gly41 (loop III)
Asp90 (b hairpin) HN CO Val94 (b 6)
Asp92 (b hairpin) HN Od1 Asp90 (b hairpin)
Asp92 (b hairpin) HN Od2 Asp90 (b hairpin)
Gly93 (b hairpin) HN CO Asp90 (b hairpin)
Ala95 (b 6) HN CO Ile35 (b 3)
Trang 10structure of copper zinc superoxide dismutase J Mol Biol 160,
181–217.
8 Parge, H.E., Hallewell, R.A & Tainer, J.A (1992) Atomic
structures of wild-type and thermostable mutant recombinant
human Cu,Zn superoxide dismutase Proc Natl Acad Sci USA
89, 6109–6114.
9 Parge, H.E., Getzoff, E.D., Scandella, C.S., Hallewell, R.A &
Tainer, J.A (1986) Crystallographic characterization of
recombinant human CuZn superoxide dismutase J Biol Chem.
261, 16215–16218.
10 Bertini, I., Mangani, S & Viezzoli, M.S (1998) Structure and
properties of copper/zinc superoxide dismutases In Advanced
Inorganic Chemistry (Sykes, A.G., ed.), pp 127–250 Academic
Press, San Diego, CA, USA.
11 Deng, H.-X., Hentati, A., Tainer, J.A., Lqbal, Z., Cyabyab, A.,
Hang, W.-Y., Getzoff, E.D., Hu, P., Herzfeldt, B., Roos, R.P.,
et al (1993) Amyotrophic lateral sclerosis and structural defects in
Cu,Zn superoxide dismutase Science 261, 1047–1051.
12 Battistoni, A., Folcarelli, S., Rotilio, G., Capasso, C., Pesce, A.,
Bolognesi, M & Desideri, A (1996) Crystallization and
pre-liminary X-ray analysis of the monomeric Cu,Zn superoxide
dis-mutase from Escherichia coli Protein Sci 5, 2125–2127.
13 Djinovic, C.K., Battistoni, A., Carrı`, M., Polticelli, F., Desideri,
A., Rotilio, G., Coda, A., Wilson, K & Bolognesi, M (1996)
Three-dimensional Structure of Xenopus laevis Cu,ZnSOD b
determined by X-ray crystallography at 1.5 A˚ resolution Acta
Cryst D52, 176–188.
14 Djinovic, K., Gatti, G., Coda, A., Antolini, L., Pelosi, G.,
Desideri, A., Falconi, M., Marmocchi, F., Rotilio, G &
Bolog-nesi, M (1992) Crystal structure of yeast Cu,Zn, superoxide
dismutase Crystallographic refinement at 2.5 A˚ resolution J Mol.
Biol 225, 791–809.
15 Djinovic, K., Gatti, G., Coda, A., Antolini, L., Pelosi, G.,
Desideri, A., Falconi, M., Marmocchi, F., Rotilio, G &
Bolog-nesi, M (1991) Structure solution and molecular dynamics
refinement of the yeast Cu,Zn enzyme superoxide dismutase Acta
Crystallogr B47, 918–927.
16 Kitagawa, Y., Tanaka, N., Hata, Y., Kusonoki, M., Lee, G.,
Katsube, Y., Asada, K., Alibara, S & Morita, Y (1991)
Three-dimensional structure of Cu,Zn, superoxide dismutase from
spi-nach at 2.0 A˚ resolution J Biochem 109, 447–485.
17 Djinovic, K., Coda, A., Antolini, L., Pelosi, G., Desideri, A.,
Falconi, M., Rotilio, G & Bolognesi, M (1992) Crystal stucture
and refinement of the semisynthetic cobalt-substituted bovine
erythrocyte superoxide dismutase at 2.0 A˚ resolution J Mol Biol.
226, 227–238.
18 Bordo, D., Matak, D., Djinovic-Carugo, K., Rosano, C., Pesce,
A., Bolognesi, M., Stroppolo, M.E., Falconi, M., Battistoni, A &
Desideri, A (1999) Evolutionary constraints for dimer formation
in prokaryotic Cu,Zn superoxide dismutase J Mol Biol 285,
283–296.
19 Rypniewski, W.R., Mangani, S., Bruni, B., Orioli, P.L., Casati, M.
& Wilson, K.S (1995) Crystal structure of reduced bovine
eri-throcyte superoxide dismutase at 1.9 A˚ resolution J Mol Biol.
251, 282–296.
20 Ogihara, N.L., Parge, H.E., Hart, J.P., Weiss, M.S., Goto, J.J.,
Crane, B.R., Tsang, J., Slater, K., Roe, J.A., Valentine, J.S.,
Eisenberg, D & Tainer, J.A (1996) Unusual trigonal-planar
copper configuration revealed in the atomic structure of yeast
copper-zinc superoxide dismutase Biochemistry 35, 2316–2321.
21 Banci, L., Bertini, I., Bruni, B., Carloni, P., Luchinat, C.,
Mangani, S., Orioli, P.L., Piccioli, M., Rypniewski, W & Wilson,
K (1994) X-ray structure, NMR and molecular dynamics of the
reduced form of copper-zinc superoxide dismutase Biochem.
Biophys Res Commun 202, 1088–1095.
22 Bertini, I., Luchinat, C & Monnanni, R (1985) Evidence of the
breaking of the copper-imidazolate bridge in
copper/cobalt-substituted superoxide dismutase upon reduction of the copper (II) centers J Am Chem Soc 107, 2178–2179.
23 Bertini, I., Luchinat, C., Piccioli, M., Vicens Oliver, M & Viezzoli, M.S (1991) 1H NMR investigation of reduced copper-cobalt superoxide dismutase Eur Biophys J 20, 269–279.
24 Bertini, I., Capozzi, F., Luchinat, C., Piccioli, M & Viezzoli, M.S (1991) Assignment of active site protons in the1H NMR spectrum
of reduced human Cu,Zn superoxide dismutase Eur J Biochem.
197, 691–697.
25 Paci, M., Desideri, A., Sette, M., Cirioli, M.R & Rotilio, G (1990) Assignment of imidazole resonances from two-dimensional proton NMR spectra of bovine Cu,Zn superoxide dismutase Evidence for similar active site conformation in the oxidized and reduced enzyme FEBS Lett 263, 127–130.
26 Banci, L., Benedetto, M., Bertini, I., Del Conte, R., Piccioli, M & Viezzoli, M.S (1998) Solution structure of reduced monomeric Q133M2 copper, zinc superoxide dismutase Why is SOD a dimeric enzyme? Biochemistry 37, 11780–11791.
27 Banci, L., Bertini, I., Del Conte, R., Mangani, S., Viezzoli, M.S & Fadin, R (1999) The solution structure of a monomeric reduced form of human copper, zinc superoxide dismutase bearing the same charge as the native protein J Biol Inorg Chem 4, 795–803.
28 Ferraroni, M., Rypniewski, W., Wilson, K.S., Viezzoli, M.S., Banci, L., Bertini, I & Mangani, S (1999) The crystal structure of the monomeric human SOD mutant F50/G51E/E133Q at atomic resolution The enzyme mechanism revisited J Mol Biol 288, 413–426.
29 Luty, B.A., El Amrani, S & McCammon, J.A (1993) Simulation
of the bimolecular reaction between superoxide and superoxide dismutase: synthesis of the encounter and reaction steps J Am Chem Soc 115, 11874–11877.
30 Banci, L., Bertini, I., Cramaro, F., Del Conte, R., Rosato, A & Viezzoli, M.S (2000) Backbone dynamics of human Cu, Zn superoxide dismustase and of its monomeric F50/EG51E/E133Q mutant: the influence of dimerization on mobility and function Biochemistry 39, 9108–9118.
31 Fisher, C.L., Cabelli, D.E., Tainer, J.A., Hallewell, R.A & Getzoff, E.D (1994) The role of arginine 143 in the electrostatic and mechanism of Cu,Zn superoxide dismutase: computational and experimental evaluation of site-directed mutants Proteins Struct Funct Genet 19, 24–34.
32 Hough, M.A & Hasnain, S.S (1999) Crystallographic structures
of bovine copper-zinc superoxide dismutase reveal asymmetry in two subunits: functionally important three and five coordinate copper sites captured in the same crystal J Mol Biol 287, 579– 592.
33 Getzoff, E.D., Cabelli, D.E., Fisher, C.L., Parge, H.E., Viezzoli, M.S., Banci, L & Hallewell, R.A (1992) Faster superoxide dis-mutase mutants designed by enhancing electrostatic guidance Nature 358, 347–351.
34 Kay, L.E., Xu, G.Y., Singer, A.U., Muhandiram, D.R & Forman-Kay, J.D (1993) A gradient-enhanced HCCH-TOCSY experiment for recording side-chains 1 H and 13 C correlations
in H 2 O samples of proteins J Magn Reson Series B 101, 333– 337.
35 Wider, G., Neri, D., Otting, G & Wu¨thrich, K (1989) A heteronuclear three-dimensional NMR experiment for measure-ments of small heteronuclear coupling constants in biological macromolecules J Magn Reson 85, 426–431.
36 Vuister, G.W & Bax, A (1993) Quantitative J correlation: a new approach for measuring homonuclear three-bond J (HNHa) coupling constants in15N enriched proteins J Am Chem Soc.
115, 7772–7777.
37 Archer, S.J., Ikura, M., Torchia, D.A & Bax, A (1991) An alternative 3D NMR technique for correlation backbone 15 N with side chain Hb resonances in larger proteins J Magn Reson 95, 636–641.