Download Mutational and spectroscopic studies of the significance of the active

www.elsevier.nl / locate / jinorgbio
Journal of Inorganic Biochemistry 80 (2000) 247–256
Mutational and spectroscopic studies of the significance of the active site
glutamine to metal ion specificity in superoxide dismutase
Aaron L. Schwartz b , Emine Yikilmaz a,b , Carrie K. Vance b ,1 , Surekha Vathyam b , Ronald L. Koder a ,
Anne-Frances Miller a,b
b
a
Department of Chemistry, University of Kentucky, Lexington, KY 40506 -0055, USA
Departments of Chemistry and Biophysics, The Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218, USA
Received 11 October 1999; received in revised form 7 March 2000; accepted 17 March 2000
Abstract
We are addressing the puzzling metal ion specificity of Fe- and Mn-containing superoxide dismutases (SODs) [see C.K.Vance, A.-F.
Miller, J. Am. Chem. Soc. 120(3) (1998) 461–467]. Here, we test the significance to activity and active site integrity of the Gln side chain
at the center of the active site hydrogen bond network. We have generated a mutant of MnSOD with the active site Gln in the location
characteristic of Fe-specific SODs. The active site is similar to that of MnSOD when Mn 21 , Fe 31 or Fe 21 are bound, based on EPR and
NMR spectroscopy. However, the mutant’s Fe-supported activity is at least 7% that of FeSOD, in contrast to Fe(Mn)SOD, which has 0%
of FeSOD’s activity. Thus, moving the active site Gln converts Mn-specific SOD into a cambialistic SOD and the Gln proves to be
important but not the sole determinant of metal-ion specificity. Indeed, subtle differences in the spectra of Mn 21 , Fe 31 and 1 H in the
presence of Fe 21 distinguish the G77Q, Q146A mut-(Mn)SOD from WT (Mn)SOD, and may prove to be correlated with metal ion
activity. We have directly observed the side chain of the active site Gln in Fe 21 SOD and Fe 21 (Mn)SOD by 15 N NMR. The very different
chemical shifts indicate that the active site Gln interacts differently with Fe 21 in the two proteins. Since a shorter distance from Gln to Fe
and stronger interaction with Fe correlate with a lower Em in Fe(Mn)SOD, Gln has the effect of destabilizing additional electron density
on the metal ion. It may do this by stabilizing OH 2 coordinated to the metal ion.  2000 Elsevier Science S.A. All rights reserved.
Keywords: Superoxide dismutase; Redox potential; Glutamine; Metal ion specificity
1. Introduction
Many Fe- and Mn-containing superoxide dismutases
(FeSODs and MnSODs) have virtually identical over-all
structures and active sites [1,2], homologous amino acid
sequences [3,4] and the ability to bind the non-native metal
ion, Mn or Fe, respectively. Nonetheless, most are active
only with their native metal ion bound [5,6]. Since both Fe
and Mn can mediate superoxide dismutation and both the
Fe- and Mn-specific proteins ((Fe)SODs and (Mn)SODs)
can support this activity, the defect resides in the interaction or match between the protein and the metal ion. We
have shown that the inactivity of Fe-substituted (Mn)SOD
(Fe(Mn)SOD) can be explained by its anomalously low
Em , which is consistent with the (Mn)SOD protein apply-
E-mail address: [email protected] (A.-F. Miller).
Present address: Department of Pharmacology and Therapeutics, University of Florida, Gainesville, FL 32610, USA.
1
ing Em tuning appropriate to Mn 31 / Mn 21 to bound Fe [7].
Thus, the SODs are an excellent system for understanding
redox tuning.
Since there are relatively few conserved amino acid
differences between (Fe)SODs and (Mn)SODs, it may be
possible to identify specific amino acids responsible for Em
tuning. The most striking conserved amino acid differences
between Fe- and Mn-SODs are a Gln at position 69 of
FeSOD 2 and Gly at the corresponding position (77) in
MnSOD, a Gln at position 146 of MnSOD and Ala at the
corresponding position (141) in FeSOD, Tyr at position 76
of FeSOD versus Phe84 of MnSOD, and Asp147 of
MnSODs versus Gly142 of most FeSODs. FeSOD in
which the Tyr76 was mutated to Phe did not acquire
2
Numbering of E. coli FeSOD is used for (Fe)SOD and that of E. coli
MnSOD is used for (Mn)SOD. Positions 69, 76 and 141 of FeSOD
correspond to positions 77, 84 and 146 of MnSOD.
0162-0134 / 00 / $ – see front matter  2000 Elsevier Science S.A. All rights reserved.
PII: S0162-0134( 00 )00086-6
248
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
Fig. 1. Cartoons of the active sites of FeSOD (left) and MnSOD (right)
from E. coli showing the active site Gln and its hydrogen bonding
21
21
˚ that
partners. In Fe SOD the distance from Gln’s N to Fe is 5.1 A,
21
˚
from Gln’s O to Fe is 6.1 A, in MnSOD the distance from Gln’s N and
˚ respectively, and in Fe 31 (Mn)SOD the
O to Mn are 4.6 and 5.7 A,
31
˚ Coordinates used
distance from Gln’s N and O to Fe are 4.4 and 5.5 A.
are those of Refs. [2,15].
Mn-supported activity, instead its Fe-supported activity
increased [8]. Thus, this residue alone is not a crucial
determinant of metal ion catalytic activity.
The Gln conserved at position 69 of FeSOD is clearly
analogous to the Gln at position 146 of MnSOD as their
side chains occupy similar positions in the active site and
its hydrogen bond network [9]. The conserved active site
Gln is believed to be central to the active site hydrogen
bonding network as it hydrogen bonds with both the
universally conserved Tyr34 responsible for the pK that
governs activity and substrate specificity [10–14] and
coordinated solvent (Fig. 1) [2,15]. The strength and
polarity of the hydrogen bond with coordinated solvent in
effect alters the degree to which the latter is protonated
which in turn can have a very large effect on the Em of the
metal ion. Indeed, Mn 31 / Mn 21 is calculated to have a
1.3-V higher Em when the coordinated solvent is protonated than when it is not [16]. This would be more than
sufficient to account for the 0.46 V difference between the
Em s of FeSOD and Fe(Mn)SOD (Table 1) [7]. Thus, we
would like to learn whether the hydrogen bonding between
Gln and coordinated solvent is different in (Mn)SOD and
(Fe)SOD, with each of Fe and Mn bound. The effects of
the active site Gln may indeed be different in the two
categories of proteins since it originates from different
directions and different elements of the SOD structure in
the Fe- and Mn-specific SODs.
When the FeSOD of Mycobacterium tuberculosis, which
has a His in the place of the MnSOD-characteristic Gln 3 ,
was mutated to place Gln or Glu there, the former mutant
preferentially bound Fe and the latter preferentially bound
Mn. No activities were reported [17]. Nonetheless, it is
interesting that Gln at position 146 does not enforce
Mn-binding. Cambialistic 4 SOD from Porphyromonas
gingivalis has also been mutated to incorporate the Gln
characteristic of MnSOD, instead of its native Gln characteristic of FeSOD [18]. This resulted in lower-than-WT
activity with Fe bound but higher-than-WT activity with
Mn bound, demonstrating that Gln at position 146 does
increase Mn-supported activity. The authors noted that the
Mn 31 optical spectrum was stronger in the mutant, suggesting the possibility that the mutant stabilizes more
Mn 31 , and has a lower Em , consistent with the hypothesis
of Vance and Miller that greater Mn-supported activity
results when the protein depresses the Em more [7]. Here
we report the complementary mutant in which a (Mn)SOD
protein is mutated to contain Gln at the position characteristic of FeSODs.
We have used EPR spectra of Fe 31 or Mn 21 to compare
their coordination geometries and strengths of zero-field
splittings in different WT and mutant protein environments. In addition we have characterized the active site
Gln’s interactions with the metal ion via direct observation
of that amino acid by NMR spectroscopy, in which the line
widths and positions of paramagnetically shifted resonances reflect the metal ion’s electronic state and geometry
(via the magnetic susceptibility tensor) and the degree to
which unpaired electron spin density is delocalized (via
hyperfine coupling). Comparison of the NMR signals of
analogous residues in two different proteins thus reveals
3
M. tuberculosis has a Gly at the position of the FeSOD-characteristic
Gln.
4
Active with either Mn or Fe.
Table 1
Activity of G69Q / Q146A-(Mn)SOD containing either Fe or Mn
Fe-SOD
Fe(Mn)SOD
Fe-G77Q,Q146A-(Mn)SOD
Mn-G77Q,Q146A-(Mn)SOD
Mn-SOD
Activity a
Per metal ion b
Fe content c
6500
0
360
3400
6800
6800 (100%)
(0%)
450 (6.5%)
4900 (71%)
6900 (100%)
0.95
0.86
0.81
0.01
0
Mn content d
e
ND
ND
0.70
0.98
Em
220 mV f
2240 mV f
a
Measured using the xanthine oxidase assay at pH 7.8 [26] which has a detection limit corresponding to less than 0.1% of native SOD activity and an
uncertainty of |2% of the measured activity.
b
Activity on a per Fe or Mn basis (as a percentage of the appropriate WT activity).
c
Fe content was measured using a ferrozine assay [22] and is expressed on a per subunit basis. The uncertainty is estimated to be 0.03 per subunit.
d
Mn content was measured by EPR after acidification of the sample, and is expressed on a per subunit basis. The detection limit is less than 1 mM Mn
or 0.005 / subunit and the uncertainty is approximately 3%.
e
None detected.
f
Versus NHE, from Ref. [7].
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
differences in their interactions with and position relative
to the metal ion, while comparison of general features of
the paramagnetically shifted 1 H NMR spectrum provides
insight into the state of the metal ion.
2. Methods
FeSOD, MnSOD and Fe(Mn)SOD were prepared as in
Refs. [7,12,19,20], respectively. The E. coli gene for
MnSOD (sodA) was cloned into the overexpression vector
pET24d (Novagen) using standard techniques and
mutagenized by PCR using the megaprimer method [21].
Both strands of the mutant gene were completely sequenced to confirm the presence of the desired Gly77 to
Gln and Gln146 to Ala mutations, and the absence of any
others. The mutant gene was overexpressed in E. coli
QC774 / DE3 and the protein was purified using the same
protocol as for MnSOD [19,20]. Fe content was measured
using the ferrozine assay of Carter [22], and Mn was
quantified by EPR at low temperature after acid hydrolysis
of the samples to release all Mn as Mn 21 and addition of
glycerol to 50% (v / v).
Selective 15 N labeling of Gln side chains was accomplished as in Ref. [23], using the E. coli Asn auxotrophic
strain JK120 [24].
EPR spectra were collected on a Bruker 300 EMX
calibrated at g952 using DPPH and at g956 using
myoglobin in 100 mM K 2 HPO 4 . EPR spectra of Fe 31
were collected at 9.47 GHz at 70 K, with 80 mW nominal
microwave power, 2 G modulation amplitude and 3000 G
wide scans centered at 2100 G obtained in approximately
20 min. EPR spectra of Mn 21 were collected as above but
with a nominal microwave power of 160 mW, 10 G
modulation amplitude and 500065000 G scans of 5.5 min
each, except for the spectrum of Mn 21 SOD which was
collected at 4 K. EPR samples were buffered with 50 mM
potassium phosphate and contained 50 mM NaCl but no
cryoprotectant. The samples of MnSOD and Mn-mut-SOD
were reduced with dithionite but the sample of
Mn(Fe)SOD required no reduction. The pH was measured
before freezing and after the sample was thawed to confirm
that the pH had been stable.
1
H NMR spectra were obtained at 300 MHz for protons
on an AMX 300 at 308C as previously described [12] using
the super-WEFT pulse sequence [25] with a rapid repetition rate to emphasize the rapidly relaxing resonances from
the paramagnetic active site. The acquisition time was 25
ms and the water resonance was saturated during the
35-ms delay between pulses, as well as the 15-ms relaxation delay between repetitions. For each NMR sample, 450
ml of 1 mM SOD in 100 mM K 2 HPO 4 pH 7.6 and 50 ml
2
H 2 O were degassed and reduced with 5 ml of 150 mM
sodium dithionite in an NMR tube fitted with a Teflon seal,
or a standard NMR tube which was then flame sealed.
15
N NMR spectra were collected on a Unity plus 500 at
249
50.655 MHz and 258C using either 60 or 908 pulses
followed by observation without 1 H decoupling, or a
super-WEFT pulse sequence [25] with 1 H decoupling
during the 0.1-s acquisition time but not the 0.7-s relaxation delay that followed or the 0.6-s delay between the two
pulses. Samples of 1 mM SOD in 50 mM potassium
phosphate at pH 7.8 or 7.5 with 10 or 25 mM NaCl and
10% 2 H 2 O were oxidized with KMnO 4 and filtered
(Mn 31 SOD) or degassed and reduced as above (Fe 21 SOD
and Fe 21 (Mn)SOD) or observed without further treatment
(apo(Mn)SOD).
3. Results
3.1. G77 Q ,Q146 A-( Mn)SOD expression and activity
The mutant protein overexpressed well but the identity
and quantity of metal ion incorporated by the protein
varied depending on the composition of the growth
medium. Therefore, in order to be able to work with
protein containing one metal ion only, bound metal ions
were removed and replaced by either Fe or Mn using
previously developed methods [7]. The resulting preparations of mutant protein containing Fe (Fe-G77Q,Q146A(Mn)SOD or Fe-mut(Mn)SOD), or Mn (MnG77Q,Q146A-(Mn)SOD or Mn-mut(Mn)SOD) were characterized with respect to both Fe and Mn content and SOD
activity in the standard assay [26]. Table 1 shows that
Mn-mut-(Mn)SOD retained 70% of WT MnSOD activity.
This assay employs low O ?2
concentrations and assesses
2
k cat /KM , which is indicative of binding steps as well as
chemical steps. Although the mutations affect residues
central to the active site, the retention of 70% of activity
suggests that the active site remains intact. Fe-mut(Mn)SOD displayed small but significant activity, corresponding upon correction for substoichiometric Fe incorporation to 7% of WT FeSOD activity. Although the
decrease in Mn-supported activity to 70% is virtually
insignificant, the increase in Fe-supported activity from 0
to 7% is dramatic. To learn what changes in the active site
may be responsible for mut-(Mn)SOD’s significantly increased Fe-supported activity, we have begun characterizing the mutant’s active site and comparing it with those of
(Fe)SOD and (Mn)SOD, with each of Fe and Mn bound.
3.2. Spectroscopy of G77 Q ,Q146 A-( Mn)SOD
In order to assess the degree to which the active site is
disrupted or deformed by relocating the Gln, we have
compared the EPR spectra of Mn 21 SOD and Mn 21 -mut(Mn)SOD. Fig. 2 shows that these spectra are similar in
general appearance, with similar features appearing in
similar places. Although good simulations are required to
support a quantitative comparison, the spectra nonetheless
suggest that neither the coordination geometry of Mn 21 ,
250
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
Table 2
pH values of samples and g values from Fe 31 EPR spectra of SODs
Sample
pH
g values
FeSOD
Fe-G77Q,Q146A-(Mn)SOD
Fe(Mn)SOD
7.65
7.4
7.53
4.81
5.00
5.00
4.02
3.86 a
3.84
3.69
3.43
3.46
a
Due to the overlapping negative portion of the non-active site
rhombic Fe signal, this g value was obtained by extrapolating the
negative portion of SOD’s signal to lower fields and noting the intersection with the best baseline (see Fig. 2).
parison. The EPR spectrum of Fe 31 -mut-(Mn)SOD is
more similar to that of Fe 31 (Mn)SOD than that of
Fe 31 SOD, based on g values (Table 2) and appearance,
neglecting the somewhat larger contribution of rhombic
non-active-site (NAS) Fe 31 to the spectrum (Fig. 3).
These indicate that the coordination geometry and electronic state of Fe 31 , like those of Mn 21 , are not altered
significantly by relocating the Gln side chain in the active
site of (Mn)SOD. The EPR signal of Fe 31 -mut-(Mn)SOD
is distinctly different from that of Fe 31 SOD however.
Fig. 2. EPR comparison of Mn 21 SOD (top), Mn 21 -mut-(Mn)SOD (center) and Mn 21 (Fe)SOD (bottom). Samples were buffered at pH 7.8 with
50 mM potassium phosphate and 50 mM NaCl.
the covalency nor the magnitude of zero-field splitting, is
changed very much upon relocation of the active site Gln
(and replacement of a Gly by an Ala). By contrast, there
are a number of clear differences between the spectra of
mutant and WT Mn 21 SOD on the one hand, and that of
Mn 21 (Fe)SOD on the other. These results are consistent
with the fact that mut-MnSOD retains 203 amino acids
identical to those of MnSOD and only two identical to
Mn(Fe)SOD. Thus the active site Gln’s interaction with
Mn 21 does not appear to be strongly sensitive to the origin
of the Gln in the amino acid sequence, but is more
dependent on the identities of the rest of the active site
amino acids.
To address possible location-specific interactions with
Fe, Fe-mut-(Mn)SOD was compared to FeSOD and
Fe(Mn)SOD. EPR was used for the oxidized state comparison and NMR was used for the reduced state com-
Fig. 3. EPR comparison of Fe 31 SOD (top), Fe 31 -G77Q,Q146A(Mn)SOD (center) and Fe 31 (Mn)SOD. Sample pH values were 7.65, 7.4
and 7.53, respectively.
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
Thus, variations in the signal appear to predominantly
reflect the identity of the rest of the protein, not the
different locations of the Gln in the amino acid sequence.
The NAS Fe 31 signal indicates that there is approximately a 1:0.4 ratio of active site to NAS Fe 31 . This
reflects the mutant protein’s tendency to loose active site
Fe when extensive effort was made to remove NAS Fe.
However it also indicates that the actual Fe-supported
activity on a per active site Fe basis is 1.4 times the
activity reported in Table 1 on a per total Fe basis, or 9%
of WT activity. Thus Fe-mut-(Mn)SOD appears to have
almost 10% WT levels of Fe-supported activity, but
certainly more than 7%.
Fig. 4 shows that the same general trends hold for the
Fe 21 state of Fe 21 -mut-(Mn)SOD, whose NMR spectrum
is more similar to that of Fe 21 (Mn)SOD than that of
Fe 21 SOD. Subtle differences between Fe 21 -mut(Mn)SOD’s and Fe 21 (Mn)SOD’s spectra between 10 and
15 ppm might be attributable to the mutated amino acids
and shifts of conserved features between 22 and 30 ppm,
and 215 to 25 ppm may reflect slight conformational
changes. However, these distinctions are significantly
smaller than the differences between Fe 21 (Mn)SOD and
Fe 21 SOD, which result in quite different appearance
between 30 and 10 ppm and distinct chemical shifts for the
ligand His’ Nd1 protons’ resonances between 30 and 90
ppm. The overall similarity in chemical shift range and line
widths in all three spectra nonetheless indicate that Fe 21 ’s
electronic structure and geometry are very similar in the
three proteins. Since the NMR spectrum is generally
similar in Fe 21 SOD and Fe 21 (Mn)SOD [20], it is not very
surprising that it is similar in Fe 21 -mut-(Mn)SOD as well.
The spectra above describe three of the four possible
metal ion and oxidation states in SOD and indicate that
they are not much changed by moving Gln146 to position
77 in (Mn)SOD. This is consistent with Mn-mut(Mn)SOD’s retention of close to full WT activity in the
standard assay. However, the acquisition of at least 7% of
the FeSOD level of Fe-supported activity indicates that the
251
subtle spectroscopic differences between WT and mut(Mn)SOD reflect electronic or other changes sufficient to
support small but significant activity.
Since the inactivity of Fe(Mn)SOD can be explained by
its anomalously low Em [7], Fe-mut-(Mn)SOD’s recovery
of some activity suggests that its Em is higher than that of
Fe(Mn)SOD. The above preliminary characterizations
demonstrate that G77Q,Q146A-(Mn)SOD retains basic
active site integrity and should be amenable to further
study, including redox titrations.
3.3. NMR spectroscopy of the Gln side chain amide in
WT Fe 21 ( Mn)SOD
The active site Gln hydrogen bonds directly with the
coordinated solvent, and may thereby exert a strong effect
on the metal ion Em and thus its SOD activity [7].
Therefore, we wish to learn how the Gln’s relation to the
metal ion correlates with enzyme activity. We have compared this relation in active Fe 21 SOD with that in inactive
Fe 21 (Mn)SOD by NMR. The Gln side chain’s close
proximity to the bound metal ion precludes its observation
by 1 H NMR, but the low gyromagnetic ratio of 15 N
permits direct observation of the side chain amide by 15 N
NMR in the presence of Fe 21 [23,30].
15
NH 3 is incorporated directly into the side chain of Gln
by E. coli while direct incorporation into Asn is prevented
by the use of an auxotrophic strain. Most other amino acid
positions derive their N indirectly from Gln. Thus, in an
15
N NMR spectrum of protein isolated from E. coli / JK120
grown in the presence of 15 NH 3 and all the amino acids
except Gln in 14 N form, only the side chains of Gln appear
[23]. The side chain N of Gln69 of Fe 21 SOD has been
observed at 158 ppm, strongly shifted from the 113- and
114-ppm resonances of Gln side chains distant from Fe 21
[23].
When MnSOD was labeled with 15 N in the side chains
of its 6 Glns, there was little leak into the side chains of
Asn, since only five strong pairs of resonances plus two
Fig. 4. NMR comparison of Fe 21 SOD (top), Fe 21 -G77Q,Q146A-(Mn)SOD (center) and Fe 21 (Mn)SOD.
252
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
Fig. 5. HSQC of [ 15 N]Gln-labeled (Mn)SOD. NMR spectra were collected at 258C on a Unity plus spectrometer at 500 MHz for 1 H and 50 MHz for 15 N.
1
H– 15 N HSQCs [40] were collected with sweep widths of 1650 Hz for 15 N and 6500 Hz for 1 H with GARP decoupling of 15 N during acquisition. Data
were processed using NMRPipe [41]. Chemical shifts were referenced to water at 4.78 ppm for 1 H and external 15 NH 4 Cl (2.9 mM) in 1 M HCl at 208C at
24.93 ppm [42].
weak ones were observed (Fig. 5). Nor was there any
detectable label leak into the backbone based on the
complete absence of any backbone resonances (between 7
1
15
and 9 H ppm and 108 and 126 N ppm, [27]), or
resonances lacking partners at the same 15 N chemical
5
shift . The selectivity of the labeling signifies that all
strong 15 N resonances observed can be safely ascribed to
Gln side chain amides.
31
The HSQC of Mn SOD in Fig. 5 is not expected to
contain signals from Gln146’s side chain 15 N as the latter
˚ from the bound Mn 31 [15] and subject to severe
is 4.6 A
paramagnetic relaxation. The other 5 Gln side chain Ns are
˚ or further from the Mn based on the crystal structure
14 A
of MnSOD [15], consistent with the five pairs of strong
resonances observed. These are poorly resolved in the 15 N
1D spectrum. However, direct detection of 15 N permits
observation of more rapidly relaxed 15 N nuclei. Indeed, an
additional resonance is observed at 192 ppm in the 15 N
spectrum of Fe 21 (Mn)SOD (Fig. 6). This was not present
in either Mn 31 SOD or apo(Mn)SOD, presumably due to
the stronger paramagnetic relaxation and lesser paramag-
netic shifting conferred by Mn 31 in the former case [28],
and the absence of paramagnetic shifting in the latter case.
Thus, the extreme chemical shift reflects proximity to, and
possible hyperfine interaction with Fe 21 .
˚
Gln146 is the only Gln with its side chain within 14 A
˚ from Fe 31 [29]. The
of Fe, and its N is only 4.4 A
assignment of the 192 ppm signal to Gln146 is
strengthened by this signal’s accelerated 1 / T 1 relaxation
rate, evinced by its resistance to saturation in a rapidrepetition rate super-WEFT spectrum [25] that partially
saturates the resonances of the other Gln side chain 15 Ns
(Fig. 7).6
Like Gln69 of Fe 21 SOD [23], Gln146 of Fe 21 (Mn)SOD
is strongly affected by Fe 21 . However, the effect is clearly
different as Gln146’s 15 N is paramagnetically shifted by
|79 ppm, almost twice as much as Gln69’s (shifted by |45
ppm). This difference could reflect differences in the
electronic state of Fe 21 , its interactions with its ligands in
general, its interactions with Gln in particular or the
geometry of the active site. The gross similarity of the 1 H
NMR spectra of Fe 21 SOD and Fe 21 (Mn)SOD indicate
similar electronic structures for the Fe 21 in the two
5
Since side chain amides have two Hs bound to a single N, their HSQC
resonances are characterized by two H resonances with identical 15 N
chemical shifts, whereas back bone amides consist of a single H bound to
an N and so occur as independent resonances that do not share a chemical
shift with any other except fortuitously.
6
The 192-ppm signal also appears broader than the other Gln side chain N
signals, indicating that it has faster 1 / T 2 relaxation, but this is not easily
quantified due to the insensitivity of the spectra and the overlap of the
diamagnetic 15 N signals.
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
Fig. 6. 15 N NMR spectra of Fe 21 SOD (top), Fe 21 (Mn)SOD (second),
apo(Mn)SOD (third) and Mn 31 SOD (bottom). 1 H coupled spectra were
collected at 258C at 50.655 MHz on a Unity plus 500 MHz spectrometer in
a 5-mm broadband probe using 908 pulse excitation, 0.125-s acquisition
times, no decoupling and 1-s delays, except for Fe 21 (Mn)SOD which was
collected with a 608 excitation pulse and 0.3-s acquisition time. A total of
59 498 scans were collected for Fe 21 SOD, 102 048 for Fe 21 (Mn)SOD,
9 917 376 for apo(Mn)SOD and 152 368 for Mn 31 SOD. Thirty-Hz line
broadening was applied to all spectra. Fe 21 (Mn)SOD was at pH 7.5 prior
to reduction and the other samples were at pH 7.8 prior to reduction or
oxidation. The chemical shift of Gln69 in Fe 21 SOD is affected by a pK of
8.5 but only moves by 10 ppm total between pH 6 and 10, and moves
very little between pH 7.5 and 7.8.
proteins and the comparable chemical shifts of the ligand
His’ indicate similar degrees of covalency with respect to
the His’ at least [20]. In apparent contrast, the 55%
difference between the two Glns’ 15 N chemical shifts 7 is
striking, and suggests alteration of a specific relation
between Fe 21 and the Gln side chain N.
It is most likely that the very different Gln chemical
shifts reflect different distances from Fe 21 to N in the
(Fe)SOD and (Mn)SOD proteins. Although a structure has
not been reported for Fe 21 (Mn)SOD, the N to Fe 31
˚ [29] vs. 5.1 A
˚ between
distance in Fe 31 (Mn)SOD is 4.4 A
˚ in Fe 31 SOD [2]).
N and Fe 21 in Fe 21 SOD (and 5.0 A
Since dipolar paramagnetic chemical shifting scales as
1 /r 3 where r is the distance to the metal ion, the different
distances to Fe suggest that the chemical shift of Gln146
will be |1.6 times that of Gln69. This accounts for a
substantial fraction of the observed ratio of 1.8. Possible
7
34 /((45179) / 2)555%
15
253
21
Fig. 7. N NMR spectra of Fe (Mn)SOD collected using (top) super
WEFT regime and (bottom) the 908-observe experiment. The 908-observe
1
spectrum was collected without H decoupling as in Fig. 5 and the super
WEFT spectrum [25] employed a 44.5-ms inversion pulse followed by
15
0.6-s delay during which fast-relaxing N signals fully recovered but
most 15 N signals decayed to close to zero magnetization, a 908 observe
1
pulse and 0.125-s acquisition (with H decoupling) followed by a 0.7-s
relaxation delay. Thus, slowly relaxing signals are suppressed relative to
fast relaxing signals. Thirty-Hz line broadening was applied to both
spectra.
displacement of Gln146’s side chain relative to Fe in
Fe(Mn)SOD upon reduction is expected to increase the
distance and decrease the ratio of the inverse cube dis˚ distance increase the ratio drops to
tances (e.g. for a 0.1 A
1.5). The balance of the extra chemical shifting could
reflect slightly different polar and azimuthal angles for the
Gln relative to the Fe 21 susceptibility tensor, and / or a
greater hyperfine contribution to paramagnetic chemical
shift. The latter and the shorter distance to Fe in Fe(Mn)SOD are both suggestive of stronger hydrogen bonding.
Weaker paramagnetic shifting than expected based on
Fe–N distances in Fe 21 SOD does not appear to support
21
reversal of the side chain N and O of Gln69 in Fe SOD.
If the side chain O instead of the NH 2 were closer to Fe 21 ,
˚ (the
then the Fe 21 to N distance would become 6.1 A
distance to O in the reported crystal structure of Fe 21 SOD
[2]). This distance results in a predicted chemical shift
ratio of 2.7, much larger than is observed. Thus, barring
cancellation of distance effects by orientational or scalar
ones, it appears that the Gln side chain retains the same
polarity in the reduced state of FeSOD as in the oxidized
state.
254
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
4. Discussion
The active site Gln is one of a few conserved differences
between Fe- and Mn-specific SODs. The Gln side chain
hydrogen bonds with the coordinated solvent [2,15] thus
tuning the degree of protonation of this ligand. This in turn
tunes the Em of the metal ion. Thus, the nature of the
interaction between the Gln side chain and the active site
metal ion is important to understanding the metal ion
specificity of SODs [7].
To the extent that the active site Gln side chain can
affect the metal ion, it is itself affected by the metal ion.
The 15 N chemical shift is indicative of both electron
density delocalization onto Gln via the hyperfine contribution, and position relative to Fe via the dipolar contribution. The very different chemical shifts of Gln146 and
Gln69 can be substantially explained by the different
distances from their Ns to Fe 21 , but may also include
different hyperfine contributions to chemical shift as well.
Thus, our NMR results extend to solution and the reduced
state an important difference between the crystal structures
of Fe 31 SOD and Fe 31 (Mn)SOD. The significantly shorter
distance between Gln and the metal ion in (Mn)SOD was
not noted in the crystallographic reports, but such positional differences may be more important than oft-noted
differences in the chemical identities of the side chains in
the active sites of (Fe)SOD and (Mn)SOD. A shorter N to
Fe 21 distance in Fe 21 (Mn)SOD (and more hyperfine
shifting) may imply stronger hydrogen bonding with
coordinated solvent in Fe 21 (Mn)SOD than in Fe 21 SOD.
Thus, the system with the higher Em , FeSOD, appears to
have weaker hydrogen bonding between Gln and the
coordinated solvent.
Similarly, mutation of the active site Gln to Asn in
human MnSOD resulted in a longer Mn to amide N
distance and a substantially increased Em [31]. The mutant
active site incorporates an additional solvent molecule
which is presumed to mediate hydrogen bond interactions
between the Asn side chain and coordinated solvent [31].
However, this indirect hydrogen bonding is likely weaker
than direct hydrogen bonding. Thus, in both E. coli and
human SOD, weaker hydrogen bonding is associated with
a higher Em relative to MnSOD. It therefore appears that H
bonding from Gln destabilizes additional electron density
on the metal ion or has the effect of diminishing proton
density around the metal ion. The active site Gln is ideally
positioned to do this by stabilizing coordinated OH 2 over
H 2 O (i.e., by increasing coordinated solvent’s acid dissociation constant).
Eq. (1a) depicts proton and electron uptake in FeSOD,
with the amino acid ligands shown only in one protonation
and oxidation state, for simplicity. Since coordinated
solvent is believed to take up a proton upon metal ion
reduction in MnSOD (Vance and Miller, unpublished) as
well as in FeSOD [32] (diagonal line in Eq. (1a)), the
dissociation constant of the proton acceptor in the reduced
state, Kred , is smaller than that in the oxidized state, Kox
and it is the difference between Kred and Kox that determines the Em at a given H 1 concentration,
(1a)
1
RT Kred 1 [H ]
]
]]]]
Em 5 EAH 1
ln
,
F
Kox 1 [H 1 ]
(1b)
where EAH is the reduction potential relating the protonated
states (1a). The larger Kox and the smaller Kred , the lower
Em should be.
If the difference between the redox tuning of the two
different SOD proteins is to be ascribed to different pK
values for coordinated solvent, then DE, the Em of a metal
ion bound to (Fe)SOD minus its potential when bound to
(Mn)SOD, can be written in terms of four pK values:
0.5 V¯ DE( Fe)SOD – ( Mn)SOD
1
Mn
(K Fe
red 1 [H ]) 3 K ox
5 60 mV log]]]]]]
Mn
1
K Fe
ox (K red 1 [H ])
(2)
where superscripts are used to identify the metal ion
specificity of the SOD protein to which a dissociation
constant pertains. Eq. (2) incorporates the simplification
that Kox 4[H 1 ] for both FeSOD and MnSOD, the difference between the potentials of Fe in the two proteins is
0.46 V (Table 1) [7], and that of Mn is even larger (Vance
Fe
Mn
and Miller, unpublished). Thus, either K Mn
ox .K ox , (K red 1
1
Fe
1
Fe
1
[H ]),(K red 1[H ]) or both. However, K red <[H ] for
Fe
pH,11 [12] so the relative magnitudes of K Mn
red and K red
should be much less important than the relative magnitudes
Fe
of K Mn
ox and K ox . Thus, stronger hydrogen bonding from
Gln to the coordinated solvent could produce the observed
trend in Em values, via its larger effect on the oxidized
state.
For the oxidized state pK values to fully account for the
redox tuning difference between (Fe)SOD and (Mn)SOD,
the pK of H 2 O / OH 2 must be |8 pH units lower in
(Mn)SOD than in (Fe)SOD, when a given metal ion is
bound. An 8 pH unit shift in a pK is somewhat larger than
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
has been reported in proteins so far, but pK shifts of 6 and
5 pH units are known [33]. Therefore, it is possible for this
mechanism to contribute a substantial fraction of the Em
tuning difference between (Fe)SOD and (Mn)SOD.
Since ionization of coordinated solvent has been assigned to a pK of 5.1 in Fe 31 SOD [34], a pK substantially
lower than 5 and as low as 23.2 is predicted for
Fe 31 (Mn)SOD. Such a pK is unlikely to be experimentally
accessible. Similarly, coordinated solvent’s pK in
Fe 21 (Mn)SOD should be very high, consistent with the
absence of a pK ascribable to coordinated solvent in the
range of pH 6–11 [20]. Finally, the predicted pK values
also indicate that proton uptake will always accompany
reduction in Fe(Mn)SOD, justifying the application of the
above equations to this case.
The distance from Gln’s N to the metal ion is a property
of the protein, not the result of the metal ion identity, since
˚ are observed in
shorter distances of 4.4 and 4.6 A
˚
Fe(Mn)SOD and MnSOD, respectively, than the 5.1-A
distance in FeSOD. This is a result of both the different
backbone origins of the Gln in (Fe)SOD and (Mn)SOD,
and the positioning of Gln’s side chain by the active site
hydrogen bond network. The former is reversed in our
G77Q,Q146A mutant, and results in restoration of small
but significant Fe-supported activity.
Our G77Q,Q146A mutant of MnSOD retains the active
site Gln, but at the backbone location characteristic of
FeSOD. Thus, it represents a very conservative but informative pair of mutations. While a crystal structure is not
yet available, we can begin evaluating the integrity of the
mutated active site by spectroscopic means. Subtle effects
associated with nuclear movements too small to be unambiguously resolved by protein crystallography, but
associated with sufficient distortion of the metal ion
electronic structure and geometry to significantly affect
activity, can be detected immediately by EPR, NMR and
other spectroscopic methods that probe the metal ion
electronics directly. Indeed, the altered metal ion coordination geometry observed in Fe(Mn)SOD crystallographically [29], was first emphasized based on EPR [20]. Moreover
the fact that only one of two crystallographic active sites
displayed the altered geometry is also explained by the
EPR data which showed that the novel coordination
geometry is characterized by a pK of 8.4 [20], very close
to the crystallization pH of 8.5 [29].
Our current results demonstrate conferral of low but
significant Fe-supported activity in (Mn)SOD upon relocation of the active site Gln. This and the work of Yamakura
[18] are the first experimental validation of the proposal of
Parker et al. [35] that the Gln is a determinant of metal ion
specificity. However, since less than 10% WT Fe-supported activity is obtained, the Gln is clearly not the sole
determinant of metal ion specificity (also [18]).
Our spectra all demonstrate that the mutant retains basic
active site integrity and metal ion binding, indicating that
the low activity is unlikely to be due to gross structural
255
factors. However, the fact that the mutant’s active site
resembles that of the (Mn)SOD parent more than the
(Fe)SOD active site which the mutant’s Gln location
emulates indicates that additional amino acids contribute to
metal ion electronics and geometry. These residues most
likely include the hydrogen bonding partners of Gln since
these position and orient the Gln side chain relative to
coordinated solvent, and they most likely include the other
‘missing’ determinants of metal ion activity. These residues may be identified by incorporating more amino acid
substitutions into our G77Q,Q146A-(Mn)SOD. It may also
be possible to correlate the subtle spectroscopic differences
between G77Q,Q146A-(Mn)SOD, (Mn)SOD and (Fe)SOD
with activity, and thus obtain insight into electronic bases
for Em tuning and activity. Since the inactivity of WT
Fe(Mn)SOD can be explained by an anomalously low Em
[7] below that required for O ?2
oxidation, we anticipate
2
that Fe-G77Q,Q146A-(Mn)SOD’s small but significant
activity will be accompanied by an increase in Em above
that of Fe(Mn)SOD.
Although the Fe-supported activity of Fe-mut-(Mn)SOD
is low compared to WT FeSOD and MnSOD, it is
comparable to the activities reported for cambialistic
SODs. The SOD from Methylomonas J. has Mn-supported
activity of 2300 u / mg protein and Fe-supported activity of
100 u / mg protein after correction for incomplete metal ion
incorporation [36], that of Propionibacterium shermanii
has activities of 900 and 700, respectively [37], Mycobacterium smegmatis SOD has activities of 5000 and 300 [38]
and Porphyromonas gingivalis SOD has Mn- and Fesupported activities of 1200 and 1000 [39]. Our Fe-supported activity of at least 450 on a per Fe basis thus falls
well within the range of activities considered biologically
significant, and our mutation in effect converts MnSOD
into a cambialistic SOD.
5. Notation
1D, one-dimensional; DPPH, a,a9-diphenyl-b-picryl hydrazine; DSS, 4,4-dimethyl 4-silapentane sodium sulfonate; EPR, electron paramagnetic resonance spectroscopy;
FeSOD, Fe-containing superoxide dismutase; (Fe)SOD, the
protein of FeSOD; Fe(Mn)SOD, Fe-containing (Mn)SOD
protein; Fe-mut-(Mn)SOD, Fe-containing G77Q,Q146A(Mn)SOD protein; G77Q,Q146A-MnSOD, mutant
MnSOD with glycine 77 mutated to glutamine and
glutamine 146 mutated to alanine; MnSOD, Mn-containing
superoxide dismutase; (Mn)SOD, the protein of MnSOD;
Mn(Fe)SOD, Mn-containing (Fe)SOD protein; mutMnSOD, G77Q,Q146A-MnSOD, mutant MnSOD with
glycine 77 mutated to glutamine and glutamine 146
mutated to alanine; NAS, non-active site; NHE, normal
hydrogen electrode; NMR, nuclear magnetic resonance
spectroscopy; PCR, polymerase chain reaction; SDS–
256
A.L. Schwartz et al. / Journal of Inorganic Biochemistry 80 (2000) 247 – 256
PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; SOD, superoxide dismutase; WEFT, watereliminated Fourier transform; WT, wild-type, not mutated
Acknowledgements
Acknowledgment is made to the donors of The Petroleum Research Fund, administered by the A.C.S., for
partial support of this research under ACS-PRF 33266AC4,3. A.-F.M. is pleased to thank the N.S.F. for financial
support (MCB9728793) and N.I.H. for generous financial
aid in acquiring an NMR spectrometer (RR 06468). AFM
also thanks the reviewers of the manuscript for exceptionally thoughtful reviews.
References
[1] W.C. Stallings, K.A. Pattridge, R.K. Strong, M.L. Ludwig, J. Biol.
Chem. 259 (1984) 10695–10699.
[2] M.S. Lah, M.M. Dixon, K.A. Pattridge, W.C. Stallings, J.A. Fee,
M.L. Ludwig, Biochemistry 34 (1995) 1646–1660.
[3] M.W. Parker, C.C.F. Blake, FEBS Lett. 229 (1988) 377–382.
[4] M.W. Smith, R.F. Doolittle, J. Mol. Evol. 34 (1992) 175–184.
[5] D.E. Ose, I. Fridovich, J. Biol. Chem. 251 (1976) 1217–1218.
[6] F. Yamakura, J. Biochem. 83 (1978) 849–857.
[7] C.K. Vance, A.-F. Miller, J. Am. Chem. Soc. 120 (1998) 461–467.
[8] S. Yamano, T. Maruyama, J. Biochem. 125 (1999) 186–193.
[9] M.W. Parker, C.C.F. Blake, J. Mol. Biol. 199 (1988) 649–661.
[10] T. Hunter, K. Ikebukuro, W.H. Bannister, J.V. Bannister, G.J. Hunter,
Biochemistry 36 (1997) 4925–4933.
[11] D.L. Sorkin, D.K. Duong, A.-F. Miller, Biochemistry 36 (1997)
8202–8208.
[12] D.L. Sorkin, A.-F. Miller, Biochemistry 36 (1997) 4916–4924.
[13] M.M. Whittaker, J.W. Whittaker, Biochemistry 36 (1997) 8923–
8931.
[14] Y. Guan, M.J. Hickey, G.E.O. Borgstahl, R.A. Hallewell, J.R.
Lepock, D. O’Connor, Y. Hsieh, H.S. Nick, D.N. Silverman, J.A.
Tainer, Biochemistry 37 (1998) 4722–4730.
[15] R.A. Edwards, H.M. Baker, G.B. Jameson, M.M. Whittaker, J.W.
Whittaker, E.N. Baker, J. Bio. Inorg. Chem. 3 (1998) 161–171.
[16] C.L. Fisher, J.-L. Chen, J. Li, D. Bashford, L. Noodleman, J. Phys.
Chem. 100 (1996) 13498–13505.
[17] K. Bunting, J.B. Cooper, M.O. Badasso, I.J. Tickle, M. Newton, S.P.
Wood, Y. Zhang, D. Young, Eur. J. Biochem. 251 (1998) 795–803.
[18] B.Y. Hiraoka, F. Yamakura, S. Sugio, K. Nakayama, Biochem. J. 345
(2000) 345–350.
[19] J.W. Whittaker, M.M. Whittaker, J. Am. Chem. Soc. 113 (1991)
5528–5540.
[20] C.K. Vance, A.-F. Miller, Biochemistry 37 (1998) 5518–5527.
[21] S. Barik, Methods in molecular biology; PCR cloning protocols:
from molecular cloning to genetic engineering, in: B.A. White (Ed.),
Mutagenesis and Gene Fusion by Megaprimer PCR, Vol. 67,
Humana Press, 1989, pp. 173–182.
[22] P. Carter, Anal. Biochem. 40 (1971) 450–458.
[23] C.K. Vance, Y.M. Kang, A.-F. Miller, J. Biomol. NMR 9 (1997)
201–206.
[24] J. Parker, J.D. Friesen, Mol. Gen. Genet. 177 (1980) 439–445.
[25] T. Inubishi, E.T. Becker, J. Magn. Reson. 51 (1983) 128–133.
[26] J.M. McCord, I. Fridovich, J. Biol. Chem. 244 (1969) 6049–6055.
[27] S. Vathyam, R.A. Byrd, A.-F. Miller, J. Biomol. NMR 14 (1999)
293–294.
[28] L. Bertini, C. Luchinat, NMR of Paramagnetic Molecules in
Biological Systems, Benjamin Cummings, 1986.
[29] R.A. Edwards, M.M. Whittaker, J.W. Whittaker, G.B. Jameson, E.N.
Baker, J. Am. Chem. Soc. 120 (1998) 9684–9685.
[30] B.-H. Oh, J.L. Markley, Biochemistry 29 (1990) 4012–4017.
[31] H. Hsieh, Y. Guan, C. Tu, P.J. Bratt, A. Angerhofer, J.R. Lepock,
M.J. Hickey, J.A. Tainer, H.S. Nick, D.N. Silverman, Biochemistry
37 (1998) 4731–4739.
[32] C. Bull, J.A. Fee, J. Am. Chem. Soc. 107 (1985) 3295–3304.
[33] D.E. Anderson, W.J. Becktel, F.W. Dahlquist, Biochemistry 29
(1990) 2403–2408.
[34] J.A. Fee, G.J. McClune, A.C. Lees, R. Zidovetzki, I. Pecht, Israel J.
Chem. 21 (1981) 54–58.
[35] M.W. Parker, C.C.F. Blake, D. Barra, F. Bossa, M.E. Schinina, W.H.
Bannister, J.V. Bannister, Protein Eng. 1 (1987) 393–400.
[36] T. Matsumoto, K. Terauchi, T. Isobe, K. Matsuoka, F. Yamakura,
Biochemistry 30 (1991) 3210–3216.
[37] B. Meier, D. Barra, F. Bossa, L. Calabrese, G. Rotilio, J. Biol.
Chem. 257 (1982) 13977–13980.
[38] F. Yamakura, K. Kobayashi, S. Tagawa, A. Morita, T. Imai, D.
Ohmori, T. Matsumoto, Biochem. Mol. Biol. Int. 36 (1995) 233–
240.
[39] F. Yamakura, R.L. Rardin, G.A. Petsko, D. Ringe, B.Y. Hiraoka, K.
Nakayama, T. Fujimura, H. Taka, K. Murayama, Eur. J. Biochem.
253 (1998) 49–56.
[40] L.E. Kay, P. Keifer, T. Saarinen, J. Am. Chem. Soc. 114 (1992)
10663–10665.
[41] F. Delaglio, S. Grzesiek, G.W. Vuister, G. Zhu, J. Pfeifer, A. Bax, J.
Biomol. NMR 6 (1995) 277–293.
[42] G.C. Levy, R.L. Lichter, in: Nitrogen-15 Nuclear Magnetic Resonance Spectroscopy, Wiley, New York, 1979.