EPR of Cu2þPrion Protein Constructs at 2 GHz Using the gtRegion
to Characterize Nitrogen Ligation
James S. Hyde,†* Brian Bennett,†Eric D. Walter,‡Glenn L. Millhauser,‡Jason W. Sidabras,†
and William E. Antholine†
†Department of Biophysics, Medical College of Wisconsin, Milwaukee, Wisconsin; and‡Department of Chemistry and Biochemistry,
University of California, Santa Cruz, California
provides at most three imidazole ligands to each copper ion to form a square-planar Cu2þcomplex. This work is concerned with
identification of the fourth ligand. A new (to our knowledge) electron paramagnetic resonance method based on analysis of the
intense features of the electron paramagnetic resonance spectrum in the gtregion at 2 GHz is introduced to distinguish between
three and four nitrogen ligands. The methodology was established by studies of a model system consisting of histidine imidazole
ligation to Cu2þ. In this spectral region at 2 GHz (S-band), g-strain and broadening from the possible rhombic character of the
Zeeman interaction are small. The most intense line is identified with the MI¼ þ1/2 extra absorption peak. Spectral simulation
demonstrated that this peak is insensitive to cupric Axand Ayhyperfine interaction. The spectral region to the high-field side of
this peak is uncluttered and suitable for analysis of nitrogen superhyperfine couplings to determine the number of nitrogens. The
spectral region to the low-field side of the intense extra absorption peak in the gtpart of the spectrum is sensitive to the rhombic
distortion parameters Axand Ay. Application of the method to the prion protein system indicates that two species are present and
that the dominant species contains four nitrogen ligands. A new loop-gap microwave resonator is described that contains ~1 mL
of frozen sample.
A double octarepeat prion protein construct, which has two histidines, mixed with copper sulfate in a 3:2 molar ratio
This work describes an improved electron paramagnetic
resonance (EPR) methodology for interpreting spectra from
Cu2þsquare-planar complexes to determine the number of
nitrogen ligands. All EPR spectra were obtained at low
microwave frequencies—either 2 or 3.3 GHz—with the
use of loop-gap resonators (LGR) (1,2). The method was
applied to a standard sample, copper histidine, which has
four equivalent nitrogen ligands. The study was then
extended to a copper-binding domain derived from the prion
protein (PrP), where the number of nitrogen ligands is uncer-
Rist et al. (3) and Rist and Hyde (4) showed that the values
of the Zeeman and hyperfine parameters for a square-planar
copper complex are strongly affected by the molecular
environment, and that rhombic distortion can occur. Further-
more, in powders and glasses containing these complexes,
line broadening (termed g-strain and A-strain) is always
encountered. In a molecular orbital formulation, broadening
arises from the distributions of bonding parameters. Froncisz
and Hyde (5,6) noted that the gk‘‘turning points’’ in the
derivative-like spectra from square-planar copper complexes
can be considered identical to pure absorption spectra from
single crystals in the parallel orientation, as was previously
pointed out for other highly anisotropic systems (7,8). Fron-
cisz and Hyde discovered that distributions of Akand gkat
these turning points, the so-called g- and A-strains, were
correlated, tending to cancel for MI¼ ?1/2 or ?3/2 and to
add for þ1/2 and þ3/2. They introduced the term 3 for the
correlation coefficient. If this term is greater than zero, the
gkMI¼ ?1/2 feature always exhibits the best resolution.
Their calculations indicated that the spectral resolution of
superhyperfine couplings to ligand nitrogen nuclei is
expected to be optimal at a microwave frequency of
2 GHz, which is near the lower end of the S-band range.
At this frequency, the g- and A-strain distributions cancel
to a degree that is determined by 3, and the number of
nitrogen ligands can, it was hypothesized, be determined.
Froncisz and Aisen (9), in a study of monocupric transferrin,
provided the first experimental demonstration of these ideas.
This theory was subsequently applied to numerous copper
systems (10–12) in studies that addressed copper binding
in PrPs. In those studies, the microwave frequency was
~3.3 GHz, which was a compromise between optimal reso-
lution and reasonable signal/noise ratio (SNR).
This methodology was found to be robust in distinguish-
ing between the cases of 0, 1, 2, or 3 nitrogen ligands, but
distinguishing between three and four nitrogen ligands has
proven difficult. In a simple stick diagram, coupling to three
equivalent nitrogen nuclei results in seven superhyperfine
lines of relative intensities 1, 3, 6, 7, 6, 3, and 1, whereas
for four nitrogen nuclei, the relative intensities of the nine-
line pattern are 1, 4, 10, 16, 19, 16, 10, 4, and 1. It is noted
that the nitrogen hyperfine coupling is nearly isotropic (4),
and stick-diagram analysis is therefore appropriate. The
SNRs have been inadequate for reliable detection of the
Submitted November 3, 2008, and accepted for publication January 26,
Editor: David D. Thomas.
? 2009 by the Biophysical Society
3354 Biophysical Journal Volume 96April 2009 3354–3362
weakest lines, and the relative intensities of the three most
intense lines are nearly the same (6/7 compared with 16/19)
for three and four nitrogen nuclei, respectively. Hyde et al.
(13) pointed out that the ratios of signal intensities of the
second and third lines from the center to the intensity of
the center line can be diagnostic in determining whether
three or four nitrogens are present.
The PrP binds Cu2þin its N-terminal octarepeat domain,
comprised of tandem PHGGGWGQ octarepeat segments.
Recent work from our laboratories demonstrates the exis-
tence of three distinct copper-binding modes depending on
the precise ratio of Cu2þto octarepeat concentrations (12).
At high occupancy, each HGGGW segment, within an
individual octapeptide, takes up a single Cu2þequivalent.
Equatorial coordination involves the histidine imidazole
and deprotonated amide nitrogens from the two glycines
that immediately follow the histidine. At low copper occu-
pancy, coordination is from imidazole nitrogens involving
three or four histidine side chains. These two binding modes,
referred to as component 1 and component 3, respectively
(see Fig. 1), are well characterized. Component 2 coordina-
tion is observed at intermediate copper concentrations;
however, the precise details of this binding mode are less
certain. EPR investigations using selectively N-methylated
peptides suggest that there are two nitrogens at equatorial
coordination sites: one from the histidine imidazole, and
one from the histidine backbone nitrogen. However, the
3.3 GHz EPR spectra derived from these constructs were
not well resolved and thus were ambiguous regarding the
exact number of equatorial nitrogens. In addition, the prion
fragment with two octarepeats has not been studied in
detail. In this study, we applied 2 GHz EPR to Cu2þ-PrP
constructs with the goal of obtaining greater resolution and
detail with regard to the equatorial coordination environ-
ment. This was a joint project conducted by the prion-
construct team at the University of California, Santa Cruz,
and the microwave resonator team at the Medical College
of Wisconsin. At the outset of the project, the sample volume
was set at 1 mL. Sample preparation and development of
a purpose-built 2 GHz LGR suitable for a 1 mL sample
were carried out at the same time.
It was expected that the increase in sample volume from
70 mL to 1 mL would more than compensate for the loss
of sensitivity at 2 GHz relative to 3.3 GHz, and that
improved resolution of the gkMI¼ ?1/2 feature would
permit determination of the number of nitrogen ligands
(three or four) for the particular prion construct of interest.
As it turned out, this expectation failed.
However, during the course of the research, what we
consider a new, successful way to determine the number of
nitrogen ligands for square-planar Cu2þcomplexes was
found based on analysis of the gtpart of the spectrum.
This new method is the primary subject of the study pre-
sented here. The idea is illustrated in Fig. 2. The top row
shows ideal simulated pure absorption Cu2þspectra at 1,
2, 3, 6, and 9.5 GHz, and the lower rows show the spectral
manifolds assigned to the four copper nuclear quantum
numbers. The integrated areas of spectra in the first row
are identical, and the integrated areas of spectral manifolds
assigned to each of the four copper nuclear transitions are
each 1/4 of the integrated area of spectra in the first row.
All spectra were simulated using the same linewidth, all
ligands were assumed to have zero nuclear spin, and only
one copper isotope was assumed to be present. Axial
FIGURE 1Copper complexes of the prion octarepeat peptide system.
a function of frequency together with simulations of the four copper hyper-
fine manifolds that form the spectra (see text for details).
Simulated pure absorption square-planar Cu2þspectra as
Biophysical Journal 96(8) 3354–3362
EPR of Cu2þ-Prion at 2 GHz3355
symmetry was assumed, with the principal axes of the
Zeeman and Cu-hyperfine interactions colinear. The radio-
frequency (RF) magnetic field was assumed to be strictly
perpendicular to the applied field. Spectral features of
interest in the following discussion are lettered in Fig. 2
Intense features are seen in the so-called gtregion of the
spectra. These features have variously been called ‘‘over-
shoot’’ lines, extra absorption (EA) peaks, or hyperfine
anomaly lines in the literature. They are associated with
MI¼ þ1/2 and þ3/2 spectral manifolds. The earliest expla-
nation of these features seems to be that of Ovchinnikov and
Konstaninov (14), who introduced the EA notation. As is
apparent, these features arise from a buildup of intensity
arising from the interplay of g- and A-anisotropies. At 6
and 9.5 GHz, two EA transitions are seen, whereas at 2
and 3 GHz, there is just one. In this study, we focus on the
feature (labeled a) seen at 2 GHz. To the high-field side of
a, there is a broad, featureless region of nearly zero intensity
in a derivative-like display (labeled b). Region b is favorable
for the observation of superhyperfine coupling to nitrogen
Variation of these couplings occurs if the nitrogen ligands
are inequivalent, which affects resolution. Using electron-
nuclear double resonance (ENDOR), Rist and Hyde (4)
found that the nitrogen nuclear superhyperfine tensor
exhibits a small degree of rhombicity, which can be expected
to be unresolved in EPR but will contribute to broadening of
Cancellation of the g- and A-strains occurs to varying
degrees in the parallel features e and f arising from
MI¼ ?1/2 and ?3/2 transitions. For all other transitions
(parallel as well as perpendicular), the g- and A-strains
add. Atis much smaller than Ak, and the spectral features
associated with Atwill, we expect, exhibit correspondingly
reduced strain broadening. At sufficiently low microwave
frequencies, the g-strain no longer dominates the residual
linewidth. Thus, as an alternative to analyzing features e
and f to count nitrogens, we propose a strategy for analysis
in the more intense perpendicular part of the spectrum. We
seek a sufficiently low microwave frequency to avoid domi-
nant g-strain broadening and to identify a spectral feature
with minimal overlap that is dominated only by residual
Atstrain. Our working hypothesis is that feature a and the
high-field side of this feature, region b, satisfy these require-
ments at 2 GHz. Peak a is a superposition of peaks c and d,
although peak c is strongly dominant. At this frequency, we
expect to see, without overlap, the intense center superhyper-
fine line and either three or four transitions to the high-field
side with intensities that match (by computer fitting) the
assignment to either three or four nitrogen ligands.
In the course of developing the methodology introduced in
this work, we considered the possibility that some amount of
rhombicity could be present, requiring that gtand Atbe
divided into components gx, gy, Ax, and Ay, respectively.
Of course, as the microwave frequency is reduced, Zeeman
rhombic distortion becomes increasingly difficult to detect.
We carried out simulations similar to those of Fig. 2, using
the parameters given by Rist et al. (3). At a microwave
frequency of 2 GHz, the low-field side of the EA transition
(Fig. 2 a) was found to be sensitive to hyperfine rhombicity,
whereas the high-field side was insensitive.
MATERIALS AND METHODS
Resonator design and construction
LGR design was carried out with the use of Ansoft (Pittsburg, PA) High
Frequency Structure Simulator (HFSS, version 10.1) software, which
provides full-wave solutions to Maxwell’s equations and also can find
resonant frequencies using an eigenmode solver. Ansoft Maxwell 3D
(version 11), which is based on a quasi-static approximation to Maxwell’s
equations, was used to determine eddy currents arising from magnetic field
modulation. Simulations were performed with a sample that had the dielec-
tric constant of ice (3r¼ 3.2 þ j0.00128) (15) in the appropriate quartz tube
(3r¼ 3.78) (see Table 1 for sample tube and resonator dimensions). Values
are also provided for the resonator used to acquire spectra at 3.3 GHz (see
Fig. 3). Both LGRs were of the one-loop/one-gap type (1) (see Fig. 3).
The dimensions of the resonator were adjusted until the resonator
frequency with the sample in place was 2 GHz. Modulation slots were
then added to the design, and simulation of field modulation penetration
was carried out using Ansoft Maxwell 3D.
An assembly drawing using three-dimensional computer-aided design
was prepared using AutoDesk (San Rafael, CA) Inventor 10, and fabrication
drawings of each part were made using the symbolic language of geometric
dimensioning and tolerancing. Most parts were made using wire electrical
discharge machining. In this technique, both the LGR gap and field
modulation slots can be as small as 0.05 mm, with precisional tolerances
of 0.001 mm. The resonator body was made from high-purity silver (see
Fig. 4). Table 1 gives measured and calculated parameters comparing the
resonator of Fig. 4 with the structure that has been used in this laboratory
for more than 20 years (2).
Relative sensitivities were determined from simulations of electromag-
netic fields under both the assumption of constant incident power and the
assumption of constant peak value of H1(the RF magnetic field in the
rotating frame) at the sample. The latter condition is equivalent to assuming
that T1 (the electron spin-lattice relaxation time) is independent of
TABLE 1Characteristics of 3.3 and 2 GHz resonators
Inner radius, R
Outer radius, r
Gap width, w
Gap separation, t
Resonator height, Z
Sample tube I.D.
Sample wall thickness
Filling factor, h
Shield inner diameter
Biophysical Journal 96(8) 3354–3362
3356 Hyde et al.
microwave frequency and that spectra are acquired in all cases at the same
degree of microwave power saturation.
Relative sensitivities at the same level of incident power can be calculated
using the data in Table 1 and the theory outlined by Feher (16); the
subscripts refer to resonators a and b of Fig. 3:
The filling factors in Table 1 were calculated using HFSS. Under
this condition, the loss in sensitivity because of the linear dependence of
the Boltzmann factor on microwave frequency is compensated for by the
increase in filling factor, and the resultant sensitivity is nearly the same at
3.3 and 2 GHz. This increase arises because a lower fraction of the 2 GHz
resonator is occupied by sample-cell walls.
Relative sensitivities at the same level of microwave field can be calcu-
latedusingthe theoryas outlinedbyHydeandFroncisz(2).Here,the sample
volume is V and the RF field is assumed to be uniform over the sample:
The increase in sensitivity at 2 GHz under the condition of constant RF field
at the sample arises from increased sample volume. From a practical
perspective, the spectra reported here have about the same SNR at 2 and
EPR spectra were calculated with the XSophe suite of programs (17), em-
ploying matrix diagonalization and a mosaic misorientation linewidth
model-based algorithm (17,18). Matrix diagonalization was required
because the condition of applicability of perturbation theory for the simula-
tion of Cu2þceases to hold at frequencies below ~2 GHz (14). Correlated
strains in g and A (5,6) were included, largely to account for the lineshapes
of the gkfeatures of the Cu2þspectra (19). The linewidth is expressed, in
frequency terms (20,21), by the formula
sRiði ¼ x;y;xÞ
are the residual linewidths due to unresolved hyperfine coupling and other
sources, and sgiand sAiare the widths of Gaussian distributions of giand
Simulations were carried out over field ranges of 600–1000 G using
4096 points and sampling 240 partitions, and four segments of the Sophe
grid (17) (see Table 2 for simulation parameters).
A copper histidine sample, which is known to have four equivalent in-plane
nitrogen ligands, was prepared to test the use of the gtspectral region to
determine the number of nitrogens. The sample was prepared by adding
63Cu (2 mM final concentration; Cambridge Isotope Laboratories, Andover,
MA) to excess (40 mM) histidine (enzyme grade; Fisher BioReagents,
Fairlawn, NJ) in 20% glycerol.
was prepared using fluorenylmethoxycarbonyl (Fmoc) methods as described
previously (12). The sequence was KKRPKPHGGGWGQPHGGGWGQ.
This section of the sequence is common to all placental mammals except
mice. The construct concentration was 600 mM in D2O/NEM buffer
(25 mM ethylmorpholine) at pH 6.5 (pD ¼ 6.1) with 20% glycerol. The
copper concentration was 400 mM of63Cu (copper sulfate). Thin-walled
(0.5 mm) quartz sample tubes with inner dimensions of 3 mm for 3.3 GHz
experiments and 9 mm for 2 GHz experiments were used.
The S-band spectrometer consists of a Varian E-9 EPR spectrometer fitted
with a microwave bridge of local design and construction operating between
passes through the coupling loop. For adjustment of the coupling, the
resonator (with sample) moves and the coupling loop is fixed. The shield
is fiberglass-epoxy-plated on the inside. Temperature-regulated gas flows
over the assembly for temperature control.
Cutaway drawing of the 2 GHz LGR assembly. The sample
for additional dimensions.)
LGR outline drawings: (a) 2 GHz; (b) 3.3 GHz. (See Table 1
Biophysical Journal 96(8) 3354–3362
EPR of Cu2þ-Prion at 2 GHz3357
1.98 and 4 GHz. A Varian flow system was used to control the temperature.
The EPR facilities are located at the National Biomedical EPR Center at the
Medical College of Wisconsin (Milwaukee, WI), sponsored by the National
Institutes of Health.
The spectrometer conditions for 2 GHz spectra for copper histidine were
as follows: microwave frequency, 1.8922 GHz; microwave power, 2.8 mW;
modulation frequency, 100 kHz; time constant, 0.128 s; modulation
amplitude, 3 G; scan time, 4 min; four scans averaged for 1000 G scans;
The spectrometer conditions for 2 GHz spectra for the cupric complex of
PrP(23–27, 60–75), Cu-KKRPK(PHGGGWGQ)2, were as follows: micro-
wave frequency, 1.925 GHz; microwave power, 10.5 mW; modulation
frequency, 100 kHz; time constant, 0.128 s; modulation amplitude, 7 G;
scan time, 4 min; five scans averaged across 1000 G scans; and ?140?C.
Corresponding values at 3.3 GHz were the same except for the following:
microwave power, 2.2 mW; modulation amplitude, 7 G; scan time, 1 min;
and 18 scans averaged across 1000 G.
The EPR spectrum of Cu2þ-histidine (Cu-His) exhibits
a resolved superhyperfine pattern at the MI¼ ?1/2 gkline,
centered at around 510 G (Fig. 5 a; the MI¼ ?3/2 line is not
shown). Seven lines would be expected for three nitrogen
ligands of Cu2þ, and nine lines for four nitrogens; the SNR
of the experimental spectrum is too poor to distinguish these
possibilities by direct observation. The pattern at gk(MI¼
?1/2) contains a center line (labeled ‘‘0’’) and therefore
were also observed in the gtregion from ~620 to 740 G.
Computer simulations of the experimental spectrum are
presented as traces b–e of Fig. 5, for which the number of
nitrogen ligands and the hyperfine interaction with63Cu in
the x-y plane were varied. Positions of superhyperfine
nitrogen transitions for the MI¼ ?1/2 gkfeature, and also
for the high-field side of the EA line, which is feature a in
Fig. 2, are indicated in Fig. 5. Fig. 6 shows, in expanded
form, the fits that have been obtained for the superhyperfine
spectra of Cu-His for three and four nitrogen ligands. Fig. 6
a is for the MI¼ ?1/2 line in the parallel region of the spec-
trum, and Fig. 6 b is a fit to the superhyperfine pattern on the
high-field side of the perpendicular region of the spectrum.
Even with the aid of computer simulation, determination of
the number of nitrogens using the gkregion alone is not
convincing. However, the outermost line on the high-field
side in the gtregion (line 4 of Fig. 6 b) is observed in
both the experimental spectrum and in the simulation with
four equivalent nitrogen ligands, but not in the simulation
with three. In addition, the ratio of the heights of line 2 to
line 0 in the gtregion is diagnostic. This ratio is sensitive
to the assignment of ligation to either three or four nitrogens,
and the fit is much better for four than for three.
TABLE 2 Simulation parameters
CuIm at 1.896 GHz
gi (i ¼ x, y, z)¼ 2.056, 2.056, 2.261
jAi(63Cu)j ¼ 2.0, 2.9, 18.5 ? 10?3cm?1
jAi(14N)j ¼ 1.43, 1.43, 1.27 ? 10?3cm?1
Linewidth, sRi¼ 0.50, 0.50, 0.25 ? 10?3cm?1
g-strain, sgi/gi¼ 0.0007, 0.0007, 0.0005
A-strain, sAi/Ai¼ ?0.5, ?0.5, ?0.8 ? 10?3cm?1
Model simulations (1–9.5 GHz)
gi¼ 2.079, 2.079, 2.298
jAi(63Cu)j ¼ 2.4, 2.4, 18.5 ? 10?3cm?1
Linewidth, sRi¼ 0.8, 0.8, 0.8 ? 10?3cm?1
Prion parameters (1.949 GHz)
gi¼ 2.061, 2.061, 2.248
jAi(63Cu)j ¼ 2.8, 2.8, 17.2 ? 10?3cm?1
jAi(14N)j ¼ 1.34, 1.34, 1.20 ? 10?3cm?1
Linewidth, sRi¼ 0.70, 0.70, 0.30 ? 10?3cm?1
g-strain, sgi/gi¼ 0.0030, 0.0030, 0.0005
A-strain, sAi/Ai¼ 0.5, 0.5, ?1.0 ? 10?3cm?1
nitrogen ligands and axial A(63Cu) (At¼ 2.4 ? 10?3cm?1). (c) Simulation
for four nitrogen ligands and axial A(63Cu) (At¼ 2.4 ? 10?3cm?1).
(d) Simulationfor three nitrogen
(Ax¼ 2.0 ? 10?3cm?1, Ay¼ 2.9 ? 10?3cm?1). (e) Simulation for
four nitrogen ligands and rhombic A(63Cu) (Ax ¼ 2.0 ? 10?3cm?1,
Ay¼ 2.9 ? 10?3cm?1). (f–i) Residuals. In traces b–e, the computed spectra
(bold lines) are overlaid on the experimental spectra (thin lines). Labeled,
dashed lines show the superhyperfine transitions used to determine the
number of ligands (lines 0–4) and to aid refinement of the Axand Ay
hyperfine terms (lines ?3 to ?6). Note the much improved fit in the range
of 550–620 G when the rhombicity of the copper hyperfine tensor is
(a) EPRspectrum of Cu-His at 2 GHz. (b) Simulationfor three
Biophysical Journal 96(8) 3354–3362
3358 Hyde et al.
In this nearly ideal system, the SNR in the parallel region
remains insufficient to distinguish between three and four
nitrogens, although this might be possible using more exten-
sive signal averaging. The known assignment to four nitro-
gens is clearly supported in the perpendicular region of the
spectrum by the analysis presented here. The experimental
evidence for line 4 and the substantially improved fits to
the intensities of lines 2 and 3 are judged to be definitive.
Residuals g and i in Fig. 5, which are for four nitrogens,
are substantially reduced in the range of 720–750 G relative
to f and h, which are for three nitrogens. This example
supports the hypothesis that spectral fitting of the gtregion
at 2 GHz can be used to determine the number of nitrogen
Although the fit of the nitrogen superhyperfine features on
the high-field side of the MI¼ 1/2 EA line seen in Fig. 5 c for
four equivalent nitrogens is very good, the fit of spectral
features on the low-field side of the EA line fails. In partic-
ular, see lines labeled ?5 and ?6. The only remaining free
parameter within the assumptions of the model is At. Spec-
tral features on the low-field side of the EA line were found
to be quite sensitive to this parameter, but the fits still
remained unsatisfactory under the assumption of an axial
copper hyperfine tensor. Therefore, the decision was made
to add a small amount of rhombicity to the model. Parame-
ters Axand Aywere swept, and a convincing fit of lines
labeled ?3 to ?6 was achieved (Figs. 5 e and 6 b). (See
residuals h and i compared with f and g in Fig. 5.) The spectra
were insensitive to rhombic distortion of the Zeeman tensor,
presumably because of the low magnetic field required for
2 GHz EPR.
The lines on the low-field side of the MI¼ þ1/2 EA line
arise from a superposition of the superhyperfine pattern that
is centered on the EA line onto patterns from other hyperfine
manifolds (see Fig. 2). The perpendicular feature of the
MI¼ ?1/2 manifold is prominent among the features of
these other manifolds. The resultant overlap of nitrogen
superhyperfine lines from different copper hyperfine mani-
folds is very sensitive to the copper hyperfine values. The
existence of overlapping superhyperfine lines makes the
2 GHz spectrum sensitive to hyperfine coupling. The molec-
ular structure that gives rise to the observed rhombicity is
unknown, but is thought to arise from steric effects of the
bulky ligand groups.
Thus, we propose that in a sample of unknown nitrogen
ligation, the most intense line in the gtregion of the spec-
trum should first be located and the number of well-resolved
lines to the high-field side in this pattern should then be
counted. If the center line is designated as ‘‘0’’, four more
lines are resolved in the case of four nearly equivalent
nitrogen donor atoms in the equatorial plane. Using this
analysis, the number of nitrogen donor atoms is determined
from the intense lines in the perpendicular spectral region
sis can be extended, if desired, to the copper hyperfine
couplings in the perpendicular orientation.
Cu2þplus excess KKRPK(PHGGGWGQ)2
The increase in spectral resolution of the ?1/2 gkfeature of
the PrP construct at 1.925 GHz compared with 3.275 GHz,
due to greater cancellation of the strain terms, is readily
evident (Fig. 7). However, the SNR is not sufficient to distin-
guish a nine-line pattern from a seven-line pattern. The reso-
lution of the superhyperfine couplings in the ?1/2 gkfeature
is also poorer (comparing Fig. 7 with Figs. 5 and 6), which
may arise because one of the four nitrogen ligands is not
equivalent to the other three ligating nitrogens.
The gkMI¼ þ3/2 and ?3/2 features in Fig. 7 are nonsym-
metric, unlike the corresponding features from the imidazole
model system (Figs. 5 and 6). This is spectroscopic evidence
for the presence of at least two species with differing coordi-
nation. These species could also lead to loss of superhyper-
In the gtregion, the resolution is improved at 1.925 GHz
relative to 3.275 GHz (Fig. 7) but is poorer relative to
lines) on the experimental spectra (thick lines) of Cu-His
in spectral regions used to determine the number of
nitrogen ligands. (a) The gkMI¼ ?1/2 region of the
spectrum. (b) the gtregion of the spectrum. Simulations
assumed either three (labeled ‘‘3N’’) or four (‘‘4N’’)
nitrogen ligands with the same parameters as for Fig. 5,
d and e, respectively. The labeled, dashed lines signify
superhyperfine transition lines in the spectra, as in Fig. 5.
Superposition of calculated spectra (thin
Biophysical Journal 96(8) 3354–3362
EPR of Cu2þ-Prion at 2 GHz3359
copper-histidine spectra (Figs. 5 and 6). The methodology
developed for interpretation of the superhyperfine structure
in the perpendicular region of the spectrum of the imidazole
model system was applied to the PrP construct using an
increased intrinsic linewidth. The most intense line corre-
sponds in position to the EA transition and is assigned to
line 0 of the nitrogen superhyperfine spectrum, and spectra
are simulated for three and four nitrogen ligands. The results
are shown in Fig. 8, a and b, where the fit was forced by
setting both peak-to-peak intensities to the same value. The
fit of transitions labeled 2, 3, and 4 is substantially better
for 4N compared to 3N. These are transitions that are partic-
ularly sensitive to the number of nitrogen ligands (13). The
spectral positions for lines 0 and ?1 are good, but the inten-
sities of these lines are in poor agreement. It is possible that
spectral contributions from spurious species with 0, 1, or 2
nitrogen ligands underlie the spectrum. A change of ligation
can be expected to shift an EA line, resulting in linewidth
increase of lines 0 and 1 of the superimposed superhyperfine
patterns. Because of these uncertainties, no attempt was
made to fit the features to the low-field side of the EA peak.
It is concluded that the dominant species present in the PrP
preparation has four nitrogen ligands, and that both gkand
gtportions of the spectrum are consistent with the presence
of more than one species.
Based on the studies presented here on the copper histidine
system in frozen solution, it is concluded that EPR at
2 GHz solves resolution problems associated with g-strain
broadening in the so-called perpendicular region of the spec-
trum. It can be speculated that unresolved coupling to nearby
protons, the anisotropy of the nitrogen nuclear hyperfine
couplings, nonequivalent nitrogen ligands, and the precise
choice of microwave frequency will determine the linewidth.
We have demonstrated that features to the high-field side
of the most intense spectral feature of copper imidazole at
2 GHz—namely, the EA line—can be well simulated by
the assumption of isotropic coupling to four nitrogen ligands.
Many spectral simulations were performed in the course
of this work, and the existence of an easily recognized,
well-defined EA peak associated with the þ1/2 copper
nuclear hyperfine manifold was robust. The þ3/2 gkturning
point is to the high-field side of the EA line, but it is weak
and diffuse. All studies performed to date support the
hypothesis that the spectral region to the high-field side of
the EA line can be used to distinguish between four-nitrogen
ligation and three-nitrogen/one-oxygen ligation.
Features to the low-field side of the EA line, including
rhombicity, are sensitive to At. The value of Athas not,
to the best of our knowledge, previously been determined
by analysis of EPR powder-type spectra. It was hypothesized
that a scan of the Atvalue to achieve a best fit of the features
to the low-field side of the most intense spectral feature
would permit determination of this difficult-to-measure
parameter. The results support the hypothesis that EPR at
2 GHz permits determination of the Axand Aycomponents
Of the four spectral features assigned to the so-called
parallel region of the spectrum, the MI ¼ ?1/2 feature
exhibits the highest superhyperfine resolution. Improved
resolution arises from a canceling of g-strain by correlated
A-strain. Acquisition at 2 GHz is favorable, and longer acqui-
sition times than were used here are recommended. Never-
theless, the method described herein will generally be
favored because of the 10-fold higher signal intensities of
the MI¼ þ1/2 EA line in the perpendicular part of the spec-
trum relative to the MI¼ ?1/2 feature in the parallel part of
and 1.925 GHz (see text for details).
Copper spectra of the double octarepeat prion peptide at 3.275
posed on the experimental 1.925 GHz spectrum of Fig. 7. (b) Spectral fit for
three equivalent nitrogens.
(a) Simulations for four equivalent nitrogen ligands superim-
Biophysical Journal 96(8) 3354–3362
3360Hyde et al.
the spectrum. It can be presumed that strains in the parallel
and perpendicular regions will differ since they arise funda-
mentally from distributions of different combinations of
bonding parameters. Little is known about strain in the
perpendicular spectral region.
The 2 GHz LGR described here was specifically designed
for this experiment. The sample volume of ~1 mL was preset
by the constraints of preparing the PrP construct. Modeling
of the resonant microwave frequency with the sample was
within 2% of the value observed. The coupling structure
was also computer-modeled, taking into account the calcu-
lated Q-value with the sample, and worked as expected.
The choice of microwave frequency was based on the
analysis of Froncisz and Hyde (5). This is the frequency
that is predicted to yield the narrowest superhyperfine
linewidth for the MI¼ ?1/2 gkfeature. Accidentally, this
choice proved to be favorable for the analysis method intro-
duced here. Fig. 2 suggests that 3 GHz would be more
favorable than 2 GHz, but Fig. 7 indicates for the protein
construct that spectral resolution in the perpendicular part
of the spectrum is substantially poorer at 3.275 GHz relative
to 1.925 GHz. The optimum microwave frequency for our
method appears to be ~2.5 GHz. It is noted that LGRs that
contain 1 mL of sample can readily be designed throughout
the 2–4 GHz band.
The spectroscopic EPR evidence presented here shows
that the dominant component present in the double octare-
peat PrP construct can be simulated using four equivalent
equatorial nitrogen ligands. In region b of the spectrum as
defined in Fig. 2, the relative intensities of lines 2, 3, and 4
are nearly the same for both the PrP construct and the Cu-His
sample that was studied in this work (compare Fig. 6 b with
Fig. 8 b). This assignment is consistent with the results of
Chattopadhyay et al. (12). Component 3 was observed in
the presence of excess octarepeat and assigned to a species
with three equatorial histidines plus either an unknown
oxygen or a nearly equivalent additional in-plane nitrogen
ligand. Component 1 was considered by Burns et al. (10).
It was formed using a stoichiometry of about one octarepeat
to one copper and contains just one histidine nitrogen. Addi-
tional nitrogen ligands are nonequivalent. The resolution of
copper superhyperfine structure in the ?1/2 gkfeature is
much poorer for component 1 than for component 3 (12),
which is consistent with the nonequivalence of the nitrogen
ligands in component 1. For the PrP system studied here,
three histidines are available for every copper ion. The spec-
troscopic data presented here support the presence of
a heterogeneous sample, with one component consisting of
four nearly equivalent nitrogen ligands (possibly all from
histidines) and the other component or components consist-
ing of complexes with either one or two nitrogens. If the
two-nitrogen component is present, nonequivalence would
be expected to result in further broadening of lines 0 and
1, as was observed. A spectrum similar to that found for
component 2 was reported for the full PrP (11). It is possible
that this component can be assigned to one of the superim-
posed spectra in this study.
At 2 GHz, the EPR spectra of Cu2þsquare-planar copper
spectra exhibit increased resolution in the gtspectral region
for several reasons: 1), g-strain is greatly reduced and no
longer dominant over other line-broadening mechanisms;
2), rhombicity of the Zeeman interaction is not apparent;
and 3), only one EA line occurs, as demonstrated in Fig. 2.
We show here for the copper-histidine system that this spec-
tral region can be reliably simulated. These simulations
permit determination of the rhombicity of the copper hyper-
fine coupling and of the number of equatorial nitrogen
ligands, which up to now has been an elusive goal. More-
over, it may be within reach to determine whether all
nitrogen ligands are equivalent. It may even be possible to
determine the anisotropy of the superhyperfine couplings,
as suggested by the residual, line i of Fig. 5. Experiments
as a function of increasing microwave frequency can be
expected to yield quantitative information about g-strain in
the perpendicular spectral region, as well as Zeeman rhom-
including experiments with the microwave field H1parallel
to H0, will provide information about the copper quadrupole
coupling, as shown previously (23). Global fitting of
multifrequency spectra can be expected to yield magnetic
parameters with increased precision. LGRs are the enabling
technology for low-frequency EPR, and further development
of this technology is possible.
This work was supported by National Institutes of Health grants R01
EB001417 and P41 EB001980 (J.S.H.), and R01 GM065790 (G.L.M).
1. Froncisz, W., and J. S. Hyde. 1982. The loop-gap resonator: a new
microwave lumped circuit ESR sample structure. J. Magn. Reson.
2. Hyde, J. S., and W. Froncisz. 1989. Loop gap resonators. In Advanced
EPR: Applications in Biology and Biochemistry. A. J. Hoff, editor.
Elsevier, Amsterdam, The Netherlands. 277–306.
3. Rist, G. H., J. Ammeter, andH. H. Gu ¨nthard.1968. Influenceof the host
lattice upon the EPR coupling parameters of copper-8-hydroxyquinoli-
nate. J. Chem. Phys. 49:2210–2217.
4. Rist, G. H., and J. S. Hyde. 1969. Ligand ENDOR of Cu-8-hydroxyqui-
nolinate substituted into organic single crystals. J. Chem. Phys.
5. Froncisz, W., and J. S. Hyde. 1980. Broadening by strains of lines in the
g-parallel region of Cu2þEPR spectra. J. Chem. Phys. 73:3123–3131.
6. Hyde, J. S., and W. Froncisz. 1982. The role of microwave frequency in
EPR spectroscopy of copper complexes. Annu. Rev. Biophys. Bioeng.
7. Hubbell, W. L., and H. M. McConnell. 1971. Molecular motion in spin-
labeled phospholipids and membranes. J. Am. Chem. Soc. 93:314–326.
8. Weil, J. A., and H. G. Hecht. 1963. On the powder line shape of EPR
spectra. J. Chem. Phys. 38:281–286.
9. Froncisz, W., and P. Aisen. 1982. The EPR spectra of copper transferrin
complexes at 2–4 GHz. Biochim. Biophys. Acta. 700:55–58.
10. Burns, C. S., E. Aronoff-Spencer, C. M. Dunham, P. Lario,
N. I. Avdievich, et al. 2002. Molecular features of the copper binding
Biophysical Journal 96(8) 3354–3362
EPR of Cu2þ-Prion at 2 GHz 3361
sites in the octarepeat domain of the prion protein. Biochemistry.
11. Burns, C. S., E. Aronoff-Spencer, G. Legname, S. B. Prusiner,
W. E. Antholine, et al. 2003. Copper coordination in the full-length,
recombinant prion protein. Biochemistry. 42:6794–6803.
12. Chattopadhyay, M., E. D. Walter, D. J. Newell, P. J. Jackson,
E. Aronoff-Spencer, et al. 2005. The octarepeat domain of the prion
protein binds Cu(II) with three distinct coordination modes at pH 7.4.
J. Am. Chem. Soc. 127:12647–12656.
13. Hyde, J. S., W. E. Antholine, W. Froncisz, and R. Basosi. 1986. EPR
determination of the number of nitrogens coordinated to Cu in
square-planar complexes. In Advanced Magnetic Resonance Tech-
niques in Systems of High Molecular Complexity, Vol. 2. N. Niccolai
and G. Valensin, editors. Birkha ¨user, Boston. 363–384.
14. Ovchinnikov, I. V., and V. N. Konstaninov. 1978. Extra absorption
peaks in EPR spectra of systems with anisotropic g-tensor and hyperfine
structure in powders and glasses. J. Magn. Reson. 32:179–190.
15. Westphal, W. B., and A. Sils. 1972. Dielectric Constant and Loss Data.
Laboratory for Insulation Research, Massachusetts Institute of
Technology, Cambridge, MA.
16. Feher, G. 1957. Sensitivity considerations in microwave paramagnetic
resonance absorption techniques. Bell System Tech. J. 36:449–484.
17. Hanson, G. R., K. E. Gates, C. J. Noble, A. Mitchell, S. Benson, et al.
2003. XSophe-Sophe-XeprView: a computer simulation software suite
for the analysis of continuous wave EPR spectra. In EPR of Free
Radicals in Solids: Trends in Methods and Applications. M. Shiotani
and A. Lund, editors. Kluwer Press, Amsterdam, The Netherlands.
18. Shaltiel, D., and W. Low. 1961. Anisotropic broadening of line-width in
the paramagnetic resonance spectra of magnetically dilute crystals.
Phys. Rev. 124:1062–1067.
19. Basosi, R., W. E. Antholine, and J. S. Hyde. 1993. Multifrequency
ESR of copper: biophysical applications. In Biological Magnetic Reso-
nance. Vol. 13. L. J. Berliner and J. Reuben, editors. Plenum, New
20. Pilbrow, J. R. 1990. Transition Ion Electron Paramagnetic Resonance.
21. Pilbrow, J. R., G. R. Sinclair, D. R. Hutton, and G. J. Troup. 1983.
Asymmetric lines in field-swept EPR: Cr3þlooping transitions in
ruby. J. Magn. Reson. 52:386–399.
22. Stahl, N., and S. B. Prusiner. 1991. Prions and prion proteins. FASEB J.
23. Liczwek, D. L., R. L. Belford, J. R. Pilbrow, and J. S. Hyde. 1983.
Elevation of copper nuclear quadrupole coupling in thiol complexes
by completionof the coordination
Biophysical Journal 96(8) 3354–3362
3362 Hyde et al.