(Received for publication, March 7, 1997, and in revised form, May 14, 1997)
From the Departments of Molecular Biology and
Biochemistry and ¶ Psychobiology, University of California,
Irvine, California 92696 and the § Universidad Nacional
Autonoma de Mexico, Facultad de Quimica, Mexico
D.F., Mexico 04510
Recent studies have implicated the amyloid A
peptide and its ability to self-assemble as key factors in the
pathogenesis of Alzheimer's disease. Relatively little is known about
the structure of soluble A
or its oligomeric state, and the existing
data are often contradictory. In this study, we used intrinsic
fluorescence of wild type A
-(1-40), fluorescence resonance energy
transfer (FRET), and gel filtration chromatography to examine the
structure of A
-(1-40) in solution. We synthesized a series of
mono-substituted fluorescent A
-(1-40) derivatives to use as donors
and acceptors in FRET experiments. We selected fluorescent peptides
that exhibit aggregation properties comparable to wild type A
for
analysis in donor-acceptor pairs; two labeled with
5-(2-((iodoacetyl)amino)ethyl)aminonaphthylene-1-sulfonic acid
at Cys-25 or Cys-34 and fluorescein maleimide at Cys-4 or Cys-7.
Another peptide containing a Trp substitution at position 10 was used
as an acceptor for the intrinsic Tyr fluorescence of wild type
A
-(1-40). Equilibrium studies of the denaturation of A
-(1-40)
by increasing concentrations of dimethyl sulfoxide (Me2SO) were conducted by monitoring fluorescence,
with a midpoint value for the unfolding transition of both the
substituted and wild type peptides at among 40 and 50%
Me2SO. A
-(1-40) is well solvated and largely monomeric
in Me2SO as evidenced by a lack of FRET. When donor and
acceptor A
derivatives are mixed together in Me2SO and
then diluted 10-fold into aqueous Tris-HCl buffer at pH 7.4, efficient
FRET is observed immediately for all pairs of fluorescent peptides,
indicating that donor-acceptor dimers exist in solution. FRET is
abolished by the addition of an excess of unlabeled A
-(1-40),
demonstrating that the fluorescent peptides interact with wild type
A
-(1-40) to form heterodimers that do not exhibit FRET. The
A
-(1-40) dimers appear to be very stable, because no subunit
exchange is observed after 24 h between fluorescent homodimers.
Gel filtration confirms that nanomolar concentrations of
14C-labeled A
-(1-40) and fluorescein-labeled
A
-(1-40) elute at the same dimeric position as wild type
A
-(1-40), suggesting that soluble A
-(1-40) is also dimeric at
more physiologically plausible concentrations.
The extracellular deposition of -amyloid in senile plaques is
one of neuropathological hallmarks of Alzheimer's disease. Biochemical
analysis of the amyloid peptides isolated from Alzheimer's disease
brain indicates that amyloid
(A
)1 1-42 is the
principal species associated with senile plaque amyloid (1), while
A
-(1-40) is more abundant in cerebrovascular amyloid deposit. A
is folded into the
-sheet structure that is characteristic of
amyloid fibrils. Amyloid plaque formation may involve two basic steps:
the initial formation of a seeding aggregate that establishes the
amyloid fibril lattice (2), followed by the elongation of the fibril by
the sequential addition of subunits (3). Some of the key parameters
that promote the assembly of amyloid fibril include high peptide
concentration, long incubation time and low pH (pH 5-6) (4-6),
solvent composition (7), and salt concentration (8). The length of the
carboxyl terminus is also critical in determining the assembly
dynamics. The A
-(1-42) isoform aggregates at a significantly
greater rate and to a greater extent at pH 7.4 than A
-(1-40).
Assembly of A
into the fibrils may also be promoted by molecules
that interact with A
and increase its rate of aggregation in
vitro including apolipoprotein E (9, 10),
1-antichymotrypsin (11), complement C1q (1), heparin
sulfate proteoglycan (12), and zinc ions (13, 14).
Although many of the parameters influencing fibril assembly have been
elucidated, relatively little is known about the structure of soluble
A. Understanding the structure of A
may provide insight into how
this peptide assembles into the amyloid fibrils characteristic of
Alzheimer's disease and other amyloidoses. Gel filtration analyses of
A
in solution have revealed the presence of multiple, discrete structures that have been variously interpreted as monomer, dimer, trimer, and higher order aggregates (8, 13, 15-17). While differences
in the concentration of the peptide, time of incubation, and the
structure of A
(16) can account for some of the discrepancies in the
oligomeric species observed in the studies cited above, it is not clear
that the identities assigned to the peaks are correct, since gel
filtration only measures the effective hydrodynamic radius of a
molecule and not its molecular mass.
Previous studies from this laboratory indicate that the smallest
species in aqueous solution elutes from a gel filtration column with an
apparent molecular mass appropriate for a dimer for both A-(1-40)
and A
-(1-42) and at higher concentrations, larger micelle-like
oligomers also exist in solution (16). The presence of these
micelle-like oligomers has been confirmed by dynamic light scattering
measurements (18, 19) and analytical ultracentrifugation (19). In this
work, we have used the intrinsic fluorescence of wild type A
-(1-40)
and fluorescence resonance energy transfer in conjunction with gel
filtration chromatography to define the oligomeric state of soluble
A
-(1-40). Here we report that soluble A
-(1-40) forms a stable
dimer at concentrations from the low nanomolar range up to the critical
concentration of approximately 25 µM, as evidenced by the
observation of efficient FRET between several different combinations of
donor and acceptor peptides. This suggests that soluble A
-(1-40)
adopts an ordered conformation in solution as a prelude to fibril
assembly and that dimerization is an initial event in amyloid
self-assembly.
All A peptide analogs were synthesized by
fluoren-9-ylmethoxy carbonyl chemistry using a continuous flow
semiautomatic instrument as described previously (5). The peptides were
purified by reverse phase high performance liquid chromatography, and
the purity and expected structure was verified by electrospray mass spectrometry. Only peptides exhibiting 90.0% or greater purity with
less than 5.0% of a single contaminant were used. Cys substitution mutants were synthesized simultaneously by the same method, except that
at locations where Cys was substituted, a portion of the resin was
coupled separately with Cys. [3H]A
-(1-40) and
[14C]A
-(1-40) were synthesized by incorporation of
Fmoc-[3H]Phe or Fmoc-[14C]Ala at positions
4 and 2 respectively, yielding specific activities of 200 mCi/mmol for
[3H]A
-(1-40) and 36 mCi/mmol for
[14C]A
-(1-40). 1,5-IAEDANS and FM were obtained from
Molecular Probes (Eugene, OR). All other reagents were of the highest
analytical grade commercially available. We use a shorthand notation to
refer to the A
-(1-40) analogs that indicates the position of the
Cys substitution with the understanding that all peptides are 40 residues long and the rest of the sequence is that of wild type A
as
described in the abbreviations list.
Since A-(1-40) was modified with a single Cys at different
positions, the sulfydryl-specific reagents FM and 1,5-IAEDANS were used
to prepare fluorescent derivatives. The A
-(1-40) analog peptide was
dissolved in 10 mM MOPS, pH 8.5, at a concentration of 25 µM (pH 7.4 in the case of fluorescein labeling).
1,5-IAEDANS or FM was added to this solution from a stock solution of
10 mM at a 20-fold (for 1,5-IAEDANS) or 5-fold (for FM)
molar excess over A
-(1-40). The reaction was allowed to proceed at
room temperature in the dark for 6 h. Free fluorophore was then
removed by filtration on a Sephadex G-25 column equilibrated with 10 mM MOPS at pH 7.4. Labeled A
-(1-40) was aliquoted,
lyophilized, and stored at
20 °C. Protein was determined by
Coomassie R protein assay reagent (Pierce). The concentrations of
1,5-IAEDANS or FM were spectrophotometrically determined by using their
molar extinction coefficients (5.7 mM
1 at 336 nm or 83 mM
1 at 490 nm, respectively). The
labeling stoichiometry of the final products was 1.0. The stoichiometry
was confirmed by laser desorption mass spectrometry that demonstrated
all of the precursor has been converted to a mass appropriate for
fluorescent peptide.
Aggregation was determined using a
sedimentation assay as described previously (5). 75 µM
[3H]A-(1-40) (specific activity 36 mCi/mmol) was
mixed with 5 µM fluorescent A
-(1-40) in either 0.1 M NaAc, pH 5.0, 0.1 M NaCl, 20 mM
Tris-HCl, pH 7.4 or 0.1 M NaCl, 20 mM Tris-HCl,
70 µM ZnCl2, and incubated for 48 h at
room temperature. The radiolabeled and fluorescent peptides were
diluted 10-fold from water stock solutions upon mixing. The samples
were centrifugated at 15,000 × g for 10 min in a
Beckman Microfuge 11. Afterward, the amount of fluorescent and
tritiated wild type A
in both the supernatant and pellet was
determined by measuring the fluorescence intensity and the radioactivity by scintillation counting.
Gel filtration analysis was
performed with a Pharmacia Superdex 75 HR 10/30 column using a Waters
490 multiple wavelength UV absorbance detector and Hewlett Packard 3250 fluorescence detector. Data were collected with a Waters Maxima
chromatography data system, with a flow rate of 0.4 ml/min. Because the
refolding experiments of A-(1-40) were carried out by diluting the
Me2SO to subdenaturing concentrations, the running buffer
used was 50 mM Tris-HCl, 0.1 M NaCl, pH 7.4 (buffer A) in the presence of 10% or 2% Me2SO. The column
was calibrated several times using the buffer A in the presence of 10%
or 2% Me2SO, the standards were chromatographed both
separately and in mixtures, obtaining the same calibration curve in all
conditions. The standards used to calibrate the column and their masses
(Da) are: thyroglobulin (670,000), bovine serum albumin (68,000),
ovalbumin (43,000), soybean trypsin inhibitor (23,000), ubiquitin
(8500), aprotinin (6500), and acetone (58). The peptides were detected
by UV absorbance at 280 nm and by fluorescence emission at 482 nm for
1,5-IAEDANS (or at 520 nm for FM) upon excitation at 336 nm (or 490 nm
for FM).
11 µM A-(1-40) or A
C25AEDANS was
incubated in increasing concentrations of Me2SO, at
24 °C. Emission spectra were recorded after 30 min incubation from
290 to 400 nm upon excitation at 280 nm (or from 346-600 upon
excitation at 336 nm for A
C25AEDANS). Emission spectra were not
recorded at longer incubation times as A
aggregates (19). For
refolding experiments, samples were incubated in 100%
Me2SO for 30 min at room temperature and refolding was
initiated by 10-or 50-fold dilution of the Me2SO solution in Tris-HCl buffered solution. The concentration of peptide ranged between 3 and 10 µM.
Absorption
measurements were measured with a Perkin Elmer Lambda 3B UV-Vis
spectrophotometer. Fluorescence spectra (excitation band pass 4 nm;
emission band pass 8 nm) were measured either on an Aminco SLM 48000 or
a SPEX Fluorolog F112A spectrofluorometer. Intrinsic Tyr fluorescence
was measured from 285 to 400 upon excitation at 275 nm. For
AC25AEDANS or in energy transfer experiments, excitation was at 330, and the spectra were obtained from 340 to 620 nm. In all cases,
emission fluorescence spectra of identical samples (without protein)
were recorded. These were subtracted from the experimental samples. The
lifetime measurements for FM were acquired using the 488 nm line of
argon ion laser for excitation using a multiharmonic frequency-domain
spectrofluorometer (Aminco 48000S).
The efficiency
(E) of fluorescence resonance energy transfer (FRET) between
probes was determined by measuring the fluorescence intensity of the
donor (AC-AEDANS or A
-(1-40)) both in the absence (Fd) and presence (Fda)
of the acceptor (A
C-FM or A
Y10W), as given by Equation 1.
![]() |
(Eq. 1) |
To
demonstrate the feasibility of the experiments employing fluorescent
A analogs, we synthesized a series of A
-(1-40) variants containing a single Cys substitution. We chose A
-(1-40) because it
is stable over the time interval required for these experiments, while
A
-(1-42) rapidly forms higher order oligomers that would complicate
the interpretation of the fluorescence data (16, 21). Cys was chosen
because of its unique chemical reactivity and its absence in the wild
type A
sequence. We initially synthesized a series of Cys
substitutions by replacing every third residue, because the Cys side
chain would be expected to alternate on opposite sides of the strand in
a
-sheet structure. The Cys-containing probe peptides were
covalently labeled with a variety of extrinsic fluorescent probes. For
this work we used A
-(1-40) Cys mutants labeled with 1,5-IAEDANS at
positions 25 and 34 and FM-labeled A
-(1-40) Cys mutants at
positions 4 and 7. Mass spectrometry confirmed the expected mass of the
final product, and the absence of the precursor peptide indicated that
the labeling reaction was complete (data not shown). We also
synthesized a probe containing Trp instead of Tyr at position 10 to use
as an acceptor for the wild type peptide intrinsic Tyr fluorescence.
Some of the other peptides synthesized were not analyzed because they
either labeled inefficiently or displayed significantly altered
aggregation properties (data not shown).
An obvious concern in modifying the structure of A to make
fluorescent derivatives is whether the modification significantly alters the structure and properties of A
. We compared the
aggregation properties of the fluorescent derivatives to wild type A
under physiological conditions where it is largely soluble
(e.g. Tris-HCl buffered saline at pH 7.4) and under
conditions that are known to promote fibril assembly (pH 5.0 and at pH
7.4 in the presence of ZnCl2) (Fig.
1). At pH 7.4 and at pH 5.0, the
sedimentation behavior of all of the fluorescent peptides used is
indistinguishable from wild type A
-(1-40). In the presence of 70 µM Zn2+, the fluorescein- and AEDANS-labeled
A
peptides aggregate to approximately 50-75% of the extent of wild
type A
. However, Trp substitution at residue 10 did not alter the
aggregation behavior in response to Zn2+. Several of the
other Cys substitution mutant peptides we synthesized were not suitable
for further analysis because either they failed to label efficiently or
the labeled peptides displayed significantly altered aggregation
properties (data not shown).
The oligomeric structure of the fluorescent peptides was characterized
by gel filtration chromatography, and we found that the fluorescent
peptides AC25AEDANS and A
C7FM elute at the same position as wild
type A
-(1-40) (Fig. 2). The elution
position corresponds to an apparent molecular mass of 9000 Da
established by the elution behavior of a series of calibration
standards as reported previously (16) (Fig. 2A,
inset). The other fluorescent peptides used in this study
also elute as a dimer (data not shown). The calibration curve also
indicates that the expected elution position for a peptide of the mass
of monomeric A
-(1-40) is well separated from the observed elution
position of dimeric A
-(1-40). Nanomolar concentrations of
14C-labeled A
-(1-40) also elute at the position
expected for a dimer (Fig. 2B). Since gel filtration only
measures the effective Stokes' radius, the elution behavior is not
definitive evidence for the interpretation that the peak represents a
dimer.
Denaturation and Renaturation of A
To more definitively determine the structure of
soluble A-(1-40) by FRET, we had to first establish conditions for
the denaturation and renaturation of A
-(1-40). Previous studies
using a combination of Fourier transform infrared spectroscopy, and
dynamic light scattering (7, 19) suggested that A
is a random coil
monomer in Me2SO. The intrinsic fluorescence of proteins
provides a signal commonly used to monitor conformational changes and
unfolding (22). In the present work we used the intrinsic Tyr
fluorescence of wild type A
-(1-40) to study the denaturation of
A
in Me2SO and its renaturation. The Tyr emission of
most native proteins and peptides is frequently small or undetectable
due to the presence of more highly fluorescent Trp residues (20, 22),
but Trp is absent in A
. The spectral properties of extrinsic
fluorescent probes can also be exploited to obtain information about
the environment surrounding the probe and to examine whether the
environment changes upon aggregation and assembly (20). The
denaturation curve for wild type A
-(1-40) showed a single, smooth
cooperative transition (Fig. 3).
Increasing concentrations of Me2SO increased the intrinsic fluorescence intensity of A
-(1-40), indicating that a significant increase in the exposure of the Tyr residue occurs in the unfolded state. The midpoint of intrinsic fluorescence changes occurred at
approximately 40% Me2SO. The emission maximum of Tyr is
not affected by Me2SO, remaining the same at all
concentrations (e.g. 308 nm) because the Tyr fluorescence
emission maximum is not sensitive to the polarity of the solvent (20).
The data have been corrected for the relatively small solvent effect of
Me2SO on free Tyr (less that 5% of the intensity change
observed for A
-(1-40)) to ensure that the curve accurately reflects
the unfolding of A
-(1-40).
We also examined the denaturation of the extrinsically-labeled
fluorescent A-(1-40) probes. For environment-sensitive fluorophores (like 1,5-IAEDANS) the emission maximum shifts to a shorter wavelength (blue shift) as the polarity of the surrounding environment decreases (20). Conversely the emission maximum shifts to the longer wavelengths (red shift) in a more polar environment. For A
C25AEDANS, a marked blue shift (42 nm) of the emission was observed upon unfolding by
Me2SO (Fig. 3), indicating that the fluorophore is
increasingly exposed to the surrounding media at increasing
Me2SO concentrations. As with the intrinsic Tyr
fluorescence, the midpoint of the blue shift of AEDANS occurred at
approximately 50% Me2SO. The unfolding transition was
recorded after 30 min of incubation in Me2SO, as it has
been reported that longer incubation times lead to an aggregated comformation of A
(19). At concentrations of Me2SO below
10%, there is little further change in the intrinsic fluorescence
emission of A
-(1-40) or the extrinsic fluorescence of A
C25AEDANS
(Fig. 3). These results indicate that AEDANS-labeled and wild type
A
-(1-40) peptides have similar stabilities and suggest that there
is relatively little change in the overall structure of A
-(1-40)
over the range of Me2SO from 0 to 10%.
The renaturation of A-(1-40) from Me2SO solution was
also examined. Upon 10-fold dilution of Me2SO into aqueous
buffer solution, the emission spectrum of wild type A
-(1-40) and
A
C25AEDANS showed the same maximum at 308 and 494 nm, respectively,
as observed for the peptides dissolved directly in buffer A, indicating
that the denatured peptide returns to the same overall structure (Fig. 4, A and B). Time
course studies indicated that the refolding of the peptide was
immediate (data not shown). Similar denaturation and renaturation
results were obtained with guanidine HCl (data not shown).
Me2SO stock solutions of peptide were employed for all of
the subsequent experiments.
Association of Fluorescent A
Fluorescence
resonance energy transfer between AEDANS and fluorescein was used
initially to monitor association of A-(1-40) monomers following
dilution of Me2SO into aqueous buffer solution. For these
experiments, A
C25AEDANS and A
C7FM Me2SO stock
solutions were mixed 1:1 (donor:acceptor), and then subsequently
diluted 10-fold in buffer A. The final concentration of the peptide was 3 µM. The resulting fluorescence spectra are shown in
Fig. 5A. Efficient FRET was
observed, as evidenced by a quenching of the donor emission at 474 nm
and an increase in the acceptor fluorescence at 520 nm, compared with
the control spectra, indicating that hybrid A
-(1-40) dimers had
formed in the mixture containing both donor and acceptor (Fig.
5A). The efficiency of FRET did not change significantly
upon subsequent incubation for 24 h (data not shown). To control
for possible effects of peptide structure on the fluorescence intensity
of labeled peptides, we carried out control measurements in which
either A
C25AEDANS or A
C7FM were individually mixed with an equal
amount of non-labeled peptide in Me2SO, and then diluted
10-fold in buffer A. The emission spectra obtained for A
C25AEDANS or
A
C7FM are shown as arithmetic sum of the individual spectra
(i.e. the expected emission in the absence of energy
transfer). We also observed efficient FRET with several other pairs of
A
-(1-40) peptides (Table I). Fig.
5B shows the spectra of the energy transfer experiment where
another donor-acceptor pair, A
C34AEDANS-A
C4FM, was used. The
efficiency of FRET for this combination is higher than that observed
for A
C25AEDANS and A
C7FM, suggesting that the peptide structure
is ordered and that the fluorophores at positions 34 and 4 may be in
closer proximity in the structure than those at positions 25 and 7. Once formed, the A
-(1-40) dimers appear to be relatively stable in
solution. If fluorescent homodimers are formed first by individually
diluting the stock solutions 10-fold into buffer A, and then the
homodimers are subsequently mixed, no resonance energy transfer is
observed over an incubation of 24 h (data not shown), indicating
that subunit exchange between homodimers is not detectable over this
period. This suggests that the dimer is very stable.
|
Several controls were conducted to ensure that the FRET observed is due
to interactions between peptides that occurs in solution with wild type
A-(1-40). After each FRET experiment, we confirmed that the
fluorescent peptide mixture eluted at the same apparent dimer position
as the wild type peptide by gel filtration chromatography (data not
shown). We also determined that FRET is nearly abolished when a 10-fold
molar excess of wild type peptide is added to the fluorescent peptide
mixture in Me2SO and then subsequently diluted (Fig.
5C), indicating that the wild type peptide can compete for the fluorescent peptides and form fluorescent and wild type
A
-(1-40) heterodimers that do not exhibit FRET. Finally, we
exploited the endogenous Tyr fluorescence of wild type A
-(1-40) as
a donor for A
-(1-40) in which Trp replaces the wild type Tyr at
position 10 (A
Y10W). Efficient FRET was also observed for the
mixture of A
-(1-40) and A
Y10W (Fig.
6). This experiment was conducted in two
different conditions: where the mixture was diluted to 10%
Me2SO (Fig. 6, A and B) and to 2%
Me2SO to dilute the solvent to a more subdenaturing
concentration (Fig. 6, C and D). A deconvolution analysis of FRET between A
-(1-40) and A
Y10W was conducted as reported (23) (Fig. 6, B and D). The efficiency
of FRET is significantly higher in 2% Me2SO than in 10%
Me2SO, suggesting that the structure in 10%
Me2SO may be partially unfolded (Table I).
Our experiments with three different donor-acceptor pairs demonstrate
that efficient FRET is observed when the peptides are mixed in
Me2SO prior to dilution in aqueous buffer, suggesting that
they form dimers in solution. It is conceivable that the aggregates may
actually represent higher order structures (e.g. trimers or
tetramers). We measured the lifetime of the FM-labeled A-(1-40) by
phase modulation frequency domain methods in the range 3 µM to 100 nM. The fluorescence lifetime data
fit to a single decay exponential that remained constant over the
concentration range examined (data not shown). This suggests that there
is a single distance between the fluorophores and that it does not change over the concentration range examined. A single distance would
be expected in a population of structurally homogeneous dimers, while
higher order aggregates would be expected to contain more than one
distance between fluorophores that would be observable as different
lifetimes. Taken together with the gel filtration data, the simplest
conclusion is that A
-(1-40) exists as a stable dimer in aqueous
saline solution at pH 7.4 over the concentration range from nanomolar
to micromolar.
The purpose of the present study was to characterize the
oligomeric structure of A-(1-40) in aqueous solution. The published literature on this subject is confusing because different distributions of monomer, dimer, trimer, tetramer, and higher order oligomers have
been reported (8, 13, 15, 16, 21, 24-30). Some of the discrepancies
can be attributed to structural differences in amyloid isolated from
brain tissue (21, 26, 27, 29) in comparison to synthetic A
(8, 13,
15, 16, 21, 25). The aggregates observed in amyloid isolated from brain
tissue may be covalent because they are stable under denaturing
conditions and may reflect peptide-peptide associations occurring
in vivo (21). Even though many of the differences in the
oligomeric state of synthetic A
can be explained by differences in
the concentrations of peptide examined and the conditions and
methodology employed, a controversy remains about whether a monomer or
dimer is the smallest structure that exists under physiological
conditions. Hilbich et al. (8) reported that a dimer is the
predominant species at pH 7.0 and at physiological salt concentrations
for A
-(10-43), in good agreement with our own observations on
A
-(1-40) and A
-(1-42) (16). Barrow et al. (15)
reported a mixture of monomer, dimer, trimer, and tetramer for
A
-(1-39) and A
-(1-42), and Roher and colleagues reported that a
monomer is the predominant species for both A
-(1-40) and
A
-(1-42) at pH 7.4 (21). While this distinction may not be
important for screening for compounds that inhibit fibril assembly, it
is important for understanding the mechanism of amyloid assembly and
critical for correctly assigning the identity of NMR cross peak
resonances in 2D structural analysis of A
, where it has been assumed
that the peptide is monomeric (17, 31).
In this study, we used fluorescence spectroscopy and FRET to clarify
the oligomeric structure of soluble A-(1-40). Fluorescence spectroscopy has been used to probe the structure and dynamics of a
wide variety of biological self-assembly or association reactions, including triose-phosphate isomerase (23), actin (32, 33), tubulin
(34), and neurofilaments (35). FRET between two fluorophores can be
used as a spectroscopic ruler for measurements of distances in the
range of 10-100 Å, distances that begin where NMR methods leave off
and that end at the diameter of a typical amyloid fibril (36). We found
that A
-(1-40) is well solvated and monomeric in Me2SO
solution, in agreement with previously published evidence (7, 19), and
we found that it renatures upon dilution in aqueous buffers. Upon
dilution from Me2SO solution, A
-(1-40) elutes as a
single peak at a position corresponding to the expected molecular
weight of a dimer and fluorescent A
-(1-40) derivatives exhibit
efficient FRET, indicating that a stable complex is formed between
donor and acceptor probes. Together, these observations suggest that
A
-(1-40) exists as dimer in solution. The fact that a single peak
is observed on gel filtration suggests that A
-(1-40) monomer is not
detectable under these conditions. It is conceivable that the monomer
and dimer might co-migrate on gel filtration, but this seems unlikely
because the A
-(1-40) peak elutes in the mid range of the
fractionation volume and the column is capable of resolving monomeric
and dimeric A
under denaturing conditions (21). The fact that a
single fluorescent lifetime was observed for A
C7FM is also
consistent with the interpretation that the peak does not contain a
mixture of monomer and dimer.
The utilization of fluorescent-labeled A-(1-40) peptides in these
studies requires altering the structure of A
-(1-40), so it is
conceivable that the properties of the probes may not be precisely the
same as A
-(1-40). Several lines of evidence argue against the
interpretation that dimer formation is an artifact of fluorescent
labeling. The elution behavior of the fluorescent peptides on gel
filtration is identical to wild type A
-(1-40), indicating that they
have the same hydrodynamic radius in solution. Denaturation-renaturation experiments also demonstrate that the stability of the fluorescent peptides is indistinguishable from the
wild type peptide. When we assayed the aggregation properties of the
fluorescent probe peptides, we found that they were nearly identical to
the wild type peptides. Finally, we employed several different
combinations of donor and acceptor peptides labeled at different
positions and all of the peptides used behaved as dimers. One of the
donor-acceptor pairs used Tyr fluorescence from wild type A
-(1-40)
as the donor and Trp fluorescence as the acceptor from the A
peptide
substituting Trp for Tyr at position 10. Tryptophan substitution has
been often used to study proteins where is not desirable to drastically
modify its structure (37). This donor-acceptor pair may be expected to
be efficient since the Förster distance for Tyr-Trp transfer is
about 10-18 Å, a size comparable to the diameter of many proteins
(20). The observation of efficient energy transfer between Tyr and Trp
confirms that wild type A
-(1-40) forms a dimer with the Trp
substitution probe.
The finding that A-(1-40) forms a stable dimer in solution suggests
that dimerization is the initial event in amyloid aggregation and that
it represents the fundamental building block for further fibril
assembly as has been proposed previously (7, 16). This model of amyloid
assembly is very similar to the model recently proposed for
immunoglobulin light chain amyloid fibrils on the basis of molecular
modeling studies (38). In this model, the two light chains align in a
parallel fashion creating a dimer with a 2-fold axis of symmetry. It
seems likely that the A
-(1-40) dimer may also have the same
arrangement because it behaves as if it is axially amphipathic with one
end polar and the other end hydrophobic (16). This implies that at
least some of the dimer may be arranged in a parallel fashion, because
if the dimer were arranged in a simple head-to-tail fashion, as has
been proposed previously (20), the hydrophobic moment of the resulting
dimer might be expected to be symmetric with respect to the ends of the
dimer. Previous CD and Fourier transform infrared spectroscopic studies
indicate that soluble A
has substantial
-sheet content, suggesting that the dimer adopts a
structure (4, 7, 15, 17).
In the amyloid light chain model, the next step in polymerization is
head to tail association of dimers related by a 90° rotation around
the 2-fold axis to form a tetramer that establishes a "proamyloid" filament lattice that is capable of propagating filaments of indefinite length. Elongation of the fibril is accomplished by the stepwise addition of dimers onto the filament. This "proamyloid" filament may correspond to the " crystallite" proposed for A
from
fibril diffraction measurements (39) and observed in atomic force
microscopy images (40). In this step, the free energy contribution of
individual amino acid side chains is effectively doubled because of the
2-fold symmetry of the interacting surfaces (38). It seems likely that A
is also capable of forming a similar stable tetramer. Discrete aggregate species migrating at the position expected for a tetramer have been observed by SDS-PAGE in samples of A
-(1-42) (5, 16). In
A
, the formation of this SDS-resistant higher order aggregate
depends on the length of the carboxyl terminus of A
(e.g.
only A
-(1-42) and A
-(1-43)) and it is also
concentration-dependent, occurring at approximately the
critical micelle concentration defined by surface tension measurements
(16). These results suggest that the formation of higher order
aggregates in A
may be mediated predominantly by hydrophobic
contacts.
The formation of amyloid fibrils in the light chain model is proposed
to proceed by the lateral association of the "proamyloid" filaments
or subfibrils (38). This step corresponds mechanistically to the
formation of the nucleating center proposed for A (41). Evidence for
the existence of subfibrils has been obtained for A
by rapid-freeze,
deep-etch electron microscopy (42) and more recently by atomic force
microscopy (40). In the light chain model, four strands are proposed to
associate in an antiparallel fashion, but it is not clear how many
subfibrils are contained within A
fibrils. The number of subfibrils
for amyloid A
fibrils is not clear, but electron micrographs show
images that appear to contain five subfibrils (42), and this number
gives the best fit in modeling the observed reflections in fiber
diffraction studies (39). It is also conceivable that this number could vary within a population of A
fibrils, and this could account for
differences in the diameter and morphology of fibrils that has been
reported (40) and the fact that sheet and ribbon morphologies are also
known to occur in synthetic A
aggregates (5, 43). Other than the
initial dimerization event, the details of this model of amyloid fibril
formation remain to be verified experimentally for A
.
Fluorescent derivatives of A-(1-40) may also prove useful for
exploring other aspects of amyloid structure. Quantitative measurements
of distances between fluorescent dipoles by FRET are also possible
(20). The fact that A
-(1-40) forms a dimer in solution simplifies
the interpretation of FRET measurements, because there is only one
distance between fluorophores in a dimer. If a sufficient number of
distance measurements are available, it may be possible to discern the
structural organization of the polypeptide within the dimer, albeit at
a lower resolution than might be achievable by x-ray crystallography or
NMR. Until now, this is the first report in which A
amyloid
intrinsic fluorescence and FRET between A
fluorescent derivatives
have been used to study amyloid structure. Different fluorescent
A
-(1-40) analogs may also be useful for mapping the
solvent-accessible surface of the amyloid fibril by quenching studies
(20). Since the aggregation state of A
has been shown to be
important for in vitro toxicity, there is a growing interest
in molecules that inhibit A
aggregation as a candidates for
potential therapeutic strategies based on blocking amyloid deposition.
Potentially interesting classes of inhibitors for therapeutic
evaluation would be molecules that bind tightly with A
monomer and
prevent dimerization and molecules that prevent oligomerization of
dimers or the extension of fibrils.
This work was made possible, in part, through access to the Laser Microbeam and Medical Program (LAMMP) and the Clinical Cancer Center Optical Biology Shared Resource at the University of California, Irvine (facilities are supported by National Institutes of Health Grants RR-01192 and CA-62203). W. G.-R. and M. S.-B. gratefully acknowledge the Center for Fluorescence Spectroscopy in the Department of Biological Chemistry at the University of Maryland School of Medicine for training in fluorescence spectroscopy. We thank Drs. Nancy Allbritton and Bruce Tromberg for many helpful discussions and suggestions.