Ultrafast Time-Resolved Studies on Fluorescein for Recognition Strands Architecture in Amyloid Fibrils

Abstract

Protein aggregation is associated with numerous devastating diseases such as Alzheimer’s, Parkinson’s, and prion diseases. Development of therapeutics would benefit from knowledge of the structural organization of protein molecules in these amyloid aggregates, particularly in their aqueous biological milieu. However, detailed structural studies to date have been mainly on the solid state and have required large quantities of purified aggregate. Moreover, these conventional methods require the aggregated assembly to remain structurally stable over days or weeks required to perform the experiment, whereas the pathologically relevant species of in vivo aggregates may be shorter lived. Here, we show the organization of protein chains in dissolved amyloid aggregates can be readily determined spectroscopically using minute quantities of fluorescein-labeled protein segments in a matter of minutes. Specifically, we investigated the possibility of using the ultrafast dynamics of fluorescein to distinguish among three categories of β-sheet geometry: (1) antiparallel in-register, (2) parallel in-register, or (3) antiparallel out-of-register. Fluorescein, the most commonly used staining dye in biology and medicine, was covalently attached to the N-termini of peptide sequences selected from a library of known amyloid crystal structures. We investigated the aggregates in solution using steady-state and time-resolved absorption and fluorescence spectroscopy. We found that the dynamics of fluorescein relaxation from the excited state revealed amyloid structure-specific information. Particularly, the nonfluorescent cation form of fluorescein showed remarkable sensitivity to local environments created during aggregation. We demonstrate that time-resolved absorption is capable of differentiating strand organization in β-sheet aggregates when strong intermolecular coupling between chromophores occurs. This approach can be useful in optical recognition of specific fibril architectures of amyloid aggregates.

Graphical Abstract

INTRODUCTION

Fluorescein belongs to a popular family of xanthene dyes and is widely used in bioresearch.^1,2 Approved by the FDA in 1976, it was accepted in the community as a safe agent for medical applications and treating patients.³ It was then a natural step forward to explore various biological systems of interest including DNA,^4,5 proteins,^6,7 or lipid membranes⁸ in conjugation with fluorescein. Because of high photoluminescence quantum yield and reproducible photophysical properties it has been used in imaging,⁹ cell staining,¹⁰ diagnostics,¹¹ eye examination,¹² sensing,¹³ drug delivery,¹⁴ and other applications.

However, fluorescein optical properties can be affected by many factors including the pH of dye environment, aggregation, and charge or energy transfer processes that may occur in conjugation with biomolecules.¹⁵⁻¹⁷ To get reliable data on fluorescein’s behavior in a complex media, such as cellular environment, it is important to know how specific conditions are affecting the optical properties of the dye. The dependence of optical absorption,¹⁸ photoluminescence quantum yield¹⁹ (PLQY), and lifetimes²⁰ of fluorescein dye had been extensively investigated from the middle 70 s, but to the best of our knowledge, not the ultrafast transient absorption in the context of dye sensitivity to amyloid polymorphism and strands organization in aggregates associated with neurodegenerative diseases.

Parkinson’s, Alzheimer’s, and prion diseases are members of a group of over two dozen known amyloid diseases. Amyloid disease patients share in common the deposition of amyloid fibrils detected post-mortem in tissues.²¹ Amyloid fibrils form when protein molecules aggregate, thereby losing normal biological functions.²² Mature fibrils from different diseases share common features such as length (>1 μm), width (8–10 nm), and general architecture of strands oriented perpendicular to the long fibril axis.²³ Furthermore, amyloids, both fibrillar and oligomeric, are composed of β-sheets. However, despite these similarities, a diversity of β-sheet geometries are possible,²⁴ and simulations suggest a diversity of geometries form at the earliest stage of aggregation, called nucleation²⁵ and morph during the transition to oligomers and fibrils. A rapid and reliable method of interrogation of β-sheet geometry is needed.

Diversity in β-sheet geometry arises from three fundamental levels of organization: (1) the directionality of neighboring β-strands within β-sheets (parallel or antiparallel), (2) the distribution of side chains between the two sheet faces (antifacial or equifacial), and (3) translational shifts of neighboring strands along the strand direction (in-register or out-of-register).²⁴ These three levels of organization can appear in any combination, producing at least eight conceivable sheet geometries. Determination of the specific sheet geometry is an emerging challenge for structural biologists, essential to the rational development of therapeutics and novel diagnostics for early detection in amyloid diseases.

We investigated the possibility of using the ultrafast dynamics of fluorescein’s excited state to distinguish among three categories of β-sheet geometry: (1) antiparallel in-register, (2) parallel in-register, or (3) antiparallel out-of-register. The third type of geometry is known to be associated with both amyloid fibrils, and toxic β-barrel-like oligomers.²⁶ We show that in ultrafast time-resolved absorption experiments, different amyloid architectures create various deactivation channels for the attached chromophores. Peptide sequences from a library of crystallographically determined amyloid fibrils were selected from these three different groups (1) GYVLGSA,²⁷ FGAILSS;²⁸ (2) GGNNQQNY,²⁹ LVEALYL,³⁰ NNQNTF;³¹ (3) ASLTVS, VAVHVF.³² We monitored changes in dynamics of photogenerated excited states in covalently attached chromophore−fluorescein. Electron microscopy in similar systems showed that chromophore attached to one of the termini can follow peptide self-assembly without disrupting the fibrillization process.³³ In contrast, chromophore photophysics can be affected by shortening the distance between neighboring dyes upon fibrillization that leads to change of microenvironment, dye aggregation, and intermolecular interactions influencing the relaxation pathway of chromophores.³⁴ Thus, to investigate fluorescein in different conditions and in conjugation with peptides, we used primarily the time-resolved absorption technique as it shows large sensitivity to conformational changes in peptides and proteins.^35,36

MATERIALS AND METHODS

Fluorescein dye was purchased from Sigma-Aldrich and functionalized peptides with fluorescein attached at the N-termini through carboxyl group were purchased from Genscript. All samples were used as obtained. The peptides for optical experiments were dissolved in water buffers and adjusted to a concentration 1 mg/mL, agitated and incubated in elevated temperature at 37 °C for 48 h. Insoluble aggregates were spun and the exact concentrations were calculated using the molar absorptivity of a fluorescein cation 53 000 M⁻¹ cm⁻¹.¹⁸ Fresh made samples were diluted ten times for optical experiments and pH was readjusted after checking the absorption spectrum and matching with pristine fluorescein spectrum at the specific pH. The ultrafast transient absorption experiments were first conducted using following peptides (1) FGAILSS; (2) LVEALYL, NNQNTF; (3) VAVHVF. Then a “blind” test was performed and unknown peptides A, B, and C were aggregated in the same conditions and recorded dynamics compared with results obtained for structures 1, 2, and 3. Proper matching validate described methodology for architecture recognition.

Fibril formation was checked by electron microscopy and β-sheet structure was confirmed by FTIR and diffraction patterns (Figure S1 and S2).

Drop casted samples on glass slides were tested using Thermo Nicolet iS10 FTIR Spectrometer equipped with Smart Diamond ATR accessory. Working in the mid infrared range of 4000 to 500 cm⁻¹.

Samples for fiber diffraction and electron microscopy were prepared by the following steps. Peptide stock solutions were diluted to 5 mg/mL in 0.1 M phosphate citrate buffer pH 4.0, yielding final buffer concentrations from 70 to 85 mM and total volumes ranging from 128 to 403 μL. Samples were shaken in microcentrifuge tubes for 14 h at 37 °C.

To prepare the fibers for alignment and diffraction, 50 μL of each sample was washed twice to remove any salt that might yield unwanted noise to the background. The washing procedure consisted of centrifuging the samples at 14 000 rpm at room temperature in a table top centrifuge for 15 min, removing all supernatant, and then resuspending the pellet in 50 μL water. After the second round of washing, the pellet was resuspended in 5 μL of water, and the suspension was pipeted between two glass rods, and allowed to dry at room temperature. Diffraction patterns from the aligned fibers were collected at the Advanced Photon Source, beamline 24-ID-C, using an X-ray wavelength of 0.9202 Å, 100% transmission, 1 s exposure, and 450 mm sample-to-detector distance. The detector was a Pilatus 6M-F pixel detector.

To prepare the samples for electron microscopy, the original fiber concentration was diluted 1:20, and 2 μL was applied to a copper grid (Ted Pella, Inc., 400 mesh, Formvar/Carbon, Product number 01754-F) that had been glow discharged. The samples were stained with uranyl acetate and images were collected on a T20 iCorr FEI electron microscope.

Luminescence lifetime measurements were performed using the time-correlated single photon counting (TCSPC) technique.³⁷ Approximately 100 fs excitation pulses with wavelength 400 nm were generated by doubling the fundamental frequency of femtosecond Ti:sapphire laser (Spectraphysics Tsunami) pulses in a commercial optical harmonic generator (Inrad). The laser repetition rate was reduced to 2 MHz by a homemade acousto-optical pulse picker to avoid saturation of the chromophore. The TCSPC system is equipped with a thermoelectrically cooled single-photon counting avalanche photodiode (Micro Photon Devices) and electronics board (Becker and Hickl SPC-630) and has an instrument response time of about 30−40 ps. The triggering signal for the TCSPC board was generated by sending a small fraction of the laser beam onto a fast (400 MHz bandwidth) Si photodiode (Thorlabs Inc.). The fluorescence signal was dispersed in an Acton Research SPC-500 monochromator after passing through a pump blocking, long wavelength-pass, autofluorescence-free, interference filter (Omega Filters, ALP series). In addition to the time-resolved detector, the monochromator was equipped with a CCD camera (Princeton Instruments PIXIS-400) allowing for monitoring of the time-averaged fluorescence spectrum. Luminescence transients were not deconvolved with the instrument response function because their characteristic time constants were much longer than the width of the system response to the excitation pulse.

Transient absorption (TA) measurements were performed using a home-built setup similar to the one described in ref 38. The output of a Ti:sapphire regenerative amplifier (Coherent Astrella, 800 nm, 120 fs, 1.2 mJ pulses with 5 kHz repetition rate) was divided with a 10/90 ratio into probe and pump beams, respectively. The probe light was attenuated to microjoule level and focused onto 2 mm thick crystalline sapphire plate in the [0001] direction to generated supercontinuum. The femtosecond white light was focused onto a sample by a parabolic mirror. The usable spectral range for the supercontinuum probe was 430−750 nm. The pump beam was routed into an optical parametric amplifier (OPA, Light Conversion TOPAS Prime with NirUVis harmonic unit). The OPA output was tuned to 430 nm to match the lowest energy absorption peak in the samples. The pump beam was aligned through a corner cube reflector mounted on a computer-controlled translation stage with 20 cm travel range and 1 μm positioning accuracy (Newport Corp.) to control the delay between the pump and the probe pulses. The pump and probe beams were focused onto the sample and overlapped spatially at a small angle. A neutral density filter wheel (New Focus Inc.) was used to adjust the pump pulse energy. After passing through the sample, the probe beam was dispersed by a Cherny-Turner monochromator (Acton Research SP-300) and detected by a Si photodiode (Oriel) connected to a current preamplifier (SRS SR570). The photodiode signal was then fed into a lock-in amplifier (SRS SR830) synchronized to the first subharmonic of the laser repetition rate (2500 Hz). An optical chopper (New Focus Inc.) clocked by the same frequency was used to modulate the pump beam during the differential transmission measurements. A second lock-in amplifier synchronized with the full repetition rate of the laser was used to monitor the probe intensity. The details on the measurement method and chirp correction can be found in ref 38. The instrument response fwhm was measured using nondegenerate two-photon absorption in a crystalline sapphire slide and was found to be about 100 fs with a minor variation at different probe wavelengths.

The samples for TA spectroscopy were placed into quartz cells with optical path length of 1 mm. To avoid observation of pump-dependent effects, such as singlet−singlet annihilation or photochemical reactions, TA transients were acquired at several, gradually reduced levels of the pump pulse energy until pump-dependent components of the TA signal disappeared.

Dynamics were fitted across the whole decay and analyzed using OriginPro 8.5 software. The data were fitted to a multiexponential decay of the form

$$I\left(t\right)=A+\sum _{i=1}^n{B}_i{\text{e}}^{-t/{\tau }_i}$$

where τ_i is the transient absorption dynamics, B_i is the amplitude of the ith decay component, and A is the background noise level. The quality of fits was determined by the reduced χ².

UV−vis absorption spectra were collected using Shimadzu UV-2401PC spectrophotometer and PL spectroscopy measurements were conducted on Photon Technology International QuantaMaster fluorometer.

RESULTS

Steady-State and Time-Resolved Absorption Spectroscopy of Fluorescein at Different pH Levels.

Fluorescein in solution can exist in four main prototropic forms (cation, neutral, anion, and dianion) in the pH range from 2 to 12 and, depending on the exact conditions, can exist in an equilibrated mixture between the several forms that make the chromophore itself a complex molecule (Figure 1a). Figure 1b shows the individual absorption spectra that were extracted by measuring UV−vis absorption at different pH values. In very acidic (pH < 1) and basic (pH > 10) environments, the UV−vis absorption spectra of samples match those of cationic and dianionic species, respectively. For intermediate pH values the spectra of neutral and anionic species can be extracted easily given that at pH = pK_ai, where i is the index denoting transitions between different ionic states, relative abundances of two prototropic species are equal.

Figure 1.

(a) Various pH-dependent ionic species of fluorescein and prototropic transition in aqueous solutions; (b) deconvoluted spectra of pure dianion, anion, neutral, and cation ionic species; (c) experimental absorption spectra of fluorescein at different pH values; (d) calculated dependence of the concentration of different ionic species with marked points representing the pH values at which fluorescein and fluorescein−peptide conjugates were studied (at pH 12 a pure dianion is present).

Because each prototropic form has different absorption spectrum and PLQY that in consequence lead to a strong variation in the fluorescein photophysics, the UV−vis absorption spectra of actual samples were measured and shown in Figure 1c. Most of the experiments with peptides were conducted at pH = 2 and in this particular case a shoulder in the absorption spectrum at about 475 nm appears due to the presence of neutral species. The other two tested pH values were pH = 7 that is extended toward the blue end of the spectrum compared to that at pH 12 due to the presence of anionic fluorescein. The concentrations of different prototropic species vs pH shown in Figure 1d were calculated on the basis of the pK_a values published in the literature.¹⁸

The reason for such strong differences among ionic species of the chromophore is not fully elucidated, but there is some indication that symmetry of the molecule can be responsible for the optical properties. For example, it was shown that the anion form loses symmetry with respect to the dianion and the angle between the plane of the xanthene moiety and the plane of the benzoate ring is changed significantly, from 90° to 70°.³⁹ It leads to distinct changes in the spectrum of skeleton vibration modes, which can strongly affect the transition moments between different states and consequently influence the optical properties in the ground as well as excited state (Figure 2). The fact that a subtle change in the fluorescein structure can significantly affect its spectral properties makes this chromophore a promising agent to study β-strand organization in amyloid fibrils where structural differences are in the subnanometer range. In the case of peptide aggregation, the symmetry of the benzoate ring can be strongly affected and fluorescein can act as a sensor of intermolecular interactions when two chromophores appear in a close proximity. We use this advantage to probe fluorescein dynamics that is sterically perturbed by the particular amyloid architecture.

Figure 2.

(a) Transient absorption spectra at 1 ps delay (solid lines) and 50 ps delay (dashed lines) and (b) dynamics of fluorescein in different pH conditions: pH 12 (red), pH 7 (gray), and pH 2 (black). The experimental data are presented as either open squares (dynamics at 460 nm) or open circles (dynamics recorded at 530 nm).

First, careful examination of pristine fluorescein dynamics is required because those data are missing in the literature. This information is required for use as a reference in comparison with amyloid-conjugated fluorescein. Figure 2a shows transient absorption spectra of the dye at three different pH levels: pH 12, pH 7, and pH 2 at time delay 1 and 50 ps. In the case of the dianion form, the ground-state bleaching (GSB) occurs between 450 and 520 nm whereas in the region 525−620 nm the stimulated emission (SE) was detected (Figure S3). A similar case was observed at pH 7 where fluorescein exists in a mixture between the anion and dianion form. At pH 2 the TA spectrum turns negative around 570 nm, indicating that above that wavelength excited-state absorption can be detected in acidic conditions. Figure 2b represents the dynamics recorded at two wavelengths: 460 (GSB) and 530 nm (SE). The dianion form at pH 12 shows a linear 5000 ps decay that is similar at both studied wavelengths and reflects the nature of the dye in basic conditions. At pH 7 the three components were resolved (1.4 ± 0.6, 22 ± 7.1, and 4000 ± 500 ps) at 460 nm and two components (2.1 and 3500 ± 400 ps) at 530 nm. The appearance of shorter lived components in neutral conditions can be related to the anion form that exists in the mixture with the dianion at pH 7. At pH 2 the relaxation was as short as 2.85 ± 0.74 ps at both studied wavelengths. It is because the cation form is nonfluorescent and dominates in acidic environments whereas contributions from other species (e.g., neutral) are below the detection resolution of the pump−probe technique. The ultrafast dynamics at pH 2 can be attributed to deprotonation of the cation form to amphiprotic water solvent,¹⁹ the process that is responsible for the nonfluorescent nature of the cation form. The long-lived components (>1 ns) in neutral and basic conditions can be attributed to fluorescence emission and/or nonradiative relaxation from neutral/anion or dianion forms.

Fluorescein Dynamics in Conjugation with Peptide in a Nonaggregated and Aggregated State.

In the next step, fluorescein dynamics were analyzed when the molecule was conjugated to the peptide LVEALYL, a segment of the B-chain of insulin. This sequence was chosen because its aggregation can be controlled by pH and there is one tyrosine residue that may affect dye photophysics, as was previously suggested.⁴⁰ The LVEALYL segment of the protein insulin does not aggregate in basic conditions due to charge repulsion between the glutamic acid (E) residue and its solvent environment (pH 12). Adjusting the pH to a level between 6 and 8 causes the formation of morphologically distinct fibrils and acidic conditions at pH below 4 favors formation of mature fibrils.⁴¹ The fibril formation and β-sheet structure in aggregates was confirmed by electron microscopy, FTIR, and X-ray diffraction (Figure S1 and S2). The micrographs indicate also that fluorescein conjugation is not affecting the fibrils morphology.

The dynamics recorded for the fluorescein−LVEALYL conjugate at pH 12 in the ground-state bleaching region of 460 nm showed behavior similar to that of the pristine dye (Figure 3), indicating that amino acid composition and aromatic residues have no influence on fluorescein dynamics when the chromophore is covalently attached to the N-termini and the peptide remains in a nonaggregated state. Similar results were obtained in the stimulated emission region at 530 nm (Figure S4).

Figure 3.

Time-resolved absorption dynamics of pristine fluorescein (black) and fluorescein−LVEALYL conjugates (red) recorded at 460 nm: (a) dynamics at pH 12 where LVEALYL is in a monomeric state (inset: fluorescein−LVEALYL conjugate); (b) dynamics at pH 7 where LVEALYL forms intermediate aggregates; (c) dynamics at pH 2 where LVEALYL is in an aggregated state;, (d) transient absorption spectrum at pH 2 (aggregated LVEALYL). Labeled sample concentrations were (a) 4.5 μM, (b) 5.3 μM, and (c) and (d) 8.4 μM.

At pH 7 a noticeable difference in dynamics of fluorescein was observed when relaxation decays of dye conjugated to the peptide and the pristine chromophore were compared (1.4 ± 0.6, 22 ± 7.1, and 4000 ± 500 ps at 460 nm). Three components were resolved at 2.7 ± 0.3, 22 ± 7.1, and 3890 ± 200 ps for LVEALYL−fluorescein at pH 7. A possible explanation is that some initial aggregates are formed at neutral pH that affect the relaxation of dye molecules. Noteworthy, the signal also rises between 80 and 150 ps. This could be a side effect of the excited state of the fluorescein anion that when attachment to the peptide becomes unstable, it loses the second proton, thereby forming a dianion that has a higher transition moment (TA signal is proportional to the transition moment) and raises the signal in the picosecond time domain.

Likewise the insulin segment, LVEALYL, forms fibrils at pH 2 and its crystal structure shows that β-strands are organized in parallel in-register 2 β-sheets.³⁰ The aggregation caused a remarkable change in the fluorescein dynamics in comparison to the case for the pristine dye; complex relaxation was revealed with three components 0.19 ± 0.05, 16.9 ± 5.2, and 107 ± 22.4 ps (Figure 3 bottom panel). Particularly interesting is the appearance of a long-lived component that may indicate the change in the fluorescein lifetime when the dye aggregates with the peptide. To further analyze the effect of aggregation on fluorescein lifetime, time-correlated single photon counting (TCSPC) experiments were performed.

Time-Resolved Fluorescence of Fluorescein and Fluorescein−Peptide Aggregates.

To determine whether peptide aggregation affects the fluorescein photoluminescence emission lifetime and efficiency, we performed steady-state fluorescence and time- correlated single photon counting (TCSPC) experiments.

In general, photophysics of xanthene dyes, and fluorescein in particular, has been extensively studied in a monomer and aggregated state.⁴²⁻⁴⁴ Aggregation was usually recognized by red-shifting of the emission spectrum and noticeable changes in the fluorescence lifetime.⁴⁵

Because the differences in dynamics of pristine fluorescein and fluorescein conjugation with LVEALYL were most pronounced at pH 2, and because such low pH conditions were previously reported to be effective for promoting aggregate self-assembly, we assumed pH 2 is an optimal value at which to study the photophysics of other peptide−fluorescein conjugates, making data interpretation feasible. The inset of Figure 4 shows that steady-state emission spectra are red-shifted in conjugates as a consequence of aggregation.⁴⁶ The lifetime data of the chromophore alone and in conjugation with peptides representing each β-sheet architecture 1, 2, and 3 showed a monoexponential decay where fitting revealed a 2 ns component for pristine dye and 2.7 ns for fluorescein in conjugation with aggregated peptides (Figure 4 and Figures S5 and S6 including explanation and a table summarizing lifetime data). The fluorescein lifetime is significantly longer than the picosecond decays measured in pump−probe experiments at pH 2. The lifetime is related to resolution and sensitivity whereby TCSPC is capable of detecting the fluorescent species even at very low concentrations (that is, below the detection limit of pump−probe technique). At pH 2, a dominating form of fluorescein is a nonfluorescent cation while a small fraction of neutral species that is fluorescent is also present (Figure 1b). Thus, we assume that the photoluminescence emission with 2 ns lifetime is predominantly from the neutral form of fluorescein.

Figure 4.

Lifetime data of fluorescein recorded at 530 nm (black) and in conjugation with three different zipper structures: (1) antiparallel in-register (blue), (2) parallel in-register (red); (3) antiparallel out-of-register (green). All samples were measured at pH 2. The inset is the steady-state fluorescence spectra with an arrow indicating red-shifting of the spectrum when fluorescein is in conjugation with peptides in the aggregated state. The representative lifetimes are shown for aggregated peptides: FGAILSS, 7.2 μM (1); LVEALYL, 7.5 μM (2); ASLTVS, 5.4 μM (3) that had no aromatic amino acid in the sequence.

Longer lifetimes in aggregated fluorescein−peptide (compared to pristine fluorescein) can be explained by a change of the solvent microenvironment that can particularly affect the cation photophysics.¹⁸ The most plausible mechanism is the restriction of proton transfer to solvent molecules as a consequence of aggregation and solvent reorganization that is manifested with longer lifetimes.⁴⁷ The TCSPC results are then in good agreement with pump−probe data where the long-lived component was resolved only in the aggregated state, thus indicating a change in the deprotonation rate. But all studied zippers showed identical lifetime values, indicating that the cation form may have some contribution to the decay trend but eventually the neutral form dominates and is similar in each amyloid architecture. Therefore, the TCSPC operating in the pico- and nanosecond time domain provide limited access to the cation form, making this particular technique adequate to study amyloid aggregation but uninformative of fibril strand organization that requires a fluorescence up-conversion system for analyzing short-lived states in the femtosecond time regime.

Transient Absorption of Fluorescein−Peptide Aggregates.

To elucidate whether β-strand organization of fibrils affects the relaxation of the cation form of fluorescein, transient absorption was employed to probe chromophore photophysics in the three sheet geometries 1, 2, and 3. The intermolecular distance between the fluorescein molecules when it is conjugated with peptides is constrained to be within Ångstrom distances whereas the exact separation is determined by the β-sheet organization in the peptide aggregates. In parallel in-register fibrils (1) the fluorescein labels are spaced 4.8 Å apart whereas in antiparallel in-register 2 the spacing is doubled to 9.6 Å. In antiparallel out-of-register (3) sheet geometry the separation between fluorescein labels is 12 Å⁴⁸ (Figure 5a). The dimensions of the fluorescein molecule (13 Å × 11 Å)⁴⁹ is relatively large compared to the separation between the strands and guarantees strong coupling between chromophores.⁵⁰ Simultaneously, the X-ray data confirm that the interspace between β-sheets is not perturbed by fluorescein conjugation (Figure S2).

Figure 5.

(a) Schematic drawing of strand architectures in the three β-sheet architectures with a chromophore attached: (1) antiparallel in-register; (2) parallel in-register; (3) antiparallel out-of-register. (b) Dynamics of pristine fluorescein (black) and fluorescein in conjugation with peptides forming different β-sheet architectures 1 (blue), 2 (red), and 3 (green) at 460 nm. The inset of (b) represents a time domain window >15 ps. (c) Dynamics at 530 nm and (d) at 600 nm. Presented dynamics in the graphs are representative decays of fluorescein conjugated with GYVLGSA, 7.7 μM (1); NNQNTF, 8 μM (2); and VAVHVF, 9.1 μM (3). Fitting of all aggregated peptides is summarized in Table 1.

The ultrafast time-resolved absorption of fluorescein constrained in three different sheet geometries measured at a particular wavelength can provide information on chromophore−chromophore interactions in the aggregated state, solvent−chromophore interactions upon reorganization of water molecules during aggregation, and the nuclear motion of bound chromophores (Figure 5b–d).

To investigate the dynamics of fluorescein attached to the specific zipper structures as well as pristine fluorescein, three wavelengths were studied (ground-state bleaching at 460 nm and stimulated emission at 530 and 600 nm) to obtain a comprehensive picture of chromophore dynamics in different fibril architectures. Representative dynamics are shown in Figure 5b–d and fitting is summarized in Table 1 for reference samples 1, FGAILSS; 2, LVEALYL, NNQNTF; and 3, VAVHVF and for “blind” test samples GYVLGSA, GGNNQQNY, and ASLTVS. Good matching of the fitting curves between reference samples and the blind test indicate that transient absorption is providing reproducible results for specific amyloid architectures (Figure S7). Dynamics of fluorescein conjugated with different peptide sequences show common features in relaxation decays that can be attributed to chromophores organization in the fibrils.

Table 1.

Summary of Dynamics at pH 2 of Fluorescein Conjugated with Aggregated Amyloid Segments Forming Different β-Sheet Structures and the Relative Contribution of Each Component in Percent

sample	460 nm (GSB) ps	530 nm (SE) ps
fluorescein at pH 2	2.85 ± 0.74	2.84 ± 0.85
antiparallel in-register (1)
GYVLGSA	0.94 ± 0.30 (65%)	353 ± 104
	17.3 ± 6.38 (35%)
FGAILSS	0.82 ± 0.32 (62%)	375 ± 110
	23.1 ± 4.83 (38%)
parallel in-register (2)
LVEALYL	0.19 ± 0.05 (30%)	342 ± 45
	16.9 ± 5.2 (37%)
	107 ± 22.4 (33%)
NNQNTF	0.68 ± 0.11 (33%)	314 ± 98
	16.7 ± 4.45 (41%)
	238 ± 102 (26%)
GNNQQNY	0.18 ± 0.06 (32%)	321 ± 107
	3.35 ± 0.64 (35%)
	74.8 ± 14.2 (33%)
antiparallel out-of-register (3)
VAVHVF	1.20 ± 0.12 (53%)	690 ± 230
	340.6 ± 49.5 (47%)
ASLTVS	0.24 ± 0.06 (51%)	730 ± 189
	309.3 ± 71.5 (49%)

Unlike in pristine fluorescein where only fast decay (2.84 ps) was recorded at 460 nm, the dynamics in group 1 fibrils required fitting with at least two components (0.94 ± 0.30 and 17.3 ± 6.38 ps for GYVLGSA and 0.82 ± 0.32 and 23.1 ± 4.83 ps for FGAILSS), which is related to a change of microenvironment and deprotonation rate due to aggregation of conjugates.

Even more complex relaxation was recorded for parallel in-register fibrils 2 NNQNTF and GNNQQNY where three components were resolved: 0.68 ± 0.11, 16.7 ± 4.45, and 238 ± 102 ps (NNQNTF) and 0.18 ± 0.06, 3.35 ± 0.64, and 74.8 ± 14.2 ps (GNNQQNY). Comparing to group 1 antiparallel in-register fibrils, in group 2, a longer lived component appeared, indicating rearrangement of covalently attached chromophores that lead to the occurrence of emissive states in the fluorescein cation.

The third group is antiparallel out-of-register fibrils 3 VAVHVF and ASLTVS that also exhibit a fast decay and a long-lived component (VAVHVF, 1.20 ± 0.12 and 340.6 ± 49.5 ps; ASLTVS, 0.24 ± 0.06 and 309.3 ± 71.5 ps). Also in group 3 the multiexponential decay can be attributed to changes in deprotonation rates (short-lived components) and conversion of fluorescein cation into emissive species (long-lived component).

By comparison of the dynamics of all three fibril groups with that of pristine fluorescein, it becomes clear that a strong coupling between chromophores must occur in the processes of aggregation for all samples studied and through-space interactions lead to significant changes in rates required for repopulating the ground state. Because the differences in fluorescein dynamics can be resolved only in the ultrafast time domain, we conclude that the cation deprotonation process is most affected, but to a different extent depending on the β-sheet geometry adopted during aggregation.

As a consequence of the change in deprotonation rates, the analysis in the stimulated emission region at 530 nm provides the information on solvent−chromophore interactions. The dynamics of three amyloid architectures exhibit 1 order of magnitude longer decays than the pristine dye and were on average 360 ± 100 ps (1), 320 ± 65 ps (2), and 710 ± 190 ps (3). The results indicate that solvent molecules were reorganized in the hydration shell of aggregates⁵¹ and restriction of proton transfer led to activation of emissive states in the cation form.⁴⁷ This observation is consistent with longer fluorescence lifetime data measured at the same wavelength but with the difference that the relaxation decays in pump−probe experiments are amyloid architecture dependent.

Further analysis of dynamics at 600 nm shows peptide structure-dependent rise of signals that can be explained using the Franck−Condon principle. It assumes that electronic transitions in both directions are much faster than nuclear motion; thus they are usually shown as vertical lines on the energy-nuclear coordinate diagram. The dynamics at 600 nm mostly refer to the emissive states that are the resulting products of the nuclear reorganization following the absorption of the photon. In each type of aggregates 1, 2, or 3, certain degrees of the nuclear motion may be restricted, affecting the rate of vibrational relaxation and formation of the state producing the fluorescence.

The dynamics at the ultrafast domain were 0.55 ± 0.1 ps for pristine dye and were visibly longer in the dye conjugated with 1 (0.78 ± 0.25 ps), 2 (3.1 ± 0.9 ps), and 3 (2 ± 0.4 ps). The transfer of magnetization between different nuclear spins depends on dipole−dipole (through-space) interactions and surroundings of the sample molecules.⁵² The larger the space between the nuclei, the less efficient is the interaction. Thus, the dynamics measured at 600 nm show that nuclei of different chromophores interact with each other in a slightly different manner in amyloid architectures.

DISCUSSION

Unlike electrostatic dyes that were reported to influence the aggregation of protein and peptides,⁵³ the impact of covalently attached chromophores on aggregation was studied to a lesser degree. Existing reports indicate that even large and bulky dyes do not interfere with aggregation, and only after long incubation time, some morphological changes can be observed.⁵⁴ Fluorescein derivatives were previously confirmed to be good candidates to study the proteins and peptides aggregation. Time-resolved and nonlinear spectroscopy were used in many configurations to analyze the structural organization of modified peptides with dyes.^45,55 The overall results obtained with applicable methodology for amyloid fibrils structure analysis indicate that fluorescein derivatives do not affect the aggregation even of short peptides. With respect to the existing literature, we suggest that nonlinear and ultrafast spectroscopy can be potentially used to analyze strands organization in amyloid fibrils.

In the present case, to elaborate on differences in recorded dynamics in amyloid−fluorescein structures, we adapt the dimer model developed by Arbeloa et al. for the pristine dye⁴² and take into consideration that the cation is sensitive to solution organization that depends on amyloid architecture. The pristine dye monomers in free dimer are in parallel planes forming an angle of 76°. In the case of amyloid−fluorescein, the fibril architectures 1, 2, or 3 sterically determine the exact angle between neighboring chromophores. The dynamics then depend both on that angle and on the symmetry of the two molecules, where the angle formed by the pair of benzoate rings with respect to the tilted xanthene moieties is most likely responsible for specific relaxation decay.

Also interesting is the relatively long decay time of fluorescein in conjugation with fibrils from groups 2 and 3. We propose that this length might be related to changes in the hydration shells^56,57 where group 2 fluorescein molecules are densely packed (only 4.8 Å separations between strands) in a ladder-type structure and access of water molecules is limited. Noteworthy, the interspace distance between chromophores in fibrils in group 2 is nearly identical with the the distance in the fluorescein dimer where two chromophores are at a distance of 4.6 Å.⁴² Therefore, a redistribution of solvent molecules must occur around the chromophores as well,⁵⁸ and the local environment becomes more hydrophobic,^59,60 creating new conditions for energy distribution in the ground and excited state that are considerably different from those in the monomer dye or even different from those of fibrils from group 1 where the interspace is twice as large (9.6 Å) and more access to water is plausible. In line with this hypothesis is the explanation of the long decay time in group 3 fibrils; in this particular architecture where sheet curvature into barrel-like structures have been observed, the chromophores can be buried inside the barrel, which shields the chromophore from solvent molecules. The cation form of fluorescein is particularly vulnerable to changes in the solvation environment and cybotactic region whereas reduced access of water molecules to the chromophore significantly decreases the rate of proton transfer from cationic species to water acceptors.

As a consequence of different rates of deprotonation, reorganization of solvent molecules in the hydration shells, and distance dependent through-space coupling of neighboring fluorescein molecules, the occurrence of different relaxation pathways is highly probable.

In summary, the ability of time-resolved absorption to analyze ground and excited states allows one to collect information on the molecular arrangement of chromophores in the fibril structures. Taking together the three wavelengths 460, 530, and 600 nm, our work suggests that, by using the microenvironment sensitive cation form of fluorescein, the time-resolved absorption methodology can be used to differentiate the fibril architecture by monitoring chromophore−chromophore interactions and their dynamics. That was confirmed by performing the “blind” test where unknown peptides A, B, and C were aggregated and decaying trends were compared with the dynamics at each wavelength obtained for structures 1, 2, and 3. Proper matching of fitting curves (within the errors due to scattering) validates that chromophore organization in “blind” test samples is the same as in the reference structures.

CONCLUSIONS

In this paper we analyze the dynamics of fluorescein in relation to aggregating peptides. Overall results confirm that chromophores organization is determined by the fibrils architecture and that relaxation depends on the rate of intermolecular interactions and solvation shell around the fluorescein−fibrils conjugates. The results indicate that ultrafast time-resolved absorption spectroscopy is capable of following the deprotonation process of the fluorescein cation that can be helpful in resolving structural differences in amyloid fibrils. In this study, the particular emphasis was put on differentiating three amyloid geometries: 1, antiparallel in-register; 2, parallel in-register; 3, antiparallel out-of-register. It may be that one of these is preferentially involved in neurodegeneration⁶¹ and this methodology may help to resolve the early stages of its aggregation. Relying on available crystallographic structures, we show that transient absorption can be useful in distinguishing different fibril structures in a solution environment that is closer to physiological conditions than the crystalline state. These results are consistent with the general efforts made on developing tools such as pump−probe microscopy⁶² or systems for single molecule detection⁶³ for investigating the structure and dynamics of molecules relying on the absorption based methods.