Failure of some proteins to adopt or maintain their native conformational state may result in a number of disorders, grouped under the name of protein misfolding diseases. These misfoldings usually lead to the conversion of detergent-soluble forms into amyloid fibrils—highly organized and generally intractable protein aggregates enriched in -sheet structure.1 One of the causes of such aggregation is a mutation in certain genes, which result in the expanded repeats of CAG trinucleotide, encoding glutamine. The abnormal expansion of the polyglutamine tract (polyQ) leads to misfolding, aggregation, inclusion body formation, and eventual neuronal cell death in nine inherited neurodegenerative disorders, including Huntington’s disease, dentatorubral pallidoluysian atrophy, spinal and bulbar muscular atrophy, and several types of spinocerebellar ataxia.2, 3, 4, 5 Efficiency of aggregation increases with the length of polyQ track in yeast and mammalian cells, whereby polypeptides with less than 39 glutamine residues fail to aggregate at all.6, 7, 8 Consistent with this notion, in Huntington’s disease polyQ repeats are expanded from normal 8 to 25 to more than 39 at the -terminus of the huntingtin protein.9, 10
Despite the fact that pathogenicity of the polyQ aggregation has been known for some time, there is poor evidence in misfolding diseases management and development of effective therapies requires better understanding of aggregates formation.11 Several fluorescence spectroscopy techniques have been applied for this purpose, such as fluorescence microscopy,12 fluorescence correlation spectroscopy,13, 14 and Förster resonance energy transfer (FRET) imaging microscopy. One of the most popular techniques, FRET has been applied15, 16, 17, 18, 19, 20 to study the involvement of other proteins in the aggregates formation and discrimination between mono- and oligomers of polyQ (also reviewed in Ref. 21). The technique is based on the nonradiative energy transfer from the excited donor molecule to another fluorescent moiety, the acceptor. For the transfer to happen, the molecules must be in close proximity (within the range of ), which makes FRET an ideal technique to study interactions between molecules.22
The fluorescence spectroscopy techniques are based on the labeling of the protein(s) of interest with further monitoring of the fluorescence signal properties and distribution. However, possible complex interactions between many fluorophores in close proximity, as well as the change of their microenvironment in aggregates, can influence the resulting photophysical properties and thorough characterization of such changes is required for the correct interpretation of the data.
In this work, we have examined the changes caused by aggregation to the average time that fluorophore molecules spend on the excited state levels on photon absorption and prior to the relaxation to the ground state—the fluorescence lifetime. At this, the relaxation through photon emission competes with a number of alternative pathways of excessive energy release given that they happen in the time scale comparable with the fluorescence lifetime. Among these are protonation, electron transfer, collisional quenching, and nonradiative energy transfer.23 In FRET, for example, the energy transfer from donor to acceptor creates an additional pathway for the depletion of the donor’s excited state levels, thus shortening the radiative decay and correspondingly lowering the fluorescence lifetime values.23 This sensitivity of the technique made it possible to probe the environment of proteins in cells in terms of its refractive index,24 pH (Ref. 25), and ion concentration,26 to perform dynamic imaging of the redox state,27 to follow signaling and trafficking,28 and to monitor protein localization and interactions.29 Fluorescence lifetime microscopy (FLIM) can also be applied for the monitoring of conformation, especially30 in combination with FRET. Previously, FLIM has also been applied in the studies of misfolding proteins.31, 32, 33
One of the most applied fluorescent markers, the green fluorescent protein (GFP) is a 238-amino-acid-long polypeptide, isolated from the jellyfish Aequorea victoria.34, 35 The chromophore of the wild-type GFP is a -hydroxybenzylidene-imidazolidinone derivative formed by an autocatalytic posttranslational cyclization and oxidation of Ser-65, Tyr-66, and Gly-67 (Ref. 36). Located in the middle of a rigid 11-stranded barrel, the chromophore is rigidly embedded in the protein matrix and thus is isolated from the bulk solvent. The barrel shielding makes the GFP almost completely insensitive to the solvent polarity, ionic strength, and viscosity. The convenience of GFP comes from the autocatalytical nature of its chromophore synthesis, so that the expression of the fusion gene of GFP and a protein of interest in a cell requires no additional cofactors, thus minimizing the impact on the cell environment and making the operation easy.
Despite the tremendous potential of the wild-type GFP, it has a few deficiencies. The formation of dimers in some conditions complicates the results interpretation and might also affect the labeled protein functionality.37 Moreover, from the two absorption peaks of wild-type GFP, the one at has higher amplitude at lower photostability as compared to the more stable peak at . A number of mutants with improved fluorescence characteristics of the GFP have been developed. A popular variant enhanced GFP (eGFP) holds a double mutation of and with the former shifting the absorption spectrum to the blue range and increasing brightness, and the latter supporting protein folding at high temperatures.38 Having the same structure with the GFP, this enhanced variant of the fluorescent protein is, however, sensitive to the acidic environment25 . This variant has been demonstrated as a well-expressed monomer and gained wide acceptance.39
We examined the changes in the fluorescence lifetime dynamics of eGFP fused to 97Q protein (97Q-eGFP) by means of FLIM in model yeast cells and found that the lifetime of the fluorophore is decreasing on aggregation. To properly assess the influence of the aggregation of the labeled 97Q on the fluorescence lifetime of eGFP, we identified other possible causes for the radiative decay dynamics. We assessed the contribution of the pH level, direct nonradiative transfer to fluorescent metabolites, and level of cellular autofluorescence and environmental changes caused by physiological cells conditions on the dynamics of the eGFP fluorescence lifetime. We also investigated existence of FRET between eGFP molecules in 97Q aggregates. The possible influence of such nonradiative transfer and refractive index on the phenomenon observed is discussed.
Materials and Methods
Plasmid and Yeast Strain Constructions
The pYES2-97Q-eGFP plasmid was based on the pYES2-103Q-eGFP, kindly provided by Sherman, was described in Ref. 40. A reduction of six glutamine residues in the expected repeat region was found on conducting the DNA sequencing of the plasmid, hence the new construct pYES2-97Q-eGFP, used throughout all the subsequent experiments. The pYES2-eGFP plasmid construct was derived from pYES2-97Q-eGFP by deleting the DNA region encoding 97Q. With polymerase chain reaction (PCR), the DNA fragment covering Gal1 promoter and Flag tag of pYES2-97Q-eGFP was amplified using EcoRI-Gal-97Q-ADE2 primer: GGAATTCCGCTTAATGGGGCGCTACAG and Gal-Flag-SmaI primer: TGCCTCCCGGGTTTGAGGGACTCGAAGGCCTTC. The PCR product was digested with Kpn I and Sma I, and then replaced the Kpn I-Sma I fragment of pYES-97Q-eGFP to create pYES2-eGFP. Each of the two plasmids (pYES2-97Q-eGFP and pYES2-eGFP) was transformed into Saccharomyces cerevisiae W303-1a (MATa ade2-1 trp1-1 ura3-1 leu2-3,112 his3-11,15 can1-100).
To make the hsp104 knock-out strain , the entire coding sequence of HSP104 gene was replaced with a KanMX gene by the one-step gene disruption method.41 The targeting DNA was amplified by PCR (HSP104-KAN primer: ATGAACGACCAAACGCAATTTACAGAAAGGGCTCTAACGATTTTGACGTTAGCTTGCCTCGTCCCCGCCGG and HSP104-KAN primer: TTAATCTAGGTCATCATCAATTTCCATACTGTCCTCATTATCGTCATCACTCGACACTGGATGGCGGCGTTAG, plasmid pUG6 as a template) and transformed into S. cerevisiae W303-1a. Transformants were selected on YPDA/G418 plates and the strain was confirmed by genomic PCR.
Development of Ade2-Proficient Strains
Both wild type W303-1a (wt) and cells had characteristic adenine synthesis blockade caused by Ade2 deficiency, which led to the accumulation of intermediates, absorbing in the range of eGFP’s fluorescence. To check the nonradiative transfer from eGFP, the adenine metabolism was restored by turning the cells into Ade2-proficient by introducing the pRS424-Flag-ADE2 plasmid construct.
For this purpose, the ADE2 gene was amplified using BamHI-Flag-ADE2 primer: CGGGATCCATGGACTACAAGGACGACGATGACAAGCTGGCGACCCTGGAAAAGCTGGATTCTAGAACAGTTGGTATA and NheI-ADE2-HA primer: CTAGCTAGCTTAGATGATGGCGTAGTCGGGCACGTCGTAGGGGTAATGGCGGCCGCCCTTGTTTTCTAGATAAGCTTC in a PCR reaction. Following BamH I and Nhe I digestion, the resulting fragment was ligated into the pYES2 plasmid, which was linearized by BamH I and Xba I to create pYES2-Flag-ADE2. Using this plasmid as a template, the Flag-tagged ADE2 gene as well as the upstream GAL1 promoter was subsequently subcloned into pRS424 plasmid vector to generate pRS424-Flag-ADE2.
Cell Growth Assay
For the assessment of 97Q cytotoxicity, a cell growth assay was carried out. Cells were precultured in synthetic dropout media containing 2% dextrose at . An appropriate amount of the cell suspension was then transferred to a fresh medium that contained 2% galactose as the sole carbon source to induce the expression of eGFP or 97Q-eGFP. After culturing, the cell density usually reached and the cells were mounted on an agarose slide to measure the cellular distribution of eGFP and its fluorescence lifetime. The influence of Ade2 deficiency was eliminated by reintroducing the expression of the ADE2 gene from pRS424-Flag-ADE2 in these cells.
Imaging System Setup
The measurements were conducted on the time-resolved two-photon fluorescence spectroscopy system, built around the modified laser-scanning microscope (IX71, Olympus Corp.). The sample was excited at by a mode-locked femtosecond linearly polarized Ti:sapphire Mira F-900 laser (Coherent, Inc.), operated in two-photon mode at a frequency. The wavelength choice is explained with the lowest autofluorescence level detected at this excitation band. The intensity of the laser power was kept within above the objective. The beam was fed into the scanning unit (FV300, Olympus Corp.) and scanned across the sample at the speed controlled externally by a function generator (AFG310, Tektronix Inc.). The numerical aperture (NA) PlanApochromat lens (Olympus Corp.) focused the beam at the cell samples and collected the fluorescence photons, which were delivered in a nondescanned mode to the photomultiplier tube (H7422-P40, Hamamatsu Photonics K.K.) mounted at the backport of the microscope. A filter ( , Semrock) was installed along with the IR cut-off filter (Edmund Optics, Inc.) in the detection channel for the selective acquisition of eGFP fluorescence. Signal synchronization and building of the time-resolved data matrix were conducted by the SPC-830 time-correlated single photon counting (TCSPC) board42 (Becker&Hickl GmbH). All the images were taken at resolution with the acquisition time varying in the range of . No significant change of the photon count rate was detected during the measurements.
For the anisotropy imaging, the horizontal polarization of the laser was utilized without additional polarizers in the excitation pathway. A Glan-laser polarizer (GL10, Thorlabs Ltd.) with a coating wavelength range of was installed in the detection pathway. The measurements in the polarization plane parallel and perpendicular to that of the laser was taken successively by manually rotating the mounted polarizer. To decrease the depolarization caused by the optics, an objective of lower NA (0.9, ) was used for anisotropy imaging. Photon distribution was registered with the TCSPC board with the same settings as for the fluorescence lifetime imaging.
Data analysis via model function fitting, instrument response function (IRF) deconvolution, and color coding were conducted with the commercially available SPCImage software package (v. 2.9, Becker&Hickl, GmbH). A mathematical model was fitted to the experimental data by iterative non-least-squares reconvolution:is the instrument response, is the baseline offset, and are the main components of the fluorescence decay. The preexponential coefficients have relative amplitudes and demonstrate the correlation between components fractions populations, are the lifetimes of all the components to the decay (in case of multiexponential one). The IRF of the setup was measured from the second harmonic generated by the periodically poled lithium niobate (PPLN) crystal. The measured full width at half maximum of the IRF was and represented the IRF of the whole system.
The quality of fitting was judged by the residual function distribution and reduced goodness-of-fit parameter calculated by the equationis the actual experimental data, is the calculated decay as already described, is the number of the data (time) points (256 time channels for the given work), and is the number of model parameters. The was minimized by the Levenberg-Marquardt search algorithm.
To remove the background and signal from dead cells with high autofluorescence levels either the regions of interest were marked so that only the pixels inside the region were used to generate the lifetime distribution histogram, or a threshold parameter was used. A pixel binning factor of 2 was applied in our analysis, so that a signal from 25 adjacent pixels was used to derive a single decay curve for further fitting. All the pixels within the specified region and with the number of the photon counts higher than the threshold defined were then analyzed and pixel frequency-weighted lifetime distribution histograms were built.
Fluorescence anisotropy data were analyzed by the custom developed program for MATLAB software package (The Mathworks, Inc.). The program parsed the TCSPC-generated files of the and perpendicular measurements, fetched the photons distribution data and summed together the number of the photons for all 256 time channels in each pixel. The numbers thus obtained were then used for the calculation of steady state anisotropy:is the correction factor applied to account for the differences in the sensitivity of the detection optics for vertically and horizontally polarized light. For our case, the value of 1 was applied since no monochromator was used and both polarization plane signals were detected with the same channel.
Statistical processing of the data was carried out with the single-factor analysis of variance (ANOVA). The similarity and diversity of the data sets were conducted by comparing the peak values of the fluorescence lifetime histograms recorded from the images.
The wt and cells with the expressed 97Q-eGFP were grown in an adenine-restricted medium (adenine final concentration of ) for , resulting in red ade2 dye overproduced and accumulated in vacuole.43 The cells were mounted on the agarose pad of a single-cavity slide and visualized by the fluorescence microscope Axioskop 2 (Carl Zeiss Microimaging Inc.) with a oil immersion objective. The differential interference contrast (DIC) images were taken for phase contrast. eGFP was excited at , and ade2 dye was excited at .
eGFP Fluorescence Lifetime Decreases in 97Q Aggregates
To study the dynamics of the eGFP fluorescence lifetime in aggregates, the fluorophore was lagated with 97Q protein with a linker of seven aminoacids. The fluorescence decays recorded for eGFP, both expressed solely and fused to 97Q protein, were well fitted with a single-exponential model, as judged by the distribution of residual functions and goodness of fit (Fig. 1 ). Although for the 97Q-eGFP expressed in wt cells, further denoted as 97Q-eGFP(wt), the goodness of fit appeared slightly increased as compared to the free eGFP in same cells [eGFP(wt)], application of a double-exponential decay model did not improve the fitting quality. The difference between the decays was well distinguishable with the one measured from the 97Q-eGFP(wt) having steeper slope.
The change in the eGFP lifetime behavior at fusion to the 97Q is well illustrated with the distribution of the averaged histograms, collected from the series of images [Fig. 2 ]. The histogram of eGFP(wt) stretches from with the peak value mean of . Ligation of eGFP to 97Q resulted in the shift of the histogram to the shorter values region. It appears slightly wider with the stretch from and the maximum mean of . The two histograms obtained are well separated and the peak value data set distribution is significantly different .
The eGFP(wt) and 97Q-eGFP(wt) also exhibited different distribution patterns in yeast cells (Fig. 3 ). Free eGFP had a diffused distribution with the exception of the vacuole, where little to no signal was detected [Fig. 3]. In contrast to that, the 97Q-eGFP exhibited grain- and cloudlike patterns, indicative of polyQ tract aggregation [Fig. 3]. With the given color coding it is evident that the aggregates exhibit shorter lifetime than the surrounding.
To reveal the role of the aggregation in the lifetime dynamics observed, the 97Q-eGFP was expressed in the strain, where the 97Q aggregation was prevented [hereinafter referred to as ]. Intracellular protein aggregation is a complex process involving microtubule-mediated protein transport to the centrosome with further formation of the aggresomes, and regulated by chaperones and ubiquitin-proteasome degradation pathway components.44, 45, 46, 47 Among major factors involved in this process are heat shock proteins, which restore native conformation and support proper interaction of proteins aggregated at high temperature or as a result of environmental stress. In particular, overexpression of heat shock proteins Hsp70 and Hsp40 inhibits the formation of large detergent-soluble inclusion bodies, leading to the conversion of the aggregates into soluble forms.48 However, for Hsp104, an opposite effect was demonstrated: under a normal growth condition, this protein can assist the formation of nonpathological aggregates, prions, and is essential for polyQ protein aggregation in yeast cells. In the absence of this chaperone, aggregation of prions and polyQ is prevented and these proteins become soluble.8, 40, 49, 50
At aggregation prevention, the averaged lifetime histogram appeared narrower than that for the 97Q-eGFP(wt), and shifted back toward the lifetime values of eGFP(wt) with the peak mean of [Fig. 2]. Similar to free eGFP, the appeared evenly distributed in the cytoplasm of cells [Fig. 3], indicative of the lack of aggregation.
Red Metabolites: Influence of Hetero-FRET and pH
The ade2 mutation, characteristic for the W303-1a strain, used in the experiments, is known to block adenine synthesis in cells and lead to the accumulation of intermediate metabolite 4-aminoimidazole ribotide. This colorless compound is synthesized in cytosol with further concentration of its derivatives in the vacuole, where it undergoes polymerization and oxidation, producing a red color.51 The aminoimidazole and its derivatives were demonstrated in previous works to absorb at the wavelengths range of . Therefore, there was a possibility of direct nonradiative energy transfer from eGFP, given the overlapping of its emission and absorption spectra of these metabolites.52 This may have caused the decrease of the eGFP lifetime as a donor.29
To examine the possibility of this energy transfer, we conducted control measurements with restored adenine metabolism by expressing Ade2 protein in W303-1a cells. The fluorescence lifetime of 97Q-eGFP in Ade2-proficient cells, hereinafter denoted as 97Q-eGFP(Ade2), exhibited peak value of , which appeared statistically identical to that of 97Q-eGFP(wt) with [Fig. 2, Table 3 in Sec. 4]. Similar to that, the fluorescence lifetime value of obtained for the 97Q-eGFP in Ade2-proficient cells was indistinguishable from the results recorded form . The similarity applies also to the spatial distribution of the 97Q-eGFP in Ade2-proficent cells: grains with shorter fluorescence lifetime in wt cells [Fig. 3], and evenly distributed pattern of longer lifetime for 97Q-eGFP in the absence of the Hsp104 chaperone [Fig. 3].
Distribution of the P values calculated by a single-tailed ANOVA test.
|eGFP(wt)||97Q-eGFP(wt)||97Q-eGFP(hsp 104Δ )||97Q-eGFP(wt Ade2)||97Q-eGFP(hsp 104Δ Ade2)|
The red ade2 pigment was also utilized in this study to investigate the influence of low pH on the eGFP lifetime values. Misfolded and aggregated proteins can hamper cellular function, and these proteins normally are degraded by lysosomes or proteasomes.53 In yeast cells, the degradation occurs in vacuoles—organelles with acidic lumen,54 which may shorten the eGFP lifetime. To investigate the contribution of pH level to the fluorescence lifetime dynamics observed, we checked the presence of eGFP in vacuole by colocalization experiments. The cells were grown in adenine restriction medium to mark the vacuole with the ade2 red pigment. The results, presented on Fig. 4 , clearly show no overlapping of the eGFP signal with the red color of ade2, originating from the vacuole.
Homo-FRET in Aggregates
The resonance energy transfer can occur also between the molecules of the same fluorophore given that there is overlap between emission and absorption of the fluorescent species. Unlike FRET between different molecules, energy transfer between like molecules (homo-FRET) is typically not manifested in changes of fluorescence lifetime or intensity.23 Recent works, however, have demonstrated that in some cases aggregation of fluorophores can still affect the rate of excited state levels depletion hence contributing to its fluorescence lifetime shortening.55, 56
Homo-FRET is usually observed with fluorescence anisotropy imaging microscopy.57, 58, 59 On excitation, a pool of molecules with the dipoles oriented collinearly to the excitation light’s electric vector will be preferentially excited, so that in the case of slow rotation, a relatively high anisotropy is expected. As the energy is transferred in a nonradiative manner from the initially excited molecules to other molecules, situated at different angles (except for ), the photons emitted will then be characterized with decreased anisotropy values.
Depicted on Fig. 5 is a comparison column chart showing the average intensity-weighted steady state anisotropy values measured from eGFP(wt), 97Q- , and 97Q-eGFP(wt). The value of obtained from the eGFP(wt) appeared to be not much lower than the maximal value of 0.57, available at two-photon excitation. This can be explained with a relatively long rotational correlation time for eGFP, so that it slightly affects its depolarization within the radiative decay time.60 A strong depolarization was observed for the 97Q-eGFP(wt): an anisotropy value as low as was recorded. As the volume and molecular mass of the aggregates are much higher than that for the free fluorophore, the rotational depolarization can be ruled out, leading us to the conclusion that homo-FRET is the main cause for the phenomenon observed. A depolarization of eGFP fluorescence as a result of homo-FRET has been also demonstrated in earlier works.61
Contrary to expectations, 97Q- exhibited a steady state anisotropy value of , which was lower than the value observed for the eGFP(wt). This can be partly explained with some probability of aggregation even in the absence of the hsp104. Earlier works have demonstrated that in 10 to 40% of the cells, the aggregates can still be found even in the absence of the chaperone.12 Indeed, in some measurements, a strong depolarization was observed, with the anisotropy values similar to that of the aggregates.
Cytotoxicity and Autofluorescence
The data on the cytotoxicity of the polyQ aggregates in yeast is controversial. In the early studies, none or little toxicity was reported even for 103Q tracts—at most, there was just a slight reduction in final culture density.8 Later studies, however, reported toxicity of expanded polyQ especially at the late stages after expression with the accumulation of the aggregated protein in the nucleus, leading to cell death with the markers, characteristic for apoptosis62, 63, 64 [mitochondrial fragmentation, reactive oxygen species (ROS) accumulation, and caspases activation). The latter, in its turn, may result in the increased level of autofluorescence intensity and perturbations in the intracellular environment, and hence affect the eGFP fluorescence lifetime.65, 66
To investigate the level of 97Q cytotoxicity in our experiment, cell proliferation of wt and yeast cells was assessed during after the induction of 97Q by measuring the culture’s optical density at . The results are presented in Table 1 and Fig. 6 . The values for all the samples were identical within the first of observation. At the 12th hour, wt cells expressing 97Q exhibited decreased values as compared to the control cells versus ). For the same time point, a decrease was also observed for the cells: expression of the 97Q resulted in the reduction of the values from to .
Results of the proliferation assay for wt and hsp 104Δ cells with and without 97Q expression.
The influence of the autofluorescence was checked on the wt and cells with and without expression of the 97Q protein. For all samples, we obtained the value of mean fluorescence lifetime with the intensity less than 10% of that recorded from the samples with eGFP. As shown in Fig. 7 , on the induction of 97Q the autofluorescence intensity increased. However, the increase was observed both for the wt and cells. Moreover, the autofluorescence, recorded from the cells expressing 97Q appeared somewhat higher than for the wt cells. Therefore, the autofluorescence perturbations could not account for the lifetime dynamics observed.
The experiments conducted revealed a decrease of the eGFP fluorescence lifetime in fusion with 97Q by as compared to the eGFP alone. Preventing aggregation of the polyQ extended tract through the elimination of Hsp104 chaperone led to the reverse of the 97Q-eGFP’s fluorescence lifetime with just slightly smaller values than that recorded from the free eGFP.
The possibility of the direct nonradiative transfer of energy from eGFP to the 4-aminoimidazole ribotide and red Ade2 dye was assessed by comparing the lifetime values of 97Q-eGFP(wt) (with the fluorescent metabolites accumulated) and 97Q-eGFP(Ade2) (with the restored purine biosynthesis and thus lack of potential FRET acceptors accumulation). Only a slight difference between both cell types was observed with obtained from 97Q-eGFP(wt) and obtained from 97Q-eGFP(wt Ade2). The measurement results of 97Q-eGFP taken from Ade2-deficient and proficient cells also appeared to be identical (Tables 2, 3 ). The data obtained proves lack of nonradiative energy transfer, meaning lack of contribution by the metabolites to the observed fluorescence lifetime decrease.
Summary of the lifetime results for the measurements taken from 97Q-eGFP in wt and hsp 104Δ cells, as well as in their Ade2-proficient analogs.
|wt||hsp 104Δ||wt Ade2||hsp 104Δ Ade2|
The autophagy of abnormal proteins to lysosomes, which are known to have acidic pH in their lumen, was shown to be not effective in our experiments, since no signal of the eGFP was detected from the vacuole. Hence, the influence of low pH on the fluorescence lifetime dynamics can also be ruled out.
Cell proliferation assay revealed slightly lower in the 12th hour on the 97Q induction in wt cells as compared to the control cells without polyQ protein. Despite the fact that some cell growth inhibition was observed, the degree of its manifestation was much lower than that reported in earlier works, which reported apoptosis induction by polyQ aggregation.64 Moreover, a reduction in the value was observed also for the cells expressing 97Q, indicating that the reduction of the cell density was caused by the expression of exogenous 97Q rather than the protein aggregation. The results obtained enable us to conclude that at this level of cytotoxicity no massive changes in the intracellular environment took place and thus it could not be the major contributor to the eGFP fluorescence lifetime changes recorded.
The reasons for the autofluorescence intensity increase on 97Q expression are unclear. However, since the increase was recorded from both wt and cells, it is clear that this also could not account for the observed fluorescence lifetime dynamics.
Reviewing all of the preceding, the decrease of the eGFP fluorescence lifetime can supposedly be attributed to the influence of the aggregation per se. Although the exact reason for the phenomenon observed requires further investigation, we suggest two possible hypotheses for the observed fluorescence lifetime dynamics: homo-FRET between eGFP molecules due to their tight packing in aggregates and changes in refractive index of the eGFP environment at aggregation.
The existence of homo-FRET was demonstrated by fluorescence anisotropy imaging. A strong depolarization in 97Q-eGFP(wt) is indicative of the eGFP molecules being in close proximity. Theoretically, homo-FRET does not affect the recorded fluorescence lifetime since the transfer time is negligible as compared to the time the molecules spend on the excited state levels, and the fluorescence properties of the acceptor are identical to those of the donor. However, in special cases a direct dependence of the CFP fluorescence lifetime on the presence of homo-FRET was demonstrated earlier even for dimers and was suggested to result from the chromophore structure.55 In the same work, however, the experiments failed to work the same dependence with eGFP. Moreover, other works, exploring homo-FRET of eGFP showed no changes in the fluorophore’s fluorescence lifetime values.61 Nevertheless, we assume that in the case of the aggregates, an increase in the number of molecules in a volume unit could lead to the continuous transfers of the energy quantum, and such an energy trap could lead to the decrease of the recorded fluorescence lifetime. A similar effect has been also reported earlier in Ref. 56.
The relation between refractive index and fluorescence lifetime was defined in 1962 by Strickler and Berg:67is the natural lifetime, is the fluorescence emission, is the extinction coefficient, is the emission spectrum in wavenumbers, and is the refractive index. Although the equation describes the dependence for the natural lifetime only, the applicability of the same model has been demonstrated in experiments. Thus, the dependence of the GFP fluorescence lifetime on the square of the refractive index in solution was demonstrated by Suhling 24 In their experiment, increasing a solution’s refractive index from 1.35 to 1.46 by changing the glycerol contents from 10 to 90%, resulted in a drop from for the GFP fluorescence lifetime. This dependence was further utilized to sense the spatial variation of the refractive index in live cells, which was demonstrated by other methods to vary from 1.35 in the cytoplasm to 1.46 to 1.60 for the plasma membrane.68, 69, 70, 71 However, in the majority of these works, FLIM was applied to reveal the spatial heterogeneities mainly within plasma membrane.72, 73, 74 A change of refractive index was also associated with the eGFP fluorescence lifetime decrease on fusion to other proteins;73 such dependence could be the reason for lack of full restoration of the free eGFP lifetime in 97Q- observed in our experiments.
Deviations from the quadratic dependence were also demonstrated in some cases.75 This, in particular, may be applicable for the molecules bound to optically thin layers characterized with the refractive index values different from the homogenous medium surrounding the fluorophore. In this case, a decrease of the fluorescence lifetime may result from the reflections of the emitted fluorescence at the interface and interference of the reflected field and the excited state dipole.73, 76 In this case, the radiative rate of a fluorophore depends on the orientation of the transition moment of the molecule with respect to the interface. Given that current data describes the aggregates as enriched in -sheet structures, this model may apply to the dynamics observed and explain the broadening of the histogram recorded from the 97Q-eGFP(wt).
The results obtained suggest that application of fluorescence spectroscopy techniques for the study of misfolded and, as a result, aggregated proteins require a careful approach and consideration of the changes in the used fluorophores photophysical properties, in particular, as demonstrated, the fluorescence lifetime. This is especially important for qualitative measurements, such as, for example, FRET, which is used for the calculation of intermolecular distance.
This work is generously supported by the National Science Council under Grants NSC 97-3112-B-010-006, NSC 97-2627-M-010-006, NSC 96-2112-M-010-001, NSC 94-2311-B-010-007, NSC95-2311-B-010-004, and NSC96-2311-B-010-001 and by the Ministry of Education under the “Aim for Top University” project.