Sci. Aging Knowl. Environ., 11 June 2003
Vol. 2003, Issue 23, p. pe14
[DOI: 10.1126/sageke.2003.23.pe14]


Monitoring Proteins in Intact Cells

Oksana Berezovska, Brian J. Bacskai, and Bradley T. Hyman

The authors are at the Alzheimer's Disease Research Laboratory, Massachusetts General Hospital, Harvard Medical School, Charlestown, MA 02129, USA. E-mail: oberezovska{at} (O.B.);2003/23/pe14

Key Words: neurodegenerative disease • protein-protein interaction • protein conformation • fluorescence resonance energy transfer • fluorescence lifetime imaging microscopy • presenilin • amyloid precursor protein


Over the past several years, research on various neurodegenerative disease processes has converged, leading to a new focus on protein misfolding as a common underlying thread that results in neuronal loss and neural system dysfunction. Autosomal dominant mutations that cause inherited forms of Alzheimer's disease, frontotemporal dementia, Parkinson's disease, and amyotrophic lateral sclerosis all lead directly or indirectly to increased aggregation and cellular deposition of certain proteins, followed ultimately by neuronal death. The proteins involved are diverse, and the specific neuronal populations affected vary, leading to distinct clinical and pathological phenotypes. For example, in the so-called "tauopathies" the microtubule-associated protein tau contributes to abnormal deposits of insoluble protein (inclusions) in neuronal and glial cells that are seen in frontotemporal dementia, as well as to the neurofibrillary tangles of Alzheimer's disease (see "Detangling Alzheimer's Disease"). The "synucleinopathies" include deposits of the presynaptic protein synuclein in Lewy bodies, the cytoplasmic inclusions found in Parkinson's disease and diffuse Lewy body disease, and in glial inclusions in multiple system atrophy. Aggregates of abnormal proteins also have been found in vulnerable neurons in certain diseases caused by the expansion of trinucleotide repeats, including Huntington's disease and inherited ataxias. In each of these instances, a unique and important feature is that the aggregates occur only in specific cellular populations, often in discrete regions of the brain and frequently within specific subcellular compartments. The same protein, expressed in other neurons and even in different subcellular compartments of a neuron that contains aggregates, does not precipitate. Understanding the underlying changes in protein conformation and the cellular and subcellular events that lead to protein deposition remains one of the great challenges in unmasking the pathophysiology of neurodegeneration.

A closely related issue follows from the fact that the "pathological" proteins do not act in isolation. Protein-protein interactions have emerged as a critical point of investigation in understanding cell signaling pathways, transcriptional regulation, and cell death cascades. Understanding the translocation of proteins, or of protein complexes, from one subcellular compartment to another is a frequent theme in these studies, and highlights the need for anatomically specific and discrete methodologies to examine protein-protein interactions.

Fluorescence Resonance Energy Transfer

The application of fluorescence resonance energy transfer (FRET) techniques to monitor the proximity of two proteins within a cell is a powerful approach to help answer questions about protein-protein interactions. The biophysical principle of FRET was established decades ago in pioneering work by Stryer (1), who applied FRET techniques to measure proximity or distance between fluorescent moieties in isolated biological systems. Current applications extend this approach to intact cells and even to histological preparations, with ever-improving resolution. The basis of FRET is the observation that, when two fluorophores that have overlapping excitation and emission spectra are in very close proximity (generally less than 100 Å), the higher-energy fluorophore (the donor) can release its energy by nonradiatively transferring energy to the longer-wavelength acceptor fluorophore (Fig. 1A). The acceptor fluorophore then emits photons at its longer wavelength. The degree to which this occurs is a function of how close the molecules are. If the fluorophores are more than about 100 Å apart, it occurs to a negligible extent; if they are very close together (the Förster radius, on the order of 50 Å), about half of the energy is transferred. These distances are specific for individual fluorophore pairs, and distance determination depends on a variety of assumptions about proximity, orientation, and appropriateness of a given donor/acceptor pair (2).

View larger version (40K):
[in this window]
[in a new window]
Fig. 1. Methods to measure FRET. (A) Spectral emission analysis measures the intensity of fluorescence emitted at 586 nm (Cy3) after excitation at 488 nm (FITC). (B) Photobleach dequenching analysis measures the change in fluorescence intensity emitted at 522 nm (FITC) after photobleaching of the acceptor (Cy3) by exposure to intense 568-nm light. (C) FLIM analysis involves measurement of the donor (FITC) fluorescence lifetime in the absence (no FRET) and in the close presence of the acceptor (Cy3). Fluorescence decay curves for non-FRETing and FRETing fluorophores are indicated on the bottom right.

The popularity of using FRET in cell-based and tissue-based assays has come with the development of newer techniques to measure its presence. This can be performed either by generating fusion proteins with endogenous fluorescent properties [such as cyan or yellow fluorescent protein (CFP or YFP)] or by using antibodies tagged with different colored fluorophores. The conceptually simplest method is to illuminate the sample with light specific for the donor fluorophore and measure any emissions from the acceptor (Fig. 1A). Several software packages are available to do this and require only slight modifications of a standard fluorescent or confocal microscope. However, "bleedthrough" emissions from the donor into the acceptor emission channel and misexcitation of the acceptor leading to direct fluorescence both increase background noise (3). If the FRET signal is not extremely robust, this noise can interfere substantially. Although computational methods are available to correct for the noise, the corrections frequently involve making additional measurements and generally diminish the sensitivity of the method.

Another commonly used technique to measure the presence of FRET is referred to as donor dequenching analysis (Fig. 1B). The reasoning is as follows: Because the donor fluorophore, in the presence of an acceptor, transfers some of the energy that would ordinarily result in emissions, the donor fluorophore is quenched by the presence of the acceptor. If the acceptor is either destroyed (for example, by photobleaching) or is moved away from close proximity with the donor (for example, in protease activity assays, where cleavage separates the donor and acceptor), the fluorescence of the donor increases. Thus, by taking "before and after" measurements, one observes an enhancement in the amount of fluorescence as the donor is "dequenched." The extent of this enhancement reflects the presence of FRET (4, 5). This technology can also be applied with relatively minor modifications to standard fluorescent or confocal microscopes. The major drawback occurs under circumstances in which either FRET is weak (leading to a small percentage increase in donor fluorescence) or some donor molecules are interacting with an acceptor, but others are not. If only a small percentage of molecules are interacting, then only those molecules are dequenched, and the resulting increase in fluorescence is "washed out" because of the high baseline of noninteracting donor molecules. Moreover, using the acceptor photobleach technique inevitably involves exposing the sample to relatively high-energy light, and care must be taken in some experiments to avoid introducing autofluorescence and other background artifacts. Nonetheless, in many experimental paradigms, the approach provides quantitative information that can even allow one to differentiate between the proximity of different domains of two interacting molecules. For example, using these techniques, we have been able to demonstrate direct interactions between the C-terminal domain of the amyloid precursor protein (APP) and the extracellular domains of the low-density lipoprotein receptor-related protein (LRP). In addition, we have detected heterotrimeric interactions of the adapter protein FE-65 with both APP and LRP (6).

Fluorescence Lifetime Imaging Microscopy

A newer and more complex technology, termed fluorescence lifetime imaging microscopy (FLIM), has recently become available to detect and quantitate the presence of FRET between two fluorophores (Fig. 1C). This approach takes advantage of observations regarding fluorescence lifetimes during FRET events. After excitation by a brief pulse of light, fluorophores have a characteristic delay before emitting a photon. A population of fluorophores results in a distribution of delay times that can be modeled as an exponential decay (7). Measuring these decay times, which are typically in the picosecond range, is technologically challenging but can be accomplished with new generations of detectors (8). An exponential fit of the observed decays permits calculation of the characteristic lifetime, {tau}. This measurement is not dependent on the concentration of the fluorophore but is sensitive to its microenvironment (2). In the presence of an acceptor fluorophore, {tau} is shortened. The amount of shortening is dependent on FRET and reflects the proximity between the donor and acceptor. Software packages have been developed that can deconvolve (or separate) decay curves into more than one exponential fit, providing distinct measurements of {tau} for both the fluorophore molecules that are not in the vicinity of an acceptor and the fluorophore molecules that are in the close proximity of an acceptor. Because the measurement of lifetimes is not concentration-dependent (2), the technique is ultrasensitive for monitoring interacting fluorophores, even if they represent only a small percentage of the total fluorescence. Moreover, because the amount of lifetime shortening is readily measured with the new sophisticated equipment, the technique provides a quantitative readout of the amount of FRET.

Most exciting from the perspective of cell biology has been the development of hardware and software packages that allow one to perform these measurements on a pixel-by-pixel basis in microscopic fields. The lifetimes can be converted into pseudocolored images providing true, exquisite subcellular resolution of different fluorescent lifetimes, which reflect protein-protein interactions. The technique is nondestructive and does not involve manipulating the acceptor fluorophore. Moreover, because certain living color donor/acceptor pairs such as CFP/YFP can be used, the potential exists to perform protein-protein interaction assays with extraordinary resolution in living systems (9, 10).

Because of the lack of concentration dependence and the enhancements in the ability to detect precise subcellular localizations, the FLIM technique has proven to be advantageous in many circumstances. Its major disadvantage lies in the relatively extensive hardware and software requirements. We use a femtosecond pulsed Ti sapphire laser (tunable from 700 to 1000 nm) to generate pulses on the order of every 12 ns (Spectra-Physics). Specific fluorophores are individually excited by means of two-photon excitation, and lifetimes of fluorophores on the order of 1000 to 3000 ps can be readily observed with Becker and Hickl FLIM hardware and software packages and a Hamamatsu microchannel plate detector. In double immunofluorescence experiments, in which one type of antibody is labeled with the fluorescent tag fluorescein isothiocyanate (FITC, the donor fluorophore) and the other with Cy3 (the acceptor fluorophore), the lifetime of the fluorescence derived from the FITC-labeled antibody is shortened from 2500 ps to about 1500 ps when in close proximity to Cy3.

Applications of FLIM Technology

As an example of this procedure, we used antibodies that recognize the C terminus of APP and the large hydrophylic loop of presenilin-1 (PS-1, also known as PSEN1) to evaluate possible sites of interaction between these molecules in neurons. PS-1 is believed to be a major component of the gamma secretase complex, which cleaves APP to generate amyloid {beta} peptide (A{beta}), a component of cerebral plaques in Alzheimer's disease. This cleavage event then initiates release of the APP intracellular signaling domain from the membrane-bound precursor protein (see Wolfe Perspective). Fluorescently labeled secondary antibodies linked to either FITC or Cy3 were used to detect primary antibodies that recognize the proteins of interest. The intensity image in Fig. 2A shows PS-1 immunoreactivity (visualized with FITC) in primary neurons derived from mouse embryos. The FLIM image in Fig. 2B is pseudocolored, reflecting FITC lifetime measurement as a color on a pixel-by-pixel basis. These lifetimes were fit with two exponential fluorescence decay curves, and average lifetimes are reported. These lifetimes are then mapped by pseudocolor over the entire image. The color-coded FLIM image shows the donor (FITC) fluorescent lifetime in each pixel. The range of the scale has been chosen to indicate lifetimes ranging from 2000 ps (red) to 2500 ps (blue). The closer the two fluorophores are, the shorter the fluorescent lifetime is. In singly labeled specimens, in the absence of the Cy3 acceptor, all the lifetimes appear blue.

View larger version (48K):
[in this window]
[in a new window]
Fig. 2. FLIM analysis of the proximity between APP and PS-1 in primary neurons. (A) Intensity image shows immunostaining pattern for PS-1. (B) Color-coded FLIM image shows the distribution of the FITC (PS-1) lifetimes within the cell, reflecting the proximity between the large hydrophilic loop of PS-1 and the APP C-terminal fragment, CTF. The closest proximity (shortest lifetime, red) between PS-1 and APP is observed near the cell surface. The colorimetric scale shows fluorescence lifetime from 2000 to 3000 ps.

From intensity measurements of antibody staining derived from images like the one shown in Fig. 2A, it is clear that APP and PS-1 colocalize throughout the cell, with a predominant colocalization in the perinuclear area [representing the Golgi apparatus and endoplasmic reticulum (ER)]. However, the proximity image shown by the FLIM technique (Fig. 2B) reveals that the closest proximity of these two molecules does not occur in the perinuclear region, but near the cell surface (red color in the FLIM image). Therefore, mapping of donor fluorescence lifetime intensity throughout the entire image allows us to identify both subcellular distribution and proximity between two molecules, even in a subcellular region where the molecules are sparse (11).

In contrast to simple double immunofluorescence experiments, therefore, the proximity assay provides a great deal of additional information. Compared to traditional protein-protein interaction assays such as coimmunoprecipitation, FLIM has several advantages. For example, it can be performed using intact cells, and information can be obtained from single cells. In addition, coimmunoprecipitation experiments frequently depend on the exact extraction conditions used to fractionate the cells, which might disrupt the protein-protein interactions being studied.

The most striking advantage of FLIM, however, is the straightforward way in which subcellular distribution can be analyzed. The result shown in Fig. 2, which reveals that APP and PS-1 are in closest proximity near the cell surface although they display a high degree of colocalization in subcellular compartments such as the Golgi apparatus and ER, helps address the so-called spatial paradox presented by Annaert and colleagues (12, 13). These investigators argued that, because APP and PS-1 are primarily colocalized in the Golgi, but A{beta} is generated to a great extent in later compartments in the secretory pathway and even in early endosomes, PS-1 might not be an important member of the complex that generates A{beta}. The current observations, based on FLIM technology, suggest instead that a relatively small population of PS-1 and APP interacts closely at or near the cell surface. Presumably it is this interaction that contributes primarily to A{beta} generation.

Thus, FRET techniques measure proximity phenomena in cells and in tissue sections, and this proximity assay can be interpreted in the context of protein-protein interactions. Although there are several ways to measure FRET, the novel FLIM technology has the following advantages: (i) it is not concentration-dependent; (ii) bleedthrough, misexcitation, and autofluorescence do not complicate analysis; (iii) it is nondestructive and therefore can be used in living cells or whole organisms; (iv) both FRETing and nonFRETing populations of fluorophores can be detected simultaneously, so that even if the interacting proteins are a relatively small percentage of the total population, the interactions can still be detected; (v) the percentage of FRETing and non-FRETing populations can be calculated under some circumstances; (vi) the degree of shortening of the lifetime is inherently a quantitative measure of proximity, and changes in this quantity by FRET reflect alterations in conformation or in the distance between the two molecules; and (vii) this information can be displayed with very high spatial resolution.

FLIM is a relatively new approach that still requires a fair amount of optimization for each application. However, this novel technique provides a great deal of information about protein-protein interactions within an intact cell, and potentially even on conformational changes in single molecules after experimental manipulations or under pathological conditions. For example, we have shown using the donor dequenching FRET technique that the conformation of synuclein is altered within Lewy bodies in the brains of patients with Parkinson's disease (14), and our recent experiments suggest that FLIM confirms and extends these types of observations. Similar studies on the conformation of A{beta} in and near different morphological types of senile plaques (15) and on tau conformation in neurons undergoing neurodegeneration in Alzheimer's disease also appear to have promise. Thus, techniques that measure fluorescence energy transfer in general, and FLIM applications in particular, may well hold promise for understanding protein conformational changes and protein-protein interactions in a wide variety of neurodegenerative processes.

June 11, 2003
  1. L. Stryer, Fluorescence energy transfer as a spectroscopic ruler. Annu. Rev. Biochem. 47, 819-846 (1978).[CrossRef][Medline]
  2. J. R. Lakowicz, Principles of Fluorescence Spectroscopy (Plenum, New York, 1999).
  3. G. W. Gordon, G. Berry, X. H. Liang, B. Levine, B. Herman, Quantitative fluorescence resonance energy transfer measurements using fluorescence microscopy. Biophys. J. 74, 2702-2713 (1998).[Medline]
  4. P. J. McLean, H. Kawamata, S. Ribich, B. T. Hyman, Membrane association and protein conformation of alpha-synuclein in intact neurons. Effect of Parkinson's disease-linked mutations. J. Biol. Chem. 275, 8812-8816 (2000).[Abstract/Free Full Text]
  5. F. S. Wouters, P. I. Bastiaens, Fluorescence lifetime imaging of receptor tyrosine kinase activity in cells. Curr. Biol. 9, 1127-1130 (1999).[CrossRef][Medline]
  6. A. Kinoshita, C. M. Whelan, C. J. Smith, I. Mikhailenko, G. W. Rebeck, D. K. Strickland, B. T. Hyman, Demonstration by fluorescence resonance energy transfer of two sites of interaction between the low-density lipoprotein receptor-related protein and the amyloid precursor protein: role of the intracellular adapter protein Fe65. J. Neurosci. 21, 8354-8361 (2001).[Abstract/Free Full Text]
  7. J. R. Lakowicz, H. Szmacinski, K. Nowaczyk, K. W. Berndt, M. Johnson, Fluorescence lifetime imaging. Anal. Biochem. 202, 316-330 (1992).[CrossRef][Medline]
  8. Q. S. Hanley, V. Subramaniam, D. J. Arndt-Jovin, T. M. Jovin, Fluorescence lifetime imaging: multi-point calibration, minimum resolvable differences, and artifact suppression. Cytometry 43, 248-260 (2001).[CrossRef][Medline]
  9. A. G. Harpur, F. S. Wouters, P. I. Bastiaens, Imaging FRET between spectrally similar GFP molecules in single cells. Nat. Biotechnol. 19, 167-169 (2001).[CrossRef][Medline]
  10. A. Periasamy, M. Elangovan, E. Elliott, D. L. Brautigan, Fluorescence lifetime imaging (FLIM) of green fluorescent fusion proteins in living cells. Methods Mol. Biol. 183, 89-100 (2002).[Medline]
  11. O. Berezovska, P. Ramdya, J. Skoch, M. Wolfe, B. Bacskai, B. Hyman, APP associates with a Nicastrin dependent docking site on the PS1/{gamma}-secretase complex in cells demonstrated by Fluorescence Lifetime Imaging. J. Neurosci., in press.
  12. W. Annaert, B. De Strooper, Presenilins: molecular switches between proteolysis and signal transduction. Trends Neurosci. 22, 439-443 (1999).[CrossRef][Medline]
  13. P. Cupers, M. Bentahir, K. Craessaerts, I. Orlans, H. Vanderstichele, P. Saftig, B. De Strooper, W. Annaert, The discrepancy between presenilin subcellular localization and gamma-secretase processing of amyloid precursor protein. J. Cell Biol. 154, 731-740 (2001).[Abstract/Free Full Text]
  14. N. Sharma, P. J. McLean, H. Kawamata, M. C. Irizarry, B. T. Hyman, Alpha-synuclein has an altered conformation and shows a tight intermolecular interaction with ubiquitin in Lewy bodies. Acta Neuropathol. 102, 329-334 (2001).[Medline]
  15. B. J. Bacskai, J. Skoch, G. A. Hickey, R. Allen, R., B. T. Hyman, FRET determinations using multiphoton fluorescence lifetime imaging microscopy (FLIM) to characterize amyloid-beta plaques. J. Biomed. Opt., in press.
  16. Supported by NIH grants AG15379, AG08487, and EB00768 and by a Pioneer Award from the Alzheimer Association.
Citation: O. Berezovska, B. J. Bacskai, B. T. Hyman, Monitoring Proteins in Intact Cells. Sci. SAGE KE 2003, pe14 (11 June 2003);2003/23/pe14

Science of Aging Knowledge Environment. ISSN 1539-6150