New Approaches To Intracellular Analysis

9.5.1. Expressed Probes

The Tsien group has developed several exciting new approaches to in situ analysis. Miyawaki et al. (1997) introduced the use of Green Fluorescent Protein variants fused to calmodulin, such that the dramatic conformational change induced in the dumbbell-shaped protein by binding calcium was transduced as a change in energy transfer efficiency (Figure 9.1). This approach combines the advantages of energy transfer assays with an expressible indicator: To measure the analyte, all one need do is express the gene within the relevant cell type (or even organism) to quantitate the analyte. Others have attempted to adapt this approach to measure other analytes such as zinc, with a range of success from modest (Barondeau et al., 2002; Pearce et al., 2000) to satisfactory (Bozym et al., 2004)(see below). Calmodulin is unusual, however, in that it undergoes a really substantial change in conformation upon calcium binding and thus donor-acceptor distance, which results in a significant change in energy

Figure 9.1. Expressible fluorescence energy transfer-based calcium sensor of Miyawaki et al. (1997). In the absence of Ca(II), the calmodulin (upper panel) is in an extended conformation such that the BFP on the left does not transfer energy, and one observes its emission. In the presence of calcium (lower panel), calmodulin adopts a more compact conformation, bringing the BFP in close proximity to the GFP, resulting in efficient energy transfer to the GFP, and reduced emission from BFP and enhanced emission from GFP. [Reproduced with permission from Miyawaki et al. (1997).]

Figure 9.1. Expressible fluorescence energy transfer-based calcium sensor of Miyawaki et al. (1997). In the absence of Ca(II), the calmodulin (upper panel) is in an extended conformation such that the BFP on the left does not transfer energy, and one observes its emission. In the presence of calcium (lower panel), calmodulin adopts a more compact conformation, bringing the BFP in close proximity to the GFP, resulting in efficient energy transfer to the GFP, and reduced emission from BFP and enhanced emission from GFP. [Reproduced with permission from Miyawaki et al. (1997).]

transfer efficiency. The R6 dependence of the efficiency on distance (Forster, 1948) means, however, that the system must be poised in terms of a carefully matched overlap integral and donor-acceptor distance [or distance distribution (Eis and Lakowicz, 1993)] to provide a significant change in transfer efficiency, and therefore signal. For instance, if the binding of the analyte changes the donor-acceptor distance from 20 to 30 A but the donor and acceptor are chosen so that "Forster distance" where energy transfer is 50% efficient is 40 A, analyte binding only results in a 14% change in transfer efficiency for a 50% increase in distance. With fluorescent donors and acceptors attached to proteins, the use of site-directed mutagenesis to position these moieties unambiguously and optimally is preferred (Thompson et al., 1996a). The change in energy transfer efficiency can be measured accurately as a change in intensity ratio, donor lifetime, or anisotropy (polarization) (Thompson et al., 1998). The flexibility of the energy transfer approach (i.e., one can induce a change in efficiency in principle by changing distance, spectra, quantum efficiency, refractive index, lifetime, and/or relative orientation) suggests that it will be even more widely used in the future.

9.5.2. Single-Molecule Detection in Cells

The development of single-molecule fluorescence detection techniques over the last several years (Moerner and Kador, 1989; Orrit and Bernard, 1990) (see Chapters 1, 2, and 4 in this book) has been very exciting in part because it potentially offers a way to look at molecules that may be very rare in the cell, and largely without having to average over an ensemble of molecules. The essence of the approach is to excite fluorescence within a relatively small (usually about one femtoliter) volume element of a medium containing fluorescent molecules at some low concentration (nanomolar), such that fluorescence above background levels will be readily collected and observed if a fluorophore happens to be within the volume element. Multiphoton excitation with a high numerical aperture objective is nearly ideal for this purpose. If the fluorescence collection is efficient, the fluorophore quantum yield is high, and the background is low, emission from individual molecules can be observed with statistically defined levels of confidence. It should be noted that this doesn't necessarily translate into determination of analytes at really low concentrations, since to ensure that a fluorophore will be observed in the tiny voxel within a reasonable time frame, one needs a nanomolar concentration of fluorophores. Nevertheless, in cells where the molecules of interest are likely to be localized to some structural feature such as the nucleus (e.g., you're looking a small fraction of the cell volume), this becomes a potentially fruitful endeavour.

Work has been ongoing at the single-molecule level in cells using the molecular beacon technology (Yao et al., 2003), but little has appeared using single indicator molecules to analyze metal ions. Several of the fluorescent probes for certain metal ions are good enough fluorophores that this should be possible in principle. At the single-molecule level, however, one is perforce no longer observing ensemble averages, but rather has entered the regime of stochastic sensing (Braha et al., 2000), where one measures concentration not by the fractional occupancy of an ensemble of sites, but by the time-averaged occupation of a single site. For low concentrations providing low fractional occupancy, this implies observing for a period of time, which at some point collides with the propensity of the fluorophore to photobleach. Use of fluorophores with much lower photobleaching propensity (like the semiconductor nanoparticles) would appear essential to such studies. The virtue of the patch-clamp-like techniques used in the existing stochastic sensors is that interrogation of the transducer is nondestructive and can be carried out for hours.

As mentioned above, an important issue in single-molecule work (indeed, in all fluorescence microscopy) is photobleaching, which is the often irreversible loss of fluorescence emission upon continuing excitation. The (photo)chemistry of the process differs for various fluorophores, but is often oxygen-dependent (Turro, 1978). It is important to remember that a fluorophore in the excited state has substantial additional energy and is thus more chemically reactive than in the ground state. Indeed, the fluorophore in the excited state often behaves entirely differently (from a chemical standpoint) from that in the ground state. For instance, it is well-known that some fluorophores that are weak bases in the ground state become strong acids in the excited state, and it is even more common that a molecule's redox properties change dramatically upon excitation. Reagents exist that counteract the effects of photobleaching, but these often are usable only in fixed and permeabilized cells. A rule of thumb in this regard is that a very good fluorophore (i.e., Rhodamine 6G) on the average will survive for a few thousand cycles of excitation and emission in solution before undergoing a destructive reaction. Since it is easy (by increasing the laser power and focusing in a small volume) to provide very intense excitation, the collection efficiency must be optimized since there is a limited opportunity to observe the fluorophore before it bleaches. In microscopes with high numerical aperture objectives, collection of fluorescence can be almost 50% efficient (Axelrod et al., 1984), whereas in ordinary fluorometers the efficiency is probably less than 10%. One answer to this is the use of metal-enhanced fluorescence, which under certain conditions is not emitted isotropically, but is emitted in a particular direction without the use of focusing optics (Lakowicz, 2004).

Finally, when studying trace analytes at the single-molecule level, one is concerned not only about the destructive effects of merely observing the fluorophore for prolonged periods, but also about the amount of time it will take the analyte to find the indicator molecule. For a small molecule present at nanomolar concentration with a diffusion-controlled association rate constant, this should take on the order of seconds in water, but with the slower (and anisotropic) diffusion within the cell and at lower analyte concentrations the binding process will be more prolonged. Within the cell, one imagines that small-molecule effectors seldom have to diffuse very far, and most likely only need to diffuse within a compartment to find their cognate binding site; effectors like steroids which must diffuse from the membrane into the nucleus perforce are not required to act quickly. Thus one imagines that it will be useful to localize the indicator within the organelle(s) of interest using chemical or molecular biological methods.

9.5.3. PEBBLES for Intracellular Measurements

An important development has been contributed by the Kopelman laboratory that promises to be extremely useful in understanding the biology of individual cells. Instead of dispersing a fluorescent indicator for a particular analyte throughout the cell or organelle of interest, the indicator is incorporated into very small (20-300 nm) beads and injected or otherwise introduced into cells. These beads are formed by emulsion polymerization and are named PEBBLES, an acronym short for "Probes Encapsulated by Biologically Localized Embedding" (Monson et al., 2003).

The incorporation of the fluorescent indicator into the PEBBLES offers a number of advantages. First, if the indicator itself is toxic or deleterious, the amount introduced into the cell is much smaller than if the indicator were present throughout the cytoplasm. A subtler, related advantage is that (for equilibrium binding indicators) the limited indicator amount means that the concentration of the analyte is not perturbed by excessive concentrations of indicator, as discussed by Dinely et al. (2002). The limited volume subtended by

Figure 9.2. Scanning electron micrograph of sol-gel PEBBLES; the scale indicates that the average size is 160 nm. [Reproduced with permission from Monson et al. (2003).]

the PEBBLE (see Figure 9.2) means that background fluorescence due to autofluorescence from cell constituents can be accurately subtracted from the emission of the PEBBLE; for indicators dispersed throughout the cell, this is much harder to do accurately. Moreover, if different PEBBLES are present having indicators whose spectral properties overlap to some degree, the crosstalk interference between PEBBLES does not require correction since the emissions are spatially resolved as well; again, this is very difficult to do if the compounds are dispersed throughout the cell and not spatially resolved. Typically, the fluorescent indicator is sequestered within the PEBBLE matrix, such that the indicator is protected from adsorption to, for instance, protein(s) in the intracellular milieu which might change its response or degrade it (Burdette et al., 2001). Also, the sequestration limits any toxicity the indicator system might have. Finally, it is straightforward to include a second fluorescent dye for intensity ratiometric measurements, an approach whose advantages are widely appreciated.

PEBBLES have been made three different ways to date. The first and most widely used approach is to form ~40-nm polyacrylamide beads by polymerization in a microemulsion formulation. The polyacrylamide gel is similar to those used for analytical electrophoretic separations, being quite hydrophilic. The detergents used to form the microemulsion can be deleterious to protein-based indicators, however. PEBBLES can also be made from so-called sol-gel formulations, which comprise a rigid, porous, transparent, glass-like material containing the indicator and polyethylene glycol. The rugged nature of the sol-gel compared to the softer polyacrylamide suggests its use in more demanding environments. While the conditions needed to form sol-gels are comparatively mild compared to those required for regular glass, the conditions may be deleterious to protein-based sensors. Finally, for a more hydrophobic PEBBLE matrix suited to hydrophobic indicators, the use of decyl methacrylate has been developed.

PEBBLES may be introduced into cells by one of four routes: They may be introduced by liposome fusion or direct injection, propelled ballistically into the cell like a bullet using a device called a "gene gun," or taken up by endocytosis. The methods differ in their degree of efficacy with a given cell line and where within the cell the PEBBLE ends up. PEBBLES can be directly injected into the cell using a micropipette. While this procedure is used routinely for in vitro fertilization, there is reason to suspect that such injections may indeed disrupt the cell. PEBBLES may be phagocytosed like bacteria or other small particles; this approach is especially valuable to study the phagosomes or lysosomes, but by itself it may not be very useful for studying other organelles such as the mitochondria or nucleus. PEBBLES can be introduced into the cell by fusion of liposomes that contain them with the cell membrane. This technique is very benign but has little target specificity. Most ingenious of all is the use of the Gene Gun, a device that transfects cells with DNA by "spitting" (using a gas jet) gold nanoparticles coated with DNA into the cell. To a degree, the penetration of the gold nanoparticle into particular compartments in the cell can be influenced by the gun parameters, but the accuracy with which the nanoparticles can be placed is very modest. Of course, a virtue of the multiple means of introduction is that if one is unsatisfactory, one of the others might work.

Several analytes have been measured using PEBBLES, including oxygen, pH, Zn, Ca, and potassium. In principle, it would appear possible that anything measured by a dispersed indicator could also be measured with PEBBLEs. One of the advantages of PEBBLES is that if the fluorescent indicator exhibits just a simple intensity change upon binding (or colliding with) the analyte, it can be readily converted into a ratiometric indicator by inclusion of a spectrally distinct fluorophore that is unresponsive together with the indicator fluorophore: Co-location in the PEBBLE maintains the concentration relationship of the two fluorophores, as well as their spatial relationship. Figure 9.3 depicts PEBBLES in rat C6

Figure 9.3. Fluorescence micrograph Rat C6 glioma cells with ballistically inserted PEBBLES. [Reproduced with permission from Monson et al. (2003).] See insert for color representation of this figure.

glioma cells containing a ratiometric oxygen indicator consisting of a mixture of ruthenium tris (dpp) together with Oregon Green; the oxygen quenches the metal complex efficiently, whereas the Oregon Green is quite insensitive. Thus the emission intensity of the complex declines with increasing oxygen concentration, whereas that of the organic fluorophore remains constant, and thus the observed ratio of the emissions is a simple function of the oxygen concentration (Monson et al., 2003).

One of the issues in the development of PEBBLE-based sensors is the means of their formation and its compatibility with biomolecules such as proteins. In particular, while the conditions are gentle, the emulsion polymerization can be deleterious to proteins since it contains substantial concentrations of surfactants, which are well known to denature proteins. Other components of this or the sol-gel or decyl methacrylate formulations may also be deleterious to certain indicator molecules. There seems to be no general solution, with individual sensor components requiring validation with a particular PEBBLE formulation on a case-by-case basis.

9.5.4. Metal Ion Biosensors

As mentioned above, while scores of low-molecular-weight fluorescent indicators have been described for a myriad of metal ions, in many cases this approach has appeared inadequate. For metal ions like Mg, Ca, K, and Na that are relatively abundant in the free form in cells, it has proven feasible to synthesize fluorescent indicator molecules (Haugland, 2005). For metal ions present at trace and ultratrace levels in the free form, such as Zn, Cu, Fe, and others, synthesizing indicators has proven more difficult (Thompson, 2005). While chelators are known that bind these metal ions very tightly, the need for selectivity in the complex matrices of the cytoplasm and extracellular medium is paramount: For a sensor, it is not enough to bind (and therefore respond to) free zinc at picomolar levels. Rather, it must bind zinc specifically in the presence of million-fold higher concentrations of calcium and magnesium. Classical chemical separations to remove interferents are of course problematic within an organism: The separation must be done in situ "on the fly" as it were. Some time ago we (Thompson and Jones, 1993) and others (Godwin and Berg, 1996; Regan and Clarke, 1990; Walkup and Imperiali, 1997) sought to adapt the extreme selectivity of biological molecules to this problem. In our case we exploited a human enzyme, carbonic anhydrase II, as a very selective ligand for zinc and other metal ions (Fierke and Thompson, 2001). We demonstrated that the protein would selectively bind these target metals at nanomolar to picomolar levels in the presence of 10 mM Ca and 50 mM Mg (the concentrations found in sea water). Looking at the structure of the active site (Christianson and Fierke, 1996) (Figure 9.4), it is easy to see that it would bind zinc with high affinity due to the position and orientation of the three histidine imidazole ligands, but would bind Ca and Mg with poor affinity, if at all. We believe that the high selectivity compared with classic chelators such as EDTA can be rationalized by considering that the ligands in the protein are held in position by hydrogen bonds to other amino acid residues, which must be displaced for other metals to bind and which therefore is energetically unfavorable (Kiefer et al., 1995b). By comparison, the chelator is much more flexible and can accommodate a greater variety of metal ions, which is undesirable in this case. We note that cyclophanes are considerably more rigid and have been adapted to some metallofluorescent indicators [notably for zinc (Kikuchi et al., 2004)], but these indicators exhibited very slow kinetics.

A unique advantage of the biosensor approach is the ease with which the protein [or other macromolecules such as RNA aptamers (Yang and Ellington, 2005)] can be modified

Figure 9.4. Active site of human carbonic anhydrase II indicating residues that are zinc ligands, as well as "second shell" residues which hold the ligand histidinyl imidazoles in position by hydrogen bonding. [Reproduced with permission of the copyright holder from Fierke and Thompson (2001).] See insert for color representation of this figure.

Figure 9.4. Active site of human carbonic anhydrase II indicating residues that are zinc ligands, as well as "second shell" residues which hold the ligand histidinyl imidazoles in position by hydrogen bonding. [Reproduced with permission of the copyright holder from Fierke and Thompson (2001).] See insert for color representation of this figure.

to improve its properties for a particular application. For instance, our collaborators, led by Professor Carol Fierke of the University of Michigan, have shown that the affinity, selectivity, and even kinetics of metal ion binding may be improved by mutagenesis of the protein, by either directed or combinatorial mutagenesis (Kiefer et al., 1995a; Huang et al., 1996; Hunt et al., 1999). Moreover, the use of site-directed mutagenesis has greatly simplified the labeling process, enabling us to optimize the response of fluorescent labels by positioning them with respect to the metal ion binding site (Thompson et al., 1996a). Finally, the use of macromolecules as transducers (as discussed above with the work of Miyawaki et al., 1997) enables the use of Forster transfer as a sensing approach, which has several advantages (Godwin and Berg, 1996; Thompson et al., 2002a; Thompson and Patchan, 1995).

We (and others) have found several means to transduce the metal recognition event as a change in fluorescence we can measure. Transduction as a simple intensity change is so prone to artifact and so hard to calibrate accurately (especially in the microscope) that it has been largely superseded by intensity ratios (Grynkiewicz et al., 1985), fluorescence lifetimes (Lippitsch et al., 1988; Szmacinski and Lakowicz, 1993) and fluorescence polarization (anisotropy) (Dandliker et al., 1973; Elbaum et al., 1996). Thus we use zinc-dependent binding of a fluorescent ligand to carbonic anhydrase and use the change in fluorescence intensities at differing wavelengths (Figure 9.5) (Thompson et al., 2000), fluorescence polarization (Elbaum et al., 1996), or fluorescence lifetime (Thompson and Patchan, 1995). For other metal ions that are colored when bound [such as Cu (II)], we can use quenching of nearby fluorescent labels due to Forster transfer if the label's emission overlaps the weak d-d absorbance band, resulting in reduced fluorescence lifetimes (Thompson et al., 1999a) (Figure 9.6). For uncolored ions, other proximity-dependent quenching mechanisms can be used; under certain conditions these mechanisms result in changes in fluorescence

Figure 9.5. Principle for fluorescence detection of zinc by apocarbonic anhydrase and ABDN (left panel) and dependence of fluorescence emission intensity ratios on free zinc concentration (right panel). Binding of ABDN to carbonic anhydrase only occurs when zinc is bound to the active site, and results in a blue shift and increase in quantum yield; the proportion of blue-shifted emission reflects the fraction of protein with zinc bound, and thus the concentration. [Reproduced from Thompson et al. (2000), with permission.]

Figure 9.5. Principle for fluorescence detection of zinc by apocarbonic anhydrase and ABDN (left panel) and dependence of fluorescence emission intensity ratios on free zinc concentration (right panel). Binding of ABDN to carbonic anhydrase only occurs when zinc is bound to the active site, and results in a blue shift and increase in quantum yield; the proportion of blue-shifted emission reflects the fraction of protein with zinc bound, and thus the concentration. [Reproduced from Thompson et al. (2000), with permission.]

Figure 9.6. Schematic of fluorescence sensing of Cu(II) by carbonic anhydrase II. Energy transfer from the fluorescent label to protein-bound Cu(II) results in quenching and a decrease in lifetime; the fraction of the shorter lifetime corresponding to the Cu-bound form is proportional to the free Cu concentration. Inset: Spectral overlap of label fluorescence emission and bound Cu absorption. [Reproduced from Thompson et al. (1999), with permission of the copyright holder.]

Figure 9.6. Schematic of fluorescence sensing of Cu(II) by carbonic anhydrase II. Energy transfer from the fluorescent label to protein-bound Cu(II) results in quenching and a decrease in lifetime; the fraction of the shorter lifetime corresponding to the Cu-bound form is proportional to the free Cu concentration. Inset: Spectral overlap of label fluorescence emission and bound Cu absorption. [Reproduced from Thompson et al. (1999), with permission of the copyright holder.]

Figure 9.7. Fluorescence anisotropies of apoN67C-ABD-T as a function of the concentrations of five metal ions. [Reproduced with permission from Thompson, et al. (1999b).]

polarization which are more convenient to measure; an example of these data is shown in Figure 9.7.

Beginning with the work of Peterson, Hirschfeld, and Seitz, there has been interest in adapting fluorescence assays to fiber optics for purposes of determining analytes in remote or inaccessible places (Thompson, 1991; Thompson, 2005; Wolfbeis, 1991). Several workers have demonstrated sensors based on changes in intensity, intensity ratio (Goyet et al., 1992), and fluorescence lifetime (Bright, 1988). We also have adapted our metal ion biosensors to use with fiber optics (Thompson et al., 1996b; Thompson and Jones, 1993; Zeng et al., 2003,2005), being particularly interested in measurements in the depths of the ocean (where sampling for analysis is slow, expensive, and prone to contamination) and in vivo in larger animals, where the skin is not transparent and the area of analytical interest may be deep within an organ or body cavity. Thus we have made measurements of Cu(II) at picomolar levels in situ in the ocean, in real time (Zeng et al., 2003), Similarly, we have measured free zinc in the brain of a large animal ischemia model in situ in real time. Recently, we replaced the off-axis parabolic mirrors in our fiber-optic apparatus (Thompson et al., 1990) with dichroic mirrors, which greatly simplified alignment of the system (Figure 9.8).

Pregnancy Nutrition

Pregnancy Nutrition

Are You Expecting? Find Out Everything You Need to Know About Pregnancy and Nutrition Without Having to Buy a Dictionary. This book is among the first books to be written with the expertise of a medical expert and from the viewpoint of the average, everyday, ordinary,

Get My Free Ebook


Post a comment