|Home | About | Journals | Submit | Contact Us | Français|
Single-molecule detection (SMD) with fluorescence is a widely used microscopic technique for biomolecule structure and function characterization. The modern light microscope with high numerical aperture objective and sensitive CCD camera can image the brightly emitting organic and fluorescent protein tags with reasonable time resolution. Single-molecule imaging gives an unambiguous bottom-up biomolecule characterization that avoids the “missing information” problem characteristic of ensemble measurements. It has circumvented the diffraction limit by facilitating single-particle localization to ~1 nm. Probes developed specifically for SMD applications extend the advantages of single-molecule imaging to high probe density regions of cells and tissues. These applications perform under conditions resembling the native biomolecule environment and have been used to detect both probe position and orientation. Native, high density SMD may have added significance if molecular crowding impacts native biomolecule behavior as expected inside the cell.
Single-molecule detection and characterization methods have opened a new line of research into the physics of biological processes. They complement and sometimes replace the more traditional ensemble-based observations. In an ensemble-averaged observation, the observable parameter is summed over its value in every state represented in the ensemble. A model for the state distribution provides an expectation value for the observable for comparison with experiment and any model distribution giving the observed value is equally valid. This top-down approach usually does not fully constrain the state distribution leading to an information deficit called “missing information.” The maximum entropy method was created to provide the probability for states in the ensemble that are consistent with observation while maximizing likelihood for all states in the system, thereby providing an unbiased representation of the missing information (Katz 1967). Single-molecule observations detect each state individually. When summed, the single observations are equivalent to the ensemble average. Individually, the observations provide the state distribution. An example is the ion channel in a cell membrane. Single-channel conduction measurements showed discrete open and closed states, the time course, and state distribution or occupation probability from which channel kinetics were deduced. Another example is the myosin motor where the power stroke impelling attached actin is the unitary event. Observing the unitary event by measuring force or protein conformation in time characterized myosin kinetics in contractility. The single-molecule, bottom-up approach has no missing information except for the obscurity created by experimental noise.
The significant advantages of single-molecule over ensemble-average measurements are offset by the experimental difficulty inherent in single-molecule detection (SMD). Single-ion-channel isolation by a patch clamp records the picoamp current in the channel (Neher and Sakmann 1976). Patch clamping turned out to be very practical for single-molecule recordings with the modernized technique a longtime standard measurement (Hamill et al. 1981). Single-ion-channel detection was reported at about the same time as the first SMD using fluorescence. Fluorescence from a single polymer molecule was detected when multiple fluorescein chromophores modified a single polyethyleneimine bound to an antibody adsorbed to a solid substrate (Hirschfeld 1976). Later, filamentous actin (F-actin) labeled with multiple chromophores was visualized in a microscope to quantitate F-actin gliding velocity when propelled by myosin (Yanagida et al. 1984). On a separate track, SMD using protein mechanical properties was invented to investigate the unitary myosin power stroke impelling attached actin during contraction (Kishino and Yanagida 1988). In the power stroke, the myosin head moiety rotates a lever-arm domain through ~70° while hydrolyzing ATP to convert chemical free energy to the mechanical work of moving actin (Eisenberg and Hill 1985; Huxley 1969). Early experimental methods utilized force fluctuations to deduce contributions from individual myosins (Ishijima et al. 1991; Yanagida et al. 1993). With optical traps (Ashkin et al. 1987), force impulses and nanometer step sizes were resolved from single myosins attached to actin (Finer et al. 1994; Simmons et al. 1996). Simultaneous SMD, with fluorescence and mechanical impulse, was reported for a fluorescent-tagged ATP bound to myosin associated with actin (Ishijima et al. 1998).
The requirement for several bound chromophores to achieve SMD relaxed as new methodologies gradually emerged to increase signal-to-noise (S/N) by lowering background intensity, increasing emitted light collection efficiency, and enhancing chromophore stability against photobleaching. The large phycoerythrin molecule was detected in a flow cell where a small volume of solution was laser-illuminated under optimized detection conditions (Nguyen et al. 1987; Peck et al. 1989). Phycoerythrin is a light-harvesting pigment initiating photosynthesis in cyanophytes and containing dozens of chromophores. Fluorescence bursts detected from the flow cell were attributed to passage of single phycoerythrin molecules through the illuminated region. The absorption frequency spectrum from a single pentacene molecule in a crystal was detected at 1.6 K (Moerner and Kador 1989). The single-molecule fluorescence excitation frequency spectrum was also recorded from the same system (Orrit and Bernard 1990). Later, a single rhodamine molecule confined to a micro-droplet was detected by fluorescence (Whitten et al. 1991).
A photostable and high yield fluorescent chromophore embedded at low density in a thin polymethylmethacrylate (PMMA) film spread on a coverslip provided the first images of an immobilize single molecule at room temperature (Betzig and Chichester 1993). To reduce background, the excitation light was confined to a small volume using a near-field scanning optical microscopy (NSOM) probe, while emission was collected by a high numerical aperture objective in a conventional far-field microscope. Raster scanning the tip over the film provided a spatially resolved image of single molecules, their dipolar emission pattern, and a map of the excitation electric field produced near the probe tip.
Room-temperature SMD opened new applications of microscopy to address the problem of single-particle tracking. In light microscopy, the point spread function (PSF) of an objective indicates the three-dimensional intensity profile for a luminous point object at focus. PSF volume is smallest for the highest numerical aperture (NA) objectives, and it indicates the resolving power of the optics. The Raleigh criterion states that two point objects are resolved when they are separated by a distance equivalent to the separation of the PSF peak from its first minimum. This distance is ~200 nm for a high NA objective at visible wavelengths. In single-particle tracking experiments, the challenge is to localize one particle at any instant with the highest precision. The significant quantity is the PSF center and how much it changes with each step. Bobroff derived the accuracy for position measurement from a signal like the PSF (i.e., signal with a finite width) and found position can be measured with far more certainty than signal width (Bobroff 1986). Position accuracy depends mainly on signal-to-noise rather than the PSF shape implying that accuracy is limited by the information content in the signal. Thompson et al. applied these concepts to single-particle tracking in two dimensions demonstrating point object position detection with an order of magnitude better certainty than PSF width (Thompson et al. 2002). Yildiz et al. tracked the position of fluorescent probes placed at different points along the rotating myosin V lever arm as the myosin moved processively along an actin filament (Yildiz et al. 2003). The 1.5 nm position accuracy distinguished between two competing models for myosin V motility. Recently, Pierobon et al. performed similar experiments in live cells (Pierobon et al. 2010).
Single-molecule tracking in three dimensions required additional development to boost axial tracking precision. The axial dimension, co-linear with the optical axis of the microscope, is less resolved than lateral dimensions due to the loss of light propagating towards the objective but at angles beyond the objective numerical aperture. The introduction of a cylindrical lens to the microscope emission pathway causes astigmatism that leads to increased single-particle localization precision in the axial dimension with a slight reduction in lateral dimension precision (Holtzer et al. 2007). A double-helix PSF is the basis for another approach producing an intensity profile from a single particle consisting of two lobes in the image plane that rotate with the axial position of the particle (Pavani et al. 2009). Finally, imaging a point source at two different focal planes indicates the intensity profile changes that depend on axial dimension (Ram et al. 2008). These new methods provide comparable precision for lateral and axial dimensions.
Lately emphasis has been on capturing the full information content of probe position from fluorescence. Ober et al. have derived expressions estimating the absolute precision limits given microscope specifications (Ober et al. 2004). Maximum-likelihood fitting of the theoretical PSF to fluorescence data obeying Gaussian or Poisson statistics conforms to different optimization schemes (Stoneking and Den Hartog 1997). Mortensen et al. confirmed that maximum-likelihood fitting for Poisson statistics is optimal with photon-counting detection (Mortensen et al. 2010).
Biological applications for fluorescence also involve detection of molecules inside a cell or within a multicellular tissue where there is less or no control over fluorescent marker concentration. In muscle for instance, myosin concentration in the sarcomere is ~300 µM giving ~180 molecules per attoL (10−18 L) or ~0.006 attoL per myosin, while a diffraction-limited illumination volume is larger than 10 attoL (Tikunov et al. 2001). Other biologically interesting molecules also localize to a portion of the cell or tissue in a high density cluster thereby defining a structure or activity center where a cellular function is carried out. Careful attention to labeling conditions and data analysis permitted single-actin rotation detection in cardiac muscle fibers with confocal microscopy (Muthu et al. 2010), however, detecting fluorescent-tagged single proteins clustered at high density usually requires (1) below diffraction limit detection volume, (2) selective activation of a subpopulation of the total fluorescent probes, or (3) a combination of (1) and (2). There has been steady progress in both areas.
Below diffraction limit illumination volumes were created using near-fields. NSOM squeezes light through a nanometer size aperture on the end of a fiber optic that is scanned over the sample (Lewis et al. 1984; Pohl et al. 1984). The near-field is nonpropagating and confined to within a few nanometers of the tip thus defining an illuminated volume that would contain on average fewer than one myosin molecule in a muscle fiber sample. An array of nanometer size apertures were also constructed in a thin metal film covering a planar substrate for detection of single molecules in high concentration (Levene et al. 2003).
Total internal reflection fluorescence (TIRF) is another near-field (also called evanescent field) technique created when excitation laser light is incident on the glass side of a glass/aqueous interface at angles greater than critical angle, θc, for total internal reflection. Although light is totally reflected, an evanescent field created in the water medium decays exponentially with distance from the interface and excites fluorophores within ~100 nm of the surface. Harrick described theory and application for various internal reflection techniques (Harrick 1979). Axelrod and coworkers combined TIRF with microscopy to take advantage of the small excitation volume created by the evanescent field and the high efficiency of microscopic optics to collect emitted light (Axelrod 1981; Axelrod et al. 1983; Burghardt and Axelrod 1981; Thompson et al. 1981). TIRF microscopies with through-a-prism or through-the-objective illumination options have complementary properties. In through-the-objective TIRF microscopy, focused (Burghardt et al. 2006a; Ruckstuhl and Seeger 2003, 2004) or unfocused (Axelrod 2001; Borejdo et al. 2006) (shown in Fig. 1) versions provide detection volumes of ~3 or ~7 attoL (10−18 L), respectively. Recently developed, around-the-objective TIRF has a narrower penetration depth (~50 nm) and other characteristics complementing previous methods (Burghardt et al. 2009a). Another TIRF microscopy variation uses a parabolic mirror objective (Ruckstuhl and Seeger 2003).
The interface affects probe emission in TIRF. Figure 2 shows an emitting fluorophore (µ) in the aqueous medium near a dielectric interface (water/glass, ignore the metal film for now). The probe dipolar emission contains both propagating (far-field) and evanescent (near-field) contributions. The near-field is perturbed by the interface into propagating waves that appear in the glass medium beyond critical angle (supercritical angle fluorescence or SAF) (Ruckstuhl and Verdes 2004). Oil immersion objectives with NA 1.3 capture SAF. Detectable SAF intensity decreases with probe distance from the interface creating a second layer of spatial selectivity for probes nearest to the interface that is in addition to the exponential decay from the exciting field.
Hellen, Axelrod, and Fulbright introduced a thin metal film to the dielectric interface in TIRF as indicated in Fig. 2 (Fulbright and Axelrod 1993; Hellen and Axelrod 1987). The metal film permits negligible excitation (green) transmission for all incidence angles except near to the surface plasmon angle, θsp, where transmission enhancement occurs due to resonant excitation of electron oscillations (surface plasmon resonance or SPR) propagating along the water/metal interface. The SPR field in the water medium is evanescent, decaying exponentially with distance from the interface. Metal film TIRF selectively detects probes within a thin layer near the interface because all detectable emission on the glass side of the interface is SAF, and fluorescence from probes within ~10 nm of the interface is quenched. Bare glass (left) and metal film (right) TIRF from a rhodamine-labeled muscle fiber are compared in Fig. 3 (Burghardt et al. 2006b). The striated pattern is due to localization of rhodamine-labeled myosin in the sarcomeres (Fig. 1). The image shows the metal-coated side has lower background and shallower detection volume.
Other modern microscope techniques achieve subdiffraction limited resolution including 4Pi excitation and detection (Gugel et al. 2004), structured illumination (SI) (Gustafsson 2005), and the stimulated emission depletion (STED) of peripheral excited molecules in a diffraction limited spot (Donnert et al. 2006). These techniques manage 10–100 nm scale detection volumes and can image the specimen over three dimensions unlike the near-field NSOM and TIRF techniques that are confined to the specimen surface.
Modern inorganic fluorescent probes are photostable and bright chromophores ideally suited to SMD. Likewise, variants of the green fluorescent protein (GFP) have proven to be highly adaptable to evolving microscopy applications.
The wild type GFP (wGFP) chromophore possesses two long wavelength absorption bands at ~400 and 480 nm with emission at ~510 nm. The 400 and 480 nm absorption peaks correspond to mixed populations of neutral phenols and anionic phenolates, respectively, from residue Y66 (Brejc et al. 1997). Polarized absorption from crystalline wGFP indicated the absorption dipole moment directions for the two visible bands relative to the atoms in the fluorophore and the crystallographic axes (Rosell and Boxer 2003). Absorption and emission dipole moments were shown to be parallel in the longer wavelength band in wGFP and wGFP-tagged proteins (Inoué et al. 2002; Rocheleau et al. 2003).
Photoactivatable GFP (PA-GFP) is a wild type variant where the isoform producing the characteristic 400 nm absorption band is preferentially stabilized by a T203H substitution (Patterson and Lippincott-Schwartz 2002). The PA-GFP has little fluorescence under 488 nm excitation until photoactivated by irradiation in its 400 nm band. Activation causes a 100-fold increase in fluorescence excited by 488 nm light (Lippincott-Schwartz and Patterson 2003). Activated PA-GFPs remain in their activated conformation for weeks.
PA-GFP in photoactivated localization microscopy (PALM) accomplishes superresolution microscopy inside a cell or tissue when probe concentration is high (Betzig et al. 2006). PA-GFP and other photoactivatable fluorescent proteins (PA-FPs) tagged target proteins produced in cells. The targeted proteins assemble into their cell structures at high density. Mild photoactivation produced a sparse concentration of fluorescent species within the high density tagged protein structure that were continuously illuminated and individually imaged until photo-bleached. The single PA-FP image was then localized to much higher precision than the diffraction limit by locating the center of the fluorescence emission with high precision. A photo-switchable organic chromophore, Cy3, and Cy5 in close proximity replaced PA-GFP in stochastic optical reconstruction microscopy (STORM) (Bates et al. 2005; Rust et al. 2006). New photo-switchable organic probes show promise for superresolution imaging in live cells (Lord et al. 2008).
We performed in situ single-myosin orientation sensing in a muscle fiber using PA-GFP. Myosin head moiety (subfragment 1 or S1) crystal structures suggested that small conformational changes in the active site induced by ATP hydrolysis are amplified to the large linear displacements by the lever-arm rotation (Geeves and Holmes 1999). Two myosin light chains bound to the lever arm, essential and regulatory (ELC and RLC), are exchangeable in a muscle fiber making them targets for probe modification for in situ investigation of lever-arm orientation (Wagner and Weeds 1977).
An orientation-sensitive spin probe was the first to be introduced to the lever arm by modification of an isolated RLC that was exchanged with the native RLC in a permeabilized muscle fiber (Arata 1990). Genetically engineered RLCs containing cysteins selectively modified at the SH groups by mono- or bi-functional organic fluorescent probes were likewise introduced to a fiber with light chain exchange. RLC dynamics was detected with time-resolved fluorescence polarization indicating lever-arm rotational movement during contraction (Corrie et al. 1999; Irving et al. 1995). Fluorescent-labeled RLC and ELC were simultaneously exchanged into fibers (Borejdo et al. 2002). We exchanged PA-GFP-tagged human cardiac RLC (HCRLC-PAGFP) with native light chain on myosin cross-bridges in permeabilized muscle fibers. Sparse photo-activation of the PA-GFP permitted detection of single-myosin lever-arm orientation across the fiber sample (Burghardt et al. 2009b).
Figure 4 depicts a microscope setup for the photo-activatable probes. The pump laser photoactivates PA-GFP during a short intense pulse of light. The argon-ion probe laser excites fluorescence from photoactivated PA-GFP. The Pockels cell (PC) rotates linear electric field polarization of the laser illumination, and the dichroic mirror (DM) reflects pump and transmits probe light. Beams are co-linear after the DM, enter the microscope epi-illumination port, and are focused to the same point on the sample by the objective. Excited fluorescence is collected by the objective. The beam splitter separates emitted light into two perpendicular linear polarizations that are imaged on separate halves of the CCD camera. Pixel size (6.45 µm) together with the evanescent field depth defines a detection volume of ~2× 106 nm3 for the 60× objective (Burghardt et al. 2006a). Figure 1 shows the sample chamber where an RLC-PAGFP exchanged muscle fiber in aqueous buffer solution contacts the coverslip and is illuminated by the evanescent field.
Figure 5 shows F‖, (fluorescence for excitation polarization parallel and emission polarization perpendicular to the fiber axis) from a portion of an RLC-PAGFP exchanged and sparsely photoactivated fiber in isometric contraction under TIRF. Activated PA-GFPs are bright spots superposed on the background fluorescence from unphotoactivated weakly fluorescing PA-GFPs. Photoactivated PA-GFPs are apparently single molecules because their density in the fiber image implies an infinitesimal probability for two photoactivated molecules to reside in one pixel. Varied single-molecule intensities indicate varied dipole orientation relative to the emission polarizer or varied proximity to the TIR surface where evanescent field intensity is maximum. An equivalent image for F‖, ‖ is collected simultaneously.
Figure 6 shows single-molecule polarization P‖ = (F‖, ‖− F‖, )/(F‖, ‖+ F‖ ) histograms for fibers in rigor and isometric contraction. The sum of two Gaussian distributions (black line) fits the data (■). The area under individual Gaussians (red or blue lines) corresponds to myosin lever-arm orientation subpopulation occupation probability. Occupation probability is the most significant characteristic, and it is also indicated in Fig. 6. Data regarding isometric contraction and rigor characterize extent and direction of lever-arm swing when transisting from the high free energy state to the low free energy state, respectively. Subpopulation curve with <P‖>~0.46 in red and predominant in isometric contraction (occupation probability 0.61) shifts to the curve with <P‖>~0.19 in blue and predominant in rigor (occupation probability 0.68). Shifting from P‖ = 0.46 to 0.19 corresponds to a dipole moment rotation that could be as large as 80° or as small as 20°. The angular range reflects the different dipole orientations that correspond to a given P‖.
The single-molecule assessment of lever-arm unitary rotation can be compared among myosin isoforms and between native and disease-linked myosin mutants. Modification of the myosin unitary rotation alters its defining functional requirement, which could play a role in disease phenotype.
Single-molecule techniques clarify averaging ambiguities unavoidable with ensemble measurement (Neuweiler and Sauer 2005), nevertheless, single molecules in vitro do not represent the native system in a cell where molecular crowding is the norm (Minton 2001). In the previous section we addressed experimental methods needed to perform SMD in crowded cells and tissues. This is a natural progression from artificial in vitro systems that is justified generally by the intuition that we should conduct experiments under native conditions. In addition, molecular crowding has a thermodynamic effect on free energy providing a specific rationale for pursuing in situ SMD.
Molecular crowding contributes substantially to the total free energy of a protein under native conditions. It changes rates and equilibria of macromolecular reactions relative to those measured in dilute conditions. A crowded environment causes preferential hydration of a protein due to the excluded volume effect thereby favoring lower surface area structures and promoting self-association (Timasheff 2002). The consequences of crowding were observed in skeletal muscle myosin where artificial crowding of the protein was shown to influence steps critical in the contraction cycle (Peyser et al. 2003). In the Mg-mediated myosin ATPase cycle without actin, dissociation of products is coupled to a 70° lever-arm swing to reprime myosin in the contraction cycle. Polyethylene glycol (PEG)-induced crowding was shown to inhibit product release. The molecular volume and surface area of the myosin in various conformations showed the M** state, just preceding product release, is smallest and should be favored with crowding. Thus crowded conditions will inhibit the product release and swing of the lever arm necessary to exit the M** conformation as observed. Cardiac and skeletal muscle myosin have evolved to function optimally under crowded conditions, hence their behavior may depend upon maintenance of the crowded native environment.
SMD with fluorescence is a commonly used microscopic measurement technique for the bottom-up characterization of biomolecule structure and function. Single-molecule imaging has circumvented the diffraction limit by facilitating single-particle localization to ~1 nm. Probes developed specifically for SMD applications extend the advantages of single-molecule imaging to high probe density regions of cells and tissues. High density SMD may have added significance if molecular crowding affects native biomolecule behavior.
The work was supported by the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) grant R01AR049277; the National Heart, Lung, Blood Institute (NHLBI) grant R01HL095572; and the Mayo Foundation.
Thomas P. Burghardt, Department of Biochemistry and Molecular Biology, Mayo Clinic Rochester, Rochester, MN 55905, USA. Department of Physiology and Biomedical Engineering, Mayo Clinic Rochester, Rochester, MN 55905, USA.
Katalin Ajtai, Department of Physiology and Biomedical Engineering, Mayo Clinic Rochester, Rochester, MN 55905, USA.