Expanding single-molecule fluorescence spectroscopy to capture complexity in biology

Abstract

Fundamental biological processes are driven by diverse molecular machineries. In recent years, single-molecule fluorescence spectroscopy has matured as a unique tool in biology to study how structural dynamics of molecular complexes drive various biochemical reactions. In this review, we highlight underlying developments in single-molecule fluorescence methods that enable deep biological investigations. Recent progress in these methods points toward increasing complexity of measurements to capture biological processes in a living cell, where multiple processes often occur simultaneously and are mechanistically coupled.

Introduction

Uncovering the diverse mechanisms of complex cellular processes is crucial to understanding life and its maintenance. The ideal description of biological processes would be a movie, in which the temporal evolution of the conformational and compositional states of the participating molecular machinery is dynamically described at atomic resolution. Various biochemical, biophysical, and computational methods have been developed to capture vignettes of a molecular process, which are then woven together to delineate an underlying molecular movie. Single-molecule fluorescence spectroscopy has become an indispensable tool to investigate molecular mechanisms of diverse biological processes. The power of single-molecule fluorescence methods lies in its ability to track structural dynamics at single-molecule resolution, linking existing structural and biochemical information into a mechanistic model of a given process.

Application of single-molecule fluorescence to complex biological processes is hindered by two major methodological limitations: First, only a small number of fluorescence channels are available as observables for monitoring local structural dynamics, blinded to correlated global rearrangements within a molecular complex. Second, high-resolution, single-molecule measurements are often performed in vitro using biochemically reconstituted components, which may perturb the native pathway. In this review, we aim to provide an overview of the recent advances in single-molecule fluorescence methods to meet these challenges, including: i) new methods to label biomolecules with stable and photophysically optimized fluorescent probes, ii) improvements in instrumentation and probing strategies to increase temporal resolution, the throughput of each experiment, and the number of observable signals, and iii) carefully designed biochemical systems that mimic in vivo conditions during single-molecule measurements. These advances will enable single-molecule fluorescence methods to capture complex biology with high precision.

New chemical methods and dyes to site-specifically label biomolecules

Single-molecule fluorescence experiments require labeling of biomolecules with photostable dyes. For most single-molecule applications, fluorescent proteins are too dim and lack the requisite photostability (although this is an advantage in super-resolution microscopy [1•,2•]). Thus, exogenous labeling with organic fluorophores is usually required for high precision measurements, with subsequent biochemical benchmarking to ensure that the labeled biomolecule retains activity. Recent advances in biochemical and chemical methods have greatly facilitated the specific labeling of proteins and nucleic acids for single-molecule fluorescence. For nucleic acids, dyes can be incorporated by chemical synthesis into shorter (up to 200 nucleotides, albeit costly) RNAs or DNAs, and then ligated to create longer fragments. An elegant example of this approach is the work by Hoskins et al. on the spliceosome [3••]: Single-molecule Förster resonance energy transfer (smFRET) is used to monitor the distance between donor and acceptor probes [4,5••]. By this approach Hoskins et al. directly monitored dynamic rearrangement during the splicing of mRNA. Other RNA labeling approaches leverage specific chemistry of modified RNAs [6], or simply hybridization of a labeled RNA or DNA to a complementary strand in the target nucleic acid [7]. Finally, site-specific labeling of RNAs using pre-labeled NTP substrates can be achieved using in vitro transcription via a hybrid solid–liquid phase transcription method [8].

Traditionally, single-cysteine mutants and labeling with dye-maleimide conjugates have been the primary approach for specific labeling of proteins. However, this method often fails or is not feasible for larger eukaryotic proteins; alternative approaches have fortunately been developed that allow multiple rounds of site-specific labeling using orthogonal chemistries (for conjugating different dyes specifically on the same protein molecule). Use of suppressor tRNAs allows incorporation of modified amino acids that can be used with CLICK or other forms of efficient aqueous chemistry for coupling [9]. Alternatively, peptides can be chemically synthesized with modified amino acids and then ligated to assemble a full-length protein, an approach pioneered by Dawson et al. [10]. Fusion of target proteins with a protein tag such as SNAP [11] or HALO [12] provides an efficient approach to protein labeling, with the caveat that protein tags (with sizes near 20 kDa) do not interfere with the biological process of interest. To circumvent the need for a large protein tag, short (10–12 amino acids) peptide tagging systems have been developed, which harness enzymes that recognize a specific peptide and perform chemistry to create a covalent dye-tag conjugate. This approach has been successfully applied to many systems, including the HIV-1 virion [13]. This powerful repertoire of labeling methods has facilitated new applications of single-molecule fluorescence described below.

In addition to novel ways of labeling target biomolecules, the fluorescent dyes themselves have been undergoing optimization, such as creating dyes with greater brightness, improved photocycles, and the elimination of off pathway triplet states. One example is the work of Blanchard et al. whereby the photostability of probes was improved by chemically attaching a triplet-state quencher directly to organic fluorophores [14]. If the size of a fluorescent probe is irrelevant to the biochemical reaction, solid-state dyes such as quantum dots and nano-diamonds may be used [15]: these dyes have far superior photophysical characteristics compared to organic fluorophores and fluorescent proteins, with high brightness and very long dye lifetimes without photobleaching. When applied judiciously, these continually evolving solid-state dyes may enable high-resolution, single-molecule fluorescence measurements.

Improvements in temporal resolution, throughput and signal collection

In addition to improvements in organic dyes and labeling methods, improved optics in collecting and processing signals is another critical advance in single-molecule fluorescence spectroscopy. One of the most exciting recent developments is the use of scientific, complementary, metal-oxide semiconductor (sCMOS) detectors for collecting emitted fluorescence signals [16••]. Traditionally, electron-multiplying, charge-coupled device (EMCCD) cameras have been used to detect single-molecule fluorescence due to their superior performance within the low photon-per-pixel regime. The temporal resolution of EMCCD cameras is typically limited to the 10 ms range, while simultaneously imaging 100–1000 molecules. Further improvements in temporal resolution come at a cost of a smaller number of imaged molecules. In recent years, sCMOS cameras now have signal-to-noise performance comparable to EMCCD cameras for observing typical single-molecule fluorescence signals, while simultaneously imaging more than 1000 molecules with a temporal resolution of 1 ms (Figure 1) [2•,16••]. The use of sCMOS cameras in single-molecule fluorescence instruments allows measurements at high temporal resolution with high-throughput. Combined with single-molecule FRET methods that can probe structural dynamics, the sCMOS instrument can detect tRNA conformational changes during protein translation in the millisecond timescale [16••].

Figure 1.

Different modes of single-molecule fluorescence spectroscopy. Fluorescent signals from molecules tethered to a surface are recorded in real time. The number of simultaneously monitored molecules are limited by the size of the camera chip. Structural dynamics within the molecular complex can be probed using smFRET, where Förster energy transfer between two fluorescent probes report their distance, typically in 20–80 Å range. Discrete quantized steps in fluorescent intensity correspond to changes in the distance between the fluorescent donor and acceptor dye molecules. Compositional dynamics can be monitored as colocalization of different molecules within the same complex, a technique referred to as colocalization single-molecule spectroscopy (CoSMoS). High background caused by direct excitation of diffusing fluorophores can be suppressed using zero-mode waveguides (ZMWs), which limit the excitation volume within the nano-fabricated structures.

Because the camera technology is the fundamental component in single-molecule fluorescence instruments, its improvement will have rapid impact on the field.

Another advance is improved suppression of fluorescent background from diffusing fluorophores by the use of zero-mode waveguides (ZMW) [17]. The conventional approach for background suppression utilizes either confocal microscopy for imaging away from the surface or total internal reflection microscopy (TIRFM) for selective imaging near the surface. If a physiologically relevant system can be established for the surface-tethered molecules, then TIRFM is preferred for single-molecule fluorescence studies for its ability to track multiple molecules continuously in parallel. Yet, the use of TIRFM limits the concentration of directly excited fluorescent molecules in solution to 5–50 nM, due to the transient excitation of diffusing fluorophores within the excitation volume. ZMW technology limits the excitation volume near the surface using pre-patterned nano-structures (Figure 1). This allows up to 10 μM of freely diffusing, fluorescently labeled molecules in solution, which is necessary to study biological phenomena involving medium to low binding affinity with high nanomolar to low micromolar dissociation constants. Combined with indirect illumination using FRET, up to a millimolar range of fluorescently labeled molecules can be included in the experiment without drowning the signal of interest [18]. Within this concentration range, study of the association and dissociation dynamics of physiologically-relevant factors such as GTP and ATP may be feasible, which are important steps in powering various molecular machines in the cell. Instruments combining ZMW-technology and sCMOS camera have allowed the study of co-translational and co-transcriptional processes with high-throughput assays at physiologically relevant conditions [19,53], necessary for investigating rare events with statistical robustness.

Biochemical developments of single-molecule assays to mimic in vivo condition

While these chemical and optical advances in single-molecule fluorescence techniques have enabled highly sensitive biophysical measurements, investigation of complex biological phenomena requires reconstitution in physiologically relevant systems amenable to single-molecule fluorescence measurements. The in vitro reconstituted systems are typically tailored to enhance the observed fluorescence signal. Reconstituting a complex biological system can be challenging because individual factors participating in the process need to be identified, biochemically purified, and included in the reconstituted system with physiologically relevant stoichiometries of the components. The purification of individual components of a biological system is often laborious, expensive and in some cases impossible, obviating the reconstituted system approach. These difficulties have been tackled by biologists and biochemists using creative and well-designed biochemical assays based on single-molecule fluorescence.

Observing biochemical activity under physiological conditions has been a standard problem in the study of protein dynamics, especially for membrane proteins. In addition to changes in protein dynamics due to the absence or presence of their protein and/or nucleic acid binding partners, the local environment of a protein of interest may also influence its structural dynamics and therefore its function within the cell. Biological processes occurring near the cell membrane such as cell signaling and viral infection have biomedical importance, and recreating an appropriate physiological condition is critical to their study. This challenge has been addressed using several general approaches for single-molecule fluorescence-based assays (Figure 2a): First, the glass surface of the microscope slide can be treated with a phospholipid bilayer [20]. Here, the protein is directly tethered to the surface using surface-immobilized, biotinylated bovine serum albumin, but surrounded with the lipid bilayer to interact freely. Second, membrane proteins can be immobilized in nanodiscs [21], which are engineered scaffold proteins that assemble into a circular belt to enclose a lipid bilayer containing the membrane protein of interest. This has allowed biophysical and structural studies of isolated membrane proteins [22]. Third, large liposome vesicles containing the protein of interest can be directly tethered to the surface by labeling the head group of lipids with biotin [13,23••,24••,25]. The second approach has been especially successful in studying viruses [13,23••,24••], as entire virions containing molecules of interest can be monitored at the single-particle level. Combined with structural and dynamic information from NMR spectroscopy, X-ray crystallography, cryoEM and/or molecular dynamic simulation [5••,26,27], smFRET-based approaches will be powerful in describing the local and global structural dynamics of diverse membrane proteins.

Figure 2.

Reconstitution of physiological conditions in single-molecule fluorescence experiments. Interaction between the molecule of interest and cellular components is reproduced via: (a) embedding molecules such as a membrane protein into a local environment that mimics the lipid bilayer (e.g. a nanodisc), (b) supplementing all biological factors needed for the mechanism studied, such as the addition of a cell lysate, and (c) fluorescent labeling of molecules within living cells and observing them in vivo. Figure of bilayer and membrane protein adapted from [20].

Along with recreating physiological conditions in vitro, the ability to purify biomolecules participating in the biological system of interest has hampered the application of single-molecule methods. In particular, the mechanisms of genetic information transfer and maintenance most often involve complex machinery with 10–100 unique factors participating in basal and regulated functions. Recent developments in cell-extract technology have considerably lowered the hurdle to study these processes (Figure 2b), exemplified by studies of pre-mRNA splicing and DNA repair. In the case of splicing, the use of cell nuclear extract supplemented with the fluorescently labeled components of the spliceosome complex has enabled observation of splicing events happening in real time without a need to purify and reconstitute individual factors [3••,28,29]. Similarly, a non-homologous end-joining mechanism to repair DNA double-strand breaks has been monitored using single-molecule fluorescence-based assays [30•,31].

Here, the factors participating in distinct stages of enzymatic progression have been identified using lysates from cells expressing different set of factors. A general use of cell extracts in reconstituted systems should improve biochemical activities in various single-molecule fluorescence assays and link insights from in vitro systems to molecular processes in vivo.

The ultimate goal of any mechanistic study is to understand biological processes occurring in vivo. Therefore, various in vivo single-molecule fluorescence spectroscopy methods have been developed recently (Figure 2c), particularly taking advantage of super-resolution imaging [1•,2•]. While in vitro measurements provide more rigorous control and achieve higher signal-to-noise ratios in general, in vivo measurements provide key information, such as biomolecule trafficking and localization within the cell. Among the host of important reports in the past couple of years, a particularly compelling use of in vivo single-molecule fluorescence imaging is tracking of translating ribosomes by labeling both the nascent peptide via a peptide-tag and the mRNA using an MS2-tag that recognizes particular RNA secondary structures [32•,33•,34•,35•]. The ability to track trafficking of translating mRNA has been yearned for a long time, and the development of a peptide-tagging approach has catalyzed multiple labs to develop translated mRNA tracking assays. Another exciting front in the application of in vivo single-molecule fluorescence methods is the study of phase-transitions and phase separations within a cell, which results from forming intermolecular clusters of RNA and/or protein molecules [36–38]. Live-cell imaging studies are currently being revolutionized by the use of lattice light-sheet microscopy in combination with adaptive optics [39••], which provides unprecedented temporal and spatial resolution of cellular processes at long timescales. Combined with the existing and new ways to introduce stable chemical dyes into live cells [40,41•], high-resolution in vivo imaging may be a future operating mode for single-molecule fluorescence studies.

Future direction for monitoring coupled biological processes

The biochemical and biophysical advances discussed above continue to facilitate investigation of complex biological processes using single-molecule fluorescence spectroscopy. However, single-molecule studies described thus far have a disadvantage in their small number of observables, inherently limited by the available channels within the visual spectrum of light. Increasing the number of simultaneously tracked factors or movements would allow a cross-correlation of observed structural dynamics, which would not only ensure the fidelity of the measurements but also allow the study of other simultaneously occurring processes and how they are coupled. Here, we highlight recently developed methods to increase the number of observables in each assay, which enables future studies of coupled biological processes.

While smFRET is a versatile tool and used as a common mode for single-molecule fluorescence studies, it provides only one observable where the structural dynamics are mapped to a single distance relating two fluorescent probes. Correlated structural dynamics are a common feature of biomolecular systems, in which multiple processes occur simultaneously and are often coupled. Correlated sequences of structural rearrangements have been successfully monitored using augmented smFRET methods such as Three-color FRET [42–45] or multiple dye-quencher methods [46•], where multiple distance changes corresponding to conformational and/or compositional dynamics are tracked within a single molecule. Alternatively, fluorescence intensities of individual dyes may be utilized to infer underlying molecular events. A change in fluorescence may be due to a change in the local environment of the molecule, the presence of amino acid side chains or nucleotides that quench or enhance the nearby fluorophore [47]. While predicting a precise change of fluorescence attached to a specific position of a protein is challenging, intensity change information can be empirically characterized and used as another tool for monitoring conformational or compositional changes during a biological progress.

Combining fluorescence measurements with other single-molecule methods such as force spectroscopy forms a powerful tool for studying biological processes (Figure 3a), as most conformational changes coupled to a biochemical reaction can be monitored and manipulated by force spectroscopy. Force spectroscopy approaches, such as optical and magnetic tweezers, have contributed unique insights to our understanding of molecular motors, where motor motion has been tracked with high spatial and temporal resolutions. By coupling mechanical observations made with force spectroscopy to compositional and/or conformational states observed using fluorescence spectroscopy, a detailed relationship between molecular state and mechanical function can be described at the single-molecule level. Fluorescence measurements have been incorporated in magnetic trap [48] and high-resolution optical tweezer experiments [49], resulting in detailed mechanical insights into helicase molecular motors [50]. Recently, a high-resolution magnetic tweezer setup that measures torque and force was coupled with fluorescence measurements [51••]. Multi-modal measurements using hybrid instrumentation not only increases the number of observables from a biological process, but also allows physical manipulation of biomolecules to reveal the underlying complex energy landscapes.

Figure 3.

Combining single-molecule fluorescence measurements with other spectroscopic or microscopic methods. (a) Fluorescence spectroscopy can be combined with a force spectroscopy method, such as rotor bead tracking shown here (figure adapted from [51••]). One end of the complex is tethered to the magnetic bead to either induce mechanical force into the system or confine the molecular movements within the convenient axis, which are detected as scattering light from the rotor bead in addition to the fluorescence signal. (b) Fluorescence tagging of a molecular complex can be used with electron cryotomography in tandem, where the fluorescence imaging of a frozen cell reveals broad localization of specific molecules, providing additional information for the electron tomography.

Finally, single-molecule methods and approaches can be combined with other structural and biophysical methods to provide correlative data. NMR spectroscopy and single-molecule methods both probe dynamics but in distinct manners and over different timescales; combining the two approaches has been powerful. The exciting evolution of electron cryomicroscopy (cryoEM) has allowed structural heterogeneity of different systems to be gleaned directly, and these states can be dynamically sampled by single-molecule methods. In addition, correlative light and electron cryotomography allows the direct merging of the two methods, albeit in frozen samples (Figure 3b) [52]. Such approaches have been powerful in conjunction with cryo-tomography to gain detailed structural insights (at a nanometer range) into cells and tissues, and correlating features such as protein fluorescence that would be observed in standard or super-resolution light microscopy. Merging multiple approaches and thereby harnessing their distinct advantages will be the key to gaining deeper insight into the mysteries of biology.