Intranucleus Single-Molecule Imaging in Living Cells

Abstract

Many critical processes occurring in mammalian cells are stochastic and can be directly observed at the single-molecule level within their physiological environment, which would otherwise be obscured in an ensemble measurement. There are various fundamental processes in the nucleus, such as transcription, replication, and DNA repair, the study of which can greatly benefit from intranuclear single-molecule imaging. However, the number of such studies is relatively small mainly because of lack of proper labeling and imaging methods. In the past decade, tremendous efforts have been devoted to developing tools for intranuclear imaging. Here, we mainly describe the recent methodological developments of single-molecule imaging and their emerging applications in the live nucleus. We also discuss the remaining issues and provide a perspective on future developments and applications of this field.

Introduction

The mammalian nucleus is a complex multicomponent and multiphase system that acts as a command center to regulate various life events inside cells. The dynamic and complicated interplay of highly diverse proteins, RNA, and chromatin within the nucleus operates over several orders of magnitude in space and time. These interactions establish and maintain cell activities, including cell fate decision, self-maintenance, differentiation, and reprogramming. Compared with biochemical experiments, statistical information obtained from largely stochastic molecular events by single-molecule imaging can overcome the loss of information due to ensemble averaging in traditional bulk imaging and generate new insights about the heterogeneity of complicated cellular activities and deepen our understanding of the underlying working mechanisms (1, 2, 3). These advantages have rendered single-molecule imaging as a powerful tool to address important questions in which randomness and inhomogeneity are inherent to the dynamics of biomolecules. Particularly, single-molecule imaging in living nuclei has emerged as a powerful and unique tool for the mechanistic study of some essential nuclear processes that otherwise are difficult to study using conventional techniques (Fig. 1). These include recent beautiful works on the repair dynamics of DNA double-strand breaks (4), mRNA transporting through the nuclear pore (5), and enhancer cluster formation in embryonic stem cells (6). However, although numerous living cell single-molecule imaging experiments have been successfully conducted at the plasma membrane as well as in the cytosol, there are a limited number of such studies in the nucleus, mainly because of difficulties with labeling and the high fluorescence background in the nucleus. This review focuses only on single-molecule optical detection in the nuclei of living cells, in particular some fundamental processes involving chromatin folding, transcription, RNA exporting, chromatin remodeling, and histone- or DNA-associated enzymatic activities. In addition, we also summarize the available and practical labeling methods and imaging modes for the biomolecules, including DNA, mRNA, and proteins.

Figure 1.

Biological objects in the nucleus that can benefit from single-molecule imaging for mechanistic studies. The chromatin DNA is packaged into high-order structure with the assistance of structural proteins, including lamins, histones, CTCF, and cohesin. Based on the loop-extrusion model, CTCF and cohesin anchor the chromatin loop and organize multiple loops into topological-associated domains (TADs). Different TADs with similar epigenetic signatures are characterized by stronger interdomain interactions and are further organized into compartments (colored shapes in the nucleus represent different compartments). Most intranuclear processes occur within these structures. The red-star-labeled objects indicate objects that have been studied by single-molecule imaging. (a) The labeling of the chromatin locus is shown. Chromatin loci in different states (cell cycles, epigenetics, and nuclear position) can have different motion behaviors. (b) RNA single molecules get through the nuclear pore complex. The transcripts leave the transcription sites and diffuse through the pathways intertwined between compact chromatin domains and compartments until the nuclear pore complex. (c) CTCF/YY1 and cohesin anchor the chromatin loop. Binding of the cohesin complex creates initial chromatin loops, which extend in both directions until anchor proteins (CTCF/YY1) are encountered. This brings within close proximity two anchor-protein-binding sites in divergent orientation, potentially leading to the dimerization of two anchor proteins. (d) The dynamics of nucleosomes and chromatin fibers is shown. Various nuclear processes, including DNA replication and transcription, involve chromatin remodeling to disrupt nucleosomes for the exposure of DNA or to exchange core histones with histone variants, such as H2A.Z or H3.3. (e) Transcription factors (TFs) search the genome by 3D diffusion and one-dimensional sliding with fast kinetics until the target site in the enhancer region is encountered. Once TFs find the target site, they will reside at the position with longer duration and then initiate the transcriptional response. Enhancer-promoter loops bring transcription factors bound to the enhancer in close spatial proximity to the promoter. This interaction is thought to be stabilized by the mediator. (f) Epigenetic effectors dynamically interact with chromatin. Epigenetic writers catalyze the addition of chemical moieties onto either histone tails or chromatin DNA. Epigenetic readers recognize specific epigenetic marks and are recruited to these regions dynamically. Epigenetic erasers remove chemical groups from the histone or DNA. (g) The chromatin remodeling process is shown. The ATP-dependent chromatin remodeling complexes bind to chromatin, either moving, ejecting, or restructuring nucleosomes. (h) DNA repair machinery are shown. The central reaction in homologous recombination is the pairing and exchange of strands between two homologous DNA molecules. A number of effectors catalyze this process. (i) Telomeres form T-loops, which consist of complicated ribonucleoprotein. Shelterin is anchored on the telomeric DNA by two duplex DNA-binding factors: TRF1 and TRF2. Telomerase adds repetitive DNA sequences to each chromosome end to maintain genome integrity. (j) A diagram shows the replication folk. Various proteins constitute the complicated machinery at the fork. The leading and lagging stands are replicated in different manners.

Single-molecule imaging and labeling strategies of intranuclear biomolecules

According to the central dogma of molecular biology, genetic information encoded in DNA is transcribed to mRNA, which is then translated to protein. All the three major components of central dogma can be studied in the living prokaryotic and eukaryotic cells by single-molecule imaging techniques. We only introduce recent technical developments and findings in eukaryotic cells. For details about prokaryotic cells, we refer the readers to recent excellent reviews on these topics (7, 8).

DNA

On the most fundamental level, almost all the processes operating in the nucleus involve the chromatin as scaffold. Individual molecules and macromolecular machinery either participate in the organization and regulation of chromatin structure or use chromatin as substrates for their enzymatic reactions. The chromatin organizational units, such as double helix DNA, nucleosomes, and topological associated domains (TADs) are engaged in transient interactions with various effector proteins. Visualization of the specific chromatin and its effector at the single-molecule level can provide detailed information about its transient interactions (Fig. 1 a). Recently, although fluorescence in situ hybridizations have been extensively used in mapping the spatial organization of chromosomal DNA (9, 10) and the distribution of mRNA (11) in fixed cells, the lack of temporal information makes these methods incapable of revealing dynamic processes in living cells. Sequence-specific labeling of chromatin loci in the living nucleus mainly relies on the fluorescent repressor and operator system (FROS) and programmable DNA-binding proteins. High fluorescence background in the nucleus requires multiple fluorescent probes to be recruited to the labeling site. FROS takes the advantage of the tandem repeats of operator sequences, which are inserted in the genome in either a targeted or random manner (Fig. 2 a ①) (12). The programmable DNA-binding proteins include the zinc-finger nucleases, the transcription-activator-like effector (TALE) (Fig. 2 a ) (13), and the clustered regulatory interspaced short palindromic repeats (CRISPR/Cas9) system (Fig. 2 a ③ and ④) (14, 15, 16). Using the nuclease-deactivated form of these proteins, site-specific labeling is achieved by adjusting the sequence of these proteins (zinc-finger nucleases or TALEs) or single-guide RNA (sgRNA) (CRISPR/Cas9) (Fig. 2 a). Comparisons of these chromosomal DNA-labeling methods have been made in our previous review (17).

Figure 2.

Labeling methods of DNA, mRNA, and proteis for live cell single-molecule imaging. (a) Live cell DNA (chromatin locus) labeling methods are shown. To achieve high signal-to-noise ratio, recruitment of multiple fluorescent proteins to one locus through various approaches is necessary. ① A working mechanism of the FROS system is shown. Operon arrays are inserted into chromosome loci, and fluorescent-protein-fused repressors can be recruited to the sites to accumulate fluorescent signals. ② Schematics of TALEs based chromatin loci imaging are shown. Multiple TALE modules, which are fused with fluorescent proteins, can be used to recognize specific chromatin regions. ③ Schematics of Cas9-based CRISPR imaging of chromosomal regions. The dCas9 molecules fused with fluorescent proteins and multiple sgRNA each targeting one specific region are coexpressed in the same cell to label regions of interest. ④ Schematics of sgRNA-based CRISPR imaging of chromatin loci are shown. The sgRNA are modified by inserting aptamers in tetraloop, loop2, or 3′ terminal. With the assistance of dCas9, the fluorescent-protein-fused RNA aptamers can be recruited to regions of interest in the genome. (b) Schematics of single mRNA molecules imaged in living cells via the MS2 coat protein method are shown. A tandem array of 24 MS2 RNA aptamers is introduced into the 3′-untranslated region of the target mRNA. The interaction between the coat proteins and hairpins can recruit 24 fluorescent proteins to the mRNA, rendering the mRNA brighter than the surrounding background. (c) Single-molecule labeling methods for protein tagging in living cells are shown. The protein of interest (POI) is genetically fused with different tags in its terminus. ① The POI is fused with various kinds of fluorescent proteins (FPs), including conventional FPs, photoactivatable FPs, and photoconvertible FPs. Three FPs can be fused with a POI to increase the brightness. ② The POI is fused with self-labeling tags, including HaloTag, SnapTag, CLIPTag, and TMPTag. The dye-conjugated ligand for the tags can be ligated to the self-labeling tag covalently. In addition, the triple self-labeling tags can be fused with the POI to increase the brightness. ③ The POI is fused with or visualized by an epitope tag. The POI is fused with multiple epitope tags (SunTag/ArrayG) to increase the brightness. The POI can also be fused with epitope tags inserted into GFP-like scaffolds for correlative light and electron microscopy. The FabLEM technique can visualize specific post-transcriptional modifications of the POI. ④ The POI is fused with multiple repeats of the small fragment form split FPs.

By fluorescent labeling of chromatin locus via various approaches, researchers have uncovered the dynamic nature of chromatin fibers on different timescales. On the short timescale, chromatin undergoes locally constrained, rapid, energy-dependent fluctuations caused by chromatin remodeling and transcription (18). These local structural fluctuations of chromatin can transiently expose buried DNA sequences and provide temporary accessibility to proteins. On the longer timescale, multiple effector proteins like transcription factors or epigenetic enzymes can rearrange the existing chromatin structure and establish new chromatin states, resulting in different degrees of chromatin condensation. These changes are always accompanied by the alteration of chromatin locus position and mobility (19).

mRNA

mRNA in the nucleus generally undergo three processes: first, the generation through transcription from a DNA template by RNA polymerase II to make pre-mRNA; second, the maturation of the pre-mRNA through splicing, capping, and polyadenylation; and third, the packaging of the mature mRNA with proteins to make a particle that is competent for export (Fig. 1 b). Studies have demonstrated that these three steps are tightly coupled spatiotemporally instead of occurring sequentially and independently (20). Interrogating the movement of RNA from the sites of synthesis within the nucleus to the cytoplasm is crucial for understanding the dynamic process of gene expression. Answering how transcription and splicing are coupled and how mRNA molecules find their tunnels and then get through the nuclear pore complex in the crowded nuclear environment requires live cell tracking of single mRNA molecules. A number of methods are now available for the labeling of mRNA (21). However, the fluorescent RNA aptamer system remains the only one that can achieve single mRNA imaging in living cells. To tag mRNA, an RNA aptamer binding protein is fused to fluorescent protein and expressed along with a target mRNA inserted with an RNA aptamer in its untranslated region (Fig. 2 b). The most commonly used aptamer systems, including MS2, PP7, and λN, all originate from bacteriophage, which can avoid the cross talk between the endogenous RNA and aptamer (22). To image mRNA at the single-molecule level, as many as 24 copies of the phage hairpin motif should be introduced into the target mRNA, rendering the mRNA considerably more fluorescent than the background and thus enabling its detection.

To address how transcription and splicing are coupled, a complete kinetic model has been constructed by using dual-color mRNA single-molecule imaging in living human cells. It was found that kinetic competition resulted in multiple competing pathways for pre-mRNA splicing (23, 24). Three-dimensional (3D) mRNA single-molecule imaging has also revealed that mRNA molecules travel within pathways intertwined between compact chromatin domains. In addition, time projections of all single-molecule trajectories showed that the diffusion paths are repeatedly used by different messenger ribonucleoproteins. This restricted diffusion of messenger ribonucleoproteins within interchromatin channels is more efficient than diffusion in 3D, as less nucleoplasmic space must be sampled, thus decreasing the diffusion dimensionality (25) (Fig. 1 b). Using a similar single-molecule tracking approach, the kinetics of mRNA transport from the nucleus to cytosol in mammalian cells was spatially and temporally resolved. The RNA exportation through the nuclear pore complex consists of three processes—docking (80 ms), transport (5–20 ms), and release (80 ms)—indicating that translocation through the channel was not the rate-limiting step (5). Additionally, single endogenous mRNA molecules can be labeled and imaged in primary mammalian cells and tissue via a transgenic mouse (26).

Most of the RNA labeled by the aptamer system are expressed from a plasmid, as tagging the multiple repetitive aptamer sequence to the endogenous chromosomal locus is laborious. The newly emerging RNA-targeting techniques, including the spCas9 and c2c2 from the CRISPR system, allow direct imaging of the endogenous mRNA without tagging in live cells (27, 28). However, single-molecule imaging of the endogenous mRNA has not been achieved using these proteins. Even if recruiting multiple spCas9 or c2c2 to the targeting RNA can be realized, the huge molecular weight will increase the burden of the labeled mRNA and thus influence the dynamics of the mRNA export and function. Therefore, finding other smaller proteins that can be programmed to bind specific RNA sequences may represent a future direction of research. Various other kinds of RNA, including long noncoding RNA, micro RNA, and enhancer RNA, exist in the nucleus. Development of new RNA single-molecule labeling methods will further shed light on the underlying mechanisms of how these RNA function.

Proteins

Many intranuclear proteins have a high copy number. Therefore, given the Abbe’s diffraction limit, the target molecule species need to be sparsely labeled to spatially resolve individual molecules. The direct fusion of conventional fluorescent proteins to the target proteins for single-molecule imaging is only possible in some circumstances because quantitatively controlling labeling density using genetic tools is quite challenging. The development of photoactivatable fluorescent proteins (PAFPs) bypassed the problem. Fusing target proteins with PAFPs allows single-particle-tracking photoactivation localization microscopy by optical control of the labeling density (Fig. 2 c ①) (29). PAFPs can also be used to perform time-correlated photoactivation localization microscopy, which allows real-time characterization of protein clustering quantitatively in living cells (30, 31). Protein post-translational modifications play a critical role in the regulation of many biological processes. In addition to monitoring the dynamics of a fused protein, the newly developed Fab-based live endogenous modification (FabLEM) method allows labeling and imaging of post-translational modifications of an untagged protein using antigen-binding fragments (Fig. 2 c ③) (32). With further optimization, FabLEM may be used for single-molecule imaging in the living nucleus.

Unfortunately, compared with organic dyes, fluorescent proteins (FPs) are much less bright, which reduces the localization accuracy and observation duration in single-molecule tracking experiments. An alternative of controlling the labeling density is to use the genetic encoded self-labeling tags, including HaloTag (33), SnapTag, CLIPtag (34), and TMPTag (Fig. 2 c ②) (35). These tags are microbial enzymes that can catalyze the covalent ligation of organic-dye-conjugated substrate, which is cell permeable and thus allows live cell single-molecule imaging. The sparse labeling of self-labeling tags can be fine-tuned by adjusting the fluorescent dye concentration. Several recently developed dyes, such as JF549, JF646, PA-JF549, and PA-JF646, which are compatible with live cell imaging, have demonstrated great performance in intranuclear live cell single-molecule imaging (36, 37, 38). Several advantages over the PAFPs, including brightness, photostablity, and monomerization, have made self-labeling tags superior in single-molecule imaging nowadays.

In addition, precise copy number control of fluorescently labeled molecules (dynamic range of ∼10,000-fold) in living cells can be achieved by simply putting a weak stop codon between the target proteins and the labeling tag to allow translational read-through (39). Unfortunately, most targeting proteins are overexpressed without tags by this method, which may also influence the kinetics of the labeled target proteins (39). In addition, this method can only be applied to add tags at the C-terminus of a protein and is incapable of adding tags at the N-terminus (39).

Although the self-labeling tag is brighter and more stable than the PAFPs, the continuous observation time of a single molecule is still limited to a certain extent. Many approaches are available to amplify the single-molecule fluorescence signal through recruiting multiple copies of fluorophores to target proteins. Such signal amplification systems allow a lower laser power and shorter exposure time to produce the same image quality, thus enabling long-term tracking of single molecules. First, direct fusion of multiple copies of FPs or self-labeling tags to a protein of interest has been used to increase signal intensity (Fig. 2 c ① and ②). So far, three copies of GFP and HaloTag have been realized, but a further increase in the copy number is challenging because of their burdensome size and folding problems. Second, fusion to the protein with small epitope tags arranged into a multimerization scaffold becomes an alternative to the direct fusion of the multiple FPs. These tools contain SunTag (40) and ArrayG (41), which can recruit as many as 24 copies of GFPs (Fig. 2 c ③). In addition, through incorporating numerous copies of peptide epitopes into GFP-like fluorescent protein scaffolds, a “spaghetti monster” was created to amplify fluorescent signal via binding to multiple IgG antibodies (Fig. 2 c ③) (42). Another similar strategy based on background rejection is implemented by bimolecular fluorescence complementation, including mEos3.2 (43), sfGFP (44), and sfCherry (45). Fusing the protein of interest (POI) with the multiple short fragment of FPs (the 11th β-sheet), the remaining of the larger fragment (the 1–10th β-sheet) can bind the array with high affinity (Fig. 2 c ④).

Most of the single-molecule tracking experiments conducted in living cell nuclei focus on transcription factors. Specific binding of transcription factors to chromosomal DNA is the initial step for promoting transcription, as it is an important approach in the regulation of gene expression (Fig. 1 e). The searching dynamics of a number of transcription factors have been investigated in living cell nuclei by single-molecule imaging techniques, including SOX2 (6, 46), TetR (47), p53 (48), GR (49), and TFIIB (50). Although each individual transcription factor (TF) has different kinetics, almost all TFs show a similar searching mechanism. TFs search for specific binding targets buried in the enormous irrelevant DNA sequences via a 3D diffusion dominant mode. In addition, intermittent and nonspecific interaction with some similar sequences occurs frequently. Furthermore, the transient interaction between TFs and chromatin is also modulated in response to various stimuli, including post-translational modifications of TFs (51). Although the binding kinetics of TFs to chromatin has been widely investigated in living cells, the underlying DNA that they bind was not imaged simultaneously. Therefore, it is difficult to distinguish which binding events represent real interactions between TFs and target sites, and it is uncertain which binding events can contribute to transcriptional output. To address these questions, multicolor single-molecule tracking should be performed to image the genomic DNA, TFs, and nascent RNA simultaneously in the future. On the other hand, almost all the transcription initiation processes involve multiple TFs and other proteins cooperating structurally and functionally. Dissecting the cooperative behaviors of multiple interacting TFs also requires single-molecule imaging of multiple species.

The dynamics of scaffold proteins that constitute chromatin has also been studied. PAFP-labeled histone was tracked in living nuclei, indicating that the nucleosome domain fluctuates in the interphase nucleus (52). The heterochromatin-rich regions showed more domains and less movement. With cell differentiation, the domains became more apparent, with reduced dynamics (52). Moreover, the mobility of loop anchor proteins CTCF and cohesin has also been investigated by single-molecule tracking experiments (Fig. 1 c). The authors find that both proteins dynamically interact with insulator elements, but CTCF binds chromatin much more dynamically than cohesin. Thus, the loop anchor complex formed by CTCF and cohesin is a rapidly exchanging “dynamic complex” rather than a typical stable complex, which leads to a “dynamic loop” model (53).

Another kind of protein that has been extensively studied in single-molecule experiments is the various enzymes in the nucleus that use chromatin as a substrate. Telomerase adds repetitive DNA sequences to each chromosome end in actively dividing cells to maintain genome integrity. The authors demonstrate that telomerase also diffuses to search for telomeres in 3D, probing each telomere thousands of times in S-phase but rarely forming a stable association (Fig. 1 i) (54). Chemical covalent modification of histones and DNA modulates higher-order chromatin structures and functions. An explanation of how this modification is established and maintained would require live cell single-molecule imaging. The tracking results suggest a novel hierarchical cooperation mechanism by which histone modifications coordinate PRC1 to target chromatin (Fig. 1 f) (55). Rapid changes in chromatin structure via the action of chromatin-remodeling complexes are thought to regulate transcription and DNA repair. Chromatin-remodeling complexes are ATP-dependent enzymes that target genomic loci to dynamically regulate local chromatin structure (Fig. 1, d and g). Using live-cell single-molecule tracking, the authors have found that PBAF binds chromatin within actively transcribed regions for shorter time periods relative to heterochromatin, and histone acetylation levels increased the frequency of PBAF revisiting to genomic foci (56). Besides, the searching mechanism of the programmable DNA-binding endonuclease Cas9 in living mammalian cells has also been resolved, providing guidelines for the genome editing applications (57).

Imaging modes for intranuclear single-molecule imaging

Single-molecule localization precision is affected by background level, especially in the case of low fluorescence signal (58). Epi-illumination mode excites out-of-focus probes and results in high fluorescence background when imaging the nucleus (Fig. 3 a). Total internal reflection fluorescence microscopy (TIRFM) has been the most popular imaging mode for live cell single-molecule imaging. However, as the penetration depth of the evanescent field in TIRFM is only ∼200 nm, TIRFM is not capable of imaging the nucleus, which is several micrometers from the bottom of the surface. Although confocal illumination mode could lower the out-of-focus background, its temporal resolution is relatively low and also causes fast photobleaching. In 2008, Tokunaga et al. presented highly inclined and laminated optical sheet (HILO), an imaging mode using an inclined beam and a focal-plan-conjugated iris (59). The HILO mode can lower background level significantly but at the cost of largely reducing the field of view (Fig. 3 b).

Figure 3.

Schematics of the illumination modes for intranuclear single-molecule imaging. (a) Wide-field microscopy is shown. (b) Highly inclined and laminated optical sheet (HILO) microscopy is shown. (c) Lattice light-sheet microscopy is shown. (d) Reflected light-sheet microscopy is shown. (e) Light-sheet Bayesian microscopy is shown. NA, numerical aperture; SBR, signal-to-background ratio; FOV, field of view. To see this figure in color, go online.

Almost at the same time, light-sheet illumination began to attract attention. Using a layer of light to illuminate the sample at the focal plane of the detection objective, light-sheet illumination mode is equipped with the merits of simultaneous low background, low light toxicity, and rapid imaging (60). However, although light-sheet mode has demonstrated great performance for imaging tissue samples and model organisms, its application to single-cell imaging has been quite challenging because of the geometric hindrance between objectives (Fig. 3 c). In 2013, Xie’s group used an AFM cantilever to reflect the light sheet at a 90° angle to cross the nucleus of a cell near the cantilever (49). This clever configuration bypasses the geometric hindrance by positioning the illumination and detection objectives so that they face each other, thus allowing use of objectives with high numerical aperture for higher optical resolution and sensitivity (Fig. 3 d). With a ∼1 μm thin beam waist, this method has an enhancement in signal-to-background ratio (SBR) of 2.5- to 5-fold compared to HILO and is able to track the movement of a single fluorescent molecule within the nucleus. However, the configuration and operation of the system is rather complicated. Later, a single-lens light-sheet microscope (61) was developed by microfabricating a chip with 45° reflective surfaces next to the sample reservoir. This design greatly simplifies the configuration and increases the throughput to achieve single-cell 3D imaging and single-molecule super-resolution imaging. However, the integrated design of the sample and mirror makes the illumination plane of the light sheet and the focal plane of the objective lens coupled, which complicates the control for z-sectioning. Although both designs could enhance SBR greatly with thin beam waist (∼1 μm), the Rayleigh length (half of the field of view) is limited (∼6 μm) because of the intrinsic properties of a Gaussian beam. In 2014, Betzig’s team developed a lattice light sheet based on multiple Bessel beams (62). It achieves high-speed, high-resolution imaging of samples with various dimensions (Fig. 3 c). Importantly, Bessel beams greatly extend the field of view to ∼50 μm. However, the design of the lattice light sheet is relatively complicated, and the geometric hindrance also limits the numerical aperture of the objectives. To tackle the geometric hindrance problem, several clever solutions have been developed (Fig. 3 e) (63, 64, 65, 66).

Further developments of light-sheet illumination for single-cell fluorescence detection would aim at a system that balances the imaging performance (SBR, scanning speed, etc.), throughput as well as the ease and stability of operation, which are important properties of light-sheet imaging mode. Additionally, besides the fluorescence background, the light distortion caused by the uneven refractive index of the cytoplasm and organelles also affects the localization accuracy. Adoption of adaptive optics into light-sheet imaging would further improve the performance of intranucleus single-molecule imaging (67).

Potential pitfalls

Regarding single-molecule fluorescence imaging in the nucleus, attention should be paid to the artifactual effects caused by intrusive labeling methods and harsh imaging conditions.

First, labeling of DNA, RNA, and proteins might interfere with the normal physiological functions of the tagged molecules. For instance, labeling tags to DNA might disturb the local chromatin structure and functions. Furthermore, binding of endonucleases (TALE, Cas9) or repressors to chromatin may interfere with gene transcription. For mRNA labeling, recruitment of multiple aptamer coat proteins to one single mRNA would increase the size of mRNA significantly and thus disturb the transport kinetics. As for the protein labeling, the potential pitfalls might originate from the difference between endogenously tagged and exogenously tagged molecules. Actually, the majority of the intranuclear single-molecule imaging has been performed on exogenous proteins, which is likely to introduce artifactual effects because of overexpression of target proteins. Accumulating evidence has suggested that it is essential to image target molecules at their physiologically relevant levels (53, 54, 68). Fortunately, genetic engineering technologies such as the CRISPR/Cas9 system can now efficiently insert labeling modules into the endogenous genomic loci of the target molecule, making the fusion molecule expressed at a physiological level (53, 54, 68).

Secondly, the unideal performance of fluorescent probes and phototoxicity could lead to imprecise or even incorrect interpretation of single-molecule data. For instance, the calculation of protein residence time depends on the photostability of the labeling fluorophore (69). There is still a large gap between the labeling performance required for intranuclear imaging and the existing probes. More bright, stable, and less cumbersome probes and labeling strategies are needed for long-term, continuous, fast, and low phototoxic intranuclear single-molecule imaging. With further optimization, unconventional fluorophores, such as nanobodies (70), quantum dots (71), conjugated semiconductor polymer dots (72), and carbon nanodots (73) may become excellent probes for live cell single-molecule imaging.

Conclusions and perspectives

As an important technique, single-molecule imaging in living cells has provided essential information about the kinetics, dynamics, and architecture of the molecules or even molecular machines that operate in living cells in the past several decades. Single-molecule tracking will cooperate with other disciplines, including biochemistry, genetics, and genomics, to help us deepen our understanding of life in a more comprehensive and quantitative manner.

To understand the complicated behaviors of some cellular processes, single-molecule imaging of multiple components simultaneously is highly demanded. For instance, robust multicolor single-molecule imaging assays are needed to resolve the dynamics of molecular machinery involved in transcription (Fig. 1 e), DNA repair (Fig. 1 h), and DNA replication (Fig. 1 j).

Regarding the imaging approaches for intranuclear single-molecule tracking, up until now, most experiments have been carried out on cultured cell lines ex vivo, making it hard to realize single-molecule imaging in living organisms due to the scattering of light. Culturing and acclimation of tissue cells in the dish may cause dramatic alteration of nuclear rigidity and chromosome organization, leading to misinterpretation of the relationship between the structure and functions. Performing single-molecule imaging in vivo in whole living organisms is a key step toward understanding the links between the dynamics of single-molecules and cell behavior at the tissue or organism level. Accordingly, large volume imaging with high imaging speed is crucial and will become an important aim for further developments in light sheet microscopy. Further, the combination of adaptive optics, light sheet microscopy, and bright genetically encoded labels will make single-molecule imaging go even deeper.

Editor: Amy Palmer.