Single-molecule fluorescence imaging of DNA maintenance protein binding dynamics and activities on extended DNA

Abstract

Defining the molecular mechanisms by which genome maintenance proteins dynamically associate with and process DNA is essential to understand the potential avenues by which these stabilizing mechanisms are disrupted. Single-molecule fluorescence imaging (SMFI) of protein dynamics on extended DNA has greatly expanded our ability to accomplish this, as it captures singular biomolecular interactions – in all their complexity and diversity – without relying on ensemble-averaging of bulk protein activity as most traditional biochemical techniques must do. In this review, we discuss how SMFI studies with extended DNA have substantially contributed to genome stability research over the past two years.

Introduction

Genome instability induced by endogenous and exogenous DNA damaging agents is known to promote cellular senescence, cell death, and a variety of human diseases. An army of enzymes counteracts genome instability through tightly regulated and highly specific DNA repair mechanisms that occur under different DNA topology and chromatin contexts. Traditional biomolecular approaches characterize these mechanisms through ensemble-averaging, which fails to capture the heterogeneity of often transient protein activity. To capture the full spectrum of protein activity and answer essential questions regarding protein dynamics, individual protein-DNA interactions must be studied.

Here, we review how single-molecule fluorescence imaging (SMFI) reveals the binding dynamics and activity of proteins on extended DNA with high spatial and temporal resolutions. We highlight how lesion search, recognition, and processing are accomplished by DNA repair proteins. Common quantitative parameters collected from SMFI of proteins on stretched DNA are summarized in Figure 1. We conclude with a brief discussion of the challenges and recent advances in the SMFI field in both in vitro and live cell models.

Figure 1.

Quantifiable parameters from SMFI of proteins on extended DNA [1,2].

DNA curtains and surface tethered DNA characterize DNA repair and DNA topology change

DNA Curtains are a versatile approach for studying protein dynamics on DNA that can be tethered at one or both ends in a flow cell. Importantly, DNA tethered to a lipid bilayer at one end through biotin-streptavidin linkage is mobile. Hydrodynamic forces align and extend DNA at metallic barriers, forming DNA Curtains with defined sequence orientation and positions (Figure 2a) [3]. Additional pedestals and DNA linkage can be added to allow both ends of the DNA to be tethered. Fluorescently labeled protein is introduced, allowing protein binding dynamics along hundreds of DNA molecules to be visualized in a single flow cell using Total Internal Reflection Fluorescence Microscopy (TIRFM). By directing the laser at the critical angle, TIRFM illuminates only a few nanometers of sample depth, restricting the volume of excited fluorophores and enhancing the signal-to-noise ratio over epifluorescence imaging (Figure 2b) [3,4].

Figure 2.

DNA Curtains reveal DNA unwinding by BLM helicase promoted by DNA2 and RPA. (a) Schematic of the DNA Curtain assay. Single-tethered dsDNA containing position-specific damage or structures of interest is aligned by buffer flow, then bound with fluorescently labeled proteins. (b) Side view of DNA Curtain flow cell with sample volume only partially illuminated by laser (blue) in TIRFM. (c) TIRFM image showing GFP-BLM (green) binds dsDNA (~50 kbp) at 3^′ overhang of free end. B: Barrier. E: DNA end. (d) Binding position distribution of GFP-BLM on the DNA. (e) Kymographs showing DNA end resection by GFP-BLM, DNA2 (unlabeled), and mCherry-RPA. White arrows denote spontaneous release of the 3^′ ssDNA end. (f) End resection velocity (left) and processivity (right) distributions of GFP-BLM and DNA2 plus or minus RPA. The center bar provides the mean and error bars show standard deviation. Adapted from Ref. [5].

Several recent applications of DNA Curtains in the DNA double-strand break (DSB) repair field highlight the advantages of this assay. During S and G2 phase, DSBs are primarily resolved through the high-fidelity homologous recombination repair (HRR) pathway. HRR utilizes a homologous DNA duplex as a template for DNA synthesis at broken DNA ends. This process is initiated through endonuclease-mediated DNA end resection and progressive unwinding by DNA helicases. Several recent studies have designed lambda DNA curtains containing a 3′ overhang at their free ends, modeling early-stage end resection. The recruitment of fluorescently labeled protein to the 3′ overhang can be studied (Figure 2c and d), plus DNA unwinding, which is evidenced by directional movement of protein along DNA curtains and progressive accumulation of the ssDNA-binding replication protein A (RPA) behind it (Figure 2e) [5–7]. Using this approach, the rate of DNA unwinding by the essential HRR factor, Bloom syndrome helicase (BLM), was measured [5]. By co-incubating BLM with known cofactors in HRR initiation, DNA2 (Figure 2f) [5] and MRE11-RAD50-NBS1 (MRN) [6] endonucleases were shown to enhance BLM translocation velocity and DNA unwinding rate. Additionally, because DNA Curtains can anchor a single DNA end, the free end dynamics can be studied throughout resection. Labeling the 3′ overhang with ATTO565 revealed BLM and DNA2 physically pull the 3’ overhang with them during DNA unwinding, arranging the ssDNA overhang into a loop, protecting it from degradation [5].

A similar approach to DNA Curtains tethers biotinylated DNA at both ends to a passivated flow cell surface as opposed to a lipid bilayer. The resulting U-shaped DNA under buffer flow is ideal for characterizing DNA loop extrusion and supercoiling by the Structural Maintenance of Chromosomes (SMC) protein family that facilitates DNA repair. Using U-shaped DNA and dual-color protein labeling, orientation-specific DNA binding by CCCTC-binding factor (CTCF) was shown to directly control cohesin-mediated DNA loop extrusion, evidenced by halting the emergence of DNA puncta (loop) that expand in size over time [8].

Addition of the intercalating SYTOX Orange (SxO) dye induces directional supercoiling in a concentration-dependent manner, and this method has been used to study SMC-mediated loop extrusion on supercoiled DNA. In the presence of ATP, condensin stabilized supercoiling-induced protrusions known as plectonemes by extruding them into a single supercoiled DNA loop [9]. Similar findings were observed by the evolutionarily related yet comparatively understudied Smc5/6 complex. Dual-color imaging revealed Smc5/6 binds and initiates DNA loop extrusion at the tip of positively supercoiled plectonemes before translocating and stabilizing at the loop bases [10,11]. Photobleaching indicates Smc5/6 primarily accomplishes this as a dimer, yet translocates along dsDNA as a monomer [11]. These studies establish Smc5/6 as an integral mediator of genome organization differentially regulated by its oligomeric state and DNA supercoiling.

Optical tweezers modulate DNA tension to probe protein binding dynamics

The combination of optical tweezers, microfluidics, and confocal microscopy leads to a versatile imaging platform allowing for sequential incubation of DNA with different proteins and modulation of DNA tension. Dual-trap optical tweezers stabilize two microsphere beads in space using the gradient forces of a highly focused laser beam [12,13]. Biotinylated DNA containing position-specific structures of interest is strung between the microspheres under regulatable tension (0.02–250 pN) [14], enabling protein search dynamics and lesion recognition to be studied (Figure 3a). The dual-trap method also enables microsphere-DNA-microsphere construct translocation between chamber channels formed by laminar flow for complete buffer exchange and tight bioreaction control, such as stepwise protein introduction. Thus, optical tweezers combined with microfluidics can explore similar protein activities as DNA Curtains, such as DNA unwinding [15], but with tighter control of protein introduction, removal, and DNA tension.

Figure 3.

Smc5/6 preferentially localizes to ssDNA-dsDNA junctions along DNA extended between optically trapped beads. (a) Schematic of dual-trap optical tweezers design. dsDNA containing a position specific nick is stretched under constant tension between two optically trapped microsphere beads in the presence of the labeled protein. Increased tension is applied to the dsDNA by increasing the distance between the beads, causing dsDNA to peel at the nick site. Resulting DNA under increased force contains position specific ssDNA and ds/ss DNA junction regions. (b) Representative kymograph of Cy3-Smc5/6 (green) and LD650-RPA (red) proteins on DNA tether under low (LF) and high force (HF). The red laser was temporarily switched off to confirm Smc5/6 signals. dsDNA (blue), ssDNA (red), and junction DNA (green) regions are given at the right. (c) Smc5/6 fluorescent signals on different types of DNA under HF. Bar heights represent group mean and error bars give standard deviation. P-value determined from two-tailed unpaired t-tests with Welch’s correction. Adapted from Ref. [22].

An ongoing question in DSB HRR is how RAD51, an essential mediator of homology search and strand invasion, successfully displaces RPA from unwound ssDNA. Recent optical tweezers studies have explored this question by co-incubating fluorescently labeled RAD51 or RPA with putative protein interactors on optically trapped ssDNA or dsDNA containing ssDNA gaps (‘gap DNA’) [16–18]. RPA displacement is measured as the loss of fluorescently labeled RPA from ssDNA over time. Using this approach, the RPA-interacting HELQ helicase [16] and SWSAP1-SWS1 complex [17] were shown to strip RPA and enhance its diffusivity on ssDNA, respectively, based on mean square displacement (MSD) analysis (Figure 1). Dual-color labeling can also be combined with optical tweezers to study protein-protein interactions. In this way, BRCA2 and RAD51 were shown to collaboratively bind and diffuse along dsDNA through a sliding mechanism [19]. While not covered in this short review, recent DNA Curtain studies have similarly used dual-color imaging and fluorescence intensity measurements to explore how HRR repair protein interactions and their oligomeric states affect protein processivity on HRR intermediates [20].

Several recent studies take advantage of optical tweezer’s ability to regulate DNA tension as a means of altering DNA structure. The Rueda lab used a unique DNA substrate containing a centrally located double hairpin modeling the HRR Holliday junction intermediate. The junction can be reversibly unfolded at higher tensions produced by increasing microsphere separation. This enabled the diffusive dynamics, binding specificity, and cleavage kinetics of endonuclease 1 at the intact versus dissolved junction to be studied in real-time [21]. DNA tension modulation has also been used to induce dsDNA melting, creating position-specific single-stranded gaps and dsDNA-ssDNA junctions (Figure 3a) [22]. Using this substrate, the Smc5/6 complex was shown to preferentially localize to dsDNA-ssDNA junctions (Figure 3b and c) [22]. Additionally, increased tension on nucleosome-studded DNA can promote nucleosome unwrapping [23,24]. This substrate recapitulates the chromatin environment proteins would encounter in the nucleus and has been used in both optical tweezers [23,25] and DNA Curtain [26,27] experiments, revealing the complexity of nucleosome-histone chaperone interactions: nucleosomes serve a barrier function and suppress chaperone diffusion [25], while chaperones also can enhance the loading of histone proteins [26] as well as the eviction of partially unwrapped nucleosomes from DNA [23,24]. Lastly, protein-induced DNA tension modulation can also be measured based on changes to the distance between optically trapped beads. The Alberti lab took advantage of this and recently demonstrated the essential DNA damage sensor PARP1 compacts nicked DNA, as evidenced by reduced bead distance in the presence of PARP1 [28]. This exciting publication showed DNA compaction appears to promote the formation of condensates of damaged DNA, PARP1, and PAR chains, to which downstream repair proteins localize to promote repair [28].

DNA tightropes capture protein binding modes and diffusivity

The DNA tightrope assay offers a simple and cost-effective approach to study protein dynamics along DNA containing position-specific damage. Lambda or ligated linear DNA (~20–50 kbp) is stretched between poly-L-lysine coated silica beads in a flow chamber by hydrodynamic buffer flow [29]. By tandemly ligating linear DNA molecules containing position-specific sequences or structures, the structure of interest occurs at regularly spaced intervals along DNA tightropes. The binding dynamics of fluorescently labeled proteins can then be monitored with Highly Inclined and Laminated Optical Sheet (HILO) microscopy (Figure 4a).

Figure 4.

UV-DDB colocalizes with SMUG1 glycosylase and enhances SMUG1 diffusivity on DNA tightropes containing abasic sites. (a) Schematic of the DNA Tightrope assay. Long DNA substrates with defined abasic sites every 2 kbp are suspended between silica beads and incubated with the labeled protein. (b) Image of co-localized (yellow, marked by white arrow) quantum dot (Qdot)-labeled SMUG1 (green) and UV-DDB on the abasic tightrope suspended between beads. (c) Kymographs of Qdot-labeled SMUG1 (top), UV-DDB (middle), and merge (bottom) show they co-localize. (d) Stacked bar graph showing the fraction of motile versus stationary and persistent versus dissociated SMUG1 in the absence (left) or presence (right) of UV-DDB on abasic DNA during 300 s observation. (e) Box and whisker plots (10–90 percentile) of diffusion coefficient (log₁₀D, left) and Anomalous diffusion exponent (α, right) for SMUG1 alone and SMUG1 with UV-DDB on abasic DNA. (f) Effects of UV-DDB on the binding lifetimes of SMUG1-DNA complexes. Data plotted as the mean ± SEM from 3 independent experiments. Adapted from Ref. [30].

The retention of glycosylases at abasic sites resulting from cleavage in base excision repair (BER) suggests glycosylase liberation may depend on additional protein factors. Recent DNA Tightrope studies from the Van Houten group have identified the DNA damage sensor UV-DDB as an important facilitator of BER [30]. UV-DDB was shown to colocalize with AAG [30] and another glycosylase, SMUG1 [31] (Figure 4b and c), increasing their diffusivity and dissociation frequencies (Figure 4d–f). This suggests UV-DDB facilitates both lesion search by glycosylases and their subsequent release from abasic sites. Similar optical tweezers work has helped further characterize the common sliding and hopping linear diffusion modes of other glycosylases, including TDG glycosylase [32].

Recent work from our lab characterized the binding dynamics of Methyl Binding Domain (MBD) proteins 2 and 3 [33] using DNA Tightropes. MBD2 and MBD3 are mutually exclusive subunits of the Nucleosome Remodeling and Deacetylase (NuRD) complex, and previous in vitro studies suggested they differentially bind methylated (m)CpG dinucleotides. Using physiologically representative DNA tightropes containing CpG and mCpG islands, we showed MBD2 binds stably to mCpG islands, while MBD3 diffuses independently of DNA methylation status [33]. This result provided the first physiologically representative in vitro evidence that several populations of NuRD complexes likely exist that differ in their diffusivity along CpG and mCpG islands depending on the MBD protein they contain.

Future perspectives: SMFI challenges, recent advances, and cellular approaches

The exciting studies mentioned above highlight the distinct advantage of SMFI in capturing and quantifying DNA repair processes that ensemble-averaging approaches would overlook. While signal-to-noise ratio optimization, photobleaching, and slowed protein dynamics due to labeling remain ongoing challenges [34], these issues are continuously mitigated by advances to SMFI technologies. The widespread adoption of TIRF and HILO microscopies maximize signal-to-noise ratios by exciting only a fraction of the sample volume. Moreover, brighter and more photostable quantum dots and dyes are constantly being developed to increase temporal resolution [35]. Additionally, research groups are tirelessly working to identify, quantify, and mitigate the risk of potential artefacts in recent technologies, such as the need for tight temperature control in optical tweezers experiments [36].

As SMFI of protein dynamics on extended DNA has become more widely adopted, fascinating applications of these approaches expand our ability to study genome maintenance mechanisms. An ever-growing number of unique nucleic acid substrates elegantly model DNA replication intermediates [37], epigenetic hallmarks [33], and compacted chromatin states [23]. Optical tweezers can transiently induce these various structures and states through mechanical manipulation and can induce both directional supercoiling [38] and various DNA repair intermediates [8,21,22]. Additionally, novel approaches in protein preparation, such as Single-Molecule Analysis of DNA-binding proteins from Nuclear Extracts (SMADNE), enable proteins to be purified in their native states with interactors that may influence their activities, preserving their more physiologically representative states [39–41]. The high spatial and temporal resolutions with which these diverse substrates and proteins can be studied make SMFI a powerful and often high-throughput approach for characterizing molecular mechanisms of DNA damage response and with that, identifying potential drug candidates that disrupt these mechanisms in cancer models [42,43].

While not discussed in this short review, live cell fluorescence imaging further advances understanding of protein dynamics on DNA by preserving the nuclear environment. The induction of cellular toxicity and specific forms of DNA damage at discrete genomic loci can be accomplished through an ever-expanding array of mechanisms, and the subsequent recruitment of fluorescently labeled proteins of interest to those damage sites can then be studied in real time. We encourage readers to explore recent publications utilizing or reviewing modern cellular approaches to characterize protein dynamics in DNA repair and replication [16,44–55]. As both healthy and diseased cell lines can be used, SMFI in live cells combined with complex mechanistic evidence from extended DNA studies have profound implications for understanding how protein activity contributes to human diseases and spurring the development of novel therapeutics.

Data availability

No data was used for the research described in the article.