Application of STED imaging for chromatin studies

Chromatin is the information center of a cell. It comprises proteins and nucleic acids that form a highly complex and dynamic structure within the nucleus. Its multiple organization levels span from micrometre to nanometre scale. For many years, the lower levels of chromatin organization have been beyond the resolution limit of fluorescent microscopy, thus impeding research on nucleus architecture, transcription, translation and DNA repair. Recent development in super-resolution fluorescence microscopy enables us to more easily observe objects at the nanometre scale and allows the study of complex cellular structures at unprecedented detail. This review focuses on the application of stimulated emission depletion microscopy for imaging two main components of the chromatin-DNA and the proteins interacting with it.


Introduction
The ability to investigate chromatin organization and dynamics is a key to the understanding of how genetic mat erial is maintained, decoded through transcription and trans mitted to daughter cells. Numerous chromatin components have been characterized by classical biochemistry and struc tural methods, providing invaluable insights on the properties of enzymes and structural proteins, and their interactions with DNA sequences [1]. However, this is insufficient for deci phering molecular mechanisms on the whole cell level, as many crucial functions (such as DNA replication, repair and transcription) are not performed by single proteins or well defined complexes, but rely on the formation of very large, heterogeneous and dynamic ensembles, involving multiple copies of different proteins and longrange DNA interactions.
In some cases, structural details of such ensembles can be gained by electron microscopy, but dynamics of these pro cesses remain mainly accessible through light microscopy [2]. Fluorescent microscopy is the most flexible method for studying dynamic processes inside the cell, but many chro matin functions are regulated by looping, bending, super coiling and other conformational changes that occur at length scales below the light diffraction limit of ~250 nm (figure 1). For many years, such structures and processes could only be visualized by indirect methods, such as FRET [3,4]. This has changed with the introduction of superresolution fluores cence microscopy techniques that now allow to see the real size of many such structures [5,6].
There are three major microscopy methods that overcome the light diffraction barrier: stimulated emission depletion (STED) microscopy [7], structured illumination microscopy (SIM) [8] and single molecule localization (SML) microscopy [9,10]. Each of them has its own advantages, requires dif ferent equipment and poses unique requirements on sample preparation and experiment design. These points have been reviewed and the methods have been compared elsewhere [6], and therefore they will not be extensively discussed here. The present review will specifically focus on one of the techniques-STED microscopy-and its impact on chro matin research. In contrast to SIM and SML, imaging by STED microscopy has an inherent advantage of immediately providing a hardwarebased image with improved resolu tion without the need for postprocessing procedures, which makes it less prone to incorrect data processing and interpreta tion. With the example of STED microscopy, we will discuss how the superresolution imaging is transforming chromatin research. We will begin by describing the fluorescent dyes that are used for labeling DNA and chromatin components. Then we will provide illustrative examples, where improved resolution offered by STED microscopy was instrumental to obtain fundamentally novel information or explain previously elusive phenomena.

Fluorophores used in STED microscopy
Chromatin components can be visualized in different ways: total DNA can be stained with sequence unspecific dyes, spe cific DNA sequences can be hybridized with fluorescently labeled complementary probes; chromatin proteins can be visualized with fluorescently labeled antibodies or as fusions with fluorescent proteins. STED microscopy poses relatively few specific requirements on a fluorophore itself: it must be photostable in order to withstand much higher laser powers than during conventional confocal imaging, and it must show efficient stimulated emission with minimal reexcitation at the wavelength of the depletion laser. The overview of fluo rophores used for chromatin imaging by STED is provided in table 1. There are no strict rules to predict compatibility of the dye with STED microscopy. The most often used STED compatible fluorophores are derivatives of rhodamines. In contrast, cyanines show limited STEDcompatibility, pro viding little gain in resolution, with the exception for DNA intercalating dyes based on cyanine scaffold (see below). Autofluorescent proteins can also be used for STED imaging, but their lower photostability compared to the organic dyes results in less acquired frames [11]. When considering fluo rescent protein tags, the proteins operating at red or farred spectral region should be the first choice, as imaging at these wavelengths induces the least phototoxicity [12][13][14].
In contrast to fluorescent proteins, interaction of organic dyes with DNA is fluorogenic -usually, DNAspecific dyes are nonfluorescent or weakly fluorescent when free in solu tion and their fluorescence dramatically increases after revers ible binding to DNA. This enhances the contrast of the images and reduces bleaching via fluorophore exchange with a pool of dye present in the medium [15].
Other requirements for small molecule fluorescent probes used for STED imaging are the same as in conventional imaging techniques: good solubility in aqueous solutions, sequence or structure specificity, high binding affinity and, ideally, turnon fluorescence upon target binding. In case of living cell imaging, low toxicity, good cell permeability and low offtargeting are essential.

In vitro DNA imaging
Fluorescent probes targeting DNA have become indispens able tools in cell biology, molecular biology, chemical biology and drug discovery fields. Taking into account the increasing number of new fluorescent probes developed each year, the availability of fluorescent DNA stains is also steadily increasing [16,17].
First successful STED imaging of DNA in vitro was demon strated by Persson et al [18] by employing the standard DNA bisintercalating dye YOYO-1 (figure 2(a) and figure 3(a)), which belongs to dimeric cyanine class of dyes. YOYO-1 shows a 500fold fluorescence enhancement upon binding to DNA. The authors demonstrated a six fold improvement in resolution over standard confocal microscopy by exciting   the dye at 470 nm and using 568 nm emission depletion laser. The increased resolution allowed to resolve the occurrence of bends and kinks along phage λ DNA. The kinks were attrib uted to the sequence specific binding of the YOYO-1 dye and to the binding of proteins or other small molecules. Moreover, it was demonstrated that a fourfold improvement of reso lution over confocal imaging could be achieved by using a 647 nm depletion laser where the YOYO-1 emission is just 3% of its maxima, thus demonstrating that the depletion laser can be applied over a wide wavelength range. YOYO-1 was also used for the accurate determination of fragment lengths produced by digesting λ DNA with the restriction enzymes R.BstEII or R.HindIII. For linear DNA measurement, DNA fragments can be stretched by capillary force, force flow or dynamic molecular combing (DMC). Kim et al has combined DMC with STED microscopy, due to its simplicity and superb reproducibility [19]. Immersing silanized coverslip in a Teflon chamber with DNA fragments followed by pulling at a constant speed of 200 µm s −1 ensured binding of one DNA end to a coverslip and rupturefree DNA stretching. Subsequent continuouswave STED imaging led to the determination of fragment lengths ranging from 117 to 23 130 bp. Importantly, fragments of similar lengths, such as 117 and 224 bp or 1264 and 1371 bp, could be distinguished. The authors proposed that the measurement precision is suf ficient for their method to have potential for the diagnosis of the genetic disorders caused by copy number variation, such as trinucleotide repeat disorder [19].

In vivo DNA imaging
Traditionally, dyes excited by UV light (DAPI and Hoechst), have been used for staining DNA in the nucleus of living and fixed cells [20,21]. These dyes are not suitable for Reprinted with permission from [27]. Copyright (2019) American Chemical Society. (c) Live cells stained with 500 µM 1 4+ and imaged by: HyVolutionconfocal imaging combined with deconvolution algorithm (dCLSM), 3DSTED, and STED. Panels (i)-(iii) show magnified images at positions marked by boxes. Fluorescence intensity profiles are generated from the lines shown in (i)-(iii). Reprinted with permission from [28]. Copyright (2017) American Chemical Society. superresolution microscopy and imaging by blue light results in significant phototoxicity [22,23]. In 2015, Lukinavičius et al introduced a farred DNA stain 6-SiR-Hoechst (figure 2(b)) (aka SiRDNA) [24] for livecell imaging. The probe was designed by modular Hoechst tagging strategy [25] and consists of STED compatible farred dye 6carboxyl silicon rhodamine [26] conjugated to Hoechst as DNA binding moiety by a short aliphatic linker [17]. 6-SiR-Hoechst is a highly photostable, fluorogenic dye showing approximately 50fold fluorescence increase upon binding to DNA and its fluores cence can be efficiently switched OFF by 775 nm STED laser. The proposed mechanism of fluorogenicity is different from the parent Hoechst 33342, and at least partially results from a shift of the equilibrium of siliconrhodamine from the nonflu orescent spirolactone to the fluorescent zwitterion form. The relationship between probe structure and properties was fur ther investigated in a recent study by Bucevičius et al which demonstrated that a minimal change in the attachment point of the dye has a dramatic effect on the properties of the final probe [27]. Specifically, it was revealed that 6carboxylrhoda mine dye containing probes, such as 6-SiR-Hoechst, exhibit dualmode binding to DNA and form a dimmer complex at high DNA concentrations, whilst 5carboxylrhodamine dye containing probes exclusively exhibit singlemode binding to DNA, which results in lower cytotoxicity and up to 10fold brighter nuclear staining. This feature was demonstrated with tetramethylrhodamine (5-TMR-Hoechst), two carbopyro nines (5-580CP-Hoechst and 5-610CP-Hoechst), silicon rhodamine (5-SiR-Hoechst) and germaniumrhodamine (5-GeR-Hoechst) (figures 2(d) and 3(b)). Employing these highly efficient probes allowed to estimate the heterochro matin exclusion zones (~165 nm) in nuclear pore region in living cells and intact chicken erythrocytes. Moreover, a wide colour selection of probes allowed two colour STED imaging of intact nucleated erythrocytes -a system where imaging is highly complicated by the haemoglobin absorption.
Another infrared DNA probe of similar design, JF646-Hoechst (figure 2(c)), consisting of JF646 fluorophore conju gated to Hoechst by a C4PEG2 linker, allowed visualization of small DNA loops in the mitotic chromosomes of fixed HeLa cells by STED microscopy [15].
In 2016, Byrne et al used ruthenium (II) metal complex luminophore as a fluorescent DNA probe for STED micros copy [29]. Transition metal complex luminophores, such as Ru (II) polypyridyl complexes, bear a number of useful properties for fluorescence microscopy: large Stokes shift, long lived emissive states usually arising from phosphores cent emission and good photostability. However, the inability to cross membranes has hampered the use of metalorganic complexes for superresolution imaging of living cells earlier.
Byrne et al have overcome this issue by conjugating Ru(II) polypyridyl complexes to cell penetrating peptides such as nuclear localisation sequences (NLS). In particular, previ ously known DNA binding metal complex [Ru(dppz)(bpy) (bpyArCOOH)] 2+ was conjugated to NLS (VQRKRQKLMP) of the transcription factor NFκB by peptide coupling to yield Ru-NLS probe (figure 2(e)). The emission of this complex switches OFF in water, but it starts to emit strongly upon binding to DNA due to the presence of the phenazine ligand. The stained nuclei of living HeLa cells were imaged by exci tation at 488 nm and STED laser at 660 nm, which overlaps with the broad ruthenium 3 MLCT emission band at 611 nm. The turnon property of the probe combined with excellent STED performance allowed the authors to identify cell cycle phases through a series of STED images of a labeled HeLa cell. Moreover, it was demonstrated that long lived phospho rescent emission makes these probes particularly amenable for timegated STED applications. Timegating the detec tion in STED microscopy serves as a spatial filter, reducing background signal and the pointspread function of the inter rogation volume to improve image resolution. A time delay is introduced after the excitation to maximize the number of fluorophores in the depletion donut that are switchedoff. Moreover, it is also a very useful approach in resolving two dyes with different excited state lifetimes even if the dyes emit light of similar wavelengths. Inspired by these results, Sreedharan et al successfully applied a dinuclear Ru (II)based complex 1 4+ (figure 2(f) and figure 3(c)) in STED microscopy [28]. Surprisingly, no cell penetrating peptides were needed as the probe was cell permeable. The complex also demonstrated fluorescence turnON upon DNA binding. Moreover, a selec tive staining of mitochondria or nucleus could be obtained by adjusting the concentration of the complexconcentrations below 5 µM stained mitochondria and at concentrations above 18 µM, mitochondrial emission was not discerned due to the more intense emission of complex bound to nuclear DNA. The biggest advantage of this STED compatible DNA probe is that it can be easily synthesised on a gram scale and no con jugation to targeting moieties is required. Because of its high photostability, this complex is compatible with 3DSTED imaging, and provides high resolution along all three axes ( figure 3(c)).
Recently, a thiophene based Zn(II) complex LC (figure 2(g)), which can specifically target live cell mitochondrial DNA by intercalation, has been described [30]. The fluo rescence of this probe turnson upon DNA binding and it is excited by 405 nm (one photon absorption) or 840 nm (two photon absorption) and is compatible with 595 nm STED laser. The authors reported that this probe exhibits low cytotox icity at low micromolar concentration range over a period of 24 h. STED imaging revealed ultradetailed structures corre sponding to mtDNA distributed within mitochondrial inner matrix with an approximate 40-70 nm separation between two stripes. Moreover, the authors claim that the photophysical properties of the LC probe can be tuned by changing a bridging unit-thiophene-and by attaching different electrondonors, thus resulting in a tuneable platform for the design of prom ising bioimaging probes for superresolution imaging.

Imaging of proteins interacting with DNA
Proteins interacting with DNA are particularly interesting objects because these complexes play a central role in gene regulation, repair and packing. Heller et al resolved single proteins interacting with DNA at high concentration using a setup comprised of: (1) optical tweezers for DNA conformation and tension control, (2) beamscanning con focal fluorescence microscope for protein imaging on the stretched DNA and (3) 1D STED nanoscopy to distinguish individual labeled proteins at high protein density on DNA. The experiments were performed on λ DNA fused to 2 poly styrene spheres stretched with optical tweezers. Full width at half maximum (FWHM) of 50 nm was obtained when imaging Atto 647N labeled restriction enzymes EcoRV and BsoBI bound to stretched DNA. The experiments revealed that at least 5 pN stretching force was required to obtain effective resolution. The combined setup was also used for investigating DNAbinding dynamics of human mito chondrial transcription factor (TFAM) labeled with Atto 647N. Time traces of single TFAM molecules at high pro tein concentrations were resolved and mean diffusion con stant D = 0.028 µm 2 s −1 was determined, which is much slower than that measured for monomeric TFAM (D = 0.08 µm 2 s −1 ), probably due to oligomerisation [31].
Similar to small molecule DNA binders, histones can be used for visualizing overall chromatin structure [32]. Imaging cells stained with antibody against core histone H3 has revealed chromatin domains of 40-70 nm size in neonatal rat ventricular myocytes, primary neurons, HeLa and HEK293 cell lines. The authors proposed that improved resolution allows tracking individual chromatin fibers, which provides an overview of the global chromatin structure and facilitates distinguishing between chromatin and interchromatin regions. Treatment of cells with hypertrophic agonist isoproterenol resulted in redistribution of staining intensities, which was interpreted as global chromatin rearrangements. Although interesting, such data lacks the information on the identity of the genomic regions undergoing change, which precludes mechanistic interpretations. More specificity can be achieved by staining one or several chromatin components (proteins and/or DNA sequences) that are involved in a particular process. For example, visualization of centromeric and pericentromeric satellite repeats by FISH (fluorescence in situ hybridization) has revealed massive dis tension of these, otherwise tightly packed DNA regions, early at the onset of cell senescence [33]. These changes were of a massive enough scale (up to 5 µm) to be detected by con ventional confocal microscopy. However, STED microscopy provided finer details on their structure: in both, proliferating and senescent cells, peri/centromeric satellite DNA comprises globular domains linked by DNA threads [34]. In proliferating cells, however, these globules are held in close proximity, while in senescent cells the threads are much more extended, suggesting loss of a factor that originally holds them together. This information offers a way to further address the mech anisms underlying these changes.
Research fields dealing with such complex processes as DNA replication, repair and recombination have greatly bene fitted from the superresolution colocalization techniques for investigation of heterogeneous ensembles that span kilobases of DNA and involve multiple copies of chromatin proteins (reviewed in [38,39]). In many cases, improved resolution allows to observe the real dimensions of the structures of interest, although in each case this must be carefully verified by comparing the obtained dimensions with those determined by either electron microscopy or other independent methods.
DNA replication in mammalian cells occurs in discrete sub nuclear locations-replication foci or factories. These facto ries are sized below diffraction limit (<250 nm), assemble during S phase and contain high local concentrations of repli cation enzymes. Green et al was the first group that analyzed DNA replication using STED microscopy [35]. They stained PCNA (proliferating cell nuclear antigen) and RPA (replica tion protein A) with antibodies. STED microscopy images revealed that replication factories are approximately 150 nm in diameter, and up to 1400 of replication foci, each containing 2-3 replication forks, exist in an early Sphase nucleus ( figure  4(a)). These numbers are up to 10fold different compared to the numbers obtained by confocal microscopy highlighting the importance of superresolution imaging. Furthermore, the replication factory size is close to the dimensions obtained by electron microscopy demonstrating the reliability of the performed measurements. Taking advantage of the increased resolution, the authors noticed ~40% reduction in number and ~30% increase in diameter of replication factories in the cells treated with the replication inhibitor hydroxyurea. Such subtle changes cannot be reliably identified by standard confocal imaging.
DNA is under constant assault by various internal and external agents which can lead to heterocyclic base, sugar or phosphodiester damage. The latter can result in difficult to repair and potentially deadly doublestrand breaks. Many studies have employed STED microscopy to study DNA doublestrand break repair. Histone H2AX phosphoryla tion on serine residue 139, known as γH2AX, is a common marker of doublestrand breaks, which is often used to esti mate the extent of DNA damage [40].
A recent study used STED microscopy to examine organi zation of nucleoprotein complexes inside the ionizing radia tioninduced foci (IRIF), which encompass megabase pair chromatin regions, containing phosphorylated histone γ H2AX and multiple DNA repair and DNA damage signaling factors [41]. Both, the mediator protein 53BP1 that is involved in the nonhomologous end joining and Rad51 protein that links complementary chromatin strands in homologous recom bination, belong to this compartment. Although γH2AX and 53BP1 colocalize on a micrometer scale in confocal images, imaging with STED showed nearly complete anticorrelation of the two signals ( figure 4(b)). An earlier study has similarly shown anticorrelation between 53BP1 and Rad51 signals [42]. This allowed building IRIF model, where each of these proteins occupy different positions in accordance to their dis tinct functions in DNA repair: 53BP1 occupies perichromatin region, but does not penetrate the dense chromatin domains marked by γH2AX, while Rad51 locates at the center, in the singlestranded DNA compartment. Such arrangement is consistent with the known role of Rad51 in the resection of DNA ends, while 53BP1 in the periphery may serve to limit the resection and stabilize the whole structure.
Another study investigated the structure of DSB repair sites by integrating the data from superresolution microscopy and wholegenome sequencing [38]. This study has found that γ H2AX foci that are identified by conventional microscopy upon xray irradiation, in reality represent higher order chro matin structures of discontinuously phosphorylated chromatin, that consist of clustered γH2AX nanofoci of 40-160 kbp and ~200 nm in size. Each of such clusters was found to contain one DSB and distribution of these γH2AX clusters changed according to the progression of repair. Majority of data pre sented in this paper was produced with 3DSIM microscopy, and imaging with STED, that provided a further 2fold improve ment in resolution, was mainly used to verify the findings.
Immunostaining of γH2AX and DNA repair protein 53BP1 was used to compare pseudo and real superresolu tion techniques: Zeiss Airyscan and Leica HyVolution 2, STED, groundstate depletion microscopy followed by indi vidual molecule return (GSDIM) and structured illumination microscopy (SIM) [36]. In agreement with previous studies, the authors found that all techniques showed much improve ment in the spatial resolution significantly below diffraction limit, ranging from 40 to 140 nm. Superresolution imaging increased precision of DNA repair foci localization but this did not always lead to identification of a higher number of foci. In addition, the authors expressed another DNA repair protein Ku80 as GFPfusion and found that γH2AX, 53BP1 and Ku70/80 repair proteins do not fully colocalize in the higher order chromatin structure.
DualImmuno FISH was used to colocalize telomeric DNA sequences, γH2AX and 53BP1 to follow the appearance and structure of telomereassociated foci in mice and human cardiomyocites upon aging [43]. This study addressed the role of irreparable telomere damage and cellular senescence in agerelated cardiac dysfunction. The high resolution of STED allowed better discrimination of telomere lengths, and enabled to resolve individual telomeres in telomere clusters.
This provided the evidence that lengthindependent telomere damage, rather than telomere shortening, underlies the induc tion of senescence in the postmitotic cardiomyocytes.
Another study used STED microscopy for investigating double strand break repair by RecAcatalyzed homologous recombination in bacteria [44]. It found that damage induced RecA foci are typically between 50 and 200 nm in length. The authors have developed a new method for detecting RecA complexes in the nucleoids spread on a glass surface, and stained with antiRecA and antiSSB antibodies. They have demonstrated that RecAK250N, a lossoffunction ATPase defective mutant, spontaneously form DNAbound complexes, which cause cytotoxicity. Furthermore, using combination of superresolution microscopy and biochemistry, they could prove that ATPase activity of RecA prevents accumulation of toxic complexes (foci) caused by binding of RecA to undam aged regions of dsDNA.
STED microscopy was used to dissect a new role of con densin in the maintenance of sister chromatids cohesion during meiosis [45]. In this case, high imaging resolution was essential to demonstrate that the reduced levels of cohesin in condensin I mutants interfere with the recruitment of synap tonemal complex central region proteins. Importantly, in the case of conventional fluorescence microscopy, the proteins under study appear to colocalize, thus making it impossible to resolve the mutant phenotype. Viral infections are followed by the replication of the viral genome. Recently, STED microscopy was applied to track DNA during lytic cycle of herpes simplex virus 1 infections [46]. The authors used immunofluorescence and FISH to observe positions of DNA, viral single strand binding protein ICP8 and host RNA polymerase II. They found that regions of DNA replication marked by ICP8 are spatially separated from regions of active transcription. In addition, they noted that the viral genome is expanding spatially during the replication process. Again, the use of STED microscopy considerably increased the precision of these observations.
Mitochondrial DNA encodes genes required for oxidative phosphorylation, rRNAs and tRNAs. mtDNA is packaged in a compact structure called nucleoid, which is composed of DNAprotein complexes. Kukat et al conducted an interesting study in which STED imaging revealed that mitochondrial nucleoids in a variety of mammalian species had a mean diam eter below ∼100 nm, which was much smaller than anticipated (figure 4(c)) [37]. The authors have used antibodies against DNA, mitochondrial transcription factor A (TFAM), or incor porated BrdU. A later study that focused on TFAM showed that mtDNA packing is a multistep process initiated by TFAM aggregation, crossstrand binding and DNA looping [47]. It was demonstrated that increased mtDNA copy number results from the increased nucleoid number, but not the change in their sizes. The authors suggested that a fundamental organ izational unit of mitochondrial nucleoid is a single copy of mtDNA packed by TFAM. Importantly, all these observa tions were backedup by electron microscopy. However, even such detailed studies do not provide a complete picture of mitochondrial nucleoid organisation. Stochastic 3D super resolution microscopy methods consistently provide wider nucleoid size distributions than those obtained by STED and suggest existence of nucleoid clusters and/or larger nucleoids of irregular shape. This discrepancy can only be resolved by imaging the same samples with different techniques in par allel (reviewed in [48]).

Points accumulation for imaging in nanoscale topography (PAINT) imaging
All types of microscopy benefit from a high number of pho tons arriving from the structure under investigation. This stimulates the development of brighter and more photostable dyes. An alternative strategy involves exchanging fluoro phores after photobleaching during the image acquisition. For the first time, such method was applied for SML micros copy and called Point Accumulation for Imaging Nanoscale Topography (PAINT) [52]. The versatility of DNA makes it an ideal PAINT probe. In particular, DNA is used as a fluoro phore carrier and affinity tag for staining structuresofinterest based on interactions between two complementary DNA strands. This has led to development of a whole family of imaging methods-DNA PAINT ( figure 5(a)) [53,54].
DNA PAINT principle was employed for visualization of multiple objects by the method called DNA exchange imaging (DEI). It uses DNAbarcoded antibodies (typically 9-10 nucle otides) against multiple targets that are detected by sequential application and gentle removing of the corresponding com plementary DNA strands, labeled with the same fluorophore. This works in combination with different imaging platforms: Exchange Confocal, ExchangeSIM, ExchangeSTED and ExchangePAINT. ExchangeSTED was applied for imaging four proteins (GFAP, pNFH, SynapsinI and Bassoon) in fixed DIV14 mouse hippocampal neurons. Localizations of target protein were determined by sequential exchange of corre sponding imager strands linked to one fluorophore-Cy3B [50]. To reduce the background, the imaging strands were labeled with fluorophores at both ends and the fluorescence of the unbound imager strands was quenched due to high local fluorophore concentration [49].
A different approach was applied for imaging DNA ori gami, where stable imager strand binding and low background was achieved by using longer complementary DNA and including denaturing agent in the washing buffer. A library of sixhelixbundle DNA structures carrying four specific single stranded DNA extensions arranged in four spots approxi mately 113 nm apart was investigated. Sequential labeling and imaging cycles resolved the labeled positions and deciphered unique barcodes on the origami structure. The specific labeled positions in the barcode were detected with 91% accuracy. This technique was also applied for cellular proteins. The combination of DNA barcoded antibody imaging resulted in excellent quality multicolor STED images (figure 5(b)) [51].

Conclusions and future perspectives
We have highlighted the most prominent examples of applica tion of STED microscopy for the imaging of DNA and DNA interacting proteins. These studies clearly demonstrate the benefits of superresolution imaging to uncovering the impor tant details of chromatin structure and function. Most of the work presented here was performed on fixed cells, but as the imaging speed increases and the labeling methods get optim ized, undoubtedly there will be more studies addressing the dynamic processes in living cells. The ongoing further devel opments towards increased resolution and imaging speed will be able to provide even more detailed picture. For example, the recently introduced MINFLUX imaging method achieves resolutions close to 1 nm and has been already applied for imaging DNA origami and tracking ribosomes in living cells [55,56]. This method has a great potential to advance our understanding of basic chromatin biology and facilitate the development of novel therapies.