DNA is rapidly shifting from being merely the subject of biological imaging to becoming the imaging medium itself. A landmark 2019 study demonstrated that short DNA molecules, diffusing through a sample and concatenating with one another, could encode spatial information so precisely that an algorithm could reconstruct a physical image of cells without any lenses, cameras, or light sources at all.1Cell. DNA Microscopy: Optics-Free Spatial Imaging of Cellular Biopolymers That concept, called DNA microscopy, is just one thread in a broader revolution. Across dozens of labs, DNA-based techniques are now being used to paint nanometer-resolution pictures of proteins, map which genes are active in which cells of a tissue, record the family trees of individual cells, and even build tiny rulers that calibrate other microscopes. Together, these approaches are rewriting the rules of what biological imaging can capture and how.
DNA Microscopy and the Idea of Optics-Free Imaging
Traditional microscopy works because photons bounce off or pass through a sample and are focused by glass lenses onto a detector. DNA microscopy replaces every part of that chain with chemistry. The sample’s RNA or DNA molecules are first tagged with unique molecular identifiers. Those tagged molecules then amplify and diffuse outward, forming overlapping clouds. Where two clouds overlap, the identifiers physically join together, creating a new combined sequence. The frequency of these concatenation events encodes how close two original molecules were: molecules that started near each other produce many joined sequences, while distant molecules produce few or none. A computational algorithm then reads all the joined sequences from a standard DNA sequencer and works backward to infer the positions of every original molecule.2Cell. DNA Microscopy: Optics-Free Spatial Imaging of Biological Specimens
The result is a spatial map built entirely from sequence data. Because the “imaging power” comes from diffusive molecular dynamics rather than from photons, the technique sidesteps every physical limitation that plagues optical microscopy: diffraction limits, photobleaching, the need for transparent samples, and expensive objective lenses. A related approach called DNA-GPS uses a grid of known “satellite” barcodes to anchor the reconstruction, achieving resolutions around 10 micrometers and near-single-cell precision across a range of barcode densities.3Cell Systems. DNA Pictures: The Future of Biological Imaging – Section: DNA GPS
In 2025, a team extended DNA microscopy into three dimensions for the first time, imaging an intact organism by building an intermolecular network of proximal barcodes throughout the specimen’s volume. Their reconstruction method inferred molecular positions across tens of millions of molecules in a single measurement, capturing transcriptomes, genotypes, and gross morphology simultaneously without any prior knowledge of the specimen’s structure.4PubMed Central. Spatial transcriptomic imaging of an intact organism using volumetric DNA microscopy That step, from flat 2D images to volumetric ones, signals that DNA microscopy is moving toward being practical for real biological questions rather than remaining a proof of concept.
DNA-PAINT and Nanometer-Scale Super-Resolution
While DNA microscopy uses chemistry to replace optics entirely, another family of techniques uses DNA to push optical microscopy far beyond its normal limits. DNA-PAINT exploits a simple trick: short, dye-labeled DNA strands (called imager strands) are designed to bind transiently to complementary target strands (docking strands) attached to the structure you want to see. Each binding event produces a brief flash of fluorescence. Because binding is random and temporary, only a few molecules light up at any given moment, allowing a camera to pinpoint each one precisely. Over thousands of frames, these pinpoints accumulate into a super-resolution image.5PubMed. Super-resolution microscopy with DNA-PAINT
The technique was first demonstrated on DNA origami nanostructures, where researchers achieved resolution below 30 nanometers by exploiting the reversible, specific binding of labeled oligonucleotides.6PubMed. Single-molecule kinetics and super-resolution microscopy by fluorescence imaging of transient binding on DNA origami A persistent limitation was high background fluorescence from unbound imager strands floating in solution, which slowed imaging speed. A fluorogenic variant solved this by engineering probes that carry both a fluorophore and a quencher; the probe is dark while floating freely and brightens up to 57-fold only upon binding its target. Base-pair mismatches between probe and target were further tuned to balance brightness with fast binding kinetics.7Nature Methods. Fluorogenic DNA-PAINT for faster, low-background super-resolution imaging
What makes DNA-PAINT especially versatile is that the “blinking” rate does not depend on the photophysics of the dye itself, unlike other super-resolution methods. It depends on the concentration and binding kinetics of the imager strands, which researchers can tune by adjusting DNA sequence, salt concentration, or strand length. This programmability opens the door to multiplexed imaging: by using several orthogonal imager-docking pairs, you can image multiple protein targets in the same cell sequentially, swapping out one imager solution for another between rounds.
Mapping Genes in Place with Spatial Transcriptomics
Knowing which genes are turned on is useful. Knowing where in a tissue those genes are turned on is transformative. Spatial transcriptomics refers to a family of methods that measure gene expression while preserving the tissue’s physical layout, and DNA-based approaches dominate this space.
One major branch uses multiplexed fluorescence in situ hybridization. MERFISH, for instance, assigns each gene a unique binary barcode and then images the tissue through multiple rounds of hybridization, each round revealing one bit of the barcode. By reading the full barcode for each RNA molecule, the method can simultaneously measure transcripts from roughly 10,000 genes in individual cells, with about 80 percent detection efficiency and a misidentification rate near 4 percent.8PubMed Central. Spatial transcriptome profiling by MERFISH reveals subcellular RNA compartmentalization and cell cycle-dependent gene expression The technique has recently been adapted even for bacteria, which are far smaller than mammalian cells, by combining it with 1,000-fold volumetric expansion to make bacterial cells large enough to resolve thousands of operons spatially.9PubMed Central. Highly multiplexed spatial transcriptomics in bacteria
A complementary branch takes an array-based approach. Tissue sections are placed onto a glass slide studded with millions of spatially barcoded capture probes. As RNA diffuses out of the tissue, it is captured and reverse-transcribed by the nearest probe, which stamps each transcript with a positional barcode. Sequencing then reveals both the identity and the approximate location of each transcript. Early implementations demonstrated this in mouse brain and human breast cancer tissue.10PubMed. Visualization and analysis of gene expression in tissue sections by spatial transcriptomics Later versions used denser bead arrays to push resolution higher, capturing RNA on spatially barcoded beads packed tightly enough to approach single-cell scale.11Nature Methods. High-definition spatial transcriptomics for in situ tissue profiling
Reading Sequences Directly Inside Intact Tissue
Rather than pulling RNA out of a tissue section and sequencing it externally, in situ sequencing methods read the genetic code right where the molecules sit. FISSEQ, one of the first genome-wide in situ approaches, converts RNA into cross-linked cDNA amplicons inside fixed cells and then sequences those amplicons directly on a confocal microscope. Unlike standard RNA-seq, it enriches for context-specific transcripts over housekeeping RNA and preserves the tissue’s architecture throughout.12PubMed Central. Fluorescent in situ sequencing (FISSEQ) of RNA for gene expression profiling in intact cells and tissues
A more recent evolution combines in situ sequencing with expansion microscopy. Expansion sequencing physically swells biological specimens in a hydrogel, separating molecules that were previously too close to distinguish. This physical expansion enables both targeted and untargeted sequencing of RNA at subcellular resolution, localizing transcripts to specific compartments like dendrites or organelles that conventional methods blur together.13PubMed Central. Expansion sequencing: Spatially precise in situ transcriptomics in intact biological systems When paired with a protein-imaging method called cycleHCR, this approach revealed an intricate network of ten distinct subcellular structures in mouse embryonic fibroblasts, showing that gene expression maps and structural maps can now be layered in the same experiment.14PubMed Central. Deep-tissue transcriptomics and subcellular imaging at high spatial resolution
Recording Cellular History with DNA Barcodes
Some of the most creative uses of DNA as an imaging tool are not about seeing what is happening right now but about reconstructing what happened in the past. Lineage tracing uses DNA barcodes, often introduced via CRISPR-based editing, to mark cells so that every descendant inherits a slightly modified version of the barcode. By reading those barcodes later, researchers can reconstruct which cells are related and how they diverged during development.
A foundational demonstration of this in mammals used multiple homing guide RNAs in mice, each generating hundreds of mutant alleles that combined into an exponential diversity of barcodes. Activation at conception and continued editing through gestation produced “developmentally barcoded” mice, allowing researchers to reconstruct the earliest lineage splits and investigate how the brain’s axes formed.15PubMed Central. Developmental barcoding of whole mouse via homing CRISPR The approach has since been adapted to other species: a CRISPR/Cas9-based barcoding method in sheep used synthesized targeting arrays introduced at multiple genome loci, generating heritable clonal markers detectable throughout embryonic development.16Scientific Reports. A method for CRISPR/Cas9-induced genetic barcoding and lineage tracing in sheep
The broader landscape of molecular recording now encompasses CRISPR/Cas systems, base editing, recombination-based approaches, and even innate variable sequences in the genome. Each strategy trades off barcode diversity against editing fidelity and timing resolution, but together they are making it possible to trace the hidden intermediate steps of development and disease progression that no snapshot technique could capture.17PubMed Central. DNA Barcoding Technology for Lineage Recording and Tracing to Resolve Cell Fate Determination
Seeing DNA’s Own Structure
All of the technologies above use DNA as a tool to image other things. But imaging DNA itself, at the level of its physical double helix, has its own frontier. Atomic force microscopy drags a nanoscale tip across a surface to build a topographic map of whatever sits on it. For DNA, this means researchers can see single molecules in liquid, under near-physiological conditions, without any labeling or staining.18PubMed Central. Atomic force microscopy-A tool for structural and translational DNA research Recent work using cobalt-mediated adsorption has achieved enough resolution in liquid to visualize the double-helix structure itself and variations within it, a feat that would have been hard to imagine a decade ago.19bioRxiv. Co2+-mediated adsorption facilitates atomic force microscopy of DNA molecules at double-helix resolution
At larger scales, cryo-electron microscopy and cryo-electron tomography are resolving how DNA is packaged into chromatin, the higher-order structure that determines which genes are accessible and which are silenced. Recent cryo-EM studies of the fundamental chromatin unit, the nucleosome, have clarified how structural details of this packaging relate to gene regulation.20PubMed. Chromatin structure meets cryo-EM: Dynamic building blocks of the functional architecture Where AFM excels at single-molecule views in native conditions, cryo-EM captures frozen-in-time snapshots at near-atomic resolution. The two complement each other more than they compete.
DNA Origami as Nanoscale Rulers
One surprisingly practical offshoot of DNA nanotechnology is its use as a calibration standard for other microscopes. DNA origami, the technique of folding long DNA strands into precise nanoscale shapes, allows researchers to place fluorescent dye molecules at exact, pre-designed positions. These “nanorulers” serve as reference structures for super-resolution microscopes, providing traceable distance measurement standards with an expanded uncertainty of roughly ±2.4 nanometers.21Scientific Reports. Using DNA origami nanorulers as traceable distance measurement standards and nanoscopic benchmark structures
The concept extends into three dimensions. DNA origami nanopillars, about 220 nanometers tall and 14 nanometers in diameter, can be immobilized upright on a surface via biotin anchoring at their base. With site-selective dye labels, researchers visualized the pillars’ structure and orientation using 3D super-resolution microscopy, establishing them as rigid, addressable scaffolds for benchmarking any 3D super-resolution setup.22Nano Letters. DNA Origami Nanopillars as Standards for Three-Dimensional Superresolution Microscopy Having a reliable, reproducible physical standard at this scale matters for comparing results across instruments and laboratories, which is a prerequisite for these technologies to move from research demonstrations into clinical use.
Computational Reconstruction and Machine Learning
Nearly every DNA-based imaging technique produces raw data that looks nothing like an image. It is a pile of sequences, or fluorescence traces, or bead-capture counts. The actual picture emerges only after computational reconstruction, and this step is becoming as important as the wet-lab chemistry.
For spatial transcriptomics, a persistent challenge is stitching together 2D tissue slices into a coherent 3D volume. One recent tool uses graph convolutional neural networks and Fourier transforms to align and reconstruct three-dimensional spatial structures from serial 2D slices, preserving spatial gene expression patterns that would be lost if each slice were analyzed in isolation.23PubMed Central. 3D reconstruction of spatial transcriptomics with spatial pattern enhanced graph convolutional neural network Another approach, scHolography, uses machine learning to project single-cell transcriptome data into spatial coordinates, reconstructing cellular neighborhoods and enabling 3D tissue visualization even when the original spatial data is incomplete or sparse.24PubMed Central. scHolography: a computational method for single-cell spatial neighborhood reconstruction and analysis
The computational challenge scales steeply. Volumetric DNA microscopy of a single organism involves tens of millions of molecules, each generating multiple sequence reads. The 3D reconstruction method used for that task, geodesic spectral embedding, was chosen specifically because it handles the sparse, noisy distance data that DNA-based proximity measurements produce.4PubMed Central. Spatial transcriptomic imaging of an intact organism using volumetric DNA microscopy As these datasets grow, the algorithms are becoming as much of a bottleneck as the chemistry, and improvements in one often unlock the other.
Applications in Cancer and Neuroscience
The practical payoff of these technologies is already visible in cancer research. Spatial omics approaches have been used to profile the tumor microenvironment, the complex ecosystem of cancer cells, immune cells, and structural tissue that determines whether a tumor grows or shrinks. In breast cancer, researchers combined single-cell RNA sequencing with spatial transcriptomics to study the transition from ductal carcinoma in situ to invasive disease. By ordering cells along a pseudotime trajectory and mapping their spatial positions, they found that the physical migration of tumor cells from ducts into surrounding tissue corresponded to specific gene expression changes linked to aggressive behavior.25PubMed Central. Spatial omics for profiling the dynamic tumor microenvironment This kind of spatially resolved molecular progression map is something no conventional biopsy or bulk sequencing could provide.
In neuroscience, DNA barcoding is being used to map not just which cells are where but where their axons project. BARseq2, an upgraded barcode sequencing method, simultaneously detects gene expression and projection patterns in the mouse brain. It revealed that a shared set of cadherin genes mediates the projections of a specific neuron class to both the motor cortex and auditory cortex, demonstrating the potential for uncovering the molecular logic behind how neural circuits are wired.26PubMed Central. Spatial transcriptomics in neuroscience Spatial transcriptomics more broadly is enabling researchers to study brain organization at a level of molecular detail that was simply inaccessible when they could only dissect regions and sequence them in bulk.
Where the Technical Limits Still Bite
For all the progress, each technique carries real trade-offs. DNA microscopy’s resolution, while sufficient to distinguish individual cells, is still far coarser than optical super-resolution methods. DNA-GPS simulations suggest resolutions near 10 micrometers under ideal conditions, which is fine for mapping cell types across a tissue but cannot resolve structures within a single cell.3Cell Systems. DNA Pictures: The Future of Biological Imaging – Section: DNA GPS DNA-PAINT achieves nanometer resolution but requires fixed, transparent samples and long acquisition times; live cells are out of reach. MERFISH can profile thousands of genes but demands many rounds of hybridization and imaging, making it slow and computationally intensive. Array-based spatial transcriptomics captures unbiased transcriptome-wide data but blurs spatial resolution because RNA diffuses before being captured.
Sequencing cost and throughput are a shared constraint. DNA microscopy, lineage tracing, and spatial transcriptomics all ultimately convert spatial information into sequence reads, and the number of reads required grows with the volume imaged, the number of genes measured, and the resolution desired. A volumetric DNA microscopy experiment on a whole organism involves on the order of tens of millions of molecules, each needing sufficient sequencing depth to reliably estimate pairwise proximity. As sequencing costs continue to fall, the balance between resolution and affordability will keep shifting, but for now, most labs face hard choices about what to measure and at what scale.
Sample preparation is another underappreciated bottleneck. Many of these techniques require fixation, permeabilization, or physical expansion of the specimen, each of which can distort the very biology being measured. Plant tissues, with their rigid cell walls, are especially challenging to process for spatial transcriptomics. And live-cell applications remain limited: delivering DNA-based probes into living cells without disrupting their behavior is an active area of engineering, but most of the high-resolution techniques described here work only on dead, fixed material.
DNA-Based Probes for Live Cells
Getting DNA probes to work inside living cells is one of the field’s harder unsolved problems. DNA nanoprobes designed for imaging microRNA inside live cells, for example, must cross the cell membrane, resist degradation by nucleases, and avoid triggering immune responses, all while remaining bright and specific enough to detect their targets at low copy numbers. Various delivery vehicles have been developed and tested for their ability to shuttle these probes across membranes while maintaining biosecurity and resilience in the complex intracellular environment.27PubMed Central. Shedding Light on DNA-Based Nanoprobes for Live-Cell MicroRNA Imaging The challenge is that the same chemical stability that makes DNA an excellent information-carrying molecule also makes it conspicuous to the cell’s defense machinery. Modified nucleotides, protective coatings, and nanoparticle carriers all help, but no universal solution exists yet. Solving this would open the door to watching gene regulation unfold in real time inside individual living cells, rather than reconstructing it from fixed snapshots after the fact.