An integrated quantitative single-objective light-sheet microscope for subcellular dynamics in embryos and cultured multicellular systems

TL;DR

The paper presents a quantitatively calibrated single-objective light-sheet platform based on oblique plane microscopy (OPM) that combines high-NA remote refocusing, tilt-invariant light-sheet scanning, and hardware-timed synchronization to enable rapid, long-term, and low-phototoxic volumetric imaging in embryos, stem cells, and organoids. It delivers an end-to-end quantitative workflow, including optical/detector characterization, deskewed 3D reconstruction, autofocus stabilization, deterministic device synchronization, and a unified Python-based acquisition/reconstruction pipeline. The authors demonstrate quantitative transcription imaging across Drosophila embryos, mouse embryonic stem cells, and gastruloids, showcasing single-locus transcription readouts, chromatin dynamics, and stable performance in complex 3D tissues. Together, the work establishes OPM as a robust, quantitatively calibrated platform for studying transcription and nuclear dynamics in living multicellular systems, with potential extensions to multi-color, chromatin-locus tracking and deeper tissue imaging.

Abstract

Quantitative imaging of subcellular processes in living embryos, stem-cell systems, and organoid models requires microscopy platforms that combine high spatial resolution, fast volumetric acquisition, long-term stability, and minimal phototoxicity. Single-objective light-sheet approaches based on oblique plane microscopy (OPM) are well suited for live imaging in standard sample geometries, but most existing implementations lack the optical calibration, timing precision, and end-to-end integration required for reproducible quantitative measurements. Here we present a fully integrated and quantitatively characterized OPM platform engineered for dynamic studies of transcription and nuclear organization in embryos, embryonic stem cells, and three-dimensional culture systems. The system combines high numerical aperture remote refocusing with tilt-invariant light-sheet scanning and hardware-timed synchronization of laser excitation, galvo scanning, and camera readout. We provide a comprehensive characterization of the optical performance, including point spread function, sampling geometry, usable field of view, and system stability, establishing a well-defined framework for quantitative volumetric imaging. To support high-throughput operation, we developed a unified acquisition and reconstruction pipeline that enables real time volumetric imaging at hardware-limited rates while preserving deterministic timing and reproducible geometry. Using this platform, we demonstrate quantitative three-dimensional imaging of MS2-labeled transcription sites in living Drosophila embryos, cultured mouse embryonic stem cells, and mESC-derived gastruloids, enabling extraction of transcriptional intensity traces across diverse biological contexts. This work establishes OPM as a robust and quantitatively calibrated single-objective light-sheet platform for transcription imaging in complex living systems.

Figures (9)

• Figure 1: OPM optical layout, geometry, and sampling. (A) Simplified schematic of the single-objective oblique plane microscopy (OPM) optical layout. An excitation beam (blue), pre-shaped by cylindrical optics, is delivered through the primary objective (OBJ1, inverted configuration) to generate a planar illumination at an oblique angle $\theta=30^\circ$ with respect to the OBJ1 focal plane. Fluorescence emission (orange) is collected by the same objective and relayed into a remote-refocusing module, where a third objective (OBJ3), tilted by the same angle $\theta$, reimages the oblique plane onto the camera sensor. A dichroic beamsplitter enables simultaneous dual-channel imaging with dedicated cameras. (B) A galvanometric mirror translates the tilted light sheet laterally through the sample while maintaining a fixed illumination angle. Each camera frame captures a single oblique optical section, and sequential acquisition of adjacent planes enables volumetric imaging. (C) During volumetric acquisition, fluorescence emission is descanned by the same galvo, resulting in a geometric shear in the raw image stack in which consecutive oblique slices are laterally displaced. Deskewing corrects this shear to reconstruct a geometrically accurate volume, yielding a coordinate system $(x',y',z')$ that is rotated with respect to the laboratory frame $(x,y,z)$. (D) The discrete sampling of adjacent oblique planes imposes constraints on the galvo step size $\delta$, which must be chosen such that pixels from consecutive frames align after deskewing. This condition relates $\delta$ to the lateral pixel size $p_{x'y'}$ and the illumination angle $\theta$; in our implementation, $p_{x'y'} = 115~\mathrm{nm}$ and $\theta = 30^\circ$, and we select $N = 5$, corresponding to an effective axial sampling of $p_{z'} = 330~\mathrm{nm}$. (E) Photographs of the assembled OPM system showing the primary objective with integrated sample incubation chamber (top left), the scanning unit containing the galvo mirror and relay optics (top center), the remote-refocusing module with objectives OBJ2 and OBJ3 (top right), and a top view of the complete microscope assembly (bottom).
• Figure 2: PSF and resolution characterization. (A) Measured point spread function (PSF) obtained by imaging sub-diffraction fluorescent beads. Upper left: the PSF is axially elongated along the laboratory $z$ axis but sampled along the oblique direction $z'$, resulting in a tilted PSF in the rotated coordinate system $(x',y',z')$. Upper right and bottom: orthogonal cross-sections $y'z'$, $x'z'$, and $x'y'$ of the measured PSF. Scale bar: $500~\mathrm{nm}$. (B) Line profiles extracted along the PSF principal axes $(x,y,z)$ with Gaussian fits yield full width at half maximum (FWHM) values of $295 \pm 30~\mathrm{nm}$ along $x$, $305 \pm 30~\mathrm{nm}$ along $y$, and $790 \pm 75~\mathrm{nm}$ along $z$ ($n > 1000$ beads). (C) Lateral ($x,y$; upper) and axial ($z$; lower) resolution measured as a function of lateral position across the field of view. (D) Lateral ($x,y$; upper) and axial ($z$; lower) resolution measured as a function of axial position within the sample volume. Together, these measurements demonstrate diffraction-limited performance over a volumetric field of view of approximately $150 \times 150 \times 40~\mu\mathrm{m}^3$.
• Figure 3: System stabilization via back-reflection–based autofocus. (A) Schematic of the autofocus optical path. Excitation light reflected at the coverslip–sample interface is redirected by a D-shaped mirror into a dedicated optical arm and imaged onto an auxiliary camera. Axial motion of the sample produces a measurable change in the back-reflection signal. (B) Determination of the OBJ1 focal plane position from back-reflection intensity. Maximum reflected intensity occurs when the coverslip–sample interface coincides with the focal plane of OBJ1, where the high-NA light sheet reaches its minimal waist. Left: interface at the focal plane. Right: interface displaced from the focal plane. (C) Calibration of absolute axial positioning. Integrated back-reflection intensity measured as a function of axial sample position (data points), with a polynomial fit (solid line) used to determine the focal plane location. (D) Reproducibility of sample repositioning to the focal plane following repeated axial displacements. (E) Calibration for active drift correction. Upper: polynomial fit of the back-reflection signal used to infer relative axial sample position. Lower: resulting linear relationship between lateral displacement of the back-reflection spot and axial sample position. (F, G) Long-term stability during extended acquisitions with autofocus feedback disabled (F) and enabled (G).
• Figure 4: Hardware synchronization of galvo, lasers, and camera. (A) Hardware architecture. A DAQ card functions as the master timing controller, generating clock-synchronized analog (galvo) and digital (camera, laser) signals. Precomputed timing sequences are loaded onto the DAQ for autonomous, deterministic execution during acquisition, ensuring fixed relative timing between galvo motion, laser illumination, and camera exposure. The host computer asynchronously acquires frames for visualization and processing. (B) Timing diagram for one volume acquisition. Command signals (black) sent to galvo (analog staircase), camera (exposure gates), and laser (illumination pulses) are shown with measured device responses (blue). Finite latencies between commands and responses require compensation to synchronize illumination with camera exposure during galvo dwell periods. (C) Measured device latencies. Quantified response times for galvo (mechanical settling), camera (trigger delay), and laser (switching time) enable compensation in the DAQ timing sequences, ensuring optimal synchronization and acquisition efficiency.
• Figure 5: Software pipeline for high-throughput volumetric imaging. Schematic of the process-based software architecture used for acquisition, visualization, and data handling. Hardware control is managed by a main process, while camera readout, preprocessing, and postprocessing are executed in independent subprocesses. Data flow proceeds as follows: (1) camera subprocesses acquire image frames into shared circular buffers and post volume-completion notifications to message queues; (2) preprocessing worker pools monitor these queues, retrieve buffer metadata, and generate maximum-intensity projections that are written to Zarr arrays; (3) the graphical user interface reads the Zarr arrays to enable real-time visualization during acquisition; (4) postprocessing workers receive queued tasks and write raw volumetric data to disk. Queue-based messaging ensures process independence, while shared-memory buffers enable zero-copy data access, supporting sustained high-throughput acquisition without software bottlenecks.