1. Introduction
Real-time observation of dynamic cellular processes within living tissues is crucial for understanding biological phenomena at the network level. Synaptic and diffusive signaling among cells on timescales of milliseconds to minutes trigger intracellular signaling cascades, causing changes in gene expression patterns in individual cells. These patterns reflect the structure of a local connectome, the network of connections among cells. Circadian clocks are dynamic oscillations in the expression of clock genes. Observing gene expression in circadian clocks provides an ideal system for probing network dynamics. It has been computationally demonstrated that the short timescale dynamics of intercellular communication can shape the spatial patterning of circadian gene expression in the suprachiasmatic nucleus (SCN), the master pacemaker tissue that regulates the circadian rhythms of the rest of the body [
1].
Organotypic culture based on a semi-permeable membrane enables a stable culture and long-term observation [
2]. By culturing tissues expressing reporters on membranes under the microscope, we can now track the evolving gene expression patterns. This method allows us to decipher the long-term functional connectome and reveal how the network coordinates cellular behaviors, such as the timing of the circadian clock in each cell [
3].
Reporter gene expression enables real-time observation of live systems by inserting a coding sequence that produces a detectable signal when a gene of interest is transcribed. Green fluorescent protein (GFP) and its variants are commonly used for optical measurement of these activities. This fluorescent reporter technique is particularly effective for revealing short-term dynamics and has been highly successful in monitoring calcium dynamics, as demonstrated with GCaMP [
4]. However, fluorescent reporters are often not suitable for long-term culture imaging due to phototoxicity caused by the ultraviolet (UV) light required for excitation. This can damage cells and the reporter molecule itself. Additionally, fluorescence has a limited dynamic range and often produces nonspecific background fluorescence that can obscure weak signals.
In contrast, bioluminescence reporters, such as luciferase, offer a distinct advantage for long-term observation. Firefly luciferase (Fluc), for example, was first adopted as a reporter of gene transcription, analogous to the β-galactosidase assay. The luciferase assay does not require histological processing, such as fixation, and live samples can be used as is for measurements. This technique relies on the bioluminescence reaction involving luciferase, luciferin, and ATP, which produces light through an enzyme-catalyzed process. In this reaction, luciferase catalyzes the oxidation of luciferin in the presence of ATP and oxygen, resulting in the formation of oxyluciferin. As oxyluciferin returns to its ground state, the excess energy is released as photons, producing visible light. Since luciferin is membrane-permeable, it can be added to the culture medium to generate light without the need for excitation light, unlike fluorescence-based reporters. The low light signal generated by luciferase can be difficult to capture without sensitive light detectors, such as the photomultiplier tubes used in luminometers. Nonetheless, while the signal and noise of fluorescent reporter detection are influenced by excitation and emission, the detection of bioluminescence reporter activity is primarily limited by the background noise of the detection apparatus [
5]. Bioluminescence does not require external excitation since light is produced within the cell as long as the substrate is present [
6]. This eliminates phototoxicity concerns and allows for continuous, long-term monitoring.
It should be noted that despite the low brightness of the signal, a single luciferase molecule, such as Fluc, produces multiple photons. This makes the absolute quantification of the number of reporter molecules difficult, though not impossible [
7]. The accepted unit for bioluminescence is therefore the relative light unit (RLU). Here, since we base the measurements on the brightness registered and digitized by the same charge-coupled device (CCD), we report bioluminescence values in terms of 16-bit pixel intensity values.
Advancements in CCDs have revolutionized the use of bioluminescent reporters for visualizing spatial gene expression, both in vitro and in vivo [
8]. The organotypic culture of the SCN demonstrated the powerful potential of bioluminescence reporter systems for systems biology, particularly in studying synchronization dynamics among circadian clock neurons at the intratissue network level [
9]. Since then, reporter imaging methods have been extensively employed to reveal network dynamics beyond the circadian timescale, including photoperiodic encoding in tissue explants [
10]. The success of bioluminescence imaging in circadian biology has led to computational investigations into dynamic network interactions [
3].
However, bioluminescence signals are inherently weak, posing a significant challenge for imaging, particularly at the tissue level where a large field of view and typically low-magnification objectives are required, which gather less light. Efforts to enhance bioluminescence imaging have focused on two key strategies: developing brighter luciferases and utilizing high-sensitivity cameras. Brighter luciferases, such as the emerald luciferase (ELuc) from the Brazilian click beetle, increase overall signal intensity [
11]. Alternatively, high-gain cameras like intensified CCDs (I-CCDs) and electron-multiplying CCDs (EM-CCDs) amplify the weak signal. Nevertheless, these approaches have limitations. Brighter luciferases can lead to a higher baseline signal, potentially reducing the signal-to-noise ratio (SNR) [
12]. EM-CCDs, while amplifying the signal, also amplify the inherent dark current noise within the camera. This can limit the overall image quality, and EM-CCDs are significantly more costly than conventional cooled CCDs [
13].
In popular photography and astronomical imaging, there is a technique that increases the light-gathering ability of the lens, known as telecompression, or more popularly, the “speed booster” [
14]. It is an adapter that sits between the sensor and the lens, reducing the lens’s magnification. This is the opposite of a teleconverter, which increases magnification. Higher magnification (telephoto) lenses collect light from a narrower field of view, concentrating the light onto a smaller area. A speed booster can then expand this field of view. Importantly, it does so while preserving the total number of photons collected. A similar principle can be directly applied to bioluminescence imaging. By using a high-magnification and high-numerical-aperture objective along with a low-magnification converting relay lens between the objective and the CCD, more photons can be gathered per single cell that emits bioluminescence.
Telecompression can provide a cost-effective solution to optical enhancement of bioluminescence imaging, especially because the lack of excitation–emission simplifies the optics. We employed this technique in our first timelapse bioluminescence imaging about a decade ago, where we revealed that single-cell oscillation periods in the SCN can differ under explant culture conditions [
12]. However, we did not focus on this optical method or rigorously quantify the signal improvements achieved through telecompression. Here, we compare the signal improvements over conventional optics to illustrate its clear benefits.
4. Discussion
We demonstrated a straightforward optical configuration that enhances signal strength in bioluminescence imaging. The microscopy setup proposed in
Figure 1C can be easily constructed using commercially available components or adapted to existing microscopy setups. In this demonstration, we used Nikon components, but we have also tested similar configurations with Olympus, Leica, Mitutoyo, and generic components, all yielding comparable improvements in signal strength. It is evident that this simple and economical construction provides clear benefits in increasing the collection efficiency of bioluminescence signals. Although this optical improvement can only be achieved with a high-NA objective lens, high-NA lenses at low magnification are often highly expensive or even unavailable. In our example, we boosted an NA of 0.45 at an effective magnification of 11×; however, it can be increased to NA 0.60 by combining a 40x objective lens with a 0.25× relay lens (
Supplementary Figure S2). Therefore, applying the telecompression principle to bioluminescence imaging offers clear advantages.
One key advantage of the enhanced light collection for circadian systems is the reduction of exposure time, which shortens the sampling interval in timelapse imaging. Circadian clocks in different parts of the body express characteristic period lengths [
19]. Even within the SCN tissue, subtle differences in period length can exist among circadian clock cells when freerunning under explant culture [
12]. Resolving these subtle differences using spectral methods like Fast Fourier Transform can be challenging when the sampling interval is limited to the standard 1 h. However, our method allows for single-cell resolution with a 15 min exposure under 4-binning (
Figure 3). This significantly boosts the spectral resolution of single-cell circadian oscillations in conventional cooled CCD cameras.
However, the telecompression approach has its limitations. As shown in the low bioluminescent sample (
Figure 2B), the bioluminescent signal must exceed the CCD’s baseline noise level for detection. While our approach can partially address this issue, signal strength can be further improved by engineering the luciferase for higher brightness [
11]. Despite this, the telecompression method offers the advantage of enhancing the signal while keeping baseline noise unchanged. On the other hand, we found that EM-CCDs amplify noise during long exposures, whereas bioluminescence images obtained using conventional cooled CCDs with the telecompression system were less noisy. Conventional CMOS (Complementary Metal–Oxide–Semiconductor) sensors have not been optimal for bioluminescence imaging due to their lower signal-to-noise ratio and limited exposure time compared to CCD sensors. The cooled scientific CMOS (sCMOS) sensors we tested did not suppress baseline noise during long exposures as effectively as cooled CCDs (see data from the Hamamatsu ORCA-Quest qCMOS in
Supplementary Figure S3. In comparison, a more conventional Sony IMX183-based imager did not return any bioluminescence image). While sCMOS sensors provide excellent images for fluorescence imaging due to their much lower read noise, their greater dark noise during long exposures makes them less suitable for bioluminescence imaging compared to CCDs. There have been recent developments in CMOS sensor technology, including improved quantum efficiency through back-illumination, reduced readout noise, and increased full well capacity, although exposure time remains limited. Using the latest engineered luciferase, NanoLuc (100-fold brighter than firefly luciferase), along with a qCMOS camera, volumetric bioluminescence imaging in
Caenorhabditis elegans was recently demonstrated through computational image restoration using deep learning [
20]. With parallel advancements in emission, collection, and detection (Equation (2)), bioluminescence shows great promise as the next reporter of choice for monitoring tissue culture systems, including organoids.
The telecompression system has another limitation when high magnification images are required. To utilize a reducing relay lens effectively, a higher magnification objective is needed, often requiring oil or water immersion beyond 40x magnification. We found that maintaining imaging with immersion objectives is impractical. Water immersion lenses are problematic due to contamination and issues with temperature and moisture control. Oil immersion lenses, on the other hand, often have a working distance too short to reach the tissue explant sample. Additionally, with the high temperatures maintained in the stage-top incubator, the immersion oil dries out over the course of the observation period, which typically lasts a week for circadian experiments. Therefore, the relay lens solution is the most suitable for effective magnification at around 10×, which is ideal for visualizing and analyzing network dynamics in cultured systems.
With the development of red-shifted luciferase substrates, such as CycLuc1 (emission wavelength 599 nm) and TokeOni (675 nm), in vivo tissue imaging of bioluminescent reporter activity is becoming more feasible [
21]. Although achieving near-infrared emission without mutations in luciferase remains challenging, these new approaches open up opportunities for deeper tissue imaging in live animals, especially with the development of the skin-transparenting technique using an orange–yellow dye, which allows for the transmission of these long-wavelength signals [
22]. Since telecompression increases the field of view, the method may also be useful for in vivo imaging, although additional considerations are required for low-wavelength optics.
Using telecompression, a technique originally used in other fields of digital photography, we constructed our system to monitor tissue-wide gene expression activities at a low cost. This method is particularly well-suited for time-lapse imaging of circadian clock activities. However, there are instances where high spatial resolution imaging of luciferase reporter expression is needed. For these purposes, adopting a technique from astrophotography known as “lucky imaging” could offer improvements [
23]. As discussed above, CMOS sensors are not advantageous for long exposures, but they can be used in lucky imaging to take multiple short exposure images, selecting and collating those that meet an SNR standard. However, time-lapse imaging requires the quantification of the signal intensity (which essentially means photon counting) over a specific exposure interval, making lucky imaging unsuitable for monitoring circadian oscillations. Nevertheless, we note that exploring methods from other fields holds great promise for advancing bioluminescence imaging techniques.