2.1.

Preparation of Coelenterazine Substrate

The substrate for Renilla- and Gaussia-based luciferases, coelenterazine (CTZ), is typically dissolved in nonpolar solvents such as ethanol or methanol. These solvents are not ideal for live cell imaging due to their inherent toxicity. We therefore recommend solubilizing CTZ in aqueous solution with the help of inert chemical agents such as $\beta$-cyclodextrin as described by Teranishi and Shimomura⁴¹ or by utilizing commercially available solvents (e.g., Fuel-Inject from Nanolight Technology). Solutions of CTZ can be made at any desired concentration, although we routinely use a stock concentration of $600\mu \mathrm{M}$ for in vitro use. After solubilization, CTZ can be aliquoted and stored at $-20°\mathrm{C}$ for several months. Since the amount of CTZ needed for in vivo applications is relatively higher, we freshly prepare CTZ before each use following the manufacturer’s recommendations. It is important to note that CTZ should be protected from light to prevent auto-oxidation.

2.2.

Cell Culture and Transfection

HEK293 cells were passaged regularly in Dulbecco’s Modified Eagle Medium (with 10% FBS, 1% penicillin/streptomycin) and seeded to 90% confluency on glass cover slips the day before transfection. HEK cells were transfected with expression vectors encoding membrane-localized Renilla luciferase¹⁶ and membrane-localized Gaussia luciferase protein¹⁸ using Lipofectamine 2000 (Invitrogen). Transgene expression was confirmed by fluorescence microscopy the following day.

Dissociated cortical neuron cultures were derived from E18 rat embryos. Cortical tissue was digested with $2\mathrm{mg}/\mathrm{mL}$ papain and dissociated by mechanical trituration before seeding onto 18-mm diameter German glass coverslips coated with $50\mu \mathrm{g}/\mathrm{mL}$ poly-d-lysine (Sigma). Neuronal cultures were grown in serum-free Neurobasal media ($\mathrm{w}/1\times$ B27 supplement, 0.5 mM glutamine) and media was changed (half-volume) every 3 to 4 days. Cortical neuron cultures were transduced with viral vectors encoding luciferase 2 to 3 days after plating.

2.3.

Live Cell Bioluminescence Imaging

Bioluminescence images were taken on an Olympus inverted fluorescence microscope equipped with a variety of objectives, c-mount adaptors, and cameras (Table 2).

Table 2

Objectives, adaptors, and cameras tested.

Cells cultured on glass coverslips were transferred to a perfusion chamber (Warner Instruments) containing phenol-free media at the time of imaging to minimize light absorption and maximize transmission of bioluminescence through the media (this is especially important for upright microscopes). The cells were then checked under fluorescence to confirm transgene expression and determine the right depth of focus for bioluminescence imaging. The microscope was then switched to an empty filter position to collect whole-spectrum bioluminescence.

Images were collected using the open-source Micromanager image acquisition software. Camera settings were standardized and optimized by cooling the chip to the lowest temperature (to minimize dark current) and maximizing the gain. Binning was used only when it was necessary to produce visible images. All of the background light sources in the room (windows, doors, and electronics) were covered with blackout material and images were collected at various exposure times (1 to 40 s). Long-term bioluminescence images were acquired using the multiacquisition feature of Micromanager and a perfusion system to deliver CTZ during the imaging period (up to 3 h). Imaging was done under controlled conditions that included using the same batch of transfected cells, CTZ concentration, media, exposure times, and imaging conditions to reduce variability between experiments. We did not attempt to normalize bioluminescence signals between different sets of experiments, although we employed the same camera settings (i.e., exposure time and binning) for a given combination of the microscope, camera, objective lens, and luciferase used. Thus, the bioluminescence signal was comparable within a given set of the experiments.

For the comparison of camera models in Fig. 3, bioluminescence imaging was repeated using an upright microscope (Eclipse E600-FN, Nikon) and a water-immersion objective lens ($40\times$, NA 0.8). CTZ was applied to the same batch of HEK cells transiently transfected with a membrane-bound Gaussia luciferase cultured on glass coverslips. Images were taken every 5 s with exposure time of 4.5 s and CTZ ($100\mu \mathrm{M}$) was bath-applied 50 s after the start of an imaging session. The camera models compared were Photometrics CoolSNAP ES [noncooled scientific charge-coupled device (CCD); digitization: 12 bit; gain $0.3\times$], Photometrics CoolSNAP fx (cooled scientific CCD; digitization: 12 bit; gain $2\times$), Hamamatsu ImagEM C9100-13 (EMCCD; digitization: 16 bit; EM gain: $1200\times$), and Andor iXon Ultra 897 (EMCCD; digitization: 16 bit; EM gain $1000\times$). Each camera was binned to roughly $256\times 256\text{pixels}$, resulting in a similar pixel size of $25.8\mu \mathrm{m}$ (CoolSNAP ES), $26.8\mu \mathrm{m}$ (CoolSNAP fx), $32\mu \mathrm{m}$ (ImagEM), and $32\mu \mathrm{m}$ (iXon Ultra 897).

Image quality was estimated by calculating the SNR of captured images at the peak luminescence time point. SNR was calculated by dividing the mean pixel intensity of the bioluminescence image to the standard deviation of the pixel intensity from a background image (or region with no cells) in ImageJ.

2.4.

In Vivo Bioluminescence Imaging

For imaging the mouse brain, an adeno-associated virus (AAV) encoding Renilla luciferase (RLuc) was injected into the cortex ($-1.58\mathrm{AP}$, $-0.75\mathrm{ML}$, $-0.8\mathrm{SI}$, and $-1.58\mathrm{AP}$, $-1.75\mathrm{ML}$, $-0.5\mathrm{SI}$) of 5- to 6-week old CD1 white mice (Harlan). After 2 weeks, the fur was shaved off the head and the animals were anesthetized with ketamine/xylazine before bioluminescence imaging. $500\mu \mathrm{g}$ of CTZ was administered either intraperitoneally or intravenously (via tail vein injection). Bioluminescence images were captured with the animals still under ketamine/xylazine anesthesia using a conventional luminescence imager (Fuji LAS-3000). Sequential images of 20 s exposure time were captured and the total signal intensity of each image was quantified in ImageJ to achieve a bioluminescence signal time course. Images at the time of maximum signal intensity are displayed showing the relative bioluminescence signal pseudocolored by a RGB lookup table in ImageJ. All procedures were performed in accordance with the US National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at Emory University.

3.1.

Live Cell Bioluminescence Imaging

Image brightness is directly related to the light-gathering power of the objective (numerical aperture, NA) and inversely related to the image magnification (M): brightness $\propto$ (NA/M).² We have therefore maximized the image brightness of our samples by optimizing three components of our microscope: the objective, camera, and intermediate optics.

In selecting an appropriate objective lens for bioluminescence imaging, we aimed at finding one with the highest NA and lowest workable magnification. In comparing various objective lenses, we found that objectives with higher NA produced brighter, higher resolution images [Figs. 1(a2) versus 1(a1)]. Lower magnification objectives were also able to produce higher resolution images [Figs. 1(b2) versus 1(b1)].

Fig. 1

Objective lens comparison. Bioluminescence images from cultured HEK cells expressing Gaussia luciferase (a1 and a2) and Renilla luciferase (b1 and b2) with the following objectives: (a1) LUCPlanFL $60\times$ 0.7 NA, (a2) LUMPlanFLN/W $60\times$ 1.0 NA for comparison of NA, (b1) UPlanFl $40\times$ oil 1.3NA, and (b2) UPlanSApo $20\times$ 0.75 NA for comparison of different magnification together with NA.

We utilized an intermediate demagnifying lens on a camera mount in order to allow more light to be focused onto a smaller area of the camera chip. Each pixel in the illuminated area of the chip therefore receives more light, producing a brighter image. We found that greater demagnification produced brighter images with sufficient spatial resolution to visualize detailed cellular morphology [Figs. 2(a) versus 2(b)]. Note that silhouetting at the periphery of the image may occur with greater demagnification, so the field of view requirements must be carefully considered.

Fig. 2

Intermediate lens comparison. Bioluminescence images of dissociated cortical neuron cultures expressing Renilla luciferase using a $0.3\times$ intermediate lens (a) and a $0.5\times$ intermediate lens (b). Both images were taken with an UAPO $40\times$ 1.35NA oil objective and a sCMOS (OptiMOS, Photometrics) camera.

There are numerous camera options currently available for low-light optical imaging. CCD cameras are the most popular choice and rely on the photoelectric effect to convert a light signal into an electrical signal. The readout noise of CCD devices was significantly reduced with the advent of electron multiplying charge-coupled devices (EMCCD), making these cameras especially well suited for low-light imaging applications. Scientific CMOS (sCMOS) devices are a relatively new type of sensor that also offers extremely low readout noise and wide dynamic range, making them a cost-effective alternative for imaging in low-light conditions. We have therefore compared several CCD, EMCCD, and sCMOS cameras for bioluminescence imaging (Fig. 3). All of the cameras tested were able to produce bioluminescence images with relatively low background and short exposure times (1 to 10 s). Even though EMCCD cameras generally outperform sCMOS devices at very low light levels, our images were qualitatively similar (most likely due to the fact that RLuc and GLuc are relatively bright luciferases). One should therefore select a camera based on the level of sensitivity required, as the price differences between EMCCD, sCMOS, and CCD cameras can be quite significant.

Fig. 3

Camera comparison. Bioluminescence images of HEK293 cells transfected with Gaussia luciferase taken with various cameras: (a) Noncooled scientific CCD (Coolsnap-ES, Photometrics); (b) cooled scientific CCD (Coolsnap-FX, Photometrics); (c): EMCCD (ImagEM C9100-13, Hamamatsu); (d): EMCCD (iXon Ultra 897, Andor). Contrast was adjusted for each camera using the highest and lowest intensity obtained in each camera. (e) Average SNR calculated at the time of peak luminescence for each camera.

Due to the relatively long-exposure times required for bioluminescence imaging, we found that it was important to reduce the amount of ambient light in the room as much as possible to reduce background noise. We routinely turned off or covered light sources (such as an arc lamp for fluorescence observation) near the microscope before bioluminescence imaging. We found that blackout curtains were especially effective at isolating the microscope and camera from any potential light contamination. Cooling the camera to the lowest temperature setting also helped reduce background by limiting dark current noise.

The danger of phototoxicity and photobleaching limits the effectiveness of long-term live-cell imaging with fluorescent molecules. In contrast, we have demonstrated that bioluminescent reporters can be used to image live cells for extended periods of time without any apparent adverse effects (Fig. 4). The substrate was supplied by a perfusion system, resulting in a long-lasting bioluminescent signal which could be detected over several hours. We did not observe any apparent adverse effects during this period, making this approach suitable for imaging cellular processes that are directed over long timescales, such as cellular trafficking, synaptogenesis, and migration.

Fig. 4

Long-term bioluminescence imaging in vitro. Time-lapse bioluminescence images illustrate the ability to conduct live cell imaging over extended periods of time without the threat of phototoxicity or photobleaching. Cortical neurons expressing Renilla luciferase were imaged over a period of 3 h. Note that the bioluminescent signal decreases over time (as substrate availability decreases) but is able to be refreshed again by the addition of more substrate. (Video 1, WMV, 1014 KB) [URL: http://dx.doi.org/10.1117/1.NPh.3.2.025001.1].

3.2.

In Vivo Bioluminescence Imaging

In vivo fluorescence imaging is often compromised by high nonspecific background from tissue and cells (auto-fluorescence). In contrast, background bioluminescence from tissues not expressing luciferase is negligible. This property of bioluminescence makes it an ideal signal for imaging whole animals where the number of luciferase-expressing target cells is generally few compared to the surrounding non-expressing tissue. We assessed the usability of Renilla luciferase for in vivo imaging and found that the bioluminescence signal was strong enough to allow detection of signals through the intact skull [Figs. 5(a) and 5(b)]. Special consideration of the route of substrate administration is needed when selecting a luciferase to use for in vivo bioluminescence imaging. Firefly luciferase (Fluc) has been frequently used for bioluminescence imaging in animals due to the relatively ease of the intraperitoneal route of substrate (D-luciferin) administration. Although we have demonstrated that coelenterazine can also be delivered intraperitoneally, Renilla luciferase is maximally effective when the substrate is administered via intravenous routes [Fig. 5(c)]. This may be due to the fact that the bioavailability of substrate in the brain is reduced when it is administered intraperitoneally, where it needs to drain through the lymphatic system before entering the bloodstream. Since beetle (e.g., firefly) and marine (Renilla, Gaussia) luciferases utilize different luciferin substrates, it is therefore feasible to multiplex them together due to absence of crosstalk between the two systems.

Fig. 5

In vivo bioluminescence imaging. Bioluminescence imaging of an adult mouse injected with AAV encoding Renilla luciferase in the cortex 2 weeks prior. $500\mu \mathrm{g}$ CTZ was delivered to the same mouse either intravenously (a) or intraperitoneally (b). Color bar depicts relative luminescent signal. (c) Bioluminescence images were taken over time and quantified in ImageJ to create a plot of bioluminescence time course.

The low background with bioluminescence makes in vivo bioluminescence imaging particularly sensitive for detecting signals from small areas over extended periods of time. In fact, bioluminescence imaging has been successfully used for applications from tracking transplanted cells^42,43 to monitoring cellular processes such as neurodegeneration, inflammation, and neurogenesis.⁴⁴ Although in vivo bioluminescence signals provide limited spatial information directly, the cell-type specificity of luciferase expression provides indirect spatial information because any detected bioluminescence should ostensibly be coming only from cells expressing luciferase.