Bioluminescence Microscopy: Design and Applications Bioluminescence Microscopy: Design and Applications

Bioluminescence imaging by microscopy is performed using an ultra-low-light imaging camera. Although imaging devices such as sensor and camera have been greatly improved over time, such improvements have not been attained commercially which are available for microscopes now. We previously optimized the optical system of a microscope for bioluminescence imaging using a short-focal-length imaging lens and evaluated this system with a conventional color charge-coupled device camera. Here, we describe the concept of bioluminescence microscope design using a short-focal- length imaging lens and some representative applications, including intracellular calcium imaging, imaging of clock gene promoter assays, and three-dimensional reconstruction of Drosophila larva. This system facilitates the acquisition of bioluminescence images of single live cells using luciferase, which is similar to fluorescence microscopy using a fluorescent protein.


Introduction
Aequorin is a calcium-specific light-emitting protein extracted from the jelly fish Aequorea [1]. Additionally, aequorin-injected eggs of the medaka (Oryzias latipes), a fresh water fish, showed a dramatic increase in free calcium during fertilization, as determined by measuring light from the eggs using a photomultiplier tube [2]. Notably, bioluminescence microscopy with an image intensifying system using a vidicon camera was performed in 1978 to show the spatial distri-bution of the free calcium in the egg [3]. This system revealed a spreading wave of high free calcium (calcium wave) during fertilization from the animal pole, as discussed with the fertilization wave of cortical changes in eggs observed by light microscopy [4]. Although the potential of low-light imaging has been recognized in physiology and developmental biology, this technique was not commonly used at that time due to a lack of commercially available instrumentation. Later, advances in detector and digital imaging processing systems facilitated the commercial production of appropriate instrumentation and made it possible for low-light imaging to be carried out using a silicon-intensifier target (SIT) tube camera or a high-sensitivity cooled charge-coupled device (CCD) camera.
Since the cloning of firefly luciferase in the late 1980s, luciferase has been used as a reporter enzyme to assay the activity of a particular gene promoter using the photon-counting luminometer method [5][6][7]. Additionally, bioluminescence microscopy of promoter activity in single cells has been performed using ultra-low-light imaging cameras, such as liquid nitrogencooled CCD cameras, photon-counting CCD cameras, or image-intensifying CCD cameras [8][9][10][11][12][13][14][15][16]. However, temporal and spatial resolution was not enough for the observation of cellular biological events and for the detection of single cells compared with that of conventional CCD cameras. Therefore, satisfactory analysis has not been achieved at the single-cell level by bioluminescence microscopy.
Recently, electron-multiplying CCD (EM-CCD) camera, which yields higher sensitivity and image quality, was commercially released and subsequently used for bioluminescence microscopy [17][18][19]. Although the image sensor of ultra-low-light imaging cameras has been greatly improved over time, such improvements have not been made commercially available for microscopes.

Microscope design
Bioluminescence microscopy is based on the detection of light emitted by living cells expressing a luciferase gene or other luminescence-related gene. Conventional microscopes are inefficient at transmitting light from the sample to the detector, necessitating long exposure times. We designed a new type of microscope for ultra-low-light imaging based on modifications to the imaging lens, vignetting, and effective field area. Figure 1 shows a diagram of an inverted microscope equipped with an infinity-corrected optical system. Light from a sample is collected by an objective lens (OB), and the sample image is created by an imaging lens (tube lens; IM) on a CCD chip. Generally, the degree of brightness (I) of an image is directly proportional to the square of the numerical aperture (NA) of the OB and is inversely proportional to the square of magnification (M) of the image, namely as I ∝ (NA/M) 2 . Therefore, a higher NA and lower M yield much brighter images. However, it is difficult to obtain both conditions. Because higher NA OB yields higher M (shorter focal length), or lower M OB yields lower NA. Thus, high NA and low M are mutual trade-offs. On the other hand, the value of NA/M is the same as the NA of the IM, geometrically denoted as NA′. Therefore, a microscope with a high NA′ (short-focal-length imaging lens) makes it possible to achieve a higher NA and lower M without further improvement of the objective lens. This was the concept on which we based the design of the bioluminescence microscope [20,21].  shows one example of the I condition to capture bioluminescence images of live cells using a microscope. The X-axis indicates the I value of the in vivo macro-imaging system, OV100 (Olympus, Tokyo, Japan), using a 0.8× objective lens with an NA varying from 0.05 to 0.25 (I = 0.004-0.098). HeLa cells transiently expressing the Luc+ luciferase gene (pGL3 control vector; Promega, Madison, WI, USA) in Hanks' balanced salt solution (HBSS; Invitrogen, Carlsbad, CA, USA) containing 1 mM D-luciferin, potassium salt (Promega) at room temperature were imaged using a CCD camera (ST-7; SBIG, Ottawa, Canada) for astronomical imaging. The exposure time was 1 min, and the cooling temperature was −20°C. The Y-axis indicates the normalized luminescence intensity of the entire area of the image captured ( Figure 2A). As shown in the graph (Figure 2B), luminescence images could be captured at I values of greater than 0.02, although the M was lower for single-cell imaging. Therefore, we designed an IM to achieve an I value of 0.02 with a higher M using a conventional OB.  Figure 3 shows diagrams of light passing from an object in an infinity-corrected optical system. In this system, light from object runs parallel between the OB and IM (Figure 3A and B). Therefore, this system is suitable for several observations because several optical elements (such as mirror units for fluorescence observations or polarizing filters) can be inserted between the OB and IM without light-pass correction for image formation. However, light from peripheral vision (ray "a") is vignetted by the imaging lens, when the distance between the OB and IM becomes longer ( Figure 3C). Therefore, vignetting can be avoided by shortening the distance between the OB and IM.  Figure 4 shows an effective field diagram on a CCD chip. Generally, the area of the CCD chip is smaller than the effective field area because peripheral vision is affected by several optical aberrations. Images on the CCD chip only show part of the light collected by the OB. If all light is collected on the CCD chip as an image by reducing the magnification using an intermediate tube lens or modified IM, the light intensity of each pixel becomes greater. Thus, image quality is sacrificed to obtain brighter images.  Figure 5 shows the inverted bioluminescence microscope used in our studies (Luminoview LV200; Olympus, Tokyo, Japan). A halogen lamp was used as the source of transmitted brightfield light. The light was directed to a sample through a condenser lens with a glass fiber. A short focal-length imaging lens (f = 36 mm, NA = 0.2) was customized based on the condition of I to capture dim bioluminescence images in this system. Normal OBs are available for observation. Using the IM, total magnification was reduced to one-fifth of the magnification of the OB because the focal distance of the IM is fixed (180 mm; Olympus) in a conventional microscope body. The distance between the OB and IM was set at 17 mm to avoid vignetting. A stage-top incubator with temperature and CO 2 gas controllers (MI-IBC-IF; Tokai Hit Co., Shizuoka, Japan) was added to the sample stage. The observation area was covered with a dark box [21,44].

The bioluminescence microscope, LV200
To evaluate the performance of LV200, bioluminescence images of U2OS cell lines stably expressing CBG99, CBR, and Luc2 beetle luciferase (Promega) were captured using an UPlanFLN 40× oil objective lens (NA = 1.30, I = 0.026) and DP70 color CCD camera (Olympus). Cells were cultured on 35-mm glass-bottomed dishes in Dulbecco's modified Eagle's medium (DMEM; Invitrogen) containing 10% fetal bovine serum and 1 mM beetle D-luciferin at 37°C with 5% CO 2 . Binning of the CCD camera was 1 × 1 (1360 × 1024 pixels), International Organization for Standardization gain was 1600, and exposure time was 2 min. Figure 6A shows the bioluminescence images of cells expressing CBG99, CBR, and Luc2, captured within 2 min using a conventional color CCD camera as green, red, and orange color, respectively. Notably, bioluminescence images could not be captured under the same conditions (stable cell lines, OB, CCD camera, and exposure time) using a conventional inverted microscope (IX70; Olympus; NA = 1.30, I = 0.001), although a 10-min exposure time was required to obtain images for the beetle luciferase-expressing cell line (Figure 6B) [21]. Despite the use of blank image subtraction, 10 min is the upper limit of exposure time for the DP70 color CCD camera due to intense background evaluation [21]. Thus, bioluminescence images of cells expressing the luciferase gene can be captured using an LV200 microscope with a 40× OB and color CCD camera. In this case, the M of the image was reduced by a power of 8 owing to the short-focallength IM, and the I value was 0.026. To equalize the I value between the LV200 and IX70 microscopes, a low M and high NA OB (e.g., 8×, NA 1.3) is required for IX70. However, an OB with such a high NA cannot be purchased commercially. To show the spatial resolution of the bioluminescence images acquired using LV200, organelletargeted images were captured using an UPlanFLN 100× oil objective lens (Olympus) and ImagEM EM-CCD camera (C9100-13; Hamamatsu Photonics, Shizuoka, Japan). The binning of the EM-CCD camera was 1 × 1 (512 × 512 pixels), EM-gain was 1024, and exposure time was 300 ms to 1 s. NanoLuc luciferase (Promega), which is 150-fold brighter than beetle luciferase [37], was used as a tag for organelle localization, similar to a fluorescent protein. Before substrate addition (12.5 μM furimazine; Promega), cells were washed with culture medium three times.    (Figure 7A), mitochondrial targeting sequence (subunit VIII of human cytochrome C oxidase, CoxVIII) (Figure 7B), endoplasmic reticulum resident protein, calreticulin with KDEL retrieval sequence (Figure 7C), or no targeting sequence (Figure 7D) in U2OS cells [21]. The NanoLuc-NLS accumulated in the nucleus of the cell, and the CoxVIII-NanoLuc and calreticulin-NanoLuc-KDEL appeared in a meshwork pattern in the cytoplasm. Thus, the nucleus and cytoplasm were discriminated clearly, and mitochondria and endoplasmic reticulum were recognized in the cytoplasm.

Applications
As examples of bioluminescence microscopy using our system (LV200), we introduce three applications: (1) calcium imaging of single cells, (2) imaging of clock gene promoter assays, and (3) three-dimensional imaging of Drosophila larva.

Intracellular Ca 2+ imaging using obelin
Obelin is a calcium-specific bioluminescent protein similar to aequorin; using obelin, intracellular calcium was imaged by ATP and ionomycine (A23187) stimulation for calcium release from intracellular membranes (mitochondria and endoplasmic reticulum) and inflow from outside of the cell, respectively. The apoobelin gene [45] was inserted into the mammalian expression vector, pCDNA3.1 (Invitrogen), and transfected into HeLa cells. HeLa cells transiently expressing apoobelin were incubated in DMEM containing 60 μM coelenterazine (Promega) for 4 h to reconstitute obelin. The cells were stimulated with 500 μM ATP, and bioluminescence images were captured using an LV200 microscope with an UplanApo 20× OB (NA = 0.70; Olympus) and an iXon EM-CCD camera (DU-897l; Andor Technology, Belfast, UK). Binning of the CCD was 2 × 2, EM gain was maximum, and the exposure time was 25 s with a 30-s interval. The cells were restimulated by 10 mM ionomycine at 20 min after ATP stimulation.   Figure 8A shows pseudocolor-coded bioluminescence images of intracellular calcium in HeLa cells at 8 min after ionomysin stimulation. Figure 8B shows a time course of the intracellular calcium response for 10 single cells using time-laps image analysis software TiLIA [46]. Calcium responses in each cell varied temporally, were broad in intensity at around 15 min after ATP stimulation, and were uniform and greater in intensity after ionomysin stimulation [47,48]. Using this imaging system, ATP-induced calcium oscillation in HEK-293 cells was confirmed using a bioluminescent calcium sensor constructed by aequorin and GFP with 1 s exposure time using a bioluminescence resonance energy transfer (BRET) system [36].

Imaging of clock gene promoter assays
The circadian rhythm is monitored by measuring the promoter activity of clock genes from individual cells as a cellular clock. However, it is impossible to resolve whether loss of circadian rhythm following stimulation is caused by dis-periodicity or dis-synchronicity in individual cells using a luminometer because the luminometer captures total luminescence from the cell population. Bioluminescence microscopy can provide clear single-cell analyses of promoter activity.  The promoter region of the Per2 clock gene in mice was inserted into the luciferase promoter vector, pGL3 (Promega), and the vector was transfected into NIH3T3 cells. Cells were cultured in CO 2 -independent DMEM (Invitrogen) containing 500 μM D-luciferin. Bioluminescence images were captured using an LV200 microscope with an UPlanApo 20× OB (NA = 0.70) and DP30BW CCD camera (Olympus) at 37°C. The binning of the CCD was 1 × 1 (1024 × 1024 pixels), the cooling temperature of the sensor chip was 5°C, and the exposure time was 5 min with 30-min interval for 28 h. As shown in Figure 9A, bioluminescence images of single cells expressing the Per2 gene were captured clearly using a conventional CCD camera. Figure 9B shows a time course of Per2 promoter activity in 10 cells selected appropriately for 48 h using TiLIA [46]; this time course allowed us to analyze synchronicity among cells [49]. Using this imaging system, Ukai et al. [23] produced photoresponsive mammalian cells by introducing the photoreceptor melanopsin and monitored the effects of photoperturbation on the state of the cellular clock. They observed that a critical light pulse drove cellular clocks into singularity behavior and proved that loss of the circadian rhythm of a cellular clock may be caused by dissynchronization of individual cells underlying singularity behavior by single-cell analysis.

Three-dimensional imaging of Drosophila larva
Because our bioluminescence microscope system utilizes a short-focal-length imaging lens, the magnification is lower and the focal depth is shallower than those of conventional microscopy systems using the same OB. Therefore, depth of field is also shallower. This is convenient for three-dimensional image reconstruction by light sectioning. Accordingly, we constructed transgenic Drosophila melanogaster carrying an armadillo (a member of the segment polarity gene) promoter and luciferase fusion gene [31]. Figure 10 shows three-dimensional images of armadillo promoter activity from insular larva of transgenic flies reconstructed from 11 sectionalized images (front and slant views). The larva was immersed in 3 mM D-luciferin for 5 min before image acquisition. Bioluminescence images of the anesthetized larva with cold treatment were captured using an LV200 microscope with an UPlanFLN 60× OB (NA = 0.90) and iXon EM-CCD camera. The binning of the CCD was 1 × 1; EM gain was maximum; and exposure time was 30 s. Eleven sectionalized images were obtained with 10-μm steps from top to bottom (0-100 μm) of the larva. After deconvolution, three-dimensional images were reconstructed using CelSens Dimension image analysis software (Olympus). As shown in the Figure 10, the expression of the armadillo was observed in the midgut from a tissue depth of 100 μm, although it needed clearing treatment of kidney tissue for imaging of 100-200 mm depth by confocal fluorescence microscopy [50].

Conclusion
In this study, we presented the concept of bioluminescence microscopy using a short-focallength IM. This system facilitates the acquisition of bioluminescence images of single live cells using luciferase, similar to fluorescence microscopy using a fluorescent protein, although M is lower than that of conventional microscopy. Furthermore, this method is applicable for studies of cellular activity at the single cell level, including analyses of signal transduction, gene expression, and embryogenesis.
As bioluminescence microscopy requires no excitation light, it leads to substantive differences from fluorescence microscopy. Bioluminescence observation lacks the phototoxicity and background autofluorescence problems associated with fluorescence observation and permits the long-term, non-lethal observation of living cells such as embryonic stem cells, iPS cells, and embryos. Table 1 summarized the substantive differences between the fluorescence and bioluminescence microscopy. Thus, bioluminescence microscopy is a powerful tool in cellular biology and complements fluorescence microscopy.