Index Introduction Materials and Methods Nipkow disk confocal unit with mercury arc lamp illumination system Spatial resolution measurements Cell preparation Gene construction High-speed [Ca2+] imaging Multi-color imaging using a half-reflective mirror Results and Discussion Comparison of illumination at the ends of the optical fiber and spatial resolution using different light sources The power and uniformity of illumination Optical sectioning of biological samples High-speed confocal imaging with excitation illumination by mercury arc lamp Multi-color imaging with a half reflective mirror Acknowledgements References

Introduction

Laser-scanning confocal microscopy has been a method of choice for examining biological structures at the sub-cellular, cellular, and tissue levels in three dimensions for several decades, due to its ability to acquire images of thin optical sections within thick samples (Swedlow et al., 2004; Hibbs, 2004). However, a single-point scanning system usually takes several seconds to scan a single viewing field, which is a disadvantage for real-time imaging. Thus, multi-point scanning confocal systems using a scanning disk have been developed to achieve high-speed imaging without losing spatial resolution (Ichihara et al., 1996). The Yokogawa CSU10 is well known as one of the most advanced scanning disk systems. In this system, an expanded laser beam is directed onto a Nipkow disk, which contains arrays made up of tandem pairs of microlenses and pinholes, arranged circumferentially at a constant angle and radially at constantly decreasing distances from the center. When the disk is spun, each microlens/pinhole pair sweeps the excitation laser beam across the object through the objective lens, producing a raster scan of multiple laser beams. This scanning system provides two advantages for confocal imaging: 1) a faster frame acquisition rate (up to 360 Hz) because of the simultaneous scanning by approximately 1200 points, and 2) a reduction of photo-induced damage, because the excitation light power of each scanning point can be reduced to as low as 1/1200 of that of the single-point scanning system (Inoué and Inoué, 2002; Wang et al., 2005). There are some other commercially available disk-scanning systems with pinholes/slits, but they are different from the Yokogawa CSU10. The Olympus DSU employs a scanning disk with silts, which allows the acquisition of quasi-confocal images; however, there is no confocal effect along the horizontal aperture of the slit. The CARV II (BD Bioscience) has a Nipkow disk with pinholes, but the pinholes are larger (CARV II; 70 mm, CSU10; 50 mm), which reduces the confocal effect. Therefore, the axial resolution for both of these systems is lower than for the CSU10 (Toomre and Pawley, 2006).

As the excitation light source, visible wavelength lasers (Argon ion, Krypton-argon ion, Helium-neon ion, and other solid-state visible lasers) have been used for both single-point and multiple-point scanning confocal systems. However, the use of lasers often limits the choice of fluorescent dyes to be excited, because irradiated laser lights contain a relatively narrow range of wavelengths. Moreover, the use of coherent laser light with a Nipkow-disk confocal system generally results in considerable degradation of the image (Fewer et al., 1998). In contrast to a laser light source, a mercury arc lamp provides incoherent light with a much broader range of wavelengths, from 250 nm to 600 nm, and is commonly used as an excitation light source for conventional wide-field fluorescent microscopes. At a certain wavelength, the axial or lateral resolution of a confocal microscope is determined primarily by the size of the pinhole and the performance of the objective and tube lenses. Thus, confocal-based optical sectioning is possible, regardless of the excitation light source. Based on the idea that the mercury arc lamp provides a much broader range of wavelengths than a laser, illumination with a mercury arc lamp would be preferable for a Nipkow disk confocal system. In spite of this, there has been no report of applying a mercury arc lamp as the light source for the Yokogawa CSU system, possibly due to the difficulty of obtaining intense and uniform illumination of the specimen.

Here we report a method for multiple-point scanning confocal microscopy using a 100 W mercury arc lamp as the light source. In this system, a large diameter (~1 mm in diameter) multi-mode optical fiber is inserted between the arc lamp and the confocal system, providing stable, intense, and uniform illumination. The excitation wavelength is determined by filter selection, as in conventional wide field epi-fluorescence microscopy systems. This microscopy system enables the visualization of multiple-molecular dynamics in living cells without limiting the choice of fluorophores. Using this system, we show both high-speed (up to 100 Hz) and multiple color confocal imaging of dynamic biological events. In addition, we discuss other suitable applications for this new, convenient confocal system.

Materials and Methods

Nipkow disk confocal unit with mercury arc lamp illumination system

Fig. 1 shows a schematic diagram of the Nipkow disk confocal unit with a mercury arc lamp illumination system. The system consists of an inverted microscope (TE2000-E, Nikon, Tokyo, Japan) and a commercially available Nipkow disk confocal unit (CSU10, Yokogawa Electric, Tokyo, Japan). A 60×, or 100× PlanApo NA1.4 oil-immersion objective lens (Nikon, Tokyo, Japan) was used to collect the images. The light from a 100 W mercury arc lamp was focused onto the input end (1.0 mm in diameter) of a multi-mode optical fiber (ST1000H; Mitsubishi Electric Cable, Tokyo, Japan) to scramble the light (Kam et al., 1993; Tani et al., 2005). The light from the fiber’s output end was then directed to the CSU10. A MAC 5000 system (Ludl, Hawthorne, NY) was used to control the filter wheels for the excitation and emission light and for shuttering the excitation light. The emitted light collected by the objective lens was focused on a cooled electron-multiplying charge-coupled device (EM-CCD) camera (ImagEM; Hamamatsu Photonics, Hamamatsu, Japan) or a cooled 3CCD camera (ORCA-3CCD; Hamamatsu Photonics). The shutter, filter wheels, and camera were controlled by the software, AquaCosmos version 2.6 (Hamamatsu Photonics).

View Details

Fig. 1. Nipkow disk confocal unit with mercury arc lamp illumination system. (A) A schematic diagram of the Nipkow disk confocal unit with mercury arc lamp illumination system. ML: 100 W mercury arc lamp, FW1: Filter wheel for excitation, L: collection lens, MMF: multi-mode optical fiber, BS: beam splitter, CSU: Yokogawa CSU10, MLD: microlens array Nipkow disk, PHD: pinhole array Nipkow disk, OBJ: objective lens, CS: cover slip and sample, FW2: filter wheel for emission, CCD: cooled CCD camera. (B) Photograph of the Nipkow disk confocal unit with mercury arc lamp illumination system.

Spatial resolution measurements

To compare the spatial resolution of a confocal system using a mercury arc lamp with that using an Ar⁺ laser, 0.1 μm-diameter fluorescent beads (TetraSpeck; Molecular Probes, Eugene, OR) were used. Excitation light through a mercury arc lamp with a 470/40 nm excitation filter or a 488-nm Ar⁺ laser (20 mW; model 532-BS-A04; Melles Griot, Carlsbad, CA) was used to collect images of the beads. Six images obtained using each excitation light were averaged and compared.

Cell preparation

HeLa cells were cultured in homemade 35-mm glass-bottomed dishes in Dulbecco’s modified Eagle’s medium (DMEM; Sigma, St. Louis MO) containing 10% fetal bovine serum (BioWest, Nuaillé, France). The cells were transfected with plasmids using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). The transfected cells were cultured for 1 to 2 days before observation.

Gene construction

The DNA sequence encoding the N-terminal 12 amino acids of cytochrome c oxidase subunit IV (CoxIV) was fused to the sequence encoding the N-terminus of mSECFP (Matsuda et al., 2008), to obtain mSECFP-mit, which specifically localizes to the mitochondrial matrix. The nuclear localization signal sequence (NLS) of SV40 large T-antigen was fused to the C-terminus of mRFP1 (Campbell et al., 2002) to obtain mRFP1-nu, which localizes to the nucleus. mSECFP-er was generated by extending the N-terminus of mSECFP with the signal peptide sequence of calreticulin and the C terminus of mSECFP with the ER retention signal sequence KDEL, which localizes to the endoplasmic reticulum. The sequences encoding mSECFP, mSECFP-mit, mRFP1-nu, and mSECFP-er were cloned into the pcDNA3 vector (Invitrogen, Carlsbad, CA). Venus-Actin was generated by replacing the EYFP gene of pEYFP-Actin (Clontech Laboratories, Palo Alto, CA) with the Venus gene (Nagai et al., 2002). The DsRed1 gene of pPKC-γ-DsRed1 (Clontech Laboratories) was replaced with the gene for its rapidly maturating variant, DsRed T3 (Bevis and Benjamin, 2002) to construct PKC-γ-DsRed T3. The construction of yellow cameleon 3.60 (YC3.60) was reported previously (Nagai et al., 2004).

High-speed [Ca²⁺] imaging

HeLa cells were incubated in DMEM containing 5 μM Oregon Green-488 BAPTA-1 AM (Molecular Probes, Eugene, OR) for 1 hour at room temperature, washed twice with PBS, and transferred to DMEM again before observation. A 470/40 nm filter (Nikon, Tokyo, Japan) was used as the excitation filter. To show the spatial propagation of the Ca²⁺ wave clearly, the obtained images were smoothed by taking the rolling average over 4 images. From the series of fluorescent images, the relative change in the intracellular Ca²⁺ was calculated from the fluorescence intensity value (F) of the cell divided by its value at time 0 (F₀). The obtained changes in F/F₀ were shown as a series of pseudo-color images.

Multi-color imaging using a half-reflective mirror

For multi-color imaging, we used a half-reflective mirror, T50 (Asahi-spectra Co., Tokyo, Japan), that transmits approximately 50% of visible wavelengths from 400 nm to 700 nm. We used the following combination of excitation and emission filters, respectively: 440/21 nm and 480/30 nm for mSECFP, 490/20 nm and 535/26 nm for Venus, and 540/30 nm and 575 nm long path filter for mRFP1 (all filters were from Omega Optical and are designated by the central wavelengths of their transmittance and band widths at half-maximal transmittance). For the multi-color time-lapse imaging of HeLa cells expressing both YC3.60 and PKC-γ-DsRed T3, we used 440/21 nm and 540/30 nm for YC3.60 and DsRed T3 excitation, respectively, and 480/30 nm, 535/26 nm, and 575 nm long pass filter for mSECFP, Venus, and DsRed T3 emission, respectively. All of the excitation and emission filters were automatically alternated by filter wheels, using the MAC 5000 system (Ludl, Hawthorne, NY).

Results and Discussion

Comparison of illumination at the ends of the optical fiber and spatial resolution using different light sources

A Nipkow disk confocal microscope requires an illumination system that provides uniform excitation light for each scanning point. To obtain uniform illumination from a mercury arc lamp, we introduced a multi-mode, large diameter (1 mm) optical fiber into the illumination system as a light scrambler (Fig. 1A). Typical images of the output end of the multi-mode optical fiber coupled to a mercury arc lamp and that of a single-mode optical fiber coupled to a 488-nm Ar⁺ laser are shown in Fig. 2A and B, respectively. The intensity profile of the image obtained from the multi-mode fiber output is shown in Fig. 2C. The light was uniformly distributed, and did not reflect the image of the mercury arc’s geometry.

View Details

Fig. 2. Comparison of the efficiency of the CSU10 with mercury arc lamp versus 488-nm Ar+ laser. (A and B) Images of the optical fiber output end coupled to a mercury arc lamp (A) or 488-nm Ar+ laser (B). (C) The intensity profiles across the center of the images in A (red line) and B (green line), normalized to the peak intensity. (D, E) The spatial resolution of the CSU10 with mercury arc lamp (red line) and 488-nm Ar+ laser (green line). (D) Horizontal intensity profiles across just the middle section of stacks of a 0.1-μm fluorescent bead. (E) Vertical intensity profiles across the center of a 0.1-μm fluorescent bead.

The excitation light from the output end of a multi-mode or single-mode fiber end was then introduced into the CSU10 Nipkow disk-type confocal scanner unit. To compare quantitatively the spatial resolution of the system using a mercury arc lamp with that using a 488-nm Ar⁺ laser line, we collected three-dimensional images of a 0.1-μm fluorescent bead. The intensity profiles of the images along the X-axis and Z-axis are shown in Fig. 2D and E, respectively. The full width at half-maximal intensity (FWHM) of the images in the xy plane was 0.27 μm for the mercury arc lamp and 0.24 μm for the 488-nm Ar⁺ laser (Fig. 2D). The FWHM of the bead image on the xz plane was 0.76 μm for the mercury arc lamp and 0.72 μm for the laser light (Fig. 2E). These results indicate no significant differences in the point-spread function on the xy and xz planes between a CSU10 disk illuminated with a mercury arc lamp versus a 488-nm Ar⁺ laser. Thus, the mercury arc lamp illumination was equivalent to the 488-nm Ar⁺ laser for acquiring confocal images through the CSU10.

The power and uniformity of illumination

We measured the power of the excitation light of the mercury arc lamp at the input and output ends of the multi-mode optical fiber and at the level of the specimen. With a 470/40 nm band-pass filter, the power at these levels was 60, 8, and 0.05 mW, and with a 540/30 nm band-pass filter, it was 150, 30, and 0.19 mW, respectively. The power of the excitation light with the 470/40 nm filter was lower than that with the 540/30 nm filter, because the mercury arc lamp has prominent emission lines in 540/30 nm but not in 470/40 nm. The applicability of excitation light with no prominent emission line to high-speed imaging will be examined later in this article.

To examine the uniformity of illumination on the specimen, we obtained an image of a fluorescent plate filled with uniformly distributed fluorescent dyes. With both the mercury arc lamp (Fig. 3A) and the 488-nm Ar⁺ laser light (Fig. 3B), the peak of intensity was at the center of the image, and about a 20% decrease in signal was seen at the periphery of the image (Fig. 3C, D). In spite of the large difference in the intensity profile at the end of the optical fiber (Fig. 2A–C), there was no significant difference in the uniformity of illumination on the specimen between the mercury arc lamp and the 488-nm Ar⁺ laser.

View Details

Fig. 3. The uniformity of illumination on a sample. (A and B) The images of a fluorescent plate obtained using the CSU10 with illumination from a mercury arc lamp (A) or 488-nm Ar+ laser (B). (C and D) The normalized intensity profiles across the center of the images in A (C) and B (D).

Optical sectioning of biological samples

For further demonstration of the use of the mercury arc lamp as an ideal light source for CSU10, we took confocal fluorescent images of pumpkin pollen grains (diameter, ~100 μm), a useful sample for testing the effect of optical sectioning by confocal systems using a mercury arc lamp (Fig. 4A) and 488-nm Ar⁺ laser (Fig. 4B). As a comparison, the same image was taken with a wide-field microscope (Fig. 4C). There was no significant difference between the images obtained using the mercury arc lamp versus the laser (Fig. 4A, B) whereas in the image taken by wide-field microscopy, the inside structure of the pumpkin pollen grain was unclear and the background fluorescence was higher (Fig. 4C) than in the confocal images. Thus, it was fully expected that optical sections of even a single living HeLa cell (~10 μm thickness) might be obtained with the CSU10 with mercury arc lamp system. To verify this, we visualized mSECFP localized to specific organelles in HeLa cells using the CSU10 with a mercury arc lamp as a light source. The condensed structures (vesicles, nucleolus) inside the cell contrasted well with the mSECFP in the cytoplasm and the nucleus (Fig. 4D), and the network structure of the endoplasmic reticulum (Fig. 4E), and the tubule structure of the mitochondria (Fig. 4F) were clearly seen. In wide-field microscopy, these structures cannot be seen clearly because of high background fluorescence from the out-of-focus planes.

View Details

Fig. 4. Optical sectioning of biological samples. (A–C) Real-color images of confocal or non-confocal optical sections of a pumpkin pollen grain. These images were obtained by the CSU10 with a mercury arc lamp (A) or 488-nm Ar+ laser (B) or by non-confocal (wide field) fluorescence microscopy (C). Scale bar represents 20 μm. (D–F) Images of HeLa cells expressing mSECFP in the cytoplasm and the nucleus (D), mSECFP in the endoplasmic reticulum (E), and mSECFP in the mitochondria (F). Scale bar represents 10 μm.

High-speed confocal imaging with excitation illumination by mercury arc lamp

The mercury arc lamp has several prominent emission lines, at 366, 405, 435, 546, and 578 nm, with peak intensities up to 6–10 times higher than the average intensity of the residual emission wavelength. These prominent emission lines can be used to easily excite fluorescent dyes such as DAPI, CFP, DsRed, and mRFP1 effectively. However, there are no prominent emission lines between 450 nm and 500 nm, which would be applicable to the excitation of various useful fluorescent probes, such as organic compound dyes, Alexa 488, the calcium ion indicator Oregon Green-488 BAPTA-1 (OGB), and the fluorescent proteins EGFP and EYFP. To test the applicability of the CSU10 system with a mercury arc lamp to the fluorescent confocal imaging of dyes that are excited by light from 450 nm to 500 nm, we examined the intracellular Ca²⁺ propagation in HeLa cells using OGB, with a sampling rate of 100 Hz. The fluorescence of OGB loaded into HeLa cells was uniformly distributed in the cytoplasm and the nucleus. After the application of 10 μM histamine (~6 sec after the start of the observation), an increase in Ca²⁺ concentration occurred at the cell periphery that rapidly propagated to the center of the cell (Fig. 5A). The time course of the change in F/F₀ ratio measured in the area enclosed by a white rectangle in Fig. 5B shows an oscillation of intracellular Ca²⁺ concentration with a frequency of approximately 0.1 Hz. This result showed that even at a wavelength of around 500 nm, where no prominent emission lines exist, the CSU10 with a mercury arc lamp could be used to visualize spatial and temporal intracellular Ca²⁺ changes with a time resolution of up to 100 Hz, the same resolution obtained with the CSU10 and a laser light source (Genka et al., 1999).

View Details

Fig. 5. High-speed confocal Ca2+ imaging by the CSU10 with mercury arc lamp. (A) A series of pseudo-colored images of HeLa cells loaded with Oregon Green-488 BAPTA-1 AM. Scale bar represents 10 μm. (B) Time course of the F/F0 ratio value at the ROI indicated in (A).

Multi-color imaging with a half reflective mirror

The wider range of excitation wavelengths emitted by the mercury arc lamp (from 400 nm to 600 nm) allows the visualization of multiple fluorescent dyes. For multi-color imaging with the CSU10, a special interference mirror with multiple transference/reflection peaks (multi-chroic mirror) has been used (Fig. 6A). Since the multi-chroic mirror is optimized only for cyan, yellow, and red fluorescent dyes, it could limit the advantage of the mercury arc lamp system in providing a broad range of wavelengths. Therefore, to achieve multi-color imaging with the CSU10 with a mercury arc lamp, we introduced a half-reflective mirror instead of the multi-chroic mirror. The transmittance spectrum of the half-reflective mirror is shown in Fig. 6B. Transmission was kept constant at approximately 50% in the 400–700 nm range of the spectrum. By using multiple combinations of excitation and emission filters, multiple images of different colors could be obtained. To compare the transmission/reflection characteristics of the two mirrors described above, the amount of light transmitted/reflected by the mirrors was expressed as areas filled with blue, yellow, and red, if the incident light was an ideal white light (i.e., if the light intensity-wavelength profile were flat). Approximately 50% of the transmitted light through the excitation filter was available for the excitation, and approximately 50% of the collected light through the objective lens could be used for signal detection. The amounts of excitation and emission light in the system using the half reflective mirror were slightly lower than obtained using the multi-chroic mirror. However, the half-reflective mirror should be a powerful choice if one needs to use various combinations of fluorescent dyes for multi-color imaging. To demonstrate the validity of using the half-reflective mirror, we transfected HeLa cells with the genes for chimeric fluorescent proteins designed to stain different subcellular structures: mitochondria (mSECFP-mit), actin (Venus-Actin), and the nucleus (mRFP1-nu). The HeLa cells were then observed using the CSU10 system with a mercury arc lamp and a half-reflective mirror (Fig. 6C). The three emission signals from the different fluorescent proteins were precisely separated from each other and merged. As shown in Fig. 6C, the three images were well aligned spatially.

View Details

Fig. 6. Multi-color imaging by the CSU10 with a half-reflective mirror. (A) Transmittance of a multi-chroic mirror optimized for three fluorescent proteins: ECFP, EYFP, and mRFP1. (B) Transmittance of the half-reflective mirror, which has about 50% transmittance in the 400–700 nm wavelength range. Areas filled with blue, yellow, or red indicate the amount of light transmitted/reflected by the mirrors and excitation filters and emission filters for ECFP, EYFP, and mRFP1. (C) Images of HeLa cells co-expressing mSECFP-mit, Venus-Actin, and mRFP1-nu. The merged image shows all three channels: green (mSECFP-mit), red (Venus-Actin), and blue (mRFP1-nu). (D) A series of pseudo-colored images of HeLa cells co-expressing YC3.60 (upper panel) and a series of fluorescent images of PKC-γ-DsRed T3 (lower panel). Scale bars in C and D represent 10 μm.

We next demonstrated time-lapse imaging of three different colors using the same microscopy system. To monitor the translocation of protein kinase C gamma (PKC-γ) and its upstream Ca²⁺ signal in a living cell, DsRed T3 fused with PKC-γ (PKC-γ-DsRed T3) and yellow cameleon 3.60 (YC3.60) were co-expressed in HeLa cells. YC3.60 is a genetically encoded Ca²⁺ indicator that changes the ratio of YFP and CFP intensity by nearly 600%, based on FRET technology (Nagai et al., 2004). PKC-γ is an isozyme of conventional PKC (cPKC) and has functional C1 and C2 domains in its regulatory region, which bind diacylglycerol and Ca²⁺, respectively (Nishikawa et al., 1997). Ca²⁺-mediated PKC-γ activation and translocation to the plasma membrane are well studied (Sakai et al., 1997). Fig. 6D shows a series of pseudo-colored images of HeLa cells expressing YC3.60 (upper panel) and PKC-γ-DsRed T3 (lower panel). The PKC-γ translocation to the plasma membrane was precisely synchronized with the increase in cytoplasmic Ca²⁺ concentration after the application of 10 μM histamine. This result agrees with the previously reported Ca²⁺-controlled transient membrane association of PKC (Violin et al., 2003).

In this paper, we demonstrated the usefulness of multiple-point scanning confocal microscopy using a 100 W mercury arc lamp as a light source. In our new system, the coupling efficiency between the multi-mode optical fiber and the CSU10 was about 0.01, which is lower than that of CARV II (0.05–0.07), another Nipkow disk-scanning confocal system (Toomre and Pawley, 2006). In spite of the low coupling efficiency, the power of the excitation light with no prominent emission lines of the mercury arc lamp was adequate for high-speed Ca²⁺ imaging at 100 Hz (Fig. 5). In addition, using CSU10 with a mercury arc lamp and a half-reflective mirror may solve the problems associated with multi-color imaging using a laser light source, such as the limitation of fluorophore choice. Many variants of fluorescent proteins have been developed and are available (Giepmans et al., 2006). In particular, red to far-red emitting bright fluorescent proteins have emerged (Merzlyak et al., 2007; Shcherbo et al., 2007). In multi-color imaging, these newly developed red fluorescent proteins are anticipated to be good FRET partners for conventional fluorescent proteins, such as ECFP, EGFP, and EYFP. When we use fluorescent protein variants with longer emission wavelengths, we find that a Xenon arc lamp is superior to the mercury arc lamp, owing to the incident wavelengths expanding to the far-red region. In such cases, a Xenon arc lamp can easily replace the mercury arc lamp as the light source in our microscopy system. Thus, this microscopy system should make multi-color imaging readily available to many researchers.

Acknowledgements

We are grateful to R.Y Tsien for the gift of the mRFP1 plasmid, and we would like to thank M. Shimizu and H. Hirukawa (Yokogawa Electric Corporation) and M. Mizuta and K. Toshimitsu (Nikon Corporation) for technical support and advice. This work was partly supported by Grants from Scientific Research on Advanced Medical Technology of the Ministry of Labor, Health and Welfare of Japan, and the Japanese Ministry of Education, Science and Technology.