Oblique plane microscopy (OPM) is a light sheet microscopy technique that uses a single high numerical aperture microscope objective to both illuminate a tilted plane within the specimen and to obtain an image of the tilted illuminated plane. In this paper, we present a new OPM configuration that enables both the illumination and detection focal planes to be swept simultaneously and remotely through the sample volume, enabling high speed volumetric imaging. We demonstrate the high speed imaging capabilities of the system by imaging calcium dynamics in cardiac myocytes in 2D at 926 frames per second and in 3D at 21 volumes per second. In the future, higher frame rate CCD cameras will enable volumetric imaging at much greater rates, leading to new capabilities to study dynamic events in cells at high speeds in two and three dimensions.
© 2011 OSA
Fluorescence microscopy has become a standard technique in modern cell biology and many studies demand optically sectioned imaging, which enables quantitative imaging of a thin slice within the sample. However, many applications would benefit from faster optically sectioned fluorescence imaging and from lower associated photobleaching and/or phototoxicity. An example of this is the study of calcium (Ca) waves and sparks in cardiac myocytes. A Ca wave typically travels at around 100 µm.s−1 , and calcium sparks are small, rapid releases of Ca2+ from intracellular stores that are the building blocks of contractile activity in the heart [1,2]. Sparks are typically ~2 µm wide, last for ~10-40 ms [3–5] and occur more often in failing hearts . Recently even smaller Ca release events, termed Ca quarks, have been reported, with dimensions in the range of 0.5-1 µm, which may represent the quantal components of Ca sparks . To the best of our knowledge, the fastest 2D optically sectioned imaging of Ca dynamics in cardiac myocytes to date was performed at up to 660 fps with 512 × 32 pixels using a high speed slit scanning confocal microscope . Even faster imaging rates are required to study Ca spark dynamics at and below the millisecond timescale and, ideally, spark production should be studied in three dimensions on a millisecond timescale.
Another example of the need for microscopy techniques exhibiting lower photobleaching and phototoxicity is long term (hours or days) time-lapse fluorescence microscopy, where it is essential to limit the total light dose to the specimen in order to ensure that sufficient fluorescence signal remains at the end of the experiment and that the sample remains viable. Often, 2D time-lapse imaging is chosen for such studies. For conventional 3D time-lapse imaging the total light dose to the specimen scales linearly with the number of optically sectioned images acquired for each image volume, which can be prohibitive for photosensitive samples.
Currently, the fastest optically sectioning microscopy techniques include multi-beam confocal microscopes  and slit-scanning confocal microscopes [8,10]. However, these imaging techniques illuminate the entire axial extent of the specimen for every image that is acquired. Furthermore, the rapid axial movement of the sample or primary objective lens that is necessary for time-lapse volumetric imaging with these techniques becomes more difficult if an objective lens with a liquid immersion medium is employed, as the mechanical motion can be coupled to the sample. Multiphoton microscopy enables optically sectioned imaging with low out-of-plane photobleaching and/or phototoxicity and is particularly advantageous when applied to imaging in the presence of optical scattering. However, multiphoton microscopy is not the most advantageous technique for all applications as it requires the use of a complex mode-locked laser source and can cause enhanced photobleaching and/or phototoxicity in the focal plane .
One way to reduce photobleaching and phototoxicity is to employ the technique of light sheet microscopy [12–15]. In its most straightforward configuration, light sheet microscopy uses a cylindrical lens to produce a thin sheet of illumination at a single axial position in the sample, which is then imaged by a second objective lens placed at 90° to the illumination plane. The key advantages are that the illumination of the specimen is minimized, as only the plane under observation is illuminated at any given time, and that each optically sectioned image can be acquired in real time by the camera without the use of any moving parts or image processing. Light sheet microscopy has been applied to studies of many different biological systems, e.g. embryo development [REMOVED HYPERLINK FIELD], but normally requires the sample to be embedded in a rigid medium such as agar in order to allow for orthogonal fluorescence excitation and detection. This drawback has been overcome through a number of elegant methods including a tilted light sheet microscopy arrangement using a water dipping objective, which has been used to achieve high speed optically sectioned imaging of neuronal activity in vivo . Another arrangement uses a conventional microscope together with a modified sample chamber that allows the orthogonal light-sheet to be introduced . However, these modifications to light sheet microscopy cannot be applied on a conventional microscope frame to image samples prepared using many standard sample preparation techniques, e.g. microscope slide and coverslip, tissue culture dish or multi-well plate.
Oblique plane microscopy (OPM) is a light sheet microscopy technique that uses the same high NA microscope objective to provide both the fluorescence illumination and detection . In OPM the excitation light is focused to a short line at the edge of the back aperture of the microscope objective in order to produce a sheet of illumination at an angle of ~60° to the conventional optical axis. Fluorescence emitted by the sample is then collected back through the same objective and correction optics are inserted into the fluorescence collection beam path so that the focal plane of the microscope is tilted by ~30°. The tilted focal plane is then co-planar with the illumination sheet to achieve light sheet microscopy using a single microscope objective to excite and collect the fluorescence.
In the previous publication on OPM , an oil immersion objective was used to image fixed specimens with relatively long image acquisition times. In addition, it was necessary to translate the specimen relative to the microscope objective in order to acquire a 3D image of the specimen. In this paper, we present the first live-cell OPM using a water immersion objective and a new OPM configuration that allows the illumination sheet and detection focal plane to be swept simultaneously through the specimen volume without perturbing the specimen or the primary microscope objective. This new configuration enables us to demonstrate OPM of live cells at up to 926 frames per second (fps) and volumetric imaging at up to 21 volumes per second (vps).
2. Experimental OPM system
The optical system employed is shown in Fig. 1 and the principle of the OPM method is described in detail in reference . Briefly, the correction optics for tilting the focal plane consist of a second microscope (T2 and O2) that is placed back to back with the first microscope (O1 and T1) to produce an intermediate image (II) of the specimen where the axial and lateral magnifications are equal, as used in the remote refocusing technique of Botcherby et al. [20–22]. OPM then uses a third microscope (O3 and T3) that is placed at an angle θ to the first two microscopes in order to produce an image of a tilted focal plane within the specimen.
In the initial publication on OPM , only the collected fluorescence passed through the image relay (T1, T2&O2 in Fig. 1). In this paper, the concept of OPM is extended so that both the collection and illumination beam paths pass through the image relay. This has the important advantage that both the light sheet and observation plane can be swept together through the sample at very high speeds by simply adjusting the axial position of the second microscope objective O2.
Excitation light provided by a solid-state 488 nm laser (Cyan Scientific, Spectra-Physics) was attenuated to provide a maximum of 0.39 mW at the back focal plane of O1. The output from the laser was initially expanded in the horizontal direction by a cylindrical telescope (not shown in Fig. 1) and passed through adjustable slit S that allowed the width of the excitation beam to be controlled. For the experiments here, the slit was set to a width of 3.5 mm. The light was then focused by cylindrical lens C1 (f = 20 mm) to produce the initial light sheet at the focal plane of O2. The light sheet is then relayed to the sample via two microscopes formed by O2, T2, T1 and O1, and the specifications for the lenses used are given in Table 1 .
The overall magnification (M) from the focal plane of O1 to O2 is chosen to be equal to the ratio in refractive index between these two planes, i.e. M = 1.33 as O1 is a water immersion objective, which ensures that the axial and lateral magnifications are equal . In order to achieve the necessary total magnification using off-the-shelf optics, it was necessary to employ a multi-element lens for T2 consisting of an achromatic doublet (f = 150 mm, Thorlabs, UK) and an aplanatic meniscus (f = −500 mm, Thorlabs, UK) separated by 97 mm in order to achieve an overall focal length of 162 mm. Fluorescence from the sample is relayed back to form an intermediate image at the focal plane of O2 and then collected by the third microscope formed by O3 and T3 onto an EMCCD camera (iXon DU860, Andor Technology, UK) with 128 × 128 pixels. The emission filter employed was a long-pass 500 nm filter (Chroma Technology, USA). When performing a sub-array readout of 128 × 64 pixels the maximum frame rate of the camera was 926 fps. The effective axial position of the light sheet and tilted observation plane in the sample was determined by the axial position of O2, which was controlled using a piezo-electric objective actuator (Physik Instrumente, Germany). For 3D time-lapse imaging the actuator was driven with a symmetrical saw-tooth wave at 21 Hz over an axial range of 48 µm (corresponding to a range of 36 µm in the sample space). The motion of O2 was synchronized to the CCD camera acquisition and a stack of 20 frames was acquired for each image volume.
The angle between the optical axis of O2 and O3 (θ in Fig. 1) was set to be 32°. As the angular acceptance of O1 (NA = 1.2, n = 1.33) is 64°, then this means that the collection half-angle is 64-32 = 32°, corresponding to a collection NA of 0.7. As can be seen in Fig. 2 , it is objective O1 that limits the collection aperture of this system. The half angle of the illumination cone of rays was chosen to be 5°, which gives a calculated sinc2 illumination beam waist FWHM of 2.5 µm.
Due to the oblique nature of the imaged plane, each resulting image stack records the fluorescence signal from a rhombus-shaped volume in the sample. In order for this to be rendered, it was necessary to ‘shear’ the raw data in software. This was achieved using a custom LabVIEW (National Instruments, USA) programme that performed the image shearing using bi-linear interpolation. Volume rendering was then performed using the maximum intensity projection option in Volocity (PerkinElmer, USA).
The spatial resolution of the system was determined in the oblique plane by imaging 43 nm fluorescent beads (FluoSpheres 505/515, Invitrogen) dried onto a microscope coverslip and subsequently re-immersed in water. The optical FWHM was measured from individual beads (n = 14) to yield an ‘in-OPM-plane’ resolution of 0.41 ± 0.05 µm (mean ± s.d.). An Andor Luca CCD camera was used for this measurement due to its smaller pixel size. The expected in-OPM-plane resolution can be calculated from the numerical aperture of the OPM collection optics (0.7), which gives a calculated PSF FWHM of 0.38 µm that compares reasonably well with the experimental measurement. We also quantified the in-OPM-plane resolution for fluorescent beads located away from the focal plane of O1 by adjusting the axial position of the sample and then adding the required amount of refocus by adjusting the axial position of O2. Our results show that we are able to refocus over an axial range of 40 µm whilst maintaining the measured PSF FWHM to within 7% of the minimum value.
The thickness of the illumination beam was measured by imaging a thin fluorescent polymer sheet coated onto a microscope coverslip. This sample was adjusted so that the fluorescent polymer sheet intersected with the oblique illumination plane in the centre of the CCD field of view. The width of the resulting stripe of fluorescence was then determined (n = 6) and used to calculate the width of the illumination sheet (in the direction perpendicular to the propagation direction) as 3.1 ± 0.2 µm (mean ± s.d.). This experimentally obtained value compares reasonably well with the calculated value of 2.5 µm and we attribute the discrepancy mainly to aberrations caused by the singlet cylindrical optics employed in the excitation beam path.
In order to demonstrate the performance of the OPM system for 2D and 3D imaging, we applied it to image calcium dynamics in acutely dissociated adult rat ventricular cardiomyocytes loaded with the Ca2+ sensitive fluorescent probe Fluo-4 using a standard protocol , and the results are shown in Fig. 3 .
The cells were initially paced using electrical stimulation and then imaged using OPM with the pacing switched off. Figure 3(a) shows an example of a spontaneous Ca wave observed during a time-lapse image acquisition of 16,000 frames at 926 fps with 128 × 64 pixels, see also Media 1. There is some vertical striping in the images shown in Fig. 3(a), which is attributed to attenuation of the excitation beam as it propagates through the sample. In the future, this effect can be overcome using the light sheet scanning approach presented by Huisken and Stainier . Figure 3(b) shows the same cell as in Fig. 3(a) imaged at 505 fps with 128 × 128 pixels during pacing, and Ca sparks are observed in the intervals between contractions (see Media 2). Spatio-temporal differences between spontaneous and field-stimulated calcium transients are seen in Fig. 3. A spontaneous calcium transient arises from a clear single point or subcellular microdomain (a), whereas the stimulated transient arises simultaneously from multiple such points distributed throughout the cell (b). In both scenarios, the decay time for the local calcium level to return to basal levels is relatively constant across different cytoplasmic domains within the cell, whereas calcium decay within bright organelles (taken to be the nuclei) is prolonged in comparison to the cytoplasmic response.
In order to characterize the events observed in Fig. 3(b), we have plotted the fluorescence intensity as a function of time for regions of interest in the cytoplasm and one of the nuclei while cells are being paced, see Fig. 4 .
In Fig. 4(a), it is clear that the fluorescence transient observed in the nucleus begins at the same time point as the transient observed in the cytosol, but takes longer to reach its peak calcium level, which is delayed by approximately 90 ms compared to the cytosol for this specific cell and under this specific pacing protocol. We have also located 8 separate spark events and their corresponding transients are shown as increases in fluorescence over the local baseline value for the spark region (ΔF/F0) in Fig. 4(b). The spark duration measured at full duration at half maximum (FDHM) was found to be 26 ± 3 ms for this cell. This corresponds closely with previous measurements using line scanning confocal microscopy at 22°C from the same species [3,4].
OPM was also used to demonstrate time-lapse 3D imaging of Ca dynamics at 21 volumes per second with 128 × 64 × 20 voxels, see Fig. 5(a) and (b) . The 3D data was rendered using maximum intensity projection and Ca sparks and electrically-stimulated Ca transients can be observed. The same features are also clearly visualised in the rendered 3D time-lapse movie included as Media 3. A Ca spark is observed in Fig. 5(a) at t = 6.72 s that is not present at t = 6.67 s (red circles). At t = 7.25 s, the start of an electrically-stimulated Ca transient occurs that peaks at t = 7.30 s, see Fig. 5(b).
The results presented above were obtained with the OPM field of view (FOV) optimized for imaging isolated cardiac myocytes. The limitation on the maximum lateral FOV is currently set by objective O2 in Fig. 1, which has a FOV of 530 µm. Owing to the magnification of 1.33 provided by O2, T2, T1 and O1, this allows for a maximum field of view of 400 µm at the specimen. Larger lateral fields of view up to the maximum of 400 µm set by O2 could be achieved by decreasing the focal length of T3 or by adding an additional telescope into the beam path between O3 and T3.
In terms of the axial field of view of the OPM system, this is determined by how accurately aberrations introduced by defocus in the first microscope (O1 and T1) are compensated by the second microscope (T2 and O2). This issue has been addressed in detail in the paper by Botcherby et al. , where a remote refocusing system providing aberration-free refocusing over an axial range of 70 µm was demonstrated. We have determined that the OPM system is able to refocus over an axial range of 40 µm whilst maintaining the measured resolution to within 7% of the minimum value. In the future, an improved optical design for the multi-element lens T2 should increase the achieved resolution.
While it is possible to perform high speed 2D optically sectioned imaging using OPM for any actuator position for O2, the 3D image acquisition rate is determined by both the frame rate of the CCD camera and the speed with which objective O2 can be translated. Our current piezoelectric objective actuator can be driven over a range of 75 µm at 55 Hz, which corresponds to a scan range of 55 µm in the specimen, and so our current actuator provides the potential for bi-directional volumetric imaging at up to 110 Hz. Therefore, in the future, higher frame rate CCD cameras will enable OPM to be demonstrated at much higher volumetric imaging rates and/or with a greater number of optical sections per volume.
Currently, the OPM acquisition software displays the raw image acquired by the CCD camera, which corresponds to an optically sectioned image of the sample acquired at an oblique angle. When operated for 3D imaging, the software only displays a single (user-selectable) image from each acquired volume corresponding to a specific oblique slice within the specimen. While these images are acquired at a different angle than in conventional microscopy, it is still straightforward to navigate a specimen rapidly and locate features. In the future it will be possible to render every 3D data set and display it to the user in real time.
In summary, OPM trades-off some of the available NA of a high NA microscope objective in order to achieve optically sectioned imaging via light sheet illumination using only a single objective. As a comparison, a confocal scanning microscope using the same primary objective (O1 in Fig. 1, NA 1.2, water immersion) would achieve a theoretical lateral resolution of 0.2 µm and an axial resolution of 0.7 µm, compared to the theoretical in-OPM-plane resolution of 0.38 µm and light sheet thickness of 2.5 µm. While the spatial resolution achieved with OPM cannot match that of a confocal microscope, the benefits of OPM are those inherent to light sheet microscopy, i.e. minimal out-of-plane photobleaching and no requirements for moving parts or image calculations in order to obtain an optically sectioned image. As we have demonstrated in this paper, these advantages allow OPM achieve high speed 3D imaging.
OPM of live cells has been presented for the first time, demonstrating optically sectioned imaging at 926 fps. In addition, the first near video-rate time-lapse 3D fluorescence imaging of calcium dynamics has been presented at 21 vps. In the future, higher speed cameras should allow events to be studied at much higher volumetric imaging rates, allowing new insights into the temporal and spatial organisation of Ca release within cardiac myocytes and into dynamic events in many other cell types. OPM is not restricted to imaging Ca dynamics, and it provides the potential to perform high speed 2D and 3D imaging of a wide range of biological specimens, ranging from single cells to small embryos. Owing to the intrinsic advantages of light sheet microscopy of low photobleaching and phototoxicity, OPM will also have significant advantages in time-lapse 3D microscopy. As it is can be used to image multiwell plates, OPM is also applicable to high-throughput fluorescence imaging studies.
The authors gratefully acknowledge support from a Royal Society Research Grant (2008/R2), an Engineering and Physical Sciences Research Council First Grant (EP/H03238X/1), and from the British Heart Foundation.
References and links
1. H. Cheng, M. R. Lederer, R. P. Xiao, A. M. Gómez, Y. Y. Zhou, B. Ziman, H. Spurgeon, E. G. Lakatta, and W. J. Lederer, “Excitation-contraction coupling in heart: new insights from Ca2+ sparks,” Cell Calcium 20(2), 129–140 (1996). [CrossRef] [PubMed]
4. H. Cheng, M. R. Lederer, W. J. Lederer, and M. B. Cannell, “Calcium sparks and [Ca2+]i waves in cardiac myocytes,” Am. J. Physiol. 270(1 Pt 1), C148–C159 (1996). [PubMed]
5. W. G. Wier, H. E. ter Keurs, E. Marban, W. D. Gao, and C. W. Balke, “Ca2+ ‘sparks’ and waves in intact ventricular muscle resolved by confocal imaging,” Circ. Res. 81(4), 462–469 (1997). [PubMed]
6. A. R. Lyon, K. T. MacLeod, Y. J. Zhang, E. Garcia, G. K. Kanda, M. J. Lab, Y. E. Korchev, S. E. Harding, and J. Gorelik, “Loss of T-tubules and other changes to surface topography in ventricular myocytes from failing human and rat heart,” Proc. Natl. Acad. Sci. U.S.A. 106(16), 6854–6859 (2009). [CrossRef] [PubMed]
8. G. Iribe, C. W. Ward, P. Camelliti, C. Bollensdorff, F. Mason, R. A. B. Burton, A. Garny, M. K. Morphew, A. Hoenger, W. J. Lederer, and P. Kohl, “Axial stretch of rat single ventricular cardiomyocytes causes an acute and transient increase in Ca2+ spark rate,” Circ. Res. 104(6), 787–795 (2009). [CrossRef] [PubMed]
9. N. Takahashi, T. Sasaki, W. Matsumoto, N. Matsuki, and Y. Ikegaya, “Circuit topology for synchronizing neurons in spontaneously active networks,” Proc. Natl. Acad. Sci. U.S.A. 107(22), 10244–10249 (2010). [CrossRef] [PubMed]
12. H. Siedentopf and R. Zsigmondy, “Uber Sichtbarmachung und Grossenbestimmung ultramikroskopischer Teilchen, mit besonderer Anwendung auf Goldrubinglaeser,” Ann. Phys. 10, 1–39 (1903).
13. A. H. Voie, D. H. Burns, and F. A. Spelman, “Orthogonal-plane fluorescence optical sectioning - 3-dimensional imaging of macroscopic biological specimens,” J. Microsc. (Paris) 170, 229–236 (1993). [CrossRef]
14. J. Huisken, J. Swoger, F. Del Bene, J. Wittbrodt, and E. H. K. Stelzer, “Optical sectioning deep inside live embryos by selective plane illumination microscopy,” Science 305(5686), 1007–1009 (2004). [CrossRef] [PubMed]
16. P. J. Keller, A. D. Schmidt, J. Wittbrodt, and E. H. K. Stelzer, “Reconstruction of zebrafish early embryonic development by scanned light sheet microscopy,” Science 322(5904), 1065–1069 (2008). [CrossRef] [PubMed]
17. T. F. Holekamp, D. Turaga, and T. E. Holy, “Fast three-dimensional fluorescence imaging of activity in neural populations by objective-coupled planar illumination microscopy,” Neuron 57(5), 661–672 (2008). [CrossRef] [PubMed]
18. R. M. Jonker, G. Eichhorn, F. van Langevelde, and S. Bauer, “Predation danger can explain changes in timing of migration: the case of the barnacle goose,” PLoS ONE 5(6), e11369 (2010). [CrossRef] [PubMed]
21. E. J. Botcherby, R. Juskaitis, M. J. Booth, and T. Wilson, “An optical technique for remote focusing in microscopy,” Opt. Commun. 281(4), 880–887 (2008). [CrossRef]