Supplementary information, Figure S1A-S1H The thickness and the uniformity of the light sheet at different DOFs. By imaging FITC-containing solution, the thickness of the light sheet generated by the P3A-DSLM setup was quantified at different DOFs. (A) Representative side views of the light sheet at the modulation intensity of TAG between % and 35%. Scale bar: 5 μm. (B D) The corresponding radial fluorescence profiles at sections of the light sheets shown in (A) at the nearend (red), in the middle (green) and at the far-end (blue) of the focal plane of the excitation objective. Scale bar: 5 μm. (E) The relationship between the TAG modulation intensity and the FWHM of the light sheet at the near-end (red), in the middle (green) and at the far-end (blue) of the focal plane of the excitation objective. TAG modulation intensity at 35% or smaller was used for biological experiments. (F H) Compensation of illumination with a Pockels cell effectively removed nonuniform illumination across the whole DOF. Before compensation, the illumination intensity decays with the penetration depth (F). After dynamically adjusting the illumination laser power using the Pockels cell, the illumination intensity became uniform across the whole penetration depth (G). The intensity profiles of lines across the whole field of view before (blue, F) and after compensation (green, G) were shown in H. Scale bar in F and G: 1 μm. 1
Supplementary information, Figure S1I-S1K Enlarged tunable range of the illumination area by using a low-magnification, low-na illumination objective. By using a low-magnification low-na illumination objective (1X, NA.3), the tunable range of the illumination area was expanded to as large as 5 μm, and no significant changes in the illumination intensity and thickness of the light sheet were observed along the full illumination field. However, the thickness of such light sheet was about.5 μm. (I) Representative side views of the light sheet at the modulation intensity of TAG between % and %. (J) The corresponding radial fluorescence profiles at sections of the light sheets shown in (I) in the middle of the focal plane of the illumination objective. (K) Tailorable light sheet size up to 5 5 μm driven by different TAG intensity modulation. Scale bars: 1 μm.
Supplementary information, Figure S1L Theoretical consideration of the axial resolution of LSFM with different thicknesses of the light sheet. Two orthotropic objectives are used in all light sheet microscopes. Therefore, the point spread function (PSF) of the detection system is determined by the excitation intensity distribution g(x, y, z) and the detection objective blur function d(x, y, z) [1]:,, g x, y, zd x, y, z P x y z (1) The definition of the x-, y-, z-directions is the same as in the main text. We can theoretically estimate the optical resolution of the system by calculating the FWHM of the PSF in the lateral direction (x, y) and axial direction (z). In the lateral direction, the illumination is homogeneous, so g(x, y) =1. Hence, the PSF in x- and y- directions are solely determined by d lateral(x, y), the lateral blurring function within the focal plane of the detection objective lens and given by the Airy formula []: J1 nrsin Plateral r dlateral r nrsin () Here, λ is the wavelength of the fluorescence, n is the diffraction index, α is the aperture angle of the objective lens and r is 3
given by: r x y (3) For fluorescence wavelength of 515 nm (EGFP) and a detection objective lens with NA of.8, we calculate the FWHM lateral of the PSF to be 396 nm. In the axial direction, the excitation is highly concentrated and the intensity distribution can be approximated by a Gaussian beam: g z z exp (4) Where σ represents the standard deviation of the Gaussian function which can be calculated by the FWHM of excitation profile in z-direction (σ ez) as the equation shown below []:. σ ez 1 FWHM ez ln (5) And the blurring function d axial (z) along z-axis is also known as []: d axial z sin nzsin nzsin (6) Hens the axial PSF is given by: sin nzsin z Paxial z exp ez nzsin (7) By changing the thickness of the light sheet (σ ez), we obtained various distributions of P axial (z), in which each FWHM axial can be calculated, respectively. The figure showed the relationship between FWHM axial (z-direction resolution) and σ ez (z-direction thickness of the light sheet). Under the wide-field illumination, σ ez approaches infinity and the axial resolution is ~1.7 μm. The thickness of common single-view LSFM that can reach the same penetration depth in x-direction as we did (17 μm) is about ~8 μm [3]. Correspondingly, their axial resolution is about 1.3~1.7 μm, only slightly improved as compared to the wide field illumination. In our P3A-DSLM setup, the thickness of the light sheet can be as thin as 8 nm-9 nm along x- direction. This significantly improves axial resolution to 7~8 nm, much higher than the existing single-view light sheet 4
microscopes. It is also clear that, while less than μm light sheet thickness significantly improves the axial resolution, thicker light sheet will contribute little to the axial resolution improvement. 5
Supplementary information, Figure S1M-S1N The relationship between the spatial resolution and the maximum DOF. 5-nm Green fluorescent beads were homogenously embedded in clear agarose gel and excited by 9 nm laser with different modulation intensities of TAG of %, 5%, 1%, 15%, %, 5%, 3% and 35%. An axial step of 1 nm was used to collect the fluorescence images of each bead along the z axis. (M) For each TAG modulation intensity, the bead images in xy plane and yz plane were averaged images of nine beads, with three beads at the near-end, the middle and the far-end of the excitation field, respectively. The equivalent pixel size on scmos camera was ~14 nm (5.6 μm/4), satisfying the Nyquist sampling requirement. Scale bars: 5 nm. (N) The lateral resolution of the system was defined as the lateral FWHM of the fluorescence beads, which ranged between 4~45 nm. 6
Supplementary information, Figure S1O-S1Q Ultrafast calcium nanosparks measured in a rat cardiac myocyte. We fused GCaMP6f to the N-terminus of triadin 1 to measure dyanmic calcium events occurring in the nanoscopic space between T- tubule and junctional sarcoplasmic reticulum ( Calcium nanosparks ) [4]. (O) By combining ultra-fast TAG axial scanning (45 khz), fast galvo-mirror vertical scanning (1 khz) with cropped scmos acquisition (48x56), the P3A-DSLM succeeded in -dimentional ultra-fast imaging (8 fps) of calcium nanosparks in the dyadic space of a single cultured rat cardiac myocyte. (P) The region circled in O showed three spark events occurred at different times. (Q) Expanded figure of the middle event at panel P. Although one-dimensional line-scan confocal imaging can achieve similar temporal resolution, our setup could provide two-dimensional information, which may be important to study the intra- and inter-junctional interactions between different ryanodine receptors. Scale bar: μm. 7
Supplementary information, Figure S1R-S1X Experimental and theoretical comparison of photobleaching between a P3A-DSLM and a P-LSM. (R) To compare photobleaching between the P3A-DSLM and the P-LSM, we must ensure that two systems have same spatial and temporal resolution, and produced fluorescence images with similar signal rates upon recording similar biology samples. In theory, we considered a P3A-DSLM with average power P, excitation intensity I, illumination NA, emission NA e, waist width w o, Rayleigh range of b and signal rate S. In comparison, the P-LSM with average power P', excitation intensity I', illumination and emission NA', waist width w o', Rayleigh range of b' and signal rate S'. The waist width is inversely proportional to NA; Rayleigh range is inversely proportional to the square of NA: P3A-DSLM: w ( NA) (8) 8
b n (9) ( NA) P-LSM: w b ( NA) n ( NA) (1) (11) where λ is the wavelength of the laser, n the refractive index of the medium. To achieve the same lateral resolution and axial laser focus size, w o in P3A-DSLM has to be equal to b' in P-LSM, and the emission NA e in P3A-DSLM has to be equal to NA' in P-LSM: w n b ( NA) ( NA) (1) NA NA (13) e From Eq. 1 we have: NA nna (14) With the experimental data shown in Supplementary Figures S1A D, the axial FWHM of the excitation in P3A-DSLM was about 8 nm. From the relationship of the FWHM to NA, FWHM p ln NA (15) we calculated the effective illumination NA as ~.3. According to Eq. 14, we need an objective with NA'~.9 in P-LSM to achieve equal axial laser focus size. Therefore, we have NA: NA'=1:3. Because the excitation intensity I is proportional to the square of NA, the peak intensity of I in P3A-DSLM is 9 times lower than I' of P-LSM recorded at the same resolution. The super-quadratic effect, which depends on the peak illumination intensity, is the main mechanism for the photodamage in nonlinear in vivo imaging. Therefore, similar to the P-DSLM, our system exhibited a much reduced peak illumination intensity and minimized photobleaching and photodamage [5]. Experimentally, we compared our P3A-DSLM with an OLYMPUS F1 P-LSM, using the same femtosecond laser (Coherent Chameleon) and Nikon 4X NA.8 objectives. The objective in the P-LSM and the detection objective in the P3A-DSLM were used with their full back aperture to achieve the same lateral resolution. The illumination objective in the P3A-DSLM were used with part of its back aperture to achieve a NA~.3. This arrangement ensured similar excitation thickness in two cases. The TAG in the P3A-DSLM was 9
maintained at % modulation to achieve 1 μm DOF, comparable to the size of the P-LSM illumination field. The 3D stacks were captured at axial step of 5 nm per slice. The temporal resolutions for both systems were maintained at 5 ms per frame and 4 s per volume. Using these systems, we continuously captured time-lapse 3D images of mt-cpyfp-transgenic C. elegans. (S-T) Time-lapse images of worms observed at different times under the P3A-DSLM (S) or the P-LSM (T). (U-V) The fluorescence intensity of worms exhibited negligible bleaching during 15 minutes of recording under the P3A- DSLM (U), while more than 5% of the YFP signal was lost due to bleaching under the P-LSM (V). (W-X) 3D images (x-y, y-z and x-z planes) of worms observed under the P3A-DSLM (W) or the P-LSM (X). Scale bars: 3 m in xy plane and 1 m in y-z and x-z planes. 1
Supplementary information, Figure S1Y Theoretical comparison of photodamage among a P3A-DSLM, a P-DSLM and a static light-sheet microscope. In [5], the authors have compared the P-DSLM with the static light-sheet. Following their precedents, we extended such comparison to the P3A-DSLM. We considered a comparison among a P3A-DSLM lightsheet created by a spherically focused Gaussian beam with a certain characteristic width w in both the z-direction and y- direction, and with a certain Rayleigh range b; a conventional P-DSLM with a width Nw in both the z-direction and y- direction and naturally an enlarged Rayleigh range of N b [3]; and a static light-sheet microscopy focused to the same width Nw in the z-direction and same Rayleigh range of N b while being uniform in the y-direction over a range of Mw. To simplify the calculation, we assume that within Rayleigh range N b, the width of the beam remains constant. In order to get the same uniform illumination field size, the P3A-DSLM need to be extended N times in x-direction using TAG tuning and M times in y-direction using GSM scanning. Similarly, the P-DSLM need to be extended M times in y-direction using GSM scanning. The excitation intensity is proportional to the average excitation power, inversely proportional to the laser beam cross-sectional area, and is thus given by: P3A-DSLM: I 1 (16) P-DSLM: I (17) Static light-sheet: 3 (18) The excitation volume in a unit time is given by: I P w P Nw P MNw 11
P3A-DSLM: v 1 w b (19) P-DSLM: Static light-sheet: v N 4 w b () 3 v MN w 3 (1) b The two-photon average excitation signal rate per spatial duty cycle is proportional to the square of the intensity times excitation volume, which are given by: P3A-DSLM: P-DSLM: Static light-sheet: P SR I V b () w 1 1 1 4 w P SR I V N b (3) 4 w 4 4 N w P SR I V MN w b (4) 3 3 3 3 4 M N w From Eqs 4, we concluded that if we acquired the same signal rate among P3A-DSLM, P-DSLM and static lightsheet with same detection efficiency SR 1=SR =SR 3, the ratio of excitation intensity is: I : I : I 1 : : 1 3 1/ 3/ N M N (5) Usually the photodamage is classified into three categories: linear, quadratic, and supra-quadratic, which correspond to effects that come from absorption of one, two, and more than two photons, respectively. With the same pulse width and the repetition rate of the femtosecond laser, the linear, quadratic, and supra-quadratic photodamage to the whole sample in the excitation pathway are proportional to IV, I V and I x V, respectively. Using Eqs 19 1 and Eq 5, we have: Linear photodamage: Quadratic photodamage: Supra-quadratic photodamage: 1/ 3/ I1V 1 : I V : I3 V3 1: N : M N (6) I V : I V : I V 1:1:1 (7) 1 1 3 3 I V : I V : I V 1: N : M N (8) x x x x4 x/1 3 x/3 1 1 3 3 For M >>N (in our experiments, M about, N about 4) and x >, P3A-DSLM will have lower linear photodamage, equal quadratic photodamage but higher supra-quadratic photodamage to the sample compared with P-DSLM and static light sheet. This is the consideration of sample exposure within the Rayleigh range volume. 1
The photodamage effect to the sample out of the illumination Rayleigh range is also important. During TAG axial scanning, imaging regions out of the Rayleigh range volume is exposed to the laser energy in the P3A-DSLM, which is absent in the P-DSLM and static light sheet configurations. The much higher effective NA used in the P3A-DSLM (5-1 times compared with NA in the P-DSLM and the static light-sheet microscope) results in much faster diverging beam profile and leads to more linear heating but negligible quadratic and supra-quadratic damage. So the total per-frame linear damage of the P3A-DSLM to the sample will be the summation of two components. It varies at different z-section intervals, field sizes and biological samples, and shall be evaluated case by case. 13
Supplementary information, Table S1 Imaging parameters for all the presented images and videos Fig.1H, Supp. Mov1 Supp. Fig. 1O Supp. Mov Fig.1J, K Supp. Mov3 Supp. Fig. 1S,W Supp. Mov4 Left Supp. Fig. 6T,X Supp. Mov4 Right Modality P3A-DSLM P3A-DSLM P3A- DSLM Sample C.elegans A rat cardiac Zebrafish on pharynx on myocyte adult day 3 adult day 3 Imaged volume(xyz) 9 45 1 6 56 1 1 14 P3A-DSLM C. elegans pharynx on adult day 3 P-LSM C. elegans pharynx on adult day 3 474 843 8 51 51 79 97 Exposure Time (ms) 1. 5 5 5 Average 5 1 1 No No Lateral pixel size ( μm).15.15.15.15.3 Axial section (z-step) size ( μm) Excitation wavelength ( nm) Excitation power out of Obj ( mw) TAG modulation intensity ( %) 1.5.5.5 85 9 8 9 9 5 1 18 5 5 3 References: 1 Engelbrecht CJ, Stelzer EH. Optics letters 6; 31:1477-1479. Born M, Wolf E. Principles of optics : electromagnetic theory of propagation, interference and diffraction of light. 7th expanded Edition. Cambridge ; New York: Cambridge University Press 1999. 3 Kopecky B, Santi P, Johnson S, Schmitz H, Fritzsch B. Dev Dyn 11; 4:1373-139. 4 Shang W, Lu F, Sun T et al. Circ Res 14; 114:41-4. 5 Truong TV, Supatto W, Koos DS, Choi JM, Fraser SE. Nat Methods 11; 8:757-76. 14