Fig 1.
Microscope setup, light sheet geometry and beam path.
(A): Computer rendering of the setup. (B): Light sheet geometry. The water-filled cuvette with the sample embedded in an FEP tube was positioned at the intersection of the illumination and detection objective’s focal planes. The sample was moved through the thin sheet of light illuminating the focal plane of the detection objective. (C): Top view of the beam path. The illumination laser was fibre-coupled into the beam expander (BE) and the light sheet formed with a cylindrical lens (CE). A mirror (M) was used to adjust the light sheet position. The light sheet was projected into the sample with an assembly of scan lens (SL), tube lens (TL) and illumination objective (IO). A red LED (LED) was used for transmission imaging. Both, fluorescence and transmission signal, were collected with the detection objective (DO), separated with a filter cube (FC) and imaged through two identical tube lenses (TL-D) simultaneously onto two cameras (BF cam, FL cam). To acquire 3D data, the imaging chamber was moved in z and the z-position of the detection objective was adjusted using motorized stages. (D): Side view of the beam path. Side and top view only differ after the illumination light passed the cylindrical lens; only this part of the beam path is illustrated.
Fig 2.
Adjustment of the detection objective position.
(A): During alignment, the imaging plane and light sheet were superpositioned in the centre of the sample. (B): A 3D image stack was acquired by moving the medium-filled chamber containing the sample through the light sheet, but imaging plane and light sheet did not overlap anymore due to refractive index mismatch between medium and air. (C): The detection objective was moved to compensate for the mismatch. (D): Colour-coded projection of ø500nm fluorescent beads acquired without moving the detection objective. Beads towards both ends of the stack were out-of-focus. (E): Same stack acquired with refocusing of the detection objective. All illuminated beads were in-focus. (F): The stack shown in (D) was resliced and planes 350-399 were maximum intensity projected (highlighted region in (D)), showing out-of-focus beads towards the end of the stack. (G): Reslice and maximum intensity projection of the highlighted region in (E). The whole stack is in focus. Scalebar 100 m.
Fig 3.
72 hpf zebrafish embryos expressing GFP in the vascular system.
Prior to imaging, all zebrafish embryos were fixed and their fluorescence signal was recovered using booster-GFP. (A): Transmission image and (B): fluorescence image of vasculature of the same embryo. (C): Overlay of fluorescence and transmission data. Fluorescence data was 3D-rendered and colour coded for depth. Dataset of 75planes, colour bar, 500 m long. (D): Fluorescence signal of the sample embedded in vectashield (blue) and agarose (red). The sample was illuminated every 30s for 15min, followed by a 15min dark period (inset). Photobleaching dominated during bright periods. For mounting in vectashield, overall photobleaching was greatly reduced and even negligible during dark periods.
Fig 4.
(A): Photograph of the setup combined from two individual photographs acquired with different polarization. All optical components were placed beneath a glass cover, which also included a laser safety interlock. In order to keep the appearance of the system as clean and clear as possible, all controllers, the Arduino board and the PC were hidden below the table. The system was operated via a control panel including seven push buttons while the sample was viewed on a monitor. (B): Usage statistics: stacks (blue) and sample moves (red). Peaks in number of acquired stacks correspond to the official opening day of the exhibition (champagne glass), the long night of open museum (moon), a stretch of extremely rainy days in Dresden (cloud), the photobleaching experiments we conducted (laser, see Fig 3) and a media festival (camera). (C): Distribution of button presses overall (blue) and on Aug 18, 2015 (red), automated photobleaching experiments were not counted towards the overall number of button presses.