Materials and Methods


SPIM setup

SPIM setup consisted of an Olympus N4X-PF low magnification air objective for detection and two cylindrical lenses aligned orthogonal to the detection for double sided light-sheet illumination. The sample was illuminated alternately from two sides. The alternating illumination was controlled via a flipper mirror (8892-K-M, New Focus, USA). The sample chamber was 3d printed and had two windows for illumination and one window for detection of the sample. The electromagnets were hold by the sample chamber and assembled in a tetrahedral geometry to enable three-dimensional control of sample’s orientation. The applied current was remotely controlled via a programmable power supply (QL355P, Aim-TTi, United Kingdom) and a manual switch. The focus of the four electromagnet-tips coincided with the sample. The sample was inserted from the top within a FEP tube orthogonal to the detection objective and to the illumination objective. To scan the sample the sample chamber was positioned on a motorized linear stage (M111.1DG, Physik Instrumente, Germany) and moved relative to the detection objective. To correct for the different path length in water and air the detection objective was also placed on a linear stage (M111.1DG, Physik Instrumente, Germany) and moved accordingly. For wide-field illumination a LED is placed behind the sample. For light-sheet illumination a single color (488nm) Coherent Sapphire laser (488-30 CDRH) was used. The wide-field and fluorescence signals were detected with a sCMOS camera (Zyla 5.5, Andor, United Kingdom).

Electromagnets
The electromagnets were custom-built and consisted of a magnetic core made from HyMu 80 alloy (National Electronic Alloys, USA) and a solenoid. The bobbin was made from Teflon and a high resistance wire was wind up hundreds of times to create a high magnetic field. The magnetic core had a diameter of 6mm and was tapered to create a strong magnetic field gradient. The inner diameter of the bobbin was 6mm that the magnetic core could be inserted.


Super-paramagnetic beads
The washed super paramagnetic beads (Dynabeads MyOne Carboxylic Acid, Invitrogen, USA) with a diameter of 2.8 µm were injected with standard glass needles into the yolk of the zebrafish embryo. The injection needle was inserted from either the vegetal pole or the lateral side. For a sufficient torque to turn the embryo it is important that the beads stay close to the yolk membrane. Injections were performed at extremely low pressure (5-10 psi) and long injection duration (100-150 ms) to avoid the dispersion of beads. By applying a strong constant magnetic field after the injection of the beads with a permanent magnet the beads are attracted and clump. This aggregation of beads preserves the single beads from translating through the yolk and eases the rotatio


Magnetic manipulator inset for a stereoscope and inverted microscopes
The magnetic manipulator inset was build from a aluminum plate with a size of 92.5x64 mm fitting on standard microscope stages. On this plate an arc holding two electromagnets that can be freely slid along the arc to control the angle between the magnets and the sample. The sample is either embedded in a glass capillary to study it on a upright microscope or placed on a coverslip and held by a Teflon holder to study the sample on an inverted microscope.


Zebrafish
TO DO (GOPI)

Sample handling for taking multi-view, multi-axes stacks
We oriented the zebrafish in the chorion and waited till it had oriented what typically took less than 10s. When the embryo was oriented we switched off the magnet or reduced the current before starting a stack to avoid any deformations by the applied force. Switching off typically resulted in a slight reorientation in a new settle position. After a few seconds when the zebrafish had settled in its new resting position we started the 3D stack of the zebrafish.

Registration of the multi-view, multi-axes SPIM data
Since the zebrafish is rotating in the chorion and the surrounding media did not change its orientation bead based registration techniques can no be used to register the multi-view, multi-axes SPIM data. To use the sample feature based registration the transformation between the different views had to be determined Preibisch 2010. We determined the transformations by using the numeric control of the orientation in the 3d viewer vaa3d and aligning two views and choosing the best transformation by eye Peng 2010. We applied the found models to the stacks in the BigDataViewer and started the feature based registration Pietzsch 2015. If the registration failed we wrote a macro to test different orientations around the by eye found best guess.