The 100 € lab: A 3-D printable open source platform for fluorescence microscopy, optogenetics and accurate temperature control during behaviour of zebrafish, Drosophila and C. elegans

Small, genetically tractable species such as larval zebrafish, Drosophila or C. elegans have become key model organisms in modern neuroscience. In addition to their low maintenance costs and easy sharing of strains across labs, one key appeal is the possibility to monitor single or groups of animals in a behavioural arena while controlling the activity of select neurons using optogenetic or thermogenetic tools. However, the purchase of a commercial solution for these types of experiments, including an appropriate camera system as well as a controlled behavioural arena can be costly. Here, we present a low-cost and modular open-source alternative called FlyPi. Our design is based on a 3-D printed mainframe, a Raspberry Pi computer and high-definition camera system as well as Arduino-based optical and thermal control circuits. Depending on the configuration, FlyPi can be assembled for well under 100 Euros and features optional modules for LED-based fluorescence microscopy and optogenetic stimulation as well as a Peltier-based temperature stimulator for thermogenetics. The complete version with all modules costs ~200 Euros, or substantially less if the user is prepared to shop around. All functions of FlyPi can be controlled through a custom-written graphical user interface. To demonstrate FlyPis capabilities we present its use in a series of state-of-the-art neurogenetics experiments. In addition, we demonstrate FlyPis utility as a medical diagnostic tool as well as a teaching aid at Neurogenetics courses held at several African universities. Taken together, the low cost and modular nature as well as fully open design of FlyPi make it a highly versatile tool in a range of applications, including the classroom, diagnostic centres and research labs.

Here, we first present the basic mode of operation including options for micropositioning 81 of samples and electrodes and demonstrate FlyPi's suitability for light microscopy and 82 use as a basic medical diagnostic tool. Second, we present its fluorescence capability 83 including basic calcium imaging using GCaMP5 [1]. Third, we survey FlyPi's suitability 84 for behavioural tracking of Drosophila and C. elegans. Fourth, we demonstrate 85 optogenetic activation of Channelrhodopsin 2 [3] and CsChrimson [14] in transgenic 86 5 larval zebrafish as well as Drosophila larvae and adults. Fifth, we evaluate performance 87 of FlyPi's Peltier-thermistor control loop for thermogenetics [15]. Sixth, we briefly 88 summarise our efforts to introduce this tool for university research and teaching in sub- 89 Saharan Africa [4], [16].  Basic camera operation and microscopy 108 To keep the FlyPi design compact and affordable yet versatile, we made use of the RPi 109 platform, which offers a range of FlyPi-compatible camera modules. Here, we use the 110 "adjustable focus RPi RGB camera" (Supplementary Table 1) which includes a powerful 111 12 mm threaded objective lens. Objective focal distance can be gradually adjusted 112 between ~1 mm (peak zoom, cf. Fig. 2D) and infinity (panoramic, not shown), while the 113 camera delivers 5 megapixel Bayer-filtered colour images at 15 Hz. Spatial binning 114 increases peak framerates to 42 Hz (x2) or 90 Hz (x4). Alternatively, the slightly more 115 expensive 8 megapixel RPi camera or the infrared-capable NO-IR camera can be used. 116 Objective focus can be set manually, or via a software-controlled continuous-rotation 117 micro servo motor (Fig. 1G). Alternatively, the RPi CCD chip can be directly fitted 118 above any other objective with minimal mechanical adjustments.  The camera can be mounted in two main configurations: upright or inverted ( Fig. 2A, B).

127
While the former may be primarily used for resolving larger objects such as adult 128 Drosophila (Fig. 2C) or for behavioural tracking (cf. Fig. 4), the latter may be preferred 129 7 for higher-zoom applications (Figs. 2D,E) and fluorescence microscopy (cf. Fig. 3), or if 130 easy access to the top of a sample is required. Here, the image quality is easily 131 sufficient to monitor basic physiological processes such as the heartbeat or blood-flow 132 in live zebrafish larvae (Fig 2F, Supplementary Video 1).

133
If required, specimens can be positioned by a 3-D printed micromanipulator [4] (Fig.   134 2B). Up to three manipulators can be attached to the free faces of FlyPi (Fig. 1D, I).

135
Manipulators can also be configured to hold probes such as electrodes or stimulation 136 devices (Fig. 1I). Like the camera objective, manipulators can be optionally fitted with    is traditionally excited around 488 nm, however there is a second and larger excitation 168 peak in the near UV [20] (Fig. 3D, but see e.g. [1] ). Here, we made use of this short-169 wavelength peak by stimulating at 410 nm to improve spectral separation of excitation 170 and emission light despite the suboptimal emission filter. Figure 3C shows the 171 fluorescence image recorded in a typical fluorescence test-slide. The RGB camera chip 172 allowed simultaneous visualisation of both green and red emission. If required, the red 173 9 channel could be limited either through image processing, or by addition of an 174 appropriate short-pass emission filter positioned above the camera. Next, using green 175 fluorescent beads (100 nm, Methods) we measured the point spread function (psf) of 176 the objective as 5.4 µm (SD) at full zoom (Fig. 3E, F). This is approximately ten times 177 broader than that of a typical state-of-the art confocal or 2-photon system [21], though 178 without optical sectioning, and imposes a theoretical resolution limit in the order of ~10 179 µm. Notably, with an effective pixel size of ~1 µm (Fig. 1E) the system is therefore 180 limited by the objective optics rather than the resolution of the camera chip such that the 181 use of a higher numerical aperture objective would yield a substantial improvement in 182 spatial resolution. It also means that at peak zoom, the camera image can be binned at 183 x4 for increased speed and sensitivity without substantial loss in image quality.

184
Next, we tested FlyPi's performance during fluorescence imaging on live animals. At 185 lower magnification, image quality was sufficient for basic fluorescence detection as 186 required for example for fluorescence based sorting of transgenic animals (screening). 187 We illustrate this using a transgenic zebrafish larva (3 dpf) expressing the GFP-based

216
One key advantage of using genetically tractable model organisms is the ability to 217 selectively express proteins in select populations of cells whose state can be precisely        The GUI is also capable of creating folders and saving files to the Raspberry Pi desktop.

434
For simplicity, the software creates a folder called "FlyPi_output" and subfolders  Video and image acquisition: All static image data was obtained as full resolution red-463 green-blue (RGB) images (2592x1944 pixels) and saved as jpeg. All video data was 464 obtained as RGB at 42 Hz (x2 binning), yielding image stack of 1296x972 pixels, and 465 saved as h264. Video data was converted to AVI using the ffmpeg package for 466 GNU/Linux (ffmpeg.org, a conversion button is added to the GUI for simplicity). All 467 further data analysis was performed in Image-J (NIH) and Igor-Pro 7 (Wavemetrics).