Live tracking of Moving samples in confocal microscopy for vertically grown roots tips

Roots navigate through soil integrating environmental signals to orient their growth. The Arabidopsis root is a widely used model for developmental, physiological and cell biological studies. Live imaging greatly aids these efforts, but the horizontal sample position and continuous root tip displacement present significant difficulties. Here, we develop a confocal microscope setup for vertical sample mounting and integrated directional illumination. We present TipTracker - a custom software for automatic tracking of diverse moving objects usable on various microscope setups. Combined, this enables observation of root tips growing along the natural gravity vector over prolonged periods of time, as well as the ability to induce rapid gravity or light stimulation. We also track migrating cells in the developing zebrafish embryo, demonstrating the utility of this system in the acquisition of high resolution data sets of dynamic samples. We provide detailed descriptions of the tools enabling the easy implementation on other microscopes.


Introduction 33
Root tips constantly explore the soil searching for water and nutrients. Their movement is propelled by 34 however the authors did not provide a comprehensive documentation of the setups that were used. 56 light coming from the LEDs ( Figure 3G, H). This allows for keeping the illumination on during image 119 acquisition. 120 One limitation of the system is that the usage of immersion objectives becomes complicated as the 121 immersion solution flows down. We overcame this by using a hair gel with a refractive index similar to 122 water. The microscope is, however, mainly intended for the acquisition of multiple position and long 123 time-lapse experiments, thus the use of any immersion liquid is problematic in a standard microscope 124 setup as well. For these experiments, we primarily employ a 20x/0.8NA or a 40x/0.95NA dry objective. 125 In summary, we developed a vertical microscope setup with controlled illumination that enables multi-126 day live imaging. The system keeps the entire functionality of the CLSM microscope, including the 127 motorized stage. This is essential for the acquisition of multiple positions during time-lapse experiments 128 and also for the tracking of the root tips, as described below. 129 130

Rapid gravistimulation experiments using the rotation stage 131
For gravistimulation experiments it is important that the sample is securely fixed. Also for the long-term 132 time-lapse experiments, the seedlings need to be well adapted to the experimental conditions. We grow 133 the seedlings in a Nunc™ Lab-Tek™ chambered coverglass. While the roots grow between the glass and 134 a block of agar, the cotyledons remain free to the air (Figure 4 A, B). A detailed sample preparation 135 protocol is provided in the materials and methods section. Ideally the sample preparation is performed a 136 few hours before imaging to give the plants time to acclimate and recover from the stress of transfer. 137 Plants can be cultured in the chambers for a period of several days. grow down along the gravity vector. Since the gravity vector cannot be modified easily, we developed a 145 microscope sample holder that can be rotated by any degree around the axis of the light path ( Figure 4C). 146 In this way, the roots can be observed before and shortly after the gravistimulation, and due to the axis of 147 rotation, the "upper" and "lower" root sides are equally accessible to imaging. The rotation stage is an 148 aluminium frame with a rotating inset that holds the sample chamber. The inset and the frame are 149 connected by a number of rings made of Teflon to provide smooth and precise sliding ( Figure 4E, F). 150 Supplemental File 1 contains a 3D CAD file of the rotation stage designed for a motorized stage 151 (Märzhäuser Scan IM). In order to minimize the time the user spends finding the roots after rotation of the 152 inset, we developed a MATLAB®-based script that calculates the new positions of the root tips ( Figure  153 4D, Supplemental File 2). The experimental procedure was as follows: First, the motor coordinates of the 154 mechanical centre of rotation had to be determined. To this end, the inset holding the sample chamber 155 was replaced with a disk into which a small hole (diameter 200 μm) had been drilled, which coincides 156 with the centre of rotation. The hole was centred in the field of view and the motor position was saved in 157 a file. Then, the disk was replaced with the sample holder and the positions of the root tips were saved. 158 After imaging the first part of gravistimulation experiment (roots in vertical position) the rotation was 159 applied. The MATLAB® script (Supplemental File 2) was executed, and output the new position of root 160 tips. The mechanical precision was good enough that the calculated positions deviate only slightly from 161 the actual ones and imaging could be continued within 3 minutes after the rotation. 162 Thus, our rotating stage enables the user to select any sequence of gravistimulations desired, and 163 subsequently a very rapid image acquisition, providing the setup necessary for high-resolution studies of 164 gravitropism. 165 166

TipTracker automatically recognizes and follows root tips during growth 167
A root tip of a 4-5 day-old Arabidopsis seedling grows approximately 50 -300 μm per hour (see below). 168 This means that it moves through the field of view of a 20x objective within 1-2 hours. To be able to 169 observe the root tips for a longer period of time, we developed the root tip-tracking program TipTracker. that is used to update the lateral sample displacement . This method makes no assumptions about the 181 shape or brightness of the samples or the type of movement and is thereby not limited to roots; it is, in 182 fact, entirely independent of the specimen and can be used for all samples that move autonomously or 183 through external forces. The growth of a root between two time points -1)+ 184 . Finally, a new list P(x, y) = P(x, y) + of the predicted positions of the roots is generated and 185 loaded into the microscope control software and the next acquisition is started ( Figure 5). This process is 186 The program is designed to follow actively growing root tips in a highly efficient manner, as we 212 demonstrate below. In case the tracking algorithm loses a sample, this can result in excessive stage 213 movements. In order to protect the objectives, we implemented a limit on the maximum degree of stage 214 movement. When one of the positions exceeds this user-defined limit, the tracking of that particular 215 position is stopped, while the other positions are further tracked. 216 Limitations are that the computer should not be interfered with during imaging, as this could confuse the 217 communication between TipTracker and the imaging software. In addition, online tracking with 218 TipTracker creates a time overhead compared to a time series that is acquired directly with the 219 microscope control software, since at each time point the acquisition is stopped, the data stored and then 220 read again. 221 In summary, our root tip-tracking program TipTracker allows for online long-term tracking of root tips or 222 other moving samples and can be easily implemented on a wide range of microscopes. 223

Long-term imaging and tracking of root tips 226
To test the ability of the TipTracker software, we imaged roots expressing the plasma membrane marker 227 successfully tracked all roots. We coupled the illumination system to a regular time switch to simulate 230 day and night. The growth rate of all roots dropped during the night period and increased again in the day 231 period ( Figure 6B). In the resulting images, the cell division in the meristematic zone and the progression 232 towards the transition and elongation zones can be observed ( Figure 6C). We took a single plane of one of 233 the datasets, cut out a small area overlapping with one of the cell files (cortex) and mounted the images 234 side-by-side as a montage ( Figure 6C). In that montage we colour-coded membranes according to their 235 appearance (first generation: yellow, second: cyan, third: magenta, fourth: green). This experiment 236

Imaging of the KNOLLE syntaxin during cell division 240
To test the tracking using higher magnification, we analysed the dynamics of the expression of the cell relocalises to pre-vacuolar compartments and is finally degraded in the vacuole, as described previously 247 (Reichardt et al., 2007). In our setup, we could follow this cycle in a given cell and measured that it lasted 248 for more than 4 hours (Figure 7, lower panel). We successfully imaged growing roots for more than 12 249 hours and captured a stack of 10 z-sections every 3 minutes. 250 Imaging of the DII-Venus after gravistimulation 251 As a next example, to test how we can visualize dynamic processes during gravitropism, we observed 252 roots during gravistimulation using the rotation stage. For this purpose, we used the DII-Venus auxin These examples of the performance of TipTracker show that the program can be used to track root tips at 278 high temporal and spatial resolution. In addition to root tips, it is also possible to track other moving 279 samples, as we demonstrate with the example of the prechordal plate movement in the zebrafish embryo. 280 It is important to note that when setting up an experiment, the users should consider the magnitude of the 281 velocity of their sample relative to the field of view of the objective being used, as well as to the temporal 282 resolution of the acquisition (Figure 10). For example, when using a 20x objective lens, we recommend 283 specifying time intervals not larger than 30 minutes, otherwise the root will escape the tracking field of 284 view before the next time point is captured. 285

Conclusions and Discussion 286
In this work, we describe in detail a confocal microscope setup with a vertical mounting. This enables 287 long-term (up to several days) live imaging with confocal resolution of seedlings growing in the natural, 288 vertical position. We also built a rotation stage that makes it possible to freely adjust the plant's 289 orientation with respect to gravity, while preserving the ability to observe it. Together with integrated 290 illumination, our setup provides growing seedlings with the optimal and controlled conditions necessary 291 for long-term imaging experiments. We provide blueprints for building the setup and a description of 292 optimized sample preparation, which is a critical step for the sensitive Arabidopsis seedlings. Together, 293 the sample preparation, illumination and the vertical position result in healthy seedlings, even in the 294 artificial conditions of a confocal microscope. 295 Furthermore, we developed the TipTracker program to automatically follow root tips for long periods of 296 time. Importantly, it can track multiple objects simultaneously while fully preserving the functionality of 297 the confocal, i.e. multiple-colour imaging and z-sections. Brightfield and fluorescence channels can be 298 used as the input for tracking. The tracking is both robust and very accurate, as exemplified by 299 Supplemental movies 7 and 8. TipTracker tracks objects only in 2 dimensions since the roots are confined 300 between the coverslip and the agar block, but if needed, 3D tracking is a straightforward extension. 301 The usage of TipTracker is not limited to vertical stage microscopes and can be used on any inverted or 302 upright microscope setup with a motorized stage. The example of zebrafish embryo development also 303 demonstrates that the tracking algorithm is not limited to root tips, and can in fact be used for all moving 304 samples, given that the algorithm makes no assumptions about the samples, such as shape, brightness, or 305 direction of movement. 306 Combining optimal growing conditions and root tip tracking we now were able to perform experiments 307 that were previously very hard to conduct. Long-term image acquisition revealed previously 308 unappreciated regularity in the cell division and elongation pattern making up the root growth. The high 309 resolution imaging of dividing cells enabled capturing the exact timing of several cytokinesis events while 310 observing the whole root meristem; or the observation of the dynamic rearrangement of the auxin gradient 311 during gravitropism making it possible to dissect the spread of auxin distribution at high spatio-temporal 312 resolution. Our setup made these findings possible, demonstrating its versatility and application to a broad 313 range of questions in developmental, cell biology and physiology. We have aimed for a very detailed 314 description that will enable other labs to implement the setup or its components, and will therefore be 315 beneficial to the Arabidopsis community as well as non-plant researchers. The voltage can be adjusted in the range of 3.5 -9.5 V. Appropriate resistors were used to reach light 338 intensities ranging from 40-180 μmol/m²/s (see Figure 3). Fish maintenance and embryo collection were carried out as previously described (Westerfield, 2007). 367 Embryos were raised in either E3 medium or Danieau's buffer, kept at 28 or 31°C and staged according to

Image analysis 375
Cell division analysis in Figure 6: A single z-section of data set number #05 was stabilized around one 376 cell file using semi-automatic motion tracking in Adobe After Effects. The image sequence was exported 377 as tif files and imported into Fiji. The area, highlighted with a dashed white box in Figure 6C