RMI-chip: a microfluidics setup for functional imaging of microbial interactions with tree roots

Coupling microfludics with microscopy has emerged as a powerful approach to study dynamics in plant physiology at cellular resolution and at higher throughput than with conventional methods. Most devices have been designed to study the model plant Arabidopsis thaliana, including root-microbe interactions. However, there is a need for microfluidic devices which enable in vivo studies of root development and root-microbe interactions in woody plants. Here, we developed the RMI-chip, a microfluidic setup that allows for continuous microscopic scale observation of real-time interactions between live Populus tremuloides (aspen tree) seedling roots and the rhizobacterium Pseudomonas fluorescens over an extended period longer than a month. Colonization dynamics in the RMI-chip are consistent with previous observations in a different experimental set-up. Also, we find that biosensors based on the rhizobacterium Bacillus subtilis can be used to monitor compositional changes in root exudates but that their application is limited by their inefficient colonization of aspen roots. Our results indicate that functional imaging of dynamic root-bacteria interactions in the RMI-chip requires careful matching between the host plant and the bacterial root colonizer.


23
The plant microbiome plays an important role in the rhizosphere [1][2][3]. Virtually all plant tissues host 24 microbes that can act as symbionts, commensals, or pathogens. Interactions between plant and microbes 25 can be beneficial, neutral, or harmful and directly influence plant growth and productivity [2,4,5]. Plant-26 growth-promoting (PGP) rhizobacteria are bacteria that exert beneficial effects on plants through direct 27 or indirect interactions with the roots [6,7]. PGP bacteria have the potential to increase the availability of 28 soil nutrients to the plant, produce metabolites such as plant hormones, elicit plant systemic defences, 29 and increase plant resistance to biotic and abiotic stresses [8,9]. In return, the plant provides 30 photosynthetically-derived carbon, such as sugars and organic acids that are consumed by rhizosphere 31 microorganisms as well as a wide range of molecular compounds acting as environmental signals for the 32 root microbiota. Microbes attach to the root surface and form microcolonies that can eventually grow 33 into larger biofilms. The formation of biofilms at root surfaces was proposed to be part of the cellular PGP 34 activities of beneficial rhizobacteria [10].

35
Understanding these complex interactions between plant roots and microbes requires the ability to track 36 their dynamics at high spatial and temporal resolution. Real time monitoring of dynamic root-microbe 37 interactions at cellular resolution is now possible using microfluidics approaches coupled with advanced 38 live imaging microscopy. Microfluidic platforms provide a powerful approach to evaluate the responses 39 of growing plant cells to external perturbations (e.g., nutrients, media flow, temperature, hydrodynamics, 40 light, and stressors) at throughputs higher than with conventional methods using pots or plates, and in 41 precisely controlled environments. Multiple microfluidics devices such as "Plant on a chip" [11], RootChip 42 [12], RootArray [13], TipChip [14], and PlantChip [15] were developed to study various aspects of the cell 43 biology of Arabidopsis thaliana, including gene expression, cell biomechanics, cellular mechanism of 44 growth and cell division (reviewed in [16]). The PlantChip device enables the continuous monitoring of 45 phenotypic changes at the cellular level and also at the whole plant level, including seed germination and 46 root and shoot growth (hypocotyls, cotyledons, and leaves) [15].
Fewer studies have used microfluidic devices to visualize the interactions of Arabidopsis roots with 48 pathogenic or beneficial microorganisms. Using the plant-in-chip platform, visualization of the attack of 49 Arabidopsis roots by pathogenic nematodes and oomycetes motile spores revealed some physiological 50 changes taking place in the host plant and the pathogen during the attack [17]. Recently, a microfluidic 51 device to track root interactions system (TRIS) revealed a distinct chemotactic behaviour of the bacterium 52 Bacillus subtilis toward the root elongation zone and its rapid colonization, and allowed real-time

59
Here, we describe a microfluidic device, called RMI-chip, that enables the direct visualization of root-60 microbe interactions taking place at early stages of tree seedling growth. We studied the interactions of 61 Populus tremuloides (trembling aspen tree) with the bacterium Pseudomonas fluorescens. These 62 interactions are biologically relevant in nature, as P. fluorescens is abundant in the rhizosphere of Populus 63 trees [1,[19][20][21], and exhibits functionality in laboratory assays. We have shown that P. fluorescens 64 promotes the growth of aspen seedlings [22], colonizes aspen seedling roots by forming dense and 65 dynamic biofilms [23], and modulates expression of anti-fungal defence response genes in roots of aspen 66 seedlings [24]. The RMI-chip device was designed to accommodate aspen seedling growth for periods up 67 to 5 weeks, and to enable direct observation of root growth and its dynamic colonisation by P. fluorescens 68 biofilms with high spatiotemporal resolution. The RMI-chip was also used to monitor the growth of rice 69 seedling roots and detect the production of reactive oxygen species (ROS) by the root using engineered 70 Bacillus subtilis strains as biosensors. We find that in the RMI-chip interactions between host plants and bacterial species are specific, consistent with ecological observations and with colonization profiles 72 observed in other experimental systems, and that formation of bacterial biofilms on root surfaces is 73 needed for extended colonization. The RMI-chip was designed to observe the growth and rhizobacterial colonization of aspen seedling roots 177 over extended periods of time. The aspen root system consists of a taproot from which smaller branch 178 roots emerge. When a seed germinates, the first root to emerge is the primary root which develops into 179 the taproot. With seedlings grown into agar-filled pipette tips (Fig. S1a), we observed that the primary 180 root quickly branched after exiting the tip (Fig. S1b). Thus, seedlings were inserted into the RMI-chip 181 shortly before the primary root reached the end of the pipette tip. Further development of the primary 182 root, including formation of branch roots, was not hampered by the pipette tip (Fig. S2a). In early studies, 183 we observed that the primary root readily entered the RMI-chip circular chamber (Fig. 1a), but we found 184 that a substantial part of the root system di not continue into the linear growth channel, instead growing 185 in circles, which caused the seedling to wither quickly. This problem was resolved by growing the seedlings 186 semi-gravitropically in a chip tilted at a 45° angle, submerged in Johnson's solution for up to a week, until 187 the root tip reached the media inlets in the growth channel (Fig. S2c). Under these conditions, most 188 primary roots entered the growth channel, at which point the RMI-chip was placed horizontally in a 189 humidity chamber and continuous media flow was supplied (Fig. 2). Regular exchange of the nutrient 190 solution surrounding the RMI-chip and the maintenance of water-saturated paper used to provide the 191 required humidity in the growth chamber, were crucial to obtain healthy seedling growth.

192
The RMI-chip has 6 independent channels with one inlet for the seedling root, two dedicated inlets with 193 filters for media and bacterial inoculation, and one common outlet (Fig. 1 a)

207
It can be challenging for plant roots to grow under constant flow. For example, without flow in a growth 208 channel, root border cells and mucilage can be observed at the root tip (Fig. S2c, Fig. S3). These cells and 209 mucilage can be washed away at high flow rates (e.g., 2 µL/min). Nevertheless, nutrient flow is needed to 210 maintain seedling growth in the RMI-chip. Therefore, we determined a minimal flow rate of nutrient 211 solution that did not affect root growth, preserving root morphology, including root cap, root hairs and 212 mucilage, while keeping out air bubbles. After several iterations, each including at least 4 seedlings, the 213 minimal flow rate was found to be 0.02 µL/min. As each channel is connected to an individual 1 ml syringe, 214 this flow rate provides enough media for up to 5 weeks. Under these conditions, the calculated average 215 laminar flow is 4 µm/s and the media is replaced every ~14 minutes in each channel. Aspen seedlings 216 exhibited a high heterogeneity in primary root growth with an average growth rate of 1 ± 0.9 mm/d. This  (Fig. S4). Finally, 233 a large closed chamber that can be filled with water was built to provide humidity during long-term 234 growth. In addition, a holder compatible with the imaging chamber was fabricated for the Nikon Ti-2 235 microscope (Fig. S4). The humidity chambers and microscope stage adapters were fabricated from PMMA

256
fluorescens SBW25 by injection of approximately 2 x 10 9 cells through an inlet port (Fig. S6a). After 257 incubation for 16 hours without flow, the flow was restored at 0.2 µL/min for 2 hours to wash away excess 258 bacterial cells, then set at 0.02 µl/min for continuous perfusion. Under these conditions, a small number 259 of SBW25 cells remained associated with the lower part of the root (Fig. S6b). Importantly, the only source 260 of carbon available for bacterial growth in the RMI-chip are the sugars and organic acids 261 photosynthetically produced by the plant. Five days after inoculation, actively dividing SBW25 cells were 262 colonizing the intercellular spaces between root epidermal cells of the root cortex, and, to a lesser extent, 263 the root surface (Fig. S7). Although this preferential colonization of crevices on the root surface may be 264 caused by the flow in the RMI-chip, it is consistent with our previous observations in a vertical plate 265 system, where SBW25 first colonizes the cell interstitial spaces along the aspen root cortex before forming 266 a variety of bacterial assemblies that range from microcolonies to highly structured biofilms [23].  Notably, we observed that a variety of SBW25 cell assemblies coexisted with biofilms on the root surface 281 (Fig. 4a). These assemblies included clusters of somewhat regularly spaced individual cells, clusters of cell 282 doublets which are presumably dividing, microcolonies and mature biofilms (Fig. S8). These assemblies 283 may reflect the different stages of biofilm formation, where cells attach to the root surface, divide, and 284 over time form more compact assemblies maturing into biofilms. Intriguingly, cells packed within a biofilm 285 appeared to have a round coccoid shape in contrast with the normal rod-like morphology of individual 286 SBW25 cells (Fig. 3a, Fig. 4a). Confocal 3D image reconstruction and side projection of the 3D volume (

291
After 30 days in the RMI-chip, fluorescent SBW25 cells became rare on the root surface and were 292 preferentially found as micro-colonies located within the spaces between root epidermal cells (Fig. S9).

293
Overall, the dynamic of root colonization by SBW25, which includes active growth and colonization of the 294 root during the first two weeks, formation of mature biofilms associated with a variety of other cell 295 assemblies, and dispersal of these biofilms and assemblies after 5 weeks, appears to be remarkably  (Fig. 5b) or ROS (Fig. 5c) by the root. Exposure of the 335 ROS biosensor to a rice primary root under the same RMI-chip conditions (i.e., no flow) resulted in a robust 336 colonization of the root and in high levels of GFP after 3 days (Fig. 5d). This finding indicates that high 337 levels of ROS are produced by rice primary root. Together, these results provide a proof-of-concept that 338 bacterial biosensors can be used in RMI-chip as a way to investigate the dynamic chemical crosstalk 339 between root and rhizobacteria.

340
We also inoculated rice seedlings with P. fluorescens SBW25 to test for ability to colonize in the RMI-chip.

341
Repeated trials showed that SBW25 cells were rapidly washed away from the root surface and remained 342 barely detectable under nutrient flow after one day (data not shown), suggesting an inability of P.