Machine-assisted cultivation and analysis of biofilms

Biofilms are the natural form of life of the majority of microorganisms. These multispecies consortia are intensively studied not only for their effects on health and environment but also because they have an enormous potential as tools for biotechnological processes. Further exploration and exploitation of these complex systems will benefit from technical solutions that enable integrated, machine-assisted cultivation and analysis. We here introduce a microfluidic platform, where readily available microfluidic chips are connected by automated liquid handling with analysis instrumentation, such as fluorescence detection, microscopy, chromatography and optical coherence tomography. The system is operable under oxic and anoxic conditions, allowing for different gases and nutrients as feeding sources and it offers high spatiotemporal resolution in the analysis of metabolites and biofilm composition. We demonstrate the platform’s performance by monitoring the productivity of biofilms as well as the spatial organization of two bacterial species in a co-culture, which is driven by chemical gradients along the microfluidic channel.


Machine-assisted cultivation and analysis of biofilms
(e, f) Optical fibre micro sensors for oxygen measurement are moveably mounted within the cannula. (h) Representative image of a cultivation gasket for anoxic biofilm cultivation with a controlled gaseous phase and online oxygen measurement inside the microfluidic chip that is enclosed in the anaerobic chamber. (g) Overview image of a typical setup for long-term cultivation. A multi-channel syringe pump, equipped with large syringes, is used for constant medium supply. The chips are connected to the syringe through three-way-valves, to enable exchange of syringes during prolonged cultivation times. The chip is placed in a temperature-controlled humidity chamber to warrant constant environmental conditions. The effluent is collected in a waste flask.  Figure S1D. The entire chip was held under anoxic conditions using nitrogen gas and flushed with air-saturated LB medium. In the standard PDMS chip (green), significant oxygen concentrations are detectable only at high flowrates (> 50 µL/min), while coverage with a glass slide led to increased oxygen levels, even at lower flowrates (> 30 µL/min). In contrast, a significantly lower depletion of oxygen was determined for the Noa81 chip, where differences between the inlet and the outlet occurred only at very slow flowrates (< 4 µL/min). (b) Oxygen concentrations measured at the outlet of a standard PDMS (red curve) or a Noa81 (black) meander chip during the cultivation of E. coli biofilms under flow conditions. The chips were kept under oxic environmental conditions at room temperature and constantly perfused with air-saturated medium (20 µL/min, LB medium). As expected, the growth of E. coli biofilms led to consumption of oxygen, which was exhausted (< 0.02 % oxygen) after 22 h or 35 h of cultivation in the Noa81 or standard PDMS chip, respectively. Examination of the chips after 40 h with optical microscopy revealed only few cells in the Noa81 chip (d), whereas a thick biofilm layer was visible in the PDMS chip (c).

Experimental details and discussion
Possible strategies to reduce the gas permeability of the cultivation chip were evaluated by abiotic oxygen monitoring experiments described above. The PDMS body was bonded to cover slips from both sides, the top and the bottom of the demolded block (double sealed meander flowcell). And secondly, flowcells were fabricated in the commercially available thiolen polymer Noa81. This material was chosen because it offers several advantages like biocompatibility 2,3 , transparency, low price and, most importantly, low gas permeability. 2 Furthermore, there are techniques available for the bonding of Noa81 structures to glass slides. 3 To characterize the oxygen content of the medium inside the channel when enclosed in an anoxic environment, abiotic experiments with online oxygen measurement were conducted. To this end, the various meander flowcells were placed in the anaerobic chamber which was constantly perfused with N 2 gas (Figure S1 h) and oxygen saturated medium was pumped through the channel. The oxygen content in the inlet and outlet chamber of the channel was measured online by optical sensors (Needle type micro sensor, OXY-10 Transmitter, Presens, Germany) and thereby the depletion of oxygen through the chip could be measured ( Figure S1 e, f, h). The oxygen sensors were calibrated by a two-point method using oxygen saturated medium as 100 % and saturated sodium dithionate solution as 0 % reference. In Figure S2a the oxygen concentrations at the inlet and outlet chamber of the meander channels in the three mentioned variations are plotted after a sufficient equilibration period at various flowrates.
To evaluate whether a self-induced oxygen gradient can be achieved in the meander flowcells, despite the high gas permeability of the PDMS, a cultivation experiment was conducted. For comparison, E. coli biofilms were cultivated in PDMS as well as in a Noa81 meander chips. The oxygen content was monitored at the outlet throughout the whole cultivation. Figure

Experimental details and discussion
The chip-to-LHS interface consisted of the flowcells, the inlet and outlet structures, connecting cannulas and the three-part cartridge. One cartridge can hold up to three cultivation chips for automated handling on the LHS ( Figure S4). The PDMS structures for inlet and outlet were produced by solution casting of PDMS. The replication masters for this process as well as the flowcell cartridges were produced by 3D-printing (Sculpteo, France). The pipetting needles of the LHS could enter in the cone shaped connector ports of the inlet structure to establish the leak-free but reversible connection between the pipetting needles and the fluidic chip ( Figure S4 c, d). The design of the PDMS inlet structure was adopted from the "microfluidics-onliquid handling station" (μF-on-LHS) concept introduced by Waldbaur et al. 4 . The outlet structure was designed suitable for storage of up to 800 µL liquid in each cavity (Figure S4 b). The accumulated liquid could be removed from the cavities by the pipetting needles. The fully assembled fluidic system (injection structure, up to three flowcells, outlet structure) was placed in the carrier to allow for automated transfer by the robotic manipulator arm (RoMa) of the LHS ( Figure S4e). The cartridge system comprised of the bottom plate with positioning recesses, a cover with access holes for the pipetting needles and a removable lid to enable optical analysis (Figure S a). The cover provided light protection and prevented lifting of the fluidic system from the bottom plate.
For automated execution of the FISH protocol, the liquid handling station Tecan Freedom Evo 200 was equipped with two temperature control systems with a coupled microplate/eppendorf tube carrier and a custom-made modification for a microplate carrier, which prevented the cartridge from lifting, when the pipetting needles were retracted. One temperature control system was constantly cooled to 4 °C and used for storage of heat-sensitive reagents (hybridization buffer, DAPI solution), the second temperature control system was initially cooled to 4 °C for the fixation step of the FISH assay and afterwards heated to 48 °C for hybridization. Temperature adjustments and all other process steps were controlled by custom-made .exe files initiated by the Evoware software. The automated FISH procedure followed the protocol from Pernthaler et al. 5 All reagents were injected to the channels by the LHS with a flowrate of 50 μL/min. Topographic representation (6.7  3 mm², height 0-100 µm, as indicated by color scale) of the 3D structure of a mixed species E. coli and B. subtilis biofilm cultured in the straight cultivation chip obtained by OCT during the automated FISH procedure captured. To analyze the structural changes of the biofilm, which may occur during the various treatments of the automated FISH procedure, images were acquired immediately before FISH (a), after fixation (b), permeabilization (c), hybridization (d) and washing (e) steps. Although slight biofilm detachment was observed in selected regions, the overall structural integrity remained preserved throughout the entire procedure.

Experimental details and discussion
To prove the structural integrity of the biofilms during the automated FISH procedure, we monitored the mesoscopic structure of the developing mixed biofilms (E. coli & B. subtilis) by means of optical coherence tomography (OCT) during the entire experiment.
Briefly, OCT is an interferometric imaging modality capable of visualizing biofilms completely 6,7 . In this study a GANYMEDE-II spectral domain OCT was applied (Thorlabs GmbH, Dachau, Germany). It acquires three-dimensional structural datasets at the mesoscale (mm-range) with high lateral (≤ 12 µm/pixel) and axial (≤ 3.1 µm/pixel) resolution at high speed, in situ, and without any sample treatment directly inside the cultivation device. Thus, the biofilm structure after each step of the automated FISH procedure was acquired and stored in a three-dimensional dataset. These datasets were analysed using ImageJ. 8 After cropping the datasets to the flow channel of the microfluidic device and binarization using Otsu's method 9 a topographic representation of the bulk-biofilm interface was calculated. Figure S5 shows the subsequent OCT images during automated FISH procedure as well as an example of an OCT image in conjunction to the FISH images itself. A lot of planktonic cells were visible before the protocol was started (green, lose parts), which were washed out in the fixation step. Subsequently, there was no obvious association between biomass loss and one specific step, rather than a continuous slight erosion of biofilm throughout the whole procedure. We observed partly strong reflexion effects on the glass bottom of the cultivation channel which prohibited the exact quantification of the attached biofilms. In the first phase (1) the stress increases due to elastic deformation of the PDMS layer. A small first stress release occurred when the PDMS is ruptured (2). Subsequently, the detected force increases again due to friction (3). When the tip of the cannula enters the channel (4), the force value drops until the glass bottom of the cultivation channel is reached (5), thus leading to a sudden increase in the detected stress. Figure S9: Image series of the probing cannula entering the cultivation channel (a) and depositing 10 µL samples into wells of a 384 well microplate plate (b). Figure S10: Graphical user interface (GUI) for the control of the biofilm sampler. The software was realized in Visual C#, an object-oriented programming language. (a) The GUI is divided into direct and script control elements. In the upper part, various elements allow the direct manipulation of the different components, i.e., positioning stages, the pumping unit, or the camera installed on the sampling probe ( Figure S7). The lower part contains the elements necessary to control the platform. Every usable lab ware is assigned to a defined ID in the dynamic representation of the robotic deck, which is also organized in six blocks. For each lab ware, the corresponding position data is stored and can be retrieved under the respective ID. Depending on the experiment, the user can freely choose between a chip holder, a microplate holder and two reaction tube holders. The images b) -e) show detailed views of the control elements for the positioning system (b), the pump (c), the loadcell (d) and the script table (e). The commonly used spreadsheet program Excel can be used to create script files for control of the biofilm sampler. Prescribed script modules can be used to enable various procedures. The modules can be customized by numerous freely definable variables, such as volume and flowrate, directly in the Excel template. The finished script is then imported to the script table (e).