Chip design
We used our micro-engineered silicone chip termed the “Obstacle chip”26, representing a proxy of a soil pore space system containing different sets of microstructures. The chip consists of an artificial pore system open on one side for inoculum, and it is designed to investigate the growth and dispersal behaviour of soil microbes (Supplementary Fig. 1a, b). The chip’s pore-space dimensions are optimized to match the dimensions of fungal hyphae, with structure widths ranging between 4 and 100 µm, and a uniform height of ~7 µm to aid microscopy, since cells are located in the same focal plane and rarely overlay. It contains five different geometric sections accessible by soil microbes via a common entry area (Supplementary Fig. 1a). The entry area consists of an open area with round pillars of 100 µm diameter at a separation of 100 µm, holding up the chip’s ceiling. It was cut open longitudinally with a scalpel prior to bonding (see below, section ‘Chip fabrication’), ensuring direct contact of the soil with the chip’s interior. The inner section comprises a combination of differently shaped channels and obstacles constituting five experimental sections, of which two were systematically examined in this study: (1) Section C: A set of channels with sharp corners of three different types (n = 12, randomly distributed): zigzag channels (90° turns with all channel sections at ±45° angle from the main growth direction), meandering square channels (90° turns with each section oriented in either the main growth direction or perpendicular to it), ‘z’-shaped channels (sharp corners diverting 135° from the previous growth direction, with channel sections in the main growth direction and at angles of 45° and 135° from it); (2) Section D: So-called ‘diamond channels,’ with a repeated combination of 10 µm wide and 400 µm long straight channels alternating with 140 µm wide diamond-shaped widenings. This channel type is replicated in 36 channels, each containing 33 diamond widenings. The widenings were used as quantification units to count bacteria and protist cell numbers, and for determination of liquid ingression, for the experiments on dispersal via fungal hyphae. Section A of the chip contains systems of hexagonal pillars of different diameters, Section B consists of straight channels with different widths, and Section E contains two types of obstacle courses comprised of complex structures. Those and the entrance system provided space for general observations. The design was constructed in AutoCad 2015 (Autodesk), in which patterns within experimental Sections A and C were randomized using a custom script from UrbanLISP (http://www.urbanlisp.com).
Chip fabrication
The microfluidic chips were moulded in PDMS on a photoresist master defined by UV lithography and bonded to glass slides, according to Aleklett et al.26. The master was made by spin coating a thick negative photoresist (SU-8 5, MicroChem Corp, USA) on a glass plate for 60 s at 1250 rpm. This generated a photoresist layer of ~7 µm. The photoresist was soft baked for 5 min at 90 °C on a hot plate, patterned by UV exposure (Karl-Suss MA4 mask aligner) and post-exposure baked. It was then developed for 3 min in mr-Dev 600 (MicroChem) and finally rinsed with isopropanol (VWR International). The PDMS slabs were produced by thoroughly mixing a PDMS base and a curing agent (both Sylgard 184, Dow Corning, USA) in a 10:1 ratio, followed by pouring the mix onto the master in a 4-mm-thick layer, and degassing it in a vacuum chamber at −25 kPa for 45 min. Then the PDMS was cured in an oven for 2.5 h at 60 °C. Once cooled, the PDMS was cut slightly larger than the designed pattern, covering an area of about 40 × 65 mm, and cut though the entry system, creating a lateral opening to the chip along the pillar system.
The PDMS slabs were bonded to glass slides. Glass slides, 55 × 75 mm and 1 mm thick (Thermo Scientific), were first cleaned with acetone, 75% ethanol and deionized water, and then dried under an air-blower. The pieces of PDMS and the glass slides were treated separately in an oxygen plasma chamber (Diener Electronic Zepto). For each chip, a glass slide was exposed to oxygen plasma under UV light for 1 min, followed by exposure of the PDMS piece for 10 s. Once both samples were plasma-treated, they were immediately brought in contact with their activated surfaces facing each other, and gently pressed to each other in the centre parts of the chip. To avoid collapse of the ceiling of the entrance, none of the chip edges were pressed. The chips were heated on a heating plate for ~15 s at 100 °C to ensure a proper bonding. After another 15 s, the chips with liquid treatments were filled with the different media using a micropipette, taking advantage of the PDMS’s temporary hydrophilia following plasma treatment so that liquids were readily drawn into its structures. The chips were filled with one of the following three treatments: (1) deionized water, (2) liquid malt medium, a complex medium to provide a nutrient-rich environment including reduced sugars such as disaccharide maltose and in lower proportion nitrogenous components such as peptides, amino acids purines and vitamins (malt extract for microbiology, Merck KGaA), or (3) chips were left empty, i.e., air-filled. The eight chips filled with liquid were then placed in a vacuum chamber for 30 min at −25 kPa to remove any bubbles. Finally, the chips were kept in sterile Petri dishes, sealed with Parafilm and stored overnight in a cold room before being dug down into or inoculated with soil.
Expt. 1: in situ incubation of chips
To evaluate the effect of different nutritional conditions on colonization of the soil chips by microbes, we evaluated three pore space filling treatments: (1) deionized water, (2) malt extract medium, or (3) air; n = 3 chips per treatment. The experimental site was a small grove of deciduous trees in the city of Lund, Sweden (55° 42′ 49.5′′ N, 13° 12′ 32.5′′ E; Supplementary Fig. 1c). The season chosen for burial of the chips was early autumn (October 2017) to guarantee a moist soil during the experiment. Groups of replicates of all three chip treatments were buried randomly within the inner parts of the grove (n = 3 chips per filling treatment). The litter layer was removed, and 20 × 20 cm holes were carefully dug into the ground with a spade. The chips were placed horizontally in the soil at a depth of 10 cm in which the PDMS chip was facing up and the glass slide down. Horizontal placement was chosen to probe a single stratum of the soil, serving as a comparable inoculum to the whole of the entry system, and to aid nondestructive recovery. The soil was carefully placed back in its original orientation, and the litter layer was placed back. A string attached to each chip was placed with its opposite end above the soil surface and attached to a pin, to guide future retrieval. There was a minimum distance of one meter between each chip replicate.
Preliminary experiments had shown that a 2-month incubation period would grant the colonization of different types of soil microorganisms and minerals, and a stabilization of the inner environmental conditions between the soil chip and the surrounding soil. Thus, after 64 days (December 2017), the chips were collected by carefully removing soil around the string leading to each chip. We carefully kept the adjacent soil atop the glass slide along the opening of the chip, to keep our artificial pore system connected to the real soil pore system, and to avoid such disturbances as hyphal tearing or evaporation of the liquid inside the chips (Supplementary Fig. 1d). We cleaned the chip windows by softly wiping them with a clean wipe and deionized water. Samples were carefully transported to the microscopy facilities, located adjacent to the burial site. The chips were harvested one at a time and analysed under the microscope immediately after collection and cleaning.
We recorded the presence or absence of the main soil microbial groups in the entry systems and in the different channels, including their furthest extent into the chips, with help of the internal rulers.
To analyse the effect of fungal hyphae on bacterial abundance, we recorded real-time videos slowly scanning along the whole length of the diamond-shaped opening channels (each 33 diamonds, Section D in Supplementary Fig. 1a; Fig. 3). The rather sparse hyphal colonization allowed us to select pairs of channels where in the first channel a hypha had proliferated far into the channel, combined with a directly adjacent channel without hyphae, n = 4. In each diamond-shaped widening we counted the number of bacterial cells, the presence or absence of fungal hyphae, and the presence or absence of liquid. After completion of all measurements, the chips were left uncovered at room temperature for 60 min to initiate air drying in the adjacent soil, in order to observe the real-time effects of drying on organisms and particles in the pore space system of the chips. The adjacent soil was re-wetted by adding 400 µl of water. The water inside the chips corresponded to the adjacent soil pore water, regressed upon evaporation, and refilled the chip structures upon rewetting of the adjacent soil.
Expt. 2–3: laboratory incubation of soil on chips
In a complementary approach, we collected soil from a lawn in Lund, Sweden, at 10 cm depth, and placed 5 g of this soil in front of the entry system of the chip. Chips received the three nutrient condition treatments as described above, air, water or malt medium (n = 2, Expt. 2). An additional set of air-filled chips was studied to quantify fungal highways (n = 3, Expt. 3). Chips were monitored under the microscope after inoculation, observation was documented with images and videos. Chips were kept in sealed Petri dishes with wet cotton cloths to maintain high humidity and were taken out for analysis only. The soil inoculum on the chips and the interior of the chips were kept moist with 500 µl of water added to the soil once a week. The artificial waterlogging event in the chips of Expt. 3 (‘fungal highways’) was achieved by adding a total of 2 ml of water to the inoculum soil over the course of a week, and the drying event was achieved by discontinuing the watering.
During Expt. 2, we recorded the abundance and the furthest extent of bacteria, protists (including the morpho-groups ciliates, flagellated, and amoeboids), and the extent of hyphal colonization into the diamond section over time. After 2 months of incubation, we measured the furthest extent of colonization into the angled channels for the organism groups bacteria, fungi, and protists. During Expt. 3, we recorded the presence and the furthest extent of hyphae, liquid, bacteria, and protists in the diamond channels over time. We also recorded the number of protists, bacteria (in categories 0, <5, <50, <250, <500), and the percentage of the pore filled with liquid (in categories 0, <10, <25, <50, <75, 100) in each diamond-shaped pore.
Microscopy and image analysis of the chips
All visual inspection and imaging were done with an inverted microscope (Nikon Diaphot 300) with 40–400× magnification. Bright-field images and real-time videos were recorded through the microscope with a digital camera (USB29 UXG M). Presence and abundance of organisms and liquid was counted by eye. Active bacterial cells were counted when exhibiting nonrandom movement, allowing to record cells down to a size of circa 0.4 µm in movement. The real cell number may be underestimated since small cells may have remained undetected at rest. The absence/presence of the main soil microbiota groups (fungi, bacteria, protists, nematodes, and microarthropods) was recorded in the entry system and in the different channels. Where present, the maximum individual extent of each soil biota group into the chip was recorded in the turning and diamond channels (Section C and D). Videos and images of particles and microbial interactions were recorded in the entry system and in the channels and pillar systems of the chip. All images shown in Fig. 1, except for Fig. 1c, h, were extracted from videos (Supplementary Movies 1–6, 8) using Adobe Premiere Pro CS6. All images shown in Figs. 1 and 3 were cropped in Adobe Photoshop CC (v. 20.0) and the image from Fig. 4 was cropped in Adobe Photoshop 2020 (v.21.2.4); contrast and brightness were adjusted to obtain optimal visual data. Scale bars were added with ImageJ Software (v. 1.56 h, NIH). False-color image legends were produced in Adobe Photoshop CC (v. 20.0). The central illustration and the final assembly of other elements of Fig. 1 were created in Adobe Illustrator 2020. Images shown in Fig. 5 were captured using the ImageJ plug-in TrackMate version 3.8.0.
Velocity analysis of particles in water flow paths
The velocity analysis of soil particles in soil solution streams was performed in ImageJ version 1.52i. A 48-s video clip of a drying event inside the chip was converted to an 8-bit file and subjected to a threshold of gray values from 0 to 175. This range was selected because it showed to be optimal in the number of moving particles captured. Particles were tracked with the Image J plug-in TrackMate version 3.8.0. A LoG detector, which applies a Laplacian of Gaussian filter to the image, was used to detect the particles. The parameter ‘Estimated blob diameter,’ which should have approximately the same size as the analysed particles, was set to 2.76 µm. The radius of the particles considered by the software then ranged from one-tenth to two times the radius. This value represents the quality of the detection and must be a positive number with higher numbers representing higher quality. This size range was chosen because particles smaller than 2 µm were frequently lost in the tracking algorithm, and particles larger than 6 µm commonly were aggregates that could break apart and move in parts. The quality histogram threshold was set above 37. The particles were tracked using the simple Linear Assignment Problem (LAP) tracker function, with linking and gap-closing conditions of 30 and 40 pixels respectively, and a maximum frame gap of 0 frames (frame rate: 18.3 frames per second). Velocity tracking in Fig. 5 is visualized as color-coded mean velocity tracking lines, showing individual velocities of the particles.
Raman scattering microspectroscopy in the chips
Raman scattering spectra of particles in the chips were recorded using a long working distance ×50/0.5 objective of a LabRAM HR Evolution confocal Raman microscope, equipped with 785 nm diode excitation laser, 600 g/mm diffraction grating, and front-illuminated thermoelectrically deep-cooled CCD camera Sincerity (Horiba Scientific). The confocal mode of the microscope was used, employing 50 μm diameter pinhole. Spectra in the spectral region 200–970 cm−1 were acquired with an acquisition time of 30 s and 32 accumulations, at full excitation laser power (neutral density filter T = 100 %, P785 = 100 mW). Spectra of the particles in the chip were compared with mineral reference spectra and identified as quartz. Other peaks in the spectra are assigned to the PDMS (Supplementary Fig. 2).
Statistics and reproducibility
Data presented in Figs. 2–5 (Supplementary Data 1) was statistically evaluated in JMP Pro 15.0 (SAS Institute Inc., Cary, USA) with the significance level set to α = 0.05, two-sided tests. Analysis was performed on log-transformed or square-root data if necessary to meet normal distribution of the residuals.
Expt. 1
The influence of the chip-filling treatment and the different channel shapes on the maximum colonization distances of fungal hyphae, bacteria, and protists was evaluated via a restricted maximum likelihood ANOVA model, with the full-factorial combination of the two factors ‘channel type’ (three levels: zigzag/square/z-shaped; maximum extent per channel type) and ‘nutrient treatment’ (three levels: air/water/malt medium), which were full-factorially distributed at each location (n = 3 per channel and filling type). The influence of the presence of fungal hyphae on the presence of bacteria in Expt. 1 was investigated in a paired set of channels with a hypha, against directly adjacent channels without hypha, n = 4. Data of four channel pairs from one air-filled chip were evaluated by a paired two-way t-test. In an alternative approach, individual pores were used as a replication base, n = 33 in 4 pairs in a restricted maximum likelihood ANOVA model depending on the presence/absence of a hypha and with pairs as random attribute factor, which produced similar results. The influence of the presence of fungal hyphae on the presence of water in the diamond widenings of the initially air-filled chips was analysed with a 2 × 2 contingency analysis and evaluated for dependency with Pearson’s Chi-squared test.
Expt. 2
A restricted maximum likelihood ANOVA model with the full-factorial combination of the two factors ‘channel type’ (three levels: zigzag/square/z-shaped; 12 of each channel type per chip) and ‘nutrient treatment’ (three levels: air/water/malt medium), and the factor ‘chip replicate’ (n = 2) as a random attribute, was used to test the influence of nutrient condition and channel shape on fungal, bacterial, and protist dispersal.
Expt. 3
A restricted maximum likelihood ANOVA model, with a full-factorial combination of the factors ‘time’ and ‘presence of hyphae’, and with the factor ‘chip replicate’ (n = 3) as a random attribute, was used to test the influence of fungal hyphae on water ingression, bacterial, and protist dispersal over time both on pore space (n = 1188 × 3 chips) and at channel replication (n = 36 × 3 chips). As the pore space dataset was strongly skewed due to the initially rare events of microbial colonization, we analysed the data in parallel with non-parametric Wilcoxon/Kruskal Wallis tests and with a logistic regression using the presence/absence of hyphae and bacteria or protists, which gave similar results in terms of which comparisons were significantly different.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
Source: Ecology - nature.com
