The transparent polymer PDMS, as often used for the fabrication of microfluidic environments, was shown to be not toxic for the brown algae Ectocarpus sp. (Evariste et al., 2012) as blended filaments inoculated on surfaces of different PDMS compositions grew as expected. Figure 2A shows a schematic of a PDMS device in side view. The key elements of the structure consist of the inlet, the main chamber, as well as the microchannels. The inlet has a diameter of 1.2 mm and enables filling of the chamber and loading of the samples. The height of the chamber is 100 μm to ensure sufficient availability of nutrients. Depending on the planned experiment, a design with a circular (Figures 2B,C) or triangular chamber (Figures 2D,E) can be chosen. The chamber is connected to numerous microchannels with a height of 20 μm. The reduction in height from the chamber to the channel constricts the specimen’s growth direction and, by that, allows for long-term imaging within a stable focal plane. While circular chambers connect to 44 channels with a width of 25 μm and allow for multiplex observations of the specimens, triangular chambers connected to 14 channels with a width of 40 μm to enable the detailed investigation of specific growth trajectory, e.g., filament undulations. The channels are separated by PDMS walls with a width of 20 μm, which is necessary to ensure proper sealing and adhesion between the PDMS device and the glass substrate. Figure 2D shows a schematic of a PDMS device with a triangular chamber. At the entrance of the channels, a small funnel guides filaments from the chamber into the channel, thereby enhancing the percentage of filaments growing into the channels. Once an entrance or channel is filled, a constriction at the end of the funnel should prevent an additional filament from entering the already occupied channel. Additionally, the constriction reduces the risk of specimens getting flushed through the channels during the initial filling of devices with wider structures.

The width chosen for the channels was a compromise of several criteria. On the one hand, narrower channels would increase the spatial constraints for Ectocarpus, which, despite being of broader interested concerning the resulting growth cell response, is not the focus of this initial investigation on the feasibility of lab-on-chip technology. Additionally, narrower channels would require a higher loading pressure, which could damage the samples or even separate the bond between the PDMS structure and the glass substrate. On the other hand, the use of wider channels bares the risk of samples passing through the microchannels rather than being collected in the chamber during loading. Overall, considering biological features as well as manufacturing constraints and technical challenges, we considered 25 μm to be a compromise that would allow the algae to have sufficient nutrients while ensuring a mechanically robust lab-on-chip device.

As to the chamber, its geometry was adjusted to allow a sufficiently large number of connected channels for each design while simultaneously limiting the volume available inside the chamber. Thus, it optimises the chance of obtaining filaments growing parallel to each other inside channels which is a crucial factor for the efficient study of slow-growing organisms like Ectocarpus. However, although no direct influence on the specimen’s growth was observed in relation to changing the inlet geometry, future studies could examine the impact of additional design parameters, such as the height or diameter of the inlet, on the specimen’s health and viability.

The lab-on-chips were fabricated by double-layer photolithography in clean room environments and by mould replication techniques. Silicon wafers were cleaned with acetone, isopropanol, and deionised water prior to mould fabrication. For the layer containing the microchannels, the photosensitive polymer SU-8 3025 (Kayaku Advanced Materials) was spin-coated on the wafers using a three-step process with a maximum angular velocity of 4500 rpm to obtain a uniform photoresist layer of 20 μm. The corresponding design of the channel layer was transferred onto the photoresist using a mask aligner (MA6/BA6, Süss MicroTec). Finally, the wafers were developed (mr-Dev 600, Micro Resist Technology GmbH) and hard baked at 150°C for 5 min to improve the stability of the channels and their adhesion to the silicon substrate. The second layer of the lab-on-chip, i.e., the elevated section containing the inlets as well as the main chamber, was fabricated through an additional photolithography process using SU-8 100 (Kayaku Advanced Materials) and a maximum angular velocity of 3000 rpm to ensure a uniform 100 μm thick layer of photoresist. Figure 2E shows an optical microscopy image of a fabricated mould with a triangular chamber to be used for PDMS replication. The observed variation in the focal plane is based on the height difference between the channel layer and the triangular chamber layer.

Prior to PDMS casting, the fabricated structures were vapour coated with (tridecafluoro-1,1,2,2-tetrahydrooctyl) trichlorosilane (CAS 78560-45-9, ABCR) and heated to 120° C for 10 min to improve the reusability of the mould. PDMS (Sylgard 184, Dow Corning) with a weight ratio of 1:10 (curing agent to pre-polymer) was mixed sufficiently for 5 min, poured over the SU-8 mould, and degassed for 30 min in a vacuum chamber. To transfer the 3D pattern into the polymer, the PDMS was cured for 1 h in an oven at 80°C. Finally, the PDMS was cut and peeled off the mould and the inlets were punched using a biopsy punch with a diameter of 1.5 mm (BP15, Vetlab). To chemically bond the PDMS and the cover glass with a thickness of 0.17 μm, both surfaces were exposed to oxygen plasma (Femto Plasma Asher, Diener Electronics) for 30 s before being brought in contact. Slight pressure was manually applied to ensure proper contact between the entirety of the lab-on-chip and the cover glass to prevent future algal specimens from growing beyond the dimensions of the corresponding microchannels.

For size comparisons, Figure 2F shows a ready-to-use lab-on-chip with 25 μm wide channels next to a €2 coin.

The chips were sterilised by UV irradiation for 30 min in a sterile laminar flow hood. Then, they were used either dry or pre-filled with sterile seawater (see below).

Ectocarpus strain CCAP 1310/4 (also named Ec32) from the Culture Collection of Algae and Protozoa was grown in natural seawater (approximately 550 mosmoles) supplemented by vitamins and microelements as described by Le Bail and Charrier (2013). Sporophyte filaments were produced from the germination of swimming mitospores (Charrier et al., 2008). Cultivation took place at 13°C under 12:12 light:dark conditions (light intensity of approximately 29 μE m^–2 s^–1), usually on the main types of plastic and glassware. Seawater was renewed every 2 weeks when algae were grown in open space environments.

The Leica DMI-8 inverted microscope was used to observe calcofluor (Ex/Em: 380/475 nm) and fluorescein isothiocyanate (FITC; Ex/Em: 495/520 nm) labelling with the corresponding filters. Confocal microscopy (TCS SP5 AOBS inverted confocal microscope, Leica, objective Plan-Apochromat 63×/1.4 Oil) was used to increase the z spatial resolution compared to image acquisition by epifluorescence microscopy.

Cellulose was stained by incubating the lab-on-chip in 0.01% calcofluor white solution (fluorescent brightener 28, F-3543, Sigma-Aldrich) for 30 min at room temperature (RT) in the dark. The solution was injected into the channels by high pressure pipetting. Observation was performed under UV light using epifluorescence microscopy after flushing the channels with fresh seawater at least twice and after incubation for at least 15 min at RT between each rinsing step.

Immunolocalisation of cell wall polysaccharides in the lab-on-chip was developed by adapting the protocol from Rabillé et al. (2019b). The main modification consisted in fixing the algal material in 4% paraformaldehyde prepared in H₂O instead of seawater. All the solutions were rinsed by high pressure pipetting into the inlet which occasionally resulted in the loss of some filaments.

Statistical evaluations concerning the specimens’ growth rates were performed as double-sided t-test with unequal variances for n = 15 specimens present in the free environment and n = 22 specimens in the microfluidic channels. The resulting p-values are reported in the corresponding paragraph. Statistical significance was determined as p ≤ 0.05.

The early development of the Ectocarpus sporophyte has been described in detail (Le Bail et al., 2008, 2010, 2011; Nehr et al., 2011; Rabillé et al., 2019a) and is summarised in Figure 3. When grown in open space environments, the Ectocarpus sporophyte is composed of microscopic, branched uniseriate filaments (Figure 3A) that form a visible tuft approximatively 4 weeks after the very initial stages (Figure 3B). Sporophytic filaments growth is initiated by an asymmetrical cell division of the zygote or mitospores (Figure 3C), forming the first apical cell that continues to grow indefinitely by tip growth (Figure 3D). A few hours later, the initial cell germinates again and gives rise to a second apical cell, which grows along the same axis as the first one but in the opposite direction. Over the span of several days, each first apical cell gradually differentiates into spherical cells. Apical growth and cell rounding generate a uniserial filament which is composed of two main cell types, i.e., elongated cells with a width of approximately 7 μm located at both ends of the filaments and circular cells with a diameter of approximately 15 μm situated at the centre of the filaments (Figure 3E). After around 10 days, the sub-apical cells branch (Figure 3F) and the resulting branches continue to grow as the primary filaments do. This repeated developmental programme results in the overall morphology as shown in Figure 3B.

We investigated whether these developmental steps were preserved in the constrained environment of the lab-on-chip devices. Ectocarpus spores were introduced into the structures as described in section “Materials and Methods.” We monitored growth in the lab-on-chip with the goal of constraining the spores to germinate in the chamber. The spores successfully germinated inside the main chamber of the microfluidic device 5 days after inoculation (Figure 4A). The ratio of cell division asymmetry, i.e., the ratio of cell divisions producing unequally sized daughter cells as depicted in Figure 3C, was consistent with what was reported in open space germination, where about 80% of spores divide asymmetrically and approximately 20% symmetrically (Le Bail et al., 2011). When grown freely in water deprived of microelements, it has previously been observed that filaments tend to develop a cell sheet, meaning that Ectocarpus filaments shifted their development from uniaxial to bidirectional growth. This major change in body plan organisation is a sign that tip growth mechanisms have been severely impaired in the absence of microelements. However, in none of the lab-on-chips with 40 or 25 μm wide channels were such patterns observed and the filaments maintained an uniaxial growth throughout the experiment (Figure 4B) which indicates a sufficient supply of microelements even inside the confined regions of the lab-on-chip devices. Calcofluor staining further showed that filament growth took place in the dome of the apical cell (Figure 5I) as previously observed in open space environments (Le Bail et al., 2008). To monitor specimen health, the growth rates of 22 filaments growing in 25 μm wide channels for 1 week were compared to the growth behaviour of filaments thriving in the open space environment of the same Petri dish (Figures 6A,B). Figures 6C,D showed that the overall growth dynamics were similar between the confined and free filaments, with filaments that grew inside the channels having an average growth rate of 2.8 μm⋅h^–1 compared to 2.41 μm⋅h^–1 for filaments growing in the open space (Supplementary Table 1; t-test p-value = 0.39). While the growth rates appeared to decrease with time independently of the environment (e.g., from 3.13 to 2.60 μm⋅h^–1 inside the microchannels), this reduction in growth speed did not reach statistical significance (t-test p-value = 0.15). For direct comparisons, Supplementary Figure 3 displays the average growth rate over time for both environments in a single graph. Overall, the observed growth rates occur with the same dynamic as previously reported, i.e., 2.5 μm⋅h^–1 (Rabillé et al., 2019a).

The rounding of the cells is a function of their maturation stage and their position along the filament. It is, therefore, an excellent qualitative parameter reflecting the fitness of Ectocarpus. Rounding of elongated sub-apical cells (Figure 3E) was observed as expected, so that spherical cells were located only in the most proximal parts of the filaments (Figure 4C). Additionally, branching, which is an additional cellular event reflecting the canonical developmental pattern as described in Figure 3, also occurred in the filaments developed within the channels (Figures 4C,D). Furthermore, we did not notice any colour changes from brown to green in the chloroplasts of cells growing inside the microfluidic chip, which is typically observed in correlation with filament death. Finally, under UV exposure, the intensity of the auto-fluorescence emitted by the chloroplasts was detected at a level similar to that of filaments growing in the open space environment (see Figures 5D,H).

Therefore, for spores germinating inside the chamber, neither the presence of the PDMS nor the enforced space limitations demonstrated a negative effect on the filament development and growth pattern, e.g., by impairing the quality of the seawater inside the confined environment or through a possible increase in mechanical stress. Altogether, the developmental pattern of the filaments growing inside the channels, i.e., apical cell polarisation, growth rate, and cell shape changes, was consistent with the pattern observed in filaments growing in an open environment.

Despite the recent advances of editing approaches of Ectocarpus (Badis et al., 2021), the expression of fluorescent reporter genes is not yet feasible. Therefore, the study of Ectocarpus cell biology requires the use of classical cytological approaches, like vital staining and immunolocalisation that are the only available non-invasive techniques allowing for the labelling of specific components in or at the surface of the algal cells. Cellulose, i.e., a rigid polymer of β(1–4) glucose present in Ectocarpus cell walls at a level of approximately 10% (Charrier et al., 2019), was uniformly labelled throughout filaments growing inside the 25 μm wide channels on both the outer and transverse cell walls (Figures 5A,B). This observation is in great agreement with previous studies for filaments grown in open space environment (Le Bail et al., 2008; Simeon et al., 2020). Similar results were also obtained for filaments in devices with 40 μm wide channels. Using confocal microscopy, the more focused images shown in Figures 5C–F confirmed the overall and homogeneous labelling of all filaments present inside channels. Both the shape of the chloroplasts highlighted by autofluorescence (Figures 5G,H) and the apical growth indicated by the new dark area at the filament tip (Figures 5I,J) confirmed that the filaments were thriving in this confined environment and, hence, supported the value of the previously derived growth rate measurements.

In a second step, we labelled alginates, i.e., a polysaccharide consisting of a mixture of guluronic and mannuronic acids linked by β(1–4) bond, which is present in brown algal cell walls at a ratio of up to 40% (reviewed in Charrier et al., 2019). The monoclonal BAM6, that recognises mannuronan-rich alginates, was used in combination with a secondary antibody coupled to the green fluorochrome FITC. The immunolocalisation protocol was applied in the lab-on-chip, aiming to label several filaments growing in parallel in separate channels. Simultaneously, free organisms present in the same Petri dish were labelled and used as positive controls. In contrast to the negative control without primary antibody (Figures 7A–C), the cell wall of both apical (Figures 7D,E) and rounding cells (Figure 7F) of labelled free-growing organisms displayed a specific signal similar to that reported in Rabillé et al. (2019b). In the 25 μm channels, Ectocarpus filaments also showed a strong signal in the dome of apical cells (Figures 7G–J) and on the flanks of rounding cells (Figure 7K). This pattern was as strong and specific as in the internal positive controls. The remaining cell walls of an empty plurilocular sporangium incidentally present within one of the channels also displayed significant labelling (Figure 7L). The red autofluorescence signal emitted by the chloroplasts further demonstrated the healthiness of the filaments prior to the formaldehyde fixation step of the immunolocalisation protocol (Figures 7A–J).

Overall, both cell biology protocols successfully labelled cellulose and alginate cell wall polysaccharides, either homogeneously (cellulose) or in specific locations (alginates) along the filaments. The labelling of the filaments was independent of the position of the specimen inside the lab-on-chip, e.g., at the entrance or exit of the channels, and displayed expected results even for samples located far away from the main chamber (Figure 7G).