Spatial organization and proteome of a dual-species cyanobacterial biofilm alter among N2-fixing and non-fixing conditions

ABSTRACT Many disciplines have become increasingly interested in cyanobacteria, due to their ability to fix CO2 while using water and sunlight as electron and energy sources. Further, several species of cyanobacteria are also capable of fixing molecular nitrogen, making them independent of the addition of nitrate or ammonia. Thereby they hold huge potential as sustainable biocatalysts. Here, we look into a dual-species biofilm consisting of filamentous diazotrophic cyanobacteria Tolypothrix sp. PCC 7712 and heterotrophic bacteria Pseudomonas taiwanensis VLB 120 growing in a capillary biofilm reactor. Such systems have been reported to enable high cell densities continuous process operation. By combining confocal laser scanning and helium-ion microscopy with a proteomics approach, we examined these organisms’ interactions under two nitrogen-feeding strategies: N2-fixing and nitrate assimilation. Not only did Pseudomonas facilitate the biofilm formation by forming a carpet layer on the surface area but also did N2-fixing biofilms show greater attachment to the surface. Pseudomonas proteins related to surface and cell attachments were observed in N2-fixing biofilms in particular. Furthermore, co-localized biofilm cells displayed a resilient response to extra shear forces induced by segmented media/air flows. This study highlights the role of Pseudomonas in the initial attachment process, as well as the effects of different nitrogen-feeding strategies and operation regimes on biofilm composition and growth. IMPORTANCE Cyanobacteria are highly interesting microorganisms due to their ability to synthesize sugars from CO2 while using water and sunlight as electron and energy sources. Further, many species are also capable of utilizing molecular nitrogen, making them independent of artificial fertilizers. In this study, such organisms are cultivated in a technical system, which enables them to attach to the reactor surface, and form three-dimensional structures termed biofilms. Biofilms achieve extraordinarily high cell densities. Furthermore, this growth format allows for continuous processing, both being essential features in biotechnological process development. Understanding biofilm growth and the influence technical settings and media composition have on biofilm maturation and stability are crucial for reaction and reactor design. These findings will help to open up these fascinating organisms for applications as sustainable, resource-efficient industrial workhorses.

might play a part in a future bioeconomy as solar cell factories due to their ability to grow without organic carbon and nitrogen compounds, even on non-arable land.
Despite these promising features, their effective biotech application is hampered due to low biomass, insufficient product titers, and reaction instabilities (4). However, growing cyanobacteria as biofilms has been demonstrated to be an effective method to address the problem of reaching high cell densities resulting in up to 52 g cdw L −1 (cdw: cell dry weight) (5), meanwhile systems working with suspended cultures could reach up to 4-8 g cdw L −1 (4,6). Biofilms consist of microorganisms embedded in a three-dimensional self-produced extracellular polymeric matrix (EPS). The bacterial communities embedded within the EPS are more resistant to stress conditions like toxic substrates/products. They are self-regenerating which makes them very appealing for biotech applications. Cells in biofilm growing communities are in close contact with each other and have established various inter-species communication strategies, which may affect the community structure (7,8). Excellent examples of this interaction dynamics are microbial mats, which can be regarded as huge biofilms composed of various specialized microbial layers, including cyanobacteria as primary producers (9).
Recently, capillary biofilm reactors (CBR) have been introduced as an easy-to-maintain system for the study of cyanobacterial biofilms in a biotechnology context (5, 10). The CBR system is a minimized tubular reactor system, which is continuously flushed with medium and thereby represents a plug flow reactor with biomass retention. It can be operated in single or multi-phase mode, the latter involving the introduction of gas segments into the medium flow. These segments serve multiple purpose. They add hydrodynamic forces to the system, which facilitate mass transfer in the biofilm and lead to stronger attachment of the cells to the surface. Moreover, the gas segments extract excess oxygen and, depending on the respective experiment, supply gaseous substrates such as N 2 or CO 2 (11).
In previous studies using a CBR, it was demonstrated that various cyanobacteria grow more biomass in the system when co-cultivated with Pseudomonas taiwanensis VLB120 (5,10). Remarkably, P. taiwanensis VLB120 was using metabolites supplied by the cyanobacterial partner as sole sources of carbon and nitrogen. Among the investigated strains, the filamentous N 2 -fixing cyanobacterium Tolypothrix sp. PCC 7712 showed an exceptional performance in terms of biomass formation and cell retention (10).
Here, we report on the spatial organization in an early-stage (young) biofilm composed of Tolypothrix sp. PCC 7712 and P. taiwanensis VLB120 cultivated in a continuous flow-through system and how different growth conditions influence this spatial organization. Previously established custom-made flow-cells (12) resembling CBR geometry served as confocal scanning laser microscopy (CLSM) compatible cultivation device. CLSM and helium-ion microscopy (HIM) allowed for monitoring the intra-biofilm organization under N 2 -fixing and non-fixing conditions. Based on the microscopic data it was possible to calculate the biovolume of the different species and conclude on population dynamics. The biovolume can be defined as the total volume of all cells, including Tolypothrix sp. and Ps_egfp cells. We show that P. taiwanensis VLB120 is highly important for the initial biofilm attachment, forming a seeding layer directly on the attachment surface. This effect is even more prominent under N 2 -fixing conditions, when the medium is lacking nitrate as a nitrogen source in addition to the organic carbon. Localization of P. taiwanensis VLB120 in close proximity to Tolypothrix sp. enhanced attachment of the organisms and prevented flush-out upon segmented flow conditions. Shotgun proteomics revealed several P. taiwanensis VLB 120 proteins exclusively abundant in N 2 -fixing conditions, which might pose engineering targets to tune biofilm attachment forces in the future.

Cultivation and maintenance of microorganisms
Pseudomonas taiwanensis VLB120_egfp (hereafter referred to as Ps_egfp; inhouse strain collection (5), carries a chromosomal integrated constitutive egfp gene for synthesis of "enhanced green fluorescence protein" (eGFP) (14) important for fluorescence imaging (see below). Ps_egfp was cultivated on LB agar media, prior to overnight cultivation in LB broth (30°C/200 rpm in a Multitron Pro Incubator, Infors HT, Switzerland). Subsequently, 200 µL of pre-culture was transferred to 20 mL M9 minimal medium and cultivated under the same conditions. This cultivation step was repeated once more before mixing the Ps_egfp with the cyanobacteria in BG11 or BG11-0, depending on the experiment. Optical density was measured at 450 nm using a spectrophotometer (Libra S11 Visible spectrophotometer, Biochrom, UK) based on previous studies (5,15,16).
Tolypothrix sp. PCC 7712 (hereafter referred to as Tolypothrix sp.) was purchased from Pasteur Culture Collection (PCC). After cultivating it from the -80 cryo stocks on BG11-0 and BG11 agar media with 25 µEm −2 s −1 illumination, it was transferred to BG11-0 or BG11 broth media, depending on the experiment, and cultivated for 5 days under standard conditions (continuous 50 µEm −2 s −1 illumination; 75% humidity) without shaking (Multitron Pro Incubator; Infors HT, Switzerland). Five milliliters of this pre-culture was then transferred to 45 mL fresh BG11-0 or BG11 media, respectively. After 3 days of cultivation under standard conditions, the entire 50 mL Tolypothrix sp. culture was suspended by flushing it through 10 mL syringes equipped with stainless steel needles to prevent floc formation. Then, the culture was left in the same media for two more days under standard conditions before mixing with Ps_egfp.

Preparation of the inoculum for flow-cell and CBR
Flow cell and CBR were operated in the same system. Only the cultivation chamber was adjusted (flow cell channel dimensions, 65 mm: 3 mm: 4.5 mm-length: height: width; or capillary dimensions, 200 mm: 3 mm-length: diameter, Fig. S1), while the operation procedure stayed the same. Cyanobacteria cultures were pelleted by centrifugation (5,000 g, 10 min, RT) and resuspended in fresh BG11-0 (for N 2 -fixing conditions) or BG11 media (for non-fixing conditions), respectively by adjusting their final Chla concentration to 8 µM. Ps_egfp cultures were washed two times with either BG11-0 or BG11 medium after measuring OD 450 of the initial culture. Cells were concentrated in the same medium. Final cell concentration was adjusted to OD 450 to 2.0. All photometric measurements were conducted with diluted samples fitting to the interval between 0.05 and 0.5 to prevent extrapolation, following Lambert-Beer's law. Consecutively, 5 mL of cyanobacteria culture (8 µM Chla) and 5 mL of Ps_egfp culture (OD 450 of 2) were mixed. After mixing, a final ratio of 4:1 ([Chla]: OD 450 ) was obtained and cultures were left for static incubation overnight at standard conditions before inoculating the flow-cells via a syringe through the inoculation port (12). Flow-cells or capillaries were covered with aluminum foil and incubated overnight in the dark. Subsequently, the media flow was started at 52 µL min −1 and the system was put under constant illumination at 50-60 µE m −2 s −1 . After 4 days of single-phase flow (only medium), the airflow was started at 52 µL min −1 (equal volume as liquid phase).

Operation of biofilm flow-cells and CBR
Biofilms were cultivated in custom-made flow cells (channel dimensions 65 mm: 3 mm: 4.5 mm-length: height: width) to analyze them via confocal laser scanning microscopy (CLSM) as described previously by David et al. A peristaltic pump (ISM939D; Ismatec, Germany) was used to pump media and filtered air to the system. Two flow-cells, each containing two channels, were operated in parallel: one with BG11-0 for N 2 -fixing conditions; one with BG11 for non-fixing conditions. Both flow-cells were flushed with the respective media for 2 hours for surface conditioning prior to inoculation. CBR experiments (capillary dimensions 200 mm: 2.8 mm-length: diameter) were conducted as described previously (10) with the same conditions applied in flow-cells experiments except for the cultivation period which was 1 month.

Confocal laser scanning microscopy
Fluorescence imaging was conducted making use of the autofluorescence of Chla and phycocyanin in Tolypothrix sp. and the eGFP reporter in Ps_egfp. The biofilms were visualized on a confocal laser scanning microscopy (CLSM) (Zeiss LSM 710 NLO, Germany) equipped with laser lines 488 nm (eGFP) and 633 nm (PC + Chla) lasers, and a LD C-Apochromat 10 ×/1.1 W objective. Filters targeting green emission (513-611 nm) and red emission (647-725 nm) was set for the detection. The images were analyzed using IMARIS 8.2 (Bitplane AG, Switzerland).
Each image was divided into three segments in z-dimension: 0-10 µm, 10-20 µm, and 20-30 µm. First, an isosurface was created using the red channel (633 nm excitation) with a quality threshold of over 30. All voxels belonging to the green channel (488 nm excitation) inside the isosurface of the red channel were set to zero via masking feature of the IMARIS 8.2 in order to prevent overestimation of total volume of eGFP cells. Subsequently, masked green channel was used to create another isosurface with a quality threshold of over 10, 12, and 14 for the depth volumes (20-30 µm), (10-20 µm), and (0-10 µm), respectively. Results of the isosurfaces for the different segments of each image were exported as a CVS file, including each isosurface area and their sum values. The derived isosurface based volume data for each image were imported to Origin 2019 (OriginLab Corporation, USA).
Microsoft Excel 2019 (Microsoft, USA) was used to organize the raw data obtained from IMARIS 8.2 before transfer to Origin 2019. Origin 2019 was used for all Whiskers box plots and the calculation of statistical parameters. T-test was applied for the comparison of biovolumes in different depths. Images were taken from six randomly chosen locations from each flow-cells containing two channels.

Biofilm analysis using helium-ion microscopy
HIM was used to analyze the biofilm grown in the CBR. Silicone was used as capillary material to allow sample preparation for the HIM. Chemical fixation of the biofilms in the capillaries was achieved by pumping 2% paraformaldehyde (PFA) dissolved in cacodylate buffer through the system after 4 days of biofilm cultivation. After the capillaries were filled with the fixation solution, they were clamped tightly and incubated at 4°C overnight. Later, capillaries were gently dehydrated in ethanol series by pumping solutions (10%, 20%, 40%, 60%, 80%, 100%) ethanol in cacodylate buffer into the capillaries and incubating each for 15 min. Then, the ethanol was replaced by a 1:1 mixture of ethanol: hexamethyldisilazane (HMDS) and subsequently pure HMDS, by incubating for 15 min, each followed by air-drying overnight. The dried capillary-samples were cut along the cylinder-axis by vibratome in order to allow for accessing the biofilm on their inside. Finally, the cut-open capillaries were glued onto standard electron microscopy stubs (10 mm) with conductive silver epoxy. Then the samples were ready for HIM imaging. For that a Zeiss Orion NanoFab (Zeiss Peabody, MA, USA) was used. The ion landing energy was set to 25 kV and the beam current amounted to approximately 0.5 pA. For image acquisition secondary electrons were detected with an Everhard-Thornley type electron detector.

Proteomics sample preparation and data analysis
For proteome analysis, biofilm biomass was removed manually from the CBR by scrubbing the walls with a syringe, and suspending it in 20 mL of BG11-0 or BG11 media, respectively. Four milliliters biofilm suspension was centrifuged (5,000 g/10 min/RT) and the resulting pellet was immediately frozen in liquid nitrogen and kept at −80°C until further processing steps.
Fifty microliters of ammonium bicarbonate solution (50 mM) was added to the frozen sample before it was subjected to three rounds of freeze-thaw followed by 10 min sonication. Centrifugation (10,000 g /5 min/4°C) removed cell debris and the protein containing supernatant was mixed with 50 mM dithiothreitol and incubated in a thermomixer (30°C/1 h/400 rpm). Reduced samples were alkylated by adding 400 mM 2-iodoacetamide (RT/1 h/400 rpm/dark) followed by the addition of 6.3 µg of trypsin (Proteomic Sequencing Grade, Promega GmbH, Germany) and overnight (37°C) incubation. The reaction was halted by adding 1 µL of formic acid (100%). Subsequently, the sample was desalted on a ZipTip-µC 18 column (Merck Millipore, Merck KGaA, Germany). Eluted samples were dried via vacuum centrifuge.
Extracted and cleaned samples were resuspended in 0.1% formic acid. Analysis was done via nano-liquid chromatography coupled to tandem mass spectrometry (nLC-MS/MS) using an Orbitrap Fusion Trihybrid MS (Thermo Scientific, Rochester, NY, USA) connected to a nano-liquid chromatograph (Dionex Ultimate 3,000RSLC, Thermo Scientific, USA). LC-MS/MS parameters were set according to Seidel et al. (17).
Proteome Discoverer (version 2.2, Thermo Fisher Scientific) was used to identify peptide spectrum matches, peptides and proteins from mass spectrometric analysis. As a search database we used the concatenated NCBI databases for Pseudomonas VLB120 (NCBI: txid69328, accession numbers CP003961.1 [genome] and CP003962.1 [plasmid]) and SequestHT as a search engine. Up to two misses of trypsin cleavage were permitted. Mass tolerances of 10.0 ppm and 0.2 Da were accepted for the detection of precursor and fragment ions, respectively. By comparing q values of hits to target and decoy databases, the Target Decoy PSM Evaluator node of Proteome Discoverer evaluated peptide spectrum matches at a false discovery rate (FDR) of 1%. A dynamic modification was chosen as oxidation of methionine residues, and a fixed modification was selected as carbamidomethylation of cysteine residues. Proteome Discoverer's Minora node was operated to compute protein and peptide abundance values applying intensity-based label-free quantification and utilizing the internal standard Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) obtained from Staphylococcus aureus. In Microsoft Excel, protein abundance rankings were generated based on abundance values retrieved from the Proteome Discoverer Minora node. Each protein was ranked within all proteins of a particular sample, with the most abundant protein receiving rank number 1.

RESULTS
To unravel the population dynamics of dual-species cyanobacterial biofilms during initial attachment, the respective organisms were cultivated in custom made flow-cells and analyzed via CLSM. Biofilm development by Tolypothrix sp. PCC 7712 and Ps_egfp was mapped via CLSM during the first 8 days of cultivation as dual-species biofilms growing in N 2 -fixing and non-fixing (nitrate containing medium) conditions. Creation of isosurfaces followed by 3-D imaging of the samples allowed for calculation of the biovolumes of Ps_egfp and Tolypothrix sp. at different depths (10, 20, 30 µm, Fig. 1A) and thus revealed the distribution of the two species over the first 8 days of biofilm development (Fig. 1). Biovolume refers to the total volume of all cells, either Tolypothrix sp. or Ps_egfp, respectively. Ps_egfp was limited in terms of carbon and nitrogen source in every condition. It was found that Ps_egfp did not grow in M9 medium when the nitrogen source was changed from NH 4 Cl to NaNO 3 under aerobic conditions (data not shown). Several Pseudomonas species have previously been reported to be capable of the assimilation of nitrate under anaerobic conditions (18)(19)(20). As demonstrated in a previous study using capillaries containing Ps_egfp and Synechocystis sp. PCC 6803, anoxic conditions cannot be achieved if organic carbon sources are insufficient for Ps_egfp to consume oxygen (5). In previous experiments, Tolypothrix sp. PCC 7712 and Ps_egfp containing biofilms always accumulated some oxygen in the gas phase (Table  S4). The lack of organic carbon source in our experiments means that Ps_egfp cannot create an anaerobic environment in which nitrate could be assimilated.

Pseudomonas sp. forms a seeding layer directly on the cultivation surface (0-10 µm)
Especially in the initial phase of biofilm formation Ps_egfp was highly abundant in the area close to the attachment surface (0-10 µm). The population did not grow and was maintained at more or less the same level also after segmented flow was started. This is not surprising, as no organic carbon source was available for Ps_egfp and organisms Please note the different ranges of the y-axis.

Research Article mSystems
were solely relying on the compounds excreted by Tolypothrix, which at this time point was not present in high numbers. The surface coverage at 0-10 µm by Ps_egfp was in the range of 3-10% after 8 days when cultivated in media containing nitrate as nitrogen source for the cyanobacteria partner (Ps_egfp is not capable of using nitrate as nitrogen source under aerobic conditions). Correlating to the distance from the flow cell surface (Fig. 1A), this value decreased to 0.1-1.3% in the most distant layer. Under N 2 -fixing conditions, this observation was even more prominent. Ps_egfp maintained a higher total biovolume even after segmented flow started (Day 5), suggesting a stronger attachment compared to nitrate-fed environments (Fig. 1B) which showed a statistically significant difference between (0-10 µm) and (20-30 µm) (Day 5, P < 0.05, Table S5). Furthermore, mean values of biovolumes (0-10 µm) were higher by approximately 2-fold in N 2 -fixing conditions after 5 days of cultivation (P < 0.05), a trend also visible in the more distant layers. These results clearly demonstrate that the abundance of Ps_egfp was reaching up to 100 times higher values close to the attachment surface (0-10 µm) than in deeper layers of the biofilm (20, 30 µm), independent of the applied growth condition.
Tolypothrix sp. on the other hand grew linearly in biovolume over the cultivation time of 8 days in all levels analyzed (Fig. 1C). It showed a reverse behavior compared to Ps_egfp, with increasing abundance towards the outer levels of the biofilm closer to the light source. When nitrate was added to the medium, deviations in biofilm biovolume values was enhanced and values fluctuated significantly (Fig. 1C) which resulted in statistically non-significant comparisons (P > 0.05, Table S5). Nevertheless, also under these conditions a linear increase in mean values of Tolypothrix sp. biovolume could be detected. The results show that Tolypothrix sp. grows faster in nitrate-fed biofilms, but is more stable when fixing molecular nitrogen. The uptake of nitrate is a less costly process for cyanobacteria than N 2 -fixation. Therefore, it was expected that cyanobacteria would grow faster in nitrate-fed biofilms. However, the big deviation observed in the biovolume content in the nitrate-fed biofilms indicate that there are strong fluctuations in detachment events under these conditions. The ratios of Ps_egfp : Tolypothrix sp. (Ps : To) reflect the above described findings (Fig.  1D). Surprisingly, Ps_egfp outcompetes Tolypothrix in terms of biovolume during this initial biofilm development stage, especially directly at the surface layer, where the ratio Ps : To is still 5:1 under N 2 -fixing conditions after 8 days of cultivation (P < 0.05, Table  S5). However, for all conditions and in all levels analyzed, the initially high ratio of Ps : To is decreasing over time, especially after segmented flow was started and it is to be expected, that in the course of biofilm development Tolypothrix will be the prominent organism.

Co-localized cells resisting to the segmented flow
Segmented flow is an important feature of operating CBRs and has been shown to be beneficial for biofilm development in previous studies (5,10,11). Therefore, the impact of segmented flow on the spatial localization of the organisms relative to each was analyzed.
Starting the airflow on day 4 and operating the system under segmented flow conditions promoted biofilm stability and reduced biovolume fluctuations significantly. Under N 2 -fixing conditions, the air segments not only increased hydrodynamic forces but also delivered molecular nitrogen for the nitrogen fixing reaction (Table S4), resulting in a more stable biofilm and an increase in Tolypothrix sp. biovolume (10). As a result of segmented flow, mean values of Ps_egfp biovolume of decreased and remained stable for three more days (Days 5-8) in nitrate-fed biofilms. This effect was especially obvious in the lower levels of the biofilm where Ps_egfp was most abundant. However, N 2 -fixing biofilms demonstrated greater resilience to segmented flow by maintaining biovolumes more or less at a constant level before and after segmented flow started.

Research Article mSystems
For better resolution, this experiment was repeated in the CBR and the biofilm was analyzed via HIM. In the HIM micrographs, Ps_egfp cells attached to the EPS structure around Tolypothrix sp. filaments are visible (Fig. 2C) on day 4 of biofilm cultivation. Moreover, a layer of Ps_egfp embedded in the EPS matrix covers the attachment surface, supporting the conclusion on the conditioning film formation of Ps_egfp. Surprisingly, after starting the segmented flow on day 4, Ps_egfp localized in immediate vicinity of Tolypothrix sp. stayed fixed in their position, while other Ps_egfp were flushed out in significant numbers ( Fig. 2A and B; Day 5; Fig. 3A). Before segmented flow started, green fluorescence was observed randomly in almost every position indicated by the blue arrow in Fig. 3, on the contrary, red fluorescence emitted by Tolypothrix sp. showed intensities only at certain positions (Fig. 3B). After the segmented flow started, the high red fluorescence intensities did not change the position pattern in contrast to the green fluorescent signal. These drastically reduced especially in the positions where no co-localization with red fluorescing Tolypothrix sp. occurred. Although this behavior was more pronounced when Tolypothrix sp. had to fix molecular nitrogen, it was also observed in the presence of NaNO 3 . Ps_egfp cells were apparently stronger attached when being localized in close proximity of Tolypothrix sp. Thus, segmented flow stabilized the overall biofilm and reduced the variations in biovolume values by flushing out loosely attached cells from the system. Furthermore, whole-image

Surface and substrate binding proteins found in N 2 -fixing biofilms
Shotgun proteomics were conducted to further access the effect N 2 -fixation has on cell attachment. To gain enough biovolume, the CBR system was used for cultivation and biofilms were cultivated for 1 month before harvest. As both species of the consortium have been completely sequenced ((21) ; Genbank accession: GCA_025860405.1 and GCA_000494915.1), it was possible to allocate many of the proteins to the respective species. Here, we focused on proteins originating from Ps_egfp to find possible explanations regarding microscopy results. In this context, several proteins originating from Ps_egfp were identified only in N 2 -fixing biofilms but not in nitrate-amended biofilms (Table 1), including some related to cell-cell or cell-surface interaction. The OmpA family protein was identified in previous studies to be important for outer membrane integrity and resistance against environmental stress (22). The peptidoglycan binding LysM domain contains a motif which binds to cell walls and it is commonly found as "cell wall binding domain" in several species (23). LysM domains can recognize and bind several different ligands containing N-acetylglucosamine (24)().
Moreover, the extracellular binding protein (PVL_06300) was detected exclusively in N 2 -fixing conditions yet similar proteins were identified in NaNO 3 fed biofilms as well. Other interesting proteins detected in only N 2 -fixing biofilms include malate dehydrogenase (PVLB_01645), PhaF protein (PVLB_02145), and polyhydroxyalkanoate (PHA)   Research Article mSystems granule-associated protein (PVLB_02140). Possible reasons behind their abundances in this specific condition are discussed in detail in the next section. Furthermore, the detachment of Ps_egfp was monitored in the CBR via plating (track-dilution method [24], Table S3) of the flow through followed by colony counts. Around 2.92 × 10 8 ± 3.20 × 10 7 and 2.03 × 10 9 ± 2.09 × 10 8 CFU Ps_egfp /L have been detected for N 2 -fixing and nitrate-amended conditions, respectively, after 1 mo of cultivation. Ps_egfp colonies were quantified also in the biofilm, resulting in 1.42 × 10 11 ± 1.36 × 10 10 and 3.48 × 10 11 ± 1.48 × 10 10 CFU Ps_egfp /L, respectively. In spite of the fact that nitrate-fed conditions produced more Ps_egfp cells in the biofilm after 1 month of cultivation, the ratio of resident to detached cells in N 2 -fixing conditions was almost twice as high than that in nitrate-fed conditions (Table S3). This supports the findings of the proteomics and microscopy studies that attachment of Ps_egfp is stronger in conditions involving N 2 -fixing by Tolypothrix sp. cells.

DISCUSSION
The main driver of this work was the observation, that in an artificial environment like a CBR, cyanobacteria develop much better biofilms in terms of biomass when co-cultivated with the chemo-heterotrophic organism Pseudomonas taiwanensis VLB120 (5, 10). The respiration activity of Ps_egfp leading to low oxygen tension in the system, was identified as a main reason for this phenomenon (5), albeit not the only one. Exchanging Ps_egfp with other aerobic heterotrophs like E. coli did not lead to enhanced biofilm formation (10). P. taiwanensis VLB 120 is a well described biofilm-forming organism applied in various biotechnological applications (11,25,26). The identification of possible communication mechanisms between two species was so far not successful (data not shown). Based on the previously published studies we hypothesized that P. taiwanensis forms a kind of seeding carpet on the first layer of the attachment surface and thereby facilitates cell attachment and biofilm growth of the cyanobacterial partner. The in-depth imaging and analysis of the initial biofilm formation phase presented here supported this hypothesis. The level directly above the attachment surface (0, 10 µm) was dominated by Ps_egfp cells, independent of whether Tolypothrix sp. was depending on nitrogen fixation or not. The Ps_egfp biovolume was almost 100 times larger directly on the attachment surface as compared to deeper levels (20, 30 µm). The results of this study clearly indicate that Ps_egfp promotes general surface attachment of consortia in the early growth phase, despite the fact that this organism is heavily limited in carbon and nitrogen compounds, respectively.
Under N 2 -fixing conditions, the overall biovolume development was less fluctuating indicating strong attachment especially of the cyanobacterial partner, although nitrate-fed biofilms had higher total biovolume values. These findings are in line with a previously published study, where it was shown, that also other cyanobacteria like Nostoc sp. had lower detachment rates under N 2 -fixing conditions (10). In a complementary CBR experiment cultivated for 1 month, it was observed that ratio of resident to detached Ps_egfp cells was elevated 2-fold in N 2 -fixing biofilms compared to nitrate-fed biofilms (Table S3). These findings are very interesting in terms of keeping the harvested energy in the cultivated system. Further studies should focus on modelling of these systems in order to understand the amount of energy being lost comparing N 2 -fixing and non-fixing conditions. It may be speculated that the specialized surface layers of the heterocysts, the specialized cells where the N 2 -fixing enzymes are located in many cyanobacterial species, are important for the enhanced attachment. As nitrogenase is highly O 2 -sensitive, heterocysts have special surface layers, which provides an O 2 -free inner compartment. According to some studies, these surface layers contain a variety of biopolymers that are responsible for protecting the structure (27,28). It may be that these biopolymers facilitate cell attachment not only of the cyanobacterium, but also in case of the Ps_egfp. Fitting to this hypothesis is the fact, that also Ps_egfp seems to attach much stronger to the surface in close proximity of Tolypothrix sp. cells.
Surprisingly, Ps_egfp outcompeted Tolypothrix in total biovolumes in the early stage biofilms in all conditions and in all levels analyzed, especially under N 2 -fixing conditions. At a first glance, this may be surprising, as Ps_egfp is limited not only in the carbon, but also in the nitrogen source. The synthesis of EPS seems to be enhanced when nitrogen is depleted in some cases like Anabaena sp. ATCC 33047 (29) and Cyanothece sp (30). It is likely that this is related to the rise in the C:N ratio, which facilitates the addition of carbon to polymers (31)(32)(33), which in turn might lead to an increase in Ps_egfp cells.
In the context of cell attachment, the hydrodynamic forces introduced into the CBR by the addition of the air segments are highly important (34). Following the start of segmented flow at day 4 of cultivation, biofilms became generally more stable in terms of showing fewer fluctuations and biovolume development was strongly enhanced under all cultivation conditions tested. The hydrodynamic force introduced into the system by the air segments might have resulted in a change in overall biofilm structure by washing out loosely attached cells (11), and an increase in biovolume of Tolypothrix sp. (Fig. 2 and 3). Under N 2 -fixing conditions, additional molecular nitrogen was introduced into the system via the air bubbles, promoting the growth of Tolypothrix sp. cells. Furthermore, the co-localized clusters of Ps_egfp cells surrounding Tolypothrix sp. filaments showed a strong attachment after segmented flow was initiated. These results are in agreement with other publications, showing that segmented flow increases the surface coverage and cell attachment in CBRs (5,10,11).
Proteomics revealed the abundance of a wide variety of proteins in Ps_egfp, which were only abundant under N 2 -fixing conditions. Of particular interest are the LysMdomain protein, a protein belonging to the OmpA family, a malate dehydrogenase, and some proteins related to PHA synthesis. The LysM-domain protein recognizes and binds several different peptidoglycan ligands that contain N-acetylglucosamine (35), identified in ,for example, Nostoc punctiforme and Anabaena cylindrica fixing nitrogen (36)(37)(38). A comparable trait in Tolypothrix sp. might explain the strong co-localization of Ps_egfp and Tolypothrix sp. in N 2 -fixing conditions. Furthermore, plants can identify their symbiotic microbes via specific LysM domains (35). It is possible that Ps_egfp and Tolypothrix sp. have a similar interaction that could explain why in N 2 -fixing biofilms superior attachment and co-localization as compared to nitrate-feeding biofilms was observed. More experimental evidence is required to support this claim.
Also interesting is a protein candidate from the OmpA family, which also was abundant only in N 2 -fixing biofilms. A recent study showed that upon deletion of the peptidoglycan-binding anchor (Pba) of the OmpA family protein P. aeruginosa formed low-density, unorganized biofilms which were very fragile and sensitive to shaking compared to the wild type (39). In addition, some studies have demonstrated that the OmpA family of proteins can act as a pore for the diffusion of small molecules (40)(41)(42). Ps_egfp might use this protein in order to increase the diffusion of necessary nutrients and also attach strongly to surfaces in N 2 -fixing biofilms.
Malate dehydrogenase, catalyzing the bilateral conversion of L-malate and oxaloacetate in the tricarboxylic acid cycle (TCA), is another enzyme of Ps_egfp that was detected to be abundant in N 2 -fixing biofilms. Its abundance under N 2 -fixing conditions is interesting because its synthesis is triggered by altering conditions like oxygenation and the type of carbon substrates available (43,44). Abundance of this protein in Ps_egfp might indicate that N 2 -fixing cyanobacteria release different carbon substrates which subsequently serve as feedstock for Ps_egfp in the biofilm compared to nitrate-assimilating Tolypothrix sp. cells.
Additionally, proteomics revealed that proteins related to PHA synthesis were abundant in only N 2 -fixing biofilms such as PhaF (PVLB_02145) and the PHA granuleassociated protein (PVLB_02140). In some bacteria, such as Pseudomonas, carbon is stored as PHA, which is composed of 6-14 carbon atoms (45). Excess carbon availability but also certain nutrient limitations may promote medium-chain PHA accumulation in Pseudomonas strains (46). There are numerous studies reporting on PHA production in Pseudomonas sp. is strongly dependent on the type and amount of organic acids and nitrogen starvation (47)(48)(49). Interestingly, Tolypothrix sp. can produce fatty acids like γ-linolenic acid, and palmitic acid (50). Combined with nitrogen starvation this may induce the respective enzymes related to PHA synthesis. However, as Ps_egfp is severely limited in our biofilm, no significant PHA synthesis is to be expected. Kelly (51) demonstrated that pyruvate dehydrogenase E1 and peptidoglycan associated lipoprotein interact with PhaF protein and localize to PHA granules in vivo. A similar protein profile was also observed in our study under N 2 -fixing conditions in biofilms composed of Tolypothrix sp. and Ps_egfp.

Conclusion
This study contributes to a better understanding of microbial interactions in artificial dual-species phototrophic biofilms containing the cyanobacterium Tolypothrix sp. PCC 7712 and P. taiwanensis VLB 120_egfp as supporter strain in a continuous cultivation system. It was possible to show, that Ps_egfp established a conditioning film on the attachment surface, despite being strongly C and N limited, facilitated the attachment of cyanobacteria partner. N 2 -fixing biofilms showed higher Ps_egfp to Tolypothrix sp. ratios during initial attachment period indicating stronger attachment of Ps_egfp in this condition. In this context, a CBR experiment showed three times higher ratio of residing to detached cells in N 2 -fixing biofilms compared to nitrate-fed biofilms in the later phase of biofilms. Additionally, co-localization of Ps_egfp and Tolypothrix sp. in close proximity prevented detachment of Ps_egfp, which was correlated with the identification of several proteins of higher abundance being relevant for cell attachment. In conclusion, showing the role of Ps_egfp s in the initial attachment process, as well as the effects of different nitrogen feeding strategies and hydrodynamic forces, may assist in optimizing biofilm photobioreactors and facilitate the application of cyanobacteria for productive processes.

ADDITIONAL FILES
The following material is available online.  2022)). They both have the same system except the cultivation chamber. Fig. S2 (mSystems.00302-23-s0002.tif). Additional CLSM images show the co-localization of species in different locations. (A) shows N 2 -fixing biofilms on day 4, (B) indicates NaNO 3 -fed biofilms. Images indicated with the yellow box were taken in a different experiment independent from the current study but they were also same conditions.  .00302-23-s0007.docx). T-test results from the comparison of differences between (a) changing depths and (b) NaNO 3 -fed and N 2 -fixing biofilms in a certain volume. Data Set S1 (mSystems.00302-23-s0008.xlsx). Complete proteomics data. Full list of Ps_egfp and Tolypothrix sp. proteins abundant N 2 -fixing and NaNO 3 -fed dual-species biofilms (FDR of peptide sequences was adjusted to < 1%) are shown in the Excel file. M1 and M2 represent biological replicates of each condition; 'w/o N' represents the N 2 -fixing, and 'w/ N' represents non-fixing biofilms. Medium preparation procedure (mSystems.00302-23-s0009.docx). Preparation procedure for the media used in the study.