For the isolation of novel unicellular alga, we used a low energy laser trap microscope (PALM MicroTweezers, Zeiss, Germany). This microscope is equipped with optical tweezers and a vacuum suction device. We have been using this to isolated individual microalga from environmental freshwater samples of a pond of the “Botanical Garden” (Hamburg, Germany, latitude: 53.5667, longitude: 9.9833).

Microalga isolates were either obtained from the MZCH¹ and newly isolated. The following isolates were cocultured for a minimum of 10 years in the strain collection: C. saccharophila (MZCH 10155), Chlorella vulgaris pv minima HH (MZCH 10162), Chlorella zofingiensis (MZCH 10156), Scenedesmus acuminatus (MZCH 10102), Scenedesmus carinatus (MZCH 10103), S. quadricauda (MZCH 10104), Micrasterias furcata (MZCH 75), M. crux-melitensis (MZCH 98), and Micrasterias muricata (MZCH 125). Recently newly isolated strains are: Chlamydomonas sp. HH 10201, Chlamydomonas sp. HH 10204, Acutodesmus sp. HH 10202, Desmodesmus sp. HH 10203, Scenedesmus sp. HH 10205, and Scenedesmus sp. HH 10206.

Cultivation of the individual alga species was done as follows: for the cultivation of C. saccharophila (MZCH 10155), we used an Ld+OHC-media with minor modifications (Aaronson and Baker, 1959). The medium was composed of 0.05 g/l Ca(NO₃)₂ × 4H₂O, 0.05 g/l MgSO₄ × 7H₂O, 0.003 g/l K₂HPO₄, 0.25 ml micronutrient solution [stock solution: 0.5 g/l FeSO₄ × 7H₂O, 1.0 g/l H₃BO₃, 0.5 g/l MnCl₂ × 4H₂O, and 5.0 g/l titriplex III (EDTA)], 1 ml/l vitamin solution (stock solution: 0.1 g/l thiamin × HCl, 0.01 g/l biotin, 0.002 g/l cyanocobalamin, and 0.001 g/l niacin), and 10 ml/l OHC solution (stock solution: 100 g/l D-glucose, 100 g/l bacto-tryptone, 100g/l liver extract, and 100g/l yeast extract). The cultivation of C. vulgaris pv minima HH (MZCH 10162), C. zofingiensis (MZCH 10156), Scenedesmus sp. HH 10205, and Scenedesmus sp. HH 10206, was carried out in Kessler and Czygan-media with minor modifications (Kessler and Czygan, 1970). The media contained 0.81 g/l KNO₃, 0.0015 g/l CaCl₂ × 2H₂O, 0.25 g/l MgSO₄ × 7H₂O, 0.47 g/l NaCl, 0.47 g/l NaH₂PO₄, 0.36 g/l Na₂HPO₄ × 2H₂O, and 1.0 ml micronutrient solution [stock solution: 0.6 g/l FeSO₄ × 7H₂O, 0.05 g/l H₃BO₃, 0.05 g/l MnCl₂ × 4H₂O, 0.8 g/l titriplex III (EDTA), 0.02 g/l ZnSO₄ × 7H₂O, and 0.02 g/l (NH₄)₆Mo₇O₂₄ × 4H₂O]. S. quadricauda (MZCH 10104), Chlamydomonas sp. HH 10201, and Acutodesmus sp. HH 10202 were cultivated in BG11 (Stanier et al., 1971; Rippka et al., 1979). This media consists of 0.038 g/l CaCl₂ × 2H₂O, 0.075 g/l MgSO₄ × 7H₂O, 0.041 g/l K₂HPO₄ × 3H₂O, 1.50 g/l NaNO₃, 0.04 g/l Na₂CO₃, 0.006 g/l citric acid, 0.006 g/l ferric ammonium citrate, 0.001 g/l triplex III (EDTA), 1.0 ml micronutrient solution [stock solution: 0.079 g/l CuSO₄ × 5H₂O, 0.049 g/l CO(NO₃)₂ × 6H₂O, 2.86 g/l H₃BO₃, 1.81 g/l MnCl₂ × 4H₂O, 0.39 g/l Na₂MoO₄ × 2H₂O, and 0.222 g/l ZnSO₄ × 7H₂O], and 4.78 g/l HEPES buffer. For thei cultivation of S. acuminatus (MZCH 10102), S. carinatus (MZCH 10103), M. crux-melitensis (MZCH 98), M. furcata (MZCH 75), M. muricata (MZCH 125), Chlamydomonas sp. HH 10204, and Desmodesmus sp. HH 10203 we used a Woods Hole MBL-media (Stein, 1979). This media contains 0.037 g/l CaCl₂ × 2H₂O, 0.037 g/l MgSO₄ × 7H₂O, 0.009 g/l K₂HPO₄, 0.013 g/l NaHCO₃, 0.085 g/l NaNO₃, 0.028 g/l Na₂SIO₃ × 9H₂O, 1.0 ml micronutrient solution (stock solution: 0.01 g/l COCl₂ × 6H₂O, 0.01 g/l CuSO₄ × 5H₂O, 3.15 g/l FeCl₃ × 6H₂O, 0.18 g/l MnCl₂ × 4H₂O, 4.36 g/l Na₂EDTA, 0.006 g/l Na₂MoO₄ × 2H₂O, 0.02 g/l ZnSO₄ × 7H₂O), 1 ml/l vitamin solution (stock solution: 0.0001 g/l thiamin × HCl, 0.0005 g/l biotin, 0.0005 g/l cyanocobalamin, and 0.001 g/l niacin), and 0.5 g/l HEPES buffer. The pH value was adjust to 7.0–7.2.

The microalgae were cultivated at 22°C in liquid medium with a final concentration of 4% CO₂ at a natural light intensity for 14 h per day and a dark period of 10 h per day.

Scanning electron microscopy (SEM) was performed as previously published (Krohn-Molt et al., 2013). Therefore, samples were fixed in paraformaldehyde (1%) and glutaraldehyde (0.25%), dehydrated by ascending alcohol series and dried at the critical point with Balzers CPD 030 Critical Point Dryer (BAL-TEC, Schalksmühle, Germany). After coating samples with gold using a sputter coater SCD 050 (BAL-TEC), scanning electron micrographs were taken with a LEO 1525 (Zeiss, Germany).

Samples were taken from three different microalga: C. saccharophila (MZCH 10155), S. quadricauda (MZCH 10104) and M. crux-melitensis (MZCH 98). At the time of sampling, the microalgae were under steady growth conditions. The used samples were collected at the same photoperiod of the hosts, after 7 h of a daylight period under stable growth condition.

For metagenome sequencing and PCR amplification, total nucleic acids were extracted from the samples using a previously published enzymatic cell lysis protocol (Krohn-Molt et al., 2013), including a sonification (30 s, amplitude 60%, cycle 0.5) and centrifugation step (5 min, 700 × g) at the beginning, to separate the bacteria from the microalga. Concentration and purity of DNA was analyzed using a Nanodrop ND-2000 instrument (PEQLAB Biotechnologie GmbH, Erlangen, Germany).

For the phylogenetic characterization of the alga-associated microbial community, genes were amplified using an amplicon barcoded sequencing protocol for MiSeq platforms. Variable regions V3 and V4 of bacteria/archaeal genes were amplified using the primer set 515F/806R (515F: 5′-AAT GAT ACG GCG ACC ACC GAG ATC TAC ACT ATG GTA ATT GTG TGC CAG CMG CCG CGG TAA-3′ and 806R: 5′-CAA GCA GAA GAC GGC ATA CGA GAT (barcode) AGT CAG TCA GCC GGA CTA CHV GGG TWT CTA AT-3′ each reverse primer contains different barcode sequences) as previously published (Caporaso et al., 2012; Klindworth et al., 2012).

The PCR contained 100 ng of template DNA/μl, 0.2 mM of each of the four deoxynucleoside triphosphates, 1.5 mM MgCl₂, 1 μM (each) primer, and 2.5 U of Taq DNA polymerase. PCR cycling conditions were as follows: initial denaturation at 94°C for 3 min, followed by 34 cycles of denaturation at 94°C for 45 s, annealing at 50°C for 60 s, and extension at 72°C for 90 s. The final extension was conducted at 72°C for 5 min. Negative controls were performed with H₂O instead of template DNA. The obtained PCR products were purified via Gel/PCR DNA Fragments Extraction Kit (Geneaid Biotech, Taiwan) as recommended by the manufacturer. Three separate PCRs were conducted for each sample. The amplified genes were sequenced and analyzed on an Illumina MiSeq following the manufacturer’s instructions. Sequencing was performed in 2 × 151 bp paired-end mode using the MiSeq Reagent Kit v3. For the analysis, Illumina’s “16S Metagenomics” workflow (Analysis software version 2.6.2.3) was executed on-instrument using default settings. The workflow’s classification step is based on Illumina’s proprietary algorithm “ClassifyReads” and a modified version of the Greengenes Database (v.13_5)².

For phylogenetic characterization of the strain collection species: C. saccharophila (MZCH 10155), C. vulgaris pv minima HH (MZCH 10162), C. zofingiensis (MZCH 10156), S. acuminatus (MZCH 10102), S. carinatus (MZCH 10103), S. quadricauda (MZCH 10104), M. furcata (MZCH 75), M. crux-melitensis (MZCH 98), M. muricata (MZCH 125), and the new isolates: Chlamydomonas sp. HH 10201, Chlamydomonas sp. HH 10204, Acutodesmus sp. HH 10202, Desmodesmus sp. HH 10203, Scenedesmus sp. HH 10205, and Scenedesmus sp. HH 10206, the 18S/ITS/26S regions were amplified using oligonucleotide primers with an additional sequencing adapter binding site (M13for-18S/AU500af: 5′-TTG TAA AAC GAC GGC CAG TGG CGC TAC ACT GAT GTA TTC AA-3′ and M13rev-28B: 5′-GGA AAC AGC TAT GAC CAT GAG GTC CGT GTT TCA AGA CGG G-3′ (Friedl, 1996; Kovarik et al., 2005; Lindstrom and Hanic, 2005; Helms et al., 2007). PCR mixtures contained 100 ng of template DNA/μl, 0.2 mM of each of the four deoxynucleoside triphosphates, 1.5 mM MgCl₂, 1 μM (each) primer, and 2.5 U of Taq DNA polymerase. Thermocycling conditions included 45 s of denaturation at 94°C, 45 s of primer annealing at 58°C, and 1 min 30 s of primer extension at 72°C. This cycle was repeated 34 times. The fragment was sequenced with automated ABI377 technology following the manufacturer’s instructions.

Sequencing of metagenomic DNA was performed on the HiSeq 2500 platform using Rapid SBS chemistry (Illumina, San Diego). Libraries were constructed applying the NEBNext^® Ultra^TM DNA Library Prep Kit for Illumina (New England Biolabs) according to the manufacturer’s protocol. Each sample was sequenced on a single lane of a HiSeq 2500 rapid paired-end run (2 × 250 cycles). Sequencing adapters from the 3′-end of sequencing reads and low quality sequences (Phred quality score below 54) were removed using Trimmomatic (Bolger et al., 2014). Trimmed reads with a length of at least 35 bp were assembled using IDBA-UD (Peng et al., 2012). The resulting contigs were binned using MaxBin (Wu et al., 2014).

Taxonomic profiling of metagenome contigs and bins was performed using AMPHORA2 using a universal marker set of 31 protein encoding phylogenetic marker genes (dnaG, frr, infC, nusA, pgk, pyrG, rplA, rplB, rplC, rplD, rplE, rplF, rplK, rplL, rplM, rplN, rplP, rplS, rplT, rpmA, rpoB, rpsB, rpsC, rpsE, rpsI, rpsJ, rpsK, rpsM, rpsS, smpB, and tsf; Wu and Scott, 2012). NCBI taxonomy IDs were mapped to phylogenetic lineages given by AMPHORA2.

For the sequences’ functional characterization, we used the IMG/MER pipeline and tools. To further analyze the biological processes linked to the individual genes and open reading frames the KEGG (Nakaya et al., 2013), the COG (Tatusov et al., 2001), and the Pfam (Finn et al., 2008) databases were employed.

Triplicates of the stationary growth phase, and after 7 h of a daylight period, of the microalgae–bacteria cultures were sonicated (30 s, amplitude 60%, cycle 0.5) and centrifuged at 700 × g for 3 min, to separate the bacteria from the microalga. In a second step, the supernatant (contains the bacteria) were again centrifuged at 11,000 × g for 5 min. To extract the total RNA, we used a hot phenol method with minor modifications (Aiba et al., 1981). Therefore, the bacterial culture was mixed with 25 ml ice-cold killing buffer (20 mM Tris–HCl, 5 mM MgCl₂, pH 7.5), centrifuged at 4,000 × g for 10 min at 4°C and shock frozen in liquid nitrogen. Cells were resuspended in 125 μl ice-cold 300 mM sucrose/10 mM sodium acetate solution (pH 5.2) and mixed with 125 μl of 2% (w/v) SDS/10 mM sodium acetate (pH 5.2). The cells were incubated for 90 s at 65°C. The next step, the supplementation of 400 μl hot phenol was followed by an incubation time of 3 min at 65°C with a brief mixture of the suspension every minute. The suspension was shock frozen in liquid nitrogen and spun down at 11,000 × g for 10 min at RT, the supernatant was mixed again with 400 μl hot phenol and the steps were repeated. Afterward, 400 μl of phenol/chloroform/isoamyl alcohol (25:24:1 v/v) was added to the supernatant and spun down. This step was repeated, followed by the addition of 400 μl chloroform/isoamyl alcohol (24:1 v/v). After a further centrifugation step at 11,000 × g for 2 min, 40 μl 3 M sodium acetate (pH 5.2) and 1 ml 100% (v/v) EtOH were added to the supernatant. The RNA precipitated overnight at -20°C. Afterward, the RNA was spun down for 20 min at 4°C, washed twice with 70% (v/v) EtOH, dried and dissolved in 200 μl DEPC-treated H₂O. The removal of simultaneously extracted DNA was achieved by using the RTS DNase^TM Kit (MO BIO Laboratories Inc., Carlsbad, United States). The RNA was cleaned and concentrated via the RNA Clean & Concentrator^TM-5 Kit (Zymo Research, Irvine, United States). Concentration and quality of the total RNA was measured by a Nanodrop ND-2000 instrument (PEQLAB Biotechnologie GmbH, Erlangen, Germany) and verified on a 1.2% formaldehyde-agarose gel. The remaining transcripts were used for cDNA library construction by Vertis Biotechnologie AG, Germany³. Libraries for Illumina sequencing of cDNA were constructed by Vertis Biotechnology AG, Germany (see text footnote 3), as described previously for eukaryotic microRNAs but omitting the RNA size-fractionation step prior to cDNA synthesis (Berezikov et al., 2006). Equal amounts of RNA samples were poly(A)-tailed using poly(A) polymerase. Then, the 5′-triphosphates were removed by applying tobacco acid pyrophosphatase resulting in 5′-monophosphate. Afterward, a RNA adapter was ligated to the 5′-phosphate of the RNA. First-strand cDNA was synthesized by an oligo(dT)-adapter primer and the M-MLV reverse transcriptase. In a PCR-based amplification step using a high fidelity DNA polymerase, the cDNA concentration was increased to 20–30 ng/μl. A library-specific barcode for multiplex sequencing was part of a 3′-sequencing adapter. The following adapter sequences flank the cDNA inserts: TrueSeq Sense primer 5′-AAT GAT ACG GCG ACC ACC GAG ATC TAC ACT CTT TCC CTA CAC GAC GCT CTT CCG ATC T-3′ and TrueSeq Antisense NNNNNN primer (NNNNNN = 6n barcode for multiplexing) 5′-CAA GCA GAA GAC GGC ATA CGA GAT-NNNNNN-GTG ACT GGA GTT CAG ACG TGT GCT CTT CCG ATC (dT25)-3′. The libraries were sequenced on an Illumina HiSeq 2500 in single-end mode with 100 cycles.

Reads in FASTQ format were trimmed with a cut-off Phred score of 20 by the program fastq_quality_trimmer from FASTX toolkit⁴. The following steps were performed with the tool READemption (Förstner et al., 2014). The poly(A)-tail sequences (introduced in the library preparation) were removed and a size filtering step was performed that eliminated reads shorter than 12 nt. The remaining reads were mapped to the reference metagenome sequences, i.e., the assemblies of the bacterial communities using segemehl (Hoffmann et al., 2009).

The quantification of read per annotated feature as well as the calculation of RPKM (Reads Per Kilobase per Million) value was conducted with READemption’s subcommand “gene_quanti.”

The relative occurrence frequency of each COG in expressed genes in comparison to all genes found in the metagenomes was evaluated by performing a Fisher’s exact test. COGs with a p-value (adjusted by the Benjamini–Hochberg procedure) less than 0.05 were considered as enriched. The source code of the scripts for the computational analysis of the RNA-Seq data and the COG enrichment analysis are deposited at Zenodo⁵.

The samples of the S. quadricauda (MZCH 10104) culture were collected in the stationary growth phase of cultivation, after 7 h of a daylight period. Triplicates of the supernatant were concentrated (1:10) via a Vacuum Concentrator (Eppendorf, Hamburg, Germany). For purification, we used the Thermo Scientific Pierce C18 Spin Columns following the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, United States). The exopolymeric substances secreted by the microorganisms were analyzed concerning their content of expressed proteins. The lyophilized samples were sent to the “Systematic Proteomics and Bioanalytics” center in Kiel, Germany⁶. Each sample was re-dissolved in 20 μl 50 mM ammonium bicarbonate buffer, pH 7.4 (ABC). 15 μl of each sample were mixed with 15 μl 4 M guanidine hydrochloride in ABC. Disulfides were reduced in 10 mM 1,4-D/L-dithiothreitol (60 min at 56°C). Alkylation was performed by addition of iodoacetamide to a final concentration of 55 mM (incubation in the dark at room temperature for 60 min). For enzymatic digestion, the solutions were incubated overnight with trypsin, purified by SPE (C₁₈ ZipTips), dried by applying vacuum in a SpeedVac (Eppendorf, Concentrator plus) and resuspended in 20 μl 3% acetonitrile/0.05% formic acid for triplicate LC-MS/MS analysis. Peptide separation was performed on a Dionex Ultimate 3000 UHPLC system equipped with an Acclaim^® PepMap 100 analytical column (2 μm, 100 Å, 75 μm × 500 mm) coupled online to a Q Exactive Plus. Chromatographic separation was performed with 0.05% formic acid as buffer A and 80% ACN/0.04% FA as buffer B with following gradient: 0 min, 5% B; 5 min, 5% B; 95 min, 70% B; 100 min, 95% B; 110 min, 95% B; 110.1 min, 5% B; 120 min, 5% B. The flow-rate was set to 300 nl/min and the column oven temperature to 45°C. Mass Spectrometry acquisition was performed in positive mode on a Thermo Q Exactive Plus Orbitrap. For MS/MS, the 10 most intensive ions were fragmented by HCD (normalized collision energy: 25, full scan MS: 300–2,000 m/z, resolution 70,000). Data analysis, using the Sequest search algorithm, was performed on Proteome Discoverer against a FASTA databases containing the metagenome data of S. quadricauda (MZCH 10104) and common impurities.

During this study, generated raw sequence data have been deposited to the European Nucleotide Archive under the BioProject number PRJEB15376⁷.

Assemblies of the three microalgae metagenomes are available via IMG/MER⁸ using the IMG ID 3300008885 for the metagenome analysis of bacterial community of C. saccharophila (MZCH 10155), the IMG ID 3300005759 for the analysis of the bacterial community of S. quadricauda (MZCH 10104), and the IMG ID 3300008886 for the analysis of the bacterial community of M. crux-melitensis (MZCH 98).

The obtained 18S/ITS/26S rRNA gene sequences have been deposited in the GenBank database under accession numbers: KX525234–KX525236, MF326553–MF326558.

Previous studies have shown that microalga obtained from strain collections are associated with various bacterial communities with a rather low diversity, and that these bacteria can form tight phycosphere biofilms (Krohn-Molt et al., 2013; Mendes et al., 2013; Knief, 2014; Knack et al., 2015).

In our study, SEM confirmed a tight adherence of relatively few bacteria to the different alga surfaces. In addition, the images also indicated that the cells produced in part net like structures (Figures 1A–C). To further analyze these fascinating microbial communities for certain alga species 16S rRNA gene amplicon, and deep metagenome sequencing was initiated.

For the 16S rRNA amplicon sequences analysis, we assigned a minimum of 550,000 reads for each sample. The clustering of bacterial 16S rRNA gene amplicons was done at a 97% similarity level. In general, the analyses resulted in the identification of 7–22 operational taxonomic units (OTUs) per strain and 261 OUTs for the pond (located in the “Botanical Garden” of the University of Hamburg) from which the new strains were isolated. Interestingly, the results of the recently isolated microalga from the pond showed no specificity (Supplementary Figure S1A). In our view, this finding supported the concept that the long-term cultivation has resulted in the selection of distinct bacterial communities. Therefore, we have focused our experiments on the analysis of selected Chlorella, Scenedesmus, and Micrasterias strains from the “MZCH” (Supplementary Figure S1B). The three analyzed Micrasterias species revealed the lowest diversity with only 7–9 OTUs, while the Chlorella and Scenedesmus microalgae species revealed in general higher numbers of OTUs ranging from 11 to 22. Within this framework, the taxonomic classification of the bacterial 16S rRNA amplicons indicated that the obtained sequences were mainly affiliated with the phyla of the α-, β-Proteobacteria, and the Bacteroidetes. The main bacterial families observed were affiliated with the Comamonadaceae, Caulobacteraceae, Chitinophagaceae, Flexibacteraceae, and Sphingomonadaceae, and included also unclassified microorganisms (Supplementary Figure S1B and Figure 1D). While in the Scenedesmus phycospheres the Comamonadaceae appeared to be the dominant family present in all samples, no clear pattern was visible for the Chlorella phycospheres. In contrast, in the analyzed microbiomes of the Micrasterias isolates, the Sphingomonadaceae (α-Proteobacteria), appeared to be the dominant family. No archaeal OTUs were observed in any of the samples.

In general, it is well known that 16S rRNA amplicon sequencing can result in a biased data analysis (Klindworth et al., 2012; Schirmer et al., 2015; Fischer et al., 2016). Thus, to verify the above made findings, we were interested in metagenome-based analyses. Therefore, in addition to amplicon sequencing, a large dataset of metagenomic DNA sequences was produced for the microbiomes of C. saccharophila, S. quadricauda, and M. crux-melitensis. For each of the three analyzed microbiomes, we generated a minimum of 567 mio reads, which were assembled to 174 Mb of DNA for C. saccharophila, 162 Mb of contiguous DNA for S. quadricauda, and 268 Mb for M. crux-melitensis (Supplementary Table S1).

To further analyze the dataset, we employed the AMPHORA2 software, which uses 31 conserved bacterial proteins as phylo genetic markers (Wu and Scott, 2012). A total of 2,800 marker genes were identified and classified in our metagenome datasets. The analysis confirmed largely the 16S rDNA amplicon results with minor differences on the family level. For the C. saccharophila metagenome, we identified mainly Sphingomonadaceae, Caulobacteraceae, Comamonadaceae, and unclassified bacteria. The S. quadricauda microbiome was also affiliated with members of the Sphingomonadaceae, Bradyrhizobiaceae, Phyllobacteriaceae, Caulobacteraceae, Comamonadaceae, and unclassified bacteria. The main colonizers of the M. crux-melitensis microbiome appeared to be Bradyrhizobiaceae, Sphingomonadaceae, Phyllobacteriaceae, Chitinophaga, and bacteria affiliated with the phylum of the Bacteroidetes (Figure 1D).

For the functional analyses, our metagenome data were analyzed using Pfam, COG, and KEGG databases. A total of 236,273 possible proteins were identified for the bacterial community of C. saccharophila, 270,387 for the microbial metagenome of S. quadricauda, and 349,091 for the microbiome of M. crux-melitensis.

In the light of the aforementioned findings, we asked which genes with relevance to plant–microbe interactions were highly transcribed in the alga microbiomes. In addition, we wanted to know, whether the highly transcribed genes were affiliated mainly with the α-Proteobacteria, the Bacteroidetes, or other phyla. Therefore, we established a set of RNA-Seq data for each of the three microbiomes. Nine individual samples were analyzed by RNA-Seq representing three independent biological samples for each of the three microbiomes (Supplementary Table S3). A minimum of 24 mio cDNA reads could be uniquely mapped for the C. saccharophila transcriptome dataset. More than 18 mio reads were uniquely aligned for the S. quadricauda cDNA and over 10 mio could be aligned for the M. crux-melitensis transcriptome. Genes with a RPKM value of 10.0 or higher where considered as expressed. In order to calculate the gene expression level between different communities, the level of transcribed COGs were used to group genes (Figure 2). For statistical analyses, the Fisher’s exact test was applied in order to detect COGs enriched in the transcriptome datasets compared to the corresponding microbiomes. Overall, microbial genes encoding ribosomal proteins, and metabolic pathways (e.g., carbohydrate-, lipid-, amino acid metabolism, and general function including energy production pathways) were among the most strongly enriched ones. In addition, we detected a high number of enriched COGs, which encoded for metabolism of cofactors, vitamins and secondary metabolites, signal transduction mechanisms, and transport mechanisms and secretion systems. The bacterial ribosomal proteins are, of course, distinct from the ribosomal rRNAs depleted from metatranscriptomic assays.

We observed that a shared set of 867 common genes were expressed in all three microbiomes (Figure 2). Overall, the strongest transcribed genes in all three microbiomes were genes with the following functions: transcription and translation including posttranslational modifications (24–35% of common genes), energy production and conversion, transport mechanisms and secretion system, cofactor metabolism, and unknown function (each 5–9% of common genes). However, in each alga microbiome a different set of specific genes was expressed. In the C. saccharophila-associated microbiome, 214 specific genes were expressed, in the M. crux-melitensis, 192 genes, and in the S. quadricauda microbiome, 362 genes (Figure 2 and Supplementary Table S4).

The main difference between the three microbiomes transcripts were observed with respect to genes affiliated with transport and secretion processes, vitamin biosynthesis, signaling pathways and flagella biosynthesis. In the C. saccharophila transcriptome, we observed that specifically genes affiliated with the type II, IV, and VI protein secretion systems were highly transcribed. For the S. quadricauda transcriptome, we also identified components of the sugar-specific permease components of ABC-type transport systems (ribose, xylose, arabinose, and galactoside). Further many bacterial genes coding for proteins involved in carbohydrate and lipid modification were strongly transcribed.

Finally, in the M. crux-melitensis transcriptome, we identified mainly genes coding for ABC-transporters, long-chain fatty acid transporters as well as biopolymer transporter genes were highly transcribed. Interestingly, the number of genes involved in translation and posttranslational modifications were overrepresented in the transcription profile compared to the C. saccharophila and S. quadricauda microbiome (Figure 3).

Secretome analyses for the S. quadricauda microbiome were performed to partially verify the metagenome and metatranscriptome-based results (Supplementary Figure S2 and Table S8).

In total, we identified 156 bacterial proteins in the supernatants. The majority of the proteins identified originated from bacteria affiliated with the α- and β-Proteobacteria. Of these, a total of 55 proteins originated most likely from Porphyrobacter, eight from Blastomonas, and six from uncultured Sphingomonadales. Furthermore, 43 proteins originated from a non-cultivated bacterium and 42 from a Mesorhizobium species present within the microbiome. These data concur mostly with the phylogenetic analyses, the metagenome and with the RNA-Seq analyses. However, they also indicate an important role of non-cultivated bacteria and a mesorhizobial isolate in this alga microbiome, which was not detected having significant transcriptional activities.

The functional evaluation of the protein data shows that mainly proteins were found in the supernatant that are linked to protein biosynthesis, transport systems, and hypothetical proteins. The largest fraction (41 hits) was linked to ribosomal proteins. Thirty-five hits contained proteins with no function assigned to yet. Furthermore, 28 proteins were identified as periplasmic components of ABC transporters indicating that transport of small molecules is an important key feature in this alga microbiome. Additionally, 13 outer membrane and peptidoglycan-associated proteins, seven superoxide dismutases, seven proteins involved amino acid biosynthesis/modifications, five flagellins, and few hits for cobalamin and TonB receptor proteins were identified.