Metagenome Survey of a Multispecies and Alga-Associated Biofilm Revealed Key Elements of Bacterial-Algal Interactions in Photobioreactors

Scenedesmus obliquus strain U169 and Chlorella vulgaris strain U126 were obtained from the collection of algae at the University of Hamburg, Hamburg, Germany (http://www.biologie.uni-hamburg.de/bzf/zeph/zephsvcke.htm). The defined medium for cultivation of algae was composed of 2 g liter⁻¹ Flory Basis Fertilizer 1 (Euflor, Germany) and supplemented with 3.22 g liter⁻¹ KNO₃. Flory Basis Fertilizer 1 is solely based on mineral compounds and was not supplemented with any vitamins. The pH was adjusted to 7.0. The algae were cultivated at 17°C in liquid medium at a natural light intensity in polyethyleneterephthalate flat-panel photobioreactors. The culture medium circulated at 1 liter min⁻¹ in the PBR system. The PBR was aerated with compressed air and flue gas obtained from a combined block heat and power station. Physical parameters and culture conditions were monitored continuously (WTW IQ Sensor Net; System 2020 XT, Germany). The light intensity was determined with a LI-190 sensor (Li-Cor, USA). For a detailed description of the reactor, see Fig. S1 in the supplemental material and reference 36.

Escherichia coli strains were grown at 37°C in lysogenic broth (LB) medium (1% tryptone, 0.5% yeast extract, 1% NaCl) (37) supplemented with appropriate antibiotics. Lactobacillus plantarum ATCC 8014 and Lactobacillus delbrueckii subsp. lactis DSM 20335 for B vitamin detection were obtained from the DSMZ, Braunschweig, Germany. Lactobacillus strains were grown on MRS medium (38) under anaerobic conditions.

Media for cultivation of individual bacterial isolates derived from the PBR biofilm community were prepared as follows. R2A medium was prepared as described previously (39), and M9, TSB, and NB media were prepared according to the method of Sambrook and Russell (37). To stimulate microbial growth, the media were in part supplemented with algal culture extracts ranging from 5% to 50% (vol/vol). The inoculated plates were incubated for 5 to 7 days at 22°C under aerobic and anaerobic conditions.

For scanning electron microscopy (SEM), biofilm 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 an SCD 050 sputter coater (Bal-Tec), scanning electron micrographs were taken with a Leo 1525 (Zeiss, Germany).

Total nucleic acids were extracted from the biofilm samples using a previously published enzymatic cell lysis protocol with some modifications (40). The samples were stirred (200 rpm) overnight in 10 ml of extraction buffer (100 mM Tris-HCl, pH 8.0, 100 mM sodium EDTA, pH 8.0, 1.5 M NaCl, 0.1% Tween 80) with the addition of 5 mg lysozyme and 0.5 mg proteinase K at 37°C. Subsequently, SDS (1 ml; 20%) was added and incubated at 65°C for 90 min. The sample was centrifuged for 10 min at room temperature and 6,000 × g. The pellet was resuspended in 10 ml extraction buffer, incubated for 10 min at 65°C, and centrifuged again. The supernatant was collected and mixed with 1 volume polyethylene glycol (PEG) 6000 (30%) and 1.6 M sodium chloride. After 2 h of incubation at room temperature, the mixture was centrifuged again for 20 min at 10,000 × g. The resulting pellet was resuspended in 2 ml TE buffer (10 mM Tris-HCI, 1 mM sodium EDTA, pH 8.0). For separating proteins and polysaccharides, 7.5 M potassium acetate was added to the mixture, achieving a final concentration of 0.75 M. After 5 min of incubation on ice, proteins and polysaccharides were centrifuged at 16,000 × g and 4°C for 30 min. The DNA was finally extracted with phenol-chloroform–isoamyl alcohol and precipitated overnight at −20°C after adding 0.7 volume isopropanol with 1/10 volume of 3 M sodium acetate. The isolated DNA was used for PCR amplifications, as well as metagenomic analyses.

For the phylogenetic characterization, 16S rRNA genes were amplified using the standard primers 27f and 1492r (41). The amplified genes were ligated into the pDrive cloning vector (Qiagen, Hilden, Germany) and transformed into chemically competent TOP10 E. coli cells (Invitrogen, Karlsruhe, Germany). The 16S rRNA gene was sequenced with automated ABI377 technology following the manufacturer's instructions.

The large-insert fosmid library was constructed according to the Copy Control fosmid library production kit manual (Epicentre Biotechnologies, Madison, WI, USA). A total of 14,976 fosmid clones harboring inserts with an average size of 35 kb were generated. The insert rate was approximately 98%.

For sequencing of metagenomic DNA on the GS FLX platform (Roche Applied Science), libraries were constructed by applying the GS Rapid Library Prep kit (Roche Applied Science) according to the manufacturer's protocol. A single library was sequenced in half a PicoTiter Plate on the GS FLX system using the titanium sequencing chemistry (Roche Applied Science, Mannheim, Germany). Raw data were processed by employing the analysis pipeline for whole-genome shotgun sequence reads and applying GS FLX System Software (version 2.3). Illumina sequencing was done using a single lane of a HiSeq 2000 paired-end run (2 sets of 100 cycles). De novo assembly of these short reads was performed using the Velvet assembly program version 1.2.08 (42).

For the sequences' functional characterization, we used the Integrated Microbial Genomes (IMG) pipeline. To further analyze the biological processes linked to the individual genes and open reading frames (ORFs), we mainly employed the KEGG (43), COG (44), and Pfam (45) databases, using a cutoff of 10⁻⁵.

16S rRNA genes were amplified using oligonucleotide primers (27f, 5′-AGA GTT GAT CMT GGC TCA G-3′, and 1492r, 5′-GGY TAC CTT GTT ACG ACT T-3′) (41). PCR mixtures contained 100 ng of template DNA per μl, 0.2 mM each of the four deoxynucleoside triphosphates, 1.5 mM MgCl₂, 1 μM (each) primer, and 2.5 U of Taq DNA polymerase. The thermocycling conditions were 45 s of denaturation at 94°C, 45 s of primer annealing at 50°C, and 1 min 30 s of primer extension at 72°C. This cycle was repeated 30 times.

Denaturing gradient gel electrophoresis (DGGE) analyses were carried out with a DCode system (Bio-Rad, Munich, Germany) using denaturing gradients of 40 to 70% denaturants. For the amplification of the 16S rRNA gene fragments, the primers 341F-GC (5′-GCA CGG GGGG CCT ACG GGA GGC AGC AG-3′, containing a 30-bp GC-rich sequence at the 5′ end) and 907R (5′-CCG TCA ATT CCT TTR AGT TT-3′) were used (46, 47). The thermocycling conditions were 40 s of denaturation at 96°C, 40 s of primer annealing at 56°C, and 1 min of primer extension at 72°C. This cycle was repeated 30 times. DGGE bands were sampled for reamplification and sequencing by punching into individual bands with sterile pipette tips and were transferred to 1.5-ml microcentrifuge tubes that contained 50 μl of distilled water. The reamplified PCR products were cleaned up using the Gel/PCR DNA Fragments Extraction kit (Qiagen, Hilden, Germany) and finally sequenced using standard technologies.

For restriction fragment length polymorphism (RFLP) analysis, the amplified (27f and 1492r) 16S rRNA genes were cloned into pDrive and transformed in E. coli cells. The insert DNA of 100 clones of each sample (T0 to T5) was amplified using oligonucleotide primers (M13-20-f, 5′-GTA AAA CGA CGG CCA GT-3′, and M13-r, 5′-CAG GAA ACA GCT ATG ACC-3′). The thermocycling conditions were 45 s of denaturation at 94°C, 45 s of primer annealing at 57°C, and 1 min 45 s of primer extension at 72°C. This cycle was repeated 30 times. The fragments were digested by the restriction enzyme HpaII. Clones that showed different patterns were sequenced using standard technologies.

Furthermore, to investigate the composition of the microbial community in the photobioreactor, rRNA sequences of the small subunit (SSU) were extracted from the unassembled FLX 454 and Illumina data using SortMeRna (48). Low-quality (<20 Ns) and short (<80-bp) reads were removed prior to further analysis. To gain insight into the community composition, sequences were directly mapped onto the most recent Silva database (SSURef 111 NR) using Bowtie2 (49). The Silva taxonomy of the best hit in the database was affiliated to the respective query sequence. For deeper analysis of the bacterial community structure, the hypervariable regions V3 and V6 of the bacterial 16S rRNA genes were extracted using V-Xtractor (50) and were also mapped on the most recent Silva database.

Exopolymeric substances (EPS) from the biofilm sample were extracted according to the method of Wingender et al. (51) with minor modifications. For this purpose, the biofilm samples were carefully removed from the PBR surface, weighed, and subsequently suspended in a ratio of 1:16 (wt/vol) in 0.14 M NaCl solution and homogenized by stirring at room temperature for 60 min. The suspension was centrifuged for 30 min at 20,000 × g and 10°C. The supernatants were twice sterile filtered using membrane filters (cellulose acetate; pore size, 0.20 μm). The filtrates were either processed immediately or aliquoted and stored until use at −20°C.

The lipid content was measured by extraction with n-hexane (52). To 0.5 ml of the EPS sample, 0.5 ml n-hexane was added and homogenized for 30 min by shaking at room temperature following a centrifugation step at 12,000 × g for 90 s. The upper phase (hexane and lipids) was transferred to a preweighed glass tube while avoiding the interphase. The solution was dried, and the glass vessel was weighed again.

The total carbohydrate content was determined using a phenol-sulfuric acid method (53) with dextran (neutral polysaccharides) and alginate (acidic polysaccharides) standards. To 0.5 ml of sample, 0.5 ml of phenol solution was added and mixed. Following this, 2.5 ml of concentrated sulfuric acid (H₂SO₄) was added, and the sample was mixed again following an incubation step at room temperature (10 min) and incubation at 30°C for 15 min. After a further 5-min incubation at room temperature, the absorbance of the samples was measured spectrophotometrically as the optical density at 480 nm (OD₄₈₀) for acidic polysaccharides (alginate) and the OD₄₉₀ for neutral polysaccharides (dextran).

Uronic acids were determined according to the method of Filisetti-Cozzi and Carpita (54) with glucuronic acid as a standard. To 0.4 ml of each sample, 40 μl solution 1 (4 M sulfamic acid [H₃SO₃N], pH 1.6) was added and mixed. After this step, 2.4 ml of solution 2 [0.075 M Na₂(B₄O₅(OH)₄) × 8H₂O dissolved in H₂SO₄] was added, mixed, and then incubated for 20 min at 100°C. The samples were cooled in an ice bath for 5 min. Then, 80 μl of solution 3 (0.15% [wt/vol] m-hydroxybiphenyl [C₁₂H₁₀O] dissolved in 0.5% [wt/vol] NaOH) was added. Finally, the samples were thoroughly mixed and incubated for 10 min at room temperature, and the absorbance was recorded as OD₅₂₅.

The protein content was measured using a modified Lowry assay (55, 56).

Furthermore, the total DNA within the EPS solution was detected spectrophotometrically after extraction from the EPS using the UltraClean Microbial DNA Kit (Dianova, Hamburg, Germany).

Lipase and esterase activities were assayed using various p-nitrophenyl (pNP) substrates and following previously published protocols (57).

For the determination of cobalamin the Difco B₁₂ assay medium was used, and vitamin B₁₂ was extracted and assayed according to the protocol published by Denter and Bisping (58) employing the L. delbrueckii subsp. lactis DSM 20335 indicator strain.

To determine the biotin content of the sample, the culture supernatant was centrifuged and sterile filtered. The biotin detection was done as described previously using the Lactobacillus growth assay and L. plantarum strain ATCC 8014 (59).

This project has been assigned the GenBank BioProject number PRJNA197241. The sequences derived from Illumina and FLX454 sequencing were deposited in the NCBI Short Read Archive under the study accession number SRP021903. The genome assembly, together with predicted gene models and annotations, is available from www.jgi.doe.gov (DOE Joint Genome Institute) under the IMG Project ID 9992.

The 16S rRNA gene sequences obtained have been deposited in the GenBank database under accession numbers KC994651 to KC994663, KC994664 to KC994680, KC994681 to KC994881, and KC994882 to KC994889.

PBRs are commonly inoculated with nonaxenic cultures of eukaryotic microalgae (14–16). In the current study, we examined a microbial biofilm in a PBR, which was inoculated with an S. obliquus and C. vulgaris culture. We monitored biofilm development over a period of 12 weeks by analyzing a total of six time points (T0 to T5) (Fig. 1). As early as 14 days after the initial transfer of the starter culture into the reactor, extensive biofilm formation was observed. The biofilm was tightly attached to the surface of the reactor and spread over the entire reactor within a period of 10 to 12 weeks, thereby strongly affecting the overall photosynthesis rates of the algae. This strong biofilm development resulted in shutdown of the PBR. Microscopic examinations of biofilm samples using SEM suggested that the bacteria were tightly attached to the algae (Fig. 2A to D). Additionally, EPS and nanowire-like filamentous structures were visible in the SEM images (Fig. 2). It is well known that EPS in biofilms mainly consists of polysaccharides, proteins, nucleic acids, and lipids (60). The concentrations of the individual components, however, vary and depend to a large extent on the different microorganisms and their medium and environmental conditions (61–63). EPS has diverse functions during biofilm development: it provides mechanical stability, builds up a three-dimensional network of polymers, and plays a role in cell-cell communication and defense against biocides (60, 64). A detailed chemical analysis of the PBR biofilm exopolymeric matrix suggested that it consisted mainly of fatty acids (10.4 mg/g), acidic and neutral polysaccharides (11.6 mg/g), and proteins (3.8 mg/g). However, only small amounts of uronic acid (<1%) and only traces of DNA were observed (Fig. 3). Since microalgae are well known to produce and release lipids (2, 5), it is not surprising that the EPS consisted largely of fatty acids.

For phylogenetic analysis, DNA from the mature biofilm (T5), the different biofilm developmental stages (T1 to T4), and the nonaxenic starter culture (T0) was extracted. The quality of the isolated DNA was sufficiently high for the construction of a large-insert metagenome library and DNA-sequencing analysis using pyrosequencing and Illumina-based technologies.

We used different strategies to estimate the microbial diversity within the developing biofilm samples and the established biofilm. We amplified 16S rRNA genes using standard primers (27f and 1492r [41]) and cloned the amplified DNA fragments. Inserts of these clones were completely sequenced. After removal of potential chimeras and duplicates, we analyzed a total of 201 unique 16S rRNA gene sequences from the mature biofilm (T5) (KC994681 to KC994881). In order to identify unique phylotypes and to estimate the bacterial richness, a rarefaction analysis using QIIME (65) was performed. This analysis suggested that the 16S rRNA gene sequences represented 28 operational taxonomic units (OTUs) based on a >99% identity cutoff for bacterial 16S rRNA genes. Rarefaction curves reached saturation at distance levels of 20% (phylum level) and 3% (species level) (Fig. 4). This analysis also suggested that the overall diversity within the biofilm community was rather limited. In addition, a Shannon-Weaver index (66) of 3.01 at a genetic distance of 1% was observed. These results, together with abundance-based coverage estimators (ACE) (67) and Chao1 richness estimates (68), indicated that a significant fraction of the bacterial species diversity was assessed. The analyzed bacterial 16S rRNA gene sequences were mainly affiliated with members of the classes Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria (Fig. 5; see Table S1 in the supplemental material). Additionally, a significant number of 16S rRNA gene hits were associated with the phylum Bacteroidetes. Bacteroidetes are among the main colonizers of the guts of humans and animals, but they are also found in many other habitats and are regarded as specialists in the degradation of high-molecular-weight organic matter (34). Overall, the most abundant bacterial organisms were members of the families Sphingomonadaceae, Caulobacteraceae, Rhizobiaceae, Comamonadaceae, Xanthomonadaceae, Sphingobacteriaceae, and Flavobacteriaceae.

Additional data from DGGE and RFLP analysis from the different time points during biofilm development and the nonaxenic starter culture supported these findings (see Fig. S2 and S3 in the supplemental material). The DGGE profiles and RFLP analysis, in combination with 16S rRNA gene sequences (DGGE, KC994651 to KC994663; RFLP, KC994664 to KC994680) derived from T0 to T5 further suggested that the phylogenetic structure of the biofilm population was rather stable over the entire time. The data also indicated that the microbes observed within the PBR biofilm were mostly already present in the starter culture used to inoculate the PBR at T0.

To further verify these data, we analyzed the DNA sequences of the mature PBR biofilm obtained by pyrosequencing and Illumina-based sequencing for the presence of hypervariable regions of rRNA gene fragments. Analysis of the V3 and V6 regions has been applied to study bacterial communities in diverse ecosystems (69–71). In total, 226 sequences from the FLX 454 (see Fig. S4A in the supplemental material) and 137,506 sequences from the Illumina (see Fig. S4C in the supplemental material) data sets were directly mapped onto the Silva database. For the analysis of the bacterial population, the hypervariable region V6 of the bacterial 16S rRNA gene was analyzed. Altogether, 41 sequences from the FLX 454 (see Fig. S4B in the supplemental material) and 558 sequences from the Illumina (see Fig. S4D in the supplemental material) data sets were included in the analysis.

In total, the results from the analysis (see Fig. S4 in the supplemental material) largely confirmed the results and the phylogenic composition estimated by using the 16S rRNA gene clone library.

However, it is notable that primer biases occurring during amplification of 16S rRNA genes (72) and the relatively high copy number of the microalgal plastid genomes, including 16S rRNA genes, resulted in slightly different species counts. The observed fungi, diatoms, and golden algae most likely are contaminants resulting from previous use of the PBRs.

Initial attempts to isolate bacteria associated with the algae by plating on different solid media (LB, tryptone-yeast [TY], Reasoner's 2A [R2A], TSB, and NB) under aerobic and anaerobic conditions, resulted in only two pure cultures. 16S rRNA gene sequencing suggested that the two isolates were affiliated with the genera Brevundimonas ( JQ661035.1; isolate A) and Paracoccus ( JQ404485.1; isolate B) (see Table S2 in the supplemental material). Members of these genera have been frequently isolated from soil and aquatic environments, but various clinical isolates are also known. Interestingly, tests in which we added various amounts of algal culture supernatant, including living cells, to the solid media did stimulate growth of another six bacterial organisms. These isolates were phylotyped by 16S rRNA gene sequencing (see Table S2 in the supplemental material). The 16S rRNA gene sequence of isolate C is most similar to that of Chryseobacterium taichungense YNB68 ( JQ071521.1). C. taichungense belongs to the Flavobacteriales and was only recently described as a novel species able to hydrolyze many chromogenic substrates (73). Isolate D could be associated with a Brevibacterium species that was not further characterized ( GQ199748.1). Brevibacteria are strictly aerobic Gram-positive microorganisms. Both isolates C and D were isolated on R2A agar plates supplemented with 50% (vol/vol) algal culture. The closest relative of isolate E is a member of the genus Roseomonas. The 16S rRNA gene sequence revealed a similarity of 78% to Roseomonas sp. ( HQ588850.1). Isolate E was cultivated in the presence of 25% (vol/vol) algal culture on NB agar plates. 16S rRNA analysis of isolate F indicated that this strain is associated with the yet-uncultured Cytophaga-Flavobacteria-Bacteroides (CFB) group bacterial clone S15D-MN13 ( AJ583211.1). The 16S rRNA gene sequence was 96% similar to the S15D-MN13 GenBank entry. Isolate G was similar to a hitherto uncultured gammaproteobacterium, clone L4 ( EU887990.1). 16S rRNA sequencing of isolate H suggested that the isolate is affiliated with Rhodococcus sp. strain 3/2 (EU041710.1). Isolates F to H were grown on NB medium supplemented with 50% (vol/vol) algal culture.

Altogether, these data suggest that only a minor fraction of the bacteria within the PBR are easily cultivable in laboratory media ( KC994882 to KC994889).

Today, our knowledge of the physiology and metabolism of complex microbial communities associated with eukaryotic microalgae is very limited. Therefore, we analyzed the partial metagenome sequences of the PBR biofilm microbial community with respect to possible gene functions. We established and assembled 164.7 Mbp of DNA sequences for the PBR biofilm community in contigs ranging from 165 bp to 203,650 bp (Table 1). Approximately 347,000 protein-coding genes could be assigned. Of all possible ORFs predicted, a total of 55,767 were similar to ORFs and genes in the KEGG database; 125,999 ORFs matched the COG database, and 35,935 ORFs could be analyzed by a MetaCyc pathway analysis (74). In addition, it should be noted that up to 1.5% of all ORFs were derived from eukaryotic DNA in the sample. Within this framework, it should be taken into account that most likely a small but significant number of ORFs are not correctly predicted due to overlapping genes/ORFs and noncorrected frame shifts. Based on the assumption that most of these 30 different bacterial species have genomes ranging from 3 to 8 Mbp, it is reasonable to speculate that the overall bacterial biofilm metagenome ranges from 90 Mbp to 240 Mbp. Therefore, the 164.7-Mbp assembled DNA represents a significant fraction of the overall bacterial biofilm metagenome. Although the available sequences do not allow a complete analysis of the physiological and metabolic functions within this bacterial community, the sequences give a good estimate of the biofilm genome structure and its metabolic potential.

The KEGG analysis suggested that the metabolic and catabolic potential of the microbes living in the PBR biofilm is highly diverse and flexible (Table 2). Genes for many of the classical pathways linked to the degradation of polysaccharides, proteins, and cellulose, but also for the degradation of aromatic compounds, were identified. With respect to the degradation of biopolymers, at least 64 amylases, 57 cellulases (mainly GH5 and M-family cellulose), 22 genes linked to the depolymerization of poly-β-hydroxybutyrate (PHB), and a high number of proteases (2,419) were observed.

Many genes involved in lipid and fatty acid catabolism could be identified. Altogether, at least 1,137 genes and ORFs encoded esterolytic and/or lipolytic enzymes, suggesting a high hydrolytic activity for fatty acids within the analyzed microbial community. This observation was supported by data from a metagenome fosmid library screening. Altogether, 14,976 clones with an average insert size of 35 kbp were screened in tributyrin-containing indicator plates. This equals about 524 Mbp of screened DNA. From this screen, a surprisingly large number of lipolytic clones were identified. A total of 68 clones were identified in a first screen using the outlined screening procedures. Further tests with cell extracts of 16 of these clones confirmed that a wide range of fatty acids with various chain lengths were turned over by extracts of the fosmid clones (Fig. 6). The majority of the clones were active on the pNP-laurate (C₁₂) but had almost negligible activities on the pNP-palmitate (C₁₆) and pNP-stearate (C₁₈) and thus most likely encoded esterases acting preferentially on shorter-chain fatty acids. However, very few clones also hydrolyzed the longer-chain fatty acid substrates and are perhaps more likely to encode true lipases (i.e., pFOS73A2, pFOS108D6, and pFOS131A4). Additional tests with culture supernatants supported this hypothesis. In these tests, we observed small but significant hydrolytic activities ranging from 0.0732 to 0.3373 milliunits per ml in the cell-free culture supernatant of the PBR. Lipolytic enzymes, including lipases (EC 3.1.1.3) and carboxylesterases (EC 3.1.1.1), are of general cell importance for membrane and fatty acid metabolism, and they are well known to be important during various pathogenic and nonpathogenic host-microbe interactions (75, 76). Thus, we speculate that the relatively high number of esterolytic and lipolytic genes is perhaps a result of the known production of lipids and fatty acids by the algae and thus has possibly resulted in an enrichment of highly lipolytic and esterolytic microbes. It is further notable that many short-chain fatty acid oxidoreductases and oxidoreductases without an assigned substrate (>600) were identified within the PBR metagenome. Some of these will most likely be important for the oxidation/reduction of the lipids derived from algae.

The metagenome sequences further suggest that the bacteria associated within the PBR were mainly heterotrophs metabolizing a wide range of carbon and energy sources provided by the algae.

The microbial lifestyle in the PBR biofilm most likely included aerobic, as well as microaerobic, growth. Altogether, 1,844 cytochrome-encoding genes were counted, suggesting that highly diverse and branched electron transport chains are encoded by the bacteria living within the PBR biofilm. Branched electron transport chains are an adaptation to varying oxygen concentrations (77) and are observed in the majority of all bacterial genomes. In addition, all the genes needed for nitrate respiration were identified. The occurrence of these genes will certainly allow a wide range of metabolic activities under varying oxygen concentrations and could be an adaptation to life in biofilms that are known to include gradients of oxygen and often even have anoxic pockets (78, 79).

Vitamin B₁₂ is one of the most complex cofactors and an essential vitamer that is required for molecular rearrangements of hydrogen or methyl groups in prokaryotic and eukaryotic cells, and it is involved in the reduction of ribonucleotide triphosphate to 2′-deoxyribonucleotide triphosphate (80, 81). Higher eukaryotes (i.e., animals, humans, and protists) require B₁₂ but do not synthesize the vitamer, and plants and fungi neither synthesize it nor use it. While plants have evolved a B₁₂-independent methionine synthase, algae require the vitamer for the function of B₁₂-dependent methionine synthase (22–24, 82, 83). Vitamin B₁₂ is synthesized by bacteria and archaea, which have evolved two distinct routes for biosynthesis under aerobic and anaerobic conditions (22). More than 20 different steps are required for the biosynthesis of the corrinoid ring structure from uroporphyrinogen III (80, 81).

The KEGG-based analysis of the predicted genes and ORFs in the PBR biofilm metagenome suggested that all the genes with relevance to B₁₂ vitamin biosynthesis were present, and many of them in multiple copies (see Fig. S5 in the supplemental material). Within this framework, it is notable that many algae are auxotrophic with respect to the production of thiamine and biotin, as well (21). Thiamine (vitamin B₁) is essential for some of the key enzymes in the primary carbohydrate and branched-chain amino acid metabolism. Altogether, we have identified at least 218 genes in the biofilm metagenome that are involved in thiamine biosynthesis (see Fig. S6 in the supplemental material) (84, 85). Biotin is a cofactor of biotin-dependent carboxylase reactions, among them acetyl coenzyme A (CoA) carboxylase, which is essential for fatty acid synthesis (86, 87). For all major genes involved in biotin biosynthesis, multiple copies were detected (see Fig. S7 in the supplemental material). Based on the KEGG analysis, almost 6.7% of all ORFs and genes encoded metabolism of cofactors and vitamin biosynthesis. Additional tests in an experimental test reactor confirmed that small amounts of the two vitamers B₁₂ and biotin could already be detected in the PBR culture supernatant 48 h after inoculation (Fig. 7 and 8). The pico- and nanomolar amounts observed were certainly sufficient to support algal growth.

Microbial biofilm formation depends to a large extent on the attachment of bacteria to surfaces, microcolony formation, and maturing of the biofilm structure (64, 88, 89). Within the PBR metagenome, many genes were identified that are essential to biofilm formation. Biofilm formation is regulated by a wide range of factors, including signaling molecules, such as autoinducer involved in quorum-sensing-dependent gene regulation. Interestingly, we have identified only a small number of genes that are involved in the synthesis of autoinducer I-type (N-acyl-homoserine lactone) molecules. N-Acyl-homoserine lactones are the key signaling molecules in the cell density-dependent system of gene regulation in many Gram-negative bacteria, and they are usually synthesized through a LuxI-like protein (90, 91). Employing the amino acid sequences of various LuxI homologues for Blastp searches, we could not detect more than a few homologues (<10) in the PBR metagenome. This finding is intriguing, and it might suggest that autoinducer I-dependent signaling may play only a minor role within the observed ecological niche.

Sequencing-based metagenome analyses in combination with function-based studies have in general significantly increased our knowledge of microbial communities and their phylogenetic makeup and genetic potential. Finally, these studies provided insight into the ecology of the analyzed communities and thereby enriched our understanding of biodiversity and biochemical functions. Many of these studies have been summarized previously (29, 92–94). Surprisingly, only a few studies have addressed metagenome technologies for the analysis of mixed-species biofilms in nonhuman interactions. The earliest studies dealt with the acid mine drainage biofilm (95) and a drinking water biofilm (96). More recently, studies have addressed microbial biofilm metagenomes of hydrothermal vent chimneys, biofilms on concrete water pipes, and an extremely acidophilic sulfur-oxidizing biofilm (97–99). Although not all studies have been listed here, it is evident that many different biotic and abiotic factors shape each of these communities and that each community clearly differs in its phylogenetic structure and genomic-information content.

In summary, our data give detailed insight into the microbial community and the metagenome of a PBR-associated microbial biofilm. Altogether, we observed 28 OTUs within the analyzed biofilm, and DGGE and RFLP analyses suggested that the community was rather stable over time. The sequence-based data and the laboratory measurement data clearly support earlier findings with respect to the bacterially produced B vitamins. Our data also suggest that bacterial growth depends in part on compounds released by the algae or bacteria within the community.

Finally, it is interesting that the bacteria in the PBR studied here encode many esterolytic and lipolytic enzymes. This finding was supported by extracting active lipolytic clones and by measuring the hydrolytic activities in the culture supernatants. Future work will have to use these data to further improve the efficiencies of PBRs with respect to CO₂ fixation and downstream processing of biofuels and/or other valuable compounds produced and released by the algae.