香京julia种子在线播放

Three A. aerophoba individuals were sampled in June 2014 by SCUBA diving in Cala Montgó, Spain (42.1140N, 3.167E) between 7.8 and 12.7 m depth. The individual sponges all possessed several oscula and grew between 3 and 15 m apart from each other on sun-exposed rock surfaces. Approximately 5 cm³ of each individual were cut with a diving knife and transferred to 50 mL centrifuge tubes under water. Samples were kept on ice and transported to the laboratory within few hours. Tissue separation and cryopreservation of all individuals was performed as previously described (Sipkema et al., 2011), using a final concentration of 25% glycerol in sterile artificial sea water [ASW, 33 g/L (Reef Crystals, Blacksburg, VA, United States)] as cryoprotectant. Cryopreserved samples were transported to Netherlands and stored at −80°C for several months. Four seawater samples (2 L each) were collected at the sampling site in proximity to the sponges and filtered over 0.2 μm pore size polycarbonate filters. Filters were stored at −20°C until DNA extraction was performed.

A cryopreserved cell suspension of A. aerophoba (Aa18) was thawed and divided into four 150 μL aliquots. Two aliquots served as total prokaryotic community controls (Cryostock samples) and were stored at 4°C for a few hours until DNA extraction as described below. To assess the viable prokaryotic community after cryopreservation (Figure 1), two aliquots were treated with a propidium monoazide dye (PMAxx, Biotium Hayward, CA, United States) following manufacturer’s instructions (here referred to as PMA samples). PMA permanently modifies DNA of membrane-impaired, dead cells and thus only DNA from viable cells with intact membranes is amenable to PCR amplification and sequencing (Nocker et al., 2007; Emerson et al., 2017). After photo-activation of the dye (using the PMA-Lite LED Photolysis Device), cells were pelleted for subsequent DNA extraction.

Total community DNA was extracted from i. cryopreserved sponge and bacteria cell suspensions, ii. filtered environmental seawater samples, iii. the four aliquots of the viability tests, iv. samples from 7 days and 4 weeks of Liquid-Solid cultivation, and v. the MiniColumn samples (Figure 1). Picked colonies were directly used for PCR without prior DNA extraction. Seawater filters were cut into small fragments and suspended in 400 μL of STAR buffer (Roche Diagnostics Corporation, Indianapolis, IN, United States). Sediment samples of the MiniColumns were mixed with 1 mL of STAR buffer and vortexed thoroughly. After letting the sediment settle, the supernatant was transferred to bead-beating tubes. For all other sample types, cells were pelleted at 10 000 g for 10 min, resuspended in 400 μL of STAR buffer and transferred to 2 mL sterile bead-beating tubes filled with 0.1 g of 0.11 mm Zirconia beads (BioSpec Products, Bartlesville, OK, United States). Cell lysis was achieved by bead-beating using a Precellys 24 tissue homogenizer (Bertin Instruments, Montigny-le-Bretonneux, FR) for two times 30 s at a speed setting of 5.5 m/s. Total community DNA was extracted using a Maxwell 16 Instrument in combination with the Maxwell 16 Tissue LEV Total RNA Purification Kit customized for microbial DNA extraction (Promega, Madison, WI, United States) following manufacturer’s instructions. The DNA was eluted in 30 μL DNAse free water and quantified using Nanodrop 2000c Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States) and gel electrophoresis.

Amplicons of the V4 region of the 16S rRNA gene were generated from extracted DNA, using a two-step PCR protocol. In the first step, PCR amplicons were generated using the primers 515f-806rB (Walters et al., 2015) with an attached linker sequence (UniTag) (van Lingen et al., 2017). PCR was performed in triplicate and the reaction mix contained 5 μL 5X Phusion HF buffer, 0.5 μL dNTPs (10 mM), 1 μL UniTag1-515f primer (5′-GAGCCGTAGCCAGTCTGC-GTGYCAGCMGCCGCGGTAA-3′) (10 μM), 1 μL UniTag2-806rB primer (5′-GCCGTGACCGTGACATCG-GGACTACNVGGGTWTCTAAT-3′) (10 μM), 0.25 μL Phusion Hot Start Polymerase (Thermo Fisher Scientific, Waltham, MA, United States), and 1–10 μL of extracted DNA (20 ng/μL). Nuclease-free water (Promega) was added to yield a total reaction volume of 25 μL. The PCR program comprised of initial denaturation at 98°C for 30 sec (10 min for colony PCR), followed by 25 cycles of denaturation at 98°C for 45 sec, annealing at 50°C for 30 sec, elongation at 72°C for 10 sec, and a final extension step at 72°C for 7 min. PCR products were visualized on agarose gel. In the second step PCR a sample-specific barcode was added as described below.

The colonies picked from the Plates experiment were also identified using 16S rRNA gene amplicon sequencing. First step PCR amplicons were generated directly by colony PCR, without prior DNA extraction. Each colony was PCR amplified following the protocol described above, in a single reaction using 1 μL of the nuclease free water template. All colony PCR reactions with a negative result were repeated. To reduce the number of samples for 16S rRNA gene profiling, on average 40 colony PCR products were pooled at approximately equimolar amounts (based on gel band intensity) into one sample for the second step PCR. For the AP_negative set of plates, one PCR product pool corresponded to the amplified colonies from one medium plate (Glc_1x – Rha_50x, Table 2). Due to the low yield of colonies from AP_positive plates, insufficient numbers of positive PCR products were available per pool, thus colonies from more than one plate were pooled to keep the number of PCR products consistent within one sample for 16S rRNA gene amplicon sequencing. These samples were labeled P13II to P19II (Table 2) and resulted in mixed carbon source and medium dilutions. All PCR products that resulted from colonies picked after 180 days of incubation were pooled into four samples for amplicon sequencing and labeled 2ndP1I to 2ndP2II (Table 2).

Second step PCR reactions were done in triplicate and contained: 10 μL 5X Phusion HF buffer, 1 μL dNTPs (10 mM), 5 μL of sample specific, mixed forward and reverse Unitag-Barcode primer, 0.5 μL Phusion Hot Start Polymerase (Thermo Fisher Scientific, Waltham, MA, United States), 28.5 μL nuclease-free water and 5 μL DNA template. The PCR program was initial denaturation at 98°C for 30 sec, followed by five cycles of denaturation at 98°C for 10 sec, annealing at 52°C for 20 sec, elongation at 72°C for 20 sec, and a final extension step at 72°C for 7 min. PCR products (∼350 bp) were purified using the HighPrep PCR product purification kit (MAGBIO GENOMICS, Gaithersburg, MD, United States) and quantified using the Qubit fluorometer BR assay kit (Molecular Probes by Life Technologies, Thermo Fisher Scientific, Waltham, MA, United States). Equimolar amounts of purified PCR amplicons were pooled into libraries and sent for sequencing using the Illumina MiSeq platform (GATC-Biotech, Konstanz, Germany).

Raw paired-end MiSeq sequencing reads were analyzed using the NG-Tax pipeline (Ramiro-Garcia et al., 2018) by filtering to reads with perfectly matching primers and barcodes, which were used to demultiplex reads by sample. Forward and reverse reads were both truncated to 100 nt and concatenated. Unique sequences (operational taxonomic units, OTUs) occurring above a minimum 0.1% relative abundance threshold per sample were picked, and subjected to non-reference based chimera checking, where the parent sequence needed to be more abundant by a 0.5 ratio than the chimeric sequence. Taxonomy was assigned to OTUs using a customized version of the SILVA_128_SSU Ref database (Quast et al., 2013). Three samples from the Plates experiment, namely Rha_10x, All_50x and P18II did not pass quality criteria and were excluded from further analyses. Demultiplexed, raw reads have been deposited at the European Nucleotide Archive (ENA) under accession number PRJEB31820¹.

The resulting biom tables and tree files were analyzed in R version 3.4.3² using the phyloseq package version 1.20.0 (McMurdie and Holmes, 2013) and the microbiome package version 1.1.2 (Lahti et al., 2017) for data import, storage, quality control, data transformations, subsetting, ordination methods and diversity analyses. OTUs classified as Chloroplasts were discarded from the analysis.

Diversity indices for total and viable community fractions were estimated as implemented in phyloseq, and significance was tested using Kruskal–Wallis rank sum test. DESeq2 (Love et al., 2014) as implemented in phyloseq was used to normalize the OTU table of viability-test samples and to detect differentially abundant taxa in duplicate samples. OTUs with p_adj < 0.01, corrected for multiple testing, were considered significantly differentially abundant.

The ape package version 5.0 (Paradis et al., 2004) was used for phylogenetic tree handling, and the picante package version 1.6-2 (Kembel et al., 2010) was used to calculate Faith’s phylogenetic diversity. Non-parametric tests on medians of phylogenetic diversity per sample group (Seawater, Sponges, Plates, MiniColumns and Liquid-Solid) were performed using the Mann–Whitney test (Mann and Whitney, 1947). Sequence counts were transformed to relative abundance per sample, and distances between samples were calculated using both weighted and unweighted UniFrac distances (Lozupone et al., 2011) as implemented in phyloseq. Principal Coordinates analyses on weighted UniFrac distance metrics were performed to visualize beta diversity differences for sample groups. The degree of dispersion in beta diversity was calculated using the betadisper function, and the adonis test with 999 permutations was used to test significance of associated variables affecting the clustering as implemented in the vegan package version 2.4-5 (Oksanen et al., 2016). Venn-diagrams displaying shared and unique OTUs per sample group were created using Venny 2.1 (Oliveros, 2007) and redrawn using Microsoft PowerPoint. Relative abundances of the top 100 cultivated taxa were visualized using the pheatmap package version 1.0.8 (Kolde, 2012) and refined using Adobe Photoshop.

Canonical correspondence analysis (CCA) as implemented in phyloseq was conducted on relative abundance data to analyze the effect of the variable Medium dilution on the Plates sample set and of the variable location within the MiniColumns sample set. Significance values were calculated using the anova.cca function as implemented in vegan. To assess the overall influence of aeroplysinin-1, relative abundances of taxa were calculated for the sum of reads of samples with or without the antibiotic. The linear model relating oxygen concentration and phylogenetic diversity was fitted as implemented in the R stats package. Ggplot2 version 2.2.1 (Wickham, 2016) was used for data visualization. Full code and input files are available on GitHub³.

The viable fraction of cryopreserved A. aerophoba cell suspensions was analyzed by comparing the differences in community composition of the total (cryopreserved samples) and viable fractions (PMA treatment of cryopreserved samples). The number of OTUs observed in the viable fractions was not significantly different to that of the total communities (n = 104 vs. 109; p = 0.1213, Kruskal–Wallis rank sum test). Similarly, no statistically significant difference was observed for the phylogenetic diversity (PD_avg = 11.5 vs. 11.9; p = 0.1213). A total of 100 OTUs (88.5%) were shared between viable and total communities, whereas nine and four OTUs were uniquely detected in cryopreserved samples and viable fractions, respectively. We used DESeq2 to identify 15 OTUs that were significantly (p_adj < 0.01) differentially abundant in viable fractions (Figure 2). In total, these OTUs accounted for a cumulative relative abundance of 12.4% in the cryopreserved community, and 4.5% in the viable fraction. Furthermore, of these 15 OTUs, only Cyanobacteria OTU-451 and several acidobacterial OTUs represented predominant members of the sponges’ cryopreserved prokaryotic community that showed a notable decrease in relative abundance in the viable fractions after PMA treatment. PAUC34f OTU-540 was the only taxon that showed a significant increase in relative abundance in the viable fractions, however, this OTU was absent or below the detection threshold in the total cryopreserved communities.

Aplysina aerophoba-derived bacteria were grown in three different cultivation experiments resulting in 63 samples referred to as “cultivation samples”: (a) a variety of plates with gelrite-solidified media tailored toward the cultivation of Poribacteria, (b) a “Liquid-Solid”cultivation where bacteria were incubated in liquid medium and subsequently transferred to petri dishes containing the same medium solified with gelrite and covered with the same liquid growth medium, and (c) in “MiniColumns” (Figure 1). Table 2 provides an overview of the 39 “Plates,” four “Liquid-Solid” and 20 “MiniColumns” samples derived from the cultivation experiments.

In order to compare the phylogenetic richness of prokaryotes within A. aerophoba, the surrounding seawater and the communities that were enriched within the different cultivation approaches, Faith’s phylogenetic diversity was calculated (Figure 3A). Phylogenetic diversity was substantially lower in samples derived from any of the cultivation experiments, indicating a strong bias toward cultivating only certain phylogenetic groups. Interestingly, this was contrasted by the observation that the total number of unique OTUs recovered by all cultivation experiments together was higher compared to that observed for the original sponge samples (Mann–Whitney Test p < 0.05) (Figure 4A). It should be noted, however, that the number of samples derived from cultivation experiments was much higher (n = 63) as compared to the number of sponge samples analyzed (n = 7).

Sponges, seawater and cultivation samples showed little overlap in prokaryotic community composition (Figure 3B), and sample type significantly (adonis R² = 0.83201, p = 0.001) explained the separation. We observed this large heterogeneity with respect to prokaryotic community composition for cultivation samples (Figure 3B). Prokaryotic communities derived from plates were found well separated from the MiniColumn samples, whereas the Liquid-Solid samples were positioned in between. Even though large heterogeneity of recovered bacteria was also observed within each of the cultivation method sample groups (Plates, Liquid-Solid, MiniColumns), the degree of dispersion in beta diversity was not significantly different (permutation test of beta dispersion F.Model = 2.1915, Pr(>F) = 0.084). This means that all groups exhibited similar levels of variation within the group, confirming that the differences between cultivation method sample groups were caused by the cultivation method and not by chance. For unweighted Unifrac analysis, a similar result was obtained (Supplementary Figure S4).

In total, the three cultivation experiments yielded 256 cultured OTUs from seven bacterial phyla, namely Actinobacteria (1.2% of cultivated OTUs), Bacteroidetes (5.1%), Firmicutes (29.7%), Planctomycetes (3.1%), Proteobacteria (56.6%), Tenericutes (0.4%) Verrucomicrobia (0.4%) and an unidentified phylum related to the Planctomycetes (3.5%). Amongst the Proteobacteria, Alphaproteobacteria (44.5%) and Gammaproteobacteria (10.9%) dominated the cultivated fraction. No Poribacteria OTUs were detected in any of the cultivation experiments. On average, 17 OTUs were detected per cultivation sample, with a minimum of seven and a maximum of 36 OTUs. This confirmed that the number of OTUs recovered per colony PCR pool from the plates was in the range of the expected numbers based on the fact that on average PCR products of 40 colonies were pooled into one sample.

Of the 60 most abundant OTUs associated to A. aerophoba, only one Synechococcus cyanobacterial OTU was also detected in seawater, and none in the cultivated fractions. Among the 60 most prominent seawater OTUs, three Synechococcus cyanobacterial OTUs were also detected in sponges, and two [Ruegeria sp. (Alphaproteobacteria) and Idiomarina sp. (Gammaproteobacteria)] were recovered by cultivation. Of the 256 OTUs recovered by cultivation, four OTUs, classified as Pseudovibrio sp. (Alphaproteobacteria), Halomonas sp. (Gammaproteobacteria), Flavobacteriaceae family and Lutimonas sp. (both Bacteroidetes) were shared between the sponge tissue and the cultivated fraction (Figure 4A). Another three OTUs, Halomonas sp., Idiomarina sp. and Hyphomonas sp. (Alphaproteobacteria) were shared between the seawater and the cultivated fraction. One OTU, Ruegeria sp. (OTU-3), the overall most dominant OTU within the Plates approach, was shared between seawater, sponge tissue and the cultivated fraction.

The different cultivation approaches shared merely 10 OTUs, classified as Ruegeria sp., Pseudovibrio sp., Microbulbifer sp. and five other Rhodobacteraceae family (Alphaproteobacteria) OTUs, indicating that most bacteria were only obtained by one cultivation approach (Figure 4B). The MiniColumns and the Liquid-Solid approach shared another five OTUs from the Rhodobacteraceae family, four of which were further classified as Pseudovibrio. The Liquid-Solid and Plates approaches shared further nine OTUs, five of which belonging to the Rhodobacteraceae family (three Pseudovibrio), and another four to the gammaproteobacterial Microbulbifer genus. The MiniColumns and the Plates approach shared another 18 OTUs belonging to the three phyla Firmicutes, Proteobacteria (Alpha, Beta and Gamma) and Actinobacteria. Interestingly, all of these except for one Pseudovibrio OTU were detected only in the water samples of the MiniColumns, as well as on the plates.Pseudovibrio sp. OTU-621 was recovered from all but one cultivation derived samples and dominated the cultivable fractions from 3.0 to 89.7% relative abundance per sample (Figure 5). This OTU was also detected in sponge samples with an average of 0.3% relative abundance. In addition to the high relative abundance of OTU-621, a large number of other OTUs from the genus Pseudovibrio were recovered from a variety of growth media and conditions. Further, Ruegeria, Bacilli (Firmicutes) and various Halomonas OTUs were also found in samples derived from many conditions. However, many recovered taxa had a preference for one of the cultivation methods. This difference was especially pronounced with respect to the exclusive presence of Clostridia (Firmicutes) OTUs in the anaerobic sediment fractions of the MiniColumns. All sediment fraction samples (especially of MiniColumn2) were also enriched for OTUs that could not be assigned to a phylum using the SILVA_128_SSU Ref database (OTU-2, 11, 17, 60, 89). The most abundant of these OTUs (OTU-17, up to 5.8% relative abundance in sediment fractions of MiniColumn2) exhibited maximally 97.5% nucleotide sequence similarities to unidentified Planctomycetes, recently detected by deep-cultivation from the calcareous sponge Clathrina clathrus (e.g., GenBank accession no. CP036425.1). Furthermore, an OTU belonging to the rarely cultivated phylum Tenericutes was detected with approximately 2% relative abundance in three samples derived from MiniColumn3. On the other hand, Flavobacteriia (Bacteroidetes) OTUs and Microbulbifer OTUs were nearly absent in the MiniColumns, but were recovered at high relative abundances from plates. A prolonged incubation time of 180 days led to the recovery of members of the Verrucomicrobia and Planctomycetes on plates. The Liquid-Solid approach resulted in high recovery of Microbulbifer and other Gammaproteobacteria from the genus Psychrobacter, of which the latter was detected solely in the Liquid-Solid cultivation.

From the different plates 1758 colonies were picked, including 1129 colonies from the AP_negative set and 629 colonies from the AP_positive set. Aeroplysinin-1 presence led to a lower number of colonies and impacted their size and morphology. Most colonies were small and translucent on AP_positive plates, whereas on AP_negative plates larger and partly pigmented colonies could be observed (Supplementary Figure S1). Since the final concentration of DMSO was only 0.3% in the inoculum suspension, it is probably safe to assume that the lower yield of colonies on AP_positive plates was due to the toxicity of the antibiotic.

The 16S rRNA gene of 1463 colonies could be successfully PCR-amplified and sequenced. While Bacteroidetes were completely inhibited by AP, most other taxa grew also in the presence of the antibiotic (Figure 5). The Rhizobium (Alphaproteobacteria) OTU-193 on P15I was the only OTU exclusively detected on an AP_positive plate. Overall, the addition of AP resulted in a consistently higher (61% ± 17% vs. 27% ± 17%) relative abundance of Pseudovibrio sp. OTU-621, while reducing the relative abundance of Ruegeria sp. OTU-3 from 54% ± 20% to 30.4% ± 13% (Supplementary Figure S2). PCoA analysis on weighted UniFrac distances revealed that AP significantly (adonis R² = 0.25784, p = 0.001) affected the composition of OTUs recovered (Figure 6).

To assess the influence of medium dilution and carbon source, samples from the Plates experiment were subset to only AP_negative since for this part of the dataset the variables medium dilution and carbon source could be consistently assigned. In total, 1x diluted media yielded 40 OTUs from the phyla Actinobacteria, Firmicutes and Proteobacteria, where else 10x and 50x diluted media resulted in 50 and 38 OTUs, respectively, from Actinobacteria, Bacteroidetes, Firmicutes and Proteobacteria. Overall, phylogenetic diversity increased significantly (p < 0.01) in 50x diluted media compared to the 1x diluted media, whereas carbon source did not significantly impact the recovered bacterial diversity (Figure 7A). Consistently, medium dilution (adonis R² = 0.25771, p = 0.015) and not carbon source (adonis R² = 0.47646, p = 0.165) was the significant variable explaining community composition patterns in constrained coordinate analysis (Figure 7B).

Certain bacterial classes were associated with specific medium dilutions such as Flavobacteriia or Bacilli, which were mainly recovered on highly diluted or less diluted media, respectively. Proteobacteria were cultivated on all medium dilutions. Phylogenetic diversity was highest in the 50x diluted media containing galactosides, followed by mucin and amino acids, of which the latter also showed the highest relative abundance of Flavobacteriia.

Since on plates, the 50x diluted amino acids medium (Aa_50x) resulted in high phylogenetic diversity diversity of cultivated bacteria and supported the growth of Flavobacteriia, this medium was selected to create liquid cultures. We observed biofilm formation on the bottom of the liquid culture wells after 7 days, upon which we transferred the cultures to solid Aa_50x medium and overlaid it with the same liquid medium for a total of 4 weeks of incubation. The cultures were dominated by Pseudovibrio OTUs after 7 days as well as after 4 weeks. Nonetheless, the Liquid-Solid cultivation resulted in the recovery of genera that were not detected in other cultivation methods, such as the gammaproteobacterial genera Colwellia, Thalassomonas, and Psychrobacter, as well as the alphaproteobacterial genera Leisingera and Kordiimonas at low relative abundances. Faith’s phylogenetic diversity was not significantly different than for the other cultivation approaches (Figure 3A), indicating a similar level of recovered species richness.

At the end of the 130 days incubation period, the oxygen gradient along the depth in the MiniColumns was measured. The oxygen concentration in the headspace of all columns was 88.4% (±6.4%) air saturation, and the four replicate columns had different oxygen profiles along the column: MiniColumn4 was completely anaerobic at the top of the water column, while other columns had oxygen concentrations around 30–40% at the water surface with a gradual decrease to complete anoxia in the sediment fractions (Figure 8A). The negative control maintained a 100% oxygen saturation along the whole depth of the column (data not shown). Despite visual differences amongst the columns (Supplementary Figure S3), identical locations in different columns promoted the growth of very similar bacterial communities (Figure 5). Location within the column was the main driver of community profile differences, with a range of Clostridia OTUs and unidentified OTUs related to the Planctomycetes representing indicative taxa for the anaerobic sediment fractions of the columns (Figure 8C). Linear model regression analysis revealed a positive, though not significant correlation of oxygen concentration with phylogenetic diversity (PD). Both high (PD = 3.56) and low (PD = 1.99) phylogenetic diversities were found for anaerobic conditions, whereas microaerophilic and aerobic parts of the columns enriched for intermediately diverse communities (Figure 8B).

Since sponge-associated microorganisms remain mostly recalcitrant to cultivation efforts, we aimed to investigate if the often necessary cryopreservation impacted the viability of sponge associated microorganisms. We did not detect significant differences between the total and viable communities after cryopreservation. We thus conclude that the majority of sponge-associated microorganisms survive sample processing and storage at −80°C for several months and that their current uncultivability is most likely not due to cryopreservation. Nonetheless, these results should be interpreted carefully, since the uncultivability of sponge-associated microorganisms makes it difficult to validate the viability staining techniques applied, which has been shown to be necessary in some cases (Chang et al., 2010; Emerson et al., 2017). To the best of our knowledge, only Esteves et al. (2016) analyzed the viable fraction of two Australian sponge species, after tissue separation and cell fractionation to obtain microbial pellets. Overall, they observed a loss of OTUs of up to 51% in the viable fraction, whereas we observed only a marginal loss. This difference might be due to the many steps of filtration and centrifugation involved in the cell fractionation as done by Esteves et al. (2016), which might negatively affect the viability of microorganisms, as well as lead to the loss of rare taxa.

Here, we described three different strategies to cultivate sponge tissue-derived bacteria, which all resulted in the recovery of communities with a very different composition compared to the original sponge samples and surrounding seawater samples (Figure 3). As negative controls in the cultivation experiments consistently showed no microbial growth, we conclude that these cultivated OTUs must be derived from rare members of the sponge community that were below the applied detection threshold.

Defining cultivability as the number of taxa that could be cultured divided by the total number of taxa detected in the sponge tissue, we report here a recovery of 2% of the sponge-associated taxa by cultivation. This falls into the range of other sponge-microbe cultivation experiments (Sipkema et al., 2011; Lavy et al., 2014; Keren et al., 2015; Esteves et al., 2016; Versluis et al., 2017), which report a recovery rate of 0.1 to 14%. Even though different sponge species harbor species-specific and highly divergent microbiota, it has been shown that their cultivable fractions are comparable (Li et al., 2007; Hardoim et al., 2014) since rare generalists are proliferating under laboratory conditions.

Despite employing newly designed media containing various unusual carbon sources, the novelty of recovered isolates was moderate. Except for the unidentified OTUs related to Planctomycetes, which were not obtained as pure isolates, the 100 most predominant cultivated OTUs had a sequence identity of >98% to the closest cultivated relative in the NCBI database. This may be partially explained by the high degree of sequence conservation of the V4 region of the 16S rRNA gene (Sun et al., 2013). Nevertheless, five OTUs that were present in the sponge tissue were also recovered by cultivation. Amongst these shared taxa, the alphaproteobacterial Pseudovibrio and Ruegeria OTUs were the most frequently cultivated. Pseudovibrio species represent low abundant sponge symbionts which are known as versatile, opportunistic bacteria capable to adapt to a wide range of cultivation conditions (Muscholl-Silberhorn et al., 2008; Bondarev et al., 2013; Versluis et al., 2017; Fróes et al., 2018). With this study we could furthermore add aeroplysinin-1 to the list of antibiotic resistances exhibited by Pseudovibrio species (Versluis et al., 2017). The genus Ruegeria is part of the abundant seawater-dwelling Roseobacter lineage (Buchan et al., 2005; Wagner-Döbler and Biebl, 2006) and has frequently been isolated from sponges and other marine environments (Mitova et al., 2004; Muscholl-Silberhorn et al., 2008; Esteves et al., 2013; Rua et al., 2014). Two other shared OTUs (OTU-168, 169) belong to the class Flavobacteriia, members of which are regularly detected in and isolated from marine sponges, however, their roles as sponge symbionts remain to be investigated (Lavy et al., 2014; Montalvo et al., 2014; Horn et al., 2016; Yoon et al., 2016; Versluis et al., 2017).

Increased incubation time of 180 days resulted in a broader detected cultivated bacterial diversity, as evidenced by the detection of Planctomycetes and Verrucomicrobia OTUs on plates. To our knowledge, this is the first report of an A. aerophoba derived member of the Verrucomicrobia phylum detected during cultivation, and OTU-204 exhibits only 91% sequence similarity to other cultured sponge-derived Verrucomicrobia: Rubritalea marina (Scheuermayer et al., 2006) and Rubritalea spongiae (Yoon et al., 2007). Planctomycetes have been obtained from A. aerophoba before (Pimentel-Elardo et al., 2003), however, the unidentified OTU-17, which was enriched to up to 5% relative abundance in sediment fractions of the MiniColumns, exhibited 97.5% sequence similarity to the closest cultivated neighbor, a member of the Planctomycetes recently obtained by deep-cultivation from the calcareous sponge Clathrina clathrus (GenBank: CP036425.1, Wiegand et al., 2019). Other close uncultured phylogenetic relatives of OTU-17 have been found associated to diseased tissue of a Caribbean coral (AF544881, 96% sequence identity, Pantos et al., 2003), as well as in seawater from 3000 m depth close to the Mariana Trench (AB703899, 96% sequence identity, Nunoura et al., 2015).

In this study, we extended the attempt of Lavy et al. (2014) to design cultivation media for the candidate phylum Poribacteria. The predicted genetic potential for the utilization of urea and organic phosphorous, as well as the potential to degrade a variety of carbohydrates was taken into account (Siegl et al., 2011; Kamke et al., 2013) to design nine defined media. Additionally, trace metal and micronutrient solutions were tailored toward meeting potential co-factor requirements of annotated Poribacteria enzymes (Siegl et al., 2011; Kamke et al., 2013). Furthermore, all media were diluted to account for Poribacteria potentially being oligotrophs, and incubation time was prolonged to 180 days to account for potential slow growth (Vartoukian et al., 2010; Prakash et al., 2013). Gellan gum was used as solidifying agent to avoid inhibitory effects of agar (Janssen et al., 2002; Overmann, 2010). However, none of the incubation conditions applied enabled the cultivation of Poribacteria. Further adaptations of the cultivation conditions might resolve this in the future, such as selective enrichment based on the predicted Wood-Ljungdahl pathway (Siegl et al., 2011). Also, the inclusion of siderophores (Vartoukian et al., 2016) or the provision of helper strains might aid in the cultivation of this sought-after bacterial candidate phylum (Morris et al., 2008; Davis et al., 2014; Pande and Kost, 2017).

We observed that the sponge-derived antibiotic aeroplysinin-1 (AP) strongly inhibited bacterial growth and led to decreased cultivated diversity on the plates. The number, size and pigmentation of colonies was negatively impacted, hinting at an overall cellular toxicity of this antibiotic (Supplementary Figure S1). AP seemed to affect taxa differently, as observed by the consistently higher relative abundance of the sponge-associated Pseudovibrio sp. OTU-621 in the presence of AP as compared to the seawater-derived Ruegeria sp. OTU-3. Ruegeria species seem to be negatively affected by AP, which supports the notion that this antibiotic contributes to protecting the sponge from seawater-derived bacterial infections (Lipowicz et al., 2013). Furthermore, all cultivable members of the phylum Bacteroidetes were completely inhibited by the antibiotic.

Within the AP_negative set, the highest media dilutions (50x) supported the growth of the most phylogenetically diverse bacterial community. This observation supports the notion that many marine organisms require oligotrophic conditions for successful cultivation and might be inhibited by increased substrate concentrations (Stevenson et al., 2004; Hanson et al., 2007; Pham and Kim, 2012), which was also demonstrated for sponge-derived bacteria (Hentschel et al., 2001; Muscholl-Silberhorn et al., 2008). Medium dilution, not carbon source, was the only significant factor driving the development of specific communities within the Plates experiment. This could indicate that many marine heterotrophic microorganisms are equipped with the genetic potential to degrade a wide range of carbohydrates, which was reported for e.g., Poribacteria, Pseudovibrio or marine Flavobacteriia (Kamke et al., 2013; Barbeyron et al., 2016; Alex and Antunes, 2018), and are more influenced by the concentration of nutrients.

Overall, the recovered prokaryotic communities clustered based on the cultivation method applied, indicating that macro-environmental conditions such as liquid (water samples in MiniColumns, Liquid-Solid cultivation) or solid medium interface had a more pronounced effect than micro-environmental conditions such as carbon source. For example, the Aa_50x medium applied in plates enriched for Flavobacteriia, while in the Liquid-Solid cultivation the same medium supported the growth of gammaproteobacterial genera Psychrobacter, Thalassomonas, and Colwellia, which were not detected in other cultivation experiments within this study. Liquid cultivation approaches have only rarely been applied to cultivate sponge-associated bacteria (Sipkema et al., 2011), and have been shown to result in lower species diversity as compared to solid agar plates (Schoenborn et al., 2004). However, liquid cultivation can result in the recovery of OTUs not detected in other cultivation methods (Sipkema et al., 2011), which we could also observe in this study.

To our knowledge, this is the first report applying a Winogradsky column approach to sponge-derived samples. Winogradsky columns are enclosed, self-sustaining microbial ecosystems, where chemical gradients create niches for different microorganisms and metabolic interactions are maintained (Madigan et al., 2014; Rundell et al., 2014).

During incubation, oxygen gradients ranging from moderately aerobic in the water columns to complete anoxia in the sediment fractions developed. Community composition varied between the aerated parts of the columns, which exhibited higher relative abundances of mostly aerobic bacteria such as Actinobacteria and Bacilli, and the anaerobic parts of the columns, which were dominated by Clostridia. The unidentified OTU-17 related to Planctomycetes was enriched to up to 5% relative abundance in the anaerobic sediment fractions of the MiniColumns, hinting toward an anaerobic lifestyle of this microbe. Most fractions of the MiniColumns also exhibited high relative abundances of Pseudovibrio, confirming the facultative anaerobic lifestyle of this bacterial genus (Shieh et al., 2004).

The overall phylogenetic diversity of enriched microorganisms was positively correlated with oxygen concentration, even though some exceptional samples exhibited high phylogenetic diversity in anoxic conditions (Figure 8). These results indicate that anaerobic cultivation approaches can yield a comparable diversity, with distinct and potentially novel recovered taxa, as indicated by Lavy et al. (2014). In their study, six out of eight novel taxa with <94% sequence identity to the closest cultivated strains were recovered from strictly anaerobic conditions. Also for A. aerophoba and other sponges, periodically occurring tissue anoxia has been observed, indicating a potential niche for anaerobic microbes (Hoffmann et al., 2005, 2008) and the need to further explore anaerobic cultivation techniques for culturing sponge-associated microorganisms.