- Open Access
Understanding microbial community dynamics to improve optimal microbiome selection
© The Author(s). 2019
- Received: 1 February 2019
- Accepted: 21 May 2019
- Published: 3 June 2019
Artificial selection of microbial communities that perform better at a desired process has seduced scientists for over a decade, but the method has not been systematically optimised nor the mechanisms behind its success, or failure, determined. Microbial communities are highly dynamic and, hence, go through distinct and rapid stages of community succession, but the consequent effect this may have on artificially selected communities is unknown.
Using chitin as a case study, we successfully selected for microbial communities with enhanced chitinase activities but found that continuous optimisation of incubation times between selective transfers was of utmost importance. The analysis of the community composition over the entire selection process revealed fundamental aspects in microbial ecology: when incubation times between transfers were optimal, the system was dominated by Gammaproteobacteria (i.e. main bearers of chitinase enzymes and drivers of chitin degradation), before being succeeded by cheating, cross-feeding and grazing organisms.
The selection of microbiomes to enhance a desired process is widely used, though the success of artificially selecting microbial communities appears to require optimal incubation times in order to avoid the loss of the desired trait as a consequence of an inevitable community succession. A comprehensive understanding of microbial community dynamics will improve the success of future community selection studies.
- Artificial microbiome selection
- Microbial communities
- Microbial ecology
- Polymer degradation
- Chitin degradation
- Ecological succession
- Microbial community dynamics
Evolution is able to act upon multiple levels of biological organisation [1-5]. It had previously been contested that a whole microbial community may be used as a unit of selection artificial microbiome selection and that a community may become progressively better at a selective process over successive transfers [2, 6-8]. The artificial selection of a measurable and desirable trait is thought to outperform traditional enrichment experiments, as it bypasses community bottlenecks and reduces stochasticity . Artificial selection has been shown to induce statistically significant responses, in both microcosm [2, 8] and computational ecosystems . While microbial communities with desirable phenotypes have been achieved, the results of these experiments have been limited and the composition of these communities has not generally been determined. The underlying mechanisms and key players in this selection have therefore not been identified, nor have the growth parameters involved (e.g. incubation time) been systematically optimised .
Microbial communities are known to go through distinct stages of community succession, where they may see large enrichments of different groups of organisms [10-16]. During the colonisation and degradation of the abundant marine polymer, chitin, three phases of community succession were previously observed: (1) selection of colonising organisms, (2) selection of chitin degraders and (3) chitin degraders are overtaken by cheaters . Cheaters, often called cross-feeders, are organisms that are not metabolically capable of carrying out a particular process themselves, but are able to benefit from the public goods generated by others [17-19]. For example, in the environment, it has been shown that the number of organisms capable of taking up the by-products of chitin hydrolysis is far higher than the number that encodes for chitinases . Furthermore, marine microorganisms ‘leak’ a large diversity of organic matter which in turn may be used by the rest of the community [19, 20] enhancing microbial interdependencies . It has however been previously observed that if the abundance of cheaters becomes sufficiently high, then access to the resource may even be blocked completely , leading to a loss of community function.
The undulations between cooperation and competition drive niche-specialisation and higher-level community organisation . The structure of the microbial community has been suggested to significantly alter the observed phenotype in artificial microbiome selection experiments . Community structure may be altered through changes in species' composition or interactions between organisms, ultimately leading to changes in the community phenotype. Computational models have shown that community structure, with  or without genetic changes , can be responsible for differences between the phenotypes of a community subjected to a directed selection and one that is randomly selected. Hence, the understanding of microbial community ecology suggests that controlling microbial community dynamics is important for achieving a high-functioning microbial community.
In the present study, we aimed to determine the mechanisms behind artificial microbiome selection and early microbial community succession in order to optimise the selection of a process, i.e. chitin degradation. Chitin is one of the most abundant polymers on Earth (i.e. the most abundant polymer in marine ecosystems) constituting a key component in oceanic carbon and nitrogen cycles . Many microorganisms are already known to degrade chitin, and the enzymes and pathways used to do so are well characterised . We found that a microbiome could be artificially evolved to achieve higher chitinase activities, but there were certain methodological caveats to this selection process. We found that the incubation time between transfers needed to be continuously optimised in order to avoid community drift and decay. Microbial community composition was evaluated, and we confirmed that, if transfer times are not continuously optimised, efficient biodegrading communities are rapidly taken over by cheaters and predators with a subsequent loss of degrading activity.
First artificial selection experiment: process optimisation
While the 9-day incubation experiment gave an overall negative trend, shortening the incubation times to the chitinase maxima drastically increased the benefits of artificial community selection, i.e. initially to 4 days (generation 15, Fig. 2b) and later to 2 days (generation 20, Fig. 2c). This suggests that the selection of an efficient chitin-degrading community shortens the time required not only to reach maximum chitinase activity, but also to enter decay due to community succession.
Microbial community succession
We carried out MiSeq amplicon sequencing of the 16S and 18S rRNA genes to characterise the microbial community succession that occurred within the first selection experiment and, by this way, gain insight into the strong variability in chitinase activity observed over time. We sequenced the communities that were used as the inoculum for each of the 20 transfers, both 9- and 4-day-long experiments, as well as the community obtained from the daily monitoring of the incubation period for transfer 20. This data was processed using both Mothur  and DADA2 workflows [28, 29], obtaining similar results (Additional file 1: Figures S3 and S4). DADA2 results are presented here as this workflow retains greater sequence information, better identifies sequencing errors and gives higher taxonomic resolution . Unique taxa are therefore amplicon sequence variants (ASVs) rather than operational taxonomic units (OTUs).
Community succession over the 4-day incubation period within transfer 20
The SIMPER analysis of the 18S rRNA gene highlighted ASV2 (Cafeteria sp., contributing to 34% of the community variation between 4 days, p = 0.016), ASV4 (Paraphysomonas, 10% variation, p = 0.023), ASV1 (Cafeteria sp., 6% variation, p = 0.392), ASV6 (Apsidica, 5% variation, p = 0.040) and ASV3 (Incertae Sedis, 5% variation, p = 0.059) as the five main ASVs contributing to 60% of the community variation over 4 days (Fig. 3c). ASV2, which was 96% similar to the bactivorous marine flagellate Cafeteria sp., was by far the most striking Eukaryotic organism, showing an increase in relative abundance from 2% on day 1 up to over 76% on day 4 (Fig. 3b, c). As observed in prokaryotes, eukaryotic phylogenetic groups also showed a large variation between the beginning and the end of the incubation period, mainly due to the increase of Bicosoecophyceae over time (i.e. from 2.6 to 89% relative abundance driven by both ASV1 and ASV2, Additional file 1: Figure S5).
Community succession over the entire artificial selection experiment
SIMPER analyses were carried out to determine the ASVs that most strongly contributed to the differences between groups (i.e. positive versus random selections and 9-day versus 4-day incubation times, Fig. 5b). For the 16S rRNA gene, the top 5 ASVs identified by the SIMPER analysis contributed to 35% of the community variation, while for the 18S rRNA gene, they accounted for 61% (Fig. 5b). The 16S rRNA gene ASVs 5, 7 and 11 (Crocinitomix, Terasakiella and Carboxylicivirga flava, respectively) presented a much higher abundance in the 4-day-positive selection than in any other selection (13%, 11% and 8%, respectively), suggesting that these species were the major contributors to the differentiation of these communities, as seen in Fig. 5a. As observed above for the 4-day incubation analysis, Cafeteria sp. (18S rRNA gene ASV1 and ASV2, both 96% similar) was again the most conspicuous eukaryotic organism. ASV2 was more abundant in the positive 4-day selection (32% of the relative abundance), while ASV1 was highest in the three other selections (70% and 82% in the positive and random 9-day selection, respectively, and 16% in the random 4-day selection; Fig. 5b).
Chitinase gene copies in artificially assembled metagenomes
Artificially assembled metagenomes, generated by PICRUSt  from the 16S rRNA gene amplicon sequences, were used to search for enzymes involved in chitin degradation: KEGG orthologs K01183 for chitinase, K01207 and K12373 for chitobiosidase, K01452 for chitin deacetylase and K00884, K01443, K18676 and K02564 for the conversion of GlcNAc to fructose-6 phosphate (Additional file 1: Figure S7 and Table S3) [32-34]. As expected from the measured chitinase activities, the shortened 4-day incubation experiment showed over 30 times more chitinase (K01183) gene copies than the 9-day incubation experiment (i.e. an average of 0.66 copies per bacterium were observed in the 4-day incubation experiment while only 0.025 copies per bacterium were observed over the same transfers in the 9-day experiment). Also, from the daily analysis of transfer 20, the chitinase activity was positively correlated with the normalised chitinase gene copy number (r2 = 0.57), with a peak in chitinase activity and chitinase gene copies on day 2 (i.e. over one chitinase gene copy per bacterium). The most striking result from this analysis was the strong bias of taxonomic groups that contributed to the chitinase and chitin deacetylase genes; chitinase genes were mainly detected in Gammaproteobacteria and some Bacteroidia, whereas the chitin deacetylase genes were almost exclusively present in Alphaproteobacteria. It is worth highlighting that the chitosanase gene (K01233), the enzyme required to hydrolyse the product from chitin deacetylation, chitosan, was not detected in any of the artificial metagenomes. Chitobiosidases (K01207 and K12373) and enzymes involved in the conversion of GlcNAc to fructose-6 phosphate (K00884, K01443, K18676 and K02564) were more widespread. Nevertheless, this data needs to be taken with caution as these were not real metagenomes.
Isolation and identification of chitin degraders
Bacterial isolates were obtained from the end of the artificial selection experiments to confirm the ability of the identified groups to degrade chitin. From the 50 isolates obtained, 20 were unique according to their 16S rRNA gene sequences. From these, 18 showed at least 98% similarity with one or more of the MiSeq ASVs (Additional file 1: Table S4) although, unfortunately, none belonged to the most abundant ASVs detected during the community analysis. The ability for chitin and GlcNAc degradation by each one of the isolates was assessed. We found that 16 of these isolates could grow using GlcNAc as the sole carbon source, but only 11 of these strains could grow on chitin (Fig. 4). The four remaining bacteria from the 20 isolated could not grow using chitin or GlcNAc. Most interestingly, all isolates from the class Gammaproteobacteria (n = 7) were capable of chitin degradation whereas only a smaller subset of isolates had this phenotype in other abundant taxonomic groups, such as Bacteroidia (1 out of 3) or Alphaproteobacteria (1 out of 8, Fig. 4).
We confirmed that both a cheater (i.e. isolate able to grow on GlcNAc but not chitin; Donghicola eburneus, Alphaproteobacteria) and a cross-feeder (i.e. isolate not capable of growth with GlcNAc or chitin; Phaeobacter gallaeciensis, Alphaproteobacteria) were only able to grow with chitin in the presence of a chitin-degrading isolate (Pseudoalteromonas shioyasakaiensis, Gammaproteobacteria; Additional file 1: Figure S8). As expected, while no growth was observed in the absence of the chitin degrader, both the cheater and the cross-feeder grew over two orders of magnitude more when co-cultured with the degrader (Additional file 1: Figure S8).
Second artificial selection experiment: implementing an improved selection process
Artificial selection of microbial communities is, in principle, a powerful and attractive technique which has surprisingly been used in only a limited number of studies to date [2, 6, 8], possibly due to the lack of success as a consequence of poor process optimisation. Here, using chitin degradation as a case study and a detailed analysis of the community succession, we show that artificial selection of microbial communities can be greatly improved by controlling the incubation times between transfers. We believe that the rapid succession of microbial community structure means transfers need to be done at the peak of the selected phenotypic activity (e.g. chitinase activity) or these get swiftly replaced by less efficient communities of cross-feeding microorganisms (i.e. â€˜cheatersâ€™ and grazers). Previous studies that have artificially selected microbial communities for a particular phenotype did not report optimisation of the incubation time between transfers [2, 6, 8] which, in our hands, would have resulted in a negative selection (Fig. 2). In agreement with our results, Penn and Harvey  suggested that the observed phenotype in artificial ecosystem selection experiments could be significantly affected by interactions between different species and therefore microbial community structure.
A comprehensive understanding of microbial ecology helps to explain the importance of the timing during transfers. Datta et al.  observed three distinct stages of community structure during the colonisation of chitin particles: (a) attachment, (b) selection and (c) succession. Each phase was characterised by having relatively higher abundances of organisms that were (a) good at attaching to chitin particles, (b) good at degrading chitin particles and (c) not able to degrade chitin, but able to benefit from others that could, i.e. cheaters and cross-feeders [17, 18, 22, 35]. During our first experiment, as communities became better and faster at degrading chitin, we were measuring the chitinase activity when the communities were in the succession rather than in the selection stage and, by that point, the active chitinolytic community had decayed and was dominated by a cross-feeding community (Figs. 3 and 4). Hence, it was only when selecting at phenotypic time optima when chitinase activity improved and the overall community differentiated from the random control communities (Figs. 5 and 6). Due to the stochasticity of complex microbial communities, this time optima is difficult to predict and continuous phenotypic monitoring is required. It is also interesting to note the selection of the grazer Cafeteria sp. (90% of the eukaryotic community), a genus of bactivorous marine flagellates that are commonly associated with marine detritus . The predator-prey dynamics postulated by Lotka, Volterra's equations would also support the need to shorten transfer times to favour the prey's growth, i.e. chitinolytic bacteria [37, 38]. We note, however, that there may be a lower limit to optimal transfer times (possibly around 2, 3 days, Fig. 2c) as sufficient time has to be given for slower growing marine taxa  to (i) develop a biofilm , (ii) initiate the hydrolysis of the polymer to access the sugars and (iii) allow the generation of sufficient biomass to overcome the dilution between transfers.
Interestingly, a strong successional pattern was observed at a higher taxonomic level. While Gammaproteobacteria dominated during the initial stages when chitinase activity was at its peak (accounting for over 70% of the prokaryotic community), other groups increased in abundance during the later stages (i.e. Alphaproteobacteria, Bacteroidia and Clostridia), similarly to the pattern previously observed by Datta et al.  and Enke et al. [22, 35]. The fact that Gammaproteobacteria are major contributors to chitin degradation is not new [41-46]. All Gammaproteobacteria isolates obtained from the end of the experiments were able to grow using chitin as the only source of carbon and energy (Fig. 4) confirming that this class is likely responsible for most of the chitinase activity observed. On the other hand, Alphaproteobacteria, the numerically dominant class of heterotrophic bacteria in surface oceans [47, 48], follow a cross-feeding and/or cheating life-strategy as five out of eight Alphaproteobacterial isolates could only use N-acetyl-d-glucosamine (GlcNAc) and only one could use chitin (Additional file 1: Table S4). The dependence of cheaters and cross-feeders on the presence of a chitin-degrader was confirmed with co-cultures (Additional file 1: Figure S8) and agreed with the results generated by others .
The PICRUSt metagenome analysis (Additional file 1: Figure S7) further confirmed that almost all chitinase gene copies were encoded by Gammaproteobacteria (i.e. 90%; almost one gene copy encoded per bacterium) and, to a lesser extent, by some Bacteroidia. Chitin is made up of molecules of GlcNAc linked by (1,4)-Î²-glycosidic bonds, and it has previously been found that initial degradation of chitin takes place predominantly by (i) chitinases which depolymerise the (1,4)-Î²-glycosidic bonds either at the ends or in the middle of chains, or (ii) chitobiosidase enzymes which also hydrolyse (1,4)-Î²-glycosidic bonds but only at the ends of chitin chains. Genes for the intracellular enzymes involved in GlcNAc utilisation (i.e. the transformation of GlcNAc to fructose-6-phosphate) were much more widespread amongst different taxonomic groups, highlighting the broader distribution of cross-feeding or cheating organisms which can benefit from the extracellular depolymerisation of chitin which generates freely available GlcNAc to the community. Alternative degradation of chitin may also occur by deacetylation and deamination of the GlcNAc amino sugar, transforming chitin into chitosan and cellulose, respectively, after which they can be depolymerised by a range of other enzymes (e.g. chitosanases or cellulases) [26, 49, 50]. While Alphaproteobacteria did not contribute to chitinase enzymes, they did potentially encode for most of the chitin deacetylases in the system, although no chitosanases were detected.
Chitinolytic organisms have previously been found to make up between 0.1 and almost 6% of prokaryotic organisms in aquatic ecosystems [17, 51], while over a third of the organisms in these habitats can utilise only the products of chitin hydrolysis (i.e. GlcNAc) [17, 52-54]. With Gammaproteobacteria being primarily responsible for the degradation of chitin here, the success of the artificial selection for an enhanced chitinolytic community was possibly achieved by the selective enrichment of this group between the beginning (5% of the prokaryotic community, within the expected range of Gammaproteobacteria, found within natural environments) [17, 51] and end of the experiment (75% of the community).
Finally, chitin degradation is a task that single microorganisms can perform efficiently, but other more laborious phenotypic traits are rarely carried out entirely by a single microorganism in nature. It is now well documented that a distribution of labour is favoured in natural microbial communities [55-59]. The detrimental effects of community dynamics and drift described here could be overcome by synthetically assembled microbial communities, preventing the system from moving away from the high-performing desired community. Nevertheless, this still requires a comprehensive understanding of the community structure and the necessity to select and isolate the microbes of interest.
Here, we have proven the validity of artificially selecting a natural microbial community to better degrade chitin, but have highlighted the caveats for achieving this goal, which require a better understanding of the ecology of the system. We found that continuous optimisation of incubation times is essential in order to successfully implement this process, as optimal communities rapidly decay due to their replacement by cheaters and cross-feeders, as well as the increase of potential predators such as grazers and, although not tested here, viruses. Hence, future artificial selection experiments should adjust transfer incubation times to activity maxima to successfully evolve enhanced community phenotypes and, eventually, allow the enrichment and isolation of microbes of interest.
The microbial community used as an inoculum was obtained from bulk marine debris collected during boat tows from both Plymouth Sound (Devon, UK; June 2016) and Portaferry (Northern Ireland, UK; August 2016).
Chitinase activity measurements
Chitinase activity was measured as the liberation of the fluorescent molecule 4-methylumbelliferyl (MUF) from three chitinase substrates (MUF-N-acetyl-β-d-glucosaminide, MUF-β-d-N,N’-diacetylchitobioside and MUF-β-d-N,N’,N’’-triacetylchitotrioside; Sigma Aldrich, UK), following the previously described method [42, 60, 61] (Additional file 1: Supplementary information). Standards curves were obtained using chitinase from Streptomyces griseus (Sigma Aldrich, UK) dissolved in sterile phosphate-buffered saline solution (pH 7.4; 0.137 M) with a highest concentration of 0.1 UmL-1 (activity equivalent to 144 μM day-1). Samples were diluted prior to measurement if they were expected to be above this range.
The process for artificial selection is depicted in Fig. 1. Briefly, 30 individual microcosms per treatment and generation were incubated in the dark under the conditions described below. At the end of each incubation period, the three microcosms with the highest chitinase activities (or three random microcosms in the case of the control) were pooled and used as the inoculum for the next generation of microcosms (n = 30). This was repeated across multiple transfers. Two artificial selection experiments were performed, the first to optimise the process and the second to implement optimal conditions and achieve a high-performing chitinolytic microbial community.
First artificial selection experiment
Incubations were carried out at 23 °C in 22 mL glass vials (Sigma Aldrich), each containing 20 mL of autoclaved seawater (collected from outside Plymouth Sound, Devon, UK; June 2016) supplemented with NaH2PO4, F/2 trace metals  (Additional file 1: Supplementary information) and 100 mg of chitin powder (from shrimp shells; Sigma Aldrich) as the sole source of carbon and nitrogen. Generation 0 was started with 200 μL of microbial inoculum. The efficiency of the selection process was assessed by comparing a ‘positive selection’ (where the three communities with highest activity were pooled and 200 μL was used to inoculate each one of the 30 microcosms of the next generation) against a ‘random selection’ (where three communities were chosen at random, using a random number generator within the Python module Random, to inoculate the following generation) to give a control against uncontrollable environmental variation . Each treatment was repeated across 20 generations with incubation times of 9 days. In parallel, treatments where incubation times were shortened to 4 days were set up after generation 15. Samples were taken from each community and stored in 20% glycerol at , 80 °C for further microbial isolation, and pellets from 1.5 ml of culture were collected by centrifugation (14,000 x g for 5 min) and stored at ,-20 °C for final DNA extraction and community analysis.
Second artificial selection experiment
A second selection experiment was set up implementing optimal transfer incubation times. Microcosms were incubated in 2 mL 96-well plates (ABgeneTM, ThermoFisher Scientific) covered by Corning® Breathable Sealing Tapes to stop evaporation and contamination while allowing gas exchange. Each well contained 1.9 mL of a custom mineral media containing MgSO4, CaCl2, KH2PO4, K2HPO4, 0.52 M NaCl and artificial seawater trace metals (Additional file 1: Supplementary information), supplemented with 10 mg of chitin powder. The microbial inoculum was 100 μL (i.e. initial inoculum and transfer between generations). Chitinase activity was measured daily. Transfer between generations was carried out just after the peak of chitinase activity had occurred, calculated as the mean chitinase activity across the 30 microcosms of the positive selection treatment. Plates were incubated in the dark at 30°C with constant shaking (150rpm). Eight days was the maximum incubation time allowed to reach maximum chitinase activity due to volume constraints. Pellets from 1.5 mL were collected by centrifugation (14,000 x g for 5 min) and stored at -20°C for final DNA extraction
DNA extraction and amplicon sequencing
DNA was extracted using the DNeasy Plant Mini Kit (Qiagen) protocol, with modifications as follows (adapted from ): 300 μL 1x TAE buffer was used to resuspend cell pellets and these were added to ~ 0.4 g of sterile 0.1 mm BioSpec Zirconia/Silica Beads in 2 mL screw cap microtubes (VWR international). Bead beating was carried out for 2 x 45 s and 1 x 30 s at 30 Hz using a Qiagen Tissue Lyser. Cell lysates were then processed in accordance with the manufacturer's instructions, with an extra centrifugation step to ensure all liquid was removed (1 min, 13,000 x g) directly before elution of samples. A Qubit® HS DNA kit (Life Technologies Corporation) was used for DNA quantification after which they were diluted to equalise the concentrations across samples. A Q5® Hot Start High-Fidelity 2X Master Mix (New England Biolabs® Inc.) was used to amplify the 16S rRNA gene v4-5 regions using primers 515F-Y and 926R , and the 18S rRNA gene v8, 9 regions using primers V8F and 1510R  (Additional file 1: Supplementary information). PCR products were purified using AmpliClean Magnetic Beads (NimaGen, The Netherlands). Index PCR was carried out using Illumina Nextera Index Kit v2 adapters. Samples were normalised using a SequelPrepTM Normalisation Plate Kit (ThermoFisher Scientific). Samples were pooled and 2 x 300 bp paired-end sequencing was carried out using the MiSeq system with v3 reagent kit. Negative DNA extraction controls and library preparation negative controls as well as chitin-only positive controls were processed and sequenced alongside samples.
Microbial community structure determination
Two different workflows were used to analyse the sequencing data: DADA2 [28, 29] and Mothur . DADA2 delivers better taxonomic resolution than other methods (e.g. Mothur) as it retains unique sequences and calculates sequencing error rates rather than clustering to 97% similarity . The resultant taxonomic units are referred to as amplicon sequence variants (ASVs) rather than operational taxonomic units (OTUs from Mothur). For the DADA2 analysis, sequencing data were processed following the DADA2 (version 1.8.0) pipeline . Briefly, the data were filtered, i.e. adapter, barcode and primer clipped, and the ends of sequences with high numbers of errors were trimmed. The amplicons were denoised based on a model of the sequencing errors and paired-end sequences were merged. Only sequences between 368, 379 for the 16S rRNA gene and 300, 340 for the 18S rRNA gene were kept and chimaeras were removed. The resulting ASVs were classified using the SILVA reference database (v132) . For the Mothur analysis , sequencing data were filtered, i.e. adapter, barcode and primer clipped, sequence length permitted was 450 bp for the 16S rRNA gene and 400 bp for the 18S rRNA gene, maximum number of ambiguous bases per sequence = 4 and maximum number of homopolymers per sequence = 8. Taxonomy assignment was performed using the SILVA reference database (Wang classification, v128)  and operational taxonomic units (OTUs) set at 97% similarity. For both processing workflows, chloroplasts, mitochondria and Mammalia were removed from the 16S rRNA gene and 18S rRNA gene datasets; eukaryotes were removed from the 16S rRNA gene dataset; and bacteria and archaea were removed from the 18S rRNA gene dataset. The average number of reads per sample was approximately 12,500 for the 16S rRNA gene and 20,000 (Mothur) or 34,000 (DADA2) for the 18S rRNA gene. Samples with less than 1000 total reads were excluded from downstream analyses. Although most analyses were carried out using relative abundance, each sample was subsampled at random to normalise the number of reads per sample, and the resulting average coverage was 92% (Mothur) or 94% (DADA2) for the 16S rRNA gene and 99% (Mothur and DADA2) for the 18S rRNA gene.
Microbial isolation and characterisation
Microbes were isolated from the final transfer of positive selection experiments by plating serial dilutions on Marine Broth 2216 (BD DifcoTM) and mineral medium plates (i.e. custom medium, Additional file 1: Supplementary information) supplemented with 0.1% N-acetyl-d-glucosamine (GlcNAc) and 1.5% agar. Colonies were re-streaked on fresh agar plates until pure isolates were obtained. The identification of isolates was carried out by sequencing the partial 16S rRNA gene (GATC BioTech, Germany) using primers 27F and 1492R  (Additional file 1: Supplementary information).
Isolates were grown in custom mineral medium supplemented with either 0.1% chitin or 0.1% GlcNAc (w/v), as sources of carbon and nitrogen, to test for chitinase activity and chitin assimilation, respectively. Growth was monitored over 14 days by measuring (i) chitinase activity (as described above), (ii) optical density at 600 nm and (iii) protein content (following the manufacturer's instructions; QuantiProTM BCA Assay Kit, Sigma Aldrich, UK). Isolates were also tested on custom mineral medium agar plates made with the addition of 0.1% chitin and 0.8% agarose. Plates were incubated at 30°C for 21 days to allow the formation of halos indicative of chitinase activity.
Co-cultures were performed using isolates Pseudoalteromonas shioyasakaiensis (chitin degrader), Donghicola eburneus (cheater, capable of growth on GlcNAc but not chitin) and Phaeobacter gallaeciensis (cross-feeder, not capable of growth on either chitin or GlcNAc). Combinations of these strains were grown in 25 cm2 tissue culture flasks with 25 mL custom mineral media (Additional file 1: Table S1) supplemented with 0.1% (w/v) of chitin. Cultures were incubated at 30 °C with shaking at 200 rpm for 3 days. Pellets from 1.5 ml of culture on days 0 and 3 were collected by centrifugation (14,000 x g for 5 min) and stored at , -20°C for DNA extraction, as above. Specific primers were designed for each of the isolates (see Additional file 1: Supplementary methods and materials), and qPCR was performed (Applied Biosystems 7500 Fast Real-Time PCR system) using 1 μL template DNA following the manufacturer's instructions for the GoTaq® qPCR Master Mix (Promega). Final primer concentrations were 0.5, 0.9 and 0.9 M, for the degrader, cheater and cross-feeder, respectively. Results were normalised to standard curves that used DNA extracted from pure cultures.
All analyses of chitinase activity and most MiSeq data analyses were carried out using custom Python scripts (Python versions 2.7.10 and 3.6.6) using the modules colorsys, csv, heapq, matplotlib, numpy, os, pandas, random, scipy, scikit-bio, sklearn  and statsmodels. SIMPER analyses and plotting of phylogenetic trees were performed in R (R version 3.3.3)  using the following packages: ape , dplyr, ggplot2, gplots, ggtree , lme4, phangorn , plotly, tidyr, vegan  and phyloseq . The top 5 ASVs identified in each SIMPER analyses were classified to their closest relative using a BLAST search of the GenBank database. Hypothetical community functions were obtained using PICRUSt in QIIME1 [31, 76] by mapping ASVs to the Greengenes database  (v13.5) at the default 97% similarity threshold. The PICRUSt analysis includes almost 35% of all ASVs, accounting for a mean relative abundance of 53%, 68% and 81% for the positive selection 9-day, 4-day and daily analyses, respectively. The Nearest Sequenced Taxon Index (NSTI) obtained for each of the taxonomic groups is available in Additional file 1: Table S3 (a full summary for each sample can be found on GitHub https://github.com/R-Wright-1/ChitinActivity). The three groups with the highest relative abundance, Gammaproteobacteria, Alphaproteobacteria and Bacteroidia (i.e. 29.3%, 26.4% and 16.6%, respectively) showed NSTI values of 0.07, 0.09 and 0.18, respectively. Sequences used for phylogenetic trees were aligned using the SILVA Incremental Alignment (www.arb-silva.de)  and mid-point rooted maximum likelihood trees were constructed using QIIME1 . All scripts can be found at https://github.com/R-Wright-1/ChitinActivity. All sequences have been deposited in the NCBI Short Read Archive (SRA) database under Bioproject PRJNA499076. qPCR data was analysed using custom Python scripts.
MiSeq amplicon sequencing was carried out by the University of Warwick Genomics Facility. We also thank Etienne Low-Decarie, for introducing us to the concept of microbial community artificial selection; Danielle Green, for collecting seawater from Portaferry, Sally Hilton, Michaela Mausz; the Warwick Genomics Facility for amplicon sequencing help and advice and Vinko Zadjelovic, Gabriel Erni Cassola, Audam Chhun; and the rest of the Christie-Oleza group, for insightful discussions and advice throughout the project. We thank SÃ©verine Rangama and Rui Abreu Pereira for help with the qPCR.
Robyn Wright was supported by a Midlands Integrative Biosciences Training Partnership PhD scholarship via grant BB/M01116X/1, Joseph Christie-Oleza by National Environment Research Council Independent Research Fellowship NE/K009044/1 and Matthew Gibson by European Research Council grant 638631.
RW and JCO designed the study. RW performed all the experiments with guidance from JCO and MIG. RW wrote the first draft of the manuscript, and all authors contributed substantially to revisions. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Johnson CR, Boerlijst MC. Selection at the level of the community: the importance of spatial structures. Trends Ecol Evol. 2002;17:83, 90.View ArticleGoogle Scholar
- Swenson W, Wilson DS, Elias R. Artificial ecosystem selection. Proc Natl Acad Sci. 2000;97:9110, 4.View ArticleGoogle Scholar
- Williams HTP, Lenton TM. Artificial selection of simulated microbial ecosystems. Proc Natl Acad Sci. 2007;104:8918, 23.View ArticleGoogle Scholar
- Goodnight CJ. Multilevel selection: the evolution of cooperation in non-kin groups. Popul Ecol. 2005;47:3, 12.View ArticleGoogle Scholar
- Goodnight CJ. Heritability at the ecosystem level. Proc Natl Acad Sci. 2000;97:9365, 6.View ArticleGoogle Scholar
- Swenson W, Arendt J, Wilson DS. Artificial selection of microbial ecosystems for 3-chloroaniline biodegradation. Environ Microbiol. 2000;2:564, 71.View ArticleGoogle Scholar
- Williams HTP, Lenton TM. Artificial ecosystem selection for evolutionary optimisation. In: Banzhaf W, Ziegler J, Christaller T, Dittrich P, Kim JT, editors. Adv Artif Life ECAL; 2007. p. 93, 102.View ArticleGoogle Scholar
- Blouin M, Karimi B, Mathieu J, Lerch TZ. Levels and limits in artificial selection of communities. Ecol Lett. 2015;18:1040–8.View ArticleGoogle Scholar
- Penn AS, Harvey I. The Role of Non-Genetic Change in the Heritability, Variation and Response to Selection of Artificially Selected Ecosystems.In Proceedings of the ninth international conference on artificial life. MIT Press. pp. 352–7.Google Scholar
- Datta MS, Sliwerska E, Gore J, Polz MF, Cordero OX. Microbial interactions lead to rapid micro-scale successions on model marine particles. Nat Commun. 2016;7:1–7.View ArticleGoogle Scholar
- Dang H, Lovell CR. Bacterial primary colonization and early succession on surfaces in marine waters as determined by amplified rRNA gene restriction analysis and sequence analysis of 16S rRNA genes. Appl Environ Microbiol. 2000;66:467, 75.View ArticleGoogle Scholar
- Ortiz-Álvarez R, Fierer N, de los Ríos A, Casamayor EO., Barberán A. Consistent changes in the taxonomic structure and functional attributes of bacterial communities during primary succession. ISME J. 2018;12:1658, 67.View ArticleGoogle Scholar
- Kiørboe T, Tang K, Grossart H, Ploug H. Dynamics of microbial communities on marine snow aggregates: colonization, growth, detachment, and grazing mortality of attached bacteria. Appl Environ Microbiol. 2003;69:3036, 47.View ArticleGoogle Scholar
- Simon M, Grossart H-P, Schweitzer B, Ploug H. Microbial ecology of organic aggregates in aquatic ecosystems. Aquat Microb Ecol. 2002;28:175, 211.View ArticleGoogle Scholar
- Kiørboe T, Grossart HP, Ploug H, Tang K. Mechanisms and rates of colonisation of sinking aggregates. Appl Environ Microbiol. 2002;68:3996, 4006.View ArticleGoogle Scholar
- Wu X, Wu L, Liu Y, Zhang P, Li Q, Zhou J, et al. Microbial interactions with dissolved organic matter drive carbon dynamics and community succession. Front Microbiol. 2018;9:1–12.View ArticleGoogle Scholar
- Beier S, Bertilsson S. Uncoupling of chitinase activity and uptake of hydrolysis products in freshwater bacterioplankton. Limnol Ocean. 2011;56:1179, 88.View ArticleGoogle Scholar
- Keyhani NO, Roseman S. Physiological aspects of chitin catabolism in marine bacteria. Biochim Biophys Acta 1473. 1999;1473:108, 22.Google Scholar
- Cordero OX, Ventouras L, Delong EF, Polz MF. Public good dynamics drive evolution of iron acquisition strategies in natural bacterioplankton populations. Proc Natl Acad Sci. 2012;109:20059, 64.View ArticleGoogle Scholar
- Christie-Oleza JA, Sousoni D, Lloyd M, Armengaud J, Scanlan DJ. Nutrient recycling facilitates long-term stability of marine microbial phototroph-heterotroph interactions. Nat Microbiol. 2017;2:1–10.View ArticleGoogle Scholar
- Morris JJ, Lenski RE, Zinser ER. The Black Queen Hypothesis: evolution of dependencies through adaptive gene loss. MBio. 2012;3:1–9.View ArticleGoogle Scholar
- Enke TN, Datta MS, Schwartzman J, Cermak J, Schmitz D, Barrere J, et al. Modular assembly of polysaccharide-degrading marine microbial communities. Curr Biol. 2019;29:1–8.View ArticleGoogle Scholar
- Dang H, Lovell CR. Microbial surface colonization and biofilm development in marine environments. Microbiol Mol Biol Rev. 2016;80:91-138.View ArticleGoogle Scholar
- Penn A. Modelling Artificial Ecosystem Selection: A Preliminary Investigation. In: Banzhaf W, Ziegler J, Christaller T, Dittrich P, Kim JT. (eds) Advances in Artificial Life. ECAL 2003. Lecture Notes in Computer Science. vol 2801. Berlin: Springer; 2003.Google Scholar
- Souza CP, Almeida BC, Colwell RR, Rivera ING. The importance of chitin in the marine environment. Mar Biotechnol. 2011;13:823, 30.View ArticleGoogle Scholar
- Beier S, Bertilsson S. Bacterial chitin degradation-mechanisms and ecophysiological strategies. Front Microbiol. 2013;4:1–12.View ArticleGoogle Scholar
- Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB, et al. Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Appl Environ Microbiol. 2009;75:7537, 41.View ArticleGoogle Scholar
- Callahan BJ, Sankaran K, Fukuyama JA, McMurdie PJ, Holmes SP. Bioconductor workflow for microbiome data analysis: from raw reads to community analyses. F1000Research. 2016;5:1492.View ArticleGoogle Scholar
- Callahan BJ, McMurdie PJ, Rosen MJ, Han AW, Johnson AJA, Holmes SP. DADA2: High-resolution sample inference from Illumina amplicon data. Nat Methods. 2016;13:581–3.View ArticleGoogle Scholar
- Hugerth LW, Andersson AF. Analysing microbial community composition through amplicon sequencing: from sampling to hypothesis testing. Front Microbiol. 2017;8:1–22.View ArticleGoogle Scholar
- Langille MGI, Zaneveld J, Caporaso JG, McDonald D, Knights D, Reyes JA, et al. Predictive functional profiling of microbial communities using 16S rRNA marker gene sequences. Nat Biotechnol. 2013;31:1814, 21.View ArticleGoogle Scholar
- Kanehisa M, Goto SKEGG. Kyoto Encyclopedia of Genes and Genomes. Nucleic Acids Res. 2000;28:27, 30.View ArticleGoogle Scholar
- Kanehisa M, Furumichi M, Tanabe M, Sato Y, Morishima K. KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Res. 2017;45:D353, 61.View ArticleGoogle Scholar
- Kanehisa M, Sato Y, Kawashima M, Furumichi M, Tanabe M. KEGG as a reference resource for gene and protein annotation. Nucleic Acids Res. 2016;44:D457, 62.View ArticleGoogle Scholar
- Enke TN, Leventhal GE, Metzger M, Saavedra JT, Cordero OX. Microscale ecology regulates particulate organic matter turnover in model marine microbial communities. Nat Commun. 2018;9:1–8.View ArticleGoogle Scholar
- Patterson D, Steinberg G, Nygaard K, Steinberg G. Heterotrophic flagellates and other protists associated with oceanic detritus throughout the water column in the mid north atlantic. J Mar Biol Assoc UK. 1993;73:67, 95.View ArticleGoogle Scholar
- Volterra V. Fluctuations in the abundance of a species considered mathematically. Nature. 1926;118:558, 60.View ArticleGoogle Scholar
- Lotka AJ. Elements of physical biology. Sci Prog Twent Century. 1926;21:341–3.Google Scholar
- Joint I, MÃ¼hling M, Querellou J. Culturing marine bacteria, an essential prerequisite for biodiscovery. Microb Biotechnol. 2010;3:564, 75.View ArticleGoogle Scholar
- Sutherland IW. Biofilm exopolysaccharides: a strong and sticky framework. Microbiology. 2001;147:3–9.View ArticleGoogle Scholar
- Das SN, Sarma PVSRN, Neeraja C, Malati N, Podile AR. Members of Gammaproteobacteria and Bacilli represent the culturable diversity of chitinolytic bacteria in chitin-enriched soils. World J Microbiol Biotechnol. 2010;26:1875, 81.View ArticleGoogle Scholar
- Lecleir GR, Buchan A, Maurer J, Moran MA, Hollibaugh JT. Comparison of chitinolytic enzymes from an alkaline, hypersaline lake and an estuary. Environ Microbiol. 2007;9:197, 205.View ArticleGoogle Scholar
- Kielak AM, Cretoiu MS, Semenov AV, SÃ¸rensen SJ, Van Elsas JD. Bacterial chitinolytic communities respond to chitin and pH alteration in soil. Appl Environ Microbiol. 2013;79:263, 72.View ArticleGoogle Scholar
- Fontanez KM, Eppley JM, Samo TJ, Karl DM, DeLong EF. Microbial community structure and function on sinking particles in the North Pacific Subtropical Gyre. Front Microbiol. 2015;6:1–14.View ArticleGoogle Scholar
- Aunkham A, Zahn M, Kesireddy A, Pothula KR, Schulte A, BaslÃ© A, et al. Structural basis for chitin acquisition by marine Vibrio species. Nat Commun. 2018;9:1, 13.View ArticleGoogle Scholar
- Giubergia S, Phippen C, Nielsen KF, Gram L. Growth on chitin impacts the transcriptome and metabolite profiles of antibiotic-producing Vibrio coralliilyticus S2052 and Photobacterium galatheae S2753. mSystems. 2017;2:1, 12.View ArticleGoogle Scholar
- Morris RM, RappÃ© MS, Connon SA, Vergin KL, Siebold WA, Carlson CA, et al. SAR11 clade dominates ocean surface bacterioplankton communities. Nature. 2002;420:806, 10.View ArticleGoogle Scholar
- Billerbeck S, Wemheuer B, Voget S, Poehlein A, Giebel HA, Brinkhoff T, et al. Biogeography and environmental genomics of the Roseobacter-affiliated pelagic CHAB-I-5 lineage. Nat Microbiol. 2016;1:1, 8.View ArticleGoogle Scholar
- Beygmoradi A, Homaei A. Marine chitinolytic enzymes, a biotechnological treasure hidden in the ocean? Appl Microbiol Biotechnol. 2018;102(23):9937–48.View ArticleGoogle Scholar
- Patil RS, Ghormade V, Deshpande MV. Chitinolytic enzymes: an exploration. Enzyme Microb Technol. 2000;26:473, 83.View ArticleGoogle Scholar
- Cottrell MT, Moore JA, David L, Kirchman DL. Chitinases from uncultured marine microorganisms. Appl Environ Microbiol. 1999;65:2553, 7.PubMedPubMed CentralGoogle Scholar
- Eckert EM, Baumgartner M, Huber IM, Pernthaler J. Grazing resistant freshwater bacteria profit from chitin and cell-wall-derived organic carbon. Environ Microbiol. 2013;15:2019, 30.View ArticleGoogle Scholar
- Nedoma J, Vrba J, Hejzlar J, Å imek K, StraÅ¡krabovÃ¡ V. N-acetylglucosamine dynamics in freshwater environments: concentration of amino sugars, extracellular enzyme activities, and microbial uptake. Limnol Oceanogr. 1994;39:1088, 100.View ArticleGoogle Scholar
- Riemann L, Azam F. Widespread N-acetyl-d-glucosamine uptake among pelagic marine bacteria and its ecological implications. Appl Environ Microbiol. 2002;68:5554, 62.View ArticleGoogle Scholar
- Ponomarova O, Patil KR. Metabolic interactions in microbial communities: untangling the Gordian knot. Curr Opin Microbiol. 2015;27:37, 44.View ArticleGoogle Scholar
- Zhang Z, Claessen D, Rozen DE. Understanding microbial divisions of labor. Front Microbiol. 2016;7:1, 8.Google Scholar
- Cavaliere M, Feng S, Soyer O, Jiminez J. Cooperation in microbial communities and their biotechnological applications. Environ Microbiol. 2017;19:2949, 63.View ArticleGoogle Scholar
- Delgado-Baquerizo M, Eldridge DJ, Ochoa V, Gozalo B, Singh BK, Maestre FT. Soil microbial communities drive the resistance of ecosystem multifunctionality to global change in drylands across the globe. Ecol Lett. 2017;20:1295, 305.View ArticleGoogle Scholar
- Hug LA, Co R. It Takes a village: microbial communities thrive through interactions and metabolic handoffs. mSystems. 2018;3:e00152, 17.PubMedPubMed CentralGoogle Scholar
- Hood MA. Comparison of four methods for measuring chitinase activity and the application of the 4-MUF assay in aquatic environments. J Microbiol Methods. 1991;13:151, 60.View ArticleGoogle Scholar
- Köllner KE, Carstens D, Keller E, Vazquez F, Schubert CJ, Zeyer J, et al. Bacterial chitin hydrolysis in two lakes with contrasting trophic statuses. Appl Environ Microbiol. 2012;78:695, 704.View ArticleGoogle Scholar
- Guillard RRL, Ryther JH. Studies of marine planktonic diatoms: I. cyclotella, Nana hustedt, and Detonula confervacea (cleve) gran. Can J Microbiol. 1962;8:229, 39.View ArticleGoogle Scholar
- Day MD, Beck D, Foster JA. Microbial communities as experimental units. Bioscience. 2011;61:398, 406.View ArticleGoogle Scholar
- Bryant JA, Clemente TM, Viviani DA, Fong AA, Thomas KA, Kemp P, et al. Diversity and activity of communities inhabiting plastic debris in the North Pacific Gyre. mSystems. 2016;1:e00024, 16.View ArticleGoogle Scholar
- Parada AE, Needham DM, Fuhrman JA. Every base matters: assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ Microbiol. 2016;18:1403, 14.View ArticleGoogle Scholar
- Bradley IM, Pinto AJ, Guest JS. Design and evaluation of Illumina MiSeq-compatible, 18S rRNA gene-specific primers for improved characterization of mixed phototrophic communities. Appl Environ Microbiol. 2016;82:5878, 91.View ArticleGoogle Scholar
- Quast C, Pruesse E, Yilmaz P, Gerken J, Schweer T, Glo FO, et al. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 2013;41:590, 6.View ArticleGoogle Scholar
- Lane DJ. 16S/23S rRNA sequencing. In: Stackebrandt E, Goodfellow M, editors. Nucleic Acid Tech Bact Syst; 1991. p. 115, 75.Google Scholar
- Pedregosa F, Weiss R, Brucher M. Scikit-learn: machine learning in Python. J Mach Learn Res. 2011;12:2825, 30.Google Scholar
- R Development Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing; 2008. http://www.R-project.org.
- Paradis E, Claude J, Strimmer K. APE: analyses of phylogenetics and evolution in R language. Bioinformatics. 2004;20:289, 90.View ArticleGoogle Scholar
- Yu G, Smith DK, Zhu H, Guan Y, Lam TT. GGTREE: an R package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol Evol. 2017;8:28, 36.View ArticleGoogle Scholar
- Schliep KP. phangorn: phylogenetic analysis in R. Bioinformatics. 2011;27:592, 3.View ArticleGoogle Scholar
- Dixon P. VEGAN, a package of R functions for community ecology. J Veg Sci. 2003;14:927, 30.View ArticleGoogle Scholar
- McMurdie PJ, Holmes S. phyloseq: An R Package for Reproducible Interactive Analysis and Graphics of Microbiome Census Data. PLoS ONE. 2013;8(4): e61217. https://doi.org/10.1371/journal.pone.0061217.View ArticleGoogle Scholar
- Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello EK, et al. QIIME allows analysis of high-throughput community sequencing data. Nat Methods. 2010;7:335, 6.View ArticleGoogle Scholar
- Desantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller K, et al. Greengenes, a chimera-checked 16S rRNA gene database and workbench compatible with ARB. Appl Environ Microbiol. 2006;72:5069, 72.View ArticleGoogle Scholar
- Pruesse E, Peplies J, Glöckner FO. SINA: accurate high-throughput multiple sequence alignment of ribosomal RNA genes. Bioinformatics. 2012;28:1823, 9.View ArticleGoogle Scholar