Stable-Isotope Probing Identifies Uncultured Planctomycetes as Primary Degraders of a Complex Heteropolysaccharide in Soil

Abstract

The exopolysaccharides (EPSs) produced by some bacteria are potential growth substrates for other bacteria in soil. We used stable-isotope probing (SIP) to identify aerobic soil bacteria that assimilated the cellulose produced by Gluconacetobacter xylinus or the EPS produced by Beijerinckia indica. The latter is a heteropolysaccharide comprised primarily of l-guluronic acid, d-glucose, and d-glycero-d-mannoheptose. ¹³C-labeled EPS and ¹³C-labeled cellulose were purified from bacterial cultures grown on [¹³C]glucose. Two soils were incubated with these substrates, and bacteria actively assimilating them were identified via pyrosequencing of 16S rRNA genes recovered from ¹³C-labeled DNA. Cellulose C was assimilated primarily by soil bacteria closely related (93 to 100% 16S rRNA gene sequence identities) to known cellulose-degrading bacteria. However, B. indica EPS was assimilated primarily by bacteria with low identities (80 to 95%) to known species, particularly by different members of the phylum Planctomycetes. In one incubation, members of the Planctomycetes made up >60% of all reads in the labeled DNA and were only distantly related (<85% identity) to any described species. Although it is impossible with SIP to completely distinguish primary polysaccharide hydrolyzers from bacteria growing on produced oligo- or monosaccharides, the predominance of Planctomycetes suggested that they were primary degraders of EPS. Other bacteria assimilating B. indica EPS included members of the Verrucomicrobia, candidate division OD1, and the Armatimonadetes. The results indicate that some uncultured bacteria in soils may be adapted to using complex heteropolysaccharides for growth and suggest that the use of these substrates may provide a means for culturing new species.

INTRODUCTION

Some major phyla of the domain Bacteria, like the Proteobacteria and Actinobacteria, are well-known from cultivation studies. Others, like Acidobacteria, Verrucomicrobia, and Planctomycetes, have few cultured species, although they make up large proportions of the bacterial 16S rRNA genes detected in environments such as soil (1). One theory proposed to explain the difficulty in culturing bacteria from these phyla is that most are K selected as opposed to r selected. Typical r-selected bacteria grow rapidly on simple monomeric substrates in nutrient-rich environments, conditions presented by most microbiological media (2, 3). On the other hand, K-selected bacteria have a more efficient cell metabolism and strong competitive ability and grow slowly on recalcitrant, complex substances (3–5). Several studies have demonstrated that members of difficult-to-culture phyla like Acidobacteria, Verrucomicrobia, and Planctomycetes can be cultivated using polysaccharides, such as xylan, xanthan, and pectin, as energy sources (e.g., see references 6 to 8). The heteropolysaccharide gellan, which is often used as a gelling agent as an alternative to agar, can also act as an energy source for some soil bacteria (9), in part explaining the observation that it improves culturability (6, 7, 10).

Most polysaccharides present in soils are plant compounds, like cellulose and hemicellulose. Cellulose is a simple polysaccharide of glucose, but hemicellulose is a complex mixture of different heteropolysaccharides, such as xylan. Besides plants, other sources of heteropolysaccharides in soil are bacteria. Bacteria produce exopolysaccharides (EPSs) to serve a variety of functions, including surface attachment, biofilm formation, desiccation resistance, and cryoprotection. EPSs may also serve as an energy reserve or as a diffusive barrier to harmful chemicals (11, 12). These substances have received scientific interest primarily due to their biotechnological applications (13). However, they are ecologically important in soils as well, for example, by contributing to soil aggregate structure and water retention (14, 15). One key function proposed for EPS is as an energy reserve, although because it is extracellular, an EPS can be grazed on by species other than the one that produces it. For example, Koch et al. (16) isolated an Acidobacteria species in coculture with the EPS-producing methanotroph Methylocella silvestris, a close relative of Beijerinckia indica, and suggested that the EPS supported the growth of the satellite bacterium. There is a vast range of chemically different bacterial EPS molecules (17), suggesting that these may form a diverse food source in microbially rich environments, such as soil.

Beijerinckia indica subsp. indica (ATCC 9039) is known for its copious production of EPS (18). Most of the EPS produced (85 to 90%) is an acidic heteropolysaccharide comprised of repeating subunits of [l-guluronic acid, β(1→3) d-glucose, and β(1→4) d-glycero-d-mannoheptose β(1→4)] (19, 20). The remaining, neutral fraction of the EPS (10 to 15%) is a mixture of d-glucose, d-mannose, l-arabinose, and l-rhamnose (19). Similar EPS compositions have been verified in related species of Beijerinckia, although the polysaccharide can be branched rather than linear (21).

Several previous studies have used ¹³C-labeled cellulose as a stable-isotope probing (SIP) substrate to identify cellulose-degrading bacteria in soil (22, 23). ¹³C-labeled cellulose can be produced by feeding [¹³C]glucose to Gluconacetobacter xylinus and harvesting the [¹³C]cellulose produced as an exopolysaccharide. However, to our knowledge, SIP has not been performed with other bacterial EPS molecules. Although these substrates are quantitatively much less important in soils than cellulose, they may form a diverse and heterogeneous food source for some members of the soil community. A preference for rare and complex substrates, such as bacterial EPS, may explain some of the vast uncultured microbial diversity in soils. Because of the copious production of EPS by B. indica, combined with its unusual sugar components, we used it in SIP experiments to examine which soil bacteria were capable of growing on a model EPS heteropolysaccharide.

MATERIALS AND METHODS

Soils.

Two sample soils were selected. One sample was near the Paint Pots (PP) in British Columbia, Canada (51°10.237′N, 116°09.527′W) (24). A spring at this site flows though sulfide mineral deposits and creates acidic soils in a process analogous to the acid mine drainage process (24). Soil was taken from the surface (0 to 5 cm) layer of an unvegetated extinct spring mound. This was a true oxic soil site well distant from the present spring. Although the soil is adjacent (5 m) to a mixed forest, it remains unvegetated due to the high acidity. This soil was chosen because the lack of vegetation was expected to limit the bacterial species adapted to grow on plant organic matter and instead favor other metabolisms. The soil pH, measured 1:1 in water, was 4.2, and the loss on ignition (550°C) was 13.3%. The second soil sample was taken from just outside Big Hill Springs Provincial Park (BHS), Alberta, Canada, located northwest of Calgary, Alberta, Canada (51°15.013′N, 114°23.156′W). The parent rock material here is primarily calcareous sandstones and carbonate-dominated glacial tills, resulting in a more neutral pH, measured to be 5.6. Soil was taken from the surface (0 to 10 cm) A horizon of a mixed forest under a poplar (Populus balsamifera). Loss on ignition was 15.4%. Soils were sampled with clean spades and placed into sterile polypropylene containers, which were sealed and stored at room temperature before the experiments began.

[¹³C]cellulose and [¹³C]EPS production.

Gluconacetobacter xylinus (ATCC 53524) was grown in buffered S & H medium (25) containing 2% (wt/vol) fully ¹³C-labeled glucose as the sole carbon source (minimum, 99%; LC Scientific Inc.). Aerobic incubations lasted for up to 1 week at 30°C under static conditions. Bacterial [¹³C]cellulose was extracted from G. xylinus as described previously (25) via boiling and two baths in a 4% NaOH solution with subsequent washing. This procedure produces cellulose with 99% purity (22). Purified cellulose pellicles were placed into 2-ml tubes with a ceramic bead and milled in a Bertin Technologies Precellys cell homogenizer for 3 cycles of 30 s each at a speed setting of 6.0 ms⁻¹. Purified cellulose was autoclaved and then freeze-dried at −80°C until use.

B. indica was grown on nitrogen-free mineral medium (26) containing 1% (wt/vol) fully ¹³C-labeled glucose as the sole carbon source. Incubations were made aerobically for up to 1 week at 22°C. The extraction and purification of EPS were modified from a strategy described by Liu and Fang (27). EPS and cells (∼15 ml) from plates of B. indica were collected and placed into 50-ml Falcon tubes, and an equal volume of sterile 10% NaCl solution was added. The tube was vortexed and centrifuged at 20,000 × g at 4°C for 30 min. The supernatant was then filtered through a 0.2-μm-pore-size membrane (VWR International Inc.) at 22°C to remove the cells, and the solution was purified with a 3,500-Da dialysis membrane (Spectrum Laboratories Inc.) at 4°C for 48 h to remove monosaccharides. The purified EPS solution was frozen at −80°C for 24 h and then freeze-dried at −80°C for 72 h. The DNA concentration and protein concentration of the freeze-dried EPS were determined on a Qubit fluorometer using a Quant-iT double-stranded DNA (dsDNA) high-sensitivity (HS) assay kit (Invitrogen) and a Quant-iT protein assay kit (Invitrogen) to determine the EPS purity. The purifications contained only 0.04% protein and 0.00074% DNA.

SIP incubation.

Five-gram (wet weight) amounts of soil from each sample (BHS and PP soils) plus 0.5% (wt/wt) cellulose or B. indica EPS (¹³C-labeled or ¹²C-labeled controls) were distributed into 120-ml serum bottles, which were sealed and made gas tight with butyl rubber stoppers. The soil was moistened with 1 ml of sterile deionized water per bottle. All samples were set up in triplicate and incubated at 22°C. Ten percent (vol/vol) ¹²CO₂ was added to the headspace of the [¹³C]cellulose or [¹³C]EPS incubation bottles to reduce cross feeding of other bacteria with ¹³CO₂. We felt that this was necessary primarily for the PP soil, which potentially has high populations of autotrophs (24). Vials with ¹³C-labeled substrates added were also uncapped and aired every 30 days in order to maintain oxic conditions and further prevent ¹³CO₂ cross feeding. The headspace CO₂ mixing ratios of soil vials incubated with either no added substrate, [¹²C]cellulose, or [¹²C]EPS (all without added CO₂) were monitored over time via gas chromatography (GC) using a thermal conductivity detector (28). The production of excess detectable CO₂ in the substrate-amended versus unamended soils was taken as evidence for mineralization of the added substrate, and soils were processed for SIP soon after CO₂ production from either substrate (G. xylinus cellulose and B. indica EPS) was evident in at least one of the two soil samples. Cellulose incubations were stopped after 18 to 30 days, and B. indica EPS incubations were stopped after 80 to 111 days.

Molecular community analyses.

DNA extraction, gradient fractionations in cesium chloride (CsCl) solution, PCR, and pyrosequencing of 16S rRNA genes were performed as described previously (28). Briefly, DNA was extracted from about 500 mg of soil before and after incubation with ¹³C- or ¹²C-labeled substrates, using a FastDNA spin kit for soil (MP Biomedicals). Additional washes using 5.5 M guanidine thiocyanate removed excess humic acids from DNA extracts. DNA yields were about 2.8 and 0.1 μg DNA per g soil for the BHS and PP soils, respectively. DNA was combined with CsCl (1.87 g ml⁻¹) and gradient buffer in ultracentrifugation tubes (13 by 51 mm). Ultracentrifugation was done at 299,065 × g in an NVT90 rotor (Optima L-100K; Beckman Coulter Inc.) at 20°C with a vacuum over 65 h. DNA was retrieved by gradient fractionation, resulting in 12 fractions of approximately 400 μl each. The density of each fraction was measured with a refractometer (AR200; Reichert) to confirm gradient formation. DNA was precipitated from the CsCl with polyethylene glycol and glycogen, washed with 70% ethanol, and eluted in 30 μl of Tris-EDTA buffer. The DNA concentration of each fraction was determined with a Qubit fluorometer using a Quant-iT dsDNA HS assay kit (Invitrogen).

The ¹²C-labeled substrate incubations were used as controls to determine the expected position of unlabeled soil DNA in the CsCl density gradients. After comparison of the CsCl gradients of soils incubated with either ¹³C- or ¹²C-labeled substrate, the 16S rRNA gene was targeted in selected fractions via PCR using FLX titanium amplicon primers 454T_RA_X and 454T_F containing the target primers 926fw and 1392r at their 3′ ends, respectively, along with the adaptors necessary for the Roche titanium chemistry. These primers target 16S rRNA genes from all three domains of life (28). Each reverse primer contained a unique 10-nucleotide identifier bar code sequence that allowed sequences to be binned according to the particular sample. The PCR mixtures and PCR amplification profile were exactly as described previously (28). Purified PCR products (∼150 ng total DNA) were sequenced on a Roche 454 (Branford, CT, USA) genome sequencer FLX instrument using titanium chemistry.

Enrichment of Planctomycetes.

In order to determine if Planctomycetes could be grown on laboratory medium containing B. indica EPS, the PP soil enrichment with B. indica EPS was allowed to incubate for a total of 18 months, and then soil slurry was sampled with a syringe and dripped onto the plate of DNMS mineral salts medium (29) that was adjusted to pH 3.8, supplemented with 0.05% (wt/vol) EPS as the sole carbon and energy source, and solidified using 15 g liter⁻¹ of gellan gum. These plates were incubated for 7 weeks under an atmosphere of 5% CO₂ in air and observed under a stereomicroscope. Colonies with the predominant morphologies (10 to 20 colonies of a particular morphology) were selected with a 22-gauge needle and transferred to a DNA extraction tube, and the microbial community was analyzed as described above.

Bioinformatics.

QIIME (version 1.8) (30) was used to analyze the sequenced 16S rRNA genes. Quality filtering and processing were performed exactly as described previously (28). A minimum quality score of 25 was used in the quality filtering step. All pyrosequencing runs for all experiments were run simultaneously in a single analysis, so operational taxonomic units (OTUs) could contain reads from across different samples. OTUs based on 97% sequence similarity were clustered, and representative sequences of predominant OTUs were analyzed by comparison of the sequences with those in the SILVA 111 reference database by use of the BLAST program (31). Eukaryotic sequences and chloroplasts made up less than 0.1% of all reads and were removed from the analysis. Phylogenetic trees were constructed using the ARB software environment (32).

Nucleotide sequence accession number.

The 16S rRNA sequences obtained in this study have been deposited in the Sequence Read Archive (SRA) database under accession number SRP056796.

RESULTS

Bacterial phyla in sample soils.

The BHS site was dominated by Actinobacteria (30.7%) and Proteobacteria (21.4%), with smaller amounts of Acidobacteria (11.3%), Chloroflexi (3.7%), and Planctomycetes (3.6%) being found (see Fig. S1 in the supplemental material). The Paint Pots soil was predominated by bacteria belonging to the Chloroflexi (25.4%), candidate division WPS2 (24.1%), and Proteobacteria (21.4%), with smaller proportions of Acidobacteria (11.8%), Planctomycetes (6.9%), and Actinobacteria (7.5%) being found (see Fig. S1 in the supplemental material).

Soil incubations.

GC monitoring of O₂ verified that the vials did not ever become anoxic during the incubations (vials containing ¹³C-labeled substrates were aired every 30 days). The carbon dioxide in the headspace of the soil incubation vials increased over time, with or without added substrates (see Fig. S2 in the supplemental material). The BHS soil plus cellulose treatment showed a much higher mean CO₂ production rate than the control soil as early as 9 days into the incubation. This treatment was therefore the first harvested, at 18 days. The total excess carbon produced as CO₂ compared to the amount of carbon produced by the controls after 18 days was 25% of the EPS carbon added. In order to have roughly comparable time courses in both soils for cellulose degradation, the PP soil plus [¹³C]cellulose treatment was harvested soon afterwards, at 30 days, despite no excess CO₂ being detected at this time compared to the amount of CO₂ in the substrate-unamended controls (see Fig. S2 in the supplemental material). The lower level of CO₂ production in this soil may indicate limited cellulose degradation compared to normal soil respiration. Nevertheless, SIP requires very little carbon assimilation (as little as 65 μg ¹³C per g soil) (33), and therefore, strong degradation of the ¹³C substrate (added at 5 mg per g soil) to CO₂ is not necessarily required to identify the active microorganisms.

The PP soil plus B. indica EPS treatment also showed a slightly increased CO₂ production rate compared to that for the control group, but this was detectable only in the PP soil and only after 60 to 80 days. Both soils to which B. indica EPS was added were therefore harvested after 80 days and again after 111 days to have more data for this substrate.

SIP results for cellulose.

DNA extracted from BHS and PP soils incubated in the presence of [¹³C]cellulose showed small shifts in the DNA density profiles compared to those for the control DNA obtained from soils incubated with [¹²C]cellulose (see Fig. S3 in the supplemental material). Heavy fractions of the DNA from soils incubated with [¹³C]cellulose (indicated in Fig. S3 in the supplemental material) were selected by comparison to the DNA in the corresponding [¹²C]cellulose controls. The phyla detected via 16S rRNA gene pyrosequencing in the heavy fractions of both soils after [¹³C]cellulose incubation were distinct from the communities detected in native soils. In the PP soil incubation, there was an increase primarily in Actinobacteria (from 7.5% to 60.5%) and Proteobacteria (from 16.4% to 33.1%). In the BHS soil incubation, there was an increase primarily in Proteobacteria (from 21.4% to 46.3%) (see Fig. S1 in the supplemental material). The most abundant OTUs, representing >1% of all reads and enriched in heavy DNA at least 25-fold compared to the amount in the original soil community, are listed in Table 1. Most of the OTUs highly enriched in the heavy DNA fractions were closely related (16S rRNA gene sequence identities, 93 to 100%) to known species of Proteobacteria or Actinobacteria. A strain 99% identical to the actinobacterium Catenulispora acidiphila made up >50% of all reads in the heavy DNA of the PP soil, and a strain 98% identical to the proteobacteria Caulobacter henricii and Caulobacter segnis was the most abundant species enriched in the BHS soil.

TABLE 1.

Relative abundance of highly enriched OTUs in heavy fractions from [¹³C]cellulose SIP experiments with PP and BHS soils^a

Soil sample and OTU	Relative % abundance in:		Fold enrichment (% abundance in heavy DNA/original soil)	Taxonomy (phylum, class)	Closest relative by BLAST analysis	% identity to the closest relative
	Heavy DNA fraction	Original soil
PP soil
8231	46.5	0.47	856	Actinobacteria, Actinobacteria	Catenulispora acidiphila	99
7718	20.7	UD	>1,900	Proteobacteria, Alphaproteobacteria	Telmatospirillum siberiense	94
10085	10.2	UD	>940	Actinobacteria, Actinobacteria	Catenulispora yoronensis	98
7534	7.47	0.21	34	Proteobacteria, Alphaproteobacteria	Bradyrhizobium elkanii	100
BHS soil
12963	11.9	UD	>1,110	Proteobacteria, Alphaproteobacteria	Caulobacter henricii	98
1448	3.09	UD	>280	Proteobacteria, Deltaproteobacteria	Phaselicystis flava	93
11150	2.84	0.12	24	Proteobacteria, Betaproteobacteria	Aquincola tertiaricarbonis	99
5445	2.48	0.098	26	Actinobacteria, Actinobacteria	Catenuloplanes crispus	100
859	1.50	UD	>138	Bacteroidetes, Cytophagia	Sporocytophaga myxococcoides	99
3553	1.28	0.033	39	Proteobacteria, Alphaproteobacteria	Brevundimonas denitrificans	100
10786	1.18	0	>108	Proteobacteria, Alphaproteobacteria	Phaeospirillum tilakii	95
6220	1.04	0.011	95	Proteobacteria, Gammaproteobacteria	Lysobacter niabensis	98

^aOnly OTUs with >1% relative abundance and enriched at least 25-fold compared to the amount in unincubated soil are shown. On the basis of a Z test of ratios, all significance values (chance of sampling these differences from the same population) are <0.001. UD, no reads were detected in the original soil; the minimum fold enrichment in these cases was calculated by assuming 1 read in the original soil.

SIP results for B. indica EPS.

DNA extracted from soils incubated with B. indica [¹³C]EPS showed pronounced shifts in density compared to DNA extracted from soils incubated with B. indica [¹²C]EPS (Fig. 1). The production of a distinct tail of heavy DNA was evident in all analyses, including duplicate runs of each soil sample at 80 days (see Fig. S4 in the supplemental material). At the phylum level, the heavy DNA fractions showed increases primarily in the relative abundance of Planctomycetes compared to that in the native soil. Planctomycetes increased from to 3.7 to 15.4% in the BHS soil community incubated with [¹³C]EPS and from 6.8% to between 14.3% and 68.2% in the two PP soil communities incubated with [¹³C]EPS (see Fig. S1 in the supplemental material). Although the PP soil initially had a large proportion of Planctomycetes (6.8%), the predominant OTUs in the original soils were not the ones enriched in the SIP experiments (see Table S1 in the supplemental material). The most abundant individual OTUs (>1% of all reads) enriched at least 25-fold in the heavy DNA are listed in Table 2, and their phylogenetic positions are noted in Fig. 2. A more complete list of all OTUs that were enriched at least 20-fold and that made up at least 0.2% of all reads is shown in Table S2 in the supplemental material. Although rarer OTUs may represent pyrosequencing and/or computational clustering errors (see Table S2 in the supplemental material), many of the OTUs were both abundant and very divergent from each other (Fig. 2), indicating that they were not artifacts.

FIG 1.

Relative DNA concentrations in CsCl gradient fractions of DNA extracted from B. indica EPS SIP experiments with BHS soil (80 days) (A) and PP soil (80 days [B] and 111 days [C]). For intuitive simplicity, the graphs are drawn with density on the y axis in the same orientation that one observes in the gradient tubes. Solid lines, incubations with [¹²C]EPS (control); dashed lines, incubations with [¹³C]EPS. The x axis indicates the relative DNA concentrations recovered from each gradient fraction, with the highest quantity detected in any gradient fraction being set equal to 1.0. The arrows indicate the representative heavy fractions that were used to fingerprint the active EPS-assimilating community.

TABLE 2.

Relative abundance of highly enriched OTUs detected in heavy-density fractions of SIP experiments of PP and BHS soils incubated with B. indica [¹³C]EPS^d

Soil sample, time of collection, and OTU	Relative % abundance in:		Fold enrichment (% abundance in heavy DNA/original soil)	Taxonomy (phylum, class)	Closest relative by BLAST analysis	% identity to the closest relative
	Heavy DNA fraction	Original soil
PP soil, 80 days
1577a	62.0	0.054	1,148	Planctomycetes, Phycisphaerae	Schlesneria paludicola	85
12292	7.08	UD	>488	Candidate division OD1		<80
9612	2.78	0.011	254	MLE1-12	Microcystis ichthyoblabe (Cyanobacteria)	87
3848	1.40	UD	>96	Armatimonadetes	Frankia alni (Actinobacteria)	87
7541	1.08	0.043	25	Planctomycetes, Planctomycetacia	Singulisphaera acidiphila	95
PP soil, 111 days
12201	5.74	0.011	434	Planctomycetes, Planctomycetacia	Aquisphaera giovannonii	93
11666b	3.52	UD	>320	Planctomycetes, Planctomycetacia	Aquisphaera giovannonii	91
BHS soil, 80 days
2613	7.18	0.010	658	Planctomycetes, Phycisphaerae	Schlesneria paludicola	84
5341	2.57	UD	>236	Verrucomicrobia, Spartobacteria	Roseimicrobium gellanilyticum	88
4249c	2.46	UD	>223	Planctomycetes, Phycisphaerae	Schlesneria paludicola	84

^aAlso includes the closely related OTUs 11375, 12969, 505, 4724, 10597, and 3328 (see Table S2 in the supplemental material).

^bAlso includes the closely related OTUs 1598, 3075, and 778 (see Table S2 in the supplemental material).

^cAlso includes the closely related OTU 10087 (see Table S2 in the supplemental material).

^dOnly OTUs with >1% relative abundance in the heavy fraction (generally >100 reads) and enriched at least 25-fold compared to the amount in unincubated soil are shown. On the basis of a Z test of ratios, all significance values (chance of sampling these differences from the same population) are <0.001. UD, no reads were detected in the original soil; the minimum fold enrichment in these cases was calculated by assuming 1 read in the original soil.

FIG 2.

Maximum-likelihood (RxaML) 16S rRNA gene sequence-based tree showing the phylogenetic positions of the most enriched B. indica EPS-degrading bacteria identified in BHS and PP soils (designated in bold as the respective OTU numbers). Bootstrap support values (100 constructions) are shown when they were >50%. The sequence alignments were from the ARB-SILVA database, and calculations were made using ARB software (32). Bar, 0.1 change per nucleotide position.

Unlike the OTUs enriched in the [¹³C]cellulose enrichment, most of the OTUs enriched in the B. indica [¹³C]EPS enrichment were not closely related to cultured bacteria, showing only 80 to 95% 16S rRNA gene sequence identities to validated species (Table 2). Besides the predominant Planctomycetes, highly enriched OTUs also included members of the Verrucomicrobia, Armatimonadetes, candidate phylum OD1, and candidate phylum MLE1-12 (related to the Cyanobacteria). Most represented deeply rooted uncultured lineages of these phyla (Fig. 2).

Although OTUs of the phylum Planctomycetes were predominant in all three B. indica [¹³C]EPS SIPs that were analyzed via sequencing of 16S rRNA genes (one with BHS soil and two with PP soil), the most predominant OTUs varied across the three analyses. In the 80-day incubations of both soils, the predominant OTUs belonged to the class Phycisphaerae, although the individual OTUs were different in each case. On the other hand, at 111 days the PP soil sample was predominated by members of the class Planctomycetacia. In all three cases, multiple OTUs were found to be enriched in the heavy DNA, and in one case (PP soil at 80 days) these included members of both the Phycisphaerae and Planctomycetacia classes. The ability to grow on B. indica EPS appears to be shared by diverse Planctomycetes species.

SIP controls.

For the visualization of DNA gradients in CsCl, samples incubated with the ¹³C substrate were always compared with samples incubated with the ¹²C substrate in order to precisely localize the heavy DNA fractions (Fig. 1). Heavy fractions were chosen where there was little detectable DNA in the control soil fractions (Fig. 1), and PCR products could not be obtained from the control fractions. This clearly shows that the heavy DNA fractions in the ¹³C incubations must have originated from the added labeled substrate, and the communities represented by these heavy DNA fractions represent ¹³C-assimilating bacteria. However, although the ¹²C-labeled controls were vital in localizing the heavy DNA fractions, we chose to use the original soil DNA rather than the incubations with ¹²C as comparative controls when calculating enrichment factors for individual OTUs. ¹²C-labeled substrate incubations are valid controls in a SIP experiment only when the SIP-targeted populations are rare compared to their level in the overall soil community (i.e., when there is a high DNA background) and a small enrichment of these bacteria will not bias the overall community. In our soils, especially the PP soil, the initial bacterial community was not abundant, and therefore, any growth in response to added substrate could cause a large bias.

Several other control analyses were also performed to verify the SIP results for the B. indica EPS experiment. These included (i) confirmation that there was a very minor effect of the bar codes used on the PCR primers, (ii) comparison of OTU abundances in the heavy DNA fractions with their abundances in the light DNA fractions of the incubations with ¹³C to verify that the pattern obtained was the same as that obtained by comparisons of the heavy fractions to the complete DNA of the original soil sample, (iii) verification that the changes in the unlabeled soil community during incubation with ¹³C-labeled substrates were minor, and (iv) examination of the degree to which the same OTUs detected in the SIP experiments could also be found directly enriched in the incubations with ¹²C without the aid of SIP. These controls are described in detail in the “Additional SIP controls” section in the supplemental material.

Enrichment of Planctomycetes.

The PP soil enrichments described in the SIP experiments were allowed to incubate for a total of 18 months and then transferred to plates of mineral salts medium supplemented with 0.05% (wt/vol) EPS as the sole carbon and energy source. After 7 weeks of incubation, small colonies (diameter, <0.5 mm) were observed to be growing on the soil. The predominant colony type was raised, slightly pink, and weakly viscous (see Fig. S5 in the supplemental material). Several of these colonies were pooled, the DNA was extracted, and community analysis was performed as described above. The sequencing products were predominated (>70 to 90% of reads) by OTU 12201 (Table 2), representing a Planctomycetes with 93% identity to Aquisphaera giovannonii. This finding verified that a predominant OTU identified in the SIP study could also be grown on solid medium with B. indica EPS as the sole energy source.

DISCUSSION

There was some overlap between the bacterial taxa identified to be potential cellulose degraders in this study and those identified in previous research examining cellulolytic communities in various soils via cellulose enrichment (34) or [¹³C]cellulose SIP (22, 23, 35–37). Previous studies also commonly identified Caulobacteraceae to be potentially cellulolytic (Caulobacter and Brevundimonas in Table 1), and in some cases, other families containing bacteria noted in Table 1, notably, Bradyrhizobiaceae, Xanthomonadaceae, and Rhodospirillaceae, were found to be potentially cellulolytic. There is considerable variation in the bacteria detected across these experiments, which is to be expected due to soil and climate differences. Most similar to our results, Verastegui et al. (37) noted a predominance of Proteobacteria (primarily Caulobacteraceae) and Actinobacteria in the active communities detected with cellulose SIP in three temperate to subarctic soils in Canada.

Most of the OTUs enriched in the heavy DNA (Table 1) of the cellulose SIP experiments were closely related to cultured, taxonomically validated species. The predominant bacterium in the heavy DNA fraction of BHS soil showed 98% identity to Caulobacter henricii. Caulobacteraceae are particularly predominant in cellulose enrichment and SIP experiments (23, 34, 36, 37), and although the cellulolytic ability of Caulobacter henricii appears to be untested, closely related species of Caulobacter can be grown on crystalline cellulose or gluco-oligosaccharides (38, 39). The Caulobacter crescentus genome possesses genes encoding cellulases, xylanases, and xylosidases (40). The bacterium most enriched in the heavy DNA of the [¹³C]cellulose experiment in PP soil showed 99% identity to the actinobacterium Catenulispora acidiphila. Although this organism was reported to be incapable of growth on cellulose (41), the complete genome sequence (42) contains two predicted endoglucanases of cellulase families 6 and 8, as well as exo-1,4-β-glucosidase for degradation of cellobiose units. It seems likely that this species does have cellulolytic activity, which was perhaps not measurable under the culture conditions used in the original taxonomic description. The predominance of this OTU in the active fraction of the PP soil enrichment (53% of all reads) suggests that among the bacteria in the PP soil, it probably plays a fundamental role in cellulose degradation.

The predictability of the cellulose SIP results on the basis of previous study and characterized cultures serves to verify our SIP procedures to some extent. This is a useful control when examining the B. indica EPS results, because in contrast to cellulose, this substance appeared to be assimilated primarily by deeply rooted lineages of uncultured bacteria. The highest similarity of any predominant OTU detected in the heavy DNA fractions of the SIP experiments using B. indica EPS was a species of Planctomycetes somewhat related (95% 16S rRNA sequence identity over the sequenced region) to the genus Singulisphaera originally described by Kulichevskaya et al. (43). Singulisphaera acidiphila is able to hydrolyze various polysaccharides, including laminarin, pectin, chondroitin sulfate, esculin, gelatin, pullulan, lichenan, and xylan (43). The strong enrichment of this bacterium also helps to validate the SIP procedure, indicating that an OTU related to a known heteropolysaccharide degrader was enriched. However, all other OTUs predominant in the heavy DNA fractions of the B. indica EPS SIP were more distantly related to cultured species.

Members of the phylum Planctomycetes were predominant in all three B. indica EPS SIP analyses (twice with PP soil and once with BHS soil), although the individual OTUs differed. There are only a few cultured species of Planctomycetes available, and knowledge of the overall diversity of this phylum is therefore limited. The phylum has two validated classes: the Planctomycetacia, which contains 11 genera, and the Phycisphaerae, formerly known as the WPS1 group (44, 45), which contains the single genus and species Phycisphaera mikurensis (46). In addition, some “Candidatus Planctomycetes” species are anaerobic ammonium-oxidizing (anammox) bacteria consisting of lithoautotrophs that reduce nitrite and oxidize ammonium to produce dinitrogen gas under anoxic conditions (45). Aside from the anammox bacteria, most cultured Planctomycetes, including Phycisphaera mikurensis, the only cultured representative of the Phycisphaerae, are organoheterotrophs. Few polymeric growth substrates have been tested on this organism, although it can grow on cellobiose and several monosaccharides (46). Members of the Phycisphaerae have been found to be associated with polymeric substrates, such as decaying wood (47). Many members of the better-known Planctomycetacia class are able to grow on heteropolysaccharides. Rhodopirellula baltica is one of the best-studied Planctomycetes. On the basis of its genome sequence, it can degrade complex polysaccharides such as fucoidan and is hypothesized to live primarily via the decomposition of complex sulfated heteropolysaccharides produced by algae (48, 49). Planctomycetacia isolated from Sphagnum peat, including species of Schlesneria, Singulisphaera, and Zavarzinella, are also able to degrade various heteropolysaccharides, although not the homopolysaccharides cellulose and chitin (50).

This body of knowledge suggests that growth on complex heteropolysaccharides and the oligo- and monosaccharides produced from their hydrolysis are major ecological niches for Planctomycetes in nature, and our results provide support for this generalization. In the samples obtained from both EPS SIP experiments at 80 days, the predominant OTUs in the heavy DNA fractions belonged to the class Phycisphaerae. The level of one OTU (OTU 1577) was increased over 1,000-fold in the heavy DNA fraction of the PP soil sample, comprising over 60% of all reads. Members of the class Planctomycetacia were also enriched in the PP soil sample at 80 days, and the most predominant OTUs in the PP soil sample at 111 days were Planctomycetacia. The divergence of the active bacteria in PP versus BHS soil samples is expected due to differences in soil chemistry, but the reason behind the divergence between the two PP soil samples taken at different times is more difficult to pin down. It could be a combination of changing primary or secondary consumers of the EPS over time (80 days versus 111 days) and stochastic factors related to the low initial abundances of the EPS-active bacteria in the soil (since the samples obtained at the two time points were from different microcosms). In any case, a predominance of B. indica EPS-assimilating Planctomycetes was a common feature of all three analyses (BHS soil at 80 days, PP soil at 80 days, and PP soil at 111 days). It seems that the ability to degrade this heteropolysaccharide is shared by diverse Planctomycetes in the classes Planctomycetacia and Phycisphaerae.

In long SIP incubations, there is always the possibility of cross feeding of bacteria that are not primary consumers of the ¹³C-labeled substrate added. We added ¹²CO₂ to minimize autotrophic cross feeding; however, cross feeding could still occur via the mono- or oligosaccharides produced from the initial EPS hydrolysis by exo- and endoglucanases or via other cell exudates or lysis products. However, unless the methodological biases are very great, secondary consumers should be of lower abundance than primary consumers. Given that the Planctomycetes made up as much as 60% of the total reads in the heavy fraction of the B. indica [¹³C]EPS SIP and that the growth of some Planctomycetes on plates containing EPS as the sole energy source could be verified, it seems likely that these are primary degraders of the added substrate rather than secondary degraders of monosaccharides or other molecules.

Some minor populations could represent secondary consumers or less predominant primary consumers of the B. indica EPS. One enriched OTU belonged to the phylum Armatimonadetes, formally known as candidate phylum OP10. The first cultivation of an OP10 bacterium was reported from decaying plant material in a thermal environment (51). A key to cultivating this strain was the use of gellan as the energy source (51), and like the B. indica EPS, gellan is a complex heteropolysaccharide. Analysis of the first Armatimonadetes genome identified 65 glycoside hydrolases, and a lifestyle whereby it scavenges a wide range of plant oligo- and polysaccharides was postulated for this bacterium (52). Another OTU from the BHS soil incubation belonged to the Verrucomicrobia order Chthoniobacterales, which is known to include common soil strains that mainly use mono-, di-, and polysaccharides (53). A member of candidate phylum OD1 (also unofficially dubbed “Parcubacteria”) was also strongly enriched. Primarily on the basis of metagenomics and single-cell genomics, a fermentative lifestyle has been proposed for some members of this phylum (54–56). OD1 bacteria are predicted to be anaerobes or facultative anaerobes producing acetate, ethanol, lactate, and H₂ as fermentation end products (55). It is, of course, impossible to extrapolate the results from studies of a few cells to an entire phylum, but the development of low-O₂ microsites in our incubations, combined with a fermentative process breaking down B. indica EPS or oligosaccharides produced by primary hydrolytic attack, is possible.

SIP can be combined with cultivation methods to track actively growing bacteria and customize media for target bacteria. The isolation of heterotrophic Planctomycetes is difficult, in part because of low growth rates (50, 57). Our incubations ran over 111 days, and we used the sensitive SIP method to detect enrichment, which overcame the limitation of slow growth and implicated these bacteria as primary players in degrading one heteropolysaccharide in soil. Due to their low growth rates, isolation of Planctomycetes in the laboratory requires special procedures, such as the use of antibiotics to reduce overgrowth of fast-growing, r-selected bacteria. Common beta-lactam antibiotics do not target Planctomycetes, possibly because they have a proteinaceous cell wall (45, 57, 58). Another option for isolation is to use very low nutrient concentrations (45, 59, 60) or a substrate for which there are few competing r-selected species, such as N-acetylglucosamine (NAG). Many Planctomycetes consume this substrate, perhaps because they grow in association with chitinolytic microorganisms that release this monomer (50, 57). As suggested by the present study, other substrates that may reduce competition with rapidly growing bacteria would be certain complex heteropolysaccharides, such as the heteropolysaccharide produced by B. indica. These are unique both in their secondary structure and in some of their monomeric components, like d-glycero-d-mannoheptose, that are rarely used in bacteriological media. The planctomycetes detected in this experiment may be able to metabolize this unusual sugar. Efforts are ongoing to isolate the bacteria detected in this study using B. indica EPS as a sole substrate.