Detection of 14C-phenol-assimilating microorganisms by CIArray analysis
For radioactive labeling of phenol-utilizing microorganisms in the activated sludge, a sample (collected in April 2011) was incubated under anoxic conditions with 13 mg L−1 (270 μCi of radioactivity) of 14C-phenol, 12 mg L−1 of 12C-phenol and 18 mg N L−1 of nitrate. Upon incubation at room temperature, phenol was degraded within 6 h (Supporting Information Fig. S1), with consumption of near-stoichiometric amounts of nitrate and minimal accumulation of nitrite.
Subsequently, ~ 200 μg (with a radioactivity of 150 nCi) of extracted DNA was hybridized to the CIArray, followed by radio-imaging to reveal spots with 14C-enriched DNA. As shown in Fig. 1a, positive signals were observed for 12 probes (of a total of 96) on the array; this suggests that microorganisms that assimilated 14C, presumably directly from phenol considering the short incubation period, were abundant in the reactor.
Figure 1. (a) Autoradiograph of the CIArray after hybridization with whole-community DNA from activated sludge sample amended with 14C-labeled phenol (13 mg L−1) under nitrate-reducing conditions. Spots marked by solid lines are probes (i.e. fosmid clones) with a positive signal and the numbers indicate probe IDs. The three spots marked by dashed lines are positive hybridization controls containing whole-community DNA. (b) Normalized abundance of one of the positive probes [Fos33; marked with an arrow in (a)] in the buoyant density-resolved DNA fractions prepared from sample incubated with 25 mg L−1 of 12C- and 13C-phenol (shown as empty and filled symbols, respectively).
Download figure to PowerPoint
For identification, 11 CIArray-positive probes (one of the 12 clones appeared to be contaminated during preparation of fosmid DNA for sequencing and was therefore omitted) as well as 12 randomly selected probes with negative signal on the CIArray were subsequently partially sequenced from both ends and the sequences compared with public databases. As shown in Table S2, most sequences possessed only low similarity to those in the NCBI nr nucleotide database. The highest identity (87%, over a 770-bp region) to known bacteria was found for one of the CIArray-negative clones, which was to the full genome of Dinoroseobacter shibae strain DFL 12 (Biebl et al., 2005). This species was originally isolated from Tokyo Bay and these bacteria likely entered the reactor via the seawater that was used to prepare the synthetic coke-oven wastewater. Other sequences also had best-matches to various taxa known to be ubiquitous in marine environments as well as several denitrifying bacteria (Table S2).
Taxonomic assignment of the fosmid clone end sequences showed that the CIArray-positive clones that could be classified at the class level (13 of 22 sequences) belonged to the Gammaproteobacteria (Table S2). For the CIArray-negative clones, sequences with class-level assignment belonged to both the Gamma- and Alphaproteobacteria. While more accurate identification was not feasible, it is anticipated that this should be possible by complete sequencing of the fosmid clones, especially considering the current rapid expansion of public sequence databases. Nevertheless, even without unambiguous identification, the sequences can be directly used for designing specific assays to track these microorganisms, for example using quantitative real-time PCR.
Validation of the CIArray results by DNA-SIP
To validate the CIArray data, real-time PCR primers were designed for the 11 CIArray-positive clones and six randomly selected negative clones, and used to quantify their abundance in density-resolved DNA fractions from the 13C- and 12C-phenol-amended samples. For the incubation with 13C-phenol (25 mg L−1) set up in parallel with the 14C-incubation, clone-specific DNA was not shifted toward higher buoyant densities (BDs) in comparison with the 12C-phenol sample for any of the positive clones (a representative result is shown in Fig. 1b). This indicated that 13C-enriched DNA was not detectable, which may have been attributed either to unreliable identification using the CIArray due to cross-hybridization, resulting in qPCR assays that fail to target microorganisms that assimilated isotope-labeled carbon, or to inferior detection sensitivity of the DNA-SIP.
To clarify this point, another SIP experiment was performed with higher concentrations of phenol. More specifically, 25 mL of sample (collected in August 2011) was incubated with 30 mg L−1 of 13C-phenol (added on day 0) and 150 mg L−1 of 13C-phenol supplemented on days 1 and 2. Nitrate was supplemented on day 0 (40 mg N L−1) and on days 1 and 2 (160 mg N L−1). Again, no increase in BD was detected for DNA obtained from the 13C-phenol incubation after degradation of 30 mg L−1 of phenol for any of the positive clones (Fig. 2a,b). After repeated amendment of phenol (150 mg L−1), a small but discernible degree of 13C-enrichment was evidenced by shouldering of the DNA peak toward higher BDs. A clearly distinguishable peak at high BD was detected for all of the positive clones after the third amendment of phenol (Fig. 2a,b; Fig. S2). For none of the CIArray-negative clones was a ‘heavy’ DNA peak observed on day 3 (Fig. 2c,d).
Figure 2. Normalized abundance of two representative CIArray-positive (a: Fos07 and b: Fos40) and -negative probes (c: Fos37 and d: Fos69) in the density-resolved DNA fractions from sample incubated with 12C- and 13C-phenol. SIP profiles after incubation with 30 mg L−1 (day 1, filled triangles), 180 mg L−1 (day 2, filled diamonds) and 330 mg L−1 (day 3, filled circles) of 13C-phenol and 330 mg L−1 of 12C-phenol (day 3, empty circles) are shown.
Download figure to PowerPoint
Taken together, the SIP analyses verified that all probes that were identified as positive on the CIArray stemmed from microorganisms that integrated isotope-labeled carbon atoms into their DNA. This could be either through direct assimilation from labeled phenol or through cross-feeding of metabolites from primary phenol degraders, with the short incubation period applied for the 14C-phenol incubation (< 6 h) suggesting that they may represent primary consumers. In any case, the complete concordance between both datasets indicated that the CIArray was reliable and specific, although more in-depth studies using pure cultures would be needed to determine the level of specificity that can be achieved using fosmid clone probes under the specified hybridization conditions.
Importantly, our analyses also indicated that the CIArray technique was more sensitive than DNA-SIP since positive detection was achieved at a phenol concentration that failed to yield a ‘heavy’ DNA fraction, even when the density gradients were assessed with fosmid-clone-specific PCR assays. The superior sensitivity of the CIArray is presumed to be due to the better resolution of labeled and unlabeled DNA through radio-imaging, as compared with isopycnic centrifugation and fractionation for SIP. Although the DNA-labeling efficiency of the 14C-assimilating bacteria could not be determined accurately since radioactivity is measured in bulk community DNA, the higher sensitivity of hybridization-based assays is in qualitative agreement with Mayali et al. (2011). In that study, it was shown that an isotopic enrichment of 0.5 atom% 13C was sufficient for detection of labeled rRNA on a microarray using NanoSIMS for imaging, compared with the >20% necessary for density gradient separation (Uhlík et al., 2009).
Finally, to obtain an independent view on the identity of the phenol-degrading microorganisms, a 16S rRNA gene clone library was prepared from the ‘heavy’ DNA fraction of the 13C-phenol-supplemented sample after degradation of 330 mg L−1 of phenol. Based on the partial (700 bp) 16S rRNA gene sequences, two operational taxonomic units (OTUs, defined at 97% sequence identity) accounted for more than half of chimera-checked sequences in the library (n = 31; Table S4). Both OTUs were assigned to the unclassified Gammaproteobacteria using the RDP naive Bayesian classifier (Wang et al., 2007) and they shared only 90–92% sequence identity with named isolates in the Greengenes database (DeSantis et al., 2006; Table S4), suggesting that they may represent novel species. Interestingly, the sequences belonging to OTU LE12 shared >98% sequence identity with an uncultured Gammaproteobacterium (accession number: AM292411) previously detected in a denitrifying toluene-degrading enrichment prepared from marine sediment (Alain et al., 2012).
In agreement with our findings, a previous study that employed rRNA-SIP for identification of phenol-assimilating bacteria in the same reactor system (Sueoka et al., 2009), also found that both gammaproteobacterial OTUs (with >99% identity to the sequences belonging to OTUs LE12 and LE05) represented more than half of 16S rRNA clones derived from ‘heavy’ rRNA fractions. However, in the latter work, a phylotype affiliated with Azoarcus was concluded to be the primary phenol degrader since it was the first to incorporate 13C-atoms. Here, Azoarcus-related sequences were not detected in the fosmid clone library, suggesting that it was not a dominant community member during the period of our study. Taken together, these data suggest that novel marine Gammaproteobacteria, probably derived from the seawater that is used to dilute the coke-oven wastewater prior to treatment, constitute a stable phenol-degrading guild in the reactor.
In summary, a novel isotope array technique – the community isotope array – was developed and validated by detection of anoxic phenol-assimilating bacteria present in activated sludge used to treat synthetic coke-oven wastewater. Validation by DNA-SIP proved that the CIArray was highly specific and also enabled positive identification at a lower substrate concentration in comparison with DNA-SIP, which may render the CIArray technique applicable to a broader range of substrates and/or environments.