Activity-based cell sorting reveals responses of uncultured archaea and bacteria to substrate amendment

To expand on previous BONCAT findings, we initially performed an experiment with E. coli, which demonstrated that translational activity as measured by BONCAT can only be detected when cultures were grown with glucose or sorbitol as the sole carbon and energy source but not with sucrose (Supplementary Fig. 1). This is consistent with the notion that E. coli is genetically incapable of sucrose utilization. Addition of chloramphenicol, an antibiotic targeting ribosome function, to the culture decreased the BONCAT fluorescence intensity to background levels, i.e., fluorescence values observed when cells were grown with HPG but without growth substrate. These results demonstrated that BONCAT can be used to study cell activity responses to substrate amendment and suggested that it could be used to study complex microbiomes.

For this benchmark study, we selected a high temperature (74 °C), alkaline (pH 8.2) hot spring (Five Sisters 5, FS5) in the Lower Geyser Basin of YNP (Fig. 1). Hot springs harbor lower complexity microbial communities compared to other environments, such as soils and marine sediments, making them an ideal system for the development of novel approaches. As indicated by 16S rRNA gene amplicon sequencing, hot spring FS5 was dominated by the archaeal candidate phylum Aigarchaeota (41.9%), with the next most abundant community members belonging to bacterial candidate phylum Fervidibacteria (9.57%) and phylum Deinococcus-Thermus (7.98%) (Fig. 2). Members of the Aigarchaeota have been detected in many high temperature (65–88 °C) hot springs over a wide pH range (2.9–9.3) [25]. Despite multiple studies describing their versatile metabolic potential, Aigarchaeota remain recalcitrant to cultivation and no experimental data on their functional activity are currently available. Reconstructed genomes suggest the capacity for aerobic respiration or anaerobic respiration with nitrate as an electron acceptor [25]. In addition, a recent study found multiple endoglucanases and β-glucosidases that might be involved with degradation of cellulose and cellobiose in an Aigarchaeota metagenome bin [26]. One of the goals of our study was to test these functional predictions using BONCAT-FACS.

We applied BONCAT-FACS combined with 16S rRNA gene sequencing to identify the active microbial community members in hot spring material when incubated in the presence of 23 different substrates or growth conditions (Fig. 1; see list of treatments in Supplementary Table 1). Because BONCAT relies on the incorporation of a synthetic amino acid as an indirect activity tracer, we were also able to test activity responses toward changes of headspace gas composition. Specifically, we tested the activity of microbes in the presence of atmospheric (21% O₂), microoxic (2% O₂), and anoxic (100% N₂) headspace conditions. Laboratory incubations were established with a slurry containing FS5 water and sediment to which HPG and substrates were added. After 48 h of incubation, biomass was resuspended in glycerol Tris-EDTA buffer [12] and samples were frozen at −80 °C for later processing. After samples were thawed, cells were extracted from the slurry, fluorescently stained using click chemistry, and sorted via FACS based on fluorescence signal. The sorted cells were lysed with repeated freeze–thaw cycles prior to PCR amplification of their 16S rRNA genes. An HPG-only incubation was used to identify cells active under close to in situ conditions (with low perturbation) and represented the baseline for comparing cellular activity under varying incubation conditions.

The microbial community of FS5 was highly active as represented by the high proportion of BONCAT-labeled events in the HPG-only control (Supplementary Table 1). Furthermore, all abundant taxa (>1% relative abundance of the total extractable community, ASVs collapsed to the genus level) were active in at least one condition, and most taxa were active under several conditions (Fig. 2). This demonstrates that BONCAT can be applied to a wide variety of phylogenetic groups, which is consistent with previous reports [5,6,7,8,9, 27]. The sorted, active fraction from incubations with different treatments contained no statistically significant difference when compared to the HPG-only control based on Bray–Curtis dissimilarity (MANOVA, p = 1) (Fig. 2). This result indicates that the microbial activity response to any treatment, as captured by BONCAT, was not large enough or consistent enough among replicates to significantly change the overall active community when compared to the HPG-only control.

Although the overall composition of the active communities did not vary significantly in response to incubation conditions, changes within the richness and evenness of the communities were detected. To further describe the community composition, the Shannon’s diversity index of each incubation was calculated (Fig. 3). For each treatment, the bulk, presort fraction (representing the total, extractable cell community), and the sorted, active cell fraction were analyzed and compared to the respective HPG-only control. Overall, the variation in the presort fraction of Shannon’s diversity indices was less variable than the sorted fraction from the same treatment incubation (Fig. 3). This demonstrated that the BONCAT-FACS approach could detect changes in the diversity of the active community due to individual cell responses prior to shifts in cell abundance occurring in presort communities. We expected little variation in the presort community because the limited time of incubation should not have allowed for an overall shift in community composition on a bulk level. However, Fervidibacteria were observed in higher proportion in most presort populations as compared to the original T₀ bulk sample. This could be attributed to either favorable growth of this yet uncultured lineage or be a result of preferential cell extraction or cell lysis during freeze–thaw cycles as compared to bulk sample DNA extraction. Alternatively, their increase in relative abundance could have resulted from sample cooling (72 to 55 °C) during transit from the field to the laboratory (4 h).

The Shannon’s diversity indices of the active, sorted fractions were more varied when standardized to the HPG-only control than the presort fractions. This indicates that the activity response among different taxa varies between treatments. For example, incubations with cellobiose or under anoxic conditions led to statistically significantly different diversity indices on average than the HPG-only control incubations (linear mixed effects model, p = 0.00349 and p = 0.0334, respectively). Cellobiose is a disaccharide of repeating glucose monomers that compose cellulose polymers. Cellobiose can be cleaved by β-1,4 hydrolases into glucose monomers which serve as initial substrate for most central metabolic pathways. The Shannon’s diversity index of cells active with cellobiose amendment was the highest we observed alluding to wide use of cellobiose increasing the evenness of active microbes in these incubations (p = 0.00349).

The perimeter of FS5 has a photosynthetic zone (<70 °C; Fig. 1), away from the sampled area, which, together with constant exchange with the atmosphere, serves as a source of O₂ in the spring (dissolved oxygen levels were 0.909 mg/L at the time of sampling but can vary 1–3 mg/L). Headspace exchange from aerobic (atmospheric air, 21% O₂) to anoxic conditions (100% N₂) in the incubations caused the active community to have a significant increase in Shannon’s diversity (p = 0.0334) compared to the HPG-only control. This is consistent with the idea that facultative anaerobes persist in the hot spring and their activity varies with oxygen availability.

The relative abundance of six taxa increased and two decreased statistically significantly in the active, sorted community in response to cellobiose-amendment based on log₂-fold change (LFC) when compared to the HPG-only sorted fraction (p < 0.10). This suggests that cellobiose has an impact on the activity of several taxa despite the hot spring lacking noticeable large carbon inputs or organic matter that would be degraded regularly in situ. Uncultured phylum BP4 had an LFC of 3.62 in response to cellobiose amendment (Fig. 4; for a list of all ASVs see Supplementary Table 3). While not much is known about BP4, these results could suggest a substrate preference for BP4 toward cellobiose or its degradation products. In contrast, candidate phylum Fervidibacteria decreased in relative abundance in the active fraction of cellobiose amendment with an LFC of −2.88 (Fig. 4). This decrease in Fervidibacteria abundance in the active fraction was surprising because this lineage had previously been proposed to have the genomic potential for complete lignocellulose degradation [28]; further, we did not detect a significant response to the addition of cellulose (p = 1). These results demonstrate the importance to functionally test genomic predictions under environmentally relevant conditions.

In anoxic incubations, Thermus increased (LFC = 3.53, Fig. 4) and represented a large portion of the sorted, translationally active community (Fig. 2). In contrast, a rare member of the Thaumarchaeota (averaged relative abundance in the HPG-only sample was 0.4%), whose taxonomic affiliation could not be resolved beyond phylum level, decreased (LFC = −4.31, Fig. 4) in activity in response to anoxic conditions. This is consistent with the observation that all cultured thaumarchaeotes, including thermophilic representatives, have a strictly aerobic lifestyle [29,30,31,32].

Sequences related to the genus Pyrobaculum exhibited the largest LFC increase in abundance of any taxon detected in the presence of cellobiose (LFC = 6.52) and under anoxic conditions (LFC = 8.12) (Fig. 4). Pyrobaculum isolates demonstrate widely varying phenotypic capabilities, including both autotrophic and heterotrophic lifestyles, which can be coupled to H₂, O₂, AsO₄³⁻, S⁰, S₂O₃²⁻, SeO₄²⁻, Fe³⁺, and NO₃⁻ respiration [33,34,35,36,37]. This metabolic versatility permits Pyrobaculum species to occupy geochemically diverse, mildly-acidic to basic (pH > ~4) geothermal environments, including hot springs throughout YNP [38] and likely explains their significant LFC increases in the two treatments.

Cellobiose and anoxic incubations were the only treatments to inform us of community-wide substrate and headspace selectivity with statistical significance. However, organisms belonging to Thermocrinis and uncultured phylum Gal15 increased in relative abundance in both the cellobiose (p = 0.0219 and 0.0488, respectively) and anoxic (p = 0.0229 and 0.0209, respectively) incubations’ sorted fractions alluding to possible favorable cultivation conditions for these organisms. Other organisms were also detected with large LFCs, though they were not statistically significant when averaged across all three biological replicates. For some samples high variability was observed, emphasizing the need for replication in order to draw biologically meaningful conclusions.

An uncultured member of the Aigarchaeota represented a smaller proportion of presort reads (31.4%) after the 2-day incubation than in the original sample (41.9%) (Fig. 2). While Aigarchaeota sequences were dominant in the presort fractions, they were not highly abundant in any active fraction (averaged 12.4%). In contrast, the BONCAT-labeled, active fraction of cells were dominated by the uncultured candidate phylum Fervidibacteria [39]. Fervidibacteria also comprised 48.0% of the total reads across all presorted, bulk communities and the active, sorted cell fraction (Fig. 2). These results could indicate a potential for slow growth or a lack of necessary growth factors for Aigarchaeota in our incubations, while Fervidibacteria maintained high activity regardless of the amendment. Previously published genomic predictions [26] informed us of possible favorable conditions for Aigarchaeota, including acetate and biotin, but none of the tested substrates stimulated a strong activity response. This suggests the importance of unknown community interactions or growth conditions of Aigarchaeota that so far have not been replicated in laboratory experiments and highlights the limits of genomic functional predictions and interpretations. Aigarchaeota have been proposed to have the genomic potential for both aerobic and anaerobic growth [25] and our experimental results support this hypothesis. Members of Aigarchaeota were most abundant in the active fractions of microoxic (20.0%) and anoxic (21.7%) treatments despite not being statistically significant (p = 1) (HPG-only = 13.3%). The ability to analyze microbial activity under varying headspace compositions demonstrates that BONCAT-FACS is an approach with great potential for studying cellular responses involving non-assimilatory pathways.

Preliminary experiments determined that 48 h was to be the optimal time for incubation of this sample type to allow for enough HPG incorporation to be detected efficiently on our cell sorter (data not shown). However, this incubation time showed high baseline activity in the HPG-only control making activity variation due to substrate amendment difficult to determine. Extended incubation times with HPG have been shown to not be detrimental to the survival or activity of microorganisms in other sample types [5,6,7, 9, 11]. However, we posit that proteins related to environmental changes or defensive or stress mechanisms could be responsible for increased protein synthesis. Longer incubation times could select for specific populations and act as pseudo-enrichment allowing “weed” organisms to obscure activity response from hard-to-culture or rare taxa. Despite our limited incubation time, we consistently observed a higher abundance of Fervidibacteria populations in presort samples, possibly attributing to higher proportions in sorted, active samples. Alternatively, sampling too early might not provide enough opportunity for organisms to respond to the conditions provided, attributing the BONCAT signal to activity independent of sample treatment, or a general lack of signal. This is a problem with any methodology surveying activity in response to substrate amendment including BONCAT and heavy water SIP [2].

Source: Ecology - nature.com