Case Study 3: Functional Grouping

an ecological function to organisms. Especially, if the metagenome and annotations for the particular organisms are incomplete, additional information on an organism’s i_{A bacterium that cannot manufacture its own food from simple molecules and needs to consume organic}

substances from an external source.

ii_{A bacterium that produces complex organic molecules required for survival from simple molecules. Often light}

Results

Figure 4.8: Time-course experiment with labeling substrate. If no unlabeled reference

sample (diamonds) is employed, the number of detected, labeled peptides decreases as less unlabeled peptide features are detected and utilized as a reference. Adding unlabeled reference samples (circles) compensate for this effect and increase the number of labeled peptides detected by MetaProSIP. Adapted from Sachsenberg et al.110.

biochemical repertoire can be deduced from its (in-)capability to metabolize certain substrates. Qualitatively grouping into active and inactive organisms (those that are able/unable to metabolize a substrate) is performed based on the presence of labeled peptides. Unlabeled peptides indicate an inactive organism. If the peptides are labeled with high RIA, the organism is likely a primary metabolizer of the substrate. Low RIA might correspond to organisms that metabolize additional substrates, including organisms that metabolize excreted molecules or scavenger organisms, that consumed biomass of labeled organisms.

Taubert et al.123 _{investigated anaerobic benzene degradation using a time series with} 13_{C-labeled CO}

2 and benzene substrates. Manual analysis of RIAs distinguished three active but functionally different groups of microorganisms. Benzene is first anaerobi- cally degraded by organisms of the Clostridium genus. Subsequently, Desulfobacteria consume the formed metabolites for biomass production. The third group of organ- isms was suspected to be not directly involved in substrate degradation but instead are scavengers which feed on dead cell mass of other bacteria. We applied the basic MetaProSIP workflow (Appendix Figure D.1 and Table D.1) to this previously manually analyzed dataset. Figure 4.9.a is based on the automatically generated heatmap. The clustering performed in

MetaProSIP

tinct RIAs. Each of them corresponding to a different incorporation behavior. To verify that these groups indeed reproduce the three reported groups from Taubert et al.123_we annotated the peptides using BlastP96and assigned phylogenetic taxa using MEGAN124 (Figure 4.9.b). In concordance with the results from Taubert et al.123 _{the high RIA} group was annotated to predominantly originate from Clostridiales. Deltaproteobacteria were also identified for the medium RIA group. The low RIA group was, as expected and previously reported, more heterogeneous since dead cell biomass can be metabolized by a range of bacterial taxa.

Figure 4.9: a color-coded heatmap showing the three RIA groups as determined by

MetaProSIP. b Annotation of groups with phylogenetic information reveals a distinct

composition of microorganisms. Group I is clearly dominated by Clostridiales, group II by Deltaproteobacteria while group III displays a more heterogeneous composition of phylogenetic taxa. Adapted from Sachsenberg et al.110_.

Results

Taubert et al. showed that protein-SIP allows tracing of elemental fluxes between two time points.

Figure 4.10: a Three groups, (high, medium and low RIA) have been detected and analyzed

with MetaProSIP at two time points. Median RIA of group II differs significantly between

t₁and t₂. b Phylogenetic annotation and biological interpretation allow reconstructing the elemental flux. Adapted from Sachsenberg et al.110_.

Figure 4.10.a displays the RIA distribution obtained for the three groups at two different time points (t1= 180 d, t2 = 300 d). While group I and III RIAs are relatively stable, group II shows a significant increase in RIA (p< 0.65·10−9) and LR (p< 0.61·10−3, LR not shown) confirmed by a two-tailed, heteroscedastic t-test. While an increase in LR can be explained by protein turnover and growth, the increase in RIA can be explained by the mode of substrate metabolism: An increase of RIA in group II indicates, that the 13_{C content of its substrate pool is increased over time. As the labeling of the externally} provided substrate is constant, the increase is likely a result of group II organisms consuming metabolites from group I or III. Because group I has the highest level of RIA and the phylogenetic annotation (Clostridiales-like) identifies them as organisms able to break the benzene ring, these are likely at the top of the degradation hierarchy. Based on their phylogenetic annotation, group III are mainly composed of scavenger organisms that consume dead biomass of group I and II organisms. They are therefore at the bottom of the degradation hierarchy with group II (Deltaproteobacteria-like) taking an intermediate position. Combining this information with biological knowledge

(e.g., group I and II are known to release and fixate CO2), parts of the elemental flux network can be hypothesized (Figure 4.10.b). A working hypothesis could be that in the elemental flux network,13C-benzene is initially degraded by group I organisms. Release and fixation of labeled CO2 between group I and group II yield an increase of RIA in group II. In addition, metabolites from group I (e.g., acetate) with increasing 13_{C content, are also incorporated in organisms of group II. Group III mainly show} an increase in RIA because they are composed of scavengers that feed on dead (and potentially labeled) organisms. In summary, we demonstrated that functional grouping based on incorporation behavior is indeed feasible. The biological interpretation of used substrate, LR and RIA shifts between organisms and time allows hypothesizing parts of the elemental flux network which may be confirmed using additional experiments. In our experiments, we only used two time points. More time points, different substrates (e.g., labeled acetate or CO2) may be possible follow-up experiments to further resolve

the elemental flux.