Bioremediation in marine ecosystems: a computational study combining ecological modeling and flux balance analysis

Abstract

The pressure to search effective bioremediation methodologies for contaminated ecosystems has led to the large-scale identification of microbial species and metabolic degradation pathways. However, minor attention has been paid to the study of bioremediation in marine food webs and to the definition of integrated strategies for reducing bioaccumulation in species. We propose a novel computational framework for analysing the multiscale effects of bioremediation at the ecosystem level, based on coupling food web bioaccumulation models and metabolic models of degrading bacteria. The combination of techniques from synthetic biology and ecological network analysis allows the specification of arbitrary scenarios of contaminant removal and the evaluation of strategies based on natural or synthetic microbial strains. In this study, we derive a bioaccumulation model of polychlorinated biphenyls (PCBs) in the Adriatic food web, and we extend a metabolic reconstruction of Pseudomonas putida KT2440 (iJN746) with the aerobic pathway of PCBs degradation. We assess the effectiveness of different bioremediation scenarios in reducing PCBs concentration in species and we study indices of species centrality to measure their importance in the contaminant diffusion via feeding links. The analysis of the Adriatic sea case study suggests that our framework could represent a practical tool in the design of effective remediation strategies, providing at the same time insights into the ecological role of microbial communities within food webs.

Keywords: Adriatic sea; PCBs; Pseudomonas putida; bioremediation; ecological network analysis; flux balance analysis.

Figure 1

Conceptual model of the Adriatic PCBs bioaccumulation network. Flows are shown with respect to a generic functional group. Mass-balanced groups are enclosed in the gray boxes, externals are shown outside. The dashed arrow from planktonic groups indicate possible indirect connections. Feeding links from discard and detritus are omitted. (A) Red arrows indicate contaminant flows mediated by feeding connections. (B) Green arrows highlight the potential propagation of bioremediation effects. Possible bioremediation scenarios are assumed at the interface between detritus and planktonic groups (microbial loop), or in the water compartment.

Figure 2

Integration of the aerobic pathway of PCBs degradation in the core metabolism of P. putida KT2440 (iJN746). BphA, biphenyl 2,3-dioxygenase (multicomponent Rieske non-heme iron oxygenases); BphB, cis-2,3-dihydrobiphenyl-2,3-diol dehydrogenase; BphC, biphenyl-2,3-diol 1,2-dioxygenase; BphD, 2,6-dioxo-6-phenylhexa-3-enoate hydrolase; mhpD, 2-keto-4-pentenoate hydratase; mhpE, 4-hydroxy 2-oxovalerate aldolase; mhpF, acetaldehyde dehydrogenase.

Figure 3

(A) Bilevel analysis on the P. putida metabolism: we study the optimal growth rates on the solution space of optimal PCBs uptake (L1), when the upper bound of the latter ranges from 0 to 15 mmol h⁻¹ gDW⁻¹. The maximum PCBs uptake rate is 10 mmol h⁻¹ gDW⁻¹, and the optimal growth rate is thus achieved for almost the whole range of PCBs uptake. (B) Single-level analysis: controlled/optimal flux of biomass and PCBs uptake rate at different oxygen levels, which in our case are determined also by different depths. The P. putida is able to keep a high growth rate also on low oxygen. The linear relationship between PCBs and oxygen uptake rates is in keeping with the fact that the uptake of PCBs depends on aerobic degradation. (C) Interdependence between toluene and PCBs uptake and corresponding phenotypic phase plane (PhPP). The red dashed line shows the trade-off between toluene and PCBs uptakes, obtained with a bilevel analysis of optimal toluene uptake (L2), over the configuration maximizing PCBs uptake (L1), by limiting the latter from 0 to 10 mmol h⁻¹ gDW⁻¹. The symmetric bilevel problem (with toluene limited from 0 to 20 mmol h⁻¹ gDW⁻¹) gives the same linear front. This tradeoff delineates two phenotypes in the PhPP analysis (L2: biomass, L1: toluene+PCBs uptakes): in the lower half (green region), we have optimal growth; in the upper half (blue region), growth is limited to 71% of the optimal growth.

Figure 4

Circular plot of the Adriatic food web in the three cases considered: PCBs bioaccumulation network without bioremediation (A; Scenario 0); at maximum bioremediation efficiency for the natural bioremediation acting on detritus and discard (B; Scenario 1); and the in-situ bioremediation acting on the water compartment (C; Scenario 2). Functional groups are located clock-wise in ascending trophic level order. Ribbons represent feeding links carrying PCBs flows. Each ribbon takes the same color as its source node (the prey), and thickness is proportional to the contribution of the source in the diet of the target node (the predator). In each group, the outmost stacked bars summarize its diet composition and its contribution to predators' diet. External and flows to detritus groups are not displayed. The top-right table lists the functional groups of the Adriatic food web and their ID numbers. Images has been obtained by using the Circos tool (Krzywinski et al., 2009).

Figure 5

Levelplots of PCBs concentrations (A,B) and flow betweenness centralities (C,D) in Adriatic species (y-axis) at increasing amounts of contaminant removed by bacterial uptake (x-axis) in the natural (A,C) and in situ (B,D) bioremediation scenarios. In the middle, the final amount of remediated flow and the corresponding PCBs uptake are reported for the two scenarios. Plots on the top of (A,B) show the evolution of the sum of PCBs in the food web at increasing degrees of bioremediation. Plots on the bottom of (C,D) show the effects of bioremediation in the link density of the bioaccumulation network.