CAMDLES: CFD-DEM Simulation of Microbial Communities in Spaceflight and Artificial Microgravity

Abstract

We present CAMDLES (CFD-DEM Artificial Microgravity Developments for Living Ecosystem Simulation), an extension of CFDEM^®Coupling to model biological flows, growth, and mass transfer in artificial microgravity devices. For microbes that accompany humans into space, microgravity-induced alterations in the fluid environment are likely to be a major factor in the microbial experience of spaceflight. Computational modeling is needed to investigate how well ground-based microgravity simulation methods replicate that experience. CAMDLES incorporates agent-based modeling to study inter-species metabolite transport within microbial communities in rotating wall vessel bioreactors (RWVs). Preexisting CFD modeling of RWVs has not yet incorporated growth; CAMDLES employs the simultaneous modeling of biological, chemical, and mechanical processes in a micro-scale rotating reference frame environment. Simulation mass transfer calculations were correlated with Monod dynamic parameters to predict relative growth rates between artificial microgravity, spaceflight microgravity, and 1 g conditions. By simulating a microbial model community of metabolically cooperative strains of Escherichia coli and Salmonella enterica, we found that the greatest difference between microgravity and an RWV or 1 g gravity was when species colocalized in dense aggregates. We also investigated the influence of other features of the system on growth, such as spatial distribution, product yields, and diffusivity. Our simulation provides a basis for future laboratory experiments using this community for investigation in artificial microgravity and spaceflight microgravity. More broadly, our development of these models creates a framework for novel hypothesis generation and design of biological experiments with RWVs, coupling the effects of RWV size, rotation rate, and mass transport directly to bacterial growth in microbial communities.

Keywords: CFD-DEM; clinostat; computational fluid dynamics; microbial community; microgravity; population dynamics; rotating wall vessel.

Figure 1

Software technology tree. CAMDLES relates to parent software via black arrows and related software via green arrows.

Figure 2

Model microbial community simulated in this study. Adapted with permission from [28]; published by Cell Reports, 2014. When grown in minimal media with lactose, Escherichia coli K12 with a metB gene knockout (E. coli ΔmetB) provides acetate as the carbon growth substrate necessary for a methionine-secreting mutant of Salmonella enterica ser. Typhimurium (S. enterica) to grow. In exchange, the S. enterica provides the methionine necessary for the E. coli to grow.

Figure 3

Free body diagram of a particle immersed in an RWV. Acceleration terms are described in Equation (3). The simulation domain is a rectangular box in a rotating reference frame.

Figure 4

Average E. coli Fold Change in Mass vs. Colony Population in Microgravity. Solid lines represent statistical means, while transparent wedges represent inner quartiles. Either densely packed spherical or randomly distributed colonies were generated at various sizes. Only one colony was generated per simulation. Note: as the population density increased, the standard deviation of growth rates increased, which was likely caused by local spatial heterogeneity within the simulation domain, increasing the likelihood of resource competition.

Figure 5

E. coli Percent Maximum Growth Rate vs. Colony Population in Microgravity. Densely packed multispecies spherical colonies were generated at various sizes. Only one colony was generated per simulation. In solid colonies, self-insulation limited the growth rate beyond a colony of 500 cells. The effect was reduced in liquid colonies.

Figure 6

Acetate diffusion. The symbol C represents the acetate concentration in units of cell equivalents/μm³; note that the color scale is different between the two panels. Compared to other simulations, the E. coli:S. enterica population ratio was increased to a 1956:44 count. Diffusion parameters were set to be equal in (a,b) at a nominal 5 × 10⁻⁶ cm²/s. (a) Two-species spherical colony; (b) random uniformly distributed cells.

Figure 7

Acetate diffusion in RWV conditions. The color scale (symbol C) represents acetate concentration in units of cell equivalents/μm³. Note that the direction of the diffusion gradient was reversed in this case compared to that found in microgravity (Figure 6a) and that the acetate concentrations decreased. The reversed gradient indicated that the growth rates were unsustainable.

Figure 8

Single-species Colonies Distanced Far apart. The E. coli:S. enterica population ratio was adjusted to optimal conditions (1895 count:105 count). (a) Coculture colonies of E. coli (green) and S. enterica (red). The symbol C represents acetate concentration in units of cell equivalents/μm³. Colonies were generated as spherical Gaussian distributions placed as far away as possible within the computational domain. The microgravity case is displayed, but the RWV case was near-identical. (b) Collection of S. enterica growth over time in tightly packed colonies. The centerline plots the average cell mass. The dark shaded region shows the inner quartiles, and the light shaded region shows the outer quartiles. The growth trajectories for the cases depicted in Figure 6a and Figure 7 are graphed here.

Figure 8

Single-species Colonies Distanced Far apart. The E. coli:S. enterica population ratio was adjusted to optimal conditions (1895 count:105 count). (a) Coculture colonies of E. coli (green) and S. enterica (red). The symbol C represents acetate concentration in units of cell equivalents/μm³. Colonies were generated as spherical Gaussian distributions placed as far away as possible within the computational domain. The microgravity case is displayed, but the RWV case was near-identical. (b) Collection of S. enterica growth over time in tightly packed colonies. The centerline plots the average cell mass. The dark shaded region shows the inner quartiles, and the light shaded region shows the outer quartiles. The growth trajectories for the cases depicted in Figure 6a and Figure 7 are graphed here.

Figure 9

E. coli Percent Maximum Growth Rate with Various Product Yield Parameters λ. If not specified, λ was set to 100% as the default. Densely packed multispecies spherical colonies were generated with a colony size of 2000 cells. Only one colony was generated per simulation. In an RWV, gravity caused the colonies to move as either a rigid unit (regarded as a solid) or each cell moved individually (regarded as liquid). This was not important to the growth rates and was only tested in: (a) an RWV vs. microgravity with low diffusion parameters like those in biofilms. * S. Enterica growth is displayed as it was the faster-growing species, and the diffusion was at an intermediate level. (b) An RWV vs. microgravity with high diffusion parameters like those in liquid media.

Figure 10

Natural Convective Flow. A spherical biofilm placed on the simulation floor exhibited solute convective flows in a 1 g gravity condition. Red velocity arrows, in units of s⁻¹, are scaled up by 1000×. The concentration field of acetate is depicted from high to low as red to blue. Fluid simulation boundaries are periodic except on the floor and ceiling of the box.