The Overflowing-Continuous Culture

Two Rushton impellers were fixed specifically on the rotating axis of the stirring device of the tank reactor (detailed Supplementary Figure 1 is given in the Supplementary Material section). The lower impeller was immersed in the culture broth as is usually done to ensure the mixing and oxygen transfer. The upper impeller was fixed just above the culture broth surface. The working volume of the bioreactor was fixed to 3 L. During the feeding phase of the process, the broth volume increased until the level of the upper impeller was reached, favoring the mixing at the gas–liquid interface. In these operating conditions, the impeller increased the foamability of the broth.

Foamability tests were done by adapting a previously described setup (Saint-Jalmes et al., 2005). The bubble column used by these authors was replaced by a stirred tank reactor. At the bottom of the tank, the air was blown at controlled rates through the sparger into the culture broth (constant bubble size). The air rate was fixed at 0.75 L⋅min^–1. Considering the culture broth, the temperature was fixed at 30°C, the pH at 6.5, the concentration in biomass at 1.3 g(DW)⋅L^–1. The mycosubtilin concentration was fixed at the maximal value that can be obtained during foam overflowing cultures, i.e., 2 mg⋅L^–1. The foam volume V_foam was deduced from foam height measured as a function of time, with a time lapse of 5 min, and compared with the volume of gas injected V_gas. The constant value of foamability K = V_foam/V_gas was measured for different stirring speeds corresponding to different peripheral speeds of the Rushton turbine, calculated at the tip of the blade.

Description of the Process. The term “overflowing continuous culture” as used in this paper describes an original continuous culture process whereby the outgoing outflow rate is divided into two separate flow rates, the outgoing flow rate in the waste vessel (F_out) and the foam overflowing flow rate (F_foam), as shown in Figure 1.

Flow sheet of the overflowing continuous culture (O-CC).

The process was separated into three successive phases exemplified in Figure 2. In phase I, cells were grown in a batch culture at the maximal specific growth rate, i.e., 0.33 h^–1. DO was controlled above 15% v/v with an adaptive stirring strategy. The phase I duration comprised between 24 and 28 h, until the observation of a decrease of the stirrer speed and a rise of the DO value, expressing a limitation in a substrate. In this phase, the volume of liquid remaining in the tank reactor decreased and the volume of the foam collected increased. This phenomenon was observed because of the foam overflowing flow rate occurring during the continuous production of mycosubtilin due to the constitutive P_repU promoter inserted in B. subtilis ATCC 6633 derivatives, BBG116 and BBG125. It has to be noticed that in the case of BBG116, which has a (Myc+++ Srf+) phenotype, the foam overflowing flow rate was also linked to the residual production of surfactin.

(A) Evolution upon time of the stirrer speed (^___) and the dissolved oxygen (^___) during an overflowing continuous culture (O-CC) of Bacillus subtilis BBG116 in the modified Landy culture medium at a temperature of 30°C, a pH of 6.5 u⋅pH, a dilution rate of 0.05 h^–1, and a dissolved oxygen value regulated above 15% v/v with an agitation cascade mode of between 200 and 400 rpm. (B) Evolution of the biomass dry weight (◆) in the stirred tank reactor. (C) Evolution of the volumes of the stirred tank reactor V_R (^___), the feed (), the collected foam (), and the waste ().

Phase II could be described as a transitory feed phase that aimed at returning to the initial broth volume in the tank reactor, which allowed the management of the foaming capacity of the broth as described at section “Geometric Configuration of the Stirred Tank Reactor”. For experiments performed in a 3-L bioreactor at a dilution rate comprised between 0.05 and 0.1 h^–1, the feed flow rate of phase II comprised between 150 and 300 ml⋅h^–1. As the overflowing foam of phase I corresponded to a volume loss comprised between 300 and 900 ml, the return to the initial broth volume took at least between 1 and 4 h. This duration was increased because the foam overflowing flow rate occurring during phase I also occurred during phase II. To gain time and return faster to the initial broth volume, the feed flow rate of phase II could be increased as long as it corresponded to a value of the dilution rate below the value of the washout rate of the microorganism, i.e., 0.33 h^–1.

Phase III corresponded specifically to the so-called O-CC phase. The feeding strategy was applied in reference to a continuous culture protocol with a given dilution rate as a reference parameter. The originality of the present process came from the outgoing flow rate that was divided into two separate flows: the outgoing flow rate in the waste vessel (F_out) and the foam overflowing flow rate (F_foam). For the setup of a long duration and sustainable foam overflowing process, it was indeed necessary to allow the pumping out of the tank reactor of the broth because it was observed that the volume of the broth in the reactor could increase punctually beyond the broth volume operating point of 3 L. This phenomenon was linked to F_foam that encountered slight variations over time. Pumping out the broth when its volume increased beyond the operating point was found to be an efficient strategy to keep constant the broth volume in the tank. In these conditions, the sum of the overflowing foam and the outgoing flow rates was equal to the feed flow rate.

Modeling of the Biomass During Overflowing-Continuous Culture and Theoretical Aspects. In phase III of the O-CC, the biomass change in the stirred tank reactor could be expressed as follows:

The process conditions of phase III were as follows:

with F_in = F_out + F_foam (4), expressing the constant volume operation. In these conditions, α + β = 1.

The modeling of the biomass extraction by the foam led to the following:

The biomass extraction model in the foam overflow was written with three main assumptions:

X_foam proportional to X_R (a ≤ 1 and b = 0), corresponding to a partial recycling process.

X_foam equal to X_R (a = 1 and b = 0), corresponding to a continuous one, without recycling.

a < 1 and b≠0, corresponding to the foaming process (Guez et al., 2007).

Taking into account equations (2–4), equation (1) could be rewritten as follows:

During phase III, the biomass concentration in the tank was shown to be pseudo constant, so dX_R/dt = 0; and the concentration of biomass in the feed vessel was X=i⁢n0 g⋅L^–1. In steady state, D=Fi⁢nVR, so that equation (1) could be rewritten and the microorganism growth rate could finally be expressed as follows:

For a mean value of the concentration of biomass in the tank, X_Rmean, a mean value of the specific growth rate, μ_mean, could be computed.