Many commercial microbial fermentation processes utilize a fed-batch approach, where a main medium is inoculated with the microbe and after a period of time, is fed with additional nutrients over the course of the bioprocess. This is due to a number of factors. First, the amount of nutrients required to produce the amount of biomass and product in many bioprocesses is simply too large to add into a batch medium due to chemical solubility constraints. Second, many microbes are subject to overflow metabolism, whereby excess substrate in the medium results in the production of overflow metabolites. This can divert flux away from the product, complicate downstream operations, and, at worst, have toxic effects on the strain.
Each microbe’s physiology is different and genetically distinct strains of a single species may have different growth rates, nutrient uptake rates, and pathway fluxes. Moreover, environmental conditions in the fermentor, such as inoculum density, media composition, and temperature setpoints may also impact growth and production kinetics. Thus, in order to maximize pathway flux and minimize overflow metabolites, optimal feed rates must be optimized not only with different strains but also with changing bioprocesses. As the bioprocess and the genetics of the strain are being optimized during development, the consummate feed rates for a strain or process also change in parallel. Therefore, in order to realize the full potential of each strain in evolving processes, it is imperative to have accompanying optimization of feed rates.
There are a number of different strategies that can be used to control the feed rates in a fed-batch bioprocess; the strategy adopted depends on available capabilities and specific microbes and processes.
Each fed-batch fermentation usually begins with a batch medium, which contains the initial nutrients required for microbial growth. Once the carbon in the batch medium is depleted, the fermentation will shift to the fed-batch phase, where a carbon source and additional nutrients are provided via an external feed. This transition can be a timed change, where the feed is initiated at a predetermined time within the fermentation. However, this strategy will result in suboptimal feeding if the strain growth or metabolism differs from the strain used to design the timing of feed initiation.
Transitions to the fed-batch portion of the fermentation can also be initiated using physiological signals from the microbe. There are several hallmarks of carbon source depletion that can be leveraged as physiological triggers to initiate feeding. These include an increase in dissolved oxygen, a decrease in OUR, or an increase in pH. When feeds are programmed to initiate upon these quantifiable signals, the initiation can be timed correctly for each strain.
Fixed feed regimes are predefined in advance of the fermentation. Feeds can be fixed at the same rate for the entirety of the fed-batch portion of the fermentation, or with discrete steps that go up and down in rate over time. Feeds can also increase in a linear or exponential fashion, or incorporate various combinations of different rate architectures throughout the fermentation.
Manual feedback control allows for maintaining the concentration of a substrate within a predetermined range using offline measurements. Fermentation operators take samples at regular intervals, measure the concentration of a media component offline, and make adjustments to the feed rate to maintain the substrate broth concentration within the predefined bounds accordingly.
Dynamic feed regimes, similar to feed triggers, leverage physiological signals from microbes in response to their environment. Microbes have been characterized to have measurable physiological responses to substrates in their environment. Signals that can be leveraged in these dynamic feed schemes include dissolved oxygen (%DO), oxygen uptake rates (OUR), pH, and respiratory quotient (RQ). Feeds can be pre-programmed to cycle on and off in response to these signals or adjust feed rates up and down to maintain a setpoint. Some commonly used organisms, such as Saccharomyces cerevisiae or Escherichia coli, have overflow metabolism profiles that are amenable to more complex feed probing strategies. For these organisms, feed rates are probed down and/or up to assess the likelihood of overflow metabolism and adjust feed rates accordingly.
While some fermentations only use a single nutrient feed, many processes incorporate multiple different feeds into the process. This can be for a variety of reasons. The product may incorporate many different elements or precursors that can be fed and controlled separately from the main carbon source. Alternatively, the carbon source can be adjusted during the fermentation in order to provide maximal flux into the desired product pathway.
Feed strategies can significantly impact the performance of a microbe. Without an optimized feed strategy, microbes may not perform to their full potential and lead to potentially incorrect assessments of strain or process performance. This is further compounded by the fact that both the process and microbes are often being optimized at the same time, leading to changes that can quickly make fixed feed regimes obsolete. While dynamic feed schemes can alleviate many of these challenges, a strategy should be selected carefully based on the physiology of the organism, and in some cases, the product being produced.
Even with a dynamic feed scheme, there can be some challenges. As mentioned above, selecting a scheme well-suited to the physiology of the microbe is important. In addition, when using quantifiable signals as a trigger, it is important to build in safety mechanisms to avoid false signals. For example, DO trends can be noisy at the beginning of a fermentation, and a feed could easily be triggered based on signal noise. Here, safety measures such as timed pauses on feed initiation and using a more reliable signal for the dynamic process can be implemented. Moreover, the development of dynamic feed strategies takes time as factors such as signal magnitude may need to be optimized empirically for each microbe and process. Allocating time to develop a robust dynamic feed scheme may be challenging when there are multiple priorities and constrained bioreactor capacity.
Best practices for feed strategies begin with an assessment of the microbe being used for the process and determining the best feeding strategy for that microbe. Developing a feed strategy with prior knowledge of the microbe's physiology can help ensure that optimal process performance is achieved with the selected strategy.
When developing a feeding strategy, it is also best practice to optimize setpoints empirically. For example, when developing a pH stat for a finely tuned feed, it can be helpful to have an understanding of the threshold that can both trigger a feeding rapidly after substrate exhaustion and avoid false triggers due to noise. If using a fixed feed regime, testing the impact of over- and under-feeding on strain performance can aid in determining the suitability of the method.
Culture’s cloud bioreactors have a number of hardware and software features that enable precise control of fermentations.
Culture offers a variety of feed control schemes. These include standard pre-defined timed feed regimes with fixed, linear, and exponential feed rates. Feeds can also be programmed with automated feed triggers to initiate feeds upon depletion of carbon in the main batch phase. Culture also offers a variety of dynamic feeds including pH-stat, DO-stat, and various pulse feed strategies. At Culture, each reactor is outfitted with continuous off-gas measurements, which enables a number of dynamic feed methods based on measurements including OUR-stat, CER-stat, and RQ-stat. Methods can be also developed on demand.
Each of Culture’s bioreactors has up to five separate feeds available. Each of these feeds is delivered via a peristaltic pump with individual scale feedback control. This results in precise weight-based delivery of feeds so that the amount of feed delivered over the course of the fermentation is as intended. Importantly, the capacity to deliver up to five feeds allows for schemes with auxiliary feeds.
When developing a feed scheme, some factors may need to be optimized empirically. These include the magnitude of signals used to trigger feeds and trials of the impact of over- and under-feeding. Culture’s capacity enables methods to be developed and fine-tuned with a number of different variables.