Chemostat

Applications

Research

Chemostats in research are used for investigations in cell biology, as a source for large volumes of uniform cells or protein. The chemostat is often used to gather steady state data about an organism in order to generate a mathematical model relating to its metabolic processes. Chemostats are also used as microcosms in ecology and evolutionary biology. In the one case, mutation/selection is a nuisance, in the other case, it is the desired process under study. Chemostats can also be used to enrich for specific types of bacterial mutants in culture such as auxotrophs or those that are resistant to antibiotics or bacteriophages for further scientific study. Variations in the dilution rate permit the study of the metabolic strategies pursued by the organisms at different growth rates.

Competition for single and multiple resources, the evolution of resource acquisition and utilization pathways, cross-feeding/symbiosis, antagonism, predation, and competition among predators have all been studied in ecology and evolutionary biology using chemostats.

Industry

Chemostats are frequently used in the industrial manufacturing of ethanol. In this case, several chemostats are used in series, each maintained at decreasing sugar concentrations. The chemostat also serves as an experimental model of continuous cell cultures in the biotechnological industry.

Technical concerns

• Foaming results in overflow with the volume of liquid not exactly constant.
• Some very fragile cells are ruptured during agitation and aeration.
• Cells may grow on the walls or adhere to other surfaces, which may be overcome by treating the glass walls of the vessel with a silane to render them hydrophobic. However, cells will be selected for attachment to the walls since those that do will not be removed from the system. Those bacteria that stick firmly to the walls forming a biofilm are difficult to study under chemostat conditions.
• Mixing may not truly be uniform, upsetting the "static" property of the chemostat.
• Dripping the media into the chamber actually results in small pulses of nutrients and thus oscillations in concentrations, again upsetting the "static" property of the chemostat.
• Bacteria travel upstream quite easily. They will reach the reservoir of sterile medium quickly unless the liquid path is interrupted by an air break in which the medium falls in drops through air.

Continuous efforts to remedy each defect lead to variations on the basic chemostat quite regularly. Examples in the literature are numerous.

• Antifoaming agents are used to suppress foaming.
• Agitation and aeration can be done gently.
• Many approaches have been taken to reduce wall growth
• Various applications use paddles, bubbling, or other mechanisms for mixing
• Dripping can be made less drastic with smaller droplets and larger vessel volumes
• Many improvements target the threat of contamination

Experimental design considerations

Parameter choice and setup

• The steady state concentration of the limiting substrate in the chemostat is independent of the influx concentration. The influx concentration will affect the cell concentration and thus the steady state OD.
• Even though the limiting substrate concentration in the chemostat is usually very low, and is maintained by discrete highly concentrated influx pulses, in practice the temporal variation in the concentration within the chemostat is small (a few percent or less) and can thus be viewed as quasi-steady state.
• The time it takes for the cell density (OD) to converge to a steady-state value (overshoot/undershoot) will often be long (multiple chemostat turnovers), especially when the initial inoculum is large. But, the time can be minimized with proper parameter choice.

Steady state growth

• A chemostat might appear to be in steady state, but mutant strain takeovers can occur continuously, even though they are not detectable by monitoring macro scale parameters like OD or product concentrations.
• The limiting substrate is usually at such low concentrations that it is undetectable. As a result, the concentration of the limiting substrate can vary greatly over time (percentage-wise) as different strains takeover the population, even if resulting changes in OD are too small to detect.
• A "pulsed" chemostat (with very large influx pulses) has a substantially lower selective capacity than a standard quasi-continuous chemostat, for a mutant strain with increased fitness in limiting conditions.
• By abruptly lowering the influx limiting substrate concentration it is possible to temporarily subject the cells to relatively harsher conditions, until the chemostat stabilizes back to the steady state (on the time order of the dilution rate D).

Mutation

• Some types of mutant strains will appear rapidly:
  • If there is a SNP that can increase fitness it should appear in the population after only few chemostat doublings, for characteristically large chemostats (e.g. 10^11 E. coli cells).
  • A strain that requires two specific SNPs where only their combination gives a fitness advantage (whereas each one separately is neutral), is likely to appear only if the target size (the number of different SNP locations that give rise to an advantageous mutation) for each SNP is very large.
• Other types of mutant strains (e.g. two SNPs with a small target size, more SNPs or in smaller chemostats) are highly unlikely to appear.
  • These other mutations are expected only through successive sweeps of mutants with a fitness advantage. One can only expect multiple mutants to arise if each mutation is independently beneficial, and not in cases where the mutations are individually neutral but together advantageous. Successive takeovers are the only reliable way for evolution to proceed in a chemostat.
• The seemingly extreme scenario where we require every possible single SNP to co-exist at least once in the chemostat is actually quite likely. A large chemostat is very likely to reach this state.
• For a large chemostat the expected time until an advantageous mutation occurs to be on the order of the chemostat turnover time. Note, this is usually substantially shorter than the time for an advantageous strain to take over the chemostat population. This is not necessarily so in a small chemostat.
• The above points are expected to be the same across different asexually reproductive species (E. coli, S. cerevisiae, etc.).
• Furthermore, the time until mutation appearance is independent of genome size, but dependent on per-BP mutation rate.
• For characteristically large chemostats, a hyper-mutating strain does not give enough of an advantage to warrant use. Also, it does not have enough of a selective advantage to be expected to always appear through random mutation and take over the chemostat.

Single takeover

• The takeover time is predictable given the relevant strain parameters.
• Different dilution rates selectively favor different mutant strains to take over the chemostat population, if such a strain exists. For example:
  • A fast dilution rate creates a selection pressure for a mutant strain with a raised maximal growth rate;
  • A mid-range dilution rate creates a selection pressure for a mutant strain with a higher affinity to the limiting substrate;
  • A slow dilution rate creates a selection pressure for a mutant strain which can grow in media with no limiting substrate (presumably by consuming a different substrate present in the media);
• The time for takeover of a superior mutant will be quite constant across a range of operation parameters. For characteristic operation values the take over time is on the order of days to weeks.

Successive takeovers

• When the conditions are right (a large enough population, and multiple targets in the genome for simple advantageous mutations) multiple strains are expected to successively takeover the population, and to do so in a relatively timed and paced manner. The timing depends on the type of mutations.
• In a takeover succession, even if the selective improvement of each of the strains stays constant (e.g. each new strain is better than the previous strain by a constant factor) – the takeover rate does not stay constant, but rather diminishes from strain to strain.
• There are cases where successive takeovers occur so rapidly that it is very difficult to differentiate between strains, even when examining allele frequency. Thus, a lineage of multiple takeovers of consecutive strains might appear as the takeover of a single strain with a cohort of mutations.

Variations

Fermentation setups closely related to the chemostats are the turbidostat, the auxostat and the retentostat. In retentostats, culture liquid is also removed from the bioreactor, but a filter retains the biomass. In this case, the biomass concentration increases until the nutrient requirement for biomass maintenance has become equal to the amount of limiting nutrient that can be consumed.