Turbidistat design

Measurements of in vivo properties of synthetic genetic circuits requires a stable, repeatable cell environment. The turbidistat is one way of achieving this repeatable cell behavior. In a turbidistat, cells grow in rich medium at normal growth rates, limited only by their own ability to replicate. Chemostats, in contrast, limit cell growth rate by using a limiting concentration of some nutrient, and thus exhibit physiological changes characteristic of nutrient starvation.

Turbidistats measure the cell density using the optical density of the medium, and performing dilution of the cell volume to maintain constant optical density. For exponential growth experiments, the optical density range is in the 0.1 to 0.5 OD range. For stationary growth phase measurements, the optical density is in the 0.5 to 3.0 range. Rarely are experiments conducted outside of these ranges.

Temperature should be controllable in a range from 18 to 42 C, although below-room-temperature operation is rarely necessary.

pH control is likely unnecessary in this environment, since at OD 0.5, cells are not metabolizing a significant fraction of the medium composition. At OD 3.0, this might become an issue, and is worth thinking about. Likely constant pH could be achieved with correct buffer composition in the medium.

Under good growth conditions, the doubling time of E. coli cells is between 15 minutes and an hour, typically around 30 minutes, although this is controlled by many factors such as the medium composition, temperature, strain, antibiotic, and plasmid. Since the closed loop feedback maintains constant OD readings, the doubling rate of cultures can be directly observed, and is an important property to record.

With a small volume (360 ul) reactor, the flow rate of new nutrient is quite low. For a nominal 30 minute doubling time, the volume is replaced each hour, so the flow rate is approximately 6 ul/minute, or 100 nl/second. Delivery volumes can be quantized, which will produce some variation in the OD measurement. With a 3.6 ul delivery volume, these variations are held to the 1% level. This is insignificant with respect to the medium chemistry, but will affect fluorescent measurements. These measurements can be compensated for computationally by recording fluorescence/OD values.

A typical use of the turbidistat system is the measurement of the input/output relationship of some genetic circuit. Typically this is done by varying some input parameter, such as concentration of an inducer chemical, and observing the behavior of the system. Common inducers include IPTG, aTc, homo-serine lactones, and arabinose. These are supplied in a wide range of concentrations, typically over 3-4 orders of magnitude. Performing dilutions of these chemicals to the correct level as input to the system may be significant challenge, although multi-stage mixing (serial dilution) can likely achieve this result easily.

In most constructs, the inducer modulates the activity of a transcriptional promoter, and the resulting transcriptional activity (measured in POPS) is the input to the system. This input is typically used to drive a reporter gene, such as yellow fluorescent protein (YFP) and to drive the input to a test circuit. The output of the circuit is similarly in the form of transcriptional events per second (POPS) and is used to drive a second reporter gene such as cyan fluorescent protein (CFP) or red fluorescent protein (RFP). Simultaneous measurements of the fluorescence of YFP and RFP provides an indirect measure of the input and output POPS of the test circuit. By varying the inducer level, a curve tracer can be effectively constructed.

Defects in this scheme include assumptions about the constant degradation rate of the fluorescent proteins, and the similar folding and fluorophor formation rates of the fluorescent proteins. Some of these problems can be eliminated by interchanging the YFP and RFP genes in the test system, and measuring the nominally identical system a second time.

Our typical experiments to date have not dynamically varied the inducer concentrations, but rather relied on static measurements. With an ability to dynamically vary inducer concentrations, the dynamics of these circuits can also be probed. An innovative approach is the use of system identification techniques, which inject noise into the test circuit, and correlate the output with the input noise signal, providing information about the magnitude and phase of the frequency response of the test system.

Simple system:
 * turbidistat with constant OD control from 0.1 to 3.0
 * temperature control at a constant 37C
 * measurement of YFP and RFP fluorescence

More complex system:
 * add serial dilution capability for dynamic changes in inducer concentrations over 3-4 orders of magnitude
 * allow sampling of the output cultures by snap freezing or cooling of reactor overflow volumes in a microtiter plate
 * allow many simultaneous cultures with different culture media

More advanced experiments for the future might include evolutionary experiments, where cultures are grown for many weeks, months, or years. For these experiments, sterility is paramount, and backup cultures taken from overflows would be necessary, both for restart from failure or contamination, and for post-run analysis of when and where evolutionary changes arose. For these long range experiments, biofilm growth on the reactor sidewalls may be an important issues, which could be dealt with by switching chambers, rather than by attempting to eliminate the problem.

OD measurements can be made cancelling out the biofilm scattering by measuriing the OD through two different path lengths, such as in a differential measurement between two emitters and detectors which are arranged in a "butterfly" configuration, measuring the straight and diagonal OD of the liquid.

References: Reference are scanned and available by searching in this bibliography and following the "call number" links,  although you will  have to contact me (TK) for the user name/password. See especially the list of references on turbidistats in the Markx paper.


 * 1) Novick50 Description of the chemostat, Science 112:715-6 (1950)
 * 2) Novick50a Experiments with the chemostat on spontaneous mutations of bacteria, PNAS 36:708-19 (1950)
 * 3) Markx91 The permittistat: a novel type of turbidostat, J Gen Microbiol 137:735-43 (1991)