Proof of concept
Hardware
Setup and Goal
To evaluate the functionality of our custom-built chemostat system, we designed and performed a proof-of-concept experiment aimed at demonstrating its ability to sustain continuous cultivation and enable in situ induction of gene expression. As a model system, we employed Escherichia coli DH10B Marionette, a strain engineered to respond to small-molecule inducers with tuneable promoter systems. The strain carried a Level 1 plasmid expressing superfolder GFP (sfGFP) under control of a vanillin-inducible promoter (EF-pVAN-sfGFP) (see plasmid, Figure 1).
Figure 1: Plasmid of EF-pVAN-sfGFP
Our goal was to confirm that the chemostat could maintain bacterial growth under steady-state conditions while allowing controlled addition of an inducer to modulate gene expression during the run. Vanillic acid served as the inducer, activating the pVAN promoter and thereby triggering sfGFP production. The experiment was conducted over a total of 51 hours to monitor system stability, growth dynamics, and induction response.
The test runs not only provided insight into the operational reliability of our DIY chemostat design, but also established a framework for future optimization of continuous culture conditions and dynamic induction experiments.
First Test Run
Procedure:
We used M9 minimal media (filter sterilized) with the following content:
- 2L M9 5x stock
- 10 mL 100 mM CaCl2
- 100 mL trace salts 100x stock
- 3.1 mM FeCl3-6H2O
- 10 mL 1 M MgSO4
- 450 mL of 20% (w/v) glucose
- 10 mL of kanamycin 50µg/mL
- added Milli-Q to 10 L
After having set up the chemostat, we commenced the run on 28.09.2025 at 3:00 p.m., and upon heating to a working temperature of 37°C, 5mL of E. coli DH10B Marionette with an OD600 of 0.4 was added. Our inflow rate was 1 mL/ min with 200 mL working volume. For a few hours, the in- and outflow were turned off and afterwards set to ~1 mL/ min for the rest of the run.
On September 30th, starting at 3:56 p.m., for a time duration of two hours we spiked in vanillic acid, dissolved in 25/75 ethanol/H2O, was added to maintain concentration of 50µM in the chemostat.
Results
Table 1: OD600 of E. coli cells during the run of the chemostat.
| Date / Time | OD600 |
|---|---|
| 29.09.2025 / 12:30 | 0.645 |
| 29.09.2025 / 13:30 | 0.295 |
| 29.09.2025 / 17:10 | 0.498 |
| 29.09.2025 / 21:50 | 0.345 |
| 30.09.2025 / 10:58 | 0.203 |
These measurements were taken by taking a sample from the chemostat and measuring OD600 with a 1 mL cuvette taking a 1 mL cuvette with just media as a blank in an Implen Nanophotometer NP80.
Table 2: Measurements of OD600 and absorption at 488nm after addition of vanillic acid.
| Time (h) | A488 | OD600 | A488 / OD600 |
|---|---|---|---|
| 0 (pre-induction) | 0.238 | 0.175 | 1.36 |
| 0.25 | 0.123 | 0.197 | 0.624 |
| 0.5 | 0.147 | 0.095 | 1.547 |
| 1 | 0.146 | 0.098 | 1.49 |
| 1.5 | 0.102 | 0.055 | 1.855 |
| 2 | 0.089 | 0.093 | 0.957 |
These measurements were taken by taking a sample from the chemostat and measuring OD600 and absorption at 488nm with a 1 mL cuvette taking a 1 mL cuvette with just media as a blank in an Implen Nanophotometer NP80. From that the relative sfGFP levels were calculated.
Figure 2: Relative levels of sfGFP after adding inducer (vanillic acid) over the course of 2 hours. Calculated by taking absorption at 488nm divided by OD600 of sample.
Figure 3: Sensor data collected during the chemostat run using the Stream Chemostat Controller software: 3.1 Temperature over time (°C). 3.2 pH over time. 3.3 Optical density (OD) over time. 3.4 CO₂ concentration over time (ppm). 3.5 Gas flow rate over time (L/h). 3.6 sfGFP fluorescence intensity measured at 500 nm.
Discussion
Throughout the chemostat test run, the optical density (OD₆₀₀) remained rather low. This suggests that the dilution rate was likely too high, causing cells to be flushed out faster than they could replicate. As a result, the biomass concentration never reached a stable steady state suitable for sustained induction. We therefore recommend waiting for the culture to reach a higher OD before commencing the experiment, or maintaining a lower flow rate during future experiments and only increasing the inflow once a higher OD has been established. However, despite these experimental shortcomings, we established that the chemostat was able to sustain growth and is suitable for continuous bioproduction. We observed some fluctuations in the sensor reading, which could be attributed to the lower-grade quality of the sensors. Despite this, during the course of the run the readings remained relatively constant, showing the reliability of our chemostat. No significant increase in sfGFP fluorescence was observed following the addition of vanillic acid. This lack of induction response can likely in part be attributed to the low biomass levels, which limited the number of cells capable of expressing the reporter protein. Additionally, the inducer concentration may have been insufficient to trigger strong promoter activation under these conditions. While we based the vanillic acid concentration on Meyer et al. 2019, who reported half-maximal induction at 26 µM for plasmid-based sensor, later information indicated that our specific promoter reaches maximal activation at approximately 100 mM (iGEM Part:BBa_K3317006). Increasing the inducer concentration could therefore yield more pronounced fluorescence in subsequent runs.
Modifications to the growth medium to enhance nutrient availability and cell growth could improve culture stability and increase the reproducibility of induction results.
Second Test Run
With the problems detected in the first test run, we went back into the lab to improve our results. When researching again, we noticed a crucial detail we missed before: the strain we use for our trial run is aucotrophic for Leucine, which means we need to supply Leucine in the M9 media to achieve growth. So we changed the composition of the M9 medium by adding 0.5mM L-Leucine. The rest of the parameters were kept the same as in the first run. This yielded constant stable growth, as depicted by the stable optical density measurements in Figure 4.
Figure 4: Optical density measurements of the second trial run
References
Escherichia coli “Marionette” strains with 12 highly optimized small-molecule sensors | Nature Chemical Biology. (n.d.). Retrieved October 7, 2025, from https://www.nature.com/articles/s41589-018-0168-3