Week 1 HW: Final Project Ideas

Living pH Homeostat

An Engineered Auxiliary Microbe for Autonomous Fermentation pH Stabilization

Concept proposal • Fall 2026

Figure 1. The concept in one picture. The auxiliary “homeostat” strain lives dispersed throughout the vessel, so it meets local pH excursions where they actually happen, rather than where the single bulk probe is mounted.

1. What would I like to biomanufacture, and why?

I would like to biomanufacture a living pH-control module: an engineered microbial strain that acts as an auxiliary population inside a fermentation. Its job would not be to make the commercial product. Instead, it would sense when extracellular pH moves outside an intended operating band and activate metabolism that pushes the environment back toward that band.

The motivation is scale-up. Large bioreactors are not perfectly homogeneous. Mixing becomes slower as volume increases, and cells can pass through short-lived local regions with different pH, dissolved oxygen, or substrate concentrations. Published scale-down studies have recreated pH differences of up to about 0.9 pH units, and CHO studies have shown that repeated excursions as small as 0.4 pH units can alter cell physiology and process performance [2–4].

Today, pH is normally controlled by a probe, controller, and acid/base addition system. My proposal is not to replace that safety-critical hardware initially. The near-term goal is to add a biological layer that can damp sustained or spatial pH excursions from within the culture itself. The long-term question is whether distributed living controllers could reduce the severity of local pH hotspots that a single bulk probe may not capture.

This is the distinction drawn in Figure 1: one probe reports a single bulk value, while a dispersed sentinel population is present inside the zones the probe cannot see.

2. What living system would I use?

Initial chassis: engineered Escherichia coli. E. coli is attractive for a first prototype because its genetics are well characterized, it grows quickly, and published synthetic pH-control circuits already exist in this organism [1,5].

The biological design would contain three functional layers:

1. pH sensing — promoters or regulatory elements whose activity changes when extracellular pH becomes too acidic or too alkaline.
2. Decision threshold — the circuit remains largely inactive inside the acceptable pH range and turns on only when a threshold is crossed.
3. Actuation — the activated pathway generates an acidifying or alkalizing effect through cell metabolism or an expressed enzyme.

Figure 2. The control logic as three states. The middle panel is the one that matters most for metabolic burden: inside the acceptable band both sensors are inactive and the circuit does close to nothing.

flowchart LR
  A["Extracellular pH<br/>excursion"] --> B["1. Sensing<br/>pH-responsive promoters"]
  B --> C["2. Decision threshold<br/>silent inside the operating band"]
  C --> D["3. Actuation<br/>acidifying / alkalizing metabolism"]
  D -->|"pushes pH back<br/>toward the band"| A

Published precedent

The key precedent is Li et al. (2020), who built a Genetic pH Shooting (GPS) system in E. coli by coupling pH-responsive promoters to acid- or base-producing genes. Their engineered strains could self-adjust extracellular pH, and the combined circuit was tested during lycopene fermentation at 3 L working volume. In that study, the acid-responsive promoter was mainly active around pH 5–6 and the alkaline-responsive promoter around pH 8–10; therefore, the paper does not demonstrate precision control around pH 7 ± 0.05 [1].

A 2026 study provides a second proof that engineered bacteria can actively modulate extracellular pH. Kuang et al. placed urease under light control in E. coli; urease generated ammonia from urea to raise pH, while native metabolism drove re-acidification when urease activity fell. The system could repeatedly switch extracellular pH states over multiple cycles [5].

3. How would I use automation?

Automation would be used mainly for design-build-test-learn screening rather than for the final large-scale pH control loop. A liquid-handling robot (for example, an Opentrons-type platform) could prepare a matrix of media conditions and promoter/circuit variants across 96-well plates, standardize inoculation, and perform repeated sampling. Plate-reader measurements of growth and reporter output, combined with direct pH measurements, would identify circuits with useful thresholds and low background activity.

flowchart TD
  S1["1. Characterize sensing alone<br/>across a pH series"] --> S2["2. Test one-sided actuation<br/>recovery from an acidic excursion"]
  S2 --> S3["3. Combine acid- and<br/>base-correcting functions"]
  S3 --> S4["4. Test as low-abundance co-culture<br/>with a production organism"]

Running the stages in this order reduces the chance of confusing sensing failure, actuator failure, and co-culture interactions with one another.

Figure 3. Where automation earns its place. Stages 2 and 3 are the throughput bottleneck, and stage 5 is where the co-culture assumption in the callout above finally gets tested.

4. How much would I need to make, and how long would it take?

4a. Meaningful prototype scale

For Homework #1, I define a meaningful first scale as enough auxiliary cells to inoculate a 10 L pilot-scale fermentation and maintain a small sentinel fraction for a 20 h run. There is no published value showing the minimum helper-cell fraction required for this architecture, so the following calculation is explicitly a design assumption rather than an established biological requirement.

4b. How long?

Under optimal rich-medium laboratory conditions, E. coli can double in about 20 min and an overnight culture can exceed 10⁹ CFU/mL [6]. If a seed started with 10⁷ viable cells, reaching 10¹¹ cells would require about 14 doublings: 14 × 20 min ≈ 4.7 h under ideal exponential growth.

That is a theoretical lower bound. Recombinant expression, lag phase, media limitations, and the pH-control circuit may slow growth, so a practical plan would allocate an overnight expansion (about 12–16 h) to prepare the helper inoculum. The 20 h production-run duration is a project design assumption, not a claim from the cited literature.

5. What do I not know yet?

1. Can the controller respond near the desired setpoint with useful resolution? The strongest published GPS demonstration used broad acidic and alkaline response windows. It does not establish ±0.05 pH control around neutral pH. New sensing elements, promoter engineering, or a deliberately wider acceptable band may be required.
2. Will the feedback dynamics be stable? Gene expression and metabolism are slower than an electronic pH controller. Delay between sensing and chemical correction could cause weak response, overshoot, or oscillation. The 2026 optogenetic work specifically used enzyme degradation to improve switching dynamics, showing that response timing is a real engineering variable [5].
3. Can an auxiliary strain control pH without damaging the production process? The helper will consume nutrients and may secrete metabolites beyond the intended actuator. Its abundance, growth rate, and metabolic burden must be constrained, and its effect on product yield and quality must be measured. This is the largest unproven assumption in the co-culture version.

10-week feasibility boundary

A complete industrial pH homeostat is not a realistic 10-week deliverable. A defensible 10-week demonstration would be narrower: show that an engineered E. coli circuit can (i) report a defined pH excursion and (ii) produce a measurable one-sided extracellular pH correction in a small, controlled culture.

A stretch goal would be to test the same circuit at low abundance in a simple two-strain co-culture and quantify whether it changes the partner strain’s growth. Precise neutral-pH control, industrial-scale gradient correction, and long-term co-culture stability should remain future research questions.

Assumptions used in this proposal

• A separate auxiliary pH-control strain can eventually be made compatible with a production strain. This is a hypothesis, not demonstrated by the cited GPS paper.
• A 1% helper-to-producer cell ratio is used only for the back-of-envelope manufacturing calculation; the functional minimum ratio is unknown.
• A production-culture density of 10⁹ cells/mL is used as an order-of-magnitude prototype assumption and will vary strongly by host and process.
• The illustrative 10 L, 20 h fermentation is a chosen prototype scale, not a literature-derived requirement.
• The first prototype aims to damp meaningful pH excursions, not maintain pH 7.00 ± 0.05.
• Conventional pH probes and acid/base addition remain the safety/reference control in early experiments.

References

1. Li C, Gao X, Peng X, et al. Intelligent microbial cell factory with genetic pH shooting (GPS) for cell self-responsive base/acid regulation. Bioresources and Bioprocessing. 2020. PMID: 33138821. https://doi.org/10.1186/s40643-020-00348-8
2. Cortés JT, Flores N, Bolívar F, Lara AR, Ramírez OT. Physiological effects of pH gradients on Escherichia coli during plasmid DNA production. Biotechnology and Bioengineering. 2016;113(3):598–611. https://doi.org/10.1002/bit.25817
3. Arulrajah P, Lievonen AE, Subaşı D, Pagal S, Weuster-Botz D, Heins A-L. Scale-down bioreactors — comparative analysis of configurations. Bioprocess and Biosystems Engineering. 2025;48(10):1619–1635. https://doi.org/10.1007/s00449-025-03182-w
4. Paul K, Böttinger K, Mitic BM, et al. Development, characterization, and application of a 2-Compartment system to investigate the impact of pH inhomogeneities in large-scale CHO-based processes. Engineering in Life Sciences. 2020;20(8):368–378. https://doi.org/10.1002/elsc.202000009
5. Kuang J, Armendarez OJ, Chang W-T, et al. Engineering an Optogenetic pH-Modulator in Bacteria. Advanced Science. 2026;13(42):e24319. https://doi.org/10.1002/advs.202524319
6. Green MR, Sambrook J. Growth and Maintenance of Escherichia coli Laboratory Strains. Current Protocols. 2021. PMCID: PMC8006063.

AI use disclosure

AI client: OpenAI ChatGPT (GPT-5.6 Sol). AI was used to critique the concept, identify relevant published precedents, distinguish evidence from assumptions, and draft/structure this Homework #1 response. Literature claims were checked against the cited papers or PubMed/PMC records.

Prompt used (abridged): “Prepare the Homework #1 document for an engineered pH homeostat. Keep statements scientific and technically meaningful, ground facts and thoughts in published research papers, avoid wild assumptions, and explicitly list every assumption used.”

Final Project