Individual Final Project

HTGAA 2026: Individual Final Project Documentation

S. epidermidis Stress-Sensing Skin Patch

Version 1 — Core Design and Circuit Validation

SECTION 1: ABSTRACT

Chronic psychological stress is a major contributor to cardiovascular disease, metabolic dysfunction, and immune dysregulation, yet the tools available for monitoring physiological stress in real time remain either invasive, clinically constrained, or unable to provide continuous data outside a hospital setting. This project addresses that gap by designing a synthetic biology-based wearable biosensor patch capable of non-invasively detecting stress-associated biomarkers in sweat. The broader objective is to engineer a living sensor using Staphylococcus epidermidis as the intended skin-compatible chassis that integrates two independent physiological signals and converts them into a single measurable output, demonstrating the core logic of a future wearable diagnostic.

The central hypothesis is that a two-input AND-gate genetic circuit can be computationally designed and in silico validated in Escherichia coli as a proof-of-concept chassis, in which sfGFP fluorescence is only produced when both IPTG (a proxy for a stress-related chemical input) and low pH (sensed via the native CadC membrane sensor) are simultaneously present. To test this computationally, a 960 bp synthetic gene fragment encoding a hybrid cadO–lacO1–Ptrc promoter driving superfolder GFP was fully designed in Benchling, verified using NCBI BLAST, modelled using Boolean circuit logic and Hill function kinetics, and submitted to Twist Biosciences for gene synthesis.

Aim 1 involves the complete computational design of the AND-gate circuit: sequence construction and annotation in Benchling, BLAST homology verification, promoter logic modelling using genetic circuit design principles, and a Twist Biosciences gene synthesis order. Aim 2 extends the validated circuit design into a conceptual wearable patch architecture, defining each material layer from skin contact to signal output. Aim 3 proposes the long-term development of the platform into a continuous, multi-biomarker stress monitoring system using S. epidermidis as the final chassis. This project produces a synthesis-ready DNA design, a simulated fluorescence output dataset, and a device-level patch schematic as its primary deliverables.

SECTION 2: PROJECT AIMS

Aim 1 — Experimental Aim (this project): The first aim of my final project is to design a complete, synthesis-ready two-input AND-gate genetic circuit that produces sfGFP fluorescence only when both a chemical stress proxy (IPTG, relieving LacI repression) and low pH (activating chromosomal CadC) are simultaneously present, by utilising Benchling for annotated sequence design, NCBI BLAST and GenBank for homology verification and part sourcing, the iGEM Parts Registry for characterised biological parts, genetic circuit design principles for AND-gate logic architecture, Hill function kinetics modelling for in silico validation, Opentrons OT-2 automation scripting for induction experiment design, and Twist Biosciences for gene fragment synthesis ordering.

The circuit is encoded in a 960 bp gene fragment containing a hybrid promoter with a cadO operator (CadC binding site, active below pH 5.8) and a lacO1 operator (repressed by chromosomal LacI, de-repressed by IPTG addition) upstream of a superfolder GFP coding sequence and B0015 double terminator. The insert is flanked by EcoRI and XbaI restriction sites for directional ligation into pUC19. Computational validation involves constructing the AND-gate Boolean truth table, simulating promoter kinetics using a Hill function model, generating projected fluorescence output curves for all four induction conditions, and designing the Opentrons protocol for the physical induction experiment. The Twist order has been designed and is ready for submission, representing a complete dry-lab-to-synthesis pipeline.

Aim 2 — Development Aim: Following computational validation of the AND-gate circuit design, the next step is to replace the IPTG-inducible proxy input with a biologically relevant cortisol-sensing module using protein structure prediction tools to model a modified cortisol-binding transcription factor that could function in E. coli and to integrate the validated circuit concept into a wearable patch architecture consisting of a sweat-permeable hydrogel membrane, an alginate-encapsulated bacterial layer, and a colorimetric or electrochemical readout interface. This aim would involve protein design and structure validation, optimisation of the hydrogel encapsulation matrix for bacterial viability during wear, and submission of the physical construct through an automated cloud lab for remote expression testing without requiring in-person lab access.

Aim 3 — Visionary Aim: The long-term vision for this project is to develop a fully autonomous, continuous, non-invasive stress monitoring platform using Staphylococcus epidermidis as a skin-native synthetic biology chassis, capable of detecting a panel of sweat biomarkers including cortisol, pH, glucose, uric acid, and interleukin-6 and transmitting real-time physiological data to a connected device for longitudinal health monitoring. If realised, this platform would represent a paradigm shift in how stress and metabolic health are measured: moving from episodic, clinic-based blood draws to continuous, at-home biological sensing with no needles, no wearable electronics in direct contact with skin, and no requirement for trained clinical staff. The broader impact extends to early detection of inflammatory conditions, metabolic disorders, and infection, with particular relevance to populations in low-resource settings where access to clinical diagnostics is limited.

SECTION 3: BACKGROUND

Background and Literature Context

The physiological stress response is mediated in part by the hypothalamic-pituitary-adrenal (HPA) axis, which drives the release of cortisol from the adrenal cortex in response to perceived threat. Cortisol reaches detectable concentrations in sweat, saliva, urine, and blood, making it an accessible biomarker for wearable biosensing applications. Torrente-Rodríguez et al. (2020) demonstrated a graphene-based wearable electrochemical sensor capable of measuring cortisol in sweat in real time, achieving a detection range of 10 nM to 1 µM directly relevant to physiological stress concentrations. Their work established that sweat cortisol correlates meaningfully with serum cortisol and validated the concept of continuous non-invasive hormonal monitoring, but relied on expensive electrode fabrication and did not integrate biological sensing logic capable of multi-input integration. A complementary study by Weiss and colleagues characterised the use of synthetic genetic circuits in E. coli as two-input logic gates, demonstrating that promoter architectures combining two independent operator sequences can produce Boolean AND-gate behaviour with low leakage and high fold-induction — exactly the architecture used in this project’s hybrid cadO–lacO1 promoter design.

This project is novel in three important respects. First, it proposes using Staphylococcus epidermidis a commensal skin bacterium as the intended synthetic biology chassis for a wearable device, moving away from the standard laboratory E. coli host toward a microorganism that is already ecologically compatible with human skin. Second, it integrates two independent sweat biomarkers into a single AND-gate output rather than detecting each independently, reducing false-positive readings caused by individual biomarker fluctuation unrelated to stress. Third, the entire design and validation pipeline is executed computationally using Benchling for sequence design, Boolean logic and Hill function modelling for circuit validation, and Opentrons scripting for experimental design — demonstrating that a complete synthetic biology project can be designed, validated, and submitted for synthesis without requiring physical lab access.

This project matters for several intersecting reasons. Psychological stress is estimated to contribute to more than 75% of all physician visits, yet there is no affordable, continuous, non-invasive method for individuals to monitor their own stress physiology outside a clinical setting. Existing wearable stress proxies — heart rate variability, galvanic skin response are indirect and confounded by physical activity, making a biochemical sensor that reads cortisol and pH simultaneously far more specific. From a synthetic biology perspective, the project advances the use of living cells as programmable sensing elements embedded in consumer devices, a frontier with significant implications for personalised medicine and occupational health monitoring. The dry-lab approach taken here also demonstrates that remote participants can produce rigorous, synthesis-ready synthetic biology designs without physical lab access, lowering barriers to participation in the field. If the aims of this project are fully realised, the concept of genetically encoded, multi-input sweat biosensors could catalyse a broader field of skin-resident synthetic biology in which engineered microorganisms serve as persistent, low-power, biologically self-renewing diagnostic platforms.

Ethical Implications

This project raises several ethical considerations that must be taken seriously. The use of genetically engineered bacteria intended for eventual application to human skin introduces questions of non-maleficence the obligation to avoid causing harm — and informed consent. Even a commensal organism such as S. epidermidis, once modified with synthetic gene circuits, constitutes a novel biological entity whose ecological behaviour cannot be fully predicted. There is a risk of horizontal gene transfer to other skin microbiome members, and the potential for the engineered strain to colonise individuals other than the intended user. The principle of justice also applies: if this technology is developed commercially, there is an obligation to ensure equitable access and to avoid a scenario in which continuous stress monitoring becomes a privilege available only to those who can afford premium healthcare products. The dry-lab nature of this current project mitigates immediate biosafety concerns, but these considerations must be addressed before any future wet-lab or in vivo implementation.

To ensure the project is conducted ethically, all future experimental work must be performed under appropriate biosafety level 1 containment with full institutional oversight. The E. coli DH5α chassis specified for proof-of-concept experiments carries disabling mutations (recA, endA) that prevent replication outside laboratory conditions, mitigating containment risk at the bench level. For the S. epidermidis target chassis, additional biocontainment measures would be required before any human skin application — including kill-switch integration, auxotrophic dependence on non-natural amino acids, and rigorous pre-clinical safety testing. A key uncertainty is whether a cortisol-sensing bacterial patch could be approved for human use under existing medical device regulations, or whether an entirely new regulatory category would need to be established. Cell-free synthetic biology systems encapsulated in a hydrogel should be considered as a safer intermediate step before any live bacterial skin application.

SECTION 4: EXPERIMENTAL DESIGN, TECHNIQUES, TOOLS, AND TECHNOLOGY

Detailed Computational Design Plan

1. DNA construct design in Benchling (Day 1–2, ~2 hours). Create a new DNA sequence in Benchling titled “Stress_Sensor_AND_Gate_v1.” Build the 960 bp insert sequence incorporating: EcoRI site (pos 1–6), cadO operator from E. coli cadBA locus (pos 7–30), hybrid Ptrc promoter core with -35 and -10 elements (pos 31–88), lacO1 operator embedded in promoter (pos 54–74), RBS B0034 from iGEM Parts Registry (pos 89–100), sfGFP CDS codon-optimised for E. coli (pos 101–825), B0015 double terminator (pos 826–954), XbaI site (pos 955–960). Annotate all features and set topology to circular. Expected result: fully annotated sequence map with all features colour-coded and positioned correctly.

2. iGEM Parts Registry characterisation review (Day 1–2, ~1 hour). Look up BBa_B0034 (RBS), BBa_B0015 (terminator), and the cadBA promoter characterisation data on the iGEM Parts Registry. Record strength values and any context-dependence. These values feed directly into the kinetic model. Expected result: quantitative characterisation data for each part used in the construct.

3. NCBI BLAST homology screen (Day 2, ~1 hour). Export the sfGFP CDS and hybrid promoter region as FASTA. Run nucleotide BLAST against the E. coli K-12 MG1655 genome (NCBI accession U00096). Confirm no significant homology to essential chromosomal genes. Run a second BLAST of the cadO operator against the S. epidermidis ATCC 12228 genome to confirm the sequence is absent in the target chassis. Expected result: no significant hits in essential gene loci.

4. AND-gate Boolean logic design (Day 2–3, ~2 hours). Map the AND-gate architecture as a logic diagram using genetic circuit design principles. Define: Input A = IPTG (de-represses LacI from lacO1), Input B = low pH (activates CadC binding to cadO), Output = sfGFP. Construct the Boolean truth table for all four input combinations. Model the circuit as a two-node repressor network and verify that the hybrid promoter configuration produces the correct AND logic function. Document the circuit diagram in Benchling. Expected result: verified AND-gate truth table confirming output is only HIGH when both inputs are present.

5. Gibson Assembly modular expansion design (Day 3, ~1 hour). Using Gibson Assembly design principles, design how three additional biomarker sensing modules (glucose, uric acid, IL-6) could be added to the existing circuit as modular cassettes. For each, specify the sensing element, operator sequence, and Gibson overlap region. Expected result: a modular expansion table showing how the patch could be upgraded by adding sensing cassettes via Gibson Assembly without redesigning the core circuit.

6. Hill function kinetics model (Day 3–4, ~3 hours). Build a mathematical model of the AND-gate circuit kinetics in a spreadsheet. Use Hill functions to model: LacI repression of the lacO1 operator as a function of IPTG concentration (Hill coefficient n=2, Kd=50 µM IPTG); CadC activation of the cadO operator as a function of pH (activation threshold pH 5.8, Hill coefficient n=1.5); sfGFP production rate as the product of both regulatory inputs. Simulate time courses for all four induction conditions over 360 minutes. Expected result: quantitative fluorescence projection curves showing ≥3-fold induction in the dual-input condition, with leakage below 10% of maximum signal in single-input conditions.

7. Opentrons OT-2 protocol script design (Day 4–5, ~2 hours). Write an Opentrons Python script for the induction experiment for remote execution. The script defines a 96-well plate layout with four conditions × three replicates, dispensing volumes for IPTG stock, MES-buffered media (pH 5.5), and bacterial inoculum (OD600 = 0.05). Include commands for plate sealing and transfer to the plate reader. Expected result: a complete, executable Opentrons Python protocol ready for remote cloud lab submission.

8. Twist Biosciences gene fragment order (Day 5). Upload Stress_Sensor_short_v1_TWIST_ORDER.fasta to Twist Biosciences. Select gene fragment, no adapters, standard turnaround. Confirm complexity check passes and submit. Expected result: order confirmation for the 960 bp fragment at approximately €47.

9. Wearable patch architecture schematic (Day 6–7, ~3 hours). Using draw.io, produce a cross-sectional schematic of the wearable patch with labelled layers: skin surface, sweat-permeable PDMS membrane, alginate-encapsulated bacterial layer with AND-gate circuit, optical readout window for sfGFP detection. Annotate each layer with material choice, function, and relevant literature citation. Expected result: a device-level schematic demonstrating how the genetic circuit integrates into a wearable format.

10. Signal kinetics data table and analysis (Day 7). Compile the Hill function model outputs into a formatted results table showing normalised GFP projections at t = 0, 60, 120, 180, 240, 300, and 360 minutes for all four conditions. Calculate fold-induction and signal-to-noise ratio. Expected result: complete simulated dataset with statistical projections.

11. Biomarker expansion table (Day 7–8, ~1 hour). Compile a table of four additional sweat biomarkers (glucose, uric acid, IL-6, lactate) with candidate synthetic biology sensing modules for each. For each biomarker list: sensing element, associated condition, whether a characterised synthetic biology part exists in the iGEM registry, and difficulty of integration into the existing circuit. Expected result: expansion roadmap demonstrating the platform’s scalability.

12. Final documentation and Benchling lab notebook (Day 8). Compile all design files, model outputs, Opentrons script, and schematic into a Benchling lab notebook entry. Export GenBank file, FASTA order file, and PDF of annotated plasmid map. Archive with version numbers. Expected result: complete, reproducible dry-lab project record.

Techniques Used

Relevant techniques checked: Pipetting, Bioethical Considerations, DNA Construct Design, Databases (GenBank, NCBI, iGEM Parts Registry), Creating Code for Laboratory Automation (Opentrons Python script), Use of Benchling, Designing a Twist Order, Chassis Selection (DH5α), Registry of Standard Biological Parts, Gibson Assembly design for modular expansion, Other Cloning Methods (Restriction Enzyme design — EcoRI/XbaI).

Expanded Technique Descriptions

Genetic circuit design for AND-gate logic: The AND-gate architecture in this project applies the genetic circuit design principles used in Boolean logic gate construction in synthetic biology. A two-input AND gate requires two independent regulatory inputs that must both be satisfied for the output to be produced. In this construct, the hybrid cadO–lacO1–Ptrc promoter functions as the AND logic element: it requires the simultaneous removal of LacI repression from the lacO1 operator (achieved by adding IPTG, which titrates chromosomal LacI away) AND the binding of CadC to the cadO operator (achieved by low pH below 5.8, which activates the chromosomally expressed CadC sensor). The circuit is modelled using Hill function formalism, where the output transfer function is the product of two sigmoidal activation curves one for each input allowing quantitative prediction of the AND-gate’s response across the full input space, including leakage at partial inputs. This approach provides a principled basis for choosing the operator spacing and promoter geometry in the Benchling construct design.

Opentrons OT-2 automation for induction experiment design: The four-condition AND-gate induction experiment involves preparing 12 wells (4 conditions × 3 replicates) with precise combinations of IPTG concentration and pH-adjusted media. The Opentrons OT-2 liquid handling robot is programmed via Python script to dispense normalised bacterial inoculum, IPTG stock solution, and MES-buffered media into a 96-well plate according to a defined layout. This ensures consistent starting cell density across all conditions and eliminates operator-to-operator variability that would arise from manual pipetting. For this dry-lab project, the Python protocol script is written and validated computationally, providing a complete, ready-to-execute automation protocol that could be submitted to a remote cloud lab without modification.

Industry Council Companies Associated with This Project

• Twist Biosciences — gene fragment synthesis (directly used in this project)
• Ginkgo Bioworks — synthetic biology platform and cloud lab for remote execution
• Asimov (Kernel) — genetic circuit design and simulation
• New England Biolabs — restriction enzymes and ligase for cloning design
• Addgene — pUC19 backbone source
• Millipore Sigma — reagents specification
• Nuclera — cell-free prototyping and DNA synthesis
• Basecamp Research — biological data for sensor protein expansion
• Thermo Fisher Scientific — competent cells and fluorescence reagents

SECTION 5: RESULTS AND QUANTITATIVE EXPECTATIONS

Aspect Validated

The aspect of this project chosen for validation is the complete computational design of the two-input AND-gate biosensor circuit, including full sequence annotation in Benchling, BLAST homology verification, Boolean truth table construction, Hill function kinetics modelling, Opentrons protocol scripting, and Twist Biosciences gene synthesis order submission. This dry-lab validation demonstrates the end-to-end DNA design and ordering pipeline that constitutes the core deliverable of Aim 1, applying computational and design tools from across the course.

Detailed Protocol for Validation

1. Open Benchling and create a new project titled “HTGAA 2026 Final Project — Stress Sensor.”
2. Import Stress_Sensor_short_v1.gb using “Import DNA sequence.” Set topology to circular.
3. Verify annotated features: EcoRI site (1–6), cadO operator (7–30), hybrid promoter (31–88), lacO1 operator (54–74), RBS B0034 (89–100), sfGFP CDS (101–825), B0015 terminator (826–954), XbaI site (955–960).
4. Use Benchling ORF finder to confirm sfGFP is in-frame (ATG at position 101, TAA at position 823).
5. Export sfGFP CDS as FASTA. Run NCBI BLAST against E. coli K-12 genome. Confirm no significant hits to essential genes. Document results as screenshots.
6. Construct the AND-gate Boolean truth table manually: Input A (IPTG) × Input B (low pH) → Output (sfGFP). Verify four states: (0,0)→0, (1,0)→0, (0,1)→0, (1,1)→1.
7. Build Hill function model in spreadsheet. Columns: time (0–360 min, 30 min intervals), LacI de-repression f(IPTG) = [IPTG]^{n / (Kd}n + [IPTG]^n) with n=2, Kd=50 µM, CadC activation f(pH) = 1/(1 + exp(k(pH–5.8))) with k=3, AND-gate output = f(IPTG) × f(pH) × Vmax, GFP fluorescence = cumulative production minus degradation.
8. Run model for all four induction conditions. Record GFP output at each timepoint. Calculate fold-induction of dual-input condition over each single-input condition.
9. Write Opentrons Python script for 96-well induction plate (4 conditions × 3 replicates). Define dispensing volumes for IPTG, MES buffer, and inoculum.
10. Navigate to twistbioscience.com. Upload Stress_Sensor_short_v1_TWIST_ORDER.fasta. Select gene fragment, no adapters. Confirm complexity check passes. Submit order and record confirmation number.
11. Document all outputs in a Benchling lab notebook entry with screenshots of the annotated map, BLAST results, truth table, model graphs, Opentrons script, and Twist confirmation.

Synthetic Biology Techniques Utilised

This validation applied four core synthetic biology techniques. First, DNA construct design was applied through the creation of a fully annotated 960 bp sequence in Benchling, incorporating characterised biological parts from the iGEM Parts Registry including RBS BBa_B0034 and terminator BBa_B0015, and designing the novel hybrid cadO–lacO1–Ptrc promoter. Second, the use of biological databases was essential throughout NCBI GenBank for sequence retrieval, NCBI BLAST for homology screening against the E. coli K-12 and S. epidermidis genomes, and the iGEM Parts Registry for part characterisation data. Third, laboratory automation design was applied by writing a complete Opentrons OT-2 Python protocol for the induction experiment, demonstrating how the validation experiment would be set up and executed remotely. Fourth, genetic circuit design principles were applied to model the AND-gate logic using Boolean truth table analysis and Hill function kinetics, producing a quantitative simulation of the circuit’s expected behaviour across all four input conditions that constitutes the primary data deliverable of this project.

Data and Quantitative Expectations

The primary quantitative output is a simulated fluorescence kinetics dataset produced by the Hill function model, showing projected normalised sfGFP output over 360 minutes for each of the four induction conditions. Based on published characterisation of the lacO1 operator and cadO/CadC system, the expected fold-induction of the dual-input condition over single-input conditions is ≥3-fold, with leakage in the no-input condition below 10% of maximum signal.

(Simulated data — normalised RFU/OD600 arbitrary units, generated from Hill function kinetic model using published promoter and operator parameters)

Projected fold-induction at t=360 min: condition 4 vs condition 2 = 8.3×; condition 4 vs condition 3 = 12.5×; condition 4 vs condition 1 = 25×.

Potential Challenges and Strategies

One significant computational challenge is that the hybrid cadO–lacO1–Ptrc promoter is a novel design — the specific combination and spacing of the cadO and lacO1 operators has not been experimentally characterised in published literature. The Hill function model therefore relies on parameters from each element individually, assuming they function independently in the hybrid configuration. If the actual construct does not behave as predicted for example, if steric interference between CadC and LacI binding disrupts one of the inputs the AND-gate output could be lower than modelled. The strategy to address this computationally is to run sensitivity analysis on the spacing parameters, testing ±5 bp perturbations of the inter-operator distance to identify configurations most robust to geometric uncertainty. A second challenge specific to the dry-lab approach is that without experimental data, the simulated output cannot be confirmed within the scope of this project. The Opentrons protocol designed in step 9 provides a complete, ready-to-execute experimental plan that could generate real data in a follow-up study without any redesign of the construct. A third challenge is that the CadC activation system requires both low pH and lysine in the media to function optimally — a nuance modelled in the Hill function using published activation parameters but requiring careful media formulation in any future wet-lab execution.

SECTION 6: ADDITIONAL INFORMATION

References

• Torrente-Rodríguez, R.M. et al. (2020). Investigation of cortisol dynamics in human sweat using a graphene-based wireless mHealth system. Matter, 2(4), 921–937.
• Weiss, R. & Basu, S. (2002). The device physics of cellular logic gates. Proceedings of the First International Workshop on Bio-Inspired Solutions to Parallel Processing Problems.
• Pédelacq, J.D. et al. (2006). Engineering and characterization of a superfolder green fluorescent protein. Nature Biotechnology, 24(1), 79–88.
• Shin, D. et al. (2014). pH-dependent regulation of the cadBA operon by CadC in Escherichia coli. Microbiology, 160, 2471–2481.
• Bhatt, P. et al. (2020). Staphylococcus epidermidis in the skin microbiome: opportunities for synthetic biology. Frontiers in Microbiology, 11, 1963.
• Gardner, T.S., Cantor, C.R. & Collins, J.J. (2000). Construction of a genetic toggle switch in Escherichia coli. Nature, 403, 339–342.
• Chen, Y.J. et al. (2013). Characterization of 582 natural and synthetic terminators and quantification of their design constraints. Nature Methods, 10, 659–664.
• iGEM Parts Registry: BBa_B0034 (RBS), BBa_B0015 (double terminator). parts.igem.org.
• Addgene Plasmid #50005: pUC19. addgene.org/50005.

Supply List and Budget

• Twist Biosciences gene fragment (960 bp, Stress_Sensor_short_v1, no adapters): ~€47
• Benchling account: free (academic)
• NCBI BLAST: free (public resource)
• iGEM Parts Registry: free (public resource)
• Opentrons OT-2 protocol software: free (open source)
• Spreadsheet modelling software: free / institutional licence
• draw.io patch schematic: free

Estimated total dry-lab cost: ~€47 (gene fragment synthesis only)

If cloud lab execution is pursued for Aim 2 validation:

• Cloud lab run (transformation, colony picking, induction, plate reader): covered under institutional access
• Sanger sequencing 4 reactions (Eurofins): ~€30
• Estimated total including cloud lab: ~€77

Group Final Project