Week 7 HW: Genetic Circuit Part II: Neuromorphic Circuits
Part 1: Intracellular Artificial Neural Networks (IANNs)
- Advantages IANNs have over traditional genetic circuits
There are several advantages: 1) IANN processing uses graded values and the output is analog, just like continuous biochemical signals being graded and variable, not digital on/off; 2) adaptability of the circuit design as any function can be designed and simulated with changed weights instead of rebuilding circuit topology; 3) weights in IANN correspond directly to a real biological interactions (activation/repression of a gene); 4) IANNs more efficiently use limited cellular resources (weights are tuned instead of gates chained).
- Describe a useful application for an IANN; include a detailed description of input/output behavior, as well as any limitations an IANN might face to achieve your goal.
There is a recent study (Hu, Sun, Qi, 2025) that reveals a mechanistic link between α-Synuclein and mitochondrial stress in Parkinson’s disease and a self-reinforcing pathogenic loop that is involved in the disease progression, and so an IANN could be useful to build on their results.
The paper showed that αSyn (its special NAC domain that targets αSyn to mitochondria) binds the mitochondrial matrix protease ClpP, which normally counteracts α-Syn pathology by stabilizing the native tetramer of α-Syn. This interaction blocks ClpP activity and causes mitochondrial proteotoxic stress, and the loss of ClpP function in turn leads to further αSyn aggregation, which forms a self-reinforcing pathogenic loop. The authors also developed a decoy peptide (CS2) that binds the NAC domain of α-Syn, prevents the interaction between α-Syn and ClpP, and restores ClpP function. Importantly, CS2 reduced α-Syn pathology and neurotoxicity in primary neurons inoculated with αSyn fibrils, PD patient iPSC-derived dopaminergic neurons, and an A53T α-Syn transgenic mouse model.
An IANN can be designed to sense pathogenic processes and produce the decoy peptide as therapy on demand. The circuit is integrates three analog inputs: 1) a proximity sensor (positive weight) that reports interaction between αSyn and ClpP (could be built with one split-reporter fragment at the N- or C-terminus of α-Syn, leaving the NAC domain free, and the other on ClpP; 2) a ClpP-activity reporter (negative weight) that signals functional protease and the system that could still be self-correcting; 3) a mitochondrial-stress reporter (positive weight). The weighted sum passed through a nonlinear activation function drives the expression of the decoy peptide CS2. A functional ClpP freed from α-Syn would restabilize the αSyn tetramer and lower the interaction signal; CS2 output increases with pathology and falls as the decoy peptide frees the ClpP protease.
Perceptron form: y= w₁x₁ - w₂x₂ + w₃x₃ + b; output drives expression of CS2. Low severity of the pathology -> y low -> the output σ(y) low -> low CS2. Rising αSyn–ClpP interaction (x1 increases), poor ClpP activity (x2 decreases), and increasing mitochondrial stress (x3 increases) increase the y and the output -> CS2 expression increases proportionally -> CS2 binds the NAC domain of α-Syn -> α-Syn cannot bind to ClpP -> ClpP disinhibited -> native α-Syn tetramer restabilized -> the proximity sensor signal (x1) decreases -> output decreases. The limitations include sensor sensitivity and proper organization of the circuit in the correct compartment. Although it’s relatively clear how the sensor should be designed, the sensor would be complex and may not be specific enough to the pathological α-Syn interaction with ClpP. As for compartmentalization, the output must work in the mitochondrial matrix, where the input event occurs, while it would be synthesized outside mitochondria, and so careful engineering is required.
- A diagram for an intracellular multilayer perceptron where layer 1 outputs an endoribonuclease that regulates a fluorescent protein output in layer 2.
Part 2: Fungal Materials
What are some examples of existing fungal materials and what are they used for? What are their advantages and disadvantages over traditional counterparts?
- packaging alternatives: custom-shaped protective packaging for agricultural waste as a replacement for expanded polystyrene foam, Styrofoam (the advantage is energy efficiency because the packaging is produced at room temperature and the packaging is compostable)
- leather alternatives: used to produce fashion products (the advantage is no animal-welfare and ecological concerns)
- sound-absorbing alternatives: used as acoustic insulation and decorative elements (the advantage is natural appearance, that density and thickness can be tuned during mycelium growth to filter particular frequencies).
What might you want to genetically engineer fungi to do and why? What are the advantages of doing synthetic biology in fungi as opposed to bacteria?
I’d engineer grown-to-shape, biodegradable, single-trip mycelium cradles for museum-to-museum loans and travelling exhibitions. The art or historical objects inside the cradles would be transported as usual, in climate-controlled vehicles. Fingi could be engineered to over-express hydrophobins, uniformly through the material, and to be a poor host for mold. The cradles can compete with the currently used carved Ethafoam because of their custom fit and environmental sustainability. The fungi could also be engineered to pass archival-safety testing.
The cradles could be produced ahead of a planned shipment of art objects; for that, they would be grown against a 3D-printed or cast replica. Also, mycelium can be grown for an artifact-supporting and moving assisting object. Cradles would be custom fit, low cost, low carbon compared to petroleum foam, and compostable.
The advantage would be that fungi already make the material that is fibrous, self-assembling, and has an architecture resistant to load, while bacteria do not. Also, the grown-to-shape feature is native to fungal growth.