Proposal for a Logic-Gated Neuro-Diagnostic Platform

What would I like to biomanufacture and why?

I want to biomanufacture a low-cost, point-of-care microfluidic diagnostic chip pre-loaded with freeze-dried biological sensors (split-reporter proteins). The chip tests a single drop of blood plasma to give a clear readout for chronic neurodegenerative diseases like Alzheimer’s or Parkinson’s.My motivation comes from both my personal life and my career direction: Personal Driver: I have close family members and friends who were diagnosed with Multiple Sclerosis (MS) and Parkinson’s disease. They were fortunate enough to get diagnosed early and start treatment, but it constantly leaves me wondering: What happens when a diagnosis comes too late? Professional Focus: As a 4th-year medical student planning to specialize in neurosurgery, I want to focus my work on clinical problems where getting an early, accurate diagnosis directly changes a patient’s outcome.When patients show early symptoms of neurodegeneration, confirming a diagnosis currently requires an expensive battery of tests—including PET/MRI neuroimaging scans, spinal taps, and costly biomarker panels. These machines cost hundreds of thousands of dollars, making definitive testing rare or non-existent in smaller municipal hospitals and developing regions.Worse, simple blood tests for Tau protein fail because of false positives. If someone gets a concussion or has a stroke, dying brain cells dump huge amounts of normal Tau protein into their blood. A standard test measuring “total Tau” will flag a simple concussion as Alzheimer’s. To solve this, my chip uses a biological AND-gate: the sensor only glows if a protein is both misfolded into a toxic shape AND chemically hyperphosphorylated. This filters out background concussion trauma and catches true chronic neurodegeneration.

2. What living system would I use?

I propose using Escherichia coli (E. coli SHuffle maybe) as the primary manufacturing host, deployed via a Cell-Free Protein Synthesis (CFPS) platform.Instead of putting live bacteria onto a diagnostic cartridge , E. coli acts as an upstream protein factory:

We introduce plasmid DNA into E. coli fused to affinity binders targeting pathological Tau.(very excited for the 2nd lecture to learn more about DNA design)

We harvest the E. coli cells, break open their membranes (cell lysis), and extract their active proteinmaking machinery and pre-folded biosensors.(here comes the 3rd lec)

This active cell-free “biological soup” is printed directly into the microchannels of plastic microfluidic chips and freeze-dried (lyophilized).When a patient’s blood plasma sample flows through the chip, it rehydrates the freeze-dried sensors. If pathological Tau is present, the split-reporter halves snap together on the target protein and glow, which a simple $20 handheld electronic reader box can measure.

Now, why SHuffle E. coli instead of standard E. coli?

This is because we’ll most definitely rely on a binder of some sort and since most protein binders rely on disulfide bonds,so we have to provide a meduim that won’t inhibit them.

knowning that a sandard BL21 E. coli will break these bonds because of its cytoplasmic reducing environment. Thus, another strain should be used , with the help of AI I found a paper about SHuffle E. coli strain that will help.

Literature Rationale:

Paper title :SHuffle, a novel Escherichia coli protein expression strain capable of correctly folding disulfide bonded proteins in its cytoplasm

This cell-free approach is grounded in the readings that were assigned by Karim & Jewett (2021), which highlights how decoupling biological machinery from living cell membranes creates stable, cold-chain-independent biosystems. Using E. coli as a high-yield factory also draws on established industrial biomanufacturing principles described by Zhang et al. (2017)

3. How would I use automation?

To be honest, as an absolute beginner to automation, this part felt a bit foggy for me at first, hopefully, in the upcoming 4th lec I get the grasp of it. However, from what I understand so far, automation can cut through a lot of unnecessary manual work, saving both time and money across two key parts of the project:
Automating Repetitive Lab Work During Prototyping :Before we can print working chips, we have to test tons of different combinations of our biosensors, target proteins, and chemical buffers to find the exact right ratio. Doing all this continuous testing by hand in the wet lab means hours of repetitive manual pipetting, which easily leads to human error and wasted reagents. On the actual diagnostic device, we need to move the patient’s blood plasma sample through the microscopic channels of the chip. If someone tries to push the fluid manually using a handheld syringe, it is very easy to press too hard or unevenly which can easily over-pressurize and damage the fragile plastic chip or create unwanted turbulent fluid flow that messes up the delicate reaction.
Instead, I would use a tiny automated micro-pump controlled by an Arduino. This automates the fluid flow, ensuring the blood plasma moves through the chip at a smooth, continuous, and steady rate without needing risky manual syringe handling.

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

Well to be honest, I had no idea on this section where would I get relevant information so I relied on AI. The volume of a each microfluidic chip is said to be 10microL , so in order to screen 2000 patients we will need 20mL of cell-free reaction mixture. I’m not sure how much time it will need since I don’t have much experience,but it shouldn’t take more than a day.

5. What don’t I know yet?

1-Affinity Binder Specificity: Can we find or engineer affinity binders (like nanobodies) that bind specifically to hyperphosphorylated residues and misfolded Tau aggregates without picking up background normal Tau monomers?

2-Shelf-Life & Reconstitution Kinetics: How long can the freeze-dried E. coli lysate sit at room temperature before losing its enzymatic activity when rehydrated with plasma?

3-Signal-to-Noise on Low-Cost Hardware: Is a 10microL sample volume big enough to generate a light signal bright enough for a $20 Arduino photodiode sensor to pick up without needing a $50,000 laboratory plate reader?(probably not)

AI used

agent:gemini

prompts:“would a diagnostic test of tau protein be feasible to make for HTBAA course project” because I was afraid I was shooting too high “wouldn’t the standard BL21 E. coli starin inhibit most protein binders that would be used ? because most binders are disulfide bonds dependent, suggest papers that is related to this problem " “how can I calculate how much tests would I need for a meaningful scale” of course I explain the test for it before

DNA sequence for detecting normal and phosphorylated tau(p-tau)

to find my work on kernel you can search the Mustafa Alwash htbaa 2026 repo which is at Adrian Filips account provided by openbioscience node.

Defining the Expression Target

i have built two split protein sensors for both normally structured tau protein and phospholyted tau(p-tau) protein, here is a structured template of what is happening in my repo:

Total Tau beta-lactamase sensor

this is the 1st sensor made for detecting normal tau in the blood. Its a split protein made up of two parts , each part follows the same general scheme which is {backbone}→{promotor}→{RBS}→{a nanobody}→{linker}→{subunit}→(terminator):

split protein A

you can see this part in this construct ,named tau, which responsible for binding to the C-terminal of tau.

split protein B

this part is at this contruct ,named tau B.

p-tau sensor

this is the 2nd sensor made for detecting phospholyted tau proteins in the blood. To be specific, this will detect only the molecules that are phosphorylated at serine 422 , whch is a type of p-tau that’s linked to Al-zheimer’s disease

split A

this constract ,named p-tau A, will be the part responsible of attachment with the phospho part of the tau

split B

is the same as the Total tau B split this constract ,tau B.

Host / Chassis: E. coli SHuffle T7 strain (for engineered redox cytoplasmic folding) and E. coli cell-free protein synthesis (CFPS) lysate.

Localization: Cytoplasm / Cell-Free Reaction Solution. No signal peptide or secretion tag is required as the reaction occurs in cell-free lysate.

desgning the genetic parts

Assembling the constructs

All split protein constructs were assembled in Kernel using the same $5’ \rightarrow 3’$ orientation scheme:

{backbone} -> {promotor} -> {RBS} -> {nanobody} -> {linker} -> {reporter} -> {terminator}

Assembled constructs in Kernel

1. Normal Tau Sensor - Split Protein A (LargeBiT):

   • {pSB1C3} -> {T7} -> {RBS A1} -> {H3-2} -> {(G4S)3} -> {BBa_I757012} -> {T7 Terminator}
   • construct

2. Normal Tau Sensor - Split Protein B (SmallBiT):

   • {pSB1C3} -> {T7} -> {RBS A1} -> {MT3.1} -> {(G4S)3} -> {BBa_I757011} -> {T7 Terminator}
   • contruct

3. pTau S422 Sensor - Split Protein A (LargeBiT):

   • {pSB1C3} -> {T7} -> {RBS A1} -> {pS422} -> {(G4S)3} -> {BBa_I757012} -> {T7 Terminator}
   • constract

4. pTau S422 Sensor - Split Protein B (SmallBiT):

   • {pSB1C3} -> {T7} -> {RBS A1} -> {MT3.1} -> {(G4S)3} -> {BBa_I757011} -> {T7 Terminator}
   • contruct

Testing plan and experimental assay

To test and validate the biosensor constructs, the system will be run in an E. coli Cell-Free Protein Synthesis (CFPS) assay

1. Cell-free expression setup Combine E. coli CFPS extract, master mix (NTPs, amino acids, energy regeneration mix, T7 RNA polymerase), and plasmid DNA (5-10 ng/µL). Co-express the sensor pairs:

Misfolded Tau pair: {H3-2-LargeBiT} + {MT3.1-SmallBiT}

pTau-S422 pair: {pS422-LargeBiT} + {pS422-SmallBiT}

Incubate at 30°C for 2-4 hours for transcription and translation.

2. Target titration and signal measurement

Add Nano-Glo substrate (Furimazine) at a 1:50 dilution to the CFPS mix.

Spike in purified target protein across a concentration gradient from 0 nM to 1 µM:

Aggregated Tau vs. monomeric Tau for the misfolded Tau sensor.

Synthetically phosphorylated Tau (pS422) vs. unphosphorylated Tau for the pTau sensor.

Record luminescent intensity at 460 nm on a plate reader over 30 minutes.

3. Experimental controls Background control: Express single constructs alone (LargeBiT only or SmallBiT only) without their complementary half to measure baseline background assembly.

Specificity control: Incubate the biosensor pairs with non-target proteins (BSA or unphosphorylated monomeric Tau) to confirm no off-target activation.

Positive control: Express full-length, intact NanoLuc driven by the {T7} promoter to set the maximum luminescence baseline in lysate.

Success metric: Target a Signal-to-Noise ratio (S/N = Luminescence_target / Luminescence_blank) greater than 10.

Project unknowns and potential risks

1. Linker spatial constraints

• The (G4S)3 linker length might not match the physical distance between Tau epitopes.
• If too short, LargeBiT and SmallBiT cannot reach each other; if too long, background noise increases.

2. Nanobody folding in cell-free lysate

• VHH nanobodies require disulfide bonds to fold properly.
• Standard E. coli lysate is reducing, which could lead to misfolded or inactive nanobodies.

3. Stoichiometry and background assembly

• Controlling equal co-expression of both plasmids in lysate is difficult.
• Excess uncomplexed reporter halves can cause non-specific background signal without target Tau present.

AI prompts

• I did relay on gemini when searching for eligible candidates for nanobodies for the detection of both normal and phosphorlyated tau protein , if not the search can go on for weeks
• Also, It did help with the concentrations and amounts mentioned in the “Testing plan and experimental assay” part of the HW

Final Project