Homework #1 · Initial Idea for Biomanufacturing

What and why

The project began with a flood-survival wearable in my Obsidian idea database, mentioned above. I wanted to explore whether a device with biomanufactured parts could automatically provide protection during accidental immersion in dangerous floodwater while remaining compact enough for everyday wear. An airbag bicycle helmet was an important initial inspiration: a small accessory that expands into a protective structure when needed.

My initial concept combined inflatable protection, flotation and a short-duration breathing apparatus. I decided this was too ambitious for a prototype, especially with the material development involved, and narrowed the focus to flotation and physical protection. Automatically deploying a breathing interface would add considerable complexity. I wanted to first work out whether the accessory could unfold reliably and provide useful flotation and protection.

I also initially considered preventing the wearer from being swept away. I am now focusing on supporting someone who may be carried by water or awaiting rescue, since flotation would not prevent drifting. Cummings et al. found that wearing a personal flotation device was associated with approximately 49% lower drowning risk in recreational boating, with an adjusted relative risk of 0.51 and a 95% confidence interval of 0.35–0.74 [10]. This gives me a reason to explore flotation, but it does not establish that my proposed wearable would be effective in floods, where forces, debris and entrapment would need separate consideration.

The device should be an everyday accessory, potentially worn around the neck, so compactness, comfort and appearance are part of whether it would be useful. The protective shapes would be manufactured in advance and folded inside it, with inflation unfolding them into position. I want the deployed structure to protect the crown, sides and back of the head, leave the face unobstructed and provide flotation around the upper body, with cushioning around vulnerable areas.

I began looking into spider silk after seeing the iNANO announcement for Anna Rising’s lecture on biomimetic spinning [17]. It suggested a possible material for the wearable that could be produced biologically. Research on recombinant silk shows promising fibre toughness, including Fan et al.’s work on aqueous spinning of recombinant spider-silk fusion proteins [11]. I wanted to investigate whether these properties could help make a lightweight reinforcing textile that packs compactly and withstands demanding use.

Inflated compartments could provide cushioning, but their construction would also need to resist tearing, abrasion and puncture. Tough fibres alone would not make a compartment durable or airtight; the membrane and seams would matter too.

I am considering recombinant spider-silk reinforcement combined with a flexible, microbially produced polymer membrane. The silk would carry and distribute loads, while the membrane would retain gas. I previously considered cellulose/PHA laminates, but performance in packaging would not be enough to establish their suitability for a repeatedly folded inflatable structure. A flexible PHA membrane remains a candidate to investigate.

I also became interested in whether the material itself could help the device change shape. Responsive regions might help initiate unfolding, alongside the pre-folded geometry, while reinforcing regions maintain support. Moisture-responsive silkworm-fibroin films provide an experimental example of actuation driven by water vapour [12], though this does not show that they would respond usefully under full immersion or be able to unfold the wearable.

I would like to understand how biological production could give different parts of the device different material properties. Alternative feedstocks and end-of-life options also interest me, although I would need to consider the processing involved and how long the materials last before assuming an environmental benefit.

I have people living or working in flood-prone communities in mind. The device would need to be practical enough to wear before an emergency occurs. Affordability would include maintenance, storage and replacement after deployment, as well as the initial material cost.

For now, I would focus on making the reinforcing textile and membrane. I would leave responsive elements for later experiments.

Living system

I then needed to understand how the silk protein could be made into fibres for the wearable.

Spider-silk reinforcement

One possibility is engineered Escherichia coli producing recombinant spider-silk proteins, or spidroins. Schmuck et al. demonstrated high-yield production of the miniature spidroin NT2RepCT in bacterial fed-batch culture, followed by purification and aqueous fibre spinning [13]. Nutrients and a carbon feed support cell growth and protein expression. After harvesting and disrupting the cells, the protein is purified, concentrated and spun into an acidic aqueous bath.

The bacteria would therefore produce the protein; making fibres and assembling them into a textile would come afterwards. For the wearable, I would need to consider how the fibres are arranged to distribute loads while keeping the textile flexible. The protein sequence and processing conditions would both influence the resulting material.

I would also need to understand how the fibres behave when wet. Asakura et al. investigated a modified recombinant spider-silk sequence and fibres with improved dimensional stability during repeated wetting and drying [14]. Changes in dimensions could affect how the inflated structure fits around the wearer, so this would be important to test.

The studies use different silk constructs and processing methods. I therefore cannot assume that one fibre would combine the toughness reported by Fan et al., the production yield achieved by Schmuck et al. and the water stability investigated by Asakura et al. [11], [13], [14].

Gas-retaining membrane

For the membrane, I am considering engineered Yarrowia lipolytica producing medium-chain-length polyhydroxyalkanoate, or PHA. Rigouin et al. produced two polymers, including a copolymer with thermoplastic-elastomer behaviour. Their cultures used glucose-containing medium supplemented with fatty-acid methyl esters, and the recovered polymers were made into films for characterisation [15].

My first question would be whether a selected film could retain gas, withstand repeated folding and form durable seams. I could compare a membrane supported within a silk textile with a bonded layered assembly. Neither the gas retention nor the compatibility of these materials is established for the wearable.

The study used chloroform to extract the polymers [15]. I would need to include recovery, solvent use and energy demand when assessing the process, rather than judging it only by the biological production step.

Responsive elements and device assembly

The shape-change idea would need its own investigation. Manikandan et al.’s moisture-responsive films were made from silkworm fibroin, rather than recombinant spider silk [12], so I could not assume the reinforcing fibres would behave in the same way. A responsive element might eventually help with deployment, but the main structure would still consist of pre-shaped, folded inflatable compartments.

The organisms would produce the materials, rather than remain alive in the finished device. Those materials would then need to be assembled into chambers and folded, with inflation hardware and a trigger added to make the wearable function.

Automation

I used Claude to help identify plausible automation approaches for this idea, since my background is in nanoscience and I do not normally work with production scale-up. An approximate reconstruction of my prompt is:

How could automation help me make and test the materials for this wearable? I am considering recombinant spider-silk reinforcement and a microbially produced flexible membrane. What would make sense for comparing materials initially, and what might change at a larger production scale? My background is in nanoscience, and I have limited experience with manufacturing and automation.

I would need to investigate these suggestions further before choosing equipment or methods.

Since the material choices remain open, I would begin by comparing production conditions and how they affect the materials. An Opentrons liquid-handling robot could prepare nutrient formulations, distribute samples and set up protein-quantification assays. These tasks fit its documented liquid-handling capabilities and demonstrated protein-assay workflows [16]. Culturing and taking measurements would require additional equipment.

A preliminary screen could compare four expression conditions and three nutrient formulations, each in triplicate:

I would compare the amount of protein recovered, rather than cell growth alone, then check promising conditions in controlled bioreactors. Automation would help prepare the samples consistently and repeat the comparison as conditions are adjusted.

Making the accessory strong but compact would also depend on how the materials are processed. Programmable pumping and fibre collection could control spinning conditions, while controlled film casting could vary membrane thickness. I would want to see how these choices affect the amount of material needed and how tightly the assembly can be folded.

Automated mechanical tests, repeated-folding fixtures and pressure logging could help compare wet strength, fatigue and leakage. Keeping each sample’s production conditions linked to its test results would help me decide what to change if it tears, weakens or loses pressure.

For larger production volumes, I would consider bioreactor control, purification and parallel fibre processing. Producing more protein or polymer would only be useful if it could still be processed into suitable fibres and membranes.

How much, how long

I do not normally estimate manufacturing quantities or production times, so I used Claude to help work through what scaling this idea might involve, based on the research I had found. The following is an approximate reconstruction of my prompt:

Using the papers I have found, help me estimate how much silk protein and PHA I would need for one wearable prototype and some test samples, and how long production might take. I study nanoscience and do not normally work with manufacturing or scale-up. Explain the main calculations and make clear which values come from the papers and which we are assuming.

I wanted a rough sense of the quantities and time involved. Without a final pattern or measurements of material losses during processing, I would treat these estimates as a starting point.

(a) Material quantity

I began with an assumed 1 m² each of reinforcing textile and membrane, counting all the individual panels. A textile weighing 200 g/m² and a membrane 50 µm thick, with an assumed density of 1,100 kg/m³, would give:

This would put the textile and membrane at around 255 g before adding inflation hardware, closures and other components. It does not yet tell me whether the accessory would be comfortable or compact enough; that would depend on the seams, folding and actual material thickness.

To allow for test samples and cutting losses, I added 25% extra material. I also assumed that 70% of the purified silk protein would become usable textile and 80% of the PHA produced inside the cells would become usable film. I do not have measured values for these efficiencies, so they are assumptions for the calculation.

Schmuck et al. reported 14.5 g of purified silk protein per litre of culture [13]. Using that yield:

Rigouin et al. reported approximately 2.9 g of PHA produced inside the cells per litre of culture [15], giving:

Under these assumptions, I would need a few tens of litres of each culture for one demonstrator and test samples. The silk value already accounts for purification, whereas the PHA still needs to be extracted from the cells. Different material variants or production conditions could change these yields.

(b) Production time

I also wanted to understand how long the steps after biological production might take. Producing enough protein would not necessarily mean the fibres could be made quickly.

Schmuck et al. calculated that 23 g of protein could supply approximately 75 hours of spinning at their laboratory flow rate [13]. The material estimate above requires about 357 g of purified protein, which would give:

This assumes continuous spinning at the reported rate. It is not a demonstrated production schedule, but it suggests that making the fibres could take much longer than I initially expected. Ten equivalent spinning lines would theoretically reduce this to about five days, although I would need to consider equipment availability and whether the fibre quality could be maintained.

The PHA study used five-day cultures [15], followed by polymer recovery and film preparation. Silk cultivation and purification, textile manufacture, assembly and testing would also take time. I would need to know more about the equipment and processing steps before giving a reliable overall schedule.

I would therefore begin with small material samples and one deployable compartment. These would let me measure how much material is needed, how much is lost during processing and how long the steps take before attempting the complete wearable.

What I don't know yet

Would the assembled materials be durable enough?

Silk interested me because of its mechanical properties, but I would need to test how the textile, membrane and seams behave together. Could they withstand abrasion when wet, puncture, impact and repeated folding while still retaining gas? I also want to understand how much material this would require, since added reinforcement could make the accessory too bulky for everyday wear.

Would the folded structure deploy as intended?

The structure would need to unfold reliably around the head and upper body, leave the face clear and support the wearer in a position that keeps their mouth and nose above water. I would begin with mannequin tests to examine the geometry and loads at attachment points, then investigate impact protection and leakage. I have not yet settled on a trigger. If responsive materials are included, I would need to establish whether they help deployment and whether everyday moisture could activate them prematurely.

Would biomanufacturing offer a practical advantage?

I found possible ways to produce these materials biologically, but I am still unsure about the cost, processing involved and how long they would last. Material recovery and spinning time would help me understand production, while packed size and maintenance requirements would affect everyday use. Solvent use and energy demand would also need consideration. I want the wearable to be affordable and practical enough that someone would already be wearing it when an emergency occurs.

What and why

I would like to make a flexible living fabric whose appearance can be changed through light scattering inside its cells. I was interested in cephalopod skin because its optical properties depend on the arrangement of biological structures. Cuttlefish leucophores, for example, produce broadband white scattering through densely packed protein-containing structures called leucosomes [18]. My idea was to use a related mechanism in a fabric, where changing what the cells express would change its appearance.

I first considered gas vesicles as the scatterers. Their gas-filled interiors have a different refractive index from the surrounding cell, which made them seem suitable. I had some doubts after reading about their behaviour under pressure and the effects of producing them. Gas vesicles collapse irreversibly above a certain pressure [19], though I do not know whether ordinary bending or the fabrication process would reach that pressure. One study also reported a decline in bacterial viability 8–16 hours after induction. Adjusting the timing of expression helped preserve viability [20], so this did not rule them out, but it meant I would have to consider when and how much the cells should produce.

I then looked into reflectin. In 2020, the Gorodetsky group expressed reflectin A1 in HEK293 cells, where it formed spheroidal structures and increased scattering. The cells also showed different optical responses under different salt conditions [21]. Two papers from 2023 examined the structures further: one used experiments and modelling to map their refractive-index distributions [22], and the other studied chemically induced changes inside mammalian cells [23].

I chose reflectin because it had already been shown to change a cell’s optical properties through structures formed inside it. The work was in mammalian cells, however. I would still need to find out whether bacteria could produce useful structures and whether encapsulating them would preserve the effect. I have not established that reflectin would be tougher or easier to express than gas vesicles.

The 2020 paper included measurements of fixed cell monolayers [21]. What I want to examine is a living bacterial layer inside a flexible material, including how thickness and cell density affect diffuse reflectance and total transmittance. I am particularly interested in whether reflectin expression could make a visible difference to how well the material hides a pattern underneath.

I would start with a fabric swatch. Since bacteria scatter light without reflectin, I would compare induced and uninduced layers of the same composition. This would help separate the effect of reflectin from scattering already present in the cells and material. I would also compare them through the outer enclosure, since that is how the swatch would actually be viewed.

Why the outer layers and compartments?

I wanted an outer layer because the living component needs to stay hydrated, but the fabric also needs to withstand handling. I imagine a thin living layer behind a transparent viewing surface, with a gas-permeable backing and mechanical support around it. The assembly would need to retain water while remaining flexible. Bioprinting could help make the hydrogel layer, but I would still need a supporting textile, films and a way to join them.

The small sealed polygons came from thinking about how the material could bend around a curved surface. Bending between compartments might allow this without requiring each compartment to deform as much. Tessellating shapes could also fit beside one another without the gaps left by separate circular pockets.

Another possibility would be to fill the compartments with cells prepared in different expression states. That would let me make a pattern without first having to demonstrate local switching inside the assembled material.

I have not decided which shape or size to use. Smaller compartments might make bending easier, but they would also mean more seals across the surface. For a square compartment with outer side length $L$ and a seal extending inward by $w$ on each side, the fraction containing the living layer would be:

With an assumed 1 mm seal, 81% of a 20 mm compartment would be available for the living layer, compared with 64% of a 10 mm compartment. The seals could still reflect or scatter light, but the cells would not control their appearance. Making the compartments smaller could therefore reduce how much of the surface can visibly change.

I also need to consider how long the cells would remain functional. Smaller compartments would contain fewer nutrients, but they would contain fewer cells too, assuming the same thickness and cell density. I cannot conclude from size alone that they would run out sooner. I would need to look at transport through the enclosure and water loss, as well as how compartment size affects bending and the visible pattern.

Living system

I would start with a laboratory strain of Escherichia coli and inducible reflectin expression. Reflectins have been produced in bacteria, including as intracellular inclusion bodies [24]. The bacterial work examined R1b and R6, while the 2020 optical study in HEK293 cells used RfA1 [21], [24]. Being able to produce one sequence does not tell me how another would assemble or scatter light.

RfA1 interests me because of its demonstrated optical effects inside cells. I would need to investigate its expression in bacteria and whether it forms structures that add enough scattering to be useful. The cells would also need to remain viable while accumulating the protein.

The bacteria would need a carbon source, nutrients, water and a signal to induce expression. They would remain inside the material after producing reflectin, so I would be working with the cells themselves rather than extracting and purifying the protein.

For the first comparison, I would prepare induced and uninduced batches and incorporate them into otherwise equivalent layers. A layer without cells would show how much the matrix and enclosure contribute to scattering. I would also include an appropriate induction control to check whether any changes come from reflectin or from other effects of induction.

Eventually, I would like to change expression after the cells have been encapsulated. The inducing signal would have to reach them through the enclosure. If I used a chemical signal, it might also diffuse between compartments through any permeable connections. Keeping the bacteria in separate compartments would not necessarily keep the signal separate.

I also cannot assume that switching expression off would return the material to its earlier appearance, since reflectin already produced could remain in the cells. Controlling protein production, changing how the protein assembles and reversing the visible effect would each need to be tested.

Automation

I considered bioprinting as a way to control the shape and thickness of the living layer. My bachelor’s project involved a biphasic bioink for 3D bioprinting of nerve-like semi-synthetic tissue, so this is an approach I have some experience with.

Liu et al. printed responsive bacterial materials using a Pluronic F127 diacrylate hydrogel ink [25]. This suggests a possible way to deposit the bacteria in patterns, although I would need to check whether the formulation is sufficiently transparent and whether reflectin still produces a useful scattering effect inside it.

The printer would make the living layer. The supporting structure, viewing surface, backing and seals would need to be made and assembled separately. I would consider printing if it helped control thickness, fill individual compartments or place cells in a pattern. For a uniform layer, casting or dispensing might be simpler.

I also found work on stretchable living devices made from hydrogel–elastomer hybrids [26], which gave me an example of how a living layer could be enclosed in a flexible structure. That study used a polyacrylamide–alginate hydrogel, however, so I could not assume its bonding and fabrication methods would work with the Pluronic ink.

The bioprinting idea and my reading of these papers were my own. I asked Claude for help with automating sample preparation, which led to the suggestion of an Opentrons liquid-handling robot. An approximate reconstruction of my prompt is:

How could I automate preparing samples with different bacterial concentrations and induction conditions for this living material? Could an Opentrons help, and what would I need to check before using it with the hydrogel?

An Opentrons could help prepare repeated combinations of cell concentration and induction conditions [27]. Whether it could also handle the viscous matrix would need checking separately.

I would use automated measurements to compare diffuse reflectance and total transmittance at different thicknesses, cell densities and expression states. Photographing the same test pattern underneath each sample would help me relate the measurements to how much of the pattern remains visible. I would measure the samples inside the complete enclosure, since the films and seals could affect their appearance.

Producing larger sheets or more swatches would also mean considering how to automate culture production, deposition, sealing and inspection. The equipment used to prepare small experimental batches might not suit those steps. I would first need to find a combination of materials and assembly methods that works consistently before deciding how to increase production.

How much, how long

I asked Claude for help with these estimates again because I have limited experience with biomanufacturing and scale-up. Learning how to approach these questions is one of the reasons I am taking this course. I used my proposed dimensions and the papers I had found as the basis for the calculations. An approximate reconstruction of my prompt is:

I want to estimate the material quantities and production time for a 20 × 20 cm swatch with a 0.5 mm living layer, then compare this with a 1 m² panel. Based on the papers I found, help me work out the amount of cell-containing material, the culture volume and the time needed for growth, expression and printing. I have limited experience with biomanufacturing and scale-up, so explain the calculations and distinguish values from the papers from assumptions we make.

(a) How much

I would begin with a 20 × 20 cm swatch, large enough to examine its appearance, compartment arrangement and bending. Assuming a 0.5 mm living layer:

Allowing 20% extra for preparation and filling losses gives approximately 24 mL per swatch. A 1 m² panel at the same thickness would need 0.5 L, or 0.6 L with the same allowance. These estimates treat the whole area as filled; seals would reduce the volume occupied by the living layer.

The swatch would also need about 0.04 m² each of viewing film and backing, plus margins and mechanical support. For the larger panel, the two films would total about 2 m². I cannot yet estimate their mass because I have not chosen the materials or thicknesses.

The culture volume is less certain because I do not know what cell density would give useful scattering. For a rough calculation, I assumed 10 g of dry-cell equivalent per litre of hydrated material, a culture producing 2 g/L and 80% cell recovery. These are calculation assumptions, not values established for this material. “Dry-cell equivalent” describes the amount of biomass; the cells would remain hydrated.

For one swatch, this gives:

The same calculation gives approximately 3.75 L of culture for the 1 m² panel. I would revise these estimates after measuring the optical response at different cell densities. If useful scattering required ten times as many cells, the culture requirement would also increase tenfold.

(b) How long

The bacterial reflectin-production paper includes overnight starter growth and overnight expression [24]. For this estimate, I allowed 16 hours for the starter, eight hours for expansion and 16 hours for expression, giving about 40 hours. These durations are planning assumptions and presume that a working strain is already available. Developing the strain and finding suitable expression conditions would require additional time.

Liu et al. reported a printing speed of 5 mm/s [25], although their nozzle sizes and structures differed from those proposed here. Using that speed as a reference, I assumed 0.5 mm line spacing and two 0.25 mm layers. Filling the swatch with parallel lines would then take:

This leaves out turns, interruptions, curing, sealing and inspection. I would also need to check whether the proposed line geometry could actually be printed at that speed.

Culture and printing would therefore add up to roughly 49 hours for one swatch, before cell recovery, the remaining assembly steps and testing. This gives me an initial estimate of those two stages, rather than a completion time for the swatch.

For a 1 m² panel, the same single-nozzle calculation gives around 222 hours of printing alone. Printing the entire layer this way may therefore be impractical at that size. If the small samples showed a useful optical response, I would compare it with wider or parallel deposition and casting into preformed compartments.

What I don't know yet

Would reflectin make a visible difference in bacteria?

The bacterial production studies and mammalian-cell optical studies do not yet answer this. I would need to compare bacterial layers with and without reflectin at different thicknesses and cell densities to see whether the additional scattering visibly changes the material.

How long would the cells remain alive and responsive?

Reflectin production, nutrient availability, water loss and cell growth could all affect the cells and their scattering. A layer might retain its appearance after the cells stop responding. I would therefore need to check whether signals reach the encapsulated cells and whether I can change the appearance more than once.

Would it work as a fabric?

I would need to test the living layer together with its enclosure, support and seals. Compartment size could affect how easily it bends, how much of the surface can change appearance, whether it leaks and how long it remains functional. The fabrication studies suggest methods I could try, but this combination has not been tested.

I am interested in whether the cells could keep changing the material’s appearance after it has been made. I want to find out how much control over appearance is possible through the structures inside the cells, and whether the effect would still be visible once they are enclosed in a fabric.