Individual Final Project

https://docs.google.com/presentation/d/1oW4llqCBtHTd_fU4BQsxu1bvHlVMxw1XPJHbaQzOp-U/edit?usp=sharing

Sequence-Programmed Hygromorphic Coatings from Engineered Elastin-Like Proteins

DNA Sequence as a Material Design Variable for Programming Environmental Response

Yao Wang | HTGAA 2026

1. Concept

What if you could write a line of genetic code and have it determine how a surface curls in the rain? This project treats DNA sequence as a material programming language. Rather than engineering biology to produce a molecule or report a signal, I use genetically encoded elastin-like proteins (ELPs) to ask a more fundamental question: can editing the sequence of a repetitive protein produce predictable, distinct material behaviors at the macroscale?

The core proposition is that DNA is the only existing system where a discrete, editable, reproducible instruction set compiles into a physical material with tunable macroscopic behavior. Unlike conventional responsive materials, where achieving different behavior requires reformulating chemistry, adjusting processing, or introducing new components, sequence-encoded materials allow variation to live in the code rather than the process. The fabrication pipeline—same vector, same bacteria, same induction, same coating method—stays identical. What changes is the material output.

ELP is the right chassis for this proof of concept because it is one of the few protein systems where the sequence-property relationship is unusually direct. ELPs are intrinsically disordered: they do not need to fold correctly to function, which means expression almost always works and the pipeline does not break when the sequence changes. Their lower critical solution temperature (LCST) transition—soluble and extended below the transition temperature, collapsed and aggregated above it—provides environmentally responsive behavior intrinsic to the material, without requiring any engineered sensor or switch. And because the transition temperature shifts predictably with sequence composition, editing the DNA directly tunes the trigger point. The distance between a sequence edit and a material behavior change is shorter, more predictable, and less failure-prone in ELPs than in virtually any other protein system.

Most responsive materials today rely on synthetic polymers, complex fabrication, or external mechanical actuation. Biologically derived responsive systems exist—whole-cell bio-actuators, for instance—but they are difficult to control and deploy. This project occupies a third space: genetically encoded material behavior without living cells in the final product. The material is dead but programmed. That framing opens a path toward adaptive biomaterials, environmentally responsive interfaces, and biofabrication systems tunable for different deployment conditions.

2. Project Aims

Aim 1: Experimental Aim — This Project

Test whether DNA-encoded ELP sequence variants produce distinct humidity-responsive material behaviors using Benchling-based DNA design, molecular cloning, E. coli expression, basic protein purification, and small-scale film/bilayer fabrication on PET substrates.

I will construct three ELP variants that differ at the guest residue position of the canonical (VPGXG) pentapeptide repeat, producing a hydrophilicity gradient: (VPGSG)₂₀ hydrophilic/serine, (VPGAG)₂₀ balanced/alanine, and (VPGVG)₂₀ hydrophobic/valine. Each construct includes an N-terminal 6xHis tag for detection and optional purification. The three variants share an identical backbone architecture and differ by a single residue per repeat unit, isolating amino acid identity as the sole design variable.

I will compare their water uptake, swelling, bending amplitude, recovery speed, and structural stability as one-sided coatings on PET film strips under controlled humidity changes. Key metrics include maximum bending angle, time to peak deformation, recovery during drying, and coating integrity over repeated wet/dry cycles.

Aim 1b: Diblock Architecture Variants — Stretch Goal

If the three single-composition variants produce distinguishable behaviors, I will additionally test diblock ELP constructs that combine two guest residues in a single chain—for example, (VPGSG)₁₀-(VPGAG)₁₀ or (VPGSG)₁₀-(VPGVG)₁₀. These diblocks test whether sequence organization, not just composition, changes film behavior, which is a stronger claim for the “sequence as design language” framework. Diblocks may exhibit phase separation, domain-specific swelling, or asymmetric film morphology that single-composition variants cannot produce. These constructs are included in the design files for feasibility assessment; synthesis and testing will proceed only if timeline and Twist compatibility allow.

Aim 2: Development Aim — Multi-Axis Sequence Programming

Aim 1 demonstrates that sequence edits along one axis, guest residue hydrophilicity, produce measurable differences in one class of material behavior: humidity response. Aim 2 would expand the system to a second, qualitatively different behavioral axis by introducing a charged guest residue.

The proposed construct is (VPGKG)₂₀, with lysine at the guest position. Lysine introduces a positively charged side chain at physiological pH, making this variant not just humidity-responsive but also pH-responsive and salt-responsive. Under Aim 2, I would test the same coating-on-PET pipeline but expose samples to acidic vs. basic conditions and varying ionic strength, comparing whether the same sequence framework can be tuned to respond to fundamentally different environmental inputs. This would transform the project from “I varied one parameter” to “I showed that sequence edits can switch which environmental variable the material responds to.”

Aim 2 would also include expanding environmental inputs for the Aim 1 variants, such as temperature and salinity, and improving film robustness, bilayer control, and material processing methods.

Aim 3: Visionary Aim — Sequence–Environment–Behavior Maps

In the long term, I envision this project becoming a platform for environmentally responsive biomaterials in which DNA sequence serves as a programmable design language. The endpoint is a searchable, combinatorial design space: given a target environmental condition and a desired material response, a designer could consult a sequence–environment–behavior map to select or computationally predict the right ELP construct. This would bring materials design closer to the logic of software—version-controlled, editable, reproducible—and enable new approaches to adaptive interfaces, climate-specific materials, and sustainable biofabrication.

3. Background

Why ELP?

Elastin-like proteins are among the few protein systems where structure is optional. Most proteins require correct folding to function; ELPs are intrinsically disordered below their transition temperature and do not depend on tertiary structure for their material properties. This means expression almost always succeeds, the pipeline tolerates sequence variation without catastrophic failure, and multiple variants can be produced in parallel with confidence. The LCST phase transition is intrinsic to the material—no engineered sensor or switch is needed—and shifts predictably with guest residue identity. ELPs also film well: the existing literature supports that they can be processed into coatings, films, fibers, and hydrogels, providing reasonable confidence that expressed protein will form usable coatings on PET.

Literature Context

Srokowski et al., “Surface and adsorption characteristics of three elastin-like polypeptides with sequence length and guest residue variations.” This study demonstrates that ELP sequence length and guest residue composition significantly alter thin film and coating surface properties, directly supporting the logic that DNA sequence variation translates into material behavior differences.

Välisalmi et al., “Highly Hydrophobic Films of Engineered Silk Proteins by a Simple Fusion Protein Concept.” This work shows that genetically engineered spider silk proteins can be processed into thin films with strongly hydrophobic surface properties, demonstrating that recombinant protein design can directly program film behavior.

Zhu et al., “Humidity-responsive self-assembly of short peptides with super-flexibility.” This study reports that short peptide materials undergo reversible structural transformation under changing humidity, driven by hydrogen bonding and conformational changes. It provides direct precedent for biomolecular humidity-responsive actuators.

Liu et al., “Spider dragline silk as torsional actuator driven by humidity.” This research demonstrates that natural protein materials can convert humidity input into large-amplitude mechanical deformation, establishing a strong precedent for using biological molecules as humidity-driven actuators.

Novelty

This project is novel because it treats DNA sequence as a design variable for programming material behavior, not merely as biological information for producing a molecule. Instead of engineering biology to express a protein and then asking what material it makes, I propose a controlled comparison: a small family of variants that differ by a single residue per repeat unit, tested through an identical pipeline, to isolate the effect of sequence on macroscale response.

The project also reframes bio-inspired responsive design. Rather than replicating whole-cell bio-actuators or mimicking biological morphology, it focuses on a minimal and testable platform where DNA-encoded peptide structure directly shapes macroscale material response. The inclusion of diblock architecture variants, Aim 1b, and a charged variant, Aim 2, extends this beyond parametric tuning toward a genuine demonstration that sequence can control qualitatively different dimensions of material behavior.

Significance

Most responsive materials rely on synthetic polymers, complex fabrication, or external mechanical components. These systems are difficult to customize at the molecular level, and variations in behavior typically require reformulating chemistry or processing. This project addresses the problem by demonstrating that variation can live in the genetic code itself: the fabrication pipeline stays identical while the material output changes.

If successful, this work establishes that even minimal sequence edits—a single amino acid substitution per repeat unit—produce measurably different humidity-responsive behaviors. This is the minimum viable proof that DNA can function as a material programming language. The broader implication is a path toward biomaterials whose behavior is version-controlled, computationally searchable, and tunable for specific deployment conditions, with applications in adaptive interfaces, environmental sensing, and sustainable biofabrication.

4. Ethical Considerations

This project raises ethical questions around biosafety, environmental responsibility, and scientific framing. Although the work uses standard laboratory expression systems and is limited to small-scale peptide/protein materials, it involves synthetic biology methods that should be guided by non-maleficence and responsibility. A key ethical concern is avoiding exaggerated claims: a sequence-programmed biomaterial is not automatically sustainable, safe, or deployment-ready. Observed material behavior may depend not only on DNA sequence but also on processing conditions and experimental context, and these limitations must be stated clearly.

All work will be conducted in contained laboratory conditions using non-pathogenic systems, with no environmental release or real-world deployment. The project will be described as a proof of concept, not an immediate solution. Uncertainties and limitations will be explicitly stated in interpreting results. Potential unintended consequences include overstating sustainability, overgeneralizing from a small number of variants, or assuming biomaterials are inherently beneficial. Alternatives include first testing similar responsive behaviors with non-engineered biological polymers or fully synthetic materials before scaling toward more complex engineered systems.

5. Experimental Design

Central Hypothesis

Changing the guest residue in DNA-encoded repetitive elastin-like proteins will alter their water uptake, swelling behavior, coating morphology, and bilayer mismatch on a passive PET substrate. As a result, different ELP sequence variants will generate distinct humidity-responsive material behaviors such as bending, curling, wrinkling, or changes in surface texture.

Sequence Design

Three ELP variants, differing only at the guest position X of the (VPGXG) pentapeptide repeat:

Stretch goal constructs, Aim 1b: Diblock variants combining two guest residues in a single chain, such as (VPGSG)₁₀-(VPGAG)₁₀ and (VPGSG)₁₀-(VPGVG)₁₀. These test whether sequence organization—not just composition—affects film behavior. Feasibility of Twist synthesis for these constructs will be assessed before ordering.

Future construct, Aim 2: (VPGKG)₂₀, a charged variant with lysine at the guest position. This variant is designed but not tested in the current project. It would introduce pH-sensitivity and salt-sensitivity, demonstrating that the same ELP scaffold can be tuned to respond to qualitatively different environmental inputs.

Detailed Protocol

1. Define the sequence design strategy and project scope, Day 1. Finalize three ELP variants with a clear sequence gradient. Expected outcome: a set of three protein variants with a strong conceptual link between sequence composition and expected material response.
2. Design three DNA-encoded ELP variants in Benchling, Days 1–2. Design constructs with N-terminal 6xHis tag, annotate inserts, confirm reading frames, check restriction sites, verify repeat organization. Expected result: three fully annotated DNA constructs ready for synthesis.
3. Prepare cloning strategy compatible with Twist Bioscience synthesis, Days 2–3. Prepare insert sequences with flanking regions for cloning into an IPTG-inducible E. coli expression vector. Select target vector backbone and ensure synthesis compatibility. This avoids cloning errors associated with repetitive DNA.
4. Submit Twist gene fragment order and finalize vector design, Day 3. Order each variant as a single Gene Fragment from Twist Bioscience, guaranteed 100–1000 ng per fragment. Three variants = three fragments, approximately $63 total. This quantity is sufficient for Gibson Assembly into pET-28a; protein yield depends on bacterial amplification, not DNA input quantity. If budget allows, order two copies of each fragment as backup. Alternatively, ordering Clonal Genes, where Twist clones and sequence-verifies for you, costs more but saves 3–5 days of cloning time. In parallel, finalize the expression vector design including promoter, RBS, His tag, antibiotic resistance, and cloning junctions.
5. Clone synthesized fragments into expression vector, 2–3 days after DNA arrival. Use Gibson Assembly, restriction enzyme cloning, or another standard method. Transform into DH5α or TOP10, plate on selective agar, pick colonies.
6. Validate constructs by colony PCR, MiniPrep, and sequencing, 1–2 days. Colony PCR, liquid culture, MiniPrep, diagnostic digestion and/or sequencing. Sequence confirmation is especially important for repetitive regions.
7. Transform verified plasmids into expression strain, 1 day. Transform into BL21(DE3) or equivalent for inducible production.
8. Perform small-scale expression tests, 2 days. Pilot expression under IPTG induction. Assess by SDS-PAGE comparing induced and uninduced samples. Goal: determine whether each construct expresses detectably and whether the protein is soluble enough for coating experiments.
9. Optimize expression conditions if needed, 1–2 days. Test induction timing and IPTG concentration. Critical: expression temperature must be 18–25°C with overnight induction, and all protein solutions must be kept at 4°C throughout handling and storage. This is not an optional optimization—it is a prerequisite for the entire downstream pipeline. ELPs expressed at 37°C are far more likely to aggregate irreversibly, and protein left at room temperature will degrade or phase-separate before coating. The project requires only small-scale material, so the yield optimization threshold is modest, but temperature discipline is non-negotiable.
10. Extract and optionally purify recombinant proteins, 2 days. Lyse cells. Use Ni-NTA affinity purification for His-tagged constructs if feasible; otherwise proceed with partially purified fractions and state this limitation. Quantify by Nanodrop or BCA assay.
11. Prepare standardized PET substrates, 1 day. Use 6 µm Mylar/PET film as the primary substrate, Premier Lab Supply, approximately $12–60 per roll. Cut into uniform strips or small rectangles. 6 µm is thin enough to respond visibly to coating-induced stress; if it proves too flimsy to handle reliably, 12 µm Mylar is the fallback. Before coating, treat the PET surface with oxygen plasma, available at MIT.nano after registration and training or by arranging brief access through a nearby microfluidics group. Plasma treatment is critical: it activates the PET surface for better protein adhesion. The coating must be applied within 30 minutes of plasma treatment, as the hydrophilic surface will revert.
12. Create protein coatings on one side of PET strips, 1–2 days. Apply each protein sample onto one side of plasma-treated PET using repeated drop-casting: deposit one layer, allow to dry completely, then deposit the next layer, repeating 3–5 times to build up coating thickness. Confirm each layer is fully dry before applying the next. This multi-pass approach produces a thicker and more continuous active layer than a single application. The result is a bilayer system: PET as passive layer, ELP coating as humidity-responsive active layer.
13. Document initial coating quality and morphology, 1 day. Photograph and image coated samples. Note differences in smoothness, cracking, opacity, aggregation, thickness uniformity, and adhesion across variants.
14. Expose coated strips to controlled humidity changes, 1–2 days. Use a sealed chamber with moisture source or standardized mist exposure. Introduce a reproducible humidity change and observe how coated PET strips respond. Different variants should swell differently and generate different bilayer mismatch.
15. Record macroscopic responses with video and images, same day. Capture time-lapse images or short videos during humidity exposure and drying. Observable readouts: bending angle, curling, lifting, wrinkling, surface texture change.
16. Quantify deformation and response kinetics, 1–2 days. Measure maximum bending angle, time to peak deformation, recovery time, curl radius, or extent of wrinkling. If bending is minimal, quantify alternative outputs such as end displacement or surface-area distortion.
17. Perform repeat-cycle testing for reversibility, 1 day. Expose samples to at least three dry/wet cycles. Record whether response is reversible, degrades, or delaminates—degradation patterns are themselves sequence-dependent outcomes.
18. Include controls, 1 day. Uncoated PET, PET with blank buffer/lysate, and if possible PET with a non-repetitive control protein. Controls distinguish sequence effects from substrate or solvent artifacts.
19. Interpret results through biological and materials frameworks, 1–2 days. Biologically: does DNA sequence function as a design variable for protein material properties? From materials: does sequence-dependent water uptake and bilayer mismatch produce distinguishable macroscale behaviors?
20. Plan for technical risk and fallback outcomes. Two-variant comparison is sufficient if one fails to express. Partially purified samples can support preliminary coating tests. Wrinkling, swelling, or coating integrity differences are meaningful even without large bending.
21. Assemble final figures, workflow diagrams, and documentation, final 2 days. Organize all design materials, cloning maps, expression data, coating images, response videos, and quantitative plots into a coherent final presentation.

Condensed Timeline

Critical Experimental Notes

Expression temperature is non-negotiable. Induce at 18–25°C overnight. Keep all protein solutions at 4°C from lysis through coating. ELPs expressed at 37°C are prone to irreversible aggregation, and protein stored at room temperature will phase-separate or degrade before it can be used. This is not an optimization variable—it is a prerequisite for the entire downstream pipeline.

Plasma treatment timing. After oxygen plasma treatment of PET, the activated hydrophilic surface begins reverting within 30 minutes. All coating must be applied within this window. Plan the workflow so that plasma treatment and drop-casting happen in immediate sequence.

Multi-layer coating protocol. A single drop-cast layer is unlikely to produce sufficient thickness for visible bilayer mismatch on 6 µm PET. Build up the coating through 3–5 repeated cycles of drop-cast → dry → drop-cast. Confirm each layer is fully dry before the next application. This is especially important for the more hydrophobic variants, which may form thinner or less continuous films per layer.

Substrate thickness selection. Start with 6 µm Mylar: thinner substrates amplify bilayer bending response. If 6 µm is too fragile to handle or cut cleanly, switch to 12 µm as the fallback. Document which thickness was used for each sample, as substrate thickness directly affects the magnitude of observable deformation.

6. Techniques

Relevant Techniques Checklist

• Pipetting
• Lab safety
• Bioethical considerations
• DNA sequencing
• DNA construct design
• Restriction enzyme digestion
• Gel electrophoresis
• DNA purification from gel
• Databases: GenBank, NCBI
• Designing a Twist order
• Protein design
• Use of Benchling
• Chassis selection: BL21(DE3)
• Plasmid preparation
• Bacterial culturing
• Bacterial processing: centrifugation, lysis, DNA purification
• Protein purification
• Gibson Assembly
• Primer design or selection
• PCR reactions

Two Key Techniques in Detail

Benchling-based DNA construct design. I will use Benchling to design, annotate, and validate all three ELP constructs before synthesis. This includes confirming the reading frame across all 20 repeat units, checking for internal restriction sites that could interfere with downstream cloning, verifying the 6xHis tag is in-frame with the start codon, and ensuring the flanking regions are compatible with the chosen cloning method and expression vector. Because repetitive sequences are prone to recombination and deletion, careful construct design in Benchling is critical to catching problems before synthesis rather than after. Benchling will also serve as the central documentation and version-control platform for the project’s DNA design files.

SDS-PAGE for expression validation. After IPTG induction, I will use SDS-PAGE to compare induced and uninduced cell lysates for each ELP variant. ELPs are repetitive and often migrate anomalously on gels—appearing at a higher apparent molecular weight than predicted—so I will calculate expected molecular weights in advance and look for bands in the appropriate range. SDS-PAGE will also allow me to assess relative expression levels across the three variants and to determine whether each protein is primarily in the soluble or insoluble fraction, which directly affects whether the downstream coating experiments are feasible. This technique is the gatekeeper between the cloning/expression phase and the materials phase of the project.

Associated Industry Council Companies

Twist Biosciences, gene synthesis; Addgene, vectors; New England Biolabs, cloning enzymes and Gibson Assembly; Opentrons, potential automation; Thermo Fisher Scientific, expression and purification reagents; Benchling, construct design platform.

7. Expected Results and Quantitative Expectations

Validation Approach

I will validate the project by designing and ordering the three ELP DNA constructs via Twist Bioscience, confirming successful cloning by colony PCR and sequencing, and verifying protein expression by SDS-PAGE. This validates the core pipeline from sequence design to protein production.

Expected Outcomes

Humidity response gradient. The serine variant (VPGSG)₂₀ is expected to show the highest water uptake and strongest swelling-driven deformation. The alanine variant (VPGAG)₂₀ should show moderate response. The valine variant (VPGVG)₂₀ is expected to show the weakest swelling, potentially with more brittle or opaque coatings due to increased hydrophobic aggregation.

Coating morphology. Different guest residues may produce visible differences in film smoothness, cracking patterns, opacity, and adhesion to PET, even before humidity testing.

Reversibility. Some variants may retain reversible actuation over multiple cycles while others degrade, crack, or delaminate—degradation patterns are themselves sequence-dependent and meaningful.

Minimum success threshold. The project is designed to be meaningful even with partial success. Two expressing variants with distinguishable humidity response is sufficient to support the core hypothesis. If dramatic bending does not occur, differences in wrinkling, swelling, coating integrity, or surface texture still constitute evidence that sequence programs material behavior.

Technical Risks and Fallbacks

Expression failure. If one variant does not express, a two-variant comparison is still sufficient. ELPs are well-established expression targets; complete failure of all three is unlikely.

Purification difficulty. Partially purified fractions can support preliminary coating tests. This limitation will be clearly stated.

Twist synthesis issues with repeats. ELP repeats are short, 15 bp per pentapeptide, and within Twist’s capabilities, but synthesis of highly repetitive regions can occasionally fail. Fallback: adjust codon usage to reduce exact repeat identity while preserving the same amino acid sequence.

Diblock synthesis feasibility. The Aim 1b diblock constructs may be more challenging for Twist due to longer repetitive regions. These are explicitly marked as stretch goals pending feasibility confirmation.

8. References

• Srokowski et al., “Surface and adsorption characteristics of three elastin-like polypeptides with sequence length and guest residue variations.” Journal of Biomaterials Science, Polymer Edition. https://pmc.ncbi.nlm.nih.gov/articles/PMC3540362/
• Välisalmi et al., “Highly Hydrophobic Films of Engineered Silk Proteins by a Simple Fusion Protein Concept.” Langmuir. https://pubs.acs.org/doi/10.1021/acs.langmuir.2c03442
• Zhu et al., “Humidity-responsive self-assembly of short peptides with super-flexibility.” CrystEngComm. https://pubs.rsc.org/en/content/articlelanding/2024/ce/d4ce00031e
• Liu et al., “Spider dragline silk as torsional actuator driven by humidity.” Science Advances. https://www.science.org/doi/10.1126/sciadv.aau9183

9. Supply List and Budget

Estimated total: $377–$838 with the Gene Fragments route; up to ~$1,175 if using Clonal Genes. Costs depend on lab availability of shared reagents and equipment.