Week 2: Wiring a Formaldehyde Alarm — Circuit Design in Kernel

The Circuit Behind the Color Change

Part 0: Set up in Kernel

Workspace: HTGAA 2026 · Repository / Folder: Dominic Orozco_TX-TL

Part 1: Define Your Expression Target

What molecule am I expressing? Two proteins, working together as a sense-and-report pair, rather than an enzyme that builds a material directly. FrmR is the sensing element — a transcriptional repressor. mCherry is the reporter — a red fluorescent protein that provides the visible/measurable output. Neither protein synthesizes the physical fiber itself (that’s a materials-science step, not a biological one); the biology’s job here is purely informational: turn “formaldehyde present” into “visible red signal.”

What is the sequence, and where does it come from?

• FrmR — the native E. coli K-12 gene, sourced from the UniProt entry FRMR_ECOLI (Swiss-Prot). Because this is the same organism as my chassis, no cross-species codon adaptation is conceptually needed (see Part 4).
• mCherry — a monomeric red fluorescent protein engineered by directed evolution from DsRed, a naturally pigmented protein from Discosoma sea anemones (Shaner et al., 2004). I did not import mCherry as a standalone sequence; I sourced it, already correctly positioned downstream of its regulated promoter, from the pre-assembled Registry part BBa_K2257007 (see Part 2–3).

What host or chassis? A cell-free system: a freeze-dried E. coli transcription–translation (TX-TL) extract, with no living cells present at the point of use. In practice, though, the plasmid DNA itself still has to be propagated and purified from living E. coli first — there’s no way around that step — so E. coli is both the indirect DNA-production host and the source organism for the cell-free extract itself.

Where does the product need to go? Cytoplasm — both FrmR and mCherry act intracellularly (or, more precisely, in-tube, inside the cell-free reaction volume). No secretion or membrane targeting sequence is needed, since the readout has to be read directly from the reaction/fiber, not exported anywhere.

Part 2: Design Your Genetic Parts

No standalone signal peptide was added — the product stays cytoplasmic, so this layer is intentionally absent from the design.

This design continues a long-running strategy in synthetic biology of repurposing natural bacterial repressor proteins as engineered logic elements — the same basic move used in one of the field’s earliest synthetic circuits, a repressor-based oscillator built from TetR, LacI, and λ cI (Elowitz & Leibler, 2000), and later systematized into a general method for mining whole families of natural repressors as circuit parts (Stanton et al., 2014). Here, FrmR plays that role, but its “logic” was borrowed pre-built from nature rather than mined computationally.

Part 3: Assemble Your Construct

The construct is a single circular plasmid, assembled in Kernel from two blocks: BBa_K2257007 (1,407 bp) and the pUC-AmpR v4 backbone (2,077 bp), for a total of 3,484 bp — matching the sum exactly, a first sanity check that no bases were accidentally duplicated or dropped when the two blocks were joined.

Read 5’→3’ within the sensor module (all elements in the same orientation):

PfrmR (promoter + RBS) → frmR (CDS) → FLAG tag → PfrmAB (promoter + RBS) → mCherry (CDS) → Terminator

…sitting on top of the backbone’s own ori + AmpR cassette, which retains its own internal, independent orientation. I confirmed in Kernel’s sequence viewer that both frmR and mCherry begin with a valid start codon (ATG) and end with a valid stop codon, in-frame, and that the two module junctions (backbone → sensor module, and sensor module back to backbone) have no gaps or overlapping bases.

Part 4: Optimize and Quality-Check

Codon optimization. I ran Kernel’s codon optimizer separately on each CDS for E. coli. In both cases, the resulting protein sequence was translated and compared back against the pre-optimization protein: identical, 0 amino-acid changes in both frmR+FLAG (101 aa) and mCherry (236 aa) — exactly as expected, since codon optimization is only supposed to swap synonymous codons, never change the encoded protein. The only structural change in either sequence was the addition of one extra codon at the very end — a second, in-frame stop codon placed immediately after the first (e.g., mCherry’s stop went from a single TAA to a tandem TGA-TAA). This is a deliberate synthetic-biology safety practice: a tandem stop still terminates translation correctly even if a future point mutation disables the first stop codon, preventing unwanted ribosomal read-through into downstream sequence. The optimizer otherwise consistently shifted codons toward E. coli’s most-used synonyms for a given amino acid (e.g., CGT over CGC/CGA/AGA for Arginine, AAA for Lysine, CAG for Glutamine) — this matters because translation speed depends on how much matching tRNA the host has available for a given codon; rare codons can stall or drop ribosomes mid-transcript, reducing yield or producing truncated/misfolded protein.

Sequence search (pairwise alignment). I ran a pairwise alignment for each CDS’s translated protein against its natural reference, rather than aligning DNA-to-DNA (which would falsely penalize synonymous codon differences):

• frmR+FLAG vs. UniProt FRMR_ECOLI (P0AAP3): 90.10% identity, but critically, 0 mutations and 91/91 matches — my 101-aa sequence covers the entire 91-aa native FrmR reference with zero amino acid substitutions; the only difference is 10 extra residues at the C-terminus, which is exactly the FLAG tag (DYKDDDDK) plus a short linker that was deliberately fused on. In other words, the sensing domain itself is unmodified, wild-type FrmR.
• mCherry vs. reference (“Designer Cells” construct): 81.00% identity (226/279 matches, 6 mutations, 4 deletions, 43 insertions). The low headline number is misleading on its own — inspecting the alignment shows the reference is 275 aa long, 39 residues longer than my 236-aa mCherry, and that extra C-terminal tail reads EGRGSLLTCGDVEENPGP..., which contains the diagnostic motif of a T2A/P2A viral self-cleaving peptide (...GDVEENPGP) followed by the start of a second, different protein. That means the “reference” I compared against isn’t a standalone mCherry — it’s a multi-protein expression cassette (mCherry-2A-X) from a different construct entirely, and the 43 “insertions” are simply that fused second half, not something missing from my design. Setting that fused tail aside, only a handful of scattered point differences remain, consistent with ordinary variation between different depositors’ mCherry sequences (there are several published mCherry variants with minor substitutions) — not evidence of an error in my construct, especially since my own mCherry independently matches the exact canonical 236-aa published sequence (Shaner et al., 2004) residue-for-residue.

Safety screen. Kernel’s biosecurity screen (SecureDNA) checks a sequence against a database of sequences associated with regulated or select-agent pathogens and toxins before it can be sent out for synthesis. Result: no flagged sequences detected. This is the expected outcome — none of frmR, mCherry, the pUC origin, or AmpR correspond to controlled bio-agent sequences, so a clean screen here confirms there’s nothing in this design that would block ordering synthesis on biosecurity grounds.

Part 5: From Design to Build

How would I build it? At ~3.5 kb and composed entirely of standard, non-repetitive parts, this construct is well within range for direct, clonal full-plasmid gene synthesis rather than requiring in-house fragment assembly. That said, Gibson assembly — isothermal, single-step joining of overlapping DNA fragments via a combined exonuclease/polymerase/ligase reaction (Gibson et al., 2009) — would be my fallback if I wanted to keep reusing this exact pUC-AmpR backbone across future sensor variants (e.g., swapping in a different reporter, or a different chemically-responsive promoter later): ordering backbone and sensor-module insert separately and Gibson-assembling them would let me mix and match modules without re-synthesizing the whole plasmid every time.

What does synthesis constrain? The terminator is deliberately a strong hairpin (that’s mechanically how rho-independent terminators work), and hairpin structures are a known source of synthesis and sequencing errors — exactly the class of error that protein-mediated correction methods, such as MutS-based mismatch filtration, were built to catch before assembly (Carr et al., 2004). A synthesis vendor would likely apply an equivalent error-correction step automatically for a region like this. Outside of the terminator, the design has no unusually repetitive stretches, extreme GC content, or long homopolymer runs that I’d expect to cause major synthesis difficulty.

Prototype vs. scale. For validating this one design, ordering the ~3.5 kb plasmid synthesized and clonally verified by sequencing a handful of colonies is the right prototyping approach — slow per construct, but low overhead for a single test. If I later want to systematically test many variants at once — different promoter strengths, alternative reporters, or other natural regulators mined the way Stanton et al. mined a whole family of prokaryotic repressors for orthogonal logic gates (Stanton et al., 2014) — the more efficient path would be chip-based, multiplexed oligo synthesis: the same parallel, microarray-derived DNA pool strategy described by Tian et al. (2004) for accurate multiplex gene synthesis, and later scaled up with selective amplification from high-fidelity microchip pools (Kosuri et al., 2010) or parallel microfluidic synthesis (Kong et al., 2007). That’s a genuinely different automation technology from what a single prototype needs — mirroring the prototype-vs-scale distinction from Homework #1.

Part 6: What Don’t I Know Yet?

1. Is the plasmid-encoded FrmR module even necessary, or does native FrmR carried over in the crude cell-free extract already regulate PfrmAB on its own? How I’d find out: run the cell-free reaction with a reporter-only variant (frmR deleted from the construct) side by side with the full construct, and compare formaldehyde-induced signal between the two. If the reporter-only version still responds, native extract FrmR is sufficient and the synthetic frmR cassette is redundant.
2. Does the freeze-dried, cell-free reaction survive being embedded in an electrospun fiber (high voltage, organic solvents during spinning)? How I’d find out: compare a fiber functionalized after spinning (reaction dispensed onto the finished mat) against one where the reaction is co-spun directly into the fiber, and measure whether formaldehyde-induced signal differs between the two — isolating whether electrospinning itself degrades the reaction.
3. Does the reporter’s sensitivity and response time on a solid fiber matrix match what’s been characterized in solution/plate format for BBa_K2257007? How I’d find out: run a formaldehyde dose-response curve in a sealed gas chamber, side by side, in a solution tube and on a spotted fiber sample, and compare the detection threshold and kinetics between the two formats.

References

Brophy, J. A., & Voigt, C. A. (2014). Principles of genetic circuit design. Nature Methods, 11(5), 508–520. https://doi.org/10.1038/nmeth.2926

Carr, P. A., Park, J. S., Lee, Y.-J., Yu, T., Zhang, S., & Jacobson, J. M. (2004). Protein-mediated error correction for de novo DNA synthesis. Nucleic Acids Research, 32(20), e162. https://doi.org/10.1093/nar/gnh160

Denby, K. J., Iwig, J., Bisson, C., Westwood, J., Rolfe, M. D., Sedelnikova, S. E., Higgins, K., Maroney, M. J., Baker, P. J., Chivers, P. T., & Green, J. (2016). The mechanism of a formaldehyde-sensing transcriptional regulator. Scientific Reports, 6, 38879. https://doi.org/10.1038/srep38879

Elowitz, M. B., & Leibler, S. (2000). A synthetic oscillatory network of transcriptional regulators. Nature, 403, 335–338. https://doi.org/10.1038/35002125

Gibson, D. G., Young, L., Chuang, R.-Y., Venter, J. C., Hutchison, C. A., & Smith, H. O. (2009). Enzymatic assembly of DNA molecules up to several hundred kilobases. Nature Methods, 6, 343–345. https://doi.org/10.1038/nmeth.1318

Kong, D. S., Carr, P. A., Chen, L., Zhang, S., & Jacobson, J. M. (2007). Parallel gene synthesis in a microfluidic device. Nucleic Acids Research, 35(8), e61. https://doi.org/10.1093/nar/gkm121

Kosuri, S., Eroshenko, N., LeProust, E. M., Super, M., Way, J., Li, J. B., & Church, G. M. (2010). Scalable gene synthesis by selective amplification of DNA pools from high-fidelity microchips. Nature Biotechnology, 28, 1295–1299. https://doi.org/10.1038/nbt.1716

Nguyen, P. Q., Soenksen, L. R., Donghia, N. M., Angenent-Mari, N. M., de Puig, H., Huang, A., Lee, R., Slomovic, S., Galbersanini, T., Lansberry, G., Sallum, H. M., Zhao, E. M., Niemi, J. B., & Collins, J. J. (2021). Wearable materials with embedded synthetic biology sensors for biomolecule detection. Nature Biotechnology, 39(11), 1366–1374. https://doi.org/10.1038/s41587-021-00950-3

Pardee, K., Green, A. A., Ferrante, T., Cameron, D. E., DaleyKeyser, A., Yin, P., & Collins, J. J. (2014). Paper-based synthetic gene networks. Cell, 159(4), 940–954. https://doi.org/10.1016/j.cell.2014.10.004

Shaner, N. C., Campbell, R. E., Steinbach, P. A., Giepmans, B. N. G., Palmer, A. E., & Tsien, R. Y. (2004). Improved monomeric red, orange and yellow fluorescent proteins derived from Discosoma sp. red fluorescent protein. Nature Biotechnology, 22(12), 1567–1572. https://doi.org/10.1038/nbt1037

Stanton, B. C., Nielsen, A. A. K., Tamsir, A., Clancy, K., Peterson, T., & Voigt, C. A. (2014). Genomic mining of prokaryotic repressors for orthogonal logic gates. Nature Chemical Biology, 10, 99–105. https://doi.org/10.1038/nchembio.1411

Tian, J., Gong, H., Sheng, N., Zhou, X., Gulari, E., Gao, X., & Church, G. (2004). Accurate multiplex gene synthesis from programmable DNA microchips. Nature, 432, 1050–1054. https://doi.org/10.1038/nature03151

Vieira, J., & Messing, J. (1982). The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencing with synthetic universal primers. Gene, 19(3), 259–268. https://doi.org/10.1016/0378-1119(82)90015-4

Part:BBa_K2257007 — composite formaldehyde-sensing device (PfrmR–frmR–PfrmAB–mCherry). Registry of Standard Biological Parts (iGEM). https://parts.igem.org/Part:BBa_K2257007

Part:BBa_K2728001 — pFrmR, an engineered formaldehyde-inducible promoter (BGIC-Global, 2018). Registry of Standard Biological Parts (iGEM). https://parts.igem.org/wiki/index.php/Part:BBa_K2728001

AI Use Disclosure

Tool: Claude (Anthropic, Sonnet 5), used through the chat interface.

Prompts (summarized; originally written in Spanish): discussed which E. coli formaldehyde-sensor parts were available in Kernel’s part libraries, asked for help interpreting search results and choosing between candidate parts (settling on BBa_K2257007 and the pUC-AmpR v4 backbone), asked what output the sensor part produces and what the backbone provides, asked for help verifying the assembled construct’s linear logic and orientation, asked for a prompt to verify the plasmid sequence with Kernel’s own AI assistant, and asked for this write-up to be restructured to follow Homework #2’s required sections (Parts 0–6), with reliable, cited supporting literature including the assigned course readings.

I selected the specific parts, built and inspected the construct directly in Kernel, ran (or will run) the codon optimization, BLAST, and safety-screen tools myself, and reviewed/edited the final write-up, table, and citations.