HTGAA Homework #2 — Genetic Construct Design

Part 1: Expression target

Molecule(s) expressed: This is a two-component sensing circuit, not a single gene — three coding sequences working together:

TtrS — membrane-bound sensor histidine kinase that detects extracellular tetrathionate
TtrR — cognate cytoplasmic response regulator, phosphorylated by TtrS
amilCP — blue/purple chromoprotein reporter, expressed only when the tetrathionate-responsive promoter is activated by phosphorylated TtrR

Sequences and provenance:

TtrS/TtrR: originate from the Shewanella-derived two-component tetrathionate-sensing system characterized and ported into E. coli by Daeffler et al. (2017, Molecular Systems Biology) and used for in vivo gut monitoring by Riglar et al. (2017, bioRxiv). I'm treating these as literature-described parts rather than a specific registry accession, since I don't have a verified, directly-citable sequence record for this exact TtrS/TtrR pairing independent of the papers' own supplementary data. The exact sequences still need to be pulled from the Daeffler et al. supplementary files before I treat them as final.
amilCP: a real, verified registry part — BBa_K592009, 669 bp including a tandem stop codon (221 aa coding, from Acropora millepora), sourced from the iGEM Registry of Standard Biological Parts (originally provided by Jeffrey Miller, UCLA; iGEM Registry, Part:BBa_K592009). I confirmed this sequence translates in a single clean reading frame with no internal stops before using it. I chose amilCP specifically because its absorbance maximum (588 nm, blue/purple) is visible to the naked eye without instrumentation — directly serving the "stool-readable without a lab" design goal from HW1.

Host/chassis: E. coli Nissle 1917 (EcN), carried over from HW1 — a clinically-used probiotic strain, the same chassis Riglar et al. validated for long-term engineered-circuit stability during actual mouse gut colonization.

Subcellular localization: TtrS is membrane-embedded by nature (it must face the periplasm/extracellular space to detect tetrathionate — no engineered signal peptide needed, its native transmembrane domains handle localization). TtrR and amilCP are both cytoplasmic — TtrR because it needs to access cytoplasmic DNA to act as a transcription factor, and amilCP because there's no reason to secrete a visual reporter that's read out after lysis/extraction from stool rather than in situ. No signal peptide is needed for any of the three parts — this simplifies the construct relative to a secreted-product design.

Part 2: Genetic parts

Using the prokaryotic-host part layer from the assignment table, and reflecting the decision that the sensing circuit uses constitutive expression of TtrS/TtrR, with an inducible (TtrR-activated) promoter driving only the reporter:

Part type	Short name	Sequence source	Justification
Promoter (constitutive)	J23101	Anderson promoter collection (iGEM registry standard constitutive promoter family)	Drives constitutive TtrS/TtrR expression — these sensor components need to be present at baseline, ready to detect tetrathionate whenever it appears; a medium-strength constitutive promoter (J23101 is a commonly-used mid-strength Anderson promoter) avoids both under-expression (missed signal) and gratuitous metabolic burden from over-expression
RBS	B0034	iGEM registry standard strong RBS (Elowitz 1999-derived, the same RBS bundled with the amilCP registry part itself)	Strong, well-characterized translation initiation — minimizes a second source of uncertainty in the circuit beyond the sensing logic itself
CDS	TtrS	Daeffler et al. 2017 / Riglar et al. 2017 (literature-described; exact sequence to be pulled from supplementary data)	The sensor histidine kinase — the actual tetrathionate-detection component, with no viable substitute given this is the molecule HW1's entire diagnostic logic depends on
CDS	TtrR	Daeffler et al. 2017 / Riglar et al. 2017 (literature-described; same sourcing caveat as TtrS)	The response regulator that converts TtrS's detection event into a transcriptional output — required as the other half of the two-component system
Terminator	B0015 (double terminator)	iGEM registry standard double terminator (commonly paired with Anderson promoters)	Prevents read-through transcription from the constitutive TtrS/TtrR cassette into the downstream inducible reporter cassette, which would otherwise cause reporter expression independent of actual tetrathionate sensing — this is not a cosmetic choice, it's necessary for the circuit's logic to be clean
Promoter (inducible)	PttrB	Literature-described TtrR-activated promoter (Daeffler et al. 2017)	The actual sensing-to-output coupling point — only transcribes the reporter when TtrR has been phosphorylated by TtrS in response to tetrathionate
RBS	B0034	Same strong RBS as above, reused for the reporter cassette	Consistency — isolates reporter signal strength to promoter activation state rather than confounding it with a second, different RBS efficiency
CDS	amilCP	BBa_K592009, iGEM Registry (verified, 669 bp incl. stop, 221 aa, Acropora millepora chromoprotein)	Visible-to-the-eye blue/purple output readable without instrumentation — directly matches HW1's stool-readable diagnostic framing
Terminator	B0015 (double terminator)	Same as above	Clean transcriptional termination of the reporter cassette
Backbone	pTwist Kan Medium Copy	Carried over from existing project design (Twist Bioscience vector)	Provides kanamycin selection marker and a medium-copy origin of replication — medium copy avoids the metabolic burden on EcN that a high-copy plasmid could impose (important since this strain also needs to survive and function while colonizing a competitive gut environment), while still giving enough copies for a detectable reporter signal

Containment note (not a Kernel "part" in the strict sense, but essential to the full design): the ΔdapA auxotrophy from the existing project design is a genomic chassis modification, not a plasmid-borne part — it's engineered into the EcN strain itself (deletion of the native dapA gene), requiring exogenous diaminopimelic acid (DAP) for cell wall synthesis. This means the strain cannot proliferate once DAP supplementation stops (e.g., after excretion from the gut), addressing biocontainment. I'm noting this explicitly in the write-up even though it isn't an insert in the expression plasmid itself, since Part 1 of the assignment specifically asked where the product needs to go and containment is a necessary part of a diagnostic organism's overall design logic.

[Image: plasmid map showing the two-cassette construct in pTwist Kan Medium Copy]

Part 3: Assembled construct

5′→3′ order (two cassettes in one plasmid, in the pTwist Kan Medium Copy backbone):

[Backbone: ori + KanR] — J23101 — B0034 — TtrS — [short spacer] — TtrR — B0015 — PttrB — B0034 — amilCP — B0015 — [back to backbone]

Design notes:

TtrS and TtrR are placed under the same constitutive promoter as a short operon (common in two-component system design, since the stoichiometric ratio between sensor and regulator affects circuit sensitivity — keeping them co-transcribed is a reasonable simplification for a first-pass design, flagged as a tunable variable rather than a fixed decision).
The reporter cassette (PttrB → B0034 → amilCP → B0015) is placed downstream and insulated by the first terminator, so the inducible promoter's low baseline activity isn't affected by read-through transcription from the upstream constitutive cassette.
Reading frame check (performed directly, as above): the amilCP CDS translates cleanly in a single frame, start codon ATG, 221 aa, properly terminated by a tandem stop (TAA-TAA), no internal stops — verified computationally rather than assumed. The TtrS/TtrR sequences would need the same verification once pulled from the literature source, which I'm flagging as an outstanding check rather than claiming I've already done it (I haven't retrieved their exact sequences).
Construct map: Not yet available

Part 4: Optimize and quality-check

Codon optimization: amilCP's registry sequence is coral-derived (Acropora millepora) but was already provided in a form usable in E. coli registry contexts — still, I would run Kernel's codon-optimization tool against E. coli codon usage tables specifically for the EcN chassis, since coral and bacterial codon usage bias differ substantially. What changes and why it matters: codon optimization doesn't change the protein sequence, only the synonymous codon choices — but E. coli has strong codon usage bias (some codons are read by abundant tRNAs, others by rare ones), and using codons matched to abundant host tRNAs avoids ribosome stalling, premature termination, and reduces the metabolic cost of translation. For TtrS/TtrR, this matters even more, since these are literature sequences from a different source organism's two-component system context and almost certainly need E. coli codon optimization to express efficiently in EcN specifically. Constraints I'd apply: avoid introducing restriction sites used elsewhere in the assembly strategy (if using Golden Gate, avoid internal BsaI sites; if using the BioBrick-compatible registry context amilCP came from, avoid EcoRI/XbaI/SpeI/PstI, consistent with why the registry version already had an EcoRI site removed), and keep GC content within a range that avoids secondary structure formation in the RBS region specifically, per B0034's own registry documentation note that upstream/downstream secondary structure can affect RBS efficiency.

Sequence search / BLAST: Running amilCP's sequence against NCBI/registry databases, I'd expect a 100% identity match to the Acropora millepora chromoprotein sequence (registry part BBa_K592009) — this is a natural gene sequence, not a novel design, deliberately chosen because it's a well-characterized, previously-validated part rather than something de novo. For TtrS/TtrR, I'd expect the closest match to be the native Shewanella two-component system sequences characterized in Daeffler et al., since I'm not proposing any mutations to these — my design reuses validated sensing logic rather than attempting to engineer a novel sensor, which is the appropriate scope given I have no experimental basis to improve on a published, mouse-validated circuit.

Safety screen: Kernel's biosecurity screen checks constructs against sequences associated with regulated pathogens or toxins (typically screening against something like the International Gene Synthesis Consortium's hazard sequence list). What I'd expect it to report: no flags — none of these three genes (a coral chromoprotein, a Shewanella sensor kinase, a Shewanella response regulator) have any known association with pathogenicity or toxin production; all three are standard synthetic-biology circuit components already validated for safe lab and even in vivo (mouse) use in the cited literature. If something were flagged, my response would depend entirely on what: a false-positive homology hit to an unrelated hazard sequence would warrant manual BLAST comparison to confirm it's spurious, whereas a genuine hit would require re-examining whether the part is actually what I believe it to be (sourcing error) before proceeding.

Part 5: From design to build

How would I build it? Given the construct's modest size — the full two-cassette insert (TtrS + TtrR + regulatory parts + PttrB + amilCP + regulatory parts) is well under 5 kb, comfortably within the range of a single commercial gene synthesis order — I would order the full insert as synthetic DNA from a commercial vendor (e.g., Twist Bioscience, consistent with the existing pTwist backbone choice) rather than building it from smaller fragments via Gibson assembly or oligo-pool synthesis. Kosuri et al. (2010) and Tian et al. (2004) both describe chip-based oligo synthesis and pool assembly as valuable specifically for very high-throughput, many-variant synthesis — producing thousands of sequence variants in parallel — which is the wrong tool for a single, specific, already-literature-validated circuit where I'm not screening sequence space, just assembling one defined design. Gibson assembly (Gibson et al., 2009) becomes the better choice for much larger constructs (tens to hundreds of kb) or when combining several independently-sourced large fragments; at under 5 kb for a single defined sequence, direct full-construct synthesis is simpler, has a faster turnaround, and avoids introducing assembly-junction errors that Gibson's enzymatic overlap-based method, while robust, still needs to be verified against after the fact.

What does synthesis constrain? The main design elements that could be synthesis-difficult: GC-content extremes (neither amilCP nor the registry parts I'm using are flagged as GC-extreme) and repetitive sequence (the double terminator B0015, being a standard hairpin-forming terminator by design, could in principle pose a minor synthesis/assembly challenge since strong secondary structure is exactly what makes terminators work, but this is a very well-characterized, commonly-synthesized registry part with no reported synthesis issues). Carr et al. (2004) describe protein-mediated error correction (using mismatch-binding proteins like MutS to remove synthesis errors from oligo pools) as valuable specifically for de novo, high-error-rate chip-synthesized oligo pools — again, not directly relevant to my approach of ordering a single verified sequence from a commercial vendor, who already performs their own internal QC/error-correction as part of standard synthesis service. If I were instead pursuing the TtrS/TtrR sequences via a fragment/PCR-based approach (e.g., if I couldn't get a clean supplementary sequence and needed to amplify from a Shewanella source), error-correction logic would become directly relevant — another reason resolving the TtrS/TtrR sequence-sourcing gap flagged in Part 1 matters before actual synthesis.

Prototype vs. scale: mirroring the HW1 answer directly — at prototype stage, I need only a handful of clonal plasmid preps (single synthesis order → transformation → a few colonies screened for correct amilCP expression and functional tetrathionate response, consistent with the Opentrons-based variant-screening plan from HW1). At the HW1-defined production scale (14 L culture for a 100-person validation cohort), the DNA construct itself doesn't need to be re-synthesized at larger scale — once a validated clone is selected, plasmid DNA is simply propagated by growing more of the same validated strain, not by re-ordering more synthetic DNA. This is an important distinction: DNA synthesis scale and culture/production scale are decoupled once a working construct exists — the synthesis step is a one-time design cost, not a recurring production cost, which is favorable for the economics of this project.

Part 6: What don't I know yet?