Skip to content

Cassette Compatibility — Computational Analysis (comp-010)

1. Question

Does the uricase (Q00511) + lactoferrin (P02788) payload pair have any cassette-design-specific issues — codon collisions, KEX2 site geometry problems, or secretion-pathway burden — that the Ward 1995 glucoamylase-KEX2 fusion architecture will not handle out of the box in A. oryzae?

This question sits between the Killshot #1 literature audit (which confirmed the dual-cassette Ward architecture is precedented in A. oryzae submerged culture) and §1.9 of validation-experiments.md (the first wet-lab feasibility test). comp-010 is design support — not a killshot, not a replacement for §1.9. It asks whether there are any sequence-level landmines in the proposed payload pair that should be designed around before gene synthesis begins.

2. Verdict

Overall cassette-design risk: LOW (Mechanistic Extrapolation; in silico only)

No blocking cassette-design issues identified for the proposed asymmetric architecture — direct-secretion for uricase, glucoamylase-KEX2 fusion for lactoferrin. The payload pair is architecturally compatible with the Ward 1995 / Huynh 2020 design as-is, with two actionable design notes and one informational finding.

Design notes (not blocking, but act before gene synthesis):

  1. Lactoferrin KEX2 internal sites: One moderate-risk internal K-R dipeptide at mature position 579 (P1'=K — cleavage reduced but possible). One non-functional site at mature position 38 (P1'=D — cleavage abolished). Monitor by SDS-PAGE in §1.9.
  2. Uricase C-terminal SKL motif: Resembles a peroxisomal PTS1 signal. Unlikely to override the amyB secretion signal in the fusion context, but verify by ELISA fractionation vs. subcellular localization in §1.9.

Informational finding (not load-bearing for proposed architecture):

  • Uricase contains one high-risk internal K-R site at residue 128 (P1'=N). This is irrelevant for the proposed direct-secretion cassette design but would require attention if uricase is later moved to a glucoamylase-KEX2 fusion.

3. Why This Matters

The koji endgame strain thesis requires two heterologous expression cassettes co-integrated in A. oryzae. Both payloads are validated independently:

  • Uricase: A. flavus uaZ expressed in S. cerevisiae = rasburicase (FDA 2001); parent gene well-characterized.
  • Lactoferrin: Human hLf expressed in A. awamori at >2 g/L (Ward 1995); correct fold confirmed at 2.2 Å (Sun 1999).

But combining them in the same strain introduces payload-pair-specific risks that the individual precedents don't resolve: Will a KEX2 site in one payload cause aberrant cleavage of the other's fusion? Does the combined disulfide load saturate ER folding? Does the human-origin lactoferrin require extensive codon optimization that blocks expression?

comp-010 computes these risks from sequence alone before committing ~$600–1,000 to gene synthesis. If a blocking design issue had been found, it would restructure §1.9 before the first dollar is spent on gene synthesis. No blocking issue was found — the experiment proceeds as designed.

4. Method Summary

Seven analyses run on the protein sequences (Python stdlib only; reproducible):

Analysis Method Inputs
Codon usage CAI proxy (geometric mean RSCU of origin-preferred codons vs. A. oryzae table); rare-codon hotspot scan (≥3 consecutive RSCU < 0.4) Q00511.fasta, P02788.fasta, a_oryzae_codon_usage.json
KEX2 geometry Internal K-R dipeptide census; P1' residue scoring vs. canonical Kex2p preference rules Q00511.fasta, P02788.fasta, kex2_site_specs.json
Secretion targeting C-terminal KDEL/HDEL scan (ER retention); C-terminal SKL family (PTS1); N-terminal R/K-L-x5-H/Q (PTS2) Q00511.fasta, P02788.fasta
Disulfide load Cysteine count + UniProt-annotated disulfide bonds; folding load index = N_SS / 16 (Huynh 2020 baseline) Sequence + UniProt annotations
N-glycosylation N-X-S/T sequon scan (X ≠ P); cross-reference to UniProt annotations Q00511.fasta, P02788.fasta
Combined burden Synthesis of analyses 1–5; concurrent-load risk for the dual-cassette scenario All above
Huynh 2020 comparison Dimension-by-dimension comparison vs. adalimumab HC + LC (PMC7257131) Literature; above outputs

All evidence level: Mechanistic Extrapolation (in silico).

5. Key Results

5.1 Codon Usage

Protein Origin Burden label CAI proxy Rare-codon hotspots
Uricase Q00511 A. flavus (fungal) LOW 1.51 0
Lactoferrin P02788 Homo sapiens (mammalian) LOW (proxy) 1.45 0

The amino-acid-level proxy scores both LOW because human and A. flavus preferred codons broadly overlap A. oryzae preferences at this resolution. This is a known methodological artifact — at the actual CDS level, human cDNA uses AU-rich 3rd-position synonyms that A. oryzae disfavors. Standard practice for mammalian-origin genes in Aspergillus: full gene synthesis with A. oryzae codon optimization. This is not optional for lactoferrin; it is the expected gene-synthesis workflow regardless of this proxy result. Uricase (fungal origin) requires minimal or no codon optimization.

5.2 KEX2 Site Geometry

Protein Internal K-R sites High-risk (P1' preferred) Moderate-risk Low-risk (P1'=D/E) Overall
Uricase Q00511 1 1 (pos 128, P1'=N) 0 0 HIGH — but not load-bearing (direct-secretion design)
Lactoferrin P02788 2 0 1 (pos 579, P1'=K) 1 (pos 38, P1'=D) MODERATE

Uricase K-R site note. Residue 128 (K128-R129-N130) is a canonical KEX2 recognition sequence with preferred P1' context. If uricase were in a glucoamylase-KEX2 fusion, this site would be cleaved — producing a 128-aa truncated fragment (N-lobe) and a 174-aa C-terminal fragment. However, the §1.9 design puts uricase on a direct-secretion cassette (PTEF1-amyB_SP-uaZ-TgpdA) — no glucoamylase fusion, no KEX2 processing step, no internal-site risk. This finding is actionable only if uricase is ever moved to a fusion architecture.

Lactoferrin K-R site note. Two sites in the mature protein (post-signal-peptide): - Position 38 (K56-R57-D58 in full sequence): P1'=D — KEX2 cleavage abolished by acidic residue. Non-functional site. - Position 579 (K597-R598-K599 in full sequence): P1'=K — KEX2 cleavage reduced (~2–3× below baseline rate). Monitor by SDS-PAGE for minor truncated species (~67 kDa). If truncation is visible, mutate K597→Q597 (conservative, preserves charge environment without introducing K-R dipeptide).

5.3 Secretion Targeting

Protein Issues Routing risk
Uricase Q00511 C-terminal SKL (resembles PTS1 peroxisomal signal) MODERATE — verify in vivo
Lactoferrin P02788 None LOW

Uricase (A. flavus uaZ) ends in ...SKL — a canonical PTS1 peroxisomal targeting signal in fungi. However, PTS1 targeting requires the protein to lack an N-terminal secretion signal; in the §1.9 design, uricase is expressed with the amyB signal peptide, which should route it into the ER secretory pathway and override PTS1 routing. Verify by anti-uricase ELISA on secreted fraction vs. cell lysate in §1.9. If peroxisomal misrouting is confirmed (uricase in cell lysate only, absent from culture supernatant), append a short C-terminal tag (3× Ala or Gly) to mask the PTS1 signal.

5.4 Disulfide Load

Protein Cysteines Disulfides (UniProt) Folding load (vs. Huynh = 1.00) Fold risk
Uricase Q00511 3 0 0.00× VERY LOW
Lactoferrin P02788 33 16 1.00× MODERATE
Combined 36 16 1.00× —

Uricase is a disulfide-free enzyme — all three cysteine residues in the A. flavus sequence are free thiols, consistent with the intracellular homotetramer biology (no ER oxidative machinery required). This means the ER PDI/ERO1 burden of the dual-cassette strain is entirely attributable to lactoferrin.

Lactoferrin's 16 disulfides (1.00× Huynh 2020 adalimumab baseline) fall within the demonstrated capacity of the A. oryzae NSlD-ΔP10 ER folding machinery. Huynh 2020 achieved 39.7 mg/L functional full-length adalimumab (16 disulfides, two chains) — confirming that this host's ER can correctly fold ~16 disulfide bonds simultaneously in a secreted mammalian-origin glycoprotein. Lactoferrin's 16 disulfides are concentrated in a single chain rather than split across two, which slightly increases the per-chain folding complexity but does not change the total oxidative load. Note: LF disulfide count corrected from 17 to 16 per Notari 2023 (PMC10465537) — see chaperone-orthogonal-stacking.md §10.2. The architecture-adjusted effective PDI load for LF is 24–40 (16 disulfides × transferrin-lobe α = 1.5–2.5 per chaperone-orthogonal-stacking.md §3.5). (Mechanistic Extrapolation; source: chaperone-orthogonal-stacking.md)

5.5 N-Glycosylation

Protein N-X-S/T sites predicted UniProt-annotated Comments
Uricase Q00511 1 (NFS at pos 191) 0 NFS likely unoccupied (fungal intracellular enzyme; no glycosylation documented)
Lactoferrin P02788 3 (NWT at pos 137, NQT at pos 478, NGS at pos 623) 3 (N137, N478, N623) All three predicted sites confirmed by UniProt; match Sun 1999 crystal structure

All three lactoferrin glycosylation sites are correctly predicted by the N-X-S/T scanner and confirmed by UniProt annotation. A. oryzae will produce high-mannose fungal N-glycans at these sites rather than the complex sialylated/fucosylated glycans of native milk lactoferrin. This difference has been characterized: recombinant Aspergillus-produced hLf with fungal glycans is ~40× less immunogenic and ~200× less allergenic than native milk hLf in BALB/c mice (Almond 2012, PMID 23012214) — and the Sun 1999 2.2 Å crystal structure confirmed the native fold is preserved with fungal glycans in A. awamori-produced Lf. Glycan divergence is likely a feature for the chronic oral dosing use case, not a concern.

5.6 Combined Secretion-Pathway Burden

Axis Uricase Lactoferrin Combined Huynh 2020 baseline
Disulfides 0 16 16 16
N-glycosylation ~0 (unoccupied) 3 3 2
Codon burden LOW (fungal) LOW (proxy) / full-opt in practice — HEAVY (both mammalian)
Folding load index 0.00 1.00 1.00× 1.00

OE dual-cassette burden is essentially equal to or slightly below the Huynh 2020 adalimumab precedent. Uricase contributes zero disulfide load (unlike adalimumab light chain which contributed 4–6 of the 16 disulfides). The combined glycosylation is 3 sites (slightly above Huynh's 2 Fc sites). Overall: the ER processing machinery faces a comparable or lighter folding load than it handled for adalimumab at 39.7 mg/L.

5.7 Huynh 2020 Baseline Comparison

Dimension OE pair Huynh 2020 adalimumab
Protein origins Fungal + Mammalian Mammalian + Mammalian
Total disulfides 16 (all on Lf) 16 (8 per chain)
Glycosylation sites 3 (Lf only) 2 (Fc N297 ×2)
Host strain NSlD-ΔP10 (recommended) NSlD-ΔP10 (required)
Architecture AmyB-KRGGG-Lf + direct uricase AmyB-KRGGG-HC + AmyB-KRGGG-LC
Titer (achieved / target) Target: 500 mg/L Lf + 100 mg/L uricase Achieved: 39.7 mg/L

Easier than Huynh: - Uricase (fungal origin) requires no codon optimization. Both adalimumab chains were human-origin and required full optimization. - Uricase contributes zero disulfide load. Adalimumab light chain added ~4–6 disulfides to the combined ER load.

Harder than Huynh: - The H01 Lf target (500 mg/L) is 12.6× the Huynh 2020 adalimumab titer (39.7 mg/L). However, the correct benchmark for lactoferrin is Ward 1995 (>2 g/L, A. awamori submerged with strain improvement) — not the Huynh antibody titer, which reflects antibody-specific dual-chain assembly constraints. 500 mg/L is well within the Ward 1995 demonstrated range for this exact protein class. - Solid-state format (primary OE target) adds format-translation risk. Huynh 2020 was submerged only. This is the dominant unresolved risk — not addressed by comp-010 (in silico) and remains load-bearing for §1.9.

Comparable to Huynh: - Same NSlD-ΔP10 host, same KRGGG linker KEX2 architecture, same dual-locus integration strategy. - Lactoferrin's per-chain disulfide complexity (~16 SS bonds in one chain) is analogous to adalimumab heavy chain's disulfide density in the Huynh 2020 system.

6. Design Recommendations for §1.9

  1. Host strain: NSlD-ΔP10 (default, per H01 Killshot #1 findings). Wild-type RIB40 is insufficient.

  2. Uricase cassette: Direct-secretion design (PTEF1 or PamyB :: amyB signal peptide :: uaZ :: TgpdA). Do NOT use glucoamylase-KEX2 fusion for uricase (internal KR128 site would cause truncation). Codon optimization optional but not required.

  3. Lactoferrin cassette: Ward 1995 / Huynh 2020 exact design (PamyB :: glucoamylase :: KRGGG :: hLf (A. oryzae codon-optimized) :: TamyB). Full codon optimization mandatory — order as codon-optimized synthetic gene from Twist or IDT, specifying A. oryzae or high-GC fungal optimization.

  4. Monitor Lf KEX2 site (pos 579): Run SDS-PAGE on secreted fraction; a ~67 kDa band indicates truncation at K579-R580. If seen: mutate K597→Q (full-sequence position) in the codon-optimized gene design.

  5. Monitor uricase routing (C-terminal SKL): Confirm uricase is in the secreted fraction (anti-uricase ELISA on culture supernatant). If primarily intracellular, add C-terminal 3×Ala linker to mask PTS1.

  6. Iron supplementation: FeCl₃ or iron citrate at 10–50 ppm added to rice substrate. Rice grain bioavailable iron (~1–3 ppm) may be insufficient for high-titer Lf folding; Ward 1995 used defined iron supplementation in submerged culture.

  7. Submerged parallel control: Run solid-state koji and submerged DPY in parallel. Solid-state format risk is the load-bearing unresolved question; the parallel control isolates format-specific failure modes from architecture-specific ones.

7. Limitations

  1. Amino-acid-level CAI proxy, not CDS analysis. Codon burden scores reflect residue-level analysis only. At CDS level, human cDNA uses synonymous codons that A. oryzae disfavors — lactoferrin requires full gene synthesis with codon optimization regardless of the LOW proxy score. This is a methodological floor, not a green light.

  2. KEX2 P1' rules are S. cerevisiae-derived. A. oryzae kexB specificity is not independently published with a full P1' matrix. Rules are inferred from Kex2p family homology. The KRGGG linker validation in A. oryzae (Huynh 2020) is the strongest available evidence for internal-site risk estimation.

  3. No structural accessibility correction for KEX2 sites. Internal KR sites buried in high-pLDDT domains have lower cleavage probability. This analysis assumes all KR sites are equally accessible — a conservative (worst-case) estimate. The lactoferrin moderate-risk site (pos 579) is in the C-lobe; if it is buried (high pLDDT in AF model), actual cleavage risk is lower than scored here.

  4. No quantitative ER capacity model. Whether simultaneous expression of both cassettes at target titers (500 mg/L Lf + 100 mg/L uricase) saturates A. oryzae ER folding machinery (BiP, PDI, ERO1) is empirically unknown. No published quantitative model exists for dual-heterologous-protein secretion in A. oryzae.

  5. Solid-state format not modeled. All sequence-level findings apply equally to submerged and solid-state formats. The Sun 2024 caveat (some proteins fail to secrete in solid-state despite secreting in submerged) is a format risk, not a sequence risk — comp-010 cannot resolve it.

  6. Codon table represents genome-wide average, not secreted-protein-specific preferences. RSCU estimates carry ~±15% error relative to highly-expressed secreted gene preferences.

8. Impact on §1.9 Priority

comp-010 does not reframe §1.9 as a confirmation experiment. §1.9 remains a feasibility gate — the primary open question is the solid-state format translation, which is a wet-lab question that no sequence-level analysis can answer. What comp-010 does:

  • Removes cassette architecture as a pre-experiment concern. No blocking sequence-level issues mean the §1.9 design (direct secretion uricase + glucoamylase-KEX2 lactoferrin in NSlD-ΔP10) can proceed as specified without redesign.
  • Provides two actionable design notes (KEX2 site 579 monitoring; C-terminal SKL verification) that should be built into the §1.9 readout plan before wet-lab starts.
  • Anchors the Huynh 2020 titer comparison. The 12.6× titer gap vs. Huynh adalimumab is not the correct benchmark for Lf — Ward 1995 >2 g/L Lf is. The Huynh baseline is relevant for confirming ER folding capacity; it is not a ceiling for Lf titer.

9. Cross-References

All findings in this analysis are Mechanistic Extrapolation — in silico sequence analysis. Wet-lab confirmation required for all design recommendations.