Methodological Decision Log

Decision: MACS will use HALDSLE repeats instead of HALKSLE.

Reason: The intended pH-switch mechanism depends on Asp/Glu-assisted histidine pKa engineering. Aspartate provides a local negative electrostatic environment that may increase histidine protonation propensity near the target pH threshold around 6.8. Lysine would introduce a positive charge and may oppose histidine protonation, weakening the intended histidine-zipper mechanism.

Alternative considered: HALKSLE, which appeared in earlier sequence files and AlphaFold input.

Interpretive note: The intended pH-switching role of the HALDSLE environment remains a design hypothesis. Sequence-level rationale and computational analysis do not substitute for experimental validation.

Decision: For pH 6.8, do not protonate all histidines. Use a selective borderline histidine set based on PROPKA pKa values and structural inspection.

Primary pH 6.8 candidates: A:282, A:289, A:296, B:272, B:293, B:324, B:329, B:332.

Reason: These residues have PROPKA-predicted pKa values near the target pH range and/or are structurally close to the HALDSLE/interface region.

pH 7.4 rule: Histidines will remain neutral unless a strong residue-specific reason exists.

Methodological note: Static PROPKA on an initial structural model provides heuristic guidance rather than dynamic, solvent-coupled pKa values in solution.

Decision: All histidines in the pH 7.4 model were assigned as neutral HIE.

Reason: pH 7.4 is above the predicted pKa values for most histidines. Therefore, doubly protonated HIP was not used.

Alternative considered: Mixed HID/HIE tautomer assignment. Deferred for simplicity; this model uses a uniform neutral HIE assignment.

Methodological note: HIE/HID tautomer choice can affect local hydrogen-bonding networks but represents a secondary variation within a neutral histidine ensemble.

Decision: pH 6.8 topology was generated with selective HIP assignments: A:282, A:289, A:296, B:272, B:293, B:324, B:329, B:332. All other histidines were assigned as neutral HIE.

Reason: Selected residues are borderline pKa/interface/motif candidates based on PROPKA and structural inspection.

Methodological note: Topology generation finalized the discrete charge model for subsequent production sampling.

Decision: The pH 6.8 model uses selective HIP assignments for eight borderline histidines.

Reason: PROPKA/Henderson–Hasselbalch analysis indicates that these residues have higher protonation propensity at pH 6.8 than at pH 7.4. The average predicted protonated fraction for the selected set is substantially higher at pH 6.8 than at pH 7.4, supporting a pH-dependent protonation gradient.

Alternative considered: Constant-pH MD, which would allow dynamic protonation/deprotonation. This is outside the current compute and software scope.

Decision: Generate 300 ns production TPR files for both pH 6.8 and pH 7.4 conditions using GROMACS 2026.1 for primary production sampling (~600 ns total).

Setup sanity check: Before long production runs, both prepared systems completed a short local post-equilibration sanity check following EM/NVT/NPT preparation. This check was used to detect gross setup instability and was not treated as evidence of convergence or mechanistic validity.

Sampling structure: Primary production sampling consists of one primary trajectory of approximately 300 ns per pH condition (pH 6.8 and pH 7.4) under classical fixed-protonation conditions, yielding approximately 600 ns total primary sampling. No completed independent production replicas were generated. Trajectory length was selected for exploratory conformational sampling and does not imply statistical convergence.

Output policy: TRR output remains disabled. Compressed XTC output is used for trajectory storage.

Decision: Use the final 100 ns of each primary production trajectory for a matched PB-MMPBSA endpoint comparison between Module A and Module B.

Method: Analyzed the final 100 ns window (approximately 200–300 ns) using 1001 frames per condition at 100 ps frame spacing. Module A was treated as receptor and Module B as ligand using gmx_MMPBSA 1.6.5 with AmberTools 23.6 PB solver.

Result: Across 1001 frames per condition, mean PB-MMPBSA ΔTOTAL was +61.43 kcal/mol at pH 6.8 (SD 59.38; SEM 1.88) and +71.15 kcal/mol at pH 7.4 (SD 41.03; SEM 1.30). The mean difference, pH 6.8 − pH 7.4, was −9.72 kcal/mol.

Interpretation: Within this endpoint-energy model, the pH 6.8 ensemble has a lower mean Module A–Module B ΔTOTAL than the pH 7.4 ensemble. This is directionally consistent with a pH-dependent difference in the modeled interface energetics.