MD Simulations

Nine runs: controls, truncation ladder, 6xHis tag, pH 5.5

All nine trajectories share one receptor, one force field, one equilibration schedule and one analysis codebase. Four things change between them: which peptide is bound, where it starts and stops, what sits on its C-terminus, and which histidines carry a charge. The map is the shape of those four choices. BoPEP4 sits at the centre; each limb is one thing we changed, and each chip on a limb is a trajectory that changed it.

The map needs JavaScript. Without it, the nine reports are listed below, grouped the same way.

Nine reports, one self-contained file each

Each report is a single self-contained file with its own 3D trajectory player and its own shareable link, so any one of them can be sent to a reader on its own. Whichever one you open, the navigation bar across the top will bring you back here.

Four columns change: ligand, residue range, C-terminus, protonation

Whether two runs can be compared directly is a question of how many columns apart they sit in this table. One column apart is a controlled comparison. Three columns apart is two experiments at once, and the set contains one of those on purpose: 9-23 at pH 5.5 crosses a truncation with a protonation change, and it is the run that failed.

A muted cell is the same as the wild type.

Four results that need all nine runs

Backbone RMSD tracks the salt bridge at Spearman +0.29

Eight of the nine runs carry a peptide, which makes this answerable with a number instead of an anecdote. Backbone RMSD over the shared 15-23 window, the kind of score constructs usually get ranked on, correlates with the bidentate Asn23-Arg487 salt bridge at Pearson +0.13 and Spearman +0.29. Against the four-contact clamp it is +0.03 the wrong way. Contact count does better and not by much: +0.26 and +0.55 respectively. Eight trajectories are not eight independent samples, so read these as descriptive, not inferential.

The practical consequence is on this page twice. Ranked on RMSD the 6xHis construct comes first in the set, and it is the one construct whose salt bridge is gone. Ranked on RMSD the failed run comes second.

The 15-23 columns count only the C-terminal nine residues, the one stretch present in all eight peptide-bearing systems. There are no whole-peptide averages here because the peptides run from 9 to 29 residues, and an average over them is not measuring the same thing twice. The apo row is mostly dashes for the same reason: with no peptide there is no clamp, no salt bridge and no contact count.

Ten receptor residues stay above 50% in all eight runs

Pool the per-residue contact occupancies across all eight peptide-bearing runs and ten receptor residues stay above 50% in every one of them, whatever the peptide is doing at the other end. Five of the ten are on the list of eight hotspots Tang and colleagues read off the AtPep1 complex; five are not. The five that are not are Ser275, Lys345, Tyr395, Phe419 and Ile463. Tang's list was assembled from one crystal structure of a different peptide, and this is what it looks like after eight trajectories of a second one.

Apo floor: receptor Cα 1.57 Å, groove 35.91 ± 0.55 Å

The apo run exists to answer a question the other eight cannot: how much of the receptor motion they report would have happened anyway. Its receptor Cα RMSD is 1.57 Å, which is below all seven peptide-bearing runs it can be compared with. Those seven span 1.58 to 2.16 Å. Swapping one peptide for another therefore moves the receptor across a 0.59 Å range, and removing the peptide altogether moves it less than that. The binding groove holds at 35.91 ± 0.55 Å with nothing in it, and the per-residue flexibility profile correlates with the bound runs at r = 0.95. Any receptor-side difference smaller than about half an ångström on the other eight pages is inside this floor.

pH 5.5 puts Glu12 on His227 at 2.75 Å in 99.9% of frames

Switching His14 and His22 on the peptide and His227 on the receptor to their protonated form, with every other atom identical, locks the Glu12 carboxylate onto the His227 ring at 2.75 ± 0.11 Å in 99.9% of frames. At pH 7 the same pair sits 5.33 Å apart and appears in 0.8%. The AtPep1 benchmark already makes that contact at neutral pH, at 3.36 Å and 76.9%, so what pH 5.5 does for BoPEP4 is recover a contact its own calibrating control has natively. The C-terminal anchor is untouched and the 15-23 window becomes the tightest of any BoPEP4 run. Future BoPEP4 simulations should default to pH 5.5 with pH 7 as the control.

The His14 pH-switch idea that motivated the run is not supported. Protonated or not, its ring nitrogens sit about 6.1 Å from the nearest receptor carboxylate, at 0.0% occupancy. The pH-sensing pair is receptor His227 with peptide Glu12.

ff19SB / OPC, 4 fs HMR, 30 ns on one A100

Every system was built on the PEPR1 ectodomain from PDB 5GR8, residues 29 to 738, and solvated in OPC water with AMBER ff19SB. Hydrogen mass repartitioning allows a 4 fs time step. Each system ran 30 ns of production after the same equilibration schedule, on one A100-SXM4-40GB at roughly 250 ns per day, for 25.9 GPU-hours and 13,500 saved frames in total. Systems are about 127,600 atoms. Every number on this page and on the nine reports comes from one analysis codebase run over all nine trajectories, so the columns mean the same thing across rows.

30 ns, one seed, nothing tested at the bench

Thirty nanoseconds is three to six orders of magnitude short of a dissociation event, so nothing here is a binding affinity, a leaving time or a free energy. The 15-23 run was designed to supply a leaving time and could not have, whatever it did. Each system has one seed, so a difference between two runs may be a difference between two relaxation paths. The acceptance gate was written before the tagged construct existed and, on that construct, measures atoms it does not have. And none of it has been checked at the bench: no binding assay, no thermal shift, no plant. Everything on this page is a hypothesis about what to build and what to measure next.

Where this feeds the rest of the project: the sequence that goes into the order is argued on the Sprints and Results page, and the release and dosing arithmetic is on the Math Model page.