The n16N-3 system

5.4.1 PLUM*

A REMD simulation of n16N-3 was carried out for 6.4 microseconds, using 55 replicas. The thermostatted temperatures were Ti 2{275.0, 275.9, 276.81, 277.71, 278.61, 279.52, 280.42, 281.32, 282.22, 283.13, 284.03, 284.93, 285.84, 286.74, 287.64, 288.55, 289.45, 290.35, 291.25, 292.16, 293.06, 293.96, 294.87, 295.77, 296.67, 297.58, 298.48, 299.38, 300.28, 301.19, 302.09, 302.99, 303.9, 304.8, 305.7, 306.61, 307.51, 308.41, 309.31, 310.22, 311.17, 312.21, 313.35, 314.64, 316.15, 318.09, 321.15, 324.77, 328.38, 331.99, 335.6, 339.21, 342.82, 346.43, 350.0_} K.

42799 frames were used for analysis of the system. The geometric clustering analysis was carried out with an RMSD cut-o↵ of 0.8 nm, and the results show a tendency towards a single, strongly favoured top cluster has emerged. The top four clusters had populations of 7.0%, 1.5%, 1.3% and 0.93%, and these are shown in fig. 5.9. The dominant core type seen in n16N-2 PLUM* (fig. 5.1a), involving -hairpins starting at SD1, has no analogue in the present system; instead, chains hook into each other, ending the hook with a turn at the end of SD2. In the top cluster, two chains hook the other chain together.

The regional clustering analysis showed very little change compared to the n16N-2 system. SD2 became even less flexible, its top cluster now representing a single strand of -structure, turning at the SD2/SD3 boundary, as seen in all the top full-system clusters, with a population of 72.1%. The two other subdomains remained as before, SD3 staying the most flexible.

The degree to which each residue of the chain is involved in interpeptide in- teractions shows the same trend in the tripeptide system compared to the dipeptide; this is shown in fig. 5.11a. The increased level of hydrogen bonding in many residues in SD1 and SD2 highlights the lower level seen in residues K4 and K5 of SD1, and P15 and Y16 at the end of SD2, both of which are turning points in the chain in every full-system and regional top cluster.

The Ramachandran plot of the system is shown in fig. 5.10, as well as a di↵erence plot comparing to n16N-2 in PLUM*. The shift towards -structure and away from ↵-structure has continued as the increased number of peptides has made -strands running alongside each other increasingly favourable. The chain- hooking form of aggregation which dominates in this system is made of -strands, and extended conformations which occupy territory on a Ramachandran plot.

5.4.2 PRIME20-like

REMD simulations were carried out on the n16N-3 system in PRIME20-like in a large box. 48 replicas were used to evenly span the temperature range [0.105,0.250], and each replica executed approximately 7⇥1010 events.

12338 trajectory snapshots at T⇤ = 0.135 were used for analysis. The full- chain clustering analysis in fig. 5.12 used an RMSD cut-o↵of 0.65 nm. The analysis shows that the chains have some ability to extend and fold together, rather than being purely fixed in a collapsed coil state. Unlike the PLUM* model, no strongly favoured structure has emerged yet. Additionally, no interpeptide interaction speci- ficity on the level of subdomains or residues can be seen.

(a) 1 (7.0%) The grey and red chains run with parallel -strands from their K4 residues, turning at the end of SD2 and di- verging in SD3. The blue chain encloses the two SD2 regions, again turning at the end of its SD2. Each chain’s interpeptide inter- actions last approximately until its final ty- rosine.

(b) 2 (1.5%)The grey and blue chains form a core similar to n16N-2’s fig. 5.1b. The third chain floats above, involving its SD2 in the core interpeptide interactions.

(c) 3 (1.3%) The grey and blue chains loop around each other, turning at the end of SD2, and being involved in interpeptide interactions throughout SD2 and early SD3. The red chain is largely self-interacting, hav- ing a -hairpin turning about residue G7, while SD3 features an ↵-helix. The red chain’s SD2 is involved in interpeptide side- chain interactions with both other chains.

(d) 4 (0.93%) Similar to 5.9c and 5.9b in having two chains loop around each other and a third nearby but outside of the loop. Turns occur consistently at residues K4 and K5, and at the end of SD2.

Figure 5.9: The four top-occurring structures for the n16N-3 system in PLUM* at 300.0K. Each structure is labelled by its rank, with the percentage population given

in brackets. N-termini are highlighted in yellow. At an RMSD cut-o↵ of 0.8 nm,

this analysis coarsely groups frames in which chains are similarly positioned, with little discrimination based on local intrapeptide secondary motifs.

Figure 5.10: Ramachandran plots of the n16N-3 system. Left shows a standard

Ramachandran heat map. The system is strongly -structure dominated. Right

shows a di↵erence heat map, comparing the( , ) coordinates visited with those of

the n16N-2 system in PLUM*. The map is similar to the di↵erence map of n16N-1

and n16N-2, shown in fig. 5.3, though the magnitude of the changes is far smaller. in an interpeptide interaction. The shift from a dimer system to a trimer system has not elicited any greater di↵erentiation of regions from the chains on this metric. Although both lines’ global minima exist around the SD2/SD3 boundary, the valley here is less pronounced than before. There is no evidence of distinct subdomains from this figure.

The Ramachandran plot of the system is shown in fig. 5.13, as well as a di↵er- ence plot comparing to n16N-2 in PRIME20-like. The di↵erences between this and smaller PRIME20-like systems remain minor compared to the large rearrangement seen with the PLUM* model.