Similarly Somavarapu and Kepp constructed a PS homology

Similarly, Somavarapu and Kepp [29] constructed a PS1 homology model using 4HYG as a template, plus modeling the TM6-TM7 intracellular segment (amino acids 273 to 374). In this study, the authors considered for the first time the mature and immature (not autoproteolyzed) forms of PS1 component, since it is well known that it is related to the active and inactive forms of GS, respectively [30]. They performed three independent atomistic MD simulations of the mature and non-mature PS1 forms during 500 ns each (a total of 3 μs) with GROMACS 5.0 [31]. Their RMSD analysis showed that PS1 reached an equilibrium state after 300 ns, demonstrating that 500 ns is an acceptable time to perform a conformational analysis. The principal component analysis (PCA) of their last 200 ns revealed the high flexibility of TM2 and TM6 and identified significant changes in TM2 tilt angles during their simulations, directing a possible function of these two TMs as “gate doors” for C99 internalization and cleavage. This opening gate mechanism agrees with the TM2-TM6 movement observed by Kong and coworkers (vide supra). They also suggested that the hydrophilic loop (HL1) that connects TM1 and TM2 might control TM2 movement by acting as a gate hinge. Moreover, the MD simulations showed that GS maturation is responsible for the highly dynamic multi-states of PS1, explaining the importance of the auto-catalytic cleavage, for the C99 processing and the alterations in the cleavage pattern caused by PS1 mutations [30]. Their studies on the autoproteolytic maturation effect, and the binding poses of a GS inhibitor (Avagacestat [32], BMS-708163) in the catalytic subunit were approached by the authors to have a first approximation towards the consideration of these factors employing a MD approach.
Initial C99-APP binding site in GS For the internalization of C99 to the active site, first GS must recognize the intramembrane substrate to translocate it to the entry site. Two initial substrate cftr channel (docking sites) have been proposed by experimental assays, which involve: (1) TM2, TM6 and TM9 of PS1 (BS1), or (2) PS1-NTF, PEN-2 and NCT ECD (BS2) (see Fig. 1a). The first binding site was described through TM-swap PS1 mutants and cross-linking experiments carried out by Tomita's group [33]. Experimental results suggested that the proximity of TM2 and TM6 to TM9 and failures in substrate recognition of TM2-and 6-swapped mutants could indicate that these TMs constitute the initial binding site. However, recent substrate photocrosslinking experiments performed by Fukumori and Steiner [21] demonstrated that C99 binds first to NCT and PEN-2 exosites and then at PS1-NTF exosite for its further translocation toward the active site. To assess these initial substrate binding site hypotheses, Han and coworkers [34] performed a coarse-grained (CG) MD simulation study with the Martini v2.2 force field [35] and atomistic MD simulations employing a GS model (5FN2 structure) without the NCT ECD. As far as we know, this is the first study that performed MD simulations employing the GS cryo-EM structure. It is noteworthy that their 5FN2 model lacked the TM6-TM7 intracellular segment, which was further considered by Sovamarapu and Kepp in a later study (vide infra). In this work, the authors evaluated GS interactions with different substrates and studied the key role of TMs flexibility in the isolated substrate and the GS-substrate complexes. A detailed clustering analysis of CG simulations showed that GS substrates have high probability of binding first to BS1 with multiple poses that allows the cleavage site recognition by GS. Furthermore, their atomistic simulations were employed to evaluate GS-substrate stability and characterize the interface interactions between them. Meanwhile, Somavarapu and Kepp [36] performed MD simulations employing the 5FN3 cryo-EM structure and its previous PS1 homology model (vide supra) to construct a more complete GS model (Fig. 1c). Three independent MD simulations of their model were evaluated during 500 ns, retrieving three different PS1 conformational states of the opening-closing mechanism (closed, semiopen and open) based on the distance between the catalytic aspartics and TM2/TM3 (Fig. 1c). A previous statistical study reported by these authors showed that PS1 mutations related to Alzheimer's disease reduce the stability and compactness of GS induced by an increase polarity of the transmembrane region [37]. The structure conformations obtained from their simulations suggested that a semiopen and open state correspond respectively to the compact and loose form of the complex. Molecular docking of three different lengths of Aβ into the different PS1 conformations also suggested that GS substrate binding may be controlled by the PS1 conformations, showing better Aβ docking scores with the semiopen conformation. Therefore, a retention time model suggests that a semiopen (compact) state produces high affinity and longest substrate retention in the active site, which could lead to the formation of shorter Aβ isoforms. On the other hand, their lower substrate affinities and shorter retention time observed in open conformations could explain the production of Aβ42 in PS1 mutants. Additionally, the authors employed their dynamic structural model to give a probable substrate recruitment route towards the catalytic site based on a stepwise transfer mechanism of a bound substrate reported by Fukumori and Steiner.