- Email: [email protected]
Recruitment of the amyloid precursor protein by γ-secretase at the synaptic plasma membrane
Accepted Manuscript Recruitment of the amyloid precursor protein by γ-secretase at the synaptic plasma membrane Martina Audagnotto, Alexander Kengo Lorkowski, Matteo Dal Peraro PII:
S0006-291X(17)32154-X
DOI:
10.1016/j.bbrc.2017.10.164
Reference:
YBBRC 38781
To appear in:
Biochemical and Biophysical Research Communications
Received Date: 6 July 2017 Revised Date:
4 October 2017
Accepted Date: 29 October 2017
Please cite this article as: M. Audagnotto, A. Kengo Lorkowski, M. Dal Peraro, Recruitment of the amyloid precursor protein by γ-secretase at the synaptic plasma membrane, Biochemical and Biophysical Research Communications (2017), doi: 10.1016/j.bbrc.2017.10.164. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Recruitment of the Amyloid Precursor Protein by γ-Secretase at the Synaptic Plasma Membrane
RI PT
Martina Audagnotto#, Alexander Kengo Lorkowski#, Matteo Dal Peraro*
Institute of Bioengineering, School of Life Sciences, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne 1015, Switzerland and Swiss
These authors equally contributed
M AN U
#
SC
Institute of Bioinformatcs (SIB), Lausanne 1015, Switzerland
AC C
EP
TE D
* Correspondence to [email protected]
1
ACCEPTED MANUSCRIPT Abstract Γ-secretase is a membrane-embedded protease that cleaves single transmembrane helical domains of various integral membrane proteins. The amyloid precursor protein (APP) is an important substrate due to its The mechanism of the
RI PT
pathological relevance to Alzheimer’s disease.
cleavage of APP by γ-secretase that leads to accumulation of Alzheimer’s disease causing amyloid-β (Aβ) is still unknown. Coarse-grained molecular
SC
dynamics simulations in this study reveal initial lipids raft formation near the catalytic site of γ-secretase as well as changes in dynamic behavior of γ-
M AN U
secretase once interacting with APP. The results strongly suggest the environmental precursor to APP binding and hints at conformational changes
1. Introduction
TE D
of γ-secretase in the nicastrin (NCT) domain upon APP binding.
The γ-secretase is a membrane-embedded protease that cleaves single transmembrane helical domains of various integral membrane proteins. Of
EP
particular interest is the sequential cleavage of the amyloid precursor protein
AC C
(APP) into variable length peptides, commonly named amyloid β (Aβ). This cleavage occurs in two steps and two proteases are involved: β-secretase first removes the ectodomain of APP and further the γ-secretase cleaves the remaining C-terminal fragment within its transmembrane (TM) domain delivering a mixture of 37 to 49 amino acid long Aβ peptides. In particular, longer Aβ peptides (i.e., Aβ42) are linked closely to Alzheimer’s disease (AD) due to their propensity to aggregate and form extracellular senile plaques in the brain [1]. Alternatively, cell surface APP can be cleaved by α-secretase to
2
ACCEPTED MANUSCRIPT release the non-amyloidogenic soluble APPα, which was observed to be neuroprotective [2]. γ-secretase is a 230 kDa complex with 20 TMs structured in four components: presenilin (PS1), presenilin enhancer 2 (PEN-2), nicastrin
RI PT
(NCT), and anterior pharynx-defective 1 (APH-1) (Figure 1) [3]. The fulllength PS is inactive and association with Pen-2 facilitates an autocatalytic cleavage of presenilin between TM6 and TM7, producing two fragments
SC
known as N-terminal fragment (NTF) and C-terminal fragment (CTF) [4-6]. Subsequently, NTF and CTF bind to form stable and active nine TMs PS1
M AN U
heterodimer [3], which it is actually the PS1 active catalytic form. At the interface between the NTF and CTF of presenilin is located the catalytic center composed by residues D257 and D385, located on TM6 and TM7 helices, which is excluded from the external surface of the enzyme. NCT is a
TE D
130 kDa type I TM protein that contains a large glycosylated ectodomain (ECD) and a single TM domain. Nicastrin is the scaffolding protein within the γ-secretase complex, and the ECD is proposed to act as substrate receptor.
EP
The other two γ-secretase components were initially identified through genetic
AC C
screening in C. elegans [7, 8]. PEN-2 directly binds PS1 and is required for its autocatalytic maturation and protease activity. APH-1 contains seven TM domains and is indispensable for γ-secretase assembly. Moreover, several studies showed that these four components are cross-regulated and downregulation or deficiency of one given component typically destabilizes the other altering the trafficking [9, 10].
3
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 1. Structure of human γ-secretase. The γ-secretase architecture and structure is
TE D
shown in surface and cartoon representation (PDB: 5A63): PS1 (orange), APH-1 (violet), PEN-2 (green) and NCT (light blue), with focus at the catalytic dyad and adjacent PAL motif. The PAL loop is highlighted in red and the close view shows its relative position compared to
EP
the catalytic residues.
AC C
During the past years, our knowledge on γ-secretase has largely increased. Residue level information on substrate docking to γ-secretase has been explored in previous biochemical experiments [11-14] suggesting that several residues scattered across the TM domains of presenilin 1 (PS1) constitute an extended binding surface for the initial substrate binding. Advancements in cryo-electron
microscopy
(cryo-EM)
have
resulted
in
high-resolution
structures of γ-secretase. A 3.4 Å cryo-EM reported 2015 (PDB: 5A63) provides a view on the atomistic organization of all four domains of the human
4
ACCEPTED MANUSCRIPT γ-secretase as well as the large ECD [3]. Despite the new insights from these studies, neither the mechanism of APP cleavage is fully understood nor a cocrystal structure of γ-secretase in complex with any of its substrates has been solved.
RI PT
Studies regarding γ-secretase nowadays aim at elucidating how APP is recognized and recruited. It has been suggested based on coarse-grained molecular simulation that the recognition and recruitment of substrate can
SC
occur on an extended surface covering PS1’s TM2/6/9 domains and PAL motif [15]. Translocation of the substrate to the active site is likely coupled
M AN U
with a large-scale conformational change. Indeed, it was suggested that a considerable conformational change in nicastrin (NCT) extracellular domain is required in order to allow the accessibility of the active site of PS. In particular, the heavily glycosylated ECD of NCT is thought to recognize the substrate N-
TE D
terminus [3] and redirect it to PS catalytic pocket for further cleavage processing. The cryo-EM structure at 3.4 Å of resolution (PDB: 5A63) does indeed provide density map for 10 out of 16 potential glycosylated sites [3];
EP
however, it is not currently known the function of these N-linked glycans and
AC C
further investigations are required. Recently, it was observed that the monomeric APP interacts with both NTF
and CTF and the flexibility of the APP influences the stability of these interactions [15]. Moreover, the non-covalent APP homodimer is protected from cleavage of γ-secretase independently from the dimerization motif [16]. Therefore, in this study we investigated the interplay between APP and γsecretase with particular attention to the specific membrane environment, namely, the effect of the lipids membrane composition.
5
ACCEPTED MANUSCRIPT Several experimental evidences associate lipid rafts to APP amyloidogenic processing. In fact, the β-secretase and γ-secretase enzymes as well as the full-length APP and the APP C-terminal fragments are localized in lipid raft domains [17-20], whereas other substrates (e.g. Notch1, Jagged2 or N-
domains
act
as
large
platforms
for
proteins
RI PT
cadherin) reside in non-raft membrane [18]. These results suggested that raft recruitment
selecting
amyloidogenic substrates. Therefore, based on lipidomics analysis of brain
SC
tissues, we modeled a realistic synaptic plasma membrane (SPM) [21], where we monitored lipids raft formation. The γ-secretase complex surrounded by
M AN U
APP proteins was embedded in this SPM model and a coarse-grained (CG) approach [22, 23] was used in order to investigate the recruitment and recognition of the APP by γ-secretase along with the raft formation with respect to the proteins’ localization. Our molecular simulations shed light on
TE D
the complex mechanism of γ-secretase APP recruitment by modeling a realistic membrane environment and by providing essential information for the
EP
further study of γ-secretase binding and inhibition modes of action.
AC C
2. Material and methods
2.1 Structural modeling. A 4.2 Å resolution cryo-EM structure (PDB ID: 5FN2) was used as the initial model for MD simulations of γ-secretase [24]. This structure was chosen, rather than the higher resolution PDB ID: 5A63 model, because it comprised of all four subunits of γ-secretase (Aph-1, Pen-2, PS1, and NCT); furthermore, γ-secretase is in complex with the nontransition-state analogue inhibitor DAPT, locking γ-secretase into a substratebinding mimic conformational state not represented in the 3.4 Å resolution
6
ACCEPTED MANUSCRIPT cryo-EM structure (PDB ID: 5A63) [3]. APP was taken from an NMR structure (PDB ID: 2LLM) that contains both the transmembrane domain and the extracellular kink [25]. Both atomic structures were coarse-grained using martinize [22].
RI PT
Γ-secretase and six randomly placed APP were embedded into a plasma membrane using insane [26]. The composition of the plasma membrane was constructed to be as similar to experimental lipidomics [21] (Table 1) as could
SC
be through the available MARTINI CG parameters [22, 23]. The dimension of
M AN U
the membrane across all replicas is 200 Å × 200 Å.
2.2 Coarse-Grained (CG) MD simulations. In the present study, all MD simulations were conducted using the GROMACS package [27]. These simulations were used to examine processes of substrate binding to γ-
TE D
secretase and to derive information about binding poses. All the systems were solvated in MARTINI CG water boxes, buffered with NaCl at 150 mM concentration.
EP
The polarized MARTINI 2.2 parameters were used for the simulations [22,
AC C
23]. All the production runs were preceded by a 5000-step energy minimization followed by a 100 ns NVT and 1 µs NPT equilibration with positions of protein beads being restrained. The temperature was set to 323 K using the V-rescale thermostat and the pressure was set to 1 bar using a semi-isotropic coupling method. A time step of 20 fs was used, a typical value employed in MARTINI simulations. Each simulation ran afterwards for 4 µs, the first 500 ns and the last 500 ns of which was used for analysis.
7
ACCEPTED MANUSCRIPT 2.3 Membrane thickness and order parameter. Two useful quantities to measure in membrane simulations are the membrane thickness and order parameters of the lipid chains. In particular, these observables are helpful in determining the presence of lipid rafts.
RI PT
Membrane thickness is here evaluated by first splitting the bilayer into its inner and outer leaflet. Each leaflet is overlaid on a grid of nodes on the xyplane where each node is equally spaced from each other. Each lipid assigns
SC
its z-position to the closest matching node. The thickness is obtained by subtracting each grid of nodes element-wise from each other. In this report,
M AN U
each node is set to have a neighborhood with the dimension of 5 Å by 5 Å. Common measurements of order in lipid bilayers are of the type that can be measured by deuterium NMR. Mathematically, lipid order parameters are defined as
1 (3 2
TE D
=
− 1)
where θ is the angle between a CD bond (in the experiment) or a CH bond (in
EP
the simulation) and the membrane normal. An order parameter of zero can describe either an unordered (isotropic) system or a perfectly ordered system
AC C
oriented at the magic angle of 54.7 with respect to the magnetic field. A value of –0.5 indicates a perfectly ordered acyl chain in all-trans conformation, rapidly rotating around the bilayer normal. Likewise, a value of 1 indicated perfect alignment with the bilayer normal.
2.4 Cross-Correlation. Dynamic cross correlation (DCC) analysis can be used to quantify the correlation coefficients of motions between atoms [28]. The DCC between the ith and jth atoms is defined by the equation:
8
ACCEPTED MANUSCRIPT ( , ) =
〈∆ ( ) ∙ ∆ ( )〉
〈‖∆ ( )‖ 〉 !〈"∆ ( )" 〉
where ri(t) denotes the vector of the ith atom’s coordinates as a function of time t, 〈∙〉 is the time ensemble average and ∆ ( ) = ( ) − ̅ where ̅ is the
RI PT
average vector of ith atom’s coordinates over the entire trajectory.
2.5 Principal Component Analysis. Principal component analysis (PCA) is a
SC
dimension-reduction procedure that reduces a large number of correlated
M AN U
variables into a smaller set of linearly uncorrelated variables called principal components (PC).
The PCs can be obtained by obtaining a set of
eigenvectors and eigenvalues from a covariance matrix that is mathematically defined as
,
1 Σ(x, y) = (() − )̅ )(* − *+) '−1
TE D
-.
The first PC has the largest eigenvalue whose physical meaning is the variance along the PC.
Each succeeding PC has the highest eigenvalue
Σ/0 = 1/0
AC C
EP
under the constraint that it is orthogonal to the preceding PCs.
9
ACCEPTED MANUSCRIPT 3. Results and Discussion 3.1 Lipid rafts form near the γ-secretase domain NCT and the active PS1. Lipid rafts are dynamic and highly ordered membrane micro-domains rich in cholesterol and sphingolipids. They are usually associated with thicker
RI PT
regions of the membrane compared to adjacent areas due to an ordering effect employed by saturated hydrocarbon chains of the lipids and cholesterol molecules [29]. These highly order micro-domains are believed to play an
SC
important role in cellular function serving as a platform for cellular processes like cell signaling, motility and protein sorting or trafficking [16, 29]. In
M AN U
particular, the combination of biochemical and magnetic immune-isolation experiments showed the coexistence of APP and γ-secretase substrates in these micro-domains [18, 30]. Moreover, raft association of γ-secretase is sensitive to acute cholesterol depletion and the absence or inhibitions of γ-
TE D
secretase activity resulted in the accumulation of APP in lipid raft domains [31]. Altogether, these results strongly suggested that γ-secretase cleavage of APP occurs in raft domains.
EP
Therefore, we first built a model of the SPM including 32 different lipids
AC C
species observed in brain tissue of AD patients [21]. This model of the SPM reflects the total lipids distribution in the presence of the AD showing an increment in sphyngolipids, ganglioside and cholesterol compared to a health brain tissue. These lipids anomalies are potentially linked with AD pathogenesis and therefore this lipidomics analysis represents a valid starting point for studying the interplay between γ-secretase and APP. Our SPM model, produced using the MARTINI force field, was thus composed of phosphatydilcoline (PC, counting for a 21% of the total lipid content),
10
ACCEPTED MANUSCRIPT phosphatidylethanolamine
(PE,
11%),
phosphatidic
acid
(PA,
0.5%),
sphyngomyelin (SM, 6%), negatively charged phosphatidylserine (PS, 2.5%) and phosphatidylinositol (PI, 2%), cholesterol (55% of the SPM) and ceramide (CE) monosialodihexosylganglioside (GM3) made up the remaining 2%. The
RI PT
lipid tails included saturated palmitoleic, polyunsaturated arachidonic and docosahexaenoic acids (Table 1).
AC C
EP
TE D
PIPC (16:0/18:2) POPC (16:0/18:1) DPPC (16:0/18:0) PAPC (16:0/20:4) PEPC (16:0/20:2) PGPC (16:0/20:1) PUPC (16:0/22:6) PRPC (16:0/24:6) PIPE (16:0/18:2) POPE (16:0/18:1) DPPE (16:0/18:0) PAPE (16:0/20:4) PQPE (16:0/20:3) PGPE (16:0/20:1) PUPE (16:0/22:6) PRPE (16:0/24:6) POPA (16:0/18:1) PGPA (16:0/20:1) PGPS (16:0/20:1) PRPS (16:0/24:6) POPI (16:0/18:1) PVPI (16:0/18:1) PAPI (16:0/20:4) PUPI (16:0/22:6) PPC (16:0/18:0) POSM (18:1/18:1) PVSM (18:1/18:1) DPSM (18:1/18:0) PGSM (18:1/22:1) PNSM (18:1/24:1) DPCE (18:1/18:0) DPG3 (18:1/18:0) CHOL
M AN U
SC
Table 1. Lipids distribution of the synaptic plasma membrane models of AD’s brain tissue. The total relative abundance of the different lipids species is given for each lipid species used. Each name corresponds to the conventional name found in the MARTINI force field.
11
SM (%) 0.36 (3) 6.93 (57) 2.55 (21) 2.92 (24) 1.58 (13) 3.53 (29) 2.07 (17) 1.95 (16) 0.12 (1) 0.49 (4) 0.12 (1) 0.49 (4) 0.12 (1) 0.97 (8) 2.43 (20) 6.20 (51) 0.24 (2) 0.24 (2) 0.73 (6) 1.82 (15) 0.24 (2) 0.24 (2) 1.46 (12) 0.24 (2) 0.36 (3) 0.36 (3) 0.36 (3) 4.26 (35) 0.12(1) 1.22 (10) 0.36 (3) 0.12 (1) 54.74 (450)
ACCEPTED MANUSCRIPT In this context, we examined the lipids raft formation with respect to the position of the γ-secretase. The top view and a cut through the planar SPM model (Fig. 2a) revealed the coexistence of the 2 phases (liquid ordered: Lo and liquid disordered: Ld) in equilibrium. The final composition of the raft-
RI PT
domain (Lo) is on average 0.51:0:20:0:28 of cholesterol/unsaturated lipids/saturated lipids, respectively, while the final composition of the Ld phase is on average 0.35:0.45:0.19. The comparison between these two membrane
SC
regions showed roughly 2 times more cholesterol in the Lo phase compared to the Ld as well as an increase in unsaturated lipids in Ld as previously
M AN U
experimentally [32] and computationally [33] observed for a simple three components lipid bilayer. The difference in lipid composition between these 2 domains is explained by diverse thickness values (Fig. 2b) and average order parameter distribution (Fig. 2c), which altogether represent the key factors for
TE D
the lipid phase evaluation. Indeed, the high thickness values are associated with high order parameters peculiar of a Lo phase, while low thickness value and low order parameters indicates an Ld region. Moreover, the difference in
EP
thickness of roughly 6 Å between Lo and Ld is in agreement with atomic force
AC C
microscopy [34] and NMR [35] data as well as molecular dynamics simulations [36]. Although the average size of lipid rafts (~9 nm in diameter) is smaller compared to the experimentally estimated one (50 nm in diameter), our MD simulations clearly showed the initial formation of a raft-like region. Interestingly, all our MD simulations showed that a greater thickness is present on the convex side of the γ-secretase structure in close proximity of the extracellular part of NCT domain and PS1 (Fig. 2b). While the exact thickness distribution varies on the whole, this local feature is present in all
12
ACCEPTED MANUSCRIPT MD replicas independently from the APP position. Indeed, liquid order parameters around these regions (Fig. 2c) mirror the state of the membrane thickness (Fig. 2b): the membrane has a greater Lo phase near the convex
TE D
M AN U
SC
RI PT
side of the γ-secretase.
Figure 2. The NCT domain of γ-secretase is localized in a raft-like domain. (A) Top-
EP
view showing the position of the γ-secretase with respect to the lipid raft composed by saturated lipids and cholesterol (red). The side-view focus (right) displayed in detail the
AC C
repartition between Ld phase composed by unsaturated lipids (green) and the Lo phase mostly composed by cholesterol (represented in van der Walls) and saturated lipids (red). (B) The probability density of the thickness, averaged on the last 500 ns, as well as (C) the average order parameter distribution, over the last 500 ns, highlighted the lipids raft formation.
3.2 APP and lipid rafts. Although APP itself is generally a raft protein, where its cleavage by the secretases is hypothesized to occur, a small portion of
13
ACCEPTED MANUSCRIPT APP has been observed to be localizing into non-lipid rafts [31]. This APP raft localization has been suggested to be regulated by cholesterol [32], one of the main components of lipid raft micro-domains. Our MD simulations confirmed these experimental data showing that APP is indeed located in 2 different
RI PT
membrane associated pools: either (i) in a raft-like domain when it is far away from the γ-secretase, or (ii) in a non-raft-like domain when it interacts with the enzyme (Fig. 3A). In fact, the averaged cholesterol distribution with respect to
SC
the proteins shows that APP resides in high-cholesterol region when not in the vicinity of γ-secretase (violet dots in Fig. 3B), while when APP is close to the
M AN U
enzyme the distribution of cholesterol around the APP is lower (orange dot in Fig. 3B).
With 6 APP present in each MD simulation allowed to freely evolve in time, we expected multiple occurrences of dimer formation; however, this was not
TE D
the case. Analysis of the trajectory shows that cholesterol has the tendency to interact preferentially with APP at the G700XXXG704XXXG708 motif (Fig. S1). Indeed the dimerization process in our MD replicas (Fig. 4) eventually occurs
EP
only in the presence of low cholesterol concentration, in line with a previous
AC C
study where we observed that cholesterol interacts with the dimerization motif G700XXXG704XXXG708 [37]. In fact, it has been experimentally observed that the production of Aβ protein is enhanced under conditions of high cholesterol concentration [38, 39]. NMR chemical shift measurements of APP in micelles of varying cholesterol concentration and MD simulations of APP homodimer [40] support the competing formation of the APP-cholesterol complex with formation of APP dimer [41].
The results indicate that cholesterol-APP
interactions have a competitive advantage over the dimer structure formation.
14
ACCEPTED MANUSCRIPT The cholesterol-APP complex could potentially promote assimilation of APP into lipids rafts where it could then interact with γ-secretase. The nature of APP being a single TM helix surrounded by a number of cholesterol moieties should make its integration into lipid rafts more likely than a bulkier
RI PT
homodimer. Furthermore, these results suggest a possible relocalization mechanism mediated by cholesterol distribution. In fact, the pool of APP found in the raft-like domains is stabilized by the interaction with cholesterol,
SC
while close to the γ-secretase the APP-membrane stability decreased in order to facilitate the further expulsion of the Aβ products upon γ-secretase
EP
TE D
M AN U
cleavage.
AC C
Figure 3. The APP localizes in two different regions of the SPM model. (A) Representative replica showed that APP either localized in a raft-like region (violet dots) or in a non-raft region (orange dots) when APP more closely interacts with the γ-secretase. (B) Average normalized cholesterol distribution showed that APP resides in high-density cholesterol when far away from the enzyme.
3.3 Interactions between APP and γ-secretase: a one-to-one relationship that involves the NCT domain. APP is a type-I membrane protein that has
15
ACCEPTED MANUSCRIPT its extracellular part mostly removed by sheddases, as α- and β-secretases [14]. Nevertheless, it is still an open question how the γ-secretase selectively recognizes the shedded N-terminal substrates and recruits them for catalysis. It was experimentally observed that NCT has a central role in the APP-PS1
RI PT
interactions. Indeed, the NCT knock-down by RNAi causes a loss of FRET signal between APP and PS1 in cells [43]. In particular, it was recently proposed that the NCT binds the ectodomains of the substrates redirecting
SC
them to the PS1, the catalytic active site of γ-secretase. This interaction was suggested to be mediated by the residue E333 in the NCT extracellular
M AN U
domain [44]. However, a recent high-resolution γ-secretase structure revealed that E333 is actually buried and on the opposite sites of the catalytic dyad formed by D257 and D385 [3, 24] (Fig. 1). Therefore, the basic mechanism of substrate recognition by γ-secretase remains controversial and requires
AC C
EP
TE D
further investigation.
Figure 4. Stable G700XXXG704XXXG708 APP dimers where observed in the SPM model. Representative snapshot of the dimeric APP conformer observed in a complex SPM model. Distance between the center of mass of the two APP monomers showed the interaction pathway over the MD simulation time.
16
ACCEPTED MANUSCRIPT CG-MD simulations revealed that G700XXXG704XXXG708 APP dimeric conformation is the stable one in this complex SPM model (Fig. 4), confirming our previous study [37]. However, we could observe in our MD simulations that when in proximity to the γ-secretase, APP is always found in the
EP
TE D
M AN U
SC
RI PT
monomeric conformation (Fig. 5).
AC C
Figure 5. The interaction between APP and γ-secretase. Representative APP-enzyme interactions, based on two different MD replicas, as a function of time. The monomeric APP interacts mainly with the NCT domain and in particular with the TM (A) as well as the extracellular (B) regions. Distances measured are the smallest pair-wise distance between NCT and APP. (C) Cross-correlation matrix and (D) the first PC from the last 500 ns of MD trajectory shown in (A) with the extracellular component of NCT omitted in the calculation. The first PC resembles an opening and closing motion of the TM domain subunits of γsecretase.
17
ACCEPTED MANUSCRIPT We noticed that the flexibility of the N-terminal of APP seems to be important for the interaction with the enzyme. In the simulations where the APP structure was retained in its NMR based L-shape conformation by harmonic potentials between interacting sites, a substrate-enzyme interaction
RI PT
mode was never detected (Fig. S2). On the other side, when this bias was removed, a preferred APP interaction was observed for the NCT domain of the γ-secretase with a distance between APP and γ-secretase around 3.5 Å
SC
(Figs. 5A-B and S2). In particular, we observed a preference for the NCT extracellular part that was neglected in previous studies [15]. On the other
M AN U
side, we never observed the interaction of APP with the catalytic residues D257 and D385 (on average in all MD simulations the distance between APP and catalytic residues is
40 Å). In fact, the active site of PS1 is hardly
accessible on the convex site of the TM horseshoe, suggesting a
recruitment.
TE D
conformational reorganization of the nicastrin domain after substrate
Conformational reorganization is further supported by the cross-
EP
correlation analysis of our CG-MD trajectories (Fig. 5C). When APP binds
AC C
near the active site on γ-secretase (Fig. 5A), APH-1 becomes anti-correlated with the subunits PEN-2 and NCT. Along with the PCA (Fig. 5D), which showed a movement parallel to the membrane, the anti-correlation suggests an opening and closing motion of γ-secretase, which could lead to the large conformational change required to expose the catalytic aspartate residues to APP for cleavage, and will be further studied when other relevant conformations of the γ-secretase will be available.
18
ACCEPTED MANUSCRIPT
In conclusion, these MD simulations performed in a complex model of the SPM revealed the importance of lipids rafts like domains in the localization
RI PT
and stability of both the substrates and enzyme. In particular, we observed that APP tends to localize in two different regions of the synaptic plasma membrane, which vary mainly by their cholesterol concentration. This in turn
SC
suggests the key role of these sterol molecules in stabilizing the substrates before recruitment by the enzyme. Moreover, our computational study reveals
M AN U
insights into the role of the NCT domain for substrate recognition endorsing the importance of this extracellular domain. Although our study is preliminary and has to deal with all the known limitations associated with the approach we chose (i.e., CG MD simulations) and further investigations are surely required,
TE D
these results provide new structural information on the early stages of substrate-enzyme association process and its specific localization at the membrane.
EP
During the last 20 years, our knowledge of the AD’s neurobiology has
AC C
increased tremendously leading to the development of several putative disease-modifying treatments. Particular attention was recently raised for the disease–modifying therapies based on the inhibition or modulation of α-, βand γ-secretases [45-48]. γ-secretase notably not only cleaves Aβ peptides but many other transmembrane proteins, for instance Notch protein. This broad range of substrates is therefore the main cause of toxicity in preclinical tests of existing γ-secretase inhibitors. Therefore, exhaustive efforts have been made to design more specific APP/γ-secretase modulators. For
19
ACCEPTED MANUSCRIPT example, it was discovered that some small molecules [49] as well as nonsteroidal anti-inflammatory drugs (NSAIDs) might modulate the APP cleavage without interfering with the cleavage of other substrates. This initial study shed light on the early stages of APP recruitment and interaction mechanism with
RI PT
γ-secretase, which would be important for further drug design investigations. Altogether, our results provide a glimpse on the proteolytic processes involved in AD pathology representing also a promising starting point for future
M AN U
4. Acknowledgment
SC
investigations on the mechanism of γ-secretase inhibitors.
M.D.P. lab is supported by the EPFL and Swiss National Science Foundation
EP
5. References
TE D
(SNF grant No. 200020_157153).
[1] X. Zhang, Y. Li, H. Xu, Y.W. Zhang, The gamma-secretase complex: from
AC C
structure to function, Frontiers in cellular neuroscience, 8 (2014) 427. [2] S. Nizari, L. Guo, B.M. Davis, E.M. Normando, J. Galvao, L.A. Turner, M. Bizrah, M. Dehabadi, K. Tian, M. Francesca Cordeiro, Non-amyloidogenic effects of alpha2 adrenergic agonists: implications for brimonidinemediated neuroprotection, Cell death & disease, 7 (2016) e2514.
[3] X.C. Bai, C. Yan, G. Yang, P. Lu, D. Ma, L. Sun, R. Zhou, S.H. Scheres, Y. Shi, An atomic structure of human gamma-secretase, Nature, 525 (2015) 212-217.
20
ACCEPTED MANUSCRIPT [4] T. Ratovitski, H.H. Slunt, G. Thinakaran, D.L. Price, S.S. Sisodia, D.R. Borchelt, Endoproteolytic processing and stabilization of wild-type and mutant presenilin, J Biol Chem, 272 (1997) 24536-24541. [5] N. Takasugi, T. Tomita, I. Hayashi, M. Tsuruoka, M. Niimura, Y. Takahashi, G. Thinakaran, T. Iwatsubo, The role of presenilin cofactors in
RI PT
the gamma-secretase complex, Nature, 422 (2003) 438-441. [6] G. Thinakaran, D.R. Borchelt, M.K. Lee, H.H. Slunt, L. Spitzer, G. Kim, T. Ratovitsky, F. Davenport, C. Nordstedt, M. Seeger, J. Hardy, A.I. Levey, S.E. Gandy, N.A. Jenkins, N.G. Copeland, D.L. Price, S.S. Sisodia,
SC
Endoproteolysis of presenilin 1 and accumulation of processed derivatives in vivo, Neuron, 17 (1996) 181-190.
[7] R. Francis, G. McGrath, J. Zhang, D.A. Ruddy, M. Sym, J. Apfeld, M.
M AN U
Nicoll, M. Maxwell, B. Hai, M.C. Ellis, A.L. Parks, W. Xu, J. Li, M. Gurney, R.L. Myers, C.S. Himes, R. Hiebsch, C. Ruble, J.S. Nye, D. Curtis, aph-1 and pen-2 are required for Notch pathway signaling, gamma-secretase cleavage
of
betaAPP,
and
presenilin
protein
accumulation,
Developmental cell, 3 (2002) 85-97.
TE D
[8] C. Goutte, M. Tsunozaki, V.A. Hale, J.R. Priess, APH-1 is a multipass membrane protein essential for the Notch signaling pathway in Caenorhabditis elegans embryos, Proceedings of the National Academy of Sciences of the United States of America, 99 (2002) 775-779.
EP
[9] B. De Strooper, Aph-1, Pen-2, and Nicastrin with Presenilin generate an active gamma-Secretase complex, Neuron, 38 (2003) 9-12. T.
Iwatsubo,
The
gamma-secretase
complex:
machinery
for
AC C
[10]
intramembrane proteolysis, Current opinion in neurobiology, 14 (2004) 379-383.
[11] Y. Imamura, N. Watanabe, N. Umezawa, T. Iwatsubo, N. Kato, T. Tomita, T. Higuchi, Inhibition of gamma-secretase activity by helical beta-peptide foldamers, J Am Chem Soc, 131 (2009) 7353-7359. [12] A.Y. Kornilova, F. Bihel, C. Das, M.S. Wolfe, The initial substrate-binding site of gamma-secretase is located on presenilin near the active site, Proc Natl Acad Sci U S A, 102 (2005) 3230-3235. [13] S. Takagi-Niidome, T. Sasaki, S. Osawa, T. Sato, K. Morishima, T. Cai, T. Iwatsubo, T. Tomita, Cooperative roles of hydrophilic loop 1 and the C21
ACCEPTED MANUSCRIPT terminus of presenilin 1 in the substrate-gating mechanism of gammasecretase, The Journal of neuroscience : the official journal of the Society for Neuroscience, 35 (2015) 2646-2656. [14] N. Watanabe, I. Image, II, S. Takagi, I. Image, II, A. Tominaga, I. Image Image, T. Tomita, I. Image, II, T. Iwatsubo, I. Image Image, Functional
RI PT
analysis of the transmembrane domains of presenilin 1: participation of transmembrane domains 2 and 6 in the formation of initial substratebinding site of gamma-secretase, J Biol Chem, 285 (2010) 19738-19746. [15] S. Li, W. Zhang, W. Han, Initial Substrate Binding of gamma-Secretase:
SC
The Role of Substrate Flexibility, ACS chemical neuroscience, (2017).
[16] E. Winkler, A. Julius, H. Steiner, D. Langosch, Homodimerization Protects the Amyloid Precursor Protein C99 Fragment from Cleavage by gamma-
M AN U
Secretase, Biochemistry, 54 (2015) 6149-6152.
[17] D.R. Riddell, G. Christie, I. Hussain, C. Dingwall, Compartmentalization of β-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts, Current Biology, 11 (2001) 1288-1293.
[18] K.S. Vetrivel, H. Cheng, S.-H. Kim, Y. Chen, N.Y. Barnes, A.T. Parent,
TE D
S.S. Sisodia, G. Thinakaran, Spatial segregation of γ-secretase and substrates in distinct membrane domains, Journal of Biological Chemistry, 280 (2005) 25892-25900.
[19] J.M. Cordy, I. Hussain, C. Dingwall, N.M. Hooper, A.J. Turner,
EP
Exclusively targeting β-secretase to lipid rafts by GPI-anchor addition upregulates β-site processing of the amyloid precursor protein, Proceedings
AC C
of the National Academy of Sciences, 100 (2003) 11735-11740. [20] R. Ehehalt, P. Keller, C. Haass, C. Thiele, K. Simons, Amyloidogenic processing of the Alzheimer β-amyloid precursor protein depends on lipid
rafts, The Journal of cell biology, 160 (2003) 113-123.
[21] R.B. Chan, T.G. Oliveira, E.P. Cortes, L.S. Honig, K.E. Duff, S.A. Small, M.R. Wenk, G. Shui, G. Di Paolo, Comparative lipidomic analysis of mouse and human brain with Alzheimer disease, The Journal of biological chemistry, 287 (2012) 2678-2688. [22] D.H. de Jong, G. Singh, W.F. Bennett, C. Arnarez, T.A. Wassenaar, L.V. Schafer, X. Periole, D.P. Tieleman, S.J. Marrink, Improved Parameters for
22
ACCEPTED MANUSCRIPT the Martini Coarse-Grained Protein Force Field, Journal of chemical theory and computation, 9 (2013) 687-697. [23] S.J. Marrink, H.J. Risselada, S. Yefimov, D.P. Tieleman, A.H. de Vries, The MARTINI force field: coarse grained model for biomolecular simulations, J Phys Chem B, 111 (2007) 7812-7824.
RI PT
[24] X.C. Bai, E. Rajendra, G. Yang, Y. Shi, S.H. Scheres, Sampling the conformational space of the catalytic subunit of human gamma-secretase, Elife, 4 (2015).
[25] K.D. Nadezhdin, O.V. Bocharova, E.V. Bocharov, A.S. Arseniev,
SC
Structural and dynamic study of the transmembrane domain of the amyloid precursor protein, Acta naturae, 3 (2011) 69-76.
[26] T.A. Wassenaar, H.I. Ingolfsson, R.A. Bockmann, D.P. Tieleman, S.J.
M AN U
Marrink, Computational Lipidomics with insane: A Versatile Tool for Generating Custom Membranes for Molecular Simulations, J Chem Theory Comput, 11 (2015) 2144-2155.
[27] M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindahl, GROMACS: High performance molecular simulations through
(2015) 19-25.
TE D
multi-level parallelism from laptops to supercomputers, SoftwareX, 1–2
[28] J.A. McCammon, Protein dynamics, Reports on Progress in Physics, 47 (1984) 1.
EP
[29] J.B. Helms, C. Zurzolo, Lipids as targeting signals: lipid rafts and intracellular trafficking, Traffic (Copenhagen, Denmark), 5 (2004) 247-254.
AC C
[30] K.S. Vetrivel, H. Cheng, W. Lin, T. Sakurai, T. Li, N. Nukina, P.C. Wong, H. Xu, G. Thinakaran, Association of γ-secretase with lipid rafts in postGolgi and endosome membranes, Journal of Biological Chemistry, 279 (2004) 44945-44954.
[31] K.S. Vetrivel, G. Thinakaran, Membrane rafts in Alzheimer's disease beta-amyloid production, Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids, 1801 (2010) 860-867. [32] A.J. Beel, M. Sakakura, P.J. Barrett, C.R. Sanders, Direct binding of cholesterol to the amyloid precursor protein: An important interaction in lipid-Alzheimer's disease relationships?, Biochimica et biophysica acta, 1801 (2010) 975-982. 23
ACCEPTED MANUSCRIPT [33] M. Audagnotto, M. Dal Peraro, Protein post-translational modifications: In silico prediction tools and molecular modeling, Comput Struct Biotechnol J, 15 (2017) 307-319. [34] H.A. Rinia, M.M. Snel, J.P. van der Eerden, B. de Kruijff, Visualizing detergent resistant domains in model membranes with atomic force
RI PT
microscopy, FEBS letters, 501 (2001) 92-96. [35] S.P. Soni, D.S. LoCascio, Y. Liu, J.A. Williams, R. Bittman, W. Stillwell, S.R. Wassall, Docosahexaenoic acid enhances segregation of lipids between : 2H-NMR study, Biophysical journal, 95 (2008) 203-214.
SC
[36] P.S. Niemela, S. Ollila, M.T. Hyvonen, M. Karttunen, I. Vattulainen, Assessing the nature of lipid raft membranes, PLoS computational biology, 3 (2007) e34.
M AN U
[37] M. Audagnotto, T. Lemmin, A. Barducci, M. Dal Peraro, Effect of the Synaptic Plasma Membrane on the Stability of the Amyloid Precursor Protein Homodimer, J Phys Chem Lett, 7 (2016) 3572-3578. [38] R. Ehehalt, P. Keller, C. Haass, C. Thiele, K. Simons, Amyloidogenic processing of the Alzheimer beta-amyloid precursor protein depends on
TE D
lipid rafts, J Cell Biol, 160 (2003) 113-123.
[39] S. Wahrle, P. Das, A.C. Nyborg, C. McLendon, M. Shoji, T. Kawarabayashi, L.H. Younkin, S.G. Younkin, T.E. Golde, Cholesteroldependent
gamma-secretase
activity
in
buoyant
cholesterol-rich
EP
membrane microdomains, Neurobiol Dis, 9 (2002) 11-23. [40] E.T. Parkin, A.J. Turner, N.M. Hooper, Amyloid precursor protein,
AC C
although partially detergent-insoluble in mouse cerebral cortex, behaves as an atypical lipid raft protein, Biochemical Journal, 344 (1999) 23-30.
[41] Y. Song, E.J. Hustedt, S. Brandon, C.R. Sanders, Competition between homodimerization and cholesterol binding to the C99 domain of the amyloid precursor protein, Biochemistry, 52 (2013) 5051-5064.
[42] K.S. Vetrivel, H. Cheng, W. Lin, T. Sakurai, T. Li, N. Nukina, P.C. Wong, H. Xu, G. Thinakaran, Association of gamma-secretase with lipid rafts in post-Golgi and endosome membranes, J Biol Chem, 279 (2004) 4494544954. [43] O. Berezovska, A. Lleo, L.D. Herl, M.P. Frosch, E.A. Stern, B.J. Bacskai, B.T. Hyman, Familial Alzheimer's disease presenilin 1 mutations cause 24
ACCEPTED MANUSCRIPT alterations in the conformation of presenilin and interactions with amyloid precursor protein, The Journal of neuroscience : the official journal of the Society for Neuroscience, 25 (2005) 3009-3017. [44] G. Zhao, Z. Liu, M.X. Ilagan, R. Kopan, Gamma-secretase composed of PS1/Pen2/Aph1a can cleave notch and amyloid precursor protein in the
the Society for Neuroscience, 30 (2010) 1648-1656.
RI PT
absence of nicastrin, The Journal of neuroscience : the official journal of
[45] D. Levitan, I. Greenwald, Facilitation of lin-12-mediated signalling by sel12, a Caenorhabditis elegans S182 Alzheimer's disease gene, Nature,
SC
377 (1995) 351.
[46] B. P Imbimbo, G. AM Giardina, γ-secretase inhibitors and modulators for the treatment of Alzheimer's disease: disappointments and hopes,
M AN U
Current topics in medicinal chemistry, 11 (2011) 1555-1570. [47] T. Tomita, Secretase inhibitors and modulators for Alzheimer’s disease treatment, Expert review of neurotherapeutics, 9 (2009) 661-679. [48] F. Panza, V. Solfrizzi, V. Frisardi, C. Capurso, A. D’Introno, A.M. Colacicco, G. Vendemiale, A. Capurso, B.P. Imbimbo, Disease-modifying
(2009) 537-555.
TE D
approach to the treatment of alzheimer’s disease, Drugs & aging, 26
[49] A. Petit, F. Bihel, C.A. da Costa, O. Pourquié, F. Checler, J.-L. Kraus, New protease inhibitors prevent γ-secretase-mediated production of
AC C
507-511.
EP
Aβ40/42 without affecting Notch cleavage, Nature Cell Biology, 3 (2001)
25
ACCEPTED MANUSCRIPT
Raft-like micro-domains formation influences interactions between amyloid precursor protein (APP) and γ-secretase
2.
γ-Secretase nicastrin extracellular domain is playing a key role in the early recruitment of APP
3.
A realistic model of the synaptic plasma membrane is key to describe APP/γ-secretase interplay in molecular simulation
AC C
EP
TE D
M AN U
SC
RI PT
1.
S0006-291X(17)32154-X
DOI:
10.1016/j.bbrc.2017.10.164
Reference:
YBBRC 38781
To appear in:
Biochemical and Biophysical Research Communications
Received Date: 6 July 2017 Revised Date:
4 October 2017
Accepted Date: 29 October 2017
Please cite this article as: M. Audagnotto, A. Kengo Lorkowski, M. Dal Peraro, Recruitment of the amyloid precursor protein by γ-secretase at the synaptic plasma membrane, Biochemical and Biophysical Research Communications (2017), doi: 10.1016/j.bbrc.2017.10.164. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Recruitment of the Amyloid Precursor Protein by γ-Secretase at the Synaptic Plasma Membrane
RI PT
Martina Audagnotto#, Alexander Kengo Lorkowski#, Matteo Dal Peraro*
Institute of Bioengineering, School of Life Sciences, École Polytechnique Fédérale de Lausanne (EPFL), Lausanne 1015, Switzerland and Swiss
These authors equally contributed
M AN U
#
SC
Institute of Bioinformatcs (SIB), Lausanne 1015, Switzerland
AC C
EP
TE D
* Correspondence to [email protected]
1
ACCEPTED MANUSCRIPT Abstract Γ-secretase is a membrane-embedded protease that cleaves single transmembrane helical domains of various integral membrane proteins. The amyloid precursor protein (APP) is an important substrate due to its The mechanism of the
RI PT
pathological relevance to Alzheimer’s disease.
cleavage of APP by γ-secretase that leads to accumulation of Alzheimer’s disease causing amyloid-β (Aβ) is still unknown. Coarse-grained molecular
SC
dynamics simulations in this study reveal initial lipids raft formation near the catalytic site of γ-secretase as well as changes in dynamic behavior of γ-
M AN U
secretase once interacting with APP. The results strongly suggest the environmental precursor to APP binding and hints at conformational changes
1. Introduction
TE D
of γ-secretase in the nicastrin (NCT) domain upon APP binding.
The γ-secretase is a membrane-embedded protease that cleaves single transmembrane helical domains of various integral membrane proteins. Of
EP
particular interest is the sequential cleavage of the amyloid precursor protein
AC C
(APP) into variable length peptides, commonly named amyloid β (Aβ). This cleavage occurs in two steps and two proteases are involved: β-secretase first removes the ectodomain of APP and further the γ-secretase cleaves the remaining C-terminal fragment within its transmembrane (TM) domain delivering a mixture of 37 to 49 amino acid long Aβ peptides. In particular, longer Aβ peptides (i.e., Aβ42) are linked closely to Alzheimer’s disease (AD) due to their propensity to aggregate and form extracellular senile plaques in the brain [1]. Alternatively, cell surface APP can be cleaved by α-secretase to
2
ACCEPTED MANUSCRIPT release the non-amyloidogenic soluble APPα, which was observed to be neuroprotective [2]. γ-secretase is a 230 kDa complex with 20 TMs structured in four components: presenilin (PS1), presenilin enhancer 2 (PEN-2), nicastrin
RI PT
(NCT), and anterior pharynx-defective 1 (APH-1) (Figure 1) [3]. The fulllength PS is inactive and association with Pen-2 facilitates an autocatalytic cleavage of presenilin between TM6 and TM7, producing two fragments
SC
known as N-terminal fragment (NTF) and C-terminal fragment (CTF) [4-6]. Subsequently, NTF and CTF bind to form stable and active nine TMs PS1
M AN U
heterodimer [3], which it is actually the PS1 active catalytic form. At the interface between the NTF and CTF of presenilin is located the catalytic center composed by residues D257 and D385, located on TM6 and TM7 helices, which is excluded from the external surface of the enzyme. NCT is a
TE D
130 kDa type I TM protein that contains a large glycosylated ectodomain (ECD) and a single TM domain. Nicastrin is the scaffolding protein within the γ-secretase complex, and the ECD is proposed to act as substrate receptor.
EP
The other two γ-secretase components were initially identified through genetic
AC C
screening in C. elegans [7, 8]. PEN-2 directly binds PS1 and is required for its autocatalytic maturation and protease activity. APH-1 contains seven TM domains and is indispensable for γ-secretase assembly. Moreover, several studies showed that these four components are cross-regulated and downregulation or deficiency of one given component typically destabilizes the other altering the trafficking [9, 10].
3
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
Figure 1. Structure of human γ-secretase. The γ-secretase architecture and structure is
TE D
shown in surface and cartoon representation (PDB: 5A63): PS1 (orange), APH-1 (violet), PEN-2 (green) and NCT (light blue), with focus at the catalytic dyad and adjacent PAL motif. The PAL loop is highlighted in red and the close view shows its relative position compared to
EP
the catalytic residues.
AC C
During the past years, our knowledge on γ-secretase has largely increased. Residue level information on substrate docking to γ-secretase has been explored in previous biochemical experiments [11-14] suggesting that several residues scattered across the TM domains of presenilin 1 (PS1) constitute an extended binding surface for the initial substrate binding. Advancements in cryo-electron
microscopy
(cryo-EM)
have
resulted
in
high-resolution
structures of γ-secretase. A 3.4 Å cryo-EM reported 2015 (PDB: 5A63) provides a view on the atomistic organization of all four domains of the human
4
ACCEPTED MANUSCRIPT γ-secretase as well as the large ECD [3]. Despite the new insights from these studies, neither the mechanism of APP cleavage is fully understood nor a cocrystal structure of γ-secretase in complex with any of its substrates has been solved.
RI PT
Studies regarding γ-secretase nowadays aim at elucidating how APP is recognized and recruited. It has been suggested based on coarse-grained molecular simulation that the recognition and recruitment of substrate can
SC
occur on an extended surface covering PS1’s TM2/6/9 domains and PAL motif [15]. Translocation of the substrate to the active site is likely coupled
M AN U
with a large-scale conformational change. Indeed, it was suggested that a considerable conformational change in nicastrin (NCT) extracellular domain is required in order to allow the accessibility of the active site of PS. In particular, the heavily glycosylated ECD of NCT is thought to recognize the substrate N-
TE D
terminus [3] and redirect it to PS catalytic pocket for further cleavage processing. The cryo-EM structure at 3.4 Å of resolution (PDB: 5A63) does indeed provide density map for 10 out of 16 potential glycosylated sites [3];
EP
however, it is not currently known the function of these N-linked glycans and
AC C
further investigations are required. Recently, it was observed that the monomeric APP interacts with both NTF
and CTF and the flexibility of the APP influences the stability of these interactions [15]. Moreover, the non-covalent APP homodimer is protected from cleavage of γ-secretase independently from the dimerization motif [16]. Therefore, in this study we investigated the interplay between APP and γsecretase with particular attention to the specific membrane environment, namely, the effect of the lipids membrane composition.
5
ACCEPTED MANUSCRIPT Several experimental evidences associate lipid rafts to APP amyloidogenic processing. In fact, the β-secretase and γ-secretase enzymes as well as the full-length APP and the APP C-terminal fragments are localized in lipid raft domains [17-20], whereas other substrates (e.g. Notch1, Jagged2 or N-
domains
act
as
large
platforms
for
proteins
RI PT
cadherin) reside in non-raft membrane [18]. These results suggested that raft recruitment
selecting
amyloidogenic substrates. Therefore, based on lipidomics analysis of brain
SC
tissues, we modeled a realistic synaptic plasma membrane (SPM) [21], where we monitored lipids raft formation. The γ-secretase complex surrounded by
M AN U
APP proteins was embedded in this SPM model and a coarse-grained (CG) approach [22, 23] was used in order to investigate the recruitment and recognition of the APP by γ-secretase along with the raft formation with respect to the proteins’ localization. Our molecular simulations shed light on
TE D
the complex mechanism of γ-secretase APP recruitment by modeling a realistic membrane environment and by providing essential information for the
EP
further study of γ-secretase binding and inhibition modes of action.
AC C
2. Material and methods
2.1 Structural modeling. A 4.2 Å resolution cryo-EM structure (PDB ID: 5FN2) was used as the initial model for MD simulations of γ-secretase [24]. This structure was chosen, rather than the higher resolution PDB ID: 5A63 model, because it comprised of all four subunits of γ-secretase (Aph-1, Pen-2, PS1, and NCT); furthermore, γ-secretase is in complex with the nontransition-state analogue inhibitor DAPT, locking γ-secretase into a substratebinding mimic conformational state not represented in the 3.4 Å resolution
6
ACCEPTED MANUSCRIPT cryo-EM structure (PDB ID: 5A63) [3]. APP was taken from an NMR structure (PDB ID: 2LLM) that contains both the transmembrane domain and the extracellular kink [25]. Both atomic structures were coarse-grained using martinize [22].
RI PT
Γ-secretase and six randomly placed APP were embedded into a plasma membrane using insane [26]. The composition of the plasma membrane was constructed to be as similar to experimental lipidomics [21] (Table 1) as could
SC
be through the available MARTINI CG parameters [22, 23]. The dimension of
M AN U
the membrane across all replicas is 200 Å × 200 Å.
2.2 Coarse-Grained (CG) MD simulations. In the present study, all MD simulations were conducted using the GROMACS package [27]. These simulations were used to examine processes of substrate binding to γ-
TE D
secretase and to derive information about binding poses. All the systems were solvated in MARTINI CG water boxes, buffered with NaCl at 150 mM concentration.
EP
The polarized MARTINI 2.2 parameters were used for the simulations [22,
AC C
23]. All the production runs were preceded by a 5000-step energy minimization followed by a 100 ns NVT and 1 µs NPT equilibration with positions of protein beads being restrained. The temperature was set to 323 K using the V-rescale thermostat and the pressure was set to 1 bar using a semi-isotropic coupling method. A time step of 20 fs was used, a typical value employed in MARTINI simulations. Each simulation ran afterwards for 4 µs, the first 500 ns and the last 500 ns of which was used for analysis.
7
ACCEPTED MANUSCRIPT 2.3 Membrane thickness and order parameter. Two useful quantities to measure in membrane simulations are the membrane thickness and order parameters of the lipid chains. In particular, these observables are helpful in determining the presence of lipid rafts.
RI PT
Membrane thickness is here evaluated by first splitting the bilayer into its inner and outer leaflet. Each leaflet is overlaid on a grid of nodes on the xyplane where each node is equally spaced from each other. Each lipid assigns
SC
its z-position to the closest matching node. The thickness is obtained by subtracting each grid of nodes element-wise from each other. In this report,
M AN U
each node is set to have a neighborhood with the dimension of 5 Å by 5 Å. Common measurements of order in lipid bilayers are of the type that can be measured by deuterium NMR. Mathematically, lipid order parameters are defined as
1 (3 2
TE D
=
− 1)
where θ is the angle between a CD bond (in the experiment) or a CH bond (in
EP
the simulation) and the membrane normal. An order parameter of zero can describe either an unordered (isotropic) system or a perfectly ordered system
AC C
oriented at the magic angle of 54.7 with respect to the magnetic field. A value of –0.5 indicates a perfectly ordered acyl chain in all-trans conformation, rapidly rotating around the bilayer normal. Likewise, a value of 1 indicated perfect alignment with the bilayer normal.
2.4 Cross-Correlation. Dynamic cross correlation (DCC) analysis can be used to quantify the correlation coefficients of motions between atoms [28]. The DCC between the ith and jth atoms is defined by the equation:
8
ACCEPTED MANUSCRIPT ( , ) =
〈∆ ( ) ∙ ∆ ( )〉
〈‖∆ ( )‖ 〉 !〈"∆ ( )" 〉
where ri(t) denotes the vector of the ith atom’s coordinates as a function of time t, 〈∙〉 is the time ensemble average and ∆ ( ) = ( ) − ̅ where ̅ is the
RI PT
average vector of ith atom’s coordinates over the entire trajectory.
2.5 Principal Component Analysis. Principal component analysis (PCA) is a
SC
dimension-reduction procedure that reduces a large number of correlated
M AN U
variables into a smaller set of linearly uncorrelated variables called principal components (PC).
The PCs can be obtained by obtaining a set of
eigenvectors and eigenvalues from a covariance matrix that is mathematically defined as
,
1 Σ(x, y) = (() − )̅ )(* − *+) '−1
TE D
-.
The first PC has the largest eigenvalue whose physical meaning is the variance along the PC.
Each succeeding PC has the highest eigenvalue
Σ/0 = 1/0
AC C
EP
under the constraint that it is orthogonal to the preceding PCs.
9
ACCEPTED MANUSCRIPT 3. Results and Discussion 3.1 Lipid rafts form near the γ-secretase domain NCT and the active PS1. Lipid rafts are dynamic and highly ordered membrane micro-domains rich in cholesterol and sphingolipids. They are usually associated with thicker
RI PT
regions of the membrane compared to adjacent areas due to an ordering effect employed by saturated hydrocarbon chains of the lipids and cholesterol molecules [29]. These highly order micro-domains are believed to play an
SC
important role in cellular function serving as a platform for cellular processes like cell signaling, motility and protein sorting or trafficking [16, 29]. In
M AN U
particular, the combination of biochemical and magnetic immune-isolation experiments showed the coexistence of APP and γ-secretase substrates in these micro-domains [18, 30]. Moreover, raft association of γ-secretase is sensitive to acute cholesterol depletion and the absence or inhibitions of γ-
TE D
secretase activity resulted in the accumulation of APP in lipid raft domains [31]. Altogether, these results strongly suggested that γ-secretase cleavage of APP occurs in raft domains.
EP
Therefore, we first built a model of the SPM including 32 different lipids
AC C
species observed in brain tissue of AD patients [21]. This model of the SPM reflects the total lipids distribution in the presence of the AD showing an increment in sphyngolipids, ganglioside and cholesterol compared to a health brain tissue. These lipids anomalies are potentially linked with AD pathogenesis and therefore this lipidomics analysis represents a valid starting point for studying the interplay between γ-secretase and APP. Our SPM model, produced using the MARTINI force field, was thus composed of phosphatydilcoline (PC, counting for a 21% of the total lipid content),
10
ACCEPTED MANUSCRIPT phosphatidylethanolamine
(PE,
11%),
phosphatidic
acid
(PA,
0.5%),
sphyngomyelin (SM, 6%), negatively charged phosphatidylserine (PS, 2.5%) and phosphatidylinositol (PI, 2%), cholesterol (55% of the SPM) and ceramide (CE) monosialodihexosylganglioside (GM3) made up the remaining 2%. The
RI PT
lipid tails included saturated palmitoleic, polyunsaturated arachidonic and docosahexaenoic acids (Table 1).
AC C
EP
TE D
PIPC (16:0/18:2) POPC (16:0/18:1) DPPC (16:0/18:0) PAPC (16:0/20:4) PEPC (16:0/20:2) PGPC (16:0/20:1) PUPC (16:0/22:6) PRPC (16:0/24:6) PIPE (16:0/18:2) POPE (16:0/18:1) DPPE (16:0/18:0) PAPE (16:0/20:4) PQPE (16:0/20:3) PGPE (16:0/20:1) PUPE (16:0/22:6) PRPE (16:0/24:6) POPA (16:0/18:1) PGPA (16:0/20:1) PGPS (16:0/20:1) PRPS (16:0/24:6) POPI (16:0/18:1) PVPI (16:0/18:1) PAPI (16:0/20:4) PUPI (16:0/22:6) PPC (16:0/18:0) POSM (18:1/18:1) PVSM (18:1/18:1) DPSM (18:1/18:0) PGSM (18:1/22:1) PNSM (18:1/24:1) DPCE (18:1/18:0) DPG3 (18:1/18:0) CHOL
M AN U
SC
Table 1. Lipids distribution of the synaptic plasma membrane models of AD’s brain tissue. The total relative abundance of the different lipids species is given for each lipid species used. Each name corresponds to the conventional name found in the MARTINI force field.
11
SM (%) 0.36 (3) 6.93 (57) 2.55 (21) 2.92 (24) 1.58 (13) 3.53 (29) 2.07 (17) 1.95 (16) 0.12 (1) 0.49 (4) 0.12 (1) 0.49 (4) 0.12 (1) 0.97 (8) 2.43 (20) 6.20 (51) 0.24 (2) 0.24 (2) 0.73 (6) 1.82 (15) 0.24 (2) 0.24 (2) 1.46 (12) 0.24 (2) 0.36 (3) 0.36 (3) 0.36 (3) 4.26 (35) 0.12(1) 1.22 (10) 0.36 (3) 0.12 (1) 54.74 (450)
ACCEPTED MANUSCRIPT In this context, we examined the lipids raft formation with respect to the position of the γ-secretase. The top view and a cut through the planar SPM model (Fig. 2a) revealed the coexistence of the 2 phases (liquid ordered: Lo and liquid disordered: Ld) in equilibrium. The final composition of the raft-
RI PT
domain (Lo) is on average 0.51:0:20:0:28 of cholesterol/unsaturated lipids/saturated lipids, respectively, while the final composition of the Ld phase is on average 0.35:0.45:0.19. The comparison between these two membrane
SC
regions showed roughly 2 times more cholesterol in the Lo phase compared to the Ld as well as an increase in unsaturated lipids in Ld as previously
M AN U
experimentally [32] and computationally [33] observed for a simple three components lipid bilayer. The difference in lipid composition between these 2 domains is explained by diverse thickness values (Fig. 2b) and average order parameter distribution (Fig. 2c), which altogether represent the key factors for
TE D
the lipid phase evaluation. Indeed, the high thickness values are associated with high order parameters peculiar of a Lo phase, while low thickness value and low order parameters indicates an Ld region. Moreover, the difference in
EP
thickness of roughly 6 Å between Lo and Ld is in agreement with atomic force
AC C
microscopy [34] and NMR [35] data as well as molecular dynamics simulations [36]. Although the average size of lipid rafts (~9 nm in diameter) is smaller compared to the experimentally estimated one (50 nm in diameter), our MD simulations clearly showed the initial formation of a raft-like region. Interestingly, all our MD simulations showed that a greater thickness is present on the convex side of the γ-secretase structure in close proximity of the extracellular part of NCT domain and PS1 (Fig. 2b). While the exact thickness distribution varies on the whole, this local feature is present in all
12
ACCEPTED MANUSCRIPT MD replicas independently from the APP position. Indeed, liquid order parameters around these regions (Fig. 2c) mirror the state of the membrane thickness (Fig. 2b): the membrane has a greater Lo phase near the convex
TE D
M AN U
SC
RI PT
side of the γ-secretase.
Figure 2. The NCT domain of γ-secretase is localized in a raft-like domain. (A) Top-
EP
view showing the position of the γ-secretase with respect to the lipid raft composed by saturated lipids and cholesterol (red). The side-view focus (right) displayed in detail the
AC C
repartition between Ld phase composed by unsaturated lipids (green) and the Lo phase mostly composed by cholesterol (represented in van der Walls) and saturated lipids (red). (B) The probability density of the thickness, averaged on the last 500 ns, as well as (C) the average order parameter distribution, over the last 500 ns, highlighted the lipids raft formation.
3.2 APP and lipid rafts. Although APP itself is generally a raft protein, where its cleavage by the secretases is hypothesized to occur, a small portion of
13
ACCEPTED MANUSCRIPT APP has been observed to be localizing into non-lipid rafts [31]. This APP raft localization has been suggested to be regulated by cholesterol [32], one of the main components of lipid raft micro-domains. Our MD simulations confirmed these experimental data showing that APP is indeed located in 2 different
RI PT
membrane associated pools: either (i) in a raft-like domain when it is far away from the γ-secretase, or (ii) in a non-raft-like domain when it interacts with the enzyme (Fig. 3A). In fact, the averaged cholesterol distribution with respect to
SC
the proteins shows that APP resides in high-cholesterol region when not in the vicinity of γ-secretase (violet dots in Fig. 3B), while when APP is close to the
M AN U
enzyme the distribution of cholesterol around the APP is lower (orange dot in Fig. 3B).
With 6 APP present in each MD simulation allowed to freely evolve in time, we expected multiple occurrences of dimer formation; however, this was not
TE D
the case. Analysis of the trajectory shows that cholesterol has the tendency to interact preferentially with APP at the G700XXXG704XXXG708 motif (Fig. S1). Indeed the dimerization process in our MD replicas (Fig. 4) eventually occurs
EP
only in the presence of low cholesterol concentration, in line with a previous
AC C
study where we observed that cholesterol interacts with the dimerization motif G700XXXG704XXXG708 [37]. In fact, it has been experimentally observed that the production of Aβ protein is enhanced under conditions of high cholesterol concentration [38, 39]. NMR chemical shift measurements of APP in micelles of varying cholesterol concentration and MD simulations of APP homodimer [40] support the competing formation of the APP-cholesterol complex with formation of APP dimer [41].
The results indicate that cholesterol-APP
interactions have a competitive advantage over the dimer structure formation.
14
ACCEPTED MANUSCRIPT The cholesterol-APP complex could potentially promote assimilation of APP into lipids rafts where it could then interact with γ-secretase. The nature of APP being a single TM helix surrounded by a number of cholesterol moieties should make its integration into lipid rafts more likely than a bulkier
RI PT
homodimer. Furthermore, these results suggest a possible relocalization mechanism mediated by cholesterol distribution. In fact, the pool of APP found in the raft-like domains is stabilized by the interaction with cholesterol,
SC
while close to the γ-secretase the APP-membrane stability decreased in order to facilitate the further expulsion of the Aβ products upon γ-secretase
EP
TE D
M AN U
cleavage.
AC C
Figure 3. The APP localizes in two different regions of the SPM model. (A) Representative replica showed that APP either localized in a raft-like region (violet dots) or in a non-raft region (orange dots) when APP more closely interacts with the γ-secretase. (B) Average normalized cholesterol distribution showed that APP resides in high-density cholesterol when far away from the enzyme.
3.3 Interactions between APP and γ-secretase: a one-to-one relationship that involves the NCT domain. APP is a type-I membrane protein that has
15
ACCEPTED MANUSCRIPT its extracellular part mostly removed by sheddases, as α- and β-secretases [14]. Nevertheless, it is still an open question how the γ-secretase selectively recognizes the shedded N-terminal substrates and recruits them for catalysis. It was experimentally observed that NCT has a central role in the APP-PS1
RI PT
interactions. Indeed, the NCT knock-down by RNAi causes a loss of FRET signal between APP and PS1 in cells [43]. In particular, it was recently proposed that the NCT binds the ectodomains of the substrates redirecting
SC
them to the PS1, the catalytic active site of γ-secretase. This interaction was suggested to be mediated by the residue E333 in the NCT extracellular
M AN U
domain [44]. However, a recent high-resolution γ-secretase structure revealed that E333 is actually buried and on the opposite sites of the catalytic dyad formed by D257 and D385 [3, 24] (Fig. 1). Therefore, the basic mechanism of substrate recognition by γ-secretase remains controversial and requires
AC C
EP
TE D
further investigation.
Figure 4. Stable G700XXXG704XXXG708 APP dimers where observed in the SPM model. Representative snapshot of the dimeric APP conformer observed in a complex SPM model. Distance between the center of mass of the two APP monomers showed the interaction pathway over the MD simulation time.
16
ACCEPTED MANUSCRIPT CG-MD simulations revealed that G700XXXG704XXXG708 APP dimeric conformation is the stable one in this complex SPM model (Fig. 4), confirming our previous study [37]. However, we could observe in our MD simulations that when in proximity to the γ-secretase, APP is always found in the
EP
TE D
M AN U
SC
RI PT
monomeric conformation (Fig. 5).
AC C
Figure 5. The interaction between APP and γ-secretase. Representative APP-enzyme interactions, based on two different MD replicas, as a function of time. The monomeric APP interacts mainly with the NCT domain and in particular with the TM (A) as well as the extracellular (B) regions. Distances measured are the smallest pair-wise distance between NCT and APP. (C) Cross-correlation matrix and (D) the first PC from the last 500 ns of MD trajectory shown in (A) with the extracellular component of NCT omitted in the calculation. The first PC resembles an opening and closing motion of the TM domain subunits of γsecretase.
17
ACCEPTED MANUSCRIPT We noticed that the flexibility of the N-terminal of APP seems to be important for the interaction with the enzyme. In the simulations where the APP structure was retained in its NMR based L-shape conformation by harmonic potentials between interacting sites, a substrate-enzyme interaction
RI PT
mode was never detected (Fig. S2). On the other side, when this bias was removed, a preferred APP interaction was observed for the NCT domain of the γ-secretase with a distance between APP and γ-secretase around 3.5 Å
SC
(Figs. 5A-B and S2). In particular, we observed a preference for the NCT extracellular part that was neglected in previous studies [15]. On the other
M AN U
side, we never observed the interaction of APP with the catalytic residues D257 and D385 (on average in all MD simulations the distance between APP and catalytic residues is
40 Å). In fact, the active site of PS1 is hardly
accessible on the convex site of the TM horseshoe, suggesting a
recruitment.
TE D
conformational reorganization of the nicastrin domain after substrate
Conformational reorganization is further supported by the cross-
EP
correlation analysis of our CG-MD trajectories (Fig. 5C). When APP binds
AC C
near the active site on γ-secretase (Fig. 5A), APH-1 becomes anti-correlated with the subunits PEN-2 and NCT. Along with the PCA (Fig. 5D), which showed a movement parallel to the membrane, the anti-correlation suggests an opening and closing motion of γ-secretase, which could lead to the large conformational change required to expose the catalytic aspartate residues to APP for cleavage, and will be further studied when other relevant conformations of the γ-secretase will be available.
18
ACCEPTED MANUSCRIPT
In conclusion, these MD simulations performed in a complex model of the SPM revealed the importance of lipids rafts like domains in the localization
RI PT
and stability of both the substrates and enzyme. In particular, we observed that APP tends to localize in two different regions of the synaptic plasma membrane, which vary mainly by their cholesterol concentration. This in turn
SC
suggests the key role of these sterol molecules in stabilizing the substrates before recruitment by the enzyme. Moreover, our computational study reveals
M AN U
insights into the role of the NCT domain for substrate recognition endorsing the importance of this extracellular domain. Although our study is preliminary and has to deal with all the known limitations associated with the approach we chose (i.e., CG MD simulations) and further investigations are surely required,
TE D
these results provide new structural information on the early stages of substrate-enzyme association process and its specific localization at the membrane.
EP
During the last 20 years, our knowledge of the AD’s neurobiology has
AC C
increased tremendously leading to the development of several putative disease-modifying treatments. Particular attention was recently raised for the disease–modifying therapies based on the inhibition or modulation of α-, βand γ-secretases [45-48]. γ-secretase notably not only cleaves Aβ peptides but many other transmembrane proteins, for instance Notch protein. This broad range of substrates is therefore the main cause of toxicity in preclinical tests of existing γ-secretase inhibitors. Therefore, exhaustive efforts have been made to design more specific APP/γ-secretase modulators. For
19
ACCEPTED MANUSCRIPT example, it was discovered that some small molecules [49] as well as nonsteroidal anti-inflammatory drugs (NSAIDs) might modulate the APP cleavage without interfering with the cleavage of other substrates. This initial study shed light on the early stages of APP recruitment and interaction mechanism with
RI PT
γ-secretase, which would be important for further drug design investigations. Altogether, our results provide a glimpse on the proteolytic processes involved in AD pathology representing also a promising starting point for future
M AN U
4. Acknowledgment
SC
investigations on the mechanism of γ-secretase inhibitors.
M.D.P. lab is supported by the EPFL and Swiss National Science Foundation
EP
5. References
TE D
(SNF grant No. 200020_157153).
[1] X. Zhang, Y. Li, H. Xu, Y.W. Zhang, The gamma-secretase complex: from
AC C
structure to function, Frontiers in cellular neuroscience, 8 (2014) 427. [2] S. Nizari, L. Guo, B.M. Davis, E.M. Normando, J. Galvao, L.A. Turner, M. Bizrah, M. Dehabadi, K. Tian, M. Francesca Cordeiro, Non-amyloidogenic effects of alpha2 adrenergic agonists: implications for brimonidinemediated neuroprotection, Cell death & disease, 7 (2016) e2514.
[3] X.C. Bai, C. Yan, G. Yang, P. Lu, D. Ma, L. Sun, R. Zhou, S.H. Scheres, Y. Shi, An atomic structure of human gamma-secretase, Nature, 525 (2015) 212-217.
20
ACCEPTED MANUSCRIPT [4] T. Ratovitski, H.H. Slunt, G. Thinakaran, D.L. Price, S.S. Sisodia, D.R. Borchelt, Endoproteolytic processing and stabilization of wild-type and mutant presenilin, J Biol Chem, 272 (1997) 24536-24541. [5] N. Takasugi, T. Tomita, I. Hayashi, M. Tsuruoka, M. Niimura, Y. Takahashi, G. Thinakaran, T. Iwatsubo, The role of presenilin cofactors in
RI PT
the gamma-secretase complex, Nature, 422 (2003) 438-441. [6] G. Thinakaran, D.R. Borchelt, M.K. Lee, H.H. Slunt, L. Spitzer, G. Kim, T. Ratovitsky, F. Davenport, C. Nordstedt, M. Seeger, J. Hardy, A.I. Levey, S.E. Gandy, N.A. Jenkins, N.G. Copeland, D.L. Price, S.S. Sisodia,
SC
Endoproteolysis of presenilin 1 and accumulation of processed derivatives in vivo, Neuron, 17 (1996) 181-190.
[7] R. Francis, G. McGrath, J. Zhang, D.A. Ruddy, M. Sym, J. Apfeld, M.
M AN U
Nicoll, M. Maxwell, B. Hai, M.C. Ellis, A.L. Parks, W. Xu, J. Li, M. Gurney, R.L. Myers, C.S. Himes, R. Hiebsch, C. Ruble, J.S. Nye, D. Curtis, aph-1 and pen-2 are required for Notch pathway signaling, gamma-secretase cleavage
of
betaAPP,
and
presenilin
protein
accumulation,
Developmental cell, 3 (2002) 85-97.
TE D
[8] C. Goutte, M. Tsunozaki, V.A. Hale, J.R. Priess, APH-1 is a multipass membrane protein essential for the Notch signaling pathway in Caenorhabditis elegans embryos, Proceedings of the National Academy of Sciences of the United States of America, 99 (2002) 775-779.
EP
[9] B. De Strooper, Aph-1, Pen-2, and Nicastrin with Presenilin generate an active gamma-Secretase complex, Neuron, 38 (2003) 9-12. T.
Iwatsubo,
The
gamma-secretase
complex:
machinery
for
AC C
[10]
intramembrane proteolysis, Current opinion in neurobiology, 14 (2004) 379-383.
[11] Y. Imamura, N. Watanabe, N. Umezawa, T. Iwatsubo, N. Kato, T. Tomita, T. Higuchi, Inhibition of gamma-secretase activity by helical beta-peptide foldamers, J Am Chem Soc, 131 (2009) 7353-7359. [12] A.Y. Kornilova, F. Bihel, C. Das, M.S. Wolfe, The initial substrate-binding site of gamma-secretase is located on presenilin near the active site, Proc Natl Acad Sci U S A, 102 (2005) 3230-3235. [13] S. Takagi-Niidome, T. Sasaki, S. Osawa, T. Sato, K. Morishima, T. Cai, T. Iwatsubo, T. Tomita, Cooperative roles of hydrophilic loop 1 and the C21
ACCEPTED MANUSCRIPT terminus of presenilin 1 in the substrate-gating mechanism of gammasecretase, The Journal of neuroscience : the official journal of the Society for Neuroscience, 35 (2015) 2646-2656. [14] N. Watanabe, I. Image, II, S. Takagi, I. Image, II, A. Tominaga, I. Image Image, T. Tomita, I. Image, II, T. Iwatsubo, I. Image Image, Functional
RI PT
analysis of the transmembrane domains of presenilin 1: participation of transmembrane domains 2 and 6 in the formation of initial substratebinding site of gamma-secretase, J Biol Chem, 285 (2010) 19738-19746. [15] S. Li, W. Zhang, W. Han, Initial Substrate Binding of gamma-Secretase:
SC
The Role of Substrate Flexibility, ACS chemical neuroscience, (2017).
[16] E. Winkler, A. Julius, H. Steiner, D. Langosch, Homodimerization Protects the Amyloid Precursor Protein C99 Fragment from Cleavage by gamma-
M AN U
Secretase, Biochemistry, 54 (2015) 6149-6152.
[17] D.R. Riddell, G. Christie, I. Hussain, C. Dingwall, Compartmentalization of β-secretase (Asp2) into low-buoyant density, noncaveolar lipid rafts, Current Biology, 11 (2001) 1288-1293.
[18] K.S. Vetrivel, H. Cheng, S.-H. Kim, Y. Chen, N.Y. Barnes, A.T. Parent,
TE D
S.S. Sisodia, G. Thinakaran, Spatial segregation of γ-secretase and substrates in distinct membrane domains, Journal of Biological Chemistry, 280 (2005) 25892-25900.
[19] J.M. Cordy, I. Hussain, C. Dingwall, N.M. Hooper, A.J. Turner,
EP
Exclusively targeting β-secretase to lipid rafts by GPI-anchor addition upregulates β-site processing of the amyloid precursor protein, Proceedings
AC C
of the National Academy of Sciences, 100 (2003) 11735-11740. [20] R. Ehehalt, P. Keller, C. Haass, C. Thiele, K. Simons, Amyloidogenic processing of the Alzheimer β-amyloid precursor protein depends on lipid
rafts, The Journal of cell biology, 160 (2003) 113-123.
[21] R.B. Chan, T.G. Oliveira, E.P. Cortes, L.S. Honig, K.E. Duff, S.A. Small, M.R. Wenk, G. Shui, G. Di Paolo, Comparative lipidomic analysis of mouse and human brain with Alzheimer disease, The Journal of biological chemistry, 287 (2012) 2678-2688. [22] D.H. de Jong, G. Singh, W.F. Bennett, C. Arnarez, T.A. Wassenaar, L.V. Schafer, X. Periole, D.P. Tieleman, S.J. Marrink, Improved Parameters for
22
ACCEPTED MANUSCRIPT the Martini Coarse-Grained Protein Force Field, Journal of chemical theory and computation, 9 (2013) 687-697. [23] S.J. Marrink, H.J. Risselada, S. Yefimov, D.P. Tieleman, A.H. de Vries, The MARTINI force field: coarse grained model for biomolecular simulations, J Phys Chem B, 111 (2007) 7812-7824.
RI PT
[24] X.C. Bai, E. Rajendra, G. Yang, Y. Shi, S.H. Scheres, Sampling the conformational space of the catalytic subunit of human gamma-secretase, Elife, 4 (2015).
[25] K.D. Nadezhdin, O.V. Bocharova, E.V. Bocharov, A.S. Arseniev,
SC
Structural and dynamic study of the transmembrane domain of the amyloid precursor protein, Acta naturae, 3 (2011) 69-76.
[26] T.A. Wassenaar, H.I. Ingolfsson, R.A. Bockmann, D.P. Tieleman, S.J.
M AN U
Marrink, Computational Lipidomics with insane: A Versatile Tool for Generating Custom Membranes for Molecular Simulations, J Chem Theory Comput, 11 (2015) 2144-2155.
[27] M.J. Abraham, T. Murtola, R. Schulz, S. Páll, J.C. Smith, B. Hess, E. Lindahl, GROMACS: High performance molecular simulations through
(2015) 19-25.
TE D
multi-level parallelism from laptops to supercomputers, SoftwareX, 1–2
[28] J.A. McCammon, Protein dynamics, Reports on Progress in Physics, 47 (1984) 1.
EP
[29] J.B. Helms, C. Zurzolo, Lipids as targeting signals: lipid rafts and intracellular trafficking, Traffic (Copenhagen, Denmark), 5 (2004) 247-254.
AC C
[30] K.S. Vetrivel, H. Cheng, W. Lin, T. Sakurai, T. Li, N. Nukina, P.C. Wong, H. Xu, G. Thinakaran, Association of γ-secretase with lipid rafts in postGolgi and endosome membranes, Journal of Biological Chemistry, 279 (2004) 44945-44954.
[31] K.S. Vetrivel, G. Thinakaran, Membrane rafts in Alzheimer's disease beta-amyloid production, Biochimica et Biophysica Acta (BBA)-Molecular and Cell Biology of Lipids, 1801 (2010) 860-867. [32] A.J. Beel, M. Sakakura, P.J. Barrett, C.R. Sanders, Direct binding of cholesterol to the amyloid precursor protein: An important interaction in lipid-Alzheimer's disease relationships?, Biochimica et biophysica acta, 1801 (2010) 975-982. 23
ACCEPTED MANUSCRIPT [33] M. Audagnotto, M. Dal Peraro, Protein post-translational modifications: In silico prediction tools and molecular modeling, Comput Struct Biotechnol J, 15 (2017) 307-319. [34] H.A. Rinia, M.M. Snel, J.P. van der Eerden, B. de Kruijff, Visualizing detergent resistant domains in model membranes with atomic force
RI PT
microscopy, FEBS letters, 501 (2001) 92-96. [35] S.P. Soni, D.S. LoCascio, Y. Liu, J.A. Williams, R. Bittman, W. Stillwell, S.R. Wassall, Docosahexaenoic acid enhances segregation of lipids between : 2H-NMR study, Biophysical journal, 95 (2008) 203-214.
SC
[36] P.S. Niemela, S. Ollila, M.T. Hyvonen, M. Karttunen, I. Vattulainen, Assessing the nature of lipid raft membranes, PLoS computational biology, 3 (2007) e34.
M AN U
[37] M. Audagnotto, T. Lemmin, A. Barducci, M. Dal Peraro, Effect of the Synaptic Plasma Membrane on the Stability of the Amyloid Precursor Protein Homodimer, J Phys Chem Lett, 7 (2016) 3572-3578. [38] R. Ehehalt, P. Keller, C. Haass, C. Thiele, K. Simons, Amyloidogenic processing of the Alzheimer beta-amyloid precursor protein depends on
TE D
lipid rafts, J Cell Biol, 160 (2003) 113-123.
[39] S. Wahrle, P. Das, A.C. Nyborg, C. McLendon, M. Shoji, T. Kawarabayashi, L.H. Younkin, S.G. Younkin, T.E. Golde, Cholesteroldependent
gamma-secretase
activity
in
buoyant
cholesterol-rich
EP
membrane microdomains, Neurobiol Dis, 9 (2002) 11-23. [40] E.T. Parkin, A.J. Turner, N.M. Hooper, Amyloid precursor protein,
AC C
although partially detergent-insoluble in mouse cerebral cortex, behaves as an atypical lipid raft protein, Biochemical Journal, 344 (1999) 23-30.
[41] Y. Song, E.J. Hustedt, S. Brandon, C.R. Sanders, Competition between homodimerization and cholesterol binding to the C99 domain of the amyloid precursor protein, Biochemistry, 52 (2013) 5051-5064.
[42] K.S. Vetrivel, H. Cheng, W. Lin, T. Sakurai, T. Li, N. Nukina, P.C. Wong, H. Xu, G. Thinakaran, Association of gamma-secretase with lipid rafts in post-Golgi and endosome membranes, J Biol Chem, 279 (2004) 4494544954. [43] O. Berezovska, A. Lleo, L.D. Herl, M.P. Frosch, E.A. Stern, B.J. Bacskai, B.T. Hyman, Familial Alzheimer's disease presenilin 1 mutations cause 24
ACCEPTED MANUSCRIPT alterations in the conformation of presenilin and interactions with amyloid precursor protein, The Journal of neuroscience : the official journal of the Society for Neuroscience, 25 (2005) 3009-3017. [44] G. Zhao, Z. Liu, M.X. Ilagan, R. Kopan, Gamma-secretase composed of PS1/Pen2/Aph1a can cleave notch and amyloid precursor protein in the
the Society for Neuroscience, 30 (2010) 1648-1656.
RI PT
absence of nicastrin, The Journal of neuroscience : the official journal of
[45] D. Levitan, I. Greenwald, Facilitation of lin-12-mediated signalling by sel12, a Caenorhabditis elegans S182 Alzheimer's disease gene, Nature,
SC
377 (1995) 351.
[46] B. P Imbimbo, G. AM Giardina, γ-secretase inhibitors and modulators for the treatment of Alzheimer's disease: disappointments and hopes,
M AN U
Current topics in medicinal chemistry, 11 (2011) 1555-1570. [47] T. Tomita, Secretase inhibitors and modulators for Alzheimer’s disease treatment, Expert review of neurotherapeutics, 9 (2009) 661-679. [48] F. Panza, V. Solfrizzi, V. Frisardi, C. Capurso, A. D’Introno, A.M. Colacicco, G. Vendemiale, A. Capurso, B.P. Imbimbo, Disease-modifying
(2009) 537-555.
TE D
approach to the treatment of alzheimer’s disease, Drugs & aging, 26
[49] A. Petit, F. Bihel, C.A. da Costa, O. Pourquié, F. Checler, J.-L. Kraus, New protease inhibitors prevent γ-secretase-mediated production of
AC C
507-511.
EP
Aβ40/42 without affecting Notch cleavage, Nature Cell Biology, 3 (2001)
25
ACCEPTED MANUSCRIPT
Raft-like micro-domains formation influences interactions between amyloid precursor protein (APP) and γ-secretase
2.
γ-Secretase nicastrin extracellular domain is playing a key role in the early recruitment of APP
3.
A realistic model of the synaptic plasma membrane is key to describe APP/γ-secretase interplay in molecular simulation
AC C
EP
TE D
M AN U
SC
RI PT
1.