Fibronectin type III domain-containing protein 5 interacts with APP and decreases amyloid β production in Alzheimer’s disease

The deposition of Amyloid-beta peptides (Aβ) is detected at an earlier stage in Alzheimer’s disease (AD) pathology. Thus, the approach toward Aβ metabolism is considered to play a critical role in the onset and progression of AD. Mounting evidence suggests that lifestyle-related diseases are closely associated with AD, and exercise is especially linked to the prevention and the delayed progression of AD. We previously showed that exercise is more effective than diet control against Aβ pathology and cognitive deficit in AD mice fed a high-fat diet; however, the underlying molecular mechanisms remain poorly understood. On the other hand, a report suggested that exercise induced expression of fibronectin type III domain-containing protein 5 (FNDC5) in the hippocampus of mice through PGC1α pathway. Thus, in the current study, we investigated a possibility that FNDC5 interacts with amyloid precursor protein (APP) and affects Aβ metabolism. As a result, for the first time ever, we found the interaction between FNDC5 and APP, and forced expression of FNDC5 significantly decreased levels of both Aβ40 and Aβ42 secreted in the media. Taken together, our results indicate that FNDC5 significantly affects β-cleavage of APP via the interaction with APP, finally regulating Aβ levels. A deeper understanding of the mechanisms by which the interaction between APP and FNDC5 may affect Aβ production in an exercise-dependent manner would provide new preventive strategies against the development of AD. Electronic supplementary material The online version of this article (10.1186/s13041-018-0401-8) contains supplementary material, which is available to authorized users.

MD runs were carried out with time steps of 2 fs and snapshots were output every 2 ps to yield 500 snapshots per nanosecond of simulation. MD simulation of 20 ns was performed for each replica, and thus the total simulation time was 0.62 µs (=20 ns × 31).
We extracted 10,000 structures of APP 672-699 from T-REMD trajectory at the lowest temperature (T = 298K) every 2ps. After the backbone Cα atoms were structurally aligned, tertiary structures of these atoms in the 10,000 snapshots were clustered into 300 categories by using the k-means clustering method. For each clustering category, the structure that has the smallest root-mean-square deviation from the cluster center was selected as a representative one. A total of 300 representative APP 672-699 structures were further used for following irisin-APP 672-699 docking simulation.

Irisin-APP 672-699 docking simulation and binding free energy estimation
Our in vitro experiments clearly demonstrated that an N-terminal region of C99 (Asp672-Gln687) is required for binding to irisin. Also, a crystallographic analysis of irisin suggested that its flexible loop regions (Ser30-Ser32, Glu55-Val58, and Ser106-Gln108) play a significant role in recognition of other proteins [17]. Based on these experimental information, we predicted a plausible binding structure of APP 672-699 to irisin as follows. ZDOCK 3.0 program [18] was used to generate candidates of the irisin-APP 672-699 complex structure. With the standard default settings, each of the 300 APP 672-699 structures was docked into irisin and the 100 first-ranked binding poses were output, generating a total of 30,000 complex structure models. Assuming that the N-terminal region of C99 (Asp672-Gln687) and the flexible loops in irisin are involved in their binding, we selected 15,009 APP 672-699 binding-mode candidates that satisfy following two conditions: (i) the flexible loop regions in irisin are located within 5Å of APP 672-699, and (ii) more than 15% of APP atoms consisting of residues Asp672-Gln687 interacted closely with irisin (<5Å). Conformational clustering of these APP 672-699 docking-poses was then performed on its backbone Cα atom coordinates to categorize them into 2,000 representative binding-modes using the k-means clustering method. In each clustering category, the docking pose that has the smallest root-mean-square deviation from the cluster center was selected as a representative one. The binding stabilities of these irisin-APP 672-699 complex structure models were assessed by molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) [19] [20] combined with MD simulation.
MD simulation of irisin in complex with APP 672-699 was performed as follows. Each of the 2,000 irisin-APP 672-699 complex structure models was set to the initial structure. Force fields for protein, water, and ion molecules were the same as described above. Water molecules were placed around the complex model with an encompassing distance of 8 Å, including roughly 19,000 water molecules. 150mM sodium and chloride ions were added to neutralize the system. After each of the fully solvated systems was energy-minimized, it was equilibrated for 100 ps under the NVT condition, and run for 100 ps under the NPT condition, with positional restraints on irisin and APP 672-699 heavy atoms. Production runs were conducted under the NPT condition (298K and 1bar) without the positional restraints, using simulation parameters described in the previous section. A 10ns production run was performed for each of the 2,000 irisin-APP 672-699 docking structure models.
The total simulation time was 20 µs (= 10ns ×2,000 docking structures). Among the 2,000 MD trajectories, we selected 1,620 in which APP 672-699 stably binds to irisin during the 10ns simulation by judging whether the complex structure after 10ns satisfies the above-described two conditions. The MM-PBSA calculation was carried out using the MMPBSA.py module [21] in the Amber12 package [22]. In this method, the protein-ligand binding free energy (ΔG bind ) is calculated according to the following equation: where ΔE gas is the molecular mechanics energy difference in the gaseous phase, T is absolute temperature, ΔS is the conformational entropy, and ΔG solv is the solvation free energy. ΔE gas and ΔG solv were calculated by the single trajectory approach [23], in which the (free) energies for "complex", "protein", and "ligand" are computed from an MD trajectory for the protein-ligand (irisin-APP 672-699 ) complex only. In contrast, ΔS was computed using three individual MD trajectories of the complex, protein, and ligand, respectively, because small-molecular size compounds and peptides exhibit larger differences in conformational flexibility between the solvated and protein-bound states [23]. The MM-PBSA calculation was performed using a set of 450 structures extracted from a trajectory from 1 to 10 ns at regular intervals. TΔS was calculated by the quasi-harmonic approximation using the same trajectory. Ionic strength for a series of calculations was set to 150 mM. This protocol was performed for each of the 1,620 trajectories. Since several initial docking poses of APP 672-699 significantly changed during the 10 ns simulation, the mean binding-structure corresponding to the resulting ΔG was calculated by averaging the structures observed from 1 to 10 (ns), named "MD-mean binding structure". After the backbone Cα atoms in irisin were structurally aligned, a total of 1,620 MD-mean binding structures of APP 672-699 were hierarchically clustered using root-mean-square deviation of the backbone Cα atoms in the Asp672-Lys687 region, and then trees produced by the clustering were cut at a height of 10Å. The binding stability of each conformational cluster was calculated by averaging the ΔG bind values corresponding to the MD-mean binding structures within it.