cGMP favors the interaction between APP and BACE1 by inhibiting Rab5 GTPase activity

We previously demonstrated that cyclic guanosine monophosphate (cGMP) stimulates amyloid precursor protein (APP) and beta-secretase (BACE1) approximation in neuronal endo-lysosomal compartments, thus boosting the production of amyloid-β (Aβ) peptides and enhancing synaptic plasticity and memory. Here, we further investigated the mechanism by which cGMP regulates the subcellular localization of APP and BACE1, finding that the cyclic nucleotide inhibits the activity of Rab5, a small GTPase associated with the plasma membrane and early endosomes. Accordingly, we also found that expression of a dominant-negative Rab5 mutant increases both APP-BACE1 approximation and Aβ extracellular levels, therefore mimicking the effects induced by cGMP. These results reveal a functional correlation between the cGMP/Aβ pathway and the activity of Rab5 that may contribute to the understanding of Alzheimer’s disease pathophysiology.

Aβ evaluation. ELISA kits (High Sensitive Human/Rat β Amyloid (42) and Human/Rat β Amyloid (40), Wako Chemicals GmbH, Germany) were used to measure Aβ peptides released into culture media from N2a cells. At the end of treatments, conditioned media were collected, spun at 1000 g for 5 min to remove cell debris, and stored at −80 °C until use. Were indicated, intracellular Aβ 42 was measured in total cell extracts. ELISA tests were carried out following the manufacturer's protocol, and the concentration of Aβ was calculated according to the standard curves prepared on the same ELISA plates.
Rab5 activity. N2a cells were grown overnight on 10 cm culture dishes and then incubated with 100 μM vardenafil (or with the same volume of DMSO) for the indicated times. At the end of treatments, Rab5 activity was analyzed with the Rab5 Activation Assay Kit (NewEast Biosciences, USA), according to the manufacturer's protocol. Briefly, an anti-active Rab5 mouse monoclonal antibody was used to immunoprecipitate the GTP-bound form of Rab5 in the cell extract. Immunoprecipitated Rab5 was then detected by immunoblot analysis using a rabbit anti-Rab5 polyclonal antibody. Immunoblot analysis. Total protein extraction from cell cultures and immunoblots were performed according to standard methods, as described previously 19 . Anti-rabbit and anti-mouse secondary antibodies were coupled to horseradish peroxidase (GE Healthcare, UK). Protein were visualized with an enzyme-linked chemiluminescence detection kit according to the manufacturer's instructions (Amersham, UK). Chemiluminescence was monitored by exposure to films, and signals were analyzed under non-saturating condition with an image densitometer (Bio-Rad, USA).

Confocal analysis.
N2a cells were grown overnight on culture slides and transiently transfected with mCherry-Rab5 mutants where indicated. After 16 hours, cells were transfected with APP:VN and BACE1:VC expressing vectors and, 6 hours later, incubated with DMSO or vardenafil (50 μM) for 16 hours. At the end of treatments, cells were permeabilized and fixed with ice-cold methanol, incubated with TO-PRO-3 Iodide (Thermo Fisher Scientific, Italy) for nuclear staining, and observed with the appropriated filters on a Leica TCS SP2 confocal microscope (planapochromat x 60 oil-immersion objective, numerical aperture 1.4).

Statistical analysis.
Results are expressed as mean ± standard error of the mean (SEM). The number of independent experiments is reported in each figure legends. Data were analyzed using one-away ANOVA followed by Dunnet's post-hoc. The level of significance was set at P < 0.05.

Results
Inhibition of endocytosis prevents the amyloidogenic effect of cGMP. Since our recent studies showed that cGMP induces a faster internalization of APP and triggers APP and BACE1 to colocalize in endosomal compartments 11,12 , we first investigated the impact of endocytosis on the cGMP-induced amyloidogenesis. To this aim, N2a cells were pre-treated with PitStop2, an inhibitor of both clathrin-dependent and clathrin-independent endocytic pathways 20 , and then exposed to the cGMP-enhancer vardenafil. As expected, the inhibition of endocytosis per se reduced the concentration of Aβ 42 peptides in conditioned media 4 (67% of control, P < 0.05), whereas vardenafil robustly increased it 11 (286% of control, P < 0.0001). Notably, the effect of vardenafil was totally prevented by PitStop2, indicating that the endocytic process is required for cGMP to increase extracellular Aβ 42 (Fig. 1a). In addition, also intracellular levels of Aβ 42 were significantly increased under vardenafil treatment, though to a lesser extent compared with the secreted peptide (130% of control, P < 0.01), providing further support to the evidence that cGMP plays a role in the processing of APP (Fig. 1b).
Rab5 activation state modulates Aβ42 levels. Given the involvement of Rab5 14 and Arf6 8 in APP and BACE1 endosomal trafficking, respectively, we evaluated the amount of Aβ 42 peptides in conditioned media of cells transiently transfected with mCherry-Rab5 WT , HA-Arf6 WT , or their constitutively active (mCherry-Rab5 CA , HA-Arf6 CA ) and dominant-negative (mCherry-Rab5 DN , HA-Arf6 DN ) mutants. Efficiency of transfections in multiple experiments was assessed by immunoblot analyses, and typical levels of expression are shown in Fig. 1c (left panels). Overexpression of either WT or mutant forms of HA-Arf6 did not alter the amount of Aβ 42 released by (2020) 10:1358 | https://doi.org/10.1038/s41598-020-58476-8 www.nature.com/scientificreports www.nature.com/scientificreports/ the cells in the culture media, and similar results were obtained in mCherry-Rab5 WT expressing samples (see the histogram in Fig. 1c). On the contrary, overexpression of mCherry-Rab5 DN increased the Aβ 42 release (130% of control, P < 0.05), while a slight, but statistically significant decrease was observed in culture media of cells transfected with mCherry-Rab5 CA (83% of control, P < 0.01) (see the histogram in Fig. 1c).

Rab5 siRNA increases both Aβ 42 and Aβ 40 peptides. In order to confirm the correlation between Rab5
and Aβ production, we induced Rab5 transient knockdown in N2a cells using specific siRNAs. Forty-eight hours after siRNA transfections, conditioned media were subjected to Aβ specific ELISA, whereas cell extracts were analyzed for Rab5 expression by immunoblot. In line with the effect exerted by Rab5 DN , silencing of Rab5 significantly increased the amount of Aβ 42 released by the cells in the culture medium (131% of non-targeting control siRNA, P < 0.05) (Fig. 1d). Of note, also the Aβ 40 species increased in the medium of Rab5 knockdown cells (136% of non-targeting control siRNA, P < 0.05), indicating that the amyloidogenic effect of Rab5 siRNA does not change the Aβ 40 /Aβ 42 ratio (Fig. 1d).
cGMP reduces GTP-Rab5 levels. We next sought to examine whether an increase of intracellular cGMP could influence the activation of Rab5. As a small GTPase, Rab5 cycles between a GDP-(inactive) and a GTP-bound form (active) 21 . Therefore, we measured GTP-Rab5 levels in cells exposed to vardenafil for different times (5, 10, 30, and 60 min). As shown in Fig. 2, an approximately 50% reduction of active GTP-bound Rab5 was already evident, although not statistically significant, after 5 min of vardenafil treatment (P = 0.088 vs control). The amount of GTP-Rab5 dropped to 40% of control after 10 min (P < 0.01) and remained depressed after 30 and 60 min of treatment (45% of control, P < 0.05, at both time points). Because the total expression of Rab5 did not change at any time of vardenafil exposure (Fig. 2, Total Rab5 panel), it is likely to assume that cGMP increased the inactive GDP-bound form of the small GTPase.

Expression of Rab5 Dn mutant increases APP-BACE1 approximation. Using the Optical
Convergence of APP and BACE1 (OptiCAB) assay 18 , we have previously demonstrated that cGMP stimulates the interaction between APP and BACE1 in endosomal compartments 12 . Here, given the effect of vardenafil on the activation state of Rab5, we took advantage of the OptiCAB assay to investigate whether the expression of Rab5 mutants could affect the APP-BACE1 approximation. To this aim, 16 hours after mCherry-Rab5 DN or mCherry Rab CA transfections, N2a cells were further transfected with APP tagged with the N-terminal fragment of the Venus fluorescent protein (APP:VN), and with BACE1 tagged with the complementary C-terminus (BACE1:VC). In this manner, the physical approximation of APP:VN and BACE1:VC allows the reconstitution of Venus protein, which becomes fluorescent. Indeed, confocal microscopy analysis showed that the expression of Rab5 DN is able to increase APP-BACE1 approximation (Fig. 3), which is suggestive of an increased interaction between BACE1 and its substrate. Notably, this effect exactly resembled that induced by the cGMP-enhancer vardenafil and was not mimicked by the expression of Rab5 CA (Fig. 3).

Discussion
In this study, we provide evidence that cGMP influences the activation state of Rab5 by favoring its inactive GDP-bound conformation. This conclusion is supported by a coherent set of data. Firstly, the amount of Rab5 bound to GTP is quickly reduced by vardenafil, a phosphodiesterase 5 (PDE5) inhibitor whose ability to enhance cGMP in N2a cells is widely documented 12 . Secondly, expression of a dominant mutant that keeps Rab5 in its GDP-bound form increases the amount of Aβ peptides released by the cells, in line with previous studies showing that cGMP stimulates Aβ production in a dose-dependent manner 11,12 . Thirdly, extracellular Aβ levels are increased when Rab5 expression is knocked down by siRNA. Finally, expression of the inactive Rab5 mutant www.nature.com/scientificreports www.nature.com/scientificreports/ increases APP-BACE1 interaction, an effect previously observed in primary neurons treated with vardenafil 12 , and further confirmed in the present study.
Taken together, these results suggest that the well-established positive effects exerted by cGMP on synaptic plasticity and memory formation (for review, see 22 ) may occur via lowering Rab5-GTP levels, thus explaining, at least in part, why the upregulation of the small GTPase is associated with neurodegenerative phenomena 13 .
Another important evidence in the present study is that the inactivation of Rab5 correlates with an increased production of Aβ peptides, an event that, based on the "amyloid hypothesis" of AD 23 , would generate detrimental consequences. Indeed, our data fit well with the notion that released Aβ boosts hippocampal activity by regulating synaptic trsnsmission 24 , and with the studies of Puzzo and collaborators, who clearly demonstrated that physiological (picomolar) concentrations of extracellular Aβ 42 are required to sustain LTP and cognitive performance 25,26 .
On the other hand, our findings seem at variance with previous reports showing that overexpression of either wild-type Rab5 27 or dominant negative Arf6-T27N 8 increases the production of Aβ peptides. It should be noted, however, that those studies have been performed in non-neuronal cells engineered to overproduce Aβ, thus under different conditions that might justify the different results.
Consistent with this hypothesis, a new transgenic mouse model of neuronal Rab5 over-activation was very recently shown to develop AD-related endosome dysfunction and AD-like deficits in axonal transport, synaptic plasticity, cognition, and neuron survival, apparently without increasing cerebral Aβ. Noteworthy, crossing this Rab5 model with APP-models of AD accelerated Aβ accumulation and cholinergic neurodegeneration, suggesting that APP mutation/elevation may change the effect of Rab5 activation on Aβ production (Nixon, R. A.  www.nature.com/scientificreports www.nature.com/scientificreports/ Collectively, our data support a scenario in which cGMP favors the endosomal interaction between APP and BACE1 by keeping Rab5 in its GDP-bound conformation, consequently leading to the production of Aβ peptides that, in turn, sustain LTP and memory formation/consolidation 12,26 .
Although we are still far away from elucidating the complete picture, novel molecular players governing the dynamics of Aβ production are beginning to emerge. Certainly, the more we learn on the physiology of APP and its processing, the closer we get to understanding AD and finding effective pharmacological treatments.
From this prospective, increasing cGMP levels with PDE5 inhibitors could represent a promising therapeutic strategy. As a matter of fact, a study investigating repeated administrations of udenafil in patients suffering from erectile dysfunction has shown beneficial effects on memory performance as well as on a battery of tests assessing executive functions 28 . For a more complete information, however, it must be said that, in healthy subjects, single vardenafil administration did not produce cognitive or information processing modifications 29 .

Data availability
The authors make materials, data and protocols associated to this article promptly available to readers without undue qualifications in material transfer agreements.