LRRK2 overexpression in the hippocampus and cortex of the AD mouse model

Previous studies have indicated that the mRNA level of LRRK2 is notably elevated in the temporal cortex of AD individuals with diffusely-distributed Lewy bodies, commonly referred to as Lewy body variant (LBV), when in comparison with AD control subjects [14]. To ascertain the potential elevation of LRRK2 expression in AD mice, the LRRK2 expression was measured in the 5xFAD mice at 8 months. As the 5xFAD mouse model is recognized for its aggressive amyloid pathology, characterized by developing plaques as early as two months of age [31], the present study employed immunohistochemistry to investigate the expression of LRRK2 in various brain regions in both 5xFAD and WT mice, namely, the olfactory bulb, neocortex, striatum, hippocampus, cerebellum, and brain stem. The results revealed a significant elevation of LRRK2 expression in the olfactory bulb (P < 0.0001), neocortex (P = 0.0066), and hippocampus (P = 0.0028) of the 5xFAD mice when compared with the WT mice (Fig. 1A, B). Intriguingly, the 5xFAD mice reported a decrease in LRRK2 expression in the brain stem (P = 0.014) and cerebellum (P < 0.0001) (Fig. 1B). Furthermore, the mRNA expression of LRRK2 was upregulated in the olfactory bulb (P < 0.0001), striatum (P < 0.0001), and hippocampus (P = 0.033) of the 8-month-old 5xFAD mice (Fig. 1C). To ascertain the expression of LRRK2 in specific cell types, LRRK2 (green) was respectively double-stained with βⅢ-tubulin, GFAP, or IBA1 (Red), which revealed the presence of LRRK2 in neurons and microglia, not in astrocytes (Fig. 1D-H). Furthermore, the co-localization analysis of LRRK2 (green) and 6E10 (red, indicating Aβ) demonstrated a significant increase in the presence of LRRK2 on the microglia that surrounded plaques (Fig. 1F, G). Collectively, these findings indicate that LRRK2 is highly expressed within neurons and microglia in the critical brain areas of AD, potentially regulating the AD-associated microglial functions.

Given the potential of inhibiting LRRK2 as a therapeutic intervention for tauopathies [27, 32] and the findings of LRRK2 upregulation in AD mice presented in Fig. 1, we hypothesized that the inhibition of LRRK2 activity may ameliorate cognitive decline in the AD mouse model. To investigate the impact of LRRK2 deficiency on cognitive alterations in mice, we crossbred two types of transgenic mice, namely the 5xFAD and LRRK2-deficient mice. Subsequently, we evaluated the cognitive and motor functions of the resulting offspring (WT, 5xFAD, 5xFAD;LRRK2, and 5xFAD;LRRK2 mice).

The Morris water maze test showed that compared with the WT mice, the AD mice exhibited a significant impairment in spatial learning and memory, reporting prolonged escape latency time during the training trials and a reduced number of crossings over the platform during the probe trial (Fig. 2A-C). However, these impairments in the AD mice were significantly reversed by the deficiency of LRRK2, in which 5xFAD;LRRK2 mice displayed a significant improvement in the acquisition performance from day 2 to day 5, suggesting that these mice learned the location of the hidden platform earlier than the 5xFAD mice (P = 0.007) (Fig. 2B). During the probe trial, wherein the platform was removed, the performance of 5xFAD;LRRK2 and 5xFAD;LRRK2 mice bore a similarity to that of WT mice, but significantly surpassed that of 5xFAD mice (5xFAD vs. 5xFAD;LRRK2, P = 0.026; 5xFAD vs. 5xFAD;LRRK2, P = 0.002) (Fig. 2C). These findings suggest that even a partial knockdown of the LRRK2 gene in mice, thus a reduced LRRK2 function, may enhance spatial learning and cognitive flexibility of AD mice.

Additionally, no discernible disparities in grip tests were observed between 5xFAD and WT mice, while the absence of the LRRK2 gene greatly amplified the grip strength of 5xFAD transgenic mice (5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. 2D). Furthermore, the results of the accelerating rotarod task, which assesses alterations in motor learning, coordination, and repetitive behavior in mice, showed that the duration of rod retention was shorter in the 5xFAD transgenic mice than in both WT and 5xFAD;LRRK2 mice (WT vs. 5xFAD, P = 0.021; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. 2E). Together, these results suggest that LRRK2 deficiency ameliorates memory deficits and promotes motor coordination in AD mice.

The formation of Aβ plaque, a critical contributor to AD pathologies, is caused by the aggregation of Aβ [33]. Thus, we speculated whether genetic inhibition of LRRK2 may yield beneficial effects on alleviating Aβ plaque formation. Immunofluorescent staining with Aβ antibody revealed a notable decrease in the quantity of Aβ plaques in the hippocampal CA1 and cortex (CTX) regions of the 5xFAD;LRRK2 (CA1: P < 0.0001; CTX: P = 0.0002) and 5xFAD;LRRK2 mice (CA1: P < 0.0001; CTX: P = 0.0006), when in comparison with the 5xFAD mice (Fig. 3A and D, E). Furthermore, the 5xFAD mice exhibited a notable increase in the size of Aβ plaques, whereas the deletion of the LRRK2 gene resulted in a reduction in the size of Aβ plaques in the AD mouse model (CA1: 5xFAD vs. 5xFAD;LRRK2, P = 0.043; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.042) (Fig. 3B, C). These findings provide compelling evidence that the absence of LRRK2 significantly attenuates the aggregation of Aβ in the mouse model of AD.

Aβ is produced through a series of enzymatic cleavages facilitated by β- and γ-secretase during the processing of the amyloid precursor protein (APP) [34]. The activity of β-secretase is facilitated by the beta-site-APP cleaving enzyme (BACE), which subsequently cleaves APP to produce soluble APPβ (sAPPβ) [35]. However, α- and β-secretases engage in a competitive relationship for APP as a substrate, exerting conflicting impacts on Aβ production. Precisely, the overexpression of α-secretase elevates sAPPα levels while decreasing the generation of Aβ peptides and plaques [36]. Moreover, the activity of α-secretase is found within a group of proteins referred to as a disintegrin and metalloproteinase (ADAM), which encompasses ADAM9, ADAM10, ADAM17/TACE, and ADAM19 [37]. Thus, we assessed the impact of LRRK2 deficiency on Aβ production, aiming to determine whether LRRK2 deficiency reduces Aβ aggregation by acting on the processing of APP.

Interestingly, in comparison with the 5xFAD mice, the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice displayed no significant alteration in the protein expression levels of full-length APP (APPfl), sAPPβ, BACE1, ADAM10, and ADAM17 during the APP processing (Fig. 3F-K). Moreover, our findings indicated that BACE1 level was elevated in the 5xFAD mice when compared with the WT mice (P = 0.0004) but was not significantly altered in the LRRK2, 5xFAD;LRRK2, and 5xFAD;LRRK2 mice when compared with the 5xFAD mice (Fig. 3H). In light of these findings, we proceeded to investigate the potential impact of LRRK2 activity loss on the expression of amyloid β-degrading enzymes, namely, insulin degrading enzyme (IDE) and neprilysin (NEP), in an AD mouse model. Consistently, the expression of IDE and NEP was not significantly different across the WT, 5xFAD, 5xFAD;LRRK2, 5xFAD;LRRK2, and LRRK2 mice (Fig. 3L, M). Together, these findings demonstrate that LRRK2 deficiency alleviates Aβ burden in the AD mouse model neither by altering APP expression and processing nor by affecting the enzymic degradation of Aβ.

In the context of AD, the presence of microglia and astrocytes is commonly associated with plaque clearance. Research has demonstrated that astrocytic and microglial cells play a crucial role in modulating neuroinflammation in AD [38], while facilitating effective communication between astrocytes and microglia, improving the degradation of Aβ aggregates [39]. Therefore, we first examined the colocalization of these cells with plaques in mouse brains by immunofluorescence staining with IBA1 (Red) and GFAP. The confocal microscopy in the hippocampal CA1 and CTX regions revealed that compared with those of the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice, microglia in the 8-month-old 5xFAD transgenic mice featured enlarged cell bodies and thicker, shorter processes with reduced branching (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. 4A and Fig. S1A), which conformed with the morphological alterations associated with reactive microgliosis. Furthermore, GFAP-positive astrocytes exhibited a distinct clustering pattern within the CA1 and CTX regions of the 5xFAD mice. However, in the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice, the astrocytes were more widely dispersed, displaying a reduced clustering around plaques (Fig. S1B) and a small averaged size in both regions of interest (ROIs) (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.042, 5xFAD vs. 5xFAD;LRRK2, P = 0.044) (Fig. S1D). Additionally, when compared with the 5xFAD mice, quantitative image analysis demonstrated a significant reduction of IBA1 immunoreactivity of microglia (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) and GFAP immunoreactivity of astrocytes (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.010, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.036, 5xFAD vs. 5xFAD;LRRK2, P = 0.015) in the CA1 and CTX region of the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice (Fig. 4B, C and Fig. S1C).

As demonstrated in the data presented above, the presence of Aβ plaques may induce the activation of astrocytes and microglia, resulting in morphological changes that are associated with reactive microgliosis. It follows that LRRK2 deletion may inhibit these alterations. Thus, we assessed the quantity of microglia and astrocytes surrounding Aβ plaques in the hippocampal CA1 region of the 8-month-old 5xFAD;LRRK2 and 5xFAD mice. Cell numbers in the proximal area (located within the central region of large plaques) and distal area (encompassing the surrounding region of large plaques) were analyzed with the three-dimensional (3D) reconstructed images. Intriguingly, the 5xFAD mice reported a higher number of IBA1- and GFAP-positive cells within the central region of plaques, whereas the 5xFAD;LRRK2 mice exhibited a significant reduction in glia in close proximity to plaques (IBA1: P = 0.002; GFAP: P = 0.0096) (Fig. 4D, E and Fig. S2A, 2B). Furthermore, distal to the plaques, a decrease was observed in astrocytes (P = 0.0056), not in IBA1-positive cells (Fig. 4F and Fig. S2C). The present findings evidence a reprogrammed microglial response in the 5xFAD;LRRK2 mice, which results in a reduction in the number of microglia around the plaques and in their activation level.

The microglial phenotypic switch towards beneficial functions involves multiple molecular mechanisms, including the activation of receptors and channels that promote anti-inflammatory microglial function, and the inhibition of synthesis of nitric oxide (NO), chemokines, and inflammatory cytokines [40]. The expression of LRRK2 in immune cells is significantly elevated and subject to strict regulation in response to immune stimulation. Moreover, studies have established its functional association with crucial pathways related to immune cell function, including cytokine release, autophagy, and regulation of immune-pathways [41]. Accordingly, we investigated the potential influence of LRRK2 deficiency on the polarization of microglia into M1/M2 phenotypes in AD mice by qRT-PCR to identify markers associated with microglial phenotypes. The results revealed a reduction in the expression of M1 phenotypic markers, including inflammatory cytokines interleukin (IL)-1α (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001), tumor necrosis factor-α (TNFα) (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001), and inducible NO synthase (iNOS) (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P = 0.0009), and a decrease in M2-like polarization signals such as IL-4 (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001), transforming growth factor beta (TGFβ) (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001), and arginase 1 (Arg1) (5xFAD vs. 5xFAD;LRRK2, P < 0.0001; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001), in the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice when compared with the 5xFAD mice (Fig. 4G). The above findings suggest that the absence of LRRK2 facilitates the microglia reprogramming by regulating microglial polarization signal and alleviating the Aβ-induced microglial immune response.

Thus, we next assessed whether the LRRK2 deficiency-induced microglial reprogramming can effectively reduce the formation and expansion of Aβ plaques by enhancing the microglial phagocytosis. Numerous studies have documented the presence of cluster of differentiation 68 (CD68) in activated phagocytic microglia, which has been linked to neuroinflammation in AD [42]. The co-localization of IBA1 with lysosome-associated membrane protein CD68 demonstrated a significantly-elevated CD68 expression in the microglia of the CA1 and CTX regions of the 5xFAD mice (Fig. 4H). Quantitative analysis indicated a reduction in the number and total volume of CD68 structures within the cells in the CA1 and CTX of the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice [CA1(number): 5xFAD vs. 5xFAD;LRRK2, P = 0.0042, 5xFAD vs. 5xFAD;LRRK2, P = 0.0007; CA1(volume): 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX(number): 5xFAD vs. 5xFAD;LRRK2, P = 0.0261; CTX(volume): 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001] (Fig. 4I-L). These findings evidence that the absence of LRRK2 markedly ameliorates chronic neuroinflammation, reduces microglial CD68 expression, and enhances the clearance of Aβ aggregates, indicating that the inhibition of LRRK2 can effectively mitigate specific detrimental AD-associated microglial responses.

Previous studies indicate that Aβ is one of the main pathogenic factors for synaptic dysfunction by directly impacting neurons or via the microglia-induced excessive synaptic pruning [43, 44]. Therefore, we further investigated whether the lack of LRRK2 also compromises the stability of synaptic structures and the complement pathway in the AD mice. The co-staining of synaptic markers synaptophysin (SPH, a presynaptic marker protein) and complement C1qa revealed a notable decrease in SPH density and significant increase in C1qa expression in the hippocampal CA1 and CTX of the 5xFAD mice (Fig. 5A). However, the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice reported a largely intact SPH density (CA1: 5xFAD vs. 5xFAD;LRRK2, P = 0.0495, 5xFAD vs. 5xFAD;LRRK2, P = 0.0045; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.0071, 5xFAD vs. 5xFAD;LRRK2, P = 0.0069) and a reduced C1qa expression (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. 5B-E). Meanwhile, consistent with the findings of the western blot analysis, the levels of synaptophysin and postsynaptic density protein 95 (PSD95, a postsynaptic marker protein) were markedly reduced in the 5xFAD mice when compared with the WT mice (PSD95: P = 0.002; synaptophysin: P = 0.004), but highly elevated in LRRK2-deficient 5xFAD mice (PSD95: 5xFAD vs. 5xFAD;LRRK2, P = 0.005, 5xFAD vs. 5xFAD;LRRK2, P = 0.003; synaptophysin: 5xFAD vs. 5xFAD;LRRK2, P = 0.003, 5xFAD vs. 5xFAD;LRRK2, P = 0.016) (Fig. 5G-I). In addition, the western blot analysis and qRT-PCR demonstrated a notable decline in the expression of C1qa (at the protein level) (5xFAD vs. 5xFAD;LRRK2, P = 0.003) and of C1qa, C3, and CR3 (at the mRNA level) within the hippocampus of the 5xFAD;LRRK2 and 5xFAD;LRRK2 mice when in comparison with the 5xFAD mice (C1qa: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; C3: 5xFAD vs. 5xFAD;LRRK2, P = 0.017, 5xFAD vs. 5xFAD;LRRK2, P = 0.007; CR3: 5xFAD vs. 5xFAD;LRRK2, P = 0.001, 5xFAD vs. 5xFAD;LRRK2, P = 0.003) (Fig. 5F, J). Given that PSD95 may serve as a critical scaffolding molecule in excitatory synapses [45], these findings suggest that the deficiency of LRRK2 potentially mitigates the AD-associated disruption in synaptic structure and may reduce the activation of the complement system in the AD pathology.

The impact of LRRK2 deficiency on synaptic ultrastructure in the hippocampal CA1 was vividly visualized by transmission electron microscopy (TEM) (Fig. 5K). Our findings showed that when compared with the WT mice, the AD mice displayed a reduction in the thickness of the postsynaptic density, which was restored in the LRRK2-deficient 5xFAD mice (5xFAD vs. 5xFAD;LRRK2, P = 0.0031; 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. 5L). Notably, the width of the synaptic cleft remained unaltered across all genetic mouse models (Fig. 5M). Taken together, our findings elucidate that LRRK2 deficiency contributes significantly to the maintenance of neuronal structure in the pathogenesis of AD, in which the absence of LRRK2 attenuates complement activation within neurons and promotes the stability of synaptic structures.

Based on our discovery of increased C1qa expression within the complement system of the AD mice and the role of C1q in promoting microglial clearance of apoptotic neurons and synaptic pruning [46], we proceeded to investigate the impact of LRRK2 deficiency on microglial engulfment of synaptic structures in the AD mice. The immunofluorescence and reconstruction techniques revealed that the quantity of PSD95 within microglial phagolysosomes of the hippocampal CA1 region was significantly elevated in the 8-month-old 5xFAD mice when compared with the WT, 5xFAD;LRRK2 and 5xFAD;LRRK2 mice (Fig. 6A). Furthermore, the 5xFAD mice reported a higher microglial phagocytosis of PSD95 when compared with the other three groups (volume: WT vs. 5xFAD, P = 0.043, 5xFAD vs. 5xFAD;LRRK2, P = 0.028, 5xFAD vs. 5xFAD;LRRK2, P = 0.0064; number: WT vs. 5xFAD, P = 0.0484, 5xFAD vs. 5xFAD;LRRK2, P = 0.0288, 5xFAD vs. 5xFAD;LRRK2, P = 0.0113; average size: WT vs. 5xFAD, P = 0.0097, 5xFAD vs. 5xFAD;LRRK2, P = 0.0256, 5xFAD vs. 5xFAD;LRRK2, P = 0.0048) (Fig. 6B-D).

Many reports indicate that the early manifestation of AD involves the disruption of neuronal activity, particularly an imbalance in the underlying excitation/inhibition (E/I) process. This phenomenon is acknowledged as a crucial connection between structural brain pathology and cognitive impairment [47]. Additionally, complement C1q has been reported to contribute to the microglial elimination of excitatory and inhibitory synapses in AD mouse models [48]. Another complement activation fragment of C3 is strongly correlated with the pathology of AD, which can be ameliorated via the inhibition of complement C3 [49], suggesting that C3 may potentially mitigate neuropathology in mouse models with mutant APP. Therefore, we examined whether the deficiency of LRRK2 may impact the expression of excitatory and inhibitory synapses, and C1qa/C3 in AD mice. The immunostaining of vesicular glutamate (VGLUT1) and GABA (VGAT) transporters revealed, in the hippocampal CA1 and CTX of the AD mice, an elevation in inhibitory synapses (VGAT immunoreactive) (CA1: 5xFAD vs. 5xFAD;LRRK2, P = 0.0013, 5xFAD vs. 5xFAD;LRRK2, P = 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.0011, 5xFAD vs. 5xFAD;LRRK2, P = 0.0001), a decrease in excitatory synapses (VGLUT1 immunoreactive) (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P = 0.0423; CTX: 5xFAD vs. 5xFAD;LRRK2, P = 0.0043, 5xFAD vs. 5xFAD;LRRK2, P = 0.0477) (Fig. 6E-H and Fig. S3A-S3D), and a wide distribution of C1qa/C3 protein around the excitatory and inhibitory synapses (Fig. 6A and Fig. S3A). Additionally, other reports suggest that the presence of deposited C3 can directly activate C3 receptors on the microglia, thereby initiating synapse elimination via phagocytosis [50]. As anticipated, the expression of C3 was upregulated in the excitatory and inhibitory synapses within the hippocampal CA1 and cortex of the AD mice. Conversely, the absence of LRRK2 diminished the overexpression of C3 in AD mice (CA1: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001; CTX: 5xFAD vs. 5xFAD;LRRK2, P < 0.0001, 5xFAD vs. 5xFAD;LRRK2, P < 0.0001) (Fig. S4). Collectively, our observations suggest that the absence of LRRK2 restores the balance of excitatory and inhibitory synapses in AD pathology, potentially preserving the synaptic function in AD mice.