Hippocampal degeneration is evident in Unc5c KI/KI mice by 18 months of age

The UNC5C T835M mutation is associated with LOAD [1], but the pathogenic mechanism of this variant in CNS neurodegeneration is unclear. To shed light on the pathophysiology of UNC5C T835M in AD, we used standard homologous recombination in ES cells to generate a targeted replacement mouse line that constitutively expresses this variant (Fig. S1A). We then analyzed the CNS phenotypes associated with UNC5C T835M expression in the targeted replacement mice, which were bred to homozygosity and aged up to 24 months. Using NeuN immunofluorescence microscopy, we found a significant reduction in hippocampal area in Unc5cKI/KI mice compared to wildtype control Unc5c+/+ at 12 and 18 months compared to 6 months (Fig. 1A, B), while Unc5cKI/KI and Unc5c+/+ cortical areas were equal at all ages, even up to 24 months of age (Fig. S1B, C). This observation may be related to higher expression levels of Unc5c in hippocampus compared to cortex [1]. We confirmed hippocampal volume loss in Unc5cKI/KI mice in vivo over time using longitudinal magnetic resonance imaging (MRI) (Fig. 1C, D). We found that ventricular volume increased significantly between 13- and 18-months in Unc5cKI/KI mice compared to Unc5c+/+ mice, (Fig. 1E, Fig.S1D) while hippocampal volume decreased significantly in Unc5cKI/KI mice, both suggesting neurodegeneration (Fig. 1F, Fig.S1E). As in the immunofluorescence microscopy-based analysis, cortical thickness was equivalent in both Unc5cKI/KI and Unc5c+/+ mice and did not change over time (Fig. 1G, H). We used diffusion tensor imaging (DTI) to measure the fractional anisotropy (FA) index, which is indicative of white matter connectivity. Between 13 and 18 months of age, Unc5cKI/KI mice had a significantly greater decrease in FA than Unc5c+/+ mice, indicating reduced gray-matter interconnectivity in Unc5cKI/KI mice (Fig. 1I, J).

Fig. 1

Unc5cKI/KI mice show significant hippocampal atrophy with age. A. Coronal sections of hemibrains immunostained for NeuN (green) in Unc5c+/+ and Unc5cKI/KI mice at 6, 12, and 18 months. Scale bar, 1.16 mm. Three different bregma positions (−1.34 mm (anterior), −1.70 mm (center), −2.06 mm (posterior)) were chosen for each animal to do the area analysis. Hippocampus is highlighted by dashed yellow region. B. Quantification of hippocampal area by ImageJ from 6–18 months. Blue circles—males; pink triangle—females. n = 4 females, 2 males (6 months), n = 5 females, 4 males (12 months), n = 6 females, 2 males (18 months). C. Schematic representation of the longitudinal MRI study. n = 8/genotype (4 males, 4 females) D. (Left) MRI slices with superimposed segmented regions of interest (Hippocampus (Blue), ventricle (red)) visualizing the changes in ventricle and hippocampus size over the course of ~ 5–6 months (13 to 18 month) for Unc5c+/+ (top) and Unc5cKI/KI (bottom) mice. (Right) Representative 3D rendered MR images are shown for both Unc5c+/+ and Unc5cKI/KI mice at 18 months. E–F. Quantification of the change over time (13 months to 18 months) in ventricular volume (E) and hippocampal volume (F). G. Representative MRI image slice showing the cortical thickness measurement. Red lines indicate various positions where the thickness was measured. Measurements (in white) in three different regions are similar in both Unc5c+/+ and Unc5cKI/KI mice at 18 months. H. Quantification of cortical thickness in Unc5c+/+ and Unc5cKI/KI mice at 13 and 18 months. I. Representative 2D MRI slices depicting the brains of Unc5c+/+ and Unc5c.KI/KI mice with superimposed fractional anisotropy (FA) patterns thresholded at ~ 0.2 (red) across the whole brain. J. Quantification of the change over time (∆) in FA/FAWT (13 months to 18 months). Statistics calculated using two-tailed unpaired student’s t-tests and ordinary one-way ANOVA using Tukey’s multiple comparison tests with Bartlett’s test correction. Data are presented as mean ± SEM. ns = non-significant. *p-value ≤ 0.05, ** p-value ≤ 0.01, *** p-value ≤ 0.001, and ****p-value of ≤ 0.0001

Synaptic protein levels and dendritic organization are significantly altered in Unc5c KI/KI mice

Since the longitudinal MRI study strongly suggested white matter atrophy in the hippocampus, we hypothesized that the Unc5cKI/KI mice had axonal and synaptic degeneration in the hippocampal region. Therefore, we assessed the levels of axonal and synaptic proteins in these mice. Immunoblot analysis of hippocampal homogenates of 18-month-old Unc5cKI/KI mice appeared to have reduced presynaptic/axonal markers, Myelin basic protein (MBP), Neurofilament/pan-axonal marker (SMI312), β-secretase (BACE1), and synaptophysin (SYP) (Fig. 2A, B-E, Fig.S2). In contrast, immunoblots for the post-synaptic marker PSD95 surprisingly showed a significant increase in the Unc5cKI/KI mice (Fig. 2A, F, Fig.S2). Post-synaptic changes were corroborated by immunofluorescence microscopy (Fig. 2G, H). Both MAP2 and PSD95 (post-synaptic markers) had increased immunostaining intensity in hippocampal CA1 (Fig. 2 G-I). Previous studies have shown that homeostatic synaptic plasticity exists to maintain the balance between pre- and post-synaptic sides of the synapse in the developing nervous system [29,30,31]. We hypothesized that if UNC5C T835M affects normal excitability or synaptic homeostasis, an abnormal decrease of the pre-synapse could cause a compensatory increase of the post-synapse. Another UNC protein, MUNC13, has also recently been implicated in controlling postsynaptic AMPA receptor density and clustering [32,33,34], suggesting a possible role for the UNC family, including UNC5 C, in the post-synaptic spine. Taken together, these data suggest that UNC5C T835M-mediated presynaptic degeneration is coupled with compensatory postsynaptic sprouting. Additionally, we observed that PSD95+ dendritic processes in the hippocampal CA1 region of the Unc5cKI/KI mice showed abnormal organization compared to Unc5c+/+, which we measured as a loss of linearity in dendrites that run parallel to each other and perpendicular to the CA1 cell layer (Fig. 2J, K). When dendritic processes are parallel to each other, the angle between them is near 0°, as observed in 3-month-old mice and the 18-month-old Unc5c+/+ mice, but in 18-month Unc5cKI/KI mice, the orientation is distributed nearly evenly, with no clear peak, indicating a loss of parallel organization. Even at 3 months in Unc5cKI/KI mice, the peak near 0° is less pronounced and there are more values in the tails of the distribution, indicating that loss of linear organization may occur early and worsen with age. Taken together, our results show that the hippocampal atrophy exhibited by Unc5cKI/KI mice is associated with reduced hippocampal presynaptic and axonal proteins, compensatory postsynaptic sprouting, and disorganized dendrites, suggestive of a neurodegenerative process.

Fig. 2

Unc5cKI/KI mice have axonal and synaptic degeneration and dendritic disorganization A. Immunoblots of presynaptic/axonal and postsynaptic proteins in the hippocampal samples of 18-month-old Unc5c+/+ and Unc5cKI/KI mice. B-F. Quantification of the immunoblots in (A) normalized to GAPDH. Blue circles—males; pink triangle – females. n = 4–5 females, 3–4 males/genotype. G. CA1 region of hippocampus stained for post-synaptic proteins such as PSD95 at 3-months and 18-months and MAP2 at 18 months in Unc5c+/+ and Unc5cKI/KI mice. H, I. Quantification of mean fluorescence intensity of PSD95 (H) and MAP2 (I). J, K Graph showing the distribution of orientation of dendrites with respect to the CA1 nuclear layer obtained at 3 months (J) and 18 months (K) in Unc5c+/+ (black) and Unc5cKI/KI (red) mice. n = 6 (Unc5c+/+), n = 7 (Unc5c.KI/KI). Statistics calculated using two-tailed unpaired student’s t-tests and ordinary one-way ANOVA using Tukey’s multiple comparison tests with Bartlett’s test correction. Data are presented as mean ± SEM. ns = non-significant. p-value ≤ 0.05, *p-value ≤ 0.01, **p-value ≤ 0.001, and ****p-value of ≤ 0.0001

Proteomic analysis reveals upregulation of oxidative stress and down-regulation of chaperone proteins in Unc5c KI/KI mice

We performed bulk proteomics on hippocampal extracts of Unc5c+/+ and Unc5cKI/KI mice at 18 months to identify in an unbiased fashion the proteins that may reveal pathways and networks with important roles in UNC5C-mediated cell death (Fig. 3A, S8). We performed Gene Ontology:Biological processes (GO:BP) enrichment analysis for significantly upregulated proteins using Database for Annotation, Visualization and Integrated Discovery (DAVID) [35]. The terms that were most significantly upregulated included oxidative phosphorylation, Alzheimer’s disease, Parkinson’s disease, Huntington’s disease (pathways leading to neurodegeneration), and endocytosis (Fig. 3B, D, S8). To corroborate our proteomic results, we quantified by immunoblot the levels of specific upregulated proteins, including Calmodulin-1 (CALM1), ubiquinol-cytochrome c reductase binding protein (UQCRB), and Capping actin protein of muscle Z-line β subunit (CAPZB) that are involved in oxidative stress pathways leading to neurodegeneration and endocytosis (underlined in red in Fig. 3D, F, S8). Notably, UQCRB and CAPZB were significantly increased in Unc5cKI/KI compared to Unc5c+/+ mice (Fig. 3G, I). Although CALM1 levels were not increased in the original Unc5cKI/KI sample (Fig. 3H), normalization against PonceauS showed that CALM1 was significantly increased in Unc5cKI/KI hippocampus (Fig.S3C). Remarkably, the down-regulated proteins by GO terms in the Unc5cKI/KI mice were chaperone, myelin sheath, and glutamatergic synapse proteins (Fig. 3C, E). We validated the proteomic results with immunoblot analysis for some of the representative proteins from each term: heat shock protein family D (HSP60) member 1 (HSPD1/HSP60) (chaperone/protein folding), Glial fibrillary acidic protein (GFAP) (myelin sheath) and Calcium voltage-gated channel auxiliary β subunit 4 (CACNB4) (glutamatergic synapses) (underlined in red in Fig. 3E, J). Notably, GFAP was significantly decreased in the Unc5cKI/KI compared to the Unc5c+/+ mice, while HSP60 and CACNB4 levels trended towards decrease in the Unc5cKI/KI mice (Fig. 3K-M, Fig.S3), which were significant upon normalization with PonceauS (Fig.S3 F, G). Previously, it has been shown that increased endocytic activity along with increased trafficking to endosomes could possibly generate Aβ that could contribute to amyloid pathology and accelerate AD [36]. Together, our findings suggest that UNC5C T835M may promote oxidative stress, which in the presence of a cytotoxic stressor, such as pathologic Aβ or tau (AD) or α-synuclein (Parkinson’s disease), may cause disease pathogenesis.

Fig. 3

Proteomics reveal upregulation of oxidative stress and down-regulation of chaperone proteins in Unc5cKI/KI mice hippocampi. A. Volcano plot of up-and down-regulated proteins obtained by TMT-MS of hippocampal homogenates of Unc5c+/+ and Unc5cKI/KI mice at 18 months (n = 5 females/genotype). Number of up-regulated (red) and down-regulated proteins (blue) are shown along with total number of proteins obtained. B, C. DAVID analysis of up-regulated proteins (B) and down-regulated proteins (C) showing the most significant GO terms. D, E. List of proteins under each significantly up-regulated (D) and down-regulated (E) biological process/GO term in DAVID analysis. F. Immunoblot analysis of hippocampal homogenates from 18-month Unc5c+/+ and Unc5cKI/KI mice for proteins (underlined in red in D) under each of the GO categories listed in D. G-I. Quantification of the immunoblots for UQCRB (G), CALM1 (H) and CAPZB (I) normalized to β-tubulin (n = 5 females/genotype). J. Immunoblot analysis of hippocampal homogenates from 18-month Unc5c+/+ and Unc5c.KI/KI mice for proteins (underlined in red in E) under each of the GO categories listed in E. K-M. Quantification of the immunoblots for GFAP (K), HSPD1 (L) and CACNB4 (M) normalized to β-tubulin (n = 5 females/genotype). Statistics calculated using two-tailed unpaired student’s t-tests. Data are presented as mean ± SEM. ns = non-significant. p-value ≤ 0.05, **p-value ≤ 0.01, ***p-value ≤ 0.001, and ***p-value of ≤ 0.0001

Unc5c KI/KI mice exhibit increased apoptosis

Since it was shown previously that UNC5C T835M increases susceptibility to death of hippocampal primary neurons [1], we employed the bioinformatic tool Polyphen-2 software (http://genetics.bwh.harvard.edu/pph2/) [37] to assess the impact of the T835M mutation on the structure and function of UNC5C protein. Polyphen-2 showed that the T835M substitution was “possibly damaging” to UNC5C protein structure with a score of 0.929 out of 1.0 (a mutation with a score of 0.0 is “tolerated” while that with 1.0 is “deleterious”; Fig. S4A), suggesting potential functional consequences that may explain the previously reported findings [1] and our own results of hippocampal atrophy suggesting neurodegeneration, as the mutation is in/near the hinge region and could affect the UNC5C open/closed state that triggers cell death or growth activities. Since cleavage of the intracellular domain of UNC5C and other UNC5 family members affect cell death via the apoptotic pathway [38,39,40], we generated an antibody against the C-terminal 400 amino acids of UNC5 C to assess how the T835M mutation affects protein levels and proteolytic processing in the hippocampus. We hypothesized that since the mutation is in the hinge region, it could cause a change in protein conformation favoring the ‘open’ state, thereby exposing the death domain that is susceptible to cleavage by activated caspase-3, triggering the apoptotic cascade. We observed that full length (FL) UNC5C levels were significantly reduced in Unc5cKI/KI hippocampal homogenates by immunoblot analysis (Fig. 4A, B, Fig.S4B). Additionally, we observed two bands at ~ 57 kD and 52 kD – denoted as “CL1 and CL2”, respectively, corresponding to the expected cleaved products of UNC5C with activated caspase-3 (boxed regions, Fig. S4C), in Unc5c+/+ and Unc5cKI/KI mice, but that were absent in Unc5c constitutive knockout mouse hippocampal homogenates (Unc5cKO/KO) (Fig. 4A, B, Fig.S4B). Both CL1 and CL2 levels were increased in Unc5cKI/KI mice, although CL1 did not reach statistical significance (Fig. 4A, B). Importantly, the individual and summed ratios of CL1 and CL2 to full-length UNC5C were elevated in Unc5cKI/KI mice, with the increase in CL2 being the main driver of the change (Fig. 4C). This observation, that signal for FL band is decreased while those of CL bands are increased, suggests a ‘precursor—product’ relationship. These results suggests that the T835M mutation might increase the cleavage of the FL UNC5 C protein into fragments that once released could in turn trigger the apoptotic cascade downstream of UNC5C when it is in its open conformation.

Fig. 4

Increased neuronal apoptosis in Unc5cKI/KI mice A. Immunoblot analysis using an UNC5C-specific antibody on 12-month-old hippocampal samples from Unc5c+/+ and Unc5cKI/KI mice. Unc5cKO/KO (Unc5c−/−) is used as a negative control. The Full-length (FL) UNC5 C band was observed around 115 kDa. Note two additional lower bands that are specific to the UNC5C antibody labeled Cleaved 1 (CL1) and Cleaved 2 (CL2) above and below the 50 k D marker, respectively. N = 10 (5 females, 5 males) B. Quantification of FL, CL1, CL2 and combined (CL1 + CL2) bands of UNC5C normalized to GAPDH and presented as arbitrary units (a.u). C. Quantification of the ratios of CL1, CL2, and combined CL1 + CL2 to FL bands. Blue circles – males; pink triangles – pink. D. Confocal images of CA1 region from 18m Unc5c+/+ and Unc5cKI/KI mice stained for TUNEL-positive neurons (NeuN, magenta; TUNEL+, red). Scale bar, 100 μm. Sections around −1.70 mm Bregma position were chosen for analysis. White boxed region in upper panels is enlarged in lower panels. Yellow arrowheads show TUNEL+ cells, of which some are NeuN+ (white arrowheads). Scale bar, 20 μm. E, F. Quantification of number of TUNEL+ neurons (TUNEL+ NeuN+) (E) and non-neuronal TUNEL+ cells (TUNEL+ NeuN−) (F) in hippocampal sections of Unc5c+/+ (black) and Unc5cKI/KI (red) mice. Blue circles—males; pink triangle—females. N = 5–7 males, n = 5–8 females/genotype/age. G. Quantification of the %NeuN covered area in the hippocampus at 18 months. n = 6 mice/genotype (2 sections/animal). H. Quantification of caspase-3 activity assay expressed as relative fluorescent units. n = 9–10 mice/genotype I. Immunoblot analysis of hippocampal homogenates from Unc5c+/+ and Unc5c.KI/KI mice for proteins involved in UNC5 C T835M-mediated apoptosis pathway at 12 months. J-N. Quantification of immunoblot signals for Protein kinase-D (PKD) (J), phospho-JNK/JNK (K), cycline-dependent kinase (CDK5) (L), NADPH oxidase (NOX1) (M), Netrin1 (NTN1) (N) normalized to GAPDH. Blue circles—males; pink triangle—females. n = 3–5 females, n = 2–5 males/genotype. Statistics calculated using two-tailed unpaired student’s t-tests, multiple t-test, two-way ANOVA using Sidak’s multiple comparisons test (for panels B and C) and ordinary one-way ANOVA using Tukey’s multiple comparison tests with Bartlett’s test correction. Data are presented as mean ± SEM. ns = non-significant. *p-value ≤ 0.05, **p-value ≤ 0.01, ***p-value ≤ 0.001, and ****p-value of ≤ 0.0001

To support this hypothesis, we measured apoptotic cell death in Unc5c+/+ and Unc5cKI/KI brains. Using a standard TUNEL (Terminal deoxynucleotidyl transferase (TdT) dUTP Nick-End Labeling) assay, we observed a significant increase in TUNEL + cells (Fig. 4D, Fig.S4D), specifically, NeuN +/TUNEL + cells (~ 60–80 cells) in the hippocampus of Unc5cKI/KI mice at 18 and 24 months (Fig. 4D, E). The number of NeuN-/TUNEL + cells, representing microglia, astrocytes, endothelial cells, oligodendrocytes, and other cells, were far fewer compared to NeuN +/TUNEL + cells (~ 10–20 cells), and remained unchanged in Unc5cKI/KI mice compared to that in Unc5c+/+ mice, indicating that only neurons are more susceptible to apoptosis in the Unc5cKI/KI mice (Fig. 4F, Fig. S4E, F). Co-staining with GFAP (astrocytes), and Iba1 (microglia) showed there were very few TUNEL + GFAP + and TUNEL + Iba1 + cells in both Unc5c+/+ and Unc5cKI/KI hippocampi (Fig. S4E, F). Additionally, NeuN-covered area was decreased by ~ 31% in the hippocampus of Unc5cKI/KI compared to Unc5c+/+ mice (Fig. 4G). The activity of caspase-3, an effector caspase in the apoptotic process, was increased in hippocampal homogenates from Unc5cKI/KI compared to Unc5c+/+ mice at 12 and 18 months of age (Fig. 4H), confirming that the UNC5C T835M mutation increases the susceptibility of neurons to cell death via apoptosis with age.

To further understand the molecular pathway of UNC5C-mediated neuron death, we performed immunoblot analysis for certain kinases known to be involved in apoptosis. Previous studies have shown that Protein Kinase-D (PKD) decreases induction of apoptosis by modulating the c-Jun N-terminal Kinase (JNK) pathway and phosphorylation of c-Jun [41]. Additionally, PKD1 has been shown to play an anti-apoptotic role in protecting neuronal cells in early stages of oxidative stress [42] by modulating JNK phosphorylation and preventing apoptosis. Since PKD1 has been shown to play a protective role in oxidative stress [43, 44], decreased PKD1 levels could lead to JNK phosphorylation and induce apoptosis via NAPDH oxidase (NOX1), as previously reported [2]. NOX1 has been associated with activated caspases in AD brains [45, 46]. Therefore, reduction in PKD levels might activate the JNK via increased phosphorylation, which then leads to elevated NOX1 levels. Consistent with increased apoptosis, we found a significant decrease in PKD in the hippocampus of Unc5cKI/KI mice (Fig. 4I, J, Fig.S4G) as well as increased phosphorylation of JNK/SAPK, which has been shown to act downstream of UNC5C T835M [2] (Fig. 4I, K, Fig.S4G). Additionally, we observed increased Cyclin-Dependent Kinase 5 (CDK5) levels in Unc5cKI/KI hippocampus (Fig. 4I, L, Fig.S4G), which have been implicated in apoptosis [47, 48]. Another study has shown that CDK5 induces c-Jun phosphorylation through activation of JNK by promoting oxidative stress [49]. Further, NADPH oxidase (NOX1) levels were increased in Unc5cKI/KI hippocampus (Fig. 4I, M, Fig.S4G), providing additional evidence of an oxidative stress environment with age in the Unc5cKI/KI mice. Taken together, our results strongly suggest that UNC5C T835M increases susceptibility to hippocampal neuron loss by creating an oxidative stress environment that leads to death via an apoptotic mechanism in vivo.

We also analyzed Netrin1 levels, which were significantly reduced at 12 months of age in the hippocampal samples from Unc5cKI/KI compared to Unc5c+/+ mice (Fig. 4I, N, Fig.S4H). Since UNC5C is a member of the “dependence” receptor family, reduced Netrin1 (ligand) could initiate the apoptotic pathway [40]. This supports the hypothesis that the T835M mutation could cause a change in protein conformation that makes UNC5C more prone to adopting the “open” conformation when Netrin1 levels decrease, thus triggering apoptosis and neurodegeneration. Cytotoxic stressors such as Aβ could exacerbate this mechanism. Furthermore, reduced Netrin1 levels are also associated with increased amyloidogenic processing of APP [50], so the UNC5C mutation could have the dual effect of both inducing apoptotic neuronal death and driving amyloid pathology through Netrin1.

Reduced GFAP levels and morphological changes are observed in astrocytes of Unc5c KI/KI mice

UNC5C is expressed in astrocytes as well as neurons [51] (https://brainrnaseq.org/?2327723709=1271088613) [52, 53]. Notably, we observed reduced GFAP levels in Unc5cKI/KI hippocampal homogenates by proteomic analysis (Fig. 3J, K). Therefore, we used immunofluorescence microscopy to assess whether the UNC5C T835M mutation affected astrocytic phenotype (Fig. 5A-C). At 12 and 18 months, we observed a significant decrease in GFAP immunofluorescence (Fig. 5D) and GFAP + coverage area (Fig. 5E) in Unc5cKI/KI mice, but no change in cell number (Fig. 5F) in the CA1 region of Unc5cKI/KI mice. This suggested that astrocytes in Unc5cKI/KI hippocampi are altered as a result of the mutation, since UNC5C is also expressed in astrocytes and our proteomics study showed that GFAP levels were down-regulated in the Unc5cKI/KI mice (Fig. 3A, E). We speculate that UNC5C signaling may modulate GFAP expression in astrocytes, and that T835M may cause the observed GFAP reduction in Unc5cKI/KI mice. GFAP comprises intermediate filaments of astrocytes and it has been shown that reduction/knockout of GFAP in astrocytes does not necessarily affect their survival [54]. Therefore, reduced GFAP levels could affect the cytoskeletal structure of astrocytes, which, in turn could affect their morphology. So, we next sought to examine the morphology of astrocytes in Unc5cKI/KI mice. We used IMARIS software to perform 3D image reconstructions to measure astrocyte process length, volume, number of branch points, number of branch terminal points, number of dendrite terminal points, and soma area (Fig. S5A-M). Although Unc5cKI/KI astrocytes exhibited more filamentous structure with increased branches, branch points and terminal points, we observed no change in the soma area of astrocytes (Fig. S5E). We speculate that astrocytes in Unc5cKI/KI mice may compensate for neuronal cell death by branching out as a way of neuroprotection. Alternatively, the astrocytic changes could be playing a role in cell death, as decreased GFAP levels could be negatively affecting astrocyte functions such as providing cytoskeletal structure to astrocytes, as well as supporting neurons and endothelial cells in the neurovascular unit [55].

Fig. 5

Astrocyte morphology is significantly altered in the Unc5cKI/KI mice. A-C. Confocal images of CA1 region of Unc5c+/+ and Unc5cKI/KI mice at 3 months (A), 12 months (B) and 18 months of age (C) immunostained for GFAP (astrocytes). Scale bar, 33 μm. Sections around −1.70 mm Bregma position were chosen for analysis. White-dashed boxed region is enlarged in A’ (3 months), B’ (12 months), and C’ (18 months). Scale bar, 50 μm. D. Quantification of mean fluorescent intensity of GFAP in the CA1 region of hippocampus at 3–6 months, 12 months and 18 months in Unc5c+/+ and Unc5c.KI/KI mice. E. Quantification of the % GFAP covered area in the hippocampus at 12 months and 18 months. Blue circles—males; pink triangle—females. n = 5–7 females, 5–7 males/genotype/age (2 sections/animal were used in the analysis). F. Quantification of number of astrocytes in the CA1 region of hippocampus at 18 months. n = 4 females, 3 males. Statistics calculated using two-tailed unpaired student’s t-tests and ordinary one-way ANOVA using Tukey’s multiple comparison tests with Bartlett’s test correction. Data are presented as mean ± SEM. Only comparisons with significant p-value are indicated. * p-value ≤ 0.05, ** p-value ≤ 0.01, ** p-value ≤ 0.001, and **** p-value of ≤ 0.0001

Microglia show increased activation in Unc5c KI/KI mice

Although microglia do not express UNC5C under normal conditions, other members of UNC5 family can be upregulated under pathological/stress conditions in cultured microglia, AD mice and AD human brain [1, 56]. In addition, neuronal apoptosis and astrocytic changes caused by the UNC5C T835M mutation could affect microglia indirectly. To assess the effects of UNC5C T835M in microglia, we performed immunofluorescence microscopy, which revealed that Unc5cKI/KI hippocampi had increased Iba1 and activated phagocytic microglial marker, CD68 compared to Unc5c+/+ mice at 18 months of age (Fig. 6A-E), demonstrating increased microglial activation. Additionally, the overall number of CD68 +/Iba1 + cells were significantly increased in Unc5cKI/KI mice, supporting increased activated phagocytic microglia in the Unc5cKI/KI mice (Fig. 6E).

Fig. 6

Microglia show increased activation in Unc5cKI/KI mice. A. Confocal microscope images of CA1 of Unc5c+/+ and Unc5cKI/KI mice at 18 months of age immunostained for Iba1 (red), CD68 (magenta) and C1q (green). Scale bar, 33 μm. B. Higher magnification of images in A showing increased activation of microglia (C1q+, CD68+) in the Unc5cKI/KI mice. Scale bar, 7 μm. C, D, F. Quantification of mean fluorescence intensity of Iba1 (C), CD68 (D), and C1q (F) in Unc5c+/+ and Unc5cKI/KI mice. E, G. Quantification of number of microglia (Iba1+) that are CD68+ (E) and both CD68+ and C1q.+ (G). Blue circles—males; pink triangle—females. n = 4–6 females, 3–5 males/genotype/age (2 sections/animal were used in the analysis). Statistics calculated using two-tailed unpaired student’s t-tests. Data are presented as mean ± SEM. Only comparisons with significant p-value are indicated. *p-value ≤ 0.05, ** p-value ≤ 0.01, *** p-value ≤ 0.001, and **** p-value of ≤ 0.0001

Previous studies have shown that increased C1q expression in microglia correlated with increased synaptic engulfment and plays a role in neurodegeneration in an Alzheimer’s disease mouse model, and increased C1q levels were observed in hippocampi of patients with multiple sclerosis [57,