Experimental animals were generated by crossing hemizygous Tg2576 mice (Taconic Biosciences, Rensselaer, NY, United States) expressing APPswe with α4^−/− mice (Charles River France; Marubio et al., 1999). Tg2576 mice were crossed with α4^−/− breeders to obtain Tg2576/α4^+/− mice, that, in turn, were crossed with α4^+/− to obtain the four experimental groups: APPswe^−/−/α4^+/+ control (Ctrl) mice, APPswe^−/−/α4^−/− (α4KO) mice, APPswe^+/−/α4^+/+ (APPswe) mice, APPswe^+/−/α4^−/− (APPswe/α4KO) mice (C57Bl6J background). Male and female 5 and 15 month-old (mo) mice were used. Mice were kept in conditioned rooms with stable temperature (21°C) and humidity (60%), on a light/dark cycle of 12 h. Food and water were available ad libitum and body weight was recorded throughout the entire observation period. All animal procedures were approved by the Committee on Animal Health and Care of the University of Modena and Reggio Emilia (protocol number: 102/2011-B) and conducted in accordance with National Institutes of Health guidelines.

Mice were anesthetized with isoflurane and sacrificed by intracardiac perfusion with cold 4% paraformaldehyde (PFA) preceded by an infusion of 50 mL of 0.9% NaCl saline containing heparin sodium (5,000 U/L); the brains were postfixed in the same solution for 12 h, rinsed in 15% sucrose in PBS for approximately 12 h and then in 30% sucrose in PBS for 1 day. Fixed brains were cut at the cryostat (20 μm thickness) and processed according to established protocols to test cell-specific AD-like alterations through immunohistochemical (IHC) analysis (Daini et al., 2022).

The following antibodies (Abs) were used: rabbit anti-human APP (1:1,000, Cell signaling, #2452), mouse anti-human Aβ/APP 6E10 (epitope human Aβ1-16, 1:500, Signet, #9300-10), mouse anti-human Aβ MoAb2 (epitope recombinant human Aβ oligomers, 1:500, Millipore, #MABN254), mouse anti-human Aβ 11A1 (epitope synthetic peptide of E22P-Aβ 10–35 part, 1:500, Tecan, #10379), rabbit anti-Ionized calcium-binding adapter molecule 1 (Iba1, 1:1,000, Wako, #019-19,741), rabbit anti-glial fibrillary acidic protein (GFAP, 1:2,000, Dako, #Z0334) and rabbit anti-synaptophysin (SYN, 1:1,000, SYSY, #101004). A pre-treatment with formic acid was performed for anti-Aβ Abs when specified by the manufacturer. Vectastain ABC-HRP (#PK-4002; #PK-4001) kits were used for peroxidase diaminobenzidine staining; slices were placed on gelatinized glass slides, dehydrated and mounted with Eukitt® mounting medium. Congo Red (optical and fluorescent dye with high affinity for the β-sheet structure and thus marker for fibrillar Aβ) and Nissl stainings were performed according to established protocols (Daini et al., 2022).

Immunolabeled brain slices were photographed on a Nikon Eclipse CiL microscope (4×, 10×, or 40× objective) through a Nikon DS-Fi3 camera under constant light conditions. All evaluations were performed on coded slides by at least two experimenters.

The semi-quantitative densitometric (i.e., specific optical density) and morphometric (i.e., count, areas, thickness, length) analyses were performed following established protocols (Zoli et al., 1990, 1992; Daini et al., 2022) using routines from the Nikon NIS-Elements D (5.21.00v) and ImageJ Fiji programs.

Congo Red- or Aβ-immunoreactive plaque analysis was performed on images obtained with a 20x objective. Plaques were manually selected within the neocortex (nCtx; Bregma between −0.7 and −0.8 mm), the entorhinal cortex (Ent; Bregma between −3.1 and −3.2 mm) and the hippocampus (Hip; Bregma between −1.9 and −2.2 mm). First, images were manually edited to remove edge artifacts, folds, and blood vessels, then subjectively thresholded. To quantify global amyloid plaque extension, the % positive area covered by all amyloid plaques within the manually outlined regions of interest (Ctx, Hip) was automatically obtained through ImageJ software. Other quantitative morphometric parameters such as plaque number, area and perimeter of every plaque in the analyzed areas, were measured through a Nikon NIS-Elements D software.

In the analysis of Nissl staining, at least three somatosensory cortex (sCtx) and Hip slices/mouse (Bregma between −0.7 and −0.8 mm and between −1.9 and −2.2 mm, respectively) were acquired with 4× objective and layer thickness quantified through the length/thickness function of ImageJ software.

For SYN immunoreactivity (ir) analysis, three slices/mouse from sCtx and Hip (Bregma between −0.7 and −0.8 mm and between −1.9 and −2.2 mm, respectively) were acquired with 10× objective and the morphometric parameters quantified through the length/thickness and area functions of ImageJ software. For densitometric analysis, acquired images were converted to 32 bit grayscale and specific optical density values were obtained by subtracting the optical density of the sampled region from the optical density of non-specific staining, i.e., corpus callosum for both cortex and CA3 analysis.

For densitometric analysis of GFAP and Iba1 immunostainings, at least three slices/mouse from cingulate cortex (cCtx; Bregma between −0.5 and −0.9), sCtx (Bregma between −0.7 and −0.8 mm) and Hip (Bregma between −1.9 and −2.2 mm) were acquired with 10× objective. Acquired images were converted to 32 bit grayscale and a thresholding procedure was applied (−20 units from the peak mean gray value) and % positive area, i.e., the area covered by pixels with gray value above the threshold, was automatically recorded by means of analyze particles function of ImageJ software.

As specifically concerns the analysis of morphometric parameters related to microglia, firstly Iba1+ cell count was performed manually, in live mode, within the area of interest by using a 40× objective; in the same section, for each Iba1+ cell, cell outline was manually selected and stereological parameters, such as area, perimeter, number of primary processes and total primary process length, were automatically recorded by means of Nikon NIS-Elements D software. Finally, microglial phenotyping and characterization was performed on the basis of the number of processes and soma volume as describe in the literature with minor changes (Franco-Bocanegra et al., 2021). Briefly, cell phenotype was attributed on the basis of the number and morphology of processes: cells with 5 or more long, thin, highly branched processes were classified as homeostatic cells, cells with reduced number (3–4) of unbranched or less branched shortened processes were classified as less ramified cells and, cells without processes or 1–2 shortened and unbranched processes were classified as amoeboid cells (Torres-Platas et al., 2014; Franco-Bocanegra et al., 2021).

Values are shown as mean ± standard error of the mean (SEM) or 95% confidence interval as appropriate. Group differences were analyzed by means of two-way ANOVA or Mann–Whitney U-test as appropriate. Proportions were analyzed by means of the Chi square test. Correlations were analyzed by means of the Spearman test. Statistical analyses were performed through SPSS software, with p < 0.05 as the level for a significant difference and 0.10 > p > 0.05 as trend for a significant difference.

In order to assess whether the absence of α4* nAChRs influences Aβ plaque load, coronal brain slices from APPswe^−/−/α4^+/+ (Ctrl), APPswe^−/−/α4^−/− (α4KO), APPswe^+/−/α4^+/+ (APPswe), and APPswe^+/−/α4^−/− (APPswe/α4KO) mice were processed by means of Congo Red standard stain (for amyloidosis detection) and IHC staining with different Abs for unaggregated, oligomeric, and fibrillar forms of human Aβ42 and unaggregated Aβ40 (anti-MoAb2 Ab) or Aβ42 neurotoxic oligomers (anti-11A1 Ab).

In 5 mo mice, no Congo Red, MoAb2+ or 11A1+ plaques were detected in any mouse group including APPswe and APPswe/α4KO. Interestingly, an anti-human APP Ab showed intracellular staining in APPswe mice which was similarly intense in APPswe/α4KO mice (Mann–Whitney U-test, p = 0.70), and was absent in Ctrl and α4KO mice (Supplementary Figure 1).

In 15 mo mice, amyloid plaques were detected in both APPswe and APPswe/α4KO but never in Ctrl or α4KO mice. To characterize amyloid deposits, we performed a morphometric analysis of plaques labeled with MoAb2 and 11A1 Abs or Congo Red stain (not shown). Figure 1 provides representative images of cortical and hippocampal MoAb2+ (Figures 1A,C,E,F, respectively) and 11A1+ (Figures 1B,D,G,H, respectively) amyloid plaques in 15 mo mice (Figures 1A–H).

Our analysis demonstrated that in APPswe mice, while the regional distribution of Congo Red, MoAb2+ and 11A1+ plaques was comparable, the morphometric features of plaques were staining-specific. The number and size of MoAb2+ plaques were much larger than those of 11A1+ plaques which were, in turn, more numerous and larger than Congo Red stained plaques (Figures 1I,J). Indeed, while 11A1+ plaque number was highly significantly correlated with the number of Congo Red stained plaques in the nCtx of APPswe mice (Spearman Rho = 0.857, p = 0.007, n = 8), the number of MoAb2+ plaques was not correlated with the number either of 11A1+ or of Congo Red plaques (not shown).

A global decrease in both MoAb2+ and 11A1+ plaque number/sampled region was observed in the forebrain of APPswe/α4KO mice in comparison with APPswe mice, that was particularly marked in nCtx (Figures 1K,L). The distribution of plaque size was, however, comparable in the two mouse groups (chi square test, MoAb2, p = 0.1254, 11A1, p = 0.9419; Figures 1M,N).

These results demonstrate a decrease in the number of amyloid deposits associated with the maintenance of size distribution suggesting that the main effect of α4KO is on plaque production since the process of their growth is substantially maintained. Detailed statistical reports are shown in Supplementary Table 1.

Since amyloid-related dendritic atrophy is described in AD and neuronal populations of the primary sCtx are severely affected in the Tg2576 mouse model of AD (Somogyi et al., 2016), we analyzed the thickness of neuronal layers in Nissl-stained sCtx as a marker of tissue atrophy.

In 5 mo mice, a significant reduction was observed in cortical layers V and VI of APPswe mice, while α4KO had no significant effect (Figure 2A).

The same reduction was confirmed in 15 mo mice and restricted to layers V and VI; moreover, a synergistic effect of APPswe and α4KO was shown in layer III (Figure 2B).

As in Tg2576 mice Aβ-related neurodegenerative changes are extended also to the Hip, we analyzed the thickness of pyramidal cell layers in the different Cornu Ammonis (CA) fields.

No significant effect of APPswe or α4KO was observed in 5 mo mice (Supplementary Figure 2A). In 15 mo mice, while APPswe induced a significant decrease in pyramidal cell layer thickness in CA1 field and intermediate part of CA3 field, α4KO had no significant effect on these parameters in any hippocampal subregion analyzed (Supplementary Figure 2B).

Taken together, these results demonstrate that APPswe expression is associated with both cortical and hippocampal atrophy without any substantial effect of α4KO. Detailed statistical reports are shown in Supplementary Table 1.

Synaptic alterations represent an important feature of AD neuropathology and reductions in neuronal processes and synaptic density strongly correlate with cognitive decline in AD. Specifically, it has been reported that elevated SYN ir in cortico-hippocampal regions of aged Tg2576 mice is associated to impaired cognitive functions and possibly to pathophysiologic synaptic processing (King and Arendash, 2002). We tested whether α4* nAChR loss influences APPswe-related increase in SYN ir at cortical and hippocampal levels of aged mice.

In all experimental groups, SYN ir was particularly intense in both hippocampal and cortical regions. While in 5 mo mice, the densitometric (specific optical density) and morphometric (total area, internal area, length) analyses did not evidence any significant difference between the groups in any brain region analyzed (not shown), in 15 mo mice several regions, such as CA3 field of Hip and cCtx (King and Arendash, 2002), showed alterations in SYN ir in APPswe mice that were partially counteracted by α4* nAChR loss (Figure 3).

At cortical level, the SYN ir appeared widely distributed in layers I-VI, except for the areas occupied by neuronal somata. APPswe induced an increase in SYN ir in the external layers of sCtx (King and Arendash, 2002) that was partially counteracted by α4 loss (Figure 3). No effect either of APPswe or α4* nAChR loss was detected outside cortico-hippocampal regions, such as the caudate-putamen (CPU; not shown).

At hippocampal level, the pyramidal cell layer showed very light staining while the areas innervated by mossy fibers showed intense ir. The analysis was focused on the CA3 field (Figures 4A–D). The densitometric analysis showed a significant APPswe-related increase in the intensity of SYN ir in the stratum radiatum (SR) and stratum oriens (SO), but not in the stratum lucidum (SL), that was counteracted by the loss of α4* nAChRs (Figure 4E).

To assess whether changes in the intensity of SYN ir are accompanied by alterations in the shape and/or size of the stained areas, we focused our analysis on the SL of the CA3 field, the subregion with the highest density of SYN+ structures. We found that a number of morphometric parameters, i.e., the area of the internal more intensely stained part of the SL layer, the length of the SL and the ratio between internal and external parts of the SL, were significantly reduced in APPswe mice, alterations that were significantly counteracted by the loss of α4* nAChRs (Figures 4F–I).

Overall, present analysis shows that α4KO counteracts a number of APPswe-associated synaptic alterations, evidenced by significant changes in the intensity and/or distribution of SYN ir, in both cerebral cortex and hippocampus. Detailed statistical analysis is reported in Supplementary Table 1.

To evaluate how APPswe age-dependently influences astrogliosis, and whether α4KO modulates this process, brain sections from Ctrl, α4KO, APPswe and APPswe/α4KO mice were labeled with anti-GFAP Ab for IHC analysis. Several brain regions were analyzed, i.e., cingulate cortex (Cg), corpus callosum (CC) and hippocampal regions, at both 5 and 15 mo of age (Figure 5).

In 5 mo mice, when there was no evidence for Aβ plaque deposition in any brain region, α4KO significantly counteracted APPswe-related increase in GFAP ir in the hilus of dentate gyrus (DG) and stratum oriens (SO) of the CA1 field while a trend for a significant difference was observed in the stratum lacunosum-moleculare of the CA1 field (Figure 5I).

The same analysis was carried out in 15 mo mice, when Aβ plaque deposition was present in both cortical and hippocampal regions. At this age, APPswe expression was associated with a significant increase in GFAP ir in the hilus of DG, CA3 and the cingulate cortex/medial corpus callosum region (CC/Cg), that was counteracted by α4KO in CA3 field (Figure 5J).

Altogether, these results show that the loss of α4* nAChRs reduces the signs of astrocyte activation that appear already in 5 mo Tg2576 mice, before Aβ plaque deposition, and this effect is maintained at later stages when amyloid neuropathology is extensive.

Similar to astrocytes, morphological changes in microglial cells, such as increased cell soma size and process retraction and thickening, have been observed in both AD and mouse transgenic models of AD (Leyh et al., 2021). The assessment of microglial number and morphological alterations was performed in both 5 and 15 mo mice to reveal whether the loss of α4* nAChRs influences distribution and morphology of microglia in APPswe mice.

The stereological analysis showed that both APPswe and α4KO induced a number of changes already in 5 mo mice in sCtx. In fact, while the number of Iba1+ cells was not different between the groups (Supplementary Figures 3A–F), the prevalence of the three principal microglial subclasses, i.e., homeostatic, less ramified and amoeboid cells, was significantly altered by both APPswe and α4KO in layer I-II and by APPswe in layer V (Supplementary Figure 4; Figures 6A–F).

Further analysis of the microglial subpopulations in layers I–II showed that APPswe induced an increase in the number of amoeboid and a decrease in the number of homeostatic cells as well as an increase in the perimeter and a decrease in the number of processes of activated cells, whereas α4KO induced a decrease in the number of processes of Iba+ cells (Supplementary Figure 4). In layer V, APPswe increased the area of less ramified cells and the perimeter of homeostatic cells, whereas α4KO increased both area and perimeter of less ramified cells (Supplementary Figure 5). Overall, both APPswe and, to a lower extent, α4KO were associated to a reduced number of homeostatic microglial population.

Several changes were also observed in 15 mo mice. We first performed an analysis of the number of Iba1+ cells in several brain regions. APPswe expression significantly increased the number of Iba1+ cells in sCtx (Figures 7A–D), DG (Figures 7E–H), and the CA3 field (not shown) but not in the CA1 field (not shown) and CPU (Figures 7I–L). α4KO did not induce any significant change in any brain region examined (Figure 7M).

Then we performed a stereological analysis of Iba1+ cells in the layers I–II and V of the sCtx by using quantitative parameters like soma area and perimeter, process length and number of processes. In layers I–II, APPswe expression did not significantly alter total cell number (Figure 8A), significantly decreased the number of processes/cell without changing process length, and increased the size of Iba1+ microglial cells, while α4KO did not induce any significant change in Iba1+ cells (Supplementary Figure 6).

In layer V, APPswe expression significantly increased the number and the size of Iba1+ microglial cells (Figure 8A), and decreased their number of processes/cell without changing process length, while α4KO did not induce any significant change in Iba1+ cells (Supplementary Figure 7).

The morphological analysis of homeostatic, less ramified and amoeboid cells demonstrated a shift toward less ramified and amoeboid Iba1+ cells in APPswe mice with respect to Ctrl or α4KO mice, that was partially reverted in APPswe/α4KO mice in layer V but not in layers I–II (Figures 8B,C).

The analysis of the morphometric features of the three classes of microglia showed similar changes in layers I–II and layer V, though more intense in layer V. APPswe expression was associated with a significant increase in the number and size of amoeboid and reactive microglia, this latter showing also a decreased number of processes, and a significant decrease in cell and process number and increase in size of resting microglia in layer I–II and, more intensely, in layer V. These alterations in microglia populations were not significantly modified by α4KO (Supplementary Figure 4).

In conclusion, while α4KO was not able to significantly counteract the morphological changes induced by APPswe expression in microglia, it could partially counteract the shift toward more reactive forms in layer V, in parallel with a reduction in plaque load (see above). Detailed statistical analysis is reported in Supplementary Table 1.