Midbrain degeneration triggers astrocyte reactivity and tau pathology in experimental Alzheimer’s Disease

Experiments complied with the ARRIVE procedures and the ethical guidelines of the European Council Directive (2010/63/EU). Experimental approval was obtained from the Italian Ministry of Health. Male and female C57BL/6N (Charles River, Italy) and female B6.129P2(Cg)-Cx3cr1^tm1Litt/J (CX₃CR-1^GFP knock-in/knock-out; Jackson Laboratory Strain #:005582) mice were used at 2–3 months of age. Heterozygous male Tg2576 mice (Taconic #APPSWE—Model 1349 tg/wt; [41]) and their Wild-Type (WT) littermates (mice with the same strain and genetic background of Tg2576, negative for APPSWE overexpression—Taconic #APPSWE—Model 1349 wt/wt) were used at 6–7 months of age.

All mice were housed with adlibitum food and water, with a 12 h light/dark cycle. Mice were housed in cages of 3–4 individuals and all cages were equipped with the same environmental enrichment.

Mice were anaesthetized with Rompun (20 mg/mL, 0.5 mL/kg; Bayer) and Zoletil (100 mg/mL, 0.5 mL/kg; Virbac; intraperitoneally i.p.) and positioned in a stereotaxic apparatus.

For Caspase-3 (Casp3) mice (C57BL/6N, CX₃CR-1^GFP or Tg2576 mice), we infused a mix of Adeno-Associated Viruses (AAVs; 0.5 µL each, flux 80 nL/min [42]: (i) AAV1-THp-iCre (5 × 10¹² viral particles/mL; Vector Biolabs) and (ii) AAV5-flex-taCasp3-TEVp (4.2 × 10¹² viral particles/mL; UNC Vector core, gift from Nirao Shah) in the left VTA (AP: −3.2, ML: −0.35, DV: −4.4; 43), to lesion Tyrosine Hydroxylase-positive (TH⁺) neurons in the VTA and Substantia Nigra pars compacta (SNpc) and 5-HT⁺ neurons in the interpeduncular nucleus (IPN). Sham mice were injected only with the AAV1-THp-iCre virus.

In a separate cohort of C57BL/6N mice, the AAV1-THp-iCre + AAV5-flex-taCasp3-TEVp mix (0.125 µL each, flux 25 nL/min) were injected in the IPN (thereafter: Casp3^IPN; AP: −3.5, ML: 0, DV: −4.7; [43]) to lesion the 5-HT⁺ neurons of the IPN while leaving intact the TH⁺ neurons of the VTA/SNpc. Sham mice were injected only with the AAV1-THp-iCre virus (Sham^IPN).

To confirm the ectopic expression of TH promoter-driven AAVs, a set of C57BL/6N mice were infused in the VTA with a mix (0.5 µL each, flux 80 nL/min) of (i) AAV1-THp-iCre and (ii) AAV5-EF1α-DIO-eYFP (4.0 × 10¹² viral particles/mL; UNC Vector core, gift from Karl Deisseroth). eYFP expression was then analyzed with confocal microscopy (see below).

For 6-hydroxy-dopamine (6OHDA) lesion (C57BL/6N mice), 6OHDA (Sigma-Aldrich; 7.6 mg/mL, calculated as free-base) was dissolved in 0.2 mg/mL ascorbic-acid (Tocris) prepared in 0.9% saline, and continuously kept on ice. Each mouse was injected with 2.5 μg in 0.4 µL (flux 40 nL/min) unilaterally in the left VTA. Thirty minutes before surgery, mice received 10 mL/kg (2.85 mg/mL as free base, i.p.) of the norepinephrine (NE) reuptake-inhibitor desipramine-hydrochloride (Sigma-Aldrich), to prevent NE fiber degeneration. Control mice were injected with Saline.

C57BL/6N and CX₃CR-1^GFP mice received the intracerebral injection at 2 months of age. Tg2576 mice were injected at 6 months of age. All mice were used 30 days following injection to ensure steady-state lesion. A separate set of Casp3 and Sham were used at 72 h following lesion; in these mice we examined the levels of Casp3 protein in the midbrain (see below).

For retrograde labelling, red retrobeads (120 nL; 12 nL/min; Lumafluor) were injected bilaterally in the Nucleus Accumbens (NAc) medial shell (AP: 1.8, ML: ± 0.4, DV: 4.1), NAc core (AP: 1.5, ML: ± 0.9, DV: 4.1) and dorsal striatum (AP: 1.0, ML: ± 1.6, DV: 2.3). Mice were examined 1 month after surgery.

For all infusions we used 1 μL Hamilton syringes (Model Neuros7001) mounted on a Pump-11 Elite Nanomite (Harvard Apparatus). When possible, the accuracy of the injection site was controlled by immunofluorescence analysis; mice with misplaced injection were excluded.

Animals were injected with L-DOPA (i.p., 10 mg/kg; Sigma-Aldrich) plus benserazide (12 mg/kg; Sigma-Aldrich) [44], or with A68930-hydrochloride (5 mg/kg; Santa Cruz), once a day for 4 or 7 days, respectively; 0.9% saline was used as control (vehicle); experiments were performed 1 h after the last injection.

Fluoxetine-hydrochloride (30 mg/kg; Tocris) was dissolved in water and delivered adlibitum in drinking bottles wrapped in tin-foil for 30 days, starting immediately after surgery. Normal drinking water was used as control.

Anaesthetized mice (Rompun/Zoletil) were transcardially-perfused with Phosphate Buffer (PB; 0.1 M, pH 7.4) followed by 4% paraformaldehyde in PB. Brains were postfixed in 4% paraformaldehyde for at least 4 h, dehydrated and cryoprotected in 30% sucrose in PB at 4 °C until sinking. Thirty μm-thick coronal sections were cut with a cryostat, and slices were collected in PB-sodium azide 0.02%. All analyses were performed in the left hemisphere, ipsilateral to the lesion.

Slices were incubated with primary antibodies in PB containing 0.3% Triton X-100 overnight at 4 °C for TH, Iba1, GFAP, hippocampal 5-HT transporter (SERT) and striatal TH/DA transporter (DAT)/SERT staining, or 3 nights for IL-1β/Iba1, IL-18/Iba1, IL-18 Receptor (IL-18R)/GFAP, p-Nuclear Factor kappa B (p-NFκB)/S100β and Microtubule-Associated Protein 2 (MAP2).

For 5-HT, slices were incubated with primary antibodies in PB containing 0.5% Triton X-100 and 10% donkey-serum for 2 nights at 4 °C.

For hippocampal TH⁺/NE transporter-positive (NET⁺) fibers, sections were incubated in citrate buffer (10 mM sodium-citrate, pH 6.0 containing 0.05% Tween-20; 20 min, 75 °C), rinsed in PB, immersed in blocking solution (5% donkey serum, 0.2% Triton X-100 in PB; 1 h, RT), and incubated with primary antibody in the same solution (overnight at 4 °C; [45]).

For NLRP3/Iba1/GFAP and NLRP3/Iba1 staining, sections were pretreated with 50% methanol (15 min, RT) immersed in blocking solution (3% bovine serum albumin, 0.1% Triton X-100 in PB, 30 min, RT) and incubated with primary antibodies in PB with 0.1% Triton X-100 (2 nights, 4 °C).

For Iba1/CD68, slices were permeabilized using 0.5% Triton X-100 in PB (45 min, RT), incubated in blocking solution (2% bovine-serum albumin, 0.5% Triton X-100 in PB; 1 h, RT) and exposed to primary antibodies in blocking solution (2 nights, 4 °C).

For C3/GFAP, slices were immersed in blocking solution (5% donkey serum, 0.1% Triton X-100 in PB; 1 h, RT) and then incubated with primary antibodies in 1% donkey serum, 0.1% Triton X-100 in PB (overnight, 4 °C).

For C3aR/Iba1/NeuroTrace analysis, slices were incubated in blocking solution (3% bovine-serum albumin, 5% goat serum, 0.5% Triton X-100 in PB; 1 h, RT) and then with primary antibodies in blocking solution (overnight, 4 °C).

For tau staining (AT8), slices were pretreated with 50% methanol (15 min, RT), immersed in blocking solution (3% bovine serum albumin, 0.1% Triton X-100 in PB) with M.O.M.® (Mouse on Mouse) Blocking Reagent (1:1000; Vector laboratories, #MKB-2213–1; 1 h, RT) and then incubated with primary antibodies in 0.1% Triton X-100 in PB (2 nights, 4 °C).

For Aβ staining (6E10), sections were pretreated with M.O.M.® (1:1000; 2 h, RT) diluted in permeabilization solution (PB with 0.3% Triton X-100) and incubated with primary antibodies overnight at 4 °C in permeabilization solution.

For eYFP expression, coronal sections containing the midbrain, Raphe and LC were stained with TH/5-HT, 5-HT and TH, respectively as described above.

For every immunofluorescence protocol, after primary antibody, slices were incubated with secondary antibodies in the same solution of primary antibody (2 h, RT) and counterstained with Blue-Fluorescent Nissl-Stain (NeuroTrace 1:200; Invitrogen) or DAPI (1:1000, Serva).

After mounted, slices were examined using a Nikon Eclipse Ti2 confocal microscope. The labeling specificity was confirmed by omission of primary antibodies and use of normal serum instead (negative controls). For quantitative analysis, images were processed simultaneously and analyzed with Fiji-ImageJ (http://imagej.nih.gov/ij/↗): after 8-bit conversion and background subtraction, the signal was quantified by measuring the relative fluorescence intensity. The F/A ratio defines mean fluorescence intensity (F) over surface area (A).

For analysis of fiber density and intensity, and protein levels, images were acquired with a 20x-objective by Z-stacks, then processed by maximum-intensity projection. All samples were captured with identical Z-stack thickness and laser settings within each analysis.

TH⁺/DAT⁺/SERT⁺ fiber intensity was quantified by setting 16 frames (100 × 100 pixel). The total fiber number per 250 µm was counted manually [45, 46].

CD68, NLRP3, IL-1β and IL-18 protein levels were measured within Iba1⁺ cells.

Automated total cell counts and analysis of overlapping regions between markers to assess colocalization were performed in Imaris XT software (Bitplane AG, Oxford Instruments, Abingdon-on-Thames, UK). For all double-positive cell counting analyses, the Surfaces function was used. Briefly, after applying background subtraction and automatic thresholding, to retain only objects within lower and upper threshold Limits, and splitting of touching objects to separate individual cells, colocalization between markers was quantified using the Overlapped Volume Surfaces option. Complement component 3 (C3) levels were quantified within GFAP⁺ cells, while MAP2 intensity was quantified by setting 12 frames (90 × 90 pixel) over the stratum radiatum. Intracellular Aβ levels were quantified by setting 10 randomly-distributed frames on the hippocampal CA1 pyramidal layer (70 × 70 pixel).

To analyze the number of Aβ plaques and the AT8⁺ area in the hippocampus, we acquired Z-stack large images with a 20x-objective. The mean number of 6E10⁺ plaques in the hippocampus from at least 3 slices/animal was quantified. To calculate AT8⁺ area, the hippocampus was manually defined, and automatic brightness thresholding was used to delineate the positive area. Hippocampal AT8⁺ area was expressed as ratio of total area analyzed in each slice/animal (modified from [47]). Intracellular AT8 levels were quantified by Z-stack with 60x-oil objective and by setting 1 frame (20 × 20 pixel).

For all analyses, quantification was done on 4 slices/mouse. Data were then averaged per mouse for figures and statistical analysis.

Primary antibodies: 5-HT (1:500; ImmunoStar #20080; RRID:AB_572263), AT8 Ser202/Thr205 (1:200; Invitrogen #1020; RRID:AB_223647), C3 (1:300; Novus Biologicals #NB200-540; RRID:AB_2744548), C3aR (1:200; HycultBiotech #HM3028; RRID:AB_2131309), CD68 (1:400; Biorad #MCA1957; RRID:AB_322219), DAT (1:400; Chemicon #MAB369; RRID:AB_2190413), GFAP (1:1000; Millipore #AB5804; RRID:AB_2109645), GFAP (1:1000; DAKO #Z0334; RRID:AB_2314535), hAPP695 (6E10) (1:500; BioLegend #803001; RRID: AB_2564653), Iba1 (1:600; Wako #019–19741; RRID:AB_839504), AIF-1/Iba1 (1:600; Novus Biologicals #NB100-1028; RRID:AB_3148646), IL-1β (1:200; R&D #AF-401-NA; RRID:AB_416684), IL-18 (1:300; MBL #D047-3; RRID:AB_592016), IL-18R⍺/IL-1 r5 (1:100; R&D #AF856; RRID:AB_355664), MAP2 (1:500; Invitrogen #MA5-12,826; RRID:AB_10976831), NET (1:500; Atlas Antibodies #AMAb91116; RRID:AB_2665806), NLRP3 (1:200; Adipogen #AG-20B-0014; RRID:AB_2490202), p-NFκB (Ser536) (1:200; Cell Signaling #3033; RRID:AB_331284), S100β (1:500; SYSY #287 006; RRID:AB_2713986), SERT (1:500; Millipore #PC177L; RRID:AB_2122553), TH (1:1000; Millipore #MAB318; RRID:AB_2201528), TH (1:500; Millipore #AB152; RRID: AB_390204).

Secondary Antibodies (Thermo Fisher): Alexa Fluor-488 donkey anti-rabbit (1:200; #A-21206; RRID:AB_2535792), Alexa Fluor-555 donkey anti-rabbit (1:200; #A31572; RRID:AB_162543), Alexa Fluor-647 goat anti-rabbit (1:200; #A-31573; RRID:AB_2536183), Alexa Fluor-488 donkey anti-mouse (1:200; #R37114; RRID:AB_2556542), Alexa Fluor-555 donkey anti-mouse (1:200; #A31570; RRID:AB_2536180), Alexa Fluor-488 donkey anti-rat (1:200; #A21208; RRID:AB_2535794), Alexa Fluor-647 goat anti-rat (1:200; #A-21247; RRID:AB_141778), Alexa Fluor-488 donkey anti-goat (1:200; #A-11055; RRID:AB_2534102), Alexa Fluor-647 donkey anti-goat (1:200; #A-21447; RRID:AB_141844), Alexa Fluor-555 goat anti-chicken (1:200; # A-21437; RRID:AB_2535858).

Exclusively for the representative confocal images, after the quantitative analysis, LUTs were equally increased at the same level for all groups of a given experiment. Quantitative analyses were performed on raw images.

Immunofluorescence sections were used to estimate: TH⁺ neurons in the left SNpc, VTA and LC, 5-HT⁺ neurons in the entire IPN; 5-HT⁺ neurons in the dorsal (dRaphe) and medial Raphe (mRaphe); Iba1⁺ cells in the hippocampus, dorsal and ventral striatum and GFAP⁺ cells in the hippocampus. The area boundaries for VTA/SNpc/LC were defined by TH, the IPN/dRaphe/mRaphe by 5-HT, the dorsal and ventral striatum and hippocampus by DAPI staining, using the 5x-objective, in accordance to Paxinos guidelines [43].

We applied an optical fractionator stereological design using the Stereo Investigator System (MBF Bioscience). A stack of MAC5000 controller modules (Ludl Electronic Products, Ltd) was interfaced with a Zeiss Microscope Axio Imager KMAT with a motorized stage and a Zeiss Axiocam 506 mono with Working High End PC. A 3D-optical fractionator counting probe (x, y, z dimension 50 × 50 × 25 μm for TH⁺ neurons, 70 × 70 × 25 μm for 5-HT⁺ neurons and 100 × 100 × 25 μm for glia cells) was applied. Cells were marked with a 100x-oil-(VTA/SNpc) or a 40x-objective (neurons in LC/IPN/dRaphe/mRaphe; glia cells in hippocampus and striatum).

The total cell number was estimated according to the formula (Eq. ):where SQ represents the cell number counted in all optically sampled fields of the ROI, ssf is the section sampling fraction, asf is the area sampling fraction and tsf is the thickness sampling fraction. 1 1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{N }=\text{ SQ }\times (1/\text{ssf}) \times (1/\text{asf}) \times (1/\text{tsf})$$\end{document} N SQ ssf asf tsf = × ( / ) × ( / ) × ( / ) 1 1 1

Microglia were imaged with a Zeiss Microscope Axio Imager KMAT with motorized stage and a camera connected to Neurolucida software (7.5v; MBF Bioscience) for quantitative 3D-analysis of the entire cell [48, 49]. Only non-overlapping cells that showed clear soma and branching were analyzed. Soma area and perimeter were measured; Sholl analysis included counting the number of dendritic intersections, nodes and endings, and dendritic lengths at fixed distances from the soma in 10 μm-spaced concentric circles originating from the soma. Analysis was done with 100x-oil objective. Nine representative cells/animal were analyzed randomly and data were averaged for each mouse.

The ipsilateral hippocampus was dissected from the entire brain and stored at −80 °C until the day of the experiment. The ipsilateral midbrain was isolated from 500 μm-frozen slices using a biopsy punch. Tissue was homogenized in RIPA buffer containing (in mM): 50 Tris–HCl pH 7.5, 150 NaCl, 5 MgCl₂, 1 EDTA, 1% Triton X-100, 0.25% sodium deoxycholate, 0.1% SDS, 1 sodium-orthovanadate, 5 b-glycerophosphate, 5 NaF and protease inhibitor cocktail; samples were then sonicated four times (five strokes of 0.5 pulse/s) and incubated on ice for 30 min [50]. Samples were centrifuged (13000 g, 20 min, + 4 °C) and the supernatant's protein concentration was determined by the Bradford method. Proteins were applied to SDS-PAGE and electroblotted on a polyvinylidene-difluoride membrane. Blotting analysis was performed using a chemiluminescence detection kit. The relative levels of immunoreactivity were determined by densitometry using ImageJ.

Primary antibodies: Caspase-3 (1:500; Cell Signaling Technology; #9662s; RRID:AB_331439), Glycogen Synthase Kinase 3β (GSK3β; 1:1000; Cell Signaling Technology; #9832s; RRID:AB_10839406); p-GSK3β (Ser9) (1:1000; Cell Signaling Technology; #9336s; RRID:AB_331405); JNK (1:1000; Cell Signaling Technology; #9252s; RRID:AB_2250373); p-JNK (Thr183/Tyr185) (1:1000; Cell Signaling Technology; #9251s; RRID:AB_331659); p-p38 (Thr180/Tyr182) (1:1000; Cell Signaling Technology; #4511s; RRID:AB_2139682); p38 (1:1000; Cell Signaling Technology, #8690s; RRID:AB_10999090); TH (1:500; Abcam, #ab112; RRID:AB_297840); β-Tubulin (1:1000; Biolegend; #801201; RRID:AB_2313773). All primary antibodies were incubated overnight except for β-Tubulin (2 h).

Secondary antibodies: goat anti-mouse IgG (1:3000; Bio-Rad; #1706516; RRID:AB_11125547), goat anti-rabbit IgG (1:3000; Bio-Rad; #1706515; RRID:AB_2617112).

Membranes were stripped using Re-Blot Plus Strong Solution (Millipore; 15 min, RT). Both groups (Shan vs Casp3; Tg Sham vs Tg Casp3) were analyzed simultaneously.

Quantification of hippocampal monoamines (DA, 5-HT, NE), and relative metabolites (DOPAC and HVA for DA, 5-HIAA for 5-HT, MOPEG for NE) was performed using an HPLC system (UltiMate® 3000, Thermo Fisher) coupled with Coulochem electrochemical detection (6011RS Ultra Coulometric Analytical Cell, Thermo Fisher). After animal sacrifice, the left hippocampus was rapidly dissected on ice and immediately stored at −80 °C until analysis. On the day of analysis frozen samples were homogenized on ice with 0.05 M HClO₄ and antioxidant solution (containing in mM: 0.27 Na₂EDTA, 100 acetic acid, 0.0125 ascorbic acid) in a 4:1 ratio. The homogenate was mechanically lysed, sonicated on ice and centrifuged (10000 rpm, 20 min, 4 °C). The supernatant was transferred into a new tube and the pellet was weighed. 20 µL of each sample were injected onto the HPLC-ECD (run-time: 60 min; flow-rate: 0.6 mL/min). Standards of each metabolite were prepared fresh on the day with same solutions and quantities as for tissue samples. The chromatographic separation was performed on a Hypersil GOLD aQ-C18 column (150 × 3 mm, 5 µm) fitted with an aQ-C18 drop-in guard pre-column (10 × 3 mm, 5 µm) maintained at 37 °C. The mobile phase consisted of 5% methanol and buffer solution (0.1 M Na-Phosphate, 0.1 mM Titriplex® III and 0.5 μM 1-Octanesulfonic-Acid Na-salt, pH 3.6 adjusted with 85% Ortho-Phosphoric-acid), filtered through a 0.22 µm cellulose-ester membrane (Millipore). The potential applied to both dual-inline flow-through micro-porous graphitic carbon-working electrodes was set at + 450 mV, with 1 nA-gain for ECRS1 and 10 nA-gain for ECRS2. The chromatograms were integrated with Chromeleon™ Software (7.0v; Thermo Fisher). The sample concentration of metabolites was calculated from the corresponding peak height, normalized to the pellet weight.

Following head dislocation, mice were decapitated and the brain was rapidly removed; 300 μm-thick parasagittal slices containing the left dorsal hippocampus were cut (Leica VT1200S vibratome) in oxygenated (95% O₂, 5% CO₂) ice-cold sucrose-based solution (in mM: 3 KCl, 1.25 NaH₂PO₄, 26 NaHCO₃, 10 MgSO₄, 0.5 CaCl₂, 25 glucose, 185 sucrose; ~ 300 mOsm, pH 7.4). Brain slices were left to recover in artificial Cerebro-Spinal Fluid (aCSF; containing in mM: 124 NaCl, 1.25 NaH₂PO₄-H₂O, 26 NaHCO₃, 3 KCl, 10 Glucose, 1 MgSO₄, 2 CaCl₂, ~ 300 mOsm, pH 7.4) at 32–34 °C for 40 min, and moved at RT for at least 30 min before recording [55].

For recording, each slice was placed under an Olympus BX51WI microscope and perfused with oxygenated aCSF (3-4 mL/min, 30-32 °C). The hippocampus was visualized with infrared differential-interference-contrast at 4x.

Recordings were performed using a MultiClamp700B amplifier, digitized with Digidata1550B and computer-saved with pClamp11 (Molecular Devices). Patch-pipettes (3–5 MΩ), filled with aCSF, were pulled from TW150F-4 glass tubes (WPI). Field excitatory postsynaptic potentials (fEPSPs), recorded in the stratum radiatum, were induced with Schaffer collateral stimulation (100 μs), acquired at 20 kHz and filtered off-line with a 10 kHz low-pass Bessel filter. Input–Output (I/O) curves of fEPSP slopes were obtained at 10 μA-stepped increasing stimulation every 30 s. The Paired-Pulse Ratio (PPR) was evaluated with pairs of stimuli (20–1000 ms interval), at half-maximal stimulation. For Long-Term Potentiation (LTP), after 20 min of test stimulation (half-maximal intensity, every 30 s) to assess slope stability, the slice was challenged with two trains at 100 Hz (1 s duration, 20 s interval) followed by test stimulation for 1 h. The LTP magnitude was evaluated as the fEPSP mean slope at 60 min after the conditioning trains, normalized to the mean baseline slope [45, 56, 57]. To assess the effect of the IL-1β receptor antagonist (IL-1β-Ra) on LTP, slices were pre-incubated with IL-1β-Ra (100 ng/mL, from freshly-thawed stock in PBS/bovine-serum albumin 0.1%; R&D Systems) in oxygenated aCSF for 1 h and recorded as above.

Coronal slices (240 μm-thickness) were used for spine density analysis. CA1 pyramidal neurons were filled with biocytin during whole-cell patch clamp. Freshly weighted biocytin (0.2%; Tocris) was added to intracellular solution containing (in mM): Cs-methanesulfonate 120, CsCl 15, NaCl 8, HEPES 10, EGTA 0.20, TEA-Cl 10, QX314-Cl 5, Mg-ATP 2, Na-GTP 0.3 (~ 275–285 mOsm, pH 7.4). Patch-pipettes used for cell filling had a tip resistance of 4-10 MΩ and a minimum filling time of 10 min was used for each cell. Access resistance was monitored throughout this time to check for accessibility to the cytoplasm. To avoid cell damage after filling, the electrode was removed from the slice by establishment of an outside-out patch.

For analysis of biocytin-filled neurons, slices bearing neurons loaded with biocytin were fixed by immersion in 4% paraformaldehyde in PB (overnight, 4 °C). Then, slices were washed three times in PB and incubated with Streptavidin (1:750; Alexa Fluor 555-conjugated #S32355, Thermo Fisher; RRID:AB_2571525) and Green-Fluorescent Nissl-Stain (NeuroTrace 1:600; Invitrogen #N21480) in PB containing 0.3% Triton X-100 (overnight, 4 °C).

The Z-stack confocal images were captured using Nikon Eclipse Ti2 confocal microscope with 60x-oil objective and 4.5 digital zoom. We analyzed spines in the distal (terminal) apical dendrites.

Spine density was assessed by counting the number of spines in at least 18 segments per neuron (4 cells, 5 animals per experimental group) chosen in branch orders 2–4 of apical dendrites extending for an additional 30–60 μm away from the starting point [58]. Imaris software (9.8.2) was utilized to create 3D reconstructions of dendritic segments using the Filaments tool. The settings used for all reconstructions is the same as in [59]. An observer blinded to the experimental groups manually edited the reconstructions to include or exclude misidentified spines.

All behavioral tests were conducted between 09:00 a.m. and 04:00 p.m.. Mice were habituated to the experimenter through daily handling sessions for one week prior to the onset of behavioral testing. In addition, animals were acclimated to the testing room for one hour immediately before the start of each experimental session.

To minimize potential olfactory interference, the chambers and objects were cleaned between animals and/or sessions using a 5% ethanol solution.

In Casp3 and Sham mice we next examined the midbrain rostro-caudally to confirm the lesion in the VTA, SNpc and IPN (Fig. 1C). The stereological counting of TH⁺ neurons in the ipsilateral VTA and SNpc revealed a significant decrease in Casp3 compared to Sham mice 1 month after the lesion (Fig. 1D). This was accompanied by reduced total TH protein levels in the midbrain (Supplemental Fig. 1F). Consistent with the ectopic activity of the TH promoter in the IPN, 5-HT⁺ neurons were also significantly reduced in Casp3 mice (Fig. 1E). Since the midbrain receives TH⁺ fibers from the LC, we also investigated whether the viral infusion could lead to the retrograde loss of TH⁺ LC neurons (Fig. 1F). Yet, no changes were observed in the number of LC TH⁺ neurons in Casp3 mice (Fig. 1G).

To confirm the reduction of midbrain monoamine inputs to the hippocampus of Casp3 mice, 1 month after surgery we analyzed the dopaminergic and serotonergic innervations in the dorsal hippocampus (Fig. 1H). The density of hippocampal TH⁺ fibers was significantly reduced in Casp3 mice (Fig. 1I). This was not due to damage to the TH⁺ innervation from the LC, as the density of NET⁺ fibers was unaltered (Fig. 1J). Consistent with the significant loss of IPN 5-HT⁺ neurons, in Casp3 mice there was also a reduction in the density of hippocampal serotonergic fibers, labelled for the 5-HT transporter (SERT; Fig. 1K). Of note, the stereological cell counting of 5-HT⁺ neurons in both the dorsal and medial Raphe showed no changes in Casp3 mice (Fig. 1L,M), confirming that the reduction of hippocampal SERT⁺ fibers is solely due to the IPN lesion.

In Line with the loss of dopaminergic and serotonergic innervation in the hippocampus, HPLC analysis of total hippocampal tissue confirmed a drop of both DA and 5-HT and their relative metabolites (DOPAC, HVA and 5-HIAA) in Casp3 mice (Fig. 1N,O; Supplemental Table 1). No changes in NE and MOPEG levels were observed (Fig. 1P; Supplemental Table 1). Thus, the midbrain lesion in Casp3 mice results in the combined depletion of both DA and 5-HT in the hippocampus.

One of the key contributors in the development of neuroinflammation is the NLRP3 inflammasome, responsible for the secretion of pro-inflammatory cytokines such as IL-1β and IL-18 [75, 76]. Therefore, we examined the number of Iba1⁺ cells expressing NLRP3 (NLRP3⁺/Iba1⁺) and the levels of NLRP3 and IL-1β. Cell counting revealed a significant increase in the percentage of NLRP3⁺/Iba1⁺ cells in Casp3 compared to Sham mice (Fig. 2E, left). Moreover, we found increased NLRP3 immunoreactivity (Fig. 2E, right), associated with higher IL-1β levels (Fig. 2F), proving a pro-inflammatory profile. Notably, similar microglial alterations were also observed in female Casp3 compared to sex-matched Sham mice (Supplemental Fig. 3D-J), indicating that the midbrain lesion induces microglia-driven hippocampal neuroinflammation irrespective of sex.

To assess the transcriptional signatures associated with microglia reactivity, we performed mRNA sequencing on FACS-sorted hippocampal microglia from Sham and Casp3-lesioned CX₃CR-1^GFP mice at 1 month post-lesion (Supplementary Fig. 4A-D). Among the 19117 genes detected, six Differentially Expressed Genes (DEGs) reached significance after FDR correction (FDR < 0.05), and 381 met a nominal threshold (p < 0,05), as illustrated in the volcano plot (Supplementary Fig. 4E). These included 138 up- and 243 down-regulated genes in Casp3 mice relative to Sham microglia, with fold changes and associated statistics detailed in Supplementary Table 2.

We next examined the full set of DEGs using ToppGene functional annotation to identify perturbed biological processes. The top enriched Gene Ontology (GO) terms within the "Biological Process" category are shown in Fig. 2G, with the complete list in Supplementary Table 3. Consistent with a shift toward a reactive phenotype, Casp3 microglia exhibited significant enrichment for terms related to microglial cell activation (GO:0001774), chemotaxis (GO:0006935), and neuroinflammatory response (GO:0150076). Additional enriched processes included apoptotic regulation (GO:0042981) and translational machinery dynamics such as cytoplasmic translation (GO:0002181), synaptic translation (GO:0140241) and ribosome assembly (GO:0042255).

Collectively, these results prove that the combined depletion of both midbrain DA and 5-HT in Casp3 mice triggers hippocampal neuroinflammation via microglia activation and NLRP3 pathway induction.

Given the well-established role of monoamines in modulating hippocampal function, we next asked whether this neuroinflammatory environment compromises neuronal dendritic structure. Morphological analysis of biocytin‑filled CA1 pyramidal neurons revealed a significant reduction in distal apical spine density in Casp3 mice relative to Sham controls (Fig. 2H). To determine whether this structural alteration translated into a functional deficit, we assessed LTP at CA3–CA1 synapses. LTP appeared unaltered in Casp3 slices (Supplemental Fig. 5A), suggesting that elevated IL‑1β levels might mask a synaptic deficit, an interpretation consistent with prior reports indicating that IL‑1β can enhance LTP [77]. Indeed, pre‑incubation with the IL‑1β receptor antagonist (IL‑1β‑Ra) unmasked a clear LTP impairment in Casp3 mice (Supplemental Fig. 5A), despite unchanged glutamate release probability (PPR) and Input/Output (I/O) curves (Supplemental Fig. 5B).

To prove that the hippocampal microglia response requires the loss of both DA and 5-HT, we examined the striatum, which receives robust dopaminergic but negligible IPN‑derived serotonergic innervation [78]. Retrograde labelling confirmed that the dorsal and ventral striatum (NAc core/shell) do not receive inputs from IPN neurons (Supplemental Fig. 6A-D). The analysis of TH, DAT and SERT markers confirmed that the dopaminergic innervation in the dorsal striatum and NAc core/shell is impaired in Casp3 mice (Supplemental Fig. 6E,F), while the serotonergic fibers are intact (Supplemental Fig. 6G,H). In contrast to what we observed in the hippocampus, there were no changes in Iba1⁺ cell number and morphology in the dorsal striatum (Supplemental Fig. 7A-E), NAc core (Supplemental Fig. 7F-J) or NAc shell (Supplemental Fig. 7K-O), suggesting that the loss of DA alone is insufficient to induce microglia reactivity in our model.

As an additional confirmation, we examined the hippocampus of mice infused in the midbrain with the catecholaminergic neurotoxin 6OHDA that induces selective loss of VTA and SNpc dopaminergic neurons but does not affect IPN 5-HT⁺ neurons (Supplemental Fig. 8A-C). Analysis of microglia in the hippocampus of 6OHDA mice showed no differences in cell number and morphology compared to saline-injected mice (Supplemental Fig. 8D-G). In line with these results, no changes were detected in the levels of CD68, NLRP3 or IL-1β in Iba1⁺ cells of 6OHDA mice (Supplemental Fig. 8H-J).

Similarly, to assess whether the selective loss of the serotonergic component alone is sufficient to elicit hippocampal neuroinflammation, a separate cohort of mice received targeted IPN injections (Casp3^IPN mice) of the AAV1-THp-iCre + AAV5-flex-taCasp3-TEVp mix, selectively lesioning 5-HT⁺ neurons while sparing midbrain dopaminergic neurons (Supplemental Fig. 9A-C). Compared with Sham-injected controls (Sham^IPN), Casp3^IPN mice showed a selective reduction of hippocampal SERT⁺ fibers and preserved TH⁺ innervation (Supplemental Fig. 9D, E). Notably, serotonergic denervation alone failed to induce microglia reactivity, as evidenced by unchanged microglia number, morphology (Supplemental Fig. 9F-I) and levels of CD68, NLRP3, or IL-1β immunoreactivity in Iba1⁺ cells (Supplemental Fig. 9J-L), arguing against a pro-inflammatory microglia response.

Taken together, these findings demonstrate that only the combined loss of midbrain DA and 5‑HT innervation synergistically drives hippocampal neuroinflammation, dendritic spine loss, and associated deficits in synaptic plasticity, key pathological features contributing to memory impairment in AD.

DA and 5-HT receptors are expressed in both microglia and astrocytes. Indeed, both monoamines negatively regulate glial cell reactivity by inhibiting the NLRP3 inflammasome and the subsequent release of pro-inflammatory cytokines [14, 15, 79, 80].

Similarly, the chronic treatment of Casp3 mice with the selective 5-HT reuptake inhibitor (SSRI) fluoxetine, started one day after lesioning and continued for 30 days (Fig. 4A), prevented microglia reactivity as shown by the reduced number and morphological complexity of Iba1⁺ cells compared to vehicle-treated Casp3 mice, reaching those of Sham mice (Fig. 4B,C; Supplemental Fig. 10C,D). Fluoxetine also attenuated the activation of the NLRP3 pathway, as demonstrated by the lower percentage of NLRP3⁺/Iba1⁺ cells and the lower levels of NLRP3 and IL-1β compared to vehicle-treated Casp3 mice (Fig. 4D,E). Collectively, these results indicate that both dopaminergic and serotonergic treatments can protect Casp3 mice from NLRP3-mediated neuroinflammation (Fig. 4F).

To assess whether the hippocampus of naïve Tg2576 mice (Fig. 5A) is affected by microglia reactivity, we examined microglia density and morphological changes. We found a greater number of Iba1⁺ cells in Tg2576 mice (Fig. 5B). These cells had also an increased number of intersections, nodes, endings, and length of their processes, as well as a significant soma enlargement (Fig. 5C, Supplemental Fig. 12A,B) accompanied by enhanced CD68 immunoreactivity (Fig. 5D). Furthermore, we observed a significant increase in the percentage of NLRP3⁺/Iba1⁺ cells in Tg2576 compared to WT mice, along with increased microglia NLRP3, IL-1β and IL-18 immunoreactivity (Fig. 5E-G). Collectively, these data indicate that Tg2576 mice exhibit hippocampal neuroinflammation mediated by microglia through the activation of the NLRP3 inflammasome pathway, which coincides with the loss of DA signaling from the VTA.

To validate our hypothesis that the enhanced loss of midbrain monoamines could aggravate the microglia response, we next examined Tg Casp3 mice, obtained by unilateral co-infusion of the AAV1-THp-iCre and AAV5-flex-taCasp3-TEVp viruses into the left midbrain of 6-month-old Tg2576 mice (Fig. 6A). Consistent with the evidence that DA and 5-HT deficiency triggers microglia-mediated neuroinflammation in the hippocampus of C57BL/6N mice (see Fig. 2), the lesion significantly exacerbated the microglia response in Tg Casp3 compared to Tg Sham mice. Specifically, stereological cell counting and Sholl analysis of Iba1⁺ cells in Tg Casp3 mice 1 month post-lesion revealed an increase in cell numbers and hyper-ramification of microglia processes (Fig. 6B,C; Supplemental Fig. 12C,D). Additionally, Iba1⁺ cells of Tg Casp3 mice showed significantly higher NLRP3, IL-1β and IL-18 immunoreactivity compared to Tg Sham mice (Fig. 6D-F), proving a worsened pro-inflammatory microglial state.

Previous works have shown that reactive microglia can activate astrocytes which, in turn, participate in the inflammatory response by activating the NFκB pathway, leading to the release of Complement 3 (C3). This cascade can further enhance the microglia reactivity impairing Aβ phagocytosis and can promote Aβ and tau pathology in neurons [82 –86]. Here, we explored the hypothesis that in Tg Casp3 mice the severe microglia reactivity induced by the midbrain lesion can lead to astrocytic response and, thus, trigger the pathogenic cycle described above.

Before doing so, we examined astrocyte reactivity in the hippocampus of naïve Tg2576 mice. Unlike microglia (see Fig. 5), astrocytes do not appear to be responsive in naïve Tg2576 at 7 months of age, as shown by count of GFAP⁺ cells and the absence of NLRP3 immunoreactivity (Supplemental Fig. 12E,F). Additionally, we examined the microglia-astrocyte crosstalk pathway involving the activation of the IL-18R in astrocytes and the subsequent phosphorylation of NFκB [87]. Consistently with the absence of astrocyte reactivity in the naïve Tg2576 hippocampus, confocal analysis of IL-18R and p-NFκB in astrocytes (immunolabelled by GFAP or S100β) revealed comparable numbers of astrocytes expressing IL-18R and p-NFκB between Tg2576 and WT mice (Supplemental Fig. 12G,H).

To further investigate the astrocyte-neuron and astrocyte-microglia crosstalk in Tg Casp3 mice in response to elevated astrocytic C3, we assessed the C3a receptor (C3aR) levels in hippocampal CA1 pyramidal neurons and microglia. Both cell types exhibited an approximately 2-fold increase in C3aR immunoreactivity (Fig. 7E), which, in the case of microglia, may explain the exacerbated microglial response observed in Tg Casp3 compared to Tg Sham mice (see Fig. 6). The astrocyte-neuron communication via the intracellular C3 pathway has been shown to drive Aβ and tau pathology as well as synaptic dysfunction in neurons, potentially through modulation of different kinases, including GSK3β and the Mitogen-Activated Protein Kinases (MAPKs) such as JNK and p38 [82, 83, 88, 89]. We therefore quantified the phosphorylated levels of GSK3β, JNK and p38 and observed a significant reduction of p-GSK3β, indicating an increased GSK3β activity in Tg Casp3 compared to Tg Sham mice (Fig. 7F). Instead, the levels of p-JNK and p-p38 were unchanged in Tg Casp3 mice (Supplemental Fig. 12J). Given the known role of GSK3β in the progression of tau pathology [90, 91], we next evaluated tau hyperphosphorylation (immunolabelled with the AT8 antibody). We found a stronger AT8 immunoreactivity in the hippocampus of Tg Casp3 mice, associated with enhanced intracellular levels of hyperphosphorylated tau (Fig. 7G). Of note, the levels of phosphorylated tau in the hippocampus of 7-month-old Tg Sham mice are similar to those of WT littermates (Fig. 7G), suggesting that midbrain lesion in Tg2576 mice drives secondary tau pathology. The increased activity of GSK3β has also been associated with reduced microtubule stabilization and aggravated Aβ pathology [92 –94]. Accordingly, we found a diffuse reduction of the dendritic MAP2 in pyramidal neurons of Tg Casp3 mice, suggestive of pathological changes in dendritic integrity (Fig. 7H), and increased intracellular Aβ levels (Fig. 7I), paralleled by the appearance of precocious Aβ plaque accumulation (Fig. 7J).

Collectively, these results demonstrate that both dopaminergic and serotonergic treatments effectively suppress microglial pro-inflammatory cytokine release, astrocyte reactivity, and the downstream tau hyperphosphorylation in the hippocampus of Tg Casp3 mice.