Animal procedures were performed in strict compliance with protocols approved by the Italian Ministry of Public Health and the local Ethics Committee of the University of Pisa, in accordance with the European Directive for the Care and Use of Animals 2010/63/EU (project license no. 39E1C.N.5Q7 approved on 30/10/2021). C57BL/6 J mice were used. Animals were maintained in a regulated environment (23 ± 1°C, 50% ± 5% humidity) with a 12-h light–dark cycle and food and water ad libitum. The here described hiPSCs are commercially available,¹ in several modifications, and are listed in several registries (e.g., https://hpscreg.eu/cell-line/UCSFi001-A). It is thus an established and commercially available cell line.

For primary mouse HNs culture, P0/P1 (postnatal day 0/1) stage mice were sacrificed and both hippocampi were explanted and dissected in a solution of ice-cold 6.5 mg/mL D-glucose in Dulbecco’s Phosphate-Buffered Saline (DPBS, Gibco, #14190-144). Tissue was digested for 10 min in 0.25% Trypsin (Gibco, #15050-065) at 37°C and for 5 min in 0.25% Trypsin added with 1% DNase 10 mg/mL (Sigma-Aldrich, #DN25) at 37°C. After centrifugation at 1,200 rpm for 1 min, mechanical dissociation was carried out by passing the cells through a Pasteur glass pipette (VWR, #612-1702) 30 to 40 times. Cells were seeded at cell density of 50,000 cells/cm² in high glucose Dulbecco’s Modified Eagle Medium (DMEM, Gibco, #21063-029) modified with 10% fetal bovine serum (FBS, Gibco, #10270-106), 100 IU/mL penicillin, 100 μg/mL streptomycin (Gibco, #15140-122), and GlutaMAX supplement (Gibco, #35050-038). Cells were seeded on 13 mm circular plastic coverslips (Sarstedt, #83.1840.002) pre-coated with 100 μg/mL Poly-L-lysine (PLL, Sigma-Aldrich, #P4707) and incubated at 37°C in a saturated humidity atmosphere containing 95% air and 5% CO₂. The medium was replaced at day in vitro 0 (DIV0) 4 h after seeding with Neurobasal-A medium (Gibco, #12349-015) modified with B27 supplement (Gibco, #17504-044), GlutaMAX, 50 IU/mL penicillin, 50 μg/mL streptomycin, and 5 μg/mL MNPs.

Human cortical neurons were derived from hiPSC line Ngn2-WTC11 as generated in (38). hiPSCs were routinely cultured in StemMACS iPS-Brew-XF medium (Miltenyi Biotec, #130-104-368) supplemented with Antibiotic-Antimycotic Solution (Merck, #A5955) on plates pre-coated with 200 μg/mL Geltrex (Thermo Fisher Scientific, #A1413302) in KnockOut DMEM (KO-DMEM, Thermo Fisher Scientific, #10829018). Cells were incubated at 37°C in a saturated humidity atmosphere containing 95% air and 5% CO₂ and regularly passaged when 80% confluent in a 1:10 ratio, using Versene (Thermo Fisher Scientific, #15040066) and StemMACS iPS-Brew-XF supplemented with 2 μM thiazovivin (Axon Medchem, #Axon1535) for the first 24 h. For pre-differentiation, on the first day (DIV0), cells were incubated in Accutase (Sigma-Aldrich, #A6964) for 5 min at 37°C; then, DPBS was added, and cells were centrifuged at 400 g for 5 min. Supernatant was discarded, and the pellet was re-suspended in StemMACS iPS-Brew-XF. Cells were seeded onto Geltrex-coated plates in pre-differentiation medium consisting of KO-DMEM/F12 (Thermo Fisher Scientific, #12660012) containing N2 supplement (Pan-Biotech, #P07-11010), non-essential amino acids (NEAA, Thermo Fisher Scientific, #11140035), 1.5 μg/mL mouse laminin (Sigma, #L2020), 10 ng/mL brain-derived neurotrophic factor (BDNF, Peprotech, #450-02), 10 ng/mL neurotrophin-3 (NT3, Peprotech, #450-03), 2 μM thiazovivin, 2 μg/mL doxycycline (Merck, #D9891), and Antibiotic-Antimycotic solution. The medium was changed daily for 2 days to fresh pre-differentiation medium without thiazovivin. The duration of the pre-differentiation protocol is 3 days (from DIV0 to DIV2). Hereafter, we refer to these cells as induced neurons (iN) when the precursors have completed the pre-differentiation protocol and are at any stage of terminal differentiation (24 ≥ DIV≥3). For terminal differentiation (DIV3), cells were dissociated as described for pre-differentiation and seeded onto 13 mm circular plastic coverslips pre-coated with 20 μg/mL Poly-D-Lysine (Sigma-Aldrich, #P7886) and 20 μg/mL Cultrex 3-D Laminin I (Biotechne, #3446-005-01) in maturation medium consisting of 50% DMEM/F12 (Thermo Fisher, #11320033), 50% Neurobasal-A, B27 supplement, N2 supplement, GlutaMax, NEAA, 1.5 μL/mL mouse laminin, 10 ng/mL BDNF, 10 ng/mL NT3, 2 μg/mL doxycycline, and Antibiotic-Antimycotic solution at a seeding density of 50,000 cells/cm². The medium was replaced 4 h after seeding by maturation medium supplemented with 5 μg/mL MNPs.

For both cell cultures, the coverslips were placed inside 35 mm Petri dishes (VWR, #734-2317) that are compatible with the insertion inside the magnetic applicator.

The MNPs used in this study are the FluidMAG-ARA (Chemicell, #4115). They are characterized by an organic iron oxide core with a diameter of 75 ± 10 nm and a magnetic saturation of 59 Am²/Kg, as stated from the supplier. They are coated with a polysaccharide shell composed of glucuronic acid, making the hydrodynamic diameter approximately 100 nm. MNPs were added to the cell growth medium at a concentration of 5 μg/mL.

Twenty-four hours after seeding, an external static magnetic field was applied to the cell cultures by placing the 35 mm Petri dishes within a toroidal magnetic applicator Halbach-type that provides a constant magnetic field gradient (46.5 T/m) in the radial centrifugal direction (27, 39). Cells were maintained inside the magnet for 48 h of stimulation.

For oTau preparation, 1 μM recombinant human hT40 protein (TAU441, Abcam, #ab191460) was incubated with 18.75 μM arachidonic acid (Cayman Chemicals, #90010.1) for 15 min at room temperature (RT) in a polymerization buffer consisting of 10 mM HEPES pH 7.6, 100 mM NaCl, 5 mM DTT, and 0.1 mM EGTA. The solution was stored at −80° C. prior to treatment, the solution was centrifuged at 1,400g for 5 min; then, oTau were administered to primary mouse HNs in conditioned growth medium at a final concentration of 50 nM, 24 h after seeding (DIV1).

For oAβ preparation, Aβ40 and Aβ42 powder (21st Century Biochemicals, #AB40-0010 and #AB42-0010) were dissolved in Hexafluoro-2-propanol (HFIP, 100%) to 1 mM concentration. Then, the HFIP was completely evaporated overnight (ON) in a Speedvac centrifugational evaporator, and the lyophilized powder aliquots were stored at −80°C. For reconstitution, directly before use, dried aliquots were dissolved in 50 mM NaOH; then, Aβ40 and Aβ42 were mixed in a 7:3 ratio and added with DPBS and 50 mM HCl obtaining a final Aβ concentration of 100 μM. To induce oAβ formation, the Aβ mixture was incubated at 37°C for 1 h; then, oAβ were administered to hiPSC-derived cortical iNs in conditioned maturation medium at a final concentration of 1 μM, 24 h and 48 h after seeding (Day 1 and Day 2 of terminal differentiation).

At DIV3 for HNs and at DIV6 for iNs, cells were fixed, immunostained, and imaged.

For HNs, after two washes with DPBS, cells were fixed in 2% w/v paraformaldehyde (PFA) and 7.5% w/v sucrose (Sigma-Aldrich, #S0389) in DPBS at RT for 20 min. Samples were washed three times with DPBS and permeabilized in 0.5% v/v Triton X-100 (Sigma-Aldrich, #X100) in DPBS at RT for 10 min. After three 3-min washes in 0.1% v/v Triton X-100 in DPBS, cells were blocked in 5% v/v goat serum (Gibco, #16210-064) and 0.3% v/v Triton X-100 in DPBS for 1 h at RT and incubated with the primary antibodies ON at 4°C in 3% v/v goat serum and 0.2% v/v Triton X-100 in DPBS. After ON incubation, samples were washed and incubated with secondary antibodies and Hoechst 33342 (Invitrogen, #H3570, dilution 1:1,000) in 3% v/v goat serum and 0.2% v/v Triton X-100 in DPBS and then washed three times in DPBS.

For iNs, after one wash in DPBS, cells were fixed in 3.7% w/v PFA and 4% sucrose in DPBS at RT for 1 h. After three washes in PBS, cells were permeabilized and blocked in 0.5% v/v Triton X-100 and 5% w/v BSA (Carl Roth, #8076.4) in DPBS at RT for 5 min. Cells were washed once in DPBS and incubated with the primary antibodies ON at 4°C in DPBS. After ON incubation, samples were washed three times in DPBS and incubated with secondary antibodies in DPBS for 2 h at RT. After two washes in DPBS, samples were incubated with NucBlue (Thermo Fisher Scientific, #R37605, dilution 1:1,000) in DPBS for 10 min at RT and washed once in deionized water.

The primary antibody used for axon length analysis is Anti-β-Tubulin III (TUBBIII, Sigma-Aldrich, #T8578, dilution 1:500). The primary antibodies used for MT stability analysis are Anti-acetylated α-tubulin (Sigma-Aldrich, #T7451, dilution 1:400) and Anti-tyrosinated α-tubulin [YL1/2] (Abcam, #ab6160, dilution 1:400). The secondary antibodies used are Thermo Fisher Scientific #A11029, #R6393, and #A11077 (dilution 1:1000).

Coverslips were mounted onto glass microscope slides (Epredia, #AA00000102E01MNZ10) using Aqua-Poly/Mount (Polysciences, #18606-20) and left to dry ON at RT protected from light.

In HNs, cell imaging for analysis of axon and dendrite length and pyknotic nuclei was performed at 10X magnification using an inverted fluorescence microscope (Nikon Eclipse Ti) equipped with Nikon DS-Ri2 camera and the help of NIS-Elements AR software version 5.11; cell imaging for analysis of MT stability was performed at 60X magnification with a laser scanning confocal microscope (Nikon AX) and the help of NIS-Elements software version 5.42.

In iNs, cell imaging for analysis of neural processes length and pyknotic nuclei was performed at 20X magnification using a fluorescence microscope (Zeiss Axioscope 5) and the ZenBlue Pro imaging software version 2.5; cell imaging for analysis of MT stability was performed using the same microscope and software, at 40× magnification.

Images were analyzed with Fiji software version 1.54i.

Length of TUBBIII-positive axons and dendrites was measured, and the number of dendrites per cell was counted manually using the NeuronJ plugin. A total of 30 non-interconnected axons per replicate were traced using the tracing tool, and their length data were collected. A total of 30 cells per replicate were used for dendrites analysis, and the average length and number of dendrites per cell data were collected.

For oligomer neurotoxicity testing, cell death was calculated as the percentage of pyknotic nuclei per replicate, analyzing at least 2,000 total nuclei or six pictures per replicate. The status (pyknotic/non-pyknotic) of nuclei was evaluated by the operator considering the level of brightness and compactness. Pyknotic and total nuclei in a picture were counted with Fiji’s CellCounter tool.

For MT stability analysis, the intensities of fluorescent signals of acetylated and tyrosinated α-tubulin were quantified as mean fluorescence ( f ¯ ). Mean fluorescence ( f ¯ ) of a specific channel in a specific region of interest (ROI) was calculated as follows: f ¯ = IntDen − ( f ¯ back ⋅ A ) A where IntDen is the integrated density (the sum of the intensities of all the pixels in the ROI), f ¯ back is the mean fluorescence of background readings, and A is the area of the ROI. All these parameters are measured by Fiji software. As a ROI, non-interconnected axons, dendrites, or somas were selected. For background readings, three circular ROIs were selected in close proximity to the analyzed axons/dendrites/somas and their fluorescence values were averaged. At least 12 cells per replicate were analyzed. Stability was calculated as the ratio between acetylated versus tyrosinated α-tubulin mean fluorescence.

For elongation and MT stability, a priori power analysis for required sample size computation was performed with GPower 3.1 software (input parameters: α = 0.05, β = 0.20, two-tailed hypothesis, effect size calculated from pilot study data). Data were plotted, and statistical analysis was performed using GraphPad Prism software, version 8.0.2. Values are reported as mean ± standard error of the mean (SEM) from four replicates. Data distribution was checked through different normality tests: Anderson-Darling, D’Agostino & Pearson, Shapiro–Wilk, and Kolmogorov–Smirnov. Specifically, when data did not show a Gaussian distribution, Mann–Whitney test was used for comparing two groups and Kruskal–Wallis test followed by Dunn’s multiple comparisons test was used for comparing three groups. When data showed a Gaussian distribution, t-test was used for comparing two groups and one-way analysis of variance (ANOVA) followed by Holm-Sidak’s multiple comparisons test was used when comparing three groups. For the comparison of the frequency distribution of dendrites number per cell, chi-squared test was applied. p-values of ≤ 0.05 were considered statistically significant.

The positive control (nano-pulling) is not plotted, but it is reported for the dataset related to iNs in Supplementary Figure S2.

Although various forms of Aβ and Tau aggregates have been shown to exhibit some level of toxicity, oligomers are currently considered to be the most harmful (40, 41). For this reason, we used oligomeric species of Tau and Aβ as stressors in our models. For Tau oligomerization, we modified the protocol from Caneus et al. (41), optimizing the concentration, 50 nM, that is able to induce a 30% shortening of axon length in mouse HNs (Supplementary Figure S1). For Aβ oligomerization, we used an oligomeric preparation of a 7:3 ratio of Aβ40/Aβ42 at 1 μM concentration since we had previously reported that it vigorously induces hallmarks of AD (17). As this oligomers mix shows reversion of toxicity after 24 h (17), treatment was repeated for 2 consecutive days.

HNs were treated with 50 nM oTau at DIV1, and analysis was performed at DIV3 (Figure 1A). To evaluate the detrimental effect of the neurotoxic oligomers, we measured the axonal length, already used in literature as an indicator of neuron functionality (42). A statistically significant delay of outgrowth of axons in the treated samples compared to control samples was observed. The axon length reduction obtained by administering 50 nM oTau to mouse HNs is 28.6% [average axon length was 127.8 ± 5.8 μm in controls and 91.3 ± 3.6 μm in oTau-treated samples, p < 0.0001 (Figure 1B)]. The average percentage of cell death was 5.3 ± 0.14% in controls and 5.4 ± 0.14% in treated samples (p = 0.32, Figure 1C).

iNs were treated with 1 μM oAβ at DIV4, and analysis was performed at DIV6 (Figure 2A). Similarly to the previous model, a reduction in the length of the neural processes was obtained. The reduction obtained was 24.6% [the average length of neural processes was 155.3 ± 8.2 μm in controls and 113.4 ± 5.3 μm in oAβ-treated samples, p = 0.0002 (Figure 2B)], and the average percentage of cell death was 8.7 ± 0.4% in controls and 8.9 ± 0.4% in treated samples (p = 0.65, Figure 2C).

These results show that our treatments impact process elongation but not cell viability, mimicking the first stage of pathology.

To assess the neuroprotective effect of mechanical stimulation, three experimental groups were established: a control group that provides the baseline and reference values of the analyzed parameters, a group treated with oligomers from which we expect a decline in these values, and a group treated with oligomers and subjected to mechanical stimulation, from which we expect a recovery of the control values if the nano-pulling exerts a neuroprotective effect.

For HNs, MNPs were applied 4 h after cell isolation and seeding, oligomer treatment was performed at DIV1, and the mechanical stimulation was performed from DIV1 to DIV3 (Figure 3A). The analysis of axon length at DIV3 showed that the reduction in the axon length due to oTau treatment was recovered when the cells were subjected to nano-pulling (Figures 3B1, 3B2, 3B3). Specifically, the average axon length in the control group was 134.6 ± 5.4 μm, in the pathological group was 92.3 ± 3.6 μm, and in the pathological group subjected to mechanical stimulation was 133.8 ± 4.9 μm (Figure 3B4) (p < 0.0001 between the pathological and control groups and p = 0.18 between the pathological group subjected to nano-pulling and the control group).

Given that Tau pathology primarily affects dendrites, we also evaluated the effects of oTau and nano-pulling in this compartment. We found that dendrite length is not influenced by oTau nor oTau in combination with nano-pulling (specifically, the average dendrite length in the control group was 22.93 ± 0.74 μm, in the pathological group was 25.65 ± 0.90 μm, and in the pathological group subjected to mechanical stimulation was 22.91 ± 0.85 μm, p = 0.1 between the pathological and control groups, and p = 1 between the pathological group subjected to nano-pulling and the control group) (Figure 3B5). However, the study of the frequency distribution of the dendrite number per cell shows a reduction by oTau treatment with respect to the control (p = 0.005) that is compensated when cells are simultaneously subjected to nano-pulling (p = 0.39 versus control) (Figure 3B6).

For iNs, MNPs were applied 4 h after the beginning of terminal differentiation (DIV3), oligomer treatment was performed at DIV4, and the mechanical stimulation was performed from DIV4 to DIV6 (Figure 4A). A similar effect was observed on the neural processes of iNs: The average length in the control group was 147.4 ± 7.6 μm, in the pathological group was 113.7 ± 4.9 μm, and in the pathological group subjected to mechanical stimulation was 175.9 ± 9.9 μm (Figure 4B) (p = 0.009 between the pathological and control groups and p = 0.18 between the pathological group subjected to nano-pulling and the control group).

These data show that mechanical stimulation with piconewton forces prevents axonal shortening induced by oTau and oAβ, restoring outgrowth to normal physiological levels.

Using the same experimental design, to assess whether the neuroprotective effect of mechanical stimulation is mediated by force-induced MT stabilization, we evaluated the stability of axonal MTs in the two models. Cells were co-immunostained for acetylated and tyrosinated α-tubulin, routinely used markers of stable and dynamic MTs, respectively (43–45), and the ratio between the two fluorescent signals was used as a measure of MT stability (Figure 3C). Analysis was performed at DIV3 for HNs and at DIV6 for iNs (Figure 4C).

Data analysis showed that in HNs treated with oTau in the absence of mechanical stimulation, there is a statistically significant destabilization of MTs in the axons compared to the control group (2.02 ± 0.05 and 1.74 ± 0.06 for control and oTau, respectively, p = 0.0065), while in presence of mechanical stimulation, there is a statistically significant increase of the stability compared to the control group (2.02 ± 0.05 and 2.55 ± 0.1 for control and oTau+nano-pulling, respectively, p < 0.0001) (Figure 3D). Similarly, in iNs treated with the oAβ in absence of mechanical stimulation, there is a statistically significant destabilization of MTs in the neural projections compared to the control group (2.13 ± 0.11 and 0.98 ± 0.05 for control and oAβ, respectively), while in presence of mechanical stimulation, there is a statistically significant increase of the stability compared to the control group (2.13 ± 0.11 and 3.93 ± 0.21 for control and oAβ + nano-pulling, respectively, p < 0.0001) (Figure 4D). This result suggests that oTau and oAβ are able to induce MT destabilization at the axonal level and nano-pulling not only compensates this loss but also increases MT stability over physiological levels, acting as a neuroprotective factor.

We also evaluated whether the stabilizing effect propagates to MT in the cell soma, using the same experimental design. Data analysis showed that in cells treated with the neurotoxic oligomers in the absence of mechanical stimulation, there is a statistically significant destabilization of MTs in the somas compared to the control group. In particular, in the HNs model, the ratio between the mean fluorescence of acetylated and tyrosinated α-tubulin found in somas of controls was 1.50 ± 0.04, while in samples treated with 50 nM oTau it was 1.32 ± 0.04 (p = 0.0017) (Figure 3F). In the iNs model, the ratio was 2.05 ± 0.13 in controls and 1.06 ± 0.07 in samples treated with 1 μm oAβ (p < 0.0001). Mechanical stimulation was able to restore the soma MT stability levels of control condition in the HNs model (ratio 1.49 ± 0.04, p = 1 compared to the control group) (Figure 3F). In the soma of oAβ-treated iNs, nano-pulling was able to increase the level of MT stability beyond that of the control group (ratio 2.71 ± 0.17, p = 0.02 compared to the control group) (Figure 4E). This result suggests that oTau and oAβ also induce MT destabilization at the soma level and nano-pulling is able to compensate for induced destabilization, restoring or overcompensating MT stability levels in soma, depending on the cell model.

Similarly in dendrites of HNs, MT stability is reduced by oTau treatment and compensated in the group simultaneously subjected to nano-pulling: the ratio between the mean fluorescence of acetylated and tyrosinated α-tubulin found in dendrites of controls was 1.68 ± 0.06, in samples treated with 50 nM oTau was 1.42 ± 0.05 (p = 0.0029 versus control), and in treated samples simultaneously subjected to nano-pulling was 1.68 ± 0.48 (p = 1 versus control) (Figure 3E). These results suggest that oTau also induce damages at the dendrites level (consisting in reduction in number and MT destabilization) and nano-pulling is able to compensate, restoring dendrites number and MT stability to a physiological level.