Magnetically guided neurite outgrowth modulated by PIEZO1 in hiPSC-derived retinal ganglion cells

Abstract

Retinal ganglion cell (RGC) axon degeneration is one of the major causes of vision loss in glaucoma, where regeneration remains limited by the central nervous system's low intrinsic repair capacity. We applied a magnetic-guided nanotechnology approach to modulate axon growth in human-induced pluripotent stem cell-derived RGCs (hiPSC-RGCs). Magnetic Neural Axon Vectors (MagNAVs)—multi granular Fe₃O₄ nanoparticles functionalized with rhodamine—were efficiently internalized and directed by an external magnetic field (MF), enhancing neurite elongation and alignment. The mechanosensitive ion channel PIEZO1 was expressed in both soma and neurites. Specifically, PIEZO1 inhibition by Dooku1 enhanced neurite elongation, and activation by Yoda1 elevated somatic calcium transients without affecting neurite outgrowth. Furthermore, two-photon calcium imaging revealed that somatic calcium activity was PIEZO1-dependent, whereas neuritic calcium signaling was more responsive to magnetic force. These results indicate that PIEZO1-mediated calcium influx may negatively regulate force-induced neurite elongation. This work demonstrates a nanomaterial-based strategy for targeted axon regeneration and highlights the mechanoregulatory role of PIEZO1.

Graphical abstract

MagNAVs were internalized by hiPSC-derived RGCs to investigate mechanosensitive regulation of neurite outgrowth. Magnetic field stimulation aligned intracellular MagNAVs to promote directional neurite extension. PIEZO1 activity modulated compartment-specific Ca²⁺ transients and negatively regulated neurite elongation, with its inhibition enhancing axonal outgrowth under magnetic guidance.

Highlights

• •Magnetic nanoparticles guide directional neurite growth in hiPSC-derived RGCs.
• •PIEZO1 inhibition enhances neurite outgrowth under magnetic stimulation.
• •MagNAVs + magnetic force synergistically promote axonal elongation.
• •Calcium imaging reveals compartment-specific PIEZO1 mechanosensitivity.
• •PIEZO1-mediated Ca²⁺ influx negatively regulates neurite extension.

1. Introduction

Retinal ganglion cells (RGCs) are located in the innermost layer of the retina and are crucial in transmitting visual information to the brain. Upon detection of visual stimuli by the eyes, light activates rod and cone photoreceptors, which relay signals to bipolar cells and transmit them to RGCs. The axons of RGCs converge to form the optic nerves, which serve as the pathway for transmitting visual information to the lateral geniculate nucleus and, finally, the visual cortex in the brain [1].

Because the optic nerve is part of the central nervous system, it has a limited capacity for regeneration following injury. Consequently, diseases that cause RGCs to become necrotic or dysfunctional—such as glaucoma [2], hereditary optic neuropathies [3], ischemic optic neuropathies [4], and demyelinating conditions [5] —can severely impair visual signal transmission. Therefore, developing strategies to enhance RGC survival and promote axonal regeneration is critical for mitigating the vision loss associated with glaucoma and other related disorders [6]. Despite difficulties, evidence indicates that RGC regeneration may be feasible, suggesting new avenues for advanced therapeutic approaches [7]. Microtubules and neurofilaments influence neurite outgrowth. Stabilization of microtubule enhances the capacity of axon regeneration in spinal cord repair [8]. In the signaling pathway, promoting mTOR and JAK/STAT signaling facilitates RGC survival and regeneration [9]. Adding calcium channel inhibitors to block acute axonal degeneration also promotes axonal regeneration [10]. Moreover, the differentiation of embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs) have emerged as novel strategies for regenerating RGCs. Evidence suggests that it is possible to transplant a derived retina into a degenerative mouse model to restore vision [11]. In addition, transplanted cells can secrete repair-related factors to influence damaged cells, thereby achieving therapeutic effects [12]. Thus, structures differentiated from iPSCs or ESCs not only can serve as sources for replacing damaged cells or tissues but also can act as healthy cells to influence the damaged cells. With fewer ethical issues and the potential to access a variety of retinal cells, hiPSC-derived retinal organoids have become a popular tool for addressing ophthalmic research problems. Specifically, hiPSC-derived retinal organoids can enhance our understanding of retinal development and facilitate the construction of comprehensive in vitro disease models. This provides researchers a versatile platform to study various aspects of eye diseases [13].

The first problem encountered in transplanting neural cells is that the axons need to be guided to the correct area. Therefore, it is necessary to have a deeper understanding of hiPSC-RGC axon growth. In these in vitro neuronal systems, growth cones—composed of actin filaments and microtubules—are highly motile structures at the tips of growing axons and dendrites [14]. They navigate the direction of neurite outgrowth by sensing and responding to environmental cues [15]. These cues include soluble substances, contact-mediated signals, and cell surface molecules. They can attract or repel the growth cones, leading to intracellular signaling and axon growth [16]. In addition to chemical factors influencing the direction of growth cone advancement and axon growth, numerous studies have found that mechanical forces are another important factor driving growth cone extension and neurite outgrowth [17,18]. After growth cones interact with these external molecules, actin filaments of growth cones bind to these molecules in the surrounding environment, generate forces that promote actin polymerization, and push the growth cones and cell membranes forward. At the same time, myosin motors on the microtubules generate traction forces that extend the growth cones toward them [19]. As the cell membranes extend outward due to actin polymerization, the myosin motors simultaneously exert a force pulling the growth cones back. Specifically, myosin II on the microtubules generates a pulling force that draws the growth cones toward the central zone. During neuronal growth, the interplay between these opposing forces regulates the overall activity and direction of neurite outgrowth [14].

Mechanical forces are integral to various stages of nervous system development, including axon extension and neuronal migration [20], and can influence individual neuronal growth by altering external mechanical environments [21]. Recent advances have leveraged magnetic nanoparticles (MNPs) and magnetic fields (MFs) as non-invasive methods to generate nano-scale pulling forces on cells, thereby promoting neurite elongation [22], modulating axonal endosome transport [23] and enhancing stem cell differentiation toward neural lineages [24]. For instance, internalizing superparamagnetic iron oxide particles in spiral ganglion neurons stimulates neurite outgrowth. It aligns axonal extension under external MFs [25] while applying these nanoparticles directly to RGC filopodia facilitates filopodia extension [26]. However, whether magnetically mediated mechanical forces can similarly drive axonal outgrowth in iPSC-derived RGCs remains unknown, representing a critical gap in our understanding of retinal regeneration.

Meanwhile, mechanosensitive ion channels such as PIEZO1 and PIEZO2 have emerged as key regulators of neuronal responses to tension, compression, and shear [27]. Although PIEZO2 is predominantly expressed in sensory tissues, PIEZO1 is more abundant in non-sensory regions [28], and both channels are present in various ocular structures—including the optic nerve head, trabecular meshwork and RGCs—where their expression is upregulated under elevated intraocular pressure [29]. Moreover, activating PIEZO1 via the agonist Yoda1 can reduce intraocular pressure in mice [30] but paradoxically suppresses RGC neurite outgrowth, whereas inhibition of mechanosensitive channels with GsMTx4 promotes neurite extension [31]. The extent to which inhibiting PIEZO1 alone can enhance neurite outgrowth remains undetermined.

Building on these observations, the present study employed hiPSC-derived RGCs to examine whether the internalization of MagNAVs (Magnetic Neural Axon Vectors) coupled with magnetic field application affects neurite outgrowth, and to explore whether PIEZO1 mediates these mechanically induced changes. Combining magnetic nanotechnology with mechanistic insights into the PIEZO1 function addresses a key knowledge gap and advances the potential for harnessing mechanical forces to promote RGC regeneration.

2. Materials and methods

2.1. Synthesis of Magnetic Neural Axon Vectors (MagNAVs)

MagNAVs were synthesized in the Department of Biomedical Engineering and Environmental Sciences at National Tsing Hua University. The detailed synthesis procedure was described in Huynh et al. [32]. Briefly, MagNAVs were prepared by dissolving ferric chloride hexahydrate (FeCl₃·6H₂O, 20 mmol, Sigma-Aldrich) and 0.5 mL of oleylamine (OA, Sigma-Aldrich) in 40 mL of ethylene glycol (EG, anhydrous, 99 %, Sigma-Aldrich) with gentle stirring at room temperature. Sodium acetate (NaAc, 40 mmol, Sigma-Aldrich) was then added to the mixture, which was stirred until a uniform brown solution was formed. This solution was stirred for an additional hour and subsequently transferred into a Teflon-lined stainless-steel autoclave for a hydrothermal reaction at 200–240 °C for 6 h. After the reaction, the mixture was cooled to room temperature, and the precipitate was collected by centrifugation at 3500 rpm for 10 min in ethanol. The collected MagNAVs were then washed three times with excess ethanol and deionized water to remove unreacted reagents and stored in ethanol at 4 °C until further use.

2.2. Characterization of MagNAVs

The morphology of MagNAVs was characterized by transmission electron microscopy (TEM; Hitachi H-7100, Japan) operated at 75 kV. Particle size distribution was analyzed from TEM images using ImageJ software (NIH, USA). The hydrodynamic diameter of MagNAVs in PBS was measured using dynamic light scattering (DLS; Zetasizer Nano ZS90, Malvern Instruments, UK; n = 3).

To confirm cellular internalization of MagNAVs, hiPSC-derived RGCs were fixed with 2.5 % glutaraldehyde in 0.1 M Sorenson's phosphate buffer, post-fixed in 1 % osmium tetroxide, dehydrated, embedded in resin, and sectioned for ultrastructural analysis by TEM. Internalized MagNAVs were observed within the cytoplasm and neurites of RGCs.

2.3. Maintenance and differentiation of hiPSCs into retinal organoids

hiPSCs were cultured at 37 °C with 10 % CO₂ in Geltrex-coated dishes using StemFlex Medium (Gibco), refreshed daily. Cells were passaged at 85 % confluence using TrypLE Express (Gibco), then reseeded in fresh Geltrex-coated dishes. For retinal organoid differentiation, a modified protocol was followed [33]. hiPSCs (1.5 × 10⁵ cells/mL) were seeded in Petri dishes with a 3:1 mix of StemFlex and Neuron Induction Medium (NIM), supplemented with ROCK inhibitor (Merck). NIM consisted of Advanced DMEM/F12, heparin (2 μg/mL), N2 supplement and non-essential amino acids (Gibco). The medium was gradually transitioned to 100 % NIM by day 6. On day 7, embryoid bodies were transferred to NIM +10 % FBS (NQBB) for attachment. From day 16, Retinal Differentiation Medium (RDM; DMEM/F12 3:1, B27 supplement without vitamin A, NEAA, and Pen/Strep) was used to promote optic vesicle formation. Vesicles were gently detached from day 20 for further culture.

2.4. Cultured retinal ganglion cells and drug treatments

To obtain isolated RGCs, optic vesicles were dissociated into single cells using both chemical and mechanical methods [34]. Initially, optic vesicles were collected into a 15 mL centrifuge tube and allowed to settle by gravity for a few minutes. The medium was then removed, and the vesicles were rinsed once with 500 μl of TrypLE. Subsequently, 1 mL of TrypLE was added, and the tube was incubated at 37 °C for 6 min. At each 2–3 min, the tube was gently shaken to prevent clumping. After 6 min, the tube was flicked, and gentle pipetting was initiated. Following this step, 500 μl of the culture medium was added, and gentle pipetting was repeated. When the optic vesicle structures were no longer visible and became the liquid form, the medium was filtered using round-bottom tubes with a cell strainer cap (352235; Falcon). The filtrate was transferred to a new tube and centrifuged for 5 min. After centrifugation, the supernatant was removed, and the cells were resuspended in 1 mL of the medium. Cells were counted and seeded into a 24-well plate coated with 0.0001 % poly-L-ornithine solution (27378-49-0; Merk) and 1.4 μg/mL laminin (354232; Corning) at a concentration of 4 × 10⁴ cells/well. After incubating for 24 h, the hiPSC-RGCs were exposed to the medium containing 10 μM Yoda1 (HY-18723; Med Chem Express) or 10 μM Dooku1 (HY-126010; Med Chem Express). Half of the medium was replaced on day 2, and the cells were fixed after 4 days in culture.

2.5. Magnetic field (MF)

The MF were generated by eight magnets, each with a strength of 2292 ± 58.7 Gauss. On the day following cell seeding in the 24-well plate, eight magnets were placed at the four corners of the short sides of the plate, with two magnets at each corner. The intensity of the MF was measured with a Tesla meter (WT108, Weite Magnetic, China).

2.6. Cell viability evaluation

The effective concentrations of Yoda1 and Dooku1 were determined using the Cell Counting Kit-8 (CCK8) assay (CK04; Dojindo Laboratories). In this assay, hiPSC-RGCs were seeded in a 96-well plate coated with poly-L-ornithine solution and laminin at a concentration of 4 × 10⁴ cells/well. After 24 h of incubation, the cells were exposed to different concentrations of Yoda1 and Dooku1 (1, 5, 10, 20, and 40 μM) for 24 h. A mixture of 10 % CCK8 and DMEM/F12 medium was added to each well and incubated at 37 °C for 4 h. Cell viability was evaluated using a microplate reader (Varioskan Flash; Thermo Fisher Scientific) at an absorbance of 450 nm.

2.7. Immunocytochemistry

After culturing for 4 days, RGCs were gently washed three times with 1xPBS and fixed with 4 % paraformaldehyde for 30 min at room temperature. The cells were rinsed with 0.1 % Triton-X 100 (9002-93-1; BioShop) in 1x PBS for 10 min and blocked with 1 % blocking buffer (W-3400; Goalbio) dissolved in 0.1 % Triton-X 100 in 1x PBS for another 10 min. The cells were then incubated with primary antibodies in the blocking buffer overnight at 4 °C. The primary antibodies included β-Ⅲ-tubulin (TUJ1; 1:1000; sc-80005; Santa Cruz Biotechnology), ATOH7 (1:300; A7273; ABclonal), PIEZO1 (1:300; 15939-1-AP; Proteintech), and RBPMS (1:200; A15395; ABclonal). On the following day, the cells were rinsed three times for 5 min each with 0.1 % Triton-X 100/1x PBS, followed by incubation with corresponding secondary antibodies (1:1000; A11001 or A11012; Invitrogen, U.S.A.) for an hour in room temperature. DAPI was then added to the cells for 20 min and the cells were rinsed three times for 5 min each with 0.1 % Triton-X 100/1x PBS. Finally, the cells on the coverslips were mounted onto slides and visualized by using confocal microscopy (Stellaris 8 confocal microscope, Leica).

2.8. Knockdown of PIEZO1 in hiPSC-RGCs

hiPSC-RGC cells were transfected with a lentiviral vector expressing a shRNA targeting human PIEZO1 (sc-93227-V, Santa Cruz Biotechnology). A scrambled shRNA lentivirus (sc-108080, Santa Cruz Biotechnology) served as a control. Cells were infected at a multiplicity of infection (MOI) of 5 for 24 h and then placed in a magnetic field for neurite growth. Knockdown efficiency was assessed by Western blotting using an anti-PIEZO1 antibody (1:1000; 15939-1-AP; Proteintech).

2.9. Flow cytometric analysis of apoptosis

Apoptosis was evaluated using a PE Annexin V/7-AAD assay kit (BD Pharmingen). Retinal organoids were dissociated into single-cell suspensions using TrypLE digestion for 6 min with gentle pipetting. After enzymatic dissociation, cells were centrifuged at 1000 rpm for 5 min and resuspended in 0.5 mL of binding buffer containing 7-AAD. The final concentration was adjusted to approximately 1 × 10⁵ cells per sample. Staining with Annexin V-PE and 7-AAD was performed according to the kit protocol. Compensation controls included unstained, Annexin V–only, and 7-AAD–only samples. Flow cytometry was carried out on an Attune™ NxT system (Thermo Fisher Scientific) using a standard two-color detection protocol, and at least 1 × 10⁴ gated events were acquired for each sample.

2.10. Neurite outgrowth measurement and quantification

To calculate the average neurite length, images of hiPSC-RGCs were acquired using confocal microscopy (Stellaris 8 confocal microscope, Leica) with a 20 × objective lens. The confocal images were color-separated: green for TUJ1, representing the neurites, and blue for DAPI, representing the nuclei. Image analysis was performed using the ImageJ plugin AutoNeuriteJ to automatically measure the neurite length. The average neurite outgrowth per cell was calculated using the following formula:

$$Neuritesoutgrowthlengthpercell\left(\mu m\right)=\frac{sumofneuriteslength\left(\mu m\right)}{cellnumbers}$$

Fifty cells were randomly selected from each well for quantification to ensure statistical reliability.

2.11. Quantification of neurite outgrowth direction

To quantify the direction of neurite outgrowth, we utilized the ImageJ plugin Directionality. Images with the longest average neurite lengths were selected and analyzed using this plugin to determine the predominant direction of neurite outgrowth across the entire image. This derived direction was then compared to the known magnetic field direction to assess the deviation of neurite outgrowth. The deviation direction was calculated using the following formula:

Deviation direction from magnetic field

$$(°)=\left|directionofneuriteoutgrowth(°)-directionofmagneticfield(°)\right|$$

2.12. Two-photon microscopic calcium imaging

For calcium imaging, hiPSC-RGC cultures were first loaded with the fluorescent calcium indicator Oregon Green™ BAPTA-1 (OGB-1). A 5 mM stock solution of OGB-1 was prepared in DMSO, and prior to imaging, it was diluted 1:100 in retinal differentiation medium (RDM) to reach a final concentration of 50 μM. The cells were then incubated at 37 °C for 1 h to allow dye uptake, after which the medium was replaced with fresh RDM to remove excess dye.

Two-photon microscopy was performed using a custom-designed system (Southport Co.) equipped with a movable objective and a 16 × , 0.8 NA Nikon N16×LWD-PF objective (Nikon Instruments, America). Excitation was provided by a 920 nm ultrafast pulsed laser (Mradian Femto sources co.) at 30–40 mW to excite the calcium indicator. Emission signals were collected by two photomultiplier tubes (PMTs; H16201P-40, Hamamatsu): one for 505–545 nm emission (channel 1) and another for 580–670 nm emission (channel 2). Image acquisition was carried out with a resolution of 512 × 512 pixels (field of view: 358.4 × 358.4 μm) at a rate of 23 frames per second, using an input range of ±0.4 mV as recorded by custom-made software (Southport Co.).

2.13. Statistical analysis

Statistical analyses of neurite outgrowth and directional deviation were conducted using GraphPad Prism version 6.0.0 (GraphPad Software, RRID: SCR_002798). Depending on the experimental design, either two-way ANOVA or one-way ANOVA followed by planned two-tailed unpaired Student's t-tests was used to compare specific groups. To evaluate the independent effects of PIEZO1 modulation and magnetic stimulation on neurite outgrowth, independent sample t-tests and Levene's tests were performed using SPSS Statistics for Windows version 20.0 (SPSS Inc., Chicago, IL, USA). The normality of neurite length distribution was assessed using the Kolmogorov–Smirnov test. All data are presented as mean ± standard error of the mean (SEM).

3. Results

3.1. Characterization of MagNAV nanoparticles

The synthesis of Magnetic Neural Axon Vectors (MagNAVs) is illustrated in Fig. 1A. Raspberry-like iron oxide particles (I.RB) were prepared via a hydrothermal reaction using oleylamine (OA) and sodium citrate as coordinating agents. These particles were subsequently conjugated with rhodamine isothiocyanate (RITC) to generate fluorescent MagNAVs. Dynamic light scattering (DLS) analysis confirmed that the nanoparticles had an average hydrodynamic diameter of approximately 201 nm (Fig. 1B), with a narrow size distribution as shown in the corresponding histogram (Fig. 1C). Transmission electron microscopy (TEM) revealed that the synthesized MagNAVs consisted of clustered iron oxide crystallites forming multigranular spherical or irregular structures ranging from 90 to 250 nm in diameter (Fig. 1D). The high-resolution contrast image also indicated porous domains between the clustered grains, consistent with hydrothermally assembled iron oxide nanostructures. Scanning electron microscopy (SEM) further confirmed the raspberry-like surface morphology and aggregation characteristics of the MagNAV clusters (Fig. 1E). Fluorescence microscopy confirmed the successful labeling of MagNAVs with RITC, allowing for visualization in cell-based assays (Fig. 1F).

Fig. 1.

Characterization of MagNAVs. (A) Schematic illustration of the synthesis of I.RB via a hydrothermal reaction using OA and sodium citrate, followed by conjugation with RITC to produce fluorescent MagNAVs. (B) DLS analysis showing a hydrodynamic diameter. (C) Histogram indicating the particle size distribution of MagNAVs. (D) TEM image displaying multigranular spherical or irregular MagNAV structures composed of clustered iron oxide crystallites (scale bar = 500 nm). (E) SEM image revealing the raspberry-like morphology and aggregated surface texture of MagNAVs (scale bar = 500 nm). (F) Fluorescence microscopy image confirming successful RITC labeling of MagNAVs for subsequent cellular visualization (scale bar = 50 μm).

3.2. Differentiation and characterization of hiPSC-Derived retinal ganglion cells

Using a chemical-based differentiation protocol [34,35], hiPSCs were induced to form retinal ganglion cells through three key stages: embryoid bodies, neural rosettes, and optic vesicles (Fig. 2A). During the first seven days, hiPSCs underwent suspension culture and progressively formed spheroid embryoid bodies containing ectodermal, mesodermal, and endodermal cells, analogous to the three germ layers in early embryonic development [36]. On day 7, the EBs were transferred to adherent culture and maintained in neuronal induction medium. By day 16, they differentiated into neural rosettes, characterized by radially arranged neural progenitor cells forming circular structures (Fig. 2A, D0–D16). Subsequently, the medium was switched to retinal differentiation medium to induce optic vesicle (OV) formation. On day 20, the developing OVs were transferred back to suspension culture, where they initially appeared flattened but transformed into three-dimensional spherical structures within a day (Fig. 2A, days 20–22), giving rise to RGCs. On day 24, the OVs were dissociated using TrypLE and plated onto dishes coated with poly-D-lysine and laminin to generate single-cell RGC cultures (Fig. 2A, day 24).

Fig. 2.

Differentiation and characterization of hiPSC-derived RGCs. (A) Schematic timeline of the hiPSC-to-RGC differentiation protocol, which progresses through embryoid body (EB) formation (D7), neural rosette development (D16), and optic vesicle (OV) formation (D22). On day 24, optic vesicles were dissociated to isolate and culture individual RGCs. (B) Immunocytochemistry at day 16 shows the expression of the neuronal marker TUJ1 (green) and the early RGC marker ATOH7 (red) within neural rosettes. Nuclei are counterstained with DAPI (blue). Scale bar = 50 μm. (C) Four days after dissociation and single-cell culture, co-expression of TUJ1 (green) and strong ATOH7 (red) confirms early RGC differentiation. Nuclei are stained with DAPI (blue). Scale bar = 50 μm. (D) After an additional 10 days of culture, mature RGCs express both the neuronal marker TUJ1 (green) and the RGC-specific marker RBPMS (red). Nuclei are stained with DAPI (blue). Scale bar = 50 μm.

To confirm RGC identity and maturation, immunofluorescence staining was performed for the pan-neuronal marker TUJ1, the early RGC marker ATOH7, and the mature RGC marker RBPMS [37,38]. At day 16, TUJ1 and ATOH7 expression was detected in neural rosettes (Fig. 2B), indicating early RGC differentiation. On day 24, ATOH7 remained strongly expressed in isolated single cells (Fig. 2C), and after 10 days of culture, robust RBPMS expression confirmed RGC maturation (Fig. 2D). The sequential expression of TUJ1, ATOH7, and RBPMS validated the progressive and successful differentiation of hiPSC-derived RGCs within 20 days. Quantification revealed that ATOH7⁺ cells accounted for 93.0 ± 2.93 % of all neurons (Fig. 2C), while RBPMS⁺ cells accounted for 83.8 ± 3.6 % (Fig. 2D). TUJ1 was expressed in nearly all neurons, validating its use as a pan-neuronal marker.

3.3. Cellular uptake and intracellular localization of MagNAVs in hiPSC-Derived RGCs

Before being applied to cells, MagNAVs were thoroughly washed with PBS and RDM, followed by sonication for at least 5 min to ensure even dispersion. On day 17 of RGC differentiation, MagNAVs were added to the neural rosette cultures at a final concentration of 0.3 mg/mL, appearing as black particulate structures under brightfield microscopy (Fig. 3A). After three days of co-culture, the adherent optic vesicle structures were dissociated and transferred to suspension culture to continue RGC differentiation. TEM (Fig. 3B) and confocal imaging (Fig. 3C) confirmed the successful internalization of MagNAVs into hiPSC-derived RGCs. To ensure that the particles were specifically taken up by RGCs, co-localization between the RITC-loaded MagNAV signal and the neurite-specific marker TUJ1 was examined. Flow cytometry was performed to assess MagNAV uptake (Fig. 3D). Quantitative analysis revealed that approximately 88.83 ± 0.15 % of hiPSC-RGCs internalized MagNAVs, compared with 0.91 ± 0.07 % in untreated controls (Fig. 3E). Taken together, these results demonstrate that MagNAVs are efficiently internalized by RGCs, localize within neuronal compartments, and extend along neurites.

Fig. 3.

Cellular uptake and intracellular localization of MagNAVs in hiPSC- RGCs. (A) Brightfield image showing black, particulate MagNAVs (0.3 mg/mL) dispersed in neural rosette cultures on day 17 of RGC differentiation. (B) TEM image revealing internalized MagNAVs within the cytoplasm of hiPSC-derived RGCs. Scale bar = 200 nm. (C) Confocal fluorescence image demonstrating successful cellular uptake of rhodamine-labeled MagNAVs (RITC; red) in RGCs, with co-staining of neurites by the neuronal marker TUJ1 (green) and nuclei by DAPI (blue). Scale bar = 50 μm. (D) Flow cytometry histograms of RITC fluorescence in control (blue) and MagNAV-treated (orange) hiPSC-RGCs. (E) Quantification showing that about 88.83 ± 0.15 % of hiPSC-RGCs were RITC-positive after MagNAV exposure compared with 0.91 ± 0.07 % in controls.

Previous studies have reported that iron oxide nanoparticles may induce ferroptosis or lysosomal stress [39]. To further evaluate the safety of MagNAVs, hiPSC-RGCs were treated with MagNAVs at day 19 of differentiation and analyzed on day 26. TEM analysis revealed intact mitochondrial double membranes with well-preserved lamellar cristae, without swelling, fragmentation, or electron-lucent (arrows). Lysosomal structures also appeared normal, showing no signs of vacuolization or increased multilamellar bodies(arrowheads) (Fig. S1A). In addition, flow cytometry with Annexin V-PE/7AAD staining was performed to quantify apoptosis (Fig. S1B). The proportion of Annexin V⁺/7AAD⁻ cells did not differ significantly between MagNAV-treated and control groups (Control: 1.65 %; MagNAVs: 1.62 %; n.s.) (Fig. S1C). Together, these findings indicate that MagNAVs do not exert detectable cytotoxic or pro-apoptotic effects under the conditions used in our neurite outgrowth assays.

3.4. MagNAV-driven directional neurite outgrowth under magnetic field stimulation

To investigate whether hiPSC-derived RGCs internalizing MagNAVs could promote neurite outgrowth and facilitate directional extension under a guiding magnetic field, RGCs were cultured in a 24-well plate with permanent magnets (2292 ± 58.7 Gauss) affixed to the outer sides of the plate to establish a defined MF. As illustrated in Fig. 4A, magnets were positioned at both ends of rows A and D, generating a directional magnetic gradient across the plate. Wells adjacent to the magnets (A1, A6, D1, and D6) exhibited excessive aggregation of MagNAVs near the magnetic source, impairing uniform field distribution and confounding neurite analysis; thus, these wells were excluded from subsequent experiments (data not shown).

Fig. 4.

Effects of MagNAVs and MF on hiPSC-RGCs neurite outgrowth. (A) Schematic diagram of the experimental setup using a 24-well plate. Permanent magnets were positioned outside wells A1 and A6 (rows A and D) to generate a directional magnetic field (MF) across the plate. The left panel illustrates the magnetic field lines acting on MagNAVs within the wells. The right panel depicts the angular orientation (azimuth) of each well relative to the adjacent magnet, with 0° defined at the edge of the magnet (A1, A6, D1, and D6). Specifically, wells A1–A3 and B1–B3 are referenced to the magnet at A1; wells A4–A6 and B4–B6 are referenced to the magnet at A6; wells C1–C3 and D1–D3 are referenced to the magnet at D1; and wells C4–C6 and D4–D6 are referenced to the magnet at D6. The values in parentheses represent the magnetic field strength (in mT) as measured by a Tesla meter. (B) Confocal images of hiPSC-RGC neurite outgrowth under different conditions: control, MagNAV without MF, and MagNAV with MF. TUJ1 (green) stains RGCs, and DAPI (blue) stains nuclei. Scale bar, 50 μm. (C, D) Distribution histograms illustrate the directionality of neurite outgrowth under different experimental conditions. In the control group (C), RGCs did not internalize MagNAV and were not affected by MF. In the MagNAVs + MF group (D), RGCs internalized MagNAVs and were guided by MFs. (E) Neurite outgrowth deviates from the MF direction. MagNAVs + MF significantly affected the direction of neurite outgrowth (control group n = 15, MagNAVs group n = 8, MagNAVs + MF group n = 11). (F) Average neurite length of hiPSC-RGCs. MagNAVs + MF stimulated neurite elongation, whereas MagNAVs alone had no significant effect (control group n = 15, MagNAVs group n = 8, MagNAVs + MF group n = 11). ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; error bars represent mean ± SEM.

RGCs were cultured for four days under conditions with or without MF, and no apparent differences in cell morphology or viability were observed between groups (Fig. 4B). Directionality histograms were generated using the Directionality plugin in ImageJ to assess neurite orientation. In the absence of a magnetic field, neurites extended in random directions (Fig. 4C). In contrast, RGCs exposed to both MagNAVs and MF displayed a significantly more directed neurite growth pattern aligned with the magnetic field vector (Fig. 4D). Quantitative analysis of the deviation angle between neurite trajectories and the magnetic field direction revealed that the MagNAV + MF group exhibited markedly smaller deviation angles compared to both the control group and the MagNAV-only (no MF) group (Fig. 4E), indicating that internalized MagNAVs effectively mediated neurite guidance under MF stimulation.

Furthermore, measurement of neurite length demonstrated that the MagNAV + MF group exhibited significantly greater neurite elongation relative to the control (Fig. 4F). These findings collectively indicate that MagNAVs enhanced neurite outgrowth only in the presence of magnetic stimulation and further promote neurite alignment along the applied magnetic field direction.

3.5. PIEZO1 involvement in neurite outgrowth of hiPSC-RGCs

To evaluate the role of PIEZO1 in neurite elongation induced by MagNAVs in conjunction with magnetic field stimulation, we employed a pharmacological approach using the selective PIEZO1 agonist Yoda1 and antagonist Dooku1. Immunofluorescence staining confirmed the expression of PIEZO1 in hiPSC-derived RGCs, showing prominent localization in the cell bodies and proximal neurites (Fig. 5A). Before investigating the functional effects of PIEZO1 modulation, cell viability assays were conducted using CCK-8 to determine appropriate working concentrations of Yoda1 and Dooku1. A decline in cell viability was observed with Yoda1 treatment at concentrations exceeding 20 μM (Fig. 5B). At the same time, Dooku1 showed acceptable cytocompatibility up to 10 μM (Fig. 5C). Based on these findings, a concentration of 10 μM was selected for both compounds in all subsequent experiments. hiPSC-RGCs were then treated with Yoda1 or Dooku1 and cultured under standard conditions. No notable morphological abnormalities were observed, indicating that neither treatment adversely affected general cell health (Fig. 5D). Quantitative analysis of neurite length revealed that Dooku1 treatment significantly enhanced neurite extension, whereas Yoda1 showed no statistically significant difference compared to the control group (Fig. 5E). These results suggest that PIEZO1 may play a regulatory role in neurite outgrowth in hiPSC-derived RGCs.

Fig. 5.

Modulation of neurite elongation by PIEZO1 in MagNAV-loaded hiPSC-derived RGCs. (A) Immunofluorescence staining showing PIEZO1 (red) expression in TUJ1-positive RGCs (green), with nuclei labeled by DAPI (blue), confirming localization in both soma and proximal neurites. The enlarged image in the red box shows the localization of PIEZO1 in cell bodies and neurites. Scale bar = 50μm. (B, C) CCK-8 assay results showing the cytocompatibility of Yoda1 (B) and Dooku1 (C) at increasing concentrations. Yoda1 showed cytotoxicity above 20 μM, while Dooku1 maintained acceptable viability up to 10 μM. (D) Representative brightfield images of RGCs following treatment with 10 μM Yoda1 or Dooku1, showing no obvious morphological changes. (E) Quantification of neurite length under different treatment conditions. Dooku1 significantly enhanced neurite outgrowth, whereas Yoda1 had no significant effect relative to control. Data are presented as mean ± SEM. ∗p < 0.05.

3.6. Synergistic effects of PIEZO1 inhibition and magnetic force on neurite growth

To further investigate the interplay between magnetically induced mechanical forces and PIEZO1-mediated regulation of neurite outgrowth, hiPSC-derived RGCs internalizing MagNAVs were treated with either the PIEZO1 agonist Yoda1 or antagonist Dooku1 under magnetic field exposure (Fig. 6A).

Fig. 6.

Synergistic effects of PIEZO1 modulation and magnetic force on neurite elongation in MagNAV-loaded hiPSC-derived RGCs. (A) Experimental workflow showing hiPSC-RGCs preloaded with MagNAVs and subjected to treatment with PIEZO1 agonist (Yoda1) or antagonist (Dooku1), with or without exposure to an external magnetic field (MF). (B) Quantification of neurite length under different treatment combinations. Yoda1 alone did not significantly affect neurite growth compared to control. Yoda1 combined with MagNAV + MF increased neurite length, but this effect was comparable to MagNAV + MF alone. Dooku1 alone and MagNAV + MF alone both significantly enhanced neurite outgrowth. Notably, the combination of Dooku1 and MagNAV + MF resulted in the greatest neurite elongation, suggesting a synergistic interaction between PIEZO1 inhibition and magnetic stimulation. Data are shown as mean ± SEM. ∗p < 0.05, ∗∗p < 0.01. (C) Western blot validating PIEZO1 knockdown 48 h after transduction with PIEZO1 shRNA lentivirus versus non-targeting control shRNA; GAPDH as loading control. (D) Neurite length per cell (mean ± SEM; dots = biological replicates). Two-way ANOVA with Sidak correction; ∗p < 0.05.

Quantitative analysis of neurite length under various treatment conditions revealed that Yoda1 alone did not significantly alter neurite outgrowth compared to the control group (Fig. 5, Fig. 6B). Interestingly, when Yoda1 was combined with MagNAV + MF treatment, neurite length was significantly increased relative to the control group (Fig. 6B); however, this increase was comparable to that observed in the MagNAV + MF group without Yoda1. These results suggest that Yoda1 could not counteract the neurite elongation effects induced by magnetic force-mediated mechanical stimulation via MagNAVs. Both Dooku1 and MagNAV + MF treatments independently promoted neurite extension (Fig. 5, Fig. 6B). When Dooku1 was combined with MagNAV + MF, a further enhancement in neurite length was observed, surpassing the effects of either treatment alone. This additive response indicates a potential synergistic interaction between PIEZO1 inhibition and magnetically applied mechanical forces (Fig. 6B).

To validate the effects of Dooku1, we used lentiviral shRNA to knock down PIEZO1 expression in cells. Western blotting confirmed effective knockdown of PIEZO1 in hiPSC-RGCs (Fig. 6C). Following PIEZO1 knockdown, neurite length remained similar to that observed with scrambled shRNA, indicating that the virus did not affect neurite outgrowth. Furthermore, a significant increase in neurite length was also observed following PIEZO1 knockdown with co-treatment of MagNAVs + MF (Fig. 6D). Together, these findings suggest that PIEZO1 may function as a mechanosensitive negative regulator of neurite elongation in hiPSC-RGCs. Inhibiting PIEZO1 activity can enhance neurite growth, particularly when combined with external magnetic forces generated by MagNAVs.

3.7. PIEZO1 activity and magnetic force differentially regulate compartmentalized spontaneous calcium transients

To evaluate the effects of PIEZO1 activity and magnetically induced mechanical force on spontaneous calcium dynamics in hiPSC-derived RGCs, two-photon calcium imaging was performed for 5 min following treatment with PIEZO1 modulators (Yoda1 or Dooku1) and/or magnetic field (MF) stimulation in cells internalizing MagNAVs. Calcium fluctuations were quantified as ΔF/F and high-pass filtered to reduce baseline noise. Transient events were defined as peaks exceeding 20 % ΔF/F. Representative calcium traces from soma and neurites under each condition are shown in Fig. 7A and B, respectively. Spontaneous calcium transients were detected in both compartments across all treatment groups. When comparing paired soma and neurite recordings from the same cell, most RGCs exhibited higher peak amplitudes in neurites than in soma (Fig. 7C). However, some cells showed the opposite trend, highlighting variability in compartmental calcium dynamics at the single-cell level. The group analysis revealed a statistically significant increase in average calcium transient amplitude in neurites compared to soma (Fig. 7C).

Fig. 7.

Compartment-specific regulation of spontaneous calcium transients in hiPSC-derived RGCs by PIEZO1 activity and magnetic force. (A, B) Representative calcium traces recorded via two-photon microscopy from the soma (A) and neurites (B) of hiPSC-RGCs treated with Yoda1, Dooku1, and/or MF following MagNAV internalization. (C) Comparison of peak calcium transient amplitudes between soma and neurites from paired recordings within the same cells. Most cells exhibited higher transient amplitudes in neurites than soma, though some showed the inverse. (D) Quantification of spontaneous calcium transient frequency in the soma. Yoda1 significantly increased transient frequency, while Dooku1 significantly decreased it. MagNAV + MF alone had no significant effect, but a trend toward further suppression was observed when combined with Dooku1. (E) Quantification of spontaneous calcium transient frequency in the neurites. Yoda1 did not significantly affect activity, whereas MagNAV + MF significantly suppressed neuritic calcium transients. Dooku1 addition did not further reduce activity beyond MF alone. Data represent mean ± SEM. ∗p < 0.05, ∗∗p < 0.01. (F) Representative immunofluorescence images of hiPSC-RGCs cultured with MagNAVs in the absence (top) or presence (bottom) of MF. Cells were stained for PIEZO1 (red), F-actin/phalloidin (green), and nuclei (blue). Without magnetic stimulation, PIEZO1 distribution remained diffuse along neurites. Scale bar = 50 μm. (G) Quantification of the PIEZO1 neurite-to-soma fluorescence intensity ratio. Data are mean ± SEM. ∗∗∗∗p < 0.0001.

In the soma, quantitative analysis of calcium transients (Fig. 7D) showed that Yoda1 significantly increased transient frequency compared to the control, while Dooku1 significantly reduced it. Treatment with MagNAVs + MF alone had no significant effect on the peak number. When combined with Dooku1, MagNAVs + MF exhibited a trend toward further reduction in calcium transients compared to Dooku1 alone, although this difference did not reach statistical significance. In contrast, calcium dynamics in the neurites followed a different pattern (Fig. 7E). Yoda1 did not significantly affect the transient frequency, whereas MagNAVs + MF alone significantly suppressed neuritic calcium activity. Adding Dooku1 to MagNAVs + MF did not result in further suppression, suggesting that mechanical stimulation may independently mediate calcium regulation in neurites. To further explore the compartment-specific distribution of PIEZO1, we performed immunofluorescence staining of PIEZO1 together with F-actin (phalloidin) in hiPSC-RGCs cultured with MagNAVs under conditions with or without magnetic stimulation. Without magnetic stimulation, PIEZO1 distribution appeared diffuse along neurites with limited enrichment at tips. By contrast, magnetic stimulation induced prominent colocalization of PIEZO1 and F-actin at neurite terminals, as shown in the magnified views of growth cones (Fig. 7F). Furthermore, we quantified PIEZO1 expression in the soma and neurites. We found that in the absence of magnetic force, PIEZO1 signaling was primarily localized in the soma. In the presence of MagNAVs plus magnetic field force (MF), PIEZO1 signaling on neurites was significantly enhanced (Fig. 7G). These results suggest that magnetic stimulation promotes the recruitment of PIEZO1 to actin-rich growth cones, supporting its role in spatially regulating neurite extension.

These results show that PIEZO1 activity and magnetic forces differentially regulate spontaneous calcium transients in a compartment-specific manner. In particular, Dooku1-induced decreases in calcium activity in both the cell body and neurites, whereas MagNAVs + MF-induced decreases in calcium activity only in the neurites. These findings support the hypothesis that PIEZO1-mediated calcium influx plays an inhibitory role in neurite elongation.

4. Discussion

In the present study, we demonstrated that internalization of MagNAVs by hiPSC-derived RGCs, followed by exposure to an external MF, not only directed the orientation of neurite outgrowth (Fig. 4D and E) but also significantly enhanced neurite elongation (Fig. 4F). These findings are consistent with prior studies showing that MNPs, when subjected to external magnetic stimulation, can exert mechanical forces to guide neurite directionality and promote extension in various neuronal cell types, including PC12, SH-SY5Y, spiral ganglion neurons, and hippocampal neurons [25,40]. De Vin et al. further reported that MNPs with larger diameters (75 nm) induced more robust neurite elongation compared to smaller particles (3 nm), highlighting the influence of nanoparticle size on mechanotransductive outcomes [41]. Additionally, when cells containing nanoparticles were exposed to a magnetic field, it induced stretching that promoted the axonal cytoskeleton's remodeling and increased the axon's microtubule density [42]. Furthermore, it was reported that mechanical tension induced by attaching magnetic nanoparticles to the membrane tips of neuronal filopodia and applying an external magnetic field can accelerate actin polymerization within filopodia and enhance filopodia growth [26]. These findings suggest that magnetic nanoparticle-mediated forces can provide an effective strategy for guiding neurite outgrowth.

While these previous studies have explored potential applications in neuroregeneration, our study yields novel results. First, we used raspberry-like, relatively large (approximately 150–200 nm) multiparticle MagNAVs. Unlike traditional superparamagnetic nanoparticles, these particles offer enhanced magnetic responsiveness and hold promise for future applications in drug delivery. Second, this is the first demonstration of magnetic nanoparticle-mediated neurite guidance in hiPSC-derived RGCs, a clinically relevant model for optic neuropathies such as glaucoma. Third, we identified PIEZO1 as a negative mechanosensitive regulator of neurite elongation and demonstrated that pharmacological inhibition of PIEZO1 synergizes with magnetic stimulation to promote axonal growth. Finally, using two-photon calcium imaging, we identified a novel compartment-specific mechanotransduction mechanism, in which somatic calcium activity is PIEZO1-dependent, while magnetic forces primarily regulate neuronal calcium activity. These advances enhance the understanding of magnetic guidance, provide mechanistic insights and translational applications, and make MagNAV a valuable tool for studying force-dependent neuronal remodeling and as a potential platform for neural regenerative medicine.

In addition, while neurite length in hiPSC-RGCs was significantly increased in the MagNAVs + MF group compared to the control (Fig. 4F), no statistically significant difference was observed between the MagNAVs alone and MagNAVs + MF groups. This contrasts with findings from Hu et al. (2021), in which internalization of superparamagnetic iron oxide (SPIO) nanoparticles alone was sufficient to promote neurite elongation in spiral ganglion neurons, with the application of magnetic fields further enhancing the length and directional growth [25]. These discrepancies may be attributed to differences in neuronal cell types, suggesting that the mechanosensitive response to nanoparticle-induced forces could vary across distinct neuronal populations. Another possible explanation lies in the differences in particle size; the MagNAVs used in our study may differ in diameter or magnetic responsiveness compared to the SPIOs employed by Hu et al., potentially affecting the efficacy of mechanotransduction and neurite extension.

The application of magnetic nanoparticles inherently requires guidance by an external magnetic field. In this study, we employed a static magnetic field of 0.229 T, which is supraphysiological but remains below the ICNIRP acute exposure limit for the general public (0.4 T) and well within occupational limits (2 T for the head/torso and 8 T for the limbs) [43]. Moreover, clinical modalities such as MRI (1.5–3 T) and transcranial magnetic stimulation (TMS, peak ∼2 T) routinely operate at substantially higher field strengths under controlled conditions [44]. These precedents underscore the translational feasibility of magnetic guidance strategies based on static magnetic fields, provided that appropriate safety safeguards are in place. Iron oxide nanoparticles as biomedical materials have been widely adopted [39]. It is widely used for MRI contrast agents, diagnostic, drug delivery, and cell labeling applications [39,[45], [46], [47]]. In addition to their intrinsic magnetic properties, iron oxide nanoparticles offer higher stability, better biocompatibility, and lower manufacturing costs than other materials, making them a popular choice for various applications [48]. In the present study, we utilized MagNAVs with a raspberry-like structure. This unique morphology allowed for the embedding of RITC onto the nanostructure, resulting in fluorescence. Their size, shape, and charge influence the efficiency and mechanisms by which these particles enter cells and the presence of functional groups on their surface [49]. The particles employed in this study were positively charged, facilitating their adhesion to the cell surface and increasing cellular uptake likelihood. With an approximate size of 150 nm, these nanoparticles demonstrated more substantial magnetic responsiveness compared to smaller magnetic nanoparticles when guided by an external magnetic field [50]. Additionally, larger particles can carry a greater payload, providing significant potential for future applications, particularly drug delivery. Under the influence of an external magnetic field, these particles can facilitate the targeted delivery of drugs to specific regions, thereby enhancing the efficacy of therapeutic interventions [50]. Our in vitro system of hiPSC-derived RGCs provides mechanistic insights into the role of magnetic guidance and PIEZO1 in neurite outgrowth. Although animal studies were not included in the present work, previous in vivo investigations have shown that iron oxide–based nanoparticles are generally biocompatible and can persist in the posterior segment of the eye without inducing structural or functional damage [51,52]. These findings support the potential applicability and safety of MagNAVs in retinal explants and animal models, underscoring the importance of future in vivo studies to validate translational potential.

The downstream signaling pathways activated after particle internalization are closely related to the mechanisms by which they are taken up by the cells [53]. According to the size and characteristics of the particles used in the present study, it is hypothesized that they entered the cells through clathrin-mediated endocytosis or micropinocytosis [49,53]. Clathrin-mediated endocytosis is the primary mechanism through which cells acquire nutrients from their surrounding environments. At the same time, macropinocytosis is a non-specific process that engulfs extracellular fluids using actin-driven protrusions to encapsulate and internalize external materials into the cell [53]. After entering the cell, these MagNAVs are transported to various locations by the cytoskeleton and motor proteins [53]. However, the exact intracellular destinations, pathways, and downstream signaling mechanisms affecting RGCs remain unclear.

Previous studies have shown that activation of PIEZO1 by the agonist Yoda1 suppresses neurite outgrowth in retinal ganglion cells (RGCs) [31]. In contrast, GsMTx4, a spider venom peptide from Grammostola spatulata known to block cation-permeable mechanosensitive ion channels, has been reported to promote neurite extension [31]. In our study, treatment with Yoda1 did not significantly inhibit neurite outgrowth in hiPSC-derived RGCs (Fig. 5E), whereas pharmacological inhibition of PIEZO1 using Dooku1 significantly enhanced neurite elongation (Fig. 5E). This apparent discrepancy may be attributed to differential responsiveness of PIEZO1 channels in the soma versus neurites. Our calcium imaging analysis revealed that Dooku1 suppressed spontaneous calcium transients in both soma and neurites, whereas Yoda1 significantly increased calcium transient frequency only in the soma, with no effect in neurites (Fig. 7). These results suggest that under our experimental conditions, PIEZO1 channels in neurites may already be in a highly active state (Fig. 7C), rendering them less responsive to further activation by Yoda1. Consequently, Yoda1 failed to elicit a significant impact on neurite outgrowth. In contrast, Dooku1 likely inhibited a substantial number of activated PIEZO1 channels, thereby facilitating neurite extension.

Yoda1 activates PIEZO1 channels and facilitates Ca²⁺ influx into the cell. Dooku1 inhibits PIEZO1 channel function, reducing Ca²⁺ influx. Research indicates that the route of Ca²⁺ entry into the cytosol affects its impact on neuronal growth [54]. Specifically, calcium entry through mechanosensitive channels inhibits neuronal growth, while influx through channels like transient receptor potential (TRP) channels and intracellular release mechanisms promote it [54]. Neurite outgrowth is influenced by growth-inhibiting and growth-promoting Ca²⁺ signaling within the growth cone. Alternatively, blocking mechanosensitive channels with drugs such as Dooku1 increases Ca²⁺ influx through non-mechanosensitive channels [54], which promotes neurite outgrowth. This mechanism may explain why Dooku1 enhances neuronal growth despite inhibiting PIEZO1 channels. Moreover, Dooku1's specific targeting of PIEZO1 channels suggests their predominant role in regulating Ca²⁺ influx during neuronal growth processes, further supporting their influence on neurite outgrowth dynamics.

Activation of the mechanosensitive ion channel PIEZO1 induces calcium influx, which has been associated with the inhibition of neurite outgrowth in various neuronal models [31,55]. However, the precise mechanisms underlying this inhibitory effect remain a subject of ongoing investigation. Several studies suggest that mechanosensitive ion channels exert a suppressive role on neurite elongation, as evidenced by enhanced neurite extension following channel blockade. For instance, inhibition of mechanosensitive channels using GsMTx4, a tarantula-derived peptide, resulted in longer neurites, possibly by redirecting calcium influx through alternative, non-mechanosensitive pathways that are more permissive for growth [54]. Similarly, Calabrese et al. (1999) reported that aminoglycoside-mediated blockade of stretch-activated ion channels in leech neurons significantly promoted axon elongation [56]. These findings support the hypothesis that PIEZO1-mediated calcium entry may negatively regulate neurite outgrowth. In addition to these channel-based mechanisms, other studies propose that PIEZO1 suppresses neurite extension through downstream calcium-dependent signaling cascades [31,57]. Growth cone navigation is known to be influenced by external cues interacting with cell adhesion molecules (CAMs) on the neuronal membrane. These interactions can elicit either attractive or repulsive responses and trigger localized calcium influx, which then activates intracellular signaling pathways involving cyclic adenosine monophosphate (cAMP), protein kinase A (PKA), and the ryanodine receptor (RyR) on the endoplasmic reticulum [57]. Activation of RyR promotes calcium-induced calcium release (CICR), a process that has been shown to support neurite outgrowth. Although there is currently no direct evidence linking PIEZO1 activation to the initiation of CICR in neurons, it is plausible that PIEZO1-mediated calcium entry fails to trigger CICR and instead activates alternative pathways that inhibit neurite elongation. These findings suggest that the effect of PIEZO1 activation on neurite dynamics may be highly context-dependent, influenced by the mode and localization of calcium entry, the cellular microenvironment, and the integration of downstream signaling pathways.

In this study, we observed that pharmacological inhibition of PIEZO1 with Dooku1 led to significantly greater neurite elongation compared to mechanical stimulation via MagNAVs and magnetic field exposure alone (Fig. 6B). This additive effect suggests that PIEZO1 modulation and mechanical force operate through partially independent pathways that converge on neurite growth, resulting in a synergistic enhancement of axonal extension. PIEZO1 is a large transmembrane protein that forms a non-selective cation channel widely expressed across multiple tissues [28]. As a mechanosensitive ion channel, PIEZO1 can be activated by various physical stimuli, including membrane tension, substrate stiffness, shear stress, tissue compression, expansion, and osmotic pressure [27]. Prior studies have demonstrated that magnetic nanoparticles internalized by neurons can generate intracellular mechanical tension when subjected to an external MF [58], raising the likelihood that PIEZO1 channels are activated in response to MF-induced mechanical stretch in MagNAV-loaded hiPSC-RGCs.

Two principal models have been proposed to explain PIEZO1 mechanogating: the “force-from-lipids” mechanism, in which alterations in local membrane tension and curvature directly induce conformational changes in the channel [59], and the “force-from-filaments” mechanism, in which cytoskeletal or extracellular matrix tethers transmit mechanical cues that activate the channel [60]. In our MagNAV-loaded hiPSC-RGCs, local pulling forces generated by MagNAVs may increase membrane tension, thereby contributing to lipid-mediated gating. In parallel, changes in F-actin dynamics within the growth cone may reflect filament-based force transmission (Fig. 7F). Consistent with this, Wang et al. reported that PIEZO1 is closely associated with the cadherin mechanotransduction complex and the actin cytoskeleton [61]. Nevertheless, our present results cannot determine which mechanism predominates. Further studies will be required to elucidate the primary pathway through which MagNAV-induced forces engage PIEZO1 gating in retinal ganglion cells.

Prior studies demonstrate that mechanosensitive PIEZO channels restrain neurite extension via growth-cone Ca²⁺ signaling. In Drosophila and mammalian neurons, Piezo activation at growth cones elicits local Ca²⁺ transients that engage a CaMKII→NOS→cGMP/PKG cascade to limit axon regeneration; conversely, Piezo loss-of-function accelerates regeneration, whereas pharmacological activation (e.g., Yoda1) suppresses it [55]. Elevated growth-cone Ca²⁺ also modulates growth-cone turning and elongation through Ca²-dependent effectors such as CaMKII [62,63]. These findings align with our observation that PIEZO1-evoked Ca²⁺ acts as a negative regulator of neurite elongation in hiPSC-RGCs.

Several limitations should be acknowledged. First, our conclusions rely on pharmacological modulation of PIEZO1 and Ca²⁺ imaging in two-dimensional hiPSC-RGC cultures, which may not fully recapitulate in vivo cytoarchitecture. Second, cell-to-cell heterogeneity in MagNAV uptake likely contributes to variability in the extent of neurite guidance and in compartment-specific (soma vs. neurite) Ca²⁺ responses. Third, PIEZO1 redistribution was assessed by endpoint immunofluorescence; more precise approaches will be required to distinguish between surface and intracellular pools and to quantify spatial dynamics with higher fidelity.

5. Conclusion

Our study demonstrates that magnetically guided MagNAVs effectively promote directional neurite elongation in hiPSC-derived RGCs. PIEZO1 plays a key role in modulating this process, where its inhibition enhances, and its activation limits, axonal growth—mainly through calcium-mediated mechanisms. These findings highlight the synergistic potential of combining mechanical stimulation with molecular modulation to enhance neural regenerative strategies in retinal and CNS repair.

CRediT authorship contribution statement

I-Ting Chen: Writing – original draft, Methodology, Formal analysis. Pei-Hsin Chu: Methodology. Shang-Hsiu Hu: Methodology. Shih-Kuo Chen: Methodology. Chuan-Chin Chiao: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Investigation, Funding acquisition, Conceptualization. Tien-Chun Yang: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Formal analysis, Data curation, Conceptualization.

Funding

This study was funded by Taiwan's National Science and Technology Council, MOST-111-2314-B-007-002-MY3 (to CCC).

Declaration of competing interest

The authors confirm that there are no known financial or personal conflicts of interest that could have influenced the work presented in this study.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.