Microcontact printing transfers a pattern consisting of biomolecules onto another surface. This technique, which is well-established in our laboratory [18], allows biomolecule deposition in µm-sized patterns. Polydimethylsiloxane (PDMS) stamps were produced using a micropatterned template from a silicon wafer referred to as a “master plate”. The surface relief of the elastomer stamp is formed by casting and curing liquid PDMS against the master plate. The raised and lowered regions of the master plate are replicated into the cured stamp. Elastomer is chosen as material for its ability to ensure conformal contact to non-planar surfaces and its hydrophobic nature, which is ideal for transferring biomolecules. In our experimental setup, the PDMS stamp adsorbs a collagen hydrogel solution loaded with peptides or proteins of our interest and transfers it onto a semipermeable extra membrane upon contact.

Numerous stamps were generated from one master plate. The creation of the master plate follows a previously described protocol [18]. The PDMS prepolymer (Sylgard 184 Silicone Elastomer Kit, Dow, Seneffe, Belgium, 01673921) consists of two components. The elastomer curing agent was mixed with the elastomer base solution in a ratio of 1:10. The blended PDMS was then poured onto the master plate positioned in the center of a petri dish. Air bubbles were removed using a desiccator connected to a vacuum pump. After curing overnight at 60 °C, the solid PDMS was peeled off the master plate, and was cut to size with a scalpel, resulting in stamps.

In this study, we loaded pORN (poly-DL-ornithine, Sigma-Aldrich, St. Louis, MO, USA, P-0671), human recombinant A$\beta$(42) (Innovagen, Lund, Sweden, SP-BA42-1), aggregated A$\beta$(42) (aggregation procedure detailed in [27]), reversed A$\beta$(42) (Innovagen, Lund, Sweden, SP-BA42R-1), P301S mutated aggTau (Abcam, Cambridge, UK, ab246003), and WT Tau441 (Abcam, Cambridge, UK, ab84700) into the collagen hydrogel solution. Bovine collagen solution type I (Sigma-Aldrich, St. Louis, MO, USA, 804592) was crosslinked with 4arm-PEG succinimidyl succinate (Sigma-Aldrich, St. Louis, MO, USA, JKA7006) during printing procedure. All components were kept sterile and on ice to prevent premature gel formation. 133 µL of collagen solution (3 mg/mL) and 20 µL of 100 mM phosphate-buffered saline (PBS, pH 7.4, autoclaved, 10 mM Na${}_2$HPO${}_4$ (Merck Millipore, Darmstadt, Germany, 106586), 137.75 mM NaCl (Roth, Karlsruhe, Germany, 3957.1), 2.68 mM KCl (Roth, Karlsruhe, Germany, 6781.3)) were mixed. Then, 1.6 µL of 1 M NaOH (Roth, Karlsruhe, Germany, 6771.3) was added to achieve the desired pH of 7.2. A fluorescent antibody (2 µL, Alexa Fluor 546 anti-rat, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, A11081) was added to aid visualization of the print. Collagen hydrogel solution was vortexed and spinned down. After preparing a stock solution of 12.5 mg/mL PEG in 10 mM PBS (pH 7.4), 25 µL of the stock solution were blended with collagen hydrogel solution. The solution was then loaded with the respective peptides/proteins (10 µL of 1 mg/mL pORN plus 10 µL of 10 mM PBS to substitute, or 20 µL of 1 mg/mL A$\beta$(42), or 20 µL of 100 µg/mL tau) or left empty (20 µl of 10 mM PBS) to serve as negative controls.

Following collagen hydrogel preparation, 15 µL of the solution was applied onto each micropatterned PDMS stamp. A coverslip (R. Langenbrinck, Emmendingen, Germany, 01-2126/1) was placed on top to distribute the ink solution on the stamp and incubated for 15 min at 37 °C to allow crosslinking of the collagen hydrogel solution. After removing the coverslip, striking off the excess solution, the stamp was inverted, and the collagen hydrogel solution was transferred onto the semipermeable extra membrane (Isopore, Merck Millipore, Darmstadt, Germany, HTTP02500) by applying pressure with 18 g coins on top overnight at 4 °C. The position of the µCP was marked with small dots using a permanent marker to aid in arranging the brain slices [18], and the membranes were sterilized under UV light for 20 min.

Organotypic brain slice cultures serve as a physiologically relevant three-dimensional ex vivo model of the brain, a technique well-established in our laboratory [16]. Organotypic brain slices were taken from postnatal day 8–10 C57BL/6 WT mice randomized between the groups, irrespective of sex. All experimental procedures were approved by the Austrian Ministry of Science and Research, and complied with Austrian guidelines on animal welfare and experimentation, following the ethical principles of the three Rs (replace, reduce, refine).

Brains were dissected under sterile conditions. After carefully removing the cerebellum, the brains were affixed onto the sample holder platform of a water-cooled vibratome (Leica, Nussloch, Germany, VT1000S) with adhesive (Loctite 401, Henkel, Düsseldorf, Germany 231435) on their newly formed caudal surface. Coronal slices, 150 µm thick at the hippocampus level, were cut in a sterile preparation medium cooled down to approx. 5 °C (pH 7.2–7.3, autoclaved, 1$\times$ MEM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA, 11012044), 5.12 mM NaHCO${}_3$ (Merck Millipore, Darmstadt, Germany, 106329)). From each mouse, six slices were obtained from the hippocampal area, horizontally halved, and the upper hippocampal part was positioned onto microcontact-printed extra membranes in cell culture inserts (Millicell, Merck Millipore, Darmstadt, Germany, PICM03050). Each well of the six-well plate (Sarstedt, Nümbrecht, Germany, 83.3920) was filled with 1 mL of sterile-filtered slice culture medium (pH 7.2, sterile filtrated, 1$\times$ MEM, 5.12 mM NaHCO${}_3$, 31.5 mM glucose (Merck Millipore, Darmstadt, Germany, 49159), 2 mM glutamine (Merck Millipore, Darmstadt, Germany, 100289), 10% heat-inactivated horse serum (HS, Gibco, Thermo Fisher Scientific, Waltham, MA, USA, 16050-122), 0.25$\times$ HBSS (Gibco, Thermo Fisher Scientific, Waltham, MA, USA, 24020091), 1$\times$ antibiotic-antimitotic solution (Sigma-Aldrich, St. Louis, MO, USA, A5955). The half slices were incubated on the microcontact-printed surfaces for two weeks at 37 °C 5% CO${}_2$, and slice culture medium was refreshed weekly. Slices were fixed in 4% paraformaldehyde (PFA, pH 7.4, filtered, Sigma-Aldrich, St. Louis, MO, USA, 158127) for 3 h and stored at 4 °C until further use.

Fluorescently labeled dextrans are frequently employed in live cell imaging studies to track both anterograde and retrograde neuronal pathways, a technique firmly established in our laboratory [28]. MiniRuby (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, D3312) is a dextran conjugate combined with the red-fluorescent dye tetramethylrhodamine. The MiniRuby crystal was positioned on the microcontact-printed surface, two to three mm away from the two-week-old brain slice, using the tip of a fine needle. The slices were then observed for a time period ranging from 1 to 7 days under a microscope (Leica, Nussloch, Germany, DMIRB) within a chamber equipped with temperature (PeCon, Erbach, Germany, tempcontrol 37-2 digital) and CO${}_2$ concentration (PeCon, Erbach, Germany, CO${}_2$ controller) control units. Observations were conducted using a Y3 filter (excitation 535/50 nm, emission 610/75 nm).

Voltage-sensitive and calcium-sensitive dyes are used to monitor the electrical activity in neurons. Voltage-sensitive dyes change their spectral properties with membrane potential shifts. Conversely, calcium-sensitive dyes alter spectral properties upon binding intracellular calcium ions (Ca${}^{2+}$), providing insight into action potential dynamics. After a 2-week culture period, organotypic brain slices underwent a 30-minute wash with serum-free medium at 37 °C in a 5% CO${}_2$ environment. Subsequently, the brain slices were loaded with the voltage-sensitive dye FluoVolt (FluoVolt Membrane Potential Kit, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, F10488) or the calcium-sensitive dye Rhod-4 (AAT Bioquest, Pleasanton, CA, USA, 21121) free-floating/inverted for 60 minutes at 37 °C and 5% CO${}_2$, following the manufacturer’s guidelines. For FluoVolt staining, the dye was diluted to 1:1000 and PowerLoad at 1:100 in serum-free medium, while Rhod-4 was used at a concentration of 10 µM, with PowerLoad (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, P10020) at 1:100 and 2.5 mM Probenecid (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, P36400) in serum-free medium. Following loading, the membranes were washed once with serum-free medium and transferred back to their inserts, where they were incubated in serum-free medium for 2 hours at 37 °C and 5% CO${}_2$. Organotypic brain slices loaded with Rhod-4 were instead incubated in serum-free medium with 2.5 mM Probenecid. Subsequently, each well of a fresh six-well plate received 100 µL of 10 mM PBS. For sections loaded with FluoVolt, PBS was additionally supplemented with 1:10 Neuro Backdrop Background Suppressor. The membranes were rinsed in 10 mM PBS and placed inverted in 100 µL PBS. The area of interest was then focused under the microscope using a 10$\times$ objective. Recording of a video was initiated, followed by the addition of 100 µL of isotonic KCl solution (pH 7.4, 140 mM KCl, 5 mM NaCl, 1.8 mM CaCl${}_2$ (Merck Millipore, Darmstadt, Germany, 1172113), 1 mM MgCl${}_2$ (Merck Millipore, Darmstadt, Germany, 105833), 20 mM HEPES (Sigma-Aldrich, St. Louis, MO, USA, H3375), 20 mM Glucose (Merck Millipore, Darmstadt, Germany, 49159)) on top of the membrane to induce depolarization. The depolarization process, achieved with a final concentration of 70 mM KCl (a dilution of 1+1 with PBS), was observed for 2–5 minutes under the microscope (standard FITC and TRITC filter sets). Following depolarization, the organotypic brain slice was washed once with serum-free medium, and the membrane was returned to its insert and incubated at 37°C and 5% CO${}_2$ for either 1 hour or 24 hours. Subsequently, sections were fixed in 4% PFA and stained for ATP1A1.

Immunostaining was employed to assess the outgrowth of neurites along the microcontact-printed surfaces using the cytoskeleton markers neurofilament (NF) and microtubule-associated protein 2 (MAP2), as well as the myelin sheath marker myelin oligodendrocyte glycoprotein (MOG). Additionally, the interaction with astrocytes was visualized through glial fibrillary acidic protein (GFAP) and the neuronal activity via membrane bound Na,K-ATPase (ATP1A1). Furthermore, an immunostaining for SIRT1 was conducted, which is an intracellular regulatory protein related to aging and age-associated diseases. Fixed slices on the membrane were transferred to a six-well plate and washed 3 $\times$ 3 min each with fresh 10 mM PBS. The brain slices were then incubated in Triton-PBS 0.1% (T-PBS) for 30 min at room temperature (RT) with gentle shaking. Subsequently, they were washed again for 3 $\times$ 3 min with fresh 10 mM PBS and blocked in T-PBS 0.1% + bovine serum albumin (BSA) 0.2% + HS 20% for 30 min with slow shaking. If the primary antibody was mouse-derived, an additional blocking step was performed through incubating the slices with mouse on mouse blocking reagent (M.O.M., Vector Laboratories, Newark, NJ, USA, MKB-2213) at a ratio of 1 drop blocking reagent per 2.5 mL 10 mM PBS for 1 h at 20 °C. Following this, the sections were incubated in T-PBS 0.1% + BSA 0.2% with primary antibodies against NF (NF-H/NF200, proteintech, Rosemont, IL, USA, 60331-1-lg, 1:500), tau (Tau-5, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, AHB0042, 1:250), MAP2 (Synaptic Systems, Göttingen, Germany, 188002, 1:1000), MOG (proteintech, Rosemont, IL, USA, 12690-1-AP, 1:1000), GFAP (Merck Millipore, Darmstadt, Germany, AB5541, 1:2000), ATP1A1 (proteintech, Rosemont, IL, USA, 55187-1-AP, 1:1000) and SIRT1 (ABclonal, Woburn, MA, USA, A0230, 1:600) at 4 °C for 48 h. After washing again 3 $\times$ 3 min with fresh 10 mM PBS, the samples were incubated with the corresponding green-fluorescent Alexa Fluor 488 anti-mouse (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, A11029, 1:400), or Alexa Fluor 488 anti-rabbit (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, A21206, 1:400), or Alexa Fluor 488 anti-chicken (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, A11039, 1:400) in T-PBS 0.1% + BSA 0.2% at RT for 1 h with gentle shaking and protected from light. The sections were washed again for 3 $\times$ 3 min with fresh 10 mM PBS, and all slices were counterstained with blue-fluorescent nuclear dye DAPI (Sigma-Aldrich, St. Louis, MO, USA, D9542, 1:10,000) in T-PBS 0.1% at RT for 1 h with gentle shaking and protected from light. After a final washing step of 3 $\times$ 3 min with fresh 10 mM PBS, the sections were mounted with Mowiol onto glass slides and left to dry overnight. Widefield microscopy was conducted using the fluorescence microscope (Olympus, Tokyo, Japan, BX61). Alexa Fluor 488 was observed in the green channel (ex 480/40 nm, em 527/30 nm) and Alexa Fluor 546 in the red channel (ex 535/50, em 610/75). Fluorescence images were captured using the connected ProgRes C14 camera (Jenoptic, Jena, Germany) and Openlab software (version 5.5.0, Improvision, Coventry, UK).

Confocal microscopy was performed using LSM 980 with Airyscan 2 (Zeiss, Oberkochen, Germany). The 63$\times$ 1.2 NA (glycerol) objective was selected, and the 488 nm laser was activated to locate the appropriate focus and region of interest. The 63$\times$ objective was further corrected for differences in coverslip thickness. The imaging mode (4Y) and dyes to be detected (Alexa Fluor 488, Alexa Fluor 546, DAPI) were selected, Airy Scan calibration was initiated, and the percentage of laser strength was adjusted accordingly. Interval in the z-direction between the first and last planes as well as sampling interval (reassessed by the online tool microscopy Nyquist rate calculator) were defined and a Z-stack composite image was acquired. Parameters for deconvolution with the Huygens Professional software (version 23.10, Scientific Volume Imaging, Hilversum, Netherlands) were set, including lens immersion (glycerin, 1.456), embedding medium (Mowiol, 1.49), signal/noise per channel (12 for each channel), max. iterations (100), and quality change threshold (0.1%). After the deconvolution, images were processed with Imaris software (version 10.0.1, Oxford Instruments, Abingdon, UK) for three-dimensional representation with added surfaces. Surface detail, threshold, and filter surface were adjusted for each image.

Propidium iodide (PI) is a widely utilized red-fluorescent nuclear stain employed to identify dead cells, as it does not permeate through living cell membranes. The inserts containing viable brain slices were gently immersed in slice culture medium containing 2 µg/mL of PI (Sigma-Aldrich, St. Louis, MO, USA, P4170) and were then incubated for 30 min at 37 °C and 5 % CO${}_2$. Subsequently, the slices underwent 3 $\times$ 3 min washes with fresh 10 mM PBS and were fixed for 3 h at 4 °C in 4% PFA. To visualize living and dead cells, DAPI counterstaining was performed on all slices. Additionally, control brain slices were exposed to 2 µL of H${}_2$O${}_2$ (Merck Millipore, Darmstadt, Germany, 1.07209.0250) per mL slice culture medium for two days to induce cell death prior to PI incubation.

Human recombinant peptides of monomeric A$\beta$(42) (1 µg, Innovagen, Lund, Sweden, SP-BA42-1), aggregated A$\beta$(42) (750 ng, aggregation procedure detailed in [27]), and reversed A$\beta$(42) as a control (1 µg, Innovagen, Lund, Sweden, SP-BA42R-1) were analyzed via Western blot. Additionally, human recombinant proteins of WT Tau441 (1 µg, Abcam, Cambridge, UK, ab84700) and mutated P301S tau aggregate (1 µg, Abcam, Cambridge, UK, ab246003) were subjected to Western blot analysis. Samples were loaded in their native state with sample buffer (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, NP0007) onto a 10% Bis-Tris precast polyacrylamide gel (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, NP0301BOX), along with a protein standard marker (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, LC5800). Sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) was conducted for 30 min at 200 V, followed by electro-transfer to a polyvinylidene difluoride (PVDF) membrane (Merck Millipore, Darmstadt, Germany, ISEQ00010) for 20 min at 25 V in a semi-dry transfer cell (Pierce Power Blot Cassette, Thermo Fisher Scientific, Waltham, MA, USA). Commercially available running buffer (Thermo Fisher Scientific, Waltham, MA, USA, NP0002) and transfer buffer (Thermo Fisher Scientific, Waltham, MA, USA, 84731) were used for separation and blotting. For detection, reagents from the anti-mouse chemiluminescent Western blot immunodetection kit (Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, WB7104) were employed. Briefly, membranes were blocked for 30 min at RT and then incubated overnight at 4 °C on a shaker with primary antibodies against A$\beta$ clone 6E10 (1:500, BioLegend, San Diego, CA, USA, 803015) and Tau-5 (1:500, Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA, AHB0042). Subsequently, blots were washed and incubated with secondary alkaline-phosphatase conjugated anti-mouse antibodies for 30 min at RT. After washing, bound antibodies were detected using CDP-Star chemiluminescent substrate solution and visualized with a cooled CCD camera (SearchLight, Thermo Fisher Scientific, Waltham, MA, USA).

An unbiased image evaluation was conducted to determine the quantity and length of neurites that had extended. Slices were considered for inclusion if they exhibited at least one protruding neurite. Only neurites surpassing a length of 100 µm and clearly situated within the microcontact-printed lanes were included in the quantitative examination. Images were captured using a widefield microscope (Olympus, Tokyo, Japan, BX61 and Leica, Nussloch, Germany, DMIRB) at a 20$\times$ magnification and color-merged to display green neurites with blue nuclei on red microcontact-printed surface. These images were captured from the boundary of the horizontal cutting side, encompassing representative microcontact-printed lanes (equivalent to a microcontact-printed area of 300 µm in width). Utilizing the NeuronJ plugin within ImageJ software (version 1.48; National Institute of Health, Bethesda, MD, USA), the images were analyzed. The pixel-to-µm ratio was calibrated based on microscope measurements, and the image type was converted to 8-bit images. Neurites were then traced by manually following along the neurite path suggested by the program until the deviation from the anticipated tracing became too pronounced. The neurite counts were converted to neurite number per mm for each brain slice, and for neurite length, an average was calculated for each brain slice. For ATP1A1, a segmented line was drawn along the outer membrane of an ATP1A1-positive outgrown fiber and the mean grey value was measured. The nuclear SIRT1 was determined by mean grey value of immuno-positive nuclei in the brain slice close to the boundary of the horizontal cutting side. Values were reported as mean $\pm$ standard error of the mean. The sample size (n) indicated the number of brain slices assessed, each obtained from different animals. Statistical significance was determined using a one-way ANOVA with Fisher’s LSD posthoc test for multiple comparisons of means, with significance denoted by p-values less than 0.05. All graphs were created with GraphPad Prism software (version 10.1.2, GraphPad Software Inc., San Diego, CA, USA).

Hippocampal organotypic half brain slices were arranged on microcontact-printed extra membranes, which were centrally laid into a cell culture insert in a six well plate filled with slice culture medium (Fig. 1A). These half brain slices were connected to the µCPs with the horizontal cutting side and straight lanes arranged at a 90° angle (Fig. 1B). Our area of interest was located along brain slice’s horizontal cutting side (Frame in Fig. 1B). µCPs were identified as 800 µm long and 30 µm in width sharp and clear lines using a red-fluorescent Alexa-546 antibody (Fig. 1C). The brain slice’s position on the µCP was easily recognized using the blue-fluorescent nuclear marker DAPI (Fig. 1D). To evaluate brain slice viability on the µCP, 2-week-old cultures were incubated with PI, revealing minimal nuclear staining (Fig. 1E), in contrast to a hydrogen peroxide-treated control (Fig. 1F,G).

Fig. 1.

Coupling of collagen-based microcontact prints (µCPs) with organotypic brain slices. (A) Coronal half slice of 150 µm in thickness at the hippocampus level arranged on the microcontact-printed extra membrane in a cell culture insert. (B) Top view schematic representation of the coupling between an organotypic brain slice (grey) and a µCP (red). The frame shows the area of interest located along brain slice’s horizontal cutting side. (C) Alexa Fluor 546 (comprised in the collagen hydrogel solution) visualizes the µCP pattern and brain slice position can be recognized through brighter background fluorescence (D) Nuclear DAPI staining shows the location of the brain slice (blue) connected to the µCP (green, Alexa Fluor 488). (E) Viability of brain cells is examined by propidium iodide (PI) staining under normal conditions. Since PI is a membrane impermeable DNA binding dye, intact and viable cells do not show any nuclear staining. (F) For control purposes, brain slices (blue, DAPI) were exposed to oxidative stress (+ H${}_2$O${}_2$). (G) Nuclei of dead brain cells are stained with PI. (H) The number and length of neurofilament (NF)-positive neurites per mm µCP are quantified on control membranes with no µCP (-), collagen-based µCPs without any load (PBS) and loaded with poly-DL-ornithine (pORN). (I) Evaluated number and length of all outgrown neurites along collagen-based µCPs co-loaded with pORN and with human beta-amyloid(42) (A$\beta$(42)) or human P301S mutated aggregated tau (Tau P301S). Values are given as mean $\pm$ SEM. Statistical significance was evaluated by one-way ANOVA with Fisher’s LSD posthoc test for multiple comparisons of means (* p 0.05). Scale bar in A = 1 cm, in C = 500 µm, in D–G = 100 µm.

Nearly no spontaneous outgrowth of neurites was visible in slices without µCPs (Fig. 1H). Since we excluded all slices that had no neurites or neurites less than 100 µm in length for quality standard reasons, we had a sample number of only 3 in the “no µCP” group (Fig. 1H). Collagen-only µCPs without any additional load but PBS already promoted nerve fiber growth along the microcontact-printed lanes when stained for NF (Fig. 1H). When pORN was loaded to the collagen-based µCP, the number of NF-positive outgrown neurites was significantly enhanced compared to collagen-only µCPs (Fig. 1H). The length of these outgrown NF-positive nerve fibers was also significantly longer when pORN was applied to the collagen-based µCP (Fig. 1H).

NF was used to stain and quantify outgrown neurites. NF effectively stained nerve fibers, demonstrating their growth along the collagen-based microcontact-printed lanes (Fig. 2A–C). To further confirm the specificity of the neurites, cultures were stained for MAP2, MOG, and GFAP. MAP2-positive nerve fibers were identified growing along the collagen-based microcontact-printed lanes (Fig. 2D). MOG-positive staining was observed along the collagen-based microcontact-printed lanes, however, the MOG staining appeared to be patchier (Fig. 2E). GFAP was utilized to visualize astrocytes, which were exclusively found in the brain slices, with some processes rearranging and extending towards the microcontact-printed lanes (Fig. 2F).

Fig. 2.

Characterization of neurite outgrowth. (A–C) Neurofilament-heavy (NF-H)-positive neurites (green, Alexa Fluor 488) extended beyond the brain slice along the microcontact print (µCP) pattern (red, Alexa Fluor 546). (D) Microtubule-associated protein 2 (MAP2) and (E) myelin oligodendrocyte glycoprotein (MOG) also stains outgrown neurites (arrows in D and E, dotted line marks the brain slice boarder, artificially overlaid red µCP). (F) Glial fibrillary acidic protein (GFAP)-positive astrocytes primarily remain within the brain slice but organize themselves along microcontact-printed lanes, too. Nuclei are stained with DAPI (blue). Scale bar in A–C = 100 µm, in D–E = 75 µm, in F = 120 µm.

To examine the effect of A$\beta$ on nerve fiber growth, human A$\beta$(42) was loaded to the collagen-based µCPs. Our data indicate that recombinant human monomeric A$\beta$(42) significantly enhanced the number of nerve fiber growth but not the length (Fig. 3A). For a control, aggregated and reversed forms of A$\beta$(42) were loaded to the collagen-based µCPs. Unlike monomeric A$\beta$(42), these forms showed no notable impact on neurite outgrowth. Co-loading pORN with A$\beta$(42) or the aggTau P301S had no effect on the number of NF-positive outgrown neurites (Fig. 1I). However, the combined load of pORN and tau P301S resulted in significant shorter outgrown neurites, as quantified by NF staining (Fig. 1I). Western blot analysis confirmed monomeric A$\beta$(42) presence at 4 kDa, aggregated A$\beta$(42) was evident by the smear at 40 kDa, while reverse A$\beta$(42) was undetected, as anticipated (Fig. 3B). No significant different neurite outgrowth was observed along collagen-based µCPs loaded with tau WT or with aggTau P301S mutated (Fig. 3C). In Western blot analysis, tau WT was identified at the expected 60 kDa size, whereas aggTau P301S mutated appeared as bands surpassing the 60 kDa mark in molecular weight, confirming their aggregated state (Fig. 3D).

Fig. 3.

Effects of human beta-amyloid (A$\beta$) and human tau on neurite outgrowth. Hippocampal vibrosections (150 µm, half slices) of postnatal day 8–10 C57BL/6 wild-type (WT) mice were cultured for 2 weeks on collagen-based microcontact prints (µCPs). Sections were then fixed and immunohistochemically stained for neurofilament (NF) using Alexa Fluor 488 and DAPI. The number of neurites and the length of processes were measured along 300 µm in width of representative microcontact-printed lanes using NeuronJ. (A) The number and length of NF-positive outgrown neurites per mm µCP loaded with human A$\beta$(42) monomeric (mon). As a control, empty collagen-based µCPs (PBS), µCPs loaded with human A$\beta$(42) reversed (rev), human A$\beta$(42) aggregated (agg) are added. (B) In Western blot analysis, A$\beta$ antibody clone 6E10 detects the mon A$\beta$42 (1 µg) at the expected size of 4 kDa and the agg A$\beta$42 (750 ng) is evident by the smear of peptides at 40 kDa. Rev A$\beta$42 (1 µg) peptide is not detectable. (C) Quantified number and length of all outgrown neurites along collagen-based µCPs loaded with human WT tau (Tau WT) and human P301S mutated aggregated tau (Tau P301S). (D) In Western blot analysis, Tau-5 antibody detects the human tau WT (1 µg) at the expected 60 kDa size. Tau P301S appeared as bands with molecular weights $>$60 kDa, verifying aggregated forms. Values (mean $\pm$ SEM) give the outgrowth larger than the spontaneous outgrowth ($>$100 µm) per mm of microcontact-printed area. Statistical significance was evaluated by one-way ANOVA with Fisher’s LSD posthoc test for multiple comparisons of means (* p 0.05).

Confocal imaging revealed several long intact nerve fibers running in parallel bundles after two weeks of incubation on µCPs (Fig. 4A,B). Long NF-positive nerve fibers formed exclusively along the microcontact-printed lanes (Fig. 4C).

Fig. 4.

Confocal imaging of neurites outgrown along microcontact-printed lanes. (A) Confocal zoomed-in image of several neurofilament (NF)-positive long side-by-side nerve fibers (green, Alexa Fluor 488) in three-dimensional representation (processed with the Imaris 10.0.1 software). (B) Associated nuclei (blue, DAPI) (C) NF-positive nerve fibers (green, Alexa Fluor 488) form exclusively within the two microcontact-printed lanes (µCP, red, Alexa Fluor 546). Scale bar in A–C = 40 µm.

To investigate the functional activity of the newly formed nerve fibers, MiniRuby crystals were introduced to the µCPs close to the brain slices (Fig. 5A). The MiniRuby appeared as a large red-fluorescent dot and was transported to the brain slice after 1 day of incubation, though not in direct contact with the brain slice (Fig. 5B). A clear red-fluorescent MiniRuby-positive area was observed at the green-fluorescent collagen-based µCPs (Fig. 5C). Some MiniRuby-positive immunoreactive spots were visible directly along the NF-positive nerve fibers on the µCP (Fig. 5D–F). Interestingly, after 7 days of incubation with MiniRuby, MiniRuby-positive cells were identified in the hippocampus (Fig. 5G–I).

Fig. 5.

Live cell imaging using the neurotracer MiniRuby. (A) Schematic representation of the MiniRuby (mR) neurotracer placement on a green microcontact print (µCP) connected to a brain half slice (grey). (B) MiniRuby (mR, red-fluorescent) is retrogradely transported to the nuclei-stained brain slice (blue, DAPI), though not in direct contact to the brain slice. (C) MiniRuby (mR, red-fluorescent) is well positioned on the µCP (green, Alexa Fluor 488). (D–F) When fixed sections are stained for neurofilament (NF) revealing outgrown neurites (green, Alexa Fluor 488), MiniRuby (mR, red-fluorescent) is taken up by the neurons (arrows) and transported retrogradely to the brain slice (blue, DAPI). (G–I) Within the brain slice, MiniRuby (mR, red-fluorescent) specifically stained neurons in the hippocampal area, which were not visible in the green channel. Scale bar in B,C = 100 µm, in D–I = 75 µm.

Voltage-sensitive dye FluoVolt alters its spectral properties in response to membrane potential shifts, while calcium-sensitive dye Rhod-4 changes its spectral properties upon binding intracellular calcium ions (Ca${}^{2+}$). Upon stimulation (70 mM KCl), FluoVolt displayed a slow response over 50 seconds, exhibiting enhanced fluorescence (Fig. 6A–C). Typically, the magnitude of fluorescence increases by 1% per mV. Conversely, Rhod-4 indicates intracellular calcium increases (Fig. 6D–F, blue arrows), decreases (Fig. 6D–F, yellow arrow) and fluctuations (Fig. 6D–F, green arrow), within a 240-second observation period following stimulation (70 mM KCl).

Fig. 6.

Voltage-sensitive (FluoVolt) and calcium-sensitive (Rhod-4) dyes to monitor the electrical activity in neurons. Two-week-old organotypic brain slices, incubated on collagen-based microcontact prints (µCPs) loaded with monomeric beta-amyloid (A$\beta$(42)), were treated with voltage-sensitive FluoVolt or calcium-sensitive Rhod-4. Following depolarization with isotonic 70 mM KCl solution, samples were video recorded for 2–5 min. (A) Outgrown neurites loaded with voltage-sensitive FluoVolt exhibited progressive activation (blue arrows) over 15 (B) and 50 seconds (C). (D–F) Samples loaded with Rhod-4 displayed that intracellular calcium levels increased (blue arrows), decreased (yellow arrow), and fluctuated (green arrow) over a 240-second period. The dashed line indicates the half slice border. Scale bar in A–F = 150 µm.

The presence of the membrane bound Na,K-ATPase serves as an indicator of neuronal activity. Neurites extending from organotypic brain slices cultured for 2 weeks on collagen-based µCPs loaded with A$\beta$(42) showed pronounced membrane-specific Na,K-ATPase (ATP1A1) staining (Fig. 7A). After depolarization with an isotonic 70 mM KCl solution and fixation after 1 hour, outgrown neurites demonstrated increased Na,K-ATPase expression (Fig. 7B). Na,K-ATPase expression returned to baseline levels when fixation was delayed until 24 hours after depolarization (Fig. 7C). However, no notable differences in Na,K-ATPase immunoreactivity were observed among neurites along collagen-based µCPs, whether unloaded (PBS) or loaded with monomeric or aggregated A$\beta$(42) (Fig. 7D).

Fig. 7.

Expression levels of Na,K-ATPase (ATP1A1) and SIRT1 showed no difference between collagen-based microcontact prints (µCPs) devoid of load (PBS) and those loaded with monomeric or aggregated beta-amyloid (A$\beta$(42)). (A) Neurites grown from organotypic brain slices (incubated for 2 weeks on collagen-based µCPs loaded with A$\beta$(42)) exhibited positive staining for Na,K-ATPase. (B) Following depolarization with an isotonic 70 mM KCl solution and fixation after 1 hour (Stim. + 1 h), outgrown neurites displayed strong Na,K-ATPase expression. (C) Increased Na,K-ATPase expression recovered 24 hours after stimulation (Stim. + 24 h). (D) Optical density (OD) measurements revealed that there was no significant difference in Na,K-ATPase immunoreactivity among neurites along collagen-based µCPs without load (PBS), monomeric, or aggregated A$\beta$(42). (E–G) Cells at the half slice border exhibited nuclear SIRT1 expression (arrows). The dashed line indicates the half slice border. (H) OD measurements of nuclear SIRT1 immunoreactivity remained consistent across collagen-based µCPs without load (PBS), monomeric, or aggregated A$\beta$(42). Values are given as mean $\pm$ SEM. Scale bar in A–C = 15 µm, in E–G = 30 µm.

Finally, an immunostaining for SIRT1 was conducted, which is an intracellular regulatory protein related to aging and age-associated diseases and may be involved in A$\beta$(42) pathology. Cells located at the half slice border clearly displayed nuclear SIRT1 expression (Fig. 7E–G). However, nuclear SIRT1 immunoreactivity remained consistent across collagen-based µCPs, irrespective of their loading status with PBS, monomeric, or aggregated A$\beta$(42) (Fig. 7H).

In our research group, organotypic brain slice cultures are well-established and have been extensively utilized, with a record of more than 60 publications, including a comprehensive review by Humpel [16]. Recently, we introduced and modified the microcontact printing technique in our lab to immobilize bioactive factors in a µm-size defined pattern, which we connected to brain slices [18]. We have successfully demonstrated the microcontact printing efficiency on various aspects, such as microglia [29], vascular structures [30], and NFG-induced nerve fibers [31]. This experience instills confidence in the robust viability of our brain slices cultured on µCPs for two weeks. During this culture period, the brain slice underwent flattening and attained transparence, serving as a reliable indicator of high quality brain slices. Additionally, a viability assay using PI reinforces the excellent brain quality and survival of the brain cells. Finally, positive immunostainings for neuronal markers unequivocally illustrates the survival of neurons in our experimental setting.

NFs are neuron-specific 10-nm filaments recognized as members of the intermediate filament family. They exhibit a relatively sparse and convoluted distribution within dendrites and perikarya. Within axons, however, they tend to be numerous, predominantly straight, unbranched, and extending to considerable length [32]. NFs are predominantly composed of three subunits, distinguished by their apparent molecular weight: NF-H (heavy, ~200 kDa), NF-M (medium, ~160 kDa), and NF-L (light, ~68 kDa) protein [32]. In our present study, we employed a commercial NF-H monoclonal antibody [33], which exhibited robust staining of nerve fibers derived from mouse slices. Especially in the confocal microscopy, this antibody distinctly highlighted nerve fibers that extended from brain slices along the collagen-based µCPs. When axons undergo radial expansion, often increasing their size by up to 10-fold, this is accompanied by a proportional increase in the number of NFs [34]. During the onset of axonal growth, NF-H expression is typically delayed compared to NF-L and NF-M [35], suggesting that positive NF-H reactivity may signify a more mature state of NFs. NF-H and NF-M stand out due to their longer tail domains that extend radially from the filament core, forming side arms. Phosphorylation of these side arms further enhances the stability of the axonal NF cytoskeleton which extensively cross-links to actin filaments and microtubules [32].

To reinforce our findings, we employed another neuronal marker, MAP2, confirming the neuronal nature. Additionally, we applied the oligodendrocyte marker MOG revealing a punctuated staining, indicative of potential myelin sheets on the emerging nerve fibers. A GFAP staining displayed reactive astrocytes only within the brain slices but not along the µCPs. Subsequently, confocal microscopy was conducted, revealing numerous long, intact nerve fibers organized in parallel bundles with a tendency toward the borders of microcontact-printed lanes.

Collagen is widely employed in numerous cell culture models due to its bioactive and biodegradable properties. At present, collagen is the foremost extracellular matrix protein employed in three-dimensional cell cultures [36]. Over the course of several years, we have utilized collagen in our experimental approaches, as elucidated in previous reviews [20, 37]. Our proficiency extends to the construction of collagen hydrogels [38, 39] or collagen-based µCPs [18] incorporated in organotypic brain slices with preserved three-dimensional neuronal architecture and microenvironment. Despite collagen being found only in basement membrane of the vasculature in the brain, it is also widely utilized as a biomaterial to promote the growth of neurites [21, 22]. Our current findings clearly demonstrate the capacity of collagen since brain slice-derived nerve fibers selectively extended and aligned along the µCP composed solely of collagen. Neurite outgrowth quantity is known to be influenced by collagen concentration through altering its fibril density and pore size [40]. An optimal concentration of 2 mg/mL collagen was identified [40], aligning with the end concentration achieved in our µCPs, which promotes neuronal differentiation and outgrowth of neurites. Neurons encapsulated in a collagen hydrogel containing decellularized brain extracellular matrix facilitated neurite development and enhanced neurite outgrowth [41]. More experiments are necessary to assess the role of collagen, which was out of focus in this study.

Compelling evidence from existing literature underscores an influence of polyamines in supporting mediated cell adhesion and nerve fiber growth due to their intrinsic positive charges [42]. In particular, the capacity of microcontact-printed polylysine in guiding axons has been extensively investigated [23, 24, 25, 43, 44, 45]. In the present study, we aimed to examine the impact of another polyamine, pORN, on neurite outgrowth. Notably, pORN and polylysine share an almost identical structure, differing only by a single atom in the length of their primary amino group carbon chain, making them functionally equivalent regarding neuron adhesion, viability, and pattern compliance [24]. The non-proteinogenic amino acid ornithine plays a crucial role in synthesizing various essential cellular products, making L-ornithine a widely utilized supplement in cell culture media [46] and pORN a widely utilized coating reagent for neuronal cell lines [26]. Our present study revealed that pORN loaded to the collagen-based µCPs significantly enhanced nerve fiber growth compared to collagen-only µCPs, in terms of number of fibers as well as fiber length. However, pORN had no additive impact on either the effects of A$\beta$(42) or aggTau. Despite some literature suggesting potential cytotoxicity associated with pORN, our study dismisses this concern, as pORN was employed at a very low concentration and showed effects similar to collagen [47].

A$\beta$ is primarily known for its involvement in AD forming extracellular plaque deposits. Different lengths of A$\beta$ peptides arise from the proteolytic cleavage of the amyloid precursor protein (APP) by the sequential action of two proteases, the ß-secretase (BACE1) and $\gamma$-secretase. The imprecise cleavage of $\gamma$-secretase at the C-terminus results in the two major isoforms A$\beta$(40) and A$\beta$(42). The 40-amino acid A$\beta$ variant displays higher solubility and is detectable in the walls of cerebral arterioles, capillaries and in the blood, potentially released from thrombocytes [48, 49]. In contrast, the larger 42-amino acid peptide has a higher propensity to form oligomers, which aggregate and accumulate in extracellular plaques. However, the physiological role of the evolutionary highly conserved A$\beta$ is not fully understood and remains a topic of ongoing research. To date, the putative roles of A$\beta$ are numerous [50], including neuroprotective effects [51] and the facilitation of synaptic plasticity due to enhancing long-term potentiation in the hippocampus [52]. Specifically, low (physiological) doses of A$\beta$ enhance memory, while high (pathological) concentrations inhibit memory function [4, 5, 53]. Soluble oligomeric A$\beta$ at physiological levels promotes dendritic complexity in hippocampal neurons, indicating that dendritic stability and dynamics are modulated by A$\beta$ [6]. However, although amyloid exposure initially stimulates neurogenesis, it fails to complete the maturation cascade [7]. Other studies suggest that A$\beta$ peptide toxicity is associated with active fibrillization processes [54]. Non-soluble A$\beta$ aggregates negatively affect the morphology of neurites and cell viability [8].

Our present findings reveal a twofold stronger stimulatory effect of human A$\beta$(42) on the formation of nerve fibers compared to solely collagen when detected by NF antibodies. In the present investigation, emphasis was placed on the larger A$\beta$(42) peptide, as it is postulated to exert a more potential impact on both pathological and physiological aspects within the brain. Notably, the human A$\beta$(42) differs from its murine counterpart by three amino acids located at the residues 5, 10 and 13 [55]. Due to this fact, human A$\beta$(42) tends to have a higher aggregation in comparison to murine A$\beta$(42). In our laboratory, human A$\beta$(42) has been extensively used to elucidate its impact on vascular support [30] and their spreading across the brain [27]. As a negative control, the collagen-based µCPs were loaded with both aggregated and reversed forms of human A$\beta$(42). In contrast to monomeric A$\beta$(42), these variations did not affect neurite outgrowth, maintaining consistency with the collagen-only µCPs. Western blot analysis revealed the presence of monomeric A$\beta$(42) at 4 kDa, while the aggregated form manifested as a smear at 40 kDa, as expected. The reversed form of A$\beta$(42) remained undetected in Western blot analysis, aligning with our expectations. Unfortunately, microcontact printing of pure A$\beta$ on semipermeable membranes is unfeasible, since collagen provides a scaffold to hold the small A$\beta$ peptides on the membrane. Co-loading A$\beta$(42) with pORN eliminated the twofold higher effect, reducing it to a level close to the baseline effect observed with collagen alone. pORN may interfere with the signaling pathways or mechanisms through which A$\beta$(42) exerts its effects, or the intense exposure to both stimuli might induce desensitization, potentially diminishing the observed effects. Collectively, our data strongly imply that the human form of monomeric A$\beta$(42) exerts a significant and positive influence on neurites, supporting the hypothesis of A$\beta$ being neuroprotective and enhancing synaptic plasticity, learning as well as memory.

Sirtuins represent a highly conserved class of nicotinamide adenine dinucleotide (NAD${}^{+}$)-dependent enzymes, primarily functioning as deacetylases [56]. Identified as seven variants (SIRT1-SIRT7), they are commonly associated with caloric restriction and aging via modulating energy metabolism, genomic stability and stress resistance [57, 58]. SIRT1, the most extensively studied, is expressed ubiquitously in the brain, predominantly within neurons and detectable in both the nucleus and cytoplasm [59]. Studies have linked aging to an increase in SIRT1 expression levels. While aging correlates with heightened SIRT1 expression, SIRT1 activity tends to decline [60], consistent with the age-related decline in NAD${}^{+}$ levels [61]. Reactivating SIRT1 could potentially mitigate AD development [62]. The potential of SIRT1 in combating AD pathology operates through diverse pathways: decreasing A$\beta$ production, reducing amyloid plaque accumulation, promoting neurogenesis, mitigating neuroinflammation, and preventing mitochondrial dysfunction [56, 59, 62].

A recent study demonstrated that SIRT1 promotes the degradation of oligomeric A$\beta$ in primary astrocytes, with endogenous SIRT1 levels increasing, following prolonged A$\beta$ treatment [63]. To investigate the role of SIRT1 in neurite outgrowth along A$\beta$(42)-loaded µCPs, we conducted immunostainings after a 2-week A$\beta$(42)-loaded µCP incubation to assess potential changes in SIRT1 expression. However, nuclear SIRT1 immunoreactivity remained unchanged among collagen-based µCPs, regardless of whether they were loaded with PBS, monomeric, or aggregated A$\beta$(42). Nonetheless, neurons positioned at the boundary of the half slice exhibited notable nuclear SIRT1 expression, highlighting the importance of SIRT1 on neuronal complexity. This observation is consistent with previous research indicating the crucial involvement of SIRT1 in synaptic plasticity and normal cognitive function [64].

The microtubule-associated protein tau exemplifies a highly dynamic protein due to posttranslational modifications at more than 50 sites, which comes at a heightened propensity for self-aggregation [65, 66]. The appearance of intracellular tau NFTs mainly composed of such aggregated hyperphosphorylated tau protein is the second defining feature of AD. Tau is encoded by the MAPT gene, comprising 16 exons that generate six alternative splicing isoforms ranging from 351 to 441 amino acids [67]. Tau protein is distinctly divided into the N-terminal part, proline-rich region (PRR), microtubule-binding domain (MTBD), and C-terminus [67]. Each contains either 3 or 4 MTBD repeats (3R or 4R) which have been shown to be essential for the ability of tau to bind to microtubules [67]. However, experiments performed by several other groups revealed that tau serves functions beyond merely stabilizing the microtubules, notably regulation of neuronal network activity [67]. In the present study, we microcontact-printed a human tau variant comprising 441 amino acids, characterized by the formation of active aggregates resulting from a mutation at position 301 (P301S). We previously conducted experiments to assess the spreading of P301S aggTau in brain slices [68]. Although tau primarily functions as a cytosolic microtubule-associated protein, it is also found in extracellular environments under physiological conditions [69]. It appears that tau can be transferred from cell-to-cell through various pathways, including vesicular mechanisms (exosomes, ectosomes, endosomes, cell membrane fusion), vesicle-free mechanisms (direct translocation across the plasma membrane known as unconventional protein secretion I) or tunneling nanotubes [69]. Human tau was employed due to potential differences in tau expression between mice and humans [70]. Although the tau sequences in mice and humans exhibit an 89% amino acid similarity, the human tau variant features an additional 11 amino acids at the N-terminal end [70].

Microcontact printing of aggTau within a collagen matrix did not lead to an increase in outgrown nerve fibers derived from brain slices. For control purposes, collagen-based µCPs were loaded with WT tau, resulting in neurite outgrowth similar to that of baseline collagen-only µCPs. In Western blot analysis, we detected WT tau at the anticipated 60 kDa position, while the P301S mutated aggTau appeared as several bands exceeding the 60 kDa mark in molecular weight, confirming their aggregated state. Upon concurrent co-loading of aggTau and pORN, we observed a synergistic effect in the number of outgrown neurites, although this effect did not reach statistical significance. However, the combined load of aggTau and pORN led to a significant decrease in the length of outgrown neurites. Since aggTau may be bioinactive, it could interfere with normal cellular processes of tau necessary for neurite elongation. Several studies indicate that tau is essential in modulating synaptic plasticity and signaling [71, 72, 73, 74]. Tau mediates changes in the cytoskeleton by binding actin to its PRR, concurrently binding to microtubules through MTBD [75]. Acting as a cross-linker between microtubules and actin filaments, tau contributes to the dynamic reorganization of the cytoskeleton network, potentially playing a crucial role in neurite formation. In fact, several studies have reported the involvement of tau in neurite outgrowth, however, the specific details regarding the implication of tau in axonal guidance remain unclear [9, 10, 11].

In our present study, we show the growth of brain slice-derived nerve fibers along microcontact-printed lanes, with the most pronounced effects observed when A$\beta$(42) was loaded. Additionally, we sought to ascertain the functional activity of these newly formed nerve fibers and their capability for retrograde substance transport. To address this, we employed the fluorescent neurotracer MiniRuby, previously used to demonstrate retrograde transport in dopaminergic meso-striatal neurons [76]. Small MiniRuby crystals were placed in proximity to the brain slice, likely to come into contact with the emerging nerve fibers. After 24 h incubation, we observed small red-fluorescent dots (presumably MiniRuby) aligning with green NF stained nerve fibers along the µCPs. This observation strongly supports the functional activity of the newly formed nerve fibers. Subsequently, we investigated whether MiniRuby could be transported into the brain slice, revealing several red-fluorescent cells within the hippocampus. This finding strongly suggests that nerve fibers uptake MiniRuby outside the brain slice and transport it deep into the brain tissue.

The voltage-sensitive dye FluoVolt indicates changes in its spectral properties upon membrane potential shifts. At resting potential (–65 mV) of neurons, there is a higher concentration of K${}^{+}$ ions inside the neuron compared to Na${}^{+}$ ions outside the cell, which is maintained by Na,K-ATPase activity. In our present study, we applied an isotonic 70 mM KCl stimulation to two-week-old brain slices, following the manufacturer’s recommendations. As the extracellular potassium concentration increases, the potassium gradient diminishes, leading to a more positive potassium equilibrium potential. Concurrently, external Na${}^{+}$ ions continue to leak into the soma, causing global depolarization to occur (+40 mV). In the slow-response action, FluoVolt enters depolarized cells, binding to proteins or membranes, and exhibiting enhanced fluorescence. However, this membrane translocation event diminishes the ability of these reporters to react to membrane potential changes and introduces a capacitive load that can impact cell health. Nevertheless, FluoVolt demonstrates a substantial response magnitude, typically in the range of 1% per mV. In the present study, we observed proof-of-principle that such a slow-response of FluoVolt occurs in newly outgrown neurites up to 50 seconds after stimulation, strongly indicating the neuronal physiological activity of the newly formed nerve fibers.

The calcium dye Rhod-4 enters the cells through passive loading, and once inside, intracellular esterases cleave the dye to its cell-impermeant, active form, which exhibits fluorescence upon binding with calcium ions. The concentration of calcium ions (Ca${}^{2+}$) outside the cell is typically thousands of times higher than inside. When the cell membrane depolarizes, voltage-gated calcium channels (VGCCs) open, enabling calcium ion influx into the cytoplasm. Depolarization can also induce the release of calcium ions from intracellular stores like the endoplasmic reticulum through calcium-induced calcium release. In our present study, Rhod-4 reveals fluctuations in intracellular calcium levels, including both increases and decreases, during a 240-second observation post-stimulation with isotonic 70 mM KCl. These increases signify depolarization events, while decreases may indicate repolarization or cell death. Again, we provide proof-of-principle that the newly formed nerve fibers respond with a calcium activation.

To demonstrate that depolarization is linked to the activity of axonal plasma membrane-associated Na,K-ATPase [77, 78], brain slices were immunohistochemically stained for Na,K-ATPase 1 hour or 24 hours after depolarization. Neurites indeed showed increased Na,K-ATPase expression following depolarization compared to non-depolarized brain slices. This increased expression returned to baseline levels 24 hours post-stimulation. However, there was no significant difference in Na,K-ATPase immunoreactivity among neurites along collagen-based µCPs without load (PBS), monomeric, or aggregated A$\beta$(42).

This study provides a proof-of-concept demonstrating the utility of collagen-based microcontact prints as an extracellular matrix for nerve fiber growth. However, certain limitations must be acknowledged. (i) Firstly, our previous studies have revealed inconsistent microcontact printing, variable rates of collagen degradation in culture and variability in organotypic brain slice culture. This potentially contributed to increased experimental variance, necessitating the implementation of exclusion criteria (i.e., fibers shorter than 100 µm were excluded) for quantitative analysis. (ii) Secondly, our current model lacks the ability to print a concentration gradient of different biomolecules, which could potentially improve the model. (iii) Thirdly, we exclusively tested the effects of human A$\beta$(42), the most potent toxic peptide causing plaques, on nerve fiber outgrowth. The exploration of other A$\beta$ variants such as human A$\beta$(40), or mouse A$\beta$ could provide further insights. (iv) While the use of postnatal day 8–10 brain slices is widespread due to enhanced cellular survival, it may not fully capture the complexity of neurodegenerative diseases manifesting in aging adults. Conducting long-term studies to investigate the effects of A$\beta$ in adult WT brain slices would provide valuable insights into the aging-related dynamics of AD. (v) Lastly, the translatability of our findings to humans remains uncertain, although we used human-homolog factors (A$\beta$(42), tau). Further investigations using human postmortem brain tissue cultures may complement our assays.

In this study, we describe an ex vivo model that enables the study of selective neurite growth affected by exogenous stimuli. This model extends our previous research, which focused on microglial migration [29] and endothelial cell migration up to vessel formation [30]. Utilizing microcontact printing in combination with organotypic mouse brain slices, we envision the development of a “triple model”, where the effects of exogenous factors can be concurrently investigated across three distinct functional assays: neurons, microglia, and vessels. This integrated approach could be further refined into a “brain-on-a-chip” functional assay, serving as a valuable platform for screening various bioactive factors, novel drugs, or medications. Such a model holds promise in contributing to the principles of the 3Rs in animal research by minimizing the number of animals used and alleviating associated pain.