Electroconductive Collagen‐Carbon Nanodots Nanocomposite Elicits Neurite Outgrowth, Supports Neurogenic Differentiation and Accelerates Electrophysiological Maturation of Neural Progenitor Spheroids

Abstract

Neuronal disorders are characterized by the loss of functional neurons and disrupted neuroanatomical connectivity, severely impacting the quality of life of patients. This study investigates a novel electroconductive nanocomposite consisting of glycine‐derived carbon nanodots (GlyCNDs) incorporated into a collagen matrix and validates its beneficial physicochemical and electro‐active cueing to relevant cells. To this end, this work employs mouse induced pluripotent stem cell (iPSC)‐derived neural progenitor (NP) spheroids. The findings reveal that the nanocomposite markedly augmented neuronal differentiation in NP spheroids and stimulate neuritogenesis. In addition, this work demonstrates that the biomaterial‐driven enhancements of the cellular response ultimately contribute to the development of highly integrated and functional neural networks. Lastly, acute dizocilpine (MK‐801) treatment provides new evidence for a direct interaction between collagen‐bound GlyCNDs and postsynaptic N‐methyl‐D‐aspartate (NMDA) receptors, thereby suggesting a potential mechanism underlying the observed cellular events. In summary, the findings establish a foundation for the development of a new nanocomposite resulting from the integration of carbon nanomaterials within a clinically approved hydrogel, toward an effective biomaterial‐based strategy for addressing neuronal disorders by restoring damaged/lost neurons and supporting the reestablishment of neuroanatomical connectivity.

A novel electroconductive collagen/carbon nanodots nanocomposite supports robust neurite outgrowth, directs neuronal differentiation without exogenous factors, and elicits the electrophysiological maturation of induced pluripotent stem cells (iPSCs)‐derived neural progenitor spheroids through enhanced firing and more active network activity.

1. Introduction

Neuronal disorders, including neurodegenerative conditions such as Alzheimer's and Parkinson's disease, as well as traumatic injuries and stroke, are associated with neuronal loss and/or the deterioration of neuroanatomical connectivity, severely affecting the quality of life of patients. The first attempts to regenerate lost neuronal functions date back to the 1970s and explored the transplantation of primary dopaminergic neurons sourced from human embryos.^{[ 1} , ² , ^{3 ]} However, inconsistent protocols, excessive dopamine transmission, low availability of fetal tissues, and ethical concerns hindered their clinical success,^{[ 4 ]} inspiring researchers to turn to emerging alternatives for scalable and traceable sources, such as embryonic stem cells and iPSCs. The well‐established protocols and the efficacy in preclinical models have made stem cells an excellent candidate for cell‐based therapies,^{[ 2 ]} but low (5–10%) survival rates, inadequate differentiation into the desired phenotype, and adverse host immune responses, have negatively affected the therapeutic outcome.^{[ 1} , ² , ³ , ^{4 ]} To address these limitations, researchers have turned to original approaches that synergistically integrate concepts of regenerative medicine with biomaterial‐based neural tissue engineering.^{[ 5} , ^{6 ]} One of the fundamental design criteria for effective clinical translation emphasizes the use of biomaterials that recapitulate key physicochemical features of the intended target tissue to recreate a biomimetic microenvironment that ultimately enhances cell survival, guides differentiation, and promotes functional integration within the host tissue. Among the material properties deemed fundamental to achieve neuroregenerative abilities (e.g., biocompatibility, native tissue‐like mechanical properties, and suitable degradation rate),^{[ 5} , ^{7 ]} recent findings have highlighted the pivotal role of electroconductivity in guiding several developmental processes associated with cell proliferation and synaptic plasticity.^{[ 8} , ⁹ , ^{10 ]} Based on this evidence, the use of electroconductive hydrogels has emerged as an effective approach to recapitulate the physicochemical and electrical microenvironment of neural tissues.^{[ 11} , ^{12 ]} In fact, while widely used hydrogels (e.g., collagen, alginate, and gelatin) offer the structural basis for a cell‐instructive microenvironment that mimics the physicochemical properties of native extracellular matrix (ECM) to enhance the adhesion, growth, and differentiation of neuronal cell populations, they are characterized by a low electrical conductivity.^{[ 13} , ¹⁴ , ^{15 ]} One of the most investigated strategies to integrate this additional functionality is the incorporation of metal‐ or carbon‐based nanomaterials.^{[ 11} , ¹² , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ^{18 ]} In this context, nanocomposites consisting of nanomaterials such as carbon nanotubes (CNTs) and graphene nanoparticles dispersed within a hydrogel matrix have been shown to promote highly desirable cellular effects, such as neuronal differentiation, axonal elongation and network formation. In particular, the exceptional electrical conductivity of CNTs and their capacity to integrate with the ECM led to enhanced electrical signaling among neurons, which is vital for the formation of functional neural networks.^{[ 17} , ¹⁹ , ^{20 ]} Likewise, nanomaterials including graphene oxide and reduced graphene oxide offer a conducive environment for cell adhesion and neurite expansion, in addition to modulating the expression of neuronal markers.^{[ 21} , ²² , ^{23 ]} In this context, previous work has addressed the controversial argument of cytotoxicity caused by CNTs and graphene nanomaterials, ultimately highlighting the importance of (i) carefully evaluating factors such as oxidative stresses, DNA damage and severe inflammatory response in the design of carbon nanomaterials and (ii) employing advanced in vitro testing platforms that mimic the in vivo microenvironment for more accurate and reliable safety data extrapolation.^{[ 24} , ²⁵ , ²⁶ , ^{27 ]} Notably, the extensive body of literature which investigated the potential of these candidate electroconductive nanocomposites for neuroregeneration, mostly employed simplified experimental conditions (e.g., conventional monolayer cultures) that do not recapitulate the native ECM and/or adopted primary immortalized cells (e.g., PC12 and SH‐SY5Y) that do not exhibit the cellular phenotype of mature differentiated neurons.^{[ 16} , ²⁸ , ²⁹ , ^{30 ]} It is well known that for central nervous system neurons, the two‐dimensional cellular microenvironment associated with cell monolayer cultures leads to aberrant cell–cell contacts and network formation, unrealistically flattens soma and growth cones and limits axon‐dendrite outgrowth.^{[ 31} , ³² , ^{33 ]} Despite the advancements in preclinical technologies (e.g., bioprinting) and approaches which enable the creation of highly controlled in vitro models and cellular constructs,^{[ 34} , ³⁵ , ^{36 ]} most reported studies in this field have employed culture systems that do not fully capture the physiological complexity and architecture of the native nervous tissue.^{[ 37} , ³⁸ , ^{39 ]} Taken together, these observations reaffirm the importance of selecting physiologically accurate in vitro testing platforms and relevant cell models, as an important early step for the development of novel neuroregenerative biomaterials and the investigation of their therapeutic potential.

In this context, we have developed a novel electroconductive nanocomposite consisting of a collagen type I matrix functionalized with carbon nanodots fabricated from glycine precursors (GlyCNDs). The rationale behind this novel strategy is that collagen has already shown great promise as a therapeutic solution for central nervous system injuries and degeneration.^{[ 5} , ³³ , ^{40 ]} When used in its hydrogel form, collagen provides unique physicochemical and biological cues that support cellular adhesion, growth and proliferation. It is also a biopolymer approved by the FDA for clinical testing in neural tissue engineering and several collagen‐based products are already commercially available.^{[ 7 ]} However, collagen's fast degradation rate in physiological conditions represents a factor that cannot be overlooked.^{[ 41 ]} In addition, native collagen cannot recapitulate the electroactive environment of nervous tissues. In parallel, the choice of CNDs, a relatively new type of carbon‐based nanoparticles (1–10 nm in diameter), is supported by the fact that they offer diverse physicochemical properties and advantageous characteristics such as biocompatibility, low cytotoxicity, electrical conductivity, ease of synthesis, abundant functional groups (e.g., amino, hydroxyl, carboxyl), and high physicochemical stability.^{[ 42} , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ , ⁴⁷ , ^{48 ]} CNDs can be synthesized from a wide variety of carbon‐containing precursors all of which dictate the resultant surface functional groups.^{[ 49 ]} Because they are a relatively new class of nanoparticles, the potential of CNDs in the context of neural tissue engineering is only beginning to be explored.

In this work, we carry out a comprehensive physicochemical characterization of the collagen‐GlyCND nanocomposite and evaluate its biological impact by employing both mouse iPSC‐derived NP spheroids (and 3D cultures of primary cortical neurons as a qualitative confirmatory cell model), specifically selected to recreate a more tissue‐like cellular microenvironment. In particular, spheroids (hereafter also referred to as neurospheres) offer several advantages over conventional culture systems, making them a more physiologically relevant model for studying neural tissue engineering. Firstly, spheroids mimic the three‐dimensional (3D) architecture of the ECM, thereby enabling more accurate cell–cell and cell–ECM interactions, which are essential to investigate cellular behavior and functions.^{[ 50 ]} Secondly, spheroid cultures better recapitulate the spatial organization and cellular heterogeneity of native tissues, providing a more realistic environment for studying neuronal differentiation and network formation.^{[ 50} , ^{51 ]} Additionally, spheroids exhibit nutrient and oxygen gradients, closely resembling in vivo conditions.^{[ 52 ]}

From a morphological point of view, significantly more neurite sprouting occurs in both spheroids and primary neurons embedded in the nanocomposite when compared to collagen and collagen with suspended GlyCNDs, showing a significantly higher branching tendency. In addition, the nanocomposite positively enhances additional functions, such as neuronal differentiation and electrophysiological maturation. In particular, the significantly lower number of cells positive for Ki‐67, a nuclear proliferation marker, the lower expression of nestin (proliferating neural precursor marker) coupled with higher expression of both β‐III‐tubulin, and MAP2 (early and mature neuronal marker), suggest that the nanocomposite accelerates neurodifferentiation of NP spheroids. Moreover, the electrical activity displayed by the neural spheroids, as determined by multielectrode arrays (MEA) measurements, was significantly higher when they were embedded in the nanocomposite (versus in collagen and in collagen with suspended GlyCNDs) in terms of both single electrode and network activity. Interestingly, only the electrophysiological activity of spheroids in the nanocomposite was significantly more sensitive to NMDA receptor (NMDAR) compared to α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid receptor (AMPAR) blockers, thereby indicating an interplay between the NMDA‐dependent network and the immobilized GlyCNDs.

2. Results and Discussion

2.1. Physicochemical Characterization

The detailed fabrication process and physicochemical characterization of the GlyCNDs have been previously reported by Naccache and co‐authors.^{[ 53 ]} Here, we carried out a complementary X‐ray photoelectron spectroscopy (XPS) analysis to evaluate their conductive properties and confirm the nature of their surface functional sites. The XPS survey scan and the deconvolution of high‐resolution XPS spectra of O1s, C1s, and N1s peaks are reported in Figure S1 (Supporting Information).

From the deconvolution of the C1s spectrum, we observed a higher proportion of sp²‐hybridized carbon atoms (C=C, 284.55 eV) compared to sp³ (C‐C, 285.61 eV), with a sp³/sp² ratio of 0.37. Conductive carbon‐based nanomaterials often exhibit a preponderance of sp² hybridized carbon, which indicates that electrons experience greater mobility, and, in turn, an enhanced overall conductivity.^{[ 54} , ^{55 ]} Moreover, the C1s spectrum featured a distinct peak at a binding energy of 288.25 eV, generally indicative of carbonyl (CO) functionalities. In addition, a peak was also identified at a binding energy of 531.87 eV in the deconvoluted O1s spectrum, indicating the presence of carboxylic acid (COOH) groups on the surface of the GlyCNDs that are expected to create amide bonds when interacting with free amines (NH₂) along the collagen chain.^{[ 56 ]}

To test the resulting binding efficacy, we incorporated the GlyCNDs in a collagen suspension at a concentration of 1 mg mL⁻¹ both in their pristine state (hereafter referred to as C_CND1) and upon (1‐ethyl‐3‐[3‐dimethylaminopropyl]carbodiimide hydrochloride) (EDC)/N‐hydroxysuccinimide (NHS) activation (hereafter referred to as CE_CND1), and compared them to collagen cross‐linked with EDC/NHS (CE). The degree of immobilization was evaluated by Raman spectroscopy. Figure  1A displays the representative Raman spectra for CE, C_CND1, and CE_CND1. A greater presence of GlyCNDs in CE_CND1 is attested by the appearance of a wide fluorescence band spanning from ≈1800 to 3600 cm⁻¹ that is characteristic of the nanoparticles (refer to Raman spectra of GlyCNDs in Figure 1A‐insert). Furthermore, the formation of new C─N bonds between the COOH groups of GlyCNDs and NH₂ groups of collagen is demonstrated by quantifying the area of the bands associated with the C─N stretching of both Amide III (1286 cm⁻¹) and II (1589 cm⁻¹) as well as the C═O stretching of Amide I (1641 cm⁻¹).^{[ 57} , ⁵⁸ , ⁵⁹ , ^{60 ]} Specifically, the average normalized area relative to the C─N stretching of the Amide III (Figure 1B) is 0.04 for CE, and it increases to 0.12 and 0.17 for C_CND1 and CE_CND1, respectively. In addition, the normalized area of the C─N stretching of the Amide II (Figure 1B) increases from 0.12 for CE to 0.24 and 0.38 for C_CND1 and CE_CND1, respectively. Lastly, the normalized area of the C═O stretching band characteristic of Amide I increases from 5.89 for CE to 6.92 and 8.23 for C_CND1 and CE_CND1, respectively.

Figure 1.

A) Representative Raman spectra of GlyCNDs (insert), CE, C_CND1, and CE_CND1. B) Quantification of the C─N stretching associated with the Amide III and Amide II groups as well as of the C═O stretching relative to Amide I, normalized against the N─C─H deformation of collagen proline ring. C) Representative FT‐IR spectra of CE, C_CND1, and CE_CND1 samples. D) Quantification of the C─N and C═O stretching vibrations in terms of integral area. E) Release profile of GlyCNDs from C_CND1 and CE_CND1. F) Atomic force microscopy (AFM) micrograph of CE_CND1 showing the spatial distribution of GlyCNDs within the collagen fibrous matrix. G) Representative depth profile extracted from AFM micrograph. Scale bar: 150 µm. All numerical data are presented as mean ± standard deviation (SD). Statistical significance was determined using one‐way analysis of variance (ANOVA) and Tukey's Honestly Significant Difference (HSD) post hoc test: ^** p < 0.01, ^* p < 0.05, non‐significant (ns) p  > 0.05.

To further confirm the successful formation of new C−N covalent bonds within the CE_CND1 condition, we carried out Fourier‐Transform Infrared (FT‐IR) spectroscopic analysis focusing on the 1200–1800 cm⁻¹ region. Figure 1C displays a representative FT‐IR spectrum for the three conditions examined. Upon peaks deconvolution, we observed an increase in the normalized area for both the C─N stretching (Amide II, 1562 cm⁻¹) and the C═O stretching (Amide I, 1678 cm⁻¹) in the CE_CND1 condition compared to both CE and C_CND1 (Figure 1D).^{[ 61} , ^{62 ]} By exploiting the native fluorescence of GlyCNDs at an excitation wavelength of 380 nm (Figure S2A, Supporting Information), we carried out release studies to further confirm the role of EDC/NHS coupling in achieving a stable immobilization of the GlyCNDs within the collagen matrix. As shown in Figure 1E, the release of GlyCNDs displayed by CE_CND1 is significantly lower compared to that of C_CND1 condition throughout the full experiment timeframe (14 days). Notably, the release percentage from CE_CND1 at day 4 (≈20%) was used to include one additional control for the cellular experiments (referred to as CE + rCND1, and further described in Section 2.3), where we supplemented the culture medium with the same amount of GlyCNDs released from the CE_CND1 condition at this time point. This enabled us to isolate the cellular effects solely due to the immobilized GlyCNDs from those associated with suspended nanoparticles.

Successively, non‐contact AFM was employed to visualize the GlyCNDs spatial arrangement within the collagen matrix (Figure 1F), characterized by randomly distributed clusters of the carbon nanodots (Figure S2B, Supporting Information). The AFM linear depth profiles (Figure 1G) confirm a size distribution of the GlyCNDs ranging from ≈6 to 16 nm, as previously reported by Naccache and co‐authors.^{[ 53 ]} Conductivity measurements assessed how the incorporation of GlyCNDs into the collagen matrix modulates the overall electrical properties of the nanocomposite. To this end, different GlyCND concentrations, ranging from 0.1 to 4.0 mg mL⁻¹, were tested and the resulting conductivity was compared to the one displayed by CE. As shown in Figure  2A (i.e., conductivity values reported at the 2‐hour mark), the relatively poor conductivity of CE (2.8 mS cm⁻¹) significantly increases as a result of the addition of the GlyCNDs up to 1 mg mL⁻¹. Interestingly, a higher concentration (i.e., 4 mg mL⁻¹) does not result in additional changes. We hypothesize that concentrations higher than 1 mg mL⁻¹ saturate the availability of free NH₂ groups of collagen, thereby resulting in unreacted GlyCNDs that ultimately become suspended and thus do not contribute to the overall conductivity. In this context, the substantial contribution of GlyCNDs is further demonstrated by considering the complete time‐dependent conductivity values gathered during the initial 2 hours (h) of collagen fibrillogenesis (Figure S2C, Supporting Information). During such process, the evolving collagen network sequesters ions and thus reduces the number of mobile charges available, leading to a gradual decline in conductivity. Although the conductivity measurements for both CE and CE_CND1 show a trend that supports this phenomenon, the nanocomposite demonstrates greater resistance to conductivity loss. Specifically, CE_CND1 only experiences a 14% decrease in conductivity, while CE conductivity declines by over 70%. Furthermore, we evaluated the chemical stability of the nanocomposite. In accordance with results already reported in the literature,^{[ 63} , ^{64 ]} we found that the presence of EDC/NHS greatly improves the resistance to both hydrolytic and enzymatic (Figure 2B) degradation of pristine collagen (hereafter referred to as C). Both CE and CE_CND1 conditions display a significantly lower mass loss due to hydrolytic degradation when compared to C. The same consideration is also valid for enzymatic degradation in the presence of collagenase IV. This increased stability of the CE_CND1 condition highlights the potential of our novel nanocomposite to overcome one of the major limitations of collagen hydrogels for tissue engineering and neural applications.

Figure 2.

Electrical, structural, and mechanical characterization of the nanocomposite and relevant control conditions for comparison. A) Quantification of the electrical conductivity as a function of variable concentrations of GlyCNDs spanning from 0 (CE) to 4 mg mL⁻¹. B) Hydrolytic (left) and enzymatic (right) degradation of C, CE, and CE_CND1. C) Viscosity of the hydrogels as a function of shear rate. D) Stiffness (left) and Young's Modulus (right) obtained via AFM indentations. All numerical data are presented as mean ± SD. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p  > 0.05.

Lastly, owing to the well‐known importance of stiffness in modulating the cellular response,^{[ 65} , ⁶⁶ , ^{67 ]} we carried out a comprehensive mechanical characterization via rheological analysis and AFM indentations with a 5 µm‐diameter spherical indenter. Figure 2C presents viscosity measurements for three distinct conditions: C, Cand CE_CND1. The results clearly demonstrate the influence of EDC/NHS cross‐linking on the rheological properties of collagen, as evidenced by the elevated viscosity of the CE sample compared to C. Such observation aligns with existing literature, which has similarly highlighted the substantial role of EDC/NHS cross‐linking in tuning the mechanical characteristics of polymeric scaffolds.^{[ 68} , ^{69 ]} In addition, the introduction of GlyCNDs does not significantly affect the rheological properties of the final construct, as CE_CND1 displayes a similar shear‐thinning behaviour when compared to CE.

The analysis of the AFM force‐indentation curves enabled the extraction of sample's stiffness and Young's modulus. As shown in Figure 2D, pristine collagen displayes the lowest stiffness (1.92 × 10⁻³ N m⁻¹) and Young's modulus (609 Pa).^{[ 70 ]} As expected,^{[ 63} , ^{71 ]} EDC/NHS cross‐linking in the CE condition results in a higher stiffness (2.28 × 10⁻³ N m⁻¹) and Young's modulus (735 Pa). In agreement with the rheological measurements, the mechanical properties are not altered by the incorporation of GlyCNDs. The nanocomposite CE_CDN1 displayes stiffness and Young's modulus values of 2.37 × 10⁻³ N m⁻¹ and 744 Pa, respectively, with no statistically significant differences with the CE condition. The results found for CE and CE_CND1 enabled us to conclude that the rheological and mechanical properties, namely viscosity and stiffness, are not expected to be a major contributing factor for the cellular results reported in the following sections.

2.2. Cytotoxicity Assays

Despite the considerable efforts to characterize the potential neurotoxicity of carbon‐based nanomaterials,^{[ 42} , ⁷² , ^{73 ]} the results reported in the literature remain controversial, likely due to interstudy variations in size, physicochemical properties, and concentration. Here, to evaluate the cytocompatibility of both collagen‐immobilized and suspended GlyCNDs with iPSC‐derived NP spheroids, we carried out a colorimetric lactate dehydrogenase (LDH) assay complemented with propidium iodide (PI) live staining of dead cells. Three different concentrations (i.e., 0.5, 1, and 4 mg mL⁻¹) of GlyCNDs were both immobilized within the collagen matrix by EDC/NHS coupling (hereafter referred to as CE_CND05, CE_CND1, and CE_CND4) and suspended in the culture medium (hereafter referred to as CE + CND05, CE + CND1, CE + CND4). Pristine collagen (C) and EDC/NHS cross‐linked collagen (CE) were used as controls. Figure  3A displays the percentage values of released LDH found for the different conditions at 6 and 72 h. The results indicate that the viability of NP spheroids is not impacted by GlyCNDs immobilized within the collagen matrix at concentrations of 1 mg mL⁻¹ or lower. In fact, the percentage values of released LDH of CE_CND05 and CE_CND1 are similar to the ones displayed by the control groups at both time intervals. Conversely, we observed that CE_CND4 significantly increases cell mortality at both time points, with LDH percentage values that doubled in comparison to those displayed by the controls. When GlyCNDs are suspended in the culture medium, significantly higher levels of LDH are observed for both time points. In particular, the cytotoxicity increases proportionally with the GlyCNDs concentration, with CE + CND4 displaying the highest LDH release values. These results were further complemented by PI staining of the NP spheroids at 72 h.

Figure 3.

The immobilization of GlyCNDs within the collagen matrix hinders potential cytotoxic effects arising from the presence of suspended nanoparticles, yielding a cell viability similar to that of pristine collagen. Different concentrations of GlyCND were both immobilized within the collagen matrix (_) and suspended in culture medium (+). C and EDCCE were employed as controls. A) LDH percentage release for the different conditions tested. B) Representative PI/DAPI staining images for control conditions C and CE, suspended GlyCNDs at a concentration of 1 and 4 mg mL⁻¹ and immobilized GlyCNDs at a concentration of 1 and 4 mg mL⁻¹. Scale bar: 25 µm. C) Cell death for the different conditions. All numerical data are presented as mean ± SD. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p  > 0.05.

Figure 3B displays representative images of controls, immobilized and suspended GlyCNDs conditions. It is evident that NP spheroids cultured in the GlyCNDs‐supplemented medium exhibit significantly higher amounts of dead cells compared to both controls and collagen with immobilized GlyCNDs. Cell death percentage (Figure 3C) was calculated as the PI‐positive (PI⁺) cells over the total amount of nuclei. Not surprisingly, PI staining mirrors the results obtained with the LDH assay. Immobilized GlyCNDs display similar cell death percentages to control groups, while suspended GlyCNDs yield higher cytotoxicity. It can be hypothesized that the amount of ─COOH groups on suspended GlyCNDs may be a leading factor that yielded such high mortality rate. In fact, it has been previously reported that such free functional sites are likely to interact with lipid molecules in cellular membranes, potentially disrupting membrane integrity and leading to cellular damage and death.^{[ 74} , ^{75 ]} Overall, our results provide additional evidence to support recent arguments that emphasize the need for immobilization of carbon‐based nanomaterials within scaffolds to mitigate undesirable outcomes that are associated with their use as suspended particles, such as bioaccumulation (Figure S2D, Supporting Information).^{[ 76} , ^{77 ]} Furthermore, the consistent LDH values found at 6 and 72 h across all conditions with suspended GlyCNDs indicate cell death onset within the first 6 h of exposure. Most importantly, the cytotoxicity assays played a crucial role in determining that immobilized GlyCNDs at a concentration of 1 mg mL⁻¹ displays viability results that are comparable to those of the control groups, while greatly improving the poor electrical conductivity of pristine collagen. For this reason, we selected CE_CND1 as the experimental condition for the cellular assays presented in the following sections.

2.3. Neurite Growth and Formation of a Complex Neuronal Network

One of the primary goals of neural tissue engineering is to develop functional scaffolds that favor neurite spreading and extension while supporting the establishment of complex neural networks. After only 5 days in culture, several neurites spreading from spheroids embedded in the CE_CND1 nanocomposite are visible, while none/few are observed in control conditions (i.e., CE and CE + rCND1) (Figure S3A, Supporting Information). To quantitatively characterize morphological differences, the spheroids were stained for β III‐tubulin and imaged via confocal microscopy at 14 days. The CE_CND1 promotes a significantly more extensive neurite outgrowth when compared to CE and CE + rCND1 (Figure  4A). The overall branching complexity was evaluated using a semiautomated Sholl analysis centered on the spheroid's body. In particular, the total number of neurite intersections was calculated using 20 µm‐spaced concentrical hemispheres (Figure 4B). As shown in Figure 4C, spheroids within the CE_CND1 condition exhibit an average number of neurite intersections of 95 ± 28, which is significantly higher than when they are placed within CE (58 ± 21) and CE + rCND1 (47 ± 18) matrices. To evaluate the neurite ramification tendency, we utilized the Schoenen index, defined as the maximum number of intersections divided by the number of primary branches exiting from the spheroid's soma.

Figure 4.

CE_CND1 matrix promotes neuritogenesis and the formation of complex neural networks. A) β‐III tubulin expression of NP spheroids embedded in CE, CE_rCND1 and CE_CND1. Scale bar: 250 µm B) semiautomated Sholl analysis performed on NP spheroids encompassing the use of 20 µm‐spaced concentric hemispheres (centered around the spheroid's soma). Scale bar: 600 µm. C) Number of neurite‐hemisphere intersections and Schoenen ramification index. D) Representative linear regression applied to the area‐normalized intersections as a function of the distance from the spheroid's soma. E) Sholl's regression index. F) Quantification of neurites’ length. G) Relative expression of GAP43. All numerical data are presented as mean ± SD. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p > 0.05.

As shown in Figure 4C, the CE_CND1 condition displays the highest average index of 3.6 ± 0.8. Interestingly, the significantly lower average value found for spheroids in the CE + rCND1 condition when compared to CE (i.e., 1.7 versus 2.1), indicates that the released GlyCNDs have detrimental effects on the formation of a complex neuronal network, thereby confirming previous similar observations with carbon nanomaterials.^{[ 17} , ¹⁹ , ²¹ , ²² , ^{23 ]} The overall increased branching complexity of spheroids in CE_CND1 is also validated by the Sholl's regression index, a parameter that quantifies the rate of decay of the number of branches with distance from the center of analysis (Figure 4D). As shown in Figure 4E, the CE_CND1 condition is characterized by the lowest index (0.3) when compared to that of the CE (1.6) and CE + rCND1 (1.8) matrices.

In addition to the increased neural network complexity, spheroids embedded in the CE_CND1 also exhibit the highest neurite length (Figure 4F) of 369 ± 113 µm. Interestingly, in the CE + rCND1 matrix, neurites are significantly longer when compared to the CE condition (i.e., 273 ± 78 versus 207 ± 53 µm). To unveil the underlying mechanisms that drive such divergent morphology, we investigated via Western blotting the expression of growth‐associated protein 43 (GAP43), a crucial protein for neuritogenesis.^{[ 78} , ⁷⁹ , ^{80 ]} As shown in Figure 4G, spheroids in CE_CND1 express a significantly higher amount of GAP43 at 14 days compared to those within the control conditions. Taken together, our findings reveal that immobilized GlyCNDs upregulated NP spheroid expression of GAP43, which is decisive in neurite outgrowth and complex neural network formation. In contrast, although GlyCNDs released in solution facilitate the development of longer neurites, they exhibited significantly lower ramifications than NP spheroids embedded in CE.

To investigate whether the ability of the nanocomposite to support neuritogenesis extends to mature neurons, we used the CE_CND1 nanocomposite as a matrix for a scaffold‐based 3D culture of mouse primary cortical neurons, confirming that the establishment of a complex neural network is observed within 7 days (Figure S3B, Supporting Information). Compared to CE, neurons grown in the CE_CND1 matrix show a significant increase in neurite outgrowth that enabled the formation of intricate networks. Notably, the morphology elicited by CE_CND1 was qualitatively comparable to the one observed in Matrigel, the gold standard matrix in neurobiology.^{[ 81 ]} The use of a different cell type offers a validation of the nanocomposite's distinctive potential to support key processes associated with neuroregeneration. The similar behaviour observed in NP spheroids and primary cortical neurons ssuggests that the nanocomposite's beneficial effects transcend specific cell types and in vitro culture systems. Both spheroids and the 3D culture of primary neurons emulate more closely than conventional 2D monolayer systems the in vivo cellular environment, thereby providing a more physiologically relevant representation of neuron‐biomaterial interactions. Taken together, our findings highlight the potential wide‐ranging applicability of the CE_CND1 nanocomposite for reconstruction of lost neuroanatomical connectivity.

2.4. Neuronal Differentiation

To expand the breadth of the investigation of the beneficial effects of the nanocomposite, we evaluated the expression levels of key neuronal maturation markers by immunofluorescence (IF) imaging and Western blotting. Figure  5A displays representative IF images of NP spheroids embedded in CE, CE + rCND1, and CE_CND1 after 2 weeks of culture. Samples were stained for Ki‐67, a nuclear proliferation marker,^{[ 82 ]} as well as for β III‐tubulin and MAP2, an early and a mature neuronal marker, respectively.^{[ 83} , ⁸⁴ , ^{85 ]} The overall expression of these proteins indicates that the differentiation of the neurospheres embedded in the CE_CND1 matrix is promoted, showing significantly lower amounts of Ki‐67‐positive cells and higher levels of β III‐tubulin and MAP2 (Figure 5B). These findings are also corroborated by the relative expression of calretinin (an early neuronal marker) and nestin (a NP marker).^{[ 83} , ⁸⁶ , ^{87 ]} In particular, spheroids cultured in CE and CE + rCND1 matrices display a higher amount of nestin when compared to the ones embedded in CE_CND1, which instead express higher levels of calretinin (Figure S4, Supporting Information). Quantitative Western blot analysis (Figure 5C) confirms IF imaging results by showing a higher relative expression of β III‐tubulin for neurospheres in CE_CND1 (0.29) when compared to CE and CE + rCND1 (0.17 each) (Figure 5D). As shown in Figure 5E, similar considerations are also valid for MAP2, with higher expression (0.48) in the CE_CND1 nanocomposite. These results support the hypothesis that the distinct physicochemical properties of the GlyCNDs (i.e., size, reactive functional groups and conductivity) play a pivotal role in promoting the neuronal differentiation of NP spheroids. In particular, the dimension of the GlyCNDs (in the 6–16 nm range), is expected to facilitate selective interactions with specific membrane receptors implicated in neuronal differentiation. In this context, previous studies showed that carbon‐based nanomaterials can enhance neuronal differentiation by binding selectively to TrkA/p75 and integrin receptors, thereby triggering downstream pathways that favor neuronal lineage commitment.^{[ 88} , ^{89 ]} In addition, the abundant reactive functional groups present on GlyCNDs, such as COOH, NH2, and C═O, can enhance the local adsorption and concentration of key neurotrophic factors such as brain‐derived neurotrophic factor (BDNF) and basic fibroblast growth factor (bFGF), providing a more favorable microenvironment for sustained neuronal differentiation.^{[ 90 ]} Finally, conductive environments have been shown to modulate vital cellular functions such as ion channel activity^{[ 90 ]} and membrane potentials,^{[ 91 ]} as well as improve intercellular communication and signaling, ultimately contributing to enhanced neuronal differentiation.^{[ 92} , ^{93 ]}

Figure 5.

CE_CND1 supported the early differentiation of NP spheroids into mature neurons. A) Representative IF images showing the expression of Ki‐67, MAP2, and β III‐tubulin in spheroids cultured in CE, CE_rCND1, and CE_CND1 for 14 days. Nuclei were stained with DAPI). Scale bar: 100 µm. B) Quantification of the percentage of Ki‐67 positive cells (Ki‐67⁺). C) Western blot analysis of β III‐tubulin and MAP2 c/d expression. D) Quantification of β III‐tubulin expression level. E) Quantification of MAP2 c/d expression level. All numerical data are presented as mean ± SD. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p > 0.05.

Taken together, our findings indicate that the immobilized GlyCNDs induce a rapid neuronal differentiation and maturation of NP spheroids. This evidence is supported by the fact that the nanoparticles released from the matrix (CE_rCND1) do not elicit any effect on differentiation, similar to the condition where they are absent (CE). The positive effects of immobilized GlyCNDs strengthen the potential and impact of carbon‐based nanomaterials in neural tissue engineering. Specifically, the accelerated neuritogenesis and the elevated expression of the mature neuronal marker MAP2 highlight the remarkable capacity of our nanocomposite to accelerate neuronal differentiation of NP spheroids (compared to previous literature on mouse iPSCs)^{[ 94} , ⁹⁵ , ^{96 ]} and ensure their complete electrophysiological maturation, two crucial aspects for biomaterial‐driven neural development and regeneration.

2.5. Electrophysiological Maturation and Neuronal Network Communication

Lastly, we evaluated the spontaneous electrical activity of NP spheroids, both at the single‐electrode and network level, by capitalizing on MEA systems (Figure  6A). Starting at day 5 and consistently throughout the entire experimental duration (28 days), NP spheroids in the CE_CND1 matrix exhibit a remarkable enhancement in single‐electrode activity. Specifically, the firing rate, which reflects the frequency of action potentials, is significantly higher than the other experimental groups (Figure 6C). Furthermore, an increased number of active electrodes (Figure 6D) and an enhanced burst frequency can be observed (Figure 6E), indicating an increased synchronized activity among the cells in the spheroids. In addition to the single‐electrode level, the CE_CND1 nanocomposite also positively influences the network burst frequency, a measure of large‐scale synchronized activity among different electrodes, which is consistently and significantly higher for the spheroids embedded in CE_CND1 (Figure 6F). This result indicates that the CE_CND1 nanocomposite facilitates the formation of functional connections among cells, yielding a more integrated neuronal network. In addition, the average number of spikes per network burst is notably higher (Figure 6 G,H), suggesting that the spheroids cultured in the nanocomposite establish more complex and intensified network communications. By day 14, neurospheres show a remarkable single‐electrode activity and the generation of intricate neuronal network connections (Figure 6B). The importance of immobilizing the GlyCNDs within the collagen matrix is further demonstrated by the electrophysiological recordings relative to the CE_rCND experimental group. In fact, released GlyCNDs did not elicit noticeable or consistent alterations from the electrical signature that was displayed by CE, neither at the single electrode or at the network level.

Figure 6.

Spheroids cultured in the CE_CND1 nanocomposite show faster and more integrated electrophysiological maturation. A) Representative image of NP spheroids cultured in MEA well plates. Scale bar: 300 µm. B) Representative activity traces at 14 days in vitro (DIV) for CE, CE_rCND1 and CE_CND1. Single detected spikes are represented by the black lines in the raster plots, while blue lines indicate single electrode burst. Neural networks are shown as pink rectangles. C) Heating map showing the weighted mean firing rate during 28 DIV. D) Number of detected active electrodes during 28 DIV. Shadowed areas represent mean ± SD. E) Single electrode burst frequency. F) Network burst frequency. G) Representative 1s‐binning raster plots showing singular spikes detected within network activity at 14 DIV for CE, CE_rCND1, and CE_CND1. H) Spikes per network activity during 28 DIV. All numerical data are presented as mean ± SD, except for (G) where mean ± standard error of the mean (SEM) are presented. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p  > 0.05.

These findings demonstrate that collagen‐immobilized GlyCNDs provide a favorable microenvironment that accelerates the electrophysiological maturation and the establishment of effective neural network formation. Together with the neuronal differentiation and the morphogenesis data reported in the previous sections, these results demonstrate that the electroconductive nanocomposite enhances the maturation of functionally active neuronal cells and promotes their communication through an intensified neural network.

To gain a deeper understanding of the underlying mechanisms that contribute to such enhanced electrophysiological activity, we investigated by IF and Western blotting the expression of synapsin (i.e., a crucial phosphoprotein known to promote the establishment and maintenance of synaptic connections by actively regulating the release of neurotransmitters) and that of NMDAr, which are primarily involved in the transmission of excitatory information.^{[ 97} , ⁹⁸ , ⁹⁹ , ^{100 ]} Figure  7A shows that NP spheroids embedded in CE_CND1 exhibit higher levels of synapsin, namely 0.27 versus 0.16 and 0.18 for the CE and CE_rCND1 groups, respectively (Figure 7C,D). On the other hand, NMDA receptor levels are comparable for the conditions tested (Figure 7B,C,E). However, acute treatment with MK‐801, a noncompetitive NMDA receptor antagonist, significantly affects the electrical activity; in particular that of the spheroids embedded in the CE_CND1 nanocomposite. As depicted in Figure 7F, upon treatment with MK‐801 on days 14 and 28, the firing rate exhibits a reduction of 63% and 61%, respectively, compared to the spontaneous baseline recording. Furthermore, MK‐801 acute treatments are also found to reduce network activity (Figure 7G). Notably, the impact of NMDAr antagonist is significantly more severe on the firing rate and network burst of the spheroids embedded in CE_CND1, indicating an overall activity that is heavily dependent on the availability of NMDAr. In comparison, acute treatment with 2,3‐dioxo‐6‐nitro‐7‐sulfamoyl‐benzo[f]quinoxaline (NBQX), a commonly used AMPA and kainate receptor blocker, yields the opposite trend, whereby neurospheres in CE_CND1 are significantly less susceptible to AMPA receptor antagonist than those in the CE and CD_rCND1 matrices at both 14 and 28 days. This differential impact of the two antagonists suggests that the electrophysiological activity of NP spheroids is mainly governed by NMDAr activation. Based on these findings, we hypothesize the presence of direct interactions between the GlyCNDs in the CE_CND1 and NMDAr alters the channel dynamics and ultimately leads to channel activation. In particular, the COOH, NH₂, and CO functional groups on the GlyCND surface could either competitively or synergistically interact with the glycine binding sites on NMDAr. In addition, it could be conceived that the anchorage of GlyCNDs on collagen fibers allows for an optimal spatial orientation that ultimately enhances binding efficiency with NMDAr. Although this theory needs further scrutiny since other surface receptors (e.g., neurotrophin and nerve growth factor) can regulate NMDAr‐dependent currents,^{[ 101 ]} our findings nonetheless indicate biomaterial‐derived effects on the NMDAr which may, either completely or in part, explain the reported cellular effects, as NMDAr activation is known to drive neurite growth, neural differentiation, and maturation as well as electrical activity.

Figure 7.

NMDA receptors greatly influence the electrophysiological signature of NP spheroids embedded in the nanocomposite. A) Representative IF images showing the 14‐day expression of synapsin and β III‐tubulin in spheroids cultured in CE, CE_rCND1, and CE_CND1 matrices. Nuclei were counterstained with DAPI. Scale bar: 100 µm. B) Representative IF images of NMDAr and β III‐tubulin in spheroids cultured in CE, CE_rCND1, and CE_CND1 matrices. Nuclei were stained with DAPI. Scale bar: 100 µm. C) Western blot analysis of synapsin and NMDAr expressions. D) Quantification of synapsin expression levels. E) Quantification of NMDAr expression levels. F) Percentage firing rate deviations from baseline recordings following acute MK‐801 and NBQX treatment performed at 14 and 28 DIV. G) Network burst frequency deviations from baseline recordings following acute MK‐801 and NBQX treatment performed at 14 and 28 DIV. All numerical data are presented as mean ± SD. Statistical significance was determined using one‐way ANOVA and Tukey's HSD post hoc test: ^** p < 0.01, ^* p < 0.05, ns p > 0.05.

3. Conclusion

In conclusion, this work presents a novel electroconductive nanocomposite consisting of a collagen type I matrix decorated with GlyCNDs, ultimately demonstrating its promising potential as a biomaterial for applications ranging from neural tissue engineering and neuroregenerative medicine to bioinks and matrices for 3D cultures and biomimetic in vitro models. In particular, after establishing an effective anchorage method for GlyCNDs within the collagen matrix through EDC/NHS coupling, we carried out a cytotoxicity assay to inform the optimization of GlyCNDs concentration toward a biocompatibility comparable to that of pristine collagen. The biological characterization of the nanocomposite was carried out with mouse iPSC‐derived NP spheroids (and, in part, with a 3D culture of primary neurons) for a more physiologically accurate representation of in vivo conditions toward enhancing the reliability and translatability of our findings. The CE_CND1 nanocomposite substantially enhances the neuronal differentiation of NP spheroids and promotes neuritogenesis, with conspicuous dendritic arborization and axonal outgrowth, ultimately facilitating the formation of functional and highly integrated neural networks. Furthermore, acute MK‐801 treatment suggests a direct interaction between collagen‐immobilized GlyCNDs and postsynaptic NMDAr.

4. Experimental Section

Synthesis

GlyCNDs were produced through a one‐step hydrothermal procedure.^{[ 53 ]} Briefly, glycine was dissolved into 8 mL of a sodium hydroxide solution (5 × 10⁻³ m) to obtain a final concentration of 4 m. Subsequently, a 100 mL solution of deionized water containing 50 × 10⁻³ m citric acid was added and the final mixture was thoroughly agitated until clear. The solution was then subjected to a hydrothermal treatment at 210 °C for 4 h under continuous stirring at 550 rpm. Once cooled, the solution was dialyzed against water for 4 days to remove unreacted precursors. Then, the GlyCNDs were washed multiple times with acetone (1:10 v/v) and dried for 48 h at 85 °C.

Nanocomposite Hydrogels Preparation

Collagen hydrogels were prepared by dilution of rat tail type I stock (Corning, USA, 9.38 mg mL⁻¹, #354249) in differentiation medium (DM) and PBS 10× to reach a final concentration of 2 mg mL⁻¹. The medium's composition is outlined in “Derivation and culture of iPSC‐derived NP spheroids” section. GlyCNDs were introduced in the collagen solution either in their pristine or activated state. For the latter, GlyCNDs were immersed in MES buffer (pH = 6.0) containing EDC and NHS (ratio 2:1) for 1 h at 4 °C. The mass ratio between the GlyCNDs and EDC was fixed at 1:1. Upon the introduction of the GlyCNDs, the pH of the collagen solution was neutralized through the addition of 1 N NaOH. The collagen hydrogels were then incubated at 37 °C to allow complete polymerization. Three GlyCNDs concentrations were tested (for both pristine and activated GlyCNDs states) based on cytotoxic data available for other carbon‐based nanoparticles in the literature (Table  1 ).^{[ 76} , ¹⁰² , ^{103 ]}

Table 1.

GlyCNDs concentrations tested in this study.

Pristine GlyCNDs
Abbreviation	EDC/NHS [× 10−3 m]	GlyCNDs [mg mL−1]
C_CND05	/	0.5
C_CND1	/	1.0
C_CND4	/	4.0
Activated GlyCNDs
CE_CND05	2.5	0.5
CE_CND1	5	1.0
CE_CND4	20	4.0

AFM Characterization

Samples were dried overnight and imaged via non‐contact AFM carried out on an Alpha300 RSA system (WITec, Germany). Surfaces (1 × 1 µm²) were scanned using the triangular Si₃N₄ Cantilevers of the DNP‐S10 chip (Bruker, USA), characterized by a nominal spring constant of 0.2 N m⁻¹, a resonant frequency of 13 kHz and nominal tip radius of 10 nm. 3D micrographs were then processed in Gwyddion^{[ 104 ]} to extract depth profile and GlyCNDs’ diameter.

XPS Characterization

GlyCNDs were analyzed by XPS with a K‐Alpha X‐ray photoelectron spectrometer (Thermo Scientific, USA). Three randomly selected regions of the sample were analyzed in triplicate, with each scan consisting of 10 runs. The results were averaged and plotted for both the survey and the high‐resolution scans.

Raman Spectroscopy

Single spectra for each experimental condition were acquired with the Raman module of the Alpha300 RSA system. Spectra were collected through a 50× objective (EC Epiplan NEOFLUAR, N.A. = 0.9, Zeiss) with an excitation wavelength of 524 nm provided by a doubled Nd:YAG laser (12.5 mW, acquisition time of 2 s). Baseline subtraction, data normalization, and Voigt (Lorentzian/Gaussian) deconvolution for the identification of vibrational components were performed in OriginPro (OriginLabs, USA). The bands related to C─N stretching of both the Amide III (1271 cm⁻¹) and Amide II (1558 cm⁻¹) groups as well as the characteristic C═O stretching bands related to Amide I, were used to confirm the immobilization of the GlyCNDs into the collagen matrix.^{[ 57} , ⁵⁸ , ⁵⁹ , ^{60 ]} The 1098 cm⁻¹ band, associated to the N─C─H deformation of the collagen proline ring,^{[ 57} , ⁶⁰ , ^{105 ]} was used as the reference peak for the normalization and comparison of spectra across conditions.

FT‐IR Spectroscopy

A Nexus 870 FT‐IR system (ThermoFisher Scientific, USA) was utilized in transmission configuration. Spectral data were collected with a 2 cm⁻¹ resolution in the 1200–1800 cm⁻¹ range. Each measurement consisted of 128 scans at a rate of 1 second per scan. Baseline correction, spectral normalization, and Voigt deconvolution for band identification were carried out by using OriginPro software (OriginLabs, USA). The 1437 cm⁻¹ band, associated with CH₂ bending, was used as the reference peak for spectra normalization. The bands related to C─N (Amide II) and C═O stretching (Amide I), centered respectively at 1562 and 1678 cm⁻¹, were employed to evaluate the newly formed covalent bonds between GlyCNDs and the collagen matrix.

Conductivity Measurements

A conductivity meter (CON 110 Conductivity/TDS Meters, Oakton Instruments, USA) was used to assess the electrical conductivity of the collagen matrices used in this work, with a variable GlyCND concentration, namely 0.5, 1.0, and 4.0 mg mL⁻¹. Samples were prepared according to the same procedure outlined above, by diluting the collagen stock solution with Neurobasal media to reach a final collagen concentration of 2 mg mL⁻¹. The matrices were incubated at 37 °C for up to 2 h to allow for complete polymerization prior to the conductivity measurements. Three samples per concentration were analyzed and compared to the control condition represented by EDC/NHS‐cross‐linked collagen.

Degradation Studies

The structural stability of the collagen matrices with variable GlyCND concentration was evaluated by carrying out both hydrolytic and enzymatic degradation studies. For the latter, the degradation solution consisted of collagenase type IV from Clostridium histolyticum (305 U mg⁻¹, # LS004188, Worthington Biochemical Corporation, USA) dissolved in phosphate‐buffered saline solution (PBS 1×) containing 2 × 10⁻³ m CaCl₂ at a concentration of 10 U mL⁻¹. In both studies, 400 µL of degradation solution (PBS 1× for hydrolytic) was added to 250 µL of the matrix. At specific timepoints, the supernatant was collected, and the collagen concentration was estimated following the microplate procedure of the colorimetric BCA assay (ThermoFisher Scientific, Pierce BCA Protein Assay Kit, # 23225, ) and calibration curves, obtained collagen standards (0–500 µg range) diluted in both degradation solutions. The percentage of the nanocomposites’ mass loss was calculated at any selected timepoint by subtracting the collagen mass in the supernatant from the initial one (i.e., 500 µg).

Rheological and AFM Measurements

A characterization of C, CE and CE_CND1 was performed via rheological measurements and AFM indentations. The viscosity profiles of fully polymerized hydrogels were extracted using a Brookfield R/S plus rheometer (Brookfield, USA) equipped with a conical spindle with a diameter of 50 mm (C50‐2). The samples were subjected to a shear rate ramp, incrementally escalating up to 500 s⁻¹.

Stiffness and Young's modulus were extracted from indentation curves performed on AFM module on an Alpha300 RSA system using a spherical silicon nitride tip with a precalibrated spring constant of 0.133 N m⁻¹ and a diameter of 5 µm (Bruker, USA, MLCT‐SPH‐5UM, cantilever E). Indentations were carried out in PBS 1× to quantify the samples’ stiffness in their hydrated state.

The indentation curves were processed with the OriginPro software. Specifically, a linear regression of the approach phase of indentation was used to extract the stiffness of the samples. The Young's modulus was quantified by fitting force‐indentation curves with the following adapted Hertz model that accounts for the geometry of the indenter:

$$F=\frac{4}{3}\frac{E}{\left(1-{\upsilon }^2\right)}{R}^{0.5}{d}^{1.5}$$ (1)

where F represents the corresponding load, E represents the Young's modulus, υ is the Poisson's ratio of the material (set at 0.5 throughout the conditions), R is the radius of the indenter, and d is the indentation depth.^{[ 106 ]}

Derivation and Culture of iPSC‐Derived NP Spheroids

Commercially available mouse iPSCs (Alstem, iPS02m, USA) were propagated in feeder‐free conditions on gelatin‐coated culture surfaces. Cultures were periodically tested for mycoplasma with Lookout mycoplasma PCR detection kit (Sigma Aldrich, MP0035, USA). iPSC maintenance medium was composed of KnockOut DMEM (Gibco, 10829018, USA) supplemented with 15% knockout serum replacement (Gibco, N10828028), 1% MEM nonessential amino acid solution (Stemcell, 07600), 200 × 10⁻⁶ m l‐glutamine (Gibco, 25030), 1% penicillin‐streptomycin (Gibco, 15070063), 100 × 10⁻⁶ m 2‐mercaptoethanol (Gibco, 31350), and 1000 U mL⁻¹ leukemia inhibitory factor (LIF, SigmaAldrich, LIF2050). Embryoid body (EB) formation was initiated by detaching iPSCs from culture surfaces using TryplE (Gibco, 12604013) and resuspending in fresh iPSC maintenance medium without LIF. Cell suspensions were transferred to AggreWell800 plates (STEMCELL Technologies, 34811, USA) treated with anti‐adherence rinsing solution (STEMCELL Technologies, 07010) and embryoid bodies were allowed to form overnight. Finally, EBs were transferred to anti‐adherence treated 6‐well plate. After 72 h, the media was switched to neuronal expansion (EM) consisting of 1:1 mixture of DMEM/F12 and Neurobasal medium (Gibco, #21103049) supplemented with 1% GlutaMAX, 1% pen/strep, 1% B‐27 Plus Supplement (Gibco, #A3582801), 0.5% N‐2 Supplement (Gibco, #17502001), 200 × 10⁻⁶ m ascorbic acid (Sigma Aldrich, #AX1775) and the following inhibitors: 5 × 10⁻⁶ m SB‐525334 (Tocris Bioscience, 3211, UK), 250 × 10⁻⁹ m dorsomorphin (Tocris Bioscience, 3093), 3 × 10⁻⁶ m Wnt agonist CHIR99021 (Millipore, SML1046, USA). After 10 days, the neurospheres expanded and formed spheroids. Prior to usage, the spheroids were passages 3× with TrypLE (ThermoFisher, #12604013).

The resulting spheroids were embedded in the collagen matrices and culture for 4 days in DM, consisting of EM without the inhibitors. The media was successively switched to maturation media (MM) consisting of DM supplemented with 100 × 10⁻⁶ m brain‐derived neurotrophic factor (BDNF, STEMCELL Technologies, #78005), 100 × 10⁻⁶ m glial‐derived neurotrophic factor (GDNF, STEMCELL Technologies, #78058) and dybutyryl‐cAMP (db‐cAMP, STEMCELL Technologies, #73882). During the spheroids culture, the maturation medium was refreshed every 3 days.

Cytotoxicity Assays

To evaluate potential cytotoxic effects associated to both immobilized and medium‐suspended GlyCNDs, the level of LDH was measured using CytoTox 96 Non‐Radioactive (Promega, G1780, USA) assay. Briefly, 50 µL aliquots were collected from the culture media at 6 and 72 h and then transfer to a 96 well plate. Subsequently, 50 µL of the Citotox96 reagent were added to each sample aliquot. The plate was then covered with tin foil and incubated for 30 min at room temperature. Finally, 50 µL of the stop solution were added to each well and the absorbance band at 492 nm was collected using a Sinergy H1 plate reader (BioTech Instruments, USA). The percent of cytotoxicity was calculated as:

$$\mathrm{Percent}\mathrm{cytotoxicity}=\frac{\mathrm{Experimental}\mathrm{LDH}\mathrm{release}\left({\mathrm{OD}}_{490}\right)}{\mathrm{Maximum}\mathrm{LDH}\mathrm{release}\left({\mathrm{OD}}_{490}\right)}\times 100$$ (2)

where the maximum LDH release was obtained by adding 10 µL of 10× Lysis solution to a negative control sample, 1 h before adding the Cytotox96a Reagent.

In addition, at 3 DIV, the samples were stained with PI (ThermoFisher, #BMS500PI) and Hoechst 33342 (ThermoFisher, R37605) and visualized under an LSM880 AxioObserverZ1 confocal microscope (Zeiss, Germany) with a Plan‐Apochromat 20× objective (NA = 0.8, Zeiss). The resulting images were processed on ImageJ to quantify the number of PI⁺ nuclei that was used as a measure of cell death.

According to the results obtained from the cytotoxicity assays, the 1 mg mL⁻¹ concentration of GlyCNDs was selected (CE_CND1) for all the subsequent cellular studies. This condition was compared with the control condition CE consisting of a collagen matrix that was cross‐linked with the same amount of EDC/NHS (i.e., 5 × 10⁻³ m) used for CE_CND1. Notably, we added an additional condition in which the spheroids were embedded in a CE matrix (CE_rCND1), supplemented with GlyCNDs dispersed in the medium to simulate the release profile previously quantified for the CE_CND1 condition. In this way, we were able to isolate the effect provided by both immobilized and dispersed unbound GlyCNDs released by the matrix. Table  2 summarizes the conditions considered for the in vitro assays.

Table 2.

Conditions tested in the following in vitro experiments.

Conditions employed for cell culture experiments
Abbreviations	EDC/NHS [× 10−3 m]	Immobilized GlyCNDs [mg mL−1]	Media‐suspended GlyCNDs [mg mL−1]
CE	5.0	/	/
CE_rCND1	5.0	/	0.2 *
CE_CND1	5.0	1.0	/

Immunohistochemistry (IHC) Staining

At 14 DIV, the spheroids were fixed in fresh 4% paraformaldehyde (PFA) at room temperature for 2 h. Fixed samples were permeabilized with 0.25% Triton‐X100 (Sigma‐Aldrich, #11332481001) and blocked with 5% horse serum (ThermoFisher, #31874) overnight at 4 °C. Samples were successively incubated with primary antibodies for 24 h at 4 °C, rinsed for a minimum of 10 times with blocking buffer, and lastly incubated overnight at 4 °C with donkey secondary antibodies. The details and working dilutions of primary and secondary antibodies are listed in Table  3 . After five rinses, the nuclei were stained with DAPI) for 4 h at room temperature.

Table 3.

List of primary and secondary antibodies, with their working dilution, used in this study.

Primary antibody	Dilution	Secondary antibody	Dilution
β‐III tubulin (Abcam, ab78078) MAP2 (ThermoFisher, PA5‐17646) Ki‐67 (ThermoFisher, 14–569) Synapsin (Phosphosolution, 1927‐SYNP) NMDAR2D extracellular (ThermoFisher, PA577425) Nestin (Abcam, ab105389) Calretinin (Abcam, ab92341)	1:1000 1:500 1:250 1:500 1:200 1:500 1:500	Anti‐goat Alexa 488 Anti‐rat Alexa 555 Anti‐mouse Alexa 594 Anti‐rabbit Alexa 647	1:500 1:500 1:500 1:500

Protein Visualization

Spheroids were imaged on LSM880 AxioObserverZ1 confocal microscope (Zeiss, Germany) through a Plan‐Apochromat 20× objective (NA = 0.8, Zeiss). The multichannel z‐stack images were successively processed in FIJI for background subtraction and the generation of a maximum projection.^{[ 107 ]}

Morphological Analysis

Spheroids were imaged on an LSM880 AxioObserverZ1 confocal microscope with a 10× EC Plan‐Neofluar (Ph1) objective (NA = 0.3, Zeiss). To evaluate neurite branching complexity and ramification, a semiautomated Sholl analysis centered around the spheroid body with a 20 µm‐step between consecutive hemispheres was performed on ImageJ (Neurite Tracer plugin).^{[ 108 ]} The coefficient (Sholl's decay index) of the linear regression model was used as a measure of the rate of decay of the number of branches with distance from the center of analysis. As a measure of branch ramification, Schoenen index (i.e., maximum number of intersections divided by the number of primary branches) was used. In addition, a semiautomated Strahler analysis was performed to extract the mean branch length. Measurements were carried out on a total of 27 samples, resulting from nine biological replicates for each of the three independent set of experiments.

Western Blotting

Spheroids cultured in CE, CE_rCND1, and CE_CND1 were harvested into cold RIPA buffer by scraping and sonication of the collagen matrix. The total protein concentration of the cell lysates was determined by BCA assay (ThermoFisher, Pierce BCA Protein Assay Kit, # 23225,). The lysates were successively boiled for 10 min at 95 °C in the sample loading buffer. The proteins were electrophoretically resolved on a 10% SDS‐PAGE gel at 100 V. Resolved proteins were transferred to PVDF membranes for 30 min at 20 V using Transblot Turbo (BioRad, USA). After washing, the membranes were blocked in 5% BSA for 1 h at room temperature. Subsequently, the PVDF membrane were blotted with primary antibodies overnight at 4 °C, washed five times with TBST buffer and incubated 2 h at room temperature with peroxidase‐conjugated secondary antibodies. After washing, the membranes were imaged by a ChemiDoc XRS+ (BioRad) system and the bands were analyzed with the ImageJ software. Primary antibodies targeting the following proteins were used: β III‐tubulin (ab78078), MAP2 (PA5‐17646), synapsin (1927‐SYNP), GAP43 (ab16053), and NMDAR2D (PA577425). The relative protein expression was calculated following normalization against the total protein expression. The total protein in each band is displayed in Figure S5 (Supporting Information).

MEA Readings

The spheroids were cultured on Cytoview MEA 48‐well plates (Axion BioSystem, M768‐tMEA‐48B‐5, USA), after precoating with poly‐L‐lysine and laminin. Baseline recording of spontaneous activity was performed in a Maestro MEA system and AxIS software (Axion Biosystems) by using a bandwidth with a filter for 10 Hz to 2.5 kHz cut‐off frequencies. Spikes were detected by using an adaptive threshold set to six times the standard deviation (SD) of the estimated noise on each electrode. Each plate rested for 1 min for acclimatization in the Maestro instrument and was then recorded for an additional 2 min. Multielectrode data analysis was performed using the Axion Biosystems Neural Metrics Tool. Bursts were identified in the data recorded from each individual electrode using an adaptive Poisson surprise algorithm. Network bursts were identified for each well using a nonadaptive algorithm requiring a minimum of 40 spikes and of 25% of active electrodes with a maximum interspike interval of 100 ms.

Based on baseline recording at 14 DIV, the most active wells electrodes (>0.1 spikes s⁻¹) were selected for treatment with either MK‐801 or NBQX. Briefly, the medium of the selected wells was complemented with either 0.1 m MK‐801 or 0.1 m NBQX. After 30 min of incubation, spontaneous neuronal activity was measured for 2 min to identify the effect of the specific blocker. After recordings, the samples were washed 5× with PBS 1× and fresh medium was added. The same procedure was followed for the treatment at 28 DIV. For each experimental condition, a total of three independent experiments were performed using a minimum of 9 wells per plate.

Culture and Imaging of Mouse Primary Cortical Neurons

Primary cortical neurons from QBM Cell Science (Ottawa, Canada ) were suspended in Neurobasal medium complemented with 1% GlutaMAX, 1% pen/strep, 1% B‐27 Plus Supplement, and 0.5% N‐2 Supplement. To incorporate the cells into the hydrogels, the cell suspension and hydrogel solution were mixed in a 1:1 ratio, achieving a final cell density of 5.0 × 10⁶ cells mL⁻¹ and a final collagen concentration of 2 mg mL⁻¹. The cell‐hydrogel mixture was then pipetted into glass bottom 96‐well plates (Greiner Bio‐One, #655892, USA) and incubated at 37 °C for 30 min to allow gelation. Successively, 200 µL of complete culture medium was added to each well. The hydrogels were maintained in culture for up to 7 days with media changes performed every 2 days. At 7 DIV, the samples were fixed in fresh 4% PFA at room temperature for 2 h. Fixed samples were permeabilized with 0.25% Triton‐X100 and blocked with 5% horse serum (ThermoFisher, #31874) overnight at 4 °C. Finally, the samples were stained for β III‐ tubulin and MAP2 and imaged on LSM880 AxioObserverZ1 confocal microscope through a Plan‐Apochromat 20× objective (NA = 0.8, Zeiss). The multichannel z‐stack images were successively processed in FIJI for background subtraction and the generation of a maximum projection.

Data Analysis

Data are reported as mean ± SD or SEM from at least three separate experiments. Data were plotted with GraphPad software, version 8.0. The normality of the distribution was assayed by different tests, such as D'Agostino‐Pearson normality test and Shapiro‐Wilk normality test. For normally distributed data, one‐way ANOVA test followed by Tukey's HSD post hoc test was used. For non‐normally distributed data, Kolmogorov–Smirnov test analyses were carried out. Significance was set at p ≤ 0.05.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Lomboni D. J., Ozgun A., de Medeiros T. V., Staines W., Naccache R., Woulfe J., Variola F., Electroconductive Collagen‐Carbon Nanodots Nanocomposite Elicits Neurite Outgrowth, Supports Neurogenic Differentiation and Accelerates Electrophysiological Maturation of Neural Progenitor Spheroids. Adv. Healthcare Mater. 2024, 13, 2301894. 10.1002/adhm.202301894

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.