Iron‐Chelated Silk Microfiber/Hydrogel Composites as Injectable, Magnetically Aligned Cell Guidance Scaffolds

ABSTRACT

In the design of biomaterial scaffolds for traumatic spinal cord injury (SCI), anisotropic composites of magnetically aligning architectures and injectable hydrogels have garnered increasing interest for their combined neuroregenerative and neuroprotective advantages, uniquely supporting both axonal orientation cues and minimally invasive delivery. Herein, ferric iron‐chelated silk fibroin microfibers (Fe³⁺‐mSF) are explored as a magnetic nanoparticle‐free alternative to existing magneto‐responsive fiber architectures. Functionalization of nascent mSF via iron chelation exhibited stability at physiological pH, with minimal iron release observed over 7‐d and 3‐month periods. Mechanically, the addition of Fe³⁺‐mSF to an existing injectable, in situ crosslinking hydrogel did not compromise its injectability, gelation kinetics, or low swelling profile. Fe³⁺‐mSF cytocompatibility was observed in SH‐SY5Y cells (neuronal analog), both when cultured with Fe³⁺‐mSF‐supplemented media and when cultured within Fe³⁺‐mSF/collagen hydrogels, and the addition of Fe³⁺‐mSF to a collagen‐only hydrogel resulted in a few significant changes in neuroregenerative biomarker expression. Finally, in a proof‐of‐principle study, magnetically‐actuated alignment of Fe³⁺‐mSF/hydrogel composites was achieved on the patient bed of a magnetic resonance imaging (MRI) machine, without obvious pulling of the fiber architecture from the hydrogel bulk. It is envisioned that this Fe³⁺‐mSF‐laden hydrogel system may be used as an injectable, in situ aligning neuroregenerative scaffold.

Ferric iron‐chelated silk fibroin microfibers (Fe³⁺‐mSF) are explored as a magnetic nanoparticle‐free alternative to existing magneto‐responsive nerve guidance architectures. Fe³⁺‐mSF release minimal iron at physiological pH, and, when incorporated into existing injectable hydrogels, do not compromise favorable mechanical properties (syringeability, gelation time, low swelling profile), exhibit cytocompatibility, and may increase the neuroregenerative potential of the scaffold system.

1. Introduction

Regenerative medicine research for neurotraumatic injuries has increasingly emphasized biomaterial‐based scaffolds and therapeutic delivery systems, especially for applications in traumatic spinal cord injuries (SCI) [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. By physically bridging the lesion and alleviating environmental inhibitors of innate neuroregeneration, these biomaterials provide biomechanical and biochemical cues to guide functionally integrative cell migration, maturation, and organization across the injury site [1, 12]. Both injectable and implantable biomaterial systems have been explored in efforts to facilitate neurorecovery post‐SCI, each offering unique therapeutic advantages. Injectable, in situ crosslinking biomaterials are advantageous in their minimally invasive delivery and high‐fidelity conformation to heterogeneous lesion geometries, limiting counterproductive exacerbation of secondary damage cascades and mitigating fibrotic capsule formation [10, 12, 19, 20, 21]. Unfortunately, in situ gelation also sacrifices control over internal architecture, limiting opportunities for aligned biophysical cues to direct trans‐lesion axonal elongation and functional integration [22]. Implantable biomaterials are advantageous because their hierarchical morphology and internal architectures can be precisely controlled to produce aligned features, which are critical for directing axonal outgrowth and reconnection across the SCI lesion [2, 10, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35]. However, surgical implantation of these systems exacerbates neuroinflammation, and imperfect fitting to the lesion site results in fibrotic scar tissue deposition, further inhibiting axonal ingrowth. As highlighted by this tradeoff, current design paradigms for injectable and implantable scaffold systems are inherently limited in their neuroregenerative and neuroprotective contributions to neurorecovery post‐SCI.

Combining the benefits of minimally invasive delivery and structured cues for lesion‐bridging neurite outgrowth, innovative neuroregenerative strategies for SCI have recently explored injectable biomaterials with in situ aligning guidance architectures, i.e., scaffold features that can be aligned post‐delivery in response to the controlled application of an external stimulus (e.g., a magnetic field) [22, 27, 30]. For instance, Tran et al. imparted magneto‐responsive properties to the self‐assembling peptide amphiphile RADA‐16I via magnetic nanoparticle (MNP) doping [35]. While the alignment of RADA‐16I fibrils is typically flow‐induced during injection, requiring a complex, potentially traumatic micromanipulator‐assisted needle retraction scheme, Tran et al. demonstrated that incorporating MNPs into the hydrogel bulk facilitates minimally invasive post‐injection alignment via an externally applied magnetic field.

Alternatively, several research groups have explored the neuroregenerative potential of in situ aligning anisotropic hydrogels, i.e., injectable hydrogel bulks into which magneto‐responsive fiber architectures are incorporated [27, 29, 30, 37, 38]. The internal architectures can be aligned in situ within the hydrogel precursor via externally applied magnetic fields, after which their alignment is then preserved via in situ hydrogel gelation. These “anisogels” are especially attractive for biomaterial‐based SCI treatments, providing the neuroregenerative benefits of rostro‐caudal guidance cues for neurite outgrowth without sacrificing the neuroprotective benefits of minimally invasive delivery. Recent publications by Rose et al. and Omidinia‐Anarkoli et al. represent the first reports of these magneto‐responsive anisogels in neuroregenerative applications. By photo‐crosslinking MNP‐doped polyethylene glycol (PEG) solutions in cylindrical molds and electrospinning MNP‐doped fibrin solutions, respectively, the authors produced magneto‐responsive microgels (1 × 10 µm) and microfibers (5 × 50 µm) for subsequent suspension in fibrin or collagen hydrogels [27, 30]. A few years later, a similar approach for anisogel synthesis was reported by Ghaderinejad et al., demonstrating the neuroregenerative potential of a polycaprolactone microfiber/alginate hydrogel composite []. Additionally, Johnson et al. developed magneto‐responsive cylindrical constructs by rolling electrospun mats of aligned, MNP‐doped poly(L‐lactic acid) nanofibers [].

The mechanism by which magneto‐responsive properties are imparted to each of these existing nerve guidance architectures is the same: doping of superparamagnetic iron nanoparticles (SPIONs) directly into the fibers. Compared with other metal‐oxide‐based MNPs, SPIONs are particularly beneficial for biomaterial development due to their high magnetic saturation, imparting substantial magneto‐responsiveness with minimal increase in the material's metal content. While SPIONs are thereby associated with reduced metal toxicity risks, both ferrous and ferric oxides exhibit greater toxicity than iron alone [39, 40, 41, 42].

To this end, ferric iron chelated silk fibroin microfibers (Fe³⁺‐mSF) are evaluated as an MNP‐free alternative to existing magnetically aligning nerve guidance architectures. Derived from Bombyx mori silkworm cocoons, silk fibroin (SF) is a naturally occurring, FDA‐approved biomaterial with an inherently fibrous structure, making it well‐suited for applications as nerve guidance architectures [44, 45, 46, 47, 48, 49, 50]. Previously, we demonstrated the feasibility of imparting magneto‐responsive properties to SF microfibers (mSF) via ferric iron chelation, achieving a comparable degree of fiber alignment in Fe³⁺‐mSF to that reported for MNP‐doped nerve guidance architectures [30, 43]. Herein, the effects of the proposed Fe³⁺‐mSF architecture on the mechanical functionality and neuroregenerative potential of existing injectable hydrogel‐only scaffold systems are investigated.

2. Results and Discussion

2.1. Ferric Iron‐Chelated Silk Fibroin Microfibers (Fe³⁺‐mSF) Release Minimal Iron

To better understand the implications of incorporating the proposed architecture into existing hydrogel‐only systems, both the short‐term and long‐term stability of the Fe³⁺‐mSF iron coordination was assessed by measuring Fe³⁺‐mSF iron release over time. Though the metal‐binding capacity of SF is well documented, the mechanisms that facilitate this bioinorganic coordination chemistry remain a subject of ongoing research [51, 52, 53, 54, 55]. With regard to iron, specifically, prior experimental investigations indicate that the coordination mechanism, and thereby the coordination stability, is dependent on the metal's oxidation state [55]. Not only does ferric iron tend to interact with different SF moieties than ferrous iron, but the more positively charged ferric iron ion (Fe³⁺) also exhibits stronger electrostatic interactions with negatively charged SF than the less positively charged ferrous iron ion (Fe²⁺) [55]. While we previously showed that iron is chelated by mSF from ferric chloride solution (FeCl₃) in the ferric oxidation state (Fe³⁺), we also demonstrated that the iron content of the synthesized product (Fe³⁺‐mSF) increases concomitantly with increasing concentration of the FeCl₃ solution, warranting investigation of how Fe³⁺‐mSF iron content affects retention of the loaded iron content (i.e., iron release) from the material [50].

Analysis of Fe³⁺‐mSF iron release via inductively coupled plasma mass spectrometry (ICP‐MS) indicated that iron release increases concomitantly with increasing Fe³⁺‐mSF iron content. However, regardless of Fe³⁺‐mSF iron content, less than 1% of the total loaded iron was released, and the mass of iron released was minimal in the context of physiologically normal iron levels observed in both the cerebrospinal fluid (CSF) and serum (Figure 1) [43].

FIGURE 1.

Cumulative Iron release from Fe³⁺‐mSF in phosphate‐buffered saline. (A) Short‐term (through 7 days) iron release as quantified via ICP‐MS; data are presented as the mean ± SEM cumulative iron release (µg) per unit mass Fe³⁺‐mSF (mg) from three independent replicates, with each condition prepared and evaluated in triplicate (N = 3, n = 3). (B) Long‐term (through 91 days) iron release as quantified via ICP‐MS; data are presented as the mean ± SD cumulative iron release (µg) per unit mass Fe³⁺‐mSF (mg) from one independent experiment with samples prepared and evaluated in triplicate (N = 1, n = 3).

In both the short‐term and long‐term studies, intraday and cumulative mass iron release from Fe³⁺‐mSF were significantly affected by the material's initial iron content (two‐way ANOVAs, Dunnett's multiple comparisons tests). Notably, all iron‐modified Fe³⁺‐mSF exhibited the greatest iron release at the 1‐d timepoint (Figure S1). After the first day, no significant differences in intraday iron release were observed from any Fe³⁺‐mSF in the short‐term study, regardless of Fe³⁺‐mSF iron content. In the long‐term study, only the “high iron” Fe³⁺‐mSF and “moderately high iron” Fe³⁺‐mSF exhibited significant differences in intraday release, both of which ceased within the duration of the 91‐d study (i.e., at the 42‐d timepoint and 35‐d timepoint, respectively). The initial burst release of iron observed from all Fe³⁺‐mSF is unlikely to be the result of residual unbound iron, as ICP‐MS analysis of the pre‐release Fe³⁺‐mSF wash steps demonstrated sufficient removal of remaining free iron post‐synthesis (Figure S2). Rather, the initial burst release may indicate weak electrostatic iron coordination mechanisms in the washed Fe³⁺‐mSF product: though prior experimental investigation has validated ferric iron chelation at oxygen‐containing residues in SF, it is possible that additional ferric iron ions are either electrostatically bound to the negatively charged mSF or otherwise entrapped in the material during binding‐induced conformational changes (e.g., ferric‐iron‐induced beta sheeting) [43, 56]. As the bio‐adsorption properties of SF are sensitive to environmental pH, the initial day‐1 burst release may have resulted from the transfer of the Fe³⁺‐mSF samples from the acidic FeCl₃ solutions in which they were synthesized to the neutral (pH 7.4) phosphate‐buffered saline (PBS) solutions in which the release studies were conducted; the subsequently diminishing intraday releases may have resulted from the stabilization of the electrostatic interactions [52, 56, 57, 58].

Consistent with the intraday release results, the greatest cumulative mass iron release was observed from the “high iron” Fe³⁺‐mSF, which exhibited significantly greater iron release than the “no iron added” Fe³⁺‐mSF control at every timepoint in both the short‐term and long‐term studies. Similarly, the “moderately high iron” Fe³⁺‐mSF and “moderate iron” Fe³⁺‐mSF exhibited significantly greater cumulative iron release than the “no iron added” Fe³⁺‐mSF control at every timepoint in the short‐term study, though only the former exhibited comparatively significant iron release in the long‐term study. No significant iron release was observed from the “moderately low iron” Fe³⁺‐mSF or “low iron” Fe³⁺‐mSF by comparison to the “no iron added” Fe³⁺‐mSF. As expected, the “no iron added” Fe³⁺‐mSF control, i.e., nascent silk fibroin microfibers, exhibited the least cumulative iron release over the duration of both the short‐term and long‐term studies. Though trace quantities of iron were unexpectedly detected in the release medium samples collected from the “no iron added” Fe³⁺‐mSF, these observations may be attributed to confounding factors including minor iron contributions from the water in which the mSF were soaked (sham “0 mM” FeCl₃ treatment) and subsequently washed, cross‐sample contamination due to the use of an autosampler during ICP‐MS measurements, and minor iron accumulation in the ICP‐MS instrument over the course of sample set analyses. Collectively, these results indicate that, while Fe³⁺‐mSF iron release in PBS increases with Fe³⁺‐mSF iron content, especially during the first 24 h, it ultimately plateaus within several weeks.

Importantly, independent of release profile, the short‐term and long‐term iron release observed from Fe³⁺‐mSF was minimal by comparison to the iron “loaded” (i.e., bound during Fe³⁺‐mSF synthesis). All Fe³⁺‐mSF released only 1% of the total bound iron content (Figure S3, Table S1), indicating that ferric iron chelation offers a stable mechanism for Fe³⁺ functionalization of mSF at physiological pH. The stability of this coordination is promising for long‐term material biocompatibility. For contextualization, cumulative intra‐lesion mass iron release of an Fe³⁺‐mSF/hydrogel composite was estimated with assumptions of a 1 mg mL⁻¹ fiber density (i.e., mass Fe³⁺‐mSF per unit volume bulk hydrogel) and a total composite volume of 2 cm³ (i.e., an average‐to‐large volume for the cavitated lesion site in SCI patients) [59]. Under these assumptions, the greatest intraday iron release observed (“moderately high iron” Fe³⁺‐mSF, day 1 of short‐term study) would translate to an intra‐lesion iron release of 0.3245 µg, while the greatest cumulative iron release observed (“high iron” Fe³⁺‐mSF, long‐term study) would translate to an intra‐lesion iron release of 0.6294 µg over 91 days. Assuming no clearance of ferric iron from the CSF into the serum and a total CSF volume of 150 mL perfusing the body at any given time, the greatest cumulative iron release observed would increase CSF iron levels by only 4.2 µg L⁻¹ [61]. Assuming clearance to the serum (5000 mL circulating volume), this effect would be even less significant, increasing serum iron levels by only 0.13 µg L⁻¹ [62]. When compared against normal iron levels for both the cerebrospinal fluid (CSF) and serum (61.01 ± 18.3 and 1332.7 ± 408.2 µg L⁻¹, respectively), these estimates offer preliminary in vitro evidence of physiologically minimal iron release from all Fe³⁺‐mSF examined in this study [60]. As Fe³⁺‐mSF magneto‐responsiveness has been shown to increase with increasing iron content, with “high iron” Fe³⁺‐mSF achieving target alignment uniformity for nerve guidance architectures, these results support continued exploration of Fe³⁺‐mSF as an in situ aligning architecture for injectable, neuroregenerative hydrogel scaffolds [30, 43].

2.2. Fe³⁺‐mSF Architecture Does Not Detract From Mechanical Functionality of Existing Injectable Hyaluronic Acid (HA) Hydrogels

To examine the effect of Fe³⁺‐mSF on the mechanical properties of an existing injectable, in situ crosslinking scaffold, the proposed architecture was incorporated into a hyaluronic acid (HA)‐based hydrogel system with demonstrated translational relevance in traumatic SCI treatment. HA‐based hydrogels have emerged as a favorable biomaterial for neuroregenerative scaffolds and cell‐delivery vehicles [12, 18, 21, 63, 64, 65, 66, 67]. In the context of chronic SCI, HA‐based hydrogels are beneficial because they both mitigate inhibitory responses and enhance endogenous neuroregenerative responses. As a primary component of native spinal cord extracellular matrix, HA hydrogels exhibit extraordinary biocompatibility, cytocompatibility, and immunomodulatory properties [68, 69, 70, 71]. HA positively engages with the ubiquitous neural cell surface receptor, CD44, to promote cell migration, proliferation, and differentiation. Furthermore, it has the potential to bind, organize, and present a wide variety of bioactive molecules employed in neuroregenerative processes [67, 68, 71, 72, 73].

While HA will not self‐assemble into a hydrogel, the requisite crosslinking mechanism offers an opportunity to tailor the properties of the final material to the desired application [74, 75]. Given the high solubility of HA and its susceptibility to rapid enzymatic degradation in vivo, HA‐based hydrogels for chronic SCI must be synthesized via highly resilient crosslinkers for increased longevity. Physical crosslinking mechanisms are disadvantageous as they yield unstable, rapidly degrading gels (days), while chemical crosslinking mechanisms are disadvantageous as potential crosslinker cytotoxicity and/or side reactivity with native proteins detract from hydrogel biocompatibility [66]. To this end, Diels‐Alder (DA) click chemistry offers a promising approach to the facile synthesis of highly biocompatible HA hydrogels. Consequently, the hydrogel system selected for material characterization studies consists of furan‐modified HA (FFA‐HA) and polyethylene glycol‐bismaleimide (PEG‐(MI)₂), in which the maleimide groups of the PEG‐(MI)₂ (dienophile) facilitates crosslinking of the furan groups of the FFA‐HA (diene) via a DA click reaction (4‐2 cycloaddition) (Figure S4) [76]. Synthesized per previously reported methodology, the FFA‐HA/PEG‐(MI)₂ system exhibits favorable properties for efficacious performance as an injectable, in situ crosslinking, lesion‐bridging scaffold for traumatic SCI [66, 67]. Additionally, given the potential for metal‐ligand interaction between HA and ferric iron ions, the use of an HA‐based system allows for a more comprehensive investigation of the effect of the added Fe³⁺‐mSF architecture on the bulk hydrogel's gross mechanical properties [74].

2.2.1. Hydrogel Syringeability is Preserved With Added Fe³⁺‐mSF Architecture

Syringeability is a critical design consideration for facilitating the minimally invasive delivery of an in situ‐crosslinking scaffold system: the translational potential of these systems is highly dependent on users’ ability to inject the material in a controlled manner with reasonable effort via a clinically relevant injection scheme [77]. Though there are currently no clinical trials evaluating injectable biomaterials for SCI, simulation of Fe³⁺‐mSF/hydrogel injection was informed by current clinical practice for delivering cell therapies and pharmacological agents to the spinal cord [78, 79, 80]. The ISO 7886‐1 standard (sterile hypodermic syringes for single use) and human factors studies identify an upper threshold injection force of 40 to 50 N as a metric of reasonable syringeability [77]. In the context of injectable biomaterials, specifically, Robinson et al. reports participants’ perceived syringeability of 5 mL biomaterial volumes when injected at a rate of 1 mm s⁻¹, ultimately correlating an injection force of 12 N or less to “easily injectable,” 12 to 38 N as “injectable with considerable effort,” 38 N to 64 N as “injectable with great effort,” and greater than 64 N as “entirely non‐injectable” [81]. In evaluating the syringeability of the Fe³⁺‐mSF/hydrogel composites, both maximum and average injection forces were calculated from the force‐displacement curve generated during simulated injection from a 0.5 mL syringe. Maximum injection force was identified as the peak force measured across the entire curve, while average injection force was calculated as the mean force measured across the final 0.5 mm of the plateau region (Figure S5).

The syringeability of the existing HA hydrogel system was not adversely affected by the addition of the Fe³⁺‐mSF architecture or by manipulating parameters of fiber density and iron content (Figure 2). When comparing the forces‐displacement curves of the Fe³⁺‐mSF/hydrogel composites to those generated for the hydrogel‐only control, only fiber density, not iron content, impacted maximum injection force (p = 0.0057 and p = 0.6783, respectively; two‐way ANOVA, Dunnett's multiple comparison test). While the maximum injection forces of both the 0.5 and 2 mg mL⁻¹ 0 mM Fe³⁺‐mSF/hydrogel composites were significantly different from that of the hydrogel‐only control, they were lower than that of the control, indicating syrinegability was not negatively impacted by the added Fe³⁺‐mSF. Furthermore, neither fiber density nor iron content of the added Fe³⁺‐mSF impacted the average injection force (p = 0.0541 and p = 0.5810, respectively; two‐way ANOVA, Dunnett's multiple comparison test). When comparing syringeability among Fe³⁺‐mSF/hydrogel composites (i.e., excluding the hydrogel‐only control condition), variations in fiber density and iron content of the added Fe³⁺‐mSF architecture did not affect the maximum or average injection forces (two‐way ANOVAs, Tukey's HSD test).

FIGURE 2.

Effect of Fe³⁺‐mSF on FFA‐HA hydrogel syringeability. (A) Maximum and (B) average force required to eject pre‐gelled Fe³⁺‐mSF/hydrogel composite from a 0.5 mL syringe fitted with a 27 G needle. Data are presented as mean ± SD from three independent experiments (n = 9). Statistical differences in injection forces observed in Fe³⁺‐mSF/hydrogel composites relative to hydrogel‐only samples were determined via two‐way ANOVA followed by Dunnett's multiple comparisons test; statistical differences among unique Fe³⁺‐mSF/hydrogel composite compositions were determined via two‐way ANOVA followed by Tukey's HSD test (^* p < 0.05).

In terms of clinical relevance, these results indicate that Fe³⁺‐mSF architecture can be incorporated into existing injectable hydrogel systems without detracting from their syringeability. The maximum and average injection forces measured for the Fe³⁺‐mSF/hydrogel composites were not only well below the threshold set by regulatory standards, but also within the “easily injectable” range established by human factor studies, supporting the material's translational potential.

2.2.2. Hydrogel Gelation is not Impeded by Added Fe³⁺‐mSF Architecture

Crosslinking kinetics represent another key design consideration in the performance of injectable hydrogel systems, determining the material's capacity to form a lesion‐bridging scaffold in situ [17, 82]. Not only should the fully crosslinked hydrogel mechanically mimic native tissue, supporting functionally integrative regeneration, but the rate of crosslinking should also allow time for injection and lesion filling without compromising hydrogel retention or apposition to native tissue [17]. To this end, the effects of Fe³⁺‐mSF on the viscoelastic properties (Figure 3) and critical gelation time (Figure 4) of FFA‐HA hydrogels were examined via oscillatory rheology. Given the minimal iron release observed (Section 2.1), Fe³⁺‐mSF iron content was varied only to the extreme levels of “no iron added” and “high iron.”

FIGURE 3.

Effect of Fe³⁺‐mSF on FFA‐HA hydrogel viscoelastic behavior (oscillatory rheology). (A, B) Shear strain sweep examining strain‐dependent changes in the storage modulus (G′, Pa) and loss modulus (G″, Pa) of pre‐formed (i.e., click‐crosslinked) hydrogel composites containing either (A) “no iron added” Fe³⁺‐mSF or (B) “high iron” Fe³⁺‐mSF at densities of 0.5, 1.0, and 2.0 mg mL⁻¹ hydrogel, with a 0 mg mL⁻¹ Fe³⁺–mSF hydrogel‐only control. Data are presented as mean ± SD from three independent experiments (N = 3). (C–F) Frequency sweep examining time‐dependent changes in the storage and loss moduli (G′ and G″, Pa) as well as the complex viscosity (η*, Pa·s) of pre‐formed (i.e., click‐crosslinked) hydrogel composites containing either (C, E) “no iron added” Fe³⁺‐mSF or (D, F) “high iron” Fe³⁺‐mSF at densities of 0.5, 1.0, and 2.0 mg mL⁻¹ hydrogel, with a 0 mg mL⁻¹ Fe³⁺‐mSF hydrogel‐only control. Data are presented as mean ± SD from three independent experiments, each with triplicate sample sets (N = 3, n = 3).

FIGURE 4.

Effect of Fe³⁺‐mSF on FFA‐HA hydrogel critical gelation time (oscillatory rheology). Oscillatory time sweep examining time‐dependent changes in the storage modulus (G′, Pa) and loss modulus (G″, Pa) of Fe³⁺–mSF/hydrogel composite during gelation to obtain the (A) critical gelation time from (B) 0 mg mL⁻¹ Fe³⁺–mSF hydrogel‐only control or (C, E, G) “no iron added” Fe³⁺‐mSF and (D, F, H) “high iron” Fe³⁺‐mSF at densities of 0.5, 1.0, and 2.0 mg mL⁻¹ hydrogel, respectively. The critical gelation times of Fe³⁺‐mSF/hydrogel composites were determined by the crossover point in the respective storage and loss moduli (G′ and G″, Pa) and are presented as the mean ± SD of three independent experiments (N = 3). Storage and loss moduli at each time point are presented as the mean of three independent experiments (N = 3).

Regarding viscoelastic properties, the added Fe³⁺‐mSF architecture did not significantly affect the storage or loss moduli of the crosslinked bulk HA hydrogel. Following strain sweeps (Figure 3A,B), frequency sweeps were performed to characterize time‐dependent viscoelastic behaviors of pre‐crosslinked Fe³⁺‐mSF/hydrogel composites when subjected to oscillatory shear strain within the identified linear viscoelastic region (Figure 3C,D). Comparison of the storage and loss moduli measured for each Fe³⁺‐mSF hydrogel composition revealed no significant differences from the hydrogel‐only control moduli at the respective frequencies (one‐way ANOVA, Dunnett's multiple comparisons test). Furthermore, no significant differences in storage or loss moduli were observed when comparing the effects of Fe³⁺‐mSF fiber density and iron content (two‐way ANOVA, Tukey's HSD test).

In addition to storage and loss moduli, complex viscosities of the Fe³⁺‐mSF/hydrogel composites were evaluated during the oscillatory frequency sweep (Figure 3E,F). No significant differences were observed across all groups, suggesting that the fibers did not impact the complex viscosity. Within all experimental groups, the complex viscosities of both the Fe³⁺‐mSF/hydrogel composites and the hydrogel‐only control were significantly affected by the applied frequency (p < 0.0001), suggesting frequency‐dependent viscoelastic behavior. As the rate of applied oscillatory shear was increased from 0.0159 Hz to 3.66 Hz, the complex viscosity of all sample compositions decreased, indicative of shear‐thinning behavior [83]. With further increases in oscillatory shear rate beyond 3.66 Hz, the complex viscosity of all sample compositions subsequently increased, consistent with the reduced capacity for energy dissipation observed in the storage and loss moduli trends at the corresponding frequencies. These observations are consistent with the shear‐thinning behavior observed in the existing hydrogel‐only system, a viscoelastic property that not only facilitates syringeability but also mimics the behavior of native spinal cord extracellular matrix [66, 82, 84].

In addition to its effects on the viscoelastic properties of the crosslinked hydrogel system, the effect of the added Fe³⁺‐mSF on the gelation kinetics was examined via oscillatory time sweeps (Figure 4 and Table 1). No significant differences were observed between the storage/loss moduli crossover point of the Fe³⁺‐mSF/hydrogel composite materials and that of the hydrogel‐only control (p = 0.7591, one‐way ANOVA, Dunnett's multiple comparisons test). Furthermore, no significant differences were observed in storage/loss moduli crossover point when comparing the different Fe³⁺‐mSF/hydrogel composite groups (p = 0.7591, one‐way ANOVA, Tukey's HSD test). The lack of significant differences in critical gelation times suggests that the addition of Fe³⁺‐mSF does not impede crosslinking of the FFA‐HA/PEG(MI)₂ hydrogel bulk, illustrating the translational potential of Fe³⁺‐mSF as an additive scaffold architecture for existing in situ crosslinking hydrogel systems. As in the hydrogel‐only control, the post‐critical gelation storage and loss moduli plateaus observed for all Fe³⁺‐mSF/hydrogel composites are consistent with mechanical properties shown to support cell attachment and proliferation in neural tissue engineering applications, further demonstrating the material's translational relevance as a functionally integrative scaffold system [10, 85].

TABLE 1.

Critical gelation time for Fe³⁺‐mSF/FFA‐HA hydrogel composites.

Sample composition	Critical gelation time (min)
Hydrogel‐only control	21.67 ± 6.01
0.5 mg mL−1, “No Iron Added” Fe3+‐mSF	18.00 ± 7.00
1 mg mL−1, “No Iron Added” Fe3+‐mSF	21.00 ± 7.76
2 mg mL−1, “No Iron Added” Fe3+‐mSF	15.50 ± 5.68
0.5 mg mL−1, “High Iron” Fe3+‐mSF	24.17 ± 6.71
1 mg mL−1, “High Iron” Fe3+‐mSF	22.50 ± 10.97
2 mg mL−1, “High Iron” Fe3+‐mSF	22.33 ± 10.98
The critical gelation times of Fe3+‐mSF/FFA‐HA hydrogel composites were determined by the crossover point in the respective storage and loss moduli (G’ and G”, Pa) and are presented as the mean ± SD of three independent experiments (N = 3).

2.2.3. Low Swelling Profile of Hydrogel is Preserved With Added Fe³⁺‐mSF Architecture

To minimize secondary damage to the surrounding tissue, lesion‐bridging biomaterial scaffolds should exhibit a low swelling profile, thereby reducing disruptions to intraspinal pressure [86, 87]. Compared to the FFA‐HA/PEG(MI)₂ hydrogel‐only control, the Fe³⁺‐mSF/hydrogel composites did not exhibit significant differences in swelling ratio, i.e., the ratio of initial mass to post‐swelling mass (Figure 5), after 24 h submerged in PBS (pH 7.4, incubated at 37°C). Not only was the swelling profile of the hydrogel preserved with the addition of the Fe³⁺‐mSF architecture, but it was also consistent with what is termed “low swelling” in the design of hydrogels for spinal cord repair (33% or less increase in swollen mass from retained fluid) [88]. Furthermore, the swelling ratios observed for all Fe³⁺‐mSF/hydrogel compositions were well below that currently observed in existing hydrogel‐based dural sealants (50%–400%) [88]. Given that the excessive swelling observed in these dural sealants has been counterproductive to their clinical implementation, this comparatively low swelling ratio is promising for translation.

FIGURE 5.

Effect of Fe³⁺‐mSF on FFA‐HA hydrogel 24‐h swelling behavior (PBS). (A) Swelling ratio of pre‐formed (click‐crosslinked) Fe³⁺‐mSF/hydrogel composites synthesized with Fe³⁺‐mSF of varied iron content (increasing with darkening grayscale) at 1 mg mL⁻¹ fiber density, with a 0 mg mL⁻¹ hydrogel‐only control (white column). Data are presented as the mean ± SD of three to five samples (N = 1, n = 3–5). (B) Swelling ratio of Fe³⁺‐mSF/hydrogel composites synthesized with Fe³⁺‐mSF of varied iron content (blue: “no iron added” Fe³⁺‐mSF, or orange: “high iron” Fe³⁺‐mSF) and varied fiber density (0.5, 1.0, and 2.0 mg mL⁻¹, increasing with darkening blue/orange color scale), with a 0 mg mL⁻¹ Fe³⁺‐mSF hydrogel‐only control (white column). Data are presented as the mean ± SD of four to five samples (N = 1, n = 4–5).

2.3. Fe³⁺‐mSF Exhibit Cytocompatibility With SH‐SY5Y Cells (Neuronal Analog)

With evidence of Fe³⁺‐mSF cytocompatibility previously demonstrated with normal human dermal fibroblasts, further cytocompatibility assessments of the proposed architecture were conducted with a commonly used neuronal analog, i.e., the neuroblastoma‐derived SH‐SY5Y cell line [43, 89]. To investigate whether Fe³⁺‐mSF exposure elicits a cytotoxic response, SH‐SY5Y cell metabolic activity levels were quantified via the resazurin assay after 1, 3, 5, and 7 d of continuous Fe³⁺‐mSF exposure. SH‐SY5Y cells were exposed to Fe³⁺‐mSF both as an independent biomaterial, with Fe³⁺‐mSF suspended in their culture media, and as an additive scaffold architecture, with Fe³⁺‐mSF incorporated into cell‐seeded collagen hydrogels (Figure 6). Collagen was selected as the bulk hydrogel component for both its relevance as a scaffold for SCI neuroregeneration and its inherent capacity to support cell adhesion [13, 90, 91]. The use of a bulk hydrogel material with demonstrated cell adhesion not only reduced confounding variables for cytocompatibility assessments but also supported later characterization of changes in cytoskeletal biomarker expression associated with the neuroregenerative outcomes of axonal outgrowth and extension (Section 2.4).

FIGURE 6.

Cytocompatibility response of SH‐SY5Y Cells to Fe³⁺‐mSF exposure in vitro. (A) Metabolic activity of undifferentiated SH‐SY5Y cells cultured on tissue‐culture treated polystyrene with 0.5 mg mL⁻¹ Fe³⁺‐mSF in complete media, normalized to metabolic activity of an untreated control (i.e., SH‐SY5Y cells cultured in growth media only, no Fe³⁺‐mSF) at respective exposure timepoint. (B) Metabolic activity of undifferentiated SH‐SY5Y cells cultured in 200 µL collagen I hydrogels containing 0.5 mg mL⁻¹ Fe³⁺‐mSF, normalized to metabolic activity of an untreated control (i.e., SH‐SY5Y cells cultured in collagen hydrogel‐only scaffold, no Fe³⁺‐mSF) at respective exposure timepoint. Metabolic activity levels were assessed via the resazurin reduction assay after 1, 3, 5, and 7 d exposure to Fe³⁺‐mSF of varied iron modification (reported in the legend). Data are presented as mean ± SEM from three independent replicates, with each condition evaluated six times per replicate when assessing Fe³⁺‐mSF as an independent biomaterial (A: N = 3, n = 6) or three to five times per replicate when assessing Fe³⁺‐mSF as an architectural component of composite collagen hydrogels (B: N = 3, n = 3–5). No statistical differences were observed relative to the non‐exposed control conditions or among varied Fe³⁺‐mSF iron content as determined by a two‐way ANOVA.

In studies assessing Fe³⁺‐mSF as an independent biomaterial, SH‐SY5Y cell metabolic activity was examined across variable Fe³⁺‐mSF iron content levels (noted in Figure 6 legend). When normalized to the non‐ Fe³⁺‐mSF ‐exposed control (SH‐SY5Y cells only), the metabolic activity of Fe³⁺‐mSF‐exposed SH‐SY5Y cells showed no significant differences from that of the non‐exposed SH‐SY5Y cells at the respective time point (p = 0.9247; two‐way ANOVA, Dunnett's multiple comparisons test) (Figure 6A). Furthermore, no significant differences were observed in the metabolic activity of SH‐SY5Y cells when compared across Fe³⁺‐mSF iron content groups and exposure time (p = 0.9729 and p = 0.0602, respectively; two‐way ANOVA, Tukey's HSD test).

With no statistically significant differences in viability observed across variable levels of Fe³⁺‐mSF iron content in the two‐dimensional studies, experimental groups were down‐selected for the three‐dimensional studies. Specifically, changes in the metabolic activity of SH‐SY5Y cells in Fe³⁺‐mSF‐exposed cultures were examined with cells cultured in collagen hydrogels containing either “no iron added” Fe³⁺‐mSF or “high iron” Fe³⁺‐mSF. The reported metabolic activity was normalized to the metabolic activity of SH‐SY5Y cells cultured in collagen‐only hydrogels at the respective time point. Neither the added Fe³⁺‐mSF architecture nor the exposure time had a significant effect on the metabolic activity levels of the SH‐SY5Y cells (p = 0.2709 and p = 0.7924, respectively; two‐way ANOVA) (Figure 6B). No significant differences were observed between the collagen hydrogel composites prepared with “high iron” mM Fe³⁺‐mSF and those prepared with “no iron added” Fe³⁺‐mSF, indicating the iron content of the Fe³⁺‐mSF did not significantly affect the cytocompatibility of the composite scaffold system.

Collectively, these results suggest that neither Fe³⁺‐mSF iron content nor Fe³⁺‐mSF exposure time affects the metabolic activity of SH‐SY5Y cells, which is consistent with the minimal iron release observed from Fe³⁺‐mSF in PBS (Section 2.1). As Fe³⁺‐mSF magneto‐responsiveness has been shown to increase with increasing iron content, the cytocompatibility observed across all Fe³⁺‐mSF groups suggests that the highest‐iron‐loading synthesis scheme examined herein (“high iron,” 150 mM FeCl₃ treatment solution) could be used to increase magnetically actuated alignment uniformity of the scaffold architecture without compromising its cytocompatibility.

2.4. Fe³⁺‐mSF may Increase Neuroregenerative Potential of Collagen Hydrogels In Vitro

Given the intended application of Fe³⁺‐mSF as a nerve guidance architecture, the effect of Fe³⁺‐mSF on the neuroregenerative potential of an existing hydrogel‐only scaffold was also evaluated in vitro. As in the cytocompatibility assessments, characterization of scaffold neuroregenerative potential was conducted using SH‐SY5Y cells seeded in Fe³⁺‐mSF/collagen hydrogel composites. Changes in the expression of several neuroregenerative biomarkers (Table 2) by SH‐SY5Y cells were quantified in response to varied parameters of the collagen‐based scaffold composition, including the addition of Fe³⁺‐mSF architecture (0 mg mL⁻¹ Fe³⁺‐mSF collagen‐only hydrogel, or 1 mg mL⁻¹ Fe³⁺‐mSF/collagen composite), iron modification of the added Fe³⁺‐mSF architecture (“no iron added” Fe³⁺‐mSF or “high iron” Fe³⁺‐mSF, if applicable), and gross scaffold organization (randomly oriented or shear aligned). The effect of varied scaffold composition on neuroregenerative biomarker expression was examined in both non‐differentiated and retinoic acid (RA)‐differentiated SH‐SY5Y cells after 8 days in culture (1‐d seeding recovery, 7‐d differentiation period).

TABLE 2.

Biomarkers of neuroregeneration and axonal regeneration.

Gene	Encoded protein	Expression	Biomarker profile
NES	Nestin	Cytoskeletal	Progenitor, “Neural Stemness”
RBFOX3	Neuronal Nuclear Antigen (NeuN)	Nuclear	Immature, Mature Neuron
ENO2	Neuronal Specific Enolase 2 (NSE)	Cytoplasmic	Immature, Mature Neuron
TUBB3	β‐Tubulin III	Cytoskeletal	Immature Neuron
MAP2	Microtubule‐Associated Protein 2	Cytoskeletal	Mature Neuron
GAP43	Growth Associated Protein 43	Growth Cone Membrane	Extending/Regenerating Axon
SYP	Synaptophysin	Pre‐Synaptic Membrane	Synaptic Neuron

With gene expression internally normalized to the respective randomly oriented collagen‐only scaffold control, few significant differences in neuroregenerative biomarker expression were observed in the non‐differentiated SH‐SY5Y cell cohort in response to varied scaffold composition (one‐way ANOVA, Tukey's HSD test); no significant differences were observed in the RA‐differentiated SH‐SY5Y cell cohort (one‐way ANOVA, Tukey's HSD test) (Table 3 and Figure 7). Unexpectedly, expression of the somatic markers NES (nestin, neural stemness marker), RBFOX3 (neuronal nuclear protein, neuronal maturation marker), and ENO2 (neuronal enolase 2, neuronal maturation marker) was not significantly affected by scaffold composition in either cohort. When differentiating SH‐SY5Y cells toward a neuronal phenotype with RA‐supplemented, serum‐deprived media, an initial upregulation and subsequent downregulation of NES is typically observed along with an overall upregulated expression of RBFOX3 and ENO2 [92, 93, 94]. While those outcomes were not observed herein, the differentiation in 3D culture systems (i.e., collagen hydrogels) with no additional media supplements (i.e., additional neurotrophic factors, including brain‐derived neurotrophic factor or N‐2 supplements) may have delayed or otherwise altered the temporal presentation of these differentiation profiles (expression was evaluated only on day 8) [90, 92, 93, 94, 95, 96].

TABLE 3.

Effect of Fe³⁺‐mSF architecture on neuroregenerative biomarker expression in non‐differentiated vs. RA‐differentiated SH‐SY5Y cells in vitro.

	Non‐differentiated		RA‐differentiated
Biomarker (Gene)	p value	Significance	p value	Significance
NES	0.9158	ns	0.4617	ns
RBFOX3	0.9814	ns	0.3455	ns
ENO2	0.4112	ns	0.3397	ns
TUBB3	0.2139	ns	0.9081	ns
MAP2	0.8250	ns	0.3410	ns
GAP43	0.9659	ns	0.6172	ns
SYP	0.0382	*	0.6755	ns

The table above reports only the results of the one‐way ANOVAs performed for the non‐differentiated and RA‐differentiated quantitative real‐time polymerase chain reaction (qRT‐PCR) data sets (ns – not significant, ^* p < 0.05); post‐hoc analyses not shown.

FIGURE 7.

Effect of Fe³⁺‐mSF architecture on neuroregenerative biomarker expression in non‐differentiated versus RA‐differentiated SH‐SY5Y cells in vitro (qRT‐PCR, internal normalization). Fold change expression of (A) somatic, (B) cytoskeletal, and (C) synaptic biomarkers associated with neuronal differentiation in Fe³⁺‐mSF/collagen‐seeded SH‐SY5Y cells (neuronal analog) when maintained in growth media (“Non‐Differentiated SH‐SY5Y Cells,” light gray) and when maintained in differentiation media (“RA‐Differentiated SH‐SY5Y Cells,” dark gray) in vitro. All normalized fold change expression values are reported relative to the SH‐SY5Y cells seeded in the collagen‐only, randomly oriented scaffold and maintained in the respective media formulation (non‐differentiation growth media or RA‐differentiation media) (i.e., internal experimental control, leftmost light gray bar and leftmost dark gray bar, respectively). Data are presented as the mean ± SEM from three independent experiments with one to three biological replicates per experiment (N = 3, n = 1–3). Accompanying graphics created in BioRender.

Similarly, expression of the presynaptic functional markers GAP43 (growth‐associated protein 43) and SYP (synaptophysin) was largely unaffected by scaffold composition, both of which are typically upregulated during differentiation toward a neuronal phenotype [92, 96, 97]. No significant differences in GAP43 expression were observed across varied scaffold compositions in either cohort. While SYP expression was significantly affected by scaffold composition in the non‐differentiated cohort (Table 3), a significant difference was observed only between two experimental groups, not between an experimental group and the control: aligned “high iron” Fe³⁺‐mSF/collagen was associated with upregulated SYP expression by comparison to the aligned collagen‐only control, but not by comparison to the randomly oriented collagen‐only control. No significant differences in SYP expression were observed across varied scaffold compositions in the RA‐differentiated cohort.

Notably, while not statistically significant, a pattern of increasing expression of cytoskeletal‐associated biomarkers was observed with Fe³⁺‐mSF exposure. In the non‐differentiated cohort, the addition of the Fe³⁺‐mSF fiber architecture, especially the “high iron” Fe³⁺‐mSF, seemed to be associated with upregulated expression of TUBB3 (β‐tubulin III), one of the microtubules in the distal growth cone of extending axons [92, 93, 97, 98, 99]. These trends were not observed in the expression level of the other examined cytoskeletal marker, MAP2 (microtubule‐associated protein 2), possibly indicating early stages of axonal outgrowth. β‐tubulin proteins homodimerize with α‐tubulin III to form the microtubules that extend an axon's growth cone, then proteins like MAP2 are synthesized to stabilize the extensions in both axonal and dendritic projections [98, 99, 100]. Not only is TUBB3 expression observed only in neuronal phenotypes (i.e., not observed in glial or microglial neural phenotypes), indicating differentiation toward a neuronal phenotype, but the β‐tubulin III protein is critical to rapid regenerative axonal extension and elongation, a key step in functional reconnection of disrupted nerve pathways [96, 99].

To examine how the scaffold‐induced neuronal differentiation outcomes compared to those induced by RA media supplementation, gene expression levels in both the non‐differentiated and RA‐differentiated cohorts were then universally normalized to that of the non‐differentiated SH‐SY5Y cells cultured in the randomly oriented collagen‐only control. Under this alternative normalization scheme, significant differences in biomarker expression were primarily observed relative to the media formulation rather than the scaffold composition (Table S3 and Figure S6).

By comparison to the non‐differentiated SH‐SY5Y cells maintained in growth media, the RA‐differentiated SH‐SY5Y cells maintained in RA‐supplemented, serum‐reduced differentiation media generally exhibited greater expression of neuroregenerative biomarkers. Regardless of scaffold composition, experimental groups in the RA‐differentiated cohort generally exhibited equivalent expression of neural stemness marker (NES) and either equivalent or significantly increased expression of neuronal somatic markers (RBFOX3 and ENO2, respectively) when compared to the respective scaffold condition in the growth media cohort, consistent with the more neuronal phenotype expected to be induced by the RA differentiation media. Consistent with previous studies examining neuronal biomarker expression in 7‐d RA differentiated SH‐SY5Y cells in 3D collagen scaffold systems, significantly upregulated TUBB3 expression was observed with no significant change in MAP2 expression, suggesting early‐stage differentiation toward a mature neuronal phenotype [90, 101]. Similarly, equivalent expression of axonal growth cone marker GAP43 and significantly upregulated expression of synaptic marker SYP were observed in the RA differentiation media cohort by comparison to the respective scaffold conditions of the growth media cohort, indicative of differentiation toward a more functionally mature neuronal phenotype [101].

Interestingly, when stratifying by scaffold composition, SH‐SY5Y cells cultured with both randomly oriented and aligned Fe³⁺‐mSF/collagen scaffolds, expression levels of MAP2 and GAP43 in the RA differentiation media cohort were not significantly greater than those observed in the growth media cohort (Figure S6 and Table S3). Furthermore, in those cells cultured with randomly and aligned “high iron” Fe³⁺‐mSF/collagen scaffolds, specifically, the upward‐trending levels of TUBB3 expression observed in the non‐differentiated SH‐SY5Y cells approached those of the RA‐differentiated. While the differences observed were not statistically significant (consistent with previously reported TUBB3 expression levels in response to 7‐d RA‐only differentiation procedures), these trends suggest that exposure to Fe³⁺‐mSF may not only promote more axonal regeneration than the collagen hydrogel alone, but also do so on a comparable scale to that achieved with RA differentiation media [92, 101, 102]. Furthermore, in comparing TUBB3 expression changes in the “high iron” Fe³⁺‐mSF/collagen scaffolds to those observed in the “no iron added” Fe³⁺‐mSF/collagen, the iron content of the architecture appears to increase TUBB3 expression further. Given that the Fe³⁺‐mSF/collagen scaffolds in this study were shear‐aligned rather than magnetically aligned (to reduce confounding effects of magnetic field exposure), this correlation is likely a response to the material itself rather than the alignment mechanism. It is unlikely that TUBB3 expression was increased by released iron, as minimal iron release was observed from the Fe³⁺‐mSF (Section 2.1). It is possible that the added ferric iron (Fe³⁺) reduced the negative charge on the surface of nascent, “no iron added” Fe³⁺‐mSF, thereby improving cell spreading and cytoskeletal attachment to the “high iron” Fe³⁺‐mSF by comparison [103, 104]. Ultimately, while statistically significant Fe³⁺‐mSF ‐associated changes were not observed in the measured neuroregenerative biomarkers, the instance of significantly upregulated SYP expression and the possible pattern of upregulated TUBB3 expression warrant continued exploration of the Fe³⁺‐mSF architecture's effect on the neuroregenerative potential of hydrogel‐only scaffold systems.

Finally, in a proof‐of‐principle study, magnetically actuated and gelation‐preserved Fe³⁺‐mSF alignment was observed in an SH‐SY5Y cell‐seeded collagen hydrogel upon exposure to the static magnetic field generated by an magnetic resonance imaging (MRI) machine, with no obvious pulling of the fibers or surrounding cells from the hydrogel bulk (Figures 8 and 9). Given the applicability of MRI‐guided injection procedures in intralesional therapeutic delivery (i.e., placement of hydrogel scaffolds), alignment without gross displacement of Fe³⁺‐mSF or the seeded SH‐SY5Y cells in the instrument's magnetic field is promising for the translational relevance of this magneto‐responsive, in situ aligning guidance architecture in neuroprotective, neuroregenerative SCI scaffold design [105, 106, 107]. Furthermore, when fluorescently stained for visualization of cell morphology (phalloidin cytoskeletal stain, DAPI nuclear stain), confocal images of these MRI‐exposed samples illustrated increased cell spreading and cytoskeletal interaction in response to the “high iron” Fe³⁺‐mSF architecture, consistent with the possible trend of TUBB3 upregulation observed in response to the high iron Fe³⁺‐mSF/collagen scaffolds (Figures 7, 8, 9).

FIGURE 8.

SH‐SY5Y cell interaction with magnet‐exposed Fe³⁺‐mSF/collagen scaffolds. Confocal microscopy images of fluorescently stained SH‐SY5Y cells seeded and differentiated in magnet‐exposed collagen hydrogel composites containing either (A, B) 0 mg mL⁻¹ Fe³⁺‐mSF, (C, D) 1 mg mL⁻¹ “no iron added” Fe³⁺‐mSF, or (E, F) 1 mg mL⁻¹ “high iron” Fe³⁺‐mSF. Images were captured at 40X magnification (A–C), with areas of interest (white boxed regions in A–C) further magnified in post‐processing (B, D, E). Phalloidin staining (red) and DAPI (blue) staining were performed to afford visualization of SH‐SY5Y cell cytoskeleton and nuclei, respectively; due to the autofluorescence of silk fibroin, Fe³⁺‐mSF were captured in both channels, but are distinguishable from cellular structures as the long, sometimes arcing rectangular features observed in panels C, D and E, F. NOTE: Fe³⁺‐mSF orientation as it appears in the figure is not representative of the alignment achieved during magnet‐exposure; post‐magnet‐exposure Fe³⁺‐mSF orientation was disrupted by sample handling during the staining and mounting processes.

FIGURE 9.

SH‐SY5Y cell interaction with non‐magnet‐exposed Fe³⁺‐mSF/collagen scaffolds. Confocal microscopy images of fluorescently stained SH‐SY5Y cells seeded and differentiated in non‐magnet‐exposed collagen hydrogel composites containing either (A, B) 0 mg mL⁻¹ Fe³⁺‐mSF, (C, D) 1 mg mL⁻¹ “no iron added” Fe³⁺‐mSF, or (E, F) 1 mg mL⁻¹ “high iron” Fe³⁺‐mSF/collagen hydrogel composites. Images were captured at 40X magnification (A–C), with areas of interest (white boxed regions in A–C) further magnified in post‐processing (B, D, E). Phalloidin staining (red) and DAPI (blue) staining were performed to afford visualization of SH‐SY5Y cell cytoskeleton and nuclei, respectively; due to the autofluorescence of silk fibroin, Fe³⁺‐mSF were captured in both channels, but are distinguishable from cellular structures as the long, sometimes arcing rectangular features observed in panels C, D and E, F.

3. Conclusions and Future Work

In this work, previously developed Fe³⁺‐mSF were investigated as a potential nerve guidance architecture to increase the neuroregenerative potential of existing hydrogel‐only scaffold systems. Minimal, physiologically negligible iron release from Fe³⁺‐mSF was observed in PBS over the course of both short‐term (7‐d) and long‐term (3‐mo) studies. Characterization of Fe³⁺‐mSF/hydrogel composites demonstrated that the addition of the Fe³⁺‐mSF architecture to existing hydrogel‐only systems did not impede key mechanical functionalities for minimally invasive scaffold delivery (syringeability, post‐injection gelation, low swelling profile). Injection forces measured for Fe³⁺‐mSF/hydrogel composites were consistent with those of the hydrogel‐only control group, remaining within “easily injectable” human factor limits and suggesting that the material's capacity for minimally invasive delivery is preserved with the addition of the Fe³⁺‐mSF architecture. Rheological characterization of Fe³⁺‐mSF/HA hydrogel composites indicated that neither the fiber density nor the iron content of the added Fe³⁺‐mSF architecture significantly altered the favorable viscoelastic behaviors or critical gelation time of the material.

In terms of effect on neuroregenerative potential, Fe³⁺‐mSF did not elicit any significant cytotoxic response from SH‐SY5Y cells (neuronal analog) in vitro after 1, 3, 5, or 7 d of exposure (resazurin reduction assay), providing evidence of the proposed architecture's cytocompatibility both as an independent biomaterial and when incorporated into an existing hydrogel system. In terms of its effect on the neuroregenerative potential of existing hydrogel‐only scaffolds, the addition of the proposed Fe³⁺‐mSF to a collagen‐only hydrogel did not significantly affect overall neuroregenerative biomarker expression in seeded SH‐SY5Y cells. However, a possible trend of TUBB3 regulation was observed in SH‐SY5Y cells when seeded on scaffolds with the added Fe³⁺‐mSF architecture, even without treatment with media‐supplemented differentiation factors (i.e., RA). As TUBB3 is a key biomarker of axonal advancement, these non‐significant trends warrant continued investigation. Future studies should examine the neuroregenerative effects of Fe³⁺‐mSF in a primary neural or neuronal cell line to further probe the outcomes observed in the SH‐SY5Y cell neuronal analog, ideally a neural stem/progenitor cell line (NSPCs) [65, 80, 108, 109, 110, 111, 112]. While the use of the SH‐SY5Y cell neuronal analog facilitated first‐pass screening of Fe³⁺‐mSF‐induced expression changes in key neuroregenerative biomarkers, NSPCs would additionally provide insight into Fe³⁺‐mSF‐driven changes in preferential differentiation outcomes among endogenous or transplanted cell populations, i.e., whether NSPCs differentiate toward neuronal cells or glial cells (e.g., astrocytes, oligodendrocytes, microglia), especially if examined with single‐cell RNA sequencing techniques. As NSPC preferential differentiation can be significantly affected by both topographical alignment cues and oxidative stress levels, quantified biomarkers should be expanded to include those relevant to iron homeostasis and oxidative stress, as well (e.g., transferrin receptor 1, TFR1; ferroportin, FPN; mitochondrial ferritin, FTMT; ferritin heavy chain, FTH) [113, 114, 115, 116].

Finally, in a proof‐of‐principle study, magnetically actuated Fe³⁺‐mSF alignment was achieved within an SH‐SY5Y cell‐seeded collagen hydrogel on the patient bed of an MRI machine, without obvious disruption of fiber or cell distribution. This work is the first to investigate Fe³⁺‐mSF as an injectable, magnetically aligning nerve guidance architecture for increasing the neuroregenerative potential of existing in situ crosslinking hydrogel scaffold systems for SCI.

4. Experimental Section/Methods

4.1. Fe³⁺‐mSF Synthesis

Magneto‐responsive silk fibroin microfibers were synthesized via SF chelation of ferric iron ions (Fe³⁺‐mSF) from aqueous ferric chloride solution (FeCl₃) as reported previously [43]. The concentrations of FeCl₃ treatment solutions (and thereby the iron contents of the synthesized Fe³⁺‐mSF) were selected to parallel those investigated in prior work, as synthesis of Fe³⁺‐mSF under these conditions was shown to facilitate fiber alignment comparable to that of existing MNP‐based nerve guidance architectures without compromising material cytocompatibility [30, 43]. “Low iron” Fe³⁺‐mSF were synthesized with 0.015 mM FeCl₃ and “high iron” Fe³⁺‐mSF were synthesized with 150 mM FeCl₃, while intermediate variants of “moderately low iron,” “moderate iron,” and “moderately high iron” Fe³⁺‐mSF synthesized with 0.15, 1.5, and 15 mM FeCl₃, respectively, with a “no iron added” Fe³⁺‐mSF control synthesized with 0 mM FeCl₃ (ultrapure water) [43].

4.2. Iron Quantification via Inductively Coupled Plasma Mass Spectroscopy (ICP‐MS)

Iron release from Fe³⁺‐mSF was quantified via ICP‐MS. To simulate iron release at physiological pH, Fe³⁺‐mSF were synthesized via treatment of mSF with FeCl₃, washed with ultrapure water, and resuspended at 1 mg mL⁻¹ in phosphate‐buffered saline (PBS, pH 7.4), aliquoted into 1.5 mL microcentrifuge tubes (1 mL per tube), and incubated at 37°C. The release medium (supernatant) was sampled at regular intervals over a “short‐term” 7‐d period (24‐h sampling interval) and a “long‐term” 3‐mo period (7‐d sampling interval), with 90% volume collected and replenished per sampling interval.

The iron content of the Fe³⁺‐mSF release supernatant was quantified using a NexION 350X ICP Mass Spectrometer (PerkinElmer, USA) interfaced with Syngistix software. The system was optimized for operation with kinetic energy discrimination (KED) via the Syngistix SmartTune function using the manufacturer‐recommended performance verification standard (NexION Setup Solution, PerkinElmer, USA), with flow rates for the nebulizer gas (Argon) and KED gas (Helium) optimized at 1.08 L min⁻¹ and 5 mL min⁻¹, respectively. Before each data collection session, a standard performance check was run to confirm appropriate analyte intensity readings and minimal analyte charging with the optimized nebulizer flow rate.

Samples were loaded into the instrument via a Perkin Elmer S10 Autosampler (initialized within and interfaced with the Syngistix software), with a 60 s sample flush, 65 s read delay, and 30 s post‐read wash. Spectral analysis was performed in peak hopping mode, with triplicate readings of 25 sweeps acquired with a dwell time of 50 ms per amu and an integration time of 1250 ms. Prior to analysis of the release supernatant samples, a linear calibration curve was generated from standard solutions of known iron concentrations in an equivalent matrix composition to that of the release supernatant samples. Standard solutions were prepared via dilution of a 1000 µg mL⁻¹ elemental iron standard (Ultra Scientific Analytical Solutions, USA) in an aqueous phosphate‐buffered saline solution (5% v/v PBS). A NIST‐certified internal reference standard (standard reference material 1643f) was read immediately after the standards to verify the accuracy of the calibration curve and again every 18 samples to monitor standard drift (i.e., analyte and/or matrix accumulation in the instrument). All release supernatant samples were prepared in triplicate.

To minimize polyatomic interference, data collection was performed in kinetic energy discrimination (KED) mode. Given the expected low iron content of the samples relative to the instrument detection limits/innate variability in baseline matrix reads, all measured iron concentrations were normalized to an internal standard of known iron content. Furthermore, due to the propensity for charge accumulation on the instrument cones, the aforementioned standard was read before and after each sample set; the pre‐ and post‐sample set standard read was averaged, and each of the measurements acquired for the respective sample set was normalized to this pre/post‐read standard. All samples were baseline corrected to the first blank standard read acquired on the first day of data collection.

To complement the quantification of iron release from Fe³⁺‐mSF, iron loading in the synthesized Fe³⁺‐mSF product was likewise quantified via ICP‐MS analysis of samples collected from the FeCl₃ treatment solution and each post‐synthesis wash volume. Fe³⁺‐mSF iron loading, i.e., the mass of iron assumed bound to the washed Fe³⁺‐mSF product, was calculated as the difference between the iron content of the respective FeCl₃ treatment solution before and after the 24‐h mSF treatment period, less the cumulative iron content of the post‐synthesis wash volumes (Equation 1 and Table S1). The initial, pre‐treatment iron content of the FeCl₃ solutions (Equation 1, m_soln(0) ) was approximated as the as‐prepared molarity of the FeCl₃ treatment solution, while the final. Post‐treatment iron content of the FeCl₃ solutions (Equation 1, m_soln(f) ) and the iron content of each of the four post‐synthesis wash volumes (Equation 1, m_w1, m_w2, m_w3, m_w4 ) were quantified via ICP‐MS analysis (Table S1). Percent mass loading was subsequently calculated relative to the theoretical initial iron content of the FeCl₃ treatment solution (Equation 2 and Table S1).

$$\begin{array}{ccc}\hfill \mathrm{mass}\mathrm{loaded}\left(\mu g\right)& =& m_{soln\left(0\right)}-m_{soln\left(f\right)}\hfill \\ & & -\left(m_{W1}+m_{W2}+m_{W3}+m_{W4}\right)\hfill \end{array}$$ (1)

$$\mathrm{mass}\mathrm{loaded}\left(\%\right)=\frac{\mathrm{mass}\mathrm{loaded}\left(\mu \mathrm{g}\right)}{m_{soln\left(0\right)}}\ast 100\%$$ (2)

4.3. Preparation of Fe³⁺‐mSF/HA Hydrogel Composite Scaffolds

FFA‐HA was synthesized via previously published methods with some modifications [76]. Briefly, 0.40 g of 151 ‐ 300 kDa HA (1 mmol carboxylate; Lifecore Biomedical Inc., Chaska, MN, USA) was dissolved in aqueous MES buffer (100 mM, pH 5.5). Once dissolved, 0.28 g (1 mmol) of 4‐(4,6‐dimethoxy‐1,3,5‐triazin‐2‐yl)‐4‐methylmorpholinium chloride (1 mmol DMTMM, Millipore Sigma, Burlington, MA, USA) was dissolved in the HA solution under constant stirring for 10 min. Then, 47.2 µL furfurylamine (∼0.5 mmol FFA, Sigma–Aldrich, St. Louis, MO, USA) was added, and the reaction proceeded for 24 h with constant stirring. Following the amidation reaction, the synthesized FFA‐HA product was dialyzed in 12–14 kDa MWCO tubing (SpectraPor, Spectrum Chemical Mfg. Corp., New Brunswick, NJ, USA) against pH 3 water for 24 h then against non‐ pH‐adjusted deionized water (PicoPure water purification system, Hydro Service and Supplies, Durham, NC) for 48 h followed by lyophilization into a dried product. The degree of substitution (DoS) of the FFA‐HA was quantified via proton nuclear magnetic resonance (¹H‐NMR) spectroscopy. Briefly, calculation of DS was performed by comparing the area under the curves of the three peaks pertaining to the furan (6.59–6.23 and 7.35–7.58 ppm) normalized to the area under the curve of the unmodified HA region (1.44–2.44 ppm, 12 contributing protons) as previously described [76]. The integral ratio on the ¹H NMR spectrum suggested that HA was grafted with furan groups on an average of 25% of the disaccharide repeats. FFA‐HA hydrogels were subsequently prepared via Diels‐Alder click chemistry using a PEG‐(MI)₂ (3000 Da, RAPP Polymere GmbH, Tülbingen, Germany) crosslinking molecule per previously described methods [76]. The 4‐2 cycloaddition reaction between the furan (diene, FFA‐HA) and maleimide (dienophile, PEG‐(MI)₂) yields a click‐crosslinked hydrogel. For this work, all hydrogels were synthesized with a final HA concentration of 1.5% w/v, and the diene‐to‐dienophile ratio was standardized at a 1:2 molar ratio of furan to maleimide.

For HA‐based hydrogel composites, the added Fe³⁺‐mSF architecture was varied in terms of fiber density and iron modification (Table 4). The effect of Fe³⁺‐mSF density was examined at levels of 0.5, 1.0, and 2.0 mg mL⁻¹, with a 0 mg mL⁻¹ hydrogel‐only control condition; the effect of Fe³⁺‐mSF iron modification was examined at the “high iron” (150 mM) level, with a “no iron added” (0 mM) control condition.

TABLE 4.

Experimental variables examined in Fe³⁺‐mSF/hydrogel composite characterization.

Experimental group name	Fiber density a	Iron content b
0 mg mL−1 (Hydrogel‐Only Control)	0.0 mg mL−1	n/a
0.5 mg mL−1 “No Iron Added” Fe3+‐mSF	0.5 mg mL−1	0 mM FeCl3 Treatment
0.5 mg mL−1 “High Iron” Fe3+‐mSF	0.5 mg mL−1	150 mM FeCl3 Treatment
1.0 mg mL−1 “No Iron Added” Fe3+‐mSF	1.0 mg mL−1	0 mM FeCl3 Treatment
1.0 mg mL−1 “High Iron” Fe3+‐mSF	1.0 mg mL−1	150 mM FeCl3 Treatment
2.0 mg mL−1 “No Iron Added” Fe3+‐mSF	2.0 mg mL−1	0 mM FeCl3 Treatment
2.0 mg mL−1 “High Iron” Fe3+‐mSF	2.0 mg mL−1	150 mM FeCl3 Treatment

^a Fiber density refers to the mass of Fe³⁺‐mSF per unit volume of total hydrogel composite precursor during Fe³⁺‐mSF/hydrogel composite synthesis.

^b Iron content refers to the amount of iron with which mSF were functionalized during Fe³⁺‐mSF synthesis (directly correlated to the concentration of FeCl₃ solution with which mSF were treated) [43].

4.4. Syringeability Assessment via Instron Compressive Testing

The syringeability of Fe³⁺‐mSF/FFA‐HA hydrogel composites was characterized in terms of the compressive force required to eject the material from a pre‐loaded syringe. Briefly, 50 µL of Fe³⁺‐mSF/FFA‐HA pre‐gel solution was loaded into a 0.5 mL syringe (Becton Dickinson, Franklin Lakes, NJ, USA) fitted with a 27G needle (0.5 in length, 0.210 mm inner diameter) and allowed to crosslink overnight at 37°C in a humidified environment (to mitigate dehydration of loaded solution in needle tip prior to injection testing). Injection force was characterized via compressive testing performed on an Instron ElectroPuls instrument fitted with a 50 N load cell. All samples were subject to two consecutive compressive tests during which the syringe plunger was depressed at a rate of 0.5 mm min⁻¹ to a displacement of 2.5 mm per test (total displacement of 5 mm). The initial compressive test served to prime the syringe, while the secondary compressive test facilitated characterization of maximum and average injection forces required to eject the material from a primed syringe (final 0.5 mm of secondary force displacement curve) (Figure S5). To this end, injection of 50 µL of Fe³⁺‐mSF/FFA‐HA hydrogel composite was performed using a 0.5 mL syringe fitted with a 27 Ga needle (0.21 mm inner diameter, 0.5 in needle length). Plunger depression (0.5 mm min⁻¹, approximately 10 µL min⁻¹) was simulated via compressive loading performed on an Instron ElectroPuls (Instron, Norwood, MA, USA), and the applied force was plotted over a displacement range of 2.5 mm.

4.5. Oscillatory Rheology

The viscoelastic properties and critical gelation time of Fe³⁺‐mSF/FFA‐HA hydrogel composites were characterized via oscillatory rheology. All rheological testing was performed with an Anton Paar Modular Compact Rheometer (MCR 302 WESP) (Graz, Austria) fitted with a heated Pelletier plate (37°C) and a humidity chamber for physiological contextualization. In characterizing viscoelastic properties, a 5 mm parallel plate upper geometry (PP05/S, Anton Paar) was used. Subsequent oscillatory time sweeps were conducted using a 25 mm parallel plate upper geometry (PP25/S, Anton Paar).

Preliminary oscillatory strain sweeps (N = 3) were conducted to identify the linear viscoelastic region (LVER) of pre‐formed Fe³⁺‐mSF/FFA‐HA hydrogel composites, within which oscillatory frequency sweeps were subsequently performed. Strain sweeps of crosslinked Fe³⁺‐mSF/hydrogel composites were conducted with an initial shear strain of 0.01% that was increased stepwise on a base‐10 logarithmic scale to a final shear strain of 100%. For the subsequent frequency sweeps, a 5% oscillatory shear strain was applied at an initial frequency of 0.0159 Hz and increased stepwise to 100 Hz on a base‐10 logarithmic scale.

To mitigate the confounding effects of strain or frequency‐dependent viscoelastic behaviors, oscillatory time sweeps were performed within the previously identified LVER (5% strain, 0.159 Hz). The critical gelation time was identified as the time at which the storage modulus (G’) overtook the loss modulus (G”), indicating that the composite material began to exhibit elastic‐dominant behavior. While convergence of the storage and loss moduli was observed in all samples, not all time sweep plots exhibited a distinct crossover point, especially those with higher fiber densities. In these samples, the starting value of the storage moduli was generally higher than that of samples with lower fiber densities, reducing the difference between the storage and loss moduli of the composite system by increasing the elastic fraction of the initial system and thereby obscuring the crossover point. Consistent observation of moduli crossover in the hydrogel‐only control sample further validates attribution of lack of moduli crossover to the confounding contribution of the Fe³⁺‐mSF. Critical gelation time for these samples was instead estimated as the time at which the storage modulus was not only greater than the loss modulus (Equation 3) but also became increasingly greater than the loss modulus (Equation 4).

$$\mathrm{\Delta }G=\left(G_t^{\prime \prime }-G_t^{\prime }\right)$$ (3)

$$\mathrm{\Delta }\mathrm{\Delta }G=\frac{\left(G_{t+1}^{\prime \prime }-G_{t+1}^{\prime }\right)}{\left(G_t^{\prime \prime }-G_t^{\prime }\right)}$$ (4)

4.6. Swelling Study

The swelling properties of Fe³⁺‐mSF/FFA‐HA hydrogel composites were characterized at physiological pH and temperature. Briefly, 100 µL Fe³⁺‐mSF/FFA‐HA hydrogel composites were synthesized per previously described methods (Section 4.3) and crosslinked overnight (24 h) at 37°C in a humidified chamber to mitigate evaporation. Post‐gelation, the Fe³⁺‐mSF/FFA‐HA hydrogel composites were weighed (initial pre‐swollen mass, m₀ ), submerged in 1 mL PBS for 24 h, and then removed from the PBS swelling solution. The Fe³⁺‐mSF/FFA‐HA hydrogel composites were again weighed with their retained fluid volume (final post‐swollen mass, m_t ), and the swelling ratio was calculated as the fraction increase in mass (Equation 5).

$$SwellingRatio=\frac{m_t}{m_0}$$ (5)

4.7. SH‐SY5Y Cell Culture Maintenance and Cytocompatibility Testing

Neuroblastoma‐derived SH‐SY5Y cells (CRL‐2266, ATCC, Manassas, VA, USA) were cultured in Dulbecco's Modified Eagle/ Ham's F12 (DMEM/F12, 1:1; Cytiva, Marlborough, MA, USA) supplemented with 10% v/v FBS, 2.5 mM L‐glutamine, 100 U mL⁻¹ penicillin/ 100 µg mL⁻¹ streptomycin, and 0.25 µg mL⁻¹ Amphotericin‐B (growth media formulation). The cultures were maintained at 37°C in a humidified incubator supplemented with 5.0% CO₂ and passaged at approximately 90% confluency using 0.25% Trypsin/EDTA. To differentiate toward a neuronal phenotype, SH‐SY5Y cells were treated with reduced serum media (1% v/v FBS rather than 10% FBS) supplemented with 10 µM all‐trans RA, Cayman Chemical Company, Ann Arbor, MI, USA) (differentiation media formulation). Post‐passaging, undifferentiated SH‐SY5Y cells were allowed to recover for 24 h in “growth” media, then switched to differentiation media for an additional 7 days.

SH‐SY5Y cells were seeded in a 96‐well plate at an initial density of approximately 3.125 × 10⁴ cells cm⁻¹ (i.e., 1.0 × 10⁴ cells per well). After 24 h in normal growth medium, the medium was replaced with growth medium containing “low iron,” “moderately low iron,” “moderate iron,” “moderately high iron,” or “high iron” Fe³⁺‐mSF at a concentration of 0.5 mg mL⁻¹, with both a “no iron added” Fe³⁺‐mSF and 0 mg mL⁻¹ Fe³⁺‐mSF (growth media only) control treatment. After 1, 3, 5, and 7 d of Fe³⁺‐mSF exposure, 100 µL of media was carefully removed from each well (avoiding Fe³⁺‐mSF aspiration), replaced with 200 µL sterile filtered resazurin solution (0.15 mg mL⁻¹ in sterile phosphate buffered saline diluted to 25 µg mL⁻¹ in growth medium), and incubated for approximately 3 h at 37⁰C, 5% CO₂ to allow reduction of resazurin to resorufin by cellular metabolites. Post‐incubation, 100 µL of supernatant was collected from each well, transferred to a black 96‐well plate (GreinerOne, non‐sterile polystyrene), and the endpoint fluorescence signal (excitation 544 nm, emission 590 nm, cutoff 570 nm) analyzed via spectrophotometry (SpectraMax M2 Multilabel Microplate Reader, Molecular Devices, San Jose, CA, USA).

To prepare cell‐seeded Fe³⁺‐mSF/collagen hydrogels, components of neutralized collagen solution, Fe³⁺‐mSF suspension (DMEM/F12), and SH‐SY5Y cell suspension (DMEM/F12) were individually prepared at 3× final concentration and combined at a 1:1:1 volumetric ratio. All reagents and component solutions were prepared on ice under aseptic conditions. Sterile rat tail type I collagen (Corning, Bedford, MA, USA) at a concentration of 3 mg mL⁻¹ was neutralized via the addition of sterile‐filtered 1 N NaOH (23 µL per mL of collagen stock solution). Fe³⁺‐mSF suspensions were prepared with “high iron” and “no iron added” Fe³⁺‐mSF at 3 mg mL⁻¹ in DMEM/F12, with a 0 mg mL⁻¹ Fe³⁺‐mSF DMEM/F12‐only “suspension” substituted for the collagen‐only hydrogels. Undifferentiated SH‐SY5Y cells were trypsinized and resuspended at 7.5 × 10⁶ cells mL⁻¹ in DMEM/F12, yielding a final density of 2.5 × 10⁶ cells mL⁻¹ when combined with the collagen and Fe³⁺‐mSF solutions. After 1, 3, 5, and 7 d of Fe³⁺‐mSF exposure, media was carefully removed from each, replaced with 200 µL sterile‐filtered resazurin solution (0.15 mg mL⁻¹ in sterile PBS diluted to 25 µg mL⁻¹ in growth medium), and incubated for 3 h at 37⁰C, 5% CO₂. Post‐incubation, 100 µL of supernatant was collected from each well, transferred to a black 96‐well plate (GreinerOne, non‐sterile polystyrene), and the endpoint fluorescence signal (excitation 544 nm, emission 590 nm, cutoff 570 nm) was analyzed via spectrophotometry (SpectraMax M2 Multilabel Microplate Reader, Molecular Devices, San Jose, CA, USA).

4.8. Quantitative Real‐Time Reverse Transcription (qRT‐PCR) Polymerase Chain Reaction

Following Fe³⁺‐mSF exposure, changes in neuroregenerative biomarker expression by both undifferentiated and RA‐differentiated SH‐SY5Y cells were assessed via qRT‐PCR. SH‐SY5Y cells were seeded in Fe³⁺‐mSF/collagen hydrogel composite scaffolds prepared in a 48‐well tissue culture plate (200 µL Fe³⁺‐mSF/collagen hydrogel composite per well, with 5.0 × 10⁵ cells per 200 µL scaffold). Scaffold composition was varied with respect to Fe³⁺‐mSF architecture addition (0 or 1 mg mL⁻¹), Fe³⁺‐mSF iron content (“no iron added” Fe³⁺‐mSF or “high iron” Fe³⁺‐mSF), and Fe³⁺‐mSF orientation (random or aligned). While the feasibility of magnetically actuated Fe³⁺‐mSF alignment was successfully demonstrated in prior work, Fe³⁺‐mSF/collagen scaffolds analyzed for neuroregenerative biomarker expression were shear aligned to reduce the confounding effects of magnetic field exposure on observed gene expression. Immediately after pipetting the cell‐laden Fe³⁺‐mSF/collagen pre‐gel solution into 24‐well plates for gelation and subsequent culture, a 25 G needle was subsequently dragged through the hydrogel precursor solution across the diameter of the well, pulling the Fe³⁺‐mSF, collagen, and SH‐SY5Y cells into alignment.

After an initial 24 h recovery period (growth media), cultures were either maintained in growth media or switched to differentiation media for an additional 7 days. Post‐expansion or post‐differentiation (day 8 of culture for both media conditions), seeded cells were lysed with and collected in Trizol (Invitrogen Corporation, Thermo Fisher Scientific, Waltham, MA, USA) for RNA extraction. Extracted and isolated RNA content was quantified (NanoDrop 2000 spectrophotometer, Thermo Scientific, Waltham, MA, USA), and cDNA was subsequently synthesized via reverse transcription (High‐Capacity cDNA Reverse Transcription Kit, Applied Biosystems, Thermo Fisher Scientific, Waltham, MA). Quantitative reverse transcription polymerase chain reaction (qRT‐PCR) primers (Table S2) were purchased from Integrated DNA Technologies, Inc. (Coralville, IA, USA), and qRT‐PCR was performed with Applied Biosystems PowerSYBR Green PCR Master Mix per manufacturer recommendations (Thermo Fisher Scientific, Waltham, MA, USA) on the QuantStudio 6 Pro Real‐Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA). For each gene of interest, change in cycle threshold values (ΔC_t) were calculated relative to a housekeeping gene (GAPDH), and change in change in cycle threshold values (ΔΔC_t) was calculated relative to the average of a 0 mg mL⁻¹ Fe³⁺‐mSF/collagen (hydrogel‐only) experimental control condition. Normalized fold change in expression of examined genes (2^−ΔΔCt) was subsequently calculated for each experimental condition.

4.9. Phalloidin/DAPI Fluorescence Staining and Confocal Imaging

SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen hydrogel composites were visualized via fluorescence staining and confocal microscopy imaging. To facilitate simultaneous magnetic alignment of multiple samples, SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen hydrogel composites were prepared within the static magnetic field generated by a MRI machine (3.0 T SIGNA Premier MRI, GE Healthcare, Chicago, IL, USA). SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen hydrogel composites were prepared in a 48‐well tissue culture plate under sterile conditions as previously described, with 200 µL of SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen hydrogel composite per well and 1.5 × 10⁶ cells per mL scaffold. Immediately after plating (before gelation), the plate was either transferred to the patient bed of the MRI machine, allowing Fe³⁺‐mSF alignment in the instrument's static magnetic field (Figure S7), or transferred to a non‐magnetic field‐exposed benchtop, maintaining as‐plated, “random” Fe³⁺‐mSF orientation as an experimental control. All SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen hydrogel composites were maintained in their respective exposure conditions for 1 h, allowing preliminary thermal crosslinking at room temperature to preserve Fe³⁺‐mSF orientation. The scaffolds were then transferred to a 37°C, 5.0% CO₂ humidified incubator for an additional 30 min to allow further crosslinking, after which 100 µL of growth media was added to each scaffold‐containing well. SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen scaffolds were maintained in growth media for 24 h then RA differentiation media for an additional 7 d, with 50% of the media exchanged daily.

Following 7‐d RA differentiation, SH‐SY5Y cell‐seeded Fe³⁺‐mSF/collagen scaffolds were washed with Dulbecco's phosphate‐buffered saline (DPBS; Corning, Inc., Corning, NY, USA), and fixed for 45 min with 4% paraformaldehyde solution (PFA, Electron Microscopy Sciences, Hatfield, PA, USA). Post‐fixation, scaffolds were washed thrice with DPBS, then the cells were permeabilized with 0.1% Triton X‐100. To avoid damaging scaffolds or disrupting cellular and architectural orientation, fluid exchange steps were performed by gently pipetting fixation agents, permeabilization agents, and wash volumes along the well plate walls. Fixed and permeabilized scaffolds were subsequently incubated with Phalloidin (1:40 dilution in DPBS; BioLegend, Inc., San Diego, CA, USA) and DAPI (1:1000 dilution in DPBS; Thermo Fisher Scientific, Waltham, MA, USA), washed once with DPBS, then transferred to a glass‐bottom MatTek dish for confocal fluorescence imaging (Leica Stellaris 8, Wetzlar, Germany). Five representative z‐stacks of each scaffold composition were captured at 20× magnification, and two or three representative z‐stacks of each scaffold composition were captured at 40× magnification.

4.10. Statistical Analysis

All statistical analyses were performed in GraphPad Prism 7. All data are presented as mean ± standard error of the mean (SEM) unless otherwise noted (i.e., mean ± standard deviation, SD). Cumulative and intraday iron release are presented as mean ± SEM for the short‐term release study data (N = 3, n = 3) and as mean ± SD for the long‐term release study data (N = 1, n = 3). Maximum and average injection forces are both presented as mean ± SD of nine replicates from three independent experiments (n = 9). For oscillatory rheology data sets, storage moduli and loss moduli values are presented as mean ± SD in amplitude sweep plots (N = 3, n = 1), mean ± SEM (N = 3, n = 3) in frequency sweep plots, and mean ± SD in time sweep plots (N = 1, n = 3). Swelling ratios are presented as mean ± SD (N = 1, n = 3). SH‐SY5Y cell cytocompatibility data are presented as mean ± SEM of three independent experiments (N = 3). qRT‐PCR data are presented as 2^−ΔΔCt with lower and upper error bounds calculated using the ΔΔC_t linear (SEM) and reported as 2^{(‐ΔΔCt – SEM)} and 2^{(‐ΔΔCt + SEM)}, respectively. Statistical analyses of qRT‐PCR data sets were performed on the ΔΔC_t values. Statistical differences were determined by an analysis of variance (ANOVA) followed by either Dunnett's multiple comparisons test (when comparing experimental groups to a control group) or Tukey's honestly significant differences (HSD) test (when comparing among experimental groups). In statistical comparison of data, a p‐value less than 0.05 was considered significant.

Author Contributions

Melissa A. Wojnowski: conceptualization (equal); methodology (lead); data curation (lead); investigation (lead); validation (lead); formal analysis (lead); funding acquisition (equal); visualization (lead); writing – original draft preparation (lead); writing – review and editing (equal)., Zephyr A. Paxton: methodology (supported); data curation (supported); formal analysis (supported); visualization (supported); writing – review and editing (equal). Donald Pellegrino: methodology (supported); data curation (supported); formal analysis (supported); visualization (supported); writing – review and editing (equal). Jeannine M. Coburn: conceptualization (equal); methodology (supported); data curation (supported); investigation (equal); validation (supported); formal analysis (supported); funding acquisition (equal); visualization (supported); writing – original draft preparation (supported); writing – review and editing (equal).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adhm71430‐sup‐0001‐SuppMat.docx.

Data Availability Statement

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