Multifunctionalized electrospun silk fibers promote axon regeneration in central nervous system

Abstract

The repair of central nerves remains a major challenge in regenerative neurobiology. Regenerative guides possessing critical features such as cell adhesion, physical guiding and topical stimulation are needed. To generate such a guide, silk protein materials are prepared using electrospinning. The silk is selected for this study due to its biocompatibility and ability to be electrospun for the formation of aligned biofunctional nanofibers. The addition of Brain Derived Neurotrophic Factor (BDNF), Ciliary Neurotrophic Factor (CNTF) or both to the electrospun fibers enable enhanced function without impact to the structure or the surface morphology. Only a small fraction of the loaded growth factors is released over time allowing the fibers to continue to provide these factors to the cells for extended periods of time. The entrapped factors remain active and available to the cells as rat retinal ganglion cells (RGCs) exhibit longer axonal growth when in contact with the biofunctionalized fibers. Compare to non-functionalized fibers, the growth of neurites increased 2 fold on fibers containing BDNF, 2.5 fold with fibers containing CNTF and by almost 3-fold on fibers containing both factors. The results demonstrate the potential of aligned and functionalized electrospun silk fibers to promote nerve growth in the central nervous system, underlying the great potential of complex biomaterials in neuroregenerative strategies following axotomy and nerve crush traumas.

1. Introduction

Regenerative medicine has emerged at the interface between biomaterial engineering and medical science. In the neurobiological field, many biomaterials have been developed to improve the regeneration of peripheral nerve damage, ^[1] yet the repair of central nerves, such as the optic nerve that have been cut or crushed, remains a major challenge. ^[2] One of the first attempts to repair sectioned central nerves was performed a hundred years ago by suturing a sciatic nerve graft at a site of an optic nerve lesion to rescue the axotomized and dying retinal ganglion cells (RGC). ^[3] However, only a few regenerative fibers emerged from the RGC and grew along the sciatic implant. This low rate of regeneration is mainly due to the inability of adult RGC, to switch to a regenerative state to regrow a functional optic nerve. ^{[4, 5]} Following injury, RGC axons in the optic nerve undergo Wallerian degeneration ^[6] during which surrounding glial cells are activated and an inflammatory response is triggered involving the infiltration of macrophages. ^[7] The central nerve then continues to degenerate in a process involving gliosis, formation of glial scar and reorganization of the matrix molecules blocking axon regeneration. ^{[8, 9]} Glia reactivity is not completely detrimental since some neurotrophic molecules produced by glial cells provide a favorable environment for axonal growth and regeneration. These include brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), ciliary neurotrophic factor (CNTF), leukemia inhibitory factor (LIF), insulin-like growth factor (IGF), nerve growth factor (NGF) and basic fibroblast growth factor (bFGF). ^{[10, 11]} Downstream signaling effects of these neurotrophic factors are mediated via their binding affinity to specific receptors such as the trk (tropomyosin-receptor-kinase) family of tyrosine kinase receptors, leading to activation of signal cascades, promoting survival and functional regulation in neurons. ^[2] The main challenge in regenerative medicine is to enhance these signals for neuronal survival that are masked in the background of noxious effects; biomaterials are of main interest to achieve this goal.

While progress has been made to generate artificial nerve regrowth systems using different approaches (hydrogels, cast materials, electrospun fibers) with different biomaterials (collagens, polyglycolic/polylactic acid, silicone, alginates, and others), no nerve guide has yet emerged that can: i) physically align the growth of regenerating nerves; ii) switch neurons to a regenerative state and promote nerve growth and axonal adhesion; iii) eliminate scar tissue formation; iv) avoid rejection by the body; v) not swell to impinge on the nerve; vi) degrade to form byproducts that are nontoxic or noninflammatory; and vii) easily be handled by surgeons.

The ability to control the morphology, chemistry structure, biocompatibility, and the remarkable mechanical properties of silk-based biomaterials make this protein a unique platform to control functional tissue outcomes. Based on recent studies, materials made from silk fibroin can address the needs of specific cellular repairs for tissue engineering of bone, cartilage, adipose tissue, blood vessels and ligaments. ^[12–20] Another application is the generation of artificial nerve guides since silk fibroin has shown to be neuro-biocompatible using hippocampal neurons ^[21] and therefore is appropriate to develop materials to treat central nervous system injuries or diseases. Silk nerve conduits have been tested for peripheral nerve regeneration with outcomes approaching autografts. ^{[22, 23]} However, due to their large diameter, such tubular materials are not fully adapted to individual growth cone guidance in the central nervous system.

Methods to electrospin silk through a completely aqueous process generating new biomaterials have been successfully developed recently. ^{[23, 24]} Electrospinning is a simple and versatile technique for fabricating nanofibrous non-woven mats. ^[25–27] Further modification of this method allows the generation of parallel arrays of fibers. ^[28] The hydrophobic nature of silk fibroin provides stability in shape and size of the fibers under physiological conditions. Moreover, the electrospinning process allows the addition of bioactive materials during the spinning process, with slow release of these entrapped compounds, keeping intact their biological functions. ^{[23, 29–31]}

The goal of the present study was to incorporate growth factors in aligned electrospun silk fibers to generate bio-active fibers and to test their ability to promote neuronal survival and neurite outgrowth in vitro. To mimic traumatic optic neuropathy, ^[32] cultures of pure postnatal RGC using the immunopanning method were performed. ^[33–38] As it occurs during optic nerve trauma, purified RGC did degenerate in a minimal medium in vitro. Only the addition of specific growth factors could limit the extent of the neurodegenerative process. ^[33–38] BDNF and CNTF were selected as neurotrophic factors as they have been extensively studied in this specific neuronal population: BDNF promotes RGC outgrowth and survival in culture, ^[37,38] and astrocyte-derived CNTF is able to switch mature RGC to a regenerative state following nerve trans-section and lens injury in vivo. ^{[4, 5]}

The present study shows that multifunctionalized and aligned silk nanofibres serve as support matrices for axons from primary RGCs culture, increase RGCs survival and stimulate their neurites outgrowth. These results suggest new types of nerve guides based on biologically active silk protein-based materials.

2. Results

2.1. Description of the electrospinning setup and morphology of aligned nanofibrous fibers

Electrospinning is a process that allows the continuous production of fibers of diameters ranging from few nanometers to few microns. A liquid jet is pulled by electrostatic forces from the tip of a droplet attached to a high voltage to a grounded collector (Fig. 1A). A bending instability triggers the jet to whip more and more as the jet travel from the tip to the collector resulting in a drastic reduction of the jet diameter to the submicron range. When the collector consists of a plate, random fibers form a non-woven mat (Fig. 1B). To guide cells in a preferential direction, an aligned functional network of nanofibers was generated using the electrospinning process combined to a rotating wheel as a collector. ^[28] This combination forced the fibers to wind around the wheel in a parallel array.

Figure 1. Control of fiber orientation depending on the speed of the collector plate.

A: Electrospinner set up. B–E: FESEM pictures of static fibers (B), fibers electrospun at 6 m s⁻¹ (C), fibers electrospun at 10 m s⁻¹ (D), fibers electrospun at 14 m s⁻¹ (E). Rate of aligned fiber increase from static collector (B) to 10 m s⁻1 (D) At higher speed (E) silk fibers start to break. The alignment of the fibers was optimal at 10 m⁻¹. Scale bars: 5 microns.

The morphological structure of electrospun fibers was observed by FESEM. As shown in Fig. 1B–E, the quality of the alignment of the fibers depended on the speed of the wheel. Fibers showed no alignment in static condition (Fig. 1B), a weak alignment at 6 m.s⁻¹ (Fig. 1C), reached optimal alignment at 10 m.s⁻¹ (Fig. 1D) and started to break down around 14 m.s⁻¹ (Fig. 1E). At optimal speed the angular deviation from the alignment was 4 degrees. The average fiber diameter of non aligned fibers was 350 nm +/− 40 nm while the average fiber diameter of optimally aligned fiber was 330 nm +/− 130 nm. The standard deviation of the aligned fibers was significantly different from the nonaligned fibers indicating a strong pulling from the rotating wheel that stretched the fibers. Unless specified, further experiments were conducted using the optimally aligned fibers obtained at 10 m.s⁻¹.

2.2. Surface properties

As the surface properties of the material can, by itself, modify the adhesion and the growth of the cells, it was therefore essential to determine by AFM the roughness of silk fibroin fibers before and after biofunctionalization (Fig. 2). Ra (the arithmetic average of the absolute values of the surface height deviations measured from the mean plane) was calculated as a measure of the roughness (Fig. 2G and 2H). The Ra of the silk/PEO fibers was about 1 nm, 3 times the value of the glass. Ra of functionalized fibers showed no significant difference from the non-functionalized fibers. This implies that the roughness of the material was not affected by the functionalization and that effects of functionalization would then come from the growth factor properties and no from any change of the physical properties of the support.

Figure 2. Fiber roughness.

A–C: FESEM pictures of Silk fiber (A), Silk/BDNF fiber (B), Silk/CNTF fiber (C); scale bars are 200 nm. D–F: AFM pictures of Silk fiber (D), Silk/BDNF fiber (E), Silk/CNTF fiber (F). G: Roughness of the surface of the Silk fiber, Silk/BDNF fiber, Silk/CNTF fiber compare to glass, increment of the Y axis is 1 nm. H: Ra related to the roughness of the surface of the Silk fiber, Silk/BDNF fiber, Silk/CNTF fiber and the glass surrounding the silk fiber. The fibers were significantly rougher than the glass. Ra of functionalized fibers showed no significant difference from the one of the non-functionalized fibers, demonstrating that the roughness of the material was not affected by the functionalization.

2.3. Functionalization

The release of growth factors from the mat was determined by ELISA (Fig. 3A). The relative amount of growth factors released was extremely low, less than 0.0005% of them were released over a week. The very small release happened during the first hour. Subsequent time points showed identical or lower amount of growth factors suggesting that the release stopped and growth factors degradation occurred thereafter. By comparing the degradation rate of the few released growth factors, it can be noticed that CNTF degradation was slower than for BDNF.

Figure 3. Release and stability of growth factors.

A: Percentage of growth factor released by the fibers. The release was very low and happened in the first hour. After the first hour, the curve was decreasing showing that there was no more release and the growth factors started to be degraded. B: percentage of growth factor left in the fibers after 1 week in PBS with 0.2% BSA compared to percentage of the growth factors left in fibers after 1 week left dry. C: Degradation of the growth factors with time in a PBS with 0.2% BSA solution. The percentage of growth factor was diminishing overtime showing the time period during which they were stable in solution. BDNF was more unstable than CNTF. Less than 50% of the BDNF could be detected after 4h while 60% of CNTF was still detectable after one week.

The lack of release could be due to an initial lack of embedding of the growth factors in the fibers during the electrospinning process. To answer this question, the presence of growth factors in the fibers was measured (Fig. 3B). Mats were dissolved in lithium bromide after 7 days of soaking in PBS with 0.2% BSA at 37°C. A control left dry for 7 days was dissolved at the same time for comparison. As shown in Fig 3B, relatively high amounts of factors were recovered from the fibers. 90% of the CNTF and 57% of the BDNF were recovered showing that the growth factors were indeed embedded in the fibers. There were no statistical differences found between the soaked and dry sample which indicates that the growth factors were trapped in the fibers and not released. The percentage of growth factors recovered from fibers loaded with BDNF was significantly lower than the percentage of CNTF. As hinted by the release experiment (Fig. 3A) this may reflect a higher stability of CNTF compared to BDNF. The process of fiber dissolution contained a 6 hours dialysis step that was carried out to wash the LiBr and could lead to a high percentage of BDNF degradation. To verify this, an experiment where both growth factors were soaked in PBS with 0.2% BSA was carried out as shown in Fig 3C. Interestingly, when growth factors were not trapped in fibers but diluted in PBS with 0.2% BSA, there was a significant difference in their respective degradation rate. BDNF almost completely degraded over the course of a day while CNTF lost only 60% of its original concentration over a week. The difference in percentage of growth factor recovery found in Fig. 3B is therefore due to a difference in stability between CNTF and BDNF. In conclusion, these results show that the BDNF and CNTF are stably embedded in the fibers. Moreover, growth factors trapped in silk fibers are protected from degradation over long period of time, suggesting that biofunctionalized silk fibers may represent a new tool to provide a neuronal growth-promoting environment at the site of a nerve lesion.

2.4. RGC cultures on non functionalized silk fibers

Pure post natal RGC cultures were used to study the ability of fully differentiated neurons from the CNS to regenerate after an axotomy. First, RGC neurons were cultured on non-oriented silk fibers in complete medium supplemented with BDNF and CNTF (MM+BC). This medium has been shown to be optimal for survival and neurite extension in this neuronal population when cultured on poly-d-lysine substrate. ^[36,37] Here we tested the effect of silk fibers on RGCs survival and neurite outgrowth in optimal medium MM+BC. Phalloidin staining of the actin filaments revealed that RGCs were able to extend long neurites (Fig 4A). While on poly-d-lysine coated surfaces, it was previously shown that neurite extention occurred randomly with extensive branching, ^[36] we did not observed this phenomena for RGC when cultured on silk fibers. On silk electrospun fibers RGC usually extended less neurites but their extensions exhibited some network organization. Superimposing pictures of the autofluorescent silk fibers to those of the RGCs demonstrated that the growth cone progressed along the silk fibers and changed direction at fiber intersections (arrows in Fig. 4A and 4B). Growth cones (arrowheads in Fig. 4A and 4B), permanently probed the environment, and selectively chose a silk fiber to further elongate a neurite at its contact. When observing ultrastructural interactions using SEM (Fig. 4C), RGCs attached to silk fibers (left insert in 4C) and extended long neurites (arrows in 4C) and growth cones (right insert in 4C) in contact with the surrounding silk fibers. Magnifications in Fig. 4D and 4E revealed that the cell soma was tightly attached to the fibers (Fig. 4D) and that growth cones established multiple contacts with the silk fibers (Fig. 4E), both via lamellipodia and filopodia (arrows in Fig. 4D and 4E). These findings demonstrated that silk is a suitable material to guide neurite direction of extension.

Figure 4. RGC culture on silk fibers.

A: Phalloidin labelling of the actin filaments in RGC shows a neurite exhibiting sharp change in direction of growth (arrow). B: Merged image composed of the TRITC-phalloidin (red), DAPI nuclear staining (blue) and autofluorescence of silk network (green) demonstrates that growth cone (arrowheads in A and B) elongates in contact with silk fibers and that sharp changes of direction correspond to silk crossroads (arrows in A and B). SEM of RGC (C–E) reveals that neurons grow in close contact with silk. Cell soma is anchored to the fiber network (left square in C and corresponding magnification in D) via lamelipodia and filopodia (arrow in D). Neurite grows along the silk guide (arrows in C) and growth cone (right square in C and corresponding magnification in E) adheres to this material via lamelipodia and filopodia (arrow in E). RGC survival and neurite extension on orientated silk fibers using different culture condition (F–I) In minimal medium (F) few RGC survive and extend neurites. Addition of BDNF (G), CNTF (H) and combination of both factors (I) in the culture medium increase neurite extension and improve the complexity of the neuronal network. Effect of growth factor addition in the culture medium on RGC survival after 3 days in vitro (J). Percentage of surviving RGC in culture is compare between minimal medium, addition of BDNF (MM+ B), addition of CNTF (MM+ C) and combined addition of BDNF and CNTF (MM+ BC). Survival in vitro is increased by a factor 3 when both factors are present. The fact that RGC respond to growtgh factor stimulation suggest that silk exhibit no toxicity towards this neuronal type. Scale bars 20 μm in A–B, 5 μm in C, 2 μm in D–E, 50 μm in F–I, Error bars reflect SEM. **P>0.01 and P<0.001, significance value for neuronal survival (J) in comparison to the minimal medium and between different growth factor compositions of medium (bracket).

The effect of neurotrophic factor addition to the media on the morphology and survival of RGCs when cultured on aligned silk fibers was observed. In minimal medium (MM) few surviving RGCs were observed, and even fewer exhibited long neurites (Fig. 4F). The addition of BDNF (Fig. 4G) or CNTF (Fig. 4H) in the medium increased neurite sprouting that mainly followed the underlying orientation of the silk fibers. The addition of both factors (Fig. 4I) resulted in a combined effect on neurite network development and complexity. In a similar manner, survival of the RGCs, observed by live/dead assay (Fig. 4J), was significantly improved by the addition of BDNF (MM+B) or CNTF (MM+C), reaching 30.9% (+/− 1.5) and 37.6% (+/− 1.0), respectively, of survival after 72h in vitro. This resulted in a dramatic increase in neuron survival when compared to the control (MM), where only 10.7% (+/− 0.9) of the RGCs survived. A significant combined effect of both factors was observed when added in the medium (MM+ BC) leading to a maximal survival rate of 60.8% (+/− 1.6) after 3 days in vitro. These results are in agreement with previously published observations where low RGC survival was found in the absence of growth factors and up to 70% survival in complete medium. ^[37] This confirmed the importance of BDNF and CNTF for RGC survival and demonstrated that electrospun silk had no negative effect on RGCs survival and neurite outgrowth. Indeed the effect of growth factors on survival rate was conserved in the presence of the silk fibers suggesting that they are a suitable biomaterial support matrix for RGCs, exhibiting no toxicity towards this primary neuronal type.

2.5. RGC cultures on biofunctionalized silk fibers

In case of an optic nerve trauma growth factors would need to be provided topically and over a long period of time to enhance nerve regeneration. Therefore, we studied here the possibility to promote RGC survival by using silk fibers loaded with growth factors rather than just added in the culture medium. Different combinations of functionalized silk fibers were studied for their ability to support RGCs in a medium devoid of growth factors (Fig. 5). Therefore, silk fibers are the exclusive source of growth factors. Two morphological parameters were assessed to identify biological effects of the functionalized silk guides: the average length of the longest neurite and the percentage of gap43 (a marker of neuronal growth cones) expressing neurons found in culture. After 3 days of culture, the effects of silk fibers loaded with BDNF (B Silk), CNTF (C Silk) and a combination of these factors (BC Silk) were compared to non-functionalized control fibers (Null Silk). On the Null Silk, 22.9% (+/− 1.8) of the RGCs presented gap43 positive neurites while on the B Silk the percentage was 57.0% (+/− 2.7) and on the C and BC silk this reached 65.9% (+/− 1.1) and 66.4% (+/−2.4), respectively (Fig. 5A). On the Null silk, the few RGCs that survived were able to extend neurites with an average maximal length of 46.9 μm (+/− 2.4). This result suggested that silk provides suitable growth support even in the absence of additional factors. Using functionalized silk fibers led to a significant increase in neurite length (Fig. 5B). On B Silk, neurite extension show a 1.8-fold increase with a length of 82.3 μm (+/− 2,8). C silks show a 2.3-fold increase in neurite length (105.2 μm +/− 1.3). The length reached 117.7 μm (+/− 2.9) on BC Silk (Fig. 5B). These measures showed that the multifunctionalization of the fibers significantly improved the rate of neurite extension. Aspect of RGC cultures after immunocytochemical stainings (Fig. 5C–F) supported these results. In the absence of functionalization few neurons survived, exhibiting short neurites (Fig. 5C). However, these rare neurites extended in contact with silk fibers as seen in the merged image (Fig. 5D). In the case of functionalized supports (Fig. 5E, F), RGCs survived massively and extended longer neurites that followed the silk guides over long distances as seen in the merged image (Fig. 5F). These results showed that RGCs were sensitive to the BDNF and CNTF entrapped in the fibers. Even if these factors were not released into the medium, they were available for the developing growth cones and conserved their neurotrophic properties.

Figure 5. RGC culture on biofunctionalized silk fibers.

A: Percentage of neurons exhibiting a gap43 positive neurite on different combinations of biofunctionalized silk. Biofunctionalization with BDNF (B silk), CNTF (C silk) and both factors (BC silk) increases the percentage of neurons expressing this axonal protein in their neurite by a factor 2,5 to 3 respectively compared to non biofunctionalized fibers (null silk). B: Neurite extension after 3 days in vitro. We measure the longest neurite of RGC cultured on simple silk (null silk) or biofunctionalized silk with BDNF (B silk), CNTF (C Silk) and multifunctionalized (BC Silk). The neurite length (expressed in μm on Y axis), is significantly increased on biofunctionalized silk. Notably, silk alone also allows some level of neurite extension, thus suggesting a mechanical effect of silk, beneficial to the growth cone in addition to the neurotrophic effect of the factors. Comparison of RGC cultures on normal electrospun silk (C, D) containing no growth factor versus biofunctionalized silk support with CNTF (E, F). Beta III tubulin staining revealed few surviving RGCs, exhibiting short neurites that follow the non-functionalized silk support (seen on the bright field of the merged image in D), Whereas growth factor functionalization (E, F) enhances neurite extension after 3 days in vitro along the functionalized silk (as seen on the bright field channel of the merged image in F). Scale bar 50 μm. Error bars reflect SEM. *P<0.05, **P<0.01 and P<0.001, significance value in comparison to the null silk and between different biofunctionalized silks (bracket) for axonal marker expression (A) and neurite elongation (B).

3. Discussion

Many strategies have been tested both in peripheral and central nervous systems to help sectioned or crushed nerves to regrow. These strategies encompass physical approaches, architectural designs and active molecule release. Structures such as hydrogels, ^[39] guidance tubes ^[40] or electrospun fibers have been evaluated and some of these technologies have been coupled with the release of active molecules. ^[41] Electrospun fibers are of interest in these types of studies as they can be tailored to possess critical properties to form nerve guides, (physical guide, biocompatibility, adherence properties). The electrospun fibers can be constructed from various polymers, such as collagen, ^[42,43] polycaprolactone ^[44,45] or poly(lactic-co-glycolic acid), ^[46,47] organized in parallel arrays ^{[43,44, 46]} and they can be loaded with active molecules. ^[43]

Such multi-level approach was pursued in the present work, where the properties of electrospun silk fibers were evaluated. Alignment of silk electrospun fibers offered support and guidance for regrowing neurites, and functionalization of the fibers with two different growth factors, CNTF and BDNF, providing the biological stimuli needed for the neuronal regeneration. Aligned fibers were obtained by coupling the electrospinning method with two techniques of alignments a rotating wheel and grounded gaps between nongrounded glass pieces. The idea of using a rotating disk for fiber alignment has been described recently using polyethylene glycol (PEG) ^[28] and has been successfully used in different variations with poly(l-lactid-co-ε-caprolactone) (PLLA-CL), ^[48] collagen, ^[49] poly-ε-caprolactone (PCL), ^[50] poly(l-lactic) acid (PLLA). The idea of using a nongrounded gap to align fibers was describe later ^[51] and works on numerous polymers and ceramics. It can be fastidious to achieve a high number of samples with the gap technique and the alignment can be lower than with a rotating wheel. One the other side, the technique of the rotating wheel is very efficient at aligning fibers but is designed to electrospin on a thin grounded metal and not on wide nongrounded pieces of glass. ^[52] In the present work, to achieve a guide for central nervous system nerves, highly aligned fibers directly deposited on glass coverslips were generated by combining both techniques. A rotating wheel covered alternatively with nongrounded and grounded gaps was used. An alignment (4 degrees of angular deviation at optimal speed) similar to what has been described in the literature with silk or other materials was observed. ^[23]

Many studies using electrospun biomaterials are performed on embryonic neurons ^[43] or stem cells lines, ^[39–44] and mostly in the peripheral nervous system. ^[23] The present data were obtained with functionalized electrospun silk fibers in a regeneration model using fully differentiated post natal neurons from the central nervous system that underwent an axotomy. This in vitro model matches therefore more closely the reality of traumatic nerve lesion in the central nervous system. First, to establish cytocompatibility of the silk nanofibers, pure primary RGCs were seeded on random silk mats. In suitable growth medium, RGCs were able to survive and grow neurites in contact with silk. Level of survival was found to be similar to what was described previously when RGC are seeded on the canonical poly-lysine and poly-lysine/laminin coating. ^[33,35–38] This demonstrated that silk nanofibers were a suitable support matrix for neuronal development, and that silk did not exhibit any toxicity towards these primary neurons in vitro as growth factors retains their effect on RGCs. The differences in the maximal survival rate: 60% on silk fibers vs 70% on poly-lysine and polylysine/laminin coatings ^[33,35–38] was likely due to the fact that silk fibers do not cover the whole glass surface homogeneously; therefore some RGCs have to contact the glass.

Surprisingly, silk fibers alone promoted slight neurite extension even in a minimal medium. This demonstrated that silk alone provides a suitable substrate to support nerves probably due to its mechanical properties acting as guidance cue as it has been shown in other systems ^[30,53] to promote neurite elongation ^[54,55] even in absence of growth factor. ^[56] As cells can sense change of roughness in the nanoscale level, growth cones may be responding to changes in surface roughness in our system. We found that the silk fiber roughness was significantly higher than the glass, although well below the described cytophobic threshold of 55nm. ^[57] However, growth factors addition enhances dramatically the survival and axon growth of these neurons demonstrating their essential contribution. As the inclusion of growth factors in the fibers did not affect the surface roughness, the differences in cell response to the growth factors was attributed to the functionalization and not to physical surface modifications of the fibers.

The functionalization of the fibers reached by immobilization of the growth factors inside the structure of the fibers resulted in weak release of the growth factors BDNF and CNTF from the fibers over time. A burst release of at least 20–30% of the factors was expected, as often occurs from electrospun fibers. ^{[23,30,58,59]} One study ^[23] reported weak release of growth factors (namely Glia Derived Neurotrofic Factor and NGF) from silk film and attributed that to the ionic interaction between the silk and the growth factor. For the B silk electrospun fibers, ionic interactions could explain the weak release as silk is negatively charged (pI 4.3) and BDNF (pI 9.1) is positively charged. However this is not the case for the C silk as CNTF is also negatively charged with an isoelectric point of 5. This unexpected result might also be due to the ratio between the silk and the growth factors. It has been reported that silk electrospun fibers loaded with epidermal growth factor (pI 4.6) have a burst release of 30% ^[30] however, the molecular weight of commercially available EGF (12 kDa) is significantly lower than CNTF (23 kDa) and BDNF (27 kDa) and this may be a factor in the stronger immobilization. Another factor that could partially explain the weak release is the absence of degradation of silk/peo fibers in PBS. ^{[30, 60, 61]} This property of those fibers implies that no growth factors can be released due to the degradation of the fibers, limiting their release process to diffusion through the fibers. The ELISA results combined with the effects of the functionalized fibers on survival and growth of neurons showed that growth factors present in the electrospun silk fibers were weakly released in the medium and exhibited a longer half life compared to factors directly added to the medium. Moreover, these factors retained their biological activity. Average neurite extension and number of RGC exhibiting regenerating axons were dramatically increased when cultured on the functionalized fibers, demonstrating the efficiency of the functionalization on axon growth and elongation. Even if we cannot rule out the fact that the initial small release of growth factors was responsible for the increase in RGC survival and neurite extension, it is highly unlikely that such low amount of factor would have any biological activity according to previously published dose dependant experiments. ^[37] It is more likely that entrapped growth factor retain their biological properties as shown previously ^[62] For example, activation of trk receptors does not require internalization of neurotrophins. Binding of neurotrophins induces a dimerization of the trk receptors that are then activated by transphosphorylation on their intracellular domains. ^[63] Therefore, multifunctionalization of electrospun silk fibers represents a major advantage compared to local application of growth factors as it protects them from rapid degradation and even avoids uncontrolled and over concentrated release that can be detrimental to retinal structure and function, causing pathological conditions such as gliosis. ^[64–66]

In the near future we aim to develop RGC culture over several weeks to demonstrate the ability of silk fibers to support their survival on the long term. However, this needs further slik guide development. The two factors, BDNF and CNTF were chosen because they are required in culture medium to support the growth of rodents RGCs. ^[33,35–37] Other candidate factors, such as NGF or GDNF, could be used as well to further enhance the outcomes reported here. ^[30,50] The approach described in the present work can also be generalized further, such as to allow the incorporation of additional components to foster matrix reorganization (e.g., matrix metalloproteases ^[67]) in order to reduce glia scaring, to increase adhesion and affinity of the growth cone to the silk (L1 Cell adhesion molecules ^[68]) and to optimize neurite regrowth ^[65,69]. In this regard, it was observed that cultured astrocytes organized axially by stretching along the silk fibers (data not shown). Silk contact may therefore reduce glia cell proliferation and resulting glia scar, further promoting growth cone elongation along this glia/silk guide resembling bands of Büngner found in the regenerative peripheral nervous system ^[70] mimicked by sciatic nerve grafts of the optic nerve ^{[71, 72]} Sciatic nerve grafts may however lead to peripheral cell proliferation brought by this permanent cell-loaded graft material, a safety concerns that can be circumvent by the use of degradable and acellular electrospun silk fibers.

4. Conclusions

To our knowledge, this is the first attempt to rescue RGC cell death and enhance their regeneration using biofunctionalized electrospun material. The present results demonstrate the possibility to generate neurite regrowth systems that combine growth support, directionality and the neurotrophic stimulation with one multifunctional biomaterial. Altogether silk fibroin fibers represent a promising implant strategy for optic nerve trauma rescue. Ongoing studies focus now on the evaluation of these new materials in an in-vivo model of optic nerve regeneration by regrouping aligned fibers obtained by longer electrospinning time in a 3D nerve-type structure to allow their surgical implantation.

5. Experimental

Preparation of aligned silk fibers

An 8wt % aqueous silk fibroin solution was prepared as previously described [24,30]. This solution was mixed with a 5 wt % polyethylene oxide (PEO) (900,000 g mol⁻¹) in a 4:1 ratio to produce a 7.5% silk/PEO solution. The silk/PEO solution was delivered through a 16G stainless steel capillary at a flow rate of 0.005 ml min⁻¹ using a Sage syringe pump (Thermo Scientific, Waltham, MA). The capillary was maintained at a voltage of 8 to 9 kV and was mounted at the center of a 10 cm diameter aluminum plate. The collector was a grounded wheel in aluminum of 20 cm diameter and 9 mm width. The outside circumference of the wheel was composed of a succession of seventy 9×9 mm squares on which 9×9 mm cover glasses were glued with a rubber cement (Elmer, Columbus, OH). We glued coverslips only every other square to leave grounded metallic gaps between the coverslips. This was done to improve the fiber alignment. Silk fibroin electrospun fibers were collected on the cover glasses. The deposition time was 5 minutes. The distance between the capillary and the top of the wheel was 17 to 20 cm. The wheel was rotated at speeds varying between 6 and 14 m.s−1 using a small dc motor (Bodine, Chicago, IL). BDNF (Peprotech, Rocky Hill, NJ), CNTF (Peprotech) or both growth factors, prepared in PBS solution, were added to the silk/PEO solution at a concentration of 1.25 μg ml⁻¹ to make the biofunctionalized fibers. Cover glass coated with silk fibers were then stored in dry atmosphere until use.

Immunopurification of RGCs and cell culture

Postnatal RGCs were isolated using a modified protocol adapted from previously published procedure in rats ^[33,34] and mice. ^[35–37] Postnatal Long Evans rats (7 days old; Medical Experimental Center, Faculty of Medicine, University of Leipzig) were sacrificed using CO₂ according to institutional guidelines. To isolate RGCs, retinae were incubated for 45 min at 37°C in D-PBS (Gibco/Invitrogen, Karlsruhe, Germany) containing 160 U.ml⁻¹ papain and 200 U.ml⁻¹ DNAse (Sigma-Aldrich, Munich, Germany). The tissues were then sequentially triturated in D-PBS containing 0.4 % BSA (fraction V, A 8806, Sigma), 650 U ml⁻¹ DNAse and 1:75 rabbit anti-rat macrophage antibody (AI-A51240, WAK Chemie Medical, Steinbach/Ts, Germany). After trituration, and filtration, the cell suspension was incubated on a subtraction plate for 30 min. For immunopanning, two subtraction plates (150 mm diameter Petri-dishes; Falcon; BD Biosciences) and one selection plate (100 mm diameter Petri-dish) were incubated for > 12 h at 4°C with 10 μg ml⁻¹ secondary antibody in 50 mM Tris-HCl (pH 9.5) (for subtraction: goat anti-rabbit IgG; for selection of rat RGCs: goat anti-mouse IgG (Jackson Immunoresearch Laboratories/Beckman Coulter). After washing, selection plates were incubated for > 4 h at room temperature with 0.2 μg ml⁻¹ primary antibody (mouse IgG anti-Thy1 clone Ox7, ab225, AbCam Cambridge, UK) and then washed with D-PBS. The cell suspension was then incubated for 45 min on the selection plate. Non-adherent cells were thoroughly washed off using D-PBS and bound cells were released by trypsination (12,000 U ml⁻¹ in EBSS for 10 min in 5% CO2 at 37°C, T9201 Sigma-Aldrich). Following inactivation of trypsin by 30% fetal calf serum (Gibco/Invitrogen), cells were cultured in Neurobasal medium (Gibco/Invitrogen) supplemented with (all from Sigma, except where indicated) pyruvate (1 mM), glutamine (2 mM; Gibco/Invitrogen), N-acetyl-L-cysteine (60 μg ml⁻¹), putrescine (16 μg ml⁻¹), selenite (40 ng ml⁻¹), bovine serum albumin (100 μg ml⁻¹; fraction V, crystalline grade), streptomycin (100 μg ml⁻¹), penicillin (100 U ml⁻¹), triiodothyronine (40 ng ml⁻¹), holotransferrin (100 μg ml⁻¹), insulin (5 μg ml⁻¹), progesterone (62 ng ml⁻¹), B27 (1:50, Gibco/Invitrogen), and forskolin (10 μM). This medium contain all component required for RGC survival with the exception of growth factors will be therefore referred to as minimal medium (MM). To support neuronal survival, this medium was further supplemented with BDNF (25 ng ml⁻¹, this medium will be referred to as MM+B), CNTF (10 ng ml⁻¹, this medium will further be referred to as MM+C) or a combination of both factors (this medium will be referred to as MM+BC). For the cell studies, 300 cells mm⁻² were grown per glass slides coated with silk fibers for 3 days in controlled atmosphere (37°C, 90% humidity, 5% CO2).

Immunostaining

Cells were fixed using parafomaldehyde solution (4% in PBS). Blocking and membrane permeabilization were performed for 20 min using a PBS solution with 30% casblock (Invitrogen) and 0.2% tween 20. Phalloidin staining of the actin filaments was performed using a rhodamine-coupled phalloidin used at 1/1000 in PBS (Invitrogen). Primary antibodies mouse anti beta III tubulin (MMS435P, Covance, Emeryville, CA) and rabbit anti gap43 (ab7462, Abcam) were diluted 1/500 an 1/1000 respectively in a PBS solution containing 5% casblock and 0.02% tween 20, and incubated overnight at 4°C. After PBS washing secondary antibodies (1/500, Alexafluor 488 and 555 developed in goat, Invitrogen) together with nuclear marker (DAPI 1/1000, Invitrogen) were incubated in PBS for 45 min at room temperature. Slides were mounted and examined with a fluorescence microscope (Axioplan2 Zeiss coupled to a Axiocam MRC5). Images were archived using Axiovision 4.6 software.

Morphometric analysis

Neuronal survival in the different culture conditions was determined using a live/dead assay (Molecular Probe). The percentage of surviving cells was calculated using 30 random fields for each culture conditions. To characterize the regenerative properties of the silk guide, the longest neurite extension per neuron on the different biofunctionalized silk fibers was measured. Pictures of 30 random fields were taken after staining with anti beta III tubulin antibody. The software Image J with the Neuron J pluggin was used to measure the longest neurite of neurons which developed projections. Neurites were mesured from soma to growth cone. Statistical analysis was performed using Student’s test to identify any significant increase in neurite extension on biofunctionalized silk compared to the control situation. Under the same conditions, the percentage of RGC growing an axon on the different silk types was also assessed by counting the number of RGC exhibiting gap43 positive neurites vs the total number of RGC present in the culture. All experiments were performed at least in triplicate.

ELISA test assay

ELISA test assays using the Quantikine BDNF kit (R&D Systems, Minneapolis, MN) and the RayBio Rat CNTF ELISA kit (RayBiotech, Norcross, GA) were performed to assess amounts of growth factors released from the fibers as well as amounts remaining in them. To measure the growth factor release, pieces of biofunctionalized mats were placed in PBS with 0.2% bovine serum albumin (BSA) at 37°C. Supernatant samples were taken after 1, 4, 10, 24, 48, 96, and 144 hours. Assuming that all the solvent evaporated after the fiber formation and knowing that a concentration of 1.25 μg ml⁻¹ of growth factors in 7.5% silk/PEO solution was used to make the mats, the mats are made of 1.7 10⁻³ wt% of growth factor and 100% release of growth factor would be 17 μg of growth factor per g of mat.

To estimate the amount of growth factor remaining in the fibers the silk electrospun mats were dissolved in 4 μl of 9.3 M LiBr per mg of mat for 30 min at room temperature. The resultant solution was then diluted down 11 times in PBS with 0.2% BSA and dialyzed against 1 L of DI water for 6 hours. Fibers were either soaked in PBS with 0.2% BSA or left dry at 37 °C for 1 week. The presence of growth factors was measured after 1 week for both conditions. Again, 100% of growth factors remaining in the fibers corresponded to 17 μg of growth factors per g of mat. To characterize the stability of the growth factors in solution, the concentration of BDNF and CNTF diluted in PBS with 0.2% BSA was measured over time. A growth factor concentration of 3.125 ng.ml⁻¹ was used. Samples were taken after 1, 4, 10, 24, 48, 96, and 144 hours and assayed with the same ELISA kits.

Atomic Force Microscopy (AFM)

Glass coverslips (Fisher Scientific) were covered with electrospun silk fibers. The glass surface appeared flat with an Ra (arithmetic average of the absolute values of the surface height deviations measured from the mean plane) lower than 0.3 nm. Atomic force images were obtained in tapping mode using silicon TESPA cantilevers from Digital Instruments (Santa Barbara, CA) on a Dimension 3100 AFM with Nanoscope V controller, also from Digital Instruments). Several scans were performed over a given surface area. These scans produced reproducible images to exclude sample damage induced by the tip. Height and phase mode images were scanned simultaneously at a fixed scan rate of 1 Hz with a resolution of 512×512 pixels. Profilometric section analyses allowed determination of local texture of fiber surfaces, taken in the direction along the length of the fiber. Ra measures and profilometric analysis were performed with the software nanoscope 7.2.

Scanning Electron Microscopy (SEM)

For electrospun silk mats, the materials were sputter-coated with platinum/palladium using a 208 HR Sputter Coater (Cressington, UK). Specimens were then examined using a field-emission scanning electron microscope (FESEM ultra 55, Zeiss, NY) at 5 kV. For RGCs, slides were fixed in 2.5% glutaraldehyde in 0.1 M PBS (pH 7.4) for 40 minutes and postfixed in 1% OsO4 in 0.1 M PBS for 60 minutes following extensive washing in PBS. Samples were dehydrated in graded series of alcohol (from 25 to 90% ethanol in PBS) for 10 minutes each. The samples were then coated with gold and examined with a Philips XL-30 scanning electron microscope at 5kV.

Statistical analysis

Unless specified, the data were expressed as means ± SD and analyzed by using one-way ANOVA. Values of p<0.05 were considered statistically significant.