Nerve conduits and growth factor delivery in peripheral nerve repair
Abstract
Abstract Peripheral nerves possess the capacity of self-regeneration after traumatic injury. Transected peripheral nerves can be bridged by direct surgical coaptation of the two nerve stumps or by interposing autografts or biological (veins) or synthetic nerve conduits (NC). NC are tubular structures that guide the regenerating axons to the distal nerve stump. Early synthetic NC have primarily been made of silicone because of the relative flexibility and biocompatibility of this material and because medical-grade silicone tubes were readily available in various dimensions. Nowadays, NC are preferably made of biodegradable materials such as collagen, aliphatic polyesters, or polyurethanes. Although NC assist in guiding regenerating nerves, satisfactory functional restoration of severed nerves may further require exogenous growth factors. Therefore, authors have proposed NC with integrated delivery systems for growth factors or growth factor–producing cells. This article reviews the most important designs of NC with integrated delivery systems for localized release of growth factors. The various systems discussed comprise NC with growth factors being released from various types of matrices, from transplanted cells (Schwann cells or mesenchymal stem cells), or through genetic modification of cells naturally present at the site of injured tissue. Acellular delivery systems for growth factors include the NC wall itself, biodegradable microspheres seeded onto the internal surface of the NC wall, or matrices that are filled into the lumen of the NC and immobilize the growth factors through physical-chemical interactions or specific ligand-receptor interactions. A very promising and elegant system appears to be longitudinally aligned fibers inserted in the lumen of a NC that deliver the growth factors and provide additional guidance for Schwann cells and axons. This review also attempts to appreciate the most promising approaches and emphasize the importance of growth factor delivery kinetics.
Introduction
Peripheral nerves possess the capacity of self-regeneration after traumatic injury. The quality of functional regeneration depends on a number of factors including size and location of the injury as well as the age of the individual. Severed nerves do not spontaneously restore their function and continuity of the nerve has to be reestablished first by microsurgical intervention such as by epineurial or individual fascicles’ suturing. In case of important loss of nerve tissue, an autograft must be interposed (Schmidt and Leach, 2003). This treatment, however, is associated with morbidity and potential neuroma formation at the donor site, besides frequent disappointing functional outcomes (de Medinaceli et al., 1993; Jaquet et al., 2001; Kim et al., 2003). An interesting alternative to direct end-to-end suturing of nerve stumps or interposing of an autograft is the insertion of a nerve conduit (NC). An NC is a tubular structure designed to bridge the gap of a sectioned nerve, protect the nerve from the surrounding tissue, e.g., scar formation, and guide the regenerating axons into the distal nerve stump. Comparative clinical studies between autograft and NC implantations in relatively short gaps revealed similar functional outcomes with both the autografts (the gold standard) and the collagen NC (Archibald et al., 1995) or polyglycolide NC (Battiston et al., 2005).
Nerve regeneration requires a complex interplay between cells, extracellular matrix, and growth factors. (Cytokines and growth factors are polypeptides that are produced by a variety of cells and often show overlapping actions. Thus, the two terms are often used synonymously, although cytokines are inducible proteins acting as humoral regulators, whereas growth factors were originally defined as substances that promote cell growth. Throughout this review, for simplicity, we will use the term of growth factors for all peptides and proteins that show a stimulatory effect on nerve regeneration.) The local presence of growth factors plays an important role in controlling survival, migration, proliferation, and differentiation of the various cell types involved in nerve regeneration (Fu and Gordon, 1997). Therefore, therapies with relevant growth factors received increasing attention in recent years. However, growth factor therapy is a difficult task because of the high biological activity (in pico- to nanomolar range), pleiotrophic effects (acting on variety of targets), and short biological half-life (few minutes to hours) (Tria et al., 1994) of these protein drugs. Thus, growth factors should be administered locally to achieve an adequate therapeutic effect with little adverse reactions. Severe adverse effects without beneficial improvement of the disease was observed, e.g., after systemic administration of ciliary neurotrophic factor (CNTF) to treat amyotrophic lateral sclerosis (ALS CNTF Treatment Study Group, 1996; Miller et al., 1996). Factors with a higher specificity, e.g., members of the neurotrophin (NT) family [nerve growth factor (NGF) (Petty et al., 1994) and NT-3 (Chaudhry et al., 2000)] may cause less severe side effects. Nevertheless, the short biological half-life of growth factors demands for a delivery system that protects and slowly releases locally the protein over a prolonged period of time. Therefore, growth factor delivery for peripheral nerve regeneration may be ideally combined with an NC.
NC are nowadays mostly made of biodegradable materials such as aliphatic polyesters or polyurethanes, collagen, chitosan, or excised vein (Table 1). Nonetheless, silicone has long been the most frequently used NC material. However, its nondegradability and relative stiffness often caused long-term complications. Depending on the material, NC are commonly produced by melt extrusion, dip coating, casting, or depositing material on a rotating mandrel. Typical NC dimensions for experimental use in small animals are inner diameters of 1–2 mm and lengths of several millimeters, depending on the experimental gap length. A comprehensive review on NC materials and manufacturing, including an early history of NC, has been published by Fields et al. (1989).
| NC material | Characteristics | Reference |
|---|---|---|
| Silicone: poly(dimethylsiloxane) | Highly elastomeric polymer, nonbiodegradable, bioinert, impermeable, hydrophobic | Lundborg et al. (1982a); Braga-Silva (1999) |
| Collagen: structural protein | Chemically cross-linked protein, enzymatically degradable, good cell interactions, NC approved by the Food and Drug Administration (FDA) | Archibald et al. (1991; 1995); Li et al. (1992); Krarup et al. (2002) |
| PGA: poly(glycolide) | Aliphatic polyester, biodegradable by hydrolysis, NC approved by FDA | Weber et al. (2000); Battiston et al. (2005) |
| PHB: poly(3-hydroxybutyrate) | Aliphatic polyester, biodegradable by hydrolysis | Young et al. (2002); Mosahebi et al. (2003) |
| PLLA: poly(L-lactide) | Aliphatic polyester, very slowly biodegradable by hydrolysis | Widmer et al. (1998); Evans et al. (2000) |
| Chitosan: β-(1-4)-linked D-glucosamine | Polysaccharide, enzymatically degradable, positively charged, good cell interactions | Itoh et al. (2003); Yamaguchi et al. (2003) |
| Polyester urethane | Elastomeric polymer, biodegradable | Borkenhagen et al. (1998) |
| PHEMA-MMA: poly(2-hydroxyethyl methacrylate-co-methyl methacrylate) | Hydrogel-forming polymer, nonbiodegradable, stiffness in the range of nerve tissue | Dalton et al. (2002); Belkas et al. (2005) |
This review focuses on localized growth factor delivery from NC, which has basically been approached in three different ways: (1) delivery of proteins from the NC lumen or NC wall directly to the target nerve; (2) seeding cells inside the NC lumen that produce the growth factors; and (3) use of gene therapy to transfect resident cells to express a certain protein (Fig. 1). In addition to reviewing these different approaches, we will also discuss some of the pros and cons of these systems and emphasize the importance of delivery kinetics on peripheral nerve regeneration. Finally, recent trends and future perspectives of growth factor delivery from NC will be highlighted. The readers should note that many other drugs, besides growth factors, have been used locally or systemically for promoting peripheral nerve regeneration. They include, e.g., triiodothyronine (Voinesco et al., 1998) and tacrolimus (FK506). Tacrolimus was originally introduced as an immunosuppressant to prevent rejection of allografts; later, it was discovered that tacrolimus also possesses nerve regenerative properties when applied systemically at sub-immunosuppressant doses on isografts. So far, tacrolimus is the only drug with demonstrated therapeutic effect on nerve regeneration after systemic administration (Gold, 1997; Chunasuwankul et al., 2002; Myckatyn and MacKinnon, 2004).
Schematic representation of ways to deliver growth factors from a nerve conduit (NC). From top to bottom: protein delivery system that directly releases growth factors to the lumen of the NC; transplantation of cells to the inside of the NC to synthesize and release growth factors; gene delivery to transfect resident cells to produce growth factors.
Challenges in the manufacturing and evaluation of growth factor delivery systems
Neuronal growth factors are challenging molecules with respect to their pharmacokinetic and pharmacodynamic properties. In addition, when formulating such growth factors in a drug delivery system, they require special attention with regard to their physicochemical stability during system manufacturing, storage, and in vitro assessment of their release into an aqueous environment. Moreover, in the context of localized delivery of such growth factors, little is known about required doses and dosing regimens or appropriate nerve defect and animal models.
The stability of growth factors is generally very limited so that denaturation and partial or complete loss of biological activity are frequent events in processing and formulating these compounds (Eng et al., 1997; Fu et al., 2000; Zhu et al., 2000). Therefore, growth factors require mild processing conditions such as ambient or low temperatures, little exposure to organic solvents, and presence of stabilizing additives. Moreover, to obtain sterile dosage forms or delivery systems for growth factors, all processes have to be performed aseptically because terminal sterilization through g-rays or heat would destroy the proteins. Aseptic processing, however, requires careful validation and is very cost intensive.
The in vitro assessment of the biological activity of formulated growth factors requires very sensitive assays, as the therapeutic dose in a delivery system is generally in the nanogram to low microgram range. For monitoring the sustained release of growth factor over several days to weeks, only a few nanograms must therefore be accurately quantifiable. Immunoassays [e.g., enzyme-linked immunosorbant assay (ELISA)] are most frequently used for such quantification, although they may suffer from inaccuracy due to conformational changes of the protein. For more accurate quantification, radiolabeled compounds may be more appropriate. Radiolabeled compounds are further useful to trace the distribution and elimination of the growth factor in vivo (Austin et al., 1997). However, neither ELISA nor radiolabel measurements provide information about the bioactivity of the protein. The bioactivity of growth factors can be determined, e.g., by measuring the neurite outgrowth of cultivated PC12 cells (Ohuchi et al., 2002) or dorsal root ganglion neurons (Gavazzi et al., 1999; Deister and Schmidt, 2006). Hence, a combination of methods is required to characterize fully the in vitro performance of a delivery system.
For testing the therapeutic efficacy of growth factor delivery systems, the use of an optimal local dose and delivery kinetics is critical, but not generally known. Whereas subtherapeutic doses will have no effect, exceeding doses or inappropriate release kinetics may lead to downregulation of the inherent production of growth factors and their receptors or to increased internalization of the receptors (Jullien et al., 2002; Geetha et al., 2005). The therapeutic testing of growth factor delivery systems further requires meaningful nerve defect and animal models as well as assessment methods. A wide variety of such models and methods have been used in past studies. Among them, the rat sciatic and rabbit facial nerves with nerve gaps of 5–15 mm were most frequently used. The gap distance is very critical for nerve regeneration, especially when the gap is bridged with an impermeable silicone NC. Silicone NC have produced relatively good results in gap lengths of less than 5 mm (Lundborg et al., 1982a; Dahlin and Lundborg, 2001), but poor or no nerve regeneration at gap lengths of 10 mm and above (Lundborg et al., 1982a; Francel et al., 1997; Heijke et al., 2001).
Assessment of nerve regeneration is frequently done by histomorphometrical evaluation, e.g., by counting myelinated axons. However, the number of myelinated axons does not generally correlate with functional parameters such as fluorogold retrolabeling, nerve conduction studies (Ahmed, 2005), muscle contraction force under nerve stimulation, or mobility of limbs as quantifiable by video gait or walking track analysis (Varejao et al., 2004). Poor functional data in comparison to histomorphometrical data can be attributed to a mismatch among axons reinnervating the target (Scott, 1996). It should be realized that different testing procedures are generally poorly correlated so that multiple methods should be used to evaluate the effect of any intervention on nerve regeneration. The lack of functional testing in many published studies prohibits the comparison of regeneration efficacy of different NC. Proper assessment of nerve regeneration finally requires the inclusion of both negative controls (empty NC) and positive controls (direct suturing or autografts). Unfortunately, published studies mostly lack the latter control group.
The frequently observed lack of accurate regeneration of axons to their original target end organ to yield complete functional recovery after nerve lesions has stimulated extensive investigations on the regulation of motor and sensory axonal growth. Regenerating axons exhibit a strong preference to grow along Schwann cell basal lamina tubes in the distal nerve stump. The generation of several collateral branches from lesioned axons is one mechanism by which regeneration accuracy is increased. Studies in rats and nonhuman primates have repeatedly shown that regenerating motor axons initially grow into both sensory (cutaneous) and motor (muscle) nerve branches, but are pruned over time from the cutaneous branch, which eventually results in preferential motor innervation (Brushart, 1988; 1993; Madison et al, 1996). This suggests that immediate mechanical guidance alone may not be crucial, but factors from the pathway itself or end organ reinnervation are very important (Brushart et al., 1998). Indeed, various factors have been described that regulate collateral generation and pruning of axons. A recent study provided evidence that Schwann cells express distinct sensory and motor phenotypes and that these phenotypes are associated with modality-specific promotion of axon regeneration (Höke et al., 2006). The phenotypes were distinguished by their characteristic patterns of growth factor expression. It is believed that growth factor production by denervated Schwann cells in the distal nerve segment is crucial for successful peripheral nerve regeneration. Whereas the Schwann cells of cutaneous nerve predominantly produce NGF, brain-derived neurotrophic factor (BDNF), vascular endothelial growth factor (VEGF), NGF, and insulin-like growth factor I, at different kinetics, glial-derived neurotrophic factor (GDNF) and pleiotrophin are upregulated substantially by both cutaneous nerve and ventral root Schwann cells (Höke et al., 2006). This important information is of course critical when designing NC loaded with growth factors. Past and present studies on neuronal growth factor delivery from NC are reviewed.
Delivery strategies and systems
Protein delivery
There are various avenues for delivering locally a protein from an NC to the site of injury, whereby the lumen of the NC is either left empty or filled with material (Fig. 2). An empty lumen has the advantage of allocating space for nerve regeneration so that the axons may selectively reinnervate the appropriate target. Conversely, a filled lumen provides a supporting structure that may favor cell ingrowth and guidance. The pros and cons of various delivery systems in connection with NC are discussed in the following and illustrated with selected examples. Growth factors that had been used to stimulate peripheral nerve regeneration are compiled in Table 2.
Protein delivery strategies comprising a nerve conduit with open (A–D) or filled lumen (E–H). Unprotected growth factor in solution in the lumen of an impermeable NC (A), growth factor encapsulated or covalently linked to the wall of NC (B), microspheres in the lumen or wall of NC (C), and subcutaneously implanted osmotic minipumps or injection devices connected via a catheter to the NC (D). Growth factors can be immobilized in a matrix through ion interaction (E), through binding to extracellular matrix domains that present the growth factors in a favorable conformation for receptor interaction (F), or suspended in a viscous gel (G). The lumen may also be filled with longitudinally aligned fibers that release growth factors and provide additional guidance to the Schwann cells and axons (H).
| Growth factor | Major target* | Reference |
|---|---|---|
| NGF: nerve growth factor | Sensory neurons, small axons | Lee et al. (2006) |
| NT-3: neurotrophin 3 | Sensory neurons, small- and medium-size axons | Midha et al. (2003) |
| BDNF: brain-derived neurotrophic factor | Sensory neurons, large axons | Terris et al. (2001) |
| GDNF: glial-derived neurotrophic factor | Motor neurons | Barras et al. (2002) |
| FGF-1: fibroblast growth factor 1 (acidic fibroblast growth factor) | Vascular endothelial cells | Cordeiro et al. (1989) |
| FGF-2: fibroblast growth factor 2 (basic fibroblast growth factor) | Ohta et al. (2004) | |
| GGF: glial growth factor | Schwann cells | Mohanna et al. (2005) |
| PDGF: platelet-derived growth factor | Schwann cells | Wells et al. (1997) |
| CNTF: ciliary neurotrophic factor | Schwann cells (injury factor) | Ho et al. (1998) |
| VEGF: vascular endothelial growth factor | Vascular endothelial cells | Hobson (2002) |
| IGF-I: insulin-like growth factor I | Inflammatory cells (anti-inflammatory) | Fansa et al. (2002) |
| LIF: leukemia inhibitory factor | Neurons (injury factor) | McKay Hart et al. (2003) |
- * Many growth factors are pleiotrophic and act on a variety of cells. Here, however, we indicated only the putatively most important targets for peripheral nerve regeneration.
Growth factor in solution inside the lumen of silicone NC
Silicone NC have been extensively used to bridge 6- to 8-mm-long gaps of rat sciatic or rabbit facial nerves. They have been instrumental for studying biological events during nerve regeneration. For example, neurotropism was demonstrated with the help of a Y-shaped silicone NC (Politis and Spencer, 1983; Seckel et al., 1986). The Y-basis of the NC was sutured to a proximal nerve stump, and one of the arms of the Y-shaped NC was sutured to the distal end of the severed nerve (Seckel et al., 1986) or filled with a polymer matrix that contained a homogenate of the severed nerve (Politis and Spencer, 1983; Seckel et al., 1986); the other arm of the NC was sutured to the distal end of a different nerve or filled with polymer matrix without nerve tissue homogenate. This design demonstrated the selective growth of axons toward diffusible factors released from distal stumps of the severed nerve. The confined space of impermeable silicone NC was further useful for assessing cellular and molecular events occurring during nerve regeneration (Lundborg et al., 1982b): (1) sampling of tissue fluids from the NC inserted between two nerve stumps revealed the production of endogenous factors after nerve transection (Lundborg et al., 1982c; Bates et al., 1995) and (2) loading the NC with exogenous factors such as the extracellular matrix components laminin, fibronectin, or collagen (Chen et al., 2000; Verdu et al., 2002), or growth factors such as NGF demonstrated growth-promoting effects.
A silicone NC filled with NGF and implanted in a rat sciatic nerve gap yielded significantly more myelinated axons in the distal part of the defect at 4 weeks after surgery than a NGF-free control NC (Rich et al., 1989). This difference between the NGF-loaded and the control NC had vanished, however, after 10 weeks, when both treatments produced almost complete regeneration (Hollowell et al., 1990). The beneficial impact of NGF inside a silicone NC persisted even up to 28 weeks in a rabbit facial nerve gap (inferior alveolar), where significantly more myelinated fibers, thicker myelin sheaths, and higher conduction velocities were found in the test vs. the control group after both 12 and 28 weeks (Bu et al., 1999). Yet, as no positive control was used, the effective long-term benefit of the NGF-loaded NC remains ambiguous because the negative control NC may have simply had a strong inhibitory effect on axonal growth. One of the few studies that compared NGF-loaded NC against an autograft (the gold standard) determined significantly more myelinated fibers in the autograft than in the NGF/NC group after 5 weeks (Spector et al., 1995). In the autograft, however, a majority of the axons were localized in the extrafascicular connective tissue and, thus, unlikely to find their way to the distal nerve stump. Conversely, the axons in the NGF/NC had unrestricted space for growing selectively into the appropriate distal nerve sheath. Therefore, one may speculate that the axons of the NGF/NC group would have achieved a higher number of hits than those of the autograft group, so that functional outcome would not necessarily have been hampered by the smaller number of axons. Unfortunately, no functional tests were performed in this study.
The use of impermeable silicone NC has nowadays been largely abandoned despite the convenience of their manufacturing, loading with soluble growth factors, and surgical handling. A major shortcoming of silicone NC is obviously their nonbiodegradability, which often caused complications so that 25–50% of NC used clinically had to be explanted because of patients’ discomfort (Braga-Silva, 1999; Lundborg et al., 2004). Presently, there is a common consensus that NC for nerve repair should be made of biodegradable materials (Lundborg et al., 2004). Another drawback of silicone NC is their restricted use for relatively short nerve gaps of up to 6–8 mm and the required large inner NC diameters to prevent nerve compression by the NC (Merle et al., 1989; Dahlin and Lundborg, 2001). These relatively large inner NC diameters may accelerate the leakage of growth factors from the lumen.
Growth factor delivery from a matrix inside the lumen of NC
The entrapment of growth factors in a matrix that is loaded into the lumen of an NC is one of many attempts to prolong localized drug availability over a period of days to weeks. Growth factors entrapped in such a way encompass NGF (Lee et al., 2003), BDNF (Terris et al., 2001; Midha et al., 2003), NT-3 (Midha et al., 2003), fibroblast growth factor (FGF)-2 (Ohta et al., 2004), VEGF (Hobson et al., 2000; Hobson, 2002), platelet-derived growth factor (PDGF) (Wells et al., 1997), glial growth factor (GGF) (Mohanna et al., 2003; 2005), and leukemia inhibitory factor (LIF) (McKay Hart et al., 2003). Frequently used matrix materials include the hydrogel-forming collagen, laminin, alginate, heparin, and heparin sulfate. These matrix materials provide a more viscous environment in the NC lumen, thereby slowing down the exchange of fluid between lumen and external tissue and related rapid loss of growth factor. Depending on their physicochemical nature, the matrix materials may also interact molecularly with the growth factors through, e.g., ionic, electrostatic, or hydrophobic interactions, or through ligand-receptor binding such as via a heparin-binding domain, thereby prolonging the release and protection against enzymatic breakdown of the growth factors. Finally, the matrices may also promote nerve growth by serving as guidance for the ingrowth of Schwann cells and axons.
The use of a matrix inside an NC without a growth factor may either enhance (Madison et al., 1985; Chen et al., 2000; Toba et al., 2002), conditionally enhance (depending on the time point of assessment and conduit type) (Madison et al., 1987), or impede nerve regeneration (Valentini et al., 1987; Terris et al., 1999; Mohanna et al., 2005) as compared with an empty NC. It has been shown that relatively weak collagen or laminin hydrogel matrices mediate superior nerve regeneration than denser hydrogels (Labrador et al., 1998). Highly viscous hydrogels may obstruct the path for axonal growth toward the distal stump. We may even speculate that a very constraining directional guidance may hamper the chance of the regenerating axons, especially the motor axon projections, to reinnervate the appropriate endoneurial tube.
The use of electrostatic interactions between growth factor and matrix material successfully controlled the release of LIF from alginate hydrogel. LIF is a basic (pI > 10) pleiotrophic ‘injury factor,’ which is strongly upregulated immediately after trauma and acts as a survival factor for both the sensory and the motor neurons. LIF (125I labeled) embedded in a calcium alginate matrix was released in vitro at a daily rate of less than 1% of the dose over several months and remained detectable in vivo in the vicinity of the alginate gel for up to 6 months (Austin et al., 1997). In connection with a NC, LIF was incorporated into an alginate matrix supplemented with fibronectin, which was then filled into the lumen of a poly(hydroxybutyrate) (PHB) NC. Upon implantation of this NC in a rat sciatic nerve gap model of delayed nerve repair (2 and 4 months predegeneration), tissue regeneration was enhanced as compared with the implant without LIF (control NC) (McKay Hart et al., 2003). Nonetheless, best treatment results were achieved with an autograft, although the differences were statistically significant only between the autograft and the control NC. PHB NC filled with alginate hydrogel were also used for the delivery of GGF. This system implanted into a long peroneal nerve gap (20 and 40 mm) in rabbits afforded significantly better regeneration than the control NC without GGF (Mohanna et al., 2005).
An interesting alternative to ionic growth factor-matrix material interaction is the specific binding of many growth factors to components of the extracellular matrix (ECM). For example, FGF (Neufeld et al., 1987; Fannon et al., 2000), GDNF (Rider, 2003), and NGF (Neufeld et al., 1987) bind to heparin and heparan sulfate. Such binding may not only protect the growth factors from proteolytic cleavage (Saksela et al., 1988) but also enhance the receptor binding and signal transduction (Barnett et al., 2002; Tanaka et al., 2002). Release of NGF from heparin, which itself was bound covalently to a fibrin matrix, was slower than the release of NGF from the fibrin matrix alone (Sakiyama-Elbert and Hubbell, 2000). When this delivery system was loaded with three different NGF doses (approximately 0.125, 0.5, 1.25 ng) and introduced into a silicone NC to bridge a 13-mm rat sciatic nerve defect, the two higher NGF doses produced similar outcomes in terms of number of regenerating nerve fibers and fiber diameter as did an autograft; the empty NC and the delivery system alone without NGF were significantly less effective (Lee et al., 2003). An alternative system consisting of FGF-2 embedded in a heparin/alginate hydrogel was used without an NC to bridge a 10-mm gap in the rat sciatic nerve. The system improved nerve regeneration significantly as compared with a gel formulation without heparin (Ohta et al., 2004). However, it remains elusive whether such a viscous gel without NC is competitive for nerve regeneration because no comparison with an autograft was made.
In summary, NC filled with a growth factor–containing matrix generally mediated better nerve regeneration than NC filled with drug-free matrix or an empty NC. However, NC filled with drug-free matrix often yielded worse results than empty NC.
Growth factor delivery from osmotic minipumps or injection devices
Osmotic minipumps are implantable, mostly cylindrical devices that typically consist of a reservoir for the drug solution and an adjacent or surrounding chamber filled with an osmotically active agent that is enclosed in a semipermeable membrane (Theeuwes and Yum, 1976). When exposed to an aqueous environment, water penetrates into the osmolyte chamber and exerts a pressure on the reservoir that delivers the drug solution generally at zero-order kinetics via a catheter, which can be directed toward the NC or nerve gap. The rate of delivery is independent of the physicochemical properties and size of the molecules delivered, but controlled exclusively by the osmotic pressure exerted on the drug-containing reservoir. Thus, osmotic minipumps can deliver quite different drugs at constant rates for typically up to 4 weeks. An osmotic minipump has been used to determine the therapeutic window of growth factors such as BDNF and GDNF. BDNF, e.g., exhibited a bimodal dose response (Boyd and Gordon, 2002); high doses (>8 mg/day) inhibited motor neuron survival in a dose-dependent manner, whereas low doses (0.5–2 mg/day) had no detectable effect after immediate repair, but reversed the negative effects of chronic axotomy 2 months prior to nerve resuture. In contrast, GDNF showed no dose dependency within the tested range (0.1–10 mg/day), but a synergistic effect in combination with BDNF (Boyd and Gordon, 2003).
When constant drug delivery from a minipump is unsuitable, but a pulsatile release desired, injection devices are most useful. They represent subcutaneously implantable drug reservoirs, which deliver the drug via a catheter to the site of injury. The drug is filled periodically into the reservoir by means of syringe and needle. The use of an injection device for daily administration of NGF (660 ng) to an epineurial suture of the rat sciatic nerve enhanced nerve regeneration without inducing any device-related adverse effects (Santos et al., 1998; 1999). Unfortunately, the authors did not profit from the unique feature of injection devices to apply different delivery regimens, such as those matching the mRNA expression pattern of NGF or its receptors after axotomy.
A drawback of both osmotic minipumps and injection devices is their nonbiodegradability, requiring surgical explantation at the end of the therapy. In addition, their quite important size and the demanding microsurgical handling for placing the catheter may become problematic in locations where space is scarce, such as in hand surgery or experimental surgery in small animals. Furthermore, the maintenance of the catheter in the correct position during nerve regeneration appears difficult (Santos et al., 1998). Therefore, their use in therapeutic applications is limited. Nonetheless, they remain interesting devices for investigating the impact of growth factors in experimental setups due to the ease of modifying and controlling the drug release kinetics.
Growth factor delivery from microspheres
Biodegradable, polymeric microspheres have been considered for the delivery of neuronal growth factors in the context of both the implantation in the brain (Menei et al., 2000; Pean et al., 2000) and the combination with NC for peripheral nerve repair. In contrast to afore discussed single-unit delivery systems, microspheres are multiple-unit dosage forms, which may facilitate dosing regimens. They might, e.g., be useful for individualized growth factor dosing (depending on nerve type, gap length, patient’s age) by loading the desired amount of drug-containing microspheres into an NC immediately before NC implantation. Moreover, the combination of different microsphere types will afford different release kinetics for one or several growth factors. Despite these potential advantages, microspheres have not been widely used in the peripheral nervous system, which might be due to the complexity and difficulties of protein microencapsulation technology in terms of protein stability and aseptic manufacturing.
Microspheres can be loaded either into the NC wall (see section below) or into the NC lumen. In the lumen, the microspheres can be filled as aqueous suspension (Xu et al., 2002) or dispersed in a hydrogel (Rosner et al., 2003). As an example, NGF has been microencapsulated in a copolymer made of phosphoesters and ethylene terephthalate (PPE) (Xu et al., 2002). NGF in vitro release from the PPE microspheres lasted for 70 days, and the release rate could be modulated by coencapsulating bovine serum albumin (BSA) (NGF to BSA ratio of 1 : 3,000). The microspheres were introduced into NC made of PPE or silicone, which were implanted in a rat sciatic nerve gap of 10 mm (Xu et al., 2003). After 3 months, the NC with the NGF-containing microspheres produced significantly more fibers and a higher fiber density than the control groups (empty NC and NC loaded with microspheres containing only BSA). Incidentally, there was no consistent difference between the PPE and the silicone NC groups, although this was not explored in detail.
Growth factor delivery from the NC wall
Growth factor release from the wall of an NC leaves the lumen open for the axonal growth, which may result in faster reinnervation of the distal nerve sheath as compared with using a filled NC. The time axons need to reinnervate their targets is indeed a critical parameter, which was correlated with the long-term functional outcome (Krarup et al., 2002). Moreover, it has been hypothesized that an open lumen provides space for a more accurate innervation of the appropriate endoneurium tube and may thereby improve the functional outcome. A drawback of an open lumen is the lack of a supporting structure for the ingrowth of Schwann cells and axons; therefore, open lumen NC may be unsuitable for long gaps.
A very simple device for delivering growth factors from NC walls consists in NGF-impregnated fibronectin mats, which are sutured around the nerve stumps to form a conduit (Whitworth et al., 1996). Such mats released NGF over 7 days in vitro and significantly enhanced rat sciatic nerve regeneration as compared with mats without NGF. However, impregnation of a matrix with growth factors affords only limited control over the release kinetics. A better release control may be achieved by embedding growth factors into NC walls with a dense matrix structure (Langer, 1998). However, the use of a single polymeric material for two different functions, namely the structural function of an NC wall and the release-controlling function of a delivery system, suffers from the inherent drawback that the mechanical and permeability properties cannot be varied independently. This may greatly limit the achievable release kinetics. Therefore, NC with delivery systems integrated in the NC wall should consist of two or more compartments. Typical multicompartment designs may consist of a supporting structure (tube) that provides the mechanical strength and an integrated delivery system in the form of drug-containing coatings on the inner or outer side of the NC wall, drug-containing microspheres embedded in the NC wall, or covalently linked drugs.
A composite NC with a drug-releasing coating layer has been made of a nonbiodegradable poly(ethylene-co-vinyl acetate) (EVAc) tube. Its luminal side consisted of a porous layer of an EVAc/BSA blend containing the FGF-2, whereas the outer layer was made of pure EVAc and served as a diffusion barrier (Aebischer et al., 1989). BSA acted as porogen and eluted from the EVAc/BSA layer upon contact with aqueous medium, thereby facilitating the release of FGF-2. This NC type was further developed for controlling the release of GDNF, NGF, and NT-3 (Barras et al., 2002; Fine et al., 2002). The different growth factors were individually embedded into a thin EVAc/BSA rod by coextrusion; the rod was then attached to a steel mandrel and this setup dip coated with EVAc solution, thereby incorporating the rod into the NC wall. The in vitro growth factor release was characterized by a marked initial burst (during first few days) followed by a constant release of approximately 1 ng/day for 1 month. The important burst release of GDNF exerted a detrimental effect on nerve regeneration of an axotomized rat facial nerve. To eliminate the excessive initial dose, the NC were preincubated for 3 days before implantation, which significantly improved the therapeutic performance (Barras et al., 2002). This indicates a specific therapeutic window of GDNF and emphasizes the importance of optimizing release kinetics.
Attempts to engineer NC with controlled growth factor release kinetics are relatively scarce. In one approach, NGF and the stabilizer BSA were incorporated by three different techniques into the wall of poly(2-hydroxyethyl methacrylate-co-methyl methacrylate) P(HEMA-co-MMA) NC, which were coated on the luminal side with a porous layer of PHEMA: (1) a NGF/BSA solution was directly placed into the lumen of the NC so that the proteins adsorbed across the entire wall; (2) a NGF/BSA solution was added to the PHEMA-polymerization mixture used to coat the luminal side of the NC wall; (3) NGF and BSA were first coencapsulated into poly(lactic-co-glycolic)acid (PLGA) microspheres, which were then suspended in the PHEMA-polymerization mixture used to coat the luminal side of the NC wall (Piotrowicz and Shoichet, 2006). NGF adsorbed to the NC wall was almost completely released within 1 day, and the amount released was approximately 1 ng NGF per centimeter NC. NGF embedded as solution or in microencapsulated form into the porous PHEMA layer was released continuously over 30 days; the total amounts released per centimeter NC were approximately 8 ng NGF (embedded solution) and 0.2 ng NGF (embedded microspheres). A major disadvantage of the proposed P(HEMA-co-MMA) NC is their nonbiodegradability, although their hydrogel character may afford less long-term irritation as compared with the nonbiodegradable silicone NC.
Biodegradable two-ply NC were made of poly(D,L-lactide) (PLA), which consisted of a porous inner PLA tube containing FGF-2 and a dense outer PLA tube acting as a diffusion barrier (Wang et al., 2003). This NC type significantly enhanced nerve regeneration across a 15-mm gap in the rat sciatic nerve as compared with control NC without FGF-2. Another biodegradable composite NC consisted of an inner chitosan tube surrounded by an outer chitin tube, and of epidermal growth factor (EGF)-containing PLGA microspheres that were placed in-between the tubes. The microspheres were immobilized in the interspace of the two tubes during NC drying because the outer chitin tube shrank considerably more than the inner chitosan tube. EGF was released in vitro over 8 weeks, and its bioactivity maintained for at least 2 weeks (Goraltchouk et al., 2006). In our own work, we developed a biodegradable delivery system consisting of a porous collagen NC coated with concentric layers of different PLGA types, in-between which GDNF solution was embedded. GDNF was released in vitro over 1 month at a rate of a few nanograms per day. Changes in the polymer type (lactide : glycolide ratio, molecular weight, end-group hydrophilicity) and stacking sequence of the layers allowed us to tailor different release kinetics (L.A. Pfister, unpublished data).
Finally, growth factors have also been covalently bound to the luminal side of the NC wall. NGF was covalently linked via carbodiimide to a gelatin-tricalcium phosphate membrane (Chen et al., 2005). The in vitro release of biologically active NGF consisted of an initial burst (approximately 60% of total amount released) followed by a zero-order release over 50 days, which was controlled through the degradation of the matrix. In another device, BDNF was covalently linked to collagen NC. Implanted in a rat sciatic nerve gap, this NC mediated better functional outcome (sciatic functional index) than direct epineurial suture (Utley et al., 1996). When BDNF and CNTF were covalently coupled to a collagen NC, the functional recovery was significantly superior to that observed with the NC containing only BDNF (Ho et al., 1998).
Generally, NC that contained embedded or covalently attached growth factors in or at their wall yielded significantly better nerve regeneration than NC without growth factors. So far, however, only one study compared a growth factor–loaded NC to an autograft and reported better functional outcome for the NC (Utley et al., 1996). Changes in the NC wall architecture further permitted to modulate the release of growth factors from NC, which represents a promising means to test the impact of different release kinetics on nerve regeneration.
Cellular therapy
Schwann cells
Cell transplantation is another strategy to create a favorable environment for nerve regeneration (Fig. 1). The most obvious cells to use are autologous Schwann cells. They represent the glia cells of the peripheral nervous system and support the axons by sheathing them with insulating myelin layers and providing neurotrophic factors. Upon injury, Schwann cells display a central role in nerve regeneration (Frostick et al., 1998). They synthesize surface cell adhesion molecules (CAMs) and build up basement membranes for the support and guidance of the sprouting axons. Furthermore, Schwann cells are competent to secrete growth factors, such as NGF, BDNF, NT-3, CNTF, and GDNF (Watabe et al., 1995; Höke et al., 2006), which ensure the survival and regeneration of the neurons.
An inherent advantage of Schwann cell transplantation over single or multiple protein delivery is the bioresponsiveness of Schwann cells to react to local environmental stimuli by secretion of a variety of appropriate growth factors. In contrast, man-made delivery systems are very limited with respect to bioresponsiveness, manageable number of growth factors, and fine-tuning of their release kinetics. A major limiting factor of autologous Schwann cells, however, is the time of at least 3 weeks required for ex vivo proliferation and purification of the Schwann cells to obtain a sufficient number of viable cells. This longsome proliferation of Schwann cells in culture may heavily impair the therapeutic benefit in situations where immediate repair is possible and essential. Moreover, the ex vivo proliferation of Schwann cells is comparatively costly (labor, time, infrastructure) and is at inherent risk of contamination. Nevertheless, Schwann cells have been extensively used in recent years and they have proven to be particularly advantageous for the repair of long gaps. In this case, Schwann cells have been injected into an autograft or used in combination with acellular allografts.
Schwann cells can be accommodated in NC by either of the following ways: (1) by injection of cells into the NC lumen in the absence of a supporting structure (Ansselin et al., 1997; Zhang et al., 2002; Koshimune et al., 2003); (2) by cell seeding within a hydrogel that is subsequently filled into the lumen of a NC (Guenard et al., 1992; Strauch et al., 2001; Galla et al., 2004); and (3) by seeding the cells into the channels of a multichannel NC or onto fibers that are longitudinally aligned inside the NC (Cheng and Chen, 2002).
In a NC lumen devoid of a supporting structure, Schwann cells cannot adhere effectually, which may result in substantial loss of cells. Nonetheless, Schwann cells were successfully accommodated in the empty lumen of NC made of collagen (Ansselin et al., 1997) or autogenous vein (Zhang et al., 2002); these cell-loaded NC enhanced nerve regeneration in an 18-mm sciatic nerve gap in rats (Ansselin et al., 1997) and in a 4-cm tibial nerve defect in rabbits (Zhang et al., 2002) as compared with cell-free NC but performed worse than autografts. Mosahebi et al. (2001) examined the importance of the Schwann cell concentration (transfected with the lacZ gene) in a PHB NC on the repair of a 10-mm gap of a rat sciatic nerve. The length of regenerated nerve, as measured after 3 weeks, increased up to a concentration of 80 × 106 cells/ml (which is about four times the concentration found in normal nerves) and slightly decreased at higher cell concentrations (Mosahebi et al., 2001). To enhance the regenerating capacity of Schwann cell–seeded NC, some authors additionally subjected the repair site to repeated low-energy ultrasonic treatment (0.2 W/cm2). Such ultrasonic treatment indeed improved nerve regeneration in Schwann cell–loaded PLGA NC implanted in a 10-mm rat sciatic nerve gap, but induced massive fibrosis in combination with silicone NC (Chang and Hsu, 2004; Chang et al., 2005).
The seeding of Schwann cells within a hydrogel that is subsequently filled into a NC was performed, e.g., with collagen (Evans et al., 2002), alginate/fibronectin (Mosahebi et al., 2003), gelatine (Hsu et al., 2004), and Matrigel (Guenard et al., 1992; Levi et al., 1997; Udina et al., 2004). Incidentally, Matrigel is a commercially available matrix consisting of laminin-rich basement membrane components. While the benefit of the sole hydrogel in NC remained ambiguous, the addition of Schwann cells significantly improved nerve regeneration. For instance, vein grafts filled with Schwann cell–containing Matrigel successfully bridged up to 6-cm-long gaps (Strauch et al., 2001).
With the aim to increase the surface for attachment of Schwann cells, NC with multiple channels were fabricated (Hadlock et al., 2000). An NC made of PLGA foam with five channels yielded thicker fibers than autografts, although the amount of neural tissue between the two groups was similar. This was due to the reduced total space available for nerve regeneration in the multichannel NC, which was only 50% of that available in an autograft. A more favorable ratio of surface area for Schwann cell attachment and migration over space for regeneration and vascularization was afforded by NC filled with longitudinally aligned fibers. Such NC were produced with fibers of poly(glycolide) or polydioxanone; the NC lumen offered space to accommodate 16 fibers with a diameter of 70 mm (Shen et al., 2001) or 100 fibers with a diameter of 12 mm (Cheng and Chen, 2002). The fibers elicited only a mild immune response (Shen et al., 2001) and promoted efficiently nerve regeneration (Cheng and Chen, 2002). Finally, Schwann cells were also seeded on a strip of submucosa from the small intestine of neonatal rats and incubated to reach confluence. The strip was then rolled into a cylinder and used to bridge a 7-mm gap of a rat sciatic nerve (Hadlock et al., 2001). The Schwann cell–loaded cylinder yielded functional recovery that was intermediate between those achieved with a cell-free cylinder and an autograft.
To shorten and facilitate the process of Schwann cell extraction from nervous tissue and in vitro expansion, which usually takes 3–8 weeks, Nilsson et al. (2005) transected the rat sciatic nerve and left it untreated for 7 days before repair. On the day of repair, 2- to 3-mm-long proximal and distal pieces of nerve were removed from the severed nerve for immediate preparation of nonneuronal cells. A total of 95,000 cells with 80% being Schwann cells were obtained and injected into a silicone NC implanted for bridging the gap (Nilsson et al., 2005). Although the NC containing the cell suspension mediated superior regeneration as compared with cell-free NC, it remains unclear whether such low numbers of Schwann cells can sustain nerve regeneration to a similar extent as do autografts. For testing the feasibility of obviating the use of autologous Schwann cells, Guenard et al. (1992) compared syngeneic and heterologous Schwann cells, which were introduced at different concentrations (40, 80, and 120 × 106 cells/ml) into a semipermeable poly(acrylonitril-co-vinylchloride (PAN-PVC) NC filled with Matrigel. The NC were implanted in rat sciatic nerve gaps of 8 mm. The NC loaded with heterologous cells elicited a strong immune response and afforded only a limited growth of myelinated axons; conversely, the NC loaded with syngeneic cells were nonimmunogenic and afforded axonal growth to the distal stump (Guenard et al., 1992). A Schwann cell density of 80 × 106 cells/ml resulted in maximal regeneration, which is in agreement with the work of Mosahebi et al. (2001), who used NC without Matrigel. In another study, which used allogeneic Schwann cells within Matrigel-containing collagen NC, systemic administration of the immunosuppressant tacrolimus (FK506) was required for improving nerve regeneration (Udina et al., 2004).
Finally, Schwann cells have also been combined with growth factors inside NC. For instance, Schwann cells were seeded into a highly porous PLGA NC that had previously been immersed in GGF solution; the NC were then implanted in a 10-mm gap of a rat sciatic nerve (Bryan et al., 2000). The combination of Schwann cells and GGF inside the NC yielded a higher myelination index and faster electrical conduction velocity, but a smaller number of axons and blood vessels than the NC loaded with GGF alone. Thus, the combination of Schwann cells and growth factor appeared to exert a synergistic effect. Future work should also consider using Schwann cells of appropriate phenotype that produce the required growth factors for motor or sensory axonal growth (Höke et al., 2006).
In summary, NC loaded with Schwann cells generally yielded better nerve regeneration than cell-free NC but did not reach the levels achieved with autografts in the few studies with such positive control. However, Schwann cell–seeded NC showed impressive nerve regeneration across gap length of up to 6 cm, and the combination of Schwann cells with a guiding structure (e.g., longitudinally aligned fibers) or exogenous growth factors may indeed represent the most promising alternative to the use of autografts for bridging long gaps.
Stem cells
Bone marrow stromal cells, also known as mesenchymal stem cells (MSC), are pluripotent cells that have the potential to differentiate into bone, cartilage, fat, and muscle cells. Moreover, MSC were also transdifferentiated successfully into neural cells (Hermann et al., 2004). As MSC can be isolated relatively easily from bone marrow aspirates and expanded in culture, they provide an interesting alternative to Schwann cell transplantation.
MSC, labeled with green fluorescent protein, were recently seeded into a Matrigel-containing chitosan NC (Zhang et al., 2005). Upon implantation of the NC into a rat sciatic nerve gap of 5 mm, functional recovery in terms of conduction velocity and sciatic functional index was significantly improved as compared with MSC-free control NC and similar to that obtained with NC loaded with Schwann cells. The similar outcome of the two cell-loaded NC groups is quite remarkable considering that only about 5% of the MSC transdifferentiated into a Schwann cell–like phenotype, while the major cell population maintained an undifferentiated phenotype, as evidenced by S100 protein staining. Similar findings were also reported by others, and it was speculated that also undifferentiated MSC may have contributed to nerve regeneration by secreting growth factors and depositing basal lamina components (Cuevas et al., 2002; 2004). Transdifferentiated MSC with Schwann cell–like characteristics have been produced in an appropriate conditioning medium (containing heregulin, basic FGF, and PDGF) and seeded into Matrigel that was subsequently filled into hollow fiber made of polyethersulfone (Dezawa et al., 2001). This device loaded with differentiated MSC significantly enhanced nerve fiber growth across a rat sciatic nerve gap of 12 mm as compared with the control device loaded with undifferentiated MSC. After 6 months, the functional recovery was significantly better with the test device than with the control NC (Mimura et al., 2004). Although the mechanism of MSC transdifferentiation and the molecular cross talks between the MSC and the peripheral nerve are by far not fully understood, MSC may become a promising and abundant source for Schwann cell–like cells to overcome the bottleneck of Schwann cell proliferation. Incidentally, stem cells have also been isolated from hair follicles and adopted Schwann cell characteristics when placed between the stumps of a transected peripheral nerve (Amoh et al., 2005). However, extraction of a high number of hair follicle stem cells seems more laborious than harvesting MSC.
Gene delivery
Gene transfer in damaged nerve or Schwann cells has also been proposed for promoting nerve repair (Fig. 1). Difficulties associated with in vivo gene transfer are low transfection efficiency and, in case of viral vectors, safety. For peripheral nerve repair, a transient vector is preferable to avoid possible adverse effects after successful regeneration. Although adenoviral vectors are considered to be transient vectors, they are incorporated into the genome at low frequency (Harui et al., 1999; Mitani and Kubo, 2002); thus, insertional mutagenesis may occur even at low doses and upon local administration. Safety and other regulatory issues (Fox, 1999; Gonin et al., 2005) as well as patients’ acceptance will therefore constitute major hurdles for gene therapy in case of non–life threatening peripheral nerve injuries.
As a proof of principle, a replication-defective adenoviral vector carrying the gene for b-galactosidase (lacZ) was injected into the transected (Joung et al., 2000) or crushed (Sorensen et al., 1998) sciatic nerve of rats. About 5% of the Schwann cells expressed lacZ, which increased up to 18% upon daily injections of the immunosuppressant cyclosporine A, suggesting an immune response toward the vector. However, cyclosporine A not only enhanced transfection efficiency by reducing the lymphatic infiltration but also reduced proliferation of Schwann cells, which may impede functional outcome (Sorensen et al., 1998). Gene expression decayed with time, but remained detectable for up to 2 months.
DNA plasmid constructs were used for an ex vivo genetic modification of Schwann cells to overexpress two isoforms of FGF-2. Silicone NC filled with Matrigel and cells expressing the high-molecular isoforms of FGF-2 (21 and 23 kDa) significantly enhanced nerve regeneration across a rat sciatic nerve gap of 15 mm (Timmer et al., 2003). However, the 18-kDa isoform of FGF-2 exerted an inhibitory effect on the myelination of axons (Haastert et al., 2006).
Appraisal of most promising delivery systems
Analysis of the pros and cons of the above described systems and strategies leads us to identify the most promising delivery systems, although such appreciation suffers from the inherent difficulty of comparing and weighting data from differently designed studies (see Challenges in the Manufacturing and Evaluation of Growth Factor Delivery Systems section). For the present appraisal, it seems appropriate to distinguish between small and large gaps.
For small nerve gaps of a length of less than 6–8 mm, the repair time has a crucial impact on the functional outcome, whereas for very long gaps of one to several centimeters, the challenge resides in the principal difficulty to bridge such a gap. For short gaps, we advocate the use of NC with open lumen rather than with matrix-filled lumen so that the axonal growth is not slowed down by physical barriers before entering the distal nerve stump. In addition to the open lumen, a multicompartmental wall structure may offer opportunities for controlling drug release kinetics over prolonged periods of time and, ideally, tailoring growth factor delivery according to biological needs.
In very long gaps, the axons require guidance and neurotrophic support. Because it will take a long time to bridge the gap and the requirements for growth factors will change with proceeding regeneration, a delivery system capable of responding to stimuli of the local environment may be necessary. Autologous Schwann cells of appropriate phenotype seeded on longitudinally aligned fibers would probably best match such requirements. The fibers provide a large area for Schwann cell adherence and strong guidance toward the distal nerve stump, whereas the Schwann cells synthesize CAMs, basal lamina components, and growth factors that should contribute to successful regeneration. Therefore, for long gaps, it may be appropriate to delay the nerve repair treatment for the time that is needed to expand ex vivo appropriate autologous Schwann cells.
Drug delivery kinetics
Gene profiling and growth factor monitoring after nerve transection highlight the importance of drug delivery kinetics in nerve regeneration. Analysis of the mRNA expression for the NTs NGF (Heumann et al., 1987), BDNF (Meyer et al., 1992), NT-3, and NT-4 (Funakoshi et al., 1993) demonstrated distinct patterns. For example, NGF mRNA expression appeared to be biphasic, peaking at 12 h after lesion followed by a decline and a second rise at day 3 (Heumann et al., 1987). This biphasic pattern may explain the observation that high doses of exogenous NGF (3 mg/day) delay the early phase of nerve regeneration in adult rats (Hirata et al., 2002). It was hypothesized that the depletion of NGF at the target receptor may trigger the regenerative response. In contrast, BDNF mRNA is not upregulated immediately after axotomy, but increases steadily after day 3 (Meyer et al., 1992). Changes in the mRNA expression of the GDNF family members and their receptors after axotomy showed a high upregulation only for GDNF and its receptor, whereas no changes were observed for neurturin, persephin, and artemin (Hoke et al., 2000).
The above examples illustrate the intricate kinetics of mRNA expression involved in nerve regeneration, which is regulated through multiple feedback loops. The distinct pattern of growth factor synthesis underlines the importance of a well-defined timing of growth factor delivery from NC for optimal recovery. Although the complexity of molecular events following peripheral nerve regeneration is not yet fully elucidated, the existing knowledge of the underlying mechanisms of the cascade triggering nerve regeneration after injury provides important clues for designing growth factor delivery regimens. Past studies have clearly neglected the issue of growth factor release kinetics, which may be attributed, in part, to the limited control over growth factor release achievable with most of today’s delivery systems; indeed, most common and readily achievable growth factor release profiles consist of a continuous release over several days or weeks with a more or less pronounced initial burst. Yet, achievement of better functional recovery after peripheral nerve trauma will most probably require the optimization of growth factor delivery kinetics.
Future perspectives
The increasing understanding of the underlying mechanisms of peripheral nerve regeneration will allow scientists to devise more appropriate NC with integrated growth factor delivery systems and/or cellular components. Single-molded NC may not give sufficient control over both the mechanical properties and the delivery of bioactive agents. More complex devices will be needed, such as multilayered NC where growth factors are entrapped in polymer layers with varying physicochemical properties or tissue-engineered NC containing viable Schwann cells.
The combination of two or more growth factors will likely exert a synergistic effect on nerve regeneration, especially when the growth factors belong to different families and act via different mechanisms. Combinations of growth factors can be expected to enhance further nerve regeneration, particularly when each of them is delivered at individually tailored kinetics. The determination and control of suitable delivery kinetics for each of several growth factors will constitute a major hurdle both technically and biologically with the biological hurdle lying in the compliance with the naturally occurring cross talk between growth factors and cells. These difficulties together with regulatory issues (safety) and costs will probably strongly confine the number of concomitantly deliverable growth factors. A solution to this problem may be the use of Schwann cells because they can synthesize several growth factors, which is critical for very long gaps, as for instance in crossfacial grafting. The combination of Schwann cells with growth factors and their controlled delivery may further improve nerve regeneration. Such a system may be made from longitudinally aligned fibers that contain and deliver the growth factor(s) and act as support for Schwann cells. In addition, gradients of growth factors may be engineered along an NC to mimic the concept of neurotropism. Concentration gradients have already been implemented in gel matrices (Cao and Shoichet, 2003; Kapur and Shoichet, 2004); it will be very instructive to hear whether such a sophisticated system is advantageous in vivo to guide axons across long gaps.
