Controlled spatial and temporal release of neurotrophic growth factor from genetically modified tissue engineering living scaffolds

Abstract

Peripheral nerve injuries (PNIs) affect thousands of patients yearly, often resulting in loss of function, sensation, and chronic pain. In critical-size defects, advanced surgical repair strategies often fail to restore full function. A key limitation is the lack of sustained, localized delivery of biological cues for axonal regeneration, such as growth factors. Glial-cell line-derived neurotrophic factor (GDNF) is known to promote axonal growth, Schwann cell migration, and neuronal survival, but uncontrolled release may cause axonal entrapment. We previously developed tissue-engineered nerve grafts (TENGs) composed of two neuronal populations connected by stretch-grown axons. In this study, we genetically modified the distal population to express human GDNF under a Tet-on inducible promoter, temporally controlling GDNF release through doxycycline administration. Modified TENGs survived implantation in a 1.5-cm rat sciatic nerve defect, supporting future studies. This approach offers a promising platform for spatially and temporally controlled neurotrophic factor delivery from tissue-engineered living scaffolds.

Introduction

Peripheral nerve injuries (PNIs) affect roughly 50,000–100,000 patients that require surgical intervention in the United States and Europe annually.^1–3 PNIs are caused by trauma, the excision of nerves and surrounding tissues, and diseases, with extremity traumas being the primary etiology, accounting for 70%–80% of PNIs.^4,5 In cases involving higher degrees of PNI including axonotmesis (i.e. crush axonal injury that leaves surrounding endoneurium intact) and neurotmesis (i.e. complete transection of nerve and supporting tissue), spontaneous recovery is generally not observed. In these cases, surgical interventions are typically required, as untreated injuries can lead to impaired motor function or limb paralysis, sensory loss, and chronic neuropathic pain. PNIs associated with long segments of nerve loss cannot undergo tension-free end-to-end repair and require grafting procedures. Autologous nerve grafts (autografts) are currently the gold standard to treat critical and longer nerve gaps of more than 3 cm.^6,7 Nonetheless, autografts require the sacrifice of a healthy donor nerve and are consequently associated with donor site morbidity, increased operative time, and limited availability. Although alternative scaffolding options such as hollow nerve conduits and acellular nerve allografts from cadavers exist, these remain inferior compared to autografts.^7–9 Indeed, a meta-analysis encompassing 130 PNI clinical studies showed that only 42.6% of patients achieved satisfactory outcome in sensory recovery and only 51.6% of patients achieved satisfactory motor outcome. ¹⁰ Thus, despite decades of advances in microsurgical techniques and nerve repair methods, functional recovery is generally not achieved.

A major barrier to robust functional recovery is insufficient biologically active cues within a physical bridge across a nerve gap. For long gap nerve injuries, such biologically active cues are necessary to drive and sustain axonal re-growth and Schwann cell infiltration across the length of the lesion. Furthermore, axonal growth from the proximal and into the distal stump needs to happen as quickly as possible since axons also need to traverse further down the distal nerve to reach their end target muscle, before the target muscle undergoes irreversible atrophy. Therefore, supplemental active cues are needed to support long-term axonal regeneration. Widespread evidence has shown that neurotrophic growth factors, such as glial-cell line-derived neurotrophic factor (GDNF), promote axonal extension, induce Schwann Cell (SC) migration, and increase motor and sensory neuron survival.^11–16 GDNF is usually upregulated after PNI and acts to enhance functional nerve regeneration, but unfortunately, endogenous GDNF levels decrease after prolonged periods of denervation.^17,18 Previous studies have exogenously delivered GDNF at the site of a nerve defect to accelerate host axonal growth. ¹⁹ Interestingly, exogenous, constitutive GDNF release has often led to deleterious effects due to excessive axonal coiling and entrapment at the site of GDNF release, known as the “candy store” effect.^20–22 For instance, in vivo rat model studies have demonstrated that 4–6 weeks of GDNF release from the distal nerve led to improved muscle recovery and neuronal survival, whereas uncontrolled GDNF expression led to the failure of axonal regeneration beyond the source of GDNF, impairing recovery.^23–25 Accordingly, the spatial and temporal distribution of GDNF must be controlled and optimized to accelerate axonal growth and regeneration.

Our group has previously developed tissue engineered nerve grafts (TENG) as implantable living scaffolds to bridge major peripheral nerve defects.^26–30 TENGs are comprised of living longitudinally aligned axonal bundles that span two parallel populations of dorsal root ganglia (DRG). The long axonal tracts are generated via so-called “stretch-growth” in custom mechanobioreactors.^28,30,31 Stretched to desired lengths in vitro, the parallel axon tracts and the DRG populations are encapsulated in collagen to create living 3D scaffolds in a columnar form.^{26–28,30,32} In vivo studies have shown that TENGs are able to recruit host axons and SCs into the nerve defect zone,^26,30,31 and can also preserve the regenerative capability of the distal tissues .^26,30 TENGs implanted to treat 1-cm defects in rats for 16 weeks and 5 cm defects in pigs for 9–11 months performed comparably to autografts based on functional restoration and axonal regeneration.^26,27,30

Building on this previous body of work, in the current study, 1.5–5 cm living TENGs were genetically modified to express human GDNF (GDNF) using Adeno-associated virus serotype 2 (AAV2), referred to as “AAV-TENGs.” To control the timing of GDNF release, we utilized tetracycline inducible promoter (TRE or Tet-on), which can be turned on with the addition of Doxycycline (Dox), a member of the tetracycline family of antibiotics that is well tolerated clinically.^33–36 In this study, we characterized the spatial and temporal regulation of the neurotrophic growth factor expression along the length of the AAV-TENGs in vitro. We also performed a pilot implantation study to assess in vivo survival of AAV-TENG neurons and axon tracts when used to bridge a critical size 1.5 cm sciatic nerve gap in a rat model.

Methods

Dorsal root ganglia harvest

Dorsal root ganglion (DRG) explants were obtained from embryonic Day 16 Sprague Dawley rats as previously described. ³⁷ Pregnant dams were euthanized via CO₂ asphyxiation followed by decapitation in accordance with the AVMA Guidelines for the Euthanasia of Animals. Pups were removed and kept in L-15 media (Life Technologies) on a cold block. Using micro-forceps, spinal cords were harvested. DRG explants were then plucked out and temporarily placed in cold Hanks Balanced Salt Solution (HBSS, Life Technologies). For longer term storage (>1 h), DRG explants were placed in Hibernate-E media supplemented with B27, Penicillin/Streptomycin, and Glutamax (ThermoFisher). As these are primary embryonic tissues directly isolated without passaging or genetic manipulation, no additional characterization (e.g. karyotyping, morphology screening, or mycoplasma testing) was performed. The use of freshly isolated, unmodified DRG has been extensively established in the field as a robust and reliable source of sensory neurons for culture studies.^38,39

Dorsal root ganglia culture

Before adding cells, tissue culture plate surfaces were treated either with 20 µg/mL poly-D-lysine (PDL; BD Biosciences) diluted in cell culture grade water (Lonza) overnight or with 500 µg/mL polyethyleneimine (PEI; ThermoFisher) diluted in cell culture grade water for 45 min. After three rinses with PBS, 20 µg/mL laminin (BD Biosciences) was added to the plate surface, followed by a 2-h incubation period. Laminin was removed and DRG media was added. After the isolation steps described above, DRG were plated in the tissue culture plates. The media for this study consisted of Neurobasal medium (ThermoFisher) supplemented with 2% B-27 (ThermoFisher), 1% fetal bovine serum (FBS, Atlanta Biologicals), 0.5-mM Glutamax (Gibco), 20-ng/mL nerve growth factor (BD Biosciences), 2.5-g/L glucose (Sigma), 0.1% penicillin-streptomycin (ThermoFisher), 20 mM 5FdU (Sigma-Aldrich), and 20 mM uridine (Sigma-Aldrich).

Adeno-associated virus-dorsal root ganglia transduction

AAV2 transduction was performed using commercially available viral vectors at the optimized concentrations as outlined in Table 1. DRG used to verify inducible GDNF expression were placed in culture media containing the vectors for 1–3 days at 37°C. DRG used for the fabrication of AAV-TENGs were placed in serum-free, vector-containing previously described hibernate media for 2 days at 4°C. After 2 days, cells were rinsed three times with hibernate media, transferred to regular culture media, and plated in custom mechanobioreactors as described below.

Table 1.

Vectors used to transduce DRG for planar plating and TENG fabrication. Viral vectors, their corresponding manufacturer, and chosen transduction concentrations.

Viral vectors	Manufacturer	Viral concentrations (vp/DRG)
AAV2-CMV-eGFP	Addgene	3.13 × 109
AAV2-CMV-mCherry	Vector Builder	9.39 × 109 for 1.5-cm TENGs 3.13 × 109 for 2.5-and 5.0-cm TENGs
AAV2-CMV-hGDNF	Vector Builder	3.13 × 109
AAV2-TRE-hGDNF	Vector Builder	1.41 × 1010

Adeno-associated virus-tissue engineered nerve graft fabrication and culture

After AAV transduction of DRG, AAV-DRG were washed with hibernation media three times. AAV-DRG were plated onto custom-built mechanobioreactors to induce stretch growth as previously described. ³⁷ The mechanobioreactors are comprised of two overlapping 33C Aclar membrane films. One of the two membranes (base membrane) remains stationary, while the other membrane (towing membrane) is incrementally pulled by a stepper motor, inducing stretch growth. Aclar membranes were pre-soaked in 1M NaOH (ThermoFisher) for 24 hrs to increase hydrophilicity of the surface, washed twice with deionized water, and coated with PDL or PEI and laminin as mentioned above. AAV-DRG expressing one gene were plated parallel to the edge on the towing membrane, followed by the placement of an AAV-DRG population expressing a different gene on the base membrane, parallel to the towing membrane population. AAV-DRG are initially plated to have a 1–2 mm gap between the two populations with around 4–8 AAV-DRG on each membrane. AAV-DRG were cultured without stretch to allow for neurite network formation for 5–6 days in culture. On the fifth to sixth day, the stepper motor was then engaged to move the towing membrane in micron-sized steps until the two populations of AAV-DRG were separated by a pre-specified distance using a custom-made software (1.5, 2.5, and 5 cm; Figure 1). The stretched culture was coated with a layer of 2.5 mg/mL followed by a layer of 3.0 mg/mL rat tail collagen type 1 (BD Biosciences) in 4× MEM, 100 mM HEPES, and 8.8 mg/mL NaHCO3. The collagen/DRG construct was then removed from the mechanobioreactor resulting in AAV-TENG constructs. AAV-TENGs are labeled as described in Table 2. Conditioned media was collected throughout the culture period for up to 14 days. GDNF ELISA was performed on the condition media per manufacturer’s protocol (R&D Biosystems).

Figure 1.

AAV-TENG Fabrication. (a-d) AAV-TENGs were fabricated from two differentially transduced populations of AAV-DRG that were placed on parallel membranes. (e-f) AAV-TENG axons were “stretch-grown” and expressed different transgenes on each side.

Table 2.

AAV-TENG identification table.

AAV-TENG ID		Towing membrane (left)	Base membrane (right)	Length (cm)
1	CMV-eGFP-mCherry	AAV-CMV-eGFP	AAV-CMV-mCherry	2.5 and 5
2	CMV-eGFP-hGDNF	AAV-CMV-eGFP	AAV-CMV-hGDNF	1.5 and 5
3	CMV-mCherry-TRE-hGDNF	AAV-CMV-mCherry	AAV-TRE-hGDNF	1.5

Visualization and quantification of differential transgene expression

CMV-eGFP-hGDNF AAV-TENGs were fabricated to express GFP in the towing membrane and GDNF in the neurons on the base membrane following the previously described methods. CMV-eGFP-mCherry AAV-TENGs were fabricated to express mCherry in neurons on the towing membrane and GFP in neurons on the base membrane following the previously described methods. After encapsulation, CMV-eGFP-hGDNF AAV-TENGs were divided in halves and cultured in two separate dishes for up to 14 days. Conditioned media samples of each culture were obtained at time of plating, twice during stretch growth, during culture after encapsulation, and before fixing. GDNF ELISA was performed per manufacturer’s protocol (R&D Biosystems).

Sustained Dox induction of adeno-associated virus-tissue engineered nerve graft transgene

AAV-TENGs 1.5 cm in length were stretch-grown with constitutively active mCherry expression on one side and inducible GDNF expression on the other side (CMV-mCherry-TRE-hGDNF). They were encapsulated in collagen, as previously mentioned and cultured for up to 14 days. Two Dox pulses were performed in culture. For each pulse, Dox (2 µg/mL) was added to the culture media of CMV-mCherry-TRE-hGDNF TENGS for 3 days at days 4 and 13. These TENGs were then washed with PBS three times and subsequently cultured in Dox free culture media for 4 days. Four days following the second Dox pulse, the AAV-TENGs were divided into halves and separately cultured in Dox containing media for 3 days. The conditioned media was collected throughout the culture period, and GDNF ELISA was performed per manufacturer’s protocol (R&D Biosystems).

Dox control of adeno-associated virus-tissue engineered nerve graft transgene expression

AAV-TENGs of 1.5 cm length were stretch-grown with constitutively active mCherry expression on one end and inducible GDNF expression on the other end (CMV-mCherry-TRE-hGDNF). 2 ng/mL Dox was added 2 days before the AAV-TENGs completed stretching, as a prime step. After stretching was completed, CMV-mCherry-TRE-hGDNF TENGs were cultured in Dox-free media for 2 days, while in the mechanobioreactor. They were then removed and encapsulated in collagen and cultured in Dox-media for 4 days, after which they were washed with PBS three times and cultured in regular Dox-free media for 16 more days. Conditioned media was collected and media was renewed every 3–4 days. At the end of the culture period, the CMV-mCherry-TRE-hGDNF TENGs were divided in half and placed in separate dishes containing Dox-media for 4 days. Conditioned media was collected, and ELISA was performed to quantify GDNF concentration per manufacturer’s protocol.

Immunocytochemistry

After culture, DRG and AAV-TENGs were fixed with 4% paraformaldehyde for 30 min. Cultures were then rinsed three times in 1× PBS and permeabilized with 0.3% Triton X-100 + 4% horse serum in PBS for 60 min. Cultures were then incubated with corresponding primary antibodies overnight at 4°C (1:500 GFP and mCherry, 1:100 GDNF), rinsed with 1× PBS and incubated with 1:1000 corresponding secondary antibodies for 2 h (488 for GFP, 587 for mCherry, and 647 for GDNF) for visualization.

In vivo surgery and implantation

All animal protocols used in this study were reviewed and approved by the Institutional Animal Care and Use Committees at the University of Pennsylvania and the Corporal Michael Crescenz VA Medical Center. All animals utilized in this study were managed in accordance to the Guide for Care and Use of Laboratory Animals and ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments). ⁴⁰ AAV-TENGs were fabricated to express GFP in the neurons on the towing membrane and GDNF in neurons on the base membrane following the previously described methods (CMV-eGFP-hGDNF). After encapsulation, AAV-TENGs were placed in 2.0 mm × 25 mm NeuroMend^TM wraps (Stryker). Wraps were then sutured shut to ensure proper TENG placement. Three adult male Sprague-Dawley rats were used as experimental subjects. Rats were anesthetized using inhaled isoflurane at 1.0–2.0 L/min (5% induction and 2.5% maintenance). The rat sciatic nerve was exposed as previously described.^27,31 Briefly, the rats’ hindlimb was shaved and a skin incision was made along the posterior thigh. A blunt dissection was made through the underlying musculature and the biceps femoris was then retracted to visualize the sciatic nerve. A 1.5 cm segment was excised and AAV-TENGs in Neuromend wraps were implanted. TENGs were implanted by inserting 1 mm of distal and proximal nerve stumps inside the wraps and sutured to the epinerium using 8-0 non-absorbable sutures. The surgical site was closed using 4-0 prolene sutures.

Nerve harvest and histology

At time of harvest, rats were overdosed with CO₂ for 8 min. The entire length of the repaired nerves, including the implanted graft, was harvested and placed in 4% paraformaldehyde for 48 h at 4°C. The tissue was immersed in 30% sucrose solution for 48 h or until fully saturated. Nerve tissue was blocked and embedded in OCT (Tissue Tek) and frozen at −80℃. Axial sections were cut with a Microm HM 550 cryotome (ThermoFisher) at 20 μm thickness, mounted on Superfrost glass slides (Fisher), and rehydrated with PBS. Sections spanning the entire nerve at approximately 160 µm intervals were labeled with the following primary antibodies diluted 1:500 in PBS containing 0.3% Triton X-100 (Sigma-Aldrich) and 4% NHS: GFP (Abcam); SMI31+32 (phosphorylated and nonphosphorylated neurofilaments, BioLegend); and GDNF (Abcam). After extensive rinsing, sections were then incubated in Alexafluor-conjugated secondary antibody solutions (ThermoFisher; 1:500 in PBS) for 2 h at room temperature. To preserve fluorescent signal, stained samples were treated with Fluormount-G (ThermoFisher) prior to coverslipping.

Microscopy and image analysis

Phase-contrast and brightfield images of AAV-TENGs were taken using a Nikon Eclipse Ti-S inverted microscope and Keyence BZ-X810 fluorescent microscope. Fluorescent images of AAV-TENGs were also captured with the Keyence and a Nikon A1RSI Laser Scanning Confocal microscope. All image analysis was performed using ImageJ.

Statistical report and analysis

All statistical analysis was performed on GraphPad Prism. Overall, normality tests were conducted via the visualization of residual histogram, residual quantile plots, and Shapiro-Wilk’s test (in which p > 0.05). Post-hoc Tukey-Kramer tests were conducted to compare means within each factor. Nonparametric tests including one-tailed Wilcoxon Signed Rank Test and Welch’s test were performed on data that did not pass normality tests. A Grubb’s outlier test was performed to remove any potential outliers from the data. The significance level was 0.05 in all analysis. Data in figures are represented by the mean ± 1 standard error.

Results

Spatial control of transgene expression along adeno-associated virus-tissue engineered nerve grafts

The spatial control of transgene expression along the length of AAV-TENGs was studied using CMV-eGFP-mCherry TENGs (Figure 2(a)). TENGs were transduced with AAV2-CMV-mCherry in one DRG population (red; right) and with AAV2-CMV-eGFP in the opposing population (green; left). The DRG neuronal clusters expressed the highest level of either GFP or mCherry compared to the axonal component of the CMV-eGFP-mCherry. As expected, the AAV-GFP transduced DRG end expressed the highest GFP signals and AAV-mCherry transduced DRG end expressed the highest red fluorescent signals (Figure 2(a)). Both GFP and mCherry signals were observed throughout the axons as well (Figure 2(b) and (c)). We also found that the longer 5 cm CMV-mCherry-eGFP TENGs also co-express GFP and mCherry in a similar manner (Figure 2(d)). DRG neuronal clusters express the highest level of either GFP or mCherry while the signals seem to decrease within the axons in a nonlinear pattern (Figure 2(e) and (f)). Low levels of GFP were observed at the right end of the AAV-TENG, and similarly, mCherry signal is also observed at the left side of the AAV-TENG.

Figure 2.

Representative AAV TENGs with Differential Expression of GFP and mCherry. (a) 2.5 cm CMV-eGFPmCherry TENG containing five mCherry-expressing DRG at the right side (base membrane) and four GFP expressing DRG aggregates at the left side (towing membrane). Scale: 1000 μm. (b) Magnified (40×) view of the GFP+ axon tracts near the DRG aggregates. Scale: 1000 μm. (c) Magnified (40×) view of the mCherry+ axon tracts near the DRG aggregates. (d) 5 cm CMV-eGFP-mCherry TENG: GFP and mCherry signal observed throughout the axonal tracts. Scale: 2000 μm. (e) Magnified view (40×) of the predominantly GFP+ axon tracts near the left side. (f) Magnified view of the predominantly mCherry+ axon tracts near the right side. Scale: 2000 μm.

Spatial Control of Glial Derived Neurotrophic Factor Release From the Adeno-Associated Virus-Tissue Engineered Nerve Graft (CMV-eGFP-hGDNF)

As described above, the expression patterns in GFP-mCherry TENGs suggest that differential AAV-DRG populations at each end of the TENGs could establish gradient profiles based on transgene expression of soluble factors across the graft. Accordingly, we developed 1.5 and 5 cm-long AAV-TENGs with an AAV-CMV-hGDNF transduced DRG population on one end (base membrane) and AAV-CMV-GFP transduced DRG population at the other end (towing membrane) (CMV-eGFP-hGDNF). The CMV-eGFP-hGDNF TENGs were then cut in half, and each half was separately cultured in regular media. The levels of GDNF present in the conditioned media were collected at 1 and 3 days after the TENGs were separated (Figure 3(a)). At both 1 and 3 days after separation, significantly higher GDNF levels were detected from the GDNF side of the TENG (Figure 3(d); Day 1: GDNF: 104.358 ± 24.885 pg/mL, GFP: 2.222 ± 8.964 pg/mL, p = 0.0079, n = 5; Day 3 GDNF: 257.777 ± 78.822 pg/mL, GFP: 3.898 ± 15.519 pg/mL, p ⩽ 0.0079, n = 5; Unpaired, Non-parametric t-test). GFP expression was observed in the GFP side of the TENG at the end of the 3 days, expressed by the DRG clusters and the axons, further illustrating spatial control of transgene expression and suggesting maintained viability of the construct (Figure 3(c) and (e)).

Figure 3.

Fabrication of AAV-TENG With Differential Expression of GFP and GDNF: (a) schematic of in vitro characterization of CMV-eGFP-hGDNF TENG post-fabrication. (b) Schematic of GDNF release from AAV-GDNF TENGs. (c) Representative overlay image of each side of the 1.5 cm AAV-TENG, with the GFP-DRG population at the left and the GDNF-DRG population at the right. Scale: 1 cm (d) Significantly higher levels of GDNF concentration were detected in conditioned media from the GDNF side compared to that of the GFP side days 1 and 3 after population separation than that of the GFP side. (p ⩽ 0.0079, n = 5) (e) GFP intensity of each side of the TENG in c, with more GFP present throughout the left side of the axons than that of the right.

To better understand GDNF release post TENG fabrication, 1.5 cm CMV-eGFP-hGDNF were stretch grown, fixed, and stained for anti-GDNF antibodies. We visualized both GFP and GDNF in the TENG aggregates and axons (Figure 4(a)). Statistically significant GFP concentration in the CMV-eGFP transduced side was detected compared to the CMV-hGDNF side (Figure 4(b)). Next, fluorescent signal of eGFP and GDNF were analyzed in differentially transduced populations, as described above. Interestingly, similar levels of hGDNF fluorescent signal were observed on both sides of AAV-TENG DRG’s (Figure 4(c)). To ensure successful GNDF expression, we performed GNDF ELISAs on cut TENGs as described above. GDNF transduction was confirmed by a higher detection of GDNF in the CMV-GDNF side compared to the eGFP side. This suggests that increased GDNF detected in ELISA in Figures 3(d) and 4(d) is secreted and diffuses away from the neuronal somata, rather than being stored in the cells or binding to local extracellular matrix postproduction. This may also be due to the fact that immunostaining is not sensitive enough to detect GDNF level differences observed via ELISA.

Figure 4.

Visual and molecular confirmation of CMV-eGFP-hGDNF TENG GDNF release. (a) 4× images of 1.5 cm CMV-eGFP-hGDNF TENGs stained for GDNF, GFP, Tuj1, and Hoechst. Scale: 1000 μm (b) Representative magnified (10×) images of individual aggregates of 1.5 cm CMV-eGFP-hGDNF TENGs stained for GFP, GDNF, and Hoechst. Left image correlates to a representative aggregate in GFP (towing) membrane, while the right correlates to a representative aggregate in GDNF (base) membrane. Scale: 100 μm (c) Relative fluorescence unit quantification of GFP and GDNF signal normalized to aggregate number per TENG (n = 3 TENGs). (d) Higher GDNF levels were detected from the GDNF side of the CMV-hGDNF TENG compared to that of the GFP side in culture. Concentration was normalized to DRG count on each side (n=3 TENGs).

Temporal Control of Glial Derived Neurotrophic Factor Release From Adeno-Associated Virus-Tissue Engineered Nerve Graft Under Dox Regulation (CMV-mCherry-TRE-hGDNF)

GDNF levels detected from AAV-TRE-GDNF DRG cultures swiftly decreased in the absence of Dox until 13 days post-AAV-transduction, at which point GDNF was nondetectable (Figure 5(a); Day 4: 108 ± 15.43 pg/mL and Day 7: 17.61 ± 6.425 pg/mL; p = 0.0079; two-tailed unpaired t-test; n = 4). DRG neurons were transduced with AAV-TRE-GDNF for 3 days and cultured in media with or without Dox for 2 days. Conditioned media collected from the Dox+ group expressed significantly higher GDNF than that from the Dox− group (Figure 5(b); Dox+: 1958 ± 637 pg/mL and DOX-: 25.8 ± 29.49 pg/mL; p = 0.0387, n = 3; Two-Tailed unpaired t-test). AAV-TRE-hGDNF transduced DRG cultured in Dox− media had low but detectable levels of GDNF, indicating initial low-grade “leakage” of the AAV-TRE-vectors or low level GDNF release from non-transduced cells in DRG (note the different y-axis scale).

Figure 5.

Temporal Control of GDNF Release From AAV-TENGs Under Dox Regulation (a). AAV-TRE-GDNF transduced DRG expressed GDNF in the absence of Dox for the first week after transduction (n = 5). Significantly lower levels of GDNF were obtained 7 days after transduction and were negligible after (p = 0.0079; n = 4) (b) DRG transduced with AAV-TRE-GDNF released significantly higher GDNF levels in the presence of Dox. (p = 0.0387, n = 3). (c) AAV-TRE-GDNF-transduced DRG pulsed with Dox twice (yellow) demonstrated sustained GDNF release for 22 days (n = 5). (d) Detected GDNF levels from each side of CMVmCherry-TRE-hGDNF TENG after Dox was reintroduced for 3 days (n = 4 CMV-mCherry, n = 2 TRE-hGDNF). (e) Significantly more GDNF was released from the GDNF side than that of the mCherry side (p < 0.0286, n = 4).

The Dox-responses of the AAV-TRE-GDNF transduced DRG were then investigated. After the first Dox pulse, DRG expressed higher levels of GDNF compared to expression levels before the pulse. The expression continued to decrease until a second pulse was added, and GDNF levels increased again. After the increase, GDNF levels continued to decrease until 22 days post-transduction (Figure 5(c)). AAV-TENGs with AAV-CMV-mCherry+ DRG on one side and AAV-CMV-hGDNF+ DRG (CMV-mCherry-hGDNF) or AAV-TRE-hGDNF+ DRG (CMV-mCherry-TRE-hGDNF) on the other side were fabricated to 1.5 cm length over the course of 2–3 weeks. After stretch-growth, TENGs were cultured under two Dox pulses (Figure 5(d)). Before any addition of Dox, no GDNF levels were detectable in the CMV-mCherry-TRE-hGDNF TENG cultures. Interestingly, only low levels of GDNF were released after the first Dox pulse (Day 5), suggesting a lagging period of GDNF expression in response to Dox. However, higher GDNF expression was detected after the second Dox stimulation compared to the GDNF levels detected in the previous cycle (Figure 5(d)). The data suggests that a separate priming Dox stimulation may be necessary before Dox-responsive transgene upregulation can occur in TENGs.

After the two Dox cycles, AAV-TENGs were divided in half, and each half of the AAV-TENGs was separately cultured in Dox+ media. Higher GDNF concentration was detected from the conditioned media cultured with the half containing the AAV-TRE-GDNF transduced DRG compared to that of the AAV-CMV-mCherry transduced DRG (Figure 5(e); GDNF side: 229 ± 150.3 pg/mL, mCherry Side: 23.67 ± 8.749 pg/mL, p < 0.0286, n = 4; Un-paired, non-parametric t-test).

The CMV-mCherry-TRE-hGDNF TENG fabrication process was modified to include a “priming” step, which involved the addition of Dox before AAV-TENGs were completely stretched to 1.5 cm (Figure 6(a)). Little to no GDNF was detected immediately after the priming step. When Dox was added to the fully stretched CMV-mCherry-TRE-hGDNF TENGs, increased GDNF levels were detected. GDNF levels decreased 6 days after the removal of Dox and became nondetectable for the rest of the 10 days (Figure 6(b); Day 4: 223.6 ± 175.8 and Day 10: 19.54 ± 19.02; p = 0.0286; Two-tailed, unpaired t-test; n = 2 before day 0, n = 4 at and after day 4). The CMV-mCherry-TRE-hGDNF TENGs were then divided in half and each half was separately cultured in Dox+ media for 4 days. GDNF was detected from the side containing AAV-TRE-GDNF-transduced DRG as opposed to the side containing AAV-CMV-mCherry transduced DRG (Figure 6(c)). TRE-GDNF AAV-TENGs released GDNF in a spatially and temporally controlled manner under Dox stimulation.

Figure 6.

GDNF Release After Dox Priming: (a) Schematic of the Dox priming step in the fabrication protocol for TRE-GDNF TENGs. (b) GDNF levels during and after the fabrication (priming included) of TRE-GDNF TENGs (n = 4). After priming, GDNF levels increased in the presence of Dox (yellow-Day 0–4) and decreased rapidly upon Dox removal (Day 4–10; n = 4). Significantly lower levels of GDNF were detected after Dox removal (p = 0.0286; n = 4). (c) GDNF levels were detected from the GDNF side of the primed TRE-hGDNF TENG compared to that of the mCherry side in Dox-present culture (n = 4).

Following in vitro characterization of differential transgene expression in CMV-eGFP-hGDNF TENGs, we next evaluated construct viability in vivo using a 1.5 cm rat sciatic nerve transection model. The sciatic nerve was transected, and a CMV-eGFP-hGDNF TENG was implanted with the GFP-transduced DRG population oriented proximally and the GDNF-transduced population distally (Figure 7(b)). At 24 weeks post-implantation, robust cell survival was observed in both graft regions. Notably, stronger GFP fluorescence was detected in the proximal region, consistent with the orientation of the GFP-labeled DRG population (Figure 7(c)). These findings demonstrate successful implantation and survival of CMV-eGFP-hGDNF TENGs in vivo, supporting the feasibility of spatially controlled transgene delivery for future studies focused on growth factor gradients.

Figure 7.

In vivo cell survival after implantation of CMV-eGFP-hGDNF TENG into a 1.5 cm gap in a sciatic nerve defect in rats after 24 weeks. (a) Schematic of sciatic nerve defect and CMV-eGFP-hGDNF TENG implantation. (b-c) Proximal and distal regions of TENG depicting the eGFP side of the TENG (b) and the hGDNF side of the TENG (c). Scale: 100 μm.

Discussion

A major barrier to effective nerve regeneration is the lack of sustained biological cues necessary for expeditious axonal growth across critical-size defects. TENGs have emerged as a promising alternative to autografts and allografts. As mentioned above, TENGs are fabricated by stretch growing the axons that span two populations of DRG plated in custom mechanobioreactors. We previously showed that TENGs serve as a cell-based bridge that encourages host axon growth and Schwann Cell infiltration in vivo. ²⁷ However, the addition of controlled gradients of growth factor across the TENGs may serve as a novel drug delivery platform for sustained and controlled neurotrophic growth factor release.

Accordingly, in the current study we genetically modified TENGs to allow for spatially and temporally controlled growth factor gradients to supplement direct axon-mediated axon regeneration presented by TENGs. In this study, we achieved differential gene transduction as seen by successful spatial and temporal release of GDNF in 1.5–5 cm TENGs. This combined cell-engineering and tissue-engineering strategy may accelerate axonal regeneration and functional recovery in critical-size peripheral nerve defects of five or more centimeters.

GDNF plays a critical role in promoting motoneuron survival,^12,15 axonal growth, ¹¹ myelination,^13,41 and remodeling of neuromuscular junctions.^16,42,43 While GDNF is upregulated and expressed by SCs, neuronal subpopulations, ⁴⁴ and skeletal muscles following PNI,^45–48 endogenous GDNF levels decrease upon persistent de-innervation and degeneration of target issues. ¹⁸ Numerous studies have exogenously introduced growth factors into the nerve stump via a variety of drug delivery systems, including hydrogel release,^49–53 chemically-engineered scaffolds,^54,55 direct viral transduction of tissues,^21,22,56 or a combination of these approaches.^24,57–59

In this study, AAV2 was used to induce expression of neurotrophic growth factor from our well-characterized TENGs. AAV2 has proven to be safe against immune response, is effective in enhancing bioavailability, and has been widely used in clinical trials^60–63 As a proof-of-concept, TENGs were first engineered to express GFP on one end and mCherry on the other end using the protocol described in Figure 1. CMV-mCherry-eGFP TENGs expressed GFP and mCherry gradients throughout the axons (Figure 2). Low levels of GFP were observed at the right end of the AAV-TENG, and similarly, mCherry signal is also observed at the left side of the AAV-TENG. This may be due to fluorescent expression in axons projecting from the opposing population, but also from the migration of AAV-transduced neurons or glial cells from one side to another at the beginning of the fabrication step, before the two distinct DRG populations were stretched apart from each other (between steps C and D in Figure 1). Next, CMV-eGFP-hGDNF TENGs were engineered to constitutively release hGDNF from the right side only, while eGFP was expressed preferentially on the left side (Figure 3). In previous studies, GDNF has been observed to undergo anterograde transport up to 17 cm/day,^43,64 and thus a local GDNF gradient is likely established within the axonal level as soon as GDNF is expressed in DRG. At the macroscopic level, GDNF is first released by cells and diffuses into collagen surrounding the cells within the nerve graft. Entrapped GDNF is subsequently released into the environment from the collagen, establishing a bulk GDNF release (Figure 3(b)). The spatial control of GDNF release from one population of DRG creates a slow gradient of GDNF release as the molecule diffuses across the TENG collagen. The potential presence of GDNF in collagen likely explains the statistically significant low GDNF concentration detected from the GFP side of the CMV-eGFP-mCherry TENG (Figure 3(d); p = 0.0079). Similarly, CMV-eGFP-hGDNF TENGs were engineered and stained to visually detect the differential expression of GDNF and GFP (Figure 4). Despite ELISA confirmation of differential GDNF expression, GDNF signal intensity was similar in all cell populations. As mentioned above, this may be due to endogenous glial cell release, protein transport across the axons, or insufficient staining sensitivity to detect pico-level changes in GDNF.

While exogenous levels of GDNF can accelerate axonal elongation of the host nerve, ⁵³ prolonged GDNF release led to adverse effects. Specifically, sustained concentrations of GDNF resulted in axonal trapping and coils at the site of GDNF release negatively affecting recovery.^21,45,56 When GDNF is released at the proximal side of the nerve stump, axonal extension was not observed ⁴³ and recovery was not achieved. ²¹ This is referred to as the candy store effect, in which axons are content and remain at the site of GDNF overexpression and hence fail to cross the nerve gap to reinnervate target organs. ²¹ Our spatial gradient design directly addresses this phenomenon by creating differential zones of GDNF expression, thereby encouraging elongation toward the distal stump rather than trapping axons mid-graft. Importantly, the dose, timing, and the location of GDNF must be regulated. After host axons have infiltrated the nerve stump as a response to exogenous GDNF, GDNF must be turned off to perpetuate axonal regeneration and re-innervation toward the distal organ.

An important design consideration was temporal ON/OFF control of GDNF release. We transduced DRG with AAV-TRE-GDNF to create this ON/OFF control. Doxycycline (Dox) was chosen as the TRE trans-activator to control GDNF release. Dox is currently used as an FDA approved antibiotic and acne treatment and has been found to be highly biocompatible, with limited data suggesting severe side effects. ⁶⁵ As well, previous studies have found that Dox on its own may mimic neuron growth factor PI3K/Akt signaling pathways by activating trkA receptors, inducing neuritogenesis, thus showing promise for use in neural injury repair. ⁶⁶ The beneficial effects of Dox have been shown in tissue engineering applications of other musculoskeletal systems, such as articular cartilage. ³⁵ Future in vivo studies will include a Doxycycline-only group of animals to control for any potential confounding effects. We found that DRG transduced with AAV-TRE-GDNF all displayed basal expression of GDNF in the first week in the absence of Dox. Nevertheless, GDNF levels drastically decreased and were not detectable by ELISA 13 days after transduction (Figure 5(a); p = 0.0079). For the CMV-mCherry-TRE-hGDNF TENG cultures, there was no GDNF detected in the absence of Dox (Figure 5(b); p = 0.0387). This was expected because the TENG stretching period, which occurred after DRG were transduced, spanned 14 days, which was longer than the leakage period determined previously. The slower rate of inducibility of Tet-on transgenes has also been observed in other studies.^25,67–70 Prior to initial Dox stimulation, there would be little to no trans-activator present, but with the second Dox pulse, it can induce sufficient trans-activators to successfully turn on transgene expression. ⁶⁸ The priming step incorporated here provides a practical strategy to overcome this limitation, distinguishing our system from prior inducible delivery approaches.^71,72

For clinical use, after the implantation of TENGs, it is crucial for GDNF treatment to begin as soon as possible, as delayed treatments could significantly impair recovery, with 1–2 week delays leading to less-than-optimal outcomes.^73–75 A Dox pulse was incorporated into the fabrication protocol of TRE-hGDNF TENGs (“prime” step), so that an additional Dox pulse can immediately drive robust GDNF expression. Indeed, after the CMV-mCherry-TRE-hGDNF TENG was primed, the Dox pulse successfully induced robust GDNF release from the GDNF side within 4 days. Furthermore, GDNF levels decreased and became nondetectable 7 days after the removal of Dox (Figure 6(b); Day 10: p = 0.0286). The lag period that occurred between the removal of Dox and the complete cessation of GDNF is likely dictated by the half-life of Dox, TRE trans-activator, GDNF mRNA, and GDNF protein levels (intracellular and entrapped in collagen).^68,76 Specifically, this lag serves as a buffer against fluctuations of Dox in vivo, as Dox levels in blood can fluctuate and vary from patient to patient due to physiological differences affecting administration, distribution, metabolism, and excretion (ADME) routes.^70,76,77 Previous studies using the Tet-on system have reported basal levels of GDNF without the presence of Dox,^24,69,78 and even low levels of GDNF released from the grafts still resulted in some degree of axonal coil and entrapment.^21,22,24 Notably, TRE-TENGs did not exhibit GDNF leakage due to the tet-on promoter. Because of this, TENGs provide on-off control of GDNF release without the potential side effects of basal levels of GDNF.

In vivo, we anticipate that AAV-TENGs will be able to establish a GDNF gradient in the nerve gap, with higher GDNF levels being expressed and released in the distal end of the stump, which will diffuse into the proximal end (GFP side) over time. Future studies will include an immunohistological examination of GDNF protein in the distal stump compared to the proximal stump to confirm the formation of an in vivo gradient. Under chemotaxis, host axons can infiltrate the TENG starting from the low GDNF proximal end and quickly reach the GDNF-rich end. The GFP+ DRG on the proximal side of the TENG can also be replaced by GDNF-expressing DRG transduced in lower doses of AAV-GDNF, allowing quicker infiltration of host axons into the TENGs. Furthermore, previous studies have suggested that different neurotrophic factors can produce various and sometimes synergistic effects on axonal regeneration.^79,80 Each side of the TENG can be engineered to express various neurotrophic factors; for example, with NGF being expressed on the proximal side to promote axonal infiltration into the stump and GDNF on the distal side to promote axonal elongation toward the distal end. With spatial regulation and design, AAV-TENGs serve as a multifunctional bioactive scaffold capable of expressing multiple therapeutic genes to induce optimal recovery. This modularity extends findings from prior studies showing distinct roles of NGF, BDNF, and NT-3, by offering a single platform capable of multiplexed spatial control.^15,79

Given that AAV-TENGs show promise in delivering growth factors in a spatially and temporally controllable manner, several considerations will need to be made when implanting these grafts in vivo such as immune response and biocompatibility of the graft. Preliminary data showed that GDNF-TENGs survived in vivo for up to 24 weeks (Figure 7). No notable side effects were observed in any of the implanted animals. It has also been previously shown in non-neuronal models that GDNF may suppress inflammatory cytokine production. ⁸¹ Similarly, previous studies implanting DRG into peripheral nerve injury models showed a decrease in macrophage infiltration when implanting allogeneic DRG only allografts compared to other cell types. ⁸² These findings support the potential of AAV-TENGs as biocompatible scaffolds for peripheral nerve regeneration. However, for eventual clinical use, patients receiving allogeneic TENGs will likely need short-term immunosuppression – at least for the time period over which host axons are regenerating across the graft – to avoid graft rejection.

Upon implantation, patients can start taking Dox to elicit GDNF release at the distal side nerve stump. Once axons have regenerated across the nerve graft to the distal end, Dox administration can be terminated, and host axons will cross the gap and regenerate and reinnervate the target organs. Furthermore, if complications occur, GDNF can be attenuated at any time during treatment based on the amount of Dox consumed. ⁸³ Studies have shown Dox can provoke physiological risks such as inflammation.^77,84 Other studies have shown that Dox, when consumed at sub-antimicrobial doses, does not lead to changes in antimicrobial susceptibility.^26,69 Dox is an antibiotic that is FDA approved and has been generally found to be safe to consume. In addition to inducing GDNF via the TRE-promoter, Dox may have a positive or negative effect on nerve physiology and health by itself.^85,86 It will be of value to study the optimal doses of Dox and their effects on PNI recovery. Further in vivo studies are under way to investigate the efficacy and safety of AAV-TENGs, effect of Dox on peripheral nerve regeneration, as well as spatial and temporal effect of AAV-TENGs on peripheral nerve regeneration. Similarly, further studies will also determine the optimal vector of GDNF gene copies per cell for optimized transduction. Our findings of long-term survival and biocompatibility are consistent with previous DRG allograft and AAV safety studies, further supporting the translational potential of AAV-TENGs.^62,81

Conclusions and future directions

Previous works have established the formidable potency of neurotrophic factors in promoting the health of peripheral nerves. Moreover, studies have suggested that nerve grafts capable of neurotrophic factor release can improve PNI regeneration and recovery. To avoid unwanted effects, the spatial and temporal release of these therapeutic candidates needs to be controlled. Using AAV and TENGs, we developed a novel, genetically modified nerve graft capable of releasing growth factors under spatial and temporal regulation. We showed that upon the addition of doxycycline, GDNF expression turned on and was released preferentially from the distal side of the AAV-TENG; also, GDNF expression completely turned off upon the removal of doxycycline.

These findings suggest that AAV-TENGs hold promise for enhancing nerve regeneration by providing an axon-based living scaffold with a superimposed GDNF gradient to synergistically support axonal growth and accelerate functional recovery following severe PNI. Indeed, the technical achievements and detailed validation described herein are critical enabling steps to allow execution of follow-on studies investigating the chronic efficacy of these AAV-TENGs in supporting and accelerating axonal regeneration, Schwann cell infiltration, target reinnervation, and functional recovery. These future studies will also investigate the spatial and temporal regulation of neurotrophic expression in vivo, and will investigate the effects of controlled expression of these factors on nerve regeneration and recovery. Overall, AAV-TENGs serve as a bioactive scaffold that may improve regenerative capacity to bridge critical size nerve defects and facilitate recovery following severe PNI.