Reducing neuroinflammation by delivery of IL‐10 encoding lentivirus from multiple‐channel bridges

Abstract The spinal cord is unable to regenerate after injury largely due to growth‐inhibition by an inflammatory response to the injury that fails to resolve, resulting in secondary damage and cell death. An approach that prevents inhibition by attenuating the inflammatory response and promoting its resolution through the transition of macrophages to anti‐inflammatory phenotypes is essential for the creation of a growth permissive microenvironment. Viral gene delivery to induce the expression of anti‐inflammatory factors provides the potential to provide localized delivery to alter the host inflammatory response. Initially, we investigated the effect of the biomaterial and viral components of the delivery system to influence the extent of cell infiltration and the phenotype of these cells. Bridge implantation reduces antigen‐presenting cell infiltration at day 7, and lentivirus addition to the bridge induces a transient increase in neutrophils in the spinal cord at day 7 and macrophages at day 14. Delivery of a lentivirus encoding IL‐10, an anti‐inflammatory factor that inhibits immune cell activation and polarizes the macrophage population towards anti‐inflammatory phenotypes, reduced neutrophil infiltration at both day 7 and day 28. Though IL‐10 lentivirus did not affect macrophages number, it skewed the macrophage population toward an anti‐inflammatory M2 phenotype and altered macrophage morphology. Additionally, IL‐10 delivery resulted in improved motor function, suggesting reduced secondary damage and increased sparing. Taken together, these results indicate that localized expression of anti‐inflammatory factors, such as IL‐10, can modulate the inflammatory response following spinal cord injury, and may be a key component of a combinatorial approach that targets the multiple barriers to regeneration and functional recovery.


| I N T R O D U C T I O N
To date, the only treatment for treating spinal cord injury (SCI) with any degree of success in clinical trials is methylprednisolone, which mitigates the inflammatory response after injury. 1 Inflammation after SCI initiates a damaging secondary injury response that establishes multiple barriers (e.g., accumulation of myelin debris, and reactive gliosis) to regeneration and restoration of function. The inflammatory cascade begins with upregulated production of inflammatory cytokines, chemokines, and reactive oxygen species (ROS), [2][3][4][5] which leads to recruitment of peripheral immune cells 6 and activation of local microglia, oligodendrocytes, and astrocytes. [7][8][9][10] Early recruitment of neutrophils (PMNs)and later macrophages-comprises the innate immune response to SCI.
PMNs and macrophages recruited to the injury can remove pathogens; however, they also propagate secondary injury, which results in death of neurons and oligodendrocytes. 11,12 Macrophages, in particular, also phagocytose debris left behind by necrotic or apoptotic cells, including inhibitory myelin debris whose slow clearance contributes to the inability of axons to regenerate. 13,14 In addition, inflammatory cytokines produced by PMNs and macrophages induce astrocytes to adopt a reactive state in which they proliferate and can increase the secretion of chondroitin sulfate proteoglycans (CSPGs). These reactive astrocytes then form an inhibitory glial scar that surrounds the injury site and prevents regenerating axons from crossing. 9,[14][15][16][17][18][19] In normal wound healing, the inflammatory response resolves over time, leaving behind regenerated, functional tissue. In contrast, SCI results in a chronic immune response characterized by the persistence of inflammatory cells, insufficient clearance of cellular debris at the injury, and formation of a robust glial scar that creates a barrier that limits regeneration. 12,20,21 In peripheral tissues, a transition in macrophage phenotype from inflammatory (M1 or classically activated) to antiinflammatory (M2 or alternatively activated) is largely responsible for dampening and resolving the immune response after injury. [22][23][24] However, this transition does not occur on a sufficient scale after SCI. Instead, M1 macrophages chronically persist in zones of axon degeneration, where they propagate inflammation and inhibit axon regeneration. [25][26][27][28][29][30] Though a preponderance of evidence has shown that M1 macrophages predominate within the injury site after SCI, M2 macrophages are present-typically peaking at day 14 after injury in rodents. 31 M2 macrophages generally have enhanced phagocytic capabilities 32,33 and can promote axon growth across inhibitory boundaries in vitro. 25 They stimulate tissue repair through attenuated production of inflammatory cytokines, reduced ROS production, and expression of pro-resolving cytokines such as IL- 10. 34 In contrast, M1 macrophages are characterized by increased production of pro-inflammatory cytokines, ROS and NO-each of which leads to neuron and oligodendrocyte toxicity. 25,35 Several factors promote M2 phenotypes including IL-4, IL-13, glucocorticoids, and IL-10. Of these, IL-10 was selected for this study due to its ability to promote neuron survival while reducing leukocyte infiltration and activation. [36][37][38] Macrophage depletion has been investigated as a potential solution to mitigate the inflammatory response, but there have been mixed results.
Treatment with liposome-encapsulated clodronate to deplete hematogenous macrophages improved partial hindlimb recovery and tissue repair in one study 39 and decreased fibrotic scarring while increasing axon numbers in a second study. 40 In contrast, antibody-mediated depletion of CD11c 1 monocytes/macrophages or conditional ablation by diphtheria toxin each reduced functional recovery. 41 These seemingly contradictory results suggest that specific macrophages phenotypes may have different effects on tissue repair and that eliminating or enhancing specific subtypes may be more beneficial than targeting the entire macrophage population.
Additionally, there may be unintended systemic or local consequences of depleting the entire population, such as increased susceptibility to infection and disease and reduced clearance of inhibitory debris.
While macrophage phenotype has often been depicted as a binary system where a cell is either M1 or M2 at any given time, it is now thought that macrophages exist along a continuous spectrum of activation states, with M1 and M2 as polar opposites. 23,42 For example, some reports have sub-divided M2 macrophages into additional subtypes (e.g., M2a, M2b, and M2c), each of which has a role in suppressing inflammation. The M2a (alternative) macrophages are involved in initial wound healing and the Th2-type (T-cell mediated) inflammatory response, while M2b (type 2) macrophages are believed to be immunoregulatory. Finally, M2c (deactivated) macrophages are immunosuppressive and facilitate matrix deposition and issue remodeling. 43,44 Despite observations of these macrophage subtypes in vitro, it remains unclear how this correlates with macrophage activation in vivo. Macrophages retain inherent capacity for plasticity along this activation spectrum, dynamically responding to changes in their local environment. 36,45 This dynamic plasticity makes macrophages a compelling target for resolving inflammation in the spinal cord to enable regeneration.
In addition to addressing neuroinflammation, therapies for spinal cord regeneration must provide physical support and guidance for regenerating axons across the injury site. SCI therapies focused solely on resolution of neuroinflammation have largely failed due to both lack of a growth-promoting substrate and accumulation of inhibitory factors at the injury. 46,47 While immunomodulatory strategies can reduce inflammation and promote axonal sparing, guidance of regenerating axons requires a permissive substrate. Initial attempts to provide a permissive bridge across spinal cord lesions implanted donor peripheral nerve grafts. 48 More recently, synthetic multiple-channel bridges have been developed to replicate the synergistic advantages of peripheral nerve grafts-physical support and guidance and a biologically active, pro-regenerative microenvironment. [49][50][51][52][53][54] We have previously demonstrated that multiple-channel bridges made from biodegradable poly (lactide-co-glycolide (PLG) support robust crossing of axons across the lesion site, resulting in formation of regenerated axon bundles after complete bridge degradation 6 months after SCI. 54,55 Longitudinal, macroscale channels within these bridges act as conduits for axons regenerating across the injury site, while scaffold microporosity supports host cell infiltration and tissue integration. 55,56 Herein, we investigate the hypothesis that localized lentiviral expression of IL-10 from multiple-channel PLG bridges will modulate the numbers, phenotypes, and proportions of leukocyte populations infiltrating the bridge to promote a "resolving" anti-inflammatory environment thought to be more permissive to regeneration. In addition to providing a substrate for regeneration, PLG bridges have also served as MARGUL ET AL. | 137 a platform for localized delivery of gene therapy vectors that can induce the expression of therapeutic proteins. 54,57 Biomaterial-mediated lentivirus delivery after SCI efficiently transduces infiltrating cells to yield a   localized, stable pattern of gene expression. 54,58 Lentiviral particles associated with the heparin-modified PLG bridges, which functions to retain   the vector locally and can increase its half-life for enhanced gene trans-fer. Infiltrating host cells are transduced by the lentivirus, with peak transgene expression occurring approximately 7 days after bridge transplantation with sustained expression for at least 8 weeks. 54,58 Expression of IL-10 will be investigated due to its neuroprotective and antiapoptotic properties, as well as its ability to skew macrophages towards anti-inflammatory phenotypes. 36,59 The studies focus on the first four weeks after injury comprising the acute, subacute, and intermediate phases of recovery, which encompasses the time over which many of the barriers to regeneration become established.

| Fabrication of multiple-channel bridges
Bridges were fabricated using a sacrificial template variation 61 of the gas foaming/particulate leaching technique, 54 as previously described. 62 Briefly, PLG (75:25 lactide:glycolide; inherent viscosity 0.76 dl/g; Lakeshore Biomaterials, Birmingham, AL, USA) was dissolved in dichloromethane (6% w/w) and emulsified in 1% poly(vinyl alcohol) using a homogenizer (PolyTron 3100; Kinematica AG, Littau, Switzerland) at 3000 rpm to create microspheres (z-average diameter 1mm). D-sucrose was caramelized, cooled, and drawn from solution with a Pasteur pipette to make sugar fibers. These fibers were coated with a mixture of PLG microspheres and salt (63-106 lm) and pressed into a salt-lined aluminum mold. The materials were then equilibrated with CO 2 gas (800 psi) for 16 hrs and then gas foamed in a custom-made pressure vessel. Bridges were subsequently cut into 2.25 mm sections and leached for 2 hrs to remove porogens. The bridges are dried overnight and stored in a desiccator.

| Heparinization of nerve bridges
Heparin coating of bridges has been shown to enhance lentiviral loading and transduction from bridges in vivo. 63

| Virus loading onto heparinized nerve bridges
Multiple additions of viruses were adsorbed onto bridges in an iterative manner in order to increase lentiviral loading. Prior to virus addition, bridges were disinfected in 70% ethanol and washed with water. After 12 min of drying time, bridges were saturated with 2 ll of virus. Bridges were then dried for an additional 12 min followed by another 2 ll of virus. Bridges were then dried for 14 additional min followed by a final 2 ll of virus. After a final 5 min of drying, bridges were stored at 2808C until used for surgery.

| Tissue processing and immunofluorescence
Spinal cord tissue was collected at days 7, 14, and 28, which were chosen to represent acute, subacute and intermediate phases of regeneration respectively. It may be noted that days 7 and 14 are at the tail end of their respective phases; however, it is not possible to extract the spinal cord and bridge together prior to day 7 as the bridge falls out due to a weak interface tissue-bridge interface. Additionally, cellular infiltration is lower at earlier time points, making flow cytometry challenging. Cell counts were then normalized to counted area in each tissue section.
Macrophage shape was characterized for each positive cell and cells were binned as fibrous/elongated, round and not ruffled with a bright F4/80 1 perimeter, or as multinucleated foreign body giant cell (FBGC).
To assess the numbers of regenerated and myelinated axons, NF200 was used to identify axons, NF200 1 /MBP 1 to determine the number of myelinated axons, and NF200 1 /MBP 1 /P0 1 to determine the amount of myelin derived from infiltrating Schwann cells. 57  in MOPS buffer, as previously described by Beck et al. 20 Cells were subsequently blocked with a solution containing 1% normal mouse and rat serum (Sigma Aldrich) and anti-mouse CD16/32 (eBioscience) and stained for viability using fixable violet dead cell stain (Invitrogen) and stained with the specific antibodies listed above. Data were acquired on a BD LSR II cytometer, and analyzed using FloJo software. Fluorescence minus one staining was used as a negative control.

| Flow cytometry of digested tissue samples
The following flow cytometry antibodies were used: v500-

| mRNA isolation and qRT-PCR analysis
To isolate mRNA, spinal cord tissues and bridge implants were explanted with 2 mm of spinal cord both rostral and caudal to the bridge. Explanted tissues were homogenized using 1 ml of Trizol reagent (Life Technologies) with a tissue grinder. RNA isolation was followed by chloroform extraction and isopropanol precipitation. 66 The extracted RNA was dissolved in 30 ll of RNase-free distilled water and RNA concentration was measured using a NanoDrop 2000C (Ther-moFisher Scientific, Newark, DE, USA) and to assure sufficient purity (A260/A280 ratios between 1.9 and 2.1 for all samples). Total isolated RNA was stored at 280 C freezer until use. cDNA was synthesized using iScript TM DC t 5 C t, sham -C t,18s-rRNA. 68

| Behavioral analysis
The Basso mouse scale (BMS) open-field locomotor test (range 0-9) was used to assess functional recovery for a period 24 weeks after SCI as previously described. 69 A baseline was determined prior to SCI, and mice were tested 3, 7, 14, 21, and 28 days. Observations and BMS scoring were performed by two trained observers at 4-min intervals. however, by day 28, the macrophage percentage increased in the bridge conditions such that they were no longer significantly less than the no bridge condition ( Figure 3B). Notably, these changes represent the percentage of CD45 1 cells that were also F4/80 1 ; thus, the increase in macrophage percentage at day 28 does not imply an increase in the total number of macrophages. Finally, delivery of IL-10-encoding lentivirus reduced numbers of infiltrating PMNs by 3.5-fold when compared to the bridges loaded with control lentivirus at Day 7 (p < .05) ( Figure 3C).

| Statistical analysis
Bridges loaded with IL-10-encoding lentivirus had similar numbers of infiltrating PMNs as in hemisections with no bridge implant. Localized IL-10 expression at the bridge did sustain a 3.5-fold reduction of PMN infil-tration through day 28 (relative to the no bridge control) (p < .05) ( Figure   3C). Additionally, IL-10 expression reduced the mean level of PMN infiltration by 2.4-fold compared to all other bridges, though this decrease was not statistically significant ( Figure 3C).

| IL-10 gene delivery from bridges increases M2 macrophage infiltration
Immunofluorescence staining of tissue sections was applied to investi-  Figure   5D). For the round macrophages, no significant differences were observed, though a strong trend (p 5 .088) toward an increase in arginase 1 round macrophages was identified for IL-10 expression relative to FLUC across all time points ( Figure 5E). Finally, for the FBGCs, a greater density of the arginase 1 cells was observed at Day 14 for the IL-10 conditions relative to the FLUC virus control ( Figure 5F).

| IL-10 delivery results in elevated arginase expression
mRNA levels for arginase were measured by qRT-PCR to complement the arginase 1 cells density assessment by immunofluorescence. At day 7, qRT-PCR revealed a 5-fold or greater increase in arginase mRNA in the IL-10 group (p < .001) (Figure 6). Increased mRNA at day 7 may  Figure 4F). At day 14, arginase expression in IL-10 bridges was elevated almost 5-fold compared to bridges loaded with FLUC-encoding lentivirus (p < .0001). At day 28, arginase mRNA levels remained more than 2-fold higher for the IL-10 condition, but this difference was not statistically significant (.08 < p < .13)

| Axon regeneration and myelination
Axon numbers and myelination were quantified at day 28 to assess whether IL-10 overexpression would prove detrimental, which has been reported in the peripheral nervous system. 71 Myelinated (NF200 1 /MBP 1 ) and unmyelinated (NF200 1 /MBP -) axons were seen throughout the bridges in animals receiving no lentivirus, FLUC-encoding control lentivirus, or IL-10-encoding lentivirus (Figure 7A-D) 28 days after SCI. NF200 1 axons were typically observed as bundles as previously reported for multichannel PLG bridges. 54,57,72 Empty bridges had approximately 800 neurites/mm 2 , and both lentiviral conditions had approximately 1100 neurites/ mm 2 . While the data suggest a trend toward greater densities of regenerating axons with lentivirus delivery, these differences were not statistically significant. Similarly, the percentage of axons that were myelinated (22-34%) or the percentage of myelination that was derived from Schwann cells (15-35%) did not vary between conditions ( Figure 7E).

| Motor function is improved with IL-10
Motor function in the ipsilateral hindlimb was characterized over the Il-10 results in elevated arginase mRNA. qRT-PCR revealed elevated arginase levels at 7 and 14 days after SCI. Levels statistically returned to baseline by day 28, but a strong trend remained (0.08 < p < 0.13). N 5 4-6 per group for PCR. Statistical analysis was completed using a one-way ANOVA with Tukey's post hoc test at each time point. Significantly different groups were denoted with letters with "a" and "b" denoting statistical significance between the two groups: day 7 (p <.001) and day 14 (p <.0001)  37,38 Despite several studies investigating IL-10 delivery for SCI, the effects of sustained localized delivery on dampening and resolving the immune response to promote regeneration have not previously been studied. 59 and are the focus of the studies herein.
We initially demonstrate that delivery of IL-10-encoding lentivirus can decrease PMN infiltration into biomaterial bridges implanted following SCI, and that though the loading of lentivirus to bridges may promote PMN infiltration, IL-10 over-expression reverses this effect.
Although PMNs are transient participants in the immune response fol-lowing injury to peripheral tissues, they can persist for weeks after injury to the spinal cord. 12,20 Furthermore, though virus delivery from biomaterials can be used to enhance nerve regeneration, 54,76 addition of virus to the biomaterial platform can increase the extent of cell infiltration. [77][78][79] Secretion of chemokines from the increased immune cells may be responsible for the increase in PMN infiltration at day 7 and 28 reported here ( Figure 3C), though increased cytokines would also be expected to simultaneously influence macrophage infiltration, which was not observed and is in agreement with a previous report. 58 Due to their primarily bactericidal role in wound healing, PMNs are not expected to provide neuroprotection and are primarily detrimental in SCI. 12 PMNs (and macrophages) release cytokines, free radicals, eicosanoids, and proteases, which are toxic to neurons and glia. In particular, superoxide, nitric oxide, and peroxynitrite are highly toxic and create irreversible damage to cellular components, inducing apoptosis. 80 However, tail vein administration of the RB6-8C5 Ly6G/Gr-1 antibody, which selectively depleted the hematogenous PMN population by more than 90%, resulted in less glial scarring, decreased spared tissue, and worse BMS functional results. 81  The BMS was used to test for differences in motor function in the hindlimb ipsilateral to SCI (n 5 15 per group). Statistical analysis was completed using a two-way ANOVA with repeated measures and a Sid ak correction for multiple comparisons. Significant differences notated with *** (p < .001) or **** (p < .0001) Figure 4, which may depend on the specific subtype and phase of regeneration (inflammatory, proliferative, or remodeling). 44 IL-10 known for its ability to upregulate expression of IL-4Ra, which may synergize with IL-4 dependent arginase expression. 86  In vivo characterization of the relationship between macrophage phenotype and morphology has similarly been conflicting. One hypothesis is that the elongated F4/80 1 cells are activated hematogenous macrophages, while the round cells are activated resident microglia.
Microglia are well known to proceed from a highly ramified morphology toward an amoeboid morphology as they become activated and phagocytic after injury. 89 Unfortunately, antibodies have not been identified to accurately distinguish macrophages from microglia using immunohistochemistry, though future studies using genetic models such as cr2 rfp ::Cx3cr1 gfp mice 90 have the potential to distinguish cell morphology, phenotype, and lineage. Alternatively, these different morphologies may represent distinct M2 phenotypes. Shechter et al. selectively ablated monocyte-derived macrophages followed by adoptive transfer of monocytes. 41 The exogenous monocytes differentiated into macrophages with an activated morphology, which was manifested by a large cell body with few to no processes, arginase expression, and IL-10 release. These factors suggested an anti-inflammatory phenotype, and the monocytes contributed to regeneration. 41 A subset of infiltrating monocyte-derived, anti-inflammatory macrophages may be essential for recovery, though these cells represent a specific subset of M2 macrophages and will require further characterization. 91,92 The round macrophages with prominent F4/80 1 borders may be undergoing fusogenesis. FBGCs have not been extensively stud- Although IL-10 did not increase axon numbers in the bridge, IL-10 robustly improved motor function after SCI (Figure 8). IL-10 may act to moderate the deleterious immune response, resulting in improved axon sparing. Increased sparing with increased IL-10 expression is consistent with previous reports, which have demonstrated that IL-10 reduces loss of neurons and oligodendrocytes directly through trophic support 38 and indirectly by limiting the immune response. 59 Additionally, in our past work, differences in numbers of regenerating axons between conditions become more pronounced at later time points. 54 Thus, improved motor function observed in the current study suggests improved sparing and plasticity instead of regeneration. While these data suggest that IL-10 overexpression from PLG bridges effectively improves function after SCI, delivery of IL-10 in combination with neurotrophic factors known to promote axon regeneration, such as neurotrophin-3 (NT-3), 54,57 may act synergistically to further improve both sparing and regeneration.
The present study indicates that immobilization of IL-10 lentivirus onto multiple channel bridges alters the immune response to create a microenvironment that is permissive to regeneration and shows promise as a translatable strategy. The bridges used are made of PLG, a biomaterial that has been utilized for decades in FDA-approved applications, including biodegradable sutures and drug delivery vehicles. 95 PLG is easily sterilized for clinical using standard techniques, most commonly g-irradiation. 96 Though safety has been cited as a concern for lentiviral vectors, there are ongoing clinical trials using lentivirus without adverse effects or evidence of insertional mutagenesis. [97][98][99] Pyrogens from virus production can be limited by purification through gravity-flow columns to yield endotoxin-free, concentrated plasmid, as done in these studies. Furthermore, good manufacturing practices (GMP) have been developed for large-scale preparation of lentivirus for clinical use. 98,100 Finally, the hemisection model of SCI, while not perfectly representative of all SCI, is a translatable platform for studying the injury site and developing treatments. While contusive injuries represent the majority of SCI in the developed world, at least 28% of cases in the US military are penetrating, 101 and in South Africa, more than 60% of SCI is categorized as penetrating. 102 Moreover, the initial deficits resulting from penetrating SCI exhibit substantially less improvement over time than those from contusive injuries. 103