The role of the immune system during regeneration of the central nervous system

Abstract

Central nervous system damage in mammals leads to neuronal cell death, axonal degeneration, and formation of a glial scar resulting in functional and behavioral defects. Other vertebrates, like fish and salamanders, have retained the ability to functionally regenerate after central nervous system injury. To date research from many research organisms has led to a more concise understanding of the response of local neural cells to injury. However, it has become clear that non-neural cells of the immune system play an important role in determining the tissue response to injury. In this review we briefly consider the mammalian response to injury compared to organisms with the natural ability to regenerate. We then discuss similarities and differences in how cells of the innate and adaptive immune system respond and contribute to tissue repair in various species.

1. Introduction

Humans are unable to functionally regenerate after central nervous system (CNS) traumas. Various animal models (i.e. mice and rats) of mechanical or contusive injury to the brain and spinal cord have provided a wealth of information regarding our inability to regenerate lost or damaged CNS tissue. However, we still do not know how to therapeutically stimulate robust functional recovery. This is due, in part, to the complexity of CNS traumas, which involves multiple cell types, both neural and non-neural, axonal degeneration, and formation of a growth inhibitory glial scar. Remarkably, other vertebrates, such as fish, frogs, and salamanders, are capable of regenerating a fully functional CNS after injury. Functional CNS regeneration in these species is characterized by the lack of reactive gliosis or a glial scar, robust axon regeneration, and a pro-regenerative glial cell response. Better understanding of the mechanisms utilized by naturally regenerating species could identify particular processes or pathways for therapeutic intervention.

Apart from the endogenous neural response to CNS trauma, both the innate and adaptive immune system play an important role in regulating a tissues response to injury. Immediately following injury, dead and necrotic cells release damage associated molecular patterns (DAMPs), reactive oxygen species (ROS), purine metabolites (ATP), mitochondrial components, and diverse polysaccharide components ^{1, 2} These factors bind to pattern recognition receptors (PRRs) (i.e. Toll-like receptors (TLRs) or P2X7) on the surface of neighboring cells, including resident immune cells, which stimulates the release of cytokines that are responsible for mobilizing an immune response ³ While this immune response is critical to protect the host from wound-related infections, functional studies have shown that this cascade of events also directly regulates the regenerative outcome.

In this review, we will briefly discuss the different response to CNS trauma in several species, including mice, zebrafish, and salamanders. We will then focus on the role of the immune system following injury. Specifically, on the role of cells of the innate immune system, the molecules they secrete, and how that influences the response of local neural cells. Finally, we will review the poorly understood and emerging role of the adaptive immune system during CNS regeneration.

2. Vertebrates Response to Central Nervous System Damage

2.1. Spinal Cord Regeneration in Naturally Regenerating Organisms

Some vertebrates, like fish, frogs, and salamanders, have the remarkable ability to regenerate the CNS after injury ⁴. Specifically, after spinal cord injury there is robust axon regeneration, which results in function restoration. Early retrograde tracing approaches in zebrafish and salamanders identified specific brain regions that project axons to the spinal cord. Following SCI, every spinal cord projecting brain region regenerates axons through the lesion. However, in zebrafish the organization of the regenerated axon tracts is slightly different from the pre-injury organization ^5-7. Instead of regenerating through the white matter, axons preferentially migrate through the spinal cord grey matter before making connections with cells caudal to the lesion ⁸. Further investigations into the regenerative potential of neuronal subpopulations using transgenic reporter animals confirmed that the organization of regenerated TH⁺ (dopaminergic) and 5-HT⁺ (serotonergic) axons is drastically different compared to uninjured spinal cords ⁹. Interestingly, dopamine released from regenerating TH⁺ axons directly regulated the neural stem cell response to injury and promoted motor neuron regeneration in the zebrafish spinal cord ¹⁰.

In addition to extensive axon regeneration, naturally regenerating organisms exhibit a pro-regenerative glial cell response to injury. After injury glial cells adjacent to the lesion proliferate and migrate to repair the damaged spinal cord and create a permissive environment for axon regeneration ^11-14. After spinal cord injury in zebrafish, GFAP⁺ glial cells adjacent to the injury adopt a bipolar, elongated morphology and form a glial bridge connecting the rostral and caudal stumps ^{13, 15}. Formation of the glial bridge is dependent on FGF and connective tissue growth factor (CTGF) signaling, and overexpression of CTGF is sufficient to speed up bridge formation and functional recovery ^{13, 15} Regenerating axons use the glial bridge as a substrate to migrate across the lesion and restore motor function. Whether these same signaling mechanisms regulate glial bridge formation in lamprey is not clear.

In contrast to zebrafish, GFAP⁺ glial cells in the salamander spinal cord do not form a glial bridge. Early reports show that after spinal cord transection, glial cells seal over the rostral and caudal stumps and migrate towards each other ^{11, 12, 14, 16}. A detailed characterization of spinal cord regeneration in newts suggests that glial cells and regenerating axons are intimately associated with each other throughout the regeneration process ¹⁴. This is slightly different than in zebrafish where the glial bridge forms first and then axons regenerate over the bridge and through the lesion. Regardless of the mechanism, the overall conclusion remains similar: glial cells adjacent to the lesion undergo a pro-regenerative response to injury, function to reconnect the spinal cord, and create a permissive/instructive environment for axon regeneration.

More recent molecular analyses have identified several signaling pathways that are important for the pro-regenerative glial cell response. Following spinal cord transection or tail amputation in zebrafish and salamanders classical developmental signaling molecules like Wnt/PCP, FGF, Notch, Sonic Hedgehog, and Sox2 play critical roles in regulating the pro-regenerative glial cell response, spinal cord patterning, asymmetric neural stem cell division, and neuronal differentiation ^17-23. Additionally, microRNAs play a critical role in promoting spinal cord regeneration after spinal cord transection and tail amputation. Inhibition of a subset of these microRNAs results in defects in glial cell proliferation, spinal cord patterning, ependymal bulb formation, and expression of various axon guidance molecules or transcription factors ^24-27. Recently, it was shown that axolotl glial cells up-regulate the non-canonical AP-1^cFos/JunB after spinal cord transection, which functions to promote regeneration and inhibit glial scar formation ²⁶. This is in contrast to damaged mammalian astrocytes, which express the canonical AP-1^cFos/cJun, which functions to promote reactive gliosis and glial scar formation ^28-30. How changes in the composition of the AP-1 transcriptional complex results in drastically different cellular responses to injury is still not clear but remains an intriguing question.

2.2. Mammalian Response to CNS Damage

Damage to the mammalian CNS results in neuronal cell death, axon degeneration, and formation of a glial and fibrotic scar that collectively lead to behavioral defects (Figure 1). There are many factors that contribute to regenerative failure in mammals, including the inability of neurons to initiate new axon growth ^31-34, as well as a lesion microenvironment that inhibits axon regeneration ^35-38. Following axon damage there is unrestricted influx of calcium into the axoplasm and breakdown of the cytoskeletal components ^{34, 39-41}. These events result in massive cytoskeletal rearrangement, activation of various signaling pathways, and proteases, such as calpains ^{39, 41, 42}. Activation of these pathways in response to axon damage are important for sealing the damaged axon and supporting growth cone formation in organisms with the innate ability to regenerate, like Aplysia and the sea lamprey ^{34, 40, 43}. However, in mammals these early signaling events fail to induce formation of new growth cones and instead result in formation of dystrophic end bulbs ³⁹. It was historically thought that dystrophic end bulbs were stagnant structures but actually there is persistent and highly dynamic remodeling of the actin and microtubule cytoskeleton, even into the chronic stages of injury ^{32, 44-46}. Remarkably, when exogenously supplied with neurotrophins up to a year after injury some dystrophic axons formed new growth cones and migrated through the lesion ^{32, 45}. Collectively, these studies suggest that damaged mammalian axons can be induced to a regenerative response if given the correct signals.

Figure 1: Inflammatory signaling after spinal cord injury.

Following spinal cord injury damaged axons degenerate, form dystrophic end bulbs, but largely fail to migrate across the lesion. Astrocytes adjacent to the injury undergo reactive gliosis and contribute to the formation of the glial scar. Dead and necrotic cells release DAMPs, which signal to local neural cells and recruit leukocytes into the lesion. Collectively, the interaction between local cells and leukocytes influences the CNS response to injury by regulating reactive gliosis, glial scar formation, and neuronal survival.

Apart from environmental cues, there are neuron intrinsic roadblocks to axon regeneration. A handful of molecules and pathways have been identified as modulators of axon regeneration after injury. Experimental activation of mTOR signaling or overexpression of various KLF transcription factors in corticospinal tract neurons promotes axon regeneration and sprouting after spinal cord injury ^47-51. Additionally, overexpression of STAT3 in dorsal root ganglion neurons enhances regeneration of the centrally projecting axons ⁵². Finally, conditional knockout of suppressor of cytokine signaling 3 (SOCS3) in retinal ganglion cells promotes axon regeneration after optic nerve crush ⁵³. However, the precise mechanisms by which these molecules stimulate axon regeneration remain unclear. An additional caveat to these experiments is that gene manipulation occurred prior to injury. An outstanding question in the field is whether and to what extent manipulation of these pathways after injury can active a regenerative response. Remarkably, forced activation of mTOR signaling in corticospinal tract neurons one year after injury is sufficient to stimulate axon regeneration through the lesion ⁵⁴. While the regenerated axons formed synaptic structures caudal to the lesion whether there was a functional improvement was not addressed ⁵⁴. How to ensure that regenerated axons form the appropriate connections with the right post-synaptic partner to restore lost function represents an additional level of complexity to the problem.

In addition to neuron intrinsic pathways, non-neuronal cells, including astrocytes, oligodendrocytes, microglia, and peripheral inflammatory cells exquisitely regulate the mammalian response to CNS damage ^{36, 37, 55-57}. Injury in the CNS result in the formation of two distinct compartments: 1) a fibrotic lesion at the core of the injury, which is comprised of fibroblasts, inflammatory cells, endothelial cells, and growth inhibitory extracellular matrix^{35-38, 58} and 2) a glial scar comprised of reactive astrocytes ^{38, 58-60}. Reactive astrocytes are the major glial cell type contributing to the glial scar. Following injury, astrocytes adjacent to the injury site proliferate, undergo reactive gliosis, up-regulate expression of the intermediate filament proteins glial fibrillary acidic protein (GFAP) ^61-63 and vimentin ⁶⁴, and migrate towards the lesion. The reactive astrocytes originate from pre-existing astrocytes that re-enter the cell cycle, proliferate, and migrate towards the lesion ^{29, 65, 66}. Lineage tracing experiments suggest ependymal cells, a putative spinal cord neural stem cell population, could give rise to astrocytes after injury, which then undergo reactive gliosis and contribute to the glial scar ^{67, 68}. Reactive astrocytes within the glial scar express axon growth inhibitory molecules such as chondroitin sulfate proteoglycans (CSPGs), semaphorins, and ephrins ^{36, 37}.

While reactive astrocytes do secrete growth inhibitory molecules, recent reports using sophisticated cell ablation and conditional gene knockout approaches suggest a protective role for the glial scar. Ablation of proliferative, scar forming reactive astrocytes led to increased axonal degeneration, a failure to repair the blood brain barrier, and decreased functional recovery compared to wild type mice ⁶⁶. The increased axonal degeneration was attributed to a massive influx of inflammatory cells that are normally restricted to the fibrotic scar at the lesion center ⁶⁶. However, ablation of reactive astrocytes led to a breakdown of the glial barrier separating the fibrotic lesion core from surrounding spinal cord tissue ⁶⁶, allowing the aberrant infiltration of inflammatory cells into the spinal cord parenchyma. Further analysis identified the transcription factor STAT3 as being necessary for induction of reactive gliosis in spinal cord astrocytes after injury ⁶⁹. Specific knockout of STAT3 from astrocytes using a GFAP-Cre phenocopied the cell ablation experiments, further highlighting a potentially protective role of reactive astrocytes early after injury ⁶⁹. While early work largely focused on the effect of reactive astrocytes on axon regeneration, it is now clear that reactive astrocytes also regulate and respond to the inflammatory response to CNS trauma. The interaction between astrocytes and inflammatory cells can have major consequences on the tissue response to injury. We will now briefly consider what is known about these interactions.

3. Reactive Astrocytes and Inflammatory Signaling

Traumatic brain and spinal cord injury results in the death of local cells and disruption of the blood brain barrier (BBB) ^{55, 56}. These events result in dead and necrotic cells releasing various DAMPs, including ATP and the DNA scaffolding protein high mobility group B1 (HMGB1), into the surrounding environment. These DAMPs will diffuse away from the lesion and bind to Toll-like receptors and other PRRs on neighboring cells and enter into circulation to stimulate a systemic inflammatory response. While the subsequent inflammatory response is essential to protect against infection, it also plays a major role in regulating the tissue response to injury (Figure 1).

While immune cells primarily express PRRs they are also expressed by astrocytes ^{70, 71}. Activation of PRR signaling in astrocytes by ATP or HMGB1 directly or indirectly activates pro-inflammatory transcriptional cascades, like NF-κB or STAT3. Subsequent NF-κB activation contributes to cytotoxic edema associated with traumatic brain injury ⁷². Paradoxically, NF-κB signaling also stimulates astrocytes to express key pro-survival, neurotrophic molecules and could also mediate a neuroprotective function ⁷³.

In addition to regulating the astrocytic response to injury, PRRs can signal in local, non-neural cells as well. HMGB1 signaling in endothelial cells stimulates vascular remodeling and BBB repair after brain injury ^{74, 75}. Additionally, HMGB1 induces IL-6 expression in microglia ⁷⁶. Increased IL-6 results in STAT3 activation in astrocytes leading to the up-regulation of aquaporin 4, which further exacerbates trauma-associated edema ⁷⁶. Interestingly, STAT3 is a key regulator of reactive gliosis and glial scar formation in astrocytes, suggesting STAT3 likely regulates multiple processes in astrocytes after injury ⁶⁹. Other signaling molecules that function in astrocytes to promote reactive gliosis and glial scar formation are SOCS3 and NF-κB ^{77, 78}. Intriguingly, these pathways play important roles in regulating and responding to inflammatory signals, suggesting that the inflammatory response and reactive gliosis are linked at the molecular level. Consistent with this hypothesis, recent reports have identified microglia as important regulators of the reactive phenotype.

Transcriptional profiling of reactive astrocytes after stroke or systemic lipopolysaccharide injections revealed distinct transcriptional differences between the resulting populations ⁷⁹. Recently, 2 subtypes of reactive astrocytes were genetically and functionally described, where A1 reactive astrocytes were induced by classically activated microglia, were pro-inflammatory, and exhibited cytotoxic effects on neighboring neurons ⁸⁰. The A1 phenotype could be induced in vivo and in vitro by exposing astrocytes to TNF-α, IL-1α, and the complement component C1q ⁸⁰. Conversely, A2 reactive astrocytes were induced after ischemic injury and secreted neurotrophic and angiogenic factors that conferred a more neuroprotective effect ⁸⁰. These studies suggest that different kinds of stimuli (i.e. ischemia, inflammation, mechanical injury, degeneration) could induce a unique reactive state in astrocytes. Whether reactive astrocytes have a predetermined state or could transition along a spectrum of phenotypes remains an intriguing question ⁵⁷.

4. An Immunological Blueprint for Nervous System Regeneration

4.1. Neutrophils and mammalian nervous system response to injury

Neurological damage in mammals results in the loss of sensory, motor, and cognitive function. Historically, research has focused on elucidating the cellular and molecular mechanisms that lead to regenerative failure in neural tissue, such as why neurons are not replaced, why axons fail to regenerate, and how reactive astrocytes and oligodendrocytes contribute to the glial scar. However, in the last few decades it is clear that the non-neural immune system plays a key role in regulating the response of neural tissue to injury ^81-84. After spinal cord injury or traumatic injury to the brain, neutrophils are the first inflammatory cell type to traffic to the lesion, appearing within hours, peaking around 24 hours, and largely cleared from the lesion 3-5 days post injury (Figure 2) ^85-90.

Figure 2: Dynamics of Leukocyte Infiltration in various species after Spinal Cord or Brain Injury.

Schematic representation of leukocyte infiltration after mammalian spinal cord injury^{91, 200}, adult zebrafish spinal cord injury,^{8, 13} larval zebrafish spinal cord injury, ¹⁵⁰ Axolotl spinal cord injury,^{151, 152} mammalian brain injury,⁸⁴ adult zebrafish brain injury,^{135, 136, 201} and larval zebrafish brain injury ¹⁴¹.

In general, neutrophil function seems to inhibit nervous system regeneration and exacerbate tissue damage ⁹¹. After injury, local neural cells respond to DAMPs released from damaged cells and additionally release various signals that recruit neutrophils to the injury site. Experimental perturbation of some of these damage responsive pathways like, leukotriene B4, cAMP, NF-κB signaling, CXCR2, and myeloperoxidase signaling results in decreased neutrophil trafficking to the lesion and an overall better histological score following injury ^92-95. In vitro experiments suggest that neutrophils exert neurotoxic effects on cultured neurons via cell-cell interactions and through the secretion of factors including MMP9, ROS, and TNF-α ^{96, 97}. Knockout models of neutrophil elastase exhibited decreased edema and fewer apoptotic cells after traumatic brain injury ⁹⁸. However there was no effect on neutrophil migration or MMP9 expression, and no long-term functional benefits.

One caveat of these experiments is that many of these genetic perturbations are not neutrophil specific, but instead target the entire myeloid lineage. Approaches utilizing antibody-based depletion of Ly6G/Gr-1⁺ neutrophils have reported a potentially beneficial role for neutrophils after spinal cord injury. Neutrophil depleted mice exhibit a much larger ROS burst after injury as well as decreased vasculogenesis and a heightened inflammatory response ^99-101. These results suggest that after spinal cord injury, neutrophils could play a neuroprotective role by mitigating ROS and the inflammatory response while promoting new blood vessel formation. Whether neutrophil depletion resulted in long-term functional benefits after spinal cord injury remains unclear. In contrast, after brain injury, neutrophil depletion appears to be beneficial. Neutrophil depletion resulted in less edema, fewer apoptotic cells, and reduced macrophage/microglia activation in response to brain injury ¹⁰². However, as with elastase knockout, there were no long-term functional improvements.

4.2. Macrophages/Microglia and mammalian nervous system regeneration

Unlike neutrophil infiltration, the kinetics of macrophage/microglia migration is slightly different after spinal cord and brain injury. In the spinal cord, macrophages/microglia appear during the sub-acute phase 2-7 days post injury and can persist within the lesion months after the initial injury (Figure 2) ^{91, 103}. Whether these chronically persistent macrophages are the result of continued monocyte trafficking or proliferation of lesion-associated macrophages is not clear ¹⁰³. However, in the brain, macrophages/microglia appear 3-7 days after injury and are largely cleared by chronic stages of injury ^{82, 84, 104}. When chronically activated macrophages/microglia persist around the lesion, this is often correlated with a poor cognitive prognosis ^{105, 106}.

In mouse models of spinal cord injury, ablation of macrophages/microglia with Clodronate loaded liposomes resulted in improved motor recovery ^107-109. However, macrophage ablation using thymidine kinase or diphtheria toxin receptor expression in CD11b⁺ myeloid cells did not result in improved repair after brain injury ^{110, 111}. However, ablation of CD11b⁺ cells even in uninjured animals resulted in an inflammatory response making it difficult to draw definitive conclusions. Inhibition of macrophage/microglia migration to the lesion by blocking CCR2 does help to combat cognitive decline, suggesting an overall detrimental role during functional recovery ^112-114. It is experimentally difficult to separate macrophages from microglia, leading many researchers to investigate the role that collective cell populations have on nervous tissue repair. However, attempts to differentiate between the two populations using various classification schemes suggest that peripheral macrophages and resident microglia do play distinct roles after nervous system injury ^{83, 115-117}.

Utilizing bone marrow chimeric mice, peripheral macrophages were GFP labeled based on expression of the chemokine receptor CX3CR1, which allowed for differentiation between peripheral macrophages (GFP⁺) and microglia (GFP⁻) ^117-121. These approaches revealed that macrophages are enriched in the fibrotic center of the lesion while microglia are largely associated with the glial scar, which has a high concentration of chondroitin sulfate proteoglycans. Interestingly, microglia express CD44, which binds to CSPGs expressed by cells in the glial scar leading to the expression of neurotrophins and an overall neuroprotective effect ¹²². Activated microglia also secrete various cytokines that stimulate local astrocytes to undergo reactive gliosis and thus facilitates formation of the glial scar ¹²³. Yet, this too could prove beneficial as the glial scar mitigates the acute inflammatory response to injury and prevents excessive secondary damage ^{66, 69, 124}. However, during chronic phases of spinal cord injury the glial scar does represent a major barrier to axon regeneration and functional restoration. It is important to mention that Ly6^hi/CX3CR1^lo pro-inflammatory peripheral macrophages are also found in the glial scar, although the functional significance of this finding is not clear ^{103, 118, 121, 125}.

Traumatic brain injury models in CX3CR1 knockout mice have fewer macrophages/microglia that infiltrate the lesion. This results in an early neuroprotective effect after injury but leads to the persistence of a chronic pro-inflammatory lesion, which is normally resolved in wild type animals ¹²⁶. The chronic pro-inflammatory lesion results in increased neuronal cell death and decreased cognitive performance. This suggests that early inhibition of CX3CR1⁺ cell migration to the lesion could be beneficial following injury. However, these cells are necessary at later time points to resolve the inflammatory response and facilitate functional recovery. It is not clear whether the same or two functionally distinct cell populations mediate these opposing processes. Following spinal cord injury, there does appear to be two temporally distinct waves of peripheral macrophage infiltration. Inhibition of the later wave results in decreased functional recovery, suggesting this later population of macrophages is beneficial to repair ⁸⁵. These differing results could be due to the heterogeneity of macrophage phenotypes ¹²⁷.

Pro-inflammatory M1 macrophages are generally thought to be detrimental to regeneration while the anti-inflammatory M2 macrophage is thought to promote regeneration ^128-130. While this system implies a binary cell state, accumulating evidence supports the notion that macrophage phenotypes reside on a spectrum between the M1 and M2 classification ¹³¹. M2 macrophages are present after spinal cord injury but only within the first 7 days and are quickly replaced by M1 macrophages, which persist into chronic stages ¹³². These results suggest that the chronic persistence of M1 macrophages could be responsible for a pro-inflammatory microenvironment that impedes functional recovery. Similarly, after traumatic brain injury there is an influx of M1 macrophages at 3 and 7 days post injury with a transient increase in M2 macrophages at 5 days post injury ¹³³. Whether these dynamics are necessary to avoid catastrophic effects due to prolonged M1 or M2 signaling is not clear. Furthermore, it would be interesting to determine if activation of a more M2-like signature during chronic stages of nervous system repair would be sufficient to confer a more regenerative response.

4.3. Inflammation and functional nervous system regeneration

Recent studies in zebrafish are beginning to shed light on the role of various immune cell types during nervous system regeneration (Figure 2). An acute inflammatory response is necessary for zebrafish brain regeneration after stab injury or neuronal ablation ^134-139. Inhibition of the inflammatory response leads to decreased neural stem cell proliferation and neuronal differentiation ¹³⁶. Circulating macrophages are essential for the re-establishment of the blood brain barrier in response to damage ¹⁴⁰. After stab injury in the larval zebrafish brain there is rapid recruitment of microglia to the lesion however tissue repair largely occurs in the absence of neutrophils ¹⁴¹. Intriguingly, microglia in the newt brain seem to play an inhibitory role during dopaminergic neuron regeneration following chemical ablation ¹⁴². Following neuronal ablation newts normally regenerate dopaminergic neurons within 30 days and this is dependent on neural stem cell differentiation ^{143, 144}. Interestingly, inhibition of the microglial response to neuronal ablation results in more TH⁺ neurons ¹⁴². While inhibition of microglia does not affect neural stem cell proliferation, it does lead to fewer apoptotic neurons suggesting that microglia normally impede neuronal survival ^{142, 145}.

Adult and larval zebrafish are able to regenerate spinal cord motor neurons after mechanical injury or chemical ablation ^{17, 146}. Specific ablation of motor neurons using the nitroreductase system ¹⁴⁷ stimulates a robust inflammatory response, which is necessary for motor neuron regeneration ¹⁴⁸. The precise mechanism by which inflammatory cells stimulate the endogenous neural stem cells to differentiate into new motor neurons is still not clear. However, stab injury in the zebrafish brain leads to expression of the chemokine receptor CXCR5 in neural stem cells and is necessary for regenerative neurogenesis ¹⁴⁹. This suggests that injury leads to differential expression of immune modulatory molecules, potentially sensitizing local cells to subsequent inflammatory signals to promote a regenerative response. While a similar reaction occurs after injury in mammals, this more often than not leads to a scarring or fibrotic response, however an important difference is the duration of the immune response in pro-regenerative species versus non-regenerative mammals.

The inflammatory response to injury also plays an important role in regulating axon regeneration after spinal cord injury. In larval zebrafish, there is a rapid and dynamic influx of macrophages/microglia and neutrophils to the lesion, which secrete TNF-α or IL-1β, respectively. Macrophages and macrophage-derived TNF-α is necessary for axon regeneration and neutrophil clearance. Genetic ablation of macrophages or CRISPR-mediated TNF-α knockout resulted in sustained neutrophil presence and chronic IL-1β signaling, which blocked axon regeneration ¹⁵⁰. Leukocyte infiltration is slightly different in adult zebrafish where macrophages/microglia persist caudal to the lesion up to 2 weeks after injury and neutrophils are only rarely observed ^{8, 13}.

Early work in the axolotl identified recruitment of macrophages or microglia to the lesion within 24 hours and persisted until 2 weeks post crush injury (Figure 2) ¹⁵¹. Whether these cells are necessary to promote regeneration is not known. A recent microarray analysis of axolotl spinal cord regeneration revealed that IL-1β is up-regulated transiently at 24 hours post injury, which appears to coincide with blood cell infiltration within the lesion ¹⁵². Whether these cells are macrophages, neutrophils, or some other cell type is not clear. While transient IL-1β signaling is crucial for zebrafish spinal cord regeneration, it is also important for fin regeneration suggesting an overall pro-regenerative effect ^{150, 153, 154}. It would be interesting to determine if early and transient up-regulation of IL-1β in axolotl plays a similar pro-regenerative role. Furthermore, the zebrafish JunB homologue junbb is necessary for injury-induced up-regulation of IL-1β ^{153, 154}. In axolotl, the non-canonical AP-1^cFos/JunB is up-regulated in glial cells after spinal cord injury and functions to promote a pro-regenerative response ²⁶. Whether this JunB-containing AP-1 complex similarly regulates IL-1β expression in axolotl glial cells is an interesting question.

5. The role of T cells during Regeneration

5.1. Effector lymphocytes during tissue repair

While a lot is known about how the innate immune system effects regeneration relatively little is known about the role of the adaptive immune system after injury ^{155, 156}. In general, it seems that the contribution of T cells to the regenerative response is highly context dependent. For example, a contusive model of brain injury in RAG^−/− mice, which lack T and B cells, showed that there is largely no effect on functional outcome ¹⁵⁷. However, it is worth mentioning that there appeared to be fewer neutrophils and macrophages around the lesion area. Additionally, inhibition of T cell trafficking to the lesion by injection of the sphingosine-1-phosphate receptor agonist FTY720 similarly had no effect on functional restoration ¹⁵⁸. However, after ischemia/reperfusion injury, RAG^−/− and severe combined immunodeficiency (SCID) mice exhibit better neurological recovery ^159-161. It has been proposed that after ischemia/reperfusion, canonical T cell cytokines like interferon-γ, IL-17, IL-21, and IL-23 could promote neuronal degeneration ¹⁵⁶. However, it is not clear why lymphocytes would be beneficial after ischemia/reperfusion but have no effect after contusive brain injuries. These potential discrepancies could be addressed using experimental approaches that specifically ablate B cells or subsets of effector and regulatory T cells.

A specific role for CD4 T cells in providing neuroprotective signals has been reported after injury and motor neuron disease. The ability of CD4 T cells to confer a protective effect is largely due to secretion of IL-4, which stimulates expression of various neurotrophic factors in local neurons ^{162, 163}. In this way, CD4 T cells delay axonal degeneration and mitigate neuronal apoptosis. Interestingly, many of the CD4 T cells that traffic to central nervous system lesions are autoreactive for particular antigens, like myelin-associated proteins ^164-167. Injection of mice with autologous dendritic cells pulsed with myelin components stimulated a strong adaptive immune response and improved functional recovery after spinal cord injury ^{168, 169}. While it is still not clear what the precise role of the adaptive immune system is after injury, it seems that therapeutic stimulation could elicit a pro-reparative response. Indeed, various approaches utilizing active and passive vaccination approaches are beginning to elucidate the molecular mechanisms that stimulate functional recovery after central nervous system injuries ¹⁷⁰.

It is also becoming clear that T cells play an important role during mammalian wound healing. The skin has a complement of tissue resident T cells, which play integral roles during immune surveillance for malignant cells, viral, and bacterial infections ¹⁷¹. Resident skin T cells are a major source of insulin-like growth factor-1 (IGF-1), which promotes wound healing after injury ¹⁷². Apart from typical effector T cells, a subset of specialized T cells that express γ and δ T cell receptor chains are important for wound healing. The use of genetic mouse models deficient for γδ T cells revealed that this cell population is a critical source of keratinocyte growth factor, which is necessary for normal wound healing ¹⁷³. Furthermore, wound beds deficient for γδ T cells have a higher proportion of myeloid-derived suppressor cells, fewer pro-regenerative M2 macrophages, and an altered cytokine profile ¹⁷⁴. However the functional outcome of these altered cellular dynamics to injury has not been reported. Additionally, whether γδ T cells play a similar role during injury response in the CNS is not clear but remains an interesting area of research.

The potential contribution of the adaptive immune system to tissue regeneration in non-mammalian models is almost completely unexplored. It is known that zebrafish and salamanders possess similar compliments of adaptive immune cells relative to humans. This suggests that these cells are present and could potentiate a pro-regenerative response to injury. As discussed above, complete and functional tissue regeneration in salamanders and fish, occurs despite a robust and well-orchestrated inflammatory response. These observations highlight that differences likely exist in which leukocytes arrive when, how long they stay, where in the injury they reside, and the specific cytokine profiles (pro- versus anti-inflammatory) they present.

5.2. Regulatory T cells during tissue repair

Regulatory T (Tregs) cells have recently been implicated as major regulators of mammalian tissue regeneration that generally seem to play a pro-regenerative or protective role. Tregs are defined by the expression of the transcription factor FoxP3 and the high affinity IL-2 receptor subunit CD25 ¹⁷⁵. Tregs play important roles in regulating effector cell functions during inflammatory responses, help to prevent autoimmunity, and have an emerging role in promoting tissue regeneration ^176-179. Similar to their role during immune responses, Tregs seem to promote tissue repair by directly regulating the inflammatory response to injury. For example, Tregs secrete TGF-β and IL-10, which mitigate the pro-inflammatory response to injury ^{180, 181}. However, Tregs also release molecules that elicit a pro-reparative response from resident cells in the lesion. Specifically, Tregs secrete the epidermal growth factor-like molecule Amphiregulin, which promotes lung, brain, and muscle repair ^182-184. Depletion of Tregs or conditional knockout of Amphiregulin in Tregs leads to a heightened pro-inflammatory response and decreased functional recovery ^182-188. Additionally, in mammals Tregs recruited to sites of demyelination directly regulate oligodendrocyte progenitor cell differentiation and re-myelination by secretion of cellular communication network factor 3 (CCN3)¹⁸⁹. Remarkably, Tregs directly regulate epithelial stem cell differentiation to stimulate hair follicle replacement after depilation ¹⁹⁰. Whether Tregs similarly influence epithelial stem cell differentiation after full thickness wounding remains to be seen. Adoptive transfer of Tregs into RAG^−/− mice or expansion of the Treg pool results in better functional recovery in a variety of injury models ^184-188. Interestingly, epidermal growth factor receptor signaling in Tregs is necessary for normal cutaneous healing after full thickness wounds ¹⁹¹. However, it is not clear if Amphiregulin secreted by Tregs then signals in an autocrine manner though EGFR to potentiate the pro-regenerative functions of Tregs. Whether Tregs in mammals exert a similar regulatory role over neural stem cells or reactive astrocytes after CNS injuries is poorly understood but remains an interesting and exciting question.

Similar to mammals, zebrafish Tregs (zTregs) promote functional regeneration after a variety of injury models. zTregs traffic to injury sites in the spinal cord, heart, and retina and function to elicit pro-regenerative response from local cells. Interestingly, the expression of pro-regenerative molecules was tissue dependent, where zTregs up-regulated neurotrophin-3 in the damaged spinal cord, insulin-like growth factor-1 in the retina, and neuregulin-1 in the heart¹⁹². Xenopus tadpoles are able to regenerate their tails after amputation, however there is a developmental stage (st45-47) where they temporarily lose this ability¹⁹³. One study suggests this refractory period coincides with the development of the immune system ¹⁹⁴. Pharmacologic and genetic perturbation in immune system development is sufficient to improve regenerative out comes ¹⁹⁴. Additionally, FoxP3 expression is highest during the post-refractory period suggesting that development of Tregs coincides with restoration of regenerative potential ¹⁹⁴. It would be interesting to use cell transplantation and cell ablation approaches to determine if a causal relationship exists between Treg development and the restoration of regeneration.

6. Perspective

The precise function of the immune system during tissue repair and regeneration remains enigmatic. It is clear that the immune system exerts a strong regulatory role after injury and it is not simply the presence or absence of an inflammatory response that dictates the reparative outcome (i.e. functional regeneration versus scarring). Instead, the cytokine profile and spatiotemporal dynamics of leukocyte recruitment to a lesion likely play integral roles in determining functional outcome.

For example, after spinal cord injury in larval zebrafish neutrophils arrive first and then macrophages infiltrate the lesion, a process which is similar to mammals ^{91, 103, 150}. However in zebrafish, both cell types are present in the lesion for a much shorter period of time than in mammals and express either IL-1β or TNF-α, respectively. In contrast, mammalian neutrophils and macrophages have both been reported to secrete IL-1β and TNF-α ^{91, 103}. Admittedly, leukocyte recruitment may be different after spinal cord injury in adult versus larval zebrafish, as neutrophils were not commonly found in the injured adult spinal cord ¹³. This is reminiscent of larval zebrafish brain regeneration where microglia but not neutrophils infiltrate the lesion, which is in contrast to mammalian brain injury where neutrophils arrive first ^{84, 141}.

In the future it will be important to conduct comparative studies across multiple species with varied regenerative abilities using standardized injury paradigms. The use of such approaches will provide valuable insight into the precise role that the immune system plays after injury. For example, the zebrafish is able to regenerate the heart after amputation but the medaka, a closely related species, is not ^195-197. Compared to zebrafish, the medaka exhibits a blunted acute inflammatory response to injury ¹⁹⁶. The inflammatory response was necessary for heart regeneration in the zebrafish and sufficient to stimulate a regenerative response in medaka ¹⁹⁶. Additionally, the African spiny mouse (Acomys spp.) is able to regenerate ear hole punches that the standard Mus musculus cannot ¹⁹⁸. Initial characterization of the inflammatory response to ear hole punch in Acomys and Mus revealed that both organisms undergo an inflammatory response to injury. However, the timing of leukocyte infiltration, the macrophage polarization state, and specific leukocyte localization in the lesion is different between the two species and likely contributes to the differences in regenerative outcome ¹⁹⁹. Collectively, these studies highlight the importance of comparative analysis across closely related species with varied regenerative capabilities. As genome editing and conditional transgenesis techniques continue to improve, such comparative studies will become more feasible and shed light on the complex role of the immune system during tissue repair and regeneration.