Remaking a connection: molecular players involved in post-injury synapse formation

Functional recovery from central nervous system (CNS) trauma depends not only on axon regeneration or compensatory sprouting of uninjured fibers but also on the ability of newly grown axons to establish functional synapses with appropriate targets. Although several studies have successfully promoted long-distance axonal regeneration in distinct CNS injury models, none of them have resulted in a viable therapeutic approach for patient recovery. A possible reason may be the lack of new synaptogenesis for re-establishing the circuitry lost after injury. Herein, we discuss how our understanding of the mechanisms that instruct synapse formation in the injured nervous system may contribute to the design of new strategies to promote functional restoration in traumatic CNS disorders.

Return to a growth-competent state: In contrast to the adult CNS, embryonic or peripheral neurons show a remarkable ability to regenerate their axons after injury. They accomplish that by implementing a pro-regenerative intracellular program that allows them to upregulate genes necessary for axon elongation (Winter et al., 2022). At the same time, regeneration-competent neurons also downregulate critical components of the synapse, such as the presynaptic active zone proteins RIM1/2 and Munc13, that are inhibitory to axon growth (Hilton et al., 2022). This suggests that the injured axon must disassemble some of the presynaptic specializations that are established during development to regenerate. Although adult CNS neurons do not spontaneously grow their axons after injury, they partially revert their transcriptomes towards a growth-competent state. Corticospinal tract (CST) neurons return to an embryonic transcriptional state after spinal cord injury in mice. However, they are only capable of sustaining this regenerative state for two weeks and therefore fail to regenerate (Poplawski et al., 2020). Nevertheless, several approaches have succeeded in reprogramming mature CNS neurons back to a growth state and inducing robust axonal regeneration via transcriptome and epigenome modifications, enhancing protein synthesis and mitochondrial trafficking, and augmenting the responsiveness of injured neurons to growth factors (Winter et al., 2022). However, the underlying mechanisms accounting for the lack of repair in different animal models of CNS injury remain largely unknown. In some cases, the regenerative phenotype does not translate into significant behavioral gains, raising the question of whether regenerating axons have the ability to re-establish lost synaptic contacts. Since injured neurons must downregulate genes related to synaptic function in order to carry out a regrowth program, can the regenerated axon differentiate back into a presynaptic specialization in the adult CNS? In the next sections, we will try to answer that question by looking at the molecular mechanisms described to promote circuit remodeling in the injured CNS.

Post-injury synapse formation in the central nervous system: The case of fibroblast growth factor 22: During development, a growing axon projects to distant target regions, where it will establish the correct synaptic contacts to construct neuronal circuits. The formation of presynaptic sites is instructed either by cell adhesive contacts with the postsynaptic partner or by soluble secreted factors that can have a neuronal or glial origin. These synaptogenic molecules promote the clustering of presynaptic active zone proteins and neurotransmitter-containing synaptic vesicles along the axon, establishing the sites where new synapses will be formed. Among these molecules, the family of fibroblast growth factors (FGFs) and their receptors play a significant role in regulating presynaptic differentiation. One member in particular, FGF22, acts as a soluble presynaptic organizer to coordinate excitatory synapse formation in the developing hippocampus (Pinto and Almeida, 2016). Interestingly, both FGF22 and its receptors FGFR1 and FGFR2 are expressed in the adult spinal cord and are required for the formation of an intraspinal detour circuit that circumvents the lesion site and allows some functional restoration in incomplete spinal cord lesions. Following a thoracic dorsal hemisection, severed CST axons sprout into the gray matter of the cervical spinal cord and make new synaptic contacts with long propriospinal neurons (LPSN). These, in turn, project to lumbar motor neurons, allowing the functional communication between the motor cortex and the lumbar spinal cord (Figure 1A). Genetic ablation of FGF22 or the targeted deletion of FGFR1 and FGFR2 in the mouse motor cortex, prevents the formation of this detour circuit, compromising functional recovery (Jacobi et al., 2015). These findings indicate that spinal relay neurons release FGF22 to induce presynaptic differentiation in injured CST axons (Figure 1A), suggesting that the mechanisms that instruct synaptogenesis during development remain operational in adulthood. Besides FGF22, other synaptogenic molecules and even guidance cues that coordinate developmental neural circuit formation were identified in the adult CNS. For instance, several members of the slit, semaphorin, synCam, neuroligin, and ephrin families were detected in the adult spinal cord, while their receptors were found to be expressed by cortical neurons (Jacobi et al., 2014). Together, these findings indicate that the adult CNS has the means to spontaneously reorganize some axonal circuits after injury and thereby promote functional recovery. However, can this endogenous capacity be enhanced to improve circuit rewiring and mitigate the severe functional consequences of CNS trauma? Overexpression of FGF22 in cervical spinal neurons through virally-mediated gene delivery increases the number of presynaptic boutons in CST axons that sprout into this region after a thoracic hemisection in mice. This synaptogenic gene therapy not only resulted in an increased number of CST contacts onto relay LPSN but also in additional excitatory interneurons, enhancing CST connectivity in the cervical spinal cord and thereby motor recovery in injured animals (Aljović et al., 2023). Despite its complexity, the presynaptic terminal forms fairly quickly, within minutes to a few hours (Pinto and Almeida, 2016). Thus, the timing of synaptogenic gene therapies must be considered in order to produce the desired effects on functional recovery. In fact, FGF22 gene therapy only resulted in significant behavioral gains when performed within the first 24 hours post-injury. A delayed intervention (5 days post-injury) failed to induce functional recovery in injured animals, probably because it passed the period when CST-LPSN detour circuits were formed (Aljović et al., 2023). This reveals a critical therapeutic window for synaptogenic strategies that aim to enhance the remodeling of axonal circuits in the injured CNS. It is noteworthy to mention that increased synaptic contacts were only observed in neurons that overexpressed FGF22, probably because secreted FGFs interact with heparan sulfate proteoglycans present in the extracellular matrix that limit their diffusion. This suggests the possibility that gene therapy with this presynaptic organizer can be used to selectively target the rewiring process of specific neuronal circuits. For instance, can FGF22 overexpression in spinal neurons located past the injury site improve the synaptic reintegration of regenerating CST axons and thus the recovery of motor function? Paraphs, a combined strategy that enhances the intrinsic growth capacity of CNS neurons and then the presynaptic differentiation of their regenerated axons in target regions, should be considered in the future.

Figure 1.

Molecular mechanisms driving post-injury synapse formation.

(A) Role of FGF22 in post-injury remodeling of spinal cord circuits. In response to a thoracic hemisection, injured CST axons sprout into the cervical spinal cord and establish synapses with LPSN that act as relays to lumbar motor circuits. The formation of this detour circuit is highly dependent on FGF22 signaling. Spinal interneurons, including LPSN, release FGF22 that acts through FGFR1 and FGFR2 to induce presynaptic differentiation in CST axons. (B) The dual role of astrocytes in synaptic repair. Reactive astrocytes in the distal part of the lesion secrete TSPs and hevin that instruct post-injury synapse formation via interaction with pre- or postsynaptic α2δ-1 and by modulating neurexin/neuroligin adhesion, respectively. Close to the injury site, reactive astrocytes can adopt a neurotoxic phenotype due to the action of pro-inflammatory mediators released by activated microglia. Created with Adobe Illustrator. C1q: Complement component subunit 1q; CST: corticospinal tract; FGF: fibroblast growth factor; LPSN: long propriospinal neurons; NL: neuroligin; Nrx: neurexin; PSD: post-synaptic density; SV: synaptic vesicle; TNF: tumor necrosis factor; TSP: thrombospondin.

Astrocyte-derived factors: Astrocytes are critical players involved in the formation of neuronal circuits during development. They coordinate synapse formation either by secreting synaptogenic factors such as thrombospondins (TSPs), transforming growth factor β1 or hevin, or through contact-dependent mechanisms via the expression of transmembrane γ-protocadherin (reviewed in Pinto and Almeida, 2016). Upon injury, astrocytes become reactive and hypertrophy. The degree of astrogliosis depends on the number of cell surface receptors for damage-associated molecular patterns, the levels of proinflammatory cytokines and chemokines, and the distance of astrocytes from the lesion core. For instance, close to the injury site, astrocytes are newly proliferated, hypertrophic and intertwine with their processes to form a barrier-like structure around the lesion center, commonly known as the astrocytic scar. In contrast, reactive astrocytes in the distal part of the lesion are hypertrophic but nonproliferative and maintain their basic structure and individual processes, contributing to the remodeling of neuronal circuits mainly through the release of TSPs (Figure 1B; Liauw et al., 2008; Bradbury and Burnside, 2019). These synaptogenic factors are large oligomeric extracellular matrix glycoproteins that induce excitatory synapse formation in different types of neurons through interaction with the voltage-gated calcium channel subunit α2δ-1, which is expressed by neurons both pre- and postsynaptically. All TSP isoforms (TSP1-TSP5) are synaptogenic (Risher and Eroglu, 2020). Following ischemic stroke in mice, TSP1 and TSP2 are re-expressed and secreted by reactive astrocytes in the ischemic cortical penumbra to induce the remodeling of lost axonal circuits. These two TSPs are not only responsible for instructing new synaptogenesis in this area but also for promoting axonal sprouting from inputs originating in the contralateral cortex. In fact, genetic deletion of both TSPs compromises this synaptic remodeling, leading to worse functional outcomes (Liauw et al., 2008). Upon facial nerve transection in mice, reactive astrocytes in the facial nucleus also up-regulate and release TSP1 in a STAT3-dependent manner, which in turn promotes the recovery of excitatory synaptic input onto surviving motor neurons (Tyzack et al., 2014). These results indicate that astrocyte-derived TSPs are critical modulators of post-injury synapse formation in the adult CNS and prompt the use of TSPs in future synaptogenic therapies for traumatic diseases. However, it is important to note that, unlike FGF22, the role of TSPs as presynaptic organizers is not completely clear. Depending on the neuronal population, TSPs can act on pre- or post-synaptic α2δ-1 to induce synaptogenesis (reviewed in Risher and Eroglu, 2020). As a result, the outcomes of synaptogenic therapy based on FGF22 and TSPs may differ since they appear to act on different neuronal subcompartments.

Astrocytes may also induce post-injury synaptic remodeling by releasing hevin. The levels of this synaptogenic factor increase in the hippocampus of rats after traumatic brain injury and are further amplified by blocking the astrocytic calcineurin/nuclear factor of activated T cells signaling pathway, implying that intrinsic factors may restrain the ability of astrocytes to promote post-injury synaptogenesis. In fact, inhibition of this pathway in astrocytes helps to restore some hippocampal synaptic function and plasticity in injured animals (Furman et al., 2016). However, it is unclear whether the observed synaptic recovery is primarily caused by the release of hevin or if other astrocyte-derived synaptogenic factors are also involved. Finally, it is relevant to mention that astrocytes have been described to play a double-edged role in the repair of the injured CNS. This most likely reflects the diverse post-injury phenotypes that astrocytes can adopt. For instance, pro-inflammatory microglia induce a neurotoxic reactive astrocyte phenotype through the release of IL-1α, tumor necrosis factor, and complement component subunit 1q. Instead of supporting neuronal survival, axonal regrowth, and synaptogenesis, these neurotoxic astrocytes induce the death of neurons and oligodendrocytes, and express several genes of the classical complement cascade known to be synaptotoxic (Liddelow et al., 2017; Figure 1B). Moreover, both reactive astrocytes and microglia have been described to engulf synapses after CNS trauma. Blocking this glia-mediated synapse elimination increases the levels of synaptic proteins and dendritic spine density in a mouse model of traumatic brain injury, leading to better functional outcomes (Shen et al., 2023). Thus, neuroinflammation can negatively impact synaptic remodeling by interfering with the reactive state of astrocytes. Strategies aiming at resolving neuroinflammatory processes in the injured CNS may enhance the ability of astrocytes to induce post-injury synapse formation, improving circuit rewiring and functional recovery.

Conclusions: Overall, these findings indicate that the developmental mechanisms that instruct synaptogenesis in the CNS are reactivated after injury to promote the remodeling of axonal circuits. Severed axons can differentiate back into functional presynaptic units using the same instructive signals as developing axons. Moreover, this endogenous capacity for synaptic repair can be enhanced by the local administration of synaptogenic molecules, improving circuit rewiring and functional recovery. Although research into post-injury synapse formation is still in its early stages, it opens the possibility of the application of synaptogenic therapy to repair the injured CNS. For instance, can a synaptogenic therapy based on the use of presynaptic organizers improve the synaptic reintegration of experimentally induced regenerated axons and thereby functional restoration? These concerns should be taken into consideration in future therapeutic strategies for traumatic CNS diseases.

This work was supported by “la Caixa” Foundation (ID 100010434) and FCT-Fundação para a Ciência e a Tecnologia, I.P. under the agreement LCF/PR/HP20/52300001 and by FCT, I.P., under projects PTDC/NAN-OPT/7989/2020, UIDB/04501/2020, UIDP/04501/2020, UIDB/04539/2020, UIDP/04539/2020 and LA/P/0058/2020 and through the individual grant SFRH/BD/139368/2018 (DT).

Additional file: Open peer review report 1 ^{(78.8KB, pdf)} .