Blog

By Xinli Xu, Rui O. Beleza, Francisco Q. Gonçalves, Sergio Valbuena, Sofia Alçada-Morais, Nélio Gonçalves, Joana Magalhães, João M. M. Rocha, Sofia Ferreira, Ana S. G. Figueira, Juan Lerma, Rodrigo A. Cunha, Ricardo J. Rodrigues, and Joana M. Marques

Excerpt from the article published in Scientific Reports 12, 14690, 29 August 2022, DOI: https://doi.org/10.1038/s41598-022-18884-4

Editor’s Highlights

• The current knowledge on the mechanisms driving circuit re-wiring in the mature central nervous system is limited.
• An example of such re-wiring processes in the adult brain is the hippocampal mossy fiber (MF) sprouting found in patients and in animal models of temporal lobe epilepsy (TLE).
• Other disorders displaying also circuit alterations are neuropsychiatric disorders affecting adults and prevalent on ageing.
• Adenosine A_2A receptors (A_2AR) control axon formation/outgrowth and synapse stabilization during development.
• The overactivation of A_2ARs is a molecular mechanism underlying synaptic remodeling in the injured brain, namely SE-induced hippocampal MF sprouting.
• The A_2AR-mediated mechanism seems intrinsic to neurons rather than involving astrogliosis and inflammation, which are also characteristic features of epilepsy that are associated with the development of seizures and with the ability to contribute to synaptic remodeling.
• A_2ARs contribute to epilepsy-related circuit remodeling, which may involve the reactivation of the ability of A_2ARs to control axon formation and outgrowth during development.

Abstract

The molecular mechanisms underlying circuit re-wiring in the mature brain remains ill-defined. An eloquent example of adult circuit remodelling is the hippocampal mossy fiber (MF) sprouting found in diseases such as temporal lobe epilepsy. The molecular determinants underlying this retrograde re-wiring remain unclear. This may involve signaling system(s) controlling axon specification/growth during neurodevelopment reactivated during epileptogenesis. Since adenosine A_2A receptors (A_2AR) control axon formation/outgrowth and synapse stabilization during development, we now examined the contribution of A_2AR to MF sprouting. A_2AR blockade significantly attenuated status epilepticus(SE)-induced MF sprouting in a rat pilocarpine model. This involves A_2AR located in dentate granule cells since their knockdown selectively in dentate granule cells reduced MF sprouting, most likely through the ability of A_2AR to induce the formation/outgrowth of abnormal secondary axons found in rat hippocampal neurons. These A_2AR should be activated by extracellular ATP-derived adenosine since a similar prevention/attenuation of SE-induced hippocampal MF sprouting was observed in CD73 knockout mice. These findings demonstrate that A_2AR contribute to epilepsy-related MF sprouting, most likely through the reactivation of the ability of A_2AR to control axon formation/outgrowth observed during neurodevelopment. These results frame the CD73-A_2AR axis as a regulator of circuit remodeling in the mature brain.

Introduction

The current knowledge on the mechanisms driving circuit re-wiring in the mature central nervous system is limited. Yet, this knowledge is fundamental to understand and correctly promote nerve regeneration, control stem cell grafting or adult-born neurons integration or prevent pathological circuit remodeling. A very well-established example of such re-wiring processes in the adult brain is the hippocampal mossy fiber (MF) sprouting found in patients and in animal models of temporal lobe epilepsy (TLE)^1,2.

The retrodirective growth of MF into the inner molecular layer of the dentate gyrus (DG) ^3,4seems to be due to a dysregulation of guidance cues involved in the establishment of normal MF connectivity during development^5,6,7,8 but the underlying molecular determinants are still unclear. BDNF has been associated to this initial axonal branching^9,10,11 but other studies observed MF sprouting in the absence of BDNF¹² or showed that BDNF is not sufficient to induce MF sprouting^13,14. which critically requires abnormal circuit activity¹⁵.

One unexplored set of candidates to regulate mature circuit remodeling, in particular hippocampal MF sprouting, are purines; purines are released in an activity-dependent manner¹⁶ and can control neuronal migration and wiring during development, particularly through adenosine A_2A receptors (A_2AR). ATP is preferentially released from MF terminals at high-frequency stimulations^17,18 and we recently reported that A_2ARs control axon formation and outgrowth of migrating cortical projection neurons during embryogenesis¹⁹. being also involved in synaptic stabilization during development²⁰. Besides, A_2ARs are located in hippocampal MF terminals¹⁸. contribute to synaptic failure and transient decrease in cell proliferation in the DG upon oxygen–glucose deprivation²¹. are active during epileptogenesis^18,22,23,24 and their pharmacological blockade or genetic deletion prevent seizure-induced neurodegeneration²³ and attenuate the progressive seizure severity in different models of kindling^23,25.

In this study, we provide evidence that A_2ARs, in particular those located in dentate granule cells are activated by ATP-derived adenosine and contribute to status epilepticus (SE)-induced hippocampal MF sprouting, most likely by inducing axon formation and outgrowth. This demonstrates a reactivation of adenosine-induced A_2AR-mediated control of brain wiring in the adult brain, impacting on circuit remodeling after SE.

Results

Pharmacological activation of A_2ARs induces the formation of secondary aberrant axons in cultured hippocampal neurons

To start addressing if A_2ARs control axon formation and/or outgrowth of hippocampal neurons, as previously observed in cortical neurons¹⁹ we first evaluated the impact of the pharmacological manipulation of A_2ARs in axon formation/outgrowth of cultured E18-rat derived hippocampal neurons. We observed that the pharmacological activation of A_2ARs with the selective agonist CGS21680 (30 nM) from DIV0 caused an increase in the number of axons per neuron, reflecting an increase in the percentage of neurons with more than one axon, analyzed at DIV3 (Fig. 1a,b). CGS21680 also increased axonal length without affecting axonal branching (Fig. 1c). The selective antagonist of A_2AR, SCH58261 (50 nM), did not significantly modify neither axon formation nor axonal elongation (Fig. 1b,c), showing that the tonic activation of A_2ARs by endogenous adenosine is absent or does not affect axon formation/outgrowth. Accordingly, the knockdown of A_2ARs, using a validated short-hairpin RNA expressing a construct against A_2AR (shRNA-A_2AR), neither significantly modified the number of axons per neuron nor axon elongation, analyzed at DIV3 (Fig. 1d and Supplementary Figs. 1, 2). The increase in the number of axons per neuron and axon length caused by CGS21680 are mediated by the activation of A_2ARs since it was not observed in the presence of SCH58261 (Fig. 1b,c) and was observed in cells electroporated with a short-hairpin RNA expressing a construct against a non-targeting control (shRNA-Control) but not in cells expressing shRNA-A_2AR (Fig. 1d).

These results show that the pharmacological activation of A_2ARs promotes axonal elongation and induces the formation of secondary abnormal axons in developing rat hippocampal neurons. Accordingly, we observed an immunoreactivity for A_2AR in non-polarized cells as well as in axonal growth cones (Fig. 1e). These findings in hippocampal neurons converge with a previous report in cortical neurons¹⁹, suggesting that A_2ARs control axogenesis in different neuronal populations.

A_2ARs contribute to status epilepticus-induced hippocampal mossy fiber sprouting

Hippocampal organotypic slices display spontaneous MF sprouting in the molecular layer of the DG^26,27,28 supporting an enhanced excitation of dentate granule cells²⁹. In hippocampal organotypic slices prepared from P5-7 mice, stimulation of MF at DIV7 elicited synaptic currents in dentate granule cells (Fig. 2). In slices cultured in the presence of the A_2AR antagonist SCH58261 (50 nM), the current density was not significantly modified (Fig. 2), indicating that the formation of these recurrent excitatory inputs at the dentate granule cells in hippocampal organotypic slices is unrelated to the activation of A_2ARs by endogenous adenosine.

Next, we evaluated the contribution of A_2ARs to hippocampal MF sprouting induced by SE. We resorted to the pilocarpine-induced model of SE in rats, which triggers a robust hippocampal MF sprouting within weeks to months after SE³⁰. Forty-two days after SE induction, a positive Timm staining demonstrated MF sprouting in the granule cell and molecular layers of the DG (Fig. 3a,b). Only pilocarpine-treated animals reaching at least stage 4 seizures were considered in this study. To appraise the contribution of A_2ARs, rats were injected daily with SCH58261 (0.1 mg/kg, i.p.) or vehicle, beginning 10 h after the onset of SE (Fig. 3a). The extent of MF sprouting was assessed by estimating the percentage of the granule cell layer plus molecular layer that presented Timm staining (Fig. 3b). In control rats that had not experienced SE, non-significant levels of Timm staining were observed in the granule cell layer and inner molecular layer either in vehicle or SCH58261-treated rats (Fig. 3c,d). In contrast, animals that had developed SE and had been administered with vehicle for 42 days, displayed a significantly denser Timm staining in the entire granule cell layer and inner molecular layer of the DG (P < 0.0001). In rats administered with SCH58261, SE still induced a significantly higher Timm staining than in control (P = 0.013 SE-SCH58261 vs.Control-SCH58261), but considerably lower than the observed in vehicle-injected rats subjected to SE (P = 0.039 for SE vs. SCH58261 interaction; P = 0.004 for SE-vehicle vs. SE-SCH58261) (Fig. 3c,d). SE-induced MF sprouting was observed along the entire septotemporal axis (Fig. 3e) both in control and, albeit with significantly lower levels, in SCH58261-treated animals (Fig. 3e).

These results indicate that A_2ARs contribute to SE-induced hippocampal MF sprouting.

A_2ARs expressed by dentate granule cell neurons contribute to mossy fiber sprouting

If the contribution of A_2ARs to SE-induced hippocampal MF sprouting is due to a direct control of axon formation/outgrowth, as observed in cultured hippocampal neurons, A_2ARs expressed in dentate granule cells should mediate this effect. Thus, we evaluated the impact of knocking down A_2ARs in dentate granule cells using lentivectors encoding the reporter gene EGFP and either shRNA-A_2AR or shRNA-Control (Refs 31, 32 and Supplementary Figs. 1, 2).

Each rat was stereotaxically injected into the left or right DG with either shRNA-Control or shRNA-A_2AR and was allowed to recover for 10 days before SE induction (Fig. 4a). Only animals reaching at least stage 4 seizures were considered in this study, and they were sacrificed 56 days after SE for the evaluation of MF sprouting in the inner molecular layer. This was performed by immunohistochemical identification of sprouted MF puncta by immunolabelling synaptoporin (Fig. 4b), a synaptic vesicle membrane protein highly enriched in MF terminals^33,34,35. The lentivectors used have been previously shown to have a limited spread within the brain parenchyma³⁶ infecting only neurons³². Accordingly, EGFP⁺-cells were all endowed with NeuN⁺-immunolabelling (Fig. 4c).

Control animals showed no MF sprouting, evidenced by the lack of synaptoporin staining in the inner molecular layer, either in the DG injected with shRNA-Control or shRNA-A_2AR (Fig. 4d). The rats that experienced SE displayed synaptoporin labelling in the granule cell and inner molecular layers (Fig. 4d). To estimate the impact of the knockdown of A_2ARs in dentate granule cells in MF sprouting, we determined a Δsprouting ratio to evaluate the correlation between the density of infected granule cells (EGFP⁺) and sprouted MF puncta density in the adjacent inner molecular layer for each image acquired from animals that had experienced SE (Fig. 4e). We found a negative correlation between the density of synaptoporin puncta in the inner molecular layer and granule cell layer and EGFP⁺-cell density in the adjacent granule cell layer region in shRNA-A_2AR-injected DG, but not in shRNA-Control-injected DG (Fig. 4e,f). In the analysis of individual animals, the Δsprouting ratio of shRNA-A_2AR-injected DG was significantly different from shRNA-Control-injected DG in 4 out of 10 animals (Fig. 4f left graph). An overall analysis of all the animals that developed SE revealed that in shRNA-Control-injected DG, the Δsprouting ratio was not different from zero, showing no impact of shRNA-Control in synaptoporin puncta density. In contrast, shRNA-A_2AR-injected DG displayed a significantly negative Δsprouting ratio (P < 0.0001 vs. hypothetical value of 0) and statistically different from the Δsprouting ratio of shRNA-Control (P = 0.0001; paired t-test). These data show that the knockdown of A_2ARs in granule cells attenuates synaptoporin puncta density in the inner molecular layer and granule cell layer, demonstrating that A_2ARs located in dentate granule cells contribute to SE-induced MF sprouting.

Impact of the pharmacological blockade of A_2ARs in SE-induced cell proliferation

MF sprouting into the inner molecular layer occurs in the absence of neurogenesis³⁷. Yet, seizures induce the generation of new dentate granule cells from the subgranular zone^38,39,40 and these post-SE born cells also contribute further to MF sprouting, in particular at later stages^41,42. Indeed, cells born after SE display hilar basal dendrites and ectopic migration, but do not send axons into the inner molecular layer until 4 weeks after the insult ^41,42 and only have a substantial contribution to MF sprouting 10 weeks after SE⁴¹.

A_2ARs may promote adult neurogenesis particularly in pathological conditions^43,44. Hence, we sought to determine if, in addition to the contribution of A_2ARs located in dentate granule cells born before SE, A_2ARs may also further promote hippocampal MF sprouting by contributing to the generation of newborn cells. To address this, we tested whether the daily administration of the selective A_2AR antagonist SCH58261 (0.1 mg/kg, i.p.), beginning 10 h after the onset of SE, affected SE-induced cell proliferation in the granule cell layer. To label mitotically active cells, rats were injected with BrdU at 24 h (200 mg/kg, i.p.), 7 days (200 mg/kg, i.p.) and 8 days (100 mg/kg, i.p.) after SE (Fig. 5a), the latter two time points being within the period of significantly increased mitosis³⁷. Consistent with previous reports^38,45,46, quantitative analysis revealed a significant increase in the number of BrdU⁺-cells in the granule cell layer in SE rats treated with vehicle for 42 days in comparison with control rats (P < 0.001; Fig. 5b,c). The vast majority of BrdU⁺-cells were co-localized with the mature neuronal marker NeuN (Fig. 5b), in agreement with previous reports describing that the majority (~ 90%) of the newly generated cells within the granule cell layer after SE differentiate into neurons^38,45,46. In control rats, SCH58261 administration per se did not significantly modify the density of BrdU⁺-cells in the granule cell layer, showing that A_2ARs do not contribute to basal adult cell proliferation originated from the subgranular zone. Regarding the impact of SCH58261 on SE-induced increase in BrdU⁺-cells in the DG, we found an interaction between SE and SCH58261 (P = 0.0428; two-way ANOVA). Yet, post-hoc analysis showed that SE-vehicle and SE-SCH58261 groups were not significantly different (P = 0.1492; Fig. 5c). On the other hand, in rats daily injected with SCH58261, SE did not induce a statistically significant increase in BrdU⁺-cells (P = 0.9722 for SE-SCH58261 vs. Control-SCH58261).

The newborn neurons generated in the subgranular zone migrate radially into the granule cell layer, mainly integrating functionally into the existing circuitry in the inner third of the granule cell layer^47,48. Following an epileptogenic insult, there is a misplacement of a significant number of cells that migrate abnormally to the outer two-thirds of granule cell layer^38,40. Accordingly, we now observed in control rats that most of the BrdU⁺-cells were located in the inner one-third of the granule cell layer and only a small portion of BrdU⁺-cells were located in the outer two-thirds of the granule cell layer (4.82 ± 0.73%; n = 7) (Fig. 5d). SCH58261 administration per se did not modify this pattern in control rats. However, SE rats treated with vehicle for 42 days displayed a significant increase in the relative percentage of BrdU⁺-cells in the outer two-thirds of the granule cell layer (8.66 ± 0.77%, n = 12), compared to controls (Fig. 5d). In contrast to the observed in SE-induced increase in BrdU⁺-cells density (Fig. 5c), SCH58261 administration did not modify the SE-induced spatial distribution of adult newborn cells in the granule cell layer (P = 0.3058 for SE x SCH58261 interaction; two-way ANOVA) (Fig. 5d). These data show that A_2ARs are not involved in SE-induced aberrant migration of adult-born cells within the granule cell layer.

Status epilepticus-induced hippocampal mossy fiber sprouting is engaged by extracellular ATP-derived adenosine

ATP is preferentially released at high-frequency stimulations and A_2ARs are mainly activated by ATP-derived adenosine through the catabolism of extracellular ATP involving ecto-5’-nucleotidase (CD73) activity in MF terminals^17,18,19. CD73, which catabolizes extracellular AMP into adenosine, is present in sprouted MF in epileptic rats²⁰ and a SE-induced increased activity of CD73 has been reported in the pilocarpine rat model of temporal lobe epilepsy^49,50. CD73 density is also enhanced in the inner molecular layer of the DG from human temporal lobe epilepsy patients⁵¹. We now measured a sustained increase of the evoked release of ATP from hippocampal nerve terminals prepared from rats 7 days after pilocarpine-induced SE in comparison with control rats, but not at 24 h or 3 days post-SE (Fig. 6a). This is coincident with the increased CD73 activity found in hippocampal nerve terminals from the same rat model of temporal lobe epilepsy⁵¹. This prompted the hypothesis that the adenosine triggering the observed A_2AR’s contribution to SE-induced hippocampal MF sprouting derives from the extracellular catabolism of ATP in a CD73-dependent manner. To test this hypothesis, SE-induced MF sprouting was evaluated in CD73-KO mice. The latency to SE onset and the seizure score during SE were similar between wild-type and CD73-KO mice (Fig. 6b). In wild-type mice, SE induced a visible MF sprouting evaluated 105 days later (Fig. 6c). However, in CD73-KO mice, SE induced a significantly lower percentage of the granule cell layer plus molecular layer displaying Timm staining (P = 0.0129 for SE-wild-type vs. SE-CD73-KO), which was not statistically different from control (P = 0.1383 for SE-CD73-KO vs. Control-CD73-KO) (Fig. 6d). This significantly lower SE-induced MF sprouting in CD73-KO mice was also observed along the entire septotemporal axis (Fig. 6e), as observed with A_2AR blockade.

Altogether, these results indicate that the adenosine triggering A_2AR-mediated MF sprouting induced by SE is derived from the CD73-mediated extracellular catabolism of ATP.

Discussion

The present study demonstrates that the overactivation of A_2ARs is a molecular mechanism underlying synaptic remodeling in the injured brain, namely SE-induced hippocampal MF sprouting. While A_2ARs did not contribute to the generation of recurrent excitatory inputs spontaneously developed in dentate granule cells of hippocampal organotypic slices (Fig. 2), the pharmacological blockade of A_2ARs significantly attenuated SE-induced hippocampal MF sprouting in a rat pilocarpine model (Fig. 3). This involves A_2ARs located in dentate granule cells since the knockdown of A_2ARs selectively in dentate granule cells reduced the density of SE-induced MF terminals in the inner molecular layer (Fig. 4). Moreover, the observation of an SE-induced increase in ATP release from hippocampal terminals, together with the prevention/attenuation of hippocampal MF sprouting in CD73-KO mice (Fig. 6), strongly suggests that A_2AR-driven MF sprouting is triggered by extracellular ATP-derived adenosine.

We recently reported that A_2ARs control neuronal polarization and axon formation of cortical projection neurons during embryogenesis¹⁹. In the present study, we found that A_2ARs also control axon formation in developing rat hippocampal neurons (Fig. 1). However, in hippocampal neurons, A_2ARs did not control the formation of the “normal” axon; instead, the pharmacological activation of A_2ARs induced the formation of aberrant secondary axons (Fig. 1). Moreover, we now demonstrated the involvement of A_2ARs in SE-induced hippocampal MF sprouting (Fig. 3). This contribution of A_2ARs to MF sprouting was only observed in pathological conditions, i.e. in SE-induced MF sprouting, but not in the recurrent excitatory inputs that develop spontaneously in hippocampal organotypic slices (Fig. 2). This indicates that spontaneous MF recurrent sprouting in hippocampal organotypic slices is mechanistically distinct from epilepsy-related hippocampal MF sprouting and argues that an increased activity of A_2ARs after SE²³ is required to trigger the abnormal axonal sprouting after SE. The observation that the knockdown of A_2ARs in granule cells was sufficient to reduce MF sprouting (Fig. 4), together with the ability of A_2ARs’ activation to induce the formation of abnormal secondary axons in cultured rat hippocampal neurons (Fig. 1) strongly suggests that A_2ARs contribute to hippocampal MF sprouting by directly promoting axon formation/outgrowth selectively during pathological conditions in the adult brain.

This newly described A_2AR-mediated mechanism seems intrinsic to neurons rather than involving astrogliosis and inflammation, which are also characteristic features of epilepsy^52,53that are associated with the development of seizures⁵⁴ and with the ability to contribute to synaptic remodeling⁵⁵. Although both astrogliosis and neuroinflammation are also controlled by A_2ARs¹⁷, it was previously reported that blocking A_2ARs abrogates SE-induced neurodegeneration without affecting astrogliosis²⁴ and it was now observed that the selective elimination of A_2ARs in hippocampal neurons was sufficient to prevent SE-induced MF sprouting. Thus, the proposed conclusion that A_2AR-mediated control of SE-induced MF sprouting seems to involve an intrinsic neuronal remodeling by the adenosine neuromodulation system, seems the most parsimonious explanation, which may entail a direct regulation of cytoskeleton, as observed in developing rat cortical neurons⁵⁶. Nevertheless, MF sprouting has been also shown to be correlated with cell loss, although it is not mandatory⁵⁷. In particular, the extension of hippocampal MF sprouting has been correlated with the loss of hilar inhibitory interneurons and mossy cells⁵⁷ found both in experimental models and in patients with temporal lobe epilepsy^{58,59,60,61,62,63}. Another study also reported a positive correlation between MF sprouting and cell loss both CA3 and CA1⁶⁴. As mentioned, the antagonism of A_2ARs prevents SE-induced neurodegeneration²³. This has been associated to an up-regulation of neuronal/synaptic A_2AR-driven increase in excitatory drive leading to calpain-mediated neurodegeneration²³. More recently, it was observed an absence of neurodegeneration upon SE in CD73-KO mice²⁴. The involvement of A_2ARs and of CD73 in hilar neurons death was not detailed^23,24. Nevertheless, there is the possibility of an additional indirect contribution of the CD73-A_2AR axis to hippocampal MF sprouting by controlling hippocampal cell damage.

This study determined that the contribution of A_2ARs to SE-induced hippocampal MF sprouting here identified entails, at least in part, A_2ARs located in dentate granule cells generated before SE (Fig. 4). But newly born cells also contribute to the abnormal retrodirective innervation of the inner molecular layer^38,41,42. There are evidences suggesting for an inhibitory action of A_2ARs in neurogenesis in postnatal and adult rodent brain, in particular in cell proliferation⁴⁴. A_2ARs were shown to contribute to the transient decrease in cell proliferation from the subgranular zone upon oxygen–glucose deprivation, most likely by contributing to the damage of immature cells²¹. On the other hand, it was recently reported that the activation of A_2ARs prevents the reduction of newborn dentate granule cells as a consequence of hearing loss induced by noise exposure⁴³. Here, we observed that the pharmacological blockade of A_2AR per se did not alter adult-born cell generation from the subgranular zone (Fig. 5). However, we found a statistically significant interaction between SCH58261 and SE in cell proliferation, but not in their positioning within the granule cell layer (Fig. 5). This suggests that SE-induced dentate granule cell proliferation, but not their migration, entails, either directly or indirectly, the activation of A_2ARs. Hence, A_2ARs may further contribute and sustain MF sprouting by promoting the generation of adult born cells induced by epileptogenic injury. The newborn cells only contribute to the abnormal innervation of the inner molecular layer upon their full maturation within 4 weeks^41,42, although immature granule cells are able to extend axons to CA3^65,66. Thus, it will be now interesting to elucidate if A_2ARs contribute to MF formation/outgrowth independently of the postsynaptic target or if A_2ARs are only involved in the growth of abnormal retrodirective axons. In addition, it is worth noting that we recently showed that A_2ARs also play a key role in GABAergic synapses stabilization²⁰. Although we cannot yet extend this role of A_2ARs to the stabilization of MF synapses, we do not discard also an eventual postsynaptic effect of A_2ARs in the stabilization/formation of MF contacts. Hence, our findings leave open the possibility that A_2ARs may control circuit remodeling in the adult brain not only by the presynaptic regulation of axogenesis, either directly or indirectly, but also through the reactivation of their ability to control synapse stabilization/maturation.

The gain of function of neuronal A_2ARs after SE to control MF sprouting provides a functional correlate between the previously reported increased density of A_2ARs in nerve terminals after SE²² and the increased density of CD73, which has been proposed to be a marker of synaptic remodeling after SE⁵¹. The present observations that the synaptic release of ATP is increased after SE (Fig. 6a) and that the knockout of CD73 dampens SE-induced MF sprouting (Fig. 6b–e) allows concluding that noxious brain conditions such as SE trigger an upregulation of the whole ATP-CD73-A_2AR axis to control MF sprouting. Although this bolstered ATP-CD73-A_2AR axis seems to be recurrently engaged in the control of brain and synaptic dysfunction in different animal models of brain diseases^24,67,68, the molecular mechanisms engaged by A_2ARs to tinker with synaptic remodeling are still not defined. The control of synaptogenesis during development involves an A_2AR-mediated control of cAMP levels in synaptic terminals²⁰, but the A_2AR-mediated control of synaptotoxicity after noxious stimuli seems instead to involve p38-mediated signaling⁶⁹. A possible A_2AR-mediated control of BDNF and TrkB function⁷⁰ is also grounded on conflicting evidence: although several evidences point towards an involvement of an up-regulation of BDNF in triggering this initial branching^{9,10,11,13,71}, this does not seem sufficient per se to induce MF sprouting^12,14. Furthermore, both the A_2AR-mediated control of synaptogenesis during development as well as the A_2AR-driven axonal elongation in cultured cortical neurons were not modified by the presence of a BDNF scavenger^20,56. Clearly, the transducing pathways operated by these pleiotropic A_2ARs to control MF sprouting still remains to be unraveled.

Since the selected model to induce a pathological-like sprouting process in the adult brain is also relevant to model epileptogenesis, it is of interest to tentatively frame the present findings in the context of epilepsy. Adenosine has been for long known as an anti-convulsant⁷² via A₁ receptors (A₁R) activation⁷³, which are mostly activated by adenosine released per se, depressing basal excitatory transmission⁷⁴. However, during epileptogenesis there is a down-regulation of A₁Rs indicated by both a decreased density of A₁Rs in excitatory synapses in animal models of epilepsy²² and a reduction of ambient levels of adenosine, mostly driven by an increased activity of adenosine kinase^75,76. In contrast, A_2ARs become prominent at high-frequencies stimulations, activated by adenosine originated from the CD73-dependent extracellular catabolism of ATP that is prominently released at high-frequencies^17,18,24. Moreover, it was reported an increased density of A_2ARs in different animal models of epilepsy^22,77,78 and in patients with epilepsy⁷⁹, together with the prominent synaptic release of ATP at seizure-like firing patterns^17,80,81,82 and an increase of the density and activity of CD73^{24,49,50,51,79}, which supports a gain of function of A_2ARs during epileptogenesis. Indeed, as mentioned, A_2ARs tether SE-induced increase in neuronal excitability to hippocampal neurodegeneration²³, an effect that is dependent on CD73 ²⁴. Moreover, although we have not directed the present study to investigate this question, it is interesting to note that several studies demonstrated anti-convulsant effects upon inhibition of A_2ARs^83,84,85,86. More importantly, the ability of the pharmacological inhibition of A_2ARs to mostly prevent the progressive seizure severity in kindling protocols^23,25,86 and to arrest long-term deleterious consequences of early-life convulsions⁸⁷, strongly indicate a relevant contribution of A_2ARs to epileptogenesis. This is in accordance with the presently concluded ability of A_2ARs to contribute to the circuit remodeling characteristic of epileptogenesis. However, despite the evidence indicating that MF sprouting forms recurrent excitatory inputs^{3,4,88,89,90,91,92}, the link between MF sprouting and progressive hyperexcitability and seizure severity is still a working hypothesis. It may be pro-epileptogenic, homeostatic or an epiphenomenon^2,93 or may be contributing to epilepsy comorbidities rather than seizure severity⁹⁴. Interestingly, A_2ARs and CD73 also contribute to some of the comorbidities associated to epilepsy^23,24. Further studies are now needed to unravel the functional consequences of the involvement of A_2ARs and CD73 in epilepsy-related circuit remodeling here identified.

Overall, the findings presented shown that A_2ARs contribute to epilepsy-related circuit remodeling, which may involve the reactivation of the ability of A_2ARs to control axon formation and outgrowth during development¹⁹. This posits A_2ARs as potential targets to manipulate aberrant synaptic remodeling during epileptogenesis, which may be extended to other disorders displaying also circuit alterations such as neuropsychiatric disorders affecting adults and prevalent on ageing.