Purinergic Signaling Participates in a Transition Between Functional States of Reactive Microglia and Controls Astrocyte-Driven Neuroinammation in the Model of Trimethyltin-Induced Neurodegeneration

Background The present study aims to explore the involvement of purinergic signaling in the rodent model of hippocampal degeneration induced by trimethyltin (TMT), which results in behavioral and neurological dysfunction similar to Alzheimer’s disease. Our study has provided novel evidence that TMT induced extracellular depositions of amyloid β, which might be the cause of the well-dened progressive hippocampal neurodegeneration and gliosis. Methods We have applied enzyme histochemistry and immunohistochemistry to study spatial and temporal patterns of ectonucleotidase NTPDase1/CD39 and eN/CD73 expression, gene expression analysis and immunochemistry to analyze cellular localization of select purinoreceptors and pro-inammatory cytokines previously associated with microglia and astrocytes activation.


Abstract Background
The present study aims to explore the involvement of purinergic signaling in the rodent model of hippocampal degeneration induced by trimethyltin (TMT), which results in behavioral and neurological dysfunction similar to Alzheimer's disease. Our study has provided novel evidence that TMT induced extracellular depositions of amyloid β, which might be the cause of the well-de ned progressive hippocampal neurodegeneration and gliosis.

Methods
We have applied enzyme histochemistry and immunohistochemistry to study spatial and temporal patterns of ectonucleotidase NTPDase1/CD39 and eN/CD73 expression, gene expression analysis and immunochemistry to analyze cellular localization of select purinoreceptors and pro-in ammatory cytokines previously associated with microglia and astrocytes activation.

Results
Our study demonstrated that all Iba1-ir microglial cells, irrespective of the cell shape and localization, upregulated NTPDase1/CD39, while the induction of eN/CD73 has been observed only at amoeboid microglia, localized within the hippocampal layers with pronounced cell death. Marked induction of P2Y 12 R and P2Y 6 R at amoeboid microglia might re ect the transition from rod to amoeboid microglia and the adaptation to the migratory and phagocytic properties of the latter. Based on the expression of the microglial polarization markers, the majority of microglia belonged to the M2-like functional state. A signi cant change in purinergic signaling components accompanied the response of reactive astrocytes, which occupied the areas with pronounced cell death. Reactive astrocytes, which markedly expressed adenosine A 2A and P2Y 1 receptors, showed massive induction of complement component C3, NF-kB and IL-1β, suggesting that astrocyte-derived in ammation might be responsible for prolonged and spreading neurodegeneration in TMT model.

Conclusion
This study put glia-associated purinergic signaling in the center of molecular pathogenesis of AD-like disease. Our ndings suggest that the ectonucleotidases and purinergic signaling play signi cant role in microgliosis, astrocyte-driven neuroin ammation and prolonged neurodegeneration in the TMT model.

Background
Glial cells are a heterogeneous class of cells in the central nervous system (CNS), playing a range of roles in the maintenance of tissue homeostasis. Among them, critical roles of microglia and astrocytes are to monitor, maintain and preserve the metabolic and structural integrity of the CNS and to respond to noxious stimuli and insults to the brain [1]. Microglial cells react to deranged CNS homeostasis by immediate changes in their morphology and function, which may progress in two directions, assigned as M1/M2 polarization states. The so-called classical or M1 activation is induced by activation of the classical complement cascade and interferon-γ (IFN-γ), featuring the massive release of pro-in ammatory cytokines, including tumor necrosis factor-α (TNF-α), interleukin (IL)-1β and IL-6, the induction of nitric oxide synthase-2 (iNOS) and the burst of reactive oxygen species (ROS) [2]. Alternatively, microglia may assume a neuroprotective M2 phenotype, characterized by elevated expression of anti-in ammatory cytokines, such as transforming growth factor β (TGF-β) and IL-10, and the induction of molecular markers, like arginase-1 (Arg1) [3], [4]. The M2 microglial cells clear apoptotic bodies, and re ne synapses through phagocytosis [5] and release protective factors, which may contribute to protection and repair.
The M1/M2 represents two extreme activation phenotypes of reactive microglia, with a full repertoire of transitional states between them. Based on a distinct combination of microenvironment in uences, alternatively activated microglia may be categorized into the subsets termed M2a, M2b, M2c, M2d microglia [6], capable to perform speci c functions in different physiological and pathological conditions [7]. In response to brain injury, astrocytes assume reactive states that may be discriminated based on the proliferation, and induction of pro-in ammatory mediators and ROS to: (i) newly proliferated astrocytes organized to forms borders around areas of tissue damage or in ammation, and (ii) non-proliferative astrocytes that retain the basic cell structure, tissue architecture, and functional interactions established in healthy tissue. With the analogy to reactive states of microglia, based on their molecular signature these two broad reactive astrocyte subtypes are also classi ed as A1 pro-in ammatory and A2 antiin ammatory astrocytes [8].
Microglia and astrocytes communicate with each other and with affected neurons by using a range of paracrine signals, including cytokines, gliotransmitters, and neuromodulators. When exposed to noxious stimuli the glial cells become a major source of adenosine triphosphate (ATP), released through the hemichannels [9], [10]. The nucleotide acts as a "danger signal" [11] via two classes of purinergic receptors, i.e. ligand-gated P2X channels and G-protein coupled P2Y receptors, widely expressed at all CNS cell types. ATP promotes microglial chemotaxis towards the challenged area [4], [12] via P2Y 12 receptors [13] and triggers microglial phagocytosis through activation of the P2Y 6 receptor [14].
Furthermore, ATP promotes the release of pro-in ammatory cytokines IL-1β, TNF-α, and IFN-γ via P2X 7 receptors [15] and induces the astrocytic release of cytokines/chemokines and glia-derived neurotrophic factor (GDNF) via P2Y 1 R [4]. The reach and duration of extracellular ATP actions are tightly regulated by membrane-bound ectonucleotidases, which rapidly hydrolyze ATP, by its sequential hydrolysis to adenosine, as the nal product. Adenosine, on the other hand, mainly exerts anti-in ammatory and immunosuppressive effects, via G-protein coupled adenosine receptors (A 1 R, A 2A R, A 2B R, and A 3 R) [16], [17].
The enzymatic breakdown of extracellular ATP and the generation of adenosine are catalyzed by two functionally coupled ectonucleotidases, namely ectonucleoside triphosphate diphosphohydrolase 1 (NTPDase 1/CD39) and ecto-5' nucleotidase (eN/CD73) [18]- [20]. In the CNS, NTPDase1/CD39, largely expressed at microglia and vascular endothelial cells [11], [21], crucially participates in the regulation of P2X and P2Y receptor signaling, via hydrolysis of ATP and ADP. The resulting AMP is hydrolyzed to adenosine by eN/CD73, which is expressed by neurons, glial cells, ependymal cells, and cells of the choroid plexus [18], [20], [21]. The two ectonucleotidases act together as an immune checkpoint, as they determine the ratio of ATP/adenosine and the in ammatory status of the tissue. Accordingly, altered function of NTPDase1/CD39 and eN/CD73 and the dysregulation of the purinergic signaling are largely implicated in the pathophysiology of several neurological diseases, including Alzheimer's disease (AD), Parkinson disease (PD), multiple sclerosis (MS), and astroglioma [22]. Therefore, NTPDase1/CD39 and eN/CD73 represent promising pharmacological targets in the treatment and control of neuroin ammatory processes [23].
The main goal of the present study is to explore the involvement of ectonucleotidases and purinergic signaling in the rodent model of hippocampal degeneration induced by trimethyltin (TMT). The toxicant induces a selective and progressive hippocampal neurodegeneration and chronic microgliosis and astrogliosis, which result in behavioral and neurological symptoms and memory dysfunction similar to AD [24]- [27]. In the previous paper, we have described particular reactive astrocyte phenotypes and their dynamic remodeling in TMT-induced neurodegeneration [28]. The TMT-induced reactive astrocyte phenotypes have been previously linked with increased activation of glial P2X 2 R [29]. Given that the involvement of purinergic signaling has not been explored in the TMT model, in the present study, we have analyzed the spatial and temporal patterns of NTPDase1/CD39 and eN/CD73 and purinergic receptors expression in the context of microglia and astrocytes activation. It was previously shown that TMT induced hippocampal neurodegeneration and gliosis, the pattern of which was comparable in adult rats of both sexes [24]- [26], [28], [30], [31]. However, given that the expression of ectonucleotidases in the brain is under the control of gonadal steroids and differs in two sexes [21], [32], [33], the study was performed in female rats, bilaterally ovariectomized three weeks prior to TMT injection [28].

Materials And Methods
A total of 80 animals were divided into two groups. On day zero, animals of TMT group received TMT (8 mg/kg dissolved in 1 mL 0.9 % w/v saline) (in the form of a single i.p injection), whereas the control group received the adequate volume of 0.9 % saline solution. The animals were returned to their cages and monitored for unusual signs of behavior until sacri ce. At 2-, 4-, 7-and 21-days post-intoxication (dpi), animals of TMT and control groups (10 animals/group) were sacri ced by decapitation (Harvard apparatus, Holliston, MA, USA).
Histochemistry, immunohistochemistry and immuno uorescence microscopy Brains (n = 5 per group) were carefully removed from the skull, xed in 4 % PFA for 24 hours, cryoprotected in graded sucrose (10-30 % in 0.2 M phosphate buffer), and stored at 4°C, as described before [21], [28]. The brains were cryosectioned in serial 25-µm thick coronal sections and the sections at 3.12 -3.84 mm antero-posterior to Bregma were air-dried and stored at -20°C until use.

Nissl staining
Alterations in hippocampal cytoarchitecture induced by TMT injection were evaluated by Nissl staining. Sections were kept in 0.5 % thionine solution for 20 min, washed in tap water, dehydrated in graded ethanol (70 % -100 %), and cleared in xylene for 2×5 min and covered with DPX-mounting medium (Sigma Aldrich, USA).

Immunohistochemistry and immuno uorescence
Slides were kept at room temperature (RT) for 30 min prior to staining. After washing in PBS, slides were put in 0.3 % H 2 O 2 in methanol for 20 min, to block endogenous peroxidase, and then immersed in 5 % donkey normal serum at RT for 1 h to block a non-speci c binding. Sections were probed with primary antibodies, overnight at 4°C in a humid chamber. After washing in PBS (3×5min), sections were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (2 h, RT in a humid chamber). The list of antibodies used for immunohistochemistry (ICH) and immuno uorescence (IF) is presented in Table 1. The immunoreaction was visualized with 3,3´S-diaminobenzidine-tetrahydrochloride (DAB, Abcam, UK), which is converted to the insoluble brown precipitate by HRP. Sections were washed in distilled water, dehydrated in graded ethanol solutions (70 % -100 %), cleared in xylene, and mounted with the use of DPX-mounting medium (Sigma Aldrich, USA). Sections were analyzed under LEITZ DM RB light microscope (Leica Mikroskopie & Systems GmbH, Wetzlar, Germany), equipped with LEICA DFC320 CCD camera (Leica Microsystems Ltd., Heerbrugg, Switzerland) and LEICA DFC Twain Software (Leica, Germany). All images were captured at 40× magni cation.
The identical protocol has been applied for double and triple immuno uorescence (IF) staining, with the omission of the methanol/H 2 O 2 step. After incubation with primary antibodies (Table 1), sections were probed with uorescence-dye labeled secondary antibodies and mounted with Mowiol (Calbiochem, La Jolla, CA). For double and triple IF staining, primary and secondary antibodies were separately applied for each labeling. Sections incubated without primary antibodies or with rat pre-immune sera were used as negative controls. Sections were analyzed by confocal laser-scanning microscope (LSM 510, Carl Zeiss GmbH, Jena, Germany), using Ar Multi-line (457, 478, 488 and 514 nm), HeNe (543 nm) and HeNe (643 nm) lasers using 63× (×2 digital zoom) DIC oil, 40× and monochrome camera AxioCam ICm1 camera (Carl Zeiss GmbH, Germany).

Immuno uorescence quanti cation
Raw multi-image immuno uorescence micrographs were used to measure integrated uorescence density and the density con ned within ve pre-de ned regions of interest (ROIs), with background uorescence subtraction for at least 3 images per ROI and n = 5 sections per group (JACoP ImageJ plugin). A degree of overlap and correlation between multiple channels were estimated by calculating Pearson's correlation coe cient (PCC) [35]. PCC is statistical parameter that re ects both co-occurrence (degree at which intensities of two channels for each pixel are beyond or above the threshold), and correlation (pixel-for pixel proportionality in the signal levels of the two channels). PCC values range from 1 (for two images whose uorescence intensities are perfectly, linearly related) to -1 (for two images whose uorescence intensities are perfectly, but inversely, related to one another). Values near zero re ect distributions of probes that are uncorrelated with one another. The results are expressed as mean PCC± SEM.  Table 2. Relative quanti cation was performed using 2 -ΔΔCt method, using cyclophilin A (CycA), as a reference gene. For each group, ve samples run in duplicate were analyzed.
Ampli cation e cacy was assessed by the generation of internal standard curves by several-fold dilutions of generated cDNA, while melting curve analysis at the end of each experiment was used to con rm the formation of a single PCR product. The results were expressed as the abundance of target mRNA/CycA-mRNA at each dpi relative to a corresponding intact control ± SD. The mRNA levels of most investigated genes showed biphasic response to TMT intoxication; initial downregulation at 2-dpi was followed with gradually upregulation and overcoming basal levels at 7-and 21-dpi (Supplementary Table  1). The transient down-regulation may be the consequence of increased corticosterone levels [36] observed in the days after TMT intoxication [37], [38].

Statistical analysis
Quantitative data were scrutinized for normality and statistically analyzed by parametric tests. The results of qPCR and PCC quanti cation obtained for 7-dpi and 21-dpi were compared to the corresponding control and analyzed using Student's t-test. Data are presented as mean ± S.E.M. For all statistical analyses, Graphpad Prism 5.04 (Graphpad) software was used.

Results
Spatio-temporal patterns of neurodegeneration, amyloid-β deposition and gliosis after TMT exposure Nissl staining, applied to explore the overall hippocampal cytoarchitecture, corroborated existing data on TMT-induced neurodegeneration (Figure 1). Brie y, cell disorganization and thinning of neuronal cell layers in CA1 and hilus/pCA3 observed already at 4-dpi, progressed to noticeable neurodegeneration in CA1 and proximal and medial CA3 (p/mCA3) at 7-dpi (Figure 1a), with almost complete disappearance of neuronal somata in CA1 and p/mCA3 regions at 21-dpi. Of note is that during the whole period after TMT exposure, the complete CA2 sector remained intact, without apparent neuronal cell loss and gliosis. On the top of cell death, immunolabeling directed to amyloid-β protein (Aβ) demonstrated the presence of Aβ depositions in neuronal cell layers of CA1 and CA3, in the form of diffuse and dense-core plaques, at 7dpi and 21-dpi, respectively (Figure 1b). Immunostaining directed to astrocyte marker GFAP showed the presence of pronounced gliosis (Figure 1c). The rst indication of astrogliosis was observed already at 2dpi [28] while the full-blown hypertrophied astrocytes were seen at 7-dpi in CA1 region and atrophy-like morphotype was observed in the hilar/CA3 region. At 21-dpi, hypertrophied/atrophied astrocytes were widespread in CA1 and CA3 sectors, and covered the hilar region in the form of gliotic scar, while CA2 remained devoid of GFAP-ir. Even greater diversity in cell morphology was observed among reactive microglia (Figure 1e). Speci cally, highly rami ed Iba1-ir cells, evenly distributed in the control hippocampal tissue (Figure 1d), were gradually transformed to peculiar rod Iba1-ir cells at 4-dpi, observed in synaptic layers of CA1 and the hilar/pCA3. Besides rod shape, a range of other reactive Iba1-ir morphotypes was observed, from hyper-rami ed to bushy/amoeboid. At 7-dpi, more rod Iba1-ir cells populated the synaptic layers in the entire CA1 and the hilar/pCA3 sectors, while cells which already entered neuronal cell layers, attained amoeboid morphology. At 21-dpi, Iba1-ir cells located just next to neuronal cell layers still displayed their elongated morphology, while more numerous Iba1-ir in neuronal cell layers displayed amoeboid morphology. Interestingly, rod Iba1-ir cells were not observed in the hilar/pCA3, whereas at the latest time point, the hilar area and granular cell layer appeared completely without Iba1-ir (Figure 1d).

Expression of NTPDase1/CD39, eN/CD73 and purinoreceptors involved in microglial reactivity
The main goal of the present study was to explore the involvement of the purinergic signaling system in the TMT-induced hippocampal neurodegeneration and gliosis. Gene expression analyses included genes encoding NTPDase1/CD39 and eN/CD73, which are the major ectonucleotidases involved in the extracellular catabolism of ATP in the brain. Two and a half-and ve-fold increase in NTPDase1/CD39-mRNA abundance were observed at 7-dpi and 21-dpi, respectively, and two-fold increase in eN/CD73-mRNA was observed at 21-dpi (Figure 2a). The identity of cells that up-regulate NTPDase1/CD39 in response to TMT was determined by enzyme histochemistry, which enables labeling of cells that exhibit the enzymatic activity in situ. In control sections, the product of ATP/ADP-hydrolyzing reactions, which corresponded to the catalytic activity of NTPDase1/CD39, was uniformly present at rami ed cells that, based on morphological criteria, belonged to resting microglia (Figure 2b). At 7-dpi, cells with ovoid cell body and retracted process, in the vicinity or within neuronal cell layers were histochemically labeled. At 21-dpi, numerous amoeboid cells in ltrated the neuronal cell layers. The obtained patterns of ATP/ADP enzyme activities closely corresponded to Iba1-ir (Figure 1d), suggesting that reactive microglial cells mostly up-regulated NTPDase1/CD39 after the exposure to TMT.
The identity of cells that induced eN/CD73 in response to TMT was determined by combining AMP-based enzyme histochemistry (Figure 2b) with eN/CD73-directed immunocytochemistry (Figure 2c). In intact hippocampal tissue, diffuse histochemical reaction and eN/CD73-ir were observed in synaptic layers, while neuronal cell layers remained unstained. After TMT exposure, the histochemical reaction and eN/CD73-ir begun to be noticed at amoeboid cells, increasingly in ltrating within neuronal cell layers from 7-dpi and afterwards. Cellular localization of eN/CD73-ir was ascertained by triple immuno uorescence directed to GFAP, Iba1, and eN/CD73 (Figure 3a). The pattern of eN/CD73ir increasingly overlapped with Iba1-ir, peaking at 21-dpi at amoeboid Iba1-ir cells in ltrated within neuronal cell layers, while the co-localization with GFAP-ir was not observed. These ndings con rmed the dominant expression and up-regulation of eN/CD73 by reactive microglial cells after TMT exposure.
The co-localization of NTPDase1/CD39 and eN/CD73 at microglial cells was demonstrated by doubleimmuno uorescence labeling, which showed the overlap of the signals corresponding to NTPDase1/CD39 and eN/CD73 at amoeboid microglial cells, while rami ed and rod NTPDase1/CD39ir cells within synaptic layers did not show eN/CD73-ir (Figure 3b). The level of overlap between the two-ir signals was quanti ed by using Pearson correlation coe cient (PCC). The raising PCC values showed increasing overlap between Iba-1/eN/CD73-ir and NTPDase1/CD39-eN/CD73-ir signals after TMT exposure, whereas negative PCC values for GFAP-ir and eN/CD73-ir corroborated the lack of astrocytic expression of eN/CD73 after TMT.
We further assessed gene expression of several purinoceptors functionally associated with chronic neuroin ammation (Figure 4). Concerning ATP/ADP-sensitive P2 receptors, we detected strong upregulation of P2X 4 R-, P2X 7 R-and P2Y 2 R-mRNA and almost 20-fold induction of P2Y 6 R-and P2Y 12 R-mRNA at 7-dpi and 21-dpi. For P1 receptors, a several-fold increase in A 3 R-mRNA at 7-dpi and 21-dpi, together with signi cant, but less conspicuous induction of A 1 R-, A 2A R-and A 2B R-mRNAs were detected at the latest time point after TMT exposure.

Functional state of reactive microglia and astrocytes
The results pointed to the marked induction of NTPDase1/CD39 by microglial cells, and the colocalization with eN/CD73 at amoeboid microglial cells after TMT exposure. It is known that resting microglia activate under the in uence of ATP and develop functional phenotype, which may be broadly categorized as pro-in ammatory or neuroprotective. Therefore, we next assessed the in ammatory status of the tissue, by determining the expression of several cytokines and in ammation markers, iNOS, C3, and arginase 1 (Arg1), which are often used to discriminate between M1/M2 polarization states. Severalfold increases in TNF-α-and IL-6-mRNA abundances at 7-dpi at the earliest, followed by strong upregulation of IL-1β-and IL-10-mRNA (Figure 5a) at the latest time point were detected. The induction of the cytokines was accompanied by prominent and lasting induction of C3-mRNA and iNOS-mRNA ( Figure  5a), without signi cant change in Arg1-mRNA (data not shown). However, neither of the tested proin ammatory cytokines (data not shown) and polarization markers was found in association with Iba1-ir, as demonstrated by double immuno uorescence (Figure 5b). However, Iba1-ir cells did show Arg1ir and ir corresponding to phagocytic marker CD68. The signals co-occurrence was observed at rod and amoeboid cells at 7-dpi and later. The induction of the chemotaxis marker P2Y 12 R was also observed at Iba1-ir cells.
The lack of expression of pro-in ammatory cytokine and markers by Iba1-ir microglial cells prompted us to explore their astroglial expression ( Figure 6). Speci cally, the signals corresponding to IL-1β, TNF-α, and IL-10 almost completely overlapped with GFAP-ir at 7-dpi and 21-dpi. The GFAP-ir cells also expressed iNOS, NF-kB, and C3, suggesting that the astrocytes were the major source of the in ammatory cytokines and the driving force of the TMT-induced neuroin ammation. With regard to purinoreceptors expression, massive induction of P2Y 1 R and A 2A R was found on GFAP-ir and C3-ir astrocytes at 7-dpi and 21-dpi, which suggested the involvement of the purinoreceptors in the pro-in ammatory astrocyte phenotype.

Discussion
The results of the present study corroborate the existing data on the spatiotemporal pattern of neurodegeneration and gliosis in the TMT model. Brie y, a single i.p. injection of TMT triggered a wave of spreading neurodegeneration and neuronal cell death starting from 2-dpi, which sequentially hit neuronal cell layers of the hilus/CA3, and CA1, while sparing the entire CA2 sector [24], [26], [28], [39]. As a result of the neuronal cell death, the entire affected sectors became gradually populated with reactive astrocytes, whose number and hypertrophy became noticeable at 4-dpi and peaked at 21-dpi, while the response of microglia slightly lagged behind. Again, the hardest hit neuronal cell layers in hilus/CA3, and CA1 sequentially became in ltrated with microglial cells of amoeboid shape, while the CA2 neuronal cell layer rested spared. Synaptic layers in the affected sectors became largely populated with rod microglia, occasionally found in a train formation. At the end of the monitoring period, hilus/CA3 was completely devoid of microglia, while the parenchyma was covered with numerous astrocytes with atrophic-like morphology [28].
On top of the neurodegeneration, the present study provided the novel evidence that TMT caused extracellular depositions of Aβ, implying that it might play a role in TMT-induced neurodegeneration, as shown in AD and several other diseases, including PD, Huntington's disease and posttraumatic brain [40]. By analogy with AD, we hypothesize that Aβ may trigger neuroin ammatory responses of microglia and astrocytes by enhancing the release of ATP and the activation of ATP-sensitive purinoceptors [41], [42].
Degradation of extracellular ATP results in extracellular adenosine level jump and enhanced adenosine receptor-mediated responses [42], which set the in ammatory status of the tissue [17]. Therefore, the present study aimed to explore the supposed parallels between the TMT model and AD, with special reference to purinergic signaling and its role in gliosis.
Keeping up with the main goal, the central nding of our study is prominent induction of NTPDase1/CD39 and eN/CD73 by reactive microglial cells after TMT exposure, whereas the pattern of the induction differed between two main morphotypes of reactive microglial cells. In physiological conditions, resting microglia express NTPDase1/CD39 [43]. After the exposure, NTPDase1/CD39 was markedly up-regulated in all Iba1-ir cells, irrespective of their shape and position, while eN/CD73 was selectively up-regulated in distinct microglial cell type. Speci cally, eN/CD73 was expressed by amoeboid Iba1-ir cells, implicating that the differential induction might be an adaption to speci c hippocampal micro-location or speci c function of the microglial subsets. The transition between functional states of reactive microglia is accompanied by the morphological transformation of the cells, and among the critical factors that trigger the transition are ATP, adenosine, vitamin E, IL-34, and chemokine fractalkine [44]- [46]. Rod microglia are usually found at early stages in AD, in association with damaged neurons and axons, and not in aggregation with other glial cells [5], which imply their protective and reparative role [47]. Rod cells are able to provide new cells and to transform into amoeboid microglia [48]. Hence, the main functional properties of amoeboid microglia are motility and phagocytosis, both of which are tightly regulated by purinergic signaling mediated by NTPDase1/CD39 and eN/CD73 and select P1 and P2 receptors [19]. Speci cally, adenosine controls the retraction of microglial processes and the shape change via A 2A Rs [49], and affects chemotaxis via P2Y 12 R/A 3 R co-activation [13], [50]. The activation of the purinoceptors, speci cally the activation of P2Y 6 R in microglial cells promotes the phagocytosis of debris generated by excitotoxic hippocampal damage [51]. Furthermore, NTPDase1/CD39 and eN/CD73 not only that catabolize extracellular ATP and provide adenosine, but function as clusters of differentiation and cell adhesion molecules, which regulate the adhesion and glial cell migration through speci c interactions with extracellular matrix components [52]. Thus, we propose that the up-regulation of NTPDase1/CD39 and eN/CD73 and the change in purinoceptors repertoire induced by TMT are speci c adaptations to the phagocytic activity and amoeboid shape of microglial cells, which migrate along with the wave of spreading neurodegeneration, from hilus towards CA3 and CA1.
With regard to the functional state and M1/M2 polarization of microglia, our data showed that Iba1ir cells co-expressed Arg1 and CD68, and did not exhibit iNOS, NF-kB, and C3, which imply the M2-like reactive phenotype [53]. The up-regulation of NTPDase1/CD39 by M2-like microglial cells was previously demonstrated in experimental autoimmune encephalomyelitis, where the induction of NTPDase1/CD39 tended to be associated with Arg1-ir and CD68-ir microglial cells [54]. M2 polarized microglia may be further divided into four distinct subsets induced by different environmental cues, among which M2d is triggered by IL-6 and develops under the in uence of adenosine [53]. Speci cally, it has been shown that activation of microglial TLR receptors induces A 2A R-and A 2B R-mediated switch in the production of cytokines, from pro-in ammatory TNF-α to the anti-in ammatory IL-10 [53], [55], [56]. Our study provided mechanistic evidence for all the critical steps in the sequential pathway, including induction of IL-6, transcriptional and functional up-regulation of NTPDase1/CD39 and eN/CD73, and induction of A 2A R and A 3 R. The trend of decreasing expression of TNF-α and IL-6 and increasing expression of IL-10 signaling, presented herein is also in the correlation with M2d reactive microglial state.
Our study also provided evidence that purinergic signaling regulates functional attributes of reactive astrocytes after TMT-induced neurodegeneration as well. Massive induction of P2X 7 R at the latest time point after TMT exposure re ected the involvement of the low-a nity ATP receptors in the delayed and lasting astrogliosis initiated by a single TMT injection. It is known that ATP locally released as a danger signal from damaged neurons acts at P2X 7 R located at proximal glial cells and initiates NOD-like receptor protein 3 in ammasome assembly and the release of IL-1β [41], [57]. However, a wave of neurodegeneration and apoptosis induced by TMT might provide conditions for a sustained ATP release and prolonged P2X 7 R activation, with the resulting postponed induction of IL-1β, together with TNF-α, IL-18, and IL-6 [41]. We have demonstrated that TMT induced IL-1β, not before 21-dpi, primarily in reactive astrocytes which co-expressed markers of the pro-in ammatory phenotype, iNOS, NF-kB, and C3. At the latest time point, these apparently pro-in ammatory astrocytes up-regulated A 2A Rs, whose aberrant expression was demonstrated in the hippocampal astroglia in AD patients and in AD mouse models [58], [59]. Furthermore, massive induction of P2Y 1 R on astrocytes postulates its signi cance in TMT-induced reactive astrogliosis, as already demonstrated in several models of AD [60], [61]. It is known that pathological conditions may induce gain-of-function in A 2A Rs signaling, due to the establishment of toxic iso-and hetero-receptor complexes with other membrane proteins and the disturbance of several intracellular signaling pathways [62]. Moreover, given that eN/CD73 provides the ligand for A 2A R activation, enhanced expression of eN/CD73 and the prolonged coupling with A 2A R had been suggested to play a role in the transition between acute and chronic neuroin ammation in several neurodegenerative disorders [17], as well it may be the case in AD-like TMT-induced neurodegeneration.

Conclusion
To sum up, our study demonstrates that purinergic signaling plays a signi cant role in microglial and astroglial responses to TMT-induced neurodegeneration. Marked and distinct changes in eN/CD73 and purinoreceptors expression accompany activation of microglia and their transition towards two prevailing morphotypes, rod and amoeboid microglia. The induction of NTPDase1/CD39, eN/CD73, and select purinoreceptors, most notably P2Y 12 and P2Y 6 , probably re ected the transition from rod to amoeboid microglia and the adaptation to the migratory and phagocytic properties of the cells, which, according to the expression of the functional polarization markers, most likely belonged to M2d functional state of the cells. The signi cant change in purinergic signaling components accompanied the response of reactive astrocytes as well, which permeated the neuronal cell layers in affected sectors. Reactive astrocytes massively up-regulated A 2A R, together with the persuasive expression of complement component C3, NF-kB, and IL-1β, indicating their harmful phenotype responsible for the prolonged and spreading neurodegeneration. Overall, the results of our study suggest that the ectonucleotidases and purinergic signaling play signi cant role in microgliosis, astrocyte-driven neuroin ammation and prolonged neurodegeneration in the TMT model (Fig 7).Finally, our results put glia-associated purinergic signaling in the center of molecular pathogenesis of AD-like disease.