Adult male 6-month-old C57BL6 were purchased from Janvier Labs (Saint Berthevin Cedex, France), and they were acclimatized for 19 days before the experiments. Adult male 6-months-old CX3CR1^+/GFP were bred in loco. Mice were housed in a controlled environment with the temperature at 21°C, controlled humidity, 12 h light/dark cycle, cages enriched by stimuli, fed ad libitum with rodent pellet diet, and free access to water. All animals included in this work and the procedures performed are approved by the Malmö/Lund Animal Ethical Committee (Dnr 5.8.18-09454/2021).

Mice received a single i.p. injection of LPS (clone O55:B5, Sigma-Aldrich, St. Louis, MO, United States) or vehicle alone and were sacrificed 1- or 30-days post-treatment. 5 mg/kg concentration of LPS was used to induce the acute and chronic response and 5, 0.5, 0.05, and 0.005 mg/kg doses were used for the dose-dependent response experiments. All experiments involved three animals in each group and were performed simultaneously. At the end of the experiments, mice were deeply anesthetized using pentobarbital and perfused with saline, followed by 4% paraformaldehyde (PFA). Brains were post-fixed in the same fixative for one overnight and then immersed in PBS with 30% sucrose until use.

Brains were cut into 25 um thick sections using a microtome (Leica SM2010R, Microtome and Microscope, Leica Microsystems, Wetzlar, Germany) and stored at −20°C in an anti-freeze solution. Brain slices were stained for IIba1 antibody (1:1000, Nordic Biolabs, Wako Chemicals, Täby, Sweden, Cat. #019-19741), CD68 antibody (1:500, Bio-Rad, Hercules, CA, United States, Cat. #MCA1957), TH antibody (1:2000, Chemicon-Millipore, Burlington, MA, United States, Cat. #AB152 and #MAB358), Gephyrin antibody (1:250, Synaptic Systems, Göttingen, Germany, Cat. #147011C3), GFAP antibody (1:500, Dako, Santa Clara, CA, United States, Cat. #Z0334), in PBS with the serum (5–10%) from the animal species of the secondary antibodies and Triton-X 100. The images were acquired using Olympus Virtual Stage 120 with extended focus imaging setting (EFI) for scan images (20x and 10x magnifications), Olympus BX53 microscopes, Shinjuku, Tokyo, Japan (20x and 60x magnification), and LEICA Stellaris 8 Dive for confocal images (40x and 63x magnification). EFI setting consists of five optical sections with a 5 um increment between two consecutive optical sections. Z-stack confocal images consist of 30 optical sections, with 0.5 um increment between two consecutive optical sections, and their 3-dimensional projection was used for analyses.

We examined 16 different brain regions: (1) Different representative nuclei, such as NAc, striatum, thalamus, hypothalamus, amygdala, VTA, SNpc, and SNpr; (2) Different regions of the cerebral cortex: primary motor and somatosensory cortex (PSM cortex), posterior parietal association areas (PPA cortex), primary visual area (visual cortex, PC), hippocampus, entorhinal cortex, posterior piriform cortex, and the simple lobule of cerebellum; (3) anterior corpus callosum was the only white matter structure considered. Regions of interest (ROIs) followed coordinates from Allen Brain Atlas, as shown in few examples in Supplementary Figure 1A.

We evaluate microglia activation considering changes in the percentage of the marked area and cell density. We performed image analyses using ImageJ software and different homemade macros adapted to our purpose. Iba 1 marked areas were detected using a set threshold able to distinguish cell soma and main branches (Supplementary Figure 1B). Cell density was measured using a threshold able to identify the soma of microglia cells. We applied several transformations and size filters to detect the cell soma specifically, as shown in Supplementary Figure 1C.

To quantify TH and Gephyrin, we used confocal images and ImageJ software with homemade macros and set thresholds. To assess GFAP and CD68 alterations in response to the exposure of LPS, we randomly selected 4–7 frames of 20x images per animal and analyzed them using ImageJ software and homemade macros. We quantified GFAP using a set threshold while we performed a semi-quantitative analysis of CD68 with variable thresholding.

Statistical analyses were performed using Graph Pad Prism 9 as software. All graphed values are shown as means with standard deviations, and statistical details are described in the results and figure legends. Generally, statistical significance was assessed using a one-way ANOVA with Tukey or Sídák correction, two-way ANOVA with Tukey correction, and unpaired student t-test. Significances are partially summarized in the graphs, and more detailed information is provided in Supplementary Tables 1–6.

To study microglia heterogeneity in C57BL6 mice, we first characterized microglia cells across the different brain regions in control mice without LPS exposure. SNpr represented the brain region with the highest percentage of the marked area by Iba1 (Figures 1A,D). Comparatively, microglia coverage in the cerebellum, VTA, and corpus callosum was the lowest, followed by the hypothalamus, thalamus, and SNpc. The remaining nine regions showed similar microglia marked area. We observed a similar situation when we analyzed the cell density (Figures 1B,D). Then, we examined microglia size and morphology, considering their branch structures. For this purpose, we performed confocal microscopy imaging in Iba1 immuno-labeled sections. We conducted cell size analysis creating ROIs around single cells. This analysis (Figure 1C) showed that microglia cells in 12 regions present no significant difference in cell size. Interestingly, the size of microglia cells in the cerebellum, corpus callosum, VTA, and SNpc was significantly smaller than in other brain regions. After that, we performed a qualitative morphological analysis, considering the complexity of branch structure. We distinguished the branches between “simple” and “complex” based on the number of branches that derived from cell soma and by the presence of further ramifications on each branch. From this analysis, we observed the simplest branch structure in VTA, cerebellum, corpus callosum, SNpc, thalamus, and hypothalamus, while all other regions showed a complex structure (Supplementary Figure 2).

To further investigate microglial heterogeneity, we performed a qualitative analysis of the cluster of differentiation 68 (CD68), a microglial-specific transmembrane glycoprotein localized primarily on lysosomal and endosomal vesicles. From this analysis, we generally observed the same pattern as for Iba1. The heterogeneity between the different brain regions seems to depend mainly on differences in microglia density. However, CD68 positive profiles were more robust in SNpr compared with other regions, while the hypothalamus and VTA showed lower amounts. Interestingly, the CD68 positive profile seems more intense in the PSM cortex and cerebellum concerning all other regions (Supplementary Figure 3). These data showed that nine of the examined regions present a similar percentage of Iba1 marked area and cell density in physiological conditions, while the other seven appeared to be more heterogeneous. Nuclei exhibited higher variability than cortical regions. Morphologically, microglia cells were heterogeneous, and their size depended primarily on the complexity of their branch structure. Moreover, further differences could be present considering other microglial features such as the lysosomal or endosomal system.

To evaluate whether microglia heterogeneity could contribute to different inflammatory responses among different brain regions, we compared microglia activation in mice treated systematically with either 5 mg/kg LPS or vehicle (PBS). We detected a significant increase of the marked area by the Iba1-positive profile (Figures 2A,E) and cell density (Figures 2B,E) in all brain regions upon LPS exposure. SNpr and the entorhinal cortex presented the most robust microglia activation. Comparatively, the activation of microglia cells in the VTA, cerebellum, and corpus callosum was the lowest, although the response was still statistically significant upon LPS treatment compared to the control mice. Although the extent of activation depends on the baseline level of the examined brain regions, these data still suggest a difference in inflammatory responses between various brain regions.

When we examined the fold changes between LPS-treated and control groups, we observed a 2–3-fold increase in microglia marked area (Figure 2C) and similar changes in microglia density (Figure 2D). The most significant increase in the microglia marked area appeared in the VTA. In contrast, the lowest changes were detected in the PSM cortex (Figure 2C). Concerning the relative changes in cell density (Figure 2D), the entorhinal cortex, VTA, and SNpc showed the highest differences, while the PSM and SNpr were the lowest. These differences in microglial activation were evident when comparing PSM and entorhinal cortices (Figure 2E). These regions present similar microglia marked area and density in control conditions. However, in LPS treated mice, microglia activation was more robust in the entorhinal cortex than PSM cortex.

In the end, we performed a semi-quantitative analysis of CD68 changes under LPS exposure to confirm a differential microglial response among different brain regions. We observed a general increase in all regions, which appeared dependent both on cell density increase and on intensity of CD68 immunoreactivity. The most evident change occurred in the SNpr, where the signal increased stronger than that in all other regions (Supplementary Figure 4). These data indicate differential microglial responses to LPS exposure among different brain regions.

To examine whether microglia heterogeneity also affects the sensitivity of inflammatory responses, we performed a dose-dependent i.p. injection of LPS (from 0.005, 0.05, 0.5 to 5 mg/kg) together with the control group. Microglia morphological changes were examined 24 h after LPS administration, and we observed significant differences in several brain regions. Iba1-positive microglia marked area (Figure 3A) and cell density (Figure 3B) increased from low doses of LPS (0.05 mg/kg), particularly in the SNpr, entorhinal cortex, NAc, and hypothalamus. In most other brain regions (such as the striatum, thalamus, PSM cortex, PPA cortex, VTA, corpus callosum, and hippocampus), only 5 mg/kg LPS induced a significant increase in the marked area and the density of microglia. In the fold change analyses, we detected significant changes in the marked area at low doses in the hypothalamus, entorhinal cortex, VTA, and cerebellum, followed by SNpc, SNpr, and NAc (Figure 3C). A similar pattern was observed for the fold increase in the density of microglia cells, although the changes of this parameter were less extent and significant (Figure 3D). Similar pattern changes in cell density and marked area were observed in most other regions, which became significant at the highest dose of LPS (Figures 3C,D). This data confirms the differential sensitivity of microglia cells to LPS exposure. We next characterized the morphological features of microglia following LPS treatment (Figure 3E; Supplementary Figures 5, 6). We observed that microglia cells in the hypothalamus, SNpr, entorhinal cortex, VTA, and cerebellum exhibited activated features, e.g., enlarged cell body with thickened processes, in the mice injected with 0.05 and 0.5 mg/kg LPS dose. However, the activation did not look uniform even in the regions with activated microglia. For example, the highest microglia activation in the hypothalamus was in proximity to the CVO, while it became gradually less extensive away from this hypothalamic sub-region (Supplementary Figure 5). Then, we semi-quantitatively compared the CD68 changes in the hypothalamus, SNpr, and entorhinal cortex in mice treated with different LPS doses and the vehicle. Interestingly, we observed evident changes already in the SNpr at the dose of 0.05 mg/kg (Supplementary Figure 7). Moreover, considering relative changes, SNpr represents the only region where CD68 showed significant changes already at 0.05 mg/kg, while the hypothalamus at 0.5 mg/kg.

To confirm that the different sensitivity observed was not dependent by a different TNFα/LPS diffusion within the brain, we examined the vessel distribution using podocalyxin as a marker. We found a higher amount of vessels in the thalamus, cerebellum, and PSM, PPA, and visual cortices (Supplementary Figures 8A,B). This pattern does not correlate with the microglial sensitivity observed in several brain regions, confirming that our data were independent of TNFα and/or LPS diffusion.

This data shows that microglia cells present innate regional differences in sensitivity to systemic inflammation induced by i.p. LPS administration.

An interesting neuroinflammatory feature induced by systemic LPS administration is the ability to convert an acute inflammatory response to a chronic one (Qin et al., 2007; Liu et al., 2008; Zheng et al., 2013). However, whether this chronic inflammation is present uniformly in different brain regions is still not well-understood. To clarify this aspect, we treated mice with 5 mg/kg LPS and vehicle (PBS) i.p. injection and sacrificed them 30 days post-treatment. We observed a slight increase in the marked area and density of microglia cells in a few brain regions, such as the amygdala, SNpr, PSM cortex, and the piriform cortex (Figure 4; Supplementary Figures 9, 10). These data indicate sustainable microglia activation in selected brain regions in the extended period (1 month) after LPS injection. Moreover, 1 month post-LPS treatment, microglia cells lost their morphology shown 24 h after the treatment. Although a higher Iba1 intensity was still present, they adopted a more ramified morphology. This data indicates differences in microglia ability to maintain the inflammatory condition in several regions 1 month after treatment.

We observed that microglia cells appeared in the SNpr with the highest density in the naive mice condition and responded most severely to the acute and chronic stimulation by LPS exposure. Then, we examined the consequences on neuronal profiles in this specific region. SNpr represents an important output nucleus in basal ganglia circuitry, therefore, we examined the subsequent neuronal loss profiles in this region. SNpr consists mainly of gamma-aminobutyric acid (GABAergic) -positive neurons with a spontaneous high-frequency activity that projects to the motor thalamus (Zhou and Lee, 2011; Lanciego et al., 2012). SNpr mainly receives GABAergic projections from the striatal-nigral bundle and glutamatergic projections from the subthalamus (Lanciego et al., 2012). Moreover, this region is enriched with dopaminergic dendrites from the neighboring SNpc region, making synapses with striatal-nigral GABAergic projections (Björklund and Lindvall, 1975; Geffen et al., 1976; Mercer et al., 1979; Cheramy et al., 1981; Lanciego et al., 2012).

We performed a double immune-labeling with antibodies against TH, the key enzyme for dopamine synthesis, and Iba1. We observed dopaminergic, TH-positive, dendritic profiles extending from the SNpc to the SNpr (Figure 5A), intermingled with Iba1 microglia cells branches (Figure 5B). Quantitative analyses with a series of confocal images showed that the density of dopaminergic dendrites significantly decreased in the medial part of the SNpr (Figures 5C,E) in mice exposed to 5 mg/kg LPS compared to the vehicle-treated mice 24 h after the injection. Surprisingly, this decrease was no longer evident 30 days post-LPS injection (Supplementary Figure 8C). Quantitative analyses of GABAergic synapse marker gephyrin presented no alterations in the SNpr of mice exposed to 5 mg/kg LPS (Figure 5D; Supplementary Figures 8D,E). These data suggest that the acute inflammatory response induced by LPS leads to transient and specific alterations of dopaminergic dendrites in SNpr.

CX3CR1 is a chemokine receptor localized in microglia and other immune cells, such as dendritic cells. Its ligand, CX3CL1 (fractalkine), is constitutively expressed in neurons (Pawelec et al., 2020). CX3CR1^+/GFP mice have been widely used in studying microglia dynamics and activation in various models of neurodegenerative diseases (Marker et al., 2010; Avignone et al., 2015). We, therefore, investigated microglia alterations in CX3CR1 partially ablated mice (CX3CR1^+/GFP). We first examined C57BL6 and CX3CR1^+/GFP age-matched mice in the naive condition (without LPS exposure). Comparing the percentage of the marked area by Iba1, we found no significant differences between wild type and CX3CR1^+/GFP mice (Figure 6A), indicating that microglia seem to be similar between these two mice in the resting condition, at least from a morphological point of view. Subsequently, we analyzed CX3CR1^+/GFP mice treated with 5 mg/kg LPS and vehicle 24 h post-injection (Figure 6B). Significant microglia activation was present in SNpr, SNpc, NAc, amygdala, visual cortex, entorhinal cortex, and hippocampus, while not appearing in the rest of the brain regions. Comparing relative changes between LPS- and vehicle-treated groups in C57BL6 and CX3CR1^+/GFP mice, we observed a general and significant decrease in relative changes in all regions examined in mice partially ablated by CX3CR1 (Figures 6C,D; Supplementary Figures 11, 12). These data suggest that partial ablation of this receptor made the microglia cell less sensitive or responsive to the LPS treatment.

To further confirm this observation, we qualitatively compared CD68 changes. No differences were observed between C57BL6 and CX3CR1^+/GFP mice in the control condition. Comparing CX3CR1^+/GFP mice treated with 5 mg/kg of LPS and relative controls, we observed a slight or no increase in CD68 signal. Comparing CD68 between C57BL6 and CX3CR1^+/GFP mice treated with 5 mg/kg LPS, we observed a clear lower CD68 expression in CX3CR1^+/GFP mice (Supplementary Figure 13).

We then examined the effect of partial ablation of CX3CR1 on the dose-dependent pattern observed in C57BL6 mice (Figure 7A). We treated C57BL6 and CX3CR1^+/GFP mice with 5, 0.5, 0.05, and 0.005 mg/kg i.p. LPS and relative vehicle. We observed that the partial ablation of CX3CR1 led to a general decrease in microglia activation upon systemic LPS administration. Despite a general decreased response at lower doses, microglia in CX3CR1^+/GFP mice were responsive to the high dose (5 mg/kg) LPS treatment in most brain regions tested, except for the thalamus, piriform cortex, and cerebellum. SNpr represented the brain region with the highest microglia sensitivity, which became significant at 0.05 mg/kg. A recent study compared the effects of LPS in SNpr of C57BL6 mice between the doses of 1 and 5 mg/kg i.p. injections. It appeared that only the higher dose could sustain the chronic microglia activation (Zhao et al., 2020). Here, we examined whether microglia activation observed in CX3CR1^+/GFP 24 h after the LPS treatment can be sustained for the long term. Thirty days after LPS injection, microglia activation was not observed in CX3CR1 partially ablated mice. No differences could be observed in all the brain regions tested between LPS-treated and vehicle-treated mice (Figure 7B). This data implied that CX3CR1 partial ablation decreased sensitivity to microglia activation and sustainability in response to systemic LPS administration.

We have also investigated astrocytic heterogeneity, which represents another essential cell type in neuroinflammation. These cells cooperate with microglial cells in the neuroinflammatory event. Astrocytes could also have a pro-inflammatory toxic phenotype and harm the surrounding tissue (Miller, 2018; Linnerbauer et al., 2020).

Firstly, we examined the distribution of astrocytes using the glial fibrillary acid protein (GFAP) as a marker. Although these cells are distributed in all brain parenchyma, they express different levels of GFAP among the different regions. For this reason, it is not possible to equally mark these cells within all regions, and their heterogeneity could be highly influenced by the different expression of this marker between the different astrocytic populations. We analyzed the marked area by GFAP positive signal in C57BL6 mice in the vehicle condition. SNpr, hippocampus, and corpus callosum presented a significantly higher density of GFAP positive cells (Figure 8A). Moreover, the morphology of these cells appeared different among different brain regions.

Then, we compared C57BL6 mice treated with 5 mg/kg LPS and the vehicle control. This analysis showed a heterogeneous increase in GFAP signal between several brain regions. We observed a significant astrocyte increase in the hippocampus, SNpc, SNpr (Figures 8B,C), and corpus callosum (Supplementary Figure 14A), but not in the other areas. Further alterations were qualitatively observed. GFAP signal increases in the hypothalamus, but intraregional differences mitigate this increase. An increase in GFAP signal was also observed in other regions such as the cerebellum (Supplementary Figure 14B) and the thalamus. However, the GFAP intensity was below the threshold used in the analyses, and the increase was not detected.

Comparing GFAP-positive cells between C57BL6 and CX3CR1^+/GFP mice, we observed a similar GFAP pattern in the naive (non-LPS treated) mice (Figure 9A). The only exception was a significant increase in GFAP marked area in SNpr in CX3CR1^+/GFP mice. Further qualitative observations revealed that this increase appeared to involve mainly in the medial part of SNpr and it seems to be due to a higher astrocytes density (Figure 9C). In the end, we investigated astrocytic activation in CX3CR1^+/GFP mice treated with 5 mg/kg of LPS compared to the vehicle treated mice. This analysis detected an increase in GFAP signal only in the hippocampus (Figures 9B,D) but not in SNpr (Supplementary Figure 14C).

Our data reveal that astrocytes also present a brain region-specific activation and confirmed a lower inflammatory response in CXRCR1 partially ablated mice compared to the mice that express a physiological amount of this receptor.

We first investigated microglia heterogeneity in terms of the marked area, cell density, and morphology in control conditions between the 16 different brain regions analyzed. Our data showed that nine areas were similar, while the other seven appeared different.

To explore microglia activation under systemic inflammation, we used a mouse model treated with a single i.p. injection of 5 mg/kg LPS. This model showed differences in microglia activation in several brain regions and the strongest increase of Iba1 expression was observed in SNpr and entorhinal cortex. Further qualitative analyses about the alterations on the lysosomal system in several microglial populations showed the most evident changes in SNpr.

However, investigating Iba1 relative changes we observed that SNpr is less subject to an increase of microglia density compared to the entorhinal cortex and other brain regions. This finding agrees with the data obtained by Abellanas and colleagues. They showed that the midbrain is the only area compared with the striatum, cortex, and hippocampus where it is possible to detect an increase of anti-inflammatory cytokines transforming growth factor-beta (TGFβ) and IL10 expression 48 h after LPS administration (Lynch et al., 2004; Arimoto et al., 2007; Tesseur et al., 2017; Abellanas et al., 2019). Abellanas, in his work, suggested the presence of negative feedback that can down-regulate the neuroinflammatory response in this area. This feedback could have a crucial role in containing the inflammatory response under a certain limit to preserve the surrounding tissue. Comparing 5 mg/kg LPS and vehicles groups, we found higher microglia relative changes in regions characterized by low microglia density in control conditions. Whether these different responses represent compensatory mechanisms it is not clear, but it confirms the presence of heterogeneous inflammatory responses between different brain regions.

Using a dose-dependent paradigm, we detected different levels of microglia sensitivity to systemic TNFα/LPS. The mechanisms behind the different microglia sensitivity observed are unknown, and further studies are needed to investigate these aspects. Moreover, these mechanisms may differ between several regions. According to our data regarding the relative changes in Iba1, the most sensitive microglia population was in the hypothalamus. In this region, in proximity to the CVOs, we observed activated microglia cells at the lowest LPS concentrations. Here, the lack of BBB exposes this cell population to the systemic bloodstream and pro-inflammatory agents more effectively than in other brain regions. The effects observed in this region could derive from higher exposure to TNFα and LPS and the synergic activations of their receptors (Vargas-Caraveo et al., 2017, 2020; Furube et al., 2018). Interestingly, different works described that microglia cells in this region could modulate hypothalamic functions, such as regulating the body metabolism (Rosin and Kurrasch, 2019; Valdearcos et al., 2019).

Our data on the dose-dependent microglia response indicates that also the microglia population in the SNpr is particularly sensitive compared to most of other regions. Abellanas’s study showed that a higher portion of microglia cells in the SNpr express pro-inflammatory receptors such as TLR4 with respect to the other regions analyzed. Furthermore, De Biase summarized their findings arguing that microglia cells in SNpr have an “injury-responsive phenotype” (De Biase et al., 2017; Abellanas et al., 2019).

Our results indicate microglial ability to perceive and react to a systemic pro-inflammatory stimulus. More studies are needed to understand whether the pattern observed could also reflect microglial cells’ sensitivity when pro-inflammatory agents are within the brain parenchyma. An example could be represented by microglial sensitivity observed in the entorhinal cortex. Our findings agree with previous observations in Alzheimer’s disease (AD) models, where microglia activation in this cortex precedes all other regions as well as amyloid plaques deposition (Janelsins et al., 2005).

Interestingly, we observed evident CD68 changes at 0.05 mg/kg LPS only in SNpr compared with the hypothalamus and entorhinal cortex, which showed an increase at the higher doses. This observation confirmed different sensitivity in microglial activation among different brain regions. However, it also pointed out that several patterns could be found using different markers. These differences could be due to different microglia pro-inflammatory phenotypes and a specific protein’s role in a particular brain region. Recently, Zhao showed that chronic inflammation depends on IL1β expression, which is prevented in the nucleotide-binding domain and leucine-rich repeat receptor containing a pyrin domain 3 (NLRP3) and IL1β deficient mice (Zhao et al., 2020). Other authors showed different IL1β expression patterns under systemic inflammatory conditions in several brain areas (Silverman et al., 2014; Abellanas et al., 2019), suggesting that also chronic inflammation could differ throughout the brain. Our data shows that this is indeed the case. We compared 5 mg/kg LPS and relative control mice 30 days after treatments, and we found a slight increase in the microglia marked area only in some regions such as the amygdala, SNpr, PSM cortex, and piriform cortex. However, microglia cells partially lost the pro-inflammatory morphology shown 24 h after LPS injection, assuming an intermediate state. Altogether, our data showed that microglia heterogeneity leads to different inflammatory responses when systematically stimulated with TNFα/LPS. Recently, systemic inflammation has been under the spotlight concerning the COVID-19 pandemic. High levels of TNFα in the bloodstream and the development of chronic neurological symptoms were widely described in the past year (Del Valle et al., 2020; Feldmann et al., 2020; Robinson et al., 2020; Marshall, 2021). Interestingly, no trace of COVID-19 was found within the brain (Marshall, 2021). Whether these symptoms are related to the systemic “storm” of cytokines during the acute stage of COVID-19 infection and the subsequent neuroinflammatory activation is still not understood.

Most of the regions identified in our work with distinct microglia responses are also involved in the two most common neurodegenerative diseases. The entorhinal cortex, piriform cortex, amygdala, and NAc are all regions involved in AD pathology, while the SN is the region mainly affected in Parkinson’s disease (PD) and related disorders. Our data suggest that the different neuroinflammatory responses could partially influence the regional neurodegenerative pattern observed in different neurodegenerative diseases.

Neurodegeneration observed within the SNpc of PD patients and other related disorders are mainly associated with losing dopaminergic neurons. This neuronal population is more sensitive to reactive oxygen species (ROS), which exacerbate the detrimental effect of pathological agents such as α-synuclein or toxins (Drechsel and Patel, 2008; Dutta et al., 2008; Gao et al., 2008; Ramsey and Tansey, 2013; Rai and Singh, 2020; Rai et al., 2020). Moreover, it has been shown that ROS species, LPS, and pro-inflammatory cytokines such as TNFα and IL1β, can trigger the activation of several transcription factors, such as nuclear factor- kb and transcription factor EB (Brady et al., 2018; Rai et al., 2021). These transcription factors in turn regulate the expression of inflammatory cytokines and autophagy system, which is crucial for several neurodegenerative diseases, such as PD and AD (Cortes and La Spada, 2019; Rai et al., 2021). In the present study, we showed that the neuroinflammatory response in SNpr can induce a rapid alteration of dopaminergic dendrites in the medial SNpr 24 h post-LPS injection. These data indicate that the inflammatory process in SNpr could directly affect the neuronal population in the neighboring SNpc region. Moreover, our data showed that this decrease in dendrites is reversible and no longer visible 1 month after LPS administration. This finding is in line with observations made by Ong et al. (2021) regarding the increase in TH activity 1 week after peripheral LPS administration. The authors did not identify an apparent reason for this early increase in TH activity. Here, we showed that it could be related to the re-establishment of the dopaminergic dendritic profile in SNpr after the damages induced by the acute inflammatory response. SNpr consists of neurons less densely compacted compared to those in the SNpc and VTA, and the region is particularly enriched by neurites from other regions and glial cells. We observed that dopaminergic dendrites in this region surround microglia cells. An interesting characteristic described is that these dendrites release dopamine in the SNpr (Björklund and Lindvall, 1975; Geffen et al., 1976; Mercer et al., 1979; Cheramy et al., 1981), which was suggested to possesses a modulatory effect on the GABAergic neurons (Waszcak and Walters, 1983; Martin and Waszczak, 1996; Windels and Kiyatkin, 2006). However, other works described the ability of this neurotransmitter to modulate also microglia cell activity (Farber et al., 2005; Yan et al., 2015; Elgueta et al., 2017; Fan et al., 2018). In this region, the inflammatory response presented the most robust activation, one of the highest sensitivities to systemic inflammation, and the ability to maintain chronic microglial activation. Additionally, other evidence points to the fact that this cell population seems to be regulated by several feedback mechanisms. These characteristics highlight the complexity and the peculiarity of microglia cells in SNpr and bring us to consider a possible physiological role. Many studies showed that microglia cells are essential for synapse modeling within the brain (Kim et al., 2013; Ikegami et al., 2019; Crapser et al., 2021). Here, we observed a rapid and transient alteration of dopaminergic dendrites with further consequences on the striato-nigral bundle. We speculate that SNpr could represent the “bridge” between the immune system and basal ganglia circuitry, which may have a physiological role in modulating its structure and activity.

During the last two decades, an increased number of studies focused on the functions of the fractalkine axis CX3CL1-CX3CR1. CX3CL1 is constitutive and abundantly expressed in neurons, while its receptor is exclusively expressed in microglia cells in the CNS and other peripheral immune cells, such as dendritic cells (Pawelec et al., 2020). Two-photon in vivo imaging demonstrated microglial dynamics and its interaction with neuronal profiles, such as dendrites, axons, and synapses in mice expressing both CX3CR1^+/GFP and Thy1-YPF. Moreover, mice completely ablated for CX3CR1 presented aberrant synaptic connections between neurons, indicating the crucial role of microglia cells in synaptic modulation (Paolicelli et al., 2011).

Many studies also showed the role of the CX3C axis in physiology and in several neurodegenerative diseases (Biber et al., 2007; Lyons et al., 2009; Wolf et al., 2013). Unlike other chemokines, CX3C is more highly expressed in the CNS than in the periphery and, for this reason, it could represent a crucial pharmacological target for neurodegenerative diseases. However, contradictory data were obtained when this axis was perturbed in models of different neurodegenerative diseases (Pawelec et al., 2020). In models of stroke, epilepsy, and amyotrophic lateral sclerosis, a decrease of CX3CR1 or a rise of its ligand seem to be protective (Soriano et al., 2002; Cipriani et al., 2011; Yeo et al., 2011; Xu et al., 2012; Fumagalli et al., 2013; Tang et al., 2014; Liu et al., 2019). The deficiency of CX3CL1 or CX3CR1 in AD models reduces β-amyloid deposition but increases tau phosphorylation and aggregation, with a consequent detrimental effect on the behavioral and cognitive deficit (Bhaskar et al., 2010; Fuhrmann et al., 2010; Lee et al., 2010, 2014; Liu et al., 2010; Maphis et al., 2015; Merino et al., 2016). In PD models based on α-synuclein over-expression, the deficiency of this receptor has led to contradictory results. The increase of CX3CL1 induced protection to dopaminergic neurodegeneration (Pabon et al., 2011; Morganti et al., 2012; Nash et al., 2015; Thome et al., 2015; Castro-Sánchez et al., 2018). Pabon et al. (2011) showed that the administration of exogenous CX3CL1 in rats treated with 6-hydroxydopamine (6-OHDA) was protective. An absence of dopaminergic neurodegeneration was observed in CX3CR1 deficient mice treated with intranasal inoculation of 6-OHDA or 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP). On the contrary, extensive neurodegeneration was observed in CX3CR1 deficient mice treated with an i.p. injection of MPTP. They found a higher activation of astrocytes than that in control mice (Tristão et al., 2016).

Comparing mice partially and totally ablated of CX3CR1, Cardona et al. (2006) found that the complete ablation increased neuroinflammatory effects and neurodegeneration mediated by systemic LPS. However, they did not compare these models with mice expressing physiological levels of CX3CR1, and it is unclear whether the differences observed are due to a real increase in microglial response in mice totally ablated or by a decrease in mice that partially express this receptor.

Here, we showed that the partial ablation of CX3CR1 can significantly decrease microglia response when mice are treated with 5 mg/kg LPS and compared with mice that express physiological levels of CX3CR1. Moreover, chronic microglial activation was not sustained in all regions analyzed 30 days after LPS administration. These differences between partial and total ablation may be due to the partial decrease of CX3CR1 without altering the amount of CX3CL1, leading to changes in the ligand-receptor ratio, with consequences in their binding equilibrium. These data suggest the importance of CX3C axis integrity and that a higher amount of CX3CL1 respect CX3CR1 could depress the inflammatory response. Nevertheless, mice partially ablated of CX3CR1 can still respond to LPS stimulus in the acute experimental paradigm, which indicates the usefulness of such a model in microglia optical imaging. However, one should also be aware of the potential insensitivity to pro-inflammatory stimuli. These data also imply the CX3C axis as a potential pharmaceutical target in neurological and neurodegenerative diseases.

Astrocytes are another cell type that cooperates with microglia during a neuroinflammatory event. Several studies showed their involvement in several animal models, including those investigated in our study (Miller, 2018; Linnerbauer et al., 2020). To have a complete overview of the inflammatory heterogeneity within the brain, we also briefly investigated this cell type. Astrocytes appeared highly heterogeneous regarding GFAP expression, cell density, and morphological characteristics. SNpr, hippocampus, and corpus callosum represent the regions most densely populated by astrocytes. Moreover, astrocytes in these regions increase significantly under inflammatory conditions induced by 5 mg/kg of LPS.

Comparing C57BL6 with CX3CR1 partial ablated mice, we observed a significant higher GFAP positive signal in SNpr, which seems to be due to an increase in astrocytes density in the medial part of this region. Moreover, we found significant astrocytic increase only in the hippocampus by investigating the inflammatory changes induced by 5 mg/kg of LPS in CX3CR1^{+ ⁣/GFP} mice and relative control. However, this increase was significantly lower than the one observed in C57BL6, confirming that partial ablation of the CX3CR axis also decreases astrocytic activation.