Lesion-associated microglia and macrophages react with massive phagocytosis for debris clearance and wound healing after LPS-induced dopaminergic depletion

Background Neuroinammation contributes to neuronal degeneration in Parkinson’s disease (PD). However, how brain inammatory factors mediate the progression of neurodegeneration is still poorly understood. Experimental models of PD have shed light on the understanding of this phenomenon, but the exploration of inammation-driven models is necessary to better characterize this aspect of the disorder. The use of lipopolysaccharide (LPS) to induce a neuroinammation-mediated neuronal loss is useful to induce reliable elimination of dopaminergic neurons. Nevertheless, how this model parallels the PD-like neuroinammation is uncertain. Methods In the present work, we used the direct LPS stereotactic injection as a model inductor to eliminate dopaminergic neurons of the substantia nigra pars compacta (SNpc) in rats and reevaluated the microanatomy of inammatory reaction three and seven days after the insult. For this, we analyzed the tissue with high resolution confocal microscopy to assess the neuronal loss, the vulnerability of dopaminergic neurons, as well as the activation of lesion-associated microglia and macrophages (LAMMs) together with the visualization of their phagocytic capacity. In addition, we set up a co-culture of BV2 microglia and PC12 dopaminergic cells to understand the role of LPS-mediated neuroinammatory toxicity versus phagocytosis. Results High-resolution 3D histological examination revealed that, although LPS induced a reliable elimination of SNpc dopaminergic neurons, it also generated a massive neuroinammatory response. This inammation-mediated injury was characterized by a damaged parenchyma occupied by a vast population of LAMMs undertaking wound compaction and scar formation. LAMMs tiled the entire lesion and engaged in long-standing phagocytic activity to resolve the injury. Additionally, modeling LPS inammation in a cell culture system helped to understand the role of phagocytosis and cytotoxicity in dopaminergic degeneration and indicated that LAMM-mediated toxicity and phagocytosis coexist during LPS-mediated dopaminergic in a rat model of dopaminergic degeneration and analyzed in detail with cellular resolution the effects on the mesencephalic parenchyma. We observed that LPS-induced dopaminergic loss causes severe tissue damage characterized by aggressive microglial activation and massive macrophage inltration within a corralled lesion showing high phagocytic capacity of dopaminergic debris, wound compaction and scar formation. These results indicate that although dopaminergic loss is reproducible and reliable, the inammation caused by the direct injection of LPS does not recapitulate the typical inammatory response observed in ageing and PD.


Background
Parkinson's disease (PD) is characterized by a dramatic loss of dopaminergic neurons of the substantia nigra pars compacta (SNpc) [1,2]. The cause of this neurodegeneration is still under scrutiny. Although the contribution of glial-mediated neuroin ammation is now generally accepted, many questions remain unanswered regarding the particular role of in ammatory mediators on dopaminergic neurons [3]. Thus, understanding neuroin ammatory response in PD is essential to design alternative therapeutic interventions [4,5]. For this to be elucidated, experimental models of parkinsonian-like neuroin ammation are necessary to understand the action of local glial cells, as well as potential in ltrated immune cells, on neurons [6].
The gram-negative bacterial membrane compound lipopolysaccharide (LPS) has long been used to promote neuroin ammatory responses in vivo [13][14][15] and in vitro [16,17]. Particularly in vitro, together with interferon gamma (IFN-γ), it is canonically used to skew cells of macrophage lineage to a proin ammatory phenotype [18,19]. In vivo, the direct injection of LPS in the central nervous system (CNS) in rodents has commonly been used to study neuroin ammation, particularly as an in ammationinduced PD model when injected directly into dopaminergic areas, speci cally the SNpc [6,20]. Inducing this proin ammatory phenotype in the CNS has been particularly useful to understand the harmful effects on neurons. Moreover, the hypothesis that bacterial-derived, LPS-mediated neuroin ammation could be behind PD pathogenesis makes this approach increasingly relevant [15].
It is clear that intranigral LPS injection generates reliable neuronal loss, which can be suitable for particular scienti c questions [6], but whether the in ammation of this model is comparable to PD-like in ammation is still unclear.
In the present work, we revisited the neuroin ammatory response generated by the direct injection of LPS in the SNpc in a rat model of dopaminergic degeneration and analyzed in detail with cellular resolution the effects on the mesencephalic parenchyma. We observed that LPS-induced dopaminergic loss causes severe tissue damage characterized by aggressive microglial activation and massive macrophage in ltration within a corralled lesion showing high phagocytic capacity of dopaminergic debris, wound compaction and scar formation. These results indicate that although dopaminergic loss is reproducible and reliable, the in ammation caused by the direct injection of LPS does not recapitulate the typical in ammatory response observed in ageing and PD.

Animals and surgeries
Thirty-six adult male Wistar rats (200-300 g, 2.5 months) were used for this study. Rats were housed in Plexiglas cages (33 x 44 x 20 cm) at the animal housing facilities (n = 5 per cage) under a 12/12 h light/dark cycle (light on at 08:00) at room temperature (RT) with water and food ad libitum. All experimental procedures were performed according to the International Guide for the Care and Use of Laboratory Animals (National Institute of Health, 1999), register # ICS-2017-007 from ethical committee of Instituto de Ciencias de la Salud from Universidad Veracruzana, and the NOM-062-ZOO-1999 and NOM-087-ECOL-SSA1-2002 from Mexican legislation. All efforts were made to minimize animal discomfort during the study.
Two sets of experiments were planned according to the different time window between the intranigral injection of LPS and the euthanasia. Rats were injected either with an intracranial injection of LPS or saline in the SNpc. The groups of animals were then perfused three or seven days after the LPS.
Three groups of animals (n = 6) were used for each set of experiments: group 1: sham (dura mater removed); group 2: saline (saline solution intranigral injection) and group 3: LPS (intranigral LPS injection). Animals were injected intranigrally with 10 µg/2 µl of LPS or saline solution at the same volume.
Intracranial surgery was done as follows. Rats were anesthetized with a single intraperitoneal dose of ketamine/xylazine (10 mg/kg/8 mg/kg) and received an intranigral injection of LPS from Escherichia Coli 055: B5) (Sigma-Aldrich; St. Louis, MO, USA) or saline. We used the following coordinates: AP + 3.0 mm, from the interaural point; ML + 2.8 mm from the intraparietal suture and DV -6.9 mm from the dura mater.
The injection ow was 0.2 µL/min maintained by a perfusion micropump (Stoelting, Wood Dale, IL, USA) according to previous procedure [21]. Once the LPS or saline solution was injected, according to each case, the needle was left inside ve minutes to allow complete diffusion, then was withdrawn slowly after which the animals were sutured.
Perfusion. Three or seven days after surgery, animals were deeply anesthetized with 100 mg/kg i.p. of During the perfusion, only the head of the animal was favored, so it was necessary to mechanically occlude the descending aortic artery.
Extraction of brain tissue. Once the perfusion was completed, the brains were immediately removed and post-xed for 24 hours at 4°C in 4% paraformaldehyde. After this, they were placed in 30% sucrose solution in 0.1 M PBS under the same storage conditions. Subsequently, brains were cut in 35 µm-thick serial sections in a cryostat (Leica CM1520, Leica Inc, Germany) in coronal plane to obtain a representative sample of all levels of the SNpc. The 35 µm slices were collected starting from the anteriormost portion of the SNpc and sequentially placed with a soft-bristled brush inside a sterile 24-well box (Nalgene), previously lled with 0.1 M PBS. Sections were placed one by one in six series; this allowed 9-10 representative sections of SNpc to be obtained for each well. The tissue was stored in cryoprotective solution at − 20°C until used.
Immuno uorescence. Tissues were washed three times for 10 minutes with PBS containing 0.13 M NaCl, 0.010 M Na 2 HPO 4 and 0.002 M NaH 2 PO 4 at pH 7.4. Next, they were immersed in citrate buffer at 60-80°C for 20 minutes and rinsed again with PBS for ve minutes and after with TBS-Triton 0.05%. Subsequently, they were incubated for one hour in 10% horse serum + 0.1% sodium azide for blocking nonspeci c sites, then washed ve minutes with horse serum 1% + sodium azide 0.1% and incubated with primary antibodies [anti-TH made in rabbit, (1:1000) for identi cation of dopaminergic neurons; anti-OX42 made in mouse (1:500) and Iba-1 (1:500) produced in rabbit for microglial cells; anti-NeuN made in mouse (1:300), to detect neuronal nuclei] in 1% horse serum solution + 0.1% sodium azide for 48 hours at room temperature (RT). After that, three 10-minute washes were done with TBS-Triton 0.05% and tissues were incubated with uorescent secondary antibodies: Alexa Fluor 488 made in goat anti-rabbit IgG (H + L), Alexa Fluor 488 made in goat anti-mouse IgG (H + L) and Alexa Fluor 555 made in goat anti-mouse IgG (H + L), for four hours (at a concentration of 1:1000 in horse serum 1% + sodium azide 0.1%) and protected from light. Next, three washes were done with PBS for 10 minutes and tissues were incubated with DAPI (1:1000) for 30 minutes to label the cell nuclei. They were rinsed with PBS three times for 10 minutes and nally were mounted on non-gelatinized slides and protected with antifading reagent (Pro-Long Gold) and coverslips.
Image Capture And Quanti cations TH and Iba-1 immunoreactivity were visualized under the multichannel epi uorescence microscope (Eclipse 90i, Nikon), adapted to a high-resolution camera (Nikon DXM1200F), using software to systematically capture images (ACT-1). Representative mosaic images of the SNpc were also taken with a high-resolution camera with a microscope with an automated stage (Eclipse TE2000-E, Nikon) covering the section with systematic images according to x and y coordinates. Image mosaics were built using a stitching tool plugin (Image J). The number of cells was quanti ed using a specialized software (Image J) and considering the physical dissector as stereological criteria.

Confocal Microscopy
TH/OX42 and TH/NeuN positive cells were visualized using a high-resolution confocal microscope (Zeiss LSM 700) with a 20x objective and capturing systematic images of the SNpc with specialized software (ZEN 2010, Zeiss). For quanti cations, TH + /OX42 + or TH + /NeuN + immuno uorescence images were captured in 3D stacks con gured with several optical sections with a 1-µm optical interval and analyzed with specialized software. In addition, DAPI was also captured as counterstaining. For numerical quanti cations, including number of cells and the number phagocytic events, we used object counter tools (Image J) and optical dissector stereological criteria. To quantify volumes and visualize phagocytosis events, we used different three-dimensional rendering software (Imaris, Bitplane and Illucida FX, Los Angeles) to create isosurfaces and blend views. To visualize Voronoi partition and crystallization, we also used specialized software applications (ImageJ and Pixelmator). Analysis and visualization of relative orescence with 5 ramps was also done with speci c application software (Image J).

Cell Culture Experiments
Cell lines PC12 and BV2 cells were obtained from the INc-UAB institutional repository. BV2 is an immortalized murine microglia cell line, which has been widely used and successfully evaluated in response to LPS and IFN-γ [22]. Both cell lines were maintained at 37 C with 5% CO 2 in their corresponding media, which for the PC12 cell line was DMEM supplemented with 7% fetal bovine serum, 7% horse serum and 0.2% penicillin-streptomycin and for the BV2 cell line was RPMI supplemented with 10% fetal bovine serum and 0.1% penicillin-streptomycin. For the different experiments, PC12 and/or BV2 were placed in a 24well culture plate and grown for 24 hours then treated with different compounds such as LPS at different concentrations, IFN-γ, or co-cultured with PC12. Twenty-four hours after the treatment, the supernatant was collected to conduct the Griess assay or to treat subsequent cultures. At the end of the experiments, cells were xed with 4% paraformaldehyde in PBS for immunocyto uorescence.

Immunocyto uorescence
The immunocyto uorescence was performed on cell cultures to simultaneously detect TH and CD11b. Cells were washed with PBS, then permeabilized with 0.02% saponin/PBS at RT and later rewashed with PBS1X. Nonspeci c sites were rst blocked with PBS1X, 0.01% saponin and 0.075% glycine and later nonspeci c sites were again blocked with PBS1X, 0.01% saponin, 0.075% glycine and 5% BSA. Coverslips containing the cells were placed in a humid chamber with 12 µL of PBS1X, 0.01% saponin, 1% BSA and the primary antibody overnight. To detect TH and CD11b, the primary antibodies were anti-TH (sheep, 1:500) (Merck; Darmstadt, Germany) and anti-CD11b (mouse, 1:1000) (Abcam; Cambridge, United Kingdom), respectively. The following day coverslips were placed again in the 24-well plate, cells were washed with PBS 1X before being placed again in a humid chamber with PBS1X, 0.01% saponin, 1% BSA and the secondary antibody for 45 minutes. In order to detect TH, the secondary antibody was donkey anti-sheep (1:1000) in green uorescence (Alexa Fluor 488) (Life Technologies; Carlsbad, CA, USA), and for CD11b detection the secondary antibody was goat anti-mouse (1:500) in red uorescence (Alexa Fluor 555) (Bio-Rad Company; Berkeley, California, USA). Coverslips were later returned to the 24-well plate to be washed with PBS1X before adding 400 µL per well of the nucleus marker, DAPI (1:1000) (Life technologies; Carlsbad, CA, USA) in PBS1X. One more wash with PBS 1X was done before the coverslips were mounted on a glass slides using antifade reagent (Prolong Gold, Life Technologies; Carlsbad, CA, USA).
For every experimental condition, a secondary antibody control was also performed, leaving one coverslip per condition without any primary antibodies.

Immunocyto uorescence Quanti cation
To quantify the number of cells in the co-culture and its controls, we used an image analysis protocol. Each coverslip was imaged using a uorescence microscope (Nikon Eclipse 90i) attached to a DXM 1200F digital camera and version 2.70 of the ACT-1 software (Nikon Corporation). With this system, we obtained 30 images per condition at 20x. Two quanti cation methods were used for the immunocyto uorescence labeling. One consisted of counting the number of cells, either PC12 or BV2, following stereological criteria and using a specialized software (Image J version 1.47, NIH, USA). The other method consisted of measuring the area of BV2 (CD11b) to estimate the activation, also using the same software (Image J version 1.47, NIH, USA).

Determination Of Nitrites By Griess Assay
The determination of nitrites by Griess assay is based on a chemical reaction in which nitric oxide in the presence of an aromatic amine produces different compounds in sequence, resulting in a pink-colored compound, easily detectable by spectrophotometry. First, the calibration curve was prepared with known concentrations of NaNO 2 (100 µM, 50 µM, 25 µM, 12.5 µM, 6.25 µM, 3.125 µM, 1.5625 µM and 0.78125 µM), and 100 µL of each solution was added into a 96-well plate for the calibration curve, as was 100 µL of the experiment samples. Next, 100 µL of the Griess reagent (0.1 g in 2.5 mL milli-Q water) was added to every well. Duplicates were done for each condition. The plate was then incubated in darkness at RT for 15 minutes. Then the plate was read with specialized microplate reader software (KC Junior, Kansas City, MO, USA) and the nitrite ion equivalents of the samples of interest were calculated according to the calibration curve prepared in the same 96-well plate.

Statistics
The results were expressed as mean ± standard error of the mean (SEM), which were calculated with suitable software (Sigma STAT or Excel, Microsoft O ce MSO, Redmond, Washington, USA). All the data were statistically analyzed with suitable software (R commander package, R software; version 3.5.2, Free Software Foundation's GNU General Public License), using one-way analysis of variance (ANOVA) in pairwise comparisons of means mode. The value of p < 0.05 was the criterion to establish differences between means.

Results
To corroborate the effectiveness of LPS as a dopaminergic neurotoxin and potential experimental model for PD, rst, we analyzed the consistency of a direct LPS injection inducing dopaminergic neuronal loss in the SNpc (Fig. 1A-C). We examined an early time window of three days to analyze the initial neuroin ammatory response and related neurodegeneration, and also at seven days to observe the potential recovery of the lesion. Precise intra-parenchymal LPS injection originated dopaminergic depletion in the targeted site, observable both three and seven days after the injection, with no apparent effect on the contralateral side (Fig. 1D). Mainly dopaminergic neurons were eliminated but, interestingly, non-dopaminergic neurons, quanti ed as NeuN + with no TH expression, also showed a similar percentage of reduction (Fig. 1E-G) questioning the speci city and the particular vulnerability of dopaminergic neurons in this scenario. At seven days, both dopaminergic and non-dopaminergic neuronal loss seemed to be slightly higher, suggesting a progression of the neuronal degeneration.
Most importantly, LPS-induced dopaminergic neuronal loss was associated with a massive accumulation of lesion-associated microglia and macrophages (LAMMs), evidenced by the marker OX42 at the injection site ( Fig. 2A, B). Reduction of dopaminergic area on the ipsilateral side was concomitant with the populating of a vast amount of OX42 + cells compared with the contralateral side (Fig. 2B). This phenomenon could be seen in LPS-injected areas both at three and seven days after injection, whereas only scant levels of activation could be seen after saline injection (Fig. 2C). Higher resolution images of the OX42 + cells in the SNpc revealed drastic morphological changes in LPS-injected parenchyma (Fig. 2C). Quanti cation of the number of OX42 + cells in the SNpc shown a dramatic increase in LAMMs in the site of injection three days after LPS delivery. Moreover, this increase persisted at seven days, together with a mild increase observed on the contralateral side (Fig. 2D). Skeletonized OX42 + cells from high-resolution images illustrate the representative morphological changes after LPS delivery. While sham and saline-injected areas show recognizable microglial morphology, LPS-injected areas show rounded-shape cells, compatible with vast monocyte in ltrate (Fig. 2E).
Examination of sections, in fact, showed a massive in ltrate of OX42 + cells from vascular structures in the SNpc area (Supplementary gure S1), demonstrating the contribution of blood-in ltrated macrophages to the LAMM population at the injection site.
A closer look at the anatomically enclosed, LPS-induced lesion site showed a densely packed LAMM area covering the entire injury (Fig. 3A). Image analysis comparison of LPS-injured and intact SNpc, using Voronoi partition or crystallization, showed a vast region of tiling at the lesion site (Fig. 3B) lling and compacting the injury. The 3D blended view also showed a honeycomb-like tiling at the LPS-targeted parenchyma (Fig. 3C). Cells were grouped membrane-to-membrane along the SNpc.
Because LPS injection caused a dramatic dopaminergic neuronal loss accompanied by a massive LAMM corralling in tissue, we designed an in vitro platform to understand the roles of microglial-mediated neuroin ammation in this scenario. First, to determine the direct toxicity of LPS, we treated dopaminergic cells with increasing concentrations, and we observed only a slight dopaminergic cell loss in some instances (Fig. 4A). Next, we separately treated microglia with the same protocol, detecting an increase in cell numbers, but a relatively low or absent in ammatory response as determined by nitrite release and microglial cell size (Fig. 4B-D).
Since a sole LPS administration does not explain the dramatic dopaminergic cell elimination, we cocultured microglia and dopaminergic cells to determine the role of the intercellular interactions. We observed a dramatic loss when combining both cell types (Fig. 4E and F), with no apparent changes when adding increasing LPS concentrations, demonstrating that the presence of microglia is essential for LPS-induced dopaminergic cell elimination (Fig. 4G and H).
Because this microglial-mediated elimination was spontaneous and apparently without an evident neuroin ammatory stimulus, we wanted to con rm that the interaction with PC12 dopaminergic cells in vitro does not trigger an in ammatory response. To do so, we analyzed microglial activation following PC12 presentation and compared it with the canonical proin ammatory stimulus LPS and IFN-γ. The LPS/IFN-γ combination increased microglial activation parameters in contrast with interaction solely with PC12, demonstrating that the encounter with dopaminergic cells does not trigger this neuro-in ammatory response (Fig. 5A-C).
Knowing that the presence of microglia is critical for the elimination of dopaminergic cells, we postulated that phagocytosis may be a major event in eliminating dopaminergic cell targets. However, to understand the importance of the microglial-mediated toxic environment, we cultured dopaminergic cells with supernatant coming from naïve or activated microglia, and we observed a substantial reduction in dopaminergic cell numbers when adding supernatant from LPS/IFN-γ-activated microglia, demonstrating that the toxic environment is also crucial for neuroin ammatory-mediated dopaminergic elimination ( Fig. 5D-F).
Because phagocytosis may be critical for injury healing of LPS-mediated dopaminergic elimination, we analyzed with high-resolution detail the phagocytic capacity of LAMMs at the rats' lesion sites. First, we identi ed a constellation pattern of TH + material in the LPS lesion area, with barely recognizable dopaminergic neuronal bodies (Fig. 6A). Alpha blending and 3D reconstructions revealed a large amount of TH + debris, within the volumetric space of the lesion, some of which was contained inside OX42 + LAMMs (Fig. 6B). Certain TH + material was identi able as neuronal bodies completely surrounded by LAMMs ( Fig. 6C and Supplementary video 1). High-resolution confocal Z scans con rmed that TH + material was completely engulfed by LAMMs at the injection site, and 3D reconstructions facilitated the visualization of these phagocytic events ( Fig. 6D and Supplementary video 2). Interestingly, the volume of TH + debris, although reduced, persisted even seven days after the lesion (Supplementary gure S2) together with an important volumetric reduction of OX42 + membranes (Supplementary gure S2), indicating a slow and prolonged initiation of the wound healing process.
In addition, although rare events, pyknotic nuclei could be seen inside OX42 + cells, suggesting potential instances of primary phagocytosis (Fig. 6F). Quanti cation of phagocytic events revealed high phagocytic activity at three days with a signi cant reduction of phagocytic events seven days after the lesion (Fig. 6G), indicating the evolving resolution of the injury. However, persistent phagocytic activity of LAMMs after LPS lesion was still detectable seven days after the injection with similar percentage of phagocytizing as on day three (Fig. 6G) evincing that elimination of dopaminergic cell fragments is a long-lasting process during the course of the wound healing.

Discussion
In the present study, we show that LPS is an effective compound producing a lesion in the brain parenchyma that effectively depletes dopaminergic neurons. Parallel to this lesion, a massive LAMM activation occurs within the wounded area, populating the entire lesion membrane-to-membrane and showing persistent phagocytosis of dopaminergic debris. This wound compaction scenario juxtaposes the in ammatory landscape seen in ageing and PD [10] or MPTP insult in humans and non-human primates [11,23], where microglial cells appear activated within a preserved and recognizable nerve tissue structure. In the case of rodent PD models, although a tangential in ltration of monocytes/macrophages may be found, the anatomical architecture of the parenchyma is not severely affected. MPTP insult in mice only causes an early (24h) and transient monocyte in ltration not altering the outcome of dopaminergic degeneration [24], microglia being the main cell phagocytizing dopaminergic neurons [25]. In 6-OHDA-injected rats, the situation is quite similar [8,26], with a preserved structure and limited monocyte in ltration [27].
Although previous reports highlight the utility of using the direct LPS injection rat substantia nigra to the rat's [28], because of the evident in ammatory response, the parallelism with PD is still questionable. The direct contact of LPS with the nerve tissue generates an in ammatory-mediated damage that signi cantly deteriorates the parenchymal structure. The resulting injury is so severe that some other structural proteins such as glial brillary acidic protein (GFAP) may not even be detected [28]. Thus, a fair comparison between 6-OHDA and LPS injection is hard to convey. The only report comparing both models uses an indirect approach, studying the LPS-induced neuroin ammation in the SNpc after a striatal injection [29]. This is actually the most adequate approach for a PD-like in ammatory reaction since injecting into the striatum avoids the direct damage caused by LPS in the SNpc but generates an indirect microglial activation while structurally preserving the area of interest [6].
The phagocytic process observed is also very different between the classical PD models and LPS. MPTP in mice induces a very speci c reaction against intoxicated dopaminergic neurons engul ng entire neuronal bodies, enclosing pyknotic nuclei [30]. In contrast, in our images here, the phagocytosis of pyknotic nuclei is rare. Rather, constellations of dopaminergic debris, which are not observable after MPTP, are being phagocytized by a large number of LAMMs in a process of clearance.
Although the damage generated by direct LPS injection is patent, the mechanisms by which it occurs are to be further investigated. Our co-cultures shed light on this phenomenon, indicating that sole delivery of LPS does not fully explain the elimination of dopaminergic toxicity, but requires the interaction of microglia for effective elimination. Probably, the presence of TLR4 receptors in myeloid cells may in part clarify this outcome, microglia being the major TLR4-expressing cell in the nerve tissue in contrast with neurons [31]. This is consistent with the idea that a massive LPS delivery within the CNS parenchyma triggers a LAMM-mediated in ammation that causes a large neuronal loss. This elimination occurs with the combination of cytotoxicity and phagocytosis, as supported by the in vitro results. This indicates that the exacerbated response of the host is key to the elimination of neurons rather than the direct toxicity of the endotoxin, consistent with previous reports [15,32].
Our results pinpoint phagocytosis as part of the wound compaction and healing of direct LPS-dependent lesion in the CNS. Consistent with our in vivo images, this phagocytosis is evident and abundant three days after the lesion and, although reduced, remains persistent at least seven days after the LPS insult.
In conclusion, it seems that microglial-mediated neuroin ammation, combining cytotoxicity and phagocytosis, is essential for the dopaminergic elimination induced by LPS. However, the direct injection of LPS into the CNS generates such a massive wound which signi cantly alters the structure of the tissue and puts in question the parallelism with the idiopathic parkinsonian in ammatory response. This group of characteristics seems to be far from the PD-like in ammation, rather being similar to other recently described scenarios for massive nerve tissue lesions [33], where clearance of debris, wound healing and compaction, mediated by LAMMs, contribute to the tissue repair.

Conclusions
Our results demonstrate that direct LPS-mediated lesion in the SNpc triggers a massive in ammatory response, characterized by LAMM tiling and phagocytosis that contributes to the wound compaction and healing. Although this model may be useful to generate a drastic depletion of dopaminergic neurons, assumptions regarding the contribution of neuroin ammation in PD from this scenario should be drawn cautiously. Novel approaches and techniques need to be further explored and modeled to better understand the neuroin ammatory contribution to dopaminergic neurodegeneration in PD.