ORIGINAL RESEARCH published: 20 November 2020 doi: 10.3389/fncel.2020.567404 Differential Roles of TREM2+ Microglia in Anterograde and Retrograde Axonal Injury Models Gemma Manich † , Ariadna Regina Gómez-López † , Beatriz Almolda*, Nàdia Villacampa, Mireia Recasens, Kalpana Shrivastava, Berta González and Bernardo Castellano Department of Cell Biology, Physiology, and Immunology, Institute of Neuroscience, Universitat Autònoma De Barcelona, Barcelona, Spain Microglia are the main immune cells of the central nervous system (CNS), and they are devoted to the active surveillance of the CNS during homeostasis and disease. In the last years, the microglial receptor Triggering Receptor Expressed on Myeloid cells-2 (TREM2) Edited by: has been defined to mediate several microglial functions, including phagocytosis, Veronica Ines Brito, University of Barcelona, Spain survival, proliferation, and migration, and to be a key regulator of a new common Reviewed by: microglial signature induced under neurodegenerative conditions and aging, also known Andrew David Greenhalgh, as disease-associated microglia (DAM). Although microglial TREM2 has been mainly The University of Manchester, United Kingdom studied in chronic neurodegenerative diseases, few studies address its regulation and Audrey Lafrenaye, functions in acute inflammatory injuries. In this context, the present work aims to study Virginia Commonwealth University, the regulation of TREM2 and its functions after reparative axonal injuries, using two-well United States established animal models of anterograde and retrograde neuronal degeneration: the *Correspondence: Beatriz Almolda perforant pathway transection (PPT) and the facial nerve axotomy (FNA). Our results beatriz.almolda@uab.cat indicate the appearance of a subpopulation of microglia expressing TREM2 after both † These authors have contributed anterograde and retrograde axonal injury. TREM2+ microglia were not directly related to equally to this work proliferation, instead, they were associated with specific recognition and/or phagocytosis Specialty section: of myelin and degenerating neurons, as assessed by immunohistochemistry and flow This article was submitted to Non-Neuronal Cells, cytometry. Characterization of TREM2+ microglia showed expression of CD16/32, a section of the journal CD68, and occasional Galectin-3. However, specific singularities within each model were Frontiers in Cellular Neuroscience observed in P2RY12 expression, which was only downregulated after PPT, and in ApoE, Received: 29 May 2020 where de novo expression was detected only in TREM2+ microglia after FNA. Finally, we Accepted: 14 October 2020 report that the pro-inflammatory or anti-inflammatory cytokine microenvironment, which Published: 20 November 2020 may affect phagocytosis, did not directly modify the induction of TREM2+ subpopulation Citation: in any injury model, although it changed TREM2 levels due to modification of the Manich G, Gómez-López AR, Almolda B, Villacampa N, microglial activation pattern. In conclusion, we describe a unique TREM2+ microglial Recasens M, Shrivastava K, subpopulation induced after axonal injury, which is directly associated with phagocytosis González B and Castellano B (2020) Differential Roles of TREM2+ of specific cell remnants and show different phenotypes, depending on the microglial Microglia in Anterograde and activation status and the degree of tissue injury. Retrograde Axonal Injury Models. Front. Cell. Neurosci. 14:567404. Keywords: IL-10 (interleukin-10), IL-6 (interleukin-6), phagocytosis, proliferation, neuroinflammation, microglial doi: 10.3389/fncel.2020.567404 clusters, P2RY12, APOE Frontiers in Cellular Neuroscience | www.frontiersin.org 1 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury GRAPHICAL ABSTRACT | In the left, after an anterograde axonal injury such as the perforant pathway transection (PPT), TREM2 is rapidly upregulated in a subpopulation of microglia located in the molecular layer. TREM2+ microglia, which can be proliferative, show co-localization with phagocytic CD68 and CD16/32, occasionally with TREM2 phagocytic ligand Galectin-3 (Gal-3), and demonstrates an important P2RY12 downregulation. In the right, after facial nerve axotomy (FNA)—a retrograde axonal injury—TREM2+ microglia is upregulated at late time-points, coinciding with the formation of phagocytic clustered microglia. Microglia clusters can be CD68+ and CD16/32+, may show expression of TREM2 ligands ApoE and Galectin-3, both molecules also observed in the surface of facial motor neurons, and can express variable levels of P2RY12. Overall, our results suggest that TREM2+ microglia is a subpopulation that phagocytes more efficiently specific cell remnants, mainly myelin and degenerating neurons, and its expression prepares the way for injury repair. INTRODUCTION conditions or in aging (Gratuze et al., 2018). A plethora of ligands bind to TREM2, including anionic ligands, such as phospholipids Microglial cells are the principal myeloid cells within the or sulfatides, lipoproteins like ApoE, β-amyloid, and also DNA central nervous system (CNS), which play an immune (reviewed in Kober and Brett, 2017). Upon ligand binding, surveillance function under homeostasis and after damage TREM2 interacts with DAP12 and results in a wide range of or pathological insult (Colonna and Butovsky, 2017; Wolf functions, including proliferation, migration, pro-survival signal, et al., 2017). A molecule that gained attention in the last lipid sensing, phagocytosis, and energy metabolism (reviewed in years by its involvement in the modulation of microglia Painter et al., 2015; Jay et al., 2017b), mainly aimed at containing is the Triggering Receptor Expressed on Myeloid cells 2 and removing apoptotic or degenerated cells produced during (TREM2), an innate transmembrane type I receptor of the neuronal damage (Takahashi et al., 2005, 2007; Hsieh et al., immunoglobulin superfamily that is expressed only by myeloid 2009; Krasemann et al., 2017; Deczkowska et al., 2018). Recently, cells (Colonna, 2003). Several studies have strongly linked single-cell RNA-sequencing analysis in the CNS tissue linked variants in the TREM2 gene with increased risk of developing TREM2 with the differentiation of a newly identified specific Alzheimer’s disease (AD), fronto-temporal dementia, and microglial subtype appearing in mice in neurodegenerative Nasu-Hakola disease and, to a lesser extent, Parkinson’s disease conditions and aging, the so-called disease-associated microglia and amyotrophic lateral sclerosis (reviewed in Jay et al., 2017b). (DAM; Keren-Shaul et al., 2017; Deczkowska et al., 2018) In the brain, TREM2 is exclusively expressed by microglia, or microglia associated to neurodegeneration (Krasemann with variations depending on the CNS region (Schmid et al., et al., 2017). These microglia play a key role in chronic 2002), and is generally upregulated in vivo upon inflammatory neurodegenerative conditions and show a unique transcriptional Frontiers in Cellular Neuroscience | www.frontiersin.org 2 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury and functional signature highly differing from homeostatic Barcelona. Every effort was made to minimize the number of microglia, characterized by the overexpression of other genes, animals used to produce reliable scientific data, as well as animal such as Clec7a, Lgals3, ApoE, and the downregulation of suffering and pain. microglial homeostatic genes, such as P2ry12 or Tmem119. The role of TREM2 has been mainly reported in chronic Perforant Pathway Transection and experimental models of AD and other chronic disorders—like Experimental Groups experimental autoimmune encephalomyelitis (EAE, reviewed in In the first part of the study, 84 WT animals were used, Karanfilian et al., 2020)—and, to a lesser extent, in different and in the second part of this study, performed with Tg acute neuroinflammatory conditions, including cuprizone- animals, 27 GFAP-IL6Tg, 45 GFAP-IL10Tg, and a total of induced demyelination, stroke, and traumatic brain injury, 72 corresponding WT mice littermates were used. Animals were among others (reviewed in Deming et al., 2018). However, subjected to wire-knife unilateral PPT, as already described in little is known about its expression after axonal CNS lesions, Jensen et al. (1999). Briefly, animals were anesthetized with an and even less about the role of TREM2 in those lesions intraperitoneal injection of a solution of ketamine (80 mg/kg) that are associated with a reparative process. Indeed, studies and xylazine (20 mg/kg) at the dose of 0.01 ml/g body weight. on microglial TREM2 function after spinal nerve injury Anesthetized mice were placed in a stereotaxic device (Kopf (Kobayashi et al., 2016) showed an upregulation of TREM2, Instrumentsr ) and a small window in the skull was created by which was also detected in the microglial transcriptome drilling made in the left side of the skull (4.6 mm dorsal to after facial nerve injury (Tay et al., 2018). Moreover, in Bregma and 2.5 mm laterally). A folded wire-knife (McHugh these and other studies based on TREM2KO and DAP12KO Milleux, m121) was inserted at an angle of 15◦ anterior and mice, reduced secretion of pro-inflammatory cytokines and 10◦ lateral. The knife was unfolded at 3.6 mm ventrally and a lower neuronal death rate were reported after axotomy the perforant pathway—formed by fibers from neurons placed (Kobayashi et al., 2015; Krasemann et al., 2017), pointing in layers II and III from the entorhinal cortex projecting to towards an important role of TREM2 in the modulation of the hippocampus—was transected retracting the knife 3.3 mm. the inflammatory response and the resolution of nerve injury Finally, the knife was folded and removed from the brain. After (Kobayashi et al., 2016). surgery, the skin was sutured with 2–0 silk and the wound was In this context, the main objective of our work is to study cleaned with iodine. Non-lesioned (NL) and lesioned animals TREM2 expression and its function in two well-established were distributed in different experimental groups and euthanized models of anterograde and retrograde axonal degeneration at 2, 3, 7, 14, and 21 days post-lesion (dpl). —with an associated reparative process—: the perforant pathway transection (PPT: transection of the principal pathway Facial Nerve Axotomy and Experimental that connects the entorhinal cortex with the hippocampus) Groups and the facial nerve axotomy (FNA), respectively. Also, we In the first part of this work, 110 WT animals were used, and explored whether changes in microglia activation by local for the second part of this study, specifically performed with pro-inflammatory or anti-inflammatory microenvironments, Tg animals, 32 GFAP-IL6Tg, 32 GFAP-IL10Tg, and a total of using transgenic mice with IL-6 or IL-10 overproduction 64 corresponding WT mice littermates were used. Animals were (Almolda et al., 2014; Villacampa et al., 2015; Recasens et al., anesthetized with a solution of ketamine (80 mg/kg) and xylazine 2019), are able to modulate TREM2 expression after PPT (20 mg/kg) injected intraperitoneally at a dose of 0.01 ml/g and and FNA. underwent a unilateral FNA as described in Almolda et al. (2014). The skin behind the right ear was shaved and cleansed with 70% ethanol. A small incision was made in the skin and both the MATERIALS AND METHODS trapezius and the anterior digastric muscles were gently separated to expose the right facial nerve. One millimeter of the facial Animals nerve main branch was resected at the level of the stylomastoid GFAP-IL6 transgenic (Tg) mice, GFAP-IL10Tg, and their foramen. Following the surgery, the skin was sutured with corresponding wild-type (WT) C57BL/6 littermates from both 5–0 nylon. Corneal dehydration was prevented by the application sexes were used in this study. GFAP-IL6Tg was constructed and of Lacri-luber eye ointment. After anesthesia recovery, the initially studied by Prof. Iain Campbell (Campbell et al., 1993), complete whisker paralysis was assessed to ensure that complete while GFAP-IL10Tg has been constructed and characterized facial nerve resection was achieved. NL and axotomized animals by our laboratory (Almolda et al., 2015). All mice were were distributed in different experimental groups and euthanized housed and bred in the Institute of Neurosciences of the at 3, 7, 14, 21, and 28 dpl. Universitat Autònoma de Barcelona under a 12 h light/dark cycle, with food and water ad libitum. All experimental 50 Bromodeoxyuridine Injections animal work was conducted according to Spanish regulations To determine microglia proliferation in the FNA model, (Ley 32/2007, Real Decreto 1201/2005, Ley 9/2003 and Real proliferative cells were labeled with 50 bromodeoxyuridine Decreto 178/2004) in agreement with European Union directives (BrdU). BrdU is a synthetic thymidine analog that incorporates (86/609/CEE, 91/628/CEE, and 92/65/CEE) and was approved into the DNA of dividing cells during the S-phase and can be by the Ethical Committee of the Autonomous University of transferred to daughter cells upon replication. Lesioned WT Frontiers in Cellular Neuroscience | www.frontiersin.org 3 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury (n = 4) animals were intraperitoneally injected with BrdU 1 h with blocking solution 3 (BS3) containing 0.2% gelatin (100 mg/kg) diluted in 0.1 M PBS (pH 7.4) every 24 h, from the in TBS with 0.1% Triton X-100 (TBS-T0.1%) or with BS2 for day of the lesion to 14 dpl, to be sacrificed afterward. pHH3. Later, sections were incubated with sheep anti-TREM2 for 48 h at RT, or either goat anti-ApoE antibody or rabbit Tissue Processing for Histological Analysis anti-pHH3 ON at 4◦ C plus 1 h at RT. After several washes Animals were deeply anesthetized with a solution of ketamine with TBS-T0.1%, sections were incubated with the corresponding (80 mg/kg) and xylazine (20 mg/kg) injected intraperitoneally secondary antibody being anti-sheep Alexa-Fluor (AF) 488, at a dose of 0.015 ml/g and perfused intracardially with 4% anti-goat AF568, anti-goat AF488 in the case of the double paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brains immunostaining combining ApoE with GFAP, or anti-rabbit were removed and post-fixed in the same fixative for 4 h at 4◦ C AF488 for 1 h at RT (Table 1). Afterward, sections were washed and, after rinsing in phosphate buffer, cryopreserved for 48 h and incubated ON at 4◦ C plus 1 h at RT with rabbit anti- in a 30% sucrose solution and subsequently frozen in ice-cold Iba1, rabbit anti-pHH3, rat anti-CD68, rat anti-CD16/32, rat 2-methyl butane solution (320404, Sigma–Aldrich). Parallel anti-Galectin-3, rat anti-CD11b antibodies, or mouse anti-GFAP free-floating coronal sections (30-µm-thick) of the brainstem (Table 1). After several washes, sections in each combination containing the facial nucleus (FN), in the case of the FNA model, were incubated for 1 h at RT with the corresponding secondary as well as parallel free-floating transversal sections (30-µm-thick) conjugated antibodies anti-rabbit AF568, anti-rat AF555 anti-rat of the brain containing the hippocampus, for the PPT model, AF647, or anti-mouse AF555 secondary antibodies (Table 1). were obtained using a CM3050s Leica cryostat, and sections were For the triple IHC protocol, these last steps were repeated stored at −20◦ C in Olmos antifreeze solution until their use. Each using a mouse anti-GFAP antibody and secondary anti-mouse experimental group was formed by three to five animals. AF488 conjugated antibody (Table 1). On the other hand, in Single Immunohistochemistry the case of TREM2 and P2RY12 double immunohistochemistry, Free-floating cryostat sections were processed for the sheep anti-TREM2 and rat anti-P2RY12 primary antibodies were visualization of TREM2, CD16/32, CD68, and phospho- incubated simultaneously 48 h at RT (Table 1), and after washes, Histone 3 (pHH3). Briefly, after 10 min of endogenous secondary anti-sheep AF488 antibody was firstly incubated for peroxidase blocking with 2% H2 O2 in 70% methanol and 1 h at RT, followed by a second incubation with an anti-rabbit later washes with Tris-buffered saline (TBS, pH 7.4), sections AF568 antibody for 1 h at RT (Table 1). Finally, in double ApoE were blocked for 1 h in either blocking solution 1 (BS1), and Iba1 or GFAP immunostaining, sections were treated for containing 0.2% gelatin (powder food grade, 1.04078, Merck) a diminishing fluorescence background as described in Schnell in TBS with 0.5% Triton X-100 (TBS-T0.5%)—in the case of et al. (1999). Thus, sections were dipped in distilled water, treated TREM2 staining—or in blocking solution 2 (BS2), containing with CuSO4 (10 mM) in ammonium acetate buffer (50 mM 10% fetal bovine serum (FBS) in TBS with 1% Triton X-100 CH3 COONH4 , pH 5.0) for 90 min, and rinsed again with distilled (TBS-T1%)—for CD16/32, CD68 and pHH3–. Then, sections water before mounting. were incubated with sheep anti-TREM2, rat anti-CD16/32, Additionally, microglial proliferation was determined by rat anti-CD68, or rabbit anti-pHH3 antibodies (Table 1). BrdU staining. DNA was denatured by first incubating sections Incubations were performed overnight (ON) at 4◦ C plus 1 h in 0.082 N HCl for 10 min at 4◦ C and then for 30 min in at room temperature (RT) in the case of CD16/32, CD68, 0.82 N HCl at 37◦ C. Sections were subsequently rinsed with and pHH3, or 48 h at RT in the case of TREM2. Sections borate buffer (pH 8.5) and 0.5% Triton X-100 in TBS. Afterward, incubated in media lacking the primary antibody were used sections were incubated for 1 h at RT using BS2, and the as negative control and spleen sections as a positive control. anti-BrdU antibody was added and incubated ON at 4◦ C plus After washes with either TBS-T0.5% or TBS-T1%, sections were 1 h at RT (Table 1). After several washes with TBS-T1%, the incubated at RT for 1 h with the corresponding biotinylated corresponding conjugated anti-rat AF555 secondary antibody secondary antibodies, followed by 1 h at RT with horseradish was added for 1 h at RT (Table 1). Then, the anti-Iba1 antibody peroxidase (HRP)-conjugated streptavidin, diluted in the was incubated ON at 4◦ C plus 1 h at RT, with a final 1 h corresponding BS (Table 1). After washes with TBS, the final incubation at RT of anti-rabbit AF488 conjugated antibody reaction was visualized by incubating sections with a DAB (Table 1). All double-labeled sections were nuclei stained with kit (SK-4100; Vector Laboratories Incorporation, Burlingame, DAPI (Table 1) for 5 min, before being coverslipped with CA, USA) following the manufacturer’s instructions. Finally, Fluorescence Mounting Medium (S-3023; Dako). sections were mounted on gelatin-coated slides, counterstained with toluidine blue if considered appropriate, dehydrated in Brightfield Microscopy Quantification graded alcohols, and, after xylene treatment, coverslipped Quantitative analysis was performed on sections stained with with DPX. single immunohistochemistry for TREM2, CD68, CD16/32, and pHH3. For each immunostaining, three to five WT, GFAP- Double and Triple Immunohistochemistry IL6Tg, and GFAP-IL10Tg animals per time post-lesion and Double and triple immunolabeling combining (1) TREM2, experimental lesion model were analyzed. ApoE or pHH3 with (2) either Iba1, CD68, CD16/32, Galectin- In the PPT model, images of two central hippocampal 3, pHH3, P2RY12, GFAP or CD11b, and (3) GFAP were sections containing the deafferented molecular layer (ML) of performed as follows: sections were washed and blocked for the dentate gyrus (DG) for each lesioned animal, and the Frontiers in Cellular Neuroscience | www.frontiersin.org 4 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury TABLE 1 | List of antibodies and reagents used in immunohistochemistry (IHC). Target antigen Host/conjugation Dilution Catalog number Manufacturer Primary antibodies ApoE Goat 1:2,500 AB947 EMD Millipore BrdU Rat 1:120 Ab6326 Abcam CD11b Rat 1:1,000 MCA74G AbD Serotec CD16/32 Rat 1:1,000 553142 BD Pharmingen CD68 Rat 1:1,000 MCA1957 AbD Serotec Galectin-3 Rat 1:500 125402 BioLegend GFAP Mouse 1:6,000 G3893 Sigma Iba1 Rabbit 1:500 GTX100042 GeneTex P2RY12 Rat 1:50 848001 BioLegend pHH3 Rabbit 1:500 06-570 EMD Millipore TREM2 Sheep 1:400 AF1729 R&D Systems Secondary antibodies Goat IgG Alexa Fluor 568 1:1,000 A11057 Invitrogen Goat IgG Alexa Fluor 488 1:1,000 A11055 Invitrogen Mouse IgG Alexa Fluor 555 1:1,000 A31570 Invitrogen Rabbit IgG Alexa Fluor 488 1:1,000 A21206 Invitrogen Rabbit IgG Alexa Fluor 568 1:1,000 A10042 Invitrogen Rat IgG Alexa Fluor 555 1:1,000 A21434 Invitrogen Rat IgG Alexa Fluor 647 1:1,000 A21247 Invitrogen Sheep IgG Alexa Fluor 488 1:1,000 A11015 Invitrogen Goat IgG Biotinylated 1:500 BA-9500 Vector Laboratories Rabbit IgG Biotinylated 1:500 BA-1000 Vector Laboratories Rat IgG Biotinylated 1:500 BA-4001 Vector Laboratories Sheep IgG Biotinylated 1:500 BA-6000 Vector Laboratories HRP-conjugated Streptavidin 1:500 SA-5004 Vector Laboratories DAPI 1:10,000 D9542 Sigma–Aldrich equivalent area for NL mice, were captured. A total of three of pHH3+ cells in the FN were manually counted using a 20× photographs were obtained for each hippocampus, at 40× objective. Then, all FN counted were photographed at 10×, magnification in the case of TREM2 and 20× magnification and the area of each FN was quantified using ImageJ software for CD16/32 and CD68 analysis. For the FNA model, at least (Wayne Rasband, National Institute of Health, Bethesda, MD, three representative sections from the brainstem of each animal, USA). pHH3 cell density was calculated from the total pHH3+ containing the central part of the contralateral and the ipsilateral cells/FN area for each animal, and results were expressed as FN, were photographed at 10× magnification. All images were pHH3+ cells/mm2 . obtained using a DXM 1200F Nikon digital camera joined to a brightfield Nikon Eclipse 80i microscope and the corresponding Confocal Microscopy Quantification ACT-1 2.20 software (Nikon Corporation). In both models, To evaluate the percentage of ramified microglial TREM2+ cells quantification of each photograph was done by using the for the PPT model, sections immunostained with Iba1 and AnalySISr software, in which the threshold corresponding to TREM2 were used. A minimum of three WT, GFAP-IL6Tg, the staining was manually set. In the case of TREM2, the and GFAP-IL10Tg animals euthanized at 3 dpl were quantified. threshold was set to quantify only TREM2 cellular staining, In this case, the total ML of DG was photographed at 40× excluding the background staining corresponding to soluble magnification with a Zeiss LSM 700 confocal microscope. From TREM2 (sTREM2). Analysis of each photograph resulted in the the total amount of microglial Iba1+ cells contained in the percentage of area covered by the immunolabeling (% Area) and deafferented ML of the DG, ramified microglial TREM2+ cells the intensity of the immunoreaction (Mean Gray Value Mean). were counted using the ‘‘Cell counter’’ plug-in from ImageJ Then, the AI index was calculated by multiplying the percentage software. Results were expressed as a percentage of Iba1+ cells of the immunolabeled area and the Mean Gray Value Mean containing TREM2 in their ramifications from the total number (Acarin et al., 1997). Results of the AI index and intensity were of Iba1+ cells (% TREM2 + Iba1+ cells/total Iba1+ cells). expressed in arbitrary units. In the FNA model, quantification of microglial clusters In the case of PPT, quantification of microglial cell expressing TREM2 and microglial TREM2 ramified clusters was proliferation was carried out on hippocampal sections performed by analyzing the double immunofluorescence for immunostained for the mitotic marker pHH3 in NL and lesioned Iba1 and TREM2. A minimum of three WT, GFAP-IL6Tg, and animals at 2, 3, and 7 dpl. The number of pHH3+ microglial GFAP-IL10Tg animals at 14 and 21 dpl were used. The number cells in the whole ML of the DG were manually counted using a of total microglial clusters in the FN, identified by Iba1+ staining, 20× objective. Data were averaged and represented as pHH3+ expressing TREM2 in microglial ramifications were manually cells/mm2 . counted using a 40× and 63× objective with a Zeiss LSM For the FNA model, pHH3 quantification was done in NL 700 confocal microscope. Results were expressed in percentage and lesioned animals at 3, 7, 14, 21, and 28 dpl. The number as TREM2 + Iba1+ microglial clusters from the total amount Frontiers in Cellular Neuroscience | www.frontiersin.org 5 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury of Iba1+ clusters (% TREM2 + Iba1+ microglial clusters/total and 20 mg/kg xylazine solution, injected intraperitoneally). Iba1+ clusters). Brains were removed quickly and cut into little pieces in a cold 0.32 M sucrose solution prepared in 0.1 M Tris.Cl buffer. Tissue Processing for Protein After homogenization with a mechanical homogenizer, myelin Quantification was separated by using a sucrose gradient in an ultracentrifuge For quantifying sTREM2, NL, 3 dpl after PPT (n = 6–8) and at 105,000 g for 45 min at 4◦ C (70Ti Rotor, Sorvall). Myelin 21 dpl axotomized WT animals (n = 10) were used. Animals was isolated from the interphase, and after a hypoosmotic were deeply anesthetized with a solution of ketamine (80 mg/kg) shock using Tris.Cl buffer and 0.32 M sucrose, pelleted myelin and xylazine (20 mg/kg) injected intraperitoneally at a dose of was weighted and reconstituted in PBS at a 100 mg/ml 0.015 ml/g and perfused intracardially for 1 min with cold 0.1 M (w/v) suspension. DPBS (pH 7.4, 14190-094; Thermo Fisher Scientific, Waltham, For myelin fluorescent labeling, conjugation with MA, USA). Then, the brain was removed from the skull, and for pHrodo-STP Ester Greenr (P35369, Thermo Fisher Scientific, NL and PPT-lesioned animals, the hippocampus, was dissected Waltham, MA, USA)—with excitation at 505 nm and emission out. For FNA-lesioned animals, the brain was excised and two at 525 nm—was used following the procedure described by 0.5-mm-thick coronal slices were obtained from the brain trunk Greenhalgh et al. (2018). Briefly, the pHrodo dye was dissolved using a Mouse Brain Matrix (Zivic Instruments). For each slice, in 75 µl dimethyl sulfoxide (DMSO, Sigma–Aldrich), and the dorsal inferior half was cut with sterile knives, and tissue afterward, 18.5 µl of myelin in a concentration of 15 mg/ml was containing the contralateral and ipsilateral FN was divided. mixed with 25 µl of pHrodo and resuspended in 206.5 µl of Samples were immediately snap-frozen individually in liquid PBS (pH 8.0). After 45 min of incubation at RT, with agitation nitrogen, and stored at −80◦ C. The tissue of two hippocampi and protected from light, labeled myelin was spun down, or three FN was pooled for protein extraction. Briefly, protein resuspended in PBS (pH 7.4), and stored at −80◦ C. extraction was performed by using a mechanical homogenizer to disrupt tissue in lysis buffer, containing 25 mM HEPES, 2% Myelin Phagocytosis Assay and Flow Igepal, 5 mM MgCl2 , 1.3 mM EDTA, 1 mM EGTA, 0.1 M PMSF, Cytometry and protease (1:100, P8340; Sigma–Aldrich) and phosphatase The capacity of TREM2+ microglia in PPT-lesioned animals inhibitor cocktails (1:100, P0044; Sigma–Aldrich), and allowing (at 3 dpl) and axotomized animals (21 dpl) to phagocyte solubilization for 2 h at 4◦ C. Afterward, samples were centrifuged myelin debris was analyzed by isolating cells with Percoll and at 6,500 g for 10 min at 4◦ C in a microcentrifuge (5415R performing a myelin phagocytosis assay followed by a flow centrifuge, Eppendorf) and the supernatants were collected cytometry procedure (Almolda et al., 2009; Greenhalgh et al., and frozen. 2018). Briefly, anesthetized animals (dose of 0.015 ml/g of an For separating the soluble protein fraction, each protein 80 mg/kg ketamine and 20 mg/kg xylazine solution, injected lysate was ultracentrifuged at 107,000 g for 30 min at intraperitoneally) were intracardially perfused for 1 min with 4◦ C in a Sorvall MTX 150 Series Micro-Ultracentrifuge. 0.1 M DPBS. The brain was removed and the area of interest, After ultracentrifugation, supernatants were collected and either hippocampus or FN, was dissected out as described concentrated by using microcentrifuge cellulose filter units above. For each tube, an individual lesioned hippocampus, (MRCPRT010, Merck Millipore). Resulting protein lysates were two NL hippocampus, or a pool of two either lesioned or collected and total protein concentration was determined contralateral NL FN were used. To obtain a cell suspension, with a commercial Pierce BCA Protein Assay kit (#23225; samples were dissociated through 160 and 70 µm meshes and Thermo Fisher Scientific, Waltham, MA, USA) according to digested for 30 min at 37◦ C using type IV collagenase (17104-019; manufacturer’s protocol. Life Technologies, Carlsbad, CA, USA) and DNAsa I (D5025; Sigma–Aldrich, St. Louis, MO, USA). Subsequently, each cellular Enzyme-Linked Adsorbent Immunoassay suspension was centrifuged at RT for 20 min at 2,400 rpm in a for sTREM2 Detection discontinuous density Percoll gradient (17-0891-02; Amersham- Quantification of sTREM2 was performed by following the Pharmacia, UK) between 1.03 and 1.08 g/ml using a Heareus manufacturer’s instructions of a Mouse TREM2 ELISA kit Multifuge 3L-R centrifuge (ThermoFisher). Myelin in the upper (ABIN429539, Antibodies-online). Briefly, standards and the layer was removed, and cells in the interphase and the clear corresponding samples were incubated at 2.5 µg/µl in a upper-phase were collected and washed in PBS + 2% FBS. pre-coated TREM2 ELISA plate. After reagent incubations, and Afterward, the myelin phagocytosis assay was performed as the addition of TMB substrate and Stop solution, results were follows: cells were resuspended in DMEM-F12 medium (ref. read at 450 nm in the microplate reader VarioskanTM Lux 31330-038; Thermo Fisher Scientific, Waltham, MA, USA) + (Thermo Fisher Scientific, Waltham, MA, USA). 10% FBS, then 3 µg of pHrodo Green-labeled myelin were added, and samples were incubated at 37◦ C for 4 h. For Isolation of Myelin and Fluorescent negative controls, the same procedure was performed without Labeling With pHrodoTM Green STP Ester adding myelin, or by resuspending cells in PBS + 2% FBS and Myelin was isolated following the protocol described by Rolfe preserving at 4◦ C. After myelin incubation, cells were washed et al. (2017). Briefly, six WT animals were dislocated under the one time with PBS + 2% FBS at 1,200 rpm for 10 min. Then, effects of anesthesia (dose of 0.015 ml/g of 80 mg/kg ketamine cells were blocked for Fc receptors by incubating them in a Frontiers in Cellular Neuroscience | www.frontiersin.org 6 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury TABLE 2 | List of antibodies used in flow cytometry. Target antigen Format Dilution Catalog number Manufacturer Fc blocker CD16/32 Purified 1:250 553142 BD Biosciences Primary antibodies CD11b APC-Cy7 1:400 557657 BD Biosciences CD45 PerCP 1:400 557235 BD Biosciences TREM2 PE 1:400 FAB17291P R&D Company solution of purified CD16/32 diluted in PBS + 2% FBS, for the nucleus of microglial cells (arrowheads in Figure 1H); 10 min at 4◦ C (Table 2). Afterward, cells were labeled for at 2 and 3 dpl, a subpopulation of microglial cells showed 30 min at 4◦ C with the combination of the following surface TREM2 expression in their ramifications (Figure 1I). antibodies: anti-CD11b-APCCy7, anti-CD45-PerCPCy5, and Quantification of TREM2+/Iba1+ microglia cells at 3 dpl anti-TREM2-PE (Table 2). In parallel, isotype-matched control revealed that approximately half of microglia co-expressed antibodies for the different fluorochromes (BD Pharmingen, TREM2 and Iba1 in their ramifications (48.1 ± 1.96%, in Switzerland) and samples that were not incubated with myelin Figure 1J). From 3 dpl onwards, microglial cells expressing were used as the negative control, while a cell suspension TREM2 in their ramifications decreased progressively. of splenocytes and cerebellum of GFAP-IL6Tg animals as the Interestingly, at 7 and 14 dpl, only a few scattered microglial positive control. Finally, cells were acquired using a FACS Canto processes were labeled, being TREM2 expression mostly flow cytometer (Becton Dickinson, San Jose, CA, USA) in the restricted again into the region next to the nucleus. At 21 dpl, corresponding fluorescent channels, and pHrodo-Green labeled TREM2 was completely restricted in a position adjacent to myelin phagocytosed by microglia could be captured in the FITC the nucleus, like in basal conditions (Figure 1F). Notably, channel. Results were analyzed using the FlowJoTM software from 2 to 14 dpl, also a diffuse staining of sTREM2 covering (FlowJo LLC). the neuropil of both the MML and the OML was observed (Figures 1B–E). Quantification of sTREM2 at 3 dpl showed an Statistical Analysis increase of this protein fraction in the lesioned compared to the Statistics were performed using Graph Pad Prismr software NL hippocampus (Figure 1K). (Graph Pad Software Inc.) and results were expressed as Mean ± Standard error of the mean (SEM). Appropriate statistical tests were used for each condition: for time-course TREM2 Is Upregulated In the Lesioned FN dynamics, one-way ANOVA with multiple comparison post hoc After FNA Tukey’s test was used, for microglial cell density Dunnet’s post After FNA, the immunohistochemical study of hoc test was used instead, for sTREM2 comparison and flow TREM2 expression showed an upregulation of TREM2 in cytometry analysis either unpaired or paired Student’s t-test was the ipsilateral FN along the different time points, in comparison used, and finally, two-way ANOVA with multiple comparison to the NL contralateral side, where the levels of TREM2 were very post hoc Sidak’s test was used to compare Tg mouse lines with low (Figures 2A–F). TREM2 immunolabeling increased from their corresponding WT. 3 dpl onwards, showed a maximum at 21 dpl, and decreased thereafter at 28 dpl, although levels were still higher than in the RESULTS contralateral NL FN (Figure 2J). Double-immunolabeled sections with TREM2 and TREM2 Is Upregulated in the Denervated Iba1 demonstrated that, in the NL contralateral side, ML of the DG After PPT TREM2 expression was concentrated in the cytoplasm of The immunohistochemical study of TREM2 expression after microglia, in a position adjacent to the nucleus (arrowheads PPT showed that, in comparison to the low to undetectable in Figure 2G). In the ipsilateral side, at 3 and 7 dpl, TREM2 expression in the ML of NL animals, a significant TREM2 expression remained located like in basal conditions upregulation in TREM2 immunolabeling in both the medial although levels of expression were higher (arrowheads in molecular layer (MML) and the outer molecular layer (OML) Figure 2H). In contrast, at 14 and 21 dpl, TREM2 staining was of the DG was observed (Figures 1A–F). A notable increase additionally found in the ramifications of a subpopulation of in TREM2 immunolabeling was found between 2 and 3 dpl, microglial cells, forming part of the so-called microglial clusters which sharply declined afterward, at 7 and 14 dpl. At or microglial nodules (arrows in Figure 2I), which consist 21 dpl, levels of TREM2 were comparable to basal conditions of accumulations of three to four microglial cells commonly (Figure 1G). described in the time-course of FNA injury (Moran and Graeber, Co-localization studies indicated that all TREM2+ cells 2004). At these time-points, almost all microglial clusters in the corresponded to microglia/macrophages, as they co-labeled FN showed TREM2+ ramified microglia (Figure 2K and arrows with Iba1 at all time-points analyzed (Figures 1H,I). Moreover, in Figure 2I). In a similar way to the observations of PPT injury, variations in the cellular location of TREM2 after injury were low diffuse staining of sTREM2 covering the neuropil of the FN observed. Whereas in basal conditions TREM2 was found was observed, especially at 14 and 21 dpl. A comparison of the concentrated in the cytoplasm, in a position adjacent to amount of sTREM2 fraction between NL and 21 dpl animals Frontiers in Cellular Neuroscience | www.frontiersin.org 7 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury Phenotypic and Functional Characterization of TREM2+ Microglial Cells After PPT and FNA TREM2 has been principally associated with microglial proliferation, phagocytosis, and DAM-triggering (Takahashi et al., 2005, 2007; Wang et al., 2015; Keren-Shaul et al., 2017; Krasemann et al., 2017). To determine if TREM2 microglial expression after PPT and FNA was related to these functions, we studied the dynamics of microglial proliferation (pHH3) and phagocytosis (CD16/32, CD68) in both models, and we compared it to the time course of TREM2 expression. We also performed co-localization studies with TREM2+ microglia using specific markers related to proliferation (pHH3, BrdU), phagocytosis (CD16/32, CD68, ApoE, Galectin-3), and DAM-triggering (Galectin-3, ApoE, and P2RY12), as well as a functional myelin phagocytosis assay. In this part of the study, we described as TREM2+ microglia the population appearing after both lesions that show an increase of TREM2 in microglial ramifications, according to our results. Expression of Proliferation Markers In the PPT paradigm, a subpopulation of TREM2+ cells in the MML and OML expressed pHH3, a mitotic marker, at both 2 and 3 dpl (Figures 3A–F), coinciding with the peak of microglial proliferation observed in this paradigm (Figure 3G). Despite all proliferative cells were TREM2+, the total amount of proliferating cells within the TREM2+ subpopulation was low. After FNA, pHH3 cell counting showed a peak of proliferation at 3 dpl, when TREM2 was located intracytoplasmatically. The study of TREM2 with pHH3 showed that, at 14 and 21 dpl, a low number of proliferating pHH3+ microglial cells was detected (Figures 4A,B). Also, accumulative BrdU for 14 dpl indicated the presence of BrdU+ microglial cells in both the parenchyma FIGURE 1 | Temporal pattern of Triggering Receptor Expressed on Myeloid and within clusters where, as previously described, TREM2+ cells-2 (TREM2) changes in microglial cells after perforant pathway transection (PPT). (A–F) Representative images showing TREM2 staining in cells were located. Proliferating cells showed variable BrdU the granular (GL) as well as the inner, medial and outer molecular layers (IML, intensity in the nucleus, indicating an uneven number of mitosis MML, and OML, respectively) of the dentate gyrus (DG) in non-lesioned (NL) (Figures 4C–E). (A) and PPT-lesioned mice from 2 to 21 dpl (B–F). Note that, while in NL and at 21 dpl TREM2 was only depicted as small rounded morphologies Expression of the Phagocytic Markers CD16/32 and (arrowheads), from 2 to 14 dpl ramified TREM2+ cells were also observed CD68 (arrows). (G) Graph showing the quantification of the AI index for TREM2 along the different time-points. Note the higher level of TREM2 at The study of Fc-receptors CD16/32 and lysosomal 2 and 3 dpl compared to NL and the posterior decrease from 3 to 7 dpl CD68 expression in microglia during PPT and FNA time-courses (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s test *p < 0.05, provided a general assessment of changes in these phagocytic ***p < 0.001 respect to NL, a p < 0.01 respect to the previous time-point). molecules. In the NL hippocampus, microglia express low (H,I) Double immunohistochemistry showing microglial cells labeled with Iba1 levels of CD16/32, being not fully coincident with microglia (red) and TREM2 (green). Note that in NL animals, TREM2 is confined to areas around the nucleus of some Iba1+ cells (arrowheads), whereas at 3 dpl, expressing intracellular TREM2 (Figures 5A–C). After the additionally, some prolongations of Iba1+ microglial cells (arrows) were also PPT lesion, the majority of TREM2+ cells observed at observed. (J) Graph showing the quantification of the percentage of TREM2+ 3 dpl, in the peak of microglia activation, presented variable cells, -those showing TREM2 staining in ramifications-, from total Iba1+ cells CD16/32 staining (arrowheads and arrows in Figures 5D–F). at 3 dpl. (K) Graph showing the quantification of sTREM2 (pg/ml) in the hippocampus of NL and PPT-lesioned mice at 3 dpl. A significant increase in At later time-points, 14 dpl, when TREM2 was restricted the amount of sTREM2 is detected in the lesioned hippocampus (unpaired to the cytoplasm in a position adjacent to the nucleus, Student’s t-test, *p < 0.05). Scale bar (A–F) = 50 µm; (H,I) = 50 µm. some microglia cells still maintained CD16/32 expression (Figures 5G–I). Furthermore, in NL hippocampus, CD68 and TREM2 were both located intracellularly and mostly coincident did not show a statistically significant increase in this protein in microglia cells (Figures 5K–M). At 3 dpl, all TREM2+ fraction after lesion (Figure 2L). microglia contained CD68, although CD68 expression was Frontiers in Cellular Neuroscience | www.frontiersin.org 8 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 2 | Temporal pattern of TREM2 changes in microglial cells after facial nerve axotomy (FNA). (A–F) Representative images showing TREM2 staining in the contralateral NL, as well as the ipsilateral sides of the FN from 3 to 28 dpl. From NL to 7 dpl TREM2 is mainly restricted to a perinuclear location (arrowheads), whereas from 14 dpl onwards TREM2 is extended to microglia ramifications and clusters (arrows). (G–I) Double immunohistochemistry showing microglial cells positive for TREM2 (green) and Iba1 (red). Note that in both the NL side and at 7 dpl, TREM2 is confined to areas around the nucleus of some Iba1+ cells (arrowheads), whereas at 14 dpl expression was detected in microglial prolongations in some microglial clusters (arrows). It is important to note that occasional Iba1 + TREM2-clusters were found at this time-point (arrowheads in I). (J) Graph showing the quantification of the AI index for TREM2 along the different time-points. (Continued) Frontiers in Cellular Neuroscience | www.frontiersin.org 9 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 2 | Continued Note the increase of TREM2 until 21 dpl and the later decrease (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s test *p < 0.05, ****p < 0.001 compared to NL; a p < 0.01 compared to the previous time-point). (K) Graph showing the percentage of microglial clusters containing TREM2+ microglial cells with expression in their ramifications at 14 and 21 dpl. (L) Graph showing the quantification of sTREM2 (pg/ml) in the FN of contralateral NL and the corresponding ipsilateral FN of mice lesioned at 21 dpl. No significant differences in the amount of sTREM2 are detected (paired Student’s t-test, p = 0.61). Scale bar (A–F) = 30 µm; (G–I) = 50 µm. FIGURE 4 | Determination of the proliferative activity in microglial clusters after FNA. (A) Graph showing the quantification of PHH3+ cell density in the FN after FNA. Note the massive increase of proliferative cells at 3 dpl, followed by a later decrease at 7 and 14 dpl, and the disappearance of proliferative cells from 21 dpl onwards (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s ****p < 0.001 compared to NL, a p < 0.001 compared to the previous time-point). (B) Representative triple immunofluorescence image combining PHH3 (green) with CD11b (blue) and GFAP (red). Note that all PHH3+ cells counted in the FN corresponded to microglia cells (CD11b+), but not astrocytes (GFAP). (C–E) Representative double-immunolabeled images combining TREM2 (green) and BrdU (red) in the FN of 14 dpl animals administered daily with BrdU. Proliferative BrdU+ microglial cells were observed inside microglial clusters (arrowheads) and outside them (arrows). Scale bar (B) = 5 µm; (C–E) = 10 µm. In line with the above-mentioned observations, the dynamics of CD16/32 expression in the PPT lesion was sharply upregulated until 7 dpl and decreased thereafter at 14 dpl, although at this time point, levels of expression were still around 100-times higher than in NL animals (Figure 5J). In contrast, CD68 expression also was upregulated at 3 dpl but then FIGURE 3 | Determination of the proliferative activity of TREM2+ microglial progressively decreased from 7 to 14 dpl. At this later time-point, cells after PPT. (A–F) Representative double-immunolabeled images combining TREM2 (green) and PHH3 (red) in the DG of PPT-lesioned animals CD68 levels were still around two-times higher than NL mice at 2 dpl (A–C) and 3 dpl (D–F). Note that at both time points PHH3+ cells (Figure 5T). showed membranal TREM2 expression in the ramifications and soma of In the contralateral NL FN, CD16/32 was not microglial cells (arrowheads). However, not all TREM2+ microglial cells with expressed in microglial cells (Figures 6A–C), while expression in their ramifications were undergoing mitosis (arrows). (G) Graph CD68 was located intracellularly and colocalized with of PHH3+ cell density (cell/mm2 ) in NL and PPT lesioned animals after 2, 3, and 7 dpl (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s intracellular TREM2 expression (Figures 6H–J). After compared to NL **p < 0.01). Scale bar (A–F) = 10 µm. FNA, the majority of clusters containing TREM2+ microglia also co-expressed both CD16/32 and CD68 (arrows in Figures 6D–F,K–M). However, the few also found in other microglia, that is, CD68 is not exclusive microglial clusters that were TREM2- also contained to TREM2+ microglial subpopulation (arrowheads in microglial staining for CD16/32 and CD68 (arrowheads in Figures 5N–P). At 14 dpl, CD68 and TREM2 were again Figures 6D–F,K–M). expressed intracellularly and mostly coincident in microglia When we studied the microglial expression pattern of (Figures 5Q–S). CD16/32 and CD68 along the different time-points after FNA Frontiers in Cellular Neuroscience | www.frontiersin.org 10 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 5 | Analysis of CD16/32 and CD68 expression in TREM2+ microglial cells after PPT. (A–I) Representative double-immunolabeled images combining TREM2 (green) and CD16/32 (red) in the ML of the DG in NL (A–C) at 3 dpl (D–F) and 14 dpl (G–I). Note that both TREM2 + CD16/32+ (arrows), some of them very intensely stained (arrowheads) and TREM2+ CD16/32- (asterisk) microglial cells were observed. (J) Graph showing the AI index of CD16/32 expression in NL and 00 PPT-lesioned mice at 3, 7, and 14 dpl (One-way ANOVA, time effect p < 0.0001; post hoc Tukey’s test compared to NL ***p < 0.001, ****p < 0.0001; a p < 0.01, a p < 0.001 compared to previous time point). (K–S) Representative double-immunolabeled images combining TREM2 (green) and CD68 (red) in the DG of NL (K–M) and PPT-lesioned animals at 3 (N–P) and 14 dpl (Q–S). Note that similarly to CD16/32, most TREM2+ ramified cells expressed also CD68 (arrows in N–P) at 3 dpl, whereas at 14 dpl TREM2 expression was restricted to the perinuclear area of CD68+ cells (arrowheads in Q–S). Also, some TREM2-CD68+ cells were found at 3 dpl (arrowheads in N–P). (T) Graph showing the AI index of CD68 expression in NL and PPT-lesioned mice at 3, 7, and 14 dpl (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s test compared to NL ***p < 0.001, ****p < 0.0001; a0 p < 0.01 compared to previous time point). Scale bar (A–I,K–S) = 50 µm. (Figures 6G,N), we found that both CD16/32 and CD68 levels ligands of TREM2 and DAM-associated molecules. The study increased rapidly at 3 and 7 dpl, respectively, remained elevated was centered on the peak of TREM2 expression after PPT, until 21 dpl, and decreased at 28 dpl (Figures 6G,N). around 3 dpl, and after FNA, at 14 and 21 dpl, compared to basal conditions. Expression of the Phagocytic Ligands ApoE and In the NL hippocampus, no expression of ApoE was Galectin-3 observed in microglia, like similarly observed in the PPT According to results obtained in previous sections, we studied paradigm at any time-point (Figures 7A–I), as its expression ApoE and Galectin-3 expression, both specific phagocytic was in all cases restricted to astrocytes (Supplementary Frontiers in Cellular Neuroscience | www.frontiersin.org 11 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 6 | Analysis of CD16/32 and CD68 expression in TREM2+ microglial cells after FNA. (A–F) Representative double-immunolabeled images combining TREM2 (green) and CD16/32 (red) in the NL (A–C) and ipsilateral FN at 21 dpl (D–F). Note that microglial clusters with low (arrowheads) or high (arrows) TREM2+ expression in their ramifications being CD16/32+ could be observed. (G) Graph showing the AI index of CD16/32 expression in the contralateral, NL, and the ipsilateral sides of the FN from 3 to 28 dpl (one-way ANOVA, time effect p < 0.0001; post hoc Tukey’s test compared to NL **p < 0.01, ****p < 0.0001, compared to the previous time-point a p < 0.05, a0 p < 0.0001). (H–M) Representative double-immunolabeled images combining TREM2 (green) and CD68 (red) in the NL (H–J) and ipsilateral FN of lesioned animals at 21 dpl (K–M). Note that similarly to CD16/32, most CD68+ microglial clusters express high levels of TREM2 (arrows in K–M), but also some clusters with low levels of TREM2 were found to be CD68+ (arrowheads in K–M). (N) Graph showing the AI index of CD68 expression in NL and FNA-lesioned mice from 3 to 28 dpl (one-way ANOVA, time effect p < 0.01; post hoc Tukey’s test compared to NL *p < 0.05, ***p < 0.001, ****p < 0.0001, respect to the previous time-point a p < 0.05). Scale bar (A–F,H–M) = 50 µm. Figure 1). Moreover, de novo expression of Galectin-3 in were observed. As far as Galectin-3 is concerned, de novo some TREM2+ microglial cells was reported at 2 and 3 dpl expression of this marker was found in some TREM2+ (Figures 8A–F). microglial clusters, although no clear relationship between After FNA, the expression of ApoE was restricted to Galectin-3 and TREM2 intensities was observed, since astrocytes in the contralateral NL side (Figures 9A–C; clusters with high TREM2 levels presented variable Galectin-3 Supplementary Figure 2), but on the ipsilateral side, ApoE intensity staining (arrows and arrowheads, Figures 10A–F). staining was observed in cells of TREM2+ microglial clusters Additionally, Galectin-3 staining on the surface of isolated at both 14 and 21 dpl (Figures 9D–I). Moreover, ApoE+ facial FMN was found at 14 and 21 dpl (dashed line circles in motor neurons (FMN, dashed line circles in Figures 9D–I) Figures 10A–F). Frontiers in Cellular Neuroscience | www.frontiersin.org 12 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 8 | Analysis of the expression of TREM2 ligand Galectin-3 after PPT. (A–F) Representative double immunofluorescence images combining Galectin-3 (red) and TREM2 (green) at 2 dpI (A–C) and 3 dpI (D–F) after PPT. Occasional microglial cells showing TREM2 expression in their ramifications contained Galectin-3 (white arrowheads) as perinuclear staining in both time-points. Note that at 3 dpI some microglial cells also presented Galectin-3 in their ramifications, indicating an increase of Galectin-3 compared to 2 dpI. Scale bar (A–F) = 10 µm. Microglial Phagocytosis of Myelin by TREM2+ Microglia After PPT and FNA FIGURE 7 | Analysis of the expression of TREM2 ligand ApoE after PPT. To study the functional phagocytosis capacity of both TREM2+ (A–I) Representative images of double immunofluorescence combining ApoE (red) and Iba1 (green) in the NL DG (A–C) and at 2 dpI (D–F) and 3 dpl (G–I) and TREM2- microglia after PPT and FNA, we performed after PPT. Note that ApoE expression could be observed on astrocyte-like a myelin phagocytosis assay and flow cytometry analysis cells (empty arrowheads), but not in microglia (white arrows) in all the of CD11b+/CD45low and CD11b+/CD45high populations time-points. Scale bar (A–I) = 10 µm. (Figures 13A–D). Our results show that in the CD11b+CD45low population of NL hippocampus and FN, formed by microglia, TREM2- cells are phagocytosing myelin debris in low percentages, in all cases below 20% (Figures 13E,M). Expression of the Homeostatic Microglial Marker Interestingly, similar percentages are maintained after lesion P2RY12 (Figures 13G,O), despite the number of cells is increased in these Because ApoE and Galectin-3 molecules were upregulated, as conditions. Therefore, our results indicate that phagocytosis it is seen in DAM, we were interested to observe whether the is not exclusive of TREM2+ cells within CD11b+/CD45low homeostatic receptor P2RY12 was downregulated, to further cells but is a minor function within this population. On the confirm similarities with the DAM phenotype. After PPT, in contrary, when we studied CD11b+/CD45low /TREM2+ cells in contrast to the high P2RY12 expression observed in all microglia NL hippocampus and FN, we could observe that phagocytosis of the NL DG; a gradual downregulation of P2RY12 levels was by microglial cells is increased, and differences between Myelin- observed in TREM2+ microglial cells of the OML and MML and Myelin+ cells observed within the TREM2- population are until 3 dpl (arrowheads, Figure 11). However, in the FNA model, hardly maintained (Figures 13F,N). Furthermore, in the lesioned different levels of P2RY12 were reported in TREM2+ microglial hippocampus, 50% of cells are phagocyting myelin (Figure 13H), clusters at both 14 and 21 dpl. Indeed, TREM2+ clusters showing whereas after FNA there is a tendency to show a higher either low or high P2RY12 were observed. Few TREM2- clusters percentage of phagocytic TREM2+ cells (Figure 13P). In the located within the same region showed high P2RY12 levels case of CD11b+/CD45high populations, formed by macrophages (Figures 12A–O). or highly activated microglia, TREM2- cells show a similar Frontiers in Cellular Neuroscience | www.frontiersin.org 13 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 9 | Analysis of ApoE expression in microglial clusters after FNA. (A–I) Representative double-immunolabeled images combining Iba1 (green) and ApoE (red) in the NL FN (A–C) and at 14 dpI (D–F) and 21 dpl (G–I). In the lesioned FN, astrocyte-like cells were easily identified by strong ApoE staining (empty arrowheads). Note that at 14 and 21 dpl, ApoE was also present in microglial clusters (white arrowheads), and in some of them with high levels of ApoE staining (white arrows), although not all clusters were positive for ApoE (data not shown). Additionally, at 21 dpl, some FMN showed ApoE staining on their surface (dashed line circles). Scale bar (A–I) = 10 µm. phagocytic capacity than the one observed in CD11b+/CD45low Transgenic Overproduction of Either cells: high percentages of non-phagocytic cells are found in NL and lesioned hippocampus and FN (Figures 13I,K,Q,S). IL-6 or IL-10 Modifies TREM2 Expression, Observation of phagocytosis in CD11b+/CD45high /TREM2+ Especially After Anterograde Degeneration population showed a completely opposed behavior, as NL and (PPT Model) lesioned hippocampus and FN contained elevated percentages, Previous studies in our laboratory (Almolda et al., 2014; over 70%, of phagocytic cells (Figures 13J,L,R,T). Altogether, Villacampa et al., 2015; Recasens et al., 2019) using two our results indicate that TREM2 is a receptor preferentially Tg animal lines with an astrocyte-targeted expression of found on phagocytotic cells but it is not exclusive of them, as the pro-inflammatory cytokine IL-6 (GFAP-IL6Tg) and also TREM2- cells can phagocytose myelin. the anti-inflammatory cytokine IL-10 (GFAP-IL10Tg), Frontiers in Cellular Neuroscience | www.frontiersin.org 14 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 10 | Analysis of Galectin-3 expression in microglial clusters after FNA. (A–F) Representative double-immunolabeled images combining TREM2 (green) and Galectin-3 (red) in the ipsilateral FN at 14 dpI (A–C) and 21 dpI (D–F). Note that at 14 and 21 dpI, clusters displaying high intensity for Galectin-3 (white arrows) could show either high or low intensity for TREM2. Likewise, clusters presenting low Galectin-3 (white arrowheads) presented variable TREM2 intensity as well. No differences in microglial cluster Galectin-3 staining were seen between 14 and 21 dpl. FMN were also positive for Galectin-3 at 14 and 21 dpI, being more easily detectable at 14 dpI (dashed line circles). Scale bar (A–F) = 10 µm. demonstrated significant differences in the microglial activation (61.97 ± 2.58%) concerning WT animals (48.1 ± 1.96%; pattern after PPT and FNA. These differences correlated with Figures 15A–D). important modifications in lesion outcomes, either neuronal After FNA, a significantly lower percentage of TREM2 stained survival or axonal sprouting. Considering that the effect of area was observed at 21 dpl in GFAP-IL6Tg mice compared to cytokines into TREM2 expression is not known and that WT mice (Figure 14E; Supplementary Figure 3K). Similar to neuronal survival and axonal sprouting are conditioned to the WT mice, in GFAP-IL6Tg and GFAP-IL10Tg mice, all or almost microglial phagocytic activity which, according to our results, all microglial clusters showed TREM2 staining in microglial can be modified by TREM2 expression, we found of great ramifications at 14 dpl and at 21 dpl (Figure 15E). interest to assess the regulation of cellular TREM2 expression in these Tg animals after both experimental models. DISCUSSION In NL conditions, we detected a qualitative increase of TREM2 intensity levels in the OML and MML of In this work, we studied TREM2 regulation in microglia the hippocampus of both GFAP-IL6Tg and GFAP-IL10Tg after axonal injury, in both anterograde and retrograde mice; but no modifications were observed in the NL FN neurodegeneration models: the PPT and the FNA (Figures 14B,D,F,H; Supplementary Figures 3A,E,I,L). respectively. Results obtained demonstrate an upregulation After PPT, a higher percentage of area occupied by TREM2 in of TREM2 expression in both models of axonal injury but the deafferented area of the ML was detected at 3 dpl in following different patterns. In the PPT, TREM2+ cells appeared both Tg mice (Figures 14A,C; Supplementary Figures 3B,F). at early time-points (2 and 3 dpl), whereas after FNA TREM2+ These differences were more pronounced in the GFAP-IL6Tg microglia were observed mainly forming clusters at 14 and mice than in GFAP-IL10Tg mice compared to their respective 21 dpl. Also, we identified singular subpopulations of TREM2+ WT. In line with these observations, at 3 dpl, a higher microglia expressing CD16/32, CD68 as well as showing de percentage of ramified TREM2+/Iba1+ microglial cells was novo expression of ApoE or Galectin-3 and diverse levels of found in GFAP-IL10Tg (74.7 ± 1.91%) and in GFAP-IL6Tg the homeostatic P2RY12 molecule. Finally, using Tg animals, Frontiers in Cellular Neuroscience | www.frontiersin.org 15 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury TREM2 expression in microglia in the FN and hippocampus of NL animals. In these regions most TREM2 labeling was restricted to a concentrated area in the cytoplasm, in a position adjacent to the nucleus of the microglial cells, coinciding with a previous study where the intracellular distribution was shown in two pools: a deposit in the Golgi complex and a population of exocytic vesicles, that would facilitate a continuous translocation to and recycled from the cell surface (Prada et al., 2006). This storage of intracellular TREM2 favors the continuous functional adaptation of these cells to the microenvironment and their rapid response to microglial activation. However, the fact that TREM2 shall be exposed to the membrane to interact with their ligands, lead us to study the TREM2+ microglial population appearing after lesions, showing elevated levels of TREM2 expression in ramifications, and according to our flow cytometry study, at least with partial TREM2 expression at the level of the cytoplasmic membrane. In our study, upregulation of TREM2 expression could be observed in both models of axonal injury, being coincident with the previous general in vivo data in chronic (Jay et al., 2017a; Sayed et al., 2018) and acute CNS inflammatory injuries (Saber et al., 2017; Scott-Hewitt et al., 2017; Wu et al., 2017), as well as in other axonal injury models (Kobayashi et al., 2016; Tay et al., 2018). Additionally, an increase in sTREM2 could be found at the peak of TREM2 microglial expression after PPT, which was not detected after FNA. sTREM2 is the result of proteolytic TREM2 cleavage by ADAM10 and ADAM17 at H157–S158 peptide bond, resulting in its liberation. Despite little is known about the role of sTREM2, this fraction has been related to the potentiation of several microglial functions including pro-inflammatory cytokine production, migration, survival, and phagocytosis (Zhong et al., 2017, 2019). Although both injury models presented expression of TREM2 in microglia, differences in the pattern of expression of FIGURE 11 | Analysis of P2RY12 expression in microglial cells after PPT. this molecule were found between anterograde and retrograde (A–I) Representative images showing P2RY12 expression in the DG of the degeneration models. Concretely, after PPT, TREM2 expression hippocampus in NL conditions (A–C), at 2 dpl (D–F), and 3 dpl (G–I) after followed a pattern coincident with microglia activation PPT. In NL conditions, TREM2 (in green) is expressed accumulated adjacent to the nucleus, and high levels of P2RY12 (in red) are found in all microglial (Recasens et al., 2019), but after FNA, TREM2 levels increased cells. After lesion, at 2 dpl, most microglia show low levels of TREM2 in at 14–21 dpl and were restricted to microglial clusters that ramifications and high levels of P2RY12 expression in all microglial cells do not coincide with microglial activation, which starts at (arrows). Also, some microglia with high levels of TREM2 expression and low 3 dpl, peaks from 7 to 14 dpl, and is downregulated afterward levels of P2RY12 microglia are observed (arrowheads), being the latter the (Villacampa et al., 2015). These results suggest that not all microglial phenotype mostly seen at 3 dpl. Scale bar (A–I) = 10µm. activated microglia express TREM2 and pointed towards a specific role of this receptor in microglial function. Differences we demonstrated that the cytokine microenvironment, which in TREM2 expression could be related to the nature of the lesion modifies microglial activation, influenced TREM2 expression, (anterograde vs. retrograde) and microglial cell function along especially in the PPT paradigm and under IL-6 overexpression. the time-course of both models. On one hand, PPT is a model In the following paragraphs, we will discuss our findings with that involves the complete transection of the perforant pathway, the current literature. that produces an anterograde axonal and terminal degeneration of these axons in the outer 2/3 of the ML in the hippocampal TREM2 Is Increased in Activated Microglial DG. Due to this lesion, a rapid and robust microglial activation Cells After Both PPT and FNA is induced in the DG at 2–5 dpl (Lynch et al., 1972; Jensen TREM2 is a receptor found in the membrane of myeloid immune et al., 1999; Ladeby et al., 2005), whose principal function cells such as macrophages, osteoclasts, and microglia (Schmid is to phagocyte myelin and neuronal debris (Nielsen et al., et al., 2002; Colonna, 2003). Under basal conditions, TREM2 has 2009), contributing to the structural reorganization and axonal been reported in microglial cells, both in vitro and in vivo, collateral sprouting initiated around 7 dpl and being complete during development and in adulthood (Schmid et al., 2002; at 14 dpl (Deller et al., 2007). On the other hand, the FNA Chertoff et al., 2013). In agreement, we observed constitutive model involves a mechanical transection of the facial nerve, Frontiers in Cellular Neuroscience | www.frontiersin.org 16 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 12 | Analysis of P2RY12 expression in microglial clusters after FNA. (A–O) Representative double-immunolabeled images combining TREM2 (green) and P2RY12 (red) in the ipsilateral NL FN (A–C) and at 14 (D–I) and 21 dpI (J–O). Note that clusters with high levels of P2RY12 (arrows) can present both high and low levels of TREM2. In the same way, clusters with low levels of P2RY12 (arrowheads) show high and low levels of TREM2. No differences were detected between all the time-points in terms of P2RY12 staining. Scale bar (A–O) = 10 µm. that leads to retrograde neuronal degeneration accompanied CD16/32, and CD68, involved in this function, points out to an by a dramatic microglial activation. During early microglia especially relevant role of TREM2+ microglia after axonal injury activation and proliferation, from 3 to 7 dpl, TREM2 expression in this function. is mainly intracytoplasmatic and does not increase its expression in ramifications. Importantly, we observe an increase in TREM2+ Cells Expressed Markers Related intracellular TREM2 in this period, which could suggest a role to Proliferation and Phagocytosis After for the intracellular TREM2 pool in microglia. Later, TREM2 is FNA and PPT upregulated in specific structures called microglial clusters, The phenotypic characterization of TREM2+ microglial cells first appearing at 14 dpl, located around FMN and involved in showed expression of both proliferation and phagocytic markers phagocytosis (Moran and Graeber, 2004). after PPT, but only phagocytic markers after FNA. The fact that TREM2 increases its expression in microglia In terms of proliferation, our results showed that, after coinciding in both models with the specific time-point related to PPT, all proliferating microglia were TREM2+. In previous phagocytosis, together with the expression of different markers, reports in DAP12KO mice, a reduced microglial increase Frontiers in Cellular Neuroscience | www.frontiersin.org 17 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 13 | Flow cytometry analysis of myelin phagocytosis by TREM2+ microglia and macrophages from NL and lesioned hippocampus and FN after PPT and FNA. (A) Representative dot-plot of CD11b/CD45 expression in cells obtained from the hippocampus of NL animals. The selected region was the population of cells used in this study. (B) Representative histogram where populations of CD11b+/CD45low cells (microglia) and CD11b+/CD45high cells (macrophages) were defined. (C,D) Representative histogram-plot showing the expression of TREM2 (C) and Myelin (D) in the population of microglia/macrophages in comparison to the corresponding isotype control. (E–H) Graphs showing the percentage of TREM2− and TREM2+ cells phagocyting myelin in NL (E,F) and at 3 dpl (G,H) in CD11b+/CD45low population after PPT. (I–L) Graphs showing the percentage of TREM2− and TREM2+ cells phagocyting myelin in NL (I,J) and at 3 dpl (K,L) in CD11b+/CD45high population after PPT. (M–P) Graphs showing the percentage of TREM2− and TREM2+ cells phagocyting myelin in contralateral NL FN (M,N) and at 21 dpl (O,P) in CD11b+/CD45low population after FNA. (Q–T) Graphs showing the percentage of TREM2− and TREM2+ cells phagocyting myelin in contralateral NL FN (Q,R) and at 21 dpl (S,T) in CD11b+/CD45high population after FNA. (E–T) Results were represented as ± SEM (Unpaired Student’s t-test, ****p ≤ 0.0001, ***p ≤ 0.001, **p ≤ 0.01, *p ≤ 0.05, $ p ≤ 0.08). after spinal nerve transection (Kobayashi et al., 2016) was without TREM2 or harboring dysfunctional TREM2 variants found. Also, TREM2KO mice showed reduced microglia (Jay et al., 2017a; Cheng-Hathaway et al., 2018; Meilandt proliferation during homeostasis and after injury (Cantoni et al., 2020). On the other hand, some authors demonstrated et al., 2015; Poliani et al., 2015; Zheng et al., 2017), and that lack of TREM2 did not affect microglia proliferation in impaired proliferation was observed in AD mouse models microglial cell cultures (Noto et al., 2010) or after spinal nerve Frontiers in Cellular Neuroscience | www.frontiersin.org 18 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury transection in DAP12KO (Guan et al., 2016). From our point of view, our current study demonstrates that the increase in TREM2+ microglia depends on the specific features of microglial response to axonal injury. In PPT, proliferation is coincident with microglia activation, and the lesion is rapidly rectified; while after FNA, activation and proliferation are immediately triggered, and phagocytosis of late degenerating FMN and injury resolution are delayed. Our BrdU study demonstrated that cells that were previously proliferating after FNA, and were not showing TREM2 expression at that time, were forming phagocytic microglial clusters, which are mainly TREM2+. Yet, further studies are needed to demonstrate a clear relationship between TREM2 expression and proliferation in axonal lesions. Despite TREM2 phagocytic role has been extensively studied in acute and chronic neurodegenerative diseases models as AD or multiple sclerosis (reviewed in Deming et al., 2018; Karanfilian et al., 2020); to the best of our knowledge, this is the first study showing a link of TREM2 with phagocytic microglia after direct or primary acute axonal injury. As far as the possible role of TREM2+ microglial cells in phagocytosis is concerned, our results indicate an important implication of TREM2 in phagocytosis in both models, thus coinciding with the view of TREM2 as a receptor involved in the triggering of an ‘‘eat- me’’ signal (Neumann et al., 2009). After both PPT and FNA, most TREM2+ cells also expressed the phagocytic-associated molecules CD16/32 and CD68, and the myelin phagocytosis assay in TREM2+ microglia showed higher percentages of phagocytosis of myelin debris compared to TREM2- cells. In our results, the pattern of expression of TREM2 in microglia FIGURE 14 | Temporal pattern of TREM2 expression in microglial cells in and its specific location, especially in microglial clusters, mice with astrocyte-targeted overexpression of IL-6 (GFAP-IL6Tg) or IL-10 (GFAP-IL10Tg) compared to their corresponding wild-type (WT) after PPT and lead us to speculate that, after FNA, TREM2+ microglia are FNA. (A–D) Graphs showing either the percentage of Area (A,C) or intensity involved in the phagocytosis of degenerating FMN close to (B,D) of TREM2 expression in the OML and MML of the DG of PPT-lesioned microglial clusters, whereas in the PPT paradigm TREM2 is GFAP-IL6Tg, GFAP-IL10Tg, and their corresponding WT littermates. recognizing myelin debris, that appears at early time-points Statistically significant differences are observed between GFAP-IL6Tg and WT in this lesion (Jensen et al., 1999). Certain specificity of animals in the percentage of Area of TREM2 expression at 3 dpl (two-way ANOVA, time effect p < 0.0001, genotype p = 0.07; interaction TREM2 for phagocytosis of myelin debris and apoptotic neurons time-genotype p < 0.01; Sidak’s multiple comparison test, ***p < 0.001) as has already been suggested based also on TREM2 ligands well as the intensity of TREM2 expression (two-way ANOVA, time effect (Gervois and Lambrichts, 2019). In the case of myelin p < 0.0001, genotype p < 0.01; Sidak’s multiple comparison test, *p < 0.05). phagocytosis, similarly to our observations, TREM2 increases Also, differences between GFAP-IL10Tg and WT mice are shown in the percentage of Area of TREM2 expression (two-way ANOVA, time effect its expression correlating to myelin debris appearance in p < 0.0001, genotype p = 0.08; no significant differences were found with demyelinating models (Piccio et al., 2007; Takahashi et al., 2007; Sidak’s multiple comparison test) and the intensity of TREM2 expression Scott-Hewitt et al., 2017). Overexpression of TREM2 in (two-way ANOVA, time effect p < 0.0001, genotype p = 0.308; no significant administered bone-marrow-derived cells improved myelin differences were found with Sidak’s multiple comparison test). (E–H) Graphs clearance in EAE-induced mice (Takahashi et al., 2005), showing either the percentage of Area (E,G) or intensity (F,H) of TREM2 expression in the ipsilateral FN of GFAP-IL6Tg and GFAP-IL10Tg while TREM2 blockade increased demyelination (Piccio lesioned animals, and their corresponding WT littermates. Differences in the et al., 2007). TREM2KO microglia showed diminished percentage of Area (two-way ANOVA, time effect p < 0.0001, genotype microglia activation and was unable to degrade myelin in effect p < 0.05; Sidak’s multiple comparison test, *p < 0.01) and intensity cuprizone models (Cantoni et al., 2015; Poliani et al., 2015; (two-way ANOVA, time effect p < 0.0001, genotype effect p = 0.06; no Nugent et al., 2020), and accordingly, the treatment with significant differences detected with Sidak’s multiple comparison test) of TREM2 in GFAP-IL6Tg compared to WT were found. In the case of an agonistic antibody against TREM2 enhanced microglial GFAP-IL10Tg, no statistically significant differences between time-points were clearance and digestion of myelin, facilitated remyelination observed at neither the percentage of Area (two-way ANOVA, time effect and preserved axonal integrity (Cignarella et al., 2020). p < 0.0001, genotype effect p = 0.883, no significant differences detected Our results also suggest that TREM2+ microglia phagocyte with Sidak’s multiple comparison test) nor the intensity (two-way ANOVA, time myelin debris more efficiently, which may facilitate a effect p < 0.01, genotype effect p = 0.06; no significant differences detected with Sidak’s multiple comparison test). rapid lesion resolution after PPT, and the proper axonal collateral sprouting. Frontiers in Cellular Neuroscience | www.frontiersin.org 19 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury FIGURE 15 | Quantification of TREM2+ microglia subpopulation after PPT and TREM2+ microglial clusters after FNA in GFAP-IL6Tg, GFAP-IL10Tg, and WT mice. (A–C) Representative double-immunolabeled images combining TREM2 (green) and Iba1 (red) in the ML of the DG at 3 dpl in WT (A), GFAP-IL6Tg (B), and GFAP-IL10Tg (C). (D) Graph showing the percentage of TREM2+ ramified cells concerning the total number of Iba1 cells at 3 dpl in WT (48.1 ± 1.96%), GFAP-IL6Tg (61.97 ± 2.58%), and GFAP-IL10Tg (74.7 ± 1.91%). A higher percentage of TREM2/Iba1 cells are found in Tg mice concerning WT at 3 dpl (one-way ANOVA, *p < 0.01; post hoc Dunnet’s test, GFAP-IL6Tg vs. WT *p = 0.01, and GFAP-IL10Tg vs. WT ***p = 0.001). (E) Graph showing the percentage of microglial clusters with TREM2+ expression in their ramifications at 14 and 21 dpl after FNA in WT, GFAP-IL6Tg, and GFAP-IL10Tg. Scale bar (A–C) = 50 µm. On the other hand, a higher avidity in phagocytosis has In line with these results, ApoE, a phagocytic ligand of been established for apoptotic or degenerating cells. In vitro TREM2, was found in TREM2+ microglia after FNA, but not studies showed inhibition of phagocytosis of apoptotic neurons after PPT. ApoE has been reported to be upregulated after after TREM2 knockdown, while on the contrary, apoptotic FNA in microglia, as we and other transcriptomic studies neurons were cleared more efficiently by TREM2+ BV2 cultured found (Tay et al., 2018). Moreover, the ApoE staining in cells (Takahashi et al., 2005; Hsieh et al., 2009). TREM2KO FMN could indicate a role of this molecule as a ligand in microglia showed also decreased phagocytic capacity, and TREM2-mediated phagocytosis in clusters, as demonstrated possibly related to this fact, a reduced age-associated neuron in other studies (Krasemann et al., 2017). Apoptotic neurons loss was found in TREM2KO aged mice (Linnartz-Gerlach et al., can bind to ApoE, and Atagi et al. (2015) demonstrated 2019). In our study, TREM2 was mainly expressed in microglial that its expression enhanced microglial phagocytosis, acting clusters, reported to phagocytose degenerating FMN (Moran as an opsonin. Consequently, studies on ApoE-KO animals and Graeber, 2004), which reinforces the role of TREM2+ cells showed decreased FMN death after FNA (Krasemann et al., in neuronal phagocytosis in this model. As reported for PPT, 2017). To our knowledge, ApoE upregulation has been TREM2+ microglia after FNA showed higher phagocytic activity described in DAM microglia, which has been associated of myelin compared to TREM2- microglia, and suggested that, with axonal dystrophy in AD models but is not induced despite being two different axonal lesions with two different in neuropathic pain (Krasemann et al., 2017; Deczkowska targets, the phagocytic capacity of TREM2+ microglia in both et al., 2018). Given our results, microglial ApoE may not models is independent of their specific lesion particularities. be induced, in specific circumstances, in TREM2+ microglia Frontiers in Cellular Neuroscience | www.frontiersin.org 20 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury of injuries affecting only axons, and consequently, TREM2+ M-CSF (Nugent et al., 2020). Given these results, the relevance microglia could be phagocyting axonal debris after PPT in an of TREM2-mediated phagocytosis may be dependent on ApoE-independent manner. the microenvironment in which phagocytosis takes place. In this work, we additionally studied another recently On the other hand, the role of sTREM2 in phagocytosis described TREM2 ligand, Galectin-3 (Boza-Serrano et al., 2019). mediated by TREM2- cells may not be excluded. Interestingly, In our observations, we found that Galectin-3 was expressed PPT showed increased levels of sTREM2 after lesion, that in TREM2+ microglial cells after both types of axonal injuries. is not observed after FNA, and are concomitant with high Galectin-3 is a molecule extensively related to microglial myelin CD68 levels. sTREM2 has been involved in the stimulation phagocytosis and clearance (Puigdellivol et al., 2020), which of β-amyloid phagocytosis and lysosomal pathways, and it reinforces this role in PPT TREM2+ microglia. Also, after also increased CD68 expression in the AD model (Zhong FNA, we observed the expression of Galectin-3 in microglia et al., 2019). Therefore it could be interesting to further and on the surface of some FMN, as occurred with ApoE. explore the role of sTREM2 in the phagocytic function of Galectin-3 is similarly described in the surface of neurons after this model. traumatic brain injury, where it has been suggested to act as Finally, we explored the possibility that TREM2+ cells an alarmin (Yip et al., 2017; Boza-Serrano et al., 2019) or an presented a downregulation of the homeostatic microglial opsonizing molecule in the phagocytosis of desialylated neurons receptor P2RY12. Downregulation of this receptor, together (Nomura et al., 2017; Puigdellivol et al., 2020), thus reinforcing with the expression of TREM2, CD68, CD16/32, Galectin- this phagocytic role for TREM2+/Galectin-3+ microglia 3, and ApoE has been defined by transcriptomic studies to after FNA. be associated with the induction of a DAM-phenotype in Apart from the phagocytic particularities, TREM2-mediated microglia (Keren-Shaul et al., 2017; Krasemann et al., 2017; phagocytosis has been qualified as an ‘‘anti-inflammatory,’’ Boza-Serrano et al., 2019). In line with these works, our results because it results in the production of anti-inflammatory showed a general P2RY12 downregulation in microglia after cytokines (Takahashi et al., 2007; Hsieh et al., 2009) due PPT in the peak of activation, which is not restricted to to an antagonistic role of TREM2 upon NF-κB activation TREM2+ microglia. However, after FNA, different patterns (Yao et al., 2019). Cuprizone studies showed that the absence of P2RY12 expression were observed in TREM2+ clusters, of TREM2 expression in microglia increases axonal injury including clusters with high P2RY12 levels, which suggested and impairs a proper repair (Cantoni et al., 2015; Poliani the induction of diverse microglial phenotypes within the et al., 2015) that can be rescued with TREM2 activating TREM2+ microglia subpopulation. Similarly, other reports antibodies (Cignarella et al., 2020). In our study, in both injury also found activated microglia with high P2RY12 expression models, inflammation is downregulated around the peak of in pre-active and chronic white matter lesions of multiple TREM2 expression (Gehrmann et al., 1991; Moran and Graeber, sclerosis and in diffuse β-amyloid plaques in human samples 2004), which indicates a reparatory role for this receptor. (van Wageningen et al., 2019; Walker et al., 2020). Still, the As suggested by the above-mentioned studies, the reparatory significance of TREM2 microglia with high P2RY12 levels remain effect could be the result of the efficient TREM2-mediated to be explored. phagocytosis accompanied by the secretion of anti-inflammatory Altogether, the phenotypic characterization of TREM2+ cytokines, that could facilitate either axonal sprouting or microglia after PPT and FNA indicates a clear involvement effective FMN clearance and neuronal regeneration, depending of this molecule in a subpopulation of phagocytosing on the model. microglia, probably aimed at recognizing myelin debris Although we centered our attention on microglial and degenerating neurons, that could clear more efficiently TREM2-mediated phagocytosis, it is important to note these cell remnants. In the PPT paradigm, TREM2+ also that CD16/32+ and CD68+ microglial cells without microglia were CD16/32+, CD68+, Galectin 3+, ApoE-, TREM2 expression were found in both injury models. Our whereas, in the FNA model, TREM2+ microglia in clusters study shows that TREM2 is only triggered in a subpopulation was CD16/32+, CD68+, Galectin-3+, ApoE+ and showed of phagocytic microglial cells and strongly suggests a TREM2- P2RY12 variable expression. independent microglial phagocytosis. Indeed, some reports indicate that TREM2 is not strictly required for microglia to Pro-inflammatory Rather Than produce the engulfment of cellular remnants, but it increases Anti-Inflammatory Local the efficient phagocytosis of apoptotic neurons (Hsieh et al., 2009; Cantoni et al., 2015). These results are following our Microenvironments Influence Microglial phagocytosis study, where lower percentages of TREM2- TREM2 Expression Mainly After PPT but independent microglial phagocytosis are observed in all Also After FNA conditions when compared to TREM2+ phagocytosis of myelin. The final part of this work was to analyze possible modifications Interestingly, a recent work assigned a role for TREM2 in lipid of TREM2 in both GFAP-IL6Tg and GFAP-IL10Tg animals after digestion, excluding its involvement in the uptake of myelin PPT and FNA. To our knowledge, few in vitro and in vivo debris during microglial phagocytosis which—according to data are available on the effect of cytokines on the regulation cell cultures assays performed with TREM2KO bone-marrow- of microglial TREM2 expression (Petković et al., 2016; Zhai derived cells in this work—may be rescued by high levels of et al., 2017; Zhong et al., 2019). Moreover, our previous studies Frontiers in Cellular Neuroscience | www.frontiersin.org 21 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury demonstrated in both Tg animals and after both injury models, changes in differential microglial activation but not to a direct differences in microglial activation pattern and injury outcome, cytokine effect. which are importantly related to phagocytic. In the PPT, a higher percentage of microglial cells expressing DATA AVAILABILITY STATEMENT TREM2 was observed in both GFAP-IL10Tg and GFAP-IL6Tg compared to WT. However, this result did not match with The original contributions presented in the study are included differences in terms of microglial proliferation observed between in the article/Supplementary Materials, further inquiries can be Tg animals (Recasens et al., 2019), suggesting that other directed to the corresponding author. mechanisms related to microglial activation, and possibly related to phagocytosis, are originating differences in delayed ETHICS STATEMENT axonal sprouting. In our previous studies in the FNA model, GFAP-IL6Tg All experimental animal work was conducted according to showed increased neuronal death and less microglial clusters Spanish regulations (Ley 32/2007, Real Decreto 1201/2005, Ley (Almolda et al., 2014), and GFAP-IL10Tg presented more 9/2003, and Real Decreto 178/2004) in agreement with European neuronal survival and microglial clusters (Villacampa et al., Union directives (86/609/CEE, 91/628/CEE, and 92/65/CEE) and 2015) compared to WT. The significant reduction in the was approved by the Ethical Committee of the Autonomous TREM2+ area observed in GFAP-IL6Tg mice at 21 dpl could University of Barcelona. be explained by the lower microglia cluster formation, although our results also indicate that TREM2 expression in microglial clusters is not directly affected by cytokine overproduction, as AUTHOR CONTRIBUTIONS almost all clusters are TREM2+ in all mouse lines. Indeed, GM, AG-L, NV, and MR designed and performed the the reduced TREM2 expression in GFAP-IL6Tg due to lower experiments, including acquiring data and analyzing data; and cluster numbers could be explained by a deficient phagocytic wrote and revised the manuscript. BA designed experiments, capacity in microglia, as already observed in cuprizone-treated analyzed data; wrote and revised manuscript. KS contributed GFAP-IL6Tg (Petković et al., 2016). Thus, our results suggest to design the experiments, analyzed data, and revised the that IL-6 overproduction downregulates TREM2 expression manuscript. BC and BG designed the research and revised the due to changes in the microglial activation pattern, which manuscript. All authors contributed to the article and approved affect cluster formation, and may not to be a direct effect the submitted version. of IL-6 on TREM2 levels. Moreover, in our study, we did not find a direct link between injury outcome and TREM2 after FNA. FUNDING As a whole, TREM2 expression in Tg animals seems to be This work was supported by the Ministerio de Ciencia e more influenced by changes in microglial activation pattern that Innovación (BFU2014-55459 and BFU2017-87843-R) to BC. indirectly affects TREM2 expression, than by a direct effect of IL-6 or IL-10 overproduction in phagocytic capacity due to changes in TREM2 microglial expression. ACKNOWLEDGMENTS We would like to thank Miguel A. Martil and Isabella Appiah for CONCLUSIONS their outstanding technical help, and Núria Barba for her help In conclusion, the results obtained in this work show that on confocal microscopy. The graphical abstract was created with TREM2 expression is upregulated after both anterograde and BioRender.com. retrograde axonal injury in a specific subpopulation of microglia. Although we have not found a temporal relationship between the induction of proliferation and TREM2 expression in SUPPLEMENTARY MATERIAL microglial cells, this unique TREM2+ microglia are associated The Supplementary Material for this article can be found with a more efficient phagocytosis of specific cell remnants, online at: https://www.frontiersin.org/articles/10.3389/fncel. such as myelin and degenerated neurons. However, it is 2020.567404/full#supplementary-material. important to highlight that the phenotypic profile of TREM2+ microglial cells in the two models studied present specific SUPPLEMENTARY FIGURE 1 | Analysis of ApoE expression in astrocytes after characteristics linked to the degree of tissue injury. Concretely, perforant pathway transection (PPT). (A–F) Representative double-immunolabeled the expression of P2RY12 is only downregulated in microglia images combining ApoE (green) and GFAP (red) in the non-lesioned (NL) dentate gyrus (DG) of the hippocampus (A–C) and at 2 dpI (D–F) after PPT. Note that after PPT, whereas ApoE expression in TREM2+ cells is ApoE expression was easily found on astrocyte-like cells in each condition (white exclusively observed in retrograde axotomy, where the axonal arrows). Scale bar (A–F) = 10 µm. injury causes neuronal death. Finally, pro-inflammatory and anti-inflammatory local CNS microenvironments by IL-6 or SUPPLEMENTARY FIGURE 2 | Analysis of ApoE expression in astrocytes after IL-10 overproduction modified TREM2 expression, mainly facial nerve axotomy (FNA). (A–I) Representative double-immunolabeled images after PPT and under pro-inflammatory conditions, due to combining ApoE (green) and GFAP (red) in the ipsilateral NL FN (A–C) and at Frontiers in Cellular Neuroscience | www.frontiersin.org 22 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury 14 dpI (D–F) and 21 dpI (G–I). Note that, in all the time-points, ApoE was GFAP-IL10Tg mice (E–H). Note that, while in NL TREM2 was only depicted as detected in astrocytes (white arrows). Scale bar (A–I) = 10 µm. small rounded morphologies (arrowheads), also found at 21 dpl, at 3 and 7 dpl ramified and occasionally at 21 dpl TREM2+ cells were also observed (arrows). SUPPLEMENTARY FIGURE 3 | Temporal pattern of TREM2 changes in (I–N) Representative images showing TREM2 staining in the contralateral NL, as microglial cells in GFAP-IL6Tg and GFAP-IL10Tg after PPT and FNA. (A–H) well as the ipsilateral sides of the FN at 14 and 21 dpl of GFAP-IL6Tg (I–K) and Representative images showing TREM2 staining in the granular (GL) as well as the GFAP-IL10Tg (L–N). In NL TREM2 is mainly restricted to a perinuclear location inner, medial and outer molecular layers (IML, MML, and OML, respectively) of the (arrowheads), whereas at 14 and 21 dpl TREM2 is extended to microglia DG in NL and PPT-lesioned mice at 3, 7, and 21 dpl of GFAP-IL6Tg (A–D) and ramifications and clusters (arrows). Scale bar (A–H) = 50 µm; (I–N) = 30 µm. REFERENCES species differences between rat and mouse. Prog. Brain Res. 163, 501–528. doi: 10.1016/S0079-6123(07)63027-1 Acarin, L., Gonzalez, B., Castellano, B., and Castro, A. J. (1997). Quantitative Deming, Y., Li, Z., Benitez, B. A., and Cruchaga, C. (2018). Triggering analysis of microglial reaction to a cortical excitotoxic lesion in the receptor expressed on myeloid cells 2 (TREM2): a potential therapeutic early postnatal brain. Exp. Neurol. 147, 410–417. doi: 10.1006/exnr. target for Alzheimer disease? Expert Opin. Ther. Targets 22, 587–598. 1997.6593 doi: 10.1080/14728222.2018.1486823 Almolda, B., Costa, M., Montoya, M., Gonzalez, B., and Castellano, B. (2009). Gehrmann, J., Schoen, S. W., and Kreutzberg, G. W. (1991). Lesion of the rat CD4 microglial expression correlates with spontaneous clinical improvement entorhinal cortex leads to a rapid microglial reaction in the dentate gyrus. in the acute Lewis rat EAE model. J. Neuroimmunol. 209, 65–80. doi: 10.1016/j. A light and electron microscopical study. Acta Neuropathol. 82, 442–455. jneuroim.2009.01.026 doi: 10.1007/BF00293378 Almolda, B., de Labra, C., Barrera, I., Gruart, A., Delgado-Garcia, J. M., Gervois, P., and Lambrichts, I. (2019). The emerging role of triggering receptor Villacampa, N., et al. (2015). Alterations in microglial phenotype and expressed on myeloid cells 2 as a target for immunomodulation in ischemic hippocampal neuronal function in transgenic mice with astrocyte-targeted stroke. Front. Immunol. 10:1668. doi: 10.3389/fimmu.2019.01668 production of interleukin-10. Brain Behav. Immun. 45, 80–97. doi: 10.1016/j. Gratuze, M., Leyns, C. E. G., and Holtzman, D. M. (2018). New insights bbi.2014.10.015 into the role of TREM2 in Alzheimer’s disease. Mol. Neurodegener. 13:66. Almolda, B., Villacampa, N., Manders, P., Hidalgo, J., Campbell, I. L., Gonzalez, B., doi: 10.1186/s13024-018-0298-9 et al. (2014). Effects of astrocyte-targeted production of interleukin-6 in Greenhalgh, A. D., Zarruk, J. G., Healy, L. M., Baskar Jesudasan, S. J., Jhelum, P., the mouse on the host response to nerve injury. Glia 62, 1142–1161. Salmon, C. K., et al. (2018). Peripherally derived macrophages modulate doi: 10.1002/glia.22668 microglial function to reduce inflammation after CNS injury. PLoS Biol. Atagi, Y., Liu, C.-C., Painter, M. M., Chen, X.-F., Verbeeck, C., Zheng, H., et al. 16:e2005264. doi: 10.1371/journal.pbio.2005264 (2015). Apolipoprotein E is a ligand for triggering receptor expressed on Guan, Z., Kuhn, J. A., Wang, X., Colquitt, B., Solorzano, C., Vaman, S., et al. myeloid cells 2 (TREM2). J. Biol. Chem. 290, 26043–26050. doi: 10.1074/jbc. (2016). Injured sensory neuron-derived CSF1 induces microglial proliferation M115.679043 and DAP12-dependent pain. Nat. Neurosci. 19, 94–101. doi: 10.1038/nn.4189 Boza-Serrano, A., Ruiz, R., Sanchez-Varo, R., Garcia-Revilla, J., Yang, Y., Jimenez- Hsieh, C. L., Koike, M., Spusta, S. C., Niemi, E. C., Yenari, M., Nakamura, M. C., Ferrer, I., et al. (2019). Galectin-3, a novel endogenous TREM2 ligand, et al. (2009). A role for TREM2 ligands in the phagocytosis of apoptotic detrimentally regulates inflammatory response in Alzheimer’s disease. Acta neuronal cells by microglia. J. Neurochem. 109, 1144–1156. doi: 10.1111/j.1471- Neuropathol. 138, 251–273. doi: 10.1007/s00401-019-02013-z 4159.2009.06042.x Campbell, I. L., Abraham, C. R., Masliah, E., Kemper, P., Inglis, J. D., Jay, T. R., Hirsch, A. M., Broihier, M. L., Miller, C. M., Neilson, L. E., Oldstone, M. B., et al. (1993). Neurologic disease induced in transgenic mice Ransohoff, R. M., et al. (2017a). Disease progression-dependent effects of by cerebral overexpression of interleukin 6. Proc. Natl. Acad. Sci. U S A 90, TREM2 deficiency in a mouse model of Alzheimer’s disease. J. Neurosci. 37, 10061–10065. doi: 10.1073/pnas.90.21.10061 637–647. doi: 10.1523/JNEUROSCI.2110-16.2016 Cantoni, C., Bollman, B., Licastro, D., Xie, M., Mikesell, R., Schmidt, R., Jay, T. R., von Saucken, V. E., and Landreth, G. E. (2017b). TREM2 in et al. (2015). TREM2 regulates microglial cell activation in response to neurodegenerative diseases. Mol. Neurodegener. 12:56. doi: 10.1186/s13024- demyelination in vivo. Acta Neuropathol. 129, 429–447. doi: 10.1007/s00401- 017-0197-5 015-1388-1 Jensen, M. B., Hegelund, I. V., Poulsen, F. R., Owens, T., Zimmer, J., and Finsen, B. Cignarella, F., Filipello, F., Bollman, B., Cantoni, C., Locca, A., Mikesell, R., et al. (1999). Microglial reactivity correlates to the density and the myelination (2020). TREM2 activation on microglia promotes myelin debris clearance and of the anterogradely degenerating axons and terminals following perforant remyelination in a model of multiple sclerosis. Acta Neuropathol. 140, 513–534. path denervation of the mouse fascia dentata. Neuroscience 93, 507–518. doi: 10.1007/s00401-020-02193-z doi: 10.1016/s0306-4522(99)00139-6 Colonna, M. (2003). TREMs in the immune system and beyond. Nat. Rev. Karanfilian, L., Tosto, M. G., and Malki, K. (2020). The role of TREM2 in Immunol. 3, 445–453. doi: 10.1038/nri1106 Alzheimer’s disease; evidence from transgenic mouse models. Neurobiol. Aging Colonna, M., and Butovsky, O. (2017). Microglia function in the central nervous 86, 39–53. doi: 10.1016/j.neurobiolaging.2019.09.004 system during health and neurodegeneration. Annu. Rev. Immunol. 35, Keren-Shaul, H., Spinrad, A., Weiner, A., Matcovitch-Natan, O., Dvir- 441–468. doi: 10.1146/annurev-immunol-051116-052358 Szternfeld, R., Ulland, T. K., et al. (2017). A unique microglia type Cheng-Hathaway, P. J., Reed-Geaghan, E. G., Jay, T. R., Casali, B. T., associated with restricting development of Alzheimer’s disease. Cell 169, Bemiller, S. M., Puntambekar, S. S., et al. (2018). The Trem2 R47H 1276.e17–1290.e17.doi: 10.1016/j.cell.2017.05.018 variant confers loss-of-function-like phenotypes in Alzheimer’s disease. Mol. Kobayashi, M., Konishi, H., Sayo, A., Takai, T., and Kiyama, H. (2016). Neurodegener. 13:29. doi: 10.1186/s13024-018-0262-8 TREM2/DAP12 signal elicits proinflammatory response in microglia Chertoff, M., Shrivastava, K., Gonzalez, B., Acarin, L., and Gimenez-Llort, L. and exacerbates neuropathic pain. J. Neurosci. 36, 11138–11150. (2013). Differential modulation of TREM2 protein during postnatal brain doi: 10.1523/JNEUROSCI.1238-16.2016 development in mice. PLoS One 8:e72083. doi: 10.1371/journal.pone. Kobayashi, M., Konishi, H., Takai, T., and Kiyama, H. (2015). A DAP12- 0072083 dependent signal promotes pro-inflammatory polarization in microglia Deczkowska, A., Keren-Shaul, H., Weiner, A., Colonna, M., Schwartz, M., following nerve injury and exacerbates degeneration of injured neurons. Glia and Amit, I. (2018). Disease-associated microglia: a universal immune 63, 1073–1082. doi: 10.1002/glia.22802 sensor of neurodegeneration. Cell 173, 1073–1081. doi: 10.1016/j.cell.2018. Kober, D. L., and Brett, T. J. (2017). TREM2-ligand interactions in health and 05.003 disease. J. Mol. Biol. 429, 1607–1629. doi: 10.1016/j.jmb.2017.04.004 Deller, T., Del Turco, D., Rappert, A., and Bechmann, I. (2007). Structural Krasemann, S., Madore, C., Cialic, R., Baufeld, C., Calcagno, N., El Fatimy, R., reorganization of the dentate gyrus following entorhinal denervation: et al. (2017). The TREM2-APOE pathway drives the transcriptional phenotype Frontiers in Cellular Neuroscience | www.frontiersin.org 23 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury of dysfunctional microglia in neurodegenerative diseases. Immunity 47, Saber, M., Kokiko-Cochran, O., Puntambekar, S. S., Lathia, J. D., and 566.e9–581.e9. doi: 10.1016/j.immuni.2017.08.008 Lamb, B. T. (2017). Triggering receptor expressed on myeloid cells Ladeby, R., Wirenfeldt, M., Garcia-Ovejero, D., Fenger, C., Dissing-Olesen, L., 2 deficiency alters acute macrophage distribution and improves recovery Dalmau, I., et al. (2005). Microglial cell population dynamics in the injured after traumatic brain injury. J. Neurotrauma 34, 423–435. doi: 10.1089/neu.20 adult central nervous system. Brain Res. Brain Res. Rev. 48, 196–206. 16.4401 doi: 10.1016/j.brainresrev.2004.12.009 Sayed, F. A., Telpoukhovskaia, M., Kodama, L., Li, Y., Zhou, Y., Le, D., et al. Linnartz-Gerlach, B., Bodea, L.-G., Klaus, C., Ginolhac, A., Halder, R., (2018). Differential effects of partial and complete loss of TREM2 on microglial Sinkkonen, L., et al. (2019). TREM2 triggers microglial density and age-related injury response and tauopathy. Proc. Natl. Acad. Sci. U S A 115, 10172–10177. neuronal loss. Glia 67, 539–550. doi: 10.1002/glia.23563 doi: 10.1073/pnas.1811411115 Lynch, G., Matthews, D. A., Mosko, S., Parks, T., and Cotman, C. (1972). Induced Scott-Hewitt, N. J., Folts, C. J., Hogestyn, J. M., Piester, G., Mayer- acetylcholinesterase-rich layer in rat dentate gyrus following entorhinal lesions. Proschel, M., and Noble, M. D. (2017). Heterozygote galactocerebrosidase Brain Res. 42, 311–318. doi: 10.1016/0006-8993(72)90533-1 (GALC) mutants have reduced remyelination and impaired myelin debris Meilandt, W. J., Ngu, H., Gogineni, A., Lalehzadeh, G., Lee, S. H., Srinivasan, K., clearance following demyelinating injury. Hum. Mol. Genet. 26, 2825–2837. et al. (2020). Trem2 deletion reduces late-stage amyloid plaque accumulation, doi: 10.1093/hmg/ddx153 elevates the Aβ42:Aβ40 ratio, and exacerbates axonal dystrophy and dendritic Schmid, C. D., Sautkulis, L. N., Danielson, P. E., Cooper, J., Hasel, K. W., spine loss in the PS2APP Alzheimer’s mouse model. J. Neurosci. 40, 1956–1974. Hilbush, B. S., et al. (2002). Heterogeneous expression of the triggering receptor doi: 10.1523/JNEUROSCI.1871-19.2019 expressed on myeloid cells-2 on adult murine microglia. J. Neurochem. 83, Moran, L. B., and Graeber, M. B. (2004). The facial nerve axotomy model. Brain 1309–1320. doi: 10.1046/j.1471-4159.2002.01243.x Res. Brain Res. Rev. 44, 154–178. doi: 10.1016/j.brainresrev.2003.11.004 Schnell, S. A., Staines, W. A., and Wessendorf, M. W. (1999). Reduction of Neumann, H., Kotter, M. R., and Franklin, R. J. (2009). Debris clearance by lipofuscin-like autofluorescence in fluorescently labeled tissue. J. Histochem. microglia: an essential link between degeneration and regeneration. Brain 132, Cytochem. 47, 719–730. doi: 10.1177/002215549904700601 288–295. doi: 10.1093/brain/awn109 Takahashi, K., Prinz, M., Stagi, M., Chechneva, O., and Neumann, H. Nielsen, H. H., Ladeby, R., Fenger, C., Toft-Hansen, H., Babcock, A. A., (2007). TREM2-transduced myeloid precursors mediate nervous tissue debris Owens, T., et al. (2009). Enhanced microglial clearance of myelin debris in clearance and facilitate recovery in an animal model of multiple sclerosis. PLoS T cell-infiltrated central nervous system. J. Neuropathol. Exp. Neurol. 68, Med. 4:e124. doi: 10.1371/journal.pmed.0040124 845–856. doi: 10.1097/NEN.0b013e3181ae0236 Takahashi, K., Rochford, C. D., and Neumann, H. (2005). Clearance of Nomura, K., Vilalta, A., Allendorf, D. H., Hornik, T. C., and Brown, G. C. (2017). apoptotic neurons without inflammation by microglial triggering receptor Activated microglia desialylate and phagocytose cells via neuraminidase, expressed on myeloid cells-2. J. Exp. Med. 201, 647–657. doi: 10.1084/jem. galectin-3, and mer tyrosine kinase. J. Immunol. 198, 4792–4801. 20041611 doi: 10.4049/jimmunol.1502532 Tay, T. L., Sagar, Dautzenberg, J., Grun, D., and Prinz, M. (2018). Unique Noto, D., Takahashi, K., Miyake, S., and Yamada, M. (2010). In vitro differentiation microglia recovery population revealed by single-cell RNAseq following of lineage-negative bone marrow cells into microglia-like cells. Eur. J. Neurosci. neurodegeneration. Acta Neuropathol. Commun. 6:87. doi: 10.1186/s40478- 31, 1155–1163. doi: 10.1111/j.1460-9568.2010.07152.x 018-0584-3 Nugent, A. A., Lin, K., van Lengerich, B., Lianoglou, S., Przybyla, L., Davis, S. S., van Wageningen, T. A., Vlaar, E., Kooij, G., Jongenelen, C. A. M., Geurts, J. J. G., et al. (2020). TREM2 regulates microglial cholesterol metabolism upon chronic and van Dam, A. M. (2019). Regulation of microglial TMEM119 and phagocytic challenge. Neuron 105, 837.e9–854.e9.doi: 10.1016/j.neuron.2019. P2RY12 immunoreactivity in multiple sclerosis white and gray matter lesions 12.007 is dependent on their inflammatory environment. Acta Neuropathol. Commun. Painter, M. M., Atagi, Y., Liu, C.-C., Rademakers, R., Xu, H., Fryer, J. D., et al. 7:206. doi: 10.1186/s40478-019-0850-z (2015). TREM2 in CNS homeostasis and neurodegenerative disease. Mol. Villacampa, N., Almolda, B., Vilella, A., Campbell, I. L., Gonzalez, B., and Neurodegener. 10:43. doi: 10.1186/s13024-015-0040-9 Castellano, B. (2015). Astrocyte-targeted production of IL-10 induces changes Petković, F., Campbell, I. L., Gonzalez, B., and Castellano, B. (2016). Astrocyte- in microglial reactivity and reduces motor neuron death after facial nerve targeted production of interleukin-6 reduces astroglial and microglial axotomy. Glia 63, 1166–1184. doi: 10.1002/glia.22807 activation in the cuprizone demyelination model: implications for Walker, D. G., Tang, T. M., Mendsaikhan, A., Tooyama, I., Serrano, G. E., Sue, L. I., myelin clearance and oligodendrocyte maturation. Glia 64, 2104–2119. et al. (2020). Patterns of expression of purinergic receptor P2RY12, a putative doi: 10.1002/glia.23043 marker for non-activated microglia, in aged and Alzheimer’s disease brains. Int. Piccio, L., Buonsanti, C., Mariani, M., Cella, M., Gilfillan, S., Cross, A. H., J. Mol. Sci. 21:678. doi: 10.3390/ijms21020678 et al. (2007). Blockade of TREM-2 exacerbates experimental autoimmune Wang, Y., Cella, M., Mallinson, K., Ulrich, J. D., Young, K. L., Robinette, M. L., encephalomyelitis. Eur. J. Immunol. 37, 1290–1301. doi: 10.1002/eji.2006 et al. (2015). TREM2 lipid sensing sustains the microglial response in an 36837 Alzheimer’s disease model. Cell 160, 1061–1071. doi: 10.1016/j.cell.2015.01.049 Poliani, P. L., Wang, Y., Fontana, E., Robinette, M. L., Yamanishi, Y., Gilfillan, S., Wolf, S. A., Boddeke, H. W., and Kettenmann, H. (2017). Microglia in physiology et al. (2015). TREM2 sustains microglial expansion during aging and and disease. Annu. Rev. Physiol. 79, 619–643. doi: 10.1146/annurev-physiol- response to demyelination. J. Clin. Invest. 125, 2161–2170. doi: 10.1172/JCI 022516-034406 77983 Wu, R., Li, X., Xu, P., Huang, L., Cheng, J., Huang, X., et al. (2017). Prada, I., Ongania, G. N., Buonsanti, C., Panina-Bordignon, P., and Meldolesi, J. TREM2 protects against cerebral ischemia/reperfusion injury. Mol. Brain 10:20. (2006). Triggering receptor expressed in myeloid cells 2 (TREM2) trafficking doi: 10.1186/s13041-017-0296-9 in microglial cells: continuous shuttling to and from the plasma membrane Yao, H., Coppola, K., Schweig, J. E., Crawford, F., Mullan, M., and Paris, D. regulated by cell stimulation. Neuroscience 140, 1139–1148. doi: 10.1016/j. (2019). Distinct signaling pathways regulate TREM2 phagocytic and NFκB neuroscience.2006.03.058 antagonistic activities. Front. Cell. Neurosci. 13:457. doi: 10.3389/fncel.2019. Puigdellivol, M., Allendorf, D. H., and Brown, G. C. (2020). Sialylation and 00457 Galectin-3 in microglia-mediated neuroinflammation and neurodegeneration. Yip, P. K., Carrillo-Jimenez, A., King, P., Vilalta, A., Nomura, K., Front. Cell. Neurosci. 14:162. doi: 10.3389/fncel.2020.00162 Chau, C. C., et al. (2017). Galectin-3 released in response to traumatic Recasens, M., Shrivastava, K., Almolda, B., Gonzalez, B., and Castellano, B. (2019). brain injury acts as an alarmin orchestrating brain immune response Astrocyte-targeted IL-10 production decreases proliferation and induces and promoting neurodegeneration. Sci. Rep. 7:41689. doi: 10.1038/srep a downregulation of activated microglia/macrophages after PPT. Glia 67, 41689 741–758. doi: 10.1002/glia.23573 Zhai, Q., Li, F., Chen, X., Jia, J., Sun, S., Zhou, D., et al. (2017). Triggering receptor Rolfe, A. J., Bosco, D. B., Broussard, E. N., and Ren, Y. (2017). In vitro phagocytosis expressed on myeloid cells 2, a novel regulator of immunocyte phenotypes, of myelin debris by bone marrow-derived macrophages. J. Vis. Exp. 130:56322. confers neuroprotection by relieving neuroinflammation. Anesthesiology 127, doi: 10.3791/56322 98–110. doi: 10.1097/ALN.0000000000001628 Frontiers in Cellular Neuroscience | www.frontiersin.org 24 November 2020 | Volume 14 | Article 567404 Manich et al. TREM2+ Microglia After Axonal Injury Zheng, H., Jia, L., Liu, C. C., Rong, Z., Zhong, L., Yang, L., et al. (2017). Conflict of Interest: The authors declare that the research was conducted in the TREM2 promotes microglial survival by activating Wnt/β-catenin pathway. absence of any commercial or financial relationships that could be construed as a J. Neurosci. 37, 1772–1784. doi: 10.1523/JNEUROSCI.2459-16.2017 potential conflict of interest. Zhong, L., Chen, X.-F., Wang, T., Wang, Z., Liao, C., Wang, Z., et al. (2017). Soluble TREM2 induces inflammatory responses and enhances Copyright © 2020 Manich, Gómez-López, Almolda, Villacampa, Recasens, microglial survival. J. Exp. Med. 214, 597–607. doi: 10.1084/jem.201 Shrivastava, González and Castellano. This is an open-access article distributed 60844 under the terms of the Creative Commons Attribution License (CC BY). The use, Zhong, L., Xu, Y., Zhuo, R., Wang, T., Wang, K., Huang, R., et al. distribution or reproduction in other forums is permitted, provided the original (2019). Soluble TREM2 ameliorates pathological phenotypes by modulating author(s) and the copyright owner(s) are credited and that the original publication microglial functions in an Alzheimer’s disease model. Nat. Commun. 10:1365. in this journal is cited, in accordance with accepted academic practice. No use, doi: 10.1038/s41467-019-09118-9 distribution or reproduction is permitted which does not comply with these terms. Frontiers in Cellular Neuroscience | www.frontiersin.org 25 November 2020 | Volume 14 | Article 567404