Microglial-to-neuronal CCR5 signaling regulates autophagy in neurodegeneration

In neurodegenerative diseases, microglia switch to an activated state, which results in excessive secretion of pro-inﬂammatory factors. Our work aims to investigate how this paracrine signaling affects neuronal function. Here, we show that activated microglia mediate non-cell-autonomous inhibition of neuronal autophagy, a degradative pathway critical for the removal of toxic, aggregate-prone proteins accumulating in neurodegenerative diseases. We found that the microglial-derived CCL-3/-4/-5 bind and activate neuronal CCR5, which in turn promotes mTORC1 activation and disrupts autophagy and aggregate-prone protein clearance. CCR5 and its cognate chemokines are upregulated in the brains of pre-manifesting mouse models for Huntington’s disease (HD) and tauopathy, suggesting a pathological role of this microglia-neuronal axis in


In brief
Festa et al. identified microglial-derived CCL3, CCL4, and CCL5 and their neuronal receptor CCR5 as inhibitors of autophagy, a process that enables degradation of the proteins causing Huntington's disease and tauopathies. Pharmacological or genetic inhibition of CCR5 rescues autophagy and neurodegeneration in mouse models of these diseases.

INTRODUCTION
Neurodegenerative diseases, including tauopathies and Huntington's disease (HD), are characterized by the accumulation of aggregate-prone proteins in the neuronal cytoplasm. These proteins, like mutant huntingtin (mHTT) (in HD) and tau (in various dementias), cause pathology via toxic gain-of-function mechanisms. 1 Neurons possess protective mechanisms to prevent the accumulation of toxic aggregate-prone proteins, like macroautophagy (henceforth autophagy), a major intracytoplasmic protein degradation pathway. In autophagy, double-membraned autophagosomes engulf portions of cytoplasm and deliver them to the lysosomes for degradation. 2 Autophagy is critical for the removal of intracellular aggregate-prone proteins causing neurodegeneration (like huntingtin and tau), and impairing autophagy triggers their accumulation and promotes their toxicities. Strengthening this concept, defects in multiple steps of the autophagy pathway have been identified in the early stages of many neurodegenerative diseases. Furthermore, genetic or pharmacological induction of autophagy enhances the clearance of aggregate-prone proteins and attenuates deleterious phenotypes observed in animal models mimicking these disorders. 3,4 Understanding the mechanisms underlying defective neuronal autophagy in neurodegeneration is key for identifying new therapeutic targets. Most studies have focused on cell-autonomous mechanisms, neglecting the potential impact that aberrant paracrine signaling derived from other brain cell types might have in modulating neuronal autophagy in disease. 5 In physiological conditions, bidirectional neuron-microglia communication via the secretion of soluble factors regulates the functions of these cells. 6 In neurodegenerative diseases, this communication becomes dysfunctional. In the prodromal phase of HD and Alzheimer's disease (the most common tauopathy), microglia become activated by switching from quiescent to pro-inflammatory states. 7,8 This results in increased expression of surface receptors and chronic secretion of cytokines, especially chemokines, recently identified as major mediators of neuroinflammation. Importantly, surface receptors for those chemokines are expressed in neurons. 9 The interactions between chemokines and neuronal receptors can activate neurotoxic pathways in neurodegeneration, but it is unclear whether they regulate neuronal autophagy. 10 In this study, we characterized a detrimental role of activated microglia, which inhibit neuronal autophagy non-cell autonomously. We identified microglial-derived CCL3, CCL4, and CCL5 and their neuronal receptor CCR5 as key mediators of this process and suitable targets for therapeutic intervention in tauopathies and HD. This microglia-neuron axis was upregulated in the brains of neurodegenerative disease mouse models and was self-sustaining, as CCL5:CCR5 autophagy inhibition also impaired CCR5 degradation itself.

RESULTS
Activated microglia regulate neuronal autophagy non-cell autonomously via secretion of CCL3, CCL4, and CCL5 To understand whether activated microglia regulate neuronal autophagy in a non-cell-autonomous fashion, we incubated mouse primary neurons in conditioned media derived from mouse primary microglia previously cultured in basal conditions or activated with lipopolysaccharide (LPS) and interferon gamma (IFNg). Recently, it has been shown that these factors provoke Alzheimer's disease-relevant changes in mouse and human induced pluripotent stem cell (iPSC)-derived microglial transcriptomes, suggesting that they phenocopy the microglial activation and the inflammatory response observed in neurodegenerative conditions. 11 Primary neurons in basal conditions, or treated with LPS and IFNg, were used as controls ( Figure 1A). We assessed neuronal autophagy by measuring LC3-II protein levels, which correlate with autophagosome load. Autophagy flux was monitored after 8-h treatment with 400 nM bafilomycin A1 (BAF), an inhibitor of the vacuolar-type ATPase (V-ATPase) proton pump, which blocks lysosomal LC3-II degradation and hence enables inference of autophagosome formation rates. No significant increase in lactate dehydrogenase (LDH) release was seen in BAF-treated cells compared with DMSO, excluding a major cytotoxic effect of this treatment in neurons ( Figure S1A).
Neurons cultured with conditioned media from activated microglia exhibited a significant reduction of LC3-II formation compared with control neurons, where LPS and IFNg were added directly to the neurobasal media ( Figure 1B). No significant changes were observed when neurons were incubated with non-activated microglial medium compared with control neurons cultured in neurobasal media ( Figure 1B). Similar results were obtained using conditioned media from BV2 cells (Figure S1B), a widely used model of immortalized microglia. Furthermore, co-culture of neurons with activated but not control BV2 cells, using cell culture inserts that allow free transfer of molecules between cells without direct contact between them, reduced neuronal LC3-II levels ( Figures S1C and S1D). These data suggest that microglia can secrete factors that inhibit neuronal autophagy.
To identify potential secreted mediators of this process, we analyzed published RNA sequencing performed on microglia from aged mice, 12 where microglial activation occurs concomitantly with decreased autophagy in the brain. 13,14 CCL3, CCL4, and CCL5 were three of the most highly upregulated factors among the chemokines and cytokines ( Figures 1C, S1E, and S1F). ELISA analysis of CCL3, CCL4, and CCL5 in conditioned media derived from control or activated microglia cells (primary and BV2) confirmed that the secretion of these chemokines was increased when microglial cells were activated with LPS and IFNg ( Figures 1D and S1G).
To investigate whether CCL3, CCL4, and CCL5 were the factors secreted by activated microglia that inhibited neuronal autophagy, we immunodepleted these chemokines from the conditioned media from activated microglia ( Figures 1E and S1G) and found that this rescued the decrease in LC3-II levels (Figure 1F). These data strongly suggest that CCL3, CCL4, and CCL5 are secreted from activated microglia and detrimentally affect neuronal autophagy.
Next, we assessed whether the decreased production of autophagosomes in CCL-3/-4/-5-treated neurons was associated with altered levels of three autophagy substrates: (1) the autophagy receptor SQSTM1, (2) aggresomes, (3) and polyQ aggregates caused by the overexpression of the EGFP-Q80 peptide. 20 CCL5-treated WT primary neurons exhibited a significant increase in SQSTM1 levels compared with untreated cells, as shown by immunoblot and by immunofluorescence ( Figures 2C and S2C). Likewise, the area occupied by the Proteostat-stained aggresomes (a dye that becomes brightly fluorescent upon binding to aggregated proteins) and the proportion of neurons with Q80-EGFP aggregates were increased when the cells were treated with CCL5 ( Figures 2D and S2D). We found similar results when neurons were treated with CCL4 and CCL3, the other two CCR5-activating ligands (Figures 2D, S2C, and S2D).
CCL5 was the most upregulated chemokine secreted by activated primary microglia. Therefore, we used it for proof-ofconcept experiments in HeLa CCR5-GFP cells. Similar to what we observed in primary neurons, CCL5 treatment in HeLa CCR5-GFP cells induced defective autophagy and altered levels of its substrates, here monitored by the levels of aggregates formed by overexpression of mHTT exon 1 21 (Figures S3A and  S3B). The accumulation of aggregates caused by CCL5 occurs in control cells but not in cells with knockdowns of the key autophagy genes ATG7/ATG10 ( Figures S3C and S3D). Importantly, changes in LC3-II and in the amounts of SQSTM1, Proteostatstained aggresomes, and Q80 or mHTT aggregates were not detected in CCR5-knockout (KO) (i.e., Ccr5 À/À ) primary neurons or CCR5-silenced HeLa cells, suggesting that CCL-3/-4/-5 impair autophagy via CCR5 ( Figures 2B-2D, S2A-S2D, S3A, and S3B).
We tested whether CCR5 agonism inhibited autophagy by activating mechanistic target of rapamycin complex (mTORC1), which has previously been linked to this GPCR. [22][23][24] activity, which inhibits autophagy, 25 was assessed by measuring the phosphorylation of its substrate p70 S6 kinase (p70S6K) and S6 ribosomal subunit, a substrate of p70S6K. CCR5 WT primary neurons or HeLa CCR5-GFP cells treated with CCL5 showed increased mTORC1 activation compared with control. In contrast, CCL5 failed to activate the mTORC1 pathway in CCR5-KO primary neurons or CCR5-silenced HeLa cells, suggesting that this chemokine signals to mTORC1 via CCR5 ( Figures 2E and S3E). mTORC1 chemical inhibition (Torin 1) ( Figures 2F and S3F) or genetic disruption (RAPTOR knockdown) ( Figure S3G) prevented the CCL5-induced decrease of LC3-II levels, indicating that mTORC1 activation underlies the autophagy defect caused by this chemokine. CCL4 and CCL3 treatment of primary neurons resulted in similar CCR5-dependent effects on the mTORC1-autophagy axis ( Figures S3H-S3K).
CCL5-induced CCR5 activation increases mTORC1 activity via PI3K-AKT-TSC2 pathway Recent studies in macrophages showed that chemokine stimulation of CCR5 activates AKT signaling pathway via Gai-PI3K. 24,26,27 Importantly, AKT can promote mTORC1 activity by phosphorylating and inhibiting its well-known suppressor  . Mouse primary neurons were treated with conditioned media from primary microglia cultured in basal conditions or previously activated by LPS (10 ng/mL) + INFg (100 ng/mL) for 24 h. Neurons alone cultured in neurobasal media with or without LPS+INFg were used as controls. (B) Western blotting and densitometry analyses of LC3-II protein levels in mouse primary neurons treated as described in (A) and incubated in DMSO or 400 nM bafilomycin A1 (BAF) for 8 h before being lysed. TUBULIN was loading control. Plots represent mean ± SEM (n = 4 independent experiments). Two-tailed, onesample t test *p < 0.05, ns, not significant. (C) Differential expression of chemokines in microglia from young and old mice. Scatterplot of Àlog 10 p value against log 2 fold-change from young to old. Data from Ximerakis et al. 12 (D) ELISA measurement of CCL3, CCL4, and CCL5 levels in conditioned media from mouse primary microglia cultured in basal conditions or activated by LPS+INFg. Plots represent mean ± SEM (n = 3 independent experiments). Two-tailed, unpaired t test *p < 0.05, ***p < 0.001. (E) Schematic of conditioned media used in experiments in (F). CCL3, CCL4, and CCL5 were immunodepleted from conditioned media derived from BV2 cells, previously activated by LPS (10 ng/mL) +INFg (100 ng/mL), using specific IgG-anti-CCL3/-4/-5. Neurobasal media supplemented with LPS+INFg or conditioned media from activated BV2 cells were incubated with non-specific IgG and used as controls. (F) Western blotting and densitometry analyses of LC3-II levels in mouse primary neurons treated with CCL-3/-4/-5-immunodepleted conditioned media from activated BV2 cells or activated BV2 and neurobasal media treated as described in (E) and incubated in 400 nM BAF for 8 h before being lysed. TUBULIN was loading control. Plots represent mean ± SEM (n = 3 independent experiments). Two-tailed, paired t test *p < 0.05. See also Figure S1.    TSC2. 28 PI3K-mediated phosphorylation of AKT (S473 and T308) and AKT-mediated phosphorylation of TSC2 (T1462) were increased in CCR5 WT mouse primary neurons and HeLa CCR5-GFP cells exposed to CCL5. CCR5 silencing in HeLa cells or CCR5 depletion in neurons abrogated the effect of chemokines on the phosphorylation of these proteins, suggesting that CCR5 is required for CCL5-mediated activation of PI3K-AKT-TSC2 signaling in both cell types and this pathway is conserved across species (Figures 3A and 3B).
(E) Western blotting of TSC2, P-p70S6K (T389), and p70S6K in HeLa CCR5-GFP cells transfected with control or TSC2 siRNA for 5 days and cultured in presence or absence of CCL5 for an additional 30 min before being lysed. GAPDH was loading control. Adjacent graphs show quantification of P-p70/p70 ratios. Plots represent mean ± SEM (n = 3 independent experiments). In (C)-(E), one-tailed, paired Student's t test *p < 0.05, **p < 0.01. ns, not significant.  (A and E) Western blotting and densitometry analyses of CCR5 levels in brain lysates from HD:Ccr5 +/+ (A) or PS19:Ccr5 +/+ (E) mice and corresponding controls. A Ccr5 À/À mouse brain lysate was used in (A) for validating the specificity of the anti-CCR5 Ab. GAPDH was loading control. Plots represent mean ± SEM (n = 5-7 mice per group).
(legend continued on next page) ll OPEN ACCESS Article CCR5 expression is altered in neurodegenerative disease mouse brains, and its genetic depletion is neuroprotective We observed elevated CCR5 protein levels in the brain lysates of pre-manifesting HD and PS19 tauopathy mice ( Figures 4A and  4E) compared with controls, and this increase occurs specifically in neurons ( Figures 4B and 4F). The specificity of the CCR5 antibody used in this study was confirmed by the absence of a specific band when brain lysates derived from CCR5 KO animals were analyzed by immunoblot ( Figure 4A) and by the lack of fluorescence signal when brain slices derived from CCR5 KO mice were immunostained and analyzed by confocal microscopy ( Figure S4A). In brains of both disease mouse models, the fluorescence intensity of IBA1, a marker for microglia activation, was increased compared with the controls ( Figures 4C and 4G), suggesting the presence of microgliosis. We also observed differences in microglial morphology in HD mouse brains. Using 3D reconstruction of z stacks, we observed that microglia appeared ramified in WT mouse brain, whereas microglia from the HD mouse brain appeared rounder and ameboid ( Figures S4B and S4C). These morphological changes are suggestive of microglia activation. 29 As we showed that microglial activation resulted in increased secretion of CCL3, CCL4, and CCL5 in cell culture (Figures 1D and S1G), we analyzed the brain concentrations of these chemokines in both HD and PS19 mouse models and found them increased compared with control animals ( Figures 4D and 4H). Thus, CCR5 activation might occur in the brains of these mice.
The prodromal phases of many neurodegenerative diseases, such as HD and tauopathies, are characterized by increased levels of inflammation in the brain associated with altered mTORC1-autophagy pathways. [30][31][32][33][34] Indeed, mTORC1 signaling, assessed by quantifying the fluorescence intensity of P-S6, was increased in the brains of pre-manifesting HD and PS19 mouse models (Figures 5A and 5B), while LC3-II levels were decreased ( Figures 5C and 5D).
CCR5 depletion significantly normalized mTORC1 signaling and LC3-II levels in pre-manifesting HD and PS19 mouse brains (Figures 5A-5D) and lowered the amount of HTT aggregates and insoluble tau in manifesting mice ( Figures 5E and 5F), likely because of autophagy activation. Of note, HTT aggregates were significantly decreased in mice expressing Ccr5 in heterozygosis, suggesting a potential pharmacological benefit in suppressing CCR5-mediated signaling in this disease ( Figure 5E). Altogether, these data suggest that CCR5 activation induces, or at least contributes to, the dysfunctional mTOR-autophagy pathway observed in the brains of pre-manifesting neurodegenerative mouse models.
Furthermore, in HD mice depleted of CCR5, the severity of tremors was significantly improved at several time points (Figure S4D). Muscle coordination and endurance, as assessed with wire maneuver and grip strength tests, were also significantly ameliorated in CCR5 KO-HD mice at 15 weeks of age ( Figures S4E and S4F). Thus, CCR5 depletion improved the motor phenotypes of HD mice in all the tests shown. Depletion of CCR5 also ameliorated cognitive deficits in the tauopathy mouse model-CCR5 KO-PS19 mice performed significantly better on the novel object recognition test ( Figure S4G).

Pathogenic tau expression modulates CCR5 and autophagy levels
Our analysis in HD and PS19 mice indicated that the CCR5mediated autophagy defect occurs in pre-manifesting animals. To reinforce the pathophysiological relevance of our finding, we analyzed CCR5 and LC3-II levels in brains of pre-manifesting rTg4510 mice, another tauopathy model. 35 These mice overexpress mutant human tau P301L protein in the forebrain controlled by the Tet-Off expression system ( Figure S5A) and develop progressive age-related neurofibrillary tangles, neuronal loss, and behavioral impairments, thus allowing investigation of pathogenic events over the course of the disease. 36 To validate the efficiency of the Tet-Off system, mice were fed with control and doxycycline diet for 6 weeks ( Figure S5B). In contrast to control-fed animals, doxycycline-fed rTg4510 mice exhibited decreased tau levels over time ( Figure S5C). At the end of the treatment, as mice of both groups were 16 weeks old and rTg4510 mice did not show tau aggregates and behavioral abnormalities, they were considered pre-manifesting and suitable for comparative analysis of CCR5-autophagy signaling. Compared with control single transgenic mice (only expressing the tetracycline-controlled transactivator [tTA]), rTg4510 mice had increased CCR5 expression and lower LC3-II levels. Both phenotypes were rescued when rTg4510 mice were fed with doxycycline (i.e., when tau expression was suppressed) ( Figures S5D and S5E). This suggests that tau drives CCR5 and autophagy perturbation, strengthening the link between alteration in the CCR5-autophagy pathway and the pathogenesis of neurodegenerative diseases, or at least tauopathies.
CCR5 inhibition by maraviroc rebalances mTORC1autophagy axis and pathologies in HD and tauopathy mouse models As CCR5 acts as a co-receptor for HIV, there has been a growing interest in developing pharmacological CCR5 inhibitors, 37 and this has led to a U.S. Food and Drug Administration (FDA)approved drug, maraviroc (MVC), that selectively blocks CCR5 (B and F) Representative confocal micrographs (maximum intensity projection) and quantification of CCR5 signal (green) in NeuN + neurons (red) of brain slices derived from HD:Ccr5 +/+ (B) or PS19:Ccr5 +/+ (F) mice and corresponding controls. Plots represent mean ± SEM. (n = 3 mice/group. n = 3 randomly selected fields/ mouse, each containing $400-600 NeuN + neurons) Scale bar, 50 mm. (C and G) Representative confocal micrographs (maximum intensity projection) and quantification of IBA1 signal (red) in brain slices derived from HD: Ccr5 +/+ (C) or PS19:Ccr5 +/+ (G) mice and corresponding controls. Plots represent mean ± SEM. (n = 5/6 mice/group. n = 3 randomly selected fields/mouse) Nuclei were counterstained with DAPI. Scale bar, 50 mm. (D and H) ELISA measurement of CCL3, CCL4, and CCL5 levels in brains from HD:Ccr5 +/+ (D) or PS19:Ccr5 +/+ (H) mice and corresponding controls. Plots represent mean ± SEM (n = 5-7 mice/group). Statistical tests on these data have been applied upon exclusion of outliers (black dots) using ROUT test (Q = 1%). All data in this figure were analyzed by using two-tailed, unpaired Student's t test, *p < 0.05, **p < 0.01, ***p < 0.001. ns, not significant. See also Figure S4. activity. 38 MVC treatment of WT mouse primary neurons prevented the mTORC1 activation caused by CCL5 stimulation of CCR5 ( Figure 6A). To monitor the effect of MVC-mediated CCR5 inhibition on autophagy flux, we used primary neurons derived from transgenic mice expressing mRFP-GFP-LC3 (here referred to as Traffic light mice [Tfl]). In this system, where LC3 is double tagged with both GFP and a red fluorescent protein (mRFP), one can distinguish non-acidified autophagosomes (red and green = yellow) from acidified autolysosomes (red only), as the GFP fluorescence is more rapidly quenched by the low lysosomal pH, thus allowing assessment of autophagy flux. Compared with control, CCL5-treated neurons exhibited a significant decrease of autolysosomes (red-only puncta), suggesting a decrease in autophagy flux, which is compatible with increased mTOR signaling. MVC significantly rescued autolysosome (red vesicles) numbers ( Figure 6B) and prevented the abnormal increase in polyQ aggregates caused by CCL5 treatment in neurons, likely due to autophagy activation ( Figure 6C).
To assess the in vivo efficacy of MVC on the mTORC1-autophagy axis and levels of neurotoxic proteins, we used doubletransgenic HD mouse model also expressing mRFP-GFP-LC3 (HD-Tfl mice). MVC was administered to HD-Tfl mice and their corresponding controls aged 8 weeks through daily intraperitoneal injections for 4 weeks ( Figure 6D). Administration of MVC significantly reduced the abnormal neuronal mTORC1 activation and rescued the decreased numbers of autophagosomes and autolysosomes observed in vehicle-treated HD-Tfl mouse brains when compared with controls. No significant changes were observed between vehicle-and MVC-treated control mice ( Figures 6E, 6F, and S6D). Furthermore, the number of aggregates was significantly reduced in the piriform and motor cortices of HD-Tfl mice after 4 weeks of treatment ( Figures 6G  and 6H), paralleling what we observed in the CCR5 KO-HD model. We did not assess behavior in these experiments, as 28 days of treatment (the maximum duration possible with our license with daily intraperitoneal injections) may be too short to have any effects on these parameters, and these HD mice only manifest some mild and initial phases of behavioral deficits at 12 weeks of age (thus limiting the power to detect effects of interventions).
We also validated the effect of MVC in the PS19 tauopathy mouse model ( Figure 7A). Similar to what we observed in HD mouse model, 4 weeks of MVC treatment in PS19 mice rescued the hyperactivation of neuronal mTOR signaling, restored LC3-II levels, and reduced the soluble and sarkosylinsoluble tau levels ( Figures 7B-7D). These MVC-induced improvements were associated with a remarkable rescue of the CA1 neuronal loss and a significant amelioration of cognitive function, as judged by performance on the novel object recognition test ( Figures 7E-7G).
Together, our data suggest that inhibition of CCR5 signaling rebalances mTORC1 activation and autophagy clearance in HD and PS19 mouse models and improves clinical manifestation of these diseases.
CCR5 is a substrate of autophagy The neuronal levels of CCR5 were increased in both HD and PS19 mice (see Figure 4). As Ccr5 mRNA levels were unchanged in the brains of both disease models ( Figures S7A and S7B), we reasoned that CCR5 could be degraded by autophagy, which is impaired in these mice. To test this hypothesis, HeLa cells depleted of the key autophagy gene ATG16L1 (ATG16 KO cells) and their WT counterpart were transiently transfected with a construct carrying the expression of human FLAG-CCR5. FLAG-CCR5 protein levels were increased in ATG16 KO cells compared with WT cells and were lowered in the ATG16 KO cells after ATG16L1 re-expression ( Figure 8A). Similar data were obtained by using ATG5 WT and ATG5 KO HeLa cells (Figure S7C). In line with these results, boosting autophagy via physiological stimuli, such as fetal bovine serum (FBS) or glucose starvation, for 10 h decreased CCR5 levels in ATG16 WT cells compared with basal conditions, but this was not seen in ATG16 KO cells ( Figure S7D). Consistently, the mTORC1 inhibitor Torin, a potent autophagy inducer, also lowered CCR5 protein levels ( Figure S7E).
To investigate whether CCR5 is an autophagy substrate in neurons, we treated primary neurons with SBI-0206965 (SBI impairs autophagosome biogenesis by inhibiting the kinase, ULK1) or BAF (V-ATPase inhibitor) that blocks autophagosome degradation. Decreased LC3-II levels upon SBI treatment and increased LC3-II levels upon BAF treatment validated the effect of these drugs on autophagy in neurons ( Figures S7F and S7G). Accordingly, SBI decreased LC3 vesicle numbers, while these numbers accumulated with BAF ( Figure 8B). Neuronal CCR5 levels (detected by an antibody validated in CCR5 KO neurons;   (B) Representative confocal micrographs (maximum intensity projection) and quantification of P-S6 signal (red) in NeuN + neurons (green) of brains derived from control and transgenic mice injected with vehicle or maraviroc are indicated. Plots represent mean ± SEM (n = 7 mice/group. n = 3 randomly selected fields/ mouse, each containing $400-600 NeuN + neurons). One-tailed, unpaired Student's t test. ***p < 0.01. Scale bar, 50 mm. (C and D) Western blotting and densitometry analyses of LC3-II levels (bottom graph) in (C) and soluble and insoluble tau (adjacent graphs) in (D) in brain lysates from control and/or transgenic mice injected with vehicle or maraviroc. GAPDH was loading control. Plots represent mean ± SEM (n = 9/10 mice per group). Onetailed, unpaired Student's t test. *p < 0.05, **p < 0.01. ***p < 0.01.
(legend continued on next page) ll OPEN ACCESS Article Figure S7H) significantly increased with both SBI and BAF treatment ( Figures 8C and 8D). Altogether, these data suggest that autophagy modulates CCR5 degradation in neurons and support a model where hyperactive CCR5 might upregulate its own expression by blocking autophagy.

CCR5 enters autophagosomes via recycling endosomes
Previous studies show that CCR5 and the transferrin receptor (TfR) co-localize in recycling endosomes 39 and that TfR is an autophagy substrate as it is recruited to the nascent autophagosomes derived from recycling endosomes and thus targeted for degradation. 40 We reasoned that CCR5 might enter the autophagy pathway in a similar way. To follow the trafficking of CCR5 from the plasma membrane to intracellular compartments during starvation, we loaded FLAG-CCR5-transfected HeLa cells with ferrofluid beads conjugated with an anti-CCR5 antibody (Ab) binding the extracellular domain of the CCR5 receptor. After 1 h of internalization in starvation media, the cells were mechanically disrupted, and the ferrofluid-CCR5 Ab-containing membranes were magnetically isolated and then analyzed by western blot. The ferrofluid-CCR5 Ab-containing membranes were enriched in recycling endosome markers, such as TfR and RAB11, and autophagosome markers, such as LC3-II and ATG9. In the unbound membranes, we found endoplasmic reticulum (ER) and early endosomal markers, such as SEC61A and EEA1 ( Figure 8E). Similar results were observed when ferrofluid fractionation was performed on starved cells loaded with beads conjugated with anti-TfR Ab (an approach we previously optimized 40 ), suggesting that both TfR and CCR5 might enter the autophagy route by trafficking to the recycling endosomes. Supporting this hypothesis, the recruitment of CCR5 and LC3 at the recycling endosomes was enhanced during FBS starvation ( Figures 8F, 8G, and S7I-S7K), and an increase in LC3 and CCR5 co-localization was also observed in this condition ( Figures S7K and S7L). Interestingly, by using proximity ligation assays (PLA), we demonstrated that CCR5 colocalizes with LC3, and this was dramatically increased after FBS starvation ( Figure 8H).

CCR5 self-regulates its expression by inhibiting autophagy
Having established that CCR5 is an autophagy substrate, we considered that CCL5:CCR5-mediated autophagy inhibition might affect CCR5 degradation and trafficking (i.e., CCR5 regulates itself by controlling autophagy). To assess this hypothesis, we cultured FLAG-CCR5-transfected HeLa cells in basal conditions or in FBS starvation media in the presence or absence of CCL5. The decrease of CCR5 levels caused by starvation-inducing autophagy was rescued by the treatment with chemokines, suggesting that CCL5 might block CCR5 autophagic turnover. The effect of CCL5 on CCR5 protein levels was abrogated in ATG16 KO cells, demonstrating that CCL5's impact on CCR5 homeostasis depends on autophagy ( Figure 8I). Of note, CCL5 treatment in starved cells did not affect CCR5 trafficking to recycling endosomes but reduced LC3-II in this compartment, as shown by immunofluorescent and ferrofluid approaches ( Figures 8F, 8G, S7I-S7K, and S7M), thus impeding autophagosome formation and CCR5 degradation.

DISCUSSION
In this study, we identified CCL3, CCL4, and CCL5 and their cognate receptor CCR5 as mediators of a new detrimental crosstalk between microglia and neurons, which suppresses neuronal autophagy via mTORC1 activation and impairs the clearance of aggregate-prone proteins. Interestingly, the activation of this axis was observed in the brain of pre-manifesting HD and tau pathology mouse models, suggesting that this neuro-inflammatory mechanism drives neurodegenerative processes at early stages of disease. This highlights the importance and challenge of understanding pathological events in neurodegenerative diseases prior to their clinical manifestation, because such processes can change during the disease course. Indeed, here, we showed that mTORC1 signaling is increased early in the disease course in HD models, while we previously reported that this pathway was attenuated in neurons with large huntingtin aggregates, appearing during the late manifesting phase of the same HD mouse model. This was due to the sequestration of mTORC1 in these aggregates as we demonstrated in cell culture, mouse model, and human samples. 30 Thus, pathological changes seen in late-stage disease cells may not reflect key events leading to the genesis of the condition. Importantly, genetic depletion or pharmacological inhibition of CCR5 by MVC reversed the defective autophagy-mediated degradation of neurotoxic proteins and the dysfunctional behavior associated with HD and tau pathology mouse models-a result that could have substantial clinical implications.
Our studies mainly focused on the harmful effects of elevated CCR5 signaling in neurons. However, as CCR5 is also expressed in other cell types (like macrophages, microglia, astrocytes), we cannot exclude that the beneficial effects associated with its inhibition might additionally be due to the suppression of its signaling outside the neuronal network.
The lack of available brain biopsies from patients affected by neurodegenerative disorders at pre-clinical stages hinders the investigation of CCLs-CCR5 signaling in human diseases. Previous studies have suggested a negative role of CCR5 in modulating memory circuit and synaptic plasticity, and association studies between the CCR5D32 polymorphism (encoding a non-functional receptor) and the development of neurodegenerative dementias have been performed. [41][42][43][44][45] No significant association was observed in these studies; however, the number of homozygote carriers analyzed in control and diseases cohorts was very small due to the low frequency of the null allele, and (E) Unbiased stereological estimates of numbers of NeuN-positive CA1 neurons in hippocampus of Ccr5 +/+ and maraviroc-and vehicle-treated PS19:Ccr5 +/+ mice were performed using optical dissector method. Plots represent mean ± SEM. Two-tailed unpaired t test; p values are shown. (F and G) Cross-sectional (F) and longitudinal (G) analysis of scores of WT, PS19-vehicle, and PS19-maraviroc-treated mice on NORT task. The graphs show novelty preference (%) in NORT. One-tailed, unpaired Student's t test in (F) and one-tailed, paired Student's t test in (G). P values are shown. In the second part of our study, we showed that CCR5 is degraded by autophagy and, similarly to TfR, enters the autophagy pathway by trafficking toward the recycling endosomes. 40 CCR5 activation by chemokines does not affect receptor trafficking but inhibits autophagosome formation at the recycling endosomes. This prevents the degradation of CCR5 by autophagy, causing its accumulation ( Figure S8). In other words, CCR5 binding to chemokines activates a self-regulatory loop in cells expressing mHTT or tau, resulting in the upregulation of CCR5 protein levels. Notably, our experiments in the Tet-Off tauopathy model suggest that removal of the mutant protein is sufficient to break this self-amplifying loop. This mechanism can explain the increased CCR5 neuronal expression observed in our neurodegenerative mouse models.
Together, our results provide insights into the roles of activated microglia in non-cell-autonomous downregulation of neuronal autophagy. This detrimental microglia-neuron crosstalk is activated early in neurodegeneration and is potentially self-sustaining but may be amenable to therapeutic interventions.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

RESOURCE AVAILABILITY
Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, David C. Rubinsztein (dcr1000@cam.ac.uk).

Materials availability
This study did not generate new unique reagents.
Data and code availability This paper analyses existing, publicly available data. These accession numbers for the datasets are listed in the key resources table. This paper does not report original code. Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request.

General considerations
Mouse studies were performed in accordance with the UK Animals (Scientific Procedures) Act with appropriate Home Office Project and Personal animal licenses and with local Ethics Committee approval. Mice were housed in individually ventilated cages with free access to standard animal food chow and water, in a climate-controlled room with a 12 h light/dark cycle.

Mice
In this study, we used three different neurodegenerative disease mouse models (HD-N171-82Q, tau PS19 and rTg4510 mice), mRFP-GFP-LC3 autophagy reporter line, here named traffic light (Tfl) mouse line and Ccr5 À/À mouse model. HD-N171-82Q mice, here referred to as HD mice (B6C3F1/J-Tg(HD82Gln)81 Dbo/J, Jackson Laboratory, Bar Harbor, ME, USA), are heterozygous mice carrying an N-terminal fragment of the first 171 amino acids of human huntingtin with 82 glutamine repeats under the mouse prion protein promoter. 55 This line has been backcrossed on a C57BL/6J background for more than 10 generations. Both male and female mice aged 6-8 weeks did not exhibit aggregates and behavioral abnormalities, therefore they were used for pre-manifesting studies. Previous work has shown that both male and female mice develop aggregates in their brains detectable by the age of 12 weeks, whereas male mice display more robust behavioral deficits. 56,57 Therefore, male mice were used for behavioral analysis and both genders were used for histological analysis.
Tau (PS19) mice, here referred to as PS19 mice (B6N.Cg-Tg (Prnp-MAPT*P301S) PS19Vle/J, Jackson Mouse Stock No 024841) are heterozygous mice expressing the T34 isoform of mutant human microtubule-associated protein tau (MAPT) driven by the mouse prion protein promoter. The transgene encodes the disease-associated P301S mutation and includes four microtubule-binding domains and one N-terminal insert (4R/1N). 58 These mice have been extensively used for studying neurofibrillary tangles, neurodegenerative tauopathy and AD. Both male and female mice aged 22-23 weeks did not exhibit aggregates and behavioral abnormalities therefore they were used for pre-manifesting studies. rTg4510 mice express a repressible form of human tau containing the P301L mutation that has been linked with familial frontotemporal dementia. The generation of rTg4510 mice was described previously. 35 Both congenic parental mouse strains, tTA activator, known as Camk2a-tTA (B6.Cg-Tg (Camk2a-tTA)1Mmay/DboJ, Jax stock no. 007004) and htau responder, also known as Tg (tauP301L)4510 (FVB-Fgf14Tg (tetO-MAPT*P301L)4510Kha/JlwsJ, Jax stock no.015815) were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). The rTg4510 mice were bred from these two different parental lines (responder has to be co-expressed with the activator) in order to activate the expression of transgene under the control of a tetracycline conditional gene expression system (tet-off, tTA). mRFP-GFP-LC3 (Traffic light (Tfl) mice were generated, as previously described. 49 mRFP-GFP-LC3 was subcloned by PCR from pmRFP-EGFP-rLC3 (gift from T. Yoshimori) into pCAGGS (gift from J. de Belleroche) which drives ubiquitous expression of the transgene under the control of a chicken beta-actin promoter and cytomegalovirus enhancer. The vector was digested with HindIII and SpeI to remove the DNA backbone and the fragment purified by gel extraction and used for microinjection into hybrid B6CBA oocytes. Founders were identified by PCR from ear biopsies and crossed to C57BL/6 mice to obtain F1s. mRFP-GFP-LC3 protein expression level was assessed by western blotting in brain, muscle and tissue and by cryo-sectioning of fresh frozen tissue and direct observations of fluorescence levels. Line 1 was selected for further study due to its good levels of protein expression and clear fluorescence in brain. This line has been backcrossed on a C57BL/6J background for more than 10 generations.
In CCR5 KO mice, also known as Ccr5 À/À mice, (B6; 129P2-Ccr5tm1Kuz/J, Jackson Mouse Stock No. 002782), the exon containing the entire coding region of Ccr5 was replaced with a neomycin resistance gene inserted by homologous recombination. 59 This line has been backcrossed on a C57BL/6J background for more than 10 generations.
To study the effect of CCR5 depletion in neurodegenerative disease, CCR5 KO mice were crossed with either HD or PS19 mouse models. To measure autophagy upon maraviroc treatment in vivo, the Tfl mouse line was crossed with the HD disease mouse line.
The following signs were used as human endpoints for the neurodegenerative disease mice, which resulted in euthanasia: marked loss of appetite and fluid intake, staring coat, hunched posture, tremor, subdued behavior, or 15% weight loss over a period of less than 3 days. The litters produced by mating of transgenic mice were genotyped at an age after 3 weeks by PCR using ear punches of individual animals, according to the protocols recommended by the Jackson Laboratory and outsourced to Transnetyx. For mRFP-GFP-LC3 mice, mRFP probe, Forward Primer: AGCGCGTGATGAACTTCGA, Reverse Primer: GCGCAGCTTCACCTTGTAGAT, was used by Transnetyx.

Cell lines
Humancervical cancer (HeLa) and immortalised murine microglial (BV2 -kindly provided by Prof. Guy Brown) cell lines were cultured in Dulbecco's modified Eagle's medium (DMEM) (Sigma-Aldrich, D6546) supplemented with 2mM L-glutamine, 100U/ml Penicillin/ Streptomycin and 10% Fetal Bovine Serum. HeLa cells stably transfected with CCR5-GFP construct were kindly provided by Prof. Olivier Schwartz and cultured in DMEM Glutamax (Gibco, 31966021) supplemented with Penicillin/Streptomycin and FBS as described above. 46 G418 (Gibco; 0.5 mg/ml) and Hygromycin (Calbiochem, 50 mg/ml) selection markers were added to the cell culture media for one week after defrosting HeLa CCR5-GFP cells. HeLa CRISPR/Cas9 ATG16L1 KO cell lines were generated as previously described and cultured as normal HeLa cells. 47 HeLa CRISPR/Cas9 ATG5 were kindly provided by R. J. Youle's laboratory. 48 HeLa and BV2 cells are derived from an adult female human and a neonatal female mouse, respectively. All cell lines were maintained at 37 C and 5% CO 2 and were regularly tested for mycoplasma contamination.

Mouse primary neurons
Primary cortical neurons were isolated from male and female C57BL/6 J (Jackson Laboratories) or CCR5 WT and CCR5 KO, or mRFP-GFP-LC3 mouse embryos at E16.5. Briefly, brains were harvested and placed in ice-cold HBSS (Sigma-Aldrich), where the meninges were removed, and the cerebral cortices were dissected. Cortices were digested in 0.25% trypsin (Gibco), followed by FBS inactivation and three washes in HBSS. After further mechanical dissociation, neurons were seeded on poly-D-lysine-(Sigma-Aldrich, P6407) coated plates. Cells were cultured in Neurobasal-A Medium Minus Phenol Red (Gibco, 12349-015) supplemented with 1X B-27 Serum-Free Supplement (50X) (Gibco, 17504-044), 2mM GlutaMAX(Gibco), sodium pyruvate (Gibco), and 1% penicillin-streptomycin at 37 C in a humidified incubator with 5% CO 2 . One-half of the culture medium was changed every 2 days. After 4-6 days of culturing in vitro, differentiated cortical neurons were either co-cultured with BV2-containing transwells or treated with the indicated compounds/conditioned media or infected with lentiviral particles.

Mouse primary microglia
Primary mixed glial cultures were obtained from 1 to 3 days old, male and female C57BL/6 mouse pups. Briefly, brains were harvested and placed in ice-cold HBSS, where the meninges were removed, and the cerebral cortices were dissected. Cells were mechanically dissociated and plated in 75 cm flasks in DMEM media (Sigma-Aldrich, D6546) supplemented with 10% FBS, 2mM GlutaMAX and 1% penicillin-streptomycin at 37 C in a humidified incubator with 5% CO 2 . Media was changed after 1 day and then every 7 days. On day 11-14, microglia cells were visible on the astrocyte layer and were dissociated by gentle manual shaking and finally seeded on poly-D-lysine coated 12-multiwell plates.
BV2 and primary neurons co-culture BV2 cells were seeded on Nunc polycarbonate cell culture inserts in multiwell plates (Thermo Fisher, 140640). On the next day, BV2 media was replaced with neurobasal medium supplemented with B27, GlutaMAX, sodium pyruvate, and penicillin-streptomycin. Transwell inserts with BV2 were transferred to the plate containing primary neurons at day 6 in vitro. BV2 were treated with LPS (10 ng/mL) +INFg (100 ng/mL) for 24 h where indicated. Neurons co-cultured with empty transwell inserts containing neurobasal media or LPS+INFg were used as controls. During the last 8 h of treatment, 400 nM of BafilomycinA1 (BAF) or DMSO was added to wells containing neurons.

METHOD DETAILS
Grip strength, wire maneuver and tremor Double transgenic male HD mice of genotype HD:Ccr5 À/À (n = 28) and HD:Ccr5 +/+ (n = 23) were tested for grip strength, wire maneuver and tremor. These tests were performed at 7, 9, 11, 13, 15 and 17 weeks of age. Wire maneuver and tremor are part of the SHIRPA battery of behavioral tests. 60 Grip strength was monitored quantitatively using a grip strength meter (Biosep, France), as previously described. 56 In the wire maneuver test, mice are tested for their ability to climb back using their hindlimbs on a horizontal wire when hung on the wire by their forelimbs. The scores were set as: 0, active grip with hind legs; 1, difficulty grasping with hind legs; 2, unable to lift hind legs; 3, falls immediately. While performing the behavioral tests, the experimenter was blinded to the genetic and treatment status of the mice.

Novel object recognition (NORT)
Novel object recognition test (NORT) is commonly used to assess changes in short-term memory tasks and was performed as previously described. 61 A cohort of double transgenic mice of genotype PS19:Ccr5 À/À (n = 19 mice: 13 females and 6 males) and PS19:Ccr5 +/+ mice (n = 21 mice: 10 females and 11 males) and a cohort of vehicle-(n = 9 mice per group: 3 males and 6 females) and maraviroc-(n = 8 mice per group: 4 males and 4 females) treated PS19:Ccr5 +/+ mice, and vehicle-treated Ccr5 +/+ mice (n = 10 mice per group: 4 males and 6 females) were tested for NORT at 10 months of age. The mice were acclimatised in the behavior room at least 30 min before the beginning of the test. The room was illuminated by sombre red light at 70-75 lux. A day prior to exposure to the object, the mice were habituated to the open field arena for 30 min. The next day, training and testing were performed. During the training phase mice are exposed to two identical objects (also called familiar objects) for 15 min. During the testing phase, mice are exposed to one familiar object and one novel object for 10 min. Training and test phases were separated by 3.5-4 h. Mice were removed from the arena and placed in their holding cage during the interval between the two phases. Both familiar (object A) and novel objects (object B) were made of same plastic material with similar sizes (height, about 5 cm), slightly different in texture (with and without ridges i.e. smooth surface) and have distinctive shapes (one is rectangular and other is square). The objects were positioned on the diagonal (i.e. one in the NW corner and one in the second SE corner) (10 cm from the walls) counterbalanced to avoid preference biases toward location. Each object was used equally as a familiar object and as a novel object to avoid preference biases toward the object. We made sure that the diagonal used during the training was the same as that used on the testing for each mouse. The arena and objects were cleaned between each trial with 70% alcohol to mask any olfactory cues. Activities were recorded using overhead cameras connected with Panlab Record-IT Media Software. Videos were analyzed manually using Behavioral Observation Research Interactive Software (BORIS). 62 Exploration was scored positive if mouse's nose was within a distance of 1 cm to the object and/or touching it with the nose, while sitting and climbing or rearing on the object was not considered exploration. Total exploration time of the familiar and novel objects was used to calculate novelty preference and discrimination index, using the equations: Novelty Preference = time spent on novel object total time exploring both objects ðnovel + familiarÞ 3 100 Discrimination Index = time spent on novel object À time spent on familiar object total time exploring both objects ðnovel + familiarÞ While performing the behavioral tests, the experimenter was blinded to the genetic status or treatment group of the mice. Statistical analysis on percent time spent on novel object was performed using two-tailed unpaired Student's t test.

Maraviroc treatment
In order to study the effect of pharmacological inhibition of CCR5 on autophagy clearance in neurodegenerative disease, a mixed cohort of HD:Ccr5 +/+ single transgenic (n = 2 females mice per group), HD:Ccr5 +/+ :Tfl À/+ double transgenic mice (n = 6 mice per group: 4 males and 2 females) and control mice (n = 4 mice per group: 2 males and 2 females) aged 8 weeks were administered with maraviroc (UK-427857) (Selleckchem, catalog no.S2003) -50 mg/kg via intraperitoneal injections (i.p.) daily for 28 days. An equal number of single and double transgenic and control mice of same sex were daily injected with vehicle. The same treatment was administered to PS19:Ccr5 +/+ (n = 10 mice per vehicle group: 6 females and 4 males) -(n = 10 mice per maraviroc group: 5 males and 5 females) aged 39 weeks. Mice in both studies were assigned to treatment group in a factorial randomised design. For instance, mice receiving the maraviroc treatment were in the same cages with mice receiving the vehicle treatment and WT littermates. Maraviroc was dissolved in DMSO to make the stock solution (100 mg/mL) and stored at À20 C. Before use, the stock solution was diluted with 40% PEG300, 2% Tween-80 and saline to 20 mg/mL. Control mice were administered with vehicle supplemented with 5% DMSO.

Antibodies
Cell treatment CCL5 (mouse 478-MR-025, human 278-RN-050 R&D systems), CCL4 (mouse 451-MB-050 R&D systems) and CCL3 (mouse 450-MA-050 R&D systems) were added to cell culture at 50nM for the indicated time points. Treatment with CCLs was performed in media supplemented with low concentration of FBS (0.5%) and was replaced every 12 h when the experiment lasted 24 h.
Conditioned media collection BV2 cells or mouse primary microglia cells were plated at a confluency of $30%. The next day, microglial media was replaced with neurobasal medium supplemented with B27, GlutaMAX, sodium pyruvate, and penicillin-streptomycin. Cells were treated with LPS (10 ng/mL) + INFg (100 ng/mL) or left untreated. After 24 h, microglial conditioned media (CM) was collected. At the point of CM collection microglia were $50-60% confluent. CM was centrifuged at 1000 rpm for 5 min to remove any cell debris. If not used immediately, CM was stored at À80 C.

Immunodepletion
The concentration of chemokines in the conditioned media was determined by ELISA. Three times the equivalent molar concentrations of biotinylated rabbit anti-murine CCL3, CCL4 and CCL5 antibodies (500-P121; 500-213; 500-118, Peprotech) was added to the conditioned media overnight at 4 C. The biotinylated antibodies complexed with chemokines were removed from the media using dynabeads-streptavidin (Dynabeads MÀ280 streptavidin, Life Technologies) followed by magnetic separation. As control, media was treated with equivalent concentrations of biotinylated-IgG.
Transfection trans IT-2020 reagent (Mirus) was used for DNA transfection, while Lipofectamine 2000 (Invitrogen) was used for siRNA transfections, according to the manufacturer's instructions. Transfection was performed in 6 well plates. pHM6-httQ74-HA construct was transfected at 0.3mg per well, while the other constructs were transfected at 1mg per well. Cells were transfected for 24-48 h, as indicated in the figure legends. For silencing experiments, cells were transfected with either a single or double round of 50nM siRNA (Dharmacon; SMARTpool siRNA). In single transfections (RAPTOR and TSC2 knockdown), cells were split 24 h after transfection, and collected three days post-transfection. For double transfections (CCR5 and ATG7+ATG10 knockdown), cells were transfected on day one and day three. Cells were split the day after each transfection and collected five days after the first transfection.

Lentivirus production & infection of primary neurons with Q80-EGFP
The lentiviral plasmid carrying the expression of Q80-EGFP fusion protein under the control of the neuron-specific synapsin promoter was kindly provided by J.B. Uney and J.L. Howarth. 53 Lentiviral particles were produced and transduced following the RNAi Consortium protocols. 66 Briefly, HEK293T packaging cells growing in 10-cm dishes were transfected with a mix of 2 mg packaging vector (psPAX2), 270 ng envelope vector (pMD2.G) and 2.7 mg of lentiviral expression vector. Lipofectamine LTX Reagent with PLUS Reagent (Thermofisher, A12621) was used as transfection reagent. After transfection, cells were cultured in high-serum medium. Cell culture medium was collected 48 h later and replaced by high-serum medium; this step was repeated 2-3 times at intervals of 24 h. Virus preparations were then concentrated by centrifugation at 160,000 g for 90 min. Viral particles were added to primary cultured neurons (DIV 5) and incubated overnight. Chemokines were added 3 days after viral infection and left for a further 24 h. Neurons were fixed in a 2% PFA plus 7.5% glucose solution and coverslips were mounted with ProLong Gold antifade reagent with DAPI (Thermo Fisher; Invitrogen P36935).

Western Blot analysis
Upon perfusion with PBS, mouse brains were harvested and split transversally: one-half was fixed and processed for immunostaining. The other half was flash frozen, homogenized in 1mL RIPA buffer [50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholatemonohydrate, 0.1% SDS], supplemented with protease (Roche) and phosphatase inhibitors cocktails (Sigma Aldrich) sonicated, and centrifuged two times at 12,000 r.p.m. for 10 min at 4 C. Total protein concentrations were assayed in each sample with the BCA-assay, according to the manufacturer's protocol (Thermofisher, 23225). The samples were dissolved in 2x Laemmli buffer with b-mercaptoethanol buffer and loaded on a gel (20 mg per lane). HeLa cells and primary neurons were directly lysed in Laemmli buffer supplemented with b-mercaptoethanol buffer. All samples protein were boiled for 5-7 min at 100 C, separated by SDS-PAGE and transferred onto PVDF membranes. The membranes were blocked with 5% non-fat milk or 3% BSA in PBS, incubated with primary antibodies overnight, followed by HRP-conjugated (GE Healthcare) or DyLight Fluors-conjugated (Invitrogen) secondary antibodies. Immunoreactive bands were visualized with an ECL enhanced chemiluminescence detection kit (GE Healthcare) using the Chemidoc Imaging System (Bio-Rad). Alternatively, the membranes were labeled with fluorescent secondary antibodies and analyzed with the LICOR-Odyssey apparatus. For re-probing, the membranes were rinsed and incubated for 30 min at 55 C in a stripping buffer (62.5 mmol/L Tris-HCl, 2% SDS, 100 mM mercaptoethanol, adjusted to pH 7.4), before incubation with new primary antibodies. In some experiments, after protein transfer, the PVDF membrane was cut into fragments, to allow for incubation with different primary antibodies. Densitometry analysis on the immunoblots was performed by using Fiji (ImageJ) or IMAGE STUDIO Lite software.
Western blots performed on cell culture samples are representative of at least three independent biological replicates. In this case, every experiment is set up with its own control, which is used to normalise the result of the corresponding biological replicate. The statistics were performed on these normalized experiments, which allowed us to monitor whether the variation observed between the conditions tested is significant.
Western blots performed on tissue samples include the complete cohort of mice analyzed in each experiment except in Figures 5C  and 5D, 7C, 7D, and S5C, where the number of mice analyzed did not fit on one single gel. In these cases, one representative western blot is shown, and the quantification of LC3-II and tau was performed as follows. LC3-II or tau bands from immunoblots processed in parallel by using the same reagents and buffers were quantified. Values were normalized on a reference sample present on each immunoblot to allow the comparison between the bands detected on the different membranes. The statistics were performed on these normalized experiments.
For the detection of endogenous CCR5, FLAG-CCR5 or CCR5-GFP, cells or tissues were collected in TBS buffer (20mM Tris-HCl pH 7.4, 150mM NaCl, 1 mM EDTA, 1mM EGTA and protease inhibitors cocktail) best-known for its structure-stabilizing ability and recommended for the extraction of transmembrane proteins such as GPCRs. 67 Tissues were homogenized using Precellys CK14 Lysing Kit (10144-554, Avantor), while cells were mechanically triturated 10 times using a 30G needle. Protein samples were then mixed with Laemmli buffer and loaded into the gel without boiling and processed for immunoblot using the anti-CCR5 Ab. In mouse tissues and derived cells, a specific band was detected approximately at 60 kDa, whereas in HeLa cells transfected with FLAG-CCR5 a specific band was found approximately at 40 kDa. The specificity of the band was validated using CCR5 KO tissues and CCR5 silencing. As previously described, CCR5 can appear at different molecular sizes, depending on the tissue where it is expressed. This might depend on differences in mRNA splicing, products of differential transcriptional regulation, or the result of post-translational modifications. 68 Tau sarkosyl extraction Soluble and insoluble tau fractionation was performed using the sarkosyl extraction protocol 35 with minor modifications. Frozen mouse brains were homogenized in TBS (20 mM Tris pH 7.4, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA supplemented with protease and phosphatase inhibitors) and disrupted by using Precellys CK14 Lysing Kit (10144-554, Avantor). Sample homogenates were ultracentrifuged at 150 000 g for 15 min at 4 C. Supernatants were collected (soluble tau fractions) and used for proteins quantification. The pellets were re-homogenized in 150 mL high salt/sucrose buffer (10 mM Tris pH7.4, 0.8 M NaCl, 10% Sucrose, 1 mM EGTA, 1 mM PMSF) and ultracentifuged, as before. The supernatants were incubated with a final volume of 1% sarkosyl for 1 h at 37 C before ultracentrifugation at 150 000 g for 1 h at 4 C. The pellets (sarkosyl-insoluble tau) were resuspended directly in 20 mL of Laemmli buffer and boiled at 100 C for 5 min. Samples were run on 10% acrylamide SDS-PAGE gels. Blots were blocked in 5% non-fat milk in PBS-T and incubated with primary antibodies.

Ferrofluid
TfR-and CCR5-enriched membrane isolation using Magnetic Microbeads (Ferrofluid EMG508, Ferrotec, USA) was performed as previous described 69 with some modifications. Cell were serum starved for 30 min, incubated with serum-free media containing Fe-TfR Ab or Fe-CCR5 Ab magnetic complex for 1 h and chased in serum free media for other 30 min. Cells were chilled on ice, rinsed with 1 mL of cold PBS, scraped in 700mL of homogenization buffer (10 mm HEPES, pH 7.2, 100 mm KCl, 1 mm EDTA, 25 mm sucrose) and disrupted by adding glass beads to the samples and vortexing at maximum speed for at least 1 min. Then, the homogenate was centrifuged at 13 000 RPM for 10 min, and the supernatant (Fe-unbound fraction) was separated from the pellet (Fe-bound fraction). The pellet was resuspended in the homogenization buffer and washed 3 times using a magnetic rack. The samples were then removed from the magnetic field and eluted in sample buffer.

ELISA
Concentrations of CCL3, CCL4 and CCL5 in conditioned media and in mice brain were determined by using commercially available murine CCL3, CCL4 and CCL5 sandwich ELISAs (Peprotech or BIOMATIK), according to manufacturer instructions. For the analysis in mouse tissue, CCL-3/-4/-5 concentrations were normalized to protein concentrations. Given the high dispersion of data values, each dataset was tested for outlier identification by applying the ROUT method (Q = 1%) via PRISM. The outliers were indicated in the graphs as black dots and were excluded by the statistical analysis.
RNA extraction and qPCR from tissue Total RNA was extracted from mouse brain using RNeasy Lipid Tissue Mini Kit (74804, QIAGEN). DNAse I treatment was performed to eliminate genomic DNA contamination. One microgram of RNA was used to perform the reverse transcriptase reaction with Super-Script III First Strand Synthesis System for RT-PCR (1880-05, Invitrogen). The synthesized cDNA was mixed with primers and SYBR Green PCR Master Mix (4309155, Applied Biosystems) and processed by real time qPCR by using CFX96 Real-Time PCR Detection System (Bio-Rad). For the analysis of Ccr5, we used: forward primer 5 0 TGCTGCCTAAACCCTGTCAT -3 0 and reverse primer 5 0 CGATCAGGATTGTCTTGCTGGA -3 0 as previously described. 19 For the analysis of Gapdh we used: forward primer 5 0 TGCACCACCAACTGCTTAGC -3 0 and reverse primer 5 0 GGCATGGACTGTGGTCATGAG -3 0 as shown before. 64 The relative changes in Ccr5 over Gapdh mRNAs were calculated using the 2 ÀDDCt formula.

LDH-release assay
Mouse primary neurons, cultured at the same density, were incubated with BAF 400nM or DMSO for 8 h and LDH-release assay was performed and quantified by using the LDH Assay Kit (Abcam ab65393), according to manufacturer instructions. Briefly, cells were precipitated by centrifugation at 600g for 30 min, cell medium was transferred to an optically clear 96-well plate and mixed with the reaction buffer for 30 min. Absorbance at wavelength 450 nm was measured by Tecan plate reader.

Analysis of brains from mouse experiments
For immunohistochemistry and fluorescent imaging, mice were anesthetized with i.p. injections of pentobarbital sodium (euthatal) and perfused with ice-cold PBS. Mouse brains were fixed in 4% PFA (or 10% Formaline) overnight at 4 C, rinsed in PBS, and incubated 2-3 days at 4 C in 30% sucrose before being snap-frozen in cryogenic Tissue-Tek OCT compound (Electron Microscopy Sciences, Hatfield, USA) and stored in À80 C.

Quantification of mRFP-GFP-LC3 vesicles in mouse brains
The analysis and quantification of autophagosomes and autolysosomes in the brains of HD:Tfl mice were performed on one hemisphere of the brain. 8mm thick sagittal sections were cut from perfused, cryo-protected mouse brain hemispheres using a cryostat. Sections were mounted on poly-lysine coated slides (cat no. 631-0107, VWR). Autofluorescence was quenched by staining the sections with 0.05% Sudan Black in 70% methanol for 5-10 min 70 Sections were mounted using ProLong Diamond Antifade Mountant with DAPI (cat no. P36962, Thermo Fisher) and imaged by an LSM710 Confocal Microscope (Zeiss) using a 633 objective. Quantitative image analysis for cerebral cortex was performed by selecting 5 random visual fields acquired with the same setting parameters (except for minor changes on Digital Offset, when needed). The number of mRFP-GFP-LC3 dots (autophagosomes, yellow) and mRFP-LC3 dots (autolysosomes, red) were counted manually. We only included in the quantification the yellow vesicles that also appeared red when we switched between the red and green channels. The number of vesicles were quantified per frame in the cortex. The experimenter was blind to the treatments or slide names or numbers. Mean values of each vesicle type (autophagosomes, autolysosomes and total vesicles) of maraviroc-treated mice were compared to the mean value of the same vesicle type of vehicletreated mice (control). Statistical analysis was performed by comparing autolysosome numbers or autophagosome numbers or total vesicle numbers for treatment vs. control.
Quantification of neuronal HTT aggregates in mouse brains 30 mm thick coronal sections derived from one brain hemisphere of HD:Ccr5 +/+ and HD:Ccr5 À/À transgenic mice or HD:Ccr5 +/+ : Tfl À/+ double transgenic mice treated with vehicle or maraviroc were analyzed for neuronal aggregates as described before. 71 Sections were labeled with anti-huntingtin antibody (clone mEM48, Millipore, MAB5374) by free-floating immunohistochemistry for 3 days. Staining was performed by peroxidase labeling using Vectastain Elite ABC reagent kit (PK6100, Vector Laboratories) and visualized with DAB reagent (SK-4100, Vector Laboratories). Mutant huntingtin inclusions were counted blind in the piriform cortex and motor cortex in 3-5 fields on five sections per animal at a magnification of 3100 (Axioimager Z2 Upright Microscope). Statistical analysis was performed on the mean percentage of cells with aggregates.