|Home | About | Journals | Submit | Contact Us | Français|
In Huntington's disease (HD), mutated huntingtin (mhtt) causes striatal neurodegeneration which is paralleled by elevated microglia cell numbers. In vitro cortico-striatal slice and primary neuronal culture models, in which neuronal expression of mhtt fragments drives HD-like neurotoxicity, were employed to examine wild type microglia during both the initiation and progression of neuronal pathology. As neuronal pathology progressed, microglia initially localized in the vicinity of neurons expressing mhtt fragments increased in number, demonstrated morphological evidence of activation, and expressed the proliferation marker, Ki67. These microglia were positioned along irregular neurites, but did not localize with mhtt inclusions nor exacerbate mhtt fragment-induced neurotoxicity. Prior to neuronal pathology, microglia upregulated Iba1, signaling a functional shift. With neurodegeneration, interleukin-6 and complement component 1q were increased. The results suggest a stimulatory, proliferative signal for microglia present at the onset of mhtt fragment-induced neurodegeneration. Thus, microglia effect a localized inflammatory response to neuronal mhtt expression that may serve to direct microglial removal of dysfunctional neurites or aberrant synapses, as is required for reparative actions in vivo.
Huntington's disease (HD) is an autosomal dominant neurodegenerative disorder involving a gradual loss of medium spiny neurons, the most prevalent cell type in the striatum [Martin and Gusella 1986]. Expansion of a translated trinucleotide (CAG) repeat in the HD gene encoding huntingtin (htt) leads to a polyglutamine expansion at the amino terminus of the resultant 348kD htt protein [HD Collaborative Research Group 1993]. Although loss of normal htt function may contribute to neurodegeneration [Cattaneo, et al. 2005], a growing understanding of HD pathogenesis suggests a gain-of-function neuronal toxicity of mutated htt (mhtt) that involves transcriptional dysregulation, mitochondrial dysfunction, and impaired synaptic transmission [Landles and Bates 2004, Panov, et al. 2002, Ross 2002]. This toxicity appears to be driven by aggregated N-terminal fragments of mhtt, rather than full-length mhtt [Cooper, et al. 1998, DiFiglia, et al. 1997, Wang, et al. 2008].
Normal htt protein contributes to vesicular transport and synaptic transmission and, as such, is highly expressed in dendrites and nerve terminals [Gutekunst, et al. 1995, Trottier, et al. 1995]. In HD, the resistance of mhtt to proteolysis and its tendency to misfold precipitate the formation of inclusion bodies (IBs) in the nucleus, cytoplasm, and neurites [Gutekunst, et al. 1999]. HD patients exhibit an early accumulation of N-terminal fragments of mhtt in non-nuclear regions of the cell [Sapp, et al. 1999] and, throughout disease, IBs in dendrites and axons (neuropil aggregates) appear more frequently than nuclear aggregates [Gutekunst, et al. 1999, Li, et al. 1999]. In support of a primary dysfunction in axons and dendrites, early stages of HD are characterized by dystrophic neurites with fewer dendritic spines and thickened, proximal dendrites [Albin, et al. 1990, Li, et al. 2001]. The progressive appearance of small neuropil aggregates correlates with a disruption in trafficking and synaptic function, as well as mitochondrial damage, microtubule destabilization, neurite retraction, and the eventual development of neurological symptoms [Li, et al. 1999, Li, et al. 2001, Smith, et al. 2005, Trushina, et al. 2004, Trushina, et al. 2003]. However, the actual contribution of aggregate formation to neuronal loss in HD has yet to be identified [Kuemmerle, et al. 1999, Saudou, et al. 1998].
Microglia are necessary for healthy brain function, as these cells clear tissue debris, remove soluble factors and aberrant proteins from the neuronal microenvironment, and respond to brain insults including neurodegenerative disease progression [Gehrmann, et al. 1995, Long-Smith, et al. 2009, Meda, et al. 1995]. A correlation between structural changes in microglia and severity of HD neuropathology has been reported in patients [Pavese, et al. 2006, Sapp, et al. 2001, Singhrao, et al. 1999], suggesting a role in disease progression. Microglial responses in the striatum are characterized by an increased activation state in pre-manifest [Tai, et al. 2007] and manifest HD [Sapp, et al. 2001], as well as the R6/2 mouse model for HD [Tai, et al. 2007].
Although microglia are intimately linked with HD neuropathology, the exact role that they play in regulating the health of mhtt-expressing neurons remains unclear. The functional interaction between diseased neurons in HD and immuno-modulatory microglia appears to represent a critical juncture in the progression and amplification of HD pathology that, if understood, could support the development of anti-inflammatory based patient treatment options. The correlation between increasing microglia cell numbers and the progression of HD pathology suggests that microglia exacerbate the pathology of diseased neurons. In the current study, we evaluated the relationship between neighboring microglia and mhtt-expressing neurons in primary cell and brain slice culture models that possess cortico-striatal neuronal connections shown to be involved in the development of the HD phenotype [Cepeda, et al. 2007]. In these model systems, the timecourse of neurite disruption and neuronal death allowed for the examination of microglial activation state and localization at progressive stages of mhtt-induced neuronal degeneration.
Sprague Dawley rats and CD-1 mice (Charles River Labs, Raleigh, NC) were obtained at postnatal day 1-2 or 10 (PND1-2 or 10) and gestational day 15 or 18 (GD15 or 18). Animals were housed in a semi-barrier facility at 21±2°C with 50±5% humidity and 12h light/dark cycle. Food and deionized, reverse osmotic-treated water were available ad libitum. All procedures were conducted in compliance with approved animal protocols from NIEHS/NIH or Duke University Institutional Animal Care and Use Committees.
Cortico-striatal neuronal cultures were prepared from GD18 rats as previously described [Kaltenbach, et al. 2010]. Briefly, dissociated striatal neurons were transfected by electroporation (Lonza, Basel Switzerland) with a fluorescent protein tag (either mCherry or yellow fluorescent protein, YFP), combined with either a control plasmid or a human htt exon-1 (N-terminus) expression construct containing a 73-polyglutamine-encoding repeat (HttN90Q73; mHtt). Cortical neurons were co-transfected with a different fluorescent tag (either YFP or cyan fluorescent protein- CFP) combined with control or mHtt plasmid. All expression plasmids were controlled by the cytomegalovirus promoter. Following electroporation, striatal and cortical neurons were mixed and plated into 96-well plates (60,000 cells per well) on a confluent bed of rat-derived astrocytes (> 97% pure). Cultures were grown in Neurobasal media supplemented with 5% fetal bovine serum (FBS; <0.3EU/mL endotoxin), 2mM Glutamax, KCL, and 5μg/mL gentamycin (all reagents from Invitrogen, Carlsbad, CA). According to preliminary evaluation of NeuN staining, the neuronal plasmid transfection efficiency is approximately 25%.
Primary cortical neurons were prepared from GD15 mice as previously described [Kraft, et al. 2004]. Cortical cells were filtered (70μm cell strainer; Millipore, Billerica MA) and plated on poly-D-lysine (PDL)-coated plates at 3.0 × 105 cells/cm2 in MEM supplemented with 10% FBS, 10% horse serum, 2mM Glutamax, and 1% penicillin/streptomycin (P/S) for 48h prior to changing to Neurobasal media with B27 supplement containing antioxidants (all reagents from Invitrogen). Dissociated cells were maintained at 37°C under 5% CO2. At 5 days in vitro (DIV), cells were biolistically co-transfected with YFP and CFP (control plasmids) or YFP and HttN90Q73 possessing a CFP tag at the C-terminus (mHtt plasmids). Using particle-mediated gene transfer as previously described [Lo, et al. 1994], gold particles (1.6μm) coated with 12μg of each plasmid were propelled at 95psi into the cultures by a Helios gene gun (Bio-Rad, Hercules, CA) at a height of 2.5cm in the absence of media (conditioned media was replaced immediately afterwards). This resulted in a transfection rate of 1-2 YFP neurons per 30,000μm2 (1-2% transfection efficiency) by 18h.
Cultures of 250μm-thick coronal slices through the striatum were prepared from PND10 rats as previously described [Crittenden, et al. 2010] and plated on 0.4μm Millipore PET membrane inserts. After a 1h recovery period from sectioning, slices were biolistically co-transfected as described for primary cortical neurons above. Slice cultures were maintained at 32°C under 5% CO2.
To examine the influence of microglia on mHtt-transfected neurons, exogenous microglia were added to primary neuronal cultures. Dissociated primary glia were generated according to the method of McCarthy and deVellis [McCarthy and de Vellis 1980] from cortical tissue of PND1-2 rats or mice with trypsin digestion, sequential filtration (165μm and 40μm nylon mesh), and plating in DMEM (Invitrogen) with 10% FBS and 1% P/S. Mixed glia cultures (~10DIV) were shaken at 35rpm for 7h at 37°C. Detached microglia were filtered (70μm) and resuspended in Neurobasal media.
microglia (0-5000 cells per well) were added to transfected neurons at 2-4h post-electroporation and cultured with the transfected neurons for 7DIV prior to evaluation of cell viability, microglial proliferation, and microglial localization to neurons.
microglia were labeled with isolectin B4-594 from Griffonia simplicifolia (IB4; 1:200 dilution; Sigma-Aldritch) during the final 1h of shaking (at 12rpm). Fluorescent microglia were washed with PBS, filtered (70μm cell strainer), and added to neuron cultures 4h post-transfection at 1.5 × 104 cells/cm2 for time-lapse microscopy of localized targeting to neurons.
Total RNA was isolated with TRIzol (Invitrogen) after micro-punch (Miltex, York, PA) excision of the striatal tissue (5 cultured slices pooled per sample; n=3). Reverse transcription was performed with 1μg total RNA using SuperScript II Reverse Transcriptase (Invitrogen). qPCR was carried out (Perkin Elmer ABI Prism 7700 Sequence Detector) using 50ng cDNA as template, 1X Power SYBR Green Master Mix (Applied Biosystems; Foster City, CA), and forward and reverse primers (Table 1). Reaction mixtures were held at 50°C for 2-min, 95°C for 10-min, followed by 40 cycles at 95°C for 15-sec and 1-min at 60°C. Amplification curves were generated with Sequence Detection System 1.9.1. Threshold cycle values were determined and mean fold changes over the 2DIV control samples were calculated according to the 2-ΔΔCT method [Livak and Schmittgen 2001] and normalization to RPL32 (see Table 1 for primer sequences).
Pooled slices (n=6 per sample) collected from individual culture preparations were lysed in RIPA buffer with protease inhibitor (Roche, Basel Switzerland), incubated on ice for 30min, and centrifuged at 13,000xg. Thirty μg of protein supernatant was separated by electrophoresis on Novex Bis-Tris gels (4-12%) using NuPage solutions under reduced conditions (Invitrogen). Protein was transferred to PVDF membranes (Invitrogen) and incubated overnight at 4°C with rabbit polyclonal antibodies to Iba1 (1:1000; Wako, Osaka Japan) or β-actin (1:5000, Abcam). Immunoblotting and visualization were performed with WesternBreeze reagents (Invitrogen). Images were acquired (Kodak Image Station 440cf; Kodak, Rochester NY) and Iba1 protein expression level was calculated as mean intensity relative to β-actin.
Cultured brain slices were fixed with 4% paraformaldehyde (PFA)/ 0.1% glutaraldehyde in PBS (18h, 4°C). Slices were cryoprotected in 30% sucrose and cryosectioned at 10μm. Dissociated neuronal cultures were fixed in 4% PFA/ 4% sucrose in PBS (15 min). Following blocking (PBS with 2% goat serum, 1% bovine serum albumin, and 0.1% Triton X-100), samples were incubated with primary antibodies (18h, 4°C). Microglia were identified with anti-Iba1 (1:250; rabbit polyclonal; Wako) or IB4 (1:100-1:200; Sigma-Aldritch), proliferative cells with anti-Ki67 (1:200; mouse monoclonal; Leica Microsystems, Bannockburn, IL), and aggregated N-terminal htt fragments [Gutekunst, et al. 1999] with anti-EM48 (1:250; mouse monoclonal; Millipore) followed by Alexa Fluor (594nm or 488nm, Invitrogen) secondary antibodies (1:200-1:500). Dissociated cells were counterstained with 0.1mM Hoescht 33258 (Sigma-Aldritch) and brain slice sections with Prolong + DAPI (Invitrogen). Images were captured by epifluorescence using a Nikon TE2000 inverted microscope (Tokyo, Japan) and processed with Metamorph™ (Universal Imaging Co., Downingtown, PA). Confocal images (LSM 5 Pascal inverted, laser-scanning confocal microscope; Carl Zeiss, Oberkochen, Germany) were processed with LSM 5 software.
Neuronal viability in composite cultures was quantified using the Cellomics Arrayscan VTI (Thermo Scientific, Pittsburgh, PA; [Kaltenbach, et al. 2010]). The timing of mHtt-induced loss of healthy YFP+ MSNs beginning at 3DIV in brain slices was confirmed [Crittenden, et al. 2010].
In 7DIV composite cortico-striatal cultures, 70-140 regions for each combination of treatment (control or mHtt transfection) and microglia supplementation (200-1000 microglia) were collected across multiple tissue culture wells and 2-4 individual cell culture preparations. Using an automated x-y stage and pre-programmed, distance-defined coordinates, the regions were selected in a “systematic-random” fashion from random start points within the culture wells and 20x images of YFP, CFP, 594 (microglial markers), and DAPI were acquired using constant exposure times. Images were thresholded and individual, immuno-positive microglia co-expressing DAPI were counted and averaged by culture. Microglia within IB4 or Iba1-stained clusters were counted only if individual cell morphology could be resolved and met cell identification criteria (≥7μm diameter and one DAPI+ nucleus).
Cell viability was analyzed using a one-way ANOVA followed by Dunnett's t-test for comparisons of independent group means. qPCR for mRNA levels, microglial cell counts in non-manipulated dissociated neuronal cultures, and neuronal branching were analyzed using two-tailed Student's t-tests. Microglia morphology and cell counts in acute slice cultures or following addition of exogenous microglia to dissociated neurons were analyzed using a two-way ANOVA with Bonferroni's post-hoc tests for independent group mean analysis. Statistical significance was set at p<0.05.
To examine the relationship between microglia and neurons, we visualized cell-cell interactions following neuronal expression of mHtt. Neuronal expression of N-terminal fragments of mhtt protein containing pathological length polyglutamine expansions initiate neuronal cell death mechanisms that are similar to those seen in genetic models [Runne, et al. 2008]. In our “composite” cortico-striatal culture system, mHtt expression results in the degeneration of approximately 60% of the transfected neurons by 7DIV relative to controls [Kaltenbach, et al. 2010]. Concurrent with neurodegeneration, an increased staining for the microglial marker, Iba1, was observed in cultures transfected with mHtt (Fig. 1A-B), correlating with a significant, 26% increase in microglia number (Fig. 1C).
To further define the relationship between microglia cell number and neuronal expression of mhtt fragments, microglia were added to composite cortico-striatal cultures. An initial seeding of 200, 500, or 1000 exogenous microglia to control cultures resulted in Iba1+ cell number increases of 2.1, 3.9, and 5.8-fold over non-seeded cultures, respectively (Fig. 1D). In comparison, Iba1+ cell numbers in mHtt-transfected cultures were increased 3.6, 7.0, and 9.5-fold. Calculation of the best-fit lines showed an approximate doubling in the slope of Iba1+ cell numbers following neuronal expression of mHtt (Control: y=0.0208x+4.84 versus mHtt: y=0.0464x+8.70). A 2x2 ANOVA revealed significant main effects of transfection and exogenous microglia, and a significant interaction between transfection and exogenous microglia. Although a robust elevation in Iba1 labeling was observed with neuronal mHtt expression, seeding number did not appear to influence the location of microglia as cells remained evenly distributed around transfected striatal neurons (Fig. 1A-B,E-J).
To determine if microglia localize to striatal neurons in a preferential manner, composite cortico-striatal cultures were evaluated. In both control and mHtt-transfected cultures, microglia did not appear to cluster at the neuronal somata; rather, they appeared to display a preferred localization along thin processes and at points of intersection or overlap between neuronal processes (Fig. 1E-J). Relative to controls, mHtt-transfected striatal neurons possessed a reduced number of primary branches and displayed bulbous swellings along fine processes (Supplementary Fig. 1). Iba1+ microglia were often observed in association with process blebs and clustered at the terminal ends of mHtt+ striatal neurites with an irregular morphology (Fig. 1J, arrows).
Proliferation is one component of the microglial response to injury. In composite cortico-striatal cultures, the mhtt fragment-induced increase in Iba1+ microglia at 7DIV appeared to be paralleled by an increase in microglial proliferation, as demonstrated by elevated Ki67 staining in Iba1+ microglia (Fig. 2; quantified in Supplementary Fig. 3A). Both Ki67+, proliferative microglia (arrows) and Ki67- microglia appeared to interact with striatal neurons primarily at neurites at varying distances from the neuronal somata, however, this preference appeared less consistent for Ki67- microglia (Fig. 2).
In cortico-striatal cultures, microglial localization at neurites did not appear to be directly related to the location of aggregated mhtt fragments. In striatal neurons, aggregated mHtt labeled with anti-EM48 was observed within fine neurites and thickened processes with bulbous swellings (Fig. 3, arrowheads), as well as within the neuronal somata (Fig. 3, arrows). Microglia appeared to be evenly distributed at aggregate-nonspecific points along fine, YFP+ neurites, rather than at sites of mHtt accumulation (Fig. 3). These microglia-neurite interactions were often some distance from any visible EM48 staining (Fig. 3C-D, inset).
Using biolistic transfection, we generated a primary neuronal culture system with a lower expression level of mHtt. This system allowed us to visualize individual processes of transfected neurons and determine whether microglial association with neurites is associated with the onset or progression of neuritic degeneration. Neurons transfected with control plasmids displayed normal neurite morphology for >72h (Fig 4A). Within 24h, mHtt+-transfected neurons displayed characteristic CFP+ mhtt aggregates in the cell nucleus and the proximal and distal neurites (Fig. 4B). The expression and size of aggregates coincided with the appearance of neurite irregularities, such as axonal swelling, re-curving of distal neurites, and deteriorating dendrite morphology by 48h (Fig. 4C). Consistent with the composite cortico-striatal cultures, localization of pre-labeled, exogenous microglia appeared to preferentially occur along neurites (Fig. 4) at mHtt aggregate-nonspecific sites (data not shown). Using time-lapsed microscopy, 24 to 48h after mHtt transfection, microglia were observed at distal processes near to the terminal ends of neurites, at sites of neurite intersection, and at positions along neurites where irregularities form (Fig. 4D-G). The severity of degeneration progressed rapidly after 48h, such that by 72h the majority of mHtt-transformed neurons displayed degenerated processes and/or fragmented cell somata (Fig. 4J). During this period of active neuronal degeneration, microglia-neuron associations appeared to remain committed to the initial neurite contact, as microglia tracked between these timepoints did not appear to migrate to an alternative site (Fig. 4H, J). In many instances, a shift in the morphological phenotype of microglia associating with degenerating neurites was observed (Fig. 4I, K).
To extend these observations to a more intact system, we examined the pattern of microglial changes occurring in a cortico-striatal slice culture system. This slice maintains inputs to striatal cells from cortical sources, but lacks extrinsic afferents from dopaminergic or thalamic sources. Consistent with previous work [Lo, et al. 1994], biolistic transformation of slice cultures resulted in selective gene expression within neurons (Fig. 5A-D) without ancillary trauma or noticeable activation of surrounding glia. Control slices expressed both YFP and CFP throughout the transfected striatal neurons (data not shown). mHtt- transfected slices expressed YFP throughout the neurons and CFP selectively at locations of aggregated mhtt in the neuronal somata and processes (Fig. 5A), as indicated by co-localization with EM48 (data not shown). Consistent with previous studies showing that transfection of slice cultures with mHtt induces progressive neurodegeneration that becomes measurably divergent from controls between 3-5DIV [Crittenden, et al. 2010], striatal YFP mRNA levels were similar between groups at 2DIV (mHtt slices= 81 ± 7.9% of controls, n=3) and decreased by mHtt at 5DIV (mHtt slices= 46 ± 17% of controls, p<0.05, n=3).
The tendency of microglia to distribute and localize to neuronal processes as observed in dissociated cell cultures was maintained in slice cultures. At 3 and 5DIV, microglia appeared uniformly distributed around the somata of labeled neurons (Fig. 5B-E), with an apparent preferential localization of IB4-labeled microglia to neuronal processes (arrows). In the mHtt-transfected slices, microglia were observed at sites of process fragmentation (Fig. 5C, E), however, the localization of microglia to neuronal processes did not appear to demonstrate preferential colocalization to CFP-labeled mhtt aggregates (arrowheads).
In HD, increases in microglial cell number are closely linked with elevations in disease severity [Sapp, et al. 2001]. To quantify and localize the observed increase in microglia endogenous to the slice following neuronal expression of mhtt fragments, IB4-labeled microglia within 100μm of YFP-labeled neuronal somata in the striatum of control and mHtt-transformed cultures were counted using a Sholl analysis (Supplementary Fig. 4A, C). In control cultures, a progressive increase was observed over time in culture, with a significant increase occurring at 5DIV. In mHtt-transformed cultures, this process was accelerated, with a significant increase in the number of microglia observed at 3DIV (Fig. 6A). This accelerated microglia increase coincided with the progression of neuronal degeneration. Similar to the cell death time course, a differential between control and mHtt slices was detected by 3DIV and exacerbated by 5DIV, with a significantly greater number of microglia seen in the mHtt-transfected slices (Fig. 6A,C). Two-way ANOVA revealed significant main effects of DIV and transfection, and a significant DIV by transfection interaction.
The increase in microglia was not observed uniformly across the slice but was specific to within the experimentally-defined radius of the neuronal process field. At 3DIV, the number of microglia observed within regions devoid of YFP expression (>250μm of any observed YFP signal) were similar across control and mHtt-transformed cultures (Fig. 6B), indicating that the treatment-related increase in microglia was dependent upon the local presence of neurons expressing mhtt fragments. When the distribution of microglia along the neurite axis was quantified, the pattern of localization in slice cultures confirmed observations of a lack of microglia homing to the cell body of neurons in dissociated cultures. In mHtt-transfected slice cultures, the number of microglia within each of the Sholl regions was significantly increased at 3 and 5DIV, as compared to controls (Fig. 6C). However, the distance-defined localization of microglia around neuronal somata was similar across DIV (data not shown) and transfection (Fig. 6D).
The increase in microglia number could have arisen from increased proliferation of microglia. Similar to observations in composite cortico-striatal cultures, staining of Iba1+ microglia nearby transfected striatal neurons was increased with expression of mhtt fragments (Fig. 7). At 3 and 5DIV, an increase in Iba1+ microglia co-expressing Ki67 (e.g. arrows in Fig. 7E-F) was observed in mHtt-transfected striatal slices, as compared to control slices (Fig. 7; quantified in Supplementary Fig. 3B), suggesting a proliferative response of microglia localized within the dendritic arbor of mHtt-transfected neurons.
Cell surface molecules expressed by microglia can reveal select changes in their functional phenotype [Gehrmann, et al. 1995]. In cortico-striatal slice cultures, at 2DIV, prior to observable neuronal injury and prior to any observed increase in microglia cell number, Iba1 mRNA levels were elevated by neuronal expression of mHtt (Fig. 8A). Iba1 is a calcium-binding adaptor molecule that is expressed by microglia at rest and at progressively higher levels with increasing activation state. This increase suggests an early response of these cells prior to microglial proliferation and neuronal death. At 3DIV, when mHtt-induced neurodegeneration is first observed, increased microglia cell numbers coincided with a 50-70% elevation in Iba1 protein levels over controls (Fig. 8B). At 5DIV, Iba1 mRNA levels remained elevated in response to mHtt, as compared to controls (Fig. 8A). mRNA levels for the scavenger receptor, Msr1, were decreased prior to neuronal death (2DIV), but returned to control levels by 5DIV (Fig. 8A). Genes associated with microglial phagocytosis (CD68), migration (CXCR3), and chemokine signaling from neurons (CX3CR1) were unchanged by mHtt transfection at both time points (Fig. 8A and data not shown).
The activation state and functional demands of microglia can be inferred, in part, according to their morphology, which ranges from process-bearing to amoeboid [Hanisch and Kettenmann 2007]. Given that microglia can be activated by cellular changes in the environment such as neuronal injury, we examined the temporal expression of distinct morphological phenotypes of microglia as a function of mHtt neuronal transfection in slices (Fig. 9). As mHtt- transfected neurons within slice cultures begin to display clear evidence of perturbation at 3DIV, the ramification states of microglia <100μm from transfected striatal neurons were analyzed at 3 and 5DIV. Since the procedure for generating slice cultures induces a level of morphological activation of microglia over subsequent DIV [Huuskonen, et al. 2005], these analyses were expressed relative to DIV- matched microglia in control cultures. Microglia surrounding mHtt- transfected striatal neurons exhibited a reduction in the number and length of their processes in comparison to microglia surrounding control neurons (Fig. 9A). To quantify the shift towards a rounded, or amoeboid, morphological phenotype, form factor (FF) analyses were conducted (Supplementary Fig. 4B, D) to calculate the process complexity of microglia <100μm from transfected neurons based on the cell perimeter, cell area, and the area of the polygon inscribed by the outermost microglial processes, as previously described [Heppner, et al. 1998, Wilms, et al. 1997]. At both time points, a significant increase in the percentage of microglia with amoeboid morphology was observed in the vicinity of mHtt-expressing neurons compared to their YFP counterparts (Fig 9B). Two-way ANOVA showed a significant main effect of both transfection and DIV. No change in the overall size of the individual, IB +4 microglia was observed, although cellular clustering was more frequent in mHtt-transfected cultures (data not shown). No mHtt-related changes in GFAP levels or astrocyte morphology were observed (Supplementary Fig. 5).
To examine the functional impact of the increased number of microglia within the dendritic arborization of mHtt-transfected neurons, the composite cortico-striatal cultures were used to directly manipulate the number of microglia. Adding exogenous, primary microglia to mHtt-expressing neurons allowed for the assessment of cell type-specific contributions to an HD-like phenotype. Supplementation of cortical and striatal co-cultured neurons with exogenous microglia on the day of electroporation and plating resulted in “dose-dependent” increases in neuronal survival that was proportional to the numbers of microglia added (Fig. 10). While neuroprotection of striatal neurons occurred in both control and mHtt cultures, the effect on mHtt- transfected neurons was significantly greater, with an increase in neuronal viability at 5x lower levels of microglial supplementation (Fig. 10A). Unlike the striatal neurons, viability of cortical neurons in the control cultures was not altered by addition of exogenous microglia. The viability of mHtt-expressing cortical neurons was significantly improved by the addition of microglia (Fig. 10B).
Previous associations between HD progression and increases in cytokines such as IL-1β, IL-6, and TNF-α have been reported [Kyrkanides, et al. 2001]. To address the potential influence of microglia on the induction of these pro-inflammatory cytokines, we examined mRNA levels for TNF-α, IL-1β, and IL-6, in the striatum of cortico-striatal slice cultures. The proliferation and increased local reaction of microglia in the mHtt- transfected slices was not accompanied by elevations in TNF-α, IL-1β, or IFNγ mRNA levels (Fig. 11A and data not shown). In response to mHtt transfection, there was no change in the expression of any of these cytokines prior to neuronal damage and, at the onset of neurodegeneration (3DIV), no changes in IL-1ß and IL-6 protein levels were observed (data not shown and Supplementary Fig. 5). However, by 5DIV, mRNA levels for IL-6 were significantly elevated in the striatum of mHtt-transfected slice cultures, as compared to controls (Fig. 11A).
The localization of microglia to neurites, independent of mhtt aggregate presence, and the observation in dissociated neuronal culture that microglia retained the initial contact association with neurites, both suggest a role for microglia in contact-dependent relationships with afflicted neurons. Activation of the cellular complement system has been reported to be involved in neuronal dysfunction in HD [Singhrao, et al. 1999] and specific signaling interactions of complement component 1q (C1q) have been implicated in synaptic/dendritic remodeling in neurodegenerative disease [Stevens, et al. 2007]. Functional C1q consists of six A, B, and C chains encoded by the genes C1qA, C1qB, and C1qC, respectively. The expression of mHtt in slice cultures resulted in a significant, 75% increase in mRNA levels for C1qA and a significant, 35% increase in C1qB mRNA levels (p<0.05), as compared to control slices at 5DIV (Fig. 11B). mRNA levels for C1qC were not significantly different between transfection conditions (Fig. 11B) and, at the onset of neurodegeneration (3DIV), C1q protein was marginally increased by 10% in mHtt-transfected cultures, as compared to controls (Supplementary Fig. 5).
Neuronal expression of N-terminal mhtt fragments containing pathological length polyglutamine expansions is sufficient to initiate HD-like phenotypes and neuronal degeneration [Runne, et al. 2008, Saudou, et al. 1998]. We now demonstrate that expression of mHtt in striatal neurons, within both primary cells and brain slice cultures, initiated a localized response of wild type microglia, including elevated numbers and activated morphological phenotypes. These proliferating microglia appeared to associate preferentially with neurites at time points representing early and late degenerative phenotypes of the neurons, but did not directly contribute to neurodegeneration. Rather, microglial supplementation provided measurable support to mHtt- transfected neurons. Thus, a functional shift of microglia prior to neuronal toxicity appears to occur in response to some unidentified factor that may be produced along dystrophic or dysfunctional neurites and is associated with downstream signaling responses, including IL-6 and C1q.
We now report a stimulatory effect of mHtt-induced neuronal dysfunction on microglia. Microglial interactions with degenerating neurons appeared to occur primarily at neurite intersections and swellings, as well as thinning processes, regardless of the local presence of aggregated mhtt. Although aggregated mhtt fragments can induce toxicity [Yang, et al. 2002], they do not correlate with the progression of HD pathology [Saudou, et al. 1998] and may even represent a protective response of the cell to soluble mhtt [Arrasate, et al. 2004]. From qualitative observations of co-labeling for microglia and mhtt aggregates, the absence of a distinct localization of microglia to sites of visible mhtt aggregation argues against a localized, aggregation-specific signal for microglial activation. However, aggregation-mediated changes which subsequently activate microglia at distant sites or an altered composition of in vitro aggregates cannot be ruled out.
Neurite intersections are unique in that they contain a large proportion of mitochondria and have a high likelihood of synaptic contacts. Thinning neuronal processes can be representative of a severe loss of dendritic spines and the associated loss of synapses, while structural sites of process bleb formation have been correlated with high Ca2+ entry after stimulation of glutamate receptors (GluRs) [Bindokas and Miller 1995]. HD patients exhibit neuronal network remodeling, including dendritic atrophy and sprouting [Graveland, et al. 1985], and one mechanism of mhtt-induced cell death involves excitotoxic overstimulation of GluRs [Tabrizi, et al. 1999]. In general, morphological changes in the neurites of mhtt-expressing neurons can be attributed to metabolic disturbances, including disrupted transcription [Nucifora, et al. 2001] and release of factors that block protein transport [Li, et al. 2001].
In models of brain injury, microglia or brain macrophages can either be recruited to a neuronal injury site [Kurpius, et al. 2007] or proliferate locally [Graeber, et al. 1988]. Depending on the source of the brain macrophages and their capacity for proliferation, the accumulation of these cells can result in differential effects on the neuronal population. Previous studies demonstrate that transient, proliferative microglia can aid in the repair and regeneration of neurons following injury, while induction of non-proliferative microglia can be associated with phagocytic activity and dendrite loss [Graeber, et al. 1998, Rappert, et al. 2004, Rogove, et al. 2002]. In human HD, activated microglia increase with disease severity and contact diseased neurons at both the cell body and processes [Sapp, et al. 2001]. Within the current culture model, neuronal expression of mhtt fragments initiated a proliferation of microglia that correlated with the apparent localization of microglia to morphologically- aberrant and dystrophic neurites, yet there was no evidence of direct neurotoxic actions. Together, the localization of microglia to these structural sites and microglial proliferation preceding overt neurotoxicity suggests that microglia are responding to early, detrimental changes within afflicted neurons rather than initiating processes which directly lead to neuronal dysfunction and death. Early changes in the synaptic health of mHtt- expressing neurons may represent one such activation signal for microglia
Microglia possess great diversity in their activation phenotypes. The acquisition of these various functional states involves a wide range of cell surface receptors and cytokines whose expression depends on the type and timing of initiating cues in the microenvironment [Perry, et al. 2010]. While microglial activation is typically associated with increased phagocytosis, many other functional and adaptive changes are employed prior to, and during, cellular activation [Napoli and Neumann 2009]. Several of these changes involve limiting or counteracting neuronal damage and dysfunction, including repair or regeneration of damaged tissue, turnover of persistent proteins, removal of glutamate, and support of neurogenesis [Hanisch and Kettenmann 2007, Neumann, et al. 2009]. Additionally, microglia are known to be involved in monitoring and pruning neuronal synapses, both on a regular basis and with injury, the latter occurring, presumably, as a neuroprotective action [Trapp, et al. 2007, Wake, et al. 2009].
Our data support the induction of an early phenotypic change in microglia in response to neuronal expression of mhtt fragments. Msr1, a scavenger receptor expressed by phagocytic microglia, is decreased, and Iba1, a microglial marker that is upregulated in response to cellular activation, is increased, at 2DIV. These events occurred prior to neuronal degenerative events and any observable difference in microglia cell numbers or proliferation around mHtt- transfected neurons. Alterations to these markers indicate a changing phenotype of microglia in association with neurons lacking any mHtt- induced neuropathological features, which precedes microglial proliferation and morphological activation.
Despite these early alterations in cell surface molecules and close association with sites along mHtt neurites, the microglia gave no evidence of actively contributing to degenerative phenotypes in mHtt-transfected neurons. Microglia were often localized to neuronal processes in both control and mHtt-transfected cultures, indicative of a normal process of neurite stabilization and remodeling. The observation that microglia were neuroprotective/ neurosupportive to primary striatal neurons, but not cortical neurons, following transfection with control plasmids suggests that a differential sensitivity to microglial actions may exist across neuronal populations derived from different brain regions. An amplification of this normal repair process appears to be required following neuronal expression of mhtt fragments, as supplementation with microglia provided neuroprotection for both striatal and cortical, mHtt-transfected neurons. Similar to observations of trophic support of normal neurons by non-stimulated microglia in culture [Zhang and Fedoroff 1996], microglia in co-culture with neurons expressing mhtt fragments were not toxic, but rather, displayed protective phenotypes that seemed to be initiated along neurites. In their study, Zhang and Federoff noted that microglial interactions improved the arborization and thickness of contacted neurites, despite the acquisition of “phagocytic” phenotypes. This observation suggested that morphological activation of microglia does not preclude their capacity for trophic support of neurons.
While normal microglia were supportive of mHtt- transfected neurons in the current studies, microglia expressing mhtt may possess a different phenotype. As microglia in HD have been shown to express mhtt [Shin, et al. 2005], it is important to recognize that the models used in the current studies recapitulate only a select facet of the disease: namely, the acute response to neuronal expression of mhtt. In the R6/2 mouse model of HD, with mhtt expression in all cells, there is a reported decrease in the overall number of microglia, with many displaying structural changes such as process swellings and fragmentation [Ma, et al. 2003]. Microglial expression of mhtt has been reported to induce a “functional overactivity” involving altered iron metabolism, perturbed kynurenine pathway regulation, and the production of excess IL-6 upon challenge [Bjorkqvist, et al. 2008, Giorgini, et al. 2008, Simmons, et al. 2007]. Such alterations would significantly alter the normal capacity of microglia to regulate the neuronal environment. Thus, it is possible that targeting the regulation of microglia toward a more normal responsive phenotype can significantly contribute to the protection of HD neurons.
Abnormalities in synapse proteins and synaptic transmission occur prior to neuron loss and contribute to early HD symptoms [DiProspero, et al. 2004, Li, et al. 2003]. Microglial processes directly contact neuronal synapses in vivo and these contacts increase in frequency with elevated neuronal activity [Wake, et al. 2009]. During the early stages of neuronal injury, signals occur at synapses to orchestrate synapse stripping and remodeling. Both activated microglia [Blinzinger and Kreutzberg 1968, Trapp, et al. 2007] and complement factors [Stevens, et al. 2007] are now shown to be essential for this process, which occurs in the absence of neuropathology [Trapp, et al. 2007]. Complement is a critical component of the innate immune response against pathogens. Upon injury, C1q produced by microglia [Lynch, et al. 2004] binds to neurons, promoting phagocytosis through its interaction with membranes of apoptotic cells, blebs, or structures otherwise identified as abnormal [Fraser, et al. 2010, Ogden, et al. 2001], including excess or aberrant synapses targeted for elimination [Stevens, et al. 2007]. Local production of complement factors by reactive microglia has been observed in the caudate of HD patients [Singhrao, et al. 1999]. We now report that mHtt expression in striatal neurons resulted in the appearance of degenerative features such as neuronal blebs, which coincided with an increased expression of C1q. This concurrent timing suggests that C1q serves as a signal to target microglia to neurites for clearance of aberrantly-functioning synapses.
Changes to the pro-inflammatory cytokine profile of microglia can have a significant impact on the health of the neurons with which they associate [Kim and de Vellis 2005]. Reciprocally, microglial production of these same cytokines (e.g. IL-6 and TNF-α) can be increased by the release of factors such as ATP from damaged or stressed neurons [Diaz-Hernandez, et al. 2009, Inoue 2002]. In physical lesion models, brain macrophages can lead to secondary neuronal and dendritic damage [Eyupoglu, et al. 2003] and neurodegeneration initiated by microglia can be driven through TNF-α signaling [Iliev, et al. 2004]. TNF-α is predominantly produced by microglia [Hetier, et al. 1990, Kraft, et al. 2009] and has been observed to be inversely associated with their phagocytic activation [Magnus, et al. 2001]. Recent work has demonstrated a dual role of C1q in modulating the cytokine environment and facilitating microglial clearance of injury-generated apoptotic neurons or neuronal blebs [Fraser, et al. 2010]. During this specific process, and in the absence of activation of the entire complement cascade, C1q elicits enhanced microglial phagocytosis and concomitantly suppresses the pro-inflammatory cytokines IL-1α, IL-1β, TNF-α, and IL-6. The absence of elevations in TNF-α or IL-1β mRNA levels within the mHtt-transfected slice cultures is consistent with suppression by C1q.
In the current slice culture model, we now demonstrate that the proliferation and apparent targeting of microglia to neurites does not involve an elevation in pro-inflammatory TNF-α. The absence of TNF-α induction suggests that microglia fail to mount a cytotoxic, pro-inflammatory response to neuronal expression of mHtt and the subsequent neurite degeneration. Although HD has previously been shown to induce production of IL-1β [Ona, et al. 1999], no elevation in this cytokine was observed in our model. CSF and plasma from HD patients shows increases in cytokines involved with the innate immune response, including IL-6, at early stages of disease [Bjorkqvist, et al. 2008, Dalrymple, et al. 2007]. IL-6 has a dual role as cytokine and growth factor and can be regulated by complement factors [Hajishengallis and Lambris 2010]. Importantly, in our model, IL-6 is not changed until well after the onset of neuronal degeneration initiated by mHtt. Rather than a microglial pro-inflammatory response, the elevations in IL-6 may represent a protective response of stressed or damaged neurons, as has been shown with injuries such as axotomy [Murphy, et al. 1999] and cerebral ischemia [Suzuki, et al. 1999].
We observed commonalities between the reaction of microglia to control neurons and to neurons expressing mHtt; however, a clear amplification of these normal responses occurred following neuronal expression of mhtt fragments. We propose that, in the early stages of the disease, microglia are stimulated and targeted to neuronal processes, where they proliferate. This targeting occurs similar to the normal microglial response to synapse loss or dysfunction and process degeneration, and is not specifically driven by the aggregation of mhtt fragments within the neurites. The suggested sequence of events involves microglial proliferation, morphological progression towards an amoeboid phenotype, the recognition of C1q on neuronal synaptic terminals and somata for phagocytosis and clearance, and an elevation in IL-6 as a potential component of the repair process. In the context of HD therapy research, the neuroprotection facilitated by these cells cautions against the assumption that microglia implicitly pose a detrimental threat to neurons. Future detailed examination into mhtt fragment-induced changes in synaptic health and the resultant amplification of normal microglial processes will contribute to defining the role of these cells during the initiating stage of disease.
The authors wish to thank Gregory Turmel, M. McLean Bolton, Bijal Shah, and Denise Dunn for their technical assistance, Drs. William Mundy and Rebekah Jakel for their review of the manuscript, and Dr. Andrew Tenner (University of California, Irvine) for her gift of C1q antibody. This research was partially funded by the Division of Intramural Research of the National Institute of Environmental Health Sciences, National Institutes of Health, Department of Health and Human Services Z# ES101623 and ES021164, the Cure Huntington's Disease Initiative Foundation, Inc., and the Hereditary Disease Foundation.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of interest
None of the authors have any actual or potential conflicts of interest