Huntington disease neuroimmune reframing: microglia transition to a pro-inflammatory DAM state before neuronal loss — early microglial CSF1R inhibition may delay striatal atrophy onset

In Huntington's disease (HD) mouse models carrying the mHTT mutation, striatal microglia undergo a transcriptional transition to a pro-inflammatory Disease-Associated Microglia (DAM) state (characterized by upregulation of Trem2, Lpl, Cst7, Apoe, and downregulation of homeostatic markers P2ry12, Cx3cr1, Tmem119) that is temporally antecedent to measurable striatal neuronal loss (>10% medium spiny neuron depletion by stereology). Furthermore, pharmacological inhibition of CSF1R (e.g., PLX5622 or BLZ945) initiated prior to DAM transition onset reduces striatal volume loss by ≥25% and delays motor symptom onset by ≥2 weeks compared to vehicle-treated HD controls, as measured in at least two independent HD mouse models (e.g., zQ175 and R6/2) across a minimum 12-week treatment window.

Minimum Viable Test (MVT) — 3-Phase Design

Phase 1 — Temporal Mapping (12 weeks, in vivo + transcriptomics):

• Use zQ175 knock-in mice (n=12/sex/genotype = 48 total) and R6/2 mice (n=10/sex/genotype = 40 total)
• Collect striatal tissue at 4 timepoints per model (pre-symptomatic, early symptomatic, mid, late)
• Perform scRNA-seq (10x Genomics Chromium) on FACS-sorted CD11b+/CD45-low microglia
• Perform NeuN+ stereological counting at each timepoint
• Primary output: DAM transition timeline vs. neuronal loss timeline

Phase 2 — CSF1R Inhibition Intervention (16 weeks, in vivo):

• zQ175 mice: begin PLX5622 (1200 ppm in chow) at 2 months (pre-DAM) vs. 6 months (post-DAM onset) vs. vehicle; n=15/group/sex
• R6/2 mice: begin PLX5622 at 3 weeks vs. 6 weeks vs. vehicle; n=12/group/sex
• Primary endpoints: striatal volume (MRI at 3T), NeuN+ stereology, rotarod performance
• Secondary endpoints: microglial density (Iba1+), DAM marker expression (IHC + qPCR), inflammatory cytokine panel (Luminex 27-plex)

Phase 3 — Mechanistic Validation (8 weeks):

• Conditional Trem2 knockout (Trem2-flox × Cx3cr1-CreERT2) in zQ175 background to test DAM-specific contribution
• Bulk RNA-seq of striatal tissue from treated vs. untreated HD mice
• Proteomics (TMT-labeled LC-MS/MS) on striatal lysates

# HD Microglial DAM Validation Pipeline — Pseudocode Architecture

## MODULE 1: Animal Management System
class HDMouseCohort:
    def __init__(self, model, genotype, sex, treatment, n):
        self.model = model  # 'zQ175' or 'R6/2'
        self.genotype = genotype  # 'HD' or 'WT'
        self.sex = sex  # 'M' or 'F'
        self.treatment = treatment  # 'PLX5622' or 'vehicle'
        self.n = n
        self.timepoints = self.assign_timepoints()
    
    def assign_timepoints(self):
        if self.model == 'zQ175':
            return [2, 4, 6, 9]  # months
        elif self.model == 'R6/2':
            return [4, 6, 8, 12]  # weeks

## MODULE 2: scRNA-seq Processing Pipeline
def process_scrna_seq(raw_fastq_dir, sample_metadata):
    # Step 1: Alignment
    run_cellranger_count(
        fastq=raw_fastq_dir,
        genome='mm10_2020A',
        expected_cells=3000,
        localcores=16,
        localmem=128
    )
    
    # Step 2: Quality control
    seurat_obj = load_10x_data(cellranger_output)
    seurat_obj = filter_cells(
        min_features=200,
        max_features=6000,
        max_mito_pct=20,
        min_counts=500
    )
    
    # Step 3: Normalization and clustering
    seurat_obj = normalize_data(method='LogNormalize', scale_factor=10000)
    seurat_obj = find_variable_features(nfeatures=2000)
    seurat_obj = scale_data(vars_to_regress=['nCount_RNA', 'percent_mito'])
    seurat_obj = run_pca(npcs=30)
    seurat_obj = run_harmony(group_by='sample_id')  # batch correction
    seurat_obj = find_neighbors(dims=1:20, reduction='harmony')
    seurat_obj = find_clusters(resolution=0.5)
    seurat_obj = run_umap(dims=1:20, reduction='harmony')
    
    # Step 4: Microglial cluster identification
    microglia_markers = ['Cx3cr1', 'P2ry12', 'Tmem119', 'Hexb', 'Sall1']
    microglia_clusters = identify_clusters_by_markers(seurat_obj, microglia_markers)
    microglia_obj = subset_seurat(seurat_obj, clusters=microglia_clusters)
    
    # Step 5: DAM scoring
    DAM_signature = {
        'DAM_up': ['Trem2', 'Lpl', 'Cst7', 'Apoe', 'Cd9', 'Itgax', 'Clec7a'],
        'DAM_down': ['P2ry12', 'Cx3cr1', 'Tmem119', 'Sall1', 'Fcrls']
    }
    microglia_obj = add_module_score(
        microglia_obj,
        features=DAM_signature['DAM_up'],
        name='DAM_up_score'
    )
    microglia_obj = add_module_score(
        microglia_obj,
        features=DAM_signature['DAM_down'],
        name='DAM_down_score'
    )
    microglia_obj['DAM_composite'] = (
        microglia_obj['DAM_up_score'] - microglia_obj['DAM_down_score']
    )
    
    # Step 6: Temporal analysis
    for timepoint in sample_metadata['timepoints']:
        tp_cells = subset_by_timepoint(microglia_obj, timepoint)
        dam_high_pct = calculate_dam_high_percentage(
            tp_cells, 
            threshold=mean(DAM_composite) + 1.5*sd(DAM_composite)
        )
        store_result(timepoint, 'dam_high_pct', dam_high_pct)
    
    return microglia_obj, dam_results

## MODULE 3: Stereological Analysis Pipeline
def stereological_analysis(ihc_images, region='striatum'):
    # Automated NeuN+ counting via StereoInvestigator API
    for section in ihc_images:
        contour = define_striatal_contour(section, atlas='Allen_Mouse_P56')
        neuron_count = optical_fractionator(
            section=section,
            contour=contour,
            counting_frame=[50, 50],  # µm
            grid_size=[150, 150],  # µm
            dissector_height=15,  # µm
            guard_zones=2  # µm top and bottom
        )
        store_count(section.animal_id, section.timepoint, neuron_count)
    
    # Calculate total neuron number per striatum
    total_neurons = sum_across_sections(
        counts, 
        section_interval=6,
        section_thickness=40  # µm
    )
    return total_neurons

## MODULE 4: Temporal Precedence Statistical Test
def test_temporal_precedence(dam_results, neuron_counts, models=['zQ175', 'R6/2']):
    results = {}
    for model in models:
        # Find first timepoint with DAM-high > 20%
        dam_onset_tp = find_first_timepoint(
            dam_results[model], 
            condition='dam_high_pct > 0.20'
        )
        
        # Find first timepoint with >10% neuronal loss
        wt_mean_neurons = mean(neuron_counts[model]['WT'])
        neuron_loss_tp = find_first_timepoint(
            neuron_counts[model]['HD'],
            condition=f'neurons < {wt_mean_neurons * 0.90}'
        )
        
        # Test: DAM onset must precede neuronal loss
        precedence_confirmed = dam_onset_tp < neuron_loss_tp
        
        # Statistical test at DAM onset timepoint
        neurons_at_dam_onset = neuron_counts[model]['HD'][dam_onset_tp]
        t_stat, p_val = ttest_ind(neurons_at_dam_onset, neuron_counts[model]['WT'][dam_onset_tp])
        
        results[model] = {
            'dam_onset_timepoint': dam_onset_tp,
            'neuron_loss_timepoint': neuron_loss_tp,
            'precedence_confirmed': precedence_confirmed,
            'p_value_at_dam_onset': p_val,
            'neurons_not_yet_lost': p_val > 0.05  # SUCCESS if True
        }
    
    return results

## MODULE 5: MRI Analysis Pipeline
def analyze_mri_volumes(dicom_dir, treatment_groups):
    for animal in treatment_groups:
        # Preprocessing
        nifti = convert_dicom_to_nifti(dicom_dir / animal.id)
        nifti_brain = skull_strip(nifti, method='BET')
        nifti_registered = register_to_atlas(
            nifti_brain, 
            atlas='Allen_Mouse_Brain_MRI',
            method='ANTs_SyN'
        )
        
        # Striatal segmentation
        striatum_mask = apply_atlas_roi(
            nifti_registered,
            roi='striatum_bilateral',
            atlas='Allen_Mouse_Brain_MRI'
        )
        striatal_volume = calculate_volume(striatum_mask, voxel_size=0.1**3)  # mm³
        store_volume(animal.id, animal.timepoint, animal.treatment, striatal_volume)
    
    # Statistical comparison
    for model in ['zQ175', 'R6/2']:
        plx_hd = get_volumes(model, 'HD', 'PLX5622')
        veh_hd = get_volumes(model, 'HD', 'vehicle')
        wt = get_volumes(model, 'WT', 'vehicle')
        
        protection_pct = (mean(plx_hd) - mean(veh_hd))