Donor bone marrow–derived macrophage MHC II drives neuroinflammation and altered behavior during chronic GVHD in mice

Neurocognitive dysfunction is present in up to 60% of patients with graft-versus-host disease (GVHD) after hematopoietic stem cell transplantation (HSCT). Adams et al describe the immune landscape of central nervous system (CNS) disease in chronic GVHD (cGVHD). While acute GVHD (aGVHD) is mediated by inflammatory mediators from resident microglia, cGVHD depends on infiltration of the CNS with donor bone marrow derived macrophages. Through analysis of the divergent immune CNS profiles in aGVHD and cGVHD, the authors identify potential targets for addressing CNS cGVHD.


Novel Object Recognition Test
The novel object recognition test was performed in accordance with a standardised protocol 1 with minor modifications to assess recognition memory. Mice were accustomed to the experimenter through daily handling (1min/mouse/day) in the week prior to testing. Mice were acclimated to the experimental room for a minimum of 30 minutes prior to each session. The arena consisted of a 40cm x 40cm x 40cm box, cleaned thoroughly with ethanol between each mouse. During the habituation phase, each mouse was allowed to freely explore the empty arena for 10 minutes. 24 hours later, each mouse was allowed to again explore the empty arena for 1 minute, followed by familiarisation with two identical objects placed in the arena, with a total of 10 minutes permitted to reach 20 seconds of object exploration. The testing session occurred on the following day, beginning with 1 minute of empty arena habituation. A familiar and a novel object, differing in texture and shape, were placed in the arena, with a total of 10 minutes permitted to reach 20 seconds of object exploration. The experiment was stopped once 20s was reached. The position of the familiar and novel objects was randomized between mice.
An overhead camera recorded the testing session for manual scoring.

Active Place Avoidance
Spatial learning and memory were assessed with the active place avoidance paradigm (APA) as previously described 2 . Mice were familiarised with the experimenter through daily handling (1min/mouse/day) in the week prior to testing. The APA arena consisted of a rotating (1rpm) grid metal platform with a surrounding transparent plastic boundary. Four different black and white visual cues (A3-size) were placed on the walls in the testing room. Mice were habituated to the arena through 20 minutes of free exploration with the shock zone turned off. Beginning 24 hours later, mice were testing for 20 minutes per day over five consecutive days, in which they were required to integrate the visual cues to avoid a 60 o shock zone, which remained consistent relative to the room. An overhead camera linked to Track Analysis software (Bio-Signal Group, version 2.36) detected the position of the animal in the arena, with entry into the shock zone causing delivery of a mild foot shock (0.5 mA, 500ms duration, 60Hz) at 1.5s intervals until the animal moved out of the zone. The number of entries into the shock zone and time spent avoiding the shock zone were calculated by Track Analysis software (Bio-Signal Group, version 2.36) were calculated as measures of spatial learning and memory.
Performance within one session is indicative of short-term memory, and performance over 5 consecutive days equates to longer-term memory.

Tissue procurement and digest
Mice were anesthetised with a ketamine (40mg/mL)/xylazine (2mg/mL) solution prior to euthanasia by transcardial perfusion of 20mL 1x PBS. After removal of the brain, either the hippocampus was dissected as previously described 3 , or a coronal segment was obtained by removing the olfactory bulbs and cerebellum. Tissue for assessing mRNA expression was immediately snap-frozen in liquid nitrogen, and samples for immunofluorescence were immersion fixed in 4% buffered formalin.
Samples for flow cytometric analysis of leukocytes were dissected as described above, followed by enzymatic digestion in collagenase IV (2mg/mL) and DNaseI (0.5mg/mL) at 37 o C for 20 minutes. Cells were pressed through a 70µm filter and then centrifuged at 1400rpm for 7 min. Cells were resuspended in 25mL 33% Percoll and centrifuged at 1700rpm for 14 min with no brake to obtain a leukocyte pellet. The top film of tissue and Percoll were then aspirated.
Samples for FACS isolation of microglia and BMDM were dissected into small pieces with a scalpel, followed by digestion in papain (Worthington) (1mg/mL) supplemented with DNaseI (0.25mg/mL) at 37 o C for 16 min, triturating every 4 min. Cells were centrifuged at 1100rpm for 10 min, and then gently pressed through a 70µm cell sieve to obtain a single cell suspension.
Cell washes were performed with RPMI supplemented with 2% bovine serum albumin (BSA).

Quantitative Real-Time PCR
The hippocampi and coronal segments of mice were isolated using the same method described above. Total RNA was extracted using Trizol Reagent (Life Technologies Australia) as per the manufacturer's instructions, and quality of RNA was analysed by the NanoDrop spectrophotometer. RT-qPCR was performed using SYBR Green PCR master mix (Life Technologies Australia) and run on an ABI Viia 7 machine (Applied Biosystems). Fold change of gene expression was calculated as 2 -ΔΔCT , and values were normalised against the expression of the housekeeping gene HPRT. Oligonucleotide sequences used are listed in Supplemental Table 3.

Histology
Skin samples were treated with 70% ethanol prior to paraffin-embedding and sectioning for haematoxylin and eosin (H&E) staining. Representative images were acquired with an Aperio AT Turbo brightfield slide scanner and viewed with the Aperio ImageScope software (v.12.4) (original magnification 20x).

Tissue preparation for immunofluorescence
Brains were cryoprotected in 30% sucrose and 0.05% sodium azide in 1xPBS for 24 hours prior to sectioning. Series of 1 in 6 free-floating coronal sections (40m) were obtained using a sliding microtome (Leica).

Immunofluorescence
Brain sections were blocked for two hours at room temperature in 5% BSA with 0.

Imaging and cell quantification
Representative multicolour confocal images were acquired with an inverted ANDOR WD Revolution spinning disk confocal microscope fitted with a Zyla sCMOS camera controlled by Metamorph imaging software. Regional images were obtained using a 20x or 40x CFI Plan Apo VC NA 0.75 WD 1.0mm objective with z-stacks separated by 5m. Microglia morphology was captured with a 100x Plan Apo (oil) NA 1.45 WD 0.13mm objective with 1m z-stack images. Images of RFP + donor T cells for quantification and representative images were obtained using a 20x CFI Plan Apo VC NA 0.75 WD 1.0mm objective with z-stacks separated by 3m. Images of synaptophysin puncta were obtained with a 60x CFI Plan Apo (water) NA 1.2WD 0.31mm objective from three consecutive sections per mouse at Bregma -1.70mm, -1.94mm, and -2.4 with 0.3m z-stacks. All image visualiation was conducted in Imaris Sequence reads were trimmed for adapter sequences using Cutadapt version 1.9 4 and aligned using STAR version 2.5.2a 5 to the Mus musculus GRCm38 reference genome assembly using the gene, transcript, and exon features model of Ensembl (release 70). Quality control metrics were computed using RNA-SeQC version 1.1.8 6 , and transcript abundances were quantified using RSEM version 1.2.30 7 . Some samples were removed prior to downstream analysis due to low mapping rates to the Mus musculus genome. All downstream RNA-seq analysis was completed in R version 4.0.2 (https://www.R-project.org/). Differential expression analysis was completed using edgeR version 3.30.3 8 . The function filterByExpr() was used with default settings (minimum of 10 counts in the smallest biological group determined by the design matrix) to remove genes with low counts. The calcNormFactors() function was used to normalise library sizes using the TMM normalization method. Log2 counts-per-million (logCPM) were generated using the edgeR cpm() function. The edgeR functions glmFit() and glmLRT()were used to fit negative binomial generalized linear models and conduct gene-wise likelihood ratio tests. Heatmaps were created using the heatmap.2 function in the gplots package version 3.1.1 9 . LogCPM values across each gene were scaled using the scale() function so that mean expression is zero and standard deviation is one. Gene ontology, pathway, and gene set analysis was conducted using Ingenuity Pathway Analysis (IPA) (QIAGEN Inc., https://www.qiagenbioinformatics.com/products/ingenuitypathway-analysis), gene set enrichment analysis (GSEA) 10