Radiotherapy induces persistent innate immune reprogramming of microglia into a primed state

Over half of patients with brain tumors experience debilitating and often progressive cognitive decline after radiotherapy treatment. Microglia, the resident macrophages in the brain, have been implicated in this decline. In response to various insults, microglia can develop innate immune memory (IIM)


INTRODUCTION
The number of long-term survivors of brain tumors is steadily growing.However, 50% to 90% of surviving patients experience an irreversible and often progressive impairment in neurocognitive function at a detrimental cost to quality of life. 1 The associated neurocognitive sequelae include a decline in different domains, such as memory, attention, executive function, and intellectual abilities, particularly in survivors diagnosed at a younger age. 2,3he degree and pattern of cognitive side effects vary among patients.3][4] Radiotherapy is a life-saving treatment in the management of primary brain tumors.However, the normal healthy brain tissue is also exposed to radiation when the target tumor volume is irradiated, leading to short-and long-term side effects. 5][8][9] Indeed, studies have reported that radiation can affect both the number and activity of microglia. 10,11However, the longterm molecular and functional state of microglia after brain irradiation remains largely unknown.Similarly to other macrophages, microglia have been recently shown to develop innate immune memory (IIM) capacity. 12,13After an initial insult, IIM comprises long-lasting molecular reprogramming that translates into a functional state in which microglia either enhance (priming, also called training) or suppress (tolerance) their response to a subsequent inflammatory challenge. 14,15The concept of microglial priming has been associated with aging and many neurodegenerative diseases, and the resultant excessive immune response has been linked to neuropathology and cognitive decline. 13,16,17Primed microglia were first demonstrated in a study showing exaggerated interleukin-1b cytokine expression in a model of murine prion disease after a systemic endotoxin lipopolysaccharide (LPS) immune challenge 18 and have been observed in several disease mouse models, including Alzheimer's disease (AD) 19 and accelerated aging with defective DNA repair. 15,20The transcriptomic profile of primed microglia was characterized by high expression of genes associated with phagosome, lysosome, and antigen presentation, 21 similar to the disease-associated microglia (DAM) profile. 22In contrast to priming, microglia can also become desensitized to a subsequent immune insult, entering an IIM state of tolerance, for instance after repeated doses of LPS. 12,13Whether radiation can affect the IIM of microglia by leading to either immune priming or tolerance upon a secondary immune challenge has not been addressed before.
In this study, we showed by transcriptional profiling that radiation induces microglial priming resulting in an enhanced inflammatory response to LPS.Additionally, we identified that this response is persistent over time and is dose and age dependent and that its extent is not affected by the administration of multiple smaller radiation doses.Importantly, a comparative transcriptomic analysis with postmortem normal-appearing (NA) non-tumor brain regions of patients with glioblastoma (GBM) that were treated with radiotherapy revealed an overlap in radiation-associated microglial priming genes, indicating that this response is likely conserved in humans.

Brain irradiation leads to microglial priming
To investigate the effect of radiation on the IIM of microglia, we irradiated the entire rat brain with 14 Gy X-rays and, after 6 weeks, exposed the animals to PBS vehicle or an LPS immune challenge.After 4 h, CD11b pos /CD45 int microglia were isolated by fluorescence-activated cell sorting, and RNA sequencing (RNA-seq) was performed (Figures 1A and S1A).We did not observe differences in the number of peripheral immune cells between control and irradiated animals, indicating that no significant infiltration of peripheral immune cells occurred (Figures S1B and S1C).Principal-component analysis (PCA) showed segregation of all four groups and distinct transcriptomic profiles (Figure 1B).We performed differentially expressed gene (DEG) analysis between the groups (Figures 1C, S1D, and S1E; Table S1).All the DEGs derived from the 4 comparisons were grouped into five different clusters by Manhattan-distance-based hierarchical clustering (Figures 1D and S1F; Table S2).The largest number of genes was in cluster 4, which consisted of genes upregulated in the irradiated groups.Interestingly, cluster 4 contained priming signature genes, such as Itgax, Clec7a, and Gpnmb (Figure 1D), and Gene Ontology (GO) analysis showed an enrichment of terms related to inflammation and immune regulation (Figure 1E).Whereas cluster 5, consisting of genes downregulated after irradiation, was enriched for genes related to cell morphology and extracellular matrix organization, as well as marker genes of homeostatic microglia that are known to be downregulated in reactive or primed microglia, including Tmem119 (Figures 1D and 1E).Together, clusters 4 and 5 indicate that radiation induces transcriptional changes consistent with microglial priming.
We next examined whether irradiated microglia react with a stronger response to a subsequent inflammatory challenge with LPS, which would confirm that microglia acquired a primed state after irradiation.Cluster 1 genes, including proinflammatory genes such as Tnf, Il1b, and Ccl4, were upregulated after LPS treatment and were significantly more upregulated in microglia derived from animals previously irradiated and exposed to LPS treatment compared to non-irradiated LPS-treated animals, showing that radiation primed the microglia (Figures 1D and 1F).GO analysis demonstrated enrichment of biological processes related to inflammation and the general immune response (Figure 1E).Moreover, we performed gene set enrichment analysis (GSEA) comparing the microglia of animals previously irradiated and exposed to LPS treatment (radiation + LPS group) to those of animals only exposed to LPS (control + LPS group) to investigate the possible consequence of radiation-induced microglial priming, which showed suppressed regulation of neuronal synaptic plasticity and affected Wnt and BMP signaling pathways (Figure S1G).
To further corroborate our findings, we compared our data to previously published microglial gene expression signatures associated with an acute, general, or primed response (Figure 2A). 21Higher expression of acute response genes was observed in microglia isolated from LPS-treated animals, while microglia isolated from irradiated animals show a higher expression of genes associated with the primed and mainly primed gene hubs described in Holtman et al. 21At 6 weeks post irradiation, a significant increase in the percentage of microglia expressing microglial priming signature genes Clec7a and Gpnmb was detected using RNA in situ hybridization (Figures 2B and 2C) and immunofluorescent staining for GPNMB (Figures 2D-2G).
To investigate morphological changes induced by irradiation in microglia, we measured 23 morphometric parameters of microglia stained for IBA1 in the cortex (gray matter) and fimbria (white matter) regions of control and irradiated rats at 6 weeks post irradiation using our semi-automated pipeline 23 (Figures 2H, 2I, and  S2).PCA indicated notable differences in the morphology of cortical microglia between control and irradiated rats (Figure 2I).Unsupervised clustering of microglia morphologies in the cortex showed that in irradiated animals, 79% of microglia clustered separately from control microglia (Figures 2J-2L and S2A).These microglia were smaller and less ramified (Figures 2J, 2K, and S2A), morphological changes commonly associated with reactive microglia. 24,25nterestingly, the percentage of microglia in irradiated animals that showed a reactive morphology is similar to the percentages of microglia that were positive for priming signature gene mRNA or protein, suggesting that primed microglia likely display a reactive morphology (Figures 2C, 2E, 2G, and 2L).Comparable morphological differences were found in microglia located in the fimbria of control and irradiated rats (Figures S2B-S2E), indicating that microglia in both gray and white matter regions were similarly affected by irradiation.Altogether, we revealed that brain irradiation induces microglial priming, resulting in an amplified microglial immune response after a subsequent inflammatory insult.

Brain irradiation induces a persistent microglial priming gene signature
Cognitive impairment in survivors of brain tumors is considered to be irreversible and progressive in nature. 1,26To investigate the persistence of radiation-induced microglial priming and whether this response could contribute to the late radiation effects, we quantified the expression of several microglial priming genes in rat cortical tissue at different time points post irradiation by qPCR (Figure 3A).We observed increased expression of most (A) Heatmap depicting expression of hub genes associated with an acute immune response and microglial priming described in literature. 21Colors indicate column Z score on the normalized read counts across samples for each gene.
(legend continued on next page) Next, to examine whether radiation-induced microglial priming also occurs to the same extent at different radiation doses, we quantified the expression level of the same microglial priming genes in rat cortical tissue at 50 weeks post irradiation with 10, 14, and 18 Gy (Figure 3B).Most microglial priming genes were significantly upregulated in the 10 Gy irradiated animals, indicating that a relatively lower radiation dose is sufficient to induce microglial priming.The expression levels of these genes seem to increase with the radiation dose, with Gpnmb and Irf7 showing a statistically significant upregulation in rats irradiated with a higher dose (Figures 3B and S3B).
We then investigated the effect of radiation dose fractionation on microglial priming gene expression, as radiotherapy treatment in patients with brain tumors is normally delivered in multiple lower radiation doses.We irradiated rats with 5 daily fractions of 5.8 Gy and compared them to an equivalent single fraction dose of 14 Gy (Figure 3C).Overall, these two treatment schedules led to a similar increase in priming gene expression (Figures 3C and S4A).
Altogether, these data provide evidence that radiation leads to persistent microglial priming and that its extent seems to be dose dependent and not altered by radiation dose fractionation.
Juvenile rats show a less pronounced microglial priming response Survivors of pediatric brain tumors are at a high risk of developing radiotherapy-induced cognitive decline. 2 To investigate whether microglial priming can also occur in juvenile rats, we irradiated the whole brain of postnatal day 28 rats with 14 Gy and assessed the expression of microglial priming genes in cortical tissue at 12 weeks post irradiation (Figure 3D).While most priming genes were significantly upregulated in both juvenile and adult rats compared to control, the relative expression of Clec7a, Lyz2, Slamf9, and Folr2 was significantly higher in adult rats compared to juvenile rats (Figures 3D and S4B).This suggests that the effect of radiation on microglial priming is age dependent.

Microglial priming might occur in the brain of patients with GBM treated with radiotherapy
To identify the effect of radiotherapy on microglia in the human brain, we analyzed bulk (total tissue) RNA-seq data of NA brain tis-sue from patients with GBM that were treated with radiotherapy (NA-GBM; Table S4). 27GSEA of DEGs enriched in NA-GBM showed a strong association with terms related to immune responses and cytokine production, indicating the contribution of reactive microglia in patients with GBM treated with radiotherapy (Figure 4A).This was not solely the consequence of an acute microglial response, as the cluster 1 genes, associated with an acute microglial LPS response (Figure 1D), were not increased in NA-GBM brain samples (Figure S5A).In contrast, cluster 4 genes, associated with a radiation-induced response and encompassing microglial priming signature genes (Figure 1D), were more abundantly expressed in NA-GBM bulk tissue (Figure 4B).Cluster 4 gene expression of NA-GBM tissue was only significantly correlated to the microglia of irradiated rats that also received LPS (Figures 4B, S5B, and S5C), indicating that microglia in the brain tissue of patients with GBM have a transcriptomic profile that is more similar to those of rats irradiated and then exposed to LPS treatment (radiation + LPS group) than to irradiated-only rats (radiation + PBS group) (Figure S5C).This might be due to a probable inflammatory response present at the time of death, as these patients died as a result of the tumor.
The increase in priming gene expression was also found by in situ hybridization of the priming genes CLEC7A and GPNMB, which showed increased mRNA levels in microglia in NA-GBM brain samples (Figure 4C).To confirm the transcriptional data at the protein level, we performed immunofluorescent staining for GPNMB in control and NA-GBM brain samples.Colocalization of IBA1 and GPNMB was significantly increased in NA-GBM brain samples (Figures 4D and 4E).These data suggest that microglial priming might occur in the brain of patients with GBM in response to radiotherapy treatment.

DISCUSSION
Here, we revealed that radiation leads to molecular changes associated with microglial priming and validated this in NA brain tissue of patients with GBM treated with radiotherapy.Importantly, we showed that radiation-associated microglial priming is a long-lasting response, which is in line with clinical observations in patients with brain tumors of irreversible and progressive neurocognitive sequelae after treatment with radiotherapy.Recently, it has been demonstrated that the molecular state of microglia can be affected by microenvironmental changes. 14J) Hierarchical clustering on principal components (HCPCs) identified two morphological clusters: a homeostatic cluster (1) and a reactive cluster (2).Representative silhouettes of microglia in clusters 1 and 2. (K) Sholl analysis (mean ± SD) of microglia in clusters 1 and 2. (L) Relative distribution of clusters 1 and 2 in control and irradiated rats.All stainings were performed in the cortex of control and irradiated rats at 6 weeks post irradiation.Bar graphs show mean ± SEM (C, E, and G).A two-tailed t test was performed to compare the two groups.n = 4 (A), 3 (B and C), or 6 (E-L) animals per group.See also Figure S2.

Report
Studies with DNA-repair-deficient Ercc1 knockout mice showed that generic and neuron-specific deletion of Ercc1 induced microglial priming, whereas astrocyte-or microglia-specific deletion of Ercc1 did not. 20,28Neurons might be more sensitive to impaired DNA repair because they preferentially express a relatively large percentage of long genes, which generally acquire more DNA lesions than shorter genes, compared to other nonneuronal cell types. 29,30These data suggest that the cause of persistent microglial priming induced by genotoxic radiation damage could be linked to neuronal DNA damage rather than resulting from a direct effect on the microglia.
The extent of radiation-associated microglial priming might also be affected by the severity and frequency of the initial radiation stimulus.2][33] In contrast to LPS, where multiple daily doses were found to induce tolerance, 12,13 exposure to multiple lower radiation doses still resulted in microglial priming.
Due to high survival rates of more than 70%, radiotherapyinduced side effects are of particular concern in survivors of pediatric brain tumors. 34Here, we observed that radiation seems to affect the expression of microglial priming genes to a lesser extent in juvenile rats compared to adult rats.An increase in proinflammatory microglia with increasing irradiation age in rodents has

of radiation on microglial priming gene expression
Heatmaps depicting gene expression of selected microglial priming genes of (A) control and been previously reported, 35,36 with a greater impact on neurogenesis and phagocytosis at a younger age 37 and a more pronounced neuroinflammatory response at an older age. 38he transcriptomic profile of primed microglia in irradiated animals is similar to those observed in DAM microglia in amyloid mouse models 22 and in (primed) microglia in other neurodegenerative diseases such as AD. 21,39A sustained release of proinflammatory cytokines and chemokines by DAMs was postulated to mediate a positive feedback loop resulting in a persistent neuroinflammatory response that can cause neuronal damage, ultimately exacerbating the progression of neurodegenerative diseases. 40,41adiation-induced reactive microglia are known to affect neurogenesis and neuronal and synaptic plasticity. 8Our data demonstrate that microglia isolated from irradiated rats show suppressed expression of genes associated with synaptic plasticity after LPS, as well as changes in Wnt and BMP pathways, which are involved in neurogenesis and neuronal plasticity. 42,43This points toward a functional consequence of microglial priming in neurons.
In line with the animal data, we observed that genes associated with the radiation response, including microglial priming genes, were upregulated in NA brain tissue of patients with GBM compared to control, indicating that microglial priming could be induced by brain cancer treatment in humans.The notion that exposure to subsequent immune challenges could contribute to the progressive worsening of pre-existing neurotoxicity is an important aspect that should be considered clinically during the follow-up of patients with brain tumors.Patients undergo other forms of immune-modulatory treatments in concomitance or after radiotherapy, including chemotherapy and immunotherapy, which could act as subsequent immune challenges and contribute to the worsening of neurotoxicities. 44actors such as immuno-modulatory treatments, the presence of systemic infectious events, or chronic inflammatory diseases should be carefully monitored in patients that received brain irradiation.Notably, primed microglia have been hypothesized to be involved in delirium, a frequent neurocognitive complication experienced in patients with brain tumors. 45,46everal therapeutic strategies, tested in preclinical and phase 1/2 clinical trials, have been aimed at modulating the activation status and/or number of microglia and macrophages specifically in brain tumors, as tumor-associated microglia and Bar graph shows mean ± SEM.A two-tailed t test was performed to compare the two groups.n = 3 individuals per group.See also Figure S5 and Tables S4 and S5.
Cell Reports 43, 113764, February 27, 2024 7 Report ll OPEN ACCESS macrophages (TAMs) are known to facilitate tumor growth and migration. 47,487][8][9] These immunomodulatory treatments should be explored for the combined goal of reducing microglial priming and TAMs.Additionally, the temporary use of anti-inflammatory agents, such as statins, non-steroidal anti-inflammatory drugs (NSAIDs), and tumor necrosis factor blockers, during an immune challenge subsequent to radiotherapy treatment might also be beneficial. 17ogether, our findings indicate that radiotherapy treatment can shape the IIM of microglia leading to persistent priming.The knowledge that radiation makes microglia susceptible to secondary systemic immune challenges opens the way to the development of lifestyle changes and therapeutic strategies aimed at addressing systemic inflammation and hence ameliorate radiotherapy-induced neurotoxicity and its impact on patients' quality of life.

Limitations of the study
Although in the present study, no significant increase in infiltrating peripheral immune cells was found in the rat brain at 6 weeks post-14 Gy irradiation (Figures S1B and S1C), immune cell infiltration was not investigated for all conditions tested in Figure 3.Despite observing a similarly marked expression in microglial priming genes from 48 h to 50 weeks post irradiation, the possibility of other immune cells affecting the qPCR results derived from bulk brain tissues of Figure 3 cannot be excluded.An additional limitation of the animal data is that only male rats were included in the study due to technical restrictions in the irradiation setup.A possible sex difference in radiation-induced microglial priming remains to be investigated.
Infiltrating peripheral immune cells might also be present in the NA brain tissue of patients with GBM, as previously reported in two cases of radiation-induced brain injury for nasopharyngeal carcinoma treatment. 56In Figure 4B, rat microglial gene expression was compared to human bulk tissue gene expression, of which microglial cells only represent a relatively small fraction.Although this is not ideal, it resulted in a low, but remarkable, correlation (Figure 4B).Moreover, these patients received not only fractionated radiotherapy but also other treatment modalities, such as chemotherapy.Therefore, in the human samples, we cannot link the increased expression of microglial priming genes solely to the effect of radiotherapy.Whether and how chemotherapy can affect the IIM of microglia remain to be determined.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:  S4.Information on the patients' gender, ancestry, ethnicity and socioeconomic status was not available, we estimate that this will have minimal impact on the study generalizability.

METHOD DETAILS Irradiation
The whole brain of the rats (young adult males, mean weight 296 g ± SD 21) was irradiated with 14 Gy X-rays (X-RAD 320, Precision X-ray).A 14 Gy single dose is equivalent to a clinically relevant dose of 56 Gy delivered in 2 Gy fractions as calculated by the biologically effective dose equation with an a/b ratio of 2. 31 Some animals were also irradiated with 10 or 18 Gy for the dose response study.
In addition, to compare the response of a fractionated dose to a single dose, rats were irradiated with 5.8 Gy daily for five days.This fractionated dose is equivalent to a single dose of 14 Gy as calculated by the biologically effective dose equation with an a/b ratio of 2. 31 For the irradiation procedure, rats were anesthetized using isoflurane (5% induction, 1.5-2% maintenance) and hung from their teeth on a specialised holder allowing the precise irradiation of the whole brain through a dedicated collimator on the dorsal side of the animal's head. 63Juvenile male rats at postnatal day 28 (P28) were irradiated in the same way using a collimator specifically designed to irradiate the whole brain of male rats at this age.The same procedure was followed in sham-irradiated controls without irradiating the animals.

LPS immune challenge
For the LPS immune challenge experiment, rats were randomly divided into one of four groups (n = 4): sham-irradiated controls that received phosphate-buffered saline (PBS) (control + PBS) or lipopolysaccharide (LPS) (control + LPS) and irradiated animals that received PBS (radiation + PBS) or LPS (radiation + LPS).Six weeks after irradiation, rats were intraperitoneally (i.p.) injected with PBS or with 1 mg/kg LPS in PBS 13,15,25 and returned to their home cage.Rats were perfused and sacrificed 4 h after the injection to allow enough time for peripheral cytokines to reach the brain 64 and the microglia to react and change their transcriptomic profile.

Sacrifice procedure
Rats used for RNA sequencing were sacrificed 4 h after LPS or PBS injections by perfusion and termination under isoflurane anesthesia (5% induction, 1.5-2% maintenance).The rats used for qPCR analysis or microscopy were sacrificed 48 h, 6 weeks, 12 weeks or 50 weeks post irradiation.At the specified time points, these rats were perfused and sacrificed under dexmedetomidine-ketamine anesthesia.Their brains were isolated and either stored in medium for subsequent microglia isolation, dissected and snap frozen, or fixed in 4% paraformaldehyde for 48 h and embedded in paraffin.

RNA isolation and RNA sequencing
Microglial RNA was isolated using the RNeasy Plus Micro Kit (Qiagen, cat#74034) according to the manufacturer's instructions.RNA concentrations were measured using Agilent ScreenTape System.The sequencing libraries were generated by Genomescan using the NEBNext Low Input RNA Library Prep Kit from Illumina (New England Biolabs, cat#E6420 S/L).In short, cDNA was reverse transcribed from mRNA with an oligo(dT) primer and a template-switching oligo (TSO) and amplified by PCR with sequencing primers.The quality, concentration and molarity of the libraries were then measured with the Fragment Analyzer HS NGS Fragment Kits (Agilent, cat#DNF-474).All libraries were pooled equimolar and sequenced on NovaSeq6000 and NovaSeq control software NCS v1.7 and the Illumina data analysis pipeline RTA3.4.4 and Bcl2fastq v2.20 for the following image analysis, base calling, and quality check.

RNA sequencing analyses
Quality control of the generated sequencing data was performed with FastQC (v0.11.9) and FastQA (v3.1.27)followed by adapter trimming and quality filtering with fastp (v0.23.2.) 57 Alignment with the Ensembl genome Rattus norvegicus (mRatBN7.2) was performed with STAR2 (v2.7.10), 58 and transcripts were counted with HTSeq (v2.0.2). 59 The following RNA-seq analyses were performed in the R environment (v.4.1.1).After filtering genes with counts <5, normalisation, transformation, and differential gene expression analysis were performed with edgeR (v3.36.0) 60,66 and limma (v3.50.3). 61The transposed form of log counts per million (logCPM) matrix was used for the principal component analysis (PCA).Differentially expressed genes (DEGs) analysis was performed with a threshold of log 2 fold change (LFC) > 2 and adjusted p value <0.05.Gene ontology (GO) enrichment analysis and gene set enrichment analysis (GSEA) were performed by using clusterProfiler (v3.0.4). 62A isolation and qPCR Brain tissue from the anterior cortex of the rats was sectioned into 40 mm thin sections, after which RNA was isolated using the RNeasy Lipid Tissue Mini kit (Qiagen, cat#74804) according to the manufacturer's instructions.RNA was transcribed to cDNA with M-MLV reverse transcriptase (Invitrogen, cat#28025013) according to the manufacturer's instructions.qPCR was performed using iQ SYBR Green Supermix (Bio-Rad, cat#170-8885) and run in triplicate on a Bio-Rad real-time PCR system.Relative mRNA expression was calculated with DDCT method using Ywhaz as internal control.Primer sequences are listed in Table S6.

Immunofluorescence staining of human brain tissue
Five mm paraffin sections of human brain samples were dewaxed, and antigen retrieval was performed by boiling the sections for 3.5 min in HistoVT One (Nacalai tesque, cat# 06380-05).The sections were incubated with Sudan Black (0.5% in 70% EtOH) for 5 min, before they were washed and blocked in PBS+ (PBS +0.3% Triton) with 2% donkey serum and 2% bovine serum albumin for an hour.The rest of the staining procedure was performed as described above for the rat brain samples with the exception of the primary antibodies, which were diluted in PBS+ instead of PBS.

Immunohistochemistry staining
Ten mm paraffin sections of rat brain samples were dewaxed and rehydrated.Next, antigen retrieval was performed by boiling the sections for 12 minutes in 10 mM sodium citrate.For all the following washing steps, the sections were washed in PBS.Endogenous peroxidase was blocked with 0.3% H 2 0 2 for 30 minutes.After washing, the sections were blocked with 5% donkey serum in PBS+ (PBS +0.3% Triton) for 1 hour and incubated with the primary antibody rabbit anti-IBA1 (WAKO, cat#019-19741, RRID:AB_839504, 1:1000) overnight at 4 C.The samples were washed and incubated with the biotinylated secondary antibody (Jackson Immuno Research, cat#711-065-152, RRID:AB_2340593, donkey anti-rabbit, 1:400) in PBS+ for 1 hour at room temperature.After washing, the sections were incubated with avidin-biotin complex solution (VECTASTAIN ABC Kit, Vector Laboratories, cat#PK-6100) for 30 min and washed again.The sections were then stained with 0.04% 3,3'-Diaminobenzidine (Merck, cat#D5637) in PBS with 0.01% H 2 O 2 for 10 min and were subsequently dehydrated with increasing ethanol concentrations and mounted.

In situ hybridisation
In situ hybridisation was performed on 5 mm paraffin sections of rat brain and 10 mm sections of unfixed frozen human brain tissue using the RNAscope Multiplex Fluorescent v2 kit (Advanced Cell Diagnostics (ACD), cat#323110), 67 which was combined with the RNA-Protein Co-Detection Ancillary Kit (ACD, cat#323180) for the rat tissue.The manufacturer's instructions were followed with a few modifications.Positive and negative control probes were included in all experiments to assess signal-to-noise ratio.
For the rat brain, the FFPE sample preparation and pre-treatment instructions were followed, which included baking, deparaffinization and quenching endogenous peroxidases with H 2 O 2 before a 15 min target retrieval using co-detection target retrieval at 98 C-102 C. The sections were then incubated with an antibody against the protein IBA1 (WAKO, cat#019-19741, RRID:AB_839504, 1:250) overnight at 4 C.After washing with PBS-T (PBS +0.1% Tween 20), the samples were fixed in 3.7% neutral buffered formalin for 30 min and washed again with PBS-T.The sections were then incubated with protease plus at 40 C for 30 min and washed with distilled water.The probes (Clec7a-C1, cat#1260411-C1; Tmem119-C2, cat#478921-C2; Gpnmb-C3, cat#553791-C3, ACD) were hybridised for 2 h at 40 C after which pre-amplification AMP1 to AMP3 probes were hybridised sequentially.The fluorescent signal was then developed and amplified with provided HRP-conjugated antibodies against the pre-amplification probes and opal dyes (opal 690 for the Clec7a probe,1:300; opal 520 for the Tmem119 probe, 1:750; opal 570 for the GPNMB probe, 1:750, Akoya Biosciences, cat#OP-000003), one probe signal at the time.This was also done to develop the signal for IBA1 with an HRP-conjugated antibody (goat-anti Rabbit, Dako, cat#P0448, 1:500) and opal dye (opal 620, Akoya Biosciences, cat#OP-000003, 1:750).The samples were counterstained with kit-provided DAPI and mounted with Prolong Diamond antifade mounting medium (Invitrogen, cat#P36961) and stored overnight in the dark.
For the human samples, the fresh-frozen sample preparation and pre-treatment instructions were followed.This included fixation of the tissue with 3.7% formalin for 1 h at 4 C followed by washing with PBS.The sections were then dehydrated with increasing ethanol concentrations and endogenous peroxidases were quenched with H 2 O 2 .After washing, the sections were baked for 30 min at 37 C to improve tissue adherence and incubated with Protease IV for 30 min at room temperature.The rest of the protocol was performed as described for the rat sections, with the exception of the protein staining, which was not performed.The following probe and opal dye combinations were used: GPNMB-C1 (cat# 413521) and opal 570, TMEM119-C2 (cat# 478911-C2) and opal 520, CLEC7A (cat# 511761-C3) and opal 690.

Microscopy and image analysis
For quantification of the CD3 and GPNMB immunofluorescence staining, imaging was performed using a Leica DM6B microscope at 40x magnification.Five images per animal of the same region of the frontal cortex were made for the quantification of the rat microglial GPNMB expression and at least seven images per animal were made of the frontal cortex for the CD3 quantification.For the quantification of human GPNMB expression in microglia, eight images of normal-appearing brain tissue of patients with GBM and matching controls were made per patient (Table S4).Microglia, positive for both IBA1 and DAPI, were counted manually, and of these microglia, the percentage of GPNMB-positive microglia was calculated.Representative microscopic images (Figures 2 and 4) and RNAscope microscopic images (Figures 2 and S5C) of the same region were taken using the Leica SP8X DLS confocal microscope at 40x magnification.TMEM119 contrast in the representative image of the radiation group was adjusted compared to control, as TMEM119 protein expression is decreased in reactive microglia.For the microglial morphology analyses, IBA1-stained slides were user-blinded and scanned with the NanoZoomer 2HT 2.0 digital slide scanner (Hamamatsu Photonics).

Microglia morphometrics
An amount of 300 microglia in the cortex (n > 25 per animal, n = 6 animals per group) and 200 microglia in the fimbria (n > 25 per animal, n = 4 animals per group) were randomly selected for analysis.These single-cell images were then processed using a semi-automatic microglial morphology pipeline. 23This resulted in a Sholl analysis and morphometric features for each microglia.A hierarchical clustering on principal components (HCPC) approach was used to identify and compare subpopulations between experimental groups.

Statistical analysis
Statistical analyses were performed using GraphPad Prism 8 (GraphPad Software Inc.).For the comparison of the Z score mean differences of multiple groups in RNA-seq, two-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test was used.For the comparison of the mean differences of relative RNA expression of multiple groups in qPCR at different time points post irradiation, radiation doses, dose fractionation schemes and ages at the time of irradiation, two-way ANOVA followed by Sidak's multiple comparisons test with log 2 transformed data, one-way ANOVA followed by Tukey's multiple comparisons test with log 2 transformed data, one-way ANOVA followed by Sidak's multiple comparisons test with log 2 transformed data, and two-way ANOVA followed by Tukey's multiple comparisons test with log 2 transformed data were performed, respectively.F-statistics was used to test the significance of regression coefficients in linear regression models in the R environment (v.4.1.1).For the comparisons of percentages of GPNMB-or Clec7a-positive microglia, a two-tailed T test was used.Experimental statistical details can also be found in the relative Figure legends.We defined a p value less than 0.05 as statistically significant.

Figure 1 .Figure 2 .
Figure 1.RNA-seq of microglia from irradiated rats subsequently challenged with LPS (A) Schematic overview of the experiment.(B) Principal-component analysis (PCA) of gene expression profiles of microglia in each condition.(C) Numbers of DEGs (log 2 fold change [LFC] > 2, adjusted p value < 0.05) in the comparisons between conditions.(D) Heatmap depicting gene expression of all the DEGs detected in the 4 comparisons in (C).Hierarchical clustering resulted in 5 clusters of genes based on the expression pattern.Colors indicate row Z score on the normalized read counts across samples for each gene.(E) Dot plot depicting Gene Ontology (GO) enrichment analysis (top 5 enriched functions of biological processes [BPs]) of the genes in clusters 1, 4, and 5.(F) Boxplots depicting the mean Z score of genes in cluster 1 for each condition.Boxes are from the first quartile to the third quartile, and lines indicate the median.n = 4 animals per group.***p < 0.001 and ****p < 0.0001.Two-way ANOVA followed by Tukey's multiple comparisons test was used for the comparison of the mean differences of multiple groups.See also FigureS1and TablesS1 and S2.

(
B and C) Representative in situ hybridization images of mRNA from priming genes Gpnmb and Clec7a and microglial markers Tmem119 (mRNA) and IBA1 (protein) (B) and quantification of Gpnmb and Clec7a expression in Tmem119 and IBA1 double-positive cells (C).(D and E) Representative images of GPNMB and IBA1 (D) and quantification of the percentage of GPNMB-positive cells within IBA1-positive cells (E).(F and G) Representative images of GPNMB and TMEM119 (F) and quantification of the percentage of GPNMB-positive cells within TMEM119-positive cells (G).TMEM119 contrast in the radiation group's representative image was adjusted compared to control, as TMEM119 expression is decreased in reactive microglia.(H) Representative images showing the morphology of IBA1-positive microglia.(I) PCA plot showing individual microglia (n = 25 per animal) of control and irradiated animals.The lines show the 95% confidence ellipses with the weighted centers of mass per group in the middle.

Figure 3 .
Figure 3. Persistence and dose-response effect of radiation on microglial priming gene expression

Figure 4 .
Figure 4. Microglial priming might occur in the brain of patients with GBM treated with radiotherapy (A) GSEA (BP) dot plot of DEGs (FC > 1.5, false discovery rate [FDR] < 0.05, RNA-seq data derived from Ainslie et al. 27 ) enriched in patients with GBM treated with radiotherapy (NA-GBM) compared to neurotypical control individuals.The list of DEGs can be found in Table S5.n = 5 individuals per group.(B) Scatterplot depicting a correlation of gene expression enrichment of cluster 4 genes in Figure 1D between the rat microglia-specific transcriptomic dataset (control + PBS vs. radiation + LPS) and the human bulk transcriptomic dataset (control vs. NA-GBM).F-statistics were used to test the significance of regression coefficients in the linear regression model.(C) Representative in situ hybridization images of mRNA from priming genes GPNMB and CLEC7A and microglial marker TMEM119.(D and E) Representative images of GPNMB protein expression and microglial marker IBA1 (D) and quantification of the percentage of GPNMB-positive cells within IBA1-positive cells in control and NA-GBM postmortem brain tissues (E).Bar graph shows mean ± SEM.A two-tailed t test was performed to compare the two groups.n = 3 individuals per group.See also Figure S5 and TablesS4 and S5.

TABLE
d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability B In situ hybridisation B Microscopy and image analysis B Microglia morphometrics d QUANTIFICATION AND STATISTICAL ANALYSIS B Statistical analysis description of the included patient samples can be found in Table