Transcriptional Remodeling of Microglia After Experimental Myocardial Infarction

Abstract

Beyond cardiac impairment, myocardial infarction (MI) affects the central nervous system (CNS), where it has been associated with neuroinflammation and cognitive dysfunction. Microglia, the resident immune cells of the CNS, are key regulators of neuroinflammatory processes. However, the transcriptional landscape of microglia following MI remains incompletely understood. We hypothesized that MI induces transcriptional remodeling in microglia that may reflect altered metabolic regulation. Male C57BL/6J mice underwent permanent LAD ligation or sham surgery. Five days post-MI, CD45-intermediate and SiglecH/CD11b-positive immune cells were isolated from cortical and subcortical regions by FACS and subjected to single-cell RNA sequencing. Complementary exploratory metabolic assays included assessment of mitochondrial mass and membrane potential as well as glucose uptake. Microglia represented the predominant immune cell population in both the cortex and subcortex. Subclustering revealed a significantly increased proportion of a “low translational” microglial subset after MI. Pseudobulk differential expression and gene set enrichment analyses demonstrated significant downregulation of translation-related pathways in cortical microglia and proteostasis-associated pathways in subcortical microglia. These transcriptional changes were accompanied by a significant reduction in mitochondrial mass and metabolic observations consistent with altered energetic regulation, although several functional readouts did not reach statistical significance. Experimental MI is associated with region-specific transcriptional remodeling of microglia, characterized by reduced expression of energy-intensive and proteostasis-related pathways. Exploratory metabolic observations are consistent with altered energetic regulation but require confirmation in adequately powered studies. These findings suggest that systemic cardiac injury is linked to microglial transcriptional adaptation in the early post-infarction phase.

1. Introduction

Myocardial infarction (MI) remains a leading cause of mortality worldwide, posing a major burden on public health [1]. While traditionally regarded as a heart-specific condition, it is now widely accepted that MI leads to systemic effects, which can also affect the central nervous system (CNS) [2,3,4]. In this regard, acute and chronic neuroinflammation have been proposed as pathophysiological correlates of MI-induced cognitive decline, depression and anxiety [5]. Neuroinflammation is critically mediated by microglia, the resident immune cells of the CNS, by maintaining CNS homeostasis and orchestrating immune responses [6,7]. Upon stimuli, microglia undergo a morphological transformation, proliferate and secrete cytokines [6]. Emerging evidence emphasizes region-specific differences in microglial identity [8]. Subcortical regions such as the hypothalamic paraventricular nucleus (PVN) regulate sympathetic outflow, and increased sympathetic tone contributes to heart failure progression after MI [9,10]. In line with this, dysregulated sympathetic signaling is a hallmark of HF and a major target of current therapies [11].

After MI, microglial activation and morphological alterations have been observed in the PVN in rats 4–16 weeks after MI, but not within the first 7 days [12,13]. Other studies reported increased reactive oxygen species and a shift towards a proinflammatory microglial phenotype in the mouse brain after MI [14]. Moreover, microglia exhibited an activated transcriptional profile involving enhanced chemotactic features as early as 3 days post-MI [3]. Our findings at day 5 do not contradict these observations but rather suggest temporal dynamics in microglial adaptation. While early phases may be characterized by inflammatory activation, the suppression of translation-related pathways observed at day 5 may reflect a subsequent metabolically adaptive or energy-conserving state following initial activation. These data support the concept of phase-dependent microglial reprogramming after MI [2]. However, the precise molecular mechanisms driving microglial responses, particularly at the transcriptional level, as well as their temporal dynamics following MI, remain incompletely understood [10].

In addition to local myocardial injury, MI induces systemic sterile inflammation, endothelial dysfunction, cytotoxic signaling, and hypoxia-related mitochondrial stress [1]. Processes promote the release of inflammatory mediators and metabolic stress signals that can affect distant organs, including the brain [6]. As microglial function is closely linked to intracellular metabolism, exposure to post-infarction inflammatory and hypoxic conditions may influence metabolic regulation of CNS-resident immune cells [15]. Such mechanisms provide a plausible link between systemic cardiac injury and microglial functional adaptation.

Thus, in this study, we characterized microglial responses 5 days after MI by single-cell RNA sequencing of cortical and subcortical regions. To explore potential functional implications of the observed transcriptional changes, we complemented transcriptomic profiling with exploratory metabolic assays, as immunometabolism critically shapes microglial activity [15].

2. Results

Mice underwent MI procedures as part of a training program. Based on the presence of a large myocardial infarct, mice were assigned to the MI group, while those without evidence of infarction were assigned to the Sham group (Figure 1). Five days post-surgery, brain tissue was collected, and cortical and subcortical regions were processed separately. Single-cell suspensions were prepared, and CD45⁺ immune cells were isolated by fluorescence-activated cell sorting (FACS). These cells were subsequently analyzed by single-cell RNA sequencing (scRNA-seq), enabling a detailed characterization of immune cell responses following MI.

2.1. Clustering of scRNA-Seq Data Identifies Microglial Subpopulations as the Predominant CNS-Resident Immune Cells

Bioinformatical analysis of cortical and subcortical single-cell suspension after MI vs. Sham identified 14 (0–13) different cell populations (Figure 2a). Integration of all samples into a single Seurat object enabled us to analyze specific subgroups independently. Information on experimental condition (MI vs. Sham) and tissue origin (cortex vs. subcortex) was retained in the metadata and therefore accessible throughout the entire analysis (Figure 2b). Cell-type annotation was performed using SingleR followed by expert-guided refinement based on cluster-specific marker genes. For each cluster, the top expressed genes according to FindAllMarkers-function were inspected to confirm and adjust the automated annotations. (Figure 2c).

As intended, following pre-FACS sorting for microglia and immune cells (CD45 and SiglecH), the dataset was enriched for immune cells. Microglia represented the dominant population and were therefore selected as the primary focus of downstream analyses. UMAP shows a clear separation of different immune cell populations. Microglia were subdivided into five subclusters by initial Seurat clustering (designated as: 0: “Homeostatic Microglia”, 1: “Inflammatory Microglia”, 2: “Low Translational Microglia”, “Low Ribosomal Reactive Microglia”; and 9: “Activated Microglia (Interferon)”).

Keeping in mind that cortical and subcortical tissue underwent identical preprocessing and bioinformatic quality control, Figure 2b illustrates that the number of cells passing quality thresholds differed substantially between cortex and subcortex (cortex: 11,185; subcortex: 5243). In contrast, the total number of cells obtained from MI and Sham conditions was approximately equal (Sham: 8122; MI: 8702), reflecting equal treatment and comparable cell yields across individual samples.

High Inter-Individual Variability in Cell-Type Composition, with Microglia Showing the Majority of DEGs After MI

After evaluating our primary dataset, we next investigated the composition of different cell populations within the samples. Considerable inter-individual variability in cell-type proportions was observed (Figure 3). Different clusters bear a wide range of differences between single samples. For instance, the subcortical sample ‘S5’ displayed a notably lower number of “Homeostatic Microglia” compared to other individuals (Figure 3b). While individual variability in cluster proportions was observed—particularly in the inflammatory microglial subset in the subcortex—these differences did not reach statistical significance at the group level, except for the increase in the “low translational” microglial subset. When comparing mean abundances across microglial subtypes, we observed significant differences between MI and Sham conditions. In both the cortex and subcortex, a distinct microglial subpopulation characterized by low ribosomal and high mitochondrial gene expression—referred to as Low Translational Microglia (Jun+, Fosb+, Ptpro+, Rpl37a-, Rps9-)—was more abundant after MI (cortex: MI 20.9% vs. Sham 15.9%, p < 0.001; subcortex: MI 20.1% vs. Sham 16.7%, p = 0.002; Wald H₀ test). This enrichment may reflect a subtle shift towards a less translationally active microglial phenotype in response to MI.

In the next step, we performed differential expressed gene (DEG) analysis between MI and Sham within each individual cluster, separately for cortical and subcortical tissues. To account for inter-individual variability and the biological dependence of cells within a sample, DEG analysis was conducted using DESeq2 [16].

Within the FACS-enriched immune cell compartment, microglial subpopulations exhibited the majority of transcriptional changes following MI (Figure 3c). Based on this observation, we grouped all microglia-defined clusters (“Homeostatic Microglia,” “Inflammatory Microglia,” “Low Translational Microglia,” “Low Ribosomal Reactive Microglia,” and “Activated Microglia (Interferon)”) and performed DEG analysis across the combined microglia population to capture transcriptional changes at a broader level. This analysis revealed 397 differentially expressed genes (DEGs) in cortical microglia and 302 DEGs in subcortical microglia. In comparison, “Activated Macrophages” displayed 60 DEGs in the subcortex, but only 3 in the cortex. Other cortical cell types showed minimal to no transcriptional alterations (“Ribosomal active B cells”: 9 DEGs). In the subcortex, DEG counts for additional immune cell types were as follows: “Activated Monocytes”: 15, “Activated T cells”: 1, “Inflammatory Granulocytes”: 1, and “Mature B cells”: 3.

2.2. Cortical Microglia Show Translational Decline After MI

To elucidate the biological relevance of DEGs identified in the pseudobulk analysis (MI: n = 6 vs. Sham: n = 5), we performed Gene Set Enrichment Analysis (GSEA) using the ClusterProfiler R package. DEGs were ranked according to both statistical significance and magnitude of differential expression (log2-fold change). A heatmap displaying the expression patterns of all DEGs from this analysis is shown in Figure 4a. Utilizing Gene Ontology Biological Processes (GOBP) [17,18], Hallmark gene sets (H) [19], and Reactome pathways (R) [20] as reference databases, we identified 13 biological processes significantly enriched among downregulated DEGs in cortical microglia following MI (Figure 4b), including: “R: Translation” (R_TRANSLATION), “R: Eukaryotic translation initiation” (R_EUKARYOTIC_TRANSL_INI), “R: Nonsense mediated decay NMD independent of the exon junction complex EJC” (R_NMD_INDEP_OF_EJC), “R: Nonsense mediated decay NMD” (R_NMD), “R: Major pathway of rRNA processing in the nucleolus and cytosol” (R_RRNA_PROC_NUCLEO_CYTO), “GOBP: Cytoplasmic translation” (GO_CYTOPLASMIC_TRANSL), “GOBP: Ribosome biogenesis” (GO_RIBOSOME_BIOGENE), “R: Activation of the mRNA upon binding of the cap binding complex and EIFS and subsequent binding to 43S” (R_MRNA_ACTIVATION_43S), “H: TNFA signaling via NFKB” (H_TNFA_SIGNALING_NFKB), “GOBP: Translation at synapse” (GO_TRANSLATION_AT_SYN), “GOBP: Ribosomal small subunit biogenesis” (GO_RIBOSOMAL_SUBUNIT_BIO), “GOBP: Ribosome assembly” (GO_RIBOSOME_ASSEMBLY), and “GOBP: Ribosomal small subunit assembly” (GO_RIBOSOMAL_SUBUNIT_AS).

Across all, GSEA flags especially translational machinery-related terms as strongly downregulated in microglia after MI. Immersing into the involved genes, e.g., noted by “Reactome: Translation”, we identified 33 DEGs containing structural ribosomal genes (e.g., “Rpl31”, “Rps2, “Rpl26”), elongation factors (e.g., “Eef1b2”, “Eef1d”) and one initiation factor (“Eeif5b”). Taken together, MI leads to a suppression of translational activity in cortical microglial cells. To quantify this displacement, we assembled these 33 DEGs into a “translation-score” (per-cell average of normalized expression) and compared this score between MI and Sham on a per-sample basis (Figure 4c). MI samples trended to a lower translation-score compared to Sham samples, consistent with reduced translational pathway enrichment at the transcriptomic level. However, this comparison did not reach statistical significance (p-value = 0.052, Wilcoxon rank sum exact test).

Heading forward to GSEA of MI-upregulated DEGs, these 4 terms have been found to be significantly concerned: “GOBP: Epithelial cell differentiation” (GO_REG_OF_EPITHELIAL_C), “GOBP: Neutral lipid metabolic process” (GO_NEUTRAL_LIPID_META), “GOBP: Regulation of epithelial cell differentiation” (GO_REG_OF_EPITHELIAL_C), and “GOBP: Lens fiber cell differentiation” (GO_LENS_FIBER_CELL_DIFF). These findings suggest that the terms of MI-upregulated DEGs reflect a combination of differentiation-related processes and a shift towards lipid metabolism, potentially indicating a metabolic transition in cortical microglia.

2.3. Subcortical Microglia Exhibit Signs of Proteostasis Impairment After MI

In parallel, subcortical scRNA-seq data was analyzed following the same workflow. Among the differentially expressed genes (DEGs) downregulated after MI (compared to Sham; Figure 5a), enrichment analysis identified several significantly associated pathways and processes, including: “R: Cellular responses to stimuli” (R_RESP_TO_STIMULI), “GOBP: Establishment of protein localization to organelle” (GO_PROT_TO_ORG_LOCAL), “R: Cytokine signaling in immune system” (R_CYTOKINE_SIGNALING), “R: Class I MHC mediated antigen processing presentation” (R_MHC_CLASS_I), “GOBP: Protein folding” (GO_PROTEIN_FOLDING), “H: TNFA signaling via NFKB” (H_TNFA_SIGNALING_NFKB), “H: p53 pathway” (H_P53_PATHWAY), “H: mTORC1 signaling” (H_MTORC1), “GOBP: Response to topologically incorrect protein” (GO_RESPONSE_MISFOLD), “R: HSP90 chaperone cycle for steroid hormone receptors SHR in the presence of ligand” (R_HSP90_SHR_LIGAND), “R: Cellular response to heat stress” (R_HEAT_STRESS), “GOBP: De novo protein folding” (GO_DE_NOVO_FOLDING), “GOBP: Chaperone mediated protein folding” (GO_CHAPERONE_FOLDING), “GOBP: Chaperone cofactor dependent protein refolding” (GO_COFACTOR_REFOLDING), “R: HSF1 dependent transactivation” (R_HSF1_TRANSACTIVATION), “R: DDX58 IFIH1 mediated induction of interferon alpha beta” (R_IFN_ALPHA_BETA), “R: Attenuation phase” (R_ATTENUATION_PHASE), and “GOBP: Protein refolding” (GO_PROTEIN_REFOLDING).

Following myocardial infarction, microglia show signs of reduced protein folding. Downregulation of pathways related to stress response, proteostasis, and protein quality control suggests a diminished ability to maintain protein homeostasis. Thus, terms including “Protein Folding” and “Chaperone mediated protein folding” are affected. Therefore, we calculated a module score based on DEGs associated with the “GOBP: Protein folding” term, demonstrating a reduced capacity for protein folding in subcortical microglia following MI (p-value = 0.004, Wilcoxon rank sum exact test) (Figure 5c).

In contrast, analyses of upregulated DEGs after MI, based on our stringent selection criteria and pseudobulk analysis, revealed no significantly enriched terms, suggesting a lack of strong biological coherence among these genes.

2.4. Mitochondrial and Metabolic Measurements Following MI

Based on the transcriptomic findings indicating reduced translation-related and proteostasis-associated pathways, we next explored whether complementary metabolic measurements would reveal alterations consistent with these observations. Given that translation is a particularly energy-intensive process and closely linked to metabolism [21,22], we then performed complementary metabolic assays to explore whether the observed decrease in translational (protein biosynthesis) activity is accompanied by alterations in cellular energy metabolism. Brain-derived microglia in a single-cell suspension were analyzed using MitoTracker staining to assess mitochondrial content per cell [23]. Microglia from MI animals displayed significantly reduced MitoTracker fluorescence 5 days post-MI, indicating decreased mitochondrial mass and suggesting altered mitochondrial content in the post-infarction state (Figure 6a). To assess mitochondrial membrane potential, we then performed tetramethylrhodamine ethyl ester (TMRE) staining [24]. After normalizing TMRE signals to MitoTracker to account for mitochondrial mass, there was a trend toward decreased membrane potential in MI microglia; however, this did not reach statistical significance (Figure 6b). These findings therefore do not support a robust alteration in membrane potential at this time point.

Following the assessment of mitochondrial membrane potential using MitoTracker and TMRE, we next investigated cellular glucose uptake via 2-NBDG staining, shifting our focus towards metabolic activity and the utilization of basic nutritive substrates. This was motivated by the rationale that impaired mitochondrial respiration could be compensated by increased glycolysis and that activated microglia characteristically upregulate glucose consumption [25]. Although no significant differences were observed between the groups, MI animals exhibited lower 2-NBDG fluorescence intensity (Figure 6c), representing a non-significant trend toward reduced glucose uptake.

In addition to the aforementioned metabolic assays, we then performed SCENITH (Single-Cell Energetic Metabolism by Profiling Translation Inhibition) to further investigate metabolic states of microglia after MI (Figure 6d–g). SCENITH profiling of microglia revealed no statistically significant differences between Sham and MI. However, microglia from MI animals exhibited a trend towards lower glucose dependency, suggesting a possible shift in substrate utilization. Mitochondrial dependency and glycolytic capacity were comparable between Sham and MI, while fatty acid oxidation (FAO) and amino acid oxidation (AAO) capacities were slightly elevated in the MI group. These findings do not demonstrate statistically significant differences between groups but show trends that are directionally consistent with the transcriptomic profile. Due to limited statistical power, these observations should be interpreted cautiously. Interestingly, the increased expression of genes associated with the ‘neutral lipid metabolic process’ observed in cortical microglia after MI (Figure 4, GSEA) was functionally supported by SCENITH profiling, which revealed a trend towards a slightly elevated FAO capacity in the MI group. Together, these observations are consistent with a possible shift in substrate preference; however, given the absence of statistical significance, this interpretation remains exploratory.

3. Discussion

Microglia are essential immune and homeostatic regulators in the CNS, and their functional state is closely linked to intracellular metabolism [26]. In the present study, single-cell analysis revealed significant alterations in translation-related pathways within cortical microglia 5 days after MI. In parallel, subcortical microglia exhibited reduced expression of proteostasis-associated pathways, indicating region-specific transcriptional responses. These transcriptional changes were accompanied by a significant reduction in mitochondrial mass and metabolic observations that were directionally consistent with altered energetic regulation, although several functional readouts did not reach statistical significance. Together, our findings demonstrate region-specific transcriptional remodeling of microglia, with exploratory metabolic observations requiring further validation.

Recent studies have shown that microglia actively respond to MI within the CNS, including the recruitment of peripheral immune cells and the establishment of a pro-inflammatory microenvironment [3]. Various therapeutic approaches targeting microglia—such as PLX3397-mediated depletion or minocycline treatment—have been explored with respect to cardiac outcomes, underscoring the clinical relevance of modulating microglial function after MI [4,27]. MI has also been shown to activate microglia in key cardiovascular regulatory centers in the brain, such as the rostral ventrolateral medulla and nucleus tractus solitarius [27]. However, the downstream functional consequences of this activation remain poorly understood [6].

Our data suggest that the observed transcriptional alterations may reflect changes in metabolic regulation; however, functional metabolic measurements in this study should be considered exploratory. We observed significantly reduced mitochondrial staining in cerebral microglia 5 days after MI. One of the core functions of mitochondrial activity lies in oxidative phosphorylation, the primary source of intracellular ATP production [28]. While non-activated microglia primarily rely on oxidative phosphorylation, activated microglia often shift towards glycolysis [29]. Microglia are known for their high metabolic flexibility, adapting mitochondrial function in response to inflammatory activation [30]. In line with this, inflammation-activated microglia have been shown to undergo reduced mitochondrial function [31]. Our data similarly show reduced mitochondrial staining in MI microglia, indicating decreased mitochondrial mass after MI. Whether this reflects functionally relevant changes in mitochondrial activity requires further investigation.

Immune metabolism provides a mechanistic interface between myocardial infarction and neuroinflammation. Systemic consequences of MI—including inflammatory mediators, neurohumoral activation, altered substrate availability, and stress signaling—can reach the brain and influence the metabolic state of microglia [3,4]. Because immune cell effector functions are tightly coupled to intracellular metabolism, such metabolic reprogramming directly shapes the capacity of microglia to regulate inflammatory signaling, protein synthesis, and stress adaptation [26]. In our study, MI was associated with reduced mitochondrial mass and suppression of translation-related pathways, while additional metabolic measurements showed non-significant trends consistent with altered substrate utilization. These findings support the presence of microglial transcriptional adaptation after MI; however, direct functional consequences for inflammatory tone or CNS homeostasis cannot be inferred from the present data. Thus, immune metabolism may function as a bridge translating peripheral organ damage into region-specific neuroimmune alterations.

The metabolic alterations observed after MI share conceptual similarities with microglial changes described in ischemic stroke, where cerebral ischemia directly induces mitochondrial dysfunction, altered substrate utilization, and inflammatory reprogramming [32]. However, in contrast to stroke, where ischemia occurs within the brain parenchyma, MI represents a peripheral ischemic event that secondarily influences the CNS through systemic inflammatory and metabolic signaling [13]. Thus, while ischemia-related stress may represent a shared mechanistic theme, the initiating triggers and spatial dynamics differ substantially between cardiac and cerebral ischemia.

In addition to mitochondrial-related observations, we identified a pronounced transcriptional shift characterized by downregulation of genes involved in protein synthesis. The consistent downregulation of genes involved in protein synthesis indicates a coordinated suppression of translation following MI. These findings point to superordinated regulatory processes, driving diverse functional adaptations beyond single-pathway regulation. This may also reflect a functional adaptation towards an energy-conserving phenotype, as protein synthesis is among the most energy-consuming cellular processes [21].

Notably, similar patterns of metabolic reprogramming have been reported in microglia exposed to amyloid-β (Aβ). Initially, these cells adopted a pro-inflammatory phenotype characterized by elevated glycolysis, which then transitions into an innate immune-tolerant state within 5 days in vitro [33]. This later phase is marked by impaired glycolysis and diminished oxidative phosphorylation [33]. In our study, reduced mitochondrial staining was accompanied by non-significant trends toward decreased glucose uptake and dependency. While these parallels are conceptually interesting, the metabolic observations in our dataset remain exploratory and require confirmation in adequately powered studies. This suggests that both Aβ and MI might act as stressors capable of inducing comparable microglial metabolic adaptations.

The present study does not allow us to distinguish whether microglial alterations are directly triggered by myocardial damage or indirectly mediated through systemic post-infarction processes. We consider the latter more likely, as MI induces profound systemic inflammatory, neurohumoral, and metabolic changes that can affect the CNS. Therefore, the observed microglial transcriptional remodeling should be interpreted as a secondary response within the context of systemic cardiac injury rather than as evidence of direct heart–brain signaling.

Limitations: It remains unclear whether the observed metabolic alterations represent a transient response to acute post-infarction stress or persist long-term. As our study focuses on an early time point (day 5), the findings likely reflect acute systemic signaling. Future longitudinal studies are required to determine whether microglial metabolic reprogramming resolves or contributes to sustained neuroinflammatory remodeling after MI. Moreover, although mitochondrial mass was significantly reduced, several metabolic readouts (including glucose uptake, mitochondrial dependency, and FAO capacity) showed consistent trends without reaching statistical significance. Functional metabolic conclusions therefore cannot be drawn from the present dataset and warrant validation in studies with increased biological replicates. Furthermore, bioinformatic outcomes are influenced by analytical strategy. Here, we applied a pseudobulk approach, treating each biological replicate as a single unit, which may emphasize biological variation given the modest sample size (6 MI vs. 5 Sham mice), but potentially limits the resolution of cell-type-specific heterogeneity. Furthermore, transcriptomic and metabolic measurements were conducted on separate sets of animals, which may introduce additional variability.

4. Materials and Methods

4.1. Animals

C57BL/6J 8–12-week-old male mice were housed in groups of 2–5 per cage under controlled environmental conditions, including an ambient temperature of 22.5 ± 2 °C, relative humidity of 50 ± 5%, and a 14/10 h light/dark cycle (lights on at 7:00 a.m.). Mice had ad libitum access to standard rodent chow (sniffSpezialdiäten, Soest, Germany) and water. All experimental procedures were conducted in accordance with the Animal Protection Law and the Directive 2010/63/EU of the European Parliament and Council. For this study, brains were obtained from animals of a training project in accordance with the 3R principle (approval number RUF-55.2.2-2532-2-1379 on 11 May 2021).

4.2. MI and Sham Operation

Surgery was carried out according to the established protocol in our department [34,35]. Briefly, mice were anesthetized with isoflurane (about 2.0 vol.%), intubated, and mechanically ventilated. After left thoracotomy, the left anterior descending (LAD) coronary artery was visualized. Afterwards, MI was induced by permanent ligation of the LAD. For perioperative analgesia, buprenorphine (0.002 mg per mouse) was administered intraperitoneally.

4.3. Echocardiography

Transthoracic echocardiography was performed using a high-frequency ultrasound system (Vevo 1100, VisualSonics, Amsterdam, The Netherlands) in order to assess cardiac function and verify successful MI. Parasternal long- and short-axis views were obtained to measure left ventricular dimensions and function under light isoflurane anesthesia.

4.4. Single-Cell Suspension

Five days after surgery, the mice were sedated with CO₂ and subsequently euthanized by cervical dislocation. Following transcardial perfusion with ice-cold PBS (supplemented with 1% heparin), murine brains were harvested. Brains were weighed and dissected on ice to separate cortex and subcortex. Tissue was minced and incubated with Accutase^® (23 °C, 15 min, 550 rpm) in a gentle shaker and not centrifuged again after the digestion step. The resulting suspension was filtered through a 40 µm mesh and washed with cold HBSS −/−. Cells were centrifuged (4 °C, 10 min, 500× g) using minimal brake settings. For myelin removal, a 40% Percoll gradient was prepared (comprising 10× HBSS, Percoll, and 1× HBSS) and mixed with the cell pellet. Samples were centrifuged (21 °C, 25 min, 800× g) without brake, so that myelin layer could be carefully removed providing cortical and subcortical single-cell suspension.

4.5. Flow Cytometry

After generating single-cell suspension, cell pellets were resuspended in FACS buffer containing 1:100 Fc block and incubated for 10 min at 4 °C in the dark. Cells were then stained with antibodies targeting CD45 (APC anti-mouse CD45.2 Antibody, BioLegend, San Diego, CA, USA; 109814), SiglecH (PE anti-mouse Siglec H Antibody, BioLegend, 129605), and a viability dye (Zombie Aqua™ Fixable Viability Kit, BioLegend, 423101). Hashtag antibodies (HTOs) were added at this timepoint to enable sample allocation. After 30 min of incubation at 4 °C in the dark, cells were washed, centrifuged (4 °C, 7 min, 400× g), and resuspended in FACS buffer. Immune cells have been sorted by gating for CD45 and SiglecH. Cytometric analyses were performed with FlowJo (BD Biosciences, San Jose, CA, USA). Cells were sorted using a BD FACS Aria III with a 100 µm nozzle.

4.6. Mouse scRNA-Seq

For single-cell RNA sequencing (scRNA-seq), SiglecH- and CD45-positive cells from murine cortex and subcortex were processed. Hashtag oligonucleotide (HTO) antibody tagging enabled sample allocation. The library preparation was performed using the 10x Genomics Chromium Next GEM Single Cell 3′ Library and Gel Bead Kit (v3.1). Single cell reverse transcription was performed using 10x Genomics Chromium technology. Sequencing was carried out on an Illumina Novaseq 6000 System, Illumina, San Diego, CA, USA. Raw sequencing data for cortex and subcortex samples were processed using CellRanger v7.0.1 with the mm10-2020-A reference genome. The cortex dataset contained 17,604 cells with a median of 2579 genes per cell and 21,818 reads per cell, while the subcortex dataset contained 7503 cells with 3208 genes per cell and 51,996 reads per cell. Sequencing saturation reached 51.8% (cortex) and 68.2% (subcortex), and the proportion of reads assigned to cells was 89.5% and 84.3%, respectively. Following data processing and further analysis was performed in R using the Seurat package [36].

4.7. Bioinformatics

After generating Seurat objects in R using the Seurat package v5.2.1 [36], sample demultiplexing was performed with the HTODemux function based on hashtag oligonucleotide (HTO) barcodes, enabling sample allocation of 6 MI and 5 Sham samples. The same preprocessing pipeline was applied to both cortical and subcortical datasets. Standard quality control steps included filtering out cells with high mitochondrial gene expression (>10%) and retaining cells with a gene count between 200 and 5000, in order to remove low-quality cells.

Subsequent analysis followed the standard Seurat workflow, including normalization (NormalizeData), identification of highly variable features (FindVariableFeatures), data scaling (ScaleData), principal component analysis (RunPCA), and clustering (FindNeighbors and FindClusters). Potential doublets were identified and removed using the DoubletFinder package [37]. Since cortical and subcortical samples were generated in separate batches, initial preprocessing was conducted independently. The resulting Seurat objects were then merged using the merge function, followed by batch correction using Harmony, accounting for both tissue origin and experimental condition (MI vs. Sham) [38]. Dimensionality reduction was performed via RunUMAP based on the first 20 principal components [39].

Cell-type annotation was based on Seurat-derived cluster identities and differential gene expression profiles obtained using the FindMarkers and FindAllMarkers functions. Manual assignment of cell types was guided by known marker genes. To support the annotation process and enhance consistency, we additionally applied the SingleR package (SingleR v2.8.0) for automated reference-based cell-type prediction [40].

For the in-depth characterization of subclustered microglia, initially indicated Seurat clusters, were further annotated based on the expression of specific marker genes. These key genes were used to gain a more detailed understanding of microglia’s diverse functional states: “Homeostatic Microglia” (Gpr34⁺, P2ry12⁺, Trem2⁺; Fos⁻, Atf3⁻), “Inflammatory Microglia” (Socs3⁺, Cd83⁺, Casp4⁺), “Low Translational Microglia” (Jun⁺, Fosb⁺, Ptpro⁺, Rpl37a⁻, Rps9⁻), “Low Ribosomal Reactive Microglia” (mt-Co1⁺, Malat1⁺, Rpl27⁻, Rpl35⁻), and “Activated Microglia (Interferon)” (Rtp4⁺, Slfn5⁺, Ifit3⁺, Ifit2⁺). To visualize specific markers across clusters, a heatmap was generated using scaled expression values for the top 20 marker genes per cluster, as identified by FindAllMarkers.

To enable region- and condition-specific comparisons, cortical and subcortical cells from MI and Sham groups were extracted using the subset function in Seurat. Differential gene expression (DEG) analysis was performed using a pseudobulk strategy combined with DESeq2, where counts were aggregated per biological replicate to account for variability in cell numbers and sample origin [16,41]. Features with an adjusted p-value < 0.05 were considered differentially expressed (DEGs).

Gene set enrichment analysis (GSEA) was conducted using the ClusterProfiler package [41]. DEGs were tested against the following gene set collections: Gene Ontology Biological Processes (GOBP) [17,18], Hallmark (H) [19], and Reactome (R) [20]. GMT files were obtained from the Molecular Signatures Database (MSigDB). Gene sets were considered significantly enriched if they met the following criteria: q-value < 0.2, adjusted p-value < 0.05, and included at least three genes per term.

The microglial translational state was evaluated by calculating a module score based on the “REACTOME_TRANSLATION” gene set. For this, we restricted the analysis to differentially expressed genes (DEGs) identified in our study that overlapped with the gene list annotated under the “R: Translation” term, as identified through ClusterProfiler enrichment analysis. The same procedure was applied to assess the microglial proteostasis state, using the “GOBP: Protein folding” gene set. Only DEGs from our dataset that overlapped with this gene ontology term, as identified via ClusterProfiler, were included in the module score calculation.

4.8. Glucose Uptake

Glucose uptake by microglia was measured using the fluorescently labelled glucose analog 2-NBDG (Biomol, 11046) as previously described [42]. Briefly, the isolated and digested CNS cell suspension was starved for 15 min at 37 °C in glucose-free RPMI 1640 medium (Roth, 9094.1) in a V-bottom 96-well plate containing up to 3 × 10⁵ cells per well. Afterwards, cells were incubated with 300 µM 2-NBDG in glucose-free RPMI medium for 30 min at 37 °C, washed once with cold PBS, and processed further for flow cytometry analysis. For staining, the cells were incubated with CD45-BV605 (BioLegend, 103139) and CD11b-PE (BioLegend, 101208) in PBS containing 0.5% BSA for 20 min at RT and then resuspended in PBS containing 0.5% BSA and 0.5 µL DAPI for dead-cell exclusion. Microglia were identified via CD11b and intermediate CD45 expression.

4.9. Single-Cell Energetic Metabolism by Profiling Translation Inhibition (SCENITH)

SCENITH was performed as previously described [43]. Briefly, the cell suspension was resuspended in RPMI 1640 standard medium (Gibco, Grand Island, NY, USA, 11875093), supplemented with 10% fetal calf serum (Sigma, Tokyo, Japan, 12133), 100 units/mL penicillin (Gibco, 15140122), 100 units/mL streptomycin (Gibco, 15140122) and 1% GlutaMAX-I (Gibco, 350050061) and plated in a V-bottom 96 well plate with up to 0.5 × 10⁶ cells per well. Subsequently, cells were incubated for 10 min at 37 °C with either medium, or medium containing either 1 µM oligomycin, 100 mM 2-DG or both, 1 µM oligomycin and 100 mM 2-DG. Afterwards, the cells were incubated at 37 °C for additional 8 min with medium containing 20 µg/mL puromycin. The cells were washed once with cold PBS and processed further for flow cytometry analysis. Cells were stained with Viability stain (ThermoFisherScientific, Waltham, MA, USA, 65-0865-18) for 10 min on ice. Subsequently, they were stained for CD45-FITC (BioLegend, 103108) and CD11b-PE (BioLegend, 101208) in PBS containing 0.5% BSA for 35 min on ice. After fixation with Foxp3/Transcription Factor Staining Buffer Set (eBioScience, San Diego, CA, USA, 00-5523-00) according to manufacturers’ recommendation, cells were stained intracellularly with anti-puromycin antibody (MerckMillipore, Darmstadt, Germany, MABE343-AF647) in IS buffer, containing 2 mM EDTA + 2% BSA + 0.5% Saponin + 0.2% Tween20 in PBS for 35 min at RT. Microglia were identified via CD11b and intermediate CD45 expression.

4.10. Assessment of Mitochondrial Membrane Potential and Mitochondrial Mass

To evaluate mitochondrial content and membrane potential in microglia, the single-cell suspension was incubated with either medium or 20 µM FCCP for 15 min in RPMI 1640 standard medium (Gibco, 11875093), supplemented with 10% fetal calf serum (Sigma, 12133), 100 units/mL penicillin (Gibco, 15140122), 100 units/mL streptomycin (Gibco, 15140122) and 1% GlutaMAX-I (Gibco, 350050061) in a V-bottom 96 well plate containing up to 3 × 10⁵ cells per well at 37 °C. Subsequently, cells were incubated with medium containing 2 µM TMRE (Invitrogen, Carlsbad, CA, USA, T669) and 100 nM MitoTracker (Thermo Fisher Scientific, M22426) for 30 min at 37 °C. Cells were washed once with cold PBS and subsequently stained with CD45-FITC (BioLegend, 103108) and CD11b-BV605 (BioLegend, 101257) in PBS containing 0.5% BSA for 20 min on RT. Subsequently, cells were resuspended in PBS containing 0.5% BSA and 0.5 µL DAPI to enable exclusion of dead cells. For the analysis of the mitochondrial membrane potential, the gMFI of TMRE of CD45 intermediate positive and CD11b positive microglia, which were treated with FCCP, was subtracted from medium-treated condition. The resulting gMFI of TMRE, as well the gMFI of MitoTracker of the medium-treated conditions, was plotted for analysis.

4.11. Illustrations

Bioinformatics-related illustrations were generated in R using Seurat, ggplot2 and pheatmap packages. All figures were assembled and finalized using CorelDRAW version v26.x (Corel Corporation, Ottawa, ON, Canada), with structural elements precisely redrawn to ensure graphical consistency.

4.12. Statistics

All bioinformatical statistical analyses were performed using R (R version 4.4.3) and SPSS (version 29.0.2.0). p-values < 0.05 were considered statistically significant. Specific statistical tests (e.g., Wilcoxon rank sum test,) are indicated in the corresponding figure legends. Statistical analysis for metabolic assays were performed using GraphPad Prism (version 10.4.1).

5. Conclusions

Collectively, our data demonstrate region-specific transcriptional remodeling of microglia 5 days after experimental myocardial infarction, particularly affecting translation-related and proteostasis-associated pathways. A significant reduction in mitochondrial mass was observed, while additional metabolic measurements showed non-significant trends consistent with altered energetic regulation. These findings indicate that systemic cardiac injury is associated with microglial transcriptional adaptation in the early post-infarction phase. Further studies are required to determine the functional and long-term consequences of these alterations.