Unravelling heart failure cellular signalling heterogeneity with spatial transcriptomics

This editorial refers to ‘Spatial transcriptional landscape of human heart failure’, by S.E. Lee et al., https://doi.org/10.1093/eurheartj/ehaf272. Heart failure (HF) is characterized by the inability of the heart to pump blood at a rate that covers the metabolic demands of the body. HF results from structural or functional heart abnormalities and often arises from conditions such as chronic coronary insufficiency, myocardial infarction, arterial hypertension, diabetes mellitus, amyloidosis, infectious diseases, chemotherapy- and radiotherapy-induced cardiomyopathy, congenital heart defects, genetic disorders, or ageing.1,2 Over 64 million people worldwide are living with HF, with an increasing prevalence due to population ageing and improved survival after an acute heart attack.1 Advances in treatment have improved outcomes for many patients, but HF remains a condition with significant long-term mortality. The lifetime risk of developing HF at age 50 ranges from 22.5% for women to 25.5% for men,3 and its prognosis remains poor, with a 5-year survival rate below 50%.1 Currently, there is no definitive cure for HF, and treatment focuses on managing symptoms, improving quality of life, and slowing disease progression. A critical hurdle in HF treatment is its heterogeneity, involving a complex interplay of genetic, molecular, and cellular mechanisms that contribute to the progressive decline in cardiac function. As cell function is primarily shaped by its transcriptome, approaches such as bulk RNA sequencing and microarrays have provided valuable insights, but often fall short in capturing the intricate cell type-specific and histologically heterogeneous nature of the failing heart. Single-cell RNA sequencing has revolutionized our understanding of disease-specific transcriptional programmes.4 However, these techniques are limited in their ability to spatially resolve gene expression within tissue landscape, a critical aspect of HF pathophysiology. Spatial transcriptomics, by preserving the tissue architecture, provides a novel opportunity to study heterogeneity, bridging the gap between histological observations and molecular data.5 Recent mapping of the human heart at early and late stages after myocardial infarction,6 and of atherosclerotic plaques from human coronary arteries,7 have uncovered novel potential therapeutic targets improving our understanding of cardiovascular pathophysiology. In this issue of the European Heart Journal, Lee et al.8 integrate spatial transcriptomics, histopathology, and clinical phenotypes to provide a comprehensive view of the pathophysiological mechanisms of the failing heart. Using the GeoMx Human Whole Transcriptome Atlas platform, they examined heart tissues from 44 patients diagnosed with dilated cardiomyopathy (DCM), ischaemic cardiomyopathy (ICM), hypertrophic cardiomyopathy with reduced or preserved ejection fraction (HCMrEF or HCMpEF), or without heart disease. HCMpEF (ejection fraction: 66 ± 5%) showed the most distinct gene expression patterns vs. other cardiomyopathies. In contrast, DCM, ICM, or HCMrEF shared a similar transcriptomic profile in both cardiomyocytes and endothelial cells, probably reflecting more the stage than the cause of the heart disease as all tissue specimens were derived from patients with severely impaired cardiac function (ejection fraction: DCM vs. ICM vs. HCMrEF: 23 ± 6% vs. 19 ± 7% vs. 36 ± 14%, respectively). This suggests a transcriptional convergence of the gene expression pattern in advanced cardiomyopathy regardless of the initial cause, a phenomenon also observed at the protein level.9 Across all cardiomyopathies, the common down-regulation of inflammatory pathways in cardiomyocytes indicates an overall reduced activation of immune signalling compared with the physiological level. These findings suggest that cardiomyocytes may undergo a state of immune suppression or exhaustion, potentially as a protective mechanism against macrophage-induced heart tissue damage.10 This finding agrees with a previous report showing that baseline myocardial inflammatory burden inversely correlates with cardiac improvement in end-stage HF patients undergoing mechanical circulatory support.11 In DCM, cardiomyocytes exhibited increased glycolysis and oxidative phosphorylation, pointing to a compensatory metabolic shift to sustain energy production. This dual up-regulation of energy pathways probably indicates an inefficient adaptation, potentially leading to energy deficits, oxidative stress, and worsening HF. The reduction in oxidative phosphorylation observed in ICM cardiomyocytes reflects the differences in metabolism between these two conditions. In contrast to cardiomyocytes, endothelial cells in both DCM and ICM showed increased tumour necrosis factor α (TNFα) signalling, suggesting a divergent inflammatory response. These findings highlight the need for cell type-specific analyses to guide precision therapies. The progression of HCMpEF to HCMrEF was marked by an up-regulation of inflammatory-related pathways in both cardiomyocytes and endothelial cells, suggesting that chronic inflammation in both cell types may drive fibrosis, myocardial damage, and vascular dysfunction. Conversely, myogenesis and oxidative phosphorylation were significantly down-regulated in HCMrEF, indicating that both impaired regeneration capacity and energy production contribute to contractile dysfunction.8 Current HCM therapies primarily target hypercontractility but do not directly address inflammation or metabolic dysfunction, suggesting a need for more targeted therapies focusing on inflammation, mitochondrial function, and cardiomyocyte survival that could potentially prevent or delay the transition to more severe complications. Notably, insulin-like growth factor-binding protein 2 (IGFBP2), which reduces the bioavailability of insulin-like growth factors 1 and 2 thereby limiting their protective effects on cardiomyocyte survival, and pyruvate dehydrogenase kinase 4 (PDK4), which shifts the metabolism in cardiomyocytes from oxidative phosphorylation toward the less efficient glycolysis, were both significantly up-regulated in HCMrEF compared with HCMpEF cardiomyocytes and should be further investigated in the disease progression context. The authors categorized the tissue segments based either on hypertrophy, degeneration, myocyte disarray, or fibrosis to identify a cell-specific gene signature linked to histological features.8 While hypertrophy exhibited few differentially expressed genes, cardiomyocyte degeneration was marked by an increase of ubiquitin carboxyl-terminal hydrolase L1 (UCHL1) involved in the degradation of damaged proteins, and heat shock protein A2 (HSPA2), a chaperone that activates protein quality control mechanisms. In disarrayed cardiomyocytes, the down-regulation of ribosomal proteins might indicate a translation defect and a reduction in global protein synthesis, a mechanism that controls inflammation and monocyte infiltration to the myocardium in ischaemia/reperfusion.12 These data reinforce the idea that a proteostatic imbalance may contribute to structural and functional decline of the failing heart. Targeting the mechanisms of proteostatic maintenance could therefore hold promise for effectively treating HF. Fibrotic regions were highly enriched with a proinflammatory endothelial cell subtype marked by increased expression of plasmalemma vesicle-associated protein (PLVAP), atypical chemokine receptor 1 (ACKR1; Duffy blood broup), and CC motif chemokine ligand 14 (CCL14), genes involved in leucocyte migration and vascular permeability, suggesting a role for this proinflammatory cluster in maladaptive remodelling. Considering that cardiac tissue biopsies from patients with a cardiomyopathy may also exhibit a normal histology, the authors recategorized myocardial segments into compensated state, i.e. a structurally and histologically preserved myocardium under stress, and uncompensated state, i.e. a morphologically abnormal myocardium.8 Oxidative phosphorylation and myogenesis were up-regulated in compensated states but down-regulated in end-stage HF. In contrast, inflammatory-related pathways were significantly up-regulated in uncompensated vs. compensated tissue, suggesting that the failure of compensatory mechanisms and increased inflammation mark a turning point in disease progression. Finally, by conducting a rare-variant analysis using the whole-exome sequencing data from the UK Biobank, the authors revealed significant associations with HF development for Tax1-binding protein 3 (TAX1BP3), cysteine-rich protein 3 (CRIP3), and 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 2 (PFKFB2), genes known for their metabolic regulation and inflammatory roles but without a previous link to HF progression.8 Since rare-variant associations are typically less susceptible to reverse causality, these genes may contribute, either directly or indirectly, to the development and progression of HF, highlighting their potential new role as therapeutic targets. Overall, this study highlights the added value of integrating clinical and histological data into spatial transcriptomics. The selection of a spatial transcriptomic platform depends on the research question and experimental objectives, considering factors such as the number of gene profiles, spatial resolution, RNA capture efficiency, and imaging area.5 GeoMx, which targets 18 000 genes, does not allow for a fully unbiased characterization of the spatial organization of gene expression. While it effectively compartmentalized cardiomyocytes, it struggled to segment densely packed or intertwined cells such as endothelial cells and fibroblasts. Furthermore, the requirement for minimum cell counts on the GeoMx platform restricted the comparison of fibroblasts between normal and failing hearts, highlighting the need for complementary technologies to improve cellular classification. Integrating in situ hybridization and transcriptome analysis at cellular or subcellular resolution is therefore needed to fully capture cell-bed heterogeneity. Spatial transcriptomics provide static snapshots of gene expression. Understanding dynamic processes, such as cell–cell interactions or immune infiltration, requires longitudinal data difficult to capture as this requires high-quality myocardial tissue samples, typically obtained through invasive endomyocardial biopsies or post-mortem specimens. Moreover, linking gene expression to functional outcomes, such as protein translation or metabolic activity, while preserving spatial context remains an open challenge. Finally, the need for an adequate bioinformatics pipeline further hinders reproducibility. Considerations from spatial data analysis methods widely explored in other fields such as geography could inspire the development of a new approach to spatial transcriptomics data analysis.13 As spatial transcriptomics continues to evolve, it holds immense potential for unravelling the complex molecular landscape of gene regulation and cell–cell communication in cardiovascular diseases, to develop precision medicine strategies (Graphical Abstract). To advance clinical utility, the creation of a spatiotemporal cell disease atlas with spatial multi-omics approaches (proteomics, transcriptomic, and metabolomics) will allow decoding of cell-specific complexity, and may reveal longitudinal interactions between cells and their microenvironment. Going deeper, future technologies that reveal the spatial epitranscriptome should be included, as the epitranscriptome controls all aspects of RNA life14 and, thus, possibly cellular signalling and cell communication, as recently suggested with the regulation of immune cell trafficking through interleukin-6 (IL-6) signalling in ICM.15 Addressing these challenges will be crucial for translating the present spatial transcriptomics into the future of precision medicine. The authors declare no disclosure of interest for this contribution. S.T-C. is supported by grants from the British Heart Foundation (PG/23/11093) and the Royal Society (RG1\241197). K.S. is supported by the German Research Foundation (Deutsche Forschungsgemeinschaft; DFG) (CRC1366 C07, project no. 394046768), the Health + Life Science Alliance Heidelberg Mannheim GmbH, and the Helmholtz Institute for Translational AngioCardioScience (HI-TAC).