NEW YORK – With the help of single-nucleus transcriptome profiling, an international team led by investigators in the US and Germany has characterized the cell types and gene expression features associated with different forms of heart failure, including those linked to a history of cardiomyopathy and specific pathogenic variants.
"Our findings hold enormous potential for rethinking how we treat heart failure and point to the importance of understanding its root causes and the mutations that lead to changes that may alter how the heart functions," co-senior and co-corresponding author Christine Seidman, director of the Cardiovascular Genetics Center at Brigham and Women's Hospital and a professor at Harvard Medical School, said in a statement.
As they reported in Science on Thursday, the researchers relied on single-nucleus RNA sequencing to profile transcriptomic patterns in almost 881,100 individual cells found in left and right ventricle heart tissue samples from 61 heart failure patients, comparing them to samples from 18 control participants.
While a subset of the heart failure patients carried pathogenic risk variants in genes contributing to conditions such as dilated cardiomyopathy (DCM) or arrhythmogenic cardiomyopathy (ACM), they noted, other cases were classified as idiopathic cardiomyopathies lacking such established risk variants.
"This is fundamental research, but it identifies targets that can be experimentally pursued to propel future therapeutics," Seidman explained. "Our findings also point to the importance of genotyping — not only does genotyping empower research but it can also lead to better, personalized treatment for patients."
The team's analyses — part of the larger Human Cell Atlas effort — highlighted some 71 transcriptional states, along with at least 10 expression-based cell clusters, including clusters representing cardiomyocytes, smooth muscle cells, fibroblasts, immune cells, and other cell types.
"States of the same cell type shared a transcriptional profile but also expressed distinct genes," the authors explained, "which implied biological differences."
Despite the fibrotic features identified in failing heart tissues, the researchers did not see a rise in fibroblast cells in the heart disease patients. Rather, the results pointed to a shift in the transcriptional states found for this cell type, including expanded extracellular matrix remodeling.
"The number of these cells stays the same," co-first author Eric Lindberg, a researcher at the Max Delbrück Center for Molecular Medicine, explained in a statement. "But the existing cells become more active and produce more extracellular matrix, which fills in the space between the connective tissue cells."
On the other hand, the investigators did see a decline in cardiomyocytes in the heart failure samples, along with enhanced representation of endothelial and immune cells.
When the team turned to machine learning methods to explore ties between the cell type clusters, cell transcriptional states, and participants' genotypes, meanwhile, it saw signs that known disease risk variants can impact everything from cell type composition to the precise pathways and processes at play in failing heart tissue.
The extracellular matrix boost found in fibroblast cells was particularly common in individuals with changes in DCM-related RBM20 genes, for example, and the RBM20 alterations tended to coincide with earlier-than-usual heart transplantation in the subset of patients with DCM. The RBM20 mutations were also implicated in enhanced dilation of the left ventricle, but lower-than-usual systolic contraction activity.
Similarly, the subset of heart failure patients with ACM-associated protein plakophilin-2 (PKP2) mutations showed an uptick in tumor necrosis factor activity, the researchers reported, noting that the PKP2 mutations coincided with diminished right ventricular heart function but relatively unaffected left ventricular function.
Along with efforts to narrow in on treatment-relevant disease biomarkers, the team demonstrated that it was possible to use artificial intelligence to work backwards from the transcriptional states and single-cell gene expression data to predict whether patients carried pathogenic germline alterations in specific cardiomyopathy-related genes.
"Because our findings indicated genotype-enriched transcripts and cell states, we harnessed machine learning to develop a graph attention network for the multinomial classification of genotypes," the authors explained. "This network showed remarkably high prediction of the genotypes for each cardiac sample, thereby reinforcing our conclusion that genotypes activate very specific heart failure pathways."