Integration of scRNA-seq data and machine learning identified HMGN2 as a lactylation-associated hub gene in heart failure, with lactylation activity selectively enhanced in fibroblasts and smooth muscle cells and HMGN2 validated as upregulated in a mouse pressure overload model.
Key Findings
Results
Fibroblasts and smooth muscle cells exhibited the highest lactylation signature scores among cardiac cell types in heart failure.
UMAP analysis was used to identify cell types and assess lactylation activity across them.
Overall lactylation activity was considerably higher in HF compared to non-failing controls.
Multiple gene set scoring methods were used to assess lactylation activity across identified cell types.
scRNA-seq data from human failing and non-failing control hearts were integrated for this analysis.
Results
Seven potential lactylation-related genes (LRGs) were identified using a combination of machine learning algorithms.
The machine learning algorithms used included Boruta, Random Forest, LASSO, and XGBoost.
Among the seven LRGs identified, HMGN2, NUCKS1, and VIM were significantly upregulated in HF.
HMGN2 demonstrated the highest consistency across datasets as a reliable biomarker.
The multi-algorithm approach was used to improve robustness of gene candidate selection.
Results
HMGN2 was connected to immune-inflammatory signaling pathways and protein homeostasis dysregulation.
Gene set enrichment analysis (GSEA) was used to characterize HMGN2-associated biological processes.
Transcription factor-miRNA (TF-miRNA) regulatory network analysis was performed to further characterize HMGN2 regulation.
These analyses linked HMGN2 to immune-inflammatory signaling pathways and protein homeostasis dysregulation.
Results
Pseudotime trajectory and intercellular interaction analyses indicated a progressive rise in lactylation scores and HMGN2 expression along fibroblast and smooth muscle cell differentiation paths.
Pseudotime trajectory analysis was used to map changes in lactylation scores and HMGN2 expression during cell differentiation.
The findings suggested potential crosstalk between fibroblasts/smooth muscle cells and immune cells.
Intercellular interaction analyses were used to identify potential communication between cell types.
The progressive rise in HMGN2 expression was observed specifically along fibroblast and SMC differentiation paths.
Results
In vivo validation using a mouse model of pressure overload-induced heart failure confirmed increased HMGN2 expression alongside cardiac hypertrophy and fibrosis.
A mouse model of pressure overload-induced HF was used for in vivo validation.
Cardiac hypertrophy and fibrosis were confirmed in the mouse model.
Increased HMGN2 expression was observed in the failing mouse hearts.
This in vivo experiment provided experimental confirmation of the computational findings regarding HMGN2 upregulation in HF.
Conclusions
This study provides the first cell-type-resolved map of lactylation in heart failure.
The authors describe the study as 'the first cell-type-resolved map of lactylation in HF.'
The integrative multi-omics approach combined scRNA-seq data with machine learning-based gene identification.
The study used data from human failing and non-failing control hearts.
HMGN2 is highlighted as 'a promising therapeutic target' based on these findings.
What This Means
Heart failure is a serious condition where the heart cannot pump blood efficiently. Recent research has suggested that changes in how cells use lactate (a byproduct of energy metabolism) and how this affects the way DNA is packaged (a process called histone lactylation) may play a role in the harmful changes that occur in failing hearts. This study used advanced single-cell gene analysis techniques combined with artificial intelligence methods to map which types of heart cells show the most lactylation activity and to identify specific genes involved in this process. The researchers found that two cell types — fibroblasts (cells involved in scar tissue formation) and smooth muscle cells — showed the highest lactylation activity in failing hearts, and that overall lactylation was much more active in heart failure compared to healthy hearts. Using four different machine learning methods, the researchers narrowed down a list of seven genes that appear to be most relevant to lactylation in heart failure. Of these, a gene called HMGN2 stood out as the most consistent and reliable marker across multiple datasets. Further analysis showed that HMGN2 is linked to inflammation and problems with protein maintenance within cells. The findings were confirmed in mice with experimentally induced heart failure, where HMGN2 levels were indeed elevated alongside the expected signs of heart damage such as thickening of the heart muscle and scarring. This research suggests that lactylation — particularly in fibroblasts and smooth muscle cells — plays an important role in heart failure, and that HMGN2 may serve as a useful marker to track the disease or potentially as a target for future treatments. This is described as the first study to map lactylation activity at the level of individual cell types in the failing heart, which could open new avenues for understanding and eventually treating heart failure.
Xu A, Li Y, Gong Q, Sun H, Shi J, Li J, et al.. (2026). Single-cell RNA-seq and machine learning identify HMGN2 as a lactylation-associated hub gene in heart failure.. PloS one. https://doi.org/10.1371/journal.pone.0357971