Result-1: RNAseq Analysis Reveals Mass Alterations in Heme Metabolism Between Normal Primary Human Hepatocytes and Hepatoma Cell Lines
Human primary hepatic metabolism is highly sensitive to laboratory manipulations and culture adaptations[32]. To investigate heme metabolism-related gene expression in primary human hepatocytes, RNAseq was performed directly from hepatocytes obtained from two healthy human donors (UVB and PDC) without any culture steps. Our experiments provided extensive coverage of primary hepatocyte transcriptomes, revealing the expression of approximately 7000 genes in human hepatocytes. Robust gene expression profiles associated with heme biosynthesis pathways were evident in healthy liver samples, facilitating crucial cellular processes such as the electron transport chain, exemplified by UQCRH, and cytochrome P450 enzymes, such as CYP2E1 (Fig. 1) (Supplemental Table S1).
To compare primary human hepatocyte metabolism with hepatoma cell line metabolism, we sequenced two liver cancer cell lines (HC04 and HepG2). RNAseq analysis contrasting primary human hepatocytes from healthy liver donors (UVB and PDC) with hepatoma cell lines (HC04 and HepG2) revealed significant alterations in heme metabolism in oncogenic transformation, as depicted in Fig. 1. Among the 6989 genes examined, 26% exhibited downregulation in cancer cell lines indicative of hepatic functional loss. Prominently, genes associated with primary liver metabolism, such as ARG1, UGT2R4, and HMGCS2, showed a downregulation rate of 81% in cancer lines. Conversely, 11% of genes showed upregulation in cancer cell lines associated with oncogenic transformations, such as CDK6 (Log2 Fold Change of expression > 1, adjusted p value < 0.05), as indicated by upregulated differentiation markers that 71% upregulated in cancer lines.
Next, the expression of heme metabolism-related genes was examined. Volcano plots illustrated distinct patterns of differential gene expression, revealing a substantial loss of major P450 cytochromes in liver cancer cell lines, with 79% downregulated in cancer lines. Components of the electron transport chain exhibited a 50% downregulation in cell lines, albeit to a lesser extent compared to P450, with most exhibiting a log2FC <= -2. Among heme biosynthesis enzymes, only the first two-step genes ALAS1 and ALAD were downregulated in liver cancer lines. With ALAS1 being the rate-limiting step of heme biosynthesis [33], our results suggest compromised hepatic heme production. Additionally, heme degradation enzymes displayed an interesting contrasting pattern of isozyme BLVRA and BLVRB gene expression, indicating that normal hepatocytes and liver cancer lines utilize different processes for heme degradation.
Importantly, while in vitro conditions typically exhibited a mild downregulation of the electron transport chain in cell lines, as observed in Fig. 1, our further in vivo analyses in the following results sections, including single-cell RNAseq (scRNAseq) of patient samples, liver cancer progression biopsy, and tumor tissue expression data, demonstrated sustained or even enhanced electron transport chain expression levels.
In summary, our findings confirm previous studies that liver heme-related drug metabolism is compromised in standard laboratory-cultured cell lines. Additionally, we report transcriptome-related compromised heme biosynthesis and aberrant heme degradation patterns.
Result − 2: Single-Cell Transcriptome Analysis Reveals Heme Rewiring Associated with Cell-to-Cell Heterogeneity in Oncogenesis in Human Liver Cancer
To further investigate the observed alterations in heme metabolism, particularly the impaired heme biosynthesis and significant downregulation of cytochrome P450 expression identified in our bulk RNAseq experiments, we analyzed a set of previously published single-cell RNA sequencing (scRNAseq) data [34, 35] from human hepatocellular carcinoma (HCC) patients. These analyses were aimed at studying details of cell-to-cell heterogeneity related to the rewiring of heme metabolism associated with cancer aggressiveness within tumors.
The patient-derived liver cancer single cells were initially reclustered into two distinct groups, more differentiated vs. more transformed, as illustrated in Fig. 2A. The segregation was based on hierarchical clustering of differentially expressed genes, with HCC_D representing more differentiated cells and HCC_T representing a more transformed (aggressive) cell population. Subsequent examination of heme biosynthesis and metabolic processes in the scRNAseq expression profiles of HCC samples revealed several key findings (Fig. 2B) (Supplemental Table S2). Firstly, genes specific to in vivo oncogenesis in the HCC_T group, absent in HepG2 cell lines, highlighted S100A family genes as early oncogenesis markers, consistent with previous conclusions [36]. Furthermore, both HCC_T and HepG2 exhibited an overall loss of expression in the cytochrome P450 gene family, indicating a commonality in this alteration across different liver cancer models. However, genes encoding Electron Transport Chain (ETC) hemoproteins remained largely unchanged during in vivo oncogenesis, suggesting that the mild downregulation of ETC in cancer lines may be an artifact of laboratory conditions.
Similar to our bulk RNAseq experiment results, single-cell RNAseq analysis (Fig. 2C) revealed downregulation of the first and second step heme biosynthesis genes ALAS1 and ALAD, respectively, in the more aggressive HCC_T cell population. This observation was consistent with the downregulation pattern observed in our primary human hepatocytes RNAseq experiments (Fig. 2D), indicating compromised heme biosynthesis in both liver cancer lines in vitro and in more transformed cancer cells in patient samples.
Result 3: Heterogeneous Distribution of Heme Metabolism Across Normal Human Liver Cell Populations
Using scRNAseq, we studied the diverse landscape of heme metabolism within different cell populations of normal human liver tissue from donor UBV. Given the crucial role of liver macrophages in heme homeostasis [37], we enriched for the macrophage Kupffer cell marker CD163 to ensure accurate identification of liver resident macrophages. In total, we generated transcriptomes of 1170 high-quality single cells using methods similar to our previous study [25]. Subsequent manifold learning and clustering analysis revealed six distinct cell populations (Fig. 3A) (Supplemental Table S3), with significant differentially expressed genes identified for further cell type characterization and functional annotations[38, 39] (Supplemental Table S4-5). For each cluster of significantly differentially expressed genes defining the single cell populations, we conducted cell type identification studies using predefined marker sets and non-parametric testing (Fig. 3B). Distinct populations in UBV liver were identified, including hepatocytes (enriched in APOC2 gene expression), Kupffer cells (enrich in CD163 gene expression), T cells (enriched in IL7R and CD8A gene expression), and NK cells (enriched in CD160 and GZMB gene expression).
Our gene expression analysis focused on heme metabolism-related genes across the identified cell populations (Fig. 3C). The rate-limiting step gene ALAS1, indicative of heme biosynthesis activity, exhibited significant upregulation in hepatocytes. Additionally, significant differences in cytochrome production were observed in hepatocytes, suggesting their dominant role in heme biosynthesis and cytochrome P450 supply. In contrast, Kupffer cells displayed an upregulated gene expression in process of heme breakdown, implying their role in heme recycling and homeostasis regulation. Furthermore, no significant differential expression was observed in genes related to the ETC across these populations, highlighting the importance of ETC function across all liver cell types.
To investigate heme homeostasis in hepatic cell proliferation within healthy livers, we conducted further analysis by categorizing single cells into potentially proliferating and non-proliferating populations based on established proliferation markers [38, 39] (Fig. 3D) (Supplemental Table S6). Our findings indicated no significant differences in heme biosynthesis activity or hemoprotein gene expression between proliferating and non-proliferating cells within healthy human liver tissue. This suggests robust and efficient maintenance of heme homeostasis across various liver cell functions.
Result 4: Correlation Between Heme Metabolic Shift and Clinical Progression in Liver Cancer
To study the clinical relevance of heme metabolism in cancer progression, we analyzed previously generated high-quality liver cancer progression data [27], indicative of dynamic metabolic rewiring during disease advancement. We examined the progression of hepatocellular carcinoma (HCC) across various stages, from preneoplastic lesions (cirrhosis and dysplasia) to advanced end stage tumors, as defined by Wurmbach et al [27]. Our gene expression pattern analysis included samples representing normal tissue (i.e., normal), non-tumor cirrhotic lesions (i.e., cirrhosis_1), dysplastic nodules of low and high grades (i.e., cirrhosis2 and 3), and different stages of HCC (i.e., HCC1-4), enabling a comparison to assess the statistical significance of observed gene expression changes in the sequence of oncogenesis. Figure 4A schematically illustrates the major differences in heme metabolism between healthy and oncogenic livers observed in HCC progression. While robust heme biosynthesis in healthy liver tissues supports vital metabolic functions, including P450 metabolism essential for cellular homeostasis, liver cirrhosis and liver cancer exhibit a marked reduction in heme biosynthesis and hemoprotein production, coupled with sustained or enhanced ETC activity.
Detailed analysis of patient liver cancer progression transcriptome data revealed distinct changes in gene expression profiles across disease stages (Fig. 4B) (Supplemental Table S7). As expected, markers of liver differentiation, such as EPHX2 [40], showed downregulation, while those associated with cancer proliferation, such as MKI67, increased. Notably, ALAS1, the rate limit gene in heme biosynthesis, was downregulated, while mid-step heme biosynthesis genes HMBS, driving intermediate porphyrin accumulation [25], showed opposing increasing expression during hepatic function degeneration, supporting the porphyrin overdrive hypothesis in oncogenesis.
Key P450 genes, such as CYP2A6 and CYP2C8 [14], displayed reduced expression levels in cancer progression, consistent with our bulk RNAseq and scRNA-seq results, while ETC-related genes, such as CYC1 and COX7B2, were upregulated during cancer progression, confirming in vivo human liver tumorigenesis sustains or boosting ETC. The heme metabolic gene expression pattern shifts indicate adaptive responses occurring within cancerous tissues, potentially contributing to disease progression by prioritizing heme supply to different processes demanding heme as a cofactor.
The clinical relevance of these metabolic alterations was evident in their correlation with patient overall survival (OS). Kaplan–Meier analysis derived from TCGA project demonstrated that heme biosynthesis and P450 expression were associated with inferior survival outcomes, while enhanced protoporphyrin IX (PPIX) accumulation and trafficking, along with increased ETC production, were linked to worse survival outcomes. Notably, increased expression of the heme export gene FLVCR1 [41, 42] was found in tumor progression (Fig. 4B), and associated with poor survival (Fig. 4C), indicating a significant role in heme and porphyrin trafficking in shaping the tumor microenvironment.
Together, these clinical data analyses, encompassing both HCC progression and patient survival data, show the potential of heme metabolic markers as prognostic indicators in liver cancer, suggesting the role of heme metabolism in driving disease progression.