Transcriptional and epigenetic landscape of Ca²⁺-signaling genes in hepatocellular carcinoma

Abstract

Calcium (Ca²⁺) signaling has a major role in regulating a wide range of cellular mechanisms, including gene expression, proliferation, metabolism, cell death, muscle contraction, among others. Recent evidence suggests that ~ 1600 genes are related to the Ca²⁺ signaling. Some of these genes’ expression is altered in several pathological conditions, including different cancer types, and epigenetic mechanisms are involved. However, their expression and regulation in hepatocellular carcinoma (HCC) and the liver are barely known. Here, we aimed to explore the expression of genes involved in the Ca²⁺-signaling in HCC, liver regeneration, and hepatocyte differentiation, and whether their expression is regulated by epigenetic mechanisms such as DNA methylation and histone posttranslational modifications (HPM). Results show that several Ca²⁺-signaling genes’ expression is altered in HCC samples; among these, a subset of twenty-two correlate with patients’ survival. DNA methylation correlates with eight of these genes’ expression, and Guadecitabine, a hypomethylating agent, regulates the expression of seven down-regulated and three up-regulated genes in HepG2 cells. The down-regulated genes displayed a marked decrease of euchromatin histone marks, whereas up-regulated genes displayed gain in these marks. Additionally, the expression of these genes is modulated during liver regeneration and showed similar profiles between in vitro differentiated hepatocytes and liver-derived hepatocytes. In conclusion, some components of the Ca²⁺-signaling are altered in HCC and displayed a correlation with patients’ survival. DNA methylation and HMP are an attractive target for future investigations to regulate their expression. Ca²⁺-signaling could be an important regulator of cell proliferation and differentiation in the liver.

Electronic supplementary material

The online version of this article (10.1007/s12079-020-00597-w) contains supplementary material, which is available to authorized users.

Introduction

The increase of cytoplasmic Ca²⁺ concentrations activates a broad range of Ca²⁺-dependent proteins, which through different pathways regulate several key cellular functions such as growth, cell cycle, metabolism, differentiation, cell death, proliferation, gene expression, muscle contraction, among others. Ca²⁺ signaling is orchestrated by ~ 1600 genes associated with 241 GO terms, including transducers, effectors, pumps, channels, and buffer proteins, highlighting its complexity and versatility (Monteith et al. 2007; Roderick and Cook 2008; Hortenhuber et al. 2017). The role of these genes is fundamental to maintain the intracellular calcium homeostasis and normal cell function. Deregulation and remodeling of the calcium signaling components can lead to several pathological conditions, including cancer. Some of the Ca²⁺ signaling genes’ expression is altered in several types of cancer and associates with the cancer hallmarks, such as sustained cell proliferation, cell death resistance, angiogenesis, invasion and metastasis, and altered cell metabolism (Prevarskaya et al. 2014; Monteith et al. 2017; Izquierdo-Torres et al. 2020). Ca²⁺ channels, such as the members of the transient receptor potential family (TRP) and the components of the store operated calcium entry (SOCE) are widely studied in cancer (Prevarskaya et al. 2014; Monteith et al. 2017). Moreover, epigenetic mechanisms such as DNA methylation and HPM have emerged as potential transcriptional regulators of the Ca²⁺-signaling genes (Izquierdo-Torres et al. 2020; Gregório et al. 2020). However, only the expression and regulation of a few genes of the Ca²⁺-signaling have been studied. Ca²⁺ signaling also plays an essential role in regulating tissue-specific mechanisms. Particularly in the liver, it is implicated in bile synthesis, glucose and lipid metabolism, and regeneration. Furthermore, it is associated with liver-specific diseases such as hepatocellular carcinoma (HCC), cholestasis, hepatitis, and nonalcoholic fatty liver disease (NAFLD) (Amaya and Nathanson 2013). Accumulation of lipid droplets in hepatocytes leads to altered Ca²⁺signaling, endoplasmic reticulum (ER) stress (closely related with Ca²⁺ homeostasis in the ER), generation of reactive oxygen spices (ROS), and the potential initiation of HCC (Ali et al. 2019). Thus, the components of the Ca²⁺ signaling could be an attractive target for future clinical investigations. However, the complete landscape of the Ca²⁺ signaling genes in the liver remains to be elucidated.

HCC is the most common liver cancer, a heterogeneous disease characterized by alterations in proliferation, differentiation, and gene expression (Forner et al. 2018; Llovet et al. 2016). In 2018, The Global Cancer Observatory reported 841,080 new cases and 781,631 deaths from liver cancer, occupying the 6^th place in incidence and the 4^th place in mortality among all cancers (Bray et al. 2018). HCC is also one of the most aggressive types of cancer, being a public health issue that needs to be approached. The expression of Ca²⁺-signaling genes such as FOSB, DNASE1L3, ECT2, CLEC3B, among others, is altered in HCC (Liu et al. 2018a, b; Wang et al. 2020; Dai et al. 2019; Chen et al. 2015). DNA methylation regulates CPS1 gene expression in HCC cell lines (Liu et al. 2011). Recently, DNA methylation is being studied as a potential therapeutic target for HCC. Guadecitabine, a second-generation DNA methyltransferase inhibitor (DNMTi), is currently under HCC clinical trial (Liu et al. 2018a, b). However, very little is known about the role of epigenetic mechanisms, such as DNA methylation and HPM, that regulate the Ca²⁺-signaling genes in HCC. Therefore, this study aimed to explore the transcriptional and epigenetic landscape of the Ca²⁺-signaling genes in HCC, liver regeneration, and hepatocyte differentiation; since these two processes share expression profiles and biological mechanisms with cancer. A better understanding of the Ca²⁺-related genes expression in HCC could lead to novel diagnostic and therapeutic approaches to address HCC.

Methods

Analysis of Ca²⁺-signaling genes expression and overall survival in HCC datasets

The list of Ca²⁺-signaling genes was based on a previously reported list of 1670 genes involved in Ca²⁺ signaling (Hortenhuber et al. 2017). Expression analysis was performed with three different HCC datasets, TCGA-LIHC, GSE14520, and GSE36376, using Phantasus (Lim et al. 2013; Roessler et al. 2010; Zenkova et al. 2018). Data were log2-transformed and quantile normalized, probes were collapsed by a maximum median probe when necessary, and gene expression analysis was performed with the limma package (Ritchie et al. 2015). The consensus up- or down-regulated genes among the three datasets were used for the OS analysis, which was performed with the TCGA-LIHC dataset with Kaplan–Meier survival plots (Menyhart et al. 2018). Patients were split into high and low expression groups (Q1 and Q4, according to the quartile of expression of the respective gene). Down-regulated genes with a Hazard Ratio (HR) < 1 and up-regulated genes with a HR > 1 were selected for further analysis. To determine whether the altered genes modulate their expression in different liver damage stages, we used the GSE6764 dataset, which includes nine different stages, ranging from normal liver to very advanced HCC (Wurmbach et al. 2007). The expression of altered genes was also analyzed in HepG2 cells and compared with normal liver, using the GSE18269 dataset (Hart et al. 2010). An adjusted p value < 0.01 was considered statistically significant.

DNA methylation analysis of TCGA-LIHC samples and HepG2 cells

Promoter methylation analysis of tumor samples was performed with the TCGA-LIHC methylation data. We used DNMIVD to analyze and visualize differentially methylated genes (DMG) in HCC samples and compared them to normal tissue (Ding et al. 2020). Then a Pearson correlation analysis was performed to determine whether promoter methylation correlates with gene expression. In both cases, a p value < 0.05 was considered statistically significant. WGBS data were visualized with four datasets (three from hepatocytes and one from HepG2 cells) from the International Human Epigenome Consortium (URLs available in Table S1) at the WashU Epigenome Browser v50.4.0 (Li et al. 2019). A gene expression analysis of the altered Ca²⁺-signaling genes in HepG2 cells treated with Guadecitabine, a second-generation DNA Methyltransferase inhibitor (DNMTi) currently under HCC clinical trial, was performed with the GSE105065 dataset, as described above. We also analyze the promoter methylation of the altered genes, using the GSE10566-GPL8490 dataset (Liu et al. 2018a, b). β values were obtained from GEO2R, and CpG annotations from 450 k Illumina Human Methylation Array were obtained with the R package COHCAPanno (Warden 2020). Analyzed probes, associated genes, and CpG position are described in Table S2.

ChIP-seq data visualization of hepatocytes and HepG2 cells

We accessed the Cistrome Data Browser from ChIP-seq data to visualize the histone euchromatin marks H3K4me3, H3K9ac, H3K27ac, and the heterochromatin mark H3K27me3, from ChIP-seq data (Zheng et al. 2019). The occupancy of RNA Pol II was also mapped in the promoter region of the Ca²⁺-signaling genes (URLs available in Table S1). A gene expression analysis of the altered Ca²⁺-signaling genes in HepG2 cells treated with the histone deacetylase inhibitors (HDACi) SAHA and TSA was performed as described above, using the GSE52232 and GSE4465 datasets, respectively (Chittur et al. 2008; Lee et al. 2014).

Expression of Ca²⁺-signaling genes during rat liver regeneration and in vitro hepatocyte differentiation

The expression of the altered Ca²⁺-signaling genes was analyzed in a murine liver regeneration model, with the GSE63742 dataset, as described above. This dataset consists of 10-time points, ranging from 0 to 168 h after partial hepatectomy, and their corresponding sham. A gene was considered up- or down-regulated if its expression was significantly different from the control liver (0 h) and their respective sham. Next, we evaluated the expression of the altered genes during hepatocyte differentiation using the GSE25046 dataset, which includes data of human embryonic stem cells (hESC), endoderm progenitors (EP), and mature hepatocytes (MH). We also analyzed the promoter methylation of the altered genes, using the GSE25047-GPL8490 dataset, as described above, and the CpG annotation from 27 k Illumina Human Methylation array (Kim et al. 2011). Finally, we explored the expression of the altered Ca²⁺-signaling genes with the GSE115469 dataset of human liver single-cell RNA-seq data, which includes expression data of different liver cell populations (MacParland et al. 2018). Results were visualized in a heatmap generated with the Single Cell^BETA Portal (Broad Institute).

Results

Altered expression of the Ca²⁺-signaling genes in HCC datasets

We analyzed the expression of ~ 1600 Ca²⁺-signaling genes using three HCC datasets (TCGA-LIHC, GSE14520, and GSE36376). A total of 614, 817, and 816 genes with altered expression were found in each dataset, respectively (File S1, sheets 1–3). From these genes, 169 down-regulated and 101 up-regulated genes were consistent among the three datasets (Fig. S1; File S1, sheets 1–3). Among the down-regulated genes, the expression of CCL23, GPM6A, FOSB, DNASE1L3, SELP, PLA2G12A, PDE2A, ANXA10, CLEC3B, RAPGEF2, RAMP3, REPS2, RGN, PON1, CPS1, MASP2, C1S, and FYN, correlates with patients’ survival, with a significant HR < 1, whereas from the up-regulated genes, MELK, ECT2, STX1A, and LPCAT1, displayed a significant HR > 1 (Table S3; File S1, sheet 4). Further analyses were focused on these 22 altered genes, and a representative heatmap of their expression is shown in Fig. 1a. Subsequently, we explored whether the expression of these genes was modulated in different liver damage stages. Results showed that ANXA10 expression decreases since cirrhosis and low dysplastic tissues, DNASE1L3, GPM6A, PDE2A, CCL23, and MELK expression was altered at early stages of HCC, whereas PLA2G12, CLEC3B, FOSB, FYN, REPS2, PON1, RAMP3, RAPGEF2, ECT2, and LPCAT1 expression was altered at final stages of HCC (Fig. 1b; File S1, sheet 5). We then analyzed the expression of the altered genes in HepG2 cells. Results showed that C1S, CPS1, DNASE1L3, PON1, GPM6A, MASP2, CLEC3B, REPS2, ANXA10, RAMP3, and RGN genes were down-regulated, and LPCAT1, MELK, and ECT2 genes were up-regulated, compared to the normal liver (Fig. 1c; File S1, sheet 6). Interestingly, these 14 altered genes in HepG2 cells showed the same pattern (down- or up-regulated) than HCC samples (Fig. 1a).

Fig. 1.

Ca²⁺-signaling genes with altered expression in hepatocellular carcinoma samples. Panel a shows a representative heatmap of the Ca²⁺-signaling genes altered in HCC obtained with the TCGA-LIHC dataset. The green and blue boxes on the top represent normal liver and HCC samples, and RNA-seq clusters 1 to 5 are highlighted in orange, blue, light green, green, and pink, respectively. Panel b shows the expression analysis of the altered Ca²⁺-signaling genes in different liver damage stages. N: normal; C: cirrhosis without HCC; Ch: cirrhosis; Ld: low dysplastic; Hd: high dysplastic, Ve: very early HCC; E: early HCC; A: advance HCC; Va: very advance HCC. Panel c shows the expression level of the Ca²⁺-signaling genes in HepG2 cells compared to normal liver. The pink box represents liver samples, and the green box represents HepG2 samples. In all panels, genes are sorted by descending t statistic value

Promoter methylation of the subset of Ca²⁺-signaling genes altered in HCC

Next, we explored whether the altered gene expression was related to changes in promoter methylation. Results showed that eight down-regulated and one up-regulated were DMG. Particularly, a decreased promoter methylation was found in seven of the eight down-regulated genes in HCC samples, whereas only the RAMP3 promoter displayed an increase in methylation. ECT2, which is up-regulated in HCC samples, also showed an increase in promoter methylation compared to normal tissue (Fig. 2a). A Pearson correlation analysis showed a negative correlation of RAMP3, RGN, MASP2, and ECT2 promoter methylation with gene expression. In contrast, SELP, PDE2A, CLEC3B, and FYN promoter methylation show a positive correlation with gene expression, and CPS1 shows no significant association (Fig. 2b; File S1, sheet 7).

Fig. 2.

Methylation landscape of the Ca²⁺-signaling genes altered in HCC. Panel a shows the β value at the promoters of the altered Ca²⁺-signaling genes in normal liver (green boxes) and tumor samples (orange boxes). Panel B displays the Pearson’s correlation for each gene. Panel c, comparison of WGBS data of the altered Ca²⁺-signaling genes between hepatocytes and HepG2 cells. Healthy hepatocytes (HPC20, HPC25, and HPC35) and HepG2 data are represented in blue and red. Panel d shows a heatmap of the gene expression analysis of HepG2 cells treated with 100 nM Guadecitabine for three days, indicated with a horizontal black bar on the top, and numbers represent days from the beginning of this treatment. Genes are sorted from top to bottom by descending t statistic value. Panel e represents the promoter methylation levels of the indicated genes in HepG2 cells treated with 100 nM Guadecitabine for three days. Red, orange, yellow, and green bars represent 0, 10, 17, and 24 days (d) from treatment initiation

We also explored the methylation landscape of the DMG, described above, in HepG2 cells and compared them with hepatocytes. CLEC3B, RGN, and ECT2 genes displayed a decrease in DNA methylation, upstream (ECT2) and downstream (CLEC3B and RGN) of the transcriptional starting site (TSS). In contrast, an increase in DNA methylation was found near the TSS of MASP2 (Fig. 2c). HepG2 cells treated with 100 nM Guadecitabine for three days showed an increase in MASP2, DNASE1L3, FOSB, C1S, PON1, REPS2, and RGN, and a decrease in MELK, ECT2, LPCAT1, and FYN expression, at least at one-time point (Fig. 2d; File S1, sheet 8). In HepG2 cells, C1S gene methylation was increased, whereas FOSB, REPS2, FYN, LPCAT1, and MELK displayed a decreased methylation (Fig. S2). Guadecitabine treatment of HepG2 cells decreased the promoter methylation of C1S, PON1, FYN, and LPCAT1 genes and increased that of REPS2 and MELK genes (Fig. 2e; File S1, sheet 9).

Histone posttranslational modifications were altered in the Ca²⁺-signaling genes

Next, we aimed to determine whether HPM, such as acetylation and methylation at the promoters of the altered Ca²⁺-signaling genes were modulated in HepG2 cells compared to hepatocytes. We found a marked decrease of euchromatin associated marks H3K4me3, H3K9ac, and H3K27ac, and RNA Pol II recruitment at the promoters of the down-regulated genes C1S, DNASE1L3, PON1, and REPS2, as well as an increase of H3K27me3. In contrast, the RGN promoter displayed an increase in these euchromatin marks (Fig. 3a). Other down-regulated genes such as CPS1, GPM6A, and CLEC3B showed decreased euchromatin marks, with no changes in H3K27me3, whereas MASP2 and RAMP3 displayed a reduction in H3K4me3 and H3K27ac, with an increase in H3K27me3. These down-regulated genes also showed a decrease in the RNA Pol II signal (Fig. 3b). ANXA10 showed a slight decrease in H3K27ac, H3K27me3, and Pol II occupancy (Fig. S3a). In contrast, overexpressed genes LPCAT1, MELK, and ECT2 showed an increase in euchromatin marks, except H3K27ac at the LPCAT1 promoter. An increase in the heterochromatin mark H3K27me3 was also observed at the promoter of these three genes. However, an increase of the Pol II signal was only found at MELK and ECT2 promoters (Fig. 3c).

Fig. 3.

Histone methylation and acetylation at the promoter of the Ca²⁺-signaling genes altered in HCC. Panel a and b show the down-regulated genes, and Panel c displays the up-regulated genes in HepG2 cells. H3K4me3, H3K9ac, H3K27ac, H3K27me3, and Pol II data are represented in red, yellow, green, blue, and black, respectively. Arrows indicate the TSS and their respective gene orientation. The Y-axis was adjusted to each gene region’s maximum signal for each pair of datasets (hepatocytes and HepG2 cells)

SAHA or TSA modulate Ca²⁺-signaling genes expression in HepG2 cells. FYN expression increased at both conditions, STX1A expression increased with SAHA, whereas PDE2A and RAPGEF with TSA. In contrast, MASP2 and RGN expression decreased under both conditions. Added to this, CPS1 and MELK expression decreased with SAHA, whereas LPCAT1 and C1S with TSA (Fig. S3b; File S1, sheets 10 and 11, respectively). Furthermore, H3K27ac decreased, and H3K27me3 increased at the FYN and STX1A promoters. In contrast, the promoters of PDE2A and RAPGEF2 showed a decrease of H3K4me3, H3K9ac, and H3K27ac, and an increase of H3K27me3 in HepG2 cells. RNA Pol II signal decreases at the promoter of these four genes (Fig. S3a).

Expression of the altered genes in rat regenerating liver, hepatocyte differentiation, and in human liver cell populations

Tissue regeneration is a major feature of the liver and involves the transition from a quiescent to a proliferative cell state. Therefore, we explored whether the altered Ca²⁺-signaling genes were modulated during liver regeneration after rat partial hepatectomy. Reps2, Dnase1l3, Rgn, Ect2, Pon1, and Pde2a expression decreased at least at a one-time point, between 6 to 120 h after partial hepatectomy. Specifically, down-regulation of Dnase1l3 was sustained from 6 to 72 h, whereas Reps2 and Rgn expression decreased between 6 to 30 h and 6 to 12 h, respectively. Pon1 expression decreased from 12 to 30 h, and Pde2a expression was down-regulated only at 36 h. Besides, Fyn, Rapgef2, Melk, Pla2g12a, and Masp2 expression increased at least at a one-time point between 6 and 120 h. Only Ect2 displayed both, decreased and increased expression since it decreased at 12 h and then increased from 24 to 120 h (Fig. 4a; File S1, sheet 12). Next, we explored whether the expression and methylation of the altered Ca²⁺-signaling genes are modulated during in vitro differentiation of hESC into hepatocytes. The expression of MELK, ECT2, PDE2A, LPCAT1, FYN, RAMP3, and CLEC3B decreased in MH compared with hESC, whereas that of C1S, RGN, CPS1, MASP2, STX1A, FOSB, DNASE1L3, CCL23, and ANXA10 increased (Fig. 4b; File S1, sheet 13). Moreover, RAMP3 and C1S promoter methylation increased, whereas RGN and MASP2 promoter methylation decreased in MH compared with hESC (Fig. 4c; File S1, sheet 14). Next, using a single-cell RNA-seq dataset of human liver cell populations, we found low expression levels of MELK, ECT2, LPCAT1, PDE2A, RAMP3, and CLEC3B, whereas C1S, RGN, CPS1, MASP2, and ANXA10, showed higher expression levels in hepatocytes (Fig. 4d). These expression patterns are similar to those observed in MH derived from hESC (Fig. 4c). The expression of C1S, CPS1, PON1, MASP2, REPS2, ANXA10, RGN, and PLA2G12A was also higher in hepatocytes compared to the other liver cell populations. DNASE1L3, GPM6A, CLEC3B, RAMP3, PDE2A, CCL23, and SELP expression was higher in central venous, periportal, and portal liver sinusoidal endothelial cells. FYN and LPCAT1 showed higher levels in alpha–beta T, gamma-delta T, and natural killer cells. FOSB expression seems similar in all cell populations, except hepatocytes that show relatively low levels, whereas RAPGEF2 expression remains similar in all cellular populations. MELK and ECT2 expression were barely visualized in gamma-delta T cells, and STX1A shows the lowest expression levels in all cellular populations (Fig. 4d).

Fig. 4.

Expression of the altered Ca²⁺-signaling genes in rat liver regeneration and hepatocyte differentiation. Panel a shows the gene expression analysis of the altered Ca²⁺-signaling genes in a rat liver regeneration model. Hepatectomy samples are highlighted with a green box, whereas sham samples are represented with an orange box. Time is sorted in hours after hepatectomy, and genes are sorted by descending t statistic value. Panel b shows the expression analysis of the altered genes during hepatocyte differentiation. hESC, EP, and MH conditions are represented by blue and green bars, respectively, and genes are sorted from top to bottom by descending t statistic value. Panel c shows the promoter methylation levels of the altered genes in hepatocyte differentiation. Red, yellow, and green bars represent hESC, EP, and MH, respectively. Results are shown as mean ± standard deviation of three independent experiments. One-way ANOVA was used for statistical analysis, *p < 0.05 and **p < 0.01. Panel d shows a heatmap from liver single-cell RNA-seq data. Liver cell populations go as follow: A, alpha–beta T cell; B, central venous liver sinusoidal endothelial cell; C, cholangiocyte; D, erythroid cell; E, gamma-delta T cell; F, hepatic stellate cell; G, hepatocyte; H, inflammatory macrophage; I, mature B cell; J, natural killer cell; K, non-inflammatory macrophage; L, periportal liver sinusoidal endothelial cell; M, plasma cell; N, portal liver sinusoidal endothelial cell

Discussion

The incidence of liver cancer has increased 2–3% a year from 2007 to 2016 and accounts for 8.2% of cancer deaths worldwide (Bray et al. 2018; Siegel et al. 2020). Since epigenetic mechanisms seem to be involved in all carcinogenesis stages, they are an attractive target for developing new therapeutic strategies (Allis and Jenuwein 2016). Some epigenetic drugs targeting DNMT or HDAC have been tested in clinical trials with good outcomes for some types of cancer, and the FDA has approved them (Crump et al. 2008; Kaminskas et al. 2005). Notably, remodeling of the Ca²⁺-signaling machinery expression has been reported in cancer, and epigenetic mechanisms have emerged as important players in regulating several Ca²⁺-signaling genes (Prevarskaya et al. 2014; Monteith et al. 2017; Aghdassi et al. 2012; Busselberg and Florea 2017; Kim et al. 2019, 2020; Venza et al. 2016; Gregório et al. 2020; Izquierdo-Torres et al. 2020). Despite the evidence, the transcriptional and epigenetic landscape of most genes involved in the Ca²⁺ signal in cancer, and particularly in HCC, is unknown.

Gene expression analysis revealed that 18 down-regulated and four up-regulated, Ca²⁺-signaling genes, whose expression correlates with HCC patients’ survival, were shared among the three HCC datasets. Our analysis corroborates previous studies showing that the expression of CCL23, FOSB, DNASE1L3, ANXA10, CLEC3B, RAMP3, RGN, PON1, CPS1, and C1S was down-regulated; whereas ECT2, MELK, and LPCAT1 were up-regulated in HCC (Liu et al. 2018a, b; Wang et al. 2020; Dai et al. 2019; Chen et al. 2015; Liu et al. 2011; Lu et al. 2019; Liu et al. 2002; Fang et al. 2018; Yamaguchi et al. 2016; Zhang et al. 2020; Dong et al. 2009; Xia et al. 2016; Morita et al. 2013). To our understanding, the altered expression of GPM6A, SELP, PLA2G12A, PDE2A, RAPGEF2, REPS2, MASP2, FYN, and STX1A in HCC has not been previously reported. Moreover, the expression of 13 down-regulated and three up-regulated Ca²⁺-signaling genes were modulated at different liver damage stages, some as early as the cirrhotic stage. ANXA10 expression negatively correlates with serum alpha-fetoprotein, and the loss of its expression was associated with grade II and III HCC tumors (Liu et al. 2002). CCL23 expression was down-regulated in HCC, and its low expression in patients correlates with poor survival (Lu et al. 2019). High levels of GPM6A and FOSB have been involved in the differentiation of neurons and normal mammary cells, respectively (Michibata et al. 2009; Milde-Langosch et al. 2003). Since GPM6A and FOSB are down-regulated in HCC, it might be interesting to study their role in hepatic differentiation. MELK, ECT2, STX1A, and LPCAT1, have also been found overexpressed in several types of tumors (Giuliano et al. 2018; Huff et al. 2013; Mansilla et al. 2009; Raja et al. 2019). Our analysis also revealed that 11 down-regulated and three up-regulated genes displayed the same alteration pattern in HepG2 cells and HCC samples. A previous study showed that 2646 down-regulated and 3586 up-regulated genes were also altered in HepG2 cells compared to hepatocytes (Costantini et al. 2013). Thus, HepG2 cells are a proper model for testing further hypothesis.

The components of the Ca²⁺-signaling are associated with the hallmarks of cancer. Silencing of PMCA2 potentiates apoptosis of MDA-MB-231 cells, whereas increased expression of PMCA2 renders T47D cells more resistant to apoptosis (Curry et al. 2016; VanHouten et al. 2010). The components of SOCE, ORA1, and STIM1 are involved in cell motility and metastasis, and their expression is altered in some types of cancer (Mo and Yang 2018). Overexpression of CACNA2D2 stimulates cell proliferation and angiogenesis in prostate cancer cells (Warnier et al. 2015). Thus, Ca²⁺-signaling genes could have an important role in all stages of tumorigenesis, and they are an attractive target for future investigations.

DNA methylation at gene promoters is typically associated with gene repression (Tough et al. 2016). Alteration in DNA methylation patterns is a well-known event in tumorigenesis, as well as other diseases. Our data show that the promoter of the down-regulated genes SELP, PDE2A, CLEC3B, RGN, CPS1, MASP2, FYN, and RAMP3, and the up-regulated ECT2 gene, were differentially methylated in HCC samples compared to normal liver. However, a negative correlation between gene expression and promoter methylation was only found for RAMP3, RGN, MASP2, and ECT2 genes, whereas SELP, PDE2A, CLEC3B, and FYN displayed a positive correlation. DNA promoter methylation does not always correlate with gene expression, as we observed in our data. In most cases, correlates with gene repression, but there are cases where genes with unmethylated promoters are silenced, and genes with a methylated promoter are transcriptionally active (Tough et al. 2016; Zinn et al. 2007). This could explain the differences in the correlation between DNA methylation and gene expression. To get a better insight into the potential role of DNA methylation in the regulation of the altered genes, we used WGBS and gene expression data from HepG2 cells. CLEC3B and RGN showed similar methylation patterns between the hepatocytes and HepG2 datasets. However, MASP2 showed an increase in the methylation signal at the TSS, whereas ECT2 displayed a decreased methylation at the TSS and upstream. This result became more interesting when we analyze the expression of the altered genes in HepG2 cells treated with Guadecitabine. MASP2 expression increased, whereas ECT2 expression decreased, and in both cases was sustained up to 21 days after treatment. Guadecitabine also increased the down-regulated genes DNASE1L3, FOSB, C1S, PON1, REPS2, and RGC expression, and LPCAT1, MELK, and FYN expression decreased. Our results also showed decreased promoter methylation of the up-regulated genes C1S and PON1 after Guadecitabine treatment, whereas down-regulated MELK displayed increased promoter methylation. Therefore, these results suggest that Ca²⁺-signaling genes could be regulated by promoter methylation in HCC. To our knowledge, CPS1 is the only gene among Ca²⁺-signaling genes, altered in HCC, whose expression by DNA methylation has been tested (Liu et al. 2011). It was recently shown, in pancreatic adenocarcinoma, that altered methylation levels correlate with the expression of the Ca²⁺-signaling genes ADRA1A, CACNA1A, CACNA1B, CACNA1H, CASQ2, HRH1, ORAI2, P2RX2, PDE1C, and PRKCB (Gregório et al. 2020); however, none of these genes were significant altered in our analysis with HCC samples. In lung cancer, the expression of S100A2 increased, and the promoter methylation decreased, showing a negative correlation between gene expression and promoter methylation (Izquierdo-Torres et al. 2020). Together, the evidence suggests that DNA methylation could be an important mechanism in regulating the Ca²⁺-signaling genes in cancer.

Cancer is also characterized by alterations in HPM. These modifications alter the nucleosome function and regulate gene expression by controlling the degree of chromatin compaction (Li and Seto 2016). However, very little is known regarding the role of HPM on the transcriptional regulation of the Ca²⁺-signaling genes under healthy and pathological conditions. Our analysis showed a good correlation between chromatin marks and gene expression in HepG2 cells compared with hepatocytes. Euchromatin marks accumulated in the up-regulated genes and decreased in down-regulated genes, except RGN, which showed an increase in all histone marks. MELK is regulated in mouse embryonic fibroblasts by H3K9ac and H3K27ac (Sheikh et al. 2015). However, there is no information on HPM of the other altered Ca²⁺-signaling genes. Thus, our analysis provides evidence of the potential role of the HPM in the transcriptional regulation of the Ca²⁺-signaling genes.

Some epigenetic drugs that target HPM have been tested in clinical trials or implemented for cancer therapy (Li and Seto 2016). SAHA and TSA treatment modulated the expression of the altered Ca²⁺-signaling genes in HepG2 cells. FYN expression increases, whereas MASP2 and RGN decreased under both conditions. Moreover, STX1A expression increased with SAHA, whereas PDE2A and RAPGEF expression increased with TSA. FYN, PDE2A, RAPGEF2, and STX1A up-regulation could be partially explained due to a lower H3K27ac at their promoters in untreated HepG2 cells compared with hepatocytes. Euchromatin marks were also increased at the RGN promoter of untreated HepG2 cells compared to hepatocytes; however, after TSA or SAHA treatment, its expression decreased. Added to this, CPS1 and MELK expression decreased with SAHA, whereas LPCAT1 and C1S decreased with TSA. Therefore, SAHA and TSA treatment negatively regulate the expression of these four Ca²⁺-signaling genes. Decreased gene expression after HDACi treatment could be explained through the inhibition of transcription elongation mediated by these molecules. HDACs are located at gene bodies and intergenic regions, and from there, they remove the acetylation marks, which in turn regulates the recruitment of elongation factors (Greer et al. 2015). Very little is known about HDACi and the expression of the Ca²⁺-signaling genes. SERCA3 (ATP2A3) expression is increased and regulated by H3K9ac and H3K27ac in breast, gastric, and HCC cancer cell lines treated with HDACi (Meneses-Morales et al. 2019; Izquierdo-Torres et al. 2019; Hernandez-Oliveras et al. 2019; Contreras-Leal et al. 2016). PMCA4b expression also increased in MCF-7 cells treated with SAHA and short-chain fatty acids (Varga et al. 2014). It has also been shown that Ca²⁺-signaling and histone modifications are tightly connected. Ca²⁺ flux through PKD2 activates CaMKII, which regulates the nuclear localization of HDAC4 (Rothschild et al. 2018). Nevertheless, further studies will be necessary to determine the role of HDACi regarding the expression of most Ca²⁺-signaling genes and their potential application in clinical trials.

The liver is unique, as it harbors the ability to regenerate, involving cellular mechanisms such as cell proliferation. During liver regeneration, Ca²⁺ signaling is remodeled, mainly through Ca²⁺-activated transcription factors, such as CREB, JUN, NF-kB, among others (Oliva-Vilarnau et al. 2018). Also, HCC and liver regeneration models, such as partial hepatectomy, showed similar expression profiles of genes associated with cell division and share biological processes such as cell proliferation, angiogenesis, and fatty acid metabolism, where genes such as FYN have a key role (Yin et al. 2018). Gene expression analysis of the altered Ca²⁺-signaling genes in a rat hepatectomy model showed that Reps2, Dnase1l3, Rgn, Pon1, and Pde2a expression decreased, whereas Masp2, Pla2g12a, Melk, Rapgef2, and Fyn increased after partial hepatectomy. Only Ect2 expression showed a fluctuation, as it decreased at 12 h, and increased after this timepoint. Pon1 expression, which is an important protein in the biotransformation of acetaldehyde and organophosphorus, decreased after partial hepatectomy (Sun et al. 2007). The oncogene ECT2 increased after partial hepatectomy in mice, showing high levels in mitotic cells, whereas FYN expression increased at 72 h after partial hepatectomy (Yin et al. 2018; Sakata et al. 2000). RAPGEF2 is important in fetal liver, particularly for hematopoiesis through RAP1 activation (Satyanarayana et al. 2010).

Also, we analyzed the expression and promoter methylation of the altered Ca²⁺-signaling genes in an in vitro human hepatocyte differentiation model. Results from MH showed that MELK, ECT2, PDE2A, LPCAT1, FYN, RAMP3, and CLEC3B expression decreased, whereas C1S, RGN, CPS1, MASP2, STX1A, FOSB, DNASE1L3, CCL23, and ANXA10 expression increased, compared with hESC. MELK, ECT2, and LPCAT1 expression is up-regulated in HCC and decreased in differentiated hepatocyte. All up-regulated genes in MH are decreased in HCC samples. Therefore, these results suggest that the expression of some altered Ca²⁺-signaling genes in HCC is modulated during hepatocyte differentiation. Interestingly, using a single-cell RNA-seq database, we found that hepatocytes isolated from liver tissue and in vitro differentiated hepatocytes shared similar expression patterns of the altered Ca²⁺-signaling genes. These results also suggest that the altered genes have liver cell-specific expression, since cellular populations displayed different expression patterns, probably due to their specialized cellular functions. Despite the evidence, further investigations are needed to fully understand the potential role of the altered Ca²⁺-signaling genes in liver biology and HCC.

In conclusion, our analyses provide evidence for previously unreported genes with potential clinical relevance, whose expression is altered, and confirm genes previously reported to be altered in HCC. We have also shown that DNA methylation and HPM associates with the expression of some of these genes in HCC samples and in HepG2 cells. The expression of the altered Ca²⁺-signaling genes was also modulated during liver regeneration, hepatocyte differentiation, and showed cell-specific expression in liver cell populations, suggesting that their function in the liver is more important than previously thought. Together these results contribute to a better understanding of the role of Ca²⁺-signaling genes in HCC and liver biology.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author’s contributions

AH-O conception and design of the study, performed computational analysis, discussed and interpreted the data, wrote the first draft, prepared tables and figures, supervised the study. EI-T and GH-M conducted data analysis, prepared tables and figures. AZ-H data interpretation, review and edited the manuscript. JS-G conception and design of the study, data interpretation, acquired funding, wrote and edited the manuscript.

Funding

This study was funded by Consejo Nacional de Ciencia y Tecnología, México (Grant No. PDCPN2015-1518) to JS-G.

Availability of data and material

Previously published RNA-seq data originated from: Roessler et al. (2010) (GSE14520), Lim et al. (2013) (GSE36376), Wurmbach et al. (2007) (GSE6764), Hart et al. (2010) (GSE18269), Liu et al. (2018a, b) (GSE105065), Chittur et al. (2008) (GSE4465), Lee et al. (2014) (GSE52232), Kim et al. (2011) (GSE25046), and GSE63742 dataset has no associated reference. Previously reported WGBS data originated from: The International Human Epigenome Consortium (URLs available in Supplementary Table 2), Liu et al. (2018a, b) (GSE10566-GPL8490), Kim et al. (2011) (GSE25047-GPL8490). Previously reported ChIP-seq data originated from Cistrome DB (URLs available in the Supplementary Table 2). Previously reported single-cell RNA-seq data originated from MacParland et al. (2018) (GSE115469).

Compliance with ethical standards

Conflict of interest

There are no conflicts of interests to disclose.