Pralatrexate represses the resistance of HCC cells to molecular targeted agents via the miRNA-34a/Notch pathway

Abstract

Metabolism-related pathways are important targets for intervention in the treatment of hepatocellular carcinoma (HCC), but few studies have reported on the combination of inhibitors of folate metabolism-related enzymes and molecularly targeted drugs for HCC. The results of the present work are the first to reveal the effects of an inhibitor of dihydrofolate reductase (DHFR), pralatrexate, on the sensitivity of HCC cells to molecularly targeted agents examined using multiple assays. In HCC cells, knockdown of DHFR or treatment with pralatrexate enhanced the sensitivity of HCC cells to molecularly targeted agents, such as sorafenib, regorafenib, lenvatinib, cabozantinib, or anlotinib. Mechanically, pralatrexate decreased the methylation rates of miRNA-34a’s promoter region to enhance the expression of miRNA-34a. Treatment with pralatrexate inhibited the expression of Notch and its downstream factors by enhancing the expression of miRNA-34a in HCC cells. In clinical specimens, the expression of miRNA-34a was negatively correlated with DHFR expression, while DHFR expression was positively correlated with the Notch intracellular domain (NICD) and downstream factors of the Notch pathway. The expression of miRNA-34a was negatively correlated with DHFR expression, while the methylation rates of miRNA-34a’s promoter were positively related to DHFR. The effect of pralatrexate on the metabolic profile of HCC cells is very different from that of small molecule inhibitors related to glycolipid metabolism. Therefore, pralatrexate upregulates the sensitivity of HCC cells to molecularly targeted drugs. These results expand our understanding of folate metabolism and HCC and can help provide more options for HCC treatment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-024-01572-2.

Introduction

Hepatocellular carcinoma (HCC) is not only one of the most common and deadly tumors of the digestive system, but it also has one of the highest morbidity and mortality rates of malignant tumors in humans [1]. HCC is the most predominant pathological subtype of liver tumors and poses a major challenge to public health systems in China and East Asia, as its most significant risk factor is acute and chronic liver disease caused by various hepatitis viruses, such as HBV (Hepatitis B virus) or HCV (Hepatitis C virus) [2, 3]. Currently, one of the main antitumor treatment strategies for the majority of HCC patients who are initially diagnosed with progressive disease is a variety of molecularly targeted drugs (i.e., oral multi-targeted small molecule protein kinase inhibitors) [4]. However, the resistance of HCC cells to these drugs affects and limits their clinical benefits [4]. The miRNA-34a/Notch pathway is an important regulator of resistance to molecularly targeted drugs in HCC cells [5–9]. The miRNA-34a promoter methylation can lead to the reduced expression or deletion of miRNA-34a (miRNA-34a is an miRNA that acts on the Notch protein), resulting in high Notch protein expression and overactivation of the Notch pathway, ultimately inducing resistance to molecularly targeted drugs [5, 10–12]. Research on the miRNA-34a/Notch pathway is therefore important for achieving safer and more effective molecularly targeted therapies for HCC.

The close relationship between the proliferation, survival, metastasis, invasion, and drug resistance of malignant tumor cells and metabolism-related mechanisms and malignant cells represented by HCC has been the focus of recent research [13–15]. Studies have focused on glucose metabolism, lipid metabolism, and nuclear receptors, such as pregnane X receptors (PXRs) [16]. This study is the first to report on the relationship between a core regulator of folate metabolism, dihydrofolate reductase (DHFR), and molecularly targeted drug resistance in HCC cells. Folate metabolism is not only an important mechanism of nutrient metabolism in HCC cells but is also closely related to macromolecular synthesis [17, 18]. Additionally, folate is the only source of intracellular 1C units, and blocking folate metabolism can inhibit DNA methylation mechanisms in cells [17, 18]. In this study, pralatrexate, a small molecule inhibitor of DHFR, was found to inhibit the methylation of the miRNA-34a promoter region, upregulate the expression of miRNA-34a, and ultimately downregulate the expression of Notch in HCC cells. Furthermore, pralatrexate was able to downregulate the expression levels of downstream drug resistance-associated factors and ultimately upregulate the sensitivity of HCC cells to a variety of molecularly targeted drugs. The present study not only expands our understanding of folate metabolism and resistance to molecularly targeted drugs in HCC cells but also helps to provide new options for molecularly targeted therapies.

Materials and methods

Clinical specimens

A total of 71 patients (from 2015 to 2019) with HCC were recruited for the present work (Table 1), and the methods and protocols were all approved by the Ethics Committee of the 920th Hospital of the PLA Joint Logistic Support Force, Kunming City 650032, Yunnan Province, P.R. China. The usages, and the experiments related these clinical specimens were performed in the 920th Hospital of the PLA Joint Logistic Support Force, Kunming City 650032, Yunnan Province, P.R. China. This study was conducted in accordance with the Declaration of Helsinki, WHO (all patients included in this work provided written informed consent before treatment). The patients underwent a hepatectomy carried out by one group of hepatobiliary surgeons at our hospital (the 920th Hospital of the PLA Joint Logistic Support Force, Kunming City 650032, Yunnan Province, P.R. China). Information on the patients is presented in Table 1. Surgically removed tissue specimens were separated from HCC tissue and para-cancerous tissue (the paired non-tumor tissues as control) via microdissection. The clinical tissues were preserved in liquid nitrogen, and samples of medium DNA and total RNA were extracted separately. The lentivirus of the Notch intracellular domain (NICD), miRNA-34a (hsa-pre-miRNA-34a), siDHFR, or DHFR were gifts Yingshi Zhang from Shenyang Pharmaceutical University and some of these have been used and described in our previous publications [19–21].

Table 1.

Baseline clinical data of 71 patients involved in the presence work

Characters	Number (%)
Median age, year (range)	54 (27–65)
Gender, male (%)	66 (92.93%)
Aetiology (%)
HBV positive	58 (81.69%)
HCV positive	4 (5.63%)
BCLC stages
A	63 (88.73%)
B	8 (11.27%)
ECOG PS (%)
0	35 (49.30%)
1	32 (45.07%)
2	4 (5.63%)
AFP (%)
Normal	16 (16.90%)
Elevated	55 (77.46%)
Extrahepatic metastasis (%)	2 (2.82%)
LN metastasis (%)	6 (8.45%)
Portal vein invasion (%)	–
Chilg-Pugh (%)
A	58 (81.69%)
B	13 (18.31%)
C	–
Median size of index tumor, cm (range)	3.5 (1.8–4.9)
Median number of index tumors	2 (1–4)
Prior local therapy (%)	None

Cell lines, agents, and cell survival examination

The HCC cell line: MHCC97-H, HepG2, LMC-3 and HepG2 2.2.15 was purchased from the National Infrastructure of Cell Line Resources, Academy of China Medicine, which houses the standard biological resources of the Chinese government. The cell lines were cultured using DMEM (Dulbecco's Modified Eagle's Medium) with 10% FBS (Fetal Bovine Serum). The DMEM and FBS were purchased from the Thermo Fisher Scientific Corporation (Waltham, MA, USA). The agents adriamycin (Cat. No. E2516), BAY-876 (Cat. No. S8452), fatostatin (Cat. No. S9785), rifampicin (Cat. No. S1764), pralatrexate (Cat. No. S1497), ketoconazole (Cat. No. S1353), sorafenib (Cat. No. S7397), regorafenib (Cat. No. S1178), lenvatinib (Cat. No. S1164), apatinib (Cat. No. S2221), cabozantinib (Cat. No. S1119), and anlotinib (Cat. No. S8726) were purchased from Selleck Corporation, Houston, Texas, USA. The agents were dissolved using dimethyl sulfoxide and diluted using DMEM. The MHCC97-H cells were cultured in DMEM with 10% FBS and treated with the indicated concentration of antitumor drugs for 48 h. The cells were then treated with the MTT agent (Thermo Fisher Scientific Corporation, Waltham, MA, USA) [22, 23]. The relative colony number was reflected by the optical density (OD) at 400 nm, and the inhibition rates of agents were calculated as follows: (OD 490 nm of the control group–OD 490 nm of the drug-treatment group)/(OD 490 nm of the control group) × 100%. The half maximal inhibitory concentration (IC₅₀ values) values of the molecular targeting agents were calculated using the methods described in previous publications [22, 23].

BSP, NGS, and AmpliSeq

For the bisulfite sequencing PCR (BSP) and next-generation sequencing (NGS), the genomic DNA of clinical tissues and cultured cells was isolated using the DNeasy Blood & Tissue Kit (Cat No. 69504; QIAGEN, Düsseldorf, Nordrhein-Westfalen, Germany) and treated with the EpiTect Bisulfite Kit (Cat No. 59104; QIAGEN, Düsseldorf, Nordrhein-Westfalen, Germany) according to the instructions provided by the manufacturer. Polymerase chain reaction (PCR) using the Platinum II Hot-Start PCR Master Mix (Cat No. 14000012; Thermo Fisher Scientific, Waltham, MA, USA) was performed for the amplification of the selected promoter region (chr1: 9152409–9152574) of miR-34a. The selected promoter region contains 7 CpG sites [20]. A schematic of the miRNA-34a promoter region selected for this segment and the CpG sites/islands contained therein is shown as Fig. 1. The primers used in the present study are listed in Supplemental Table 1. The PCR products were directly sequenced using the Ion Torrent PGM, Life Technologies, Thermo Fisher Scientific, Waltham, MA, USA. The methylation rates were calculated according to previous publications. For AmpliSeq, Ion AmpliSeq Library Kit 2.0 and custom panels (Cat. No. 4480441; Thermo Fisher Scientific, Waltham, MA, USA) were utilized in the construction of an RNA library in accordance with the instructions provided by the manufacturer. RNA (10 ng) isolated from cultured cells or clinical tissues was reverse transcribed, and synthesized cDNA was amplified using Ion AmpliSeq RNA custom panels and a Library Kit (Cat. No. 4480441; Thermo Fisher Scientific, Waltham, MA, USA). The primers used in the present study are listed in Supplemental Table 1. Ma et al.’s (2020) procedure for calculating methylation rates was used [20]. Figure 1 indicates the selected miRNA-34a’s promoter region and the CpG islands. The AmpliSeq library was purified using AMPure XP beads (Cat. No. A63881; Beckman Coulter, Atlanta, GA, USA). Qubit 2.0 (Qubit HS DNA kit [Invitrogen, Thermo Fisher Scientific, Waltham, MA, USA]), Ion 318 Chip (Cat. No.4488150; Thermo Fisher Scientific, Waltham, MA, USA), Torrent Suite 5.6 and Ion Reporter 5.6 software (Life-technology, Thermo Fisher Scientific), and Ion Torrent PGM (Cat. No.4462921; Thermo Fisher Scientific, Waltham, MA, USA) were used to perform the sequencing. The related expression level (folds of control) and the heatmap images were calculated according to previous publications [24, 25]. The target genes’ panel includes lipid metabolism-related genes, glucose metabolism-related genes, metabolism-related drug resistance genes, and Notch downstream genes (including some cellular pro-survival and anti-apoptotic factors and epithelial–mesenchymal transition [EMT] regulators) [20, 24, 25]. The primers used in the present study are listed in Supplemental Table 1. For AmpliSeq, three replicate wells were set up for each sample, and three biological replicates (i.e. the same qPCR experiment was done three times) were performed by Housekeeping gene (endogenous control) using β-Actin to standardise the samples for relative quantification of genes’ expression.

Fig. 1.

The CpG island of the selected region of miRNA-34a promoter. In this study, the segment chr1: 9152409–9152574, containing seven CpG sites, was selected based on a method by Ma (Pharmacol Res. 2020 Oct;160: 105071)

Biochemical examination

Metabolic examinations were carried out according to previous publications to determine the effects of the agents on the cells’ metabolic features. Glycolytic activity examinations were performed, and glucose uptake (Glucose Uptake Assay Kit [Colorimetric], [ab136955], Abcam, Cambridge, UK), lactate (Lactate-Glo Assay Kit, Promega), ATP (ATP Assay Kit (colorimetric/fluorometric) (ab83355), Abcam, Cambridge, UK) production, and LDH activity (Lactate Dehydrogenase Activity Assay Kit, Cat. No. MAK066, Sigma-Aldrich, Merck Corporation, Darmstadt, Rhine-Main, Germany) were assessed. Based on these values, the inhibition rate was calculated as follows: (control group’s biochemical values–experimental group’s biochemical values)/(control group’s biochemical values) × 100%. The results are shown as heatmap images.

Drug metabolism rate assay

MHCC97-H cells were obtained in culture, and cell samples were collected at serial time points (0, 4, 8, 12, 24, and 48 h) after treating the cells with a 1 μmol/L dose of sorafenib for approximately 12 h or with a 0.3 μmol/L dose of adriamycin for approximately 12 h. After vortex shaking fragmentation of the cell samples, the drug was extracted from the cell fragmentation products using the organic solvent acetonitrile for detection by liquid chromatography-mass spectrometry (LC–MS/MS). The methodology for LC–MS/MS was based on the previous publications. LC–MS/MS makes it possible to determine the amount of sorafenib [26] or adriamycin [27] in a cell sample at each time point. Based on this amount, a retention curve was fitted to obtain the half-life values (t_1/2 values) of the drugs in the cells [28].

The in vivo tumor model

MHCC97-H cells were obtained in culture, and cell were collected to prepare as single-cell suspension. Single-cell suspensions were inoculated directly under the skin of nude mice to form subcutaneous tumours in nude mice (5-week-old female mice, housed under SPF conditions, BalB/c strain thymus-deficient mice, purchased from Beijing Si-Bei-Fu Biotechnology Co.). Formulation for oral administration (formulation for oral administration) of Sorafenib and Pralatrexate was configured (the drug powder was weighed using a 1/100,000 precision balance, dissolved using the organic solvent dimethylsulfoxide, and thereafter diluted using sterilised saline containing polyethylene glycol and Tween 80, and finally (ensuring that the content of all these organic solvents was less than 2 parts per thousand), the animals were administered the drug every other day for about 15 doses (oral administration), and at the end point the animals were collected and tumour tissues were obtained, measured and weighed [29, 30]. Tumour tissues were numbered (each animal formed one tumour tissue, i.e., 10 tumour tissues per group), and total protein samples were extracted from tumour tissues No. 2, 4 and 8 for protein immunoblotting. At the same time, all tumour tissues were subjected to BSP-NGS as well as AmpliSeq analysis for thermograms (heat map).

Western blot and cellular sub-faction

In this study, there were two main types of samples for western blot, one was cultured MHCC97-H cells and the other was subcutaneous tumour tissue from nude mice. For MHCC97-H cells, the MHCC97-H cells were cultured and transfected with vectors. The cell samples were collected and mixed thoroughly with SDS-PAGE uploading/lysis buffer in a boiling water bath, then centrifuged at 4 °C 12000 rpm, and the supernatant was collected as the total protein sample. Thereafter, the collected samples were sequentially subjected to SDS-PAGE, membrane transfer (PVDF membrane) and closure. Thereafter, the PVDF membrane was sequentially incubated with primary antibody (monoclonal antibody against DHFR and loading control GAPDH) and secondary antibody, and finally, chemiluminescence was performed. For subcutaneous tumours in nude mice, the tumour tissue was fully soaked with RIPA lysate, then cut into pieces using ophthalmic forceps and ophthalmic scissors, then mixed with SDS-PAGE uploading/lysis buffer and subjected to a boiling water bath (at this time, a small amount of 5% SDS was added additionally to the samples), and then centrifuged at 4 °C at 12,000 rpm, and the supernatant was collected to be the total protein samples. Thereafter, the collected samples were sequentially subjected to SDS-PAGE, membrane transfer (PVDF membrane) and closure. Thereafter, the PVDF membrane was sequentially incubated with primary antibody (monoclonal antibody against NICD [Notch 1 monoclonal antibody with antigenic epitope at the C-terminus, capable of recognising and detecting the NICD shear body of Notch protein, kindly provided by zhang yingshi, Shenyang Pharmaceutical University.] and loading control GAPDH) and secondary antibody, and finally, chemiluminescence was performed.

Nucleoplasmic separation/sub-fraction of cells was studied on the basis of Western blot [31]. MHCC97-H cells were obtained in culture and treated with 2 μmol/L dose of Pralatrexate for 30 min, 1 h, and 2 h after which the cells were collected and broken using sonication. Thereafter, the samples were centrifuged at 4 degrees Celsius at 800 rpm for about 3 min to obtain the nuclear fraction, and then centrifuged at 4 degrees Celsius at 12,000 rpm for about 10 min to obtain the cytoplasmic fraction. The two fractions were thoroughly mixed with SDS-PAGE upsampling buffer, followed by boiling water bath for about 15 min, and finally they were subjected to western blot. NICD in the nucleus was detected using Notch1 antibody (C-10) (sc-373891 of Santa Cruz, USA); Notch 1 protein in the cytoplasm was also detected using Notch1 antibody (C-10) (sc-373891 of Santa Cruz, USA) via differentiation by molecular weight size (Notch 1 full-length protein exceeds the 220 kDa marker; NICD is near the 110 kDa marker). β-Actin (β-Actin antibody (2A3) [sc-517582]) was selected as an indicator of cytoplasm and Lamin A (Lamin A antibody (4A4) [sc-517581]) was selected as an indicator of nucleus.

In the WB experiment, SDS-PAGE was performed with the position of the leading edge of bromophenolan, and the pre-stained Marke (the main bands were 220, 110, 78, 53, 37, and 21 kDa) was used as a reference for the transmembrane experiments; after the completion of the transmembrane experiments, PVDF membranes near the corresponding protein band were cut down according to the position of the pre-stained marker. After the membrane transfer experiment was completed, the PVDF membrane near the corresponding protein bands was cut out according to the pre-stained marker position, and the antibody was incubated in the antibody incubation bag. Unmodified original photographs were obtained by scanning X-ray film or fluorescence imaging.

Ethics declarations

The collection and usage of clinical specimens in present work, were all approved by the Ethics Committee. This study was conducted in accordance with the Declaration of Helsinki, WHO (all patients included in this work provided written informed consent before treatment).

Statistical analysis

Statistical analyses were performed using SPSS software (software version 9.0, IBM Corporation, Armonk, NY, USA). Data analysis for comparisons between groups was performed using Bonferroni’s correction with two-way analysis of variance (ANOVA) methods. The IC₅₀ values of the antitumor drugs’ effects on MHCC97-H cells or the t1/2 of the drugs in cultured cells were calculated using GraphPad Prism 8.0 software. A P < 0.05 was considered statistically significant.

Results

DHFR modulates the sensitivity of HCC cells to molecularly targeted agents

First, to examine whether DHFR could modulate the sensitivity of HCC cells to molecularly targeted agents, MTT assays were performed. The results showed that the overexpression of DHFR enhanced the resistance of MHCC97-H cells to sorafenib, a typical molecularly targeted agent of HCC: the IC₅₀ value of sorafenib in MHCC97-H cells increased from 1.02 μmol/L (control) to 5.38 μmol/L (DHFR overexpression). When DHFR was knockdown, the IC₅₀ value of sorafenib in MHCC97-H cells increased from 1.11 μmol/L (Control siRNA) to 0.13 μmol/L (siDHFR, the small interference RNA of DHFR) (Table 2). To further examine the effect of DHFR on HCC cells, the MTT assays were performed in some other HCC cells (LMC-3, HepG2 and HepG2 2.2.15). As shown in Fig. 2, overexpression of DHFR induced the resistance of HCC cells to sorafenib, the IC₅₀ values of sorafenib on HCC cells increased from 0.91 (LMC-3 cells) (Fig. 2A), 1.16 (HepG2 cells) (Fig. 2B) and 1.25 (HepG2 2.2.15 cells) (Fig. 2C) to NA (in MTT experiments, 50% inhibition was not achieved at the maximum concentration [3 µmol] of sorafenib). Knockdown of DHFR via its siRNA enhanced the sensitivity of HCC cells to sorafenib, the IC₅₀ values of sorafenib on HCC cells increased from 0.91 µmol (LMC-3 cells) (Fig. 2A), 1.16 µmol (HepG2 cells) (Fig. 2B) and 1.25 µmol (HepG2 2.2.15 cells) (Fig. 2C) to 0.025 µmol (LMC-3 cells), 0.037 µmol (HepG2 cells) and 0.013 µmol (HepG2 2.2.15 cells).

Table 2.

DHFR modulates the sensitivity of HCC cells to molecularly targeted agents

Groups	IC50 values of Sorafenib on MHCC97-H cells (μmol/L)
Control	1.02
DHFR	5.38
siDHFR	0.13
DHFR + miRNA-34a	0.25
siDHFR + NICD	5.67

Fig. 2.

DHFR modulates the sensitivity of HCC cells to sorafenib. HCC cells MHCC97-H (A), HepG2 (B), HpeG2 2.2.15 (C) were cultured and transfected DHFR to overexpress DHFR and siDHFR to knockdown DHFR’s expression. The cells were treated with the indicated concentrations of sorafenib and for MTT assays. The effect of DHFR overexpression or knockdown on the antitumor activation of sorafenib on HCC cells were shown as effect-dose curves and IC₅₀ values. The DHFR level in HCC cells were measured by the western blot

Next, the effects of DHFR on the Notch pathway were examined. As shown in Fig. 3A–E, overexpression of DHFR increased the promoter methylation of miRNA-34a (Fig. 3A, B) (A schematic of the miRNA-34a promoter region selected for this segment and the CpG sites/islands contained therein is shown as Fig. 1), decreased the expression of miRNA-34a (Fig. 3A, B), and enhanced the expression of the NICD (Fig. 3C) and downstream factors of the Notch pathway (N-cadherin and survivin) (Fig. 3D, E). However, knockdown of DHFR decreased the promoter methylation of miRNA-34a (Fig. 3A, B), increased the expression of miRNA-34a (Fig. 3A, B), and decreased the expression of the NICD (Fig. 3C) and downstream factors of the Notch pathway (N-cadherin and survivin) (Fig. 3D, E). Further overexpression of NICD or overexpression of miRNA-34a was performed in the presence of DHFR overexpression or knockdown. The results showed that overexpression of miRNA-34a under the premise of DHFR overexpression, at this time, DHFR was also able to up-regulate the methylation of the promoter region of miRNA-34a, but could not up-regulate the expression of NICD, N-Cadherin or suvivin (Fig. 3A–E). The expression level of DHFR was shown as Fig. 3F. At the same time, overexpression of NICD under the premise of knocking down the expression of DHFR, at this time, knocking down of DHFR was still able to down-regulate the methylation rates of the promoter of miRNA-34a, and up-regulate the expression of miRNA-34a, but could not downregulate the expression level of NICD, N-Cadherin or suvivin (Fig. 3A–E).

Fig. 3.

DHFR enhanced the activation of Notch pathway via miRNA-34a. MHCC97-H cells were obtained in culture and transfected with control, DHFR, siDHFR, miRNA-34a, NICD or SiDHFR + NICD or DHFR + miRNA-34a, respectively. thereafter, cells were collected and total DNA samples were extracted separately for BSP-NGS assay; total RNA samples were extracted for AmplySeq assays were performed to detect miRNA-34a expression (A), miRNA-34a promoter region methylation (B), NICD expression (C), N-Cadherin expression (D), Survivin expression (E), and DHFR expression (F) in the HCC cells. (G) The MHCC97-H cells were transfected with vectors and for western blot to measure the expression of DHFR or NICD. The groups included control, DHFR, siDHFR, DHFR + miRNA-34a and siDHFR + NICD (G). Results are shown as bar graphs (A-F), and images of western blot. In the WB experiment, SDS-PAGE was performed with the position of the leading edge of bromophenolan, and the pre-stained Marke (the main bands were 220, 110, 78, 53, 37, and 21 kDa) was used as a reference for the transmembrane experiments; after the completion of the transmembrane experiments, PVDF membranes near the corresponding protein band were cut down according to the position of the pre-stained marker. After the membrane transfer experiment was completed, the PVDF membrane near the corresponding protein bands was cut out according to the pre-stained marker position, and the antibody was incubated in the antibody incubation bag. Unmodified original photographs were obtained by scanning X-ray film or fluorescence imaging. * p-value < 0.05 compared with the control group. * p-value < 0.05 compared with control group

Next, the western blot was performed. As shown in Fig. 3G, transfection of DHFR vectors enhanced the expression level of DHFR in MHCC97-H cells, enhanced the expression level of DHFR and NICD. Knockdown of DHFR (siDHFR) repressed the expression level of DHFR and NICD. On this basis, miRNA-34a was overexpressed at the same time as DHFR overexpression, at which time DHFR overexpression was unable to up-regulate the expression level of NICD; and NICD was overexpressed at the same time as knockdown of DHFR, at which time siDHFR was not able to knockdown the expression of NICD. Moreover, in conjunction with the DHFR/Notch pathway data, we also further tested the specificity of DHFR/Notch pathway-related factors affecting the killing of MHCC97-H cells by sorafenib (Table 2). The results are shown in Table 2: transfection of miRNA-34a while DHFR was overexpressed, DHFR was no longer able to induce resistance to sorafenib in MHCC97-H cells; at the same time, overexpression of NICD while DHFR was knocked down, at this time siDHFR was not able to up-regulate the sensitivity of MHCC97-H cells to sorafenib (Table 2). Therefore, DHFR could modulate the sensitivity of HCC cells to molecularly targeted agents via the Notch pathway.

Pralatrexate enhanced the sensitivity of HCC cells to molecularly targeted agents

Next, the effects of pralatrexate, a typical small inhibitor of DHFR, on the sensitivity of HCC cells to molecularly targeted agents were examined. As shown in Fig. 4, treatment with pralatrexate significantly enhanced the antitumor activation of the molecularly targeted agents sorafenib (Fig. 4A), regorafenib (Fig. 4B), cabozantinib (Fig. 4C), anlotinib (Fig. 4D), apatinib (Fig. 4E), lenvatinib (Fig. 4F), and donafenib (Fig. 4G) on HCC cells. The combination of these drugs with pralatrexate resulted in the significant downregulation of the IC₅₀ values of MHCC97-H cells compared to the control group, and the higher the dose of pralatrexate, the more significantly the IC₅₀ values of the molecular targeted agents were downregulated (Fig. 4A–G). These results indicate that pralatrexate enhanced the sensitivity of HCC cells to molecularly targeted agents.

Fig. 4.

Pralatrexate enhanced the sensitivity of MHCC97-H cells to molecular targeted drugs. MHCC97-H cells were obtained in culture, after which MHCC97-H cells were pretreated with solvent control (control [solvent control]), 0.5, 1 and 2 μmol/L doses of Pralatrexate for approximately 6–8 h, followed by a series of concentration gradients (10, 3, 1, 0.3, 0.1, 0.03 and 0.01). L, 1, 0.3, 0.1, 0.03, and 0.01 μml/L) of molecularly targeted drugs (sorafenib, regorafenib, Anlotinib, Apatinib, Carbozantinib, and Lenvatinib) (A–G) for approximately 8 h. The cells were then treated with a series of concentration gradients (10, 3, 1, 0.3, 0.1, 0.03, and 0.01 μml/L) of Cells were treated for approximately 48 h, after which the inhibition of cell survival by the different groups of drugs was measured by MTT assay. A curve was fitted based on the inhibition rate and IC₅₀ values were obtained

Pralatrexate repressed the activation of the Notch pathway

Based on the previous results, we examined the expression of pralatrexate on HCC resistance-related factors and found that pralatrexate downregulated the expression of the cell pro-survival and anti-apoptosis-related factors of survivin, cIAP1, and cIAP2 (Fig. 5A), downregulated the expression of the EMT-related factors N-cadherin, Vimentin, Twist, and Snail (Fig. 5A), and upregulated the expression of E-cadherin (Fig. 5B). Pralatrexate also inhibited the expression of drug metabolism-related resistance genes mdr-1, cyp3a4, and bcrp in MHCC97-H cells, but the effects of pralatrexate on the expression of these genes were weaker than their effects on the expression of pro-survival, anti-apoptosis-related and EMT-related factors (Fig. 5A). The effects of pralatrexate on NICD expression were further examined, and the results showed that pralatrexate inhibited the expression of NICD (Fig. 5A). Furthermore, pralatrexate did not affect the expression levels of glycolysis-related factors HIF-1α and GLUT1, nor did it affect the expression levels of fatty acid synthesis-related factors (ACC, ACLY, FASN, or ACS) (Fig. 5A). Thus, pralatrexate was able to inhibit Notch pathway activity in HCC cells.

Fig. 5.

The effect of drugs on drug-resistance related factors in MHCC97-H cells. MHCC97-H cells were obtained in culture and thereafter treated with 2, 1, 0.5 doses of Pralatrexate; 1, 0.1 doses of Fatostatin; 1, 0.1 μmol/L doses of BAY-876, and then by AmplySeq to detect the expression level of the target gene Panel. The genes involved were (1) NICD and Notch downstream pro-survival anti-apoptotic related factors cIAP-1, cIAP-2 and Survivin; EMT related factors N-Cadherin, Vimentin, E-Cadherin, Snail and Twist; (2) drug metabolism and clearance related drug resistance genes The results were based on the expression levels of these factors (1 in the control group and the corresponding expression levels in each group were upregulated or downregulated relative to Control [folds of control]) in heat maps or bar charts

Fatostatin (a typical inhibitor of fatty acid synthesis/a typical inhibitor of SREBP-1) and BAY-876 (a typical inhibitor of glucose consumption/a typical inhibitor of GLUT1) were also used as important controls. As expected, fatostatin inhibited the expression of EMT-related factors and fatty acid synthesis-related factors (ACC, ACLY, FASN, or ACS) but did not affect the expression of pro-survival/anti-apoptosis-related factors and drug metabolism-related drug resistance factors (Fig. 5A). Fatostatin also inhibited the expression of glycolysis-related factors HIF-1α and GLUT1 to a certain extent (Fig. 5A). BAY-876 could inhibit the expression of EMT-related factors and decrease the expression levels of glycolysis-related factors HIF-1α and GLUT1; however, it did not affect the expression of fatty acid synthesis-related factors, pro-survival/anti-apoptosis-related factors, and drug metabolism-related drug resistance factors (Fig. 5A). Treatment of pralatrexate, Fatostatin and BAY-876 enhanced the expression of E-Cadherin in MHCC97-H cells (Fig. 5B). These results further confirmed the specificity of pralatrexate’s effects.

The effects of pralatrexate on the metabolic features of HCC cells

Based on evidence that folate metabolism is important for HCC cells, the effects of pralatrexate on the metabolic features of HCC cells were examined. As shown in Fig. 6, treatment with pralatrexate did not inhibit the Warburg effect of HCC cells compared to BAY-876 or fatostatin. Moreover, as shown in Table 3, pralatrexate treatment affected the metabolism of sorafenib in HCC cells and decelerated the metabolism or clearance of sorafenib in MHCC97-H cells; the t_1/2 of sorafenib was prolonged. Compared to rifampicin, a typical agonist of PXR, or ketoconazole, a typical antagonist of PXR, the effect of pralatrexate on the half-life of sorafenib in MHCC97-H cells was much weaker. However, the corresponding rifampicin treatment significantly shortened the half-life of sorafenib in MHCC97-H cells by accelerating the metabolism and clearance of sorafenib (Table 3). Furthermore, ketoconazole significantly prolonged the half-life of sorafenib in MHCC97-H cells by slowing the metabolism and clearance of sorafenib (Table 3). A similar trend was seen with these drugs acting on adriamycin (a representative of cytotoxic chemotherapeutic drugs for HCC) (Table 3). As an important control, fatostatin and BAY-876 were not affected by the expression of metabolism-related resistance genes or by the rate of drug metabolism (Table 3). This suggests that the effects of pralatrexate on EMT occurred via the Notch pathway rather than affecting HCC cellular glycolipid metabolism, but the mechanism of pralatrexate’s effects on PXR downstream resistance genes is unclear.

Fig. 6.

The effect of drugs on the metabolic feature of MHCC97-H cells. MHCC97-H cells were obtained in culture and thereafter treated with 2, 1, 0.5 doses of Pralatrexate; 1, 0.1 doses of Fatostatin; 1, 0.1 μmol/L doses of BAY-876, and then by biochemical analysis to detect the metabolic feature of MHCC97-H cells. The selected metabolism-related features were: ATP content, glucose uptake, LDH activity, and lactate content. The results were plotted on the basis of the mean value of the inhibition rate (%) of each group for the control group on a heat map

Table 3.

The effect of the metabolism or clearance of antitumor drugs in MHCC97-H cells

Groups		Sorafenib	Adriamycin
		t1/2 values (hours)
Control		18.78	13.25
Pralatrexate	0.5	19.62	14.10
	1	20.42	16.45
	2	23.71	17.97
Fatostatin	0.1	18.39	14.01
	1	18.59	13.83
BAY-876	0.1	18.24	13.22
	1	18.32	13.48
Rifampicin	1	13.67	10.73
	10	9.34	5.94
Ketoconazole	1	22.49	19.52
	10	30.53	27.18

Pralatrexate decreased the methylation of miRNA-34a’s promoter to decrease the activation of the Notch pathway

Next, the specificity of the effects of pralatrexate/DHFR was further confirmed. As shown in Fig. 3, transfection of miRNA-34a in MHCC97-H cells downregulated the expression of NICD, as well as survivin and N-cadherin; this did not affect the methylation of the promoter region of miRNA-34a or the expression level of DHFR (Fig. 3). Transfection of NICD in MHCC97-H cells did not affect the methylation of the promoter region of miRNA-34a, the expression level of miRNA-34a, or the expression level of DHFR (Fig. 3). Overexpression of NICD (SiDHFR + NICD) along with the knockdown of DHFR still reduced the promoter methylation of miRNA-34a and upregulated miRNA-34a expression but did not have an effect on N-cadherin and survivin expression (Fig. 3). Overexpression of DHFR together with miRNA-34a (DHFR + miRNA-34a) resulted in DNFR still being able to upregulate the promoter methylation of miRNA-34a, but miRNA-34a reversed the effect of DHFR on the Notch pathway (Fig. 3).

Next, the results in Fig. 7 show that after the direct expression of NICD in HCC cells, the cells were then treated with pralatrexate, which was also able to upregulate miRNA-34a expression or decrease the methylation rates of miRNA-34a’s promoter region. However, it was no longer able to affect the expression levels of NICD and Notch downstream factors. Similar results were obtained from the MTT assays (Table 4). Therefore, pralatrexate decreased the methylation of miRNA-34a’s promoter to decrease the activation of the Notch pathway.

Fig. 7.

The specificity of Pralatrexate on MHCC97-H cells. MHCC97-H cells were obtained in culture. The cells were treated with Pralatrexate or transfected with NICD, respectively. thereafter, cells were collected and total DNA samples were extracted separately for BSP-NGS assay; total RNA samples were extracted for AmplySeq assays were performed to detect miRNA-34a expression (A), miRNA-34a promoter region methylation (B), NICD expression (C), N-Cadherin expression (D), Survivin expression (E), and DHFR expression (F) in the cells. Results are shown as bar graphs, *p value < 0.05 compared with the control group

Table 4.

Overexpression of NICD almost blocked the effect of Pralatrexate on sorafenib in MHCC97-H cells

Groups	Pralatrexate	sorafenib
		IC50 values (μmol/L)
Control	–	1.15
	0.5	0.36
	1	0.10
	2	0.02
NICD	–	4.88
	0.5	4.85
	1	4.68
	2	4.55

Clinical expression level of DHFR in HCC specimens

The above results were obtained from cell-based experiments. To further confirm the roles of pralatrexate and the mechanisms related to folate metabolism in HCC, the expression levels of these factors in the clinical specimens were examined (Baseline information of the patients were shown as Table 1). As shown in Fig. 8, the expression of DHFR was much higher in HCC clinical specimens compared to paired non-tumor tissues (Fig. 8A). The expression level of miRNA-34a was significantly lower in HCC tissues than in paracancerous tissues, and the trend of methylation in the promoter region of miRNA-34a was reversed (Fig. 8B, C). Moreover, in HCC tissues, DHFR expression levels were negatively correlated with miRNA-34a (Fig. 8D–F), positively correlated with the methylation rates of the miRNA-34a promoter (Fig. 8D–F), positively correlated with NICD expression levels (Fig. 8D–F), and positively correlated with drug resistance-related factors downstream of Notch (survivin Fig. 8G or N-cadherin Fig. 8H). This further supports the role of DHFR in the Notch pathway.

Fig. 8.

The expression of DHFR and other factors in clinical specimens. Total DNA samples were extracted from 71 pairs of HCC or corresponding paraneoplastic-non-tumour tissues for BSP-NGS to detect methylation of the promoter region of miRNA-34a; total RNA samples were extracted for AmplySeq to detect the expression levels of the corresponding factors. A the expression level of DHFR; B the expression of miRNA-34a; C the methylation of miRNA-34a promoter region; D–H the co-relation between the expression level of DHFR with NICD (D), miRNA-34a (E), the methylation of miRNA-34a’s promoter region (F); Survivin (G); N-Cadherin (H). *p value < 0.05

Pralatrexate can increase liver cancer sensitivity to molecularly targeted agents in vivo

We first examined the effects of high, medium and low doses of the drug on the tumourigenic effects of MHCC97-H cells in nude mice. The results were firstly shown in the Supplemental Fig. 1, where 0.2 mg/kg (low dose), 0.5 mg/kg (medium dose) and 1 mg/kg (high dose) of sorafenib were administered orally by gavage. The results showed that sorafenib at 0.2 mg/kg (low dose) had weak subcutaneous tumourigenic inhibitory activity against MHCC97-H cells in nude mice, for this reason sorafenib at 0.2 mg/kg (low dose) was selected for further experiments.

On this basis, the results of oral gavage administration of Pralatrexate at 5 mg/kg (low dose), 10 mg/kg (medium dose) and 20 mg/kg (high dose) showed that pralatrexate only possessed significant anti-tumour activity at higher doses (e.g. 20 mg/kg) (Supplemnetal Fig. 2). For 5 mg/kg Pralatrexate, although it did not have significant cytotoxicity (5 mg/kg pralatrexate itself is less cytotoxic at this dose) (Supplemental Fig. 2A-C), it was able to inhibit the methylation of the promoter region of miRNA-34a, up-regulate the expression of miRNA-34a, and down-regulate the expression of factors related to the Notch pathway (Supplemental Fig. 2D and E). Therefore, 5 mg/kg of Pralatrexate was selected for further experiments.

Eventually, Sorafenib and Pralatrexate were used to administer alone or in combination, respectively, while replying to the expression of NICD. The results, as shown in Fig. 9, showed that the antitumour activity of either 5 mg/kg of Pralatrexate or 0.2 mg/kg of sorafenib alone was weak, but the combination of 5 mg/kg of Pralatrexate or 0.2 mg/kg of sorafenib was able to significantly increase the antitumour activity of sorafenib at this time (Fig. 9A–C). When NICD expression was restored in MHCC97-H cells, pralatrexate was still able to reduce the methylation of the promoter region of miRNA-34a and up-regulate the expression of miRNA-34a (Fig. 9D, E); however, at this time, Pralatrexate was no longer able to inhibit the expression of NICD and Survivin (Fig. 9F), and up-regulate the anti-tumour activity of sorafenib (Fig. 9). The results were shown as histograms as well as western blot.

Fig. 9.

Pralatrexate enhanced the sensitivity of MHCC97-H cells to sorafenib. MHCC97-H cells were obtained in culture, and the cells were transfected with the appropriate vectors and then inoculated subcutaneously in nude mice to form subcutaneous tumour tissue. Sorafenib at a dose of 0.2 mg/kg and Pralatrexate at a dose of 5 mg/kg were administered to nude mice by oral gavage, either alone or in combination. Results are shown as size and weight of tumour tissue (A, B) and photographs of tumour tissue (C). (D) Tumour tissues were subjected to AmplySeq assay to determine the expression of miRNA-34a; (E) Tumour tissues were subjected to BSP-NGS assay to determine the rate of methylation of the promoter region of miRNA-34a; and (F) Protein immunoblotting assay was carried out on three representative tumour tissues. In the WB experiment, SDS-PAGE was performed with the position of the leading edge of bromophenolan, and the pre-stained Marke (the main bands were 220, 110, 78, 53, 37, and 21 kDa) was used as a reference for the transmembrane experiments; after the completion of the transmembrane experiments, PVDF membranes near the corresponding protein band were cut down according to the position of the pre-stained marker. After the membrane transfer experiment was completed, the PVDF membrane near the corresponding protein bands was cut out according to the pre-stained marker position, and the antibody was incubated in the antibody incubation bag. Unmodified original photographs were obtained by scanning X-ray film or fluorescence imaging. *p value < 0.05

The influence of miRNA-34a might DHFR expression in MHCC97-H cells

Next, how miRNA-34a might influence DHFR expression was examined. As shown in Fig. 10, transfection of miRNA-34a in HCC cells MHCC97-H was not able to affect the expression of DHFR.

Fig. 10.

miRNA-34a could not affect the expression of DHFR in MHCC97-H cells. MHCC97-H cells were obtained in culture and cells were assayed for protein immunoblotting after cell transfection with the appropriate vectors, including control miRNA or hsa-pre-miRNA-34a. DHFR or GAPDH were detected, respectively. In the WB experiment, SDS-PAGE was performed with the position of the leading edge of bromophenolan, and the pre-stained Marke (the main bands were 220, 110, 78, 53, 37, and 21 kDa) was used as a reference for the transmembrane experiments; after the completion of the transmembrane experiments, PVDF membranes near the corresponding protein band were cut down according to the position of the pre-stained marker. After the membrane transfer experiment was completed, the PVDF membrane near the corresponding protein bands was cut out according to the pre-stained marker position, and the antibody was incubated in the antibody incubation bag. Unmodified original photographs were obtained by scanning X-ray film or fluorescence imaging

The sub-cellular fraction assays

The aforementioned results detected the protein expression of NICD, and to further clarify the specificity of the action of Pralatrexate, further sub-cellular fraction assays were performed. As shown in Fig. 11, Notch 1 could be detected only in the plasma of MHCC97-H cells, whereas NICD could be detected only in the nucleus. Treatment of MHCC97-H cells with Pralatrexate for very short periods of time (30 min, 1 h, and 2 h) did not affect the protein expression of Notch 1 and did not affect the distribution of NICD in the nucleus. Therefore, Pralatrexate could not affect the cleavage of Notch in MHCC97-H cells.

Fig. 11.

Pralatrexate does not affect Notch 1 protein cleavage in MHCC97-H cells. MHCC97-H cells were obtained in culture, and after treatment of MHCC97-H cells with a dose of 2 µmol per litre of Pralatrexate (30 min, 1 h, and 2 h time points), nucleoplasmic isolation/sub-cellular fraction assays were performed to detect Notch 1 protein and NICD in the cytoplasm as well as in the nucleus, respectively. The results are shown as images of the bands from the Western blot. In the WB experiment, SDS-PAGE was performed with the position of the leading edge of bromophenolan, and the pre-stained Marke (the main bands were 220, 110, 78, 53, 37, and 21 kDa) was used as a reference for the transmembrane experiments; after the completion of the transmembrane experiments, PVDF membranes near the corresponding protein band were cut down according to the position of the pre-stained marker. After the membrane transfer experiment was completed, the PVDF membrane near the corresponding protein bands was cut out according to the pre-stained marker position, and the antibody was incubated in the antibody incubation bag. Unmodified original photographs were obtained by scanning X-ray film or fluorescence imaging

Discussion

Research into the molecular mechanisms of resistance of HCC cells to molecularly targeted drugs can help prolong patient survival and improve patients’ quality of life [4]. This study is the first to report that interfering with folate metabolism can upregulate the sensitivity of HCC cells to a variety of molecularly targeted drugs. In mammalian cells, folic acid taken up through food is reduced to dihydrofolate (DHF), which is then reduced by DHFR to tetrahydrofolate (THF), participating in the transfer of a carbon unit. In other words, the cell’s THF is coupled to the methionine cycle by the action of 5-methyltetrahydrofolate-homocysteine methyltransferase (MTR) to produce the key substrate for the regulation of methylation, S-adenosylmethionine (SAM), which is the direct source of the methyl group (1C unit) in methylation regulation [17, 18]. Pralatrexat, an inhibitor of DHFR, induces the expression of miRNA-34a by inhibiting the methylation of the promoter region of miRNA-34a, ultimately inhibiting the activity of the Notch pathway. In this study, pralatrexate was able to upregulate the sensitivity of HCC cells to molecularly targeted drugs. We used a variety of techniques and experimental designs to validate the specificity of pralatrexate’s actions on miRNA-34a/Notch. These results extend our knowledge of folic acid metabolism with HCC treatment.

Dividing cells require activated 1C units to synthesize and package nucleic acids [17, 18]. DNA methylation is an important epigenetic regulatory mechanism that induces the silencing of gene expression by methylation of the CpG site in the promoter region of specific genes [32]. DNA methylation has been studied only for various isoforms of DNA methyltransferases, but the mechanism of folate metabolism is also closely related to DNA methylation [33]. In HCC cells, the mechanism of action of Notch pathway-induced drug resistance is that, as a cellular stress response factor, Notch protein activation promotes cellular resistance to exogenous injury by inducing the expression of downstream pro-survival and anti-apoptotic proteins, thereby inducing drug resistance in HCC cells [34, 35]. This includes radiation therapy, radiofrequency ablation therapy, and anti-tumor drug therapy, all of which are capable of acting as stressors to induce Notch pathway activity or stressor response factors, such as EMT, in HCC cells [34, 35] The Notch pathway and its downstream EMT-related mechanisms have also been clearly observed in clinical studies as key regulatory mechanisms for patients’ resistance and prognosis to various antitumor treatments [36]. MiRNA-34a acts on Notch proteins and is able to reverse the resistance of malignant cells, such as HCC, to various antitumor treatment strategies. The deletion of miRNA-34a expression in HCC cells is an important mechanism in the resistance of HCC cells to molecularly targeted drugs. MiRNA-34a promoter region methylation can cause downregulation or deletion of miRNA-34a expression. The results of this study link folate metabolism to the miRNA-34a/Notch pathway, a finding that has important implications. It is worth mentioning that there are four isoforms of Notch (Notch1 to Notch4), with differences in their extracellular segments and a highly conserved intracellular segment (NICD); therefore, transfection of NICD in cells in this study also avoided the surrogate interaction between Notch proteins to ensure the specificity of the results.

PXRs are another regulator of HCC metabolism-related drug resistance mechanisms [37]. As a nuclear receptor, PXR can be activated by a variety of endogenous or exogenous ligands in HCC cells, which in turn induces the expression of downstream genes cyp3a4 or mdr-1 [38, 39]. These downstream genes are Phase I, Phase II, or Phase III metabolizing enzymes that protect the body during substance metabolism, organism detoxification, or exogenous substance clearance [38, 40]. In HCC cells, antitumor drugs as exogenous toxicants are damaging factors, and these genes can accelerate the rate of metabolism and clearance of antitumor drugs in HCC cells, ultimately inducing HCC cells to be resistant to antitumor drugs and exert a protective effect on HCC cells [26]. In HCC cells, the molecularly targeted drug sorafenib can function as a ligand to induce the transcription factor activity of PXR, which in turn induces the expression of the PXR downstream resistance genes mdr-1 or cyp3a4, ultimately accelerating its own metabolism and clearance mechanisms in HCC cells and inducing drug resistance [26]. In this study, the inhibitors of DHFR did not significantly affect the expression of PXR-associated resistance genes and did not significantly affect the rate of sorafenib metabolism and clearance in HCC cells compared with rifampicin or ketoconazole. Thus, it appears that pralatrexate may act indirectly on PXR compared to rifampicin or ketoconazole, which act directly on PXR. The mechanism of action of pralatrexate in regulating drug resistance genes downstream of PXR needs to be further explored, and we speculate that the Notch pathway may cross-talk with PXR.

The liver is one of the most important digestive glands and the most important metabolic organ in the body, and HCC cells derived from normal liver cells have unique metabolic properties. Such metabolic abnormalities are closely associated with the progressive metastatic invasion and tissue microenvironment of HCC [40]. In HCC tissues under hypoxia, low partial pressure of oxygen, or changes in osmotic pressure conditions, the cells often uptake a large amount of glucose through GLUT1 [41, 42]. HCC cells have an exaggerated glucose uptake capability due to the intensive anaerobic glycolysis (Warburg effect) required to maintain survival in the harsh tumor microenvironment in HCC tissues [41, 42]. The resistance of HCC cells to antitumor drugs is closely correlated with cellular metabolic features through the overwhelming anaerobic glycolytic activity, and the enhanced EMT takes place afterward [43, 44]. Thus, glucose metabolism is a potential target for developing antagonistic therapy for chemoresistance. The Warburg effect of HCC cells is closely related to glycolipid metabolism. Fatostatin was used in this study as an inhibitor of lipid metabolism and BAY-876 as an inhibitor of glycolipid metabolism [45]. The inhibitors of folate metabolism selected for this study had significantly different effects compared to fatostatin or BAY-876. Pralatrexate did not affect the expression of factors related to glycolipid metabolism, nor did it affect the glycolipid metabolism of HCC cells; its effect occurred via the Notch pathway.

Importantly, in this study, miRNA-34a degraded Notch mRNA through post-transcriptional gene expression silencing, which in turn down-regulated Notch pathway activity. However, the Notch pathway function is very important and the regulatory mechanism is complex and multifaceted. Proteolytic processing: The cleavage of Notch1 to produce NICD depends on the activity of proteases like ADAM10/17 and γ-secretase [46, 47]. Ligand presence: The presence and binding of Notch ligands (Delta-like or Jagged) are necessary for the activation of the Notch1 receptor [48, 49]. NICD protein stability: NICD undergoes rapid degradation through the ubiquitin–proteasome pathway [50, 51]. For this reason this study also tested the specificity of Pralatrexate's action as much as possible. MHCC97-H cells were treated with Pralatrexate for short periods of time (30 min, 1 h, and 2 h) and then subjected to sub-cellular fractionation experiments. Changes in the content of Notch 1 protein in the cytoplasm and NICD in the nucleus could reflect the shearing of Notch protein. Our results showed that Pralatrexate did not affect the content of Notch 1 protein in the cytoplasm and NICD in the nucleus, so Pralatrexate should not affect Notch shearing. At the same time Pralatrexate does not have any structural similarity to a drug like the ubiquitination inhibitor MG132, so it is also highly unlikely that Pralatrexate is involved in Notch-related ubiquitin regulation.

Conclusion

In this study, we investigated and reported the molecular mechanism by which DHFR regulates HCC resistance to molecularly targeted drugs: DHFR is able to induce methylation of the promoter region of miRNA-34a, up-regulate the activity of the Notch pathway, and ultimately induce drug resistance in HCC cells. This study establishes a link between folate metabolism and HCC drug resistance, which not only expands our understanding of folate metabolism, especially metabolism-associated cancerous modulation mechanisms, but also helps to provide more choices for HCC treatment.

Author contributions

Y.J., H.J.J. and Y.M. wrote the main manuscript text. Y.J. Q.M.L, and B.S.S. prepared Figs. 1, 2, 3, 4, 5. X.KL. and J.H.W. prepared Figs. 6, 7, 8. Z.Y.L., Y.M. and H.J.J. prepared Figs. 9, 10, 11. All authors reviewed the manuscript.

Funding

There was no funding.

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

The materials directly related to human body are the clinical specimens of HCC or the paired non-tumor tissues. Written informed consent and authorisation was obtained from the patient for the collection, containing and usage of the clinical specomens. The use of these clinical specimens was approved by the Ethics Committee of the 920th Hospital of the PLA Joint Logistic Support Force, Kunming City 650032, Yunnan Province, P.R. China. The usages, and the experiments related these clinical specimens were performed in the 920th Hospital of the PLA Joint Logistic Support Force, Kunming City 650032, Yunnan Province, P.R. China in accordance with the Declaration of Helsinki (World Health Organization). For the animal ethic declaration, the nude mouse model was used as experimental animals in the presence work. The animal experiments were performed in the First medical center of Chinese PLA General Hospital, Beijing 100853, P.R. China. The experimental design, methods and the protocol of animal experiments were reviewed and approved by the Institutional Animal Care and Usage Committee (IACUC) of the First Medical Center of General Hospital of Chinese PLA. All the animal experiments were performed in accordance with the U.K. Animals Act, 1986 (Scientific Procedures) guidelines.

Competing interests

The authors declare no competing interests.