Antimalarial Drug Pyrimethamine Plays a Dual Role in Antitumor Proliferation and Metastasis through Targeting DHFR and TP

Pyrimethamine (2,4-diamino-5-p-chlorophenyl-6-ethyl-pyrimidine, Pyr) has been clinically used as antimalarial drugs (1). Pyr exerts its antimalarial effect by targeting plasmodium dihydrofolate reductase (pDHFR; ref. 2). DHFR is an essential enzyme in the synthesis of folic acid, which is a cofactor required for DNA synthesis. In addition to its antimalarial effects, Pyr exhibits the activity of inducing apoptosis of tumor cells through cathepsin B–dependent and caspase–dependent apoptotic pathways (3, 4). Pyr can also inhibit the STAT3 pathway in breast cancer cells (5). Pyr also has a broad range of effects in non–small cell lung cancers (6). However, the target of Pyr has not been elucidated before.

Human DHFR (hDHFR) is a core enzyme in folate metabolism. It plays a key role in the biosynthesis of nucleic acids and is closely associated with thymidylate synthase in purine and pyrimidine production (7–9). Given these characteristics, hDHFR is a crucial target in anticancer drug development. In fact, DHFR inhibitors, such as methotrexate (MTX), have been applied in cancer treatment (10). The effect of Pyr on hDHFR has not been previously reported.

MTX is extensively used in chemotherapy for several cancer types, including lung cancer, leukemia, lymphoma, breast cancer, and head and neck cancers (11–13). Previous studies showed that MTX treatment may also result in undesirable side-effects. For example, MTX might induce lethal interstitial lung diseases, including pulmonary fibrosis in some cases (14). High doses of MTX can also inflict structural and functional injury to the gastrointestinal tract (15), cause inflammatory response, and alter absorptive capacity (16–18). Some in vitro studies have shown that MTX may induce EMT of epithelial cells. MTX can promote the migration and invasion of RLE/Abca3 cells and increase the expression of TGF-β (19). MTX can also inflict damage on alveolar epithelial cells and promote the epithelial–mesenchymal transition (EMT) of epithelial cells (20, 21). During EMT, cells lose their typical epithelial characteristics and acquire mesenchymal traits (22). Cancer cells undergoing EMT lose their cell–cell connection, cell–matrix contact, and normal epithelial polarity while gaining mesenchymal characteristics. These modifications may enhance the migratory and invasive ability of cancer cells.

Given that hDHFR is a target of antitumor drug development and pDHFR is a target of Pyr in plasmodium, we first investigated whether Pyr demonstrates antitumor activity by inhibiting hDHFR in tumor cells. We found that Pyr not only inhibits the proliferation of cancer cells but also suppresses the migration of lung cancer cells, whereas MTX could only inhibit the proliferation of cancer cells. These results suggested that DHFR is not the only target of Pyr. We found that Pyr might play a dual role in antitumor proliferation and migration by synergistic targeting DHFR and thymidine phosphorylase (TP). TP is a nucleoside-metabolizing enzyme that has a crucial association with tumor migration and invasion (23).

ClustalX was used to blast the primary structure of Human dihydrofolate reductase (hDHFR) and Plasmodium falciparum Dihydrofolate Reductase (pDHFR). The secondary structure elements alignment of pDHFR and hDHFR was generated by the Esprint 3.0 server. Three-dimensional structures were aligned by using PYMOL. The crystal structures of the hDHFR–MTX complex (PDB code 1u72) and pDHFR–Pyr complex (PDB code 1j3j) were downloaded from the Protein Data Bank. Molecular docking was performed using Schrodinger software. MTX in hDHFR–MTX complex was extracted from crystal structures, and the pocket was used as the central docking location.

Energy minimizations and MD simulations were performed with the Pmemd module of the Amber 14 package. To simulate the normal physiological reaction temperature, the entire MD system was gradually heated to 310 K. Periodic boundary conditions were used in the NPT ensemble and the SHAKE algorithm was applied to constrain all covalent bonds that involved hydrogen atoms. The cutoff values for nonbonded interactions were set at 10 Å. Finally, the RMSD of the initial structure from the simulated positions was used to evaluate the stability of the entire simulation.

The binding-free energies (ΔGbind) of the ligands with proteins were calculated through the MM–PBSA procedure in AMBER14. The binding-free energy for each molecular species (complex, protein, and ligand) was computed by using the equation ΔG_bind = G_complex − (G_protein + G_ligand).

The cancer cell lines NCI-H460, NCI-H446, A549, HepG2, MHCC97L, LLC, MCF-7, ASPC-1, PCNA-1, SGC-7901, HT-29, SW480 and PC-3 were obtained from KeyGen Biotech (Nanjing, China) in 2013 and authenticated by STR genotyping. Mycoplasma was analyzed using the Mycoplasma qPCR Detection Kit (Sigma) before experiment. Cells were grown in medium supplemented with 10% FBS and maintained at 37°C in a humidified atmosphere containing 5% CO₂.

The effects of Pyr and MTX on cell viability were determined through the MTT (3-(4,5)-dimethylthiahiazo (-z-y1)-3,5-di-phenytetrazoliumromide) assay. Pyr and MTX were purchased from Meilun Biotechnology Co., LTD., and the chemical structures of them were showed in Supplementary Materials (Supplementary Fig. S1). A total of 5 × 10³ cells were seeded in 96-well culture plates. Then, the cells were treated with various concentrations MTX and Pyr (0–200 μmol/L). After 24, 48, and 72 hours of incubation, the cells were stained with MTT. Then, the culture medium was removed, and cells were lysed using DMSO. Finally, the optical density (OD) values of the solution were determined at 570 nm by using a microplate reader (Multiskan FC, Thermo Scientific). Data were analyzed using GraphPad and a log plot of cell viability (%) against the concentrations of drugs was constructed. The IC₅₀ was also calculated from the plot.

Cell proliferation assays were performed with the real time cell analyzer (RTCA). Background impedance was measured with 50 μL of culture medium. NCI-H460 and A549 cells were seeded into plates (E-plate 16, ACEA Biosciences) with 100 μL of medium per well. Subsequently, the plates were monitored on the xCELLigence RTCA Dual Plate instrument (ACEA Biosciences) at 37°C in a humidified atmosphere with 5% CO₂. Pyr (15 μmol/L) and MTX (30 μmol/L) were added to the plate after the cells entered the logarithmic growth period. The experiments were repeated three times.

Live/dead fixable dead cell stain kits (Invitrogen) were used to evaluate the effect of Pyr on cells in accordance with the manufacturer's instructions. Cell viability was analyzed through flow cytometry (Millipore guava easyCyte).

An Annexin V-FITC/PI apoptosis detection kit (Nanjing Kaiji Biotechnology Development Co., Ltd.) was also used to evaluate the effect of Pyr on cell apoptosis in accordance with the manufacturer's protocol. The cells (1 × 10⁶) were evenly spread in a 6-well plate and the drug was added after adherence. After 24 hours, the cells were collected, Annexin V-FITC and PI were sequentially added according to the instructions. Experiments were repeated three times.

For the wound-healing assay, NCI-H460 and A549 cells were grown on 24-well plates to 100% confluence. A 100 μm wound was scratched using sterile pipette tips, and the exfoliated cells were washed off three times with PBS, and then Pyr (7.5 or 15 μmol/L) or MTX (15 μmol/L) was added to cells cultured in serum-free medium. Cell migration ability was assessed by measuring the movement of cells in the scratches in the wells. The wound closure rate after 24 and 48 hours was measured and normalized to length at 0 hours. After 48 hours, images of the wounds were acquired under light microscopy (Nikon, Japan). The relative length values of the individual wounds were counted according to the normalized length of 0 hours.

In this assay, 24-transwell plates were used. A total of 5 × 10⁴ cells were placed on the top chamber inserts, which were coated with Matrigel (BD Biosciences). After incubation with Pyr (7.5 or 15 μmol/L) or MTX (15 μmol/L) for 24 hours, the cells were stained with 0.1% crystal violet. Invading cells were visualized and counted in six randomly selected fields under an inverted microscope (×100).

After treatment with different drugs, proteins were extracted from NCI-H460 cells and analyzed through western blot analysis. After the culture medium was aspirated, each dish was washed with PBS, and protein lysis buffer was added (containing protease and phosphatase inhibitors) to extract the proteins. The proteins were separated by 10% polyacrylamide gel electrophoresis, transferred to a polyvinylidene fluoride (PVDF) membrane (the PVDF membrane was activated by methanol), and blocked with 5% skim milk. Proteins were incubated with primary antibodies against β-actin (Affinity, 1:5,000), E-cadherin (Affinity, 1:1,000), Vimentin (Affinity, 1:1,000 dilution), Ki-67 (Affinity, 1:1,000), MEK2 (Affinity, 1:1,000), ERK2 (Affinity, 1:1,000), and GAPDH (Affinity, 1:5,000). The samples were incubated with primary antibody overnight in a rotator at 4°C. Blots were further incubated with horseradish peroxidase–labeled secondary antibodies (Affinity, 1:5,000). Finally, target proteins were visualized using ECL substrate reagents (Millipore).

NCI-H460 cells seeded on a cell-climbing slice were incubated for 24 hours with Pyr (7.5 or 15 μmol/L) or MTX (15 μmol/L) in 24-well culture plates. The cells were fixed in 3.7% paraformaldehyde for 15 minutes and then treated with 0.1% Triton X-100 for 10 minutes, after which the cells were incubated with 3% BSA for 30 minutes. The cells were then incubated overnight with primary antibodies at 4°C. Cells were washed four times with PBS and incubated for 30 minutes with secondary antibodies. Finally, the cells were covered with DAPI for 15 minutes. Proteins were visualized through confocal microscopy (Nikon, Japan).

C57BL/6J mice (male, 5–6-weeks-old) were maintained in a specific pathogen-free animal care facility. The mice were allowed to acclimate for 7 days before the experiment. All animal studies were carried out in accordance with National Institutes of Health Animal Use Guidelines and the current Chinese Regulations and Standards for the Use of Laboratory Animals. All animal procedures were approved on the basis of guidelines of the Animal Ethics Committee of the Tianjin International Joint Academy of Biotechnology and Medicine. Lung cancer xenografts were established by subcutaneously injecting 1 × 10⁷ cells (suspended in saline) into the flanks of the mice. After the tumors volume reached approximate 100 mm³, the mice were randomly divided into four groups (n = 5). Pyr (7.5 or 15 mg/kg), MTX (7.5 mg/kg) or saline were orally administered to the mice once a day. Tumor volume and body weight were measured daily after tumor inoculation. Tumor volumes were calculated in accordance with the formula V = ab²/2 (a = length and b = width). After 2 weeks of treatment, all mice were euthanized. The xenografts and lungs were resected and measured. Metastases in lung tissues were observed by using a stereoscopic microscope and detected through hematoxylin/eosin staining.

Tumor and lung tissues were fixed in 10% formaldehyde, dehydrated, and embedded in paraffin wax. Then, 4-μm sections of the tissues were stained with hematoxylin and eosin. Digital images were acquired under microscopy (Nikon, Japan).

Tissues were deparaffinized and rehydrated through incubation with xylene and decreasing concentrations of ethanol. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide. The microwave antigen retrieval technique was used for antigen retrieval. Samples were incubated overnight with primary antibodies at 4°C after blocking with rabbit polyclonal anti–E-cadherin, rabbit polyclonal anti-vimentin, rabbit polyclonal anti-MMP2, and rabbit polyclonal anti-MMP9. All antibodies were obtained from Affinity and diluted at the rate of 1:50. Brown-stained cytoplasm, nuclei, or membranes in cells were considered positive. Staining intensity was scored as follows: none (0), weak brown (1+), moderate brown (2+), and strong brown (3+). The percentage of stained cells was divided into five classes: 0 for negative cells, 1 for 1%–25%, 2 for 25%–50%, 3 for 50%–75%, and 4 for >75%.

Biacore 3000 instrument (GE Healthcare) was used in the experiment. TP was immobilized on CM5 sensor chips in accordance with the instructions provided with the Biacore Amini Coupling Kit. Pyr was diluted in running buffer at different concentrations and injected into TP-immobilized CM5 sensor chips. The concentrations of Pyr were 0, 0.25, 0.5, 1, 2, 4, and 8 μmol/L. The surface of the control chip was prepared in the same manner for data correction. BIA evaluation software was adopted for data analysis.

Thermal shift assay (TSA) was performed using SYPRO Orange (Life Sciences) as the shift reporter dye. Briefly, 11.4 μg of protein was incubated with Pyr at a ratio of 1:10 or 1:20 for 20 minutes, dye was added, and the reactions were monitored in real time (Bio-Rad MiniOpticon; excitation, 490; emission, 575 nm) from 29°C to 95°C with a rate of change of 1°C/min. The melt curve is represented as normalized data and calculated as d(fluorescence)/d (temperature).

TP activity was reflected by intercellular thymine concentration, which was detected through LC–MS–MS (24). NCI-H460 cells were incubated for 24 hours with Pyr (30 μmol/L), 5UIR (30 μmol/L), and MTX (30 μmol/L). Next, 1 × 10⁷ cells were lysed with ice-cold 80% methanol. After centrifugation, 0.05 μg of U-¹³C₁₀ and U-¹⁵N₂ thymidine (Sigma) were added to the supernatant as the internal control. The polar metabolites in the supernatant were separated, dried, and reconstituted with the LC mobile phase. Intercellular thymidine level was measured through LC–MS–MS in the positive-ion mode and expressed as ng/1 × 10⁷ cells. All experiments were repeated independently at least twice.

The methods of wound-healing assay, invasion assays and western blot were same as mentioned above. A549 cells were used in the experiments. Cells were divided into four groups: control (treated with solvent), TP (treated with TP 10 ng/mL), Pyr (treated with Pyr 15 μmol/L) and Pyr+TP (Pyr 15 μmol/L combined with TP 10 ng/mL).

Proteomics analysis was used to identify the differentially expressed proteins of A549 cells treated versus non-treated with Pyr (7.5 μmol/L), which were significantly regulated (|logFC|>1.5) in the samples treated with Pyr. To initially explore which functions and pathways have changed in the Pyr groups, we used metascape website (http://metascape.org/) to perform GO and KEGG enrichment analysis. Protein-protein interaction (PPI) network was analyzed using STRING website (www.string-db.org/) and Cytoscape software. To get more reliable data, we only chose the interactions of the combination score >0.9. To further study which proteins play a greater role in the PPI network, CentiScape 2.2 plug-in module of Cytoscape was performed to calculate the degree of connectivity in the PPI network. To better understand the biological significance of the PPI network, MCODE plug-in module was used to select most significant (MCODE score >10) sub-modules.

Commercially available Human DHFR ELISA Kit (Wuhan Elabscience Biotechnology Co., Ltd.) was used for assaying of the activity of the DHFR in lung cancer cells. NCI-H460 cells were used to test the activity of DHFR. The cells (1 × 10⁶) were evenly spread in a 6-well plate and the drug was added after adherence. After 24 hours, the cell supernatant was collected, centrifuged at 1,000 × g for 20 minutes to remove impurities and cell debris, and the supernatant was collected. Then the activity of DHFR was tested according to the product manual.

The data of DHFR expression levels were obtained from “Human Protein Atlas” database (https://www.proteinatlas.org/ENSG00000228716-DHFR/cell). Survival data of patients with lung cancer were downloaded from TCGA. There are 982 patients with lung cancer and 12 normal ones. Overall survival and disease-free survival was analyzed according to the expression of TP or DHFR in 982 cases of lung cancer patient (using the Kaplan–Meier method and evaluated using the log-rank test). According to the data, the FPKM (fragments per kilobase of exon per million fragments mapped) value of more than 2.8 is classified as "high," and the FPKM value of less than 2.8 is classified as “low.” GraphPad was used for mapping.

Pyr is a known inhibitor of pDHFR. However, its effect on hDHFR has not been verified. We used DHFR assay kits to test the effect of Pyr on hDHFR in lung cancer cells. Our results showed that both MTX and Pyr could inhibit the activity of DHFR in lung cancer cells (Fig. 1A). MTX, a clinically used chemotherapy drug targeting DHFR, exerts an inhibitory effect on different tumor cells. So we detected the inhibitory effect of Pyr on different tumors cells. Fig 1B and C showed the inhibitory effects of MTX and Pyr on various types of tumor cells in vitro. Our findings showed that Pyr showed inhibitory activity to a variety of cell lines, such as MCF-7, NCI-H460, NCI-H446 and so on. The expression levels of DHFR in several cell lines from the Human Protein Atlas Database are displayed in Fig. 1D. The results of correlation analysis showed that the IC₅₀ values of MTX and Pyr for cancer cells were positively correlated with the expression level of DHFR (Fig. 1E and F).

We aligned orthologous pDHFR and hDHFR sequences (downloaded from UniProt) to identify the residues and secondary structures that were shared by these enzymes. pDHFR and hDHFR shared low amino acid sequence identity (Fig. 2A). Aligning the secondary and three-dimensional structures of hDHFR and pDHFR revealed that the structures of the two enzymes are mainly differentiated by an α-helix (circled in red), which is consistent with previous report (25). The root–mean–square deviation (RMSD) obtained by aligning the three-dimensional structures of the proteins was 0.679 (Fig. 2B). On the basis of the alignment results, we docked Pyr into the active sites of hDHFR and pDHFR. The conformations and orientations exhibited by Pyr in the active centers of hDHFR and pDHFR are almost identical. The docking score of Pyr and hDHFR is −7.483, and the docking score of Pyr and pDHFR is −7.140 (Fig. 2B). Therefore, Pyr may have similar binding capacities for hDHFR and pDHFR. We ran 50 ns molecular dynamics (MD) simulations for the complexation of hDHFR with Pyr, pDHFR with Pyr, and hDHFR with MTX. We analyzed the RMSD values provided by the simulations to illustrate the dynamic stability of the three complexes and to ensure the rationality of the following analysis (Fig. 2C). The RMSD of each system tended to converge. This tendency indicated that the systems are stable and in equilibrium. To further compare the binding of Pyr to hDHFR and pDHFR, we calculated the binding-free energies of all three systems by using the MM-PBSA program in AMBER. As shown in Fig. 2D, the binding capacities of Pyr for hDHFR and pDHFR are almost the same. The binding capacity of Pyr–hDHFR complex was weaker than MTX–hDHFR complex. The hydrogen bond interactions that occur between the drug and protein play an important role in the binding of the inhibitor to kinase in the three protein-inhibitor systems. Our results showed that Pyr and MTX formed stable hydrogen bonds with Ile 7, Tyr 121, and Val 115 in hDHFR. The above results suggested that similar to MTX, Pyr can act as a hDHFR inhibitor.

We used the MTT assay to detect the effect of Pyr on cell viability at 24, 48, and 72 hours. As shown in Fig. 3A, the IC₅₀ values of Pyr for NCI-H460 cells at 24, 48, and 72 hours treatment are 98.17, 64.31, and 37.60 μmol/L, respectively. As shown in Fig. 3B, the IC₅₀ values of Pyr for A549 cells at 24, 48, and 72 hours treatment are 83.37, 40.57, and 28.07 μmol/L, respectively. Next, we used a real-time cell analyzer (RTCA) to demonstrate the effects of Pyr on the proliferation of NCI-H460 and A549 cells. We found that Pyr inhibited the proliferation of NCI-H460 and A549 cells in a dose-dependent manner (Fig. 3C and E). We also used the live/dead assay to explore the effect of Pyr (15 and 30 μmol/L) on NCI-H460 and A549 cells. The percentage of cell death increased in a dose-dependent manner (Fig. 3D and F). We used an Annexin V-FITC/propidium iodide (PI) kit to examine the effect of Pyr on cell apoptosis and found that Pyr can effectively induce apoptosis in the NCI-H460 and A549 cells in a dose-dependent manner (Fig. 3G and H). The western blot analysis results also showed that Pyr treatment decreased the expression level of Ki67, which is a marker of cell proliferation (Fig. 3I).

We performed the wound-healing assay to investigate the ability of Pyr to inhibit the migration of NCI-H460 (Fig. 4A) and A549 (Fig. 4B) cells. The migration ability of cells increased after 48 hours of treatment with MTX. By contrast, wounds were widened under Pyr treatment. This behavior indicated that Pyr treatment inhibited the migration of cancer cells. We also detected the effect of Pyr on cell invasiveness (Fig. 4C). The number of cells that invaded through the Matrigel-coated filter decreased under Pyr treatment relative to the control. We used western blot analysis and immunofluorescent staining to detect the effect of Pyr and MTX on the expression of EMT markers E-cadherin and vimentin. We found that Pyr decreased the expression of vimentin and increased the expression of E-cadherin in NCI-H460 cells (Fig. 4D and E), whereas MTX exerted the opposite effect. The same results were observed in A549 cell lines (Supplementary Figs. S2 and S3).

We examined the effect of Pyr on Lewis lung cancer (LLC) xenografts in C57BL/6J mice. The body weights of mice in the MTX treatment group decreased relative to those of mice in the model group. No significant change was noted between the model group and Pyr treatment group (Fig. 5A). Tumor growth was suppressed in the MTX and Pyr treatment groups relative to that in the model group (Fig. 5B and C). As shown in Fig. 5D–F, the number of metastatic tumor nodes on lung surfaces drastically decreased after Pyr treatment. Meanwhile, the extent of lung metastasis increased after MTX treatment. Immunohistochemical results of tumor tissues revealed that E-cadherin expression levels were higher in the Pyr treatment group than those in the control groups and the expression levels of vimentin, MMP2, and MMP9 decreased in the Pyr treatment groups (Fig. 5G). Western-blot analysis showed the same results (Supplementary Fig. S4).

Pyr could not only inhibit the proliferation of lung cancer cells like MTX, but also inhibit EMT, migration, and invasion of lung cancer cells. Thus, we speculated that Pyr may have another EMT-associated target in tumor cells. Pyr is a pyrimidine analog. TP plays an important role in tumor migration, and invasion. The chemical structures of Pyr and thymidine were similar. So we hypothesized that TP may be another target of Pyr. We performed molecular docking simulations to compare the binding scores of the TP inhibitors (TPI, 5-Iodouracil, and 5-fluorouracil), Pyr, and MTX. We found that Pyr and TP inhibitors have similar docking scores and that MTX cannot enter the active center of TP (Fig. 6A; Supplementary Fig. S5). We used LC–MS–MS to detect thymine concentration in cells treated with 5UIR and Pyr, which can reflect TP activity. The results showed that Pyr inhibited the activity of TP (Fig. 6B). Moreover, we verified the interaction between Pyr and TP through the Biacore assay (K_d = 6.19 ± 0.78 μmol/L; Fig. 6C). The TSA assay also showed that Pyr could bind with TP (Supplementary Fig. S6).

To further explore the inhibitory effect of Pyr on TP, we conducted a wound-healing assay. Our results showed that TP induces the migration and invasion of lung cancer cells. Pyr inhibited the TP-induced migration and invasion of cancer cells (Fig. 6D and E). Compared with the control treatment, TP promoted the expression of ERK2 and MEK2, whereas Pyr suppressed the expression of ERK2 and MEK2 (Fig. 6F). TP increased the expression of vimentin and decreased the expression of E-cadherin, and Pyr reversed the changes of EMT markers, which showed that Pyr inhibited the EMT induced by TP (Fig. 6F).

Proteomics analysis was used to identify the differentially expressed proteins of A549 cells treated or non-treated with Pyr. Gene Ontology (GO) analysis results showed that the differential proteins were enriched in the functions of pyridine nucleotide metabolic process, apoptotic signaling pathway, regulation of cell-cycle G₂–M phase transition, cell cycle phase transition, and extracellular matrix organization. The KEGG pathway analysis revealed that the differential proteins were mainly involved in several pathways, including programmed cell death, focal adhesion, extracellular matrix organization, collagen formation, collagen degradation, and cell-cycle pathway. Protein–protein interaction (PPI) network was shown in Fig. 7B. The hub proteins were marked in red, such as HSP90AA1, CKAP5, NEDD4L, and CPSF1. The sub-networks (MCODE score > 10) from PPI network were shown in Fig. 7C–E. The functions of the sub-networks mainly associated with cell-cycle phase transition, protein ubiquitination, collagen degradation, and nucleic acid binding. These results indicated that Pyr affected not only the proliferation-related pathways (e.g., pyridine nucleotide metabolic process, apoptotic signaling pathway, and cell-cycle pathway) of tumor cells, but also migration and invasion-related pathways (eg. Focal adhesion, Extracellular matrix organization, Collagen formation, and Collagen degradation) of tumor cells, which is consistent with the functions of Pyr observed in vivo and in vitro.

We analyzed the expression of DHFR and TP in tissues from 982 patients with lung cancer included in the TCGA database. We found that the expression levels of DHFR and TP in lung cancer tissues were significantly higher than that in normal human lung tissues (Fig. 8A and B). The results from the ULCAN database revealed that the mRNA levels of the two proteins in lung cancer tissues were elevated relative to those in normal lung tissues (Fig. 8C). We also analyzed the samples in the TCGA database on the basis of pathology grade and clinical stage. The mRNA expression of DHFR was positively correlated with clinical phase I, clinical phase II, and clinical phase III. And the mRNA expression of TP was positively correlated with clinical phase II and clinical phase III (Fig. 8D). We analyzed the effect of DHFR and TP expression on survival status. DHFR and TP overexpression were associated with poor prognosis (Fig. 8E–G). To further investigate the relationship between DHFR/TP expression and EMT, we analyzed the correlation between the mRNA expression levels of DHFR/TP and EMT markers E-cadherin (gene name: CDH1) and vimentin (gene name: VIM). We found that TP was positively correlated with vimentin expression and negatively correlated with E-cadherin expression. However, the expression levels of DHFR and EMT markers were not correlated (Fig. 8H).

Pyrimethamine (Pyr) is a pyrimidine derivative, which interferes with the regeneration of tetrahydrofolic acid from dihydrofolate by targeting DHFR of the plasmodium. Because pDHFR and hDHFR possess similar three-dimensional structures, so Pyr was used for anticancer drug research. Our results showed that Pyr could bind to hDHFR, but the binding capacities of Pyr and hDHFR were weaker than MTX and hDHFR. We found that Pyr could effectively inhibit the proliferation of many cancer cell lines, and the effect is equivalent to MTX in vitro, which suggested that DHFR was a driving force for tumor cell proliferation, and even mild inhibition could significantly affect the proliferation of cells. Besides, we found that Pyr inhibited the EMT, migration, and invasion of lung cancer cells. We further demonstrated that TP might be another target protein of Pyr, which plays an important role in EMT.

DHFR, a folate-dependent enzyme that is related to DNA synthesis in cancer cells, has become a crucial target enzyme of antitumor drugs (26). DHFR positively regulates the proliferation of tumor cells, and its expression is markedly elevated in tumor cells. MTX is an antitumor drug that targets DHFR and exhibits inhibitory activity against lung cancer, breast cancer, acute leukemia, and other malignancies (27–30). Pyr is an antimalarial drug targeting plasmodium dihydrofolate reductase (pDHFR). It can induce tumor cell apoptosis through the bilateral mechanisms of caspases and cathepsins (3). Pyr can also inhibit tumor growth by regulating the activity of matrix metalloproteinases (MMPs) and telomerase (4, 31). Pyr can influence the activity of STAT3 in TUBO and TM40D-MB metastatic breast cancer cells (5). The target protein of Pyr in human cancer cells has not been reported. Our results showed that the tertiary structure of pDHFR is highly similar to that of hDHFR. Pyr can block the proliferation of tumor cells by binding to hDHFR.

Although MTX can inhibit the proliferation of tumor cells, it induces some adverse side effects. Methotrexate (MTX) can lead to alveolar epithelial cell injury followed by pulmonary fibrosis as a result of the recruitment and proliferation of myofibroblasts. MTX induces the EMT-like phenotype of A549 cells accompanied by increased interleukin-6 (IL-6) and transforming growth factor (TGF)-β1, and enhance the migration of A549 cells (20, 21). MTX can also induce the EMT of type II alveolar epithelial RLE/Abca3 cells like TGF-β1 in vitro (19). Our results showed that low doses of MTX can cause the migration and invasion of lung cancer cells and promote EMT of A549 and NCI-H460 cells in vitro. In LLC xenografts in C57BL/6J mice, MTX increased the metastatic tumor nodes on lung tissue.

We found that Pyr not only inhibits the proliferation of lung cancer cells but also blocks the EMT, migration, and invasion of lung cancer cells. Therefore, we speculated that Pyr inhibited the EMT of lung cancer cells through other target. The chemical structure of Pyr is similar to that of thymidine, a substrate of TP, which indicate that TP may be a target of Pyr. The molecular docking simulations and Biacore assay results revealed that Pyr has binding ability with TP. Pyr can also inhibit the enzymatic activity of TP.

TP is closely related to the migration, invasion, and EMT of tumor cells and is an important enzyme in the pyrimidine pathway. TP expression in tumor tissue is elevated relative to that in normal tissue, and low TP expression is associated with prolonged patient survival (32, 33). TP can stimulate the migration of human endothelial cells by specifically activating integrins α5β1 and αVβ (34). TP exerts a chemotactic effect on endothelial cells in vitro, induces tumor angiogenesis, and promotes tumor cell migration in vivo (35–37). TP facilitates cell matrix degradation and tumor invasion by promoting MMP2/9 expression. TP affects the expression of MMPs through the MAPK/Erk2 pathway (38). Given that the activation of the Erk pathway plays an important role in EMT (39), TP may also influence the EMT of cancer cells. In this work, we verified that TP could induce the EMT of lung cancer cells. Because the induction of EMT is the primary mechanism by which epithelial cancer cells acquire malignant phenotypes that promote metastasis, TP may become a target for the development of anti-tumor metastasis drugs. In this work, we found that Pyr inhibits TP-induced ERK2 and MEK2 expression and that Pyr reverses the EMT of cancer cells induced by TP. The results of the TCGA data analysis indicated that patients with lung cancer with high TP expression have poor prognoses and suggested that TP can be a target in the development of drugs against lung cancer.

In conclusion, Pyr not only targets DHFR to repress the proliferation of lung cancer cells but also inhibits EMT of lung cancer cells through the TP/MEK2/ERK2/MMPs pathway and thereby impeding the migration and invasion of tumor cells (Fig. 8I). Pyr may have potential clinical applications as a novel dual effective anti-lung cancer drug.

No potential conflicts of interest were disclosed.

Conception and design: H. Liu, Y. Qin, T. Sun, C. Yang

Development of methodology: Y. Qin, Q. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): H. Liu, Y. Qin, D. Zhai, J. Yang, K. Li, L. Yang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Gu, Y. Tang, W. Zhong

Writing, review, and/or revision of the manuscript: H. Liu, Y. Qin, D. Zhai, S. Chen, T. Sun, C. Yang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Meng, Y. Liu

Study supervision: C. Yang

This study was founded by Foundation for National Natural Science Funds of China (Grant No. 81572838 and 81703581), National Science and Technology Major Project (Grant No. 2018ZX09736005), Tianjin science and technology innovation system and the condition of platform construction plan (Grant No.14TXSYJC00572), Post-doctoral innovative talent support program (Grant No. BX20180150).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.