Introduction

Due to the high diversity, natural compounds have gained much attention as drug leads for treatment of diseases1. Periplogenin (PPG) is a natural compound extracted from traditional Chinese herb Cortex Periplocae which is widely used for the treatment of autoimmune diseases2. It has been reported to inhibit histamine release of mast cells in antigen-sensitized rats and can also ameliorate skin lesions and inflammation in psoriasis-like murine models3,4. In addition, PPG has shown antiproliferative activities in various cancer cells4,5,6,7,8,9. PPG inhibits cell growth and triggers apoptosis by induction of reactive oxygen species in human colorectal cancer cells whereas in esophageal squamous cell carcinoma it was reported to directly bind to STAT3 and thereby strongly inhibit its phosphorylation at position Tyr705 to suppress cell growth and induce apoptosis5,6 The findings of anti-tumor activity of PPG highlights the potential applications of PPG for the treatment of other cancer types.

Prostate cancer is one of the most lethal cancer types for males10. It is well established that the initial growth of prostate cancer is dependent on androgens, which makes hormonal therapy with chemical or surgical castration the favorable treatment option11,12. However, most patients eventually progress to a castration-independent phase, and the treatment of patients with metastatic castration-resistant prostate cancer remains a clinical challenge12,13,14. This has led to the urgent search for new drugs and exploration of new anti-cancer targets against prostate cancer.

An unbiased forward genetic screening method was established to investigate the molecular mechanism of anti-tumor drugs, leading to the identification of novel anti-tumor targets15. In this study we used this screening technology to demonstrate that PPG suppresses growth of DU145 prostate carcinoma cells by docking to an ATP1A1 protein pocket and forming a hydrogen bond with T804. This highlights the possible application of PPG as an anti-prostate cancer drug and identifies ATP1A1 as a potential therapeutic target.

Materials and methods

Cell culture

DU145 (human prostate cancer), DLD-1 (human colorectal cancer), U-87MG (human glioblastoma) and MCF-7 (human breast cancer) cell lines were purchased from American Type Culture Collection (ATCC) and cultured with Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum and penicillin–streptomycin. All cells were maintained at 5% CO2, 37 °C.

Cell viability assay

Cells were seeded in 96 well plates and cultured for 24 h to 80% confluence. Then, cells were treated with PPG (Desite, DY0084). Dimethyl sulfoxide (DMSO, Sigma, D8418) was used as negative control. Cell viability was measured and quantified by CellTiter-Glo® Luminescent Cell Viability Assay (Promega, G7572) in accordance with the manufacturer’s instructions.

Forward genetic screens

The forward genetic screens were carried out as previously described15. Briefly, DU145 cells were trypsinized and treated with 0.1 mg/ml N-ethyl-N-nitrosourea (ENU, Sigma, N3385) for 2 h in medium at room temperature under rotation. The same procedure was followed to the cells without ENU treatment as control. Then, cells were seeded to 15 cm diameter culture dish at 2 × 106 cells/dish and cultured for 24 h. Periplogenin in medium at the indicated final concentrations was applied to the cells for selection of resistant cells. PPG resistance of emerging cell clones was validated by cell viability assay.

Sequencing and data analysis

Genomic DNA of resistant clones was extracted using TIANamp Genomic DNA kit (TIANGEN, DP304), exome-enriched libraries were generated by SureSelect Human All Exon V6 kit (Agilent, 51900-8863) and sequenced using NovaSeq 6000 paired-end 150. Exome sequencing depth of each sample was at least greater than 150X and data were analyzed as previously described15. In brief, clean reads were aligned to the reference genome (hg19) with bwa (v.0.7.15). Reads were converted to bam format with samtools (v.1.3.1) and sorted with Picard (2.25.0). GATK (v.3.8) Realigner TargetCreator was then used to generate a list of positions for GATK IndelRealigner. After indexing, base recalibration was performed with GATK Base Recalibrator and recalibrated reads were printed with GATK PrintReads. Samtools and GATK pileups were then used to identify variants in the respective samples. After discarding indels, the frequency of the mutation sites in the control sample was less than 0.05, the read depth of the mutation sites in each resistant clone was not less than 10, and the gene frequency was not less than 0.2. After variant annotation with snpEff (v.4.2) variants relating to a moderate or high effect in protein coding were kept.

Molecular docking

The full-length amino acid sequence of human ATP1A1 protein (UniPort: P05023) was downloaded from the UniPort database. The three-dimensional (3D) structure of the predicted full-length ATP1A1 protein (AF-P05023-F1) was obtained from the AlphaFold protein Structure Database, the structure of periplogenin was downloaded from PubChem (CID 10574). The PyMol™ Molecular Graphic System (Version 2.3.0) was used to label the mutated amino acid sites on the 3D structure of the protein and to predict the drug-protein binding pocket based on UniPort annotation. The molecular docking process was carried out using AutoDock Vina 1.1.2 following the parameters: center_x = 18.118, center_y = 5.762, center_z = -37.883, size_x = 52.0, size_y = 62.0, size_z = 48.0, exhaustiveness = 64, num_modes = 9.

CRISPR genome editing

The construction of cell lines with ATP1A1 gene point mutation was performed in DU145 with CRISPR/Cas9 technology as previously described16. DNA sequences of donor homology arms and small guide RNAs were designed with Invitrogen™ TrueDesign™ Genome Editor and synthetized in GENEWIZ. Small guide RNAs were further prepared using GeneArtTM Precision gRNA Synthesis Kit (Thermo Fisher, A29377). Prepared small guide RNAs, corresponding homology arm DNA, and TrueCutTM Cas9 protein v2 (Thermo Fisher, A36498) were co-transfected to DU145 cells using Lipofectamine™ CRISPRMAX™ Cas9 Transfection Reagent (Thermo Fisher, CMAX00001). Transfected cells were maintained in medium containing periplogenin for the selection of resistant cell clones. Resistant cell clones were transferred to 12 well plate and subjected to PCR amplification and Sanger sequencing. The primers targeting ATP1A1 specific region used for the amplification were as follows: Forward 5′- AAGATCTTATTCAGATACTG-3′ and Reverse 5′-CACTACTATCGATCTGTGAC-3′. The sequences of synthesized DNA were followed: 804 mutation Donor homology arm DNA 5′-TTCCTGATATTTATTATTGCAAACATTCCACTACCCCTGGGGGCTGTCACCATCCTCTGCATTGACTTGGGCACTGAC-3′, and 804 mutation guide RNA 5′-GCAAACATTCCACTACCACT-3′.

Reverse transcription real-time quantitative PCR (RT-qPCR)

RT-qPRC was performed as previously described with minor changes17,18,19. Briefly, cells were harvested, and RNA was extracted with TRIzol (Thermo Fisher, 15596026CN). The reverse transcription was carried out with Hiscript III Reverse Transcriptase (Vazyme, R302-01). qPCR was carried out using PowerTrack™ SYBR Green master mix (Thermo Fisher, A46112). The GAPDH was used as control for normalization. The primers of GAPDH were: Forward 5′-AGGTCGGAGTCAACGGATTT-3′ and Reverse 5′-AGTTAAAAGCAGCCCTGGTGA-3′. The primers for ATP1A1 were: Forward 5′-AGTATGAGCCTGCAGCTGTT-3′ and Reverse 5’-TGCACGAGCAGATGTTAATCC-3′.

Western blotting

Cells were washed with PBS and harvested. Total protein was extracted with a total protein extraction kit (Epizyme, PC201) and separated with an ExpressPlus™ PAGE gel (Genscript, M42012) and then followed by transferring to a methanol treated polyvinylidene difluoride (PVDF) membrane (Thermo Fisher, 88518). After blocking with 5% milk for 1 h at room temperature, the membrane was incubated with an anti-ATP1A1 antibody (Abways, CY5159) at 4 °C overnight and then probed with a secondary antibody conjugated with horseradish peroxidase (HRP) for 1 h at room temperature.

Measurement of Na+/K+ -ATPase activity

Cell lysates were harvested, and the protein concentrations were determined using BCA Protein Assay Kit (KeyGEN Biotech, KGB2101). Na+ /K+ -ATPase activity was measured according to the manual of the Na+ /K+ -ATPase activity assay kit (Sangon Biotech, D799642).

Animal trial

Male BALB/c nude mice (4–6 weeks old) were purchased from Shanghai Lingchang Biotechnology Co., Ltd. 5 × 106 DU145 cells with T804A point mutation in ATP1A1 (DU145-804MUT group, n = 18) or wild type DU145 cells (DU145 group, n = 18) in a 100 μL 1:1 mixture of PBS and matrigel were subcutaneously injected in the right axillary region of mouse. When the tumor volume reached 100mm3, each group was randomly divided into the following three subgroups (n = 6 per subgroup): DMSO and PPG (10 mg/kg and 20 mg/kg). PPG or DMSO were administered via oral gavage daily.

Statistical analysis

Results were presented as means plus standard deviations. Statistical analysis was done with PRISM 9 (GraphPad Software, USA) using 2-tailed Student’s t test or One-way ANOVA. P-values for scoring genes in the whole exome sequencing data (Fig. 1B) were calculated based on the number of unique mutations identified per gene using cumulative binominal distribution function (CDF). The total number of identified mutations in the dataset was quantified using a threshold of 10 reads and 0.2 allele frequency. The average number of mutations per clone (201.2) was divided by the exome size (30 M) and a p-value to hit a single gene at least once was calculated using CDF, this accounts for larger genes being hit more likely compared to shorter genes. Then for every gene with at least one mutation (1,572 genes), the p-value to find at least a consistent number of each gene's unique mutations in 9 clones was again calculated using CDF. Each experiment was repeated at least three times unless otherwise specified.

Fig. 1
figure 1

Forward genetic screen identifies ATPAT1 as target of periplogenin. (A) Schematic of the forward genetic screen approach. (B) Forward genetic screen results. The -log10(p-value) of the analysis in the screen was plotted on the y-axis and 1572 candidate genes were organized by chromosome position on x-axis, the size of the dot exhibits the clone numbers containing the identified mutations of corresponding gene. (C, D) Molecular docking of ATP1A1 with periplogenin. (E) The effect of periplogenin treatment on the Na+ /K +-ATPase activity. The cell lysates were derived from wild-type DU145 cells (WT) and CRISPR/Cas9 engineered DU145 cell clones with T804A substitution (T804A). * and ** indicate p values of < 0.05, < 0.01, respectively.

Full size image

Results

Periplogenin shows anti-tumor activities in multiple cancer cell lines

To uncover the direct molecular target of PPG, we first identified a cell model amenable to subsequent forward genetic target identification screens. Therefore, the cytotoxicity of PPG was determined in DU145, DLD-1, U-87MG, and MCF-7 cells by a cell viability assay. The 90% inhibitory concentration (IC50) values of periplogenin for DU145, DLD-1, U-87MG, and MCF-7 cells were 2.897 µM, 8.712 µM, 3.008 µM, and 20.025 µM, respectively (Supplementary Fig. 1A). The IC90 value of PPG in DU145 was calculated to be lower than to DLD-1, U-87MG, and MCF-7 cells (Supplementary Fig. 1A). Since the applied forward genetic screening approach requires highly effective killing, DU145 cells were further characterized by studying time- and concentration-dependent PPG toxicity to define screening conditions (Supplementary Fig. 1B).

ATP1A1 is identified as target of periplogenin

Next, a forward genetic screen was carried out in the selected DU145 cells. Cells were treated with ENU, a chemical mutagen that causes point mutations throughout the cell genome and facilitates the generation of drug-resistant clones. These clones were then selected for using PPG treatment of 25 µM and 50 µM, respectively (Fig. 1A). Non-mutagenized cells were treated with the same PPG concentration function as a negative control for the screen. After 4 weeks post-treatment, we observed 9 cell clones under selection of PPG in the ENU mutagenized population, while no colonies were observed in the non-mutagenized cell population. The resistance of all cell clones was confirmed by cell viability assay (Supplementary Fig. 2).

To identify the resistance-causing genetic underpinnings, we performed next-generation whole exome sequencing. Our analysis revealed that all examined clones harbor a point mutation in ATP1A1 (Fig. 1B and Supplementary Table 1). The obtained mutations affected the amino acids A108, N129, L800, and T804 (Table 1). Notably, one clone (clone 4) harbored two mutations in the ATP1A1 coding region, A108E and T804I (Table 1).

Table 1 Identified mutations of ATP1A1 gene in resistant cell clones.
Full size table

Identified amino acid positions are essential for PPG activity

To explore the role of the identified amino acid position in PPG activity, we generated the predicted 3D structure of ATP1A1 protein with labeling of the mutated amino acids. Interestingly, although distributed over distinct functional domains, A129, L800, and T804 are located in a potential drug binding pocket (Supplementary Fig. 3A-E). We further assessed the possible binding region of PPG with ATP1A1 by performing molecular docking analysis. The analysis revealed PPG docked to the anticipated ATP1A1 pocket with a binding affinity of − 9.8 kcal/mol, including A129, L800, and T804 (Fig. 1C). Strikingly, PPG was able to form a hydrogen bond (2.2 Å) within the binding pocket of ATP1A1 protein to T804 (Fig. 1D). Since A108 is distant to the potential pocket and clone 4 harbors an additional T804I mutation, it is likely that the A108E alteration is just a bystander mutation without or with limited effect on PPG resistance. To validate the modelling data, we performed CRISPR/Cas9-mediated genome editing to mutate the potential hydrogen bond donor/acceptor by introducing the T804A substitution. Under selection of periplogenin, we observed and confirmed cell clones with T804A substitution using sanger sequencing. We randomly selected two cell clones with T804A substitution for the subsequent experiments.

We performed the in vitro Na+ /K+ -ATPase activity assay to confirm the direct effect of PPG on ATP1A1 protein. The results revealed that PPG inhibits the activity of ATP1A1 in a dose-dependent manner (Supplementary Fig. 5). The T804A mutation of ATP1A1 protein shows no obvious effect on the activity. Importantly, engineered ATP1A1 protein demonstrates better resistance to PPG treatment in the Na+ /K+ -ATPase activity than wild type ATP1A1 protein (Fig. 1E). Meanwhile the treatment of PPG does not lead to expression change of ATP1A1 protein (Supplementary Fig. 4D).

Next, we validated the resistance in two independent T804A engineered clones using a cell viability assay. Both clones are highly resistant to PPG treatment at all tested concentrations (Fig. 2A). The resistance could neither be attributed to altered ATP1A1 expression levels nor to a general effect on cell growth in both clones (Supplementary Fig. 4A-C).

Fig. 2
figure 2

DU145 cells with the T804A substitution are resistant to periplogenin treatment, both in vitro and in vivo. (A) CRISPR/Cas9 engineered DU145 cell clones with T804A substitution (T804A) were treated with periplogenin for 72 h in the indicated dose and compared to wild type DU145 cells (WT). (B) Comparison of relative tumor volumes among all groups, the tumor volume was measured using caliper every 2 days and calculated as length x width2 × 0.5, then was normalized to tumor volume of 0 day post-administration, the relative tumor volumes of the last time point were compared. (C, D) Comparison of tumor weight and size. * and ** indicate p values of < 0.05, < 0.01, respectively, ns, not significant.

Full size image

Periplogenin suppresses tumor growth in DU145 xenograft tumor model

We next assessed whether the identified single amino acid substitution could ameliorate PPG anti-cancer activities in vivo. To this end, we compared DU145 WT to DU145 ATP1A1 (T804A) cells in xenograft studies. While PPG treatment dramatically reduced the relative tumor volume and tumor weight in DU145 WT xenograft models, DU145 ATP1A1 (T804A) cell-based models were largely insensitive to PPG treatment in this setup (Fig. 2B–D). Notably, none of the mouse groups showed significant changes in body weight (Supplementary Fig. 6).

In conclusion, PPG inhibits cell growth of multiple cancer cell lines including DU145, DLD-1, U-87MG, and MCF-7. The inhibition of PPG to prostate cancer cell DU145 results from direct binding to ATP1A1 protein by forming a hydrogen bond with T804, revealing mode of action at the amino acid level.

Discussion

Natural compounds have been extensively explored as anti-tumor therapeutics. Amongst them, PPG has been reported to show therapeutic effects in various cancer types in vitro and in vivo5,6,20. Mechanistically, PPG was reported to induce reactive oxygen species to inhibit the growth of human colorectal cancer cells and it also directly interacts with STAT3 protein to suppress the growth of esophageal squamous cell carcinoma5,6. Nevertheless, the direct molecular target of PPG, in other cancer models, remained largely elusive.

In the current study, we show that PPG significantly inhibits cell growth in various cancer cells including DU145 (human prostate cancer), DLD-1 (human colorectal cancer), U-87MG (human glioblastoma) and MCF-7 (human breast cancer). Unbiased forward genetic screens reveal that PPG suppresses DU145 prostate cancer cell growth in vitro and in vivo by targeting ATP1A1 via a hydrogen bonding to T804.

The Na+/K+-ATPase is composed of four α-, three β-, and one γ- subunit isoforms and establishes a transmembrane sodium gradient by exchanging three sodium ions with two potassium ions in each pump cycle, by utilizing ATP. It is further described to act as intracellular signal transducer and membrane receptor21,22,23,24,25,26. ATP1A1, α1 subunit, contains the ATP and cation binding sites and is the catalytic subunit of the enzyme27. Mutations in ATP1A1 have been associated with altered cell proliferation and tumorigenesis in aldosterone-producing adenomas28,29,30. In addition, the aberrant expression of ATP1A1 is widely observed in multiple types of tumors. Clinically, the expression of ATP1A1 is significantly decreased in human renal cell carcinoma tissues compared to non-tumor tissues whereas abnormalities in expression are observed in breast cancer samples31,32. All these reports indicate ATP1A1 might play an important role in tumorigenesis.

In this study, we show direct interaction between PPG and ATP1A1 to account for PPG antiproliferative activities, which further indicates ATP1A1 could be a druggable target in anti-tumor treatment. Presumably, PPG binds to ATP1A1’s catalytic subunit and inhibits the enzymatic function leading to a disturbed ion homeostasis and the induction of apoptosis33.

We still observe minimal growth inhibition by PPG when DU145 cells carry an A804T substitution in ATP1A1, both in vitro and in vivo. This could be explained by an attenuated affinity between PPG and ATP1A1 or direct binding of PPG to secondary targets such as STAT3 and others.

In summary, our study demonstrates that PPG exhibits growth inhibitory effects to DU145 cells in vitro and in vivo by directly binding to ATP1A1 and forming a hydrogen bond with T804. Unravelling the molecular mode of action of PPG at amino acid resolution not only allows to induce rationally designed compound modifications, but also to systematically select most relevant indications in the future.