NS398 as a potential drug for autosomal‐dominant polycystic kidney disease: Analysis using bioinformatics, and zebrafish and mouse models

Abstract Autosomal‐dominant polycystic kidney disease (ADPKD) is characterized by uncontrolled renal cyst formation, and few treatment options are available. There are many parallels between ADPKD and clear‐cell renal cell carcinoma (ccRCC); however, few studies have addressed the mechanisms linking them. In this study, we aimed to investigate their convergences and divergences based on bioinformatics and explore the potential of compounds commonly used in cancer research to be repurposed for ADPKD. We analysed gene expression datasets of ADPKD and ccRCC to identify the common and disease‐specific differentially expressed genes (DEGs). We then mapped them to the Connectivity Map database to identify small molecular compounds with therapeutic potential. A total of 117 significant DEGs were identified, and enrichment analyses results revealed that they are mainly enriched in arachidonic acid metabolism, p53 signalling pathway and metabolic pathways. In addition, 127 ccRCC‐specific up‐regulated genes were identified as related to the survival of patients with cancer. We focused on the compound NS398 as it targeted DEGs and found that it inhibited the proliferation of Pkd1 −/− and 786‐0 cells. Furthermore, its administration curbed cystogenesis in Pkd2 zebrafish and early‐onset Pkd1‐deficient mouse models. In conclusion, NS398 is a potential therapeutic agent for ADPKD.


| INTRODUC TI ON
Autosomal-dominant polycystic kidney disease (ADPKD) is the most common inherited kidney disease with a prevalence of approximately 1 in 1000 to 2500. 1 In addition to its main clinical characteristic of uncontrolled progression of renal cysts, other manifestations include hypertension, neural aneurysm, cardiac valvular disease and metabolic disorders; thus, it is recognized as a systemic disease. [2][3][4] Most patients develop symptoms at the age of 60 years, and renal function declines as the disease progresses. By the age of 70 years, approximately 70% of patients require renal replacement therapy. 5 Although tolvaptan, the first FDA-approved drug for ADPKD, has demonstrated efficacy in slowing the decline rate of renal function, its high price and adverse effects limit its use. 6 ADPKD is mainly caused by mutations in Pkd1 (85%) and Pkd2 genes, which encode polycystin 1 (PC1) and polycystin 2 (PC2) proteins, respectively. When PC1 or PC2 is missing, the levels of intracellular calcium ions decrease, whereas those of cyclic AMP increase.
This affects downstream signalling pathways, cell proliferation and the secretion of cyst fluid, ultimately leading to the formation and growth of cysts. [7][8][9] The hyperproliferation of the cyst-lining epithelium is an important hallmark of ADPKD. The main mechanisms for the excessive proliferation of cells are as follows: (1) the activation of proliferation-related pathways, including mitogen-activated protein kinase (MAPK)/extracellular regulated protein kinases (ERK), phosphoinositide 3-kinase (PI3K)/Akt and mammalian target of rapamycin (mTOR) pathways; and (2) the suppression of the inhibition of cell proliferation due to loss-of-function mutations in PC1 and PC2. 8,9 Previous studies have found that ADPKD and tumours share many mechanisms and signalling pathways; thus, ADPKD has been called 'neoplasia in disguise'. 10,11 Although some studies have reported a lower risk of cancer in patients with ADPKD, presumably owing to differences in detection methods, follow-up durations and sample sizes, it is widely accepted that patients with ADPKD have a higher risk of tumours, especially kidney tumours. [12][13][14][15][16][17] Among the types of kidney tumours, clear-cell renal cell carcinoma (ccRCC) is the most frequent histological subtype, accounting for 75% of all kidney tumour cases. 2 Considering the similarities between ADPKD and cancer, some researchers have proposed that the therapeutics applied in cancer may be repurposed in ADPKD. Moreover, a better understanding of the critical processes that are similar between ADPKD and cancer can promote the repurposing of cancer therapeutics for ADPKD. 9 We have previously developed the Renal Gene Expression Database (RGED), a bioinformatics database comprising comprehensive gene expression datasets of studies on renal diseases published in the NCBI Gene Expression Omnibus (GEO) database. 18 The Connectivity Map database (CMap), which shows the relationship among diseases, genetic perturbation and drug action, was constructed by Lamb et al. in 2006. 19 CMap contains a reference database that includes gene expression profiles derived from cultured human cells treated with multiple perturbagens. It provides a method to screen a ranked list of small molecular compounds based on the connectivity scores, which are calculated by the Kolmogorov-Smirnov statistic, by comparing disease gene expression patterns with the reference database. A significantly negative correlation between the disease-related gene expression pattern and the gene profile after compound treatment suggests that the compound has therapeutic potential for the disease. Therefore, using the CMap database, we can discover new indications for existing drugs. [19][20][21][22] To clarify the similarities and dissimilarities between ADPKD and ccRCC and screen and identify candidates for ADPKD therapy, we designed an integrated bioinformatics approach based on the aforementioned rationale and conducted corresponding experiments for validation. Thus, our results will provide insights into the relationship between ADPKD and ccRCC, as well as a new therapeutic strategy for ADPKD.

| Microarray datasets
We collected the gene expression profiles of renal cystic tissues of patients with ADPKD and tumour tissues of patients with ccRCC from RGED (http://rged.wall-eva.net) and the high-throughput gene expression datasets from GEO (http://www.ncbi.nlm.nih.gov/geo/).  Significant GO terms and KEGG pathways were defined as p < 0.05.

| Function enrichment, protein-protein interaction (PPI) network and survival analysis
The interaction between proteins was identified using the String database (http://strin g-db.org, v10.0). Kaplan-Meier plots and log-rank test results were retrieved using the cgdsr, survival and survminer package in R. The high-expression group was defined as those with expression value > 75% quantile, whereas the low-expression group was defined as those with expression value < 25% quantile. p < 0.01 was set as the cut-off value.

| Connectivity Map analysis
The CMap (build 02, https://porta ls.broad insti tute.org/CAMP/) provides gene expression profiles from cultured human cells treated with bioactive small molecules and can be used to discover functional connections among diseases, genetic perturbation and drug action. 22 Therefore, we queried the connectivity scores between the profile of interest and the instances in the database based on the Kolmogorov-Smirnov statistic. The connectivity score is a value between −1 and +1, similar to correlation coefficients, with a negative score representing a counteracting effect of the compound on the expression pattern of the input profile, which in this case was the common DEGs. The cut-off value was defined as a connectivity score < −0.7 and P value < 0.05. The control group was administered the same volume of DMSO.

| Methylthiazolyldiphenyl-tetrazolium bromide (MTT) assays
Pkd1 −/− and 786-0 cells were seeded in culture plates and incubated for 12 h at 33 and 37°C, respectively. NS398 was then administered for 1, 2 or 3 days, respectively. DMSO was used as the vehicle for the compound. After treatment with MTT (Invitrogen-Life Technology) for 4 h, DMSO was added to dissolve the crystallized products.
Absorbance at 490 nm was measured. The measurement was repeated in six duplicate wells.

| Cell cycle analyses
Cell cycle staining was performed using a propidium iodide (PI) cell cycle kit according to the manufacturer's instructions (Multi Sciences Biotech). Briefly, cells were harvested, washed twice with phosphate-buffered saline and stained with 50 µg/ml PI. Cells were then counted in a flow cytofluorometer based on red emissions at 630 nm, and data were analysed using a Beckman Coulter CyAn ADP analyzer.

| EdU staining
EdU assays were conducted following the manufacturer's instructions (Beyotime). After adding 2× EdU working solution to each well, 24-well plates were placed in the incubator (33°C for Pkd1 −/− or 37°C for 786-0 cells) for 4 h. After fixing with 4% formaldehyde, the click reaction cocktail (0.2 ml) was added to each well, and the DNA was stained with Hoechst. The samples were imaged using Olympus BX-51 fluorescence microscopy.

| Western blot analyses
A column tissue and cell protein extraction kit (Minute) was used to extract the proteins of cultured and processed cells and mouse kidney tissues. Protein was quantified using a BCA kit (Thermo Fisher Scientific) following the manufacturer's instructions. The protein concentration of each sample was balanced by double-distilled H 2 O (ddH 2 O) and 5× protein loading buffer (Yamei) was added to the protein sample. The solution was boiled in a 95°C dry heater for 15 min.
An equal amount (20 μl) of protein was loaded onto an SDS-PAGE gel and electrophoresed, and separated proteins were transferred onto a polyvinylidene fluoride membrane. After incubation with the appropriate primary and secondary antibodies, development was performed using an enhanced chemiluminescence reagent (Thermo Fisher Scientific). Densitometric analysis for the determination of relative protein expression was performed using Image lab system (Bio-Rad Laboratory), with GAPDH as a loading control.

| Animal studies
Wild-type zebrafish (Nüsslein-Volhard Laboratory) were maintained, mated, and staged as described previously. 24 Pkd2 morpholino (Pkd2 The experiment was independently conducted three times. The early-onset PKD mouse model was generated as described previously. 26 Briefly, tamoxifen (10 mg/kg) was intraperitoneally injected into C57/BL6 Pkd1fl/fl: tamoxifen-Cre mice at postnatal day 10 (P10) to induce Pkd1 deficiency. NS398 was administered every other day via intraperitoneal injection from P13 at 20 μg/g body weight. Mice were killed by isoflurane using a rodent gas anaesthesia machine (Yuyan Machine) at P30; blood was collected by retro-orbital puncture and kidneys were harvested.
The serum blood urea nitrogen (BUN) concentration was measured with a urea assay kit (BioAssay Systems). The kidneys were weighted and collected for morphological analysis and Western blot analysis.
The study protocol was approved by the Second Military Medical University's Animal Care and Use Committee (approval number SMMU-ACUC-20150430).

| Morphological analysis
Kidney sections were fixed in 4% paraformaldehyde buffered solution and embedded in paraffin. Then kidneys were sliced into 2μm-thick sections and stained with haematoxylin and eosin (HE).
Aperio XT and Leica Aperio (Leica Biosystems) were applied to scan all specimens. The size and shape of kidney cysts were quantitatively measured from the sagittal section of the kidney. The cyst index (CI) was determined as previously described by Yang et al. 26 Briefly, the whole kidney image was divided into 20-30 small regions and the cystic area in each split image and the entire kidney area were measured using Image Pro Plus (v6.0). The CI was determined by dividing the sum of the cystic area by the whole kidney area.

| Statistical analyses
Data are presented as the mean ± standard error of at least triplicates or as a representative of three separate experiments. Significance was determined using a two-sided Student's t test. p < 0.05 was considered statistically significant.

| Identification of DEGs from GEO datasets
Detailed information about the GEO datasets is summarized in  down-regulated DEGs were found to be common to both datasets ( Figure 1B). From the PPI network graph, two modules were related to cell cycle and energy metabolism ( Figure 1C). Heatmaps showed the expression profiles of the common DEGs ( Figure 1D).

| Connectivity Map identified NS398 as a potential therapeutic compound
Using the CMap database, three compounds were identified as potential compounds that reverse the transcriptional changes shared by ADPKD and ccRCC, as they had negative mean connectivity scores (Table 2). Among them, the drug NS398, which has been studied in laboratory experiments, has an inhibitory effect on tumour progression. Thus, it was selected as a novel drug for further studies to confirm its value in ADPKD treatment.

| NS398 inhibits cell proliferation in vitro
We observed an inhibitory effect on cell proliferation in both   Figure 2F).

| NS398 down-regulates cell cycle proteins and proliferation-related signalling pathways
The abundance of cyclin D1 and p21, as representative proteins in-  Figure 4B).
Similarly, we observed the attenuation of cystogenesis in earlyonset Pkd1 conditional knockout mice. NS398 treatment remarkably attenuated the increase in kidney size in early-onset Pkd1 conditional knockout mice ( Figure 4C,D). The kidney weight-to-body weight ratio ( Figure 4E) and the CI ( Figure 4F) in NS398-treated mice were both markedly reduced. The decreased rate of cyst growth was accompanied by improved renal function, as evidenced by BUN levels ( Figure 4G). Western blot analysis indicated that the levels of phospho-and total Akt and ERK significantly decreased in NS398-treated Pkd1 conditional knockout mice ( Figure 4H).
Therefore, NS398 inhibited cyst growth in both Pkd2 zebrafish and early-onset Pkd1-deficient mouse models and improved the renal function in early-onset Pkd1-deficient mouse model.

| DISCUSS ION
In this study, we searched the RGED and NCBI-GEO databases to retrieve datasets pertaining to ADPKD and ccRCC and conducted a comprehensive bioinformatics analysis to unravel their convergences and divergences. Although many studies have investigated the similarities and dissimilarities between ADPKD and ccRCC using laboratory studies, to the best of our knowledge, this is the first report to unravel their links using high-throughput research. Thus, our findings provide a better understanding of their critical mechanisms.
Our results showed that upregulated common DEGs are mainly The most significant difference between ADPKD and cancer is the invasive and metastatic nature of cells observed in the latter.
Most deaths in patients with cancer are related to uncontrolled invasion and metastasis, which is a complex process. 35 Therefore, we also analysed the specific upregulated DEGs in ccRCC and found that the high expression of PPP1R18, PLAUR, TMEM44, JAK3, PTTG1 and ENTPD1 was related to the poor prognosis of patients. PLAUR, JAK3, PTTG1 and ENTPD1 have been reported to be associated with ccRCC, whereas to the best of our knowledge, no study has reported the association of PPP1R18 and TMEM44 with ccRCC. 36   in the treatment of ADPKD and/or tumours. Basic and clinical studies are warranted to validate our results.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are publicly available