Prostate cancer is the second leading cause of noncutaneous cancer-related deaths in American men. Androgen deprivation therapy (ADT), radical prostatectomy, and radiotherapy remain the primary treatment for patients with early-stage prostate cancer (castration-sensitive prostate cancer). Following ADT, many patients ultimately develop metastatic castration-resistant prostate cancer (mCRPC). Standard chemotherapy options for CRPC are docetaxel (DTX) and cabazitaxel, which increase median survival, although the development of resistance is common. Cancer stem-like cells possess mesenchymal phenotypes [epithelial-to-mesenchymal transition (EMT)] and play crucial roles in tumor initiation and progression of mCRPC. We have shown that low-dose continuous administration of topotecan (METRO-TOPO) inhibits prostate cancer growth by interfering with key cancer pathway genes. This study utilized bulk and single-cell or whole-transcriptome analysis [(RNA sequencing (RNA-seq) and single-cell RNA sequencing (scRNA-seq)], and we observed greater expression of several EMT markers, including Vimentin, hyaluronan synthase-3, S100 calcium binding protein A6, TGFB1, CD44, CD55, and CD109 in European American and African American aggressive variant prostate cancer (AVPC) subtypes—mCRPC, neuroendocrine variant (NEPC), and taxane-resistant. The taxane-resistant gene FSCN1 was also expressed highly in single-cell subclonal populations in mCRPC. Furthermore, metronomic-topotecan single agent and combinations with DTX downregulated these EMT markers as well as CD44+ and CD44+/CD133+ “stem-like” cell populations. A microfluidic chip-based cell invasion assay revealed that METRO-TOPO treatment as a single agent or in combination with DTX was potentially effective against invasive prostate cancer spread. Our RNA-seq and scRNA-seq analysis were supported by in silico and in vitro studies, suggesting METRO-TOPO combined with DTX may inhibit oncogenic progression by reducing cancer stemness in AVPC through the inhibition of EMT markers and multiple oncogenic factors/pathways.

Significance:

The utilization of metronomic-like dosing regimens of topotecan alone and in combination with DTX resulted in the suppression of makers associated with EMT and stem-like cell populations in AVPC models. The identification of molecular signatures and their potential to serve as novel biomarkers for monitoring treatment efficacy and disease progression response to treatment efficacy and disease progression were achieved using bulk RNA-seq and single-cell-omics methodologies.

Prostate cancer is the second leading cause of noncutaneous cancer-related deaths in American men (www.cancer.org). The androgen receptor (AR) signaling pathway plays a pivotal role in prostate development and homeostasis, as well as in the progression of prostate cancer (1). Androgen deprivation therapy (ADT), following radical prostatectomy or radiotherapy, remains the main treatment for more advanced cases of castration-sensitive prostate cancer (CSPC). In many patients, ADT effectively suppresses prostate cancer during the first 12–24 months (1–3). However, many patients ultimately develop resistance and show metastatic spread, that is, metastatic castration-resistant prostate cancer (mCRPC) disease (4). The 2020 estimates of transition from non-castrate to mCRPC are approximately 15%, with a mortality rate of 19.5% (4). Neuroendocrine differentiation following ADT to aggressive treatment-resistant neuroendocrine prostate cancer (NEPC) has been estimated at >25% (5). Standard treatment options for CRPC/NEPC include sipuleucel-T, abiraterone acetate plus prednisone (AA/P), or chemotherapy with docetaxel (DTX; refs. 6–8). Cabazitaxel, AA/P, enzalutamide, and radium-223 are approved treatments of CRPC, often following DTX (6, 7). These combination treatments can increase median overall survival (OS) by approximately 1 year (9). Resistance development, characterized by increased PSA levels, is almost universal, resulting in a progression-free survival rate of around 0% after 3 years, often accompanied by significant side effects (10–12). Furthermore, African American (AA) men are more likely to be diagnosed or progress more rapidly to aggressive forms of mCRPC compared with other ethnicities (seer.cancer.gov). Moreover, overall treatment options remain limited, and survival is poor.

Several groups, including ours, have demonstrated that the presence of cancer stem-like cells (CSC) in tumors displaying mesenchymal phenotypes (epithelial-to-mesenchymal transdifferentiation, EMT) and CD44+/CD133+ cells, with self-renewal and differentiation capacities, play crucial roles in tumor progression and development of mCRPC (13–16). Furthermore, tumor heterogeneity and host-tumor microenvironments with divergent genetic profiles and molecular signatures represent a challenge for current therapies (17). Therefore, it is necessary to develop new therapeutic approaches to overcome drug resistance in European American (EA) and AA patients with prostate cancer, specifically for mCRPC/NEPC, to improve efficacy and increased OS.

Low-dose continuous drug exposure using metronomic-like (METRO) chemotherapy involves frequent administration of chemotherapeutic agents at low or fractionated doses at close intervals over prolonged periods of time (18, 19). METRO is an emerging treatment option that has shown promise for various cancer types, including prostate cancer (20–23). In our previous study, we reported that metronomic topotecan treatment (METRO-TOPO) was 2.4- to 18-fold more potent (P < 0.05) compared with conventional topotecan (CONV-TOPO) treatment in prostate cancer cell lines (22). We then performed animal study with METRO-TOPO versus CONV-TOPO and demonstrated that by day 17, METRO-TOPO treatment resulted in significantly (P ≤ 0.05) smaller tumor volume (65.4% ± 11.2%) compared with control (136% ± 14%) and CONV-TOPO–treated animals (138% ± 10%) in an aggressive xenograft tumor model of human prostate cancer implanted in male NCr athymic mice (22). This overall antitumor activity of METRO-TOPO was maintained through the end of the study, animals receiving METRO-TOPO dosing regimens had significantly (P ≤ 0.05) smaller tumor volumes (54.8% ± 16.5%) than animals receiving CONV-TOPO dosing (144% ± 11%) or the control group (207% ± 26%; ref. 22). To further investigate the impact of low-dose therapy, the effect of METRO-TOPO, CONV-TOPO intravenous administration on tumor volume was compared with implantation of ALZET micro-osmotic pumps in nude mice after tumor xenografts reached 200–300 mm3 (22). ALZET pumps were primed to deliver 2.45 mg/kg/day to achieve plasma concentrations at the experimentally determined IC50, that is, 4–5 ng/mL and 0.10 mg/kg/day (4% of the IC50 concentration; ref. 22). After 21 days of treatment, animals receiving “metro-like” dosing at 2.45 and 0.1 mg TOPO/kg/day had significantly (P ≤ 0.05) smaller tumor volumes compared with ALZET control animals with no observable toxicities (22). Molecular analysis revealed METRO-TOPO treatment antitumor activity was associated with the inhibition of major cancer pathway genes, angiogenesis, increasing tumor hypoxia, or normalizing the tumor vasculature to improve blood flow and drug delivery (21, 22). Several other studies have reported that low-dose oral TOPO is potent for patients with cancer (24–26). However, the clinical benefit of METRO-TOPO and a comprehensive understanding of its mechanism of action are not fully known.

Therefore, we performed pretreatment versus posttreatment bulk and single-cell RNA sequencing (scRNA-seq) to identify differentially expressed genes (DEG) for prostate cancer subtypes of EA and AA origin and potential molecular pathways associated with METRO-TOPO activity in aggressive variant prostate cancer (AVPC; mCRPC, NEPC, and EMT) at the tumor and subclonal levels and to gain molecular insights into METRO-TOPO activity. Furthermore, our RNA-seq and scRNA-seq data showed that METRO-TOPO treatment was potentially effective against the development of prostate cancer subclones with a reduction of treatment-resistant stem-like genes. Using in vitro model systems of treatment-refractory and treatment-emergent AVPC, we demonstrated that the METRO-TOPO showed efficacy as a single agent, as shown previously. We showed synergistic activity in combination with conventional taxanes (CONV-DTX) and METRO-TOPO. A microfluidic chip-based confined cell invasion assay was performed. The assay recapitulates the dimensionality of pores and longitudinal channel-like tracks encountered by cancer cells during migration to investigate the effect of METRO-TOPO on cancer cell invasion and stemness. Flow cytometry and cell sorting with prostate cancer stemness-specific antibodies (CD44, CD133) were accomplished to identify differential percentages of stem-like cell populations (CD44+, CD133+, and CD++) in all prostate cancer subtypes, including EA and AA, to evaluate the impact of METRO-TOPO in eroding “stem-like” (CD44+/CD133+) cell subpopulations. The effect of METRO-TOPO treatment on EMT gene expression (GE) and protein expression was determined, and a comparative analysis with whole-genome transcriptomics data and reverse matching DEGs from patients with prostate cancer were used to examine the potential clinical significance of METRO-TOPO treatments.

Therefore, using an innovative approach that integrated advanced molecular techniques [next-generation mRNA sequencing (mRNA-seq) and single-cell-omics technologies] with microfluidics, flow cytometry, MoFlo XPD Flow high speed-based cell sorting, cytotoxicity profiling, immunoblotting as in vitro studies, and patient databases (The Cancer Genome Atlas/TCGA for in silico validation), we conclude that METRO-TOPO has the potential to improve the clinical outcome in AVPC chemotherapy by enhancing the therapeutic efficacy of standard-of-care drugs and abrogating the possibilities of development of drug resistance. Such an evidence-based approach promises to minimize the chances of trial failures and improve the probability of clinical success.

Chemicals and Reagents, details are provided in the Supplementary Materials and Methods.

Human Prostate Cancer Cell Lines

ARLow mCRPC/NEPC (PC3, PC3M, DU145), ARHigh EA mCSPC (LNCaP, 22Rv1, VCaP), ARHigh EA mCRPC (C4-2B), ARHigh AA mCSPC (MDA-Pca-2b), and normal prostate (RWPE1, RWPE2) cell lines were obtained from ATCC. AA-ARHigh CSPC (RC77T, RC165T) and AA-ARLow mCRPC (RC43T) were obtained from our collaborator's lab (Dr. Clayton Yates, Sidney Kimmel Comprehensive Cancer Center, Johns Hopkins University School of Medicine, Baltimore, MD; ref. 27). The taxane-resistant prostate cancer cell lines PC3-TXR and DUTXR were obtained from our collaborator's lab (Dr. Amit K. Mitra, Department of Drug Discovery and Development, Harrison College of Pharmacy, Auburn University, Aurburn, AL). The taxane-resistant cell lines PC3-TXR and DUTXR were generated using dose escalation with taxanes over time (28). Cell lines were authenticated at the source and tested randomly at regular intervals for provenance and cell lineage at the Center for Pharmacogenomics and Single-Cell Omics (AUPharmGx) using GenePrint 24 System. All cell lines were Mycoplasma negative and cultured with appropriate media (details provided in Supplementary Materials and Methods) at 37°C, 21% O2, and 5% CO2 in a humidified cell incubator (Thermo Fisher Scientific). Cells were subcultured when they reached approximately 80%–90% confluence. Cell lines were used until passage 30. After removing from the cryo-storage (passage 2–3), cells were cultured and passaged for 2–3 weeks before beginning studies.

Treatment Schedules for mRNA-seq and scRNA-seq

Three-dimensional (3D) spheroids were harvested and transferred into a 96-well plate in 100 μL of recommended media. After 2 days of acclimation, 100 μL of media ± drug/well were added to each well containing a spheroid. On days 3 and 5, 100 μL of media/well were removed and replaced with 100 μL of fresh media with/without (±) drug/well. On off-media exchange days, 10 μL of media/well was removed and replaced with 10 μL of media ± drug. METRO-TOPO dosing was simulated by using 20x TOPO directly spiked into the wells at 10 μL in 190 μL of media. The CONV-TOPO was added as a bolus dose on day 0. METRO/extended exposure (EE)-TOPO treatment was given daily as a fractionated dose at 1/7th the CONV-TOPO. Total topotecan dosing was 100 nmol/L during each week of therapy. Control and treated spheroid were collected and stored in RNAlaterer (Qiagen) for RNA-seq. Fresh/live samples were used for scRNA-seq.

Pretreatment and Posttreatment Tumor mRNA-seq

GE of all prostate cancer and normal prostate cell lines at baseline (no treatment) was assessed by RNA-seq. Furthermore, the effects of METRO-TOPO and CONV-TOPO exposure for 6 weeks on ARLow mCRPC/NEPC PC-3 tumor model (3D spheroid) were assessed using RNA-seq. Pre- and post-drug exposure, as described above, tumor cells were harvested, and high-quality RNA was extracted using QIAshredder and RNeasy kit (Qiagen). RNA concentration and integrity were assessed using a Nanodrop-8000 spectrophotometer and Agilent 2100 Bioanalyzer. An RNA integrity number threshold >8 was applied, and RNA-seq libraries were constructed using Illumina TruSeq RNA Sample Preparation kit v2. Libraries were then size-selected to generate inserts of approximately 200 bp. RNA-seq was performed on Illumina's NovaSeq platform using a 150 bp paired-end protocol with a depth of >20 million reads per sample. Average quality scores were above Q30 for all libraries in both R1 and R2 (29). AA cell line RNA was isolated from cultured cells using TRIzol Reagent (Sigma Life Sciences) following the manufacturer's protocol (27). Library preparation, quality control, and sequencing of extracted RNA were performed by the Center for Pharmacogenomics and Single-Cell Omics (AUPharmGX).

RNA-seq Data Analysis

RNA-seq data analysis was performed using a command line–based analysis pipeline (DEseq2 and edgeR) and Partek Flow software. Briefly, reads were preprocessed and mapped to the hg38 human genome build using the STAR Aligner tool. Next, mapped read counts were counts per million (CPM) normalized, and differential Gene Expression Profile (GEP) analysis was performed. Genes with mean fold change >|1| and P < 0.05 were considered the threshold for reporting significant DEGs. Heat maps were generated using unsupervised hierarchical clustering (HC) analysis based on top DEGs. Sequencing data on AA cell lines were compiled as FastQ files for downstream analysis (27).

scRNA-seq

scRNA-seq analysis was assessed for prostate cancer cell lines at baseline and TOPO-METRO (EE vs. CONV for 6 weeks, as described above) in the PC-3 tumor model (3D spheroid). Automated single-cell capture and cDNA synthesis were performed using the 10X Genomics Chromium platform. scRNA-seq–based GE analysis was performed on an Illumina HiSeq 2500 next-generation sequencing (NGS) platform by paired-end sequencing technique at 2*125 bp and 100 cycles using v3 chemistry.

scRNA-seq Data Analysis

scRNA-seq datasets were obtained as matrices in the Hierarchical Data Format (HDF5 or H5). CellRanger, Seurat, and Partek Flow software packages were used to preprocess the data analysis. Highly variable genes were selected for clustering analysis based on a graph-based cluster approach. The visualization of cell populations was performed by t-distributed stochastic neighbor embedding (t-SNE) and Uniform Manifold Approximation and Projection (UMAP) for dimension reduction analysis for biomarker-based identification of subclones representing CSC-EMT, taxane-resistant cells, potential METRO-TOPO targeted subclones, and METRO-TOPO treatment-induced erosion of these subclones.

Ingenuity Pathway Analysis

Ingenuity Pathway Analysis (IPA; Qiagen) was performed using top DEGs to reveal molecular pathways/mechanisms, upstream regulator molecules, downstream effects, biological processes, and predicted causal networks governing prostate cancer subtypes and METRO-TOPO functions in AVPC (30).

Patient Samples

In this study, the interactive web portals UALCAN and Gene Expression Profiling Interactive Analysis (GEPIA) were used to compare transcriptome data on target candidate pathway genes with tumor metastasis and patient survival from the prostate expression data matrix (31, 32). GE profiles for patients with prostate cancer included in TCGA database were used. Furthermore, GE data on deidentified patients with prostate cancer were extracted from the Genomic Data Commons server of TCGA database (Genomic Data Commons server/cancergenome.nih.gov). UALCAN (http://ualcan.path.uab.edu/index.html), a web-based tool for analyzing TCGA RNA-seq and clinical data, was used to evaluate the association of GE and patient survival in 31 types of cancers, including prostate cancer (30). Furthermore, GEPIA (http://gepia.cancer-pku.cn/), an interactive web-based tool for survival analysis based on GE, was also used. According to the characteristics of gene normalization, GEPIA allowed two different genes to be input at the same time for survival analysis (32).

Flow Cytometry Detection of CD44+, CD133+, and CD44+/CD133+ Cells

Prostate cancer cells were labeled with a binding buffer containing stemness markers CD44, CD133, or both antibodies. Cells were collected and quantified as CD44+, CD133+, and double-positive cells using a Beckman Coulter Analytical Flow Cytometer—a CytoFLEX LX flow cytometer at 50,000 events/measurement. “Stem-like” cell populations were measured by assessing the shift in the mean fluorescence intensity by flow cytometry. Data were analyzed (gates were set), normalizing to unstained cells of each prostate cancer subtype. Ghost Dye Red 780 was used to detect live cells in each cell population. Furthermore, ARLow cells were seeded in 6-well plates and exposed to CONV-TOPO, METRO-TOPO, and CONV-DTX+METRO-TOPO as combination treatments calculated on the basis of our earlier studies (22, 29). We have determined IC50 for MERO-TOPO, CONV-TOPO, and CONV-DTX as single agents for 48 and 72 hours for all prostate cancer cell lines (21, 22, 28). Plated cells were exposed to TOPO and DTX at the estimated 3-(4, 5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) IC50/2 for each treatment protocol (CONV and METRO) for flow cytometry. For combination, prostate cancer cell lines were treated with estimated MTT IC50, IC50/2 for CONV-DTX and MTT IC50, IC50/2 for METRO-TOPO (Supplementary Table S5). After 72 hours, cells were labeled and quantified for CD44+, CD133+, and double-positive cells with the protocol above. All Flow Cytometry data were analyzed by using FlowJo software (https://www.flowjo.com/).

In Vitro Cytotoxicity (MTT) Assay

In vitro cytotoxicity assays were performed using the MTT assay. Briefly, cells were plated in a 96-well culture plate at 2 × 103 cells/well and incubated for 24 hours at 37°C with 5% CO2. We have determined IC50 for MERO-TOPO, CONV-TOPO, and CONV-DTX for 48 and 72 hours for all prostate cancer cell lines (21, 28). The specific IC50 and IC50/2 are provided in Supplementary Table S5. Cells were treated at the IC50 and IC50/2 estimated by MTT for each treatment protocol (CONV and METRO) for single agent (21, 22, 28). For combination studies, each prostate cancer cell line was treated at IC50 and IC50/2 estimated by MTT for CONV-DTX and METRO-TOPO. Following 72 hours incubation, MTT was added, and absorbance was measured at 550 nm using a Synergy Neo2 Microplate Reader. All experiments were performed in triplicate with five replicate wells for each concentration. Percent change in MTT staining (a surrogate for cell survival) relative to untreated controls was calculated at each drug concentration.

Caspase 3/7 Activity Assay

Cell death by apoptosis was measured using a Caspase-Glo 3/7 kit. Briefly, cells were seeded and treated according to the MTT protocol (described above; Supplementary Table S5; refs. 22, 29). Following 72 hours incubation, Caspase-Glo 3/7 reagent was added, and luminescence was measured using a Synergy Neo2 Microplate Reader. The apoptotic level of each treatment group was normalized to the no drug treatment (baseline) caspase 3/7 for each cell line.

Assessment of Cellular Morphology

Cellular morphology was assessed after cells were seeded at 2.5 × 104 cells (mL/well) in 6-well plates and exposed to TOPO-CONV, TOPO-METRO, CONV-DTX as a single agent and in combination with CONV-DTX+METRO-TOPO according to MTT protocol (described above; Supplementary Table S5; refs. 22, 29). After 72 hours, cells were stained with crystal violet and washed with PBS. Three areas with approximately equal cell densities were identified in each well. Images were captured with an Agilent Cytation5 digital cell imaging system using a 4X objective and Texas Red filter with 559-34 Excitation, 630-69 Emission, and 585 DM. Images were analyzed using ImageJ software (https://imagej.nih.gov/ij/) in a double-blind manner.

Aldefluor Activity (Aldehyde Dehydrogenase) Assay

Aldehyde dehydrogenase (ALDH1) activity was assessed using an Aldefluor assay kit. Briefly, 1 × 104 prostate cancer cells were harvested and resuspended in Aldefluor assay buffer containing the ALDH substrate, BODIPY-amino acetaldehyde (BAAA). Negative control samples were treated with diethylamino-benzaldehyde (DEAB), an inhibitor of ALDH1 enzymatic activity. Cells were incubated at 37°C and suspended in Aldefluor assay buffer. The brightly fluorescent ALDH+ cells were detected by BD LSR II flow cytometry.

Immunoblotting

Pretreatment and Posttreatment cells were lysed in RIPA Lysis and Extraction Buffer (Thermo Fisher Scientific). Briefly, cells were seeded and treated according to the MTT protocol (described above; Supplementary Table S5; refs. 21, 22, 28). Quantification of proteins was performed using a Bradford assay with a BSA protein standard kit (Bio-Rad). Equal amounts of protein were loaded onto 4%–15% Criterion TGX Stain-Free Precast Gels. Proteins were separated under reducing conditions and then transferred to a polyvinylidene difluoride membrane using a Bio-Rad Semi-dry Blotting Apparatus. Nonspecific binding was limited by incubating the membrane in blocking buffer [2.5% (w/v) casein, pH 7.6, 150 mmol/L NaCl, 10 mmol/L TRIS-HCl, and 0.02% sodium azide]. Membranes were then incubated with primary antibodies for the targeted gene/protein (1:1,000), followed by secondary antibodies (1:10,000). Immunoreactivity was detected using enhanced chemiluminescence Western Blotting substrate. Images were captured on a Gel Doc EZ Gel Documentation System using ImageLab software (Azure Biosystems). Densitometry analysis was performed using ImageJ software (https://imagej.nih.gov/ij/).

Microfluidic Cell Migration Assay

The fabrication of the polydimethylsiloxane (PDMS)-based microchannel device using standard multilayer photolithography and replica molding has been reported earlier (33). Briefly, cells were seeded and treated according to the MTT protocol (described above; Supplementary Table S5; refs. 22, 23,34). Pretreated and posttreated 1–1.5 × 105 cells were introduced into the cell seeding inlet line of the microfluidic channel via pressure-driven flow and were allowed to adhere for 10 minutes at 37°C, 5% CO2. Medium supplemented with 10% FBS was added into the chemoattractant inlet line to trigger cell entry into confining [(W) width × (H) height = 3 × 10 μm2] or partially confining (W × H = 10 × 10 μm2) microchannels. The devices were placed on an automated Nikon Ti2 Inverted Microscope equipped with a Tokai stage-top incubator. Cell entry into the channels was recorded via time-lapse microscopy. Images were recorded every 10 minutes for 8 hours with a 10x/0.45 NA Ph1 objective. The percentage of cell entry into microchannels was calculated by measuring the number of cells entering the channels divided by the total number of cells seeded adjacent to the microchannel entrances.

Colony Formation Assay

DUTXR cells were seeded in a 6-well plate at 2.5 × 104 cells (mL/well), incubated overnight, and treated with CONV-TOPO, METRO-TOPO, CONV-DTX as a single agent, and CONV-DTX+METRO-TOPO in combination according to the MTT protocol (describe above; Supplementary Table S5; refs. 22, 29). The cells were harvested from a 24-well plate at 1,000 cells/well and incubated for 1–2 weeks. The colonies were fixed with 100% methanol and stained with crystal violet. Images were taken of cell colonies using an EVOS FL digital cell imaging system (Thermo Fisher Scientific). Images were recorded in brightfield and phase-contrast modes at 1X magnification and analyzed using ImageJ software https://imagej.nih.gov/ij/.

FACS

Cells were stained with stemness markers (CD44), sorted using CD44+ versus CD44 and plated for treatment. After overnight incubation, cells were treated with CONV-TOPO, METRO-TOPO, CONV-DTX, and the combination CONV-DTX+METRO-TOPO according to the MTT protocol (described above; Supplementary Table S5; refs. 22, 29). Cell cytotoxicity, cellular morphology, and apoptosis were estimated using MTT, imaging, and caspase 3/7 assays.

Statistical Analysis

All tests were two sided. Differences with P values <0.05 were considered significant. ANOVA was performed for continuous outcomes, and the Benjamini–Hochberg multiple testing methods were used as a post hoc test. The two-group t test was used to perform differential GE analysis between groups and detect the DEGs. Genes with mean fold changes>|1| and P < 0.05 were considered as the threshold for reporting significant differential GE. Finally, heat maps were generated using unsupervised HC analysis based on the top DEGs. Kaplan–Meier curves were generated using survival time and censored data to plot survival against high versus low expression of significantly associated genes. All statistical analyses were performed using R v4.1.0 and GraphPad Prism v9.0.

Data Availability Statement

The datasets generated during or analyzed during the current study are available at the NCBI Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/), accession number GSE233341 or from the corresponding author upon reasonable request.

GEP Revealed Signatures for Each Prostate Cancer Subtype in EA versus AA Cell Lines

Baseline GEP identified DEGs for each prostate cancer subtype ARHigh/mCSPC (LNCaP, VCAP, 22RV1), ARLow/mCRPC/NEPC (PC-3, PC-3M, DU145) and normal prostate cell lines (RWPE1 and RWPE2; Fig. 1AI; Supplementary Fig. S1A). Among the top 100 (FDR < 0.05, P < 0.05) DEGs for ARLow/mCRPC/NEPC cell lines were the majority of EMT markers (upregulated) such as Vimentin (VIM), hyaluronan synthase-3 (HAS3), Laminin subunit beta-3 (LAMB3), S100 calcium binding protein A6 (S100A6), Serpin-E2 (SERPINE2), insulin-like growth factor binding protein-4 (IGFBP4), Annexin A2 (ANXA2), Annexin A2 pseudogene-2 (ANXA2P2), G protein-coupled receptor class C-5A (GPRC5A), and CD55, whereas downregulated genes were included endothelial transcription factor GATA2, nerve growth factor inducible VGF/VGF, insulin-like growth factor binding protein-2 (IGFBP2), solute carrier family 44-A4 (SLC44A4), guanylate cyclase-1 alpha-1 (GUCY1A1), NK3 homeobox-1 (NKX3-1), kinesin-1A (KIF1A), prune homolog-2 with BCH domain (PRUNE2), transmembrane serine protease-2 (TMPRSS2), and acid phosphatase-3 (ACP3; Table 1; Supplementary Fig. S1B). In addition, GEPs between ARHigh/mCSPC-EA (LNCaP, VCAP, 22RV1) and ARHigh/mCSPC-AA (MDA-Pca-2b, RC77T, RC165T, and RC43T) showed discrete DEGs (Fig. 1AII). Top EMT markers VIM, HAS3, CD55 were downregulated in all AA cell lines. Furthermore, ANXA2, ANXA2P2F, S100A6 were downregulated in MDA-Pca-2b and upregulated in RC77T, RC165T and RC43T. Additional EMT markers AHNAK2, TGFB1, TGFBR2, UCHL1, ANXA3, and LOXL2 were also downregulated in MDA-PCa-2b and upregulated in other AA cell lines (RC77T, RC165T, and RC43T; Fig. 1AIII). Fascin actin-bundling protein-1/FSCN1, a taxane-resistant gene, was identified as a top DEG among all prostate cancer subtypes (Table 1).

IPA based on the DEGs revealed activation of cell mobility, morbidity, mortality, migration, invasion, survival, viability, development of vasculature-vasculogenesis, and angiogenesis were the key pathways (z score 8.175 to −7.005) associated with the development of more aggressive forms (ARLow/mCRPC/NEPC) of prostate cancer subtypes (Supplementary Fig. S2A). Furthermore, oxidative phosphorylation, p70S6K, unfolded response, TREM1, NFκB, ERK5, IL8, BAG2, Integrin, and VEGF signaling were identified (z score 5.574 to −4.811) as top canonical pathways for ARLow/mCRPC/NEPC prostate cancer (Supplementary Fig. S2B). IPA also predicted HIF1α (P < 9.43E-06) and EMT (P < 5.63E-06) as key pathways associated with prostate cancer development in AA (Supplementary Fig. S2C).

GEP Identified Unique DEGs for Taxane-resistant ARLow/mCRPC/NEPC Lines (DUTXR and PC3-TXR)

RNA-seq identified DEGs between ARLow/mCRPC/NEPC (PC-3, DU145) and its taxane-resistant version ARLow/mCRPC/NEPCtaxane-resistant (PC-3TXR and DUTXR). GEP data from baseline DUTXR and PC3-TXR indicated 3,289 and 2,215 genes were uniquely expressed in PC-3TXR and DUTXR, respectively. Furthermore, GEP also identified 2,968 commonly expressed genes in both sensitive and TXR cell lines (Supplementary Fig. S3; Supplementary Table S6A and S6B). The top common genes C1orf116, ICAM1, KRT7, SLC43A3, TGM2, TUBB2B between DUTXR versus DU145 and PC-3TXR versus PC-3 were determined using RNA-seq analysis.

scRNA-seq Identified EMT Markers in ARLow and Acquired Taxane-resistant Prostate Cancer Subtypes

scRNA-seq analysis identified higher expression of CSC markers, EMT markers (HAS3 and its receptor CD44), and “stem-like” subclonal cell populations in ARLow/mCRPC/NEPC (PC3, PC3M, and DU145) compared with ARHigh/mCSPC (22RV1, LnCaP) prostate cancer cell lines (Fig. 1BII). A higher expression of additional EMT markers (AHNAK2, LOXL2, S100A6, TGFB, TGFBR2, UCHL1, VIM, ANXA2, ANXA2P2, ANXA3, CD55, CD109) in ARLow/mCRPC/NEPC (PC3, PC3M and DU145) compared with ARHigh/mCSPC (22RV1, LnCaP) prostate cancer cell lines was also noted (Fig. 1BIII and IV). There was expression variation of each EMT marker among all ARLow/mCRPC/NEPC prostate cancer subtypes, for example, CD44 expression was greater in PC3 and PC3M compared with DU145 (Fig. 1BII–IV). CD55 expression was also higher in PC-3, whereas CD109 was greater in PC-3M (Fig. 1BII–IV). A comparison of other major EMT markers between PC3 and PC-3M revealed EZH2, Snail/SNAI1, Slug/SNAI2, and TWIST1, were expressed highly in PC-3M compared with PC-3, yet both are ARLow/mCRPC/NEPC prostate cancer subtypes (Fig. 5A). Minute levels of EMT markers were also expressed in ARHigh/mCSPC cell lines (22RV1 and LNCaP; Fig. 1BII–IV). Notably, FSCN1 associated with taxane-resistant expressed highly in single-cell subclonal populations in ARLow/mCRPC/NEPC prostate cancer (PC3, PC3M, and DU145; Fig. 1BV). In addition, the intratumor heterogeneity within various cell lines is revealed by single-cell clusters (representing subclonal populations) determined by t-SNE/UMAP analysis based on the expression of top biomarkers for each subcluster. The relative representation of each subclone between the prostate cancer cell lines was also provided using comparative pie charts. Our results showed several clusters which are unique to each prostate cancer cell line, in addition to the shared subclusters possibly representing gene signatures common to the biology of prostate cancer (Fig. IBVI).

Next, we compared the EMT markers between taxane-sensitive (DU145) and the clonally derived acquired taxane-resistant mCRPC cell line DUTXR. Some common overlap in GE was observed between the DUTXR and DU145 subclones, as expected (Fig.1CI). However, the subclonal analysis showed a greater percentage of cells not expressing EMT markers in taxane-sensitive DU145 cells compared with the taxane-resistant DUTXR cell line (Fig. 1CII–IV). This finding demonstrated that a greater percentage of DUTXR cells exhibited top EMT markers compared with DU145. The expression of VIM, LOXL2, TGFB, TGFBR2, UCHL1, ANXA2P2, CD55, and CD109 was greater in taxane-resistant DUTXR subclonal populations compared with taxane-sensitive DU145 (Fig. 1CII–IV). While the expression of other prostate cancer EMT markers HAS-3, AHNAK2, TGFBR2, ANXA2, and ANXA3 was higher in DU145 compared with DUTXR (Fig. 1CII–IV). Overall, the expression of EMT markers in subclonal cell populations was greatest in DUTXR.

Top genes representing single-cell clusters that were shared and which were unique between DU145 and DUTXR were further investigated using IPA pathway analysis and upstream prediction. The following clusters were unique to DU145: Cluster 1, Cluster 3, Cluster 5, and Cluster 7. Cluster 8 was shared between both the lines (Fig. 1CV). On the other hand, the following clusters were enriched in DUTXR: Cluster 2 (IPA predicted activation of PPARA), Cluster 4 (ERK/MAPK signaling, MSP-RON Signaling, upregulation of FOXM1, ESR1, CEBPB, XBP1, and NPM1), Cluster 6 (ATM signaling and MYC; Fig. 1CV; Supplementary Fig. S3BI and S3BII).

RNA-seq versus scRNA-seq Analysis Identified Common and Unique DEGs Among all Prostate Cancer Subtypes

The analysis of RNA-seq and scRNA-seq data revealed 70 genes common to all prostate cancer subtypes, while scRNA-seq and RNA-seq analysis identified 582 and 73 unique DEGs, respectively (Supplementary Fig. S4; Supplementary Table S7). Among the common genes were EMT markers, for example, VIM, AHNAK2, ANXA1, ANXA2, CD109, CD44, CD59, GPRC5A, HAS, S100A6, TGFBR2. Moreover, the taxane-resistant gene FSCN1 was also found to be a common gene in both cohorts of prostate cancer cell line models (Supplementary Table S7).

The IPA based on the common DEGs between RNA-seq and scRNA-seq predicted cell invasion, movement, neoplasia, migration, transformation, metastatic, and growth as key pathway-associated prostate cancer aggressiveness (Supplementary Fig. S5A). Furthermore, causal network pathway for common DEGs include LY2109761 (TGFβ receptor inhibitor), IGF2R (receptor for both insulin-like growth factor 2), IL11RA (cytokine), SEMG2 (semenogelin proteins), CCR2 (chemokine receptor type 2), TRAF3IP3 (mediates cell growth by modulating the c-Jun N-terminal kinase signal transduction pathway), GGTI-2154 (inhibitor of geranylgeranyltransferase I), L778123 (inhibitor of FPTase and GGPTase-I), TGFB1, and CXCL5 (Supplementary Fig. S5B). Upstream regulators for common DEGs were also identified, THBF1, RAF1, AGT, ER receptor, OSM, SLC15A4, JUN, and IL1B (Supplementary Fig. S5C). The S100 family, hepatic fibrosis, Rho family, RHOGDI, and integrin singling pathway signaling were identified as top canonical pathways for prostate cancer (Supplementary Fig. S5D).

In Silico Analysis Using Prostate Cancer Patient Transcriptomes Validated the Clinical Relevance of Top EMT Markers

An in silico analysis using TCGA dataset showed that the top gene, FSCN1, was significantly associated with disease-free survival (DFS), Kaplan–Meier curves (Fig. 2AI), with HR 2.2 (P = 0.00051). Low expression of FSCN1 was associated with longer DFS. In contrast, high expression of FSCN1 was associated with poor DFS. Another top EMT marker, TGFB1, showed similar trends. Low expression was associated with longer DFS, whereas high expression was associated with poor DFS (P = 0.00019, HR = 2.3; Fig. 2AII).

Flow Cytometry Analysis Identified “Stem-like” (CD44+, CD133+, and CD44+/CD133+) Cells in All (EA and AA) Prostate Cancer Subtypes

CD44 is a well-known marker for “stem-like” cells in prostate cancer (35). The percentage of CD44+ cells was greater in ARLow/mCRPC/NEPC compared with ARHigh/mCSPC. The percentage of CD44+ cells was 76.5% in PC-3, 90.3% in PC-3M, and 69.0% in DU145, whereas they were 5.60% in LNCaP, 6.59% in VCaP, 4.51% in 22RV1, and 8.50% in C4-2B cell lines (Fig. 2B; Table 2). Interestingly, taxane-resistant DUTXR showed the greatest percentage of “stem-like” cells, 96.4% CD44+. Furthermore, CD133 was identified as another marker (nonabundant) for “stem-like” cells in prostate cancer (35). Therefore, we measured CD133 and CD44/133 in all prostate cancer subtypes; the results showed a similar trend (Fig. 2B; Table 2). In the AA prostate cancer subtype (MDA-PCa-2b and RC77T/E), “stem-like” CD44+ cell proportions were 16.6% and 5.69%, respectively (Fig. 2B; Table 2).

Immunoblotting Showed Higher EMT Protein Expression in ARLow/mCRPC/NEPC and Taxane-resistant ARLow/mCRPC/NEPC Compared with ARHigh/mCSPC Prostate Cancer Subtypes

Our RNA-seq DEGs, scRNA-seq, in silico, and in vitro flow cytometry results indicated that aggressive ARLow/mCRPC/NEPC and taxane-resistant ARLow/mCRPC/NEPC prostate cancer have greater expression of EMT markers and “stem-like” cell population. On the basis of these findings, we determined the protein expression of the top markers associated with EMT (CD44, ALDH1, Oct-4, TGFB, Nanog, and Sox2) using immunoblotting in all prostate cancer cell lines. These data indicated that the protein expression of these top EMT markers was concurrently higher in ARLow/mCRPC/NEPC (PC-3, PC-3M, DU145) compared with ARHigh/mCSPC (LNCaP, VCaP, 22RV1; Fig. 2C). Furthermore, consistent upregulation of these top EMT proteins was observed in taxane-resistant DUTXR and PC-3TXR compared with ARLow/mCRPC/NEPC (PC-3, PC-3M, DU145) and ARHigh/mCSPC (LNCaP, VCaP, 22RV1) cells (Fig. 2C). EMT proteins were also expressed in the AA (MDA-PCa-2b and RC77T/E) cell line. However, their expression level was greater compared with ARHigh/mCSPC (LNCaP, VCaP, 22RV1), and lower than ARLow/mCRPC/NEPC prostate cancer (PC-3, PC-3M, DU145) and taxane-resistant ARLow/mCRPC/NEPC (DUTXR and PC-3TXR; Fig. 2C) prostate cancer cells. Densitometry plots showed significant differential protein expression among all prostate cancer subtypes (P values, P ≤ 0.05; Fig. 2C).

Metronomic Topotecan in Combination with DTX Reduced Cell Growth and Cell Density in ARLow/mCRPC/NEPC and Acquired Taxane-resistant ARLow/mCRPC/NEPC Prostate Cancer Subtypes

Previously we reported the cytotoxicity of CONV-TOPO, CONV-DTX, and METRO-TOPO for all cell lines (22, 29). In this study, the effect of CONV-TOPO, CONV-DTX, METRO-TOPO, and CONV-DTX+METRO-TOPO administration on ARLow/mCRPC/NEPC (PC-3, PC-3M, DU145) cell lines, were assessed by MTT assay at their estimated IC50 and IC50/2 drug concentrations (describe in Supplementary Table S5) for each drug at 72 hours. In vitro cytotoxicity showed decreases in cell survival or growth after 72 hours of drug treatment in all prostate cancer cell lines. In all cell lines, METRO-TOPO (lower dose) was more potent at reducing cell growth compared with CONV (high dose) treatments at IC50 and IC50/2 (Fig. 3A; Supplementary Fig. S6A). Furthermore, the greatest reduction in cell growth was observed following CONV-DTX + METRO-TOPO combination treatment in all prostate cancer cell lines tested (Fig. 3A; Supplementary Fig. S6A). The effect of CONV-TOPO, CONV-DTX, METRO-TOPO, and CONV-DTX+METRO-TOPO administration on taxane-resistant ARLow/mCRPC/NEPC (DURXR) prostate cancer subtypes. Results showed similar trends, greater potency was observed in METRO-TOPO and combination treatments.

The effect of METRO-TOPO on cell density and cellular morphology was determined. In agreement with MTT assays, micrographs of prostate cancer cells exposed to all dosing regimens showed decreases in cell density compared with control cells (Fig. 3C; Supplementary Fig. S6C). Treatment with METRO-TOPO resulted in a greater reduction of cell density compared with CONV-TOPO and CONV-DTX (Fig. 3C; Supplementary Fig. S6C). Importantly, we observed greater cell death in combination treatment, CONV-DTX+METRO-TOPO, compared with single drug treatments with CONV-TOPO, CONV-DTX, or METRO-TOPO (Fig. 3C; Supplementary Fig. S6C).

Metronomic Topotecan in Combination with DTX Induced Apoptosis in ARLow/mCRPC/NEPC and Acquired Taxane-resistant ARLow/mCRPC/NEPC Prostate Cancer Subtypes

The effect of CONV-TOPO, CONV-DTX, METRO-TOPO, and CONV-DTX+METRO-TOPO administration on apoptosis was determined by assessing caspase 3/7 (a marker of apoptosis) activity. The level of apoptosis was estimated at the IC50 and IC50/2 (dose described in Supplementary Table S5) in all cell lines following 72 hours exposure in ARLow/mCRPC/NEPC (PC-3, PC-3M, and DU145) and acquired taxane-resistant ARLow/mCRPC/NEPC (DUTXR). A significant increase in caspase 3/7 activity following each treatment compared with controls (Fig. 3BI–IV). A greater increase in apoptosis was observed post-METRO-TOPO treatment (at 72 hours IC50 and IC50/2 dose for metronomic topotecan) compared with CONV-TOPO and DTX (at 72 hours IC50 and IC50/2 dose for conventional topotecan and DTX). The greatest treatment-induced apoptosis was observed in combination with CONV-DTX+METRO-TOPO (at 72 hours IC50 and IC50/2 dose for conventional DTX and IC50/2 dose for metronomic topotecan) in all cell lines (Fig. 3BI–IV; Supplementary Fig. S6B). The relative increase (fold change) in caspase 3/7 activity following each treatment reinforced our findings (Supplementary Table S1).

scRNA-seq and RNA-seq Analysis Indicated Downregulation of EMT Markers in ARLow/mCRPC/NEPC Prostate Cancer Tumor Models Following Metronomic Topotecan Treatment

scRNA-seq analysis identified differential subclonal populations (clusters) representing control (no drug treatment), 6-week EE-TOPO (extended exposure of topotecan), and 6-week CONV-TOPO treatment (Fig. 4A). Further scRNA-seq analysis showed long-term EE-TOPO exposure downregulated expression of EMT markers and “stem-like” subclonal populations in the prostate cancer tumor model (Fig. 4B). We identified nine subclonal populations (clusters) based on scRNA-seq in control, whereas followed by EE-TOPO treatment, three (clusters two, four, and nine) subclonal populations (clusters) disappeared. These clusters include EMT markers (S100A9, ESRP1, ASPM, EPCAM, CLDN7, CDH1, INPP4B, CD70, FN1, CDH11) and drug metabolism or resistance markers (SERPINE1, ESM1, TOP2A; Fig. 4B). Further analysis showed top EMT marker HAS3 expression was downregulated by EE-TOPO treatment (54% to 47%) compared with CONV-TOPO treatment (upregulated 54% to 59%; Fig 4C). Another EMT marker, TGFB1, showed similar trends (Supplementary Fig. S7). Taxane-resistance marker FSCN1 expression increased less following EE-TOPO (69%–71%) compared with CONV-TOPO (69%–84%) treatment (Fig. 4B). RNA-seq analysis for control (no drug treatment), 3-days, and 6-week EE-TOPO versus CONV-TOPO treatment showed more reduction (downregulation) of top EMT markers, such as CD55 and HAS3 by EE-TOPO compared with CONV-TOPO treatment (Fig. 4D).

Metronomic Topotecan Treatment Reduced “Stem-like” Cell Load in ARLow/mCRPC/NEPC Prostate Cancer Subtypes

ARLow/mCRPC/NEPC (PC-3, PC-3M, DU145) were treated with CONV-TOPO, METRO-TOPO, and combination CONV-DTX+METRO-TOPO treatments. Pretreated and posttreated prostate cancer cell lines were stained with stemness markers CD44, CD133, and both CD44/133 antibodies. All cell lines showed a greater reduction in the CD44high (“stem-like”) cell population following METRO-TOPO treatment compared with CONV-TOPO treatment. Furthermore, combination treatment reduced the highest percent of the CD44high cell population (Fig 5B; Supplementary Fig. S8A and S8B; Table 2). Next, we sorted CD44+ cells from PC-3M and treated them with CONV-TOPO, CONV-DTX, METRO-TOPO, and combination with METRO-TOPO+CONV-DTX. Cell-sorted PC-3M-CD44+ cell populations showed the highest cell survival reduction following combination treatment compared with other treatments (Supplementary Fig. S9A). The caspase 3/7 assay also showed that combination treatment induced the greatest apoptosis (6.22-fold) compared with other treatments (Supplementary Fig. S9B; Supplementary Table S4).

Flow cytometry analysis identified ALDH in PC-3 and PC-3M, indicating the presence of a “stem-like phenotype” (Fig. 5C). We also showed ALDH expression was greater in PC-3M compared with PC-3. To evaluate the effect of treatment on stemness-related or EMT protein expression, we performed posttreatment immunoblotting for all treatments in the PC-3M cell line. EMT proteins (CD44, ALDH1, Oct-4, TGFB, Nanog, and Sox2) were significantly (P ≤ 0.05) downregulated following METRO-TOPO compared with CONV-TOPO treatment in PC-3M. Combination treatment exhibited the greatest downregulation (17%–56%) of all EMT proteins compared to other treatments (Fig. 5D; Supplementary Table S2).

A Microfluidic Screen Showed Metronomic Topotecan Treatment is Potentially Effective Against Cell Invasion in ARLow/mCRPC/NEPC Prostate Cancer Subtypes

This experiment allowed us to study the effect of drug and dosing schedule as a single agent and in combination on tumor cell motility through microchannels of dimensions that mimic the size of confining pores and channel-like tracks encountered by migrating cells in vivo (36). Figure 5EI–II (and Supplementary Videos) showed that cell entry into confining (W × H = 3 × 10 μm2) microchannels was greater in PC-3M compared with PC-3, suggesting that PC3-M cells are more invasive into mechanically challenging microenvironments. For confined (W × H = 10 × 10 μm2) microchannels, cell entry was greater for PC-3M compared with PC-3 but the difference was not significant.

The effect of drug treatment on PC-3M migration was accessed. PC3-M entry into confining microchannels was markedly suppressed upon TOPO-METRO single-agent and TOPO-METRO+CONV-DTX combination treatment. Of note, combination therapy resulted in a higher reduction in the invasiveness of PC-3M compared with other treatments (Fig. 5EIII).

Metronomic Topotecan in Combination with DTX Reduced Decreased “Stem-like” Cell Load in Acquired ARLow/mCRPC/NEPCtaxane-resistant Prostate Cancer Subtypes

The DUTXR cell line was treated with CONV-TOPO, METRO-TOPO, and combination CONV-DTX+METRO-TOPO. Pretreated and posttreated cells were stained with stemness marker antibodies CD44 and CD133 individually, and together (as described above), and analyzed by flow cytometry. These results showed a greater reduction in CD44high (“stem-like”) cell populations by METRO-TOPO (81.0%) compared with CONV-TOPO treatments (71.5%). Furthermore, combination treatment showed the greatest reduction (63.4%) of the “stem-like” CD44high cell populations (Fig. 6A; Table 2).

Immunoblotting results demonstrated that EMT proteins were also downregulated significantly in METRO-TOPO compared with CONV-TOPO treatment in ARLow/mCSPC/NEPCtaxane-resistant (DUTXR) cell lines. Combination treatment (CONV-DTX+METRO-TOPO) exhibited the greatest downregulation (26%–61%) of EMT proteins compared with other treatments (Fig. 6C; Supplementary Table S3).

Finally, cell-sorted DUTXR-CD44+ populations were treated with CONV-TOPO, METRO-TOPO, CONV-DTX, and combination (CONV-DTX+METRO-TOPO) treatments. A marked suppression in cell survival was observed in the combination treatment compared with other groups (Fig. 6DII). In addition, the caspase 3/7 assay showed that combination treatment induces the greatest level of apoptosis (3.28-fold) relative to other treatments (Fig. 6CII; Supplementary Table S4), where CD44 cells showed nonsignificant differences for all treatments (Fig. 6CIII–IV; Supplementary Table S4).

Microscopic images revealed CONV-TOPO, METRO-TOPO, CONV-DTX, and combination (CONV-DTX+METRO-TOPO) treatment effect on CD44+ sorted cell populations. Results showed significantly higher cell death in METRO-TOPO (6%) and combination treatment (3.2%) compared with CONV-TOPO (54%) and CONV-DTX (31%) treatments (Fig. 6E).

Metronomic Topotecan in Combination with DTX Reduced Proliferative Capacity in Acquired ARLow/mCRPC/NEPCtaxane-resistant Prostate Cancer Subtypes

Next, we evaluated the potential effect of CONV-TOPO, CONV-DTX, METRO-TOPO, and combination (METRO-TOPO+CONV-DTX) treatment on the proliferative capacity of the DUTXR cells using the colony-forming assay. TOPO-METRO treatment alone significantly reduced colony number (16.4%) as well as colony size when compared with control or CONV-TOPO (67.0%) or CONV-DTX (33.5%) treatment. Furthermore, colony-forming assay results also showed the combination of METRO-TOPO+CONV-DTX treatments further reduced the colony numbers (9.53%) and size (Fig. 6B).

Drug development for aggressive, lethal treatment-resistant prostate cancer poses a significant challenge with few therapeutic successes (4, 12). We used scRNA-seq and bulk RNA-seq as an approach to demonstrate that EMT and cancer “stemness” signatures are key pathways to developing metastatically aggressive prostate cancer, including castration-resistant and taxane-resistant tumors in EA and AA. As shown in Fig. 1BVI and 1CV, the intratumor heterogeneity within various cell lines was revealed by single-cell clusters (representing subclonal populations) determined by t-SNE or UMAP analysis based on the expression of top biomarkers for each subcluster. Our results showed several clusters which are unique to each prostate cancer cell line, in addition to the shared subclusters possibly representing gene signatures common to the biology of prostate cancer. Furthermore, using a comparative analysis of RNA-seq and scRNA-seq from prostate cancers, we identified overexpression of VIM, AHNAK2, ANXA1, ANXA2, CD109, CD44, CD59, GPRC5A, HAS, S100A6, TGFBR2 in both cohorts. Importantly, the taxane-resistant gene FSCN1 was also identified as a common gene in both cohorts of prostate cancer cell lines.

We identified “stem-like” cell populations (CD44+ and CD44+/CD133+) in aggressive metastatic prostate cancer subtypes, which was confirmed by the RNA-seq and subclonal overexpression of several EMT markers (VIM, HAS3, S100A6, ANXA2P2, ANXA2, ANXA3, AHNAK2, LOXL2, TGFB, TGFBR2, UCHL1, CD44, CD55, and CD109). We also identified top DEGs and subclonal populations between the taxane-sensitive and taxane-resistant prostate cancer subtypes. These data provide greater insights into the acquisition of drug resistance and possibly novel treatment targets.

Earlier, we demonstrated that METRO-TOPO single-agent treatment is effective against prostate cancer animal xenograft model and cell lines (22). Here, we have extended our previous findings and have shown that METRO-TOPO is highly synergistic in combination with CONV-DTX. We performed pretreatment versus posttreatment scRNA-seq and RNA-seq analysis, which revealed that METRO-TOPO treatment abrogates stem-like cell types (representing NEPC and EMT phenotypes) in lethal prostate cancer. METRO-TOPO also reduced acquired taxane resistance by downregulating EMT gene and protein and by reducing the “stem-like” CD44+ cell population. Furthermore, in silico validation with TCGA prostate cancer patient cohort databases established the relevance of top DEGs in patient survival. Using comparative analysis with whole-genome transcriptomics data from patients with prostate cancer, we concluded that METRO-TOPO has the potential to be clinically effective based on the reverse matching of DEGs.

Importantly, we identified HAS3 as one of the top EMT markers in RNA-seq and scRNA-seq for aggressive prostate cancer that is downregulated by TOPO-METRO treatment. Hyaluronan (HA) is an important constituent of the stem cell niche (34, 35). CD44 is the major HA receptor and EMT marker in prostate cancer. High expression of HAS3 (4.8-fold upregulation) resulted in the secretion of large amounts of HA bound to CD44+ in the “stem-like” cell population and plays a critical role in the development CSCs by regulating cell adhesion, migration, proliferation, differentiation, cancer metastasis, and multidrug-resistant (37, 38). In our study, HA was upregulated 4.8-fold in lethal prostate cancer subtypes. We also identified high levels of expression (4.07-fold upregulation) of EMP1 in our aggressive prostate cancer subtypes which regulates the expression of CD44 to promote stemness (39). Furthermore, CD44 was upregulated upon TGFB1-induced EMT (40), and our study reported overexpression of TGFB1 in aggressive prostate cancer (3.41-fold in GE and ∼2-fold protein). Furthermore, we identified a higher percentage of TGFB1 overexpressed subclonal population (scRNA-seq) in aggressive taxane-resistant DUTXR compared with taxane-sensitive DU145 prostate cancer subtypes. We reported that METRO-TOPO (6-week EE) treatment reduces TGFB1 expression in subclonal populations. On the basis of evaluation of TCGA database, the downregulation of TGFB1 is beneficial and associated with increased patient survival.

Previous studies have reported that CD44 expression is greater in prostate cancer, and involved in cancer cell proliferation, invasion, migration, and drug resistance (41). Some studies also suggested that CD44 plays a vital role in cancer stemness, specifically in prostate cancer (42, 43). We observed that CD44+ cells (96%) frequency and protein expression were greatest in taxane-resistant mCRPC cell lines. Furthermore, we identified a higher percentage of CD44-overexpressed subclonal population (scRNA-seq) in aggressive taxane-resistant DUTXR compared with taxane-sensitive DU145 prostate cancer subtypes. The ability to reduce stemness offers a potential strategy to treat aggressive and resistant prostate cancer. In our study, METRO-TOPO and combination (METRO-TOPO+CONV-DTX) treatment downregulated CD44+ (71.7% and 63.4%) cells along with protein expression (55% and 29%). We also observed high percentages of CD44+ cell populations in mCRPC cell lines and METRO-TOPO following single agent and in combination therapy with DTX reduced the amount of CD44+ cells significantly after treatment. Furthermore, METRO-TOPO treatment significantly decreased cell invasion and colony formation in aggressive and taxane-resistant forms of prostate cancer. Further studies employing CRISPR knockdown of HAS3 (and other top genes) are needed to further explore the significance of these novel mechanisms and their impact on cancer stemness and its association with METRO-TOPO treatment-related efficacy.

Here, we have identified VIM as a top common EMT marker from RNA-seq and scRNA-seq for aggressive prostate cancer subtypes. Earlier studies also identified its role in invasion and metastasis in prostate cancer (16, 44–46). We also showed that METRO-TOPO (6-week EE) treatment reduces VIM expression in subclonal populations.

Furthermore, SNAI1 plays a critical role in the aggressiveness of prostate cancer by increasing the expression of CD44 and vimentin (47). In our study, we detected greater expression of these markers in the subclonal population (scRNA-seq analysis) and RNA-seq GE (5.42-fold upregulation) of mCRPC/NEPC. SNAl is associated with cell proliferation and cell invasion in prostate cancer (41). We observed high expression (scRNA-seq) of SNAl1 and SNAl2 in mCRPC (PC-3 and PC-3M) prostate cancer subtypes. Our microfluid-based migration assay showed high cell invasion for these prostate cancer cell lines. Furthermore, METRO-TOPO as a single agent and in combination with DTX significantly reduced cell invasion in the mCRPC cell line model.

Furthermore, CD55 also promotes prostate cancer cell survival and metastatic lesion formation (48, 49). Recent studies have identified CD109 as a new marker for invasive breast and prostate carcinoma (50). Our scRNA-seq and RNA-seq analysis agree with this finding. We observed greater expression of CD55 and CD109 (4.46- and 3.43-fold upregulation, respectively) in aggressive mCRPC/NEPC and taxane-resistant mCRPC/NEPC subtypes. We also identified a higher percentage of subclonal cell populations in aggressive taxane resistance prostate cancer subtypes overexpressing CD55 and CD109.

Earlier studies reported that ANXA3, ANXA2, and its pseudogene ANXA2P2 are overexpressed in various cancers, including prostate cancer, and facilitated transcription of the stemness genes Nanog, Sox2, and Oct4 (51–53). AHNAK2 upregulation is correlated significantly with advanced grades of various cancers and was associated with EMT (54). Our scRNA-seq analysis detected greater expression of ANXA3, ANXA2, ANXA2P2, and AHNAK2 in aggressive prostate cancer. Our RNA-seq data also corroborated with scRNA-seq finding (4.02-, 4.51-, 4.58-, and 4.26-fold upregulated, respectively) in aggressive mCRPC/NEPC and taxane-resistant mCRPC/NEPC subtypes. We also identified greater protein expression levels of Nanog, Sox2, and Oct4 genes in mCRPC/NEPC and taxane-resistant mCRPC/NEPC prostate cancer subtypes. METRO-TOPO downregulated Oct-4 (64%), Sox2 (60%), and Nanog (39%) in mCRPC/NEPC (PC-3M) prostate cancer cell line model. METRO-TOPO also downregulated Oct-4 (41%), Sox2 (56%), and Nanog (49%) in taxane-resistant mCRPC (DUTXR) prostate cancer cell line model. The combination of METRO-TOPO and DTX exhibited similar results. A downregulation of Oct-4 (57%), Sox2 (47%) and Nanog (17%) in mCRPC/NEPC (PC-3M), and downregulation of Oct-4 (28%), Sox2 (61%), and Nanog (35%) in taxane-resistant mCRPC/NEPC (DUTXR) prostate cancer cell line model was observed after combination treatment.

Next, ALDH1 is another stemness marker in prostate cancer and overexpressed in mCRPC (55). Our studies also showed greater expression of ALDH1 in mCRPC/NEPC and taxane-resistant mCRPC/NEPC prostate cancer subtypes. Similarly, TOPO-METRO as a single agent and combination TOPO-METRO+CONV-DTX treatment downregulated ALDH1 protein (70%, 46%, respectively) in mCRPC/NEPC (PC-3M) cell line and (50%, 26%, respectively) in taxane-resistant mCRPC/NEPC (DUTXR) aggressive prostate cancer subtypes.

Interestingly, RNA-seq and flowcytometry data revealed that EMT markers are enriched in EA compared with AA prostate cancer cell lines. An earlier study showed KRT80, RUNX, GATA6, SERPINB13, and SLC6a14 are important markers for prostate cancer progression (27), whereas we also identified KRT80, RUNX as well as GATA2, GATA2-AS1 in our study. We also found KRT20 is upregulated in EA compared with AA cell lines, although in future, we need further explore its significance in prostate cancer progression.

Recent studies identified FSCN1 as a taxane-resistant marker in several solid tumors, including prostate cancer (56). We identified FSCN1 as a key upregulated gene (4.71-fold) for taxane resistance in aggressive prostate cancer. Our results also showed that the treatment of METRO-TOPO (6-weeks EE) downregulates FSCN1 expression in subclonal populations in aggressive mCRPC/NEPC 3D tumor models, whereas CONV-TOPO resulted in increased expression (P < 0.05). Furthermore, TCGA patient cohort also revealed lower expression of FSCN1 associated with better DFS (P = 0.00051).

We also identified C1orf116, TUBB2B are common genes in taxane-resistant prostate cancer subtypes. C1orf116 is a novel EMT biomarker for prostate cancer, and our study showed 17-fold and 24-fold downregulation of this gene in taxane-resistant prostate cancer subtypes DUTXR and PC-3TXR, respectively. Contrastingly, TUBB2B was upregulated in both taxane-resistant prostate cancer cell lines, approximately 12-fold in DUTXR and approximately 43-fold in PC-3TXR. An earlier study revealed downregulation of C1orf116 is associated with poor prognosis in patients with lung and prostate cancer (57). TUBB2B is the major constituent of microtubules, the primary target for taxane-based drugs. TUBB2B has also been associated with DTX resistance in prostate cancer (58, 59).

Cell invasion is one of the major characteristics of EMT transition and cancer progression (60). Our microfluidic-based invasion assay supported the reduction in cell invasion potential, followed by METRO-TOPO treatments. Earlier, we demonstrated the increased potency of METRO-TOPO in animal models. However, in this study, our approach focused on in vitro model systems to better identify the mechanistic underpinnings of metronomic administration (two-dimensional and 3D tumor models) prostate cancer. Further preclinical validation and single-cell multi-omics strategies using CRISPR-based knockout xenograft models and patient-derived organoids/xenografts will be necessary. Our approach in this study promotes the understanding of subclonal molecular pathways underlying the differential patterns of prostate cancer aggressiveness and drug response among various prostate cancer subtypes. Overall, our study identified novel mechanisms of action that can serve as a pipeline to advance METRO-TOPO as a potent clinical-trial-ready therapeutic option for the management of lethal prostate cancer with stem-like features. These mechanisms will be key in identifying patients with molecular signatures that are sensitive to treatment, and, most importantly, can identify biomarkers that can be used to monitor treatment effectiveness and disease progression.

C.C. Yates reports personal fees from Riptide Biosciences, Riptide Biosciences, QED Therapeutics, Regeneron, and PreludeDX outside the submitted work. No disclosures were reported by the other authors.

T. Mitra Ghosh: Conceptualization, resources, data curation, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. S. Mazumder: Data curation, software, formal analysis, validation, investigation, visualization, methodology. J. Davis: Investigation, visualization, methodology. J. Yadav: Investigation, visualization, methodology. A. Akinpelu: Investigation, visualization, methodology. A. Alnaim: Software, investigation, visualization, methodology. H. Kumar: Investigation, methodology. R. Waliagha: Validation, investigation, visualization, methodology. A.E. Church Bird: Software, visualization, methodology. S. Rais-Bahrami: Writing-original draft. R.C. Bird: Writing-original draft, writing-review and editing. P. Mistriotis: Software, investigation, visualization, methodology, writing-original draft, writing-review and editing. A. Mishra: Visualization, methodology. C.C. Yates: Resources, writing-original draft, writing-review and editing. A.K. Mitra: Conceptualization, resources, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing-original draft, writing-review and editing. R.D. Arnold: Conceptualization, resources, data curation, supervision, funding acquisition, writing-original draft, project administration, writing-review and editing.

We thankfully acknowledge the Center for Pharmacogenomics and Single-Cell Omics and Flow Cytometry and High-Speed Cell Sorting Laboratory, Auburn University, for support with analysis.

Note: Supplementary data for this article are available at Cancer Research Communications Online (https://aacrjournals.org/cancerrescommun/).

1.
Wen
S
,
Niu
Y
,
Lee
SO
,
Chang
C
.
Androgen receptor (AR) positive vs negative roles in prostate cancer cell deaths including apoptosis, anoikis, entosis, necrosis and autophagic cell death
.
Cancer Treat Rev
2014
;
40
:
31
40
.
2.
Magnan
S
,
Zarychanski
R
,
Pilote
L
,
Bernier
L
,
Shemilt
M
,
Vigneault
E
, et al
.
Intermittent vs continuous androgen deprivation therapy for prostate cancer: a systematic review and meta-analysis
.
JAMA Oncol
2015
;
1
:
1261
9
.
3.
Hamdy
FC
,
Donovan
JL
,
Lane
JA
,
Mason
M
,
Metcalfe
C
,
Holding
P
, et al
.
10-year outcomes after monitoring, surgery, or radiotherapy for localized prostate cancer
.
N Engl J Med
2016
;
375
:
1415
24
.
4.
Scher
HI
,
Solo
K
,
Valant
J
,
Todd
MB
,
Mehra
M
.
Prevalence of prostate cancer clinical states and mortality in the United States: estimates using a dynamic progression model
.
PLoS One
2015
;
10
:
e0139440
.
5.
Yadav
SS
,
Li
J
,
Stockert
JA
,
Herzog
B
,
O'Connor
J
,
Garzon-Manco
L
, et al
.
Induction of neuroendocrine differentiation in prostate cancer cells by dovitinib (TKI-258) and its therapeutic implications
.
Transl Oncol
2017
;
10
:
357
66
.
6.
Heidenreich
A
,
Bastian
PJ
,
Bellmunt
J
,
Bolla
M
,
Joniau
S
,
van der Kwast
T
, et al
.
EAU guidelines on prostate cancer. Part II: Treatment of advanced, relapsing, and castration-resistant prostate cancer
.
Eur Urol
2014
;
65
:
467
79
.
7.
Cornford
P
,
Bellmunt
J
,
Bolla
M
,
Briers
E
,
De Santis
M
,
Gross
T
, et al
.
EAU-ESTRO-SIOG guidelines on prostate cancer. Part II: treatment of relapsing, metastatic, and castration-resistant prostate cancer
.
Eur Urol
2017
;
71
:
630
42
.
8.
Hall
SJ
,
Klotz
L
,
Pantuck
AJ
,
George
DJ
,
Whitmore
JB
,
Frohlich
MW
, et al
.
Integrated safety data from 4 randomized, double-blind, controlled trials of autologous cellular immunotherapy with sipuleucel-T in patients with prostate cancer
.
J Urol
2011
;
186
:
877
81
.
9.
De Bono
JS
,
Oudard
S
,
Ozguroglu
M
,
Hansen
S
,
MacHiels
J-PP
,
Kocak
I
, et al
.
Prednisone plus cabazitaxel or mitoxantrone for metastatic castration-resistant prostate cancer progressing after docetaxel treatment: a randomised open-label trial
.
Lancet
2010
;
376
:
1147
54
.
10.
Galletti
G
,
Leach
BI
,
Lam
L
,
Tagawa
ST
.
Mechanisms of resistance to systemic therapy in metastatic castration-resistant prostate cancer
.
Cancer Treat Rev
2017
;
57
:
16
27
.
11.
Annala
M
,
Vandekerkhove
G
,
Khalaf
D
,
Taavitsainen
S
,
Beja
K
,
Warner
EW
, et al
.
Circulating tumor DNA genomics correlate with resistance to abiraterone and enzalutamide in prostate cancer
.
Cancer Discov
2018
;
8
:
444
57
.
12.
Wang
Y
,
Chen
J
,
Wu
Z
,
Ding
W
,
Gao
S
,
Gao
Y
, et al
.
Mechanisms of enzalutamide resistance in castration-resistant prostate cancer and therapeutic strategies to overcome it
.
Br J Pharmacol
2021
;
178
:
239
61
.
13.
Li
P
,
Yang
R
,
Gao
W-Q
.
Contributions of epithelial-mesenchymal transition and cancer stem cells to the development of castration resistance of prostate cancer
.
Mol Cancer
2014
;
13
:
55
.
14.
Contreras
HR
,
López-Moncada
F
,
Castellón
EA
.
Cancer stem cell and mesenchymal cell cooperative actions in metastasis progression and hormone resistance in prostate cancer: potential role of androgen and gonadotropin‑releasing hormone receptors (Review)
.
Int J Oncol
2020
;
56
:
1075
82
.
15.
Zhou
J
,
Wang
H
,
Cannon
V
,
Wolcott
KM
,
Song
H
,
Yates
C
.
Side population rather than CD133(+) cells distinguishes enriched tumorigenicity in hTERT-immortalized primary prostate cancer cells
.
Mol Cancer
2011
;
10
:
112
.
16.
Mazumder
S
,
Mitra Ghosh
T
,
Mukherjee
UK
,
Chakravarti
S
,
Amiri
F
,
Waliagha
RS
, et al
.
Integrating pharmacogenomics data-driven computational drug prediction with single-cell RNAseq to demonstrate the efficacy of a NAMPT inhibitor against aggressive, taxane-resistant, and stem-like cells in lethal prostate cancer
.
Cancers
2022
;
14
:
6009
.
17.
Hanahan
D
,
Weinberg
RA
.
Hallmarks of cancer: the next generation
.
Cell
2011
;
144
:
646
74
.
18.
Kerbel
RS
,
Shaked
Y
.
The potential clinical promise of “multimodality” metronomic chemotherapy revealed by preclinical studies of metastatic disease
.
Cancer Lett
2017
;
400
:
293
304
.
19.
Simsek
C
,
Esin
E
,
Yalcin
S
.
Metronomic chemotherapy: a systematic review of the literature and clinical experience
.
J Oncol
2019
;
2019
:
5483791
.
20.
Fontana
A
,
Galli
L
,
Fioravanti
A
,
Orlandi
P
,
Galli
C
,
Landi
L
, et al
.
Clinical and pharmacodynamic evaluation of metronomic cyclophosphamide, celecoxib, and dexamethasone in advanced hormone-refractory prostate cancer
.
Clin Cancer Res
2009
;
15
:
4954
62
.
21.
Nelius
T
,
Klatte
T
,
de Riese
W
,
Haynes
A
,
Filleur
S
.
Clinical outcome of patients with docetaxel-resistant hormone-refractory prostate cancer treated with second-line cyclophosphamide-based metronomic chemotherapy
.
Med Oncol
2010
;
27
:
363
7
.
22.
Mitra Ghosh
T
,
White
J
,
Davis
J
,
Mazumder
S
,
Kansom
T
,
Skarupa
E
, et al
.
Identification and characterization of key differentially expressed genes associated with metronomic dosing of topotecan in human prostate cancer
.
Front Pharmacol
2021
;
12
:
736951
.
23.
Aljuffali
IA
,
Mock
JN
,
Costyn
LJ
,
Nguyen
H
,
Nagy
T
,
Cummings
BS
, et al
.
Enhanced antitumor activity of low-dose continuous administration schedules of topotecan in prostate cancer
.
Cancer Biol Ther
2011
;
12
:
407
20
.
24.
Merritt
WM
,
Danes
CG
,
Shahzad
MMK
,
Lin
YG
,
Kamat
AA
,
Han
LY
, et al
.
Anti-angiogenic properties of metronomic topotecan in ovarian carcinoma
.
Cancer Biol Ther
2009
;
8
:
1596
603
.
25.
Wang
D
,
Margalit
O
,
DuBois
RN
.
Metronomic topotecan for colorectal cancer: a promising new option
.
Gut
2013
;
62
:
190
1
.
26.
Jedeszko
C
,
Paez-Ribes
M
,
Di Desidero
T
,
Man
S
,
Lee
CR
,
Xu
P
, et al
.
Postsurgical adjuvant or metastatic renal cell carcinoma therapy models reveal potent antitumor activity of metronomic oral topotecan with pazopanib
.
Sci Transl Med
2015
;
7
:
282ra50
.
27.
Mori
JO
,
White
J
,
Elhussin
I
,
Duduyemi
BM
,
Karanam
B
,
Yates
C
, et al
.
Molecular and pathological subtypes related to prostate cancer disparities and disease outcomes in African American and European American patients
.
Front Oncol
2022
;
12
:
928357
.
28.
Takeda
M
,
Mizokami
A
,
Mamiya
K
,
Li
YQ
,
Zhang
J
,
Keller
ET
, et al
.
The establishment of two paclitaxel-resistant prostate cancer cell lines and the mechanisms of paclitaxel resistance with two cell lines
.
Prostate
2007
;
67
:
955
67
.
29.
Mitra Ghosh
T
,
Kansom
T
,
Mazumder
S
,
Davis
J
,
Alnaim
AS
,
Jasper
SL
, et al
.
The andrographolide analogue 3A.1 synergizes with taxane derivatives in aggressive metastatic prostate cancers by upregulation of heat shock proteins and downregulation of MAT2A-mediated cell migration and invasion
.
J Pharmacol Exp Ther
2022
;
380
:
180
201
.
30.
Krämer
A
,
Green
J
,
Pollard
J
,
Tugendreich
S
.
Causal analysis approaches in Ingenuity Pathway Analysis
.
Bioinformatics
2014
;
30
:
523
30
.
31.
Chandrashekar
DS
,
Bashel
B
,
Balasubramanya
SAH
,
Creighton
CJ
,
Ponce-Rodriguez
I
,
Chakravarthi
BVSK
, et al
.
UALCAN: a portal for facilitating tumor subgroup gene expression and survival analyses
.
Neoplasia
2017
;
19
:
649
58
.
32.
Zheng
H
,
Zhang
G
,
Zhang
L
,
Wang
Q
,
Li
H
,
Han
Y
, et al
.
Comprehensive review of web servers and bioinformatics tools for cancer prognosis analysis
.
Front Oncol
2020
;
10
:
68
.
33.
Wisniewski
EO
,
Mistriotis
P
,
Bera
K
,
Law
RA
,
Zhang
J
,
Nikolic
M
, et al
.
Dorsoventral polarity directs cell responses to migration track geometries
.
Sci Adv
2020
;
6
:
eaba6505
.
34.
Mazumder
S
,
Mitra Ghosh
T
,
Mukherjee
UK
,
Chakravarti
S
,
Amiri
F
,
Hemmati
F
, et al
.
Integrating pharmacogenomics data-driven prediction with bulk and single-cell RNAseq to demonstrate the efficacy of an NAMPT inhibitor against aggressive, taxane-resistant, and stem-like cells in lethal prostate cancer
.
Cancers
2022
;
14
:
6009
.
35.
Rudzinski
JK
,
Govindasamy
NP
,
Asgari
A
,
Saito
MS
,
Lewis
JD
,
Jurasz
P
.
Preferential interaction of platelets with prostate cancer cells with stem cell markers
.
Thromb Res
2021
;
206
:
42
51
.
36.
Wong
BS
,
Shea
DJ
,
Mistriotis
P
,
Tuntithavornwat
S
,
Law
RA
,
Bieber
JM
, et al
.
A direct podocalyxin-dynamin-2 interaction regulates cytoskeletal dynamics to promote migration and metastasis in pancreatic cancer cells
.
Cancer Res
2019
;
79
:
2878
91
.
37.
Qu
C
,
Rilla
K
,
Tammi
R
,
Tammi
M
,
Kröger
H
,
Lammi
MJ
.
Extensive CD44-dependent hyaluronan coats on human bone marrow-derived mesenchymal stem cells produced by hyaluronan synthases HAS1, HAS2 and HAS3
.
Int J Biochem Cell Biol
2014
;
48
:
45
54
.
38.
Chanmee
T
,
Ontong
P
,
Kimata
K
,
Itano
N
.
Key roles of hyaluronan and its CD44 receptor in the stemness and survival of cancer stem cells
.
Front Oncol
2015
;
5
:
180
.
39.
Ahmat Amin
MKB
,
Shimizu
A
,
Ogita
H
.
The pivotal roles of the epithelial membrane protein family in cancer invasiveness and metastasis
.
Cancers
2019
;
11
:
1620
.
40.
Bhattacharya
R
,
Mitra
T
,
Ray Chaudhuri
S
,
Roy
SS
.
Mesenchymal splice isoform of CD44 (CD44s) promotes EMT/invasion and imparts stem-like properties to ovarian cancer cells
.
J Cell Biochem
2018
;
119
:
3373
83
.
41.
Li
W
,
Qian
L
,
Lin
J
,
Huang
G
,
Hao
N
,
Wei
X
, et al
.
CD44 regulates prostate cancer proliferation, invasion and migration via PDK1 and PFKFB4
.
Oncotarget
2017
;
8
:
65143
51
.
42.
Palapattu
GS
,
Wu
C
,
Silvers
CR
,
Martin
HB
,
Williams
K
,
Salamone
L
, et al
.
Selective expression of CD44, a putative prostate cancer stem cell marker, in neuroendocrine tumor cells of human prostate cancer
.
Prostate
2009
;
69
:
787
98
.
43.
Di Stefano
C
,
Grazioli
P
,
Fontanella
RA
,
De Cesaris
P
,
D'Amore
A
,
Regno
M
, et al
.
Stem-like and highly invasive prostate cancer cells expressing CD44v8–10 marker originate from CD44-negative cells
.
Oncotarget
2018
;
9
:
30905
18
.
44.
Zhao
Y
,
Yan
Q
,
Long
X
,
Chen
X
,
Wang
Y
.
Vimentin affects the mobility and invasiveness of prostate cancer cells
.
Cell Biochem Funct
2008
;
26
:
571
7
.
45.
Singh
S
,
Sadacharan
S
,
Su
S
,
Belldegrun
A
,
Persad
S
,
Singh
G
.
Overexpression of vimentin: role in the invasive phenotype in an androgen-independent model of prostate cancer
.
Cancer Res
2003
;
63
:
2306
11
.
46.
Sethi
S
,
Macoska
J
,
Chen
W
,
Sarkar
FH
.
Molecular signature of epithelial-mesenchymal transition (EMT) in human prostate cancer bone metastasis
.
Am J Transl Res
2010
;
3
:
90
9
.
47.
Deep
G
,
Jain
AK
,
Ramteke
A
,
Ting
H
,
Vijendra
KC
,
Gangar
SC
, et al
.
SNAI1 is critical for the aggressiveness of prostate cancer cells with low E-cadherin
.
Mol Cancer
2014
;
13
:
37
.
48.
Loberg
RD
,
Day
LL
,
Dunn
R
,
Kalikin
LM
,
Pienta
KJ
.
Inhibition of decay-accelerating factor (CD55) attenuates prostate cancer growth and survival in vivo
.
Neoplasia
2006
;
8
:
69
78
.
49.
Loberg
RD
,
Wojno
KJ
,
Day
LL
,
Pienta
KJ
.
Analysis of membrane-bound complement regulatory proteins in prostate cancer
.
Urology
2005
;
66
:
1321
6
.
50.
Hasegawa
M
,
Hagiwara
S
,
Sato
T
,
Jijiwa
M
,
Murakumo
Y
,
Maeda
M
, et al
.
CD109, a new marker for myoepithelial cells of mammary, salivary, and lacrimal glands and prostate basal cells
.
Pathol Int
2007
;
57
:
245
50
.
51.
Jeun
M
,
Park
S
,
Kim
Y
,
Choi
J
,
Song
SH
,
Jeong
IG
, et al
.
Self-normalized detection of ANXA3 from untreated urine of prostate cancer patients without digital rectal examination
.
Adv Healthc Mater
2017
;
6
.
52.
Yee
DS
,
Narula
N
,
Ramzy
I
,
Boker
J
,
Ahlering
TE
,
Skarecky
DW
, et al
.
Reduced annexin II protein expression in high-grade prostatic intraepithelial neoplasia and prostate cancer
.
Arch Pathol Lab Med
2007
;
131
:
902
8
.
53.
Lu
H
,
Xie
Y
,
Tran
L
,
Lan
J
,
Yang
Y
,
Murugan
NL
, et al
.
Chemotherapy-induced S100A10 recruits KDM6A to facilitate OCT4-mediated breast cancer stemness
.
J Clin Invest
2020
;
130
:
4607
23
.
54.
Wang
M
,
Li
X
,
Zhang
J
,
Yang
Q
,
Chen
W
,
Jin
W
, et al
.
AHNAK2 is a novel prognostic marker and oncogenic protein for clear cell renal cell carcinoma
.
Theranostics
2017
;
7
:
1100
13
.
55.
Samy
ALPA
,
Bakthavachalam
V
,
Vudutha
M
,
Vinjamuri
S
,
Chinnapaka
S
,
Munirathinam
G
.
Eprinomectin, a novel semi-synthetic macrocylic lactone is cytotoxic to PC3 metastatic prostate cancer cells via inducing apoptosis
.
Toxicol Appl Pharmacol
2020
;
401
:
115071
.
56.
Li
X
,
Han
X
,
Wei
P
,
Yang
J
,
Sun
J
.
Knockdown of lncRNA CCAT1 enhances sensitivity of paclitaxel in prostate cancer via regulating miR-24–3p and FSCN1
.
Cancer Biol Ther
2020
;
21
:
452
62
.
57.
Parsana
P
,
Amend
SR
,
Hernandez
J
,
Pienta
KJ
,
Battle
A
.
Identifying global expression patterns and key regulators in epithelial to mesenchymal transition through multi-study integration
.
BMC Cancer
2017
;
17
:
447
.
58.
Liotti
A
,
La Civita
E
,
Cennamo
M
,
Crocetto
F
,
Ferro
M
,
Guadagno
E
, et al
.
Periprostatic adipose tissue promotes prostate cancer resistance to docetaxel by paracrine IGF-1 upregulation of TUBB2B beta-tubulin isoform
.
Prostate
2021
;
81
:
407
17
.
59.
Cánovas
V
,
Puñal
Y
,
Maggio
V
,
Redondo
E
,
Marín
M
,
Mellado
B
, et al
.
Prostate tumor overexpressed-1 (PTOV1) promotes docetaxel-resistance and survival of castration resistant prostate cancer cells
.
Oncotarget
2017
;
8
:
59165
80
.
60.
Clarke
MF
,
Fuller
M
.
Stem cells and cancer: two faces of eve
.
Cell
2006
;
124
:
1111
5
.
This open access article is distributed under the Creative Commons Attribution 4.0 International (CC BY 4.0) license.

Supplementary data