Abstract

Background

   Cancer-associated fibroblasts (CAFs) have been reported to play a
   significant role in the development and metastasis of various tumors;
   however, research on their role in promoting prostate cancer (PCa)
   metastasis under castration conditions remains unclear.

Methods

   In this study, we utilized quantitative reverse transcription
   polymerase chain reaction (qRT-PCR) to detect the expression
   differences of microRNA-196b-5p (miR-196b-5p) in the exosomes secreted
   by CAFs before and after castration. We further characterized the
   transcriptional regulatory landscape through RNA sequencing combined
   with bioinformatics databases. In vitro and in vivo experiments were
   conducted to determine the role of miR-196b-5p in promoting tumor
   migration and metastasis. The dual-luciferase reporter assay, RT-PCR
   analysis, and Western blot analysis confirmed that miR-196b-5p targets
   HOXC8 in prostate cancer. Additionally, transwell assays and Western
   blot analysis were performed to elucidate the role and specific
   mechanisms of HOXC8 in tumor metastasis.

Results

   By analyzing the expression differences of miRNAs in the exosomes
   secreted by CAFs before and after castration, along with relevant data
   from databases, we found that miR-196b-5p is highly secreted by CAFs
   after castration. miR-196b-5p promotes the migration and metastasis of
   prostate cancer cells. Subsequently, through RNA sequencing analysis
   and experimental validation, we determined that miR-196b-5p targets
   HOXC8. This interaction activates the NF-κB pathway, leading to the
   upregulation of epithelial-mesenchymal transition (EMT)-related protein
   expression, thereby driving the metastasis of prostate cancer.

Conclusions

   Our study elucidates a specific mechanism by which CAF-derived exosomes
   promote prostate cancer metastasis via miR-196b-5p regulation,
   contributing to the identification of therapeutic targets for managing
   tumor metastasis following castration.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s13062-025-00667-2.

   Keywords: Prostate cancer, Exosome, Cancer-associated fibroblasts,
   miR-196b-5p, HOXC8, NF-κB

Background

   Currently, prostate cancer (PCa) is the most common male malignancy
   worldwide and the second leading cause of cancer-related deaths in men
   [[44]1, [45]2]. The majority of PCa cases are classified as
   adenocarcinomas, characterized by glandular formation, luminal
   differentiation, and androgen receptor (AR) expression. Androgen
   deprivation therapy (ADT) is a primary treatment for PCa; however,
   nearly all patients experience relapse after an initial response,
   leading to the development of castration-resistant prostate cancer
   (CRPC) [[46]3]. CRPC is often associated with metastasis and results in
   poorer prognosis. Recently, increasing evidence underscores the
   critical role of the tumor microenvironment (TME) and cancer-associated
   fibroblasts (CAFs) in various cancers [[47]4–[48]6]. In prostate
   cancer, CAFs play a crucial role in the development of castration
   resistance; however, the role and mechanisms by which CAFs influence
   the progression and metastasis of PCa under castration conditions
   remain unclear [[49]6, [50]7].

   In our previous study, we found that exosomes derived from CAFs
   significantly enhance the metastatic capabilities of PCa cells in a
   castration environment compared to an androgen-rich environment
   [[51]8]. Exosomes are extracellular vesicles that can transfer miRNAs,
   mRNAs, lncRNAs, and proteins between cells, participating in altering
   the tumor microenvironment, growth, and progression, thereby serving as
   key regulators of cellular signaling [[52]9–[53]11]. Numerous studies
   have reported that exosomes derived from CAFs can promote tumor
   progression [[54]12–[55]15]. Therefore, our research aims to
   investigate whether exosomes from CAFs post-ADT influence the
   progression and metastasis of PCa.

   Our preliminary study revealed that there were no significant changes
   in the quantity of CAF-derived exosomes after ADT [[56]8].
   Consequently, we focused on the components within these exosomes,
   particularly the most abundant miRNAs. Reanalyzing the miRNA sequencing
   data from CAF-derived exosomes, We observed a significant elevation of
   miR-196b-5p in exosomes derived from post-ADT CAFs (Fig. [57]S1A).
   miR-196b-5p has been reported to promote cell migration and invasion in
   various malignancies [[58]16–[59]20]; however, its role in PCa remains
   underexplored. This finding sparked our great interest.

   We transfected PCa cells with miR-196b-5p mimics and demonstrated that
   overexpression of miR-196b-5p significantly enhanced the
   epithelial-mesenchymal transition (EMT), migration, and invasion
   capabilities of PCa cells. We identified homeobox C8 (HOXC8) as a
   target gene of miR-196b-5p in PCa cells. HOXC8 encodes a transcription
   factor containing a homeodomain, and its dysregulation is associated
   with various cancers [[60]21–[61]24]; however, its role in PCa
   metastasis remains unexplored. Further in vitro experiments revealed
   that HOXC8 inhibits EMT by suppressing the activation of the NF-κB
   pathway. These findings suggest that miR-196b-5p from CAFs promotes
   cancer cell metastasis through targeting the HOXC8/NF-κB axis. Our
   results reveal a novel mechanism by which CAFs facilitate tumor
   metastasis following ADT, highlighting the need for further development
   of miRNA-based therapies for metastatic castration-resistant prostate
   cancer (mCRPC).

Materials and methods

Isolation of primary fibroblasts

   Neoplastic and normal prostate tissues were obtained from PCa patients
   treated with radical prostatectomy and patients without PCa undergoing
   total cystectomy, respectively, at the Department of Urology, Shanghai
   General Hospital. The method for isolating primary fibroblasts followed
   previously described protocols [[62]25]. Briefly, prostate tissues were
   initially minced and subjected to enzymatic digestion with
   hyaluronidase and collagenase for one hour. Fibroblasts were
   subsequently isolated via discontinuous gradient centrifugation.

Cell culture

   hTERT PF179T CAF and NCI-H660 cell lines were purchased from American
   Type Culture Collection (ATCC, USA), while LNCaP, DU145, HEK 293T, PC3,
   RWPE-1, VCaP, and C4-2B cell lines were obtained from Cell Bank of
   Shanghai Institute of Cells, Chinese Academy of Science (Shanghai,
   China). All cell lines were authenticated by (GENEWIZ, China). The
   isolated primary fibroblasts, hTERT PF179T CAFs, DU145, and HEK 293T
   cells were cultured in DMEM medium (Gibco, USA) supplemented with 10%
   fetal bovine serum (FBS, Gibco, USA) and 1% penicillin-streptomycin
   (Hyclone, USA, Cat# SV30010). LNCaP, VCaP, and C4-2B cells were
   cultured in RPMI-1640 medium supplemented with 10% FBS and 1%
   penicillin-streptomycin. NCI-H660 cell line was cultured in RPMI-1640
   medium supplemented with 5% FBS, 0.005 mg/mL insulin, 0.01 mg/mL
   transferrin, 30 nM sodium selenite, 10 nM hydrocortisone, 10 nM
   beta-estradiol, and 1% penicillin-streptomycin. PC3 cell line was
   maintained in F-12 K medium containing 10% FBS and 1%
   penicillin-streptomycin. RWPE-1 cell line was cultured in
   Keratinocyte-SFM medium (Invitrogen, USA, Cat# 17005-042), supplemented
   with 5 ng/mL human recombinant epidermal growth factor (EGF) and
   0.05 mg/mL bovine pituitary extract (BPE). All cell lines were cultured
   at 37 °C in a 5% CO₂ incubator.

Transfection

   The lentiviral plasmids encoding miR-196b-5p, HOXC8, and their negative
   controls were obtained from Genomeditech (Shanghai, Chian). Cells were
   infected with miR-196b-5p lentivirus according to the manufacturer’s
   instructions and were selected using puromycin. For HOXC8
   lentivirus-infected cells, blasticidin was used for cell selection. The
   miR-196b-5p mimics, miR-196b-5p negative control, and miR-196b-5p
   inhibitor were synthesized by RiboBio (Guangzhou, China), with
   sequences listed in Table [63]S1. The siRNA for HOXC8 and the negative
   control were designed and synthesized by Genomeditech, with sequences
   available in Table [64]S2. For miRNA, siRNA, or miRNA transfection,
   Lipofectamine 3000 Transfection Reagent (Invitrogen, USA) was used
   according to the manufacturer’s protocol.

Immunofluorescence

   Cells were seeded on coverglasses one day prior to staining. The
   following day, they were fixed with 4% paraformaldehyde at room
   temperature for 20 min and then permeabilized with 0.1% Triton X-100 at
   4 °C for 15 min. The cells were then blocked with 5% donkey serum at
   room temperature for 2 h, followed by incubation overnight at 4 °C with
   primary antibodies: α-SMA (Abcam, UK, Cat# ab7817), FAP (ABclonal,
   China, Cat# A6349), or β-Actin (Servicebio, China, Cat# GB15001). The
   cells were then incubated in the dark at room temperature for 30 min
   with Alexa Fluor 647-conjugated antibody (Abcam, Cat# ab150075) or
   Alexa Fluor 488-conjugated antibody (Abcam, Cat# ab150109). DAPI
   (Beyotime, China) was used to stain the nuclei. Fluorescent microscopy
   (Leica, Germany) was used to observe the cells.

Exosome isolation

   The extraction of exosome was performed as previously described
   [[65]8]. Briefly, cell culture supernatants were collected and
   sequentially centrifuged at 300 × g for 10 min, 2,000 × g for 10 min,
   and 10,000 × g for 30 min. The supernatants were then filtered through
   a 0.22 μm filter (Millipore, USA) and ultracentrifuged at 100,000 × g
   for 70 min at 4 °C. Pellets were resuspended in PBS, ultracentrifuged
   again at 100,000 × g for 70 min, resuspended in PBS, and stored at
   − 80 °C. Exosome release was inhibited using 20 µM GW4869 (Sigma
   Aldrich, USA).

Dual-luciferase reporter assay

   Luciferase reporter assays were performed using luciferase reporter
   constructs containing the NC 3’UTR, wild-type HOXC8 3’UTR, or mutant
   HOXC8 3’UTR regions, synthesized by Genomeditech, with sequences
   available in Table [66]S3. In brief, 293T cells were seeded in 24-well
   plates and co-transfected with the luciferase reporter, followed by
   transfection with the mimic and luciferase vector. After 48 h,
   luciferase activity was measured with the dual-luciferase reporter
   assay kit (Genomeditech).

RNA extraction and RT-qPCR

   Follow the previously mentioned steps to perform RNA extraction and
   conduct qRT-PCR for mRNA and miRNA. In brief, reverse transcription was
   performed using the PrimeScript RT reagent kit (Takara, Japan),
   followed by quantitative real-time PCR with TB Green Premix Ex Taq
   (Takara) on a QuantStudio 7 Flex system (Applied Biosystems, USA). mRNA
   levels were quantified using the ΔΔCt method, with β-actin serving as
   the reference for mRNA and U6 serving as the reference for miRNA. The
   primers used in this analysis are listed in Table [67]S4.

Western blotting

   Total cellular proteins were extracted using M-PER (Thermo Scientific,
   USA) with inhibitor of protease and phosphatase (NCM Biotech, China,
   Cat# P002) and quantified using a BCA assay kit (Thermo Scientific).
   Approximately 30 micrograms of total protein were loaded onto SDS-PAGE
   gels and subsequently transferred to 0.22 μm nitrocellulose membranes
   (Cytiva, USA). The membranes were blocked at room temperature with 5%
   BSA for 1.5 h, followed by overnight incubation at 4 °C with primary
   antibodies. The primary antibodies included anti-E-cadherin (Cell
   Signaling Technology, USA, Cat# 14472 S), anti-N-cadherin (Proteintech,
   China, Cat# 66219), anti-Vimentin (Abcam, UK, Cat# ab92547),
   anti-Phospho-IκBα (Cell Signaling Technology, Cat# 2859), anti-IκBα
   (Cell Signaling Technology, Cat# 4812), anti-p65 (Cell Signaling
   Technology, Cat# 8242), anti-Histone H3 (Cell Signaling Technology,
   Cat# 9715), anti-Snail (Cell Signaling Technology, Cat# 3879 S),
   anti-Slug (Cell Signaling Technology, Cat# 9585 S), anti-HOXC8
   (Proteintech, Cat# 15448), and anti-β-actin (Servicebio, China, Cat#
   GB15001). Secondary antibodies were applied at room temperature for 1 h
   the following day, followed by visualization.

Cell viability test

   Cell viability was assessed using the Cell Counting Kit-8 (CCK-8)
   assay. Ten microliters of CCK-8 reagent (NCM Biotech, China, Cat#
   C6005) were added to each well of a 96-well plate. The plate was
   incubated at 37 °C, 5% CO₂ incubator for 1.5 h, after which the
   absorbance was measured at 450 nm using a spectrophotometer.

Transwell assay and wound healing assay

   For the Transwell migration assay, DU145, PC3, or LNCaP cells were
   seeded in the upper chamber containing serum-free medium, while the
   lower chamber was filled with medium supplemented with 10% FBS. After
   incubating the plates at 37 °C, 5% CO₂ incubator for 48 h, the cells
   were fixed with 4% paraformaldehyde and stained with 0.1% crystal
   violet. Images were then captured using a fluorescence microscope
   (Leica, Germany). DU145 or PC3 cells were seeded in six-well plates one
   day prior to the wound healing assay. After incubating for 24 h, the
   cells were washed with PBS, and a linear scratch wound was created. The
   plates were then incubated at 37 °C, 5% CO₂ for 24 h, and images were
   captured using a fluorescence microscope (Leica, Germany).

Colony forming

   Cells were seeded in six-well plates at a density of 400 to 1,000 cells
   per well in each experimental group and incubated for 7 days. After
   incubation, cells were washed once with PBS, followed by the addition
   of 1 mL of 4% paraformaldehyde to fix the cells for 60 min. The cells
   were then washed again with PBS, and 1 mL of crystal violet solution
   was added to stain the cells for 15 min. After staining, the cells were
   washed several times with PBS and photographed.

Xenograft models and bioluminescence

   Male BALB/c nude mice aged 4 to 6 weeks underwent orchiectomy. DU145
   luciferase cells were injected via the tail vein. Tumor metastasis was
   monitored using a bioluminescent in vivo imaging system (IVIS, USA) at
   4 weeks post-injection. Mice were anesthetized with 1.5% isoflurane in
   air and received an intraperitoneal injection of D-luciferin (200 µL,
   15 mg/mL in PBS) prior to imaging. After 8 weeks post-tumor injection,
   the mice were euthanized, and lung tissues were excised for imaging.
   All animal procedures were conducted by laboratory animal guidelines
   and were approved by the Animal Care and Use Committee of Shanghai
   General Hospital.

Immunohistochemistry

   Tumors from mice were collected and fixed in 4% paraformaldehyde,
   followed by dehydration using a gradient of ethanol concentrations.
   Subsequently, tumors were embedded in paraffin and processed for
   sectioning. Immunohistochemical staining was performed using anti-HOXC8
   antibody (Proteintech, China, Cat# 15448), and the expression of HOXC8
   in the specimens was quantified using the immunohistochemical IRS
   scoring system.

Bioinformatic data acquisition

   Transcriptomic sequencing data and prognostic information for TCGA-PRAD
   were obtained from the TCGA database
   ([68]https://portal.gdc.cancer.gov/). Sequencing data for [69]GSE70770
   and [70]GSE64333 were sourced from the GEO database
   ([71]https://www.ncbi.nlm.nih.gov/geo/) [[72]26, [73]27]. The
   expression of MIR196B in neuroendocrine prostate cancer (NEPC) versus
   castration-resistant prostate cancer (CRPC) was compared using
   Beltran’s dataset [[74]28].

Transcriptome sequencing and differential gene expression analysis

   Total RNA was isolated using TRIzol Reagent (Invitrogen, USA) following
   the manufacturer’s protocol. RNA purity and concentration were assessed
   using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA), while
   RNA integrity was evaluated with an Agilent 2100 Bioanalyzer (Agilent
   Technologies, USA). Transcriptomic analyses were performed by OE
   Biotech (Shanghai, China). Differential gene expression analysis was
   conducted using the DESeq2 package (Version 1.36.0) in R, with a
   significance threshold of P < 0.05 and|log2 fold change| > 1.”

Statistical analysis

   Statistical analyses were conducted using Prism version 9.5.0.
   Differences between two groups were assessed using the Student’s
   t-test, with a P-value less than 0.05 considered statistically
   significant. Data are presented as mean ± standard deviation (SD).

Results

Expression of miR-196b-5p in primary CAFs and PCa cell lines

   We successfully isolated and cultured primary CAFs from three PCa
   samples, as well as primary normal fibroblasts from three different
   normal prostate samples. Immunofluorescence was utilized to validate
   the phenotype of the primary CAFs and primary normal fibroblasts. The
   expression of fibroblast activation protein-α (FAP) is a key
   characteristic of CAF activation [[75]29]. Notably, while detectable in
   activated fibroblasts participating in wound healing processes, FAP
   exhibits no expression in other healthy tissues [[76]30]. FAP-positive
   CAFs contribute to shaping an immunosuppressive tumor microenvironment
   by secreting various cytokines [[77]31, [78]32]. Currently, FAP is one
   of the most promising therapeutic targets in CAFs. The staining
   demonstrated that both the primary CAFs and the hTERT PF179T CAF had
   significantly higher FAP expression compared to primary normal
   fibroblasts, highlighting the activated fibroblast phenotype in these
   cells (Fig. [79]1A). Following total RNA extraction, we observed that
   miR-196b-5p expression was significantly elevated in both primary CAFs
   and the hTERT PF179T CAF compared to primary normal fibroblasts
   (Fig. [80]1B). Further analysis of baseline miR-196b-5p expression in
   multiple PCa cell lines and RWPE-1 cell line revealed that miR-196b-5p
   levels were significantly higher in various prostate cancer cell lines,
   especially in androgen-insensitive cell lines such as DU145, PC3 and
   NCI-H660 (Fig. [81]1C).

Fig. 1.

   [82]Fig. 1
   [83]Open in a new tab

   Identification of miR-196b-5p expression levels in primary fibroblasts
   and prostate cell lines. (A) Immunofluorescence staining of α-SMA and
   FAP in primary normal prostate fibroblasts, primary prostate CAFs, and
   hTERT PF179T CAFs (scale bar = 100 μm). (B) The expression levels of
   miR-196b-5p in primary normal prostate fibroblasts, primary prostate
   CAFs, and hTERT PF179T CAFs measured by RT-PCR. (C) The expression
   levels of miR-196b-5p in the normal prostate epithelial cell line
   RWPE-1 and various prostate cancer cell lines measured by RT-PCR. Data
   are presented as mean ± SD, representing triplicate measurements.
   (Student’s t-test, ***P < 0.001)

miR-196b-5p significantly elevated in CAFs exosomes after ADT and transferred
to PCa cells

   To investigate how castration conditions affect the secretion of
   miR-196b-5p by CAFs, we treated CAFs with media containing DHT as well
   as control media without DHT. The results showed that CAF-derived
   exosomes treated with ETOH had significantly higher miR-196b-5p levels
   compared to DHT-treated CAF-derived exosomes (Fig. [84]2A). We then
   treated DU145 and LNCaP cell lines with the aforementioned exosomes and
   observed that the ETOH-treated CAF-derived exosomes significantly
   elevated miR-196b-5p levels in PCa cells compared to DHT-treated
   CAF-derived exosomes (Fig. [85]2B). To confirm whether miR-196b-5p was
   transferred from CAFs to PCa cells via exosomes, we transiently
   transfected CAFs with Cy3-labeled miR-196b-5p and co-cultured them with
   DU145 cells for 24 h. Fluorescence microscopy revealed red signals
   (Cy3-labeled miR-196b-5p) and green signals (β-actin) in DU145, with
   the red signal disappearing after CAFs were treated with the exosome
   inhibitor GW4869 (Fig. [86]2C). Additionally, we extracted exosomes
   from CAFs transfected with Cy3-labeled miR-196b-5p as well as from
   control CAFs, and co-cultured them with DU145 cells for 24 h. The
   results showed that Cy3 signals was detectable in DU145 cells. And we
   obtained similar results in LNCaP cells (Fig. [87]S2A-B). These results
   indicate that CAFs secrete significant amounts of miR-196b-5p via
   exosomes under ADT, and transfer it to PCa cells.

Fig. 2.

   [88]Fig. 2
   [89]Open in a new tab

   Exosomes transfer miR-196b-5p from CAFs to PCa cells. (A) Expression of
   miR-196b-5p in exosomes from DHT-treated CAFs and matched ETOH-treated
   CAFs measured by RT-PCR. (B) Expression of miR-146a-5p in LNCaP and
   DU145 cells measured by RT-PCR at 24 h after treating with exosomes
   (25 µg/mL) derived from DHT-treated CAFs or ETOH-treated CAFs. (C) CAFs
   were transiently transfected with Cy3-labeled miR-control or
   miR-196b-5p and co-cultured with DU145 cells for 24 h following GW4869
   treatment. Fluorescence microscopy was used to observe green signals
   (β-actin) and red signals (Cy3) in DU145 cells (scale bar = 50 μm). (D)
   Exosomes were isolated from the supernatant of CAFs transfected with
   Cy3-labeled miR-196b-5p or miR-control, and exosomes (25 µg/mL) were
   incubated with DU145 cells for 24 h. Fluorescence microscopy was used
   to detect green signals (β-actin) and red signals (Cy3) in DU145 cells
   (scale bar = 50 μm). Data are presented as mean ± SD, representing
   triplicate measurements. (Student’s t-test, **P < 0.01, ***P < 0.001,
   ****P < 0.0001)

Transcriptome analysis revealed an elevated expression of miR-196b-5p in PCa
samples

   We conducted bioinformatics analysis to assess the expression levels of
   miR-196b-5p across multiple PCa datasets. The results indicated that
   tumor samples in the TCGA-PRAD database exhibited significantly higher
   levels of miR-196b-5p expression compared to normal samples
   (Fig. [90]3A). Similarly, the [91]GSE64333 dataset also indicated that
   miR-196b (precursor of miR-196b-5p) levels were significantly elevated
   in tumor samples (Fig. [92]3B). Additionally, the [93]GSE70770 dataset,
   which includes CRPC samples, showed that the expression of miR-196b was
   significantly higher in CRPC samples than in HSPC samples
   (Fig. [94]3C). Furthermore, in the [95]GSE70770 dataset, the expression
   level of miR-196b was correlated with the patients’ pathological
   staging, showing higher expression in samples from patients at the
   T3-T4 stage compared to those at the T2 stage (Fig. [96]3D). Although
   not statistically significant (p = 0.067), the high miR-196b-5p
   expression group was associated with poorer biochemical recurrence-free
   survival (BCRFS) based on the TCGA-PRAD dataset (Fig. [97]S1B).
   Interestingly, in the Beltran (2016) dataset, the levels of miR-196b
   were significantly higher in NEPC samples compared to those in CRPC,
   suggesting that miR-196b-5p may play a role not only in the progression
   from HSPC to CRPC but also potentially from CRPC to NEPC.

Fig. 3.

   [98]Fig. 3
   [99]Open in a new tab

   Expression landscape of miR-196b-5p across multiple prostate cancer
   datasets. (A) Differential expression of miR-196b-5p between tumor and
   normal samples from the TCGA-PRAD database. (B) Analysis of
   differential expression of miR-196b in tumor and normal samples from
   the [100]GSE64333 dataset. (C) Differential expression of miR-196b
   between CRPC and HSPC samples from the [101]GSE70770 dataset. (D)
   Comparative differential expression analysis of miR-196b between tumor
   samples at stages T3–T4 and T2 based on the [102]GSE70770 dataset

miR-196b-5p promotes cell migration and invasion in PCa cells by activating
EMT

   Our previous research has confirmed that exosomes derived from CAFs
   under ADT can enhance the migration and invasion of PCa cells by
   activating EMT both in vivo and in vitro [[103]8]. Notably, various
   miRNAs, especially miR-196b-5p, were upregulated in exosomes secreted
   by CAFs after castration. To examine whether the increased levels of
   miR-196b-5p influence the migration and invasion of PCa cells, we
   transfected PCa cell lines with miR-196b-5p mimics and conducted
   Transwell and wound healing assays. The results indicated that the
   expression of miR-196b-5p facilitated the migration and invasion of
   both DU145 and LNCaP cells (Figs. [104]4A-D, and [105]S4A).
   Subsequently, We performed Western blot analysis of EMT proteins in
   DU145 and LNCaP cell lines infected with lenti-miR-196b-5p. The results
   revealed that, in the group infected with lenti-miR-196b-5p, the
   relative expression of N-cadherin, Snail, and Slug was increased, while
   the relative expression of E-cadherin decreased, with no significant
   changes observed in Vimentin expression (Fig. [106]4G-H). Furthermore,
   we transfected miR-196b-5p inhibitor into the PC3 cell line, which has
   a high baseline expression of miR-196b-5p (Fig. [107]4I). The results
   from Transwell and wound healing assays demonstrated that the reduced
   expression of miR-196b-5p inhibited the migration and invasion of PC3
   cells (Figs. [108]4J-K, Fig. [109]S4B), with Western blotting showing
   increased expression of E-cadherin and decreased expression of
   N-cadherin, Snail, and Slug. In summary, miR-196b-5p promotes migration
   and invasion in PCa cells by activating EMT.

Fig. 4.

   [110]Fig. 4
   [111]Open in a new tab

   miR-196b-5p promotes cell migration and invasion in PCa cells via
   activating EMT. (A) and (B) Transwell assays performed to assess the
   effect of miR-196b-5p on migration (A) and invasion (B) in DU145 cells
   (scale bars = 25 μm). (C) and (D) Transwell assays performed to assess
   the effect of miR-196b-5p on migration (C) and invasion (D) in LNCaP
   cells (scale bars = 25 μm). (E) and (F) Efficiency of miR-196b-5p
   overexpression in DU145 and LNCaP cells was measured by RT-PCR. (G) and
   (H) Western blot utilized to assess the effect of miR-196b -5p on EMT
   marker protein in DU145 and LNCaP cells, respectively. Each set of
   three lanes in the figure represents three biological replicates. (I)
   Knockdown efficiency of miR-196b-5p in PC3 cells by miR-196b-5p
   inhibitor measured using RT-PCR. (J) and (K) Transwell assays performed
   to assess the effect of miR-196b-5p inhibitor on migration and
   invasion, respectively (scale bars = 25 μm). (L) Western blot analysis
   used to detect EMT marker protein in PC3 cells. Data are presented as
   mean ± SD, representing triplicate measurements. (Student’s t-test,
   *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001)

Identification of transcripts targeted by miR-196b-5p in PCa

   To clarify the impact of miR-196b-5p on the gene expression profile, we
   conducted RNA sequencing analysis on DU145 cells transfected with
   lenti-miR-196b-5p or lenti-miR-control. Our result indicated
   significant changes in the transcriptome due to miR-196b-5p,
   identifying a total of 1,009 differentially expressed genes (DEGs),
   including 519 upregulated genes and 490 downregulated genes
   (Fig. [112]5A). The heatmap also demonstrated that the DEGs could
   effectively differentiate between miR-196b-5p overexpressing samples
   and control samples (Fig. [113]5B). To identify potential targets of
   miR-196b-5p in PCa cells, we employed multiple public databases for
   microRNA target gene prediction. This analysis revealed that HOXC8,
   HOXA9, RDX, IGF2BP1, and GATA6 were highly correlated target genes
   (Fig. [114]5C). We intersected these five genes with the 490
   downregulated DEGs identified in our RNA sequencing data, resulting in
   HOXC8 (homeobox C8) being the only highly relevant gene (Fig. [115]5D).
   HOXC8 encodes a transcription factor containing a homeobox domain, and
   its dysregulation is associated with various tumors; however, its role
   in PCa remains unexplored. To verify whether miR-196b-5p can directly
   target the 3’ UTR of HOXC8, we designed luciferase reporter vectors
   containing either wild-type or mutated HOXC8 3’ UTR regions for
   dual-luciferase reporter assays (Fig. [116]5E). The results showed that
   co-transfection with miR-196b-5p mimics significantly reduced the
   luciferase activity of the HOXC8 WT 3’ UTR compared to co-transfection
   with miR-NC mimics in HEK-293T cells. In contrast, no significant
   direct interaction was observed between miR-196b-5p and the HOXC8 MUT
   3’ UTR-containing vector (Fig. [117]5F). Subsequently, we assessed the
   RNA and protein levels of HOXC8 in DU145 and LNCaP cell lines,
   revealing a notable downregulatory effect of miR-196b-5p on HOXC8
   expression. (Fig. [118]5G-J). In conclusion, we demonstrated that
   miR-196b-5p targets the particularly interesting gene HOXC8 in PCa,
   which is a focus point of our study.

Fig. 5.

   [119]Fig. 5
   [120]Open in a new tab

   HOXC8 is a direct target of miR-196b-5p in PCa cells. (A) Volcano plot
   of DEGs. (B) Heatmap of DEGs. (C) miR-196b-5p target genes predicted by
   the MIRWalk, TargetScan, and miRDB databases. (D) Intersection of
   significantly downregulated DEGs with predicted miR-196b-5p target
   genes. (E) Predicted target sequence of miR-196b-5p in the 3′-UTR of
   HOXC8, along with the construction of wild-type and mutant HOXC8 3′-UTR
   luciferase reporter constructs in HEK-293T cells. (F) The relative
   luciferase activities were measured after co-transfection of miR-NC
   mimics or miR-196b-5p mimics and a luciferase vector encoding NC,
   wild-type, or mutant HOXC8 3′-UTR region. (G) and (H) HOXC8 mRNA levels
   in DU145 and LNCaP cells transfected with lenti-miR-196b-5p. (I) and
   (J) HOXC8 protein levels transfected with lenti-miR-196b-5p in DU145
   and LNCaP cells. Data are presented as mean ± SD, representing
   triplicate measurements. (Student’s t-test, *P < 0.05, **P < 0.01,
   ****P < 0.0001)

HOXC8 inhibits EMT in PCa cells

   To investigate whether HOXC8 expression is also associated with cell
   migration and invasion, we transfected PCa cell lines with
   lenti-HOXC8-OE or lenti-HOXC8-control. The expression efficiency of
   HOXC8 was thereafter validated using RT-PCR and Western blot analysis
   (Fig. [121]6A-D). We subsequently examined the impact of HOXC8
   expression on the levels of EMT proteins and found that overexpression
   of HOXC8 in both DU145 and LNCaP cell lines led to a significant
   increase in E-cadherin expression (Fig. [122]6E-F). Through Transwell
   assays and wound healing assays, we discovered that overexpression of
   HOXC8 significantly inhibited the migration and invasion of PCa
   cells(Fig. [123]6G-J). Considering the important role of HOXC8 as a
   transcription factor, we performed RNA sequencing analysis on DU145
   cells transfected with lenti-HOXC8-OE or lenti-HOXC8-control, with
   relevant DEGs and transcription factor prediction analyses presented in
   Fig. [124]S6. To further investigate whether miR-196b-5p exerts its EMT
   activating effects by inhibiting HOXC8 expression in PCa cells, we
   restored HOXC8 expression in PCa cells transfected with
   lenti-miR-196b-5p. Western blot analysis indicated that HOXC8
   overexpression could reverse the effects of miR-196b-5p in DU145 and
   LNCaP cells (Fig. [125]7A-B). Importantly, similar results were
   obtained when HOXC8 was knocked down using siRNA in PCa cells,
   resembling the effects of miR-196b-5p transfection (Fig. [126]7C-F).
   These experiments suggest that miR-196b-5p activates EMT in PCa cells
   by downregulating HOXC8.

Fig. 6.

   [127]Fig. 6
   [128]Open in a new tab

   HOXC8 inhibits EMT in PCa cells. (A) and (B) Western blot analysis to
   assess the efficiency of HOXC8 overexpression in DU145 and LNCaP cells,
   respectively. (C) and (D) RT-PCR analysis to assess the efficiency of
   HOXC8 overexpression in DU145 and LNCaP cells, respectively. (E) and
   (F) Western blot assessment of EMT marker protein in DU145 and LNCaP
   cells transfected by HOXC8-control or HOXC8-OE. (G) and (H) Evaluation
   of the effects of HOXC8 on migration (G) and invasion (H) in DU145
   cells by Transwell assays (scale bars = 25 μm). (I) and (J) Evaluation
   of the effects of HOXC8 on migration (I) and invasion (J) in LNCaP
   cells by Transwell assays (scale bars = 25 μm). Data are presented as
   mean ± SD, representing triplicate measurements. (Student’s t-test,
   *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001)

Fig. 7.

   [129]Fig. 7
   [130]Open in a new tab

   miR-196b-5p promotes EMT in PCa cells by targeting HOXC8. (A) and (B)
   Western blot analysis to assess the protein levels of E-cadherin,
   N-cadherin, Snail, Slug, and HOXC8 in DU145 and LNCaP cells transfected
   with lenti-miR-control, lenti-miR-196b-5p, lenti-miR-196b-5p
   HOXC8-control, or lenti-miR-196b-5p HOXC8-OE. (C) and (D) Relative
   expression levels of HOXC8 mRNA in DU145 and LNCaP cells transfected
   with siHOXC8. (E) and (F) Western blot analysis to assess the protein
   levels of E-cadherin, N-cadherin, Snail, Slug, and HOXC8 in DU145 and
   LNCaP cells following transfected with siHOXC8

ADT-induced miR-196b-5p activates EMT and suppresses HOXC8 expression in vivo

   To confirm the effect of miR-196b-5p on PCa metastasis in vivo
   following castration, we induced androgen deprivation by performing
   surgical castration on BALB/c nude mice. DU145 luciferase cells were
   transfected with miR-196b-5p and injected via the tail vein. Tumor
   metastasis was monitored using small animal in vivo imaging. Consistent
   with in vitro experiments, the luciferase signal and the number of lung
   metastases in the miR-196b-5p group were significantly higher than
   those in the control group (Fig. [131]8A-B). Additionally, IHC analysis
   confirmed that the tumors in the control group exhibited significantly
   stronger HOXC8 staining compared to the miR-196b-5p group
   (Fig. [132]8C).

Fig. 8.

   [133]Fig. 8
   [134]Open in a new tab

   miR-196b-5p promotes metastasis of PCa cells in vivo. (A)
   Bioluminescence images of BALB/c nude mice that injected with DU145
   luciferase cells treated with either miR-196b-5p mimics or miR-NC
   mimics. The color scale bar depicts the photon flux emitted from the
   mice. (B) Tumor foci in lungs of BALB/c nude mice after injecting DU145
   cells treated with either miR-196b-5p mimics or miR-NC mimics. (C)
   Immunohistochemical staining of HOXC8 in tumors from mice after
   injecting DU145 cells treated with either miR-196b-5p mimics or miR-NC
   mimics. (Student’s t-test, *P < 0.05, **P < 0.01)

miR-196b-5p activates NF-κB pathway by targeting HOXC8

   To further investigate the downstream mechanisms of miR-196b-5p
   targeting HOXC8, we performed KEGG pathway enrichment analysis and GO
   enrichment analysis on the DEGs in DU145 cells transfected with
   miR-196b-5p (Figs. [135]9A and [136]S5A-B). The KEGG pathway enrichment
   result indicated that the DEGs were primarily enriched in the NF-κB
   signaling pathway. The GSEA enrichment analysis of the NF-κB signaling
   pathway is shown in Fig. [137]S5C. Considering the increasing evidence
   suggesting that the HOX protein family plays an important role in
   regulating the NF-κB pathway [[138]33], we conducted a series of
   experiments to investigate the effects of miR-196b-5p and HOXC8 on this
   pathway. The expression of IκBα (NF-κB inhibitor α) and p-IκBα
   (phosphorylated IκBα) were assessed through western blot analysis. The
   results showed that the expression of p-IκBα protein was upregulated in
   DU145 and LNCAP cells transfected with miR-196b-5p, and this change was
   reversed in cells restored HOXC8 (Fig. [139]9B-C). To further confirm
   the activation of the NF-κ pathway, we evaluated the expression of
   nuclear protein P65 in DU145 and LNCAP cells. In prostate cancer (PCa)
   cells transfected with miR-196b-5p, nuclear P65 expression was
   significantly elevated; however, the restoration of HOXC8 notably
   decreased nuclear P65 protein levels (Fig. [140]9D-E).

Fig. 9.

   [141]Fig. 9
   [142]Open in a new tab

   miR-196b-5p exhibits its functions by activating NF-κB signaling
   pathway in PCa cells (A) KEGG enrichment analysis of DEGs. (B-E)
   Western blot analysis of IKBα, p-IKBα, and nuclear protein P65 levels
   in DU145 and LNCaP cells transfected with lenti-miR-control,
   lenti-miR-196b-5p, lenti-miR-196b-5p HOXC8-control, or
   lenti-miR-196b-5p HOXC8-OE. (F-I) Western blot analysis of E-cadherin,
   N-cadherin, Snail, Slug, IKBα, p-IKBα, and nuclear protein P65 levels
   in PCa cells treated with NF-κB inhibitor after transfection with
   lenti-miR-control, lenti-miR-196b-5p, lenti-miR-196b-5p HOXC8-control,
   or lenti-miR-196b-5p HOXC8-OE

   Furthermore, the addition of BAY 11-7082 (an NF-κB pathway inhibitor)
   to PCa cells transfected with miR-196b-5p revealed that BAY 11-7082
   exhibited effects similar to those of HOXC8, inhibiting IκBα
   phosphorylation and P65 nuclear translocation (Fig. [143]9F-I).
   Surprisingly, although BAY 11-7082 has been reported as an IκBα
   phosphorylation inhibitor [[144]34], our findings suggest that it
   appears to inhibit total IκBα protein. Notably, the utilization of BAY
   11-7082 to inhibit the NF-κB signaling pathway counteracted the effects
   of miR-196b-5p on EMT activation, as evidenced by an increase in
   E-Cadherin expression and a decrease in N-Cadherin, Snail, and Slug
   expression (Fig. [145]9F-G). The expression levels of nuclear P65 in
   DU145 and LNCaP cell lines are shown in Figs. [146]9H and I,
   respectively, with the corresponding quantification presented in Fig.
   [147]S7. Collectively, these experimental results strongly indicate
   that miR-196b-5p activates EMT by inhibiting HOXC8, thereby activating
   the NF-κB signaling pathway (Fig. [148]10).

Fig. 10.

   [149]Fig. 10
   [150]Open in a new tab

   The flowchart of this study, created with BioRender.com

Discussion

   Increasing evidence suggests that the biological behaviors of cancer
   cells, such as proliferation, invasion, and metastasis, are strongly
   influenced by the characteristics of stromal cells in the
   microenvironment [[151]35–[152]37]. CAFs, as the most abundant
   component in the tumor microenvironment, play a crucial role in the
   progression to CRPC following ADT for prostate cancer [[153]38,
   [154]39]. Studies have shown that CAFs secrete growth factors and
   cytokines that promote the proliferation, migration, and invasion of
   tumor cells [[155]40, [156]41]. Additionally, CAFs synthesize and
   remodel the extracellular matrix, creating a barrier that hinders the
   penetration of drugs and immune cells, thereby reducing the
   effectiveness of tumor therapy [[157]42].

   The extensive heterogeneity of CAFs in prostate cancer is a challenge
   that limits their clinical therapeutic translation. ScRNA-seq analysis
   can categorize CAFs into different molecular subtypes, which may have
   distinct biological and clinical significance in tumor progression and
   immune-suppressive microenvironments [[158]43]. Research indicates that
   androgen deprivation therapy-induced SPP1^+ myCAFs are key stromal
   components driving the development of castration-resistant prostate
   cancer. These myofibroblasts can paracrine SPP1 to activate the ERK
   signaling in tumor cells, promoting castration resistance in prostate
   cancer [[159]39]. Furthermore, the interaction between CAFs and
   prostate cancer stem cells (PCSCs) is also important for the
   progression and therapeutic resistance of prostate cancer. CAFs promote
   the growth and survival of PCSCs by releasing signaling molecules and
   altering the surrounding environment, and vice versa, PCSCs may also
   influence the characteristics and behavior of CAFs through the
   production of various molecules [[160]44]. This crosstalk mechanism may
   be crucial for the development of progression and therapeutic
   resistance in prostate cancer. In our research, we found that CAFs
   post-castration treatment have an exuberant secretory function and can
   mediate cancer cell metastasis through the exosomal pathway. This
   finding helps explain the clinical phenomenon of metastasis in prostate
   cancer patients after ADT and is significant for targeting CAFs-derived
   exosomes in combination with ADT treatment to improve the prognosis of
   advanced prostate cancer patients.

   MiRNAs are the key component of exosomes derived from CAFs and have
   been demonstrated to play a critical role in the progression of various
   cancers, including prostate cancer [[161]45–[162]47]. MiRNAs are widely
   dysregulated in CRPC and have the potential to serve as therapeutic
   targets and biomarkers [[163]48–[164]51]. In our study, we identified
   that miR-196b-5p as a crucial factor promoting metastasis in PCa. It
   has been reported that in non-small cell lung cancer, miR-196b-5p
   promotes tumor proliferation, migration, and invasion by directly
   targeting the tumor suppressor genes FAS, GATA6, and TSPAN12 [[165]18,
   [166]52]. In laryngeal squamous cell carcinoma, miR-196b enhances tumor
   cell proliferation and invasion by targeting SOCS2 [[167]53].
   Additionally, the entire miR-196 family has been shown to promote cell
   migration and invasion in oral cancer cells [[168]54]. These findings
   indicate that miR-196b-5p plays a pro-tumorigenic role across various
   cancers.

   HOX transcription factors are among the most widely studied target
   genes of the miR-196 family [[169]55–[170]57]. The HOX family comprises
   a group of evolutionarily conserved genes that regulate crucial
   embryonic developmental processes, such as anterior-posterior axis
   formation and organ morphogenesis, while also playing significant roles
   in cell apoptosis, proliferation, and migration [[171]58]. Different
   HOX proteins exhibit varying functions in different tumors, acting as
   either oncogenes or tumor suppressors. HOXA5 is well-known as a tumor
   suppressor in breast cancer and solid tumors like colon cancer
   [[172]59], yet it does not exhibit the same effect in leukemia
   [[173]60]. Conversely, HOXA9 is an important oncogene in leukemia
   [[174]61], while its role in breast cancer is contrary [[175]62]. In
   our study, we employed luciferase assays and western blot analysis to
   identify HOXC8 as a direct target of miR-196b-5p, demonstrating that it
   regulates the expression of EMT-related proteins through the NF-κB
   pathway. There is a well-documented connection between the expression
   of HOX family proteins and NF-κB signaling, where HOX transcription
   factors and NF-κB mutually regulate each other via various mechanisms
   [[176]63]. For instance, a study highlighted that HOXA9 can inhibit
   NF-κB signaling by reducing its ability to bind to DNA [[177]64]. Our
   study is the first to elucidate the role of HOXC8 protein in promoting
   tumor metastasis and regulating the NF-κB signaling pathway, offering
   novel insights that may contribute to the development of effective
   therapies for CRPC.

   Our study also has some limitations. First, the potential mechanism
   behind the increase in miR-196b-5p in CAF-derived exosomes following
   ADT remains unclear. Second, the precise mechanism by which HOXC8
   influences NF-κB is yet to be elucidated. HOX transcription factors
   have been reported to interact with NF-κB through various mechanisms,
   including direct physical interactions, transcriptional regulation, and
   modulation of upstream regulators [[178]65–[179]67]. The specific
   regulatory mechanism between HOXC8 and NF-κB requires further
   investigation.

Conclusion

   In conclusion, we conducted an in-depth investigation into the
   oncogenic role of miR-196b-5p derived from CAF exosomes following ADT,
   along with its specific mechanism of action. Our study revealed that
   its target, HOXC8, is closely associated with the NF-κB pathway and EMT
   regulation. These novel findings may pave the way for further
   exploration of the pro-carcinogenic mechanisms of CAFs in prostate
   cancer and the identification of potential therapeutic targets.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [180]13062_2025_667_MOESM1_ESM.tif^ (1.4MB, tif)

   Supplementary Material 1: Supplementary Fig. 1 (A) The volcano plot
   illustrates the highly expressed miRNAs in exosomes derived from CAFs
   treated with DHT. (B) Kaplan-Meier analysis comparing BCRFS differences
   between the high miR-196b-5p group and the low miR-196b-5p group. (The
   high-expression and low-expression groups were categorized by the
   optimal cutoff value.) (C) The differential expression of miR-196b
   between NEPC and CRPC samples in the Beltran (2016) cohort.
   [181]13062_2025_667_MOESM2_ESM.tif^ (9.9MB, tif)

   Supplementary Material 2: Supplementary Fig. 2 (A) CAFs were
   transiently transfected with Cy3-labeled miR-control or miR-196b-5p, as
   well as with Cy3-labeled miR-196b-5p treated with GW4869, and then
   co-cultured with DU145 cells for 24 h. Fluorescence microscopy was used
   to observe the green signals (β-actin) and red signals (Cy3) in LNCaP
   cells (scale bar = 50 μm). (B) Exosomes were isolated from the
   supernatant of CAFs transfected with Cy3-labeled miR-196b-5p or
   miR-control, and exosomes (25 µg/mL) were added to DU145 cells and
   incubated for 24 h. Fluorescence microscopy was used to detect the
   green signal (β-actin) and red signal (Cy3) in LNCaP cells (scale bar
   = 50 μm).
   [182]13062_2025_667_MOESM3_ESM.tif^ (3.8MB, tif)

   Supplementary Material 3: Supplementary Fig. 3 (A) and (B) Cell
   viability levels were assessed using the CCK-8 assay in DU145 and LNCaP
   cells following transfection with lenti-miR-control or
   lenti-miR-196b-5p. (C) and (D) Cell proliferation levels were evaluated
   through a colony-forming assay in DU145 and LNCaP cells after
   transfection with lenti-miR-control or lenti-miR-196b-5p.
   [183]13062_2025_667_MOESM4_ESM.tif^ (4.9MB, tif)

   Supplementary Material 4: Supplementary Fig. 4 (A) Cell migration
   ability was evaluated in DU145 cells using a wound healing assay
   following transfection with miR-NC mimics or miR-196b-5p mimics. (B)
   Cell migration ability was assessed in PC3 cells using a wound healing
   assay after transfection with a miR-196b-5p inhibitor or control. (C)
   Cell migration ability was measured in DU145 cells using a wound
   healing assay after transfection with HOXC8-OE or control.
   [184]13062_2025_667_MOESM5_ESM.tif^ (2.6MB, tif)

   Supplementary Material 5: Supplementary Fig. 5. (A) Chord diagram of GO
   enrichment analysis for upregulated genes in DU145 cells overexpressing
   miR-196b-5p. (B) Chord diagram of GO enrichment analysis for
   downregulated genes in DU145 cells overexpressing miR-196b-5p. (C) Gene
   Set Enrichment Analysis of the NF-kB signaling pathway.
   [185]13062_2025_667_MOESM6_ESM.tif^ (5MB, tif)

   Supplementary Material 6: Supplementary Fig. 6. (A) Volcano plot of
   differentially expressed genes in DU145 cells overexpressing HOXC8. (B)
   Heatmap of differentially expressed genes in DU145 cells overexpressing
   HOXC8. (C) Sankey diagram of differential transcription factor
   (TF)-target gene interactions: from left to right, the first column
   represents transcription factor families, the second column represents
   differentially expressed transcription factors, and the third column
   represents differentially expressed target genes. Lines indicate
   relationships between TF families, TFs, and target genes. (D)
   Distribution of differential TF families: the x-axis represents TF
   families, and the y-axis represents gene counts. Dark blue bars
   represent all genes regulated by each TF family; yellow bars represent
   differentially expressed TFs regulated by each family; pink bars
   represent upregulated differentially expressed TFs; and light blue bars
   represent downregulated differentially expressed TFs. (E) Circular plot
   of the top 30 genes in the protein-protein interaction (PPI) network.
   Red nodes represent upregulated differentially expressed genes, and
   blue nodes represent downregulated differentially expressed genes. The
   size of each gene node reflects the number of associated genes.
   [186]13062_2025_667_MOESM7_ESM.tif^ (1.2MB, tif)

   Supplementary Material 7: Supplementary Fig. 7. (A) The expression
   levels of nuclear P65 in each group of DU145 cells. (B) The expression
   levels of nuclear P65 in each group of LNCaP cells. Data are presented
   as mean ± SEM, representing triplicate measurements.
   [187]13062_2025_667_MOESM8_ESM.docx^ (14.1KB, docx)

   Supplementary Material 8: Table S1 The miR-196b-5p mimics and matched
   control sequences used in this study.
   [188]13062_2025_667_MOESM9_ESM.docx^ (14.2KB, docx)

   Supplementary Material 9: Table S2 siRNA sequences used in the study.
   [189]13062_2025_667_MOESM10_ESM.docx^ (14.4KB, docx)

   Supplementary Material 10: Table S3 The plasmid sequences used for the
   dual-luciferase reporter assay.
   [190]13062_2025_667_MOESM11_ESM.docx^ (14.2KB, docx)

   Supplementary Material 11: Table S4 Primer sequences used in our study.

Acknowledgements