Secretome analysis
Proteomic analysis of culture medium from the PCa cell lines, described in Supplemental Materials “Cell culture and maintenance’’, was performed as previously described (23). In brief, the PCa cell lines were cultured under the recommended conditions until they reached 60% confluency. Then, the media were replaced with serum-free media, and cells were incubated for 48 h. The culture media were then collected and lyophilized. The lyophilized media were dissolved in 10 mM ammonium bicarbonate containing 4 M urea, and proteins were desalted by acetone precipitation. The precipitated protein was resuspended in 25 mM ammonium bicarbonate containing 4 M urea and 0.1% RapiGest detergent (Nihon Waters, Tokyo, Japan) and subsequently digested with trypsin for 16 h at 37°C after reduction, alkylation and dilution. The resulting peptides were desalted using C18 Stage Tips (24) and analyzed on an LTQ Orbitrap Velos (Thermo Fisher Scientific) equipped with a reverse-phase LC system. Peptides were detected sequentially in positive ion mode for MS/MS in data-dependent scanning mode and identified using Proteome Discoverer 2.5 software (Thermo Scientific) and the Swiss-Prot human database (www.uniprot.org/proteomes/UP000005640) with the following parameters: enzyme, trypsin; peptide mass tolerance, ± 5 ppm; fragment mass tolerance, ± 0.5 Da; maximum missed cleavage sites, 2; variable modifications: oxidation of methionine, acetylation and/or loss of methionine at N-terminus; and static modification: carbamidomethylation of cysteine. Mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium (PXD045369, http://www.proteomexchange.org/) via the jPOST partner repository (JPST002261, https://jpostdb.org/).
Structure modeling with AlphaFold2
Structure predictions for pre-pro-GDF15, GDPP and mGDF15 were generated by the AlphaFold2 (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb, accessed on 4 June 2023) model using the relevant online resources with their default settings (25, 26).
Open source RNA-sequencing analysis
RNA-seq transcriptome data of various cancer patients, including 493 PCa, 407 bladder cancer, 510 renal cell carcinoma, 1082 breast cancer, 484 lung cancer, 592 colon cancer, 443 melanoma, 412 gastric cancer, 366 liver cancer, 527 uterine cancer, 181 esophageal cancer, 515 head and neck carcinoma, 514 glioma and 177 pancreatic cancer patients, were downloaded from the TCGA database in 2018.
Immunofluorescence staining
LNCaP cells were seeded in 2-well chamber slides (5712-002, IWAKI) at a density of 3 × 105 cells/1.5 ml/well and incubated overnight at 37°C in a humidified atmosphere containing 5% CO2. The cells were then washed with PBS and fixed with 4% paraformaldehyde for 15 minutes on ice. After permeabilization with 0.1% Triton X-100 (87361, Muto Pure Chemicals Co., Ltd.) in PBS at room temperature for 15 minutes, the cells were incubated with primary antibodies diluted in PBS-T overnight at 4°C. The primary antibodies used were a rabbit anti-GDPP polyclonal antibody (HPA011191, Sigma‒Aldrich, 1:200) and a mouse anti-mGDF15 monoclonal antibody (sc-515675, Santa Cruz Biotechnology, 1:50). After PBS washes, the slides were incubated with the appropriate secondary antibodies, Alexa Fluor 488 goat anti-mouse secondary antibody (A-11001, Invitrogen) and Alexa Fluor 568 goat anti-rabbit secondary antibody (A-11011, Invitrogen), both diluted in PBS-T (1:500), for 1.5 hours at room temperature. The slides were then washed with PBS-T at 22°C and counterstained with ProLong Gold Antifade reagent with DAPI (P36931, Invitrogen). The stained LNCaP cells were examined using a fluorescence microscope (BZ-X710, KEYENCE). Rabbit polyclonal IgG (NBP2-24891, Novus) and mouse monoclonal IgG (ab18469, Abcam) were used as isotype controls for the respective antibodies.
Human sample collection and data
We collected human serum, plasma and tissue samples, such as PCa and bone samples, from healthy donors and PCa patients at Osaka University from December 2012 to December 2022. Total 386 patients (30 healthy donors, 60 localized PCa patients, 15 mCRPC patients without BM, 80 mCRPC patients with BM, 22 mCRPC patients with BM whose blood samples were collected over time and 179 PCa patients who underwent radical prostatectomy), we collected clinical information retrospectively. Whole blood (2.0–7.0 ml) was collected directly into Venoject Ⅱ EDTA-2Na tubes (TERUMO) for plasma samples, and whole blood (2.0–7.0 ml) was collected directly into Venoject Ⅱ tubes (TERUMO) for serum samples. Within three hours of collection, all plasma samples were centrifuged sequentially at 900 and 20,000 × g for 10 min each, and the supernatants were stored at − 80°C as plasma. All serum samples were centrifuged at 3000 rotations per minute (rpm) for 5 min, and the supernatants were stored at -80°C as serum. Serum PSA (Beckman Coulter), ALP (Shino-Test Corporation), BAP (IDS, Inc.), TRACP 5b (Nittobo Medical), LDH (FUJIFILM Wako Pure Chemical Corporation), OC (Tosoh), mGDF15 (R&D Systems) and PⅠNP (USCN) levels were measured in the same blood samples. Bone scan index (BSI) was assessed within two months of both blood collection time points.
Bone scintigraphy
All PCa patients were injected intravenously with 740 MBq of 99mTc MDP to evaluate the existence of BM. Three hours after injection, a whole-body bone scan was performed with a gamma camera equipped with a low-energy high-resolution parallel hole collimator in anterior and posterior views. The raw image data set was analyzed with the software package BONENAVI version 2, based on a personal database in Japan. This CAD system was used to calculate the BSI, which was calculated as a percentage of the sum of all spots classified as bone metastases in the patient's body. When the attending physician deemed it necessary especially for CRPC patients, it was taken about once every three month and their data was retrospectively analyzed.
Establishment of an ELISA system to measure GDPP
Anti-GDPP monoclonal antibodies targeting the GDF15 propeptide, namely, GD11-13 and GD01-62, were generated using a plasmid DNA immunization method, as we reported previously (27). These antibodies specifically recognize the central region of GDPP. To detect GDPP, a combination of GD11-13, immobilized on magnetic microparticles, and GD01-62, labeled with alkaline phosphatase, was employed. AIA-CL reagent (Tosoh) was developed based on the two-step sandwich enzyme immunoassay technique. Using the fully automated chemiluminescent enzyme immunoassay system (AIA-CL2400, Tosoh), sample dispensing, immunoreaction, B/F separation, substrate addition, and luminescence detection were performed automatically, and results were obtained in approximately 15 minutes.
Cell culture and maintenance
LNCaP and DU145 cells were purchased from RIKEN BRC CELL BANK, 22Rv1 and PC3 cells were purchased from the American Type Culture Collection (ATCC). All cell lines were maintained in basal culture medium (RPMI1640) (Nacalai Tesque) with 10% fetal bovine serum (FBS), 100 U/mL penicillin G, and 0.1 µg/mL streptomycin in a humidified incubator set to 37°C and 5% CO2. PC3-Luc2 cells were also purchased from the ATCC and maintained in basal culture medium [Ham’s F-12K Kaighn’s medium, Gibco™; 10% FBS; 8 µg/mL Blasticidin S (Invitrogen)] in a humidified incubator set to 37°C and 5% CO2. MC3T3-E1 cells (RIKEN BRC Cell Bank) were maintained in basal culture medium (αMEM, Nacalai Tesque) with 10% FBS, 100 U/mL penicillin G, and 0.1 µg/mL streptomycin, MLO-Y4 cells (Kerafast) were cultured on type I collagen-coated dishes (Corning) and maintained in basal culture medium (αMEM with 5% heat inactivated FBS, 5% calf serum, 100 U/mL penicillin G, and 0.1 µg/mL streptomycin), and OSC14C cells (Cosmo Bio) were suspended in osteoclast culture medium (OSCMW and OSCMM, Cosmo Bio). HOB (PromoCell, lot number #469Z022, from cancellous bone/femoral head tissue collected from a 78-year-old Caucasian man) was cultured in osteoblast growth medium (C-27001, PromoCell), and OSC15C (Cosmo Bio, lot number #VJ2-F-OSH) was cultured in osteoclast wash medium (OSCMW, Cosmo Bio) and growth medium including receptor activator of NF-κB ligand and macrophage-colony stimulating factor (OSCMW, Cosmo Bio). HOB was used for functional analysis with a maximum of five passages allowed for cell culture. OSC14C and OSC15C differentiation into mature osteoclasts was confirmed by TRAP staining. A cell scraper (99002, Techno Plastic Products) was used to scrape off the cells.
Analysis of secreted proteins in cell culture media
To analyze the secreted proteins in culture media, we seeded 5×105 cells in 2 ml of serum-free medium into a 6-well dish and collected the culture medium 24 hours after seeding. This medium was passed through a filter (Millex-GV, SLGVR33RS, Merck), and the filtrate was collected after centrifugation at 6000×g for 30 minutes using a centrifugal concentrator (Vivaspin, VS2091, SARTORIUS). The GDPP concentration in the culture medium for each sample was measured in triplicate. In total, 5 × 105 LNCaP cells were seeded in 6-well dishes, and the medium was changed 24 hours later. Then 25 µM furin inhibitor (#14965, Cayman Chemical) was added for 24 hours, and the whole-cell lysate and culture medium were collected.
Sodium dodecyl sulfate‒polyacrylamide gel electrophoresis (SDS‒PAGE) and western blotting
For SDS‒PAGE, sample buffer containing 10% 2-mercaptoethanol was added to whole-cell lysates, generated using RIPA Lysis Buffer (Santa Cruz Biotechnology), or culture media, and proteins were resolved on 10% polyacrylamide mini gels (TEFCO). Afterward, proteins were transferred onto a polyvinylidene difluoride membrane using a semidry transfer system (Thermo Fisher Scientific). The membrane was then probed with the indicated specific antibodies that were utilized for immunological analysis: GDPP (1:1000, HPA011191, Sigma‒Aldrich), mGDF15 (1:1000, LS-C383688, LSBio), and β-actin (1:5000, 4967S, Cell Signaling Technology). The membrane was incubated with a horseradish peroxidase-conjugated secondary antibody against rabbit immunoglobulin (1:5000, Cell Signaling Technology). Finally, the membrane was subjected to detection with enhanced chemiluminescence western blotting detection reagents (Nacalai Tesque) and visualized using the ChemiDoc XRS Plus system (Bio-Rad) as a chemiluminescence detector.
Development of human recombinant GDPP
The sequence of human GDPP with a Strep-tag at the N-terminus was cloned and inserted into an expression vector, and the resulting plasmid was amplified and utilized to transfect Expi293 mammalian cells for Strep-GDPP expression. The transfected cells were cultured, and the culture medium was collected. The recombinant GDPP protein was purified from the culture medium using a Strep-tag purification kit (IBA Lifesciences) according to the manufacturer’s instructions.
Immunohistochemical studies
Both human and mouse bone metastasis specimens were demineralized using Tris-ethylenediaminetetraacetic acid (EDTA) demineralization solution until tissue softening was observed, followed by paraffin fixation. Immunohistochemical staining was performed in 4 µm-thick paraffin-embedded tissue samples. The human sample sections were treated with EDTA buffer (pH 9.0) and activated by warming at 125°C for 30 seconds using a Pascal pressure chamber (S2800, Dako) for antigen activation treatment. Endogenous peroxidase activity was blocked by incubating the sections with 0.3% hydrogen peroxide for 5 min, followed by overnight incubation with primary antibodies against GDPP (1:200; HPA011191, Sigma‒Aldrich) at 4°C, and staining was performed using DAB substrate (MK210, TaKaRa). Finally, the sections were counterstained with hematoxylin. In the mouse tibial bone tissue specimens, antigen activation treatment was performed with 3-fold diluted Proteinase K Ready-to-use (S3020, Dako). The sections were incubated overnight at 4°C with a primary antibody against the osteoblast marker OC (M188, Takara, diluted 100 times), followed by incubation with secondary anti-rat antibody (714311, Nichirei Bioscience, Inc.). Staining was performed using DAB substrate (MK210, TaKaRa). Osteoclasts were stained using a commercially available TRAP Staining Kit (AK04F, Cosmo Bio). Osteoclasts were identified as TRAP-positive multinucleated (three or more nuclei) cells, and osteoblasts and osteoclasts were counted on the trabecular bone matrix surface in three randomly selected fields of view using light microscopy (BZ-X710, KEYENCE).
RNA interference
For knockdown of GDF15 using small interfering RNA (siRNA), cells were transfected with 10 nM of either targeting FlexiTube GeneSolution (GS9518, Qiagen) or negative control Stealth RNAi™ (12935112, Invitrogen) using Lipofectamine® RNAiMAX Reagent (13778075, Invitrogen) for 24 hours. Then, the medium containing siRNA and transfection reagent was replaced with fresh medium. Following validation of GDF15 knockdown confirmed by western blotting method, functional assays were performed.
Cell proliferation assay
PC3 cells and HOB transfected with either siRNA targeting GDF15 or negative control for 72 h were reseeded in medium supplemented with 10% FBS in 96-well plates at 1 × 103 cells/100 µL/well and 1.3× 103 cells/100 µL/well, respectively; DU145 cells were seeded in the same medium in 96-well plates at 1 × 103 cells/100 µL/well. The cells were incubated for 1 hour at 37°C in a humidified 5% CO2 atmosphere, and then 0.1 µl MT Cell Viability Substrate (G9712, Promega) and 0.1 µl NanoLuc® Enzyme (G9712, Promega) were added to each well. Luminescence was measured with a GloMax® Explorer System (GM3510, Promega) according to the manufacturer’s instructions after 24, 48, and 72 hours in a humidified incubator set to 37°C and 5% CO2; this timepoint was set as 0 hours, and GDPP was added at this point. The assay was repeated three times for each experimental group.
Wound-healing assay
PC3 cells transfected with either siRNA targeting GDF15 or negative control siRNA for 72 h were reseeded in 6-well plates at 6 × 105 cells/2 mL/well, and DU145 cells were seeded in 6-well plates at 6 × 105 cells/2 mL/well. Cells were grown to a monolayer, and a wound was created by scraping the cell layer using a sterile 200-µL yellow pipette tip when the cells reached approximately 90% confluence. Detached cells were removed by washing plates with PBS and adding fresh culture medium supplemented with 10% FBS to each plate. Cells were treated with or without GDPP at this point and then incubated at 37°C with 5% CO2. Cell migration was evaluated using a fluorescence microscope (BZ-X710, KEYENCE) at 0 h and 20 h after wound generation and quantified by measuring the size of the recovered area using ImageJ 1.53e. The assay for each experimental group was repeated three times.
Cell invasion assay
PC3 cells transfected with either siRNA targeting GDF15 or negative control siRNA for 72 h and DU145 cells were seeded into the upper chambers of Corning® BioCoat™ Matrigel Invasion Chambers (pore size—8 µm) (354480, Corning) (5 × 104 cells/well) in serum-free medium, with medium supplemented with 10% FBS present in the lower chamber. GDPP was added immediately after cell seeding. After 36 h of incubation at 37°C and 5% CO2, the cells that had penetrated into the Matrigel matrix were fixed and stained using a Diff-Quik stain kit (Sysmex, Kobe, Japan), and cell invasion was quantified using a fluorescence microscope (BZ-X710, KEYENCE). The analysis was repeated three times for each experimental group.
Quantitative real-time PCR
Total RNA was extracted from HOB (treated with or without GDPP for 3 days) and OSC15C cells (treated with or without GDPP for 12 days) using the ISOSPIN Cell & Tissue RNA Kit (NIPPON GENE). After verification of RNA quality by NanoDrop One (Thermo Fisher Scientific), the RNA was subjected to cDNA synthesis using the Prime Script RT reagent Kit (Perfect Real Time; TaKaRa Bio). Quantitative RT‒PCR was performed using the Thermal Cycler Dice Real Time System (TP800; Takara) and THUNDERBIRD™ Next SYBR® qPCR Mix (TOYOBO). Target gene expression was normalized to that of the housekeeping gene GAPDH using the delta-delta Ct method. The primers used for the PCRs are listed in Table S1, and all PCRs were performed in triplicate for each sample.
ALP staining and assessment of activity
HOB was seeded in 24-well plates and cultured on type I collagen-coated dishes (Corning) at 1×105 cells/0.5 ml/well with or without GDPP. After 5 days of incubation, for ALP staining, the cells were washed with PBS and fixed for 20 min with 10% formalin at room temperature. After fixation, the cells were incubated with an Alkaline Phosphatase Staining kit (AK20, Cosmo Bio Co., Ltd.) for 20 minutes at 37°C according to the manufacturer’s instructions. The total percentage of ALP+ cells was determined using a fluorescence microscope. To measure ALP activity, HOB was seeded to 24-well plates and incubated with or without GDPP for 5 days. Then, WCLs were obtained as stated in the Supplementary Materials section “Sodium dodecyl sulfate‒polyacrylamide gel electrophoresis (SDS‒PAGE) and western blotting’’, and ALP activity was measured using p-nitrophenylphosphate as the substrate for the alkaline phosphatase test (QFAP-100, BioAssay System) according to the manufacturer’s instructions. The optical density (OD) of WCLs was measured in a 96-well plate at 405 nm with a microplate reader (iMARK™, Bio-Rad). The assay was repeated three times for each experimental group.
Alizarin Red S staining and bone mineralization quantification
Human primary osteoblasts (HOB) was seeded in 24-well plates and cultured on type I collagen-coated dishes (354408, Corning) at 3×104 cells/0.5 ml/well using Osteoblast Mineralization Medium (C-27020, PromoCell) with or without GDPP for 21 days. The cells were subsequently washed with PBS, fixed, stained and digested with an Alizarin Red S staining kit (BMK-R009, BMK, Bio Mirai Kobo). The total area of staining was quantified using a fluorescence microscope, and after staining, the OD of the eluted Alizarin Red S solution was measured in a 96-well plate at 405 nm with a microplate reader (iMARK™, Bio-Rad). The assay was repeated three times for each experimental group.
TRAP staining and resorption pit assay
Human-derived primary osteoclast precursor cells (OSC15C) was seeded in 24-well plates at 2×105 cells/0.5 ml/well with or without GDPP for 12 days. TRAP staining was performed using a TRAP staining kit (AK04F, Cosmo Bio) according to the manufacturer’s instructions, and the total number of TRAP-positive cells with ≥ 3 nuclei was determined using a fluorescence microscope as the average of five randomly observed fields of view. The resorption pit assay was performed using the Bone Resorption Assay Kit (BRA-48KIT, PG Research). OSC15C cells were seeded in a 48-well Bone Resorption Assay Plate 48 (BRA-48P, PG Research) at 8×104 cells/0.5 ml/well with or without GDPP for 12 days. The bone resorption area was quantified using a fluorescence microscope as the average of five randomly observed fields of view. The assay was repeated three times for each experimental group.
Animal experiments with bone metastasis of prostate cancer
Male NOD.CB17-PrkdcSCID/J mice, 5–6 weeks of age, were purchased from Japan Charles River Laboratories, Inc. All mice were euthanized under anesthesia using isoflurane. For intratibial implantation, 1 × 106 PC3-Luc2 cells, purchased from ATCC, were suspended in 5 µl of VitroGel Hydrogel Matrix (VHM01S, TheWell Bioscience LLC) and 5 µl of phosphate-buffered saline (PBS). Mice were anesthetized with isoflurane, and the cell suspension was directly injected into the intramedullary cavity of the right tibia. The cavity was reached by drilling into the cortical bone of the tibial tuberosity using a 22 G needle (NN-2232R, TERUMO) with a 1 ml 29G syringe containing a needle (08299, NIPRO). Then, the skin was closed with a 6–0 suture. Mice in the treatment group were subcutaneously injected with recombinant GDPP (refer to the section “Development of human recombinant GDPP’’) dissolved in saline at a concentration of 0.1 mg/kg every other day. Control mice were injected with an equal amount, and subsequent tumor growth was evaluated weekly with bioluminescence analysis via an In Vivo Imaging System (IVIS® Lumina II, Caliper). The mice were randomized into two groups for experiments: the control (n = 10) and GDPP groups (n = 10).
Micro-CT and IVIS imaging
The mice were imaged to visualize luciferase activity immediately after injection and were monitored weekly using IVIS® imaging. Bioluminescence images of tumor-bearing mice were acquired with an IVIS Spectrum 10 minutes after intraperitoneal injection of D-luciferin (XLF-1, Summit Pharmaceutical International Corporation, 100 mg/kg), with an exposure time of 10 seconds. As a quality control measure, the photon flux was measured by quantifying the number of highlighted pixels within a circular region of interest (ROI) for each mouse in the supine position. These values were then normalized to the signal intensity obtained immediately after xenografting in the same area (day 0) of each mouse. Thus, all mice had an arbitrary starting normalized bioluminescence signal intensity of 1. This normalization was performed using Living Image® software version 4.2 (Caliper Life Sciences, Inc.) following the manufacturer's instructions. Bioluminescence imaging was employed to assess tumor burden and the localization of PC3-Luc2 cells. The bone-destructive phenotype caused by PC3-Luc2 cells was visually evaluated macroscopically. Additionally, three-dimensional images were constructed to confirm the presence of bone infiltration using R_mCT2 software (Rigaku) for micro-CT imaging.
Statistics
The statistical analyses were performed using JMP Pro (v.17.0.0; SAS Institute, NC, USA). Univariate analysis included two-tailed Student's t-test and the Mann‒Whitney U test. Multiple comparisons were assessed using the Tukey‒Kramer method to compare several treatments. Spearman's rank correlation coefficient was used as a measure of the strength of the correlation between the two variables. Statistical significance was defined as p < 0.05. The optimal cutoff value for diagnosis was determined from the receiver operating characteristic curve using the Youden index, and sensitivity and specificity for diagnosis were calculated based on each optimal cutoff value. The cutoff values for the parameters used in the prognostic analysis are the respective medians.