Introduction
Approximately 90% of men with advanced prostate cancer (PCa)
1 present with bone metastases. Bone is the preferential site for PCa metastasis and is associated with increased risk of fractures, spinal cord compression, and severe pain [
1]. Although current standard-of-care bone- and osteoclast-targeted therapies, such as radium-223 and denosumab, delay tumor growth and cancer-induced bone disease, respectively, these have been unsuccessful in eliminating and preventing bone metastatic PCa (BM-PCa) growth in bone [
2]. Patients with localized disease have a significantly better outcome than BM-PCa patients, with 5-year survival being reduced from nearly 100% for localized disease to less than 3% of bone metastatic disease [
1]. To date, bone metastatic PCa (BM-PCa) remains incurable.
PCa cells thrive in the bone microenvironment by hijacking the coupled process of bone remodeling characterized by osteoclast-mediated bone degradation and osteoblast-mediated bone formation. Heightened bone turnover results in the release of bone sequestered factors such as transforming growth factor beta (TGFβ) [
3] that promotes increased cancer cell survival and growth. While many studies have focused on the interplay between PCa cells, osteoblasts, and osteoclasts, it is increasingly evident that a large array of cell types in the bone microenvironment including mesenchymal stem cells (MSCs), hematopoietic stem cells (HSCs) and immune cells can contribute to the progression of these bone metastases [
4].
Previously, we observed that PCa can regulate bone formation by recruiting MSCs, which in turn differentiate into bone-forming osteoblasts [
5]. We also examined the differential effect of PCa cells on bone marrow MSC gene expression and observed a significant induction of interleukin (IL-8) in MSCs in response to PCa conditioned media. IL-8 is known to contribute to osteoclast formation but also is a potent chemoattractant for neutrophils [
6]. Neutrophils and neutrophil precursors are the most abundant immune cell in bone, ~ 60% and ~ 40% of the human and mouse bone marrow compartment, respectively [
7,
8]. Neutrophils are generated in the bone marrow at rates of 10
11 cells daily and are regularly released into circulation [
9]. Bodily infection or tissue damage results in a systemic gradient of chemokines being released, including IL-8 and CXCL5, which causes rapid neutrophil expansion and mobilization from the bone marrow or blood circulation into tissues followed by neutrophil secretion of a number of effector molecules, including bactericidal enzyme-containing granules, reactive oxygen species (ROS), and neutrophil extracellular traps (NETs), mesh-like scaffolds of decondensed DNA and granule enzymes [
10]. In the past decade, emerging evidence has demonstrated a role for neutrophils in cancer progression, albeit conflicting, with data indicating both anti- and pro-tumor properties [
11]. Recently, the existence of a pro-tumoral/“N2” and an anti-tumoral/“N1,” phenotype in the tumor microenvironment, has been described with emergence and dominance of the N2 subtype being regulated by TGFβ [
12]. Surprisingly, given their abundance in the bone marrow, the role of neutrophils in the progression of bone metastatic prostate cancer has not been examined thus far.
In the current study, we established the presence of neutrophils in human BM-PCa biopsies and demonstrated that PCa cells stimulate neutrophil migration, oxidative burst, and NET formation, properties associated with neutrophil activation. Reciprocally, direct co-culture assays and live cell imaging revealed that neutrophils induce PCa apoptosis. In support of this, depletion of neutrophils in vivo accelerated the growth of BM-PCa in two independent mouse models of bone metastatic prostate cancer. This phenomenon appears to be regulated by neutrophil inhibition of STAT5 signaling in PCa. Additionally, we observed that the cytotoxic neutrophil phenotype was diminished as BM-PCa progresses in bone. Collectively, these findings are the first to demonstrate an anti-tumor role for neutrophils in the prostate tumor-bone microenvironment and demonstrate that BM-PCa progresses in bone via evasion of neutrophil-mediated PCa death.
Materials and methods
Tissues and cell lines
Luciferase-expressing cell lines were generated by the Cook Laboratory using a lentiviral luciferase reporter (Qiagen), according to protocol. To collect conditioned media (CM), cell lines were rinsed with Phosphate Buffered Saline (PBS) to remove serum, complete medium was replaced with serum-free medium and cells were incubated overnight. CM was collected 18–20 h later by brief centrifugation to remove cellular debris and stored at 4 °C until usage. To minimize media-based changes, all CM was collected in RPMI. Total protein content of CM was measured using BCA assay (ThermoFisher) to ensure equal protein concentrations for treating neutrophils. Fresh media was collected every 2 weeks for experimental use.
Neutrophil isolation from bone marrow
Human bone marrow was processed using a modified Ficoll density centrifugation protocol. Briefly, bone marrow was washed in neutrophil isolation buffer at 7:1 ratio to bone marrow. This suspension was filtered with a 70-μM filter to remove bone fragments. Diluted bone marrow was slowly pipetted onto 15 mL of Ficoll-Paque in a 50-mL conical tube and centrifuged at 445 × g for 35 min at room temperature with no centrifuge brake. The granulocyte layer (the lower layer) was carefully removed and washed twice with buffer, and cells were counted for subsequent experiments. For mouse neutrophil isolation, mouse tibia and femurs were removed from male C57BL/6 mice. Bones were cleared of tissue and muscle, and the epiphysis was removed and discarded. A hole was made in the bottom of a 0.65 mL tube, and one bone was placed individually per tube. This tube was placed into a 1.5-mL tube for bone marrow collection, where it was then centrifuged at high speed for < 5 s. The bone marrow was re-suspended in 1 mL of neutrophil isolation buffer and filtered using a 70-μM filter. Bones from each mouse were pooled and counted, and the EasySep Mouse Neutrophil Enrichment (Stem Cell Technologies) protocol was followed, as per manufacturers’ instructions. Neutrophil purity was validated using flow cytometry for specific markers: human (CD11b+, CD14−, CD15+, CD16+, CD10+) and mouse (CD45+, CD11b+, Ly6Ghi). All neutrophil described functional assays utilized mouse bone marrow neutrophils, and major findings were validated in human neutrophils.
Neutrophil isolation buffer consists of 1X PBS, 2% FBS and 2 mM EDTA. RPMI complete media consists of RPMI (Hyclone), 10% FBS (Peak Serum), and 1% penicillin/streptomycin. DMEM complete media consists of DMEM (Hyclone), 10% FBS (Peak Serum), and 1% penicillin/streptomycin. Luciferase-expressing C42B and PAIII were supplemented with 10 μg of puromycin, and LNCaP with 5 μg puromycin (Gibco). For oxidative burst assays, cells were treated with either serum-free CM, lipopolysaccharide (LPS; 5 μg/ml, Sigma), or phorbol 12-myristate 13-acetate (PMA; 5 nM, Sigma) in serum-containing (10%) RPMI. For TβRI assays, primary neutrophils were incubated in CM supplemented with RepSox (5 nM; Selleckchem).
Neutrophil migration assay
Permeable 24-well transwell migration chambers were used with a pore size of 5 μm (Costar; Ref # 3422). Primary neutrophils were isolated from mouse hind limbs and 1 × 105 were seeded in 250 μL serum-free media in the inserts and 650 μL of conditioned media was added to the bottom of the respective wells, and then inserts were carefully lowered into the wells ensuring that no bubbles formed between the membrane and top of the conditioned media. Neutrophils were incubated for 1 h at 37 °C, then inserts were rinsed in 1x PBS and fixed overnight in methanol at − 20 °C. Fixed inserts were stained with hematoxylin for quantitation of number of migrating neutrophils per insert. Because ~ 90% of the neutrophils completely migrated through the inserts, neutrophils in the lower chamber media were counted using Trypan Blue Exclusion assay. For investigating the importance of IL8/CXCL1 in PCa-mediated neutrophil migration, bone-derived mouse neutrophils were pre-treated with an antibody to mouse CXCR2 (50 nM; Cayman Chemicals) for 30 min and CellTracker Green (Invitrogen) prior to addition to inserts and allowed to migrate for 1 h towards specific PCa CM. After inserts were removed, the wells were imaged using an EVOS FL Auto microscope (Invitrogen; AMAFD1000) at 10x to quantify neutrophils that completely penetrated the membrane.
Immunofluorescence
Paraffin-embedded patient bone specimens and mouse tibia bone sections were dewaxed and hydrated through an alcohol gradient. Antigen retrieval for human and mouse specimens was performed using Tris–EDTA buffer (pH 9) in a pressure cooker for 6 min. Tissues were then blocked in 10% serum in 1X Tris-buffered saline (TBS) for 1 h prior to overnight incubation with primary antibodies (Cytokeratin (dilution 1:500), Sigma C2562; phospho-histone H3 (dilution 1:200), Cell Signaling 9701L; Neutrophil Elastase (dilution 1:200), Abcam ab68672; Myeloperoxidase (dilution 7.5 μg/ml), R&D AF3667). Following washes, species-specific secondary AlexaFluor antibodies were incubated 1:1000 on the tissues for 1 h at room temperature (A11029, A11036, A21206, and A11057). Fluorescent images were taken on a Zeiss Axio Imager.Z2 at 20x and quantified using ImageJ. Phospho-histone H3 was quantified as the number of positive cells compared to the total number of cells per image as determined by DAPI staining. STAT5 was quantified as the amount of positive STAT5 signal per total area of tumor tissue. A threshold for STAT5 signal was set for all images analyzed, and the percent of positive pixels was obtained for each image. Using cytokeratin, the total tumor area in pixels was also obtained. By multiplying the percentage of STAT5 and cytokeratin-positive pixels with the image area, we were able to obtain tumor area and STAT5 positive area. Dividing the calculated STAT5 area by the cytokeratin area yielded the percent of STAT5 per tumor area in each image.
Flow cytometry
Isolated neutrophils were transferred to FACS buffer (2% FBS in 1X PBS) at 1 × 106 cells in 200 μL. For tumor studies, bone marrow was flushed from tumor-bearing and saline-injected tibiae using a syringe and excess cells were further flushed out of the marrow with EasySep buffer. Cells were treated for 2–3 min in 1X Red Blood Cell Lysis Buffer (BioLegend) at room temperature, washed in 1X PBS, and counted for antibody staining. For staining, cells were incubated on ice with 1 μL per 106 cells in mouse or human TruStain Fc block (Biolegend) for 10 min. Fluorophore-conjugated antibodies were added at a maximum of 1 μL per 106 cells (Human-APC/Cy7-CD11b, PE/Cy7-CD14, FITC-CD15, PerCp-Cy5.5-CD10; Mouse- APC-CD45, FITC-CD11b, PE-Ly6G, PerCp-Cy5.5-Ly6C). Cell viability dye, Live/Dead (Invitrogen), was added at a concentration of 0.2 μL per 106 cells. Stained cells were incubated with antibody on ice in the dark for 20 min and rinsed with 1X PBS. Cells were fixed by incubation in 1% formaldehyde in 1XPBS for 30 min in the dark and rinsed with 1X PBS. Prior to analysis, cells were reconstituted in FACS buffer. For all analyses, single cells were gated and marker expression analyzed in myeloid cells.
Oxidative burst assay
Primary mouse neutrophils were isolated from whole bone marrow (EasySep). Neutrophils were incubated under sterile conditions in serum-free conditioned media from LNCaP, C42B, PC3, BPH, and RWPE-1 or appropriate serum-free base medium for an hour at 37 °C in flow cytometry tubes. For a positive control of neutrophil activation and oxidative burst, neutrophils were treated with lipopolysaccharide (LPS; 5 μg/mL) for 30 min in serum-containing RPMI. After incubation in prostate CM, dihydrorhodamine 1,2,3 (DHR123; 25 μM) was added to neutrophils in each CM condition, allowed to incubate for 30 additional minutes at 37 °C and cell fluorescence was analyzed via flow cytometry. Immediately prior to flow analysis, DAPI was added to each condition to measure cell viability. For a longitudinal analysis of oxidative burst, primary neutrophils were placed in a 96-well plate in prostate CM and DHR123 immediately added to each well. Green fluorescence/oxidized DHR was measured over time using an Incucyte S3 Live-Cell Analysis System Imager.
For analysis of NET formation, primary mouse neutrophils were incubated for 2 h in prostate CM (LNCaP, C42B, PC3, BPH, and RWPE-1) or complete RPMI supplemented with PMA (100 nM) as a positive control. Sytox Green dye (500 nM; Sigma) was added to each condition, and after 30 min, images were taken of each well by an EVOS FL Auto microscope. The number of green fluorescent secreted DNA traps/NETs was measured as a percentage of total cells to distinguish between dying cells that absorbed the Sytox Green dye. CM-treated neutrophils were compared to neutrophils incubated in serum-free media RPMI.
Real-time qPCR
For analysis of neutrophils gene expression, RNA was isolated from neutrophils which had been incubated for 3 h at 37 °C, in PCa CM supplemented with 2% FBS. RNA was extracted using Trizol. RNA (1 μg) was used to synthesize cDNA using qSCRIPT Super mix (Quantabio) and PCR performed using Perfecta SYBR Green FastMix (Quantabio). PCR was run using Bio-Rad CFX Real-Time System. PCR conditions were as follows for all primer sequences (Supplemental Table 1): Step 1: 95° 30 s; Step 2: 95° 5 s, 57° 15 s, 72° 10 s, 95° 10 s (× 39 cycles); Step 3: Melt curve 65° 95° increments of 0.5° for 5 s.
PCa co-cultures with neutrophils
Luciferase-expressing LNCaP or C42B cells were plated at 40,000 cells/well in a 24-well plate, in triplicate per condition. Twenty-four hours later, neutrophils were isolated, re-suspended in complete medium, and plated in direct contact with cancer cells at a ratio of 10:1 (neutrophils/cancer). After incubation overnight, neutrophils were removed, and cancer cell viability was measured with Trypan Blue Exclusion assay, using a hemacytometer.
Incucyte S3 live-cell analysis and NanoLive
C42B and LNCaP co-cultures were performed using the above conditions, with modifications in cell density. C42B and LNCaP cells were plated at 9000 cells/well in a 96 well plate in triplicate, per condition. Mouse neutrophils were isolated the following day and plated at a ratio of 10:1 (neutrophils/cancer). The 96-well plate was placed into an Incucyte S3 Live-Cell Analysis System S3 Live-Cell Imager after neutrophils were added. Live-cell images were taken every 10 min and data quantified over a 24-h period. Three replicates of each cell line and condition were averaged, and data were analyzed using the Incucyte S3 Live-Cell Analysis System S3 analysis software (Essen biosciences). For imaging using NanoLive, C42B cells were plated in 6-well plates, and 24 h later, mouse bone marrow-derived neutrophils were added at a 10:1 neutrophil/cancer ratio and imaged for 1 h, as described.
Real-time Glo MT cell viability assay
Neutrophils were incubated in triplicate at 100,000 per well in PCa CM with 2x Real-Time Glo reagent (Promega). MT Cell Viability Substrate and NanoLuc® Enzyme were added in equal volumes to culture media to create the 2x Real-time Glo reagent. This media was added directly to the cells at time zero, and luminescence was read at the indicated time points using a luminometer.
In vivo mouse models
Luciferase-expressing C42B, PAIII, and LNCaP cells were grown to confluence in complete media (base media, 10% fetal bovine serum, 1% penicillin/streptomycin). Cells were trypsinized and washed three times with 1X PBS and filtered using a 70-μM nylon filter. Cells were counted and reconstituted for injection of appropriate cell numbers per 20 μL volume per mouse (500,000 cells-LNCaP and C42B; 50,000-PAIII). Fox Chase SCID Beige male mice (Charles River) were anesthetized with isoflurane, and the right tibia was injected with 20 μL of cells. An equal volume of PBS was injected into the contralateral limb as a control for the intratibial injection. For imaging of tumor burden via bioluminescence, mice were given 10 μL/gram of D-luciferin (15 mg/mL) (Gold Bio) IP imaged using the IVIS Spectrum imager (Perkin-Elmer) on each day of antibody treatment. Luciferase signal was quantified 15 min after injection, using the Living Image Software per manufacturer’s instructions. Using bioluminescence intensity, mice were randomized into 2 groups to receive either (
n = 5/group): a rat IgG2A isotype control antibody or Anti-Ly6G clone 1A8 (BioXcell) to ensure there were no differences in tumor burden at the start of neutrophil depletion. For LNCaP and C42B studies, mice received an intraperitoneal (IP) dose of 400 μg 1A8 on day 3 post-inoculation, and subsequent doses 200 μg twice per week to maintain low neutrophil numbers for the remainder of the study (for 2 weeks), based on previous experiments [
13]. Depletion efficiency was determined by flow cytometry of bone marrow and spleen. For late-stage tumor depletion, C42B cells were intratibially injected in SCID Beige mice (
n = 10/group) and after approximately 2 weeks were treated with isotype control or 1A8 antibody for the remainder of the study. For PAIII, mice received an initial dose of 100 μg antibody beginning on day 1 post-tumor inoculation and every other day for the remainder of the experiment. IVIS imaging was used for longitudinal measurement of tumor burden. To examine tumor-associated neutrophils throughout prostate cancer growth in bone, male SCID Beige mice (
n = 15) were intratibially injected with 500,000 C42B cells in each limb and neutrophils were isolated from each tibia at 3 different time points after injection (week 1, week 2, and week 4). For control tumor-naïve neutrophil collections, additional mice (
n = 5) were injected with saline. Mice were pooled per group and functional analyses performed: PCa co-culture assay, T cell suppression assay and cell viability.
T cell proliferation assay
T cell proliferation assays were performed with neutrophils recovered from the bones of tumor-bearing and control mice as previously described [
14]. Naïve CD4
+ T cells were isolated from the spleens of C57BL/6 mice by negative selection (Biolegend) and labeled with eFluor670 cell proliferation dye (5 μM; ThermoFisher), according to the manufacturer’s instructions. T cells were plated at 1 × 10
4 per well in a round bottom 96-well plate in RPMI-1640 supplemented with 10% FBS, 1%
l-glutamine, 1% HEPES, 1% penicillin/streptomycin, 0.1% β-mercaptoethanol and 100 ng/ml recombinant mouse IL-2 (BioLegend). Control or tumor-associated neutrophils isolated at week 1 and 4 post-intratibial injection were added at 1:1 or 5:1 ratio (T cells:neutrophils) to CD4
+ T cells subjected to polyclonal stimulation with anti-CD3/anti-CD28 Dynabeads (Gibco). TANs were isolated from tumor-bearing tibia, as described, and control neutrophils were isolated from saline-injected tibia.
Trichrome staining
Tibias isolated from mice were fixed in 10% neutral buffered formalin then stored in 70% ethanol. Tibia were decalcified in 14% EDTA buffer for 2 weeks at 4 °C, changing the buffer every 2–3 days. Tibia were then embedded in paraffin and sectioned on a microtome to 5 μM sections. Tissues were cleared of paraffin in 2 changes of xylene and hydrated through an alcohol gradient. Slides were then incubated in Bouin’s fixative (Ricca, 1120-16) for 1 h at 55 °C. Slides were rinsed in running dH2O until clear, then stained with hematoxylin (Ricca, 3530-32) for 5 min. To blue the hematoxylin, slides were dipped in 1x TBS for 30 s, then stained in Gomori’s Trichrome solution (Volu-Sol, VXT-032) for 20 min. Slides were then transferred to freshly made 0.5% acetic acid solution for 2 min. The slides were rinsed in dH2O until clear, then dehydrated, cleared in xylene, and mounted with Permount (Fisher, SP15-500). Images were taken at 4x, 10x, and 20x using an EVOS FL Auto microscope (Life Technologies, AMAFD1000). Areas of osteogenesis were quantified as percent of trabecular bone over total marrow area using ImageJ starting 500 μm below the growth plate and continuing for 2 mm.
Proteome profiler human phospho-kinase array
C42B and LNCaP cells were plated at 100,000 cells/well in a 6-well plate in triplicate, per condition. Mouse neutrophils were isolated the following day and plated at a ratio of 10:1 (neutrophils/cancer). 24 h later, the neutrophils were removed and protein was collected from lysed cancer cells for array analysis, performed per the manufacturer’s instructions (R&D Systems). Image J was used to measure array densitometry.
Immunoblot analysis
Whole-cell extracts were lysed in RIPA buffer and Halt protease and phosphatase inhibitor cocktail (ThermoFisher) was added as per manufacturers’ instructions. Protein concentration was determined using BCA assay (ThermoFisher). Protein lysates were fractionated by SDS-PAGE and transferred to a nitrocellulose membrane using a transfer apparatus according to the manufacturer’s protocols (Bio-Rad). After incubation with 5% nonfat milk in 1X TBST for 1 h, the membrane was washed once with TBST and incubated with appropriate antibodies (STAT5 (1:1000; Cell Signaling #94205), β-actin (1:1000; Cell Signaling #4970) or GAPDH (1:2000; Cell Signaling #5174)) diluted in 1X TBST with 5% milk. Phosphorylated STAT5a/b (1:1000; Cell Signaling #4322) was diluted in 1X TBST with 5% BSA. Membranes were incubated with rotation at 4 °C overnight. After incubation, primary antibody was removed and membranes were washed three times for 10 min each in 1X TBST and incubated with a 1:5000 dilution of horseradish peroxidase-conjugated anti-rabbit antibodies (Cell Signaling) for 1 h at room temperature. Blots were washed with 1X TBST three times and developed with the Clarity Western System (Bio-Rad) according to the manufacturer’s protocols. Blots were imaged on a digital CCD developer (Azure Biosystems).
Generation of STAT5 knockdown cells
STAT5A gene expression was reduced in C42B PCa cells using a STAT5A-targeted HuSH-29 shRNA lentiviral vector, which expresses red fluorescent protein (RFP). C42B cells were transfected using Lipofectamine, and RFP-positive cells were purified by FACS. As a control, C42B were transfected with a non-targeted scrambled sequence HuSH-29 shRNA vector. Three knockdown clones A, C and D were examined in direct co-cultures with mouse neutrophils in comparison to scrambled control C42B.
Statistical analysis
Statistical analyses (t test, ANOVA) were performed using GraphPad Prism (GraphPad Software, Inc). Error bars represent standard error from the mean (SEM).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.