ATAD2 Drives Prostate Cancer Progression to Metastasis

Abstract

Metastasis accounts for the overwhelming majority of cancer deaths. In prostate cancer and many other solid tumors, progression to metastasis is associated with drastically reduced survival outcomes, yet the mechanisms behind this progression remain largely unknown. ATAD2 (ATPase family AAA domain containing 2) is an epigenetic reader of acetylated histones that is overexpressed in multiple cancer types and usually associated with poor patient outcomes. However, the functional role of ATAD2 in cancer progression and metastasis has been relatively understudied. Here we employ genetically engineered mouse models of prostate cancer bone metastasis, as well as multiple independent human cohorts, to show that ATAD2 is highly enriched in bone metastasis compared to primary tumors and significantly associated with the development of metastasis. We show that ATAD2 expression is associated with MYC pathway activation in patient datasets and that, at least in a subset of tumors, MYC and ATAD2 can regulate each other’s expression. Using functional studies on mouse bone metastatic cell lines and innovative organ-on-a-chip bone invasion assays, we establish a functional role for ATAD2 inhibition in diminishing prostate cancer metastasis and growth in bone.

Introduction

Prostate cancer is a heterogenous disease in which prognosis is highly impacted by the development of metastasis (1). This occurs mainly in bones and entails high patient morbidity and mortality (2–4). Identifying the factors that determine how prostate cancer can metastasize to bone is thus of utmost importance for risk stratification and disease management as well as for designing novel therapeutic strategies. Progression to metastatic disease is most frequently observed after local therapy and subsequent androgen deprivation, termed metastatic castration resistant prostate cancer, or mCRPC. In a minority of patients (5–10%) metastasis may be evident before androgen deprivation (termed metastatic hormone sensitive disease, mHSPC) either at diagnosis (‘de novo’, synchronous metastatic disease) or after local therapy (metachronous metastatic disease) (5, 6). The mechanisms behind this progression are likely multifactorial and many remain largely unknown (7–9). Genetically engineered mouse models (GEMMs) have provided multiple cues into how this progression occurs (10–12). These models allow the study of spontaneous cancer development and progression in a genetically controlled system, including in immunologically intact whole organisms. We and others have shown in multiple models that genetic alterations may be necessary for cancer to progress to metastatic disease, yet other factors cooperate to ultimately determine metastasis outcomes. These include cell intrinsic activation of critical transcriptional modulators (ie, MYC, ETV4, CITED2) (13–17) and signaling pathways (ie, NFKB, Wnt, MAPK) (18–21) as well as expansion of stromal cell subtypes (22, 23).

Expression of ATAD2 (ATPase family AAA domain containing 2) has been linked to tumor progression and metastasis in multiple tumor types (24–26). However, except for a few studies in lung metastasis (27, 28), proof of its functional regulation of metastasis is largely lacking. ATAD2 functions as an epigenetic reader of acetylated histones (26, 29, 30) and is considered a modulator of the transcriptional activity of multiple transcription factors, including oncogene MYC and the androgen receptor (AR) (31, 32), both critical regulators of prostate cancer progression.

Although its precise biological function is still unclear, in prostate cancer ATAD2 is regulated by androgens and may in turn interact with AR to modulate its signaling in favor of cancer cell survival (32, 33). Moreover, ATAD2 may affect the epigenetic landscape of prostate tumors by favoring open chromatin conformations that are enriched upon cancer progression (34), as well as through regulation of other key epigenetic modulators, such as EZH2 (35).

Previous studies including our own have put forward MYC as a critical regulator of both early and late progression of prostate cancer (13, 15, 36, 37). Indeed, using a unique GEMM that enables quantitative studies on progression of prostate cancer to bone metastasis, we found MYC to be necessary but not sufficient for metastasis. Furthermore, based on integration of transcriptional profiles in this model and in human datasets, we have derived a gene signature downstream of MYC and RAS, called META16, that includes ATAD2 and is enriched in metastasis versus primary tumors as well as associated with worse metastasis-free survival (15).

Given the clinical relevance of bone metastasis in prostate cancer, the incomplete understanding of its molecular mediators and the lack of functional validation of the relevance of ATAD2 for metastasis, we studied whether ATAD2 drives prostate cancer progression to metastasis, including bone. By performing functional studies in human and mouse models and correlative studies on multiple human datasets, we show that ATAD2 is regulated by MYC, is consistently upregulated in metastasis compared to primary tumors, and that its inhibition is a promising therapeutic strategy for metastatic prostate cancer.

Materials and Methods

Mouse studies and analyses of genetically engineered mouse models.

All experiments using animals were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at Columbia University Irving Medical Center and Icahn School of Medicine at Mount Sinai. All mice were housed in pathogen-free barrier conditions under 12-h light–dark cycles and with temperature and humidity set points at 20–25 °C and 30–70%, respectively. As our focus was prostate cancer, only male mice were used. NP (RRID:IMSR_JAX:033751) and NPK (RRID:IMSR_JAX:033761) mice were bred with a conditional EYFP reporter, as previously described (12, 15, 38). Mice were induced to form tumors at 2–3 months of age by administration of tamoxifen (Sigma-Aldrich INC, St Louis, MO) using 100 mg kg−1 (in corn oil) once daily for four consecutive days. Mice were euthanized when their body condition score was <1.5 or when they experienced body weight loss ≥20% or signs of distress, such as difficulty breathing or urinary obstruction, except NP mice which were dissected one year after induction.

6–8-week-old male athymic nude-Foxn1^nu mice used for intracardiac studies were purchased from Envigo (Boyertown, PA, RRID:IMSR_ENV:HSD-069) and NOD-SCID mice (NOD.CB17-Prkdcscid/J, strain 001303, RRID:IMSR_JAX:001303) used for intratibial studies from Jackson Laboratories (Bar Harbor, ME).

Immunohistochemistry

For histological and immunohistochemical analyses, tissues were fixed in 10% formalin (Fisher Scientific Company, Pittsburgh PA). Bones were decalcified for 3 weeks in 15% EDTA (pH 7.0) solution (Sigma-Aldrich). Histopathological and immunohistochemical analyses were conducted on 3-μm paraffin sections. Images were captured using an Olympus VS120 whole-slide scanning microscope. Immunohistochemistry was performed with citrate-based antigen retrieval (Vector Labs, Newark, CA, H-330) using the VECTASTAIN^® Elite^® ABC HRP Kit (Vector Labs, PK-6100). Antibodies used were anti-human ATAD2 ab244431 1/250 (Abcam Inc, Waltham MA, RRID:AB_3674856) and anti-mouse ATAD2 cs50563 1/200 (Cell Signaling Technology Inc, Danvers MA, RRID: AB_3674855).

Cell lines

22Rv1 (ATCC, Manassas, VA, CRL-2505, RRID:CVCL_1045), LNCaP (ATCC, CRL-1740, RRID:CVCL_0395), PC-3 (ATCC, CRL-1435, RRID:CVCL_0035) and DU145 (ATCC, HTB-81, RRID:CVCL_0105) cells were grown and maintained in RPMI 1640 (ATCC, 30–2001) supplemented with 10% FBS (Sigma-Aldrich), whereas HEK-293FT (Invitrogen Waltham, MA, R700–07, RRID:CVCL_6911) were cultured in DMEM-10% FBS (Fisher). Cell lines were authenticated by STR profiling, passaged twice-weekly, used within 30 passages and tested negative for Mycoplasma using Universal Mycoplasma Detection Kit (ATCC #30–1012K). Clonal cell derivatives capable of target gene overexpression by CRISPRa were generated as previously described (17).

Mouse NP and NPK cell lines were derived from advanced primary tumors of their respective genotypes as described previously (15, 38) and grown in RPMI 1640 supplemented with 10% FBS.

RNA extraction and qRT-PCR analysis

RNA from cell lines was extracted with TRIZOL reagent (Thermo Fisher, 15–596-026), reverse-transcribed with SuperScript^™ III First-Strand Synthesis SuperMix (Thermo Fisher 11–752-050), and RNA expression measured by quantitative real time PCR (qRT-PCR) using the QuantiTect SYBR Green PCR kit (Qiagen, Germantown, MD, 204145). Sequences of all primers used in this study are provided in Supplementary Table S2.

Western Blotting

Total protein extracts were prepared by lysis of cells with 1X radioimmunoprecipitation assay (RIPA) buffer (0.1% SDS, 1.0% deoxycholate sodium salt, 1.0% Triton X-100, 0.15 M NaCl, 10 μmol/L Tris-HCl (pH 7.5), 1 mmol/L EDTA) with fresh 1% protease inhibitor (#1697498; Roche Basel), PMSF (10837091001, Sigma-Aldrich) and 1% phosphatase inhibitor (#P2850; Sigma-Aldrich). Protein lysates (20 μg per lane) were resolved by SDS page, followed by immunoblotting with primary antibodies anti-human ATAD2 ab244431 1/250 (Abcam), anti-mouse ATAD2 cs50563 1/500 (Cell Signaling), anti-MYC ab32072 1/1000 (Abcam, RRID:AB_731658) and anti-B-actin cs4970 1/20000 (Cell Signaling, RRID:AB_2223172), anti-GAPDH cs5174 1/1000 (Cell Signaling, RRID:AB_10622025), secondary antibodies (anti-Rabbit IgG HRP Sigma NA934 1/10000, RRID:AB_2722659) and visualized using an ECL Plus Western Blotting Detection Kit (Fisher RPN2232). X-ray films were developed and scanned, or membranes scanned with an iBright 1500 imager (Invitrogen). Band intensities were quantified relative to B-actin or GAPDH using ImageJ 1.54g software (RRID:SCR_003070).

Description of human patient cohorts and datasets.

All patients consented before inclusion and all studies using human tissue specimens were performed according to protocols approved by the Human Research Protection Office and Institutional Review Board at CUIMC. Only male patients were involved as the focus of our study was prostate cancer. Anonymized human tissue specimens were obtained from Columbia University Irving Medical Center’s (CUIMC cohort) Molecular Pathology Shared Resource of the Herbert Irving Comprehensive Cancer Center, as two sets of samples from independent patients. The first set (15, 17) consisted of five bone metastatic resections and ten primary prostate cancer tumors (Gleason score 9) from surgical resections of patients with advanced prostate cancer that were used for RNA isolation and qPCR studies. The second set consisted of 10 radical prostatectomy samples and 4 bone metastasis biopsies from patients with prostate cancer, used for IHC studies. RNA was extracted using miRNeasy mini kit (Qiagen) and RT–qPCR was performed using the QuantiTect SYBR Green PCR kit (Qiagen).

Analysis of ATAD2 expression in human prostate cancer tumors and metastases was also performed by staining of two independent tissue microarrays (TMA). These studies were approved by the Institutional Review Board of Memorial Sloan Kettering Cancer Center or University of Michigan. The HICCC TMA was previously described (39) and consisted of 338 samples that were stained and scored. These included 110 benign tissues and primary tumors of Gleason Scores 6 (n=111), 7 (n=69), 8 (n=30) and 9 (n=18). The UW TMA was obtained through the Prostate Cancer Biorepository Network (PCBN) and consisted of 51 bone and 70 visceral metastases obtained from 45 patients at rapid autopsy at the University of Washington. 50 bone and 65 visceral samples were successfully stained and scored, each represented by three cores, and the average of the three cores was considered for every sample. Immunostaining was quantified by the percentage of positively stained nuclei.

Two independent cohorts with clinical outcome data retrieved from the Decipher GRID registry, the MAYO cohort, GSE62116 (40) and the JHMI cohort, GSE79957 (41) were used for metastasis-free survival analyses. Patients in the MAYO cohort (n = 235) had undergone radical prostatectomy between 2000 and 2006; median follow-up was 7 years with 73 patients developing metastasis. The JHMI cohort is a case-cohort of 260 men who had undergone radical prostatectomy between 1992 and 2010 at intermediate or high risk and received no additional treatment until the time of metastasis; median follow-up was 9 years with 99 patients developing metastasis. Both cohorts were profiled on a Human Exon 1.0 ST Array and hybridization was conducted in a Clinical Laboratory Improvement Amendments-certified laboratory facility (Veracyte Inc, San Diego, CA). For comparison of ATAD2 mRNA expression in primary and metastatic prostate cancer, normalized microarray expression values were downloaded from GEO (GSE35988 (42) and GSE3325 (43)) and analyzed with two-tailed t-test.

Correlation of ATAD2 expression and MYC signaling was performed using data downloaded directly from cBioportal (44) using the TCGA firehose legacy (45) (n=491 primary tumors), SU2C/PCF dream team (SU2C (46) polyA cohort, n= 266 metastases) and FHCRC (47) (n= 149 metastases) datasets. Z-score sums from genes in the Myc Hallmarks v1 gene set from MSigDB (https://www.gsea-msigdb.org/gsea/msigdb), were used to estimate MYC pathway levels.

Lentivirus production and transduction

HEK-293FT cells were plated at ~90% confluency into 150mm dishes 24 hours before transfection, in DMEM-10%FBS culture media. 30ug of lentiviral vector (Sigma-Aldrich) along with 24ug psPAX2 (Addgene Watertown, MA, RRID: Addgene_12260) and 12ug pMD2.G (RRID: Addgene_12259) packaging plasmids were diluted in OPTIMEM-I media (ThermoFisher) and mixed with 60ug/ml polyethilenimine (Sigma-Aldrich) for transfection into HEK-293FT cells. After overnight incubation, the media was replaced with DMEM-1%FBS for 48hours, centrifuged at 300g for 5 minutes and filtered through a 0.45um filter to obtain the virus supernatant that was mixed with 0.8 ug/ml polybrene (Sigma-Aldrich) to infect target cells. Tissue culture media was changed 48 hours after viral infection and supplemented with 2 ug/ml puromycin for 72 hours. shRNA vectors used were pLKO.1 (Sigma-Aldrich): non-targeting control SHC002, shMYC#1 TRCN0000039640, shMYC#2 TRCN0000039642, shATAD2#1 TRCN0000161812, shATAD2#2 TRCN0000159158, shATAD2#3 TRCN0000158789 and shAtad2 (mouse) TRCN0000098627. CRISPRa was performed in 22Rv1 cells as previously described (17), using pHRdSV40-dCas9–10xGCN4_v4-P2A-BFP (RRID: Addgene_60903) and pHRdSV40-scFv-GCN4-sfGFP-VP64-GB1-NLS (RRID: Addgene_60904) plasmids to derive clonal derivatives showing high and homogenous levels of gene activation. sgRNA sequences targeting ATAD2 (or non-targeting controls) were cloned into BstXI and BlpI (NEB, New England Biolabs, Ipswich, MA) digested pU6-sgRNA EF1Alpha-puro-T2A-BFP vector (RRID: Addgene_60955) by oligonucleotide annealing and ligation as previously described (17). Sequences of all sgRNAs used in this study are provided in Supplementary Table S2.

Colony Formation Assays

One-thousand human or 200 mouse cells were plated in triplicate in 6-well plates and allowed to grow for two weeks before fixation in 10% formalin and staining with 0.5% crystal violet. Colonies were quantified with ImageJ software (https://imagej.nih.gov/ij/) using the ‘Analyze particle’ tool with size 50-infinity and circularity 0.4 – 1.0. Assays were performed with a minimum of two independent biological replicates.

Intratibial and intracardiac implantation studies

For monitoring tumor growth in bone, PC-3 cells (1 × 10⁶ cells in 10 μl of PBS) were injected into the tibiae of male NOD-SCID mice. A small longitudinal skin incision was made across the knee capsule, the tip of a scalpel was used to drill a hole into which cells were injected, sterile surgical bone wax (CP Medical, Norcross, GA) was used to seal the hole and the skin was then closed with wound clips. Tumor growth was monitored biweekly by bioluminescence imaging using an IVIS Spectrum Optical Imaging System (Perkin Elmer, Waltham, MA), following intraperitoneal injection of 150 mg /kg d-luciferin (PerkinElmer). Images were quantified using Living Image Software (PerkinElmer).

For intracardiac metastasis assays, mouse NPK bone metastatic cells (15) (1 × 10⁵ cells in 100 μl of PBS) were injected percutaneously into the left heart ventricle of nude mice (male, Taconic) and mice were euthanized 12–14 days after injection. At the time of sacrifice, YFP-positive tumors and metastases were visualized by ex vivo fluorescence using an Olympus SZX16 microscope (Ex490–500/Em510–560 filter).

Organ-on-a-chip Invasion Assay

Engineered bone tissues were made from bovine calf metacarpals (Lampire Biological Laboratories, 19D24003) that were sectioned into rectangular scaffolds (4 mm wide x 8 mm long x 1 mm thick) and fully decellularized as detailed previously (48, 49) The decellularization process removed all bovine cellular components, leaving just the bone extracellular matrix and bone architecture. The bones were lyophilized and the only bone scaffolds weighing between 12 and 18 mg/scaffold were included, resulting in the starting material with uniform porosity for engineering bone tissue. After sterilization in 70% ethanol overnight and 24 hours of incubation in Dulbecco’s Modified Eagle Medium (DMEM) the bone scaffolds were seeded with human cells.

Human bone-marrow derived mesenchymal stem cells (MSCs) (Lonza) were infused into the bone scaffolds (4 × 10⁵ cells per scaffold) by suspending the cells in 40 μL of medium (DMEM supplemented with 10% (v/v) HyClone fetal bovine serum (FBS), 1% penicillin/streptomycin, and 1 ng/mL of basic fibroblast growth factor, bFGF), and letting the cells attach for 2 hours before adding additional media (2 mL per well). To support the differentiation of the MSCs into osteoblasts within the engineered bone matrix, the MSC seeded bone was cultured in osteogenic medium consisting of low glucose DMEM supplemented with 1 μM dexamethasone (Sigma-Aldrich), 10 mM β-glycerophosphate (Sigma-Aldrich), and 50 μM L-ascorbic acid-2-phosphate (Sigma-Aldrich).

The osteogenic differentiation process continued for three weeks, with media changes three times a week. To create osteolytic bone, we then infused primary monocytes into the osteoblastic bone scaffolds (49). CD14+ monocytes were obtained by negative selection (EasySep Human Monocyte Isolation Kit, Stem Cell Technologies, 19359) from peripheral blood mononuclear cells (PBMC) isolated from buffy coats of human blood (fully deidentified samples obtained from the New York Blood Center) via density gradient centrifugation with Ficoll-Paque PLUS (GE Healthcare, 17–1440-02). Purified monocytes were seeded into the engineered bone tissues (4 × 10⁵ cells per scaffold) by suspending the cells in 40 μL of osteolytic medium for two hours. The engineered bone tissues were then cultured for a week in 2 mL of osteolytic media (Minimum Essential Medium Eagle Alpha modification (α-MEM, Sigma, M4526) supplemented with 10% (v/v) heat-inactivated HyClone FBS, 1% penicillin/streptomycin, l-Glutamine (Gibco, 25030–081), 20 ng mL⁻¹ Recombinant Human M-CSF (PeproTech, 300–25) and 40 ng mL⁻¹ Recombinant Human sRANK Ligand (PeproTech, 310–01)), with media changes and fresh cytokines every three days.

The engineered bone tissues were then placed into a chip designed for inter-organ communication that we recently developed (48). Briefly, engineered vascular barriers are formed by seeding 1.5 × 10⁵ MSC cells and 4 × 10⁵ human umbilical venous endothelial cells (HUVEC) on custom made transwell inserts. After the cells attached to the transwell barrier and formed a confluent monolayer, they were placed into the 1-tissue multi-organ chip and exposed to increasing levels of shear stress (0.31 dynes cm⁻² for 12 hours, 0.63 dynes cm⁻² for 24 hours, 1.88 dynes cm⁻² for 24 hours). The ramping of shear creates a confluent and quiescent vascular barrier lining the bottom of the chamber with engineered bone. After the engineered bone is added to the chamber (directly above the vascular barrier), circulating primary tumor cells were introduced into the vascular reservoir and allowed to circulate underneath the vascular barrier at a hydrodynamic shear stress of 1.88 dynes cm⁻²).

The cancer cells were labelled (LuminiCell Tracker 670, Sigma-Aldrich) to enable downstream tracking and allowed to circulate for 4 weeks. Media changes occurred every other day by replacing 1 mL of medium from the vascular reservoir with fresh vascular medium (EGM-2, Lonza) and 1 mL of medium from the engineered bone tissue reservoir with osteolytic medium specified above.

Intravasation of circulating cancer cells within the bone tissues was tracked using the IVIS Spectrum Optical Imaging System (Perkin-Elmer) in Columbia University’s Oncology Precision Therapeutics and Imaging Core. The chips with engineered tissues were placed next to one another in the same field of view and imaged using an IVIS 200 Spectrum device. The corresponding IVIS Spectrum software (Perkin-Elmer) was used to analyze the images by converting the signal to the normalized Radiant Efficiency (Emission light [photons/sec.cm² str]/ Excitation light [μW/cm²]). We measured the fluorescence of the labelled cancer cells within the engineered bone tissues by selecting the same region of interest for each tissue and quantifying the sum of the radiant efficiency of all fluorescent pixels within the region of interest.

Statistical analyses

Fisher’s exact test was used to compare two groups of categorical data, two-tailed t-test or one-way ANOVA (with Dunnett’s multiple comparisons) were used to compare two or more groups of numerical data after testing for normal distribution using D’Agostino & Pearson or Kolmogorov-Smirnov tests. Non-parametric tests were used if a normal distribution test was not passed, as stated in the figure legends. Kaplan–Meier survival analysis was conducted using GraphPad Prism software and analyzed with log-rank test. Correlation studies were performed with Spearman’s test using a two-tailed p-value. For all box plots, boxes show the 25th–75th percentile, center lines show the median and whiskers show the minimum–maximum values. In all bar graphs and dot-plots means are represented and error bars show the standard deviation. GraphPad Prism software (v.10.0) (RRID:SCR_002798) was used for statistical calculations and data visualization. BioRender license was issued to the Department of Oncological Sciences at Mount Sinai School of Medicine.

Data availability

The data generated in this study are available within the article and its supplementary data files. Expression profile data analyzed in this study were obtained from Gene Expression Omnibus (GEO) at GSE35988, GSE3325, GSE62116, GSE79957, GSE178869, GSE117430 or using cBioportal (44).

Results

ATAD2 is progressively expressed in metastatic prostate cancer.

We first evaluated ATAD2 expression in primary tumors from different PCa GEMMs of varying metastatic potential, in which tumors are induced in luminal cells of the adult mouse prostate by virtue of the Nkx3–1^CreERT2 allele (50). Immunohistochemical (IHC) staining of indolent tumors (from Nkx3–1^CreERT2; Pten^flox/flox, termed “NP” mice) showed low levels of nuclear ATAD2 staining (mean= 4.9%, n= 6), as shown in Figure 1A and Supplementary Table S1. In contrast, tumors from highly metastatic “NPK” (NP-Kras^LSL-G12D) mouse models (that include ~45% penetrance to bone(15)) showed significantly higher levels of ATAD2 expression, in both primary tumors (mean= 34.3%, n= 9, P= 0.0002) and unmatched metastases (mean= 36.1%, n= 5, P= 0.0006). Interestingly, when NPK tumors were harvested early during prostate cancer progression (16) (“Early Prostate” in Figure 1A, ie 1 – 2 months after tumors induction) before any evidence of established metastasis, ATAD2 expression was low as in indolent tumors (mean= 4.6%, n=4). This upregulation of ATAD2 expression upon progression of this highly metastatic model coincides with histological transition to carcinoma and with MYC overexpression, (previously reported in (15)) as illustrated in Supplementary Figure S1A. Thus, the correlation of ATAD2 expression with metastatic propensity in different PCa GEMMs suggests that ATAD2 may play a functional role in the development of metastasis.

Figure 1: ATAD2 is expressed in human and mouse metastatic prostate cancer and associated with the development of metastasis.

A) Schematic diagrams of different GEMMs that recapitulate the progression of prostate cancer (green) from indolent (non-lethal) NP tumors, early pre-metastatic NPK tumors (ie, before development of metastasis) and late, highly metastatic tumors that include metastasis to bone. Representative histological sections stained by H&E and immunohistochemistry for ATAD2 are shown, including a bone metastasis from a late NPK mouse. The region delimited by boxes is amplified in the inset to show nuclear staining. Scatter plots show the quantification of ATAD2 nuclear positivity in tumors from each category, with P values for one-way ANOVA with Dunnett’s multiple comparisons test compared to NP. B) Representative histological sections stained by H&E and immunohistochemistry for ATAD2 on whole tissue primary tumors obtained from radical prostatectomies as well as metastases from bone biopsies. C) ATAD2 IHC on tissue microarray (TMA) cores from a representative negative primary tumor and a metastasis with high ATAD2 expression. The region delimited by boxes is amplified to show nuclear staining. D) Stacked bar graphs showing quantification of ATAD2 positivity in different TMA tissues. The P value shown is for Fisher’s exact test comparing ATAD2 positivity (using a 1% expression cutoff) in primaries versus metastases. E,F) Box plots showing relative expression of ATAD2 in primary tumors and metastasis from three different patient cohorts. E) CUIMC cohort analyzed by qRT-PCR. F) Publicly available datasets from microarray expression data. P values in E,F are for unpaired, double sided T-test. G) Kaplan-Meier metastasis-free survival curves of ATAD2 expression stratified by quartiles in the Mayo (left) and JHMI (right) cohorts of primary tumors with prolonged follow-up data. P values were estimated using a log-rank test. Scale bars represent 50um. Mouse diagrams were created with BioRender.com.

We next examined the protein expression of ATAD2 in clinical specimens by IHC in several patient cohorts. Using whole tissue sections from 10 radical prostatectomies and 4 bone biopsies obtained at CUIMC, we observed nuclear ATAD2 staining to be mostly negative in primary tumors (<0.1% positivity). In contrast, three out of four bone metastases showed strong and high (>10%) nuclear positivity in tumor cells (Figure 1B). We used tissue microarrays (TMA) to quantify ATAD2 expression in tissues spanning the progression of PCa, as shown in Figure 1C. This included the ‘HICCC TMA’ (39) containing 76 benign tissue samples (with 110 cores) and 92 primary tumors (with 228 cores, Gleason scores 6 to 9) as well as the ‘UW TMA’ containing 115 metastases from 45 patients in a rapid autopsy program at University of Washington (containing three cores each of 50 bone and 65 visceral metastases). Table 1 summarizes the histoclinical characteristics of these samples. Strikingly, ATAD2 staining was negative in all 338 localized (benign and malignant) samples whereas over a third of metastases (37%, 43/115) showed strong nuclear positivity in at least 1% of cells, thus confirming a significant enrichment in expression in metastases compared to localized tumors (P<0.0001, Fisher’s exact two-tailed test, Figure 1D). Additional IHC images representative of different ATAD2 positivity scores are shown in Supplementary Figure S1B–D. Of note, we did not detect differences in expression between different metastatic sites, or association with survival outcomes or response to standard of care treatments (androgen depravation (ADT), enzalutamide or abiraterone), as shown in Supplementary Figure S1E.

Table 1:

Characteristics of patients in TMA Cohorts

TMA:	HICCC	UW
Clinical Parameter	All patients (n=92)	All patients (n=45)	ATAD2 Negative (n=25)	ATAD2 Positive (n=20)	P value
Gleason score no. (%)	of tissue cores* (n=228)	of primary tissue at diagnosis**
6	111 (49)	2 (5)	0 (0)	2 (11)	0.5307a
7	69 (30)	16 (37)	9 (38)	7 (37)
8	30 (13)	7 (16)	4 (17)	3 (16)
9–10	18 (8)	18 (42)	11 (46)	7 (37)
PSA at diagnosis (ng/ml) (median, range)	8.1 (0–118)	11.3 (0.68–1316)	17.4 (0.68–701)	9 (4.4–1316)	0.2955b
Age at diagnosis (years) (median, range)	61 (40–74)	63 (43–77)	63 (43–77)	63 (45–75)	0.9026c

^*malignant tissue cores at radical prostatectomy

^**Gleason score information was missing from 2 patients

^atwo-sided Fisher’s exact test comparing Gleason scores ≤7 vs >8;

^btwo-tailed Mann Whitney;

^ctwo-tailed T test

To evaluate the expression of ATAD2 mRNA in human prostate cancer, we performed qPCR analysis of a patient cohort using RNA extractions from 10 primary tumors and 5 bone metastases (CUIMC cohort, independent from samples used for IHC analysis), as well as analysis of two publicly available datasets (42, 43). These findings showed that ATAD2 mRNA levels were significantly enriched in metastases in all of these cohorts (P=0.0127 for CUIMC, P<0.0001 for GSE35988, ‘Grasso et al’ (42) and P=0.0012 for GSE3325, ‘Varambally et al’ (43), Figures 1E,F).

To ask whether ATAD2 is significantly associated with risk of metastasis, we used two independent prostatectomy cohorts with extensive clinical outcome data (MAYO (40) and JHMI (41); Figure 1G). Patients in MAYO (n = 235) had undergone radical prostatectomy between 2000 and 2006 with a median follow-up of 7 years during which time 76 patients developed metastasis. Patients in JHMI (n = 260) had undergone radical prostatectomy between 1992 and 2010 with a median follow-up of 9 years and 99 patients developed metastasis. Indeed, patients with tumors that scored in the 4^th quartile of ATAD2 expression (ie, the top 25%) showed a significantly worse metastasis-free survival compared to the rest of the patients (P<0.0001 and P=0.0023 for MAYO and JHMI, respectively, Figure 1G). Overall, ATAD2 expression in prostate cancer patients is overexpressed in metastases compared to primary tumors, and expression in primary tumors is significantly associated with metastasis survival.

ATAD2 expression may be regulated by MYC.

MYC signaling is among the most commonly activated pathways in bone metastasis compared to primary tumors in both the NPK model (15) as well as in human tumors (15, 37). Both ATAD2 and MYC reside near each other in a chromosomal location that is frequently amplified in prostate and other cancers. Moreover, ATAD2 has been previously shown to interact with MYC and modulate its transcriptional activity (31). We thus studied the relationship between ATAD2 expression and MYC pathway activation in several human datasets using cBioportal (44). ATAD2 expression was positively and significantly correlated with MYC pathway signaling in metastases from two datasets, SU2C (46) (Spearman’s R= 0.4072, P<0.0001, n= 266) and FHCRC (47) (R= 0.4810, P<0.0001, n= 149), but not in primary tumors from TCGA (n=491) (Figure 2A,B). Stratification of the latter cohort into quartiles showed that only the 4^th quartile of ATAD2 expression was positively and significantly correlated with MYC pathway signaling (R= 0.3982, P<0.0001, Figure 2A right), consistent with our previous results showing that high expression in primary tumors is associated with the development of metastasis (Figure 1G). This suggests that MYC activation may lead to ATAD2 expression in mCRPC. In support of this, using publicly available datasets from ENCODE (51) we show that the promoter of ATAD2 and an upstream cis-regulatory region are both enriched for the H3K27Ac histone mark, suggestive of active promoters/enhancer regions, and bound by MYC in prostate cancer cell lines, (22Rv1 (52) and LNCaP (53), Figure 2C).

Figure 2: Association between MYC and ATAD2 expression.

A) Correlation between ATAD2 expression and MYC pathway activation in two metastatic (SU2C, n=266 and FHCRC, n=149) and one primary (TCGA, n=491) prostate cancer patient cohorts. MYC pathway activity was estimated based on the sum of Z-scores from genes in the MYC Hallmarks v1 gene set of MYC targets. R and P values are derived from Spearman’s test and the linear regression line is shown. The fourth graph on the right depicts linear regression lines on the TCGA cohort subdivided by quartiles (Q1-Q4) of ATAD2 expression, and R values are shown below for each quartile. B) Bar graph summarizing correlations between ATAD2 and MYC mRNA to MYC pathway in the three datasets. Only significant correlations (P<0.01) are shown. C) Capture of the UCSC genome browser (GRCh33/hg38) showing ~60kb region on human chromosome 8 around ATAD2 promoter (y axis = reads per kilobase per million reads). UCSC genes (blue) at the top. Putative ENCODE candidate Cis-Regulatory Elements (cCREs) are shown at the bottom, with color codes according to ENCODE classification of regulatory signatures: red= promoter-like, orange = proximal enhancer-like and yellow = distal enhancer-like. Embedded ChIP-seq tracks were obtained from Holmes et al for 22Rv1 (H3K27ac, MYC and input) and from Augello et al for LNCaP (MYC) cell lines. Orange arrows point to two putative MYC regulatory elements in the ATAD2 promoter region. D) Western blot analysis of ATAD2 expression upon MYC knockdown in human LNCaP and PC-3 cells. The numbers on top of the ATAD2 bands show their expression relative to shControl and are summarized as % inhibition in scatter plots on the right. E) Heatmap summarizing qRT-PCR analysis of MYC, ATAD2 and three well-known MYC targets after MYC knockdown in LNCaP and PC-3 cells. Color scale depicts fold change downregulation compared to shControl. Only significant changes are colored (P<0.05 using one-way ANOVA adjusted for multiple comparisons with Dunnett’s test compared to shControl).

To functionally validate whether MYC regulates the expression of ATAD2, we used two shRNAs to knockdown MYC in four human prostate cancer cell lines. This resulted in downregulation of ATAD2 protein and mRNA levels in LNCaP and PC-3 cells (including well-known MYC targets, Figure 2D,E) but not in 22Rv1 and DU145 cells (Supplementary Figure S2A), suggesting that MYC may indeed regulate the expression of ATAD2 in a subset, but not all, of prostate tumors.

ATAD2 inhibition impacts prostate cancer proliferation and metastasis.

Having established that ATAD2 is overexpressed in advanced stages of prostate cancer, we set out to determine whether it would serve as a valid therapeutic target to diminish prostate cancer growth and metastasis. We first performed shRNA-mediated knockdown of ATAD2 to study its effects on tumor proliferation. Knockdown of ATAD2 using three shRNAs significantly impaired cell proliferation in four human prostate cancer cell lines compared to shControl, irrespective of AR status, including AR positive androgen dependent LNCaP cells, AR positive androgen independent 22Rv1 cells, as well as AR negative PC-3 and DU145 cells, as measured by colony formation assays (P<0.0001 one-way ANOVA with Dunnett’s multiple comparisons test, n=3 technical replicates) (Figure 3A and Supplementary Figure S3A). Next, we selected PC-3 cells because of their well-studied growth in bone and labeled them with GFP and Luciferase to facilitate in vivo tracking of tumor growth. We then injected shControl and shATAD2-treated cells into the bones (tibae) of immunodeficient SCID mice to study tumor growth in bone. We chose shATAD2#2 since it shows modest effects on in vitro proliferation (Figure 3A), allowing us to collect enough cells for in vivo injection. Notably, knockdown of ATAD2 resulted in slower tumor growth in bones of mice compared to shControl (Figure 3B) as measured by longitudinal luciferase tracking, (P=0.0023, two-way ANOVA with Sidak’s multiple comparisons test to day 22, 2 replicate experiments, n=5 mice each), suggesting that ATAD2 inhibition may be a viable therapeutic strategy to impair tumor growth in bones.

Figure 3: ATAD2 knockdown impairs proliferation of prostate cancer cells.

A) Colony formation assays of LNCaP and PC-3 human cell lines after knockdown of ATAD2, stained with crystal violet. Asterisks denote P values for one-way ANOVA with Dunnett’s multiple comparisons test, n=3 technical replicates. **: P<0.001, ****:P<0.0001. B) Intratibial growth assays of PC-3 cells labeled with GFP-Luciferase (2 replicate experiments, n=5 each). Representative IVIS bioluminescence images of shControl and shATAD2 tumors after intratibial injection, along with tumor growth curves quantified by longitudinal bioluminescence imaging. P-value shown for day 22 was estimated by two-way ANOVA with Sidak’s multiple comparisons against shControl. C) Representative H&E, ATAD2 and MYC IHC staining of intratibial PC-3 grafts treated with shControl or shATAD2. Scale bars represent 200um. Scatter plots on the right show quantification of percent tumor nuclei positive for ATAD2 or MYC, with P values derived from two-tailed unpaired t-test and n= 3.

Histological analysis of the tumor lesions in bone confirmed sustained ATAD2 knockdown in shATAD2 treated tumors compared to shControl (P= 0.0081 by two tailed t-test, n=3 tumors, Figure 3C). Interestingly, we also observed significant downregulation of MYC expression in shATAD2 intratibial xenografts (P= 0.0001 by two tailed t-test, n=3, Figure 3C), suggesting that ATAD2 may in turn regulate MYC expression. We thus measured MYC expression and three well-known MYC targets in LNCaP and PC-3 cells treated in vitro with shATAD2. Whereas this resulted in a significant downregulation of MYC expression (Supplementary Figure S3B) and of its downstream targets (Supplementary Figure S3C) in LNCaP cells, loss of ATAD2 did not show evidence of diminishing MYC levels or transcriptional activity in PC-3 cells, showing discrepancies between the in vivo and in vitro effects in this cell line that would be interesting to understand in future studies.

Given its significant association to metastasis in human prostate cancer specimens, we next examined whether ATAD2 could play a functional role in the development of bone metastasis. First, we used state-of-the-art engineered human tissue on a chip to model invasion of tumor cells across a vascular barrier and into an engineered bone (Figure 4A,B). As we have previously described (17, 48), tumor cells are introduced into a circulating flow that is separated by a vascular barrier from a chamber containing engineered bone. Cells are cultured in this way for 4 weeks, allowed to migrate through the endothelial lining into the bone chamber, and the resulting invading cells are quantified by fluorescence emission (Figure 4A). We first knocked-down Atad2 in our highly bone metastatic NPK cell lines and assessed the effects on invasion into engineered bone. Compared to non-metastatic NP cells (15), which express low levels of ATAD2 (Figure 4B), NPK cell lines showed a significant increase in invasion to bone, which was completely abolished by knockdown of Atad2 resulting in comparable invasion to NP cells (P<0.0001 for NP and shAtad2, one-way ANOVA with Dunnett’s multiple comparisons test to NPK shControl, n=3) (Figure 4B).

Figure 4: ATAD2 is a driver of prostate cancer metastasis.

A) Schematic of the organ-on-a-chip assay. B) Scatter plots showing invasion into engineered bone of indolent mouse NP cells as well as metastatic NPK cells treated with shControl or shAtad2. Corresponding expression levels of ATAD2 are shown in western blots. P values are for one-way ANOVA with Dunnett’s multiple comparisons test, n=3 technical replicates in duplicate independent experiments. C) Intracardiac assay after ATAD2 knockdown in NPK cells. Scatter plots show the number of ensuing bone (left) and liver (right) metastasis upon shControl or shAtad2 treatment of NPK cells injected intracardially into the circulation of nude mice (P values from two-tailed t test n=23 in three independent experiments). Representative epifluorescence images of bones and livers are shown, where metastases are evident by YFP-fluorescence. The corresponding brightfield image of each organ is shown underneath, with opacity set to 20%. Scale bars represent 0.1cm. Representative H&E, YFP and ATAD2 IHC staining of bone metastases ensuing from intracardiac injection of NPK shControl cells are also shown. Scale bars represent 200um.

We also used CRISPRactivation (CRISPRa) (54) to overexpress ATAD2 from its endogenous locus in 22Rv1 cells and observed a significant increase in invasion to bone using two independent sgRNAs targeting ATAD2, compared to non-targeting sgControl cells (P=0.004 for sgATAD2#1 and P=0.0006 for sgATAD2#2, one-way ANOVA with Dunnett’s multiple comparisons test, n=3) (Supplementary Figure S4A).

In contrast with knockdown in NP cell lines (Supplementary Figure S4B), shAtad2 treatment in NPK cells resulted in reduced proliferation rates in vitro (P<0.0001, two-tailed test compared to shControl, Supplementary Figure S4C). Moreover, this markedly reduced MYC protein levels (P= 0.0279, two-tailed t-test, Supplementary Figure S4D), supporting our hypothesis that ATAD2 may regulate MYC expression at least in certain tumors.

In order to directly assess the effect of ATAD2 on metastasis, we treated our highly bone metastatic NPK cells with shControl or shATAD2 for 5 days and then performed experimental metastasis assays using intracardiac injections into the circulation of immunodeficient nude mice. Remarkably, ATAD2 knockdown in NPK tumors resulted in a significant reduction in the number of ensuing bone and liver metastasis (Figure 4C, P=0.0024 for bone and P=0.0277 for liver metastases, two-tailed t test, n=23 in three replicate experiments), establishing ATAD2 as a driver of prostate cancer metastasis. Representative epifluorescence images of bones and livers as well as histological confirmation of bone metastasis by H&E and IHC for YFP and ATAD2 are shown in Figure 4C. Of note, despite potent knockdown in shAtad2 (Figure 4B, Supplementary Figure S4D) 50% (3/6) of histologically examined bone metastases expressed high levels of ATAD2 comparable to shControls (Supplementary Figure S4E), implying that a subset of shAtad2 tumors have either escaped knockdown or regained ATAD2 expression and further supporting a role for ATAD2 in bone metastasis.

Discussion

The occurrence of metastasis is a crucial determinant of prognosis in prostate cancer as well as in many other solid tumors (1). Defining the molecular principles of this process is therefore an urgent unmet need that may have critical translational implications for the development of novel therapies and biomarkers. In this study we have leveraged mouse models of spontaneous prostate cancer progression to bone metastasis as well as multiple independent patient cohorts to establish a functional role for ATAD2 in metastasis. We have focused on bone, the most clinically relevant metastatic site in prostate cancer, by leveraging unique mouse models and state of the art tissue engineering technologies.

Our work shows that ATAD2 is overexpressed in advanced stages of prostate cancer progression and identifies MYC as a critical regulator of its expression in a subset of tumors. Since MYC overexpression may occur in early primary tumors (36) before ATAD2 expression is evident, other mechanisms may cooperate with MYC to drive ATAD2 expression in advanced tumors, including previously reported AR and E2F transcription factors (32, 33). One potential mechanism is co-amplification of both the MYC and ATAD2 genomic loci, given their proximity in chromosome 8q, a region frequently amplified in later stages of progression (46).

We further show that genetic downregulation of ATAD2 diminishes proliferation and metastasis in advanced prostate tumors, thus establishing ATAD2 as a promising therapeutic target for advanced disease. Future studies should address the feasibility of achieving this, for example, through recently developed small molecule inhibitors that specifically bind the ATAD2 bromodomain (55–58). We note that although our shRNA studies show that ATAD2 affects tumor proliferation in multiple prostate cancer cell lines, LNCaP cells are particularly sensitive to this inhibition. Given that shATAD2 causes a strong downregulation of MYC and, conversely, shMYC results in a strong downregulation of ATAD2 in this cell line, this positive feedback loop may perhaps explain the pronounced antiproliferative effects of ATAD2 knockdown in these cells. Indeed, a similar observation has been previously made in ovarian and breast cancer, where cell lines dependent on MYC were particularly vulnerable to ATAD2 knockdown (59). Alternatively, given the well-established androgen sensitivity of LNCaP cells and the previously reported modulation of AR signaling by ATAD2 in this context (32), further studies should address whether modulation of AR signaling is involved in preferential sensitivity to ATAD2 knockdown.

In most cancer types, ATAD2 overexpression occurs early in localized disease and is associated with poor prognosis (24, 26, 31, 34) through mechanisms involving TGF-B (27), E2F (60, 61), MYC (31), SOX10 (62) or CENPE (28). In prostate cancer, however, our study and those of others (29, 32, 34) suggest that ATAD2 overexpression occurs later in cancer progression, in a subset of advanced, metastatic tumors. Thus, understanding the common and tumor specific mechanisms of how ATAD2 impacts cancer progression in different cancer types or disease stages is an important area for future studies. Indeed, the few studies that have directly focused on how ATAD2 regulates metastasis have implicated TGF-B and CENPE pathways in metastasis to lung (27, 28). Our findings support the notion that, at least in certain tumor contexts, ATAD2 may also in turn regulate MYC signaling, as had previously been described (31), adding to modulation of chromatin accessibility and AR signaling (32, 34) as mechanisms whereby ATAD2 promotes prostate cancer progression.

We acknowledge that metastasis is a multistep process and that ATAD2 could play a role in multiple parts of the so-called metastatic cascade. In this manuscript, we have focused on studying its role in later stages of metastasis, once cells have entered the circulation and bone colonization. These critical steps are incompletely understood and may be particularly clinically relevant for the development of new therapeutic strategies targeting established metastasis, a currently unmet clinical need.

In summary, our manuscript provides the first evidence that ATAD2 is linked to metastatic progression in prostate cancer and proposes ATAD2 as a promising therapeutic target.

Implications:

Our study highlights ATAD2 as a driver of prostate cancer progression and metastasis and suggests it may constitute a promising novel therapeutic target.