Abstract
The retrotransposon-derived paternally expressed gene 10 (PEG10) protein is ordinarily expressed at high levels in the placenta. Recently, it was discovered that PEG10 isoforms promote the progression of prostate cancer to a highly lethal androgen receptor (AR)-negative phenotype. The presence of PEG10 in other subtypes of prostate cancer has not been explored and a utility for PEG10 overexpression has not been developed. Here, we found that in addition to AR-null disease, PEG10 was also expressed in prostate cancer with constitutively active AR-splice variants. A molecular genetic imaging strategy for noninvasive imaging of AR-splice variant prostate cancer was developed by utilizing the cancer specificity of the PEG10 promoter to drive the expression of reporter genes. Plasmid insertion of a PEG10 promoter sequence optimized for enhanced output upstream of a reporter gene allowed detection of prostate cancer by near-infrared and positron emission tomography imaging after systemic administration of the plasmid in vivo. PEG10 expressing subcutaneous xenograft and intratibial tumor models were imaged by both modalities using this molecular genetic imaging strategy. This study demonstrates a preclinical proof-of-concept that the PEG10 promoter is a powerful and specific tool that can be utilized for noninvasive detection of aggressive prostate cancer subtypes.
PEG10 is expressed by prostate cancer with constitutively active AR-splice variants that can be exploited for noninvasive molecular imaging of this aggressive prostate cancer subytpe.
Introduction
Prostate cancer is a prevailing disease affecting 1 in 9 men in the United States (1). Once prostate cancer has metastasized to the bone and soft tissue, few effective treatment options exist. Androgen deprivation therapy (ADT) is the clinical standard for the treatment of metastatic disease (2). ADT suppresses the endogenous production of androgen by the testes, leading to decreased signaling through the androgen receptor (AR). Signaling through the AR axis is required for the development and normal function of the adult prostate as well as the growth and survival of prostate cancer cells (3). Although initial response rates are high, all men will eventually fail ADT and develop castration-resistant prostate cancer (CRPC; ref. 4). Aberrant reactivation of the AR signaling axis, due to AR overexpression, constitutively active AR-splice variants and the endogenous production of testosterone by the cancer cell, is a salient feature of CRPC (5). The second-generation antiandrogens abiraterone and enzalutamide have recently demonstrated clinical advantages in patients with CRPC treated with second-line hormonal agents and docetaxel (6). Resistance to these therapies inevitably occurs and many men develop a highly lethal subtype of non-AR–driven disease that possess neuroendocrine differentiation commonly referred to as treatment-induced neuroendocrine prostate cancer (NEPC; refs. 7, 8). Treatment options for NEPC are limited, consisting of taxanes and platinum agents (7). Thus, there is an urgent need for the development of effective therapies for this subtype of drug-resistant disease.
The development of novel therapeutics and the appropriate tailoring of existing therapies for NEPC is hindered by the inability to accurately quantify disease burden by molecular imaging (9). Compared with prostate adenocarcinoma, visceral metastases are frequent in NEPC and no imaging modality has proven superior at imaging both osseous and visceral metastases in NEPC (10, 11). Positron emission tomography (PET) radiotracers targeting the cell surface protein prostate-specific membrane antigen (PSMA) will not work for imaging this disease subtype due to the absence of PSMA expression in NEPC (12–14). One imaging strategy for NEPC that is noninvasive and allows for the detection of cancer with high specificity and low background is molecular-genetic imaging. Molecular-genetic imaging relies on the transcriptional mechanics of the disease rather than the targeting of a cell surface antigen or a metabolic protein (15). In this strategy, the promoter of a cancer-specific gene is used to drive the expression of a reporter gene within the cancer cell. This promoter-reporter gene system is placed in a DNA plasmid that is administered locally or systemically. Once the plasmid is delivered to the cancer cell, the cancer-specific promoter guides the expression of reporter genes such as herpes simplex 1 thymidine kinase (HSV1-TK), which can be detected via PET imaging after the administration of a radiolabeled nucleoside substrate (15). Preclinical research using molecular-genetic imaging has previously been successful at detecting cancer models using cancer-specific promoters to drive reporter gene expression (16, 17). In addition, molecular genetic imaging can be further exploited for theranostic purposes by the expression of suicide genes such as cytosine deaminase (18). Plasmid-based therapeutics and their delivery agents are currently in clinical trials that makes molecular-genetic imaging potentially a translational tool (16, 17, 19).
Elevated paternally expressed gene 10 (PEG10) expression in prostate cancer (20) has been reported and recently its upregulation was discovered in the transdifferentiation of AR-driven prostate adenocarcinoma to non-AR–driven NEPC (21). PEG10 is a retrotransposon-derived gene primarily expressed in the placenta and is crucial for embryonic development (22). We investigated the expression of PEG10 in healthy prostate tissue, primary adenocarcinoma, metastatic CRPC, and AR-null prostate cancer. Our analysis discovered that PEG10 expression was not restricted to AR-null disease but was also found in a subset of AR-positive CRPC including disease expressing AR splice variants. Harnessing the overexpression of PEG10, we used the transcriptional mechanisms of PEG10 to create a molecular-genetic imaging tool for the detection of these highly lethal prostate cancer subtypes. The imaging strategy incorporated an optimized PEG10 promoter and a two-step transcriptional amplification element for enhanced output. Our final imaging construct allowed for the precise tumor detection in PEG10-positive CRPC animal models by near-infrared (NIR) and PET/CT imaging.
Materials and Methods
IHC
PEG10 IHC analysis was performed on a 120 Case High Grade Race Disparity tissue microarray (TMA) constructed from African American and Caucasian patients with informed written consent and on a LuCaP patient-derived xenograft tissue TMA acquired from the Prostate Cancer Biorepository Network (PCBN). The liver biopsy analyzed for PEG10 staining was acquired using a University of Minnesota Human Subjects Division approved IRB protocol for tissue acquisition (IRB#1604M86269) and with informed written consent from the patients. IHC was performed on the formalin-fixed paraffin-embedded tissue sections using (1:500) rabbit anti- PEG10 (Novus Biologicals, NBP2-13749) and (1:100) rabbit anti-androgen receptor SP107 (Sigma). Unstained sections (4 μm) were deparaffinized and rehydrated using standard methods. For antigen retrieval, slides were incubated in 6.0 pH buffer (Reveal Decloaking reagent, Biocare Medical) in a steamer for 30 minutes at 95–98°C, followed by a 20-minute cool down period. A serum-free blocking solution (Sniper, Biocare Medical) was placed on sections for 30 minutes. Blocking solution was removed and slides were incubated in primary antibody diluted in 10% blocking solution/90% TBST. The antibody was used according to the manufacturer's protocol.
Genomic analysis
Chromatin immunoprecipitation sequencing (CHiP-seq) data were obtained on the basis of previous literature (23). RNA-seq fragments per kilobase million (FPKM) values in Fig. 1B were obtained using the methods described in ref. 24. Expression values in Fig. 2E were obtained using the publicly available expression profiling by array GSE41784. Microarray data was extracted from previously published studies of a set of metastatic tumors from men with castration-resistant prostate cancer (Fig. 2I; ref. 25) and patient-derived xenograft (PDX) models of prostate cancer (Fig. 2F; ref. 26). Both datasets are available in the Gene Expression Omnibus under accessions GSE77930 and GSE93809.
PEG10 expression in primary prostate cancer tumors versus metastasis. A, IHC staining for PEG10 in primary prostate cancer tumors graded Gleason 4+3, 4+4, and 4+5. Scale bars, 200 μm. B, RNA-seq analysis of PEG10 expression in primary tumor versus metastasis. Multiple raw FASTQ RNA-seq datasets were obtained via DbGAP and aligned and transcripts quantified via a uniform pipeline, enabling cross-experimental comparisons. Data are reported as FPKM. C, Significance was determined by unpaired two-tailed Welch t test after a significant F-test and Grubb outlier test were used in PEG10 IHC staining in a case study of a patient with AR− liver metastasis. Scale bars, 4 mm, 60 μm, and 100 μm left to right. D, IHC staining of PEG10 in a case study of patient with AR+ adenocarcinoma graded Gleason 4+5. Scale bars, 60 μm. **, P < 0.01. Results are expressed in mean + SEM.
PEG10 expression in primary prostate cancer tumors versus metastasis. A, IHC staining for PEG10 in primary prostate cancer tumors graded Gleason 4+3, 4+4, and 4+5. Scale bars, 200 μm. B, RNA-seq analysis of PEG10 expression in primary tumor versus metastasis. Multiple raw FASTQ RNA-seq datasets were obtained via DbGAP and aligned and transcripts quantified via a uniform pipeline, enabling cross-experimental comparisons. Data are reported as FPKM. C, Significance was determined by unpaired two-tailed Welch t test after a significant F-test and Grubb outlier test were used in PEG10 IHC staining in a case study of a patient with AR− liver metastasis. Scale bars, 4 mm, 60 μm, and 100 μm left to right. D, IHC staining of PEG10 in a case study of patient with AR+ adenocarcinoma graded Gleason 4+5. Scale bars, 60 μm. **, P < 0.01. Results are expressed in mean + SEM.
PEG10 is present throughout diverse prostate cancer phenotypes. A, CHiP-seq analysis of the AR binding to PEG10. All cells were pretreated with CSS. The full-length AR-expressing R1-AD1 cells were treated with DHT for AR activation and AR variant–expressing R1-567 cells were treated with vehicle. B, PEG10 expression in LNCaP and CWR-R1 cells after treatment with CSS or CSS+R1881. The PEG10 expression for each cell line was normalized relative to their corresponding FBS controls. Significance was determined by two-tailed Student t test (n = 3). C, PEG10 mRNA in prostate cancer cell lines expressed relative to housekeeping gene 18S5. D, PEG10 staining in LuCaP PDXs where a is LuCaP 145.1, a model of NE+, AR− prostate cancer, b is LuCaP 86.2, an adenocarcinoma (AC) prostate cancer model, AR-V7+, and c is LuCaP 78, an AC model with only wild-type AR expression. Scale bars, 300 μm in full image and 60 μm in magnified. E, AR regulation of PEG10 expression in R1-AD1 cells. PEG10 expression levels are represented as an expression value of RNA analyzed by Illumina Beadchips. AR-Off cells were treated with siRNA targeting AR exon 1. AR-V On Only cells were treated with siRNA targeting AR exon 7. Significance was determined using the unpaired two-tailed Student t test (n = 3). F, PEG10 and ONECUT2 expression represented in a microarray heatmap of early passage LuCaP PDXs microarray. The samples are sorted by the adenocarcinoma or NE status. G, Microarray PEG10 expression represented as log2 median-centered ratio in adenocarcinoma and NE PDXs presented microarray (F). Significance was determined using Welch unpaired t test after using an F-test and Grubb test. H, Negative correlation of PEG10 and AR in LuCaP microarray in F. Pearson correlation test, P = 0.0085. Red line, linear regression. I, PEG10 and ONECUT2 expression represented in a CRPC microarray (n = 171 tumors from 63 men). Samples are sorted by the AR and NE status. J, PEG10 expression represented as log2 median-centered ratio in patient tumor samples represented in microarray (I) sorted by the AR and NE status. Significance was determined by one-way ANOVA with Dunnett posttest. K, Negative correlation between PEG10 and AR in CRPC microarray in I. Pearson correlation test, P < 0.0001. Red line, linear regression. *, P < 0.05; ****, P < 0.0001, n.s., not significant. Results in B and E are expressed as mean + SEM. Results in G and J are expressed minimum to maximum, with all points shown.
PEG10 is present throughout diverse prostate cancer phenotypes. A, CHiP-seq analysis of the AR binding to PEG10. All cells were pretreated with CSS. The full-length AR-expressing R1-AD1 cells were treated with DHT for AR activation and AR variant–expressing R1-567 cells were treated with vehicle. B, PEG10 expression in LNCaP and CWR-R1 cells after treatment with CSS or CSS+R1881. The PEG10 expression for each cell line was normalized relative to their corresponding FBS controls. Significance was determined by two-tailed Student t test (n = 3). C, PEG10 mRNA in prostate cancer cell lines expressed relative to housekeeping gene 18S5. D, PEG10 staining in LuCaP PDXs where a is LuCaP 145.1, a model of NE+, AR− prostate cancer, b is LuCaP 86.2, an adenocarcinoma (AC) prostate cancer model, AR-V7+, and c is LuCaP 78, an AC model with only wild-type AR expression. Scale bars, 300 μm in full image and 60 μm in magnified. E, AR regulation of PEG10 expression in R1-AD1 cells. PEG10 expression levels are represented as an expression value of RNA analyzed by Illumina Beadchips. AR-Off cells were treated with siRNA targeting AR exon 1. AR-V On Only cells were treated with siRNA targeting AR exon 7. Significance was determined using the unpaired two-tailed Student t test (n = 3). F, PEG10 and ONECUT2 expression represented in a microarray heatmap of early passage LuCaP PDXs microarray. The samples are sorted by the adenocarcinoma or NE status. G, Microarray PEG10 expression represented as log2 median-centered ratio in adenocarcinoma and NE PDXs presented microarray (F). Significance was determined using Welch unpaired t test after using an F-test and Grubb test. H, Negative correlation of PEG10 and AR in LuCaP microarray in F. Pearson correlation test, P = 0.0085. Red line, linear regression. I, PEG10 and ONECUT2 expression represented in a CRPC microarray (n = 171 tumors from 63 men). Samples are sorted by the AR and NE status. J, PEG10 expression represented as log2 median-centered ratio in patient tumor samples represented in microarray (I) sorted by the AR and NE status. Significance was determined by one-way ANOVA with Dunnett posttest. K, Negative correlation between PEG10 and AR in CRPC microarray in I. Pearson correlation test, P < 0.0001. Red line, linear regression. *, P < 0.05; ****, P < 0.0001, n.s., not significant. Results in B and E are expressed as mean + SEM. Results in G and J are expressed minimum to maximum, with all points shown.
Cell lines
Prostate cancer cell lines 22Rv1, DU145, PC3, and LNCaP were obtained from the ATCC and were maintained according to ATCC guidelines. MR42D cells were a gift from Dr. Amina Zoubeidi (Vancouver Prostate Centre, Vancouver, Canada) and maintained with 10 μmol/L enzalutamide. CWR-R1 cells were a gift from Dr. Scott Dehm (University of Minnesota, Minneapolis, MN). HT-29 cells were a gift from Dr. Hiroshi Hiasa (University of Minnesota, Minneapolis, MN). All cell lines were verified by short tandem repeat analysis and analyzed for Mycoplasma contamination prior to our studies.
R1881 Treatment of CWR-R1 and LNCaP cells
A total of 106 cells per well were plated in 6-well plates in full growth media and incubated overnight. Media were then changed to 10% charcoal-stripped serum (CSS)-DMEM and cells were incubated for 24 hours. Control cells were kept in FBS. Cells were then treated with the following: control: full growth (FBS) media/DMSO, CSS: 10% CSS media/DMSO, CSS + R1881: 10% CSS Media/10 nmol/L R1881 in DMSO. Cells were incubated for 24 hours and the same treatment was reapplied after 24 hours. The total incubation in R1881 was 48 hours, after which, cells were collected for RNA extraction.
Quantitative RT-qPCR
RNA was extracted from each cell line using 106 cells with a RNeasy Kit (Qiagen). RNA was converted to cDNA using the High Capacity RNA to cDNA Kit (Applied Biosystems). TaqMan RT-PCR was performed using the TaqMan Universal PCR Master Mix (Applied Biosystems) and the following TaqMan Gene Expression probes: PEG10; Hs00248288_s1, 18S5 ribosomal RNA; Hs03928985 for a normalization control, and a custom HSV1-TK probe (27). TaqMan probe in the supplemental was PEG10; Hs01122880. qPCR was performed on a StepOnePlus Real-Time PCR System Instrument (Applied Biosystems). Data was analyzed using the comparative Ct method (fold change = 2−ΔΔCt; ref. 28).
Plasmids
A pGL3 Basic vector (Promega, E1751) was used as the backbone for all of the cloning. Primers for the full-length PEG10 promoter cloning from PBMC cDNA and truncated 1 KB promoter were based on previous literature (Supplementary Table S1; ref. 29). MluI and XhoI were the cut sites used for the promoter insertion. The conventional single construct two-step transcriptional amplification was designed according to previous literature and synthesized by GenScript (30–32). The TSTA element was designed head-to-tail or “unidirectional.” The system can be used head-to-head or head-to-tail (bidirectional) according to patent US7527942B2. Both orientations have been tested in literature. Our GAL4-VP16 fusion protein consisted of GAL4 amino acids 1-147 and two consecutive VP16 domains, amino acids 413–456. Some literature uses VP16 amino acids 413–454. We used amino acids 413–456 based on literature that states that single most crucial aspect VP16 is located between residues 429 and 456 (32). A linker was used to fuse the two proteins (PEFLQPGGS). A pause site of 33 base pairs was placed downstream of the consecutive GAL4-VP16(x2) sequences (no linker was used between the consecutive VP16 sequences; ref. 33) followed by five GAL4 DNA-binding sites (cggagtactgtcctccg) each separated by two base pairs (ag; ref. 34). An adenovirus minimal promoter was placed 16 bp after the last GAL4-binding site (23 bp from GAL4-binding site to TATA box). The advanced two-step transcriptional amplification system was designed according to previous literature and synthesized by GenScript (31). Our A. TSTA is identical to our TSTA other than the addition of polyglutamines and rat glucocorticoid receptor protein between the GAL4-binding domain and VP16 sequences. A map of the PEG10 1KB promoter with the A.TSTA can be found in Supplementary Fig. S1. BglII and HindIII were used as the insert cut sites for the TSTA elements. To the best of our knowledge, this particular system may be unique due combining the polyglutamines and rat glucocorticoid receptor and using two VP16 domains rather than one VP16 like in the original advanced TSTA. For fluorescence studies, the firefly luciferase encoding gene in pGL3 basic was replaced with a near-infrared protein 682 (iRFP682). piRFP682-N1 was a gift from Vladislav Verkhusha (Addgene plasmid # 45459). For in vivo PET studies, the Luc gene was replaced with HSV1-TK (cloned from pLV-SFFV-HSVTK, Imanis Life Sciences, DNA1052). Correct insertion of DNA fragments was verified by Sanger sequencing and gel restriction analysis.
Luciferase assays
On day 1, cells were plated in 96-well plates (104 cell/well). On day 2, cells were transfected with 90 ng of experimental plasmid DNA and 9 ng of control pRL-TK per well with 0.24 μL GeneJuice (Millipore). Seventy-two hours posttransfection (or day 5), cells were lysed using the passive lysis buffer from Promega. Luciferase activity was quantified using the Dual-Luciferase Reporter Assay System (Promega). Each experimental firefly luciferase output (LUC) was normalized to its respective Renilla luciferase (REN) control output. Relative luciferase units (RLU) are LUC/REN.
In vitro iRFP imaging
On day 1, 104 cells were plated in a 96-well plate. On day 2, cells were transfected with plasmids containing the iRPF682 gene. Ninety nanograms of DNA and 0.24 μL of GeneJuice (Millipore) were used for the transfection. Cells were incubated for 72 hours and visualized on the Odyssey Infrared imaging system (LI-COR) using the 700 nm channel.
Xenograft models
All animal studies were approved by the University of Minnesota Institutional Animal Care and Use Committee. For the subcutaneous models, 3- to 4-week-old hsd:athymic mice were purchased from Envigo. For each mouse, 106 CWR-R1 or HT-29 cells were suspended in 200 μL of a 1:1 mixture of Matrigel (Corning) and 1× PBS. The cells were implanted subcutaneously into the flanks of the mice using a 25 gauge needle. The tumors were allowed to grow until visible by naked eye to start imaging experiments. To create the CWR-R1 intratibital model, 3- to 4-week-old hsd:athymic mice were purchased from Envigo. A total of 2.5 × 105 of CWR-R1 cells in 1× PBS were injected into the tibia of one leg. Tumors were allowed to form for three weeks before near-infrared imaging. The same tumors were used for PET imaging 5 weeks post intratibial injections.
Systemic in vivo DNA delivery
Plasmid DNA was prepared with EndoFree Plasmid Kit (Qiagen). Endotoxin level was ensured as <0.1E U/μg DNA. For the delivery of the plasmids in vivo, low molecular weight l-PEI–based cationic polymer, in vivo-jetPEI (Polyplus transfection) was used as gene delivery reagent for tail vein intravenous administration. A ratio of 6 was used for nitrogen to phosphate (N/P = 6) was used for all injections. Forty micrograms of plasmid DNA and 4.8 μL of 150 mmol/L in vivo-jet PEI were combined according to the manufacturer's instructions to form the DNA polyplex in a total volume of 400 μL for each mouse. Acute toxicity was observed initially when 200 μL was used as the final volume for each systemic tail vein injection. Increasing the volume to 400 μL decreased the acute toxicity (only 1 mouse was found with acute toxicity out of all of the experiments presented in this work). Plasmid DNA/PEI complex was delivered using a 26 gauge needle.
In vivo fluorescence imaging
Mice were imaged with the IVIS Spectrum (Caliper/Xenogen) at the University of Minnesota–University Imaging Center. Mice were placed on special low fluorescence diet, TC.97184 (Envigo). For each imaging session, mice were under a 2.0% isoflurane/oxygen mixture. For the filters, 675 nm/720 nm were used. Living Image 4.5 software was used for image acquisition and analysis. For region of interest (ROI) analysis, a circle of the same size was used on all mice and placed in regions of high fluorescence in each tumor.
PET/CT imaging and data analysis
For subcutaneous CWR-R1 model.
On day 1, mice were injected with the plasmid/PEI complexes (40 μg DNA). 72 hours post DNA delivery, 124I-FIAU (purchased from 3D imaging, Arkansas) was injected into the mice intravenously (150 μCi per animal).
For intratibial CWR-R1 model.
On day 1, mice were injected with the plasmid/PEI complexes. Forty-eight hours post DNA delivery, 124I-FIAU (purchased from 3D imaging, Arkansas) was injected into the mice intravenously (250 μCi per animal). Twenty-four hours post isotope injection, the animals were imaged using Siemens Inveon microPET/CT at the University of Minnesota–University Imaging Center. Acquisition time was 40 minutes. Animals were kept under 2% isoflurane/oxygen mixture throughout the duration of the scan. For the 2D image analysis, Inveon Research Software was used. AMIRA was used for 3D reconstruction.
Statistical analysis
Data analysis was performed on GraphPad Prism 7 (GraphPad Software Inc.). Quantitative PCR results were analyzed in Excel. Statistical significance was determined using the unpaired two-tailed Student t test, unpaired two-tailed Welch t test where the variances are shown to be different via F-test, one-way ANOVA, or two-way ANOVA. Only two-tailed tests were used. Results are depicted as mean + SEM unless stated otherwise. All P values of <0.05, <0.01, <0.001, and <0.0001 were considered significant. Pearson correlation coefficient was used to determine correlation between genes (Pearson ρ). The symbols used to represent the P values were: ns, nonsignificant, P > 0.05; *, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001; ****, P ≤ 0.0001. The test used in each statistical analysis is specified in the figure legends.
Data and material availability
All data associated with this study are present in the paper or Supplementary Materials.
Results
PEG10 expression is restricted to metastatic prostate cancer
IHC analysis was performed on a tissue microarray (TMA) of primary prostate adenocarcinoma obtained from African American and Caucasian patients. The antibody for IHC was validated using placenta as a positive control and healthy human tissue sections were stained to document specificity (Supplementary Fig. S2.). In the race disparity TMA, no positive staining for PEG10 was observed in any of the primary prostate adenocarcinoma cores (0/120) even across a spectrum of high Gleason scores. (Fig. 1A). Next, we analyzed publicly available RNA-seq datasets (24, 35, 36) to quantify PEG10 expression between hormone-naïve primary prostate tumors and metastatic tumors (Fig. 1B). Analysis of these data found a significant increase in PEG10 mRNA expression in metastatic disease compared with primary tumors (P < 0.01). We then analyzed a liver biopsy from a patient who demonstrated radiographic progression while on abiraterone in the presence of declining PSA. Here, strong PEG10 staining was observed in the liver biopsy while staining for the AR was completely absent (Fig. 1C). The original biopsy from the prostate of this patient prior to prostatectomy documented aggressive Gleason 4 +5 disease that was absent for PEG10, but positive for AR by IHC (Fig. 1D).
Regulation of PEG10 by full-length AR and AR variants
On the basis of previous reports, the lack of AR expression in neuroendocrine prostate cancer (NEPC) appeared to promote PEG10 expression (21). In the CWR-R1–derivative cell line R1-AD1 (23), ChIP-seq analysis demonstrated that the wild-type, androgen-activated AR was bound to the PEG10 gene. Conversely, insignificant binding to the PEG10 gene was observed in ChIP-seq data from vehicle-treated R1-AD1 cells and in R1-D567 cells that only express the constitutively active AR variant, ARv567es (Fig. 2A). We next used RT-PCR to test the androgen regulation of PEG10 mRNA in prostate cancer cells lines treated with the synthetic androgen R1881 (Fig. 2B). LNCaP cells, which express full-length wild-type AR, displayed an increase in PEG10 expression when grown in CSS. The addition of R1881 to the CSS media reduced the PEG10 expression in LNCaP back to baseline, indicating that active AR represses PEG10 expression. CWR-R1 cells, which express wild-type AR and AR splice variants (37, 38), did not display R1881-mediated repression of PEG10 mRNA. Our data indicate that the constitutive activity of AR variants expressed in CWR-R1 cells may be dominant over full-length AR in regulating PEG10 expression.
PEG10 expression is not confined to AR-negative prostate cancer
We discovered that cell lines previously reported to express elevated levels of PEG10 (DU145 and PC3) did not exhibit the highest expression levels out of the prostate cancer cell lines analyzed by qPCR (Fig. 2C). The qPCR data found that AR+ cell lines with a castration-resistant phenotype (39, 40), CWR-R1, 22Rv1, and MR42D, had elevated PEG10 expression levels. Both CWR-R1 and 22Rv1 cells express AR splice variants in addition to full-length AR (37, 39), while MR42D cells express wild-type AR but are indifferent to AR signaling (41). The analysis of cell line PEG10 expression was also used to determine models and controls for studies of PEG10 promoter activity.
IHC staining of PEG10 in a tissue microarray of LuCaP patient-derived xenografts (PDX) found strong straining in the PDXs representing AR-null NEPC, including LuCaP 145.1 (Fig. 2D). Strong staining was also observed in the AR splice variant–positive model LuCaP 86.2, while no staining for PEG10 was observed in the AR+ adenocarcinoma model LuCaP 78 (26). The IHC results of PDX staining for PEG10 support our hypothesis that there may be a path of alternative regulation of PEG10 expression by AR splice variants that results in elevated PEG10 expression. To test this directly, we investigated microarray data from CWR-R1 cells transfected with siRNA-targeting AR exon 7, which ablates expression of full-length AR but not AR-V7, or CWR-R1 cells transfected with siRNA targeting AR exon 1, which ablates expression of both full-length AR and AR-V7 (Fig. 2E). PEG10 expression was higher in CWR-R1 cells transfected with exon 1–targeted siRNA versus exon 7–targeted siRNA, indicating that AR-V7 activates PEG10 expression. Collectively, analysis of gene expression and IHC with PDX tissue support the hypothesis that lack of AR is not the only factor responsible for PEG10 upregulation and that AR-Vs may positively regulate PEG10 expression. To our knowledge, this is the first data demonstrating positive regulation of PEG10 expression by AR splice variants. Subsequently, a microarray of early-passage LuCaP PDX models was analyzed for PEG10 expression (Fig. 2F). PEG10 was elevated in NEPC PDX models that are AR− (Fig. 2G) P < 0.0001. However, based on the heatmap, it should be noted that some AR+ PDXs did in fact exhibit elevated PEG10 that may be due to the presence of AR splice variants (LuCaP 86.2 and 147; refs. 23, 42). Aside from these individual examples, overall the PDX microarray did show a negative correlation between PEG10 and AR (Pearson ρ = −0.5242), supporting previous literature that wild-type AR does in fact play some role in downregulating PEG10 expression (Fig. 2H).
PEG10 expression weakly correlates with downregulated AR in a patient microarray
We next studied PEG10 in gene expression microarray data of 171 CRPC tumors from 63 patients with metastatic CRPC (Fig. 2I; ref. 25). This DNA microarray demonstrated evidence of large variability in PEG10 expression among tumors that are AR+, NE−. PEG10 expression was elevated in AR+/NE+ and AR−/NE+ tumors compared with the AR+/NE− tumor population (Fig. 2J). This observation supports previous literature on PEG10 elevation in NEPC (21), but also shows that PEG10 expression can be high in AR+ tumors. AR−/NE, which has been termed double-negative CRPC, displayed less PEG10 expression than NE+ tumors. The metastatic CRPC microarray revealed that in clinical samples, the negative correlation of PEG10 and AR was slightly weaker (Fig. 2K) than in the LuCaP PDX models based on Pearson correlation coefficient (Pearson ρ = −0.4217). The correlation between AR and PEG10 represented by Pearson correlation coefficient supports that PEG10 expression correlates with AR downregulation but also supports our hypothesis that AR variants and/or other factors may play a role in regulating PEG10 expression.
ONECUT2 has a similar expression trend with PEG10
ONECUT2 was recently identified as a regulator of lethal prostate cancer by suppressing AR-dependent signaling (43, 44). The recent literature documents that PEG10 expression in the 22Rv1 in vivo model decreased after inhibition of ONECUT2 and increased in vitro after ONECUT2 overexpression in LNCaP and C4-2 cells. In line with this, microarray data from patients with CRPC demonstrated that elevation of ONECUT2 expression occurred in AR+, NE+ and AR−, NE+ tumor populations but not in double negative prostate cancer (Fig. 3A). The trend in expression was similar to that of PEG10. There was a positive correlation between PEG10 and ONECUT2 (Fig. 3B; Pearson ρ = 0.5292) and a negative correlation between AR and ONECUT2 (Fig. 3C; Pearson ρ = −0.4199). ONECUT2 expression in LuCaP PDX microarray was also elevated in NEPC compared with adenocarcinoma (Fig. 3D).
ONECUT2 correlates with PEG10 expression patterns. A, ONECUT2 expression represented as log2 median-centered ratio in the CRPC microarray (Fig. 2I) sorted by AR and NE status. Significance was determined by one-way ANOVA with Dunnett posttest. B, Positive correlation between PEG10 and ONECUT2 in patient microarray (Fig. 2I). Pearson correlation test, P < 0.0001. Red line, linear regression. C, Negative correlation between ONECUT2 and AR in mCRPC microarray (Fig. 2I). Pearson correlation test, P < 0.0001. Red line, linear regression. D, ONECUT2 expression represented as log2 median-centered ratio in adenocarcinoma and NE PDXs presented microarray (Fig. 2F). Significance was determined using Welch unpaired t test after using an F test and Grubb test. Significance in A was determined using the unpaired two-tailed. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001; n.s., not significant. Results in A and D are expressed minimum to maximum, with all points shown.
ONECUT2 correlates with PEG10 expression patterns. A, ONECUT2 expression represented as log2 median-centered ratio in the CRPC microarray (Fig. 2I) sorted by AR and NE status. Significance was determined by one-way ANOVA with Dunnett posttest. B, Positive correlation between PEG10 and ONECUT2 in patient microarray (Fig. 2I). Pearson correlation test, P < 0.0001. Red line, linear regression. C, Negative correlation between ONECUT2 and AR in mCRPC microarray (Fig. 2I). Pearson correlation test, P < 0.0001. Red line, linear regression. D, ONECUT2 expression represented as log2 median-centered ratio in adenocarcinoma and NE PDXs presented microarray (Fig. 2F). Significance was determined using Welch unpaired t test after using an F test and Grubb test. Significance in A was determined using the unpaired two-tailed. *, P < 0.05; ***, P < 0.001; ****, P < 0.0001; n.s., not significant. Results in A and D are expressed minimum to maximum, with all points shown.
The PEG10 promoter is a powerful transcriptional tool
After verifying the presence of PEG10 in lethal forms of prostate cancer, we decided to incorporate the transcriptional power of the PEG10 promoter in a molecular-genetic approach to image prostate cancer. The promoter was first optimized for enhanced transcriptional output. A full-length (∼2 kb) promoter (PEG102KB) and a truncated ∼1 kb promoter (PEG101KB) were cloned into a pGL3-Basic vector (∼4.8 kb) possessing a luciferase reporter gene. The transcriptional efficiency of the two promoter variants was initially evaluated in AR-null cell lines using the full-length AR cell line LNCaP as a control for low PEG10 expression (Fig. 4A). It was discovered that the 1 kb promoter was significantly stronger than the full-length PEG10 promoter in the three cell lines analyzed. Equal mass of each construct were used for transfection. Because the 1 kb construct was approximately 15% smaller than the pGL3 vector with the 2 kb promoter, 15% more copies of the construct were delivered of PEG101KB than PEG102KB. The 1 kb promoter resulted in more than 15% more luciferase signal, which proved that the 1 kb promoter was intrinsically more efficient in prostate cancer cells and the results were not because of more plasmid copies delivered. Our findings were supported by a previous report in the literature (29), which suggested that there must be repressor elements in the 5′ end of the promoter and those elements are eliminated in the PEG101KB construct.
Transcriptional analysis of the PEG10 promoter. A, Luciferase activity in relative luciferase units (RLU) of the full-length PEG10 promoter (PEG102KB) and the truncated 1KB promoter (PEG101KB) in LNCaP, DU145, and PC3. B, Comparison of the luciferase activity of PEG101KB promoter with TSTAPEG101KB and A.TSTAPEG101KB in LNCaP, DU145, PC3. C, Comparison of the luciferase activity of PEG101KB promoter with TSTAPEG101KB and A.TSTAPEG101KB in CRPC cell lines MR42D and 22Rv1. N = 6–8 in A and B. E, Near-infrared detection of the iRFP682 in cells transfected with various constructs. iRFP682 is a construct with no promoter. Significance was determined using the unpaired two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; n.s., not significant, was used to indicate significance between PEG102KB and PEG101KB, PEG101KB and TSTAPEG101KB, PEG101KB and A.TSTAPEG101KB; P < 0.05 (a) and P < 0.001 (c) were used to represent significance comparing TSTAPEG101KB and A.TSTAPEG101KB. Results are expressed in mean + SEM.
Transcriptional analysis of the PEG10 promoter. A, Luciferase activity in relative luciferase units (RLU) of the full-length PEG10 promoter (PEG102KB) and the truncated 1KB promoter (PEG101KB) in LNCaP, DU145, and PC3. B, Comparison of the luciferase activity of PEG101KB promoter with TSTAPEG101KB and A.TSTAPEG101KB in LNCaP, DU145, PC3. C, Comparison of the luciferase activity of PEG101KB promoter with TSTAPEG101KB and A.TSTAPEG101KB in CRPC cell lines MR42D and 22Rv1. N = 6–8 in A and B. E, Near-infrared detection of the iRFP682 in cells transfected with various constructs. iRFP682 is a construct with no promoter. Significance was determined using the unpaired two-tailed Student t test. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; n.s., not significant, was used to indicate significance between PEG102KB and PEG101KB, PEG101KB and TSTAPEG101KB, PEG101KB and A.TSTAPEG101KB; P < 0.05 (a) and P < 0.001 (c) were used to represent significance comparing TSTAPEG101KB and A.TSTAPEG101KB. Results are expressed in mean + SEM.
To further enhance activity of the truncated PEG10 promoter, we added a two-step transcriptional amplification (TSTA) element. (Supplementary Fig. S3.). There are various established versions of the TSTA in literature. In our study, (Fig. 4B) we tested the conventional-TSTA (TSTA; ref. 30) and the advanced-TSTA (A.TSTA; ref. 31). The conventional TSTA contains a gene coding for the fusion protein of GAL4-binding domain and a VP16 activation domain from the herpes simplex virus 1 activator VP16 downstream of the promoter. On the basis of previous literature (45), we opted for two VP16 domains in the fusion protein in our construct. The fusion protein binds to an array of 5 GAL4-binding sites, which increases transcription of the reporter gene (30). The A.TSTA contains additional polyglutamines and rat glucocorticoid receptor in the fusion protein, increasing transcription even further (31). Increased luciferase output was observed after the addition of both the TSTA to the 1 kb promoter (TSTAPEG101KB) and A.TSTA (A.TSTAPEG101KB) compared with the 1 kb promoter alone. Comparison of the two elements side by side showed that the transcriptional output for the A.TSTAPEG101KB was significantly more powerful than the TSTAPEG101KB based on luciferase assay (Fig. 4B). We then compared the output of PEG101KB, TSTAPEG101KB, and A.TSTAPEG101KB (Fig. 4C) in AR+ CRPC cell lines and found that the addition of A.TSTA increased the transcriptional power of the PEG10 promoter. An empty vector control was performed in all cell lines (Supplementary Fig. S4). On the basis of these data, A.TSTAPEG101KB facilitated the highest transcriptional output. The colon cancer cell line HT-29 exhibited no response to the addition of the TSTA elements even though that cell line has PEG10 expression similar to LNCaP (Fig. 4D).
The luciferase reporter gene in our construct was replaced with the near-infrared fluorescence reporter protein 682 (iRFP682; excitation 663 nm, emission 682 nm; ref. 46). A transfection protocol similar to the one used in the luciferase assay was performed on CWR-R1, PC3, and LNCaP cells and the transcriptional output of the PEG10 promoter constructs was visualized via fluorescence cell imaging (Fig. 4E). This was a qualitative approach performed to visualize expression of iRFP682. The PC3 cell line appeared to have greater fluorescence intensity than the CWR-R1 even though CWR-R1 expressed higher levels of PEG10 mRNA. The decreased expression of iRFP682 in CWR-R1 was the possible result of lower transfection efficiency of the transfection reagent used compared with PC3 or different posttranscriptional regulation of the reporter gene in the cell line that are beyond promoter control. These combined experiments documented that the A.TSTAPEG101KB promoter construct was powerful, resulting in its selection for subsequent in vivo experiments.
Prostate cancer detection by NIR optical imaging using the PEG10 promoter
Mice bearing subcutaneous CWR-R1 xenografts were injected via tail vein with either A.TSTAPEG101KB iRFP682 plasmid or TSTAPEG101KB iRFP682 plasmid using in vivo-jet PEI. The mice were imaged starting at 24 hours postinjection of the polymer-coated plasmid constructs. The images (Fig. 5A and B) of mice injected with the two plasmids demonstrated that the promoter alone was strong enough to drive expression levels of the fluorescent protein sufficient for detection. The addition of the A.TSTA to the plasmid resulted in significantly stronger signal compared with the promoter alone (Fig. 5C) at the 72-hour time point (P = 0.0374) based on ROI analysis of the two groups. In a separate experiment, mice were injected with the A.TSTAPEG101KB iRFP682 and tumors were excised at 96 hours postinjection. Ex vivo visualization of the tumors documented high expression of the iRFP682 protein (Fig. 5D). HT-29 subcutaneous xenograft mice were used as the negative control based on our previous in vitro results documenting no transcriptional activity in the luciferase assay. The HT-29 mice were imaged at 24, 48, and 72 hours postinjection of the two constructs (Fig. 5E and F). When normalized to the same minimum and maximum values as CWR-R1 (Fig. 5A and B), no signal was visible in the HT-29 xenografts. IHC analysis of excised HT-29 tumors documented an absence of PEG10 protein (Supplementary Fig. S5).
In vivo near-infrared fluorescence molecular imaging with PEG10 promoter–guided expression of iRFP682. Representative images of mice with CWR-R1 subcutaneous xenografts injected with A.TSTAPEG101KB iRFP682 (n = 3; A) and PEG101KB iRFP682 (n = 4; B) across different time points postinjection. C, ROI signal presented in mean ± SEM. Significance was determined by matching two-way ANOVA (construct P = 0.0374) with Bonferroni posttest (*, P < 0.05). D, Tumors excised from mice injected with A.TSTAPEG101KB iRFP682 72 hours post intravenous administration (n = 3). E and F, Representative images of mice with HT-29 subcutaneous xenografts injected with A.TSTAPEG101KB iRFP682 (n = 4; E) and PEG101KB iRFP682 (n = 4; F) across different time points postinjection. G, Representative image of mice with intratibial tumors in one of the tibias injected with A.TSTAPEG101KB (n = 4) across different time points postinjection. H, ROI signal presented in mean ± SEM of the SWR-R1 intratibial legs and healthy legs. Significance was determined by matching two-way ANOVA (construct P = 0.0234) with Bonferroni posttest (*, P < 0.05).
In vivo near-infrared fluorescence molecular imaging with PEG10 promoter–guided expression of iRFP682. Representative images of mice with CWR-R1 subcutaneous xenografts injected with A.TSTAPEG101KB iRFP682 (n = 3; A) and PEG101KB iRFP682 (n = 4; B) across different time points postinjection. C, ROI signal presented in mean ± SEM. Significance was determined by matching two-way ANOVA (construct P = 0.0374) with Bonferroni posttest (*, P < 0.05). D, Tumors excised from mice injected with A.TSTAPEG101KB iRFP682 72 hours post intravenous administration (n = 3). E and F, Representative images of mice with HT-29 subcutaneous xenografts injected with A.TSTAPEG101KB iRFP682 (n = 4; E) and PEG101KB iRFP682 (n = 4; F) across different time points postinjection. G, Representative image of mice with intratibial tumors in one of the tibias injected with A.TSTAPEG101KB (n = 4) across different time points postinjection. H, ROI signal presented in mean ± SEM of the SWR-R1 intratibial legs and healthy legs. Significance was determined by matching two-way ANOVA (construct P = 0.0234) with Bonferroni posttest (*, P < 0.05).
To test the ability of our imaging system to detect small lesions in complex microenvironments, we repeated our NIR imaging studies in an intratibial (i.t.) xenograft model that mimics prostate cancer bone metastasis (47). The i.t. model was created by injecting CWR-R1 cells into one tibia of each mouse. Three weeks post tumor inoculation, the mice were injected with the A.TSTAPEG101KB iRFP682 plasmid and imaged at 24, 48, and 72 hours (Fig. 5G). A NIR signal was observed in the CWR-R1 i.t. legs at all of the time points imaged. We compared the signal of the healthy legs of each mouse to the i.t. legs by using same-size ROIs around the tibias (Fig. 5H). The i.t. legs had significantly more near-infrared fluorescence due to the construct as determined by two-way ANOVA (P = 0.0234) and Bonferroni post hoc analysis at the 48- and 72-hour time points.
Prostate cancer detection by PET/CT imaging via the PEG10 promoter
Encouraged by our NIR optical imaging results, we decided to use our transcriptional technology to detect prostate cancer in vivo using the clinically relevant imaging modality PET/CT. To develop a PET/CT imaging strategy, we opted to use herpes simplex virus 1 thymidine kinase (HSV1-TK) as the reporter gene. HSV1-TK works by phosphorylating radiolabeled pyrimidine nucleoside derivatives such as 5-[124I]iodo-2′-fluoro-2′-deoxy-1-β-D-arabino-5-iodouracil (124I-FIAU; ref. 48). 124I-FIAU is a poor substrate for mammalian thymidine kinase; however, cells expressing HSV1-TK can phosphorylate and trap high levels of the radiolabeled nucleoside substrate (Supplementary Fig. S6A; ref. 49). Using HSV1-TK as the reporter gene, we were to able image subcutaneous CWR-R1 xenografts 72 hours postinjection of A.TSTAPEG101KB HSV1-TK plasmid construct with 124I-FIAU. To control for off-target expression of HSV1-TK and localization of 124I-FIAU, nontumor-bearing mice were injected with A.TSTAPEG101KB HSV-TK plasmid construct (Fig. 6A). Off-target localization of free 124I from deiodinated FIAU was seen in the thyroid (Ty) and in the bladder (B) as anticipated (50, 51). Both Ty and B signal can be seen in a representative image of a nontumor-bearing mouse as visualized in the 2D (Fig. 6A, a) and 3D reconstructed images (Fig. 6A, b). The mice in the CWR-R1 xenograft group exhibited the same nonspecific uptake in the Ty and B, as documented in the 2D (Fig. 6B, a) and 3D reconstructed images of a representative mouse (Fig. 6B, b). To confirm that the signal in the tumor was the result of HSV1-TK expression, subcutaneous tumors were analyzed by qPCR and IHC. Both HSV1-TK and PEG10 were detected (Supplementary Fig. S6B and S6C). Mice with CWR-R1 i.t. tumors were injected with the A.TSTAPEG101KB HSV1-TK construct. Forty-eight hours post DNA injection, mice were injected with 124I-FIAU and imaged the next day. An 124I-FIAU signal was clearly visible in the tumor-bearing tibias, while no signal was present in the healthy bone by 2D analysis (Fig. 6C, a). 3D reconstruction of the imaging data further confirmed that a signal was present in the tumor model when bone slicing was performed (Fig. 6C, b).
In vivo PET/CT imaging with PEG10 promoter–guided expression of HSV1-TK. A, Control tumor-less mice injected with A.TSTAPEG101KB HSV1-TK and imaged 24 hours postinjection with 124I-FIAU (n = 3). a, Representative 2D images; b, representative 3D images. B, Mice with CWR-R1 subcutaneous tumors injected with A.TSTAPEG101KB HSV1-TK (n = 3) and imaged 24 hours postinjection with 124I-FIAU. a, Representative 2D images; b, representative 3D images. Ty, thyroid; B, bladder; T, tumor. C, Intratibial CWR-R1 mice injected with A.TSTAPEG101KB HSV1-TK (n = 3) and imaged 24 hours postinjection with 124I-FIAU. a, Representative 2D images; b, representative 3D images.
In vivo PET/CT imaging with PEG10 promoter–guided expression of HSV1-TK. A, Control tumor-less mice injected with A.TSTAPEG101KB HSV1-TK and imaged 24 hours postinjection with 124I-FIAU (n = 3). a, Representative 2D images; b, representative 3D images. B, Mice with CWR-R1 subcutaneous tumors injected with A.TSTAPEG101KB HSV1-TK (n = 3) and imaged 24 hours postinjection with 124I-FIAU. a, Representative 2D images; b, representative 3D images. Ty, thyroid; B, bladder; T, tumor. C, Intratibial CWR-R1 mice injected with A.TSTAPEG101KB HSV1-TK (n = 3) and imaged 24 hours postinjection with 124I-FIAU. a, Representative 2D images; b, representative 3D images.
Discussion
The goal of this study was to understand PEG10 expression in prostate cancer and develop a method for detecting aggressive subtypes of prostate cancer by exploiting the disease-specific expression of PEG10. The expression of PEG10 in prostate cancer was previously only documented in NEPC (23). By analyzing prostate cancer models and clinical specimens, we found that PEG10 is expressed in CRPC (cell lines, PDXs, microarray) in addition to AR-null NEPC. Previous research hypothesized that PEG10 expression was strongly associated with the absence of AR (21). We found a negative correlation between the expression of AR and PEG10 in the LuCaP PDX models and mCRPC patient microarray analysis. Analysis of the AR variant–expressing cell lines proposed that AR splice variants played a role in PEG10 expression. Recent findings have shown that ONECUT2 acts as a suppressor of AR activity, a survival factor, and a driver of a NE differentiation in prostate cancer. We explored the trends of PEG10, AR, and ONECUT2 expression by microarray analysis. Our analysis showed a negative correlation between AR and ONECUT2, a positive correlation between ONECUT2 and PEG10, and elevated ONECUT2 in NEPC. The data supports published findings that ONECUT2 indirectly regulates PEG10 through suppression of AR or directly regulates PEG10 by binding to its promoter (43, 44). The recent literature (43, 44) reported that PEG10 was a marker of neuroendocrine differentiation that was present in highly lethal disease subtypes.
By harnessing the transcriptional specificity of the PEG10 promoter, we were able to develop a molecular genetic strategy for detecting prostate cancer by NIR and PET/CT imaging. Tissue-specific promoters are often weak when compared with constitutively active viral promoters such as CMV. In the past, TSTA has been used to improve the output of inherently weak tissue-specific promoters (30). On the basis of our results, the PEG10 promoter by itself was not a weak promoter and could achieve strong prostate cancer–specific expression. Although the promoter was sufficient on its own, the A.TSTA element significantly improved the signal in vitro. Although there were a few instances where the A.TSTA did not add significant improvement, it did contribute to a significant increase in gene production in the majority of the cell lines and assays. We managed to optimize a stronger promoter than its intrinsic version without compromising specificity, thus creating a powerful imaging tool that detects prostate cancer for a greater duration of time compared with other reported promoter-guided imaging technologies in the literature (16, 17). In this study, we used our optimized PEG10 promoter to drive the expression of reporter genes for NIR optical and nuclear imaging (iRFP682 and HSV1-TK). Our PEG10 molecular imaging approach was sensitive enough to detect small bone lesions in the intratibial bone metastasis model generated from the CWR-R1 cell line. Overall, the PEG10 promoter showed promise of utility across a variety of different prostate cancer models.
Most of the molecular genetic imaging studies reported in the literature relied on using adenoviral (Ad) delivery of the reporter constructs rather than the systemic injection of a polymer-coated plasmid as in our study. Previously, the promoters of prostate cancer–specific genes such as probasin (52) and PSA (53, 54) were used to drive tissue-specific expression of reporter genes by molecule genetic imaging. Studies using the PSA promoter also (53, 54) utilized the TSTA elements to enhance transcription within the adenovirus, similarly to our previous work on using the promoter of the metabolic protein a-methylacyl CoA racemase promoter to detect prostate cancer in vivo (55). Although recombinant viruses have been widely used as vectors for gene delivery, a number of limitations exist when using viruses in this capacity including difficulty in production, poor reproducibility, immunogenicity, insertional mutagenesis into the human genome and poor bioavailability. The imaging technology that we developed in this study has potential for clinical translation. While the use of the PEG10 promoter for molecular imaging is novel, plasmid therapeutics in humans and the use of cationic reagents for plasmid delivery are not. The PEI reagent used in this study is under investigation in a clinical trial for plasmid delivery in France (56) and Israel (56). Noncomplexed DNA without a transfection reagent is also in clinical trials for the treatment of pancreatic (57) and ovarian cancers (58). A plasmid CRPC vaccine applied by intradermal injection in the United States has also shown success in a clinical trial (59) and there are a number of prostate cancer plasmid vaccines currently being investigated in clinical trials. We believe the limiting factors in plasmid-based agents are the delivery agents and delivery route. Linear PEI is known to cause toxicity, and we observed some adverse effects in our mice after systemic administration (60), but we were successful at decreasing toxicity by increasing the total volume of administration.
Our finding that PEG10 expression was present in AR-null and AR splice variant prostate cancer underscores the potential importance of PEG10 as a therapeutic and imaging target. The loss of AR and the emergence of AR splice variants represent two important resistance mechanisms that arise during treatment. Effective therapies do not exist for these prostate cancer subtypes and the ability to monitor patient response to therapy radiographically is critical to the development of novel therapies. We have demonstrated proof-of-concept that the PEG10 promoter can be utilized as a molecular genetic imaging tool. By harnessing the cancer specificity of the PEG10 promoter, it is possible that this strategy could be employed for theranostic purposes by the expression of suicide genes such as cytosine deaminase and radioviral therapy using conditionally replicative viruses in the future.
Disclosure of Potential Conflicts of Interest
P.S. Nelson is an advisor (paid consulting) at Janssen and Astellas. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: M. Shapovalova, A.M. LeBeau
Development of methodology: M. Shapovalova, Y. Li, A.M. LeBeau
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Shapovalova, J.K. Lee, Y. Li, D.J. Vander Griend, P.S. Nelson, A.M. LeBeau
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Shapovalova, J.K. Lee, D.J. Vander Griend, I.M. Coleman, A.M. LeBeau
Writing, review, and/or revision of the manuscript: M. Shapovalova, I.M. Coleman, P.S. Nelson, S.M. Dehm, A.M. LeBeau
Study supervision: S.M. Dehm, A.M. LeBeau
Acknowledgments
We would like to thank Colleen Forster from the Histology and Research Laboratory at the University of Minnesota for IHC analysis. We also received great technical animal support from Meri DuRand and Angela Blum. We are grateful for Dr. Thomas Pengo for assisting us with creating 3D images. We would also like to acknowledge the University Imaging Centers at the University of Minnesota (Minneapolis, MN) for assisting with imaging experiment coordination. This research was supported by a Prostate Cancer Foundation Young Investigator Award (to A.M. LeBeau), a Masonic Cancer Center Brainstorm Award (to A.M. LeBeau), Prostate Cancer Foundation Challenge Awards (to A.M. LeBeau and S.M. Dehm), the Minnesota Partnership for Biotechnology and Medical Genomics Infrastructure Award (MNP IF 16.05 to A.M. LeBeau), NIH/NCI CA090628 Paul Calabresi K12 Award (to A.M. LeBeau), NIH/NCI R01 CA237272 (to A.M. LeBeau), NIH/NCI R01 CA233562 (to A.M. LeBeau), and NIH/NCI R01 CA174777 (to S.M. Dehm). The Prostate Cancer Biorepository Network is funded by the Department of Defense Prostate Cancer Research Program Awards nos. W81XWH-14-2-0182, W81XWH-14-2-0183, W81XWH-14-2-0185, W81XWH-14-2-0186, and W81XWH-15-2-0062. This work was also supported by the Office of the Assistant Secretary of Defense for Health Affairs, through the Prostate Cancer Research Program Award W81XWH-17-1-0275 (to M. Shapovalova).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.