Background
Adrenocortical carcinoma (ACC) is a rare (0.5–2 cases per million/year) endocrine malignancy that carries a poor prognosis at diagnosis due to its propensity to metastasize before detection. Even with aggressive surgical and oncologic therapy, the 5-year survival rate is an abysmal 16–38% [
1‐
4]. A major reason for the lack of an effective targeted treatment strategy for ACCs is an inadequate understanding of the molecular pathogenesis of the disease [
3,
4].
Genetic and epigenetic dysregulations of the WNT, p53, and IGF2 pathways appear to dominate various cancer-driving anomalies in the majority of ACCs [
5‐
7]. Recent findings from comprehensive genetic analyses of ACCs confirmed a signature role for WNT dysregulation in the origin and/or progression of ACCs [
4,
6,
8,
9]. Physiologically, both canonical and non-canonical WNT signaling pathways play global and zone-specific roles in the development, differentiation, and homeostasis of the adrenal gland [
10,
11]. In particular, endocrine homeostasis of the adrenal glomerulosa and fasciculata zones is largely controlled by WNT-differentiation signaling mediated by the Wnt4-Fz1/2-Dvl3-β-Catenin-SF1 axis [
12‐
16]. Regulatory components of this proposed adrenal cortex-specific Wnt4 axis include the secretory factors, frizzled-related protein 1 (SFRP1) and the putative tumor suppressor, DKK3 [
14,
17,
18]. Aberrant WNT signaling has been well-established in the origin of many tumor types and is strongly associated with stabilization of β-catenin in the cytoplasm and/or in the nucleus and constitutive activation of WNT target genes [
19,
20]. Similar stabilization and nuclear accumulation of β-catenin is seen in benign adrenocortical adenomas (ACAs) and frequently in malignant ACCs [
10,
21]. However, only 10% of ACCs with constitutively active β-catenin carry mutations in the β-catenin gene (
CTNNB1), suggesting alternate mechanisms of aberrant WNT activation, including dysregulation of WNT inhibitors such as Wif-1 [
22]. Other WNT regulatory mutations found in ACCs include
PRKAR1A [
23] and recently identified
KREMEN1 and
ZNRF3 gene deletions [
8,
24].
Although implicated in zonal differentiation and hormone biosynthesis [
14,
25], a definitive role for the ubiquitous WNT inhibitor DKK3 in promoting functional differentiation and/or blocking tumor dedifferentiation of the adrenal cortex has yet to be clarified. The inhibitory role of DKK3 in WNT signaling is context-dependent and appears to be influenced by a repertoire of cell surface receptors and co-expressed ligands [
26]. DKK3, a 38 kDa secreted glycoprotein with an N-terminal signal peptide, is also implicated in eliciting distinct intracellular roles in addition to its secretory functions [
27]. Reduced DKK3 expression is observed in a variety of solid tumors, and re-expression studies in multiple cancer cell types mostly resulted in cell cycle arrest and/or apoptosis, strongly suggesting a global tumor suppressor role for this WNT regulator (reviewed in [
26]). Furthermore, ectopic expression of DKK3 in a variety of cancer cell types stifled aggressive malignant behavior, reversed epithelial-mesenchymal transition (EMT), and impaired cell motility, pointing towards a comprehensive dedifferentiation-blocking role for DKK3 [
28,
29]. This study investigates a potential tumor suppressor role for the implicated adrenal differentiation factor DKK3 in blocking dedifferentiation of adrenocortical cells.
Methods
Tissue acquisition
Written informed consent was obtained from patients prior to surgical resection of adrenal tissue according to protocols approved by Institutional Review Boards at (a) Yale University, New Haven, CT, USA, (b) Heinrich Heine University Düsseldorf, Düsseldorf, Germany, and (c) Karolinska Institutet, Stockholm, Sweden. Tissue samples were flash-frozen (FF) in liquid nitrogen and stored at −80 °C until processed for study. Specimens displaying unequivocal histopathological characteristics of ACCs (n = 38) and histologically normal adrenal tissue (n = 14) samples excised with ACAs were selected for study. Consecutive unstained/hematoxylin & eosin (H&E) stained 5 μM sections of formalin-fixed, paraffin-embedded (FFPE) tissue samples underwent immunohistochemistry analyses. All samples were histopathologically confirmed by experienced endocrine pathologists before processing.
DNA, RNA, and protein preparation
Genomic DNA and total RNA were isolated from FF samples using AllPrep DNA/RNA/Protein Mini Kit (Qiagen) as per manufacturer’s recommendations. Quantity and quality of prepared nucleic acids were assessed by spectrophotometry (NanoDrop Technologies, Inc.). Total protein from cultured cells was extracted using Laemmli buffer (BioRad) as cell lysis buffer; protein concentrations were quantified using Pierce BCA Protein Assay Kit (ThermoFisher Scientific) and GloMax multidetection system (Promega), as per manufacturer’s instructions.
Gene expression analysis
Total RNA (100 ng) was reverse transcribed using iScript cDNA synthesis kit (Bio-Rad) as per manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was performed in triplicate using TaqMan PCR master mix with FAM fluorophore and probe/primer pairs specific to human DKK3 (Hs00951307_m1), FOXO1 (Hs01054576_m1), and RPLP0 (Hs99999902_m1) (ThermoFisher Scientific) according to manufacturer’s cycling conditions using CFX96 thermal cyclers (Bio-Rad). Gene expression levels were normalized to mean RPLP0 expression levels. Relative gene expression values were calculated using recommended Livak method (Bio-Rad). Fold-change expression values were calculated by base-two logarithmic transformations of relative gene expression values.
For pathway-focused gene expression analysis, (a) RT
2 Profile PCR Array Human WNT Signaling Pathway and (b) RT
2 Profiler PCR Array Human Transcription Factors were used according to protocol outlined in RT
2 Profiler PCR Array Handbook (Qiagen). Briefly, 100 ng of DNA-free RNA from each sample was used for 84 target genes listed in gene lists (available at
www.qiagen.com) using 96-well RT
2 profiler array format D. cDNA was prepared using RT
2 first strand kit and amplified using RT
2 SYBR Green Mastermix (both from Qiagen) using CFX96 thermal cycler. Differential expression of target genes was calculated using ∆∆C
T method on data web portal at
www.SABiosciences.com/pcrarraydataanalysis.php.
Methylation-specific PCR
Methylation status of CpG island A of
DKK3 promoter (Chr11:12029737–12030841) was assessed by MethylScreen technology using EpiTect Methyl II PCR Assay (Qiagen) as previously described [
30]. Briefly, 125 ng of genomic DNA was mock-digested or digested with methylation-sensitive and methylation-dependent restriction enzymes individually or together, and methylation status of target DNA sequence was measured using qRT-PCR with probes specific to target
DKK3 promoter sequence. C
T values were converted into percentages of unmethylated, intermediate-methylated, and hypermethylated CpG values using a quantitation algorithm from EpiTect Methyl II PCR Assay Handbook (Qiagen). Tissue samples were designated as hypermethylated (>5% alleles with hypermethylation), intermediate-methylated (>5% alleles with intermediate methylation), or unmethylated (no methylation detected).
DNA copy number analysis (CNA) by qRT-PCR
DNA from 27 ACC samples that passed specified test quality criteria were analyzed in quadruplicate with TaqMan Copy Number Assay using a primer / probe pair specific to target gene DKK3 or housekeeping gene RPPH1. Normal adrenal tissue was used for diploid (2n) reference. Copy numbers were predicted using CopyCaller software v2.0 (ThermoFisher Scientific). TaqMan Copy Number Assay used was Hs00228043_cn. Target gene DKK3 located on Chr.11:11989984 on NCBI build 37. Housekeeping gene Ribonuclease P RNA Component H1, RPPH1 located on Chr.14:20811565 on NCBI build 37.
Immunofluorescence (IF) detection of proteins
Five μM-thick FFPE sections were processed for immunofluorescence detection of DKK3 and β-catenin proteins as described previously [
31]. Goat anti-DKK3 polyclonal (SC14959; 1:100 dilution) or mouse anti-β catenin monoclonal (SC47778; 1:200 dilution) primary antibodies and anti-goat FITC (fluorescein isothiocyanate) and anti-mouse TR (Texas Red) secondary antibodies (1:1000) were used, followed by Ultracruz mounting agent containing 4′,6-diamidino-2-phenylindole (DAPI) (all from Santa Cruz Biotechnology, Inc.) for indirect immunodetection. A Zeiss AX10 confocal microscope with AxioVision 4.8 program was used for IF analysis, and photomicrographs were taken at a total magnification of 100× or 400×, as noted.
Cell culture, expression vectors, transfections, and western blot detection
American Type Culture Collection (ATCC)-authenticated human ACC cell lines SW-13 (CCL-105) and NCI-H295R (CRL-2128) were maintained in growth conditions recommended by ATCC, as reported previously [
31]. For DKK3 treatments, a working concentration of 5 μg/mL (in PBS) of human recombinant DKK3 (R&D Systems) was used. RNAi silencing was carried out with 3 unique 27-mer siRNA duplexes (designated siA, siB, and siC) targeting
DKK3 (Human) and
FOXO1 (Human) transcripts. Universal scrambled negative control siRNA was used as non-specific control (all from Origene). Lipofectamine2000-mediated transfection was carried out in Opti-MEM according to manufacturer’s recommendations (ThermoFisher Scientific) in 6-well plates with starting densities of 50,000 cells/well for SW-13 and 80,000 cells/well for NCI-H295R. Transfection medium was replaced with regular growth medium after 24 h of transfection. Cells were lysed for RNA extraction (after 24 h) or protein extraction (after 48 h), and assays were done 48 h post-transfection.
Myc-DDK tagged pCMV6-Entry, pCMV6-Entry/GFP, and pCMV6-Entry/DKK3 plasmid vectors (Origene) were used for transient and stable expression. Transient transfection was carried out in Opti-MEM medium using Lipofectamine2000 according to manufacturer’s recommendations (ThermoFisher Scientific) in 6-well plates with starting densities of 50,000 cells/well for SW-13 and 80,000 cells/well for NCI-H295R cells. Cells were transfected one day after plating. Transfection medium was replaced with appropriate growth medium 6 h post-transfection, and cells were assayed for cell behaviors 24 h post-transfection. Total cell numbers and viability were calculated by staining cells with 0.4% Trypan Blue (ThermoFisher Scientific) and counting with hemocytometer (Hausser Scientific Co.). Experiments were performed in triplicate, and parallel pCMV6-Entry/GFP transfections were used to determine transfection efficiency.
Stable Geneticin (G418)-resistant pCMV6-Entry, pCMV6-Entry/GFP, and pCMV6-Entry/DKK3 transfected clones were selected in 800 μg/mL G418-containing growth medium (ThermoFisher Scientific). Multiple clones were then pooled into populations to avoid expression variability and selection bias between clones. Established populations designated SW-Neo (from pCMV6-Entry transfections) and SW-DKK3 (expressing Myc-DDK/DKK3) were compared to parental SW-13 cells to determine effects of constitutive DKK3 expression on SW-13 cells’ malignant properties. Constitutive DKK3 expression was confirmed via qRT-PCR using TaqMan primer/probe pairs (ThermoFisher Scientific) and Western blotting using anti-DKK3 mAb (1:500; Abcam), anti-mouse-HRP (Santa Cruz Biotechnologies, Inc.), Mini-PROTEAN TGX gel, PVDF blotting membrane (Bio-Rad), and enhanced chemiluminescence (ECL) detection reagents (ThermoFisher Scientific) as per manufacturer’s protocols. Unless specified, 100 μg protein was loaded per well of 4–10% SDS gels (Bio-Rad). Equal protein loading was confirmed by staining PVDF membranes with GelCode Blue Safe Protein stain (ThermoFisher Scientific) after chemiluminescence detection.
Flow cytometric analysis of cell cycle
SW-13, SW-Neo, and SW-DKK3 cells were fixed in cold 70% ethanol for 30 min at 4 °C, washed twice with PBS, treated with ribonuclease (100 μg/mL), and stained with propidium iodide (PI; 50 μg/mL in PBS). Using bandpass filter 605 nm (for PI), forward and side scatter were measured in a BD LSRII Flowcytometer. Pulse processing was used to exclude cell doublets from the analysis. FlowJo software was used to analyze the best Gaussian distribution curve to each peak for the cell populations of G0-G1 and G2-M.
Cell invasion, migration, adhesion, and clonogenic growth assays
To assess invasive proficiencies, 100,000 SW-13, SW-Neo, or SW-DKK3 cells were allowed to invade through Matrigel from upper chambers containing serum-free medium to lower chambers containing 10% FBS medium in BD BioCoat Matrigel invasion chambers (BD Biosciences). After 24 h, Matrigel was removed, and invaded cells were fixed in 3.7% formaldehyde/PBS (10 min), stained with 0.05% crystal violet (30 min), and counted at 100X magnification with light microscope. Matrigel invasion assay was performed twice in triplicate chambers. In migration assays, 100,000 cells were allowed to migrate through 8 μM-pore size modified Boyden Chambers (BD Biosciences) from upper chambers containing serum-free medium to lower chamber with 10% FBS medium. After 4 or 8 h, cells that migrated to lower side of the membrane were fixed, stained, and counted as above.
Cell adhesion assays were carried out in 6-well plates. One hundred thousand cells were seeded per well, allowed to grow overnight, washed with warm PBS, and incubated with 0.5 mL of 0.25% Trypsin-EDTA for 1 min; Trypsin-EDTA was then removed, plates were tapped gently to remove loosely attached cells, cells were washed with 10% FBS medium, fixed, stained and counted as above. For clonogenic growth assays, cells were seeded in 6-well plates at low densities (5,000 cells/well) and allowed to grow 7 days in appropriate growth medium (SW-Neo and SW-DKK3) with medium change every 3 days. On day 7, cells were washed with PBS, fixed, and stained as above. Colonies with 12 ± 2 or 4 ± 2 cells were counted as separate groups and averaged from 6 wells. Experiments were repeated 3 times, and data from a representative experiment is presented.
Statistical analysis
Normal distribution of continuous variables was assessed using D’Agostino and Pearson omnibus tests. Normally distributed variables were analyzed using 2-tailed t test; Mann–Whitney U test was used for non-normally distributed variables. For variables with greater than 2 dependent values, a 1-way analysis of variance and Kruskal-Wallis tests were used for normally and non-normally distributed populations, respectively. Matched continuous variables were compared using Pearson correlation. Survival data were assessed by Kaplan-Meier methods, and differences were compared by Mantel-Cox test. Statistical analyses were performed using Prism 6 (GraphPad Software).
Discussion
DKK3 expression is down-regulated in many human cancers, including that of the thyroid, lung, prostate, colon, breast, and liver [
32,
33,
36,
43], but its regulation in ACC is unclear. In this study, we utilized comprehensive genetic, epigenetic, and functional approaches to identify and characterize a potential tumor suppressor role for DKK3 in adrenal carcinogenesis. Our study showed a significant decrease in DKK3 expression in 70% (25/37) of ACCs, strongly suggesting a tumor suppressor role for DKK3 in human adrenal tissue. Whether the observed silencing in malignant samples represents an earlier dedifferentiation or a later malignancy-promoting event needs to be determined. Despite the relatively small cohort size, this study did not find an association between DKK3 silencing and prognosis, unlike in gastric cancer [
35]. Of note, the majority of this cohort of ACCs was previously shown not to harbor mutations in
DKK3 or
FOXO1 genes while <10% carried beta-catenin mutations [
24].
Epigenetic modifications, including promoter methylation and chromatin condensation, have been proposed as major DKK3 silencing mechanisms in a variety of tumors [
43]. This study also supports a role for promoter hypermethylation in DKK3 silencing in ACCs. Interestingly, DKK3 expression was also significantly decreased in many samples with intermediate methylation (48%), suggesting that even intermediate levels of methylation may be adequate to silence DKK3 expression. Whether the
DKK3 promoter methylation observed in this study is a component of the global methylation changes observed in ACCs [
9,
40] or a specific
DKK3 gene-targeted event needs to be clarified. A large proportion of the ACC study cohort with non-methylated promoters but with reduced DKK3 expression led us to seek alternate mechanisms for DKK3 down-regulation in ACC. In light of recent findings that gene copy number variations may contribute to adrenocortical carcinogenesis [
8,
24], we analyzed a portion of our samples for
DKK3 gene copy number variations. The majority of samples identified with
DKK3 copy loss also had significantly reduced DKK3 expression. Only a handful of these samples had concurrent promoter methylation, indicating a possible independent role for gene copy loss in causing DKK3 down-regulation in ACC. One ACC sample with 6 copies of
DKK3 and a hypermethylated promoter had significantly reduced expression of DKK3, suggesting that copy number variations may occur earlier in ACC oncogenesis than gene-specific methylation events.
Statistical correlation to patient characteristics and outcomes did not reveal any prognostic association of reduced DKK3 expression in ACC patients, although reduced DKK3 expression was found to trend non-significantly toward female gender. This study did not reveal a relationship between DKK3 expression and aldosterone biosynthesis, as reported earlier [
25]. In addition, no significant correlation was observed in our tumor cohort between DKK3 expression, metastasis, and tumor grade.
We used functional approaches to characterize the effects of DKK3 on human ACC cells. Silencing of DKK3 in SW13, a human ACC cell line with intact and inductile WNT signaling and endogenously expresses DKK3, did not affect growth or viability of cells but resulted in reduced clonogenic growth and increased motility, consistent with a tumor suppressor role for DKK3 [
31,
41]. In contrast, exogenous addition of DKK3 to SW-13 cells resulted in increased motility, suggesting distinct roles for intracellular and secreted DKK3s. This observation is consistent with recent suggestions that DKK3 potentially has distinct intracellular signaling partners independent of canonical WNT-β-catenin circuitry [
26]. Overall, intracellular DKK3 appears to confer a more differentiated phenotype to SW-13 cells. Whether the observed DKK3-promoted more differentiated phenotype is through (a) reactivation of the proposed adrenocortical differentiation pathway [
14], (b) blocking of malignancy signaling networks, or (c) activation of a novel redifferentiation pathway needs to be clarified.
Light microscopic analysis revealed drastic changes in organization of cell outgrowths on the edges of slow-moving SW-DKK3 cells. While parental SW-13 and SW-Neo cells produced an overwhelming number of dynamic filopodia that confer polarity and promote directional movement (41–43), SW-DKK3 cells showed predominantly lobopodia, indicative of multipolar spreading and hence arrested motility [
44‐
46]. Although DKK3 has previously been shown to influence migratory and invasive phenotypes in multiple cancer cell types, an association of cell surface modifications that can impact cell mobility has not been shown. The mechanism(s) that elicit the observed changes in cell extension repertoire need to be investigated further. The association of the dedifferentiated phenotype and loss of DKK3 expression in ACC, combined with the re-acquisition of a relatively more differentiated phenotype in SW-13 ACC cells overexpressing DKK3, suggest a global differentiation role for DKK3 in adrenal cortex and the possibility that DKK3 could serve as a re-differentiation therapeutic target.
To explore potential pathways involved in eliciting the observed DKK3-promoted redifferentiated phenotype of ACC cells, we compared the expression pattern of 84 human transcription factors. Of the 3 transcription factors found to be over-expressed in SW-DKK3 cells (
ID1, JUN and FOXO1),
FOXO1 secured our immediate attention for 3 primary reasons: (1)
FOXO1 is known to promote functional differentiation of myofibroblasts [
47], (2)
FOXO1 inhibits osteosarcoma malignancy via WNT inhibition [
48], and (3)
FOXO1 transcription has been suggested in response to steroid hormones [
49]. We hypothesized that intracellular DKK3 promotes cellular differentiation signaling encompassing cellular spreading and stifled motility, at least in part, via
FOXO1 up-regulation.
FOXO1 RNAi silencing resulted in partial reversal of the motility suppression, suggesting that
FOXO1 may indeed play a role in DKK3-promoted redifferentiation of ACC cells. Based on the inverse relationship observed in ACC tissue between DKK3 and beta-catenin expression, it can be assumed that DKK3/FOXO1 regulation of malignant behavior of SW-13 cells is mediated through beta-catenin signaling. However, the precise DKK3-FOXO1 signaling circuitry in the context of adrenocortical differentiation needs to be investigated further.
Acknowledgments
We would like to thank Drs. Irene Esposito and Henry Mayringer of Department of Pathology, Heinrich Heine University, Düsseldorf, Germany for their contributions.