ADNP Is a Therapeutically Inducible Repressor of WNT Signaling In

Published OnlineFirst November 30, 2016; DOI: 10.1158/1078-0432.CCR-16-1604

Cancer Therapy: Preclinical Clinical Cancer Research ADNP Is a Therapeutically Inducible Repressor of WNT Signaling in Colorectal Cancer Cristina Blaj1, Agnes Bringmann1, Eva Marina Schmidt1, Manuela Urbischek1, Sebastian Lamprecht1, Thomas Frohlich€ 2, Georg J. Arnold2, Stefan Krebs2, Helmut Blum2, Heiko Hermeking1,3,4, Andreas Jung1,3,4, Thomas Kirchner1,3,4, and David Horst1,3,4

Abstract

Purpose: Constitutively active WNT signaling is a hallmark Results: ADNP was overexpressed in colon cancer cells with of colorectal cancers and a driver of malignant tumor high WNT activity, where it acted as a WNT repressor. Silencing progression. Therapeutic targeting of WNT signaling is dif- ADNP expression increased migration, invasion, and prolifera- ficult due to high pathway complexity and its role in tissue tion of colon cancer cells and accelerated tumor growth in homeostasis. Here, we identify the transcription factor xenografts in vivo. Treatment with subnarcotic doses of ketamine ADNP as a pharmacologically inducible repressor of WNT induced ADNP expression, significantly inhibited tumor growth, signaling in colon cancer. and prolonged survival of tumor-bearing animals. In human Experimental Design: We used transcriptomic, proteomic, patients with colon cancer, high ADNP expression was linked to and in situ analyses to identify ADNP expression in colo- good prognosis. rectal cancer and cell biology approaches to determine its Conclusions: Our findings indicate that ADNP is a tumor function. We induced ADNP expression in colon cancer suppressor and promising prognostic marker, and that ketamine xenografts by low-dose ketamine in vivo.Clinicalassocia- treatment with ADNP induction is a potential therapeutic approach tions were determined in a cohort of 221 human colorectal that may add benefit to current treatment protocols for patients cancer cases. with colorectal cancer. Clin Cancer Res; 23(11); 2769–80. 2016 AACR.

Introduction tissue, while those with lower WNT levels are frequently more central within the tumor and appear phenotypically more differ- Colorectal cancer is a major cause of cancer morbidity and entiated (6, 7). Because of these findings, high WNT signaling mortality, ranking third in cancer incidence among men and activity is assumed to be a driving force of colon cancer invasion women (1). Most colorectal cancers are initiated by mutations and progression, making it an attractive potential target for in APC or b-Catenin that lead to overactivation of the WNT therapeutic intervention. However, because WNT signaling is signaling pathway in these tumors (2). Despite this mutational required for various physiologic processes including adult tissue activation, WNT signaling in colorectal cancer remains regulated and stem cell homeostasis, efforts in targeting this central pathway on high levels, resulting in distinct tumor cell subpopulations in clinical settings is complicated, and serious side effects may be with relatively low or high WNT activity (3). Colon cancer cell anticipated (8). subpopulations with high WNT levels were attributed certain Activity-dependent neuroprotector homeobox (ADNP) was characteristics such as more mesenchymal phenotypes and puta- initially identified in brain tissue and encodes a ubiquitously tive cancer stem cell traits, express markers that are linked to tumor expressed zinc finger homeobox protein with transcription factor invasion, and therefore are thought to be crucial drivers of colon activity (9, 10). Most knowledge on ADNP function is related to cancer progression (4, 5). These tumor cells are typically located at the central nervous system where it is required for brain formation the infiltrative tumor edge where they invade the surrounding and cranial neural tube closure (11). It also assumes protective roles against cognitive defects in neurodegenerative disease (12). Moreover, ADNP has been shown to reduce the expression of 1Pathologisches Institut, Ludwig-Maximilians-Universitat€ Munchen,€ Germany. genes involved in regulation of transcription, organogenesis, and 2Laboratory for Functional Genome Analysis (LAFUGA), Gene Center, Lud- neurogenesis, and is suggested to interact with chromatin-remo- 3 wig-Maximilians-Universitat€ Munchen,€ Germany. German Cancer Consortium deling complexes that are associated with cellular differentiation 4 (DKTK), Heidelberg, Germany. German Cancer Research Center (DKFZ), Hei- (13). In regard to cancer, a previous report demonstrated over- delberg, Germany. expression of ADNP in proliferative tissues and several different Note: Supplementary data for this article are available at Clinical Cancer malignancies, including colon cancer, and since ADNP depletion Research Online (http://clincancerres.aacrjournals.org/). reduced the viability of certain cancer cells, it suggested a possible Corresponding Author: David Horst, Pathologisches Institut der LMU, Thal- association with tumorigenesis and cell survival (10). However, € kirchner Str. 36, Munchen 80337, Germany. Phone: þ4989-2180-73611; Fax: the contribution of ADNP to human cancer and its functional role þ4989-2180-73727; E-mail: [email protected] in malignancies is still poorly understood. doi: 10.1158/1078-0432.CCR-16-1604 Focusing on transcription factors linked to WNT signaling, we 2016 American Association for Cancer Research. here identify ADNP as a negative regulator of WNT in colorectal

www.aacrjournals.org 2769

Blaj et al.

reference genome. Significantly upregulated genes were char- Translational Relevance acterized according to signaling pathways using PANTHER High WNT signaling activity is the main driver of tumor version 10.0 (www.pantherdb.org). Expression data are acces- invasion and progression in colorectal cancer. Here, we iden- sible through GEO (GSE79395). tify ADNP as a negative regulator of WNT signaling that can be pharmacologically induced by treatment with low-dose keta- Immunoblotting and proteome analysis mine. WNT suppression through ADNP induction may Immunoblotting of whole-cell lysates of colon cancer cells was improve treatment of patients with colorectal cancer. done as described previously (3). Antibodies are listed in Sup- plementary Table S3. For mass spectrometry (MS) proteome analysis, cells were lysed, sonicated, and centrifuged through QIA-Shredder devices (Qiagen). Ten micrograms of total protein was used for further cancer and reveal its role as a tumor suppressor with effects on reduction, alkylation, and trypsinization. For separation, an tumor cell proliferation, migration, invasion, as well as on tumor EASY-nLC 1000 chromatography system connected to an Orbi- growth. Furthermore, we demonstrate that ADNP can be thera- trap XL instrument (Thermo Scientific) was used. Raw data files in vivo peutically induced in colon cancer , slowing tumor growth were processed with the Homo sapiens subset of the UniProt and progression, and highlight its potential as a prognostic database and MaxQuant V1.5.1. Proteins with log2 fold changes predictor in human patients with colon cancer. of 0.6 at P values < 0.05 were considered relevant. Detailed methods for RNA-Seq and MS are described in Supplementary Materials and Methods Materials and Methods. Gene expression datasets, TCGA data, and GSEA Three sets of differentially expressed genes from colon cancer Proliferation, migration, and invasion assays cells with low and high WNT activity were screened for consis- To assess cell proliferation, 5 104 cells per well were seeded on tently deregulated genes (3, 5). From The Cancer Genome Atlas conductive microtiter plates (E-Plate 16) and monitored for up to (TCGA) database (https://gdc.cancer.gov/), RNA-Seq data of 41 150 hours using an xCELLigence DP instrument (ACEA Biosci- normal mucosa samples and 457 colon cancer samples were ence). For transwell migration and invasion assays, 8 mm ThinCert retrieved. Within the cancer sample data, Pearson correlations of cell culture inserts (Greiner Bio-One) were used, which for inva- ADNP expression and expression of 20,531 genes within this sion were coated with 100 mL of 1 mg/mL growth factor–depleted dataset were calculated and genes were ranked accordingly. GSEA Matrigel (Corning). A total of 1 105 cells/well were seeded in analyses were conducted using this ranked gene list against serum-free medium in the upper insert chambers and after 24 curated sets of upregulated WNT targets derived from Nusse and hours DMEM with 10% FBS was added to the bottom chambers of colleagues (web.stanford.edu/group/nusselab/cgi-bin/wnt/tar- the inserts. For HCT116 and SW1222 cells, inserts were removed get_genes) and Herbst and colleagues (14). Gene sets are listed after 1 or 3 days for migration, and 3 or 5 days for invasion, in Supplementary Table S1. Heatmaps for individual factors were respectively. Cells were fixed, stained with crystal violet, residual drawn with GENE-E (Broad Institute). cells from the top chamber were removed, and photomicrographs were taken. For quantification, staining was dissolved in 250 mLof Cell culture and in vitro treatments 30% acetic acid, and absorbance was measured at 590 nm on a HEK293 and HCT116 cell lines were obtained from ATCC, Varioskan instrument (Thermo Scientific). SW1222 from the Ludwig Institute for Cancer Research (New York, NY), and primary colon cancer cells from the Human Tumor xenografts and in vivo treatments Tissue and Cell Research Foundation. Cell lines were cultured Mouse experiments were reviewed and approved by the Regier- in DMEM containing 10% FBS, 100 U/mL penicillin, and 0.1 ung von Oberbayern. To determine effects of ADNP depletion, 8 mg/mL streptomycin (Biochrom). For in vitro experiments, 106 SW1222 ADNP knockdown or control cells were suspended primary tumor cells were grown as spheroids as described in 200 mL of a 1:1 mixture of PBS and growth factor–depleted previously (7). For WNT induction or inhibition, cells were Matrigel, and injected subcutaneously into age and gender treated with 20 ng/mL WNT3a (R&D Systems), 20 mmol/L matched 6- to 8-week-old NOD/SCID mice (Jackson Laboratory). LiCl, or 10 mmol/L XAV939 (both Sigma-Aldrich), respectively. Tumor growth was measured over time using calipers. Matched In vitro ketamine treatment was done at concentrations of 200 mice carrying ADNP knockdown and control xenografts were mmol/L (Ratiopharm). Transfections and transductions for sacrificed when knockdown tumors reached volumes of 500–900 ADNP knockdown, overexpression, CRISPR/Cas9 genome edit- mm3. For in vivo treatment studies, subcutaneous xenografts were ing, and luciferase assays are described in Supplementary grown as described above using SW1222, HCT116, or primary Materials and Methods. colon cancer cells. Mice were randomly assigned to control or treatment groups when tumor volumes reached 100 mm3. Gene expression analyses Ketamine (20 mg/kg in PBS) or as control PBS were administered RNA was isolated from cell lines using TRIzol (Invitrogen) daily intraperitoneally until tumors reached volumes of 1,000– and used for quantitative real-time PCR, as described previ- 1,300 mm3. ously (3), using the primers listed in Supplementary Table S2. For RNA-Seq, libraries were constructed using the Encore Com- IHC, immunofluorescence, and imaging plete RNA-Seq library system (NuGEN) according to manufac- IHC was done on 5-mm tissue sections, as described previously turer's protocol, sequencing was done on a HiSeq 1500 instru- (7), using the antibodies listed in Supplementary Table S3. ment, and processed data were mapped to the hg19 human Scoring of ADNP in colon cancer cases was done in a manner

2770 Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research

ADNP Represses WNT in Colorectal Cancer

blinded for clinical outcome, and cases were classified semiquan- ses (GSEA) revealed that ADNP in colon cancer samples strongly titatively for overall ANDP expression intensity, ranging from correlated with genes enriched for WNT target gene signatures absent or barely detectable to strong overexpression. Scoring for (Fig. 1B). On the protein level, ADNP was overexpressed at the nuclear b-Catenin was done as described previously (3). Staining infiltrative tumor edge of colorectal cancers, where it colocalized intensities of ADNP in tumor cells with low or high b-Catenin with strong nuclear b-Catenin expression, a marker for colon were quantified with ImageJ (NIH, Bethesda, MD). cancer cells with high WNT activity (Fig. 1C; Supplementary For immunofluorescence, cultured cells were fixed in 4% para- Fig. S1A and S1B). Moreover, in adenomas of patients with FAP formaldehyde for 10 minutes, permeabilized with 0.2% Tri- (n ¼ 23 adenomas), known to carry APC gene mutations (15), tonX100 for 15 minutes, blocked with 3% BSA in PBS for 30 ADNP was overexpressed when compared with normal mucosa minutes, and then incubated for 60 min at room temperature with (Fig. 1C). In addition, even in normal mucosa, epithelial cells at ADNP Ab (1:50), AXIN2 Ab (1:100), or DNMT1 Ab (1:100). the crypt base, where WNT activity is increased (16), showed Secondary Alexa Fluor 488- or 555–conjugated antibodies (Invi- slightly increased ADNP expression (Supplementary Fig. S1C). trogen) were used for visualization and nuclei were counter- These findings demonstrated consistent upregulation of ADNP in stained with DAPI (Vector Laboratories). Confocal fluorescence colorectal cancer cells and precursor lesions with high WNT images were taken on a LSM 700 laser scanning microscope using activity on mRNA and protein levels, and suggested a possible the ZEN software (Carl Zeiss). regulation of ADNP by WNT. To test whether ADNP expression directly responds to WNT, we stimulated WNT signaling in HEK293 cells by the GSK3b inhib- Clinical samples and statistical analyses itor lithium chloride (LiCl) or by WNT3a, and then evaluated Resection specimens of patients diagnosed with FAP as well as ADNP mRNA and protein levels. Surprisingly, although TOPflash colorectal cancer specimens from patients that underwent inten- luciferase assays and overexpression of b-Catenin and AXIN2 tionally curative surgical resection between 1994 and 2006 at the indicated strong induction of WNT signaling, there were no LMU were drawn from the archives of the institute of pathology. significant effects on ADNP expression (Fig. 1D and E). Also, For colorectal cancer specimens, inclusion criteria were localized suppressing WNT activity by silencing b-Catenin in colorectal colorectal adenocarcinomas with absence of nodal (N0) or dis- cancer cell lines had no significant impact on ADNP protein levels tant metastasis (M0) at the time of diagnosis (UICC stage I and II). (Fig. 1F). Hence, although ADNP expression strongly coincided Follow-up data were recorded prospectively by the Munich with WNT signaling activity in colon cancer, it apparently is no Cancer Registry. Specimens and data were anonymized, and the direct WNT target gene. need for consent was waived by the Institutional Ethics Commit- tee of the Medical Faculty of the LMU. Colorectal cancer tissues ADNP is a repressor of WNT signaling in colon cancer were assembled into tissue microarrays (TMA) with representative To obtain initial insights into ADNP function in colorectal 1-mm cores, including tumor edges and tumor centers of each cancer, we next analyzed the effects of ADNP knockdown on gene case. The final colorectal cancer collection consisted of 221 cases expression in HCT116 colon cancer cells using RNA-seq. Deple- of which in 43 cases (19%), patients had died from their tumor tion of ADNP by siRNA caused 1.4-fold or more deregulation of within the follow-up period. For tumor-specific survival analysis, 4.82% of the transcriptome. Interestingly, upregulated genes were colorectal cancer–attributed deaths were defined as clinical end- significantly more frequent than downregulated genes (3.69% vs. points. For analysis of disease-free survival, tumor progression 1.13%, P < 0.0001), suggesting predominantly repressive func- after surgical resection was the clinical endpoint, documented as tions of ADNP on the transcriptome (Fig. 2A). To identify poten- either tumor recurrence or metastasis. Survival was analyzed by tial targets of ADNP repression, we therefore focused on genes that the Kaplan–Meier method and groups were compared with the were upregulated by ADNP knockdown and screened them for log-rank test. Cox proportional hazards model was used for functional and pathway associations using the PANTHER analysis multivariate analysis. Statistics were calculated using SPSS (IBM). tool. Surprisingly, these analyses revealed that WNT signaling was the pathway most prominently related to genes affected by ADNP Results depletion (Fig. 2B). We therefore compared our gene expression ADNP is overexpressed in colon cancer cells with high WNT dataset with WNT target gene signatures using GSEA, and found signaling activity that indeed upregulated WNT target genes were significantly To identify transcription factors that are linked to WNT signal- enriched upon ADNP depletion (Fig. 2C), with overexpression ing in colon cancer, we comparatively analyzed three previously of typical WNT targets such as DNMT1, CD44, AXIN2, and TCF7 published gene expression profiles of colon cancer cell subpopu- (Fig. 2D). These findings suggested repression of WNT signaling lations with low and high WNT activity (3, 5). Among the few by ADNP. genes that were consistently upregulated in tumor cells with high Next, to determine effects on the proteome level, we used mass WNT signaling, we identified ADNP as the only overexpressed spectrometry analysis and identified deregulated proteins after gene encoding a transcription factor (Supplementary Table S4). ADNP silencing. Consistent with our transcriptome data, we Direct comparison confirmed that increased ADNP expression found that upregulated proteins were significantly more abun- coincided with high WNT target gene expression and, conversely, dant than downregulated proteins (2.8% vs. 1.43%, P < with repression of genes linked to tumor cell differentiation 0.0001, Fig. 2E). Furthermore strong correlation between our (Fig. 1A). We then analyzed an independent data set of 41 normal transcriptome and proteome results among 58 factors that were mucosa samples and 457 colon cancers from The Cancer Genome significantly deregulated in both datasets indicated consistency Atlas (TCGA) and found that ADNP mRNA expression on average within these analyses (Pearson r2 ¼ 0.52, P ¼ 0.001, data not was 2.73-fold increased in colon cancer compared to normal shown). Among proteins that showed most significant upregula- mucosa (Supplementary Table S4). Gene Set Enrichment Analy- tion were the WNT target DNMT1, a known driver of cell

www.aacrjournals.org Clin Cancer Res; 23(11) June 1, 2017 2771

Blaj et al.

r ADNP-Positive correlation

ADNP-Negative correlation

Figure 1. ADNP is overexpressed in colon cancer cells with high WNT activity but not affected by WNT manipulation. A, Heatmaps of ADNP, selected WNT targets, and differentiation factors in three data sets (D1–D3) of colon cancer cells with high and low WNT activity. B, GSEA for genes ranked by Pearson correlation (Pearson r) to ADNP expression for two WNT target gene signatures by Herbst and colleagues (green curve: NES ¼ 1.70, P < 0.001) and Nusse and colleagues (orange curve: NES ¼ 1.36, P ¼ 0.09) in 457 RNA-Seq datasets of colon cancer from TCGA. C, IHC on serial sections of colon cancer (top; scale bar, 100 mm) and a colonic adenoma of an FAP patient (bottom; scale bar, 200 mm) illustrate upregulation of ADNP in areas with increased b-Catenin staining (arrows). D and E, Dual-luciferase assays with TOPﬂash reporter constructs, immunoblotting, and qRT-PCR results on indicated proteins or genes after stimulation of HEK293 cells with WNT3a (D) or LiCl (E). P values are t test results, data are mean SD, n 3. F, Immunoblotting of indicated proteins after transfection of HCT116 and SW1222 colon cancer cells with siRNA against b-Catenin. Numbers below immunoblots indicate fold change by densitometry.

proliferation and interaction partner of b-Catenin, as well as ADNP silencing increased expression of active-b-Catenin and the TALIN-1, a recently identified key node of WNT signaling with WNT targets DNMT1, AXIN2, and LEF-1 (Fig. 3A). Immunoflu- roles in cell migration, invasion, and angiogenesis of human orescence of HCT116 cells confirmed these results with cyto- cancers (Fig. 2F; Supplementary Table S5; refs.17, 18). plasmic and membranous, or nuclear increase of AXIN2 and We further addressed the effects of ADNP on WNT signaling in DNMT1, respectively (Fig. 3B). In addition, upon ADNP silenc- HCT116 and SW1222 colon cancer cells in vitro. In both cell lines, ing, both cell lines showed elevated WNT activity when assessed

2772 Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research

ADNP Represses WNT in Colorectal Cancer

A B

Figure 2. ADNP depletion shows derepressive effects on transcriptome, proteome, and WNT signaling in colon cancer cells. A–D, Gene expression analyses for ADNP knockdown in HCT116 cells. % Gene hits A, Heatmap results of genes with significantly (P < 0.05) differential expression and 1.4 or more fold change. C D Rows represent gens and columns represent biological replicates. B, Top ten results of PANTHER analysis showing frequencies of upregulated genes linked to pathways indicated. C, GSEA with genes ranked by fold change for WNT target gene signatures by Herbst and colleagues (green curve: NES ¼ 1.54, P < 0.001) and Nusse and colleagues (orange curve: NES ¼ 1.36, P ¼ 0.04). D, Heatmap of selected WNT targets among differentially expressed genes. E and F, Proteome analyses for ADNP knockdown in HCT116 cells. E, Heatmap results of proteins with significant (P < 0.05) differential expression. Rows represent proteins and columns represent biological replicates. F, Volcano plot of protein expression. EF Red and blue dots indicate proteins with significant up- or downregulation (P < 0.05; abs. fold change >1.5), respectively. Arrows highlight most significantly deregulated proteins. Legends on heatmaps indicate fold change.

by TOPflash luciferase reporter assays (Fig. 3C). In contrast, and also generated ADNP knockout cells by CRISPR/Cas9 overexpression of ADNP reduced b-Catenin and LEF-1 genome editing (Fig. 4A; Supplementary Fig. S2A). Importantly, (Fig. 3D), and repressed WNT signaling in TOPflash assays ADNP depletion or knockout caused dramatic increases in Trans- (Fig. 3E). These results further corroborated the hypothesis that well cell migration and invasion of both colon cancer cell lines as ADNP negatively regulates WNT signaling in colorectal cancer determined by Boyden chamber assays (Fig. 4B and C; Supple- with repression of factors related to cell proliferation and other mentary Fig. S2C). In addition, ADNP depletion or knockout malignant traits of colorectal cancer cells. significantly increased cell proliferation of both cell lines as determined by impedance measurements (Fig. 4D; Supplemen- ADNP represses malignant traits and tumor growth of colon tary Fig. S2D). Of note, these effects could be counteracted in cancer ADNP knockout cells by concomitant depletion of b-Catenin, Because WNT and its associated factors DNMT1 and TALIN-1 DNMT1, or TALIN-1 (Supplementary Fig. S2B–S2D). Moreover, are known to regulate proliferation, migration, and invasion of ADNP overexpression showed opposite effects on invasion and colorectal cancer (17, 18), we tested the functional relevance of migration (Supplementary Fig. S3). To further assess effects in vivo, ADNP loss for these malignant traits in HCT116 and SW1222 we then injected 8 106 SW1222 cells with and without stable colon cancer cells. We depleted ADNP by two different shRNAs ADNP knockdown subcutaneously into flanks of NOD/SCID

www.aacrjournals.org Clin Cancer Res; 23(11) June 1, 2017 2773

Blaj et al.

A B

D E

Figure 3. ADNP represses WNT signaling in colon cancer in vitro. A–C, Effects of ADNP or control (Ctrl) knockdown by siRNA on HCT116 and SW1222 colon cancer cells, harvested 48 hours after transfection. A, Immunoblotting of indicated proteins on whole-cell lysates. B, Representative confocal immunofluorescence images of HCT116 cells for indicated proteins and DAPI as nuclear counterstain. Scale bars, 20 mm. C, Dual-luciferase assays for HCT116 and SW1222 colon cancer cells, simultaneously transfected with indicated siRNAs and TOPflash reporter constructs. D and E, Effects of transient ADNP overexpression by transfection of HCT116 and SW122 with p4.2-hADNP compared with empty p4.2 vector for 24 hours. D, Immunoblotting of indicated proteins on whole-cell lysates. E, Dual-luciferase assays for HCT116 and SW1222 colon cancer cells, simultaneously transfected with p4.2-hADNP or p4.2 and TOPflash reporter constructs. Numbers below immunoblots indicate fold change by densitometry. P values are t test results, data are mean SD, n 3.

mice and measured tumor growth over time. In line with our upon ADNP knockdown in vivo, explaining these observed differ- in vitro data, tumor xenografts derived from SW1222 cells with ences in tumor growth and size (Fig. 4F). Taken together, these ADNP knockdown grew signiﬁcantly faster, yielding larger data showed strong effects of ADNP depletion on migration, tumors, with darker color due to increased intratumoral hemor- invasion, and proliferation of colon cancer cells in vitro as well rhage, when compared with tumor xenografts with normal as tumor growth in vivo, and suggested a tumor suppressor ADNP levels (Fig. 4E). Immunohistochemical staining of these function of ADNP through repression of WNT signaling in tumors for Ki67 revealed that proliferation was increased colorectal cancer.

2774 Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research

ADNP Represses WNT in Colorectal Cancer

Induction of ADNP by subnarcotic ketamine suppresses tumor and sex were not linked to ADNP (Supplementary Table S6). growth in vivo Including these variables and b-Catenin expression levels into Previous studies indicated that ADNP expression can be phar- proportional hazards regression analyses revealed that ADNP macologically induced by subnarcotic doses of ketamine in expression was an independent predictor of favorable outcome cortical neurons (19, 20). Exploiting this potential, we treated (Supplementary Table S7). Collectively, high ADNP expression colon cancer cell lines and primary colon cancer cells, which had was a marker of good prognosis in patients with colorectal cancer, high endogenous WNT activity (Supplementary Fig. S4), with which is in agreement with its tumor-suppressive function in low-dose ketamine and observed slight but steady increases of this malignancy. ADNP protein levels by immunoblotting (Fig. 5A). To determine the effects of ketamine-induced ADNP induction on WNT signaling, we then subjected these cell lines and primary colon cancer Discussion cells to TOPflash luciferase assays and indeed observed significant Constitutive activation of WNT signaling by inactivating muta- reductions in WNT activity under ketamine treatment (Fig. 5B). tions in APC is hallmark of most colorectal cancers and drives Importantly, these relative repressive effects of ketamine on WNT tumor progression via target genes that promote cell proliferation, activity were decreased in ADNP-knockout cell lines, while the invasion, and spawn putative cancer stem cell traits (2, 4, 5). effects of a direct WNT inhibitor (XAV939) remained unchanged Therefore, identifying effectors or regulators of WNT signaling, suggesting that WNT repression by ketamine in part depended on which are involved in governing these malignant traits, may hold ADNP induction (Supplementary Fig. S5A and S5B). In addition, keys for a deeper understanding of colon cancer biology, and the similar to the effects of ADNP overexpression, ketamine reduced development of more effective therapies (21). In this context, we migration and invasion of colon cancer cells (Supplementary here identified upregulation of the transcription factor ADNP in Fig. S5C). Next, we treated NOD/SCID mice bearing colon cancer colon cancer cells with high WNT signaling activity. Although cell line or primary colon cancer xenografts with subnarcotic doses differential expression of ADNP in colon cancer cell subpopula- of ketamine and observed tumor growth over time. Ketamine tions with high and low WNT activity was relatively small, we treatment significantly slowed tumor growth (Fig. 5C) and pro- demonstrate high consistency of this finding on mRNA and longed tumor survival (Fig. 5D). These findings implicate that protein levels. However, a direct regulation of ADNP through through induction of ADNP and WNT repression, subnarcotic WNT remained unclear, as modulation of WNT signaling in colon doses of ketamine can inhibit colorectal cancer growth and tumor cancer and other cells yielded no measurable effects on ADNP progression in vivo. expression. How ADNP expression itself is regulated in colon cancer cells, therefore, still needs to be determined, keeping in High ADNP expression predicts good outcome of patients with mind that ADNP is known to engage in autoregulatory feedback colorectal cancer loops, as others report (22). Finally, we tested for ADNP expression and disease outcome of Consistent upregulation of ADNP in colon cancer cells with patients with colorectal cancer. Using IHC, we scored overall high WNT activity prompted us to further investigate its ADNP expression in a collection of 221 stage I and II human functional relevance. Unexpectedly, ADNP depletion caused colorectal cancers with recorded clinical follow-up data. ADNP significant upregulation of WNT target genes, as assessed on expression varied substantially between cases, ranging from neg- multiple levels of the transcriptome, proteome, for individual ative or barely detectable (score 0) to strong expression (score factors,andinreporterassays,while,inlinewiththesefindings, 3; Fig. 6A; Supplementary Table S6). Using Kaplan–Meier statis- overexpression of ADNP in colon cancer cells showed opposite tics, we found that differential ADNP expression was strongly effects. ADNP therefore exhibited a previously unknown func- linked to cancer-specific survival. All patients with strong ADNP tion in suppressing WNT activity in colon cancer. Specifically, expression completely survived their follow-up period (score 3, ADNP knockdown caused overexpression of WNT targets such 5-year survival rate 100%). In contrast, moderate (score 2, 5-year as DNMT1 and the recently identified WNT signaling node survival rate 88.7%), weak (score 1, 5-year survival rate 78.2%), TALIN-1, which are reported drivers of tumor cell proliferation, and barely detectable or negative ADNP expression (score 0, invasion, and migration, respectively (17, 18). Indeed, we 5-year survival rate 60.4%) were significantly linked to cancer- found that these attributes of malignancy were strongly specific deaths at increasing frequencies (Fig. 6B). Testing for unleashed in colorectal cancer cells under ADNP knockdown disease-free survival yielded comparable, yet slightly less stark or knockout in vitro, and that ADNP depletion strongly results (Fig. 6B). We then categorized ADNP expression into low enhanced tumor growth in vivo. ADNP therefore acted as a (scores 0 and 1) and high (scores 2 and 3), and tested for an overall tumor suppressor gene in colorectal cancer and our data suggest association with b-Catenin expression levels (Supplementary Fig. that this may mainly be transduced through WNT repression. S6). Although not significant, high ADNP expression tended to be Because of the strongest ADNP expression in tumor cells with more frequent in cases with high nuclear b-Catenin (Supplemen- high WNT activity, its function may thus partially mirror that of tary Table S6). Interestingly however, high and low ADNP expres- WNT feedback inhibitors such as AXIN2 (23). However, sion separated survival probabilities particularly well in the subset although these effects may be mediated through chromatin of colorectal cancer cases with high nuclear b-Catenin expression remodeling, since ADNP has been shown to interact with (Fig. 6C). We then evaluated co-occurrences of ADNP expression SWI/SNF complexes that also are known to impact on WNT and other clinical/pathologic variables and found that T-stages signaling (13, 24), the exact mechanism of ADNP function, and were significantly associated with different ADNP expression the detailed dynamics of its WNT and tumor-suppressive effects scores, with a tendency of lower T-stages linked to higher ADNP in colon cancer yet remain to be determined. Moreover, since expression. Also, low tumor grade tended to associate with low ADNP knockdown also upregulated genes of other pathways ADNP expression, while the other core clinical variables age driving colon cancer progression, such as angiogenesis and

www.aacrjournals.org Clin Cancer Res; 23(11) June 1, 2017 2775

Downloaded from clincancerres.aacrjournals.org on September 28, 2021. © 2017 American Association for Cancer Research. 2776 lnCne e;2(1 ue1 2017 1, June 23(11) Res; Cancer Clin lje al. et Blaj Downloaded from D A E F Published OnlineFirstNovember30,2016;DOI:10.1158/1078-0432.CCR-16-1604 clincancerres.aacrjournals.org C B on September 28, 2021. © 2017American Association for Cancer Research. -Pos lnclCne Research Cancer Clinical Published OnlineFirst November 30, 2016; DOI: 10.1158/1078-0432.CCR-16-1604

ADNP Represses WNT in Colorectal Cancer

Figure 5. Low-dose ketamine induces ADNP, represses WNT activity, and slows tumor growth of colon cancer xenografts. A and B, Effects of in vitro treatment of SW122, HCT116, and primary colon cancer cells (P-Tu) with 100 mmol/L ketamine. A, Immunoblotting for indicated proteins on indicated time points after addition of ketamine to culture media. Numbers below immunoblots indicate fold change by densitometry. B, Dual-luciferase assays for indicated colon cancer cells transfected with TOPﬂash reporter constructs under treatment with ketamine or PBS as control. Data are mean SD, P values are t test results, n 3. C and D, Impact of daily treatment with ketamine (20 mg/kg) or PBS as control on xenograft growth of indicated colon cancer cells or primary colon cancer (P-Tu), shown as growth curves (C) and tumor-speciﬁc survival in Kaplan–Meier plots (D). P values are t test results and data are mean SE in (C) or log-rank test results in D.

EGFR signaling (25, 26), we hypothesize that tumor suppressor ADNP can be pharmacologically induced in neurons by keta- effects of ADNP may additionally transduce through other mine (19, 20) and our data demonstrate how this aspect may pathways than WNT. translate into a therapeutic approach for colorectal cancer.

Figure 4. ADNP depletion increases migration, invasion, and in vivo tumor growth of colon cancers. A–D, Effects of stable ADNP depletion by two different shRNAs against ADNP (sh1/2 ADNP) versus unspecific control shRNA (sh Ctrl) on HCT116 and SW1222 colon cancer cells. A, Immunoblotting for indicated proteins. Numbers below immunoblots indicate fold change by densitometry. B and C, Representative micrographs (left) and quantification (right) of migrated or invaded tumor cells in Transwell assays for indicated cell lines. D, Representative proliferation kinetics based on cell quantification by impedance measurements. Data are mean, n 3, P values are t test results. E and F, Effects of stable ADNP depletion (sh2 ADNP, n ¼ 8) compared with control transduction (sh Ctrl, n ¼ 8) of SW1222 colon cancer cells on xenograft tumor growth in vivo. E, Photograph and growth curves of SW1222 colon cancer xenografts transduced with indicated shRNA constructs. P values are t test results, data are mean SE. F, IHC for ADNP and the proliferation marker Ki67, and quantification of Ki67 in xenograft tumors. Data are mean SD, P values are t test results, n ¼ 8. Scale bars, 50 mm.

www.aacrjournals.org Clin Cancer Res; 23(11) June 1, 2017 2777

Blaj et al.

A Low High

Figure 6. Loss of ADNP expression indicates poor prognosis in colorectal cancer. A, Assessment of ADNP immunostaining in a collection of 221 primary human - colorectal cancers. Tumors were - assigned semiquantitative expression scores from 0 (no or barely detectable ADNP staining) to 3 (strong ADNP staining) and accordingly categorized as ADNP low (score 0–1) and ADNP high (score 2–3). Scale bar, 100 mm. B, Kaplan–Meier plots for different ADNP expression scores for tumor-speciﬁc C survival and disease-free survival indicate signiﬁcant poorer outcomes with decreasing ADNP expression. C, Kaplan–Meier plots for low and high High Low ADNP expression in all cases, and in colon cancer subsets with low or high expression of nuclear b-Catenin. P values indicate log-rank test results.

- Ratios on curves indicate the number of events over the number of patients per group. -

Treatment with subnarcotic ketamine induced ADNP, suppressed for patients with colorectal cancer with advanced disease by WNT signaling, inhibited migration and invasion of colon cancer repressing WNT through ADNP induction with low-dose keta- cells, and significantly slowed tumor growth of colon cancer mine may quite easily be tested as an add-on to existing treatment xenografts in vivo. Although the mechanism of ADNP induction regimens, as this substance and its pharmacologic characteristics by ketamine is currently unknown (19), we demonstrate that the are well studied (30). Of course, side effects of this treatment are to effects of ketamine treatment were similar to those of ADNP be carefully evaluated and strictly balanced with potential ther- overexpression, and decreased in ADNP knockout cells, suggest- apeutic benefits, especially since others reported adverse effects of ing that the tumor suppressive effects of ketamine in part depend ketamine for patients with other malignancies, such as breast on ADNP-mediated WNT repression. These findings are of specific cancer (31). interest when considering that transcription factors are usually Although ADNP was consistently linked to nuclear b-Catenin difficult drug targets due to their binding promiscuity and the expression within individual colorectal cancers, we were surprised intrinsically disordered nature of their binding sites (27). More- to find only a weak and nonsignificant overall association of over, direct targeting of WNT signaling poses substantial chal- ADNP and nuclear b-Catenin expression in our tissue collection. lenges due to complexity of its signaling cascade and cross talk We hypothesize that this may be due to different genetic back- from various other signaling pathways (8, 28, 29). The benefit grounds of colorectal cancers with differential influence of other

2778 Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research

ADNP Represses WNT in Colorectal Cancer

signaling pathways on WNT activity that may not completely management and treatment options for patients with colorectal reflect in nuclear b-Catenin expression levels (2, 32). However, cancer. in support of the idea that ADNP functions as a tumor suppressor, we found that high ADNP expression predicted better Disclosure of Potential Conflicts of Interest outcomes for cancer and disease-free survival in human colo- T. Kirchner reports receiving speakers bureau honoraria from AstraZeneca rectal cancer, while this was independent of other core clinical and Merck. No potential conflicts of interest were disclosed by the other authors. variables. Because ADNP levels separated different outcomes particularly well in cases with high nuclear b-Catenin expres- Authors' Contributions sion that have been associated with more aggressive behavior in Conception and design: C. Blaj, D. Horst some studies (33), ADNP may predominantly exhibit its pro- Development of methodology: C. Blaj, D. Horst Acquisition of data (provided animals, acquired and managed patients, tective role in this subset of colorectal cancer cases. As more provided facilities, etc.): C. Blaj, A. Bringmann, E.M. Schmidt, M. Urbischek, than 50% of colorectal cancers progress and/or develop metas- S. Lamprecht, T. Frohlich,€ G.J. Arnold, S. Krebs, H. Blum, H. Hermeking, A. Jung, tases during the course of the disease, markers predicting T. Kirchner, D. Horst prognosis and individual risk may guide personalized therapy Analysis and interpretation of data (e.g., statistical analysis, biostatistics, regimens (34). In this context, patients with low-stage colorec- computational analysis): C. Blaj, E.M. Schmidt, M. Urbischek, S. Lamprecht, € tal cancer but loss of ADNP expression may benefitfrom T. Frohlich, G.J. Arnold, S. Krebs, A. Jung, T. Kirchner, D. Horst Writing, review, and/or revision of the manuscript: C. Blaj, E.M. Schmidt, increased clinical attention and intensified or adjuvant treat- S. Lamprecht, T. Frohlich,€ S. Krebs, H. Hermeking, A. Jung, T. Kirchner, D. Horst ment protocols. In regard to ketamine treatment, since indi- Study supervision: D. Horst vidual colon cancers showed significantly different expression levels of ADNP, it may be a useful biomarker in predicting Acknowledgments therapy response, a hypothesis to be addressed in further We are grateful to Anne Kuchler€ and Sabine Sagebiel-Kohler for experimental preclinical and eventually clinical trials. assistance, Jutta Engel for clinical follow-up data, Vivien Bubb for ADNP In conclusion, we here identified ADNP as a transcription factor overexpression vectors, and the HTCR for primary colon cancer cells. that is overexpressed in colon cancer cells with high WNT signaling activity, counteracts WNT activity in these tumor cells, and Grant Support exhibits tumor suppressor functions. These characteristics may be This work was supported by a grant from the Deutsche Forschungsge- therapeutically exploited, since ADNP is inducible by low-dose meinschaft (HO4325/4-1 to D. Horst). The costs of publication of this article were defrayed in part by the payment of ketamine treatment which reduces tumor growth in preclinical advertisement in vivo page charges. This article must therefore be hereby marked in xenograft models . Moreover, in human colorectal cancer accordance with 18 U.S.C. Section 1734 solely to indicate this fact. patients, ADNP predicts superior clinical outcome. We propose that these potentials of ADNP as a prognostic marker and ther- Received June 24, 2016; revised November 18, 2016; accepted November 21, apeutic target may be considered in further trials to improve 2016; published OnlineFirst November 30, 2016.

References 1. Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin 11. Pinhasov A, Mandel S, Torchinsky A, Giladi E, Pittel Z, Goldsweig AM, et al. 2010;60:277–300. Activity-dependent neuroprotective protein: a novel gene essential for 2. The Cancer Genome Atlas Network. Comprehensive molecular character- brain formation. Brain Res Dev Brain Res 2003;144:83–90. ization of human colon and rectal cancer. Nature 2012;487:330–7. 12. Vulih-Shultzman I, Pinhasov A, Mandel S, Grigoriadis N, Touloumi O, 3. Horst D, Chen J, Morikawa T, Ogino S, Kirchner T, Shivdasani RA. Pittel Z, et al. Activity-dependent neuroprotective protein snippet NAP Differential WNT activity in colorectal cancer confers limited tumorigenic reduces tau hyperphosphorylation and enhances learning in a novel potential and is regulated by MAPK signaling. Cancer Res 2012;72: transgenic mouse model. J Pharmacol Exp Ther 2007;323:438–49. 1547–56. 13. Mandel S, Gozes I. Activity-dependent neuroprotective protein constitutes 4. Brabletz T, Jung A, Spaderna S, Hlubek F, Kirchner T. Opinion: migrating a novel element in the SWI/SNF chromatin remodeling complex. J Biol cancer stem cells - an integrated concept of malignant tumour progression. Chem 2007;282:34448–56. Nat Rev Cancer 2005;5:744–9. 14. Herbst A, Jurinovic V, Krebs S, Thieme SE, Blum H, Goke B, et al. Com- 5. Vermeulen L, De Sousa EMF, van der Heijden M, Cameron K, de Jong JH, prehensive analysis of beta-catenin target genes in colorectal carcinoma cell Borovski T, et al. Wnt activity deﬁnes colon cancer stem cells and is lines with deregulated Wnt/beta-catenin signaling. BMC Genomics regulated by the microenvironment. Nat Cell Biol 2010;12:468–76. 2014;15:74. 6. Kirchner T, Brabletz T. Patterning and nuclear beta-catenin expression in 15. Smith KJ, Johnson KA, Bryan TM, Hill DE, Markowitz S, Willson JK, et al. the colonic adenoma-carcinoma sequence. Analogies with embryonic The APC gene product in normal and tumor cells. Proc Natl Acad Sci U S A gastrulation. Am J Pathol 2000;157:1113–21. 1993;90:2846–50. 7. Cernat L, Blaj C, Jackstadt R, Brandl L, Engel J, Hermeking H, et al. 16. Humphries A, Wright NA. Colonic crypt organization and tumorigenesis. Colorectal cancers mimic structural organization of normal colonic crypts. Nat Rev Cancer 2008;8:415–24. PLoS One 2014;9:e104284. 17. Bayerlova M, Klemm F, Kramer F, Pukrop T, Beissbarth T, Bleckmann A. 8. Kahn M. Can we safely target the WNT pathway? Nat Rev Drug Discov Newly constructed network models of different WNT signaling cascades 2014;13:513–32. applied to breast cancer expression data. PLoS One 2015;10:e0144014. 9. Bassan M, Zamostiano R, Davidson A, Pinhasov A, Giladi E, Perl O, et al. 18. Bostanci O, Kemik O, Kemik A, Battal M, Demir U, Purisa S, et al. A novel Complete sequence of a novel protein containing a femtomolar-activ- screening test for colon cancer: Talin-1. Eur Rev Med Pharmacol Sci ity-dependent neuroprotective peptide. J Neurochem 1999;72: 2014;18:2533–7. 1283–93. 19. Brown BP, Kang SC, Gawelek K, Zacharias RA, Anderson SR, Turner CP, 10. Zamostiano R, Pinhasov A, Gelber E, Steingart RA, Seroussi E, Giladi E, et al. et al. In vivo and in vitro ketamine exposure exhibits a dose-dependent Cloning and characterization of the human activity-dependent neuropro- induction of activity-dependent neuroprotective protein in rat neurons. tective protein. J Biol Chem 2001;276:708–14. Neuroscience 2015;290:31–40.

www.aacrjournals.org Clin Cancer Res; 23(11) June 1, 2017 2779

Blaj et al.

20. Turner CP, Gutierrez S, Liu C, Miller L, Chou J, Finucane B, et al. Strategies 28. Niehrs C. The complex world of WNT receptor signalling. Nat Rev Mol Cell to defeat ketamine-induced neonatal brain injury. Neuroscience Biol 2012;13:767–79. 2012;210:384–92. 29. Voronkov A, Krauss S. Wnt/beta-catenin signaling and small molecule 21. Anastas JN, Moon RT. WNT signalling pathways as therapeutic targets in inhibitors. Curr Pharm Des 2013;19:634–64. cancer. Nat Rev Cancer 2013;13:11–26. 30. Kurdi MS, Theerth KA, Deva RS. Ketamine: current applications 22. Aboonq MS, Vasiliou SA, Haddley K, Quinn JP, Bubb VJ. Activity-depen- in anesthesia, pain, and critical care. Anesth Essays Res 2014;8: dent neuroprotective protein modulates its own gene expression. J Mol 283–90. Neurosci 2012;46:33–9. 31. He H, Chen J, Xie WP, Cao S, Hu HY, Yang LQ, et al. Ketamine used as an 23. Lustig B, Jerchow B, Sachs M, Weiler S, Pietsch T, Karsten U, et al. Negative acesodyne in human breast cancer therapy causes an undesirable side feedback loop of Wnt signaling through upregulation of conductin/axin2 effect, upregulating anti-apoptosis protein Bcl-2 expression. Genet Mol Res in colorectal and liver tumors. Mol Cell Biol 2002;22:1184–93. 2013;12:1907–15. 24. Wang X, Haswell JR, Roberts CW. Molecular pathways: SWI/SNF (BAF) 32. Ormanns S, Neumann J, Horst D, Kirchner T, Jung A. WNT signaling and complexes are frequently mutated in cancer–mechanisms and potential distant metastasis in colon cancer through transcriptional activity of therapeutic insights. Clin Cancer Res 2014;20:21–7. nuclear beta-Catenin depend on active PI3K signaling. Oncotarget 25. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2014;5:2999–3011. 2011;144:646–74. 33. Chen Z, He X, Jia M, Liu Y, Qu D, Wu D, et al. beta-catenin overexpression in 26. Normanno N, De Luca A, Bianco C, Strizzi L, Mancino M, Maiello MR, et al. the nucleus predicts progress disease and unfavourable survival in colo- Epidermal growth factor receptor (EGFR) signaling in cancer. Gene rectal cancer: a meta-analysis. PLoS One 2013;8:e63854. 2006;366:2–16. 34. Langan RC, Mullinax JE, Raiji MT, Upham T, Summers T, Stojadinovic A, 27. Dunker AK, Uversky VN. Drugs for `protein clouds': targeting intrinsically et al. Colorectal cancer biomarkers and the potential role of cancer stem disordered transcription factors. Curr Opin Pharmacol 2010;10:782–8. cells. J Cancer 2013;4:241–50.

2780 Clin Cancer Res; 23(11) June 1, 2017 Clinical Cancer Research

ADNP Is a Therapeutically Inducible Repressor of WNT Signaling in Colorectal Cancer

Cristina Blaj, Agnes Bringmann, Eva Marina Schmidt, et al.

Clin Cancer Res 2017;23:2769-2780. Published OnlineFirst November 30, 2016.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-16-1604

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2017/01/21/1078-0432.CCR-16-1604.DC2

Cited articles This article cites 34 articles, 7 of which you can access for free at: http://clincancerres.aacrjournals.org/content/23/11/2769.full#ref-list-1

Citing articles This article has been cited by 1 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/23/11/2769.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/23/11/2769. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.