The DACH1 gene is frequently deleted in prostate cancer, restrains prostatic intraepithelial neoplasia, decreases DNA damage repair, and predicts therapy responses

Prostate cancer (PCa), the second leading cause of death in American men, includes distinct genetic subtypes with distinct therapeutic vulnerabilities. The DACH1 gene encodes a winged helix/Forkhead DNA-binding protein that competes for binding to FOXM1 sites. Herein, DACH1 gene deletion within the 13q21.31-q21.33 region occurs in up to 18% of human PCa and was associated with increased AR activity and poor prognosis. In prostate OncoMice, prostate-specific deletion of the Dach1 gene enhanced prostatic intraepithelial neoplasia (PIN), and was associated with increased TGFβ activity and DNA damage. Reduced Dach1 increased DNA damage in response to genotoxic stresses. DACH1 was recruited to sites of DNA damage, augmenting recruitment of Ku70/Ku80. Reduced Dach1 expression was associated with increased homology directed repair and resistance to PARP inhibitors and TGFβ kinase inhibitors. Reduced Dach1 expression may define a subclass of PCa that warrants specific therapies.


INTRODUCTION
Prostate cancer (PCa), the second leading cause of death in American men, is a genetically heterogeneous disease, likely reflecting distinct genetic drivers [1]. While substratification of PCa into genetic subtypes forms the basis of rational therapy for PCa, current diagnostic tools fail to reliably distinguish aggressive tumors from non-aggressive ones in order to predict therapeutic response [2], and the lack of markers to stratify PCa cases into low-and high-risk groups results in overtreatment of 20-42% of patients [3]. New biomarkers are urgently needed for therapeutic stratification. A better molecular understanding of the disease is necessary to develop novel targeted therapies for metastatic PCa.
Defects in DNA damage repair (DDR) pathways are a hallmark of human cancer, with somatic events present in up to 20% of primary PCa [1], including BRCA2 [4], which participates in homology-directed DNA repair (HR). Defective HR due to defects in BRCA1 or BRCA2 has led to the use of poly(adenosine diphosphate(ADP)-ribose) polymerase (PARP) inhibitors in prostate cancer therapy [5]. Target-region sequencing, array-based gene expression, copy number variation (CNV) analysis, and whole-genome sequencing of tumors have reported several PCarelated genomic alterations, including copy number gains of 8q, and copy number losses of 3p, 8p, 10q, 13q, and 17p [6][7][8]. In PCa, known genetic drivers for tumor initiation include PTEN and NKX3.1 deletions, rearrangements/fusions of multiple genes (including TMPRSS2 and the oncogenic ETS transcription factor, ERG) [8], and predisposing genetic factors (including germline DNA-repair gene mutations) [9], (reviewed in [1]). Loss of heterozygosity or deletion also occurs within the 13q21 region in PCa, may include BRCA2, and is associated with highgrade prostate cancer [10][11][12].
In addition to genetic drivers of PCa, hyperactivity of the androgen receptor, inflammation [13], TGFβ activity, and DNA damage contribute to tumor progression [14]. Transforming growth factor β (TGFβ) has tumor-inhibitory activity in the early stages of prostate tumorigenesis but promotes migration, epithelial-mesenchymal transition (EMT), invasion, and metastasis in latestage disease [15,16]. DNA-dependent protein kinase (DNA-PK) is a serine/threonine kinase that, with Ku70, Ku80, XRCC4, ligase IV, and Artemis, drives non-homologous end joining (NHEJ) repair [17]. The heterodimer of Ku70 and Ku80 binds to double-strand breaks (DSBs) and recruits and activates the catalytic subunit DNA-PKC, which in turn recruits the XRCC4/ligase IV heterodimer that is responsible for rejoining the break. Inactivation of the Ku70 or Ku80 genes in 5 mice leads to hypersensitivity to radiation, malignant transformation [18,19], and an associated increase in HR [20,21], as binding to both ends of a two-ended DSB stabilizes contacts between Ku heterodimers, tethering the DNA ends and preventing access by the HDR machinery [21].
The Drosophila Dac gene was initially cloned as a dominant inhibitor of the hyperactive EGFR, Ellipse [22]. The human Dachshund1 (DACH1) gene encodes a DNA-binding protein similar to the winged helix/Forkhead subgroup of the helix-turn-helix family. Cyclic amplification and selection of target (CAST), together with ChIP, identified DACH1 DNA binding sequences that resemble Forkhead binding sites [23]. Furthermore, expression of the DACH1 gene is reported to be reduced in PCa, and DACH1 overexpression inhibited PCa cell line growth [24].
Given the importance of identifying molecular genetic events governing PCa onset and progression, in the work reported here we investigated the role of the Dach1 gene in PCa progression in transgenic mice. We identify a novel role for DACH1 in maintaining genomic stability through the regulation of DNA repair. A reduced abundance of DACH1 in human prostate cancer, associated with either gene deletion or promoter DNA methylation, correlated with poor outcomes. In prostate OncoMice, prostate-specific deletion of the Dach1 gene enhanced prostatic intraepithelial neoplasia (PIN), and was associated with increased TGFb activity and DNA damage. Mechanistically, we show that DACH1 is recruited to sites of DNA damage, augmenting recruitment of Ku70/Ku80, restraining homologous end joining (HR). Reduced Dach1 expression was associated with resistance to PARP inhibitors and TGFb kinase inhibitors. Our results suggest reduced Dach1 expression may define a subclass of PCa that warrants specific therapies. 6
To assess the relationship between DACH1 homozygous deletions and outcomes, we queried copy number and overall survival data for three cBioPortal cohorts (TCGA PanCancer Atlas 2018, SU2C 2019, and MCTP) (n=667 tumor samples). Using a somatic copy number threshold of -2 to segregate the data into samples with altered vs. unaltered DACH1, we generated a Kaplan-Meier plot for overall survival (Fig. 1D) (Fig. 1D).
In addition to gene deletion, the abundance of DACH1 mRNA may be affected by DNA methylation. We assessed DACH1 DNA methylation and RNAseq expression data from firebrowse.org, for the n=333 primary tumor samples and n=43 adjacent tissue normal samples in the cBioPortal TCGA 2015 cohort [26]. Of 29 Illumina 450K probes associated with DACH1, 26 had substantially complete data, and 15 of these probes were within ~3kb of the DACH1 transcriptional start site (TSS) (Fig. S3B). Beta values (β) estimate DNA methylation level using the ratio of intensities between methylated and unmethylated alleles, and are continuous variables between 0 and 1, with 0 being unmethylated and one fully methylated. Probe cg13726218, which cBioPortal associates with DACH1, had both a wide range of beta values (Fig. S3B, probe indicated by a red asterisk) and a strongly-negative Spearman coefficient with DACH1 RNA-Seq expression (rho=-0.41, FDR=7.6x10 -14 ) (Fig. 1E). The additional fourteen of the 15 probes in proximity to the TSS had narrow ranges of low beta values in primary tumor samples (Fig. S3B).
Median beta values were similar in primary tumors and adjacent tissue normals for 25 of the 26 probes (Fig. S3C).
The relationship between cg13726218 DNA methylation and gene expression could be expressed as a linear negative correlation ( Fig. S3; n=308 DACH1-diploid samples, Pearson correlation cor = -0.40, p = 2.9x10 -13 , alternative hypothesis: true correlation ¹ 0). As a nonlinear alternative, a fit Generalized Additive Model (GAM) had a minimized generalized cross-validation (GCV) score of 0.83 and an adjusted R 2 of 0.16 (Fig. S3E). Collectively, these results indicate that increasing promoter DNA methylation contributed to a reduction in DACH1 mRNA.
Immunostaining for DACH1 protein was conducted in tissue microarrays (TMAs) of 68 cases in triplicate 1 mm biopsy cores of primary prostate cancer with matched normal/benign tissue. DACH1 was positive in all normal/benign prostate epithelia but was negative in 9 of the 68 cases (13.2%). 89% (8/9) of DACH1-negative cases were of high grade (Gleason 4+3=7 or higher) whereas of DACH1-positive cases only 54% (32/59) were high grade (likelihood-ratio test p=0.034). Representative images of DACH1-positive and -negative cases of prostate cancer are shown in Fig. S4.

DACH1-deletion tumors constitute a prostate cancer subtype.
Using DACH1 copy number data for samples from The Cancer Genome Atlas (TCGA) [26], we identified a group of 29 of 333 tumors with DACH1 homozygous deletion (DACH1 subtype, Fig. 2A). In a prior study, androgen receptor (AR) activity varied widely and in a subtype-specific manner, with SPOP and FOXA1 mutant tumors having the highest levels of AR-induced transcripts [26]. Comparison of AR activity levels (using an AR score derived from the expression of AR target genes [26]) showed a significant increase in the DACH1-deletion group as compared to normal samples (P=2x10 -5 by t-test) and ERG mutation groups (P=0.003 by t-test) ( Fig. 2A). AR activity expressed as a z score for each prostate cancer subtype also showed an increase in the DACH1 deletion group as compared to normal samples (Fig. S5A). However, the levels of AR mRNA and protein were not statistically significantly different (p>0.05) between DACH1 genotypes (Fig. 2B).

Dach1 deletion in the prostate promotes prostatic intraepithelial neoplasia (PIN). Given that
DACH1 and RB1 deletion may co-occur in prostate cancer (Fig. 1B), we sought to determine the functional significance of DACH1 gene deletion, independently of RB1, in the onset and progression of PCa, using a murine model. Dach1 homozygous null mice die at birth. Given this, in order to determine the role of endogenous Dach1 in mediating prostate transformation, multigenic mice were generated by crossing conditional Dach1 gene deletion mice [30] with prostate-specific Cre transgenics, using Probasin-Cre4 (Pb-Cre4) transgenic mice, which express Cre in both the basal and luminal prostatic epithelia, then further intercrossing with the TRAMP model of prostate cancer (Fig. 3A). The TRAMP model has been extensively characterized, and TRAMP mice develop PIN after 12 weeks. To follow efficient temporal and spatial regulation of Cre recombination in vivo, we intercrossed these bi-transgenic mice with a double-fluorescent Cre reporter mouse that, prior to Cre-mediated excision, expresses membrane-targeted tandem dimer Tomato (mT) and, after excision, expresses membrane-targeted green fluorescent protein (mG).
The mice thus cointegrate four transgenes (Fig. 3A). 3T3 cells derived from Dach1 -/mouse showed deletion of the Dach1 gene did not reduce the abundance of pRB (Fig. S8). Rather the abundance of pRB Ser807/811 was increased by Dach1 gene deletion when normalized to the protein loading control lamin B1 (Fig. S8). Genomic analysis of tail DNA by PCR confirmed the presence of the transgenes in these mice (Fig. S9A). Cre-induced GFP was expressed in the prostate epithelium ( Fig. S9B), whereas no GFP was observed in the absence of Cre recombinase ( Induction of an epithelial-mesenchymal transition (EMT) plays a role in both PCa metastatic progression and resistance to treatment [31]. DACH1 was detectable in human PCa cell lines (Fig. 4E,F), including PC3 cells, in which TGFb induced EMT, as evidenced by induction of the mesenchymal marker vimentin (Fig. 4G), and an increase in the proportion of cyclin D1 located in the cytoplasm. In addition, PC3 cell lines stably expressing DACH1 showed inhibition of TGFb target gene expression (Fig. 4H, blue arrows) [30]. In order to examine further the interaction between DACH1 and Ku70/Ku80 a mass spectrometry analysis was conducted. DACH1 protein complexes were prepared from HEK 293T cells transfected with a FLAG-DACH1 expression vector (Fig. S12A). DACH1-associated proteins were resolved on a 4-10% Tris-HCl gel and silver-stained. The proteins recovered from the gel were subjected to in-gel tryptic digestion and sequential MS/MS. Two excised bands, corresponding to 70 and 80 kDa were identified as ATP-dependent DNA helicase 2 subunit Ku70 and ATP-dependent DNA helicase 2 subunit 2 (Ku80). 293T cells were transfected with expression vectors for FLAG-tagged DACH1 wild type, FLAG-DACH1 DS domain deletion (DDS), FLAG-DACH1 C term (Fig. S12B). Immune precipitation of FLAG-tagged DACH1 coprecipitated Ku70 and Ku80. Therefore, we considered the possibility that DACH1 may interface DACH1 enhances DNA repair and homologous recombination. Nuclear bodies marked by the DNA damage response protein p53 binding protein 1 (53BP1) [32] participate in the cellular response to DNA damage. 53BP1 relocates to nuclear foci within minutes after exposure of cells to ionizing radiation (IR). Consistent with the finding that endogenous DACH1 governs the DDR, shDACH1 increased basal and ATO-induced 53BP1 nuclear foci formation in LNCaP cells (Fig.   7A).
The comet assay is a surrogate assay for measuring double-stranded DNA breaks in the cell.
When cells are electrophoresed in neutral pH, the image looks like a comet with a distinct head composing intact DNA and a tail consisting of damaged DNA, primarily double-strand breaks (DSBs). In doxorubicin-treated 3T3 cells (2 μM, 18 hrs) the mean comet tail moment was 22 in Dach1 +/+ cells, but 59 in Dach1 −/− cells (Fig. 7B,C), indicating that endogenous Dach1 increases repair of damaged DNA.

DISCUSSION
Gene deletion within the 13q21 region has been associated with high-grade prostate cancer.
In an independent cohort enriched for early-onset prostate cancer [27] respectively. In the current studies, we identified homozygous deletions of DACH1 in between 3 and 18% of prostate cancers in six distinct cohorts. DACH1 is located within the 13q21.31-q21.33 region, which is deleted in poor-prognosis prostate cancer [10][11][12]. Herein, homozygous DACH1 deletions correlated with reduced overall survival (medians of 84 vs. 120 months, N=667). DACH1 deletion co-occurred with deletion of SPOP and BRCA2, consistent with TCGA analysis demonstrating multiple pathways may be disrupted in a given tumor type [35]. The abundance of DACH1 was also reduced by DNA methylation, and low DACH1 gene expression was significantly correlated with earlier biochemical recurrence. DACH1 is located on chromosome 13q. In a subset of DACH1-deleted prostate cancers, we showed that DACH1 was co-deleted with RB (in 5 databases, co-homozygous-deletion of RB with DACH1 occurred in 15%, 5.3%, 0%, 1.4%, and 9%, Fig. S2). The current studies were therefore designed to test the importance of the additional DACH1 deletion in the presence of inactivated RB. Such an approach is necessary to define the additional impact of DACH1 in the onset and progression of PCa. Our studies were therefore designed to distinguish whether DACH1 deletion was an "innocent bystander", given the genomic proximity of DACH1 to RB. The TRAMP mouse[36-39], one of the most widely accepted in PCa research [40], inactivates pRB via the SV40 Large T antigen, and recapitulates with ~100% penetrance multiple aspects of the human disease, including prostatic intraepithelial neoplasia (PIN) lesions, and multifocal invasive carcinoma; and emulates histological and molecular events of human PCa [41]. With time, beyond which our studies were conducted, progression to castration-resistant prostate cancer [42][43][44], including castrate-resistant PCa (CRPC) [45] and the aggressive therapy-induced neuroendocrine prostate cancer tNEPC [42][43][44], which occurs with metastasis to distant organs [40] including the skeleton [46]. Analysis of the prostates of transgenic mice at 15 weeks demonstrated that genetic deletion of Dach1 in TRAMP mice correlated with increased Ki-67, histological features of PIN progression, increased TGFb activity, and increased evidence of DNA damage. An increased PIN in the DACH1 prostate-specific deletion samples (PbCre:Dach1 fl/fl ) was demonstrated by increased cellular proliferation, loss of cellular polarity (black arrow) (Fig. S10), nuclear enlargement (white arrow) (Fig. S10), and the presence of nucleoli (gray arrow) (Fig. S10). Dach1 deletion in the prostate resulted in PIN but not tumorigenesis in the time frame assessed. Thus, deregulation of Dach1 alone, like CHD[47], ERG1 [48], or ETV1 [49], is insufficient to drive prostate cancer in the mouse prostate. The genetic disruption of Dach1 did not disrupt RB1 gene expression and was associated with enhanced pRB Ser807/811P . In PCa cell lines, the reintroduction of DACH1 expression reduced RB abundance (Fig. S8B). These findings suggest that the effect of Dach1 deletion in the prostate is not mediated by coincident loss of RB1. These results are the first to show that endogenous Dach1 restrains features of tumorigenesis in vivo, and are consistent with prior correlative studies showing reduced DACH1 abundance in tumors of the brain, ovary, lung, uterus, non-small cell lung cancer, hepatocellular carcinoma, breast, and prostate cancer [24,50].
In prior tissue culture-based analysis, DACH1 retrained TGFb signaling by association with NCoR/SMAD4 [51,52]. The current studies extend these findings by showing that endogenous DACH1 restrains TGFb signaling in the murine prostate in vivo. TGFb signaling was the most upregulated pathway by gene expression analysis in the prostate of prostate-specific Dach1-deletion prostate cancer oncomice. TGFb activity was induced in the Dach1 knockout prostate tissue in vivo and assessed by the induction of SMAD2 Ser465/467P . Conversely, increased expression of DACH1 in PC3 cells reduced TGFβ activity and expression of TGFβ target genes.
In proliferation assays, DACH1 deletion-mediated TGFb hyperactivation led to relative resistance to the growth-inhibitory effects of the TGFbKRi in the presence of doxorubicin. In the current studies, the inhibition of TGFb with the TGFbRki (LY2157299) reduced comet tail formation, consistent with recent studies in which Tgfb reduced DNA repair and increased comet tail formation in a lung cancer cell line [53]. The reduction in mean tail movement by the TGFbRki (LY2157299) was defective in the Dach1 -/cells, consistent with increased TGFb signaling in the Dach1 -/cells. Increased TGFβ signaling has been strongly linked to PCa [54][55][56] promoting therapeutic resistance, cell invasiveness and tumor metastasis [57]. The role of TGFb hyperactivation in DACH1 deletion prostate cancer warrants further analysis.
In the work reported here, DACH1 was shown to promote the repair of damaged DNA.
Prostate-specific deletion of the Dach1 gene in TRAMP mice correlated with increased function is associated with PARP resistance and restoration of HR [59,60]. Mechanistically, Ku proteins bind to both ends of a two-ended DSB, stabilizing contacts between Ku heterodimers and tethering the DNA ends, thereby preventing access to the HDR machinery [21]. The mechanism by which DACH1 binding to Ku70/Ku80 induces PARP resistance remains to be further determined; however, Dach1 -/cells appear to be functionally defective in Ku protein recruitment.
Inactivation of the Ku70 or Ku80 genes in mice leads to hypersensitivity to radiation and malignant transformation [18,19]. Herein, DACH1 shRNA enhanced the radiation sensitivity to DNA-PK inhibition in colony assays. Collectively, these studies show that a loss of Dach1 abundance is associated with resistance to PARPi and TGFbKi, with enhanced sensitivity to radiation and doxorubicin. In addition, a growing body of evidence has identified somatic and, more recently, germline mutations of DNA repair genes in PCa [64]. The ongoing identification and testing of additional genes governing DNA repair may enhance the precision of guided targeted therapies of PCa.

Genetic and Epigenetic Analysis.
To generate profiles of SNP6-based GISTIC2 G-scores along the GRCh37/hg19 human reference genome, and on chromosome 13, we downloaded a 'scores.gistic' file for n=492 TCGA legacy PCa (PRAD) primary tumors from firebrowse.org, and an hg19 chromosome length file from genome.ucsc.edu. We generated graphics with a custom R script, and used the R package karyoploteR v1.14.1 to generate the banded chromosome 13 graphic.
To compare DACH1 copy number gene status in between primary vs. metastatic lesions, homozygous and heterozygous deletions in primary and metastatic tumors in different datasets, we used cBioPortal (http://www.cbioportal.org/) [65].
To assess the relationship between DACH1 homozygous deletions and outcomes, we queried mutation, copy number and overall survival data for three cBioPortal cohorts (TCGA PanCancer Atlas 2018, SU2C 2019, and MCTP), which offered data for a total of n=667 tumor samples. Using a copy number threshold of -2 to segregate the data into samples with altered vs. unaltered DACH1, we generated a Kaplan-Meier plot for overall survival using R's survival v3.2-7 package, and calculated median times for altered vs. unaltered DACH1 with a custom R script.
Comparative analysis of DACH1 and PTEN gene status in all PCa patients and between primary vs. metastatic lesions in different datasets was performed using data from cBioPortal (http://www.cbioportal.org/). Selected studies were identified based on query criteria and analyzed using default parameters [65]. Data for analysis of the association between low DACH1 expression with AR activity was derived from the supplementary data of [26], grouped by mutation status of ERG, ETV1/4/FLI1, FOXA1, "other" and adjacent tissue normal samples. Additional DACH1 deletion data for 333 samples from the study were downloaded from cBioPortal [65] and DACH1 homozygous deletions were defined as a thresholded GISTIC value of -2. All tumor and normal groups were tested for differences in AR score, AR mRNA, and AR protein levels, using twosample two-tail Student's t-test; results with p<0.05 were considered significant.
To assess the effect of DNA methylation on DACH1 gene expression, we downloaded from firebrowse.org the Illumina 450K DNA methylation data and RSEM gene expression data for n=498 TCGA 'legacy' PCa (PRAD) primary tumors and n=50 adjacent normal tissues. For the 26 DACH1 DNA methylation probes for which DNA methylation data was largely complete, we 18 calculated Spearman correlations between DNA methylation beta values and gene-level RSEM expression in primary tumors with R's cor.test, then used R's p.adjust to correct the p-values for multiple hypothesis testing. We assessed scatterplots of DNA methylation beta vs. DACH1 gene expression for all 26 DNA methylation probes (data not shown). For each probe, we calculated median beta values for primary tumors and adjacent normal tissue. We compared locations of DNA methylation probes to the DACH1 gene structure by transforming the probe locations into a 'bed' file and the Spearman correlation coefficients into a 'bedgraph' file (genome.ucsc.edu/FAQ/FAQformat.html), and then displaying both in the UCSC hg19 genome browser (data not shown). This analysis identified that 15 of the 26 DACH1 DNA methylation probes were near that gene's TSS, including probe cg13726218 (chr13:72438250). This probe had both the largest negative correlation with DACH1 gene expression, and a wide range of beta values (Figs. S3A, B). Together, these three factors suggested that this probe represents how DNA methylation can influence DACH1 gene expression, and this was the probe that cBioPortal reported for DACH1.
To assess the relationship between low expression of DACH1 and outcomes, we downloaded DACH1 RSEM expression Z-scores (considering all samples), for the TCGA cohort [26], i.e., for the n=290 of 333 tumor samples that had RSEM data. We defined samples with "altered" DACH1 expression as those in which the gene's Z-score was below -2.0, which is the default threshold value at cBioPortal. Using the PanCancer outcomes [28], we segregated the samples by low vs not-low RSEM. The R survival v3.2-10 package returned a Kaplan-Meier logrank P value of 0.028 for Progression Free Interval (PFI) outcome data. Biologicals) were used as described [66]. ImageJ software was used in IHC quantification.

Microarray analysis.
Total RNA was prepared from ventral prostates of 3 Dach1 WT (Probasin-Cre-Dach1 wt/wt ROSA26 mT/mG -TRAMP) and 3 Dach1 KO (Probasin-Cre-Dach1 fl/fl ROSA26 mT/mG -TRAMP) (15w) using the RNeasy kit from Qiagen following the manufacturer's instructions (Qiagen). RNA was labeled for hybridization with mouse Clariom D arrays (Applied Biosystems), and analysis was performed as previously described [30,67]. Gene set enrichment analysis for upstream regulators responsible for a significant number of changed genes was done using QIAGEN's Ingenuity® Pathway Analysis software (IPA®, QIAGEN Redwood City, www.qiagen.com/ingenuity) with the "Upstream regulator" option; regulators that passed the P <0.01 threshold, had at least ten significantly affected target genes, and had predicted activation states (|Z|>0.5) were reported. After coverslipping, the slides were scanned using Pannoramic 250 (3DHISTECH Ltd., Budapest, Hungary), and images were generated using Slideviewer software. Likelihood-ratio tests were used for statistical analysis of DACH1 status and Gleason score. type I (TGF-βRI) kinase inhibitors LY2157299 and LY363947 were bought from Selleckchem. 21 The expression vectors encoding DACH1 [69], Ku70 (RFP-Ku70) and Ku80 (RFP-Ku80) [70], shDACH1 [71] were previously described. The EGFP-DACH1 expression plasmid was made by inserting the human DACH1 cDNA into the HindIII and BamHI sites of the pEGFP-C1 vector.
Western blot analysis. Whole-cell lysates, nuclear lysates, or cytoplasmic lysates were separated by 8%-11% SDS-PAGE gel and the proteins were transferred to a nitrocellulose membrane for Western blotting, as previously described [30,67]. The bands were detected using the enhanced chemiluminescence detection system (Thermo Fisher Scientific #34578). The following antibodies Cell Proliferation and Comet Assays. Cells were seeded into 96 well plates in normal growth medium, and cell growth was measured daily by methylene blue assay [72]. Neutral pH comet assays were conducted as previously described [73] using the CometAssay Kit (Trevigen). After treatment with 2 µM doxorubicin or control for 18 hrs, cells were harvested and mixed with lowmelting temperature agarose. After lysis, electrophoresis was conducted at 1V/cm for 20 minutes.
Visualization involved SYBR Gold dye for 30 minutes and a Nikon C2+ Confocal Microscope with a 20× objective. Average tail moments from 55-72 cells per sample were obtained using OpenComet software (http://www.cometbio.org/index.html) [74]. agitation. The digestion products were applied onto a MALDI plate as described [75].
DNA repair assays. The DNA repair reporter assays for homologous repair (DR-GFP) assays in U2OS cells were conducted as previously described [33,34]. pCAGGS-NZEGFP, a plasmid encoding expressed GFP, was a transfection efficiency control. The DNA repair activity was shown as (RSceI-RpCAGGS)/RNZEGFP. RI-SceI, RpCAGGS, and RNZEGFP represent the ratio of GFP positive cells in I-SceI, pCAGGS-BSKX (vector control for I-SceI expression plasmid), and NZEGFP transfected cells respectively.    [26], showing candidate genetic drivers ERG, ETV1/ETV4/FLI1, SPOP, FOXA1, and unknown. Samples with DACH1 homozygous (deep) genetic deletions (29/333) are shown as an additional subtype. The AR score (the average of the AR target gene expression) refers to a group of AR-responsive genes [26], and together with the expression Z-score of the AR target genes, are shown as colorimetric scales. The AR score-based gene names are shown. The androgen receptor (AR) activity, inferred by the induction of AR target genes, was increased in DACH1 homozygous ('deep') deletion PCa compared with normal (P=2x10 -5 by t-test) and ERG mutation groups (P=0.003 by t-test). (B). AR mRNA and AR protein levels, shown for each DACH1 deletion sample, were not significantly different. (C). The iCluster [29], mRNA cluster, and SCNA (somatic copy-number alteration), and DNA methylation status are shown for the PCa classified by the corresponding gene deletion subtypes. (D). DACH1 homozygous deletions were enriched for iCluster 2 and 3 [29], mRNA cluster 2 (P=0.0003 by Fisher exact test, SCNA ("more" somatic copy-number alteration, P=0.0004 by Fisher exact test), but not for DNA methylation. Representative immunohistochemistry with results shown as mean ± SEM for Ki-67 (n=20, 4 separate mice for each genotype, 5 views per mouse) (D), Beclin 1 (n=9, 3 separate mice for each genotype, 3 views per mouse) (E); and AR (n=15 for Dach1 wt/wt mice, 3 separate mice, 5 views per mouse) (n=12 for Dach1 fl/fl mice, 3 separate mice, 2 views for one mouse and 5 views for other two mice) (F). Scale bars, 50 μm. A Student's t-test was performed for all comparisons.