key: cord-0322577-t1y1m5jc
authors: Whitlock, Nichelle C.; Trostel, Shana Y.; Wilkinson, Scott; Terrigino, Nicholas T.; Hennigan, S. Thomas; Lake, Ross; Carrabba, Nicole V.; Atway, Rayann; Walton, Elizabeth D.; Gryder, Berkley E.; Capaldo, Brian J.; Ye, Huihui; Sowalsky, Adam G.
title: MEIS1 down-regulation by MYC mediates prostate cancer development through elevated HOXB13 expression and AR activity
date: 2019-11-20
journal: bioRxiv
DOI: 10.1101/848952
sha: 9a6cc4dd6b342a2a6ebbf5bd003c757fb7b40720
doc_id: 322577
cord_uid: t1y1m5jc

Localized prostate cancer develops very slowly in most men, with the androgen receptor (AR) and MYC transcription factors amongst the most well-characterized drivers of prostate tumorigenesis. Canonically, MYC up-regulation in luminal prostate cancer cells functions to oppose the terminally differentiating effects of AR. However, the effects of MYC up-regulation are pleiotropic and inconsistent with a poorly proliferative phenotype. Here we show that increased MYC expression and activity are associated with the down-regulation of MEIS1, a HOX-family transcription factor. Using RNA-seq to profile a series of human prostate cancer specimens laser capture microdissected on the basis of MYC immunohistochemistry, MYC activity and MEIS1 expression were inversely correlated. Knockdown of MYC expression in prostate cancer cells increased expression of MEIS1 and increased occupancy of MYC at the MEIS1 locus. Finally, we show in laser capture microdissected human prostate cancer samples and the prostate TCGA cohort that MEIS1 expression is inversely proportional to AR activity as well as HOXB13, a known interacting protein of both AR and MEIS1. Collectively, our data demonstrate that elevated MYC in a subset of primary prostate cancers functions in a negative role in regulating MEIS1 expression, and that this down-regulation may contribute to MYC-driven development and progression.

Locally advanced prostate cancers harbor a limited number of recurrently altered genes whose expression change at the earliest stages of tumor development. These include down-regulation of the tumor suppressors NKX3-1 and PTEN (often due to genomic deletion), up-regulation of ERG (due to fusion with TMPRSS2), and up-regulation of MYC, which often co-occurs with a single-copy gain of chromosome 8q24 [1] [2] [3] [4] [5] . Interestingly, up-regulation of MYC in most neoplastic tissues is a very early event that contributes to self-renewal and proliferation, but localized prostate cancer (PCa) is not hyperproliferative and focal amplification of MYC is rare [6, 7] . In part, the effects of the androgen receptor (AR) in terminally-differentiated luminal prostate cells are disrupted by MYC and other cofactors including FOXA1 and HOXB13 to re-engage proliferative processes during tumorigenesis [7] [8] [9] [10] [11] .

Recently, increased awareness that the vast majority of prostate cancers are indolent has led to increased molecular profiling of tumor biopsies prior to definitive treatment. Although MYC expression has been observed in the cancer precursor high-grade prostatic intraepithelial neoplasia, MYC expression in indolent-appearing tumor cells predicts the presence of higher-grade disease and is associated with poor differentiation [3] [4] [5] . In localized prostate cancers, up-regulated MYC has been further associated with alterations in nucleoli structure, concomitant with increased biogenesis of ribosomal RNA, increased purine metabolism and expression of the telomere RNA subunit TERC [12] [13] [14] . These effects contrast sharply with the phenotype of highly amplified MYC, which is enriched in metastatic and treatment-resistant prostate cancers, and mirrors the role of MYC in other cancer types via effects on AKT to contribute towards cell division and survival [15] [16] [17] .

Nonetheless, even in primary PCa, MYC protein expression is diffuse and heterogeneous [3] .

We sought to examine the genetic contribution of MYC to the development and progression of primary PCa in the context of dysregulated growth by using anti-MYC immunohistochemistry and performing laser capture microdissection on populations of human prostate tumor cells with varying 6 expression of MYC protein. We show using transcriptome profiling that increased MYC activity is inversely correlated with expression of Myeloid Ecotropic viral Insertion Site 1 (MEIS1), a transcription factor that interacts with and regulates the activity of HOX homeodomain transcription factors, including HOXB13 [18] [19] [20] . We determined that MYC binding to the MEIS1 locus decreases as MYC levels increases, and that MEIS1 expression is negatively correlated with HOXB13 expression and AR activity.

This research was conducted in accordance with the principles of the Declaration of Helsinki. All 

Gene-level normalized FPKM expression values for The Cancer Genome Atlas (TCGA) RNAseq data were downloaded from the NCI Genomic Data Commons (https://gdc.cancer.gov). Cases were filtered for tumor samples within each organ type and cases with missing gene expression values were removed. Single-sample gene set enrichment analysis was performed on the GenePattern server (https://cloud.genepattern.org) using the ssGSEAProjection module version 9.1.1 with the following parameters: weighting exponent: 0.75; combine mode: combine.add; sample normalization method: none. For each TCGA tumor type, ssGSEA projection values were obtained for a 17-gene MYC-independent proliferation signature and a 54-gene MYC activity signature [21] .

Serial sections of formalin-fixed, paraffin-embedded tissues were cut at 6 μm thickness onto 

Stained slides were scanned using the 20× objective (Plan-Apochromat, NA 0.8) with brightfield illumination on an AxioScan.Z1 (Zeiss). All slides were reviewed by a board-certified surgical pathologist following the 2014 ISUP guidelines [22] . Whole slide .CZI files of Ki67-and MYCstained slides were imported into Definiens Developer XD 64. The magnification for each analysis was set to 40×, with 0.11 μm/pixel for both Ki67 and MYC solutions. IHC stain was identified as brown chromogen. The tumor cells were counted using their nuclear stain.

For Ki67, composer magnification was set to 6× with 12 training subsets at segmentation level 9.

Segments were classified as tumor or stroma, with normal glands excluded. Within the tumor pattern, cellular analysis magnification was set to 10× with 12 training subsets, and within nuclear detection, thresholds were 0.2 for hematoxylin and 0.5 for brown chromogen with typical nuclear size set to 30 μm. Nucleus classification: low vs. medium at 0.65; medium vs. high at 0.86.

For MYC analysis, composer magnification was set to 7.5× with 12 training subsets at segmentation level 7. Segments were classified as tumor or stroma, with normal glands excluded.

Within the tumor pattern, cellular analysis magnification was set to 10× with 12 training subsets, and within nuclear detection, thresholds were 0.1 for hematoxylin and 0.45 for brown chromogen: 0.45 with typical nuclear size set to 30 μm. Nucleus classification: low vs. medium at 0.5; medium vs. high at 0.7.

The total number of positively stained nuclei were reported, along with distribution of low-, medium-, and high-intensity stained nuclei for each tumor focus. A percent positive index score was calculated using a weighted average divided by the total number of nuclei, where index = [ (1 × nuclei stained low) + (2 × nuclei stained medium) + (3 × nuclei stained high) ] ÷ (3 × total nuclei).

RNA knockdown of MYC expression was achieved using the SMARTvector lentiviral shRNA system (Dharmacon) and a standard second-generation packaging system. hCMV-TurboGFP 

Laser capture microdissection (LCM) was performed as previously described [23] . For withinpatient analyses (high and low MYC levels in the same patient), cases were selected based on having regions of differential MYC staining intensity by IHC, but concordant ERG and PTEN status by IHC. Areas of high MYC were defined by moderate (++) or high (+++) staining in at least 30% of cancer cells, while areas of low MYC were defined by weak (+) or absent (-) staining in at least 75% of cancer cells, with intense staining limited to 10% of cancer cells. LCM was guided by review of serially-sectioned slides stained with MYC, visualized using ZEN Browser (Zeiss) on an adjacent monitor as reference. High or low MYC cancer glands from each case were collected on separate caps and a photomicrograph was acquired for the purposes of estimating tumor cell purity in each sample. For between-patient analyses, MYC status was ascertained by IHC and assigned to the foci of tumor cells subjected to LCM. The distinction between within-patient and between-patient analyses is that all within-patient cases harbored both MYC-high and MYC-low populations of tumor cells.

RNA was extracted using the RNeasy FFPE Kit (Qiagen) following the manufacturer's protocol with modifications. Briefly, the film to which LCM cells adhered was removed using a disposable blade, immersed into Buffer PKD with proteinase K, digested for 15 min at 56°C, followed by a 15minute incubation at 80°C. After centrifugation and DNase treatment to remove genomic DNA, concentrated RNA was purified using RNeasy MinElute spin columns and eluted with 30 μL RNasefree water. RNA yields were quantified using RiboGreen reagent (Life Technologies).

RNA was purified using the RNeasy Plus Mini Kit (Qiagen) and quality tested using the High Sensitivity RNA ScreenTape (TapeStation 4200, Agilent Technologies). Expression levels of MYC and MEIS1 transcript were quantified using TaqMan Fast Virus 1-Step Master Mix (Life Technologies) using PrimeTime qPCR Probe Assays from Integrated DNA Technologies in a 3:1 ratio of primer to 5′ 6-FAM and 3′ TAMRA labeled probe. Sequences of the primer-probe sets are listed in Supplementary Table 1 . qPCR was performed in duplex with GAPDH endogenous control TaqMan assays (Life Technologies), and relative transcript levels were quantified using the 2 -ΔΔCT method [24] .

Total cell lysates were isolated using RIPA buffer (Pierce) containing protease and phosphatase inhibitors (Pierce). 15 μg of soluble protein per well were separated by SDS-PAGE and transferred to nitrocellulose membranes. Blots were blocked for 1 hour in 5% nonfat dry milk in TBS/0.05% Tween-20 (TBS-T) and incubated with primary antibodies at 4°C overnight. Primary antibodies used were anti-MYC (Y69; ab32072, Abcam, 1:2000 dilution) and anti-GAPDH (6C5; MAB374, Millipore Sigma, 1:1000 dilution). After washing with TBS-T, the blots were incubated with antirabbit or anti-mouse horseradish peroxidase-conjugated secondary antibody (Jackson ImmunoResearch, 1:10,000 dilution) for 1 hour and washed 4 times. Chemiluminescence was detected with Western Lightning Plus ECL (Perkin Elmer) and recorded using a ChemiDoc Imaging System (Bio-Rad).

ChIP was performed according to the ChIP-IT High Sensitivity Kit (Active Motif) protocol.

Briefly, cultured LNCaP cells with control or MYC knockdown constructs were crosslinked, quenched, lysed, and the chromatin sheared to a size between 200-1000 bp. Drosophila melanogaster chromatin and antibody spike-in controls were added per the protocol. For each ChIP, 30 μg of chromatin were incubated with anti-MYC antibody (ab32072, Abcam, 1:100 dilution) on an end-to-end rotator overnight at 4°C. Antibody-bound protein/chromatin complexes were immunoprecipitated with Protein A/G beads, reverse-crosslinked, and DNA purified. Quality of ChIP-enriched DNA was assessed using the ChIP-IT qPCR Analysis Kit (Active Motif).

Purified ChIP DNA was assembled into libraries using the TruSeq ChIP Library Preparation Kit (Illumina) according to the manufacturer's protocol, pooled in equimolar ratios, and sequenced on a NextSeq 500 (Illumina) with 76 cycles of single-end sequencing. All samples had 90% of bases at Q30 or above, with yields between 26-42 million pass filter reads per sample. Samples were adaptortrimmed with Trim Galore (https://github.com/FelixKrueger/TrimGalore) and reads were mapped to the human genome GRCh37 and Drosophila melanogaster genome dm3 with BWA-MEM [25] .

Aligned reads were deduplicated using Picard Tools (https://broadinstitute.github.io/picard). Peak calling was performed using MACS2 [26] callpeak with the parameters -g hs and -q 0.01, differential binding of peaks was ascertained using DiffBind [27] with default parameters, and motif analysis was performed using HOMER [28] findMotifsGenome.pl with the parameters -size 200 and 13 -len 8,10,12. To normalize for technical and genome-wide variability, reads mapping to dm3 were used to calculate reads per by million mapped dm3 reads (RRPM) as described [29] .

Whole transcriptome profiling of LCM tissue was performed at the CCR Illumina Sequencing Facility using the Illumina HiSeq v4 sequencing system to a depth of 50 million reads (50 cycles, paired-end). Sample reads were trimmed for adapters using Trim Galore before alignment to the GRCh37 reference genome using STAR [30] with default parameters. Aligned reads were quantified using featureCounts [31] with default parameters, and differential expression analysis between MYC-high and MYC-low samples was performed using edgeR [32] . The log 2 counts per million (CPM) values from the output were used for downstream analysis.

From the TCGA PRAD dataset used in the pan-cancer analysis, the 20 cases with highest MYC activity signature and the 20 cases with the lowest MYC activity signature were re-analyzed using the same pipeline as tissue for determining differentially expressed genes. FASTQ files for those cases were retrieved from the NCI Genomic Data Commons via access to dbGaP phs000178.

Gene set enrichment analysis (GSEA) was performed in clusterProfiler [33] comparing MYChigh versus MYC-low LCM foci and TCGA cases, run against gene sets in the Molecular Signature Database (v6.2).

Statistical analyses were performed using GraphPad Prism 8 for Mac. Statistical tests used and relevant variables are indicated in the legend of each figure.

14 RESULTS

Although MYC orchestrates a broad range of biological functions, it has been shown in many cancers that MYC drives tumorigenesis by potentiating or stimulating cell growth and proliferation [34] . To assess the relationship between MYC activity and proliferation across cancer types, we evaluated MYC activity and cell proliferation rate in The Cancer Genome Atlas (TCGA) pan cancer cohort using single-sample gene set enrichment analysis (ssGSEA) with MYC activity and proliferation signatures [21] . As shown in Figure 1A Figure 1 that we achieved ~50% reduction in MYC transcript and MYC protein expression. As anticipated, knockdown of MYC was associated with longer intervals between cell division in vitro (Fig. 1D ) but the decrease in cell growth was not proportional to the decrease in MYC expression. The relationship between MYC and proliferation in other cancer types also stipulates a role for MYC as a universal amplifier of transcription, [36] alleviating constraints on cell growth and proliferation [37] .

However, we observed the opposite: the total amount of cellular RNA increased when MYC was knocked down in LNCaP cells (Fig. 1D) . Together, these findings indicate that the cellular behavior of MYC in prostate cancer contrasts with other tumor types, and that MYC does not act solely in a proliferative capacity.

We hypothesized that analysis of tumors with different levels of MYC expression would identify genes that may contribute to MYC activity in PCa pathogenesis. Using samples of human radical prostatectomy specimens stained with anti-MYC, we identified concomitant regions of high and low MYC expression in the same patients. We subjected these within-patient sets of tumor foci to laser capture microdissection ( Fig. 2A) , controlling for PTEN and ERG status by confirming their concordance within each case. Of the 19 foci microdissected, we designated 42% MYC-high (n = 8) and 58% MYC-low (n = 11). We performed RNA-seq on these cases and derived a limited gene set of 303 up-or down-regulated genes, with only 17 genes showing expression changes of 4-fold or more with adjusted P values less than 0.1 (Fig. 2B) . Consistent with our observation of increased mRNA in LNCaP cells expressing MYC knockdown hairpins (see Fig. 1D ), we observed more genes downregulated than upregulated in MYC-high vs. MYC-low tumor foci.

To further refine genes of potential interest, we analyzed a second dataset comprised of whole transcriptome sequencing from 499 primary tumors in the TCGA PRAD cohort. Using the mRNA expression values of genes in the MYC activity signature (see Fig. 1A) , we compared the top 20 and bottom 20 cases based on average median absolute deviation-modified z-score ( Supplementary Fig.   2 ), and established a second list of differentially expressed genes (Fig. 2C ). As these cases were compared between patients rather than within-patient, a far greater number of genes were differentially expressed (1064) with fold-change of at least 4 and adjusted P values less than 0.05.

When comparing differentially expressed genes between MYC-high vs. MYC-low tumors, MYC expression was consistently greater in the MYC-high group in the LCM primary PCa cohort (Fig. 2C and Supplementary Table 2 Table 3 ). Without any lower-bound cut-off on expression fold-change, we identified 305 and 6,907 significant differentially expressed genes between MYC-high and MYC-low prostate tumors in the LCM tissue and TCGA PRAD cohorts, respectively (Fig. 2E ). Of these, only 220 were in common at a false discovery rate threshold of 0.1 for LCM tissue and 0.05 for TCGA PRAD (Fig. 2E , and Supplementary Table 4 ).

Given that we observed many more genes downregulated in MYC-high tumors than we would have expected if MYC were functioning as a genome-wide transcriptional amplifier, we hypothesized that collective analyses of gene expression might reveal coordinated down-regulation of biological processes that may contribute to MYC-driven PCa tumorigenesis. We therefore performed gene set enrichment analysis (GSEA) using both the LCM and TCGA cohort datasets as inputs, limiting our analyses to gene sets containing at least one differentially expressed gene shared by the TCGA and LCM cohorts. As depicted in Figure 3A , our initial comparative analyses demonstrated significant and concordant enrichment of 1737 gene sets (Supplementary Table 5 ). To narrow our focus further, we refined our search only to include gene sets that incorporated at least one target of master regulators on the premise that increased MYC activity would misregulate a multitude of pathways. These analyses identified 10 gene sets, a preponderance of which were associated with development and survival (Table 1) .

Of note, gene sets related to NOTCH, RUNX1, HOXA9, and TNF were negatively enriched in MYC-high vs. MYC-low tumors. This suggested that one or more transcription factors exert effects in opposition to MYC, and may by guiding MYC-mediated misregulation of these pathways. In particular, the transcription factor MEIS1 was identified as a common and important regulator in each of the aforementioned pathways. For example, MEIS1 has been shown to regulate genes in the NOTCH pathway [40] and sensitizes cells to TNF [41] . Moreover, MEIS1 is essential for the expression of genes driven by the HOXA9-NUP98 fusion in acute myeloid leukemia [42] [43] [44] .

Therefore, we examined whether MEIS1 expression was associated with increased MYC activity as a proxy for its role in PCa development.

We observed that MEIS1 expression was reduced in MYC-high cases compared to MYC-low tumors, for both the LCM and TCGA cohorts (Supplementary Tables 2 and 3 , respectively). At a case-by-case level, MEIS1 expression was negatively correlated with MYC (Fig. 3B ) and MYC activity (Fig. 3C) in the LCM and TCGA cohorts, respectively. To assess whether this association occurred in unselected populations of primary PCa, we analyzed RNA-seq data from an additional 69 microdissected tumor foci and the entire TCGA PRAD cohort (n = 499). Although weaker than the MYC-selected cohorts, negative associations were still observed (Fig. 3D-E) .

Recently, Bhanvadia et al., postulated that higher MEIS1 expression conferred a less aggressive PCa phenotype [19] . Based on these results, we hypothesized that repression of MEIS1 expression by MYC contributes to MYC-driven PCa. Using LNCaP cells with a non-targeting hairpin as MYChigh and LNCaP cells with three different MYC-targeting hairpins as MYC-low (see Supplementary   Fig. 1) , we performed ChIP-seq against MYC to generate genome-wide site maps and ascertain chromatin occupancy at MEIS1. In these cells, we had observed increased overall transcriptional output (see Fig. 1D ), and we controlled for this phenomenon using dm3 chromatin spike-in controls.

We observed that proportional MYC recruitment to MEIS1 was increased relative to global binding in each of the three MYC shRNA lines relative to control (Fig. 4A) . Motif enrichment analysis at MYC-ChIP peaks showed modest decreases at canonical MYC binding sites upon MYC knockdown (Fig. 4B) . We then sought to determine if the observed increase in chromatin occupancy translated to altered MEIS1 transcription by qRT-PCR. Knockdown of MYC resulted in increased abundance of the MEIS1 transcript, consistent with the observed increase in MYC recruitment at MEIS1 (Fig. 4C) .

Together, these results support the role of MYC in the negative regulation of MEIS1 in primary PCa.

In PCa, MEIS1 functions to direct transcriptional specificity and activity of HOXB13 and act as a negative regulator of AR [20, 45] . Therefore, we next determined whether HOXB13 expression or AR activity were altered in the context of MYC activity. Indeed, there was a significant positive correlation between MYC and HOXB13 mRNA levels in both the entire LCM tissue cohort (Fig. 4D , top) and entire TCGA PRAD cohort (Fig. 4E, top) . We also observed a positive correlation between MYC mRNA levels and the ssGSEA scores for AR activity [46] in both cohorts, although statistical significance (P < 0.05) was reached solely in the LCM cohort ( Fig. 4D-E; bottom) . Moreover, there was a significant negative correlation between MEIS1 and HOXB13 expression in both cohorts ( Fig.   4F -G, top) which was also observed between MEIS1 expression and AR activity (Fig. 4H-I) . Taken together, our data suggest that in MYC-high tumors, PCa development is mediated by increased AR activity and HOXB13 expression resulting from MEIS1 down-regulation. DISCUSSION In many cancer types, the role of amplified MYC in mediating tumorigenesis has been linked to genes involved in ribosomal biogenesis, universally upregulated transcription, proliferation, and reprogramming cells to a pluripotent state [47] . A subset of advanced prostate cancers also harbor amplified MYC, but it is distinct from the upregulated MYC that is a hallmark of many localized prostate cancers [3, [14] [15] [16] . In the current study, we used transcriptome profiling to assess subpopulations of prostate tumors based on differential MYC protein expression and MYC activity, and we similarly compared differentially expressed genes and pathways within the larger prostate TCGA cohort based on MYC activity. Finding that increased MYC activity was inversely proportional to overall transcription, we focused on downregulated pathways, identifying a negative correlation between MYC activity and MEIS1 expression, with MYC directly involved in the negative regulation of MEIS1 as demonstrated by knockdown and chromatin immunoprecipitation analyses. The inverse association between MEIS1 expression and AR extended further to HOXB13 expression, indicating that in a subset of primary PCa, decreased expression of MEIS1 may be necessary for AR and HOXB13 to drive tumor development and progression.

From its discovery as an oncogene to the subsequent challenges associated with targeting MYC pharmacologically, efforts have shifted in identifying targetable MYC effector genes or other targetable co-factors that are necessary for MYC activity [48, 49] . Efforts to dissect functions of MYC have frequently relied on cancer models in which MYC levels rise up to 20-fold, in contrast to 1-2 fold physiological elevation of MYC expression in prostate cancer [2, 6, 50] . Not surprisingly, the phenotypes associated with MYC up-regulation differ. For example, MYC expression in the activation of lymphocytes [36] or in Burkitt's lymphoma [37] is associated with a universal amplification of transcription, while we observed that in LNCaP prostate cancer cells with upregulated MYC, knocking down MYC by less than 50% with shRNA consistently increased the total amount of RNA produced per cell. 20 The differences in MYC function in prostate cancer extend to the long-standing relationship between MYC and proliferation [47] . Here, we report a series of prostate cancer tissues, seriallysectioned and stained with anti-MYC and anti-Ki67 antibodies, that show a weak proportional relationship. While up to 50% of luminal prostate cancer cells were positive for MYC expression, proliferation measured by Ki67 affected less than 5% of cells. We show a similar relationship in the prostate TCGA cohort and prostate cancer cell lines, suggesting that while proliferating cells may harbor MYC activity, MYC alone is not sufficient for proliferation in prostate cancer. This is generally in agreement with findings that MYC expression alone is not reflective of an increased proliferative fraction of prostate cancer cells [5] .

A key finding from our study is the negative relationship between MYC activity and MEIS1 expression. In the context of MYC's role as a universal transcriptional amplifier, a target gene of MYC-mediated repression could simply considered a technical anomaly reflecting unequal numbers of cells used for analysis [36] . However, we show that transcriptional signatures of MYC activity and MEIS1 are inversely correlated in two large independent cohorts, which is largely consistent with the finding that tumors and prostate cancer cells with increased MEIS1 expression show decreased enrichment of MYC target gene sets by GSEA [19] . Importantly, we further demonstrate here with ChIP-seq that the effects on MEIS1 expression are due in part to increased MYC occupancy at the MEIS1 locus, such that lower levels of MYC resulted in repositioning of MYC at specific sites. This finding is in disagreement with the model that increased MYC expression may have strengthened its activity at a MYC-recruited repressor [36] .

Previously, Bhanvadia et al. [19] reported that tumors with increased MEIS1 are potentially less aggressive, based on studies of LAPC-4 prostate cancer cells expressing shRNA against MEIS1 and that tumors with lower levels of MEIS1 were at greater risk of biochemical recurrence. HOXB13, a homeodomain transcription factor, has been shown to regulate AR activity while shRNA against HOXB13 in LAPC4 cells inhibits their growth [18] . Based on these prior findings, as HOXB13 physically interacts with MEIS1 [18] , tumors expressing less MEIS1 would be expected to display greater HOXB13 expression and AR activity. Indeed, we show an inverse relationship between MEIS1 expression and AR activity in two independent cohorts, which is consistent with previous observations of a positive correlation between MYC expression and AR activity [7] .

In summary, our analysis of MYC-expressing prostate tumors demonstrates an inverse relationship with MEIS1 expression, which in turn is negatively correlated with HOXB13 expression and AR activity. Mechanistically, our data demonstrate that MEIS1 is a directly repressed target of MYC, and via effects on HOXB13 link MYC activity to AR activity. The potential clinical significance of the inverse MYC/MEIS1 relationship warrants further investigation as AR-directed therapies are introduced earlier in the clinical course of disease, and MEIS1 levels may indicate potential sensitivity to treatment. TABLES 

The Molecular Taxonomy of Primary Prostate Cancer. Cell

Integrative genomic profiling of human prostate cancer

Gleason Score 7 Prostate Cancers Emerge through Branched Evolution of Clonal Gleason Pattern 3 and 4. Clin Cancer Res

PTEN loss and chromosome 8 alterations in Gleason grade 3 prostate cancer cores predicts the presence of un-sampled grade 4 tumor: implications for active surveillance

Nuclear MYC protein overexpression is an early alteration in human prostate carcinogenesis

MYC and Prostate Cancer

A positive role of c-Myc in regulating androgen receptor and its splice variants in prostate cancer

The molecular and cellular origin of human prostate cancer

The homeodomain protein HOXB13 regulates the cellular response to androgens

Current perspectives on FOXA1 regulation of androgen receptor signaling and prostate cancer

c-Myc Antagonises the Transcriptional Activity of the Androgen Receptor in Prostate Cancer Affecting Key Gene Networks

Myc-dependent purine biosynthesis affects nucleolar stress and therapy response in prostate cancer

Alterations in nucleolar structure and gene expression programs in prostatic neoplasia are driven by the MYC oncogene

MYC drives overexpression of telomerase RNA (hTR/TERC) in prostate cancer

Neoadjuvant-Intensive Androgen Deprivation Therapy Selects for Prostate Tumor Foci with Diverse Subclonal Oncogenic Alterations

Prostate epithelial cell of origin determines cancer differentiation state in an organoid transformation assay

HOXB13 interaction with MEIS1 modifies proliferation and gene expression in prostate cancer

MEIS1 and MEIS2 Expression and Prostate Cancer Progression: A Role For HOXB13 Binding Partners in Metastatic Disease

HOXB13 mutations and binding partners in prostate development and cancer: Function, clinical significance, and future directions. Genes & diseases

A library of gene expression signatures to illuminate normal and pathological lymphoid biology

The 2014 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma

Clonal Progression of Prostate Cancers from Gleason Grade 3 to Grade 4

Analysis of Relative Gene Expression Data Using Real-Time Quantitative PCR and the 2−ΔΔCT Method

Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM

Model-based Analysis of ChIP-Seq (MACS)

DiffBind: differential binding analysis of ChIP-seq peak data

Simple combinations of lineage-determining transcription factors prime cisregulatory elements required for macrophage and B cell identities

Quantitative ChIP-Seq normalization reveals global modulation of the epigenome

STAR: ultrafast universal RNA-seq aligner

featureCounts: an efficient general purpose program for assigning sequence reads to genomic features

Bioconductor package for differential expression analysis of digital gene expression data

clusterProfiler: an R Package for Comparing Biological Themes Among Gene Clusters

Targeting oncogenic Myc as a strategy for cancer treatment

In vitro and in vivo model systems used in prostate cancer research

c-Myc is a universal amplifier of expressed genes in lymphocytes and embryonic stem cells

Transcriptional amplification in tumor cells with elevated c-Myc

Mechanisms of MYC stabilization in human malignancies

Overexpression of C-MYC oncogene in prostate cancer predicts biochemical recurrence. Prostate Cancer And Prostatic Diseases

Meis1 coordinates a network of genes implicated in eye development and microphthalmia

Meis1 programs transcription of FLT3 and cancer stem cell character, using a mechanism that requires interaction with Pbx and a novel function of the Meis1 C-terminus

Nup98-HoxA9 immortalizes myeloid progenitors, enforces expression of Hoxa9, Hoxa7 and Meis1, and alters cytokine-specific responses in a manner similar to that induced by retroviral co-expression of Hoxa9 and Meis1

NUP98-HOXA9 induces long-term proliferation and blocks differentiation of primary human CD34+ hematopoietic cells

Deregulation of the HOXA9/MEIS1 axis in acute leukemia. Current opinion in hematology

MEIS1 functions as a potential AR negative regulator

Genomic Strategy for Targeting Therapy in Castration-Resistant Prostate Cancer

MYC on the path to cancer

c-Myc target genes involved in cell growth, apoptosis, and metabolism

BET bromodomain inhibition as a therapeutic strategy to target c-Myc. Cell

Analysis of gene expression during myc oncogene-induced lymphomagenesis in the bursa of Fabricius