JAZF1: A Metabolic Regulator of Sensitivity to a Polyamine-Targeted Therapy

Abstract

Identifying and leveraging unique points of metabolic dysregulation in different disease settings is vital for safe and effective incorporation of metabolism-targeted therapies in the clinic. In addition, it has been shown identification of master metabolic transcriptional regulators (MMTR) of individual metabolic pathways, and how they relate to the disease in question, may offer the key to understanding therapeutic response. In prostate cancer, we have previously demonstrated polyamine biosynthesis and the methionine cycle were targetable metabolic vulnerabilities. However, the MMTRs of these pathways, and how they affect treatment, have yet to be explored. We sought to characterize differential sensitivity of prostate cancer to polyamine- and methionine-targeted therapies by identifying novel MMTRs. We began by developing a gene signature from patient samples, which can predict response to metabolic therapy, and further uncovered a MMTR, JAZF1. We characterized the effects of JAZF1 overexpression on prostate cancer cells, basally and in the context of treatment, by assessing mRNA levels, proliferation, colony formation capability, and key metabolic processes. Lastly, we confirmed the relevance of our findings in large publicly available cohorts of prostate cancer patient samples. We demonstrated differential sensitivity to polyamine and methionine therapies and identified JAZF1 as a MMTR of this response.

Implications: We have shown JAZF1 can alter sensitivity of cells and its expression can segregate patient populations into those that do, or do not highly express polyamine genes, leading to better prediction of response to a polyamine targeting therapy.

Figure 1.

Variable sensitivity to polyamine therapy in prostate cancer. A, Baseline patient tumors were sectioned and either flash frozen or treated on dental sponges for 7 days with vehicle control or two doses combination therapy (1 μmol/L BENSpm + 1 nmol/L MTDIA, or 10 μmol/L BENSpm + 10 nmol/L MTDIA). Baseline tumor (never treated) were used for RNA-seq to determine signatures of responders and nonresponders. B, Dose-dependent response to polyamine-targeted therapies as determined by measurement of polyamine levels revealed 6 responders (green) and 3 nonresponders (gray). C, IHC stained slides for SSAT indicated a change in index scores between vehicle control and 10 μmol/L BENSpm + 10 nmol/L MTDIA for responders and nonresponders. D, Unsupervised hierarchical clustering of samples based on DEGs show clear delineation between responders (green) and nonresponders (gray). E, Unsupervised hierarchical clustering of transcriptomic data from TCGA PRAD patients stratified into Q1 (nonresponders, gray) and Q4 (responders, green) based upon the 82-gene ex vivo signature scores across 1,843 metabolic genes. F, Comparison of TCGA predicted nonresponders to responders revealed nonresponders have lower levels of polyamine genes, compared with the responders (size of the triangle indicates FC, red = significantly upregulated, blue = significantly downregulated, gray = unchanged transcript). G, Master regulator enrichment of the ex vivo signature reveal potential transcriptional regulators of resistance to polyamine-targeted therapies.

Figure 2.

MMTRs determine adaptation to polyamine-targeted therapies. A, CWR22-RV1 and LNCaP C4–2 cell lines were cultured with MTA (vehicle), single or combination therapy for varying time points. RNA was then collected and used for RNA-seq and then used for further MMTR analysis and gene set enrichment analysis. B, Volcano plots of RNA-seq data from LNCAP C4–2 cells comparing control-treated and combination-treated (BENSpm + MTDIA) at 1, 24, and 96 hours. (Significant genes are red). C, Example of MMTR analysis of the DEGs from the C4–2 96-hour combination versus vehicle treatment revealed master regulators for the time point. This was then overlapped with the master regulators of the other cell line, time points, and treatments to generate a (D) word cloud (right), where the size of the transcriptional regulator is indicative of how frequently it occurred, which is quantitated for the top 9 MMTRs (left). E, Further MMTR analysis of the polyamine and methionine cycles, the current targets of our drugs, similarly enriched for JAZF1 as a master regulator. *All experiments were a biological n = 3 and represent mean ± SD Significance calculated by Student t test, unless otherwise indicated.

Figure 3.

JAZF1 overexpression confers resistance to polyamine-targeted therapies. A, JAZF1 overexpression leads to increased growth over 8 days for androgen sensitive (LNCaP) and insensitive (C4–2 and RV1) cell lines without (left) and with (right) combination therapy. B, Overexpression of JAZF1 resulted in alterations of genes associated with response to treatment such as decreases n SSAT (target of BENSpm) and SMOX and PAOX (ROS producers) in both androgen-independent (blue and purple, top) and dependent (green, bottom) cell lines. C, Basal levels of ROS and ROS induction with treatment are decreased with JAZF1 overexpression as compared with parental controls in both LNCaP C4–2 (blue, top) and LNCaP (green, bottom) cell lines (D) Both basal and combination treated SSAT activity is reduced across all cell lines with JAZF1 overexpression as compared with parental controls. E, UPLC analysis of intracellular spermine and spermidine pools after 8 days of treatment indicates that the BENSPm and combination therapy induced distortion of the spermidine/spermine ratio abrogated with JAZF1 overexpression. F, Representative images of colony formation with combination therapy in the parental cell lines (−JAZF1) and JAZF1 overexpression (+JAZF1) cell lines, quantitated by CV. G, CV quantification at day 25 of the colony formation assay comparing the ratio of absorbance present in the combination/vehicle treated groups for each of the parental and JAZF1 overexpressing cell lines, multiple t tests were performed to assess significance in the ratio between parental controls and the JAZF1 overexpressing cell lines. *All experiments were a biological n = 3 and represent mean ± SD Significance calculated by Student t test, unless otherwise indicated.

Figure 4.

JAZF1 overexpression reveals altered metabolism and predicts differential polyamine metabolism in TCGA Data. A, CWR22-RV1 and LNCaP-C42 parental cell lines and those overexpressing JAZF1 were treated with vehicle control or combination therapy for 1 hour and 96 hours to assess the downstream transcriptional changes cells occur when JAZF1 levels are high prior to and during polyamine-targeted therapy. B, Unsupervised hierarchical clustering on 1,843 metabolic genes demonstrating distinction between the RV1 parental (purple) and JAZF1 (orange) overexpressing cell lines treated with vehicle (yellow) or combination (green) treatment or time points. C, Metabolic pipeline analysis revealed 16 metabolic pathways commonly altered (yellow) as a result of JAZF1 overexpression, and 9 metabolic pathways (green) that remained altered with combination therapy, all of which were a subset of the 16 originally altered pathways. D,JAZF1 high (purple) and low (green) patients are well segregated on the basis of the global expression of 1,843 metabolic genes. Quartiles also tend to closely follow high versus low designation (E) JAZF1 high patients, overall, have lower expression of transcripts within the polyamine biosynthetic pathway, indicating they may not be primed for response to polyamine-targeted therapies. F, Correlation matrix between JAZF1 expression levels in TCGA PRAD and polyamine metabolism genes. Node size, −log P value; red, positive correlations; blue, negative correlations.

Figure 5.

Increased JAZF1 corresponds with decreased predicted sensitivity to combination therapy. A, Unsupervised hierarchical clustering of transcriptomic data from TCGA PRAD patients stratified into JAZF1 expression high, top 25% (purple) and low, lower 75% (green) and ex vivo score predicted resistant, top 50% (gray) and predicted sensitive, lower 50% (green) based upon the 82-gene ex vivo signature genes. B, Venn diagram demonstrates a significant overlap of the patients with increased levels of JAZF1 (purple) and predicted resistance to polyamine therapies by ex vivo score (gray). Significance determined by hypergeometric test. C,JAZF1 and ex vivo prediction scores were positively correlated in the metastatic samples of the Taylor and colleagues dataset.

Figure 6.

Proposed effects of JAZF1 on polyamine biosynthesis and combination therapy. JAZF1 decreases the levels of SSAT, and induction of SSAT enzymatic activity induced by BENSpm. It also decreases the transcriptional levels of PAOX and SMOX, the major ROS producers, thereby decreasing both basal ROS accumulation and induction by therapy. This leads to drug resistance and cellular proliferation with therapeutic intervention.