Wei, Lixuan

Abstract Prediction of the response of cancer patients to different treatments and identification of biomarkers of drug response are two major goals of individualized medicine. Here, we developed a deep learning framework called TINDL, completely trained on preclinical cancer cell lines (CCLs), to predict the response of cancer patients to different treatments. TINDL utilizes a tissue-informed normalization to account for the tissue type and cancer type of the tumors and to reduce the statistical discrepancies between CCLs and patient tumors. Moreover, by making the deep learning black box interpretable, this model identifies a small set of genes whose expression levels are predictive of drug response in the trained model, enabling identification of biomarkers of drug response. Using data from two large databases of CCLs and cancer tumors, we showed that this model can distinguish between sensitive and resistant tumors for 10 (out of 14) drugs, outperforming various other machine learning models. In addition, our small interfering RNA (siRNA) knockdown experiments on 10 genes identified by this model for one of the drugs (tamoxifen) confirmed that tamoxifen sensitivity is substantially influenced by all of these genes in MCF7 cells, and seven of these genes in T47D cells. Furthermore, genes implicated for multiple drugs pointed to shared mechanism of action among drugs and suggested several important signaling pathways. In summary, this study provides a powerful deep learning framework for prediction of drug response and identification of biomarkers of drug response in cancer. The code can be accessed at https://github.com/ddhostallero/tindl.

Abstract Androgen receptor (AR) is expressed in 80% to 90% of estrogen receptor α–positive (ER+) breast cancers. Accumulated evidence has shown that AR is a tumor suppressor and that its expression is associated with improved prognosis in ER+ breast cancer. However, both a selective AR agonist (RAD140) and an AR inhibitor (enzalutamide, ENZ) have shown a therapeutic effect on ER+ breast cancer, so the potential for clinical application of AR-targeting therapy for ER+ breast cancer is still in dispute. In this study, we evaluated the efficacy of ENZ and RAD140 in vivo and in vitro in AR+/ER+ breast cancer models, characterizing the relationship of AR and ER levels to response to AR-targeting drugs and investigating the alterations of global gene expression and chromatin binding of AR and ERα after ENZ treatment. In the AR-low setting, ENZ directly functioned as an ERα antagonist. Cell growth inhibition by ENZ in breast cancer with low AR expression was independent of AR and instead dependent on ER. In AR-high breast cancer models, AR repressed ERα signaling and ENZ promoted ERα signaling by antagonizing AR. In contrast, RAD140 activated AR signaling and suppressed AR-high tumor growth by deregulating ERα expression and blocking ERα function. Overall, analysis of the dynamic efficacies and outcomes of AR agonist, and antagonist in the presence of different AR and ERα levels reveals regulators of response and supports the clinical investigation of ENZ in selected ER+ tumors with a low AR/ER ratio and AR agonists in tumors with a high AR/ER ratio. Significance: The ratio of androgen receptor to estrogen receptor in breast cancer dictates the response to AR-targeted therapies, providing guidelines for developing AR-directed treatment strategies for patients with breast cancer.

Abstract Understanding the function of non-coding genomic sequence variants represents a challenge for biomedicine. Many diseases are products of gene-by-environment interactions with complex mechanisms. This study addresses these themes by mechanistic characterization of non-coding variants that influence gene expression only after drug or hormone exposure. Using glucocorticoid signaling as a model system, we integrated genomic, transcriptomic, and epigenomic approaches to unravel mechanisms by which variant function could be revealed by hormones or drugs. Specifically, we identified cis-regulatory elements and 3D interactions underlying ligand-dependent associations between variants and gene expression. One-quarter of the glucocorticoid-modulated variants that we identified had already been associated with clinical phenotypes. However, their affected genes were ‘unmasked’ only after glucocorticoid exposure and often with function relevant to the disease phenotypes. These diseases involved glucocorticoids as risk factors or therapeutic agents and included autoimmunity, metabolic and mood disorders, osteoporosis and cancer. For example, we identified a novel breast cancer risk gene, MAST4, with expression that was repressed by glucocorticoids in cells carrying the risk genotype, repression that correlated with MAST4 expression in breast cancer and treatment outcomes. These observations provide a mechanistic framework for understanding non-coding genetic variant-chemical environment interactions and their role in disease risk and drug response.

AbstractAlternative polyadenylation (APA) is a common genetic regulatory mechanism that generates distinct 3′ ends for RNA transcripts. Changes in APA have been associated with multiple biological processes and disease phenotypes. However, the role of hormones and their drug analogs in APA remains largely unknown. In this study, we investigated transcriptome‐wide the impact of glucocorticoids on APA in 30 human B‐lymphoblastoid cell lines. We found that glucocorticoids could regulate APA for a subset of genes, possibly by changing the expression of 142 RNA‐binding proteins, some with known APA‐regulating properties. Interestingly, genes with glucocorticoid‐mediated APA were enriched in viral translation‐related pathways, while genes with glucocorticoid‐mediated expression were enriched in interferon and interleukin pathways, suggesting that glucocorticoid‐mediated APA might result in functional consequences distinct from gene expression. For example, glucocorticoids, a pharmacotherapy for severe COVID‐19, were found to change the APA but not the expression of LY6E, an important antiviral inhibitor in coronavirus diseases. Glucocorticoid‐mediated APA was also cell‐type‐specific, suggesting an action of glucocorticoids that may be unique to immune regulation. We also observed evidence for genotype‐dependent glucocorticoid‐mediated APA (referred to as pharmacogenomic‐alterative polyadenylation quantitative trait loci), providing potential functional mechanisms for a series of common genetic variants that had previously been associated with immune disorders, but without a clear mechanism. In summary, this study reports a series of observations regarding the impact of glucocorticoids on APA, raising the possibility that this mechanism might have implications for both disease pathophysiology and drug therapy.

Abstract The human angiotensin-converting enzyme 2 (ACE2) and transmembrane serine protease 2 (TMPRSS2) proteins play key roles in the cellular internalization of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the coronavirus responsible for the coronavirus disease of 2019 (COVID-19) pandemic. We set out to functionally characterize the ACE2 and TMPRSS2 protein abundance for variant alleles encoding these proteins that contained non-synonymous single-nucleotide polymorphisms (nsSNPs) in their open reading frames (ORFs). Specifically, a high-throughput assay, deep mutational scanning (DMS), was employed to test the functional implications of nsSNPs, which are variants of uncertain significance in these two genes. Specifically, we used a ‘landing pad’ system designed to quantify the protein expression for 433 nsSNPs that have been observed in the ACE2 and TMPRSS2 ORFs and found that 8 of 127 ACE2, 19 of 157 TMPRSS2 isoform 1 and 13 of 149 TMPRSS2 isoform 2 variant proteins displayed less than ~25% of the wild-type protein expression, whereas 4 ACE2 variants displayed 25% or greater increases in protein expression. As a result, we concluded that nsSNPs in genes encoding ACE2 and TMPRSS2 might potentially influence SARS-CoV-2 infectivity. These results can now be applied to DNA sequence data for patients infected with SARS-CoV-2 to determine the possible impact of patient-based DNA sequence variation on the clinical course of SARS-CoV-2 infection.

Approximately one‐third of patients with metastatic castration‐resistant prostate cancer (CRPC) exhibited primary abiraterone resistance. To identify alternative treatment for abiraterone nonresponders, we performed drug discovery analyses using the L1000 database using differentially expressed genes identified in tumor biopsies and patient‐derived xenograft (PDX) tumors between abiraterone responders and nonresponders enrolled in PROMOTE trial. This approach identified 3 drugs, including topoisomerase II (TOP2) inhibitor mitoxantrone, CDK4/6 inhibitor palbociclib, and pan‐CDK inhibitor PHA‐793887. These drugs significantly suppressed the growth of abiraterone‐resistant cell lines and PDX models. Moreover, we identified 11 genes targeted by all 3 drugs that were associated with worse outcomes in both the PROMOTE and Stand Up To Cancer cohorts. This 11‐gene panel might also function as biomarkers to select the 3 alternative therapies for this subgroup of patients with CRPC, warranting further clinical investigation.

Abstract Background: Androgen receptor (AR) is widely expressed in ER+ breast cancers. Accumulated evidence showed that AR is a tumor suppressor in ERα+ breast cancers. Enzalutamide (ENZ), an AR antagonist, blocks AR signalling, and was reported to inhibit estrogen-driven tumor growth as effectively as tamoxifen. Hence, this study was performed to further examine the mechanism of action of ENZ in ERα+ breast cancers. Methods: The AR gene was knocked out using the CRISPR-Cas9 system. Four crRNAs targeting AR and trans-activating crRNA were co-transfected to Cas9-positive MCF7 and T47D cells, two ERα+ breast cancer cell lines with relatively low AR levels based on RNA seq. ESR1 knock down was performed using a lentivirus system. MCF7 and T47D cells were infected with the ESR1-shRNA lentivirus and selected by puromycin for 2 weeks. Single clones of knockdown or knockout cells were tested using western blot. MCF7 and T47D cells were collected for RNA-seq and ChIP-seq after treated with DMSO or 10µM ENZ for 6 hours in complete medium. All sequences were aligned to the Human Reference Genome (hg38, UCSC). Differential gene expression was performed using limma package and upstream analysis was done in Ingenuity Pathway Analysis software. ChIP peaks were called using MACS2 and overlapping was analyzed using ChIPpeakAnno package in R. Tritium labeled ENZ or estradiol were incubated with ERα overnight. The bound radioactivity was measured using a Beckmann LS 6500 liquidscintillation counter. Results: We found the cell growth inhibition effect of ENZ was little impaired by AR gene knock out in MCF7 and T47D cells as was expected. As ENZ has been shown to inhibit estrogen-driven tumor growth, we performed studies to understand the mechanism of ENZ.We performed RNA-seq and identified 193 differential expressed genes (76 up and 117 down) between DMSO and ENZ-treated MCF7 cells (FC>2 or <0.5, FDR<0.05). Upstream analysis showed that ESR1 might be the upstream regulator of those differentially expressed genes and ESR1 signaling was predicted to be inhibited after ENZ treatment (P=1.69×10-10). To examine the alteration of ERα genomic binding, we performed global ERα ChIP-seq and found 39.58% (8,681/21,933) of ERα-bound sites were abolished by ENZ and the binding intensity was significantly decreased upon ENZ treatment in 12,868 shared peaks. Moreover, radiolabled binding assays showed that ENZ displaced estrogen binding to ERα and directly bound to ERα with a KD= 399.60nM. We further observed a reduction of proliferation inhibition of ENZ in ESR1-KD cells compared with control cells. This suggests that ENZ inhibits ER+ cell growth by displacing E2 from ERα binding sites and subsequently diminishes ERα genomic binding and blocks ERα signaling in ER+ cancers. Conclusion: We found that enzalutamide inhibits cell growth not through AR but by antagonizing ERα activity in MCF7 and T47D ER+ cells. These results reveal ER is the main factor affecting enzalutamide efficacy in these two ERα+ breast cancer cell lines. Further clinical trials are needed to investigate AR antagonists in ERα+ breast cancers. Citation Format: Lixuan Wei, Jia Yu, Huanyao Gao, Huan Zhang, Thanh Nguyen, Yayun Gu, Richard M. Weinshilboum, James N. Ingle, Liewei Wang. Estrogen receptor is a target of enzalutamide in ER+ breast cancer [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr PD1-03.

Background and Purpose Acamprosate is an anti‐craving drug used for the pharmacotherapy of alcohol use disorder (AUD). However, only some patients achieve optimal therapeutic outcomes. This study was designed to explore differences in metabolomic profiles between patients who maintained sobriety and those who relapsed, to determine whether those differences provide insight into variation in acamprosate treatment response phenotypes. Experimental Approach We previously conducted an acamprosate trial involving 442 AUD patients, and 267 of these subjects presented themselves for a 3‐month follow‐up. The primary outcome was abstinence. Clinical information, genomic data and metabolomics data were collected. Baseline plasma samples were assayed using targeted metabolomics. Key Results Baseline plasma arginine, threonine, α‐aminoadipic acid and ethanolamine concentrations were associated with acamprosate treatment outcomes and baseline craving intensity, a measure that has been associated with acamprosate treatment response. We next applied a pharmacometabolomics‐informed genome‐wide association study (GWAS) strategy to identify genetic variants that might contribute to variations in plasma metabolomic profiles that were associated with craving and/or acamprosate treatment outcome. Gene expression data for induced pluripotent stem cell‐derived forebrain astrocytes showed that a series of genes identified during the metabolomics‐informed GWAS were ethanol responsive. Furthermore, a large number of those genes could be regulated by acamprosate. Finally, we identified a series of single nucleotide polymorphisms that were associated with acamprosate treatment outcomes. Conclusion and Implications These results serve as an important step towards advancing our understanding of disease pathophysiology and drug action responsible for variation in acamprosate response and alcohol craving in AUD patients.

Background: Trastuzumab (Herceptin) is the key systemic therapy for HER2-positive breast cancer. However, the initial response rate is limited to approximately 50% in patients. Moreover, most patients, especially at an advanced stage, eventually develop acquired resistance. Understanding the mechanisms of trastuzumab resistance is crucial for achieving better treatment outcome in this group of patients. Methods: A trastuzumab-resistant (TR) cell line was developed using the BT474 HER2-positive breast cancer cell line. Whole-transcriptome expression array was performed and the TR-related gene NDUFA4L2 was identified by differential expression analysis between BT474 and BT474-TR. Mitochondrial localization of NDUFA4L2 was confirmed by immunofluorescence and western blotting using mitochondrial fractionation. Mitochondrial function and energy metabolism were evaluated using Seahorse, ATP production, and lactate production assays, and cellular reactive oxygen species (ROS) levels were determined using DCFDA. NDUFA4L2 expression in patients was evaluated by immunohistochemistry, and relapse-free survival was analyzed using the Kaplan–Meier method. Results: NDUFA4L2 was highly expressed in the TR HER2-positive breast cancer cell line. High expression level of NDUFA4L2 was associated with shorter relapse-free intervals in trastuzumab-treated HER2-positive breast cancer patients. Overexpression of NDUFA4L2 enhanced Warburg effects, enhanced aerobic glycolysis, reduced oxygen consumption, and lowered ROS production. Mechanistically, overexpression of NDUFA4L2 facilitated mitochondrial relocalization of HER2 and suppressed ROS production, thus rendering cancer cells more resistant to trastuzumab treatment. Conclusions: We identified NDUFA4L2 as a new biomarker and potential therapeutic target for TR HER2-positive breast cancer.

Abstract Background: Androgen receptor (AR) is expressed in 60%-90% of breast cancers. The role of AR in breast cancer largely depends on estrogen receptor (ER) status and remains controversial. In ER+ cancers, AR expression is associated with improved prognosis. Enzalutamide (ENZ), an AR antagonist, impairs AR signaling, inhibits ER+/AR+ breast cancer cell proliferation and has been shown to mitigate resistance to anti-estrogen therapies. A recent study identified that RAD140, an AR agonist, suppressed the growth of ER+/AR+ breast cancer via stimulating AR signaling, resulting in down regulation of ERα. Both AR agonists and antagonists are effective in treating ER+ breast cancers. Identifying the subgroup of patients who most likely will benefit from an individual drug is important in clinical practice. Here, we sought to characterize the relative AR to ER levels and their relationships to drug response in ER+ breast cancers.Methods: We evaluated AR and ER expression levels among 68 samples with ER+ breast cancers. On each tumor, IHC staining for AR and ER were performed and results were scored by multiplying the percentage of positive cells by the intensity. TCGA-RPPA (reverse phase protein array) dataset was used to verify AR protein distribution in ER+ cancers. We created MCF7 and T47D cells with doxycycline-inducible AR and ER expression system to manipulate the relative AR to ER ratios and determined the antitumor activity of ENZ and RAD140 by cell proliferation assays. The levels of AR and ER in cells were measured by western blot. Linear regression was used to test the association between dose-response area under curve (AUC) and AR, ER levels. Drug preference was modeled against AR/ER expression levels (AR/ER ratios) using logistic regression.Results: Among 68 ER+ breast cancers, 69.12% were AR-positive. More than 2/3 cases (48 of 68) had more ER than AR expressed. The AR to ER ratios varied from 0 to 6. In the TCGA cohort, 84.15% of 347 ER+ patients had AR/ER ratios less than 1.The range of AR to ER ratios in TCGA dataset were comparable with our data (0.004 to 4.210). In our cell line models, the AR/ER ratios were controlled between 0.19 to 4.05. We found that the AUC of RAD140 was negatively associated with AR protein levels (P=0.008) and AR/ER ratios (P=0.013), and not significantly associated with ER expression levels. On the other hand, the AUC of ENZ was significantly negatively associated with ER protein (P=0.0016) but postiviely associated with AR/ER ratio (p=0.037). Preferred treatment comparing efficacy of the two drug can be best determined at extremes of AR/ER ratios. Using clinically relevant dosages, our model predicted that ENZ (10µM) would be a preferred treatment choice and have a better treatment efficacy compared with RAD140 when the AR/ER ratio was ≤ 0.42, whereas RAD140 (1µM) would be the preferred choice with a better treatment efficacy compared with ENZ when AR/ER ratios were ≥ 3.1 The efficacy preference of the two drugs are equivocal for AR/ER ratios between 0.42 and 3.10. Conclusion: We developed preclinical models using AR and ER expression levels to predict AR-targeting drug response. The results support the use of RAD140 in AR high patients and those with an AR/ER ratio >3.10, and enzalutamide in AR low patients and those with an AR/ER ratio <0.42. Equipoise on choice of drug was found for AR/ER ratios of 0.42-3.10. RAD140 and enzalutamide are compelling candidates for monotherapy or combination with anti-estrogen therapies in ER+/AR+ breast cancer. Future clinical validation of the models and therapeutic effect is warranted. Citation Format: Lixuan Wei, Huanyao Gao, Jia Yu, Duan Liu, Huan Zhang, Thanh Nguyen, Marie R Passow, Jodi M Carter, Richard M Weinshilboum, James N Ingle, Liewei Wang. Computational modeling of androgen receptor (AR) and estrogen receptor as predictive biomarkers of response to AR agonists and antagonists [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PS5-36.