Volume 30, Issue 3, Pages 459-473 (September 2016) Feedback Activation of Leukemia Inhibitory Factor Receptor Limits Response to Histone Deacetylase Inhibitors in Breast Cancer  Hanlin Zeng, Jia Qu, Nan Jin, Jun Xu, Chenchu Lin, Yi Chen, Xinying Yang, Xiang He, Shuai Tang, Xiaojing Lan, Xiaotong Yang, Ziqi Chen, Min Huang, Jian Ding, Meiyu Geng  Cancer Cell  Volume 30, Issue 3, Pages 459-473 (September 2016) DOI: 10.1016/j.ccell.2016.08.001 Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 1 HDAC Inhibition Induces Cytokine-Cytokine Receptor Pathway Reprogramming and STAT3 Activation (A) Cell sensitivity to SAHA treatment. Cells were treated with SAHA at gradient concentrations for 72 hr and IC50s were measured using CCK8 assay. (B) A scatterplot of IC50s in solid tumor and hematologic tumor cell lines to SAHA treatment. Horizontal lines represent average of IC50s shown in (A). ∗∗∗p < 0.001. (C) Heatmaps of gene-expression data from microarray analysis. MDA-MB-231 cells were treated with 5 μM SAHA in triplicate for 8 hr and the global gene-expression change was measured by microarray. Gene-expression alteration in the treated group was normalized by corresponding control treated with vehicle DMSO. Significant genes were determined by Student's t test and a threshold cutoff of p < 0.01, 3-fold change. Red, induced; green, repressed; log2-based scale. (D) KEGG pathway analysis of genes affected by SAHA as described in (C). (E) Heatmaps showing affected genes involved in cytokine-cytokine receptor interaction described in (C). (F) Phosphokinase antibody arrays. MDA-MB-231 cells were treated with 5 μM SAHA for 8 hr and the cell lysates were subjected to phosphokinase antibody array. (G) Immunoblotting of STAT3 phosphorylation. MDA-MB-231 cells were treated with indicated compounds for 8 hr. (H) STAT3 transcriptional activity. MDA-MB-231 cells were transfected with STAT3 luciferase reporter construct for 48 hr followed by SAHA treatment for 8 or 12 hr. The luciferase activity was measured. ∗∗p < 0.01. Error bars represent means ± SD from triplicates or three independent experiments. See also Figure S1; Tables S1 and S2. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 2 Feedback Activation of STAT3 Limits Responses to HDAC Inhibition in Breast Cancer Cells (A) Cell growth assay. MDA-MB-231 cells were transfected with non-targeting control (NC) or STAT3 siRNAs for 48 hr followed by SAHA treatment for a further 72 hr. Cell growth change was measured by counting cell numbers. (B) Cell apoptosis analysis. MDA-MB-231 cells transfected with indicated siRNAs for 48 hr were treated with SAHA for a further 24 hr. Apoptotic cells were detected by annexin V/PI dual staining. Knockdown efficiency was measured by immunoblotting. Acetylated tubulin (Ac-Tubulin) was blotted as an internal control for HDAC inhibition. (C and D) Anti-apoptotic signaling alteration. MDA-MB-231 cells were treated with SAHA for indicated time (C) or were transfected with indicated siRNAs for 48 hr followed by SAHA treatment for a further 8 hr (D). Anti-apoptotic signaling alteration was detected by immunoblotting analysis. (E) Basal level of phospho-STAT3 or STAT3 transcriptional activity versus SAHA sensitivity in breast cancer cells. Upper panel: waterfall plot of IC50s of SAHA in 21 breast cancer cell lines. Cells were treated with SAHA at gradient concentrations for 72 hr and IC50s measured using CCK8 assay. Lower panel: basal level of phospho-STAT3 and transcriptional activity in seven most responsive and seven most non-responsive cells. STAT3 phosphorylation was detected by immunoblotting, semi-quantified by densitometry, and normalized by that of GAPDH. STAT3 transcriptional activity was measured using a luciferase reporter assay and normalized by positive control constitutively expressing firefly luciferase. (F) SAHA-induced STAT3 status change versus cell sensitivity in breast cancer cells. Cells described in (E) were treated with SAHA at indicated concentrations for 8 hr. STAT3 phosphorylation was detected by immunoblotting. Lysates from each cell line, as separated by dashed lines, were blotted individually. STAT3 transcriptional activity was measured as described in (E) and normalized by that of the untreated group. (G) Association between STAT3 activation and SAHA sensitivity in breast cancer cells. Upper panel: waterfall plot of IC50s of SAHA in 21 breast cancer cell lines. Alteration of phospho-STAT3 upon SAHA treatment was differentiated by indicated colors. Lower panel: scatterplot of IC50s. Horizontal lines represented average of IC50s. ER, estrogen receptor; PR, progesterone receptor. Error bars represent means ± SD from triplicates or three independent experiments. ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S2. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 3 LIFR Upregulation Is Required for STAT3 Feedback Activation Caused by HDAC Inhibition (A) LIFR mRNA level alteration. MDA-MB-231 cells treated with SAHA for 8 hr were subjected qRT-PCR analysis. (B) LIFR protein level alteration. Protein levels in MDA-MB-231 cells treated with SAHA for indicated time were analyzed by immunoblotting. (C–E) STAT3 phosphorylation change. STAT3 activation was examined by immunoblotting. (C) Cells were treated with SAHA for 24 hr with or without indicated neutralizing antibodies for the last 60 min. L, low dose; H, high dose. Immunoglobulin G was used as a control. (D) Cells transfected with either non-targeting control (NC) or LIFR siRNA for 48 hr were treated with SAHA (5 μM) for a further 8 hr. (E) MDA-MB-231 cells were treated with SAHA alone for 8 hr or in combination with indicated cytokines (50 ng/mL) for the last 30 min. (F) STAT3 transcriptional activity. MDA-MB-231 cells were transfected with STAT3 luciferase reporter constructs for 48 hr followed by the treatment with SAHA for 24 hr with or without LIF stimulation for the last 30 min. Luciferase activity was normalized by that in the vehicle group. (G) Cell apoptosis analysis. MDA-MB-231 cells transfected with indicated siRNAs for 48 hr were treated with SAHA for a further 24 hr. Apoptotic cells were detected by annexin V/PI dual staining. (H) LIFR alteration versus SAHA sensitivity. LIFR level in the responsive and non-responsive cell subset as described in Figure 2E was examined by immunoblotting. Lysates from each cell line, as separated by dashed lines, were blotted individually. Lower panel: scatterplot of LIFR alteration. LIFR level was semi-quantified by densitometry and normalized by that of GAPDH. LIFR level change post SAHA treatment was normalized by that of vehicle group. Horizontal lines represented the average value of seven cell lines. (I) Baseline LIFR level versus SAHA sensitivity in breast cancer cells. Endogenous LIFR level was examined and semi-quantified as described in (H). IL6(R), interleukin-6 (receptor); IL11(R), interleukin-11 (receptor). Error bars represent means ± SD from triplicates. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S3. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 4 BRD4 Is Required for HDAC Inhibition-Induced LIFR Upregulation (A) Histone acetylation in LIFR promoter. MDA-MB-231 cells were treated with SAHA (5 μM) for 8 or 12 hr before being subjected to ChIP assay using anti-acetyl-histone H3K9 (Ac-H3K9) antibody followed by qPCR analysis using primers targeting indicated LIFR promoter region. (B) BRD4 enrichment in LIFR promoter. MDA-MB-231 cells were treated with SAHA (5 μM) for 12 hr before being subjected to ChIP assay using anti-BRD4 antibody. qPCR analysis was performed using the primers as used in (A). (C) LIFR mRNA level change. MDA-MB-231 cells were transfected with BRD4 siRNA or non-targeting control (NC) for 48 hr followed by SAHA treatment for a further 8 hr. Samples were analyzed by qRT-PCR assay. (D–F) LIFR and STAT3 phosphorylation change. MDA-MB-231 cells transfected with indicated siRNAs for 48 hr (D and E) or pretreated with JQ1 for 24 hr (F) were exposed to SAHA for a further 8 hr. LIFR and STAT3 phosphorylation change was detected by immunoblotting. (G) LIFR mRNA level change. Cells treated as in (F) were subjected to qRT-PCR assay. (H and I) Cell apoptosis analysis. MDA-MB-231 cells transfected with indicated siRNAs for 72 hr (H) or pretreated with JQ1 for 24 hr (I) were exposed to SAHA treatment for a further 24 hr. Apoptotic cells were measured by annexin V/PI dual staining. (J) Cell growth analysis. Cell growth of MDA-MB-231 cells after indicated treatment was measured by cell counting (72 hr) or clonogenic assay (10 days). Error bars represent means ± SD from triplicates. ∗∗p < 0.01, ∗∗∗p < 0.001. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 5 Concurrent Inhibition of HDACs and JAK1 Exhibits a Synergistic Effect in Breast Cancer Treatment (A) STAT3 phosphorylation change. Cells transfected with JAK1, JAK2 siRNA, or non-targeting control (NC) for 48 hr were exposed to SAHA (5 μM) for a further 8 hr. STAT3 activation and knockdown efficacy were examined by immunoblotting. Acetylated tubulin (Ac-Tubulin) was blotted as an internal control for HDAC inhibition. (B) Cell apoptosis analysis. MDA-MB-231 cells transfected with indicated siRNA for 48 hr were treated by SAHA for a further 24 hr. Apoptotic cells were detected by annexin V/PI dual staining. (C) STAT3 phosphorylation change. MDA-MB-231 cells were treated with SAHA (10 μM) alone or in combination with JAK2-selective inhibitor BSK805 (BSK, 5 μM) at or JAK1/2 pan-inhibitor INCB018424 (INCB, 10 μM) for 8 hr. STAT3 activation was examined by immunoblotting. (D) STAT3 transcriptional activity. MDA-MB-231 cells were pretreated with INCB for 12 hr followed by SAHA treatment for a further 8 or 12 hr. STAT3 transcriptional activity was measured by luciferase reporter assay. (E and F) Anti-apoptotic signaling alteration. MDA-MB-231 cells were pretreated with INCB for 12 hr followed by SAHA treatment for the times indicated. Apoptotic proteins were detected by immunoblotting. (G) Cell apoptosis analysis. MDA-MB-231 cells were treated with SAHA (10 μM) or/and INCB (10 μM) for 24 hr. Apoptotic cells were analyzed using annexin V/PI dual staining. (H) Cell growth assay. MDA-MB-231 cells were treated with SAHA alone or in combination with INCB for 72 hr. Cell growth inhibition was assessed by cell counting. (I) Synergistic anti-cancer efficacy in vivo. MDA-MB-231 xenograft model or 4T1 allograft model was treated with SAHA (100 mg/kg) and INCB (100 mg/kg) alone or in combination daily for 14 days. Endpoint tumor growth inhibition rate was analyzed using two-way ANOVA. Error bars represent means ± SEM (n = 5 mice per group). Error bars represent means ± SD from triplicates except (I). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. See also Figure S4. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 6 Combined HDAC and JAK Inhibition Is a Potential Treatment for TNBC The therapeutic efficacy of SAHA or combinations was assessed in eight breast cancer patient-derived models (PDX). Mice were orally treated with SAHA (150 mg/kg) and INCB (100 mg/kg) alone or in combination daily for up to 4 weeks. Tumor tissues were collected at 8 hr after last dosing and subjected to immunoblotting or immunohistochemistry analysis. (A) Waterfall plot of SAHA response in individual PDX models. Tumor response was indicated by endpoint tumor growth inhibition (TGI) rate. (B) Basal STAT3 status in the PDX models shown in (A). Intratumoral STAT3 and phospho-STAT3 levels in vehicle group of each model were examined by immunoblotting, semi-quantified by densitometry, and normalized by that of GAPDH in each sample. Shown are the relative STAT3 and phospho-STAT3 levels individually normalized by phospho-STAT3 level in model BR-0028. Error bars represent means ± SD from three representative samples. (C) Tumor response to SAHA treatment alone (SAHA) versus tumor response to SAHA-INCB combination (Combo). Tumor response is presented as endpoint relative tumor volume of each model. ∗p < 0.05. (D) Waterfall plot of ΔTGI. ΔTGI of each model was obtained by subtracting endpoint TGI (SAHA) from TGI (SAHA-INCB). The horizontal dashed line indicates a cutoff of 20% TGI increase. (E) Intratumoral phospho-STAT3 change stimulated by SAHA in the models shown in (D). Phospho-STAT3 level was examined and semi-quantified as described in (B). Phospho-STAT3 level change in SAHA-treated group was normalized to that of vehicle group. The resultant phospho-STAT3 fold change in each PDX model was individually compared with that of model BR-1458 (indicated by the horizontal dashed line in the left panel). Error bars represent means ± SD from three representative samples. Representative blots of two models are shown in the right panel. (F) Tumor growth curve of model BR-0028. Mice were treated with SAHA (150 mg/kg), INCB (100 mg/kg), and JQ1 (50 mg/kg) alone or in indicated combinations daily for 28 days. Growth curve was plotted by measuring the relative tumor volume twice per week. ∗p < 0.05. (G) Molecular alterations in model BR-0028. Tumor samples as described in (F) were collected at 8 hr after the last dosing and intratumoral molecular changes were detected using immunohistochemistry analysis. Scale bar, 20 μm. (H) Proposed working model. HDAC inhibition increases histone acetylation at LIFR gene promoter, which recruits BRD4 and activates LIFR transcription. Upregulated LIFR, in the presence of LIF family cytokines (such as LIF, OSM, CT1, etc.), activates JAK1-STAT3 signaling and promotes transcription of downstream anti-apoptotic genes BCL-2 and MCL-1, causing reduced response to HDAC inhibitors. LIF, leukemia inhibitory factor; OSM, oncostatin M; CT1, cardiotrophin-1. Error bars represent means ± SEM (n = 5 mice per group) except (B and E). See also Figure S5 and Table S3. Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions

Cancer Cell 2016 30, 459-473DOI: (10.1016/j.ccell.2016.08.001) Copyright © 2016 Elsevier Inc. Terms and Conditions