Targeting the BRD4/FOXO3a/CDK6 axis sensitizes AKT inhibition in luminal breast cancer

Abstract

BRD4 assembles transcriptional machinery at gene super-enhancer regions and governs the expression of genes that are critical for cancer progression. However, it remains unclear whether BRD4-mediated gene transcription is required for tumor cells to develop drug resistance. Our data show that prolonged treatment of luminal breast cancer cells with AKT inhibitors induces FOXO3a dephosphorylation, nuclear translocation, and disrupts its association with SirT6, eventually leading to FOXO3a acetylation as well as BRD4 recognition. Acetylated FOXO3a recognizes the BD2 domain of BRD4, recruits the BRD4/RNAPII complex to the CDK6 gene promoter, and induces its transcription. Pharmacological inhibition of either BRD4/FOXO3a association or CDK6 significantly overcomes the resistance of luminal breast cancer cells to AKT inhibitors in vitro and in vivo. Our study reports the involvement of BRD4/FOXO3a/CDK6 axis in AKTi resistance and provides potential therapeutic strategies for treating resistant breast cancer.

Fig. 1

BET inhibitors increase the growth suppressive effects of AKT inhibitor (AKTi). a, b Breast cancer cell lines were treated with BET inhibitors (JQ1 or MS417, 1 μM) or AKTi (MK2206, AZD5363, and GSK690693, 1 μM), or in combination for 96 h. The growth suppressive effects were measured by either cell count (a) or MTT analyses (b). Data presented are representative of three experiments performed in quadruplicate as the mean ± SD. ^*/^□/^⁑ p < 0.01 when treatment group containing JQ1 or MS417 plus either MK2206 (*), or AZD5363 (□), or GSK690693 (⁑) is compared with the corresponding group with single inhibitor. c Expression of BRD4 family members was knocked down by siRNA in breast cancer cell lines, followed by treatment with various inhibitors (1 μM) for 96 h as described above. The growth suppressive effects were reflected by cell count assay. NTC stands for non-target control of siRNA (scramble). Data presented are representative of three experiments performed in triplicate as the mean ± SD. ^*/^‡/^⁑ p < 0.01 when group with BRD4-knockdown plus either MK2206 (*), or AZD5363 (^‡), or GSK690693 (⁑) is compared with other knockdown groups in combination with same inhibitor

Fig. 2

Prolonged treatment with AKTi induces FOXO3a acetylation and BRD4 association. a BT474 and T47D cells were treated with 1 μM MK2206 for 6 h, the cellular localization of FOXO3a (green) and SirT6 (red) were analyzed by immunofluorescent staining. Nuclei stained with DAPI (blue). Scale bar, 50 μM. b Schematic diagram showing the domain structure of FOXO family members. Sequence alignment indicates the highly conserved GK-X-GK motif among K5/8 of histone H4, K73/76 of Twist, K242/245 of FOXO3a, in which the lysine residues have been reported to be acetylated in vivo. c BT474 and T47D cells were treated with AKTi for 4 day, cell extracts were subjected to immunoprecipitation (IP) with FOXO3a antibody, the acetylation of endogenous FOXO3a and its interaction with BRD4 were analyzed by western blot using pan acetylated-lysine and BRD4 antibodies. d BT474 cells were treated with 1 μM MK2206 for various time intervals. Cell extracts were prepared for IP with FOXO3a antibody, the acetylation of endogenous FOXO3a and bound BRD4 were analyzed by western blotting. e BT474 cells were treated with 1 μM MK2206 for 4 days. Cell extracts were prepared for IP with FOXO3a antibody in a buffer containing either 2 μM JQ1 or MS417. Western blottings were used to detect FOXO3a acetylation and its interaction with BRD4. f BT474 cells were treated with 2 μM of TSA and 4 mM of Nicotinamide overnight. Cell extracts were prepared and FOXO3a were IP in a buffer containing either JQ1 or MS417 as described above. FOXO3a acetylation and its interaction with BRD4 were analyzed by western blotting

Fig. 3

BD2 of BRD4 recognizes and interacts with K242/245 di-acetylated FOXO3a. a HA-tagged FOXO3a was co-expressed with Flag-tagged BD1 and BD2 in HEK293 cells treated with 2 μM TSA and 4 mM Nicotinamide overnight. After IP with anti-HA antibody, FOXO3a acetylation and its association with BD1 and BD2 were examined by western blotting using pan acetylated-lysine and anti-Flag antibodies. b Wild-type (WT) or K242R/245 R mutant (KR) of FOXO3a were co-expressed with Flag-tagged BRD4 in HEK293 cells treated with 1 μM MK2206 for 2 day. After IP with anti-HA antibody, FOXO3a acetylation and its association with BRD4 were examined by western blotting using pan acetylated-lysine and anti-Flag antibodies. c WT or KR mutant of FOXO3a was co-expressed with Flag-tagged BD2 in HEK293 cells treated with 2 μM TSA and 4 mM Nicotinamide overnight. After IP with anti-HA antibody, FOXO3a acetylation and its association with BD2 were examined by western blotting. d The FOXO3a-binding selectivity of the BD1 and BD2 of BRD4. 2D 1H-15N HSQC spectra of BRD4-BD1 or BD2 in the free form (black) and in the presence of a FOXO3a peptide (residues 238–249, NPDGG-K242ac-RG-K245ac-APR) that is non-acetylated, or contains single or di-acetylated K242 or K245 with protein:ligand molar ratio of 1:5 (red). Other FOXO members were also included for analyses. e Left and middle panel: Stereo ribbon diagram of the 3D solution structure of the BRD4-BD2 bound to a di-acetylated K242ac/K245ac FOXO3a peptide (cyan). Side chains of key residues engaged at the protein/peptide interactions are depicted and color-coded by atom type. Right panel: surface-filled representation of the 3D structure of the BRD4-BD2 highlights its same mode of ligand recognition for FOXO3a (cyan) and Twist (yellow) peptides. f FOXO3a was co-expressed with full-length WT- or HN-BRD4 in HEK293 cells treated with 1 μM MK2206 for 2 day. After FOXO3a and BRD4 were immunoprecipitated, the associated BRD4 and FOXO3a were examined by western blotting

Fig. 4

AKTi induces FOXO3a acetylation by disrupting the SirT6/FOXO3a interaction. a Flag-tagged Sirt family members were co-expressed with HA-FOXO3a in HEK293 cells. After IP with HA antibody, the bound Sirt members were examined by western blotting using anti-Flag antibody. b After BT474 and T47D cells were treated with various AKTi for 4 day, cell extracts were subjected to IP with FOXO3a antibody, binding of endogenous SirT6 was analyzed by western blotting. c Expression of endogenous SirT6 was knocked down by siRNA in BT474 cells, followed by treatment with 1 μM MK2206 for 2 days. After IP with FOXO3a antibody, endogenous FOXO3a acetylation was analyzed by western blotting using pan acetylated-lysine antibody. d Acetylated FOXO3a was purified from TSA and Nicotinamide-treated BT474 cells by FOXO3a antibody and then incubated with purified recombinant SirT6 protein in a deacetylation assay as described in the Experimental Procedures. The acetylated state of FOXO3a was assessed by western blotting using pan acetylated-lysine antibody. The immuno-purified FOXO3a and SirT6 used in this assay were analyzed by western blotting. e WT or TM mutant (with T32, S253, and S315 of FOXO3a changed to alanine, which confers FOXO3a localized exclusive in the nucleus) of FOXO3a were expressed in HEK293 cells followed by treatment with 1 μM MK2206. After IP FOXO3a, the association of SirT6 was analyzed by western blotting using anti-Flag antibody. f BT474 and T47D cells were pre-treated with 1 μM MK2206 followed by stimulation with 10 ng/ml of IGF-1 for 4 h, cell extracts were subjected to IP with SirT6 antibody, phosphorylation of endogenous SirT6 was analyzed by either specific SirT6-S338 phosphorylation or phospho-AKT substrate antibody by western blotting, respectively. g BT474 and T47D cells were treated with 1 μM MK2206, cell extracts were subjected to IP with AKT antibody, the bound SirT6 was analyzed by western blotting. h WT, S338A, or S338E mutant of SirT6 were expressed in HEK293 cells followed by treatment with or without 1 μM MK2206. After IP FOXO3a, the bound SirT6 was analyzed by western blotting using anti-Flag antibody

Fig. 5

AKTi induces CDK6 expression. a A proposed model illustrating the interplay of FOXO3a-AKT-SirT6, which leads to the formation of the FOXO3a-BRD4 complex (left panel). WT, S338A, or S338E mutant of SirT6 were expressed in indicated cells followed by treatment with 1 μM MK2206 for 4 days. The growth suppressive effects were examined by cell count analyses. SirT6 expression was confirmed by western blotting. Data presented are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 (MK2206-treated group is compared with control group without treatment); ^#p > 0.05 (MK2206-treated S338A-SirT6 group is compared with control group without treatment); and ^**p < 0.01 (MK2206-treated S338E-SirT6 group is compared with MK2206-treated vector and WT-SirT6 groups). b BT474 and T47D cells were treated with 1 μM MK2206 for 3 days, differentially expressed genes with more than 2.5-fold changes on RNA-sequencing analysis (GSE118148) from both cell lines were used to identify potential FOXO3a target genes shown in the heat-map. c BT474, T47D, and ZR75 cells were treated with 1 μM MK2206 for 3 days. The mRNA and protein levels of CDK6 were analyzed by RT-PCR and western blotting, respectively. Data presented are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when MK2206-treated group is compared with control group. d BT474 cells were treated with 1 μM MK2206 for different time intervals. CDK6 expression was analyzed by western blotting. e E2F driven luciferase reporter was expressed in indicated cells treated with MK2206, JQ1, Palbociclib (PD), or in combination. Luciferase activities were determined by dual luciferase assay. Data presented are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when MK2206-treated group is compared with other groups. f BT474 and T47D cells were treated with MK2206 (1 μM) in the absence or presence of JQ1 or MS417 (1 μM). CDK6 expression was analyzed by western blotting. g BT474 cells were treated with three different AKTi (1 μM) in the absence or presence of JQ1 or MS417 (1 μM). CDK6 expression was analyzed by western blotting

Fig. 6

CDK6 is a targeted gene of FOXO3a/BRD4. a Schematic depiction of the CDK6 promoter and CDK6 reporter luciferase constructs used. Two consensus FOXO3a-binding motifs are indicated. b Effects of WT and KR mutant of FOXO3a, BRD4 on the activity of CDK6 promoter luciferase in the presence or absence of MK2206. c Effects of WT-FOXO3a, BRD4 on the activation of WT or mutant CDK6 promoter luciferase in the presence of MK2206. For b and c, ^*/#/**p < 0.01. d FOXO3a, BRD4, H4ac4, and RNA-Pol II association at the CDK6 promoter as assessed by ChIP. ^*/⁑p < 0.01 when FOXO3a-ChIP or H4ac-ChIP in either MK2206-treated group or MK2206 plus JQ1-treated group is compared with control group. ^#/‡p < 0.01 when BRD4-ChIP or RNA-Pol II-ChIP in MK2206-treated group is compared with control group. e Cells treated with MK2206 were analyzed by sequential ChIP. The first-round of immunoprecipitation used anti-IgG, anti-FOXO3a or anti-BRD4 antibodies, followed by a second-round of immunoprecipitation using anti-IgG, anti-FOXO3a, or BRD4 antibodies. Pulldown DNA was analyzed by RT-PCR using specific CDK6 promoter primers and nonspecific primers. ^*/#/‡/⁑p < 0.01 when individual group with specific antibodies (FOXO3a and BRD4) used in the first-round of ChIP is compared with the corresponding group using IgG in the first-round of ChIP. f Effects of FOXO3a-knockdown, MK2206 treatment, or both on the association of FOXO3a, BRD4, RNA-Pol II, and H4ac at the CDK6 promoter as assessed by ChIP. ^*/#/⁑p < 0.01 when specific ChIP in MK2206-treated group is compared with the corresponding ChIP in control, or FOXO3a-knockdown or combination groups. g WT or KR mutant of FOXO3a was expressed in BT474 cells with endogenous FOXO3a knockdown. Rescue CDK6 expression was analyzed by western blotting; growth inhibition was assessed by cell count. ^*p < 0.01 when WT-FOXO3a group is compared with either vector or KR-FOXO3a group. h WT or HN mutant of BRD4 was expressed in BT474 cells with endogenous BRD4 knockdown. Rescue CDK6 expression and growth inhibition were assessed as in g. ^*p < 0.01 when WT-BRD4 group is compared with either vector or HN-BRD4 group. From b to h, data are representative of three experiments performed in triplicate as the mean ± SD

Fig. 7

Inhibition of CDK6 confers drug sensitivity to AKTi. a Left panel: MK2206-resistant cell lines were established by growing T47D and ZR75 cells in increasing concentrations of MK2206 for 4 months. Inductions of CDK6 expression was examined by western blot. Middle panel: the growth suppressive effect of MK2206 was presented by the percentage of S-phase decrease in parental or resistant cells after 2 μM MK2206 treatment for 4 days. Data are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when resistant cells were compared with parental cells. Right panel: the growth inhibition of these cells responded to the combination of MK2206 with JQ1 or Palbociclib was measured by cell count assays. Data are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when MK2206-treated group was compared with untreated group in parental cells; ^#p > 0.05 when MK2206-treated group was compared with untreated group in resistant cells. b Left panel: induction of CDK6 expression in parental and CDK6-overexpressing cells were examined by western blotting. Middle panel: the growth suppressive effect of MK2206 was presented by the percentage of S-phase decreased in parental or CDK6-overexpressing cells after 2 μM MK2206 treatment for 4 days. Data are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when CDK6-expressing cells were compared with vector control cells. Right panel: cell count assay was used to detect the growth inhibition effects of the combination of MK2206 with JQ1 or Palbociclib. Data are representative of three experiments performed in triplicate as the mean ± SD. ^**p < 0.01 when MK2206 plus JQ1 treatment group in CDK6-expressing cells was compared with corresponding group in parental cells. c Lapatinib (1 μM) also induced FOXO3a acetylation, its association with BRD4 and CDK6 expression in BT474 cells as analyzed by IP-western blotting. The growth inhibition effects of the combination of Lapatinib with JQ1 or Palbociclib on BT474 cells were reflected by cell count assays. Data are representative of three experiments performed in triplicate as the mean ± SD. ^*p < 0.01 when combinatory treatment group was compared with Lapatinib-treatment group

Fig. 8

The FOXO3a-BRD4-CDK6 axis is critical for AKTi resistance in vivo. a, b BT474 cells were injected into athymic mice subcutaneously. When tumors from mice reached 250 mm³, mice were divided into six groups (six mice/group) and treated with solvent control, MK2206 (90 mg/kg), JQ1 (50 mg/kg), Palbociclib (100 mg/kg), or in combinations, respectively. The size of tumor was recorded every 3 days. Data are represented as a mean ± SEM from six mice. ^*/#p < 0.01 when combinatory treatment group is compared with either control or single-agent treatment group. c Tumor weight was also measured from above. Data are represented as a mean ± SEM from six mice. Data are represented as a mean ± SEM from six mice. ^*/⁑p < 0.01 when combinatory treatment group is compared with either control or single-agent treatment group. ^‡p > 0.05 when both combinatory treatment groups were compared. d, e The 343 surgical specimens of breast cancer were immuno-stained using antibody against CDK6 and the control serum (data not shown). Images with consecutive IHC staining of CDK6 and H&E staining in four cases of breast tumors are shown (Scale bar = 100 μm). Recurrence rate was calculated according to CDK6 scores and clinical information of these patients. f A proposed model illustrating the formation of FOXO3a-BRD4 complex at the enhancer/promoter of CDK6, which leads to the transcriptional activation of CDK6 gene in response to AKTi. Either targeting FOXO3a-BRD4 complex by BET inhibitors or inhibition of CDK6 by Palbociclib reverses the AKTi resistance