Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 34 (6), 893-905.e8

Loss of the FAT1 Tumor Suppressor Promotes Resistance to CDK4/6 Inhibitors via the Hippo Pathway

Affiliations

Loss of the FAT1 Tumor Suppressor Promotes Resistance to CDK4/6 Inhibitors via the Hippo Pathway

Zhiqiang Li et al. Cancer Cell.

Abstract

Cyclin dependent kinase 4/6 (CDK4/6) inhibitors (CDK4/6i) are effective in breast cancer; however, drug resistance is frequently encountered and poorly understood. We conducted a genomic analysis of 348 estrogen receptor-positive (ER+) breast cancers treated with CDK4/6i and identified loss-of-function mutations affecting FAT1 and RB1 linked to drug resistance. FAT1 loss led to marked elevations in CDK6, the suppression of which restored sensitivity to CDK4/6i. The induction of CDK6 was mediated by the Hippo pathway with accumulation of YAP and TAZ transcription factors on the CDK6 promoter. Genomic alterations in other Hippo pathway components were also found to promote CDK4/6i resistance. These findings uncover a tumor suppressor function of Hippo signaling in ER+ breast cancer and establish FAT1 loss as a mechanism of resistance to CDK4/6i.

Keywords: CDK4/6 inhibitors; FAT1; Hippo pathway; RB1; YAP; abemaciclib; breast cancer; drug resistance; palbociclib; ribociclib.

Figures

Figure 1.
Figure 1.. Genomic alterations correlated with CDK4/6 inhibitor resistance.
(A) The significance of association of altered genes with progression-free survival (PFS) on CDK4/6i based on Cox proportional hazard models. Color indicates statistical significance by (q value < 0.05) and size of the circle reflects the frequency of alteration in the cohort. All q values are calculated based on Benjamini and Hochberg method correction of log-rank p values. (B, C, D) PFS of patients receiving CDK4/6i and with tumors harboring functional alterations in PIK3CA (B), CCND1 amplification (C), and RB1 loss (D) as compared to patients whose tumors were wild-type for these lesions (gray). Hazard ratios (HR) and 95% confidence intervals (95% CI) are based on Cox proportional hazard models stratified by regimen. All p values as indicated, log-rank test. See also Figure S1 and Tables S1-S4.
Figure 2.
Figure 2.. FAT1 loss and clinical resistance.
(A) Frequency of type of FAT1 alterations by type of tumor sample (metastatic vs primary) in 1501 ER+ breast cancer cases (B) The pattern, frequency, and type of genomic alterations in key breast cancer genes of the tumors presented in panel A by different classes of FAT1 alterations, comparing the FAT1 homozygous deletion or truncating mutations (left) with FAT1 missense mutations (right). (C) progression-free survival (PFS) of patients receiving CDK4/6i with tumors harboring FAT1 missense mutations (green), homozygous deletion or truncating mutations (blue) as compared to patients whose tumors were wild-type for these lesions (gray). Hazard ratios (HR) and 95% confidence intervals (95% CI) are based on Cox proportional hazard models stratified by regimen. All p values as indicated, log-rank test. (D) Baseline and post-progression PET-CT scan of the liver lesion in a patient (PFS 2.5 months) on palbociclib and fulvestrant. (E) Alignments of sequence reads on chromosome 4 in the pre-treatment lung lesion (top) and post-treatment liver lesion (bottom) showing paired ends encompassing the chr4:175870481–187630480 deletion (blue) along with sequence reads having close to average insert size (grey). The deletion is represented schematically by the red connected vertical lines on the chromosome ideogram. In the intermediate panels, the orange and grey bars display all the transcripts predicted for ADAM29 and FAT1 in Ensembl while the corresponding grey density plots above each sequence alignment window show the base level coverage (scale on the y-axis to the left). The ADAM29-FAT1 fusion was detected in the post-treatment liver tumor (bottom, supported by 46 paired-reads, average depth of coverage 791x) but was not called in the pre-treatment lung tumor (detected 2 paired-reads, average depth of coverage 1309x). In all panels, genomic coordinates are Kb as displayed immediately below the chromosome ideogram. (F) Schematic representation of the ADAM29-FAT1 fusion inferred from the mapping positions to the paired ends and based on the Ensembl canonical transcripts of the two genes. See also Figure S2 and Tables S1-S3.
Figure 3.
Figure 3.. FAT1 loss promotes resistance to CDK4/6 inhibitors.
(A) Proliferation of FAT1-sh, FAT1-CR, and parental MCF7 cells without drug treatment. Data are represented as mean ± SD; n = 3. (B) Proliferation of FAT1-loss and parental MCF7 cells exposed to 50 nM of abemaciclib. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test of day 35 data with Dunnett’s method correction compared with parental. (C) Immunoblotting of total and phosphorylated Rb in parental and FAT1-loss cells. (D) IC50s of abemaciclib, palbociclib or ribociclib in different MCF7 cell models. IC50s were calculated based on day 5 data of various doses of drug treatment. See also Figure S3.
Figure 4.
Figure 4.. FAT1 suppresses CDK6 expression.
(A) Relative mRNA level of indicated genes normalized to RPLP0 in FAT1-CR, FAT1-sh, Renilla-sh, and parental MCF7 cells. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with parental or Renilla-sh. (B) Immunoblotting of CDK4/6 and cyclin D in parental and FAT1-loss MCF7 cells. (C) Relative mRNA level of indicated genes normalized to RPLP0 in FAT1-CR and parental CAMA-1 cell. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with parental. (D) mRNA expression of CDK6 in the cohort of patients with ER+ breast cancer (TCGA provisional dataset) harboring FAT1 wild-type or deletion alteration was analyzed and p values were calculated using two-tailed Student’s t-tests. (E) mRNA expression of CDK4 and CDK6 in the cohort of patients with ER+ breast cancer (TCGA provisional dataset) was analyzed and p values were calculated using two-tailed Student’s t-tests. For (D) and (E), the bottom and top of the boxplot represent the lower and upper quartiles, respectively, and the band near the middle of the box represents the median. The upper whisker represents the upper quartile + 1.5 x interquartile range (IQR) and the lower whisker represents the lower quartile - 1.5 x IQR. (F) Ratio of mRNA expression of CDK6 normalized to that of CDK4 in multiple CDK4/6i-resistant (red/orange) and sensitive (blue) ER+ breast cancer models. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with MCF7. (G) Proliferation of parental and FAT1-CR-1 cells, constitutively expressing CDK6-shRNA, or CDK6shRNA plus CDK6 re-expression, exposed to 100 nM of abemaciclib. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with parental. (H) Tumor growth curve of PDXs in response to ribociclib treatment (200 mg/kg). (I) Representative immunohistochemical (IHC) images of PDX samples with different sensitivity to CDK4/6 inhibitors stained with FAT1 or CDK6 antibodies. Scale bars 50 μm. (J) Representative IHC images of human breast tumors with different FAT1 alterations stained with FAT1 or CDK6 antibodies. Scale bars 200 μm. See also Figure S4.
Figure 5.
Figure 5.. CDK6 is regulated by Hippo signaling in ER+ breast cancer cells
(A) Relative mRNA expression of indicated genes normalized to RPLP0 in FAT1-CR and parental MCF7 cells. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA statistical test with Dunnett’s method correction compared with parental. (B) Immunoblotting of indicated proteins in parental and FAT1 ablation cells with YAP1 and/or TAZ knockdown. (C) Proliferation of parental and FAT1-CR-1 cells, constitutively expressing YAP1-shRNA and/or TAZ-shRNA, with or without 100 nM of abemaciclib treatment. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with FAT1-CR-1. (D) ChIP-qPCR assay for YAP/TAZ showing induction of binding to CDK6 promoter upon FAT1 knockdown. Data are represented as mean ± SD; n = 3. All p values are based on two-tailed Student’s t-tests. (E) Immunofluorescent images of YAP (green) localization in parental MCF7 and FAT-sh-A cells treated with DMSO or 10 μM verteporfin. DAPI included as a nuclear stain (blue). Scale bars, 20 μm. (F) Phosphorylation and expression of Hippo pathway components in parental and FAT1 ablation MCF7 cells. (G) Gene enrichment analysis of RNA-seq result using GSEA. The relative expression of FAT1, CDK6 and CTGF in RNA-seq is summarized at the bottom. See also Figure S5 and Table S5.
Figure 6.
Figure 6.. Genomic alterations in the Hippo signaling pathway and CDK4/6i resistance.
(A) Immunofluorescence images for FAT1 (green) and NF2 (red) in MCF7 cells. DAPI was used as a nuclear stain (blue). Scale bars, 20 μm. (B) Immunoblotting for cyclin D-CDK4/6 and Rb in parental and NF2 knockout (NF2-CR) MCF7 cells in the absence or presence of 100 nM abemaciclib. (C) Relative mRNA expression of indicated genes normalized to that of RPLP0 in NF2-CR and parental MCF7 cells. Data are represented as mean ± SD; n = 3. All p values are based on one-way ANOVA test with Dunnett’s method correction compared with parental. (D) Proliferation of NF2-CR and parental MCF7 cells in the presence of 50 nM abemaciclib. Data are represented as mean ± SD; n = 3. All p values one-way ANOVA statistical test of day 5 data with Dunnett’s method correction compared with parental. (E) Pattern and frequency of likely pathogenic mutations and focal amplifications or deletions targeting components of the Hippo pathway including FAT1, LATS1, LATS2, and YAP1 and Cell Cycle pathway including RB1, and CDK6. (F) Association of altered pathways with PFS on CDK4/6i. Hazard ratios (HR) and 95% confidence intervals (9% CI) are based on Cox proportional hazard models stratified by regimen. Error bars represent 95% CIs. See also Table S6.

Similar articles

See all similar articles

Cited by 18 articles

See all "Cited by" articles

Publication types

MeSH terms

Feedback