Inhibition of Lapatinib-Induced Kinome Reprogramming in ERBB2-Positive Breast Cancer by Targeting BET Family Bromodomains

Abstract

Therapeutics that target ERBB2, such as lapatinib, often provide initial clinical benefit, but resistance frequently develops. Adaptive responses leading to lapatinib resistance involve reprogramming of the kinome through reactivation of ERBB2/ERBB3 signaling and transcriptional upregulation and activation of multiple tyrosine kinases. The heterogeneity of induced kinases prevents their targeting by a single kinase inhibitor, underscoring the challenge of predicting effective kinase inhibitor combination therapies. We hypothesized that, to make the tumor response to single kinase inhibitors durable, the adaptive kinome response itself must be inhibited. Genetic and chemical inhibition of BET bromodomain chromatin readers suppresses transcription of many lapatinib-induced kinases involved in resistance, including ERBB3, IGF1R, DDR1, MET, and FGFRs, preventing downstream SRC/FAK signaling and AKT reactivation. Combining inhibitors of kinases and chromatin readers prevents kinome adaptation by blocking transcription, generating a durable response to lapatinib, and overcoming the dilemma of heterogeneity in the adaptive response.

Figure 1. Lapatinib induces dynamic kinome responses

(A) Flow chart for experimental design. (B) MIB/MS kinome activation dynamics over 48 h of 300nM lapatinib treatment in SKBR-3 cells. Ratios greater than 1 indicate increased MIB-binding (increased activity) and values less than 1 indicate decreased MIB-binding (decreased activity) relative to control cells treated with DMSO. Data presented is the average of four biological replicates. (C) MIB-binding dynamics suggest reactivation of ERBB2 in SKBR-3 cells but continued suppression in BT474 cells. (D) Hierarchical clustering of MIB-binding ratios in SKBR-3 cells identifies clusters of dynamic kinase behavior. (E) Dynamics of a select set of kinases illustrates multiple behaviors in response to lapatinib. Four points graphically indicate 0, 4, 24, and 48 h MIB-binding. Blue, inhibited; red, activated. (F) Western blots validate MIB/MS results and identify upregulation of ERBB3, DDR1, FRK, and PKCδ and increased activation of FAK, SRC family kinases (SFKs), and STAT3 in response to lapatinib. AKT and ERK1/2 are inhibited at 4 h but become reactivated over 72 h. Treatment with 1µM lapatinib re-inhibits EGFR and ERBB2 but has little effect on other kinases. (G) BT474 MIB-binding dynamics of tyrosine kinases and MEK/ERK and AKT/mTOR pathways. (H) Western blots indicate upregulation of ERBB3, INSR, and IGF1R total levels and increase in SFK phosphorylation after lapatinib treatment in BT474 cells. Also see Figures S2

Figure 2. MIB/MS and RNA sequencing define heterogeneity in the adaptive response

(A) Statistically significant MIB-binding changes after 48 h lapatinib treatment based on Benjamini-Hochberg adjusted p-values at FDR of 0.05 and standard deviation in five ERBB2+ cell lines depicted graphically. Kinome trees reproduced courtesy of Cell Signaling Technology, Inc. (www.cellsignal.com). (B) Lapatinib-induced MIB-binding changes of tyrosine kinases illustrates differences between cell lines and identifies common response of INSR and IGF1R activation. (C) Principal component analysis identifies kinases that drive the variation in kinome response. Kinases captured in at least 3 out of 4 MIB/MS runs per cell line (67 kinases) were used. (D) Lapatinib induces 2-fold changes up or down in 18-20% of expressed mRNA transcripts. (E) Gene ontology terms enriched in commonly upregulated mRNAs between SKBR-3 and BT474 cells ide ntifies glucose regulation and transcriptional regulation as most significant processes. (F) Kinase mRNAs upregulated by lapatinib at least 2-fold by RNAseq in SKBR-3 and BT474. Hatched bars indicate RTKs and red bars indicate common upregulated kinases. Also see Figures S2 and S3

Figure 3. Cell lines exhibit variability in kinases that drive growth in the presence and absence of lapatinib

(A) Lapatinib combined with increasing doses of BMS754807 (IGF1R/INSR inhibitor) causes an increase in AKT phosphorylation in SKBR-3 but a decrease in BT474 relative to lapatinib treatment alone after 24 h. (B) BMS754 inhibits ERBB2/3 phosphorylation as a single agent, and when combined with lapatinib causes a further inhibition of AKT phosphorylation after 24 h. (C) Colony formation assays indicate heterogeneity in the kinases that contribute to growth. IGF1R/INSR inhibition has an additive effect with lapatinib in MDA361. Dasatinib is additive in BT474 and MDA361 but does not significantly enhance growth inhibition of other lines. SKBR-3 cells display synergism between lapatinib and PF228 (FAK inhibitor) or BGJ398 (FGFR inhibitor) but other cell lines show varying degrees of growth-inhibition by FAK or FGFR inhibition alone and in combination with lapatinib. SKBR-3, BT474, and HCC1419 treated for 4 weeks, HCC1954 and MDA361 treated for 5 weeks. Lapatinib doses: 100nM SKBR3; 30nM BT474; 10nM HCC1419; 300nM HCC1954 and MDA361. Data presented is mean ± SD of three technical replicates. * indicates significant difference from lapatinib alone (p≤0.05). (D) MIB/MS profile of SKBR-3 and BT474 cells after 48h treatment with 300nM Lapatinib, 30nM Dasatinib, or the combination. Dasatinib inhibits MIB-binding of multiple tyrosine kinases, but not FAK1 and FAK2 in SKBR-3 cells. (E) Western blots after 48 h demonstrate Dasatinib inhibits Lapatinib-induced SFK phosphorylation but increases FAK phosphorylation. (F) Western blots after 48 h indicate PF228 inhibits FAK and SFK phosphorylation, but increases AKT and ERK1/2 phosphorylation. (G) Western blots indicate FGFR inhibition alone slightly reduces AKT and ERK phosphorylation at 4 h, but elicits strong reactivation by 48 h. Combination with lapatinib indicates FGFRs regulate ERBB signaling and SFK and FAK phosphorylation. 300nM Lapatinib and 300nM BGJ398 added directly to media at 0 h. Media was not changed throughout experiment. (H) MIB/MS analysis of 300nM lapatinib, 300nM BGJ398, or the combination after 48 h indicates FGFRs regulate multiple lapatinib-induced TKs. (I) Matrix of p-values comparing growth inhibition of lapatinib alone versus lapatinib + kinase inhibitor in colony formation assays. Red, significant (p≤0.05); blue, not significant (p≥0.05).

Figure 4. Multiple unrelated kinases contribute to the growth of lapatinib-resistant cells

(A) Parental or 300nM lapatinib-resistant (LapR) SKBR-3 and BT474 cells were transfected with siRNAs against GAPDH (control) or ERBB receptors and cultured for 96 h. Both parental lines are strongly growth-inhibited by ERBB2 and ERBB3 knockdown. LapR cells are less dependent on ERBB2 but remain similarly dependent on ERBB3. (B) MIB/MS long tail plots of most-activated kinases in LapR SKBR-3 and BT474 cells, relative to parental cells. Kinases in red are commonly over-activated in SKBR-3 and BT474. (C) MIB/MS profile of tyrosine kinases from LapR SKBR-3 and BT474 cells. LapR SKBR-3 cells display enhanced MIB-binding of ERBB3, DDR1, FGFR2, MET, FRK, and SRC. LapR BT474 have increased activity of multiple FGFRs, EPHA7, IGF1R, MERTK, MET, LYN, and FAK1. Data presented is mean of two biological replicate MIB/MS experiments. (D) Western blots indicate RTK upregulation in LapR SKBR-3 cells and reactivation of AKT/ERK signaling. LapR BT474 cells display suppressed activity of AKT and ERK relative to parental cells. P, parental; R, LapR. (E) 96 h siRNA knockdown in LapR SKBR-3 cells indicates slight dependency on ERBB family, and a stronger dependency on DDR1, FGFR2, and CDK5. BT474 cells are growth-inhibited by ERBB3, FGFR2, and CDK5 knockdown. Data presented in A and E is mean ± SD of six technical replicates. Also see Figure S4

Figure 5. BET bromodomain inhibition suppresses lapatinib-induced kinome reprogramming and arrests growth

(A) Kinome reprogramming leads to transcriptional upregulation of multiple alternative kinases capable of reactivating or bypassing ERBB2-directed signaling. We hypothesize by inhibiting the BET family of bromodomain-containing acetylation readers, we can prevent the adaptive response at an epigenetic level. (B) Western blots demonstrate JQ1 (BET family bromodomain inhibitor) suppresses lapatinib-induced ERBB3 phosphorylation and expression at 300nM, and inhibits reactivation of AKT in SKBR-3 cells. 48h treatments. (C) Western blots indicate JQ1 blocks protein expression of multiple kinases involved in lapatinib resistance, and leads to a decrease in ERBB family, SFK, FAK, and PKCδ phosphorylation. JQ1/lapatinib combinations inhibit AKT and p70 S6K phosphorylation more than Lapatinib alone and increase cleavage of PARP. 48h treatments. (D) qRT-PCR after 24h treatment shows JQ1 inhibits mRNA transcription of multiple RTKs involved in resistance (ERBB3, DDR1, FGFR2, MET) and suppresses lapatinib-mediated induction. (E) 8-day growth curves demonstrate JQ1/lapatinib combination prevents growth of SKBR-3 and BT474 cells. Data presented is mean ± SD of six technical replicates. (F) JQ1 in combination with lapatinib suppresses colony formation of SKBR-3 and BT474 cells after 4-weeks. (G) BET family bromodomain inhibitors (JQ1, I-BET762, I-BET151) suppress colony formation of HCC1419 cells more so than kinase inhibitors in combination with lapatinib in 4-week colony formation assays. (H) Western blots indicate AKT inhibition (MK2206) induces RTK expression and ERK signaling alone and in combination with lapatinib. BET bromodomain inhibition alone does not sustain inhibition of signature kinases, and only in combination with lapatinib suppresses the adaptive response. 8-day treatment with 30nM lapatinib, 100nM MK2206, and 300nM JQ1. Data presented in D, F, and G is mean ± SD of three technical replicates. Also see Figures S5 and S6

Figure 6. JQ1 modulates lapatinib-induced transcription and inhibits epigenetic regulation of signature kinase genes

(A) RNAseq indicates JQ1 affects 11% of expressed genes 2-fold or more in SKBR-3 cells after 48 h treatment. (B) Figures 6C-E refer to JQ1 effect on lapatinib-regulated genes as indicated. (C) JQ1 downregulates 27% of the 1009 lapatinib-induced genes by at least 2-fold from the lapatinib-induced mRNA level. (D) 1000 genes not affected by lapatinib treatment display a similar up- or down-regulation profile in the lapatinib+JQ1 combination compared to JQ1 alone. (E) 964 genes at least 2-fold downregulated by lapatinib are mostly unaffected by JQ1 as compared to JQ1 alone or JQ1 effects on lapatinib-upregulated genes. (F) qRT-PCR demonstrates siRNA-mediated knockdown of BRD2 and BRD3 enhances transcription of ERBB2, ERBB3, FGFR2 and DDR1. Knockdown of BRD4 suppresses ERBB3 and DDR1 transcription, similar to JQ1. 24h siRNA knockdown, then 24h drug treatment; 300nM JQ1, 300nM lapatinib. Data presented is mean ± SD of three technical replicates. (G) ChIP-PCR indicates JQ1 inhibits BRD4 promoter occupation in the absence of lapatinib. Loss of BRD4 and elongating RNA Polymerase II (pS2-Pol2) from upstream elements is maximal when JQ1 is combined with lapatinib. 4h treatments with 300nM lapatinib and 300nM JQ1 in SKBR-3 cells. Data presented is mean of three biological replicate experiments. Also see Figure S7

Figure 7. BET bromodomain inhibition suppresses signature kinases and arrests growth in lapatinib-resistant cells

(A) RNAseq after 8-day treatment of lapatinib-resistant (LapR) SKBR-3 cells with lapatinib + 300nM JQ1 or 1µM I-BET151 indicates transcriptional suppression of the majority of tyrosine kinases. (B) mRNA fold changes in outlier kinases identified by PCA (Figure 2C) indicates BET inhibitors suppress the majority of kinases that drive variation in the kinome response. (C) MIB/MS analysis of the top 20 most-activated kinases in LapR SKBR-3 cells following 8 days treatment with 300nM JQ1 or 1µM I-BET151 indicates the majority of kinase activity is inhibited or blocked. (D) Western blots of LapR SKBR-3 and BT474 cells treated with 300nM JQ1 or 1µM I-BET151 in combination with 300nM lapatinib show suppression of signature kinase expression and phosphorylation. (E) 4-week colony formation assays demonstrate JQ1 suppresses colony formation and arrests growth of LapR SKBR-3 and BT474 cells in the presence of lapatinib. Data presented is mean ± SD of three technical replicates. (F) LapR BT474 cells are moderately growth-inhibited by combinations of lapatinib and other kinase inhibitors, but growth is completely suppressed by lapatinib and bromodomain inhibitors (300nM JQ1, 1µM I-BET762, or 1µM I-BET151, even more effectively than a triple kinase inhibitor combination (lapatinib+dasatinib+PF228). (G) Growth of LapR BT474 cells is arrested with 300nM JQ1 or 1µM I-BET762, but only in the presence of lapatinib. Data presented in F and G is mean of six technical replicates ± SD.