A fluorescence-based sensor for calibrated measurement of protein kinase stability in live cells

Abstract

Oncogenic mutations can destabilize signaling proteins, resulting in increased or unregulated activity. Thus, there is considerable interest in mapping the relationship between mutations and the stability of signaling proteins, to better understand the consequences of oncogenic mutations and potentially inform the development of new therapeutics. Here, we develop a tool to study protein-kinase stability in live mammalian cells and the effects of the HSP90 chaperone system on the stability of these kinases. We determine the expression levels of protein kinases by monitoring the fluorescence of fluorescent proteins fused to those kinases, normalized to that of co-expressed reference fluorescent proteins. We used this tool to study the dependence of Src- and Raf-family kinases on the HSP90 system. We demonstrate that this sensor reports on destabilization induced by oncogenic mutations in these kinases. We also show that Src-homology 2 and Src-homology 3 domains, which are required for autoinhibition of Src-family kinases, stabilize these kinase domains in the cell. Our expression-calibrated sensor enables the facile characterization of the effects of mutations and small-molecule drugs on protein-kinase stability.

Keywords: HSP90; cell signaling; fluorescent sensors; kinase inhibitors; kinases; oncogenes; protein stability.

FIGURE 1

An expression‐calibrated, ratiometric fluorescent protein sensor to monitor protein kinase abundance in live cells. (a) Schematic of the fates of an unstable kinase domain in the cell. Autoinhibitory domains (AD) often physically stabilize kinase domains by constraining their dynamics. HSP90 and CDC37 collaborate to stabilize a large proportion of kinase domains in the cell, which can be blocked by HSP90 inhibition by small molecules (HSP90i, HSP90 inhibitor). Unstable kinase domains are likely surveilled by ubiquitin ligases in the cell, leading to their proteolytic degradation by the 26S proteasome (26S) when ubiquitylated (U, ubiquitin moieties). (b) Schematic of the coding sequence of the lentiviral vector used in this study and the workflow for producing stable cell lines used in this study. A ribosome skipping sequence (tandem 2A, Td2A) allows for BFP (BFP) and GFP (GFP) to be expressed as separate proteins from the same transcript. The GFP sequence is appended to a linker and kinase/kinase domain (KD). The expression cassette is packaged into lentivirus and used to transduce immortalized RPE1 cells expressing human telomerase (hTERT). After transduction, polyclonal populations of cells expressing the sensor can be isolated by FACS. (c) Flow cytometry analysis of RPE1/hTERT cells expressing ARAF from the bicistronic sensor with and without the addition of an HSP90 inhibitor (HSP90i).

FIGURE 2

Analysis of HSP90‐dependent stability in Raf family kinases. (a) Schematic of Raf kinases tested in this experiment. Note the absence of N‐terminal regulatory domains in BRAF^∆N. (RBD, Ras binding domain; CRD, cysteine rich domain; P, phosphorylation sites). (b) A simplified model of BRAF activation in cells. Autoinhibited BRAF bound to MEK and 14‐3‐3 dimer is recruited to the membrane and activated by GTP‐loaded Ras GTPases, dislodging the RBD and CRD from the autoinhibited kinase to promote (homo/hetero)dimerization on a single 14‐3‐3 dimer. (c) Expression of BRAF^∆N in cells yields a highly active dimeric complex in vitro. It is unknown if this active version of BRAF is bound by HSP90/CDC37. (d) Plot of levels of Raf kinases and mutants with and without addition of an HSP90i (1 μM tanespimycin), as measured by flow cytometry. (n = 3 biological replicates). (e) As in 2E, a dose‐response plot of normalized levels of Raf kinases and variants (n = 3 biological replicates).

FIGURE 3

Src‐homology 2 and Src‐homology 3 domains stabilizes expression of the c‐Src kinase domain. (a) Schematic illustrating autoregulation by the Src module. Docking of the SH2 and SH3 domains to the kinase autoinhibit kinase activity in Src‐family kinases. (b) Expression of Src, is higher when expressed with corresponding SH3, SH2, and tail domains. (c) Plots showing both c‐Src^KD and c‐Src are modestly degraded when cells are treated with HSP90i (ganetespib). Levels of Src^KD significantly decreased compared to c‐Src in untreaded cells (p < 0.01, Mann–Whitney U test, n = 3).

FIGURE 4

Analysis of HSP90‐dependent stability in Src‐family kinases. (a) Flow cytometry analysis of c‐Src^KD and Lck^KD in the calibrated sensor in response to varying concentrations of HSP90i (tanespimycin). The outlined curve indicates data from untreated cells. (b) Plot of data collected from a similar experiment as in panel 3C demonstrating that Lck^KD is degraded to a much greater extent than c‐Src^KD in response to HSP90i (tranespimycin).

FIGURE 5

Activating mutations destabilize and a small molecule inhibitor stabilizes the c‐Src kinase domain. (a) Illustration of a crystal structure of the c‐Src kinase domain bound to nucleotide. The two residues mutated are shown in the insets. (b) Both the T341I and E381G activating mutations destabilize the c‐Src kinase domain as shown by decreased kinase domain levels when expressed from the calibrated sensor. (c) Treatment of c‐Src constructs in panel 5B with dasatinib stabilizes the c‐Src^E381G but not c‐Src^T341I or c‐Src^KD. (d) Dose‐response curve of c‐Src mutant levels as a function of dasatinib concentration. Error bars represent the standard error of the mean (n = 3). (e) Plot of replicate variant c‐Src^KD levels with and without 10 nM dasatinib. Groups were analyzed by one‐way Brown–Forsythe ANOVA with Dunnett's T3 multiple comparisons test to generate p‐values on chart.