Translating Biochemical STK-Assays into Cells
Cisbio’s KinEASE platform utilizes a minimal reagent set of three promiscuous peptide substrates (STK-peptides) sharing a common Ser-phosphorylation site and one corresponding phospho-specific antibody (pSTK) to support a panel of >100 biochemical Ser/Thr-kinase assays (Fig.
]). In vitro,
the incubation of purified kinases with biotinylated peptide substrates leads to substrate phosphorylation quantifiable by TR-FRET [28
]. We sought to translate this biochemical assay into a cellular format by first coexpressing Ser/Thr-kinases with reengineered cellular STK-peptide substrates –subsequently referred to as cSTK-pep– and then quantifying intracellular peptide phosphorylation using a simple homogeneous addition-only protocol.
Fig. (1) Schematic of biochemical and cellular TR-FRET Ser/Thr-kinase assays. (A) Biochemical KinEASE assay. (B) and (C) Cellular Ser/Thr-kinase assays. Target kinases are overexpressed in combination with cSTK peptide substrates (B) or full-length protein substrates (more ...)
To this end, we flanked the STK peptide sequences with N-terminal Myc- and C-terminal Flag-epitopes. The resulting oligopeptides were fused to hAGT carrier proteins (“SNAP-tag”) [22
] to support their stable cellular expression and to have an option for a future direct functionalization of intracellular cSTKpep-1, 2 and 3 substrates (Fig.
). The added epitope tags enable the quantification of either total substrate expression or substrate phosphorylation. To have a reference for peptide based cellular Ser/Thr-kinase assays we developed MEK1 as a model full-length protein substrate of kinases such as B-Raf or B-Raf(V600E), an activated oncogenic kinase mutant and established cancer drug target [31
Case Study 1: B-Raf – Peptide vs Protein Substrates and Pharmacological Assay Validation
When the cSTKpep-3 substrate was expressed alone in HEK293 cells the observed basal phosphorylation levels derived from endogenous Ser/Thr-kinase activities were low, barely detectable by TR-FRET with weak phospho-specific bands being detectable by western blotting (Fig. , and ). Similar observations were made with peptide substrates cSTKpep-1 and cSTKpep-2 (Fig. and data not shown).
Fig. (8). Induction of CamK2δ activity. Transient transfection reactions were performed in 384-well plates as indicated. HEK 293 were reverse transfected with 50ng CamK2δ and 50ng (black bars) or 25ng (gray bars) cSTKpep-1. CamK2δ activity was induced through 5min ionomycin (more ...)
Cotransfection of cSTKpep-3 with B-Raf(V600E) resulted in a several fold increase in substrate phosphorylation by both western blotting and TR-FRET (Figs. , and ). Moreover, B-Raf(V600E)-induced cSTKpep-3 phosphorylation was both dependent on the kinase activity as well as the presence of the target phosphorylation site since no cSTKpep-3 phosphorylation was detected when using a kinase-inactive mutant version of B-Raf(V600E), B-Raf(V600E)-KD or a Ser/Ala-mutant of cSTKpep-3 (cSTK pep-3-SA) (Fig. and ).
As expected, cSTKpep-3 phosphorylation could be modulated by both the amounts of substrate and kinase plasmids transfected as well as by the activation state of the transfected kinase (Fig. ). Comparable results were obtained when using kinase-inactive, MEK1(KD), as a substrate (Figs. and ); cotransfection of MEK1 with B-Raf(V600E) resulted in increased MEK1 phosphorylation as detected by both western blotting and TR-FRET.
These results demonstrate that promiscuous Ser/Thr-kinase peptide substrates can be expressed in cells with low detectable levels of background phosphorylation and that significant increases in peptide substrate phosphorylation can be achieved through the overexpression of a target Ser/Thr-kinase. Since both the degree of background phosphorylation as well as kinase activities depend on the amounts of expression constructs applied in the transient transfection reactions, careful titrations of both constructs need to be performed to obtain optimal assay windows.
For the pharmacological evaluation we generated double stable cells constitutively expressing either cSTKpep-3, wild-type MEK1(WT) or kinase-inactive MEK1(KD) substrates in combination with Tet-inducible B-Raf(V600E). Kinase inhibitors were then tested in 384-well or 1536-well formats. In all cases short compound treatment times of around 2h were employed to minimize cytotoxicity related compound interference (Figs. and , Table ).
Fig. (3) Pharmacological validation of Raf(V600E)/cSTKpep-3 and Raf(V600E)/MEK(WT) TR-FRET assays. IC50 curves were determined for B-Raf inhibitor A in 384-well and 1536-well plates using double stable cell lines. Open circles: DMSO, solid squares: B-Raf inhibitor (more ...)
IC50 values of B-Raf inhibitors are strongly correlated between 384-well B-Raf(V600E) cSTKpep-3 and MEK1(WT) TR-FRET and Ba/F3 proliferation assays (Table ).
IC50 Profiles of B-Raf(V600E) TR-FRET and Ba/F3 Assays. Standard deviations for data are listed in parentheses (n=4-6).
Overall, cellular TR-FRET assays performed very robustly with B-Raf inhibitor A in both 384-well and 1536-well plates for MEK(WT), MEK(KD) or c-STKpep-3 substrates with Z’=0.6-0.9. Moreover, the observed potencies of a panel of selective in-house B-Raf inhibitors correlated very well in cSTKpep-3 and MEK1(WT) TR-FRET-assays (R2=0.86, Table , Fig. ). Good agreement was also observed when comparing TR-FRET assay results with historical Ba/F3-proliferation data for B-Raf(V600E) (R2cSTKpep-3/BaF3=0.81, R2MEK1/BaF3= 0.94, Table , Fig. and ). None of these B-Raf inhibitors affected basal cSTKpep-3 and MEK1(WT) background phosphorylation in the absence of prior B-Raf(V600E) induction (data not shown). Finally, a selection of unrelated kinase inhibitors, including Wortmannin and LY29002 did not show any activity in either the cSTKpep-3 or MEK1(WT) TR-FRET assays (Table ).
An Emerging Cellular Ser/Thr-Kinase Assay Panel. Confirmed hits obtained from cSTKpep cDNA screens. fold activation was calculated as TR-FRET signal with kinase and cSTKpep cotransfection divided by cSTKpep alone.
While B-Raf selective compounds compared very well in the above assays (Table
), pronounced discrepancies were reproducibly observed for the activities of the highly selective MEK1/2 inhibitor PD0325901 [34
] and staurosporine, a non-selective kinase inhibitor with only weak activity against Raf-kinases in vitro
. Both compounds had comparable activities when using cSTKpep-3 or MEK1(WT) as B-Raf(V600E) substrates, but were inactive when using MEK1(KD). Similar observations of substrate dependent inhibitor profiles in cellular MEK-phosphorylation assays were recently made using an ELISA readout [36
]. The underlying reasons for these differences are still being investigated. One possible explanation is the existence of a positive feedback loop driving ERK1/2 mediated (hyper)activation of B-Raf(V600E), although the detailed mechanism is still unclear and factors including the cellular context and predominant Raf isoforms involved may have an impact [29
]. According to this model, PD0325901 and staurosporine indirectly affect B-Raf(V600E) activity on cSTKpep-3 and MEK1(WT) by targeting downstream kinases and thereby antagonizing the activating feedback loop. Overexpression of MEK1(KD) on the other hand would exert the same effect as small molecule MEK1/2 inhibition, thereby rendering the residual B-Raf(V600E) activity insensitive to further PD0325901 inhibition. Regardless of the underlying mechanistic reasons for the observed discrepancies between B-Raf targeted TR-FRET assays, it is interesting to note the distinctions between the MEK1(WT) and MEK1(KD) assays. The nature and activation state of functional protein substrates can impact the cellular activities of upstream drug targets and the activity profiles of putative inhibitors. This is a potentially complicating factor that needs to be investigated when designing cellular enzyme assays around bioactive substrates [15
]. Biologically inactive peptide substrates on the other hand would be expected to have a comparatively minor effect on intracellular signaling pathways.
Case Study 2: Interrogating the AKT Pathway and Probing for Compound Interference
For a second case study we turned to AKT and its upstream activation pathway. AKT-dependent intracellular cSTKpep-3 phosphorylation could be detected by western blotting and by TR-FRET (Fig. and ). In the absence of other efficacious AKT inhibitors staurosporine was used as a tool compound and found to effectively inhibit AKT-dependent cSTKpep-3 phosphorylation (Fig. ). Its cellular potency (IC50 ~100nM) agreed with the in vitro activity against AKT (IC50~100nM). Comparable IC50 values were found for incubation periods ranging from 30min (IC50~240nM) to 2h (IC50~100nM) (data not shown).
To determine if staurosporine-induced cytotoxicity and the subsequent loss of kinase substrate expression affected the outcome of the cellular assays, we aimed to separate the effect of staurosporine on cSTKpep-3 phosphorylation from its effect on total cSTKpep-3 expression (Fig. ). After short 2h treatment staurosporine exclusively affected cSTKpep-3 phosphorylation, but not cSTKpep-3 expression levels. After an extended 48h treatment stauros-porine affected phospho- as well as total-cSTKpep-3 levels with equal apparent potency, suggesting that the occurrence of cytotoxicity inappropriately interfered with the assay readout.
These observations underline the importance of short compound treatment times to avoid high rates of uninteresting hits in an HTS context and further highlight the benefit of the presented cellular assay format over cell proliferation or reporter gene assays that typically require compound incubation times from several hours to days.
Next, we aimed to interrogate the AKT pathway in greater detail by separating AKT activation as indicated by Ser473 phosphorylation, from signaling events depending on activated AKT such as cSTKpep-3 phosphorylation (Fig. ). To this end we developed another TR-FRET assay using His-tagged AKT to monitor Ser473 phosphorylation.
Fig. (7). Targeting the AKT-pathway. (A) TR-FRET assays were developed for the quantification of either p-AKT(Ser473) or p-cSTKpep-3. (B) p-AKT(Ser473) and (C) p-cSTKpep phosphorylation was quantified after treatment with Staurosporine (open triangles), Wortmannin (open squares), LY294002 (more ...)
As predicted for inhibitors of AKT activation, the PI3K inhibitors LY294002 and Wortmannin displayed comparable potencies in both AKT phosporylation (Fig.
) and cSTKpep-3 phosphorylation (Fig.
) assays. Moreover, the observed potencies of Wortmannin and LY294002 in the low nanomolar and low micromolar concentration ranges respectively were in line with their reported in vitro
and cellular activities on PI3K [37
]. Staurosporine activity on the other hand differed by a factor of ~100fold when quantifying p-AKT (IC50
~6μM) versus p-cSTKpep-3 (IC50
~60nM). These cellular activities mirror the in vitro
potency of staurosporine on PI3K (IC50
~9μM) and AKT (IC50
Together, these experiments demonstrate that the activity of overexpressed wild-type Ser/Thr-kinases can be responsive to upstream signaling events and that assays can be designed in the proposed format to interrogate a cellular pathway at different intersection points.
Seeding a Cellular Ser/Thr-Kinase Assay Panel
Having shown proof-of-concept for two important drug targets AKT and B-Raf, we sought to identify additional Ser/Thr-kinases amenable to our cellular assay approach through cDNA screening. A kinase focused sublibrary of GNF’s genome scale cDNA collection [25
] comprising approximately 800 Ser/Thr- and Tyr-kinase clones was arrayed into microtiter plates. A panel of cSTKpep-1, 2 and 3 substrates was then screened besides a number of similarly designed SNAP-peptide constructs containing alternative substrate sequences, including Crosstide (GRPRTSSFAEPG), MBPtide (FFKKIVTPRTPPP), CREBtide (LRREIL SRRPSYRK) and the p38α-activation loop (LARHTDDEMTGYVA). cDNA hits resulting in >2fold S/N ratios relative to substrate-alone controls were picked for further activity and sequence confirmation. Between all 7 screens the three cSTKpep substrates combined scored ~86% of all hits (57/66) and overall higher S/N ratios than the other peptide substrates tested. cSTKpep-3 returned the majority of hits (37/57) followed by cSTKpep-2 (12/57) and c-Kinase-1 (8/57). Ultimately, 22 distinct Ser/Thr-kinases were identified as inducers of cSTKpep phosphorylation (Table
). Of these 22 Ser/Thr-kinases, 13 were already known members of the biochemical KinEASE-panel, whereas 9 kinases were newly identified. Nine out of the 13 previously known kinases preferred the same STK peptide substrate in vitro
and in cells (Table
). No Tyr-kinases were found to stimulate cSTKpep phosphorylation, suggesting that no indirect Tyr-kinase induced activation of endogenous Ser/Thr-kinase pathways triggered cSTKpep phosphorylation. Overall, these observations are consistent with the idea that the intracellular kinase-cSTKpep interactions were direct.
Future Outlook and Opportunities
While a number of kinases already represented on the biochemical KinEASE panel were picked up through DNA screening, others were missed. Several reasons might help explain these findings. Preactivation of Ser/Thr-kinases either through physiologic stimuli or via protein engineering, e.g. through the deletion of autoinhibitory domains or activating mutations, may be required beyond their plain overexpression to induce cSTKpep phosphorylation. The subcellular localization of kinases or their substrates, cell cycle dependency, cellular backgrounds and the dynamics of signaling pathways all may play a role. By rescreening our kinase cDNA library under modified assay conditions, e.g. in the presence of external stimuli, nuclear targeted substrates, different cell types, or alternatively through the design of an entirely new cDNA library of activated kinase constructs we aim to further expand the current panel of cellular Ser/Thr-kinase assays in the future. In the case of CamK2δ we found that transient activation of kinase activity and subsequent cSTKpep-1 phosphorylation was ionomycin inducible (Fig. ). We believe that assay conditions can be found to activate additional Ser/Thr-kinases in a similar manner.