TPCK inhibits AGC kinases by direct activation loop adduction at phenylalanine-directed cysteine residues

Abstract

N-alpha-tosyl-l-phenylalanyl chloromethyl ketone (TPCK) has anti-tumorigenic properties, but its direct cellular targets are unknown. Previously, we showed TPCK inhibited the PDKl-dependent AGC kinases RSK, Akt and S6K1 without inhibiting PKA, ERK1/2, PI3K, and PDK1 itself. Here we show TPCK-inhibition of the RSK-related kinases MSK1 and 2, which can be activated independently of PDK1. Mass spectrometry analysis of RSK1, Aktl, S6K1 and MSK1 immunopurified from TPCK-treated cells identified TPCK adducts on cysteines located in conserved activation loop Phenylalanine-Cysteine (Phe–Cys) motifs. Mutational analysis of the Phe–Cys residues conferred partial TPCK resistance. These studies elucidate a primary mechanism by which TPCK inhibits several AGC kinases, inviting consideration of TPCK-like compounds in chemotherapy given their potential for broad control of cellular growth, proliferation and survival.

1. Introduction

The ability of cells to perceive and appropriately respond to their microenvironments is crucial to essentially all cellular processes. Protein kinases are key modulators and drivers of these processes and therefore have become attractive targets for drug discovery. The functionally diverse but structurally conserved family of kinases termed the AGC kinases [1] (as they contain relatives of protein kinase A (PKA), protein kinase G (PKG) and protein kinase C (PKC)) includes isoforms of Akt (protein kinase B (PKB)) [2], p70 ribosomal S6 kinase (S6K) [3], p90 ribosomal S6 kinase (RSK) [4], mitogen- and stress-activated protein kinase (MSK) [5], and serum- and glucocorticoid-induced protein kinase (SGK) [6]. Despite divergent regulation, these kinases are all activated by the common kinase regulatory mechanism of phosphorylation at the autoinhibitory activation loop [7].

3-Phosphoinositide-dependent kinase 1 (PDK1) has been shown to phosphorylate the activation loops of several AGC kinase family members [1,8]. Additionally PDK1 phosphorylates its own activation loop [9]. Despite the emergence of PDK1 as a critical regulator of multiple AGC kinases with oncogenic potential, few studies have evaluated PDK1 as a potential target for cancer therapy. However, PDK1's value as a target for anti-cancer therapy is certainly under consideration [10–12]. Previously we reported that TPCK disrupts PDK1 signaling to Akt, S6K1, and RSK [13]. However, the inhibitory mechanism was unclear given TPCK did not directly affect the phospshotransferase activity of PDK1. Here we investigated the inhibitory effects of TPCK toward the non-PDK1 dependent kinases MSK1 and MSK2 and found that they were TPCK sensitive. We next used mass spectrometry to analyze AGC kinases immunoprecipitated from TPCK-treated cells and found that TPCK-inhibited kinases showed direct TPCK adduction at phenylalanine-directed cysteine (Phe–Cys) residues in their activation loops. Mutation of the conserved cysteine or phenylalanine led to the generation of kinases partially TPCK-resistant. This study thus describes a primary mechanism by which TPCK inhibits specific AGC kinases and describes novel TPCK resistant AGC kinase alleles. Furthermore we discuss the potential productivity of developing inhibitors such as TPCK that could simultaneously target conserved Phe–Cys motifs in several AGC kinases. Such a drug could dampen multiple signaling arms of a cell in a state of hyperactivity and thereby reduce the chances of such cells developing resistance, a major thorn in the side of targeted cancer therapies in our emerging era of personalized medicine.

2. Materials and methods

2.1. Plasmids

The plasmids encoding triple hemagglutinin (HA₃)-tagged RSK1 [14] and HA-S6K1 [15] have been described. Bacterial expression plasmids encoding GST-S6 (rat) Lys218-Lys249 [16], GST-BAD [16], GST-RSK1-D1 K/R have been described. Myc-tagged PDK1 in pCDNA3 was a gift of P. Hawkins. HA-Akt in pCMV6 was a gift of P. Tsichlis. Flag-tagged MSK1 was a gift of R. Janknecht. The cloning of HA₃-MSK2 was as follows: EST clone AA576979 from Genome Systems was used to generate a ³²P-labeled probe that was used to screen 1.7 × 10⁶ plaques from a human T-cell cDNA library in Lambda ZAPII (Stratagene) at 45,000 plaques per 15 cm dish. The primers used to generate the probe were 5′-GCTCAGAGCTG-GATGTGG-3′ (sense) and 5′-TCGGCGTACAGGATGTTC-3′ (anti-sense). Twelve positive clones were ultimately identified and the longest clone contained the start codon and extended ∼1,9000 bp and included an internal NotI site. The 3′-end of MSK2 from the internal NotI site and extending past the stop codon ∼600 bp was obtained from Genome Systems EST clone AI831613. The full length sequence was assembled first in pBluescript SK-. The 5′-untranslated region was removed while inserting a BglII site at the 5′-end using PCR amplification from the initiating codon and past an internal EcoRI site. The PCR primers were 5′-AGAGATCTATGGGGGAC-GAGGACGAC-3′ (sense) and 5′- GGAATTCTTAGGAGGGGGGCAGGG GGCCGTT-3′ (anti-sense). The 5′-BglII, 3′-EcoRI fragment was subcloned into the BamHI, EcoRI sites of pKH3 in frame with the triple HA-tag (at the 5′-end). The remainder of MSK2 (from the internal EcoRI site, past the stop codon and including the 3′ untranslated region to an EcoRI site in pBluescript) was subcloned into the pKH3 that already contained the BglII-EcoRI fragment of MSK2 described above, using the EcoRI fragment excised from pBluescript-MSK2. The insert was sequenced and we found all nucleotides to be identical to those published under GenBank accession AJ010119 except for a T2237C (numbering is from AJ010119) substitution that did not change the amino acid sequence.

2.2. Mammalian cell culture, transfection and lysis

E1 A-transformed Human embryonic kidney (HEK 293E) cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% FBS, penicillin and streptomycin. HEK293E cells were transfected using calcium phosphate precipitation. Twenty four hours post-transfection, cells were starved in DMEM containing 20 mM HEPES for 16–18 h prior to stimulation and as indicated in the figure legends. After stimulation, cells were washed twice with cold phosphate-buffered saline (PBS) and then lysed in 10 mM KPO4, 1 mM EDTA, 10 mM MgCl₂, 50 mM β-glycerophosphate, 5 mM EGTA, 0.5% Nonidet P-40, 0.1% Brij 35, 0.1% deoxycholic acid, 1 mM sodium orthovanadate, 1mM phenylmethylsulfonyl fluoride, 5 μg/ml leupeptin, 5 μg/ml of pepstatin A. The cell extracts generated were centrifuged at 14,000 rpm for 10 min to remove cell debris and the clarified supernatant was used for immunoprecipitations, immunoblotting or protein kinase assays.

2.3. Immunoprecipitations, kinase assays and immunoblotting

Cell lysates were incubated with endogenous or anti-epitope tag antibodies for 2 h and then with 20 μl of a 1:1 mixture of protein G-sepharose and protein A-sepharose beads for an additional hour at 4 °C. Beads were washed three times with lysis buffer and the immunoprecipitates were used for kinase assays or immunoblotting. Kinase assays for RSK1-4, MSK1-2 and S6K1 were performed using GST-S6 as a substrate. Kinase assays for Akt used GST-BAD as a substrate. Kinase assays for PDK1 used RSK-D2 K/R as a substrate. Kinase assays were allowed to proceed for 10 min. at 30 °C prior to stopping the reaction with sample buffer as described previously [13]. The reaction products were subjected to SDS–PAGE and dried gels were exposed to X-ray film. For immunoblots, immune complexes or cell lysates were subjected to SDS–PAGE and transferred to nitrocellulose membranes. Blocking, primary and secondary antibody incubations of immunoblots were performed in TBST (10 mM Tris, pH 7.4, 150 mM NaCl and 0.1% Tween 20) supplemented with 5% (wt/vol) dry skim milk powder. Horseradish peroxidase-conjugated donkey anti-rabbit and anti-mouse antibodies were used to facilitate detection by enhanced chemiluminescence (ECL) and exposure to X-ray film.

2.4. Mass spectrometry

In-gel tryptic digests were performed as described previously [17] and extracted peptides were subjected to LC-MS/MS in a linear ion trap-FT (LTQ-FT) hybrid mass spectrometer (Thermo Electron, San Jose, CA) set up as described previously [18]. Tandem mass spectra were searched using SEQUEST against individual kinase amino acid sequences requiring no enzyme specificity, a 20 ppm precursor mass tolerance and allowing for the following differential mass modifications: methionine oxidation (+15.9949); phosphorylation on serine, threonine and tyrosine (+79.9663); acrylamide adduction on cysteine (+71.0371); and TPCK adduction (C₁₇H₁₇ NO₃S) on cysteine (+315.0929). The precursor mass spectra for all peptides of Akt and S6K1 collected in the FT were observed to be ∼+8.5 ppm off theoretical values and thus were appropriately adjusted.

3. Results

3.1. TPCK inhibits PDK1-dependent and PDK1-independent AGC kinases

We demonstrated previously that TPCK inhibited the activation of the PDK1 targets RSK, Akt and S6K1 without affecting the activity of PI3K, PDK1, PKA and ERK1/2 [13]. While the requirement for PDK1 in PKA activation has been controversial, the activation of the RSK-relatives MSK1 and MSK2 have been shown to be PDK1-independent [1]. We therefore tested the effect of TPCK on the activation of MSK1 and MSK2 to determine if TPCK's effect on AGC kinases might be PDK1-dependent. In Fig. 1 we first show our evaluation of the effect of TPCK on the activation of RSK, Akt, S6K and MSK1. Serum-starved HEK 293 cells were pre-treated with TPCK before stimulation with either EGF or insulin. The endogenous AGC kinases were then immunoprecipitated and their phosphotransferase activities were analyzed by kinase assays using purified substrates. TPCK robustly inhibited the activation of RSK, S6K, Akt and MSK1 as shown by their inability to phosphorylate their recombinant substrates (Fig. 1A). We also analyzed other isoforms of RSK (2–4) and TPCK was effective in inhibiting the kinase activity of all of them including the phosphorylation of their N-terminal activation loops, which are known to be phosphorylated by PDK1 (data not shown). The effect of TPCK on the activation of MSK2 using an HA-tagged MSK2 construct found MSK2 to be TPCK-sensitive (Fig. 1B). We also confirmed that PDK1 itself was TPCK-resistant (Fig. 1C).

Fig. 1.

TPCK inhibits PDK1-dependent and PDK1-independent AGC kinases (A) TPCK inhibits the PDK1-dependent kinases RSK, Akt and S6K1 (A and B) as well as the non-PDK1-dependent AGC kinase MSK1 and MSK2 (C) without inhibiting PDK1 kinase activity. HEK293E cells were serum-starved for 18 h, and treated with either EGF (50ng/ml) for 10 min. or Insulin (100 nM) for 30 min. (as indicated) with or without pretreatment of cells with TPCK for 30 min. as indicated. Endogenous kinases (RSK1, MSK1, Akt1 and S6K1) or exogenous kinases (HA-MSK2 and Myc-PDK1) were immunoprecipitated with their respective antibodies. These immunoprecipitates were incubated with either GST-S6, GST-Bad or GST-RSK-D2 K/R (as indicated) in a kinase reaction containing [γ – ³²P] ATP for 10 min. The samples were subjected to SDS–PAGE, and the gel was autoradiographed. The lysates were immunoblotted for pAkt, pS6, ERK1/2, HA or Myc as indicated.

3.2. TPCK modifies a conserved phenylalanine-directed cysteine residue in the activation loop of AGC kinases

The inhibition of PDK1 signaling to its downstream targets in the presence of TPCK without affecting the activity of PDK1 was intriguing. We envisaged that the TPCK-mediated inhibition of AGC kinases may be due to direct adduction of TPCK to structurally vulnerable AGC family members. To test this we immunopurified transfected RSK1 from TPCK-treated HEK293 cells. The immunopurified RSK1 was subjected to SDS-PAGE, in-gel digestion and mass spectrometry analysis (Fig. 2A). Using a high resolution mass spectrometer we detected a TPCK adduct at Cys241 of RSK1 (Fig. 2A). This result prompted us to investigate if TPCK formed adducts on other TPCK-sensitive AGC kinases. When analyzed by mass spectrometry we detected TPCK adduct formation on AKT, S6K1 and MSK1. The adduct formation occurred at Cys310 of Akt (Fig. 2B), Cys231 of S6K1 (Fig. 2C) and Cys241 of MSK1 (data not shown). This is the same conserved activation loop cysteine which was found modified in RSK1.

Fig. 2.

TPCK adducts directly to RSK1, Akt1 and S6K1 at phenylalanine-directed cysteine residues in their respective activation loops. (A) TPCK directly adducts to Cys241 of RSK 1. HEK293E cells were transfected with HA-RSK1 as indicated. Cells were starved (18 h) and stimulated with EGF (50 ng/ml) for 10 min, with or without pretreatment of cells with 50 μM of TPCK for 30 min. RSK1 in the lysate was immunoprecipitated with α-HA antibodies and the immune complex was subjected to SDS–PAGE. The gel band corresponding to RSK1 from cells pre-treated with TPCK and stimulated with EGF (inset, upper panel) was excised, digested with trypsin and analyzed by liquid-chromatography tandem MS (LC-MS/MS). A part of the lysate was analyzed for RSK1 and ERK phosphorylation (inset, lower panels). Shown is a low energy collision-induced dissociation (CID) tandem mass spectrum of a tryptic peptide from the activation loop of RSK 1 with a TPCK adduct on Cys241 indicated by an asterisk. Also indicated are the observed and calculated masses for the triply-charged ion species. The mass difference is indicated in parts per million (ppm) mass/charge (m/z). “#” indicates oxidation of methionine. (B) Cells were transfected with HA-Akt1, starved (18 h) and stimulated with Insulin (100 nM) for 30 min after pretreatment of cells with 50 μM TPCK for 30 min. Anti-HA immune complexes were treated as in A. Shown is a low energy CID tandem mass spectrum of a tryptic peptide from the activation loop of Akt1 with a TPCK adduct on Cys310 indicated by an asterisk. (C) Cells were transfected with HA-S6K1, starved (18 h) and stimulated with Insulin (100 nM) for 30 min after pretreatment of cells with 50 μM TPCK for 30 min. Anti-HA immune complexes were treated as in A. Shown is a low energy CID tandem mass spectrum of a tryptic peptide from the activation loop of S6K1 with a TPCK adduct on Cys231 indicated by an asterisk.

3.3. Partial TPCK resistance is observed in RSK1 and MSK1 following mutation of activation of phenylalanine or cysteine residues

In order to test that the TPCK-mediated inhibition of the AGC kinases is due to the modification of the conserved cysteine residues, we introduced either alanine- or serine-encoding missense mutations at Cys241 of RSK1. We also generated a RSK1 construct encoding a phenylalanine to leucine mutation at Phe240. Each of these mutations led to partial TPCK resistance with the Cys241 Ala exhibiting the most resistance (Fig. 3A). Furthermore, a MSK1 Cys241Ala mutant exhibited almost complete resistance to the inhibitory effect of TPCK (Fig. 3B). The kinases Akt and S6K1, on the other hand, become catalytically inactive upon mutation of the activation loop cysteine residue (data not shown), making it impossible to test the extent of TPCK inhibition on these mutant kinases. Nevertheless, together these data suggest that TPCK adduction to phenylalanine-directed cysteine residues in the activation loops of most AGC kinases is the primary means of their inhibition by TPCK. The proposed reaction mechanism of TPCK adduction to cysteine is presented in Fig. 3C and is consistent with the adduct we identified using high resolution mass spectrometry.

Fig. 3.

Partial TPCK resistance is observed in RSK1 and MSK1 following mutation of phenylalanine or cysteine residues in the activation loop. (A) HEK293E cells were transfected with either wild-type (WT) or C241A, C241S or F240L mutants of RSK1. The cells were serum starved (18 h), and stimulated with EGF (50 ng/ml) for 10 min, with or without pretreatment with 50 μM of TPCK for 30 min. HA-RSK1 constructs were immunoprecipitated and phosphotransferase activity of wildtype and mutant HA-RSK1 kinases were analyzed in kinase assays using GST-S6 as a substrate. The lysates were immunoblotted for RSK1 and ERK1/2. (B) Cells were treated as in A except that transfections were with Flag-tagged wildtype or Cys214Ala MSK1, and immunoprecipitations used anti-Flag resin. (C) Schematic representation showing the reaction mechanism leading to TPCK adduction.

4. Discussion

Approaches toward personalized medicine are becoming increasingly attractive and variably pursued. However, in challenges such as cancers, even exquisitely personalized therapies can suffer setbacks due to resistance. A striking example of this is the development and administration of the active BRaf inhibitor, PLX4032, which has high selectivity and therapeutic potential against tumors with the specific BRaf Val600Glu mutation [19]. While PLX4032 has shown dramatic BRafV600E-positive tumor reduction in only a very few weeks [19], cancer resistance to the drug is the dominant outcome [20]. Generally, when resistance emerges a common counter measure is a treatment regime using cocktails of inhibitors, thereby targeting multiple steps in the hyperactive pathway [21].

PDK1 lies at the apex of several arms of tumorigenic pathways given its critical role in the activation of RSK, Akt, PKC, SGK and S6K. These pathways are significantly intertwined and cross-talk and compensatory mechanisms are now beginning to be understood [22]. Drugs that could simultaneous dismantle these pathways theoretically could have significant advantages. One type of cancer resistance can be the upregulation of a parallel pathway to overcome the inhibition of the targeted pathway. This can easily be understood in terms of the difference between effects due to endogenous or overexpressed kinases. For example, when determining if a particular kinase is responsible for the phosphorylation of a given target three important tools are commonly used: overexpression, RNAi-mediated silencing and pharmacological inhibition. This is because if the overexpressed kinase causes the substrate in question to be phosphorylated this is useful in that it denotes that the kinase can lead to that substrate's phosphorylation, but the invariable and important next question is whether or not that effect is an “artifact” of overexpression, meaning a “normal” cell may not behave that way. Understanding normal cell biology is of great importance but cancer cells are not “normal” cells and obviously need to be treated for what they have become and not for what they once were. Thus, if a cancer cell is being driven by hyperactive PI3K-Akt signaling and a PI3K inhibitor successfully dampens Akt activity, the cancer might, for example, evolve to acquire hyperactive RSK activity and now RSK begins to “abnormally” phosphorylate Akt substrates as compensatory mechanism directed at maintaining the survival and proliferative capacity of the tumor cell [22]. In spite of the discussion of potential side-effects, the recent development of PDK1-specific inhibitors [23–25] is of great interest and have been the subject of several interesting articles [10–12].

We have shown that TPCK adducts to and inhibits the activation of PDK1-dependent (as well as PDK1-independent) AGC kinases with Phe–Cys motifs in their activation loops. An exhaustive search of the human kinome [26] for Phe–Cys-containing activation loops identified eight groups within the AGC kinase family (Akt, SGK, PKN, RSK, MSK, S6K, PKG, and PKC) and four groups within the CamK kinase family (MARK, NIM, NuaK, TSSK and PASK) that harbor these motifs and thus are likely sensitive to TPCK (Fig. 4). While inhibition of this set of kinases would certainly have pleiotropic effects, early studies in mice prone to breast cancer exhibited reduced tumors and no toxicity when given 1 mg per week of TPCK for over forty weeks [27]. Furthermore, in the DMBA-insulted and phorbol ester-promoted tumor model, TPCK greatly inhibited tumor formation when applications were applied to the skin [27]. Thus, in spite of, or perhaps thanks to, TPCK's inhibition of multiple kinases it has shown anti-tumorigenic properties. Given TPCK's ability to target multiple AGC kinases it may be less easily overcome by cancer resistance, particularly if used as part of a chemo-therapeutic cocktail.

Fig. 4.

Multiple sequence alignment of all human kinases with Phe–Cys motifs in their activation loops. The black bar denotes the Phe–Cys motif with the black arrow denoting the site of TPCK adduction to cysteine residues. The Gray arrow indicates the site of activation loop phosphorylation. The asterisk indicates amino acid identity, while “.” and “:” indicate increasing levels of amino acid conservation.

Finally, the generation of TPCK-resistant alleles of AGC kinases may provide mechanisms to better decipher AGC kinase-dependent signaling pathways in normal cells. Additionally, kinases could be generated to become TPCK sensitive and thereby provide new tools for signal transduction studies. In conclusion, having identified the primary mechanism by which TPCK inhibits AGC kinases enlightens both chemotherapeutic approaches as well as the use of TPCK as a tool to understand signal transduction mechanisms.

Abbreviations

TPCK

N-alpha-tosyl-l-phenylalanyl chloromethyl ketone

PDK1

3-phosphoinositide dependent protein kinase-1

AGC

related to protein kinase A, protein kinase G and protein kinase C

RSK

ribosomal S6 kinase

S6K

S6 kinase

PKA

protein kinase A

ERK

extracellular signal regulated kinase

PI3K

phosphatidylinositol-3 kinase

MSK

mitogen and stress-activated protein kinase