Reconstitution of Modular PDK1 Functions on Trans-Splicing of the Regulatory PH and Catalytic Kinase Domains

Abstract

The serine-threonine protein kinases PDK1 and PKB each contain a pleckstrin homology (PH) domain that binds the membrane-bound phosphatidylinositol 3,4,5-triphosphate [PI(3,4,5)P₃] second messenger and is required for PDK1-catalyzed phosphorylation and activation of PKB. While X-ray structures have been reported for the individual regulatory PH and catalytic kinase domain constructs of both PDK1 and PKB, diffraction quality crystals of full length constructs have yet to be obtained, likely due to conformational heterogeneity. In developing alternative approaches to understanding the potential role of conformational dynamics in regulating PKB phosphorylation by PDK1, an efficient in vitro method for protein trans-splicing was developed, which utilizes the N- and C-terminal split inteins of the gene dnaE from Nostoc punctiforme [(N)NpuDnaE] and Synechocystis sp. strain PCC6803 [(C)SspDnaE], respectively. For conjugating the regulatory PH domain to the catalytic kinase domain of PDK1, the recombinant trans-splicing fusion constructs KINASE(AEY)-(N)NpuDnaE-His₆ and GST-His₆-(C)SspDnaE-(CMN)PH were designed, PCR assembled, overexpressed, and affinity purified. The cross-reacting (N)NpuDnaE and (C)SspDnaE inteins generated full length spliced-PDK1 with k_obs = (2.8 ± 0.3) × 10⁻⁵ s⁻¹ and with ≤5% of any competing trans-cleavage reactions. Spliced-PDK1 was efficiently purified to ≥95% homogeneity from the reaction mixture by subsequent His₆ affinity and ion exchange chromatography steps. In vitro kinase assays and phosphopeptide mapping studies confirmed that spliced-PDK1 retained the ability to co-localize and selectively phosphorylate Thr-309 of PKBβ in a PI(3,4,5)P₃-dependent manner. The high-level production and reconstitution of functional spliced-PDK1 establishes the feasibility of incorporating domain-specific biophysical probes for spectroscopic studies of regulatory PH domain mediated catalytic specificity.

INTRODUCTION

Protein kinases comprise the largest enzyme family, with ∼518 being encoded by the human genome (1). The large number of serine-threonine protein kinases (∼80% of total protein kinases) reflects the large number of intracellular signal transduction pathways required in regulating proper cellular growth, survival, and proliferation. The complexity in elucidating intracellular signaling networks is compounded by analyses indicating that individual protein kinases may target and phosphorylate numerous different protein substrates (2, 3). Further complexities arise in the observation that individual protein kinases can be intricately regulated by multiple phosphorylation events catalyzed by numerous upstream kinases, as well as auto-catalytically. On review of protein kinase sequence alignments, it becomes readily apparent that signaling complexity can be largely accounted for by a distinguishingly diverse arrangement of regulatory domains, which may exist contiguously N- or C-terminal to an otherwise structurally homologous catalytic kinase domain (4). Often such domains may serve in proper localization, but many have been shown to play direct roles in regulating kinase activity either intrinsically or through participation in docking-based substrate recognition.

While the functions of various regulatory domains of different protein kinases are steadily being recognized, progress towards molecular understanding of conformational interactions within full length multi-domain kinases remains hampered. To date, no three-dimensional structure has been reported for any full length monomeric multi-domain serinethreonine protein kinase. Rather for example, X-ray structures have been reported for the isolated catalytic kinase domains of multi-domain AGC kinases including PDK1¹ (5), PKBβ (6, 7), PKCθ (8), and PKCι (9); multi-domain MAP kinases including B-RAF (10), MEK1,2 (11), MK2 (12), and MSK1 (13); and multi-domain cell-cycle related kinases including CHK1 (14), Aurora A (15), and Aurora B (16). Likewise, X-ray or NMR structures have been reported for corresponding isolated regulatory domain constructs including the PH domains of PDK1 (17), PKBα (18), and PKBβ (19); the C1 domain of PKCγ (20); the C2 domains of PKCα (21, 22), PKCβ (23), PKCδ (24, 25), and PKCε (26); the CR1 domain of RAF1 (27); the FHA domain of CHK2 (28); and the PBD domain of PLK1 (29, 30). The observation that suitable crystals for X-ray diffraction studies of full length multi-domain monomeric protein kinases have not been obtained suggests an important role for intra-domain-domain conformational dynamics. Thus alternative methodologies will be required for advanced understanding of such processes.

One very promising approach towards better understanding conformational interactions within full length multi-domain kinases involves the use of intein-based technologies to generate a native peptide bond between given pairs of contiguous regulatory and catalytic kinase domains (31-34). Successful peptide conjugation would facilitate domain-specific incorporation of biophysical probes (e.g., fluorophores and paramagnetic spin labels) and stable isotopes (e.g., ¹⁵N and ¹³C), enabling spectroscopic characterization of intra-domain-domain conformational dynamics in solution. While a large number of small C-terminal peptides have been successfully ligated to their cognate proteins, few cases have been reported in which two large functional domains have been conjugated to reconstitute regulatory function in a monomeric multi-domain native enzyme (34). To begin addressing the possibility that intein-based technologies may be used to reconstitute a monomeric multi-domain protein kinase for spectroscopic functional studies, we selected the human serine-threonine AGC protein kinase, PDK1, for which X-ray structures are known for both the N-terminal catalytic kinase (5) and the C-terminal PH domains (17). The C-terminal PH domain is required for PI(3,4,5)P₃-dependent co-localization and phosphoryaltion of PKB (35).

In this paper, we report efficient protein trans-splicing of the regulatory PH and catalytic kinase domains of human PDK1 by using naturally split DnaE inteins from two different organisms. In addition, spliced-PDK1 is shown to retain the ability to co-localize and selectively phosphorylate Thr-309 of PKBβ in a PI(3,4,5)P₃-dependent manner. The high-level production and reconstitution of functional spliced-PDK1 establishes the feasibility of generating full length PDK1 with either (i) segmental ¹⁵N-isotopic labeling of the PH domain for NMR studies or (ii) simultaneous site-directed modifications of the PH and kinase domains with small molecule donor and acceptor fluorophores for FRET-based studies of conformational transitions related to docking-based catalytic specificity.

EXPERIMENTAL PROCEDURES

Materials

PDK1-Tide (acetyl-KTFCTGTPEYLAPEVRREPRILSEEEQEMFRDYIADWC) and Crosstide (acetyl-GRPRTSSFAEG) were from 21st Century Biochemicals, Inc. (Marlboro, MA). [γ-³²P]ATP was from MP-Biomedical (Irvine, CA). 1,2-dioleoyl-sn-glycerol-3-phosphocholine (PC), 1,2-dioleoyl-sn-glycerol-3-phospho-L-serine (PS), and sn-1-stearoyl-2-arachidonyl D-phosphatidylinositol-(3,4,5)-triphosphate [PI(3,4,5)P₃] were from Echelon Biosciences (Salt Lake City, UT). Full length His₆-PKBβ and His₆-PDK1 (residues 51–556) were expressed using the Bac-to-Bac® Baculovirus Expression System (Invitrogen, Carlbad, CA) and His₆ affinity purified as described (36). Affinity purified His₆-PKBβ was fully dephosphorylated by treatment with Lambda protein phosphatase (λPP) as described (36).

PDK1 Trans-Splicing Strategy

Figure 1 shows the protein engineering strategy designed for trans-splicing of the regulatory PH and catalytic kinase domains of human PDK1. This strategy derives from recent reporting of the highly efficient in vivo cross trans-splicing reaction catalyzed between the N-terminal DnaE intein from Nostoc punctiforme [(N)NpuDnaE] and the C-terminal DnaE intein from Synechocystis sp. strain PCC6803 [(C)SspDnaE] (37). One additional advantage of utilizing (N)NpuDnaE is that it has been demonstrated to be more tolerant of amino acid substitutions at the at the C-terminal side of the splice junction (37). Since three-dimensional structures have been reported for the N-terminal catalytic kinase domain (residues 51–359) (5) and the C-terminal regulatory PH domain (residues 409–556) (17), C385 was chosen as the optimal site of conjugation. In utilizing the (N)NpuDnaE and (C)SspDnaE cross trans-splicing intein pair, ³⁸²QFG³⁸⁴ at the N-terminal side of the splice junction was replaced with ³⁸²AEY³⁸⁴. In addition, a conservative Q387N mutation gave ³⁸⁵CMN³⁸⁷ residues at the C-terminal side of the splice junction, which has been shown to be equally effective as the native CFN sequence required for trans-splicing with the native (N/C)SspDnaE pair (37). In order to facilitate purification of spliced-PDK1 from cleaved and uncleaved DnaE intein containing fragments, His₆ affinity tags were joined to the corresponding DnaE inteins. Addition of GST significantly increased the overall protein yield and solubility over that obtained by expression of the His₆-(C)SspDnaE-(CMN)PH construct.

Figure 1.

(A) Protein trans-splicing reactants. The N-terminal kinase domain of PDK1 (residues 51-381, black oval) is fused to the (N)NpuDnaE trans-splicing intein (gray square) with a C-terminal His₆ affinity tag (open square). The N-terminal three residues flanking the splice junction (Q382, F383, and G384) were mutated to the preferred AEY sequence required in trans-splicing. The C-terminal PH domain of PDK1 (residues 388-556, black oval) is fused to the (C)SspDnaE trans-splicing intein (gray square) with an N-terminal GST-His₆ tag (open square). Whereas the His₆ tag was used for affinity purification, the GST tag significantly improved both expression yield and solution stability. For the native C-terminal three residues flanking the splice junction (C385, M386, and Q387), only Q387 was mutated to yield the preferred CMN sequence optimal for cross trans-splicing of the (C)SspDnaE intein with the (N)NpuDnaE intein. The (N)NpuDnaE and (C)SspDnaE split inteins associate with high affinity and catalyze trans-splicing of the regulatory PH and catalytic kinase domains by forming a native peptide bond between Tyr-384 and Cys-385. (B) Protein trans-splicing products. The primary domain arrangement of PDK1 is regenerated, which contains the Q382A, F383E, G384Y, and Q387N mutations that flank the spice junction. The affinity tagged trans-splicing inteins are indicated. Spliced-PDK1 is separated from the cleaved (N)NpuDnaE-His₆ and GST-His₆-(C)SspDnaE split inteins, as well as any uncleaved intein fusion constructs by His₆ affinity purification. The calculated molecular weights (MW) of each trans-splicing reactant and product are indicated.

Baculovirus-Mediated Expression of KINASE(AEY)-(N)SspDnaE-His₆

Experimental procedures describing PCR engineering to generate an EcoRI–XbaI cDNA fragment containing the KINASE(AEY)-(N)NpuDnaE-His₆ fusion construct are given as described (Supporting Information). A sequence verified EcoRI–XbaI fragment containing the KINASE(AEY)-(N)NpuDnaE-His₆ construct (Figure 1) was removed from the pCR®-Blunt IITOPO® cloning vector and ligated into pFastBac™1 vector (Invitrogen) to generate recombinant baculovirus using the Bac-to-Bac® Baculovirus Expression System (Invitrogen). Recombinant P2 viral stocks were used to infect 200 mL spinner flask cultures of Sf9 cells in the mid-logarithmic phase of growth (2 × 10⁶ cells/mL) at a multiplicity of infection (MOI) of 2 viral particles/cell. The infected cells were incubated at 27 °C for 72 h and harvested by centrifugation for 10 min at 4 °C at 3000 rpm in a Beckman tabletop centrifuge. The cells were re-suspended in 50 mL of buffer (50 mM Tris-HCl, pH 7.5, 300 mM NaCl, 50 mM imidazole, 1 mM 2-mercaptoethanol, 1mM benzamidine, and 0.2 mM PMSF), and were immediately stored at −80 °C. The re-suspended frozen cells were allowed to thaw on ice for 1 h, which resulted in efficient cell lysis. Cell debris was pelleted by centrifugation for 30 min at 4 °C at 12,000 rpm. The supernatants containing the soluble components of the cell lysate were collected, and the KINASE(AEY)-(N)NpuDnaE-His₆ construct was His₆-affinity purified from the soluble lysate as described below.

Bacterial Expression of GST-His₆-(C)SspDnaE-(CMN)PH

Experimental procedures describing PCR engineering to generate the SacII–KpnI fragment containing the (C)SspDnaE-(CMN)PH fusion construct are given as described (Supporting Information). A sequence verified SacII–KpnI fragment containing the (C)SspDnaE-(CMN)PH construct was removed from the pCR®-Blunt II-TOPO® cloning vector and ligated into the pET-41b protein expression vector (Novagen), which provided N-terminal dual GST-His₆ affinity tags to yield the GST-His₆-(C)SspDnaE-(CMN)PH construct (Figure 1). The protein expression vector was transformed into the Rosetta (DE3) protein expression strain of E. coli. Cells were grown in 1 L Luria Broth (LB) containing kanamycin (50 μg/mL) and chloramphenicol (37 μg/mL) at 37 °C until an OD of 0.4. Then the cell culture was transferred to 15 °C and grown until an OD of 0.8 before induction with 0.5 mM isopropyl-β-D-thiogalactopyranoside (IPTG) for 16 h. The cells were harvested by centrifugation at 3500g for 20 min and then resuspended in 3 mL PBS buffer per gram cells (10 mM Na₂HPO₄ and 1.8 mM KH₂PO₄, pH 7.3, 140 mM NaCl, 2.7 mM KCl, 2 mM dithiothreitol, 1 mM benzamidine, and 0.2 mM PMSF). The cells were lysed using the EmulsiFlex®-C3 high pressure homogenizer (Avestin, Inc.), and particulates were removed from the lysate by centrifugation for 20 min at 14,000 rpm in the SS-34 Sorvall rotor. The GST-His₆-(C)SspDnaE-(CMN)PH construct was GST-affinity purified from the soluble lysate as described below.

His₆- and GST-Affinity Purification

The soluble Sf9 insect cell lysate containing the KINASE(AEY)-(N)NpuDnaE-His₆ recombinant fusion construct (Figure 1) was directly loaded by FPLC (1 mL/min) onto a 5 mL size Ni²⁺ Sepharose HiTrap HP affinity column (Amersham) equilibrated at 4 °C in 50 mM Tris-HCl, pH 7.5, 500 mM NaCl, 50 mM imidazole, 1 mM 2-mercaptoethanol. The soluble E. coli lysate containing the GST-His₆-(C)SspDnaE-(CMN)PH recombinant fusion construct (Figure 1) was directly loaded by FPLC (1 mL/min) onto a 5 mL size GST HiTrap HP affinity column (Amersham) equilibrated at 4 °C in PBS buffer (10 mM Na₂HPO₄ and 1.8 mM KH₂PO₄, pH 7.3, 140 mM NaCl, 2.7 mM KCl, and 2 mM dithiothreitol). The columns were subsequently washed until the absorbances at 260 nm and 280 nm returned to baseline. The KINASE(AEY)-(N)NpuDnaE-His₆ construct was eluted by linear increasing of the imidazole concentration from 50 to 500 mM at 1 mL/min for 30 min. The GST-His₆-(C)SspDnaE-(CMN)PH construct was eluted with buffer containing 50 mM Tris-HCl, pH 8.0, 2 mM dithiothreitol, and 20 mM glutathione. Chromatographic fractions (1 mL) containing the designated fusion protein were detected by absorbance at 280 nm and analyzed by SDS-PAGE. Fractions containing recombinant protein were pooled and subjected to gel exclusion chromatography for exchange to the protein trans-splicing buffer (50 mM Tris-HCl, pH 7.5, 500 mM NaCl, 0.5 mM TCEP, 0.5 mM EGTA, 1 mM EDTA, and 5% (v/v) glycerol). Protein was concentrated using a VivaSpin 20, 6, or 2 mL ultrafiltration concentrator (10 kDa molecular weight cutoff, VivaScience). Protein concentrations were determined by BioRad protein assay. Protease inhibitors (0.5 mM benzamidine and 0.1 mM PMSF) were added to each purified protein construct. A ∼40 μM stock solution of the KINASE(AEY)-(N)NpuDnaE-His₆ construct was prepared using ∼2 mg/mL total protein with ≥95% purity. A ∼40 μM stock solution of the GST-His₆-(C)SspDna-(CMN)PH construct was prepared using ∼2.8 mg/mL total protein with ≥60% purity. These solutions were stored at −80 °C.

PDK1 Trans-Splicing

Frozen aliquots containing ∼40 μM of either the KINASE(AEY)-(N)NpuDnaE-His₆ or the GST-His₆-(C)SspDna-(CMN)PH trans-splicing constructs were thawed at 21 °C. After 10 min microcentrifugation of each tube, the reaction was initiated by combining the clarified supernatants to yield ∼20 μM of each construct. The reaction tube was incubated either at 4 °C or 21 °C, and at varying times (0, 1, 3, 6, 9, 12, and 18 h) 9 μL aliquots containing ∼18–20 μg total protein were removed and quenched by addition to 9 μL of 2× SDS sample buffer, heated at 95 °C for 2 min, and placed on ice for SDS-PAGE analysis of trans-splicing progress. At the end of the time course, the reaction mixture was buffer exchanged by gel exclusion chromatography to 50 mM Tris-HCl, pH 7.5, 500 mM NaCl, 50 mM imidazole, and 1 mM 2-mercaptoethanol, the equilibration buffer for His₆ affinity chromatography. The cleaved (N)NpuDnaE-His₆ and GST-His₆-(C)SspDna products and any uncleaved KINASE(AEY)-(N)NpuDnaE-His₆ and GST-His₆-(C)SspDna-(CMN)PH reactants were removed by passage over a 5 mL size Ni²⁺ Sepharose HiTrap HP affinity column (Amersham). The column was subsequently washed until the absorbances at 260 nm and 280 nm returned to baseline. Flow-through fractions containing spliced-PDK1 were desalted and directly loaded by FPLC (0.5 mL/min) onto a Tricorn™ Mono Q™ 5/50 GL anion exchange column (Amersham) equilibrated at 4 °C in Tris-HCl buffer, pH 7.5, containing 2 mM 2-mercaptoethanol. Upon linear increasing of the NaCl concentration from 0 to 1 M at 1 mL/min for 1 h, ≥95% purified spliced-PDK1 eluted between 0.1–0.2 M NaCl. Fractions containing purified spliced-PDK1 (∼35 μM or ∼2 mg/mL) were combined, and protease inhibitors (0.5 mM benzamidine and 0.1 mM PMSF) were added before storage of either 50 μL or 1 mL aliquots at −80 °C.

SDS-PAGE and Western Analysis

Protein samples in SDS sample buffer were heated at 95 °C for 2 min and cooled on ice. Analytical SDS-PAGE was performed on 4-20% gradient polyacrylamide gels (GradiPore) developed at 150 V (constant) for 1 h or until the tracking dye reached the bottom of the slab. Coomassie staining was used to visualize total protein. Densitometric analysis was performed using the ImageQuant software (Molecular Dynamics). For Western analyses, protein was transferred from the gel to Nitrocellulose Membrane Filter Paper Sandwich (Invitrogen) in a semidry blotting apparatus using 0.7% (v/v) acetic acid as transferring solution. According to the manufacturer's instructions, purified KINASE(AEY)-(N)NpuDnaE-His₆ was probed with phospho-PDK1 (Ser241) polyclonal rabbit antibody (Cell Signaling Technology); and His₆-PKBβ was probed with either phospho-PKB (Thr-308) or phospho-PKB (Ser473) polyclonal rabbit antibodies (Cell Signaling Technology). Detection of all immuno-protein complexes was carried out using secondary anti-rabbit antibody conjugated to horseradish peroxidase (HRP) and LumiGLO® chemiluminescent reagent and peroxide.

Kinetic Assays and Phosphopeptide Mapping

V_max and K_m values at 30 °C and percent inactivation over time of purified 20 μM KINASE(AEY)-(N)NpuDnaE-His₆ in protein trans-splicing buffer at 21 °C was measured by quantifying transfer of ³²P-radiolabel from [γ-³²P]ATP to the model PDK1-Tide substrate as described (36). PI(3,4,5)P₃-dependent phosphorylation of His₆-PKBβ by spliced-PDK1 and measurement of the resulting increase in His₆-PKBβ-catalyzed transfer of ³²P-radiolabel from [γ-³²P]ATP (100 μM) to the model Crosstide substrate (50 μM) was carried out as described (36). For both assays, one unit of activity was defined as the amount of enzyme required to catalyze phosphorylation of 1 nmol of peptide substrate in 1 min.

The ability of spliced-PDK1 to selectively phosphorylate Thr-309 in the activation loop of His₆-PKBβ was tested in a 1 mL reaction mixture containing 2 μM of each enzyme (60 μg or 1 nmol) in 50 mM Tris-HCl buffer, pH 7.5, 1 mM 2-mercaptoethanol, 10 mM MgCl₂, 0.2 mM sodium vanadate, and PC/PS/PI(3,4,5)P₃ vesicles. The reaction was initiated by addition of 100 μM [γ-³²P]ATP (∼500 cpm/pmol) and allowed to proceed with continuous shaking at 30 °C for 40 min. The reaction was quenched by addition of urea to 6 M. Control assays were carried out in parallel in which either spliced-PDK1, PI(3,4,5)P₃ in the phospholipid vesicles, or both were omitted.

³²P-radiolabeled His₆-PKBβ was affinity-purified from the quenched reaction mixture by addition of 200 μL of Ni-NTA agarose resin and shaking overnight at room temperature. After low speed centrifugation, the supernatant was removed and the resin was washed twice with 1 mL of 50 mM Tris-HCl, pH 7.5, containing 6 M urea, 50 mM imidazole, 300 mM NaCl, and 1 mM 2-mercaptoethanol. ³²P-radiolabeled His₆-PKBβ was eluted from the Ni-NTA agarose resin pellet with 100 μL of 50 mM Tris-HCl, pH 7.5, containing 6 M urea, 500 mM imidazole, 300 mM NaCl, and 1 mM 2-mercaptoethanol; and approximately 40–50% of enzyme (400–500 pmol) was recovered in the 100 μL elute. First, protein concentration was determined with the Bio-Rad protein assay kit, using 10 μL of the eluted enzyme. Second, the specific radioactivity of the ³²P-radiolabeled His₆-PKBβ construct (SA^PKBβ, cpm/pmol) was determined from radioactivity detected by scintillation counting of the known amount of enzyme (∼40–50 pmol) in 10 μL of the eluted enzyme. The mole fraction of total phosphorylated sites (f_mol^tot) was calculated according to f_mol^tot = SA^PKBβ/SA^ATP.

To determine site-specific phosphorylation, the remaining amount (∼80 μL with ∼320–400 pmol) of affinity-purified His₆-PKBβ was used for trypsin digestion and HPLC resolution of ³²P-radiolabeled peptides. First, cysteine residues were reduced by addition of 0.5 mL of 50 mM Tris-HCl, pH 7.5, containing 20 mM 2-mercaptoethanol. Free cysteines were protected from re-oxidation by subsequent addition of 50 μL of 500 mM iodoacetamide incubation in the dark at room temperature for 20 min. Excess unreacted iodoacetamide was depleted from the reaction mixture by further addition of 30 μL of 500 mM 2-mercaptoethanol and incubation at room temperature for 5 min. Proteolytic digestion was carried out by addition of 1 μg trypsin and incubation at 37 °C for 5 h, followed by subsequent addition of 1 μg trypsin and incubation at 37 °C overnight. The individual reaction mixtures (∼200 μL) containing digested ³²P-radiolabeled His₆-PKBβ constructs were diluted to 1 mL with solvent A [0.1% (v/v) trifluoroacetic acid in water]. The 1 mL samples were directly loaded by HPLC (0.5 mL/min) onto a μRPC C2/C18 column (Amersham) equilibrated in 100% solvent A. The column was subsequently washed in this buffer for 5 min, and the peptides were eluted by linear increasing solvent B [80% acetonitrile and 0.07% (v/v) trifluoroacetic acid in water] from 0% to 50% in 1 h at a flow rate of 0.4 mL/min, while 300 μL were collected for each fraction. ³²P-radiolabeled peptides were detected by scintillation counting of 30 μL of each chromatographic fraction.

The identity of the ³²P-radiolabeled peptide was confirmed by MALDI-TOF mass spectrometric analysis. The 300 μL fraction containing the ³²P-radiolabeled peptide was evaporated to dryness and resuspended in a small volume (10–30 μL) of 50% acetonitrile to yield a ³²P-peptide concentration of 10 pmol/μL, as indicated by the amount of radioactivity; 1 μL of saturated 2,5-dihydroxybenzoic acid (2,5-DHB) was added to 1 μL of the concentrated ³²P-peptide, and 1 μL of this mixture containing ∼5 pmol of the ³²P-peptide was placed on a stainless steel MALDI-TOF target plate and allowed to dry. MALDI-TOF mass spectra were acquired on a Biflex IV MALDI-TOF mass spectrometer (Bruker Daltronics) in either linear (larger peptides) or reflectron (smaller peptides) mode. A N₂ laser was used to desorb/ionize the matrix/analyte material. Calibration was performed using angiotensin II (monoisotopic mass [MH+] 1046.5423 Da), angiotensin I (monoisotopic mass [MH+] 1296.6900 Da), bombesin (monoisotopic mass [MH+] 1619.8229 Da), and adrenocorticotropic hormone clip 18–39 (monoisotopic mass [MH+] 2465.2027 Da) (Sigma-Aldrich, St. Louis, MO).

RESULTS AND DISCUSSION

Development of PDK1 Cross Trans-Splicing Strategy

It has been underreported but overestablished that a majority of human protein kinases are poorly expressed and purified as inclusion bodies from both bacterial and yeast protein expression strains under a wide variety of growth and induction conditions. In addition, such kinase preparations have proven difficult to re-solubilize into stable and active conformations. To date, baculovirus-mediated protein expression in either Sf9 or Sf21 insect cells has been the most effective method for generating high levels of soluble and active human protein kinases (36). Although the reason is not entirely clear, we observed that high production of the catalytic kinase domain constructs of PDK1, S6K1, and PKBβ, typically achieved by baculovirus-mediated expression in Sf9 insect cells, was severely attenuated when these constructs were fused with a C-terminal active intein (e.g., Mxe GyrA or VMA inteins available from New England Biolabs, Inc.) (unpublished results). In addition, the small amounts of expressed fusion protein (≤0.5 mg/200 mL Sf9 insect cell culture) were observed to be ≥80% hydrolyzed. Therefore, only very small quantities of N-kinase-intein fusion protein (≤0.1 mg/200 mL Sf9 insect cell culture) were available for thiolysis of the intein to generate the C-terminal thioester required for chemical ligation to the N-terminal cysteine residue of a C-terminal peptide fragment. Thus, we reasoned that high-level baculovirus-mediated expression in Sf9 insect cells of human catalytic kinase domains for engineered ligation could be better achieved if fused to an inactive intein fragment of a protein trans-splicing intein pair.

To date, the naturally occurring split Ssp DnaE intein (38-40) and the artificially split Ssp DnaB and Sce VMA inteins (41, 42) have been best characterized with regard to achieving protein trans-splicing activity in the absence of protein refolding routines typically required of artificially split inteins. However in all these cases, protein trans-splicing under native conditions has been compromised by competing cleavage of the protein trans-splicing reactants upon mixing, resulting in typical yields of only 40–50% spliced protein product. Encouragingly, highly efficient protein trans-splicing (≥98%) of two B1 domains of the IgG binding protein (GB1) was recently demonstrated when the His₆-GB1-(N)NpuDnaE and (C)SspDnaE-GB1 cross reacting trans-splicing constructs were co-expressed in E. coli (37). Since it is difficult to access contributions of endogenous chaperones and the reducing environment to the observed high splicing efficiency in E. coli, it became of immediate interest whether this highly efficient in vivo ’cross’ trans-splicing strategy would translate to in vitro conditions.

Preparation of Cross Trans-Splicing Constructs

The KINASE(AEY)-(N)NpuDnaE-His₆ fusion construct of PDK1 (Figure 1) was generated by PCR and subcloned into the pFastBac™1 vector for generation of recombinant baculovirus using the Bac-to-Bac® Baculovirus Expression System (Invitrogen). Using MOI = 2 and a time of harvest 72 h after post infection, KINASE(AEY)-(N)NpuDnaE-His₆ was overexpressed and His₆ affinity purified, typically yielding 10 ± 2 mg from initial infection of 4 × 10⁸ total Sf9 insect cells (Table 1). Figure 2A shows that the KINASE(AEY)-(N)NpuDnaE-His₆ was efficiently purified to ≥95% homogeneity (lane 3), as judged by Coomassie blue staining of 4–20% SDS-PAGE. Purified KINASE(AEY)-(N)NpuDnaE-His₆ migrates with an apparent molecular weight of 51 kDa, which agrees with its calculated molecular mass of 50,802 Da. Similar to the recombinant catalytic domain construct of PDK1 [His₆-PDK1(ΔPH), residues 51–359] (5), Western analysis showed that the KINASE(AEY)-(N)NpuDnaE-His₆ construct was purified in its Ser-241 phosphorylated and catalytically active form (Figure 2A, lower panel). The values of V_max = 200 ± 15 U/mg and ^TideK_m = 70 ± 10 μM measured KINASE(AEY)-(N)NpuDnaE-His₆-catalyzed phosphorylation of the model PDK1-Tide substrate were in the same range as those measured for native His₆-PDK1 (36).

Table 1.

Purification of recombinant KINASE(AEY)-(N)NpuDnaE-His₆ from Sf9 insect cells^a

Purification (Step)	Volume (mL)	Concentration (mg/mL)	Yield (mg)	Purification (fold)
Crude lysate	50	2.2 ± 0.2	110 ± 10	N/A
Ni Sepharose	10	1.0 ± 0.2	10 ± 2	11

^aAll values are reported for purification from initial infection of 200 mL of insect Sf9 cells (2 × 10⁶ cells/L) with recombinant baculovirus.

Figure 2.

SDS-PAGE analysis of purification of protein trans-splicing reactants. (A) KINASE(AEY)-(N)NpuDnaE-His₆ was His₆ affinity purified from Sf9 insect cell lysate. (B) GST-His₆-(C)SspDnaE-(CMN)PH was GST affinity purified from E. coli lysate. Lane 1 shows the total lysate; lane 2 shows protein species that were not retained by the affinity column; and lane 3 shows the affinity purified recombinant protein trans-splicing construct. Molecular weight markers are indicated. Proteins were visualized by Coomassie staining.

The (C)SspDnaE-(CMN)PH fusion construct was cloned into pET-41b vector downstream of a GST-His₆ tag, yielding the coding sequence for the GST-His₆-(C)SspDnaE-(CMN)PH fusion protein (Figure 1). Recombinant protein expression of cell cultures (OD of 0.8) were induced with 0.5 mM IPTG for 16 h at 15 °C. Cell lysate from 1 L culture was collected by homogenization, and GST affinity purification typically yielded 17 ± 3 mg total protein (Table 2). Figure 2B shows that the GST-His₆-(C)SspDnaE-(CMN)PH construct was purified to ∼60% homogeneity (lane 3), as judged by Coomassie blue staining of 4–20% SDS-PAGE. Affinity purified GST-His₆-(C)SspDnaE-(CMN)PH showed anomalous slightly faster migration corresponding to an apparent molecular weight of 49 kDa compared with its calculated molecular mass of 51,214 Da. The ∼40% impurities retained after GST affinity purification of GST-His₆-(C)SspDnaE-(CMN)PH from the E. coli are primarily composed of a mixture of four protein species that migrate with apparent molecular weights of approximately 24, 25, 26 and 37 kDa, which were efficiently removed in subsequent chromatographic steps used to purify spliced-PDK1 from the trans-splicing reaction (see below).

Table 2.

Purification of GST-His₆-(C)SspDnaE-(CMN)PH from E. coli^a

Purification (Step)	Volume (mL)	Concentration (mg/mL)	Yield (mg)	Purification (fold)
Crude lysate	25	14 ± 2	350 ± 50	N/A
Ni Sepharose	10	1.7 ± 0.3	17 ± 3	21

^aAll values are reported for purification from expression in 1 L of E. coli.

PDK1 Cross Trans-Splicing Kinetics and Yield

Figure 3A shows the progress of protein trans-splicing that occurs over 12 h. The 10 mL protein trans-splicing reaction (21 °C) was initiated by combining 2 mmol of each of the protein trans-splicing reactants, KINASE(AEY)-(N)NpuDnaE-His₆ (species I) and GST-His₆-(C)SspDna-(CMN)PH (species II) (∼20 μM each). SDS-PAGE with Coomassie staining shows decreasing band intensities corresponding to consumption of the trans-splicing reactants (species I and II) with a concomitant increasing band intensity corresponding to formation of full length spliced-PDK1 (species III). In addition, increasing band intensities were observed for formation of the cleaved affinity tagged intein constructs, GST-His₆-(C)SspDna (species IV) and (N)NpuDnaE-His₆ (species V). Densitometric analysis of multiple reaction gels indicated that PDK1 trans-splicing follows apparent first-order kinetics with k_obs = (2.8 ± 0.3) × 10⁻⁵ s⁻¹ (Figure 3B). This rate is approximately 10-fold slower than the k_obs = (3.0 ± 0.4) × 10⁻⁵ s⁻¹ reported for trans-splicing of the maltose binding protein and thioredoxin utilizing the native (N/C)SspDnaE intein pair, which was also performed at room temperature (40). However, the fit of the kinetic data (Figure 3B) also indicated that the cross trans-splicing strategy facilitates a limiting reaction extent approaching 90 ± 10%. This is corroborated by observation of the faint band intensity indicated for KINASE(AEY) (species VI ≤ 5%), which results from trans-cleavage or hydrolysis of KINASE(AEY)-(N)NpuDnaE-His₆. The extent of the competing trans-cleavage reaction is significantly lower than those typically reported for the native (N/C)SspDna intein pair (38-40). Disappointingly, the trans-splicing reaction was much less efficient when performed at 4 °C.

Figure 3.

(A) Progress of PDK1 trans-splicing. Affinity purified KINASE(AEY)-(N)NpuDnaEHis₆ and GST-His₆-(C)SspDnaE-(CMN)PH protein trans-splicing reactants were mixed and allowed to react at 21 °C. Aliquots of the reaction mixture were removed and analyzed by SDS-PAGE, which indicates the distribution of reactants and products at varying times of 0 h (lane 1), 1 h (lane 2), 3 h (lane 3), 6 h (lane 4), 9 h (lane 5), and 12 h (lane 6). (B) Time courses for KINASE(AEY)-(N)NpuDnaE-His₆ inactivation (□) and yield of spliced-PDK1 (•) at 21 °C. (C) Preparative amounts of trans-splicing reactants were mixed (lane 1) and allowed to react at 21 °C for 18 h (lane 2). Lane 3 shows the protein species in the reaction mixture that were retained by His₆ affinity chromatography. Lane 4 shows purified spliced-PDK1 (≥95%), which was resolved by MonoQ ion exchange chromatography from trace amounts of protein species that were not retained by His₆ affinity chromatography.

Although the cross trans-splicing reaction at 21 °C should approach 90% yield, it was not reasonable to allow the reaction to proceed to such limits, as a k_obs = (4.4 ± 0.4) × 10⁻⁶ s⁻¹ was determined for KINASE(AEY)-(N)NpuDnaE-His₆ inactivation at 21 °C (Figure 3B). Thus, a reaction time of 18 h was selected as an optimal time for maximizing the yield of spliced-PDK1, while retaining significant enzyme stability and function (Figure 3C, lane 2). The reaction mixture was then buffer exchanged and subjected to His₆ affinity chromatography. Figure 3C (lane 3) shows that the cleaved affinity tagged intein constructs, GST-His₆-(C)SspDna (species IV) and (N)NpuDnaE-His₆ (species V), as well as the remaining uncleaved KINASE(AEY)-(N)NpuDnaE-His₆ and GST-His₆-(C)SspDna-(CMN)PH reactants were effectively retained by the column. In addition, three of the four contaminating protein impurities (24, 25, and 26 kDa) that co-purified with GST-His₆-(C)SspDna-(CMN)PH (Figure 3C, lane 3) were retained, suggesting that these may likely result from proteolytic degradation prior to cell lysis. Spliced-PDK1, the contaminating species (37 kDa), and the small amount of trans-cleaved KINASE(AEY), which were not retained by the His₆ affinity column were readily resolved by MonoQ ion exchange chromatography. Figure 3C (lane 4) shows that spliced-PDK1 is purified to ≥95% homogeneity. For this optimized preparative procedure, ∼1 mmol of purified spliced-PDK1 could be readily obtained from 2 mmol of each of the starting reactants.

PI(3,4,5)P₃-Dependent Phosphorylation of PKBβ by Spliced-PDK1

The best characterized function of the C-terminal PH domain of PDK1 is its ability to bind the PI(3,4,5)P₃ second messenger, causing its membrane co-localization with PKB, which also contains a PH domain that selectively binds PI(3,4,5)P₃ (43-53). At the membrane, PDK1 catalyzes phosphorylation of the critical residue in the activation loop of PKB, which is required for downstream PKB signaling activity. For the three known isoforms of PKB, the phosphorylation sites correspond to Thr-308, Thr-309, and Thr-305 in PKBα, PKBβ, and PKBγ, respectively. In vitro characterization of PDK1-catalyzed site-specific phosphorylation of PKB isoforms, which utilize phosphoinositide-containing vesicles, demonstrated an absolute requirement for PI(3,4,5)P₃-containing vesicles, as well as functional PH domains in both PDK1 and PKB (45, 46, 50). Since protein trans-splicing of the regulatory PH and catalytic kinase domains of PDK1 required mutation of ³⁸²QFG³⁸⁴ to ³⁸²AEY³⁸⁴ at the N-terminal side of the splice junction, as well as the Q387N mutation to give ³⁸⁵CMN³⁸⁷ at the C-terminal side of the splice junction (Figure 1), it was very important to assess whether spliced-PDK1 retained the characteristic ability to specifically phosphorylate and activate PKB in a PI(3,4,5)P₃-dependent manner.

First, full length His₆-PKBβ was obtained by baculovirus-mediated expression and purification from Sf9 insect cells (36). As we have previously demonstrated, His₆-PKBβ purifies with minimal catalytic activity to the Crosstide substrate (∼0.3 U/mg) and displays very small but detectable amounts of both Thr-309 and Ser-474 phosphorylation, as detected by Western analysis (Figure 4A, lane 1). Previous isoelectric focusing studies, as well as ion exchange chromatography could not resolve the phosphorylated isoforms, suggesting that the small detectable amount of phosphorylated His₆-PKBβ likely consists of a mixture of enzyme monophosphorylated either at Thr-309 or Ser-474 (36). Nevertheless, homogeneous unphosphorylated His₆-PKBβ with slightly lower Crosstide activity (∼0.2 U/mg) was readily generated by treatment with λPP (Figure 4A, lane 2). Western analysis further showed that λPP-treated unphosphorylated His₆-PKBβ did not catalyze autophosphorylation at either Thr-309 or Ser-474 when incubated in autophosphorylation reaction mixtures either in the absence (Figure 4A, lane 3) or presence of PI(3,4,5)P₃-containing vesicles (Figure 4A, lane 4). Most importantly, Western analysis detected significant Thr-309 phosphorylation of His₆-PKBβ only when incubated with spliced-PDK1 in the presence of PI(3,4,5)P₃-containing vesicles; and His₆-PKBβ catalytic activity to Crosstide increased ∼10-fold (Figure 4A, lanes 5 and 6). Affinity purification and phosphopeptide mapping of His₆-PKBβ from the phosphorylation reaction mixture containing both spliced-PDK1 and PI(3,4,5)P₃-containing vesicles (Figure 4A, lane 6) confirmed that spliced-PDK1 catalyzed site-specific Thr-309 phosphorylation of His₆-PKBβ (Figure 4B), as no other ³²P-radiolabeled peptides were detected. The mole fraction of total phosphorylated sites (f_mol^tot) was calculated to be 0.65 ± 0.15, in agreement with previous reports (50).

Figure 4.

Site-specific Thr-309 phosphorylation by spliced-PDK1. (A) Preparation of His₆-PKBβ (lane 1); preparation of λPP-treated unphosphorylated His₆-PKBβ (lane 2); and affinity purified λPP-treated His₆-PKBβ after 40 min incubation at 30 °C in phosphorylation reaction mixtures containing [γ-³²P]ATP-Mg²⁺ (lane 3), [γ-³²P]ATP-Mg²⁺ in the presence of PC/PS/PI(3,4,5)P₃ vesicles (lane 4), [γ-³²P]ATP-Mg²⁺ in the presence of spliced-PDK1 (lane 5), and [γ-³²P]ATP-Mg²⁺ in the presence of both spliced-PDK1 and PC/PS/PI(3,4,5)P₃ vesicles (lane 6). SDS-PAGE and Western analysis were used to visualize Thr-309 and Ser-474 phosphorylated His₆-PKBβ. Activities were determined for His₆-PKBβ-catalyzed phosphorylation of 100 μM of Crosstide (.). (B) After 40 min incubation at 30 °C in the phosphorylation reaction mixture containing [γ-³²P]ATP-Mg²⁺, affinity purified λPP-treated His₆-PKBβ was digested with trypsin and subjected to reversed-phase HPLC. Scintillation counting of the individual fractions detected the ³²P-radiolabeled peptide that eluted near 28% acetonitrile, which was identified by MALDI-TOF (m/z = 2469.6) to be the monophosphorylated tryptic peptide containing Thr-309 (³⁰⁹pTFCGTPEYLAPEVLEDNDYGR³²⁹).

PROSPECTIVE AND SUMMARY

Large efforts are being directed towards design of potent and selective inhibitors of well established protein kinase drug targets (54, 55). Since the overwhelming majority of protein kinase inhibitors bind in or near the ATP binding pocket shared by the structurally homologous catalytic domain shared by all kinases, very few serine-threonine protein kinase inhibitors have been clinically approved due to their broad specificity and overall high toxicity (56-58). Therefore, we are developing the hypothesis that serine-threonine protein kinase inhibitor selectivity may be better achieved by designing compounds that target interfacial clefts and crevices formed between contiguous regulatory and catalytic kinase domains, thereby ‘allosterically’ stabilizing inactive or autoinhibited conformations of multi-domain protein kinases (59). In fact, validation of this hypothesis is exemplified by the recent discovery of the small molecule inhibitors, Akt-I-1 and Akt-I-1,2, which showed remarkable isozyme specificity towards PKB/Akt isozymes and did not significantly inhibit other closely related kinases (60). Most importantly, only the full length PKB/Akt isozymes containing the regulatory PH domain were sensitive to inhibition, suggesting an ‘allosteric’ mechanism in which high selectivity results from inhibitor binding at a site that stabilizes a PH domain mediated autoinhibited form of the kinase.

The inability to obtain diffraction quality crystals for full length constructs of any of the large number of multi-domain serine-threonine protein kinase drug targets suggests a large degree of conformational heterogeneity, which likely results from changes in relative orientations of contiguous regulatory and catalytic kinase domains. In order to better facilitate structure-based discovery or design of highly selective inhibitors (e.g., Akt-I-1 and Akt-I,2), development of alternative methodologies will be required to better understand autoinhibitory contiguous domain-domain interactions; and domain-specific incorporation of biophysical probes would greatly facilitate spectroscopic studies of these interactions in solution. Disappointingly, the use of intein-mediated chemical ligation for such purposes is largely precluded, as we found it difficult to purify from Sf9 insect cells adequate quantities of a number of protein kinases when fused with active inteins (e.g., Mxe GyrA and VMA). While the use of naturally [e.g., (N/C)SspDnaE] or artificially split intein pairs [e.g., (N/C)SspDnaB] circumvents decreased protein stability and intein self cleavage issues during baculovirus-mediated expression in Sf9 insect cells, only 40–50% protein trans-splicing yields are typically achieved. On the basis of the recently reported highly efficient in vivo protein trans-splicing (≥98%) catalyzed between split DnaE inteins from two different organisms (37), we implemented and demonstrated that the ‘cross’ trans-splicing strategy was optimal for in vitro ligation of the regulatory PH and catalytic kinase domains of PDK1; and spliced-PDK1 was shown to retain the ability to specifically phosphorylate Thr-309 and activate PKBβ in a PI(3,4,5)P₃-dependent manner. This bioconjugation strategy may prove useful for design of spectroscopic experiments to detect and characterize the elusive frontier of potential allosteric drug targeting sites, which may exist at interfacial clefts and crevices in flexible multi-domain monomeric protein kinases.