|Home | About | Journals | Submit | Contact Us | Français|
Phosphoinositide kinases such as PI3-kinase synthesize lipid second messengers that control diverse cellular processes. Recently, these enzymes have emerged as an important class of drug targets, and there is significant interest in discovering new lipid kinase inhibitors. We describe here a procedure for the high-throughput determination of lipid kinase inhibitor IC50 values. This assay exploits the fact that phosphoinositides, but not nucleotides such as ATP, bind irreversibly to nitrocellulose membranes. As a result, the radiolabeled lipids from a kinase assay can be isolated by spotting the crude reaction on a nitrocellulose membrane and then washing. We show that diverse phosphoinositide kinases can be assayed using this approach and outline how to perform the assay in 96-well plates. We also describe a MATLAB script that automates the data analysis. The complete procedure requires 3–4 h.
Phosphatidylinositol (PI) is an abundant component of all eukaryotic cell membranes. In response to upstream signals, lipid kinases phosphorylate PI at specific positions on the inositol head group to generate a spectrum of lipid second messengers (Fig. 1). These differentially phosphorylated lipids bind to and regulate the activity of diverse effector proteins, including protein kinases, ion channels, guanine-nucleotide exchange factors, phospholipases and adaptor proteins. In this way, lipid kinases are able to control a wide range of cellular processes.
Recent interest in phosphoinositide signaling has been driven by studies of the class I PI3-kinases (p110α, p110β, p110δ and p110γ). These enzymes are activated by receptor tyrosine kinases and G-protein-coupled receptors to phosphorylate phosphatidylinositol-4,5-bisphosphate (PI(4,5)P2), thereby generating the second messenger phosphatidylinositol-3,4,5-trisphosphate (PIP3; Fig. 1). PIP3 acts as a docking site at the plasma membrane for signaling proteins that include the protein kinases 3-phosphoinositide-dependent protein kinase-1 (PDK1) and Akt; these kinases are activated by PIP3 to promote nutrient uptake, suppress apoptosis and drive cell proliferation1. PI3-kinase signaling is antagonized by PTEN2, a lipid phosphatase that dephosphorylates PIP3 to generate PI(4,5)P2. The PI3-kinase pathway is frequently activated in solid tumors, and mutations in p110α, Ras (an upstream PI3-kinase activator3–5) and PTEN are among the most common genetic alterations in cancer6–9. In addition, there is considerable evidence that p110δ and p110γ may be useful targets for the treatment of inflammation and autoimmune diseases10–15. Together, these observations have stimulated widespread interest in identifying selective PI3-kinase inhibitors16.
In addition to the well-characterized class I PI3-kinases, several other families of lipid kinases have been identified (Fig. 1). These include the class II and III PI3-kinases, which synthesize phosphatidylinositol-3-phosphate (PI(3)P) and phosphatidylinositol-3,4-bisphosphate (PI(3,4)P2); two families of PI4-kinases (designated type II and type III) that phosphorylate PI to generate phosphatidylinositol-4-phosphate (PI(4)P); and a family of phosphatidylinositol phosphate kinases (PIP-kinases) that synthesize primarily PI(4,5)P2. Furthermore, these lipid kinases are antagonized by phosphatases that dephosphorylate specific positions on the inositol ring (Fig. 1). Together, these lipid kinases and phosphatases coordinate the synthesis of a complex pool of second messengers that regulate diverse aspects of cell biology.
Purified lipid kinases are assayed in vitro to characterize their enzymatic properties or identify small molecule inhibitors. Traditionally, lipid kinase activity has been assayed by monitoring phosphate transfer to lipid via thin-layer chromatography (TLC). In this approach, the kinase is allowed to phosphorylate lipid in the presence of [γ-32P]ATP; radiolabeled lipids are separated from [γ-32P]ATP by extracting into organic solvent followed by TLC; and the radioactivity in isolated TLC spots is detected by phosphorimaging or scintillation counting. This approach is highly sensitive, as phospholipid isomers are chromatographically separated from each other as well as from residual ATP. A limitation of this assay, however, is that it is labor intensive and of low-throughput due to the need for extraction and TLC steps. Recently, fluorescence-based assays for the class I PI3-kinases have been developed that overcome some of these difficulties through the use of protein domains that bind to PIP3 (ref. 17).
We have initiated an effort to discover pharmacological inhibitors of specific lipid kinases and characterize the selectivity of these molecules across the lipid kinome18,19. To this end, we have developed a simple lipid kinase assay to facilitate the high-throughput determination of lipid kinase inhibitor IC50 values on a laboratory scale. This assay exploits the fact that phosphoinositides, but not nucleotides such as ATP, bind irreversibly to nitrocellulose membranes through hydrophobic interactions. As a result, it is possible to isolate the radiolabeled lipids from a kinase assay by spotting the crude reaction on a nitrocellulose membrane and then washing (Fig. 2). Using this approach, we have assayed diverse lipid kinases in a single format and determined several thousand inhibitor IC50 values.
We also describe here a MATLAB script (‘Spot’) that automates the analysis of phosphorimager data from this assay. By using this script, hand-spotted radioactivity can be counted quickly, uniformly and with minimal user intervention. The script, which utilizes MATLAB’s image analysis and statistics toolboxes, is available as MATLAB source code or as compiled executables for mac and pc platforms from http://www.ucsf.edu/shokat/SPOT.htm.
We have previously reported the profiling of lipid kinase inhibitors using this assay18. We describe here representative data illustrating features of this approach. Ten lipid kinases were assayed against a panel of phosphoinositides to characterize their lipid substrate specificity (Fig. 3). The observed substrate preferences in this assay closely mirror the reported biochemical selectivities of these enzymes. The three PI4-kinases each utilize exclusively PI as a substrate20–23. The PIP-kinases PI(4)P5-KIα and PI(4)P5-KIβ preferentially phosphorylate PI(4)P, whereas PI(5)P4-KIIβ phosphorylates PI(5)P24–26. The class I PI3-kinases (p110α, p110β, p110δ and p110γ) phosphorylate either PI or PI(4,5)P2 in a ratio that varies across the four isoforms. PI(4,5)P2 is the primary substrate of these enzymes in vivo, but PI is used preferentially in vitro under many assay conditions. The relative phosphorylation of these two substrates in vitro is known to depend on the identity and composition of other lipids in the membrane bilayer27. Overall, these data show that diverse lipid kinases that preferentially utilize at least four different phosphoinositide substrates can be assayed using the membrane capture approach.
Signal strength in this assay depends on length of time that the kinase reaction is allowed to proceed and the amount of enzyme used (Fig. 4). We have not observed saturation of the phosphoinositide binding capacity of nitrocellulose under the assay conditions described here, although, as in any enzymatic assay, it is possible to deplete substrate at sufficiently high enzyme concentrations. The optimal concentration of kinase for this assay depends on the specific activity of the enzyme under the conditions used, and this varies greatly across lipid kinases (see General assay considerations).
Figure 5 depicts typical data for the determination of IC50 values of two inhibitors (PIK-108 and PIK-93) against a lipid kinase (p110α). In this experiment, inhibitor was arrayed in 5 µl of 10% DMSO across eight rows of a 96-well PCR plate; PCR plates are used to facilitate handling small reaction volumes (25 µl), thereby minimizing the consumption of kinase and [γ-32P]ATP. The inhibitor was aliquoted in threefold dilutions to achieve final assay concentrations that range from 50 to 0.0003 µM, thereby spanning the predicted IC50 value for the drug. Kinase was then added to each well of the plate in 10 µl of a solution containing kinase reaction buffer, lipid and BSA. To initiate the kinase reaction, 10 µl of a solution containing 2 µCi [γ-32P]ATP was added to each well of the plate (adjusted to achieve a final ATP concentration of 10 µM in the assay). The kinase reaction was allowed to proceed for 30 min, at which point the reaction was terminated by spotting 4 µl from each well onto a nitrocellulose membrane. All liquid transfers were performed using a multi-channel pipettor to initiate and terminate the assay consistently. The membrane was then washed four times for 15 min each with 100–200 ml of wash solution (1 M NaCl/1% phosphoric acid). The last wash was allowed to proceed overnight. After allowing the membrane to dry, it was exposed to a phosphorimager screen to generate the raw data shown in Figure 5b. Quantification of these spots using the MATLAB script ‘Spot’ was performed as shown in Figure 6 to yield the dose–response data shown in Figure 5a. IC50 values for each inhibitor were obtained by fitting a sigmoidal dose–response curve to these data using the Prism software package.
The specific activity of the lipid kinase is the major determinant of the signal-to-background ratio in this assay. Extended washing removes the vast majority of radioactive ATP and phosphate from the nitrocellulose membrane, but trace amounts of radioactive material will adhere to the membrane independent of kinase activity. For relatively active lipid kinases such as the class I PI3-kinases or the PI4-kinases, we have found that a final kinase concentration in the assay of ~ 1 µg ml−1 (~ 5 nM) typically yields a signal-to-background ratio greater than 10. At this concentration, ~ 2 µg of enzyme is sufficient to assay a 96-well plate, which is cost efficient for most purposes. Other lipid kinases, such PI(5)P4-KIIβ, have very low specific activity under all assay conditions we have tested, and for these kinases it may be necessary to use 50- to 100-fold more enzyme to achieve similar results. Similar considerations apply to the analysis of kinase activity from immunoprecipitates. However, as the 96-well assay format described below requires a homogenous source of enzyme, immunoprecipitated kinases must be eluted from the beads before assaying in 96-well plates.
For each kinase preparation, it is necessary to empirically determine the appropriate enzyme concentration before starting high-throughput assays. As a general guideline, one should aim for a radioactive signal that is at least tenfold greater than negative controls in which either the kinase or the lipid has been omitted. During the optimization of enzyme concentration, it is also advisable to test several different reaction lengths (e.g., 30, 60 and 120 min) to assess the stability of the enzyme and the potential for enhanced signal-to-background ratio at longer reaction times.
The signal-to-background ratio in this assay is largely independent of the concentration of radioactive ATP, because the signal and background are both expected to increase as the concentration of radioactive ATP is increased (the background is caused by radioactivity adhering nonspecifically to the nitrocellulose membrane). The concentration of radioactive ATP primarily determines the overall signal strength of the assay and, therefore, the length of time that the membrane must be exposed to the phosphorimager screen. We typically use a radioactive ATP concentration of 0.1 Ci µl−1 with a total ATP concentration of 10 µM. If desired, the radioactive ATP concentration can be decreased at least fivefold without impairing the assay. [γ-33P]ATP can also be substituted for [γ-32P]ATP.
The ATP concentration used in this assay influences the IC50 value that is measured for a kinase inhibitor. For an ATP competitive kinase inhibitor, the IC50 value is related to the Ki of the inhibitor, the KM,ATP of the kinase and the ATP concentration by the Cheng–Prusoff equation28,29: IC50 = Ki(1 + [ATP]/KM,ATP). At ATP concentrations below the KM,ATP of the kinase, the IC50 value approximates the Ki. At ATP concentrations above the KM,ATP of the kinase, the measured IC50 value exceeds the Ki. The KM,ATP has been reported for many protein and lipid kinases28. Because the intracellular ATP concentration is 1–5 mM30,31, most kinase inhibitors appear less potent in cells than when assayed at lower ATP concentrations (typically 10–100 µM) in vitro.
The source and physical form of the lipids used for the assay can be important in some cases. Commercially available phosphoinositides may be purified from natural sources or synthesized chemically. We have found that both natural and synthetic lipids are compatible with the membrane capture assay, although in some cases natural lipids are less expensive. Synthetic lipids are often available in several hydrocarbon lengths (e.g., C8 and C16); we recommend using the longer hydrocarbons to ensure irreversible binding to the nitrocellulose membrane. The purity of commercial phosphoinositides is variable, and impurities in commercial preparations have led to the misassignment of lipid kinase substrate specificity26. For experiments in which the purity of the phosphoinositide is critical, commercial phosphoinosites should be independently characterized by HPLC or TLC. In this respect, the membrane capture assay does not distinguish between different phosphoinositides that may be generated by a kinase that phosphorylates a single lipid at multiple sites. These considerations are less relevant for the determination of inhibitor IC50 values using well-characterized lipid kinases.
The protocol outlined below describes how to perform lipid kinase assays in either individual microcentrifuge tubes or 96-well plates. The former protocol is appropriate for determining the optimal enzyme concentration and troubleshooting the assay. The 96-well plate procedure facilitates the rapid determination of inhibitor IC50 values.
Troubleshooting advice can be found in Table 1.
The membrane capture procedure can be used to screen small-molecule inhibitors against many different lipid kinases using a single assay format. The assay is of relatively high-throughput, yet requires no automation. Following the procedure described here, a single technician can expect to comfortably assay approximately ten 96-well plates in a day. As the principles of the assay are simple, many variations are possible for different applications.
Z.A.K. is a Howard Hughes Medical Institute Fellow of the Life Sciences Research Foundation. M.E.F. is an ARCS Foundation Fellow. We acknowledge funding from NIH training grant GM08284. K.M.S. received funding from the Howard Hughes Medical Institute. The research of T.B. and A.B. was supported by the Intramural Research Program of the National Institute of Child Health and Human Development of the National Institutes of Health. We thank James Hurley for the generous gift of PI(5)P4-KIIβ.
Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions