Background: The 90-kDa heat-shock proteins (Hsp90) have rapidly evolved into promising therapeutic targets for the treatment of several diseases, including cancer and neurodegenerative diseases. Hsp90 is a molecular chaperone that aids in the conformational maturation of nascent polypeptides, as well as the rematuration of denatured proteins. Discussion: Many of the Hsp90-dependent client proteins are associated with cellular growth and survival and, consequently, inhibition of Hsp90 represents a promising approach for the treatment of cancer. Conversely, stimulation of heat-shock protein levels has potential therapeutic applications for the treatment of neurodegenerative diseases that result from misfolded and aggregated proteins. Conclusion: Hsp90 modulation exhibits the potential to treat unrelated disease states, from cancer to neurodegenerative diseases, and, thus, to fold or not to fold, becomes a question of great value.
Protein kinases, the enzymes responsible for phosphoryl transfer from a chemical donor such as adenosine triphosphate (ATP) to a peptide or a protein acceptor, are integral enzymes in signaling cascades, play crucial roles in numerous cellular processes, and are of fundamental importance in systems biology. In addition, aberrant kinase activities are commonly associated with disease states, making kinases important therapeutic targets in current drug development initiatives. Therefore, understanding kinase activation dynamics is of utmost biological and clinical importance. Accurate and physiologically relevant methods to quantify kinase activities are needed to understand the intricate dynamics of kinase activation and inactivation. This chapter describes the design, evolution, and application of fluorescent-based Ser/Thr/Tyr kinase activity sensors that take advantage of chelation-enhanced fluorescence (CHEF). These sensors are compatible with physiological conditions, are selective for specific protein kinases, and provide real-time kinetic information regarding kinase activity.
Most kinase inhibitor drugs target the binding site of the nucleotide cosubstrate ATP. The high intracellular concentration of ATP can strongly affect inhibitor potency and selectivity depending on the affinity of the target kinase for ATP. Here we used a defined chemoproteomics system based on competition-binding assays in cell extracts from Jurkat and SK-MEL-28 cells with immobilized ATP mimetics (kinobeads). This system enabled us to assess the affinities of more than 200 kinases for the cellular nucleotide cofactors ATP, ADP, and GTP and the effects of the divalent metal ions Mg(2+) and Mn(2+). The affinity values determined in this system were largely consistent across the two cell lines, indicating no major dependence on kinase expression levels. Kinase-ATP affinities range from low micromolar to millimolar, which has profound consequences for the prediction of cellular effects from inhibitor selectivity profiles. Only a small number of kinases including CK2, MEK, and BRAF exhibited affinity for GTP. This extensive and consistent data set of kinase-nucleotide affinities, determined for native enzymes under defined experimental conditions, will represent a useful resource for kinase drug discovery. PMID: 23215245
Full text Figures and data Side by side Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Aurora B kinase, a key regulator of cell division, localizes to specific cellular locations, but the regulatory mechanisms responsible for phosphorylation of substrates located remotely from kinase enrichment sites are unclear. Here, we provide evidence that this activity at a distance depends on both sites of high kinase concentration and the bistability of a coupled kinase-phosphatase system. We reconstitute this bistable behavior and hysteresis using purified components to reveal co-existence of distinct high and low Aurora B activity states, sustained by a two-component kinase autoactivation mechanism. Furthermore, we demonstrate these non-linear regimes in live cells using a FRET-based phosphorylation sensor, and provide a mechanistic theoretical model for spatial regulation of Aurora B phosphorylation. We propose that bistability of an Aurora B-phosphatase system underlies formation of spatial phosphorylation patterns, which are generated and spread from sites of kinase autoactivation, thereby regulating cell division. https://doi.org/10.7554/eLife.10644.001 eLife digest Cell division is a highly organized process that involves a series of major changes. First, the cell’s chromosomes are copied and arranged at the middle of the cell. Then, the pairs of copied chromosomes are separated and pulled towards opposite ends of the cell and, finally, the cell splits in two. These steps are mainly regulated by modifications to proteins, and enzymes called protein kinases play an important role because they add phosphate groups to, or phosphorylate, so-called 'substrate' proteins to change their activities. Other enzymes called phosphatases are also important because they remove the phosphate groups from the substrates to reverse the effects. The kinase Aurora B is required for several steps during cell division and has been widely studied. This kinase is enriched in specific locations within the cell, for example at the centromere regions of the chromosomes as they line up at the cell’s center. However, Aurora B phosphorylates substrates located at distant sites on the chromosome, with less phosphorylation at sites farther from the centromere. The level of phosphorylation also changes as chromosomes become aligned. Aurora B can activate itself and this ability was suspected to help this spatiotemporal regulation. However, it was not clear how the observed gradients of kinase activity might form. Zaytsev, Segura-Peña et al. set out to answer this question by first mixing in a test tube purified Aurora B and an inhibitory phosphatase. This revealed that this kinase-phosphatase system is 'bistable', meaning that it has two stable states, low or high kinase activity, and that these states could switch in response to small changes in enzyme concentrations. Further experiments showed that this system has a kind of memory such that the level of activity (low or high) persists for a range of concentrations and depends on the system’s prior history. Zaytsev, Segura-Peña et al. then showed that both of these properties, the two stable states and the memory, exist in dividing human cells, and then went on to develop a mathematical model of how such bistability could set up gradients of Aurora B kinase activity. At the sites of highest concentration at the centromere, Aurora B can overcome inhibition by phosphatase and activates itself, as in the test tube. This activity spreads to more distant locations as active kinase molecules activate neighboring kinase molecules, establishing the areas with a high state of activity. As the local Aurora B concentration decreases further from the centromere, the phosphatase switches Aurora B into the low activity state, establishing a steep gradient of kinase activity in a region where its substrates that are important for chromosome segregation are located. Importantly, the shape and location of this gradient are predicted to depend on forces that stretch the lined chromosomes apart, offering a plausible mechanism to explain phosphorylation changes in response to tension. These theoretical insights and experimental approaches could be used to study other coupled kinase-phosphatase systems. https://doi.org/10.7554/eLife.10644.002 Introduction Aurora B, a component of the chromosomal passenger complex (CPC), is an essential kinase that is highly enriched at different intracellular locations from which it regulates cell division: it localizes initially at the inner centromere and subsequently at the anaphase spindle midzone (Carmena et al., 2012). Accumulating evidence indicates that Aurora B is capable of phosphorylating substrates that are located at a significant distance from its major binding sites. In anaphase, a long-range phosphorylation gradient is established around the spindle midzone (Fuller et al., 2008; Tan and Kapoor, 2011), but extending well beyond major sites of kinase localization (Figure 1A). This phosphorylation gradient controls the stability and length of the central spindle (Ferreira et al., 2013; Uehara et al., 2013), chromosome decondensation and nuclear envelope reassembly (Afonso et al., 2014). Similar distance-dependent phosphorylation is observed prior to anaphase onset, but at this stage Aurora B localizes to chromatin with highest concentration at the inner centromere, where CPC binding sites are enriched. During metaphase the primary targets for Aurora B, such as the microtubule-binding protein Hec1/Ndc80, are located hundreds of nanometers away at the outer kinetochore. As in anaphase, phosphorylation is lower on substrates positioned farther from Aurora B binding sites, indicating existence of a gradient of Aurora B activity (Keating et al., 2009; Liu et al., 2009; Welburn et al., 2010; DeLuca et al., 2011; Suzuki et al., 2014). Interestingly, changes of position of as little as 30–50 nm are associated with different levels of phosphorylation of both endogenous and exogenous Aurora B substrates at kinetochores (Welburn et al., 2010; Suzuki et al., 2014), indicating that the spatial regulation of Aurora B activity is very precise. Figure 1 with 1 supplement see all Download asset Open asset Spatial phosphorylation patterns in mitotic HeLa cells. (A) A HeLa cell expressing the chromatin-targeted Aurora B sensor and Aurora B-mCherry was imaged in anaphase, 10 min after addition of an Mps1 inhibitor, reversin, to increase occurrence of lagging chromosomes. The FRET ratio image shows the YFP/CFP emission ratio, color-coded as indicated. Scale bar is 5 µm. The plot shows normalized sensor phosphorylation (left axis) calculated from the FRET ratio data (see Materials and methods) and Aurora B localization signal (right axis) along the white lines which were drawn along the spindle axis in images on the left. (B) HeLa cells expressing CENP-B-FKBP, mCherry-INbox-FRB and miRNAs to deplete endogenous FKBP and INCENP, and the chromatin-targeted Aurora B sensor. Cells were treated with the kinesin-5 inhibitor STLC to generate monopolar spindles, then imaged live during rapamycin addition to induce INbox and Aurora B recruitment to centromeres. Images show INbox recruitment (bottom panels) and the YFP/CFP emission ratio (top panels) for one cell. Graph shows the FRET emission ratio averaged over chromatin in multiple cells (n≥10) treated at 3 min (arrow) with or without rapamycin. FRET ratio = 1.3 (horizontal dotted line) represents maximal Aurora B activity in cells with no INCENP depletion. The experiment was repeated three times with similar results. https://doi.org/10.7554/eLife.10644.003 One model to explain such well-controlled long-range spatial activity is by a specialized pool of Aurora B localized in close proximity to its targets (Krenn and Musacchio, 2015). At the outer kinetochore, for example, the observed gradient of substrate phosphorylation could correspond to the outermost region of the localization gradient of chromatin-bound kinase (Liu et al., 2009), or reflect the ability of Aurora B to reach these substrates by an elongated INCENP tether, the CPC component that directly binds Aurora B and is important for its mitotic functions (Samejima et al., 2015). In this view, the less abundant but proximally located Aurora B pool plays a more physiologically important role than the distant centromeric pool. Support for the kinetochore pool model comes from experiments in budding yeast, which show that the centromere localized pool of Aurora B (Ipl1) can be removed without major consequences for mitotic progression (Campbell and Desai, 2013). In several other systems, however, disrupting CPC targeting to centromeres leads to strong mitotic defects (Vader et al., 2006; Tsukahara et al., 2010; Wang et al., 2010; 2012; Yamagishi et al., 2010), suggesting that the centromere-localized pool is essential for normal cell division. An alternative model to explain how Aurora B activity is controlled at distances away from its most abundant localization sites is that this pattern depends on a biochemical crosstalk between the bound Aurora B and its cytosolic pool, which recent quantitative measurements estimate as ~25% of total Aurora B (Mahen et al., 2014). Cytosolic gradients of another mitotic regulator, RanGTP, play important roles in regulating spindle assembly (O’Connell and Khodjakov, 2007; Kalab and Heald, 2008), and a similar mechanism could contribute to long-range Aurora B activity. In this reaction-diffusion model, activation of Aurora B takes place at sites with high kinase concentration, such as the inner centromere or anaphase spindle midzone (Lampson and Cheeseman, 2011). These sites exchange quickly with a cytosolic pool (Fernández-Miranda et al., 2010), so they could serve as a source of active kinase, which has been proposed to spread to distant targets via diffusion (Fuller et al., 2008; Wang et al., 2011). However, it is not clear whether a gradient based only on the diffusion of soluble activated kinase from the inner centromere could account for changes in Aurora B substrate phosphorylation within the length scale of the kinetochore (Krenn and Musacchio, 2015). In contrast, bistable reaction-diffusion systems can in principle exhibit complex spatial patterns and support sharp boundaries of system components (Kapral and Showalter, 1995; Lobanova and Ataullakhanov, 2003; Liehr, 2013). Bistable homogeneous systems (i.e., with mixing) can switch between the alternative states, characterized by high and low activity, with no intermediate states. Furthermore, unlike in regular trigger systems, in bistable systems the high and low states can co-exist, leading to hysteresis, when the output of the system depends on its prior history (Martinov et al., 2000; Angeli et al., 2004; Tsyganov et al., 2012; Noori, 2013). Published results indicate that Aurora B kinase could in principle engage in complex non-linear behaviors. Most importantly, Aurora B can activate itself via phosphorylation of its activation loop and of a conserved TSS motif in the C-terminus of INCENP (Bishop and Schumacher, 2002; Honda et al., 2003; Yasui et al., 2004; Sessa et al., 2005; Kelly et al., 2007; Xu et al., 2010). Conversely, phosphatase can inactivate the kinase by dephosphorylating sites on Aurora B and INCENP (Sessa et al., 2005; Kelly et al., 2007; Rosasco-Nitcher et al., 2008), which could potentially help to shape the spatial gradient of Aurora B activity. Whether these reactions can lead to bistability in a coupled Aurora B-phosphatase system has not been investigated. Here, we examine the mechanisms that control Aurora B activity using cellular and simplified in vitro systems and mathematical modeling. First, we designed a novel molecular system to control Aurora B localization in cells, to directly test the importance of the centromeric pool of Aurora B in long-range activity. Second, we used purified components to reconstitute a simplified coupled Aurora B kinase-phosphatase system in vitro and showed that it exhibits bistability and hysteresis in the physiological range of Aurora B concentration. Because the complex, non-linear dynamics of reaction-diffusion systems and their spatial behavior are not intuitive, we constructed quantitative models to assist analysis of homogeneous biochemical reactions and formation of phosphorylation patterns in cells. We then developed experimental methods to analyze bistability and hysteresis of Aurora B-dependent phosphorylation in live mitotic cells, linking our biochemical findings with Aurora B regulation in cells. With these multiple approaches we provide strong evidence for a model in which spatiotemporal regulation of Aurora B is governed by a bistable reaction-diffusion mechanism. Results Concentrating Aurora B at centromeres leads to phosphorylation of distant chromatin substrates Because experiments in budding yeast have raised questions about whether concentrating Aurora B at centromeres is necessary for its mitotic function (Campbell and Desai, 2013), we designed an experiment to measure phosphorylation in live human cells while manipulating Aurora B localization with temporal control. We made a cell line that inducibly knocks down endogenous INCENP, while expressing an INbox construct that can bind and activate Aurora B (Sessa et al., 2005) but does not interact with other CPC components. The Aurora B–INbox complex is sufficient for enzymatic activity but does not localize to any particular intracellular structure because it does not form the full CPC. To control localization, we used rapamycin-based dimerization (Putyrski and Schultz, 2012), with FRB fused to INbox and FKBP fused to the centromere protein CENP-B (Figure 1—figure supplement 1A). FKBP and FRB are domains that dimerize in the presence of rapamycin. This system allows us to measure immediate effects in live cells within minutes of concentrating Aurora B at centromeres. To monitor changes in Aurora B kinase activity at a distance from sites of localization at centromeres, we used a FRET-based biosensor targeted to chromatin by fusion to histone H2B (Fuller et al., 2008). When endogenous INCENP is replaced with INbox, which is freely diffusing in the cytosol, phosphorylation is uniformly low, indicating that the cytosolic kinase pool on its own is incapable of maintaining high kinase activity along chromosome arms. Addition of rapamycin led to INbox recruitment to centromeres within minutes, accompanied by sensor phosphorylation; importantly the signal was visible all over the chromatin (Figure 1B). For these experiments cells were arrested in mitosis with a kinesin-5 inhibitor, so that chromosomes were positioned radially around a monopolar spindle with centromeres oriented toward the center (Mayer et al., 1999). With this arrangement of chromosomes, a transient phosphorylation gradient was evident extending from centromeres, similar to previous experiments in which Aurora B activity was manipulated by global inhibition (Wang et al., 2011). Similar results were observed for cells arrested with nocodazole (Figure 1—figure supplement 1B). Thus, concentrating Aurora B at centromeres of a mammalian cell is necessary and sufficient to regulate kinase activity at distal cellular locations, warranting further investigation of the kinetic mechanisms of Aurora B autoactivation. Reconstitution of Aurora B kinase autoactivation in vitro demonstrates both cis and trans components Highly concentrated centromeric kinase may become a source of active kinase for establishing spatial patterns if Aurora B can robustly activate itself in trans, i.e. intermolecularly (Sessa et al., 2005; Kelly et al., 2007; Lampson and Cheeseman, 2011). To determine the kinetic constants for Aurora B autoactivation, we measured phosphorylation in vitro in real time using purified recombinant Aurora B kinase with an INbox fragment, which is sufficient for kinase autoactivation (Sessa et al., 2005; Rosasco-Nitcher et al., 2008) (see Materials and methods and Figure 2—figure supplement 1). With purified kinase, the INCENP TSS motif, an established autophosphorylation site associated with kinase activation (Bishop and Schumacher, 2002; Honda et al., 2003; Sessa et al., 2005), was phosphorylated, as determined by immunoblotting with a phospho-specific antibody (Salimian et al., 2011; Figure 2—figure supplement 1D). This phosphorylated kinase was highly active, as shown using a chemosensor composed of a peptide containing an Aurora kinase substrate consensus site conjugated to a sulfonamido-oxine (Sox) fluorescent probe (Figure 2—figure supplement 2) (Gonzáles-Vera et al., 2009). Phosphorylation-induced increase in fluorescence of the chemosensor was followed in real time with a spectrofluorimeter, and the Michaelis–Menten, Lineweaver–Burk and Hanes–Woolf plots were analyzed (see Materials and methods), giving KM = 320 μM and kcat = 19 s-1, similar to a previous report for Aurora A kinase (Gonzáles-Vera et al., 2009). To examine activity of Aurora B in the dephosphorylated state, we incubated the kinase with λ phage phosphatase, which has previously been reported to dephosphorylate INCENP (Rosasco-Nitcher et al., 2008), and observed loss of INCENP phosphorylation (Figure 2—figure supplement 1D). Phosphonoacetic acid was then added to inhibit the phosphatase (Reiter et al., 2002) and chemosensor phosphorylation was measured. The dephosphorylated Aurora B kinase was two orders of magnitude less active than the phosphorylated Aurora B, consistent with previous studies (Eyers et al., 2005; Sessa et al., 2005), so we refer to this kinase state as partially active. Next, we sought to determine the kinetic parameters of Aurora B autoactivation. At 10–30 nM of partially active kinase, chemosensor phosphorylation was barely detected. This finding is consistent with our results using INbox replacement in cells with no rapamycin, since this low concentration range was reported for cytosolic Aurora B (Mahen et al., 2014). At 0.16–1.5 µM kinase, chemosensor phosphorylation increased nonlinearly with time, indicating autoactivation (Figure 2A, Figure 2—figure supplement 2G). Previous studies have reported that this autoactivation takes place in trans (Sessa et al., 2005; Rosasco-Nitcher et al., 2008) (Figure 2B), predicting that the coefficient for this increase vs. kinase concentration is close to 2 when plotted on a logarithmic scale. The measured slope in our experiments with low kinase concentrations was 1.23 ± 0.02 (Figure 2C), implying that the partially active Aurora B can activate itself in cis, i.e. intramolecularly (Figure 2B). Figure 2 with 3 supplements see all Download asset Open asset Aurora B kinase autoactivation in vitro. (A) Phosphorylation of 20 µM chemosensor by the indicated concentrations of partially active Aurora B kinase. Data are averages of N = 2 experiments for each kinase concentration; error bars are SEMs. Black lines are theoretical fittings with the reaction scheme in panel E. (B) Molecular scheme for Aurora B autoactivation in trans or in cis. A and A* denote partially active (dephosphorylated) and active kinase; S and P indicate substrate and product (unphosphorylated and phosphorylated chemosensors, respectively). (C) Coefficient k for the quadratic phase of chemosensor phosphorylation by partially active Aurora B kinase vs. kinase concentration (A) plotted on a log-log scale. Line is linear fit. (D) Diagram of the experimental procedure to evaluate Aurora B autoactivation at high kinase concentration (4 µM). Experimental graph on the right shows changes in concentration of active Aurora B, calculated as described in Materials and methods. Data points are mean ± SEM for N≥4 experiments. Solid line is theoretical fitting with the reaction scheme in panel E. Dashed line is theoretical fit using the analytical solution for A*(t) for the reaction scheme with only in cis activation of Aurora B. (E) Molecular scheme for the Aurora B kinase two component autoactivation in the presence of chemosensor and the corresponding reactions, see system equation 2 in Materials and methods. All other symbols are listed in Tables 1 and 2. https://doi.org/10.7554/eLife.10644.005 To reveal the in trans component, we carried out experiments using high concentration of partially active Aurora B, mimicking its clustering at cellular binding sites. At high kinase concentration the chemosensor becomes depleted quickly, so we modified our assay to uncouple the Aurora B autophosphorylation reaction from the activity measurement with the chemosensor (Figure 2D). With 4 µM kinase, kinase activity increased strongly with time, and the best-fit curve based only on in cis autoactivation provided a poor fit (Figure 2D), confirming the presence of the in trans component. With a computational model combining both reactions (Figure 2E), we generated a global fit to experimental curves in Figure 2A,D and determined molecular constants for the two-component autoactivation mechanism for Aurora B kinase (Table 2, Materials and methods). This model demonstrates that kinase autoactivation in cis dominates over the trans-activation during initial activation at low kinase concentration (Figure 2—figure supplement 2, panels H and I). A coupled Aurora B kinase-phosphatase system exhibits bistability and hysteresis in silico Our findings above imply that if Aurora B kinase, phosphatase and ATP are mixed together, two reactions should take place simultaneously: Aurora B autoactivation and its inactivation by phosphatase. We constructed a quantitative model for such a coupled kinase-phosphatase system (Figure 3A), which takes into account the determined molecular constants for two-component Aurora B autoactivation and a Michaelis–Menten mechanism for a phosphatase with variable enzymatic constants. Solving the differential equations describing this system in silico (see Materials and methods) reveals that at high kinase concentration three steady-state solutions could coexist (Figure 3B). Figure 3C shows region of bistability in the parametric plane of Aurora B kinase-phosphatase concentrations. Bistability arises when Aurora B kinase concentration exceeds 4 μM, and further increasing Aurora B concentration broadens the range of permissible phosphatase concentrations. In this region, a homogeneous mixture of kinase and phosphatase can exist in one of two stable states with different kinase activity, high or low, depending on initial conditions (Figure 3C). This prediction is important because, as we will show later, bistable behavior is essential for accurate regulation of Aurora B kinase activity away from sites of high kinase concentration. Figure 3 with 1 supplement see all Download asset Open asset Theoretical analysis of the coupled Aurora B kinase-phosphatase system. (A) Molecular scheme for the coupled system and the corresponding reactions. For reactions 1 and 2 see Figure 2E; see Table 1 and 2 for more details. (B) Steady-state solutions for concentration of active Aurora B kinase as a function of phosphatase concentration (Equation 4 in Materials and methods). For 8 µM Aurora B, three steady states can co-exist: two stable states with high and low activities and one unstable state (dashed line), corresponding to the region of bistability. (C) Bistability region in the parametric plane of phosphatase and total (phosphorylated and not) Aurora B kinase concentrations. In this region the model has two coexisting stable steady-state solutions, while enzymatic concentrations outside this region lead to only one steady state. Colored lines correspond to the solutions shown in panel B for active kinase. (D) Theoretical predictions for the changes in concentration of active Aurora B kinase, plotted as a fraction of total kinase concentration, for two different initial conditions. The initial concentration slightly higher than the threshold (horizontal line) has a steady-state solution with a larger fraction of active kinase (high state). The fraction of active Aurora B kinase declines when its initial concentration is below the threshold (low state). Calculations were done for 8 µM total Aurora B kinase and 0.47 µM phosphatase. (E) Simulation of perturbations to reaction with 0.47 µM phosphatase and 8 µM total Aurora B kinase. Active kinase is added 3 times as indicated (vertical arrows). The system returns to the steady state with low Aurora B kinase activity until the threshold is exceeded. (F) Hysteresis loop in the kinase-phosphatase system with 8 µM kinase. Phosphatase concentration was initially low, so almost all Aurora B kinase was active. As the phosphatase concentration was gradually increased up to 0.8 µM, the steady-state concentration of active Aurora B kinase decreased (top line with downward arrow). Different solutions were obtained when phosphatase concentration was decreased gradually back to 0 µM (lower line with two upward arrows). https://doi.org/10.7554/eLife.10644.009 As expected for the bistable regime, increasing concentration of active Aurora B above a threshold causes this biochemical system to switch between two states with no intermediate steady-states (Figure 3D,E). The model also predicts hysteresis in the region of bistability. Hysteresis becomes evident at intermediate levels of phosphatase concentration (e.g. 0.4 μM in Figure 3F), when almost the entire kinase pool can be either phosphorylated (high kinase activity state) or dephosphorylated (low kinase activity state) depending on the prior state of this reaction mixture. Importantly, we find that these non-linear regimes are determined mostly by the parameters of the two component Aurora B autoactivation mechanism, but not by the enzymatic constants of the protein phosphatase (Figure 3—figure supplement 1; see Materials and methods). Thus, Aurora B kinase, coupled with an inactivating phosphatase, is predicted to exhibit robust hysteresis and bistability. Aurora B kinase-phosphatase bistability and hysteresis observed in vitro are in a quantitative agreement with theoretical predictions Using the reconstituted in vitro system, we next designed an experiment to test the prediction of our theoretical model that at high Aurora B concentration the same mixture of kinase and phosphatase will result in different degrees of Aurora B activity depending on the initial conditions. We combined Aurora B kinase (8 μM), ATP (4 mM) and variable concentrations of λ phage phosphatase, such that both Aurora B activation and inactivation could take place simultaneously. Importantly, these reactions were carried out for two different initial conditions: using either the active kinase or active kinase pretreated with phosphatase (Figure 4A). All other reactants in these two mixtures were adjusted to achieve the same final concentrations for all components, including the phosphatase. The progress of these reactions was followed by taking samples at the indicated times; phosphatase inhibitor was then added to stop kinase dephosphorylation and kinase activity was measured via the initial rate of chemosensor phosphorylation. As expected, at low phosphatase concentration (0.25 µM) the partially active kinase gradually activated itself, reaching a steady-state with high activity, while the active kinase slightly lost its activity (Figure 4B, top graph). At high phosphatase concentration (0.5 µM, bottom graph) the active kinase was overpowered by the phosphatase and became gradually inactivated, reaching a level close to the fully dephosphorylated kinase. Importantly, the model accurately predicted the behavior for these two reactions and when intermediate phosphatase concentration was used (lines in Figure 4B). At 0.45 µM phosphatase, the kinase that was initially active robustly retained its high state (red data points, Figure 4B, middle graph), while the activity remained low for the kinase that was initially in the low state (blue data points, Figure 4B middle graph). These outcomes demonstrate bistability because in both of these enzyme mixtures the final concentrations of all components were identical and the reactions were allowed to proceed long enough to reach the steady-states (120 min). Similar experiments were carried out for additional phosphatase concentrations, and the steady-state levels of active Aurora B were obtained by averaging measurements for ≥60 min incubation times. These data, plotted as a function of phosphatase concentration in Figure 4C, define the bistable region for the homogeneous system in vitro, in quantitative agreement with model predictions. In this range of concentrations, the coupled kinase-phosphatase system exhibits hysteresis, with different activity levels observed depending on the initial conditions. Figure 4 Download asset Open asset Reconstitution of the coupled Aurora B kinase-phosphatase system in vitro. (A) Diagram of the experimental procedure to study bistability and hysteresis. Active kinase was preincubated with phosphatase (PPase) in the absence of ATP to generate partially active kinase, and ATP was added at time = 0 ('initially low' experiment). In a parallel experiment, the same reagents were used but active kinase, phosphatase and ATP were mixed together at time = 0 ('initially high' experiment). Samples were taken to analyze kinase activity until the corresponding steady states were reached. (B) Experimental results (dots) for changes in kinase activity vs. incubation time for 8 µM kinase
Heat shock protein 90 (Hsp90) represents a promising therapeutic target for the treatment of cancer and other diseases. Unfortunately, results from clinical trials have been disappointing as off-target effects and toxicities have been observed. These detriments may be a consequence of pan-Hsp90 inhibition, as all clinically evaluated Hsp90 inhibitors simultaneously disrupt all four human Hsp90 isoforms. Using a structure-based approach, we designed an inhibitor of Grp94, the ER-resident Hsp90. The effect manifested by compound 2 on several Grp94 and Hsp90α/β (cytosolic isoforms) clients were investigated. Compound 2 prevented intracellular trafficking of the Toll receptor, inhibited the secretion of IGF-II, affected the conformation of Grp94, and suppressed Drosophila larval growth, all Grp94-dependent processes. In contrast, compound 2 had no effect on cell viability or cytosolic Hsp90α/β client proteins at similar concentrations. The design, synthesis, and evaluation of 2 are described herein.
Compound 2 (KU-32) is a first-generation novologue (a novobiocin-based, C-terminal, heat shock protein 90 (Hsp90) inhibitor) that decreases glucose-induced death of primary sensory neurons and reverses numerous clinical indices of diabetic peripheral neuropathy in mice. The current study sought to exploit the C-terminal binding site of Hsp90 to determine whether the optimization of hydrogen bonding and hydrophobic interactions of second-generation novologues could enhance neuroprotective activity. Using a series of substituted phenylboronic acids to replace the coumarin lactone of 2, we identified that electronegative atoms placed at the meta-position of the B-ring exhibit improved cytoprotective activity, which is believed to result from favorable interactions with Lys539 in the Hsp90 C-terminal binding pocket. Consistent with these results, a meta-3-fluorophenyl substituted novologue (13b) exhibited a 14-fold lower ED50 for protection against glucose-induced toxicity of primary sensory neurons compared to 2.
The Hsp90 chaperone machine is required for the folding, activation, and/or stabilization of more than 50 proteins directly related to malignant progression. Hsp90 contains small molecule binding sites at both its N- and C-terminal domains; however, limited structural and biochemical data regarding the C-terminal binding site is available. In this report, the small molecule binding site in the Hsp90 C-terminal domain was revealed by protease fingerprinting and photoaffinity labeling utilizing LC–MS/MS. The identified site was characterized by generation of a homology model for hHsp90α using the SAXS open structure of HtpG and docking the bioactive conformation of NB into the generated model. The resulting model for the bioactive conformation of NB bound to Hsp90α is presented herein.
