CHIP participates in protein triage decisions by preferentially ubiquitinating Hsp70-bound substrates Marta Stankiewicz 1 , Rainer Nikolay 1, *, Vladimir Rybin 2 and Matthias P. Mayer 1 1 Zentrum fu ¨ r Molekulare Biologie der Universita ¨ t Heidelberg (ZMBH), DKFZ–ZMBH Alliance, Heidelberg, Germany 2 European Molecular Biology Laboratory, Heidelberg, Germany Introduction CHIP consists of an N-terminal TPR (tetratricopeptide repeat) domain, which binds to the C-terminal EEVD motif that is present in all cytosolic Hsp70 and Hsp90 chaperones, a central helical domain, which is essential Keywords chaperones; Hsp70; Hsp90; protein triage; ubiquitination Correspondence M. P. Mayer, Zentrum fu ¨ r Molekulare Biologie der Universita ¨ t Heidelberg (ZMBH), DKFZ–ZMBH Alliance, Im Neuenheimer Feld 282, 69120 Heidelberg, Germany Fax: +49 6221 545894 Tel: +49 6221 546829 E-mail: M.Mayer@zmbh.uni-heidelberg.de *Present address Biochemisches Institut der Universita ¨ t Zu ¨ rich, Winterthurerstrasse 190, 8057 Zu ¨ rich, Switzerland (Received 3 May 2010, revised 8 June 2010, accepted 14 June 2010) doi:10.1111/j.1742-4658.2010.07737.x The E3 ubiquitin ligase CHIP (C-terminus of Hsc70-interacting protein) is believed to be a central player in the cellular triage decision, as it links the molecular chaperones Hsp70 ⁄ Hsc70 and Hsp90 to the ubiquitin proteaso- mal degradation pathway. To better understand the decision process, we determined the affinity of CHIP for Hsp70 and Hsp90 using isothermal titration calorimetry. We analyzed the influence of CHIP on the ATPase cycles of both chaperones in the presence of co-chaperones and a substrate, and determined the ubiquitination efficacy of CHIP in the presence of the chaperones. We found that CHIP has a sixfold higher affinity for Hsp90 compared with Hsc70. CHIP had no influence on ADP dissociation or ATP association, but reduced the Hsp70 cochaperone Hdj1-stimulated sin- gle-turnover ATPase rates of Hsc70 and Hsp70. CHIP did not influence the ATPase cycle of Hsp90 in the absence of co-chaperones or in the pres- ence of the Hsp90 cochaperones Aha1 or p23. Polyubiquitination of heat- denatured luciferase and the native substrate p53 was much more efficient in the presence of Hsc70 and Hdj1 than in the presence of Hsp90, indicat- ing that CHIP preferentially ubiquitinates Hsp70-bound substrates. Structured digital abstract l MINT-7904367: CHIP (uniprotkb:Q9UNE7) and HSP 90-beta (uniprotkb:P08238) physically interact ( MI:0915)bymolecular sieving (MI:0071) l MINT-7904785: HSP 90-beta (uniprotkb:P08238) and p23 (uniprotkb:Q15185) bind ( MI:0407)bymolecular sieving (MI:0071) l MINT-7904047: CHIP (uniprotkb:Q9UNE7), HSP 90-beta (uniprotkb:P08238) and p23 (uni- protkb: Q15185) physically interact (MI:0915)bymolecular sieving (MI:0071) l MINT-7903424: Alpha-lactalbumin (uniprotkb:P00711), HSP70 (uniprotkb:P08107) and CHIP (uniprotkb: Q9UNE7) physically interact (MI:0915)bymolecular sieving (MI:0071) l MINT-7903354: CHIP (uniprotkb:Q9UNE7) and HSC70 (uniprotkb:P11142) bind (MI:0407) by isothermal titration calorimetry ( MI:0065) l MINT-7903373: CHIP (uniprotkb:Q9UNE7) and HSP90-beta (uniprotkb:P08238) bind ( MI:0407)byisothermal titration calorimetry (MI:0065) Abbreviations CHIP, C-terminus of Hsc70-interacting protein; Hsp70, 70 kDa heat shock protein; Hsp90, 90 kDa heat shock protein; Hsc70, 70 kDa heat shock cognate; ITC, isothermal titration calorimetry; MABA-ADP ⁄ MABA-ATP, N 8 -(4-N¢-methylanthraniloylaminobutyl)-8-aminoadenosine 5¢-di ⁄ triphosphate; RCMLA, reduced carboxymethylated a-lactalbumin; TPR, tetratricopeptide repeat. FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3353 for dimerization, and a C-terminal U-box domain, which is responsible for interaction with E2 ubiquitin- conjugating enzymes [1,2]. This interaction with molec- ular chaperones suggests that CHIP is involved in the triage decision [3–5]. Hsp70 chaperones are essential components of the cellular quality control network, interacting with virtually all misfolded proteins, pre- venting their aggregation and assisting their refolding into the native state [6–8]. Hsp90 chaperones have also been shown to bind to misfolded proteins [9], but their main essential function in all eukaryotic cells is believed to be interaction with a large number of regu- latory proteins, called client proteins, including recep- tors, protein kinases and transcription factors [10–12]. The question therefore arises as to whether CHIP pref- erentially interacts with Hsp70 to mark misfolded proteins, which cannot be refolded to the native state, for degradation, or whether it assumes a more regulatory role by interacting mainly with Hsp90 to ubiquitinate signaling proteins. It has been shown in vitro and in cell culture that sev- eral Hsp90 clients are ubiquitinated in a CHIP-depen- dent manner, and that this ubiquitination depends on CHIP’s U-box domain [13–17]. In some cases, CHIP appears to directly bind its substrates and ubiquitinate them in a chaperone-independent manner [18–20]. Such specific substrate recognition is a typical feature of E3 ubiquitin ligases; however, the majority of CHIP’s sub- strates described so far are bona fide Hsp70 and Hsp90 clients, and their degradation depends on the presence of the chaperones [13,21–23]. Among the CHIP substrates are many regulatory proteins that are ubiqui- tinated and degraded even in the presence of an Hsp90 inhibitor such as geldanamycin [21]. For most if not all functions, Hsp90 cooperates with Hsp70 in a chaperone cycle, which was first proposed for steroid hormone receptors and involves a number of co-chaperones [24,25]. Steroid hormone receptors first interact with Hsp40 and Hsp70. The dimeric pro- tein Hop (Hsp70-Hsp90 organizing Protein), which has separate TPR domains for binding to Hsp70 (TPR1) and Hsp90 (TPR2a), assembles the early client com- plex with Hsp70 and Hsp90. Hop and Hsp70 are then replaced by p23 and a TPR domain-containing pept- idyl-prolyl-cis ⁄ trans-isomerase (e.g. the 51 and 52 kDa FK506-binding proteins FKBP51 or FKBP52). The mature complex decays with a half life of approxi- mately 5 min, and the hormone receptor re-enters the cycle by binding to Hsp40 and Hsp70. As CHIP can interact with Hsp70 and Hsp90, it is not clear whether ubiquitination of the chaperone substrate occurs while the substrate is bound to Hsp70 or Hsp90. Another intriguing question is how the triage decision is made. As CHIP is a TPR-containing co-chaperone, it com- petes with numerous other TPR-containing co-chaper- ones for binding to Hsp70 and Hsp90 [26]. Here we provide new insights into the triage decision by assessing the physical interaction of CHIP with Hsp70 ⁄ Hsc70 and Hsp90, and analyzing the functional consequences for the chaperone substrates. Results CHIP–chaperone interaction and competition with Hop To determine how the triage decision is made, we first addressed the question of protein affinities and cellular concentrations. CHIP directs proteins to the degrada- tion pathway, and Hop is an essential co-chaperone for protein folding. Because they both interact with the same C-terminal EEVD motif of Hsc70 and Hsp90, we determined the affinities of CHIP for Hsc70 and Hsp90 using isothermal titration calorimetry (ITC) ( Fig. 1A,B). We also investigated the interaction of CHIP with heat shock-induced Hsp70, to determine whether CHIP has a preference for this homolog to enhance quality control processes during heat shock. CHIP’s affinity for Hsp90 (K D = 0.38 ± 0.04 lm) was approximately six times higher than that for Hsc70 (K D = 2.3 ± 0.3 lm), and two and a half times higher than that for Hsp70 (K D = 0.95 ± 0.01 lm). The affinities of CHIP for Hsc70 and Hsp70 were in the same range as the value measured for the Hop– Hsc70 interaction (K D = 1.5 ± 0.2 lm) using surface plasmon resonance spectroscopy [27]. In contrast, for the interaction of Hop with Hsp90, a K D value of 0.1 ± 0.02 lm was determined by surface plasmon resonance spectroscopy, which is only one quarter of the value measured here for the CHIP–Hsp90 interac- tion. These results indicate that Hop and CHIP com- pete efficiently with each other for binding to Hsc70 ⁄ Hsp70 when the C-termini of Hsc70⁄ Hsp70 are limiting when the concentration of Hsc70/Hsp70 is lower than the combined concentration of CHIP and Hop, but Hop appears to be at an advantage com- pared to CHIP when binding to Hsp90. The cellular concentrations of Hsc70⁄ Hsp70, Hsp90, CHIP and Hop where determined by quantitative Wes- tern blot using HEK293, a commonly used epithelial cell line, and Jurkat cells, which are a model for acute T-cell leukemia (Fig. 1C and Table 1). The values determined for Hsc70 ⁄ Hsp70 (0.9 and 0.4% of total protein for HEK293 and Jurkat cells, respectively) and Hsp90 (0.6 and 0.8% of total protein for HEK293 and Jurkat cells, respectively) were somewhat lower than CHIP preferentially ubiquitinates Hsp70 substrates M. Stankiewicz et al. 3354 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS the values of 1–2% reported for other cell lines [27a]. For Hop, we determined a relative amount of 0.2% in both cell lines. In contrast, the relative amount of CHIP varied significantly, being 0.07% in HEK293 and only 0.01% in Jurkat cells. As the total protein concentration was 154 mgÆmL )1 in HEK293 cells [28] and 127 mg ÆmL )1 in Jurkat cells [29], the concentra- tions of Hsc70 ⁄ Hsp70 are 20 and 7 lm, those of Hsp90 are 11 and 12 lm, those of Hop are 5 and 4 lm, and those of CHIP are 3.1 and 0.4 lm in HEK293 and Jurkat cells, respectively. Given these concentrations of Hsc70, Hsp90, CHIP and Hop and the K D values, approximately 8.5 and 1.4% of Hsc70 and 4.6 and 0.7% of Hsp90 molecules have CHIP bound to their C-termini at any given moment in HEK293 and Jurkat cells, respectively. As there are a large number of TPR domain proteins in higher eukaryotic cells, many of which bind to Hsp90 with a similar affinity as CHIP and Hop do [30,31], the amount of Hsp90 occupied by CHIP is probably much lower than estimated above. Fewer TPR proteins have been shown to bind Hsc70 ⁄ Hsp70 [26]. Therefore, based on our affinity determination and quantitative Western blots, the amount of Hsc70 occupied by CHIP is estimated to be 1–9% (in Jurkat and HEK293 cells). Influence of CHIP on substrate binding of Hsp70 RING and U-box E3 ligases do not transfer ubiquitin themselves but generally bring substrates and E2 ubiquitin-conjugating enzymes in close proximity by binding to both proteins. It has been shown that CHIP has the ability to bind substrates [18–20]. If CHIP also contacts substrates when bound to Hsp70, it might increase the stability of the Hsp70–substrate complex, thereby allowing more time for ubiquitin transfer by the E2 enzyme. We therefore assessed whether CHIP affects the equilibrium dissociation constant (K D ) or the dissocia- tion rate constant (k off ) of the Hsp70–substrate complex by analyzing the formation of complexes of Hsp70 with reduced carboxymethylated a-lactalbumin Fig. 1. Interaction of CHIP with Hsc70 and Hsp90 and in vivo concentrations of the chaperones and co-chaperones. (A,B) Deter- mination of the interaction parameters of the CHIP–Hsc70 (A) and CHIP–Hsp90 (B) complexes using isothermal titration calorim- etry. (C) Quantitative immunoblot for deter- mination of the in vivo concentrations of Hsp70 ⁄ Hsc70, Hsp90, CHIP and Hop in HEK293 and Jurkat cells. Various amounts of purified protein (15–400 ng, left panels) and cleared protein extracts (10–100 lg) of HEK293 (middle panels) and Jurkat cells (right panels), as indicated, were separated by SDS ⁄ PAGE and analyzed by immunoblot- ting with specific antisera. The upper bands detected in vivo for CHIP and Hop most likely represent phosphorylated variants of the proteins [72,73]. Table 1. Relative amounts of CHIP, Hop, Hsp70 and Hsp90 in HEK and Jurkat cells. The relative amounts of chaperone and co-chaperones were determined by quantitative immunoblotting as shown in Fig. 1C using purified proteins as standards. Percentage of total protein HEK Jurkat CHIP 0.072 ± 0.014 0.01 ± 0.0002 Hop 0.20 ± 0.01 0.18 ± 0.0001 Hsp70 0.94 ± 0.001 0.38 ± 0.007 Hsp90 0.60 ± 0.01 0.80 ± 0.05 M. Stankiewicz et al. CHIP preferentially ubiquitinates Hsp70 substrates FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3355 (RCMLA), a model chaperone substrate [32], in the presence and absence of CHIP using gel filtration and 3 H-RCMLA (Fig. 2A,B). When Hsp70 was pre-incu- bated with CHIP, the amount of RCMLA bound to Hsp70 decreased by approximately 40% (Fig. 2B). This was not observed with the CHIP-K30A variant, which does not bind to the C-terminal EEVD motif of Hsp70 and Hsp90 [21], suggesting that CHIP does not compete for Hsp70’s substrate binding pocket but affects the chaperone–substrate interaction in an indi- rect way. This result clearly indicates that CHIP does not prolong the half life of the high-affinity Hsp70– chaperone complex. To investigate substrate release, we chased the pre- formed Hsp70– 3 H-RCMLA complexes with unlabeled RCMLA in the absence and presence of CHIP. As shown in Fig. 2C, CHIP did not influence substrate release by Hsp70, suggesting that the decrease in detectable Hsp70–RCMLA complex is due to a decreased association rate. CHIP bound to the C-ter- minus of Hsp70 possibly creates a steric hindrance to substrate binding. We did not observe any direct inter- action of CHIP with the chaperone substrate RCMLA, suggesting that substrate binding by CHIP may be a specific interaction limited to certain proteins. In con- clusion, CHIP did not increase the half life of the Hsp70–RCMLA complex by directly stabilizing the chaperone–substrate interaction, but instead decreased the amount of Hsp70-bound RCMLA by 40%. Influence of CHIP on the ATPase cycle of Hsp70 As nucleotide exchange by Hsp70 is rate-limiting for substrate release under physiological ATP concentra- tions, CHIP could also affect the half-life of the Hsp70–substrate complex by altering the ATPase cycle of Hsp70 proteins. It has been reported that CHIP decreases the ATPase rate stimulated by the J-domain co-chaperones Hdj1 and Hdj2 but not the basal ATPase rate under steady-state conditions [14,33]. In the absence of a J-domain co-chaperone, c-phosphate cleavage is rate-limiting in the ATPase cycle of Hsp70 proteins [34,35]. In the presence of a J-domain protein, nucleotide exchange becomes rate-limiting [36,37]. The CHIP-induced reduction of the Hdj1 ⁄ Hdj2-stimulated ATPase rate of Hsc70 could therefore be caused by a reduced nucleotide exchange, which in turn would increase the dwell time of the substrate on the Hsp70 chaperone. To address this point, we analyzed the influence of CHIP on ADP dissociation from and ATP association with Hsc70 and Hsp70 using the fluorescent nucleotide analogs N 8 -(4-N¢-methylanthra- niloylaminobutyl)-8-aminoadenosine 5¢-di ⁄ triphosphate (MABA-ADP ⁄ MABA-ATP) [38] and stopped-flow instrumentation. To measure the basal ADP dissocia- tion rate, Hsc70 or Hsp70 were pre-incubated with MABA-ADP in the absence or presence of a 20-fold excess of CHIP, and subsequently mixed with an Fig. 2. CHIP reduces the affinity of Hsp70 for a model substrate without affecting the dissociation rate. (A) Size-exclusion chroma- tography of 3 H-RCMLA (reduced carboxymethylated a-lactalbumin) after pre-incubation in the absence or presence of Hsp70 and CHIP as indicated. (B) Quantification for the size-exclusion chromatogra- phy experiments shown in (A). 3 H-RCMLA was pre-incubated with the indicated proteins. The amount of radioactivity in elution volume 9–11.5 mL, in which the RCMLA–Hsp70 complex elutes, is shown relative to the radioactivity in elution volume 12–15.5 mL, in which free RCMLA elutes. (C) Dissociation of the RCMLA–Hsp70 com- plex. 3 H-RCMLA was pre-incubated with Hsp70 before addition of CHIP where indicated. At time point 0, a fivefold excess of unla- beled RCMLA was added. The complex was analyzed by size-exclu- sion chromatography at various time points. The dissociation rate constants (k off ) were determined by fitting a single exponential decay function to the data. The inset shows the dissociation rate constants in the absence and presence of CHIP (mean ± SEM of two independent determinations). CHIP preferentially ubiquitinates Hsp70 substrates M. Stankiewicz et al. 3356 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS excess of ATP. As shown in Fig. 3A, CHIP had no influence on the basal ADP dissociation rates of Hsc70 and Hsp70. These results did not explain the CHIP- mediated decrease in the Hdj1⁄ Hdj2-stimulated steady- state ATPase rate. In vivo nucleotide exchange factors such as Bag-1 accelerate ADP dissociation by Hsc70 and Hsp70 [37,39]. It has been reported that CHIP and Bag-1 interact with each other [40]. We therefore determined whether CHIP could influence Bag-1-stimulated nucle- otide exchange. To address this question, we pre-incu- bated Hsc70 ⁄ Hsp70 with MABA-ADP in the absence and presence of a large excess of CHIP, and subse- quently rapidly mixed the reaction mixture with Bag-1 and an excess of ATP. As expected, Bag-1 stimulated the ADP dissociation rate by approximately 20-fold at stoichiometic concentrations. Even a large excess of CHIP only slightly decreased the Bag-1-stimulated ADP dissociation rate of Hsc70 and Hsp70 (Fig. 3A). No effect of CHIP on the Bag-1-stimulated ADP dis- sociation rate was observed when CHIP was added together with Bag-1 instead of pre-incubated with the Hsp70 protein (data not shown). Therefore, the reported interaction of CHIP and Bag-1 has no strik- ing effect on the nucleotide release function of Bag-1. To analyze the second step of nucleotide exchange, ATP association, we pre-incubated Hsc70 or Hsp70 in the absence and presence of Bag-1 and a 20-fold excess of CHIP, and subsequently mixed the reaction mixture with MABA-ATP. As shown in Fig. 3B,C, CHIP did not slow down ATP association significantly in the absence or presence of Bag-1. Instead we observed a slight increase in ATP association rate for Hsc70 in the presence of CHIP. As neither ADP dissociation nor ATP association are negatively affected by CHIP, the reduction in the ATPase activity must be due to an effect on c-phos- phate cleavage. To verify this hypothesis, we performed single-turnover ATPase experiments. The basal ATPase rate of Hsp70 proteins is very low but can be stimulated by a J-domain co-chaperone at high concentrations (> 10-fold). As shown in Fig. 3D, high concentrations of CHIP had no effect on the basal single-turnover ATPase rate but decreased the Hdj1-stimulated ATPase Fig. 3. CHIP affects Hdj1-stimulated c-phosphate cleavage by Hsc70 ⁄ Hsp70 but not nucleotide exchange. (A) MABA-ADP dissoci- ation rates of Hsc70 and Hsp70 in the absence and presence of CHIP and Bag-1. (B) Fluorescence traces of MABA-ATP association with Hsc70 in the absence and presence of CHIP and Bag-1. (C) Association rates for MABA-ATP in the absence and presence of Bag-1 and CHIP. The columns show the rates for the fast phase (k 1 ) and the slow phase (k 2 ) of a fit of a two-phase exponential equation to the traces in (B) and additional data. (D) Single-turnover ATPase rates of Hsc70 and Hsp70 in the absence and presence of Hdj1 and CHIP as indicated. M. Stankiewicz et al. CHIP preferentially ubiquitinates Hsp70 substrates FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3357 rate, consistent with previous steady-state ATPase data [14,33]. Taken together, we found no evidence that CHIP influences the chaperone cycle of Hsp70 proteins to prolong the life-time of the substrate–Hsp70–CHIP complex and thereby to increase the possibility of recruitment of the E2 ubiquitin-conjugating enzyme and ubiquitination. In contrast, CHIP decreased the Hdj1-triggered c-phosphate cleavage, thereby decelerat- ing transition from the low-affinity to the high-affinity substrate-binding state. Influence of CHIP on the ATPase activity and co-chaperone binding of Hsp90 We next determined whether CHIP influences the ATPase cycle of Hsp90 in order to increase the possibil- ity of ubiquitination of an Hsp90-bound client protein. We performed steady-state ATPase assays of Hsp90 in the absence and presence of CHIP, and in the absence and presence of Aha1 and p23, two co-chaperones that are known to influence the ATPase activity of Hsp90. We obtained a value of 1.2 ± 0.1 · 10 )3 s )1 for the basal ATPase activity of Hsp90. This rate was stimu- lated fivefold by a threefold excess of Aha1 over Hsp90, and inhibited to 50% of the basal rate by a tenfold excess of p23, consistent with published data [41,42]. As shown in Fig. 4, CHIP did not significantly affect the basal ATPase rate of Hsp90, and also had no influence on the Aha1-stimulated or p23-inhibited rate, even at a tenfold excess over Hsp90. A previous study suggested that p23 competes with CHIP for binding to Hsp90 [13]. As CHIP did not reduce the inhibitory effect of p23 on the ATPase activity of Hsp90, we determined whether this is due to the inability of CHIP to bind to Hsp90 in the pres- ence of p23. We therefore incubated Hsp90 with CHIP and p23 and used gel filtration to analyze the com- plexes formed. As evident from Figs 5A,B and S1, CHIP forms a stable complex with Hsp90 and p23 and does not prevent p23 binding to Hsp90. In contrast, Hop reduced binding of p23 to Hsp90, consistent with previous observations [43]. Aha1 also reduced binding of p23 to Hsp90, and CHIP could not reverse this effect of Aha1. Taken together, CHIP did not reduce the ATP hydrolysis rates of Hsp90. As ATP hydrolysis leads to substrate release [44], CHIP should not increase the half-life of Hsp90–client complexes. Unfolded proteins are more efficiently ubiquitinated in the presence of the Hsp70 system As CHIP interacts with both Hsp70 and Hsp90 and the interaction is mutually exclusive, we wished to directly compare the two chaperone systems in terms of their influence on the efficiency of CHIP-mediated ubiquitination of a substrate. It has already been shown, that both systems are able to support ubiqui- tination in vitro, but quantitative time-resolved ubiq- uitination experiments are necessary to compare their relative efficiency. We pre-incubated the chaperone substrate firefly luciferase in the presence of various concentrations of Hsc70 plus Hdj1 or Hsp90 at 43 °C, and subsequently shifted the temperature to 30 °C before adding CHIP, the E2 enzyme UbcH5c, the E1 ubiquitin-activating enzyme and ubiquitin. In the presence of Hsc70 and Hdj1, ubiquitination was very efficient even at low chaperone:luciferase ratios Fig. 4. CHIP has no influence on the ATPase rate of Hsp90. (A) Steady-state ATPase rate of Hsp90 in the absence or presence of the indicated concentrations of CHIP. (B) Steady-state ATPase rate of Hsp90 in the absence or presence of the indicated concentra- tions of p23 and CHIP. (C) Steady-state ATPase rate of Hsp90 in the absence or presence of the indicated concentrations of Aha1 and CHIP. CHIP preferentially ubiquitinates Hsp70 substrates M. Stankiewicz et al. 3358 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS (1 : 1), but not in the absence of Hdj1 (Fig. 6A,B, left panel). In contrast, ubiquitination in the presence of Hsp90 was not very efficient and required high con- centrations of chaperone. This indicates that many more substrate molecules can be successfully ubiquiti- nated per one Hsc70 ⁄ CHIP complex than per one Hsp90 ⁄ CHIP complex. Time-resolved experiments also showed that CHIP-mediated polyubiquitination was faster in the presence of the Hsc70 ⁄ Hdj1 system than in the presence of Hsp90 (Fig. 6B, right panel). Interestingly, when we used the lysine-free variant of ubiquitin (Ubi-K0) to prevent polyubiquitination, we also detected multiple bands of luciferase in the pres- ence of Hsc70 and Hsp90, indicating that both chap- erones allow attachment of ubiquitin to several lysines of luciferase. These data suggest that, several lysines of the substrate are modified even in the pres- ence of wild-type ubiquitin. As multiple bands of ubiquitinated luciferase are already visible at the 20 min time point in the case of wild-type ubiquitin but appear later in the case of the K0 ubiquitin vari- ant, polyubiquitination may occur in the presence of Hsp70 with a certain processivity, or alternatively the lysines in ubiquitin (presumably Lys48) are better substrates for ubiquitination than lysines in the sub- strate. In addition, the co-chaperones of Hsp70 regulate the reaction in a dynamic manner. Hdj1 strongly enhanced ubiquitination, as mentioned above (Fig. 6A), but Bag-1 reduced the ubiquitination efficacy (Fig. 6C, left panels). In contrast, neither p23 nor Aha1 had an impact on the basal Hsp90-dependent ubiquitination (Fig. 6C, right panels). The presence of Hop reduced the ubiquitination of luciferase for both Hsc70 and Hsp90; however, the effect was observed only after shorter time periods, and polyubiquitinated species accumulate after longer time periods, despite increas- ing Hop concentrations (Fig. 6D). Both systems gener- ate substrates with multiple ubiquitinated sites, as shown for the reaction using a lysine-free ubiquitin mutant (Fig. 6B). However, unfolded proteins are more efficiently ubiquitinated in the presence of the Hsp70 system. Ubiquitination of a native chaperone substrate protein Hsp70 and Hsp90 not only interact with misfolded proteins but also with native or near-native proteins. To investigate CHIP-mediated ubiquitination of a native protein substrate, we chose the tumor suppres- sor p53, which has been shown to interact with Hsp70 and Hsp90 [45,46]. At 25 °C, p53 was efficiently mono-ubiquitinated by CHIP in the absence of chaper- ones (Fig. 7A). Neither Hsc70 ⁄ Hdj1 nor Hsp90 enhanced this ubiquitination reaction. At 37 °C, Hsc70 ⁄ Hdj1 but not Hsp90 stimulated CHIP-mediated polyubiquitination of p53. Aha1 and p23 had only minor effects on CHIP-mediated ubiquitination in the presence of Hsp90 (Fig. 7B). Taken together, as in the case of luciferase, ubiquitination of the native Fig. 5. Influence of CHIP on p23 binding to Hsp90. Hsp90 and p23 were incubated in the absence or presence of CHIP, Hop and Aha1 as indicated, and subsequently separated by size-exclusion chroma- tography on a Superose TM 12 10 ⁄ 300 column (GE Healthcare, Frei- burg, Germany), and analyzed by SDS ⁄ PAGE and Coomassie Blue staining. (A) Representative SDS gels. (B) Quantification of the gels shown in (A), Fig. S 1 and additional data: the bands representing p23 were quantified in all lanes. The bar graph shows the amount of p23 co-eluting with Hsp90 (lanes 4–6) relative to the total amount of p23 (sum of all lanes). M. Stankiewicz et al. CHIP preferentially ubiquitinates Hsp70 substrates FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3359 chaperone substrate p53 was more efficient in the pres- ence of Hsc70 and Hdj1 than in the presence of Hsp90 and its co-chaperones Aha1 or p23. Discussion Our study shows that CHIP cooperates with Hsc70 and Hsp90 in a rather passive manner. As CHIP has no effect on substrate dissociation, ADP dissociation or ATP association, it does not increase the half-life of an Hsp70–substrate complex to provide more time for recruitment of the E2 ubiquitin-conjugating enzyme. Similarly, CHIP had no influence on the ATPase cycle of Hsp90 to prolong the ATP-bound state, the ATP- bound state has a high affinity for substrates. We con- clude that CHIP to sample available Hsc70–substrate and Hsp90–substrate complexes in a stochastic process and thereby occasionally effects ubiquitination. Sub- strates that are efficiently folded or refolded and there- fore spend a relatively short time in complex with Hsc70 or Hsp90 have only a small chance of being ubiquitinated. In contrast, substrates that cannot be folded efficiently and consequently cycle on and off the chaperones continuously, or remain bound to chaper- one for a prolonged time interval, will eventually be ubiquitinated by CHIP and targeted for degradation. Such substrates may be misfolded proteins such as heat-denatured luciferase, which we have used in our study, or de novo folding substrates as the cystic Fig. 6. CHIP-mediated ubiquitination of a denatured substrate is more efficient in the presence of Hsc70 and Hdj1 than in the pres- ence of Hsp90. (A–D) Immunoblots of SDS ⁄ PAGE -separated ubiq- uitination reactions using a luciferase-specific antiserum. (A) Time course of ubiquitination of heat-denatured firefly luciferase in the presence of Hsc70 and in the presence and absence of Hdj1. Heat- denatured firefly luciferase was ubiquitinated in the presence of 50 n M E1, 1 lM UbcH5c, 1 lM CHIP (except lane 1), 100 lM ubiqu- itin (except lane 2) and 5 l M Hsc70, in the absence (lanes 3–8) and presence (lanes 9–14) of 5 l M Hdj1 for 1–20 min as indicated. (B) Comparison of CHIP-dependent poly- and multi-ubiquitination effi- ciency in the presence of Hsc70 ⁄ Hdj1 and Hsp90. Left panel, ubiq- uitination of firefly luciferase at various concentrations of Hsc70 (0.2–6 l M) with 5 lM Hdj1 and various concentrations of Hsp90 (0.2–6 l M) as indicated. Luciferase was heat-denatured in the pres- ence of the chaperones, and the ubiquitination mix consisting of 50 n M E1, 1 lM UbcH5c, 1 lM CHIP and 100 lM ubiquitin was added. Right panel, CHIP-dependent ubiquitination of heat-dena- tured luciferase in the presence of 5 l M Hsc70 plus 5 lM Hdj1 (lanes 9–16) or 5 l M Hsp90 (lanes 17–24) with wild-type ubiquitin (lanes 9–12 and 17–20) or the lysine-free ubiquitin variant Ubi-K0, in which all lysines were replaced by arginines (lanes 13–16 and 21–24) for 10–80 min as indicated. (C) Ubiquitination of firefly lucif- erase in the presence of various chaperones and co-chaperones. Lanes 1–12: ubiquitination of luciferase in the presence of 5 l M Hsc70 plus 5 lM Hdj1 and the absence (lanes 1–6) or presence (lanes 7–12) of 5 l M Bag-1 for 5–120 min as indicated. Lanes 13 to 25: ubiquitination of luciferase in the presence of 5 l M Hsp90 and the absence of co-chaperones (lanes 13–17) or the presence of 5 l M Aha1 (lanes 18–21) or 5 lM p23 (lanes 22–25) for 5–40 min as indicated. Lane 13 shows the ubiquitination of luciferase in the absence of CHIP but the presence of Hsp90. (D) CHIP-dependent ubiquitination of heat-denatured firefly luciferase in the presence of 5 l M Hsc70 plus 5 lM Hdj1 (lanes 1–5) or 5 lM Hsp90 (lanes 6–16) and increasing concentrations of Hop (0–23 l M) for 20 and 120 min as indicated. CHIP preferentially ubiquitinates Hsp70 substrates M. Stankiewicz et al. 3360 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS fibrosis transmembrane regulator CFTR, a slow-fold- ing variant (CFTRDF508) of which is known to be efficiently degraded and has been shown to be ubiquiti- nated in a CHIP-dependent way [14]. Such a mecha- nism is also consistent with the phenotype of CHIP ) ⁄ ) -knockout mice, which accumulate aggregated proteins [47]. The small amount of proteins that are ubiquitinated erroneously is the price to be paid for efficient quality control. Such a mechanism is reminis- cent of the quality control in the endoplasmic reticu- lum, where newly synthesized glycoproteins are folded in the calnexin ⁄ calreticulin cycle [48]. Misfolded glyco- proteins are bound in turn by the chaperones calnexin or calreticulin and the folding sensor UDP-glucose- glycoprotein-glucosyltransferase. Proteins that fold properly exit this cycle. Glycoproteins that remain in the cycle for an extended period of time have a high probability that their N-linked glycans will be trimmed by a-1,2-mannosidase I, marking the protein for degra- dation by the ER-associated degradation pathway. The affinity of dimeric CHIP for dimeric Hsp90 (0.38 lm) was approximately sixfold higher than its affinity for monomeric Hsc70 (2.3 lm). This result sug- gests binding of the dimeric CHIP to both C-termini of the dimeric Hsp90, in agreement with a recent amide hydrogen exchange study analyzing the interac- tion of Hsc70 and Hsp90 with CHIP [49]. A K D of 2.4 lm was found previously for the interaction of CHIP with a peptide comprising the ten C-terminal residues of Hsp90 [1]. This value most likely represents the K D for the initial binding of one TPR domain to a single EEVD motif of the Hsp90 dimer in a two-step sequential binding mechanism. K D values in the high nanomolar range have also ben obtained for the inter- action of other TPR proteins with Hsp90 [27,30]. Therefore, TPR domain proteins compete efficiently with CHIP for binding to Hsp90, and only a small amount of Hsp90 is bound to CHIP at equilibrium. This contrasts with the situation for Hsc70 ⁄ Hsp70, whose C-termini interact with only the TPR domain proteins Hop and CHIP [26]. As the concentration of Hsc70 is greater than the concentrations of Hop and CHIP together, changes in the CHIP concentration change the concentration of the Hsc70–CHIP complex, making the system very sensitive to CHIP concentra- tions. Despite the lower affinity of CHIP for Hsc70 compared to Hsp90, CHIP is more frequently in com- plex with Hsc70 in the cell than with Hsp90. We further demonstrate that ubiquitination of heat- denatured luciferase is much more efficient in the pres- ence of Hsc70 and Hdj1 than in the presence of Hsp90. This observation suggests that misfolded proteins at least are targeted to the ubiquitin ⁄ proteasomal path- way through the Hsp70 system rather than through the Hsp90 system. Such a mechanism might also be true for bona fide Hsp90 clients once an Hsp90-specific inhibitor is added. This has been indicated by data for the glucocorticoid receptor, which was found to co-localize with Hsp70 and CHIP after addition of gel- danamycin but not with Hsp90 and FKBP52 [50]. However, if CHIP is over-expressed ectopically or as a consequence of a pathological process, even Hsp90- bound clients may be ubiquitinated and degraded [13– 17]. As all Hsp90 clients, which are degraded upon CHIP over-expression, are also substrates of Hsc70, the overproduced CHIP may act on the Hsc70–client complex rather than the Hsp90–client complex. Our data with p53 support this hypothesis. Therefore, it Fig. 7. Chip-mediated ubiquitination of the native chaperone sub- strate p53. (A) Temperature dependence of p53 ubiquitination. p53 was ubiquitinated in the absence and presence of Hsc70 ⁄ Hdj1, Hsp90 or both at 25 °C (left panel) or 37 °C (right panel). Immuno- blot of SDS ⁄ PAGE-separated ubiquitination reactions using a p53-specific antiserum. (B) The Hsp90 co-chaperones Aha1 and p23 have no influence on CHIP-mediated ubiquitination of p53. p53 was ubiquitinated in the presence of Hsp70 ⁄ Hdj1, Hsp90 and Aha1 and p23 as indicated. M. Stankiewicz et al. CHIP preferentially ubiquitinates Hsp70 substrates FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3361 remains unclear whether CHIP can selectively ubiquiti- nate folded Hsp90-associated clients, thereby perform- ing regulatory functions in the cell. If such a function of CHIP exists, it seems to be of minor importance, as the stability of several bona fide Hsp90 substrates is not affected in CHIP ) ⁄ ) mouse embryonic fibroblasts [51]. The CHIP ) ⁄ ) mice show phenotypes related to abnormal protein aggregation [47] rather than break- down of signaling pathways (compare with FKBP52 knockout mice [52,53]). All these facts speak in favor of CHIP being an E3 ubiquitin ligase with low sub- strate specificity that is responsible for the clearance of hopeless cases of protein folding. However, the role of CHIP in direct substrate binding and its E4 ligase function [54,55] remain puzzling. The native chaperone substrate p53 was only mono- ubiquitinated by CHIP at 25 °C, and the chaperones did not enhance this ubiquitination nor stimulate poly- ubiquitination at this temperature. At 37 °C, Hsc70 ⁄ Hdj1 but not Hsp90 stimulated CHIP-mediated polyubiquitination. NMR experiments with p53 core domain showed that the core domain starts unfolding at 37 °C and binds concomitantly to Hsp90 [56]. The Hsc70 ⁄ Hdj1-stimulated polyubiquitination may there- fore be due to recognition of unfolded regions within p53 by the chaperone. Therefore, ubiquitination of p53 appears to be very similar to ubiquitination of the denatured firefly luciferase. Our study also clarified the previously observed effect of CHIP on the Hdj1 ⁄ Hdj2-stimulated steady- state ATPase rate of Hsc70 [14,33]. We show here that nucleotide exchange of Hsc70 is not affected by CHIP. In contrast, CHIP decreased the Hdj1-stimulated c-phosphate cleavage, as demonstrated by single-turn- over ATPase experiments. CHIP slows down the transition of Hsc70 from a low-affinity state with high substrate association and dissociation rates to a high-affinity state with low substrate dissociation rates. CHIP therefore counteracts the targeting function of the J-domain protein. The molecular basis for this observation could be a reduced association rate for substrates. It was shown previously that Hsp70 pro- teins require two signals for highly efficient hydrolysis of ATP: one signal provided by the J-domain and a second signal provided by the substrate [57–62]. High concentrations of some J-domain proteins can provide both signals by interaction with the substrate binding pocket as well as the ATPase domain [59,63–65]. As CHIP reduces the affinity for substrates without affect- ing the dissociation rate, substrate association is conse- quently reduced. This in turn reduces the substrate signal for ATP hydrolysis. It may be advantageous if substrates do not associate with Hsc70 when CHIP is already bound. Such a mechanism would prevent ubiq- uitination of a substrate that has not had the opportu- nity to refold. In summary, our results suggest the model shown in Fig. 8. Proteins in an intermediate folding state after de novo synthesis at the ribosome or proteins misfolded under stressful conditions are bound by Hsp70s in an Hdj-dependent manner and folded ⁄ refolded to the native state. Proteins that do not fold or that are diffi- cult to fold are released and rebound by Hsp70s sev- eral times (black symbols and arrows in Fig. 8). In a stochastic process, CHIP associates with Hsp70–sub- strate complexes and recruits the E2 conjugating enzyme for ubiquitination of the substrate. As Hsp70s are approximately 10–40 times more abundant than CHIP, only approximately 1–10% of the Hsp70– substrate complexes will be bound by CHIP with possible ubiquitination of the substrate. Efficiently folding substrates (gray symbols and arrows in Fig. 8) have only a small chance of being ubiquitinated. Hsc70–CHIP complexes are less likely to bind misfold- ed proteins. The likelihood of at least one round of refolding is thereby increased. This model of the triage decision allows sufficient time for refolding attempts by the chaperones, keeping the amount of erroneously degraded chaperone substrates low. Any increase in CHIP concentration due to physiological or patho- physiological processes will increase the clearance rate for damaged proteins, at an increased cost of degrad- ing proteins that are still useful. Experimental procedures Protein expression and purification Human CHIP was produced in Escherichia coli and puri- fied by a combination of cation- and anion-exchange chromatography as described previously [2]. Human Hdj1 and human Bag-1 were purified as described previously [37]. Human Hop was purified from over-producing E. coli strains as described previously [66,67]. All proteins were quantified as described previously [68] using the Bio-Rad reagent (Bio-Rad Laboratories, Mu ¨ nchen, Germany). Human wild-type Hsp90b, Hsc70, Hsp70 and Aha1 were expressed with an Ulp1 cleavable N-terminal His6–Smt3 tag in E. coli for 5 h at 30 ° C (20 °C for Aha1) and purified as described previously [69] with some modifications. The cells were lysed in a French press in 25 mm HEPES ⁄ KOH pH 7.5, 150 mm KCl, 5 mm MgCl 2 , 5% glycerol, 5 mm b-mercaptoethanol. Hsp90, Hsc70 and Hsp70 were further purified on a Resource-Q column (GE Healthcare, Freiburg, Germany). Hsp90 and Aha1 were further purified on a Superdex 200 Hiload 16 ⁄ 60 column (GE Healthcare). CHIP preferentially ubiquitinates Hsp70 substrates M. Stankiewicz et al. 3362 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS [...]... repeat-containing protein that interacts with heat shock proteins and negatively regulates chaperone functions Mol Cell Biol 19, 4535– 4545 34 Ha J-H & McKay DB (1994) ATPase kinetics of recombinant bovine 70 kDa heat shock cognate protein and its amino-terminal ATPase domain Biochemistry 33, 14625–14635 FEBS Journal 277 (2010) 3353–3367 ª 2010 The Authors Journal compilation ª 2010 FEBS 3365 CHIP preferentially. .. combined with the ubiquitinating enzymes The samples were incubated at 30 °C for the indicated time periods, centrifuged (16 200 g, 10 min, 4 °C), and the reaction was stopped using SDS loading buffer The proteins were separated on SDS ⁄ PAGE and analyzed by Western blot using a luciferase-specific polyclonal antiserum For ubiquitination of p53, 0.5 lm of the full-length protein were incubated under the... cycles of binding and release by the chaperones and spend a relatively short time in the chaperone-bound state These substrates are only inefficiently ubiquitinated by CHIP Substrates that are not refolded efficiently by the chaperones (black symbols) are repeatedly bound and released by the Hsp70 system (black arrows), and therefore spend a longer time on the chaperones Random sampling of Hsp70 by CHIP entails... Brain Res Mol Brain Res 123, 27–36 51 Morishima Y, Wang AM, Yu Z, Pratt WB, Osawa Y & Lieberman AP (2008) CHIP deletion reveals functional redundancy of E3 ligases in promoting degradation of both signaling proteins and expanded glutamine proteins Hum Mol Genet 17, 3942–3952 52 Yang Z, Wolf IM, Chen H, Periyasamy S, Chen Z, Yong W, Shi S, Zhao W, Xu J, Srivastava A et al (2006) FK506-binding protein. .. by CHIP entails eventually ubiquitination and targeting for degradation If CHIP is bound to Hsp70, substrate binding to Hsp70 is reduced (light gray arrows) p23 was expressed as a N-terminal 6His fusion in the same way, and the histidine tag was removed by digestion with thrombin protease (Serva, Heidelberg, Germany) The thrombin and the histidine tag were removed using a Resource-Q column UbcH5c was... The second step of ATP binding to DnaK induces peptide release J Mol Biol 263, 657–670 39 Hohfeld J & Jentsch S (1997) GrpE-like regulation of ¨ the Hsc70 chaperone by the anti-apoptotic protein BAG-1 EMBO J 16, 6209–6216 40 Demand J, Alberti S, Patterson C & Hohfeld J (2001) Cooperation of a ubiquitin domain protein and an E3 ubiquitin ligase during chaperone ⁄ proteasome coupling Curr Biol 11, 1569–1577... al CHIP preferentially ubiquitinates Hsp70 substrates Fig 8 Model for the role of CHIP in the triage decision The chaperone systems encounter various substrates, especially under heat shock conditions (HS): substrates that can be refolded (gray symbols and gray arrows) and substrates that cannot be refolded by the chaperones (black symbols and black arrows) Substrates that are efficiently refolded by. .. Biol Chem 283, 8877–8884 CHIP preferentially ubiquitinates Hsp70 substrates 70 Bosl B, Grimminger V & Walter S (2005) Substrate binding to the molecular chaperone Hsp104 and its regulation by nucleotides J Biol Chem 280, 38170– 38176 71 Ali JA, Jackson AP, Howells AJ & Maxwell A (1993) The 43-kilodalton N-terminal fragment of the DNA gyrase B protein hydrolyzes ATP and binds coumarin drugs Biochemistry... involved in Hsp90 binding Biochim Biophys Acta 1783, 1003–1014 Supporting information The following supplementary material is available: Fig S1 Gel filtration analysis of Hsp90, CHIP, p23 and combinations thereof This supplementary material can be found in the online version of this article Please note: As a service to our authors and readers, this journal provides supporting information supplied by the... Hsp90 ATPase activity by tetratricopeptide repeat (TPR)-domain co-chaperones EMBO J 18, 754–762 32 Palleros DR, Welch WJ & Fink AL (1991) Interaction of hsp70 with unfolded proteins: effects of temperature and nucleotides on the kinetics of binding Proc Natl Acad Sci USA 8, 5719–5723 33 Ballinger CA, Connell P, Wu Y, Hu Z, Thompson LJ, Yin LY & Patterson C (1999) Identification of CHIP, a novel tetratricopeptide . CHIP participates in protein triage decisions by preferentially ubiquitinating Hsp70-bound substrates Marta Stankiewicz 1 , Rainer Nikolay 1, *,. RCMLA, suggesting that substrate binding by CHIP may be a specific interaction limited to certain proteins. In con- clusion, CHIP did not increase the half