DIRECTING AKT AND GSK3β: MOLECULAR INSIGHTS INTO CELL SIGNALING AND SURVIVAL GORDON P. MEARES

Transcription

1 DIRECTING AKT AND GSK3β: MOLECULAR INSIGHTS INTO CELL SIGNALING AND SURVIVAL by GORDON P. MEARES RICHARD S. JOPE, COMMITTEE CHAIR GAIL V.W. JOHNSON MATHIEU LESORT SCOTT WILSON J. MICHAEL WYSS A DISSERTATION Submitted to the graduate faculty of The University of Alabama at Birmingham, in partial fulfillment of the requirements for the degree of Doctor of Philosophy Birmingham, Alabama 2007

2 DIRECTING AKT AND GSK3β: MOLECULAR INSIGHTS INTO CELL SIGNALING AND SURVIVAL GORDON P. MEARES CELL BIOLOGY ABSTRACT Proper regulation of survival signaling is critical for all organisms. One important signaling cascade involved in the coordinated effort to control signals influencing cell fate is the phosphatidylinositol 3-kinase (PI3K)-Akt-glycogen synthase kinase 3β (GSK3β) signaling cascade. Following activation by growth factors the PI3K-Akt pathway promotes cell survival and, cell death is facilitated following inhibition of Akt. Many factors influence the function of the Akt-GSK3β signaling dyad, including phosphorylation, protein complex formation and subcellular localization. Previous work had demonstrated that the 90kDa heat shock protein, HSP90, directly binds and stabilizes Akt. Furthermore, several heat shock proteins including HSP90 and HSP105 can modulate cell survival following various insults. Therefore, the goals of this research were to test if heat shock proteins, either directly or indirectly, could regulate signaling to GSK3β, modulate GSK3β-mediated apoptosis, or direct nuclear localization of GSK3β. The HSP90 inhibitor geldanamycin was used to block HSP90 function followed by examination of insulin like growth factor-1 (IGF-1) or insulin-induced signaling. HSP90 inhibition enhanced growth factor-induced phosphorylation and activation of Akt and subsequent phosphorylation and inactivation of GSK3β, indicating that HSP90 normally acts to buffer the magnitude of growth factor signaling. Hsp90 dampened Akt signaling by facilitating phosphatase-mediated dephosphorylation of Akt. ii

3 Endoplasmic reticulum (ER) stress, in opposition to growth factors, promotes apoptotic signaling. GSK3 and HSP105 were found to interact and both modulate ER stress-induced apoptotic signaling. Knockdown of HSP105 attenuated ER stress-induced caspase-3 activation down stream of the proapoptotic molecule C/EBP homologous protein (CHOP), indicating that HSP105 facilitates ER stress-induced caspase-3 activation. GSK3 also facilitates ER stress-induced apoptotic signaling, however GSK3 is acting upstream of CHOP expression. Thus HSP105 and GSK3 interact and regulate distinct points in the ER stress-induced apoptotic signaling pathway. Many of the actions of GSK3β are influenced by tightly controlled nuclear localization. Nuclear localization of GSK3β is controlled by a cohort of factors including a bipartite nuclear localization sequence (NLS), intramolecular interactions and stimulusinduced release from protein complexes. Overall, this work reveals novel roles for heat shock proteins in the regulation of survival signaling. Moreover, the molecular mechanism regulating the nuclear localization of GSK3β were discovered. iii

4 ACKNOWLEDGMENTS I thank my mentor and graduate advisor Dr. Richard Jope for guiding my transition from a student to a scientist. I thank my committee members Dr. Gail Johnson, Dr. Mathieu Lesort, Dr. Scott Wilson, and Dr. J. Michael Wyss for their advice over the years. I also thank my parents for support and inspiration. Finally, I thank my wife, Sheri, who keeps me from getting lost. iv

5 TABLE OF CONTENTS Page ABSTRACT... iii ACKNOWLEDGMENTS...v LIST OF FIGURES... vi INTRODUCTION...1 Akt...1 Glycogen synthase kinase 3β (GSK3β)...2 Heat shock proteins (HSPs)...5 HSP HSP Endoplasmic reticulum (ER) stress...10 HEAT SHOCK PROTEIN 90 (HSP90) DAMPENS AND DIRECTS SIGNALING STIMULTED BY INSULIN-LIKE GROWTH FACTOR-1 AND INSULIN...16 GSK3 AND HSP105 INTERACT AND REGULATE ER STRESS-INDUCED APOPTOTIC SIGNALING...40 RESOLUTION OF THE NUCLEAR LOCALIZATION MECHANISM OF GLYCOGEN SYNTHASE KINASE-3: FUNCTIONAL EFFECTS IN APOPTOSIS...81 CONCLUSIONS REFERENCES v

6 LIST OF FIGURES Figure Page INTRODUCTION 1 Growth factor signaling to Akt A simplified view of the unfolded protein response...15 HEAT SHOCK PROTEIN 90 (HSP90) DAMPENS AND DIRECTS SIGNALING STIMULTED BY INSULIN-LIKE GROWTH FACTOR-1 AND INSULIN 1 Inhibition of HSP90 by geldanamycin promotes IGF-1-induced Akt activation Inhibition of HSP90 by geldanamycin promotes insulin-induced Akt activation Inhibition of HSP90 promotes IGF-1-induced intracellular signaling Inhibition of HSP90 blocks dephosphorylation of Akt...39 GSK3 AND HSP105 INTERACT AND REGULATE ER STRESS-INDUCED APOPTOTIC SIGNALING 1 GSK3 associates with HSP HSP105 interacting proteins identified by mass spectrometry HSP105 knockdown and growth factor signaling to GSK3β HSP105 and UPR-induced signaling HSP105 knockdown attenuates ER stress-induced caspase-3 activation Knockdown of HSP105 does not promote cell survival GSK3 regulates ER stress-induced CHOP expression and caspase-3 activation...80 RESOLUTION OF THE NUCLEAR LOCALIZATION MECHANISM OF GLYCOGEN SYNTHASE KINASE-3: FUNCTIONAL EFFECTS IN APOPTOSIS vi

7 1 Identification and mutagenesis of an NLS in GSK3β Intracellular distribution of wild-type and mutant GSK3β The NLS consisting of amino acids from GSK3β is sufficient for nuclear localization and is highly conserved The N-terminal tail of GSK3β promotes nuclear localization NSL-deficient GSK3β can rescue GSK3β knockout MEFs from TNF-induced cell death GSK3 dimers are not detectable and do not contribute to nuclear localization Cytosolic GSK3 is bound in protein complexes and released by stimuli CONCLUSIONS 1 Growth factor signaling and ER stress converge on GSK3β to control cell fate vii

8 Introduction Akt Akt, also known as protein kinase B, is a central player in several signaling pathways that regulate diverse cellular functions; including cell survival, metabolism and transcription (Fayard et al. 2005). Akt was originally identified as a proto-oncogene and quickly was recognized as a serine/threonine kinase (Staal 1987; Jones et al. 1991). Since then, three isoforms of Akt have been identified and the importance of Akt in survival signaling has become clear (Brazil and Hemmings 2001). As with many kinases, Akt activation is dependent on phosphorylation. In growth factor-stimulated cells, this phosphorylation occurs downstream of phosphatidylinositol 3-kinase (PI3K) (Burgering et al. 1995). Stimulation with growth factors leads to activation of PI3K, which phosphorylates phosphoinositides leading to activation of the upstream kinase 3- phosphoinositide-dependent protein kinase 1 (PDK1) and recruitment of Akt to the plasma membrane via the pleckstrin homology domain of Akt (Andjelkovic et al. 1997). PDK1 is then responsible for the activating phosphorylation at threonine 308 in the activation loop of Akt (Alessi et al. 1997). An additional phosphorylation on Akt occurs at serine 473 to stabilize the active conformation (Yang et al. 2002) and, together with phosphorylation at threonine 308, results in maximal activation of Akt (Alessi et al. 1996). The kinase responsible for phosphorylation of serine 473 remained elusive for many years with several kinases being proposed. Current evidence indicates that SIN1, a component of the mammalian target of rapamycin complex 2 (TORC2) is responsible for phosphorylation of serine 473 of Akt (Jacinto et al. 2006). In addition to serine/threonine phosphorylation, Akt activity is promoted by phosphorylation on at least two tyrosines 1

9 near the activation loop (Chen et al. 2001). Figure 1 illustrates growth factor signaling leading to Akt phosphorylation. Akt promotes survival by the simultaneous promotion of pro-survival proteins and the inhibition of pro-apoptotic proteins. Survival promoting factors activated by Akt include nuclear factor-κb (NF-κB) (Kane et al. 1999) and β-catenin (Tian et al. 2004), among others. Akt inhibits a host of pro-apoptotic molecules, such as caspases, Bad, Bax, forkhead-box class O (FoxO), and GSK3β (Fayard et al. 2005). Clearly the breadth of Akt s influence on cellular function and survival is dependent on the vast number of substrates Akt phosphorylates. Particularly important and relevant to the studies presented here is the regulation of GSK3β by Akt. Glycogen synthase kinase 3β GSK3β was discovered nearly 30 years ago as a kinase capable of phosphorylating and inhibiting glycogen synthase (Embi et al. 1980). Two isoforms, GSK3α and GSK3β, have been identified and are encoded by different genes (Woodgett 1990). Since the discovery of GSK3β, it has been found to phosphorylate a steadily increasing number of substrates placing it into several signaling pathways. The importance of GSK3β as a signaling molecule has also implicated it with several widespread diseases. GSK3β affects many fundamental cellular functions, such as the cell cycle, gene transcription, and cytoskeletal dynamics. This is due to the ability of GSK3β to phosphorylate key proteins governing these processes. GSK3β, when active, is also a proapoptotic protein that inhibits several transcription factors that support cell survival (Grimes and Jope 2001). 2

10 GSK3β is constitutively active, but its activity can be regulated by site specific phosphorylation, protein complex formation and intracellular distribution (Grimes and Jope 2001). Phosphorylation on serine 9 reduces GSK3β kinase activity (Southerland et al 1993). Although several kinases have been found to phosphorylate serine 9 of GSK3β (Grimes and Jope 2001), Akt seems to be the dominate kinase in the phosphorylation of the serine 9 residue on GSK3β. When active, GSK3β may phosphorylate unprimed substrates at serine or threonine residues that are usually followed by a proline, though more often GSK3β phosphorylates primed substrates. Primed substrates are proteins that have already been phosphorylated by some other kinase on a serine four residues C-terminal from the ser/thr site for GSK3β. A primed GSK3β phosphorylation site would appear as S/T-X- X-X-S[P] (Wang et al. 1994). This type of substrate recognition adds another layer of complexity to GSK3β signaling, it allows potential substrates to remain localized with GSK3β but not be phosphorylated until an appropriate signal leads to priming of the substrate. Based on crystallographic data it was determined that the serine 9 phosphorylation is inhibitory because the phosphoserine is a pseudosubstrate for the primed substrate-binding site that includes arginine 96 of GSK3β. Thus, GSK3β binds to itself, effectively blocking the docking site for some of the substrates of GSK3β (Dajani et al. 2001). While phosphorylation of serine 9 leads to inhibition of GSK3β, phosphorylation of tyrosine 216 increases its activity (Wang et al. 1994). The regulation of tyrosine 216 phosphorylation is not fully understood, but appears to be via HSP90-dependent 3

11 autophosphorylation (Lochhead et al. 2006; Cole et al. 2004). Additionally, tyrosine phosphorylated GSK3β was shown to preferentially localize to the nucleus (Bhat et al. 2000). Although a great deal of GSK3β regulation depends on phosphorylation it is only one of multiple regulatory mechanisms. GSK3β also can be regulated by protein complex formation. An excellent example of this regulation is the involvement of GSK3β in the highly conserved Wnt signaling pathway. In the absence of Wnt ligand, GSK3β is found in a complex with APC and Axin where it is able to bind and phosphorylate β-catenin, targeting it for proteasomal degradation (Hart et al. 1998). In the presence of Wnt ligand, the GSK3β protein complex is restructured to exclude APC (Adenomatous polyposis coli) and Axin, allowing GSK3β to bind Frat1/GBP (GSK3 binding protein) and disheveled, thereby blocking GSK3β s accessibility to β-catenin, resulting in the accumulation of β-catenin (Li et al. 1999; Thomas et al. 1999). Another way GSK3β is regulated is by intracellular localization. In general, GSK3β is a predominantly cytosolic enzyme, thus providing a physical separation between GSK3β and its nuclear substrates. When GSK3β transits to the nucleus it gains access to many important molecules. Within the nucleus GSK3β promotes the functions of the tumor suppressor protein p53 (Watcharasit et al. 2002). Additionally, GSK3β phosphorylates cyclin D1 to facilitate nuclear export and degradation of cyclin D1 which promotes cell cycle progression (Diehl et al. 1998). Similarly, GSK3β phosphorylates to facilitate nuclear export of the transcription factors nuclear factor of activated T-cells (NF-AT) (Beals et al. 1997) and Snail (Zhou et al. 2004). Of interest to us is the dynamic 4

12 distribution between the nucleus and cytosol. There are a few mechanisms used by proteins to enter the nucleus, but all enter via the nuclear pore complex (NPC). Proteins of approximately 40 kda and smaller may enter the nucleus by diffusion, while the entry of larger proteins is regulated. If a protein that enters the nucleus is too large for diffusion, it generally contains a nuclear localization sequence (NLS) or is carried in by a protein containing an NLS. GSK3β is 47 kda (Woodgett 1990) and as such rests at the threshold of diffusion and facilitated nuclear transport. Despite this, it is clear that the process is dynamically regulated and thus not diffusion mediated. Under basal conditions, little GSK3β is found in the nucleus. When cells are subjected to proapoptotic insults, such as heat shock or staurosporine, GSK3β rapidly translocates to the nucleus where many GSK3β substrates reside (Bijur and Jope 2001; Jope and Johnson 2004). Furthermore, nuclear GSK3β was shown to be highly active (Bijur and Jope 2003) and following apoptotic insults the translocation of GSK3β to the nucleus occurs prior to caspase activation, suggesting a role for nuclear GSK3β in apoptosis (Bijur and Jope 2001). Although the mechanisms regulating nuclear import were previously unknown, it was demonstrated that nuclear export of GSK3β was at least partially mediated by FRAT1 in a CRM1-dependent fashion (Franca-Koh et al. 2002). Additionally, through overexpression studies, it was demonstrated that Axin2 could sequester GSK3β in the cytosol (Yook et al. 2006). Included in the work presented here is identification of the mechanisms regulating the nuclear localization of GSK3β. Heat shock proteins Molecular chaperones, such as heat shock proteins (HSPs), are required by many proteins in order for them to fold properly and become functionally active. There is a 5

13 wide array of chaperone proteins present in eukaryotic cells. Chaperone proteins can be very general in their activity or they can have limited specificity, such as for a family of proteins or for proteins colocalized with the chaperone in an organelle. Furthermore, most chaperones are highly conserved and work by the same basic mechanism; the utilization of energy-dependent conformational changes to assist their client proteins (Soti and Csermely 2002). Once thought of as just protein folding machines, HSPs are now recognized to regulate many signaling components, especially kinases and apoptotic pathways (Beere 2005; Gaestel 2006). HSP90 One such molecular chaperone is the 90 kda heat shock protein. HSP90 is highly conserved, is generally expressed at levels high enough to account for 1% of total cellular protein, and mediates diverse biological functions, ranging from suppression of the deleterious effects of genetic mutations to directing molecular switches in signaling pathways (Rutherford and Lindquist 1998). HSP90 has the unique ability to suppress phenotypic manifestations of genetic mutations, and has thus been proposed to act as a genetic buffer. Using Drosophila and Arabidopsis it was demonstrated that loss of HSP90 function resulted in the appearance of numerous and diverse phenotypes. Furthermore, the phenotypes were heritable, dependent on the genetic background of the organism and, if selected for, became HSP90-independent. From these findings it was suggested that HSP90 suppresses the effects of mutations until environmental factors influence HSP90 function, allowing the mutant phenotype to appear (Rutherford and Lindquest 1998; Queitsch et al. 2002). 6

14 HSP90 function appears to be dictated by nucleotide binding. If ATP is bound, HSP90 and its co-chaperones promote proper folding, maturation and release of its client proteins. When ADP is bound, the conformation of HSP90 is changed and this targets some HSP90 client proteins to the proteasome for degradation (Schneider et al. 1996; Neckers 2002). Akt is one of HSP90 s client proteins. Many of the studies to identify HSP90 client proteins used the benzoquinone antibiotic geldanamycin (or its analogs), which specifically inhibits HSP90 family members by binding to the ATP binding pocket (Whitesell et al. 1994; Grenert et al. 1997; Prodromou et al. 1997; Stebbins et al. 1997). This binding in the ATP binding pocket induces the ADP bound conformation, promoting the assembly of the HSP90 chaperone complex that targets client proteins for degradation (Schneider et al. 1996; Neckers 2002). HSP90 inhibitors lead to the proteasome-dependent degradation of Akt between 6 and 24 hours following HSP90 inhibition (Basso et al. 2002). In addition to Akt, several other proto-oncogenes require active HSP90 for their stability. Based on this, HSP90 inhibitors are currently in clinical trials for the treatment of several cancers (Neckers 2006). It was clearly demonstrated that Akt stability is dependent on HSP90 function; however the role of HSP90 in Akt/GSK3β signaling was not previously examined. In the following chapter this important question will be addressed. HSP105 HSP105 (also referred to as HSP110) is a major mammalian heat shock protein expressed in particularly high levels in the brain where it is localized in both the cytoplasm and the nucleus. There are two isoforms, HSP105α and HSP105β, which are 7

15 likely the result of alternative RNA splicing. The two isoforms are identical except that the β isoform is missing a stretch of 44 amino acids (Ishihara et al. 1999). HSP105 shares a large region of homology with HSP70, including an ATP binding domain, but does not appear to have the ATPase activity of HSP70 (Ishihara et al. 1999; Yamagishi et al. 2002). Chaperones are often found in complexes with other chaperones and cochaperones. In this regard HSP105 is no different; it forms a complex with HSP70/HSP40. When chaperones team up, it usually enhances the overall chaperoning ability, but it was shown that HSP105 actually suppresses the refolding ability of HSP70/HSP40 (Yamagishi et al. 2000). This raised the interesting possibility that HSP105 may counteract the action of HSP70. The modulation of HSP70-mediated protein folding by HSP105 is suppressed by phosphorylation of serine 509 on HSP105 by protein kinase CK2 (CK2) (Ishihara et al. 2003a). Although the work of Yamagishi and Ishihara suggests that HSP105 reduces HSP70 function, it has clearly been demonstrated that HSP105 functions as a nucleotide exchange factor for HSP70, thus promoting its activity (Dragovi et al. 2006; Raviol et al. 2006). In addition to being part of the HSP70 chaperone machinery, HSP105 has been reported to associate with several other proteins. Initial bacterial and yeast twohybrid screenings identified α-tubulin, cofilin, dyenin light chain 2A, α-adducin, plateletactivating factor, acetylhydrolase α-subunit, and phosphoglycerate kinase 1 as HSP105 interacting proteins (Saito et al. 2004). Of these, only the interaction with α-tubulin has been functionally characterized. HSP105 was reported to suppress disorganization of microtubules during heat shock via its interaction with α-tubulin (Saito et al. 2003). 8

16 Additionally, HSP105 interacts with the truncated androgen receptor with an extended polyglutamine tract and suppresses aggregation and apoptosis (Ishihara et al. 2003b). Similarly, HSP105 interacts with mutant Cu/Zn superoxide dismutase (SOD1) to prevent aggregation (Yamashita et al. 2007). Virtually every major family of heat shock proteins has been implicated in the regulation of apoptosis. In almost all cases HSPs have been reported to suppress apoptosis at multiple points of the signaling cascade. Release of cytochrome C, formation of the apoptosome, cleavage of procaspases, and apoptosis inducing factor (AIF) activity are all negatively affected by certain HSPs (Gabai et al. 1998; Beere 2001; Ravagnan et al. 2001). In the case of HSP105, overexpression was shown to confer protection against oxidative and genotoxic stress-induced apoptosis in PC12 cells (Hatayama et al. 2001). HSP105 also attenuates staurosporine-induced apoptosis in HeLa cells (Yamagishi et al. 2006). Although HSPs are generally believed to promote cell survival, the major HSPs, including HSP90, HSP70 and HSP60, are needed for the proper execution of apoptosis (Arya et al. 2007). Consistent with this, when HSP105α was overexpressed in mouse embryonal F9 cells there was enhanced apoptosis in response to oxidative stress. Activation of p38 was enhanced in HSP105α overexpressing cells, but not c-jun N-terminal kinase (JNK) and the inhibition of p38 decreased down stream markers of apoptosis, including cytochrome C release and caspase activation (Yamagishi et al. 2002). In line with other high molecular weight HSPs, HSP105, in general, is prosurvival but can also promote proapoptotic signaling. 9

17 Endoplasmic reticulum stress The endoplasmic reticulum (ER) is a membranous organelle that provides a unique environment for the proper folding and maturation of secreted and membranebound proteins. However, with this specialized environment come the challenges of maintaining homeostasis. Eukaryotic cells have evolved an elegant but complex system to preserve or restore proper protein folding within the ER. Despite this, diseases and environmental factors can perturb ER homeostasis, resulting in accumulation of misfolded proteins and leading to the condition termed ER stress. The response to ER stress involves multiple pathways as the cell attempts to combat the accumulation of misfolded proteins in the ER. If the cell s capacity for misfolded proteins is overwhelmed, apoptotic cell death will ensue. Therefore it is imperative that the cell responds to misfolded proteins with a careful balance of both survival and apoptotic signals. The initial response to ER stress is referred to as the unfolded protein response (UPR) (Figure 2). An important role of the UPR is the transduction of signals from the ER to other parts of the cell, including the cytosol and nucleus. There are three main trans-er membrane proteins that transmit signals from the ER during the UPR; PKR-ER like kinase (PERK), inositol requiring enzyme 1 (IRE1), and activating transcription factor 6 (ATF6). Interestingly, each of these enzymes is regulated by the ER resident HSP70 homologue glucose regulated protein 78 (GRP78/BIP) (Szegezdi et al 2006). Under basal conditions GRP78 binds to the ER lumenal domain of PERK and maintains it in an inactive state. Following accumulation of misfolded proteins in the ER, GRP78 is recruited to the misfolded proteins and away from PERK. This results in 10

18 oligomerization and autophosphorylation of PERK leading to its activation (Bertolotti et al. 2000). Active PERK then phosphorylates serine 51 of the eukaryotic initiation factor 2α (eif2α) leading to attenuation of protein translation (Harding et al. 1999). Blocking protein synthesis during ER stress is thought to promote survival by reducing the influx of new proteins into the ER, thus giving the ER a chance to eliminate (either through refolding or degradation) the misfolded proteins (Harding et al. 2000). Although most protein synthesis is down-regulated by phosphorylation of eif2α, synthesis of the activating transcription factor 4 (ATF4) is up-regulated, leading to expression of stress responsive genes (Harding et al. 2003). Among these is the proapoptotic protein C/EBP homologous protein (CHOP) that facilitates ER stress-induced apoptosis. Several nonmutually exclusive mechanisms have been proposed for how CHOP promotes apoptosis: activation of a pathway that de-represses eif2α (Marciniak et al. 2004), increased oxidation of proteins (Marciniak et al. 2004), and down regulation of Bcl-2 (McCullough et al 2001). Similar to PERK, IRE1 is a trans-er membrane kinase held inactive by GRP78. Following release of GRP78, IRE1 oligomerizes, autophosphorylates, and becomes active (Bertolotti et al. 2000). Once active, IRE1 recruits and activates JNK, thus coupling the UPR to stress activated kinases (Urano et al. 2000). In addition to kinase activity, IRE1 also has endoribonuclease activity which is used during ER stress to splice the mrna of the transcription factor X box-binding protein 1 (XBP1). IRE1-mediated splicing of XBP1 mrna removes 26 nucleotides, giving rise to active XBP1, which then promotes expression of ER-resident molecular chaperones, including GRP78 (Yoshida et al. 2001). Furthermore, IRE1 is regulated on the cytoplasmic side of the ER by the 11

19 apoptosis mediators Bax and Bak, which function to suppress both IRE1-mediated JNK activation and XBP1 splicing (Hetz et al. 2006). ATF6 is an ER membrane-bound transcription factor sequestered by GRP78 (Shen et al. 2002). Following release from GRP78, ATF6 moves from the ER to the golgi where it is cleaved by site 1 and site 2 proteases (Shen et al. 2002; Ye et al. 2000). The ATF cleavage product, p50atf6, then translocates to the nucleus where it promotes expression of ER stress responsive genes, including XBP1, which feeds back to the IRE1-mediated arm of the UPR. The UPR is a concerted effort by the cell to restore ER homeostasis and survive. However, as mentioned above, ER stress must also promote apoptotic signaling in the event the UPR is overwhelmed. As with many apoptotic paradigms, ER stress-induced apoptosis primarily utilizes the intrinsic (mitochondrial) pathway leading to activation of caspase-9 and caspase-3 (Boyce and Yuan 2006). However, ER stress can lead to caspase-independent cell death (Zhang and Armstrong 2007). Furthermore, the steps leading from the ER to the execution phase of apoptosis are poorly understood, but clearly involve GSK3β. Inhibition of GSK3 suppresses caspase 3 activation and subsequent PARP cleavage induced by the ER stress-inducing agents thapsigargin (an ER Ca ++ ATPase inhibitor) and tunicamycin (a glycosolation inhibitor) (Song et al. 2003). Several groups have since confirmed the obligatory role of GSK3β in ER stress-induced apoptosis using a variety of cell types, ER stress models, and GSK3 inhibitors (Brewster el al. 2006; Chen et al. 2004; Kim et al. 2004; Takadera et al. 2006). Despite considerable evidence that GSK3β has a crucial role in many apoptotic conditions, little 12

20 is known about the substrates involved or the mechanism by which GSK3β facilitates apoptosis. Overall, the data in the following chapters provides considerable insights into mechanisms regulating GSK3 through phosphorylation, cellular localization, and protein interactions, 13

21 GF plasma membrane PIP2 PIP3 PIP3 P P P P Adaptor p85 p110 PI3K LY PH PDK1 T308 P PH HSP90 Akt P Geldanamycin S473 Active TORC2 T308 P PH Akt HSP90 P Geldanamycin S473 P FoxO P GSK3α P GSK3β P Bad Fig 1. Growth factor signaling to Akt. Growth factors such as IGF-1 or insulin bind the receptor at the plasma membrane resulting in activation of phosphatidylinositol 3-kinase (PI3K), the target of the inhibitor LY PI3K then phosphorylates phosphatidylinositol 3,4-bisphosphate (PIP 2 ) to form phosphatidylinositol 3,4,5-trisphosphate (PIP 3 ). Akt, stablilized by HSP90, is then recruited to the membrane via its plekstrin homology (PH) domain where it is phosphorylated at threonine 308 by phosphoinositide-dependent kinase-1 (PDK1). Akt is further phosphorylated by the target of rapamycin complex 2 (TORC2) at serine 473 resulting in maximal activation. Akt then phosphorylates to inhibit several proapoptotic proteins including GSK3. 14

22 Misfolded proteins GRP78 GRP78 GRP78 ER lumen IRE1 ATF6 PERK Cytosol XBP1 mrna splice XBP1 mrna cleaved ATF6 Protein synthesis P eif2α XBP1 ATF4 Chaperones CHOP Apoptosis Nucleus Fig 2. A simplified view of the unfolded protein response. Following accumulation of misfolded proteins in the ER, GRP78 is recruited to the misfolded proteins and pro- and anti-apoptotic signaling is initiated. Inositol-requiring enzyme 1 (IRE1) is activated to splice Xbox binding protein 1 (XBP1) mrna leading to expression of an active transcription factor to promote expression of ER chaperones. Activating transcription factor 6 (ATF6) proceeds to the golgi where it is cleaved and released to translocate to the nucleus where it promotes expression of ER chaperones and C/EBP homologous protein (CHOP). (PKR)-like ER kinase (PERK) is activated to phosphorylate eif2α, blocking most protein synthesis and selectively increasing ATF4 expression leading to upregulation of CHOP, which then promotes apoptosis. 15

23 Heat shock protein 90 (HSP90) dampens and directs signaling stimulated by insulin-like growth factor-1 and insulin by Gordon P. Meares. Anna A. Zmijewska, and Richard S. Jope* FEBS Letters 574, Copyright 2004 by Elsevier B.V. Used with permission Formatted for dissertation 16

24 Abstract Heat shock protein-90 (Hsp90) buffers cells from genetic mutations and environmental stresses. To test if this capability reflects a normal physiological function of Hsp90 to buffer cellular signals, the effects of Hsp90 inhibition were measured on activation of Akt. Inhibition of Hsp90 with geldanamycin amplified Akt phosphorylation induced by insulin-like growth factor-1 (IGF-1) or insulin, indicating that Hsp90 normally buffers these signals. Furthermore, with IGF-1 stimulation Hsp90 inhibition increased p38 activation, produced additive activation of p90rsk, and slightly increased the duration of ERK1/2 activation. Hsp90 dampened Akt signaling by facilitating phosphatase-mediated dephosphorylation of Akt. Thus, Hsp90 not only buffers the cellular effects of mutations and stresses, but also buffers the magnitude and duration of activation of proliferative and survival-promoting signaling responses. 17

25 1. Introduction The chaperone protein, heat shock protein-90 (Hsp90), recently has been recognized as a cellular buffer against deleterious gene mutations and environmentally stressful conditions [1]. In organisms as disparate as flies and plants, functionally eliminating the action of Hsp90 revealed hidden genetic mutations that had been buffered by the chaperone, indicative of its crucial role in balancing stability with the capability for change [2-4]. This revelation followed the defining role of heat shock proteins in general as proteins capable of minimizing the impact of toxic environmental insults, including their namesake heat shock as well as many other potentially lethal conditions [6-8]. The capacity of Hsp90 to buffer the impact of genetic and environmental influences raised the question of whether this is a specialized function mobilized under duress, or if it is part of the normal physiological function of Hsp90 that was initially recognized by the severe outcomes that occurred following unprotected exposure to mutations or insults [2, 3]. In other words, does Hsp90 buffer nontoxic cellular signals as well as those that may be toxic? In order to address the potential buffering capacity of Hsp90 in nontoxic cellular signaling mechanisms, the present study focused on signaling leading to activation of the protein kinase Akt (also known as protein kinase B), a well-characterized client protein of Hsp90. Stimulation of a variety of receptors for growth factor, insulin, and other signaling molecules, activates a phosphatidylinositol 3-kinase (PI3K)-mediated signaling pathway that leads to activation of Akt via dual phosphorylation on serine-473 and threonine-308 on Akt [9, 10]. Many studies have shown that Akt is complexed with Hsp90 [11-13]. Long-term (e.g., 24 hr) inhibition of Hsp90 with the Hsp90-selective 18

26 inhibitor geldanamycin or its analogs [14, 15] causes degradation of Akt through the proteasomal pathway [16-21]. However, previous investigations did not examine the ramifications of inhibiting Hsp90 on the immediate signals causing activation of Akt. Therefore, this study tested if Hsp90 has a buffering capacity in receptorstimulated activation of Akt as well as other outcomes of receptor activation. The results show that inhibition of Hsp90 amplified activation of Akt following stimulation with insulin-like growth factor-1 (IGF-1) or insulin. These results suggest that Hsp90 buffers physiological as well as toxic signaling activities. 2. Materials and methods 2.1. Cell culture and treatments Human neuroblastoma SH-SY5Y cells and human embryonic kidney HEK293 cells were grown as described previously [22]. Cells were grown in serum-free media for 24 h prior to treatment with geldanamycin, 25 µm radicicol, 10 µm (Alexis, San Diego, CA), 50 ng/ml IGF-1 (Serologicals, Purchase, NY), 100 nm insulin (Sigma, St. Louis, MO), or 25 µm 17-allylamino-17-demethoxygeldanamycin (17-AAG; generously supplied provided by Dr. Robert J. Schultz, Developmental Therapeutics Program, National Cancer Institute) Immunoprecipitation To immunoprecipitate Akt, 3.5 µg of Akt1 antibody was incubated with 25 µl of 50% protein G-sepharose in binding buffer (100 mm sodium phosphate, ph 7.2) for 2 h at 4 o C followed by two washes with binding buffer. The cell extract (100 µg protein) was precleared with 25 µl 50% protein G-sepharose for 2 h at 4 o C. The extract was centrifuged, and the supernatant was incubated with the Akt1-conjugated protein G- 19

27 sepharose for 18 h at 4 o C. The immobilized immune complex was washed 3 times with lysis buffer, mixed with 40 µl of Laemmli sample buffer, placed in a boiling water bath for 5 min, and the samples were immunoblotted for Hsp90 and Akt Immunoblotting Cells were washed twice with phosphate-buffered saline and were lysed with lysis buffer (20 mm Tris, ph 7.5, 150 mm NaCl, 2 mm EDTA, 2 mm EGTA, 0.5% NP-40, 1 mm sodium orthovanadate, 100 µm phenylmethanesulfonyl fluoride, 0.1 µm okadaic acid, 50 mm sodium fluoride, and 10 µg/ml each of leupeptin, aprotinin, and pepstatin). The lysates were sonicated and centrifuged at 20,800 x g for 15 min. Protein concentrations were determined by the bicinchoninic acid method (Pierce, Rockford, IL). Where indicated, for the preparation of mitochondria and cytosolic fractions the harvested cells were homogenized by nitrogen cavitation (200 p.s.i., for 5 min) in a cell disruption bomb (Parr Instrument Co., Moline, IL) and fractions were prepared as previously described [22]. Samples were mixed with Laemmli sample buffer (2% SDS), placed in a boiling water bath for 5 min, proteins were resolved in SDS-polyacrylamide gels, transferred to nitrocellulose, and incubated with antibodies to phospho-ser473-akt, phospho-thr308-akt, phospho-ser9-gsk3β, phospho-thr180/tyr182-p38, phospho- Ser380-PTEN (phosphatase and tensin homolog deleted on chromosome ten), phospho- Thr202/Tyr204-ERK1/2 (extracellular signal-regulated kinases 1 and 2), phospho- Ser241-PDK1 (phosphoinositide-dependent protein kinase 1), phospho-ser380-p90rsk (Cell Signaling, Beverly, MA), heat shock factor-1 (HSF-1; NeoMarkers, Fremont, CA), total Akt (Sigma), total GSK3β (BD-PharMingen/Transduction Laboratories, San Diego, CA), or Hsp90β (Stressgen, Victoria, British Columbia, Canada).. Immunoblots were 20

28 developed using horseradish peroxidase-conjugated goat anti-mouse, or goat anti-rabbit IgG, followed by detection with enhanced chemiluminescence, and the protein bands were quantitated with a densitometer. 3. Results The transcription factor HSF-1 is a well-characterized Hsp90 client protein. When complexed with active Hsp90, HSF-1 is retained in the cytosol, but it translocates to the nucleus when Hsp90 is inhibited or recruited to bind other proteins [23, 24]. Therefore, the nuclear level of HSF-1 can provide a readout of inhibition of Hsp90, and this was used to determine the rate of inhibition of Hsp90 after treating of SH-SY5Y cells with geldanamycin. In untreated SH-SY5Y cells, little HSF-1 was present in the nucleus, whereas treatment with 1 µm geldanamycin caused a rapid accumulation of HSF-1 in the nucleus (Figure 1A). There was a large increase evident within 10 min, and maximal accumulation of nuclear HSF-1 was attained after 20 to 40 min of geldanamycin treatment. This fast accumulation of HSF-1 in the nucleus confirms that the inhibition of Hsp90 by geldanamycin is rapid. To test if Hsp90 buffers the magnitude of Akt activation by intracellular signaling, SH-SY5Y cells were stimulated with IGF-1 (50 ng/ml) to activate the PI3K/Akt signaling pathway with or without pretreatment with 10 µm geldanamycin. IGF-1 treatment substantially increased Akt phosphorylation, as indicated by large increases in the levels of both phospho-thr308-akt and phospho-ser473-akt (Figure 1B). The dual phosphorylation of Akt occurred within 5 min of treatment with IGF-1 and remained above control levels through 20 min of incubation. The same treatment with IGF-1 after geldanamycin pretreatment resulted in much larger increases in the dual 21

29 phosphorylation of Akt. Quantitative analyses showed that with geldanamycin treatment the rapid stimulation (5 min) of phospho-thr308-akt was nearly double the values in cells stimulated with IGF-1 alone (Figure 1C). Similar increases caused by geldanamycin were evident 10 and 20 min after stimulation with IGF-1. Treatment with IGF-1 also stimulated an increase in the phosphorylation of the Akt substrate GSK3β on serine-9, and this was greater in cells pretreated with geldanamycin (Figure 1B), verifying that the phosphorylated Akt was active. These results indicate that inhibition of Hsp90 allowed more Akt to be recruited and activated by the IGF-1-stimulated signaling pathway. Contrasting with geldanamycin's facilitation of rapid signaling to Akt, long-term treatment with geldanamycin (24 hr) to inhibit Hsp90 caused a large decrease in the level of Akt in SH-SY5Y cells (Figure 1D), as reported previously in SH-SY5Y and other cell types [16, 19]. With this treatment there also was no detectable phosphorylated Akt and there was a large decrease in the level of phospho-ser9-gsk3β, indicative of inactivation of Akt, whereas the total level of GSK3β was unaffected, indicating that the loss of Akt stems from its role as a Hsp90 client protein. These responses demonstrate that Akt in SH-SY5Y cells responds similarly to inhibition of Hsp90 following long-term treatment with geldanamycin as has been observed in other cells. To test if the promotion of Akt activation by geldanamycin-induced inhibition of Hsp90 is a general response, we examined another cell type, HEK 293 cells, and used a different stimulant, insulin. Treatment with insulin (100 ng/ml) caused a rapid and robust increase in the phosphorylation of Akt at both Thr308 and Ser473 in HEK cells (Figure 2). Quantitative analysis of several experiments demonstrated that pretreatment with geldanamycin increased the insulin-stimulated Akt phosphorylation by approximately 22

30 four-fold (Figure 2). Thus, as occurred with IGF-1 stimulation in SH-SY5Y cells, inhibition of Hsp90 allowed greater activation of Akt by the insulin-stimulated signaling pathway in HEK 293 cells. The previous results revealed that geldanamycin treatment promoted signaling to Akt induced by receptor-saturating concentrations of IGF-1 or insulin. Lower concentrations of IGF-1 were used to compare geldanamycin's ability to potentiate activation of Akt at various signal strengths. Using 5, 10, 25, or 50 ng/ml IGF-1 to stimulate SH-SY5Y cells, it was evident that even at the lowest concentration with the weakest signal, geldanamycin greatly enhanced IGF-1-induced activation of Akt (Figure 3A). In addition to geldanamycin, two other inhibitors of Hsp90, 17-AAG and radicicol, also potentiated IGF1-induced phosphorylation of Akt and the subsequent increase in phospho-ser9-gsk3β (Figure 3B). Although geldanamycin inhibits Hsp90, it has been reported not to cause the dissociation of Hsp90 from Akt (11). We tested if that also held true with IGF-1 stimulation by measuring the amount of Hsp90 that coimmunoprecipitated with Akt. Equal amounts of Hsp90 co-immunoprecipitated with Akt following treatment with geldanamycin, IGF-1, or both agents (Figure 3C), consistent with previous conclusions that Hsp90 and Akt remain in a complex regardless of the inhibitory effect of geldanamycin. Although Akt is predominantly a cytosolic enzyme, we recently found that it translocates to the mitochondria following stimulation [22]. Therefore, we tested if the potentiating effect of geldanamycin was selective for either of these pools of Akt. In the cytosol, the dual phosphorylation of Akt was increased in cells treated with both geldanamycin and IGF-1 (Figure 3D). Two effects are evident in mitochondria: the 23

31 translocation of Akt into the mitochondria and the increase in Akt phosphorylation. The level of Akt in the mitochondria increased following IGF-1 treatment, and this mitochondrial import of Akt was unaltered by geldanamycin pretreatment. However, after geldanamycin treatment there were large increases in both phosphorylation sites of Akt that accumulated in the mitochondria. Thus, the IGF-1-stimulated phosphorylation of both cytosolic and mitochondrial pools of Akt were increased by geldanamycin pretreatment. The activation of Akt is regulated by PDK1 and by PTEN. Phosphorylation of threonine-308 of Akt is mediated by PDK1, a kinase thought to be constitutively active that phosphorylates Akt after Akt translocates to the 3-phosphoinositides in the plasma membrane synthesized by the action of PI3K [9]. We tested if geldanamycin treatment altered the active state of PDK1 which can be detected by immunoblot measurement of the active phospho-pdk-1. Treatment with IGF-1 alone or with geldanamycin did not change the level of phospho-pdk1 (Figure 3E), indicating that the facilitation of Akt phosphorylation by geldanamycin was not due to increased activation of PDK1. Acting in opposition to PI3K, PTEN mediates the hydrolysis of 3-phosphoinositides, thereby reducing signaling to Akt, and is active when phosphorylated on serine 380 (Figure 3A). As with PDK1, neither treatment with IGF-1 alone nor with geldanamycin plus IGF-1 altered the level of phospho-pten, indicating that increased Akt activation after geldanamycin treatment was not due to inhibition of PTEN. These results support the hypothesis that inhibition of Hsp90 does not alter the magnitude of the signal directed to Akt, but most likely reflects a direct interaction of Akt with Hsp90. 24

32 The previous results indicate that active HSP90 limits signaling-induced phosphorylation of Akt since this response to IGF-1 or to insulin was greater following inhibition of Hsp90 with geldanamycin. Therefore, we examined if inhibition of HSP90 with geldanamycin influenced the activation of other kinases after stimulation with IGF- 1, including p38, ERK1/2, and p90rsk (Figure 3E). Each of these proved to give a unique signature following stimulation with IGF-1 alone and in combination with geldanamycin. There was virtually no activation of p38 following stimulation with IGF- 1, but in the presence of geldanamycin there was clearly strong phosphorylation of p38 stimulated by IGF-1 (Figure 3E). The phosphorylation of p90rsk differed from the other kinases in that alone both IGF-1 and geldanamycin produced equivalent large increases, and the two together caused approximately an additive effect (Figure 3E). ERK1/2 was rapidly phosphorylated following IGF-1 treatment, which was maximal after 5 min and then diminished after 10 and 20 min. Pretreatment with geldanamycin did not alter the maximal phosphorylation of ERK1/2 but slightly reduced the rate of dephosphorylation at 10 and 20 min after IGF-1 treatment (Figure 3E). Thus, Hsp90 has widespread buffering effects on the activities of signaling pathways, and these vary in magnitude and duration. Thus, inhibition of Hsp90 with geldanamycin causes not a general facilitation of IGF-1-induced signaling, but selectively promotes specific downstream signals. Inhibition of Hsp90 with geldanamycin could increase stimulant-induced Akt phosphorylation either by blocking the dephosphorylation of phosphorylated Akt or by enhancing the phosphorylation process. To test if dephosphorylation was affected, we measured the dephosphorylation of phospho-thr308-akt that followed inhibition of PI3K with LY Stimulation with IGF-1 activated Akt, this was potentiated by 25

33 geldanamycin pretreatment, and a subsequent incubation for 2.5 min with 10 µm LY caused an approximately 80% dephosphorylation of phospho-thr308-akt (Figure 4A and B). Pretreatment with geldanamycin increased the level of phospho- Thr308-Akt stimulated by IGF-1, but subsequent treatment with 10 µm LY caused only a 33% decline. These results indicate that inhibition of Hsp90 with geldanamycin inhibits the phosphatase-mediated dephosphorylation of phospho-akt leading to increased levels in the presence of geldanamycin. 4. Discussion The captivating concept of genetic buffering by Hsp90 [5] has garnered widespread experimental and theoretical support as a mechanism capable of "concealing the effects of genetic and environmental variation on phenotype" [25]. Such an intriguing concept naturally raises many questions. Among those that we found especially interesting are the questions of precisely how such buffering is achieved, and whether this mechanism is somehow reserved only for aberrant signaling activities. By definition, buffering implies a dampening of the magnitudes of fluctuations. Therefore, we examined two hypotheses. First, that inhibition of Hsp90 would allow a greater magnitude of signaling activity than in cells with active Hsp90. Second, that the affected signaling need not be aberrant, as there is not an obvious mechanism by which Hsp90 could distinguish aberrant from normal signaling. To test these hypotheses, we focused on a signaling pathway that is well-characterized as providing growth- and survivalpromoting cellular signals, that induced by IGF-1 or insulin leading to activation of Akt. The results show that buffering by Hsp90 can indeed be achieved by dampening the 26

34 magnitude of intracellular signals, and also demonstrate that this buffering action is not limited to aberrant signals. Perhaps the most predominant post-translational modification that is used to modulate the activity of enzymes is phosphorylation. In the case of Akt, dual phosphorylation on Thr308 and Ser473 greatly increases the catalytic activity of Akt, allowing it to efficiently phosphorylate many known substrates [9, 10]. This study of Akt showed that Hsp90 limits the signaling activity of Akt by facilitating the removal of the regulatory phosphates. This was evident because inhibition of Hsp90 greatly diminished this dephosphorylation deactivating process, allowing a signal (the dual phosphorylation of Akt) of greater magnitude than that achieved in the presence of active Hsp90. There is already solid evidence that protein phosphatases associate with Hsp90 and that Hsp90 can exert a regulatory influence on their activities, since several reports have demonstrated that Hsp90 binds to a number of protein phosphatases, such as protein phosphatase 2A [26], protein phosphatase 2B [27], and protein phosphatase 5 [28]. Thus, our data indicates that Hsp90 facilitates phosphatase-mediated dephosphorylation of Akt, thereby buffering the magnitude of the phosphorylation-dependent signaling activity that can be achieved. This regulatory interaction is very similar to the Hsp72-induced suppression of c- Jun N-terminal kinase (JNK) phosphorylation and activation that has been described in detail by Sherman and colleagues [29, 30]. These investigators demonstrated that Hsp72 promotes the activity of a JNK phosphatase to limit the phosphorylation of JNK. Dissociation of Hsp72 from the phosphatase by damaged proteins inactivated the phosphatase, allowing increased phosphorylation and activation of JNK [31]. Similarly, 27

35 our data indicates that Hsp90 promotes the activity of an Akt phosphatase and that inhibition of Hsp90 with geldanamycin impairs the ability of Hsp90 to promote the activity of the phosphatase, thereby allowing enhanced phosphorylation of Akt. This buffering of Akt phosphorylation could rely on an interaction between Hsp90 and a phosphatase or Hsp90 could affect the conformation of Akt to modulate its accessibility to phosphorylation or dephosphorylation. Compared with the rapid responses of Akt-associated signaling described here in which inhibition of Hsp90 is shown to promote the phosphorylation of Akt, quite different results are obtained in studies of the longer-term effects of inhibition of Hsp90. The formation of a complex between Hsp90 and Akt was first reported by Sato and colleagues [11]. Subsequently, most research on the Hsp90-Akt interaction has focused on regulation of the stability of Akt, because inhibition of Hsp90 for several hours promoted the degradation of Akt [16]. Consequently, many reports have verified that Akt levels are down-regulated following six or more hours of inhibition of Hsp90 [16-21]. However, the initial effects of inhibition of Hsp90 on intracellular signaling-induced phosphorylation of Akt remained uninvestigated, and the current results demonstrate that the initial outcome of inhibiting Hsp90 is facilitation of the stimulant-induced phosphorylation of Akt. Thus, although geldanamycin-bound Hsp90 causes destabilization of Akt, this follows a more immediate facilitation of the phosphorylation of Akt. In addition to Hsp90 dampening signaling associated with phosphorylation of Akt, Hsp90 also was found to regulate IGF-1-induced phosphorylation of several other intracellular signaling kinases. These experiments revealed a variety of regulatory actions 28

36 of Hsp90 on the magnitude and/or duration of signaling-induced phosphorylation of p38, p90rsk, and ERK1/2. While the dampening of Akt signaling by active Hsp90 was found to be due to the facilitation of dephosphorylation, the mechanisms underlying the modulation of p38, P90RSK, and ERK1/2 by Hsp90 remain to be determined. These multiple effects of Hsp90 on different kinases reveals the individuality of its modulatory influences on different intracellular signaling proteins, actions that may contribute to the capability of Hsp90 to buffer against genetic and environmental stresses by limiting the magnitude of signaling fluctuations within cells. The capability of Hsp90 to regulate phosphorylation of Akt reported here adds to a growing number of modulatory actions accomplished by Hsp90 binding to client proteins, the diversity of which likely underlies its ability to buffer cellular events. Hsp90 is currently known to associate with over 100 proteins (reviewed in [32]). These proteins include kinases, receptors, transcription factors, and a variety of others. Hsp90 has multifaceted effects on its variety of client proteins. These include assistance in the maturation or stabilization of clients, such as the viral and cellular proto-oncogene Src [33, 34], maintenance of clients in inactive conformations, as with glucocorticoid receptors [35], and other actions. Closest to the action reported here, inhibition of Hsp90 with geldanamycin previously was found to cause a rapid activation of the doublestranded RNA-activated kinase PKR which coincided with dissociation of Hsp90 from PKR [36], differing with the stable association of Hsp90 with Akt even after treatment with geldanamycin. Although Hsp90 limits phosphorylation of Akt, it is interesting that Hsp90 can function as a molecular scaffold to facilitate Akt-dependent phosphorylation of endothelial nitric oxide synthase [12] and of telomerase reverse transcriptase [37]. 29

37 Thus, the interactions of Hsp90 and Akt are complex and include attenuated phosphorylation of Akt but facilitated stability and phosphorylation by Akt of certain targets. Thus, Hsp90 has a diverse array of functions that often serve to regulate the signaling activities of its clients, regulatory actions that may underlie its capacity to buffer genetic and environmental influences on cellular metabolism. In summary, Hsp90 was shown to dampen signaling associated with activation of Akt as well as several other intracellular signaling kinases. These actions may contribute to the capability of Hsp90 to buffer against genetic and environmental stresses by limiting the magnitude of signaling fluctuations within cells. Acknowledgements This research was supported by the National Institutes of Health. 30

38 References [1] Young, J.C., Moarefi, I. and Hartl, F.U. (2001) J. Cell Biol. 154, [2] Rutherford, S.L. and Lindquist, S. (1998) Nature 396, [3] Queitsch, C., Sangster, T.A. and Lindquist, S. (2002) Nature 417, [4] Milton, C.C., Huynh, B., Batterham, P., Rutherford, S.L. and Hoffmann, A.A. (2003) Proc. Natl. Acad. Sci. USA 100, [5] Sangster, T.A, Lindquist, S. and Queitsch, C. (2004) Bioessays 26, [6] Nollen, E.A.A. and Morimoto, R.I. (2002) J. Cell Science 115, [7] Macario, A.J.L. and de Marcario, E.C. (2000) Int. J. Clin. Lab. Res. 30, [8] Takayama, S., Reed, J.C. and Homma, S. (2003) Oncogene 22, [9] Brazil, D.P. and Hemmings, B.A. (2001) Trends Biochem. Sci. 26, [10] Brunet, A., Datta, S.R. and Greenberg, M.E. (2001) Current Opin. Neurobiol. 11, [11] Sato, S., Fujita, N. and Tsuruo, T. (2000) Proc. Natl. Acad. Sci. USA 97, [12] Fontana, J., Fulton, D., Chen, Y., Fairchild, T.A., McCabe, T.J., Fujita, N., Tsuruo, T. and Sessa, W.C. (2002) Circ. Res. 90, [13] Fujita, N., Sato, S., Ishida, A. and Tsuruo, T. (2002) J. Biol. Chem. 277, [14] Whitesell, L., Mimnaugh, E.G., De Costa, B., Myers, C.E. and Neckers, L.M. (1994) Proc. Natl. Acad. Sci. USA 91, [15] Grenert, J.P., Sullivan, W.P., Fadden, P., Haystead, T.A., Clark, J., Mimnaugh, E., Krutzsch, H., Ochel, H.J., Schulte, T.W., Sausville, E., Neckers, L.M. and Toft, D.O. (1997) J. Biol. Chem. 272, [16] Basso, A.D., Solit, D.B., Chiosis, G., Giri, B., Tsichlis, P. and Rosen, N. (2002) J. Biol. Chem. 277, [17] Hostein, I., Robertson, D., DiStefano, F., Workman, P. and Clarke, P.A. (2001) Cancer Res. 61,

39 [18] Nimmanapalli, R., O'Bryan, E., Kuhn, D., Yamaguchi, H., Wang, H.G. and Bhalla, K.N. (2003) Blood 102, [19] Kim, S., Kang, J., Hu, W., Evers, B.M. and Chung, D.H. (2003) Int. J. Cancer 103, [20] Beliakoff, J., Bagatell, R., Paine-Murrieta, G., Taylor, C.W., Lykkesfeldt, A.E. and Whitesell, L. (2003) Clin. Cancer Res. 9, [21] Doong, H., Rizzo, K., Fang, S., Kulpa, V., Weissman, A.M. and Kohn, E.C. (2003) J. Biol. Chem. 278, [22] Bijur, G.N. and Jope, R.S. (2003) J. Neurochem. 87, [23] Shi, Y., Mosser, D.D. and Morimoto, R.I. (1998) Genes Dev. 12, [24] Zou, J., Guo, Y., Guettouche, T., Smith, D.F. and Voellmy, R. (1998) Cell 94, [25] Rutherford, S.L. (2000) Bioessays. 22, [26] Yoshinari, K., Kobayashi, K., Moore, R., Kawamoto, T. and Negishi, M. (2003) FEBS Lett. 548, [27] Someren, J.S, Faber, L.E, Klein, J.D. and Tumlin, J.A. (1999) Biochem. Biophys. Res. Commun. 260, [28] Silverstein, A.M., Galigniana, M.D., Chen, M.S., Owens-Grillo, J.K., Chinkers, M. and Pratt, W.B. (1997) J. Biol. Chem. 272, [29] Gabai, V.L., Meriin, A.B., Mosser, D.D., Caron, A.W., Rits, S., Shifrin, V.I. and Sherman, M.Y. (1997) J. Biol. Chem. 272, [30] Gabai, V.L. and Sherman, M.Y. (2002) J. Appl. Physiol. 92, [31] Meriin, A.B., Yaglom, J.A., Gabai, V.L., Zon, L., Ganiatsas, S., Mosser, D.D., Zon, L. and Sherman, M.Y. (1999) Mol. Cell. Biol. 19, [32] Pratt, W.B. and Toft, D.O. (2003) Exp. Biol. Med. 228, [33] Xu, Y. and Lindquist, S. (1993) Proc. Natl. Acad. Sci. USA 90, [34] Xu, Y., Singer, M.A. and Lindquist, S. (1999) Proc. Natl. Acad. Sci. USA 96, [35] Pratt, W.B. (1993) J. Biol. Chem. 268, [36] Donze, O., Abbas-Terki, T. and Picard, D. (2001) EMBO J. 20,

40 [37] Haendeler, J., Hoffmann, J., Rahnam, S., Zeiher, A.M. and Dimmeler, S. (2003) FEBS Lett. 536,

41 A nuclear HSF-1 Geldanamycin (min) B Phospho-Thr308-Akt Phospho-Ser473-Akt Phospho-Ser9-GSK3β IGF-1 (min) Geldanamycin C 2.5 * * 2 Phospho-Thr308-Akt Arbitrary area units * Ctl GA GA+ IGF IGF 5 min GA+ IGF IGF 10 min GA+ IGF IGF 20 min 34

42 D Geldanamycin 0 24 hr Geldanamycin 0 24 hr Geldanamycin 0 24 hr Geldanamycin 0 24 hr Total Akt Phospho-Ser473-Akt Total GSK3β Phospho-Ser9-GSK3β Figure 1. Inhibition of Hsp90 by geldanamycin promotes IGF-1-induced Akt activation (A) SH-SY5Y cells were treated with 1 µm geldanamycin for min. Nuclear fractions were prepared and nuclear levels of HSF-1 were measured by immunoblot analysis. (B) SH-SY5Y cells were treated with 50 ng/ml IGF-1 for 0, 5, 10, or 20 min, with or without pretreatment with 10 µm geldanamycin for 30 min. Levels of phospho- Thr308-Akt, phospho-ser473-akt, and phospho-ser9-gsk3β were measured by immunoblot analysis. (C) Quantitative values for phospho-thr308-akt were obtained from three independent experiments and are Means ± SEM. *p < Ctl = control untreated samples; GA = geldanamycin. (D) SH-SY5Y cells were treated with 10 µm geldanamycin for 24 hr followed by measurements of total Akt, phospho-ser473-akt, total GSK3β, and phospho-ser9-gsk3β by immunoblot analysis. 35

43 Phospho-Thr308-Akt Phospho-Ser473-Akt Insulin (min) Geldanamycin * * 1.5 * Phospho-Thr308-Akt arbitrary area units Ctl GA GA+ INS INS 5 min GA+ INS INS 10 min GA+ INS INS 20 min Figure 2. Inhibition of Hsp90 by geldanamycin promotes insulin-induced Akt activation HEK 293 cells were treated with 100 nm insulin (INS) for 0, 5, 10, or 20 min, with or without pretreatment with 10 µm geldanamycin (GA) for 30 min. Levels of phospho- Thr308-Akt and phospho-ser473-akt were measured by immunoblot analysis. Quantitative values for phospho-thr308-akt were obtained from three independent experiments and are Means ± SEM. *p < 0.05 compared to matched samples not treated with geldanamycin. 36

44 A Phospho- Thr308-Akt Phospho- Ser473-Akt IGF-1 (ng/ml) Geldanamycin B Phospho- Thr308-Akt Phospho- Ser9-GSK3β Ctl AAG IGF AAG + IGF Ra Ra + IGF C IP: Akt HSP90 Akt No IP Ctl GA IGF GA + IGF D Phospho- Thr308-Akt Phospho- Ser473-Akt Total Akt Cytosol Ctl GA IGF GA+ IGF Mitochondria Ctl GA IGF GA+ IGF 37

45 E Phospho-PDK1 Phospho-PTEN Phospho-p38 Phospho-p90RSK Phospho-ERK1/2 IGF-1 (min) Geldanamycin Figure 3. Inhibition of HSP90 promotes IGF-1-induced intracellular signaling (A) SH-SY5Y cells were pretreated with 10 µm geldanamycin for 30 min followed by stimulation for 10 min with 0, 5, 10, 25, or 50 ng/ml IGF-1. Phospho-Thr308-Akt and phospho-ser473-akt were examined by immunoblot analysis. (B) SH-SY5Y cells were pretreated with 25 µm 17-AAG or 25 µm radicicol (Ra) for 30 min followed by stimulation for 10 min with 5 ng/ml IGF-1 and phospho-thr308-akt and phospho-ser9- GSK3β were measured. (C) SH-SY5Y cells were pretreated with 10 µm geldanamycin (GA) for 30 min followed by treatment with 50 ng/ml IGF-1 for 10 min. Akt was immunoprecipitated, and the amount of Hsp90 that co-immunoprecipitated was measured. The amount of Akt that was immunoprecipitated from each sample also is shown. (D) SH-SY5Y cells were pretreated with 10 µm geldanamycin (GA) for 30 min followed by treatment with 50 ng/ml IGF-1 for 15 min. The cytosolic and mitochondrial fractions were prepared and immunoblotted for phospho-thr308-akt, phospho-ser473- Akt, and total Akt. Ctl = control untreated samples. (E) SH-SY5Y cells were treated with 50 ng/ml IGF-1 for 0, 5, 10, or 20 min, with or without pretreatment with 10 µm geldanamycin for 30 min. Levels of phospho-ser241-pdk1, phospho-ser380-pten, phospho-thr180/tyr183-p38, phospho-ser380-p90rsk, and phospho-thr202/tyr204- ERK1/2, were measured by immunoblot analyses. 38

46 A Phospho-Thr308-Akt Phospho-Ser473-Akt Geldanamycin IGF LY B 3.5 Phospho-Thr308-Akt arbitrary units * 0 LY IGF LY IGF + GA Figure 4. Inhibition of Hsp90 blocks dephosphorylation of Akt. Cells were treated with 10 µm geldanamycin (GA) 30 min prior to stimulation with IGF- 1 (50 ng/ml) for 10 min. Following IGF-1 treatment cells were treated with 10 µm LY for 2.5 min. (A) Levels of phospho-thr308-akt and phospho-ser473-akt were measured by immunoblot analyses. (B) Quantitative values for phospho-thr308- Akt were obtained from three independent experiments and are Means ± SEM. *p<0.05 compared to matched sample not treated with LY

47 GSK3 AND HSP105 INTERACT AND REGULATE ER STRESS-INDUCED APOPTOTIC SIGNALING by Gordon P. Meares, Anna A. Zmijewska and Richard S. Jope 40

48 ABSTRACT Stress of the endoplasmic reticulum (ER stress) is caused by the accumulation of misfolded proteins, which occurs in many neurodegenerative diseases. ER stress can lead to adaptive responses or apoptosis, both of which follow activation of the unfolded protein response (UPR). The transcription factor CHOP is a key determinant in this decision, when ER stress is sufficiently large or prolonged UPR signaling activates CHOP and this leads to apoptotic signaling and cell death. We found that HSP105 is intimately involved in ER stress-induced signaling as it binds both glycogen synthase-3 (GSK3), a promoter of ER stress-induced apoptosis, and GRP78, a key component of the UPR. Knockdown of HSP105 did not alter UPR signaling and CHOP activation after ER stress, but diverted signaling to a caspase-3-independent cell death pathway, indicating that HSP105 is necessary for apoptotic signaling after UPR activation by ER stress. Like HSP105, its binding partner GSK3 also was necessary for apoptotic signaling emanating from the UPR, as inhibition of GSK3 greatly reduced ER stress-induced CHOP expression and caspase-3 activation. Thus, HSP105 appears to chaperone the responses to ER stress through its interactions with GRP78 and GSK3, and without HSP105 cell death following ER stress proceeds by a non-caspase-3-dependent process. Furthermore, by regulating ER stress-induced CHOP expression GSK3 provides a fulcrum to balance UPR-associated survival pathways, which inhibit its activity, with apoptotic signaling, which is enhanced when GSK3 is activated. Thus, both HSP105 and GSK3 regulate the outcome of the UPR induced by ER stress. 41

49 Introduction The endoplasmic reticulum (ER) is a membranous organelle that provides a unique environment for the proper folding and maturation of secreted and membranebound proteins. Diseases, environmental factors, and pharmacological agents can perturb ER function and lead to the accumulation of misfolded proteins and ER stress (Boyce and Yuan, 2006; Lindholm et al., 2006). When the ER stress is severe enough to preclude recovery from excessive accumulation of misfolded proteins, cells undergo controlled death by apoptosis (Szegezdi et al., 2006; Yoshida, 2007). This apoptotic method of cell death is important because it evolves internally, enabling the intracellular inactivation of potentially toxic molecules that could be released in necrotic cell death. Eukaryotic cells have evolved an elegant but complex system to detect and react to misfolded proteins within the ER, encompassing both adaptive and apoptotic signals (Rutkowski and Kaufman, 2004; Boyce and Yuan, 2006; Lindholm et al., 2006). Thus, cells attempt to rectify the misfolded proteins while at the same time preparing for apoptosis should the ER stress be overwhelming (Szegezdi et al., 2006; Yoshida, 2007). The response to ER stress involves activation of the highly conserved unfolded protein response (UPR). The UPR includes activation of the trans-er membrane molecules PKR-ER like kinase (PERK), inositol requiring enzyme 1 (IRE1), and activating transcription factor 6 (ATF6). Each of these proteins is held inactive by GRP78 in the ER under basal conditions, and is activated following release when GRP78 is recruited away to misfolded proteins (Bertolotti el al., 2000; Shen et al., 2002). Activation of these UPR mediators leads to attenuation of translation and increased expression of a subset of stress-responsive genes(schroder and Kaufman, 2005; Yoshida, 2007). 42

50 The survival of cells following activation of the UPR depends on the severity and duration of the stress and is largely determined by the proapoptotic molecule C/EBP homologous transcription factor protein (CHOP, also called GADD153) (Rutkowski et al., 2006). The expression of CHOP, a member of the C/EBP family of transcription factors, is upregulated following ER stress by the ATF4 and ATF6 transcription factors (Oyadomari and Mori, 2004). CHOP has been reported to promote apoptosis via upregulation of GADD34, an eif2α phosphatase, to restore protein synthesis (Marciniak et al., 2004), increased oxidation of proteins (Marciniak et al., 2004), and suppressed expression of anti-apoptotic Bcl-2 (McCullough et al., 2001). Conversely, CHOP deficiency protects from ER stress-induced apoptosis (Zinszner et al., 1998). CHOP is a short-lived protein, so when cells are able to dissipate misfolded proteins, CHOP is rapidly degraded and its expression is reduced, allowing cells to survive. However, prolonged or severe ER stress maintains high CHOP levels to promote apoptotic signaling that induces activation of the executioner caspase, caspase-3, and cell death (Rutkowski et al., 2006). In addition to CHOP, many other proteins have been identified that regulate the ultimate decision between adaptation and death following ER stress (Xu et al., 2005). One of these is glycogen synthase-3 (GSK3), a constitutively active Ser/Thr kinase composed of two similar, but not identical, isoforms, GSK3α and GSK3β (Jope and Johnson, 2004). We previously found (Song et al.,, 2002) that GSK3 inhibitors reduced apoptosis induced by several agents that cause ER stress, including thapsigargin (an ER Ca ++ -ATPase inhibitor) and tunicamycin (an N-linked glycosylation inhibitor), which has been confirmed in numerous studies (Chen et al., 2004; Hiroi et al., 2005; Kim et al., 43

51 2005; Srinivasan et al., 2005; Brewster el al., 2006; Takadera et al., 2006). Despite this considerable evidence that GSK3 promotes intrinsic apoptotic signaling following ER stress, the underlying mechanisms remain uncertain (Beurel and Jope, 2006). Heat shock proteins (hsps) also have been identified as regulators of the ER stress response (Marcu et al., 2002; Oyadomari et al., 2006). Hsps are molecular chaperones that escort and regulate the actions of many intracellular proteins. Hsps bind newly synthesized or misfolded proteins to optimize folding and function, whereas irreparably damaged proteins are directed by hsps to sites of degradation (Young et al., 2004). Additionally, hsps regulate multiple steps in the apoptotic signaling cascade (Beere, 2005). Together these actions likely underlie the widely reported antiapoptotic, or survival-promoting, capabilities of hsps. Additionally, hsps play an important role in cellular signaling by ensuring that many key signaling proteins maintain optimally active conformations (Caplan et al., 2007). Thus, hsps generally are considered as immensely helpful proteins that ensure proteins fold and function properly and damage is controlled. Although HSP105 is one of the major mammalian heat shock proteins, little is known about its actions and client proteins compared with the well-known HSP70 and HSP90. HSP105 is a nucleotide exchange factor for HSP70 (Dragovi et al., 2006; Raviol et al., 2006) and prevents aggregation of mutant androgen receptors (Ishihara et al., 2003) and Zn/Cu superoxide dismutase (Yamashita et al., 2007). HSP105 attenuates staurosporine-induced apoptosis (Yamagishi et al., 2006), but overexpressed HSP105α in mouse embryonal F9 cells enhanced apoptosis in response to oxidative stress (Yamagishi et al., 2002a). However, the role of HSP105 in response to ER stress has not previously been examined. 44

52 If a cell is lethally damaged, especially during ER stress when many proteins are misfolded, we hypothesized that there must be a mechanism to ensure that mediators of the apoptotic cascade remain functional in a destructive environment. A logical candidate for this role is a molecular chaperone, a protein that serves to optimize protein and cellular functions even though in this case the outcome is cell death. In this report we present evidence that one function of HSP105 is to chaperone the process of apoptosis following ER stress. Therefore, the present study tested the effects of HSP105 on ER stress-induced apoptosis. We identified HSP105 and GSK3 as interacting proteins that each supports ER stress-induced apoptosis. This data demonstrates that ER stress-induced caspase-3 activation requires HSP105 and is facilitated by GSK3 through upregulation of CHOP. MATERIALS AND METHODS Cell Culture- SH-SY5Y human neuroblastoma cells where grown in RPMI 1640 medium supplemented with 10% horse serum and 5% fetal clone II (HyClone, Logan, UT). Cells were placed in serum-free media approximately 16 h prior to experimental treatments. HEK293 cells were grown in Dulbecco s modified eagle medium supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA) and 15 mm HEPES (Cellgro, Herndon, VA). All media were supplemented with 2 mm L-glutamine, 100 units/ml penicillin, and 100 µg/ml streptomycin (Cellgro), in humidified, 37 C chambers with 5% CO 2. Cells were treated for the indicated times with thapsigargin, tunicamycin (Alexis, San Diego, CA), camptothecin, LiCl (Sigma Chemical Co., St. Louis, MO), kenpaullone, 45

53 6-bromoindirubin-3-oxime (Calbiochem, San Diego, CA), and CHIR99021 (University of Dundee, Dundee, Scotland). Tissue preparation-adult, male C57BL/6 mice (Frederick Cancer Research, Frederick, MD), were decapitated, and brains were rapidly dissected in ice-cold saline. The cerebral cortex was homogenized in ice-cold lysis buffer containing 10 mm Tris-HCl, ph 7.4, 150 mm NaCl, 1 mm EDTA, 1 mm EGTA, 0.5% NP-40, 10 µg/ml leupeptin, 10 µg/ml aprotinin, 5 µg/ml pepstatin, 1 mm phenylmethanesulfonyl fluoride, 1 mm sodium vanadate, 50 mm sodium fluoride, and 100 nm okadaic acid. The homogenates were centrifuged at 20,800xg for 10 min to remove insoluble debris. Immunoblotting- Cells were washed twice with phosphate-buffered saline (PBS) and were lysed with IP lysis buffer (20 mm Tris, ph 7.5, 150 mm NaCl, 2 mm EDTA, 2 mm EGTA, 0.5% NP-40, 1 mm sodium orthovanadate, 100 µm phenylmethanesulfonyl fluoride, 0.1 µm okadaic acid, 50 mm sodium fluoride, and 10 µg/ml each of leupeptin, aprotinin, and pepstatin). The lysates were sonicated and centrifuged at 20,800xg for 15 min. Protein concentrations were determined by the bicinchoninic acid method (Pierce, Rockford, IL). Samples were mixed with Laemmli sample buffer (2% SDS) and placed in a boiling water bath for 5 min. Proteins were resolved in SDS-polyacrylamide gels, transferred to nitrocellulose, and the membranes were incubated with antibodies to phospho-ser9-gsk3β, phospho-ser473-akt, phospho-ser51-eif2α, total eif2α, cleaved caspase-3, cleaved poly-(adp-ribose) polymerase (PARP) (Cell Signaling, Danvers, MA), total GSK3β, active Bax (6A7)(BD-PharMingen/Transduction Laboratories, San Diego, CA), HSP90, HSP70, KDEL (Stressgen, Victoria, BC, Canada), CHOP/GADD153, GRP78 (Santa Cruz Biotechnology, Santa Cruz, CA), total GSK3 46

54 (Upstate Biotechnology Inc., Lake Placid, NY), HSP105 (Novocastra Laboratories, Newcastle upon Tyne, UK), MCL1 (Biovision, Mountain View, CA), GSK3α (Southern Biotech, Birmingham, AL), total Bax (a kind gift from Dr. Tong Zhou, University of Alabama at Birmingham), β-actin, or V5-tag (Sigma). Immunoblots were developed using horseradish peroxidase-conjugated goat anti-mouse, goat anti-rabbit, or donkey anti-goat IgG, followed by detection with enhanced chemiluminescence, and the protein bands were quantified with a densitometer. Subcellular Fractionation- Nuclear and cytosolic fractions were prepared as previously described (Bijur and Jope, 2001) with minor modifications. Cells were washed twice with PBS and then harvested in 200 µl lysis buffer (10 mm Tris, ph 7.5, 10 mm NaCl, 3 mm MgCl 2, 0.05% NP-40, 1 mm EGTA, 1 mm sodium orthovanadate, 100 µm phenylmethanesulfonyl fluoride, 0.1 µm okadaic acid, 50 mm sodium fluoride, and 10 µg/ml each of leupeptin, aprotinin, and pepstatin). Cells were centrifuged at 2700xg for 10 min at 4ºC. The supernatant was centrifuged at 20,800xg for 15 min at 4ºC to obtain the cytosolic fraction. The pellet, containing nuclei, was washed twice in 200 µl wash buffer (5 mm HEPES, ph 7.4, 3 mm MgCl 2, 1 mm EGTA, 250 mm sucrose, 0.1% BSA, with protease and phosphatase inhibitors). The pellet was then resuspended in wash buffer and layered on top of 1 ml 1 M sucrose (with protease and phosphatase inhibitors), and centrifuged at 2700xg for 10 min at 4ºC. The nuclear pellet was washed in lysis buffer containing 0.05% NP-40. The nuclear proteins were extracted by resuspending the pellet in nuclear extraction buffer (20 mm HEPES, ph 7.9, 300 mm NaCl, 1.5 mm MgCl 2, 0.2 mm EDTA, 0.1 mm β-glycerophosphate, 1 mm sodium orthovanadate, 100 µm phenylmethanesulfonyl fluoride, 0.1 µm okadaic acid, 50 mm sodium fluoride, and 47

55 10 µg/ml each of leupeptin, aprotinin, and pepstatin) and incubating on ice for 30 min. After extraction, the nuclear samples were centrifuged at 20,800xg for 15 min at 4ºC, and the supernatant was retained as the nuclear extract. Immunoprecipitation- Immunoprecipitations were performed using sheep anti-mouse IgG Dynabeads (Invitrogen, Carlsbad, CA). The Dynabeads, 10 µl of slurry per sample, were washed twice in PBS/0.1% BSA, and all washes were performed using a magnetic particle separator. The beads were incubated with 1 µg anti-gsk3β antibody, 1 ug anti- V5 antibody or 1 ug non-immuno isotype matched IgG (Southern Biotech) in PBS/BSA at 4ºC with end-over-end mixing overnight. The beads were then washed twice in PBS/BSA, followed by the addition of cell lysate (200 µg protein). The beads plus lysate were incubated at 4ºC for 2 h with end-over-end mixing. The beads were then washed three times with PBS/BSA. Alternatively, some immunoprecipitations of GSK3α or GSK3β were performed using anti-gsk3α- or anti-gsk3β-conjugated sepharose beads (Southern Biotech), by incubating µl of beads with cell lysate for 2 h at 4ºC with end-over-end mixing. The beads were then washed 3-4 times in IP lysis buffer. Lentivirus and shrnai- Lentiviral mediated RNAi was performed using the Lentilox3.7 vector, which co-expresses egfp, generously provided by Dr. Luk Van Parijs (Massachusetts Institute of Technology). The oligonucleotides containing the HSP105 target sequence (underlined) that were used are: sequence #2, 5 TGAAGGAGAGGACCAAGCTATTCAAGAGATAGCTTGGTCCTCTCCTTCTTTTTTC and sequence #4, 5 TGCAGTAGCCAGAGGATGTGTTCAAGAGACACATCCTCTGGCTACTGCTTTTTTC, and the reverse compliment of each oligonucleotide with an additional AGCT at the 5 end. 48

56 Sense and anti-sense oligonucleotides were annealed and ligated into the XhoI, HPA1 cut Lentilox3.7 vector. Lentiviruses were produced using protocols generously provided by Dr. James Lah (Emory University). Protocols were adapted for a four vector system. Twelve 150 mm dishes of 293FT cells (Invitrogen) were co-transfected with 300 µg of the Lentilox3.7 vector, with or without the RNAi hairpin insert, plus 150 µg each of the packaging vectors plp1, plp2, and plp/vsvg (Invitrogen) by the calcium phosphate method. The media was changed approximately 16 h after transfection, and the cells were cultured an additional h. The media was then collected, centrifuged at 500xg for 5 min, and filtered through a 0.45 µm filter. The virus-containing media was then centrifuged at 50,000xg for 2 h at 16ºC, the supernatant was discarded, and the pellet was resuspended in 500 µl PBS/tube and incubated on ice for 1 h. The viral suspension was pooled and centrifuged again at 50000xg for 2 h at 16 o C. The supernatant was discarded and the pellet was resuspended in 150 µl of PBS, aliquoted, and stored at -80ºC. The virus titers were determined by transducing HEK293 cells with 10-fold serial dilutions. 72 h post transduction the percentage of GFP positive cells was determined by flow cytometry and the titer was calculated based on the original number of cells. Titers were approximately 1x10 8 1x10 9 transducing units/ml. For HSP105 overexpression, the same procedure was used for virus production using the plenti6/v5 vector (Invitrogen) carrying the HSP105α coding region, and titered using blasticidine selection. The HSP105α was cloned by RT-PCR from the mrna of HEK293 cells. Immunocytochemistry- For cleaved caspase-3 staining, cells were plated in 24 or 48 well plates and treated as indicated. The cells were fixed in 4% paraformaldehyde for 30 min at 4ºC, then washed with PBS. The cells were incubated in blocking/permeabilization 49

57 buffer (PBS, ph 7.2, with 1% BSA, 0.2% non-fat milk, and 0.3% Triton-X 100) for 30 min at room temperature. The cells were then incubated overnight with the cleaved caspase-3 antibody diluted 1:50 in blocking buffer. The cells were washed several times with PBS, then incubated with Texas red-conjugated secondary and 1 µg/ml DAPI for 1 h at room temperature, following by thorough washing with PBS. Three random fields in each well from at least three independent experiments were imaged using a Nikon fluorescence microscope with a 20X objective and quantified using Image ProPlus software. To examine cell morphology, cells were plated on poly-d-lysine-coated coverslips and treated as indicted followed by fixation. The coverslips were washed with PBS and then incubated with 100 ng/ml Hoechst for 15 min at room temperature. XBP-1 splicing- The method to examine XBP-1 splicing was adapted from Calfon et al (2002). Cellular mrna was isolated using an RNeasy mini kit (Qiagen, Germantown, MD) following the manufacture s protocol. A fragment of XBP-1 was amplified by RT- PCR using the Superscript III one-step RT-PCR system (Invitrogen) with primers flanking the IRE1 cleavage site. The primers used were 5 AGAAGGCGCTGAGGAGGA and 5 CTGAGAGGTGCTTCCTCG. Following 40 cycles of PCR, the product was digested by PstI to remove the unspliced XBP-1 products; IRE1-mediated splicing removes the PstI site. The DNA was then resolved on a 2% agarose gel containing ethidium bromide and imaged using a molecular imager system and Quantity One software (Bio-Rad, Hercules, CA). Cell viability- Cells were grown in 24 well plates and treated as indicated. Following treatments, the media was collected and centrifuged at 1000xg for 5 min. An equal volume of media containing 1% triton X-100 was added to the wells and incubated at 50

58 37ºC for 30 min to lyse the remaining cells, followed by centrifugation at 1000xg for 5 min. The lactate dehydrogenase (LDH) assay (Roche) was then performed in duplicate on both samples from all wells. The two values obtained for each well were added together and used as the total available LDH and the percent release was calculated based on the amount of LDH in the non-detergent media divided by the total LDH obtained by lysing the cells with Triton X-100. RESULTS HSP105 client proteins HSPs are important modulators of cellular responses to misfolded protein accumulation, as occurs with ER stress, but few client proteins specific to HSP105 are known. Two approaches were used to identify HSP105-associated proteins, coimmunoprecipitation methods and the use of mass spectrometry. We tested if HSP105 associates with GSK3 because much recent evidence demonstrated that GSK3β activation contributes to ER stress-induced apoptotic signaling. To test if these proteins interact in vivo, GSK3β was immunoprecipitated from mouse cerebral cortex homogenates and immunoblotted for HSP105, which demonstrated that HSP105 co-precipitated with GSK3β (Fig. 1A). To verify this interaction, we used SH- SY5Y human neuroblastoma cells stably expressing a low level of HSP105 with a V5 tag (HSP105-V5) to allow immunoprecipitation with an antibody to V5, since we were unable to successfully immunoprecipitate HSP105 with the commercially available antibodies tested. Consistent with the in vivo data, GSK3β co-precipitated with HSP105- V5 (Fig. 1B). 51

59 To assess if stress influences the association of HSP105 with GSK3β, SH-SY5Y cells were subjected to heat shock by incubation at 45 C for 0 to 120 min, followed by immunoprecipitation of GSK3β and immunoblotting HSP105. The amount of HSP105 that co-immunoprecipitated with GSK3β was increased within 7 min of heat shock and the increased association remained evident for approximately 60 min of heat shock (Fig. 1C), suggesting that HSP105 might protect GSK3β from heat-induced damage. During apoptosis induced by heat shock or other stimuli, GSK3β can rapidly translocate to the nucleus (Bijur and Jope, 2001). Therefore, we tested if cytosolic or nuclear GSK3 associates with hsp105 and if association in each compartment was altered following heat shock. SH-SY5Y cells were subjected to heat shock for 15 or 30 min, followed by fractionation to isolate nuclei and cytoplasm. HSP105 was detected in both the cytosol and nucleus, and its subcellular distribution was not affected by heat shock (Fig. 1D), whereas heat shock increased nuclear levels of GSK3β (Fig. 1E) and GSK3α (Fig. 1F). Cytosolic GSK3β and GSK3α co-immunoprecipitated HSP105, which was modestly increased by heat shock, and increased amounts of HSP105 co-immunoprecipitated with nuclear GSK3α/β in proportion to the increased levels of GSK3α/β in the nucleus. Thus, HSP105 associates with both isoforms of GSK3 in both the cytosol and the nucleus, and heat shock-induced cell stress increases their association in both compartments. To identify additional HSP105-interacting proteins, HEK293 cells were transiently transfected with HSP105-V5 followed by immunoprecipitation with an anti- V5 antibody, resolution of proteins by electrophoresis, and identification of coimmunoprecipitated proteins by mass spectrometry. The proteins identified by MALDI- TOF included HSP70, which was previously reported to bind HSP105 (Hatayama et al., 52

60 1998), and four novel HSP105-interacting proteins (Fig. 2A). Of the new HSP105- interacting proteins that were identified, two, Sec16L and GRP78, are associated with the ER. Sec16L (KIAA0310) functions in the export of proteins from the ER (Bhattacharyya and Glick, 2007). GRP78 is particularly interesting because it is a key regulator of the cellular response to ER stress. To verify this interaction, HSP105-V5 was immunoprecipitated from HEK293 cells and immunoblotted for GRP78, which demonstrated that the two proteins were associated (Fig. 2B). We tested if ER stress or inhibition of GSK3 with the selective inhibitor lithium (Klein and Melton, 1996) influences the interaction between HSP105 and GRP78 using for immunoprecipitation an antibody to KDEL, the ER retention signal present on ER resident proteins. GRP78 was found to be the only KDEL-containing protein that co-immunoprecipitated with HSP105 (data not shown), and immunoprecipitation with anti-kdel co-immunoprecipitated HSP105 (Fig. 2C). SH-SY5Y cells were treated with tunicamycin to induce ER stress, without or with pretreatment with the GSK3 inhibitor lithium, followed by immunoprecipitation of KDEL and immunoblotting for HSP105 (Fig. 2C). Neither ER stress nor GSK3 inhibition had a major influence on the GRP78-HSP105 association, suggesting they are constitutively bound. Thus, several new partners of HSP105 were identified, and three of them, GSK3, Sec16L, and GRP78, have important roles in ER function and cellular signaling in response to ER stress. HSP105 regulates caspase-3 activation independently of the UPR Based on our findings that HSP105 is associated with proteins involved in the response to ER stress, GSK3 and GRP78, we hypothesized that HSP105 might play a role 53

61 in this signaling cascade. To examine the functional effects of HSP105, shrnai was used to knock down its levels in cells using an RNAi vector that co-expresses GFP to allow identification of transduced cells. SH-SY5Y cells were stably transduced with lentivirus carrying either the empty RNAi vector or a vector with either of two RNAi sequences specific to HSP105. The level of HSP105 was effectively knocked down by lentivirus-mediated RNAi, with a greater reduction achieved with RNAi sequence #2 compared with sequence #4, and specificity was confirmed by the lack of any effect on the levels of two other hsps, HSP90 and HSP70 (Fig. 3A). The efficiency of transduction with vector HSP105 RNAi #2 was near 100%, while the efficiency of HSP105 RNAi #4 was slightly lower, as determined by fluorescence microscopy (Fig. 3B) and flow cytometry (not shown). Since GSK3 is associated with HSP105, we used cells with knocked down levels of HSP105 to test if HSP105 regulated the serine-phosphorylation of GSK3 after Akt activation or the dynamic flux of nuclear levels of GSK3. Vector cells and cells with knockdown of HSP105 were treated with insulin-like growth factor-1 (IGF-1) and the serine-phosphorylation of Akt and GSK3β were examined by immunoblot analyses using specific phospho-dependent antibodies (Fig. 3C). Stimulation with IGF-1 induced rapid increases in the phosphorylation of both Akt and GSK3β, and these responses were equivalent in cells with or without HSP105. We previously identified mechanisms regulating the nuclear localization of GSK3, which include cytoplasmic sequestration by GSK3 binding partners (Meares and Jope, 2007). Therefore, we tested if reduction of HSP105 influenced the influx or efflux of nuclear GSK3. Cells were placed in serum-free media to promote GSK3 nuclear import, and subsequently stimulated with IGF-1 to 54

62 promote GSK3 export, followed by fractionation and immunoblotting of nuclear GSK3α/β (Fig 3D). Reduction of HSP105 had no affect on the nuclear import or export of GSK3. Thus, although HSP105 binds GSK3, it is not required for growth factor signaling to serine-phosphorylate GSK3 or for the nuclear flux of GSK3. To investigate the role of HSP105 in responses to ER stress, we first verified that SH-SY5Y cells respond appropriately to ER stress. Cells were treated with tunicamycin (Fig. 4A) or thapsigargin (Fig. 4B) for 3 to 6 h, followed by immunoblotting for ER stress and apoptotic markers. Treatment with either ER stress-inducing agent resulted in phosphorylation of eif2α and up-regulation of CHOP, confirming that the UPR was activated. ER stress also led to cleavage (activation) of caspase-3 and cleavage of the caspase-3 substrate PARP. Thus, ER stress-inducing agents activate the UPR and the apoptotic pathway in SH-SY5Y cells. To test if HSP105 regulates the UPR or the subsequent apoptotic signaling cascade, ER stress was induced thapsigargin or tunicamycin in control and HSP105 knockdown cells. RT-PCR analysis following thapsigargin treatment revealed that XBP-1 mrna splicing induced by ER stress was not overtly affected by HSP105 knockdown (Fig. 4C). Consistent with this, reduction of HSP105 had no impact on the phosphorylation of eif2α (Fig. 4D) or CHOP expression (Fig. 4E) following ER stress induced by tunicamycin treatment. In contrast to the unaltered ER stress-induced UPR signaling and CHOP expression after HSP105 knockdown, the subsequent apoptotic signaling was highly dependent on HSP105. Treatment with thapsigargin induced caspase-3 activation in wildtype and vector control cells, but caspase-3 activation was greatly attenuated in HSP105 55

63 knockdown cells (Fig. 4F). These results were verified by immunostaining for active caspase-3, which demonstrated that in HSP105 knockdown cells treated with tunicamycin or thapsigargin, active caspase-3 staining was greatly reduced compared to vector cells (Fig. 4G). In contrast to the profound block by HSP105 knockdown on ER stress-induced caspase-3 activation, HSP105 knockdown did not alter caspase-3 activation induced by the DNA damaging agent camptothecin. Quantification of the immunocytochemical measurements demonstrated that active caspase-3 was significantly reduced in HSP105 knockdown cells compared with vector cells treated with tunicamycin or thapsigargin, but not with camptothecin (Fig. 5). These results were further substantiated by immunoblotting for cleavage of the well-characterized caspase-3 substrate PARP. As with cleaved caspase-3, knockdown of HSP105 greatly reduced PARP cleavage induced by tunicamycin or thapsigargin, but not by camptothecin (Fig. 5). Additionally, when HSP105 was reduced by an alternative HSP105-specific shrnai, a similar reduction in tunicamycin-induced PARP cleavage was observed (Fig. 5A), thus confirming that the observed affects are due to HSP105 knockdown. Thus, HSP105 knockdown selectively attenuated ER stress-induced apoptotic signaling, and but did not cause a generalized inhibition of caspase-3 activation. This demonstrates that HSP105 is an important and specific mediator of ER stress-induced caspase-3 activation. To further assess the role of HSP105 in ER stress-induced apoptotic signaling, events up-and down-stream of caspase-3 activation were investigated. Following apoptotic stimuli the anti-apoptotic protein MCL1 is rapidly degraded and the proapoptotic protein Bax becomes active (Gelinas and White, 2005). Surprisingly, knockdown of HSP105 had no effect on either tunicamycin-induced reductions in MCL1 56

64 levels (Fig. 6A) or Bax activation (Fig. 6B). However, consistent with the reduction in active caspase-3, tunicamycin-induced chromatin condensation was blocked in cells with reduced HSP105 (Fig. 6C). The lack of caspase-3 activation and nuclear condensation after ER stress in cells with the level of HSP105 knocked down raised two possibilities: the cells were better able to adapt to ER stress and survive, or the cells were driven to an alternative pathway of cell death. Measurements of cell viability clearly supported the latter conclusion. With tunicamycin treatment, knockdown of HSP105 was not able to prevent loss of cell morphology, suggesting these cells were dying. (Fig. 6D).Cell death assessed by measurements of LDH release (Fig. 6E) confirmed this observation. Thus, HSP105 is required for caspase-3-mediated apoptosis following ER stress-induced activation of the UPR, and in cells with decreased HSP105 an alternative cell death pathway is utilized. GSK3 is critical for ER stress-induced apoptotic signaling Several studies have clearly established that GSK3β facilitates ER stress-induced caspase activation, but the responsible mechanisms remained to be clarified (Chen et al., 2004; Hiroi et al., 2005; Kim et al., 2005; Srinivasan et al., 2005; Brewster el al., 2006; Takadera et al., 2006). To examine the role of GSK3 in this signaling pathway, cells were treated with tunicamycin without or with lithium pretreatment to inhibit GSK3 (Klein and Melton 1996). Consistent with our previous findings, inhibition of GSK3 with lithium reduced tunicamycin-induced caspase-3 activation (Fig. 7A). Lithium treatment did not alter an early event in the UPR, the phosphorylation of eif2α, but lithium greatly reduced the up-regulation of CHOP induced by tunicamycin treatment (Fig. 7A). To verify that 57

65 active GSK3 is required for ER stress-induced CHOP expression, cells were pretreated with three other structurally distinct small molecule inhibitors of GSK3, followed by treatment with tunicamycin. All three GSK3 inhibitors reduced tunicamycin-induced CHOP expression by greater than 75% (Fig. 7B). These results demonstrate for the first time that GSK3 promotes upregulation of CHOP after ER stress, a mechanism that likely contributes to the facilitation by GSK3 of ER stress-induced apoptosis. DISCUSSION ER stress-specific responses are critical for cellular adaptation and survival and, when these fail, for the proper execution of apoptosis (Szegezdi et al., 2006; Yoshida, 2007). In the present study we identified HSP105, as well as its binding partner GSK3, as crucial components of the apoptotic program following ER stress. Although each of these two proteins facilitate the apoptosis cascade after ER stress, they do so by different mechanisms. GSK3 was discovered to promote the expression of CHOP, a key transcription factor important in the final determination of cell fate, as sufficient CHOP activation leads to cell death by apoptosis (Zinszner et al., 1998; Rutkowski et al., 2006). HSP105, on the other hand, acts downstream of CHOP activation to ensure the apoptotic cascade proceeds to activation of the executioner caspase, caspase-3. Hsps constitute a large family of proteins involved in many cell functions, but relatively few protein partners, or specific functions, of HSP105 have been identified despite its being one of the major cellular hsps. HSP105 previously was found to associate with HSP70 and HSP40 and to suppresses the ability of HSP70/HSP40 to promote protein refolding (Yamagishi et al., 2000; Ishihara et al., 2003). Although these 58

66 studies indicated that HSP105 reduces the function HSP70, in opposition to this HSP105 also was reported to be a nucleotide exchange factor for HSP70, thus promoting its activity (Dragovi et al., 2006; Raviol et al., 2006). In addition to interacting with HSP70, bacterial and yeast two-hybrid screenings identified several HSP105-interacting proteins, including α-tubulin, cofilin, dynein light chain 2A, α-adducin, platelet-activating factor, acetylhydrolase α-subunit, and phosphoglycerate kinase 1 (Saito et al., 2003; Saito et al., 2004). Of these, only the interaction with α-tubulin was functionally characterized, as HSP105 was reported to suppress disorganization of microtubules during heat shock via its interaction with α-tubulin (Saito et al., 2003). Additionally, HSP105 interacts with the truncated androgen receptor with an extended polyglutamine tract (Ishihara et al., 2003), and with mutant Cu/Zn superoxide dismutase (Yamashita et al., 2007). In the current study, we identified several new binding partners of HSP105, including Huntingtininteracting protein 1, Mortalin, Sec16L, GSK3, and GRP78. Since each of the latter three proteins is involved in some aspect of ER function, HSP105 may be particularly involved in chaperoning events in the ER, especially responses to ER stress since GRP78 plays a central role in the UPR. However, despite the clear association with GRP78, knockdown of HSP105 had no detectable affect on the UPR signaling. Since HSP105 is an abundant protein, it is possible that low levels of HSP105 remaining after shrnai-mediated knockdown are sufficient for chaperoning GRP78-mediated functions. Alternatively, HSP105 may regulate other GRP78-linked actions in the ER, perhaps acting as a nucleotide exchange factor for GRP78 since HSP105 is a nucleotide exchange factor for HSP70 (Dragovi et al., 2006; Raviol et al., 2006), which is highly homologous with GRP78. 59

67 GSK3 also was identified as a HSP105-binding protein. These proteins associated in mouse brain, and in both the nucleus and cytosol of SH-SY5Y cells. The two isoforms of GSK3, GSK3α and GSK3β, share a high degree of homology in their large central kinase domains, but less so in their N-terminal and C-terminal regions (Woodgett, 1990). Since HSP105 binds both isoforms, it is likely that HSP105 binds within their homologous region. Although many kinases are associated with members of the heat shock family of chaperones (Caplan et al., 2006), HSP105 is the first molecular chaperone found to associate with GSK3 except during translation (Lochhead et al., 2006). The association of GSK3 with HSP105 increased during heat shock, raising the possibility that HSP105 may have a protective action towards GSK3 during cell stresses that have the potential to destabilize GSK3. However, the effect of the HSP105-GSK3 interaction on ER stress signaling remains to be identified. While it is possible that the influence of HSP105 on caspase-3 is through the well characterized GSK3-dependent promotion of apoptotic signaling, it is as likely that HSP105 affects a relatively small pool of GSK3 whose function is still unknown. In addition to binding HSP105, GSK3 was found to promote apoptotic signaling, as inhibitors of GSK3 reduced caspase-3 activation induced by ER stress, in accordance with previous reports of this effect (Chen et al., 2004; Hiroi et al., 2005; Kim et al., 2005; Srinivasan et al., 2005; Brewster el al., 2006; Takadera et al., 2006). GSK3 has been reported to facilitate apoptosis induced by many different stimuli by multiple mechanisms, including promoting mitochondrial disruption, destabilization of microtubules, and inhibition of survival-promoting transcription factors (Beurel and Jope 2006). In addition to those pro-apoptotic actions of GSK3, of particular significance for 60

68 ER stress we found that GSK3 was necessary for ER stress-induced upregulation of the proapoptotic transcription factor CHOP. This may be an especially crucial action during ER stress considering the recent demonstration that the outcome of ER stress, adaptation or apoptosis, is largely determined by the level of CHOP (Rutkowski et al., 2006). Since inhibition of GSK3 reduces CHOP activation, this should raise the threshold for apoptosis after ER stress, thereby contributing to the reduced apoptosis that occurs in the presence of GSK3 inhibitors. This central regulatory role of GSK3 in ER-stressed cells also provides a bridge between growth factor signaling through the phosphatidylinositol 3-kinase (PI3K)/Akt pathway, which inhibits GSK3, and UPR-associated apoptotic signaling. Thus, when growth factor-induced PI3K/Akt signaling is low, GSK3 activity is elevated, allowing CHOP activation by the UPR and the subsequent activation of apoptotic signaling. However, with growth factor signaling maintaining GSK3 inhibited by phosphorylation on its regulatory serine residues, CHOP activation is reduced to allow the UPR-mediated adaptation to ER stress and cell survival. The most surprising function of HSP105 identified in this study is that it is required for caspase-3 activation following ER stress. Hsps are usually considered to be anti-apoptotic proteins that assist in adaptation and recovery from potentially lethal insults. However, HSP105 has another role, as it was essential for caspase-3 activation after ER stress. ER stress causes accumulation of misfolded proteins, which may threaten the ability of proteins in the apoptotic signaling cascade to function properly in the absence of a chaperone. Therefore, it is possible that HSP105 is not directly a proapoptotic protein, but simply performs its normal function, even during apoptosis, as a chaperone to maintain enzymes in active conformations, some of which may be pro- 61

69 apoptotic. This type of regulation has been demonstrated between HSP60 and procaspase-3 (although not in the context of ER stress) in which HSP60 is thought to maintain pro-caspase-3 in an activation or cleavage-competent state (Xanthoudakis et al., 1999; Samali et al., 1999). HSP105 might have a similar, but stress-specific, role acting on proteins necessary for activation of caspase-3, such as through upstream molecules such as GSK3. It is interesting to note that elevated HSP105 previously was found to be associated with activated caspase-3 during development (Evrard et al., 1999; Evrard et al., 2000; Gashegu et al., 2007). Alternatively, the finding that knockdown of HSP105 reduced caspase-3 activation could be interpreted as indicating that HSP105 does not facilitate caspase-3 activation but instead suppresses an alternative death pathway. Thus, in the absence of HSP105 this alternative pathway may be de-repressed and activated, so cell death ensues independently of caspase-3. This would be consistent with the report that in other apoptotic conditions overexpression of HSP105 blocked PARP cleavage, a caspase-3 substrate, indicating that HSP105 can reduce signaling to caspase-3 activation (Hatayama et al., 2001). However, others found overexpression of HSP105 enhanced PARP cleavage (Yamagishi et al., 2002b). These findings could indicate that the role of HSP105 in regulating caspase-3 activation is stimulus-specific, with HSP105 being necessary for caspase-3 activation after ER stress but not with all other apoptotic insults, or overexpressed HSP105 may engage in actions not performed by endogenous HSP105. Our data is consistent with either model, as we observed reduced capase-3 activation after ER stress but no appreciable change in cell death in cells with knocked down HSP

70 In summary, we identified several new HSP105-interacting proteins, including GSK3. HSP105 was found to be necessary for the proper activation of caspase-3 following ER stress. Additionally, we identified a novel role for GSK3 in the regulation of CHOP during ER stress. Thus, both GSK3 and HSP105 are important modulators of the ER stress response by facilitating apoptosis through the appropriate pathways. Acknowledgments We thank Dr. Van Parijs for the Lentilox3.7 vector, Dr. Tong Zhou for antibodies, Dr. B. Clodfelder-Miller for preparation of cerebral cortical lysates, and Dr. James Lah (Emory University) for providing virus production protocols. This research was supported by grants from the National Institutes of Health. 63

71 REFERENCES Bertolotti A, Zhang Y, Hendershot LM, Harding HP, Ron D Dynamic interaction of BiP and ER stress transducers in the unfolded-protein response. Nat Cell Biol. 2: Beere HM Death versus survival: functional interaction between the apoptotic and stress-inducible heat shock protein pathways. J Clin Invest. 115: Beurel E, Jope RS The paradoxical pro- and anti-apoptotic actions of GSK3 in the intrinsic and extrinsic apoptosis signaling pathways. Prog Neurobiol. 79: Bhattacharyya D, Glick BS Two mammalian Sec16 homologues have nonredundant functions in endoplasmic reticulum (ER) export and transitional ER organization. Mol Biol Cell. 18: Bijur GN, Jope RS Proapoptotic stimuli induce nuclear accumulation of glycogen synthase kinase-3β. J Biol Chem. 276: Boyce M, Yuan J Cellular response to endoplasmic reticulum stress: a matter of life or death. Cell Death Differ. 13: Brewster JL, Linseman DA, Bouchard RJ, Loucks FA, Precht TA, Esch EA, Heidenreich KA Endoplasmic reticulum stress and trophic factor withdrawal activate distinct signaling cascades that induce glycogen synthase kinase-3β and a caspase-9- dependent apoptosis in cerebellar granule neurons. Mol Cell Neurosci. 32: Calfon M, Zeng H, Urano F, Till JH, Hubbard SR, Harding HP, Clark SG, Ron D IRE1 couples endoplasmic reticulum load to secretory capacity by processing the XBP-1 mrna. Nature. 415: Caplan AJ, Mandal AK, Theodoraki MA Molecular chaperones and protein kinase quality control. Trends Cell Biol. 17: Chen G, Bower KA, Ma C, Fang S, Thiele CJ, Luo J Glycogen synthase kinase 3β (GSK3β) mediates 6-hydroxydopamine-induced neuronal death. FASEB J. 18: Dragovic Z, Broadley SA, Shomura Y, Bracher A, Hartl FU Molecular chaperones of the Hsp110 family act as nucleotide exchange factors of Hsp70s. EMBO J. 25:

72 Evrard L, Vanmuylder N, Dourov N, Glineur R, Louryan S Cytochemical identification of HSP110 during early mouse facial development. J. Craniofac. Genet. Dev. Biol. 19: Evrard L, Vanmuylder N, Dourov N, Hermans C, Biermans J, Werry-Huet A, Rooze M, Louryan S Correlation of HSP110 expression with all-trans retinoic acidinduced apoptosis. J. Craniofac. Genet. Dev. Biol. 20: Gashegu J, Ladha R, Vanmuylder N, Philippson C, Bremer F, Rooze M, Louryan S HSP110, caspase-3 and -9 expression in physiological apoptosis and apoptosis induced by in vivo embryonic exposition to all-trans retinoic acid or irradiation during early mouse eye development. J. Anat. 210: Gelinas C, White E BH3-only proteins in control: specificity regulates MCL-1 and BAK-mediated apoptosis. Genes Dev. 19: Hatayama T, Yasuda K, Yasuda K Association of HSP105 with HSC70 in high molecular mass complexes in mouse FM3A cells. Biochem Biophys Res Commun. 248: Hiroi T, Wei H, Hough C, Leeds P, Chuang DM Protracted lithium treatment protects against the ER stress elicited by thapsigargin in rat PC12 cells: roles of intracellular calcium, GRP78 and Bcl-2. Pharmacogenomics J. 5: Ishihara K, Yamagishi N, Saito Y, Adachi H, Kobayashi Y, Sobue G, Ohtsuka K, Hatayama T Hsp105alpha suppresses the aggregation of truncated androgen receptor with expanded CAG repeats and cell toxicity. J Biol Chem. 278: Jope RS, Johnson GVW The glamour and gloom of glycogen synthase kinase-3. Trends Biochem Sci. 29: Kim AJ, Shi Y, Austin RC, Werstuck GH Valproate protects cells from ER stressinduced lipid accumulation and apoptosis by inhibiting glycogen synthase kinase-3. J Cell Sci. 118: Klein, P. S., Melton, D. A A molecular mechanism for the effect of lithium on development. Proc. Natl. Acad. Sci. USA 93: Lindholm D, Wootz H, Korhonen L ER stress and neurodegenerative diseases. Cell Death Differ. 13: Lochhead PA, Kinstrie R, Sibbet G, Rawjee T, Morrice N, Cleghon V A chaperone-dependent GSK3β transitional intermediate mediates activation-loop autophosphorylation. Mol. Cell. 24:

73 Marciniak SJ, Yun CY, Oyadomari S, Novoa I, Zhang Y, Jungreis R, Nagata K, Harding HP, Ron D CHOP induces death by promoting protein synthesis and oxidation in the stressed endoplasmic reticulum. Genes Dev. 18: Marcu MG, Doyle M, Bertolotti A, Ron D, Hendershot L, Neckers L Heat shock protein 90 modulates the unfolded protein response by stabilizing IRE1α. Mol Cell Biol. 22: McCullough KD, Martindale JL, Klotz LO, Aw TY, Holbrook NJ Gadd153 sensitizes cells to endoplasmic reticulum stress by down-regulating Bcl2 and perturbing the cellular redox state. Mol Cell Biol. 21: Meares GP, Jope RS Resolution of the Nuclear Localization Mechanism of Glycogen Synthase Kinase-3: FUNCTIONAL EFFECTS IN APOPTOSIS. J Biol Chem. 282: Oyadomari S, Mori M Roles of CHOP/GADD153 in endoplasmic reticulum stress. Cell Death Differ. 11: Oyadomari S, Yun C, Fisher EA, Kreglinger N, Kreibich G, Oyadomari M, Harding HP, Goodman AG, Harant H, Garrison JL, Taunton J, Katze MG, Ron D Cotranslocational degradation protects the stressed endoplasmic reticulum from protein overload. Cell. 126: Raviol H, Sadlish H, Rodriguez F, Mayer MP, Bukau B Chaperone network in the yeast cytosol: Hsp110 is revealed as an Hsp70 nucleotide exchange factor. EMBO J. 25: Rutkowski DT, Kaufman RJ A trip to the ER: coping with stress. Trends Cell Biol. 14: Rutkowski DT, Arnold SM, Miller CN, Wu J, Li J, Gunnison KM, Mori K, Sadighi Akha AA, Raden D, Kaufman RJ Adaptation to ER stress is mediated by differential stabilities of pro-survival and pro-apoptotic mrnas and proteins. PLoS Biol. 4:e374. Saito Y, Yamagishi N, Ishihara K, Hatayama T Identification of α-tubulin as an hsp105α-binding protein by the yeast two-hybrid system. Exp Cell Res. 286: Saito Y, Doi K, Yamagishi N, Ishihara K, Hatayama T Screening of Hsp105αbinding proteins using yeast and bacterial two-hybrid systems. Biochem Biophys Res Commun. 314: Samali A, Cai J, Zhivotovsky B, Jones DP, Orrenius S Presence of a pre-apoptotic complex of pro-caspase-3, Hsp60 and Hsp10 in the mitochondrial fraction of jurkat cells. EMBO J. 18:

74 Schroder M, Kaufman RJ The mammalian unfolded protein response. Annu Rev Biochem.74: Shen J, Chen X, Hendershot L, Prywes R ER stress regulation of ATF6 localization by dissociation of BiP/GRP78 binding and unmasking of Golgi localization signals. Dev Cell. 3: Song L, De Sarno P, Jope RS Central role of glycogen synthase kinase-3β in endoplasmic reticulum stress-induced caspase-3 activation. J Biol Chem. 277: Srinivasan S, Ohsugi M, Liu Z, Fatrai S, Bernal-Mizrachi E, Permutt MA Endoplasmic reticulum stress-induced apoptosis is partly mediated by reduced insulin signaling through phosphatidylinositol 3-kinase/Akt and increased glycogen synthase kinase-3β in mouse insulinoma cells. Diabetes. 54: Szegezdi E, Logue SE, Gorman AM, Samali A Mediators of endoplasmic reticulum stress-induced apoptosis. EMBO Rep. 7: Takadera T, Yoshikawa R, Ohyashiki T Thapsigargin-induced apoptosis was prevented by glycogen synthase kinase-3 inhibitors in PC12 cells. Neurosci Lett. 408: Woodgett, J. R Molecular cloning and expression of glycogen synthase kinase- 3/factor A. EMBO J. 9: Xanthoudakis S, Roy S, Rasper D, Hennessey T, Aubin Y, Cassady R, Tawa P, Ruel R, Rosen A, Nicholson DW Hsp60 accelerates the maturation of pro-caspase-3 by upstream activator proteases during apoptosis. EMBO J. 18: Xu C, Bailly-Maitre B, Reed JC Endoplasmic reticulum stress: cell life and death decisions. J Clin Invest. 115: Yamagishi N, Nishihori H, Ishihara K, Ohtsuka K, Hatayama T Modulation of the chaperone activities of Hsc70/Hsp40 by Hsp105α and Hsp105β. Biochem Biophys Res Commun. 272: Yamagishi N, Saito Y, Ishihara K, Hatayama T. 2002a. Enhancement of oxidative stressinduced apoptosis by Hsp105α in mouse embryonal F9 cells. Eur J Biochem. 269: Yamagishi N, Ishihara K, Saito Y, Hatayama T. 2002b. Hsp105α enhances stress-induced apoptosis but not necrosis in mouse embryonal F9 cells. J. Biochem. 132: , 67

75 Yamagishi N, Ishihara K, Saito Y, Hatayama T Hsp105 family proteins suppress staurosporine-induced apoptosis by inhibiting the translocation of Bax to mitochondria in HeLa cells. Exp Cell Res. 312: Yamashita H, Kawamata J, Okawa K, Kanki R, Nakamizo T, Hatayama T, Yamanaka K, Takahashi R, Shimohama S Heat-shock protein 105 interacts with and suppresses aggregation of mutant Cu/Zn superoxide dismutase: clues to a possible strategy for treating ALS. J Neurochem. (in press) Yoshida H ER stress and diseases. FEBS J. 274: Yoshida H, Matsui T, Yamamoto A, Okada T, Mori K XBP1 mrna is induced by ATF6 and spliced by IRE1 in response to ER stress to produce a highly active transcription factor. Cell. 107: Young JC, Agashe VR, Siegers K, Hartl FU Pathways of chaperone-mediated protein folding in the cytosol. Nat Rev Mol Cell Biol. 5: Zinszner H, Kuroda M, Wang X, Batchvarova N, Lightfoot RT, Remotti H, Stevens JL, Ron D CHOP is implicated in programmed cell death in response to impaired function of the endoplasmic reticulum. Genes Dev. 12:

76 A HSP105 GSK3β IP GSK3β B HSP105 HSP105-V5 - + Neg 1 2 Ctl cerebral cortex GSK3β HSP105 IP: V5 C HSP105-V5 - + HSP105 IP GSK3β D GSK3β HS 45 (min) Neg Ctl cytosol nucleus HSP105 HS 45 (min)

77 E IP: GSK3β cytosol nucleus HSP105 GSK3β HS 45 (min) Neg Ctl F IP: GSK3α cytosol nucleus HSP105 GSK3α HS 45 (min) Neg Ctl Figure 1. GSK3 associates with HSP105. A, GSK3β was immunoprecipitated (IP) from cerebral cortical homogenates of two C57BL/6 mice followed by immunoblotting for HSP105 and GSK3β. B. SH-SY5Y cells stably expressing HSP105-V5 were generated by lentiviral transduction, and the expression level was examined by immunoblotting for HSP105. HSP105-V5 was immunoprecipitated using an anti-v5 antibody from non-transduced (-) or HSP105-V5 transduced (+) cells, followed by immunoblotting for GSK3β and HSP105. C. SH-SY5Y cells were subjected to heat shock (HS) by incubation in a 45ºC water bath for min, followed by immunoprecipitation of GSK3β and immunoblotting for HSP105 and GSK3β. D. SH- SY5Y cells were placed in 45ºC water bath for 0-30 min, and then cytosolic and nuclear fractions were prepared and immunoblotted for HSP105. E. SH-SY5Y cells were placed in a 45ºC water bath for 0-30 min, cytosolic and nuclear fractions were prepared, followed by immunoprecipitation of GSK3β and immunoblotting for HSP105 and GSK3β. F. SH-SY5Y cells were placed in a 45ºC water bath for 0-30 min, cytosolic and nuclear fractions were prepared, followed by immunoprecipitation of GSK3α and immunoblotting for HSP105 and GSK3α. Neg. Ctl. (negative control) is non-immuno isotype matched IgG. 70

78 A HSP105 interacting proteins identified by MALDI-TOF Chain A, Heat shock protein 70kDa protein 42kDa ATPase N-terminal domain Sec16L (KIAA0310) Huntingtin interacting protein 1 (HIP1) Heat shock 70 kda protein 9B (Mortalin-2/GRP75) 78 kda glucose regulated protein (GRP78/BIP) B IP: V5 C IP: KDEL Bip/GRP78 Heavy Chain HSP105-V5 - + HSP105 GRP94 GRP78 Tunc(h) Lithium Figure 2. HSP105 interacting proteins identified by mass spectrometry. A. HEK293 cells were transiently transfected with HSP105-V5. HSP105-V5 was then immunoprecipitated from 3 mg of cell lysate, followed by SDS-PAGE and coomassie staining. Several bands were extracted from the gel and the proteins were identified by MALDI-TOF. B. HSP105-V5 was immunoprecipitated with a V5 antibody from HEK293 cells transiently transfected with HSP105-V5 (+) or nontransfected cells (-), followed by immunoblotting for GRP78. C. SH-SY5Y cells were pretreated with or without 20 mm lithium for 30 min followed by treatment with 4 µg/ml of tunicamycin (Tunc) for 1 or 5 h. GRP78 was immunoprecipitated using an anti-kdel antibody followed by immunoblotting for HSP105 and KDEL, which recognizes GRP94 and GRP78. 71

79 A HSP105 HSP90 HSP70 HSP105 RNAi vector #2 #4 B Vector HSP105 RNAi #2 HSP105 RNAi #4 Phase GFP 72

80 C P-S473-Akt Vector HSP105 RNAi #2 P-S9-GSK3β IGF (min) D Nuclear GSK3α GSK3β S SF IGF S SF IGF Vector HSP105 RNAi #2 Figure 3. HSP105 knockdown and growth factor signaling to GSK3β. A. Stable SH-SY5Y cells were generated by lentiviral transduction using the empty shrnai vector, that still expresses GFP, or with one of two shrnai directed to HSP105. The RNAi were designated HSP105 RNAi #2 and HSP105 RNAi #4. The efficiency of knockdown of HSP105 and the effects on HSP90 and HSP70 levels were measured in immunoblots. B. The efficiency of transduction was monitored by examining GFP expression by fluorescent microscopy. C. Vector or HSP105 RNAi #2 expressing cells were treated with 50 ng/ml IGF-1 for 0-30 min followed by immunoblotting for phospho-ser473-akt and phospho-ser9- GSK3β. D. Vector or HSP105 RNAi #2 expressing cells were maintained in serum (S) or placed in serum-free (SF) media for 2 h, then incubated with or without 50 ng/ml IGF-1 for 30 min. The nuclei were isolated and the level of GSK3α/β measured by immunoblot. 73

81 A P-eIF2α B P-eIF2α CHOP CHOP Cleaved caspase-3 Cleaved PARP β-actin Tunicamycin (hr) Cleaved caspase-3 Cleaved PARP β-actin Thapsigargin (hr) C Spliced XBP-1 Thapsigargin (hr) Vector HSP105 RNAi #2 HSP105 RNAi #4 D p-eif2α eif2α Tunicamycin (min) Vector E HSP105 RNAi #2 CHOP Tunicamycin (hr) F Cleaved caspase-3 HSP Vector HSP105 RNAi #2 Thapsigargin (hr) SH-SY5Y Vector HSP105 RNAi #2 74

82 G Control Tunicamycin Thapsigargin Camptothecin Vector DAPI Cleaved Caspase-3 HSP105 RNAi #2 Cleaved Caspase-3 DAPI Figure 4. HSP105 and UPR-induced signaling. A. SH-SY5Y cells were treated with 4 µg/ml tunicamycin for 0 to 6 h followed by immunoblotting for phosphoeif2α, CHOP, cleaved caspase-3, cleaved PARP, and β-actin. B. SH-SY5Y cells were treated with 2 µm thapsigargin for 0 to 6 h followed by immunoblotting for phosphoeif2α, CHOP, cleaved caspase-3, cleaved PARP, and β-actin. C. To examine XBP-1 splicing, vector or HSP105 RNAi expressing cells were treated with 2 µm thapsigargin for 2 or 4 h. The cellular RNA was isolated and spliced XBP-1 examined by RT-PCR. D. Vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 0 to 90 min, then immunoblotted for phospho-eif2α or total eif2α. E. Vector or HSP105 RNAi #2 cells were treated with 4 µg/ml tunicamycin for 0 to 4 h, then immunoblotted for CHOP. F. Untransduced SH-SY5Y cells, vector cells, or HSP105 RNAi #2 expressing cells were treated with 2 µm thapsigargin for 2 or 3 h, then immunoblotted for cleaved caspase-3 and HSP105. G. Vector or HSP105 RNAi#2 expressing cells were treated with 4 µg/ml tunicamycin, 2 µm thapsigargin, or 1 µm camptothecin, for 3 h, then stained for active caspase-3 or DAPI and examined by fluorescence microscopy. 75

83 A Vector HSP105KD B Vector HSP105KD Active Caspase-3 (percent control) * * Active Caspase-3 (percent control) * * * Tunicamycin (hr) Thapsigargin (hr) Cleaved PARP HSP105 β-actin Vector HSP105 RNAi #2 Cleaved PARP HSP105 β-actin Vector HSP105 RNAi #2 Tunic (hr) Thaps (hr) Vector HSP105 RNAi #4 Cleaved PARP HSP105 Tunic (hr)

84 C 3000 Vector HSP105KD 2500 Active Caspase-3 (percent control) Cleaved PARP HSP105 β-actin Campt (hr) Vector Camptothecin (hr) HSP105 RNAi # Figure 5. HSP105 knockdown attenuates ER stress-induced caspase-3 activation. A. Vector cells or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 0 to 4 h and active caspase-3 examined by immunocytochemistry. Quantitative values are means ± SEM, n=3, p Cleaved PARP, HSP105, and β- actin were examined by immunoblot. Vector cells and HSP105 RNAi #4 expressing cells were treated with 4 µg/ml tunicamycin (Tunic) for 0 to 4 h and immunoblotted for cleaved PARP and HSP105. B. Vector cells or HSP105 RNAi #2 expressing cells were treated with 2 µm thapsigargin for 0 to 4 h and active caspase-3 examined by immunocytochemistry. Quantitative values are means ± SEM, n=3, P Cleaved PARP, HSP105, and β-actin were examined by immunoblot. C. Vector cells or HSP105 RNAi #2 expressing cells were treated with 1 µm camptothecin for 0 to 4 h and active caspase-3 examined by immunocytochemistry. Quantitative values are means ± SEM, n=3, p<0.05. Cleaved PARP, HSP105, and β-actin were examined by immunoblot. 77

85 A MCL1 β-actin Tunic (hr) Vector HSP105 RNAi #2 B IP: 6A7 Active Bax Total Bax Vector HSP105 RNAi #2 IP: 6A7 Active Bax Total Bax Tunic (hr) Tunic (hr) C Control Tunicamycin 4h Tunicamycin 8h Vector HSP105 RNAi 78

86 D Control Tunicamycin 4h Tunicamycin 8h Vector HSP105 RNAi E 35 Vector HSP105 RNAi #2 30 LDH release (% total) Tunicamycin (hr) Figure 6. Knockdown of HSP105 does not promote cell survival. A. Vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 3 or 4 h, and immunoblotted for MCL1 and β-actin. B. To examine Bax activation, vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 0 to 4 h. The active conformation specific Bax antibody 6A7 was used to immunoprecipitate active Bax followed by immunoblotting for total Bax in the immunoprecipitate and the cell lysate. C. To examine nuclear morphology, vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 4 or 8 h, then fixed and stained with Hoechst and examined by fluorescence microscopy. D. To examine cell morphology, vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 4 or 8 h, then fixed and examined by phase contrast microscopy. E. Vector or HSP105 RNAi #2 expressing cells were treated with 4 µg/ml tunicamycin for 0 to 16 h, and cell death was assessed by measuring the release of LDH. Quantitative values are means ± SEM, n=3. 79