Abstract

Introduction

Heat shock proteins (HSP) are a large group of chaperones which are proteins that assist in protein folding, stabilize proteins and help to refold denatured proteins, processes that are dependent on hydrolysis of ATP (1). If proper folding is not possible, they also aid in protein degradation. The major groups of HSPs are shown in table 1. Chaperonins form a sub-class of HSPs and are characterized by a stacked double-ring structure forming barrels (Xu et al. 1997). Inside the barrel structures, they contain hydrophobic residues for client binding (Lindquist 1986). The prototypes of chaperonins are GroEL/GroES (large and small proteins of the GroE operon in E. coli, mutations of which affect the growth of lambda phage by interfering with assembly of its head protein E) in bacteria (Georgopoulos et al. 1973; Sternberg 1973b, a; Hendrix 1979; Yamamori & Yura 1980; Fayet et al. 1989) and Hsp60/Hsp10 proteins in eukaryotic cells (Johnson & Craig 1997).

Originally, HSPs were described as proteins that were up-regulated after elevated temperatures (Lindquist 1986). Meanwhile it is recognized that HSPs are involved in the response to all kinds of stress reactions that disturb proper protein conformation such as reactions to chemicals like ethanol, arsenite, cadmium, zinc, copper, mercury, sulfhydryl reagents, calcium ionophores, steroid hormones, chelating agents, viruses and many more (Lindquist 1986). Of course, HSPs are strongly induced by DNA damage since this type of stress leads to mutations that often interfere with proper protein folding (Fornace et al. 1988). Since HSPs stabilize DNA binding proteins it is not surprising to detect genomic instability in HSP70 deficient mice (Hunt et al. 2004). Because also many undamaged proteins need assistance in folding, nearly all physiological processes require HSPs. Indeed, with the help of protein-protein interaction (PPI) studies, proteins could be identified as interaction partners that contribute to the following processes: transcription, mRNA splicing, translation, cell cycle control, DNA repair, apoptosis, intracellular transport, development, immune response, lipid and carbohydrate metabolism, cellular signaling, protein modification and many more (Gong et al. 2009; Tsaytler et al. 2009; Gano & Simon 2010; Echeverria et al. 2011; Hartson & Matts 2012; Taipale et al. 2014). HSPs are also involved in protein transport across membranes (e.g. of mitochondria or endoplasmic reticulum).

Several types of chaperones may act together, dependent on the type of protein or type of damage. One model suggests that client proteins (e.g. proteins in translation dependent on the stage of maturation) are first bound to Hsp70, then to chaperonins (Johnson & Craig 1997), then to more specialized proteins of the Hsp90 group (Hartl 1996; Johnson & Craig 1997). An alternative pathway, independent of HSP70 or HSP90, involves binding to CCT/TriC (chaperonin containing T-complex polypeptide/ TCP-1 ring complex) proteins. This pathway is used by filamentous proteins like actin and tubulin (Johnson & Craig 1997) but also other types of proteins. There are also multichaperone complexes like HSP70/HSP90. Complex formation in this case is mediated by the adapter protein and co-chaperone Hop (HSP organizing protein) which binds to the peptide sequence EEVD at the C-terminus of both proteins (Chen & Smith 1998; Scheufler et al. 2000; Brinker et al. 2002). Often, the client proteins remain bound to a HSP, but they may be released once they are stable on their own. Co-chaperones mediate substrate specificity, regulate activity of client proteins or recruit chaperones to specific locations in order to perform special functions like their roles in clathrin uncoating, synaptic vesicle fusion, or regulating cytoskeleton functions (Young et al. 2003b).

There are a lot of excellent reviews on the family of HSP proteins in general (Hartl 1996; Wegele et al. 2004; Calderwood et al. 2006), but also on special subfamilies like HSP70 (Bukau & Horwich 1998) or HSP90 (Csermely et al. 1998; Pearl & Prodromou 2000; Taipale et al. 2010; Erlejman et al. 2014a), their role in signaling protein movement (Pratt et al. 2004) and on co-chaperones (Pratt et al. 2004; Davies & Sanchez 2005; Cioffi et al. 2011; Sivils et al. 2011; Storer et al. 2011).

This review is focused on HSP90AB1, primarily in humans. Because of overlapping functions and lack of discrimination in the past, the term HSP90 is used when it is not clear which member of the HSP90 protein family was studied.

Classification

Based on several suggestions (Chen et al. 2005; Chen et al. 2006a), lastly by Kampinga et al. (Kampinga et al. 2009), the human genome organization (HUGO) gene nomenclature committee (HGNC, http://www.genenames.org/genefamilies/HSP) recognizes 5 human families of HSPs (Table 1). At the moment, all these families encompass 97 genes.

With 5 genes listed, the HSP90/HSPC family is the smallest among all human HSPs. One of them is a pseudogene (HSP90AA3P), therefore four real HSP’s of this family remain (Table 2) which have characteristic localizations in the cell. However, these localizations are only a general rule. Under certain circumstances, these proteins may be also found in other compartments of the cell and even extracellularly (Suzuki & Kulkarni 2010).

Cloning of HSP90AB1

Human HSP90AB1 (Rebbe et al. 1987) was cloned based on homology to HSP90AA1 which was the first HSP90 to be purified, as reported in 1982 (Welch & Feramisco 1982). Both proteins share 60 % overall homology and several regions of 50 amino acids (aa) or more share greater than 90 % homology (Rebbe et al. 1987). Mouse HSP90AB1 was cloned using the corresponding Drosophila cDNA as a hybridization probe (Moore et al. 1987; Hoffmann & Hovemann 1988). It codes for a protein containing 724 aa. The majority of the HSP90A proteins occur as homodimers (Minami et al. 1991). Human HSP90AB1 has 724 aa versus 732 aa in HSP90AA1. With the N-terminal methionins removed, which is usually the case, these numbers are 723 and 731 respectively (Lees-Miller & Anderson 1989b). In HSP90AB1, the amino acids ESEDK is removed between the phosphorylation sites (Lees-Miller & Anderson 1989b). In addition, the amino acids TQTQDQPME at the N-terminal end of HSP90AA1 are replaced by VHHG in HSP90AB1 (Lees-Miller & Anderson 1989b).

HSP90AB1 has the unique signature sequence LKID (residues 71–74) that is not present in other HSPs (Chen et al. 2006a).

Protein domains of HSP90

HSP90 is composed of five domains (Obermann et al. 1998; Chen et al. 2006a):

• -

  N-terminal domain (NTD)

• -

  Linker region/Charged domain 1 (CD1)

• -

  middle domain (MD)

• -

  charged domain 2 (CD2)

• -

  C-terminal domain (CTD)

These main domains as well as their amino acid limits are shown in figure 1 for HSP90AB1 and the 732aa translation product of HSP90AA1. Chen et al. proposed further subdomains and shifts of domain borders based on sequence conservation or variability after analysis of a large number of HSP90 proteins of different species (Chen et al. 2006a).

Open in a separate window

Figure 1

Domain structure of human HSP90AA1 (732aa translation product, Ensembl transcript ID ENST00000216281) and HSP90AB1 (ENSEMBL transcript ID ENST00000371554). Amino acid borders of the main domains are given above each molecule. Co-chaperones that bind to the respective regions are depicted below. NTD, N-terminal domain; CTD, C-terminal domain; MD, middle domain.

The NTD contains the ATP-binding site which is essential for HSP function (Prodromou et al. 1997a; Stebbins et al. 1997; Obermann et al. 1998) and binds several co-chaperones (see below). The ATP binding site shares structural homology with other ATPases in the GHKL (gyrase, HSP90, histidine kinase and MutL) superfamilly (Pearl & Prodromou 2006).

The CD1 influences binding of other proteins to the NTD (Scheibel et al. 1999). Deletion of the CD1 domain is lethal in yeast (Hainzl et al. 2009). In addition, the sensitivity for Aha1-mediated ATPase acceleration declines, binding of p23/PTGES3 is lost and the ability to facilitate client activation is reduced (Hainzl et al. 2009). This flexible domain facilitates intermolecular rearrangements on the milliseconds timescale between a state in which the NTD is docked to the middle domain and a state in which the N-domain is more flexible (Jahn et al. 2014). The MD is the main region for client binding, but also the co-chaperone Aha1/AHSA1 (see below) binds to this domain (Table 3).

The CD2 is only characterized by DNA analysis but not functionally defined (Chen et al. 2006a).

The CTD is the dimerization domain of HSP90. In addition, five C-terminal residues (Met-Glu-Glu-Val-Asp, MEEVD motif) make up a highly conserved tetratricopeptide repeat (TPR) domain binding site which mediates interaction with many co-chaperones (Bose et al. 1996; Owens-Grillo et al. 1996; Young et al. 1998).

Co-chaperones

As already mentioned, many HSP functions are influenced by co-chaperones. They have four major functions (Taipale et al. 2010):

• -

  They coordinate the interplay between HSP90 and other chaperone systems, such as HSP70

• -

  stimulate or inhibit ATPase activity of HSP90

• -

  recruit specific classes of clients and other co-chaperones and

• -

  have an enzymatic activity.

Some important co-chaperones are listed in table 3 and shortly characterized in the text below. The co-chaperone binding regions on HSP90 are depicted in figure 1.

The co-chaperone Cdc37 (cell division cycle 37) acts as an active scaffold molecule that allows binding of HSP90 with its C-terminal region and client proteins like kinases with its N-terminal domain. Cdc37 arrests the ATPase cycle of HSP90 by inserting an arginine side chain into the ATP binding pocket, thereby facilitating protein loading (Roe et al. 2004).The activation of several client protein kinases is regulated by chaperone-targeted dephosphorylation of Cdc37 (Vaughan et al. 2008).

Cdc37 interacts with Cdk4 (cyclin dependent kinase 4) and Cdk6, but not with Cdc2 (cell division cycle 2), Cdk2, Cdk3 or Cdk5 (Lamphere et al. 1997; Vaughan et al. 2006). The Cdc37/HSP90 complex stabilizes Cdk4 (Stepanova et al. 1996). In addition, Cdc37 facilitates complex formation between Cdk4 and Cyclin D1 in vitro (Lamphere et al. 1997). Since p16 competes with Cdc37 for binding to Cdk4 it is likely that part of its inhibitory effect is the destabilization of Cdk4 by removal of Cdc37 (Lamphere et al. 1997). However, it has been found that Cdk9 (formerly called PITALRE due to its characteristic amino acid motif) interacts both with Hsp70 and with a HSP90/Cdc37 complex during Tat activation of HIV-1 transcription (O’Keeffe et al. 2000).

The p23/PTGES3 (prostaglandin E synthase 3) protein known as co-chaperone for HSP90 proteins is also a cytoplasmic terminal PGE2 (prostaglandin E2) synthase (cPGES). P23 binds to the N-terminal domain of HSP90 (Karagoz et al. 2011). P23 deficient mice display a phenotype characteristic of mice with GR (glucocorticoid receptor) deficiency, suggesting that its main function is that of the co-chaperone and not its role in prostaglandin synthesis (Lovgren et al. 2007). Expression of the co-chaperone p23 enhances estrogen receptor (ER)-mediated transcription of genes dependent on the presence of estrogen response elements (ERE) (Oxelmark et al. 2006). P23 also has an interesting role in RNA interference (RNAi). RNAi is a mechanism to control gene expression by regulating mRNA levels. The specificity of this mechanism is determined by small double-stranded RNA molecules, the siRNA (small interfering RNA molecules). The guide strand of this siRNA, which is complementary to a section of the target mRNA, is incorporated into the RNA-induced silencing complex (RISC). The catalytic component of the RISC is the endoculease Argonaute (Ago) (Agrawal et al. 2003). It has been shown in vitro with plant proteins that HSP90 binds to Ago1 (Iki et al. 2010). RISC assembly is driven by the ATPase activity of HSP90 in a Hsc70 (heat shock cognate 70)/HSP90 complex (Iwasaki et al. 2010). The HSP90 co-chaperones p23 and FKBP52 interact with human Ago2 (hAgo2) and HSP90 before small RNA and RISC loading takes place (Pare et al. 2013). In addition, the co-chaperones Aha1 and Cdc37 are necessary for the mechanism of RNAi (Pare et al. 2013). HSP90 modifies Ago2 conformation in order to bind siRNA (Miyoshi et al. 2010).

SGT1 (Small glutamine-rich TPR-containing protein 1) interacts with both HSP90AA1 and HSP90AB1. Nuclear SGT1 is increased upon HSP90 inhibition with geldanamycin and is associated with apoptosis in HeLa cells (Yin et al. 2006).

AHA1 (activator of heat shock 90kDa protein ATPase homolog 1) interacts with both HSP90AB1 and HSP90AA1 (Echeverria et al. 2011). However, in a model of 293T cells, it hast been shown that it preferentially interacts with (the middle domain of) HSP90AA1 (Synoradzki & Bieganowski 2015). It is involved in a broad range of cellular processes such as regulation of signal transduction, RNA metabolism, cell development, regulation of ubiquitin-protein ligase actvity, glycolysis, induction of apoptosis and nucleo-cytoplasmic transport (Echeverria et al. 2011). AHA1 interacts with Importin-4 (IPO4) and karyopherin alpha 5 (Kpna5) whereas an interaction with exportin1 (Xpo1) could not be demonstrated by immunoprecipitation (Echeverria et al. 2011). AHA1 binds to HSP90 with its N-terminal domain, thereby stimulating ATP binding of HSP90 (Meyer et al. 2004). Silencing of p53 leads to elevated AHA1 levels. In addition, increased AHA1 levels, both by p53 silencing and by AHA1 overexpression, lead to increased HSP90 ATPase activity and enhanced cytochrome P450 levels both under basal conditions and after treatment with the carcinogen benzo[a]pyrene (Kochhar et al. 2014). Furthermore, after AHA1 and HSP90 activation, also Akt and GSK3beta are activated and Wnt target genes are expressed. Together with p53 rescue intervention, these experiments demonstrate a connection between the p53 and Wnt pathways (Okayama et al. 2014).

HOP (Hsp-organizing protein) mediates interaction between HSP70 and HSP90 proteins (Chen & Smith 1998). The TPR1 and TPR2A domains of Hop bind to the C-terminal (M)EEVD motifs of Hsp70 and HSP90 respectively (Scheufler et al. 2000). As explained below, this complex plays a role in stabilizing mutant p53 (King et al. 2001). Hop is also the first protein recruited to HSP90 in the chaperone cycle (see below).

CHIP/STUB1 (C-terminus of HSc70-interacting protein /STIP1 homology and U-box containing protein 1) is an E3 ubiquitin ligase that interacts and destabilizes fibroblast growth factor receptor 3 (FGFR3) (Laederich et al. 2011). FGFR3 stability is strongly dependent on HSP90 and Cdc37 (Laederich et al. 2011). Another client degraded by CHIP is PTK6 (protein tyrosine kinase 6), if not protected by HSP90 (Kang et al. 2012). PTK6 plays a role in tumor biology as described below. Mutations in KCNQ4 (potassium channel, voltage gated KQT-like subfamily Q, member 4) lead to a subtype of autosomal-dominant deafness. Mutant KCNQ4 is degraded by CHIP and protected by the HSP40-HSP70-HOP-HSP90(AB1 and AA1) pathway (Gao et al. 2013). Inducible HSP90AA1 and the constitutive HSP90AB1, had opposite effects on the cellular level of the KCNQ4 channel: HSP90AB1 over-expression partially restored KCNQ4 surface expression in cells mimicking heterozygous conditions in deafness patients (Gao et al. 2013). In TGF-beta signaling HSP70 and HSP90 differentially regulate complex formation of SMAD3 and CHIP. Increased levels of HSP70 and a decrease of HSP90 increased CHIP-mediated degradation of SMAD3 whereas the opposite changes in HSP levels stabilized SMAD3. HSP90AB1 is somewhat more effective in stabilizing SMAD3 than HSP90AA1. Consequently, HSP90AB1 overexpression leads to a higher activity of the TGFbeta pathway as measured with a luciferase reporter assay (Shang et al. 2014). In contrast, extracellular HSP90AB1 inhibits TGFbeta1 by binding to LAP (latency-associated peptide), thereby preventing the release of active TGFbeta1 from its complex with LAP and decreasing cell proliferation in a cell culture model with MG63 osteosarcoma cells (Suzuki & Kulkarni 2010).

FKBP (FK506 binding proteins) are members of the immunophilin family of proteins which are proteins that bind to immunosuppressants like FK506 or rapamycin. They encode proteins with peptidyl-prolyl cis-trans isomerase (PPIase) activity. FKBPL (FKBP-like) interacts with the HSP90-steroid hormone receptor complex (McKeen et al. 2008). Mutations of FKBPL are associated with azoospermia most likely due to a decrease in androgen receptor (AR) DNA binding caused by the destabilization of this protein (Sunnotel et al. 2010).

FKBPL influences the cytoskeleton through a PPIase domain that interacts with the dynamin motor protein dynamitin (McKeen et al. 2008). In addition, FKBPL binds to CD44 and inhibits cell migration, as shown in endothelial cells and breast cancer cells (Yakkundi et al. 2013). A high level of FKBPL and the FKBPL-interacting protein RBCK1 (RanBP-type and C3HC4-type zinc finger-containing protein 1) correlates with increased patient survival in breast cancer (Donley et al. 2014), suggesting decreased migration and consequential inhibition of metastasis as an underlying mechanism. The role of FKBP52 in RNAi has been mentioned above (Pare et al. 2013). Like FKBPL, FKBP52 interacts with steroid hormone receptors. It does not only stabilize them, but is also necessary for their nuclear translocation (Storer et al. 2011). The interaction between FKBP52 (FKBP4) and HSP90 is regulated by phosphorylation of FKBP52 by casein kinase II (CKII) (Miyata et al. 1997). GR and mineralocorticoid receptor (MR) hormone binding activity and their activity upon gene transcription is higher in association with FKBP52 compared to PP5, whereas FKBP51 has an inhibitory effect (Davies et al. 2005; Gallo et al. 2007). As already mentioned for FKBPL, loss of FKBP52 in mice also leads to azoospermia (Storer et al. 2011). FKBP52 influences microtubule dynamics by isomerization of the microtubule binding protein Tau to its cis configuration, which enhances its dephosphorylation by PP5 (protein phosphatase 5) (Cioffi et al. 2011). Ret51 (a variant of the Ret transmembrane tyrosine kinase) activation by both glial cell line-derived neurotrophic factor (GDNF) and nerve growth factor (NGF) triggers the formation of RET51/FKBP52 complex (Fusco et al. 2010). Mutations in Ret51 and FKBP52 that disrupt this complex have been found in rare cases of parkinson’s disease (PD) (Fusco et al. 2010).

Overexpression of FKBP52 increases aggregation of alpha-synuclein, commonly observed in patients with PD (Gerard et al. 2010). However, FKBP52 seems to play an opposite role in Alzheimers disease (AD). Overexpressioin of FKBP52 in Drosophila suppressed the toxicity of amyloid beta, a protein which forms aggregates in AD (Sanokawa-Akakura et al. 2010). During early neural differentiation, FKBP52 enhances and FKBP51 inhibits neurite outgrowth (Quinta et al. 2010). Whereas in undifferentiated cells, FKBP52, HSP90 and p23 are located in the perinuclear region, they are rapidly dispersed in the cytoplasm upon induction of differentiation (Quinta & Galigniana 2012). Neuronal damage induces redistribution of these three proteins to the perinuclear region. During adipocyte differentiation, FKPB51 rapidly increases whereas FKBP52 decreases. In addition, FKPB51 translocates from the mitochondria to the nucleus, a process which is dependent on PKA (protein kinase A) (Toneatto et al. 2013). Also in other cell types, like 3T3 fibroblasts, FKBP51 has been found in the mitochondria, which is also true for GR, but this subcellular localization and the trafficking of both proteins is independent from each other (Gallo et al. 2011). FKPB51 protects cells from oxidative stress and translocates to the nucleus upon exposure to oxidative stress (Gallo et al. 2011). FKBP38 (not mentioned in the table) has an anti-apoptotic function by anchoring the pro-apoptotic proteins Bcl-2 and Bcl-x(L) to mitochondria (Shirane & Nakayama 2003).

Cyp40 (cyclophilin 40), also an immunophilin, is involved in steroid hormone receptor assembly. It prefers HSP90 over HSP70 binding (Carrello et al. 2004). Cyp40 binds to the DNA repair protein Ku70 and to Rack1 (receptor of activated protein kinase C 1) (Park et al. 2011). Rack1 promotes stabilization of HIF-1alpha (hypoxia inducible factor 1 alpha) in prostate cancer cells (Yu et al. 2014). However, RACK1 can also down-regulate HIF-1alpha after cobalt chloride treatment of MCF-7 breast cancer cells (Park et al. 2011).

PP5 (protein phosphatase 5, also abbreviated PPP5C) is a serine/threonine protein phosphatase that functions in signaling pathways that control cellular responses to stress, steroid hormones and DNA damage (Chinkers 2001; Yang et al. 2005; Hinds & Sanchez 2008). PP5 phosphatase activity is auto-inhibited in a state unbound to HSP90 or fatty acids by a special conformation of the TPR domain and a C-terminal subdomain. Binding to HSP90 or fatty acids (e.g. arachidonic acid) to the TPR domain activates phosphatase activity (Yang et al. 2005). PP5 is up-regulated in breast cancer, as shown by immunohistochemical staining of a tissue microarray (TMA) (Golden et al. 2008).

The translocation machinery (or “translocase”) of the outer mitochondrial membrane (TOM) is a multiprotein complex containing, among other proteins, TOM20 and TOM70 (named after their approximate molecular weight). Tom70 preferentially binds pre-proteins with internal targeting sequences (Brix et al. 1997). TOM70 binds to a multichaperone complex including HSP70 and HSP90 via specialized TPR domains in which three TPR domains are organized into a superhelical structure (Scheufler et al. 2000; Young et al. 2003a). Proteins are translocated through the import pore to the TIM (translocase of the inner mitochondrial membrane) machinery in an ATP-dependent manner (Young et al. 2003a). Pre-proteins with N-terminal pre-sequences are translocated with the help of Tom20, a process which is less dependent on chaperones (Brix et al. 1997).

TPR2 binds to Rad9 (Xiang et al. 2001), which is part of the Rad1-Hus1-Rad9 complex that is important for DNA-damage induced cell-cycle checkpoint activation (Parrilla-Castellar et al. 2004). The J-domain (DNAJ homology domain) of TPR2 regulates TPR2-Rad9 interaction (Xiang et al. 2001) and thereby potentially the stability of the checkpoint complex.

Deletion of CPR7 (7^th cyclophilin homolog, interacting with the yeast global transcriptional regulator Rpd3) caused severe growth defects when combined with mutations that decrease the amount of Hsp90 or Hop. Cpr7 null mutations caused defects in GR and pp60/v-src kinase activity (Duina et al. 1996b). In addition, Cpr7 deletion resulted in a significant impairment of cell division (Duina et al. 1996a). Both Cpr6 and Cpr7 interact with the ribosome and may be involved in the regulation of protein synthesis (Tenge et al. 2015).

The HSP90 homodimer can bind several co-chaperones at once, even different co-chaperones that bind to the TPR domain. For example, a HSP90(2)-FKBP52(1)-p23(2)-HOP(2)-complex could be demonstrated by immunoprecipitation, chromatographic methods and dynamic light scattering (Hildenbrand et al. 2011).

The chaperone cycle

After binding of the client protein to the HSP90 homodimer (open configuration), ATP binds to the NTD and induces transition of the complex to a closed ATP bound form. During proper folding, the conformation closes further and the NTDs dimerize transiently, coupled to ATP hydrolysis (Prodromou et al. 2000). Finally, the configuration opens up, the folded client protein is released and ADP dissociates (figure 2).

Open in a separate window

Figure 2

Simplified chaperone cycle. ATP and ADP bound states during protein folding as well as open configuration of HSP90 homodimer are shown. NTD, N-terminal domain; CTD, C-terminal domain; MD, middle domain.

For proper functioning of this simplified chaperone cycle, several co-chaperones are necessary. The interaction with co-chaperones occurs in a ordered, sequential manner (Li et al. 2011). The first co-chaperone entering the cycle is the HSP90 ATPase inhibitor Hop (Chen & Smith 1998), that binds to the Hsp90 homodimer. Together with Hop or after that, Hsc70 binds. After that, a PPIase, such as Fkbp51, Fkbp52 or Fkbp37 binds to the opposite side of the HSP90 dimer. HSP90 converts to a closed conformation after binding of ATP and binding of p23 (Johnson et al. 1994; Johnson & Toft 1995; Freeman et al. 2000; McLaughlin et al. 2006; Forafonov et al. 2008), releasing Hop and HSC70 from the complex, leaving space for another PPIase molecule. After ATP hydrolysis, all remaining co-chaperones are released from the HSP90 dimer which converts to an open conformation. This allows the start of another HSP90 cycle (Li et al. 2011). In a cell-free system with progesterone receptor (PR) as a client protein it could be shown that such an assembly/disassembly cycle has a t1/2 of about 5 min (Smith 1993).

Posttranslational modifications

The tyrosine (Y) phosphorylation of Hsp90AA1 and Hsp90AB1 induced by LPS in endothelial cells is mediated by pp60src (Barabutis et al. 2013). Mass spectrometry identified Y309 as a major site of Y phosphorylation on Hsp90AA1 (and Y300 on Hsp90AB1) (Barabutis et al. 2013). Phosphorylation of Ser225 and Ser254 (counted after removal of the first methionine) of HSP90AB1 inhibited binding of AHR (arylhydrocarbon receptor) (Ogiso et al. 2004). The phosphorylation of HSP90 is inhibited by 17-allyl-amino-demethoxy-geldanamycin (17-AAG) (Barabutis et al. 2013). DNA-activated protein kinase (DNA-PK), a protein that is important for DNA repair, phosphorylates two threonine residues in HSP90AA1. Since HSP90AB1 lacks these threonine residues, no DNA-PK dependent phosphorylation occurs in HSP90AB1 (Lees-Miller & Anderson 1989a).

Methylated HSP90AB1 increases the proliferation of cancer cells (Hamamoto et al. 2014). Methylation of lysines 531 and 574 of HSP90AB1 occurs by SMYD2 (SET/MYND domain containing 2) histone H3 lysine (K) 36-specific methyltransferase (KMT) (Donlin et al. 2012; Hamamoto et al. 2014). Methylated HSP90AB1 is associated with increased proliferation of cancer cells (Hamamoto et al. 2014). Acetylation seems to inhibit HSP90 function (Yang et al. 2013) as shown in a model of Gaucher disease. This is a lysosomal storage disorder caused by mutations in the enzyme glucocerebrosidase (GCase), leading to a misfolded enzyme. HSP90AB1 is essential for the degradation of misfolded GCase by guiding it to a valosin-containing protein (VCP)-associated proteasomal degradation pathway (Dai & Li 2001; Zhong et al. 2004; Yang et al. 2013). Inhibition of HDAC (histone deacetylases) keeps HSP90 in an acetylated state and inhibits GCase degradation, thereby alleviating the disease phenotype (Yang et al. 2013). Conversely, HSP90 also influences HDAC levels. Hsp90 interacts with HDAC6 and is a key effector of HDAC6 levels (New et al. 2013). Nitric oxide (NO) induces S-nitrosylation of HSP90AB1 at its C-terminus and thereby reduces its activity (Retzlaff et al. 2009). Obviously, also citrullination occurs on HSP90 proteins since autoantibodies against citrullinated HSP90AA1 and HSP90AB1 are detected in patients with rheumatoid arthritis-associated interstitial lung disease (RA-ILD), but not in patients with RA without ILD, MCTD (mixed connective tissue disease) and IPF (interstitial pulmonary fibrosis) (Harlow et al. 2013).

Inhibition of HSP90

Derivatives of geldanamycin, like 17-dimethylaminoethylamino-17-demethoxy-geldanamycin (17-DMAG) or 17-AAG bind to the ATP pocket in the NTD of HSP90 (Grenert et al. 1997; Prodromou et al. 1997b; Stebbins et al. 1997; Roe et al. 1999). The HSP90 inhibitor radicicol, a macrocyclic antifungal antibiotic, also binds to the ATP binding pocket of HSP90 (Schulte et al. 1998; Roe et al. 1999). Celastrol, a triterpenoid compound that inhibits HSP90AB1 activity, alters the three-dimensional structure of the co-chaperone p23 resulting in a more selective destabilization of steroid receptors (Chadli et al. 2010) or acts by disrupting the HSP90-Cdc37 complex (Zhang et al. 2009). Celastrol also inhibits the production of matrix metalloproteinases (MMPs), inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2) from chondrocytes derived from patients with osteoarthritis (Ding et al. 2013). This effect may be mediated by direct interaction of HSP90AB1 with these proteins or by its interaction with NF-kappaB (nuclear factor kappaB) (Ding et al. 2013). The natural product drug BTIMNP_D004 inhibits, dependent on cell type, HSP90AB1 and/or HSP90AA1 and a number of proliferation-associated signaling proteins (Samadi et al. 2009). A kind of non-specific HSP inhibition is exerted by free radicals formed during oxidative stress. They can lead to a cleavage of HSP90AB1 at two positions near the N-terminus of the protein, which occurs through a Fenton-Reaction (Beck et al. 2012).

Inductors of HSP90(AB1)

As already described, there is a great number of stress types that are able to induce HSP protein levels. In this paragraph, some studies are summarized that show a specific induction of HSP90AB1. Infection is a basic type of stress and therefore a common reason for HSP induction. Increased HSP90AB1 has been detected after influenza infection in infected cells (Wahl et al. 2010). Bacterial challenge in the abalone Haliotis tuberculata leads to an up-regulation of HSP90AB1 (Travers et al. 2010). LPS treatment or challenge with src-adenovirus lead to an increased phosphorylation of HSP90AB1 (Y300) in endothelial cells (Barabutis et al. 2013).

Necrosis causes an inflammatory reaction in order to clean up the damaged region. Extracellular HSP90AB1 can elicit such a response independently from IL-8 and MCP-1 in cultured retinal pigment epithelial (RPE) cells. A cell-impermeable geldanamycin derivative can inhibit the production of inflammatory cytokines in this model (Qin et al. 2011). H. pylori induces the translocation of HSP90AB1 from the cytosol to the membrane causing interaction of HSP90AB1 and Rac1, which leads to the activation of NADPH oxidase and production of ROS in gastric epithelial cells (Cha et al. 2010).

Heat shock or increased expression of HSP90AB1 restore reduced infectivity of HIV with incompletely processed capsid-spacer protein 1 (CA-SP1) (Joshi & Stoddart 2011; Joshi et al. 2013). Increased thermotolerance of hepatoma cells (Reuber H35 cells) in late G1/early S-phase as compared to G0 phase correlated with enhanced basal synthesis of HSP65, HSP68 and HSP90AB1 as well as with increased phosphorylation of non-histone chromosomal proteins (van Dongen et al. 1986). Not only heat stress but also cold stress can induce HSPs. HSP90AB1 is up-regulated during hibernation in the hearts of ground squirrels and protects them from ischemia-reperfusion injury and apoptosis (Grabek et al. 2011).

Benzo[a]pyrene (B[a]P), a cancer promoting agent, induces HSP90AB1 mRNA (Sadeu & Foster 2013). Chronic opioid treatment leads to increased HSP90AB1 transcript and protein levels. This contributed to an enhanced responsiveness of adenylate cyclase to forskolin treatment. Thus, HSP90AB1 contributes to opioid induced adenylate cyclase sensitization (Koshimizu et al. 2010). Overexpression of TRIM14 (T-cell receptor interacting molecule 14) leads to increased levels of HSP90AB1 (Nenasheva et al. 2014). Trim14 increases mesodermal differentiation and inhibits ectodermal differentiation of mouse ES cells (Novosadova et al. 2009).

Also normal developmental processes as shown in Drosophila can induce HSP (Lindquist 1986). Induction of HSP90AB1 occurs during adipogenesis in a cell culture model (Fromm-Dornieden et al. 2012).

Role in molecular evolution

Mutations can cause defects in protein folding. Their stability can often only be maintained by binding to HSPs. In addition, molecular evolution by gene duplication and subsequent progressive mutation is only conceivable with the help of a strong and versatile chaperone machinery (Rutherford & Lindquist 1998; Yahara 1999; Jarosz & Lindquist 2010; Jarosz et al. 2010).

Diseases with down-regulated HSP90

Hsp90ab1, expressed in oligodendrocyte precursor cells, has been shown to be down-regulated in a rat model of acute spinal cord injury (ASCI) (Zhou et al. 2014). HIV virus causes a progressive decline in CD4+ T-cells. This involves an aging-like phenotype with decline of telomerase activity. This is, at least partly, caused by a reduction of HSP90AB1 which is mediated by the HIV-protein Tat (Comandini et al. 2013). In a rat model of radiation-induced fibrosing alveolitis, it has been shown that HSP90AB1 was progressively down-regulated in radiation-damaged lung cells other than mast cells (Haase et al. 2014).

Diseases with up-regulated HSP90

HSP90 is increased in systemic sclerosis (SSc) of the skin. Inhibition of Hsp90 by 17-DMAG inhibited canonical TGF-beta signaling and completely prevented the stimulatory effects of TGF-beta on collagen synthesis and myofibroblast differentiation (Tomcik et al. 2014). Thus, HSP is pro-fibrotic in this model. Conversely, HSP90 inhibitors repress TGFbeta1 signaling (Noh et al. 2012) by a mechanism dependent on Smurf2-mediated degradation of TGFbetaRII. This lead to a decreased expression of alpha-smooth muscle actin, fibronectin, and collagen I and a decrease of renal fibrosis in an animal model (Noh et al. 2012).

DNA damage recognition/DNA repair/Free radicals

The tumor suppressor protein p53 is normally degraded after binding to the E3 ubiquitin ligase MDM2 (mouse double minute 2 homolog). Mutant p53 is commonly overexpressed in cancer cells. HSP90AB1 binds to mutant p53 in cancer cells and stabilizes it (Sepehrnia et al. 1996). P53 binds with its DNA binding domain to the middle and CTD of HSP90AB1 (Muller et al. 2004). It has been demonstrated that the complex is only stable in the presence of HSC70, HSP40, HOP and ATP (King et al. 2001). Degradation of wild type p53, which is initiated by MDM2, is inhibited by the co-chaperone TPR2 that binds p53 and thereby blocks complex formation with MDM2 (Kubo et al. 2013). In the latter complex, no HSP90 has been described (Kubo et al. 2013). After exposure to ionizing radiation, inhibition of HSP90 by a geldanamycin analogue inhibits p53 activation and subsequent apoptosis in human peripheral blood mononuclear cells (Fukumoto & Kiang 2011). In a cell-free system, HSP90AA1 has been shown to stabilize wild-type p53 and to maintain binding of p53 to the p21 promoter sequence (Walerych et al. 2004).

The stability of DNA-PKcs is dependent on the activity of HSP90, since inhibition of the HSP90 with radicicol decreased the expression of DNA-PKcs in HeLa cells (Falsone et al. 2005). HSP90(AA1 and AB1) specifically binds to REV1 (reversionless 1), a Y-family polymerase that causes lack of UV-induced reversion of the arg4–17 ochre allele in yeast and plays a central role in mutagenic translesion DNA synthesis (TLS) (Lawrence & Christensen 1976; Ohmori et al. 2001; Pozo et al. 2011). HSP90 inhibition disrupts the interaction between REV1 and monoubiqitinated PCNA (proliferating cell nuclear antigen), a molecule that recruits REV1 to damaged DNA sites. Thus, HSP90 inhibition suppresses TLS-mediated mutagenesis (Pozo et al. 2011). HSP90AB1 inhibition decreased IL-1beta induced NO (nitric oxide) production in chrondrocytes (Calamia et al. 2011).

Acknowledgments

Abbreviation

Footnotes

References