*Review* **Extracellular Heat Shock Protein-90 (eHsp90): Everything You Need to Know**

**Daniel Jay <sup>1</sup> , Yongzhang Luo <sup>2</sup> and Wei Li 3,\***


**Abstract:** "Extracellular" Heat Shock Protein-90 (Hsp90) was initially reported in the 1970s but was not formally recognized until 2008 at the 4th International Conference on The Hsp90 Chaperone Machine (Monastery Seeon, Germany). Studies presented under the topic of "extracellular Hsp90 (eHsp90)" at the conference provided direct evidence for eHsp90's involvement in cancer invasion and skin wound healing. Over the past 15 years, studies have focused on the secretion, action, biological function, therapeutic targeting, preclinical evaluations, and clinical utility of eHsp90 using wound healing, tissue fibrosis, and tumour models both in vitro and in vivo. eHsp90 has emerged as a critical stress-responding molecule targeting each of the pathophysiological conditions. Despite the studies, our current understanding of several fundamental questions remains little beyond speculation. Does eHsp90 indeed originate from purposeful live cell secretion or rather from accidental dead cell leakage? Why did evolution create an intracellular chaperone that also functions as a secreted factor with reported extracellular duties that might be (easily) fulfilled by conventional secreted molecules? Is eHsp90 a safer and more optimal drug target than intracellular Hsp90 chaperone? In this review, we summarize how much we have learned about eHsp90, provide our conceptual views of the findings, and make recommendations on the future studies of eHsp90 for clinical relevance.

**Keywords:** extracellular Hsp90; stress; mechanism of action; wound healing and cancer

### **1. Introduction**

For decades, the Heat Shock Protein-90 (Hsp90) family proteins have been recognized as ATP binding-dependent molecular chaperones inside almost all types of cells throughout evolution. This understanding has served as an indisputable foundation for both laboratory research and cancer clinical trials targeting the intracellular function of the Hsp90 family proteins [1–6]. Meanwhile, a cell-surface form of Hsp90-related molecule was reported as early as the late 1970s with several publications that appeared to challenge the definition of Hsp90 as an exclusively intracellular chaperone. The question was first raised in the 1990s by Csermely and colleagues, who stated that "the major cellular function of Hsp90 is probably not its chaperone behaviour, but its dynamic participation in the organization and maintenance of the cytoarchitecture" [7], although the exact nature of the dynamic participation was not further elaborated. Throughout the following decade, however, few in the Hsp90 field credited the possible existence of a non-chaperone form of Hsp90 and regarded the reported extracellular or secreted Hsp90 as artifacts, such as leakage by a small number of dying cells in culture. A breakthrough emerged in the 2000s when several laboratories independently demonstrated a critical role for secreted Hsp90 in various pathophysiological processes such as cancer cell invasion and wound healing. Increasing lines of evidence are raising the possibility that cell surface-bound, exosome-anchored, or

**Citation:** Jay, D.; Luo, Y.; Li, W. Extracellular Heat Shock Protein-90 (eHsp90): Everything You Need to Know. *Biomolecules* **2022**, *12*, 911. https://doi.org/10.3390/ biom12070911

Academic Editor: Chrisostomos Prodromou

Received: 2 June 2022 Accepted: 24 June 2022 Published: 29 June 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

simply free-secreted Hsp90 may serve as safer and more effective therapeutic targets than their intracellular counterparts in cancer and other inflammatory human disorders. This is especially relevant considering that targeting the intracellular ATP-dependent chaperone function of Hsp90 has encountered setbacks in clinical trials. Several generations of small molecule inhibitors have entered numerous cancer clinical trials since 1999, but to date none have received FDA approval [5,6]. In this review article, we provide a comprehensive walk-through of the discovery, characterization, mechanism of action, and evaluation by animal models and human patients of what is now collectively referred to as "extracellular Hsp90" (eHsp90). More importantly, we offer our answers for the fundamental question of why eHsp90 is chosen by evolution for duties that cannot be served by other conventional extracellular factors.

#### **2. History of eHsp90 Discovery**

In the late 1970s, several laboratories independently reported a glucose-regulated 90 kDa protein both on the surface and in the conditioned medium of tumour virus-infected mouse and human fibroblast cells [8–12]. In 1983, Hughes et al. provided direct evidence that Hsp90 protein is located on the external surface of macrophage and mouse embryo 3T3 cells [13]. Srivastava et al. then reported a membrane-associated 96-kDa protein in chemically induced sarcoma cells [14] and Ullrich et al. showed that a Hsp90-related protein was detected on the external surface of both Meth A tumour and NIH3T3 cells using antibody binding to the cells at 4 ◦C that prevented membrane internalization [15]. A follow-up study by Thangue and Latchman showed cell surface accumulation of Hsp90 in HSV-infected cells [16]. While the findings of these studies were intriguing, they may have not resonated at the time. The observations of extracellular Hsp90 in these studies were thought to be due to intracellular Hsp90 being released non-specifically from dead or dying cells, and there was little preclinical or clinical relevance available. In 1992, Erkeller Yuksel et al. reported that the external surface expression of Hsp90 was a feature of about 20% of the patients with systemic lupus erythematosus (SLE) and it correlated with the severeness of the disease [17]. After the finding, Thomiadou and Patsavoudi reported a 94-kDa "neuron-specific cell surface antigen" recognized by the monoclonal antibody 4C5 [18]. The 94-kDa protein was later identified as an Hsp90-related protein by the same group using mass spectrometry [19]. The take-home message of these studies was the association of the surface or secreted Hsp90 with inflammatory diseases and tissue development. The findings of these earlier studies were, however, largely overlooked by the Hsp90 community due to lack of evidence for active secretion and the undefined function of surface-bound or secreted forms of Hsp90.

In the early 2000s, two laboratories, which had never studied Hsp90 before, were independently searching for secreted proteins that support two distinct and mechanistically related pathophysiological processes, tumour cell invasion and skin wound healing. In 2004, Jay's group at Tufts University first reported the identification of a secreted protein from the conditioned medium of a fibrosarcoma cell line, HT-1080, and showed that the secreted protein promoted tumour cell invasion in vitro by activating the matrix metalloproteinase 2 (MMP2) [20]. In 2007, Li's laboratory at the University of Southern California reported the purification of a secreted protein from the conditioned medium of hypoxiastressed primary human dermal fibroblasts and keratinocytes through chromatography and showed that this secreted protein strongly stimulated skin cell migration in vitro and promoted wound healing in mice [21,22]. The common protein involved in tumour cell invasion in vitro and wound healing in vivo was identified as the secreted form of Hsp90α. Additional publications on secreted Hsp90α have since emerged and are beginning to receive attention. To provide a common terminology that covers the meanings of "cell surface-bound", "cell-released", "cell-secreted" Hsp90α, Isaacs and colleagues recommended "eHsp90" for extracellular Hsp90 which has since become widely accepted by the Hsp90 community [23]. Over the past 20 years, there have been two dozen excellent

review articles on eHsp90, especially eHsp90α [24–41], which we use as the stepping-stones for construction of this article.

#### **3. eHsp90**α **vs. eHsp90**β**: Who Calls the Shots and Why?**

In comparison to over 70 reports on eHsp90α as of May 2022, several studies in the past have also reported the presence of either eHsp90β alone or the eHsp90α and eHsp90β proteins together in the conditioned medium of various cell types [42,43]. This leads to the question of whether eHsp90β also has extracellular functions. If one only considers the functionality of a protein in its purified form as the ultimate evidence, the answer is negative. Cheng et al. showed that human recombinant Hsp90α (hrHsp90α), but not hrHsp90β, stimulated human keratinocyte migration [22]. Jayaprakash et al. showed that hrHsp90α, but not hrHsp90β, promoted wound healing in pigs [44]. Zou and colleagues demonstrated that the intravenous injection of hrHsp90α, but not hrHsp90β, protein strongly promoted tumour formation and lung metastasis in mice [45].

The human Hsp90α and Hsp90β proteins differ by a total of 100 amino acid residues along their respective 732- (Hsp90α) and 724- (Hsp90β) amino acid sequences, including 58 conservative and 42 non-conservative amino acid substitutions, in addition to 12 amino acid deletions in Hsp90β. The highest variations between Hsp90α and Hsp90β occur within the linker region (LR) and a part of the middle domain (M), with only 61% amino acid identity, the location where Li's group identified the functional "F-5" fragment of eHsp90α that promotes wound healing [46,47]. Mouse genetic studies also showed distinct and non-compensating roles for Hsp90α and Hsp90β during development. Voss et al. reported that Hsp90β gene knockout causes a defect in placental labyrinth formation, resulting in mouse embryonic lethality on E10.5 [48]. In contrast, mice with either chaperone-defective mutations in Hsp90α [49,50] or complete Hsp90α knockout [51] showed indistinguishable difference in their phenotypes from their wild-type counterparts. The straightforward interpretations were that (1) Hsp90β is more critical than Hsp90α during mouse development and (2) Hsp90α is not required under homeostasis.

#### **4. Is eHsp90 Secreted by Living Cells on Purpose or Leaked by Dead Cells by Accident?**

While it is technically difficult to prove that eHsp90, specifically eHsp90α, does not result from the leakage of intracellular Hsp90 from a small number of dead cells in culture, several lines of evidence strongly support that eHsp90α is actively secreted. Studies showed that the quantity of eHsp90α was less or undetectable from secreted molecules of normal cells under physiological conditions in vitro, i.e., in serum-containing and pH-balanced medium under normoxia at 37 ◦C. In comparison, several fold higher eHsp90α proteins became detectable from the conditioned medium of the same cells under a variety of medically-defined stress signals including reactive oxygen species (ROS), heat, hypoxia, gamma-irradiation, UV, and tissue injury [28,52,53]. In contrast, many tumour cells constitutively secrete eHsp90α due to intrinsic oncogenes, such as overexpressing HIF-1α [21], or mutant forms of tumour suppression genes including p53 [42,54]. Eustace et al. showed only Hsp90α and not Hsp90β in the conditioned medium of tumour cells, suggesting a specific secretion of Hsp90α rather than the non-specific release of both forms from dead cells [20]. Cheng and colleagues showed that while both TGFα and EGF bind and signal through EGFR and both promote cell survival and cell growth, only TGFα stimulates the Hsp90α translocation to plasma membrane and secretion to the extracellular environment by primary human keratinocytes, the most critical cell type for skin wound healing [22]. Finally, the stress-induced secretion of Hsp90α was further substantiated by an in vivo observation that skin injury caused up to a 10-fold increase in eHsp90α deposition into the wound bed in a time-dependent fashion. As shown in Figure 1, skin injury in pigs causes an accumulatively increased deposition of eHsp90α into the wound bed in a time-dependent fashion. Since the location with increased anti-Hsp90α antibody staining includes areas in the skin dermis that does not have the continued presence of cells, the

*Biomolecules* **2022**, *12*, x 4 of 20

massive staining cannot be explained by increased intracellular Hsp90α [55]. This finding provides the first in vivo evidence of biological stress-induced Hsp90α secretion. This finding provides the first in vivo evidence of biological stress-induced Hsp90α secretion.

bed in a time-dependent fashion. Since the location with increased anti-Hsp90α antibody staining includes areas in the skin dermis that does not have the continued presence of cells, the massive staining cannot be explained by increased intracellular Hsp90α [55].

**Figure 1.** Tissue damage induces massive deposition of eHsp90α into the wound bed**.** Pig skin is biologically closest to the human skin. 1.5 cm × 1.5 cm full-thickness excision wounds were created **Figure 1.** Tissue damage induces massive deposition of eHsp90α into the wound bed. Pig skin is biologically closest to the human skin. 1.5 cm × 1.5 cm full-thickness excision wounds were created in the indicated area of pig torso. Full wedge (2 cm) biopsies cross the wound were made on the indicated days and immediately frozen on dry ice. Sections of the biopsies were stained with an anti-Hsp90α antibody. The red arrows point out the locations of the specific antibody staining (brown). Quantitation of the staining in blue boxes was done using Gabriel Landini's "color deconvolution" and ImageJ analysis. The intensity readings were converted to Optical Density (OD) (The image was taken from reference [55] with permission).

So far, the studies on mechanism(s) of eHsp90α secretion have raised more questions than answers. Two laboratories reported that the secretion can be regulated by either phosphorylation or the C-terminal amino acid EEVD motif of the Hsp90α protein [56,57]. Luo's laboratory further identified a critical role for Rab coupling protein (RCP) in mutp53 induced Hsp90α secretion [54]. Several studies suggested that eHsp90α is secreted via exosomes, based on the observation that DMA (Dimethyl amiloride), an inhibitor of the exosome secretion pathway, blocks Hsp90α secretion both in HIF-1-overexpressing tumour cells and TGFα-stimulated human keratinocytes cells, where Hsp90α was associated with isolated exosome fractions. Therefore, eHsp90α is secreted via the non-classical exosome trafficking pathway [58–61]. Guo and colleagues further identified the proline-rich Akt substrate of 40 kDa (PRAS40) as the unique downstream effector that mediates TGFαstimulated Hsp90α secretion via exosomes [62]. Tang and colleagues have recently made an interesting observation that eHsp90α is located at the external surface of tumour cellsecreted exosomes [63]. However, recent observations suggest that approximately 90% of both normal and tumour cell- secreted Hsp90α is not associated with secreted exosomes isolated by ultracentrifugation (C. Cheng, X, Tang and W. Li, unpublished; A. Bernstein and D. Jay, unpublished). Taken together, eHsp90α is a secreted protein by cells under either internal or external stress. This general understanding is depicted in Figure 2, where eHsp90α promotes tissue repair under physiological conditions or promotes tumorigenesis under pathological conditions, defined as a "double-edged sword" by Hence and colleagues [30]. So far, the studies on mechanism(s) of eHsp90α secretion have raised more questions than answers. Two laboratories reported that the secretion can be regulated by either phosphorylation or the C-terminal amino acid EEVD motif of the Hsp90α protein [56,57]. Luo's laboratory further identified a critical role for Rab coupling protein (RCP) in mutp53-induced Hsp90α secretion [54]. Several studies suggested that eHsp90α is secreted via exosomes, based on the observation that DMA (Dimethyl amiloride), an inhibitor of the exosome secretion pathway, blocks Hsp90α secretion both in HIF-1-overexpressing tumour cells and TGFα-stimulated human keratinocytes cells, where Hsp90α was associated with isolated exosome fractions. Therefore, eHsp90α is secreted via the non-classical exosome trafficking pathway [58–61]. Guo and colleagues further identified the proline-rich Akt substrate of 40 kDa (PRAS40) as the unique downstream effector that mediates TGFα-stimulated Hsp90α secretion via exosomes [62]. Tang and colleagues have recently made an interesting observation that eHsp90α is located at the external surface of tumour cell-secreted exosomes [63]. However, recent observations suggest that approximately 90% of both normal and tumour cell- secreted Hsp90α is not associated with secreted exosomes isolated by ultracentrifugation (C. Cheng, X, Tang and W. Li, unpublished; A. Bernstein and D. Jay, unpublished). Taken together, eHsp90α is a secreted protein by cells under either internal or external stress. This general understanding is depicted in Figure 2, where eHsp90α promotes tissue repair under physiological conditions or promotes tumorigenesis under pathological conditions, defined as a "double-edged sword" by Hence and colleagues [30].

in the indicated area of pig torso. Full wedge (2 cm) biopsies cross the wound were made on the indicated days and immediately frozen on dry ice. Sections of the biopsies were stained with an anti-Hsp90α antibody. The red arrows point out the locations of the specific antibody staining (brown). Quantitation of the staining in blue boxes was done using Gabriel Landini's "color deconvolution" and ImageJ analysis. The intensity readings were converted to Optical Density (OD) (The

*Biomolecules* **2022**, *12*, x 5 of 20

image was taken from reference 55 with permission).

**Figure 2.** Secretion of eHsp90α by normal cells under medically defined stress and by tumour cells driven by oncogenic signals. Almost all kinds of medically defined stresses have been shown to **Figure 2.** Secretion of eHsp90α by normal cells under medically defined stress and by tumour cells driven by oncogenic signals. Almost all kinds of medically defined stresses have been shown to trigger eHsp90α secretion in a wide variety of cell types. Tumours have either constitutively activated oncogenes or mutant tumour suppressor genes that each triggers eHsp90α secretion even in the absence of environmental stress cues. The mechanisms by which the stress and oncogenic signals cause Hsp90 secretion remain largely unstudied, in which exosome-mediated secretion of Hsp90α only accounts for 10% of the total secreted Hsp90α in both normal and tumour cells. The reported optimal working concentration for the full-length eHsp90α protein was around 3–10 µg/mL.

#### **5. Two Main Biological Functions of eHsp90**α

#### *5.1. Promoting Cell Survival under Ischemic Stress*

Shortly after tissue injury, the broken blood vessels clot and cells in the injured tissue encounter an ischemic (paucity of nutrient and oxygen) environment. The immediate challenge the cells face is survival, at least temporarily, by adapting a self-supporting mechanism. Similarly, when tumour cells invade surrounding tissues too quickly and temporarily outstrip the nearest blood vessel for 150 mirom or more, they similarly face the stress of ischemia [64]. Under these conditions, the tumour cells must find an autocrine cycle to survive without the help from blood vessels. Bhatia and colleagues showed that topical application of hrHsp90α to burn wounds in pigs prevented heat-induced skin cell apoptosis around the hypoxic wound bed [65]. Dong and colleagues demonstrated that eHsp90α protected tumour cells from hypoxia-triggered apoptosis, whereas neutralizing eHsp90α function with a monoclonal antibody enhanced hypoxia-induced tumour cell apoptosis [66]. Gao and colleagues reported that extracellular supplementation with hrHsp90α (10 µM or ~1 mg/mL) protein promoted rat bone mesenchymal stem cell (MSC) survival and prevented cell apoptosis under ischemic conditions by activating the Akt and ERK kinases [67]. Cheng et al. showed that 10 µg/mL hrHsp90α protein stimulated the maximum migration of primary human skin cells [22] and Dong et al. reported that a similar dosage of hrHsp90α prevented the death of Hsp90α-KO MDA-MB-231 cells under hypoxia [66]. Nonetheless, the conformation and composition of the eHsp90α under the above circumstances is less clear in comparison to its intracellular counterpart.

#### *5.2. Promoting Cell Motility (Not Growth) during Tissue Repair and Tumour Invasion*

The initial indication that eHsp90α regulates cell migration was reported in the 1990s by Patsvoudi's group, who showed that a monoclonal antibody against a mouse granule cell surface antigen called 4C5 inhibited the cell migration during cerebellar development [68,69]. The same group later confirmed by immunoprecipitation followed by mass spec that the 4C5 antigen is related to Hsp90 protein [19]. During the same period, Jay's group showed that eHsp90α from the conditioned medium of tumour cells was required for tumour cell invasion via activation of MMP2 in vitro [20]. The direct evidence that eHsp90α protein alone acts as a *bona fide* pro-motility factor came from Li's group that demonstrated hrHsp90α, but not hrHsp90β, stimulated primary human dermal fibroblasts and keratinocyte migration in the total absence of serum factors. Moreover, the pro-motility effect of hrHsp90α could reach approximately 60% of the total pro-motility of 10% FBS-containing medium. Under similar conditions, however, hrHsp90α showed little mitogenic effect on cell growth. More surprisingly, both the wild type and ATPase-defect mutant proteins of Hsp90α bind the cell surface receptor LRP-1 (low-density lipoprotein receptor-related protein 1) and had compatible prom-motility effects on the same cells [21,22].

#### **6. Mechanisms of Action by eHsp90**α

By and large, there have been two major parallel mechanisms of action proposed for eHsp90α [28]. The central debate is whether eHsp90α still acts as an ATP-dependent chaperone outside the cell or alternatively acts as a previously unrecognized signalling molecule no longer dependent on ATP hydrolysis. Eustace and colleagues tested DMAG-N-oxide, a cell membrane-impermeable geldanamycin/17-AAG-derived inhibitor that targets the ATPase activity of Hsp90, and showed that it inhibits tumour cell invasion [20]. Similarly, Tsutsumi and colleagues showed that the DMAG-N-oxide inhibitor reduced the invasion of several cancer cell lines in vitro and lung colonization by B16 melanoma cells in mice [70]. Furthermore, Sims et al. showed that blocking ATPase using ATP-gamma S actually increased the ability of hrHsp90α to activate MMP2 in vitro [71]. In particular, a recent elegant study from Bourboulia's group showed that TIMP2 and AHA1 act as a molecular switch for eHsp90α that determines the inhibition or activation of the eHsp90α client protein MMP2 [72]. Song and colleagues showed that Hsp90α, but not Hsp90β,

stabilized MMP2 and protected it from degradation in tumour cells in an ATP-independent manner and was mediated by the middle domain of Hsp90α binding to the C-terminal hemopexin domain of MMP2 [73]. Taken together, these studies suggest that the N-terminal ATP-binding domain and the intrinsic ATPase of Hsp90α remain essential for eHsp90α function outside of the cells. Results of other studies from different laboratories also supported the "eHsp90α chaperone mechanism" via their extracellular client proteins, most noticeably MMP2, MMP9, and TLR, just to mention a few. To avoid redundance in this special issue, we refer readers to two excellent review articles, a prior one by Wong and Jay [32] and the current one in this special issue by Bourboulia and colleagues for more detailed analysis of this mechanism.

On the other hand, the ATPase-independent mechanism has largely focused on the so-called "eHsp90α > LRP-1" signalling pathway [28]. Li's laboratory utilized both deletion and site-directed mutagenesis to narrow down the essential epitope along the 732-amino acid human eHsp90α for supporting the pro-survival, pro-motility, and pro-invasion activity of eHsp90α in vitro and in vivo. First, Cheng and colleagues reported that the ATPasedefective mutants, Hsp90α-E47A (~50% ATPase activity), Hsp90α-E47D (ATPase-defect), and Hsp90α-D93N (ATPase-defect), showed an indistinguishable degree of pro-motility activity from the Hsp90α-wt protein on primary human skin cells in vitro [22]. Second, they narrowed down the pro-motility activity to a 115-amino acid fragment called F-5 (aa-236 to aa-350) between the LR (linker region) and the M (middle domain of human) Hsp90α, as previously mentioned. They demonstrated that the F-5 peptide alone promoted skin cell migration in vitro and wound healing in vivo as effectively as the full-length Hsp90αwt [46]. Third, they illustrated the so-called "eHsp90α > LRP-1" signalling pathway as: (1) the subdomain II in the extracellular part of the low-density lipoprotein receptor-related protein-1 (LRP-1) that receives the eHsp90α signal; (2) the NPVY, but not NPTY, motif in the cytoplamic tail of LRP-1 that connects the eHsp90α signalling to the serine-473, but not threonine-308, phosphorylation in Akt kinases and (3) activated Akt1 ang Akt2 trigger cell migration [47]. Finally, within the F-5 fragment, Zou and colleagues identified a dual lysine motif (Lys-270/Lys-277) that are evolutionarily conserved in all members of the Hsp90α subfamily but absent in all Hsp90β subfamily members. Mutations at the lysine residues eliminated Hsp90α's ability to promote cell migration in vitro and tumour formation in vivo. Substitutions of the two different amino acids at the corresponding sites in Hsp90β granted Hsp90β with pro-motility activity like Hsp90α [45]. These authors presented an illustration of the F-5 fragment and the dual lysine motif locations in a schematic monomer structure of Hsp90α, as shown in Figure 3, which shows a potential target in eHsp90α for therapeutics. These findings suggest that the N-terminal ATPase domain and the C-terminal dimer-forming and co-factor-binding domain are dispensable for eHsp90α function. More interestingly, Gopal showed a novel crosstalk mechanism involving the eHsp90α-LRP1 dependent regulation of EphA2 function, in which the eHsp90α-LRP1 signalling axis regulates AKT signalling and EphA2 activation during glioblastoma cell invasion [23]. In addition, Tian et al. showed that clusterin served as an eHsp90α modulator to synergistically promote EMT (epithelial-to-mesenchymal transition) and tumour metastasis via LRP1 [74]. Besides binding to MMP2 and LRP-1, Garcia et al. reported that eHsp90α binds to the type I TGFβ receptor to stimulate collagen synthesis, which provides pavement for cell attachment and migration [75]. Nonetheless, the chaperone-dependent mechanism, such as the activation of MMP2, and the chaperone-independent mechanism, such as F-5 binding the LRP-1 receptor, do not necessarily have to be mutually exclusive, as schematically depicted in Figure 4, which may represent two parallel mechanisms of action by eHsp90α. The selectivity and specificity of these two pathways under various pathophysiological conditions remain to be further studied.

*Biomolecules* **2022**, *12*, x 8 of 20

further studied.

further studied.

**Figure 3.** The F-5 fragment is located at the surface of the Hsp90α protein. Based on the previously evaluated crystal structure of a monomer Hsp90α protein with the NTD (green), MD (blue) and CTD (red) domains, the F-5 fragment containing Lysine-270 and lysine-277 is located in the unstructured linker region (LR) between the NTD and MD domains. Inhibitors such as monoclonal antibodies, targeting the dual lysine residues (in enlarged box), are potential anti-tumour therapeutics. **Figure 3.** The F-5 fragment is located at the surface of the Hsp90α protein. Based on the previously evaluated crystal structure of a monomer Hsp90α protein with the NTD (green), MD (blue) and CTD (red) domains, the F-5 fragment containing Lysine-270 and lysine-277 is located in the unstructured linker region (LR) between the NTD and MD domains. Inhibitors such as monoclonal antibodies, targeting the dual lysine residues (in enlarged box), are potential anti-tumour therapeutics. **Figure 3.** The F-5 fragment is located at the surface of the Hsp90α protein. Based on the previously evaluated crystal structure of a monomer Hsp90α protein with the NTD (green), MD (blue) and CTD (red) domains, the F-5 fragment containing Lysine-270 and lysine-277 is located in the unstructured linker region (LR) between the NTD and MD domains. Inhibitors such as monoclonal antibodies, targeting the dual lysine residues (in enlarged box), are potential anti-tumour therapeutics.

necessarily have to be mutually exclusive, as schematically depicted in Figure 4, which may represent two parallel mechanisms of action by eHsp90α. The selectivity and specificity of these two pathways under various pathophysiological conditions remain to be

necessarily have to be mutually exclusive, as schematically depicted in Figure 4, which may represent two parallel mechanisms of action by eHsp90α. The selectivity and specificity of these two pathways under various pathophysiological conditions remain to be

**Figure 4.** Two proposed mechanisms of action by eHsp90α. eHsp90α acts via an ATPase-dependent or ATPase-independent mechanism, which is determined by different binding partners, as shown. It is possible that the two mechanisms take place in parallel and work synergistically to achieve the ultimate goal under pathophysiological conditions. **Figure 4.** Two proposed mechanisms of action by eHsp90α. eHsp90α acts via an ATPase-dependent or ATPase-independent mechanism, which is determined by different binding partners, as shown. It is possible that the two mechanisms take place in parallel and work synergistically to achieve the ultimate goal under pathophysiological conditions. **Figure 4.** Two proposed mechanisms of action by eHsp90α. eHsp90α acts via an ATPase-dependent or ATPase-independent mechanism, which is determined by different binding partners, as shown. It is possible that the two mechanisms take place in parallel and work synergistically to achieve the ultimate goal under pathophysiological conditions.

#### **7. Preclinical Studies of eHsp90α 7. Preclinical Studies of eHsp90α 7. Preclinical Studies of eHsp90**α

eHsp90α has been studied in several human disease models in animals, including cancer, wound healing, idiopathic pulmonary fibrosis (IPF) and wasting syndrome (WS). The implication of eHsp90α in blood circulation supporting tumour metastasis in a eHsp90α has been studied in several human disease models in animals, including cancer, wound healing, idiopathic pulmonary fibrosis (IPF) and wasting syndrome (WS). The implication of eHsp90α in blood circulation supporting tumour metastasis in a eHsp90α has been studied in several human disease models in animals, including cancer, wound healing, idiopathic pulmonary fibrosis (IPF) and wasting syndrome (WS). The implication of eHsp90α in blood circulation supporting tumour metastasis in a number of animal models is especially encouraging considering the long-term and heavy emphasis of Hsp90 on cancer.

#### *7.1. Wound Healing*

When tissue is injured and the broken blood vessels clotted, all of the cells surrounding the wound bed face ischemic stress, as previously described. Bhatia and colleagues made full-thickness skin wounds in pigs, biopsied the wounds over time, and immunostained the tissue samples with an anti-Hsp90α antibody. They found a massive, time-dependent increase in the antibody staining in both the epidermis and dermis [56] (see Figure 1). Song and Luo reported that eHsp90α localized on blood vessels in the granulation tissue of wounded skin and promoted angiogenesis during wound healing in mice [76]. A series of studies from Li's group demonstrated that topical application of hrHsp90α, but not hrHsp90β, strongly promoted closure of trauma (excision), burn, and diabetic wounds in mice and pigs. In reserves, topically administered antibodies against eHsp90α blocks wound closure [21,44,55,65,77]. Bhatia and colleagues carried out a clever study by taking advantage of Hsp90α transgenic mice where the Hsp90α's intracellular chaperone function is nullified but the truncated Hsp90α protein still contains the entire F-5 region. They found that these mice heal skin wounds as efficiently as their wild-type counterparts, indicating that the chaperone function of Hsp90α is dispensable. However, topical application of mAb 1G6D7 against eHsp90α inhibited the wound healing, suggesting an essential role for eHsp90α instead [55]. As previous mentioned, the wound healing-promoting effect of the full-length eHsp90α is entirely replicable by the F-5 fragment [46], which is currently undergoing clinical trials for the treatment of diabetic foot ulcers.

#### *7.2. Tissue Fibrosis*

While Hsp90α enhances wound healing, excess eHsp90α in the injured lung may do more harm than good [78]. Pulmonary fibrosis is characterized by overactivated lung fibroblasts and massive collagen deposition by the cells at the injured site. Using BLMinduced pulmonary injury and the fibrosis mouse model, which represents failed wound healing, Dong and colleagues showed that mAb 1G6-D7, a monoclonal antibody against eHsp90α, inhibited eHsp90α function and significantly protected against BLM-induced pulmonary fibrosis by ameliorating fibroblast overactivation and ECM production [79]. The same group proposed a possible mechanism by which eHsp90α links the ER stress to the PI-3K-Akt pathway [80]. Ballaye and colleagues reported the significant increase of both eHsp90α and eHsp90β in the circulation of patients with idiopathic pulmonary fibrosis (IPF), and the higher levels correlated with disease severity. They found that eHsp90α signalled through LRP-1 to promote myofibroblast differentiation and persistence in a rat ex vivo model [81]. Together, the above studies argue that the specific inhibition of eHsp90α is a promising therapeutic strategy to reduce pro-fibrotic signalling in IPF.

#### *7.3. Wasting Syndrome*

Wasting syndrome (WS) refers to the unwanted weight loss of more than 10 percent of a person's body weight, with diarrhea, weakness, and fever that can last up to 30 days. WS is often a sign of disease, such as cancer, AIDS, heart failure, or advanced chronic obstructive pulmonary disease. "Cachexia", characterized by muscle wasting, is a major contributor to cancer-related mortality. A recent study by Zhang et al. reported elevated serum Hsp70 and Hsp90 in Lewis lung carcinoma (LLC)-bearing mice. The tumour-released and exosome-bound eHsp90 and eHsp70 were both necessary and sufficient to induce muscle wasting in a syngeneic tumour mouse model [82]. These studies may suggest clinical value in inhibiting eHsp90 for WS.

#### *7.4. Tumorigenesis*

Given the specific supporting role of eHsp90α in cancer, and the failure of many clinical trials using pan- inhibitors targeting all intracellular Hsp90 chaperone members, several groups have reported on the benefit of selectively inhibiting eHsp90α for reducing tumour metastasis and improving patient survival. Stellas et al. reported that intraperitoneal injection with monoclonal antibody (100–200 µg per mouse daily) against 4C5 antigen

(cell surface eHsp90α) into C57BL/6 mice 24 h following tail vein injection with B16-F10 melanoma cells reduced tumour lung colonization and improved the survival of the mice in reference to placebo-treated mice [83]. These authors reported a similar finding using a human breast cancer xenograft model and showed that the antibody disrupted interactions of eHsp90 with MMP2 and MMP9 [84]. Tsutsumi and colleagues showed that DMAG-N-oxide (a membrane-impermeable version of 17-DMAG inhibitor) blocked lung colonization by B16 melanoma cells in nude mice [71]. Results of the study suggest that the N'-terminal ATP-binding of Hsp90α is still required for eHsp90α function. Using a different anti-Hsp90α monoclonal antibody, Song and colleagues showed the dose-dependent inhibition of tumour growth and angiogenesis of B10-F10 cells in nude mice [74]. Using orthotopic breast cancer mouse models, Hou and colleagues showed that the injection of hrHsp90α protein increased primary tumour lymphatic vessel density and sentinel lymph node metastasis. In reverse, injection of another independent anti-Hsp90α neutralizing antibody reduced 70% of lymphatic vessel density and 90% of sentinel lymph node metastasis [85]. Using the well-known human triple negative breast cancer cell line, MDA-MB-231, xenograft mouse model, two laboratories independently reported a critical role for eHsp90α in tumour growth and lung metastasis. Stivarou and colleagues showed that injection with an antibody against 4C5 antigen (Hsp90) inhibited both *de novo* tumour growth and growth of already established mammary tumours [86]. Zou and colleagues demonstrated that injection with hrHsp90α, but not hrHsp90β, protein rescued the tumorigenesis of Hsp90αknockout MDA-MB-231 cells in nude mice. More interestingly, the authors showed that the ATPase-defective Hsp90α (Hsp90α-D93N) protein showed exactly the same effect as the wild type Hsp90α on tumour formation and lung metastasis. In reverse, injection with the monoclonal antibody mAb1G6-D7, not only blocked de novo tumour formation and lung metastasis, but also significantly reduced (~35%) the continued growth of already formed tumours [45]. Consistently, Secli el al recently reported that "Morgana", a co-chaperone of eHsp90α, induced cancer cell migration through TLR2, TLR4, and LRP1. A monoclonal antibody targeting Morgana inhibited mouse breast cancer cells, EO771, from metastasizing to the lung in C57BL/6 mice [87]. Milani and colleagues established mouse models with human acute lymphoblastic leukemia (ALL) and showed that the background plasma level of eHsp90α was below 1ng/mL blood in healthy mice, whereas the plasma level of eHsp90α was elevated into the 100–150 ng/mL range within two months in a fashion that closely correlated with the increased percentage of hCD45+ cells, a monitoring marker of ALL, in the blood, bone marrow, liver, and spleen of the animals [88]. A recent study by Luo's group showed that PKM2 (pyruvate kinase M2)-like eHsp90α is secreted by lung cancer cells and detected in blood samples of human cancer patients. The injection of mouse recombinant PKM2 protein into blood circulation promoted tumour metastasis to the lung via binding to integrin β1 [89]. Since PKM2 is associated with Hsp90α inside cells [90], it is possible that the secreted PKM2 is in complex with eHsp90α, which remains to be experimentally confirmed.

#### **8. Clinical Studies of eHsp90**α **in Patients with Cancer and Inflammatory Disorders**

Since 2008, close to two dozen clinical studies have compared the eHsp90α levels in blood circulation between healthy humans and patients with various types of cancers and other inflammatory diseases. The cancers from the patients include all of the NCI (National Cancer Institute, USA)-listed major human cancers. Due to space limitations, we are unable to describe each of the individual studies and their findings in detail. Rather, we chose to summarize the common findings of these studies, i.e., elevated plasma eHsp90α in circulation, in Table 1. While the exact amount of plasma/serum eHsp90α markedly vary (from pg/mL to mg/mL) among different reports (though they all used ELISA-based detection methods), a majority of the studies showed a statistically significant increase in cancer patients compared to normal patients in a range of sub µg/mL. More intriguingly, the higher levels of plasma eHsp90α closely correlated with the later stages of the diseases, such as metastasized tumours. These studies raise the possibility of utilizing plasma

eHsp90α as a new serum marker for cancer detection and therapeutic targeting, as well as for other chronic inflammatory diseases in humans.

**Table 1.** Summary of clinical studies on plasma eHsp90 in blood circulation \*.



\* Note: The reported original data on plasma Hsp90 from patients varied dramatically from pg/mL to mg/mL, while the reasons remain unclear. Two presentations, "range" and "average", by the original studies were adopted here. Nonetheless, higher plasma Hsp90 levels in patients' blood are evident. IQR: Interquartile range (IQR).

Taking tumorigenesis as an example, the recognized five steps of tumour development include gene mutations, hyperplasia, dysplasia, primary tumour formation, and tumour metastasis [119–122]. The vast majority of the United States Food and Drug Administration (FDA)-approved oncology drugs (>1000 by the end of 2020) target primary tumours, even though cancer patients die predominantly from metastasis [123]. These drugs extend patients' survival for variable periods of time, but many lose efficacy shortly after several months of treatment due to new mutations generated in the tumours. On the other hand, tumour metastasis begins with local expansion and invasion of the tumour at the primary organ driven by oncogenic signals with tumour microenvironmental assistance. Tumour cells migrate away from their origin and infiltrate into new surrounding tissues in which the tumour cells intravasate into the nearest blood circulation or the lymphatic system. After entry into the circulation, the tumour cells become known as circulating tumour cells (CTCs). Continued distal metastasis requires the tumour cells to survive and disseminate via the blood circulation, so-called hematogenous metastasis. Only a small number of CTCs successfully extravasate by crossing the endothelial barrier, leaving the circulation, and entering a distant organ. Thus, identification of a plasma factor that provides critical assistance for CDC to achieve the ultimate success of metastasis could lead to the soughtafter target for next generation of anti-tumour therapeutics. If elevated plasma eHsp90α

in cancer patients proves to promote CTC survival and dissemination through blood circulation during metastasis, interruption of the plasma eHsp90α function by antibodies that target the F-5 region of eHsp90α, as schematically proposed in Figure 5, would be an attractive approach to slow down tumour metastasis and buy time for patients to eliminate the primary tumours via surgery and the currently available therapies. For the next few years, the potential importance of the plasma eHsp90α reported in human cancer patients must be carefully studied by engineering the pathological plasma eHsp90α levels in Hsp90α-knockout animal models. *Biomolecules* **2022**, *12*, x 13 of 20 human cancer patients must be carefully studied by engineering the pathological plasma eHsp90α levels in Hsp90α-knockout animal models.

**Figure 5.** Plasma eHsp90α as a potential target for therapeutics to block tumour metastasis. Findings of the clinical studies shown in Table 1 have raised an exciting possibility that monoclonal antibody therapeutics against plasma eHsp90α block tumour metastasis. Since plasma eHsp90α is low and unessential for homeostasis, targeting plasma eHsp90α in cancer patients may prove to be safer and more effective than targeting the intracellular Hsp90α and Hsp90β. **Figure 5.** Plasma eHsp90α as a potential target for therapeutics to block tumour metastasis. Findings of the clinical studies shown in Table 1 have raised an exciting possibility that monoclonal antibody therapeutics against plasma eHsp90α block tumour metastasis. Since plasma eHsp90α is low and unessential for homeostasis, targeting plasma eHsp90α in cancer patients may prove to be safer and more effective than targeting the intracellular Hsp90α and Hsp90β.

#### **9. Is eHsp90α a More Effective and Safer Drug Target Than Intracellular Hsp90? 9. Is eHsp90**α **a More Effective and Safer Drug Target than Intracellular Hsp90?**

As mentioned at the beginning of this article, over the past two decades, intracellular Hsp90 chaperones (Hsp90α, Hsp90β, and possibly other related chaperones) have been targeted by at least 18 small molecule inhibitors binding to the N-terminal ATP/ADP binding site of the proteins in more than 60 cancer clinical trials [4–6]. To date none has received FDA approval for clinical treatment of human cancers due to various speculative reasons [124]. A recent study raised a serious and previously overlooked concern that there might be a complete lack of a druggable window between tumour and normal tissues for ATP-binding inhibitors. Tang and colleagues showed a wide range of Hsp90 expression in different host organs which further exhibited a wide range of toxicity to an ATP-binding inhibitor and heterogenous responses against the conversional theory to the same ATP-binding inhibitor among different tumour cells. These findings could seriously complicate patient and biomarker selections, toxicity readout, and efficacy of the drug candidates for clinical trials [125]. In contrast to the essential role of the intracellular Hsp90, especially Hsp90β for cell and organ homeostasis, the requirement of eHsp90α for life has not been reported. Instead, only when tissue homeostasis is broken, such as during wound healing or disease occurrence such as tumour growth, does eHsp90α then come into the picture. To support this notion, Bhatia and colleagues showed that selectively As mentioned at the beginning of this article, over the past two decades, intracellular Hsp90 chaperones (Hsp90α, Hsp90β, and possibly other related chaperones) have been targeted by at least 18 small molecule inhibitors binding to the N-terminal ATP/ADP binding site of the proteins in more than 60 cancer clinical trials [4–6]. To date none has received FDA approval for clinical treatment of human cancers due to various speculative reasons [124]. A recent study raised a serious and previously overlooked concern that there might be a complete lack of a druggable window between tumour and normal tissues for ATP-binding inhibitors. Tang and colleagues showed a wide range of Hsp90 expression in different host organs which further exhibited a wide range of toxicity to an ATP-binding inhibitor and heterogenous responses against the conversional theory to the same ATP-binding inhibitor among different tumour cells. These findings could seriously complicate patient and biomarker selections, toxicity readout, and efficacy of the drug candidates for clinical trials [125]. In contrast to the essential role of the intracellular Hsp90, especially Hsp90β for cell and organ homeostasis, the requirement of eHsp90α for life has not been reported. Instead, only when tissue homeostasis is broken, such as during wound healing or disease occurrence such as tumour growth, does eHsp90α then come into the

blocking eHsp90α by antibodies delayed wound healing [55]. Similarly, CRISPR-knockout of the Hsp90α gene selectively eliminated the ability of the MDA-MB-231 tumour cells

tumorigenicity of Hsp90α-KO tumour cells could be fully rescued by extracellular supplementation with hrHsp90α proteins in an ATPase-independent fashion [45]. Therefore,

picture. To support this notion, Bhatia and colleagues showed that selectively blocking eHsp90α by antibodies delayed wound healing [55]. Similarly, CRISPR-knockout of the Hsp90α gene selectively eliminated the ability of the MDA-MB-231 tumour cells to invade a Matrigel barrier and form tumours in mice. More remarkably, the defective tumorigenicity of Hsp90α-KO tumour cells could be fully rescued by extracellular supplementation with hrHsp90α proteins in an ATPase-independent fashion [45]. Therefore, in theory, drugs targeting eHsp90 should achieve higher efficacy and pose minimum toxicity to patients. A schematic representation of this simplified thought is depicted in Figure 6. *Biomolecules* **2022**, *12*, x 14 of 20 in theory, drugs targeting eHsp90 should achieve higher efficacy and pose minimum toxicity to patients. A schematic representation of this simplified thought is depicted in Figure 6.

**Figure 6.** A major difference between targeting eHsp90α and targeting intracellular Hsp90 in cancer. Cytotoxicity and lack of a clear therapeutic window under tolerable dosages have been the major hurdles for ATP-binding inhibitors of Hsp90, especially Hsp90β, in cancer clinical trials. In contrast, selectively targeting eHsp90α with membrane impermeable drug candidates has immerged as a new therapeutic strategy in cancer and beyond. **Figure 6.** A major difference between targeting eHsp90α and targeting intracellular Hsp90 in cancer. Cytotoxicity and lack of a clear therapeutic window under tolerable dosages have been the major hurdles for ATP-binding inhibitors of Hsp90, especially Hsp90β, in cancer clinical trials. In contrast, selectively targeting eHsp90α with membrane impermeable drug candidates has immerged as a new therapeutic strategy in cancer and beyond.

#### **10. Why Is eHsp90α Co-Opted for Extracellular Duties? 10. Why Is eHsp90**α **Co-Opted for Extracellular Duties?**

Our current understanding of this fundamental question remains little beyond speculation. An entry point to understand the question is the fact that Hsp90 maintains an unusually high expression level in almost all cell types. Although the statement of "1–2% Hsp90 of total cellular proteins" has been used for decades, this number did not come from direct experimental measurements, but rather from estimations. The first quantitation of the cellular Hsp90 protein was completed by Sahu and colleagues in 2012. Using classical biochemical techniques, these authors demonstrated that Hsp90 accounted for 2– 3% of the total cellular proteins among four normal cell lines and 3–7% of the total cellular proteins among four cancer cell lines tested [126]. More surprisingly, a recent study involving 12 (eight tumour and four normal) cell lines reported a much greater variation in the total cellular Hsp90 (α and β) expression, a range of 1.7% to 9% among non-cancer cell lines and different mouse organs and a range of 3 to 7% among the tumour cell lines [125]. If we take the general assumption that a given type of human cell expresses 1/3 of its total 30,000 protein-coding genes, the percentage of the Hsp90 expression is at least several hundred times higher than the rest of the 9999 cellular gene products. The question is why Our current understanding of this fundamental question remains little beyond speculation. An entry point to understand the question is the fact that Hsp90 maintains an unusually high expression level in almost all cell types. Although the statement of "1–2% Hsp90 of total cellular proteins" has been used for decades, this number did not come from direct experimental measurements, but rather from estimations. The first quantitation of the cellular Hsp90 protein was completed by Sahu and colleagues in 2012. Using classical biochemical techniques, these authors demonstrated that Hsp90 accounted for 2–3% of the total cellular proteins among four normal cell lines and 3–7% of the total cellular proteins among four cancer cell lines tested [126]. More surprisingly, a recent study involving 12 (eight tumour and four normal) cell lines reported a much greater variation in the total cellular Hsp90 (α and β) expression, a range of 1.7% to 9% among non-cancer cell lines and different mouse organs and a range of 3 to 7% among the tumour cell lines [125]. If we take the general assumption that a given type of human cell expresses 1/3 of its total 30,000 protein-coding genes, the percentage of the Hsp90 expression is at least several hundred times higher than the rest of the 9999 cellular gene products. The question is why

a particular gene product must be given such a spatial privilege. Evolution would not have tolerated such an abundant storage of a protein if functioning as an intracellular chaperone were its sole duty, as Csermely and colleagues have long argued [7]. We spec-

lar duty of chaperones, such as stabilization of HIF-1α [51], whereas the vast majority of

a particular gene product must be given such a spatial privilege. Evolution would not have tolerated such an abundant storage of a protein if functioning as an intracellular chaperone were its sole duty, as Csermely and colleagues have long argued [7]. We speculate that a smaller portion of Hsp90α is required to work with Hsp90β for the intracellular duty of chaperones, such as stabilization of HIF-1α [51], whereas the vast majority of eHsp90α is stockpiled for supply to tolerate environmental insults, such as tissue injuries, that take place all the time. The second possible answer is that eHsp90α provides unique properties that are absent from conventional extracellular factors such as cytokines, growth factors, or ECMs. Li's group showed that topical recombinant eHsp90α protein promoted normal wound healing far more effectively than the (only) FDA-approved growth factor therapy (RaranexTM, PDGF-BB). Their study showed that eHsp90α overrides the inhibitory effect of TGFβ family cytokines, which are abundantly present in fresh wounds. To the best of our knowledge, eHsp90α is the first molecule with this unprecedented property [46]. Third, an effective wound-healing agent is one that must recruit all three types of skin cells (epidermal, dermal, and endothelial) to close the wound. However, all growth factors show selectively targeted cell type(s). This limitation has made any single growth factor therapy less effective in the multi-cell process of wound healing. PDGF-BB only acts on dermal fibroblasts, but not epidermal keratinocytes and dermal microvascular endothelial cells, as the latter do not express either PDGFRα or PDGFRβ [46]. These findings may explain why Raranex has shown limited efficacy in clinic, even with several thousand times higher concentration of PDGF-BB (100 µg/g gel) than found in human circulation (0–15 ng/mL). In contrast, eHsp90α acts as a common pro-motility factor for all three types of skin cells involved in wound healing and shows a far stronger effect than PDGF-BB in wound healing [46,65,78]. For similar reasons, eHsp90α may also have an advantage over conventional extracellular factors in cancer invasion. For instance, Hanahan and Weinberg in their heavily cited review on cancer pointed out that one of the most recognized tumour-suppressing effects comes from the anti-growth signal by TGFβ [127]. To sabotage the inhibitory effect of TGFβ, only a small number of tumours choose to mutate either the type II (TβRII) or type I (TβRI) TGFβ receptor or their downstream effector, Smad4, which forms a complex with activated Smad2/3 to regulate gene expression. How the rest of human tumours bypass the TGFβ's inhibitory signals has never been discussed. We argue that these tumours secrete eHsp90α to override TGFβ inhibition.

#### **11. Conclusions and Perspective**

It has been the second decade since the official recognition of eHsp90α as a new research branch of Hsp90 in 2008. Since then, all-round progress, including mechanisms of secretion and action, biological function, therapeutic epitope identification, preclinical evaluation, and clinical relevance of eHsp90α, has been reported around the globe. If we have to provide a single outstanding take-home message to the readers, it would undoubtably be the exciting consensus that eHsp90α is not required for homeostasis but remains an essential player under pathological conditions and crisis. Therapeutically targeting eHsp90α in blood circulation represents a particularly exciting modality due to its ease-of-access, safety, and likely increased efficacy compared to targeting intracellular or nuclear Hsp90. For the next decade, the central challenge is to prove the clinical relevance of eHsp90α, such as in tissue injury, fibrosis, and tumorigenesis, and to concurrently establish the druggable window for targeting eHsp90α in human disorders for therapeutic development.

**Funding:** This study was supported by NIH grants GM066193 and GM067100 (to W.L.), NIH grant CA183119 (to D.J.) and Self-Topic Fund of Tsinghua University No. 20211080002 (to Y.L.).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Acknowledgments:** We thank many of our previous lab colleagues who made contributions to the work described in this review. We apologize if we failed to acknowledge every publication on eHsp90, while this review was being written.

**Conflicts of Interest:** The authors declare that they have no conflict of interest.

### **References**


## *Review* **Organismal Roles of Hsp90**

**Patricija van Oosten-Hawle**

Department of Biological Sciences, The University of North Carolina at Charlotte, Charlotte, NC 28223, USA; pvanoost@uncc.edu

**Abstract:** Heat shock protein 90 (Hsp90) is a highly conserved molecular chaperone that assists in the maturation of many client proteins involved in cellular signal transduction. As a regulator of cellular signaling processes, it is vital for the maintenance of cellular proteostasis and adaptation to environmental stresses. Emerging research shows that Hsp90 function in an organism goes well beyond intracellular proteostasis. In metazoans, Hsp90, as an environmentally responsive chaperone, is involved in inter-tissue stress signaling responses that coordinate and safeguard cell nonautonomous proteostasis and organismal health. In this way, Hsp90 has the capacity to influence evolution and aging, and effect behavioral responses to facilitate tissue-defense systems that ensure organismal survival. In this review, I summarize the literature on the organismal roles of Hsp90 uncovered in multicellular organisms, from plants to invertebrates and mammals.

**Keywords:** Hsp90; organismal; cell nonautonomous; proteostasis; stress response; inter-tissue stress signaling

### **1. Introduction**

Cellular protein homeostasis depends on the integrity and function of its proteome, of which molecular chaperones play an indispensable role to maintain it. Heat shock protein 90 (Hsp90) is an essential and evolutionary conserved molecular chaperone, that except for archea, is found in all kingdoms of life [1]. Hsp90 is crucial for the viability and growth of eukaryotic cells and organisms, and it is one of the most abundant cellular proteins known to date, representing ~2% of total protein in a cell [2]. This abundance of Hsp90 is required to sustain the wide range of cellular processes it is implicated in, by chaperoning components involved in cellular signal transduction events such as protein kinases and steroid hormone receptors [3–8].

Hsp90 requires ATP function for its activity to help facilitate folding of client proteins, and this function is regulated and controlled by a variety of co-chaperones in a contextspecific manner [7,9], as well as post translational modifications [10–13]. Briefly, each Hsp90 dimer binds ATP in its "open conformation", which is followed by dimerization of the N-terminal domains of each protomer, allowing ATP hydrolysis. This subsequently leads to release of the folded and mature client protein, resulting in the open conformation of the Hsp90 dimer [7,8,14]. Each step along the Hsp90 chaperone cycle is finetuned by co-chaperones, such as, e.g., Cdc37/p50, which promote binding to kinase clients [15–17], or Aha1 and p23 that are involved in the regulation of Hsp90 ATP hydrolysis [14,18,19]. For more detailed information on the Hsp90 structure and regulation by co-chaperones, the reader is referred to articles and reviews specifically discussing this topic, including this Special Issue [8,20–22].

Because of its involvement in a wide range of cellular processes, Hsp90 supports an expansive network of more than 300 confirmed client proteins encompassing at least 5% of total proteins in yeast cells, and this number is similarly increased in multicellular organisms and mammals (https://www.picard.ch/downloads/Hsp90interactors.pdf; accessed 20 December 2022). Consequently, it is perhaps not surprising that Hsp90 is involved in almost every cellular process from cell cycle and a multitude of signal transduction

**Citation:** van Oosten-Hawle, P. Organismal Roles of Hsp90. *Biomolecules* **2023**, *13*, 251. https://doi.org/10.3390/ biom13020251

Academic Editor: Chrisostomos Prodromou

Received: 20 December 2022 Revised: 24 January 2023 Accepted: 25 January 2023 Published: 29 January 2023

**Copyright:** © 2023 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

pathways to protein trafficking, transcriptional processes and genomic stability [23–29]. In multicellular organisms, Hsp90's involvement is expanded accordingly, and evidence in the past two decades has shown that Hsp90's role reaches far beyond maintenance of signaling proteins: Its involvement ranges from development and evolution to intercellular stress signaling, aging responses and innate immunity, as well as neuronal function and behavior (Figure 1). In this review, I will highlight these organismal roles of Hsp90 which have been uncovered in different multicellular model systems, including plants, invertebrates and mammals. [23–29]. In multicellular organisms, Hsp90's involvement is expanded accordingly, and evidence in the past two decades has shown that Hsp90's role reaches far beyond maintenance of signaling proteins: Its involvement ranges from development and evolution to intercellular stress signaling, aging responses and innate immunity, as well as neuronal function and behavior (Figure 1). In this review, I will highlight these organismal roles of Hsp90 which have been uncovered in different multicellular model systems, including plants, invertebrates and mammals.

involved in almost every cellular process from cell cycle and a multitude of signal transduction pathways to protein trafficking, transcriptional processes and genomic stability

*Biomolecules* **2021**, *11*, x FOR PEER REVIEW 2 of 14

**Figure 1.** Organismal roles of Hsp90 in different multicellular model systems. In metazoans, such as *C. elegans*, *Mus musculus*, *D. melanogaster*, *Danio rerio* and *A. thaliana*, Hsp90 acts in diverse biolog-**Figure 1.** Organismal roles of Hsp90 in different multicellular model systems. In metazoans, such as *C. elegans*, *Mus musculus*, *D. melanogaster*, *Danio rerio* and *A. thaliana*, Hsp90 acts in diverse biological processes to ensure organismal proteostasis.

#### ical processes to ensure organismal proteostasis. **2. Hsp90 in Organismal Development and Evolution**

#### **2. Hsp90 in Organismal Development and Evolution**  *2.1. Development*

*2.1. Development*  The coordination of cell proliferation and differentiation is crucial for proper development. Hsp90's central role in growth and development is profound, as its client proteins regulate almost all phases of the cell cycle. These include PI-3/AKT, NFkB and MAP kinase pathways, which drive progression through G1/S and G2/M checkpoints through transcriptional routes converging on Cyclin D, and Cyclin B and E [30]. Hsp90 also regulates various key cell cycle regulators directly, including Cdk1, Cdk2, Cdk4 and Cdk6 [31– 34]. Furthermore, check point kinases Wee-1 and Myt-1 depend on Hsp90 function [35– 37]. Later stages of mitosis and cytokinesis also depend on Hsp90 via mitotic regulators Survivin and Aurora B [38,39]. Because of Hsp90 influencing the cell cycle at multiple levels, either directly or indirectly, Hsp90 function is indispensable not only for organismal development but also for tumor cell progression, which was recognized early on through targeted inhibition of Hsp90 function using ansamycin inhibitors such as gel-The coordination of cell proliferation and differentiation is crucial for proper development. Hsp90's central role in growth and development is profound, as its client proteins regulate almost all phases of the cell cycle. These include PI-3/AKT, NFkB and MAP kinase pathways, which drive progression through G1/S and G2/M checkpoints through transcriptional routes converging on Cyclin D, and Cyclin B and E [30]. Hsp90 also regulates various key cell cycle regulators directly, including Cdk1, Cdk2, Cdk4 and Cdk6 [31–34]. Furthermore, check point kinases Wee-1 and Myt-1 depend on Hsp90 function [35–37]. Later stages of mitosis and cytokinesis also depend on Hsp90 via mitotic regulators Survivin and Aurora B [38,39]. Because of Hsp90 influencing the cell cycle at multiple levels, either directly or indirectly, Hsp90 function is indispensable not only for organismal development but also for tumor cell progression, which was recognized early on through targeted inhibition of Hsp90 function using ansamycin inhibitors such as geldanamycin and 17-AAG [2,40–42]. Due to Hsp90 being involved in signaling pathways promoting cancer cell progression, it has become an attractive and well-established therapeutic cancer target, with Hsp90 inhibitors being continually developed and reviewed in clinical trials in an ongoing basis [43].

danamycin and 17-AAG [2,40–42]. Due to Hsp90 being involved in signaling pathways promoting cancer cell progression, it has become an attractive and well-established therapeutic cancer target, with Hsp90 inhibitors being continually developed and reviewed in clinical trials in an ongoing basis [43]. Most eukaryotic systems have two different cytosolic Hsp90 isoforms, with the exception of *C. elegans* that has only one cytosolic isoform (HSP-90/DAF-21). In mammals and yeast, the stress-inducible Hsp90α is encoded by the gene HSP90AA1 in humans (HSP82 in yeast), and the constitutively expressed Hsp90β is encoded by HSP90AB1 in humans and HSC82 in yeast. Although both isoforms share extensive sequence identity, their cellular

Most eukaryotic systems have two different cytosolic Hsp90 isoforms, with the exception of *C. elegans* that has only one cytosolic isoform (HSP-90/DAF-21). In mammals and yeast, the stress-inducible Hsp90α is encoded by the gene HSP90AA1 in humans

humans and HSC82 in yeast. Although both isoforms share extensive sequence identity,

functions are not completely identical and this is also demonstrated in their developmental requirements in an organism. For example, the Hsp90β knockout mouse shows early embryonic lethality [44], whereas this is not the case for mice lacking Hsp90α, which are viable but exhibit a failure of spermatogenesis and become sterile [45]. Moreover, while both isoforms are mutually expressed in most tissues in the mouse, the heart and muscle were found to harbor reduced levels of Hsp90α compared to Hsp90β [45]. Interestingly, zebrafish contains two Hsp90α genes, called Hsp90a1 and Hsp90a2. Hsp90a1 is crucial for myofibril organization in skeletal muscle development, whereas Hsp90a2 has no effect on muscle development [46]. Coherent with observations in vertebrates, the only cytosolic Hsp90 isoform in the invertebrate *C. elegans* (HSP-90) is crucial for myosin folding and muscle development, as RNAi-mediated knockdown leads to disrupted myosin filaments and motility defects [47,48]. Mammals such as mice, however, appear to require higher threshold levels of Hsp90 to promote stress adaptation and survival of the organism compared to yeast. This is accomplished through an internal ribosome entry site (IRES) in the 5'UTR of the Hsp90ab1 mRNA that can reprogram Hsp90 translational levels in stressed conditions [49].

As observed in other multicellular model organisms, depletion of Hsp90 by RNAi in *C. elegans* leads to morphological and transcriptional changes, including developmental changes to the gonad, vulval structures and oocyte development [50]. Indeed, Hsp90 regulates the meiotic prophase to metaphase transition during oocyte development by ensuring *wee-1* kinase functionality, which results in reduced fertility in the worm [51]. Interestingly, besides transcriptional changes that demonstrate induction of the heat shock response, Hsp90 RNAi at the whole animal level also leads to induction of an innate immune response, by altering expression levels of innate immune genes primarily expressed in the intestine [50].

The role of Hsp90 in *C. elegans* development is further highlighted through its involvement in *dauer* formation, TGFβ- and Notch signaling. The *C. elegans* TGF-β pathway regulates a decision between reproductive development and arrest at a larval stage known as *dauer* that is suited for survival under conditions of environmental stress. Hsp90 itself is implicated in *dauer* formation through its interaction with two components of the TGF-β pathway, TGF-β-RI (DAF-1 in *C. elegans*) and TGF-β-RII (DAF-4 in *C. elegans*) [52]. Hsp90 also regulates the functionality of a DAF-11/guanyl cyclase signaling pathway in sensory ciliae and amphid neurons that controls *dauer* formation in response to environmental cues, in parallel to the TGF-β pathway [53,54], as well as chemosensory behaviors [54,55]. Germline proliferation in *C. elegans* requires signaling from the somatic gonad to the germline, which is mediated by GLP-1 (a Notch orthologue) [56]. If GLP-1/Notch signaling is defective through mutations in the *glp-1* gene, germline stem cells prematurely exit mitosis and enter meiosis to form gametes, resulting in reduced germline proliferation and sterility [57]. Hsp90 has been identified as a regulator of Notch signaling that suppresses defective GLP-1/Notch signaling and promotes germline proliferation [58]. Strikingly, *C. elegans* depleted for Hsp90 by RNA interference or using an HSP-90(I461N) mutant leads to the formation of a proximal germline tumor, despite its reduced function and reduced GLP-1 signaling [57,58]. Further research in solving this paradox will be required to better illuminate the complex tissue-specific and organismal functions of the Hsp90 chaperone system in *C. elegans* and other metazoans.

#### *2.2. Hsp90 as a Capacitor of Organismal Evolution*

The physiological requirement of Hsp90 for the growth and development of model organisms was obvious early on using *Saccharomyces cerevisiae* [35,59], but was further highlighted using *Drosophila melanogaster* [60], where point mutations of the *Drosophila* Hsp90 (Hsp83) gene are lethal as homozygotes [61]. Although heterozygous mutant combinations are viable as adults, they are associated with sterility due to defects in microtubule dynamics during spermatogenesis [61]. Further experiments in *Drosophila* designed to identify suppressors of signal transduction *Sevenless* and *Raf* pathway mutants recovered Hsp90 mutants. Progeny of these Hsp90 mutants resulted in developmental abnormalities that, dependent on the genetic background, affected different morphological structures of the fruit fly [60,62]. This discovery led to further demonstration of Hsp90's importance in *Drosophila* spermatogenesis and germline development [61]. It was one of the cornerstones that defined Hsp90's prominent role in evolution, which was established as the "Hsp90 capacitor hypothesis" by the Lindquist lab [63].

The capacitor hypothesis demonstrated that diverse pathways become sensitive to the effects of genetic variation when Hsp90 function is compromised due to environmental stress, pharmacological inhibition or genetic mutation. It showed that Hsp90 functions in a wide variety of morphogenetic processes that are apparent in all model organisms tested, from yeast to vertebrates [63–65]. For example, the diverse phenotypes associated with Hsp90 impairment in *Drosophila* are deformed eye and thickened wing phenotypes [63], whereas in *Arabidopsis thaliana*, this leads to altered leaf and cotyledon shapes [64]. Similar consequences were observed in zebrafish upon reduced Hsp90 expression [66]. In *C. elegans*, the expression level of Hsp90 in particular varies during *C. elegans* embryonic development, causing embryos with stronger induction of Hsp90 to be less affected by mutation, thus buffering genetic variation [67]. However, individuals of a population with increased stress resistance due to higher Hsp90 levels show a "trade-off" with lowered reproductive potential, whereas worms with lower stress resistance are associated with higher reproductive fitness. The reason for this is thought to be a bet-hedging strategy, which is beneficial in ever-changing environments, that ensures survival of the population as a whole [67,68]. The reduced reproductive fitness due to increased Hsp90 expression perhaps highlights the requirement for Hsp90 expression levels to be tightly regulated due to its important role in germline development [57].

Importantly, phenotypic traits revealed upon temporary Hsp90 impairment can be selected for over several generations and become fixed in following generations, establishing Hsp90's crucial role in evolution. An example for this in a natural setting was provided by the cavefish *Asyanax mexicanus*, where cryptic variation in eye size was masked by Hsp90 in the ancestral river but revealed when fish were kept in caves that challenged the Hsp90 system due to low-salinity water [65]. This even plays a role in human disease, as is the case in Fanconi anemia (FA), a complex autosomal recessive human cancer predisposition syndrome that results in point mutations of 19 genes involved in the FA genome maintenance pathway [69]. The function of less severe FANCA mutants was preserved by Hsp90 binding, which maintained FA pathway function but became destabilized and sensitive to genotoxic stress upon Hsp90 impairment [69].

However, while the evolutionary capacitor hypothesis relies on the potential of cytosolic protein instability that can be exposed upon Hsp90 inhibition, other contributions were shown to underlie Hsp90-dependent transcriptional mechanisms and chromatin structure [25–28,70,71]. While these are seemingly different mechanisms leading to the variety of Hsp90 buffered traits, it is perhaps a combination of multiple Hsp90-dependent genetic as well as epigenetic mechanisms working in concert.

### **3. Hsp90-Dependent Regulation of Organismal Proteostasis, Stress and Aging**

#### *3.1. Hsp90 in the Regulation of Cell Nonautonomous Stress Signaling*

Hsp90, together with its co-chaperone machinery, is an integral part of the cellular network that safeguards proteostasis. As with other chaperones, Hsp90 expression is regulated by the stress transcription factor Heat Shock Factor 1 (HSF1) and is increased in response to environmental challenges that initiate the cytosolic heat shock response (HSR) [72]. This is accomplished in a negative feedback mechanism, whereby under normal conditions, HSF1 is sequestered by a multichaperone complex including Hsp90 and Hsp70 in an inactive monomeric form [73]. Proteotoxic stress conditions that increase the amount of misfolded proteins in the cell recruit the chaperones away from HSF1 towards the accumulating pool of misfolded proteins and releases HSF1 monomers. This in turn allows HSF1 monomers to form homotrimers that translocate to the nucleus, where they bind

to heat shock elements (HSE) that induce molecular chaperones (heat shock proteins), as well as trafficking and proteolytic genes, in order to restore cytosolic proteostasis. Once the levels of Hsp90 and other chaperones have sufficiently increased in the cytosol to refold damaged proteins, they are recruited back to HSF1.

However, expression of Hsp90 can also, directly or indirectly, be regulated by other transcription factors in addition to HSF-1. For example in *C. elegans*, the GATA transcription factor PQM-1 responds to local changes in Hsp90 expression levels as a mediator of transcellular chaperone signaling, but also regulates Hsp90 expression itself [74]. In addition to HSF1, Hsp90 also regulates the function of the FOXO orthologue DAF-16 isoform A by facilitating its translocation into the nucleus upon heat stress and reduced ILS [75].

In metazoans, the stress-dependent induction of HSF-1 transcriptional activity also depends on intercellular stress signaling responses. In *C. elegans*, temperature alterations are sensed by two thermosensory AFD neurons that control temperature-dependent behaviors. This is accomplished through the action of the guanylyl cyclase GCY-8 that is specifically expressed in AFD neurons, and which controls HSF1-dependent induction of the HSR in distal cells in order to restore proteostasis at the organismal level. Neuronal control of proteostasis in response to acute temperature challenges is, however, uncoupled from aging-related responses via a GPCR thermal receptor GTR-1 expressed in chemosensory neurons [76]. The *C. elegans* nervous system relays the signal to distal organs via the neurotransmitter serotonin, thus involving serotonergic neurocircuitry [77,78]. However, astrocyte-like cells in the nervous system can also regulate the cell-nonautonomous HSR in an HSF-1 dependent manner that does not rely on known neurotransmitters but instead requires small clear vesicle release [79]. Non-neuronal tissues such as muscle and gut cells can equally relay information of temperature changes to thermosensory AFD neurons via estrogen signaling through the nuclear hormone receptor NHR-69 [80], an orthologue of the human HNF4 transcription factors that are clients of Hsp90 [81].

Interestingly, heat shock leads to rapid induction of HSF1 activity in the *C. elegans* germline [77,82,83] and HSF1 is required for gametogenesis in invertebrates and vertebrates [72,84]. Like Hsp90, HSF1 is required for germline proliferation and fecundity, relying on Insulin/IGF-1 signaling in the soma that nonautonomously activates HSF-1 in the germline [85], although whether Hsp90 is involved in this regulation is currently unknown.

However, Hsp90 is itself involved in relaying signals from one tissue to another, particularly when its expression levels are altered in the gut or the nervous system, an organismal stress signaling response known as Transcellular Chaperone Signaling (Figure 2) [86,87]. Enhancement of Hsp90 capacity in the gut or the neurons leads to a compensatory transcriptional inter-tissue response, regulated via the transcription factor PQM-1, that induces Hsp90 expression in other distal cell types and primarily muscle cells [74,88]. This protects against the age-associated debilitating consequences of misfolded proteins expressed in muscle cells, including human amyloid beta protein or endogenously expressed metastable myosin [74,88]. How this transcriptional response is relayed from one tissue to another, however, depends on tissue context. Transcellular chaperone signaling from neurons to the muscle requires glutamatergic signaling and relies on the c-type lectin *clec-41* that associates with AMPA receptor in glutamatergic neurons (Figure 2A) [74]. Increased Hsp90 expression in the gut is relayed via the secreted immune peptide *asp-12* which leads to transcriptional upregulation of Hsp90 in muscle cells (Figure 2A) [74].

teostasis and stress resilience via Transcellular Chaperone Signaling.

membrane-associated guanylate cyclase that relays the signal received from the intestine to the muscle cell nucleus where the homeodomain transcription factor CEH-58 induces Hsp70 expression (Figure 2B) [87]. Thus, there is a difference in intercellular-signaling components which depend on the tissue-type perceiving altered Hsp90 expression levels. This argues for multiple and complex layers of responses that cannot be answered by one particular molecular mechanism, at least not in a multicellular organismal setting [87]. This demonstrates that in metazoans, local Hsp90 capacity can regulate organismal pro-

**Figure 2.** Transcellular chaperone signaling pathways. (**A**) Overexpression of Hsp90 in the nervous system mediates upregulation of Hsp90 in muscle cells via PQM-1, CLEC-41 and glutamatergic neurotransmission. Overexpression of Hsp90 in the intestine relays the signal to upregulate Hsp90 in muscle cells via PQM-1 and ASP-12. The transcription factor regulating Hsp90 in muscle cells in response to TCS has not been determined. The organismal consequences are increased health span and reduced protein aggregation in the muscle tissue. (**B**) Knockdown of Hsp90 in the intestine relays the signal to muscle cells via the secreted lipases TXT-4 and TXT-8. There, TXT-1 signals to the transcription factor CEH-58 to induce Hsp70 expression, resulting in increased longevity and stress resilience. HSF-1 functions as a suppressor of this process. Comparable organismal effects as a result of local induction of the HSF-1 mediated **Figure 2.** Transcellular chaperone signaling pathways. (**A**) Overexpression of Hsp90 in the nervous system mediates upregulation of Hsp90 in muscle cells via PQM-1, CLEC-41 and glutamatergic neurotransmission. Overexpression of Hsp90 in the intestine relays the signal to upregulate Hsp90 in muscle cells via PQM-1 and ASP-12. The transcription factor regulating Hsp90 in muscle cells in response to TCS has not been determined. The organismal consequences are increased health span and reduced protein aggregation in the muscle tissue. (**B**) Knockdown of Hsp90 in the intestine relays the signal to muscle cells via the secreted lipases TXT-4 and TXT-8. There, TXT-1 signals to the transcription factor CEH-58 to induce Hsp70 expression, resulting in increased longevity and stress resilience. HSF-1 functions as a suppressor of this process.

HSR is also observed in mammals via neuroendocrine signaling. For example, rats undergoing restraint stress have higher cortisol levels secreted by the pituitary gland which signals to activate HSF-1 in the adrenal glands in the kidney to induce Hsp70 expression [89], although how Hsp90 itself could potentially be involved in this response is currently not known and will require further research. One question that often arises is whether Hsp90 is secreted as part of inter-tissue stress signaling in an organism. Hsp90 secretion has been observed in tissue-culture in response to a variety of stress conditions as well as in cancer cells [90]. Clinically, skin injury promotes Hsp90α secretion and potentiates wound healing in tissue-culture, pigs and dogs [91,92]. However, secreted, extracellular Hsp90 has not been observed as a signaling component itself involved in inter-tissue stress signaling in an organism. In fact, secretion of Hsp90 was not detected in *C. elegans* overexpressing Hsp90 in different cell types [88]. For a more detailed review on the roles of secreted, extracellular Hsp90, the reader is referred to reviews by Li and colleagues [93] in this Special Issue on Hsp90. On the other hand, when Hsp90 levels are reduced by tissue-specific RNA interference in the gut, a compensatory signaling mechanism elevates Hsp70 expression in distant cells (Figure 2B). This is, however, not mediated by a mechanism that relies on HSF1 to activate a canonical HSR, but depends on a homeodomain transcription factor, CEH-58. HSF1 transcriptional activity is suppressed upon gut-specific Hsp90 depletion, and induction of Hsp70 relies on a different intercellular signaling cue involving TXT-1, a membrane-associated guanylate cyclase that relays the signal received from the intestine to the muscle cell nucleus where the homeodomain transcription factor CEH-58 induces Hsp70 expression (Figure 2B) [87]. Thus, there is a difference in intercellular-signaling components which depend on the tissue-type perceiving altered Hsp90 expression levels. This argues for multiple and complex layers of responses that cannot be answered by one particular molecular mechanism, at least not in a multicellular organismal setting [87]. This demonstrates that in metazoans, local Hsp90 capacity can regulate organismal proteostasis and stress resilience via Transcellular Chaperone Signaling.

*3.2. Hsp90-Dependent Regulation of Lifespan and Aging*  Comparable organismal effects as a result of local induction of the HSF-1 mediated HSR is also observed in mammals via neuroendocrine signaling. For example, rats undergoing restraint stress have higher cortisol levels secreted by the pituitary gland which signals to activate HSF-1 in the adrenal glands in the kidney to induce Hsp70 expression [89], although how Hsp90 itself could potentially be involved in this response is currently not known and will require further research.

One question that often arises is whether Hsp90 is secreted as part of inter-tissue stress signaling in an organism. Hsp90 secretion has been observed in tissue-culture in response to a variety of stress conditions as well as in cancer cells [90]. Clinically, skin injury promotes Hsp90α secretion and potentiates wound healing in tissue-culture, pigs and dogs [91,92]. However, secreted, extracellular Hsp90 has not been observed as a signaling component itself involved in inter-tissue stress signaling in an organism. In fact, secretion of Hsp90 was not detected in *C. elegans* overexpressing Hsp90 in different cell types [88]. For a more detailed review on the roles of secreted, extracellular Hsp90, the reader is referred to reviews by Li and colleagues [93] in this Special Issue on Hsp90.

#### *3.2. Hsp90-Dependent Regulation of Lifespan and Aging*

Consistent with a growth-promoting role, substantial depletion of Hsp90 by RNAimediated knockdown can lead to growth defects and larval arrest, and even shorten lifespan [50,75]. The developmental defects associated with Hsp90 RNAi are morphological changes to the gonad and vulva, induction of the HSR and changes to the muscle ultrastructure [50]. Importantly, however, mild impairment of Hsp90 either by RNAi or pharmacological inhibition leads to lifespan extension and enhances health span [94]. This was shown in a pharmacological geroprotector screen using *C. elegans* that identified two Hsp90 inhibitors, Tanespimycin and Monorden, that extended lifespan and improved health of the nematode throughout the course of aging [94]. The study found that both inhibitors acted through HSF1 to induce the age-defying and health span-inducing effects in the worm. This is consistent with HSF1's role in promoting longevity [95–97]. Similar to mild Hsp90 impairment by inhibitors, moderate depletion of Hsp90 RNAi in the gut also enhances lifespan and stress resilience in *C. elegans* without any developmental issues [87]. Similar observations were made in vertebrates, where transient knockdown of Hsp90 during embryonic development in zebrafish results in cold stress resistance in adult animals [98]. Interestingly, the Hs90 co-chaperone p23 acts in key longevity pathways to regulate lifespan in a temperature-dependent manner [99]. At elevated temperatures, p23 mutation extends lifespan through DAF-16 and HSF1 signaling pathways. Short-lived phenotypes depend on the DAF-12 steroid receptor signaling pathway [99], with DAF-12 being a type II nuclear receptor that resembles the human thyroid receptor and is a known client of the p23-HSP90 complex [100]. Apart from being involved in the key longevity pathways, ILS/IGF-1 signaling and HSF-1 signaling pathways, Hsp90 is also involved in the regulation of SIRT1 in both *C. elegans* and mammalian cells [101]. Thus, Hsp90 is unique, as it is a major facilitator that ensures the efficacy of all signaling processes maintaining organismal health and promoting survival.

#### **4. Pathogen Response and Innate Immunity**

The involvement of Hsp90 in immune responses is manifold, as it is implicated in the adaptive as well as innate immunity pathways in almost all organisms. In plants, R proteins are client proteins of Hsp90, which is important for the defense response against microbial pathogens [102,103]. The activation of R proteins results in local cell death to limit pathogen proliferation. Because of this, R protein activation also needs to be tightly controlled to avoid tissue damage, which is regulated by Hsp90 [102,103].

In the invertebrate *C. elegans*, which does not have an adaptive immune response, Hsp90 plays an important role in the innate immune response via HSF1. For example, mutant Hsp90, as well as heat shock, causes release of HSF1 from Hsp90, resulting in HSF1 initiating expression of antimicrobial peptide genes [104,105]. Coherently, depletion of Hsp90 by RNA interference also induces an innate immune transcriptional response that was proposed to be similar to the immune response after *C. elegans* exposure to *Pseudomona aeruginosa* [50]. Similarly, pathogen-infected wax moths treated with Hsp90 inhibitor 17-DMAG were protected by an increased immune response [106]. This breadth of Hsp90 being implicated in a process that mediates innate immunity via HSF1 activation demonstrates the importance and conservation of Hsp90 in the innate immune response. In mammals, Hsp90 is implicated in the presentation of antigen to T-cells and activation of macrophages [107]. Hsp90 mediates antigen presentation in target antigen-presenting cells (APC) by facilitating endocytosis of bound polypeptides [108,109]. These generated antigenic peptides are presented to MHC-I/II by Hsp90 [110]. Extracellular Hsp90 can also bind to peptide antigens to facilitate uptake of the Hsp90 antigen complex by endocytosis [107]. After the antigen is internalized, intracellular Hsp90 facilitates further

processing of these peptides to the proteasome for degradation [107]. Interestingly, Hsp90 also regulates the reactivation of the human immunodeficiency virus (HIV-1) via regulation of the PKC/ERK MAPK pathways, which influences replication and gene expression of the virus [111]. In the response to pathogens, extracellular Hsp90 can act as a damageassociated molecular pattern (DAMP) signal that regulates the production of cytokines in response to pathogenic infection and inflammation [112]. This involvement of p38 and ERK MAPK pathways in response to pathogens was shown to require Hsp90 for their function through direct interaction of Hsp90 with MAP kinases p38 and ERK in evolutionary diverse organisms [16,113,114]. Hsp90 also plays an important role in growth, development and virulence of parasitic pathogens itself, such as the parasitic protozoa *Plasmodium falciparum* [115] and *Toxoplasma gondii* [116,117]. This makes Hsp90 a high-value drug target to inhibit the parasite's growth and infection cycle in humans [117]. In summary, the role of Hsp90 in the adaptive and innate immune response is vast, and the reader is referred to specialized reviews on this topic for more detailed information (e.g., [110,118]).

#### **5. Neuronal Signaling and Behavior**

Considering the wide range of client proteins dependent on Hsp90 function, it is perhaps unsurprising, but nevertheless fascinating, to find it involved in neuronal signaling and function. Some of the first experimental evidence demonstrating a role for Hsp90 in neuronal function stems from research in *C. elegans*. Hsp90 is crucial for the function of specific chemosensory amphid neurons required to sense pheromones and other attractants. It was proposed that Hsp90 accomplishes this through interaction and stabilization of the transmembrane guanylyl cyclase DAF-11, which regulates cGMP levels, a prominent second messenger in *C. elegans* chemosensory transduction [54,55].

In mice, Hsp90 is required for the constitutive trafficking of glutamatergic AMPA-type receptors into synapses during their continuous cycling between synaptic and non-synaptic sites, as well as efficient neurotransmitter release at the presynaptic terminal [119]. In addition to its role in neuronal signaling, Hsp90 chaperones the pro-regenerative dual leucine zipper kinase (DLK), a critical neuronal sensor that drives axon regeneration, degeneration and neurological disease in Drosophila and mammalian neurons [120]. This suggests a vital role for Hsp90 in axon injury signaling, as well as neuronal function that is evolutionary conserved in both vertebrates and invertebrates.

With this importance for neuronal signaling, is it possible that Hsp90 could be involved in the regulation of behavioral responses that facilitate survival during stress conditions? There is at least one example in *C. elegans* that provides direct evidence supporting such a role. Exposure of nematodes to high concentrations of volatile compounds, such as benzaldehyde and diacetyl, induces toxicity and food avoidance behavior [121]. However, preconditioning with benzaldehyde activates stress responses mediated via DAF-16, SKN-1 and HSP-90 in non-neuronal cells that confer increased stress resilience and behavioral tolerance [121]. Another example is provided by the heat stress-induced activation of HSF1, which regulates behavioral responses through estrogen signaling from non-neuronal cells to thermosensory neurons [80]. *hsf-1* mutants are defective in their thermotactic response towards temperature, i.e., migration towards cultivation temperature. Expression of wild type HSF-1 in muscle or intestinal cells rescued this behavioral defect via activation of the NHR-69 nuclear hormone receptor involved in estrogen-like signaling [80], which is a client of Hsp90, as mentioned earlier. Thus, Hsp90, through its role in multiple stress-responsive signaling pathways, may influence behavioral outputs in order to promote survival during environmental stress conditions.

#### **6. Outlook and Conclusions**

As a chaperone safeguarding the functionality of clients involved in almost every cellular signaling process, Hsp90 is essential for cellular homeostasis. At the organismal level, intercellular signaling processes that require the involvement of Hsp90 may be underlying the organismal coordination of extra- and intracellular signaling networks

between and across different tissues and organs. Especially at the organismal level, many open questions remain to fully comprehend the organismal biology of Hsp90, particularly with regard to intercellular stress signaling.

For example, (1) is there is a tissue map or tissue hierarchy allowing highly coordinated signaling responses to occur? We know that stress signaling can be regulated via both the nervous system and non-neuronal cell types, with, e.g., muscle and gut cells transmitting feedback information to the nervous system or even suppressing stress responses in different cell types and organs. (2) What is the tissue-specific Hsp90 interactome in an organism and how are potential Hsp90 interactors of these tissue-specific networks contributing to intercellular stress signaling? (3) Is there a role of extracellular Hsp90 in intercellular signaling processes? (4) If the tissue-specific expression levels of Hsp90 can affect stress responses in distant tissues, is there a naturally occurring/physiological condition that alters Hsp90 expression levels to induce transcellular chaperone signaling? (5) As Hsp90 function is tightly regulated by co-chaperones and post translational modifications [13], we currently do not know how co-chaperones of the Hsp90 machinery and its PTMs are involved in organismal proteostasis. For example, it can be envisioned that stress responses and intercellular stress signaling pathways are similarly influenced and perhaps finetuned through tissue-specific PTMs and co-chaperone networks. (6) How do the organismal roles of Hsp90 affect diseases, including neurodegenerative diseases and cancer, in a tissue- and disease-specific context?

Thus, the involvement of Hsp90 in almost all aspects of organismal biology, from development to aging, stress adaptation, evolution and different diseases including cancer and neurodegenerative diseases, places it at the nexus of a plethora of cell nonautonomous signaling processes. The challenge for future research will be to navigate through these inter-tissue signaling pathways in a comprehensive manner to understand their increased complexity in the multicellular setting of an organism.

**Funding:** This work was supported by laboratory start-up funds to P.v.O.-H. from the University of North Carolina at Charlotte.

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

**Acknowledgments:** The Figure was created using Biorender.com.

**Conflicts of Interest:** The author declares no conflict of interest.

#### **References**


**Disclaimer/Publisher's Note:** The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
