Clusterin, Other Extracellular Chaperones, and Eye Disease

Abstract

Proteostasis refers to all the processes that maintain the correct expression level, location, folding and turnover of proteins, essential to organismal survival. Both inside cells and in body fluids, molecular chaperones play key roles in maintaining proteostasis. In this article, we focus on clusterin, the first-recognized extracellular mammalian chaperone, and its role in diseases of the eye. Clusterin binds to and inhibits the aggregation of proteins that are misfolded due to mutations or stresses, clears these aggregating proteins from extracellular spaces, and facilitates their degradation. Clusterin exhibits three main homeostatic activities: proteostasis, cytoprotection, and anti-inflammation. The so-called “protein misfolding diseases” are caused by aggregation of misfolded proteins that accumulate pathologically as deposits in tissues; we discuss several such diseases that occur in the eye. Clusterin is typically found in these deposits, which is interpreted to mean that its capacity as a molecular chaperone to maintain proteostasis is overwhelmed in the disease state. Nevertheless, the role of clusterin in diseases involving such deposits needs to be better defined before therapeutic approaches can be entertained. A more straightforward case can be made for therapeutic use of clusterin based on its proteostatic role as a proteinase inhibitor, as well as its cytoprotective and anti-inflammatory properties. It is likely that clusterin works together in this way with other extracellular chaperones to protect the eye from disease, and we discuss several examples. We end this article by predicting future steps that may lead to development of clusterin as a biological drug.

1. Introduction

The term proteostasis refers to the maintenance of the proteome in a conformation, concentration and location that is required for their correct function and is critical for the maintenance of organismal viability (Balch, Morimoto et al. 2008). These processes operate in both the intracellular and extracellular environments, but the better characterized systems relate to the intracellular environment which has been the focus of decades of research. Despite the observation that proteins aggregate when their concentration exceeds solubility (Chiti and Dobson 2006), many proteins normally function at the upper limit of their solubility (Tartaglia, Pechmann et al. 2007). Consequently, small changes in protein concentration or solubility (due to mutations or a change in the environment) can tip the delicate balance leading to aggregation and deposition.

A significant recent study used RNAi to target genes that encode the secreted proteome in the nematode worm Caenorhabditis elegans and discovered 57 regulators of extracellular protein aggregation (Gallotta, Sandhu et al. 2020). This study went on to show that C. elegans responded to treatment with the Bacillus thuringiensis pore-forming protein Cry5B, mimicking a pathogen attack, by increasing the expression of components of extracellular proteostasis and thereby limiting the in vivo aggregation of extracellular proteins. The overexpression of components of extracellular proteostasis delayed aging and rendered worms resistant to Cry5B. Based on these results, the authors proposed that enhanced extracellular proteostasis contributes to systemic host defence by maintaining a functional secreted proteome and reducing proteotoxicity (Gallotta, Sandhu et al. 2020). Two important implications of this study are firstly that the correct functioning of extracellular proteostasis has important life or death consequences for organisms, and secondly that these protective mechanisms can involve multiple protein elements working together.

Molecular chaperones are proteins that assist the conformational folding or unfolding and the assembly or disassembly of other macromolecular structures. They have emerged as key players in proteostasis systems, where they perform a variety of roles including inhibiting protein aggregation, maintaining the solubility of and refolding misfolded proteins, and protein trafficking. The tendency to aggregate increases as proteins misfold in response to stress, and the many chaperones are specifically involved in maintaining the solubility of proteins subjected to stress. Recent advances suggest that an age-related decline in proteostasis capacity allows the manifestation of various protein-aggregation diseases, including Alzheimer's disease and Parkinson's disease (Hartl, Bracher et al. 2011). The pathologies of the so-called “protein misfolding diseases” are associated with the aggregation of misfolded proteins.

Knowledge of extracellular chaperones has lagged well behind that of their intracellular counterparts (Wyatt, Yerbury et al. 2013). Nevertheless, in recent years it has become clear that there is a growing family of abundant proteins in the extracellular fluids of animals that are able to bind to and inhibit the aggregation of proteins that are misfolded due to mutations or stresses. Furthermore, the extracellular chaperones are strongly implicated in clearing these aggregating proteins from extracellular spaces and facilitating their degradation (Dabbs, Wyatt et al. 2013, Afshari, Igo et al. 2017).

The major focus of this article is on CLU¹ (clusterin), which was the first extracellular mammalian chaperone identified, discovered in the laboratory of one of the senior co-authors (Wilson) (Humphreys, Carver et al. 1999). CLU continues to be the best studied of the extracellular chaperones to date. The protein is nearly ubiquitously expressed in tissues and is present in most biological fluids (Wyatt, Yerbury et al. 2013). CLU was first described in a 1983 publication as a secreted glycoprotein present in ram rete testis fluid that enhanced aggregation (‘clustering’) of cells in vitro (Blaschuk, Burdzy et al. 1983, Fritz, Burdzy et al. 1983). The protein was subsequently studied independently in numerous other biological and disease contexts, and as a result it acquired different names. Based on DNA sequencing data, it is now known that CLU is identical to SP-40,40 (serum protein 40,40) found in the soluble membrane attack complex of complement, as well as in immune deposits in glomerulonephritis (Murphy, Kirszbaum et al. 1988, Tsuruta, Wong et al. 1990). Other aliases for CLU include Apolipoprotein J (ApoJ), a protein associated with lipoprotein particles in human serum (de Silva, Stuart et al. 1990, James, Hochstrasser et al. 1991), sulfated glycoprotein-2 (SGP-2), the major secreted product of rat Sertoli cells (Tsuruta, Wong et al. 1990), and the protein translated from testosterone-repressed prostate message-2 (TRPM-2), which is upregulated in the regressing rat ventral prostate (Leger, Montpetit et al. 1987). At the inaugural International Workshop on Clusterin held in Cambridge, England in 1992, participants agreed to the name “clusterin” (HUGO nomenclature: CLU), acknowledging the original reports of its identification (Wilson and Easterbrook-Smith 2000).

Much of the early work on CLU focused on documenting changes in its expression associated with a variety of diseases and stress conditions (e.g. cancer, retinal degeneration and apoptosis). Based on expression patterns, a plethora of proposed functions was put forward, many based on limited evidence. In some cases, this led to contradictory paradigms - for example, increased CLU expression associated with different models of apoptosis was variously ascribed to a pro- or an anti-apoptotic function [e.g., (Jones, Meerabux et al. 1992, Hsu, Yang et al. 2000) (Viard, Wehrli et al. 1999)]. After extensive earlier work testing CLU for a variety of potential biological functions, the Wilson team reported their discovery of the small heat shock protein-like chaperone activity of CLU, making it the first known secreted mammalian chaperone abundant in human body fluids (Humphreys, Carver et al. 1999).

CLU expression in the eye was first described in 1992 in association with retinal degenerative disease (Jones, Meerabux et al. 1992). It is now known that CLU is expressed in essentially all healthy eye tissues, and CLU protein accumulates in eye fluids. We previously reviewed knowledge about CLU relevant to the eye and eye disease from an historical perspective, and discussed in detail new findings from our team about the role of CLU in ocular surface disease due to dry eye (Fini, Bauskar et al. 2016). More recently, we expanded upon our studies on CLU in dry eye (Fini, Jeong et al. 2020). Here we have reframed and updated the basic science of CLU from the original article and we discuss CLUs potential value as a biotherapeutic for several different eye diseases. We conclude by introducing other extracellular chaperones that might work together with CLU in the eye.

When we commenced writing this article, we performed a search of Scopus using the term “clusterin” and identified more than 4,500 articles. We refer the reader to the numerous excellent review and perspective articles on the biochemistry of CLU, a selection of which are listed here (Jenne and Tschopp 1992, May and Finch 1992, Rosenberg, Dvergsten et al. 1993, Rosenberg and Silkensen 1995, Wilson and Easterbrook-Smith 2000, Jones and Jomary 2002, Trougakos and Gonos 2002, Koltai 2014, Yerbury, Ooi et al. 2016); (Satapathy and Wilson 2021).

2. Clusterin Gene and Protein Product

In humans, a single CLU gene of nine exons is located on chromosome 8. The sequence is highly conserved across species, showing 70–80% identity at the amino acid level amongst mammals (Jones and Jomary 2002). Transcription results in an mRNA of ~ 2 kb, which encodes a primary polypeptide chain of 449 amino acids.

Figure 1 is a schematic of the CLU molecule based on information deduced from sequence analysis and biochemical studies. An N-terminal signal peptide of 22 amino acids is removed in the endoplasmic reticulum to produce a protein with a predicted mass of ~ 50 kDa. Subsequently, CLU is proteolytically cleaved in the trans-Golgi by a furin-like protease to form two anti-parallel polypeptide chains of similar size connected at a central core by 5 disulfide bonds (Rohne, Prochnow et al. 2014). Six predicted N-linked glycosylation sites clustered around the disulfide-bonded core were confirmed by mass spectroscopy (Kapron, Hilliard et al. 1997). This results in a secreted glycoprotein with an apparent mass of 75–80 kDa by sodium dodecyl sulfate-polyacrylamide electrophoresis (SDS-PAGE), although the actual mass is ~ 58–63 kDa, of which 17–27% consists of carbohydrate (Kapron, Hilliard et al. 1997).

Figure 1. Predicted Human Clusterin Protein.

The 22-mer secretory signal peptide is proteolytically cleaved from the 449-amino acid precursor polypeptide chain and subsequently the chain is cleaved again between residues Arg227-Ser228 to generate an alpha-chain and a beta-chain. These are assembled in anti-parallel fashion to generate a heterodimeric molecule in which the cysteine-rich centers (red boxes) are linked by five disulfide bonds (black rectangles) and flanked by five predicted amphipathic alpha-helices (yellow boxes) as in (Kapron, Hilliard et al. 1997). The seven motifs for N-linked glycosylation are indicated (white spots). Amino acid numbering for the N- and C-termini, and the predicted sites for N-linked glycosylation are indicated. Schematic adapted from (Wilson and Easterbrook-Smith 2000, Bailey, Dunker et al. 2001, Jones and Jomary 2002, Fini, Bauskar et al. 2016).

Other N-terminally truncated CLU protein isoforms have been proposed, including one thought to localize to the nucleus (e.g. (Leskov, Klokov et al. 2003, Trougakos, Djeu et al. 2009)), however, the mRNA transcripts encoding these alternative isoforms do not collectively exceed 0.34% of the total CLU transcript pool even under stress conditions (Prochnow, Gollan et al. 2013). Furthermore, two independent reports have demonstrated that, during endoplasmic reticulum stress, full-length CLU is released from the secretory system to the cytosol and that the CLU released is comprised of a range of species with different apparent masses, attributable to different levels of glycosylation associated with their origins from the endoplasmic reticulum (less conjugated sugars) or Golgi (more conjugated sugars) (Nizard, Tetley et al. 2007, Li, Zoubeidi et al. 2013). Lastly, none of the previous studies reporting alternative CLU protein isoforms ever directly confirmed the molecular identity of an endogenously-expressed CLU protein species found outside the secretory system – for example, by demonstrating the lack of the normal N-terminal sequence associated with the full-length protein. Therefore, the existence of physiologically relevant levels of alternative CLU protein isoforms remains to be confirmed.

Sequence analyses have predicted secondary structures in CLU but these have yet to be experimentally validated. For example, sequence analysis predicts that the glycosylated, disulfide-bonded core of CLU is flanked by five amphipathic alpha-helices (Bailey, Dunker et al. 2001) to produce an armed molecule with regions of native disorder, resulting in a dynamic, molten globule-like structure with the capacity to bind a variety of different molecules (Bailey, Dunker et al. 2001). This includes hydrophobic regions exposed on misfolded proteins, important for CLU function as a molecular chaperone (Bailey, Dunker et al. 2001, Dabbs, Wyatt et al. 2013). CLU also binds to a variety of specific protein ligands, including the SC5b-9 complex of complement, and immunoglobulins (Wilson and Easterbrook-Smith 2000) and histones (Cunin, Beauvillain et al. 2016).

The three-dimensional structure of CLU is unknown. Owing to its extensive post-translational modifications, CLU has proved largely intractable to conventional structural approaches such as X-ray diffraction. There have been no crystal structure determinations for CLU, and only limited analyses by mass spectrometry (Kapron, Hilliard et al. 1997, Stewart, Aquilina et al. 2007) and nuclear magnetic resonance (Poon, Treweek et al. 2002). Similarly, the structural elements responsible for the molecular chaperone activity of CLU are not yet known. Importantly, the recent report by the Wilson team of a high-yield expression/purification system for fully post-translationally modified chaperone-active recombinant CLU has opened the possibility of using mutagenesis for the first time to identify these elements (Satapathy, Dabbs et al. 2020).

3. Biological Activities of Clusterin

3.1. Proteostasis

Prior to the discovery that CLU was a molecular chaperone, a report appeared describing a 14-bp DNA element that is strictly conserved between CLU gene proximal promoters from different vertebrate classes. This region is specifically recognized by HSF1 (heat shock factor 1) transcription factor and in transient expression assays, mediates heat shock-induced gene expression (Michel, Chatelain et al. 1997). HSF1 was well known for its ability to activate the expression of a family of intracellular chaperones, the heat shock proteins (Miller and Fort 2018). As mentioned above, early reports described the upregulation of CLU expression in many different models of disease and stress. These and other observations were used to support the suggestions that (i) the HSF1 response element explained the high sensitivity of CLU expression to stress conditions, and (ii) clusterin was an extracellular version of a heat shock protein (Michel, Chatelain et al. 1997). Two years later, the latter hypothesis was confirmed by the ilson team when they showed that CLU specifically inhibits protein aggregation in vitro (Humphreys, Carver et al. 1999).

Depending on the individual protein, and the prevailing physical and chemical conditions, misfolded proteins can interact to form aggregates lacking any definable structure (amorphous aggregates), or assemble into beta sheet-rich ordered fibrillar structures (amyloid), as shown in (Khurana, Gillespie et al. 2001, Bettens, Brouwers et al. 2012). In vitro assays demonstrated that CLU is able to potently inhibit both forms of protein aggregation, as illustrated in Figure 2, and is classified as a "holdase" chaperone (Humphreys, Carver et al. 1999, Poon, Easterbrook-Smith et al. 2000, Yerbury, Poon et al. 2007). The term holdase means that, like the intracellular small heat shock proteins, CLU binds to misfolded proteins in an ATP-independent manner, to stabilize them in a soluble form and prevent their aggregation.

Figure 2. Protein Aggregation Assays.

Assays measuring the effect of CLU on the heat-induced amorphous aggregation of two enzymes, glutathione-S-transferase (GST) and catalase. Absorbance at 360 nm (A³⁶⁰), as a function of time (min). A, GST (200 mg/ml) or mixtures of GST (200 mg/ml) and CLU (at the concentrations indicated) heated at 60°C; B, catalase (200 mg/ml) or mixtures of catalase (200 mg/ml) and CLU (at the concentrations indicated) heated at 60°C. Concentrations of CLU in mixtures: ♢, 0 mg/ml; ▷, 25 ug/ml; ☐ 50 ug/ml; ◪ , 100 ug/ml. The results shown are representative of three independent experiments. From (Humphreys, Carver et al. 1999), with permission.

One demonstration of the physiological significance of the chaperone function of CLU is the observation that its immunodepletion from human blood plasma renders plasma proteins susceptible to stress-induced precipitation (Poon, Rybchyn et al. 2002). Furthermore, CLU knockout mice have increased tissue damage after heat shock (Bailey, Aronow et al. 2002), myosin-induced autoimmune myocarditis (McLaughlin, Zhu et al. 2000), or post-ischemic brain injury (Wehrli, Charnay et al. 2001). Aging CLU knockout mice develop protein deposits in the kidney leading to glomerular neuropathy, which directly implicates CLU in the clearance of misfolded proteins (Rosenberg, Girton et al. 2002).

Secreted CLU has been suggested to facilitate the clearance of extracellular misfolded proteins and the aggregates they form by binding to them and promoting their internalization via specific cell surface receptors (Wyatt, Yerbury et al. 2011) for subsequent degradation via the ubiquitin-proteasome system and/or autophagy (Nizard, Tetley et al. 2007, Zhang, Kumano et al. 2014). Injection of complexes formed between CLU and misfolded protein into rats, showed their fast uptake by hepatocytes and subsequent trafficking to lysosomes for degradation (Wyatt, Yerbury et al. 2011). It has also been proposed that CLU works together with the plasminogen activation system to promote the clearance of pathological protein aggregates from the extracellular space (Constantinescu, Brown et al. 2017). Amorphous protein aggregates activate plasmin in vitro and activated plasmin in turn facilitates the proteolytic breakdown of protein aggregates into heterogeneously sized plasmin-generated protein fragments, which contain regions of exposed hydrophobicity (Constantinescu, Brown et al. 2017). It has been proposed that CLU forms part of an extracellular protein quality control system that helps to maintain proteostasis (Yerbury, Stewart et al. 2005), as outlined in Figure 3.

Figure 3. Model for Roles of Extracellular Chaperones in Proteostasis.

Misfolded proteins in body fluids can bind directly to cell surface receptors for internalization and subsequent breakdown in lysosomes. Alternatively, misfolded client proteins (either as monomers or potentially toxic oligomers) can be bound by an extracellular chaperone (EC) such as CLU which neutralizes their toxicity, inhibits their further aggregation, and facilitates their receptor-mediated uptake and intracellular degradation within lysosomes. Lastly, CLU can bind to cytotoxic plasmin-generated protein fragments (PGPFs) released by the proteolytic degradation of amorphous protein aggregates, to neutralize their toxicity and facilitate their uptake/degradation as described before.

In addition to its chaperone activity, CLU is also proteostatic by virtue of its ability to inhibit the activity of specific members of the matrix metalloproteinase (MMP) family, a group of secreted or membrane-associated proteinases that are key players in tissue morphogenesis and remodeling (Page-McCaw, Ewald et al. 2007). This was shown first for MMP25 (Matsuda, Itoh et al. 2003). CLU was found to inhibit activity of MMP25, but not MMP14 or MMP2. In a later study, the Fini team identified CLU as an MMP9-interacting protein in a yeast-two hybrid screen (Jeong, Ledee et al. 2012). CLU bound very strongly to the truncated form of MMP9 lacking the pro-domain, with an affinity constant of 2.63 nmol/L. CLU had an even higher affinity for pro-MMP9 than activated MMP9. CLU was found to inhibit the enzymatic activity of MMP9, comparing quite favorably to the synthetic small molecule inhibitor SB-3CT. In this study, CLU also inhibited enzymatic activity of MMP2 and MMP3, and to a lesser extent, MMP7. The physiologic relevance of MMP inhibition was demonstrated by showing that CLU inhibited MMP9-mediated dissolution of tight junctions in human epithelial cell cultures (Jeong, Ledee et al. 2012). CLU also binds the extracellular proteinase CTSK (cathepsin K) with high affinity [(K(d) = 0.5-0.6 nM] (Novinec, Lenarcic et al. 2012). In this case however, CLU binding does not alter the activity of the enzyme, but instead increases its stability. The interaction also results in liberation of CTSK from inhibition by excess substrate. Thus, CLU would likely increase net proteolytic activity due to CTSK.

3.2. Cytoprotection

In general, CLU levels are low in unstressed cells, but are elevated in different stress conditions and diseases (Michel, Chatelain et al. 1997, Viard, Wehrli et al. 1999). Induction of CLU expression during programmed cell death was first reported for the prostate undergoing involution following castration as a treatment for prostate cancer. Testosterone-repressed prostate message-2 (Leger, Montpetit et al. 1987) and sulfated glycoprotein-2 mRNA (Bettuzzi, Hiipakka et al. 1989) were independently cloned from the prostate undergoing involution following castration by two different groups. Sequence analysis showed they were identical to one another and to CLU. Later studies identified CLU induction in many other organ systems undergoing massive apoptosis (e.g. (Buttyan, Olsson et al. 1989)).

Analysis of CLU expression in retinal degenerative disease provided the first information on the cytoprotective role of CLU. In the earliest study, differential hybridization screening of a human retinal cDNA library revealed an increase in CLU expression in diseased retina (Jones, Meerabux et al. 1992). Subsequently, increased CLU expression was localized to apoptotic photoreceptors in retinal degenerative disease in humans (Jomary, Neal et al. 1993) and also in a mouse model (Jomary, Ahir et al. 1995, Agarwal, Jomary et al. 1996). An increase in CLU expression was also seen in light-induced retinal damage in rats (Wong, Kutty et al. 1994, Jomary, Darrow et al. 1999). These studies linked CLU to cell death by apoptosis but did not establish whether its role was causal or protective.

To define the relationship between progressive photoreceptor degeneration and CLU upregulation, the spatiotemporal distribution and level of CLU mRNA was assessed in a rat model of retinal degeneration. Expression was increased in the retinal pigment epithelium and inner retina, but not detected in the photoreceptors. These results indicate that increased CLU mRNA is not causally involved in apoptotic mechanisms of photoreceptor death (Jomary, Darrow et al. 1999).

To address the question of CLU’s role in apoptosis directly, transgenic mice were generated in which a rat CLU transgene was expressed in photoreceptor cells under the transcriptional control of the human IRBP (interphotoreceptor binding protein) promoter. A reduction in apoptotic staining in the transgenic retinas was observed from birth to postnatal day 15. These results provided the first evidence that CLU is cytoprotective (Jomary, Chatelain et al. 1999).

The cytoprotective activity of CLU has been well studied in connection with cancer cell resistance to chemotherapeutics (Djeu and Wei 2009, Matsumoto, Yamamoto et al. 2013, Toren and Gleave 2013, Koltai 2014). CLU is also highly expressed in retinoblastoma, a childhood cancer that begins in the retina (Song, Jun et al. 2013). As with other cancers, apoptosis of retinoblastoma cell death due to treatment with cisplatin was prevented by co-treatment with, or over-expression of CLU.

Both reduced and elevated levels of secreted CLU are associated with disease states. The former likely represents dysfunction, and the latter a compensatory stress response (Wyatt, Yerbury et al. 2013). CLU is found in molecular aggregates associated with diseases of protein and lipid deposition, such as Alzheimer’s disease and atherosclerosis (Chiti and Dobson 2006). The presence of CLU in these insoluble aggregates is thought to represent a local overloading of CLU’s capacity to inhibit protein aggregation (Chiti and Dobson 2006). However, if CLU can attain a sufficient concentration threshold, it potently inhibits protein aggregation and provides substantial cytoprotection (Yerbury, Poon et al. 2007).

CLU also protects cells against the toxicity of protein oligomers formed during amyloid aggregation implicated in a variety of neurodegenerative diseases, such as Alzheimer's and prion diseases (Yerbury, Poon et al. 2007, Narayanan, Redfern et al. 2013). CLU further protects the toxicity of proteolytic fragments (Constantinescu, Brown et al. 2017). This protective effect of CLU may occur via binding to exposed hydrophobic regions.

CLU has also been shown to protect cells through effects on intracellular signaling pathways. In one mechanism, CLU binds to members of the low-density lipoprotein receptor-related protein (LRP) family including LRP2 (Kounnas, Loukinova et al. 1995), LPR8 and VLDLR (Leeb, Eresheim et al. 2014). Binding induces activation of AKT and promotes cell survival (Carver, Rekas et al. 2003). It has also been suggested that CLU inhibits proapoptotic signal transduction by interacting with surface receptors (Carver, Rekas et al. 2003), suppressing TP53-activating stress signals, inhibiting BAX activation and CYCS (cytochrome C) release (Trougakos, Lourda et al. 2009), and promoting translocation of NF-kappaB to the nucleus to activate the expression of pro-survival genes (Zoubeidi, Ettinger et al. 2010).

Mutations in the CLU gene has been identified as an important risk locus for Alzheimer's disease (Lambert, Heath et al. 2009, Bettens, Brouwers et al. 2012). Functional analyses suggest that reduced secretion of the CLU protein may be the mode of action for three CLU coding mutations (Bettens, Vermeulen et al. 2015), however, the functional consequences of Alzheimer's disease-associated CLU mutations remain to be determined. CLU concentration in cerebrospinal fluid is low compared to other bodily fluids, suggesting protective activity could be easily overwhelmed and that supplementation might be of therapeutic value (Bettens, Vermeulen et al. 2015).

3.3. Anti-Inflammation

A third major activity of CLU is anti-inflammation. One of the first anti-inflammatory mechanisms identified was CLU binding to lipoproteins in the circulation. CLU (then called ApoJ) was found to associate specifically with subclasses of high density lipoproteins (HDLs) that protect against atherosclerosis (de Silva, Stuart et al. 1990, Jenne, Lowin et al. 1991). CLU is important for stabilizing protein components of the HDLs, thus maintaining their atheroprotective properties (Navab, Hama-Levy et al. 1997, Riwanto, Rohrer et al. 2013). Low density lipoproteins (LDLs) and very low-density lipoproteins (VLDLs) contain a single copy of APOB (apolipoprotein B), and transport cholesteryl ester cargo to tissues expressing LDL receptors. CLU binds to aggregated LDL in human plasma and plays a protective role against LDL aggregation (Martinez-Bujidos, Rull et al. 2015). CLU is also distributed in aortas with atherosclerotic lesions (Ishikawa, Akasaka et al. 1998) where it prevents endothelial cell activation and limits the proinflammatory response (Urbich, Fritzenwanger et al. 2000, Navab, Anantharamaiah et al. 2005, Van Lenten, Navab et al. 2008).

Also discovered early on was the capacity of CLU (then called SP-40,40) to inhibit complement activation (Murphy, Kirszbaum et al. 1988, Murphy, Saunders et al. 1989, Choi, Nakano et al. 1990). The complement system is a part of the immune system that enhances (complements) the ability of antibodies and phagocytic cells to clear pathogenic microbes and damaged cells, promote inflammation, and attack the cell membrane of the pathogen (Rus, Cudrici et al. 2005). When triggered by various activators, proteases of the system catalyze cleavage of specific proteins to release cytokines and initiate an amplifying cascade of further cleavages. The result is stimulation of phagocytes to clear foreign and damaged material, inflammation to attract additional phagocytes, and activation of the cell-killing membrane attack complex. CLU was identified in glomerular immune deposits as part of the membrane attack complex of complement (Murphy, Kirszbaum et al. 1988). Purified CLU was shown to inhibit C5b-6-initiated hemolysis in a dose-dependent manner (Murphy, Saunders et al. 1989) by binding to complement component SC5b-9 (Choi, Nakano et al. 1990).

An example providing support for an anti-inflammatory role of CLU are studies in pulmonary disease models. In a mouse asthma model, the total number of immune cells in bronchoalveolar lavage fluid and the lung was increased in Clu knockout mice as compared to wild type mice. Of these immune cells, inflammatory dendritic cells and monocyte populations in the lung were significantly increased, accompanied by increased levels of various chemokines (Hong, Kwon et al. 2016). In a mouse model for hyperoxic acute lung injury, which is characterized by excessive pulmonary inflammation, a greater injury and mortality rate was observed in Clu knockout mice as compared with wild type mice. Treatment with recombinant CLU attenuated hyperoxia-induced apoptosis (Hong, Kim et al. 2021).

Another example is degenerative joint disease. A gene expression profiling study of arthritic tissues identified CLU as one of the most differentially expressed genes in rheumatoid arthritis Devauchelle, Marion et al. 2004). CLU knockdown with small interfering RNA promoted production of the inflammatory cytokines IL6 and IL8 (Devauchelle, Essabbani et al. 2006). Clusterin serum levels are elevated in patients with early rheumatoid arthritis and predict disease activity and treatment response (Kropackova, Mann et al. 2021).

In a recent study that included co-author Wilson, CLU was found to dampen the autoimmune response (Cunin, Beauvillain et al. 2016). CLU binding to apoptotic cells potentiated their phagocytosis by macrophages. In a model of apoptotic cell-induced autoimmunity, and relative to control mice, CLU knockout mice developed symptoms of autoimmunity, including the generation of anti-dsDNA antibodies, deposition of immunoglobulins and complement components within kidneys, and splenomegaly.

4. Clusterin in Eye Diseases Characterized by Insoluble Extracellular Deposits

Table 1 lists diseases of the eye characterized by insoluble extracellular deposits, in which CLU protein has been consistently identified as an abundant component. As noted above, the presence of CLU in insoluble extracellular deposits has been interpreted to mean that its capacity to perform its chaperone function is overwhelmed in the disease state. The specific diseases are discussed more below. We summarize these examples in Figure 4.

Table 1.

Protein Misfolding Diseases of the Eye

Disease	Disease Genes	Deposit Type	Aggregating Proteins	Colocalizing Chaperones
Retinal Degenerative Disease
Age-Related Macular Degeneration	C3, C5	Amyloid	APOA1, VTN, C3, C5	APOE, CLU, CRYAA, CRYAB
High Tension Glaucoma
Pseudoexfoliation syndrome/glaucoma	LOXL1	Amorphous	ELN, FBN1, FBN2	CLU
Corneal Dystrophies
Lattice corneal dystrophy type I	TGFBI	Amyloid	TGFBI fragment, HTRA1	APCS, APOE, CLU
Granular corneal dystrophy type I	TGFBI	Amorphous	Intact TGFBI	APCS, CLU
Granular corneal dystrophy type II	TGFBI	Amorphous and amyloid	Dependent on deposit type	APCS, CLU
Fuch’s endothelial corneal dystrophy	COL8A2, TCF4, TCF8, SLC4A11, LOXHD1, AGBL1	Amorphous, fibrillar	Collagens, TGFBI	CLU
Gelatinous drop-like corneal dystrophy	TACSTD2	Amyloid	TACSTD2, LFN	CLU

Figure 4. Clusterin in Eye Diseases Characterized by Insoluble Extracellular Deposits.

Eye structures where clusterin is found in extracellular deposits associated with specific diseases, as discussed in section 4.

4.1. Age-Related Macular Degeneration

Age-related macular degeneration (AMD) is a retinal degenerative disease and a leading cause of vision loss in people of age 50 and older. About 90 percent of all people with the disease have “dry” AMD, a condition in which layers of the macula (including the photoreceptors and the retinal pigment epithelium) get progressively thinner, with loss of function. This is called atrophy. About 10 percent of all cases of AMD become “wet” AMD, a condition in which new blood vessels grow in the choroid layer behind the retina. The new vessels are weak, and they leak fluid, lipids and blood. This can cause scar tissue to form and retinal cells to stop functioning.

Immune response and oxidative injury have been implicated in the pathogenesis of AMD, however, comprehensive understanding of the disease process is lacking [reviewed in (Anderson, Mullins et al. 2002)]. In its early stages, AMD is characterized by the presence of “drusen”, i.e., extracellular deposits that accumulate between the basal surface of the retinal pigment epithelium and Bruch's membrane. A wide spectrum of amyloid structures have been identified (Isas, Luibl et al. 2010). Drusen are comprised of carbohydrates, zinc, and proteins common to extracellular deposits associated with several different diseases including atherosclerosis, elastosis and amyloidosis [reviewed in (Apte 2021)]. Various complement components (Mullins and Hageman 1999, Mullins, Russell et al. 2000, Anderson, Ozaki et al. 2001, Crabb, Miyagi et al. 2002, Malek, Li et al. 2003, Li, Chung et al. 2005, Lengyel, Flinn et al. 2007, Wang, Clark et al. 2010), as well as CLU (Crabb, Miyagi et al. 2002) are consistent components of drusen. Consistent with this, DNA sequence variants in complement genes are associated with increased disease risk (Anderson, Mullins et al. 2002, Edwards, Ritter et al. 2005, Gold, Merriam et al. 2006, Yates, Sepp et al. 2007, Fagerness, Maller et al. 2009, Zhan, Larson et al. 2013). Complement gene polymorphisms associated with increased risk for AMD hinder natural mechanisms for negative regulation of the complement system [reviewed in (Qin, Dong et al. 2021)], suggesting that CLU’s activity as a complement inhibitor might reduce disease risk.

Biomarkers would be valuable for diagnosing and grading AMD progression. A recent study identified candidate biomarkers for AMD using an unbiased proteome analysis of the aqueous humor (Rinsky, Beykin et al. 2021). CLU and SERPINA4 (kallistatin) were significantly increased in AMD compared to controls, and CLU, SERPINA4 and TF (transferrin) showed differential expression across the stages of the disease. The level of CLU appears to distinguish eyes at the atrophic and neovascular stages of AMD. Thus CLU may be a valuable biomarker for AMD diagnosis and grading of progression.

4.2. Pseudoexfoliation Glaucoma

Pseudoexfoliation syndrome (PEX) is an age-related systemic syndrome that manifests mainly as elevated intraocular pressure (Ritch and Schlotzer-Schrehardt 2001, Elhawy, Kamthan et al. 2012). Pathology is associated with the gradual deposition of “PEX” material from the lens, onto anterior chamber structures including the lens capsule, ciliary body, zonules, corneal endothelium, iris and pupillary margin. Chronic ocular hypertension leads to optic nerve damage characteristic of glaucoma. Pseudoexfoliation glaucoma (PEXG) is the most common type of secondary open-angle glaucoma (OAG) worldwide (Ritch and Schlotzer-Schrehardt 2001, Elhawy, Kamthan et al. 2012).

PEX material has been variably described as “fluffy,” “flakes” and “fibrils”. It is an amorphous substance, but looks fibrillar by electron microscopy (Ritch and Schlotzer-Schrehardt 2001)). Numerous extracellular matrix proteins have been demonstrated to be an integral part of PEX deposits, identified using immunological and proteomics techniques (Ovodenko, Rostagno et al. 2007). CLU has been reproducibly identified as a component of PEX deposits (Zenkel, Kruse et al. 2006, Hardenborg, Botling-Taube et al. 2009, Sharma, Chataway et al. 2009, Zenkel and Schlotzer-Schrehardt 2014). Significantly, CLU mRNA expression was found to be reduced in the iris, lens, and ciliary processes of PEX patients (Zenkel, Poschl et al. 2005) and CLU protein was reduced as well (Zenkel, Kruse et al. 2006). This is consistent with the idea that an insufficiency of CLU contributes to the deposition of extracellular deposits in the eye.

Variants in the gene encoding LOXL1, an enzyme involved in cross-linking elastin fibers, are highly associated with PEXG in most populations. Two SNPs in LOXL1 confer a higher than 99% population attributable risk for PEXG in the Nordic population. LOXL1 risk alleles are also frequently found in controls, arguing that other genetic and/or environmental factors are necessary for the disease to fully manifest (Wiggs, Kang et al. 2018). One of these may be degree of ultraviolet light exposure (Stein, Pasquale et al. 2011, Pasquale, Kang et al. 2014). Interestingly, CLU associates with altered elastic fibers in human photoaged skin and prevents ultraviolet light-induced elastin aggregation in vitro (Janig, Haslbeck et al. 2007). Common CLU variants appear to contribute modestly to PEXG risk, although larger datasets are needed to confirm (Fan, Pasquale et al. 2015).

4.3. Corneal Dystrophies

Corneal dystrophies are inherited disorders that affect corneal transparency and refraction due to deposition of insoluble protein material extracellularly (Klintworth 2008, Moshirfar, Bennett et al. 2021). Deposits are localized in different corneal layers and exhibit different morphologies, depending on the gene and specific mutation involved. While the mutated protein is ultimately responsible for the deposits, it is not always a component of the deposits.

A group of rare autosomal dominant corneal dystrophies are due to mutations in TGFBI (Transforming Growth Factor Beta-Induced) (Moshirfar, Bennett et al. 2021, Moshirfar, West et al. 2021). The wildtype TGFBI protein, which is highly abundant in the cornea, binds to both collagens and integrins and appears to inhibit cell adhesion. Three disease forms are characterized by morphology of TGFBI-containing stromal deposits, which are predicted by the specific gene mutation (Nielsen, Poulsen et al. 2020). Lattice corneal dystrophy type I (LCD1) presents with linear/branching opacities in the corneal stroma that form a ropy lattice structure and the deposits are amyloid in character (Moshirfar, West et al. 2021). In contrast, granular corneal dystrophies (GCD) present with discrete, irregularly shaped opacities of early onset. In both type 1 (GCD1) and type 2 (GCD2, Avellino) forms of the disease, deposits have an amorphous character; amyloid deposits can also appear in some patients with GCD2 (Han, Choi et al. 2016). CLU has been identified in all deposits (Nielsen, Poulsen et al. 2020).

Laser capture microdissection of deposits and tandem mass spectrometry analyses have revealed that deposits all contain the mutant TGFBI protein (Korvatska, Henry et al. 2000, Karring, Runager et al. 2012, Courtney, Poulsen et al. 2015, Venkatraman, Dutta et al. 2017). However, while the amorphous deposits consist largely of intact TGFBI, proteolytic fragments of TGFBI are found in amyloid deposits. The serine proteinase HTRA1 is also found in the amyloid deposits, suggesting it promotes formation of the amylogenic peptides. A recent study employing x-ray crystallography demonstrated a new dimer interface in the crystal packing of amylogenic TGFBI protein mutants, which is not present in wildtype TGFBI protein or in a mutant that forms amorphous aggregates (Nielsen, Gadeberg et al. 2021).

Gelatinous drop-like corneal dystrophy (GDLD) is a rare autosomal recessive disease characterized by subepithelial and stromal amyloid deposits (Kaza, Barik et al. 2017). The consensus in the scientific community is that the amyloid originates from the tear film, as it also contains the abundant tear protein LFN (lactoferrin) (Takaoka, Nakamura et al. 2007). The disease-causing mutation is in the TACSTD2 gene (Klintworth, Sommer et al. 1998), which regulates expression of genes encoding the tight junction-related proteins ZO1, OCN, DSP and claudins. It is thought that TACSTD2 mutations result in disruption of epithelial barrier function, allowing tear proteins (such as LFN) to migrate underneath the epithelium (Kawasaki and Kinoshita 2011). Aggregation of LFN as amyloid may be due to the stress of epithelial damage. Deposits contain CLU (Nishida, Quantock et al. 1999).

Fuchs’ Endothelial Corneal Dystrophy (FECD) is a genetically heterogeneous disease that is the most common cause of corneal endothelial dysfunction, accounting for 36% of corneal transplantations performed in the USA (Sarnicola, Farooq et al. 2019). FECD is characterized by thickening of Descemet’s membrane, “guttae” or collagenous deposits that accumulate posterior to Descemet’s membrane and loss of corneal endothelial cells (Price, Mehta et al. 2020, Moshirfar, Somani et al. 2021). While several mechanisms have been proposed, the pathogenesis remains unknown.

The rare early-onset form of FECD is ascribed to mutations in the gene for COL8A2 encoding a collagen component of Descemet’s membrane. The more common late-onset form of FECD has been linked to mutations in several genes including SLC4A11, a Na⁺-dependent pH modulator that may regulate endothelial cell survival through antiapoptotic mechanisms; TCF4 and TCF8, which encode transcription factors that regulate TGF-beta signaling and epithelial-mesenchymal transition; LOXHD1, encoding a protein thought to be involved in targeting proteins to the plasma membrane; and AGBL1, which encodes a glutamate decarboxylase (Liu, Zheng et al. 2021).

CLU expression has been demonstrated in human corneal endothelium from FECD eyes by both polymerase chain reaction (PCR) and immunohistochemistry (Dota, Nishida et al. 1999). CLU rotein is localized to the center of guttae (Jurkunas, Bitar et al. 2008, Jurkunas, Bitar et al. 2009, Weller, Zenkel et al. 2014). Interestingly, wild-type TGFBI co-localizes with CLU in guttae, indicating its aggregation in insoluble deposits is not solely due to mutation. Late-onset FECD has been associated with common CLU polymorphisms (Gu, Wei et al. 2011, Kuot, Hewitt et al. 2012).

5. Clusterin as a Biotherapeutic for Eye Diseases

The important molecular chaperone activity of CLU, and its increased expression in response to cellular stress, suggests that supplementation of CLU levels could have therapeutic potential in diseases of extracellular deposits in the eye (Wiggs, Kang et al. 2018). It is possible that supplementation of CLU levels might slowly resolve insoluble extracellular deposits already deposited in diseased eye tissues. It seems even more likely that supplementation at the early stages of disease could slow the accumulation of such extracellular deposits. However, the role of CLU in diseases of extracellular deposits in the eye needs to be better defined before therapeutic approaches involving CLU can be entertained. A more straightforward case can be made for therapeutic use of CLU based on its other proteostatic role as a proteinase inhibitor, as well as its cytoprotective and anti-inflammatory properties. For example, exogenously-added CLU protein exhibited a protective effect against oxidative stress-induced cell death in corneal endothelial cells in culture (Shin, Kim et al. 2009). Below we discuss some other examples suggesting therapeutic use of CLU in specific eye disease. We summarize these examples in Figure 5.

Figure 5. Clusterin as a Biotherapeutic for Eye Diseases.

Eye structures where clusterin may have therapeutic uses for specific diseases, as discussed in section 5.

5.1. Retinal and Choroidal Vascular Diseases

Retinal and choroidal vascular diseases constitute the most common causes of moderate and severe vision loss in developed countries. They can be divided into retinal vascular diseases such as diabetic retinopathy (DR), in which there is leakage and/or neovascularization from retinal vessels, and subretinal neovascularization characteristic of wet AMD, in which new vessels grow into the normally avascular outer retina and subretinal space. The blood-retinal barrier controls the entry of fluid and electrolytes into the extracellular space. The inner and outer blood-retinal barrier are formed by the retinal vascular endothelium and the retinal pigment epithelium, respectively. Tight junctions between cells maintain barrier function [reviewed in (Campochiaro 2013)].

Supplementation of human retinal vascular endothelial cell cultures with CLU protein was shown to protect against ischemia-induced loss of tight junction proteins (Kim, Yu et al. 2007). Moreover, in a cell culture model of DR ischemia induced by treatment with advanced glycation end product, supplementation of cultures with CLU protein was cytoprotective (Kim, Kim et al. 2010). CLU protein also protected retinal pigment epithelial cells against oxidative stress (Kim, Kim et al. 2010). Intravitreal injection of CLU protein into the eyes of mice effectively inhibited hyperpermeability caused by treatment with VEGF (vascular endothelial growth factor). In eyes of mice with streptozotocin-induced diabetes, intravitreal CLU injection protected tight-junctions of the retinal vasculature (Kim, Kim et al. 2010). These findings suggest that CLU protein may have therapeutic potential in the treatment of blood-retinal barrier breakdown due to DR and wet AMD.

5.2. Stem Cell Transplantation at the Ocular Surface

Limbal stem cell deficiency is a clinical condition of the ocular surface characterized by damage of corneal limbal stem cells, which results in an impairment of corneal epithelium turnover and in an invasion of the cornea by the conjunctival epithelium (Figueiredo, Glanville et al. 2021). Renewal of the corneal epithelium occurs from a reservoir of adult stem cells located in the limbus (Davanger and Evensen 1971, Schermer, Galvin et al. 1986); reviewed in (Daniels, Dart et al. 2001). The slow-cycling limbal stem cells give rise to transit-amplifying daughter cells that migrate centripetally towards the center of the cornea, while also dividing to generate cells that migrate upward in the epithelial cell layers and differentiate. In patients with limbal stem cell deficiency, the conjunctivalization of the cornea is associated with visual impairment and cornea transplantation has poor prognosis for recurrence of the conjunctivalization.

In the first attempts to restore the supply of epithelial stem cells in limbal stem cell deficiency, limbal/conjunctival tissue was transplanted from the fellow eye (Kenyon and Tseng 1989). These cells were isolated and expanded on feeder layers of mouse 3T3 fibroblasts (Tseng, Kruse et al. 1996). Intriguingly, the colony forming efficiency of corneal limbal epithelial stem cells was significantly enhanced by growth on a 3T3 cell feeder layer transfected with an expression vector encoding CLU (Okada, Kawakita et al. 2011). This was found to stimulate feeder cell production of the growth-promoting cytokine, HGF.

Like other adult stem cells, corneal limbal epithelial stem cells exclude the DNA-binding dye Hoechst 33342, and can thus be isolated using as a “side population” using a technique fluorescence-activated cell sorting (Scharenberg, Harkey et al. 2002). The use of side population cells for reconstruction of various tissues damaged by trauma or disease has been widely explored in recent years (e.g., (Gangavarpu and Huss 2011, Hertsenberg and Funderburgh 2015). In one study, side population cells isolated from mouse lacrimal and salivary glands were transplanted into the glands of mice made hypo-functional by irradiation. The secretions from both glands in the recipient mice were restored within 2 months of transplantation, although the transplanted cells did not appear to expand. Side population cells isolated from salivary glands of CLU knockout mice had no therapeutic potential, whereas lentiviral transduction of CLU restored function. CLU directly inhibited oxidative stress and oxidative stress-induced cell damage in these cells (Mishima, Inoue et al. 2012).

Together these findings suggest CLU protein and/or CLU gene therapy could be useful for enhancing stem cell production in culture, as well as their transplantation in the eye.

5.3. Ocular Surface Disease in Dry Eye

Dry eye is a common, age-related, multifactorial disease of the ocular surface characterized by tear film dysfunction, ocular surface inflammation and desiccation (Craig, Nichols et al. 2017). Evaporative dry eye, the most prevalent subtype, is primarily due to Meibomian gland dysfunction, which leads to lipid-poor tears that evaporate too quickly. Aqueous-deficient dry eye is characterized by a lack of fluid secretion due to lacrimal gland dysfunction. Most patients present with a mixed subtype.

One rare type of aqueous-deficient dry eye is caused by LYZ (lysozyme) amyloidosis, a hereditary disease involving deposits containing a mutant form of LYZ (reviewed in (Brorsson, Kumita et al. 2010). LYZ is produced by the lacrimal gland and is one of the most abundant proteins in tears. Dry eye is often the first symptom of the disease, preceding diagnosis by several years (Barreiros, Galle et al. 2013). Disease-causing LYZ variants convert in vitro from normally soluble and functional proteins into intractable fibrillar aggregates (Booth, Sunde et al. 1997, Morozova-Roche, Zurdo et al. 2000). The reasons for dry eye symptoms have not been investigated, but deposition of amyloid in the lacrimal gland could lead to inflammation and tissue damage, resulting in reduced secretion of the aqueous component of tears.

Desiccation induces misfolding and aggregation of proteins, whether or not they have predisposing mutations (Chakrabortee, Boschetti et al. 2007, Koshland and Tapia 2019). A recent study provides evidence that protein aggregation occurs in the tears of dry eye patients, although the specific proteins involved have not been identified (Azharuddin, Khandelwal et al. 2015). It is not known whether this is a factor in disease, but protein aggregation certainly holds the potential to exacerbate inflammation and contribute to symptoms of dry eye.

Chaperones are known to protect against desiccation-induced protein misfolding (Xu, Dai et al. 2020) and it seems very likely that CLU and other extracellular molecular chaperones help to maintain protein solubility in the tears. In fact, CLU significantly inhibits the aggregation of amyloidogenic LYZ variant I56T at molar ratios of CLU to LYZ as low as 1:80 (Kumita, Poon et al. 2007). However, we found that the concentration of CLU in tears is reduced in human aqueous-deficient dry eye, suggesting that the capacity to protect may become overwhelmed in disease (Yu, Bhattacharya et al. 2018). Thus, supplementation of the tears with CLU could be of therapeutic value in maintaining tear protein solubilization under the desiccating conditions of dry eye. CLU’s capacity to inhibit MMPs, as well as its cytoprotective, anti-inflammatory and immunosuppressing properties could also provide substantial value in dry eye, by protecting the ocular surface epithelia and the epithelial barrier and ramping down inflammation.

To test the potential efficacy of CLU as a biotherapeutic for dry eye, we designed a study in which CLU levels were perturbed while mice were subjected to desiccating stress (Bauskar, Mack et al. 2015). Perturbation was accomplished in two ways. First, we supplemented CLU at the ocular surface via topical drops in wild-type mice. Second, we used Clu knockout mice. Ocular surface and barrier damage were assessed by vital dye staining. In wild-type mice, drops containing human plasma CLU, recombinant human CLU, recombinant mouse CLU, or a bovine lbumin control, were delivered topically, 4 times/day in a dose-response experiment. In all cases, CLU at a concentration of 1-3 ug/mL was found to prevent ocular surface damage or ameliorate pre-existing damage, bringing vital dye staining to baseline. In contrast, the ocular surface was found to be more vulnerable to desiccating stress in heterozygous CLU knockout mice as compared to wild-type littermates (we did not examine homozygous knockout mice).

We further evaluated CLU’s capacity to protect the ocular surface by 1) terminal deoxynucleotidyl transferase dUTP nick end-labeling (TUNEL) assay on corneal tissue sections to quantify cells undergoing apoptosis, and 2) by SDS-polyacrylamide gel electrophoresis of proteins extracted from the ocular surface epithelia, to evaluate proteolysis of barrier proteins. CLU treatment reduced apoptosis of ocular surface epithelial cells due to desiccating stress. CLU treatment also reduced proteolytic damage to two different ocular surface barrier proteins: OCLN, a component of the paracellular barrier, and LGALS3, a component of the transcellular barrier. The former is consistent with CLU’s activation of anti-apoptotic pathways and the latter is consistent with i ts action as a MMP9 inhibitor. Significantly, CLU addition to cultured OcS epithelial cells was found to inhibit proinflammatory cytokine-stimulated expression of MMP9. This suggests that CLU acts to inhibit both MMP9 activity and expression (Bauskar, Mack et al. 2015).

Finally, we measured the concentration of natural CLU present in mouse tears. We observed that the CLU concentration in normal mouse tears was somewhat above the effective supplementation threshold (~5 ug/mL) but dropped by about 1/3^rd in mice subjected to desiccating stress. We also observed in heterozygous CLU knockout mice that the tear CLU concentration was about half (Bauskar, Mack et al. 2015).

We conclude that natural tear CLU protects the ocular surface from desiccating stress and that topical CLU supplementation can be effective when natural levels are reduced (Bauskar, Mack et al. 2015). Taken together, our findings support the use of topical CLU as a therapeutic to treat ocular surface disease in dry eye.

5.4. Retinitis Pigmentosa

As discussed above, analysis of CLU expression in retinal degenerative disease provided the first information on the cytoprotective role of CLU, including in Retinitis Pigmentosa (RP) a human genetic disorder that causes progressive loss of vision. It is one of the most common inherited diseases of the retina (retinopathies), estimated to affect 1 in 3,500 to 1 in 4,000 people in the United States and Europe (Anonymous 2020). RetNet (https://sph.uth.edu/retnet/) currently lists 31 autosomal dominant mutations, 66 autosomal recessive mutations, and 3 X-linked mutations that cause the disease. Onset of symptoms is generally gradual and often begins in childhood. The underlying mechanism involves the progressive death and loss of rod photoreceptor cells in the back of the eye. This is generally followed by loss of cone photoreceptor cells. Diagnosis involves an examination of the retina for dark pigment deposits. There is currently no effective treatment or cure and complete blindness is common (O'Neal and Luther 2020).

Inspired by our findings on the therapeutic potential of CLU in ocular surface disease, our colleagues at the University of California (USC) investigated the possible value of CLU in protecting rod photoreceptors in RP (Vargas, Kim et al. 2017). The study employed the homozygous S334ter-line-3 of albino Sprague Dawley rats, a transgenic RP prototype with a rhodopsin mutation caused by truncated murine opsin gene at Serine residue 334 (S334ter-line-3). Intravitreal treatment with CLU resulted in robust preservation of rod photoreceptors at rat postnatal days 30, 45, 60 and 75, as compared to age-matched saline-injected controls. CLU treatment activated AKT and STAT3 in photoreceptors, which significantly reduced activity of apoptosis-promoting BAX. CLU treatment also activated STAT3 in Müller cells, suggesting an indirect pathway for CLU protection of rod photoreceptors. In a follow-up study, it was found that CLU treatment also suppressed the upregulation of NOS1 (neuronal nitric oxide synthase) , a known cause of cone cell death (Vargas, Yamamoto et al. 2021). These results suggest that CLU could be used therapeutically to prolong survival of photoreceptors in human RP.

6. Other Extracellular Chaperones in the Eye

6.1. Tear Proteins

CLU may not be the only extracellular chaperone acting to protect the eye, and in fact in every other cell type and body fluid examined, it has been found that multiple chaperones are present and may work together. Molecular chaperones comprise about 10 % of the total protein mass of human cells (Finka and Goloubinoff 2013) and more than 300 molecular chaperones have been identified in humans (Brehme, Voisine et al. 2014). CLU is now recognized to be only one of a growing family of proteins with extracellular chaperone or chaperone-like activity found in plasma and other bodily fluids, including A2M (alpha-2-macroglobulin), APOE (apolipoprotein E), APCS (Amyloid P Component, Serum), caseins, FGA (fibrinogen alpha chain) and HP (haptoglobin) [reviewed in (Dabbs, Wyatt et al. 2013)]. Others include ALB (albumin) (Marini, Moschini et al. 2005, Finn, Nunez et al. 2012), two lipocalin-like proteins: ORM1 (α1-acid glycoprotein) (Zsila 2010) and PTGDS (lipocalin-type prostaglandin D synthase) (Kanekiyo, Ban et al. 2007, Kannaian, Sharma et al. 2019), and SPARC (Chlenski, Guerrero et al. 2011). As shown in Table 1, APOE and APCS have been identified in extracellular deposits in eye disease.

Interestingly, A2M (Wyatt, Constantinescu et al. 2013) is a broad-spectrum protease inhibitor, with well-known action against MMPs (Baker, Edwards et al. 2002). This supports the idea that extracellular chaperones serve as multifunctional proteins, with proteinase inhibition being a common role.

Most plasma proteins are also present in tears, although in different amounts (Zhou and Beuerman 2012). We recently measured the concentration of CLU in normal human tears as 31±14 ug/ml (Yu, Bhattacharya et al. 2018), while the most recently reported concentration of CLU in human blood serum was determined to be 101 ± 42 ug/mL (Morrissey, Lakins et al. 2001). In the most comprehensive mass spectrometry analysis of tears, 1543 proteins were identified (Zhou, Zhao et al. 2012). A review of this list by our team revealed many proteins with extracellular chaperone or chaperone-like activity, here listed in order of abundance: A2M, HP, CLU, FGA, APOE and APCS. In addition, among lower abundance proteins, we identified heat shock proteins normally found intracellularly, e.g. HSP70 and HSPA5 (Yu, Bhattacharya et al. 2018).

6.2. Crystallins

Another group of extracellular molecular chaperones in the eye, previously thought to be entirely intracellular are the alpha-crystallins, which serve as primary structural components of the lens. Like CLU, alpha-crystallins CRYAA (alpha-crystallin B) and CRYAB (alpha-crystallin B) are evolutionarily related to small heat shock proteins and prevent aggregation of client proteins by means of their ATP-independent holdase activity (Horwitz 1992). This activity plays a crucial role in maintaining protein solubility in the lens with age. Cataracts are due to aggregation of crystallin proteins, thought to be due to accumulation of crystallin post-translational modifications, which eventually overwhelm the proteostasis network (Carver, Ecroyd et al. 2018). Recently, a class of small molecules that bind CRYAA and CRYAB and reverse their aggregation in vitro was identified. The most promising compound improved lens transparency in mouse models of hereditary cataract. It also partially restored protein solubility in the lenses of aged mice in vivo and in human lenses ex vivo (Makley, McMenimen et al. 2015).

Besides the lens, crystallins are also expressed in other cell types throughout the body, including in the eye. Moreover, CRYAB is secreted from retinal pigment epithelial cells (D'Agostino, Scerra et al. 2019). As shown in Table 1, both CRYAA and CRYAB have been identified as components of drusen in AMD (Mullins and Hageman 1999, Mullins, Russell et al. 2000, Anderson, Ozaki et al. 2001, Crabb, Miyagi et al. 2002, Malek, Li et al. 2003, Li, Chung et al. 2005, Lengyel, Flinn et al. 2007, Wang, Clark et al. 2010). Retinal pigment epithelial cells lacking CRYAB are more susceptible to oxidative and endoplasmic reticulum stress compared with normal retinal pigment epithelium [e.g., (Dou, Sreekumar et al. 2012)]. Furthermore, retinal pigment epithelial cells-overexpressing CRYAB are resistant to apoptosis (Sreekumar, Spee et al. 2012). These findings suggest that CRYAB, as an intracellular chaperone, plays a cytoprotective role against multiple stress stimuli that can contribute to retinal pigment epithelial cell death, which results in drusen formation and geographic atrophy. On the other hand, studies using knockout mice provide evidence that CRYAB contributes to development of subretinal fibrosis (Ishikawa, Sreekumar et al. 2016). Mechanistic experiments revealed that CRYAB promotes epithelial-mesenchymal transition of retinal pigment epithelial cells by acting as a molecular chaperone for transcription factor SMAD4. Thus, CRYAB appears to play both positive and negative roles in AMD.

6.3. Myocilin

Another abundant protein in which mutations result in a misfolding disease in the eye is MYOC (myocilin). MYOC cDNA was first isolated by Polansky and colleagues from cultures of trabecular meshwork cells of the conventional aqueous outflow pathway, as a transcript highly-induced by treatment with glucocorticoids (Polansky, Fauss et al. 1997) and was determined to be identical to the gene mutated at the GCL1A genetic locus causing an inherited form of high-tension primary open angle glaucoma (POAG) (Stone, Fingert et al. 1997). MYOC protein is widely distributed in ocular tissues and other organ systems, however, POAG is the only known pathology caused by MYOC mutations (Tamm 2002, Borras 2014). A toxic gain-of-function disease mechanism has been hypothesized (Wiggs and Vollrath 2001, Joe, Sohn et al. 2003, Liu and Vollrath 2004, Yam, Gaplovska-Kysela et al. 2007, Kwon, Fingert et al. 2009, Suntharalingam, Abisambra et al. 2012).

Trabecular meshwork cell expression of wild-type MYOC is induced by a variety of stresses, including physical stretch, heat shock, and proinflammatory cytokines (Tamm 2002). Disease variants, distributed across the 30 kDa olfactomedin domain cause MYOC to be sequestered intracellularly instead of being secreted (Joe, Sohn et al. 2003, Liu and Vollrath 2004, Yam, Gaplovska-Kysela et al. 2007), aggregating as detergent-insoluble species (Zhou and Vollrath 1999) exhibiting numerous characteristics of amyloid (Lieberman and Ma 2021). Clearance of MYOC disease variants appears to be hindered by their aberrant interactions with the intracellular molecular chaperone HSP90B1 (Lieberman and Ma 2021). This causes endoplasmic reticulum stress, activation of the unfolded protein response (Joe, Sohn et al. 2003, Wang, Zhuo et al. 2007), and ultimately, ocular hypertension (Zode, Kuehn et al. 2015). Significantly, treatment with a chemical chaperone reduced intraocular pressure, supporting the idea of protein aggregation as part of the disease mechanism (Zode, Kuehn et al. 2015).

Recently it was suggested that MYOC protein has chaperone-like activity, protecting the enzymatic activity of a client protein against thermal inactivation in a concentration-dependent manner (Anderssohn, Cox et al. 2011). The Fini team showed that over-expression of MYOC in trabecular meshwork cells inhibited expression of inflammatory cytokines (Itakura, Peters et al. 2015). In a shotgun proteomic analysis, MYOC was found to coimmunoprecipitate with the MMP inhibitor TIMP3. Using purified proteins, it was shown that MYOC markedly enhanced the inhibitory activity of TIMP3 toward MMP2 (Fini 2017, Joe, Lieberman et al. 2017). These findings provide additional support for the idea that extracellular molecular chaperones share a set of multifunctional activities.

6.4. Newly Identified Extracellular Chaperones

Recently, the Wilson team designed and implemented a novel strategy to discover new extracellular chaperones present in human blood (Geraghty, Satapathy et al. 2021). The strategy, which was based on the well-known propensity of chaperones to bind to regions of hydrophobicity exposed on misfolding proteins, employed a destabilized protein that misfolds at 37°C as “bait” to capture extracellular chaperones on magnetic beads. Proteins eluted from the beads were identified by mass spectrometry. Seven abundant serum proteins were selected for further analysis of chaperone activity. Five of these proteins were shown to specifically inhibit protein aggregation: VTN (vitronectin), SERPINA5 (plasminogen activator inhibitor-3), F2 (prothrombin) CIR (complement component 1r), and CIS (complement component 1s). VTN and SERPINA5 inhibited both aggregation of the Alzheimer's Abeta peptide to form fibrillar amyloid, and aggregation of CS (citrate synthase) to form unstructured (amorphous) aggregates. In contrast, F2, C1R and C1S inhibited the aggregation of Abeta, but did not inhibit CS aggregation. This is the first report that molecular chaperones have differing abilities to inhibit the amorphous and amyloid-forming protein aggregation pathways (Geraghty, Satapathy et al. 2021).

SERPINA5 is a widely expressed proteinase inhibitor of broad specificity suggested to play roles in hemostasis, host defense, and male reproduction (Yang and Geiger 2017). This adds to the list of extracellular chaperones that can also function as proteinase inhibitors. Interestingly, F2 has the opposite function as a proteinase that cleaves fibrinogen to fibrin, while C1R and C1S are proteinases of the antibody-dependent classical complement pathway (Mortensen, Sander et al. 2017). VTN (also known as “S protein”) binds to complement membrane attack complexes that fail to insert into membranes, to form the soluble SC5b-9 complex. As discussed above, CLU also associates with this complex (Murphy, Kirszbaum et al. 1988, Tsuruta, Wong et al. 1990). Thus, many extracellular chaperones appear to be multifunctional, and share activities that can work together to resolve inflammatory and associated processes that cause disease.

7. Future Directions and Conclusions

As we detail in this article, the molecular chaperone CLU has several homeostatic activities that enable it to protect cells and tissues. We have discussed here some examples of eye diseases in which there is strong support for the potential use of CLU as a biotherapeutic. We predict that other possible uses in the eye will continue to emerge. In support of this idea, a recent conference abstract reported that adenoviral gene therapy to introduce CLU into the aqueous outflow pathways of the mouse eye lowered intraocular pressure by maintaining optimal cell-matrix interactions (Pattabiraman 2019).

CLU may not be the only extracellular chaperone acting to protect the eye, and we have discussed others that may collaborate with CLU to protect cells and tissues. Likely more will be found, and our team has begun looking for chaperones among tear proteins. While the protein composition of tears is highly complex, a small number of highly abundant and moderately abundant proteins, including LYZ (lysozyme), LTF (lactoferrin), LCN1 (tear lipocalin) and LACRT (lacritin) are estimated to comprise more than 90% of the total tear protein by mass (Zhou and Beuerman 2012). The high concentrations of these proteins in tears, as well as some of their structural and biochemical properties, suggest they could function as molecular chaperones.

To develop extracellular chaperones as drugs, pure amounts must be prepared in large quantities. A range of valuable products are manufactured from blood plasma, including immunoglobulins, albumin, and clotting factors. The Wilson lab routinely purifies CLU from blood for laboratory research use [originally described in (Humphreys, Carver et al. 1999)]. However, products produced by recombinant DNA technology offer a safer option than plasma-derived products because they avoid potential blood-borne transmission of infectious diseases.

Recombinant protein drugs are part of a broader class of drug products called “biologics”, that represent the largest group of new products under development by the biopharmaceutical industry (Ho and Chien 2014). Biologics are held to rigorous safety standards by regulatory agencies, and their defined nature makes it possible to test their efficacy in well-controlled clinical trials. Biologics are promising therapeutics for inflammatory and neovascular eye diseases. A number of such drugs have been developed in the last decade and are currently on the market (Posarelli, Arapi et al. 2011, Joseph, Trinh et al. 2017).

CLU is highly conserved across evolution, exhibiting 76% sequence identity (79% similarity) between humans and mice. Importantly, recombinant human CLU and recombinant mouse CLU were equally effective in preventing ocular surface disease in our mouse dry eye model (Bauskar, Mack et al. 2015). This is important, as safety testing of CLU drug products in animal models is a necessary step in drug development. The Wilson team recently described a rapid and efficient method to produce structurally and functionally-validated recombinant human CLU at high yields (Satapathy, Dabbs et al. 2020). The process must still be adapted for compliance with Good Manufacturing Practices, however, this is an important first step towards manufacturing CLU for clinical use. Importantly, our new process for making large amounts of recombinant CLU also enables us to dissect the molecular structure of clusterin to identify protein domains important to function. This offers the possibility of producing functionally-enhanced engineered proteins for therapies. We predict that a therapeutic CLU drug product will be available within the next decade, and that its first uses will be to treat diseases of the eye.

Highlights.

• Clusterin was the first extracellular mammalian chaperone identified

• Clusterin has proteostatic, cytoprotective and anti-inflammatory activities

• Insufficient clusterin may result in failure to resolve extracellular deposits

• Clusterin’s homeostatic activities suggest it as a biotherapeutic for eye disease

• Clusterin may work together with other molecular chaperones found in the eye

Publication Statement.

This article has not been published previously (except in the form of an abstract, a published lecture or academic thesis) and it is not under consideration for publication elsewhere.Publication is approved by all authors and individuals acknowledged, and tacitly or explicitly by the responsible authorities where the work was carried out. If accepted, it will not be published elsewhere in the same form, in English or in any other language, including electronically without the written consent of the copyright-holder.