CLN3, at the crossroads of endocytic trafficking

Abstract

The CLN3 gene was identified over two decades ago, but the primary function of the CLN3 protein remains unknown. Recessive inheritance of loss of function mutations in CLN3 are responsible for juvenile neuronal ceroid lipofuscinosis (Batten disease, or CLN3 disease), a fatal childhood onset neurodegenerative disease causing vision loss, seizures, progressive dementia, motor function loss and premature death. CLN3 is a multipass transmembrane protein that primarily localizes to endosomes and lysosomes. Defects in endocytosis, autophagy, and lysosomal function are common findings in CLN3-deficiency model systems. However, the molecular mechanisms underlying these defects have not yet been fully elucidated. In this mini-review, we will summarize the current understanding of the CLN3 protein interaction network and discuss how this knowledge is starting to delineate the molecular pathogenesis of CLN3 disease. Accumulating evidence strongly points towards CLN3 playing a role in regulation of the cytoskeleton and cytoskeletal associated proteins to tether cellular membranes, regulation of membrane complexes such as channels/transporters, and modulating the function of small GTPases to effectively mediate vesicular movement and membrane dynamics.

1. Introduction

The neuronal ceroid lipofuscinoses (NCLs, Batten disease) are a group of inherited metabolic disorders that primarily lead to childhood onset neurodegeneration, with clinical symptoms that include seizures, vision loss, progressive loss of cognitive and motor function, and premature death [1]. A main pathological feature is the progressive accumulation of autofluorescent lysosomal storage material in cells within and outside of the central nervous system (CNS) [1]. Originally subclassified by age of onset, thirteen genetic NCL subtypes have now been defined [2]; therefore each subtype is now referred to by its gene nomenclature [3]. Autosomal recessive CLN2 and CLN3 disease (originally classified as classical late-infantile and juvenile NCL, respectively) are the most common subtypes in most demographic regions [1]. While there are significant efforts to develop enzyme replacement and gene therapies for some forms of NCL [4, 5], and a CNS-directed enzyme replacement therapy has been approved for CLN2 disease [6], patient care is limited to palliative treatment in most forms of NCL.

CLN3 disease is usually first recognized with new onset of vision loss between four and ten years of age. Progressively declining cognitive and motor function, emergence of seizures, and behavioral and psychiatric disturbances are typical in the progression of CLN3 disease, and most patients succumb to the disease in their late teens or early twenties [7, 8]. Linkage mapping led to the identification of the CLN3 gene in 1995, which localizes to chromosome 16p11.2 [9]. The CLN3 protein is a multipass transmembrane protein with 6 membrane spanning domains, with the N- and C-termini oriented facing the cytosol [10, 11]. The genetics of CLN3 disease and biochemical features of the CLN3 protein have been extensively reviewed elsewhere [2, 12–16]. The purpose of this mini-review is to highlight recent advances delineating molecular pathways in which CLN3 likely functions, and we will discuss how the loss of CLN3 function in these pathways may impact cellular physiology in CLN3 disease.

2. Cellular phenotypes in CLN3-deficiency models

The identification of the CLN3 gene facilitated substantial efforts towards the creation of model systems to investigate CLN3 cell biology and the consequences of loss of CLN3 function on the molecular, cellular and whole organism levels. The application of numerous overexpression, knock-out and targeted mutation strategies to generate disease models has given rise to a diverse collection of tools for this important area of CLN3 disease research. These include lower organism models, such as yeast [17–19], Caenorhabditis elegans [20], Dictyostelium discoideum [21], Drosophila melanogaster [22], and zebrafish [23], as well as mammalian models, such as mice [24–28] and a mini pig model that is in development [29]. Reprogramming and gene editing technologies have also led to the establishment of a number of human in vitro stem cell model systems [30–36]. For a selection of comprehensive reviews covering models for CLN3 disease research, see [37–42].

The CLN3 protein has been reported to localize to numerous subcellular sites including endoplasmic reticulum (ER), Golgi, mitochondria, plasma membrane, endosomes and lysosomes [43–57]. Some of the reported localization results may have been influenced by systems that utilized overexpressed and/or fusion tags on the CLN3 protein or were influenced by non-specific antibodies, as discussed in [13, 58]. Nevertheless, an overwhelming majority of those studies support that CLN3 traffics through the secretory pathway following its synthesis, ultimately being targeted to endosomes and lysosomes possibly via the plasma membrane.

Consistent with the trafficking route of CLN3 and its primary localization to endosomes and lysosomes, a robust body of evidence has documented abnormalities in the secretory, plasma membrane and endolysosomal systems in CLN3 deficiency models. These include alterations in lysosomal pH [59–61], Ca²⁺ homeostasis and signaling [21, 62–65], defects in osmoregulation [66–68], changes in lipid microdomain properties [69, 70], altered endocytosis and trafficking of cell surface receptors [52, 71–74], disrupted sorting of lysosomal-targeted proteins [52, 56, 75], and defects in autophagy [76–79]. In the following sections, we will discuss how these abnormalities may be explained by an emerging picture regarding CLN3 protein interactions within its trafficking pathway.

3. Recent insights into pathomechanisms from protein-protein interaction (PPI) studies

CLN3 regulation of ion channels/transporters

Ion channels and transporters are integral membrane proteins that play roles in a wide array of biological activities, such as regulation of vascular function, cardiac function, renal function, and synaptic transmission [80]. The well-documented role that ion channel and ion transporter dysfunction plays in human diseases, such as epilepsy [81], cardiovascular disease [82, 83], kidney disease [84, 85], and neurodegenerative disease [86–89], underscores their importance in human biology.

Ion flux plays a critical role in endolysosomal regulation and function [90]. As noted above, a number of studies have demonstrated that loss of CLN3 leads to ion imbalance. This includes abnormal lysosomal pH regulation [60, 61, 91] (although normal lysosomal pH has also been reported in some CLN3 disease cell models [92]), altered Ca²⁺ homeostasis and signaling [21, 62–65], and altered biometals homeostasis [93]. While it has not been demonstrated that CLN3 directly functions to transport ions, several possible molecular mechanisms by which CLN3 may impact ion transport and signaling have emerged from PPI studies.

CLN3’s role in calsenilin-mediated modulation of K+ channels and Ca²⁺ dependent cell death

One of the first proteins shown to interact with CLN3 was calsenilin [94]. Calsenilin, also named KChIP3 (K⁺ Channel Interacting Protein) and DREAM, is a neuronal Ca²⁺ sensor that binds and regulates ion channels, including Kv4 channels, voltage dependent Ca²⁺ channels, and the Inositol trisphosphate 3 receptor (IP3R) (reviewed in [95]). Calsenilin also binds presenilins 1 and 2 to enhance gamma secretase activity and programmed cell death [96, 97], and in the nucleus binds downstream regulatory element (DRE) containing genes (hence the pseudonym DREAM) to repress transcription [98]. In a search for interactors of calsenilin using a classical yeast two-hybrid approach, Chang et al., pulled out a fragment of CLN3 and later validated the interaction at the endogenous level in SH-SY5Y human neuroblastoma cells, also further mapping the interaction to the C-terminus of CLN3 [94]. The interaction of CLN3 with calsenilin was Ca²⁺ dependent, with a reduced interaction in the presence of Ca²⁺. Moreover, Chang et al suggested that the interaction of CLN3 and calsenilin conferred anti-apoptotic properties to SH-SY5Y cells because in the absence of CLN3, these cells were more sensitive to Ca²⁺-dependent cell death mediated by calsenilin, and overexpression of CLN3 protected against Ca²⁺-dependent cell death [94].

More recently, Seifert et al confirmed the ability of CLN3 to exert an effect on calsenilin-mediated processes. In this 2020 study, the authors demonstrated that CLN3 could modulate calsenilin/KChIP3-mediated Kv4.2 function [99]. In HEK293 cells, overexpression of CLN3 reduced the amount of calsenilin/KChIP3 bound to the Kv4.2 channel. Furthermore, in a co-expression paradigm where co-expression of calsenilin/KChIP3 and Kv4.2 in HEK293 cells modulates Kv4.2 current properties, the addition of CLN3 co-expression dampened the effect of calsenilin/KChIP3 on the Kv4.2 current properties. Seifert et al. speculated that CLN3, by virtue of its binding interaction with calsenilin/KChIP3, interferes with KChIP3 association with Kv4.2 channels to negatively regulate their function [99]. Furthermore, this effect was Ca²⁺ dependent, whereby the KChIP3-Kv4.2 interaction and KChIP3-induced effects on Kv4.2 channel properties were stabilized by the presence of Ca²⁺. The demonstration that a disease-associated missense variant and a C-terminal deletion version of CLN3 had reduced or no effect on KChIP3-mediated modulation of Kv4.2 suggested this modulation of Kv4.2 channels may have some disease relevance. However, an important caveat of these studies was that Seifert et al did not observe a direct interaction between CLN3 and calsenilin/KChIP3 [99], in contrast to the findings of Chang et al. This discrepancy was most likely due to the use of differently tagged versions of the overexpressed proteins, as discussed by Seifert et al in their report [99]. Most notably Seifert et al used GFP-tagged CLN3 in their co-immunoprecipitation analysis, while Chang et al used untagged CLN3 [94]. It is well documented that tags can sometimes interfere with efficient CLN3 folding and trafficking [16, 46].

Taken together the findings from Chang et al and Seifert et al suggest that CLN3 functions as a direct modulator of Ca²⁺-dependent calsenilin function in both ion channel regulation and Ca²⁺-dependent cell death (Figure 1). In addition to the need for further delineation of the precise molecular mechanisms by which CLN3 modulates calsenilin-mediated function, many important questions remain to be answered surrounding these key discoveries in the study of CLN3 protein function: Does CLN3 also modulate other calsenilin-mediated processes, such as presenilin/γ-secretase cleavage of amyloid precursor protein (APP) and nuclear translocation and transcriptional repression? What is the role of the calsenilin activities that CLN3 modulates in the CLN3 neurodegenerative disease process? Notably, overexpression of wildtype CLN3 (but not disease-associated mutant CLN3) has been reported to interfere with gamma-secretase mediated cleavage of amyloid precursor protein (APP) in HEK293 cells, which is a presenilin-dependent process [100, 101]. Notch2-associated changes have also been reported in the cerebellum of Cln3-deficient mice [102], and Notch signaling was inhibited in a CLN3 overexpression Drosophila model [22]; presenilin is a key regulator of notch signaling [103, 104]. These observations indicate investigation of a potential role for CLN3 in calsenilin-mediated presenilin function is warranted.

Figure 1. Hypothetical model depicting CLN3-regulation of calsenilin function.

Under low Ca²⁺ conditions, CLN3 can bind calsenilin which is protective from Ca²⁺-induced cell death and calsenilin regulation of Kv4.2. In high Ca²⁺ conditions, calsenilin binds Ca²⁺ and dissociates from CLN3, where it is then able to interact with other proteins like Kv4.2 to influence their trafficking and/or function. A pool of CLN3 localized at the plasma membrane (PM) (not shown) may alternatively be involved in Kv4.2 channel regulation. Cal=calsenilin, Lys=lysosome

CLN3 association with plasma membrane Na,K-ATPase and the plasma-membrane associated cytoskeletal protein, β-fodrin

In another yeast 2-hybrid screening approach, using an N-terminus CLN3 bait, an interaction with the plasma membrane-associated Na,K-ATPase (both the ß1 and ß2 isoforms) was uncovered [105]. In this study, co-immunoprecipitation of overexpressed full length CLN3 validated the novel yeast 2-hybrid interaction. Interestingly, the cytoskeletal protein ß-fodrin (or ß-II-spectrin) and GRP78/BiP also co-immunoprecipitated with the CLN3-Na,K-ATPase complex [105]; both ß-fodrin and GRP78 are previously known interactors of Na,K-ATPase, and ß-fodrin was also identified as a candidate interactor in the CLN3 screen [105]. Na,K-ATPase ion transport function is important for maintaining membrane potential and for regulation of Ca²⁺ homeostasis. A function beyond ion transport has also been more recently established; Na,K-ATPase also plays a key role as a cellular signaling receptor in a number of important signaling pathways including Src and MEK signal transduction pathways [106, 107].

Uusi-Rauva and colleagues examined the plasmalemmal pumping activity of the Na,K-ATPase in CLN3-deficient primary mouse neurons and saw no appreciable difference compared to wildtype primary mouse neurons [105]. However, a difference in the relative amount of plasma membrane associated Na,K-ATPase and in its rate of endocytosis following ouabain treatment (which blocks the activity of the Na,K-ATPase pump) significantly differed in the wild-type versus CLN3-deficient primary mouse neurons. As a result, the authors speculated that CLN3 may modulate endocytosis of Na,K-ATPase through regulation of the fodrin cytoskeleton, a process that may also involve Grp78/BiP. Consistent with this hypothesis was an altered fodrin cytoskeleton in Cln3 knockout mouse brain and in CLN3 patient fibroblasts [105]. Grp78/BiP, a well-known chaperone of the endoplasmic reticulum (ER)[108, 109], has also been found to be associated with the plasma membrane [110], and a potential role for Grp78/BiP in oubain-induced endocytosis of the Na,K-ATPase has been suggested [111].

Role of CLN3 in regulating the actin cytoskeleton

In addition to the interaction of CLN3 with ß-fodrin already discussed here, CLN3 has also been reported to interact with the cytoskeletal associated protein myosin-II-b [112]. Using a modified yeast 2-hybrid approach (CytoTrap yeast 2-hybrid) and a C-terminal CLN3 bait, Getty et al identified an interaction with myosin-II-b that was verified with myc-tagged overexpressed CLN3 in NIH/3T3 fibroblasts [112]. Myosin-II-b is one of three different isoforms of non-muscle myosin heavy chain, which is an important regulator of the actin cytoskeleton, functioning in cell polarity, cell migration, synaptogenesis and synaptic plasticity (reviewed in [113]); myosin-II-b is the main non-muscle myosin II isoform in neuronal synapses [114].

A functional interaction of CLN3 and myosin-II-b was further supported by observations in Cln3 knock-out mouse embryonic fibroblasts (MEFs), where wound healing was significantly compromised compared to wild-type MEFs, similar to the effect seen when wild-type MEFs were treated with the non-muscle myosin inhibitor, blebbistatin [112]. Moreover, blebbistatin treatment did not alter the wound healing in the Cln3 knock-out MEFs. Interestingly, the authors further analyzed the effect of chloroquine treatment on wound healing given CLN3 is primarily an endolysosomal protein. Disruption of lysosomal function by chloroquine significantly delayed cell migration following scratch wounding of the monolayer of wild-type MEFs, while it completely ablated cell migration following scratch wounding of the monolayer of Cln3 knock-out MEFs, indicating that lysosomal activity and CLN3 were required for proper MEF cell migration in the wound healing assay [112].

Thus, CLN3 may serve as a tether between the endolysosomal membrane and the actin cytoskeleton by binding interaction with ß-fodrin and myosin-II-b [105, 112]. Notably, disruption of the actin cytoskeleton and defects in cell migration in CLN3 deficiency models have also been reported by other groups, lending additional support to this possible role for CLN3 [65, 72].

Owing to its involvement in synaptic function and plasticity, non-muscle myosin II plays a role in a wide range of neurologic disorders including neurodevelopmental disorders and neurodegenerative disease (reviewed in [113, 115]). Non-muscle myosin II is also expressed in glia where it plays a role in neuroinflammation and myelination (reviewed in [115]). Hence, the signaling pathways that modulate non-muscle myosin II function are under investigation for their potential as therapeutic targets for neurologic disease, including PAK and ROCK inhibitors (reviewed in [113, 116–118]. Notably, in a proteomics analysis of presynaptic-enriched synaptosomes isolated from Cln3 knock-out mouse brain, the “Signaling by Rho family GTPases” pathway in IPA analysis was significantly upregulated as compared to wild-type control synaptosomes, which included the ROCK2 protein [119]. Hypothesizing this may reflect synaptic degeneration, the authors analyzed whether ROCK2 inhibition by RNAi knockdown or pharmacological inhibition using fasudil would alter CLN3-induced neurodegeneration in a Drosophila eye model. Crossing CLN3-overexpressing flies, which display a small eye phenotype due to degeneration of photoreceptors [22], with two independent ROCK2 ortholog RNAi fly lines significantly rescued the CLN3-induced small eye degenerative phenotype [119]. Treatment with ROCK inhibitor, fasudil, was less effective but also significantly dampened the CLN3 overexpression induced degeneration [119]. While further studies are required to determine the potential for targeting ROCK signaling in CLN3 disease, these findings nevertheless provide additional support for a CLN3 role in mediating actin cytoskeleton dynamics.

Role of CLN3 in endocytosis

Several studies have linked CLN3 to various forms of endocytosis [52, 69, 71, 72]. Endocytosis is an important cellular process that can internalize molecules, down-regulate cell surface proteins and modulate cell growth and movement [120, 121]. Three major subtypes of endocytosis have been identified; clathrin-mediated endocytosis (CME), clathrin-independent endocytosis (CIE), and macropinocytosis (fluid-phase endocytosis). CME is the most well studied, and as the name implies, requires the formation of clathrin-coated vesicles at the plasma membrane (PM) [122, 123]. CIE is mediated by a variety of mechanism including caveolae [124]. The formation of caveolae requires lipids such as cholesterol and sphingolipids, along with proteins called caveolins [125]. In cells lacking CLN3, caveolae dependent endocytosis was defective [69]. This was most likely due to the lack of proper sorting of caveolin from the Golgi to the PM, highlighting a potential role of CLN3 at the Golgi.

Although actin is required for all forms of endocytosis, it plays a particularly important role in macropinocytosis (Reviewed in [126]). In this form of endocytosis, actin polymerization results in the formation of membrane ruffles which form large vacuoles, leading to the internalization of extracellular fluids and molecules [127]. CDC42, a member of the Rho family of small GTPases, is a regulator of actin polymerization, and plays crucial roles in both CIE and macropinocytosis [128, 129]. During macropinocytosis, GTP-loaded CDCD42 is recruited to specific regions of the PM via an interaction with phosphatidylinositol (3,4,5)-trisphosphate (PI(3,4,5)P3), which drives actin polymerization and the formation of filopodia near the forming macropinocytosis invagination [130]. In order for CDC42 to cycle to a GDP bound form, the GAP ARHGAP21 (also known as ARHGAP10) is recruited to the site. This recruitment is mediated by the small GTPase Arf1 [131]. It is important to point out that CDC42 must cycle between GDP and GTP bound states in order to function. Although depletion of ARHGAP21 results in more GTP-loaded CDC42 with increased actin polymerization, it also leads to reduced macropinocytosis [131]. CLN3 deficient murine brain microvascular endothelial cells (MBEC) had defects in macropinocytosis. This was due to an elevated amount of GTP-loaded CDC42 [72]. Further work demonstrated that the increase in GTP-loaded CDC42 was due to a decrease in PM localized ARHGAP21. Furthermore, it was demonstrated that ARF1 GTP loading was reduced in CLN3 deficient MBEC cells [72]. As Arf1 is required to recruit ARHGAP21 to the PM, this explains the observed increase in GTP loaded CDC42. How CLN3 regulates Arf1 GTP loading is yet to be determined. One possibility is that CLN3 is required to recruit the GEF responsible for Arf1 activation at the PM, although the identity of that protein remains to be elucidated. Alternatively, CLN3 may act as a scaffold, regulating the spatiotemporal interaction of Arf1 recruited ARHGAP21 with CDC42.

Role of CLN3 in regulating Rab GTPases

Small GTPases are small molecular switches that regulate various cellular functions. Five families of small GTPase are expressed in mammalian cells; Ras, Rho, Rab, Arf and Ran (Reviewed in [132]). Each family is associated with specific cellular functions including cell growth, differentiation and survival (Ras), cytoskeleton organization and vesicular trafficking (Rho), regulation of vesicular trafficking (Rab and Arf), and control of nucleocytoplasmic transport (Ran). Small GTPases function by interacting with downstream effectors proteins. A key regulation of effector protein binding is accomplished by turning on the small GTPase by binding to GTP. Activation is modulated by members of the guanine nucleotide exchange factor (GEF), which loads GTP onto the small GTPase. In the majority of cases, activated small GTPases are membrane bound. GTP loaded small GTPases can then bind an effector protein to mediate function. On the other hand, GTPase-activating proteins (GAPs) can facilitate the hydrolization of GTP to GDP, which turns the small GTPase off, and returns it to the cytosol (Reviewed in [133]).

At the trans-Golgi Network (TGN), the lysosomal sorting receptors (LSRs) are responsible for the sorting and trafficking of soluble lysosomal resident proteins (cargo) such as cathepsin D and prosaposin to lysosomes [134, 135]. This is accomplished via the interaction of the cytosolic tail of the receptor with adaptor protein-1 (AP-1), which subsequently binds clathrin (Figure 2). The spatiotemporal recruitment of AP-1 to the membranes of the TGN is regulated by the small GTPase Arf1, which both stabilizes and opens AP-1 for membrane binding and interaction with the LSR. The monomeric clathrin adaptors, the Golgi-localized, gamma adaptin ear-containing, ARF-binding proteins (GGAs), have also been implicated in this process. CLN3 has been implicated in membrane trafficking, but the molecular details have not been well studied [19, 52, 56]. In cells depleted of CLN3, export from the TGN of CI-MPR was significantly delayed, resulting in defective lysosomal enzyme function [56]. Further supporting a role for CLN3 in sorting soluble lysosomal cargo proteins, an unbiased screen using a proteomics approach demonstrated reduced levels of several soluble lysosomal proteins in cells lacking functional CLN3 [92]. The mechanism behind the delayed exit of CI-MPR from the TGN has not been elucidated. CLN3 can interact with Rab1a, a small GTPase localized to the Golgi apparatus [136]. Rab1a is implicated in the recruitment of Golgi-specific brefeldin A-resistance guanine nucleotide exchange factor 1 (GBF1), a GEF for Arf1 [137]. The recruitment of GBF1 leads to the recruitment of the GGA family of adaptors [138]. Alternatively, through a GEF cascade with Brefeldin A-inhibited guanine nucleotide-exchange protein 1 (BIG1), the recruitment of GFB1 can also recruit AP-1. Further investigation will be required to determine the molecular role of CLN3 in sorting at the TGN.

Figure 2. Schematic representation of cellular pathways modulated by CLN3.

CLN3 can regulate Golgi export by potentially regulating AP-1 recruitment, receptor retrieval from late endosomes with retromer, and macropinocytosis by regulating the actin cytoskeleton.

Of the small GTPase family, the Rab family comprises the largest number, with 60 encoded in the mammalian genome (Reviewed in [139]). Rab GTPases play key roles in the formation, movement and tethering of trafficking vesicles at various cellular locations. CLN3 was shown to interact with Rab7A, a small GTPase localized to late endosomes [136, 140]. Rab7A performs many key functions including regulating endosome-to-Trans Golgi network (TGN) trafficking, the degradation of endocytic cargo and the fusion of endolysosomes with autophagosomes [141–144]. These various functions require that Rab7A interact with various effector proteins. Modulation of effector binding is at least partially regulated by the GEF Mon1/Ccz1, which is thought to activate Rab7A resulting in the membrane recruitment of Rab7A to late endosomes where it can interact with an effector protein [145, 146]. Termination of effector binding is mediated by the GAPs TBC1D2 [147], TBC1D5 [143, 148] and TBC1D15 [149, 150], which are able to shut off Rab7A, returning it to the cytosol.

Retromer is a cytosolic trimeric protein complex composed of Vacuolar protein sorting (VPS) 26, VPS29 and VPS35 [151, 152]. Dysfunction of retromer has been implicated in a variety of neurodegenerative diseases including Alzheimer’s disease [153], Parkinson’s disease [73, 154] and Amyotrophic lateral sclerosis (ALS) [155]. More recent work has shown that retromer dysfunction also occurs in various forms of NCL, including CLN3 and CLN5 [136, 156]. Retromer is recruited to late endosomes, where it interacts with the cytosolic tail of CI-MPR and sortilin [151, 157] (Figure 3). This interaction is required to sort the LSRs out of the late endosome, and return them to the TGN [151, 152]. In cells lacking functional retromer, the LSRs remain trapped in late endosomes and are eventually degraded in lysosomes. GTP-loaded Rab7A is required to recruit retromer to late endosomes, a step necessary for efficient retromer binding to the LSRs [142, 143]. In HeLa cells lacking CLN3, the Rab7A/retromer and retromer/sortilin interactions are significantly weakened [136]. This is not due to lack of activation of Rab7A, as it is still membrane bound, suggesting that its GTP loading is occurring. CLN3 appears to function as a scaffold protein, ensuring efficient interactions between Rab7A with retromer, and retromer with sortilin [136]. It is noteworthy to point out that other NCL proteins, including CLN6, CLN7 and CLN8, are also required to efficiently sort soluble lysosomal cargo proteins, suggesting a possible common disease mechanism across multiple forms of NCL [158–160].

Figure 3. Retrieval of the lysosomal sorting receptors requires CLN3.

CLN3 coordinates the efficient interactions between Rab7, retromer and the lysosomal sorting receptor, sortilin at late endosomes. In cells lacking functional CLN3, these interactions are deficient, resulting in the lysosomal degradation of the receptors

Autophagy is known to be dysfunctional in various CLN3 model systems [76, 78, 79]. Lack of autophagosome/lysosome fusion has previously been observed in CLN3-deficient mouse cerebellar cells [63]. The lack of autophagosome/lysosome fusion could be due to dysfunctional fusion machinery, or due to lack of movement of organelles favouring fusion. Rab7A can regulate both these processes by interacting with different effector proteins. Pleckstrin homology domain-containing family M member 1 (PLEKHM1) is a Rab7A effector that functions as a tethering factor, mediating the fusion of late endosomes with lysosomes and autophagosomes with lysosomes [161]. Rab-interacting lysosomal protein (RILP) is also a Rab7A effector and plays a role in both fusion and minus end transport of lysosomes and autophagosomes [162–164]. In cells lacking CLN3, the Rab7A/PLEKHM1 interaction is disrupted, resulting is significant delays in EGFR degradation. However, the Rab7A/RILP interaction was not affected [136]. Although dysfunction in retromer activity resulting in defective lysosomal function could affect autophagy, lack of lysosome/autophagosome fusion in CLN3-deficient cells, due to decreased PLEKHM1 function, could as well.

The functional regulation of small GTPases by CLN3 appears to be emerging as a significant role for this integral membrane protein. It will be interesting to determine if CLN3 regulates other small GTPases implicated in other cellular pathways. Does CLN3 function mainly as a scaffold protein, or does it regulate signal via other mechanisms? Further work will be required to determine its roles.

CLN3 in lysosomal positioning

In various models of CLN3 disease, the localization of lysosomes within cells is affected. In HeLa cells overexpressing CLN3 bearing a disease-causing mutation, CLN3^E295K, lysosomes were more compact, and localized towards the perinuclear region. This suggested a role for CLN3 in moving lysosomes in a plus end direction, towards the periphery [140]. In order to determine how CLN3 could be implicated in organelle positioning, a co-immunoprecipitation strategy was used. CLN3 was found to interact with RILP, the dynactin subunit p150^Glued, dynein and the kinesin-2 subunit KIF3A [140]. The first three proteins form a complex required for minus-end transport (towards the peri nuclear space) of lysosomes and autophagosomes, while kinesin-2 has been shown to function in plus-end transport (transport towards the periphery) of lysosomes [162–165]. On the other hand, mouse cerebellar cells harbouring the most common disease causing mutation, a ~1-kb deletion (Cln3^Δex7/8 cerebellar cells), lysosomes were more dispersed to the cell periphery [52], a finding consistent with the observation that lysosomes were more distal from the cell bodies in CLN3-deficient iPSC-derived neurons from CLN3 patients [166]. How can these seemingly opposing results be explained? In the HeLa cell experiments, wild-type and mutant CLN3 was expressed in cells containing endogenous CLN3. On the other hand, the mouse cerebellar cells and CLN3-patient iPSC-derived neurons lack wild-type CLN3 and could only express mutant CLN3 from the endogenous locus, and most likely at much reduced levels due to non-sense mediated decay of the mutant Cln3/CLN3 mRNA [167]. Additionally, the experiments involved systems with different CLN3 mutations. It is possible that the mutations affect interactions with motor proteins differently. As the plus and minus end motors often co-purify with lysosomes together, the different CLN3 mutations could affect these interactions differently[168]. Finally, the observed differences could be due to the nature of the cells themselves. Organelle position is much more crucial and regulated in neurons compared to other cell types.

4. Molecular discoveries shed light on possible impacts on differing cell types

Function of CLN3 in cells of the CNS: Glial Cells

The CNS contains four types of glial cells, microglia, astrocytes, oligodendrocytes and progenitor cells. These cells were once thought to function as a glue, keeping the brain together, but have been shown to play major roles in the CNS and have been implicated in a variety of diseases.

Microglia are the resident immune cells of the brain and perform several functions including cytokine release and phagocytosis [169]. Microglia dysfunction has been associated with neurodegenerative diseases including Alzheimer’s disease, and it has been reported that microglia can clear Aβ fragments via macropinocytosis [170], Microglia dysfunction has also been reported in rare lysosomal diseases such as Niemann-Pick type C and CLN3 disease [171, 172]. How could defects in CLN3 function affect these microglia processes? As these cells play important roles in phagocytosis and macropinocytosis, defects in CLN3 function could severely affect these processes. Although phagocytosis in microglia has not been shown to be affected in models of CLN3 disease, this process is affected in other cell types such as retinal pigment epithelium cells [36]. Neuroinflammation occurs in several neurodegenerative diseases, including CLN3 disease [173]. Microglia play an important role in mediating inflammation through many mechanisms, although the most studied is the NLRP3 inflammasome complex. This complex senses external and internal stimuli, which leads to the activation of caspase-1, resulting in the activation of pro-inflammatory cytokines and rapid cell death [174]. This complex can be degraded via autophagy, which enables cells to down regulate inflammation. Indeed, following activation, components of the inflammasome are ubiquitinated and interact with p62 [174]. It has been shown that microglia isolated from CLN3-deficient mice are more primed towards inflammasome activation and caspase-1 activation, compared to wild-type microglia [171]. Since lysosomes in CLN3-deficient cells are dysfunctional affecting autophagy, inflammation associated with CLN3 disease could be due to prolonged inflammasome signaling. In CLN3 disease, two possible functions of microglia could be impaired, resulting in neuroinflammation.

Astrocytes are the most abundant glia cells in the CNS and perform a variety of functions including ion homeostasis [175]. Along with microglia, astrocytes play an important role in maintaining neuronal health, and their dysfunction has been associated with lysosomal diseases [176]. Astrocytosis and morphological changes in astrocytic end feet have been reported in CLN3-deficient mouse brain [177–179]. A recent study also found significant impairments in the function of astrocytes derived from the brain of CLN3-deficient mice. In this study, the authors demonstrated that these astrocytes had defects in their actin cytoskeleton, and impairments in their calcium signaling abilities, roles associated with CLN3 function [65]. An additional study also reported abnormalities in calcium signaling and metabolism in astrocytes isolated from CLN3-deficient mice [64]. Furthermore, using a co-culture system of mouse-derived astrocytes and cortical neurons, Parviainen et al showed that astrocytes derived from CLN3-deficient mice had a negative impact on the survival and shape of neurons from both wild-type and CLN3-deficient mice [65]. This points to the role of glia, and particularly astrocytes, in the neurodegeneration observed in CLN3 disease. From a molecular view, the role of CLN3 in modelling the actin cytoskeleton and its role in modulating ion channels and lysosomal targeting could explain the phenotypes observed in the mouse studies.

Neurons

Neurons are polarized cells that contain a cell body (soma), dendrites and an axon. In certain cases, the axon can extend up to 1 meter in length. Each region contains a subset of organelles. The sorting of organelles into these distinct regions is highly regulated, as is the transport of organelles in axons. For example, lysosomes are positioned throughout the soma and along axons. However, those lysosomes are not all the same. Lysosomes in the soma are more acidic, contain more degradative enzymes, and have the ability to degrade material much more that more peripheral lysosomes [180]. CLN3 could play a role in this polarized sorting by favouring minus or plus end transport of lysosomes in neurons. Consistent with this hypothesis, overexpressed CLN3 was more abundantly associated with retrograde moving, Rab7-positive vesicles (versus anterograde moving vesicles) in primary hippocampal mouse neurons [57]. As already discussed previously, CLN3 binds to both plus end and minus end motors, and in mouse cerebellar cells harbouring a disease-causing mutation of CLN3, and in iPSC-derived neurons from CLN3 patients, lysosomal positioning was affected [52, 166]. In neurons, autophagosome formation occurs in the distal axons, which are then transported in a minus-end direction in order to fuse to lysosomes [181]. Rab7A has been implicated in the movement of endolysosomes in both directions and is required for autophagosome/lysosome fusion [161, 182]. Combined with potential defects in lysosomal positioning, and autophagosome/lysosome fusion, and the need to move proteins and organelles over greater distances, neurons could be more affected by CLN3 mutations than other cell types. CLN3 has also been shown to localize to vesicles in neuronal synaptic regions, where it may play a role in regulation of synaptic activity [183], consistent with a growing number of reports documenting alterations in neurodevelopment and/or neurotransmission in CLN3 disease models [34, 119, 184–188]. Thus, CLN3-deficient neurons may be especially vulnerable to dysfunctioning lysosomal and synaptic pathways, which, combined with functional defects in glial cells, may explain the prominent impact CLN3 disease has on nervous system function.

Function of CLN3 in cells outside of the CNS

Although CLN3 disease primarily affects nervous system function, the CLN3 gene is ubiquitously expressed, and indeed lysosomal storage pathology is found in many cell types, including those found in peripheral organ systems [1, 9]. In fact, an impact on cardiac function has more recently been recognized as a clinically significant component of CLN3 disease, particularly in later disease stages [189, 190]. Intriguingly, relatively high levels of CLN3 expression (compared to brain) are seen in the muscle, kidney and digestive system in multiple organisms [9, 20, 57, 66].

Epithelial cells are of particular interest for understanding CLN3 function, as a number of studies have shown abnormalities in epithelial cells in various organs due to CLN3 deficiency. These include the retinal pigment epithelium, principal cells of the kidney, and clear cells of the epidydimis of the male reproductive tract [36, 66, 191]. An important functional role for CLN3 in epithelial cells, which, like neurons, are a polarized cell type, would be consistent with CLN3’s role in regulation of phagocytosis and endocytosis as well as in osmoregulation, membrane organization and regulation of the actin cytoskeleton, all of which may make epithelial cells vulnerable to CLN3 deficiency. Consistent with this notion, it has recently been suggested that a reduction in phagocytosis of photoreceptor outer segments by the retinal pigment epithelium may significantly contribute to the vision loss in CLN3 disease [36].

Vascular endothelial cells have also been shown to express high levels of CLN3 in mouse studies [27, 69]. In the brain, these cells are a central component of the blood-brain-barrier, which is compromised in Cln3 loss of function mouse models [69, 72, 192]. As discussed previously in this review, MBEC cells lacking CLN3 displayed defects in micropinocytosis, as well as defects in membrane microdomain fluidity and an abnormal response to osmotic stress, which were at least partly attributed to misregulated Arf1-CDC42 signaling and actin-driven processes [69, 72].

Further work to better understand the impact of CLN3 deficiency in cells outside of the CNS, together with ongoing studies aimed at establishing a more comprehensive CLN3 disease natural history will contribute important knowledge towards a broader understanding of CLN3 function and CLN3 disease pathogenesis.

5. Conclusions

Through a number of protein-protein interaction (PPIs) studies using different techniques (summarized in Table 1), the interactome of CLN3 has provided clues to its function. Emerging themes of CLN3 function include the regulation of cytoskeletal/cytoskeletal associated proteins (actin cytoskeleton and dynein motor-RILP) to tether cellular membranes, regulation of membrane complexes such as channels/transporters, and modulating the function of small GTPases to effectively mediate vesicular movement and membrane dynamics (such as fission, fusion, chemotaxis/cell migration). As technology develops, we have access to novel tools to learn more about the function of CLN3. For instance, proximity labelling techniques such as BioID (proximity-dependent biotin identification) and APEX (engineered ascorbate peroxidase) have been developed that have several advantages over traditional methods including detection of interactions with proteins expressed in their native environments, in live cells, and detection of transient interactions [193]. By engineering the tag into the endogenous locus using CRISPR/Cas9 technology, CLN3 interactomes using these methods could detect PPIs without over expressing CLN3, and with CLN3 localized to the membrane. This could result in new interacting partners missed in previous unbiased screens and identify new roles for CLN3. Further studies focusing on other interacting partners already identified, or to be identified, could shed further light on CLN3 function. A more robust understanding of CLN3, its interacting partners and function should lead to the identification of novel therapeutic targets, and possible treatments for children affected by CLN3 disease.

Table 1. Summary of CLN3 interactions.

The table shows known interacting partners of CLN3, and the method that demonstrated the interaction. For information on yeast 2-hybrid, please see [195]. Co-immunoprecipitation (co-IP). Bioluminescence resonance energy transfer (BRET) is a method to detect protein-protein interactions in live cells, in real time and with proteins expressed in their native environment. For more information, please refer to [196]. Several additional reports of hits in CLN3 PPI screens or of fragment binding interactions are not included in the table, given the limited validation done on those hits (see references [71, 197, 198]).

Interacting Protein	Method	Reference
Calsenilin	Yeast 2-Hybrid, co-IP	Chang et al., 2007 [94]
Na,K-ATPase	Yeast 2-Hybrid, co-IP	Uusi-Rauva et al., 2008 [105]
β-Fodrin	Yeast 2-Hybrid, co-IP	Uusi-Rauva et al., 2008 [105]
GRP78/BiP	co-IP	Uusi-Rauva et al., 2008 [105]
myosin-II-b	Cytotrap Yeast 2-Hybrid, co-IP	Getty et al., 2011 [112]
SBDS	Cytotrap Yeast 2-Hybrid, co-IP	Vitiello et al., 2010 [194]
Rab1A	BRET	Yasa et al., 2020 [136]
Vps26A	BRET, co-IP	Yasa et al., 2020 [136]
Rab7A	co-IP, GST pulldown BRET, co-IP	Uusi-Rauva et al., 2012 [140] Yasa et al., 2020 [136]
sortilin	BRET, co-IP	Yasa et al., 2020 [136]
RILP	co-IP	Uusi-Rauva et al., 2012 [140]
p150Glued (Dynactin subunit)	co-IP	Uusi-Rauva et al., 2012 [140]
Dynein	co-IP	Uusi-Rauva et al., 2012 [140]
KIF3A (Kinesin-2 subunit)	co-IP	Uusi-Rauva et al., 2012 [140]

SBDS=Shwachman-Bodian-Diamond syndrome protein

Highlights.

• Autosomal recessive mutations in CLN3 lead to CLN3 disease

• CLN3 disease is a fatal neurodegenerative disease that primarily affects children

• CLN3 is a membrane protein localized to the secretory and endolysosomal pathways

• Loss of CLN3 function leads to defective autophagy and lysosomal function

• CLN3 is implicated in regulating the cytoskeleton, ion channels, and trafficking