ABSTRACT
Most secretory proteins travel through a well-documented conventional secretion pathway involving the endoplasmic reticulum (ER) and the Golgi complex. However, recently, it has been shown that a significant number of proteins reach the plasma membrane or extracellular space via unconventional routes. Unconventional protein secretion (UPS) can be divided into two types: (i) the extracellular secretion of cytosolic proteins that do not bear a signal peptide (i.e. leaderless proteins) and (ii) the cell-surface trafficking of signal-peptide-containing transmembrane proteins via a route that bypasses the Golgi. Understanding the UPS pathways is not only important for elucidating the mechanisms of intracellular trafficking pathways but also has important ramifications for human health, because many of the proteins that are unconventionally secreted by mammalian cells and microorganisms are associated with human diseases, ranging from common inflammatory diseases to the lethal genetic disease of cystic fibrosis. Therefore, it is timely and appropriate to summarize and analyze the mechanisms of UPS involvement in disease pathogenesis, as they may be of use for the development of new therapeutic approaches. In this Review, we discuss the intracellular trafficking pathways of UPS cargos, particularly those related to human diseases. We also outline the disease mechanisms and the therapeutic potentials of new strategies for treating UPS-associated diseases.
Introduction
According to the classic principle of protein secretion, cargo proteins travel by using the conventional pathway from the endoplasmic reticulum (ER) to the Golgi complex, from which they subsequently move to the trans-Golgi network (TGN) and finally to the plasma membrane (Lee et al., 2004). This process is initiated by recognition of a signal peptide (also known as ‘leader sequence’) at the N-terminus or transmembrane domain of cargo proteins, followed by sequential budding and fusion of vesicular carriers (Bonifacino and Glick, 2004). Each step of the secretion pathway is under the control of a number of regulatory proteins. Correct regulation of the classic secretory pathway is imperative for the life and health of the organism (Viotti, 2016). Since the discovery of vesicular exocytic mechanisms, this classic protein secretion pathway involving ER-to-Golgi transport has been considered the only standard mechanism to move proteins out of the cell. However, discoveries over the last two decades have shown that an increasing number of proteins use alternative secretory pathways that do not involve the ER-to-Golgi transport (Malhotra, 2013; Ponpuak et al., 2015; Rabouille, 2017). These alternative pathways include the extracellular secretion of cytosolic proteins that do not bear a signal peptide (i.e. leaderless proteins) (Rubartelli, 1997), and cell-surface trafficking of transmembrane proteins via a Golgi-bypassing route. These pathways are collectively referred to as unconventional protein secretion (UPS) (see Box 1).
Most secretory proteins reach their destination via the ER–Golgi-target organelle route, which is referred to as the ‘classic’ or ‘conventional’ protein secretion pathway. These secretory proteins contain a signal peptide (the ‘leader sequence’) that directs their translocation into the lumen or to the ER membrane. The newly synthesized proteins then exit the ER at an ER-exit site (ERES) through coat protein complex II (COPII)-coated vesicles and, so, reach the Golgi network before being dispatched to the plasma membrane, lysosomes, endosomes or peroxisomes (Gee et al., 2018; Viotti, 2016). Fusion of vesicular intermediates and organelles is mediated by soluble N-ethylmaleimide-sensitive factor (NSF) attachment protein (SNAP) receptor proteins (SNAREs), Rab proteins and their regulators (Mellman and Warren, 2000).
In addition to the above-mentioned conventional pathway, eukaryotic cells also utilize unconventional protein secretion (UPS) for protein sorting and delivery. Initially, the term ‘unconventional secretion’ was used for the release of cytoplasmic proteins that lack a signal peptide, i.e. leaderless proteins, into the extracellular medium. Later, it was found that some transmembrane proteins that are synthesized in the ER, reach the plasma membrane via a route that bypasses the Golgi complex (Nickel and Rabouille, 2009). As described in the text, Rabouille colleagues divided UPS into four types, i.e. Type I, II and III UPS of leaderless proteins, and Type IV UPS of Golgi-bypassing transmembrane proteins (Rabouille et al., 2012). However, the ABC transporter-mediated Type II pathway is not well studied and needs additional validation (Rabouille, 2017).
With a few exceptions, most UPS pathways are induced by various cellular stresses, such as nutrient starvation (Cruz-Garcia et al., 2014), mechanical stress (Schotman et al., 2008), inflammation (Schroder and Tschopp, 2010) and ER stress (Gee et al., 2011; Jung et al., 2016). Notably, many disease conditions are associated with various stresses at the cellular or organismal level (Fulda et al., 2010), indicating the potential of UPS as a promising emerging target for the development of novel therapeutics to treat associated human disease.
The number of defined UPS-related diseases continues to expand. For example, sterile inflammation related to Alzheimer's disease, allergic and autoimmune diseases, and diabetes can be a trigger to induce unconventional protein secretion (Agosta et al., 2014; Chen et al., 2015; Freigang et al., 2013; Gardella et al., 2002; Schroder and Tschopp, 2010). Heat shock proteins (HSPs) that are secreted unconventionally play a pivotal role in the immunomodulation, proliferation, angiogenesis and invasiveness of cancer (Rodriguez et al., 2009; Sarikonda et al., 2015; Zhang et al., 2012). A number of autophagy components, the mutations of which are involved in various diseases (Jiang and Mizushima, 2014), participate in UPS of numerous cargo proteins (Duran et al., 2010; Kinseth et al., 2007; Manjithaya et al., 2010). Several mutant transmembrane proteins, whose associated trafficking defects to the cell surface cause inherited genetic disorders, such as cystic fibrosis and congenital hearing loss can, alternatively, reach the plasma membrane by Golgi-bypassing UPS (Gee et al., 2011; Jung et al., 2016). In addition, UPS has even been shown to be essential for microorganisms to mediate their extracellular release and exert biotrophic pathogenicity (Shoji et al., 2014).
Recent developments in the field have contributed to significant advances in the understanding of the molecular processes involved in UPS (Daniels and Brough, 2017; Pompa et al., 2017; Ponpuak et al., 2015; Rabouille, 2017; Santos et al., 2017). Nevertheless, a comprehensive description that adequately explains the role in UPS in disease pathogenesis is currently lacking. In this Review, we will define emerging roles for UPS in disease pathogenesis and highlight the possibility of novel therapeutics that target UPS.
Disease-associated unconventional secretion pathways
Overview and classification
With the increased scientific understanding of the alternative secretion pathways, several researchers have attempted to systematically classify the UPS pathways. For example, Deretic and co-workers have classified autophagy-associated protein secretion pathways (Ponpuak et al., 2015). However, because their review also addresses the overlap between autophagy and conventional protein secretion, we will adopt here the UPS classification proposed by Rabouille and colleagues (Rabouille, 2017; Rabouille et al., 2012). According to this, UPS pathways can be divided into four types with the UPS of leaderless proteins being sub-divided into three. They are: (Type I) direct secretion or pore-mediated translocation across the plasma membrane (Ding et al., 2016; Zacherl et al., 2015), (Type II) ATP-binding cassette (ABC) transporter-based secretion (McGrath and Varshavsky, 1989) and, (Type III) membrane-bound organelle (autophagosome/endosome)-based secretion (Duran et al., 2010; Kinseth et al., 2007). The fourth type is the Type IV UPS of leader-sequence-containing transmembrane proteins, which are synthesized in the ER and reach the plasma membrane by bypassing the Golgi (Gee et al., 2011). Because the nature of ABC-transporter-mediated UPS (Type II) is not fully characterized, we will consider Type I and Type II as a single category of leaderless non-vesicular UPS (Fig. 1). We will discuss these three categories of UPS with a focus on their link to human disease (summarized in Table 1).
Protein secretion pathways. The conventional secretion pathway involves the ER-to-Golgi transport of cargo (blue and yellow circles). However, a number of proteins use alternative secretory pathways, known as unconventional protein secretion (UPS) that do not involve ER-to-Golgi transport. Leaderless cytosolic proteins (orange circles) can be secreted by the cell via non-vesicular routes, such as through a membrane pore and the ABC transporter (Leaderless non-vesicular UPS). For example, UPS of FGF2 involves pore formation within the membrane, Tec kinase, ATP1A1 and extracellular heparan sulfate proteoglycan. In addition, a-factor of Saccharomyces cerevisiae requires the ABC transporter Ste6P. Another subset of leaderless cytosolic proteins, including IL-18, IL-33, α-synuclein, amyloid-β and IDE, can be secreted through autophagy-associated vesicles, such as lysosomes and late endosomes (Leaderless vesicular UPS). In addition, transmembrane proteins (red circles) can reach the plasma membrane by bypassing the Golgi (Golgi-bypassing UPS). Examples for this are αPS1 integrin, Mpl, and ΔF508-CFTR, which can be transported to the plasma membrane by GRASP-dependent UPS.
Protein secretion pathways. The conventional secretion pathway involves the ER-to-Golgi transport of cargo (blue and yellow circles). However, a number of proteins use alternative secretory pathways, known as unconventional protein secretion (UPS) that do not involve ER-to-Golgi transport. Leaderless cytosolic proteins (orange circles) can be secreted by the cell via non-vesicular routes, such as through a membrane pore and the ABC transporter (Leaderless non-vesicular UPS). For example, UPS of FGF2 involves pore formation within the membrane, Tec kinase, ATP1A1 and extracellular heparan sulfate proteoglycan. In addition, a-factor of Saccharomyces cerevisiae requires the ABC transporter Ste6P. Another subset of leaderless cytosolic proteins, including IL-18, IL-33, α-synuclein, amyloid-β and IDE, can be secreted through autophagy-associated vesicles, such as lysosomes and late endosomes (Leaderless vesicular UPS). In addition, transmembrane proteins (red circles) can reach the plasma membrane by bypassing the Golgi (Golgi-bypassing UPS). Examples for this are αPS1 integrin, Mpl, and ΔF508-CFTR, which can be transported to the plasma membrane by GRASP-dependent UPS.
In addition to the pathways mentioned above, there are other cell processes that can be considered to be UPS. For example, transport through the intercellular channels that function as pathways for the cellular spreading of macromolecules, including pathogens such as viruses and prion-like proteins, could be viewed a specialized form of UPS (see Box 2). Additionally, the senescence-associated secretory phenotype (SASP) is characterized by the secretion of pro-inflammatory and matrix-degrading molecules from senescent cells and is associated with a number of human diseases, including atherosclerosis, cancer and inflammatory diseases (Coppé et al., 2008). Because many SASP factors are leaderless proteins, UPS mechanisms are thought to be involved in their extracellular secretion (see Box 3).
Macromolecules, such as proteins and RNA, can be also transported directly to other cells through the intercellular channels. A well-known example are plasmodesmata, intercellular channels in plants and some algae (Knox and Benitez-Alfonso, 2014). The viral replication complexes of the Tobacco mosaic virus exhibit cell-to-cell movement through plasmodesmata, which can be regarded as a kind of UPS (Heinlein, 2015). In animals, tunneling nanotubes (TNTs) have been suggested to serve as the main pathway for such macromolecular transport (Gerdes et al., 2013; Rustom et al., 2004). Of note, the tunneling transport through TNTs can be used as a pathway for the cellular spreading of pathogens, such as viruses and prion-like proteins. For example, influenza A virus has been recently reported to spread through TNTs (Kumar et al., 2017). Additionally, TNT-meditated spreading of α-synuclein, a prion-like aggregated protein, uses lysosomal vesicles and has been associated with Parkinson's disease (Abounit et al., 2016; Gousset et al., 2009). However, further research is needed to evaluate the precise nature of TNTs and determine whether this type of protein transport can be specifically modulated to develop therapeutic strategies to treat human diseases.
Cellular senescence is thought to be a program of arrested proliferation and altered gene expression that can be triggered by many stresses (Kang et al., 2015). Compared to the culture medium of quiescent cells, that of senescent cells is enriched with secreted proteins. These characterize the so-called senescence-associated secretory phenotype [SASP; also known as the senescence messaging secretome (SMS)], and include interleukins (i.e. IL-1α, IL-1β and IL-6), chemokines [i.e. IL-8 and growth-regulated alpha protein (CXCL1)], growth factors [i.e. basic fibroblast growth factor (bFGF) and hepatocyte growth factor (HGF)] and extracellular proteases [i.e. matrix metalloproteinase (MMP)-1, MMP-3 and MMP-13]. SASP is associated with several human diseases, including atherosclerosis, chronic kidney disease, Crohn's disease and various cancers (He and Sharpless, 2017). Although recent studies on SASP have identified several specific transcriptional regulators, including GATA4 (Kang et al., 2015), the intracellular secretory mechanisms of the underlying factors are poorly understood. Because many SASP factors, such as IL-1β and IL-8, are leaderless proteins, UPS mechanisms are thought to play a role in SASP. Characterizing the UPS route of SASP factors and identifying their regulatory components will be of great importance to future research of ageing.
Leaderless non-vesicular UPS – Types I and II
The leaderless non-vesicular class of UPS includes cytoplasmic leaderless proteins that are secreted directly out of the cell either through plasma membrane pores (Type I) (Steringer et al., 2012) or through ABC transporters (Type II) (McGrath and Varshavsky, 1989). One of several typical triggers for this type of UPS is inflammation, which leads to the extracellular release of diverse cytokines that do not possess a signal peptide (Schroder and Tschopp, 2010). A well-known example of a protein that utilizes this UPS is interleukin (IL)-1β, which is mainly expressed in myeloid cells, such as macrophages and monocytes. Initially, IL-1β is produced as a 31-kDa inactive form that is cleaved by caspase-1 into the 17-kDa mature form. The latter is then recruited by the intracellular NACHT-domain-, LRR-motif- and PYD-containing protein3 (NLRP3) component of the inflammasome, thus engaging the immune response (Schroder and Tschopp, 2010). It appears that multiple UPS pathways can mediate IL-1β secretion (see below, ‘Leaderless vesicular UPS - Type III’), depending on inflammatory conditions and cell type. Secretion of IL-1β from macrophages following inflammation is mediated by a type of UPS that requires hyper-permeabilization of the plasma membrane (Bergsbaken et al., 2009; Martín-Sánchez et al., 2016). The precise mechanism of this membrane hyper-permeabilization is not yet fully understood (Rabouille, 2017), but the N-terminal domain of gasdermin-D, one of the regulators of pyroptosis that is also produced by caspase-1-mediated cleavage following initiation of inflammation, has been proposed to be involved in the formation of the membrane pore (Ding et al., 2016). The innate immune response evoked by cytokines, such as IL-1β was originally thought to be the first line of defense against non-self (e.g. microorganisms) and to serve as a sophisticated system to sense danger signals (Pitanga et al., 2016). However, elevated local or systemic levels of IL-1β have been associated with a number of hereditary or acquired human diseases, such as cryopyrin-associated periodic syndrome (Schroder and Tschopp, 2010). Therefore, preventing the overt secretion of IL-1β by modulating UPS pathways would be a potential new therapeutic strategy to overcome these inflammatory diseases (Table 1).
Another important example of Type I UPS is the translocation of a cargo through the plasma membrane via self-made lipidic pores. Examples are fibroblast growth factor 2 (FGF2) and the HIV-TAT protein, both of which are recruited to the cytoplasmic leaflet of the plasma membrane by interaction with phosphatidylinositol (4,5)-bisphosphate (PIP2) and then undergo self-oligomerization; this, sequentially, induces membrane insertion, pore formation, and extracellular translocation of FGF2 and HIV-TAT (Debaisieux et al., 2012; Rabouille, 2017; Zeitler et al., 2015). Phosphorylation of FGF2 by Tec kinases has been shown to be essential for PIP2-mediated translocation of FGF2 (Ebert et al., 2010). However, many questions persist concerning this process, including the source of the energy for self-oligomerization, formation of the membrane pore, and translocation across the membrane. Recently, the Na+/K+-ATPase subunit α1 (ATP1A1), an α-chain of the Na+/K+-ATPase heterotetramer (Shull et al., 1986), was identified as a regulatory factor for FGF2 secretion that is involved in recruitment of FGF2 to the plasma membrane leaflet (Zacherl et al., 2015). Interestingly, the secretion of FGF2 is inhibited by the lack of extracellular heparan sulfate proteoglycans, which may function as extracellular traps for FGF2 (Nickel, 2007). FGF2 has been shown to be crucial for the development of the central nervous system and adult neurogenesis, and proposed as a therapeutic target for various neurodegenerative diseases, including Alzheimer's disease, Parkinson's disease, multiple sclerosis and traumatic brain injury (Woodbury and Ikezu, 2014). Therefore, a modulation of the regulatory proteins that are involved in the UPS of FGF2, such as ATP1A1 or proteoglycans, might have therapeutic potential for neurodegenerative diseases. Accumulating evidence suggests that extracellular HIV-TAT, which is also secreted through the Type I UPS occurring through self-made pores, acts as a viral toxin and plays a key role in the progression of acquired immune-deficiency syndrome (AIDS) (Debaisieux et al., 2012). Therefore, inhibition of UPS regarding HIV-TAT could be beneficial to the treatment of AIDS.
It has been suggested that some leaderless cargos can be secreted out of the cell by ABC transporters. Since the initial discovery in 1989 that a-factor, the mating pheromone of Saccharomyces cerevisiae, is secreted from the cell by the ABC transporter Ste6p (McGrath and Varshavsky, 1989), several other proteins, including the m-factor of Schizosaccharomyces pombe (Christensen et al., 1997) and hydrophilic acylated surface protein B (HASPB) of Leishmania species (Denny et al., 2000; Maclean et al., 2012), have been found to be exported by ABC transporter-mediated translocation. Leishmania HASPB is a leaderless protein that is localized to the extracellular face of its plasma membrane. Leishmaniasis is caused by protozoan parasites of Leishmania species and spreads through the bites of certain types of sandfly. During Leishmaniasis pathogenesis, HASPB has been suggested to be required for either initial parasite transmission to the host or the establishment of parasites within host macrophages (Maclean et al., 2012). β-COP and δ-COP, subunits of the COPI coatomer complex that participates in the conventional secretion pathway, are known to be involved in the unconventional secretion of HASPB (Ritzerfeld et al., 2011). However, further studies are needed to elucidate the precise molecular mechanisms of UPS of HASPB, and to answer the question of the link between these COPI subunits and ABC transporters. Interestingly, HASPB is modified by lipidation, including N-myristoylation and palmitoylation, which is required for trafficking to the cell surface (Denny et al., 2000). Therefore, the study of how the lipidation event is involved in the secretion and function of HASPB might contribute to the future development of therapeutics.
Leaderless vesicular UPS – Type III
The leaderless vesicular class (Type III) of UPS involves leaderless cytoplasmic proteins that are packaged into membrane-bound organelles, including autophagosomes, lysosomes, endosomes and exosomes. Thus far, numerous studies have examined the substrate cargo proteins of leaderless vesicular UPS pathways and their transport mechanisms (Dupont et al., 2011; Ejlerskov et al., 2013; Lotze and Tracey, 2005). Although the transport process of individual cargo proteins varies greatly, autophagy-related vesicular structures appear to be an important transport carrier in Type III UPS (Manjithaya and Subramani, 2010). The best-known role of autophagy is that of a degradative canonical pathway that contributes to nutrient recycling and cellular defense by digesting cytoplasmic components during starvation (Galluzzi et al., 2014), as well as aggregated proteins (Rogov et al., 2014), damaged organelles (Yamano et al., 2014) and invading pathogens (Deretic et al., 2015). In addition to this classic role in maintaining cell health, mounting evidence suggests that autophagy also plays a role in UPS (Duran et al., 2010; Manjithaya and Subramani, 2010). This type of secretory autophagy has been shown to deliver several types of cytoplasmic (Type III UPS) and transmembrane proteins (Type IV UPS) to the cell surface for their secretion (Ponpuak et al., 2015).
Indeed, several cytokines and inflammatory mediators use autophagy components as a means of their unconventional secretion, suggesting a close link between inflammation and autophagy (see Table 1) (Manjithaya and Subramani, 2010). Recent studies have shown that the secretion of IL-1β, particularly that from non-immune cells, is mediated by vesicular structures, including those of the autophagy and endosome pathways (Dupont et al., 2011; Zhang et al., 2015). However, the general role of autophagy in IL-1β secretion remains ill-defined. For example, autophagy has been suggested to inhibit both the cleavage of pro-IL-1β by caspase-1 and the activity of the NLRP3 inflammasome, which may lead to the inhibition of IL-1β secretion (Harris and Rubinsztein, 2011). With regards to the clinical significance of unconventional IL-1β secretion during inflammation, the plasticity of IL-1β secretion under different conditions and cell types should be considered. Because, as discussed above, a significant portion of IL-1β secretion is mediated by non-vesicular UPS in macrophages, it remains to be determined whether inhibiting vesicular UPS of IL-1β in non-macrophage cells has therapeutic potential for human inflammatory diseases.
Another distinct example of Type III UPS is the secretion of high-mobility group box 1 (HMGB1) protein, a leaderless nuclear protein that is secreted into the extracellular space in an unconventional manner (Lotze and Tracey, 2005). It has been shown that autophagy components, such as ATG5 (Dupont et al., 2011) and secretory lysosomes (Gardella et al., 2002), are involved in the active secretion of HMGB1, although HMGB1 can also be passively secreted from necrotic cells (Scaffidi et al., 2002) or apoptotic cells (Qin et al., 2006) through permeabilized membranes after its dissociation from chromosomes (Falciola et al., 1997). HMGB1 is a main mediator of endotoxin shock (Wang et al., 1999) and acts on several immune cells to trigger inflammatory responses in the form of a damage-associated molecular pattern (DAMP), which then initiates and perpetuates a noninfectious inflammatory response through the activation of its receptors, which include the Toll-like receptors (TLRs) TLR2 and TLR4, and the receptor for advanced glycation and end product (RAGE, also known as AGER) (Scaffidi et al., 2002). Other functions of extracellular HMGB1 are in the maturation of dendritic cells, the production of pro-inflammatory cytokines in myeloid cells, the induction of cell adhesion molecules in endothelial cells, and the progression of cancer (Sims et al., 2010). Therefore, the modulation of HMGB1 secretion is a potential way to treat its associated diseases.
In addition to IL-1β and HMGB1, several other inflammatory mediators, such as IL-18 and IL-33, have been shown to be secreted by UPS, particularly by autophagy-associated mechanisms (Deretic et al., 2012; Murai et al., 2015). Potential therapeutic approaches that might alleviate excessive inflammatory responses by modulating UPS include the following strategies: (i) inhibition of caspase-1-mediated cleavage of IL-1β to treat diseases, such as gout, atherosclerosis and diabetes (Burns et al., 2003; Schroder and Tschopp, 2010); (ii) inhibition of cathepsin G and elastase-mediated cleavage of IL-33 to prevent inflammatory diseases, for instance, arthritis and chronic obstructive pulmonary disease (COPD) (Cayrol and Girard, 2009; Lefrancais et al., 2012); (iii) inhibition of calpain-mediated cleavage of IL-1α to provide therapy for diseases including stroke and atherosclerosis (Zheng et al., 2013) and; (iv) inhibition of caspase-1 and proteinase-mediated cleavage of IL-18 to provide therapy for diseases such as age-related macular degeneration (Gu et al., 1997; Sugawara et al., 2001).
Secretory autophagy may play a role in the extracellular transport of aggregation-prone proteins, such as α-synuclein and amyloid-β (Ejlerskov et al., 2013; Nilsson et al., 2013). α-Synuclein, particularly in its aggregated forms, has been implicated in the pathogenesis of Parkinson's disease and other related neurological disorders (Kahle, 2008; Polymeropoulos et al., 1997). The extracellular secretion of α-synuclein may result in cell-to-cell transmission of protein aggregates that occur in many neurodegenerative disorders (Lee et al., 2005, 2010). Therefore, reducing of the unconventional secretion of α-synuclein by inhibiting autophagy possibly has therapeutic potential for Parkinson's disease (Ejlerskov et al., 2013). However, the role of autophagy in α-synuclein secretion was later challenged by other researchers who showed that autophagy inhibition promotes α-synuclein secretion (Lee et al., 2013; Lee and Lee, 2016). In the case of amyloid-β, the main component of the amyloid plaques found in the brains of Alzheimer patients, it is unclear whether reducing extracellular amyloid-β secretion would be beneficial for the treatment of Alzheimer's disease. For example, inhibiting autophagy through neuron-specific deletion of Atg7 in mice aggravated the neurotoxic phenotype due to the accumulation of intracellular amyloid-β aggregates (Nilsson et al., 2013). Interestingly, the secretion of insulin-degrading enzyme (IDE), a main endogenous amyloid-β-degrading enzyme that is released from astrocytes, has been shown to be mediated by autophagy-based UPS (Son et al., 2016). Therefore, stimulation of IDE secretion from astrocytes by modulating the UPS pathway constitutes a potential strategy to cope with Alzheimer's disease (Son et al., 2016).
Golgi-bypassing UPS – Type IV
Proteins undergoing Golgi-bypassing UPS have a signal peptide that is recognized by the signal recognition particle (SRP), which recruits the protein to the ER while it is being synthesized on the ribosome (Blobel and Dobberstein, 1975; Walter et al., 1981). In contrast to conventional secretion cargos that travel through the entire Golgi, these UPS cargos bypass at least part of the Golgi complex on their way to the cell surface (Gee et al., 2018). Cargos for Golgi-bypassing UPS include position-specific antigen subunit alpha 1 (αPS1) integrin in Drosophila (Schotman et al., 2008, 2009), myeloproliferative leukemia virus oncogene (Mpl) (Cleyrat et al., 2014), the ion channel cystic fibrosis transmembrane conductance regulator (CFTR) and the ion-transporting membrane protein pendrin (Gee et al., 2011; Jung et al., 2016), as well as the ciliary membrane proteins polycystin-2, the M2 mutant of Smoothened (Hoffmeister et al., 2011) and peripherin 2 (Tian et al., 2014). In addition, CD45 (PTPRC), connexin 26, connexin 30, pannexin 1, pannexin 3, serglycin and scramblase 1 have also been shown to reach the cell surface by unconventional secretion (Baldwin and Ostergaard, 2002; Martin et al., 2001; Merregaert et al., 2010; Penuela et al., 2007; Qu et al., 2009; Scully et al., 2012). Although all Golgi-bypassing plasma-membrane proteins are typically referred to as Golgi-bypassing UPS cargos, the individual trafficking routes for the different proteins are not identical (Fig. 2). For example, peripherin 2 reaches cilia through COPII-dependent exit from the ER, distinguishing it from other UPS cargos that rely on COPII-independent routes, such as Mpl (Cleyrat et al., 2014) and CFTR (Gee et al., 2011).
Known molecular arrangements involved in UPS of transmembrane proteins bypassing the Golgi. Immature Mpl involved in myeloproliferative cancer reaches the plasma membrane by a GRASP55- and ATG5-mediated pathway. ΔF508-CFTR, which causes cystic fibrosis because of a trafficking defect, can be transported to the plasma membrane by a pathway involving GRASP, ATGs (ATG1, -5, -7 and -8), SEC16A and IRE1α. The substitution mutant H723R-Pendrin, which causes congenital hearing loss, can reach the plasma membrane by a route that is mediated by HSP70, DNAJC14, RAB18 and IRE1α. Polycystin-2, whose mutations cause autosomal-dominant polycystic kidney disease, is transported by RAB8A-mediated UPS. The M2 mutant of Smoothened which can induce cancer, is transported to the plasma membrane by RAB8A-mediated UPS. Peripherin 2, whose mutations cause ophthalmic diseases, can be transported to the plasma membrane by COPII-mediated UPS.
Known molecular arrangements involved in UPS of transmembrane proteins bypassing the Golgi. Immature Mpl involved in myeloproliferative cancer reaches the plasma membrane by a GRASP55- and ATG5-mediated pathway. ΔF508-CFTR, which causes cystic fibrosis because of a trafficking defect, can be transported to the plasma membrane by a pathway involving GRASP, ATGs (ATG1, -5, -7 and -8), SEC16A and IRE1α. The substitution mutant H723R-Pendrin, which causes congenital hearing loss, can reach the plasma membrane by a route that is mediated by HSP70, DNAJC14, RAB18 and IRE1α. Polycystin-2, whose mutations cause autosomal-dominant polycystic kidney disease, is transported by RAB8A-mediated UPS. The M2 mutant of Smoothened which can induce cancer, is transported to the plasma membrane by RAB8A-mediated UPS. Peripherin 2, whose mutations cause ophthalmic diseases, can be transported to the plasma membrane by COPII-mediated UPS.
From a therapeutic perspective, it is of great interest that some disease-causing membrane proteins that have defects in protein folding or cell-surface trafficking, such as CFTR and pendrin mutants (Gee et al., 2011; Jung et al., 2016), can be transported to the plasma membrane by UPS pathways. CFTR is an epithelial anion channel, and its loss of function as a result from genetic mutations causes cystic fibrosis (MIM 219700) and several other epithelial diseases, such as bronchiectasis and chronic pancreatitis (LaRusch et al., 2014; Lee et al., 2003). The most common disease-causing mutation of CFTR is the deletion of Phe at position 508 (ΔF508). Pendrin, a protein encoded by SLC26A4, is a transmembrane protein that exhibits Cl− to HCO3− or Cl− to I− exchange activity in the inner ear, thyroid follicles and renal cortical collecting ducts (Mount and Romero, 2004). Mutations in SLC26A4 cause non-syndromic recessive deafness with an enlarged vestibular aqueduct (deafness autosomal recessive 4, DFNB4, [MIM 600791]) and Pendred syndrome (PDS, [MIM 274600]) (Everett et al., 1997; Li et al., 1998), a common cause of hereditary hearing loss in humans. Specifically, p.H723R (His723Arg) is one of the most prevalent pathological mutations (Dossena et al., 2009; Lee et al., 2014). Both the lack of Phe508 in CFTR (ΔF508-CFTR) and substitution of His723 for Arg in pendrin (H723R-Pendrin) result in protein misfolding, retention in the ER and subsequent degradation by the ER-associated degradation (ERAD) pathway (Ward et al., 1995; Yoon et al., 2008). Consequently, only negligible amounts of ΔF508-CFTR and H723R-Pendrin reach the plasma membrane, and most of the ion-transporting activity at the cell surface is lost (Amaral, 2004; Yoon et al., 2008). Although the mutant proteins have some defects in protein folding, they retain a certain level of functional activity if they reach the cell surface (Gee et al., 2011; Jung et al., 2016). Therefore, many research efforts are invested in approaches that facilitate the membrane targeting of ΔF508-CFTR and H723R-Pendrin.
Notably, the blockade of conventional ER-to-Golgi trafficking, which induces ER stress and the unfolded protein response (UPR), has been shown to also evoke unconventional cell-surface trafficking of CFTR and pendrin (Gee et al., 2011; Jung et al., 2016). It is, however, difficult to adopt a direct activation of ER stress as a therapeutic strategy because it is likely to give rise to many unfavorable side effects (Yoshida, 2007). The basic mechanisms through which CFTR and pendrin reach the cell membrane appear to be similar. For example, expression of both ΔF508-CFTR and H723R-Pendrin at the plasma membrane was abolished by the knockdown of IRE1. This finding indicates that IRE1 plays a major role in the UPS of CFTR and pendrin mutants, when considering the three UPR signaling arms consisting of IRE1, PERK and ATF6 (Gee et al., 2011; Jung et al., 2016). The molecular mechanism of how IRE1 facilitates Type IV UPS is not clearly understood; however, a recent report has shown that IRE1 augments the expression and function of SEC16A, which forms the ER exit sites for the UPS of ΔF508-CFTR (Piao et al., 2017), suggesting that SEC16A is a downstream target of IRE1-mediated upregulation of UPS. In contrast to the finding that the same ER stress signals can activate UPS of both CFTR and pendrin, some of the key molecular factors that are involved in the UPS of these two membrane proteins are different. For example, Golgi reassembly stacking proteins (GRASPs) are required for the UPS of CFTR (Gee et al., 2011), whereas the HSP70 co-chaperone DNAJC14 is involved in the UPS of pendrin (Jung et al., 2016) (see below, ‘Potential therapeutic targets’). Theoretically, an activation of the cargo-specific pathway would be more desirable to develop therapeutics for each disease as it might help to minimize the adverse events caused by the activation of common pathways that might also induce UPS of untoward cargos.
In addition to CFTR and pendrin, several other transmembrane proteins that are associated with human diseases can arrive at the plasma membrane by UPS. Polycystin-2, mutations of which are associated with autosomal-dominant polycystic kidney disease (Mochizuki et al., 1996), is transported to cilia by UPS; this is mediated by RAB8A, a small GTPase, which plays a role in vesicular traffic from the trans-Golgi network to the plasma membrane (Hoffmeister et al., 2011). Activating mutations of Smoothened can induce medulloblastoma, basal-cell carcinoma, pancreatic cancer and prostate cancer by unregulated activation of the hedgehog pathway (Rubin and de Sauvage, 2006), and this protein is also transported to the cell surface via UPS (Hoffmeister et al., 2011). Patients with myeloproliferative neoplasms (MPNs) often carry either an activating form of JAK2 or mutations in the ER resident protein calreticulin (Klampfl et al., 2013; Nangalia et al., 2013), which leads to accumulation of immature Mpl in the ER (Kralovics et al., 2003; Moliterno et al., 1998). It has been reported that immature Mpl utilizes an unconventional autophagic secretory pathway to reach the cell surface (Cleyrat et al., 2014). Therefore, either the activation or inhibition of their unconventional secretion could have therapeutic potential for any associated diseases (Table 1).
Potential therapeutic targets
Although the precise nature and mechanisms of UPS are not yet fully understood, the accumulated knowledge thus far suggests that research into the following three areas would be especially helpful in identifying druggable targets for human diseases, as well as to further elucidate the underlying mechanism of UPS.
First, a promising research area of UPS is the identification of the vesicular carrier involved in leaderless vesicular and Golgi-bypassing UPS. Results from previous studies have provided a list of candidates for vesicular carriers that include COPII-coated vesicles (Tian et al., 2014), autophagy vesicles (Cleyrat et al., 2014; Gee et al., 2011), lipid droplets (Jung et al., 2016) and endosomes (Zhang et al., 2015). It is also possible that multiple vesicular systems are involved in a single UPS event. For example, it has been suggested that both autophagosome and endosome/multivesicular body (MVB) components are sequentially involved in the vesicular UPS of IL-1β (Zhang et al., 2015). Our recent results also indicate that early autophagosomal components, MVBs and RAB8A-dependent recycling vesicles sequentially mediate the ER stress-induced UPS of CFTR (our unpublished observation). To be able to modulate the UPS pathway with a clear therapeutic potential for diverse human diseases, the characterization of the vesicular system involved in the UPS of each cargo of interest and their regulatory processes are of paramount importance.
Second, GRASPs have emerged as an interesting regulator of the UPS pathway of various cargo proteins. GRASPs were initially identified as factors required for the stacking of Golgi cisternae by using in vitro assays (Barr et al., 1997). Two isoforms, GRASP55 and GRASP65, exist in vertebrates (Barr et al., 1997; Shorter et al., 1999). Interestingly, several studies have demonstrated that GRASPs are involved in Golgi-bypassing UPS in both invertebrate and vertebrate models, although they were initially described as Golgi-associated proteins (Cleyrat et al., 2014; Kinseth et al., 2007; Schotman et al., 2008). For example, the GRASP homologs in Dictyostelium and Drosophila, respectively, mediate the transport of acyl-CoA-binding protein (AcbA) and α-integrin at specific developmental stages through an unconventional Golgi-independent route (Gee et al., 2011; Kinseth et al., 2007; Schotman et al., 2008). Furthermore, as discussed above, GRASPs participate in the UPS of the mammalian transmembrane proteins CFTR and Mpl (Cleyrat et al., 2014; Gee et al., 2011). However, the precise roles of GRASPs in UPS are largely unknown. Interestingly, monomerization and ER re-localization of GRASP55 by phosphorylation on its serine 441 residue appear to be critical for the UPS of CFTR (Kim et al., 2016). Therefore, in order to be able to develop therapeutics for cystic fibrosis, a feasible approach might be to screen for small-molecule compounds that can induce phosphorylation and, thus, activation of GRASP55.
Third, increasing evidence suggests that molecular chaperones, such as heat shock 70 kDa proteins (HSP70s), HSP90s and members of the DNAJ-protein family (also known as HSP40s), play a role in the UPS of diverse cargos (Jung et al., 2016; Zhang et al., 2015). The correct folding of secretory and transmembrane proteins is ensured by ER quality control (ERQC) systems (Vembar and Brodsky, 2008). During endoplasmic-reticulum-associated protein degradation (ERAD), misfolded proteins that do not pass ERQC are retro-translocated to the cytoplasm and degraded in an ubiquitin-dependent process by the proteasome. Molecular chaperones have crucial roles in the recognition of misfolded proteins and the heat shock cognate protein 70 (Hsc70; encoded by HSPA8), a member of the HSP70 family, is one of the important chaperones involved in this process. However, HSP70 chaperones do not function alone. In addition to protein folding and degradation, HSP70s are involved in a myriad of biological processes, including protein–protein interaction and intracellular trafficking of various proteins (Kampinga and Craig, 2010; Young et al., 2003). Much of the functional diversity of HSP70 is thought to be driven by cofactors, including chaperone DNAJ proteins (Kampinga and Craig, 2010). Indeed, as mentioned above, the DNAJ protein DNAJC14 plays a crucial role in the UPS of misfolded pendrin (Jung et al., 2016). It appears that the activation of the chaperone machinery stimulates both UPS and the ERAD pathway to relieve the protein burden in the ER during ER stress (Ron and Walter, 2007). Among the many co-chaperones, DNAJC14 appears to be specialized in assisting Hsc70 during UPS, whereas other co-chaperones, such as CHIP, are involved in ERAD (Meacham et al., 2001). Although Hsc70 is required for the UPS of ΔF508-CFTR, DNAJC14 is not (Jung et al., 2016), suggesting the presence of unknown co-chaperone that is specific for UPS of CFTR. In addition, it has been shown that the Hsc70 co-chaperone DNAJC5 participates in Type III UPS of misfolded cytosolic proteins (Xu et al., 2018). Interestingly, DNAJC14, which mediates the UPS of transmembrane proteins, has ER-membrane-localizing transmembrane domains, whereas DNAJC5, which is involved in the UPS of cytosolic proteins, does not have any transmembrane domains (Kampinga and Craig, 2010). Identifying cargo-interacting domains in these proteins and determining the principles of how these co-chaperones recognize their UPS substrate cargos will be important in order to develop strategies for cargo-specific modulation of UPS for therapeutic applications.
Conclusion and perspectives
UPS mechanisms are found in all organisms, including yeast, fungi, plants, Drosophila and mammals (Rabouille, 2017). It is useful to classify UPS pathways and define similarities in these complex mechanisms of UPS. However, the current classification system based on cargo protein category may not be sufficient to encompass the complexities of UPS. For example, the UPS of IL-1β has characteristics of both non-vesicular (Type I) and vesicular (Type III) UPS. Additionally, Type III vesicular UPS of IL-1β involves components of the autophagosome and GRASPs (Dupont et al., 2011; Zhang et al., 2015), which are also involved in Type IV UPS of transmembrane proteins Mpl and CFTR (Cleyrat et al., 2014; Gee et al., 2011), pointing to some overlap between the different pathways. Even within UPS of transmembrane proteins, the processes involved exhibit considerable diversity depending on the particular cargos (Fig. 2). Therefore, the identification of the molecular machinery and the processes that govern the UPS of each specific cargo is essential in developing therapeutic strategies for related diseases. Broad questions remain about (a) the molecular nature of carriers at the plasma membrane that mediate non-vesicular UPS of cytosolic cargos, (b) types of vesicular carrier involved in the vesicular UPS and, (c) characteristics and intramolecular determinants of transmembrane proteins that are transported by using Golgi-bypassing UPS. More specifically, advancing recent findings related to mechanisms of UPS will have immediate therapeutic potential, particularly by identifying the crucial regulators that divert secretory autophagy from degradative canonical autophagy, by investigating the molecular switch that activates the UPS function of GRASP from its classic Golgi membrane-tethering function, and by identifying the types of molecular chaperone involved in UPS and determining how these chaperones recognize UPS cargos.
Insights of the above aspects will bring us closer to understand why these proteins are transported unconventionally via various routes and how exactly the different UPS pathways are involved in the pathogenesis of human diseases. This continued line of investigation will, ultimately, open a new era of therapeutics for numerous human diseases, including cancers, inflammatory diseases and some genetic disorders, such as cystic fibrosis and Pendred syndrome.
Acknowledgements
We thank Dong Soo Chang for the assistance with illustrations.
Footnotes
Funding
This work was supported by grant 2013R1A3A2042197 from the National Research Foundation, the Ministry of Science, ICT & Future Planning, Republic of Korea, and grant HI15C1543 of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI), funded by the Ministry of Health & Welfare, Republic of Korea.
References
Competing interests
The authors declare no competing or financial interests.