Unconventional Protein Secretion in the Central Nervous System: Mechanisms and Roles in Physiology and Disease

ABSTRACT

Extracellular secretion of neurotransmitters, proteins, and peptides by cells of the central nervous system underpins neurological function and homeostasis. Decades of elegant research have illuminated the molecular mechanisms and machinery that support the release of neurotransmitters via synaptic vesicle exocytosis, as well as the secretion of signal‐peptide bearing proteins through the endoplasmic reticulum (ER)‐Golgi based secretory pathways. However, it is now increasingly appreciated that signal‐peptide lacking “leaderless” proteins can also be secreted via ER‐Golgi‐independent mechanisms collectively termed unconventional protein secretion (UcPS). In this review, we highlight the physiological and pathological consequences of UcPS in the central nervous system. UcPS supports the secretion of aggregation‐prone proteins such as α‐synuclein and mutant huntingtin, pro‐inflammatory mediators including interleukin‐1β and high mobility group box protein 1, and neuroprotective or angiogenic factors such as fibroblast growth factor 2. Furthermore, several retroelement‐derived proteins, encoded by ancient genomic elements with structural homology to retroviruses, are also secreted via unconventional pathways, and are thought to regulate essential CNS processes such as synaptic plasticity. These diverse cargoes underscore the functional range of UcPS in neuronal and glial biology. We summarize current understanding of the major UcPS pathways used by CNS cells. These mechanisms include plasma‐membrane pore‐mediated release facilitated by proteins such as gasdermin‐D, as well as vesicular routes in which UcPS cargoes enter organelles of the autophagic and endolysosomal systems that subsequently fuse with the plasma membrane to enable extracellular release. Finally, we discuss key unresolved questionRecent evidence from HeLa cells suggests regarding the regulation of UcPS, including the molecular features that target select leaderless cargoes toward UcPS, how the balance between conventional secretion and UcPS shifts under cellular stress, and the current understanding of the diverse molecular machinery that mediates the vesicular form of UcPS.

Neurons and glia secrete and respond to numerous proteins that are released independently of the conventional ER–Golgi pathway. These unconventionally secreted proteins exert diverse pathogenic, immunomodulatory, and neuroprotective functions and are therefore implicated in multiple neurodegenerative diseases. They originate in the cytosol and exit the cell through membrane pores or vesicle‐mediated pathways. The molecular mechanisms that mediate and regulate unconventional protein secretion are diverse and cargo‐specific.

1. Introduction

The physiology of the brain is enabled by communication between the many cell types that comprise the central nervous system. Synaptic transmission, defined as the release of neurotransmitters by the pre‐synaptic neuron that engage receptors on the post‐synaptic targets, remains the most extensively characterized mode of neuronal communication (Neher and Brose 2018; Rizzoli 2014; Sudhof 2013), but it represents only one component of a broader signaling landscape. Neurons, astrocytes, microglia, and oligodendrocytes engage in continuous crosstalk that regulates neuronal activity, tissue homeostasis, proteostasis, and immune surveillance (Middeldorp et al. 2023; MacLean et al. 2022; Wittendorp et al. 2004). These interactions are mediated by a wide range of extracellular signals, including peptides and proteins, which are secreted through multiple, mechanistically independent secretory pathways (Sugita 2008; Rudolf et al. 2011; Dieni et al. 2012; van Ziel et al. 2019; Andrews and Chakrabarti 2005; Vardjan et al. 2013; Veh and Luningschror 2026).

Most secreted proteins exit cells via the conventional secretory pathway in which signal‐peptide bearing polypeptides are co‐translationally inserted into the endoplasmic reticulum (ER) and transit through the Golgi and are packed into secretory vesicles which fuse with the plasma membrane to release packaged cargo (Barlowe and Helenius 2016; Viotti 2016). In neurons and neuroendocrine cells, regulated secretory granules such as dense core vesicles, which originate from the trans‐Golgi network, can be stored for extended periods of time and undergo stimulus‐dependent exocytosis (Anantharam and Kreutzberger 2019; Burgoyne and Morgan 2003; van den Pol 2012). Proteins co‐translationally inserted into the ER are therefore competent for secretion through constitutive or regulated ER‐Golgi‐dependent pathways. However, a substantial fraction of the extracellular proteome cannot be explained by these conventional secretory pathways. Proteins lacking both ER‐targeting signal peptides and transmembrane domains are synthesized in the cytoplasm and cannot access the ER through co‐translational insertion. Nevertheless, select proteins lacking ER‐targeting signal‐peptides and transmembrane domain lacking proteins, including α‐synuclein and the cytokine interleukin‐1β, are nevertheless detected outside cells, indicating the existence of alternative routes of export.

Throughout this review, we use unconventional protein secretion (UcPS) as a cargo‐defined umbrella term for the extracellular release of such signal peptide‐ and transmembrane domain‐lacking proteins, which we refer to as UcPS cargoes (Cohen et al. 2020; Rabouille 2017). This definition does not imply a single export mechanism. Rather, UcPS cargoes can be released through multiple mechanistically distinct pathways, including direct translocation across the plasma membrane, extracellular vesicle‐associated export, secretion through autophagy‐ and lysosome‐related compartments, plasma‐membrane pore‐mediated export associated with inflammatory pathways, and transfer within specialized extracellular retrovirus‐like particles (Neel et al. 2024; Pollock et al. 2024; Jiao et al. 2024). These regulated export pathways must also be distinguished from passive extracellular release caused by loss of plasma‐membrane integrity, although both mechanisms can contribute to the extracellular detection of cytoplasmic proteins. We therefore use UcPS to organize the diverse pathways shown to mediate the extracellular release and transfer of signal‐peptide lacking proteins in the CNS, while distinguishing their underlying molecular mechanisms.

These mechanisms enable the extracellular delivery of UcPS cargoes involved in stress signaling, angiogenesis, innate immune responses, and proteostasis. Intercellular communication in the CNS therefore extends beyond classical neurotransmitters to include the release of cytoplasmic proteins that influence neighboring cells. Examples include the cell‐to‐cell spread of aggregation‐prone proteins such as α‐synuclein (Burbidge et al. 2022) and the secretion of cytoplasmic pro‐inflammatory cytokines such as interleukin‐1β from microglia (Pike et al. 2021). The pathways that mediate their release can therefore propagate inflammatory and pathological signals across neural circuits and thus function both as markers of disease and as modulators of neurodegenerative and neuroinflammatory processes (El‐Agnaf et al. 2006). Unconventionally secreted proteins (UcPS cargoes) have been identified in both neuronal and glial populations and exhibit functions that are either neuroprotective or neurodegenerative (Veh and Luningschror 2026; Nakamura et al. 2024). This review examines UcPS in neuronal and glial cells by first highlighting the roles of UcPS cargoes in neurological physiology and disease and then outlines current knowledge and outstanding questions regarding the regulation of their secretion.

2. UcPS Influence on Neurological Physiology and Disease

2.1. UcPS of Aggregation‐Prone Proteins and the Extracellular Propagation of Neurodegenerative Disease

Neurological diseases caused by the spread of misfolded proteins, referred to as proteinopathies, include synucleopathies, tauopathies, and polyglutamine disorders. In each case, there is intercellular spread of cytoplasmic, aggregation‐prone proteins, including α‐synuclein, tau, and mutant huntingtin, via UcPS (Peng et al. 2020; Figure 1A). α‐synuclein, an intrinsically disordered membrane‐binding protein enriched at presynaptic terminals, is released from neurons in both soluble and aggregated forms (Lee et al. 2005; Fussi et al. 2018; Jang et al. 2010; Lee, Cho, et al. 2013; Tyson et al. 2016). Extracellular aggregate clearance can occur by enzymatic degradation or phagocytosis by microglial cells (Caron et al. 2021; Fu et al. 2016; Lee et al. 2008; Qiu et al. 1998; Long‐Smith et al. 2009), a process which is impaired by aging, and insufficient clearance can lead to dysfunction and disease propagation (Ogawa et al. 2000; Sung et al. 2005; Hong et al. 2024; Freeman et al. 2013). In addition, extracellular α‐synuclein is a proinflammatory stimulator of immune cells (Fellner et al. 2013; Kim et al. 2016, 2013; Zhang et al. 2005), and extracellular aggregates can seed aggregation in recipient cells (Uemura et al. 2020; Rey et al. 2019; Volpicelli‐Daley et al. 2014; Luk et al. 2009; Steiner et al. 2011), further contributing to intercellular spread (Figure 1B).

FIGURE 1.

Unconventional protein secretion (UcPS) in the CNS mediates both pathogenic and neuroprotective effects. (A) UcPS of aggregation‐prone cytosolic proteins, such as α‐synuclein or tau, promotes their extracellular release and facilitates the propagation of neurodegenerative pathology. (B) Export of damage‐associated molecular patterns (DAMPs), including aggregated proteins, HMGB‐1, heat shock proteins, and some galectins, activates glial cells, promoting further DAMP and cytokine release. (C) UcPS also supports neuroprotective processes, including angiogenesis (e.g., FGF2 secretion) and extracellular degradation of protein aggregates (e.g., via IDE). Created with BioRender.com.

Polyglutamine‐expanded mutant huntingtin displays a similar phenomenon. Mutant huntingtin is released more efficiently than the wild‐type protein, a process associated with intracellular aggregation and proteostatic stress (Trajkovic et al. 2017; Bonavita et al. 2023; Ahat et al. 2022). Enhanced extracellular appearance of mutant huntingtin is therefore a feature of Huntington's disease pathology and may contribute to redistribution of proteotoxic material within neural tissue (Trajkovic et al. 2017; Ananbeh et al. 2021). Tau protein accumulation defines a class of neurodegenerative disease broadly classified as tauopathies, with Alzheimer's disease being the most prominent. Deposition of insoluble tau aggregates called neurofibrillary tangles is thought to be contributing factors in the progression of Alzheimer's disease (Han et al. 2017; Maphis et al. 2015; Goedert and Spillantini 2017).

Collectively, these observations indicate that neurons can disseminate aggregation‐prone proteins through UcPS. Neuronal proteostasis is normally maintained through intracellular degradative pathways, including the ubiquitin‐proteasome system and macroautophagy, and failure or saturation of these leads to persistence of toxic species (Dennissen et al. 2012; Hohn et al. 2020; Nixon and Rubinsztein 2024). Under such conditions, UcPS is thought to provide an additional route through which aggregation‐prone proteins can exit the cell. Among the pathways proposed to contribute to this process is misfolding‐associated protein secretion (MAPS), a proteostasis‐linked export route for cytosolic misfolded proteins that bypasses the conventional ER‐Golgi pathway (Lee et al. 2016). MAPS is thought to function alongside other UcPS mechanisms that release aggregation‐prone proteins from neurons and glia and are discussed below in Section 3.

Extracellular release of these proteins may transiently reduce intracellular proteotoxic burden by distributing aberrant proteins away from the producing cell. However, as secretion‐based clearance is inherently non‐cell‐autonomous, secreted proteins can persist in the extracellular space, engage, and inflame neighboring neurons or glia, and seed aggregation in recipient cells. UcPS in the CNS thus functions as a double‐edged mechanism that enables export of proteotoxic cargo, while simultaneously facilitating the spread of neurodegeneration‐associated proteins.

2.2. Export of Damage Associated Molecular Patterns (DAMPs) and Innate Immune Modulation

Neurodegenerative conditions are frequently accompanied by neuroinflammation, either as a primary driver of neural injury or as a consequence of degeneration (Ransohoff and Brown 2012). Glial‐derived secreted factors play a central role in initiating and sustaining these inflammatory states. Misfolded or aggregated neuronal UcPS cargoes can directly activate glial cells through two major routes (Figure 1B): (1) by direct engagement of pattern recognition receptors, such as the activation of toll‐like‐receptor‐1/2 by extracellular oligomeric α‐synuclein and tau fibrils (Daniele et al. 2015; Dutta et al. 2023; Xiang et al. 2015; Lee, et al. 2013), and (2) internalization of these aggregates into glia where they disrupt endolysosomal membranes and trigger the activation of the NLR family pyrin domain containing 3 (NLRP3) inflammasome, a cytoplasmic damage sensor that triggers the release of pro‐inflammatory cytokines (Pike et al. 2021; Freeman et al. 2013; Flavin et al. 2017; Gordon et al. 2018; Codolo et al. 2013). These events establish a pro‐inflammatory milieu that primes surrounding cells for further inflammatory amplification.

A key component of this amplification is the release of damage associated molecular patterns (DAMPs) by stressed neurons and glial cells (Castro‐Gomez and Heneka 2024). DAMPs are endogenous, non‐microbial molecules, including proteins and lipids, that activate immune receptors. Release of DAMPs often precedes cell death and reflects an adaptive stress response, and persistent DAMP signaling through pattern‐recognition receptors on microglial and astrocytes drives transcription reprogramming toward pro‐inflammatory states, promoting neuronal dysfunction across neurodegenerative and neuroinflammatory conditions (Kaur et al. 2023; Venegas and Heneka 2017).

High mobility group box 1 (HMGB1) is a well‐characterized CNS DAMP that can be released by cell death or as a UcPS cargo released from neurons and glia in response to tauopathies, ischemia, and other inflammatory states within the brain (Gaikwad et al. 2021; Festoff et al. 2016; Paudel et al. 2018; Hayakawa et al. 2012; Shi et al. 2022; Maulik et al. 2026). HMGB1 is among the most abundant nuclear proteins, where it binds DNA, altering nucleosome geometry, and regulates chromatin organization and transcription (Muller et al. 2004; Bonaldi et al. 2002). In response to cellular damage and oxidative stress, HMGB1 subcellular distribution shifts from the nucleus toward the cytosol prior to secretion (Brambilla et al. 2018; Sun et al. 2018; Zhang et al. 2026; Qiu et al. 2008). This process is regulated by post‐translational modifications and the redox state of HMGB1 (Stavely et al. 2022; Seol et al. 2024; Kwak et al. 2019). The extracellular activity of HMGB1 is also redox dependent. Distinct oxidation states influence its capacity to promote chemotaxis or cytokine production, while terminally oxidized HMGB1 is considered immunologically inactive (Venereau et al. 2012). Its effects therefore depend on both its redox state and the receptor context of the responding cell. Similar to early reports in monocytes showing that lysosomal exocytosis facilitates the secretion of HMGB1 (Gardella et al. 2002), HMGB1 release from glioblastoma cells is dependent on autolysosomal compartments (Li et al. 2022). Neurons can also secrete HMGB1 following injury, where HMGB1 laden extracellular vesicles have been visualized by electron microscopy (Sun et al. 2018; Kaya et al. 2023).

Extracellular HMGB1 engages TLR4 and RAGE receptors to activate NF‐κB signaling, reactive oxygen species production, and inflammasome priming in glial cells (Wang et al. 2020). This has been shown to promote microglial activation and persistence of inflammatory states, suppress resolution mechanisms, and exacerbate neuronal stress by impairing synaptic integrity and intracellular homeostatic pathways, further neurodegeneration (Gaikwad et al. 2021; Gao et al. 2011).

In contrast to DAMPs, cytokines such as interleukin‐1β (IL‐1β) represent regulated immune mediators whose extracellular appearance is tightly controlled (Gerke et al. 2024; De Maio 2011; Matsunaga et al. 2020; Steelman and Li 2014). IL‐1β exemplifies a cytokine that lacks a signal peptide and therefore relies on UcPS mechanisms for secretion (Zheng et al. 2026; Martin‐Sanchez et al. 2016; Anna Rubartelli et al. 1990). Inflammasome activation in microglia and astrocytes results in caspase‐1–dependent maturation of pro‐IL‐1β, followed by secretion through either plasma membrane pore‐mediated or vesicular routes, as discussed below in Section 3. Extracellular IL‐1β contributes broadly to neuroinflammation through multiple mechanisms, including increased blood‐brain barrier permeability, allowing peripheral immune cells and signals into the CNS, as well as directly modulating neuronal and glial responses (Gajtko et al. 2020; Ferrari et al. 2004; Hewett et al. 2012). The inhibition of IL‐1β signaling within the CNS leads to improved cognitive performance in mouse models of neuroinflammation, including Alzheimer's Disease and Parkinson's Disease (Phani et al. 2012; Towers et al. 2019; Kitazawa et al. 2011). Importantly, IL‐1β has context‐dependent roles and is also involved in beneficial processes such as remyelination and synaptic pruning (Ferrari et al. 2004; Zhang et al. 2022; Mason et al. 2001).

Additional UcPS cargoes with immunomodulatory potential include heat shock proteins (HSPs) and galectins (Danzer et al. 2011; Uddin et al. 2021; Kim et al. 2012; Kim et al. 2020; De Maio and Vazquez 2013; Chen et al. 2019). Extracellular HSP70 is neuroprotective in mouse models of Alzheimer's disease (Kim et al. 2020; Bobkova et al. 2014), yet it can act as a DAMP activating TLR4 (Mo et al. 2022). HSP90 can also be secreted in the CNS during neuroinflammatory conditions and is more consistently associated with detrimental effects in CNS disease models (Uddin et al. 2021; Alberti et al. 2021; Wang et al. 2017). Galectins are a family of proteins characterized by their glycan binding domains, and secreted galectins are similarly multifunctional. Galectins are implicated as regulators of neurogenesis as well as inflammation (Imaizumi et al. 2011; Metz et al. 2016; Cengiz et al. 2019). There are multiple reports of galectin‐3 expression being elevated in neurodegenerative disease including Alzheimer's disease, amyotrophic lateral sclerosis, and Parkinson's disease (Ramos‐Martínez et al. 2022; Soares et al. 2021). Multiple Sclerosis is associated with increased galectin‐9 abundance in patients (Burman and Svenningsson 2016), while the loss of galectin‐8 in mouse models of MS exacerbates symptoms (Pardo et al. 2017). Whether galectin secretion serves primarily protective, pathogenic, or compensatory functions appears to depend on the specific galectin paralog, cell type, and disease context, making this an area of active investigation rather than settled consensus.

2.3. Neuroprotective Functions of UcPS: Stress Adaptation, Repair, and Proliferation

In contrast to the pathological spread of aggregation‐prone proteins and the initiation and propagation of neuroinflammation described in Sections 2.1 and 2.2, several UcPS cargoes support neural resilience by regulating extracellular proteostasis, modulating oxidative stress responses, and promoting tissue repair programs such as angiogenesis and neurogenesis (Figure 1C).

A clear example is insulin degrading enzyme (IDE), a cytosolic protease that is secreted via UcPS (Qiu et al. 1998; Zhao et al. 2009; Son et al. 2016) and exerts well‐established neuroprotective effects. IDE contributes to extracellular proteostasis by degrading amyloid‐β peptides (Qiu et al. 1998; Zhao et al. 2009; Vekrellis et al. 2000), thereby limiting the formation of extracellular aggregates that are a hallmark of Alzheimer's disease. Astrocytes are the predominant source of IDE and secretion of IDE is known to be triggered by exposure of astrocytes to amyloid‐β peptides (Son et al. 2016). This response establishes a feedback mechanism in which the presence of aggregation‐prone peptides triggers secretion of a protease capable of degrading them. Consistent with this protective effect, lower serum IDE is correlated with more severe cognitive impairment in Alzheimer's disease patients (Tian et al. 2023). As such, in addition to fibrillar aggregation‐prone proteins triggering release of inflammatory molecules, they can simultaneously elicit compensatory UcPS pathways that enhance extracellular proteostasis.

In addition to proteostatic stress, oxidative stress is a common driver of dysfunction in neurodegenerative disorders (Olufunmilayo et al. 2023). Shifts in redox homeostasis can generate excessive reactive oxygen species that perturb mitochondrial function and oxidize proteins and lipids, which go on to activate inflammatory signaling cascades in neurons and glia. Oxidative stress has been shown to trigger the release of numerous UcPS cargoes including fibroblast growth factor 1 (FGF1) (Jackson et al. 1992), as well as the redox responsive protein PARK7 (DJ‐1) (Page et al. 2025). These proteins are expressed by neuronal and glial cell types (Klimaschewski and Claus 2021; Myers et al. 1995; Ito et al. 2005) and FGF1 increases in the serum and CSF of Alzheimer's patients (Mashayekhi et al. 2010), while FGF2 is increased in the CSF of MS patients (Sarchielli et al. 2008). In preclinical models of Huntington's disease and Parkinson's disease, FGFs have disease‐modifying effects (Klimaschewski and Claus 2021; Liu et al. 2021), including promoting neurogenesis and angiogenesis, and thereby promoting repair from injury (Klimaschewski and Claus 2021; Liu et al. 2021; Chen et al. 1994). PARK7 is more abundant in the frontal cortex and substantia nigra in idiopathic Parkinson's disease, and mutations in the PARK7 gene are associated with early onset Parkinson's disease (Bandopadhyay et al. 2004; Nuytemans et al. 2010). PARK7 is secreted by astrocytes, and extracellular PARK7 has been proposed to protect neurons from oxidative damage, including toxicity mediated by oxidized dopamine (Niere et al. 2016; Zhang et al. 2016; Solheim et al. 2026). Recent evidence from HeLa cells suggests that PARK7 is secreted through a lysosomal pathway initiated by autophagy, followed by fusion of lysosomes with autophagosomes and subsequent extracellular release through SEC22B/STX3/STX4‐dependent vesicle fusion with the plasma membrane. Whether this pathway similarly regulates PARK7 secretion in neuronal and glial cells remains to be determined (Dash et al. 2025, 2024).

Together these factors illustrate how UcPS can serve pro‐survival roles alongside pathological functions, underscoring the context‐dependence of UcPS outcomes in the CNS.

2.4. The Senescence Associated Secretory Phenotype (SASP) in the CNS

Cellular senescence is a durable state associated with growth arrest and the development of SASP, which collectively describes the release of a broad range of inflammatory factors from senescent cell populations (Wang, Han, et al. 2024; He and Sharpless 2017). Cellular senescence can be induced by a variety of cell stressors, including activation of the DNA damage response, mitochondrial stress/reactive oxygen species, and paracrine spread of senescence mediated by components of the SASP (Wang, Han, et al. 2024; He and Sharpless 2017). Key elements of the SASP cascade are driven by UcPS. The secretion of both IL‐1α and IL‐1β is dependent on activation of inflammatory caspases (Wiggins et al. 2019; Muela‐Zarzuela et al. 2024) and both are released by senescent cells. IL‐1α signaling has been shown to drive the expression and release of other SASP cytokines, including IL‐6 and IL‐8 through IL‐1 receptor (IL‐1R) signaling (Orjalo et al. 2009). IL‐1R signaling leads to expression of secretory products like IL‐6 and IL‐8, which are released by classical secretory pathways, and other products like HMGB1, which are released via UcPS (Gorgoulis et al. 2019; Hernandez‐Segura et al. 2017). While SASP broadly involves the release of classically secreted factors like IL‐6 and IL‐8, the core initiators and amplifiers of this cascade (IL‐1α, IL‐1β, and HMGB1) strictly depend on unconventional pathways. The long‐lived nature of senescent cells may reflect the utilization of gasdermin‐D (GSDMD) release pathways that do not culminate in pyroptosis or alternative, poorly characterized secretory mechanisms. In either case, UcPS of IL‐1α and IL‐1β are critical drivers of the SASP phenotype. Critically, the SASP is known to promote the spread of cellular senescence in a paracrine fashion to neighboring cells, including cells of the CNS (Acosta et al. 2013). Although cellular senescence and the SASP were first characterized and studied in fibroblasts, numerous key cell types of the CNS can adopt a senescent phenotype and have been implicated in the pathogenesis of many diseases, particularly neurodegenerative diseases, including microglia, astrocytes, and neurons (Walton et al. 2020; Li et al. 2025; Hussain et al. 2025; Han et al. 2020). Notably, although neurons are, by nature, terminally differentiated, they are known to acquire multiple senescent markers in Alzheimer's Disease (AD) that may contribute to the propagation of amyloid pathology in AD and other neurodegenerative diseases (Walton et al. 2020). As such, senescent cells are implicated in the pathogenesis of numerous diseases of the CNS, and both the induction and propagation of the SASP are known to be driven by factors that are released from cells via UcPS.

2.5. Retroelement‐Mediated UcPS in the Nervous System

The landscape of UcPS has also been expanded to include a number of retroelement derived proteins that have been evolutionarily exapted to mediate intracellular communication between cell types in the brain. These proteins originate from ancient genomic elements with structural similarity to retroviruses and retain the capacity to form higher‐order assemblies that behave analogously to viral particles while functioning within endogenous cellular contexts. These include activity‐regulated cytoskeleton‐associated protein (Arc), Paternally Expressed Gene 10 (PEG10), and Paraneoplastic Ma Antigen 2 (PNMA2) (Frank and Feschotte 2017; Frank et al. 2022). While these pathways are essential for physiological processes such as synaptic plasticity, as seen with Arc (Ashley et al. 2018; Pastuzyn et al. 2018), their dysregulation or aberrant secretion can also drive neuropathological states, exemplified by the autoimmune triggers associated with PNMA2 (Xu et al. 2024; Gröger et al. 2021; Gruchot et al. 2019).

Arc is the quintessential example of a repurposed retroelement facilitating non‐cell‐autonomous signaling. Derived from ancient Ty3/gypsy retrotransposons, Arc retains a Gag‐like architecture that allows it to self‐assemble into virus‐like capsids (Pastuzyn et al. 2018). Once released, these Arc EVs function as intercellular communication vessels. These capsids can be conceptualized as nanoscale ribonucleoprotein assemblies that encapsulate RNA cargo and protect it from degradation during intercellular transfer. Upon uptake by recipient neurons, Arc capsids release Arc mRNA for local translation. This delivered protein then triggers the endocytosis of surface AMPA receptors, effectively weakening the synaptic strength.

While it is known that PEG10 is indispensable for mammalian placental morphogenesis, the physiological implications of its conserved expression in the central nervous system remain largely uncharacterized (Abed et al. 2019). It maintains the ancestral ability to self‐assemble into virus‐like capsids that specifically package their own encoding mRNA (Campodonico et al. 2024). Although much of the current mechanistic understanding of PEG10 relies on non‐neuronal evidence, observing trends in its retained retro‐like properties allows us to hypothesize that PEG10 similarly supports RNA‐based intercellular communication within the CNS. However, the specific neural recipient cell types and the downstream functional consequences remain to be directly validated. The retrotransposon derived activity of PEG10 has been implicated in Angelman syndrome and Amyotrophic Lateral Sclerosis (ALS) through the modulation of the neuronal transcriptional landscape (Black et al. 2023). Therefore, the PEG10 UcPS underscores a critical paradigm in neurobiology: the evolutionary trade‐off of repurposing retroelements. It highlights that virus‐like transport in the brain is not only a driver of synaptic plasticity but also a process requiring constant homeostatic surveillance.

The PNMA family consists of several neuron‐specific proteins, including PNMA1, PNMA2, and PNMA3, that are predominantly expressed in the central nervous system. Like Arc and PEG10, the PNMA genes evolved from the domestication of ancient Ty3 retrotransposons, specifically co‐opting the Gag homology domains that allow for the formation of higher order protein assemblies (Xu et al. 2024; Iwasaki et al. 2013; Madigan et al. 2024). While their endogenous roles in the brain are still being elucidated, they have gained significant clinical attention due to their involvement in paraneoplastic neurological syndromes (PNS). These syndromes arise when immune responses directed against tumor antigens cross‐react with structurally similar neuronal proteins, leading to collateral damage within the nervous system. The release of these naked capsids appears to be a double‐edged sword. In the immune privileged environment of the brain, they likely participate in yet to be defined signaling pathways. Unlike soluble monomeric proteins, the fully assembled PNMA2 capsids act as potent triggers for the adaptive immune system, which can manifest as deficits in learning and memory.

Overall, the UcPS of retroelement derived proteins illustrates a diverse repertoire of evolutionary co‐option within the mammalian brain. In this regard, these proteins may represent an emerging class of proteins capable of mediating intercellular communication in the CNS. This expands the conceptual framework of neuronal communication beyond classical neurotransmitter release to include the transfer of macromolecular cargo such as RNA between cells. Retroelements and endogenous retroviruses which, despite being replication defective, may still express viral proteins, comprise substantially more of the human genome than is allocated to conventional genes driving protein expression (de Koning et al. 2011). This implies that a large fraction of the genome encodes proteins with latent capacities for self‐assembly and macromolecular packaging. While the degree to which other Gag‐related genes have been similarly co‐opted evolutionarily to mediate similar functions remains unclear, it is predicted that there are over 100 Gag‐like gene products that may be expressed and used to drive similar modes of intracellular communication (Campillos et al. 2006). Looking forward, the viral‐like biochemistry of these proteins offers a promising blueprint for bioinspired drug delivery systems. By leveraging their natural capacity to package and protect mRNA cargo, engineered versions of Arc or PEG10 capsids could serve as endogenous, low immunogenicity vehicles for neural gene therapy (Malfavon‐Borja and Feschotte 2015). This aligns with broader efforts in RNA‐based therapeutics, where efficient intracellular delivery and cell‐type specificity remain major technical barriers. Researchers have already begun to demonstrate that these Gag‐like proteins can be engineered to package specific mRNA cargo and can even be pseudotyped with various envelope proteins to achieve cell type specific delivery (Madigan et al. 2024; Gu et al. 2022; Segel et al. 2021). However, the pathological immunogenicity observed with PNMA2 highlights a critical engineering hurdle; successful therapeutic application will require precise structural modifications to antigenic domains to ensure these capsids can navigate the systemic immune system without triggering an immune response. Beyond immunogenicity, deploying these engineered retro‐like systems presents enormous technical barriers. Future research must resolve challenges regarding in vivo targeting specificity within complex neural networks, as well as overcome significant scalability and manufacturability limitations before clinical application is feasible.

3. Unconventional Protein Secretion Pathways

Two major routes have been described for UcPS (Figure 2). (1) Pore‐mediated secretion, in which proteins exit through plasma‐membrane pores formed either by the cargo itself, as in the case of FGF2 (Temmerman et al. 2008; Nickel 2011; Sparn et al. 2022), or by pore‐forming proteins such as gasdermin‐D downstream of inflammasome activation (Evavold et al. 2018; Xia et al. 2021). (2) Vesicular, pore‐independent secretion, in which cytoplasmic proteins are incorporated into organelles that subsequently fuse with the plasma membrane to release their contents (Neel et al. 2024; Semino et al. 2018). While these two routes appear to be mechanistically distinct, secretion of UcPS cargoes can occur via either pathway. For example, interleukin‐1β secretion has been shown to occur via both routes, with the dominant mechanism varying by cell type and stimulus (Semino et al. 2018; Karmakar et al. 2020). Below, we review the evidence for and mechanisms governing these pore‐mediated and vesicular UcPS pathways in neuronal and glial cells.

FIGURE 2.

Diverse routes of unconventional protein secretion (UcPS). (A) Fibroblast growth factor‐2 (FGF2) monomers self‐assemble at the plasma membrane to form a translocation pore, through which FGF2 monomers traverse to exit the cell. (B) Upon activation of the NLRP3 inflammasome, pro‐caspase‐1 is proteolytically processed into active caspase‐1, which cleaves both GSDMD and pro–interleukin‐1β (IL‐1β). The N‐terminal fragment of gasdermin D oligomerizes to form membrane pores that permit release of mature IL‐1β. (C) Cytosolic cargo is sequestered within phagophores that mature into autophagosomes; rather than undergoing lysosomal degradation, these vesicles, or autophagosome‐derived hybrid compartments, traffic to and fuse with the plasma membrane, resulting in extracellular release of their contents. (D) The endoplasmic reticulum–associated deubiquitinase USP19 recognizes intrinsically disordered proteins and, in conjunction with a cytosolic chaperone complex, mediates their delivery into endosomal compartments. These intermediates subsequently fuse with lysosomes and traffic to the plasma membrane, where fusion facilitates secretion of misfolded cargo. (E) A TMED10/HSP90‐containing chaperone complex recognizes and unfolds cytosolic substrates, promoting their channeling into the ER–Golgi intermediate compartment (ERGIC) via TMED10. Cargo is subsequently exported via a poorly defined vesicular intermediate that ultimately fuses with the plasma membrane to enable secretion. Created with BioRender.com.

3.1. Pore‐Mediated UcPS From Neurons and Glial Cells

A subset of UcPS cargoes translocate across the plasma membrane via protein‐conducting pores (Rabouille 2017). Elegant studies of fibroblast growth factor 2 (FGF2) secretion, a process that promotes angiogenesis, established this protein as the prototypical cargo that revealed this mechanism (Temmerman et al. 2008; Nickel 2011; Sparn et al. 2022). FGF2 promotes its own export by oligomerizing and forming transient membrane pores that permit its translocation. This process is initiated by phosphorylation of FGF2 followed by recruitment to the inner leaflet of the plasma membrane through interaction with the cytosolic domain of the α‐subunit of the Na,K‐ATPase (Zacherl et al. 2015). This interaction facilitates subsequent binding of FGF2 to the anionic phospholipid PI(4,5)P₂, which promotes FGF2 oligomerization and membrane pore formation (Temmerman et al. 2008; Steringer et al. 2012; Lolicato et al. 2022). FGF2 is subsequently captured through interactions with heparan sulfate proteoglycans on the outer leaflet, forming a transmembrane pore (Steringer et al. 2012; Kaur et al. 2025; Figure 2A). These mechanisms have largely been established by studying secretion of FGF2 from HeLa and CHO cells. Astrocytes are the major source of secreted FGF2 in the CNS (Kirby et al. 2013; Kajitani et al. 2012; Delgado‐Rivera et al. 2009), and extending these findings to primary astrocytes and other CNS cells that secrete FGF2 could reveal novel molecular targets to modulate neurogenesis. In addition to FGF2, phosphorylated Tau (Katsinelos et al. 2018; Merezhko et al. 2018) has been reported to use a similar mechanism, including binding to anionic phospholipids and glycosaminoglycans and insertion into lipid bilayers in reconstituted systems.

In addition to UcPS cargo‐driven pore formation, GSDMD mediates the secretion of cytoplasmic cytokines, including IL‐1β and IL‐18, downstream of inflammasome activation (Evavold et al. 2018; Xia et al. 2021). Inflammasomes are cytoplasmic multiprotein complexes that sense a wide range of sterile and infectious signals in the CNS and activate inflammatory caspases (Xu and Nunez 2023; Xu et al. 2025). GSDMD contains an N‐terminal membrane pore–forming domain that is autoinhibited through intramolecular interaction with its C‐terminus (Shi et al. 2015). This autoinhibition is relieved by caspase‐1–mediated cleavage (Shi et al. 2015; Ding et al. 2016), after which the N‐terminal fragment binds phospholipids on the inner leaflet of the plasma membrane (Balasubramanian et al. 2024), oligomerizes, and forms protein‐conducting pores. These pores permit the release of cytoplasmic proteins, including IL‐1β and IL‐18 (Evavold et al. 2018; Xia et al. 2021; Figure 2B).

Formation of plasma membrane N‐GSDMD pores can also trigger pyroptosis, an inflammatory lytic cell death (Liu et al. 2016; Chen and Broz 2024). Consequently, distinguishing bona fide pore‐mediated secretion from passive release due to pyroptotic lysis is essential. This is typically addressed using two criteria: (i) absence of lactate dehydrogenase (LDH) release, which reflects membrane rupture rather than GSDMD pore activity, and (ii) persistence of cytokine secretion in the presence of osmoprotectants such as glycine, which delay or prevent pyroptotic lysis without inhibiting N‐GSDMD pore function (Evavold et al. 2018; Zanoni et al. 2016). Indeed, confirming lysis‐independent release by monitoring cellular viability and the retention of large cytoplasmic molecules, while demonstrating secretion in parallel, is key to defining the regulated release of a UcPS cargo.

Lysis‐independent, N‐GSDMD–dependent secretion is well established in peripheral macrophages but remains insufficiently resolved in CNS cell types. In microglia, inflammasome activation, GSDMD processing, and release of IL‐1 family cytokines (IL‐1α, IL‐1β, IL‐18, IL‐33) have been clearly demonstrated (Luciunaite et al. 2020; Friker et al. 2020; Yoon et al. 2025; Freeman et al. 2017); however, these studies do not rigorously distinguish between lysis‐independent secretion and cytokine release secondary to pyroptotic membrane rupture. This distinction is particularly important in the CNS, given the unique ontogeny and turnover dynamics of CNS myeloid populations following inflammatory challenge (Yamasaki et al. 2014), including long‐lived, self‐renewing microglia as well as infiltrating monocyte‐derived macrophages (Fliegauf et al. 2026).

Consistent with a lysis‐coupled mechanism, microglial IL‐1α release triggered by Toxoplasma gondii infection is inhibited by glycine (Batista et al. 2020), indicating that cytokine release in this context is driven by membrane rupture rather than selective pore‐mediated export. In such settings, pyroptosis‐coupled cytokine release may promote recruitment and replacement by peripheral myeloid cells. In contrast, bona fide lysis‐independent secretion would be expected to remodel the CNS inflammatory milieu and influence neuronal function in the absence of overt cell loss. Furthermore, while GSDMD is best characterized as a UcPS conduit, gasdermin E, a related isoform, has also been shown to serve this function in peripheral myeloid cells (Zhou and Abbott 2021) and is expressed in neurons and other CNS‐resident cell types (Neel et al. 2023), suggesting a potential role in UcPS within the CNS.

Beyond microglia, inflammasome activation has been reported in neurons (Panicker et al. 2022; Kaushal et al. 2015), and astrocytes (Freeman et al. 2017; Liu and Chan 2014), with detection of secreted IL‐1 family cytokines that lack signal peptides. However, definitive evidence for pyroptosis‐independent release in these CNS cell types, either ex vivo or in vivo, remains to be defined.

3.2. Vesicle Mediated UcPS

While gasdermin pore–mediated release can be coupled to inflammatory cell death, secretion via fusion of vesicular intermediates with the plasma membrane provides an inherently lysis‐independent mechanism for UcPS. Multiple lines of evidence support the existence of a vesicular, pore‐independent pathway. Several cargoes lacking direct affinity for PI(4,5)P₂, including galectin‐3 and HMGB1, are nevertheless secreted, indicating that they are unlikely to self‐form pores in a manner analogous to FGF2. Moreover, UcPS can occur in the absence of inflammasome activation, where gasdermin‐mediated pore formation is not engaged (Nakamura et al. 2024). Supporting a vesicular mechanism, UcPS cargoes, including α‐synuclein, HMGB1, and IL‐1β, localize to autophagic and endolysosomal compartments (Burbidge et al. 2022; Sawai et al. 2025), and disruption of autophagosome formation or endolysosomal function, either pharmacologically or via genetic ablation of key regulators, inhibits their secretion (Burbidge et al. 2022; Sawai et al. 2025). Multiple, distinct vesicular routes have been proposed; here, we consider the sequential stages through which UcPS cargoes progress, including entry into vesicular compartments, trafficking to the cell surface, and membrane fusion enabling protein release.

Entry into vesicular intermediates during UcPS occurs through multiple mechanistically distinct pathways (Figure 2C–E). In neurons and glia, at least four routes have been described: sequestration within newly forming autophagosomes (Sawai et al. 2025), translocation across the ER–Golgi intermediate compartment (ERGIC) membrane (Jiao et al. 2024), endocytic uptake into endolysosomal compartments (Burbidge et al. 2022), or direct entry into endolysosomal compartments (Wu et al. 2023). The relative contribution of each route depends on cargo, cell type, and physiological or stress state. This is particularly relevant in the CNS, where long‐lived post‐mitotic cells must balance proteostasis, stimulus‐coupled secretion, and damage control within highly polarized membrane systems.

One major route is referred to as secretory autophagy. Early studies of Acb1 secretion in yeast showed that disruption of core autophagy genes inhibits secretion of this cytoplasmic protein, implicating autophagosome‐related membranes in UcPS (Duran et al. 2010). In the CNS, analogous evidence has accumulated for α‐synuclein. Autophagosome formation machinery is required for α‐synuclein secretion in cortical neurons, where stimulation enhances secretion alongside increased autophagic flux (Nakamura et al. 2024; Sawai et al. 2025). α‐Synuclein co‐localizes with LC3+ puncta, and secretion is inhibited by pharmacological suppression of autophagy, genetic ablation of Beclin‐1, or knockdown of autophagosome regulators in SH‐SY5Y cells (Nakamura et al. 2024; Sawai et al. 2025). These findings indicate that a fraction of α‐synuclein is incorporated into phagophore‐derived intermediates, and that this step is required for UcPS, consistent with an autophagosome‐machinery dependent secretory process.

In parallel, secreted fibrillar α‐synuclein enters human midbrain neurons via endocytosis and damages endolysosomal membranes, triggering the recruitment of galectin‐3 and activation of autophagosome formation machinery (Burbidge et al. 2022). This leads to engulfment of damaged organelles containing fibrillar α‐synuclein within hybrid endosome–autophagosome structures, which are proposed to fuse with the plasma membrane and promote aggregate spread. Thus, monomeric and fibrillar α‐synuclein access autophagosomal intermediates through distinct mechanisms: basal sequestration of a cytosolic protein versus stress‐induced engulfment of damaged, aggregate‐containing compartments. In neurons, endolysosomal injury caused by proteopathic assemblies can therefore secondarily drive UcPS by routing damaged organelles toward release‐competent compartments. Importantly, this mechanism is distinct from translocation‐based UcPS. Following engulfment of the damaged endolysosomal compartment, fibrillar α‐synuclein remains enclosed within the lumen of the resulting amphisomal compartment and is released in association with extracellular vesicles rather than by direct translocation across a membrane (Figure 2C–E). We do not discuss EV‐mediated release mechanisms in detail here, as these pathways have been extensively reviewed in the context of the CNS, and we direct readers to several recent reviews (Ikezu et al. 2024; Filannino et al. 2024; Manolopoulos et al. 2025; Wang, Zhang, et al. 2024).

A mechanistically distinct route operates at the ERGIC and is mediated by TMED10, which is proposed to act as a protein conducting channel (Zhang et al. 2020). TMED10 acts as a UcPS cargo receptor, requiring oligomerization to form a translocation competent channel (Zhang et al. 2020). The chaperone HSP90 binds both TMED10 and UcPS cargoes, including IL‐1 family cytokines, and is proposed to unfold UcPS cargoes prior to entry into the ERGIC (Zhang et al. 2020). While TMED10 dependence was initially demonstrated in HEK293T and HeLa cell lines, recently, IL‐33 (a signal‐peptide lacking member of the IL‐1 family) secretion from astrocytes has also been shown to require TMED10 (Jiao et al. 2024). A key unresolved issue is how cargo translocated into the ERGIC is prevented from entering conventional secretory trafficking. Evidence from HeLa cells suggests that ERGIC‐associated Rab2A interacts with TMED10 to promote sorting into UcPS intermediates rather than forward trafficking to the cis‐Golgi (Ge et al. 2023). Whether analogous sorting steps operate in highly polarized CNS cells remains unknown.

Misfolding‐associated protein secretion (MAPS) constitutes an alternative pathway in which misfolded proteins, including Tau and α‐synuclein, are targeted to endolysosomal compartments that subsequently fuse with the plasma membrane. Mechanistic studies, largely performed in non‐neuronal systems, indicate that this process depends on recognition of misfolded, ubiquitinated proteins by the ER‐associated, cytoplasm‐facing deubiquitinase USP19 (Lee et al. 2016), followed by engagement with a chaperone complex comprising HSC70 and DNAJC5 (Xu et al. 2018), and direct entry into late endolysosomal membranes. However, the identity of the translocation machinery remains unknown, and a role for TMED10 has not been established. In contrast to chaperone‐mediated autophagy, which degrades cytosolic proteins that directly enter lysosomes, MAPS is independent of LAMP2A. DNAJC5 is required for the secretion of misfolded α‐synuclein in human iPSC‐derived midbrain dopaminergic neurons (Wu et al. 2023), suggesting that disruption of this pathway may constrain aggregate propagation. However, whether MAPS is protective or maladaptive is likely to depend on the proteostasis context, as secretion may relieve intracellular proteotoxic stress while also promoting extracellular aggregation propagation.

There is ongoing investigation into shared machinery between MAPS and UcPS. SNARE complex members including SNAP23 or STX4, as well as vesicle sorting GTPases like RAB8A are implicated in regulating both degradative and secretory outcomes of α‐synuclein aggregates (Sawai et al. 2025). In glia, sorting receptor SORL1 is downregulated in microglia during Alzheimer's disease, and its deletion in stem cell derived microglia impairs both lysosomal degradation and lysosomal exocytosis, demonstrating these pathways are intertwined (Mishra et al. 2025). Thus, while the mechanistic understanding of MAPS, particularly in neurological contexts, remains incomplete, emerging evidence suggests that it may share machinery with other UcPS pathways and should be considered within the broader framework of proteostatic regulation.

3.3. Maturation and Fusion With the Plasma Membrane

While multiple organelles have been proposed as entry sites for UcPS cargo, not all are known to fuse directly with the plasma membrane to enable secretion. As noted above, several studies suggest that UcPS cargo initially enters autophagosomes, which may subsequently fuse with the plasma membrane; however, direct evidence for such fusion remains limited. A recent study using TIRF microscopy visualized mScarlet‐tagged LC3 at the plasma membrane of neurons under autophagic stress and reported punctate LC3 signals with kinetics consistent with SNARE‐mediated fusion events (Palumbos et al. 2025). These observations support a model in which autophagosome‐derived vesicles fuse with the plasma membrane. However, LC3 alone does not distinguish canonical autophagosomes from amphisomes or other LC3‐positive intermediates, and further validation using additional markers would strengthen assignment of vesicle identity.

Moreover, while these events are consistent with membrane fusion, TIRF imaging does not resolve membrane topology, and complementary approaches are required to define the orientation and fate of LC3 and associated cargo following fusion.

Among the organelles proposed to mediate UcPS, endosomes and lysosomes are well established to possess intrinsic plasma membrane fusion capacity. Accordingly, several models posit that autophagosomes fuse with endosomes or lysosomes prior to subsequent fusion with the plasma membrane (Burbidge et al. 2022). Recently, convergence on the lysosome as a central intermediate has been hypothesized (Neel et al. 2024), a compelling hypothesis given its role in integrating endosomal and autophagic pathways.

Within the context of UcPS, distinct SNARE combinations have been proposed to act at different stages of the pathway. VAMP7/STX4/SNAP23 has been proposed to mediate lysosome‐to‐plasma membrane fusion, consistent with a role in direct exocytic release from lysosome‐related compartments. By contrast, SEC22B/STX3/STX4/SNAP23 has been implicated in cargo release at a downstream step of secretory autophagy (Palumbos et al. 2025). In this model, the vesicular SNARE SEC22B interacts with STX3, STX4, and SNAP23 at the plasma membrane to facilitate fusion and release of UcPS cargo from secretory autophagosomes (Kimura et al. 2017). Similarly, in HeLa cells, SEC22B does not impact autophagosome‐lysosome fusion, but is essential for the secretion of PARK7 (Wang, Han, et al. 2024). These findings suggest that specific SNARE assemblies may operate at mechanistically distinct fusion during UcPS. However, the assignment of SEC22B exclusively to the terminal plasma membrane fusion step remains uncertain. SEC22B is enriched at the ER–Golgi intermediate compartment and functions canonically in ER–Golgi trafficking, potentially positioning it upstream of late exocytic events. Indeed, in neurons, it has instead been reported to promote plasma membrane expansion via a non‐fusogenic mechanism (Petkovic et al. 2014). Whether the SEC22B/STX3/STX4/SNAP23 complex mediates secretory autophagosome fusion with the plasma membrane in neuronal or glial cells therefore remains to be determined.

3.4. Retroelement‐Mediated UcPS in the Nervous System

In addition to investigation of the pathways mediating UcPS of proteins encoded by canonical host genes, the secretion of retroelement derived proteins is also being intensely investigated. These Gag‐like assemblies circumvent the canonical ER–Golgi secretory pathway, instead undergoing extracellular release via mechanisms analogous to those utilized by retroviral particles. Importantly, this viral‐like capsid transfer is mechanistically distinct from other established UcPS routes. Whereas pore‐mediated release (Section 3.1) relies on the direct membrane translocation of individual monomers, and vesicular UcPS (Sections 3.2 and 3.3) involves engulfing soluble or aggregated proteins into endolysosomal or autophagic organelles, retroelement secretion demands complex, higher‐order structural assembly prior to cellular exit. The UcPS of these capsids relies on specialized recruitment of the host cell's membrane‐sculpting machinery. In this context, membrane deformation and vesicle formation are driven by proteins that can sense and stabilize specific lipid curvatures, enabling the formation of outward buds from the plasma membrane. UcPS of Arc is mediated by its interaction with IRSp53 (Insulin Receptor Substrate p53), an Inverse‐BAR (I‐BAR) domain protein, which has also been shown to similarly enhance the release of HIV‐1 Gag from cells (Inamdar et al. 2021). IRSp53 acts as a critical molecular adapter that senses and induces negative membrane curvature and facilitates the budding of Arc into extracellular vesicles (EVs) (Ravens et al. 2024). I‐BAR domain proteins bind phosphoinositide‐rich regions of the plasma membrane and generate protrusive, outward curvature, a geometry that is well‐suited for vesicle scission away from the cytosol. This IRSp53‐dependent pathway allows Arc to bypass the classical secretory route, instead utilizing direct plasma membrane budding to exit the neuron. Thus, the IRSp53‐mediated Arc secretion represents a sophisticated co‐option of cellular membrane remodeling to enable viral‐like intercellular communication in the central nervous system (Hantak et al. 2021; Zappulli et al. 2016).

Sharing a common Ty3/mdg4 retrotransposon ancestry with Arc, PEG10 represents a functionally distinct axis of UcPS in the mammalian brain (Iwasaki et al. 2013; Segel et al. 2021). Similar to Arc UcPS, PEG10 is primarily characterized by its encapsulation within membrane‐enveloped EVs, which provide a protected environment for the delivery of genetic cargo to recipient cells (Shiura et al. 2023). These vesicles shield RNA cargo from extracellular nucleases and can facilitate uptake through endocytic pathways in recipient cells. The UcPS of PEG10 is tightly coupled to the cell's proteostatic machinery (Pandya et al. 2021). In the central nervous system, PEG10 protein levels and its subsequent secretion are strongly dependent on the E3 ubiquitin ligase UBE3A. This regulatory link is of significant clinical importance; the loss of UBE3A‐mediated degradation leads to the accumulation and potentially increased secretion of PEG10, a mechanism recently implicated in the pathophysiology of Angelman syndrome (Pandya et al. 2021). This establishes a direct connection between protein quality control pathways and the regulation of intercellular RNA transfer mechanisms in neurons.

PNMA2 exhibits a unique UcPS profile that distinguishes it from the membrane enclosed export of Arc and PEG10. Recent structural analyses reveal that PNMA2 spontaneously self‐assembles into small, icosahedral virus‐like capsids with an outer diameter of approximately 20–21 nm and is secreted from neurons and tumor cells as non‐enveloped capsids. Icosahedral symmetry reflects a highly ordered assembly of repeating subunits, a structural principle commonly observed in viral capsids that maximizes stability while minimizing genetic coding requirements. This mechanism allows them to exit the cell in a manner reminiscent of non‐enveloped viruses like adeno‐associated viruses (AAVs). In contrast to vesicle‐mediated export, this mode of release does not involve a surrounding lipid bilayer, exposing the capsid surface directly to the extracellular environment. Thus, while Arc and PEG10 utilize membrane enveloped vesicles for physiological signaling, PNMA2 is distinctively released as a small, non‐enveloped or naked capsid.

4. Conclusion and Outstanding Questions

Much of the mechanistic insight into UcPS derives from studies in myeloid cells or immortalized lines. Evidence in primary CNS cell types remains comparatively limited, although recent studies have begun to identify conserved features across cell types. Current data support a model in which CNS UcPS cargoes access secretory intermediates through at least four mechanistically distinct entry routes: (i) phagophore capture during autophagosome biogenesis, (ii) endocytic uptake, (iii) TMED10‐dependent translocation across the ERGIC membrane, and (iv) HSC70/DNAJC5‐mediated delivery into endolysosomal compartments. A central unresolved question is how cargo is selectively routed among these pathways in different neural cell types. For proteins such as α‐synuclein, routing likely depends on conformational state, subcellular localization, and the proteostatic status of the cell, including responses to membrane damage. In contrast, for cytokines such as IL‐33, access to chaperone‐assisted translocation machinery at the ERGIC may be the dominant determinant. Defining these sorting principles in neurons and glia is essential for understanding how UcPS contributes to both physiological intercellular communication and the propagation of neuroinflammatory and neurodegenerative pathology.

Despite the identification of multiple UcPS regulators, factors uniquely dedicated to this pathway, without broader roles in intracellular trafficking, remain poorly defined. For example, TMED10 and ATG7 influence both UcPS and conventional secretion or global proteostasis (Li et al. 2023). Accordingly, vesicular UcPS is best conceptualized as a repurposing or rerouting of trafficking and degradative pathways toward secretion. Although multiple organelles carrying UcPS cargo have been described, it remains unclear whether these pathways converge on a shared intermediate prior to secretion, or whether a common trigger initiates release. Lysosomal convergence has emerged as a compelling model; however, lysosomal exocytosis is Ca²⁺‐dependent, and it remains unresolved whether stimuli that induce UcPS uniformly elevate intracellular Ca²⁺ levels.

The structural determinants and post‐translational modifications that target proteins for UcPS also remain incompletely defined. No universal feature analogous to a signal peptide has been identified. Instead, cargo selection appears context dependent. For example, palmitoylation of DNAJC5 regulates vesicular secretion of α‐synuclein (Wu et al. 2023). IL‐1β secretion requires proteolytic maturation, and mature IL‐1β is unfolded by an HSP90‐dependent chaperone complex and translocated through a TMED10 oligomer at the ERGIC and also has higher affinity toward GSDMD relative to full‐length pro‐IL‐1β. Systematic analysis of post‐translational modifications across UcPS substrates remains an important area for investigation. Proteomic approaches, combining genetic or pharmacologic perturbation of defined UcPS machinery with secretome profiling, offer a tractable strategy to identify cargoes that utilize shared export routes (Poschmann et al. 2022; Tang et al. 2026, 2023; Abbineni et al. 2022).

Given the central role of misfolded protein aggregates in multiple CNS diseases, UcPS represents a pathway with broad pathogenic and therapeutic implications. Secretion of misfolded proteins may alleviate proteotoxic stress at the level of individual cells while promoting intercellular spread and broader tissue dysfunction. This trade‐off introduces a fundamental layer of complexity in neurodegenerative disease and warrants careful consideration when targeting UcPS therapeutically, as the consequences are likely to be cargo dependent. Inhibiting secretion may reduce pathological propagation while exacerbating intracellular proteotoxic stress. It remains unclear how engagement of UcPS pathways intersects with the unfolded protein response, or how sequestration of aggregates into secretory intermediates impacts cellular fitness. As molecular determinants of aggregate secretion are defined, each component represents a potential point of therapeutic intervention across proteinopathies. Conversely, the UcPS pathways present an opportunity to enhance the secretion of neuroprotective factors. Proteins such as PARK7, IDE, and select galectins exert protective effects in neurodegenerative contexts. Targeted modulation of their production and UcPS may therefore represent an alternative strategy for therapeutic intervention.

Author Contributions

Steven A. Krauklis: conceptualization, investigation, writing – original draft, writing – review and editing. Sevnur Kömürlü Keçeli: conceptualization, investigation, writing – original draft, writing – review and editing. Prabhodh S. Abbineni: conceptualization, investigation, writing – original draft, writing – review and editing, supervision. Edward M. Campbell: conceptualization, investigation, writing – original draft, writing – review and editing, supervision.

Funding

This work was funded by National Institutes of Health (NIH), National Institute of General Medical Sciences grant R00GM141268 (P.S.A.).

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.