Damage-associated molecular patterns (DAMPs)/alarmins are extracellularly released from the cell upon various types of stress and exert inflammatory or inflammation-related actions. Representative DAMPs/alarmins include high mobility group box 1 (HMGB1) protein, heat shock proteins, extracellular cold-inducible RNA-binding protein (eCIRP) and various isoforms of S100 proteins ^[1][2][3][4][1,2,3,4]. As most of them are known to have important physiological roles in the cell ^[5][6][7][8][5,6,7,8], the mechanisms underlying the extracellular release of DAMPs/alarmins have attracted the concerns of many investigators. Of interest is the fact that most of these proteins are released in a way of non-classical and non-vesicular release, which is distinct from exocytosis, as seen in neurotransmitters and peptide hormones. Current studies have revealed that some of representative DAMPs/alarmins use the release modes via gasdermin D (GSDMD) and mixed-lineage kinase domain-like pseudokinase (MLKL) pores ^[9][10][9,10].

The classical mode of secretion mechanisms has been discussed as exocytosis. It is a regulated type of exocytosis of hormones, neurotransmitters and digestive enzymes. In neurons and endocrine cells, a small percentage of secretory vesicles are fused with the plasma membrane upon cell stimuli, whereas the majority remains in a filamentous network of synapsins or actin in the case of neurons and endocrine cells, respectively, in reserve for subsequent stimulation [11]. In these mechanisms, several Ca²⁺-binding proteins are involved in molecular processes, such as tethering, docking, priming and fusion, in which Ca²⁺ sensor synaptotagmin-1, vesicular SNARE synaptobrevin (and homologue) and target SNAREs syntaxin/SNAP-25 (and homologues) play key roles in vesicle docking and fusion to the plasma membrane of nerve endings. Details have been reported elsewhere [12].

Unlike regulated exocytosis, there is a non-classical or constitutive vesicular exocytosis during the secretion of materials, such as collagen and extracellular matrix proteins (fibroblasts) [13] or bone matrix proteins (osteoblasts) [14]. Calcium signaling and cytoskeletal dynamics are also involved via a variety of signaling pathways and cellular processes. There are two constitutive types of vesicular release: exosome release and lysosome-mediated release. They are generated via a fusion of multivesicular bodies (MVBs) with plasma membranes and released into the extracellular space [15]. Exosomes play roles as carriers of several molecules, such as DNA, RNAs (miRNAs and small RNAs), proteins (signaling proteins and heat shock proteins) and lipids. Some of them are incorporated into other cells and exert biological actions [15]. The biogenesis of exosomes begins with the budding of endosomes, which is followed by the formation of MVBs and either the extracellular release of exosomes or their degradation at lysosomes. The manner of extracellular release uses the process of vesicular docking and fusion with SNAREs complexes and the endosomal sorting complex required for transport (ESCRT), consisting of ESCRT-0, I, II and III and the ATPase Vps4 complex [16]. On the other hand, lysosomes are also part of extracellular release. Digested and waste materials in lysosomes are extracellularly released into and contained in the extracellular matrix or are fused with the membrane.

Bacterial endotoxin lipopolysaccharides (LPS), when administered to the body at high levels, cause septic shock. LPS is known to stimulate Toll-like receptor 4 and produce proinflammatory cytokines such as IL-1β and TNFα via the activation of NF-κB. LPS is also reported to produce IL-1β and IL-18 via the activation of caspase-1 and release these cytokines through a GSDMD pore in the plasma membrane [17]. LPS-mediated cytokine (IL-1β or IL-18) release is mediated through a membrane pore made of GSDMD N-terminus peptide oligomers, which are cleaved by activated caspase-1. The GSDMD pore also causes a membrane rupture and pyroptotic cell death, which further enhances cytokine release. However, the GSDMD pore does not allow the release of the high mobility group protein B1 (HMGB-1) caspase-1 p20 subunit due to its pore size limit [18]. On the other hand, HMGB-1 is released from necrotic, necroptotic, pyroptotic, and ferroptotic cells ^[19][20][21][19,20,21]. Necroptosis occurs in various diseases including tumor necrosis factor (TNF)-mediated systemic inflammation and ischemic reperfusion injury. Activated TNF receptor-1 (TNFR-1) triggers signaling complex 1, composed of TNFR associated via the death domain (TRADD), TNFR-associated factor 2 (TRAF2), receptor-interacting protein kinase 1 (RIPK1), cellular inhibitors of apoptosis (cIAPs), the linear ubiquitin chain assembly complex (LUBAC), transforming growth factor-β-activated kinase 1 (TAK1), and the inhibitor of κB kinase (IKK) complex, which activate NF-κB and mitogen-activated protein kinase (MAPK) ^[22][23][22,23]. When NF-κB activation is blocked, TRADD is dissociated from complex I, forms complex II and activates caspase 8, which promotes apoptosis. When caspase 8 activity is further blocked, complex II evolves into a necrosome, composed of RIPK1, RIPK3 and MLKL, leading to the phosphorylation of MLKL and the subsequent formation of the membrane pore made of MLKL oligomers. These mechanisms cause a necroptosis and membrane rupture for the release of larger DAMPs/alarmins, such as HMGB-1 [23]. Regarding HMGB-1 release, it is also reported that all-thiol type HMGB-1 is released from the ruptured membrane, while disulfide type HMGB-1 is released by lysosome-mediated exocytosis [24].

3. New Type of Non-Classical and Non-Vesicular Release

3.1. Identification of Prothymosin α Causing Cell Death Mode Switch

Researchers discovered prothymosin α (ProTα), which inhibits necrotic neuronal death, from a conditioned medium of cortical neurons [25] and observed that it is released from neurons and astrocytes upon starving or ischemia–reperfusion stress in a unique non-classical and non-vesicular manner [26].  When freshly prepared cortical neurons from 17-day-old embryonic rat brains were cultured in serum-free (no supplement) and low-density (1 × 10⁵ cells/cm², LD) conditions (Figure 1A), more than 80% of neurons died in a manner of necrosis within 12 h (Figure 1B). Transmission electron microscopy (TEM) showed that there is a decrease in the electrical density of cytosol, a damaged plasma membrane and swollen mitochondria, while substantially no change in the nucleus, all of which indicate necrosis. The nature of neuronal death was also characterized by cytochemical analyses using propidium iodide (PI, necrosis marker). Of interest is the finding that neurons cultured in 5 × 10⁵ cells/cm² (HD) showed an increase in survival activity; apoptotic features, such as nuclear fragmentation (TEM); and immunocytochemical changes, including the externalization of annexin V and activated caspase-3 [25].  When the conditioned medium (CM) from neurons in the HD culture was added to the LD culture, the survival activity was markedly increased (Figure 1B), and the PI signal was largely inhibited. Using a cytochemical assay with PI staining for the screening, the researchers purified the necrosis-inhibitory factor from the CM and identified that it was nuclear protein ProTα [25].  The addition of recombinant ProTα reproduced not only the inhibition of necrotic features but also caused apoptotic features (Figure 1C). It should be noted that the apoptosis caused by ProTα is prevented by the addition of growth factors, such as brain-derived neurotrophic factor (BDNF), as shown in an in vitro culture study using cortical neurons [25] and an in vivo study in a rat middle cerebral artery occlusion (MCAO) model [27] and a mouse retinal ischemia–reperfusion model, respectively [28].

3.2. Serum-Free Starvation-Induced Extracellular Release of ProTα

In the serum-free culture of primary neurons, ProTα release into the CM started as early as 1 h after the start of culturing, and it was time-dependent till 12 h [25]. Immunocytochemical and Western blot analyses revealed that the level of ProTα in the nuclei of neurons and astrocytes was markedly decreased (Figure 2). At 3 h, cell contents were reduced to a little less than 50% of the initial cell contents of cortical neurons and astrocytes, while CM levels were a little more than 50% of the initial cell contents [26]. However, in any condition, ProTα was not detected in the cytosol (Figure 2). As no PI signals in the nucleus were observed in serum-free cultured neurons at 3 h [25], ProTα release is unlikely caused by a plasma membrane rupture. As released ProTα is considered to have beneficial survival effects in the brain [29], it is interesting to examine the mode of ProTα release. As ProTα levels in the CM from the culture of neurons and astrocytes were not affected by the pretreatment with brefeldin A, which inhibits protein transport from the endoplasmic reticulum to the Golgi complex [30], it is presumed that the mode of ProTα release is not vesicular. For the purpose of detailed cellular and molecular mechanisms of ProTα release, researchers used C6 glioma cells to study the serum deprivation-induced extracellular ProTα release since the release was not also affected by brefeldin A [26].