Understanding the mechanism of IL-1β secretion

1Interleukin-1β (IL-1β) is a potent pro-inflammatory cytokine that is crucial for host-defence responses to infection and injury. It is also the best characterised and most studied of the 11 IL-1 family members. It is produced and secreted by a variety of cell types although the vast majority of studies have focussed on its production within cells of the innate immune system, such as monocytes and macrophages. It is produced as an inactive 31 kDa precursor, termed pro-IL-1β, in response to molecular motifs carried by pathogens called ‘pathogen associated molecular patterns’ (PAMPs). PAMPs act through pattern recognition receptors (PRR's) on macrophages to regulate pathways that control gene expression. Induction of pro-IL-1β expression is generally referred to as a priming step, and is an inefficient secretion stimulus. The primed cell must now encounter a further PAMP, or DAMP (danger associated molecular pattern, endogenous molecules released from dead cells) to induce the processing and secretion of an active IL-1β molecule.

2Pro-IL-1β is cleaved by the pro-inflammatory protease caspase-1. The activation of caspase-1 occurs via recruitment to a multi-protein complex termed the inflammasome (Fig. 1A); a molecular scaffold composed of adaptor molecules, a cytosolic pattern recognition receptor, and pro-caspase-1. The best characterised inflammasome, is formed by the cytosolic PRR NLRP3. NLRP3 has a tripartite structure with a PAMP/DAMP sensing C-terminal leucine rich repeat (LRR), a central nucleotide binding (or NACHT) domain and an N-terminal pyrin domain (PYD). The PYD domain of NLRP3 recruits the adaptor molecule apoptosis-associated speck-like protein containing a caspase recruitment domain (ASC) via a homotypic interaction with its PYD domain. Likewise, pro-caspase-1 is recruited to ASC via a homotypic interaction of CARD domains facilitating caspase-1 activation. Following caspase-1-dependent processing of pro-IL-1β, mature IL-1β is rapidly secreted from the cell. However, how IL-1β is secreted from the cell is not clear. Multiple observations and proposals have been made that do not allow the formation of a single unified mechanism of secretion, but rather suggest that there are multiple mechanisms. These of course may not be mutually exclusive, but may each make a specific contribution to IL-1β-dependent inflammation. Here, we attempt to summarise these observations and to provide a rationale for the existence of the multiple mechanisms proposed.

3The endoplasmic reticulum (ER) and Golgi apparatus together form the endo-membrane system responsible for the targeting of the vast majority of proteins to the extracellular space and to specialised sub-cellular compartments. So typical is this mechanism for protein export, it has attained the status of being the ‘conventional’ pathway when compared to the secretion of a handful of proteins that utilise alternative ‘non-conventional’ routes of cellular exit. Typically proteins are co-translationally translocated into the ER, although some are post-translationally translocated. The first step in the co-translational pathway of translocation is the detection of a signal sequence at the N-terminus of the nascent peptide, as it emerges from the ribosome, by the signal recognition particle (SRP). The SRP bound to the ribosome and nascent protein chain then binds to the SRP receptor (SR) anchored on the ER membrane and subsequently the nascent peptide chain is translocated through the protein conducting channel formed by sec61α and sec61γ subunits into the ER lumen.

4Once in the lumen of the ER the signal peptide is removed from the nascent protein chain which then, with the aid of chaperones, assumes a correctly folded state and accumulates at the ER exit site by virtue of its binding to the coat protein complex II apparatus (COPII). The COPII complex facilitates the budding of COPII-coated vesicles (that contain the cargo protein) from the ER and that subsequently fuse with the Golgi complex thereby depositing the cargo protein for the next stage of the secretory pathway. The cargo proteins continue their journey through the Golgi receiving further post-translational modifications before they are packaged and transported to their final destination; another organelle, or exocytosis from the cell. In order to maintain organelle homeostasis, and for retrieval of proteins back to the ER, Golgi membrane needs to be recycled, and this is achieved via the budding, retrograde transport, and fusion of COPI coated vesicles. COPI-mediated retrograde transport is inhibited by the fungal metabolite brefeldin A (BFA), and treatment of cells with BFA causes the collapse of the Golgi and its redistribution into the ER, and thus completely inhibits conventional protein secretion.

5That, in essence, is the route taken by conventionally secreted proteins to exit the cell (Fig. 1B). The knowledge that IL-1β lacked a signal peptide led Anna Rubartelli, in 1990, to ask whether IL-1β utilised the conventional pathway of secretion by treating LPS-activated monocytes with BFA. Whilst the secretion of IL-6 and TNFα were inhibited by BFA treatment, the secretion of IL-1β was actually enhanced, thus providing the first direct evidence that IL-1β follows a non-conventional route of protein secretion.

6When discussing the literature on the secretion of IL-1β, it is important to consider the stimulus inducing release. IL-1β is released in response to many PAMPs and DAMPs which can activate a variety, or in some cases multiple, PRR's to form inflammasomes. As discussed below, the mechanism of secretion may be influenced by stimulus type and strength. Here we summarise some of the best known and most widely used stimuli that induce IL-1β secretion. Extracellular ATP acting via the P2X7 receptor induces caspase-1-dependent release of IL-1β, and is dependent upon the formation of the NLRP3 inflammasome. The toxins nigericin and maitotoxin, as well as infection with Staphylococcus aureus also induce NLRP3-dependent IL-1β release. Monosodium urate (MSU) crystals, calcium pyrophosphate dihydrate (CPPD) crystals, cholesterol crystals, silica crystals and aluminium salts, and aggregated beta-amyloid are also known activators of NLRP3. Infection of macrophages with Salmonella typhimurium, Shigella flexneri, Legionella pneumophila, and Pseudomonas aeruginosa all induce IL-1β secretion via activation of an NLRC4 inflammasome. The NLRP1 inflammasome is activated by muramyl dipeptide (MDP), and in mouse, NLRP1b-inflammasome is activated by Bacillus anthracis lethal toxin. The DNA sensing receptors AIM2 and RIG-I also form inflammasomes. Infection of macrophages with Listeria monocytogenes induces the activation of caspase-1 via NLRP3, NLRC4 and AIM2 inflammasomes.

7Prior to Rubartelli's observations in 1990, it was known that IL-1β is expressed as a precursor and that it lacks a signal sequence, and also, that it is absent from the ER and the Golgi of LPS-activated monocytes as determined by immunoelectron microscopy. IL-1β is translated on free polyribosomes associated with the cytoskeleton, and not membrane-bound polyribosomes. The vast majority of IL-1β in LPS-activated monocytes localises to the cytosol, although a fraction resides in vesicles and is protected from tryptic digestion. These IL-1β-containing vesicles were subsequently identified as being of endolysosomal in nature. It was proposed that a fraction of cellular IL-1β is targeted to lysosomes for degradation, but that this fraction can be rescued by triggering lysosome exocytosis and thus secretion of IL-1β. Indeed, co-incubation of LPS-stimulated monocytes with the protease inhibitors pepstatin and leupeptin enhances the release of IL-1β supporting this hypothesis. How does cytosolic IL-1β get sequestered into vesicles? In 1990 Rubartelli suggested autophagy may provide a route and recent studies suggest that this may in fact be the case. Autophagy is a mechanism whereby damaged organelles or proteins in the cytosol become enclosed in a double membrane structure forming an autophagosome. These vesicles subsequently fuse with lysosomes forming autolysosomes resulting in the proteolytic degradation of their contents. LPS treatment of macrophages induces the recruitment of IL-1β to autophagosomes. When autophagy is inhibited this IL-1β is secreted; when autophagy is activated the sequestered IL-1β is degraded. Thus a fraction of cellular IL-1β targeted for degradation can be rescued and redirected to the extracellular environment.

8Another route that a fraction of cellular IL-1β can take out of the cell is via the shedding of microvesicles from the plasma membrane. This was originally observed in P2X7-receptor-stimulated LPS-treated THP-1 cells, and thrombin activated platelets. In THP-1 cells shedding of IL-1β-containing microvesicles is preceded by flip of the lipid phosphatidyl serine (PS) to the outer leaflet of the plasma membrane, and in astrocytes requires the activation of acid sphingomyelinase. The IL-1β contained in shed microvesicles is bioactive and may be released following contact with IL-1 receptor (IL-1RI) expressing cells. IL-1β-containing microvesicles have also been isolated from P2X7-receptor-stimulated LPS-treated microglia, astrocytes, and dendritic cells. Shed microvesicles from dendritic cells, containing IL-1β and caspase-1, express P2X7-receptors in their membrane. ATP-stimulation of these microvesicles induces the release of their contents and thus provides a mechanism for how this protected IL-1β can be released at target sites to elicit cellular responses.

9IL-1β can also be secreted in a protected form by being packaged and secreted via exosomes. Exosomes are small vesicles (50–80 nm vs. 100–600 nm for microvesicles ) that are secreted from multi-vesicular bodies (MVB's) or late endosomes. They are formed by the inward budding of the MVB membrane and contain cytosol. Thus cytosolic IL-1β could be sorted into the forming exosome via sorting protein complexes, although this has yet to be shown. The secretion of exosomes following P2X7-receptor stimulation of LPS-treated macrophages is dependent upon ASC and NLRP3, but independent of caspase-1. What is the function of protected IL-1β? The fact that much of the IL-1β secreted from the cell appears to be directly available (see below), and that IL-1β has a very short half life in plasma, suggests that protected IL-1β is destined for signalling processes at sites distant to the local inflammatory lesion. Both shed microvesicles and exosomes from antigen presenting cells contain MHC II molecules and induce immunomodulatory effects. Thus IL-1β could modulate immune responses induced by MHC II antigenic peptides carried by these vesicles.

10A consequence of caspase-1 activation in macrophages following infection by NLRC4-activating pathogenic bacteria (see Section 3.1) is a rapid, and caspase-1-dependent cell death called pyroptosis. Pyroptosis is a pro-inflammatory form of cell death that causes an infected macrophage to kill itself, and at the same time release pro-inflammatory cytokines such as IL-1β and another caspase-1 substrate IL-18. The rapid caspase-1-dependent pyroptotic cell death caused by S. typhimurium, as well as the caspase-1-dependent clearance of NLRC4-activating pathogens in vivo do not depend upon IL-1β processing, although that does occur. It seems that pyroptosis serves principally to eliminate the intracellular niche required for pathogen growth. Pyroptosis of macrophages following infection with S. typhimurium occurs after the caspase-1-dependent formation of pores (1.1–2.4 nm in diameter) in the plasma membrane. These pores cause the dissipitation of ionic gradients and the osmotic lysis of the cell. This osmotic lysis can be inhibited by the presence of glycine, yet pore formation and IL-1β release is not blocked. These caspase-1-dependent pores may provide a conduit through which IL-1β passes to reach the extracellular space.

11However, cell death associated with IL-1β release is not only dependent upon infection with NLRC4-activating pathogens. Over 20 years ago it was reported that stimulation of LPS-primed peritoneal macrophages with ATP or allospecific cytotoxic T-lymphocytes induces cell death in addition to IL-1β processing and release. 30 min incubation with the NLRP3-activating stimulus ATP causes LPS-primed murine peritoneal macrophages to round up and bleb, which is closely followed by the release of the cytolytic marker lactate dehydrogenase (LDH). Although release of mature IL-1β precedes release of LDH, is does raise the suggestion that IL-1β release signals a commitment to cell death. ATP-induced death of LPS-treated mouse peritoneal macrophages is caspase-1-dependent. Incubation of macrophages with glycine inhibits ATP-induced cell lysis, but not the release of mature IL-1β, similar to the phenomenon observed following S. typhimurium infection described above. In LPS-activated monocytes treated with the NLRP3-activating stimuli, heat killed S. aureus, caspase-1 and IL-1β localise to the plasma membrane, prompting the suggestion that caspase-1 may gate a membrane pore through which IL-1β transits. Bone marrow derived macrophages transduced to express the IL-1 family member IL-36α, and a non-cytokine Green Fluorescence Protein (GFP), when stimulated with LPS and ATP release IL-1β, IL-36α and GFP in parallel, suggesting release is due to a change in membrane integrity. Thus, there are many similarities between NLRC4-mediated pyroptosis and NLRP3-mediated cell death following P2X7-receptor stimulation. The mechanisms of release described in the previous sections however are reported to be non-cytolytic (see Sections 3.2.1 and 3.2.2) even though the use of ATP as a secretion stimulus is common. The possible reasons for the differences in IL-1β release and commitment to cell death are discussed in detail below.

12Cell lysis per se does not automatically induce release of processed IL-1β. LPS-stimulated murine peritoneal macrophages release exclusively pro-IL-1β when injured by scraping, excessive heat, freeze thaw or oxidative injury by H2O2 treatment. Likewise, when treated with the Ca2+ ionophore A23187, or the detergent saponin, unprocessed pro-IL-1β is the only form of IL-1 found extracellularly. Although not biologically active, the release of pro-IL-1β following necrotic cell death may not be without relevance. In models of sterile tissue injury and acute arthritis, a caspase-1-independent activation of IL-1β is reported. A number of neutrophil derived proteases are known to cleave pro-IL-1β into biologically active molecules, with a demonstrated role for proteinase 3 in acute arthritis. Thus, released pro-IL-1β following tissue injury and cell necrosis could be cleaved extracellularly by proteases derived from infiltrating neutrophils resulting in an IL-1β-dependent inflammatory response independent of caspase-1. Pro-IL-1β can also be processed into biologically active molecules by proteases from S. aureus and Candida albicans suggesting that extracellular processing may also occur at inflammatory lesions caused by infection.

13Cell type, species, source and concentration of PAMPs and DAMPs, not to mention their nature and the duration of the stimulation are all factors that introduce variation and that may account for the predominance of one mechanism of secretion over another in many of the studies discussed above. Other important factors are likely to include the cells microenvironment, where temperature and pH, redox balance, osmolarity, culture conditions and time in culture influence IL-1β release, as does the polarisation state of the macrophage (i.e. M1-M2). Thus the strength of the inflammatory input as perceived by the cell may vary widely from one study to another. It may thus be possible to view the different observations on the mechanisms of IL-1β secretion as belonging to a spectrum, or continuum, of release (Fig. 2). Depending upon the influencing factors described above, a cell may be induced to release a low level of IL-1β without a commitment to cell death. As the inflammatory insult escalates more mechanisms become engaged or employed to the extent that the cell can no longer maintain the latency of the plasma membrane and a commitment to cell death results. For example, microvesicle shedding is suggested as an early and rapid mechanism for the release of IL-1β, although it is followed by a slower, non-protected release of IL-1β from the cell. Following ATP-induced P2X7 receptor activation on macrophages and dendritic cells both exosomes and shed microvesicles are detected in the culture supernatant. Lysosome exocytosis during pyroptosis does occur, although in this case is suggested as an attempt by the cell to repair caspase-1-dependent membrane pores rather than as a mechanism of cytokine secretion. These studies suggest that multiple mechanisms may be engaged for the release of IL-1β within the same cell population.

14Although essential for resistance to infections, IL-1β also exacerbates damage during chronic disease and acute tissue injuries. Thus the release mechanisms of IL-1β represent a therapeutic target. ATP-induced IL-1β release is abolished in macrophages isolated from caspase-1 KO mice. Given that processing and secretion are thus so closely coupled, perhaps the most effective way of inhibiting IL-1β release is by inhibiting caspase-1. Peptide inhibitors corresponding to the cleavage site on IL-1 (YVAD) are effective and widely used inhibitors of caspase-1, and block release of IL-1β. Indirect inhibitors of caspase-1 are also effective. The sulfonyl urea drug glyburide, and related sulfonyl ureas inhibit ATP-induced IL-1β release by inhibiting the NLRP3 inflammasome. Glyburide also inhibits the accumulation of IL-1β into secretory lysosomes. Lysosomal membrane rupture and cathepsin B activity are important for the activation of the NLRP3 inflammasome in response to some stimuli, where inhibition of cathepsin B attenuates IL-1β release. In the past many diverse pharmacological effectors such as the phospholipase inhibitor bromoenol lactone, the protein phosphatase inhibitor okadaic acid and hypertonic buffer, anion transport inhibitors, alkylating agents, redox active drugs, histone deacetylase inhibitors, acid sphingomyelinase inhibitors, have been reported to inhibit IL-1β release. Given that the mechanisms of action and targets of many of these drugs remain unclear there is great potential for further discovery and exploitation of this pathway.

15IL-1β is a potent pro-inflammatory cytokine produced by cells of the innate immune system. It is produced without a signal sequence and does not follow the conventional route of protein secretion, but rather employs one or more non-conventional pathways of secretion. In an attempt to consolidate the many disparate observations on IL-1β release from the literature, and to provide a rationale for their existence, we have suggested that all mechanisms are part of one continuum of secretion, or a spectrum (Fig. 2), where the routes of secretion employed are dictated by the strength of the inflammatory stimulus and thus the levels of IL-1β required extracellularly to mount an effective inflammatory response. We propose that there are, in general, three categories of secretion mechanism. For their classification in this review we have labelled them as Rescue and redirect, protected release and terminal release. We discuss evidence to suggest that more than one of these mechanisms can be engaged within the same cell population at any one time adding credence to our proposal.

16Rescue and redirect is based on Rubartelli's observation that there is a fraction of cellular IL-1β localised to vesicles of endolysosomal nature that is targeted for degradation, but that can be redirected to the extracellular space following an appropriate secretion stimulus. The regulated secretion of lysosomes is crucial for cells of hemopoietic linage and can occur in response to a variety of stimuli. It is non-lytic and given that only a small fraction of cellular IL-1β is localised to vesicles, such a mechanism will likely be engaged when the extracellular requirement for IL-1β is low, or perhaps to supplement extracellular IL-1β provided by other mechanisms. Protected release is based upon, firstly the observation that bioactive IL-1β can be found in shed microvesicles from the plasma membrane, and subsequently that it is also found in secreted exosomes. Since IL-1β has a very short half life in plasma, it makes sense to suggest that protected IL-1β is destined for sites distant to the inflammatory lesion. In ATP stimulated THP-1 cells there is, in addition to microvesicle contained IL-1β, vesicle free IL-1β released into the supernatant. Thus, protected release occurs alongside additional mechanisms providing locally available IL-1β. Terminal release involves a commitment to cell death and occurs, we propose, under conditions of extreme inflammatory stress. Such a mechanism seems geared for the rapid release of large quantities of active IL-1β directly across a disintegrating plasma membrane. It may be possible not to exceed a threshold of plasma membrane latency before the commitment to cell death becomes irreversible.

17Due to the diversity of secretion stimuli, culture conditions, the cell types that secrete IL-1β and the difficulty of their genetic manipulation, elucidating the precise mechanisms of IL-1β secretion remains a considerable challenge. By viewing the non-conventional secretion of IL-1β as a continuum dependent upon the extracellular requirement of IL-1β, it has been possible to rationalise a context in which all mechanisms contribute to the secretion of IL-1β. Given that there is still so much biology to be discovered about IL-1β and that there is a collective curiosity within the field about how its released, the picture of IL-1β release mechanisms is likely to be dynamic with many future insights yet to come.