Unanticipated Loss of Inflammasomes in Birds

Abstract

Inflammasomes are multiprotein complexes that form in response to ligands originating from pathogens as well as alterations of normal cell physiology caused by infection or tissue damage. These structures engage a robust inflammatory immune response that eradicates environmental microbes before they cause disease, and slow the growth of bona fide pathogens. Despite their undeniable utility in immunity, inflammasomes are radically reduced in birds. Perhaps most surprising is that, within all birds, NLRP3 is retained, while its signaling adapter ASC is lost, suggesting that NLRP3 signals via a novel unknown adapter. Crocodilian reptiles and turtles, which share a more recent common ancestor with birds, retain many of the lost inflammasome components, indicating that the deletion of inflammasomes occurred after birds diverged from crocodiles. Some bird lineages have even more extensive inflammasome loss, with songbirds continuing to pare down their inflammasomes until only NLRP3 and CARD8 remain. Remarkably, songbirds have lost caspase-1 but retain the downstream targets of caspase-1: IL-1β, IL-18, and the YVAD-linker encoding gasdermin A. This suggests that inflammasomes can signal through alternative proteases to activate cytokine maturation and pyroptosis in songbirds. These observations may reveal new contexts of activation that may be relevant to mammalian inflammasomes and may suggest new avenues of research to uncover the enigmatic nature of the poorly understood NLRP3 inflammasome.

Significance.

Inflammasomes are critical innate immune signaling pathways that result in the inflammatory regulated cell death pyroptosis, with the NLRP3 inflammasome remaining mysterious despite decades of research. Birds have eliminated key components of this pathway that are essential for NLRP3 signaling in mammals while preserving key output proteins, suggesting birds may hold clues to the enigmatic function of this protein in mammals.

Introduction

Regulated cell death is a strategy employed by organisms from bacteria to humans to defend against intracellular pathogens (Cookson and Brennan 2001; Jiang et al. 2020; Johnson et al. 2022). This strategy removes the host cell niche and releases soluble factors to promote the elimination of the pathogen. One form of regulated cell death is pyroptosis, which is initiated when signaling platforms called inflammasomes detect the compromise of the cytosol. These multiprotein complexes promote the dimerization of caspase-1 which ultimately cleaves inflammatory cytokines and the pore-forming protein gasdermin D (GSDMD) to enact the inflammatory pyroptotic cell death (Broz and Dixit 2016). Inflammasomes detect diverse signs of pathogen manipulation of the cytosol, including cytosolic microbe-associated molecular patterns such as, DNA, LPS, and flagellin, as well as altered cell physiology such as inappropriate protein degradation (Boyden and Dietrich 2006; Franchi et al. 2006; Miao et al. 2006; Fernandes-Alnemri et al. 2010; Rathinam et al. 2010; Hagar et al. 2013; Kayagaki et al. 2013; Sandstrom et al. 2019). The downstream signaling that results from inflammasome activation plays a vital role in the elimination of many pathogens (Maltez and Miao 2016; Nozaki et al. 2022; Li et al. 2023).

Mammals can encode all the following inflammasomes: NLRP1, 3, 6, 10, 12, NLRC4, CARD8, pyrin (encoded by MEFV) and AIM2 (Nozaki et al. 2022). These inflammasomes, with the exception of NLRC4, CARD8, and NLRP1, require the adapter protein ASC (encoded by PYCARD) to recruit caspase-1 (Fig. 1). In our exploration of caspase-1 activation in birds, we discovered that PYCARD was not present, and that only the NLRP3, CARD8, and NLRP1 inflammasomes remained. Our current understanding of NLRP3 activation in mammals requires ASC, so we sought to understand the context of this loss by exploring which inflammasomes are encoded by many species (supplementary table S1, Supplementary Material online).

Fig. 1.

Cartoon depicting inflammasome signaling in humans. Inflammasomes are comprised of an activated sensor that oligomerizes using homotypic interactions between PYD–PYD or CARD–CARD. This sensor oligomer can then recruit ASC, which further oligomerizes, or recruit caspase-1 directly. The CARD8 inflammasome is an exception to this rule, which directly recruits caspase-1, and the NLRC4 inflammasome, which is capable of directly recruiting caspase-1, though NLRC4 typically recruits ASC. In mice, NLRP1b can directly recruit caspase-1.

Inflammasomes are structures that serve to amplify signals of cytosolic compromise into a rapid and effective immune response. These structures are composed of sensors, adapters, and sometimes accessory proteins that oligomerize to form large, multiprotein complexes that can recruit and dimerize caspase-1 into its active form. Once caspase-1 is activated, it proceeds to cleave IL-1β and IL-18 to mature these cytokines into their active forms, and also cleave GSDMD to free the N-terminal pore-forming domain. The released GSDMD N-terminus is free to oligomerize and form pores in the cell membrane (Devant and Kagan 2023). GSDMD is only present in mammals. In animals that diverged before the speciation of mammals, caspase-1 cleaves the ancestral gasdermins. We recently showed that in birds, reptiles, and amphibians, all of which do not encode GSDMD, caspase-1 cleaves GSDMA (Billman et al. 2024). In more ancestral animals, like coral, caspase-3 cleaves GSDM_in to cause pyroptosis (Jiang et al. 2020). Gasdermin proteins have been identified in fungi, bacteria, and in sponges (Accession CAI7994498.1) (Daskalov et al. 2020; Johnson et al. 2022). Therefore, pyroptosis is an ancient form of regulated cell death.

The cytokines IL-1β and IL-18 are present in the cytosol of stimulated cells with an inactivating N-terminal domain that is removed by caspase-dependent maturation. These cytokines are released through the GSDMD pore, where they potentiate the immune response (Devant and Kagan 2023). All vertebrates encode IL-1β, with suggesting that immunologically differentiating between intracellular and extracellular infection is of great utility.

Results

Dynamic Changes in Inflammasomes Observed Across Species

Mammals can encode all the following inflammasomes: NLRP1, 3, 6, 10, 12, NLRC4, CARD8, PYRIN (MEFV), and AIM2. Other inflammasomes are proposed, for example, NLRP7 and NLRP9, but as there are fewer reports to support these we will not explore them here. Some of these inflammasomes are not found in all mammals, for instance, mice do not encode CARD8. To better understand which clades of life encode specific inflammasomes, we performed BLAST searches on representative species with well annotated genomes from diverse clades of animals (Fig. 2). The most conserved inflammasome was NLRP3, with identified proteins or homologs found in Cionia intestinalis (vase tunicate (sea sponge)) and Petromyzon marinus (lamprey). As a comparison, we also included a number of conserved innate immune proteins, like MyD88 and TLRs, which were found in all searched animal genomes.

Fig. 2.

Presence of inflammasome components in metazoans. The topology of the tree on the left side of the figure is derived from OneZoom, and is not depicted to a defined scale. Note that birds, though they share a common ancestor with crocodilians, turtles, and lizards, have far fewer inflammasomes. Clades of related animals are highlighted by distinct background colors. The presence or absence of genes are indicated by colored boxes, with the absence of a box indicating the gene is not present, green indicating an annotated gene, blue indicating genes annotated as “-like” and purple indicating similar genes that encode proteins with significant architectural changes such as new, or absent domains. For example, the green anole CARD8 gene uniquely encodes a protein with a C-terminal domain identified as a PYD instead of the typical CARD. Hashes indicate sequences annotated as low quality genes by NCBI. Most striking is the lack of ASC in all birds.

Caspase-1 was found in all clades searched besides the vase tunicate. Among the inflammasomes that activate caspase-1 in mammals, many are present in specific clades, whereas others are almost universally conserved.

Among the inflammasomes, NLRP3 or a related NLRP3-like protein was found in all searched species. Numerous similar genes were identified in many species, suggesting duplication of the common ancestor of this gene (Table 1). Because NLRP3 and NLRP12 are among the most closely related NLRP proteins, it may be difficult to definitively determine if the identified similar proteins arose as a result of a duplication of NLRP3 or NLRP12, or another similar gene especially in animals with a distant common ancestor with humans. CARD8- and NLRP1-like proteins, which act as decoy proteins to release an active UPA-death domain fragment that forms an inflammasome (Sandstrom et al. 2019; Wang et al. 2021), were found in all clades except vase tunicate and lamprey. Thus, guarding critical pathways by using decoy protein domains is an ancient form of immune defense. This guard strategy was first described in plants before later being discovered in mammals (Coll et al. 2011; Gaidt et al. 2021).

Table 1.

Enumeration of NLRP3-like proteins in diverse species. Identified gene copies of NLRP3-like and NLRP12-like proteins in reference genomes of selected species

Common	Species	NLRP3	NLRP12	Total
Human	Homo sapiens	1	1	2
House mouse	Mus musculus	1	1	2
Large flying fox (bat)	Pteropus vampyrus	1	1	2
Horse	Equus caballus	1	1	2
Bottlenose dolphin	Tursiops truncatus	1	1	2
Cow	Bos taurus	1	1	2
Gray short-tailed opossum	Monodelphis domestica	1	1	2
Platypus	Ornithorhynchus anatinus	2	2	4
Zebra finch	Taeniopygia guttata	1	0	1
Golden eagle	Aquila chrysaetos chrysaetos	1	0	1
Chicken	Gallus gallus	1	0	1
Chilean tinamous	Nothoprocta perdicaria	1	0	1
American alligator	Alligator mississippiensis	6	6	12
Painted turtle	Chrysemys picta bellii	15	2	17
Green anole	Anolis carolinensis	2	1	3
Common wall lizard	Podarcis muralis	2	2	4
Tropical clawed frog	Xenopus tropicalis	3	2	5
Two-lined caecilian	Rhinatrema bivittatum	6	1	7
West African lungfish	Protopterus annectens	4	1	5
Coelacanth	Latimeria chalumnae	18	2	20
Spotted gar	Lepisosteus oculatus	12	3	15
Zebrafish	Danio rerio	40	39	79
Great white shark	Carcharodon carcharias	20	1	21
Sea lamprey	Petromyzon marinus	11	4	15
Vase tunicate	Ciona intestinalis	2	0	2

NLRC4 in concert with NAIP detects bacterial T3SS components or flagellin to form an inflammasome. This inflammasome was also found in species as ancient as sharks, with an accompanying NAIP identified in all species that encoded NLRC4.

The NLRP10 and pyrin inflammasomes appear first in mammals, and are generally well preserved, however, AIM2 was frequently absent in mammals. As noticed previously, AIM2 is absent in bats (Ahn et al. 2016) and platypus (Cridland et al. 2012). We found that AIM2 was also absent in Tursiops truncatus (dolphin) and Bos taurus (cattle) supporting a previous report (Brunette et al. 2012). Previous analysis suggest that in cow, sheep, llama, dolphin, dog, and elephant AIM2 is a pseudogene (Brunette et al. 2012; Cridland et al. 2012). This pseudogene is annotated as a low quality protein, and in both cow and dolphin does not encode an N-terminal PYD. However, in both cow and dolphin, other PYHIN gene family proteins are encoded that are IFI16-like (Cridland et al. 2012; Kumar et al. 2019).

Of all of our observations, most surprisingly we found a remarkable lack of identifiable inflammasomes in birds, discussed in more detail below. Curiously, many of these proteins were present in more recent common ancestors, like crocodilians, turtles, and lizards. Some inflammasome components were identifiable in subsets of these groups, like NLRC4 found in Alligator mississippiensis (common alligator) and Anolis carolinensis (Carolina anole) but not their turtle and lizard partners, respectively. This demonstrates the dynamic background in which inflammasomes can be lost. It is important to note that our results do not exclude the possibility that these animals possess inflammasomes that have diverged significantly and thus fail to be identified by sequence homology, or have developed entirely novel inflammasome-like structures. Additionally, our method of searching for the presence of genes primarily by BLAST brings with it limitations in our ability to identify pseudogenes and depends on the completeness and quality of the annotations of the genomes searched. The presence of caspase-1 and caspase-1 like proteases in all clades searched suggests that inflammasomes are an ancient and conserved function in animals. One of the outputs of inflammasome assembly—pyroptosis—is seen in organisms as evolutionarily distant as coral, mollusks, and bacteria (Jiang et al. 2020; Johnson et al. 2022; Qin et al. 2023).

Conservation of NLRP3 Domains in Metazoans

NLRP3 in humans is composed of five regions from N-terminus to C-terminus: pyrin domain (PYD), FISNA, NACHT, winged-helix, and LRRs (Fig. 3). This inflammasome activates in response to diverse signals, including potassium efflux (Muñoz-Planillo et al. 2013), membrane integrity fluctuations, endosomal trafficking (Chen and Chen 2018), and lysosomal rupture (Katsnelson et al. 2016). A unifying upstream event that NLRP3 detects in all of these scenarios, if such an event exists, has yet to be discovered, despite extensive study of this inflammasome. Most of these domains are conserved in species that share a distant common ancestor with humans, suggesting that each domain has a critical function. The history of this protein can be gleaned from the various forms seen in diverse species found in our searches. Though the important FISNA domain was not identified by the NCBI Conserved Domain Database (CDD), significant sequence identity was found in sequences without the annotation, suggesting that, though not identified by CDD, the function of this domain may be retained.

Fig. 3.

Domains found by CDD search in selected NLRP3-like proteins. Gene length is not to scale. The presence of a colored region indicates that the NLRP3 gene from the indicated species contains the corresponding domain listed at the top of the figure, except the presence of an N-terminal CARD in sea lamprey, and N-terminal disordered region in zebra finch. Despite the lack of identifiable FISNA domains in many species, there is significant sequence similarity retained, suggesting that these proteins retain the function of this domain.

In mammals, Ornithorhynchus anatinus (platypus) encodes two NLRP3-like proteins. The first has an N-terminal PYD (“Pyrin_ASC-like”, cd08321) without an identified FISNA domain, though conserved residues were present, and an HD2 domain (“NLRC4 helical domain HD2”, pfam17776) following the WH domain (“NOD2 winged-helix domain”, pfam17779). The second has an N-terminal CARD domain (“CARD_ASC_NALP1”, cd08330) the FISNA domain (“Fish-specific NACHT associated domain”, pfam14484), and the HD2 domain. Only the first of these NLRP3 proteins are further explored here (Fig. 3). The FISNA domain allows the critical conformational changes required for human and mouse NLRP3 activation (Tapia-Abellán et al. 2021; Xiao et al. 2023). Though a FISNA domain was not identified in this isoform of NLRP3 in the platypus, the protein is 28% identical and 53% similar in the region identified as FISNA in human NLRP3 when aligned by ClustalO (supplementary fig. S1, Supplementary Material online), suggesting the domain may still confer the ability to make conformational changes despite CDD not identifying this domain in our sequences (Sievers et al. 2011).

In reptiles and amphibians, the HD2 domain as seen in the platypus was identified in all NLRP3 proteins, though the FISNA domain was not identified in Chrysemys picta bellii (painted turtle), Podarcis muralis (common wall lizard), and Xenopus tropicalis (tropical clawed frog). Additionally, in crocodiles and turtles, numerous copies of NLRP3 were identified. For example, A. mississippiensis encodes four copies in sequence on chromosome 2, and additional copies on chromosomes 8, 11, and 15.

Beginning with the common ancestor of tetrapods, each more ancient common ancestor shares this expansion of NLRP3-like proteins. Additionally, at the common ancestor with fish, an important distinction can be seen, the N-terminal domain of NLRP3 is not always identified as a PYD. Protopterus annectens (West African lungfish) NLRP3-like has no identified N-terminal domain. Latimeria chalumnae (coelacanth) have 20 NLRP3/NLRP12-like proteins, some of which have a similar structure to the human NLRP3 with an N-terminal PYD.

However, Danio rerio (zebrafish) does encode an NLRP3 protein that contains all the N-terminal PYD, FISNA, NACHT, WH, HD2, and LRRs, with an additional C-terminal stonustoxin subunit alpha domain (“SPRY_PRY_SNTX”, cd12891). This NLRP3 was determined to function similarly to human NLRP3 by recruiting ASC and caspase-1, leading to gasdermin cleavage and pyroptosis (Li et al. 2020). The majority of Carcharodon carcharias (great white shark) NLRP3-like proteins use an N-terminal CARD; however, there are two isoforms that do not have an identifiable N-terminal oligomerization domain, similar to the architecture of mammal NLRC proteins. This region may represent a new interaction interface that NLRP3-like proteins could use to interact with ASC to form an inflammasome.

Petromyzon marinus NLRP3-like proteins have an N-terminal CARD domain, with PYDs identified. Additionally, the NLRP3-like protein identified in C. intestinalis lacked an identifiable N-terminal PYD domain. This is in agreement with the lack of an identifiable ASC in these species. Thus, these NLRP3-like may be considered as NLRC-like proteins.

We only identified caspase-4/5/11 in mammals, though caspase-4/5/11 can be difficult to distinguish from caspase-1 in species that have duplicated caspase-1 like proteases (Devant et al. 2021). For example, zebrafish use caspy2 to detect cytosolic LPS, which is distinct from mammalian caspase-4/5/11 in that it encodes an N-terminal PYD (Yang et al. 2018).

Birds Have Deleted Most Inflammasome Genes

Birds were the only group that encoded fewer inflammasomes than the group they recently diverged from, the crocodilians and turtles. The most remarkable of the missing inflammasome components was the lack of ASC, paired with the maintenance of NLRP3 (Fig. 2). In mammals, NLRP3 uses its PYD to recruit the PYD of ASC, which is required to oligomerize to activate caspase-1. Most bird NLRP3 genes retain their PYD; however, so there is no apparent downstream adapter through which it could signal. We used four birds with well-sequenced genomes to represent broad categories of these animals, with paleognaths represented by Nothoprocta perdicaria (Chilean tinamou), fowl by Gallus gallus (chicken), birds of prey by Aquila chrysaetos chrysaetos (golden eagle) and songbirds by Taeniopygia guttata (zebra finch). None of these birds encoded ASC, but some variation between these species existed in other known inflammasomes.

NLRC4 is well understood to activate caspase-1 in an ASC-independent manner but was absent in most birds, despite animals with as distant common ancestors as the shark encoding homologs of this protein. Though NLRC4 was only found in the golden eagle of our representative birds species, this gene was identified in a number of extant birds, suggesting that this gene has been lost numerous times in the speciation of birds. CARD8, which has known ASC-independent activity (Ball et al. 2020; Robert Hollingsworth et al. 2021), is highly conserved in birds. NLRP1 is also found in most birds. NLRP1b from mice also activates caspase-1 in an ASC-independent manner (Broz et al. 2010; Guey et al. 2014; Van Opdenbosch et al. 2014; Ball et al. 2020). We speculate that the bird NLRP1 proteins would act analogous to their murine homologs and signal through their CARD domains directly to the CARD of caspase-1.

When searching all 451 annotated bird genomes in NCBI, only 8 ASC sequences are identified. Of these, only six are full length proteins with an N-terminal PYD, and a C-terminal CARD, one is a partial protein with these domains, and one is a partial protein with a truncated CARD (supplementary table S2, Supplementary material online). These sequences are found throughout the Aves lineage, with representatives from hawks, parrots, and petrels (Fig. 4). This suggests that the general absence of ASC in birds is not a loss by descent, but rather, ASC was lost repeatedly throughout the speciation of extant birds, and the loss of ASC granted a fitness benefit to birds in many niches.

Fig. 4.

Birds have lost ASC in multiple independent events. The presence and absence of ASC is overlaid on a phylogenetic tree of bird species. Tree topology is derived from Stiller et al. (2024). The “X” symbols indicate the absence of any ASC sequences in the derivative species. The “O” symbols indicate that the ASC protein identified contains a PYD and CARD domain (CARD-like death domain in kiwi and budgerigar). The red slash indicates a sequence of lower quality. In emu, the identified sequence encodes only a CARD domain and a short 32 AA N-terminal domain.

The Curious Case of NLRP3 in Songbirds

All birds searched encoded NLRP3 in a syntenic locus that is in a different genetic location of NLRP3 in mammals. The retention of NLRP3 is astounding because birds have lost ASC. In humans and mice, NLRP3 signals through its PYD using a homotypic pyrin–pyrin interaction with the PYD of ASC. The CARD domain of ASC then binds the CARD of caspase-1 and leads to caspase-1 activation. Therefore, it is a mystery how bird NLRP3 can signal in the absence of ASC.

Upon alignment and when comparing domains, the songbird NLRP3 stood out because the typical N-terminal pyrin domain was replaced with a novel domain. This region was also not classified by CDD as a death domain family (cd08304, which includes PYD, CARD, DED, and Death). Additionally, BLAST searches (when excluding Aves) return no hits when using the first 180 amino acids of the zebra finch NLRP3 as a query, indicating that this is a novel protein domain. By InterPro scan and IUPred3 (Erdős et al. 2021), in zebra finch, this region is disordered. We then searched all 77 sequences identified as NLRP protein in songbirds, and 6 identified as NLRP3 in falcons—the birds with the most recent common ancestor with songbirds—and found that none of these sequences had an N-terminal death domain family sequences, but did contain all the other domains of NLRP3: NACHT, HD, HD2, and LRRs. When we searched the most recent common ancestor of the next clade, that is composed of hawks, eagles, owls, and woodpeckers, we were able to identify the N-terminal PYD in every sequence (supplementary table S3, Supplementary Material online). Nevertheless, all of NLRP3 genes are syntenic, suggesting that they all arose from the ancestral NLRP3 gene. Some songbirds NLRP3 genes are annotated as NLRP12, such as T. guttata, because of changes in sequences of their N-terminal domains, we think this analysis is incorrect based on their synteny with other bird NLRP3 genes. This suggests that NLRP3 has been modified in a monophyletic group that comprises songbirds and falcons, and additionally that the PYD can be replaced by a disordered domain.

Caspase-1 is Absent in Songbirds

Surprisingly, no birds in the songbird family encode a recognizable caspase-1. Caspase-1 is the canonical inflammasome-activated protease. The entire clade, when using the 363 bird genomes from the 10KBirdGenomes project (Armstrong et al. 2020; Feng et al. 2020), was apparently missing genomic reads from the N-terminal region of NLRP3. Furthermore, the boundaries of the gap for those species that lack portions of the CASP1 gene share similar or identical 5′ and 3′ boundaries which suggests a deletion by descent versus a gap across species (Fig. 5). We must acknowledge the formal possibility that the lack of caspase-1 in songbird genomes could occur if this region is particularly difficult to sequence specifically in songbirds.

Fig. 5.

Songbirds have lost caspase-1. Alignment of 22 birds from the 10 K bird genomes project at the caspase-1 region in Gallus gallus suggest that songbirds have lost caspase-1. The tree on the left depicts the phylogeny between the selected species, with the topology derived from Stiller et al. (2024). Species names grouped with a label on the left indicate birds from a single clade. The regions of the genome that encode exonic portions of caspase-1 in Gallus gallus are highlighted with a vertical bar. Sequences from individual species are aligned using the Cactus MAF algorithm. The bars indicate sequence identity at a block of bases of the aligned bigMaf file with respect to the chicken caspase-1 sequence, with a taller bar indicating greater sequence identity. The absence of a bar indicates a gap in the sequence. This alignment demonstrates the absence of the genome that encodes exonic sequence of caspase-1 in songbirds. Exons 4, 5, and 6 correspond to the enzymatically active p20 region of caspase-1. Abbreviations of bird clades are as follows: Paleo., Paleognathae; Gallo., Galloanseres; Columb., Columbimorphae; Opisth., Opisthocomiformes; Curso., Cursorimorphae; Phaetho., Phaethoquornithes; Afro., Afroaves; Falco., Falconiformes; Psitt. Psittaciformes.

That songbirds retain the downstream targets of caspase-1—IL-1β, IL-18, and gasdermin A (GSDMA)—suggests that a core component of the inflammatory output of inflammasome activation is maintained. This observation suggests that the inflammasomes and the end products are linked by an unknown signaling pathway. It seems likely that these birds use NLRP3 to assemble into a signaling platform that activates a downstream protease.

In those birds that no longer encode caspase-1, we see two curious adaptations. First, there is a switch from in GSDMA from an FASD-like cleavage site to a YVAD-like cleavage sites in the linker region (Billman et al. 2024). Although both of these sequences are good caspase-1 cleavage targets, YVAD is classically the most specific caspase-1 target sequence that is used in synthetic inhibitors for its specificity (Talanian et al. 1997). This is curious given the switch to YVAD occurs concomitantly with the loss of caspase-1. Second, we observed a duplication a duplication of caspase-8 as caspase-18 in T. guttata, N. perdicaria, G. gallus, and A. c. chrysaetos. This duplication is preserved in all bird genomes scanned in this study (supplementary fig. S2, Supplementary Material online). In mammals, caspase-8 also is known to activate IL-1β at the same residue that is cleaved by caspase-1 in specific contexts (Maelfait et al. 2008; Bossaller et al. 2012; Gringhuis et al. 2012). Therefore, it may be that in birds a hypothetical ability of ancestral caspase-8 to cleave GSDMA, IL-1β, and IL-18 has been moved to caspase-18, and that this made caspase-1 dispensable (it should be noted that this is a highly speculative possibility). The contexts where caspase-8 can cleave caspase-1 targets proteins is currently unclear in mammalian immunology; perhaps additional analysis of bird caspase-18 might shed light on the possible relevance of this pathway.

In considering how NLRP3 signals to caspase-1 without ASC in all birds besides songbirds, we searched for novel PYD containing proteins. We searched for protein with an N-terminal PYD, (“PAAD/DAPIN/Pyrin domain”, pf02758) in birds and found a short protein that is composed of two homotypic interaction domains: an N-terminal DED/PYD, and a C-terminal death domain with similarity to RAIDD/PIDD/FADD/ankyrin/ankyrin 2 death domains (cd01670, cd08319, cd08779, cd08306, cd08317, and cd08804). This is often annotated in songbirds as “CRADD protein”; however, unlike mammalian CRADD this protein does not contain a CARD. This protein is the only protein in songbirds that encodes a PYD in the InterPro database. A protein with the same domain architecture was also identified in coelacanth, reptiles, caecilian amphibians, and marsupial mammals. However, this protein does not appear to directly connect NLRP3 to caspase-1, because this protein is also conserved in songbirds, where NLRP3 lacks a PYD. Therefore, the adapter for NLRP3 in birds remains unknown.

Discussion

Inflammasomes are among the best tools that cells have at their disposal to detect infection by intracellular pathogens. This platform efficiently leads to cytokine release and the elimination of the intracellular host cell niche. Surprisingly, these signaling platforms have changed significantly over time. We observe the accrual of new inflammasomes in mammals, which each lead to this inflammatory output. Conversely, birds have radically changed the network of available inflammasomes, which leads to speculation about which components are strictly necessary for inflammasome activation (Fig. 6).

Fig. 6.

In mammals, NLRP3 requires ASC to recruit and activate caspase-1. Nearly all birds do not encode ASC, but retain both the upstream NLRP3 and downstream caspase-1, GSDMA, IL-1β, and IL-18. In songbirds, caspase-1 is absent, but the NACHT-LRRs domain of NLRP3 is retained, along with IL-1β, IL-18, and GSDMA.

Whereas birds have lost nearly all inflammasomes, it has been noted previously in bats that AIM2 was lost (Ahn et al. 2016, 2023; Irving et al. 2021). This was proposed as a way for bats to deal with the metabolically intensive activity of flight. However, we also observe that AIM2 is absent in platypus, cow, and dolphin in the mammalian lineage. Therefore, we hypothesize that all these species do not encounter certain classes of pathogens that the AIM2 inflammasome eliminates. Alternatively, these species could have evolved novel defenses to this type of pathogen, including the duplicated IFI16-like proteins found in cow and dolphin. The inflammasome pathway proteins that are retained in birds shine a light on what may be the most indispensable proteins. Together, the NLRP3 and CARD8 inflammasomes may cover the minimal breadth of pathogen detection that cannot be further pared down with a significant fitness loss, or alternative inflammasomes.

Adequate fitness in the face of the dramatic loss of important inflammasomes reveals that these inflammasomes did not serve a role in immune defense that was sufficient to select for the maintenance of these genes. We conclude that birds must have new resistance mechanisms against pathogens that are cleared by inflammasomes in other animals. How might this resistance come to pass? In the context of a new host pathogen interface, perhaps related to flight, the pathogens that once required inflammasomes for control are now no longer encountered. Alternatively, such pathogens, could be adequately controlled by a completely different, unknown mechanism. Birds have significant alterations in their immune system, for example the bursa of Fabricius is found only in birds. What might appear as an immunological regression may in reality be a pruning down to only the those inflammasomes essential for the bird–pathogen interface.

The lack of ASC in birds could be due to the deletion of this gene being the simplest mechanism to inactivate multiple inflammasomes in one evolutionary step, thus conferring a fitness benefit to the inflammasome-deficient ancestor. In this scenario, inflammasomes could be freely lost by accruing mutations. However, the maintenance of a conserved NLRP3 protein presents a conundrum: how does NLRP3 signal without ASC? There are at least four possible explanations. First, and simplest, is the possibility that the PYD of NLRP3 in birds evolved to directly interact with the CARD of bird caspase-1, thereby functioning analogously to the signaling CARD domains of mammal NLRC4 and NLRP1.

A second possibility is that bird NLRP3 may actually be signaling similar to mammal NLRP3 signaling events that have been observed, albeit rarely. Though nearly all NLRP3 activities in human and mice require ASC, some ASC-independent roles have been reported, though these activities are sometimes disputed. The lytic cell death downstream of tyrosine kinase-mediated activation of NLRP3 is partially dependent on NLRP3 but does not require ASC (Neuwirt et al. 2023). However, in this context, IL-1β processing still required ASC. This is similar to the ASC-independent activation of NLRC4, which, without ASC, cannot mature IL-1β (Mariathasan et al. 2004). Thus, the unknown ASC-independent NLRP3 signaling pathway may also be conserved in mammals, but functioning only under specific circumstances.

A third, formal possibility is that NLRP3 is present but simply inactive in songbirds. One aspect of the genomic architecture that could support this notion is that NLRP3 may be resistant to deletion because of its genetic location with exons either within or adjacent to CEND1 in songbirds. Mice deficient in this gene develop ataxia and motor defects. Therefore, if NLRP3 deletion caused a defect in CEND1 expression, it would be highly detrimental. Nevertheless, we think the universal conservation of NLRP3 in birds suggests the protein remains functional. Songbirds do not have an identifiable death domain on the N-terminus of NLRP3. The N-terminal domain is predicted to be disordered in T. guttata. This domain may represent a functional disordered region, as seen in CARD8 (Chui et al. 2020), which is notably conserved in songbirds. This suggests active evolution of NLRP3, which supports the conclusion that NLRP3 is functional in birds.

A fourth possibility arises from the striking observation that birds also maintain CARD8 in their genomes, suggesting that NLRP3 and CARD8 together form a minimal inflammasome state. The presence of NLRP3 and CARD8 in the minimal inflammasome state is striking. In one of the original papers describing the NLRP3 inflammasome, CARD8 immunoprecipitated with the NACHT domain of NLRP3 and was dependent on the CARD8 FIIND domain, though this interaction was prevented by the presence of NLRP3 LRRs (Agostini et al. 2004) in an inactivated state. It may be that mammal CARD8 serves as a guard protein that monitors for ubiquitylation of NLRP3, and would activate via functional degradation in response to effectors bring ubiquitylating enzymes into the proximity of NLRP3. If this is the case, in the absence of ASC, as seen in birds this interaction may additionally be a persistent functional signaling mechanism. Birds may have modified the interface of the NACHT-dependent interaction of NLRP3 and CARD8, that releases an active UPA-CARD fragment of CARD8 upon activation of NLRP3. In other words, an ancestral CARD8 guard could be evolved to become a signaling adapter in birds.

The evolution of inflammasomes described here guide us toward hypotheses that would change our understanding of the NLRP3 inflammasome. Birds, because they lack ASC, are the ideal model organism for uncovering the ASC-independent functions of NLRP3 and CARD8. We hypothesize that these functions may direct us to the unifying signal of this enzyme in mammals, which has thus far eluded immunologists for decades.

Materials and Methods

Genome Searches

Initial searches were performed using iterative NCBI PSI-BLASTp, last searched December 18th, 2023. To select an isoform among multiple low E score hits, proteins were selected based on the following priority, proteins of known function (Danio rerio and Mus musculus), then proteins identified by similar domain structure using NCBI Conserved Domain Database and NCBI Conserved Domain Architecture Retrieval Tool (Geer et al. 2002; Marchler-Bauer and Bryant 2004; Marchler-Bauer et al. 2011, 2015, 2017) using the appropriate NCBI taxid.

Tree Topology

Tree topology for trees including mammals, birds, and other species was derived from OneZoom (Rosindell and Harmon 2012). Tree topology for bird only trees was derived from Stiller et al. (2024).

Software

Diagram displaying the interactions of inflammasome genes was generate using PlantUML V1.2024.0. Box plots displaying the presence and absence of inflammasome genes were generated using R version 4.3.0 and ggplot2 3.4.4.

Identification of conserved domains were performed using InterPro (Jones et al. 2014; Paysan-Lafosse et al. 2023) and the NCBI Conserved Domains Database (Geer et al. 2002; Marchler-Bauer and Bryant 2004; Marchler-Bauer et al. 2011, 2015, 2017).

Alignments were performed using ClustalO (Sievers et al. 2011).

The alignment of bird genomes is available as part of the Bird 10 K Alignment Tracks on the UCSC Genome Browser from the 10 K Bird Genomes Project (Feng et al. 2020).

Visualization of alignments was performed using Jalview (Waterhouse et al. 2009), conservation scores are derived from the AMAS method of MSA analysis (Livingstone and Barton 1993).

Supplementary Material

Supplementary material is available at Genome Biology and Evolution online.

Data Availability

All sequences used for the construction of the diagrams found in this manuscript are attached in.fasta format, and in a spreadsheet with the NCBI accession.