The Clinical Continuum of Cryopyrinopathies

Abstract

Objective

The cryopyrinopathies are a group of rare autoinflammatory disorders that are caused by mutations in CIAS1, encoding the cryopyrin protein. However, cryopyrin mutations are found only in 50% of patients with clinically diagnosed cryopyrinopathies. This study was undertaken to investigate the structural effect of disease-causing mutations on cryopyrin, in order to gain better understanding of the impact of disease-associated mutations on protein function.

Methods

We tested for CIAS1 mutations in 22 patients with neonatal-onset multisystem inflammatory disease/chronic infantile neurologic, cutaneous, articular syndrome, 12 with Muckle-Wells syndrome (MWS), 18 with familial cold-induced autoinflammatory syndrome (FCAS), and 3 probands with MWS/FCAS. In a subset of mutation-negative patients, we screened for mutations in proteins that are either homologous to cryopyrin or involved in the caspase 1/interleukin-1β signaling pathway. CIAS1 and other candidate genes were sequenced, models of cryopyrin domains were constructed using structurally homologous proteins as templates, and disease-causing mutations were mapped.

Results

Forty patients were mutation positive, and 7 novel mutations, V262A, C259W, L264F, V351L, F443L, F523C, and Y563N, were found in 9 patients. No mutations in any candidate genes were identified. Most mutations mapped to an inner surface of the hexameric ring in the cryopyrin model, consistent with the hypothesis that the mutations disrupt a closed form of cryopyrin, thus potentiating inflammasome assembly. Disease-causing mutations correlated with disease severity only for a subset of known mutations.

Conclusion

Our modeling provides insight into potential molecular mechanisms by which cryopyrin mutations can inappropriately activate an inflammatory response. A significant number of patients who are clinically diagnosed as having cryopyrinopathies do not have identifiable disease-associated mutations.

The cryopyrinopathies (also known as cryopyrin-associated periodic syndromes [CAPS]) are rare autoinflammatory diseases that have been recently characterized at the molecular level with the identification of mutations in the gene CIAS1 (cold-induced autoinflammatory syndrome 1). They include familial cold autoinflammatory syndrome (FCAS), Muckle-Wells syndrome (MWS), and neonatal-onset multisystem inflammatory disease (NOMID; also known as chronic infantile neurologic, cutaneous, articular syndrome [CINCA syndrome]). Although these disorders have been classified as distinct entities, patients often present with overlapping symptoms that include fevers, urticarial skin rash, varying degrees of arthralgia/arthritis, neutrophil-mediated inflammation, and an intense acute-phase response. In general, patients with NOMID/CINCA syndrome (MIM no. 607115) have the most severe clinical phenotype, patients with FCAS (MIM no. 120100) have the least severe phenotype, and patients with MWS (MIM no. 191100) present with an intermediate phenotype (1). Several patients have overlapping phenotypes, e.g., MWS/FCAS and MWS/NOMID.

NOMID/CINCA syndrome is characterized by nearly continuous symptoms of inflammation presenting first during the neonatal period or early infancy, with a migratory and nonpruritic urticaria-like rash and fever. Later, patients develop symptoms of central nervous system (CNS) inflammation, including chronic aseptic meningitis, increased intracranial pressure, cerebral atrophy, seizures, and sometimes mental retardation. Approximately 80% of patients develop progressive sensorineural hearing loss and ocular changes, including conjunctivitis, anterior and posterior uveitis, optic disc edema, and optic nerve atrophy with progressive vision loss. Another feature of the disease is a highly characteristic and sometimes disabling arthropathy caused by overgrowth of the patella and epiphyses of the long bones. Approximately 20% of patients with NOMID/CINCA syndrome die before reaching adulthood (2).

MWS usually presents with more episodic attacks of inflammation associated with a generalized urticaria-like rash, fever, malaise, arthralgia, and progressive hearing loss. Twenty-five percent of patients also develop serum amyloid A amyloidosis (3,4). FCAS (also known as familial cold urticaria) is characterized by cold-induced, daylong episodes of fever associated with rash, arthralgia, headaches, and less frequently conjunctivitis, but without other signs of CNS inflammation. Symptoms begin within the first 6 months of life, and generalized cold exposure is the predominant trigger of the inflammatory attacks (1).

The cryopyrinopathies are caused by dominantly inherited or de novo mutations in CIAS1, a gene that encodes a protein called cryopyrin (also known as PYPAF1 or NALP3) (3–7). Cryopyrin belongs to the family of NLR (nucleotide-binding domain, leucine-rich repeat) proteins (also known as CATERPILLER, NALP, or PYPAF), with 14 identified members (8). These proteins have 3 common structural domains: PYD (an N-terminal pyrin domain), a central NACHT domain (also known as NBD or NOD [nucleotide-binding or oligomerization domain]), and a C-terminal leucine-rich repeat (LRR) domain. The NACHT domain consists of 2 or 3 NACHT subdomains and NAD (a NACHT-associated domain). Evolutionary analysis has shown that cryopyrin is highly conserved among primates and other mammals, including the mouse. With the exceptions of V198, V262, and M662, all amino acids mutated in CAPS patients are conserved in the primate and mouse proteins, suggesting that the functions of these genes have been under strong evolutionary selection (9).

To date, >40 known disease-associated mutations have been reported, and most of them are found in exon 3 of CIAS1, encoding the NACHT subdomains and NAD domain (10). Recently, the first non–exon 3 mutations in patients with NOMID/CINCA syndrome (G755R, G755A, and Y859C) were identified (11–13). All but 1 of the known CIAS1 mutations are missense changes, suggesting that loss-of-function mutations have a different phenotype. Saito and coworkers described a NOMID/CINCA syndrome patient with somatic mosaicism for a mutation in exon 3 of CIAS1, with varying frequencies of the mutant allele in whole blood and leukocytes. Those authors proposed that somatic mosaicism may account for at least some of the undetected mutations in CIAS1 mutation–negative patients (14).

Cryopyrin is part of a multiprotein complex termed the inflammasome (15,16) that plays a critical role in the regulation of intracellular host defense in response to bacterial toxins and RNA, small antiviral compounds (imiquimod [R837] and resiquimod [R848]), and uric acid crystals. Cryopyrin is essential for activation of caspase 1 and processing of interleukin-1β (IL-1β) and IL-18 cytokines (17–20). The mechanism by which cryopyrin mutations cause inflammatory disease is still elusive, although in vitro studies suggest that the mutant protein is constitutively active (21,22). Functional studies of macrophages from NOMID and MWS patients show a constitutive increase in IL-1β and IL-18 secretion (7,16,23). As a result of this finding a new therapy targeting the IL-1β pathway, the IL-1 receptor antagonist anakinra, has been initiated in patients with these disorders (24), and has caused a rapid and dramatic improvement of symptoms and a decrease in levels of acute-phase reactants (12,25–31).

In this study we analyzed 55 NOMID/MWS/FCAS probands who have not been previously reported, and identified 7 novel mutations. Fifteen patients were mutation negative and the rest were carriers for previously reported mutations. We also summarized the results of genetic testing for CIAS1 mutations in the North American cohort of 195 patients clinically diagnosed as having NOMID/MWS/FCAS. In the absence of an experimentally determined cryopyrin structure, we used modeling to explore the nature of CAPS-associated mutations on cryopyrin.

PATIENTS AND METHODS

Patients

Patients included in this study fulfilled clinical criteria for NOMID/CINCA syndrome, MWS, or FCAS (31,32). For participation in the study all patients (or parents or legal guardians if the patient was a minor child) provided written informed consent as approved by the respective institutional review boards. Among the NOMID/CINCA syndrome patients, 10 were Caucasian, 10 were Hispanic, 1 was African American, and 1 was Asian. All FCAS and MWS patients were of Caucasian ancestry.

Molecular analysis

All coding exons of CIAS1 were screened for mutations, using primers that have been described previously (5,7). Mutation detection was performed by fluorescent sequencing with BigDye Terminator version 3.1 chemistry (Applied Biosystems, Foster City, CA) on an ABI 3100 or 3700 automated sequencer. The sequencing data were analyzed with Sequencher 4.5 (Gene Codes, Ann Arbor, MI). The allele frequencies of novel mutations were evaluated in a panel of 376 or 750 Caucasian control DNA samples from the North American Rheumatoid Arthritis Collection, using mass spec-trometry (homogeneous MassExtend assay; Sequenom, San Diego, CA) or in a panel of 100 normal control DNA samples from a North American blood bank (5), by automated sequencing. The allele frequencies for novel single-nucleotide polymorphisms (SNPs) identified in the candidate genes were evaluated in different panels of control samples (from Caucasian patients in the North American Rheumatoid Arthritis Collection, from the Coriell human diversity panel, and from African American patients in the New York Cancer Project panel).

Sequence analysis and modeling of 3-dimensional structure

Sequences of exon 3 of cryopyrin were aligned with NACHT and NAD homologies of NALP1, NALP6, NALP10, and NALP12, using Sequoia (33). The consensus secondary structure was determined by comparing the predictions for all 5 proteins using multiple programs (34–37). Models were generated by threading the cryopyrin sequence onto a known structure with Swiss-Model (38). Models for both the pyrin domain and the LRR domain were generated from known sequence homologs; however, no structures were available for obvious homologs of the NACHT and NAD domains of cryopyrin. Using threading algorithms (39,40), multiple members of the AAA+ ATPase family were top hits for the first NACHT subdomain for cryopyrin, NALP2, NALP3, CARD4, and the mouse MHCII transactivator. These models were the basis of cycles of automated and manual threading and structure minimization with Swiss-Model and the Swiss PDB Viewer. All molecular images were generated using PyMol (41).

RESULTS

Novel CIAS1 mutations

We analyzed 22 patients with NOMID/CINCA syndrome, 12 with MWS, 18 with FCAS, and 3 probands with MWS/FCAS overlap, for mutations in CIAS1. The initial screening included sequencing the 1,753-bp exon 3 of CIAS1, where most of the previously reported mutations were found. The exon 3 mutation–negative patients were screened for mutations in the other 8 exons of CIAS1.

We identified 7 novel CIAS1 mutations, occurring in 5 patients with NOMID/CINCA syndrome, 1 with MWS, and 3 with FCAS (Table 1). Seven NOMID/CINCA syndrome patients were carriers for previously described CIAS1 mutations (4 patients with the D303N mutation and 1 patient each with the T436N, G569R, and L632F mutations). Ten patients were mutation negative. Their clinical presentations were similar to those of mutation-positive patients, and they all had an excellent response to treatment with anakinra. Eight MWS probands had previously described mutations, including 2 with R260W and 6 with T348M. Three MWS patients were mutation negative. Three FCAS patients were carriers for 2 novel mutations (Table 1), while 2 lacked mutations. The remaining 13 FCAS probands were carriers for L353P (8 patients), M659K (1 patient), A439V (2 patients), and L305P (2 patients). Three patients had MWS/FCAS overlap phenotypes; 2 carried R260W and 1 had D303N.

Table 1.

Newly identified CIAS1 mutations and clinical symptoms of patients with NOMID/CINCA syndrome, MWS, and FCAS^*

Diagnosis/age (years)/sex	Cold-triggering factor	Skin rash	CNS involvement†	Joint involvement‡	Deforming arthropathy	CIAS1 mutation	Control chromosomes
FCAS/57/M	Yes	Yes	No	Yes	No	C259W (777 T>G)	0/196
NOMID/11/F	No	Yes	Yes	Yes	No	V262A (785 T>C)	0/950
NOMID/1.5/F	No	Yes	Yes	Yes	No	L264F (790 C>T)	0/924
NOMID/9/F	No	Yes	Yes	Yes	Yes	L264F (790 C>T)	0/924
NOMID/13/F	No	Yes	Yes	Yes	No	V351L (1051 G>C)	0/924
NOMID/16/M	No	Yes	Yes	Yes	No	F443L (1329 C>G)	0/930
MWS/7/M	No	Yes	No	Yes	No	F523C (1568 T>G)	0/296
FCAS/62/F	Yes	Yes	No	Yes	No	Y563N (1687 T>A)	0/188
FCAS/40/F	Yes	Yes	No	Yes	No	Y563N (1687 T>A)	0/188

^*NOMID/CINCA syndrome = neonatal-onset multisystem inflammatory disease/chronic infantile neurologic, cutaneous articular syndrome; MWS = Muckle-Wells syndrome; FCAS = familial cold-induced autoinflammatory syndrome; CNS = central nervous system.

^†Eye involvement and/or sensorineural hearing loss and/or chronic meningitis.

^‡Arthralgia and/or arthritis.

Potential reduced-penetrance CIAS1 mutations

In a family presenting with an atypical autoinflammatory syndrome we identified the R488K mutation, which was previously reported in a Spanish family with FCAS-like symptoms (42). Further investigation of the family in the present study revealed 2 asymptomatic family members who were carriers for R488K, supporting the notion that this is a reduced-penetrance CIAS1 mutation as proposed earlier (42). We screened 740 Caucasian control chromosomes and found R488K in 1 of 740; we thus estimated an R488K allele frequency of 0.0014 in the healthy Caucasian population.

The National Institutes of Health (NIH) Clinical Research Center and the Department of Pediatrics at the University of California San Diego are 2 major US referral centers for patients with various autoinflamma-tory diseases, including cryopyrinopathies, familial Mediterranean fever (MIM no. 249100), tumor necrosis factor receptor–associated periodic syndrome (MIM no. 1420680), hyperimmunoglobulinemia D with periodic fever syndrome (MIM no. 260920), and the syndrome of pyogenic arthritis with pyoderma gangrenosum and acne (MIM no. 604416). We screened 125 unrelated patients with clinically uncharacterized autoinflammatory disease for mutations in exon 3 of CIAS1. This was a cohort of patients who were referred to the NIH periodic fever clinic with a history of recurrent fevers with or without localized inflammation persisting for >6 months, and who tested negative for mutations in the clinically indicated periodic fever genes.

Ten patients, including 1 with NOMID/CINCA syndrome, were found to carry the Q703K (c.2107C>A) missense change (allele frequency 0.04). We evaluated whether Q703K may be a novel disease-associated mutation with variable expressivity. We found 1 homozygous subject and 34 carriers for the 2107C>A nucleotide transversion in a panel of 374 control Caucasian DNA samples (allele frequency 0.05).

We investigated the allele frequency for one of the earliest reported CIAS1 mutations, V198M, in a large collection of DNA samples from Caucasian subjects. This mutation was originally identified in an FCAS family that was later found to also have an E525K mutation (5), and was subsequently reported in Spanish (42) and French (4) FCAS families, in a British family with MWS and 2 British patients with uncharacterized periodic fevers (3), and in 3 German patients with atypical periodic fevers (43). V198M has been considered to be a reduced-penetrance CIAS1 mutation because it was found in asymptomatic family members and in control DNA samples. We did not identify V198M in any of 125 patients with clinically uncharacterized autoinflammatory disease. We screened a panel of 742 Caucasian DNA samples and found 11 carriers among 1,484 control chromosomes (allele frequency 0.0074).

Candidate gene screening in mutation-negative NOMID/CINCA syndrome patients

In our entire cohort of NOMID/CINCA syndrome patients, 49% were negative for CIAS1 mutations. Based on the report by Saito et al (14), we considered the possibility that some mutations might be missed due to undetected somatic mosaicism. However, we found no evidence of somatic mosaicism in our CIAS1 mutation–negative patients; therefore, we believe it is an uncommon mechanism for the cryopyrinopathies (13).

In a subset of CIAS1 mutation–negative NOMID/CINCA syndrome patients, we further screened for mutations in the NACHT domains of 5 NALP proteins that are either highly homologous to cryopyrin or have similar tissue expression profiles. We also performed mutational analysis of other genes that encode proteins that are either described as part of the inflammasome protein complex or are involved in regulation of the caspase 1/IL-1β signaling pathway (Table 2). We identified many missense substitutions in these genes, including 10 novel nonsynonymous nucleotide changes, but all of them proved to be polymorphisms (Table 2). SNPs identified in African American patients with NOMID/CINCA syndrome were evaluated in samples from African American controls.

Table 2.

Candidate genes screened for mutations in a subset of CIAS1 mutation-negative patients^*

Candidate gene	RefSeq DNA†	Exons screened	No. of patients	Nonsynonymous SNPs and deletions identified in NOMID/CINCA syndrome patients‡	Allele frequencies of novel SNPs in control chromosomes§
NALP1/DEFCAP	NM_033004.2	Exon 4 (1,705 bp)	7	T246S (rs11651595; c.737C>G)
				I601F (c. 1801A>T)¶	I601F: T allele frequency in 348 P1 samples = 0.02 T allele frequency in 94 P2 samples = 0.02 T allele frequency in 135 P3 samples = 0.15
				T782S (c.2345C>G)¶	T782S: G allele frequency in 375 P1 samples = 0.06 G allele frequency in 95 P2 samples = 0.03 G allele frequency in 146 P3 samples = 0.10
NALP2/PYPAF2	NM_017852.1	Exon 6 (1,567 bp)	7	T221M (rs17699678: c.662C>T)
				R364K (rs4306647; c.1091 G>A)
				T529A (c. 1585A>G)¶	T529A: G allele frequency in 371 samples = 0.23 G allele frequency in 89 P2 samples = 0.10 G allele frequency in 134 P3 samples = 0.16
NALP4/PYPAF4	NM_344443	Exon 3 (1,576 bp)	4	A144T (rs441827; c.430G>A) E383D (rsl7857373; c.1149G>C)
NALP6/PYPAF5	NM_38329.1	Exon 4 (1,756 bp)	7	M163L (rs6421985; c.487A>C)
				P244A (c.730C>G)¶	P244A: G allele frequency in 376 P1 samples = 0.18 G allele frequency in 85 P2 samples = 0.12 G allele frequency in 140 P3 samples = 0.05
NALP10/PYNOD	NM_76821.3	Exons 1-2	4	Nonsynonymous SNPs not identified
NALP12/PYPAF7	NM_44687.1	Exon 3 (1,702 bp)	5	D142N (c.414G>A)¶	D142N: A allele frequency in 371 P1 samples = 0.002 A allele frequency in 95 P2 samples = 0.01 A allele frequency in 143 P3 samples = 0.04
				F402L (c.ll96C>G)H	F402L: G allele frequency in 363 P1 samples = 0.07
CARD8/CARDINAL/TUCAN	NM_014959	Exons 1-8	12	Ex1: C10X (rs2043211; c.30T>A)
				Ex3: I68V (rs11881179; c.202A>G)
				Ex2: c.126_127insAA¶	c.126_127insAA all frequency in 371 P1 samples = 0.05
				Ex4: A200T (c.598G>A)¶	A200T: A allele frequency in 375 P1 samples = 0.001
				Ex4: P219R (c.656C>G)¶	P219R: G allele frequency in 352 P1 samples = 0.007 G allele frequency in 143 P3 samples = 0.11
				Ex5: P280L (c.839C>T)¶	P280L: no carriers in 304 P1 samples no carriers in 94 P2 samples T allele frequency in 141 P3 samples = 0.007
ICEBERG	NM_021571.2	Exon 1	7	Nonsynonymous SNPs not identified
IPAF/CARD12/NOD3	NM_021209.3	Exon 5	6	Nonsynonymous SNPs not identified
IL1-RN	NM_000577.3	Exons 1-5	5	Nonsynonymous SNPs not identified
PSEUDO-ICE/COP	NM_052889.2	Exons 1-3	10	Nonsynonymous SNPs not identified
PYCARD/ASC	NM_013258.3	Exons 1-3	9	Nonsynonymous SNPs not identified

^*The following single-nucleotide polymorphisms (SNPs) were identified in African American patients with neonatal-onset multisystem inflammatory disease/chronic infantile neurologic, cutaneous articular syndrome (NOMID/CINCA syndrome): I601F and T782S (NALP1), D142N (NALP12), P219R and P280L (CARD8). SNPs P244A (NALP6), F402L (NALP12), and A200T, and c.l26_127insAA (CARD8) were identified in Caucasian patients with NOMID/CINCA syndrome.

^†Uses a GenBank file as reference sequence (coding DNA).

^‡Using a dbSNP identifier as a reference; “c.” indicates a coding DNA sequence.

^§P1 = Caucasian control samples (376 samples); P2 = Human Diversity Panel control samples (96 samples); P3 = African American samples (150 samples).

^¶Novel nonsynonymous SNPs that were ruled out (proved to be polymorphisms) by screening control chromosomes.

Multiple sequence alignment of the NACHT do-main of cryopyrin with homologous human proteins

We investigated the potential molecular impact of CAPS-associated CIAS1 mutations on cryopyrin function, using a molecular modeling approach. The vast majority of mutations underlying FCAS, MWS, and NOMID are missense mutations in the NACHT and NACHT-associated (NAD) domains of cryopyrin. Although there are several mutation clusters in exon 3 of CIAS1, suggesting the presence of functionally important sites, there is no apparent correlation between mutation cluster position and disease severity. For instance, the mutation cluster adjacent to the conserved Walker B motif (Mg²⁺-binding site) at residues 300–303 includes NOMID-, MWS-, and FCAS-associated mutations (Figures 1 and 2A).

Figure 1.

Sequence alignment of cryopyrin with NALP1, NALP6, NALP10, and NALP12 for most of exon 3 of cryopyrin encoding the NACHT and NACHT-associated domains. Asterisks below the alignments show positions with perfect identity. AAA+ domain functional motifs that are identifiable in cryopyrin are shown in red. Disease-causing mutations are shown above the sequence; boxed mutations are those newly identified in this study. The consensus secondary structure prediction for all proteins from multiple prediction programs is depicted using bars for α-helices (α1–α22) and arrows for β-strands (β1–β6). Secondary structural elements shown in dark green correspond to the first subdomain of NACHT, those shown in light green correspond to the second (and possibly third) subdomain(s) of NACHT, and those shown in yellow-green correspond to the NACHT-associated domain.

Figure 2.

A, The domain and exon arrangements of cryopyrin displayed with the positions of point mutations implicated in familial cold-induced autoinflammatory syndrome (FCAS), Muckle-Wells syndrome (MWS), neonatal-onset multisystem inflammatory disease (NOMID), and other phenotypes. Domains are displayed as colored boxes; exons are numbered below and delineated by horizontal lines. Point mutations responsible for various clinical syndromes are displayed from the weakest phenotype (FCAS) on the light yellow portion of the gradient to the strongest phenotype (NOMID) on the red portion of the gradient. B, Three-dimensional models of the pyrin domain, subdomain 1 of NACHT, and the leucine-rich repeats (LRRs). Disease-causing mutations are displayed as spheres at the Cα positions, color-coded as yellow for FCAS, orange for MWS, red for NOMID, and gray for other. For amino acid positions identified in multiple syndromes, the most severe phenotype is indicated. The pyrin domain of cryopyrin was threaded onto the pyrin domain of human ASC (Protein Data Bank [PDB] code no. 1ucp [52]). The first domain of NACHT was generated using the structure of apoptotic protease–activating factor 1 (APAF1; PDB code no. 1z6t [53]). APAF1 and cryopyrin deviate between α5 and β4 and were modeled using consensus secondary structure predictions. This region includes the positions of the G326E and S331R mutations (red spheres that do not map to the common mutation surface). The LRR domains were modeled using the porcine ribonuclease inhibitor structure (PDB code no. 2bnh [54]). C, Superimposition of the model for NACHT subdomain 1 on the hexameric assembly of the AAA+ ATPase HslU (PDB code no. 1doo [48]). Most of the mutations (colored as in B) map to the inner, concave surface of the hexameric ring. In other AAA+ ATPases, this surface is where AAA+ subdomains 2 and 3 tend to contact subdomain 1 (46). As in other AAA+-ATPases, the hexameric assembly places nucleotide-binding sites at the interfaces between adjacent protomers. D, Location of the disease-causing mutations in cryopyrin, suggesting a model for inappropriate inflammasome assembly. In wild-type cryopyrin, opening of the protein to expose the pyrin and LRR domains is normally triggered due to an external stimulus. The open state of cryopyrin is presumably the active one that assembles a functional inflammasome. With mutations in the NACHT domain, the closed state is destabilized due to mutations at the interface of the hinge, tending to favor the activated and open state. The G755 mutations might similarly drive inflammasome assembly by producing a “kink” in the regular LRR structure that inappropriately exposes the LRR domains either to more readily open the structure or to become competent to directly assemble the inflammasome.

Phylogenetic studies have identified human NALP1, NALP6, NALP10, and NALP12 proteins as being closely related to cryopyrin (9). Recent studies have shown that NALP1, NALP6, and NALP12 are expressed in hematopoietic tissue; similar to findings in cryopyrin, overexpression of these proteins shows that they also regulate caspase 1 activity (8). Thus, we aligned the protein sequence of the cryopyrin NACHT domain with its most homologous human proteins, NALP1, NALP6, NALP10, and NALP12, to evaluate whether the CAPS-associated mutations affect amino acids that are conserved between these proteins (Figure 1). The sequence alignment showed that only 5 CAPS-associated mutations affected highly conserved amino acid residues (C259W, G301D, D303N, A352V, and L632F); 3 mutations were associated with the most severe NOMID phenotype, while 2 were associated with the mild FCAS phenotype.

Molecular impact of CIAS1 mutations on protein function, determined based on predicted modeling for cryopyrin

CIAS1 mutations (Figure 2A) were placed onto 3-dimensional structural models of human cryopyrin threaded onto templates of homologous structures (Figure 2B). The modeling of the N-terminal part of the NACHT homology (NACHT subdomain 1) was possible because of the robust identification of this portion of the NACHT homology as a member of the AAA+ family of proteins, and is consistent with previous models generated for this portion of the NACHT domain (44,45). These models demonstrate that the majority of mutations that can be placed in the model are located on one face of subdomain 1; for most AAA+ proteins, this face is where additional domains from the same protein can interact with subdomain 1 (46). This intraprotein domain–domain interaction could result in a closed or nonfunctional state of a modular protein such as cryopyrin.

More C-terminal portions of the NACHT and NACHT-associated domains do not robustly identify any domain by sequence homology or threading; however, the second NACHT of AAA+ proteins is typically 4-helix bundles, and the secondary structural predictions would be consistent with this (Figure 1), although the possibility of inserted secondary structural elements in the NACHT family cannot be ruled out. Notably, the pattern of disease-causing mutations affecting amino acids at conserved surfaces might continue outside of the modeled regions. For example, the mutations causing α-helix 12 amino acid substitutions (T436P, T436N, T436I, A439V, A439T, A439P, F443L [Figure 1]) are all spaced by 3 or 4 residues and would define one face of this α-helix.

Most AAA+ proteins assemble into functional hexamers (47). To gain additional insight into the disease-causing amino acid substitutions, we superimposed the NACHT subdomain 1 model onto the structure of the HsIU chaperone, which is one of the AAA+ proteins whose structure was experimentally determined in a hexamer assembly (48). Surprisingly, the surface containing most of the disease-causing amino acid substitutions in the monomer model mapped to a single inner surface of the assembled hexameric surface, which would be consistent with the notion of functional multimerization of cryopyrin through the NACHT domain (Figure 2C).

Close analysis of the NOMID mutations affecting the LRR domains, G755R and G755A, indicated that these mutations may function similarly to NACHT domain mutations, by inappropriately exposing LRRs to assemble inflammasomes (Figure 2B). The internal packing between leucine-rich repeats of the LRR domain forces every second helix to accommodate a side chain from an adjacent helix in a pocket lined by a glycine. Any other residue at the glycine position would be larger and would disrupt the LRR packing, potentially kinking the arc of the LRR domain or generating a flexible joint in the LRR at the repeat containing the substitution. G755 is the first glycine in this position in the LRR domain, and amino acid substitutions may decouple the orientation of the LRR domain from the closed/open status of the NACHT domains (Figure 2D). The fact that mutations of equivalent glycines in later repeats have not been observed may suggest that they may be too disruptive to the LRR binding surface or may not displace enough of the surface to direct inflammasome assembly.

DISCUSSION

The cryopyrinopathies represent a spectrum of autoinflammatory diseases that are caused by mutations in a single gene, and in which severity of the phenotype is most likely influenced by a number of other genetic and environmental modifiers. We report 7 novel NOMID/MWS/FCAS-causing mutations in exon 3 of CIAS1. All of these mutations cause missense nucleotide substitutions and have not been found in any of several hundred control chromosomes. Thus, with this report the total number of CIAS1 disease-associated mutations is 56 (10).

Our entire NOMID/CINCA syndrome cohort includes 39 patients who fulfill clinical criteria for diagnosis. In the past, we reported 8 mutation-positive and 9 mutation-negative patients (7,49,50); herein we report an additional 12 mutation-positive and 10 mutation-negative cases. The mutation-negative patients were screened for mutations in all exons of CIAS1, thus ruling out mutations in the coding part of cryopyrin as a cause of the disease. However, our mutation screening method, although extensive, did not rule out the presence of mutations in the noncoding regulatory sequences of CIAS1. Thus, CIAS1 mutations were identified in only 51% of cases (20 of 39), suggesting significant genetic heterogeneity among patients with NOMID/CINCA syndrome.

Similarly, 25% (3 of 12) of the probands in our MWS cohort lacked mutations in the coding sequence of CIAS1 exons. In the remaining patients, we found 3 known CIAS1 mutations (R260W, A352V, and T348M). Interestingly, the prevalence of MWS appears to be lower in North America than that reported in Europe. This may result from phenotype assignment preference in the disease continuum or the presence of larger families with MWS in Europe and larger families with FCAS in North America, resulting in a selection bias.

Over the past few years we have identified 137 FCAS patients who belong to 24 families. The vast majority of them (114 patients from 12 families) are carriers for the L353P mutation. At least 3 of these families share a common ancestor, and 1 family has been confirmed by genotyping to be independent (51). Of the remaining patients, 21 are carriers for the following CIAS1 mutations: C259W, L305P, A439V, E525K, Y563N, E627G, and M659K. Two FCAS patients are mutation negative.

In this study, we evaluated allele frequencies for 3 CIAS1 mutations that may be considered polymorphisms, V198M, R488K, and Q703K, in a large panel of DNA samples from Caucasian controls. For V198M and R488K, we found allele frequencies of 0.0074 and 0.0014, respectively, among 742 and 370 Caucasians. Previous studies identified V198M in control DNA samples from 0 of 109 North American Caucasians, 1 of 130 UK Caucasians, 0 of 150 Spanish Caucasians, and 2 of 48 Asian Indians (3,5,42), and thus have raised the question of whether V198M is a true disease-associated mutation. R488K was not identified in any of 150 Spanish control DNA samples; however, it was found in 1 unaffected member of a Spanish FCAS family (42). In the current study, which is the largest survey of healthy Caucasian controls to date, neither variant was found at the 1% allele frequency required to be considered formally as a polymorphism. Nevertheless, given that the V198M allele frequency approaches this level and that we have recently observed a patient with V198M and periodic fever that did not respond to anakinra (Kastner DL, et al: unpublished observations), we would approach the clinical significance of this variant with caution. Considering that these substitutions have been associated with nonspecific autoinflammatory pheno-types, in vitro functional assays may be useful in understanding their pathogenic consequences. Based on the similar allele frequency of Q703K in patients and control cohorts (0.04 versus 0.05; P = 0.84), we conclude that Q703K is unlikely to be pathogenic.

The vast majority of mutations underlying FCAS, MWS, and NOMID are dominant mutations. This is consistent with the notion that cryopyrinopathies are caused by gain-of-function mutations (21) and is quite different from the spectrum of frameshift and nonsense mutations observed in diseases driven by loss of function. We thus sought to understand the structural impact of these activating mutations. Our modeling of the first NACHT subdomain yielded essentially similar results using 2 different templates (Cdc6p from Pyrobaculum aerophilum [Protein Data Bank code no. 1fnn] and apoptotic protease–activating factor 1 [Protein Data Bank code no. 1z6t]). The findings were also consistent with results of previous studies (44,45) in that most of the mutations in this subdomain, including C259W, V262A, L264F, and V351L identified in this study, fall on one face of the monomer (Figure 2B). Mapping these mutants onto a model of a cryopyrin hexamer revealed the surprising result that a single, solvent-exposed surface is affected (Figure 2C). This suggests that the primary impact of these mutations is to disrupt a surface and differs from previous proposals that mutations function via defects in nucleotide hydrolysis, conformational changes, or multimerization.

Although the possibility that the mutated surface in the hexamer interacts with the LRRs cannot be ruled out (8), comparisons with other AAA+ proteins suggest that other NACHT subdomains may interact at this surface. Interaction between NACHT subdomains is consistent with patterns of mutations in other NACHT domains that suggest that another surface is disrupted (Figure 1). Similarly, the effect of the G755A/R mutations in the LRRs may be quite similar to those in NACHT subdomains, even if the mutations affect cryopyrin in a structurally different way (Figure 2D).

These models are consistent with the notion that known mutations share a common mechanism whereby a closed and inactive conformation of cryopyrin is disrupted, leading to activation of the inflammasome complex and subsequent activation of caspase 1 and increased IL-1β and IL-18 secretion (Figure 2D). The fact that most of the disease-causing mutations occur in the NACHT and NACHT-associated domains suggests that the gain of function requires intact pyrin and LRR domains, or that most mutations in the pyrin and LRR domains cannot activate cryopyrin. These disease-causing mutations do not need to completely disrupt the inactive state of cryopyrin, but rather could simply influence the normal equilibrium between active and inactive forms of the molecule and the ease by which inflammasome assembly is activated by external stimulus.

Our models, however, cannot directly address the clinically important problem of the severity of any particular mutation. And the question remains as to whether the CAPS mutations show a clear correlation between genotype and phenotype. If so, then their defects can be understood primarily in the context of cryopyrin structure–function relationships. If not, then other genetic factors likely play a role in the disease process. Addressing the genotype/phenotype correlation is complicated by the small number of patients, reporting bias, and variable phenotyping by clinicians. Despite these limitations, the phenotypes of many patients with known mutations have been adequately described in the literature and can be used as a basis for this analysis.

Analysis of all known mutations indicates that mutational severity is not correlated with mutation cluster position (Figure 1), the specific residue mutated (for example, R260 in Table 3), or conservation among CIAS1 orthologs (Figure 1). Thus, we focused solely on multiply observed mutations (Table 3). At both ends of the disease severity continuum, there are mutations with relatively consistent phenotypes. For example, L353P and L305P tend to be associated with milder pheno-types, whereas F309S and Y570C tend to be associated with more severe phenotypes. Other mutations, including those affecting A439, R260, and D303, have significantly variable severities, strongly suggesting that other genetic factors can play a role in disease severity.

Table 3.

Genotype/phenotype of reported CIAS1 mutations^*

aa	Mutation	Other		FCAS		MWS/FCAS		MWS		MWS/NOMID		NOMID		Refs.
		Families	Patient	Families	Patient	Families	Patient	Families	Patient	Families	Patient	Families	Patient
R488	K	1†	2	1	1									42, P
V198	M	4	5‡	2†§	3			1¶	3					3, 5, 42, 43
L305	P			4	19									3, 42, P
L353	P			12#	114									51, P
M659	K			2	4									10, P
A439	V			3**	7			-	3	-	2			5, 10, P
	T							l	4					4
	P									1	1			10
R260	W			1	12	4	13	6	34					4, 10, 16, 27, P
	L									2	2			45
	P											1	1	45
D303	N					1	2	1	1	1	1	9	10	4, 6, 7, 10, 23, 42, 45, P
	G											1	1	45
T348	M							10	18	5	5			4, 10, 28, 42, 45, P
G569	R							1	2			1	1	4, 25
F523	L									1	1	1	1	7
	C							1	1					P
G326	E									1	1	1	1	10
L632	F									1	1	1	1	10
L264	F									1	1	1	1	P
	H											1	1	7
T436	I									1	1			10
	N											1	2	6
	P											1	1	10
F309	S											2	2	6, 10
G755	R											1	1	12
	A											1	1	13
Y570	C											6	6	7, 10, 14
	F											1	1	10

^*Amino acid (aa) residues for which mutations are only reported once are excluded. Family data are the total number of reported families with ≥1 affected patient; patient data are the total number of reported patients. P = present study (see Table 1 for other definitions).

^†Unaffected family members with mutation.

^‡Uncharacterized periodic fevers.

^§One family also carried E525K.

^¶Patients in family with features of FCAS, MWS, and NOMID.

^#Three families confirmed (others presumed) to have common ancestor; 1 family confirmed independent.

^**One family with intrafamilial variable phenotype.

In conclusion, we have identified a number of novel disease-causing mutations in CIAS1, and our structural modeling of cryopyrin suggests a common disease mechanism. But most importantly, several lines of evidence presented here, including the finding of patients without cryopyrin mutations, the lack of a clear genotype/phenotype correlation for many mutations, and the potential reduced penetrance of V198M and R488K, strongly suggest the presence of additional genetic factors that initiate or modulate the cryopyrinopathies. Identification of these genes will be important for understanding innate immunity as well as providing new targets for controlling these diseases.