Clinical Profile

The patient, a 28-year-old European young man, was born from non-consanguineous healthy parents. He has experienced chronic undifferentiated psychosis since the age of 18. At 21 years old, he began to suffer from severe inflammatory crises during psychotic decompensations, characterized mainly by fever, joint and abdominal pain, though no direct correlation between the two conditions has been established.

His psychiatric illness manifests as episodes of delirium with psychomotor agitation, hallucinations and dissociative symptoms. He was first hospitalized at the age of 18 due to progressive psychiatric disturbances, initially without any noted inflammatory syndrome. The daily treatment for his psychosis was initially 2 mg of risperidone, then it was switched to 100 mg of clozapine, 15 mg of aripiprazole, 20 mg of diazepam, 500 mg of quetiapine, and 10 mg of tryptophan hydrochloride. Since age 21, he has been hospitalized four times due to psychiatric decompensations, each coinciding with severe inflammatory episodes. The inflammatory crises were characterized by arthralgias, arthritis of large joints (such as knees), adenopathies and abdominal pain without diarrhea. He had no hepatosplenomegaly. He also presented with scarlatiniform erythema, mainly on the front side of the trunk, lasting at least 5 days and erythematous edema of the left foot (Figure S1A). He exhibited marked biological signs of inflammation, including highly elevated levels of C-reactive protein (CRP) up to 350 mg/l, anemia and leukocytosis. He also had moderate hyperferritinemia (ranging from 482 to 662 ng/ml). Nonsteroidal anti-inflammatory drugs (NSAIDs) were trialed during the first hospitalization but showed limited efficacy. Corticosteroids were effective during the initial disease flares, with significant improvement in inflammatory markers: for example, a reduction in C-reactive protein (CRP) from 350 mg/L to 64 mg/L within 10 days during one flare (Fig. 1a).

Fig. 1

Patient’s biological and molecular data. (a) Patient’s CRP, Hb (hemoglobin) and ferritin blood levels are plotted. In each chart, the blue bars represent the periods of treatment with anakinra. The horizontal dashed lines represent the normal ranges for CRP, Hb and ferritin. (b) Plasma cytokine levels during and immediately following the flare and compared to five healthy donors. The results from healthy individuals were presented as median. ND: not detected. (c) Sanger chromatograms showing the NLRC4 genotype at codon 133 of the patient (right) who carries the c.398C>T, p.Thr133Ile variation in a mosaic state and of a healthy individual (left)

The patient underwent an initial brain MRI (with gadolinium enhancement and diffusion) that showed no abnormalities. Two subsequent brain MRIs and a cerebral CT scan were also performed and were normal. A lumbar puncture was performed due to acute inflammation and psychiatric symptoms. Cerebrospinal fluid analysis was within normal limits: white blood cell count was 3/mm³, red blood cell count was 310/mm³ (attributed to a traumatic tap), glucose level was 3.11 mmol/l, protein level was 0.66 g/l, and lactic acid was 1.3 mmol/l (within normal range). All microbiological cultures were sterile. The patient underwent two comprehensive immunological evaluations, at the age of 22 and 24 years. These included tests for antinuclear antibodies, antineutrophil cytoplasmic antibodies, and antibodies against extractable nuclear antigens, which were all negative. Additionally, there was no evidence of activation of the complement system. Tests for thyroid-specific autoantibodies and onconeural antibodies—including antibodies against N-methyl-D-aspartate (NMDA) receptors in the plasma—were also negative. The most recent episode, occurring at age 27, was notable as it was the first time to include right cervical cutaneous cellulitis and a pleural effusion (Figure S1B). The patient also presented with a chondritis of the left ear. Notably, this episode did not coincide with a psychotic decompensation. Following the identification of the NLRC4 mosaic variation, the treatment with anakinra (Kineret) was initiated during a flare, resulting in a rapid decrease in CRP from 228 mg/L to 40 mg/L within three days. Anakinra was administrated at a dose of 100 mg once daily for seven days, followed by 100 mg every other day for 14 days, and subsequently every third day for another 14 days. Due to the patient’s poor treatment adherence, it was agreed that anakinra would only be used on an as-needed basis to manage attacks.

Increased Cytokine Secretion in the Patient’s Plasma

To determine the cytokine profile of the patient, we quantified inflammatory cytokines in plasma samples collected at two different time points. The first plasma collection was conducted during one of the patient’s flares, while the second collection took place six month later, immediately following another flare. Both plasma samples were collected while the patient was not receiving any anti-inflammatory treatment. As expected, patient’s plasma levels of pro-inflammatory cytokines IL-1β (202.04 pg/ml), IL-18 (3332.59 pg/ml), IL-6 (147.92 pg/ml) and TNF-α (185.27 pg/ml) were elevated during flares compared with healthy individuals (IL-1β, IL-6 and TNF-α not detectable; IL-18 median, 378 pg/ml). Immediately following the flares, plasma IL-1β (50 pg/ml) and IL-6 (41 pg/ml) levels were slightly higher than in healthy individuals, while TNF-α was not detectable. In contrast, we detected markedly elevated plasma levels of IL-18 (5262 pg/ml), immediately after the flare, in comparison to healthy individuals (median, 595 pg/ml) (Fig. 1b, Table S1).

Detection of a Widespread Somatic NLRC4 Variant in Leukocyte Subpopulations

With the aim of identifying a molecular cause for the disease presented by the patient, we carried out a deep-targeted sequencing of an AID gene panel. We identified a somatic missense variant (c.398C>T, p.Thr133Ile) in the NLRC4 gene (Fig. 1c), with a VAF of 5% in peripheral blood (Figure S2A, B). The variant was confirmed by Sanger sequencing (Fig. 1c). The p.Thr133Ile variant is absent from the public GnomAD v4.1.0 database. It is located within the Nucleotide Binding Domain (NBD) of NLRC4 (Fig. 2a) and involves a residue that is invariant across species (Fig. 2b). To determine the location of Thr133 in the 3D structure of NLRC4, we relied on data from the crystal structure of the murine Nlrc4 complexed to an ADP molecule (PDB: 4kxf). Structural modeling revealed that the Thr133 residue is close to the site where the ADP molecule resides (Fig. 2c).

Fig. 2

NLRC4 mosaic variants on the domain organization model and a partial 3D structure of the protein. (a) Domain organization model of the NLRC4 protein on which are shown the NLRC4 mosaic variant p.Thr133Ile (in red) identified in this study and the previously reported variants p.Ser171Phe, p.Thr177Ala and p.His443Gln (in blue); CARD (caspase activation and recruitment domain), NBD (nucleotide-binding domain), HD1 (helical domain 1), WHD (winged-helix domain), HD2 (helical domain 2) and LRR (leucine-rich repeats domain). (b) Clustal Omega multiple sequence alignment of NLRC4 protein across various species, indicating the position of the Thr133Ile variant within a highly conserved region of NLRC4. (c) Ribbon plot of the mouse Nlrc4 protein highlighting Thr133 (in red) and Ser171, Ser177 and His443 (in blue). The adenosine diphosphate (ADP) molecule is shown in green

To study the cellular distribution of the identified NLRC4 mosaic variant, different cell types were subsequently isolated from the patient’s PBMCs (i.e., neutrophils, monocytes, B and T lymphocytes). NGS data showed a wide distribution of the variant allele in the studied cells with a VAF ranging from 3% in neutrophils to 6% in B cells. The p.Thr133Ile variant was also identified in the patient’s urine-derived DNA with a VAF of approximatively 5%, suggesting an early embryonic variant occurrence (Figure S2C). Notably, both epithelial cells and leukocytes can be present in urine; therefore, the detected variant may not be restricted to epithelial cells alone.

Impact of NLRC4 Variants on ASC Speck Formation and Activation of the NF-κB Pathway

To investigate the pathogenicity of the identified variant in NLRC4, we performed in vitro functional assays in 293T ASC-GFP_C1-FLAG cells. The previously reported NLRC4 mosaic variants p.Ser171Phe [16, 18] and p.Thr177Ala [17] were also studied. The p.Ser171Phe variant, which had been functionally characterized [18], was used as positive control in our studies. No functional characterization was previously performed for the p.Thr177Ala variant [17].

First, we assessed the impact of these three NLRC4 variants on ASC speck formation, a hallmark of inflammasome activation. 293T ASC-GFP_C1-FLAG cells were transiently transfected with an empty vector (pEV), pNLRC4_WT, pNLRC4_Thr133Ile, pNLRC4_Ser171Phe or pNLRC4_Thr177Ala. ASC specks were quantified, revealing their significant increase in cells transfected with pNLRC4_Thr133Ile as compared to cells transfected with pNLRC4_WT. A similar increase was observed with each of the other two mutated NLRC4 plasmids, pNLRC4_Ser171Phe and pNLRC4_Thr177Ala (Fig. 3a and Figure S3).

Fig. 3

ASC-GFP speck formation and NF-κB activity induced by NLRC4 variants. a. Percentage of ASC-GFP speck-positive cells upon transfection with empty vector (pEV), pNLRC4_WT, pNLRC4_Thr133Ile, pNLRC4_Ser171Phe or pNLRC4_Thr177Ala (n = 4). b. NF-κB activity in cells co-transfected with a NF-κB reporter construct, Renilla, pNLRC4_WT and each of the pNLRC4 mutants, with pNAIP when indicated (n = 4). Statistically significant P values (< 0.05) are denoted as *, ** and *** for < 0.05, < 0.01 and < 0.001, respectively. The data are presented with SEM error bars

We subsequently evaluated the ability of these three NLRC4 mutants to activate the NF-κB pathway, which regulates the expression of genes involved in immune and inflammatory responses, including various inflammasome components such as NLRC4 [21]. To this end, cells were transfected with pEV, the plasmid encoding NLRC4_WT or plasmids encoding each mutant, and then analyzed using a NF-κB-dependent luciferase reporter system. Co-transfection with pNAIP was performed to trigger the formation of the NRLC4-NAIP inflammasome in vitro. Activation of the NF-κB pathway was significantly higher with pNLRC4_Thr133Ile, pNLRC4_Ser171Phe or pNLRC4_Thr177Ala than with pNRLC4_WT, in keeping with a gain-of-function effect of each of these variants on the NF-κB pathway. In addition, co-expression of NAIP resulted in a 3- to 4-fold greater activation of the NF-κB pathway in cells expressing mutated forms of NLRC4 than in cells expressing NLRC4_WT (Fig. 3b, Table S2).

Location of the Studied NLRC4 Variants on the 3D Structure of the Protein and Impact of NLRC4 Mosaicism Levels on Age of Onset and Severity of the Disease

To test for a possible correlation between the intra-protein localization of NLRC4 somatic variants and the patients’ clinical features, we first reviewed the literature in relation with all previously reported mosaic variants (Fig. 2a and Table 1). To our knowledge, only three variants have been reported in original articles. Two of these variants, like the one identified in this study, are located in the NBD: c.512 C > T, p.Ser171Phe [16, 18] and c.529 A > G, p.Thr177Ala [17]. The third variant, c.1329 C > G, p.His443Gln, is located in the WHD [19]. To further examine the intra-protein localization of the mosaic variants, we relied on data from the crystal structure of the murine Nlrc4 protein complexed to ADP [3]. Strikingly, all the residues so far involved in the mosaic variants (i.e., Thr133, Ser171, Thr177 and His443) are located in the vicinity of the ADP molecule (Fig. 2c).

Table 1 Summary of reported cases with somatic mosaic NLRC4 variants

The p.Ser171Phe has been described in two unrelated patients: one patient with a VAF of 25%, who suffered from a severe syndrome of perinatal autoinflammation and HLH-like symptoms that led to his death at 2 months of age [16]; the second patient, with a VAF of 2–4%, had recurrent episodes of febrile myalgias starting at the age of 47 years and accompanied by arthralgias, asthenia, diffuse abdominal pain, diarrhea and mesenteric adenopathies [18]. The patient with the NLRC4 p.Thr177Ala variant and a VAF of 35% had typical features of NOMID, which include severe growth restriction, mental retardation and a facial appearance typical of NOMID [17]. Lastly, the patient carrying the NLRC4 p.His443Gln variant with a VAF of 5% presented with a late-onset AID in her early 60s (Table 1) [19]. Taken together, these data reveal that both the age of onset and the severity of the disease are correlated with the VAF of the NLRC4 mosaic variants.