An autoinflammatory neurological disease due to interleukin 6 hypersecretion

Autoinflammatory diseases are rare illnesses characterized by apparently unprovoked inflammation without high-titer auto-antibodies or antigen-specific T cells. They may cause neurological manifestations, such as meningitis and hearing loss, but they are also characterized by non-neurological manifestations. In this work we studied a 30-year-old man who had a chronic disease characterized by meningitis, progressive hearing loss, persistently raised inflammatory markers and diffuse leukoencephalopathy on brain MRI. He also suffered from chronic recurrent osteomyelitis of the mandible. The hypothesis of an autoinflammatory disease prompted us to test for the presence of mutations in interleukin-1−pathway genes and to investigate the function of this pathway in the mononuclear cells obtained from the patient. Search for mutations in genes associated with interleukin-1−pathway demonstrated a novel NLRP3 (CIAS1) mutation (p.I288M) and a previously described MEFV mutation (p.R761H), but their combination was found to be non-pathogenic. On the other hand, we uncovered a selective interleukin-6 hypersecretion within the central nervous system as the likely pathogenic mechanism. This is also supported by the response to the anti-interleukin-6−receptor monoclonal antibody tocilizumab, but not to the recombinant interleukin-1−receptor antagonist anakinra. Exome sequencing failed to identify mutations in other genes known to be involved in autoinflammatory diseases. We propose that the disease described in this patient might be a prototype of a novel category of autoinflammatory diseases characterized by prominent neurological involvement.


Background
Autoinflammatory diseases are rare monogenic or multifactorial (complex) illnesses, characterized by apparently unprovoked inflammation without high-titer autoantibodies or antigen-specific T cells [1,2]. Autoinflammatory diseases may cause neurological manifestations, such as meningitis and hearing loss [3,4], but they are also characterized by non-neurological manifestations, such as cutaneous and articular symptoms and signs [1,5]. Among the already recognized autoinflammatory disorders are autosomal dominant diseases related to NLRP3 (CIAS1) mutations, including chronic infantile neurologic cutaneous articular syndrome (CINCA) and Muckle-Wells syndrome (MWS), and the autosomal recessive (AR) familial Mediterranean fever (FMF) due to MEFV mutations. There are, however, patients with autoinflammatory phenotypes without known genetic mutations [2]. Here, we pinpoint an autoinflammatory disease due to selective IL-6 hypersecretion and characterized by predominant central nervous system (CNS) involvement.

Case presentation
In March 2006, a 24-year-old Italian male patient was admitted for chronic headache (more severe in the morning), intermittent, low-grade fever and recurrent episodes of double vision that first occurred in March 2004. He also had a single tonic-clonic seizure in July 2004, for which no anti-epileptic therapy was started. His familial and past histories were unremarkable, except for chronic recurrent osteomyelitis of the mandible manifesting with repetitive mandibular pain and swelling, since the age of 20. Neurological examination revealed bilateral papilledema and increased deep tendon reflexes. No extra-neurological signs were present. Laboratory investigations revealed raised inflammatory markers and mild anemia. Brain magnetic resonance imaging (MRI) showed diffuse T2-weighted hyperintensity in the white matter, patchy involvement of the basal ganglia and thalami, and arachnoid cysts. Brain computed tomography (CT) showed tiny subcortical calcifications (Figure 1 and Additional file 1). Lumbar puncture revealed increased opening pressure (40 cmH 2 O), raised proteins and mild pleocytosis (see Table 1 for details). Despite extensive investigations (see Additional file 1), no infection was found and no autoimmune marker was demonstrated, except for fluctuating (negative to 1:640) titers of antinuclear antibodies (ANAs). Steroid therapy, started in October 2006, caused rapid clinical improvement with disappearance of headache, fever, papilledema, double vision, and mandibular osteomyelitis. However, headache reappeared when prednisone was tapered down to 30 mg/day, despite the co-administration of azathioprine 200 mg/day, and the patient developed progressive hearing loss, postural tremor and cushingoid features. Laboratory investigations still showed raised inflammatory markers, mild anemia, elevated cerebrospinal fluid (CSF) white-cell count (with a variable number of polymorphonuclear leukocytes), and proteinuria (1 g/24 h). Renal biopsy revealed only fusion of podocyte processes.
Chronic autoinflammatory disease was hypothesized, despite the lack of extra-neurological manifestations except for mandibular osteomyelitis.

DNA Sampling and genetic analyses
Genomic DNA for gene sequencing was isolated from peripheral blood mononuclear cells (PBMCs) using standard methods, and then stored at +4°C.   and 3 0 -UTR, and their flanking intron regions were amplified by the PCR method, purified and directly sequenced. Primers and PCR conditions are available on request. The c.864C>G (p.I288M) mutation in NLRP3 and the c.2282G>A (p.R761H) mutation in MEFV were also searched for in the patient' s family members (that is, the father, mother, two sisters and one brother) by direct sequencing.

Exome analysis
To prepare a next-generation sequencing library, genomic DNA was extracted from PBMCs, and randomly fragmented by a Covaris Adaptive Focused Acoustics ™ (AFA) instrument to obtain an average fragment size of 200 to 300 base pairs. Then, the genomic DNA fragments were end-repaired, adenylated, linked to Illuminaspecific adapter sequences, and amplified by PCR. Exome capture was performed using the TruSeq ™ Exome Enrichment Kit (Illumina, San Diego, CA, USA). Sequencing was performed on the Illumina Genome Analyzer HiSeq2000 following the manufacturer's instructions. The average coverage was 20-fold. The sequencing reads were aligned to the human reference genome hg19 using the Burrows-Wheeler Aligner (BWA) software. We analyzed the data using the Genome Analysis Toolkit (GATK) variant pipeline with respect to 1,000 Genomes Data, Single Nucleotide Polymorphism Database (dbSNP) and in-house data, and we used the SnpEff software to predict the effects of sequence variants.

Body fluid collection
CSF and/or serum of the patient, patient family members and unrelated controls were used for laboratory analyses on the same day of the sampling or were stored at −40°C in the serum and CSF bank of the Neurological Institute 'Carlo Besta'. Control CSF was obtained from patients with non-inflammatory neurological diseases (that is, brain tumors, neurodegenerative and heredometabolic diseases).
Determination of cytokines and IL-6-R , whereas the level of IL-6 mRNA was investigated by real-time PCR using standard methods, and 18S as housekeeping gene. In the monocytes of the patient and ten healthy controls, membrane-bound IL-6R was roughly quantified by western blot using an antihuman IL-6 Rα antibody (Human IL-6 R alpha Affinity Purified Polyclonal Ab, Goat IgG, AF-227-NA, R&D Systems) and anti-beta-actin antibody as internal control (Monoclonal Anti-β-Actin antibody produced in mouseclone AC-15, purified immunoglobulin, buffered aqueous solution, catalogue number A1978, Sigma-Aldrich, Milan, Italy; www.sigmaaldrich.com).

Cell cultures
Monocytes were obtained from the peripheral blood of the patient, the patient's family members, and healthy controls. In order to determine the IL-1β and IL-6 secretion in vitro, monocytes were cultured in RPMI Medium 1640 (Invitrogen, Carlsbad, CA, USA; www.invitrogen. com) containing 10% fetal bovine serum (FBS, Gibco -Invitrogen, Carlsbad, CA, USA) for approximately 18 hours at 37°C/5%CO 2 , and then they were activated by 1 μg/ml of lipopolysaccaride (LPS, catalogue number L6529, Sigma-Aldrich) for 3 hours at 37°C in fresh RPMI Medium 1640. For the evaluation of the sole IL-1β secretion, the monocytes were subsequently incubated in the presence of 1 mM ATP (catalogue number A6419, Sigma-Aldrich) for an additional 15 minutes. Supernatants were collected for quantification of IL-1β and IL-6; for the evaluation of IL-6 mRNA levels, RNA was isolated from the monocytes using the QIAamp RNA Blood Mini Kit (catalogue number 52304; Qiagen, Milan, Italy; www.qiagen.com); for the evaluation of membrane-bound IL-6R, proteins were isolated using Laemmli Buffer.

Statistical analyses
Statistical analyses were performed using GraphPad Prism 5.00 for Windows, GraphPad Software, San Diego California USA, www.graphpad.com.

Ethical approval and patient consent
Anakinra was administered off-label, the experimental treatment with tocilizumab was approved by the independent ethics committee of the IRCCS Foundation, C. Besta Neurological Institute, and the patient provided written informed consent for both treatments.

Results
A novel c.864C>G (p.I288M) NLRP3 missense mutation and the previously described c.2282G>A (p.Arg761His) MEFV mutation were found. No mutation was found in the TNFRSF1A and MVK genes, which are associated with other autoinflammatory diseases, and karyotype and array-CGH analysis (see Additional file 1) were normal [1,6]. The p.I288M mutation -not reported in the INFEVERS registry (http://fmf.igh.cnrs.fr/ISSAID/ infevers/) -is located in exon 3, where most of the pathogenic NLRP3 mutations are located; the isoleucine residue at position 288 is conserved in mammals and birds, and its substitution with methionine is nonconservative. Prediction using the sorting intolerant from tolerant (SIFT) software indicates that the p.I288M substitution is of probable functional importance to the protein. Moreover, only one MEFV mutation was found in a subset of FMF patients, although FMF is traditionally considered an autosomal recessive disease [7], and both NLRP3-related diseases and FMF are IL-1β activation disorders, which may cause aseptic meningitis [3,4]. Hence, the combination of these two mutations could actually be pathogenic. However, segregation analysis in the proband's family showed that neither the p.I288M NLRP3 mutation alone nor the combination of the two mutations segregated with the phenotype (Figure 2). Moreover, i) there was no response to the IL-1 receptor antagonist anakinra (Amgen Inc), which is efficacious in IL-1β activation disorders; ii) circulating IL-1β concentration was similar in the serum of the patient (0.19 ± 0.3 pg/ml, n=12) and ten unrelated healthy controls (0.16±0.3 pg/ml); iii) no IL-β was detected in the CSF of the patient, even with a high sensitivity immunoassay; iv) no difference in IL-1β secretion was observed comparing patient and control monocytes under resting conditions and after LPS alone or LPS followed by ATP stimulation ( Figure 3A) [8]. Therefore, atypical IL-1β activation disorders were forcibly excluded. Therefore, we measured CSF2/GM-CSF, IFN-γ, IL-6, IL-8, IL-17α, and TNF-α in the patient's serum samples, and we found that, among these cytokines, only IL-6 was increased. Indeed, the IL-6 concentration evaluated in seven independent samples of the patient's serum, each obtained ≥4 weeks apart, was 37.20 ± 18.17 pg/ml (nv ≤10 pg/ml). On the basis of this finding, and given that the patient has been complaining of neurological symptoms, we also measured IL-6 in patient's CSF samples, and we found that its concentration was much higher than in serum. Indeed, the CSF IL-6 concentration evaluated in five independent samples of the patient's CSF, each obtained ≥4 weeks apart, was 5,432 ± 2,742 pg/ml (nv ≤10 pg/ml) ( Figure 3B) [9]. For comparison, the CSF IL-6 concentration evaluated in 17 patients with non-inflammatory neurological diseases (that is,thirteen with brain tumors and four with neurodegenerative diseases) was 10.4 ± 7.8 pg/ml, with a value ≤10 pg/ml in eleven of seventeen patients. Finally, we measured both IL-8 and TNF-α in at least one patient's CSF sample, and we found that TNF-α was undetectable, whereas IL-8 was increased (490.3 pg/ml, nv<32), Figure 2 Absence of co-segregation of the mutations I288M inNLRP3(CIAS1) and R761H inMEFVwith the disease phenotype in the patient's family members. probably because IL-8 is induced by IL-6 from CSF mononuclear cells [10]. Overall, these data suggest that IL-6 played a key role in the pathogenesis of the disease under investigation.
As a consequence, the patient was treated with the anti-IL-6 receptor (IL-6R) monoclonal antibody tocilizumab (Roche), 8 mg/kg every 28 days. Ten days after the first tocilizumab administration, the blood inflammatory markers normalized (Table 1). Moreover, the steroid intake was rapidly reduced without worsening of symptoms. Progressive disappearance of cushingoid features and postural tremor was observed and the patient definitely stopped complaining about headache after dividing the daily prednisone dose (12.5 mg/day) between morning (7.5 mg/day) and evening (5 mg/day). However, CSF parameters remained abnormal, and serum IL-6 concentration, which was evaluated in 23 independent samples of the patient's serum (each obtained about 4 weeks apart), markedly increased to 178.8±84.14 pg/ml (nv ≤10.0pg/ml) ( Figure 3B). This increase is probably due to the inhibition of the clearance of IL-6 from serum by tocilizumab. In fact, the main elimination pathway of IL-6 from serum may be from the binding of IL-6 to IL-6R, including the soluble form. As a consequence, IL-6 may accumulate in serum when the anti-IL-6R monoclonal antibody tocilizumab binds to the IL-6-binding site of its receptor [11]. In contrast, no significant change in the IL-6 concentration was found in four independent samples of the patient's CSF (mean value 7221±2212 pg/ml) ( Figure 3B), but we detected a reduction of CSF IL-8 concentration in two independent samples of the patient's CSF (from 490.3 pg/ml before tocilizumab to 157.0 pg/ml and, then, 53.0 pg/ml (nv<32.0) under tocilizumab). This reduction suggests that tocilizumab may diffuse into CSF, and inhibit the IL-6-mediated effects (though slightly).
The high CSF IL-6/serum IL-6 ratio suggests that IL-6 is produced in the CNS and/or CSF, from which it diffuses into blood ( Figure 3B and Table 1). To further corroborate this hypothesis, the IL-6 secretion between patient and control monocytes under resting conditions and after activation with LPS was compared, and no significant difference was observed ( Figure 3C). Moreover, no clear-cut difference was found between the levels of IL-6 mRNA after LPS stimulation (see Additional file 2).
To identify the cause of the chronic IL-6 hypersecretion, we searched for mutations in the IL6 gene, including its 5 0 -and 3 0 -UTRs, and in LIN28A and MIRLET7A3, which encode for molecules involved in the IL-6 regulatory pathway [12], with no mutation found. We ruled out mutations in the LPIN2, PSMB8 and SH3BP2 genes by using exome analysis, despite these genes having been associated with autoinflammatory diseases with different phenotypes. We investigated the serum concentration of soluble IL-6R (sIL-6R), which mediates the IL-6 clearance, but no clear-cut abnormality was found before or under tocilizumab (see Additional file 3) [11]. Moreover, we found that the amount of membrane-bound IL-6R in the patient's monocytes was similar to that of healthy controls (not shown). Finally, there was no infection by Human Herpes Virus 8 (HHV8), which contains an IL-6 homologous gene [13].

Discussion
We report a case of chronic inflammatory disease, which can be considered as the prototype of a novel category of autoinflammatory diseases characterized by primary CNS involvement. The key manifestations are chronic meningitis, progressive hearing loss, diffuse white-matter hyperintensity, and persistently raised inflammatory markers. The pathogenic mechanism of this disease is a selective and chronic IL-6 hypersecretion. Indeed, no other cytokine among those investigated was consistently increased in CSF and serum, and the anti-IL-6R monoclonal antibody tocilizumab was partially able to counteract the phenotype. The persistence of abnormal CSF parameters is likely due to the low permeability of the blood brain barrier for tocilizumab.