Abstract
Objective. The SAPHO syndrome (synovitis, acne, pustulosis, hyperostosis, and osteitis) is a rare disorder that mainly affects bone and skin. Chronic multifocal osteitis is the main diagnostic feature. Genetic studies of HLA genes have shown no role for these class II antigens, whereas studies of 2 mouse models (cmo and Lupo) point to a role of the PSTPIP2 gene. We analyzed the PSTPIP2 gene in patients with SAPHO syndrome.
Methods. In a cohort of 38 patients with SAPHO we analyzed PSTPIP2 and 2 other candidate genes, NOD2/CARD15 (Crohn's disease occurs in about 10% of SAPHO patients), and LPIN2 (clinical similarities of SAPHO with Majeed syndrome).
Results. Rare variants of the 3 genes observed in patients with SAPHO were not specific or were not found more frequently compared to controls, suggesting no major pathogenetic role of these genes in the SAPHO syndrome.
Conclusion. We found no association between PSTPIP2, NOD2, and LPIN2 variants and the SAPHO syndrome.
The SAPHO syndrome (synovitis, acne, pustulosis, hyperostosis, and osteitis) is a rare disorder that mainly affects bone and skin, with multifocal osteitis as the main diagnostic feature. The broad spectrum of clinical features in the SAPHO syndrome, and their highly varied combinations, raise diagnostic difficulties, especially in patients with only bone lesions1–3. This results in a probable underestimation of the frequency of the disease. Several factors have been implicated in the development of the SAPHO syndrome, such as Propionibacterium acnes infection4–8 and impaired immune responses, but the etiopathogenesis is still largely unknown. Despite the pathological characteristics that the SAPHO syndrome shares with some spondyloarthropathies, such as axial skeleton involvement, skin lesions, and inflammatory bowel disease (IBD)9–11, some authors underline the absence of certain typical features of spondyloarthropathies, such as a strong association with HLA-B27, familial segregation, male predominance, and typical radiographic lesions12.
We recently reported strong humoral and cellular proinflammatory responses in a group of 29 patients with SAPHO syndrome, as shown by high interleukin 8 (IL-8) and IL-18 plasma levels and hyperproduction of the 2 cytokines by purified polymorphonuclear neutrophils (PMN) ex vivo. P. acnes-specific PMN deactivation was also observed11. We and others have found that anti-tumor necrosis factor-α therapies (infliximab and etanercept) can be beneficial in patients with refractory SAPHO syndrome11,13–16.
Genetic factors have been implicated in SAPHO syndrome, based on familial clustering17–19. However, genetic studies of class II HLA antigens revealed no role for HLA-B27, HLA-Cw6, or HLA-DR9,10,20.
The recent description of mutated genes in 2 mouse models, cmo (chronic multifocal osteomyelitis, p.Leu98Pro)21 and Lupo (macrophage infiltration, paw osteolysis, and ear necrosis, p.Ile282Asn)22, that share some manifestations with the human SAPHO syndrome has provided new insights into the molecular basis of the SAPHO syndrome. These 2 models involve nonsynonymous homozygous mutations in the proline serine threonine phosphatase interacting protein 2 gene (pstpip2). Moreover, low levels of pstpip2 have been linked to abnormal macrophage functions, suggesting that pstpip2 mutations might be related to the bone destruction and skin inflammation observed in cmo and Lupo mice22,23.
SAPHO-related human diseases point to other possible candidate genes. Majeed syndrome is an autoinflammatory bone disease that shares with SAPHO syndrome features of chronic recurrent multifocal osteitis and inflammatory dermatosis23,24. Recently, LPIN2 mutations have been implicated in this disease25,26; lipin 2 catalyzes the transformation of phosphatidic acid (PA) into diacylglycerol. As the intracellular PA concentration regulates the mTOR (mammalian target of rapamycin), cell cycle, and RANKL (receptor activator of nuclear factor-κB) pathway27, a lipin 2 abnormality might account for the altered bone proliferation observed in Majeed syndrome.
Because of the high prevalence (10%) of chronic IBD in patients with SAPHO10,28–30 we also studied the main gene involved in Crohn's disease (CD), NOD2/CARD15. NOD2 is an intracellular receptor of muramyl dipeptide (MDP), which is the minimal common component in the peptido-glycans of Gram-negative and Gram-positive bacteria; this molecule is thus thought to serve as a general sign of bacterial infection31,32. NOD2 gene mutations lead to intestinal tolerance breakdown, and this might result in increased microbial translocation and a strong inflammatory response in the gut of patients with IBD33,34. Interestingly, in vitro studies have shown that NOD2-activating agonists can induce IL-8 production by monocytes and dendritic cells35.
We investigated whether SAPHO syndrome is associated with PSTPIP2, LPIN2, or NOD2 polymorphism in 38 French patients.
MATERIALS AND METHODS
Patients and phenotyping
The study population consisted of consecutive patients attending the Rheumatology Department, Bichat teaching hospital, Paris, France. We studied 38 patients with the SAPHO syndrome (36 unrelated patients and 2 affected siblings of 2 families) who had typical osteoarticular involvement (with at least one site of condensing and/or hypertrophic osteitis) and who met the diagnostic criteria proposed by Benhamou, et al36. A standardized questionnaire was completed for each patient, which included the date of birth, gender, family history, geographic origin, age at onset, age at diagnosis, smoking habits (smoker/ex-smoker or non-smoker), the number and location of bone lesions, the number of involved joints (peripheral and/or axial), skin lesions (acne, palmoplantar pustulosis, psoriasis vulgaris), and intestinal involvement (CD and/or ulcerative colitis). Patients with associated inflammatory diseases such as IBD and pyoderma gangrenosum were excluded. The patients all had rheumatic disease, with at least one osteoarticular inflammatory site, identified by physical examination and/or magnetic resonance imaging. Specific treatments included nonsteroidal antiinflammatory drugs, prednisone, and/or methotrexate. No patient was receiving biological therapies. Patients were informed of the purpose of the study and gave their written informed consent. All the procedures were conducted in accord with our institutional ethical guidelines.
As a control group, 162 unaffected and unrelated west European Caucasians were screened for the polymorphisms found in the patients. These subjects were recruited from CEPH (Centre d'Étude du Polymorphisme Humain), which maintains a database for genetic markers that have been typed in the CEPH reference family resource for linkage mapping37,38.
Genetic analysis
Genomic DNA was isolated from peripheral blood leukocytes by using the QIAamp DNA Blood Midi kit (QIAGEN GmbH, Hilden, Germany) following manufacturer's recommendations.
Coding sequences and intron-exon junctions of PSTPIP2 (GenBank NM_024430), LPIN2 (GenBank NM_014646), and NOD2 (GenBank AF_178930) were direct sequenced.
Polymerase chain reaction (PCR) and sequencing conditions for PSTPIP2, NOD2, and LPIN2 are available on request. Primers are described in Table 1. Purified PCR amplification products were sequenced using the BigDye Terminator Cycle Sequencing Kit v.1.1 (Applied Biosystems, Foster City, CA, USA) according to manufacturer's instructions, and were resolved on an ABI 3100 automated sequencer (Applied Biosystems). Sequence data were aligned with SeqScape 2.0 software and compared to the published sequences of PSTPIP2, LPIN2, and NOD2.
Statistical analysis
Chi-square test was used for case-control association analysis. The Mann-Whitney test and ANOVA, respectively, were used to compare qualitative and quantitative variables between groups. The threshold of statistical significance was set at p < 0.05. Correlations between clinical or biological variables and presence of one or several variants were sought using the Spearman coefficient or Mann-Whitney test.
RESULTS
Clinical description of patients
We investigated 38 patients (30 female, 8 male) with SAPHO syndrome. The syndrome was sporadic in 34 cases and familial in 4 cases (2 families). Mean age at onset was 31.3 ± 2.3 years (range 12–65). Osteitis was unifocal in 17 patients (44.7%) and multifocal in 21 patients (55.26%). The mean number ± SEM of bone lesions was 1.9 ± 0.3 (range 1–8), and the mean number of involved joints [anterior chest wall (ACW), pelvis, or distal joints] was 1.9 ± 0.3 (range 1–5). Skin lesions were present in 34 (89%) patients and consisted of palmoplantar pustulosis in 19 cases (50%), severe acne in 8 cases (21%), and psoriasis vulgaris in 20 cases (52.6%). Thirteen patients (34%) had 2 concomitant skin lesions (Appendix 1). No patient had detectable rheumatoid factor, anti-citrullinated cyclic peptide antibodies, antinuclear antibodies, or anti-extractable nuclear antigen antibodies, as described11, in spite of elevated autoantibody frequency observed in other forms of inflammatory rheumatism39,40. No patient had CD.
Genetic analysis
Twenty-six variants of PSTPIP2, NOD2, and LPIN2 were found in the 38 patients (Table 2), 4 of which were new. The frequency of variants was studied in 162 controls if the relevant data were not available in the dbSNP database (http://www.ncbi.nlm.nih.gov/SNP/snpLocus.html).
1. PSTPIP2 variants. The PSTPIP2 gene has different alternative splicings (exon 1, exon 2, exon 10, exon 11), as described41 and/or in databases (Figure 1). The ΔE2 and ΔE10 isoforms are in-frame, whereas the ∆E11 isoform encodes a different COOH-terminal part, starting at amino acid 246. ∆E11 (BC035395, protein Q9H939-2) has been reported in databases but has not been confirmed at the protein level. ∆E10 lacks the tryptophan 206 that is believed to be essential for PEST-type PTP binding.
All the coding sequences of PSTPIP2 were analyzed and plotted on the full-length cDNA (NM_024430). Among the 6 sequence variations, 2 were intronic (c.248 –20 T>C, c.516+7 A>G), one was synonymous (c.663 T>C, p.Cys221Cys), one was nonsynonymous (c.964 A>G, p.Asn322Asp), and 2 were located in the 3' untranslated region (c.1005+26 G>T and c.1005+45 T>C). The c.1005+26 G>T may induce a nonconservative substitution of the last amino acid of the ∆E11 isoform (c.934 G>T, p.Gly312Cys). Five variants were linked to a common haplotype (haplotype H: c.248C-c.516G-c.964G-c.1005+ 26T-c.1005+46C). This haplotype was found to be heterozygous in 5 patients and homozygous in one patient. Its frequency in our SAPHO patient population was 9% (7/76). One patient had the c.663 T>C, p.Cys221Cys allele in the heterozygous state, and the remaining 31 patients were wt/wt. As all but one of these single-nucleotide polymorphisms (SNP) (c.663 T>C, p.Cys221Cys) were reported in dbSNP, but with no frequency information, the 162 controls were screened for these variants. We observed a similar allele frequency for all SNP (see Table 2). In particular, haplotype H showed a frequency of 13% (43/324) in controls, which was not significantly different from the frequency in SAPHO patients (chi-square test = 0.928).
2. LPIN2 variants. Among the 2 heterozygous sequence variations, 6 affected intronic regions of the LPIN2 gene (c.288+63 G>A, c.1168 T>C, c.1456+29 A>G, c.1793+27 C>G, c.1938+14 G>A, and c.2546+51 T>A) and one was a synonymous substitution (c.2223 C>T, p.Ala741Ala). All but 2 had been described (c.1793+27 C>G and c.1938+14 G>A). These 2 intronic variants were not predicted to be deleterious, based on splice site prediction software (http://www.fruitfly.org/seq_tools/splice.html and http://genome.cbs.dtu.dk/services/NetGene2; accessed 5 November 2009) and were not screened for in controls. Finally, 2 heterozygous nonsynonymous substitutions were found in SAPHO patients (p.Ala331Ser and p.Glu601Lys; Table 2). These 2 variants were not described in dbSNP and were not found by direct sequencing in any of the 162 controls. However, they were mentioned in the "infevers" database (autoinflammatory mutation online registry), in 2 patients with psoriasis42. These 2 mutations do not affect the same region as the putative nonsynonymous mutation described in Majeed syndrome (p.Ser734Leu)25,26 and were not predicted to be deleterious by the various prediction programs (Appendix 2). However, those substitutions affected residues conserved through evolution. The 2 SAPHO patients with these 2 variants also had psoriasis.
3. NOD2 variants. A total of 11 variants were found in 38 patients. Ten had been described43, and the new variant was p.Gly1032Asp. The frequencies of the NOD2 variants are reported in Table 3. Variants of the NH2-terminal part of the protein showed a frequency similar to that found in the SNP database (c.74 –25 G>T, c.534 C>G, c.802 C>T, c.1377 C>T). They are all located within the CARD and NBD domains of NOD2 (Figure 2). The SNP showed a high frequency (between 23% and 42%), and most of the SAPHO patients and controls had at least one variant in the heterozygous state. The second part of the NOD2 protein bore fewer variations (frequency below 4%), except for p.Val955Ile (10%). This latter variant has been considered neutral43. Six rare variants were observed in the SAPHO patients: one (p.Ala611Ala) was synonymous (considered a neutral variant in previous studies) and 5 were nonsynonymous (p.Arg684Trp, p.Arg702Trp, p.Ala755Val, p.Val955Ile, p.Gly1032Asp). One variant (p.Arg702Trp, SNP8) has been linked to CD (DCM, disease-causing mutation)44, and 2 others have been potentially linked to CD (p.Arg684Trp and p.Ala755Val)43. Finally, one patient had a new variant (p.Gly1032Asp), in the heterozygous state, which affected the COOH-terminal part of the NOD2/CARD15 protein. Only one of them affected a conserved amino acid (p.Ala755Val) (Appendix 2). Protein impact prediction software clearly identified p.Arg702Trp and p.Arg684Trp as probably deleterious. The rare variant p.Gly1032Asp identified here does not affect a conserved residue and was not predicted to be deleterious.
4. Rare NOD2 variants. Five SAPHO patients had rare NOD2 variants (p.Arg684Trp, p.Ala755Val, p.Gly1032Asp) and/or a DCM (p.Arg702Trp) in the heterozygous state (Table 3). Four patients had one variant, and one patient had 2 variants (p.Arg702Trp and p.Ala755Val); DNA from these subjects' parents was not available to confirm that variants affected both alleles of the NOD2 gene. The frequency of rare variants and DCM in the SAPHO population was not significantly different from reported frequencies in healthy controls (6/76 = 8% vs 21/206 = 10%)43,44. The numbers of rare variants and NOD2 DCM did not differ significantly between our SAPHO and CD patients and controls studied by Lesage, et al (chi-square = 0.96; Table 3).
Familial studies
Three genes were analyzed in 4 patients from 2 pedigrees (Family 1, P1, P2, P3, and Family 2, P4; Figure 3). Both families were Sephardic Jews from North Africa. No specificities were found in terms of bone or skin lesions.
In family 1 the 3 SAPHO patients were siblings. Their parents had no clinical signs. The 2 affected girls (P1 and P2) had SAPHO syndrome, and P1 also had psoriatic arthritis. Their brother, P3, had SAPHO syndrome and psoriasis. Another brother had ulcerative colitis but not SAPHO syndrome. Age at onset and clinical characteristics of these patients are reported in Table 4. No rare variants of NOD2, PSTPIP2, or LPIN2 were found in P1, P2, and P3.
In family 2 the patient (P4) had a sister with the SAPHO syndrome and a brother with CD. The parents were free of SAPHO signs. DNA from the affected sister was not available. Patient P4 had a typical form of SAPHO syndrome, with no clinical signs of CD. Disease was moderate, and onset occurred at age 27 years. No rare variants were detected in P4, including NOD2 DCM.
DISCUSSION
SAPHO syndrome is an uncommon form of inflammatory spondyloarthritis of unknown cause. Susceptibility genes may be involved, as rare familial clusters have been reported. Because previous studies showed no HLA gene involvement10,20, we chose to undertake an association study with 3 candidate genes: (1) PSTPIP2 gene involved in murine autoinflammatory bone disorders similar to those seen in the human SAPHO syndrome; (2) the LPIN2 gene involved in Majeed syndrome (autosomal recessive inheritance), which is also close to the SAPHO syndrome; and (3) the NOD2 gene associated with CD.
PSTPIP2 was involved in the cmo and Lupo mouse models21,22, which share some features of human SAPHO syndrome. Bone lesions in cmo mice resemble those observed in SAPHO syndrome, with both acute and chronic inflammatory processes, abnormal bone resorption, and deformities. The main difference with the SAPHO syndrome is the location of the lesions: long bones (limbs and clavicles) are affected by the inflammatory process in 75% of SAPHO patients but are spared in cmo mice10,11,24. In addition, skin lesions in cmo mice are seen only in the ears, also affecting the cartilage, while palmoplantar pustulosis and severe acne are the most common inflammatory skin disorders in the syndrome10,23. Gastrointestinal inflammation (CD or ulcerative colitis) has also been reported in patients with the SAPHO syndrome10,25–27 but not in cmo mice21. The Lupo mouse model has a phenotype primarily affecting the distal appendages, with edematous and swollen toes that, in advanced stages, can adhere together and become osteolytic with localized necrosis. Such features are never observed in SAPHO patients. Analysis of the PSTPIP2 coding sequence revealed no specific variants in the 38 SAPHO patients studied, thus confirming results obtained in 10 SAPHO patients45. A rare common haplotype, with a potential non-synonymous substitution of an alternatively spliced form (∆E11 isoform), had the same frequency in our patients as in our healthy controls. Thus, to date, no human diseases have been linked to an abnormal PSTPIP2 protein (also known as macrophage actin-associated tyrosine phosphorylated protein, MAYP).
We also studied LPIN2, as it is involved in Majeed syndrome, a condition that shares with SAPHO syndrome the presence of chronic recurrent multifocal osteomyelitis. However, dyserythropoietic anemia and neutrophilic dermatosis are not seen in SAPHO syndrome25,26. SNP of the LPIN2 gene were similarly frequent in the patients and controls42. LPIN2 haplotype analysis was not possible in our study because the parents' DNA was not available. Nevertheless, we found 2 rare variants (p.Ala331Ser and p.Glu601Lys) in the heterozygous state in 2 SAPHO patients with psoriasis, an association observed in almost 50% of SAPHO patients. These 2 variants are listed in the "infevers" database42; they affect residues conserved through evolution but were not predicted to have a deleterious effect on protein function. The "infevers" database currently lists 5 rare LPIN2 variants found in patients with psoriasis (p.Ala331Ser, P348Leu, Lys387Glu, Leu504Phe, Glu601Lys). Moreover, the LPIN2 locus had been weakly associated to psoriasis susceptibility locus, PSORS1046. This suggests that LPIN2 may account for the psoriasis component of SAPHO in these 2 patients. However, the role of LPIN2 in the inflammatory phenotype of the Majeed syndrome and in psoriasis is still unclear. Lipin 2 has a putative role in fat metabolism and mitosis and, more interestingly, it may control inflammation following oxidative stress47. This latter property could be of particular importance in SAPHO syndrome, as we have observed PMN hyperactivation in this setting11.
Finally, we also studied the NOD2 gene, as it is involved in CD (present in 10% of SAPHO patients), as well as in chronic recurrent osteomyelitis, a childhood form of the SAPHO syndrome48. However, rare variants and described DCM were not found in excess relative to the controls.
Altogether, we did not find a higher frequency of SNP in the 3 tested genes in the SAPHO syndrome versus healthy controls. Moreover, these genes were not specifically involved in familial cases of the SAPHO syndrome, as no rare variants were described in the 4 studied probands. Our findings therefore show no involvement of these 3 genes in the pathogenesis of SAPHO syndrome. However, this negative result could be due to limited statistical power, as only 38 SAPHO patients could be tested and as we focused on SNP in coding regions and exon/intron junctions.
Recent studies suggest that SAPHO syndrome could be related to PMN dysfunction. Ferguson, et al49 recently reported a SAPHO-like family (a child and his mother) with subnormal PMN intracellular production of reactive oxygen species (ROS). We failed to detect such ROS underproduction with 2 other biological approaches (extracellular ROS assay and chemoluminescence) in 29 SAPHO patients11. Intracellular ROS production thus needs to be investigated in non-familial SAPHO patients to confirm its possible role. Another recent study showed a high level of IL-1ß production by whole-blood leukocytes in 9 children with a severe SAPHO-like phenotype due to mutations in the IL1RN gene encoding the IL-1-receptor antagonist50. Finally, we reported that P. acnes stimulated IL-8 release by PMN from SAPHO patients (6-fold vs controls), further supporting a role of PMN dysregulation in this disease11. Interestingly, in our present study, we also found that PMN from patients with at least one rare PSTPIP2, NOD2, or LPIN2 variant had an intermediate capacity to release IL-8 on bacterial stimulation (3- to 4-fold increase, n = 9) compared with patients without rare variants (6-fold increase, n = 27, data not shown). Even if the 3 genes are not associated with SAPHO pathogenesis, they might influence the clinical course by modulating PMN IL-8 production on bacterial stimulation. Indeed, the 9 SAPHO patients with intermediate IL-8 production after bacterial stimulation had earlier clinical onset than patients with no rare variants (22.7 ± 2.8 yrs vs 34.8 ± 2.3 yrs, respectively), but the sample was too small for meaningful statistical analysis. This observation needs to be confirmed in a larger population in order to detect a potential link between rare PSTPIP2, NOD2, or LPIN2 variant and PMN IL-8 production, and between IL-8 production and the clinical course of patients with SAPHO.
Appendix
Footnotes
- Accepted for publication August 24, 2009.