Abstract
Objective. Systemic lupus erythematosus (SLE) is a multisystem autoimmune disorder. Early-onset, familial, and/or syndromic SLE may reveal monogenic pathologies. The aim of this study was to examine genetic associations in patients with early-onset or familial SLE.
Methods. We enrolled 7 SLE cases (from different families) with disease onset ≤ 5 years of age and family history consistent with an autosomal recessive inheritance. Whole exome sequencing (WES) was performed in 6 index cases. Suspected variants were confirmed by Sanger sequencing. We did not perform WES in 1 patient who had features similar to the first 3 cases; only the exons of C1QA, C1QB, and C1QC were screened with Sanger sequencing.
Results. We demonstrated 2 novel and 3 previously reported variants in genes associated with SLE: a homozygous non-sense alteration (c.622C>T/p.Gln208Ter) in C1QA in 2 patients; homozygous non-sense alteration (c.79C>T/p.Gln27Ter) in C1QC in 1 (novel variant); homozygous missense alteration (c.100G>A/p.Gly34Arg) in C1QC in 1; homozygous missense alteration (c.1945G>C/p.Ala649Pro) in C1S in 1 (novel variant); and homozygous frameshift alteration (c.289_290delAC/p.Thr97Ilefs*2) in DNASE1L3 in 1 patient. Further, in 1 patient, we determined a strong candidate variant in HDAC7 (histone decetylase 7).
Conclusion. Five patients had homozygous alterations in genes coding early complement proteins. This may lead to decreased clearance of apoptotic bodies. One patient had DNASE1L3 variant, which functions in the clearance of self-antigens. In 1 patient, we determined a novel gene that may be important in SLE pathogenesis. We suggest that monogenic causes/associations should be sought in early-onset and/or familial SLE.
Systemic lupus erythematosus (SLE) is a multisystem autoimmune disorder. Different genetic factors, immunologic defects, and environmental factors play roles in SLE pathogenesis1,2. SLE is a heterogeneous disease. Thus, despite genome-wide association studies (GWAS) and case-control studies using candidate gene approaches, only around 15% of SLE heritability can be explained3.
Early-onset SLE, familial SLE, and syndromic SLE are rare presentations of the disease. Analyzing the genetic causes in these forms can help us to understand the pathogenic mechanisms in SLE4. Several monogenic causes of SLE or SLE-like phenotype have been described to date. Examples include genes important for apoptosis (e.g., FASLG), DNA damage repair (e.g., TREX1), clearance of self-antigens (e.g., DNASE1L3), and nucleic acid sensing and type I interferon (IFN) overproduction (e.g., TREX1, TMEM173)5. In addition, deficiencies in complement components may predispose to SLE through the inability to clear apoptotic cells (C1q deficiency) and negative selection of self-reactive B cells (C4 deficiency)6.
Our aim was to investigate genetic causes in cases of early-onset SLE whose family history suggested an autosomal recessive inheritance.
MATERIALS AND METHODS
Patients
We enrolled 7 pediatric SLE cases who had early disease onset (before or at 5 yrs of age) and a family history consistent with an autosomal recessive inheritance (affected siblings or parenteral consanguinity). The pedigrees of the patients are shown in Figure 1 and Figure 2. All patients fulfilled the Systemic Lupus International Collaborating Clinics (SLICC) criteria7. Demographic data, clinical manifestations, laboratory and histopathological findings, treatment, and outcome (the most recent SLICC/American College of Rheumatology Damage Index8) were documented. The subjects’ written consent was obtained according to the Declaration of Helsinki (1964) and the study was approved by the ethics committee of Hacettepe University (GO 14/368-09).
Whole exome sequencing (WES) was performed on all patients except Patient 4. Sanger sequencing of C1Q genes was performed in Patient 4 because he had similar characteristics (normal C3, C4 levels, and negative anti-dsDNA) with Patients 1–3.
DNA extraction, library preparation, and Ion Proton sequencing
Genomic DNA was extracted by standard phenol-chloroform protocol from peripheral blood of all patients and family members after the receipt of the informed consent. Ion AmpliSeq Exome RDY Kit and Ion Proton sequencing platform allowed for the capturing, amplification, and sequencing. Emulsion PCR was performed on Ion OneTouch 2 instrument using the Ion PI Hi-Q OT2 200 Kit. Enrichment of ion sphere particles (ISP) was performed using the One Touch ES module and ISP were loaded onto Ion PI chips, then sequenced according to the manufacturer’s instructions (Thermo Fisher Scientific).
Bioinformatic analysis
Exome sequencing was performed on 6 pediatric patients with SLE. Base coverage depth was 62× and 94% of the mapped reads were aligned over a target region in average. Torrent Mapping Alignment Program was used for alignment to hg19/GRCh37 human reference genome. Raw sequence file was processed using the Torrent Suite software v5.0.5 for generating mapped reads, analyzing coverage data, and variant calling. Variants with a coverage < 5× for single-nucleotide polymorphisms (SNP) and 10× for INDEL were excluded from further analysis. Phred Quality Scores were specified as Q15 for SNP variants and Q20 for INDEL. All variants that passed the quality score filters were annotated with the Ion Reporter 5.2 software. We applied variant filtration steps to select and filter the variants detected with exome sequencing.
For each sample, 48,492 variants were identified on average, including single-nucleotide variants and small INDEL. After discarding common polymorphisms (minor allele frequency > 0.01), a mean of 3542 variants remained. After that, exonic and nonsynonymous substitutions were selected, which we used for in-house filtering. The alterations that were in the boundaries of runs of homozygosity (ROH) regions ≥ 1.5 Mb within all chromosomes were extracted. We used AgileVCFMapper, a software that performs mapping of disease loci by SNP genotyping using variant caller format file obtained from exome sequence data, to identify the ROH in index cases9. As a final step, variants that were observed in ExAC Browser (in homozygous state) were ruled out. All Human Gene Mutation Database–linked mutations were collected with Ingenuity Variant Analysis software (IVA/QIAGEN Bioinformatics).
Sanger sequencing
Previously reported and novel variants identified with WES were further verified by conventional sequencing using the BigDye Terminator v.3.1 Cycle Sequencing Kit, and sequencing products were applied onto ABI 3130 genetic analyzer (Thermo Fisher Scientific). The Sanger sequencing was also performed to evaluate cosegregation within family members. All protein coding exons of C1QA, C1QB, and C1QC were sequenced in Patient 4. Primer pairs are listed in Supplementary Table 1, available from the authors on request.
RESULTS
The clinical and laboratory characteristics of the patients are summarized in Table 1 and Table 2. The characteristics of Patient 1 and 4 have been presented recently10. We identified 2 novel alterations as a homozygous non-sense variant in C1QC (Q27X/c.79C>T) in 1 patient (Patient 2) and a missense variant in C1S (A649P/c.1945G>C) in another patient (Patient 5). These 2 alterations were not present in ExAC Browser or other public databases such as dbSNP (Build 151) for default global population and Exome Variant Server (ESP6500SI-V2) for European and African American populations neither in homozygous state nor as heterozygous carrier. In addition, these changes were not present in 245 clinically unrelated Turkish individuals who constitute the in-house database established within the Hacettepe Exome Project. Open access variant functional prediction software tools were used to estimate the severity of these sequence variants in the genome. A649P alteration on C1S was predicted as “polymorphism” with MutationTaster and “damaging/probably damaging” with both PolyPhen2 and SIFT. Q27X non-sense variant on C1QC was predicted as “disease causing” with MutationTaster. Our study also demonstrated a homozygous non-sense alteration in C1QA (Q208X/c.622C>T) in 2 patients (Patients 1 and 4), a homozygous missense alteration in C1QC (G34R/c.100G>A; Patient 3) and a homozygous frameshift alteration in DNASE1L3 (T97Ifs*2/c.289_290delAC; Patient 6) in 1 case each, which are previously reported by different studies11,12,13. All indicated variants were confirmed with Sanger sequencing for index cases as well as their affected siblings. Heterozygosity for parents and genotypes of healthy siblings were also shown (Figure 1 and Figure 2).
Following WES, within 15 different chromosomes, about 126 Mb ROH regions, which are larger than 1.5 Mb, were identified for Patient 7 in total. Variant filtering revealed 6 homozygous missense variants (SNV) in different protein coding genes, which met the filtering criteria (Supplementary Table 2, available from the authors on request): a c.1472C>T in the ubiquitin specific peptidase 17-like family member 2 (USP17L2), a c.887A>C in the olfactory receptor family 5 subfamily M member 10 (OR5M10), a c.163C>T in the macrophage-expressed gene 1 (MPEG1), a c.28T>C in the Calpain 1 (CAPN1), a c.238C>T in the histone decetylase 7 (HDAC7), and a c.488T>C in the NK2 Homeobox 4 (NKX2-4). Sanger sequencing of the index case and additional family members suggested that the most probable candidate variant segregating with the affected phenotype was the HDAC7 p.R80C variant (chr12:48192588G>A; Figure 3A and 3B). We should stress that both Patient 7 and her 3.5-year-old sister were homozygous for OR5M10, MPEG1, and CAPN1 variants. Her sister did not have any signs or symptoms of SLE. However, it is too early to decide whether her sister will be affected because our index case had symptoms when she was 5 years old.
The patients with C1Q variants (Patients 1–4) have some common features such as skin involvement, negative anti-dsDNA, and normal C3 and C4 levels (Table 1 and Table 2). CH50 levels were low in all these patients. The patients with C1Q and C1S variants (Patients 1–5) improved with the addition of fresh frozen plasma therapy to conventional SLE treatment. There was no history of recurrent infections in these patients. Of note, currently, Patient 3 (with C1Q variant) experienced a flare of alopecia, malar rash, and oral aphthosis. Patient 5 (with C1S variant) had a flare of rash, oral ulcers, and neurologic involvement.
Regarding family histories, Patients 1 (C1Q variant), 2 (C1Q variant), and 5 (C1S variant) had affected siblings. The sister of Patient 1 had disease onset at 3 years of age, skin involvement (malar and discoid rash, photosensitivity), oral ulcers, nephritis, arthritis, normal C3 and C4 levels, positive antinuclear antibody (ANA), anti-SSA, and negative anti-dsDNA antibody. The brother of Patient 2 had disease onset at 4.5 years of age, malar and discoid rash, oral ulcers, arthritis, positive ANA, and hypocomplementemia. Patient 5 had an affected sister and brother. Her sister had disease onset at 5 years of age, malar and discoid rash, normal C3 and C4 levels, positive ANA, and anti-SSA antibodies. Her brother had disease onset at 5.5 years of age, oral ulcers, arthritis, neurologic involvement, normal C3 and C4 levels, positive ANA, anti-SM, and U1-RNP antibodies.
The patient with DNASE1L3 alteration (Patient 6) had a severe disease course. She received cyclophosphamide, methotrexate, and corticosteroid treatments. She is stable now on corticosteroid and hydroxychloroquine treatment with rare flares of skin and eye manifestations. It should be noted that the mother of Patient 6 had asthma and postpartum cardiomyopathy.
The patient with HDAC7 variant also had a severe disease course. She has been stable for 7 months while taking oral prednisolone, mycophenolate mofetil, and hydroxychloroquine. We performed Sanger sequencing for segregation of the indicated variant in HDAC7 gene within family members. Her parents were heterozygous while 2 siblings were carriers or negative for HDAC7 variant, which was located in a highly evolutionary conserved region of HDAC7 (Figure 3C and 3D).
None of the other family members except the ones mentioned above described features associated with autoimmunity.
DISCUSSION
In our study, Patients 1–5 had homozygous alterations in the genes coding for early complement proteins (C1Q and C1S), Patient 6 had a DNASE1L3 alteration, and we found a novel candidate variant in HDAC7 in Patient 7.
Primary complement defects have long been known to lead to an increased susceptibility to SLE14. Complement deficiencies (especially early components of the classical pathway) causing SLE have been one of the classic examples of single-gene alterations leading to the phenotype of a multifactorial disease4. Such monogenic disorders tend to present early in life or have familial causes because they are transmitted in an often-autosomal recessive fashion. Although C1q, C1s, and C1r complement deficiencies are rare, these are associated with a high risk of developing pediatric SLE, estimated at around 90% for C1q and 65% for C1r/s4.
Three different genes (C1QA, C1QB, C1QC) on the short arm of chromosome 1 encode for the C1q protein15. C1q is the first subcomponent of the classical complement pathway16,17,18. The genes for C1s and C1r are located on the short arm of chromosome 1219. C1r/s interacts with the hinge region of C1q to form the activated C1 complex. Then, C1s in this complex activates C4 and further activation of the classical complement pathway occurs. When C1 complex is not functional, the classical complement pathway cannot be activated and consumption of C4, C2, and C3 is reduced. This may result in normal to high levels of these complement proteins in serum, which was the case in Patients 1–420.
Early complement proteins are important in decreasing the number of autoantigens by their role in the clearance of apoptotic material21. C1q has also been shown to inhibit in vitro IFN-α production by plasmacytoid dendritic cells22. As a result, the loss of function in the early components of the complement pathway causes a tendency to SLE development.
To date, there are around 104 published cases with homozygous C1q deficiency23–30. The clinical presentations vary considerably; however, 2 common presentations were SLE or SLE-like disease and recurrent bacterial infections31. SLE or SLE-like disease is present in around 87% of cases, while recurrent bacterial infections occurred in 42%27,28,29,31. Among the patients with SLE or SLE-like disease, cutaneous rash was present in 70% of cases and renal involvement in around 22%27,28,29,31,32. Neurological involvement occurred in around 15% of patients, while oral ulcer was present in 18% and arthralgia/arthritis in 11%27,28,29,31. Immunologically, most C1q deficient patients had normal C3 and C4 levels, positive ANA, and extractable nuclear antigen antibodies (ENA; especially anti-Ro/SSA) and a low frequency of positive anti-dsDNA27,28,29,31. Patients 1–4, who were homozygous for C1Q variants, had SLE skin manifestations, while none had renal involvement. Two had arthritis and 1 had neurological involvement. All had normal C3 and C4 levels and negative anti-dsDNA, while 3 had positive ANA and ENA (anti-Sm and anti-Ro/SSA). The C1Q variants in Patients 1–4 were previously reported ones except 1 sequence variant.
Fewer than 10 cases with homozygous C1s deficiency have been reported so far31,33,34. Recurrent viral, bacterial, and fungal infections and mortality at a young age are common in these patients31,33,34. SLE is less frequent than patients with C1Q variants and this may be due to early mortality. In SLE cases, cutaneous and renal involvement is common. Patient 5, who was homozygous for a novel C1S variant, had both skin and renal involvement.
DNASE1L3, on the short arm of chromosome 3, encodes for 1 of the 3 human homologs of DNase1, and this enzyme functions as an endonuclease, cleaving both single-stranded DNA and dsDNA35,36,37. This enzyme probably takes a central role in SLE pathogenesis, because impaired removal of endogenous DNA from apoptotic cells or exogenous DNA derived from pathogens can trigger the formation of self-reactive DNA-specific antibodies38. DNase1-deficient mice generated by Napirei, et al presented features of SLE39. Yasutomo, et al described 2 SLE patients with heterozygote mutations in DNASE140. In 2016, Sisirak, et al reported that DNASE1L3-deficient mice rapidly developed autoantibodies to DNA and chromatin, followed by an SLE-like disease41. The p.Thr97Ilefs*2 (c.289_290delCA) mutation in our patient was previously reported in 3 patients with hypocomplementemic urticarial vasculitis syndrome, 1 of whom developed SLE at age 5.5 years11. The authors have demonstrated severely reduced levels of DNASE1L3 in homozygotes as expected because of non-sense–mediated mRNA decay triggered by the premature termination codon11. Al-Mayouf, et al also reported a deletion in DNASE1L3, which is associated with SLE in the study of 7 SLE families with a probable autosomal recessive pattern of inheritance38. They also showed that protein encoded by the mutant DNASE1L3 completely lacked DNase activity38. In fact, a GWAS peak was observed on 3p14.3 to be significantly associated with the risk of developing SLE42. This peak was attributed to PXK gene, which encodes the Phox homology domain-containing serine-threonine kinase. However, subsequent studies did not replicate the PXK association in SLE43,44. Because DNASE1L3 is only 140 kb upstream of PXK, the variants in this gene might have also generated the GWAS peak, as speculated by Al-Mayouf, et al38. When we look at the clinical characteristics of the reported SLE patients with DNASE1L3 variants (n = 18)11,38; all of them have hypocomplementemia, positive ANA, and anti-dsDNA, while most had early disease onset, nephritis, and positive antineutrophil cytoplasmic antibodies. Our patient (Patient 6) also had early disease onset, hypocomplementemia, positive ANA, anti-dsDNA, and nephritis.
HDAC are chromatin-modifying enzymes that are involved in regulation of many aspects of cell biology including tissue differentiation, autophagy, apoptosis, migration, mitosis, and angiogenesis45. There are 11 different HDAC genes encoding HDAC proteins grouped by their homology to yeast orthologues: class I (HDAC 1–3, 8), class IIa (HDAC 4, 5, 7, 9), class IIb (HDAC 6 and 10), and class V (HDAC 11)45. The role of individual HDAC proteins has not been fully identified. HDAC inhibitors have previously been considered for the treatment of SLE based on in vitro cell culture and in vivo studies in immune models of SLE; however, clinical efficacy remains to be established46. HDAC inhibition increases the production and suppressive function of regulatory T cells and decreases dendritic cell function. In addition, it is effective in the reduction of Th1- and Th17- inducing cytokines, which are important in SLE pathogenesis. In the case of identified alteration affecting the function of HDAC, the immune balance may favor autoimmunity, which may cause SLE. We will proceed with the functional study for this new candidate variant that we identified in 1 patient.
We have found alterations in early complement proteins, DNase enzyme, and a new candidate gene, HDAC7, in pediatric patients with early-onset SLE and family history suggesting autosomal recessive inheritance. WES is an effective method for identifying clinically significant exonic variants. However, there are some limitations for evolutionary conserved regulator DNA elements in untranslated, intronic, and intergenic regions that may be associated with the disease. Whole genome sequencing can also identify small copy number variations and mitochondrial DNA mutations, as well as exonic SNV missed by WES47. The definition of these single-gene disorders will shed further light on disease mechanisms.
Footnotes
Supported by Hacettepe University Research Fund with grant numbers of 014 D12 101 004-819 and TAY-2015-7335.
- Accepted for publication May 16, 2018.