Functional Analysis of RELN S2486G Mutation and its Contribution to Pathogenesis of Ankylosing Spondylitis

1Department of Medical Genetics, Faculty of Medical Sciences, Tarbiat Modares University, Tehran, Iran 2Australian Regenerative Medicine Institute, Monash University, Clayton, VIC 3800, Australia 3Department of Molecular Genetics, Faculty of Biological Sciences, Tarbiat Modares University, Tehran, Iran 4Rheumatology Research Center, Tehran University of Medical Sciences, Tehran, Iran 5Inflammation Research Center, Tehran University of Medical Sciences, Tehran, Iran


Introduction
Ankylosing spondylitis (AS; OMIM:106300) is a chronic inflammatory disease affecting the skeletal systems, mainly the spine and sacroiliac joints. 1 The chronic inflammation in bone and cartilage tissues causes irreversible damages and remodels the joints' architecture. 2 The important manifestations of AS are excessive bone formation localized in sacroiliac joints and expanded chronic inflammation. 3 Additionally, the precise etiopathogenesis of the disease has not been completely uncovered. 4 In a previous study, we reported a novel heterozygous mutation in the RELN (OMIM: 600514) gene, c.7456 A>G; p.Ser2486Gly, which was associated with AS revealed by a co-segregation study in a large Iranian family. 5 The encoded protein, reelin protein, is a large extracellular matrix protein that plays a crucial role in neuronal migration during brain development. 6,7 Different genetic polymorphisms/variations in the RELN gene have been detected and indicated that they are associated with various disorders such as schizophrenia (OMIM: 181500), autistic spectrum disorders (ASD; OMIM:209850), Alzheimer disease (OMIM: 104300), temporal lobe epilepsy (OMIM: 616436), and abnormal development of the cerebral cortex and hippocampus. 8,9 Accumulating evidence has shown the fluctuation of reelin concentration in brain autopsies and also in the blood samples derived from patients affected by neuropsychiatric disorders such as schizophrenia, bipolar diseases, and ASD. [10][11][12] In a nutshell, although various studies confirm the roles of reelin in neurogenesis, the peripheral roles, e.g. in inflammatory pathways, are still unclear.
In humans, the expression of RELN is detectable in the liver, adrenal pituitary cells, fibroblast-like synovial cells, and also osteocyte cells. [13][14][15] On the other hand, the expression of RELN is in the lowest (or even undetectable) level in blood samples. Abnormal expression of RELN has been detected in different conditions, e.g. various revealed the association between the expression of the RELN gene with rheumatoid arthritis (RA; OMIM: 180300) and also other inflammatory disorders. 14,16,17 Accumulation of reelin has been detected in synovial fluid and also in sera of RA patients. 14 In 2012, Du et al showed that reelin could promote Pla2g7 gene expression through stimulating very-lowdensity lipoprotein (VLDL) receptors in mice. 18 This report was the first evidence suggesting the inflammatory role of reelin, although this pathway had not been confirmed in humans. The enzyme of platelet-activating factor-acetylhydrolase (PAF-AH) encoded by the PLA2G7 gene has an anti-inflammatory role that leads to plateletactivating factor (PAF) inactivation. 19 PAF increases the level of numerous important inflammatory factors, e.g. cyclooxygenase 2 (Cox2) and tumor necrosis factor-α (TNF-α), which play a central role in AS pathogenesis. 20 Furthermore, PAF functions as the main modulator in cartilage degeneration. 21 Reelin is composed of eight EGF-like domains separating each constituent tandem repeats (R1 to R8) into two distinct subdomains: A and B ( Figure 1A, 1B). 22 EGF-like domains are typically composed of six conserved cysteine residues that form three intra-domain disulfide bridges. 23 These domains play an important role in orchestrating protein-protein interactions ( Figure 1C), 24 protein folding, 25 and receptor binding process. 26 Substitutions in such domains have potential effects on protein secretion. 27 The S2486G mutation is located in the receptor-binding site of the reelin (6 th EGF-like domain). The central fragment of reelin, namely R3 to R6, is not only sufficient for reelin attachment to its specific receptors (VLDLR and ApoER2) but also takes a center stage in activating downstream signals transductions. 28 Various studies have been performed to cast light on the production and secretion of reelin in cell supernatant and lysates by constructing recombinant reelin plasmids containing receptor binding domains (R5-R6) and C-terminal region (CTR). 29 Other studies have elucidated that the presence of the CTR of reelin (which is comprised of 32 amino acids) is imperative for efficient downstream signal transduction. 29,30 In this study, we investigated the potential effect of S2486G mutation on reelin secretion; additionally, we evaluated its effect on the PLA2G7 gene expression and PAF-AH concentration by constructing a subregion of Human RELN containing the R5-R6 domains (wild-type and S2486G mutant) and the CTR. For the first time, we demonstrated that the S2486G mutation not only leads to a reduction of reelin secretion in CHO-K1 and HEK-293T cells, but also causes a significant decrease of both the PLA2G7 gene expression and the PAF-AH protein level in THP-1-monocyte/macrophages cells.

Structures and Alignments
The identified mutation, S2486G, is located in the 6 th EGF-like domain of the reelin. The protein families and domains were analyzed using ScanProsite 31 and sequence alignments of the human reelin protein were recruited using ClustalW. 32 The protein structure and mutation impacts on protein flexibility and stability were analyzed by PyMOL 33 and I-mutant V2.0 34 after the structure was built based on the Phyre2 35 and SWISS-PROT 36 in a distinct PDB file. The RAMPAGE online tool was applied to check the detailed residue-by-residue stereochemical quality based on a Ramachandran plot. 37 Plasmids Constructs PVP22/myc-His2 expressing plasmid (Invitrogen, Groningen, The Netherlands) was utilized to construct recombinant plasmids containing R5-R6 domains of reelin carrying on wild-type and S2486G mutant alleles and CTR. The cDNA of K562 cells (ATCC CCL 243) was used to amplify R5-R6 domains (amino acids 1917-2673) and CTR segment (amino acids 3365-3460) using PfuUltra High-Fidelity DNA Polymerase (Stratagene, La Jolla, CA, USA). The sequence of primers utilized in this study is described in Table S1. The VP22 gene was cut by EcoRI/XbaI restriction enzymes and replaced by CTR (Control plasmid); then, the R5-R6 fragments were added to the constructs employing HindIII/EcoRI restriction enzymes (R5-R6/CWT).

Site-Directed Mutagenesis
Site-directed mutagenesis using 'Overlap Extension PCR' was applied to introduce the mutation into the R5-R6 fragment. Two mutant primers that had common nucleotides site, around 13 bp, were designed for this purpose (Table S1). Then, two fragments were amplified from R5-R6/CWT construct by F-mutant and R-R5/ R6-primers to produce a fragment with 570-bp size. R-mutant and F-R5/R6-primers were utilized to create a fragment of 1714 bp. The fragments were isolated from the agarose gel using the High Pure PCR Product Purification Kit (ROCHE, Chicago, USA) and utilized as a template for amplifying a fragment with 2271 bp through external primers (R5/R6-Forward and R5/R6-Reverse primers). In the first step, the PCR condition was 30 s at 95°C, 30 seconds at 60°C, 30 seconds at 72°C, for 15 cycles (without primers) which was followed by 30 seconds at 95°C, 30 seconds at 52°C, 120 seconds at 72°C, for 35 cycles (including primers) and 10 minutes of final extension at 72°C. The PCR products were cloned followed by digestion by HindIII/EcoRI restriction enzymes. The produced constructs were assessed by Sanger sequencing to verify the introduction of the mutation and also any additional mutations due to possible PCR errors.
We used the GST 6XHis tag ELISA kit (ab128573, Abcam, Chicago, USA) and 6XHis tag antibody (ab9108, Abcam, Chicago, USA) with a sensitivity of 0.5 ng/mL for 6XHis-tagged protein. All the reagents were applied based on standard protocols. Finally, the standard curve was achieved by applying a serial dilution of the standard recombinant GST-6XHistag. This assay was performed in duplicate for both standard and test samples. The supernatants and cell lysates of non-transfected cells were used as negative controls.

PLA2G7 Gene Expression
The biphasic monocyte/macrophage cells of THP1 (ATCC ® TIB-202™; Rockville, MD, USA) were cultured in RPMI 1640 supplemented with 10% FBS, and 1X penicillin-streptomycin in an incubator with 5% CO 2 and 95% humidity. After 14 days, adherent cells in the macrophage phase were detached using Trypsin/EDTA 0.5%. After washing with PBS, almost 8×10 5 cells were used in five distinct groups. The first plate was utilized as control and another plate was treated with supernatant obtained from the CHO-K1 cells without transfection after 48h as 'Control Conditioned Medium (CCM)'. The media derived from transfected CHO-K1 cells with control, R5-R6/CWT, and R5-R6/CMUT plasmids were respectively added to the three remaining plates after normalization.
RNA extraction was achieved using RNX-Plus Solution (CinnaGen, Tehran, Iran) 3 hours after the treatment of the cells. Then, RNA concentration was measured with Nanodrop 2000 ® (Thermo Fisher Scientific, Wilmington, DE, USA) and the synthesis of cDNA was performed using the TAKARA kit (Takara, Otsu, Japan) in 25 μL reaction mix. All primers were designed by Primer3.0 online web tool 38 and qRT-PCR was performed for each sample. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as the reference gene and consequently, data were analyzed by applying the ΔΔCT method. 39 In this study, the Human PLA2G7 ELISA kit (ab235643, Abcam, USA) with a sensitivity of 45 pg/mL was used to evaluate the concentration of the PAF-AH enzyme in the supernatant of THP1 cells 16h after treatment with Control, CCM, R5-R6/CWT, and R5-R6/CMUT. ELISA assays were obtained from three independent experiments that were performed in duplicate for all samples and standards. The supernatants and cell lysates of non-transfected cells were used as a negative control for this test.
Statistical Analysis Statistical analyses were performed using SPSS v.25.0 (IBM Corp., Armonk, NY, USA) and GraphPad Prism v.7.0 (GraphPad Software Inc, San Diego, CA, USA). Differences between experimental conditions were analyzed by considering P values <0.05 as statistically significant.

In-Silico Prediction
To investigate the impact of the S2486G mutation on reelin secretion, we constructed two expressing plasmids designed as R5-6/C WT and R5-6/C MUT. The S2486G is located in the EGF-like domain of the sixth repeat of reelin protein and located near the second cysteine which is required for forming a disulfide bridge ( Figure 1B). In-silico prediction by I-Mutant2.0 34 showed that the missense S2486G mutation can change protein folding or stability. Additionally, the prediction of O-linked and N-linked glycosylation was achieved based on the average surface accessibility and composition profile of pattern algorithms of GlycoEP. 40 This result indicates that Serine of 2486 is a potential residue for O-linked glycosylation (residue score: -1.519, P value <0.05).
We used the RAMPAGE online tool to show the detailed residue-by-residue stereochemical quality by the Ramachandran plot. The modeled structure of reelin R5-R6 indicated almost 98% of residues in the most favored regions, around 2% of residues in the allowed regions, and only 0.3% of residues in the outlier regions, which suggested that the modeled structure of reelin R5-R6 was acceptable.

Reelin Secretion
The reelin secretion in the cell lysates and supernatants, which were collected at different times after transfection from CHO-K1 and HEK-293T cells (Figure 2A, 2B), was evaluated by GST 6XHis tag ELISA kit employing 6XHis tag antibody. Similarly, in-vitro analysis of some identified mutations in the RELN gene related to ASD has demonstrated that those mutations have similar effects on the secretion of the full-length or a subregion of the reelin reading frame. 41 The signal of 6XHis neither was detected in the conditional medium nor cell lysate from non-transfected CHO-K1 and HEK-293T cells. The cellular concentration of the R5-R6/CWT per 2×10 6 CHO-K1 cells was 5.93 ± 0.9 ng.mL -1 at 24 hours and 18.33 ± 2.08 ng.mL -1 at 48 hours and this amount in HEK-293T cells was 4.16 ± 1.04 ng.mL -1 at 24 hours and 14.4 ± 1.8 ng.mL -1 at 48 hours. While the intracellular concentration of R5-R6/ CMUT per 2×10 6 CHO-K1 cells was 3.3 ± 0.81 ng.mL -1 at 24 hours and 13.6 ± 1.52 ng.mL -1 at 48 hours, this amount was 1.58 ± 0.65 ng.mL -1 at 24 hours and 7.7 ± 2.06 ng.mL -1 at 48 hours in HEK-293T cells (Table 1). These results indicate that the intracellular concentrations of R5-R6/CWT at 24 hours and 48 hours after transient transfection of both cells were significantly more than those of the mutant.

PLA2G7 Gene Expression and PAF-AH Concentration
The impact of S2486G on the expression of PLA2G7 and the concentration of the encoded protein, PAF-AH, was assessed. Our data showed that the treatment of approximately 8×10 5 THP1 cells with R5-R6/CWT conditioned medium including 20 ng/mL for 3h increased the PLA2G7 gene expression (fold change: 2.53) ( Figure  3A and 3B). A similar treatment, except for 16 hours, increased the PAF-AH level to 2.56 fold in comparison with base cells. By treating THP1 cells with R5-R6/CMUT conditioned medium, on the other hand, these levels diminished lower than THP1 base levels of PLA2G7 and PAF-AH. These findings suggested that S2486G mutation caused a reduction in the promotion of the PLA2G7 gene and therefore, led to a reduction in the PLA2G7 mRNA level lower than the level of THP1 cells. In sum, results demonstrated that S2486G can be implied as a 'Gain of Function' mutation through an unknown pathway. For a detailed comparison, refer to Table 2.

Discussion
AS is a chronic inflammatory disease in which bone formation and cartilage degeneration are influenced by various mediators. Indeed, the etiopathogenesis of this disease, which is a prototype spondyloarthropathy, has not yet been deciphered. In this study, we aimed to investigate the possible disruptive effects of RELN S2486G mutation which warrant AS etiology through inflammatory mediators and PAF-AH.
To begin with, we showed that the S2486G mutation in the 6 th EGF-like domain of reelin reduces the secretion of reelin in both CHO-K1 and HEK-293T cell lines. As shown in Figures 2A and 2B, the whole level of reelin in both conditioned medium and cell lysate were less than the whole levels of wild-type reelin. Since the cloned fragments were expressed under the regulation of a quite strong promoter (cytomegalovirus promoter), the observed differences are probably due to the translational or post-translational modifications. I-Mutant2.0 predicted that the stability of the folded reelin protein is reduced. Serine is a polar amino acid while glycine is non-polar 42 ; thus, their substitution can affect the protein folding. The endoplasmic reticulum, as a quality control apparatus, has a major role in packaging and exporting the correctly folded proteins. 43 One of these quality control mechanisms is O-glycosylation in EGF-like domains. It has been demonstrated that serine residues, which are located between the second and third cysteine of the EGF-like domain (like S2486), are the target site of O-glycosylation by protein O-fucosyltransferase (POFUT1) enzyme which adds O-fucose to this site and thereby leads to stabilizing, accelerating of protein folding and secretion. 44 These data are predicted by GlycoEP. Therefore, the S2486G substitution can abolish this action and result in a misfolded protein which will be disrupted by the endoplasmic reticulum. However, we believe that this uncertainty will soon be removed by further studies. Our data suggest that reelin can be involved in  inflammation. It has been shown that the reelin levels are increased in both serum and synovial fluid of RA patients and could be used as a diagnostic marker. 45 Indeed, the liver is the main source of reelin in the blood. 46 Despite the pathophysiology of RA, which is based on bone destruction through activation of osteoclasts, 47 the main characterization of AS is cartilage erosion and formation of new bones in sacroiliac and facet joints mediated by osteoblast cells. 48 The potential role of reelin protein in the development of cartilage has been illustrated 49 ; the corticosterone hormone has a pivotal role in bone loss and leads to the elevation of reelin expression in the murine osteocyte cell line. 50,51 Other complementary studies have revealed the association of several variants in the RELN gene with otosclerosis, a type of conductive hearing loss, following abnormal bone remodeling in the otic capsule of the ear. 52 Additionally, it has been confirmed that reelin promotes Pla2g7 gene expression by stimulating macrophage VLDL receptors in mice. For the first time in humans, we detected the same pathway. PLA2G7 encodes PAF-AH 53 which belongs to the phospholipase A2 family. 54 Inflammatory mediators have a pivotal role in triggering inflammation in AS. 55,56 The PAF-AH protein hydrolyzes PAF, which is a potent pro-inflammatory agent and plays a staple role in inflammation, by releasing inflammatory eicosanoids and TNF-α from leukocytes and endothelial cells. Consequently, this process leads to more PAF releasing and causes increased platelet activation. 57 PAF-AH is generated largely by hepatocytes and macrophages; this protein is broadly scattered in human plasma, blood cells, and a variety of tissues. 58 Besides, this enzyme reduces the inflammatory responses and this protein is a well-known biomarker for cardiovascular disease. [59][60][61] In the in-vitro analysis, we showed that the S2486G caused the reduction of concentration and expression of PAF-AH and PLA2G7, respectively. This can justify their roles in expanded inflammation which is associated with AS pathophysiology. It has been illustrated that PAF-AH can regulate epithelial-mesenchymal transition, as well. Therefore, it could control the initiation of bone regulation in preosteoblasts. 62,63 Likewise, it has been elucidated that β-arrestin2 can upregulate PLA2G7 and inhibits osteoclastogenesis in vitro which resulted in decreased bone resorption in vivo by regulating RANKL/ OPG production and ephrins mRNAs. 64,65 Thus, PLA2G7 functions as a double-edged sword by playing a role in inflammation and osteogenesis, indirectly.
Herein, for the first time, we showed that S2486G is a pleiotropic mutation that reduced reelin secretion as well as decreased PLA2G7 gene expression. Even though by in-vitro analysis, we showed that S2486G results in reduced secretion of PAF-AH, other studies need to shed light on other molecular mechanisms, other networks, and important interactions of reelin in inflammatory pathways using appropriate animal models. In a nutshell, our data show the involvement of reelin in AS etiology through inflammation and also probably osteogenesis. We believe that the findings of this study hopefully broaden the horizons toward better understanding the molecular mechanisms contributing to AS. One inherent weakness of this study is that we should have evaluated the gene and protein expression using cartilage cell lines, in different stages, e.g. from chondrocytes to osteocytes, to show the involvement of reelin in chondrogenesis/osteogenesis pathways. We believe that future studies could fruitfully explore these issues.
In conclusion, this report points to the first association of the RELN gene with AS disease. The S2486G situated in the receptor binding site of reelin may alter the inflammatory or probably osteogenesis pathway for AS manifestations. However, further studies should be conducted to cast light on reelin roles in inflammation.

Authors' Contribution
MG, EEG: Conceptualization. EEG, ER and ARB: Methodology and data curation . EEG and ER: Software, writing the paper and

Conflict of Interest Disclosures
The authors declare no conflict of interest.

Ethical Statement
The study protocol was approved by the local medical ethics committee of Tarbiat Modares University, Tehran, Iran. The study was performed during 2016-2018 at Tarbiat Modares University, Tehran, Iran.