FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W4384704670> ?p ?o ?g. }

• W4384704670 abstract "Full text Figures and data Side by side Abstract Editor's evaluation Introduction Results Discussion Materials and methods Appendix 1 Appendix 2 Data availability References Decision letter Author response Article and author information Metrics Abstract Ossification of the posterior longitudinal ligament of the spine (OPLL) is an intractable disease leading to severe neurological deficits. Its etiology and pathogenesis are primarily unknown. The relationship between OPLL and comorbidities, especially type 2 diabetes (T2D) and high body mass index (BMI), has been the focus of attention; however, no trait has been proven to have a causal relationship. We conducted a meta-analysis of genome-wide association studies (GWASs) using 22,016 Japanese individuals and identified 14 significant loci, 8 of which were previously unreported. We then conducted a gene-based association analysis and a transcriptome-wide Mendelian randomization approach and identified three candidate genes for each. Partitioning heritability enrichment analyses observed significant enrichment of the polygenic signals in the active enhancers of the connective/bone cell group, especially H3K27ac in chondrogenic differentiation cells, as well as the immune/hematopoietic cell group. Single-cell RNA sequencing of Achilles tendon cells from a mouse Achilles tendon ossification model confirmed the expression of genes in GWAS and post-GWAS analyses in mesenchymal and immune cells. Genetic correlations with 96 complex traits showed positive correlations with T2D and BMI and a negative correlation with cerebral aneurysm. Mendelian randomization analysis demonstrated a significant causal effect of increased BMI and high bone mineral density on OPLL. We evaluated the clinical images in detail and classified OPLL into cervical, thoracic, and the other types. GWAS subanalyses identified subtype-specific signals. A polygenic risk score for BMI demonstrated that the effect of BMI was particularly strong in thoracic OPLL. Our study provides genetic insight into the etiology and pathogenesis of OPLL and is expected to serve as a basis for future treatment development. Editor's evaluation This study builds on previous work to explore the genetic causes of ossification of the posterior longitudinal ligament of the spine (OPLL). A meta-Genome wide association study is conducted to increase detection power and a disease subtype analysis is completed that provides new information on if all sites of OPLL have uniform causes. Using additional open-source data, the GWAS results are explored further to find putatively causative genes and to explore causative co-existing conditions such as obesity for OPLL. Overall this study is by far the most complete genetic exploration of this disease to date and is instructive for future studies that will lead to treatments for this condition. https://doi.org/10.7554/eLife.86514.sa0 Decision letter eLife's review process Introduction Ossification of the posterior longitudinal ligament of the spine (OPLL) is an incurable disease with progressive heterotopic ossification. It can occur at any spine level from the cervical to the lumbar spine, and ossified ligaments compress the spinal cord and roots, leading to a severe neurological deficit (Matsunaga and Sakou, 2012). OPLL is a common disease; however, its frequency varies depending on the region of the world; high in Asian countries (0.4–3.0%), especially Japan (1.9–4.3%), compared with Europe and the United States (0.1–1.7%) (Matsunaga and Sakou, 2012; Ohtsuka et al., 1987; Sohn and Chung, 2013; Yoshimura et al., 2014). However, its etiology and pathogenesis remain unknown. Histological studies suggest that OPLL develops through endochondral ossification (Sato et al., 2007; Sugita et al., 2013). In recent years, OPLL has been reported to have different clinical characteristics depending on the affected region: higher body mass index (BMI), earlier-onset of symptoms, and more diffuse progression of OPLL over the entire spine in the thoracic type of OPLL (T-OPLL) than in the cervical type (C-OPLL) (Endo et al., 2020; Hisada et al., 2022). This fact suggests that there may be differences in etiology and pathogenesis for each subtype of OPLL. Currently, there is no therapeutic or preventive measure for OPLL other than surgery to decompress the spinal cord and roots. Therefore, it is necessary to clarify its etiology and pathogenesis to develop effective measures to prevent and treat OPLL. OPLL is assumed to be a polygenic disease where complex genetic and environmental factors interact. Epidemiological studies have reported the relationship between OPLL and various other traits, especially type 2 diabetes (T2D) (Akune et al., 2001; Kobashi et al., 2004), high BMI (Hou et al., 2017; Kobashi et al., 2004), low inorganic phosphate, X-linked hypophosphatemic rickets (Chesher et al., 2018), and increased C-reactive protein (Kawaguchi et al., 2017). Of these traits, T2D has been the focus of attention for a long time (Akune et al., 2001; Kobashi et al., 2004). Furthermore, because of the high incidence within families and close relatives in previous epidemiological studies, genetic factors have long been considered in OPLL development (Matsunaga et al., 1999; Sakou et al., 1991; Terayama, 1989), although there are no previous papers evaluating the heritability of OPLL such as twin studies. To understand the genetic factors associated with OPLL, we previously conducted a genome-wide association study (GWAS) and found six significant loci (Nakajima et al., 2014). In subsequent in silico and in vitro functional studies, we identified RSPO2 as a susceptibility gene for OPLL, and the role of Wnt signaling in the pathogenesis of OPLL was clarified (Nakajima et al., 2016). However, the pathogenesis of this condition remains largely unknown. In this study, to clarify the etiology and pathogenesis of OPLL, we conducted a meta-analysis of GWASs and various post-GWAS analyses. We identified 14 significant loci, including 8 previously unreported susceptibility loci. Using a gene-based analysis (de Leeuw et al., 2015) and summary data-based Mendelian randomization (SMR) (Zhu et al., 2016), we identified three candidate genes for each. Using a genetic correlation analysis and a subsequent Mendelian randomization (MR) study, we identified a causal effect of high BMI on OPLL. A polygenic risk score (PRS) of BMI demonstrated the heterogeneity of the impact of obesity on OPLL subtypes. Results Novel susceptibility loci in OPLL We conducted three GWASs (set 1–3) in the Japanese population (Supplementary file 1). After quality control of single-nucleotide polymorphism (SNP) genotyping data, we performed imputation and association analyses independently for each GWAS. Subsequently, we performed a fixed-effects meta-analysis combining the three GWASs (ALL-OPLL: a total of 2010 cases and 20,006 controls; Figure 1—figure supplement 1) and identified 12 genome-wide significant loci (p<5.0 × 10−8) (Figure 1). The genomic inflation factor (λGC) was 1.11 and showed slight inflation in GWAS; however, the intercept in linkage disequilibrium (LD) score regression (Bulik-Sullivan et al., 2015) was 1.03, indicating that inflation of the statistics was mainly from polygenicity and minimal biases of the association results (Figure 1—figure supplement 2). Figure 1 with 17 supplements see all Download asset Open asset Meta-analysis of genome-wide association studies (GWAS) identified 14 significant loci in ossification of the posterior longitudinal ligament of the spine (OPLL). Manhattan plot showing the -log10 p-value for each single-nucleotide polymorphism (SNP) in the meta-analysis. The values were plotted against the respective chromosomal positions. The horizontal red line represents the genome-wide significance threshold (p=5.0 × 10–8). Red and blue points represent the SNPs in the new and known loci, respectively. Next, we conducted a stepwise conditional analysis to detect multiple independent signals. We detected two additional independent signals that showed genome-wide significance after conditioning (Supplementary file 2): rs35281060 (12p12.3, p=1.04 × 10−10) and rs1038666 (12p11.22, p=2.37 × 10−10) (Figure 1—figure supplement 3F–I). We also detected one additional signal (rs61915977, 12p11.22 p=1.39 × 10−6) that reached locus-wide significance (p<5.0 × 10–6) (Supplementary file 2). Thus, the meta-analysis and conditional analysis identified 14 genome-wide significant OPLL loci, including 8 novel loci. Significant associations of the six previously reported loci (Nakajima et al., 2014) were observed in the present study (Table 1, Figure 1, Figure 1—figure supplement 3). The estimated proportion of the phenotypic variance explained by all the variants used in the study was 53.1% (95% confidence interval [CI] 40.6–65.6%), indicating that OPLL has a high heritability. The lead variants of the 14 loci explained 6.5% of the phenotypic variance. Together with the LD score regression results, OPLL is a highly polygenic disease. Table 1 Genome-wide significant loci in ossification of the posterior longitudinal ligament of the spine. SNPCHRPosition(Region start- end)GeneNovel/knownREFALTOPLLpOR(95% CI)Phet†GWAS 1GWAS 2GWAS 3ALT freq.pOR(95% CI)ALT freq.pOR(95% CI)ALT freq.pOR(95% CI)casecontrolcasecontrolcasecontrolrs46659722 (p23.3)27598097(26598097–28598097)SNX17(intronic)NovelTCALL7.00E-091.23(1.15–1.32)0.180.4830.4339.91E-071.27(1.16–1.40)0.4740.4253.73E-041.26(1.11–1.43)0.4410.4305.65E-011.05(0.88–1.26)Cervical5.38E-051.25(1.12–1.39)0.920.4810.4331.19E-031.25(1.09–1.44)0.4690.4251.59E-021.24(1.04–1.48)---Thoracic3.49E-021.14(1.01–1.28)0.500.4780.4333.51E-021.24(1.01–1.51)0.4540.4253.05E-011.16(0.87–1.53)0.4410.4305.65E-011.05(0.88–1.26)rs9274856 (p21.1)44538139(43529797–45538139)LOC105375075,SUPT3H(intergenic)KnownGAALL2.30E-090.76(0.70–0.83)0.250.8240.8641.22E-070.72(0.64–0.82)0.8430.8606.39E-020.86 (0.73–1.01)0.8290.8727.98E-030.74(0.59–0.92)Cervical3.77E-030.82(0.71–0.94)0.460.8350.8645.95E-030.79(0.66–0.93)0.8460.8602.40E-010.87(0.70–1.09)--Thoracic7.48E-060.72(0.62–0.83)0.920.8180.8642.63E-030.69(0.55–0.88)0.8150.8604.16E-020.71(0.51–0.99)0.8290.8727.98E-030.74(0.59–0.92)rs3748108 (q23.1)109096029(108022775–110588327)RSPO2(upstream)KnownGAALL1.03E-150.75(0.70–0.81)0.930.3230.3879.56E-100.74(0.68–0.82)0.3280.3852.72E-050.77(0.68–0.87)0.3290.3952.06E-030.76(0.64–0.90)Cervical6.04E-080.75(0.67–0.83)0.140.3370.3876.95E-040.79(0.69–0.91)0.3000.3857.42E-060.67(0.56–0.80)--Thoracic2.66E-070.73(0.65–0.82)6.6E-020.2820.3872.81E-060.62(0.50–0.75)0.3660.3854.85E-010.91(0.70–1.19)0.3290.3952.06E-030.76(0.64–0.90)rs18982878 (q23.3)117579970(116484907–118588193)LINC00536,EIF3H(intergenic)KnownACALL2.18E-100.80(0.75–0.86)0.160.6050.6682.90E-090.75(0.69–0.83)0.6250.6648.33E-030.85(0.75–0.96)0.6330.6641.85E-010.89(0.74–1.06)Cervical1.10E-020.87(0.78–0.97)0.510.6330.6681.61E-020.85(0.74–0.97)0.6410.6642.92E-010.91(0.77–1.08)--Thoracic2.18E-040.80(0.71–0.90)0.100.5840.6687.40E-050.68(0.56–0.82)0.6370.6643.80E-010.88(0.67–1.16)0.6330.6641.85E-010.89(0.74–1.06)rs3550524811 (q14.2)86830927(85724086–87887931)TMEM135(intronic)NovelTAALL6.75E-100.81(0.75–0.86)0.440.6240.6651.76E-040.84(0.76–0.92)0.5940.6597.06E-060.76(0.67–0.85)0.6040.6491.90E-020.81(0.68–0.97)Cervical1.06E-040.81(0.73–0.90)2.7E-020.6400.6651.03E-010.89(0.78–1.02)0.5770.6593.30E-050.70(0.60–0.83)--Thoracic4.53E-040.81(0.72–0.91)0.650.6050.6657.62E-030.77(0.64–0.93)0.6350.6594.67E-010.90(0.69–1.19)0.6040.6491.90E-020.81(0.68–0.97)rs3528106012 (p12.3)19976182(18955794–20077000)AEBP2,LINC02398(intergenic)NovelTGTALL1.39E-120.79(0.74–0.84)0.580.4510.5003.50E-060.81(0.74–0.88)0.4510.5062.92E-050.77(0.69–0.87)0.4290.5054.50E-040.73(0.61–0.87)Cervical1.06E-050.80(0.72–0.88)0.430.4560.5002.74E-030.82(0.72–0.93)0.4450.5068.81E-040.76(0.64–0.89)--Thoracic1.48E-060.75(0.67–0.85)0.380.4240.5005.18E-040.72(0.60–0.87)0.4820.5063.94E-010.89(0.69–1.16)0.4290.5054.50E-040.73(0.61–0.87)rs1084144212 (p12.2)20213600(20077000–21247540)LINC02398(ncRNA_intronic)KnownTCALL1.03E-120.78(0.73–0.84)0.610.4220.4891.07E-080.77(0.70–0.84)0.4240.4806.60E-050.78(0.69–0.88)0.4180.4567.56E-020.85(0.71–1.02)Cervical1.57E-080.74(0.67–0.82)0.710.4130.4892.87E-060.73(0.64–0.83)0.4200.4801.40E-030.76(0.65–0.90)--Thoracic1.80E-040.80(0.71–0.90)0.600.4170.4891.91E-030.74(0.62–0.90)0.4320.4801.32E-010.82(0.62–1.06)0.4180.4567.56E-020.85(0.71–1.02)rs1104952912 (p11.22)28471504(27300776–28800000)CCDC91(intronic)KnownCTALL1.01E-130.77(0.72–0.83)0.630.5690.6296.72E-090.76(0.69–0.83)0.5640.6271.31E-050.76(0.67–0.86)0.5720.6015.63E-020.84(0.70–1.00)Cervical2.57E-060.78(0.70–0.87)0.890.5750.6293.06E-040.78(0.69–0.90)0.5660.6272.55E-030.77(0.65–0.91)--Thoracic9.93E-060.77(0.68–0.86)0.290.5410.6296.68E-050.68(0.57–0.82)0.5770.6271.15E-010.80(0.61–1.05)0.5720.6015.63E-020.84(0.70–1.00)rs103866612 (p11.22)29085005(28800000–30107711)CCDC91,FAR2(intergenic)NovelGAALL5.09E-100.81(0.76–0.87)0.060.5730.6091.43E-030.86(0.79–0.95)0.5320.6138.18E-080.72(0.64–0.81)0.5530.6012.03E-020.81(0.68–0.97)Cervical5.48E-050.81(0.74–0.90)0.290.5690.6091.12E-020.85(0.75–0.96)0.5460.6139.33E-040.76(0.65–0.89)--Thoracic2.89E-060.76(0.68–0.85)0.220.5510.6099.54E-030.79(0.65–0.94)0.4960.6133.43E-040.62(0.48–0.81)0.5530.6012.03E-020.81(0.68–0.97)rs1115773314 (q21.3)50727523(49727523–51729133)L2HGDH(intronic)NovelGAALL2.65E-081.21(1.13–1.29)0.580.4630.4232.90E-041.18(1.08–1.30)0.4780.4197.52E-051.27(1.13–1.43)0.4600.4267.18E-021.17(0.99–1.38)Cervical5.28E-041.20(1.08–1.32)0.740.4610.4231.20E-021.18(1.04–1.34)0.4680.4191.60E-021.22(1.04–1.44)--Thoracic1.48E-031.20(1.07–1.34)0.190.4460.4232.53E-011.11(0.93–1.34)0.5170.4192.89E-031.49(1.15–1.93)0.4600.4267.18E-021.17(0.99–1.38)rs5825559814 (q23.2)62131805(61131805–63131805)FLJ22447,HIF1A-AS1(intergenic)NovelCTALL2.16E-080.81(0.75–0.87)0.760.2760.3191.75E-040.83(0.75–0.91)0.2780.3241.67E-030.81(0.71–0.92)0.2720.3244.88E-030.76(0.63–0.92)Cervical2.19E-030.84(0.75–0.94)0.360.2870.3196.19E-020.87(0.76–1.01)0.2710.3249.53E-030.79(0.65–0.94)--Thoracic1.36E-050.75(0.66–0.86)0.960.2540.3194.37E-030.73(0.59–0.91)0.2700.3248.52E-020.77(0.57–1.04)0.2720.3244.88E-030.76(0.63–0.92)rs18964674215 (q25.3)88017055(87017055–89017055)AGBL1,LINC00052(intergenic)NovelGAALL2.13E-082.03(1.59–2.61)0.420.0260.0122.49E-072.31(1.68–3.17)0.0170.0117.34E-021.57(0.96–2.58)0.0200.0127.05E-021.85(0.95–3.60)Cervical3.25E-052.14(1.50–3.07)0.670.0250.0122.88E-042.27(1.46–3.53)0.0210.0113.81E-021.92(1.04–3.56)--Thoracic1.77E-021.72(1.10–2.70)0.570.0220.0126.48E-021.88(0.96–3.68)0.0100.0117.99E-010.83(0.20–3.46)0.0200.0127.05E-021.85(0.95–3.60)rs37698937616 (q22.1)69854329(68854329–70854329)WWP2(intronic)NovelTTAGALL1.08E-080.79(0.73–0.86)0.450.6600.6934.48E-050.80(0.71–0.89)0.6770.7021.28E-020.83(0.72–0.96)0.6390.6997.28E-040.71(0.58–0.87)Cervical2.70E-040.80(0.71–0.90)0.830.6580.6932.65E-030.79(0.68–0.92)0.6730.7023.87E-020.81(0.66–0.99)--Thoracic4.10E-070.71(0.62–0.81)0.810.6310.6935.18E-040.68(0.54–0.84)0.6630.7021.07E-010.77(0.56–1.06)0.6390.6997.28E-040.71(0.58–0.87)rs614044220 (p12.3)7829397(6713042–8882559)MIR8062,HAO1(intergenic)KnownCAALL2.70E-141.39(1.28–1.51)0.070.2050.1501.41E-111.48(1.32–1.66)0.1970.1533.33E-051.38(1.18–1.60)0.1550.1435.10E-011.08(0.85–1.38)Cervical4.47E-081.42(1.25–1.61)0.610.2040.1503.67E-061.46(1.24–1.71)0.1960.1533.06E-031.36(1.11–1.67)--Thoracic2.45E-021.19(1.02–1.39)0.220.1970.1505.64E-031.39(1.10–1.76)0.1530.1539.38E-011.01(0.71–1.46)0.1550.1435.10E-011.08(0.85–1.38) SNP, single-nucleotide polymorphism; CHR, chromosome; REF, reference; ALT, alternative; OPLL, ossification of the posterior longitudinal ligament of the spine; GWAS, genome-wide association study; OR, odds ratio; CI, confidence interval; ALL, cervical + thoracic + others. *Gene in or near region of association. † Phet was derived from a Cochran’s Q-test for heterogeneity. Adjacent to lead variants in the novel loci, we found several candidate genes (Figure 1) reported to be related to osteogenesis and could be connected to OPLL development. TMEM135 (transmembrane protein 135), a gene in the newly identified significant locus (11q14.2), is a multi-transmembrane protein with seven transmembrane helices of high confidence. It is more strongly expressed in multipotent adipose tissue-derived stem cells committed to osteoblastic cells than the adipogenic lineage (Scheideler et al., 2008). WWP2 (WW domain-containing E3 ubiquitin-protein ligase 2), the nearest gene to rs376989376 (the lead SNP in 16q22.1), was recently reported to serve as a positive regulator of osteogenesis by augmenting the transactivation of RUNX2, a master regulator of osteoblast differentiation as well as for chondrocyte maturation during skeletal development (Zhu et al., 2017). All lead SNPs and SNPs in high LD (r2 > 0.8) with them in previously unreported significant loci were in intron or intergenic regions, and none of them were exonic variants (Supplementary file 3). To prioritize putative causal variants, we conducted a Bayesian statistical fine-mapping analysis for significant loci using FINEMAP (Benner et al., 2016). The lead SNPs had the highest posterior probability (PP) in any significant region, and two of them were higher than 0.5: rs4665972 (2p23.3, p=0.548) and rs1038666 (12p11.22, p=0.533) (Supplementary file 4). Statistical power analysis We examined the statistical power for minor allele frequency (MAF) and odds ratio of lead SNPs within the 14 independent significant regions in GWAS meta-analysis for ALL-OPLL. The results showed that all had a power greater than 0.5 for a significance level of p-value = 5 × 10−8, and nine had a power greater than 0.8 (Figure 1—figure supplement 4). Enrichment in genes involved in bone metabolism We conducted a gene set enrichment analysis implemented in FUMA (Watanabe et al., 2017). We found significant enrichment in the set related to bone mineral density (BMD): BMD of the heel (p=8.60 × 10−8), pediatric lower limb (p=9.24 × 10−5), and pediatric total body less head (p=2.68 × 10−4) (Supplementary file 5), compatible with the critical roles of bone metabolism in OPLL. However, we observed no significant enrichment in BMD in adults measured by dual-energy X-ray absorptiometry in this analysis. Identification of novel candidate genes missed by the GWAS meta-analysis To identify other candidate genes, we conducted a gene-based association analysis (de Leeuw et al., 2015; Watanabe et al., 2017). We found three additional genes significantly associated with OPLL: EIF3E, EMC2, and TMEM135 (Figure 2, Supplementary file 6). EIF3E and EMC2 are in the same locus most strongly associated with OPLL as RSPO2 (8q23.1.). EIF3E (eukaryotic translation initiation factor 3 subunit E) encodes a protein that is a component of the eukaryotic translation initiation factor 3 (eIF-3) complex, which functions in and is essential for several steps in the initiation of protein synthesis (Lee et al., 2015; Masutani et al., 2007). A proteomics study in a rat model of heterotopic ossification reported that Eif3e was upregulated in ossified tissues and may be involved in tissue ossification by regulating hypoxia-inducible factor (HIF) signaling, which has an important role in osteogenesis (Wei et al., 2022). EMC2 (endoplasmic reticulum membrane protein complex subunit 2) encodes a part of the endoplasmic reticulum membrane protein complex (EMC) that functions in the energy-independent insertion of newly synthesized membrane proteins into the endoplasmic reticulum membrane, an essential cellular process (Chitwood et al., 2018; O’Donnell et al., 2020). However, basic experiments evaluating the effects of EMC2 on ligament and bone tissue have not been reported, and the mechanisms involved in OPLL are unknown. On the other hand, this analysis reinforced the possible involvement of TMEM135 in the development of OPLL. Figure 2 Download asset Open asset Gene-based association analysis identified five significantly associated genes in ossification of the posterior longitudinal ligament of the spine (OPLL). Manhattan plot showing the -log10 p-value for each gene in the analysis. The values were plotted against the respective chromosomal positions. The horizontal red lines represent significance threshold (p=5.0 × 10–8). The lack of exonic variants suggests that altering gene expression levels is a key function of OPLL-associated variants. By searching expression quantitative trait loci (eQTL) data in all available tissues in GTEx (Consortium, 2015), we found 26 transcripts with cis-eQTL variants associated with OPLL signals; of these, 20 transcripts were in the novel loci (Supplementary file 7). Furthermore, SMR (Zhu et al., 2016) revealed a total of 10 gene–tissue pairs (three unique genes, namely, RSPO2, PLEC, and RP11-967K21.1) that surpassed the genome-wide significance level (PSMR < 8.4 × 10–6) without heterogeneity (PHEIDI < 0.05) (Supplementary file 8). RSPO2 is located in the most significant locus in GWAS meta-analysis, and its functions related to OPLL were elucidated in a past study (Nakajima et al., 2016). PLEC is expressed in various tissues, including muscles and fibroblasts (Consortium, 2015), and PLEC deficiency causes epidermolysis bullosa simplex with muscular dystrophy (OMIM 226670) (Smith et al., 1996), in which osteoporosis frequently develops (Chen et al., 2019). Since increased expression of PLEC was estimated to have a causal effect on OPLL (Figure 1—figure supplement 5, Supplementary file 8), these results suggest that PLEC is a likely causal gene of OPLL. As for RP11-967K21.1, its function in OPLL development is currently unknown and is expected to be elucidated in future studies. Cell groups and cell types related to OPLL We conducted partitioning heritability enrichment analyses to investigate cell groups related to OPLL. We observed significant enrichment in the active enhancers of the connective/bone cell group and the immune/hematopoietic cell group (Supplementary file 9). We then analyzed each cell type belonging to these groups and found significant enrichment of H3K27ac in chondrogenic differentiation cells (Supplementary file 10). These results concord with previous findings that in OPLL chondrocyte differentiation in the endochondral ossification process occurs (Sugita et al., 2013) and provide new insights into the involvement of immune system cells in OPLL development, which has received little attention to date. Subtype analyses of OPLL Subtype-stratified GWAS meta-analyses were also conducted: cervical (C)-OPLL (820 cases and 14,576 controls) and thoracic (T)-OPLL (651 cases and 20,007 controls). Subsequently, we identified three significant loci for C-OPLL and nine significant loci for T-OPLL (Figure 1—figure supplements 6–9, Supplementary file 11). Of these loci, one in the C-OPLL analysis and nine in the T-OPLL analysis were not identified in the analysis of ALL-OPLL and other OPLL subtypes. However, most of the lead SNPs in these significant loci were rare variants. We cannot determine that these are the causal variants based on the present results alone, but there was an interesting variant among them. rs74707424, a leading SNP in the significant locus (19p12), is located in the 3′-untranslated region of the ZBTB40 gene. In a recent study using primary osteoblasts of mouse calvaria, Doolittle et al. reported that Zbtb40 functions as a regulator of osteoblast activity and bone mass, and knockdown of Zbtb40, but not Wnt4, in osteoblasts drastically reduced mineralization (Doolittle et al., 2020). We did not find significant genes in the gene-based analysis (data not shown). Expression of candidate genes in the spinal ligament in humans and mice We then examined the expression of 23 genes of interest (ALL-OPLL GWAS: 19 genes; gene-based analysis: 2 genes; and SMR: 2 genes) in the spinal ligament, a target tissue of OPLL (see Supplementary file 12 for detailed information on the number of genes). We used the deposited RNA-sequencing (RNA-seq) data in the spinal ligament (yellow ligament) in patients with OPLL and cervical spondylotic myelopathy (CSM) (GSE188760), and chondrogenic differentiation of human spinal ligament cells and controls (GSE188759) (Tachibana et al., 2022). Both data sets included 20/23 genes, of which we found 14/20 (70%) and 15/20 (75%) expressed in spinal ligament tissue in GSE188760 and GSE188759, respectively. In addition, the expression tended to be different between OPLL patients and CSM patients. Especially, the expressions of WWP2, EIF3H, and SNX17 showed nominally significant differences (see ‘Materials and methods’ and Figure 1—figure supplement 10). WWP2 was more highly expressed in chondrogenic differentiated ligament cells than in undifferentiated ligament cells (see ‘Materials and methods’ and Figure 1—figure supplement 11). The expression of WWP2 has a positive effect on bone formation (Scheideler et al., 2008). We should revalidate these results with larger data sets in the future. Next, to further explore expressions of these genes in single-cell levels, we used deposited single-cell RNA sequencing (scRNAseq) data of Achilles tendon cells in murine ossification models: burn/tenotomy heterotopic ossification model (GSE126060) (Sorkin et al., 2020) and Achilles tendon puncture model (GSE188758) (Tachibana et al., 2022). The Uniform Manifold Approximation and Projection (UMAP) identified 13 and 9 clusters in GSE126060 and GSE188758, respectively (Figure 1—figure supplements 12–15). Both data sets contained information on the same 14/23 genes. We confirmed that 12/14 (85.7%) of these genes are expressed in both mesenchymal and immune-related cells (macrophage, dendritic cell, and lymphocyte) in both data sets. These results are concordant with the results of partitioning heritability analysis, suggesting that not only the mesenchymal cells which differentiate into ligament and chondrocyte cells but also the immune cells are involved with ligament ossification. We also conducted the same analyses for the candidate genes uniquely found in T- and C-OPLL and found the expression of most of the genes in ligamentous tissues. Causality of high BMI on OPLL Epidemiological studies have suggested a relationship between OPLL and various other diseases and traits (Akune et al., 2001; Endo et al., 2020; Kawaguchi et al., 2017; Kobashi et al., 2004), particularly with T2D (Akune et al., 2001; Kobashi et al., 2004). We investigated their relationship with OPLL using the GWAS data. We first calculated the genetic correlation between OPLL and 96 complex traits (mean number of around 130K) (Akiyama et al., 2019; Akiyama et al., 2017; Ishigaki et al., 2020; Kanai et al., 2018; see Supplementary file 13 for the traits analyzed). We found a positive genetic correlation between OPLL and BMI and T2D. The genetic correlation estimate (rg) was higher in the BMI group than in the T2D group. In addition, we identified new negative correlations between cerebral aneurysms (Figure 3, Supplementary file 13). Figure 3 Download asset Open asset Genetic correlation between ossification of the posterior longitudinal ligament of the spine (OPLL) and other complex traits. Significant positive correlations with body mass index (BMI) and type 2 diabetes, and negative correlations with cerebral aneurysm were observed. Error bars indicate 95% confidence intervals. Red color gradations represent the level of p-value. Noted by asterisk is the significant correlation (false discovery rate [FDR] < 0.05). Next, we conducted a two-sample MR using summary data from GWASs (Akiyama et al., 2017; Bakker et al., 2020; Kemp et al., 2017; Spracklen et al., 2020) to assess the causal effects of these significant traits in genetic correlation analysis on OPLL (Evans and Davey Smith, 2015; Lawlor et al., 2008; Figure 4—figure supplement 1, Supplementary file 14). The result of genetic correlation analysis for osteoporosis barely did not reach the false discovery rate (FDR)-corrected significance level, but we included it in the MR evaluation because there have been several previous reports of a strong trend toward whole-body ossification in OPLL patients (Hukuda et al., 1983; Mori et al., 2016; Nishimura et al., 2018; Yoshii et al., 2019). In the analysis, we used BMD, the main diagnostic criteria item for osteoporosis. BMD in the spine may reflect artifacts from OPLL itself, but higher BMD was also reported in patients with OPLL in the femur and the radius, a non-weight-bearing bone (Sohn and Chung, 2013; Yamauchi et al., 1999). The significant causal effect of increased BMI on ALL-OPLL was estimated using the inverse variance weighted (IVW) method and the weighted median method (Figure 4, Figure 4—figure supplement 2, Supplementary file 15). The average pleiotropic effect of the MR-Egger regression intercept was close to zero (MR-Egger intercept = 0.005, p=0.581), indicating no evidence of the influence of directional pleiotropy (Figure 4—figure supplement 2, Supplementary file 16). We also assessed potential bias in the MR with a leave-one-out analysis and funnel plots; however, we did not identify any obvious bias (Figure 4—figure supplement 3). In contrast, we could not find any causal effects of T2D on ALL-OPLL with any MR methods (Figure 4, Figure 4—figure supplement 4, Supplementary file 15). As for BMD, we found a weak but significant causal effect of increased BMD on ALL-OPLL using multiple MR methods (Figure 4, Figure 4—figure supplement 5, Supplementary file 15), and the involvement of factors that stimulate bone formation in OPLL was suggested. Regarding cerebral aneurysms, the direction of the beta estimates differed accordi" @default.
• W4384704670 created "2023-07-20" @default.
• W4384704670 date "2023-05-22" @default.
• W4384704670 modified "2023-09-26" @default.
• W4384704670 title "Decision letter: Genetic insights into ossification of the posterior longitudinal ligament of the spine" @default.
• W4384704670 doi "https://doi.org/10.7554/elife.86514.sa1" @default.
• W4384704670 hasPublicationYear "2023" @default.
• W4384704670 type Work @default.
• W4384704670 citedByCount "0" @default.
• W4384704670 crossrefType "peer-review" @default.
• W4384704670 hasBestOaLocation W43847046701 @default.
• W4384704670 hasConcept C105702510 @default.
• W4384704670 hasConcept C169760540 @default.
• W4384704670 hasConcept C205383261 @default.
• W4384704670 hasConcept C2777604421 @default.
• W4384704670 hasConcept C2779492255 @default.
• W4384704670 hasConcept C2780775167 @default.
• W4384704670 hasConcept C60644358 @default.
• W4384704670 hasConcept C71924100 @default.
• W4384704670 hasConcept C86803240 @default.
• W4384704670 hasConceptScore W4384704670C105702510 @default.
• W4384704670 hasConceptScore W4384704670C169760540 @default.
• W4384704670 hasConceptScore W4384704670C205383261 @default.
• W4384704670 hasConceptScore W4384704670C2777604421 @default.
• W4384704670 hasConceptScore W4384704670C2779492255 @default.
• W4384704670 hasConceptScore W4384704670C2780775167 @default.
• W4384704670 hasConceptScore W4384704670C60644358 @default.
• W4384704670 hasConceptScore W4384704670C71924100 @default.
• W4384704670 hasConceptScore W4384704670C86803240 @default.
• W4384704670 hasLocation W43847046701 @default.
• W4384704670 hasOpenAccess W4384704670 @default.
• W4384704670 hasPrimaryLocation W43847046701 @default.
• W4384704670 hasRelatedWork W2003746388 @default.
• W4384704670 hasRelatedWork W2068118938 @default.
• W4384704670 hasRelatedWork W2094258375 @default.
• W4384704670 hasRelatedWork W2108473458 @default.
• W4384704670 hasRelatedWork W2122581773 @default.
• W4384704670 hasRelatedWork W2138369697 @default.
• W4384704670 hasRelatedWork W2273334842 @default.
• W4384704670 hasRelatedWork W2330893367 @default.
• W4384704670 hasRelatedWork W69667030 @default.
• W4384704670 hasRelatedWork W3127199871 @default.
• W4384704670 isParatext "false" @default.
• W4384704670 isRetracted "false" @default.
• W4384704670 workType "peer-review" @default.