Proteome-wide Mendelian randomization identifies circulating proteins causally associated with childhood obesity

Proteome-wide Mendelian randomization identifies circulating proteins causally associated with childhood obesity | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Proteome-wide Mendelian randomization identifies circulating proteins causally associated with childhood obesity Despoina Manousaki, Raphael Avocegamou, Basile Jumentier, Kaossarath Fagbemi, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6189469/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background Childhood obesity is a major public health problem, affecting one in 5 youths. We aimed to characterize biomarkers for pediatric obesity among circulating proteins using Mendelian randomization (MR). Methods We utilized genome-wide significant cis- protein quantitative trait loci (pQTL) from large proteomic GWAS as instruments for circulating protein levels. Cis -pQTL effects on childhood body mass index (BMI) were retrieved from a European GWAS of 39,620 children. MR Wald ratios were calculated to estimate causal effects of each protein on childhood BMI. Sensitivity analyses, including colocalization and phenome-wide association studies (PheWAS), were performed for the candidate proteins to test for violation of the MR assumptions. Replication analyses were conducted using independent GWAS datasets, complemented by reverse MR and tissue enrichment analyses. Findings Among 535 tested proteins, three colocalized and demonstrated decreasing effects on BMI: endoglin (ENG; MR beta: -0.07, 95% CI [-0.10, -0.04], P = 4.4×10⁻⁵), fatty acid binding protein 4 (FABP4; MR beta: -0.33, 95% CI [-0.50, -0.16], P = 1.3×10⁻⁴), and Nectin-like protein-2 (MR beta: -0.26, 95% CI [-0.37, -0.15], P = 5.45×10⁻⁵). Reverse causation was identified for FABP4, suggesting a compensatory mechanism. Conclusion We identified ENG, FABP4, and Nectin-like protein-2 as potential causal blood biomarkers or drug targets for pediatric obesity. Health sciences/Diseases/Endocrine system and metabolic diseases/Obesity Health sciences/Endocrinology/Endocrine system and metabolic diseases/Obesity proteins pediatric obesity Mendelian randomization causal inference Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Childhood obesity is a global health issue( 1 ) affecting almost one in five children and has been acknowledged as a serious public health concern due to its high morbidity rates( 2 ). In comparison to children with a normal weight, children with obesity, defined as having a body mass index (BMI) two standard deviations above the age and sex-adjusted median, have an increased likelihood of living with obesity in adulthood( 3 ) and are at a higher risk of developing long-term cardiometabolic, psychosocial and musculoskeletal complications ( 4 ) ( 5 ). The identification of early biomarkers for pediatric obesity is key for the development of age-specific screening tools or new therapies. With the advance of high throughput proteomics, circulating proteins represent a valuable source for biomarker discovery, because their circulating abundances are measurable and possibly modifiable. For instance, the Glucagon-like peptide-1 (GLP-1) is a peptide detected in both the intestines and the blood that has been identified as a therapeutic target for both adult and pediatric obesity( 6 ). However, measuring serum proteins is limited by prohibitive costs and the small sample sizes of the available pediatric cohorts. Moreover, observational proteomic studies are plagued by measurement errors, and bias due to unmeasured confounding and reverse causation( 7 ). Mendelian randomization (MR) is a method providing an instrumental variable framework to mitigate the above biases and infer causality( 8 , 9 ). This approach utilizes genetic variants, randomly allocated at conception, as instruments for a biomarker to evaluate the causal impact of this biomarker on a disease or a trait. MR leans on three pivotal assumptions( 8 , 9 ). First, the genetic instrument must be robustly associated with the exposure, known as the relevance assumption. Second, the genetic instrument should lack association with confounding factors influencing the relationship between the exposure and the outcome, known as the independence assumption. Third, the genetic instrument should not influence the outcome through alternative pathways unrelated to the exposure, termed the exclusion restriction assumption. The violation of the last assumption is known as horizontal pleiotropy. Recent expansive genome-wide association studies (GWAS) have identified optimal genetic instruments for circulating protein levels within the gene encoding the protein, termed cis -protein quantitative trait loci (cis-pQTL)( 10 – 12 ). Their proximity to genes encoding proteins make the cis -pQTL ideal MR instruments, by minimizing the possibility of horizontal pleiotropy. Prior MR studies have explored causal relationships between circulating protein levels and a variety of complex diseases and traits( 13 – 16 ). In this study, we undertook an integrative proteogenomic analysis to systematically identify potential biomarkers for pediatric BMI. First, using cis -pQTL as MR instruments for up to 535 proteins, we estimated the causal effect of genetically altered levels of these proteins on pediatric BMI among 39,620 children from a large European GWAS( 17 ). We then prioritized the candidate proteins through sensitivity analyses addressing potential horizontal pleiotropy and testing Bayesian colocalization and replication in independent cohorts. Furthermore, we identified target tissues through enrichment analyses. Methods Our study adheres to the MR-STROBE( 8 , 9 ) checklist (Supplementary Material) and did not require ethics approval. Study Exposures cis -pQTL GWAS The instrumental variables ( cis -pQTL) for circulating proteins were derived from three proteomic GWAS in adults of European descent( 18 ) and one GWAS in children of European descent 21 . Cis -pQTL were identified as single-nucleotide polymorphisms (SNPs) that were independently associated with the protein levels (p ≤ 5 x 10 − 8 ) and located within 1 Mb of the transcription start site of the protein coding gene. To satisfy the first MR assumption, we retained cis -pQTL with an F-statistic > 10 defining a strong MR instrument. The measurements of circulating proteins in the Ferkingstad et al.( 12 ) (N = 35,559), and Suhre et al.( 19 ) GWAS (N = 1000) were conducted using the SomaLogic platform, while the Folkersen et al( 20 ) GWAS (N = 21,758) and the Niu et al.( 21 ) pediatric GWAS (N = 2,147) employed the Olink platform (Supplementary Table 1). Study Outcomes BMI GWAS To evaluate the association between cis -pQTL and pediatric BMI (hereinafter referred to as BMI), we retrieved their effects from a large BMI GWAS by the Early Growth Genetics Consortium( 17 , 22 ) on 61,111 children aged from 2 to 10 years. For our MR study, we used data from the discovery GWAS meta-analysis involving 39,620 children of European descent (Supplementary Table 1)( 17 ). Statistical Analyses Mendelian randomization and sensitivity analyses We performed two-sample MR analyses implemented in the “TwoSampleMR” R package (version 0.5.8)( 23 ), using the Wald ratio to estimate the effect on BMI for the majority of proteins with a single cis -pQTL instrument. To compute the Wald ratios, SNP-exposure effects were used against SNP-outcome effects to compute a MR estimate reflecting the effect (beta) of one standard deviation increase in the level of each protein on standardized BMI values. The GWAS summary statistics for genetic instruments underwent harmonization, aligning them with alleles in the outcome GWAS inferred using allele frequency data. Palindromic variants with an intermediate minor allele frequency (MAF > 0.42) were excluded. The Benjamini-Hochberg method( 24 ) to compute false discovery rate (FDR) was used to control for multiple testing. MR effects exhibiting an FDR-corrected p-value < 0.05 (corresponding to a p-value of < 1.34 x 10 − 4 ) were considered significant. Colocalization analyses MR estimates might be confounded by linkage disequilibrium (LD), when the SNP-instruments are not causal for the outcome, but instead they are inherited in the same haplotype block (in LD) with a causal SNP. Using colocalization analysis implemented in the coloc R package( 25 ), we assessed the posterior probability of a genomic region containing a causal variant influencing both the candidate protein level and BMI, examining all SNPs with a minor allele frequency > 0.01 within 1 Mb of the cis -pQTLs of the candidate protein. Within the coloc package, we employed default priors of the 'coloc.abf' function, setting the prior probability of the exposure having a causal variant and the prior probability of the outcome having a causal variant at 1.0x10 − 4 , and the prior probability of the exposure and the outcome sharing the same causal variant at 1.0x10 − 5 . The results provided posterior probabilities for 4 different scenarios (H0: no association of the genomic locus with either trait; H1: association with BMI but not with the protein; H2: association with the protein but not with BMI; H3: association with BMI and the protein through two different causal SNPs and H4: association with BMI and the protein via one shared causal SNP). A colocalization probability (p4) > 75% was considered robust evidence of colocalization( 26 ). Testing for confounding and horizontal pleiotropy To test the second and third MR assumption, we investigated potential pleiotropic effects or associations with confounders of the cis -pQTL of our candidate proteins, by undertaking phenome-wide association studies (PheWAS) for these cis -pQTL in OpenTargets( 27 , 28 ) (retrieved July 08, 2024). Since BMI accounts for both fat and lean body mass, we undertook a two-step MR( 29 ) testing potential mediation by adiposity (% fat mass) of the effect of our candidate proteins on BMI. To do this, in the absence of child-specific GWAS on adiposity, we used summary level results from a GWAS for body fat percentage (%BF) in N = 155,961 adults of European descent by Hübel et al ( 30 ) (Supplementary Table 1). The two-step MR approach involved a first step, which examined the causal MR relationship between protein levels and %BF, while the second step assessed the MR association between %BF and the outcome, BMI. The indirect MR effect of the proteins on BMI, mediated through %BF, was estimated by calculating the product of the effect of proteins on %BF and the effect of %BF on BMI outcomes. The standard error of this indirect effect and its significance were determined using the Sobel test( 31 ) . Replication and reverse MR analyses To further validate our significant MR associations, we pursued two replication approaches. First, we conducted MR analyses utilizing independent cis-genetic instruments identified within the UK Biobank Pharma Proteomics Project (UKB-PPP) study( 32 ) (Sun et al GWAS). This study involved Olink-based measurements of 2,923 unique proteins( 32 ) in a cohort of 34,557 individuals of European ancestry. Additionally, we replicated MR analyses for the candidate proteins by utilizing a proteomic GWAS meta-analysis by Emilsson et al. (N = 3,200 European adults)( 33 ). While there was partial participant overlap between this study and the GWAS by Sun et al.( 32 ), protein measurements in the Emilsson et al. GWAS were conducted using the SomaLogic platform, enabling a significantly larger number (11,000) of protein measurements ( 34 ). In our replication MR studies, we also tested effects of the candidate proteins on adult BMI. To do this, we used data from the Yengo et al GWAS on N = 693,529 European adults( 18 ) (Supplementary Table 1). Finally, for proteins prioritized by our main MR analysis, we undertook a reverse MR to identify if BMI affects levels of the candidate circulating protein and not the opposite. For these analyses, we utilized genome-wide significant (GWAS p-value < 5 x 10 − 8 ) and independent SNPs as instruments for pediatric BMI from the Vogelezang et al ( 17 ) GWAS. Effects of these SNPs were retrieved from the Suhre et al., Folkersen et al., and Ferkingstad et al., proteomic GWAS. MR estimates were computed using the inverse variance weighted (IVW) method and three other pleiotropy-robust methods (MR-Egger, weighted median, and weighted mode)( 35 – 37 ) . MR Power Analysis We assessed the power of our main MR study to detect the identified protein effects on BMI based on the variance explained by the protein-increasing alleles of the cis-pQTLs, an alpha level of 1.34 x 10 − 4 , and the sample size of the pediatric BMI GWAS using the method described by Brion et al.( 38 ) Protein-Protein Interaction, Pathway Enrichment and tissue expression analyses To investigate the functions of the candidate proteins and explore their interactions, we utilized the Gene Multiple Association Network Integration Algorithm (GeneMANIA) tool( 39 , 40 ) For each identified protein, we conducted pathway and process enrichment analyses based on their corresponding genes using various gene ontology resources facilitated by Metascape ( https://metascape.org/ ). Additionally, we performed enrichment analysis via FUMA version 1.5.2 ( https://fuma.ctglab.nl ) ( 41 ). We investigated whether the genes encoding candidate proteins are targets for existing drugs within the OpenTargets repository ( https://www.opentargets.org ). To explore whether the cis -pQTL associated with the candidate protein biomarkers for pediatric BMI showed evidence of being expression quantitative trait loci (eQTLs), we generated a gene expression heatmap using Genotype-Tissue Expression (GTEx) v8 through FUMA( 42 ). Results Mendelian randomization An overview of the design of our MR study is depicted in Fig. 1 . Following identification of cis -pQTL instruments for 2,130 proteins, we performed data harmonization (Supplementary Table 3), which enabled MRs for 535 circulating proteins. All cis-pQTL had an F-statitic > 10. Results of these MR analyses are also provided in Supplementary Table 3. Three MR associations reached an FDR-corrected significant p-value below 0.05 (Fig. 2 a). These are the endoglin (ENG; MR beta: -0.07, 95% CI= [-0.10, -0.04], P = 4.4 x10 − 5 ), the fatty acid binding protein 4 (FABP4; MR beta: -0.33, 95% CI= [-0.5, -0.16], P = 1.3 x10 − 4 ), and the Nectin-like protein-2 (MR beta: -0.26, 95% CI= [-0.37, -0.15], P = 5.45 x10 − 5 ), all demonstrating a negative effect on standardized BMI per one standard deviation increase in their blood level. No significant MR associations were found using cis -pQTL from the pediatric proteomic GWAS( 33 ) (Supplementary Table 3). Sensitivity analyses testing the MR assumptions (confounding and horizontal pleiotropy assessment) Colocalization analyses for the three MR-prioritized proteins showed evidence of colocalization with H4 exhibiting the highest posterior probabilities for ENG (cis-pQTL: rs651007, H4 = 96%), followed by Nectin-like protein 2 (cis-pQTL: rs11215406, H4 = 95%) and FABP4 (cis-pQTL: rs77878271, H4 = 76%) (Fig. 3 , Supplementary Table 4). By undertaking PheWAS analyses in OpenTargets, we identified that the genetic instruments of ENG, FABP4, and Nectin-like protein 2 have been associated in GWAS to various traits linked to childhood obesity, including anthropometric, pubertal, sociodemographic, and nutritional traits (Fig. 2 b and Supplementary Tables 5–8). This may have introduced potential bias to our MR studies, due to these traits acting as confounders of the MR associations or introducing horizontal pleiotropy in the MR instruments of the three candidate proteins. Replication and Reverse MR analyses Using cis- genetic instruments identified within the UK Biobank Pharma Proteomics Project (UKB-PPP) study( 32 ) and the proteomic GWAS meta-analysis by Emilsson et al( 33 ) (Fig. 1 ), only the association for ENG replicated (Supplementary Table 9). Additionally, by testing effects of the candidate proteins on adult BMI, using data from the GWAS by Yengo et al ( 18 ) (Fig. 1 ), we observed a MR effect on adult BMI for Nectin-like protein 2, exhibiting a consistent direction and magnitude of effect as observed in our pediatric MR analysis (beta = -0.08, p = 1.973×10 − 9 ), while ENG obtained a suggestive p-value (p = 0.049). We could not replicate the result for FABP4 since the cis -pQTL for this protein (or a proxy) was not found in the adult BMI GWAS (Supplementary Table 9). We then conducted reverse MR studies, utilizing pediatric BMI as exposure and the candidate proteins as outcomes. Using 16 SNPs as instruments for BMI following clumping and data harmonization, our reverse MR analyses revealed evidence of a reverse association of BMI with FABP4 (MR beta 0,24 (95% CI [0.17, 0.33], p = 1.803 x 10 − 9 in the IVW method and MR beta 0,28 (95% [0.17, 0.40], p = 1.814x10 − 6 in the Weighted Median method). Conversely, the results for Nectin-like protein 2 in the reverse MR analyses were not significant. The MR-Egger intercept of these analyses confirmed the absence of horizontal pleiotropy. However, we note that we were unable to retrieve effects of the BMI SNPs in the GWAS for the ENG protein which prohibited undertaking reverse MR testing for this protein (Supplementary Table 10). Two-Step Network Mendelian Randomization Analysis Our two-step network MR analysis indicated that the effects of the candidate proteins on BMI were not mediated by %BF (Supplementary Table 11). Consequently, we concluded that the impact of these proteins on BMI is likely independent of their effects on adiposity. MR power analysis Our power analysis indicated that, at a FDR-adjusted alpha of 1.34 x 10 − 4 , we had over 80% power to detect the identified in our MR effect on BMI for two of the three prioritized proteins (ENG and Nectin-like protein 2). Additionally, our MR analysis using instruments from the sole childhood proteomic GWAS indicated a power of only 13% to detect the largest beta coefficient obtained in these MR results (Supplementary Table 2). Pathway and Enrichment analyses To delve deeper into the functionality of the three candidate proteins, we used them in subsequent pathway and enrichment analyses. Our GeneMANIA analysis revealed that the genes associated with the three candidate proteins exhibit physical interactions, co-expression, co-localization, or shared biological pathways (Fig. 4 , Supplementary Tables 12–14). Further exploration with Metascape highlighted associations of FABP4 and CADM1, the gene encoding the Nectin-like protein 2, with cancers (head and neck), while ENG was linked to Congenital malformations of the circulatory system. Using FUMA, we found that ENG is predominantly expressed in vascular walls, while FABP4 and CADM1 showed overexpression in the brain and adipose tissue (Fig. 5 , Supplementary Table 13). In the OpenTargets database (Supplementary Table 5), ENG emerged as a target for an existing drug, Carotuximab. Carotuximab, also known as TRC105, is a monoclonal antibody tested in clinical settings primarily for its anti-angiogenic properties in cancer treatment, underscoring a potential drug repurposing relevance of ENG( 43 ). Discussion Childhood obesity is an important public health problem with substantial implications for long-term health outcomes and impacts on healthcare systems worldwide. In this study, we sought to uncover biomarkers specific to pediatric obesity among circulating proteins using a comprehensive approach based on Mendelian randomization and supported by colocalization and a series of sensitivity analyses. Our findings identified three circulating proteins, ENG, FABP4, and Nectin-like protein 2, which exhibited protective effects on childhood obesity with Nectin-like protein 2 also demonstrating an effect on adult BMI in the same direction. By undertaking a reverse MR analysis, we identified a probable compensatory effect of pediatric obesity on levels of FABP4. Our two step MR highlighted that the impact of these proteins on BMI is likely not explained by direct effects on adiposity. Our PheWAS indicated possible pleiotropic bias in all three MR associations, therefore our results should be interpreted with caution. However, these findings contribute to the existing literature on the potential roles of ENG, FABP4, and Nectin-like protein 2 in metabolic regulation and adipose tissue function. Among the three prioritized proteins, ENG emerged as the protein with the strongest evidence of association with pediatric BMI. Also, ENG is a known drug target in humans (Supplementary Table 10). Endoglin (ENG) serves as a coreceptor for transforming growth factor (TGF)-β1 and TGF-β3, and it is highly expressed on cell membranes of endothelial cells (ECs) and syncytiotrophoblasts ( 44 ). Beyond its canonical role in angiogenesis and vascular development, ENG is expressed by various cell types, including endothelial cells, vascular smooth muscle cells, fibroblasts, hepatic stellate cells, and activated macrophages( 45 – 48 ). In our analysis, we found that ENG expression was enriched in vascular walls. While endothelial dysfunction associated with adult obesity contributes to the development of cardiovascular diseases such as hypertension, atherosclerosis, and metabolic syndrome, our reverse MR study did not provide evidence of effects of childhood BMI on ENG levels. However, evidence also suggests that endothelial dysfunction can precede and potentially contribute to the development of obesity, especially in children and adolescents( 49 ), which is in line with our findings. The above underscore the importance of further functional studies to elucidate the role of ENG in pediatric or adult obesity. FABP4, also known as Fatty Acid Binding Protein 4 or AP2 (adipocyte protein 2), is a circulating protein showing a protective effect against pediatric obesity in our MR analysis. Interestingly, our reverse MR analysis indicates a positive causal effect of pediatric BMI on FABP4 levels, which is in contrast with the direction of effect in our main MR findings. FABP4 is primarily expressed in adipocytes and macrophages, playing a significant role in lipid metabolism and inflammatory responses( 50 ). Its expression is often upregulated in individuals with obesity and correlates with insulin resistance and metabolic disorders, under the action of inflammatory cytokines( 51 ). When fabp4 deficient mice were placed on a high-fat, high-caloric diet, the total weight gain in these mice was higher than that in wildtype controls( 52 ). Another study showed that pharmacological inhibition of fabp4 in mouse models can prevent atherosclerosis and type 2 diabetes( 50 ). In our study, we observed overexpression of FABP4 in adipose tissue, indicating its potential involvement in central and peripheral mechanisms of energy balance regulation( 53 ). The upregulation of FABP4 levels in obesity, as indicated by our reverse MR analysis and previous literature( 54 ), suggests a possible compensatory effect (canalization) underlying the complex interplay between adipose tissue function and metabolic homeostasis. Nectin-like protein 2 is another circulating protein negatively associated with both pediatric and adult BMI in our MR analyses. As a cell adhesion molecule belonging to the Nectin family, it is involved in various cellular processes, including cell-cell adhesion, migration, and signaling and has been implicated in tumorigenesis in humans and mice ( 55 ). While its specific role in metabolic regulation remains elusive, Nectin-like protein 2 has been implicated in diverse physiological functions across different tissues and organs( 56 ). For instance, it plays a significant role in immune recognition by acting as a ligand for the immune receptor CRTAM, which is predominantly expressed on cytotoxic lymphocyte cells, implicating it in the intricate balance between cell adhesion and immune system interactions( 57 ). This dual functionality of Nectin-like protein 2 as a regulator of cellular processes and an immune regulator may indicate its broader role in metabolic homeostasis, potentially affecting pathways linked to pediatric obesity. Our study has several strengths. A principal advantage is our combined approach based on MR and colocalization, further complemented by replication MRs in independent cohorts, sensitivity analyses for pleiotropy, and pathway and enrichment analyses. These complementary approaches enhanced our ability to uncover causal associations by integrating diverse types of data and to elucidate shared biological pathways among the candidate proteins and increase the credibility of the identified causal associations. This replication across various protein measurement platforms and study populations underscores the reliability and generalizability of our findings. Furthermore, the roles of ENG, FABP4, and Nectin-like protein 2 in critical biological processes such as vascular function, lipid metabolism, immunity, and adipose tissue regulation highlight the complex, and to some extent distinct etiology of childhood obesity from that of obesity presenting later in life. Nevertheless, our study presents some limitations. Despite leveraging large-scale GWAS datasets and robust statistical methods, the inherent constraints of MR analysis, such as potential pleiotropy and unmeasured confounding, cannot be entirely excluded( 58 ), especially given the results of our PheWAS analysis. The broad PheWAS associations of the genetic instruments of the three proteins suggest that the results of our causal analysis should be interpreted with caution( 59 ). In this direction, variations in the levels of these proteins might affect BMI through associations with pubertal timing, lifestyle habits or the built environment rather than being themselves the underlying causes of childhood obesity. The two step MR showing absence of mediation by adiposity further implies the presence of pleiotropic pathways explaining the MR effects of these proteins on childhood BMI. Further complicating our analysis, the SNP instruments utilized were derived from adult GWAS data, inherently assuming that the genetic variants influencing protein levels in adults are applicable to the pediatric population. While we also used SNP -instruments from the only pediatric proteomic GWAS, the small sample size of this GWAS has likely precluded detection of positive associations for the tested proteins. Moreover, our analysis was restricted to European populations, limiting the generalizability of our findings to other demographic groups. In conclusion, our study identified three circulating proteins as potential causal biomarkers and drug targets for pediatric obesity using Mendelian randomization. The identification of ENG, FABP4, and Nectin-like protein 2 opens new avenues for research and clinical translation in the field of pediatric obesity. Future studies should focus on validating these findings in diverse pediatric populations, elucidating the underlying biological mechanisms of these molecules, and exploring gene-informed therapeutic interventions to mitigate the burden of pediatric obesity. Declarations Data availability All data used in this study are publicly available. The list of GWAS used and full summary statistics available are presented in the supplementary material (Supplementary Table 1). Code availability The R codes employed to generate the results of the MR and colocalization analyses are available on a GitHub repository ( https://github.com/Raphjacksun7/Proteome_MR_Analysis ) and on Zenodo( 60 ). Ethics declarations This study did not require ethics approval. Competing interests The authors declare no competing interests. D.M. is a Fonds de Recherche du Quebec-Santé (FRQS) Junior 1 Scholar and has received a Career Development Award from ENRICH (Empowering Next-Generation Researchers in Perinatal and Child Health). Author contributions D.M. conceived the study and supervised the analyses. R.A. drafted the manuscript and performed the analyses. All authors contributed in study design, reviewing and writing the manuscript. All authors critically reviewed and approved the final version of the manuscript. References Etelson D, Brand DA, Patrick PA, Shirali A. Childhood Obesity: Do Parents Recognize This Health Risk? Obes Res. 2003;11(11):1362–8. Dietz WH. Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics. 1998;101(3 Pt 2):518–25. Dehghan M, Akhtar-Danesh N, Merchant AT. Childhood obesity, prevalence and prevention. Nutr J. 2005;4(1):24. Daniels SR. The Consequences of Childhood Overweight and Obesity. Future Child. 2006;16(1):47–67. French SA, Story M, Perry CL. Self-Esteem and Obesity in Children and Adolescents: A Literature Review. Obes Res. 1995;3(5):479–90. Kavarian PN, Mosher TL, Abu El Haija M. Use of glucagon-like-peptide 1 receptor agonist in the treatment of childhood obesity. Curr Opin Pediatr. 2024;36(5):542–6. Yazdanpanah N, Yazdanpanah M, Wang Y, Forgetta V, Pollak M, Polychronakos C, et al. Clinically Relevant Circulating Protein Biomarkers for Type 1 Diabetes: Evidence From a Two-Sample Mendelian Randomization Study. Diabetes Care. 2022;45(1):169–77. Skrivankova VW, Richmond RC, Woolf BAR, Davies NM, Swanson SA, VanderWeele TJ, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ. 2021;375:n2233. Skrivankova VW, Richmond RC, Woolf BAR, Yarmolinsky J, Davies NM, Swanson SA, et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA. 2021;326(16):1614–21. Sun BB, Maranville JC, Peters JE, Stacey D, Staley JR, Blackshaw J, et al. Genomic atlas of the human plasma proteome. Nature. 2018;558(7708):73–9. Pietzner M, Wheeler E, Carrasco-Zanini J, Cortes A, Koprulu M, Wörheide MA, et al. Mapping the proteo-genomic convergence of human diseases. Science. 2021;374(6569):eabj1541. Ferkingstad E, Sulem P, Atlason BA, Sveinbjornsson G, Magnusson MI, Styrmisdottir EL, et al. Large-scale integration of the plasma proteome with genetics and disease. Nat Genet. 2021;53(12):1712–21. Yao C, Chen G, Song C, Keefe J, Mendelson M, Huan T, et al. Genome-wide mapping of plasma protein QTLs identifies putatively causal genes and pathways for cardiovascular disease. Nat Commun. 2018;9(1):3268. Chong M, Sjaarda J, Pigeyre M, Mohammadi-Shemirani P, Lali R, Shoamanesh A, et al. Novel Drug Targets for Ischemic Stroke Identified Through Mendelian Randomization Analysis of the Blood Proteome. Circulation. 2019;140(10):819–30. Lu T, Forgetta V, Greenwood CMT, Zhou S, Richards JB. Circulating Proteins Influencing Psychiatric Disease: A Mendelian Randomization Study. Biol Psychiatry. 2023;93(1):82–91. Yoshiji S, Butler-Laporte G, Lu T, Willett JDS, Su CY, Nakanishi T, et al. Proteome-wide Mendelian randomization implicates nephronectin as an actionable mediator of the effect of obesity on COVID-19 severity. Nat Metab. 2023;5(2):248–64. Vogelezang S, Bradfield JP, Ahluwalia TS, Curtin JA, Lakka TA, Grarup N, et al. Novel loci for childhood body mass index and shared heritability with adult cardiometabolic traits. PLoS Genet. 2020;16(10):e1008718. Yengo L, Sidorenko J, Kemper KE, Zheng Z, Wood AR, Weedon MN, et al. Meta-analysis of genome-wide association studies for height and body mass index in ∼700000 individuals of European ancestry. Hum Mol Genet. 2018;27(20):3641–9. Suhre K, Arnold M, Bhagwat AM, Cotton RJ, Engelke R, Raffler J, et al. Connecting genetic risk to disease end points through the human blood plasma proteome. Nat Commun. 2017;8:14357. Folkersen L, Gustafsson S, Wang Q, Hansen DH, Hedman ÅK, Schork A, et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat Metab. 2020;2(10):1135–48. Niu L, Stinson SE, Holm LA, Vonsild Lund MA, Fonvig CE, Cobuccio L, et al. Plasma Proteome Variation and its Genetic Determinants in Children and Adolescents [Internet]. 2023 [cited 2024 Jul 12]. Available from: http://medrxiv.org/lookup/doi/10.1101/2023.03.31.23287853 Warrington NM, Beaumont RN, Horikoshi M, Day FR, Helgeland Ø, Laurin C, et al. Maternal and fetal genetic effects on birth weight and their relevance to cardio-metabolic risk factors. Nat Genet. 2019;51(5):804–14. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7:e34408. Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Ser B Stat Methodol. 1995;57(1):289–300. Giambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet. 2014;10(5):e1004383. Foley CN, Staley JR, Breen PG, Sun BB, Kirk PDW, Burgess S, et al. A fast and efficient colocalization algorithm for identifying shared genetic risk factors across multiple traits. Nat Commun. 2021;12(1):764. Ghoussaini M, Mountjoy E, Carmona M, Peat G, Schmidt EM, Hercules A, et al. Open Targets Genetics: systematic identification of trait-associated genes using large-scale genetics and functional genomics. Nucleic Acids Res. 2021;49(D1):D1311–20. Ochoa D, Hercules A, Carmona M, Suveges D, Gonzalez-Uriarte A, Malangone C, et al. Open Targets Platform: supporting systematic drug–target identification and prioritisation. Nucleic Acids Res. 2021;49(D1):D1302–10. Burgess S, Daniel RM, Butterworth AS, Thompson SG, the EPIC-InterAct Consortium. Network Mendelian randomization: using genetic variants as instrumental variables to investigate mediation in causal pathways. Int J Epidemiol. 2015;44(2):484–95. Hübel C, Gaspar HA, Coleman JRI, Finucane H, Purves KL, Hanscombe KB, et al. Genomics of body fat percentage may contribute to sex bias in anorexia nervosa. Am J Med Genet Part B Neuropsychiatr Genet Off Publ Int Soc Psychiatr Genet. 2019;180(6):428–38. Sobel Test. In: The SAGE Encyclopedia of Communication Research Methods [Internet]. 2455 Teller Road, Thousand Oaks California 91320: SAGE Publications, Inc; 2017 [cited 2024 Dec 6]. Available from: https://methods.sagepub.com/reference/the-sage-encyclopedia-of-communication-research-methods/i13518.xml Sun BB, Chiou J, Traylor M, Benner C, Hsu YH, Richardson TG, et al. Plasma proteomic associations with genetics and health in the UK Biobank. Nature. 2023;622(7982):329–38. Emilsson V, Ilkov M, Lamb JR, Finkel N, Gudmundsson EF, Pitts R, et al. Co-regulatory networks of human serum proteins link genetics to disease. Science. 2018;361(6404):769–73. Gold L, Ayers D, Bertino J, Bock C, Bock A, Brody EN, et al. Aptamer-Based Multiplexed Proteomic Technology for Biomarker Discovery. Gelain F, editor. PLoS ONE. 2010;5(12):e15004. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512–25. Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304–14. Hartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46(6):1985–98. Brion MJA, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42(5):1497–501. Franz M, Rodriguez H, Lopes C, Zuberi K, Montojo J, Bader GD, et al. GeneMANIA update 2018. Nucleic Acids Res. 2018;46(W1):W60–4. Mostafavi S, Ray D, Warde-Farley D, Grouios C, Morris Q. GeneMANIA: a real-time multiple association network integration algorithm for predicting gene function. Genome Biol. 2008;9 Suppl 1(Suppl 1):S4. Fitzgerald J, Fahey L, Holleran L, Ó Broin P, Donohoe G, Morris DW. Thirteen Independent Genetic Loci Associated with Preserved Processing Speed in a Study of Cognitive Resilience in 330,097 Individuals in the UK Biobank. Genes. 2022;13(1):122. Watanabe K, Taskesen E, van Bochoven A, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;8(1):1826. K. Seon B, Haba A, Matsuno F, Takahashi N, Tsujie M, She X, et al. Endoglin-Targeted Cancer Therapy. Curr Drug Deliv. 2011;8(1):135–43. Possomato-Vieira JS, Khalil RA. Mechanisms of Endothelial Dysfunction in Hypertensive Pregnancy and Preeclampsia. In: Advances in Pharmacology [Internet]. Elsevier; 2016 [cited 2024 Jun 5]. p. 361–431. Available from: https://linkinghub.elsevier.com/retrieve/pii/S105435891630031X Bot PTG, Hoefer IE, Sluijter JPG, Van Vliet P, Smits AM, Lebrin F, et al. Increased Expression of the Transforming Growth Factor-β Signaling Pathway, Endoglin, and Early Growth Response-1 in Stable Plaques. Stroke. 2009;40(2):439–47. St-Jacques S, Forte M, Lye SJ, Letarte M. Localization of Endoglin, a Transforming Growth Factor-β Binding Protein, and of CD44 and Integrins in Placenta during the First Trimester of Pregnancy1. Biol Reprod. 1994;51(3):405–13. Meurer, Wimmer, Leur, Weiskirchen. Endoglin Trafficking/Exosomal Targeting in Liver Cells Depends on N-Glycosylation. Cells. 2019;8(9):997. Lastres P, Bellon T, Cabañas C, Sanchez-Madrid F, Acevedo A, Gougos A, et al. Regulated expression on human macrophages of endoglin, an Arg‐Gly‐Asp‐containing surface antigen. Eur J Immunol. 1992;22(2):393–7. Kajikawa M, Higashi Y. Obesity and Endothelial Function. Biomedicines. 2022;10(7):1745. Furuhashi M, Saitoh S, Shimamoto K, Miura T. Fatty Acid-Binding Protein 4 (FABP4): Pathophysiological Insights and Potent Clinical Biomarker of Metabolic and Cardiovascular Diseases. Clin Med Insights Cardiol. 2014;8s3:CMC.S17067. Hotamisligil GS, Johnson RS, Distel RJ, Ellis R, Papaioannou VE, Spiegelman BM. Uncoupling of Obesity from Insulin Resistance Through a Targeted Mutation in aP2 , the Adipocyte Fatty Acid Binding Protein. Science. 1996;274(5291):1377–9. Steen KA, Xu H, Bernlohr DA. FABP4/aP2 Regulates Macrophage Redox Signaling and Inflammasome Activation via Control of UCP2. Mol Cell Biol. 2017;37(2):e00282-16. Prentice KJ, Saksi J, Hotamisligil GS. Adipokine FABP4 integrates energy stores and counterregulatory metabolic responses. J Lipid Res. 2019;60(4):734–40. Queipo-Ortuño MI, Escoté X, Ceperuelo-Mallafré V, Garrido-Sanchez L, Miranda M, Clemente-Postigo M, et al. FABP4 dynamics in obesity: discrepancies in adipose tissue and liver expression regarding circulating plasma levels. PloS One. 2012;7(11):e48605. Ogita H, Rikitake Y, Miyoshi J, Takai Y. Cell adhesion molecules nectins and associating proteins: Implications for physiology and pathology. Proc Jpn Acad Ser B. 2010;86(6):621–9. Galibert L, Diemer GS, Liu Z, Johnson RS, Smith JL, Walzer T, et al. Nectin-like Protein 2 Defines a Subset of T-cell Zone Dendritic Cells and Is a Ligand for Class-I-restricted T-cell-associated Molecule. J Biol Chem. 2005;280(23):21955–64. Zhang S, Lu G, Qi J, Li Y, Zhang Z, Zhang B, et al. Competition of Cell Adhesion and Immune Recognition: Insights into the Interaction between CRTAM and Nectin-like 2. Structure. 2013;21(8):1430–9. Lawlor DA, Harbord RM, Sterne JAC, Timpson N, Davey Smith G. Mendelian randomization: Using genes as instruments for making causal inferences in epidemiology. Stat Med. 2008;27(8):1133–63. Dong SS, Zhang K, Guo Y, Ding JM, Rong Y, Feng JC, et al. Phenome-wide investigation of the causal associations between childhood BMI and adult trait outcomes: a two-sample Mendelian randomization study. Genome Med. 2021;13(1):48. Avocegamou R. Proteome_MR_Analysis [Internet]. Zenodo; 2025 [cited 2025 Mar 8]. Available from: https://zenodo.org/doi/10.5281/zenodo.14969499 Additional Declarations There is NO conflict of interest to disclose Supplementary Files STROBEMRchecklistfillable.docx STROBE-MR_checklist_filled SupTables.xlsx SUPPLEMENTARY MATERIAL Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6189469","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":435489158,"identity":"48bf0736-0707-45f5-a1df-4ec0d84eded3","order_by":0,"name":"Despoina Manousaki","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0002-4133-0618","institution":"University of Montreal","correspondingAuthor":true,"prefix":"","firstName":"Despoina","middleName":"","lastName":"Manousaki","suffix":""},{"id":435489159,"identity":"68183990-84ab-40ff-a9a7-6dc5ef5686c2","order_by":1,"name":"Raphael Avocegamou","email":"","orcid":"","institution":"University of Montreal","correspondingAuthor":false,"prefix":"","firstName":"Raphael","middleName":"","lastName":"Avocegamou","suffix":""},{"id":435489160,"identity":"cd2a724b-f4e5-45a5-9c6f-c38e35f57847","order_by":2,"name":"Basile Jumentier","email":"","orcid":"","institution":"Research Center of the Sainte-Justine University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Basile","middleName":"","lastName":"Jumentier","suffix":""},{"id":435489161,"identity":"a37fa413-b328-4759-87e1-56d03ee54a36","order_by":3,"name":"Kaossarath Fagbemi","email":"","orcid":"","institution":"Research Center of the Sainte-Justine University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kaossarath","middleName":"","lastName":"Fagbemi","suffix":""},{"id":435489162,"identity":"d27e172a-836d-414a-985c-75c17810b9a2","order_by":4,"name":"Isabel Gamache","email":"","orcid":"","institution":"Research Center of the Sainte-Justine University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Isabel","middleName":"","lastName":"Gamache","suffix":""}],"badges":[],"createdAt":"2025-03-09 15:35:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6189469/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6189469/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":81012485,"identity":"2fb72b31-eec7-4b74-90ff-d18976ba5673","added_by":"auto","created_at":"2025-04-21 08:28:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":494331,"visible":true,"origin":"","legend":"\u003cp\u003eOverview of study. Mendelian randomization (MR) was conducted using the Wald ratio to estimate the effect of each circulating protein on pediatric BMI. Colocalization analyses and evaluation of horizontal pleiotropy through annotation and phenome-wide association study (PheWAS) were conducted to verify MR assumptions. Significant associations were replicated using other proteomic studies and an adult BMI GWAS. Gene expression enrichment analyses were conducted to identify potential target tissues and cell types.\u003c/p\u003e","description":"","filename":"Figure11.png","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/b40059134d9310de11759224.png"},{"id":81013514,"identity":"47412f7c-170e-49ba-9747-b270735be266","added_by":"auto","created_at":"2025-04-21 08:36:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":104018,"visible":true,"origin":"","legend":"\u003cp\u003eSummary of the study’s results. \u003cstrong\u003ea.\u003c/strong\u003eForest plot illustratIing the MR associations between the three candidate proteins and pediatric BMI. The β coefficients indicate the impact of a 1 SD increase in the protein level, as determined by its genetic instrument, on standardized BMI. The color code indicates the GWAS source of the genetic instrument. \u003cstrong\u003eb.\u003c/strong\u003e Candidate protein prioritization. Candidate proteins are ordered by posterior probability of colocalization (Colocalization panel). A posterior probability of colocalization \u0026gt; 80% (indicated by ++) was considered strong evidence of colocalization, while a posterior probability of colocalization \u0026gt; 50% (indicated by +) was considered suggestive evidence of colocalization. Associations supported by strong or suggestive colocalization evidence were replicated using additional GWAS (Replication panel). The N of associations with confounding traits of the genetic instrument of each protein in the PheWAS analyses is depicted in the left panel (Horizontal Pleiotropy).\u003c/p\u003e","description":"","filename":"Figure21.png","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/c3a90b0fe7112bb52cf04cde.png"},{"id":81012493,"identity":"29079f2c-6c36-452b-8db9-b70d254203ef","added_by":"auto","created_at":"2025-04-21 08:28:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":415612,"visible":true,"origin":"","legend":"\u003cp\u003eColocalization of genetic associations with candidate proteins and the pediatric BMI. The lead \u003cem\u003ecis\u003c/em\u003e-genetic instruments are indicated. Genetic variants located in a ±500kb window centered around each genetic instrument are plotted with their significance in respective studies with the corresponding instrument. For each target protein, the posterior probability of colocalization (PP.H4) and the posterior probability of co-existence of two distinct causal variants (PP.H3) are indicated.\u003c/p\u003e","description":"","filename":"Figure31.png","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/3e495af7b077ce66ef599965.png"},{"id":81013516,"identity":"39e80e6f-e856-45c9-bae6-9309e72bd5c4","added_by":"auto","created_at":"2025-04-21 08:36:56","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":529613,"visible":true,"origin":"","legend":"\u003cp\u003eThe genes of the candidate proteins are depicted as larger inner circles, while genes identified through the GeneMANIA extension are shown as smaller outer circles. Interaction types are color-coded: red lines indicate Physical interactions, purple lines denote Co-expression, green lines represent Genetic Interactions, and blue lines signify Pathwayinvolvement. Node sizes, shown in dark gray, reflect the weights of these genes, with larger nodes indicating higher weights.\u003c/p\u003e","description":"","filename":"Figure41.png","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/d1c9505b165cf4d4727e39d4.png"},{"id":81013517,"identity":"83f74628-fe83-4fef-aca9-175981fae123","added_by":"auto","created_at":"2025-04-21 08:36:56","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":226320,"visible":true,"origin":"","legend":"\u003cp\u003eGraph \u003cstrong\u003ea\u003c/strong\u003e displays the normalized expression values (zero-mean normalization of log2-transformed expression), where darker red indicates higher relative gene expression within each label, and darker blue signifies lower expression. Graph \u003cstrong\u003eb\u003c/strong\u003e depicts the -log10 p-values for Differentially Expressed Genes (DEGs) in each dataset. 'Up-regulated' refers to over-expression, while 'Down-regulated' indicates under-expression.\u003c/p\u003e","description":"","filename":"Figure51.png","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/9bab29d32d184efd0c6f86fd.png"},{"id":82706623,"identity":"16d43fe2-8e41-434b-a4b1-16b2c7b497d3","added_by":"auto","created_at":"2025-05-14 10:38:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2558420,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/d3361221-5ea2-4dde-9c73-99ff482bf884.pdf"},{"id":81012487,"identity":"d4294b83-8a00-4a66-8106-44bc0a9a0942","added_by":"auto","created_at":"2025-04-21 08:28:56","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":42167,"visible":true,"origin":"","legend":"STROBE-MR_checklist_filled","description":"","filename":"STROBEMRchecklistfillable.docx","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/4b3ae750c0001b4eb879a7f9.docx"},{"id":81012489,"identity":"526fd09b-0a60-4edd-905f-8448c4d2ad30","added_by":"auto","created_at":"2025-04-21 08:28:56","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":211399,"visible":true,"origin":"","legend":"SUPPLEMENTARY MATERIAL","description":"","filename":"SupTables.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6189469/v1/5de8aa01ce0d2cd3c406ebbe.xlsx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e conflict of interest to disclose","formattedTitle":"Proteome-wide Mendelian randomization identifies circulating proteins causally associated with childhood obesity","fulltext":[{"header":"Introduction","content":"\u003cp\u003eChildhood obesity is a global health issue(\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) affecting almost one in five children and has been acknowledged as a serious public health concern due to its high morbidity rates(\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). In comparison to children with a normal weight, children with obesity, defined as having a body mass index (BMI) two standard deviations above the age and sex-adjusted median, have an increased likelihood of living with obesity in adulthood(\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) and are at a higher risk of developing long-term cardiometabolic, psychosocial and musculoskeletal complications (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe identification of early biomarkers for pediatric obesity is key for the development of age-specific screening tools or new therapies. With the advance of high throughput proteomics, circulating proteins represent a valuable source for biomarker discovery, because their circulating abundances are measurable and possibly modifiable. For instance, the Glucagon-like peptide-1 (GLP-1) is a peptide detected in both the intestines and the blood that has been identified as a therapeutic target for both adult and pediatric obesity(\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). However, measuring serum proteins is limited by prohibitive costs and the small sample sizes of the available pediatric cohorts. Moreover, observational proteomic studies are plagued by measurement errors, and bias due to unmeasured confounding and reverse causation(\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMendelian randomization (MR) is a method providing an instrumental variable framework to mitigate the above biases and infer causality(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). This approach utilizes genetic variants, randomly allocated at conception, as instruments for a biomarker to evaluate the causal impact of this biomarker on a disease or a trait. MR leans on three pivotal assumptions(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e). First, the genetic instrument must be robustly associated with the exposure, known as the relevance assumption. Second, the genetic instrument should lack association with confounding factors influencing the relationship between the exposure and the outcome, known as the independence assumption. Third, the genetic instrument should not influence the outcome through alternative pathways unrelated to the exposure, termed the exclusion restriction assumption. The violation of the last assumption is known as horizontal pleiotropy.\u003c/p\u003e \u003cp\u003eRecent expansive genome-wide association studies (GWAS) have identified optimal genetic instruments for circulating protein levels within the gene encoding the protein, termed \u003cem\u003ecis\u003c/em\u003e-protein quantitative trait loci (cis-pQTL)(\u003cspan additionalcitationids=\"CR11\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Their proximity to genes encoding proteins make the \u003cem\u003ecis\u003c/em\u003e-pQTL ideal MR instruments, by minimizing the possibility of horizontal pleiotropy. Prior MR studies have explored causal relationships between circulating protein levels and a variety of complex diseases and traits(\u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn this study, we undertook an integrative proteogenomic analysis to systematically identify potential biomarkers for pediatric BMI. First, using \u003cem\u003ecis\u003c/em\u003e-pQTL as MR instruments for up to 535 proteins, we estimated the causal effect of genetically altered levels of these proteins on pediatric BMI among 39,620 children from a large European GWAS(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). We then prioritized the candidate proteins through sensitivity analyses addressing potential horizontal pleiotropy and testing Bayesian colocalization and replication in independent cohorts. Furthermore, we identified target tissues through enrichment analyses.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eOur study adheres to the MR-STROBE(\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e) checklist (Supplementary Material) and did not require ethics approval.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy Exposures\u003c/h2\u003e \u003cp\u003e \u003cb\u003ecis\u003c/b\u003e \u003cb\u003e-pQTL GWAS\u003c/b\u003e \u003c/p\u003e \u003cp\u003eThe instrumental variables (\u003cem\u003ecis\u003c/em\u003e-pQTL) for circulating proteins were derived from three proteomic GWAS in adults of European descent(\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e) and one GWAS in children of European descent\u003csup\u003e21\u003c/sup\u003e. \u003cem\u003eCis\u003c/em\u003e-pQTL were identified as single-nucleotide polymorphisms (SNPs) that were independently associated with the protein levels (p\u0026thinsp;\u0026le;\u0026thinsp;5 x 10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e) and located within 1 Mb of the transcription start site of the protein coding gene. To satisfy the first MR assumption, we retained \u003cem\u003ecis\u003c/em\u003e-pQTL with an F-statistic\u0026thinsp;\u0026gt;\u0026thinsp;10 defining a strong MR instrument. The measurements of circulating proteins in the Ferkingstad et al.(\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e) (N\u0026thinsp;=\u0026thinsp;35,559), and Suhre et al.(\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e) GWAS (N\u0026thinsp;=\u0026thinsp;1000) were conducted using the SomaLogic platform, while the Folkersen et al(\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e) GWAS (N\u0026thinsp;=\u0026thinsp;21,758) and the Niu et al.(\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e) pediatric GWAS (N\u0026thinsp;=\u0026thinsp;2,147) employed the Olink platform (Supplementary Table\u0026nbsp;1).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStudy Outcomes\u003c/h3\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eBMI GWAS\u003c/h2\u003e \u003cp\u003eTo evaluate the association between \u003cem\u003ecis\u003c/em\u003e-pQTL and pediatric BMI (hereinafter referred to as BMI), we retrieved their effects from a large BMI GWAS by the Early Growth Genetics Consortium(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e) on 61,111 children aged from 2 to 10 years. For our MR study, we used data from the discovery GWAS meta-analysis involving 39,620 children of European descent (Supplementary Table\u0026nbsp;1)(\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eStatistical Analyses\u003c/h3\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003eMendelian randomization and sensitivity analyses\u003c/h2\u003e \u003cp\u003eWe performed two-sample MR analyses implemented in the \u0026ldquo;TwoSampleMR\u0026rdquo; R package (version 0.5.8)(\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e), using the Wald ratio to estimate the effect on BMI for the majority of proteins with a single \u003cem\u003ecis\u003c/em\u003e-pQTL instrument. To compute the Wald ratios, SNP-exposure effects were used against SNP-outcome effects to compute a MR estimate reflecting the effect (beta) of one standard deviation increase in the level of each protein on standardized BMI values. The GWAS summary statistics for genetic instruments underwent harmonization, aligning them with alleles in the outcome GWAS inferred using allele frequency data. Palindromic variants with an intermediate minor allele frequency (MAF\u0026thinsp;\u0026gt;\u0026thinsp;0.42) were excluded. The Benjamini-Hochberg method(\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e) to compute false discovery rate (FDR) was used to control for multiple testing. MR effects exhibiting an FDR-corrected p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 (corresponding to a p-value of \u0026lt;\u0026thinsp;1.34 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e) were considered significant.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eColocalization analyses\u003c/h2\u003e \u003cp\u003eMR estimates might be confounded by linkage disequilibrium (LD), when the SNP-instruments are not causal for the outcome, but instead they are inherited in the same haplotype block (in LD) with a causal SNP. Using colocalization analysis implemented in the coloc R package(\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e), we assessed the posterior probability of a genomic region containing a causal variant influencing both the candidate protein level and BMI, examining all SNPs with a minor allele frequency\u0026thinsp;\u0026gt;\u0026thinsp;0.01 within 1 Mb of the \u003cem\u003ecis\u003c/em\u003e-pQTLs of the candidate protein. Within the coloc package, we employed default priors of the 'coloc.abf' function, setting the prior probability of the exposure having a causal variant and the prior probability of the outcome having a causal variant at 1.0x10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e, and the prior probability of the exposure and the outcome sharing the same causal variant at 1.0x10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e. The results provided posterior probabilities for 4 different scenarios (H0: no association of the genomic locus with either trait; H1: association with BMI but not with the protein; H2: association with the protein but not with BMI; H3: association with BMI and the protein through two different causal SNPs and H4: association with BMI and the protein via one shared causal SNP). A colocalization probability (p4)\u0026thinsp;\u0026gt;\u0026thinsp;75% was considered robust evidence of colocalization(\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eTesting for confounding and horizontal pleiotropy\u003c/h3\u003e\n\u003cp\u003eTo test the second and third MR assumption, we investigated potential pleiotropic effects or associations with confounders of the \u003cem\u003ecis\u003c/em\u003e-pQTL of our candidate proteins, by undertaking phenome-wide association studies (PheWAS) for these \u003cem\u003ecis\u003c/em\u003e-pQTL in OpenTargets(\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e) (retrieved July 08, 2024). Since BMI accounts for both fat and lean body mass, we undertook a two-step MR(\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e) testing potential mediation by adiposity (% fat mass) of the effect of our candidate proteins on BMI. To do this, in the absence of child-specific GWAS on adiposity, we used summary level results from a GWAS for body fat percentage (%BF) in N\u0026thinsp;=\u0026thinsp;155,961 adults of European descent by H\u0026uuml;bel et al (\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e) (Supplementary Table\u0026nbsp;1). The two-step MR approach involved a first step, which examined the causal MR relationship between protein levels and %BF, while the second step assessed the MR association between %BF and the outcome, BMI. The indirect MR effect of the proteins on BMI, mediated through %BF, was estimated by calculating the product of the effect of proteins on %BF and the effect of %BF on BMI outcomes. The standard error of this indirect effect and its significance were determined using the Sobel test(\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e) .\u003c/p\u003e\n\u003ch3\u003eReplication and reverse MR analyses\u003c/h3\u003e\n\u003cp\u003eTo further validate our significant MR associations, we pursued two replication approaches. First, we conducted MR analyses utilizing independent cis-genetic instruments identified within the UK Biobank Pharma Proteomics Project (UKB-PPP) study(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) (Sun et al GWAS). This study involved Olink-based measurements of 2,923 unique proteins(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) in a cohort of 34,557 individuals of European ancestry. Additionally, we replicated MR analyses for the candidate proteins by utilizing a proteomic GWAS meta-analysis by Emilsson et al. (N\u0026thinsp;=\u0026thinsp;3,200 European adults)(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e). While there was partial participant overlap between this study and the GWAS by Sun et al.(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e), protein measurements in the Emilsson et al. GWAS were conducted using the SomaLogic platform, enabling a significantly larger number (11,000) of protein measurements (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). In our replication MR studies, we also tested effects of the candidate proteins on adult BMI. To do this, we used data from the Yengo et al GWAS on N\u0026thinsp;=\u0026thinsp;693,529 European adults(\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e) (Supplementary Table\u0026nbsp;1).\u003c/p\u003e \u003cp\u003eFinally, for proteins prioritized by our main MR analysis, we undertook a reverse MR to identify if BMI affects levels of the candidate circulating protein and not the opposite. For these analyses, we utilized genome-wide significant (GWAS p-value\u0026thinsp;\u0026lt;\u0026thinsp;5 x 10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e) and independent SNPs as instruments for pediatric BMI from the Vogelezang et al (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e) GWAS. Effects of these SNPs were retrieved from the Suhre et al., Folkersen et al., and Ferkingstad et al., proteomic GWAS. MR estimates were computed using the inverse variance weighted (IVW) method and three other pleiotropy-robust methods (MR-Egger, weighted median, and weighted mode)(\u003cspan additionalcitationids=\"CR36\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e) .\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eMR Power Analysis\u003c/h2\u003e \u003cp\u003eWe assessed the power of our main MR study to detect the identified protein effects on BMI based on the variance explained by the protein-increasing alleles of the cis-pQTLs, an alpha level of 1.34 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e, and the sample size of the pediatric BMI GWAS using the method described by Brion et al.(\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e)\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eProtein-Protein Interaction, Pathway Enrichment and tissue expression analyses\u003c/h2\u003e \u003cp\u003eTo investigate the functions of the candidate proteins and explore their interactions, we utilized the Gene Multiple Association Network Integration Algorithm (GeneMANIA) tool(\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e) For each identified protein, we conducted pathway and process enrichment analyses based on their corresponding genes using various gene ontology resources facilitated by Metascape (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://metascape.org/\u003c/span\u003e\u003cspan address=\"https://metascape.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Additionally, we performed enrichment analysis via FUMA version 1.5.2 (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://fuma.ctglab.nl\u003c/span\u003e\u003cspan address=\"https://fuma.ctglab.nl\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) (\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). We investigated whether the genes encoding candidate proteins are targets for existing drugs within the OpenTargets repository (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.opentargets.org\u003c/span\u003e\u003cspan address=\"https://www.opentargets.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). To explore whether the \u003cem\u003ecis\u003c/em\u003e-pQTL associated with the candidate protein biomarkers for pediatric BMI showed evidence of being expression quantitative trait loci (eQTLs), we generated a gene expression heatmap using Genotype-Tissue Expression (GTEx) v8 through FUMA(\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eMendelian randomization\u003c/h2\u003e \u003cp\u003eAn overview of the design of our MR study is depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Following identification of \u003cem\u003ecis\u003c/em\u003e-pQTL instruments for 2,130 proteins, we performed data harmonization (Supplementary Table\u0026nbsp;3), which enabled MRs for 535 circulating proteins. All cis-pQTL had an F-statitic\u0026thinsp;\u0026gt;\u0026thinsp;10. Results of these MR analyses are also provided in Supplementary Table\u0026nbsp;3. Three MR associations reached an FDR-corrected significant p-value below 0.05 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). These are the endoglin (ENG; MR beta: -0.07, 95% CI= [-0.10, -0.04], P\u0026thinsp;=\u0026thinsp;4.4 x10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e), the fatty acid binding protein 4 (FABP4; MR beta: -0.33, 95% CI= [-0.5, -0.16], P\u0026thinsp;=\u0026thinsp;1.3 x10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e), and the Nectin-like protein-2 (MR beta: -0.26, 95% CI= [-0.37, -0.15], P\u0026thinsp;=\u0026thinsp;5.45 x10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e), all demonstrating a negative effect on standardized BMI per one standard deviation increase in their blood level. No significant MR associations were found using \u003cem\u003ecis\u003c/em\u003e-pQTL from the pediatric proteomic GWAS(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e) (Supplementary Table\u0026nbsp;3).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003eSensitivity analyses testing the MR assumptions (confounding and horizontal pleiotropy assessment)\u003c/h2\u003e \u003cp\u003eColocalization analyses for the three MR-prioritized proteins showed evidence of colocalization with H4 exhibiting the highest posterior probabilities for ENG (cis-pQTL: rs651007, H4\u0026thinsp;=\u0026thinsp;96%), followed by Nectin-like protein 2 (cis-pQTL: rs11215406, H4\u0026thinsp;=\u0026thinsp;95%) and FABP4 (cis-pQTL: rs77878271, H4\u0026thinsp;=\u0026thinsp;76%) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Supplementary Table\u0026nbsp;4).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBy undertaking PheWAS analyses in OpenTargets, we identified that the genetic instruments of ENG, FABP4, and Nectin-like protein 2 have been associated in GWAS to various traits linked to childhood obesity, including anthropometric, pubertal, sociodemographic, and nutritional traits (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and Supplementary Tables\u0026nbsp;5\u0026ndash;8). This may have introduced potential bias to our MR studies, due to these traits acting as confounders of the MR associations or introducing horizontal pleiotropy in the MR instruments of the three candidate proteins.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eReplication and Reverse MR analyses\u003c/h2\u003e \u003cp\u003eUsing \u003cem\u003ecis-\u003c/em\u003egenetic instruments identified within the UK Biobank Pharma Proteomics Project (UKB-PPP) study(\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e) and the proteomic GWAS meta-analysis by Emilsson et al(\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), only the association for ENG replicated (Supplementary Table\u0026nbsp;9). Additionally, by testing effects of the candidate proteins on adult BMI, using data from the GWAS by Yengo et al (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), we observed a MR effect on adult BMI for Nectin-like protein 2, exhibiting a consistent direction and magnitude of effect as observed in our pediatric MR analysis (beta = -0.08, p\u0026thinsp;=\u0026thinsp;1.973\u0026times;10\u003csup\u003e\u0026minus;\u0026thinsp;9\u003c/sup\u003e), while ENG obtained a suggestive p-value (p\u0026thinsp;=\u0026thinsp;0.049). We could not replicate the result for FABP4 since the \u003cem\u003ecis\u003c/em\u003e-pQTL for this protein (or a proxy) was not found in the adult BMI GWAS (Supplementary Table\u0026nbsp;9).\u003c/p\u003e \u003cp\u003eWe then conducted reverse MR studies, utilizing pediatric BMI as exposure and the candidate proteins as outcomes. Using 16 SNPs as instruments for BMI following clumping and data harmonization, our reverse MR analyses revealed evidence of a reverse association of BMI with FABP4 (MR beta 0,24 (95% CI [0.17, 0.33], p\u0026thinsp;=\u0026thinsp;1.803 x 10\u003csup\u003e\u0026minus;\u0026thinsp;9\u003c/sup\u003e in the IVW method and MR beta 0,28 (95% [0.17, 0.40], p\u0026thinsp;=\u0026thinsp;1.814x10\u003csup\u003e\u0026minus;\u0026thinsp;6\u003c/sup\u003e in the Weighted Median method). Conversely, the results for Nectin-like protein 2 in the reverse MR analyses were not significant. The MR-Egger intercept of these analyses confirmed the absence of horizontal pleiotropy. However, we note that we were unable to retrieve effects of the BMI SNPs in the GWAS for the ENG protein which prohibited undertaking reverse MR testing for this protein (Supplementary Table\u0026nbsp;10).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section2\"\u003e \u003ch2\u003eTwo-Step Network Mendelian Randomization Analysis\u003c/h2\u003e \u003cp\u003eOur two-step network MR analysis indicated that the effects of the candidate proteins on BMI were not mediated by %BF (Supplementary Table\u0026nbsp;11). Consequently, we concluded that the impact of these proteins on BMI is likely independent of their effects on adiposity.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003eMR power analysis\u003c/h2\u003e \u003cp\u003eOur power analysis indicated that, at a FDR-adjusted alpha of 1.34 x 10\u003csup\u003e\u0026minus;\u0026thinsp;4\u003c/sup\u003e, we had over 80% power to detect the identified in our MR effect on BMI for two of the three prioritized proteins (ENG and Nectin-like protein 2). Additionally, our MR analysis using instruments from the sole childhood proteomic GWAS indicated a power of only 13% to detect the largest beta coefficient obtained in these MR results (Supplementary Table\u0026nbsp;2).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003ePathway and Enrichment analyses\u003c/h2\u003e \u003cp\u003eTo delve deeper into the functionality of the three candidate proteins, we used them in subsequent pathway and enrichment analyses. Our GeneMANIA analysis revealed that the genes associated with the three candidate proteins exhibit physical interactions, co-expression, co-localization, or shared biological pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e, Supplementary Tables\u0026nbsp;12\u0026ndash;14). Further exploration with Metascape highlighted associations of FABP4 and CADM1, the gene encoding the Nectin-like protein 2, with cancers (head and neck), while ENG was linked to Congenital malformations of the circulatory system. Using FUMA, we found that ENG is predominantly expressed in vascular walls, while FABP4 and CADM1 showed overexpression in the brain and adipose tissue (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e, Supplementary Table\u0026nbsp;13).\u003c/p\u003e\u003cp\u003eIn the OpenTargets database (Supplementary Table\u0026nbsp;5), ENG emerged as a target for an existing drug, Carotuximab. Carotuximab, also known as TRC105, is a monoclonal antibody tested in clinical settings primarily for its anti-angiogenic properties in cancer treatment, underscoring a potential drug repurposing relevance of ENG(\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eChildhood obesity is an important public health problem with substantial implications for long-term health outcomes and impacts on healthcare systems worldwide. In this study, we sought to uncover biomarkers specific to pediatric obesity among circulating proteins using a comprehensive approach based on Mendelian randomization and supported by colocalization and a series of sensitivity analyses. Our findings identified three circulating proteins, ENG, FABP4, and Nectin-like protein 2, which exhibited protective effects on childhood obesity with Nectin-like protein 2 also demonstrating an effect on adult BMI in the same direction. By undertaking a reverse MR analysis, we identified a probable compensatory effect of pediatric obesity on levels of FABP4. Our two step MR highlighted that the impact of these proteins on BMI is likely not explained by direct effects on adiposity. Our PheWAS indicated possible pleiotropic bias in all three MR associations, therefore our results should be interpreted with caution. However, these findings contribute to the existing literature on the potential roles of ENG, FABP4, and Nectin-like protein 2 in metabolic regulation and adipose tissue function.\u003c/p\u003e \u003cp\u003eAmong the three prioritized proteins, ENG emerged as the protein with the strongest evidence of association with pediatric BMI. Also, ENG is a known drug target in humans (Supplementary Table\u0026nbsp;10). Endoglin (ENG) serves as a coreceptor for transforming growth factor (TGF)-β1 and TGF-β3, and it is highly expressed on cell membranes of endothelial cells (ECs) and syncytiotrophoblasts (\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). Beyond its canonical role in angiogenesis and vascular development, \u003cem\u003eENG\u003c/em\u003e is expressed by various cell types, including endothelial cells, vascular smooth muscle cells, fibroblasts, hepatic stellate cells, and activated macrophages(\u003cspan additionalcitationids=\"CR46 CR47\" citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). In our analysis, we found that \u003cem\u003eENG\u003c/em\u003e expression was enriched in vascular walls. While endothelial dysfunction associated with adult obesity contributes to the development of cardiovascular diseases such as hypertension, atherosclerosis, and metabolic syndrome, our reverse MR study did not provide evidence of effects of childhood BMI on ENG levels. However, evidence also suggests that endothelial dysfunction can precede and potentially contribute to the development of obesity, especially in children and adolescents(\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e), which is in line with our findings. The above underscore the importance of further functional studies to elucidate the role of ENG in pediatric or adult obesity.\u003c/p\u003e \u003cp\u003eFABP4, also known as Fatty Acid Binding Protein 4 or AP2 (adipocyte protein 2), is a circulating protein showing a protective effect against pediatric obesity in our MR analysis.\u003c/p\u003e \u003cp\u003eInterestingly, our reverse MR analysis indicates a positive causal effect of pediatric BMI on FABP4 levels, which is in contrast with the direction of effect in our main MR findings. \u003cem\u003eFABP4\u003c/em\u003e is primarily expressed in adipocytes and macrophages, playing a significant role in lipid metabolism and inflammatory responses(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). Its expression is often upregulated in individuals with obesity and correlates with insulin resistance and metabolic disorders, under the action of inflammatory cytokines(\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e). When \u003cem\u003efabp4\u003c/em\u003e deficient mice were placed on a high-fat, high-caloric diet, the total weight gain in these mice was higher than that in wildtype controls(\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e). Another study showed that pharmacological inhibition of \u003cem\u003efabp4\u003c/em\u003e in mouse models can prevent atherosclerosis and type 2 diabetes(\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e). In our study, we observed overexpression of \u003cem\u003eFABP4\u003c/em\u003e in adipose tissue, indicating its potential involvement in central and peripheral mechanisms of energy balance regulation(\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e). The upregulation of FABP4 levels in obesity, as indicated by our reverse MR analysis and previous literature(\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e), suggests a possible compensatory effect (canalization) underlying the complex interplay between adipose tissue function and metabolic homeostasis.\u003c/p\u003e \u003cp\u003eNectin-like protein 2 is another circulating protein negatively associated with both pediatric and adult BMI in our MR analyses. As a cell adhesion molecule belonging to the Nectin family, it is involved in various cellular processes, including cell-cell adhesion, migration, and signaling and has been implicated in tumorigenesis in humans and mice (\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e). While its specific role in metabolic regulation remains elusive, Nectin-like protein 2 has been implicated in diverse physiological functions across different tissues and organs(\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). For instance, it plays a significant role in immune recognition by acting as a ligand for the immune receptor CRTAM, which is predominantly expressed on cytotoxic lymphocyte cells, implicating it in the intricate balance between cell adhesion and immune system interactions(\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). This dual functionality of Nectin-like protein 2 as a regulator of cellular processes and an immune regulator may indicate its broader role in metabolic homeostasis, potentially affecting pathways linked to pediatric obesity.\u003c/p\u003e \u003cp\u003eOur study has several strengths. A principal advantage is our combined approach based on MR and colocalization, further complemented by replication MRs in independent cohorts, sensitivity analyses for pleiotropy, and pathway and enrichment analyses. These complementary approaches enhanced our ability to uncover causal associations by integrating diverse types of data and to elucidate shared biological pathways among the candidate proteins and increase the credibility of the identified causal associations. This replication across various protein measurement platforms and study populations underscores the reliability and generalizability of our findings. Furthermore, the roles of ENG, FABP4, and Nectin-like protein 2 in critical biological processes such as vascular function, lipid metabolism, immunity, and adipose tissue regulation highlight the complex, and to some extent distinct etiology of childhood obesity from that of obesity presenting later in life.\u003c/p\u003e \u003cp\u003eNevertheless, our study presents some limitations. Despite leveraging large-scale GWAS datasets and robust statistical methods, the inherent constraints of MR analysis, such as potential pleiotropy and unmeasured confounding, cannot be entirely excluded(\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e), especially given the results of our PheWAS analysis. The broad PheWAS associations of the genetic instruments of the three proteins suggest that the results of our causal analysis should be interpreted with caution(\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e). In this direction, variations in the levels of these proteins might affect BMI through associations with pubertal timing, lifestyle habits or the built environment rather than being themselves the underlying causes of childhood obesity. The two step MR showing absence of mediation by adiposity further implies the presence of pleiotropic pathways explaining the MR effects of these proteins on childhood BMI. Further complicating our analysis, the SNP instruments utilized were derived from adult GWAS data, inherently assuming that the genetic variants influencing protein levels in adults are applicable to the pediatric population. While we also used SNP -instruments from the only pediatric proteomic GWAS, the small sample size of this GWAS has likely precluded detection of positive associations for the tested proteins. Moreover, our analysis was restricted to European populations, limiting the generalizability of our findings to other demographic groups.\u003c/p\u003e \u003cp\u003eIn conclusion, our study identified three circulating proteins as potential causal biomarkers and drug targets for pediatric obesity using Mendelian randomization. The identification of ENG, FABP4, and Nectin-like protein 2 opens new avenues for research and clinical translation in the field of pediatric obesity. Future studies should focus on validating these findings in diverse pediatric populations, elucidating the underlying biological mechanisms of these molecules, and exploring gene-informed therapeutic interventions to mitigate the burden of pediatric obesity.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003eData availability\u003c/h2\u003e \u003cp\u003eAll data used in this study are publicly available. The list of GWAS used and full summary statistics available are presented in the supplementary material (Supplementary Table\u0026nbsp;1).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section2\"\u003e \u003ch2\u003eCode availability\u003c/h2\u003e \u003cp\u003eThe R codes employed to generate the results of the MR and colocalization analyses are available on a GitHub repository (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/Raphjacksun7/Proteome_MR_Analysis\u003c/span\u003e\u003cspan address=\"https://github.com/Raphjacksun7/Proteome_MR_Analysis\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and on Zenodo(\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e\u003ch2\u003eEthics declarations\u003c/h2\u003e \u003cp\u003eThis study did not require ethics approval.\u003c/p\u003e\u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors declare no competing interests. D.M. is a Fonds de Recherche du Quebec-Sant\u0026eacute; (FRQS) Junior 1 Scholar and has received a Career Development Award from ENRICH (Empowering Next-Generation Researchers in Perinatal and Child Health).\u003c/p\u003e\u003ch2\u003eAuthor contributions\u003c/h2\u003e \u003cp\u003eD.M. conceived the study and supervised the analyses. R.A. drafted the manuscript and performed the analyses. All authors contributed in study design, reviewing and writing the manuscript. All authors critically reviewed and approved the final version of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eEtelson D, Brand DA, Patrick PA, Shirali A. Childhood Obesity: Do Parents Recognize This Health Risk? Obes Res. 2003;11(11):1362\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDietz WH. Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics. 1998;101(3 Pt 2):518\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDehghan M, Akhtar-Danesh N, Merchant AT. Childhood obesity, prevalence and prevention. Nutr J. 2005;4(1):24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaniels SR. The Consequences of Childhood Overweight and Obesity. Future Child. 2006;16(1):47\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFrench SA, Story M, Perry CL. Self-Esteem and Obesity in Children and Adolescents: A Literature Review. Obes Res. 1995;3(5):479\u0026ndash;90.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKavarian PN, Mosher TL, Abu El Haija M. Use of glucagon-like-peptide 1 receptor agonist in the treatment of childhood obesity. Curr Opin Pediatr. 2024;36(5):542\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYazdanpanah N, Yazdanpanah M, Wang Y, Forgetta V, Pollak M, Polychronakos C, et al. Clinically Relevant Circulating Protein Biomarkers for Type 1 Diabetes: Evidence From a Two-Sample Mendelian Randomization Study. Diabetes Care. 2022;45(1):169\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSkrivankova VW, Richmond RC, Woolf BAR, Davies NM, Swanson SA, VanderWeele TJ, et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): explanation and elaboration. BMJ. 2021;375:n2233.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSkrivankova VW, Richmond RC, Woolf BAR, Yarmolinsky J, Davies NM, Swanson SA, et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA. 2021;326(16):1614\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun BB, Maranville JC, Peters JE, Stacey D, Staley JR, Blackshaw J, et al. Genomic atlas of the human plasma proteome. Nature. 2018;558(7708):73\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePietzner M, Wheeler E, Carrasco-Zanini J, Cortes A, Koprulu M, W\u0026ouml;rheide MA, et al. Mapping the proteo-genomic convergence of human diseases. Science. 2021;374(6569):eabj1541.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFerkingstad E, Sulem P, Atlason BA, Sveinbjornsson G, Magnusson MI, Styrmisdottir EL, et al. Large-scale integration of the plasma proteome with genetics and disease. Nat Genet. 2021;53(12):1712\u0026ndash;21.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYao C, Chen G, Song C, Keefe J, Mendelson M, Huan T, et al. Genome-wide mapping of plasma protein QTLs identifies putatively causal genes and pathways for cardiovascular disease. Nat Commun. 2018;9(1):3268.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChong M, Sjaarda J, Pigeyre M, Mohammadi-Shemirani P, Lali R, Shoamanesh A, et al. Novel Drug Targets for Ischemic Stroke Identified Through Mendelian Randomization Analysis of the Blood Proteome. Circulation. 2019;140(10):819\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLu T, Forgetta V, Greenwood CMT, Zhou S, Richards JB. Circulating Proteins Influencing Psychiatric Disease: A Mendelian Randomization Study. Biol Psychiatry. 2023;93(1):82\u0026ndash;91.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoshiji S, Butler-Laporte G, Lu T, Willett JDS, Su CY, Nakanishi T, et al. Proteome-wide Mendelian randomization implicates nephronectin as an actionable mediator of the effect of obesity on COVID-19 severity. Nat Metab. 2023;5(2):248\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVogelezang S, Bradfield JP, Ahluwalia TS, Curtin JA, Lakka TA, Grarup N, et al. Novel loci for childhood body mass index and shared heritability with adult cardiometabolic traits. PLoS Genet. 2020;16(10):e1008718.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYengo L, Sidorenko J, Kemper KE, Zheng Z, Wood AR, Weedon MN, et al. Meta-analysis of genome-wide association studies for height and body mass index in \u0026sim;700000 individuals of European ancestry. Hum Mol Genet. 2018;27(20):3641\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuhre K, Arnold M, Bhagwat AM, Cotton RJ, Engelke R, Raffler J, et al. Connecting genetic risk to disease end points through the human blood plasma proteome. Nat Commun. 2017;8:14357.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFolkersen L, Gustafsson S, Wang Q, Hansen DH, Hedman \u0026Aring;K, Schork A, et al. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat Metab. 2020;2(10):1135\u0026ndash;48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNiu L, Stinson SE, Holm LA, Vonsild Lund MA, Fonvig CE, Cobuccio L, et al. Plasma Proteome Variation and its Genetic Determinants in Children and Adolescents [Internet]. 2023 [cited 2024 Jul 12]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://medrxiv.org/lookup/doi/10.1101/2023.03.31.23287853\u003c/span\u003e\u003cspan address=\"http://medrxiv.lookup/doi/10.1101/2023.03.31.23287853\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWarrington NM, Beaumont RN, Horikoshi M, Day FR, Helgeland \u0026Oslash;, Laurin C, et al. Maternal and fetal genetic effects on birth weight and their relevance to cardio-metabolic risk factors. Nat Genet. 2019;51(5):804\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7:e34408.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBenjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Ser B Stat Methodol. 1995;57(1):289\u0026ndash;300.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiambartolomei C, Vukcevic D, Schadt EE, Franke L, Hingorani AD, Wallace C, et al. Bayesian test for colocalisation between pairs of genetic association studies using summary statistics. PLoS Genet. 2014;10(5):e1004383.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFoley CN, Staley JR, Breen PG, Sun BB, Kirk PDW, Burgess S, et al. A fast and efficient colocalization algorithm for identifying shared genetic risk factors across multiple traits. Nat Commun. 2021;12(1):764.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGhoussaini M, Mountjoy E, Carmona M, Peat G, Schmidt EM, Hercules A, et al. Open Targets Genetics: systematic identification of trait-associated genes using large-scale genetics and functional genomics. Nucleic Acids Res. 2021;49(D1):D1311\u0026ndash;20.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOchoa D, Hercules A, Carmona M, Suveges D, Gonzalez-Uriarte A, Malangone C, et al. Open Targets Platform: supporting systematic drug\u0026ndash;target identification and prioritisation. Nucleic Acids Res. 2021;49(D1):D1302\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBurgess S, Daniel RM, Butterworth AS, Thompson SG, the EPIC-InterAct Consortium. Network Mendelian randomization: using genetic variants as instrumental variables to investigate mediation in causal pathways. Int J Epidemiol. 2015;44(2):484\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH\u0026uuml;bel C, Gaspar HA, Coleman JRI, Finucane H, Purves KL, Hanscombe KB, et al. Genomics of body fat percentage may contribute to sex bias in anorexia nervosa. Am J Med Genet Part B Neuropsychiatr Genet Off Publ Int Soc Psychiatr Genet. 2019;180(6):428\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSobel Test. In: The SAGE Encyclopedia of Communication Research Methods [Internet]. 2455 Teller Road, Thousand Oaks California 91320: SAGE Publications, Inc; 2017 [cited 2024 Dec 6]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://methods.sagepub.com/reference/the-sage-encyclopedia-of-communication-research-methods/i13518.xml\u003c/span\u003e\u003cspan address=\"https://methods.sagepub.com/reference/the-sage-encyclopedia-of-communication-research-methods/i13518.xml\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun BB, Chiou J, Traylor M, Benner C, Hsu YH, Richardson TG, et al. Plasma proteomic associations with genetics and health in the UK Biobank. Nature. 2023;622(7982):329\u0026ndash;38.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEmilsson V, Ilkov M, Lamb JR, Finkel N, Gudmundsson EF, Pitts R, et al. Co-regulatory networks of human serum proteins link genetics to disease. Science. 2018;361(6404):769\u0026ndash;73.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGold L, Ayers D, Bertino J, Bock C, Bock A, Brody EN, et al. Aptamer-Based Multiplexed Proteomic Technology for Biomarker Discovery. Gelain F, editor. PLoS ONE. 2010;5(12):e15004.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBowden J, Davey Smith G, Haycock PC, Burgess S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHartwig FP, Davey Smith G, Bowden J. Robust inference in summary data Mendelian randomization via the zero modal pleiotropy assumption. Int J Epidemiol. 2017;46(6):1985\u0026ndash;98.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrion MJA, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42(5):1497\u0026ndash;501.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFranz M, Rodriguez H, Lopes C, Zuberi K, Montojo J, Bader GD, et al. GeneMANIA update 2018. Nucleic Acids Res. 2018;46(W1):W60\u0026ndash;4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMostafavi S, Ray D, Warde-Farley D, Grouios C, Morris Q. GeneMANIA: a real-time multiple association network integration algorithm for predicting gene function. Genome Biol. 2008;9 Suppl 1(Suppl 1):S4.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFitzgerald J, Fahey L, Holleran L, \u0026Oacute; Broin P, Donohoe G, Morris DW. Thirteen Independent Genetic Loci Associated with Preserved Processing Speed in a Study of Cognitive Resilience in 330,097 Individuals in the UK Biobank. Genes. 2022;13(1):122.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWatanabe K, Taskesen E, van Bochoven A, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;8(1):1826.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK. Seon B, Haba A, Matsuno F, Takahashi N, Tsujie M, She X, et al. Endoglin-Targeted Cancer Therapy. Curr Drug Deliv. 2011;8(1):135\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePossomato-Vieira JS, Khalil RA. Mechanisms of Endothelial Dysfunction in Hypertensive Pregnancy and Preeclampsia. In: Advances in Pharmacology [Internet]. Elsevier; 2016 [cited 2024 Jun 5]. p. 361\u0026ndash;431. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://linkinghub.elsevier.com/retrieve/pii/S105435891630031X\u003c/span\u003e\u003cspan address=\"https://linkinghub.elsevier.com/retrieve/pii/S105435891630031X\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBot PTG, Hoefer IE, Sluijter JPG, Van Vliet P, Smits AM, Lebrin F, et al. Increased Expression of the Transforming Growth Factor-β Signaling Pathway, Endoglin, and Early Growth Response-1 in Stable Plaques. Stroke. 2009;40(2):439\u0026ndash;47.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSt-Jacques S, Forte M, Lye SJ, Letarte M. Localization of Endoglin, a Transforming Growth Factor-β Binding Protein, and of CD44 and Integrins in Placenta during the First Trimester of Pregnancy1. Biol Reprod. 1994;51(3):405\u0026ndash;13.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMeurer, Wimmer, Leur, Weiskirchen. Endoglin Trafficking/Exosomal Targeting in Liver Cells Depends on N-Glycosylation. Cells. 2019;8(9):997.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLastres P, Bellon T, Caba\u0026ntilde;as C, Sanchez-Madrid F, Acevedo A, Gougos A, et al. Regulated expression on human macrophages of endoglin, an Arg‐Gly‐Asp‐containing surface antigen. Eur J Immunol. 1992;22(2):393\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKajikawa M, Higashi Y. Obesity and Endothelial Function. Biomedicines. 2022;10(7):1745.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFuruhashi M, Saitoh S, Shimamoto K, Miura T. Fatty Acid-Binding Protein 4 (FABP4): Pathophysiological Insights and Potent Clinical Biomarker of Metabolic and Cardiovascular Diseases. Clin Med Insights Cardiol. 2014;8s3:CMC.S17067.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHotamisligil GS, Johnson RS, Distel RJ, Ellis R, Papaioannou VE, Spiegelman BM. Uncoupling of Obesity from Insulin Resistance Through a Targeted Mutation in \u003cem\u003eaP2\u003c/em\u003e, the Adipocyte Fatty Acid Binding Protein. Science. 1996;274(5291):1377\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSteen KA, Xu H, Bernlohr DA. FABP4/aP2 Regulates Macrophage Redox Signaling and Inflammasome Activation via Control of UCP2. Mol Cell Biol. 2017;37(2):e00282-16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePrentice KJ, Saksi J, Hotamisligil GS. Adipokine FABP4 integrates energy stores and counterregulatory metabolic responses. J Lipid Res. 2019;60(4):734\u0026ndash;40.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQueipo-Ortu\u0026ntilde;o MI, Escot\u0026eacute; X, Ceperuelo-Mallafr\u0026eacute; V, Garrido-Sanchez L, Miranda M, Clemente-Postigo M, et al. FABP4 dynamics in obesity: discrepancies in adipose tissue and liver expression regarding circulating plasma levels. PloS One. 2012;7(11):e48605.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOgita H, Rikitake Y, Miyoshi J, Takai Y. Cell adhesion molecules nectins and associating proteins: Implications for physiology and pathology. Proc Jpn Acad Ser B. 2010;86(6):621\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGalibert L, Diemer GS, Liu Z, Johnson RS, Smith JL, Walzer T, et al. Nectin-like Protein 2 Defines a Subset of T-cell Zone Dendritic Cells and Is a Ligand for Class-I-restricted T-cell-associated Molecule. J Biol Chem. 2005;280(23):21955\u0026ndash;64.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang S, Lu G, Qi J, Li Y, Zhang Z, Zhang B, et al. Competition of Cell Adhesion and Immune Recognition: Insights into the Interaction between CRTAM and Nectin-like 2. Structure. 2013;21(8):1430\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLawlor DA, Harbord RM, Sterne JAC, Timpson N, Davey Smith G. Mendelian randomization: Using genes as instruments for making causal inferences in epidemiology. Stat Med. 2008;27(8):1133\u0026ndash;63.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDong SS, Zhang K, Guo Y, Ding JM, Rong Y, Feng JC, et al. Phenome-wide investigation of the causal associations between childhood BMI and adult trait outcomes: a two-sample Mendelian randomization study. Genome Med. 2021;13(1):48.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAvocegamou R. Proteome_MR_Analysis [Internet]. Zenodo; 2025 [cited 2025 Mar 8]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://zenodo.org/doi/10.5281/zenodo.14969499\u003c/span\u003e\u003cspan address=\"https://zenodo.doi/10.5281/zenodo.14969499\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"proteins, pediatric obesity, Mendelian randomization, causal inference","lastPublishedDoi":"10.21203/rs.3.rs-6189469/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6189469/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground\u003c/b\u003e\u003c/p\u003e \u003cp\u003eChildhood obesity is a major public health problem, affecting one in 5 youths. We aimed to characterize biomarkers for pediatric obesity among circulating proteins using Mendelian randomization (MR).\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods\u003c/b\u003e\u003c/p\u003e \u003cp\u003eWe utilized genome-wide significant \u003cem\u003ecis-\u003c/em\u003e protein quantitative trait loci (pQTL) from large proteomic GWAS as instruments for circulating protein levels. \u003cem\u003eCis\u003c/em\u003e-pQTL effects on childhood body mass index (BMI) were retrieved from a European GWAS of 39,620 children. MR Wald ratios were calculated to estimate causal effects of each protein on childhood BMI. Sensitivity analyses, including colocalization and phenome-wide association studies (PheWAS), were performed for the candidate proteins to test for violation of the MR assumptions. Replication analyses were conducted using independent GWAS datasets, complemented by reverse MR and tissue enrichment analyses.\u003c/p\u003e\u003cp\u003e\u003cb\u003eFindings\u003c/b\u003e\u003c/p\u003e \u003cp\u003eAmong 535 tested proteins, three colocalized and demonstrated decreasing effects on BMI: endoglin (ENG; MR beta: -0.07, 95% CI [-0.10, -0.04], P\u0026thinsp;=\u0026thinsp;4.4\u0026times;10⁻⁵), fatty acid binding protein 4 (FABP4; MR beta: -0.33, 95% CI [-0.50, -0.16], P\u0026thinsp;=\u0026thinsp;1.3\u0026times;10⁻⁴), and Nectin-like protein-2 (MR beta: -0.26, 95% CI [-0.37, -0.15], P\u0026thinsp;=\u0026thinsp;5.45\u0026times;10⁻⁵). Reverse causation was identified for FABP4, suggesting a compensatory mechanism.\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusion\u003c/b\u003e\u003c/p\u003e \u003cp\u003eWe identified ENG, FABP4, and Nectin-like protein-2 as potential causal blood biomarkers or drug targets for pediatric obesity.\u003c/p\u003e","manuscriptTitle":"Proteome-wide Mendelian randomization identifies circulating proteins causally associated with childhood obesity","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-21 08:28:51","doi":"10.21203/rs.3.rs-6189469/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6dff7ac3-7a02-4fe6-8751-bc5201a4d0ae","owner":[],"postedDate":"April 21st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":46375102,"name":"Health sciences/Diseases/Endocrine system and metabolic diseases/Obesity"},{"id":46375103,"name":"Health sciences/Endocrinology/Endocrine system and metabolic diseases/Obesity"}],"tags":[],"updatedAt":"2025-05-14T10:30:17+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-21 08:28:51","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6189469","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6189469","identity":"rs-6189469","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}