The mediating role of the genetically predicted N6, N6, N6-trimethyllysine levels in the association between HLA DR on CD14‐ CD16+ monocytes and ankylosing spondylitis Authors: Danyun Hua, Lu Wang, Na Li, Xiang Xu, Xiaohu Yin Published: 2024 ABSTRACT This study explores the hidden connection between HLA DR on CD14‐ CD16+ monocytes and ankylosing spondylitis (AS), with a particular emphasis on investigating and measuring the impact of 1091 blood metabolites as potential mediators. We harnessed the power of summary-level data extracted from a comprehensive genome-wide association study to delve into the intricate relationship between genetically predicted HLA DR on CD14‐ CD16+ monocytes (3621 cases) and AS (1193 cases and 374,621 controls). Furthermore, we employed a two-step Mendelian randomization (MR) methodology to elucidate the extent to which blood metabolites contribute to the effects observed in CD14‐ CD16+ monocytes, ultimately influencing the development of AS. This methodological approach provides a comprehensive and rigorous exploration of the interplay between blood metabolites and AS, shedding light on the underlying mechanisms governing this intricate association. Through MR analysis, our investigation revealed an increase in HLA DR on CD14‐ CD16+ monocytes within plasma, which correspondingly led to a reduction in the incidence of AS. The primary MR analysis yielded an odds ratio of 0.64 with a 95% confidence interval spanning from 0.53 to 0.78, underscoring the protective effect of elevated HLA DR on CD14‐ CD16+ monocytes against the development of AS. Furthermore, our study found no compelling evidence to suggest that AS exerts any discernible influence on HLA DR on CD14‐ CD16+ monocytes. Instead, our investigation identified N6, N6, N6-trimethyllysine levels (TML), a blood metabolite, as the sole mediator in the relationship between HLA DR on CD14‐ CD16+ monocytes and AS. Notably, the genetic prediction of AS mediated by TML accounted for a substantial ‐2.98% proportion of the observed variance. Our investigation has delineated a causal association between HLA DR on CD14‐ CD16+ monocytes and AS. Specifically, HLA DR on CD14‐ CD16+ monocytes exhibited a protective effect against the development of AS. Conversely, AS mediated by TML emerged as a risk factor, though the precise impact of HLA DR on CD14‐ CD16+ monocytes on AS pathogenesis remains enigmatic. It is imperative to embark on further investigations into potential mediators. In a clinical setting, it is imperative to carefully monitor the patient’s HLA DR on CD14‐ CD16+ monocytes levels. 1. Introduction Ankylosing spondylitis (AS) is a chronic inflammatory rheumatic disorder[1] which primarily targets the spine and sacroiliac joints,[2] often culminating in persistent lower back discomfort and a notable diminution of spinal mobility.[3] It is estimated that severe debilitation afflicts approximately one-third of those diagnosed with AS.[4] Epidemiological studies indicate that this condition impacts roughly 1 in every 200 adults in Europe.[5]The therapeutic landscape for systemic symptoms of AS includes nonsteroidal anti-inflammatory drugs, azathioprine, methotrexate, and biologic disease-modifying anti-rheumatic drugs, all of which have demonstrated efficacy in mitigating systemic symptoms of the disease.[6] The impact of AS on patient quality of life and mental health is substantial; however, timely diagnosis and intervention can significantly improve the disease. The precise etiology and pathogenesis of AS remain elusive,[7] with current hypotheses suggesting a multifactorial origin involving genetic predispositions, infectious agents, and immune system dysregulation. Recent research has increasingly focused on bone immunology as a pivotal aspect of AS. Studies have revealed intricate interactions between bone cells, such as mesenchymal stem cells, osteoblasts, and osteoclasts, and immune cells, including T cells and B cells, which are believed to play a significant role in the pathogenesis of AS.[8] Monocytes, particularly critical in the context of inflammatory joint conditions, have garnered notable attention.[9] MR analysis suggests that CD14-CD16+ mononuclear cells may contribute to the pathogenesis of AS.[10] However, the specific mechanisms by which HLA DR on CD14‐ CD16+ monocytes is implicated in AS remain unclear and warrant further investigation. Metabolites, the intricate tapestry of small molecules that act as intermediates or final products in the myriad pathways of metabolism, are influenced by a constellation of factors, ranging from genetic makeup to lifestyle choices and the presence of disease.[11] In the wake of technological progress, the detection of a vast spectrum of these small molecule metabolites within bodily fluids and tissues has unfurled new horizons for early diagnosis, ongoing monitoring, and a deeper comprehension of disease pathogenesis.[12] Metabolomics has become an increasingly valuable tool in the field of immunological research. For instance, in the early detection of rheumatoid arthritis (RA), the employment of autoantibodies against citrullinated proteins and rheumatoid factors stands out for their exceptional sensitivity and specificity. Consequently, the identification of 1091 blood metabolites[13] offers promising insights into their potential roles as mediators in immune cell function and disease processes. Mendelian randomization (MR) is a methodological approach in genetic epidemiology designed to deduce potential causal relationships from observed associations. Adhering to the principle of random and immutable allele assignment at conception, MR parallels the structure of traditional randomized controlled trials. This methodology possesses the distinctive advantage of mitigating potential confounders, thereby enhancing the robustness of causal inferences. Utilizing MR, we aimed to investigate the existence of a causal link between HLA DR on CD14‐ CD16+ monocytes and AS. Additionally, this study sought to identify specific metabolites that may mediate the effects of HLA DR on CD14‐ CD16+ monocytes in the context of AS. 2. Methods 2.1. Study design Our analysis utilized publicly accessible data that received approval from the institutional review boards of the respective studies, obviating the need for additional ethical sanctions. The comprehensive results of this research, including supplementary materials, are detailed in this article and its supplements. This investigation employed a two-sample, bidirectional MR approach to examine the bidirectional causal relationship between HLA-DR expression on CD14‐ CD16+ monocytes and AS. Within this framework, we adeptly designated single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) for our rigorous analysis.[14] In our MR study, outlined in Figure 1, we utilized genetic variants to investigate the causal effect of HLA DR on CD14‐ CD16+ monocytes and AS. A comprehensive PhenoScanner search was implemented to evaluate all known phenotypes associated with the genetic instruments employed in our analysis. This process was integral in determining what we have termed the “total effect” of these genetic variants. Diagrams illustrating associations examined in this study. (A and B) The total effect between HLA DR on CD14‐ CD16+ monocytes and AS. (A) Is the total effect using genetically predicted HLA DR on CD14‐ CD16+ monocytes as exposure and AS as outcome. (B) Is the total effect using genetically predicted AS as exposure and HLA DR on CD14‐ CD16+ monocytes as outcome. The total effect was decomposed into: (i) indirect effect using a two-step approach (where c is the total effect of HLA DR on CD14‐ CD16+ monocytes on 1091 blood metabolites, and d is the effect of 1091 blood metabolites on AS) and the product method (c × d) and (ii) direct effect (a′ = a ‐ c × d). Proportion mediated was the indirect effect divided by the total effect. AS = ankylosing spondylitis. 2.2. GWAS summary data sources This investigation draws upon a spectrum of publicly available data, with a specific focus on individuals of European descent as participants in the Genome-Wide Association Studies (GWAS). The genetic insights into HLA DR on CD14‐ CD16+ monocytes are informed by the pioneering work of Yiheng Chen et al. Complementarily, genetic data relevant to AS were acquired from a Finnish database. Additionally, this study delves into an array of 1091 blood metabolites, a dataset meticulously compiled by Valeria Orru et al. A detailed exposition of these research data can be found in Table 1. Table 1Summary of the genome-wide association study data.Exposure/outcomeData sourcesSample ethnic originYearHLA DR on CD14‐ CD16+ monocytes https://gwas.mrcieu.ac.uk/ European2020AS https://www.finngen.fi/fi European20231091 blood metabolites https://www.ebi.ac.uk/gwas/home European2023AS = ankylosing spondylitis. Summary of the genome-wide association study data. AS = ankylosing spondylitis. 2.3. Instrumental variable selection and data harmonization In this study, our selection of genetic instruments encompassed SNPs achieving genome-wide significance (P < 5 * 10‐8). In scenarios devoid of such SNPs, we resorted to SNPs below this level of significance (P < 1 * 10‐5) as provisional IVs. This selection was guided by stringent criteria of linkage disequilibrium, defined by a window size of 10,000 kb and an r2 < 0.001, drawing upon linkage disequilibrium estimates from the 1000 Genomes Project, specifically European samples. Where an SNP relevant to the exposure was not mirrored in the outcome dataset, proxy SNPs were deftly employed, identified through meticulous LD tagging. To ensure the robustness of our MR analysis, palindromic and ambiguous SNPs were meticulously excluded from the instrumental variables. The strength of each instrumental variable was quantified using the F statistic, calculated as [(N‐2)*R2]/(1‐R2),[15] where N denotes the sample size. Instrumental variables exhibiting weak statistical strength, indicated by F-statistics lower than 10, were methodically removed from our analysis. Finally, we searched all SNPs associated with exposure and their proxies in databases such as PubMed, Google Scholar, and the GWAS Catalog to identify SNPs linked to confounding factors. We manually removed these SNPs to avoid potential pleiotropic effects. 2.4. Statistical analysis All MR analyses in this study were conducted using R software (version 4.2.3), employing the “TwoSampleMR” package. The principal analytical approach was inverse variance weighted (IVW) to ascertain the causal relationships between exposure factors and outcome variables. To validate these results, both MR-Egger and Weighted Median methods were applied. Distinct from IVW, the MR-Egger approach incorporates the consideration of an intercept term. The weighted median method, when at least 50% of the instrumental variables used are valid, is capable of providing a consistent estimate.[16] 2.5. Mediation analysis In this investigation, we conducted a detailed mediation analysis through a two-step MR framework to ascertain which among 1091 blood metabolites mediate the causal pathway from HLA DR on CD14‐ CD16+ monocytes phenotype and subsequently to AS outcomes. This analysis facilitated the partitioning of the overall effect into 2 components: an indirect effect, mediated through these metabolites, and a direct effect, occurring in the absence of such mediators.[17] Specifically, the influence of HLA DR on CD14‐ CD16+ monocytes phenotype in the context of AS was dissected into (1) direct effects, illustrating the unmediated impact of HLA DR (depicted as “a” in Fig. 1) and (2) indirect effects, where RA acts as a mediator in this causal chain (illustrated as “c*d” in Fig. 1). The proportion of the effect that is mediated was quantified by calculating the ratio of the indirect effect to the total effect. 2.6. Sensitivity analysis In our investigation, the phenomenon of gene pleiotropy was discerned through the lens of the MR-Egger regression intercept, where P < .05 emerged as a harbinger of horizontal pleiotropy’s presence. The Cochran Q test was employed to evaluate the heterogeneity among IVs. P > .05 suggests an absence of heterogeneity, warranting the application of the fixed-effect IVW method. Conversely, P < .05 indicates significant heterogeneity, necessitating the use of the random-effects IVW approach.[18] We also applied the MR pleiotropy residual sum and outlier test (MR-PRESSO) methods to assess the presence of and the outlying SNPs, and reassess the effect estimates. 3. Results 3.1. Association of HLA DR on CD14‐ CD16+ monocytes with AS The association of HLA DR on CD14‐ CD16+ monocytes did not achieve genome-wide significance for SNPs, with the significance threshold for these SNPs being lower than that typically observed across the entire genome. Consequently, these SNPs (P < 1 * 10‐5) were selected as IVs. Following the exclusion of non-informative and palindromic SNPs, 19 SNPs associated with HLA DR on CD14‐ CD16+ monocytes (see Table S1, Supplemental Digital Content, http://links.lww.com/MD/O177) and 87 SNPs associated with AS were identified for use as IVs. The F-statistics for these IVs all exceeded the threshold of 10, and the mean F statistic value was 53.79, which demonstrated a low risk of instrument bias. To elucidate the causal relationship between genetically predicted HLA DR on CD14‐ CD16+ monocytes and AS, we employed 3 statistical methods: IVW, MR-Egger, and weighted median analyses (Figs. 2 and 3). Forest plot to visualize causal effect of each single SNP on total AS risk. AS = ankylosing spondylitis, SNP = single nucleotide polymorphism. Forest plot to visualize the causal effects of N6, N6, N6-trimethyllysine levels with HLA DR on CD14- CD16+ monocytes and AS. AS = ankylosing spondylitis. Across 3 MR methodologies, a consistent and significant inverse association was observed between HLA DR on CD14‐ CD16+ monocytes and AS. Specifically, using IVW method, the odds ratio (OR) for HLA DR on CD14‐ CD16+ monocytes was 0.64 (95% confidence interval [CI], 0.53–0.78, P < .0001). Additionally, MR-Egger regression yielded an OR of 0.52 per standard deviation increase (95% CI, 0.35–0.77, P < .01), and the weighted median approach reported an OR of 0.75 for each standard deviation increase associated with HLA DR on CD14‐ CD16+ monocytes (95% CI, 0.63–0.90, P < .01). However, our MR analysis did not support reverse causality between genetically predicted HLA DR on CD14‐ CD16+ monocytes and AS, with an OR of 1.01 using the IVW method (95% CI, 0.99–1.04; P = .25). These findings are detailed in Figure 3. 3.2. Association of HLA DR on CD14‐ CD16+ monocytes with 1091 blood metabolites In the context of exploring the genetic determinants of disease, our study scrutinized the association between 1091 blood metabolites and genetic variants, revealing that these metabolites did not attain genome-wide significance for SNPs, with a threshold of P < 1 * 10‐5, thereby necessitating their use as IVs in our analysis. Utilizing a rigorous two-step approach, we identified N6, N6, N6-trimethyllysine levels (TML) as being causatively linked to HLA DR CD14‐ CD16+ monocytes and AS. After the exclusion of palindromic SNPs lacking appropriate proxies, a total of 16 SNPs achieved genome-wide significance and were subsequently employed as IVs (refer to Table S2, Supplemental Digital Content, http://links.lww.com/MD/O177), each with F statistics exceeding 10, indicating strong instrument validity. Employing instrumental variable analysis, including IVW, MR-Egger, and weighted median methods, we observed a statistically significant inverse relationship between genetically predicted HLA DR on CD14‐ CD16+ monocytes and the risk associated with TML (IVW:OR = 0.95 [95% CI, 0.91–0.99], P < .05), as illustrated in Figure 3. While the P-values from MR-Egger and weighted median analyses exceeded the .05 threshold, indicating a lack of statistical significance, the directional consistency across these methods with the IVW results underscores a potentially meaningful biological relationship warranting further investigation. 3.3. Association of TML levels with AS In our analysis, TML did not achieve genome-wide significance for SNPs, with a significance threshold set below that typically observed across the whole genome (P < 1 * 10‐5). This led to the selection of these SNPs as IVs. Following the exclusion of palindromic SNPs lacking suitable proxies, we identified 19 SNPs that met the criteria for genome-wide significance, thereby serving as IVs for our analysis, as detailed in Table S3, Supplemental Digital Content, http://links.lww.com/MD/O177. The robustness of these instrumental variables is underscored by F statistics exceeding 10 for each. As depicted in Figure 3, our genetic prediction analysis revealed a significant inverse association between TML and the risk of developing AS, as evidenced by IVW method (OR = 0.77, 95% CI, 0.60–0.99; P < .05). This negative correlation was consistently observed across different estimation methodologies, including IVW, MR-Egger, and the weighted median approach, indicating a coherent direction of effect across these analytic strategies. 3.4. Proportion of the association between HLA DR on CD14‐ CD16+ monocytes and AS mediated by TML We investigated TML as a mediator in the pathway linking HLA DR on CD14‐ CD16+ monocytes to AS. Our analysis revealed that HLA DR on CD14‐ CD16+ monocytes was correlated with increased levels of TML, which, in turn, was associated with a heightened risk of developing AS. As illustrated in Figure 4, TML was found to contribute to a ‐2.98% reduction in AS risk associated with HLA-DR on CD14‐ CD16+ monocytes. Despite TML’s role as a mediator in the pathway from HLA DR on CD14‐ CD16+ monocytes to AS, the direction of mediation by TML is contrary to the overall effect direction. Schematic diagram of the TML mediation effect. 3.5. Sensitivity analysis Multiple sensitivity analyses were conducted to assess the potential for pleiotropy. Cochran Q test and analysis of the funnel plot indicated heterogeneity in the association between HLA DR on CD14‐ CD16+ monocytes and AS mediated by specific SNPs; however, this heterogeneity did not compromise the integrity of the IVW results. Notably, such heterogeneity was absent among the remaining SNPs, as detailed in Table S4, Supplemental Digital Content, http://links.lww.com/MD/O177. Furthermore, the MR-Egger test’s Egger intercept did not reveal any significant statistical deviation (P > .05), supporting the inference that the SNPs under consideration were not subject to horizontal pleiotropy (Table S4, Supplemental Digital Content, http://links.lww.com/MD/O177). The robustness of the overall causal estimate attributed to each SNP was further confirmed through a leave-one-out analysis (Fig. S1, Supplemental Digital Content, http://links.lww.com/MD/O176), wherein systematic MR analysis, excluding 1 SNP at a time, yielded consistent results. This consistency underscores the significant contribution of all SNPs to the causal inference. 4. Discussion Recent MR analyses have elucidated the relationship between immune cell composition and the risk of AS, demonstrating a decrease in AS risk with an increased proportion of HLA DR on CD14‐ CD16+ monocytes. These findings are in agreement with our own research, which further establishes a causal link between HLA DR on CD14‐ CD16+ monocytes and AS. Utilizing MR analysis grounded in data from existing GWAS, our study explored the association between HLA DR on CD14‐ CD16+ monocytes and AS, revealing that the metabolite TML acts as a mediator in this causal pathway. Notably, our analysis indicates that the mediation effect of TML is contrary to the overall direction of effect. Specifically, our findings suggest that genetically predicted HLA DR on CD14‐ CD16+ monocytes correlates with an increased risk of AS, with an OR of 0.64 (95% confidence interval [CI], 0.53–0.78), and that TML mediates ‐2.98% of this effect. This finding is not contradictory. It demonstrates that TML-induced upregulation of HLA-DR expression on CD14‐ CD16+ monocytes represents a distinct subset, potentially contributing to the increased incidence of AS. The significant results in IVW and other analyses attest to our findings’ robustness, while some of the adverse result of the MR-Egger approach raise some concerns. Although the MR-Egger method is more sensitive to detecting violations of the instrumental variable assumptions, the influence of strong variants might have biased the results and have inflated type 1 error rates.[19] In that case, the results of IVW analysis are more reliable. Our study found that increased expression of HLA-DR on CD14‐CD16+ monocytes reduces the incidence of AS. Additionally, elevated levels of N6, N6, N6-trimethyllysine also correlate with a decreased incidence of AS. An increase in HLA-DR on CD14‐ CD16+ monocytes results in a reduction of N6, N6, N6-trimethyllysine levels. This suggests that HLA-DR on CD14‐CD16+ monocytes increases N6, N6, N6-trimethyllysine levels, thereby reducing the risk of developing AS. However, reduced HLA-DR expression diminishes this protective effect. By intervening in this pathway, such as by administering N6, N6, N6-trimethyllysine, we can artificially elevate its concentration in the blood, potentially reducing the incidence of AS. Our findings are predictive and cannot yet be linked to the role of nonclassical monocytes in AS, as this area remains unexplored. Currently, no experimental data support these findings, necessitating future research. Monocytes, as critical components of the innate immune system, play a pivotal role in maintaining vascular homeostasis and serving as primary responders to acute infectious agents. AS is a chronic inflammatory autoimmune disease of uncertain etiology. Current research highlights the significant role of inflammatory and immune factors in its pathogenesis. CD14‐ CD16+ monocytes, classified as nonclassical monocytes, are recognized for their anti-inflammatory properties and their role in maintaining vascular homeostasis. These monocytes serve as the first line of defense against pathogens and are known to exert a protective effect in chronic diseases, correlating positively with disease burden. Emerging evidence suggests a correlation between nonclassical monocytes and the progression of arthritis,[20] challenging the conventional understanding of their anti-inflammatory capacity.[21] Furthermore, numerous studies have implicated monocytes in the inflammation characteristic of chronic conditions, diverging from the anticipated outcomes of our overarching analysis. Through an investigation of the metabolite TML, we identified its indirect effects as causative risk factors, aligning with several current findings. This discovery underscores the necessity for additional research to elucidate the complex roles of TML in AS. TML, a prevalent posttranslational modification of lysine residues found across a wide array of species from plants to animals, emerges as a significant blood metabolite with a notable presence.[22] Recent research underscores its critical involvement in the pathogenesis of AS and its potential diagnostic utility for this condition.[23] Despite these intriguing findings, the exploration of TML’s role in AS remains sparse, presenting a compelling case for deeper investigational pursuits to elucidate its functions more clearly. This study is subject to several limitations that warrant mention. Initially, the analysis was confined to a European demographic, thus constraining its generalizability. Additionally, the absence of detailed individual-level data precluded the opportunity for nuanced stratified analyses within the study population. Moreover, despite conducting extensive sensitivity analyses, the biological effects of these significant SNPs are not available to fully filter out pleiotropy. A further consideration is the adoption of a relatively lenient threshold for result evaluation, potentially introducing an increased risk of false-positive findings. Lastly, our findings reveal that the genetic prediction of TML-mediated AS contribution is minimal and negative at ‐2.98%, highlighting the necessity for further research to identify and quantify the impact of additional mediators. 5. Conclusion In conclusion, our investigation establishes a causal link between HLA DR on CD14‐ CD16+ monocytes and AS, albeit with a modest and inverse contribution from TML-mediated effects. This finding suggests that the precise impact of HLA DR on CD14‐ CD16+ monocytes in the pathogenesis of AS remains to be fully elucidated. Consequently, there is a pressing need for further exploration into additional risk factors that could serve as potential mediators in this relationship. From a clinical perspective, individuals exhibiting reduced HLA DR on CD14‐ CD16+ monocytes should be closely monitored for the potential onset of AS, underscoring the importance of vigilant patient care in managing this condition. Author contributions Conceptualization: Danyun Hua, Xiang Xu. Data curation: Lu Wang. Formal analysis: Na Li. Investigation: Lu Wang, Na Li. Methodology: Danyun Hua, Lu Wang, Xiaohu Yin. Project administration: Xiang Xu, Xiaohu Yin. Resources: Na Li, Xiang Xu. Software: Lu Wang, Xiang Xu. Supervision: Danyun Hua. Validation: Danyun Hua. Writing – original draft: Danyun Hua, Xiaohu Yin. Writing – review & editing: Danyun Hua. Supplementary Material