Identification of genetic association between mitochondrial dysfunction and knee osteoarthritis through integrating multi-omics: a summary data-based Mendelian randomization study

Knee osteoarthritis (KOA) is a common musculoskeletal disease among middle-aged and elderly individuals, characterized by chronic pain and functional disability [1]. It involves pathological changes in all joint tissues, including synovitis, cartilage extracellular matrix degradation, chondrocyte apoptosis, subchondral bone sclerosis, and osteophyte formation [2]. Globally, KOA accounts for approximately 80% of all OA cases. The onset of KOA symptoms peaks between the ages of 55 and 64, and the prevalence increases with age, significantly impacting patients’ quality of life and well-being [3, 4]. Although previous studies have investigated various aspects of the pathogenesis of KOA, such as abnormal mechanical load, oxidative stress, inflammatory factors, and mitochondrial dysfunction, the underlying mechanism is not fully understood [5]. Therefore, further research is necessary to deepen our understanding of the pathogenesis of KOA.

Mitochondria serve as the primary sites of energy production in eukaryotic cells and generate adenosine triphosphate (ATP) through oxidative phosphorylation. Additionally, mitochondria play crucial roles in various cellular functions, such as oxidative stress, biosynthesis and metabolism, and intercellular communication [6]. These highly dynamic cellular organelles continuously reshape their morphology through fusion and fission to adapt their functions. When mitochondrial dynamics are imbalanced, mitochondrial dysfunction occurs, leading to cell degeneration or death [7]. KOA is characterized by chondrocyte damage influenced by multiple factors, with aging closely related to its development. Mitochondrial dysfunction is an important marker of cell aging. Ageing-related mitochondrial dysfunction damages cartilage by releasing reactive oxygen species (ROS), leading to the progression of KOA [8]. During mitochondrial dysfunction, excessive ROS accumulation inhibits the oxidation of cysteine to cysteine sulfinic acid (Cys-SO2H) or sulfonic acid (Cys-SO3H), thereby accelerating disease progression by damaging cellular pathways [9]. Substantial evidence has shown that mitochondrial autophagy is involved in KOA pathogenesis. Impaired mitochondrial autophagy results in excessive ROS accumulation, contributing to KOA [10]. Jiang et al. found that Cpt1a knockout regulated the aging phenotype of chondrocytes through PINK1-mediated mitophagy [11]. Liu et al. found that alpha-ketoglutarate (alpha-KG) significantly improved mitochondrial respiration in chondrocytes, enhanced mitochondrial autophagy, and delayed articular cartilage degeneration [12]. Moreover, numerous studies have shown that mitochondrial-related genes, such as monoamine oxidase B (MAOB) and mitochondrial protein mitofusin 2 (MFN2), are related to cell aging and OA [13, 14]. In addition to the 37 key mitochondrial genes, the associated genome contained more than 1000 additional nuclear genes. The susceptibility of these genes indicated that the mitochondria are easily affected by external environmental factors [15]. Mitochondrial dysfunction is a complex process. Although the promotion of KOA by mitochondrial dysfunction has been extensively studied, the specific mechanism remains unclear, necessitating further research.

Mendelian randomization (MR) is a novel method, which uses genetic variations as instrumental variables (IVs) to offer inferences regarding the causal relationships between exposures and outcomes. Compared with randomized controlled trials (RCTs), it can effectively eliminate the interference of confounding factors and reverse causation, resulting in greatly improved accuracy and credibility of results [16]. As an extension of the MR analysis, summary data-based MR (SMR) method was developed to facilitate the inference of genetic association between diverse molecular traits (e.g., gene expression, DNA methylation, or protein abundance) and diseases of interest [17]. Compared with traditional MR analyses, SMR, using the top cis-quantitative trait loci (QTL) as the exposure, can yield higher statistical power when the exposure and outcome are obtained from two independent samples with large sample sizes [17]. Notably, in recent years, as large-scale genome-wide association study (GWAS) and molecular QTL data continue to be explored, we have opportunity to integrate them to explore susceptibility risk genes for different diseases. Up to now, limited SMR studies had identified the potential genetic association between mitochondrial dysfunction and KOA. Therefore, this study aimed to explore the genetic relationship between mitochondrial dysfunction proxied by mitochondrial-related genes and KOA using multi-omics data (including gene expression (eQTL) and protein QTL (pQTL)) data.

Using large-scale QTL data and GWAS summary data, we performed SMR analysis to examine the association of genetically proxied mitochondrial dysfunction with KOA risk. By combining SMR analysis and HEIDI test, the expression levels of 2 mitochondrial-related gene were found to be associated with KOA risk. Specifically, elevated gene expression levels of IMMP2L increased the risk of KOA, whereas increased gene expression levels of AKAP10 decreased the risk of KOA. Colocalization analysis demonstrated that AKAP10 and IMMP2L shared the same genetic variant with KOA. Additionally, consistent results were found in replication study, musculoskeletal tissues, and SMR-multi test, further suggesting the reliability of our findings and revealing the important role of mitochondrial dysfunction mediated by these genes in the pathogenesis of KOA.

AKAP10 gene, located on human chromosome 17, encodes a multi subunit protein that binds to the regulatory subunit of protein kinase A (PKA) [33]. Previous studies confirmed its important role in humans [34]. In recent decades, genetic variants of AKAP10 have been associated with cardiac arrhythmias[35], breast cancer[36], and preterm birth [37]. However, limited reports exist on the relationship between AKAP10 and OA. In the present study, we found that increased AKAP10 expression was associated with a reduced risk of KOA. The association can be reasonably explained through two possible mechanisms. First, AKAP10 contributes to the cholinergic/vagal signaling pathway and increases vagal sensitivity [35]. Notably, the presence of cholinergic nerves in most joint tissues, such as cartilage and subchondral bone, has been well described in a systematic review [38]. Some studies have found that vagus nerve activation can inhibit the production of cytokines (e.g., tumor necrosis factor (TNF), which may ameliorate joint inflammation [39]. Nicotine, an exogenous stimulator of the cholinergic system, has been shown to alleviate OA pain and slow cartilage degradation induced by sodium iodoacetate in mice [40]. Several in vitro tests have suggested that treatment with acetylcholinesterase inhibitors prevents cartilage degeneration and inflammation [41, 42]. Second, Kim et al. demonstrated that AKAP10 increased lipopolysaccharide-induced nitric oxide (NO) production [43]. NO, present in almost all types of human cells, has been shown to potentially have a protective effect on chondrocytes [44]. Surprisingly, NO was found to potentially relieve OA-related pain through blood flow, neurotransmitter pathways, opioid receptor pathways, and anti-inflammatory pathways [45]. Early subchondral bone loss and advanced subchondral bone sclerosis are important pathologic characteristics of OA [46, 47]. Therefore, the inhibition of excessive bone resorption mediated by osteoclasts in the early stages and abnormal bone formation mediated by osteoblasts in the late stages are currently the main therapeutic strategies of interest [48]. An older study by MacIntyre et al. revealed that NO inhibits the proliferation of osteoclasts and bone resorption in mice [49]. Many subsequent studies have found similar results [50, 51]. Notably, high concentrations of NO produced by NO donors or pro-inflammatory factors have been shown to be effective in inhibiting osteoblast growth and differentiation [52, 53]. Taken together, AKAP10 may slow the progression of OA by promoting the cholinergic/vagal nervous system and NO production.

IMMP2L encodes a protein responsible for cleaving signal peptide sequences of cytochrome c1 (CYC1) and mitochondrial glycerophosphate dehydrogenase 2 (GPD2) [54]. IMMP2L has previously been investigated in relation to ovarian aging [55] and autism[56]; however, limited research has been conducted on its association with KOA. Our results provide evidence that IMMP2L is a risk factor for KOA. Consistent with our findings, Lawther et al. recently observed decreased ROS in IMMP2L knockout mice [57]. Considering the close relationship between KOA and ROS, this finding appears plausible. More mechanisms underlying IMMP2L-induced KOA remain to be investigated.

The primary strength of our study is that we used a MR design, which diminished the bias resulting from confounders and reverse causality of traditional observational studies. In addition, evidence for genetic association was further strengthened by the success of colocalization analysis and HEIDI test. More importantly, consistent results were observed in replication study, musculoskeletal tissues, and SMR-multi test, suggesting the reliability of our findings. However, there are still several limitations. First, although the GWAS summary data for KOA in discover study we used were largest available, participants involved in this study were not exclusively of European ethnicity. It may lead to some bias. Surprisingly, fewer than 3% of the participants were of non-European ancestry; thus, we believe that the results remain robust. Second, there was a certain unmeasurable sample overlap that existed between the KOA GWAS datasets used in the primary and replication studies. It is caused by the inherent limitations of the development and updating of GWASs, as majority of the current large OA GWASs have been expanded and updated by adding new cohorts to the original data [22, 23, 58, 59]. Thus, we had to acknowledge this limitation that cannot be mitigated. In addition, probably due to the pQTL data have not been fully exploited yet, we failed to investigate any mitochondrial-related proteins associated with KOA. We expect more researchers to follow up with further exploration of pQTL data. Furthermore, the candidate genes we identified for KOA have small effect size and need to be carefully considered in clinical application. Finally, there were no available independent GWAS summary data that directly represent mitochondrial dysfunction, so we used genetically predicted mitochondrial-related genes to represent it.

In summary, our SMR and colocalization analyses examined the genetic association between mitochondrial dysfunction proxied by mitochondrial-related genes and KOA and revealed the important role of mitochondrial dysfunction mediated by these genes in the pathogenesis of KOA. Furthermore, these identified candidate genes offer the possibility of clinical drug target development for KOA. Further research was needed to validate our findings.