Autophagy-related genes analysis reveals potential biomarkers for prediction of the impaired walking capacity of peripheral arterial disease

Atherosclerosis of the arteries supplying the legs is commonly labeled as lower extremity peripheral arterial disease (PAD). PAD is a common disorder and the third leading cause of atherosclerotic cardiovascular morbidity, following coronary artery disease (CAD) and stroke [1]. It can manifest as intermittent claudication when walking or, in the most severe form, critical limb ischemia. Limb ischemia defined as chronic ischemic rest pain, ulcers, or gangrene of the lower extremity, which is a major cause of limb amputation, and a harbinger of cardiovascular mortality in PAD [1, 2]. Age, tobacco use, diabetes, hypertension, hypercholesterolemia, and a sedentary lifestyle are the major risk factors for PAD [3].

Autophagy is a cellular recycling process in response to various stressors (e.g., oxidative stress, hypoxia, and starvation). It has been implicated in several fundamental biological processes, including aging, immunity, development, tumorigenesis, vascular disease, cell death and differentiation [4, 5]. In the cardiovascular system, autophagy is a key regulator of homeostasis, responding to physiological and pathophysiological stimuli. Although recycling of cellular organelles is generally viewed as a beneficial process, insufficient and excessive levels of autophagy can lead to premature cell death (apoptosis) [6]. To date, there is increasing evidence that basal autophagy plays an essential role in protecting endothelial cells and smooth muscle cells from cell death and the development of vascular disease, particularly heart failure and atherosclerosis [5, 7]. Current research suggest that autophagy is also a fundamental process for cardiovascular homeostasis, health, and aging [8]. To the best of our knowledge, despite the protective role of autophagy in various diseases, its role in the vascular system is poorly understood and is entirely unknown in the context of PAD [9]. Exploring and uncovering potential autophagy-related genes and the autophagy level in PAD as well as a functional linkage with relevant molecular processes of vascular pathophysiology may provide potential biomarkers for assessing and monitoring disease risk and prognosis.

Recently, Biros et al. [10, 11] uploaded a gene expression omnibus (GEO) dataset GSE57691, which revealed the differentially expressed genes in the artery wall tissue from 49 patients with abdominal aortic aneurysm, 9 patients with chronic lower limb ischemia and 10 healthy controls. In the present study, we reevaluated this dataset for potential implications for lower limb ischemia, which served as the basis for an extensive new analysis. First, we explored the differentially expressed autophagy-related genes in PAD patients from the GSE57691 data set and validated target genes expression in peripheral blood samples from our Lauflabor (WalkByLab) registry participants. Then protein–protein interaction (PPI) analyses, Gene Ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) enrichment analysis were performed to investigate the biological functions of these autophagy-related genes. Subsequently, western blot was performed to assess the level of autophagy in peripheral blood mononuclear cells (PBMCs) of WalkByLab participants by analyzing the degradation of autophagic marker proteins (beclin-1, P62, and LC3B). Here, a `pro-inflammatory´ phenotype and immune activation is an important feature of PAD [12]. Therefore, we next performed single sample gene set enrichment analysis (ssGSEA) to investigate the immune microenvironment of the artery wall of PAD patients and link molecular inflammatory processes with autophagy-related genes. Finally, we analyzed and validated potential biomarkers for predicting impaired walking capacity as assessed by Gardner treadmill testing used in clinical practice. In particular, the Gardner treadmill testing is regarded as the gold standard for physical performance in the diagnoses of PAD status, but it requires appropriate equipment, time, and a professionally trained team, which is why it is rarely used in clinical practice. In summary, the aim of this study was to (1) gain insight into the relationship between autophagy effector molecules and PAD, (2) link autophagy to inflammatory processes in the development of PAD, and (3) identify novel biomarkers that can reliably predict walking ability and PAD status in clinical practice.

Autophagy is a cellular self-recycling process, which is involved in both vascular homeostasis as well as physiological and pathological arterial remodeling. Physiological autophagy serves as a protective mechanism to maintain normal cardiovascular function, whereas disordered autophagy contributes to the development and progression of diseases [19]. It seems to play a protective role in the early phase of atherosclerosis, as deletion of autophagy genes (Atg5) accelerates the progression of atherosclerosis in murine models [20]. In clinical studies, protein levels of autophagy‑associated protein beclin1 and light chain 3 (LC3) were decreased in PBMCs from patients with acute myocardial infarction compared with healthy controls or patients suffering from stable angina pectoris [21]. Based on these findings, it is speculated that acute cardiovascular events and increased severity of disease might lead to a reduction in autophagy and vice versa [6, 21]. However, the role of autophagy in cardiovascular disease development is not entirely clear, in particular, there are little study related to PAD.

In this study, we identified 20 potential autophagy-related genes in PAD through bioinformatic analysis, and validated changes in the expression of these genes in peripheral blood samples from the WalkByLab registry trial. All identified genes were expressed at low levels or had the tendency of decreasing in PAD patients’ PBMCs, which indicated a role of autophagy-related genes in PAD. The GO and KEGG enrichment analysis further revealed that these autophagy-related genes might be involved in autophagy activity and control of autophagy activation. Indeed, western blot analysis demonstrated that autophagic marker proteins beclin1 and LC3B levels were significantly decreased in the PBMCs of PAD patients. Beclin1 and LC3B are common biomarkers of autophagy, which are widely used to evaluate the level of autophagy. Our study now provided evidence that PAD is linked to a reduced expression level of these autophagy-related genes and a decreased autophagy marker protein level.

Inflammation is one of the main features of PAD. Autophagy plays an important role in inflammation by regulating the development, homeostasis, and survival of inflammatory cells such as macrophages, neutrophils, and lymphocytes. Autophagy also influences the transcription, processing, and release of a variety of cytokines and is regulated by cytokines [22]. Autophagy can trigger a dynamic response to inflammation. Active autophagy suppresses excessive inflammatory response by inhibiting inflammasome activation, facilitating damage-associated molecular patterns and damaged mitochondria clearance, and destroying inflammatory mediators [23]. Suppressed autophagy resulted in an enhanced inflammatory response [24].

Thus, we explored the overall characteristics of the immune microenvironment of artery tissue from PAD patients and healthy populations and the association with autophagy-related genes. We found higher level of immune cell infiltration and more active immune-related pathways in PAD and that the CCR interaction was the most active immune-related pathway negatively associated to autophagy-related genes. This seemed to imply that the decreased autophagy will increase inflammation in PAD. Here, the cytokines include the chemokines, the PDGF family, and the TGF-β family [25]. Among CCR interactions, chemokines are best known for their ability to stimulate the migration of leukocytes (e.g. PBMCs) and chemokines in particular are known to play an important role in the pathophysiology of cardiovascular disease [26]. Chemokines bind to specific G protein-coupled receptors and exert distinct functions, such as regulating the activation of leukocytes and coordinating their trafficking to the sites of inflammation [27]. For example, chemokine RANTES level is a useful marker of CAD severity because elevated chemokine RANTES level in patients with stable angina may predict the high risk of plaque formation in early stages of atherosclerosis [28]. Therefore, we measured the expression profiles of human chemokines in plasma from PAD patients and healthy individuals to identify a potential role of chemokines in PAD pathophysiology. ELISA analysis of the WalkByLab participants showed that human chemokine GRO and NAP2 was highly expressed in PAD which was also demonstrated by a strict propensity score-matched analysis after adjustment for covariate of age. In addition, our analysis indicated that GRO and NAP2 may be promising diagnostic biomarkers for PAD, as reflected by their respective AUC values of 0.670 and 0.752. Moreover, our analysis showed that the levels of GRO and NAP2 were not associated with MACE (acute events), therefore indicating their specificity for a systemic PAD-related pathophysiology. Interestingly, our study also revealed a negative correlation between GRO and NAP2 levels and the expression of autophagy-related genes KLHL24, VAMP3, HSPA5, and ST13, suggesting a potential link between the dysregulation of autophagy and chemokine-mediated inflammation in PAD. Further investigation is needed to elucidate the mechanistic relationship between autophagy and GRO/NAP2 signaling in the context of PAD pathophysiology.

We next examined other crucial characteristics of PAD pathophysiology. Important cofactors and clinical parameters for the evaluation of PAD are endothelial function measured by FMD and walking distance measured by a Gardner treadmill test. Our study demonstrated that plasma GRO and NAP2 levels were not correlated with participants’ FMD, but were negatively correlated with the pain-free walking and maximum walking distance in Gardner treadmill testing. Besides, GRO and NAP2 were able to predict poor walking capacity (walking time < 6 min in Gardner treadmill testing) with respective AUC values of 0.701 and 0.743. Although FMD is a very valuable clinical parameter, it is also very sensitive to a variety of external influences, such as biological age, diet, or exercise level of the patient [29]. In clinical practice FMD is therefore not used for the primary diagnosis of PAD. However, the Gardner treadmill walk test is considered the gold standard to accurately determine the stage of PAD according to the Fontaine stage. To our knowledge, a role of GRO and NAP2 for PAD in the context of walking distance and walking capacity has never been demonstrated before. Previous studies have mainly focused on the role of GRO and NAP2 in cancer development and metastasis of various malignancies [30], whereas few reports have focused on cardiovascular disease, and even fewer in the context of PAD. In the study of Ku et al. [31], upregulated NAP2 was significantly associated with severity of coronary artery stenosis in patients with diabetes. Recently Wang et al. showed that elevated plasma NAP2 level was independently associated with critical limb ischemia, not with diabetes [32], however a connection with the walking capacity was not considered. Until now the deregulation of GRO was never shown before in the context of PAD. Chemokines GROα, GROβ, and GROγ constitute the three members of the GRO family [33], which together with NAP2, share the chemokine receptor CXCR2 as a common binding receptor [34]. GROγ was reported to be upregulated in acute coronary syndromes and related to the severity of the inflammatory response [35]. Moreover, a recent study demonstrated that CXCL3 (the coding gene of GROγ protein) gene was highly expressed in acute ischemic stroke and it can be recognized as a predictor for more infarct volume [36]. In summary, chemokines GRO and NAP2 appear to be useful biomarkers in PAD, particularly in predicting impaired walking ability.

Finally, we further analyzed and processed plasma GRO or NAP2 levels with the relevant clinical parameters obtained by LASSO analysis to build nomogram models for the prediction of poor walking capacity (walking time < 6 min in Gardner treadmill testing). The NAP2-derived nomogram showed a well predictive performance, with an AUC of 0.860, while the GRO-derived nomogram had a slightly lower AUC of 0.854. These evidence demonstrated that plasma NAP2 indeed could be used as a predictive biomarker and the derived nomogram model simply and accurately predicted impaired walking capacity in patients with PAD.

The here presented study has certain limitations that need to be acknowledged. Firstly, the sample size was relatively small, and therefore, it is important to exercise caution when generalizing the results. Secondly, this study was retrospective, and as such, there is a possibility of selection bias. Nevertheless, the retrospective nature of the study was necessary to pave the way for future prospective studies. Thirdly, a correlation between autophagy-related genes and chemokines with walking ability was discovered, suggesting that autophagy may play a role in the progression of PAD by regulating chemokines and inflammation. However, this potential mechanism requires further validation in animal models, and thus, future studies are necessary to shed more light on this issue.

In conclusion, our study shed light on the interrelationships of autophagy and autophagy-related genes in the development of PAD, whcih were previously unknown. Our comprehensive analysis identified a set of 20 autophagy-related genes that were expressed at lower levels in PAD and demonstrated that the autophagy levels of PBMCs were significantly decreased in patients with PAD. This finding has significant implications for future clinical strategies and for deepening the understanding of disease onset and progression, as well as providing novel biomarkers. These results will serve to guide the clinic in developing new prognostic strategies for the evaluation of PAD. Our bioinformatics analysis further revealed a link between vascular inflammation and the effect of cytokines, particularly chemokines, with autophagy-related genes. This was also further validated in subsequent experiments. We found that GRO and NAP2 levels were elevated in the plasma of PAD patients, with a mild-to-moderate negative correlation between them and the autophagy-related genes KLHL24, VAMP3, HSPA5, and ST13. The effects of GRO and NAP2 on walking capacity in PAD have never been proven before. Our study reveals a mild-to-moderate negative correlation between chemokines and pain-free walking distance and maximal walking distance in the Gardner treadmill test. This suggests that reduced autophagy may influence the onset and progression of PAD (especially, the walking capacity) by affecting chemokines (inflammation). Finally, we constructed nomogram models that can predict poor walking capacity. The results showed that NAP2 is a promising biomarker for PAD patients, and its derived nomogram had stronger predictive potential than the GRO derived nomogram. This nomogram provides a valuable tool for clinicians to improve their understanding of a patient's disease risk and prognosis, thereby facilitating informed clinical decision-making. In all, our study provides new insights into the molecular mechanisms underlying PAD and may have implications for the development of new diagnostic and therapeutic approaches for PAD.