Introduction
Arteriovenous (AV) fistula is a preferred type of vascular access for hemodialysis in patients with end-stage renal disease given its superior patency rate. However, various studies reported a poor 1-year primary patency rate of AV fistula not exceeding 60–65% (Dixon et al., 2002; Tordoir et al., 2003; Arhuidese et al., 2018). Approximately 40% of AV fistula are no longer functional 2 years after placement (Gibson et al., 2001; Arhuidese et al., 2018). The stenotic lesions in the venous outflow tract as the major late complication contribute to AV fistula failure. Percutaneous balloon angioplasty (PTA) is usually applied to maintain AV fistula patency because it reduces the need for surgical intervention, AV fistula abandonment, or AV fistula creation of more costly AV graft, given the historically greater need for maintenance intervention (Stolic, 2013). However, the restenosis after PTA occurs frequently once the first PTA is launched and that leads to repeat operative procedures until abandonment, which increases medical cost and aggravates social burden. It is relatively important to prevent stenosis after AV fistula creation and to prolong the time of primary stenosis occurrence for the management of AV fistula.
The exact pathophysiology of stenosis in AV fistula remains unclear. After AV fistula creation, the increased arterial flow alters hemodynamics state that strengthens vessel wall shear stress (Remuzzi and Ene-Iordache, 2013) and could result in a complex multidirectional and reciprocating flow in the inner side of the venous segment (Browne et al., 2015). The disturbed flow, with low and reciprocating wall shear stress, induces the selective expression of prooxidant, procoagulant, proinflammatory, and proapoptotic genes in endothelial cells (Dai et al., 2004; Chiu and Chien, 2011) and also stimulates vascular smooth muscle cell (VSMC) migration and proliferation. All the above alterations in AV fistula could result in intimal hyperplasia (Fan et al., 2010; Ene-Iordache et al., 2013). The development of intimal hyperplasia involves several vascular biological processes such as inflammation, uremia, hypoxia, shear stress, and thrombosis (Stracke et al., 2002; Roy-Chaudhury and Lee, 2007; Roy-Chaudhury et al., 2009; Wasse et al., 2012; Simone et al., 2014). However, the molecular mechanisms of intimal hyperplasia in AV fistula remain poorly understood and probably include endothelial signaling (Jones et al., 1998; Caplice et al., 2007), inflammatory and coagulation (Nath et al., 2003), extracellular matrix regulators (Misra et al., 2010), and growth factors and cell molecules (Yang et al., 2014; Sadaghianloo et al., 2017). Undoubtfully, there is an urgent need to understand the mechanism of intimal hyperplasia and find candidate target biomarkers to prevent stenosis after AV fistula creation.
The bioinformatics analysis of the microarray data is the most commonly used and successful technique to explore the pathogenesis of complicated diseases and to identify therapeutic targets. However, limited studies utilize human gene expression profiling to explore the molecular mechanism of pathophysiology in AV fistula. The microarray data of GSE39488 (Hashimoto et al., 2013) from the Gene Expression Omnibus (GEO) database, comparing the venous outflow segment of AV fistula exposed to overflow with the normal cephalic vein in hemodialysis patients, provide the opportunity for us to investigate regulatory pathways, essential genes, and their associated networks in the venous segment of AV fistula.
Therefore, we aimed to identify the crucial gene expressed in the venous outflow tract of AV fistula based on a bioinformatics approach. The exposure of the overflow of blood in AV fistula certainly and inevitably leads to intimal hyperplasia in the venous outflow tract (Hofstra et al., 1995; Tabbara et al., 2016). Thus, the identification of relevant genes might also offer better insight into the therapeutic targets of intimal hyperplasia to prevent stenosis after AV fistula creation.
Discussion
Although AV fistula is the recommended option for hemodialysis patients, stenosis due to intimal hyperplasia is the main obstacle for maintaining AV fistula patency. Thus, the analysis of gene expression of overflow AV fistula could provide insight into the molecular mechanism of intimal hyperplasia.
Microarray technology is one of the important ways to explore gene expression related to complex disorders and could help us to predict target genes for treatment. In this study, we screened 185 DEGs containing 111 upregulated genes and 74 downregulated genes in the AV fistula group compared with the control group. These identified DEGs were subjected to the GO and KEGG enrichment analysis to acknowledge the potential biological process and signaling pathway. GSEA was used to explore the potentially activated biological processes and signaling pathways between the AV fistula group and the control group with the advantage of avoiding undetectable, small changes in the expression of single genes (Mootha et al., 2003). Subsequently, we subjected the identified DEGs to the STRING database, which exploits the complex interactions between DEGs based on evidence from experiments and repositories. In the PPI network, the nodes represented proteins, the edges between nodes represented their interactions, and the top 10 genes ranked by MCC indicating high weight between DEGs in the network were filtered as hub genes that might be important in the regulation of biological activity in AV fistula. The gene set of a signaling pathway to which a gene belongs suggests that it potentially participates in the pathway. The intersection of hub gene and four crucial gene set of signaling pathway help us to determine in which pathway the hub gene participated. Our study identified two hub genes as core genes that potentially participated in the MAPK signaling pathway.
The GO and KEGG enrichment helped us to explore the involvement of the 185 identified DEGs in BP, MF, CC, and pathways and to determine the functional annotation of these genes. As shown in Supplementary Tables 1, 2, the GO enrichment of upregulated genes in the venous segment of AV fistula was mainly involved with cell adhesion, intracellular signal transduction, and angiogenesis, while the downregulated genes were primarily enriched in the biological function relevant to apoptosis (Kong et al., 2001; Schumacher et al., 2010; Poon et al., 2014; Barth et al., 2017). Those results were in line with the previous report and crucial for intimal hyperplasia. After AV fistula creation, the hemodynamics change in outflow vein results in the short-term and long-term vascular remodeling (Remuzzi and Ene-Iordache, 2013), and sustained expression of several molecules that cause the activation of a series of pathways stimulating intimal hyperplasia (Chiu and Chien, 2011; Zafrani and Ince, 2015). Shear stress can improve the expression of surface adhesion molecules (i.e., vascular cell adhesion protein 1) (Chiu and Chien, 2011). Mechanical injury in the artery could activate a mild adventitial angiogenic response and aggravate neointima formation (Chiu and Chien, 2011). Neoangiogenesis has also been observed in the AV fistula rat model (Chiu and Chien, 2011). The activation of apoptosis in VSMCs can reduce intimal hyperplasia (Aizawa et al., 2001).
The KEGG enrichment analysis screened a total of five significant pathways based on the DEGs. Simultaneously, GSEA identified 46 significant gene set/pathways in AV fistula and 14 significant gene set/pathways in normal cephalic vein. All those significant pathways in AV fistula were concentrated on proliferation, apoptosis, cytokine–cytokine interaction, intimal hyperplasia, inflammation response, and cancer-related pathways. Interestingly, MAPK signaling pathway, NOD-like signaling pathway, Cell Cycle, and TGF-beta signaling pathway have been reported to play a prominent role in the development of intimal hyperplasia. The dysregulation of the MAPK signaling pathway leads to uncontrolled cell proliferation, migration, and differentiation. Recent studies reported that the MAPK signaling pathway is associated with intimal hyperplasia by the ERK1/2 signaling pathway (Qu et al., 2020). The inhibition of the MAPK pathway decreases the thickening of the intima and limits vascular remodeling (Ge et al., 2013; Yang et al., 2020). NOD-like receptor signaling pathway could trigger a family of pattern recognition receptors that mediate the initial innate immune response to cellular injury and stress. The activated signal molecule of NOD-like receptor family such as NLRP3 inflammasome could be aggregated to form an inflammasome complex and to enhance neointima formation (Xia et al., 2014; Koka et al., 2017). Cell cycle regulatory factors could promote VSMC migration and proliferation in the pathogenesis of intimal hyperplasia (Wei et al., 1997; Zuckerbraun et al., 2003). A study that detected the cell cycle regulators in 18 failed human AV fistula demonstrated that the regulators positively regulating the cell cycle progression were significantly higher in AV fistula than normal vein and arteries (de Graaf et al., 2003). TGF-beta is a family of regulatory proteins containing many multifunctional cytokines, such as TGF-beta, bone morphogenetic proteins (BMPs), activins, inhibins, and mycostatin (Bobik, 2006). The regulatory proteins of TGF-beta family participate in a wide range of biological processes such as cell growth and differentiation (ten Dijke et al., 1996). The expression of TGF-beta in neointimal and medial layers is higher in stenosed hemodialysis fistula (Stracke et al., 2002). The previous study has revealed that the overexpression of TGF-beta stimulates intimal hyperplasia and extracellular matrix accumulation (Nabel et al., 1993). Additionally, targeting the TGF-beta mRNA degradation attenuates neointima growth (Yamamoto et al., 2000).
From the analysis of the PPI network, 10 DEGs with the highest MCC score were filtered as hub genes, namely, EGR1, EGR2, EGR3, NR4A1, NR4A2, ATF3, CXCR4, CCL4, CYR61, and DUSP1. As shown in Supplementary Figure 2, the expression level of the 10 hub genes was notably increased in the patients with AV fistula compared with the normal controls. Early growth response genes are the family of zinc-finger transcription factors comprising of four members, namely, EGR1, EGR2, EGR3, and EGR4. They are expressed in many different cell types and could be rapidly induced in response to mitogens, differentiation, apoptotic signals, and tissue injury (Decker et al., 1998). EGR1 is crucial for intimal hyperplasia and vascular remodeling after vascular injury (Huang et al., 2010), balloon injury (Han and Liu, 2010), arterial hypertension (van der Feen et al., 2016), and chemical damage (Vazquez-Padron et al., 2010). The inhibition of EGR1 expression attenuates intimal hyperplasia through the downregulation of TGF-beta and the upregulation of nitric oxide synthase (NOS) (Liu et al., 2008; Han and Liu, 2010). In our study, an augmented expression level of EGR1 might enhance intimal hyperplasia in the venous segment of AV fistula exposed to high blood flow, indicating EGR1 could be one novel regulatory biomarker in AV fistula. EGR2 and EGR3 were essential for the regulation of immune response and the development of inflammation (Li et al., 2012; Sumitomo et al., 2013) but have not been proved to be associated with intimal hyperplasia especially in AV fistula.
The NR4A subfamilies of nuclear receptors such as NR4A1 and NR4A2 become significant regulators in the developing neointima and play a critical role in several aspects of vascular remodeling such as cell viability, proliferation, and inflammation (Zhao and Bruemmer, 2010). Some studies indicated that the activation of NR4A1 and NR4A2 could inhibit VSMC proliferation and protect against neointima formation (Pires et al., 2007; Bonta et al., 2010). But Liu et al. found that Nur77(NR4A1)-targeted siRNA could mitigate autologous vein graft stenosis (Liu et al., 2015). In our study, the expression level of NR4A1 and NR4A2 in the AV fistula group is higher than that in the control group, which were consistent with the result in the autologous vein graft stenosis model (Liu et al., 2015) but not in the cuff-induced neointima formation model (Pires et al., 2007) and the carotid artery ligation-induced neointima model (Bonta et al., 2010). That is because the hemodynamics state in AV fistula might be similar to the autologous vein graft stenosis model.
CYR61 plays a role in cell proliferation, differentiation, angiogenesis, apoptosis, and extracellular matrix formation. The knockdown of CYR61 suppresses intimal hyperplasia in a rat artery balloon injury model and regulates the proliferation of VSMCs (Lee et al., 2007; Matsumae et al., 2008). The expression of CYR61 increased in our study, which might enhance cell proliferation and intimal hyperplasia in AV fistula.
CXCL12/CXCR4 chemokine ligand/receptor axis mediates the mobilization of smooth muscle cell (SMC) progenitors, driving injury-induced neointimal hyperplasia. Endothelial CXCR4 deficiency significantly increased wire injury-induced neointima formation in carotid arteries (Noels et al., 2014). The inhibition of CXCR4 could effectively prevent neointima formation (Hamesch et al., 2012). Our dataset showed an augmented expression of CXCR4, which might activate CXCL12/CXCR4 chemokine ligand and subsequently result in neointimal hyperplasia in AV fistula. The inhibition of CCL4 could attenuate the expression of E-selectin, vascular cell adhesion molecule-1 (VCAM-1), and intercellular adhesion molecule-1 (ICAM-1) (Chang et al., 2020) and further attenuate intimal hyperplasia and reduce inflammation (Yasukawa et al., 1997; Gotoh et al., 2004). Additionally, the G-protein-coupled receptor (CCR5) is activated by the inflammatory chemokines CCL3, CCL4, and CCL5, but only CCL4 exhibits selectivity for CCR5, which stimulates intimal hyperplasia in vessels (Maguire et al., 2014). Therefore, the high expression of CCL4 in AV fistula might potentially activate CCR5 or promote the expression of E-selectin, VCAM-1, and ICAM-1 and subsequently contributes to intimal hyperplasia. The activation of ATF3, which is involved in the complex process of the cellular stress response, is a pivotal target of neointima formation and VSMC proliferation in response to wire injury and balloon injury in vascular cells (Huang et al., 2014).
DUSP1 is a dual-specific phosphatase that regulates the MAPK pathway which controlled a vast of cellular processes. The endothelial exposure of high shear stress induces the persistent expression of DUSP1 and mitigates the activities of p38 MAPK and JNK pathway (Zakkar et al., 2008). Recent studies have revealed that the attenuation of ERK induced by DUSP1 inhibits intimal hyperplasia by reducing proliferation and inducing apoptosis of SMCs (Kim et al., 2011).
As mentioned earlier, each of 10 hub genes (EGR1, NR4A1, ATF3, NR4A2, EGR2, DUSP1, EGR3, CXCR4, CCL4, and CYR61) plays a very important role in intimal hyperplasia which is the culprit of AV fistula stenosis causing the loss of hemodialysis access in the patients with end-stage renal disease. Although the molecular mechanisms and pathways between 10 hub genes and intimal hyperplasia are diverse, those 10 genes might provide us various effective targets to suppress intimal hyperplasia with the design of gene-therapeutic drug-eluted balloons or screening specific medicine to prevent AV fistula stenosis in the future study.
Furthermore, GSEA showed that the biological processes and pathways in AV fistula with the high expression of DUSP1 were primarily involved with MAPK signaling pathway, i.e., MAPK signaling-mediated pathway such as TGF-beta receptor signaling, regulation of ERK1 and ERK2 cascade, and JAK/STAT signaling (Meyer and Levine, 2014; Zhang et al., 2014; Li et al., 2019; Hawke et al., 2020), and others involved with cell apoptosis. However, the biological processes and pathways in AV fistula with the low expression of DUSP1 were majorly relevant to cell proliferation such as DNA replication, mitotic nuclear division, and inflammation containing the response to interferon-γ, immunity, ABC transporters, and cell adhesion. As discussed previously, DUSP1 mediated the phosphorylation process in negative regulation of MAPK signaling pathway and ERK cascade such as ERK1 and ERK2 as the downstream of MAPK signaling pathway (Abraham and Clark, 2006). TGF-beta has been shown to induce the expression of DUSP1, which leads to the inactivation of JNK or p38 MAPK signaling pathway (Xiao et al., 2002; Jono et al., 2003; Mikami et al., 2006). The inhibition of MAPK signaling pathway could suppress cell proliferation and intimal hyperplasia (Ge et al., 2013; Qu et al., 2020; Yang et al., 2020). In our study, the biological processes and pathways in AV fistula with the low expression of DUSP1 were partly concentrated on cell proliferation, indicating the probable promotion of cell proliferation by the p38 MAPK pathway. Based on the data mentioned above, we can speculate that DUSP1 is a crucial gene in AV fistula and potentially participates in cell proliferation and intimal hyperplasia.
NR4A1 also belongs to the MAPK signaling pathway. GSEA screened several biological processes and pathways in patients with the high expression of NR4A1 in AV fistula, such as MAPK signaling and TGF-beta signaling, ErbB signaling, regulation of ERK1 and ERK2 cascade, all of which are mediated by MAPK signaling pathways (Nie and Chang, 2007; Li et al., 2019; Hawke et al., 2020). JAK-mediated signaling may partly rely on the activation of the MAPK signaling axes (Meyer and Levine, 2014). But the biological processes and pathways in patients with the low expression of NR4A1 were mainly involved with inflammation such as cell adhesion, ECM–receptor interaction, and immune network. The role of NR4A1 in the regulation of vascular remodeling by the MAPK signaling pathway remains unknown. ERK1/2 as the downstream of MAPK signaling pathway could upregulate NR4A1 expression, and MAPK phosphatase-1 could negatively regulate the induction of NR4A1 dependent on ERK1/2 (Mori Sequeiros Garcia et al., 2015). In macrophages, the p38 MAPK signaling pathway could promote the expression of NR4A1 and subsequently protects against inflammation (Shao et al., 2010). The activation of NR4A1 could diminish the NF-κB activation partly dependent on p38 MAPK activity. Furthermore, the activation of NF-κB in endothelial cells importantly contributes to inflammation (Baldwin, 1996). Therefore, we assumed that NR4A1 potentially participates in the MAPK signaling pathway and negatively regulates inflammation in AV fistula. Particularly, we found that the biological process in AV fistula with the lower expression of NR4A1 was majorly relevant to inflammation, which is consistent with our assumption.
Based on the microarray data of GSE 39488 (Hashimoto et al., 2013), Hashimoto et al. had reported several significant biological processes and pathways. Those reported biological processes in AV fistula, such as biological developmental processes and glycosaminoglycan binding, were different from our enrichment results such as cell adhesion, signal transduction, inflammatory response, apoptotic process. Our result tends to be much more relevant to intimal hyperplasia. The major enriched pathways in their study are the TGF-beta signaling pathway and the cytokines and inflammatory response pathway, and they were similar to our results. Furthermore, their study absents some pathways contained in our results such as MAPK signaling, NOD-like signaling, ABC transporters, and cell cycle. This difference may be attributed to different approaches. The enrichment analysis of pathway in their study was performed just based on identified DEGs, but this method is not sensitive enough to detect the subtle differences between the expression of individual genes (Mootha et al., 2003). GSEA was developed to focus on the changes of expression in a priori defined gene sets and resolves the problem of the undetectable, small changes in the expression of single genes (Mootha et al., 2003). We combined the ordinary enrichment analysis of KEGG pathway with GSEA to screen significant pathways. Additionally, their study did not identify any gene crucial for regulating biological processes in the outflow tract of AV fistula comparing with the normal vein. The best advantage of our study was the identification of 10 hub genes that potentially participate in intimal hyperplasia and 2 core genes that belong to the MAPK pathway. With the analysis of the GSE39488 dataset, the study from Jie et al. (2020) already found one pathway, i.e., NF-κB, and two hub genes that belong to NF-κB, and it is good for exploring the therapeutic management of AV fistula. However, the molecular mechanism of intimal hyperplasia varies; our study tries to explore the potential molecular with the same dataset (i.e., GSE39488) by a different procedure and cutoff value and could provide a new insight into future experimental validation. In addition, the cutoff value in the study by Jie et al. (2020) is too strict, and only 45 selected DEGs were used for the establishment of PPI networks resulting in a selection bias of the identification of hub gene. Furthermore, GSEA between the higher and lower expression of the DUSP1 and NR4A1 in our study helps us know what biological processes and pathways the two crucial genes might regulate, but the study from Jie et al. (2020) lacks this procedure.
Our study provided more information about the alteration of gene expression and biological processes in AV fistula. Future studies will try to validate whether the activation of DUSP1 could attenuate cell proliferation, whether NR4A1 possesses the ability to downregulate the inflammation level to inhibit intimal hyperplasia in vitro and in vivo, and how to participate in the activity of MAPK signaling pathway. However, some limitations were still presented in this bioinformatics study. First, although bioinformatics analysis was useful for an initial screen of crucial genes that participate in intimal hyperplasia, we cannot get any valid conclusion before sufficient validation. Moreover, how the crucial genes DUSP1 and NR4A1 to participate in the MAPK signaling pathway had not been verified in an experimental model to recognize the real activity in AV fistula. Second, the sample size of this dataset was relatively small. The acquisition of venous tissue after fistula is really difficult. The clinician could obtain the venous tissue of AV fistula only when this AV fistula is abandoned or abnormal and the patient has the opportunity to create the next new AV fistula at the ipsilateral limb. This dataset (i.e., GSE39488) was the biggest microarray data focused on human AV fistula tissue in the GEO database. Although the identification of DEGs and GSEA could not reflect the real activity in overflow AV fistula, this study could still provide the knowledge of gene expression in intimal hyperplasia under the circumstance lacking adequate evidence in humans, particularly in patients with uremia. Much more gene expression analysis studies with large sample sizes in humans are urgent to provide more strong evidence. Besides, experimental validation is required to know the potential role of DUSP1 and NR4A1 in intimal hyperplasia.