Panx1 expression was upregulated in invasive PA samples
We collected 56 surgical PA samples and divided them into invasive and noninvasive groups according to Knosp classification. The expression of Panx1 in the two groups was detected by RT-qPCR, and the results showed that the expression level of Panx1 mRNA was significantly upregulated in the invasive group when compared with the non-invasive group (Fig. 1A). Similarly, IHC results showed that Panx1 protein mainly expressed in the cytoplasm and membrane of tumor cells, and the invasive tumors presented with high staining intensity (Fig. 1B, left panel). Further statistical results confirmed significantly upregulated expression of Panx1 protein in invasive PA samples (Fig. 1B, right panel), which supported an important role for Panx1 in PA invasion.
Panx1-OE promoted the invasion of PA cell lines
To further investigated the role of Panx1 in the invasive PA, we established a model of Panx1-OE PA cells using GH3 and MMQ cell lines, RT-qPCR and western blotting confirmed stable upregulation of Panx1 mRNA and protein in Panx1-OE cells compared with vector cells (Fig. 1C). Subsequently, CCK-8 assays were performed to investigate whether Panx1 impacts the proliferation of GH3 cells. Probenecid (PBN), a specific and selective Panx1 channel inhibitor [11], was added to the Panx1-OE GH3 cells, MMQ cells and Vector. The results demonstrated that neither Panx1 nor PBN could impact the proliferation of PA cell lines (Fig. 1D). Then, transwell assays were performed to investigate whether Panx1 impact the invasion of GH3 and MMQ cells. We found that compared with Vector, Panx1-OE remarkably promoted invasion of GH3 cells and PBN could reverse the effect (Fig. 1E). Likewise, Panx1-OE also promoted the invasion of MMQ cells (Fig. 1F).
Panx1-OE triggered a series of metabolites changes
It has been well established that Panx1 acts as membrane channel, resulting intracellular metabolite release and extracellular metabolite influx. The up-regulated expression of Panx1 mediated changes of intracellular and extracellular metabolites, which may cause dysregulated cellular biological behaviors [12]. To investigate the mechanism by which Panx1 promoted the invasion of PA cells, we used metabolomics analysis to assess differences in metabolites between Panx1-OE GH3 cells and wild-type GH3 cells. We found that levels of various metabolites were significantly increased in Panx1-OE GH3 cells, including adenosine, L-lactate, pyruvic acid, dUMP, L-alanine and glyceraldehyde 3-phosphate, while levels of ATP and uracil were decreased (Fig. 2). Based on KEGG pathway analysis, the differentially expressed metabolites were mainly enriched in 12 pathways with a p value<0.05 and gene count ≥ 3, of which the enrichment ratio of four pathways were at least 12, including the Warburg effect, pentose phosphate pathway and gluconeogenesis (Fig. 3A). In addition, according to KEGG pathway analysis, pathways could be divided into 5 groups including organismal systems (e.g. glucagon signaling pathway, insulin secretion and prolactin signaling pathway), metabolism (e.g. metabolic pathway, glycolysis/gluconeogenesis), human diseases (e.g. type II diabetes mellitus, Parkinson’s disease), genetic information processing (sulfur relay system) and environment information processing (HIF-1, cAMP, and AMPK signaling pathway and ABC transporters) (Fig. 3B).
Panx1-mediated ATP release promoted GH3 cell invasion by activating P2X7R
It has been reported in many cells, including various tumor cells, that Panx1 could promote cell migration by releasing ATP to the extracellular space [13]. According to the results of our metabolomics analysis, which showed downregulation of intracellular ATP levels, we proposed the hypothesis that Panx1 could promote invasion of GH3 cells by releasing ATP to the extracellular matrix. We measured the extracellular ATP concentration and found that Panx1-OE could significantly increase extracellular ATP concentration; nevertheless, when incubated with PBN, extracellular ATP of Panx1-OE GH3 cells decreased obviously (Fig. 4A). Subsequently, we further investigated whether ATP could influence the invasion of GH3 cells. Compared with Panx1-OE GH3 cells group, when added ATPase (MedchemExpress, HY-19808) into Panx1-OE GH3 cells group, the effect of Panx1-OE on invasion of GH3 cells decreased. Moreover, BzATP trithylammonium salt (BzATP), a kind of ATP analogue, was used to activate P2XR in Vector GH3 cells group and exerted an invasive effect on GH3 cell similar to the Panx1-OE group (Fig. 4B). It has been reported that the role of ATP as an extracellular transmitter of cell surface receptors is mediated by ATP-gated ionotype P2XR to purinergic transmission, which is believed to have a wide range of important functions, including activation of intracellular signaling pathways to trigger a series of physiological and pathological changes [14-16]. We then investigated which type of receptor predominantly affected the invasion of GH3 cells. When treated with PPADS tetrasodium (PPADS, broad-spectrum P2XR antagonist), 5-BDBD (P2X4R specific antagonist) and JNJ-47965567 (JNJ, P2X7R specific antagonist), the number of invasive Panx1-OE GH3 cells was decreased, especially in the JNJ-treated group (Fig. 4C, left panel). This was also confirmed by statistical analysis of cell number (Fig. 4C, right panel), which indicated that P2X7R plays a more important role in the promotion of GH3 cell invasion by Panx1.
It has been reported that Panx1 could directly interact with actin-related protein 3 (Arp3) to impact the actin cytoskeleton and invasion of tumor cells [17-18]. Therefore, we detected the relationship between Panx1 and Arp3, and the Co-IP results showed that Panx1 might not straightly interact with Arp3 in GH3 cells (Fig. 4D). Moreover, to further investigate whether Panx1 could influence the expression of Arp3 protein through ATP-P2X7R pathway, we assessed the level of Arp3 protein after adding PBN, JNJ, and BzATP into GH3 cells respectively. The results illustrated that the protein levels of Arp3 did not change regardless of the antagonists or agonists added to GH3 cells (Fig. 4E).
Activated P2X7R facilitated Ca2+ influx and increased MMP-2/9 levels
P2X7R could be activated by ATP and open a pore on the cell membrane for influx of sodium and calcium ions, and efflux of potassium ions, allowing small molecules to pass freely. This could further affect the stability of cell membrane skeleton and fluidity [19]. In our study, we found that antagonists (PBN, JNJ and ATPase) inhibited Ca2+ influx into cytoplasm, while BzATP could trigger Ca2+ influx (Fig. 5A). It is well known that cytosolic Ca2+ is an important secondary messenger and regulates several cell functions via Ca2+ binding proteins. We detected two general Ca2+ binding proteins, secretagogin and calmodulin kinase II (CaMK II). Western blotting results confirmed that expression of secretagogin protein levels in Panx1-OE GH3 cells obviously increased compared with vector groups. PBN, JNJ, and ATPase could reverse this effect in Panx1-OE GH3 cells. Similarly, the expression of secretagogin was significantly upregulated in the vector GH3 cells after addition of BzATP to the medium (Fig 5B). However, the expression of CaMK II was not affected by the above factors (Fig 5B). We further detected the expression of matrix metalloproteinases (MMPs), which have been confirmed to play an important role in PA invasion [20-21]. As a result, the concentration of MMP-2 and MMP-9 proteins was assessed in GH3 cells when overexpressing Panx1, and when incubated in agonists and antagonists, respectively. The results showed that MMP-2 and MMP-9 increased respond to Panx1-OE and P2X7R activation; in contrast, they decreased when PBN, ATPase or JNJ were added (Fig. 5C). These findings suggest that regulation of Ca2+ influx by Panx1 may influence the expression of secretagogin, thereby promoting invasion of GH3 cells.
The Panx1-ATP-P2X7R signaling pathway might promote invasion of GH3 cells by remodeling the actin cytoskeleton
Tumor cell migration is a multi-stage process, in which the actin cytoskeleton participates through actin polymerization or depolymerization, cell adhesion, and myosin contraction. The actin cytoskeleton includes globular actin (G-actin) and actin filaments (F-actin). G-actin is a type of monomeric protein, which requires Ca2+ binding and induces the formation of F-actin, further influencing cell migration capacity [22-24]. To investigate whether the Panx1-ATP-P2X7R signaling pathway could modulate the actin cytoskeletal network, immunofluorescence staining was used, and suggested that actin cytoskeleton of Panx1-OE GH3 cells group was significantly more robust and tight, while Vector group cells presented a moderate and loose actin network. When adding PBN, ATPase, and JNJ to block the Panx1-ATP-P2X7R signal, the effect of Panx1-OE on the actin cytoskeleton could be reversed, while BzATP functioned similarly to Panx1-OE to make actin cytoskeleton become more robust (Fig. 6A). However, the expression of actin protein levels were not affected by Panx1, PBN, JNJ, and BzATP. (Fig. 6B). Reports have shown that cytochalasin B (cyto B) mainly acts on the cytoskeleton, promotes its depolymerization, destroys its structure, and affects cell proliferation, apoptosis, and movement [25-26]. Subsequently, in order to verify whether the status of the actin cytoskeleton related to GH3 cell invasion, cytoB was used to destroy the actin cytoskeletal structure and invasion of GH3 cells was assessed by transwell assay. The results showed that invasion of Panx1-OE GH3 cells decreased sharply when cultured with cyto B (Fig. 6C), indicating the importance of the actin cytoskeleton in the invasion of PA.