Mortalin caused angiogenesis and sorafenib resistance in HCC cells. Firstly, we evaluated effects of mortalin on the angiogenic abilities in three human HCC cell lines, HepG2, Hep3B, and HuH7SR. Here, the tube formation ability in mortalin-knockdown cells were significantly reduced compared with MOCK group, while forced expression of mortalin exhibited the opposite phenomenon (Figs. 1A and 1B, and additional files, Fig. S1). Next, we further investigated the effects of mortalin on sorafenib resistance in the above-mentioned three HCC cell lines. Similarly, knockdown of mortalin decreased the efficiency of sorafenib, while overexpression of mortalin caused the opposite effects. The IC50s of sorafenib in MOCK, mortalin-knockdown, and mortalin-overexpression groups were HepG2 (7.682 vs. 4.214 vs. 12.33), Hep3B (6.362 vs. 2.927 vs. 12.6), and HuH7SR (20.14 vs. 10.29 vs. 26.06) (Fig. 1C). Collectively, these results indicated that mortalin might be an effective factor which could improve the angiogenic abilities and contribute to sorafenib resistance in HCC cells. Moreover, mortalin might cause the above-mentioned effects in p53-dependent and p53-independent manners.
Mortalin regulated phosphorylated modification of cancer associated proteome in HCC cells.
In addition to the classic role of mortalin-mediated inactivation of p53, it also improves cancer cell survival and enhances tumour progression (especially in drug-resistant tumours) via other ways like the synergistic effects with human telomerase reverse transcriptase and heterogeneous nuclear ribonucleoprotein K [17]. Here, we firstly employed a search tool for the retrieval of interacting genes (STRING) database to further predict the mortalin-regulated potential downstream signal transduction mechanisms. Mortalin was taken as the central molecule, and then constructed a protein-protein interaction via the generation of 20 most frequently altered neighbor interactors and 30 indirect interactors around it (additional files, Fig. S2). Next, we employed the database for annotation, visualization and integrated discovery (DAVID) to conduct KEGG pathway analysis based on the above-mentioned 50 factors. The enrichment results showed 8 pathways related to the functions of mortalin identified and the oxidative phosphorylation was ranked the first (Fig. 2A). So, to further clarify the effects of mortalin on proteomics phosphorylation modification, a phospho-antibody microarray was used in MOCK and mortalin-knockdown HuH7SR cells. We identified a protein spectrum with whose phosphorylation levels were more than 1.5 fold changed (50% increased or 33% decreased) compared with MOCK group (Fig. 2B). KEGG enrichment analysis of the array indicated that, the PI-3K/Akt pathway was ranked the first (Fig. 2C and additional files, Fig. S3). Furthermore, many of these phosphorylated proteins are vital to tumor progression, and knockdown of mortalin resulted in the extensive suppression of phosphorylation of proteome. As shown in Fig. 2D, the phosphorylation of VEGFR, IGFR, FGFR and EGFR, the receptors with extracellular signal transduction function declined obviously. The results also revealed a reduction in phosphorylation of many key components crucial for PI-3K/Akt (PI-3K-p85, AKT1), MAPK (MEK1, MKK3/6), NF-κB (IKKβ, IκBα, p65, p105/50), JAK-STAT (JAK1/2, STAT1), et al. Meanwhile, several factors involved in the negative regulation exhibited the opposite trend. Collectively, these results indicated that, mortalin could regulate the phosphorylated modification of cancer associated proteome, and that, the PI-3K/Akt might be a key downstream pathway regulated by mortalin in high angiogenic activity and sorafenib-resistance HCC cells.
Identification of PI-3K/Akt as an important downstream factor, regulating VEGF and GSK3/β-catenin in HCC cells
Due to the PI-3K/Akt pathway ranked the first in the KEGG enrichment, we further investigated the underlying mechanisms via GO analysis based on proteins enriched in PI-3K/Akt signaling pathway. As shown in Fig. 3A, the top biological process was negative regulation of apoptotic. Next, we analyzed the phosphoproteins with the most significant changes in PI-3K/Akt signaling pathway, and listed those downregulated or upregulated at least 2-fold compared with the MOCK group. Here in mortalin-KD cells, the phosphorylation of AKT1, VEGFR2, and Bcl-XL, et al were significantly decreased, while the phosphorylation of p27kiP1and GSK3β, et al were significantly decreased (Fig. 3B). Then, we further investigated the differential phosphorylation protein sites enriched in VEGF/VEGFR, GSK3/β-catenin and anti-apoptosis signing, and confirmed that, the above-mentioned three signal pathways were all inactivation in in mortalin-KD cells (Fig. 3C). Based on these results, we speculated that, the PI-3K/Akt-regulated VEGF/VEGFR and β-catenin/anti-apoptosis, were the two mainly neoplastic biological processes involved in mortalin-caused angiogenesis and sorafenib resistance in HCC.
Verification of microarray and bioinformatics results in vitro and in vivo
Verifying the above-mentioned hypothesis, we firstly determined the functions of mortalin in regulating PI-3K/Akt, VEGF/VEGFR, and β-catenin/anti-apoptosis pathways, as determined by the levels/phosphorylation of PI3K-p85 (Tyr458), Akt (Ser473), VEGFR2 (Tyr951), GSK3β (Tyr216), β-catenin (Thr41/Ser 45), Bcl-XL, and apoptosis. As shown in Figs. 4A and 4B, knockdown of mortalin decreased the phosphorylation (activation) of PI3K/Akt/VEGFR2; but increased the phosphorylation of GSK3β, which in turn inactivated the β-catenin/Bcl-XL, and enhanced the apoptosis. However, forced expression of mortalin in HepG2 and Hep3B cells showed the opposite effects. Moreover, knockdown of mortalin attenuated the secretion of VEGF (Fig. 4C). Next, we performed in vivo experiments to further confirm our results. As shown in Figs. 4D and 4E, treating the xenografts with sorafenib alone mildly inhibited tumor growth and angiogenesis. However, combining with knockdown of mortalin facilitated the sorafenib-caused inhibition of tumor growth and angiogenesis, and enhanced the apoptosis. Western blot also confirmed the inhibition of PI-3K/Akt, VEGFR, and β-catenin/anti-apoptosis pathways in mortalin-KD group (Fig. 4F). Collectively, these results suggested that, knockdown of mortalin blocked the PI-3K/Akt, which in-turn inactivated the VEGF/VEGFR and β-catenin/anti-apoptosis signal pathway, leading to the attenuation of angiogenesis and sorafenib resistance in HCC cells.
The clinical significance of mortalin in HCC
We then evaluated the expression of mortalin in HCC patients. As shown in Figs. 5A and 5B, compared with adjacent non-tumor liver tissues, a considerable elevation of mortalin was observed in HCC tissues. We then investigated the association between mortalin level and angiogenesis. The HCC specimens were divided into “mortalin-low” vs. “mortalin-high” groups according to the IHC-Q-Scores. The number of microvessels in mortalin-high HCC tissues was significantly more than those in mortalin-low tissues (Fig. 5C). Moreover, a significant positive correlation was found between mortalin levels and number of microvessels (Fig. 5D). We also found that there was a positive correlation between mortalin and 14-3-3η (an angiogenesis and sorafenib resistance inducer, which was confirmed by our previous study [14, 15]) (Fig. 5E). Finally, Kaplan-Meier survival analysis showed that HCC patients in “mortalin-high” group had the worse overall survival and recurrence-free survival than those in “mortalin-low” group (Fig. 5F). The survival analysis of 34 advanced recurrent HCC patients (received combined sorafenib treatment and transarterial chemoembolization therapy) exhibited the same trend (Fig. 5G). These results indicated that, mortalin might be regarded as a potential risk factor of HCC, predicting poor prognosis and sorafenib resistance.
Identified a novel potential chemical inhibitor for mortalin: CaA
MKT-077 (C21H22ClN3OS2) is a classic chemical inhibitor of mortalin, functions via abolishing mortalin-p53 interactions, but did not change the expression of mortalin [18]. Previous studies revealed that, caffeic acid phenethyl ester (CAPE, C17H16O4, a specific inhibitor of NF-κB activation) was also able to induce disruption of mortalin-p53 complexes, leading to nuclear translocation and activation of p53, and accompanied with a decrease in the expression of mortalin via transcriptional inactivation [19]. However, the effects of these chemicals on the stability of mortalin protein, remain unclear. Our previous study revealed that, as the same as CAPE, CaA was also an effective NF-κB inhibitor [20], and that NF-κB was an up-stream transcriptional regulator of mortalin [21]. Here, CaA decreased the expression of mortalin mRNA in HepG2, Hep3B, and HuH7SR cells (additional files, Fig. S4). Moreover, via a computer docking (SYBYL-X software), we found that the basic skeleton structure of CAPE, CaA (C9H8O4) could bind to mortalin, and the binding region of CaA and mortalin was the same as MKT-077 (Fig. 6A). Thus we adopted a novel approach by combining both IP and LC/MS techniques to reveal the interaction between CaA and mortalin. The mortalin was immunoprecipitated with its specific antibody and its CaA-binding status was determined by LC/MS to investigate if the immunoprecipitation complex contained CaA. As shown in Fig. 6B, the retention time of CaA was 3.29 min, and the area of CaA in positive control group (total protein, supernatant) was 9.6 × 105. The area of CaA in immunoprecipitated complex (experimental group, sample) was 7.5 × 105. We almost did not detect the presence of CaA in negative control groups (lysis buffer and residual supernatant). Moreover, CaA treatment enhanced the ubiquitination of mortalin in HuH7SR cells (Fig. 6C), decreased the expression/phosphorylation of mortalin, PI3K-p85 (Tyr458), Akt (Ser473), VEGFR2 (Tyr951), β-catenin (Thr41/Ser 45), and Bcl-XL (Fig. 6D). Finally, we used a previously established MHCC97H cells xenografts model, which was treated by CaA [22, 23], to confirm the effects of CaA on the expression/phosphorylation of the above-mentioned factors in vivo. As shown in Fig. 6E, CaA treatment inhibited the mortalin, PI-3K/Akt, VEGFR2, and the β-catenin/Bcl-XL. Collectively, these results indicated that, CaA inhibited the mortalin at both transcriptional (indirectly, might via blocking NF-κB) and post-transcriptional (directly, might via targeting and inducing the ubiquitination and degradation of mortalin) modifications in HCC cells.