eIF3a is associated with vemurafenib sensitivity
To explore the potential function of eIF3a in cancer therapy, gene set enrichment analysis (GSEA) was used for pathway enrichment analysis. KEGG analysis showed that the KRAS signaling pathway was significantly enriched after knockdown of eIF3a (Figure 1a). It has been reported that reactivation of RAS is associated with vemurafenib resistance [27]. Based on this viewpoint, we hypothesize that eIF3a may be involved in vemurafenib resistance. To test this hypothesis, we firstly compared the level of eIF3a in A375 and A375R cells, which is less responsive to vemurafenib than A375 cells (Figure 1b), and found that there is a significant downregulation of eIF3a in A375R cells (Figure 1c). We further found that knockdown of eIF3a by siRNA decreased the responses of human melanoma cells to vemurafenib, as demonstrated by CCK8 assay and colony formation assay (Figure 1d-e). Furthermore, overexpression of eIF3a rendered A375R cells more sensitive to vemurafenib (Fig. 1f). To further support the idea that eIF3a is associated with vemurafenib sensitivity, A375 cells were transfected with different amounts of siRNA, followed by vemurafenib treatment. Fig.1g-h showed that the level of eIF3a was positively correlated with the sensitivity of vemurafenib. These data suggest that the reduction of eIF3a in melanoma may result in vemurafenib resistance.
In addition, we also evaluated the effect of eIF3a on the anti-tumor activity of dabrafenib, another BRAF inhibitor approved by the FDA to treat melanoma [28]. Consistent with the results obtained with vemurafenib, silencing of eIF3a also reduced the cytotoxicity of dabrafenib (Figure 2a-b), and the expression of eIF3a was positively correlated with the toxicity of dabrafenib (Figure 2c-d). These experiments demonstrated the involvement of eIF3a in regulating the sensitivity of dabrafenib in the human melanoma cells.
eIF3a affects the sensitivity of vemurafenib by activating ERK signaling
Next, we investigated the molecular mechanism involved in the regulation of eIF3a in sensitivity to vemurafenib. Figure 1a shows that inhibiting eIF3a actives the RAS signaling pathway. It’s well known that ERK is the classical downstream of RAS signaling pathway, and the aberrant activation of ERK plays a central role in vemurafenib resistance [29]. Consistently, we also verify that the A375R cells harbored higher p-ERK and lower eIF3a compared with A375 cells (Figure 3a). Therefore, we next determined whether eIF3a regulates the sensitivity of vemurafenib by modulating the activity of ERK. As expected, inhibiting eIF3a by either siRNA or L-mimosine, a small molecule inhibitor of eIF3a, significantly increased the phosphorylation of ERK in melanoma cells (Figure 3b-c). Figure 3d-e further demonstrated the negative correlation between the expressions of eIF3a and p-ERK. Moreover, the regulation of eIF3a on p-ERK was also observed in non-small cell lung cancer and breast cancer cell lines (Figure S1). The ectopic expression of eIF3a resulted in a reduction in phosphorylation of ERK (Figure 3f). To further validate the effect of eIF3a on the activation of ERK, we re-introduced eIF3a into eIF3a knockdown cells by transfecting an eIF3a overexpression plasmid, and then measured p-ERK. As shown in figure 3g, the p-ERK was increased in the cells when eIF3a expression was knocked down, and eIF3a overexpression blocked the up-regulation of p-ERK in the cells subjected to silencing of eIF3a expression. Furthermore, knockdown of eIF3a in A375 cells could mitigate the suppression effects of vemurafenib on p-ERK, which revealed that vemurafenib-induced MAP kinase signaling blocking was attenuated by silencing eIF3a (Figure 3h). And by combination with an ERK inhibitor, the insensitivity to vemurafenib after knockdown of eIF3a was overcame (Figure 3i-j). These results indicated that the loss of eIF3a contributed to the activation of ERK, in turn conferring BRAF inhibitor resistance.
eIF3a suppresses ERK activity through upregulating the expression of PPP2R1B
We next sought to investigate the molecular mechanism underlying eIF3a knockdown-mediated the activation of ERK. It has been reported that the MAPK pathway can be dephosphorylated by numerous protein phosphatases, for example, reduction of dual specificity phosphatase 4 (DUSP4) and dual specificity phosphatase 6 (DUSP6) results in activation of the MAP kinase pathway [30-32]. Therefore, we wondered whether there was a phosphatase involved in regulating eIF3a knockdown-mediated ERK activation. In order to assess this hypothesis, we firstly measured the expressions of proteins phosphatases after silencing eIF3a by mass spectrometry. Among those examined, PPP2R1B, a regulatory subunit of the PP2A complex [33], was decreased in eIF3a-knockdowned cells compared with that in control cells (Figure 4a).
Consistent with the results of proteomics, knockdown or inhibition of eIF3a by siRNA or small-molecule inhibitor caused a decrease in PPP2R1B and an increase in p-ERK (Figure 4b-c), and overexpression of eIF3a resulted in PPP2R1B elevation (Figure 4d). Figure 4e further demonstrated that eIF3a could regulate the expressions of PPP2R1B. In addition, the regulation of eIF3a on PPP2R1B protein levels was also determined in non-small cell lung cancer cells (Figure S2). Next, we examined whether there is a regulatory effect between PPP2R1B and ERK. As shown in figure 4f, silencing of PPP2R1B caused a significant increase in p-ERK, and overexpression of PPP2R1B induced a remarkable reduction in p-ERK (Figure 4g). To further determine whether PPP2R1B mediates the regulation of eIF3a in the activity of ERK, we overexpressed PPP2R1B after silencing eIF3a. Figure 4h shows that re-introduction of PPP2R1B reversed the upregulation of p-ERK by eIF3a deficiency.
We further investigated the functional role of PPP2R1B in vemurafenib resistance, and found that PPP2R1B knockdown reduced the sensitivity of vemurafenib in A375 cells, and overexpression of PPP2R1B significantly enhanced the anti-tumor effect of vemurafenib (Figure S3a-b). Furthermore, ectopic expression of PPP2R1B in cells silenced eIF3a partially restored sensitivity to vemurafenib (Figure S3c). These findings support that eIF3a controls the action of ERK though regulating the PPP2R1B expression, thus determining the sensitivity of vemurafenib in melanoma cells.
eIF3a up-regulates the expression of PPP2R1B by promoting its translation
Next, we wanted to explore how eIF3a affects the expression of PP2R1B. We further examined the mRNA level of PPP2R1B after knockdown or overexpression of eIF3a. The results showed that silencing or overexpression of eIF3a did not change the mRNA level of PPP2R1B (Figure 5a-b), indicating that eIF3a regulates PP2R1B at the post-transcription level. Given that eIF3a is a translation regulator, we then examined whether eIF3a controls the expression of PPP2R1B through regulating its translation. Figure 5c indicates that the mRNA of PPP2R1B could bind to the eIF3a protein. In addition, and luciferase assay shows silencing of eIF3a inhibited PPP2R1B translation (Figure 5d). Furthermore, HCQ, an autophagy inhibitor, or MG132, a proteasome inhibitor, had no effect on the down-regulation of PPP2R1B in the cells with eIF3a knockdown (Figure 5e-f). These results indicate that eIF3a could regulate the translation of PPP2R1B.
The association of eIF3a and PPP2R1B is validated in melanoma patient samples
To further validate the association between eIF3a and PPP2R1B in human melanoma patient samples, we first assessed protein expression levels of eIF3a and PPP2R1B by the Human Protein Atlas (https://www.proteinatlas.org/). The results showed that the expression of eIF3a was positively related to that of PPP2R1B in melanoma patients (Figure 6a). We further analyzed the correlation between eIF3a and PPP2R1B by TCGA database. Consistently, the strong positive correlation between eIF3a and PPP2R1B was observed in melanoma, lung cancer and breast cancer (Figure 6b-d). In addition, we also analyzed the relationship between eIF3a and PPP2R1B in primary, metastasis and uveal melanoma by using the TIMER web server (https://cistrome.shinyapps.io/timer/), and found the expression of eIF3a was positively associated with PPP2R1B (Figure 6e).