Molecular Pathways: The eIF4F Translation Initiation Complex—New Opportunities for Cancer Treatment

Among the different steps in gene expression, cytoplasmic mRNA translation is an essential process that leads to protein synthesis. Although global translation rates are generally higher in cancer cells, it is now acknowledged that subsets of mRNAs are specifically regulated at the translation level. Excellent reviews have recently been published on the role of translation in cancer (1–5). Here, we focus on the eIF4F complex, its role in chemoresistance, and its targeting with small-molecule inhibitors.

The interaction between eIF4F and the 7-methylguanosine “cap” (m⁷G) located at the 5′ end of all mRNAs is critical to directly recruit the 40S ribosomal subunit to mRNAs through a set of protein–protein interactions and to unwind RNA secondary structures located in the 5′ untranslated region (5′UTR) of mRNAs. The eIF4F complex comprises the eIF4E cap-binding protein, the eIF4A DEAD box RNA helicase, and the eIF4G scaffolding protein (Fig. 1A). eIF4A utilizes ATP hydrolysis to unwind and resolve RNA secondary structures. Although ATP hydrolysis is necessary to the unwinding action, it also releases eIF4A from the mRNA, meaning it can use another substrate and, thus, recycle the available eIF4A to increase the rate of translation. Finally, eIF4G is a scaffold protein for the assembly of the eIF4F complex. The activity of the eIF4F complex is tightly controlled by its interaction with several proteins, including the eIF4A-binding proteins eIF4B, eIF4H, and programmed cell death 4 (PDCD4; eIF4B and eIF4H stimulate, whereas PDCD4 inhibits eIF4A); the eIF4E-inhibitory proteins 4EBP1–3; and many eIF4G-interacting proteins (e.g., the poly(A) binding protein PABP).

Not all mRNAs are similarly selected by the eIF4F complex. eIF4E is implicated in the translation of long and highly structured mRNAs. Of these mRNAs, many encode proteins involved in cell-cycle progression, cell growth, or angiogenesis (e.g., MYC, CCDN1, ODC1, VEGF, FGF2) or more generally cancer-related genes (2). The mRNAs that require eIF4A for their translation were characterized using transcriptome-scale ribosome footprinting (6). Such mRNAs, which are limited in number, harbor a particularly long 5′UTR with guanine-rich motifs that form G-quadruplexes, such as the 12-nucleotide (CGG)₄ motif, that form a 4-stranded structure. Importantly, most of these mRNAs encode for oncogenes, transcription factors, epigenetic regulators, and kinases, whereas housekeeping genes do not display G-quadruplexes and do not require eIF4A for their translation.

The eIF4F complex is located at the convergence of several cell signaling pathways involved in carcinogenesis, including the PI(3)K/AKT/mTOR pathway and the RAS/RAF/MEK/ERK/MNK MAPK pathway (Fig. 1B). When phosphorylated by mTORC1, the 4EBP proteins are unable to bind eIF4E, enabling the formation of an effective eIF4E–eIF4G complex. mTORC1 is also responsible for the phosphorylation of the S6K1/2 kinases, which phosphorylate (i) the eIF4A-inhibitory protein PDCD4, relieving the inhibitory activity of PDCD4 on eIF4A, and (ii) eIF4B, allowing it to interact with eIF4A to enhance its helicase activity. In the MAPK pathway, ERK influences the translation via the activation of the RSK kinases that target PDCD4 and S6, independently of the S6K kinases. MNK, downstream of ERK, controls the phosphorylation of eIF4E on a single site (Ser209) through its interaction with eIF4G. Strong evidence links eIF4E phosphorylation with tumorigenesis, invasion, and metastatic progression in cells and in mouse models (7–10).

In parallel with these phosphorylation events, the expression of the eIF4F complex components is also regulated. For instance, the MYC transcription factor, one of the most frequently activated oncogenes in human cancers, increases the transcription of all genes encoding components of the eIF4F complex (eIF4E, eIF4A, and eIF4G), thereby controlling protein translation. Other transcription factors can also regulate the transcription of the individual components of the translation complex following stimulation by various growth factor pathways (Supplementary Table S1).

The eIF4F complex contributes to many of the hallmarks of cancer, such as sustained proliferative signaling, evasion of growth suppression, resistance to programmed cell death, replicative immortality, angiogenesis, invasion, and metastasis. Each of the individual components of the complex has been described as a prognostic indicator. Expression levels of the eIF4F complex components and their inhibitors as well as phosphorylation can be linked with the aggressiveness of histological subtypes of cancers, poor disease outcome and survival, and response to treatment (Supplementary Table S2).

During the last decade, it has been demonstrated that the activity of the eIF4F complex contributes to drive resistance to many types of therapies used as treatment in cancer. One of the first examples was shown in Eμ-MYC hematopoietic stem cells transfected with retroviral vectors expressing eIF4E. Lymphoma cells overexpressing the cap-binding protein are highly resistant to the DNA-damaging agent doxorubicin compared with controls (11), and this observation has since been extended to other types of therapies. Knockdown of eIF4E results in enhanced chemosensitivity to cisplatin and antimitotic microtubule stabilizers (e.g., paclitaxel, docetaxel) in triple-negative breast cancer cells (12). In addition, increased expression of miR141, which targets eIF4E, has also been observed in an acquired model of docetaxel resistance in breast cancer (13).

eIF4E overexpression or amplification also promotes resistance to PI(3)K/AKT/mTOR inhibitors (e.g., AZD8055, BEZ235) in immortalized mammary epithelial cells or colon cancer cells (14, 15), and ectopic expression of eIF4E leads to resistance to inhibitors of receptor tyrosine kinases (e.g., trastuzumab, cetuximab, erlotinib) in breast cancer xenografts (16).

Furthermore, phosphorylation of eIF4E has been implicated in resistance to cisplatin in breast cancer cell lines and immortalized keratinocytes. Interestingly, this resistance to cisplatin is abolished in cancer cells that no longer have an interaction between p-eIF4E and 4E-T, which mediates eIF4E nuclear import, indicating that phosphorylation of eIF4E and its interaction with 4E-T are involved in the tolerance to increased DNA damage (17).

The eIF4A-inhibitory protein PDCD4 can also contribute to chemoresistance. Indeed, reexpression of PDCD4 sensitizes glioblastoma multiforme cells to doxorubicin via Bcl-xL inhibition (18), and, conversely, low PDCD4 expression is associated with resistance to paclitaxel and doxorubicin (19).

eIF4A itself is not directly involved in resistance mechanisms, but deregulation of its activity leads to chemosensitivity in many cancer types, as illustrated in Supplementary Table S1. Inhibition of eIF4A binding to mRNA, of its recycling, or increase of its ATPase activity contributes to sensitization in many murine cancer models and highlights the importance of this initiation factor in this process (Supplementary Table S1). This aspect will be expanded further in the following section.

eIF4A cofactors, eIF4B and eIF4H, are also involved in chemosensitivity. Overexpression of both eIF4H isoforms inhibited caspase activity following cisplatin and etoposide treatment in murine NIH3T3 cells (20). In addition, eIF4B is overexpressed in cisplatin/fluorouracil-resistant gastric tumors (21). Finally, resistance to anti-BRAF and anti-MEK therapies is associated with a prominently active eIF4F complex in a BRAF(V600)-mutated context (22).

The first strategy to decrease eIF4F activity has been to target eIF4E, which is the least abundant factor of the complex. Targeting eIF4E with an antisense oligonucleotide (4EASO) has shown a significant antineoplastic effect, where tumor growth in a prostate xenograft model was suppressed, as was the formation of vessel-like structures, suggesting an additional antiangiogenic effect (23). Clinical trials with this inhibitor produced few adverse effects but no significant clinical response on tumors (24). Therefore, although targeting eIF4E appears to be an attractive treatment, its effect as a single agent, at least using the aforementioned antisense technology, was not effective.

Another strategy to block eIF4E activity is to target the eIF4E–cap interaction. The pronucleotide 4Ei-1 (N-7 benzyl guanosine monophosphate tryptamine phosphoramidate pronucleotide) in combination with nontoxic levels of gemcitabine has been trialed in breast and lung cancer cells, which resulted in chemosensitization of the cell lines (25).

Specifically disrupting the eIF4E–eIF4G interaction has yielded promising results. The first compound used was 4EGI-1, identified by a high-throughput screening assay in 2007 (26). This drug induced apoptotic cell death in several tumor cell lines in vitro (26, 27) and promoted tumor regression of breast or melanoma cancer xenografts in vivo (28), whereas another eIF4E–eIF4G inhibitor, 4E1RCat, promoted tumor-free survival when used in combination with doxorubicin (29).

Three classes of eIF4A inhibitors have been reported so far. Flavaglines, hippuristanol, and pateamine A all originate from natural products that display potent anticancer effects in vivo.

Rocaglamide (flavaglines) was isolated in 1982 from Asian medicinal plants based on their potent antileukemic activities (30). Since then, more than 100 natural flavaglines, such as rocaglaol or silvestrol, have been identified, and many have been shown to display potent anticancer effects in murine cancer models (31, 32). The most studied is silvestrol; unfortunately, this compound shows poor bioavailability coupled with high sensitivity to multidrug resistance (33). Gratifyingly, more drug-like compounds that are insensitive to multidrug resistance displaying enhanced in vivo anticancer activities have been reported. For instance, FL3 was shown to overcome the resistance to BRAF inhibitors in mouse models of metastatic melanoma (22).

Many of the studies listed in Supplementary Table S3 have demonstrated that flavaglines strongly potentiate in vivo the antitumor effects of chemotherapeutic agents, in particular in mouse models of chemoresistant cancers.

Remarkably, flavaglines have also been shown to bind the scaffold proteins prohibitins, blocking their interaction with CRAF, which results in inhibition of the RAS/CRAF/MEK/ERK signaling pathway that is critical to the survival of the cancer cells (34). However, the identification of a drug-resistant and functional eIF4A1 allele that abolishes the cytotoxicity of flavaglines upon introduction into cells using the CRISPR/Cas9 technology suggests that eIF4A is the prime target of flavaglines in most of the cancers (35).

Flavaglines were shown to block eIF4A recycling due to its increased binding to mRNAs (36). The direct interaction with eIF4A was shown using affinity chromatography (37) and chemogenomic profiling in yeast (38). As mentioned in the previous section, mRNAs that require eIF4A for their translation encode for cancer-related proteins. Hence, this observation clarifies why eIF4A inhibitors display a cytotoxicity that is specific to cancer cells. In contrast, it has been shown that flavagline sensitivity is poorly related to the presence of the G-quadruplexes in the 5′UTR but depends strongly on polypurine sequences in these regions (39).

Hippuristanol is a complex polyoxygenated steroid originally isolated in 1981 from coral (40). This compound allosterically inhibits the binding of mRNA to eIF4A (41, 42). Recent biophysical studies using FRET indicate that hippuristanol locks eIF4AI in a closed conformation to inhibit RNA unwinding (43). In vivo studies showed that hippuristanol significantly inhibits the growth of primary effusion lymphoma in xenograft mice (44), suppresses T-cell tumor growth (45), and induces a synergistic response with a Bcl-2 inhibitor (ABT-737), resulting in the induction of apoptosis in lymphoma or leukemia cells (46). Hippuristanol has also been shown to induce cell-cycle arrest and apoptosis in vitro by reducing the expression of cell-cycle regulators (such as cyclin D1/D2, CDK4, and CDK6) or prosurvival factors (such as Bcl-xl; ref. 45). Moreover, it is capable of reversing drug resistance in PI(3)K/AKT/mTOR-dependent tumors (46).

Pateamine A is a complex macrolide that was isolated from a marine sponge in 1991 and demonstrated in vitro cytotoxicity against leukemia cells (47). Pateamine A prevents eIF4A heterodimerization with eIF4G but, surprisingly, enhances the helicase and ATPase activities of eIF4A (48, 49). Exploration of the structural requirements of pateamine A for its pharmacologic activities led to the identification of desmethyl, desamino pateamine A as a structurally simplified analogue that significantly induced tumor regression in two mouse models of melanoma (50).

On the basis of their consistent anticancer activity, eIF4F complex inhibitors should be considered for further clinical development. It will be important to define biomarkers to determine which subgroup of patients will be sensitive to these inhibitors. Some reports are already showing that response to treatment can be predicted using the eIF4F complex, and using prognostic factors combined with newer inhibitors may yield better responses to treatment.

Although targeting eIF4E has shown impressive effects on tumor progression in vivo, its clinical application has to be improved. Combining eIF4E inhibitors with other therapies seems a promising strategy to be tested [phase II trials of 4E-ASO in combination with established chemotherapies are ongoing (NCT01234038 and NCT01234025)]. Furthermore, the use of both in vitro assays and in vivo mouse models are paving the way to develop new combinations of eIF4F inhibitors with validated chemotherapies.

The reviewed studies on flavaglines, hippuristanol, and pateamine A strongly suggest that eIF4A is a valid target in oncology. The promise of these compounds is poised to promote the advancement of derivatives of these natural products toward the clinic. It also highlights the resurgence of natural products in oncology. Indeed, the advent of targeted therapies in the 1990s placed the clinical study of anticancer agents from natural products in limbo for a decade, until it appeared that targeted therapies would not fulfill expectations for many solid tumors. Thus, since 2007, 12 novel natural product derivatives have been approved to treat cancers, indicating that natural products continue to provide valid opportunities to treat unmet medical needs.

No potential conflicts of interest were disclosed.

Conception and design: H. Malka-Mahieu, M. Newman, C. Robert, S. Vagner

Writing, review, and/or revision of the manuscript: H. Malka-Mahieu, M. Newman, L. Désaubry, C. Robert, S. Vagner

Study supervision: S. Vagner

We apologize to all the colleagues who have made contributions in the field and could not be cited owing to space constraints.