The Recurrent Liver MAN2A1-FER Oncoprotein Lacks Kinase Activity: Implications for the Use of Tyrosine Kinase Inhibitors

The MAN2A1-FER fusion gene frequently is found in esophageal adenocarcinoma and hepatocellular carcinoma.¹ FER is a ubiquitously expressed nonreceptor tyrosine kinase, closely related to FES and to the product of the v-fes and v-fps viral oncogenes, responsible for feline and avian sarcomas, respectively. These 2 oncogenes have the property to transform fibroblasts and induce tumor formation in mice.^2,3 However, until recently, there was a lack of evidence supporting their role as oncogenes in human beings.⁴ The recent discovery of the recurrent chromosomal rearrangement leading to the MAN2A1-FER fusion is a major finding.⁵ Although only a limited number of fusion genes have been characterized in liver cancers, MAN2A1-FER was identified in 15.7% of hepatocellular carcinomas. In addition, it is present in 26% of esophageal adenocarcinoma, 16.8% of non-small-cell lung cancer, 7.1% of glioblastoma multiforme, 5.2% of prostate cancer, and 1.7% of ovarian cancer.¹

The mannosidase alpha class 2A member 1 (MAN2A1-FER) oncoprotein increased the proliferation and invasiveness of cancer cell lines, and induced spontaneous liver cancer in a PTEN-deficient mouse model.¹ In addition, the study suggested that the oncogenic mechanism of action of the fusion protein was associated with increased FES related (FER) catalytic activity. Given the sensitivity of FER to clinical-grade Anaplastic lymphoma kinase (ALK) and MET inhibitors, MAN2A1-FER fusion emerged as an attractive therapeutic target for the treatment of liver cancer and other malignancies.^6,7

MAN2A1-FER arises from chromosomal rearrangements on chromosome 5, bringing together the first 13 exons of the MAN2A1 gene, with the 3’ portion of FER encoding for the tyrosine kinase catalytic domain. The rearrangements always occur within the same introns; hence, all resulting MAN2A1-FER fusion proteins share the same amino acid composition. At the protein level, the oncoprotein merges the N-terminus of MAN2A1 with the C-terminal amino acids 572–822 of FER (Figure 1A). Intriguingly, MAN2A1-FER lacks the first portion of the kinase domain that structures the adenosine triphosphate (ATP) binding lobe, including the first 2 residues of the conserved GXGXXG glycine-rich loop, one of the most highly conserved sequence motifs in protein kinases (Figure 1B and Supplementary Figure 1). Indeed, all protein kinases share invariable residues required for ATP binding.⁸ Therefore, it was surprising that MAN2A1-FER showed increased kinase activity compared with the FER protein. To clarify this issue, we obtained the expression construct of MAN2A1-FER from the authors of the original publication¹ and we analyzed the kinase activity of this fusion protein. The autophosphorylation of MAN2A1-FER (Figure 1C), and the phosphorylation of cortactin, a FER substrate (Figure 1D), were assessed. In both assays, there was no evidence for kinase activity of MAN2A1-FER oncoprotein in contrast to the wild-type (WT) FER. In addition, we generated a kinase-dead (KD) variant of MAN2A1-FER through in vitro mutagenesis of the conserved K591 residue in the ATP-binding domain of FER, which is essential for ATP binding and, consequently, kinase activity.⁹ A comparative analysis of the WT MAN2A1-FER and KD version of MAN2A1-FER through immunoblotting showed the same absence of phosphorylation signals (Figure 1C and D). This strongly suggested that MAN2A1-FER is a catalytically inactive kinase.

Figure 1.

The MAN2A1-FER oncoprotein is devoid of kinase activity. (A) Schematic representation of MAN2A1 and FER domains are shown above and the MAN2A1-FER fusion protein is shown below. Amino acids VILGELLGK of the FER kinase domain are excluded in the MAN2A1-FER fusion. (B) Alignment of protein sequences showing the conserved GXGXXG motif (in red) and invariant lysine (K in bold) of the ATP binding lobe of tyrosine kinases. (C) Autophosphorylation of WT FER, MAN2A1-FER, and MAN2A1-FER-KD proteins. Immunoprecipitated FER proteins were probed using an antiphosphotyrosine antibody (p-Tyr). FER and MAN2A1-FER protein expression are shown (Flag) and tubulin was used as a control. (D) Phosphorylation of cortactin substrate. Flag-tagged cortactin was co-expressed with FER, MAN2A1-FER, or MAN2A1-FER-KD proteins as indicated. After immunoprecipitation, phosphorylated proteins were detected using an antiphosphotyrosine antibody (top panel), and all transfected proteins using an anti-Flag antibody (bottom panel). (E) In vitro kinase assay on MAN2A1-FER. Immunoprecipitated FER and MAN2A1-FER variants were incubated with poly-Glu-Tyr peptide as a substate. Kinase activity was quantified using the ADP-Glo Kinase Assay kit (Promega). Statistical analysis was performed using analysis of variance. (F) Colony formation assay using 3T3 cells transfected with control plasmid (pcDNA3), MAN2A1-FER, or MAN2A1-FER-KD. Colonies were monitored after 10 days. Left: Representative pictures of the plates showing colonies. Right: Quantification of the number of colonies. Statistical analysis was performed using analysis of variance. FBAR, Fes/Cip4 homology Bin/amphiphysin/Rvs; FX, F-BAR extension; IP, immunoprecipitation; SH2, Src Homology 2.

To definitively conclude, we next performed a quantitative in vitro kinase assay using poly-Glu-Tyr peptides as substrate, as performed in the previous study. WT FER was used as a positive control, whereas the kinase-deficient KD forms of FER and MAN2A1-FER were evaluated as negative controls. As depicted in Figure 1E, MAN2A1-FER did not show kinase activity, mirroring the kinase-deficient controls.

Finally, we also repeated the colony-formation assay used by Chen et al¹ to assess the enhanced clonogenicity induced by the MAN2A1-FER oncoprotein. Consistent with their findings, MAN2A1-FER oncoprotein conferred a 2-fold increased colony formation capacity to the transfected cells (Figure 1F). Remarkably, this feature was not dependent on the kinase activity as shown by the KD mutant (Figure 1F). We concluded that MAN2A1-FER is a catalytically deficient tyrosine kinase.

The initial published characterization of MAN2A1-FER oncoprotein proposed that the oncogenic activity of the fusion protein was linked to FER catalytic activity. As a consequence, MAN2A1-FER now is considered a target for tyrosine kinase inhibitor–based therapy.^7,8 This conclusion was supported by in vitro kinase assays and the use of crizotinib, an inhibitor of the ALK, MET, ROS Proto-oncogene 1 (ROS1), and récepteur d’origine nantais (RON) receptor, which also target many other protein kinases including FER. This mechanism, however, was surprising given the absence of the first 9 amino acids of the N-terminal lobe of FER catalytic domain in the fusion protein, which included a conserved amino acid residue, part of an invariant ATP-binding motif. Our results show that the MAN2A1-FER oncoprotein is devoid of kinase activity. We thus argue that the oncogenic feature of MAN2A1-FER fusion is independent of the kinase activity. This issue is of crucial importance for basic, preclinical, and clinical research because small molecules that target FER are available, but those would have no direct inhibitory effects on the oncoprotein. It should be noted, however, that kinase inhibitors still may have antitumor activity owing to the off-target effects of these molecules on other protein kinases. Alternatively, proteolysis targeting chimera (PROTAC) methods that aim to target FER have been developed successfully.¹⁰ Finally, our findings also imply that the MAN2A1 portion, responsible for localization of the fusion protein to the Golgi apparatus, may be responsible for the oncogenic properties of the fusion protein rather than the FER catalytic activity.

In conclusion, our data strongly support that MAN2A1-FER lacks kinase activity and, therefore, it cannot be targeted by small-molecule kinase inhibitors.