We have shown that the putative oncogene eEF1A2 is upregulated in a high proportion of breast tumours. This upregulation is considerably more significant in ER-positive tumours. There is little or no detectable expression of eEF1A2 in normal breast tissue. It is not yet known whether this overexpression results from amplification of the
EEF1A2 gene in all cases; in the study of ovarian tumours by Anand et al [
6] at least one tumour showed overexpression in the absence of gene amplification, suggesting that there are other mechanisms by which the gene can be upregulated. There is a strong association between ER-positivity and eEF1A2 overexpression which is worthy of further study.
There also appears to be a weak correlation between the absence of p53 mutations and eEF1A2 overexpression. It is possible that eEF1A2 is not upregulated in tumours with p53 mutations because wild-type p53 is required for expression of eEF1A2 in certain cell types; it has been shown that p53 can upregulate expression of eEF1A1 [
11], and the p53 binding sites identified in the gene encoding eEF1A1 are shared with that encoding eEF1A2 (unpublished observations). On the other hand it is conceivable that upregulation of eEF1A2 expression rather than p53 mutation is an alternative route for tumours to evade apoptosis in certain cancers.
The basis for the oncogenicity of eEF1A2 is still unclear. We, like Anand et al, have shown that the levels of eEF1A1 in tumours which over-express eEF1A2 are unchanged (data not shown), suggesting that these tumours might have a greater capacity for protein synthesis. However, it has been known for many years that eEF1A is in excess over the other components of the translation elongation apparatus [
12], so eEF1A is unlikely to be rate-limiting in protein synthesis. eEF1A1 has been shown to determine the susceptibility of a number of independent cell lines to chemical- and UV-induced transformation [
13] and has been identified as an actin binding protein in rat breast tumour cells, where it was found to be more highly expressed in metastatic than non-metastatic cells [
14]. It is not yet clear whether these properties are shared with eEF1A2, but the availability of specific antibodies that distinguish between the two isoforms should allow us to shed light on this. One hypothesis is that the non-canonical ("moonlighting") properties of eEF1A1 [
15] and eEF1A2 differ so that, for example, the way eEF1A2 interacts with the cytoskeleton might differ from that of eEF1A1 and affect the properties of cells which are expressing high levels of both isoforms. It has been shown, for example, that forced overexpression of eEF1A affects the cytoskeleton in both
S. pombe and
S. cerevisiae [
16,
17]. Alternatively, it has been shown that eEF1A1 and eEF1A2 differ in terms of their response to apoptotic agents [
18]; the finding that eEF1A2 is anti-apoptotic, at least in certain conditions, has obvious implications for the possible role of eEF1A2 in tumourigenesis. The observation that eEF1A2 expression is seen in the majority of cell lines, regardless of the tissue of origin, suggests that eEF1A2 expression may be triggered by the general process of transformation. This idea is strengthened by the fact that most of the few cell lines which do not express eEF1A2 tend to be untransformed, such as NIH3T3 cells (Figure
1).
The presence of increased levels of eEF1A2 in breast tumours may provide a useful new diagnostic marker. Further, eEF1A2 may prove to be a feasible target for therapeutic intervention. It has already been shown that growth-factor mediated eEF1A1 expression can be blocked with anti-EGF antibodies [
19]; it would be of interest to examine the response of eEF1A2 to similar antibodies. Investigations into non-canonical functions of eEF1A molecules may shed new light on mechanisms of oncogenicity.