In an attempt to identify new cellular targets of Adv5 E1A, we utilized the proteomic approach of TAP. Our investigation identified several well-established cellular targets of E1A in addition to novel E1A-interacting proteins, the forkhead transcription factors FOXK1 and FOXK2. E1A interaction with FOXK1/K2 occurs through the C-terminal region of E1A (exon 2). The E1A sequences involved in the interaction of FOXK1/K2 are conserved only among the species C (nononcogenic) Advs (4
). Unlike the FOXK1/K2-interacting module, the DYRK(1A/1B)/HAN11 binding sequences are located within the CR4 region conserved among different species of human Advs. Interaction with DYRK(1A/1B)/HAN11, like interaction with CtBP1/2, may be a common feature of E1A proteins of all primate Advs. Interestingly, the E6 proteins of low-risk β-HPVs (HPV21 and HPV14) also interacted with FOXK1/K2 through a sequence similar to that of Adv5 E1A.
Our results suggest that interaction of FOXK1/K2 with E1A may be mediated, at least in part, through the FHA (forkhead-associated) domain. The FHA domain is a phosphoprotein binding domain that predominantly recognizes phospho-Thr ([p-Thr] in addition to p-Ser and p-Tyr) residues within diverse proteins (39
). We have identified a Thr-Ser-containing sequence motif within Adv5 E1A and HPV21/14 E6 that is essential for interaction with FOXK1/K2 transcription factors. Our results show that both Thr and Ser residues are essential. Between these critical residues in E1A, Ser-219 and Ser-231 were reported to be phosphorylated (57
). We note that the phosphorylation site at Ser-219 is the consensus phosphorylation site for DYRK1A/1B, which suggests that DYRK1A/1B could possibly alter E1A activities by phosphorylation. Different phosphorylation prediction algorithms suggest that Thr-229 is not an optimal phosphorylation site, consistent with our inability to detect phosphorylation at this site (data not shown). Nonetheless, the critical importance of the Thr residue is evident from our mutational analyses of E1A and of HPV21 E6 and from the inability of HPV20 E6 (which contains a Glu substitution instead of Thr) to interact with FOXK1/K2 (Fig. and ).
Previous results from our laboratory reported enhanced transforming, tumor-inducing, and metastasis activities associated with E1A exon 2 mutants (56
). These results led us to forge the concept that E1A exon 2 possesses a transformation- and oncogenesis-restraining activity (8
). Part of this activity was linked to the extreme C-terminal region of E1A which interacts with CtBP (6
). Our present studies have linked two other cellular protein complexes, FOXK1/K2 and DYRK(1A/1B)/HAN11 (in addition to CtBP1/2), with transformation- and tumorigenesis-restraining activities of the E1A C-terminal (exon 2) region. Our results suggest that E1A interaction with FOXK1/K2 results in retardation of E1A-induced cell proliferation and oncogenic transformation. The E1A mutant (dl
1132) deficient in interaction with FOXK1/K2 consistently induced increased proliferation, transformation, and tumor formation compared to wt E1A, albeit not as robustly as mutants defective in interaction with DYRK(1A/1B)/HAN11 and CtBP1/2 (dl
1134, and dl
1135). The present results on the transforming activity of mutant dl
1132 (defective in interaction with FOXK1/K2) are in good agreement with a different mutant that contained a deletion encompassing the region deleted in dl
). In the present study, we demonstrate that deletion mutants dl
1133 and dl
1134, defective in interaction with DYRK(1A/1B)/HAN11 are hypertransforming, suggesting that E1A interaction with the DYRK(1A/1B)/HAN11 complex may inhibit cell proliferation and transformation. Certain hypertransforming E1A mutants previously identified by the Quinlan laboratory (13
) map in the C-terminal region deleted in mutants d
l1133 and dl
1134 (aa 241 to 270). It would be interesting to determine whether the E1A mutants reported by the Quinlan group are deficient in interaction with DYRK(1A/1B)/HAN11. Thus, E1A interaction of all three cellular protein complexes [i.e., FOXK1/K2, DYRK(1A/1B)/HAN11, and CtBP1/2] may additively contribute to the suppression of transformation and tumorigenesis since deletion encompassing all three protein interaction regions of the E1A C terminus resulted in the most potent transforming activity (56
The Adv E1A C-terminal region may exert its inhibitory effect on cell transformation by deregulation of cell differentiation-related activities via recruitment of the three different cellular protein complexes (Fig. ). Steve Frisch reported that E1A was able to suppress the tumorigenic activity of several human cancer cell lines by engendering an epithelial phenotype (23
). Subsequently, the mesenchymal-to-epithelial transformation activity of E1A was linked to interaction with CtBP (29
) and relief of transcriptional repression of cellular genes involved in the regulation of cell proliferation, differentiation, and apoptosis (28
Like the CtBP1/2 complex, the FOXK1/K2 complex may modulate cellular pathways involved in cell proliferation, differentiation, and apoptosis. In mammalian cells, FOXK1 appears to be a stem cell maintenance factor (reviewed in reference 41
). Disruption of FOXK1 was reported to result in upregulation of differentiation-related genes in a muscle stem cell population (44
). In contrast to FOXK1, there are no significant studies of the activities of FOXK2. The yeast forkhead proteins Fkh1 and Fkh2 appear to be homologs of mammalian FOXK1/K2 because they both contain FHA domains, which are not typically found in other forkhead family members. Several reports indicate that Fkh1/2 have a definitive role in transcriptional regulation of the genes involved in the G2
/M transition (65
). In addition, the activities of Fkh1/2 have been linked to pseudohyphal growth, a form of cell transformation in yeast (65
). Although transcriptional repression and activation functions of mammalian FOXK1/K2 are possible, only the transcriptional repression activity of FOXK1 has been studied in some detail (21
). Thus, Adv5 E1A and HPV21/14 E6 may deregulate transcriptional activities of FOXK1/K2 either by sequestration or by redirection to other target genes to facilitate cell cycle withdrawal and promotion of cell differentiation.
The DYRK(1A/1B)/HAN11 complex may also function at the crossroads of cellular proliferation and differentiation. Mammalian DYRK1A has also been shown to interfere with the Notch signaling pathway and the NFAT and Gli transcriptional programs, which are linked to cell fate and differentiation activities (3
). DYRK1B has been shown to regulate differentiation and apoptosis (reviewed in reference 22
). Lastly, the activity of yeast Yak1p has been linked to pseudohyphal growth (64
). Genetic studies of zebra fish have linked the DYRK1A cofactor HAN11 with certain differentiation programs (47
). Therefore, the E1A C-terminal region may exert its effects on cell proliferation and transformation by disruption of the cellular activities of these protein kinases and HAN11. As evidenced from genetic studies of yeast which revealed significant cross talk between Yak1p and Fkh1/2 pathways (33
), there may be cross talk between all three pathways deregulated by the E1A C-terminal region to exert a concerted effect on cell proliferation and transformation.
Of considerable significance is our discovery of a commonality between Adv5 E1A and HPV21/14 E6 proteins which target FOXK1/K2 through a conserved S/T sequence motif to suppress cell transformation. We have shown that HPV21 E6 can dampen Ras-cooperative transformation induced by the hypertransforming E1A mutant that is deficient in interaction with FOXK1/K2 (dl
1132) while an HPV21 E6 mutant (S8 → A) that is unable to interact with FOXK1/K2 is not able to do so. These results suggest that the E6 proteins of HPV that complex with FOXK1/K2 may also suppress transformation by retarding cell proliferation. What might be the functional consequence of such an activity in the natural history of these HPVs? HPV21 and HPV14, in addition to various other HPVs, are associated with benign cutaneous epidermodysplasia verruciformis (EV) lesions (9
). The E6 proteins of HPV21 and HPV14 may promote differentiation by inhibiting cell proliferation to facilitate replication of these epithelium-specific viruses. Since about half of the EV patients develop squamous cell carcinoma (49
), the transformation-restraining activities of the N-terminal region of HPV21/14 E6 may also suppress the oncogenic activities of other HPVs present in EV lesions to promote productive viral infection rather than oncogenic transformation. The E6 proteins of HPV21/14 and Adv5 E1A might target FOXK1/K2 to manipulate the delicate balance between proliferation and differentiation to create an optimal cellular state for viral replication in epithelial cells. The shared mechanism may be important in the natural history of these different epithelium-specific viruses.