In this study, we have shown that (a) although 14-3-3 proteins are predominantly localized in the cytoplasm, a large number of ligands are localized within the nucleus; (b) endogenous 14-3-3, fully competent to bind ligands, can be trapped in the nucleus by inhibiting Crm1-dependent nuclear export with LMB; (c) the leucine-rich putative NES sequence in 14-3-3, despite its structural homology to known NES sequences (Rittinger et al., 1999
) and its ability to function as an NES in isolation when fused to GFP (unpublished data), functions primarily in phosphoprotein binding and not as an NES in the context of the intact 14-3-3 molecule; (d) a mutant 14-3-3, which cannot bind to ligands, homes to the nucleus; (e) the nuclear 14-3-3 ligand FKHRL1 is phosphorylated at its major 14-3-3 binding site within the nucleus before its export into the cytoplasm; and (f) at least for FKHRL1, rapid export from the nucleus to the cytoplasm requires both phosphorylation/14-3-3 binding and NES sequences within the bound ligand. In addition, we have identified NLS sequences within FKHRL1 that are responsible for its nuclear localization in the absence of phosphorylation. These data suggest that for nuclear FKHRL1, the sequence of events after growth factor stimulation involves phosphorylation at the 14-3-3 binding site within the nucleus, where ligand-free 14-3-3 molecules are located, followed by rapid nuclear export that requires both phosphorylation/14-3-3 binding and intrinsic NES sequences in FKHRL1. Once FKHRL1 has been exported to the cytoplasm, phosphorylation/14-3-3 binding may play an additional role in preventing nuclear reimport possibly by masking the FKHRL1 NLS.
Our observation that many potential 14-3-3 binding substrates are present in the nuclear fraction and that FKHRL1 is phosphorylated at its major 14-3-3 binding site within the nucleus suggests that kinases, such as AKT, Chk1, and Chk2, may act preferentially within this compartment, conferring 14-3-3 binding to some of their substrates. Furthermore, the observations that LMB treatment led to accumulation of endogenous 14-3-3 in the nucleus and that a K49E mutant of 14-3-3, which does not bind to ligands, localizes preferentially within the nucleus suggests that unliganded 14-3-3 normally transits to the nuclear compartment where binding to some ligands can occur. It would also be interesting to investigate the rate of import of 14-3-3 into the nucleus to address whether dissociation of ligands from preformed cytoplasmic complexes determines the rate of 14-3-3 nuclear import. We found that elimination or mutation of the leucine-rich sequence within helix αI disrupts binding to numerous phosphorylated protein ligands. This is in excellent agreement with the results of more focused mutational studies reported by others examining 14-3-3 binding to Raf and Cdc25 (Thorson et al., 1998
; Wang et al., 1998
; Kumagai and Dunphy, 1999
; Zeng and Piwnica-Worms, 1999
). Thus, the primary and general function of this α-helix is to directly participate in ligand binding. Furthermore, the observation that this leucine-rich sequence is incapable of maintaining unliganded 14-3-3 in the cytoplasm is incompatible with its proposed function as a true NES (Lopez-Girona et al., 1999
). Instead, our present study of FKHRL1 and previous studies on Cdc25 (Kumagai and Dunphy, 1999
; Yang et al., 1999
; Davezac et al., 2000
) indicate that export of the 14-3-3–bound complex involves NES sequences that are intrinsic to the ligands.
We reported previously an interaction between 14-3-3 and Crm1 within eukaryotic cell lysates that was disrupted by a 14-3-3 binding peptide (Rittinger et al., 1999
). At that time, we interpreted this as evidence that the leucine-rich region within helix αI could interact with either ligands or Crm1 but not both simultaneously. In light of our current findings, the interaction that we observed between 14-3-3 and Crm1 was most likely mediated through the 14-3-3–bound ligands, which were competed off by the 14-3-3 binding peptide. In agreement with this, we have been unable to show a direct interaction between 14-3-3 and Crm1 when both are expressed in Escherichia coli
Our findings are consistent with several different functions for AKT and 14-3-3 in regulating the activity and subcellular localization of FKHRL1. We have shown that AKT-mediated phosphorylation of FKHRL1 at the 14-3-3 binding site within the nucleus results in accelerated nuclear export of FKHRL1. This role for phosphorylation/14-3-3 binding in nuclear export is novel and complements recent results from other investigators who reported, while our paper was under revision, that AKT phosphorylation of the FKHRL1 homologs AFX and FKHR can inhibit nuclear reimport (Brownawell et al., 2001
; Rena et al., 2001
). In addition, Brownawell et al. (2001)
proposed that phosphorylation at the 14-3-3 binding site did not affect nuclear export, since a triple phosphorylation site mutant of AFX that could not bind to 14-3-3 was able to shuttle between nuclei in heterokaryon fusion experiments. However, this study did not examine the time dependence of the import or export process, and our data in indicates that phosphorylation at the 14-3-3 binding site is critical for rapid nuclear export of FKHRL1. Therefore, it is likely that AKT phosphorylation of Forkhead family members affects both nuclear export and import: phosphorylation/14-3-3 binding increases the rapid export of FKHRL1 from the nucleus in a manner that requires the intrinsic NES sequences on FKHRL1, whereas phosphorylation near the NLS site may disrupt FKHRL1 subsequent nuclear reimport. Furthermore, Cahill et al. (2001)
showed that phosphorylation of the C. elegans
FKHRL1 homologue daf-16
by AKT together with 14-3-3 binding directly inhibited its DNA binding ability. Thus, AKT- and 14-3-3–mediated inhibition of FKHRL1 function is likely to involve multiple events, first by blocking DNA binding and then by ensuring a complete sequestration of this transcription factor in the cytoplasm away from their nuclear target genes.
We find that a K49E mutant of 14-3-3 is unable to bind to ligands or be exported from the nucleus, despite its ability to form heterodimers with endogenous WT 14-3-3 in vivo and in vitro. This finding underscores the important role of ligand binding by each monomer within the 14-3-3 dimer for 14-3-3 function. A requirement for dimeric 14-3-3 has also been demonstrated for both 14-3-3–mediated Raf activation (Tzivion et al., 1998
) and for 14-3-3–mediated disruption of daf-16
DNA binding after AKT phosphorylation (Cahill et al., 2001
) using a mutant 14-3-3 in which the dimerization interface was disrupted.
Our functional data shown in , examining FKHRL1 export as a function of time, shows an increase in nuclear export upon FKHRL1 phosphorylation and 14-3-3 binding, which suggests that 14-3-3 may enhance the export process. Exactly how 14-3-3 accomplishes this remains to be defined. 14-3-3 binding might facilitate the formation of productive complexes between FKHRL1 and regulatory components of the nuclear export machinery, perhaps through inducing conformational changes in the 14-3-3–bound ligand. Recent data from Brownawell et al. (2001)
on AFX and our unpublished observations on FKHRL1 suggest, however, that 14-3-3 binding does not cause an increase in the bulk association of these transcription factors with Crm1, though both proteins are grossly overexpressed in these assays. It has not been technically possible to measure the kinetics with which the association occurs between the endogenous proteins within cells. Alternatively, 14-3-3 binding could facilitate the ability of FKHRL1–Crm1 complexes to bind to endogenous Ran-GTP to trigger nuclear export. Another possibility is that 14-3-3 binding protects phosphorylation sites from the action of phosphatases and/or thereby permits additional modifications of FKHRL1, which are themselves involved in the export process.
We have focused this study specifically on one particular 14-3-3 ligand, FKHRL1, but studies on other 14-3-3 binding ligands such as HDAC4, -5, and -7 and Cdc25 are consistent with the idea that cytoplasmic relocalization involves both 14-3-3 proteins and ligand NES and NLS sequences (Kumagai and Dunphy, 1999
; Yang et al., 1999
; Davezac et al., 2000
; Grozinger and Schreiber, 2000
; Wang et al., 2000
; Kao et al., 2001
). Furthermore, not all 14-3-3–bound ligands are destined for cytoplasmic sequestration. For other molecules, 14-3-3 binding may preferentially sequester them within the nucleus as has been reported for the catalytic subunit of human telomerase (Seimiya et al., 2000
) and the HOX11 homeobox transcription factor Tlx-2 (Tang et al., 1998
). Thus, it appears that the ligand, rather than 14-3-3, dictates the resulting subcellular location (Muslin and Xing, 2000
). In this manner, 14-3-3 functions as a type of “molecular chauffeur” where the destination of the 14-3-3–bound complex is determined by instructions contained within the sequence and structure of the bound cargo rather than through any intrinsic properties of 14-3-3. The ultimate fate of such cargo is then determined by 14-3-3–mediated alterations in the kinetics of dynamic nuclear-cytoplasmic transport.