The nuclear localization of a number of growth factors, cytokine ligands and their receptors has been reported in various cell lines and tissues. These include members of the fibroblast growth factor (FGF), epidermal growth factor and growth hormone families. Accordingly, a number of nuclear functions have begun to emerge for these protein families. The demonstration of functional interactions of these proteins with the nuclear import machinery has further supported their functions as nuclear signal transducers. Here, we review the membrane- trafficking machinery and pathways demonstrated to regulate this cell surface to nucleus-trafficking event and highlight the many remaining unanswered questions. We focus on the FGF family, which is providing many of the clues as to the process of this unusual phenomenon.
That growth factor receptors are found in the nucleus is now no surprise; solid lines of evidence demonstrate that fibroblast growth factor (FGF), epidermal growth factor (EGF), growth hormone (GH) and a number of other ligands, as well as their receptors [receptor tyrosine kinases (RTKs)], are targeted to the nucleus and convincingly localized therein (Table 1 and Figure 1) (1–7). However, what remains a conundrum is how these transmembrane (TM), cell-surface receptors and their external ligands end up in the nucleus. Viruses and toxins demonstrate that transport routes from the cell surface to the nucleus exist (8,9). Similarly, translocation of proteins and complexes between the cytoplasm and nucleus, via nuclear pores, is now a well-defined mechanism (10). While there is still no simple or complete schema for the nuclear translocation of growth factor receptors, recent studies provide insights worthy of overview at this time. The FGF family of ligands and receptors offer many of the recent updates in this story.
Table 1. Signalling ligands and receptors found in the nucleus
The FGFs are a family of 23 growth factors with varied roles in development, wound healing and metastasis (11,12). Fibroblast growth factors interact with high-affinity phosphotyrosine kinase receptors (FGFRs), a family of type-I TM proteins encoded by four genes, and with low(er) affinity to heparan sulphate proteglycans (HSPGs), to form a ternary FGF : FGFR : HSPG-signalling complex (13,14). Alternate splicing of many FGF ligands and all FGF receptors generates a substantial number of potential signalling complexes, able to elicit varied responses including migration, differentiation and cell proliferation (11,13,15). Following activation of this complex at the cell surface, both ligand and receptor are internalized (16). In some instances, this FGF/FGFR complex is subsequently translocated to the nucleus for functions that ultimately and largely regulate cell proliferation (17,18).
The Continuing Conundrum …
In any consideration of how cell-surface receptors end up in the nucleus, one of the most confronting and still unresolved issues is how TM growth factor receptors can escape from, and cross, the lipid bilayer. Evidence from several studies shows that some intact or full-length FGF and EGF receptors can be recovered from the nucleus (1,5–7,18,19). These TM receptors, initially assumed to be irrevocably inserted into the lipid bilayer, can move out of membranes and exist as matrix-associated proteins in cell nuclei, where they have been localized by a number of methods, including electron microscopy (Table 1). Overall, this remains a prickly concept and one still without a ready mechanism to explain it. The closest parallel is perhaps the sec61 translocon in the ER membrane responsible for translocating misfolded proteins into the cytoplasm (20,21); however, this, or an equivalent machinery, has not yet been implicated in the membrane translocation of cell-surface receptors. Finally, some receptors possess an atypical TM domain, which may facilitate their escape from the lipid bilayer. Studies examining substitutions of the TM domain between FGFR1-4 suggest the presence of an atypical TM domain in FGFR1 that facilitates its release into the cytosol (74), although FGFR2/3 enter the nucleus without this apparent feature (58,60).
A neat trick to avoid the conundrum is one wherein the cytoplasmic tails of receptors are cleaved and can then migrate as soluble peptides into the nucleus. This is known to occur in some instances, notably with the EGF family receptor ErbB4, which is cleaved by γ-secretase to release a nuclear-destined tail fragment (22). However, this more attractive mechanism is not universal; no such cleavage has been reported for many of the receptors recovered from the nucleus. What is becoming clear from a number of recent studies, however, is that internalization of growth factors, cytokines and their receptors into the endocytic pathway plays a key role in their subsequent nuclear translocation. A schematic diagram for a proposed routing of these proteins from the cell surface to nucleus via the endocytic pathway is presented in Figure 2. Additionally, the proteins implicated in the nuclear translocation of the FGF ligands and receptors are summarized in Table 2.
Table 2. Compartments, signals and machinery implicated in nuclear translocation of FGF and FGFR
Function in translocation
ARF6, ADP-ribosylation factor 6; CK2, casein kinase 2; FGF, fibroblast growth factor; PKC, protein kinase C.
Activation of cell-surface RTKs by ligand binding typically involves dimerization, autophosphorylation of the RTKs and initiation of downstream signalling (23). Studies on EGFR demonstrate how subsequent internalization targets activated receptors for degradation and downregulation of signalling, or for a variety of other trafficking fates (24,25). Moreover, in some cases, ligand/receptor complexes actually initiate signalling from within the endosome (25). While we have a growing understanding of their endocytic trafficking, we can now add the nucleus as another possible destination for internalized receptors.
One important question regarding nuclear translocation of cell-surface receptors is whether such proteins are indeed derived from the cell surface. To this end, biotinylation experiments first revealed that at least some of the nuclear FGFR1 pool derive from the cell surface and, that multiple spliceforms of the receptor, biotinylated at the cell surface, could be recovered in cell nuclei after ligand stimulation (7). Using alternative approaches, we have also recently demonstrated a key role for internalization of surface FGFR1, as well as its traffic through the endosomal pathway, for nuclear translocation (26). The dynamin and ARF6-dependent internalization of surface FGFR1 is required for its appearance in the nucleus (Table 2). Interestingly, inhibition of clathrin-dependent endocytosis through the use of an Eps15 mutant protein was not sufficient to block nuclear translocation of FGFR1, suggesting that a clathrin-independent or at least an Eps15-independent pathway may be involved in nuclear translocation (Table 2). In other work, a lipid and/or caveolin-dependent mechanism has been suggested for internalization and nuclear translocation (27,28). Furthermore, as perturbation of the GTPase cycle of Rab5 also blocked nuclear translocation of FGFR1, it appears that not only internalization of surface FGFR1, but also traffic into and, potentially, out of the early endosome is required for subsequent FGFR1 nuclear translocation (Table 2) (26). Importantly, the role of internalization of other surface receptors (such as the EGF and GH receptors) for subsequent nuclear translocation remains to be elucidated.
Endocytosis is also important for the nuclear translocation of exogenous ligands. Evidence for this was produced by Olsnes and colleagues who engineered a CaaX box sequence onto FGF ligands and used farnesylation of the FGF fusion protein as a measure of movement out of vesicular compartments, into the cytoplasm or nucleus of cells (29). The authors suggest that farnesylation of FGF alone would be insufficient to confer membrane attachment, rather than nuclear import. This also served to show that exogenous FGF ligand, rather than resident intracellular pools, entered the nucleus. Such exogenous FGF ligand is internalized into EEA1-positive early endosomal compartments and, at later time points, partially accumulates in the juxtanuclear recycling compartment (28,30). Similar to FGFR1 nuclear import, the internalization of FGF-1 into endosomal compartments occurred as a dynamin-sensitive and partially clathrin-independent event (27).
More recent studies have explored how FGF ligands cross the endosomal membrane to access the cytoplasm, from where they are available for nuclear import. The movement of FGFs into the cytoplasm was found not to be disrupted by agents that perturb the cytoskeleton, nor by Brefeldin A, which was used to block retrograde transport to the Golgi complex (30,31). While the acidification of the endosome was also found not to be important for translocating FGF across the membrane, the membrane potential generated by the vacuolar proton pump, or in its absence, by vesicular Na+/K+- ATPase, was required to translocate exogenous FGF into the cytoplasm (see also Table 2). Interestingly, the transit of FGF-1 and FGF-2 across endosomal membranes appears to be most efficient during the G1 phase of the cell cycle, and although the significance of such processing is unclear, it is accompanied by N-terminal cleavage of these ligands (30). Although extensive unfolding of proteins has been noted during some cases of cytosolic import, such as for diptheria toxin, this is not the case for transit of FGFs across endosomal membranes (32). The domains or sequences in FGFR and other receptors that facilitate translocation first into the cytosol, and then into the nucleus, have remained elusive. Both cytosol and nuclear import of FGF ligand require PI3K signalling yet do not require the functional tyrosine kinase domain (of FGFR4), instead they rely on the COOH-terminal fragment of this receptor (27,30,33–35).
It remains unclear which endosomal compartment provides the environment for transit of ligand or receptor into the cytoplasm. While the translocation of Heparan sulphates (also part of the FGF complex) into the cytoplasm occurs from late endosomal membranes, it remains to be formally demonstrated that this compartment is also a major destination for internalized FGFs. Indeed, current data would fit better with transit of FGF ligands across the membranes of early or recycling endosomes. In any case, translocation of FGFs across cell membranes is not unique to nuclear translocation; FGF-1 (lacks signal peptide) is non-classically secreted from cells through a mechanism involving direct translocation across the plasma membrane [for review, see (36)].
As both receptor and ligand for the FGF family are able to translocate to the nucleus following ligand stimulation, it is tempting to speculate that these proteins will be internalized and jointly translocated to the nucleus. Indeed, receptor-mediated internalization of FGF ligand (by FGFR and/or HSPGs) is required for nuclear translocation of FGFs (33,37,38). Similarly, pairing of an EGF family ligand with the EGFR has been suggested as a functional transcriptional unit within the nucleus itself (1).
Into the Nucleus
Once in the cytosol, growth factors and their receptors would potentially be free to interact with nuclear import machinery. The most well-understood mechanism for regulated nuclear import of proteins involves interaction between nuclear localization signals (NLSs) in cargo or associated adaptors and the importin family of proteins with the Ran small GTPase to facilitate active import (39). This machinery and the mechanisms involved in this import process have been recently reviewed in this journal and will henceforth be mentioned only in the context of growth factor receptors and ligands (10,39).
This importin pathway has been implicated in the nuclear import of both the EGF and FGF families of ligands and receptors (Table 2). As small polypeptide growth factors are mostly below the size threshold for diffusion through the nuclear pore (<40–45 kDa) (40), their nuclear accumulation is not an overly surprising finding. Interestingly, some FGFs (FGF-1,-2,-3,-12 and -13) contain classical NLSs. Others (FGF-1,-2,-11–13 and -18), which lack signal peptides, may exist solely as intracellular ligands (FGF-11–14) (11,41,42). The nuclear localization of numerous other growth factors and their receptors has been well documented over the last 15 years. A list of these proteins, the methods utilized in their detection, as well as nuclear-binding proteins and nuclear targets are summarized in Table 1. The phosphorylation status of FGF-1 in the cytoplasm is regulated by PKC, which in turn regulates its nucleocytoplasmic shuttling (Table 2) (43). Within the nuclei, FGFs have been detected in both the nucleoplasm and the nucleoli, regulated in FGF-2, at least, by a bipartite nuclear/nucleoli NLS (44).
In the case of cell-surface receptors the mechanisms of nuclear import are less clear. Functional NLSs have been detected in the cytoplasmic tail of a number of EGF (ErbB) receptor family members (1,45). However, no detectable or conventional NLS has been reported in FGFRs. Despite this, nuclear translocation of FGFR1 is dependent on the function of importin β (18), suggesting that the receptor has a cryptic or an as yet unidentified NLS or that import is mediated via an adaptor protein. The presence of an NLS in various ligands has resulted in the proposal that the receptor may be ‘piggy-backed’ into nuclei through import of cognate ligand (46). Interestingly, ligands can also stimulate movement of RTKs both within the nucleus and for nuclear export, and such is the case for ErbB3, emphasizing that these receptors can functionally move both in and out of nuclei (45).
Nuclear Functions for Ligands and Surface Receptors
Once in the nucleus, FGFs have demonstrated roles in signalling and transcription. In some instances, the nuclear targeting of FGFs is required to elicit their full mitogenic response, which can be mitigated experimentally by mutation of the NLS in FGF-1 (38,47). Interestingly, the loss of the NLS does not disrupt the interaction of FGF-1 with its cognate receptor, nor of receptor tyrosine phosphorylation and transcription of its downstream target c-Fos (47). Accordingly, introduction of FGF-1 into cells lacking FGFRs (via fusion to diptheria toxin) is able to stimulate DNA synthesis (38). Both FGF-1 and FGF-2 interact with casein kinase 2 (CK2), directing the kinase activity of CK2 towards the rDNA transcriptional component, nucleolin (Table 2) (48,49). The nuclear roles of FGFs thus appear to involve cell proliferation and DNA transcription, and accordingly, FGFs have been detected in rapidly proliferating tissues, such as at sites of injury (50) and in cell lines of metastatic origin (51).
Nuclear roles in regulating transcription for cell-surface receptors are also beginning to emerge. FGFR1 has been localized by both biochemical and microscopic methods to the nuclear matrix and splicing-rich speckles, respectively, suggesting specific roles in gene transcription and subsequently splicing (6,7,52). Nuclear targeting of FGFR1 by substituting its signal peptide for an NLS is sufficient to initiate DNA synthesis and transcription of c-Jun, an activator of CyclinD1 (38). Interestingly, the removal of the kinase region of nuclear-targeted FGFR1 ablates this effect, suggesting function as a tyrosine kinase within the nucleus (38,53). Similar to nuclear-targeted ligand, however, cell replication proper requires cosignalling from cell-surface FGFR (18,38). Nuclear FGFR1 directly interacts with ribosomal S6 kinase (RSK) and modulates its activity towards cAMP response elements (CRE) transcriptional regulator complexes, to regulate the transcription of genes involved in proliferation and differentiation (53–55). It remains controversial whether nuclear-localized receptors can activate the same cascades as they do at the cell surface, such as ERK, PI3K or p38MAPK (18,55). How cessation of signalling from these receptors occurs once activated and translocated to the nucleus remains unknown. Whether the same complement of phosphatases that regulates these receptors at the cell surface, and in endosomes, also performs a similar role in the nucleus remains to be explored.
A nuclear role for members of the EGFR family has also been recently demonstrated. Nuclear-localized EGFR physically binds to AT-rich sequences in the promoter region of CyclinD1, inducing its transcription and subsequent cell-cycle progression (1). Nuclear ErbB2, another EGF family receptor, interacts with the promoter region of COX2, a metabolic enzyme important for cytokine production, and demonstrates a positive correlation between nuclear transactivation, COX2 transcription and breast tumour progression (2). As a number of nuclear RTKs are intimately involved in proliferation and, by implication, cell-cycle progression, it is tempting to speculate that the nuclear localization of RTKs may in some instances be oncogenic. Indeed, there is increased expression of ErbB2 and COX-2 in many types of human cancer (56,57).
There is notable disharmony in the kinetics of nuclear translocation among different receptors. This fact alone suggests that multiple pathways for import may exist. For instance, rapid (approximately 5 min) induction of nuclear GH receptor is achieved after ligand stimulation (3), while FGF stimulation induces nuclear appearance of FGFR1 at later timepoints (approximately 1 h) (7,26). This raises the idea of a ‘knock on’ effect, as a possible scenario for rapid deployment of receptors into the nucleus. In this, ligand-activated cell surface receptors would remotely induce the nuclear translocation of additional receptor pools from intracellular locations. To our knowledge, there is little information to either support or exclude this scenario. The large number of splice variants of the FGFR genes, including secreted or intracellular pools (13,58), further suggests the potential for co-operativity between receptor pools.
The actual proportion of receptor that gets translocated to the nucleus is a consideration. Relatively low levels of plasma membrane staining are observed for some RTKs, yet give staining patterns that suggest abundant nuclear-localized receptor after ligand stimulation (1,7,26). Biochemical measures suggest that only a subset of total receptors translocate to nuclei (7). This disparity remains unexplained.
A number of growth factor receptors also undergo heterophilic interactions with adhesion proteins, such as N- and, potentially, E-cadherin, in the case of FGFR1 (26,59). Accordingly, the adhesive and polarization status of epithelial cells has been recently demonstrated to regulate nuclear import of FGFR1 and ErbB3 (26,45). A future challenge thus involves delineating the physiological and developmental contexts for nuclear translocation of RTKs. The recent report that nuclear FGFR2 plays a role in the development of male gonads in mice is beginning to shed light on this process (60).
Finally, and in keeping with the predilections of this field, we raise vesicular or membrane transport as an unsupported, but possible mechanism for nuclear translocation of receptors. Anatomical studies have long recorded the presence of intranuclear vesicles and membranes in lower vertebrates and in cells of the reproductive system (61,62), although there is certainly no current evidence to link these structures with nuclear receptors. More generally, the dynamic formation of the nuclear envelope during mitosis involves vesicle fusion and considerable restructuring of the outer (ER) and inner leaflets of the nuclear envelope (63). Interestingly, a recent study shows that overexpressed, truncated forms of FGFR4 are localized on large, lamaellar membranes associated with the nuclear envelope and ER and positioned within nuclei (63). While this is the first demonstration of membrane-associated FGFR in the nucleus, it is not seen with wild-type receptor and is yet to be demonstrated in a physiological context. Hence, while nuclear translocation via membrane-bound vesicles would tidily answer many of the remaining conundrums and would provide plausible, alternative explanations for some of the experimental data so far, it remains only notional at present.
The focus on the FGF ligands and RTKs for much of this article is not intended to underscore the importance of other growth factors and their receptors in the nucleus. Indeed, a multitude of other systems, including, but not limited to GH/prolactin, EGF, VEGF, NGF and IFNγ undergo nuclear translocation of their receptor and ligands and are listed in Table 1. The study of the FGF family has revealed a number of surprising mechanisms and has highlighted the questions remaining to be answered. As it becomes clear that signalling of these receptors to downstream effectors is intimately regulated by both their subcellular and membrane microdomain organization, so too must we consider the mechanism and consequence of their nuclear translocation.
We apologize to those authors whose work was not cited in the pursuit of brevity and focus on FGF. We thank members of the Stow laboratory for invaluable intellectual discussions and thank David Jans, Michael Waters and Becky Conway-Campbell for sharing unpublished data and ideas. David M. Bryant holds an Australian Postgraduate Award scholarship, and Jennifer L. Stow is supported by a fellowship and grants from the National Health and Medical Research Council of Australia.