Characterization of a Novel Tripartite Nuclear Localization Sequence in the EGFR Family*

Aberrant expression of epidermal growth factor receptor (EGFR) is present in many human tumors. Several reports have shown that EGFR is translocated into the nucleus during liver regeneration and in several types of cells and tissues such as placenta and thyroid. Nuclear EGFR is associated with transcription, DNA synthesis, and DNA repair activity and serves as a prognostic marker in breast carcinoma and oropharyngeal squamous cell cancer. However, the nuclear localization sequence (NLS) of EGFR has not been extensively examined. In this study, we have shown that the juxtamembrane region of EGFR harbors a putative NLS with three clusters of basic amino acids (RRRHIVRKRTLRR (amino acids 645–657)) that mediates the nuclear localization of EGFR. We found that this newly characterized tripartite NLS is conserved among the EGFR family members (EGFR, ErbB2, ErbB3, and ErbB4) and is able to move each to the nucleus. Further, this tripartite NLS could also mediate the nuclear localization of other known cytoplasmic proteins such as pyruvate kinase. We have demonstrated that mutating one of the three basic amino acid clusters (R or K → A) leads to significant impairment of the nuclear localization of EGFR and that of a green fluorescent protein-pyruvate kinase-NLS reporter protein. Our results show that this tripartite NLS is distinct from the traditional mono- and bipartite NLS and reveal a mechanism that could account for the nuclear localization of membrane receptors.

Aberrant expression of epidermal growth factor receptor (EGFR) is present in many human tumors. Several reports have shown that EGFR is translocated into the nucleus during liver regeneration and in several types of cells and tissues such as placenta and thyroid. Nuclear EGFR is associated with transcription, DNA synthesis, and DNA repair activity and serves as a prognostic marker in breast carcinoma and oropharyngeal squamous cell cancer. However, the nuclear localization sequence (NLS) of EGFR has not been extensively examined. In this study, we have shown that the juxtamembrane region of EGFR harbors a putative NLS with three clusters of basic amino acids (RRRHIVRKRTLRR (amino acids 645-657)) that mediates the nuclear localization of EGFR. We found that this newly characterized tripartite NLS is conserved among the EGFR family members (EGFR, ErbB2, ErbB3, and ErbB4) and is able to move each to the nucleus. Further, this tripartite NLS could also mediate the nuclear localization of other known cytoplasmic proteins such as pyruvate kinase. We have demonstrated that mutating one of the three basic amino acid clusters (R or K3 A) leads to significant impairment of the nuclear localization of EGFR and that of a green fluorescent protein-pyruvate kinase-NLS reporter protein. Our results show that this tripartite NLS is distinct from the traditional mono-and bipartite NLS and reveal a mechanism that could account for the nuclear localization of membrane receptors.
Evidence is emerging to suggest that either full-length or fragmented EGFR family members can be shuttled from the plasma membrane to the nucleus (8 -20). Putative nuclear localization sequences (NLSs) of EGFR family members are thought to be required for the nuclear function of EGFR as a transcriptional co-activator (9,13). A classical monopartite NLS, RRRRHSP, resembling the NLS for the simian virus 40 (SV40) large T (LT) antigen has been identified in the C terminus of ErbB3 (11). Putative NLSs for ErbB2 and ErbB4 that mediate the nuclear localization of either full-length ErbB2 or truncated ErbB4 have also been identified in the juxtamembrane (JM) region (14,21). However, whether these putative NLSs within the JM region of EGFR family members could transport cytoplasmic proteins to the nucleus has not been addressed.
Molecules are shuttled between the nucleus and cytoplasm via several well studied pathways. Small molecules can diffuse freely through the nuclear pores, but nuclear localization of large molecules is generally mediated by NLSs, which are rich in basic amino acids (22)(23)(24). Two types of NLS have been identified thus far, one consisting of a monopartite sequence of basic amino acid residues (e.g. 126 PKKKRKV 132 in the SV40 LT antigen (25)) and the other a bipartite sequence of two clusters of basic amino acids (e.g. 305 KRALPNNTSSSPQPKKK 321 in p53 protein (26,27)). NLS-bearing molecules are thought to be transported into the nucleus through their forming of a complex with importin ␣/␤ (28,29) or importin ␤ alone (30,31). Importin ␣ is an armadillo repeat-containing protein that has an arginine-rich N-terminal domain. The armadillo repeats form a grooved structure to which the basic NLS could bind (32). Importin ␤ is responsible for docking to the nuclear pore complex and for translocating the NLS-bearing molecules through the pore (33)(34)(35)(36). Several proteins can be transported into the nucleus by directly binding to importin ␤, including ribosomal proteins (37) and the viral proteins HIV-1, Tat, and Rev (30). Here we have characterized a novel tripartite NLS, an NLS containing three clusters of basic amino acids that is conserved within the JM region of all EGFR family members, is capable of targeting cytoplasmic proteins into the nucleus, and is responsible for the nuclear localization of EGFR family members.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection Conditions-All cell lines were maintained in Dulbecco's modified Eagle's medium/F-12 with 10% fetal calf serum and 100 g/ml penicillin and streptomycin. The cells were serum-starved for 20 -24 h and then stimulated with EGF in various concentrations for different times. Transfection was done with the cationic liposome SN as previously described (13,38). Briefly, cells were grown on Petri dishes or slides overnight and incubated with plasmid-liposome complex in serum-free medium for 4 h, after which the medium was replaced by complete medium and the cells incubated at 37°C for 24 -48 h.
Plasmid Constructs-All of the EGFR expression plasmids were amplified by PCR from pcDNA3.1-EGFR, a kind gift from Dr. David James (Mayo Clinic, Rochester, MN). The pcDNA6A-EGFR plasmid, expressing full-length human EGFR with a C-terminal Myc-His 6 tag, was constructed by PCR with the forward primer 1F-H (5Ј-ATTAAGCTTCGGGGAG-CAGCGATG-3Ј) and the reverse primer 3630R-X (5Ј-CCT-TCTAGATGCTCCAATAAATTCACTG-3Ј). The DNA fragments were digested with HindIII and XbaI and cloned into the corresponding sites of the pcDNA6A vector (Invitrogen). A Myc-tagged, EGFR intracellular domain (ICD)-expressing plasmid was constructed with DNA fragments corresponding to amino acids 644 -1186 of EGFR. The PCR products were digested with KpnI and XbaI and subsequently cloned into the corresponding sites of the pcDNA6A vector. To generate NLS mutants of EGFR, specific alanine mutations were constructed by site-directed mutagenesis using either the QuikChange XL site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions or a two-round PCR method (39). The resulting PCR products were cloned into the pcDNA6A vector. The NLSm1 mutant ( 645 AAA 647 ) has a mutation in the first basic amino acids cluster, the NLSm12 mutant ( 645 AAA-HIVAAA 653 ) has a mutation in the first two basic amino acid clusters, and the NLSm3 mutant (RRRHIVRKRTLAA) has the first two basic clusters intact and a mutation in the third. We also created a dNLS mutant, in which all three basic amino acid clusters of the EGFR NLS ( 645 RRRHIVRKRTLRR 657 ) are deleted and the two amino acids (LE) inserted at the deletion junction. The dNLS mutant of EGFR was constructed by inverting PCR with the two primers 2004R-X (5Ј-GATCTCGAGCA-TGAAGAGGCCGATCCC-3Ј) and 2044F-X (5Ј-GATCTCG-AGCTGCTGCAGGAGAGGG-3Ј). The PCR product was then digested with XhoI and ligated with T4 DNA ligase.
The plasmid GFP-ICD, expressing human EGFR amino acids 640 -1186 with an N-terminal green fluorescent protein (GFP) tag, was constructed by using the forward primer 1990F-B with a BspEI site, (5Ј-AGGTCCGGAATCGGCCTCTTCATG-3Ј) and the reverse primer 3633R-H harboring a HindIII site (5Ј-CCCAAGCTTCATGCTCCAATAAATTC-3Ј). The amplified DNA fragments were digested with BspEI and HindIII and cloned into the corresponding sites of a pEGFP-N1 vector (Clontech). The GFP-ICD NLS mutants were PCR-amplified with the same primers and cloned. The plasmid GFP-JM, expressing the human EGFR JM region, was constructed by using PCR with the forward primer 1990F-B and the reverse primer 2136R-X (5Ј-TATTCTAGAATTCAGTTTCCTTC-AAG-3). The DNA fragments were digested with BspEI and XbaI and cloned into the pEGFP-N1 vector.
Cellular Fractionation and Western Blotting Analyses-Cellular fractionation was performed as described previously (9). Briefly, cells were washed twice with ice-cold phosphate-buffered saline, harvested by scraping with a rubber policeman, and lysed in a lysis buffer (20 mM HEPES, pH 7.0, 10 mM KCl, 2 mM MgCl 2 , 0.5% Nonidet P-40, 1 mM Na 3 VO 4 , 10 mM NaF, 1 mM phenylmethanesulfonyl fluoride, 2 g/ml aprotinin). After incubation on ice for 10 min, the cells were homogenized by 20 strokes in a tightly fitting Dounce homogenizer. An aliquot of cells was checked for cell lysis by light microscopy to confirm that Ͼ98% of cells had been lysed. The homogenate was centrifuged at 1,500 ϫ g for 5 min to sediment the nuclei. The supernatant was then centrifuged at a maximum speed of 16,100 ϫ g for 20 min, and the resulting supernatant formed the non-nuclear fraction. The nuclear pellet was washed three times with lysis buffer to remove any contamination from cytoplasmic membranes, and the purity of the nuclei was confirmed by light microscopy. To extract nuclear proteins, the isolated nuclei were resuspended in NETN buffer (150 mM NaCl, 1 mM EDTA, 20 mM Tris-Cl, pH 8.0, 0.5% Nonidet P-40, 1 mM Na 3 VO 4 , 10 mM NaF, 1 mM phenylmethanesulfonyl fluoride, and 2 g/ml aprotinin), and the mixture was sonicated briefly to aid nuclear lysis. Nuclear lysates were collected after centrifugation at 16,100 ϫ g for 20 min at 4°C. To obtain whole-cell lysates, cells were lysed in NETN buffer with sonication; lysates were then centrifuged at 16,100 ϫ g for 20 min. Protein concentrations were determined by using the Bradford method (Bio-Rad), and then the samples were mixed with Laemmli sample buffer and heated at 95°C for 5 min. Samples were subjected to SDS-PAGE on 8 or 10% polyacrylamide gels, and the proteins were transferred to nitrocellulose membranes. Prestained molecular mass standards for electrophoresis were obtained from Bio-Rad. Membranes were probed with monoclonal or polyclonal antibodies that were followed by horseradish peroxidase-labeled secondary antibodies. Immunoreactive protein bands were detected with an enhanced chemiluminescence reagent (Pierce or Amersham Biosciences). When required, the intensity of the expression proteins was quantified with a computer-assisted scanning densitometer equipped with Quantity One software (Bio-Rad). The antibodies used in this study were as follows: anti-EGFR (Santa Cruz Biotechnology), anti-lamin B (Calbiochem), anti-␣-tubulin (Sigma), anti-actin (Sigma), antic-Myc (Roche Applied Science), and anti-GFP (Clontech). All secondary antibodies were obtained from Vector Laboratories and Jackson ImmunoResearch Laboratories.
Confocal Analyses-Cultured cells were washed three times with phosphate-buffered saline, fixed with 4% paraformaldehyde for 15 min, permeabilized with 0.2% Triton X-100 for 5 min, and incubated with 2% bovine serum albumin for 1 h. Cells were then incubated with the primary antibodies overnight at 4°C, washed and further incubated with the appropriate secondary antibody, and tagged with either fluorescein isothiocyanate or Texas Red. Nuclei were stained with Topro 3 before mounting. Cells expressing GFP fusion proteins were subjected to nuclear staining after blocking with bovine serum albumin. Confocal fluorescence images were captured with a Zeiss CLM510 laser microscope. The images were captured in the middle sections of the nuclei.

EGF Stimulation Induces Accumulation of EGFR in the
Nucleus-The detection of EGFR in cell nuclei was first reported in regenerating liver and adrenocortical carcinoma (41,42). Nuclear expression of EGFR was subsequently detected in many other cell types and tumors, including the uterine cervix, bladder, breast, skin, and thyroid (9,(43)(44)(45)(46). Nuclear EGFR was found to be associated with transcription, DNA repair, and DNA synthesis activity (9,13,15,19,(47)(48)(49). To establish a reliable system for the functional analysis of nuclear localization of EGFR, we analyzed the cytoplasmic-tonuclear distribution of EGFR in a breast cancer cell line (MDA-MB-468) after stimulation with 10, 25, 50, or 100 ng/ml EGF for 30 min (Fig. 1A). Nuclei were then isolated by cellular fractionation and analyzed for EGFR level. The nuclear EGFR levels of the EGF-stimulated cells were about three times higher than those of unstimulated cells. To assess EGFR redistribution at different time points, we treated cells with 100 ng/ml EGF and measured nuclear EGFR levels at 0, 30, 60, and 240 min later. Nuclear EGFR levels were found to be elevated at 30 min, reached a plateau at 60 min, and declined toward the base line at 240 min after EGF treatment (Fig. 1B). Because physiological EGF levels typically range from 10 to 100 ng/ml (50), we reexamined the nuclear localization of EGFR at different times after stimulation with EGF at 25 ng/ml, a physiological concentration. A similar expression profile was observed, except that the 25 ng/ml EGF treatment led to EGFR being retained in the nucleus longer than did the 100 ng/ml EGF treatment (Fig. 1C).
We also used confocal microscopy to confirm the nuclear localization of EGFR in response to EGF. In the absence of EGF stimulation, the EGFR protein remained mostly at the cell membranes. As early as 10 min after treatment with 10 ng/ml EGF, yellow signals were detected, indicating co-localization of EGFR and DNA, in the nucleus and in the perinuclear membranes (data not shown; see Refs. 13, 19, and 51).
Identification of a Nuclear Localization Sequence within EGFR-Nuclear localization signaling is usually mediated by specific, short sequences of basic amino acids within nuclear proteins. Computer analysis of the EGFR sequence with the POSRTII program revealed the sequence RRRH (amino acids 645-648) to be a putative NLS. We also noted a polybasic sequence in the intracellular JM region of EGFR A, Western blots show dose-dependent increase in nuclear EGFR protein in response to EGF. MDA-MB-468 cells (70% confluence) were treated with the indicated concentrations of EGF at 37°C for 30 min and subjected to biochemical fractionation to separate nuclei from non-nuclear material. The nuclear (40 g) and non-nuclear (15 g) extracts were subjected to SDS-PAGE and Western blotting. EGFR levels were higher in the nuclear fraction after EGF stimulation than in unstimulated cells. Loading controls were ␣-tubulin (non-nuclear marker) and lamin B (nuclear marker). Densitometric analysis of Western blots showed that treatment with 25 ng/ml EGF led to 3.2 times more nuclear EGFR being present than in untreated cells. B and C, time course of EGFR nuclear localization. MDA-MB-468 cells were stimulated with 100 ng/ml EGF (B) or 25 ng/ml EGF (C) at 37°C for various times, after which the cells were fractionated and the nuclear and non-nuclear fractions were subjected to Western blotting with the indicated antibodies. C, plot of nuclear EGFR levels in the cells from B over time. short exp., 5 times shorter.
(RRRHIVRKRTLRR, amino acids 645-657) ( Fig. 2A) that was conserved in the EGFR family of proteins, and we thus proposed it as a putative NLS (9, 13, 18, 52). To test whether the EGFR protein contains such an NLS, we labeled cDNA fragments of EGFR ICD, JM, or the putative tripartite NLS with an enhanced GFP tag and then transiently transfected the constructs into HeLa cells. The GFP-ICD localized mostly in the nucleus with only weak cytoplasmic expression (Fig.  2B). Both the GFP-JM and GFP-NLS exhibited distinct nuclear accumulation, but the GFP control vector displayed diffuse cytoplasmic localization accompanied by a basal level of nuclear accumulation. To better quantify the cytoplasmic-tonuclear distribution of GFP fusion proteins, cells expressing the different GFP fusion proteins were subjected to biochemical fractionation. The results revealed that most of the GFP-JM and GFP-NLS proteins were in the nuclear fractions (Fig.  2C), consistent with our findings from confocal microscopy. To see whether the tripartite NLS could direct cytoplasmic proteins into the nucleus without the influence of endogenous EGFR, we tested Chinese hamster ovary cells, which do not express EGFR, and obtained similar results (data not shown). These results suggest that the nuclear localization patterns observed were not caused by endogenous EGFR. They further indicate that this tripartite 13-amino-acid sequence, containing three clusters of two or three continuous basic amino acid residues, functions as an NLS to drive GFP-tagged protein trafficking to the nucleus, similar to the classical monopartite or bipartite NLS.
All Three Clusters of Basic Amino Acids of the Tripartite NLS Are Required for Efficient Nuclear Translocation-The nucleocytoplasmic trafficking of large molecules is normally mediated by classical NLSs (22). To determine whether the tripartite NLS for EGFR functions as a bona fide NLS for transferring large cytoplasmic proteins into the nucleus, we fused this 13-amino-acid sequence to the C terminus of a known cytoplasmic protein, chicken PK, which had been tagged with GFP (40). As expected, we found that the control GFP-PK was predomi- nantly found in the cytoplasm, whereas the GFP-PK-NLS protein was clearly visible in the nucleus (Fig. 3A, a versus b). A positive control, GFP-PK-LTNLS (which expressed a fusion protein containing the monopartite NLS of SV40 LT (PKKKRKV)), was also expressed in the nucleus (data not shown). These results indicated that the EGFR NLS is a functional NLS that can lead a large cytoplasmic protein into the nucleus.
To determine whether all three of the basic amino acid clusters within the tripartite NLS are required for nuclear localization, we generated three mutants in which one of the three clusters had been mutated to alanines (Fig. 3C, NLSm1, NLSm2, and NLSm3). This substitution reduced but did not completely abolish nuclear localization; all of the mutants displayed significant cytoplasmic expression but had detectable nuclear signals of GFP-PK (Fig. 3A, c-e). Similar results were obtained with the GFP-NLS protein; mutations in any one of the three clusters of the tripartite NLS affected its nuclear translocation activity (data not shown). The biochemical fractionation experiments revealed decreases in the nuclear expression of the EGFR tripartite NLS mutants in comparison with wild-type tripartite NLS (Fig. 3B). An EGFR tripartite NLS mutant containing the first two clusters of basic amino acids but lacking the third cluster (GFP-PK NLSd3) also showed strong cytoplasmic expression but weaker nuclear expression than GFP-PK-NLS (Fig. 3A, b versus f and Fig. 3B). Collectively, these results dem-onstrate that the 13-residue sequence (amino acids 645-657 of EGFR) can function as an independent NLS and that all three clusters of basic residues are critical for this activity.
We also generated site-directed mutations in the basic amino acid clusters of GFP-ICD to determine whether that tripartite NLS is essential for EGFR protein nuclear targeting. The construct with mutations in the first two clusters (GFP-ICD-NLSm12) displayed a weaker nuclear expression pattern compared with that of GFP-ICD-NLSm1 (Fig. 4A), indicating that the NLS within the ICD is required for the nuclear localization of EGFR protein. We also examined the subcellular distribution patterns of different epitope-tagged fusion proteins by biochemical fractionation. ICD-Myc was easily detected in nuclear fractions, but both NLS mutants (NLSm1 and NLSm12) showed decreased nuclear expression (Fig. 4B). To further examine NLS functional activity in the fulllength EGFR protein, we constructed plasmids expressing EGFR with NLS mutations (mutations in the first basic cluster, NLSm1; mutations in the first and second basic clusters, NLSm12; deletion of entire NLS, dNLS). Cells expressing the different NLS mutants were subjected to biochemical fractionation and analyzed by immunoblotting (Fig.  4C). All three EGFR mutants showed similar cytoplasmic expression, but the EGFR NLSm1 mutant showed a 64% decrease in nuclear EGFR, and the EGFR NLSm12 mutant showed a 78% decrease in nuclear EGFR level, relative to the wild-type control. The expression pattern of the EGFR dNLS mutant was similar to that of the EGFR NLSm12 mutant, with barely detectable nuclear EGFR. These results provide further evidence that the tripartite NLS is essential for the nuclear localization of the EGFR protein.
Tripartite NLS Sequences Are Conserved in the EGFR Family but Vary in Their Nuclear Targeting Activity-Although an NLS has been identified in the C-terminal region of ErbB3 (11), we found that our newly identified tripartite NLS was highly conserved in the JM region of EGFR family members (Fig. 5A). In ErbB2 and ErbB4, amino acids within this JM region have been shown to have NLS activity (10,14,21,53). To examine whether this novel tripartite NLS is equally capable of moving large cytoplasmic proteins into the nucleus in all EGFR family members, we fused the GFP-PK protein with each of the tripartite NLS sequences within the JM region of each EGFR family member, transfected HeLa cells with the constructs, and examined the subcellular localization of the tagged protein with con- A, subcellular localization of GFP-PK, GFP-PK-NLS, and GFP-PK-NLS mutant proteins. HeLa cells were transiently transfected with the indicated GFP-PK fusion constructs. GFP-PK remained in the cytoplasm, but the tripartite NLS of human EGFR effectively transported the GFP-PK-NLS fusion protein into the nucleus. Top rows show the GFP signal; bottom rows show merged images of GFP (green) and DNA (red). The images represent the pattern expressed by Ͼ70% of the GFP-PK fusion protein-expressing cells. The molecular mass of all GFP-PK and GFP-PK fusion proteins was apparently exceeded by the passive-diffusion size of the nuclear pore complex. Scale bar, 10 m. B, biochemical fractionation and Western blotting of HeLa cells from A. C, summary of EGFR mutants. Amino acid residues are indicated with the single letter code. The basic amino acids are underlined. Red letters indicate the mutated amino acid residues within the NLS.
focal microscopy. Interestingly, the nuclear import activity varied among the EGFR family members (Fig. 5B). Specifically, the GFP-PK proteins fused with the different tripartite NLSs of EGFR, ErbB2, and ErbB4 showed similar subcellular expression patterns. The activity of the ErbB3 tripartite NLS was much weaker, but this NLS could still move GFP-PK proteins into the nucleus to a greater extent than could GFP-PK alone. To quantify the respective tripartite NLS activities, we subjected the transfected cells to biochemical fractionation (Fig. 5C). We found that the relative nuclear expression of GFP-PK-ErbB3 tripartite NLS was ϳ20% of that of GFP-PK-EGFR tripartite NLS; the nuclear expression of GFP-PK-ErbB4 tripartite NLS was greater than that of GFP-PK-EGFR tripartite NLS or that of GFP-PK-ErbB2 tripartite NLS. Thus, in the current study, we have characterized a novel tripartite NLS that triggers the nuclear translocation of EGFR family members.

DISCUSSION
In this report, we characterized a putative NLS within the JM region of EGFR family members and found that these NLSs could independently mediate the movement of cytoplasmic proteins into the nucleus. Most nuclear proteins are targeted to the nucleus by mono-or bipartite basic amino acid sequences in an NLS (22,23,54). The typical monopartite NLS is a cluster of basic residues starting with proline and followed by five residues, of which at least three are either lysine or arginine. The bipartite pattern begins with two basic residues followed by a 10-residue spacer and finally a cluster in which at least three of five residues are lysine or arginine. The SV40 LT (25) and p53 protein (26,27) are well characterized examples of monopartite and bipartite NLSs, respectively. In addition to these two types of NLSs, other non-classical NLSs have been identified in cellular and viral proteins; an example is the M9 sequence, a 38-amino-acid-long fragment identified in the hnRNP (heterogeneous nuclear RNP) A1 and A2 proteins (55,56).
The novel tripartite NLS we found in EGFR family members contains three clusters of basic amino acids. Mutations in any one of these three clusters affected the nuclear localization activity of the NLS (Fig. 3), and double mutations of the first two clusters completely abolished that activity (Fig. 4, B and C). Notably, even though the tripartite NLS was conserved among EGFR family members, the nuclear localization activity varied among those family members (Fig. 5). The activity of the tripartite NLS of ErbB4 was the strongest, followed by EGFR and ErbB2; that of ErbB3 was the weakest. The continuity of three or four basic amino acids in the first or second cluster could be critical for mediating the nuclear localization of these EGFR family members. Interestingly, except for ErbB4, the EGFR family members could all be translocated into the nucleus as full-length proteins; the ErbB4 in the nuclei was present as an intracellular fragment (10,57). In addition to the full-length ErbB2, an alternative initiation form of ErbB2 has also been detected in the nucleus (58). ErbB3 has an additional monopartite NLS identified in the C-terminal region (11); it is possible that the tripartite and monopartite NLSs have additive effects on the nuclear localization of the full-length ErbB3 protein.
Others have reported that the first two clusters of the ErbB4 tripartite NLS were sufficient to target GFP protein into the FIGURE 4. Mutations of NLS affect EGFR nuclear targeting. A, HeLa cells were transfected with DNA encoding the indicated GFP-EGFR ICD fusion protein or NLS mutants. Localization of the GFP fusion proteins was analyzed by confocal microscopy as described in the legend to Fig. 2. Scale bar, 10 m. B, HeLa cells were transfected with DNA (8 g) encoding EGFR full-length, ICD, or NLS mutants (with a Myc tag at the C terminus). After 40 h of transfection, the cells were biochemically fractionated and subjected to Western blotting with the indicated antibodies. Cotransfection of a plasmid encoding GFP-JM (2 g) was used as a transfection efficiency control. C, full-length EGFR constructs (8 g) and a plasmid encoding GFP-JM (2 g) were transiently transfected into HeLa cells. After 40 h of transfection, the cells were biochemically fractionated into nuclear and non-nuclear fractions. Nuclear (40 g) and non-nuclear (15 g) protein extracts were subjected to SDS-PAGE and Western blotting. Exogenously expressed EGFR was detected by means of a Myc tag at the C terminus. Relative amounts of nuclear and non-nuclear EGFR were quantified and normalized against the amounts of lamin B (nuclear marker) and tubulin (non-nuclear marker).
nucleus (21). We found similar results with the EGFR tripartite NLS; however, when only the first two clusters of EGFR tripartite NLS were fused with GFP-PK, a large cytoplasmic protein nuclear targeting activity was weak at best (Fig. 3). Therefore, moving macromolecules (such as EGFR) from the cytoplasm into the nucleus seems to require the full tripartite NLS. Notably, truncated RTKs such as ErbB2 and ErbB4 can be translocated into the nucleus through different mechanisms (e.g. alternative initiation of translation and cleavage) (10,58). In our study, we did not observe truncation of nuclear EGFR (Fig. 4B). However, it is not clear whether the truncated forms of EGFR and ErbB3 could be detected under different conditions.
In previous studies, the nuclear transport of EGFR was shown to involve endocytosis and importin ␤1 (51). Details of the molecular mechanisms underlying endocytic internalization and endosomal sorting of ErbB2 have been reported (14). Activation of RTKs at the plasma membrane typically results in rapid endocytosis and subsequent passage of the receptors through a network of endosomal compartments. One puzzle that remains is how the membrane-bound receptor escapes from the endosome membrane. One hypothetical model posits that EGFR contained in endosomes could be released into the cytoplasm via an endoplasmic reticulum-associated degradation system (59), and one recent study showed that the Sec61 translocon was involved in the nuclear translocation of EGFR (20). Another unresolved question is how long it takes for RTKs to move from the plasma membrane to the nucleus upon ligand stimulation. In our study and others (9, 51), EGFR nuclear accumulation was detectable within 10 min of EGF stimulation. One group showed that EGFR can lead to membrane ruffling and the formation of vesicletubule structures toward the cell center as early as 5 min after EGF stimulation (60). It will require extensive investigation to establish whether different types of endocytosis mechanisms are involved in EGFR nuclear localization. Our results further support the concept that membrane RTKs not only act at the cell membrane to elicit traditional signaling pathways upon stimulation by their ligands, but they can also be translocated into the nucleus.
In this study, we have demonstrated that the tripartite NLSs in the EGFR protein family can function to independently target the nuclear transport of heterogeneous proteins. In addition to EGFRs, many other membrane receptors have been detected in the nucleus, including TrkA/nerve growth factor receptor, transforming growth factor-␤ type I receptor, vascular endothelial growth factor receptor-2, fibroblast growth factor receptor, interleukin-1 receptor, and interferon-␥ receptor (61)(62)(63)(64)(65)(66). Ligands for most of these receptors have also been found in the nucleus (65,67,68 (Fig. 5A). This consensus tripartite NLS seems to be unique for the EGFR family, as it is not present in the above-mentioned receptors. The 13-amino-acid tripartite NLS of the EGFR protein has been shown to be required for receptor dimerization (69). However, other studies showed that mutation of EGFR tripartite NLS or deletion of the ErbB2 tripartite NLS did not affect receptor protein membrane localization and activation of MAPK signaling (13,14). Therefore, further studies are needed to determine whether dimerization of EGFR protein might contribute to its nuclear localization or is required for NLS functional activity.
The functional significance of cell surface receptor nuclear localization has been described in several recent reviews (15)(16)(17)(18). EGFR was initially shown to bind chromatin (41); subsequently, nuclear EGFR was shown to bind to an AT-rich minimal consensus sequence in the promoter of cyclin D1 (9), and nuclear HER2 to the specific sequence HAS (HER2-associated sequence) (12). Nuclear EGFRs may also regulate transcription of specific genes by physically interacting with various nuclear FIGURE 5. Tripartite NLS sequences are conserved within the EGFR family but vary in their nuclear targeting activity. A, sequence elements of the tripartite NLS regions of the EGFR family members. Basic residues are shown in red. These accession-numbered proteins are from the National Center for Biotechnology Information, P00533 (EGFR), P46246 (ErbB2), P21860 (ErbB3), and Q15303 (ErbB4), except that the numbers for the EGFR sequence do not count the signal peptide. B, representative images of HeLa cells transfected with DNA encoding different GFP-PK tripartite (tp) NLS constructs illustrate the subcellular localization of the various fusion proteins. Localization of the GFP-PK fusion proteins was analyzed by confocal microscopy as described in the legend to Fig. 3. Scale bar, 10 m. C, HeLa cells were transfected with the indicated DNA constructs for 36 h followed by biochemical fractionation and Western blotting with the indicated antibodies. Relative amounts of fusion proteins in the nuclei were determined by densitometry.
partners. For example, we found that nuclear EGFR in complex with STAT3 regulates the transcription of inducible nitric-oxide synthase (13) and interacts with E2F1 to activate the b-Myb gene (70). Others have shown that nuclear ErbB4 binds to STAT5a and activates the ␤-casein gene (21) and the estrogen receptor corepressor N-CoR in the developing brain (57). Notably, we found that the novel tripartite NLS not only drove GFP protein into the nucleus but also targeted the fusion protein to certain subnuclear structures (Fig. 2B). The GFP-ICD, in contrast, was excluded from the subnuclear structures. It is interesting to speculate that the sequence motif within the EGFR protein or other cellular factors might control or contribute to EGFR subnuclear shuttling and DNA binding. Because nuclear EGFR may contribute to the development and progression of cancer, developing small molecules to modulate its tripartite NLS activity could be a worthwhile endeavor.