Identification of the nuclear localization signals within the Epstein-Barr virus EBNA-6 protein

EpsteinBarr virus (EBV) is a DNA tumour virus that has the capacity to infect and transform B-lymphocytes. Following infection, the co-ordinate expression of a number of viral nuclear antigens (EBNAs) and membrane proteins (LMP-1 and -2) acts to maintain B-cell growth and transformation (for review see Kieff, 1996). EBNA-6 is essential in the transformation and immortalization process.

Analysis of the EBNA-6 amino acid sequence has revealed features common to many viral and cellular transcription factors. These include a region which resembles a basic DNA-binding domain adjacent to a potential leucine-zipper motif (b-zip) and a transactivation domain rich in glutamine/proline residues, which has similarities to the mammalian transcription factor Sp1 (Bain et al., 1996; Radkov et al., 1997; Marshall & Sample, 1995). The EBNA-6 protein has also been identified as an immortalizing oncoprotein which can cooperate with activated (Ha-)ras in cotransformation assays and can override Rb-mediated pathways (Allday et al., 1993). EBNA-6 is a hydrophilic, proline-rich, charged protein that is targeted exclusively to the cell nucleus, and cellular fractionation experiments have shown that EBNA-6 is associated with the nuclear matrix, and to a lesser extent is present in the nucleoplasm (Petti et al., 1990). Sample & Kieff (1990) showed that EBNA-6 localized to subnuclear granules within the cell nucleus.

Subsequent to translation, the fate of a protein depends largely on whether its amino acid sequence contains sorting signals directing the protein to specific cellular locations. The active transport of proteins in both directions across the nuclear envelope requires the presence of specific targeting sequences within the protein (Gorlich et al., 1996). There are at least three types of nuclear localization signals (NLS) on proteins and these signals are characteristically rich in the basic amino acids lysine and arginine and usually contain proline (Dingwall & Laskey, 1991). The first type of NLS is a continuous stretch of amino acids, which can be located almost anywhere in the protein sequence. The archetypal NLS is that of the SV40 large tumour antigen (PPKKRKV) (Huber et al., 1996). The second type of NLS is a bipartite sequence, or a signal patch, that consists of a three-dimensional arrangement on the protein surface. An example of this is Xenopus laevis nucleoplasmin, where there are two clusters of basic amino acid residues separated by an intervening 1012 aa spacer (Dingwall & Laskey, 1992). The third type are less well conserved sequences with few basic residues such as that of the adenovirus E1A (KRPRP) (Lyons et al., 1987). However, nuclear localization signals can be masked by phosphorylation close to or within an NLS. This phosphorylation inactivates the NLS through charge or conformational effects (Hennekes et al., 1993; Ohta et al., 1989). NLS are not cleaved off after transport into the nucleus, which is presumably because nuclear proteins need to be imported repeatedly, once after every cell division (Chaudhary & Courvalin, 1993). Proteins containing NLS are recognized by the transport machinery and are imported through the nuclear pore complex, whereas proteins lacking NLS remain in the cytoplasm (Moll et al., 1991).

EBNA-6 has several localized concentrations of arginine or lysine residues that are potential signatures of nuclear localization and although it is recognized that EBNA-6 is targeted to the nucleus, the NLS responsible have not been identified.

Construction of GFP-EBNA-6 deletion mutants.
The correct orientation and insertion of fragments in each of the deletion constructs of EBNA-6 were confirmed by DNA sequencing and immunoblotting.

pBS-GFP-E6.
The EBNA-6 cDNA fragment (B95-8 virus) was excised from vector pGBT9-E6 (Young et al., 1997) using BsrGI and BstEII. The excised fragment was then Klenow treated and ligated into pBS-GFP that had previously been restricted with BsrGI and Klenow treated. (pBS-GFP was a gift from M. Vogel, Medicine, Microbiology and Hygiene, Regensburg University; pBS-GFP-E6 was originally prepared by D. Young, QIMR.)

pBS-GFPL.
The pBS-GFP plasmid was adapted to include a linker containing the restriction enzyme sites BsrGI, StuI, BamHI, EcoRV, HpaI, XhoI and NotI. This was prepared by annealing 1 µg of two self-complementary 41 base oligomers (Custom-made; Life Technologies, Australia) and then ligating this fragment into pBSGFP that had been restricted with BsrGI and NotI.

pBS-GFPL-E6Δ1206.
Plasmid pBS-GFP-E6 was cut with HpaI and NotI and the E6 fragment was inserted into pBS-GFPL cut with HpaI and NotI.

pBS-GFP-E6Δ1395/601992.
The pBS-GFPL-E6Δ1206 construct was digested with ClaI and NarI, resulting in a 618 bp fragment, which was Klenow treated and then ligated to pBSGFPL that had been HpaI restricted. This resulted in a fusion of GFP with aa 396600 of EBNA-6 containing NLS24.

pBSGFP-E6Δ184992.
The vector pBS-GFP-E6 was digested with XbaI, Klenow treated, and then digested with EcoRV and religated resulting in a fusion of GFP with aa 22183 of EBNA-6 which contained NLS1.

pEBO-GFP-E6Δ396625.
NLS24 were removed from EBNA-6 by excising a 696 bp fragment from the vector pBSGFP-E6 by restriction with ClaI and NarI. The restricted ends were then Klenow treated and religated. The plasmid was then digested with HindIII and NotI and the EBNA-6 region inserted into EBO-pLPP, restricted with the same enzymes. This resulted in removal of aa 396625 of EBNA-6.

pBS-GFPL-E6Δ1206/396625.
The pBSGFPL-E6Δ396625 construct was digested with HpaI and NotI, and the resulting fragment was inserted into pBS-GFPL that had been restricted with HpaI and NotI. This resulted in the expression of a fusion of GFP with EBNA-6 aa 207395 joined with aa 626992 and contained NLS5.

pBS-GFP-E6Δ1206/396625/941992.
The 1·5 kb StuI fragment, excised from pBS-GFPL-E6Δ1206/396625, was then ligated into pBS-GFPL previously restricted with HpaI. This resulted in the expression of a GFP fusion linked to aa 207395 joined with aa 626940 of EBNA-6, which lacked all computer-predicted NLS.

Site directed mutagenesis.
In vitro mutagenesis of double-stranded DNA templates was performed as described in Sambrook & Russell (2001) with some modifications. Briefly, a PCR reaction was performed with 550 ng plasmid DNA and 125 ng of each primer using Pfu Turbo polymerase under the following buffer conditions: 20 mM Tris/HCl pH 8·8, 10 mM KCl, 10 mM (NH₄)₂SO₄, 2 mM MgSO₄, 0·1 % Triton X-100, 0·1 mg nuclease-free BSA ml^-1 in a 50 µl reaction. Cycling conditions were as follows: 95 °C 30 s, 55 °C 1 min, 68 °C 2 min per kb of plasmid DNA for 1618 cycles, followed by 72 °C for 10 min. PCR primers are listed in Table 1. Following the PCR reaction plasmid DNA was digested with 20 U DpnI for 3 h at 37 °C. The PCR mixture (4 µl) was then transformed into E. coli DH5α competent cells by electroporation and the bacteria were then plated onto LB plates containing ampicillin. Resultant bacterial colonies were selected, plasmid DNA was prepared and confirmation of mutagenized bases determined by restriction enzyme digestion and DNA sequencing. A list of the resultant NLS mutations is shown in Table 2.

Table 1. Primers used for site-directed mutagenesis

Table 2. Summary of EBNA-6 NLS mutants Amino acid residues are indicated with the single-letter code. Bold letters indicate the mutated amino acid residues within each NLS.

Cell lines, maintenance and DNA transfection.
HeLa cells were maintained in RPMI 1640 supplemented with 10 % foetal calf serum, benzylpenicillin (0·7 mg ml^-1) and streptomycin (1 mg ml^-1) at 37 °C in a 5 % CO₂ atmosphere. HeLa cells were transfected using ExGen 500 (Progen) according to the manufacturer's protocol. Briefly, cells were plated 24 h prior to transfection at a cell density of 2·4x10⁵ cells per well in 3 ml RPMI 1640 containing 10 % FCS. Plasmid DNA (5 µg) was diluted with 300 µl 150 mM NaCl and then 15·5 µl ExGen 500 was added. After vortexing and incubation for 10 min at room temperature, the mixture was added directly to the HeLa cells. The plates were centrifuged at 1200 r.p.m. for 5 min and then placed into an incubator. Cells were analysed for transient gene expression at 24 h after incubation at 37 °C in a 5 % CO₂ atmosphere.

Direct fluorescence microscopy.
HeLa cells expressing GFP-E6 fusion proteins were grown on coverslips and fixed with ice-cold acetone (2 min, -20 °C), and then mounted with Vectashield mounting medium onto glass slides. HeLa cells were scanned using either a Bio-Rad MRC 600 confocal microscope with images acquired using COMOS (Confocal Microscope Operating Software Version 6.03) or a Leica confocal microscope, model TCS SP2 and images recorded using Leica Confocal Software. The acquired images were then analysed and processed for presentation using CAS (Confocal Assistant Software Version 3.10), in which single optical sections are shown.

Nuclear localization signals in EBNA-6
Five potential NLS, as defined by either the bipartite consensus (Robbins & McMichael, 1991) or the Nakai consensus (Nakai & Kanehisa, 1992), were identified within the EBNA-6 protein sequence using PSORT (Pedro's BioMolecular Research Tools ). NLS1, which comprises multiple overlapping pattern 4 sequences, is located within the N-terminal region of EBNA-6, aa 7280 (RIRRRRRRR). The second NLS, aa 412418 (KKPRK), has overlapping pattern 4 and pattern 7 sequences as has the third NLS (NLS3) at aa 494500 (PPSRRRR). A bipartite NLS (NLS4) is located at aa 533549 (RKHQDGFQRSGRRQKRAA). The fifth NLS, which has overlapping pattern 4 and pattern 7 sequences, is located at the C-terminal end of the protein (NLS5) and encompasses aa 939945 (PKKRPRVE).

Constructs of EBNA-6, linked to GFP, were prepared such that they contained at least one of these NLS, as well as one construct which lacked all of the computer-predicted NLS, to determine whether other previously unrecognized NLS were present in the protein. The integrity of the constructs was determined by DNA sequencing and immunoblotting. To ensure easy visualization of both the nucleus and cytoplasm the constructs were transfected into HeLa cells. All of the constructs were transiently expressed in HeLa cells and the cellular location of the fusion proteins was determined by confocal microscopy. In contrast to GFP alone, which was distributed diffusely in both the cytoplasm and the nucleus, each of the constructs containing a predicted NLS localized to the nucleus of cells. However, the fusion protein in which all computer-predicted NLS were removed (GFP-E6Δ1206/396625/941992), was present only in the cytoplasm of cells (Fig. 1). These results indicated that the predicted NLS present in EBNA-6 were likely to be functional and that there were unlikely to be any additional NLS present within the EBNA-6 coding region (at least within the sequence included in GFP-E6Δ1206/396625/941992).

(37K): Fig. 1. Schematic representation of the full-length EBNA-6 and a series of EBNA-6 deletions fused to GFP and their subcellular localization. The locations of computer-predicted NLS are indicated. HeLa cells were transiently transfected with constructs expressing the fusion proteins and were analysed 24 h post-transfection by confocal fluorescence microscopy.

Mutagenesis of the EBNA-6 NLS
To determine if the predicted NLS within each of the GFP-EBNA-6 constructs was responsible for their nuclear localization, site-directed mutants in the basic residues within each of the predicted NLS were generated. The mutations generated within each of the NLS are shown in Table 2. Because of the number of consecutive arginine residues present in NLS1 two mutants were generated, one with three of the arginine residues replaced with alanine residues and the second mutant with five of the arginine residues replaced with alanine residues. Each of the mutagenized constructs was transiently transfected into HeLa cells and the cellular localization of the mutant proteins was determined by confocal microscopy (Fig. 2). Substitution of the arginine residues at position 75, 77 and 78 by alanine residues (NLS1a mut) had no effect upon nuclear localization and additional substitutions of arginine residues 76 and 79 (NLS1b mut) was required to abrogate the nuclear localization of the GFP-E6aa184992 protein.

(23K): Fig. 2. Cellular localization of GFP-EBNA-6 deletions following mutation of NLS sequences. HeLa cells were transiently transfected with plasmids expressing each of the EBNA-6 constructs and 24 h post-transfection the subcellular localization of each of the fusion proteins was determined by confocal fluorescence microscopy.

Mutation of NLS3 had no effect upon the nuclear localization of the GFP-E6aa396600 protein whereas substitution of lysine residue 414 and arginine residue 416 within NLS2 by alanine and glycine residues, respectively, was sufficient to abrogate the nuclear localization of the GFP-E6aa396600 protein. This result confirmed that the sequence KKPRK was a functional NLS and that NLS3 and NLS4 were nonfunctional. Substitution of arginine residue 941 by glycine and proline residue 942 by alanine was sufficient to destroy NLS5 and prevent the nuclear localization of the GFP-E6aa643992 protein. EBNA-6 is required for the transformation of B-lymphocytes and is likely to function as a transcriptional regulator as sequence analysis revealed a region homologous to the basic leucine-zipper motif that is found in many mammalian transcription factors. Studies have shown changes in viral and cellular genes, such as IL-1β (Krauer et al., 1998), pleckstrin (Kienzle et al., 1996), LMP1 and CD21, a B-cell activation antigen (Sample et al., 1994), following expression of the EBNA-3 family proteins. EBNA-6 has also been shown to interact with a variety of cellular proteins, including the human metastatic suppressor protein Nm23-H1, a 152 aa 17 kDa nuclear protein highly conserved in eukaryotes (Subramanian et al., 2001), histone deacetylase through the N terminus of EBNA-6 (Radkov et al., 1999), RBP-Jκ/2N (Robertson et al., 1995) and prothymosin alpha (ProTa), which is a small non-histone highly acidic nuclear protein that localizes to regions of the nucleus that are involved in transcription (Subramanian & Robertson, 2002). EBNA-6 can activate the human B-myb promoter through the E2F response element, can co-operate with (Ha-)ras in co-transformation assays and can override a pRb-mediated pathway that inhibits proliferation. EBNA-6 also plays a role in the disruption of cell-cycle checkpoints (Parker et al., 1996).

The EBNA-6 protein is targeted exclusively to the cell nucleus, and cellular fractionation experiments have shown that it is present in the nucleoplasm and associates with the nuclear matrix while localization studies demonstrate that it localizes to discrete subnuclear granules within the cell nucleus (Petti et al., 1990; Sample & Kieff, 1990). There are five potential NLS containing arginine and lysine residues within EBNA-6. The data presented here show that NLS1, NLS2 and NLS5 are all independently capable of targeting protein to the nucleus. The pattern 4 C-terminal NLS5 sequence (KRPR) is the shortest known nuclear targeting signal, and is also found in EBNA-1, EBNA-2, and in adenovirus E1A (Lyons et al., 1987; Ambinder et al., 1991; Le Roux et al., 1993). The removal of all NLS (pGFP-E6Δ1206/396625/941992) resulted in the truncated EBNA-6 protein being present solely throughout the cytoplasm, indicating that there are no additional functional NLS within residues 207395 or 626940 (Fig. 1). In addition, mutation of NLS1, NLS2 and NLS5 resulted in GFP-EBNA-6 fusion proteins being cytoplasmic, indicating that no additional functional NLS were present in sequences 22183, 396600 or 941992 (Fig. 2). Taken together the results revealed that, other than NLS1, NLS2 and NLS5, it is unlikely that EBNA-6 contains any other functional NLS.

Two types of EBV exist (type-I or type-II) which show sequence divergence within the genes encoding the EBNA-LP, -2, -3, -4 and -6 gene products (Adldinger et al., 1985; Dambaugh et al., 1984; Sample et al., 1986; Sculley et al., 1989). Since EBNA-6 is a nuclear protein it might be expected that there would be conservation of the functional NLS between the two virus types. Computer analysis of the Ag-876 type-II EBNA-6 protein sequence showed conservation of NLS14 but not NLS5 (Fig. 3), suggesting that the type-II EBNA-6 protein probably only has two functional NLS (NLS1 and NLS2). Intriguingly, the sequences for the two nonfunctional NLS (NLS3 and NLS4) were almost perfectly conserved in the type-II EBNA-6 protein, raising the possibility that they may serve another function. NLS3 is contained within the repression domain of EBNA-6, while NLS4 is within the DP-103 interaction domain and these sequences may play crucial roles within each of these domains rather than functioning as NLS.

(34K): Fig. 3. Comparison of EBNA-6 NLS in Type I and II isolates of EBV. Amino acid residues are indicated with the single-letter code. Bold letters indicate differences in sequence between each NLS. Underline indicates that this NLS was not computer-predicted.

It is not understood why a single protein contains multiple NLS. However, it has been suggested that multiple NLS may function more efficiently (Roberts et al., 1987; Knauf et al., 1996), or that different NLS may have different specificities in different cell types (Liu et al., 1998). Alternatively, some NLS can also function under different conditions, such as one of the multiple NLS in XPG nuclease (Xeroderma pigmentosum type G), which can regulate the localization of the protein to the nucleus following UV irradiation (Knauf et al., 1996). Proteins with multiple NLS include the tumour suppressor menin (Guru et al., 1998), the high-mobility group transcription factors SRY and SOX9 (Sudbeck & Scherer, 1997), the proto-oncogene c-Abl (transforming gene of Abelson murine leukaemia virus) (Wen et al., 1996), tumour suppressor p53 (Shaulsky et al., 1990) and human c-Myc, which contains one strong and one weak NLS (Dang & Lee, 1988). The presence of multiple NLS in EBNA-6 could be needed if, prior to transport into the nucleus, EBNA-6 interacts with proteins in the cytoplasm or if EBNA-6 is modulated by phosphorylation, either of which could mask one or more of the NLS. Other possible reasons for multiple NLS could involve differential cellular regulation, or simply enhancement of nuclear accumulation (Roberts et al., 1987). The authors would like to acknowledge Paula Hall and Grace Chowjnowski, Queensland Institute of Medical Research, for assistance with confocal microscopy. This work was supported by grants from the NHMRC, Australia.