Nuclear Import of Hepatic Glucokinase Depends upon Glucokinase Regulatory Protein, whereas Export Is Due to a Nuclear Export Signal Sequence in Glucokinase*

Hepatic glucokinase (GK) moves between the nucleus and cytoplasm in response to metabolic alterations. Here, using heterologous cell systems, we have found that at least two different mechanisms are involved in the intracellular movement of GK. In the absence of the GK regulatory protein (GKRP) GK resides only in the cytoplasm. However, in the presence of GKRP, GK moves to the nucleus and resides there in association with this protein until changes in the metabolic milieu prompt its release. GK does not contain a nuclear localization signal sequence and does not enter the nucleus in a GKRP-independent manner because cells treated with leptomycin B, a specific inhibitor of leucine-rich NES-dependent nuclear export, do not accumulate GK in the nucleus. Instead, entry of GK into the nucleus appears to occur via a piggy-back mechanism that involves binding to GKRP. Nuclear export of GK, which occurs after its release from GKRP, is due to a leucine-rich nuclear export signal within the protein (300ELVRLVLLKLV310). Thus, GKRP appears to function as both a nuclear chaperone and metabolic sensor and is a critical component of a hepatic GK translocation cycle for regulating the activity of this enzyme in response to metabolic alterations.

Hepatic glucokinase (GK) 1 plays an essential role in hepatic glucose metabolism by catalyzing the high S 0.5 phosphorylation of glucose (1)(2)(3)(4)(5)(6). Hepatic GK activity is regulated by both transcriptional and post-transcriptional mechanisms (7)(8)(9)(10)(11)(12). The post-transcriptional regulation of hepatic GK involves binding to the GK regulatory protein (GKRP). The interaction of these two proteins, which is stimulated by fructose 6-phosphate and inhibited by fructose 1-phosphate, competitively inhibits the phosphorylation of glucose by GK (13). Under basal glucose conditions (ϳ5.5 mM) hepatic GK is bound largely to GKRP. However, after exposure to either high glucose (10 -30 mM) or fructose (50 M to 1 mM), GK is released from GKRP and exists in an unbound condition (14,15). Immunocytochemical localization studies have established that GKRP exists largely in the nucleus of hepatocytes (16), whereas the subcellular location of GK varies depending on the metabolic state of the cell. In the cytoplasm GK exists in an unbound state. However, in the nucleus GK is bound to GKRP (16 -18).
The nuclear import and export of proteins occurs via the nuclear pore complex, which has an exclusion limit for proteins of 40 -60 kDa (19 -22). Proteins that are smaller than the exclusion limit are generally able to diffuse freely both into and out of the nucleus, whereas proteins greater than this size must be actively transported. Active transport through the nuclear pore complex involves the recognition of either nuclear localization signal (NLS) or nuclear export signal (NES) sequences (23,24). The mechanism(s) whereby hepatic GK, which has a molecular mass of 50 kDa, is able to both enter and leave the nucleus has not been elucidated. Indeed, until recently the enzyme was generally assumed to reside only within the cytoplasm of cells.
To determine the molecular mechanisms involved in the nuclear-cytoplasmic translocation of GK, we studied the role of GKRP in the translocation of GK fusion proteins using both HeLa cells and yeast as heterologous cell systems. These studies clearly demonstrate that GK lacks a NLS and thus is unable to enter the nucleus by itself. Instead, entry into the nucleus depends on GKRP, presumably by a piggy-back mechanism. In addition, identification of a NES sequence within GK suggests that upon release of binding to GKRP in response to metabolic cues, the enzyme is exported from the nucleus via an active process. Together, these results suggest a nuclear-cytoplasmic transport cycle for GK in the hepatocyte that involves both GKRP and a NES sequence in GK. GKRP appears to play multiple roles in this transport cycle. In addition to its established function as a metabolic sensor and allosteric inhibitor of GK, GKRP also appears to function as a nuclear chaperone that binds to and carries GK into the nucleus. TC (XhoI and EcoRV ends); and G5, 5Ј-GAT* ATC* GTG CGC CGT GCC TGT GAA AGC G and 5Ј-GAA TTC CTG GGC CAG CAT GCA AGC CTT (EcoRV and EcoRI ends). The altered bases are indicated by an asterisk, and the underlined bases denote a new restriction site. By utilizing the naturally occurring PstI and BalI sites, as well as the introduced sites, GK cassettes designated G1-G5 were obtained. Similarly, DNA fragments from the C-terminal half of rat HKII (Gen-Bank TM accession number M68971) corresponding to the GK sequence cassettes were amplified by reverse transcription-PCR of RNA from AR42J cells, a rat pancreatic tumor cell line (ATCC: CRL1492). The primers used for the preparation of HKII cassettes, designated H1-H5, are as follows: H1, 5Ј-CTG CAG CAC GAG CAG CTT CTG GAG GTT and 5Ј-GGT ACC ATG CAT AAC CTC CTG TGG (PstI and KpnI ends); H2, 5Ј-GGT ACC GGG GAA GAG CTC TTC GAC CAC and 5Ј-T GGC CAC TGT GTC ATT CAC CAC GGC AAC (KpnI and BalI ends); H3 5Ј-TG GCC ACT ATG ATG ACT TGT GGC TAC G and 5Ј-CTC GAG CCG CAA GTC ATC CAG GCA GCC A (BalI and XhoI ends); H4, 5Ј-CTC GAG TTT GAT GTT GCT GTG GAT GAG C and 5Ј-GAT ATC GCT GTC ATC GCA CGT GCT CTC C (XhoI and EcoRV ends); and H5, 5Ј-GAT ATC GTG AAG GAG GTG TGC ACT G and 5Ј-GAA TTC TCT CTG CCC AGC CTC CCG GAT (EcoRV and EcoRI ends). GK-EGFP fusion genes were made by ligating a PstI-EcoRI fragment of GK (containing cassettes G1-G5) together with the 5Ј sequences of hepatic GK (as a HindIII-PstI fragment) and with EGFP (as a EcoRI-NotI fragment from pEGFP-N2 (CLONTECH)) into the HindIII and NotI sites of pcDNA3 (Invitrogen). The GK-cHKII-EGFP chimeras were made by replacing the appropriate GK cassette in the GK-EGFP parent vector with the corresponding HKII cassette. Because the generation of new restriction sites resulted in amino acid substitutions in some of the HKII cassettes, the cHKII-EGFP construct was made by direct PCR amplification of HKII using the H1-forward and H5-reverse primers.
For the yeast two-hybrid analysis the GKRP cDNA fragment, including entire coding region, was amplified by linker-primer PCR from pSG5-GKRP thereby introducing a NdeI site at the 5Ј end and a BamHI site at the 3Ј end. The NdeI-BamHI fragment was ligated into pAS2-1 (CLONTECH) to generate the GAL4-binding domain (BD)-GKRP fusion gene. To express the GAL4 activation domain (AD) fusion protein, each PstI-EcoRI fragment from the GK, cHKII or GK-cHKII chimeras was ligated into the NcoI and EcoRI sites of pACT2 (CLONTECH) with the NcoI-PstI fragment of GK obtained from the fragment amplified by PCR using the primer sets G1.
The function of the two putative NESs of GK was tested using a fusion protein containing a segment of GK (amino acids 299 -359) linked to EGFP. Linker-primer PCR was used to introduce EcoRI and BamHI sites on the 5Ј and 3Ј ends, respectively, of a GK cDNA fragment (bases 962-1144). The resulting EcoRI-BamHI fragment was ligated via an XhoI-EcoRI adapter into XhoI and BamHI sites of pEGFP-N2. The adapter included the Kozak consensus sequence and start codon for translation initiation (CGCCACCATG). Mutagenesis of the putative NESs was performed using a QuikChange Site-directed Mutagenesis kit (Stratagene).
Cell Culture, DNA Transfections, and Confocal Microscopy-HeLa cells were grown at 37°C with 5% CO 2 atmosphere in Dulbecco's modified Eagle's medium containing 10% fetal calf serum and 25 mM glucose. Cells were seeded in 35-mm glass bottom microwell dishes (Mat-Tek) at a density of 1.5-2 ϫ 10 5 cells/dish 1 day before transfection. Transfections were performed using SuperFect reagent (Qiagen) according to the supplier's instructions. After transfection, cells were cultured for 14 -20 h, and then expressed proteins were localized either by direct fluorescence or by immunocytochemistry. Indirect immunofluorescent staining for GK and GKRP was performed as described previously (27) using sheep anti-GST-GK antibody (28) and rabbit anti-rat GKRP antibody. 2 Cy3-conjugated donkey anti-sheep IgG and Cy5-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.) were used as secondary antibodies. Actin was revealed by staining with monoclonal anti-actin antibody, C4 (Roche Molecular Biochemicals) after fixation with cold methanol.
Images were acquired using a Zeiss 410 Confocal Laser Scanning Microscope (Carl Zeiss). Nuclear/cytoplasmic ratios were determined by digital image analysis using NIH Image 1.61 software. Briefly, a 79 pixel area, which was approximately the area of an average nucleolus, was used to measure mean pixel intensities in the cytoplasm and in the nucleoplasm for each cells studied. From these values, the ratio of the nuclear intensity to the cytoplsmic intensity (N/C ratio) was calculated.

RESULTS
GKRP Alters the Subcellular Location of GK-Because hepatocytes express both GK and GKRP we established a heterologous cell system in which the nuclear translocation of GK could be studied without interference by endogenously expressed proteins. HeLa cells, which do not express either GK or GKRP (data not shown) and which have served as a model for nuclear translocation studies of other proteins, were used. Hepatic GK and GKRP were each transiently expressed, and their subcellular locations were each determined by immunocytochemistry. When expressed by itself, GK was found almost completely in cytoplasm (Fig. 1A). In contrast, when GKRP was expressed by itself, it was detected both in the nucleus and cytoplasm (Fig.  1B). However, when GK and GKRP were co-expressed they both were found to be located predominately in the nucleus ( Fig. 1, C and D). Thus, in HeLa cells GKRP causes GK to change from being exclusively cytoplasmic in location to being predominately nuclear in location. At the same time, the presence of GK caused GKRP to become more concentrated in the nucleus.
Analysis of the Nuclear Translocation of GK-cHKII-EGFP Chimeras in HeLa Cells-To determine whether GK and GKRP need to interact for GK to be translocated to the nucleus and to determine what regions of GK are necessary, we made and tested a series of fusion proteins between GK and the C-terminal half of hexokinase II (cHKII). HKII has a molecular mass of ϳ100 kDa and consists of two related halves, each of which is highly similar to GK (31, 32) and both of which are catalytically active (33). Chimeric proteins between GK and cHKII have been reported to be catalytically active, thereby indicating that portions of GK can be replaced with corresponding regions of cHKII without causing dramatic changes in the overall structure of this enzyme (34). Thus in this study, GK, except for 25 amino acids from N-terminal end, was divided into five cassettes (amino acids 26 -117 (G1), 118 -207 (G2), 208 -271(G3), 272-365 (G4), and 366 -464 (G5)). Each cassette was replaced with its counterpart in cHKII, thereby generating five different GK-cHKII chimeras (C1, C2, C3, C4, and C5) as shown in Fig.  2. In this experiment, all of the GK-cHKII chimeras, as well as both wild type hepatic GK and cHKII, were linked with enhanced green fluorescent protein (EGFP). This allowed us to directly observe the subcellular location of these proteins in cells, both with and without GKRP. GK-EGFP was located predominately in the cytoplasm when expressed alone but became almost entirely nuclear in location when expressed with GKRP (Fig. 3). Thus, addition of EGFP to the C-terminal end of GK did not adversely affect the GKRP-dependent nuclear translocation of GK. When expressed by itself cHKII-EGFP was also localized mainly in the cytoplasm. However, cHKII-EGFP did not translocate into the nucleus in the presence of GKRP (Fig. 3). The C1-C4 GK-cHKII-EGFP chimeras were found to be totally cytoplasmic in location when expressed either with or without GKRP. However, the C5 GK-cHKII-EGFP chimera retained the ability to translocate into the nucleus in the presence of GKRP (Fig. 3). These results suggest that nearly 70% of GK may be necessary for GKRP-dependent nuclear translocation and retention (Fig. 2).
Analysis of the Binding of GK-cHKII Chimeras to GKRP in a Yeast Two-hybrid Assay-To further assess whether the GKRPdependent nuclear translocation of GK depends upon the binding of GK with GKRP, we established a yeast two-hybrid assay in which interactions of the two proteins could be quantitatively assessed. Both GK and GKRP were co-expressed in yeast as fusion proteins with GAL4 AD or BD, respectively, and their interactions were assessed by measuring ␤-galactosidase activity. Because the interaction between GK and GKRP in yeast two-hybrid system was inhibited by glucose in medium at a concentration required for growth of yeast (45-100 mM), 3 YPGE medium that contains glycerol and ethanol instead of glucose as carbon sources was used. The interaction with GKRP was detected only in the C5 GK-cHKII chimera (Fig. 4A) among 5 GK-cHKII chimeras (Fig. 2). Thus, the only GK mutant that was able to bind to GKRP in yeast was also the only mutant that was able to cause the translocation of GK to the nucleus in HeLa cells (Fig. 3). In all cases, the proper expression of fusion proteins in yeast was verified by Western blot analysis. The lack of detectable interactions between the C1-C4 GK-cHKII chimeras and GKRP was not due to deficiency of protein expression (Fig. 4B). Thus, these results are interpreted as indicating that the nuclear translocation of GK depends upon binding to GKRP.
To determine whether these putative NES sequences were actually functional, we made a series of EGFP fusion proteins containing amino acids 299 -359 of GK because both NES were present within this portion of the protein. Alanine substitution mutations were introduced at positions 306, 307, and 309 (underlined) within the 300 ELVRLVLLKLV 310 region and at positions 355 and 357 of the 347 QIHNILSTLGLR 358 region. The substitution of alanines for leucines within functional NES sequences has been shown previously to eliminate their function (36).
This series of GK (299 -359)-EGFP fusion proteins, as well as EGFP, were expressed in HeLa cells and the subcellular location of these proteins was assessed by microscopic digital image analysis. EGFP without any GK sequence was distributed nearly equally between the cytoplasm and nucleus ( Fig. 6A; N/C ratio ϭ 1.09 Ϯ 0.024), presumably because of its small size (27 kDa), which is well below the exclusion size of the nuclear pore complex. In contrast, GK 299 -359 -EGFP had a N/C ratio of 0.95 Ϯ 0.019, indicating that this fragment of GK was acting either to diminish the entry of the protein into the nucleus or increasing its extrusion from the nucleus. Although the ability of the GK 299 -359 fragment to extrude EGFP from the nucleus may appear to be much less than that of the entire sequence of GK (Fig. 6, B and C, and Table I), the two cannot be directly compared because of the difference in the sizes of the fusion proteins (34 versus 77 kDa) which would affect the ability of these different proteins diffuse into the nucleus. Indeed, the higher N/C ratio of GK 299 -359 -EGFP compared with GK 1-456 -EGFP is likely to be due simply to greater entry of GK 299 -359 -EGFP into the nucleus. Nonetheless, given that the differences in the N/C ratios determined in this manner were not very large, we proceeded to mutate both of the putative NES sequences within GK 299 -359 -EGFP. Mutation of both NESs (GK 299 -359 -EGFP(mt1)) caused a nearly equal distribution of the protein between the nucleus and cytoplasm (Fig. 6D). In contrast, mutation of only the 347 QIHNILSTLGLR 358 sequence (GK 299 -359 -EGFP(mt2)) resulted in a protein that, again, had a greater cytoplasmic localization (Fig. 6E). Mutation of the 300 ELVRLVLLKLV 310 sequence in GK 299 -359 -EGFP(mt3) abolished this effect ( Fig. 6F and Table I). Thus, these results indicate that the amino acids 300 ELVRLVLLKLV 310 within GK do, indeed, constitute a functional NES sequence.
GK Does Not Enter the Nucleus in the Absence of GKRP-Inspection of the amino acid sequence of GK did not reveal motifs that met the consensus for any known NLS sequences. Indeed, the fact that GK is found only in the cytoplasm in the absence of GKRP strongly suggests the lack of a functional NLS within this enzyme. However, we considered it possible that GK might enter the nucleus in the absence of GKRP (via either diffusion or some other active mechanism) but then be extruded immediately by the NES in GK, thereby making it appear to always reside in the cytoplasm. Thus, to rule out the possibility of entry of GK into the nucleus in the absence of GKRP, we tested the effect of leptomycin B (LMB), an inhibitor of leucine-rich NES-dependent nuclear export (39,40). HeLa cells expressing GK were treated with LMB for 8 h, and then both GK and actin were localized by immunostaining. Actin, which has two functional NES sequences, served as positive control for the effect of LMB because actin enters the nucleus by diffusion and accumulates there when NES-mediated nuclear export is inhibited using LMB (38). After LMB treatment GK remained in the cytoplasm, whereas actin was detected in the nucleus (Fig. 7). Thus, this experiment provides direct evidence against a GKRP-independent mechanism for entry of GK into the nucleus. DISCUSSION The entry and exit of proteins into and out of the nucleus is a highly regulated process. The entry of proteins of a size greater than the exclusion limit of the nuclear pore complex depends on the presence of a NLS sequence within the protein.
In the absence of an NLS sequence, proteins that are too large to enter the nucleus by themselves may do so via interactions with other proteins, e.g. a piggy-back mechanism. Similarly, the export of proteins from the nucleus is an active process that depends upon a different motif called a NES sequence. Together, the presence or absence of either NLS or NES sequences, as well as the functional state of these motifs (because they can be masked by conformational changes or binding of other proteins) causes dynamic alterations in the subcellular distribution of many proteins. Because both NLS and NES sequences, as well as the proteins involved in transporting proteins to and from the nucleus, are increasingly well understood, we made use of this knowledge to study the mechanisms that are involved in the translocation of GK in the liver.
Essential Role for GKRP in the Nuclear Translocation and Retention of GK-In a HeLa cell model system, the co-expression of GKRP with GK was found to be required for GK to be able to enter the nucleus. In the absence of GKRP, GK resides only within the cytoplasm. However, in the presence of this protein GK is able to enter and be retained in the nucleus. Analysis of a series of GK-cHKII chimeric proteins showed that only those chimeras that are able to bind to GKRP are transported into the nucleus. These results clearly indicate that binding to GKRP is essential for both the nuclear translocation and accumulation of GK. Previous studies have shown that GKRP functions to anchor GK in the nucleus (16). However, these studies suggest that GKRP has the additional function of a chaperone that mediates the nuclear entry of GK.
HeLa cells were used in this study because they lack both GK and GKRP and have been used widely for studies of nuclear translocation mechanisms. However, HeLa cells do not fully mimic the function of hepatocytes. For instance, the intracellular location of GK in HeLa cells was not modulated by changes in the glucose concentration or addition of other metabolic factors (data not shown). This may explain why HeLa grown in a medium containing 25 mM glucose showed subcellular localization of GK and GKRP that was similar to that previously observed for hepatocytes cultured under basal glucose level (ϳ5.5 mM). The reason for the differences between HeLa cells and hepatocytes is not known but could be due to many factors such as the lack of the type 2 glucose transporter, as well as other proteins that determine rate of glycolytic flux.
Previous studies, which have detected GKRP largely in the nucleus, have led to the suggestion that GK is able to enter and exit the nucleus by itself (16). However, our results are not consistent with such a model. First, although GK has several clusters of basic amino acids, none of them matches the consensus sequences for either an SV40 or bipartite NLS. Second, GK resides exclusively in the cytoplasm of HeLa cells in the absence of GKRP. Third, GK does not accumulate in the nucleus even when nuclear export is inhibited with LMB. Thus, these findings do not support entry of GK into the nucleus in the absence of GKRP, either via an active or passive mechanism. Instead, our observation that GKRP, in the absence of GK, is found in both the nucleus and cytoplasm but changes to being largely nuclear in the presence of GK suggests that GKRP is an active participant in mediating the nuclear entry of GK. Further evidence that GKRP actively participates in the entry of GK into the nucleus was obtained by our studies of a set of GK-cHKII chimeras. The ability of these proteins to be transported and retained in the nucleus of HeLa cells was found to be dependent upon their ability to bind to GKRP, as determined by a yeast two-hybrid assay.
Thus, based on the data we have obtained we suggest a new model for the nuclear-cytoplasmic movement of GK (Fig. 8). In this model, GKRP functions as a nuclear chaperone that carries GK into the nucleus via a piggy-back mechanism. Piggy-back mechanisms have been shown to be involved in the nuclear import of several other proteins, such as the 46-kDa subunit of DNA polymerase ␣-primase (41) and the interleukin-5 receptor (42).
Although we have not demonstrated that GKRP is able to exit the nucleus after releasing GK, we think this is probably the case. Previous studies have suggested that the turnover rate of GKRP is slower than that of GK (12). Thus, unless GKRP is itself able to transit out of the nucleus, there might not be enough of the protein in some situations to be able to chaperone all molecules of GK back into the nucleus. Previously, GKRP has been thought to be located constitutively in the nucleus based on observations in primary hepatocytes (16). However, within hepatocytes, which also express GK, it is possible that GKRP is present only transiently within the cytoplasm, where it is more difficult to detect than in the nucleus.
It is also possible that specific metabolic traits of hepatocytes may affect the subcellular location of GKRP. For instance, the binding of GKRP to GK is promoted by fructose 6-phosphate and inhibited by fructose 1-phosphate (13,43). Because these hexose phosphates interact with GKRP not GK, two different conformations of GKRP have been proposed: one form binds fructose 6-phosphate and GK, and the other binds fructose 1-phosphate (44,45). Thus, different conformational variations   might also be able to affect the subcellular localization of GKRP, thereby providing a possible explanation for why GKRP is readily detected in the cytoplasm of HeLa cells but not hepatocytes.
Nuclear Export of GK-Leucine-rich NES sequences were first identified in the HIV-1 Rev protein (35) and in the cellular protein kinase A inhibitor (36). In these studies, the ability of this sequence motif to affect the subcellular location of a protein was demonstrated by nuclear microinjection of NES-conjugated bovine serum albumin or IgG. In this study we identified a NES in GK by transfection of EGFP fusion proteins. Although this method is technically simpler than the microinjection of larger fusion proteins into the nucleus, it has the disadvantage of depending upon diffusion of protein into the nucleus. Nonetheless, this approach enabled us to test the function of the two putative NES sequences in GK by measuring the N/C ratio of a series of EGRP-tagged fusion proteins. The results obtained demonstrate that one of two putative NES sequences within GK is functionally active. The presence of a functional NES in GK helps to explain why, in the absence of GKRP, GK is always detected in the cytoplasm. Also, the presence of a NES in GK may explain how the protein is quickly extruded from the nucleus after its release from GKRP.
Based on the crystal structure of the structurally related yeast hexokinase B, GK has been predicted to fold into a small domain and large domain, which are connected by hinge region with a deep cleft (46,47). The 300 ELVRLVLLKLV 310 region in GK is predicted to lie within the large domain far from the cleft where the glucose-binding site resides. Thus, given the probable location of the NES in GK, glucose binding probably does not obstruct NES-dependent nuclear export. The binding sites for GKRP have previously been suggested to be located in the small domain and the hinge region of GK (48,49), some of which overlap with glucose-binding site. However, when a portion of the large domain of GK (amino acids 272-365) that contains the NES sequence was replaced with cHKII sequences, the resulting chimeric protein did not show any affinity to GKRP in yeast two-hybrid assay. Thus, our studies suggest that there may be a large interface between GK and GKRP because only a small portion of the protein was found to be dispensable when tested in this manner. However, additional studies are necessary to be certain of this.
In the model shown in Fig. 8, when GK is in the nucleus, it binds to and is inhibited by GKRP. We speculate that GKRP may mask the NES in GK to ensure the GK-GKRP complex remains in the nucleus and is not rapidly exported. However, when blood glucose levels rise after feeding, GK is released from GKRP and is exported from the nucleus via an NES-dependent manner. Within the cytoplasm GK is catalytically active and converts glucose to glucose-6-phosphate. However, as the blood glucose levels begin to fall, GKRP binds cytoplasmic GK and moves it back into the nucleus, thereby completing a translocation cycle. Although these studies provide insights into the mechanisms that control GK translocation to and from the nucleus, they do not actually reveal the purpose of this translocation cycle. Indeed, the functional importance of the nuclear sequestration of GK will require the generation and characterization of GKRP knock-out mice.