Allosteric Activation of Human Glucokinase by Free Polyubiquitin Chains and Its Ubiquitin-dependent Cotranslational Proteasomal Degradation*

Human glucokinase (hGK) is a monomeric enzyme highly regulated in pancreatic β-cells (isoform 1) and hepatocytes (isoforms 2 and 3). Although certain cellular proteins are known to either stimulate or inhibit its activity, little is known about post-translational modifications of this enzyme and their possible regulatory functions. In this study, we have identified isoforms 1 and 2 of hGK as novel substrates for the ubiquitin-conjugating enzyme system of the rabbit reticulocyte lysate. Both isoforms were polyubiquitinated on at least two lysine residues, and mutation analysis indicated that multiple lysine residues functioned as redundant acceptor sites. Deletion of its C-terminal α-helix, as part of a ubiquitin-interacting motif, affected the polyubiquitination at one of the sites and resulted in a completely inactive enzyme. Evidence is presented that poly/multiubiquitination of hGK in vitro serves as a signal for proteasomal degradation of the newly synthesized protein. Moreover, the recombinant hGK was found to interact with and to be allosterically activated up to ∼1.4-fold by purified free pentaubiquitin chains at ∼100 nm (with an apparent EC50 of 93 nm), and possibly also by unidentified polyubiquitinated proteins assigned to their equilibrium binding to the ubiquitin-interacting motif site. The affinity of pentaubiquitin binding to hGK is regulated by the ligand (d-glucose)-dependent conformational state of the site. Both ubiquitination of hGK and its activation by polyubiquitin chains potentially represent physiological regulatory mechanisms for glucokinase-dependent insulin secretion in pancreatic β-cells.

The glucose-phosphorylating enzyme glucokinase (GK) 2 (ATP:D-hexose 6-phosphotransferase, EC 2.7.1.2) plays a piv-otal role in the regulation of glycolytic flux in hepatocytes and pancreatic ␤-cells at physiological millimolar concentrations of glucose (Glc). This 50-kDa size monomeric hexokinase catalyzes the phosphorylation of D-glucose to form glucose 6-phosphate with MgATP 2Ϫ as the phosphoryl donor and is characterized by a low affinity for Glc ([S] 0.5 ϳ 8.0 mM), a positive kinetic cooperativity of Glc binding (n H ϳ 1.7), and no feedback inhibition by its product.
Besides its expression in hepatocytes (isoforms 2 and 3) as a cytoplasmic and nuclear enzyme, where its translocation and activity is regulated by the glucokinase regulatory protein and the metabolic state of the cell (1), human GK is also expressed in pancreatic ␤-cells (isoform 1, the neuroendocrine isoform) mainly as a soluble cytoplasmic enzyme. In pancreatic ␤-cells GK has been found to be partitioned between the cytoplasm and the insulin secretory granules as a peripheral membrane protein (2)(3)(4) and in a regulated manner, i.e. mediated by Glc/ insulin. In pancreatic ␤-cells, GK acts as the glucose sensor (5), a concept supported by the finding that complete GK deficiency leads to neonatal diabetes (6,7). More than 200 different mutations have been identified in the hGK gene (8), and most of them lead to reduced enzyme activity and are associated with mild diabetes, maturity-onset diabetes of the young type 2 (MODY2). Others are characterized by an in vitro thermal instability when expressed as recombinant glutathione S-transferase (GST) fusion proteins (9). A few activating mutations have also been identified, leading to a hypoglycemic hyperinsulinism of infancy (8).
In both hepatocytes and pancreatic ␤-cells GK is activated by the bifunctional enzyme 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (10). Little is known about covalent posttranslational modifications of GK and their possible regulatory functions, and the molecular and cellular mechanisms involved in its degradation/turnover are also poorly understood. In this study, we have identified isoforms 1 and 2 of hGK as novel substrates for the ubiquitin (Ub)-conjugating enzyme system of the rabbit reticulocyte lysate. The functional implications of this posttranslational modification have been studied in vitro with reference to the major roles ubiquitination plays in regulating a broad array of basic cellular processes (11) and in par-ticular in relation to the described regulatory function on the Glc-induced insulin synthesis and secretion in pancreatic ␤-cells (12)(13)(14)(15).
Plasmid Constructs, Expression, and Purification of Recombinant Forms of hGK-Plasmid constructs, protein expression, and purification are described in supplemental methods.
In Vitro Expression of hGK-hGK in the pcDNA3.1ϩ or pcDNA3.1/HisC vector was expressed in a coupled in vitro transcription/translation system (TNT T7 Quick-coupled Transcription/Translation System; Promega) in the presence of [ 35 S]Met. The reaction contained in a final volume of 50 l: 1 l of [ 35 S]Met (10 Ci), 2 g of plasmid DNA, 20 mM dithiothreitol, and 10 M Ub (in addition to the endogenous content of Ub) and 40 l of rabbit reticulocyte lysate. The standard incubation time was 90 min at 30°C, and the reaction was quenched by adding 1 mM cold Met. His 6 -tagged hGK was alternatively isolated by affinity purification using the MagZ Protein Purification System (Promega).
In Vitro Ubiquitination of hGK-In vitro ubiquitination of His 6 -hGK, hGK, and its mutant forms was performed in a reconstituted rabbit reticulocyte lysate (RRL) system from Promega at 30°C for 90 min in a final volume of 50 l containing 15 l of purified [ 35 S]Met-labeled His 6 -hGK or hGK protein, 12.5 M Ub (in addition to the endogenous content of the lysate), 2 M Ub aldehyde, and 0 -50% (v/v) RRL. Subsequently, the high molecular mass reaction products were separated on SDS-PAGE (10% gel) and analyzed by autoradiography and densitometry. The generation of high molecular mass forms of unlabeled purified GST-hGK was alternatively studied in the same reconstituted RRL system after cleavage (by factor X a ) of the GST fusion partner. The reaction contained in a final volume of 50 l: 1.7 g of hGK protein, 50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 2 mM dithiothreitol, 2 mM ATP, 2 M Ub aldehyde, 12.5 M Ub, and 0 -50% (v/v) RRL. The reaction products were separated on SDS-PAGE (10% gel) and subjected to immunoblot analysis.
Immunoblot Analysis-High molecular mass forms of unlabeled recombinant hGK proteins (isoforms 1 and 2) generated in the in vitro ubiquitination assay were separated by SDS-PAGE (10% gel) and immunoblotted by standard procedures. The primary Ab (anti-GK) and the secondary Ab (peroxidaseconjugated anti-rabbit IgG) were diluted 1:2000. Ub-conjugated proteins were detected with an anti-Ub primary Ab that recognizes both unconjugated and conjugated mono-and polyubiquitin and peroxidase-conjugated anti-mouse IgG as the secondary Ab, both diluted 1:1500. The ladder of reference proteins was MagicMark TM XP Western Protein Standard (Invitrogen) with nine recombinant proteins, range 20 -220 kDa. The enhanced chemiluminescence detection method was employed to develop the immunoblots.
INS-1 and HepG2 cells were lysed, GK immunoisolated from the cytosolic fractions by Protein A-Sepharose, electrophoresed, and immunoblotted with anti-ubiquitin or anti-GK Ab. For details, see supplemental methods.
Assay of hGK Activity-In the standard assay the catalytic activity of purified recombinant hGK was measured in a reaction volume of 1 ml containing: 25 mM sodium Hepes, pH 7.4, 25 mM KCl, 2.5 mM MgCl 2 , 1 mM dithiothreitol, 0.1% (w/v) bovine serum albumin, 2.5 mM ATP, 1 mM NAD ϩ , 0.35 units of glucose-6-phosphatase, 0.5 g of recombinant hGK, and 50 mM D-glucose. The time course of the NADH formation was followed at 340 nm in a thermostated cuvette (37°C) of the Hewlett Packard photodiode array spectrophotometer (Agilent 8453), and the activity was calculated from the linear slope.
Intrinsic Tryptophan Fluorescence-Intrinsic tryptophan fluorescence was performed on a PerkinElmer LS-50B instrument at 25°C in a buffer containing 20 mM Hepes, 100 mM NaCl, and 1 mM dithiothreitol, pH 7.0, and a protein concentration of 0.03 mg/ml. The excitation and emission wavelengths were 295 and 340 nm, respectively, with slit widths of 4 and 7 nm.
Equilibrium Binding of Polyubiquitin Chains to hGK and hGK⌬C24-The chromatographic holdup assay was performed essentially as described by Charbonnier et al. (16). For details, see supplemental methods.

RESULTS
Structural Analysis-An analysis of the three-dimensional structure of hGK revealed the sequence EEGSGRGAALVSAVA at positions 442-456 at the C-terminal ␣-helix (Fig. 1A). This sequence is homologous to the core ubiquitin-interacting motif (UIM) sequence eeeXXXAXXXSXXe, where e is a negatively charged residue, is most often the hydrophobic residues Leu, Ile, or Ala, and X is any amino acid (17,18). Interestingly, the helix length, and thus the conformation of this site, changes upon binding of Glc and an allosteric activator (Fig. 1B), and its orientation relative to the specifically interacting helix 6 (residues 204 -217) changes as well (Fig. 1C). 3 In general, UIM sites promote ubiquitination as well as binding of polyubiquitin chains (17,18), and both possibilities were addressed in the present study.
In Vitro Synthesis and Posttranslational Ubiquitination of hGK-As seen from Fig. 2, lane 1, [ 35 S]methionine (Met)-labeled His 6 -hGK synthesized in the coupled in vitro transcription-translation system of the RRL is a substrate for the Ubconjugating enzyme system of the lysate. Several high molecular mass bands were observed in addition to the fulllength His 6 -hGK (56 kDa). Some minor low molecular mass bands were regularly observed, presumably representing incomplete chains (19) because their presence was not affected by protease or proteasome inhibitors. An identical electrophoretic pattern was observed with a construct of WT-hGK (Fig. 2, lane 2), but of slightly lower molecular mass. The apparent molecular masses of the individual radioactive bands were 56, 67.5, 74, and 79 kDa for His 6 -hGK and 53, 64.5, 71, and 76 kDa for WT-hGK, respectively, and the high molecular mass bands amounted to ϳ17% of the total. The [ 35 S]Met-labeled His 6 -hGK was purified by affinity binding to and gradient elution from Ni 2ϩ -chelated magnetic beads. The 56-kDa His 6 -hGK was preferentially eluted at 1 M imidazole (Fig. 2, lane 3). Using the purified His 6 -hGK as the substrate in the in vitro RRL system the level of Ub conjugates was found to increase in proportion to the concentration of the RRL lysate used in the incubation (supplemental Fig. S1A, lanes 1-6, and S1B).
To further characterize the Ub conjugates of hGK, purified recombinant hGK (isoforms 1 and 2) was used as substrate for the Ub-conjugating enzyme system of the RRL. When analyzed by two-dimensional electrophoresis in a high-resolution polyacrylamide gradient gel, immunodetection with anti-GK revealed that the high molecular mass forms were positioned in a diagonal pattern at a progressively higher pI than the unmodified hGK, as expected for a polyubiquitinated protein (Fig. 3, A  and B). Immunodetection with anti-ubiquitin Ab revealed a similar pattern (Fig. 3A, inset). The conjugates in these experiments represented ϳ50% of the total immunoreactive protein in the gels for both isoforms. Interestingly, the molecular mass position of hGK⅐Ub 1 and hGK⅐Ub 2 revealed a double spot. The respective spots demonstrated the same pI but different mobilities as SDS denatured proteins, presumably due to different conformations, thus indicating two putative ubiquitin acceptor sites in hGK.
To possibly identify the target residue(s) for ubiquitination, the 22 lysine residues common to the isoforms were individually mutated to arginine (Arg) (supplemental Table  S1) and the mutant forms were expressed in vitro as His 6tagged and [ 35 S]Met-labeled proteins in the RRL system. For all mutant forms, identical electrophoretic patterns of Ub conjugates were observed, as shown for WT-hGK in Fig. 2. The lysine residues are distributed rather uniformly along the protein, with some residues closely spaced in the linear sequence of flexible loop structures or closely spaced in the three-dimensional structure (20). Hence, multiple lysine residues of hGK presumably function as redundant acceptor sites. When the hGK⌬C24-truncated form, devoid of the putative UIM site (Fig. 1), was expressed in the RRL system the double bands observed for His 6 -hGK⅐Ub 1 and His 6 -hGK⅐Ub 2 of the WT-hGK were replaced by single bands (Fig.  4, A and B). This finding indicates that the ubiquitination of one of the target sites is coupled to the UIM site. Interestingly, the C-terminal-truncated form revealed a total loss of catalytic activity (data not shown).
Proteasomal Degradation of Newly Synthesized and Ubiquitinated hGK-Proteasome-dependent degradation of hGK was studied co-translationally by expressing [ 35 S]Met-labeled His 6 -hGK for only 30 min in the standard RRL system. Then its stability was followed in a proteasome-and ATP-enriched RRL system, in the absence and presence of the proteasome inhibi- 3 The helix nomenclature. In this study we used the MolMovDB (available at molmovdb.mbb.yale.edu/molmovdb/) to demonstrate the regions of variable secondary structure in the two determined crystal structures of hGK (20). 17  tor MG-132. As seen from Fig. 5, the total intensity of the hGK signal (radioactivity), including His 6 -hGK and its ubiquitinated forms, increased ϳ1.8-fold (n ϭ 5 and p ϭ 0.008) in the presence of MG-132. Thus, inhibition of the proteasomal activity resulted in a stabilization of the newly synthesized enzyme. The degree of inhibition observed may to some extent have been   affected by the relatively high concentration of nonspecific Ub conjugates in the RRL representing competitive substrates in terms of proteasomal degradation (21).

Equilibrium Binding of Polyubiquitin Chains and Catalytic
Activation of WT-hGK-To further characterize the C-terminal UIM site (Fig. 1), we studied the equilibrium binding of polyubiquitin chains (mainly Ub 5 -Lys-48 linked) to GST-WT-hGK or GST-hGK⌬C24 immobilized on glutathione-Sepharose 4B using a chromatographic holdup assay. The amount of bound and free analyte was measured by SDS-PAGE (Fig. 4C), which revealed that polyubiquitin chains bind to GST-WT-hGK in the absence of Glc (Fig. 4D, columns 1 and 2), but not to GST-hGK⌬C24 lacking the UIM site (Fig. 4D, columns 3 and 4). Moreover, in the presence of 50 mM Glc, the binding of the analyte to GST-WT-hGK was markedly reduced (columns 5 and 6) as expected from the lower accessibility of the key interacting residues of the UIM site in the ligand-bound conformation of WT-hGK (Fig. 1D).
As seen from Fig. 6A and 6B, column 2, free polyubiquitin chains also stimulated the catalytic activity of recombinant hGK in a concentration-dependent manner and in the low nanomolar concentration range. An ϳ1.4-fold stimulation (n ϭ 4) was observed at 97 nM Ub 5 -Lys-48, and the [S] 0.5 value for Glc was slightly reduced, i.e. from 8.62 Ϯ 0.22 to 8.18 Ϯ 0.27 mM (data not shown). A similar small reduction was observed in the K d value for the equilibrium binding of Glc at 25°C, i.e. from 4.8 Ϯ 0.06 to 4.3 Ϯ 0.15 mM, as determined by intrinsic trypto-  1, 2, 5, and 6) and GST-hGK⌬C24 (lanes 3 and 4) as described for the chromatographic holdup assay in the absence (lanes 1, 2, 3, and 4) and presence (lanes 5 and 6) of 50 mM D-glucose, free polyubiquitin (mainly Ub 5 Lys-48-linked) in lanes 2, 4, and 6 and bound plus free Ub 5 in lanes 1, 3, and 5 as analyzed by SDS-PAGE (10%) and immunoblot detection by the anti-Ub Ab. D, densitometric quantitation of the respective Ub 5 band intensities in Fig. 4C for GST-WT-hGK in the absence (columns 1 and 2) and presence of 50 mM Glc (columns 5 and 6), and GST-hGK⌬C24 (columns 3 and 4). Each column represents the average of three experiments, and the error bars indicate the resulting S.D. Statistical significance was determined using the Student's t-test; **, p Ͻ 0.01.  phan fluorescence (data not shown). It is also seen that monoubiquitin was much less efficient in stimulating the activity. In addition, it is seen from Fig. 6B that the reticulocyte lysate, containing a plethora of Ub conjugates (22) and free polyubiquitin chains (supplemental Fig. S2A), has a similar (ϳ1.4-fold) stimulatory effect.

DISCUSSION
In pancreatic ␤-cells the ubiquitin-proteasome pathway has been found to have a regulatory role in the Glc-stimulated insulin release (15), Glc-stimulated (pro)insulin synthesis (13), and in the biogenesis and surface expression of the ATP-sensitive K ϩ (K ATP ) channels (12) as well as in maintaining a normal function of the voltage-dependent calcium channel (14). In the present study, we demonstrate that hGK of the pancreatic ␤-cell (isoform 1) and the hepatocytes (isoform 2) are posttranslationally modified by multiple moieties of Ub. Two-dimensional electrophoresis (Fig. 3, A  and B) demonstrate the formation of hGK⅐Ub 1-3 on at least two putative target lysine residues where the double spot of SDS-denatured hGK⅐Ub 1 and hGK⅐Ub 2 in the second dimension represent two ubiquitinated species with presumably different conformations having the same pI. A similar heterogeneity (double bands) has been reported for in vitro ubiquitinated luciferase (23) as well as for Lys-48-Ub 2 and Lys-48-Ub 3 synthesized in vitro (24). However, on site-directed mutagenesis, in which 22 of the lysine residues in hGK were individually mutated to Arg and expressed in vitro, no specific acceptor site could be identified. Multiple lysine residues in hGK, therefore, seem to function as redundant acceptor sites as previously exemplified by cyclin B (25) and cyclin A (26). It should be noted that the mutant forms tested were catalytically active with a specific activity in the range of 65-99% of WT-hGK (data not shown). Deletion of the 24 C-terminal amino acids, with the UIM site, resulted in polyubiquitination at an apparently single site (Fig. 4A), indicating that the ubiquitination of one site is promoted by the UIM site as previously shown for UIM-containing proteins (27).
Polyubiquitination often serves as a signal for targeting cytoplasmic and nuclear proteins to the proteasome for subsequent degradation (reviewed in Ref. 11). We found that the 30-min in vitro translated [ 35 S]Met-labeled His 6 -hGK and its ubiquitinated forms were unstable in a proteasome-and ATP-enriched reticulocyte lysate degradation assay that was sensitive to the proteasome inhibitor MG-132 (Fig. 5). It has recently been proposed (28) that the key targeting step for proteasome-mediated degradation is the conjugation of multiple short ubiquitin chains, independent of linkage type. This finding may explain the proteasomal degradation of newly synthesized hGK in our cell-free system, suggesting that its degradation may function as a regulatory mechanism in the homeostatic control of the cellular GK protein expression as demonstrated for several long-lived proteins (29). In hepatocytes (rat) expressing most of the total GK the sequestration and degradation of cytoplasmic GK occurs by the autophagosomal-lysosomal pathway at a rate of 3.5%/h and a half-life of 12.7 h (30). So far, we have no experimental data to support that ubiquitinated hGK is preferentially targeted to the autophagosomal-lysosomal pathway. An active autophagocytosis has been demonstrated in mouse ␤-cells (31,32) and in NIT insulinoma cells (33), including autophagy of insulin secretory granules, and may in this way be involved in the turnover of the membrane-bound form of GK. In certain diabetes-associated mutations in hGK (MODY 2) a reduced stability of the enzyme has been considered as a possible mechanistic explanation for the hyperglycemia and reduced Glc-stimulated insulin secretion (9). Because ubiquitination of misfolded proteins associated with cytoplasmic chaperones may be degraded predominantly through the ubiquitinproteasome system (34), further studies are in progress to investigate this possibility for selected loss-of-function mutant forms of hGK.
Ubiquitin-binding domains are found in proteins that function in a vast range of cellular events, including the activation of kinases in the nuclear factor-B signaling pathway (35). Here we demonstrate that free poly-Ub chains (Ub 5 -Lys-48-linked) allosterically activate hGK at low nanomolar concentrations (Fig. 6, A and B), i.e. at ten times lower concentrations than for monoubiquitin. From Fig. 6A it is seen that the catalytic activity of hGK is increasingly stimulated by Ub 5 in the concentration range of 20 -100 nM, with an apparent EC 50 value of 93 nM (Fig. 6A). It is shown that the polyubiquitin chains act by an equilibrium binding to the UIM site (Fig. 1), as demonstrated by the holdup binding assay (Fig. 4, C and D). The UIM site is located in the highly mobile C-terminal part including helix 17/19, which specifically interacts with helix 6 (Fig. 1C) both in the "super-open" (helix 17) and in the closed (helix 19) form of hGK. Interestingly, seven activating mutations of hGK have presently been characterized, all clustered in this defined area, and have been considered to represent an allosteric activator site (36). Two of the mutations (Y214C and Y215A) are located in helix 6 and three (V455M, A456V, and A460R) in helix 17/19. Moreover, several synthetic organic compounds, which activate the enzyme by a V max and [S] 0.5 effect, have been found to interact with the same site (i.e. Val-452 and Val-455) (20,37) and thus considered to represent potential antidiabetic drugs (20,37). The putative physiological endogenous activator interacting at this site has, however, still to be discovered.
Our data support the conclusion that polyubiquitin chains, either free or conjugated to proteins (Fig. 6), may represent such a physiological allosteric activator. Interestingly, the maximum stimulation of catalytic activity by Ub 5 -Lys-48 (V max ) was similar to that recently reported for the synthetic allosteric activator RO-28-1675 (ϳ1.5-fold) (9). Within an in vivo context, free polyubiquitin chains or polyubiquitinated proteins in GK-expressing cells (supplemental Fig. S2) may have a similar function. Thus, the level of Ub is reported to be 10 -20 M in a variety of cultured cell lines and in rabbit reticulocytes (38 -40), where the Ub conjugates represent ϳ80% of the total Ub level (38). Furthermore, free polyubiquitin chains represent a substantial portion of the total Ub pool in vivo (41), as in the reticulocyte lysates used in the present study (supplemental Fig. S2A).