INTRODUCTION

UDP-glucose (UDP-Glc)¹ is an essential metabolite in several cellular processes: the synthesis of glycogen (1, 2), the synthesis of the carbohydrate moiety of glycolipids (3), glycoproteins (4, 5), and proteoglycans (6), the entry of galactose into glycolysis (7), and the synthesis of UDP-glucuronic acid (UDP-GlcUA) (8, 9) (see Fig. 1). Furthermore, it plays a crucial role for the "quality control" of newly synthesized glycoproteins taking place in the endoplasmic reticulum (10). Here it is used by the UDP-Glc:glycoprotein glucosyltransferase, which recognizes and glucosylates misfolded glycoproteins (11-13). The added glucose is then removed by glucosidase II (14), and a deglucosylation-reglucosylation cycle occurs until the protein is correctly folded to allow its exit from the endoplasmic reticulum (10, 15).

A UDP-Glc deficiency occurs in insulin-dependent tissues of diabetic organisms (16-18), probably as a consequence of the impaired glucose transport into the cells. A low level of UDP-Glc also occurs in cultured cells incubated under hypoxic conditions (8, 9) or in glucose-deficient medium (19-22).

We previously isolated a mutant cell line, Don Q (23), which is the only mammalian cell reported to have a persistently low level of UDP-Glc (24). Don Q was obtained after mutagenesis with ethyl methanesulfonate of wild type Chinese hamster lung fibroblasts (referred to here as Don wt) and selection for resistance to Clostridium difficile toxin B (TcdB) (23). This cytotoxin produces collapse of the actin cytoskeleton via a glucosylation of the small GTPases Rho, Rac, and CDC42, which are members of the Ras superfamily (25). Don Q is as sensitive as the wild type to other endocytosed toxins like diphtheria, pertussis, and ricin (23) but is also resistant to the Clostridium sordellii lethal toxin (24). The latter cytotoxin is a glucosyltransferase that has as targets the small GTP-binding proteins Ras, Rap, and Rac (26, 27). The resistance of Don Q to the glucosyltransferase toxins from C. difficile and C. sordellii (23, 24) is explained by the finding that these toxins use UDP-Glc as a cofactor (28).

The synthesis of UDP-Glc (and other UDP-hexoses) depends on the amount of glucose-6-phosphate (Glc-6-P) and UTP available in the cell (see Fig. 1). Two enzymes are needed for UDP-Glc synthesis: phosphoglucomutase (EC 5.4.2.2), which interconverts Glc-6-P and glucose-1-phosphate (Glc-1-P), and UDP-glucose pyrophosphorylase (UDPG:PP; EC 2.7.7.9), which catalyzes the formation of UDP-Glc from UTP and Glc-1-P (see Fig. 1). Because the reaction catalyzed by UDPG:PP is near equilibrium (29-31), a persistently defective UDP-Glc production could be due to a lowered transport of glucose across the plasma membrane, a UTP deficiency, or a defect in either phosphoglucomutase or UDPG:PP. The aim of this work was to clarify the reason for the UDP-Glc deficiency in Don Q.

EXPERIMENTAL PROCEDURES

A mutant cell denoted as Cdt^RQ (23) (referred to here as Don Q) of the diploid Chinese hamster lung Don fibroblasts (ATCC CCL-16) was seeded into 96-well plates (Nunc, Roskilde, Denmark). Cells were cultivated at 37 °C in Eagle's medium (Flow Laboratories, Irvine, Scotland) supplemented with 10% fetal bovine serum, 5 mM L-glutamine, penicillin (100 units/ml), and streptomycin (100 µg/ml) in a humid atmosphere containing 5% CO₂. After 2 days, when the cells had grown to confluent monolayers, C. perfringens phospholipase C (PLC) (32) was added at a dose of 0.12 µg/ml, corresponding to 10 TCD₅₀ (tissue culture dose 50%) (23). After 1 week, the PLC was replaced with fresh medium, and after less than 2 weeks more, one clone of revertant cells was detected and expanded (denoted as Cdt^RQ-Rev and here referred to as Don QR).

Don wt, Q, and QR were seeded in 96-well plates (500-1000 cells/well), cultivated to 90% confluency, and exposed to serial 10-fold dilutions of C. difficile TcdB (500 µg/ml) (33, 34) or C. novyi -toxin (100 µg/ml) (35) in 200 µl of medium/well. Cell viability was measured 24 h later using a neutral red assay (36). Briefly, the cells were incubated for 2 h with 200 µl/well of neutral red (Sigma) (50 µg/ml) dissolved in culture medium supplemented as above. The incorporated dye was extracted from the cells with 100 µl of acetic acid:ethanol:water (1:50:49), before reading at 540 nm in a microtiter plate reader. All experiments were performed at least three times, with six replicate samples in each experiment.

The uptake of 2-deoxy-D-[6-³H]-glucose was measured essentially as described (37). Briefly, cells were cultivated in 96-well plates, rinsed three times with phosphate-buffered saline (PBS), and incubated with 2-deoxy-D-[6-³H]-glucose (2 µCi/ml) (Amersham Corp.) in glucose-free medium with or without cytochalasin B (Sigma) (5 µg/ml). After 5, 10, 20, 25, and 35 min at 37 °C, the cells were washed three times with ice-cold PBS and lysed in situ by incubation with 10 µl of 0.4% SDS for 10 min. Scintillation mixture (190 µl) (Hisafe, Wallac) was added, and the radioactivity in the wells was counted in a microtiter plate -counter. The specific uptake was calculated as the difference between uptake in the absence and in the presence of cytochalasin B (37) and related to the amount of cellular protein. Cells in parallel wells were lysed, and the amount of protein was determined using the Bio-Rad DC protein assay (Bio-Rad) using bovine serum albumin as a standard. All experiments were performed at least three times, with six replicate samples in each experiment.

The metabolites were extracted from confluent monolayers of three 75-cm² cell culture flasks (Nunc) of each cell line using a dual-phase extraction method (38). The methanol-water phase was freeze-dried and resuspended in a buffer with 8 mM Na₂HPO₄ and 2 mM KH₂PO₄, pH 7.2 (prepared with D₂O). The acquisition of the spectra was done as described elsewhere (39). The chemical shifts were referred to 3-(trimethylsilyl)propionic-2,2,3,3-d₄ acid (0 ppm). UDP-Glc, UDP-GlcUA, UDP-Gal, Gal-1-P, and Glc-1-P were quantified by integration of the double doublets assigned to the protons of the anomeric carbon centered at 5.636, 5.618, 5.601, 5.493, and 5.453 ppm, respectively. UTP was quantified by integration of the doublet assigned to the meta-uracylic proton centered at 7.975 ppm, whereas Glc-6-P was quantified by integration of the doublet assigned to the proton of the anomeric carbon centered at 5.250 ppm. The amount of each metabolite was related to the amount of cellular protein determined in lysates from parallel flasks. All experiments were performed at least twice.

The activity of UDPG:PP was measured essentially as described (40). Briefly, confluent cell monolayers, one 75-cm² flask of each cell line, were washed twice with PBS and mechanically removed. Cell pellets were resuspended in 200 µl of lysis buffer (0.08 M Tricine-HCl, 0.03% Triton X-100, 0.1 mM phenylmethylsulfonyl fluoride (Sigma), 10 µg/ml leupeptin (Boehringer Mannheim Scandinavia AB, Bromma, Sweden), pH 7.4)). Protein concentration in cell lysates was determined as above. Cell lysates (1-20 µg protein) were incubated with UTP (Sigma) (5 mM) and [¹⁴C]-Glc-1-P (302 mCi/mmol) (Amersham Corp.) at 37 °C for 30 min. The reaction was stopped by heating to 100 °C for 2 min. Residual [¹⁴C]-Glc-1-P was removed by incubation with 0.5 units of Escherichia coli alkaline phosphatase (Sigma) at 45 °C for 30 min. An aqueous suspension of charcoal (20 mg/ml) was added to absorb the UDP-[¹⁴C]-Glc formed, and after 10 min the charcoal was washed three times with water. Finally, the charcoal pellets were extracted with 90 µl of acetic acid (10%), and 2 ml of scintillation mixture was added before counting. All experiments were performed at least twice, with three replicate samples in each experiment.

Total RNAs from Don wt, Q, and QR cells were isolated using the RNeasy kit from Qiagen GmbH (Hilden, Germany), and the concentrations were determined spectrophotometrically as absorbance at 260 nm. Reverse transcription was performed as follows: 1.5 µg of the purified RNA was incubated with 2 µl of oligo(dT) (50 µM) (the volume was adjusted to 10 µl with distilled water) at 70 °C for 5 min and then chilled on ice. Then 1 µl of avian myeloblastosis virus-reverse transcriptase (24 U/µl) (Seikagaku America Inc., Ijamsville, MD), 4 µl of dNTP (2.5 mM), 0.5 unit of ribonuclease inhibitor (40 units/µl) (Boehringer), and 2 µl of 10 × reverse transcriptase-first strand buffer were added, and the volume was adjusted to 20 µl with distilled water. After heating the reaction mixture to 95 °C for 5 min and incubation for 60 min at 40 °C, the resulting cDNAs were stored at 20 °C until use. Primers (5-TCGGAGGAGAAAGAACACAT-3 and 5-AGGCAATCAAAGACCATCAC-3) to amplify a 1913-base pair fragment that encompasses the complete coding region of UDPG:PP were designed according to the human UDPG:PP cDNA sequence (GenBank^TM accession number L14430; Ref. 58). PCR was performed with 3 µl of cDNA, 4 µl of dNTPs (2.5 mM), 3 µl of each primer (10 µM), 2.5 µl of Dynazyme (2 units/µl) (Finnzymes Oy, Espoo, Finland), and 5 µl of Dynazyme buffer (10×), the volume being adjusted to 50 µl with distilled water. After 3 min at 95 °C, the mixture was incubated for 27 cycles at 95 °C for 30 s, 50 °C for 30 s, and 72 °C for 2 min. For semiquantitative RT-PCR, a 648-base pair fragment was amplified for 20 cycles using the primers 5-TAGAAGAAGGCAAAGAGTAT-3 and 5-TCCAGAAACAGTGAGGTGAT-3, which were designed according to the sequence of the human UDPG:PP cDNA.

For Northern blot analysis the DNA fragment of the Chinese hamster UDPG:PP used as a probe was obtained by PCR amplification with the same primers used for semiquantitative RT-PCR. The fragment was run on a 1% low melting temperature agarose gel, stained with ethidium bromide, excised under UV transillumination, and purified using the BandPret kit (Pharmacia Biotech Inc.). The purified fragment was then labeled with [-³²P]CTP using the kit Ready to go (Pharmacia) according to the manufacturer's instructions.

Samples containing 20 µg of total RNA from each cell line were denatured, run on low formaldehyde (2.5% v/v) agarose gels, and blotted onto a positively charged nylon membrane (Pharmacia) by capillary transfer, using 10 × SSC as transfer buffer. The membrane was cross-linked by 1-min exposure to low wavelength UV, and prehybridizations were performed for 1-2 h at 55 °C in Rapy-hyb buffer (Amersham Corp.). Hybridizations were performed at 55 °C for 18-20 h as described (41). The membrane was washed three times in 1 × SSC/1% SDS for 20 min at room temperature. Hybrid bands were detected using a PhosphorImager (Molecular Dynamics Inc., Sunnyvale, CA), and radioactivity was quantified using the Image Quant Program (Molecular Dynamics Inc.).

For Western blot analysis confluent cell monolayers were washed twice with PBS and mechanically removed. Cell pellets were resuspended in 200 µl of 0.4% SDS containing 0.1 mM phenylmethylsulfonyl fluoride and 10 µg/ml leupeptin. Protein concentration in cell lysates was determined as above. Cell lysates (30 µg protein/lane) were submitted to electrophoresis under reducing conditions on a SDS-polyacrylamide gel (42) (10% w/v), and separated proteins were electroblotted (43) on to 0.45-µm nitrocellulose membrane (Bio-Rad). The nitrocellulose membrane was saturated with 2% bovine serum albumin and 0.1% Nonidet P40 in PBS (blocking solution) for 2 h at room temperature before a 2-h incubation with rabbit antibodies against potato UDPG:PP in blocking solution. Unbound antibodies were removed by washing four times (5-10 min each) with blocking solution before incubation with an anti-rabbit IgG-alkaline-phosphatase conjugate (Dako A/S, Glostrup, Denmark) for 2 h at room temperature. The nitrocellulose membrane was washed as described above and incubated with nitro blue tetrazolium (Sigma) (250 µg/ml) and 5-bromo-4-chloro-3-indolyl phosphate (Sigma) (56 µg/ml) in 500 mM NaCl, 20 mM Tris buffer, pH 7.5. Immunoreactive bands were detected using a densitometer and quantified using the Image Quant Program.

The PCR-amplified products were purified as described above. Sequence analysis was performed on the GeneAmp PCR System 9600 using the Applied Biosystems Prism^TM dye terminator cycle sequencing kit (Perkin-Elmer/Applied Biosystems Division, Foster City, CA) according to the manufacturer's instructions. Briefly, 10-30 ng/µl (3-6 µl) of the gel-purified PCR product were added to a MicroAmp^TM reaction tube (Perkin-Elmer) containing 3.2 pmol of sequencing primer, 8.0 µl of premixture (containing buffer, dNTPs, dye-labeled ddNTPs and AmpliTaq), and distilled water to a final volume of 20 µl. After the initial denaturation at 96 °C for 2 min, the reaction mixture was incubated for 25 cycles at 96 °C for 10 s, 50 °C for 5 s, and 60 °C for 4 min. Sequence analysis was performed on both strands. The sequencing primers were designed according to the sequence of the human UDPG:PP cDNA. For the coding strand they were: 5-CTTACACCAATCAGACTTTT-3, 5-ATTAAAACAAGAGGGACATC-3, 5-ATACTCTTTGCCTTCTTCTA-3, 5-TCAGGCCTCCATCCAAAGTC-3, 5-TCCAGAAACAGTGAGGTGAT-3, and 5-AGGCAATCAAAGACCATCAC-3. For the noncoding strand they were: 5-TCGGAGGAGAAAGAACACAT-3, 5-TTACAACAGCAACCTCACAT-3, 5-CAAAAGCTACAATACAGATG-3, 5-TAGAAGAAGGCAAAGAGTAT-3, 5-GAGTCGTTTTCTGCCTGTGA-3, and 5-GGGAACAGTTATCATTATTG-3. The extension products were purified by ethanol precipitation and evaporated to dryness under reduced pressure (SpeedVac®, Savant Instruments, Holbrook, NY). Each sample was resuspended in 4 µl of loading buffer (5:1 deionized formamide/50 mM EDTA, pH 8.0), heated for 2 min at 90 °C, and loaded onto a model 373A Sequencer (PE/ABI) according to the manufacturer's instructions. The sequence analysis of both strands was performed at least twice for each cell line.

A multiple sequence alignment was performed using the program Clustal W (44). Secondary structure elements were predicted (45) and summarized for each alignment position. Hydropathy plots were calculated using a sliding 6-residue window (46). Motifs (47) were defined as regions with at least 11 strictly conserved residues among 30 consecutive positions. The molecular modelling and picture of the model were made with the ICM program (version 2.6, 1996, Molsoft, Metuchen, NJ).

RESULTS AND DISCUSSION

Don Q is 10⁴ times more resistant to the TcdB-induced cytopathogenic effect and 10⁵ times more sensitive to the cytotoxic effect of PLC than Don wt (23). By culturing Don Q in the presence of PLC, we isolated a revertant cell, Don QR, which has regained the sensitivity to TcdB and the relative resistance to PLC displayed by Don wt (Fig. 2A and data not shown).

Dose response of Don wt, Q, and QR cells to treatment with C. difficile TcdB or C. novyi -toxin.

Because Don Q has a deficiency of UDP-Glc, a metabolite required for the entry of galactose into glycolysis (Fig. 1), we compared the ability of the three cell lines to grow on galactose as the only energy source. Don QR grew almost equally well as Don wt, whereas Don Q was not able to survive under this condition (not shown). The regained TcdB sensitivity of Don QR and its capacity to metabolize galactose indicated that the UDP-Glc deficiency in Don Q was at least partially compensated in Don QR. In this work we used the cell lines Don wt, Q, and QR to determine the reason for the low UDP-Glc level in Don Q.

To determine whether Don Q has a general deficiency in UDP-hexoses, its sensitivity to C. novyi -toxin was studied. This toxin catalyzes the incorporation of GlcNAc into Rho subfamily proteins using UDP-N-acetylglucosamine as a cofactor (48). The three cell lines showed the same dose response for development of the -toxin-induced cytopathogenic effect (Fig. 2B). Because Don Q was not more resistant to -toxin, the low level of UDP-Glc in this cell appears not to be associated with a general deficiency in UDP-hexoses. This result also implies that Don Q does not have a UTP deficiency, because this metabolite is a precursor for the synthesis of UDP-N-acetylglucosamine (Fig. 1).

The facilitated diffusion of glucose across the plasma membrane is a limiting step in the synthesis of UDP-Glc and glycogen (49-51). Therefore, the uptake of 2-deoxyglucose, a nonmetabolizable glucose analogue, was measured in the three cell lines (Fig. 3). The 2-deoxyglucose uptake in Don Q and Don wt was the same, whereas in Don QR it was slightly increased. This result indicated that the low level of UDP-Glc in Don Q is not due to a defective glucose uptake.

The cellular levels of the metabolites of the UDP-hexose pathways in the three cell lines were measured by ¹H NMR spectroscopy (Table I). The UDP-Glc level in Don Q was 26% of that in Don wt, in agreement with our previous measurement using reverse phase high pressure liquid chromatography (24). Furthermore, Don QR was found to have a UDP-Glc level of 64% of that found in Don wt. This result indicated that the UDP-Glc deficiency of Don Q has been only partially compensated in Don QR. In agreement with this, the level of UDP-GlcUA in Don QR lay between those of Don wt and Don Q (Table I).

Table I. Levels of glucose metabolites in Don wt, Q, and QR cells Metabolites from the three cell lines were extracted and 1H NMR spectra were obtained as described under "Experimental Procedures." Protein concentrations were determined in parallel flasks. The values represent the means of two determinations. Metabolite Concentration Don wt Don Q Don QR nmol/mg protein UDP-Glc 5.0 1.3 3.2 UDP-GlcUA 1.2 0.2 0.5 Gal-1-P 1.2 4.0 0.8 UDP-Gal 2.5 1.0 2.1 Glc-6-Pa 130.0 126.0 123.0 Glc-1-P 0.8 5.2 0.7 UTP 3.9 3.8 4.0 a Quantified together with traces of glucose.

The assignments of the proton on the anomeric carbon of UDP-Gal and Gal-1-P to double doublets in the region from 5.7 to 5.4 ppm of the the ¹H NMR spectra were performed for the first time (Fig. 4A). Don Q has the highest level of Gal-1-P and the lowest level of UDP-Gal compared with Don wt and QR (Table I), indicating an impaired capacity to interconvert these metabolites. The low UDP-Glc level in Don Q explains this impairment, because UDP-Glc is a required cofactor for this interconversion (Fig. 1).

When glucose is incorporated into a cell, it is immediately phosphorylated to Glc-6-P by hexokinases (52) and two steps later converted to UDP-Glc (Fig. 1). The three cell lines showed similar levels of Glc-6-P (Table I), indicating that the synthesis of this metabolite is not impaired in Don Q. Normally, the cellular level of UDP-Glc reflects the concentration of Glc-6-P (52). However, Don Q has a lower level of UDP-Glc than Don wt and QR, despite having the same Glc-6-P concentration. This finding indicates that there is an impairment in the UDP-Glc synthesis in a step downstream of Glc-6-P. Because the level of Glc-1-P in Don Q was not lower than in Don wt and QR (Table I), we concluded that the reaction catalyzed by phosphoglucomutase is not impaired in Don Q. The three cell lines showed a similar level of UTP (Table I), confirming that there is no deficiency of this precursor. Our results therefore indicated that the low level of UDP-Glc in Don Q is not due to a deficiency in Glc-6-P, Glc-1-P, or UTP. Because Don Q has altered levels of Glc-1-P and UDP-Glc compared with Don wt and QR, we decided to measure the activity of UDPG:PP, the enzyme that interconverts these metabolites.

The activity of UDPG: PP was measured in the three cell lines. Don Q showed 4% and Don QR showed 56% of the UDPG:PP activity found in Don wt (Table II). These results indicated that Don Q has an impaired capacity to synthesize UDP-Glc, which has been partially restored in Don QR. In agreement with this, the UDP-Glc level in Don QR lay between those found in Don wt and Q (Table I). The residual UDPG:PP activity in Don Q together with an increased production of Glc-1-P may explain why the UDP-Glc level is not lower than 26% in this cell. We conclude that the reason for the UDP-Glc deficiency in Don Q is a defect in its production, because of a lowered UDPG:PP activity. This conclusion has been further substantiated by the finding that Don Q cells transfected with the bovine UDPG:PP cDNA were reverted with respect to their response to TcdB.²

Table II. Activity of UDPG:PP in Don wt, Q, and QR cells Cell lysates were prepared, protein concentration was determined, and enzymatic activity was measured as described under "Experimental Procedures." The values represent the means ± S.D. of three determinations. Cell line Activity nmol UDP-Glc/min × mg protein Don wt 2.71  ± 0.31 Don Q 0.11  ± 0.01 Don QR 1.52  ± 0.14

The lowered UDPG:PP activity in Don Q and QR could result from a decreased gene transcription, an increased mRNA degradation, or a mutation in the coding region of the gene. The amount UDPG:PP mRNA in Don wt, Q, and QR was similar, as determined by semiquantitative RT-PCR and Northern blot (not shown). Furthermore, the amount of UDPG:PP was also similar in the three cell lines, as determined by Western blot analysis using polyclonal antibodies against the potato enzyme (not shown). These results suggested that the low UDPG:PP activity in Don Q is not due to a decreased gene transcription or an increased mRNA degradation. Therefore, we decided to establish the complete nucleotide sequence of the UDPG:PP cDNAs from the three cell lines to determine whether there was a mutation in the coding region of the gene associated with the low enzymatic activity.

In this work, we established the nucleotide sequence of the Chinese hamster UDPG:PP cDNA for the first time. The deduced amino acid sequence of the corresponding protein consists of 507 amino acid residues and shows homology with other eukaryotic UDPG:PPs (Fig. 5).

A comparison of the complete cDNA sequences from the three cell lines was performed. Don Q cDNAs have one single point mutation, a guanine to adenine transition at position 347 of the coding sequence (Fig. 6A), which changes amino acid residue 115 from glycine to aspartic acid. This glycine residue is strictly conserved among eukaryotic UDPG:PPs (position 133 in the alignment of Fig. 5).

The type of mutation found in the UDPG:PP gene in Don Q was the expected according to the agent used to generate this mutant cell. Ethyl methanesulfonate is a monoalkylating agent, which favors guanine to adenine transitions (53-55). This agent ethylates guanine nucleotides at the O-6 position creating a modified nucleotide that pairs preferentially with thymidine rather than cytosine (56, 57). During the following replication round this change places an adenine in the coding strand.

Sequence analysis of cDNAs from Don QR showed an equimolar mixture of guanine and adenine at position 347 (Fig. 6A). Corresponding results were obtained by sequencing the noncoding strand (Fig. 6B). The sequence analysis data strongly suggest that the mutation is present in homozygous form in Don Q and is reverted in only one of the alleles in Don QR. In agreement with this, the enzymatic activity of UDPG: PP in Don QR lies between those of Don wt and Don Q (Table II). We conclude that the G115D substitution is responsible for the impairment in the UDPG:PP activity and thereby for the UDP-Glc deficiency in Don Q.

The primary structure of several eukaryotic UDPG:PPs has been reported (58-64), but the three-dimensional structure has not been determined for any of them. We compared the sequences of all known eukaryotic UDPG:PPs to identify the conserved residues, which should correspond to those important for the structure/function of these enzymes. Alignment of nine UDPG:PPs revealed a pairwise residue identity of 97.0-99.6% between mammalian UDPG:PPs and 33.0-53.7% between mammalian and nonmammalian enzymes. There are 89 residues strictly conserved among UDPG:PPs from a variety of evolutionary diverse organisms (Fig. 5). Most of the conserved residues are glycine (16 glycine residues are invariant), a typical finding in proteins with similar three-dimensional structure (65-67). There is no strong similarity between the UDPG:PPs from plants, mammals, and fungi in the first 120 residues of the N-terminal regions. The strictly conserved residues are grouped into short segments distributed throughout the rest of the polypeptide chains (Figs. 5 and 7). Three clearly defined motifs (47) can be identified at alignment positions 129-159, 262-292, and 414-444 (Figs. 5 and 7). A strictly conserved lysine located within the third motif (at position 443 in the alignment shown in Fig. 5) has been shown to be essential for the UDPG:PP activity of the potato enzyme (68).

Predictions of secondary structure were performed and summarized (Fig. 7) for the nine eukaryotic UDPG:PPs included in the alignment of Fig. 5. The similarities in the predicted secondary structures cover most of the entire sequences (Fig. 7). Hydropathy curves for the known eukaryotic UDPG:PPs were superimposed (Fig. 7). The agreements between the hydropathy curves are also extensive, with largely parallel patterns at segments covering alignment positions 120-240, 270-390, and 410-470 (Fig. 7). Taken together, the similarities in the sequences, predicted secondary structures, and hydropathy curves suggest that all eukaryotic UDPG:PPs have a similar three-dimensional structure.

According to the secondary structure predictions of eukaryotic UDPG:PPs (Fig. 7), the segment encompassing alignment positions 132-143 constitutes a reverse turn. The conserved glycine substituted by the mutation found in Don Q UDPG:PP is located at this predicted turn. The secondary structure prediction (45) was not changed in the mutated enzyme (not shown). In contrast, comparison of hydropathy plots (46) of the wild type and the mutated sequences showed that the substitution causes a local shift toward a lower hydrophobicity (not shown).

The largest segment of strictly conserved residues among eukaryotic UDPG:PPs (KLNGGLG) encompasses alignment positions 129-135 (Figs. 5 and 7). Three out of these seven residues are glycine, which indicates stringent space restrictions (69), suggesting structural/functional importance of this stretch. The point mutation found in Don Q UDPG:PP substitutes the second glycine of the 129-135 stretch. Interestingly, a part of this sequence (NGGLG) binds Glc-1-P and comprises part of the active site in glycogen phosphorylases and E. coli maltodextrin phosphorylase (70-72). The three-dimensional structure of the active site peptides NGGLG (Fig. 8A) superimpose precisely in these enzymes (73). In glycogen phosphorylase, the second glycine of this peptide (Gly-135) makes two hydrogen bond contacts with Glc-1-P (72). The sequence identity between these conserved stretches of phosphorylases and UDPG:PPs raises the possibility that the stretch NGGLG is also part of the Glc-1-P binding site in eukaryotic UDPG:PPs.

Molecular modelling of a G135D substitution in glycogen phosphorylase (Fig. 8B) shows that space restrictions would occur at the active site due to its interaction with the Glc-1-P and the pyridoxal phosphate. Assuming that the NGGLG stretch in the Chinese hamster UDPG:PP has a similar conformation to that of glycogen phosphorylase, the steric hindrance caused by the G115D substitution would explain the impaired activity of the mutated enzyme. Alternatively, Gly-115 may be important for the local flexibility and/or conformational stability of UDPG:PPs. Regardless of the molecular mechanism, our data demonstrate that the change in the bulky side chain of residue 115 and the additionally introduced negative charge of the aspartic acid have a significant effect on the UDPG:PP activity. Furthermore, they suggest that the conserved Gly-115 has an important role for the activity of these enzymes.

In the Chinese hamster UDPG:PP gene we have identified a point mutation (G115D) that dramatically impairs the enzymatic activity of the protein product, leading to a persistently low UDP-glucose level. UDP-glucose deficiency occurs in insulin-dependent tissues of diabetic organisms (16-18), and the consequences of this deficiency have not been established. A low level of UDP-Glc also occurs in cultured cells incubated under low glucose concentration (19-22) or hypoxia (8, 9), two of the hallmarks of ischemia. We hypothesize that a low UDP-Glc level occurs in ischemic tissues and suggest that the cells Don Q and QR constitute an excellent model to determine the molecular consequences of a UDP-Glc deficiency.