High Affinity Binding and Allosteric Regulation ofEscherichia coli Glycogen Phosphorylase by the Histidine Phosphocarrier Protein, HPr*

The histidine phosphocarrier protein (HPr) is an essential element in sugar transport by the bacterial phosphoenolpyruvate:sugar phosphotransferase system. Ligand fishing, using surface plasmon resonance, was used to show the binding of HPr to a nonphosphotransferase protein in extracts ofEscherichia coli; the protein was subsequently identified as glycogen phosphorylase (GP). The high affinity (association constant ∼108 m −1), species-specific interaction was also demonstrated in electrophoretic mobility shift experiments by polyacrylamide gel electrophoresis. Equilibrium ultracentrifugation analysis indicates that HPr allosterically regulates the oligomeric state of glycogen phosphorylase. HPr binding increases GP activity to 250% of the level in control assays. Kinetic analysis of coupled enzyme assays shows that the binding of HPr to GP causes a decrease in the K m for glycogen and an increase in the V max for phosphate, indicating a mixed type activation. The stimulatory effect of E. coliHPr on E. coli GP activity is species-specific, and the unphosphorylated form of HPr activates GP more than does the phosphorylated form. Replacement of specific amino acids in HPr results in reduced GP activation; HPr residues Arg-17, Lys-24, Lys-27, Lys-40, Ser-46, Gln-51, and Lys-72 were established to be important. This novel mechanism for the regulation of GP provides the first evidence directly linking E. coli HPr to the regulation of carbohydrate metabolism.

The bacterial phosphoenolpyruvate:sugar phosphotransferase system (PTS) 1 (1,2) plays an important role in the uptake and concomitant phosphorylation of simple sugar substrates by a reaction involving a chain of phosphocarrier proteins. In the case of glucose transport in Escherichia coli, there are three soluble PTS components (enzyme I (EI), histidine phosphocarrier protein (HPr), and enzyme IIA Glc (IIA Glc )). Glu-cose uptake requires sequential phosphoryl transfer from four PTS proteins, as follows: P-enolpyruvate 3 EI 3 HPr 3 IIA Glc 3 enzyme IIB, C Glc 3 glucose. In this pathway, HPr acts as part of a phosphoryl shuttle between EI and sugar-specific proteins.
Since the PTS was discovered in E. coli nearly 30 years ago (3), additional studies have documented other metabolic roles of this system. The PTS regulates chemotaxis to PTS sugars, adenylate cyclase, and certain non-PTS permeases (2). Genetic and biochemical studies implicate the crr gene product (IIA Glc ) as the mediator of some of these regulatory phenomena. The current model suggests that IIA Glc and/or P-IIA Glc interact with target proteins and that the ratio of P-IIA Glc to IIA Glc may be the determining factor in PTS-mediated regulation. Until now, there has been no direct evidence for the participation of the two general PTS proteins (EI and HPr) in the regulation of other metabolic pathways in enteric bacteria except that GTP binding to HPr may mediate the GTP-dependent stimulation of the activity of the adenylyl cyclase complex in E. coli (4).
E. coli HPr is a small, monomeric protein with a M r , predicted from its amino acid sequence, of 9119 (5). Its threedimensional structure has been determined by x-ray crystallographic and two-and three-dimensional nuclear magnetic resonance (NMR) analysis (6 -8). Phosphoryl transfer from P-EI (His-189) to HPr is at His-15 (9). In Gram-positive bacteria, HPr is phosphorylated at a second site, Ser-46, by a metabolite-activated, ATP-dependent protein kinase and dephosphorylated by a P i -dependent phosphatase (10). The Ser-46 phosphorylation site appears to play a role in the regulation of catabolite repression in Bacillus subtilis (11). HPr has also been shown to regulate glycerol kinase by a phosphorylationdephosphorylation mechanism in Gram-positive bacteria such as Enterococcus faecalis (12).
In this study, ligand fishing, using surface plasmon resonance 2 spectroscopy, was employed to discover the high affinity binding of HPr to a protein in E. coli extracts. The protein was purified and established to be glycogen phosphorylase (GP). E. coli GP has a M r , predicted from its amino acid sequence, of 93,144, and it has been suggested, from gel filtration chromatography experiments, to exist as a dimer (13). GP is constitutively expressed (14) and is encoded (glgP) on the glgCAP operon, which also includes genes for ADP-glucose pyrophosphorylase and glycogen synthase, respectively (15). GP was overexpressed and purified. A variety of physical studies established that HPr is an allosteric effector of GP and that it specifically stimulates GP activity by a mixed type of mechanism. Several amino acids in HPr important for GP activation were identified. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. § The first two authors contributed equally to this work. ** To whom correspondence should be addressed: National Institutes of Health, Bldg. 36, Room 4C-11, Bethesda, MD 20892. Tel.: 301-496-2408; Fax: 301-480-0182; E-mail: alan@codon.nih.gov. 1 The abbreviations used are: PTS, phosphoenolpyruvate:sugar phosphotransferase system; SPR, surface plasmon resonance; RU, response units; EI, enzyme I of the PTS; IIA Glc , enzyme IIA Glc of the glucose PTS; IIA Fru , enzyme IIA Fru of the fructose PTS; NPr, the E. coli phosphocarrier protein that may control the state of phosphorylation of IIA Ntr ; DTP, the diphosphoryl transfer protein of the E. coli PTS; GP, glycogen phosphorylase; HPr, histidine phosphocarrier protein; P-HPr, phosphorylated HPr; HPLC, high pressure liquid chromatography; RP-HPLC, reverse phase-HPLC; PAGE, polyacrylamide gel electrophoresis.
Before the present studies were carried out, there was no conclusive evidence that associated HPr with regulatory functions in Gram-negative bacteria, including E. coli; further, it was previously thought that E. coli and other prokaryotic GPs are not subject to regulation (16,17).

EXPERIMENTAL PROCEDURES
Materials-Rabbit muscle glycogen phosphorylase a and b were from Sigma. Yeast glucose-6-phosphate dehydrogenase, yeast hexokinase, rabbit muscle phosphoglucomutase, and restriction enzymes SalI and NdeI were obtained from Boehringer Mannheim. Pfu DNA polymerase was a Stratagene product. Mark 12 TM Wide-Range protein standards were purchased from NOVEX. EI and IIA Glc from E. coli were purified after hyperexpression by a previously described method (18). HPr was expressed and purified using GI698 transformed with expression vector pSP100 as described recently (19). HPrs from B. subtilis and Mycoplasma capricolum were gifts from Dr. Peng-Peng Zhu. A collection of HPrs carrying single replacements was a gift from Dr. E. Bruce Waygood. Dr. Jonathan Reizer generously provided essentially pure E. coli NPr as well as E. coli DTP (approximately 50% pure).
Protein Sequencing-Two samples of the HPr-binding protein (10 g) were dried in a Speed Vac concentrator (Savant, Farmingdale, NY). The resulting residues were dissolved in 50 l of 8 M urea, 0.4 M NH 4 HCO 3 and subjected to reduction, alkylation, and proteolytic digestion with (a) 1.0 g of endoproteinase Lys-C (Boehringer Mannheim) or (b) 1.0 g of modified trypsin (Promega, Madison, WI) according to the method of Stone and Williams (20). The resulting digests were separated by RP-HPLC on a narrowbore (2.1 ϫ 250 mm) Vydac 218TP52 column/guard column combination (Separations Group, Hesperia, CA) with elution at 0.25 ml/min at 35°C, utilizing the gradient described by Fernandez et al. (21) on a System Gold HPLC equipped with a model 507 autosampler, model 126 programmable solvent module, and model 168 diode array detector (Beckman, Fullerton, CA). Solvent A was 0.1% trifluoroacetic acid in water, and solvent B was 0.1% trifluoroacetic acid in acetonitrile. Column effluent was monitored at 215 and 280 nm. Fractions were collected at 30-s intervals and stored at Ϫ70°C. Selected fractions from the trypsin digest containing mixtures of peptides were pooled, concentrated in a Speed Vac concentrator, and further purified by a second RP-HPLC step on a narrowbore (2.0 ϫ 250 mm) YMC ODS AQ column (YMC, Morris Plains, NJ). Elution was at 0.25 ml/min at 35°C on the same HPLC system utilizing the previously described gradient and buffers. Fractions (125 l) containing peptides were applied in 30-l aliquots to a Biobrene-treated glass fiber filter (Applied Biosystems, Foster City, CA) and dried prior to amino acid sequencing on a model 477A pulsed liquid protein sequencer equipped with a model 120A phenylthiohydantoin analyzer (Applied Biosystems), using methods and cycles supplied by the manufacturer. Data were collected and analyzed on a model 610A data analysis system (Applied Biosystems). Amino acid sequences were searched in the GCG-Swiss Protein Data base (University of Wisconsin Genetics Computer Group, Madison, WI).
Measurement of Protein-Protein Interaction-The interactions of glycogen phosphorylase and PTS proteins were monitored by SPR detection using a BIAcore optical biosensor (Pharmacia Biotech Inc.). 2 PTS proteins were immobilized onto the carboxymethylated dextran surface of a CM5 sensor chip. Enzyme I (70 l, 20 g/ml), HPr (70 l, 100 g/ml), or IIA Glc (70 l, 100 g/ml) in coupling buffer (10 mM sodiumformate, pH 4.0) was allowed to flow over a sensor chip at 10 l/min to couple the protein to the matrix by a NHS/EDC reaction (70 l of mix). Unreacted N-hydroxysuccinimide was inactivated by injecting 70 l of 1 M ethanolamine-HCl, pH 8.0. A blank surface was prepared by activation and inactivation of the sensor chip without any protein immobilization. Assuming that 1000 resonance units (RU) corresponds to a surface concentration of 1 ng/mm 2 , EI, HPr, and IIA Glc were immobilized to a surface concentration of 2.9, 1.4, and 2.0 ng/mm 2 , respectively.
The standard running buffer was 10 mM Hepes (pH 7.4), 150 mM NaCl, and 0.005% Tween 20, and all reagents were introduced at a flow rate of 10 l/min. The sensor surface was regenerated between assays by injecting 10 l of water to remove bound analyte. Kinetic parameters for the interaction of E. coli glycogen phosphorylase with immobilized HPr were determined using the BIA-evaluation 2.1 software (Pharmacia).
Cloning of Glycogen Phosphorylase Gene-Primers possessing the synthetic restriction enzyme sites NdeI, located 3 base pairs upstream from the ATG start codon (in boldface type) (5Ј-TTTCAGGAAACGC-CCATATGAATGCTCCGT-3Ј) of the glgP gene (encoding glycogen phosphorylase), and SalI, located 1 base pair downstream from the TAA stop codon (in boldface type) (5Ј-CGTTCTATTTATTGGTCGACTTA-CAATCTC-3Ј), were constructed (restriction sites underlined). Using these primers, the glgP gene from pJF02 (13) was amplified for 30 cycles at 94°C, 1 min (denaturing), 50°C for 2 min (annealing), and 72°C for 3 min (extension) in a Perkin-Elmer DNA thermal cycler. The pJF02 was a gift from Dr. M. Inouye. After digestion, the NdeI-SalI fragment (2454 base pairs) was inserted into the corresponding sites of pRE1 (22), and the recombinant plasmid (pGP, 7399 base pairs) ( Fig. 1) was electroporated with a Bio-Rad E. coli pulser into E. coli strain GI698 (23), which encodes the gene for the repressor under control of the trp promoter.
Expression of Glycogen Phosphorylase-Cells grown overnight were used to inoculate 1 liter of synthetic medium (23) supplemented with 50 g/ml ampicillin to an A 600 ϭ 0.25-0.30, and the culture was incubated at 30°C. When the culture reached an A 600 ϭ 0.4, tryptophan (100 g/ml) was added to induce glgP expression. The cells were harvested at an A 600 of 1.8 (7-8 h after induction) and washed once with 25 mM Tris⅐HCl, pH 7.5, containing 100 mM NaCl, and the cell pellet was stored at Ϫ20°C.
Purification of Glycogen Phosphorylase-The steps of purification were followed by SDS-polyacrylamide gel electrophoresis (SDS-PAGE). GP activity was determined as described below and protein concentration was determined by the Bradford method (24) (see Table I). The cell pellet was resuspended in 25 mM Tris⅐HCl, pH 7.5, containing 100 mM NaCl and then passed three times through a French press at 10,000 p.s.i. The lysate was cleared of cell debris by centrifugation at 17,000 ϫ g for 100 min (crude extract). The majority of GP (approximately 80%) was found in the insoluble portion of the lysate due to inclusion body formation associated with the high level of GP expression. Approximately 15% of the protein in the crude extract corresponded to GP. The crude extract was chromatographed through an FPLC Mono-Q 10/10 column (Pharmacia) using a gradient of 100 -500 mM NaCl (total volume, 80 ml). Fractions at approximately 450 mM NaCl were enriched in glycogen phosphorylase according to analysis using SDS-PAGE (4 -20% gradient gel, Novex). These fractions were pooled and concentrated in a 10 K Macrosep centrifugal concentrator (Filtron) in the presence of 1 mM dithiothreitol and 0.1 mM phenylmethylsulfonyl fluoride. At this stage, the GP was approximately 75% pure. The concentrated Mono-Q fractions were chromatographed through a Superose 12 column (1.6 ϫ 50 cm) using 25 mM Tris⅐HCl, pH 7.5, with 100 mM NaCl. The fractions were analyzed, pooled, and concentrated as described for the Mono-Q column and then stored at Ϫ80°C. Table I shows the details of the purification. The overall purification was 7.7-fold from the crude extract to the final preparation. From 1 liter of culture (wet weight of cells, 3.24 g), a yield of 16.24 mg (24) of approximately 95% pure GP was obtained.
Glycogen Phosphorylase Activity Assay-Glycogen phosphorylase activity was measured using a phosphoglucomutase-and glucose-6-phosphate dehydrogenase-coupled enzyme assay (25). The reaction mixture (1 ml) contained 50 mM sodium phosphate, pH 7.1, 1 mM MgCl 2 , 0.2% glycogen, 50 nM glucose 1,6-diphosphate, 1 mM 2-mercaptoethanol, 0.6 mM NADP, 10 g/ml phosphoglucomutase, 2 g/ml glucose-6-phosphate dehydrogenase, and 5 g/ml glycogen phosphorylase, unless noted otherwise. Two minutes after the initiation of the reaction with GP, the A 340 , a measure of the increase in NADPH, was recorded every minute for a total of 10 min in a Pharmacia LKB Ultrospec Plus Spectropho- tometer. A unit of glycogen phosphorylase activity is defined as the reduction of 1 mol of NADP/min at 25°C.
Gel Shift Assay-PAGE was used to demonstrate the shift of GP mobility due to its interaction with HPr in a nondenaturing discontinuous system (3% stacking gel in Tris⅐HCl, pH 6.7, and 6% resolving gel in Tris⅐glycine, pH 8.3). Tris⅐glycine (pH 8.3) was used as the running buffer, and the temperature was maintained at 15°C using a NOVEX Thermoflow temperature controller. A binding mixture (10 l) of 100 mM Tris⅐HCl, pH 7.5, 2 mM MgCl 2 , 10 mM KCl, 1 mM dithiothreitol, 200 g/ml glycogen phosphorylase, and 40 g/ml of HPr (approximately a 2:1 molar ratio of HPr to GP) was allowed to incubate at room temperature for 10 min. This mixture was combined with 3 l of 5 ϫ loading buffer (0.01% bromphenol blue, 0.5 M Tris⅐HCl, pH 6.8, 50% glycerol) and electrophoresed on a pre-equilibrated gel for 1 h at 70 V, followed by 2 h at 100 V. For analysis of gel shifts associated with mutant HPrs (see Table III), each gel consisted of 10 lanes and was loaded in the following manner: lanes 1 and 10, GP control; lane 2, GP plus wild-type HPr; lanes 3-9, GP plus the specified mutant HPrs. After electrophoresis, a xerographic copy of the stained gel (magnification: ϫ 4) was used for measurement of gel shifts. A reference line was drawn to connect the bands in lanes 1 and 10 (glycogen phosphorylase controls). The distance (in mm) was measured between the gel-shifted bands and the reference line, and from these values, a ratio was calculated for the distance the mutant HPr shifted GP in comparison with the shift produced by wild-type HPr.
Analytical Ultracentrifugation-Equilibrium ultracentrifugal analyses were performed using a Beckman XL-A analytical ultracentrifuge; a four-hole rotor at 32,000 rpm at 20°C was used for the characterization of HPr, and an eight-hole rotor at 6,300 rpm at the same temperature was used for all of the studies on GP. Carbon-filled epon double-sector centerpieces were used with filling volumes of 180 l, giving column heights of approximately 5 mm. For each sample, three solutions with an approximately 1:2:3 initial concentration ratio (in the range of 100 ng of GP) were used so that the resultant data could be fit globally to optimize constraints on the values of the fitting parameters. The experiments were performed using 25 mM Tris⅐HCl, pH 7.5, with 100 mM NaCl and 1 mM dithiothreitol. Buffer densities were calculated using the tabular values of Laue et al. (26).
Each experiment was run until the concentration gradients were invariant within the limits of experimental detectability for scans taken 24 h apart. Run times varied from 66 to 92 h. Sixteen replicate measurements were made at each radial position with radial increments of 0.02 mm. The data were edited and compared using Beckman software. Data analysis was performed using MLAB (Civilized Software, Bethesda, MD) for simultaneous weighted nonlinear least squares curve fitting of the three data sets from each experiment. Partial specific volumes were calculated from amino acid sequence data using the consensus values of Perkins (27).
The glycogen phosphorylase experiments were performed in pairs: E. coli GP with and without HPr; E. coli GP with two different mutant HPrs (K40A and K72E); and rabbit muscle GP with and without HPr.
To ensure an optimal effect on GP resulting from its presence, HPr was used at a 10-fold greater molar concentration than GP. This was practical because HPr at these concentrations does not have significant absorbance at the wavelength of 280 nm used to measure the GP concentration distributions. It would have been rather difficult to measure the binding of HPr to GP in the presence of the self-association reaction (28), since HPr has a molecular mass less than one-tenth that of GP (9119 g mol Ϫ1 versus 93,144 g mol Ϫ1 ) and because GP undergoes reversible self-association as well as possibly binding HPr. Thus, the studies reported here describe the effects of HPr on the self-association of glycogen phosphorylase under conditions that should approximate near saturation of the HPr binding site or sites on GP.

Immobilized HPr Complexes with a Factor in E. coli Extracts-
The technique of ligand fishing by surface plasmon resonance (29) was used to search for proteins that interact with and might be regulated by HPr. SPR detects a change in refractive index resulting from the interaction of a soluble protein with another protein adsorbed to a surface. HPr, immobilized on a CM-5 sensor chip (see "Experimental Procedures"), was used as the bait. When crude extract (600 l of 100 g/ml protein) from a stationary phase culture of E. coli strain IF-1 (sensorgram a of Fig. 2) was allowed to flow over the immobilized HPr for 10 min, there was a detectable increase in SPR response. Since the PTS activity of HPr involves phospho- Real time interaction analysis of proteins with immobilized HPr. Crude extract of E. coli IF-1 (100 g protein/ml) (sensorgram a), purified protein factor (8 g/ml) (sensorgram b), and crude extracts (10 g protein/ml) of E. coli strain SB221/pJF02 before (sensorgram c) and 4 h after (sensorgram d) expression of glycogen phosphorylase by the addition of 1 mM isopropyl-1-thio-␤-D-galactopyranoside (dissolved in the standard running buffer) were sequentially allowed to flow over the HPr surface to test for binding as described under "Experimental Procedures." The arrows indicate the starting points of the injections.

TABLE I Purification of E. coli glycogen phosphorylase
The total activity, total amount of protein, and specific activity were determined for the soluble steps of the purification scheme for GP. See "Experimental Procedures" for details of the purification and the GP assay. Chen and Segel (41) described the purification of GP and calculated that the activity of a purified preparation of the enzyme would be 1.2 units/mg. Their assays were performed in the direction of glycogen synthesis, at 37°C (expected to exhibit twice as much activity as the present assays carried out at 25°C), in the presence of 5Ј-AMP (expected to increase the activity by a factor of 2), and in the presence of added Na 2 SO 4 (expected to increase activity by a factor of 10). Correction for these modifications to the assay methodology should result in a specific activity of 0.03, approximately 5% of the activity shown in Table 1. Yu et al. (13) described the purification of GP after hyperexpression from a plasmid. Their assays were performed in the direction of glycogen synthesis, at 30°C, and in the presence of 5Ј-AMP. Correction for the differences in methodology of their reported specific activity (0.19 units/ mg) should result in a specific activity of 0.06, approximately 10% of the activity shown in Table I. Assuming that GP is equally active when assayed in either direction, the present preparation is 10 -20 times more active than previously reported. ryl acceptance from EI and the sequential phosphoryl transfer to enzymes IIA, it seemed possible that the detected interaction might be due to formation of HPr⅐EI or HPr⅐IIA Glc complexes. When purified EI or IIA Glc (600 l of 10 g/ml of each protein) was exposed to the HPr surface for 10 min in the BIAcore, no interaction was detectable (data not shown). Furthermore, in vitro phosphorylation of EI or IIA Glc by preincubation with P-enolpyruvate (for EI) and catalytic amounts of EI and HPr (for IIA Glc ) did not promote binding to immobilized HPr (data not shown). It is worth noting that the association constant between EI and HPr was recently reported to be about 10 5 M Ϫ1 by isothermal titration calorimetry (30). It is unlikely that an interaction with a K a in this range would be detected by SPR. Therefore, the observed interaction with crude extract suggested that there is another endogenous E. coli protein that binds to HPr, presumably with a higher affinity than either enzyme I or IIA Glc .
Purification of the Binding Factor-The approach to purification of the factor that interacted with HPr was based on SPR analysis; fractions from each purification step were examined for binding to immobilized HPr using the BIAcore. E. coli strain IF-1 was grown in 2 liters of Luria broth at 37°C to early stationary phase. The harvested cell pellet was washed and resuspended in buffer A (25 mM Tris⅐HCl, pH 7.5, containing 100 mM NaCl); this suspension was passed twice through a French press at 10,000 p.s.i. The lysate was ultracentrifuged at 100,000 ϫ g for 90 min to remove cell debris, and the supernatant solution was fractionated by ammonium sulfate precipitation. The fractions (30 -40% saturation) showing binding affinity toward immobilized HPr were combined, dialyzed against buffer A, and chromatographed through a DE-52 column (25 ϫ 300 mm) using a gradient of 100 -500 mM NaCl (1 liter). Fractions eluting around 0.4 M NaCl demonstrated interaction with immobilized HPr; these fractions were pooled and concentrated in a 10 K Macrosep centrifugal concentrator (Filtron). The concentrated pool was further purified on a Superose 12 column (1.6 ϫ 50 cm) using buffer A as an eluent. These fractions were analyzed, pooled, and concentrated as described for the DE-52 chromatography. At this step, the HPr-binding protein was estimated by SDS-PAGE to be about 100 kDa in size and approximately 50% pure. As a final purification step, a phenyl-Superose 10/10 column (Pharmacia) was used with 10 column volumes of ammonium sulfate gradient (1-0 M) in 25 mM Tris⅐HCl buffer, pH 7.5. The fractions at the end of the gradient, which showed interaction with HPr, were pooled and concentrated. The purity was about 90% as judged by 4 -20% SDS-PAGE (NOVEX).
This nearly homogeneous protein was tested for binding to different immobilized protein surfaces prepared as described under "Experimental Procedures." The protein factor (8 g/ml) was allowed to flow (rate of 10 l/min) over each surface for 10 min. The protein showed the expected binding to immobilized HPr (Fig. 2, sensorgram b) but not to immobilized EI or IIA Glc (data not shown). A blank surface (see "Experimental Procedures") also showed no interaction with the purified protein factor.
The HPr-binding Protein Is Glycogen Phosphorylase-The purified binding protein was subjected to proteolysis (trypsin or endoproteinase Lys-C) followed by sequencing analysis (see "Experimental Procedures"). The resultant sequences were processed by FASTA analysis to determine their identity to known sequences in the SWISS-PROT data base.
Selected fractions from the RP-HPLC separation of the trypsin digest were further purified by a second RP-HPLC step. Tryptic peptides, eluting at 72.2 and 74.6 min on the second column, were subjected to automated Edman degradation. The following amino acid sequences, respectively, were found: XY-VDCQDKVDELYELQEEWTAK and ESPDYXLEYGNPXEFK.
The observed sequences represented fragments expected from cleavage on the carboxyl side of basic residues (Arg or Lys; R or K). FASTA analysis revealed the sequences to be identical to residues 762-783 and 167-182 of E. coli GP (p13031 in the SWISS-PROT data base).
The RP-HPLC separation of the endoproteinase Lys-C digest (see "Experimental Procedures") yielded peaks that were mixtures of peptides. Upon amino acid sequencing, several of these peaks yielded multiple sequences that were later deduced to be identical to sequences in E. coli GP (data not shown). The correspondence of the sequences as well as the molecular weight of the HPr-binding protein to E. coli GP suggested strongly that the HPr-binding factor is GP.
Overproduction of E. coli Glycogen Phosphorylase Increases Binding of Crude Extract to Immobilized HPr-It was anticipated that, if the HPr-binding factor were GP, overexpression of that protein would result in enhanced HPr binding to crude extracts. Strain SB221 harboring pJF02, an expression vector for GP, was used for this purpose. Fig. 2 shows sensorgrams obtained by monitoring the flow of crude extracts (600 l of 10 g/ml of protein) from cells before induction (sensorgram c) and 4 h after induction with 1 mM IPTG (sensorgram d) (13). Crude extract from induced cells showed a much greater interaction with immobilized HPr; this data provides further support for the idea that the HPr-binding protein is glycogen phosphorylase.
The Interaction between HPr and Glycogen Phosphorylase Is Highly Specific-Since E. coli GP shares a high degree of identity (greater than 40%) in amino acid sequence with GP from other organisms and E. coli maltodextrin phosphorylase (17), it was of interest to test the binding of HPr to those proteins. Purified E. coli GP showed high affinity binding, but neither rabbit muscle phosphorylase a or b nor E. coli maltodextrin phosphorylase showed any affinity toward an HPr surface (data not shown for E. coli maltodextrin phosphorylase; see Figs. 4 and 5 for rabbit enzymes). A competition approach was taken to evaluate the specificity of the protein interaction with GP. Fig. 3 shows that if free E. coli HPr was included in the test solution with GP, the binding of GP to immobilized HPr was reduced. E. coli GP (2 g/ml) was preincubated with the following proteins (2 g/ml) in the standard running buffer: B. subtilis HPr (B-HPr) or M. capricolum HPr (M-HPr), E. coli EI (EI), E. coli IIA Glc (IIA Glc ), or E. coli HPr (HPr), and each mixture was allowed to flow (10 l/min) over the immobilized HPr surface. The response units (10-min monitoring period) were compared with the signal obtained when GP was tested alone. E. coli HPr competed with immobilized HPr for the binding with GP and gave about 60% competition compared with the control. Preincubation of HPr (10 g/ml) with 2 g/ml GP gave about 95% competition (Fig. 3). However, HPrs from other bacterial strains did not compete for GP binding. Neither EI nor IIA Glc at either 2 g/ml (Fig. 3) or 10 g/ml (data not shown) influenced the binding of GP to immobilized HPr, as expected from the results of direct binding to immobilized HPr. It should be noted that if GP bound to EI and/or IIA Glc at unique sites in addition to binding to HPr, then the response would have been enhanced. These data support the conclusion that the binding of E. coli GP is highly specific for E. coli HPr.
Phosphorylation of HPr Enhances Its Interaction with Glycogen Phosphorylase-Frequently, the phosphorylation state of a regulatory protein modulates the activity of its interacting partner (2). Accordingly, the effect of phosphorylating HPr on its affinity for GP was checked by a competition experiment (Fig. 3). When HPr (0.2 g/ml) was preincubated with GP under conditions where it became phosphorylated (see Fig. 3  legend), it gave about a 30% decrease in the signal for GP binding, while 2 g/ml of phosphorylated HPr gave a 97% decrease in binding. The competition with phosphorylated HPr was significantly greater than with unphosphorylated HPr. A comparison of the two competition curves suggests that phosphorylated HPr (P-HPr) has about 4 times higher affinity toward GP than dephospho-HPr. The observation that P-HPr competes with HPr for binding to GP indicates that P-HPr binds to the same or an overlapping site on GP as does HPr. When EI was phosphorylated by incubation with P-enolpyruvate and Mg 2ϩ , it had no effect on the binding of GP to immobilized HPr (data not shown), suggesting that immobilized HPr could not be phosphorylated.
Kinetic Analysis of the Glycogen Phosphorylase-HPr Interaction-Kinetic parameters for the binding of HPr to GP were determined using HPr immobilized to a sensor chip in the BIAcore system. Five different concentrations (1, 2, 4, 8, and 16 g/ml) of purified GP were used for the binding analysis (Fig. 4). The signal increased as a function of analyte (GP) concentration. As a control, rabbit muscle GP (8 g/ml) gave no response (see legend to Fig. 4). The dissociation constant (K D ) for the HPr⅐GP interaction, using the BIAevaluation 2.1 software, was determined to be approximately 1.7 ϫ 10 Ϫ8 M, assuming interaction of the monomeric forms of both GP and HPr.
Interaction between HPr and Glycogen Phosphorylase Results in an Electrophoretic Mobility Shift-The biomolecular interaction studies, using the BIAcore, indicated a high affinity interaction of HPr and GP. It was therefore anticipated that such a stable complex might survive under electrophoretic conditions. Consequently, gel shift experiments using nondenaturing polyacrylamide gel electrophoresis (Fig. 5)  To confirm the species specificity of phosphorylase binding to E. coli HPr, two different forms of rabbit muscle phosphorylase were run on the gel. Neither phosphoryl-ase a (lanes 3 and 4, Ϫ and ϩ HPr, respectively) nor b (lanes 5 and 6, Ϫ and ϩ HPr, respectively) showed an HPr-dependent gel shift. This is consistent with the finding that rabbit muscle GP shows no evidence for interaction with HPr by SPR (Fig. 4). Furthermore, EI or IIA Glc , regardless of their phosphorylation states, did not retard GP (data not shown). Interestingly, when GP was run with HPr under phosphorylating conditions (HPr was incubated with a catalytic amount of EI and 1 mM P-enolpyruvate, data not shown), the band corresponding to the GP⅐HPr complex was smeared between the position characteristic of the complex and that of free GP. The demonstration of a species-specific HPr-dependent gel mobility shift of GP is in full agreement with the data obtained using SPR.
The Oligomeric State of Glycogen Phosphorylase Is Modulated by HPr-Sedimentation equilibrium experiments were carried out to evaluate the possibility that HPr binding to GP is associated with a structural change. The results of the various ultracentrifugal experiments are summarized in Table II. From the first pair of experiments, it is apparent that the binding of HPr to E. coli GP enhances monomer-dimer association almost 4-fold and dimer-tetramer association 66-fold. Mutant HPrs, defective in interaction with GP (Table III), show significantly different associative behavior with GP. These mutant forms of HPr do appear to be bound on the basis of a significantly better sum of squares for the fit of the data to the binding model when compared with the sum of squares for the fit to the nonbinding model. For the K40A mutant, the mono- FIG. 4. Concentration dependence for the interaction between E. coli GP and immobilized HPr. The indicated concentrations of GP (in g/ml), in standard running buffer, were allowed to flow over an immobilized HPr surface for the determination of kinetic parameters for the interaction between GP and HPr. As a control, 8 g/ml of rabbit muscle glycogen phosphorylase (RM) was exposed to the surface. The association constant was calculated using the BIAevaluation software. The arrow indicates the starting points of the injections. mer-dimer association is about 500-fold weaker, while the dimer-tetramer association is 2-fold stronger than for wild-type HPr. For the K72E mutant, the association for dimer formation is over 300-fold weaker, while that for tetramer formation is virtually the same as for wild-type HPr.
In contrast to E. coli GP, rabbit muscle GP exhibits a differ-ent mode of association, undergoing a monomer-tetramer-octamer association in both the absence and presence of HPr. It was not possible to determine whether or not HPr was bound on the basis of the fitting statistics. However, the effect of the presence of HPr on the associative behavior of the mammalian enzyme in Tris buffer is quite minimal; at best about a 2-fold

TABLE III Interaction of mutated HPrs with glycogen phosphorylase
Wild-type and mutant HPrs were examined for binding of HPr to glycogen phosphorylase (gel shift analysis) and activation of GP (GP activation assays). Gel shift results are recorded as the ratio of the distance that a mutant HPr shifted GP in comparison to the shift observed with wild-type HPr (see "Experimental Procedures"). A statistical analysis (using Origin, version 4.0, Microcal) showed a mean gel shift by wild-type HPr of 3.0 Ϯ 0.21 mm (S.D.) (n ϭ 9). The gel shift data for those mutations exhibiting a ratio of mutant to wild-type of Յ0.3 are highlighted in boldface type. GP activation data represents the ratio of the stimulation of GP activity by a mutant HPr compared to the activation by wild-type HPr (see "Experimental Procedures"). Numbers in parentheses correspond to the activation ratio at a lower concentration of HPr (1 g/ml) than that used in the standard assay (10 g/ml) (see Fig. 7). A statistical analysis (using Origin, version 4.0, Microcal) of the activation of GP by wild-type HPr showed a mean activation ratio (ϩHPr/ϪHPr) of 2.5 Ϯ 0.26 (S.D.) (n ϭ 18). The activation data for those mutations exhibiting a large decrease (Ն50%) in GP activation are highlighted in boldface type. . For the construction of the HPr(H15E) expression vector, pSP200, the coding sequence of E. coli HPr (nucleotides 205-486 of the ptsH gene) was amplified by polymerase chain reaction using pDS20 (18) as a template. The forward mutagenic primer 5Ј-GGAAATCATATGTTCCAGCAAGAAGT-TACCATTACCGCTCCGAACGGTCTCGAGACCCGC-3Ј contained engineered NdeI and XhoI restriction sites (underlined) and a mutagenic site (bold) to change His-15 to Glu, and the reverse primer 5Ј-AAAGAACCCGGGTTATTACTCGAGTTC-3Ј contained an engineered stop codon (in bold) in addition to a pre-existing stop codon (in italics) and SmaI and XhoI restriction sites (underlined). The NdeI-and SmaI-cut polymerase chain reaction product was gel purified and ligated into the corresponding sites of the vector pRE2 (18). E. coli strain GI698 was transformed with the recombinant plasmid as previously described (23). Strain GI698/pSP200 was grown in defined medium, and HPr(H15E) expression was induced with tryptophan (23) and HPr(H15E) was purified as described (19). decrease in monomer-tetramer association in the presence of HPr and even less of an effect on the tetramer-octamer association is observed (Table II). These data suggest that HPr is probably not bound to the mammalian enzyme.
The differences between E. coli and rabbit muscle GP with respect to the interaction with HPr are graphically represented in Fig. 6, which presents the mole fractions of monomer and the various oligomers as a function of total concentration (expressed as mol of monomer). These graphs clearly demonstrate the difference in associative properties of E. coli (panels A and B, Ϫ and ϩ HPr, respectively) and rabbit muscle (panels C and D, Ϫ and ϩ HPr, respectively) GP and demonstrate the significant effect of HPr only on the bacterial enzyme.
E. coli HPr Specifically Activates E. coli GP-A phosphoglucomutase-and glucose-6-phosphate dehydrogenase-coupled enzyme assay (see "Experimental Procedures") was used to determine the effect of HPr on GP activity. This assay measures the activity in the physiological direction (glycogen degradation). The glucose 1-phosphate produced by GP promotes the conversion by the coupling enzymes of NADP to NADPH (detected by the change in absorbance at 340 nm).
The activity of E. coli GP was substantially increased by E. coli HPr (to 250% of the control value) (Fig. 7). The stimulation of GP activity elicited by E. coli HPr was not observed with HPrs from B. subtilis or M. capricolum (data not shown). Other purified PTS proteins from E. coli (EI and IIA Glc ) also had no effect on the GP activity. Further, phosphorylation of EI by P-enolpyruvate (1 mM) did not affect GP activity (data not shown). None of the E. coli PTS proteins, HPr, IIA Glc , EI, or P-EI had a significant effect on the activity of rabbit muscle GP a (active form) or b (inactive form) (data not shown). Therefore, E. coli HPr specifically activates E. coli GP.
HPr Modifies the Kinetic Behavior of GP-The stimulation of GP activity by HPr was saturable (Fig. 8A); 1 g/ml of HPr gave 50% of maximal activation, while maximal activation was observed at 5 g/ml.
The effect of HPr interaction with GP on the kinetic param-eters was evaluated. When the concentration of glycogen was varied up to a concentration of 1% (Fig. 8B), there was no indication, as previously observed (13), that GP activity was approaching saturation. HPr (10 g/ml) appeared to increase the affinity for glycogen. Consequently, the relative activation of GP by HPr was greater at lower glycogen concentrations. A reciprocal plot (Fig. 8B, inset) indicated that HPr decreased the K m for glycogen about 5-fold. The concentration of phosphate was varied up to 80 mM (Fig.  8C). The response pattern was different from that observed for variation of glycogen concentration. In this case, HPr addition (10 g/ml) resulted in a similar stimulation of the activity (to approximately 250% of the control) at all phosphate concentrations studied. This indicates that HPr affects the V max for phosphate. A reciprocal plot (Fig. 8C, inset) indicated that HPr increased V max (ϳ5-fold).
Phosphorylation of HPr Abolishes Its Stimulatory Activity-P-enolpyruvate (1 mM), EI (2 g/ml), and HPr were added to the coupled enzyme assay to determine if phosphorylation of HPr would affect its stimulatory activity. This reaction mixture was preincubated for 10 min before the assay was initiated by the addition of GP. The assay mixture containing P-HPr exhibited approximately 20% as much stimulation of GP activity as dephospho-HPr (Fig. 8A). The addition of IIA Glc (1 g/ml) to the incubation mixture containing P-HPr (resulting in the accumulation of P-IIA Glc ) did not change the degree of stimulation (data not shown). In the experiment shown in Fig. 8B, P-HPr addition resulted in no stimulation of GP activity at any concentration of glycogen tested, while the study in Fig. 8C showed P-HPr to have approximately 50% as much stimulation as dephospho-HPr at all concentrations of phosphate tested. Since it is difficult to reproducibly generate a sample of P-HPr totally devoid of the dephospho-form, it is likely that the variability in apparent activation by P-HPr is a reflection of the degree of contamination with the dephospho-form and that P-HPr itself does not activate GP.
Mutation of HPr Abolishes Its Regulatory Activity-To deter-FIG. 6. Mole fractions of monomeric and oligomeric species as a function of total concentration for E. coli and rabbit muscle glycogen phosphorylase in the absence and presence of HPr. The total concentration is represented as mol of monomer and presented as the base Ϫ10 logarithm of this concentration in all four graphs. Graphs A and B are for the E. coli GP in the absence and presence of HPr, respectively; graphs C and D are for the rabbit muscle GP in the absence and presence of HPr, respectively. mine which amino acids in HPr are important for the binding and activation of GP, 51 mutant HPrs were examined by both gel shift and GP activity analysis. The gel shift results (Table  III) are presented as the ratio of the distance that a mutant HPr shifted the GP band relative to the distance that wild-type HPr shifted the GP band. The GP activation data are expressed as the ratio of the increase in GP activity due to a mutant HPr relative to the activity increase observed with the wild-type HPr. GP activation by HPrs was tested at two levels: 1 g/ml, a level that leads to 50% of maximal activation by wild-type HPr (data not shown); and 10 g/ml, a level that results in maximal activation of GP ( Fig. 8A and Table III).
Several of the mutant HPr proteins (R17G, R17E, K24Q, K24E, K27E, K27S, K40A, S46D, and Q51E), shown in boldface letters in Table III, produced gel shift ratios of Յ0.3. In the case of Arg-17, mutation to Lys maintained the wild-type pattern (gel shift ratio ϭ 0.9), indicating that a positive charge in that region is important for production of a complex with properties similar to that of wild-type HPr. The differences in the gel shift behavior of the three mutations at Arg-17 were paralleled by their capabilities to activate GP. The R17K protein elicited both a greater gel shift response and a greater activation of GP than did either the R17G or the R17E mutant HPr.
Mutation of HPr at Lys-24 to either Gln or Glu essentially abolished the capability of the HPr to produce a gel shift. Further, there was a drastic reduction in the capacity of these mutated HPrs to promote activation of GP. The K24E mutant was essentially inactive, and the K24Q mutant required a concentration of 10 g/ml to produce approximately 60% as much activation response as wild-type HPr. Using a concentration of HPr of 1 g/ml, the K24Q mutant HPr was only 10% as effective as wild-type HPr (see numbers in parentheses, Table  III), suggesting that the affinity of this mutant protein for GP is considerably reduced. It appears from this analysis that Lys-24 is very important in the binding to and activation of GP.
Two mutations at Lys-27 (K27E and K27S) were examined; both exhibited a similar substantial decrease in the gel shift ratio (0.3). Nevertheless, at concentrations of HPr (40 g/ml) and GP (200 g/ml) corresponding to a 2:1 ratio of HPr to GP, there was a single, distinct complex of HPr and GP observed on the gel. The significantly different location of the complex on the gel implies that these mutant proteins form complexes with GP that are structurally different than the GP⅐wild-type HPr complex. The behavior of these two mutated proteins with respect to GP activation was similar to that observed with the R17G mutation, suggesting an important role for Lys-27 of HPr in the activation of GP.
Structural changes in HPr at Lys-40 (K40A and K40F) resulted in pronounced decreases in gel shift and GP activation properties. The GP complexes with these mutated proteins migrated to positions similar to that observed with the K27E and K27S proteins. Proteins mutated at Lys-40 were less effective in GP activation than either the R17G or K24Q mutants when tested at 1 g/ml (shown in parentheses), but they produced similar activation responses to these mutants when tested at 10 g/ml. Apparently, the K40A and K40F mutant proteins are severely impaired with respect to interaction with GP. Sedimentation equilibrium analysis (Table II) suggested that the K40A mutant protein does complex with GP.
A number of HPrs carrying mutations at Ser-46 were evaluated for gel shifts and activity stimulations with GP. Mutation of Ser-46 to Ala or Asn was associated with wild-type behavior in the gel shift assay. While the S46A mutant protein showed a modest (approximately 20%) decrease in GP activation, the S46N protein was considerably less potent in GP activation (activation ratio ϭ 0.3). Similar patterns of GP activation were observed with S46N, S46E, and S46R mutant proteins; however, the S46R protein produced a complex that shifted in gels more than did the wild-type HPr⅐GP complex, and the S46E HPr⅐GP complex shifted only 60% as much as the control. The S46D HPr was unique in that it exhibited only a small gel shift (30% of the wild-type shift) and only a slight activation of GP. These assays indicate that Ser-46 plays an important role in the activation of GP and that, depending on the mutation, a variety of complexes with different properties in the gel shift assay can form.
Two mutant forms of HPr with changes in Gln-51 behaved differently in the two test systems. When the Gln residue was changed to Thr, the HPr behaved exactly as did the wild-type HPr. However, when the change was to Glu, the HPr⅐GP com- FIG. 7. Activation of GP by HPr. The effects of HPr, EI, and IIA Glc (all at 10 g/ml) on E. coli glycogen phosphorylase activity were determined using the coupled enzyme assay described under "Experimental Procedures." The control is the activity observed with no protein additions. plex shifted only 20% as much as did the complex with wildtype HPr. The effects of the mutation to Glu on GP activation were also substantial. Activation ratios of 10% at 1 g/ml HPr and 50% at 10 g/ml HPr were observed. This was a behavior similar to that of some of the other mutant proteins such as K40F, where the affinity for GP appears to be considerably decreased by the mutation. These determinations suggested that, while Gln (a polar residue) at position 51 of HPr is not essential for appropriate interaction with GP, substitution with negatively charged residues, such as Glu or Asp, leads to important changes in the nature of the interaction with GP.
The Lys residue at position 72 of HPr shows a unique behavior upon mutagenesis. The gel shift is only slightly decreased (10 -20%) as a result of replacement by Arg or Glu. However, there is a substantial difference in the behavior of the two mutants in the GP activation test. The K72R mutant shows a decrease of 20% in GP activation, but the K72E mutant is essentially incapable of activating GP. Sedimentation equilibrium analysis of the K72E mutant protein (Table II) suggested that it does interact with GP but does not have as pronounced an effect on the association properties of GP as does the wild-type HPr. Consequently, it appears that the presence of a basic residue at position 72 is important for effective activation of GP.
Interestingly, the C-terminal Glu (residue 85) of HPr appears to play an inhibitory role in the activation of GP. Replacement of this residue by Lys results in a 50% increase in the HPr-dependent stimulation. The resultant complex also shifts in a gel 1.4 times as great a distance as does the wild-type complex. DISCUSSION While there was no history suggesting a role for HPr in metabolic regulation in enteric bacteria, we had a theoretical rationale to explore this possibility. HPr is both a phosphocarrier and a regulator in Gram-positive bacteria. Further, the cellular concentration of HPr in E. coli is significantly higher than necessary for its role as a phosphocarrier. The small size of HPr (less than 10 kDa) makes it an ideal potential regulatory factor. This perspective stimulated us to embark on ligand fishing experiments aimed at detecting a protein(s) exhibiting high affinity binding to HPr. This search turned up a protein with the expected tight binding; after purification to near homogeneity, sequencing experiments identified glycogen phosphorylase as the target of the HPr interaction.
The interaction between GP and HPr is highly specific; only E. coli HPr, not HPrs from Gram-positive bacteria or mycoplasma, interacts with (Fig. 3) and activates (Fig. 7) E. coli GP. A comparison of HPr primary sequences (2,31,32) shows that enteric HPrs are very similar to each other (those from E. coli and Salmonella typhimurium are identical); however, they differ considerably from those of Gram-positive bacteria and mycoplasma except around the site of phosphorylation (His-15). The three-dimensional structures of HPrs from mycoplasma (32), Gram-negative (7,8), and Gram-positive bacteria (33) have been determined; while the folding topology in all cases is an open-face ␤-sandwich, there are significant differences in the three-dimensional structures of various HPrs (34). Consequently, it is not surprising that only E. coli HPr interacts with E. coli glycogen phosphorylase.
In addition to HPr, E. coli synthesizes two other phosphocarrier proteins that accept a phosphoryl group from EI. These are the low molecular weight NPr (35,36) and DTP; DTP combines a pseudo-HPr activity with that of IIA Fru (37,38). Since the NPr sequence is most homologous to that of the HPr-like protein of (0.00346 units), whereas the data in B and C are presented as activities in units.  1-1%) in the presence and absence of HPr or P-HPr (10 g/ml) (B), and the amount of phosphate (2.5-80 mM) in the presence and absence of HPr or P-HPr (10 g/ml) (C) was varied in the coupled enzyme assay mixture to determine their effects on GP activity. The activity in A is expressed as the increase over the basal activity Alcaligenes eutrophus (39), which has been proposed to play a role in the regulation of poly-␤-hydroxybutyrate metabolism (40), the possibility was considered that NPr might function in an analagous manner in E. coli by regulating GP activity. This was ruled out. While E. coli HPr stimulates GP activity, neither NPr nor DTP has any effect on E. coli GP activity. 3 To learn the importance of individual amino acid residues in HPr involved in the binding and activation of GP, 51 mutant HPrs were examined. Those residues where mutation resulted in a significant change in the gel shift of the HPr⅐GP complex and resulted in a stimulation of GP activity of less than 50% compared with wild-type HPr included Arg-17, Lys-24, Lys-27, Lys-40, Ser-46, Gln-51, and Lys-72. The x-ray crystallographic structure (8) shows that these residues are on the surface of HPr and distributed over a large area (Fig. 9). Five of the seven residues have a positive charge, while the other two amino acids are polar, suggesting that electrostatic interactions are important for assembly of the normal, active HPr⅐GP complex. It is noteworthy that Ser-46 is the site of a regulatory phosphorylation in Gram-positive bacteria; thus, Ser-46 is important in regulatory functions in both Gram-positive and Gramnegative bacteria. Clearly, x-ray crystallographic studies of the GP⅐HPr complex and mutagenic studies on GP will enhance our understanding of the details of the interaction of GP with HPr.
Several lines of evidence indicate that the E. coli HPr-dependent allosteric regulation of GP is specific for the E. coli enzyme. Although eucaryotic GPs exhibit a high degree of sequence identity to E. coli GP (17), rabbit muscle GP does not interact with E. coli HPr as evidenced by SPR (Fig. 4), gel shift analysis (Fig. 5), or sedimentation equilibrium studies (Fig. 6); additionally, rabbit muscle GP activity is not stimulated by HPr.
Both GP and maltodextrin phosphorylase are involved in the breakdown of glycogen in E. coli. Since there is extensive sequence conservation in these two proteins from E. coli (17), it was important to determine whether HPr interacted with maltodextrin phosphorylase as well as glycogen phosphorylase. The absence of HPr binding to maltodextrin phosphorylase emphasizes the high degree of specificity of this protein-protein interaction.
As a consequence of hyperexpression of glgP, we were able to develop a simple purification scheme for E. coli GP. The specific activity of the purified protein was greater than that previously reported (13, 41) (see Table I) but still considerably lower than that of rabbit muscle GP. We have found that this purified GP preparation is allosterically activated by the phosphocarrier protein HPr. The binding of HPr to GP increases the activity to approximately 250% of the basal level. Double reciprocal plots show that HPr-dependent activation leads to an increase (approximately 5-fold) in the affinity for glycogen. When the concentration of phosphate is varied, HPr promotes an approximately 5-fold increase in the V max . Sedimentation equilibrium studies show that HPr changes the state of oligomerization of GP.
The cellular concentration of HPr in E. coli is in great excess over that of GP (41,42). The phosphocarrier function of HPr in the PTS allows the protein to exist in both phospho-and dephospho-forms. Substantial activation of GP by HPr requires that it be in the dephospho-form (Fig. 8A). Other studies (Fig.  3) indicated that P-HPr has a higher affinity for GP than does dephospho-HPr. It is therefore reasonable to assume that, in vivo, E. coli GP is always bound to HPr and that physiological 3 M. Sondej and A. Peterkofsky, unpublished results. perturbations that shift the ratio of HPr to P-HPr should lead to changes in the activity of GP. Some mutations in HPr that involve the change of a positively charged side chain to one with a negative charge (R17E, K24E, K27E, K72E) result in a decrease in the capability of the HPr to activate GP (Table III). Perhaps the conversion of dephospho-HPr to phospho-HPr reduces the potential for GP activation by a similar mechanism, presumably not favoring the optimally active conformation. It is worth noting that the residues Lys-24, Lys-27, and Ser-46 of HPr appear to be essential both for activation of GP and for the previously studied (4) interaction of HPr with GTP. These three residues occupy a limited surface region on HPr (see Fig. 9).
As is the case in higher organisms, bacterial glycogen probably serves as a reservoir of energy and carbon. The mechanism that regulates the activity of E. coli GP differs from that controlling rabbit muscle and yeast GP. Rabbit muscle GP b (inactive form) is converted to an active form (GP a) by either binding of 5Ј-AMP or covalent phosphorylation at Ser-14 by phosphorylase kinase (43). The N-terminal region of yeast GP is 39 amino acid residues longer than the rabbit muscle enzyme and lacks a serine at residue 14 (17). Instead, the phosphorylation occurs in this extended region at Thr-30 by either a phosphorylase kinase or a cAMP-dependent protein kinase, neither of which recognize the rabbit muscle enzyme (44,45). The N-terminal region of E. coli GP is 10 amino acid residues shorter than the rabbit muscle enzyme and, like yeast GP, lacks the Ser-14 phosphorylation site (17). Until now, it was believed that E. coli GP is subject to only minimal activity regulation; in contrast to the previously reported 30 -40% increase in E. coli GP activity by 2 mM 5Ј-AMP (13,41), 5Ј-AMP only produced an approximately 10% increase in GP activity in our preparation. The studies presented here establish that the high affinity interaction of E. coli GP with HPr is the major basis for activity regulation.
The genes encoding enzymes involved in glycogen metabolism (glycogen synthase and glycogen phosphorylase) are linked on a single operon that is regulated by the cAMP⅐CRP complex. Consistent with the concept that cAMP levels rise in stationary phase (46), it has been reported that glg operon expression is increased in stationary phase (47). The coincident preferential accumulation of glycogen in stationary phase (14) may reflect the preponderance of HPr in the phospho-form. When stationary phase cells enter a new round of growth, a number of metabolic changes resulting in the preferential degradation of glycogen might occur. The decrease in cellular cAMP level characteristic of the entry into logarithmic growth phase (46) would turn off the expression of the glg operon, and glycogen synthase activity might preferentially decay. The uptake of a PTS sugar should promote a shift in the state of HPr in the direction of dephospho-HPr, resulting in an activation of GP (Fig. 10). Consequently, effective recovery from the stationary phase of growth may be enhanced by the energy derived from the utilization of stored glycogen (48).
In summary, the novel finding that the activity of E. coli GP is regulated by the phosphocarrier protein, HPr, has been described. The regulation of GP activity by the state of phosphorylation of HPr is proposed to be the bacterial analogy to the covalent phosphorylation-dephosphorylation cascade characteristic of eucaryotic GPs.