The Aspartyl Replacement of the Active Site Histidine in Histidine-containing Protein, HPr, of the Escherichia coliPhosphoenolpyruvate:Sugar Phosphotransferase System Can Accept and Donate a Phosphoryl Group

The active site residue, His15, in histidine-containing protein, HPr, can be replaced by aspartate and still act as a phosphoacceptor and phosphodonor with enzyme I and enzyme IIAglucose, respectively. Other substitutions, including cysteine, glutamate, serine, threonine, and tyrosine, failed to show any activity. Enzyme I K m for His15 → Asp HPr is increased 10-fold andV max is decreased 1000-fold compared with wild type HPr. The phosphorylation of Asp15 led to a spontaneous internal rearrangement involving the loss of the phosphoryl group and a water molecule, which was confirmed by mass spectrometry. The protein species formed had a higher pI than His15 → Asp HPr, which could arise from the formation of a succinimide or an isoimide. Hydrolysis of the isolated high pI form gave only aspartic acid at residue 15, and no isoaspartic acid was detected. This indicates that an isoimide rather than a succinimide is formed. In the absence of phosphorylation, no formation of the high pI form could be found, indicating that phosphorylation catalyzed the formation of the cyclization. The possible involvement of Asn12 in an internal cyclization with Asp15 was eliminated by the Asn12 → Ala mutation in His15 → AspHPr. Asn12 substitutions of alanine, aspartate, serine, and threonine in wild type HPr indicated a general requirement for residues capable of forming a hydrogen bond with the Nε2 atom of His15, but elimination of the hydrogen bond has only a 4-fold decrease ink cat/K m .

the function of accepting and donating a phosphoryl group from enzyme I to the sugar-specific enzymes II was established for Escherichia coli. These events result in phosphoprotein formation with a N⑀ 2 -P-histidine in both enzyme I (6) and the enzyme IIA sugar domains (7) and a more unstable N ␦1 -P-histidine in HPr (8,9). The structure of HPr from a number of species is now well established from both x-ray diffraction and NMR spectrometry approaches. The overall structure of the HPrs is described as an open-faced ␤-sandwich with a ␤␣␤␤␣␤␣ fold (see reviews in Refs. 10 -12). In order to help specify phosphoryl transfer to the N ␦1 atom of the His 15 imidazole ring, various residues have been proposed to form hydrogen bonds with the N⑀ 2 atom of the His 15 imidazole. In E. coli, the involvement of a glutamate residue was suggested (9) and was identified as Glu 85 by the first 1 H NMR structure (13) and the 2.0 Å resolution x-ray structure (14), the latter showing that the interaction was with the C-terminal ␣-carboxylate. Replacement or deletion of Glu 85 did not indicate a significant role for this residue (15). NMR spectral properties of His 15 in E. coli HPr were consistent with hydrogen bonding to the N⑀ 2 atom (16), and van Nuland et al. (17) suggested Asn 12 as the most likely residue. Subsequently, in the 2.5 Å resolution structure of the complex of the Jel42 monoclonal antibody Fab fragment with HPr, the Asn 12 side chain was found hydrogen-bonded to the N⑀ 2 atom of His 15 (18).
Asn 12 in E. coli HPr has been investigated because it is a site of deamidation, which occurs through the formation of a succinimide to form aspartate and isoaspartate at residue 12 (19). The succinimide formation (Fig. 1A), especially at Asn-Gly pairs in a sequence leads to about 70% isoaspartic acid formation (20 -23). This unusual amino acid can be repaired to aspartic acid by protein carboxylmethyltransferase (L-isoaspartate-(D-aspartate) O-methyltransferase) in peptides (24), and the effective repair of residue 12 isoaspartic acid in HPr has been described (25). The substitution of aspartate or isoaspartate at residue 12 has modest effects on the phosphocarrier role of HPr (19).
The formation of succinimides can occur at aspartate residues but usually under different conditions than those that prevail for asparagine residues. A stable succinimide has been characterized in somatotropin where cyclization of an aspartyl residue under acidic conditions allows the isolation of the succinimide that is labile at alkaline pH (26). A structure of a succinimide formed by an aspartyl residue at acidic pH has recently been described in lysozyme (27). In both cases, the hydrolysis of the succinimide yielded both isoaspartic and aspartic acid. The mechanism of succinimide formation is not the only route by which deamidation can occur (20 -23), and among the possibilities is the formation of an isoimide shown in Fig. 1B.
In this paper, we describe the effects of other substitutions at residue 12 and the lack of an absolute requirement for a histidine at residue 15 of HPr. Aspartate substitutes for histidine at residue 15 in phosphoryl acceptance and transfer, albeit inefficiently. In addition, the P-aspartyl residue leads to a spontaneous chemical rearrangement with the characteristics of catalyzed isoimide formation.

EXPERIMENTAL PROCEDURES
Materials-Enzyme I, the enzymes II sugar , and DNA oligonucleotides were obtained as described previously (28). Ampholytes were from Amersham Pharmacia Biotech. DEAE paper (D81) was from Whatman. The Quik-Change site-directed mutagenesis kit was obtained from Stratagene. [ 32 P]Phosphoenolpyruvate (PEP) was produced as described previously (29).
Mutagenesis-His 15 3 Asp and His 15 3 Glu HPr mutants were produced by Dr. J. W. Anderson. Asn 12 3 Asp HPr has been described (19), and all other mutations were produced by the Quik-Change sitedirected mutagenesis kit according to the manufacturer's instructions. All mutations were in the ptsH gene incorporated into pUC19 (15). His 15 3 Ala HPr was obtained from M Scholtz (Texas A & M).
Protein Expression and Purification-HPr and mutant HPrs were expressed in E. coli strain ESK108, which is ptsH (28), using the pUC(ptsH) plasmids with HPr expression under the control of its own promoter. Homogeneous protein was produced as described previously (15). Yields were 50 -500 mg of protein/30 g of wet weight of cells.
Isolation of His 15 3 Asp HPr Derivatives-His 15 3 Asp HPr (3 mg) was phosphorylated at 37°C for 10 min in 10 mM potassium phosphate buffer, pH 7.0, with 5 mM PEP and 0.1 mg enzyme I. The three forms of His 15 3 Asp HPr, phosphorylated (lower pI), unphosphorylated, and cyclized (higher pI) were separated by anion exchange chromatography, Mini-Q-Sepharose column, and a Amersham Pharmacia Biotech Gradifrac system at 4°C. The reaction mixture was loaded with 10 mM citrate-phosphate buffer, pH 4.6, and the column was eluted at 2 ml/min with a 20-ml gradient to 0.07 M NaCl in the same buffer. Fractions (0.5 ml) were collected, and protein elution was monitored at 214 nm.
N-terminal Sequencing-Protein sequencing was performed using an Applied Biosystems, Inc. model 471A sequencer equipped with a model MG5 microgradient pump and a blot cartridge for polyvinylidene difluoride-type membranes. Data were acquired and analyzed using an Applied Biosystems Inc. model 601A data system (30). The sequencing was carried out by Dr. S. Mackenzie (Plant Biotechnology Institute, National Research Council of Canada, Saskatoon, Saskatchewan, Canada).
Mass Spectrometer Analysis-Mass spectrometry was performed using a Perseptive Biosystems Voyager ELITE matrix-assisted laser di-FIG. 1. Mechanisms of formation of succinimide and isoimide rings. A, mechanism of succinimide formation according to Geiger and Clarke (21). Not shown is the formation of small amounts of D-aspartyl and D-isoaspartyl that also occurs. B, mechanism of isoimide formation (23). C, proposed mechanism for isoimide formation by P-Asp 15 in HPr involving hydrogen bonding with the amide nitrogen of residue 16. ode ionization-time of flight spectrometer at the Plant Biotechnology Institute. The samples were run in linear mode. 15 3 Asp HPr were grown by the hanging drop vapor diffusion method at 14°C. Washing and seeding of microcrystals was used (31). Crystals formed in 0.1 M citrate phosphate buffer, pH 4.4, and 20 -25% saturated ammonium sulfate. Crystals of the high pI form of His 15 3 Asp HPr, containing a putative cyclized Asp 15 were grown similarly. Synchrotron diffraction data for His 15 3 Asp HPr were collected with a Brandeis CCD detector at the Brookhaven National Laboratory (Upton, NY). For the high pI form, data were collected at the Photon Factory (Tsukuba, Japan) using a wavelength of 1.0 Å and a screenless Weissenberg camera. The data were processed using DENZO and SCALEPACK (32). The structures were solved as has previously been described for Ser 46 3 Asp HPr (33) using the Amore suite of programs (34) with molecular replacement with wild type HPr (11). Refinement was performed using the X-PLOR 3.1 package (35).

Crystallization of HPr Mutants and Determination of the Tertiary Structures of Mutant HPrs-Crystals of His
Construction of ptsH Strains and Enzyme IIA glc Production-The gene for enzyme IIA glc , crr, was isolated by polymerase chain reaction from pTSHIC9 (36) and introduced into pT7-7 (37) using the NdeI and BamHI restriction endonuclease sites. Enzyme II glc was expressed in E. coli strain ESK262, which is Kan R ::ptsH, following mid-log phase induction by 0.5 mM isopropylthiogalactoside. This strain was constructed by ligating the Kan R gene from pUC4 into the PstI restriction endonuclease site in ptsH in pAB65 (38). The linearized plasmid was used to transform E. coli strain DPB271 (39), which is recD, and a ptsH gene replacement derivative was selected by kanamycin resistance. This E. coli strain ESK150 had no HPr detectable by assay, phosphorylation, or immunoreactivity as determined by standard methods (40). E. coli strain was a derivative of strain BL21 plysS (37), which was transduced with P1-phage grown on strain ESK150 to produce a Kan R ::ptsH strain ESK262. Enzyme I was also overproduced in this strain using a similar plasmid. 2 Protein Methylation-His 15 3 Asp HPr and derivatives were assayed for methyl accepting ability by incubation with S-adenosyl-L-[methyl- 14 C]methionine and L-isoaspartate-(D-aspartate) O-methyltransferase. Wild type HPr was used as a control. The assays were performed by J. D. Lowenson and S. Clarke (UCLA) as described previously (25).
Other Methods-Standard methods have been described for characterization of HPr:protein determinations (15,28), isoelectric focusing, (19), SDS-polyacrylamide gel electrophoresis and autoradiography (42), rates of phosphohydrolysis (9), and enzyme I and enzyme IIA sugar assays (15,28,43). 15 3 Asp HPr Can Act as a Phosphoacceptor-The following substitutions for His 15 in E. coli HPr were made: alanine, asparagine, aspartate, cysteine, glutamate, glutamine, serine, threonine, and tyrosine. Except for His 15 3 Asp HPr, none of these mutations showed any detectable activity when tested for activity with enzyme I by: (a) a spectrophotometric assay for enzyme I activity (43); (b) [ 32 P]-protein labeling by PEP detected by SDS-polyacrylamide gel electrophoresis and autoradiography (42); (c) a gel shift of a band on an isoelectric focusing (IEF) gel because of the introduction of the phosphoryl group (9); and (d) in vivo complementation of the fermentation negative phenotype of the ptsH strain, E. coli ESK108.

His
His 15 3 Asp HPr, when incubated with PEP, enzyme I, and Mg 2ϩ , revealed two new species, one with a lower pI and another with a higher pI (Fig. 2, A and B). The formation of the high pI form was not efficient at room temperature (22°C) but was readily detected at 37°C (Fig. 2C). When [ 32 P]PEP was used, the lower pI band was shown to contain [ 32 P]phosphate by autoradiography; the higher pI band did not have a phosphoryl group (Fig. 2, D and E). Enzyme I phosphotransfer activity was measured using His 15 3 Asp HPr and was shown to have a K m of 66 M for His 15 3 Asp HPr (wild type HPr K m 6 M) and a V max that was 0.1% of that obtained with wild type HPr.

P-Asp HPr Can
Act as a Phosphodonor-The impairment of the enzyme I reaction was large, and thus much higher amounts of enzyme I were used, greater than 100-fold compared with equivalent experiments with wild type HPr. Assays of the enzymes II sugar with His 15 3 Asp HPr would require impractical amounts of enzyme I to meet the requirements of independence of the enzyme II reaction from P-HPr generation (43). For this reason, sugar phosphorylation was not measured. In the experiment described below, which showed enzyme IIA glc phosphorylation using [ 32 P]PEP, the protein preparations required the purification of all the PTS proteins from strains of E. coli that did not produce HPr. When this was done, phosphorylation of enzyme IIA glc that was dependent upon the presence of His 15 3 Asp HPr could be shown (Fig. 3). To assess reactions with other enzymes IIA sugar , an in vivo approach was used. When His 15 3 Asp HPr was overproduced in vivo, it would not complement sugar fermentation in the ptsH strain, E. coli ESK108. His 15 3 Asp HPr has a 10,000-fold impairment; Ser 46 3 Asp HPr is the next most impaired HPr described, ϳ1000-fold, and its overproduction in vivo results in delayed fermentation (33). His 15 3 Asp HPr is not effective in the overall function of the PTS.
Characterization of the High pI Species of His 15 3 Asp HPr-The pH stability of the different pI species was found by carrying out the phosphorylation reaction and then putting samples at different pHs and following the progress of the loss of species by IEF. These results showed that the high pI form was more stable at acidic pH and that the P-Asp15 HPr was more stable at alkaline pH.
The deamidation events at Asn 38 and Asn 12 in HPr (19,25) suggested that the higher pI form might be a succinimide ring (Fig. 1). This form results in the loss of a water molecule from the protein, a net loss of 18 mass units. Moreover, when the succinimide hydrolyzes, the normal distribution of products is about 70% isoaspartyl and 30% aspartyl. The resulting HPr species have very similar pIs and have been distinguished by the detection of doublet bands on IEF gels (19). In gels such as shown in Fig. 2, there was no indication of doublet bands for any of the pI species. No increase in protein carboxymethyl transferase methylating activity could be detected with His 15 3 Asp HPr or the derivatives following phosphorylation. The methylation reaction requires isoaspartyl residues. In addition, N-terminal sequencing of His 15 3 Asp HPr preparations and the derivatives obtained after dephosphorylation gave normal recoveries of an aspartyl residue at position 15. If an isoaspartyl residue forms, the sequencing reactions do not proceed through the isoaspartyl residue (22).
The high pI form of His 15 3 Asp HPr was purified as described under "Experimental Procedures." Mass spectroscopy showed a molecular species with 18 mass units less than His 15 3 Asp HPr (Fig. 4). The isolated high pI form was sequenced from the N terminus, and the sequencing did not proceed beyond residue 14. Incubation of the isolated high pI form at pH 9 led to reversion to the normal His 15 3 Asp HPr, and N-terminal sequencing identified only an aspartyl residue at position 15 with normal recoveries. When phosphorylation of the reverted form was carried out as described in Fig. 2, the appearance of the phosphorylated and higher pI species was the same (results not shown). These results confirm an unusual structure at residue 15 and the reversibility of the whole process. These findings are consistent with either a succinimide ring formation followed by a very constrained hydrolysis reaction to yield only aspartate or the formation of an isoimide from which hydrolysis would always yield an aspartyl residue (Fig. 1).
Stability of P-Asp HPr-Phosphohydrolysis of P-Asp HPr was investigated at several pHs. The comparisons with P-His HPr are given in Fig. 5.
Tertiary Structure of His 15 3 Asp HPr-The structure of His 15 3 Asp HPr was determined as described under "Experimental Procedures." Crystallographic parameters are shown in Table I. The 1.5 Å resolution structure of His 15 3 Asp HPr is essentially the same as the 2.0 Å resolution structure of wild type HPr (14). However, His 15 3 Asp HPr had the two differences found in both the 1.6 Å resolution structure of Ser 46 3 Asp HPr (33) and the 2.5 Å resolution structure of wild type HPr bound to the Fab fragment of the HPr-specific monoclonal antibody Jel42 (18); neither the tight ␤-turn involving Asn 12 nor any torsion angle strain at residue 16 was found. The Asp 15 residue was well defined (Fig. 6A). One of the oxygen atoms of the Asp 15 carboxyl group is in essentially the same position as the N ␦1 atom in the His 15 imidazole ring of wild type; the relative distances between the position of the His 15 N ␦1 atom in wild type and the positions of the two Asp 15 carboxyl oxygen atoms in His 15 3 Asp HPr are 1.0 and 2.0 Å (Fig. 6B). The C-terminal carboxyl group of Glu 85 , which has been found hydrogen-bonded with the N⑀ 2 atom of His 15 , is found in the same position in His 15 3 Asp HPr (Fig. 6B) as described in wild type and Ser 46 3 Asp HPrs (14,33), but no hydrogen bond is formed. The side chain of Asp 15 is involved in no hydrogen bonds.
In addition to this structure, the high pI form, which contained the putative cyclized form, was crystallized, and diffraction data were collected within 10 days. The unit cell and refinement parameters are very similar to the normal His 15 3 Asp HPr (Table II). Only a well defined aspartyl residue was found at residue 15 (Fig. 6A) and in essentially the same position as in the normal His 15 3 Asp HPr (Fig. 6C).
The Double Mutant of HPr: His 15 3 Asp and Asn 12 3 Ala-A novel cyclic compound involving Asn 12 and Asp 15 was possible. To eliminate the potential involvement of Asn 12 in the formation of the high pI form, the double mutant of HPr, His 15 3 Asp and Asn 12 3 Ala was made and purified. In wild type HPr, the Asn 12 3 Ala mutation causes a small impairment in the enzyme I reaction (Table II). In His 15 3 Asp HPr, the Asn 12 3 Ala mutation leads to no detectable change in the formation of P-Asp HPr or the high pI species, and the IEF gel is essentially the same as presented in Fig. 2.
Deamidation and Cyclization at Residue 15-In order to follow more closely the events at position 15, the complications with respect to deamidation events at Asn 12 and Asn 38 were eliminated by creating the following two triple mutants Asn 12 3 Ala, His 15 3 Asp, and Asn 38 3 Ala, and Asn 12 3 Ala, His 15 3 Asn, and Asn 38 3 Ala. The events of either cyclization and/or deamidation were carried out at 60°C and at pH 5.0, 7.0, 8.0, and 10.0 followed by separation of products on IEF gels as shown in Fig. 7. These gels show that there was no indication of either cyclization or deamidation even after 90 min at either pH 5.0 or pH 10.0, suggesting that the location at residue 15 has no unusual propensity to either cyclize or produce a succinimide specifically. Deamidation of Asn 15 would have caused a band shift on the IEF gels, and none was observed even though the main chain angle and side chain 1 dihedral angle of Asp 15 (and presumably Asn 15 ) were near optimal for succinimide formation.
Properties of Asn 12 Substitutions-The following mutants were made in wild type HPr: Asn 12 3 Ala, Asn 12 3 Ser, and Asn 12 3 Thr. Each was expressed and purified, and kinetic parameters for enzyme I were determined. The results are presented in Table II. DISCUSSION The active site of HPr has two conserved residues, His 15 and Arg 17 , and various investigations of the structure of HPr indicate that a residue with hydrogen bonding potential should be found at residue 12 in HPrs from several species (18,19,44,45). Substitution of Asn 12 in E. coli HPr by either serine or threonine has little effect on the phosphorylation of HPr by enzyme I. Serine is found in Bacillus subtilis HPr and threonine in Staphylococus aureus and Streptococcus faecalis HPrs. The removal of the hydrogen bonding potential by replacement with alanine results in only modest changes (Table II), very similar to that previously reported for Asn 12 3 Asp HPr. The approximately 4-fold change in k cat /K m will presumably affect the physiological efficiency of HPr, but the hydrogen bonding potential of residue 12 does not appear to be a major requirement for the mechanism of phosphoryl transfer. These modest changes in activity concur with the lack of direct evidence in NMR spectra for a hydrogen bond between His 15 and residue 12 (45-48, 50, 51).
In contrast to the flexibility of requirement at residue 12, it was expected that histidine would be an absolute requirement at residue 15. However, His 15 3 Asp can be phosphorylated, and donate phosphate to at least IIA glc but with much reduced efficiency. The aspartyl residue is a partial structural analogue of histidine as shown in Fig. 6; one of the carboxyl oxygen atoms is structurally equivalent to the N ␦1 atom in histidine. Phosphoryl transfer between P-histidines and acyl phosphates is well established. Acetate kinase, in which a ␥-glutaminyl phosphate is formed (52), interacts with enzyme I to form a N⑀ 2 -P-histidine (53,54). Reactions in chemotaxis involve transfers of phosphoryl groups between N⑀ 2 -P-histidine in CheA (41) and aspartyl phosphate in CheY (49), which is an example of two component sensor systems.
An interesting aspect of the aspartyl substitution of His 15 is that phosphorylation catalyzes formation of a cyclized compound. Cyclization reactions to form succinimides are established for both asparagine and aspartate, and the production of isoaspartyl from the hydrolysis of succinimides is established (20 -23). Although isoaspartyl formation is favored, constraints in a protein structure might cause the formation of only aspartyl or only isoaspartyl. It has been proposed (20, 23) that a second form of cyclization can occur to yield an isoimide (Fig. 1), FIG. 5. pH stability of P-Asp HPr. Phosphohydrolysis (20 min at 37°C) of P-Asp HPr (q) was compared with P-His HPr (wild type, f) and measured as described previously (9). and the subsequent hydrolysis of this yields only aspartyl residues. There is no indication that isoaspartyl acid is formed from the cyclization of Asp 15 , and the stimulation of cyclization by phosphorylation has been anticipated (23). When either asparaginyl or aspartyl residues are at residue 15 in HPr, there is no indication of rapid cyclization in the absence of phosphorylation. This suggests that the phosphorylation of aspartate 15 is the pathway by which the cyclization to an isoimide is catalyzed.
In wild type HPr, the phosphoryl group bound to the N ␦1 atom of His 15 has hydrogen-bonding interactions with the amide nitrogens of residues 16 and 17. As the one of the carboxyl oxygens of the aspartyl at residue 15 is an analogue of the N ␦1 atom, it is reasonable to assume that this is why phosphorylation occurs and that the same interactions between the phosphoryl group and the amide nitrogens of residues 16 and 17 occurs. This would mean that the phosphoryl group is hydrogen-bonded to the amide nitrogen of residue 16, which would be involved in succinimide formation. However, it is proposed that this interaction (Fig. 1C) leads to a concerted reaction, in which the phosphoryl group extracts the proton from the amide nitrogen of residue 16, with the consequence that the phosphoryl group is a better leaving group, and that the carbonyl of residue 16 is more reactive in attacking the carbonyl of residue 15 to form the isoimide.
The complete lack of detection of significant amounts of isoaspartic acid and the ability to isolate a cyclic intermediate with 18 mass units less than His 15 3 Asp HPr are difficult to reconcile with the formation of a succinimide. It is suggested that the structure formed is an isoimide, which has not been found before in proteins.  15 3 Asp HPr. A, the active site residue Asp 15 and the electron density. The electron density map is from the 1.8 Å resolution structure of His 15 3 Asp HPr isolated high pI form in which the putative cyclization reverts to Asp 15 . B and C, comparison of the active sites of HPr. B, the active sites of wild type HPr structure (11) with His 15 and His 15 3 Asp HPr structure (1.6 Å resolution structure) are compared by superimposition of the whole HPr structures. C, similarly normal His 15 3 Asp HPr and the putative cyclized form are compared.  1-4) and HPr with triple mutations, Asn 12 3 Ala, Asn 38 3 Ala with either His 15 3 Asp (lanes 5-8) or His 15 3 Asn (lanes 9 -12). In each group, the first sample is unheated, and the next three were heated at 60°C at pH levels of 5.0, 7.0, and 10.0 for 90 min. Samples were placed on IEF gels, which would show deamidation (lower pI as in lanes 3 and 4) or cyclization (higher pI).