The enediolate analogue 5-phosphoarabinonate as a mechanistic probe for phosphoglucose isomerase.

A stable analogue has been prepared of the enediolate anion believed to occur transiently in the reaction of phosphoglucose isomerase. This compound, 5-phosphoarabinonate, is the strongest known competitive inhibitor of the enzyme (Ki = 3 times 10(-7) M below pH 7). A distinctive pH dependence of binding, also found for two other aldonic acid omega-phosphates, 6-phosphogluconate and 4-phosphoerythronate, involves pertubation of a pKa from 7.0 in the free enzyme to 9.0 in the enzyme-inhibitor complex. This perturbation may reflect a catalytically advantageous increase in basicity which occurs around the transition state of the normal enzymatic reaction.

A stable analogue has been prepared of the enediolate anion believed to occur transiently in the reaction of phosphoglucose isomerase. This compound, &phosphoarabinonate, is the strongest known competitive inhibitor of the enzyme (Ki = 3 x lo-' M below pH 7). A distinctive pH dependence of binding, also found for two other aldonic acid w-phosphates, 6-phosphogluconate and 4-phosphoerythronate, involves perturbation of a pK. from 7.0 in the free enzyme to 9.0 in the enzyme-inhibitor complex. This perturbation may reflect a catalytically advantageous increase in basicity which occurs around the transition state of the normal enzymatic reaction.
The mechanism for the base-catalyzed interconversion of aldoses and ketoses via an enediolate anion was originally proposed by Wohl and Neuberg (1) and revived by Topper (2) for phosphoglucose isomerase. A general base at the active center of the enzyme is thought to transfer stereospecifically a carbon-bound proton of the substrate between C-l and C-2. This mechanism is supported by two types of evidence: isotope experiments, reviewed by Rose (3) and by us (4), and studies with transition state analogues. 1 The first transition state analogue for an aldose-ketose isomerase was recognized by Wolfenden (5). He suggested that the carboxylate anion of the inhibitor, phosphoglycolate, is a specific analogue of the enediolate anion intermediate in the triose phosphate isomerase reaction. The transition state analogue approach and cases of its application have been discussed by Wolfenden (5) and Lienhard (6). It has been known for some years that 6-phosphogluconate is a strong competitive inhibitor of phosphoglucose isomerase with a pH dependence of binding greatly different from that of the substrates (7)(8)(9). Similarly, a strong dependence on pH has been described by Wolfenden  cis-1,2-enediolate intermediate (see Fig. 1) in the phosphoglucose isomerase reaction (11). This paper reports the confirmation of this prediction that &phosphoarabinonate should be a potent inhibitor of phosphoglucose isomerase. The corresponding four-carbon analogue, 4-phosphoerythronate, has also been prepared and studied. Preliminary accounts of this work have been given (12,13).

Materials
Phosphoglucose isomerase was isolated from rabbit muscle and purified to homogeneity by column chromatography on CM-Sephadex as described previously (14). graphed as a single, phosphate containing spot on cellulose thin layer plates (see "Methods").
An indirect indication of the purity of the Sphosphoarabinonate was given by the coincidence of the competitive inhibition results obtained with inhibitor samples synthesized and purified by the two different routes (see legend to Fig. 4).
A relatively specific color reaction for 5-phosphoarabinonate was found in the chromotropic acid test (15,28). About 1 mM 5-phosphoarabinonate in concentrated sulfuric acid plus 0.1% 4,5-dihydroxynaphthalene-2,7-disulfonic acid gave a green color upon heating for 15 min at loo", which turned to red after 2 hours at 100". The full spectrum for the colored product of 5-phosphoarabinonate with chromotropic acid (after 2 hours of heating) showed X,,. at 463 nm (c = 1400 liter mol-' cm-'), x mln at 695 nm, and a relative X,,, at 546 nm. Fig. 3 shows the 13C nuclear magnetic resonance spectra of potassium arabinonate and of trisodium 5-phosphoarabinonate synthesized from fructose B-phosphate. The lower trace is the spectrum of authentic potassium arabinonate.
The carboxylate at C-l is assigned to the peak at J = 171.2 ppm. The three peaks between J = 76.4 and 78.0 ppm are assigned to positions 2 through 4, which are all hydroxymethylene carbons. The peak at the highest field, J = 61.6 ppm, is attributed to the C-5 hydroxymethyl carbon. These assigments are consistent with known values for chemical shifts (29). The upper spectrum shows sodium 5-phosphoarabinonate synthesized from fructose 6-phosphate. The C-5 peak is converted to a doublet by the 31P nucleus of the phosphate group attached to this carbon. There is also peak multiplicity at around 77 ppm within the C-2 through C-4 cluster, which may represent splitting of the C-4 signal by phosphorous; two-and three-bond carbon-phosphorous couplings of this type have been observed with cyclic nucleotides (30). To determine the pK, values of the aldonic acid phosphates, 10 mg of 5phosphoarabinonate or 6-phosphogluconate in a volume of 2 ml were acidified and titrated with 1 N NaOH in a Radiometer TTTlc/ABUlc titrator system equipped with microelectrodes.
At 30" values of 3.7 and 6.3 for pK, and pKs, respectively, were found for both compounds.
Inhibition Studies on Phosphoglucose Isomerase-Results from competitive inhibition experiments were plotted according to the procedure of Dixon and Webb (31). In plots of this type upward breaks in the curve are caused by ionizations in the enzyme-inhibitor complex and downward breaks by ionizations in free enzyme or free inhibitor. Fig. 4 shows a Dixon plot for the inhibition of rabbit muscle phosphoglucose isomerase by 5-phosphoarabinonate.
The theoretical straight-line intersections indicate pK, values of 7.0 and 9.0. Below pH 7, K, is 3 x lo-' M. Fig. 5 shows a similar Dixon plot for 4-phosphoerythronate, fitted to the pK, values of Fig. 4. The curve is displaced downward, the K1 being 2 x 10e6 M below pH 7. The fact that the K1 of Sphosphoarabinonate below pH 7 is 10es times the best value for substrate Kd, rising to l/10 Kd at pH 9, supports the reasoning that this inhibitor is a good analogue of the cis-1,2-enediolate anion in the enzymatic reaction.
As can be seen from Fig. 6 (which summarizes all the inhibitor binding results from this laboratory) the four-, five-, and six-carbon enediolate analogues (for structures see Fig. 7) bind to phosphoglucose isomerase in a manner very different from that of inhibitors without carboxyl groups at C-l (32).  Both 6-phosphogluconate and 4-phosphoerythronate bind to the enzyme less well than 5-phosphoarabinonate.
A reasonable explanation is that the former must experience some steric crowding to position the C-l carboxylate correctly, while the latter may need to be reoriented to achieve proper binding of C-l and the phosphate group. It is notable that all three aldonic acid phosphates show the same pH dependence, involving pK, values of 7.0 and 9.0. The very tight binding of 5-phosphoarabinonate and the pH dependence of binding peculiar to the aldonate inhibitors provide strong confirmation of our orginal supposition ( Fig. 1) that 5-phosphoarabinonate should be a potent transition state analogue' of phosphoglucose isomerase.
The two pK, values involved in aldonate binding are most readily explained by assigning, as did Wolfenden for triosephosphate isomerase (lo), the one at 7.0 to the catalytically active base which governs V,,,,, (8). The second pK, at 9.0, which was not seen with triosephosphate isomerase (5), we interpret to reflect an upward perturbation of the pK, at 7.0, which is caused by the binding of the enediolate form or its analogues.
According to a previously postulated mechanism (8), maximum catalytic activity requires the protonation of a residue with a pK, equal to 9.3 and the deprotonation of another with a pK, equal to 6.9. On first sight these two pK, values could be the same as those observed to govern the binding of 5-phosphoarabinonate.
For the lower pK, such identity is reasonable since in both cases the pK, values at around pH 7 are assigned to the free enzyme. However the higher pK, at about pH 9 is associated with the enzyme-inhibitor complex in the case of 5-phosphoarabinonate binding but with the free enzyme in the case of productive catalysis. It is not possible to assign a lysine t-amino group the pK, value of 9.0 which governs binding of the enediolate analogue, because this fails to explain two observations: (a) the involvement of the lysine in ring-opening and (b) the uptake of a proton upon formation of the enzyme-5-phosphoarabinonate complex above pH 7.

DISCUSSION
Studies with transition state analogues for triosephosphate isomerase (5,33) suggest that planar double-bond geometry and a negative charge at C-2 are the essential features of the enediolate intermediate which must be incorporated into a good analogue inhibitor, such as first shown by Wolfenden for phosphoglycolate (5). In the case of phosphoglucose isomerase, 5-phosphoarabinonate, which possesses these features, binds to the enzyme and perturbs from 7 to 9 the pK of what may be the base responsible for proton transfer. This pK shift places constraints on the possible chemical identity of the base catalyst. Regarding the mechanism of interaction between enzyme and ligand, Pauling's original theory (34) has been extended to suggest (5,6,35) that transition state analogues may owe their tight binding to direct interaction with catalytic groups at the active site.
The binding of 5-phosphoarabinonate to phosphoglucose isomerase shows two aspects which call for explanation: its very low K, and its unusual pH dependency. Below pH 7 the K, is lo3 times smaller than K, for the substrates, suggesting that the specific interactions between enzyme and this inhibitor must be quite different from those in the Michaelis complexes. group also to interact electrostatically with the enediolate species and its analogues, thereby increasing their binding. This amino acid residue has recently been implicated in epoxide ring opening of the active site reagent 1,2-anhydromannitol 6-phosphate (37), and it was suggested that the critical side chain might be the c-amino portion of a lysine residue, possibly that involved in hemiacetal ring opening (4,8). Alternatively, we propose that it may be an arginine side chain, which has the additional function of liganding the oxygens of C-l and C-2 of the substrates and of the enediolate intermediate in the cis configuration. In addition to the tight binding of 5-phosphoarabinonate to phosphoglucose isomerase (the highest value of Ki is still 10 times smaller than K, for the substrates), there exists a second type of interaction between enzyme and inhibitor at the active site. This interaction is manifested in the shift from 7.0 to 9.0 of a pK, which may be the same pK, involved in catalysis (8). If one assigns the shifted pK, to the catalytically active base, it can be interpreted as the result of a direct electrostatic effect or of a conformational change which alters the environment of the base. Both interpretations are compatible with the base being a histidine imidazole group as advocated by our laboratory (4).
Shifts in pK, of this type have been postulated to occur in the hemoglobin alkaline Bohr effect (38) and in the activation of chymotrypsinogen.
In both cases a pK shift occurs on breaking a noncovalent imidazole-carboxyl bond: in the Bohr effect the magnitude of the pK, change is 1.6 units (39); with chymotrypsinogen it is 1.2 units (40). Such a pK, shift is difficult to envisage for phosphoglucose isomerase if the base is a glutamyl y-carboxylate group as suggested by O'Connell and Rose (37). Since this carboxylate group would already have its pK, perturbed from 4 to 7, it seems unlikely that it could then be perturbed upward by another 2 units as a consequence of interaction with a second carboxyl. Nevertheless, the data of these authors indicate the presence at the active site of a glutamate residue possessing an unusually nucleophilic y-carboxyl group. However, the heat of ionization associated with the base catalyst in rabbit muscle phosphoglucose isomerase has been estimated at 7700 cal/mol (8). This value argues against the identification of the base with a simple carboxyl, for which AHi would be expected to be between -1500 and +1500 cal/mol, while the value found is typical for the N-H bond in the imidazolyl portion of a histidine residue.
An attractive explanation for the apparent high reactivity and the abnormal pK, of a glutamate y-carboxyl group would be the formation of an imidazole-carboxyl pair similar to that first seen by Blow et al. (41) in the x-ray structure of the active site of chymotrypsin, which shows a AH," of 9600 cal/mol (42), a value of the same magnitude as that seen for phosphoglucose isomerase. It should be noted, however, that the side chain reactivities and the catalytic functions of phosphoglucose isomerase and of the serine proteases differ considerably. Furthermore, the charge localization in the protease imidazolecarboxyl pairs is unresolved (43, 44); therefore, such a pair, if it occurs in phosphoglucose isomerase, may be quite different from that seen in chymotrypsin.
One possible interpretation is that substrate interaction with the active site may induce transient conformational changes which alter the geometry of an imidazole-carboxyl pair and thence its pK,. A transient increase in basicity could in tXs manner enhance catalysis (45). Thus the pK, shift from 7.0 to 9.0, seen upon inducing phosphoglucose isomerase with 5-phosphoarabinonate into its enediolate binding conformation, may represent part of a larger increase in basicity at the transition state. Such an increase in basicity would explain the enzyme's ability to catalyze at physiological pH the abstraction from substrate of a carbon-bound proton, a reaction which otherwise occurs only in strong alkali (4).
This explanation for the mode of binding of aldonic acid o-phosphates to phosphoglucose isomerase is supported by the observation of a conformational change with 6-phosphogluconate. Upon treatment of crystals of the pig muscle enzyme with the inhibitor, a small region of intensity difference is observed in x-ray diffraction patterns (46). We conclude that the binding of 5phosphoarabinonate and its four-and six-carbon homologues mimics that of the metastable enediolate anion to the active site of phosphoglucose isomerase. This mode of binding is specific for aldonate inhibitors and may hinder the ionization of a proton from an imidazole-carboxyl pair, thereby raising its apparent pK, from 7 to 9. This pK, perturbation may occur by direct electrostatic interaction as well as by induced conformational changes, and it may represent a catalytically advantageous increase in basicity which occurs only transiently in normal catalysis. It will be of interest to see if the present conclusions about the active site of phosphoglucose isomerase are confirmed by the x-ray crystallographic studies now in progress at the University of Bristol.