Discussion
Many of the compounds tested show an uncompetitive inhibition behavior indicative of an interaction with the P-site. An interaction of the 1,3-diketones with the e-amino group of the active site lysine can reasonably be postulated to be responsible for the uncompetitive inhibition behavior of our compounds (see also [6], [7]) :
Assuming that the diketones interact with the active site lysine only on the keto function which is in the "correct" distance compared to the carboxylate function two tautomeric forms can be postulated. The tautomer 1 is analogous to the intermediate (E-iminium ion) obtained by the fixation of the first substrate on the enzyme. This tautomeric form has been postulated by Jaffe based on her NMR results.[1] The positive charge is stabilized in the case of a 1,3-diketone by H-bonding with the second keto group. This tautomer formally results from the release of HO-. The tautomer 2 is clearly more stable and results from a formal release of water. The deprotonation of the first tautomer leads to the second form. In solution the two tautomeric forms 1 and 2 are in rapid equilibrium with a strong probably almost exclusive preference for the tautomer 2. The rate for the tautomerisation and the equilibrium constant between the two tautomers for the inhibitor bound to the active site will certainly be different from the behavior in solution.
At this stage we can postulate three scenarios :
The e-amino group of the active site lysine attacks the carbonyl group forming the tetrahedral intermediate which eliminates water after protonation by some acid function present at the active site to give the tautomer 1.
The mechanism is the same as above forming the tautomer 1 first, which is then deprotonated by a base and gives the tautomer 2. The life time of the tautomer 1 is difficult to evaluate.
The tetrahedral intermediate can eliminate water directly by a base catalyzed process where one of the diastereotopic protons of the activated methylene group is removed by a judiciously placed base. In this process the tautomer 2 is formed directly without passing the iminium ion. The process could be induced by two separate functional groups at the active site: an acidic group for the protonation of the OH-group and a base for the deprotonation of one of the protons at the methylene group. Alternatively the same functional group could play both roles. This functional group could function as proton relay between the OH-group and the methylene proton.
The e-amino group of the active site lysine has to be present in its deprotonated form to be able to function as nucleophile capable of attacking the keto function of the bound substrate. The substrate in solution at the optimal pH for PBGS of E. coli will be present in its zwiterionic form (ammonium carboxylate). After the condensation to form the Schiff base the enzyme-substrate complex will be present in the form of the iminium ion stabilized by a hydrogen bridge to the free amino group of 5-aminolevulinic acid, as has been shown by the elegant NMR-experiments of Jaffe.[1] The environment of the active site has to be optimally adapted to stabilize the formation of the Schiff base and to allow the sequence of proton transfer reactions necessary for this process. It is interesting to note that the pKa of the 1,3-diketones is quite similar to the pKa of a primary amine. Protonation and deprotonation of the 1,3-diketones should therefore be possible provided that the inhibitor is bound correctly at the active site.
1,3-diketones can in principle be recognized as well at the P- and at the A-site. As well the P-site as well as the A-site of the enzyme have to be able to recognize a keto function. 1,3-diketones are inhibitors containing two keto groups so they could potentially interact at the both sides at the same time :
The 4,6-dioxoheptanoic acid (1) is by far the best inhibitor tested with a Ki = 1.4 µM and an uncompetitive or mixed inhibition behavior. In order to evaluate the influence of the different steric and electronic arguments a series of compounds derived from 1 have been synthesized and tested :
At first we wanted to evaluate the importance of the 1,3-diketo function for the recognition. The 2,4-pentandiol (13), which can be considered to be the tetrahydro-derivative of 2,4-petandione, was tested. The 2,4-pentandiol (13) as the mixture of diastereoisomers was tested and no inhibition effect could be observed up to concentrations of 50'000 µM. The weak chelating ability of the diol 13 and its capacity to form H-bridges are not creating important enough interactions with the active site to make it an inhibitor. At the A-site where a Zn2+-oxygen coordinative interaction has been postulated the 2,4-pentandiol (13) could and should show some interaction. The absence of any inhibitory effect is an indication that the coordination to the A-site Zn has to be relatively weak.
The 2,5-hexandione (14), a compound which contains two keto functions, but in a 1,4-arrangement shows an uncompetitive inhibition behavior with quite a big Ki-value (Ki = 24'900 µM). The Ki-value is nearly one order of magnitude bigger than the values determined for 2,4-pentandione (11) and for levulinic acid (Ki = 2'220 µM) which can be considered to be the 1-oxa-2,5-hexandione. On top of that levulinic acid exhibits a competitive inhibition behavior. Levulinic acid has been the standard inhibitor for PBGS for a long time.[8] Earlier tests had shown that the carboxylate is important for the recognition, for instance methyl levulinate, where the carboxylate has been replaced by an ester function, is no more an inhibitor.[9] Comparing the electrophilicity of the keto function of 2,5-hexandione (14) with that of the keto function of levulinic acid should the 2,5-hexandione (14) should be more reactive towards the e-amino group of the lysine at the P-site assuming that the inhibitor will be present in the tautomeric form 1. Assuming an interaction as well with the A- as with the P-site we would conclude that 1,3-diketones display the optimal distance between the two keto functions for the recognition.
The 2,4-pentanedione (11) shows an uncompetitive or mixed inhibition behavior with a moderate Ki-value (Ki = 3'680 µM). The 2-acetylcyclohexanone (8) is an uncompetitive inhibitor as well with an inhibition constant of Ki = 5'520 µM. This result is consistent with an increase in steric hindrance in comparison with 2,4-pentandione (Ki = 3'680 µM). The effect of the increased steric hindrance is limited probably due to the fact that the molecule is bound to the P-site and there must therefore be space available for the second substrate. As the compound is mainly present in its enolized form the molecule is flat and the six membered ring could imitate the pyrrole ring of the product.
The acetylacetic acid (12) (=1-oxa-2,4-pentandione) shows competitive inhibition behavior with a very weak inhibition constant Ki = 39'500 µM. The 3-oxo-1,6-hexandioic acid (=7-oxa-4,6-dioxoheptanoic acid) shows the same characteristics in the inhibition tests (Ki = 10'400 µM; competitive inhibition behavior). Those two results are in favor of the formation of the tautomer 2. Replacing the second keto function by a carboxylate deactivates the acidity of the a hydrogen and should thereby reduce the inhibition potency of those compounds. This is in accordance with the observed results. For the tautomer 1 the presence of a carboxylate in the a-position should stabilize the iminium ion. This is in contradiction with the observed inhibition behavior.
Compounds 15 to 20 are derivatives of methyl 4,6-dioxoheptanoate (1). They contain one of the keto groups blocked in its enol form. Comparing the pairs of regioisomers 15 and 16, 17 and 18 and 19 and 20 we hoped to be able to evaluate the importance of the "correct" distance between the carboxylate function and the keto function. At the same time the electrophilicity of the keto function is strongly influenced by the sort of substituent introduced in the b-position The influence of the electronic characteristics should therefore be possible. In the case of the compounds 15, 16, 17 and 18 the formation of H-bonding towards the iminium ion is still possible. Unfortunately a series of experimental problems prevented us from obtaining reliable Ki-values.
The two 0-methylated compounds methyl 4-methoxy-6-oxo-4-heptenoate (15) and methyl 6-methoxy-4-oxo-5-heptenoate (16) are subject to hydrolysis under the conditions of the test (at least 10% of methyl 4,6-dioxoheptanoate are formed within 30 minutes). It is not possible to determine correctly the importance of each compound for the observed inhibition result. The Ki-values for the compounds 15 and 16 can not be calculated from the experimental results because both compounds are much weaker inhibitors than the product of hydrolysis, the methyl 4,6-dioxoheptanote (2).
The methyl 4-amino-6-oxo-4-heptenoate (17) is a slow binding inhibitor. The isomer methyl 6-amino-4-oxo-5-heptenoate (19) does not show the same behavior. Unfortunately 19 forms a pyrrole with 5-aminolevulinic acid which disturbs the Ehrlich reaction used for the enzyme test.
The 2,4-dioxopentanoic acid (9) is an uncompetitive inhibitor with an inhibition constant Ki = 323 µM. It is two order of magnitude larger than 4,6-dioxoheptanoic acid (1) but one order of magnitude better than Ki for 2,4-pentandione (11) or for levulinic acid. The 2,4-dioxopentanoic acid (9) can be considered either as a derivative of levulinic acid. In this case the P-site of the enzyme will be able to recognize the carboxylate function of 2,4-dioxopentanoic acid and will then form the iminium ion on position 4 which probably will be transformed by deprotonation into the enamine conjugated with the keto function in position 2. Because the iminium ion has been shown to be in its (E)-configuration, the second keto group is placed on the wrong side to stabilize conjugated system obtained by deprotonation by H-bonding. An alternative explanation would be that the inhibitor interacts as carboxylate substituted derivative of 2,4-pentadione (11). This allows the stabilization of the iminium proton and places the a-proton of the inhibitor in the basic area of the enzyme. But the carboxylate will be forced to be present in the basic area of the enzyme as well. This unfavorable interaction should considerably reduce the affinity of this inhibitor towards the enzyme. In view of this argument we prefer the first hypothesis where we consider the 2,4-dioxopentanoic acid (9) as an analogue of levulinic acid.
The 3-acetyllevulinic acid (10) (=3-(carboxymethyl)-2,4-pentadione) is a surprisingly a poor competitive inhibitor (Ki = 15'000 µM). The Ki-value is similar to the value determined for unstabilized 3-substituted levulinic acid [9] but a factor of five bigger than the value determined for the 2,4-pentadione (11). The result is best explained assuming that the levulinic acid part overrules the 1,3-diketo part of the molecule. If this is the case then the P-site of the enzyme is already occupied by the first substrate and the introductions of the additional acetyl function creates steric hindrance. We hoped that 3-acetyllevulinic acid (10) would function as model for the intermediates postulated for the Shemin or Jordan II mechanism.[7]. Each of the keto function could interact with on of the active site. The high value for the inhibition constant for 10 is a strong indication to discard the hypothesis for the binding of this compound to the active site as a model for intermediate.
The most important results were obtained with the compounds which are structurally closely related to the 4,6-dioxoheptanoic acid (1). Changing the carboxylate against the ester function increased the value of the inhibition constant by a factor of 200. The Ki-value was 319 mM for the methyl ester of 4,6-dioxoheptanoic acid (2). This decrease in inhibition efficiency is in accordance with earlier observations from our and other laboratories [3], [9].
The 4-oxo-4-(2-oxocyclohexyl)butyric acid (5) is an uncompetitive inhibitor with an inhibition constant Ki = 337 µM. This result is somewhat surprising because the inhibition constant is of the same order as for the ester 2 and not comparable with the 4,6-dioxoheptanoic acid (1). The fact that the inhibition constants for the two diketones 11 and 8 (cyclohexyl derivative of 11) are very similar whereas in the case of 1 and 5 there is a difference of over 200 is not easy to rationalize.
The nitro analogue of 4,6-dioxoheptanoic acid (1), 1-nitro-3,5-dioxohexandione (3), is an uncompetitive inhibitor with Ki = 14.5 µM. It has to be taken in account that this compound is not perfectly stable under the condition of the test. In view of the Ki-value which is only a factor 10 higher than the value determined for the 4,6-dioxoheptanoic acid (1) is a good indication that the 1-nitro-3,5-dioxohexandione (3) is recognized at the P-site as well. It seems safe to conclude that the change of carboxylate to nitro function is not as favorable for the P-site as it is in the A-site where we observed a Ki-value which is a factor of 60 smaller for the nitro derivative compared to the carboxylate [9].
The nitro analogue of 4-oxo-4-(2-oxocyclohexyl)butyric acid (5) the 2-(3-nitro-propionyl)cyclohexanone (4) shows surprisingly competitive inhibition behavior with a Ki = 16.5 µM. This value is better than the value determined for 4-nitro-2-butanone (Ki = 28 µM) and is practically the same than the value determined for 1-hydroxy-4-nitro-2-butanone, which are both competitive inhibitors as well [10]. It is not only astonishing that the change from a carboxylate derivative 5 to a nitro derivative 4 changes the type of inhibition but it is also remarkable that the Ki-values is a factor of 20 smaller for the nitro compound. It is known [9] that the replacement of a carboxylate group by a nitro function in an inhibitor bound to the A-site of the enzyme increases the inhibition efficiency. At the same time the additional steric hindrance to the cyclohexyl ring has to be compensated. 2-(3-nitro-propionyl)cyclohexanone (4) in solution is nearly exclusively present in its enol form in CDCl3. The enol form is certainly capable of a better complexation to ZnA. However this is probably not a sufficient explanation because the 2,4-oxopentanoic acid (9) is an uncompetitive inhibitor, therefore preferentially binding to the P-site of the enzyme. 9 has only a moderate inhibition constant despite the fact that 98% of the molecule are present in its enol form in CDCl3, which should be at least as favorable for chelation to ZnA as is the case for compound 4.
In an effort to answer the question how important the tautomeric form 2 might be for the inhibition results obtained with the different 1,3-diketones, we synthesized compounds which should bind only to the P-site of the enzyme and which should at the same time be unable to form the vinylogous amide (tautomeric form 2). Two compounds fully alkylated between the two keto groups were synthesized. The introduction of a methyl group in this position has undoubtedly a steric effect on the binding of the inhibitor, which is difficult to asses beforehand.
The 5,5,7-trimethyl-4,6-dioxooctanoic acid (7), the tetramethylated analogue of 4,6-dioxoheptanoic acid (1) presents kinetics data that we interpret as characteristic for a mixed inhibition with a very poor inhibition constant Ki = 44'000 µM. This compound is still sterically demanding but more flexible compared to 6. The reduction of the steric influence seems to be the key to have access to the P-site. We are exploiting this way to generate new blocked 1,3-diketones.
The methyl blocked form of 4-oxo-4-(2-cyclohexyl)butyric acid, the racemic 4-oxo-4-(2-(2-methyl)cyclohexyl) butyric acid (6), exhibits a poor competitive inhibition with an inhibition constant Ki = 30'000 µM. The fact that we observe competitive inhibition and such a high value for the inhibition constant is a strong indication that the compound 6 binds to the A-site. The introduction of the additional keto function should facilitate the formation of the iminium ion. Obviously this electronic effect is not strong enough to obtain an inhibitor for the P-site of the enzyme. An alternative explanation could be that the additional methyl group creates sufficient steric hindrance, that binding to the P-site is not any more possible. At the moment we can not distinguish between these two possibilities. If our interpretation that 6 is an inhibitor of the A-site only is correct, the low efficiency of 6 is in accordance with all the facts obtained so far. As we have shown that substrate-like inhibitors recognized at the A-site of the enzyme are sensitive to steric hindrance (see also the discussion of the inhibition results of 10) the Ki-value obtained for 6 is in line with our earlier results.