br Substrate Specificity Discrimination and
Substrate Specificity, Discrimination, and Binding Energy In L-Cysteinesulfinic acid to catalytic efficiency, which can be gauged in reference to an absolute scale of ‘catalytic perfection’ 10, 11, specificity is a relative concept because it requires comparison between given alternative substrates. In fact, specificity is formally defined as the ability of an enzyme to discriminate between two potential substrates, in the presence of both compounds 12, 13. In a biological context, specificity entails acting on a single substrate in preference to a multitude of other metabolites in the cell. Intuitively, this can be a very difficult exercise. For example, an enzyme intended to be specific for aspartate should discriminate against (among others) asparagine, glutamate, homoserine, homocysteine, phosphoserine, alanine, L-malate, oxaloacetate and succinate, all of which are common metabolites with obvious structural similarities to the proper (cognate) substrate. Discrimination (and hence specificity) does not depend simply on the relative affinity of the substrates for the enzyme, indeed it is acknowledged that the substrates should be compared based on the ratio of kcat/Km values for their reactions  – a ratio also called the discrimination factor . A pertinent question therefore is whether there are intrinsic limits to this ratio. One potential way to address this issue is through application of transition state theory. According to this theory, the logarithm of kcat/Km is proportional to the free-energy difference (ΔG≠) between the free enzyme and substrate and the transition state complex  (Box 2). Thus, whenever transition state theory is applicable, differences in kcat/Km between different substrates reflect their different binding energies in the transition state. In addition, the amount of binding energy provided by simple groups is finite. Hence, when comparing substrates with similar structures, the difference in binding energies (ΔΔG≠) must also be finite, relatively small, and, in some cases, calculable. To illustrate this point, I next analyze three exemplary cases in which an enzyme is asked to discriminate against alternative substrates that are either slightly smaller or slightly larger than the cognate substrate (Figure 1).
Theoretical and Empirical Limits of Substrate Specificity That substrate specificity is inherently limited was first glimpsed by Pauling 60 years ago  in relation to the process of aminoacyl-tRNA synthesis. He focused in particular on the case of isoleucyl-tRNA synthetase that must distinguish between isoleucine and valine. The two amino acids differ only by one methyl group, and the difference in binding energy between them could therefore not be greater than the energy provided by the terminal CH3 group of L-Ile. This has been estimated to be worth at most 3kcal/mol ( and references therein), which sets an upper limit of about 160-fold in the discrimination between valine and isoleucine by simple binding. Such a relatively low discrimination factor seems to be biologically unacceptable, in fact extant isoleucyl-tRNA synthetases possess a distinct proofreading function that selectively deacylates any mis-aminoacylated tRNAIle, achieving a better accuracy in an energy-expensive manner . Evidently, the limit above does not apply solely to aminoacyl-tRNA synthetases, but is general for enzymes that must distinguish between two substrates differing by only one methyl group. Pairs of this type are not infrequent among common metabolites. A survey of the Brenda database  shows that the observed discrimination index of extant enzymes only rarely reaches the theoretical limit (Figure 1A). One can also try to calculate the theoretical limits of discrimination in cases when the alternative substrate lacks a hydroxyl group compared to the cognate substrate. The binding energy provided by a hydroxyl depends essentially on the ability of the group to form hydrogen bonds, and the thermodynamic stability of hydrogen bonds within macromolecules has been addressed in several mutagenesis studies. The maximum energy associated to hydrogen bonding in an uncharged donor–acceptor pair is ∼2kcal/mol, whereas a hydrogen bond to a charged partner can be worth up to 5kcal/mol . Because interactions with more than one charged partner would generate electrostatic interference, and because hydroxyl groups can form up to three hydrogen bonds, one rough estimate is that an extra hydroxyl group could be worth up to 9kcal/mol of binding energy, or a factor of ∼4×106-fold in selectivity . Figure 1B shows a set of data from the literature pertaining to this case. Although the sample may be somewhat biased (because catalytic parameters for substrates showing very low reactivity are less likely to be determined), it nevertheless suggests that discrimination is often much lower than the theoretical maximum (Figure 1B and Table S2 in the supplemental information online).