|Home | About | Journals | Submit | Contact Us | Français|
We used a series of dATP and dGTP analogues to determine how DNA polymerase I from Bacillus stearothermophilus (BF), a prototypical A family polymerase, uses N-1, N2, N-3, and N6 of purine dNTPs to differentiate between right and wrong nucleotide incorporation. Altering any of these nitrogens had two effects. First, it decreased the efficiency of correct incorporation of the resulting dNTP analogue, with the loss of N-1 and N-3 having the most severe effects. Second, it dramatically increased misincorporation of the resulting dNTP analogues, with alterations in either N-1 or N6 having the most severe impacts. Adding N2 to dNTPs containing the bases adenine and purine increased polymerization opposite T, but also tremendously increased misincorporation opposite A, C, and G. Thus, BF uses N-1, N2. N-3, and N6 of purine dNTPs both as negative selectors to prevent misincorporation and as positive selectors to enhance correct incorporation. Comparing how BF discriminates between right and wrong dNTPs with both B family polymerases and low fidelity polymerases indicates that BF has chosen a unique solution vis-à-vis these other enzymes, and, therefore, that nature has evolved at least three mechanistically distinct solutions.
A key issue facing DNA polymerases is how to minimize incorporation of incorrect dNTPs in order to avoid the potentially deleterious effects of mutagenesis. For “high fidelity” DNA polymerases, the measured error frequencies vary from around 10-3 - 10-5 errors per dNTP polymerized (1, 2). However, despite substantial study by a variety of techniques including protein mutagenesis, substrate mutagenesis, and high resolution structural approaches (3-11), the key features that differentiate right and wrong dNTPs as well as the mechanisms that polymerases employ to accomplish this differentiation remain rather unclear.
Several different mechanisms that could allow a DNA (or RNA) polymerase to accurately copy a template have been proposed. Discrimination based on a requirement to form Watson-Crick hydrogen bonds was one of the first mechanisms proposed, and several polymerases may well employ this mechanism (12-15). For example, both human and herpes DNA primases, two very low fidelity enzymes, appear to require formation of Watson-Crick hydrogen bonds in order to efficiently polymerize a NTP. The low fidelity DNA pol η and DNA pol IV from Sulfolobus solfataricus also did not efficiently polymerize several dNTPs incapable of forming Watson-Crick hydrogen bonds. A second model posits that the shape of the base pair formed by the incoming dNTP and template base provides the key factor for discriminating between right and wrong (d)NTPs (16-20). Indeed, Kool and coworkers showed that both T7 DNA polymerase and Klenow Fragment, two high fidelity A family polymerases, will efficiently and accurately generate base-pairs between adenine and 2,4-difluorotoluene nucleotides. Importantly, the shape of the latter base closely mimics thymine. Furthermore, they showed that varying the shape of the modified toluene significantly altered the efficiency of polymerization in a manner consistent with the “shape selectivity” model. However, several groups have shown that Klenow Fragment also has the ability to efficiently generate base-pairs between nucleotides whose bases cannot form a correctly shaped base-pair (7, 9, 21-23). Finally, a combined use of both positive and negative selectivity has been proposed for two B family replicative polymerases, pol α and herpes DNA polymerase (7, 24, 25). In this model, the enzyme used specific chemical features of the incoming dNTP to both prevent misincorporation (N-1, N-3, and N6 in the case of dATP) and enhance correct incorporation (formation of Watson-Crick hydrogen bonds).
The large fragment of DNA polymerase I from Bacillus stearothermophilus (BF) 1 is an A family, high fidelity polymerase (26). It exhibits substantial homology with other A family polymerases that have been characterized kinetically and structurally, including Klenow Fragment of DNA polymerase I (E. coli), T7 DNA polymerase, and KlenTaq from Thermus thermophillus (3, 27-30). BF has been extensively characterized structurally, with high resolution structures of the apoenzyme, BF-DNA binary complexes, and BF-DNA-dNTP ternary complexes available (4, 5, 31, 32). Indeed, BF even retains catalytic activity in the crystal such that the products of multiple rounds of replication can be observed using X-ray crystallography.
We have examined the interactions of BF with a panel of purine dNTP analogues to better understand how this enzyme uses N-1, N2, N-3, and N6 to discriminate between right and wrong dNTPs. Remarkably, all 4 nitrogens played major roles in both preventing misincorporation and promoting correct incorporation by BF. Furthermore, addition of N2 to dATP and related purines dNTPs enhanced both correct and incorrect polymerization of the resulting dNTPs.
BF was prepared as previously described (5). All reagents were of the highest quality commercially available. Radiolabeled nucleotides were purchased from Perkin Elmer and unlabeled nucleotides were either purchased from Sigma and Trilink Biotechnologies, or synthesized as previously described (7, 21, 24). Synthetic oligonucleotides were purchased from Biosearch Technologies and their concentrations were determined spectrally. All materials from commercial sources were used without further purification.
All reactions were carried out under steady-state conditions. Assays contained enzyme (2-10 nM, depending upon the analogue), 500nM 5'-32P-labeled primer-template, 50mM tris(hydroxymethyl)aminomethane, HCl salt, pH 8.0, 10mM MgCl2, 1mM dithiothreitol, 0.1mg/ml bovine serum albumin, 2.5% glycerol, and various concentrations of dNTPs or dNTP analogues. Polymerization reactions were initiated by the addition of enzyme and, depending on the ability of BF to polymerize the analogue, the reactions were incubated for 5-20 minutes at 37 °C. Reactions were quenched with a 15uL of gel loading buffer (90% formamide). Control experiments showed that the rates remained constant over the times used. Polymerization products were separated by denaturing gel electrophoresis (20% Acrylamide, 7.5M Urea) and analyzed via phosphorimagery (ImageQuant). Kinetic parameters and errors were determined by fitting the data to the Michaelis-Menten equation as previously described (24).
In order to understand how BF discriminates between right and wrong dNTPs, we examined polymerization of a panel of purine dNTP analogues containing modified bases (Figure 1). The modifications ranged from relatively minor (Eg., loss of N-1, N2, N-3, or N6 and/or conversion to another group) to severe (Eg., loss of N-1, N-3 and N6 from adenine plus addition of a CF3 or NO2 group). Polymerization of both the natural and analogue dNTPs was measured using primer-templates of defined sequence (Table 1). With one exception, the four primer:templates vary only in the identity of the template base being copied, thereby minimizing the possibility of DNA sequence affecting the results. In the case of DNAt, the second single-stranded template base was also changed to eliminate the possibility of consecutive dATP polymerizations. Consistent with the reported high fidelity of BF, the enzyme efficiently discriminated against polymerization of incorrect, natural dNTPs on these primer templates (Table 2). In this, and in all subsequent tables, discrimination opposite a template T or C is defined by how much less efficiently BF polymerized the tested dNTP than it polymerized either dATP (15 μM-1 min-1) or dGTP (3.4 μM-1 min-1), respectively. Opposite a template A or G (i.e., correct TTP and dCTP incorporation), we defined discrimination by comparing (analogue) dNTP incorporation to the average kcat/KM for correct incorporation of dATP and dGTP opposite T and C, respectively (9.2 μM-1 min-1).
Incorporation of a series of hydrophobic purine dNTPs was measured opposite the 4 natural template bases under steady-state conditions (Table 3). In most cases, BF polymerized the analogues opposite a natural template base no better than or only slightly more efficiently than an incorrect, natural dNTP. In a few cases, however, BF incorporated the analogue dNTP significantly faster than a natural, incorrect dNTP (Eg., 6-trifluoromethylbenzimidazole dNTP opposite either A or G).
We then examined the roles of 4 nitrogens in a purine, N-1, N2, N-3, and N6 by systematically varying their presence or absence in purine analogues. Adding N-3 to dZTP (4-methylbenzimidazole dNTP) to generate dQTP (1-deaza-6-methylpurine dNTP), a substitution that replaces an electropositive hydrogen with a free pair of electrons, had only modest effects - at most, the average efficiency of polymerization increased very slightly (Table 3). 2 In contrast, adding N-3 (purine numbering system) to benzimidazole dNTP and thereby generating 1-deazapurine dNTP, dramatically increased the efficiency of incorporation, even though a base pair between any natural base and 1-deazapurine will still lack any Watson-Crick hydrogen bonds and the shape will be distinct from a canonical base-pair (Tables 3 and and4).4). Similarly, adding N-3 to 3-deazapurine dNTP also increased polymerization of the resulting purine dNTP opposite A, T, and C by factors of 2.5, 21, and 8, respectively, while marginally decreasing polymerization opposite G (by 10%, Table 4).
We further probed the role of N-3 by removing it from two canonical, high fidelity nucleotides, dATP and dGTP (Table 4). The resulting dNTPs, 3-deaza-dATP and 3-deaza-dGTP, differed from dATP and dGTP in two significant ways. First, they were polymerized much less efficiently opposite their complementary template nucleotides, T and C, respectively. Second, the efficiency of misincorporation of 3-deaza-dATP and 3-deaza-dGTP often increased substantially compared to their parent compounds. For example, BF misincorporated 3-deaza-dGTP opposite either A or G ≥20-fold more efficiently than it misincorporated dGTP. Curiously, in two cases, polymerization of 3-deaza-dGTP opposite T and 3-deaza-dATP opposite G, removing N-3 decreased misincorporation. Thus, N-3 appears to have two distinct roles, enhancing incorporation and, at least within the context of adenine and guanine, managing misincorporation frequencies.
The role of N2 was examined within the context of both guanine and adenine-related bases (Table 4). Removing N2 from dGTP, thereby generating dITP, both decreased the efficiency of polymerization opposite C and increased misincorporation, particularly opposite A and T. The 10-fold decrease in polymerization efficiency opposite C corresponds to a ΔG of 1.4 kcal mol-1, in the range of what one would expect for a hydrogen bond. Adding N2 to dATP, thereby forming 2-amino-dATP, had unexpectedly large effects during polymerization opposite all 4 natural template bases. BF polymerized 2-amino-dATP opposite a template T slightly more efficiently than it polymerized dATP. More surprisingly, the presence of N2 significantly decreased the ability of BF to identify 2-amino-dATP as wrong opposite A, C, and G. Depending upon the mismatch examined, BF misincorporated adding 2-amino-dATP 37- to 85-fold more efficiently.
In contrast to the large effects upon adding N2 to dATP, adding N2 to purine dNTP and 1-deazapurine dNTP had much milder effects (Table 4). BF polymerized 2-aminopurine dNTP more efficiently opposite all 4 template bases by 3- to 11-fold than purine dNTP. Comparing polymerization of 1-deazapurine dNTP and 2-amino-1-deazapurine dNTP showed that the presence of N2 affected polymerization opposite A, T, and G by <4-fold, and only increased polymerization opposite C by a factor of 8. Thus, the effects of N2 appear very dependent upon the rest of the base.
N6 of adenine normally forms a hydrogen bond with O4 of thymine. Converting dATP into purine dNTP resulted in a 23-fold decreased efficiency of polymerization opposite T (Table 4). Additionally, BF misincorporated purine dNTP much more efficiently opposite the other three template bases. The loss of N6 most severely impacted misincorporation opposite C, where misincorporation increased by a factor of 22.
In a different context, removing N6 from 1-deaza-dATP to generate 1-deazapurine dNTP, the loss of N6 had quite different effects than with dATP. Compared to 1-deaza-dATP, BF incorporated 1-deazapurine dNTP with similar efficiency opposite A, only 2-fold less efficiently opposite T, but 13-and 6-fold less efficiently opposite C and G, respectively (Table 4). Comparing these two sets of data indicate that the effects of N6 vary substantially depending upon the rest of the base.
The role of exocyclic substituents at C-6 was further probed by replacing N6 with either an electron withdrawing Cl or a slightly electron donating CH3 (Table 4). As occurred upon replacing N6 with H, both of these latter two replacements inhibited polymerization opposite T and increased misincorporation opposite A, C, and G. Comparing the effects of replacing N6 shows that Cl and CH3 replacements gave less misincorporation opposite A than replacement with H, while the CH3 gave more misincorporation opposite G and less misincorporation opposite C than replacement with H and Cl.
N-1 normally forms a Watson-Crick hydrogen bond with NH-3 of thymine. Not surprisingly, removing N-1 from either purine dNTP or dATP significantly impairs polymerization opposite T (Table 4). Additionally, BF misincorporates 1-deaza-dATP much more efficiently than dATP, indicating that N-1 plays a role in preventing misincorporation. In contrast, converting purine dNTP into 1-deazapurine dNTP had variable effects on fidelity. It decreased misincorporation opposite C by 9-old, had little effect on misincorporation opposite G (25% increase), and increased misincorporation opposite A by 4-fold. Thus, just like the effects of N2 and N6, the impact of N-1 depends upon the rest of the base.
We examined the roles of N-1, N2, N-3, and N6 for both adenine- and guanine-based dNTPs during polymerization by BF. Removing any one of these nitrogens had two effects - decreased polymerization of the resulting dNTP within the context of a correct base-pair, and enhanced misincorporation. Importantly, the effects of each nitrogen depended greatly on the structure of the rest of the base.
The effects of specific functional groups exhibited a tremendous dependence upon the rest of the base. For example, with most bases examined, the presence of N-3 greatly increased polymerization, but not in the case of converting 4-methylbenzimidazole dNTP into 1-deaza-6-methylpurine dNTP. Analogously, adding N2 to both dATP and purine dNTP increased incorporation, whereas adding N2 to 1-deazapurine dNTP had minimal effects. Or, removing N-1 from dATP increased misincorporation opposite A, C, and G, but removing N-1 from either purine dNTP or 1-deazapurine dNTP inhibited polymerization. Using another A family polymerase, Klenow Fragment, Romesberg et al. have also shown that the effects of modifying a specific functional group can vary depending upon the rest of the base (35). Modifying furo[2,3-c]pyridin-7(6H)-one dNTP into furo[2,3-c]pyridine-7-thiol dNTP versus modifying furo[3,2-c]pyridin-4(5H)-one dNTP into furo[3,2-c]pyridine-4-thiol dNTP (Figure 2), both of which involve converting an oxygen into sulfur, had very different effects on polymerization of the thiol-containing dNTPs, up to >3000-fold in the case of polymerization opposite a template C. This base dependence, in combination with each functional group playing a role in both enhancing correct incorporation and preventing, suggests that BF uses a holistic, integrative approach in its interactions with the base of the incoming dNTP. Rather than using specific chemical features of the base for specific functions, BF “sees” and “interprets” the entire base when deciding whether or not to polymerize an incoming dNTP. This approach differs from the B family enzymes pol α and HSV pol where specific functional groups serve precise purposes independent of the rest of the base (7, 24, 25).
The ability of N-1 (dATP), N2 (dGTP), and N6 (dATP) to drive correct polymerization is consistent with a key role for Watson-Crick hydrogen bonds. Removing N6 from dATP reduced correct incorporation by 23-fold, while the loss of N2 from dGTP reduced correct incorporation by 10-fold. These decreases correspond to losses of 1.9 and 1.4 kcal mol-1 in transition state stabilization, respectively, consistent with the loss of a hydrogen bond. Replacing N-1 of dATP with a CH decreased polymerization opposite T by 680-fold, or 4.0 kcal mol-1. While the much greater decrease in polymerization due to removing N-1 might indicate that the hydrogen bond involving N-1 is stronger than those involving either N6 of adenine or N2 of guanine, it may also result from the replacement of a favorable interaction with an unfavorable interaction. Whereas N-1 of adenine is electron rich and ideally suited to interact with the electron deficient HN-3 of thymine, the hydrogen of HC-1 of 1-deazaadenine will also be electron deficient such that its interactions with HN-3 of thymine may become energetically repulsive.
While the effects of deleting N2 from dGTP were analogous to the effects of losing either N-1 or N6 from dATP, the consequences of adding N2 to dATP were completely unexpected. The slightly enhanced polymerization of 2-amino-dATP opposite T presumably results from the extra hydrogen bond one can form between N2 of the purine dNTP and O2 of T. Consistent with this hypothesis, adding N2 to purine dNTP also resulted in more efficient polymerization opposite T. More remarkably, adding N2 to dATP significantly reduced the capacity of BF to identify the resulting 2-amino-dATP as wrong opposite any template base. Indeed, adding N2 to dATP largely overcomes the negative consequences of incorrect hydrogen bonding groups at N-1 and N6 and/or consequences of altered base-pair shape. While the formation of a hydrogen bond between N2 of 2-amino-dATP and O2 of C could account for this misincorporation event, how N2 increases misincorporation opposite A and G is less clear. Since preventing misincorporation ultimately requires a polymerase to not recognize an incoming dNTP as correct - i.e,, the dNTP does not bind in a polymerizable conformation - the presence of N2 on dATP allows the 2-aminoadenine to bind in the active site in a manner that BF does not effectively recognize as wrong.
Why N-3 significantly alters both correct and incorrect dNTP polymerization remains unclear. Similar to BF, Klenow Fragment also polymerized 3-deaza-dGTP less efficiently than dGTP (10-fold), hence it seems likely that these effects will be general to A family polymerases (36). Curiously, in structures of E-DNA-dNTP closed complexes with various A family polymerases (KlenTaq, T7 DNA polymerase, and BF), N-3 is stacked against a highly conserved Phe (3, 29, 32). Potentially, interactions between the electron rich N-3 and electron deficient edge of this Phe could mediate the kinetic effects. Alternatively, the effects of N-3 might occur via an interaction with the protein that occurs prior to closed complex formation, such as when the template base resides within the pre-insertion site (37).
BF could incorporate a variety of dNTPs containing hydrophobic bases opposite all 4 natural bases. However, it generally discriminated against polymerizing these dNTPs quite strongly, indicating that hydrophobicity alone is not sufficient to drive rapid polymerization of unnatural dNTPs. Using a series of trifluorobenzimidazole dNTPs, all of whose bases will have similar hydrophobicity and size, BF clearly found the 6-trifluorobenzimidazole dNTP to be the most attractive substrate. Indeed, the enzyme polymerized this dNTP opposite a natural template base 4- to 80-fold more efficiently than an incorrect, natural dNTP. This clear preference for the 6-trifluoromethylbenzimidazole dNTP suggests that shape can contribute to base selectivity, although not necessarily in the way that one would predict. However, it should be noted that these bases will have different dipoles, which may also contribute to the enzymes preferences.
An outstanding, and often hotly debated question is the relative importance of base-pair shape versus Watson-Crick hydrogen bonds to DNA polymerase fidelity. Among the A family polymerases, different studies have led to very different conclusions (17, 38, 39). Part of this discrepancy may result from different A family polymerases using different mechanisms. Yet, studies have clearly shown that Klenow Fragment, the paradigm for polymerases using shape, can efficiently polymerize dNTP analogues whose base cannot form a correctly shaped base-pair with the template base being replicated (7, 21-23). These apparently conflicting observations can be reconciled if the extent to which the polymerase requires Watson-Crick hydrogen bonds and/or correct shape depends upon the overall chemistry and structure of the base. As noted earlier, the effects of Watson-Crick hydrogen bonding groups vary depending upon the base. Additionally, Waksman and colleagues showed that the A family polymerase KlenTaq replicates the 4 natural template bases with different dynamics, and that the structures of the four correct E-DNA-dNTP ternary complexes vary depending upon the template base (29, 30). Both of these results again suggest that the identity of the template base affects kinetic and/or mechanistic details of the reaction.
Different families of polymerases use at least 3 different mechanisms for discriminating between right and wrong dNTPs. Two very low fidelity RNA polymerases, human and herpes DNA primase appear to only incorporate efficiently those NTPs that can form Watson-Crick hydrogen bonds with the template base (12, 13). Likewise, two low fidelity DNA polymerases, pol η and DNA pol IV from Sulfolobus solfataricus, also may require formation of Watson-Crick hydrogen bonds to efficiently incorporate a dNTP (14, 15). Thus, direct employment of Watson-Crick hydrogen bonds may be a general strategy of low fidelity polymerases. Two B family polymerases, pol α and herpes DNA polymerase, use a combination of positive and negative selectivity to identify incoming dNTPs as right or wrong (7, 24, 25). Unlike the low fidelity enzymes, they often efficiently incorporate dNTPs whose bases lack Watson-Crick hydrogen bonding groups. These polymerases do not require N-3 of purine dNTPs for correct incorporation, but do use N-3 to prevent misincorporation. N-1 both helps prevent misincorporation and enhances correct incorporation, while N6 plays a lesser role. N2 plays a minor role during correct polymerization of dGTP, but is critical for minimizing polymerization opposite A. Since B-family polymerases have well conserved active sites (10, 40, 41), these general observations will probably apply to most B-family enzymes.
BF polymerase, a prototypical A family polymerase, appears distinct from these two groups of enzymes. Similar to the B family enzymes, BF uses N-1, N2, and N6 to both increase correct incorporation and prevent misincorporation - i.e., enzymes from both families use a combination of positive and negative selectivity. The similarity, however, ends there. While BF occasionally incorporates dNTPs containing hydrophobic bases much more rapidly than an incorrect natural dNTP, it more commonly incorporates hydrophobic dNTP analogues no better than a natural, incorrect dNTP. Klenow Fragment, another A family polymerase, also usually, but not always, polymerizes hydrophobic dNTP analogues less efficiently than the B family enzymes, with the key exception of 2,4-dihalotoluenes opposite A (7, 21). For both Klenow Fragment and BF, N-3 is critical for rapid correct polymerization of dGTP and dATP ((36) and vide infra), whereas B family polymerases primarily use N-3 to prevent misincorporation. Unlike B family polymerases, the effects of N-1, N2, N-3, and N6 depend heavily on the rest of the base. Additionally, these 4 nitrogens prevent or enhance different sets of mismatches for the different enzymes. Most notably, some B family polymerases specifically misincorporate 2-aminopurine dNTPs opposite C, likely due to formation of a hydrogen bond between the purine N2 and O2 of C (42).
A key unanswered question is what chemical features of the active sites from polymerases allow different families of enzymes to use different mechanisms to accomplish the common task of discriminating between right and wrong dNTPs? Additionally, how do different polymerases from the same family tune their active sites for different levels of fidelity? Besides providing fundamental insights into polymerase function, answering this question could lead to the development of novel polymerase inhibitors, a medically extremely important class of compounds.
This work was supported by a grant from the NIH (GM54194) to RDK.
1Abbreviations used: 2-Aminoadenine-2'-deoxyriboside triphosphate, 2-aminoadenine dNTP, d2AATP; 2-Amino-1-deazapurine-2'-deoxyriboside triphosphate, 2-amino-1-deazapurine dNTP, d2A1DPTP; 2-Aminopurine-2'-deoxyriboside triphosphate, 2-aminopurine; dNTP, d2APTP Benzimidazole-2'-deoxyriboside triphosphate, benzimidazole dNTP, dBTP; BF, Large fragment of DNA polymerase I from Bacillus stearothermophilus; 6-Chloropurine-2'-deoxyriboside triphosphate, 6-chloropurine dNTP, d6ClPTP; 1-Deazaadenine-2'-deoxyriboside triphosphate, 1-deazaadenine dNTP, d1DATP; 3-Deazaadenine-2'-deoxyriboside triphosphate, 3-deazaadenine dNTP, d3DATP; 3-Deazaguanine-2'-deoxyriboside triphosphate, 3-deazaguanine dNTP, d3DGTP; 1-Deazapurine-2'-deoxyriboside triphosphate, 1-deazapurine dNTP, d1DPTP; 3-Deazapurine-2'-deoxyriboside triphosphate, 3-deazapurine dNTP, d3DPTP; DNA polymerase α, pol α; DNA polymerase η, pol η; 4-Methylbenzimidazole-2'-deoxyriboside triphosphate, 4-methylbenzimidazole dNTP, dZTP; 6-Methyl-1-deazapurine-2'-deoxyriboside triphosphate, 6-methyl-1-deazapurine dNTP, dQTP; 6-Methylpurine-2'-deoxyriboside triphosphate, d6MePTP; 5-Nitrobenzimidazole-2'-deoxyriboside triphosphate, d5NO2BTP; 6-Nitrobenzimidazole-2'-deoxyriboside triphosphate, d6NO2BTP; Purine-2'-deoxyriboside triphosphate, purine dNTP, dPTP; 4-Trifluoromethylbenzimidazole-2'-deoxyriboside triphosphate, d4CF3BTP; 6-Trifluoromethylbenzimidazole-2'-deoxyriboside triphosphate, 6-trifluoromethylbenzimidazole dNTP, d6CF3 BTP; 7-Trifluoromethylbenzimidazole-2'-deoxyriboside triphosphate, d7CF3BTP.
2In these studies, we have only compared the efficiency of polymerization of each analogue (VMAX/KM). BF is a moderately processive enzyme, hence DNA dissociation limits the steady-state turnover rate after incorporation of a correct dNTP. Since we do not know the rate-limiting step during polymerization of the analogues, the VMAX values for polymerization of different dNTPs cannot be directly compared. Likewise, KM and KD cannot safely be compared due to the potential for different rate limiting steps.