Models for two ketoreductases were created and used to predict the stereoselectivity of the enzymes. One was based on the crystal structure of
In 1999, 33% of all dosage-form drug sales in the USA were of single enantiomers [
Recently Zhu et al. thoroughly studied a carbonyl reductase from red yeast (
There are two main goals of this work. One was to use the X-ray crystal structure of SSCR to build a model that could predict most of the stereoselectivity seen for this enzyme in the literature. As SSCR is a highly promiscuous enzyme that often results in high stereoselectivity, elucidating its behavior can increase its value as a tool for asymmetric synthesis. The second goal is to use a homology model for YOL151w to predict the stereoselectivity of this enzyme for new substrates of interest. In this paper we discuss the modeling work on SSCR and the homology model for YOL151w built from the X-ray structure of SSCR.
The 3D X-ray structure of SSCR was obtained from the Protein Data Bank (ID: 1Y1P). There are two molecules in the asymmetric unit of the crystal structure of SSCR, both nearly identical in geometry; as a result only subunit B (following the procedure laid out by Cundari et al.) [
Substrates were drawn and minimized using the Hartree-Fock method with a 6-31
The homology model of YOL151w was built based on the enzyme crystal structure of
In the analysis of the SSCR enzyme, substrates were grouped into four classes based on the functional groups they possess. Examination of the lowest energy structure that met the docking criteria (see above) was performed to determine the stereochemistry that would result from the docking; this was compared to the experimental literature results. Stereochemistry was determined from docking geometry by examining the orientation of the carbonyl group in relation to the hydride source on the cofactor. The docked geometry allowed for determination of the face (Re/Si) that would be attacked by the hydride on the cofactor and thus for prediction of the stereochemistry of the product (see Scheme
Mechanism of stereoselectivity for NAD(P)H-dependent ketoreductase (1Y1P) and homology model of YOL151w. Group priorities are based on Cahn-Ingold-Prelog rules and assumed in the schemeto be OH > R1 > R2. B represents any residue capable of donating a hydrogen atom.
Data from the docking of 26 aryl ketones (ArKs) are shown in Table
Aryl ketones (ArK).
|
||||||||
---|---|---|---|---|---|---|---|---|
ID name | Compound name | R1 | R2 |
|
|
ee (%) | Prelog | Prediction correct |
|
||||||||
ArK1* | Acetophenone | H | CH3 | −39.9 | −37.8 | 42 (R) | Anti | Y |
ArK2* | Propiophenone | H | CH2CH3 | −40.6 | −38.1 | 28 (R) | Anti | Y |
ArK3* | 1-Phenylbutan-1-one | H | (CH2)2CH3 | −44.2 | −44.4 | 88 (S) | Prelog | Y |
ArK4* | 1-Phenylpentan-1-one | H | (CH2)3CH3 | −45.5 | −47.3 | 87 (S) | Prelog | Y |
ArK5* | 1-Phenylhexan-1-one | H | (CH2)4CH3 | −47.0 | −50.7 | 34 (S) | Anti | Y |
ArK6* | 1-Phenylheptan-1-one | H | (CH2)5CH3 | −49.7 | −53.1 | 27 (S) | Anti | Y |
ArK7* | 2-Methyl-1-phenylpropan-1-one | H | CH(CH3)2 | −44.4 | −41.5 | 98 (R) | Anti | Y |
ArK8* | 2,2-Dimethyl-1-phenylpropan-1-one | H | C(CH3)3 | −47.6 | NS | 98 (R) | Anti | Y |
ArK9† | 2-Chloro-1-phenylethanone | H | CH2Cl | −39.9 | −45.3 | 98 (S) | Anti | Y |
ArK10* | Cyclopropyl(phenyl)methanone | H |
|
−46.2 | −41.5 | 96 (R) | Anti | Y |
ArK11* | Cyclopropyl(4-fluorophenyl)methanone | 4′-F |
|
−49.8 | NS | 98 (R) | Anti | Y |
ArK12* | 4-Chlorophenyl(cyclopropyl)methanone | 4′-Cl |
|
−52.2 | −33.7 | 98 (R) | Anti | Y |
ArK13† | 1-(4-Fluorophenyl)ethanone | 4′-F | CH3 | −40.4 | −36.2 | 46 (R) | Anti | Y |
ArK14† | 1-(4-Chlorophenyl)ethanone | 4′-Cl | CH3 | −44.7 | NS | 14 (R) | Anti | Y |
ArK15† | 1-(4-Bromophenyl)ethanone | 4′-Br | CH3 | −44.6 | NS | 42 (R) | Anti | Y |
ArK16† | 1- |
4′-CH3 | CH3 | −42.7 | NS | 59 (R) | Anti | Y |
ArK17† | 1-(4-Methoxyphenyl)ethanone | 4′-OCH3 | CH3 | −42.9 | NS | 57 (R) | Anti | Y |
ArK18† | 1-(4-(Trifluoromethyl)phenyl)ethanone | 4′-CF3 | CH3 | −46.7 | NS | 17 (S) | Prelog | N |
ArK19† | 1-(2-chlorophenyl)ethanone | 2′-Cl | CH3 | −42.6 | −37.6 | 15 (R) | Anti | Y |
ArK20† | 1- |
2′-CH3 | CH3 | −43.3 | −41.3 | 70 (R) | Anti | Y |
ArK21† | 1-(2-Methoxyphenyl)ethanone | 2′-OCH3 | CH3 | −49.6 | −49.5 | 99 (R) | Anti | Y |
ArK22† | 1-(3-Chlorophenyl)ethanone | 3′-Cl | CH3 | −40.5 | −40.8 | 66 (R) | Anti | N |
ArK23† | 1- |
3′-CH3 | CH3 | −42.3 | NS | 92 (R) | Anti | Y |
ArK24† | 1-(3,5-Bis(trifluoromethyl)phenyl)ethanone | 3′,5′-(CF3)2 | CH3 | −45.2 | NS | 99 (R) | Anti | Y |
ArK25† | 1-Tetralone | −42.4 | −37.7 | 94 (R) | Prelog | Y | ||
ArK26† | 6-Methyl-4-chromanone | −46.2 | −45.7 | 99 (R) | Prelog | Y |
NS = no structure found meeting the requirements.
Aliphatic ketones (ApK).
|
||||||||
|
||||||||
ID name | Compound name | R1 | R2 |
|
|
ee (%) | Prelog | Prediction correct |
|
||||||||
ApK1 | Heptan-2-one | n-Pentyl | CH3 | −43.3 | −41.2 | 30 (S) | Prelog | N |
ApK2 | Octan-2-one | n-Hexyl | CH3 | −46.8 | NS | 44 (S) | Prelog | N |
ApK3 | Nonan-2-one | n-Heptyl | CH3 | −47.3 | NS | 4 (R) | Anti | Y |
ApK4 | 1-Adamatyl methyl ketone | 1-Adamantyl | CH3 | −45.1 | −46.9 | >99 (S) | Prelog | Y |
ApK5 | Octane-3-one | n-Pentyl | CH2CH3 | −45.1 | −42.5 | 72 (R) | Prelog | Y |
NS = no structure found meeting the requirements.
Alpha-ketoesters (AKE).
|
|||||||
|
|||||||
ID name | Compound name | R1 |
|
|
ee (%) | Prelog | Prediction correct |
|
|||||||
AKE1 | Ethyl 2-oxo-2-phenylacetate | Phenyl | −51.4 | −51.7 | 99 (S) | Anti | Y |
AKE2 | Ethyl 2-(4-cyanophenyl)-2-oxoacetate | 4-Cyanophenyl | −54.4 | −64.7 | 82 (S) | Anti | Y |
AKE3 | Ethyl 2-(4-fluorophenyl)-2-oxoacetate | 4-Fluorophenyl | −53.8 | −39.6 | 74 (S) | Anti | N |
AKE4 | Ethyl 2-(4-chlorophenyl)-2-oxoacetate | 4-Chlorophenyl | NS | −47.9 | 63 (S) | Anti | Y |
AKE5 | Ethyl 2-(4-bromophenyl)-2-oxoacetate | 4-Bromophenyl | −52.2 | −55.9 | 56 (S) | Anti | Y |
AKE6 | Ethyl 2-oxo-2- |
4-Methylphenyl | −53.1 | −55.0 | 88 (S) | Anti | Y |
AKE7 | Ethyl 2-(3,5-difluorophenyl)-2-oxoacetate | 3,5-Difluorophenyl | NS | −41.7 | 43 (S) | Anti | Y |
AKE8 | Ethyl 4-methyl-2-oxopentanoate | Isopropyl | −50.2 | −44.7 | 99 (R) | Prelog | Y |
AKE9 | Ethyl 4,4-dimethyl-2-oxopentanoate |
|
−54.5 | −46.4 | 99 (R) | Prelog | Y |
NS = no structure found meeting the requirements.
Beta-ketoesters (BKE).
|
|||||||
|
|||||||
ID name | Compound name | R |
|
|
ee (%) | Prelog | Prediction correct |
|
|||||||
BKE1 | Ethyl 4-chloro-3-oxobutanoate | Chloromethyl | −46.8 | −61.2 | 97 (S) | Anti | Y |
BKE2 | Ethyl 3-oxopentanoate | Ethyl | −50.2 | −50.3 | 61 (R) | Anti | N |
BKE3 | Ethyl 4-methyl-3-oxopentanoate | Isopropyl | −52.0 | −63.9 | 99 (S) | Anti | Y |
BKE4 | Ethyl 4,4-dimethyl-3-oxopentanoate |
|
−47.6 | −59.5 | 99 (S) | Prelog | Y |
BKE5 | Ethyl 4,4,4-trifluoro-3-oxopentanoate | Trifluoromethyl | −43.1 | −58.4 | 90 (S) | Anti | Y |
BKE6 | Ethyl 3-oxo-3-phenylpropanoate | Phenyl | −54.4 | −57.8 | 56 (S) | Prelog | Y |
NS = no structure found meeting the requirements.
Lowest energy complex of SSCR (1Y1P) and ArK1. ArK1 is colored green, and the hydrogen on NADPH involved with reduction is shown as a ball.
The stereoselectivity of carbonyl reductases can often be predicted by Prelog’s rule, which states that the stereochemistry can be determined by looking at the size of the two R groups. This rule states that the enzyme has a large and small pocket that makes up the active site in which the substrate binds and controls the stereochemistry of the product based on the geometry of the substrate. The SSCR enzyme seems to follow the anti-Prelog rule (as was noted by others [
Asymmetric reduction of ketones according to the anti-Prelog rule for discrimination of the faces of carbonylic groups by the enzymes. Note: R1 is larger in size than R2.
This enzyme displays the anti-Prelog rule 21 out of 26 times for the ArKs (Table
In the docking simulations of the ArKs with SSCR, the carbonyl oxygen on the ArKs participates in hydrogen bonding with SER133 and TYR177 for all of the complexes. The energies generally show that the lowest energy geometry leads to the observed stereochemistry. The docking model predicts the wrong stereochemistry for ArK18 and ArK22. Both ArK18 and ArK22 have halogen atoms present and the docking model may be poorly reproducing the interaction that is occurring with the halogen atoms. The correlation between enantiomeric excess and energy difference seen in the model is 0.62, and there is an even better correlation between lowest energy conformation and major enantiomer seen experimentally; the correlation value is 0.74. Using the lowest energy docked geometry, more of the stereochemistry of the products was correctly predicted than that obtained from simply using the Prelog rule (noting that this enzyme displays anti-Prelog behavior).
While SSCR has anti-Prelog rule behavior with ArKs, it is interesting to note that it has Prelog rule stereoselectivity with the ApK substrates (Table
In the results for the AKEs (Table
The BKEs showed primarily anti-Prelog behavior with two exceptions: BKE4 and BKE6. Again, with these molecules the sizes of the two R-groups on each side of the carbonyl are approximately the same. For the model predictions, only the lowest-energy docked structure for BKE2 predicted the incorrect stereoselectivity. The energies between the pro-R and pro-S docked structure were very close, but there is no clear reason why the docked simulation did not prefer one over the other.
Overall 46 compounds were docked and the predicted stereochemistry was compared to the literature values. Using the lowest energy geometries that are also capable of undergoing reaction (ones whose geometry had the carbonyl group close enough to the hydride source and was close enough to hydrogen bond to two of the catalytic residues), only 6 were incorrectly predicted compared to 13 if the enzyme is assumed to prefer an anti-Prelog docking geometry. Half of the incorrectly predicted stereochemistries were on compounds containing halogen atoms, which may indicate a weakness in the model for highly electronegative atoms.
As with the modeling of SSCR, substrates were grouped into classes based on their functionality and docked in the homology model of YOL151w. Examination of the lowest-energy structure in which the docking criteria were met was performed to determine what stereochemistry would result from the docking. This predicted stereoselectivity was then compared to the experimental literature results.
The first group of substrates contained seven
Beta-ketonitrile (BKN).
|
|||||||
|
|||||||
ID name | Compound name | R |
|
|
ee (%) | Prelog | Prediction correct |
|
|||||||
BKN1 | 5-Methyl-3-oxohexanenitrile | Isobutyl | −46.2 | NS | 99 (R) | Prelog | Y |
BKN2 | 3-Cyclohexyl-3-oxopropanenitrile | Hexyl | NS | −38.2 | 99 (S) | Prelog | Y |
BKN3 | 3-Oxo-3-phenylpropanenitrile | Phenyl | NS | −48.1 | 99 (S) | Prelog | Y |
BKN4 | 3-(4-Fluorophenyl)-3-oxopropanenitrile | 4-fluorophenyl | NS | −62.5 | 99 (S) | Prelog | Y |
BKN5 | 3-(4-Chlorophenyl)-3-oxopropanenitrile | 4-chlorophenyl | NS | −41.3 | 78 (S) | Prelog | Y |
BKN6 | 3-(4-Methoxyphenyl)-3-oxopropanenitrile | 4-methoxyphenyl | −43.6 | NS | 99 (S) | Prelog | N |
BKN7 | Methyl 4-(2-cyanoacetyl)benzoate | 2-cyanoacetyl | NS | NS | 74 (S) | Prelog | N |
NS = no structure found meeting the requirements.
Data from the docking of AKE are shown in Table
Alpha-ketoesters (AKE).
|
|||||||
|
|||||||
ID name | Compound name | R |
|
|
ee (%) | Prelog | Prediction correct |
|
|||||||
AKE10 | Ethyl 2-oxobutanoate | Ethyl | −54.4 | −42.3 | 52 (S) | Prelog | N |
AKE11 | Ethyl 2-oxopentanoate | n-Propyl | −46.5 | −46.2 | 98 (R) | Anti | Y |
AKE12 | Ethyl 2-oxo-4-phenylbutanoate | PhCH2CH2 | NS | −56.3 | 98 (S) | Anti | Y |
NS = no structure found meeting the requirements.
Beta-ketoesters (BKE).
|
|||||||
|
|||||||
ID name | Compound name | R |
|
|
ee (%) | Prelog | Prediction correct |
|
|||||||
BKE1 | Ethyl 4-chloro-3-oxobutanoate | Chloromethyl | −46.3 | −34.8 | 98 (R) | Prelog | Y |
BKE2 | Ethyl 3-oxopentanoate | Ethyl | −55.3 | −47.3 | 98 (S) | Prelog | N |
BKE7 | Ethyl 3-oxobutanoate | Methyl | −51.2 | −52.0 | 98 (S) | Prelog | Y |
BKE8 | Ethyl 3-oxohexanoate | n-Propyl | −56.5 | −47.9 | 98 (S) | Prelog | N |
NS = no structure found meeting the requirements.
Overall for the homology model, 14 compounds were docked and the predicted stereochemistry was compared to the literature values. Using the lowest energy geometries that are also capable of undergoing reaction, 5 were incorrectly predicted. The correlation between enantiomeric excess and energy difference seen in the model is 0.01, and there is a correlation between the lowest energy conformation and major enantiomer seen experimentally; the correlation values is 0.45. This is about twice the failure rate of the model based on the crystal structure of SSCR and is an indication that the homology model is not as reliable as using a known structure. As a result, the determination of the crystal structure of YOL151w would be a significant advancement for modeling this highly promiscuous and synthetically useful enzyme.
Two KRED computational models were developed and used to predict the enzyme (SSCR and YOL151w) stereoselectivity for a variety of substrates. For SSCR the crystal structure (PDB ID: 1Y1P) was used to develop the model used in the docking studies. This model proved adequate for predicting the stereochemistry of docked substrates, especially for nonhalogen containing substrates. While predicting the major enantiomer was generally successful, the model could not predict the enantiomeric excess. The second model was a homology model for YOL151w that was based on the crystal structure of the related enzyme (SSCR, 1Y1P). This model was less successful at predicting the stereochemistry resulting from the reduction of carbonyl groups in the enzyme. This is not surprising as building the homology model adds another opportunity for deviations between reality and the model to occur. We plan to build the model for YOL151w from the X-ray structure when it becomes available (attempts are currently being made to obtain the structure).
The authors declare that there is no conflict of interests regarding the publication of this paper.
This work was supported by NSF-RUI Grant CHE-0848708 from the Organic and Macromolecular Chemistry Program. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors’ and do not necessarily reflect the views of the National Science Foundation.