The HMGRs of different organisms are multimers of a species-specific number of identical monomers. High-resolution crystal structures have been solved for the Class I human enzyme (HMGRH
] and for the Class II enzyme of Pseudomonas mevalonii
], including protein forms bound to either the HMG-CoA substrate or the coenzyme (NADH or NADPH) or both, or bound to statin drugs, which are potent competitive inhibitors of HMGR activity and thus lower cholesterol levels in the blood [8
]. As reviewed in detail by Istvan [10
], structural comparisons reveal both similarities and significant differences between the two classes of enzyme. The human HMGR has three major domains (catalytic, linker and anchor), whereas the P. mevalonii
HMGR has only the catalytic domain (Figure ).
Both HMGRH and HMGRP have a dimeric active site with residues contributed by each monomer, and a non-Rossman-type coenzyme-binding site (a three-dimensional structural fold that contains a nucleotide-binding motif and is found in many enzymes that use the dinucleotides NADH and NADPH for catalysis). The core regions containing the catalytic domains of the two enzymes have similar folds. Despite differences in amino-acid sequence and overall architecture, functionally similar residues participate in the binding of coenzyme A by the two enzymes, and the position and orientation of four key catalytic residues (glutamate, lysine, aspartate and histidine) is conserved in both classes of HMGR.
Unlike the central cores, the amino- and carboxy-terminal regions of the catalytic domains show little similarity between the human and P. mevalonii HMGR structures. The active site of HMG-CoA reductase is at the interface of the homodimer between one monomer that binds the nicotinamide dinucleotide and a second monomer that binds the HMG-CoA. In human HMGR, the catalytic lysine is found on the monomer that binds the HMG-CoA and comes from the so-called cis-loop (a section that connects the HMG-CoA-binding region with the NADPH-binding region). In contrast, the P. mevalonii HMGR lacks the cis-loop and the catalytic lysine is contributed by the monomer that binds the nicotinamide dinucleotide. HMGRP crystallizes as a trimer of dimers (which are composed of identical subunits), but HMGRH crystallizes as a tetramer (of identical units). HMGRP uses NADH as a coenzyme, whereas HMGRH uses NADPH, but mutation to alanine of the aspartyl residue of HMGRP that normally blocks binding of NADPH can allow NADPH to serve - albeit poorly - as the coenzyme for HMGRP. A 180° difference in the orientation of the nicotinamide ring of the coenzyme suggests that that the stereospecificity of the HMGRH hydrogen transfer is opposite to that of HMGRP.
Comparisons between the HMGRP
structures reveal an overall similarity in how they bind statins, which inhibit activity by blocking access of HMG-CoA to the active site. There is a considerable difference in specific interactions with inhibitor between the two enzymes, however [8
], accounting for the almost 104
-fold higher Ki
values for inhibition of HMGRP
by statin relative to the inhibition of HMGRH
is the equilibrium constant for an inhibitor binding to an enzyme). There are significant differences in the regions of the two proteins that bind statins. In both enzymes the portion of the statin that resembles HMG (see Figure ) occupies the HMG portion of the HMG-CoA-binding pocket, and the non-polar region partially occupies a portion of the coenzyme-A-binding site. For HMGRP
, this impairs closure over the active site of the 'tail' domain that contains the catalytic histidine.
Structures of lovastatin, a statin drug that competitively inhibits HMGR, and of HMG-CoA. It can be seen that the portion of the drug shown here at the top resembles the HMG portion of HMG-CoA.