High resolution x-ray and neutron crystallographic studies of Escherichia coli dihydrofolate reductase

by (Brad Cooper), 1976 -

Dihydrofolate Reductases (DHFRs) have been identified in nearly every proteome and are essential for most biosynthetic pathways involving one-carbon transfer reactions due to their recycling of tetrahydrofolate (THF). They catalyze the NADPH-dependent reduction of dihydrofolate (DHF), producing THF. Inhibition of DHFR ultimately depletes cellular pools of THF; causing a reduced supply of thymine nucleotides for DNA synthesis, resulting in genomic instability and cell death. Therefore, DHFRs remain important drug targets in antimicrobial and chemotherapeutic treatments. Despite exhaustive investigation of E. coli chromosomal DHFR, controversy persists over the dynamics of regulatory loops (the Met20, the ?F-?G, and the ?G-?H) and the nature of the interaction between methotrexate (MTX), a tight-binding anti-cancer drug, and Asp 27, the only ionizable residue in the active site. Also of importance is the ionization state of Asp 27 in the apoenzyme and other complexes. Hydrogen atoms (H) likely play a critical role in DHFR ligand binding and catalysis, yet are difficult to directly visualize. High resolution X-ray and neutron crystallography have been utilized in this dissertation to provide accurate positions of H within the DHFR active site and to probe dynamics of the enzyme. The ultrahigh resolution X-ray structures of DHFR/MTX (1.0Å; chapter 4), apo DHFR (1.05Å), and DHFR/MTX/NADPH (1.4Å; both chapter 5) have been solved. Novel features were observed in the electron density maps, including the ability to model the Met20 loop in the apoenzyme as closed (reported disordered previously) and alternate side chain conformations in all the structures. The high data-to-parameter ratio of the apoenzyme and the MTX data sets allowed anisotropic B-factor refinement and fullmatrix refinement to calculate carboxylate bond lengths and estimates of their deviations. iv The apoenzyme has highly different bond lengths for its Asp 27 carboxylate, thus, it is neutral at physiological pH. The carboxylate bond lengths of the Asp 27 in both the monomers of the asymmetric unit of the DHFR/MTX crystal are nearly equal, suggesting it is charged at physiological pH. If H is substituted for deuterium (D), neutrons are especially powerful probes due to D’s strong positive scattering length. To assign protonation states to the MTX and the Asp 27 by the direct identification of D, a neutron structure has been solved to 2.2Å resolution from nearly 80% complete data collected on a 0.3mm 3 crystal (chapter 4). Prerequisite to the neutron experiment was the growth and D2O-soaking of large-volume crystals (chapter 3). The DHFR/MTX cocrystal possesses the largest primitive unit cell and is the smallest D2O-soaked crystal used successfully in a neutron diffraction experiment. This is the 11th novel protein ever to be solved by neutron crystallography (the 16th total structure). Nearly 2/3 of the amide backbone has undergone H/D exchange, an indicator of protein dynamics. However, monomer B, where the Met20 loop is closed, is ~10% more exchanged than monomer A, where the Met20 loop is partially occluded. Based on results from D occupancy refinement and analysis of the neutron maps, it is concluded that the MTX N1 is protonated when bound to DHFR. Paired with the X-ray data, this is new strong evidence that the Asp 27•MTX interaction is ionic in nature. To increase the signal-to-noise ratio in future neutron experiments, perdeuterated protein has been produced and its D enrichment measured by mass spectrometry. X-ray data (to 1.2Å) has now been collected on a perdeuterated DHFR/MTX cocrystal and it is isomorphous to the native cocrystals (chapter 3). v