The Molecular Perspective: Methotrexate
Chemotherapy is the art of selective poisoning. Many of the poisons used in modern chemotherapy rely on the observation that most cells in the body are not actively dividing, so substances that block some step in cell reproduction will have a disproportionately severe effect on growing cancer cells relative to most healthy tissues. Virtually every step of cell growth and division is targeted by some chemotherapeutic drug today.
The enzyme dihydrofolate reductase (DHFR), shown in Figure 1, was the first enzyme to be targeted by chemotherapy. It plays a supporting role, but an essential role, in the synthesis of thymine nucleotides. The leading role is played by the enzyme thymidylate synthase, which adds a methyl group onto uracil to form thymine. This may seem like a small change, but the extra hydrophobic bulk of this methyl group is essential for proper discrimination of thymine from the other three bases by various transcription factors, repressors, enhancers and other DNA-binding proteins. The methyl group is delivered to thymidylate synthase by folate, a small cofactor molecule that is an essential vitamin for human beings. In the course of the reaction, as the methyl is transferred from folate to uracil, the folate molecule is oxidized. This is where DHFR makes its entry: DHFR restores folate to its reduced state, ready for the next round of thymine synthesis.
Dihydrofolate reductase. DHFR is a small enzyme with a large active site. Folate, shown in green, binds adjacent to the cofactor NADPH, shown in orange. The reaction occurs where they touch at the center, in a pocket buried deep within the enzyme. The human enzyme is illustrated, using coordinates from PDB entry 1DLS. DHFR enzymes from other organisms are similar in size and shape, with large active site grooves to bind folates and NADPH. But small species-specific differences may be targeted by antibiotic drugs such as trimethoprim, which blocks bacterial DHFR more effectively than human DHFR. Methotrexate, on the other hand, binds strongly to DHFR from many species, and thus is useful as an antineoplastic agent but not as an antibiotic.
If the action of DHFR is blocked, the cell dies. When treated with methotrexate, the level of reduced folate plummets, and synthesis of thymine and purine nucleotides grinds to a halt. (There is evidence that methotrexate also blocks other steps in nucleotide synthesis, magnifying these effects.) As the available thymine nucleotides are depleted, the normally low levels of uracil nucleotides grow, because thymidylate synthase cannot perform its job. Uracil is then erroneously added to growing DNA chains in place of thymine. Ultimately, this halts DNA synthesis, and promotes fragmentation of the DNA as repair enzymes unsuccessfully attempt to remove the many faulty nucleotides.
Aminopterin, a slightly modified version of methotrexate, was the first drug administered to inhibit DHFR. As with many drugs, it mimics the natural substrate. Both aminopterin and methotrexate closely resemble folate, but bind a thousand times more strongly to the active site, effectively blocking the normal binding of folate. Comparing these molecules to folate, we would expect that they would bind similarly, with methotrexate utilizing all of the features of recognition that have been optimized over evolution of the enzyme. When crystallographic structures of DHFR with methotrexate and with folate were obtained, however, a surprise was found. The environment of the active site was nearly identical in both, but the terminal pteridine ring of methotrexate was flipped relative to that of folate, as shown in Figure 2.
Methotrexate and folate in the active site. These illustrations show a cross-section through the DHFR active site. The view is roughly perpendicular to Figure 1, and the half of the enzyme that binds to NADPH has been removed. The enzyme is shown as spacefilling spheres, and folate (left) and methotrexate (right) are shown in ball-and-stick representation, with carbon atoms in green. The enzyme reduces the terminal pteridine group of folate (the two fused six-membered rings), seen here at the center. Notice the alignment of the two blue nitrogen atoms in folate with the two red oxygen atoms in the enzyme. These two hydrogen bonds, along with a series of interactions with bridging water molecules (not shown), are the basis of recognition. The folate ring contains a carbonyl oxygen (pointing up in red) which is replaced by an amino group in methotrexate (in blue). As a result, the methotrexate ring binds in a flipped orientation, as seen on the left. Again, two nitrogen atoms on the drug pair with the two oxygen atoms on the protein, forming a strong, stable complex, but they are a different pair compared with the two nitrogen atoms used in folate. PDB entries 1DLS and 1DHF were used for the illustrations.
Methotrexate and folate in the active site. These illustrations show a cross-section through the DHFR active site. The view is roughly perpendicular to Figure 1, and the half of the enzyme that binds to NADPH has been removed. The enzyme is shown as spacefilling spheres, and folate (left) and methotrexate (right) are shown in ball-and-stick representation, with carbon atoms in green. The enzyme reduces the terminal pteridine group of folate (the two fused six-membered rings), seen here at the center. Notice the alignment of the two blue nitrogen atoms in folate with the two red oxygen atoms in the enzyme. These two hydrogen bonds, along with a series of interactions with bridging water molecules (not shown), are the basis of recognition. The folate ring contains a carbonyl oxygen (pointing up in red) which is replaced by an amino group in methotrexate (in blue). As a result, the methotrexate ring binds in a flipped orientation, as seen on the left. Again, two nitrogen atoms on the drug pair with the two oxygen atoms on the protein, forming a strong, stable complex, but they are a different pair compared with the two nitrogen atoms used in folate. PDB entries 1DLS and 1DHF were used for the illustrations.
We often think, given the plethora of structural and genomic data, that we understand enzyme action, even to the level of prediction and design. This is pure molecular hubris: little surprises, as additional data are collected, continually challenge our knowledge. We see this again and again. We may reason that a small change in a natural substrate, such as the change of a folate carboxyl to a methotrexate amino, will create a new, better inhibitor. Surprisingly often, however, when this slightly modified molecule is tested, it binds in an unexpected manner. And then, against all expectations, the new orientation may be even better than the one we expected! Biological molecules are subtle—we still have far to go in understanding.
Additional Reading
Davies JF, Delcamp TJ, Prendergast NJ et al. Crystal structures of recombinant human dihydrofolate reductase complexed with folate and 5-deazafolate. Biochem 1990;29:9467-9479.
Allegra CJ, Grem JL. Antimetabolites in Cancer, Principles and Practice of Oncology. Lippincott-Raven Publishers: Philadelphia, 1997;1:432-452.
