Structural basis of methotrexate and pemetrexed action on serine hydroxymethyltransferases revealed using plant models

Abstract

Serine hydroxymethyltransferases (SHMTs) reversibly transform serine into glycine in a reaction accompanied with conversion of tetrahydrofolate (THF) into 5,10-methylene-THF (5,10-meTHF). In vivo, 5,10-meTHF is the main carrier of one-carbon (1C) units, which are utilized for nucleotide biosynthesis and other processes crucial for every living cell, but hyperactivated in overproliferating cells (e.g. cancer tissues). SHMTs are emerging as a promising target for development of new drugs because it appears possible to inhibit growth of cancer cells by cutting off the supply of 5,10-meTHF. Methotrexate (MTX) and pemetrexed (PTX) are two examples of antifolates that have cured many patients over the years but target different enzymes from the folate cycle (mainly dihydrofolate reductase and thymidylate synthase, respectively). Here we show crystal structures of MTX and PTX bound to plant SHMT isozymes from cytosol and mitochondria-human isozymes exist in the same subcellular compartments. We verify inhibition of the studied isozymes by a thorough kinetic analysis. We propose to further exploit antifolate scaffold in development of SHMT inhibitors because it seems likely that especially polyglutamylated PTX inhibits SHMTs in vivo. Structure-based optimization is expected to yield novel antifolates that could potentially be used as chemotherapeutics.

Figure 1

Structural formulas of tetrahydrofolate (THF), methotrexate (MTX), and pemetrexed (PTX). The compounds are divided into P-, B-, and E-moieties, as referenced throughout the text.

Figure 2

Dependence of AtSHMT2 and AtSHMT4 activity on pH. The initial velocity of the SHMT forward reaction was measured at different pH values with AtSHMT2 and AtSHMT4 (0.2 μM) using a fixed concentration of one substrate while varying the concentration of the other substrate. The fixed L-serine concentration was saturating. THF fixed concentration was giving the maximum activity when varying L-serine. All experimental points are the average ± standard deviation of three independent measurements. Panel (A) AtSHMT2, 30 mM L-serine; Panel (B) AtSHMT2, 180 µM THF; Panel (C) AtSHMT4, 30 mM L-serine; Panel (D) AtSHMT4, 30 µM THF at pH 6.5 and 7.0, 50 µM THF at pH 8.0, 80 µM THF at pH 8.5 and 100 µM THF at pH 9.5. Reactions were carried out at 30 °C.

Figure 3

AtSHMT2 and AtSHMT4 inhibition by MTX and PTX. Panel (A) shows data for AtSHMT2 inhibition by MTX; (B) AtSHMT4 by MTX; C, AtSHMT2 by PTX; D AtSHMT4 by PTX. Secondary plots of slopes as functions of antifolate concentrations were obtained from fitting of inhibition data (Fig. S2). Intercept on the X-axis gives an estimate of the inhibition constant (K_i) related to antifolate binding to the enzyme-glycine complex. Since it is known that the mitochondrial matrix has an alkaline pH (pH 8) as compared to the cytosol (pH 7.3), inhibition measurements were performed in 20 mM KPi buffer at pH 8.0 for AtSHMT2 and pH 7.3 for AtSHMT4 at 30 °C. Measurements were carried out varying 5-formyl-THF concentration, while keeping glycine at 10 mM and antifolates at different fixed concentrations. Units of slopes are μM.

Figure 4

Electrostatic potential and antifolate binding site. AtSHMT2-MTX complex (tetramer) is shown with marked subunits (A–D).

Figure 5

MTX and PTX binding by SHMTs. Green mesh (in A,C,E,G,I) represents Polder electron density maps contoured at 4.5 σ calculated for 5 Å volume around the antifolates. Panels (B,D,F,H,J) show networks of interactions between SHMTs and antifolates. Internal (LLP) or external aldimines are yellow, free serine is in magenta, residues from another protein subunit than indicated in caption, involved in interactions with MXT or PTX, are marked by asterisks.

Figure 6

Comparison of MTX, PTX, and THF binding modes. Positioning of THF (black, ball-and-stick) was acquired by superposing THF complex with murine SHMT (PDB ID: 1eji; chain A). EA-poses are shown for AtSHMT2-MTX complex (chain A), and AtSHMT2-PTX complex (chain B).

Figure 7

Detailed binding mode of PTX to AtSHMT2. PTX (orange bonds) interacts with the protein via an extensive network of hydrogen bonds (dashed, black lines; distances are given in Å), mostly mediated by water molecules (light blue circles). Hydrophobic interactions are indicated by “eyelashes”. Chain B of AtSHMT2-PTX complex is shown; EG, ethylene glycol molecule.

Figure 8

Comparison of hcSHMT, hmSHMT, AtSHMT2, and AtSHMT4. In Panel (A), bars above the sequences indicate residue conservation. Residues interacting with P-, B-, and E-moieties are marked with blue squares, yellow circles, and red triangles, respectively, and labeled with a corresponding letter. Panels (B,C) show pairwise structural comparisons of cytosolic AtSHMT4 (gray and teal subunits; blue MTX) with hcSHMT (coral, green), and of mitochondrial AtSHMT2 (gray, teal; orange PTX) with hmSHMT (coral, green; navy PTX). Residues and their numbering are given for plant/human isoforms. Antifolates are shown in ball-and-stick representation.