Discovery of a bacterial 5-methylcytosine deaminase

Abstract

5-Methylcytosine is found in all domains of life, but the bacterial cytosine deaminase from Escherichia coli (CodA) will not accept 5-methylcytosine as a substrate. Since significant amounts of 5-methylcytosine are produced in both prokaryotes and eukaryotes, this compound must eventually be catabolized and the fragments recycled by enzymes that have yet to be identified. We therefore initiated a comprehensive phylogenetic screen for enzymes that may be capable of deaminating 5-methylcytosine to thymine. From a systematic analysis of sequence homologues of CodA from thousands of bacterial species, we identified putative cytosine deaminases where a "discriminating" residue in the active site, corresponding to Asp-314 in CodA from E. coli, was no longer conserved. Representative examples from Klebsiella pneumoniae (locus tag: Kpn00632), Rhodobacter sphaeroides (locus tag: Rsp0341), and Corynebacterium glutamicum (locus tag: NCgl0075) were demonstrated to efficiently deaminate 5-methylcytosine to thymine with values of kcat/Km of 1.4 × 10(5), 2.9 × 10(4), and 1.1 × 10(3) M(-1) s(-1), respectively. These three enzymes also catalyze the deamination of 5-fluorocytosine to 5-fluorouracil with values of kcat/Km of 1.2 × 10(5), 6.8 × 10(4), and 2.0 × 10(2) M(-1) s(-1), respectively. The three-dimensional structure of Kpn00632 was determined by X-ray diffraction methods with 5-methylcytosine (PDB id: 4R85 ), 5-fluorocytosine (PDB id: 4R88 ), and phosphonocytosine (PDB id: 4R7W ) bound in the active site. When thymine auxotrophs of E. coli express these enzymes, they are capable of growth in media lacking thymine when supplemented with 5-methylcytosine. Expression of these enzymes in E. coli is toxic in the presence of 5-fluorocytosine, due to the efficient transformation to 5-fluorouracil.

Scheme 1

Figure 1

Active site structure of CodA from E. coli. (A) Residues involved in the binding of the divalent cation in the active site are conserved in all enzymes from cog0402 of the amidohydrolase superfamily (PDB id: 1K6W). (B) Mode of binding of isoguanine in the active site of CodA (PDB id: 3RN6). (C) Mode of binding of phosphonocytosine in the active site of CodA (PDB id: 3O7U).

Scheme 2

Figure 2

Sequence similarity network (SSN) diagram of CodA homologues at a BLAST E-value cutoff of 10^–140. Groups are labeled by their representative purified protein: (a) CodA (b0337); (b) Kpn00632; (c) Rsp0341; and (d) NCgl0057.

Figure 3

Sequence alignment of CodA, Kpn00632, Rsp0371, and NCgl0075. Residues presented in Figures 1, 4, and 5 are highlighted.

Figure 4

Metal center of Kpn00632 with various ligands bound in the active site. (A) 5-methylcytosine (PDB id: 4R85); (B) 5-fluorocytosine (PDB id: 4R88; and (C) phosphonocytosine (PDB id: 4R7W).

Figure 5

Active site and ligand binding residues of Kpn00632 and the D314S mutant of CodA from E. coli. (A) Kpn00632 bound with 5-methylcytosine; (B) Kpn00632 bound with 5-fluorocytosine; (C) Kpn00632 bound with phosphonocytosine; and (D) CodA-D314S (PDB: 1RAK) bound with 5-fluoro-4-S-hydroxy-3,4-dihydropyrimidine.

Figure 6

Growth of E. coli thymine auxotrophs supplemented with 5-methylcystosine and enzymes capable of deaminating 5-methyl cytosine to thymine. The experimental conditions are as follows: 400 μM 5-methylcytosine (×); 100 μM arabinose (■); 100 μM arabinose and 100 μM thymine (*); 100 μM arabinose and 100 μM 5-methylcytosine (◆); and 100 μM arabinose and 400 μM 5-methylcytosine (+). (A) Empty pBAD322c vector; (B) CodA from E. coli; (C) Rsp0341; and (D) Kpn00632.

Figure 7

Toxicity of 5-fluorocytosine to E. coli in the presence of enzymes capable of deaminating 5-fluorocytosine to 5-fluorouracil. The experiment conditions include the following: induction with 100 μM arabinose (×); 100 μM arabinose and 50 μM 5-fluorocytosine (■); 100 μM arabinose and 500 μM 5-fluorocystosine (*); 100 μM arabinose and 5 μM 5-fluorocystosine (◆). (A) Empty pBAD322c vector; (B) CodA from E. coli; (C) CodA-D314A mutant; and (D) Kpn00632.