INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Enzymes responsible for deamination reactions are widespread throughout intermediary metabolism and serve to incorporate and recycle nitrogen among key metabolites essential for DNA and protein synthesis. Some are members of an amidohydrolase protein superfamily, which catalyze at least 30% of the steps in four intermediary metabolic pathways (22). Recently, a new class of bacterial amidohydrolases has been identified; they catalyze the hydrolytic displacement of amino groups and chlorine substituents from s-triazine ring compounds (22, 32).

The s-triazine compounds have numerous applications throughout industry and agriculture (10, 19, 25, 30). Those containing N-alkyl substituents, like atrazine (2-chloro-4-N-ethylamino-6-N-isopropylamino-1,3,5-triazine), have been applied successfully as herbicides (3). Atrazine and analogous chlorinated s-triazines were initially considered to be incompletely metabolized by microorganisms (10, 19). However, 40 years after the initial introduction of atrazine into the environment, bacteria with the ability to completely mineralize this herbicide have been isolated (12, 26, 31, 40). Subsequently, bacteria were shown to initiate atrazine metabolism via dechlorination to yield hydroxyatrazine (2, 8, 12, 26, 31, 40). In 1996, the dechlorinating enzyme atrazine chlorohydrolase (AtzA) was purified and shown, via [¹⁸O]water experiments, to catalyze a hydrolytic displacement reaction (Fig. 1) (13).

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Comparison of the reactions catalyzed by melamine deaminase (TriA) from Pseudomonas sp. strain NRRL B-12227 (A) and AtzA from Pseudomonas sp. strain ADP (B).

The substrate specificity of AtzA from Pseudomonas sp. strain ADP was recently investigated (35). AtzA catalyzes the hydrolytic removal of a chlorine or fluorine substituent but does not remove cyano, azido, methoxy, thiomethyl, or amino substituents from compounds structurally analogous to atrazine. AtzA is also not active with any of the pyrimidine substrates tested (35).

Melamine (2,4,6-triamino-1,3,5-triazine), a related s-triazine that predates the use of atrazine (29), is also metabolized by soil bacteria. Worldwide production of melamine in 1994 was estimated to be 900 million lb (21). Melamine is most commonly used in the production of melamine-formaldehyde resins, which are used in laminates, adhesives, fire retardants, molding compounds, coatings, and concrete plasticizers (29). Prior to the identification of atrazine-mineralizing bacteria, Cook and Hutter isolated melamine-metabolizing pseudomonads (11). One of these, Pseudomonas sp. strain NRRL B-12227, catalyzes consecutive hydrolysis of the three amino substituents of melamine, producing the intermediates ammeline, ammelide, and cyanuric acid (Fig. 1). Another bacterium, Pseudomonas sp. strain NRRL B-12228, was unreactive with melamine but catalyzed deamination of ammeline to ammelide and of ammelide to cyanuric acid (11). Genes for ammeline and ammelide deamination, trzB and trzC, respectively, have been cloned from Pseudomonas sp. strain NRRL B-12228 (17). Detailed restriction site pattern analysis revealed conservation of trzC but not trzB in Pseudomonas sp. strain NRRL B-12227 (17). The genes encoding the enzyme for melamine or ammeline deamination in Pseudomonas sp. strain NRRL B-12227, however, were not reported. Furthermore, Pseudomonas sp. strain NRRL B-12227 was shown not to metabolize atrazine (11).

Given the similarity in structure between melamine and atrazine and their similar hydrolytic metabolism, it was hypothesized here that AtzA gene probes and antibodies might be used to identify the melamine deaminase gene and protein, respectively, in Pseudomonas sp. strain NRRL B-12227. Using this strategy, the melamine deaminase gene, designated triA, was identified, cloned, and sequenced. The melamine deaminase gene from Pseudomonas sp. strain NRRL B-12227 was 99% identical to the atzA gene from Pseudomonas sp. strain ADP. The cloned melamine deaminase was expressed in Escherichia coli DH5 and shown to catalyze the deamination of melamine and ammeline. Melamine deaminase had no activity with any of the chlorotriazine substrates tested. Taken together with the known substrate specificity of AtzA, these studies identified two nearly identical proteins that catalyze clearly distinct biochemical reactions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains, plasmids, and growth conditions. Pseudomonas sp. strains NRRL B-12227 and NRRL B-12228 were provided by Richard Eaton (U.S. Environmental Protection Agency, Gulf Breeze, Fla.). Pseudomonas sp. strains NRRL B-12227 and NRRL B-12228 have similar restriction site patterns for the genes trzC and trzD, which convert ammelide to biuret (17). Rhodococcus corallinus NRRL B-15444R was obtained from the U.S. Department of Agriculture National Center for Agriculture Utilization Research (Peoria, Ill.). Pseudomonas or Rhodococcus strains were grown in R salt minimal medium (36) with either glucose or glycerol as the carbon source, respectively, and appropriate triazines as the sole nitrogen source. Rhodococcus strains were grown at 30°C, while Pseudomonas and E. coli strains were incubated at 37°C. E. coli DH5 transformants were grown on Luria-Bertani media (33) containing the appropriate antibiotic. The atzA clone E. coli DH5(pMD4) was grown in media with chloramphenicol (30 µg/ml), while E. coli DH5 containing pUC18-derived clones was grown in media with ampicillin (100 µg/ml).

Chemicals and reagents. Atrazine, ammeline, and ammelide were generously provided by Syngenta Crop Protection (Greensboro, N.C.). Melamine was purchased from Sigma (St. Louis, Mo.), and cyanuric acid was from Fluka (Ronkonkoma, N.Y.). The other triazines used in this study, aminoatrazine, cyanoatrazine, and azidoatrazine, were synthesized in our laboratory by Gilbert Johnson as previously described (35).

DNA manipulation and Southern hybridizations. Total genomic DNA was isolated as previously described (33) and digested with EcoRI, HindIII, BamHI, and AvaI in four independent reactions. A double digestion using EcoRI and BamHI was also performed. DNA was separated on a 0.7% agarose gel and transferred onto a nylon membrane. Southern hybridizations were performed under stringent conditions as previously described (33), using a 1.9-kb AvaI fragment from pMD4 containing the atzA gene as a probe (15).

Cloning of triA. The triA gene was isolated by using the PCR technique. Total genomic DNA from Pseudomonas sp. strain NRRL B-12227 was used as template for the PCR. Custom primers (Integrated DNA Technologies, Coralville, Iowa) were designed using the Primer Design package (version 2.01; Scientific and Educational Software, State Line, Pa.) and were based on regions external to and flanking the atzA gene from Pseudomonas sp. strain ADP. An EcoRI restriction site was added to the forward primer (atzA-87Eco, TGC GGG ATG ACC GAA TTC CGG TGC AGG TTT TTC GAT G), and a HindIII restriction site was added to the reverse primer (atzA1700Hindcomp, TTT CCT CAA GGG GCG GCG GAA GCT TCA ACG GCG TCA TTT C). A 1.5-kb PCR fragment was obtained using Pfu Turbo DNA polymerase (Stratagene, La Jolla, Calif.) as recommended by the manufacturer. The resulting PCR fragment was gel purified on a 0.8% agarose gel and was isolated using the Geneclean II system (Bio 101, Inc., Vista, Calif.). The triA gene was cloned into the EcoRI and HindIII sites of pUC18, and the resulting plasmid, pJS3, was transformed into Maximum Efficient E. coli DH5 (Gibco BRL, Gaithersburg, Md.). The melamine degradation phenotype of transformed cells was confirmed by incubating cell extracts with melamine, followed by analysis using high-pressure liquid chromatography (HPLC), as described below.

DNA sequencing. Plasmid DNA was isolated as previously described (33). The nucleotide sequence on both strands was determined in duplicate using a PRISM ready-reaction dideoxy terminator cycle sequencing kit (Perkin-Elmer Corp.) and an ABI model 373A DNA sequencer (Applied Biosystems, Foster City, Calif.). The Genetics Computer Group sequence analysis software package (Madison, Wis.) was used for all DNA and protein sequence comparisons and alignments.

Preparation of cell extracts. Overnight cultures were centrifuged at 14,000 × g for 2 min at 4°C. Cell pellets were resuspended in ice-cold 10 mM phosphate buffer (pH 7). Cell suspensions were subjected to sonication with a Biosonik sonicator (Bronwill Scientific, Rochester, N.Y.) at 80% intensity, followed by three freeze-thaw cycles. Lysed cell suspensions were centrifuged at 14,000 × g for 15 min at 4°C to obtain cell extracts.

Partial enzyme purification and Western hybridization. Overnight cultures were centrifuged at 10,000 × g for 10 min and resuspended in 25 mM morpholinepropanesulfonic acid (MOPS) (pH 7). Cell suspensions were passed through a French pressure cell as previously described (13). After three additional freeze-thaw cycles, cell debris was removed by centrifugation at 15,000 × g for 100 min at 4°C. A 0 to 20% ammonium sulfate precipitation of the supernatant was used as partially purified enzyme. AtzA was purified as previously described (13). The partially purified melamine deaminase and purified AtzA were dialyzed against 25 mM MOPS (pH 7) containing 0.5 g of iron sulfate per liter, followed by dialysis in metal-free 25 mM MOPS (pH 7) buffer.

Ammelide competition studies. Enzymatic reactions were done using 100 µM melamine or ammeline in the presence or absence of ammelide (100 µM). Initial rates were determined at 22°C with crude extracts of E. coli DH5(pJS3). Reaction mixtures with melamine as the substrate contained 12.4 µg of protein per ml, while reaction mixtures with ammeline required 124 µg of protein per ml to obtain 8 to 10% conversion of substrate to product within 30 min. Triazine degradation was monitored at 0, 5, 15, 20, and 30 min after substrate addition by HPLC analysis, as described below.

Determination of substrate range. Partially purified enzyme from E. coli DH5(pJS3), isolated as described above, was incubated with the compounds listed in Table 1 at a final concentration of 8 µg of total protein per ml. Samples were assayed by HPLC at 16 and 24 h as described below. HPLC sensitivity limited detection of rates to those greater than 10 pmol/min/mg of protein. For those triazines that were substrates for melamine deaminase, initial rates of deamination and specific activities were determined from duplicate samples.

Melamine deaminase and AtzA competition experiments. Initial rates of melamine deamination and atrazine dechlorination were determined in triplicate samples in the presence of equal molar concentrations of various triazines. Specific activities were calculated based upon partially purified protein and highly purified enzyme concentrations for melamine deaminase and AtzA, respectively.

Analytical procedures. HPLC analysis was performed on a Hewlett-Packard HP 1100 Series Chromatographic system equipped with a diode array detector and interfaced with an HP Chemstation (revision A.05.01). Melamine and its metabolites were separated either on an Alltech Inertsil 5µ (150 by 4.6 mm) phenyl column (Deerfield, Ill.) with an isocratic aqueous mobile phase consisting of 5 mM sodium octane sulfate in 0.05% H₃PO₄ (pH 2.8) and detected at 200 nm or on a Merck LiChrosorb RP-8 5µ (250 by 4 mm) column (Gibbstown, N.J.) with an acetonitrile (ACN)-water gradient as follows: 5 min, 3% ACN; 5-min linear gradient to 50% ACN; 5-min linear gradient to 100% ACN; 3-min linear gradient to 3% ACN; and 3% ACN for 2 min. An Alltech Adsorbosphere C₁₈ 5µ (250 by 4.6 mm) column or a mixed-mode C₈-anion 5µ (250 by 4.6 mm) column was used to detect the alkylated triazines, including aminoatrazine, atrazine, and hydroxyatrazine, with an ACN-water linear gradient as previously described (7). Compounds with no detectable transformation after 24 h are listed below the detection limit of 0.01 nmol/min/mg.

Nucleotide sequence accession number. The triA sequence has been entered into GenBank under accession number AF312304.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of an AtzA homolog in Pseudomonas sp. strain NRRL B-12227. Total genomic DNA from Pseudomonas sp. strain NRRL B-12227 was digested with BamHI, EcoRI, HindIII, AvaI, and EcoRI-plus-BamHI. The digested DNA was separated on a 0.7% agarose gel and subjected to Southern hybridization analysis. In each case, a single fragment of 14, 12, 10, 4, or 12 kb, respectively, hybridized to an atzA probe (data not shown).

View larger version (16K): [in this window] [in a new window]   FIG. 2.   Western blotting of an SDS-PAGE protein gel using antibodies against the AtzA protein. Lanes from left to right: 1, E. coli(pMD4) positive control for AtzA; 2, prestained protein standards; 3, Pseudomonas sp. strain ADP positive control for AtzA; 4, Pseudomonas sp. strain NRRL B-12228; 5, E. coli (TrzA) negative control; 6, Pseudomonas sp. strain NRRL B-12227; and 7, R. corallinus sp. strain NRRL B-15444R (TrzA).

Cloning of the triA gene To determine whether the AtzA homolog was responsible for melamine degradation in Pseudomonas sp. strain NRRL B-12227, the gene was cloned by using the PCR technique and high-fidelity polymerase Pfu Turbo. A 1.5-kb PCR product was cloned into pUC18, and the resulting plasmid, pJS3, was transformed into E. coli DH5 cells. Cell extracts of the transformed E. coli cells showed melamine degradation activity by HPLC analysis. Control E. coli DH5 cell extracts failed to degrade melamine, suggesting that the cloned homolog in pJS3 was responsible for melamine degradation.

DNA sequence of triA The cloned DNA in pJS3 was sequenced. Sequence analysis indicated a single open reading frame 1,425 nucleotides in length. The open reading frame was named triA (GenBank accession no. AF312304) and encodes melamine deaminase. Comparison of gene and protein sequences of triA and atzA revealed that these two genes were 99% and 98% identical at the nucleotide and amino acid levels, respectively. The 9 nucleotide differences between the two genes correspond to 9 amino acid changes in the protein sequences (Fig. 3). To establish that no additional mutations were introduced through PCR, three overlapping 0.6-kb fragments were also amplified directly from Pseudomonas sp. strain NRRL B-12227 genomic DNA and sequenced in duplicate in both directions. The smaller overlapping fragments were identical to the sequence obtained from pJS3.

View larger version (55K): [in this window] [in a new window]   FIG. 3.   Amino acid sequences of AtzA and melamine deaminase (TriA) (GenBank accession no. AF312304). Boxes denote amino acid residues that differ between the two sequences.

Melamine deaminase-catalyzed melamine hydrolysis products. Cell extracts containing melamine deaminase were incubated with melamine (100 µM) for approximately 16 h. Formation of a white precipitate indicated production of less soluble products, which coeluted on HPLC with ammeline and ammelide. Mass spectral analysis of products confirmed that melamine deaminase catalyzes the hydrolysis of melamine to ammeline and ammelide (results not shown). To determine if melamine deaminase was also capable of deaminating ammeline, cell extracts of E. coli(pJS3) were incubated with ammeline. HPLC analysis of enzymatic reactions revealed formation of ammelide. Incubation of ammelide with melamine deaminase failed to produce any detectable products, including cyanuric acid. These results indicated that melamine deaminase is capable of removing two of the three amino groups from melamine, producing ammelide as a final product (Fig. 4). Formation of ammelide was 15-fold slower than that of ammeline in cell extracts. Competition experiments, in which ammelide was added to reaction mixtures with melamine and ammeline, indicated that ammelide failed to influence melamine deaminase deamination activities.

View larger version (16K): [in this window] [in a new window]   FIG. 4.   Time course of melamine and ammeline deamination by cell extracts prepared from the TriA clone E. coli(pJS3). Error bars represent the standard error of the mean; n = 2.

Substrate specificity of melamine deaminase. In addition to melamine and ammeline, melamine deaminase catalyzed the deamination of other s-triazines. Partially purified melamine deaminase was used throughout these studies. Table 1 summarizes the specific activities for those compounds that were substrates for melamine deaminase. Deamination was confirmed in each case by detection of the corresponding hydroxylated product except in the case of 2-chloro-4,6-diamino-s-triazine (CAAT). The proposed product, 2-amino-4-chloro-6-hydroxy-s-triazine (CAOT), is unstable in water and hydrolyzes to ammelide. Mass spectometry confirmed the presence of ammelide as the final product of the reaction. An initial enzyme-catalyzed dechlorination reaction would produce ammeline, which was not detected.

Melamine deaminase competition experiments. The inhibition of melamine deamination activity (n = 3) by various triazines was determined, using partially purified enzyme. The percent inhibition of melamine deamination rates in the presence of equimolar concentrations of other triazine compounds differed. Aminoatrazine inhibited melamine turnover sixfold more than atrazine (71.0% ± 0.6% versus 12% ± 2%). This suggests that melamine deaminase may have active site determinants that bind an amino substituent of the triazine ring. Ammeline and CAAT both inhibited melamine deamination by approximately 50% (47% ± 3% and 49% ± 4%), suggesting that melamine, ammeline, and CAAT bind to similar extents. Therefore, the differences in the rates in Table 1 among these substrates are probably due to differences in turnover and not in substrate binding.

AtzA competition experiments. Inhibition of AtzA (n = 3) by aminotriazines was also investigated. Atrazine dechlorination rates were reduced (73% ± 6%) by aminoatrazine, but dechlorination was not inhibited by either melamine (0% ± 2%) or CAAT (0% ± 3%). This suggests that AtzA requires the N-alkyl group on the s-triazine ring for efficient substrate recognition. This differs from melamine deaminase, which was active with both N-alkylated and nonalkylated aminotriazines. The high degree of inhibition of melamine catalysis in the presence of aminoatrazine (71.0% ± 0.6%) suggests that melamine deaminase may bind the alkylated triazines better than it binds nonalkylated ones. However, the low degree of inhibition observed in the presence of atrazine (12% ± 2%) suggests that the active site of melamine deaminase does not favor a chlorine substituent. These observations suggest the presence of a proton-donating group at the active site that might assist in binding the amino-leaving group of the substrate.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Homologous proteins catalyzing different reactions are being discovered at an increasing rate with genomic research focusing attention on the interplay between macromolecular sequence and function. As one part of functional genomics efforts, homologous proteins are being classified into superfamilies, some of which serve different biological functions by virtue of catalyzing a range of biochemical reactions (6, 9, 20, 22, 34, 45). For example, the amidohydrolase superfamily has members that catalyze ring deamination, amide bond hydrolysis, phosphotriester hydrolysis, and dechlorination reactions (22). AtzA had previously been shown to be a member of the amidohydrolase superfamily (32); in this study, melamine deaminase was added. In this superfamily and in others, members that catalyze different reactions are generally divergent to the extent that amino acid sequence identity is less than 50%. This underlies current genome annotation efforts where functional assignments based on >50% sequence identity are considered to be reasonably sound. The present finding that proteins with >98% sequence identity catalyze different reactions in different metabolic pathways is highly exceptional. Moreover, no vestigial reactivity could be detected with either enzyme. AtzA was previously shown to have no deaminase activity, even with a structural analog of atrazine containing an amino group in place of a chlorine substituent (35). In this study, melamine deaminase was observed not to catalyze dechlorination with chlorinated aminotriazine substrates. Moreover, data presented in this study are consistent with the presence of at least two areas of substrate recognition within the active site of melamine deaminase and one in AtzA. Both enzymes appear to contain a binding surface for an N-alkyl group, but melamine deaminase appears to contain an additional site for binding the amino leaving group.

While it is surprising that enzymes with such high sequence identity catalyze different reactions, it is well known that microbes adapt to metabolize new chemical compounds and that enzymes evolve in that context. This point is illustrated with the homologous enzymes enoyl-coenzyme A (CoA) hydratase and 4-chlorobenzoyl-CoA dehalogenase, which catalyze the addition of water to a double bond and hydrolytic dehalogenation of a chlorinated aromatic ring compound, respectively (34). These two enzymes show 56% amino acid sequence identity in pairwise comparison and have very similar three-dimensional structures (6, 45), yet neither member of the pair will catalyze the other's reaction. Similarly, plant fatty acyl desaturases and hydroxylases have been observed to share 81% sequence identity but catalyze different reactions (9). Each enzyme is thought to contain a binuclear iron cluster, which serves as the site of oxygen activation to catalyze either hydroxylation or electron abstraction, respectively (37). This commonality of structure and mechanism with a divergence during the latter part of the enzyme catalytic cycle is an emerging theme that will provide insight for studies in functional genomics and enzyme evolution.

The amidohydrolase superfamily fits this paradigm in that its best-characterized members are known to share a common ()₈ barrel structure (22) and a common catalytic mechanism. For example, adenosine deaminase (39, 43, 44) and phosphotriesterase (4-5) catalyze their reactions via metal activation of water for nucleophilic attack on their substrates. AtzA and melamine deaminase are members of this superfamily and may share this common mechanism, although this remains to be established experimentally. The s-triazine hydrolase, TrzA, from R. corallinus NRRL B-15444R is also a member of the amidohydrolase superfamily (22). It catalyzes both deamination and dechlorination reactions with triazine ring substrates (27). The s-triazine hydrolase, which shares 44% and 43% sequence identity with melamine deaminase and AtzA, respectively, catalyzes melamine deamination but is not active in the dechlorination of atrazine (27).

Since the initial identification of AtzA in Pseudomonas sp. strain ADP, atrazine metabolism genes have been detected in other bacteria, including Rhizobium sp. strain PATR (8), Alcaligenes sp. strain SG1 (K. Boundy-Mills, unpublished), Agrobacterium radiobacter J14a (40), Ralstonia pickettii D (M. L. de Souza, N. R. Plechacek, L. P. Wackett, M. J. Sadowsky, and B. L. Hoyle, Abstr. 98th Gen. Meet. Am. Soc. Microbiol., abstr. Q-195, p. 453, 1998), Clavibacter michiganensis ATZ1 (2, 12), and Pseudaminobacter sp. strain C147 (41). Sequence analysis has indicated that the AtzA genes in various atrazine-degrading bacteria have greater than 99% sequence identity to AtzA from Pseudomonas sp. strain ADP within a 0.5-kb region (14). The identity of the atzA homologs present in diverse genera, the plasmid localization of atzA genes, and the identification of IS1071-like sequences flanking the atrazine degradation genes suggest that horizontal gene transfer and transposition may have contributed to the spread of the atrazine degradation genes among bacteria (16). The present study shows that the global distribution of highly identical genes extends beyond the catabolism of atrazine and includes a bacterium with a metabolic pathway to catabolize melamine. Since melamine has been used industrially for approximately 80 years and atrazine for 40 years (29), these observations are consistent with recent evolutionary adaption of the relevant genes.

Sequence comparison of triA and atzA indicates a lack of silent mutations, suggesting that strong selective pressure occurred relatively quickly (18, 42). Moreover, the 9 amino acid differences are positioned throughout the proteins, indicating that catalytic differences are most likely not due to an alteration in tertiary structure but rather due to localized changes. Under conditions of weak selection and due to genetic drift, silent mutations that do not lead to changes in the amino acid sequence evolve, resulting in a trend towards preferred codon usage (1, 24, 28). It has been estimated that only 10% of the amino acid differences between species are driven by positive selection (28) and that relatively few substitutions that lead to amino acid replacements are accepted and maintained (23). This suggests that the evolutionary divergence of melamine deaminase and AtzA occurred relatively recently and under conditions of strong selective pressure.