Horizontal Gene Transfer Involved in the Convergent Evolution of thePlasmid-Encoded Enantioselective 6-Hydroxynicotine Oxidases
S. Schenk,* K. Decker
Institute of Biochemistry and Molecular Biology, Albert Ludwig University, Hermann-Herder-Strasse 7, D-79104 Freiburg, Germany
Received: 6 March 1998 / Accepted: 15 July 1998
Abstract. The D- and L-specific nicotine oxidases areflavoproteins involved in the oxidative degradation ofnicotine by the Gram-positive soil bacteriumArthrobac-ter nicotinovorans.Their structural genes are located ona 160-kbp plasmid together with those of other nicotine-degrading enzymes. They are structurally unrelated at theDNA as well as at the protein level. Each of these oxi-dases possesses a high degree of substrate specificity;their catalytic stereoselectivity is absolute, although theyare able to bind both enantiomeric substrates with a simi-lar affinity. It appears that the existence of these enzymesis the result of convergent evolution. The amino acidsequence of 6-hydroxy-L-nicotine oxidase (EC 1.5.3.6)as derived from the respective structural gene shows con-siderable structural similarity with eukaryotic mono-amine oxidases (EC 1.4.3.4) but not with monoamineoxidases from prokaryotic bacteria including those of thegenusArthrobacter.These similarities are not confinedto the nucleotide-binding sites. A 100-amino acid stretchat the N-terminal regions of 6-hydroxy-L-nicotine oxi-dase and human monoamine oxidases A possess a 35%homology. Overall, 27.0, 26.9, and 25.8% of the aminoacid positions of the monoamine oxidases ofAspergillusniger (N), humans (A), and rainbow trout(Salmo gaird-neri) are identical to those of 6-hydroxy-L-nicotine oxi-dase (Smith–Waterman algorithm). In addition, the G+Ccontent of the latter enzyme is in the range of that of
eukaryotic monoamine oxidases and definitely lowerthan that of theA. nicotinovoransDNA and even that ofthe pAO1 DNA. The primary structure of 6-hydroxy-D-nicotine oxidase (EC 1.5.3.5) does not reveal its evolu-tionary history as easily. Significant similarities arefound with a mitomycin radical oxidase fromStreptomy-ces lavendulae(23.3%) and a ‘‘hypothetical protein’’from Mycobacterium tuberculosis(26.0%). It is pro-posed that the plasmid-encoded gene of 6-hydroxy-L-nicotine oxidase evolved after horizontal transfer froman eukaryotic source.
The conservation of essential domains of proteins is in-creasingly recognized as an important mechanism inevolution. Close structural relations between proteins ofprokaryotic and proteins of eukaryotic organisms that areengaged in pathways or functions of general biologicalimportance have been repeatedly observed. Severalmechanisms have been proposed to account for this phe-nomenon: The related structures may be descendants ofa common ancestral protein; many kinases and dehydro-genases bear witness to this (Labedan and Riley 1995).In recent years, evidence has accumulated that they may
* Present address:The Scripps Research Institute, Department of CellBiology, SBR 12, 10550 North Torrey Pines Road, La Jolla, CA 92037,USACorrespondence to:K. Decker;e-mail: [email protected]
also be the result of a horizontal transfer of genetic ma-terial from one species to another including transfer fromeukaryotic to prokaryotic organisms, e.g., glyceralde-hyde-3-phosphate dehydrogenase (Bork and Doolittle1992) and cytidine deaminase (Koonin et al. 1996). Plas-mids appear to be major carriers of genetic material inthis process (Souza and Eguiarte 1997). Sometimes,however, analogous metabolic functions are also fulfilledby proteins that are structurally unrelated; evidently, theyevolved from different ancestral molecules. In thesecases, the operation of a convergent evolution is invoked(Doolittle 1994).
Arthrobacter nicotinovorans[formerly known asAr-throbacter oxidansbut reclassified in 1992 (Kodama etal. 1992)], an aerobic soil bacterium, carries the 160-kbplasmid, pAO1 (Brandsch et al. 1982); it contains theknown genes responsible for the ability of this organismto utilizeD- or L-nicotine as the sole source of carbon andnitrogen for growth (Eberwein et al. 1961). The firstenzyme of the nicotine pathway, nicotine dehydrogenase(EC 1.5.99.4), introduces a hydroxyl group into the 6-position of the pyridine ring of bothL- and D-nicotine(Hochstein and Rittenberg 1959; Decker et al. 1961).The oxidation of the 6-hydroxynicotines by 6-hydroxy-L-nicotine oxidase and 6-hydroxy-D-nicotine oxidase, re-spectively, leads to the optically inactive 6-hydroxy-N-methylmyosmine (Richardson and Rittenberg 1961) thathydrolyzes spontaneously to 3-(6-hydroxypyridyl)-(g-N-methylaminopropyl)-ketone. The latter compound is fur-ther hydroxylated at the 2-position of the pyridine ringby ketone dehydrogenase (EC 1.5.99.—) (Richardsonand Rittenberg 1961; Decker and Bleeg 1965). 6-Hydroxy-L-nicotine oxidase consists of two identicalsubunits, each bearing one noncovalently bound FADmolecule; in contrast, the monomeric 6-hydroxy-D-nicotine oxidase carries the cofactor FAD covalentlybound to a histidine (Decker and Brandsch 1991).
The presence of these enantiomer-specific enzymesposes several intriguing questions. SinceD-nicotine isnot produced naturally [D-nornicotine, however, has beenreported inNicotianaplants (Kisaki and Tamaki 1961)],the synthesis of an absolutely enantiomer-specific en-zyme is surprising. Thus, it was of interest to find outwhether the L- and D-specific enzymes are geneticallyrelated, i.e., descendants of a common ancestor. Alter-natively, these enzyme activities may have arisen fromproteins with related functions in other organisms, in-serted into pAO1 by horizontal gene transfer and evolvedto serve the specific purpose in bacterial nicotine degra-dation.
The structural gene of 6-hydroxy-D-nicotine oxidase(6-hdno) has been sequenced previously (Brandsch et al.1987). The amino acid sequences of 6-hydroxy-L-nicotine oxidase and 6-hydroxy-D-nicotine oxidase as de-rived from the structural genes did not reveal any sig-nificant resemblance. However, a search of databases
showed striking similarities of 6-hydroxy-L-nicotine oxi-dase and monoamine oxidases of exclusively eukaryoticorigin. The enzymatic oxidation of monoamines is usu-ally associated with mammals as a mechanism to disposeof bioactive monoamines such as hormones and neuro-transmitters (Bach et al. 1988). But monoamine oxidasesare also found in fish (Chen et al. 1994), fungi such asAspergillus sp. (Schilling and Lerch 1994), and evensome microorganisms, e.g.,Arthrobacter globiformis(Tanizawa et al. 1994; Choi et al. 1995),Klebsiella aero-genes(Sugino et al. 1992), andEscherichia coli(Aza-kami et al. 1992). Usually, the major substrates of theseflavoproteins are primary amines. 6-Hydroxy-L-nicotineoxidase, however, oxidizes the circular secondary andtertiary amines, 6-hydroxy-L-nornicotine and 6-hydroxy-L-nicotine, respectively. Mechanistically, the similarity isevident. The first step in all these amine oxidations is theabstraction of a hydride ion from the carbon atom boundto nitrogen; the intermediary enamine structure gives riseto ammonia and an oxo compound (aldehyde) in the caseof monoamine oxidases and to a secondary amine and anoxo compound (ketone) in the course of 6-hydroxy-nicotine oxidation.
This study tries to evaluate the contributions of vari-ous evolutionary pathways to the formation of theplasmid-associated stereoselective 6-hydroxynicotineoxidases. It compares the structural and functional rela-tionships of 6-hydroxy-L-nicotine oxidase and 6-hydroxy-D-nicotine oxidase with each other and withother amine-oxidizing flavoproteins.
Materials and Methods
Bacterial Strains, Media, and Culture Conditions. A. nicotinovoransDSM 420 was grown in a minimal medium containing 0.2%L-nicotine(Eberwein et al. 1961).E. coli JM 109 was used as host for the recom-binant plasmids pLN55 carrying the N terminus, and pLC120 with theC terminus of the 6-hydroxy-L-nicotine oxidase structural gene (6-hlno)(Schenk 1996).E. coli cells were cultured at 37°C overnight in Luria–Bertani (LB) medium supplemented with 150mg ampicillin/ml (finalconcentration).
Preparation of the Plasmid pAO1 DNA.The procedure of choiceconsisted in a combination of a modified alkaline lysis (Sambrook et al.1989a) and the protocol described previously (Brandsch and Decker1984), followed by CsCl gradient centrifugation (Sambrook et al.1989b).
Amplification, Cloning, and Sequencing of a PCR Fragment Com-prising the 6-hlno Gene.The 160-kb pAO1 ofA. nicotinovoranswaschosen as the target sequence for the synthesis of a PCR fragmentcomprising the 6-hlno gene. The oligonucleotide primers that flank the6-hlno gene were derived from the known partial amino acid sequence.The primers carriedEcoRI sites at their 58 ends to allow further cloning(pr-Eco/NN, pr-Eco/LC). The major PCR product as determined byagarose gel electrophoresis was the expected 1700-bp fragment. It waspurified using a QIAEX gel extraction kit.
The PCR product of 1700 bp wasEcoRI-restricted, leading to a1200- and a 550-bp fragment. Analysis with restriction endonucleases
179
revealed the presence of a singleEcoRI site in 6-hlno. The fragmentswere cloned into the appropriately prepared plasmid vector pTZ 18,giving rise to the recombinant plasmids pLN 55 and pLC 120. FreshCaCl2-competentE. coli JM 109 cells were transformed using standardCaCl2 transformation procedures. White colonies were screened byEcoRI restriction endonuclease digestion. Colonies bearing inserts ofthe correct sizes were expanded overnight. The plasmid DNA wasprepared using the Nucleobond AX 500 Kit (Macherey-Nagel, Du¨ren,Germany).
The sequencing of the 6-hlno gene was performed by the ‘‘cycle-sequencing method’’ (Circum Vent Thermal Cycle DNA SequencingKit; New England Biolabs, Schwalbach, Taunus, Germany). For se-quencing the forward strand 150–200 ng of the 160-kb plasmid astemplate and 58 end-labeling with [g-32P]ATP was used. The reversestrand of the 6-hlno gene was sequenced several times, using the wholePCR mixture, the 1700-bp PCR fragment, and the recombinant con-structs pLN55 and pLN 120. When applying the PCR fragments orpLN55 and pLN 120, [a-35S]ATP was used. The thermal cycler pa-rameters for the three-step method had to be adjusted for forward andreverse sequencing.
Southern Blot (Descending Blot).To establish the restriction en-zyme map, CsCl-purified pAO1 DNA was restricted with the desiredenzyme in a volume of 500ml for 3–4 h. Restriction digestions weresubjected to electrophoresis in a horizontal 1% agarose gel and thentransferred to Hybond N+ filters (Amersham Buchler, Braunschweig,Germany) by a descending transfer method (Lichtenstein et al. 1990).
Sequence Alignments.Comparisons of DNA sequences and theirderived amino acid sequences were performed using the BLAST familyof programs (Altschul et al. 1990) and the SSEARCH Genestream[Smith–Waterman (sw)-algorithm] (Smith and Waterman 1981; Pear-son 1991); for the representation of amino acid alignments, theCLUSTAL W and BOXSHADE 3.21 programs; and for the topologicaltree algorithm the PHYLIP program.
The DNA and the derived protein sequences of 6-hydroxy-L-nicotine oxidase are deposited in the EMBL Nucleotide Sequence Da-tabase, accession number AJ 223 391.
Determination of Enzyme Activities.Enzymatic activity of 6-hydroxy-L-nicotine oxidase in cell extracts was determined spectropho-tometrically (Bruhmuller et al. 1972). The assay of flavin nucleotide-dependent oxidations of different substrates yielding hydrogenperoxide was detected by concomitant peroxidase-catalyzed luminoloxidation (Hinkkanen and Decker 1983). Activity staining of 6-hydroxy-L-nicotine oxidase on nondenaturing gels was done by incu-bation with phosphate buffer, pH 7.0, containing 1 mg 6-hydroxy-L-nicotine, 100ml peroxidase solution (1 mg protein/ml deionized water),and 800ml o-dianisidine solution (1 mg/ml methanol). The presence of6-hydroxy-L-nicotine oxidase resulted in the development of a browncolor.
Results
Structural Features of the 6-hlno Gene
The 550-bp fragment obtained byEcoRI cleavage of thePCR product (1700 bp) contains the start site of the6-hlno gene and part of the structural gene of nicotine
dehydrogenase (ndh), confirming the close proximity ofthe 6-hlno and the ndh gene (Grether-Beck et al. 1994).The results obtained by digestion of the whole pAO1DNA with several restriction endonucleases (Fig. 1) arein line with the proposed localization and sequence of the6-hlno gene.
The plasmid-encoded structural 6-hlno gene showsconspicuous differences from other bacterial genes, eventhose of the genusArthrobacter (Schenk 1996). TheG+C content of DNA, [(G+C)/(A+T+G+C)], is charac-teristic for a given species and considered as a taxonomicfeature (Schlegel 1985). It varies among bacteria be-tween 25 and 75%; in DNA ofEnterobacteriaceae(Davis et al. 1980), 50–58%; and inStreptomycetes,69–78% (Goodfellow and Cross 1984). The G+C content ofmost species of the genusArthrobacterwas found to bein the range of 59–66% (Jones and Keddie 1992). Theoverall G+C content ofA. nicotinovoranshas been de-termined as 62.4% (Kodama et al. 1992). Most of theplasmid-encoded genes of the nicotine degradation path-way in this organism have significantly lower G+C con-tents: that of 6-hdno was calculated as 57.5%; that of thendh subunits as 56.6% (ndhA), 56.8% (ndhB), and55.8% (ndhC); and that of the kdh (structural gene ofketone dehydrogenase) subunits as 57% (kdhA), 54.9%(kdhB) (Schelling 1995), and 56% (kdhC) (Hoelz andDecker, unpublished). This observation argues in favorof an import of the pAO1 plasmid genes from sourcesoutside of the genusArthrobacter.The 6-hlno structuralgene, however, is even lower than that; its G+C contentof 54.6% is remarkably similar to that of the mao genesof trout (53.7%) andA. niger (55.0%). Furthermore, thecodon usage in the 6-hlno gene does not resemble thatobserved in many strongly or weakly expressed genes ofE. coli (Grosjean and Fiers 1982; Ikemura 1981). Sur-prisingly, it also did not agree with the codon usage inthe other characterized enzymes of the nicotine degrada-tion pathway, 6-hydroxy-D-nicotine oxidase, nicotine de-hydrogenase, and ketone dehydrogenase.
The open reading frame as confirmed by restrictionanalyses encodes a monomeric polypeptide of 425 aminoacids with a calculatedMr of 46,264.5 (46,984.5 includ-ing 1 mol FAD) per subunit. No significant stretches ofopen reading frame were found in the complementarystrand.
6-Hydroxy-L-nicotine oxidase and 6-hydroxy-D-nicotine oxidase are both FAD-containing oxidases; theycatalyze the same type of reaction and form the sameproduct. Although they are absolutely stereospecific with
Fig. 1. Restriction enzyme map andlocalization of the 6-hlno gene. The nucleotidesequence of the 1.8-kb fragment containing the6-hlno gene was obtained by thermal cyclesequencing. The ndh, moa, and tnp genes havebeen identified by Grether-Beck et al. (1994)and Mene´ndez et al. (1995, 1997).
180
regard to their catalytic properties, they bind both D- andL-substrates with similar affinities (Table 1). Minor dif-ferences exist in the usage of electron acceptors (Deckerand Dai 1967) and FAD-binding and transition states(Pust et al. 1989). The sequences of the structural genesincluding the regulatory sites as well as those of theamino acids, however, possess very weak similarities, ifat all; an overall alignment is not possible. It is veryunlikely that these ‘‘enantiozymes’’ (Bru¨hmuller et al1972) are derived from a common ancestral protein.
Proteins with Sequence Similarity to6-Hydroxy-D-Nicotine Oxidase
6-Hydroxy-D-nicotine oxidase was found to have a mar-ginal structural relation to the mitomycin radical oxidaseof Streptomyces lavendulae(Mr, 46,300; EC 1.5.3.—)(August et al. 1994); 23.3% identical amino acid posi-tions (sw: 154) were seen. Furthermore, an unidentifiedframe fromMycobacterium tuberculosispossesses 26%identical amino acid positions (sw: 280) in a line-up of285 residues. Obviously, these data do not allow anyconclusions with regard to related or progenitor proteins.
Proteins with Sequence Similarity to6-Hydroxy-L-Nicotine Oxidase
The amino acid sequence of 6-hydroxy-L-nicotine oxi-dase exhibits considerable similarity with that of flavin-containing monoamine oxidases ofeukaryotic origin.These relationships are not confined to the dinucleotide-binding sites; they occur also within and across the otherprotein domains. In Fig. 2 the corresponding sequencesare aligned and shaded. No similarities were found withprokaryoticmonoamine oxidases, e.g., maoA and maoCof Klebsiella aerogenes(Sugino et al. 1992), maoA ofE.coli (Azakami et al. 1992), or even monoamine oxidases[histamine oxidase (Choi et al. 1995) and phenylethyl-amine oxidase (Tanizawa et al. 1994)] of a closely re-lated species,Arthrobacter globiformis.Some other re-
corded structural identities are restricted mostly to theFAD-binding sites. Interestingly, no remarkable similari-ties to the other FAD-containing nicotine-degrading en-zymes were found.
The most obvious resemblances were observed be-tween the amino acid sequence of 6-hydroxy-L-nicotineoxidase and those of the eukaryotic monoamine oxidasesfrom Aspergillus niger(MAO-N, 55.6 kDa; 27.1% over-all identical amino acid positions) (Schilling and Lerch1994),Salmo gairdneri(rainbow trout; 56.6 kDa; 25.8%)(Chen et al. 1994), human monoamine oxidases-A and-B (monomer, 60 kDa; 26.9 and 24.7%, respectively)(Bach et al. 1988), and other eukaryotic monoamine oxi-dases, e.g., rat (A, 25.8%) and cow (25.3%). Accordingto Sander and Schneider (1991), the cutoff point for sig-nificant homology is close to 25% identical amino acidsfor longer segments. Striking similarities were seen inthe N-terminal region (including the FAD-bindingbabfold): 41% in a 100-amino acid stretch with monoamineoxidase-N and 35% with human monoamine oxidase-A.Also, in the C-terminal part of 6-hydroxy-L-nicotine oxi-dase, the amino acid sequence 360–423 possesses a highdegree of structural similarity to the fungal, fish, andhuman enzymes. This domain contains the cysteine resi-due (marked⇓ in Fig. 2) that serves as the covalentattachment site for FAD in the human and trout mono-amine oxidases. Monoamine oxidase-N and 6-hydroxy-L-nicotine oxidase do not bind FAD covalently; in theseenzymes, the pentapeptide(SGGCY)with the bindingcysteine is missing. Within this C-terminal similarity do-main (401–423 in 6-hydroxy-L-nicotine oxidase), threeinvariant glycine residues are conserved among the eu-karyotic monoamine oxidases (marked↓ in Fig. 2). Thissequence shows some analogy to other flavoproteins andis suggested to interact with the phosphate groups of theflavin cofactor (Powell 1991). In 6-hydroxy-L-nicotineoxidase, the first glycine is still conserved; four residuesupstream, another glycine is found in 6-hydroxy-L-nicotine oxidase and human and trout monoamine oxi-dase but not in monoamine oxidase-N. A region in thecenter part of the mammalian enzymes (187–230 inmonoamine oxidase-A) has been proposed as part of theactive site (markedc and b in Fig. 2); it is highlyconserved (95% identity) in monoamine oxidase-A and-B from various sources (Hsu YP et al. 1989) excludingmonoamine oxidase-N. But amino acids 173–213 of6-hydroxy-L-nicotine oxidase possess a 27.3% identitywith the mammalian segment.
Overall sequence similarities also exist between 6-hy-droxy-L-nicotine oxidase and some prokaryotic FAD-containing enzymes that oxidize nitrogenous com-pounds. Putrescine oxidase ofMicrococcus rubens(52kDa) (Ishizuka et al. 1993) shows a strong similarity inthe N-terminal domain [41.6% in a 77-amino acid (aa)overlap]. The FAD-containingL-amino acid oxidasefrom Cyanobacterium synechococcusPCC6301 (39.2
Table 1. Properties of 6-hydroxy-D-nicotine oxidase and 6-hydroxy-L-nicotine oxidase (Decker and Brandsch 1991)
Reactivity with oxygen Yes Yes1e acceptors No No2e acceptors Yes No
Intermediate flavin radicalWith S2O4
2− Anionic (red)With hn + EDTA Anionic (red)
181
Fig. 2. Multiple sequence alignment between 6-hydroxy-L-nicotineoxidase and the related eukaryotic monoamine oxidases. The consensusis given relative to the 6-hydroxy-L-nicotine oxidase sequences. Iden-tical residues aredark-shaded;conserved residues arelight-shaded.Inthe consensus line, identical residues in all six sequences are marked
with astar,and conserved residues are marked with adot.The covalentattachment site for FAD in the human and trout monoamine oxidases ismarked⇓, the three invariant glycine residues,↓. The proposed activesite is framed byc andb.
182
kDa) (Bockholt et al. 1995) catalyzes the oxidativedeamination of the basic amino acidsL-arginine, L-lysine, L-ornithine, andL-histidine. A 31-amino acidoverlap in the putative FAD-binding regions (6-hydroxy-L-nicotine oxidase, aa 12–42;L-amino acid oxidase, 2–32) shows 41.9% identical positions. Structural similari-ties also exist with some yet unidentified proteins, e.g., a‘‘hypothetical 51.4-kDa protein’’ derived from an openreading frame ofSynechocystisPCC6803, gene slr0782(25%; sw, 324), an open reading frame, identified onDNA chromosome 4 ofArabidopsis thaliana(21%; sw;219), the ‘‘Fig. 1 protein’’ ofMus musculus(23%; sw;211), and the BC542A protein ofBacillus cereus(22.7%; sw, 195).
To check whether the structural similarities between6-hydroxy-L-nicotine oxidase and monoamine oxidase aswell as putrescine oxidase andL-amino acid oxidasemanifest themselves in corresponding enzymatic activi-ties, highly purified 6-hydroxy-L-nicotine oxidase wasincubated with serotonine (for human monoamine oxi-dase-A activity), epinephrine or phenylethylamine (hu-man monoamine oxidase-B), putrescine and spermidine(putrescine oxidase),L-arginine, L-lysine, L-ornithine,andL-histidine (L-amino acid oxidase), respectively. Noactivity was observed with either substrate. The struc-tural similarities are evidently insufficient to materializein overlapping enzymatic activity. The same negativeresult was obtained when highly purified 6-hydroxy-L-nicotine oxidase isolated on nondenaturing polyacryl-amide gel electrophoresis was immersed in a mixture ofthe respective substance,o-dianisidine and peroxidase(data not shown). The potentially competitive action ofthese substrates was tested using the standard activityassay for 6-hydroxy-L-nicotine oxidase. The presence ofeach of these compounds (80mM) did not lead to anysignificant inhibition of 6-hydroxy-L-nicotine oxidation.
Discussion
The phylogenetic origin of genes encoded in plasmidsonly is yet poorly understood. Usually, plasmid-instructed enzymes serve one of two functions: to enablethe plasmid carrier to deal with a hostile environment,e.g., by conferring antibiotics resistance; or to enlarge thelimiting food supply by opening a new metabolic nichethrough enzymatic degradation of exotic substrates. Inmost instances the plasmid-borne genes are derived fromgenes evolved to confer general mechanisms such ashydrolysis, oxygenation, and hydroxylation.
The nicotine degradation plasmid, pAO1, encodes atleast two distinct enzymes catalyzing hydroxylationsteps, nicotine dehydrogenase introducing an OH groupinto the 6-position and ketone dehydrogenaseinto the 2-position of the pyridine ring. Both enzymeswere shown to have significant structural resemblancesto xanthine dehydrogenases (oxidases) including the re-quirement for FAD, Fe/S clusters, and Mo/molyb-dopterin as cofactors (Kretzer et al. 1993; Schelling1995). This suggests that they have a rather unspeci-fic dehydrogenase/oxidase for a common ancestor.Nevertheless, the possibility cannot be excluded that bac-teria acquired DNA units from an eukaryotic source atsome point in the past and developed them indepen-dently.
Now evidence is presented that a third enzyme in thispathway, 6-hydroxy-L-nicotine oxidase, appears to havea common heritage with enzymes present in mammaliancells, the monoamine oxidases. In contrast to the dehy-drogenating hydroxylases, however, 6-hydroxy-L-nicotine oxidase appears to be totally unrelated to mono-amine oxidases ofprokaryotic origin or of selectedarchaeal open reading frames (Fig. 3).
Fig. 3. Structural relationship of6-hydroxy-L-nicotine oxidase and selectedeukaryotic, prokaryotic, and archaeal proteins.The archaeal proteins are derived from openreading frames ofMethanococcus jannaschii(Bult et al. 1996) andArchaeoglobus fulgidus(Klenk et al. 1997) that showed the highestdegree of homology to 6-hydroxy-L-nicotineoxidase in partial amino acid sequences:‘‘heterodisulfide reductase,’’ subunitA/‘‘methylviologen-reducing hydrogenase,’’subunit delta (Methanobacteriumthermoautotrophicum) identified onArchaeoglobus fulgidus(AF0662);‘‘monoamine oxidase C’’ (MaoC)(Mycobacterium tuberculosis), identified onA.fulgidus (AF2313); ‘‘sarcosine oxidase,’’subunit alpha (soxA) (Corynebacteriumsp.),identified onA. fulgidus(AF0273); ‘‘thiaminebiosynthetic enzyme’’ (thi1) (Zea mays),identified onMethanococcus jannaschii(MJ0601).
183
Considering the relatively small abundance of nico-tine as a natural substrate, it is remarkable that the en-zymes dealing with its degradation developed such ahigh degree of substrate specificity. Not only is 6-hydroxy-L-nicotine oxidase unable to oxidize the mono-amines preferred by monoamine oxidases A and B, re-spectively, but also it is inactive with anyL- or D-aminoacid and, except for 6-hydroxy-L-nornicotine, with struc-turally related compounds such as 6-hydroxyanabasine(Decker et al. 1961).
If the homologous domains of 6-hydroxy-L-nicotineoxidase and the eukaryotic monoamine oxidases hadoriginated prior to the divergence of bacteria and eukary-otes, i.e., some 1.5–2.0 billion years ago, then one wouldexpect such sequence segments to be found not only infungi, plants, and animals, but also in many bacteria(Doolittle 1995). However, no structural resemblance ex-ists between the eukaryotic and any of the known pro-karyotic monoamine oxidases (Fig. 3). Furthermore, the6-hydroxy-L-nicotine oxidase sequence has a higher de-gree of similarity to the animal sequences than would beanticipated for convergent evolution during this long pe-riod.
Two prokaryotic enzymes were found to possess sig-nificant similarities to 6-hydroxy-L-nicotine oxidase, adiamine oxidase and anL-amino acid oxidase. Could oneof these enzymes be an ancestor of 6-hydroxy-L-nicotineoxidase or an intermediary between the eukaryotic pro-teins and the pAO1-encoded enzyme? In the former case,one would expect the structural similarity to 6-hydroxy-L-nicotine oxidase to be greater than that between 6-hydroxy-L-nicotine oxidase and the eukaryotic mono-amine oxidases. In the latter instance, the two enzymesshould have a higher degree of similarity to the eukary-otic enzymes than 6-hydroxy-L-nicotine oxidase. Figure3 seems to rule outL-amino acid oxidase as the precur-sor, but does not allow conclusions with regard to therelationship of 6-hydroxy-L-nicotine oxidase and putres-cine oxidase. It is just as likely that the latter two pro-karyotic enzymes evolved independently from a com-mon eukaryotic ancestor protein. At any rate, the strikingsequence similarity to the eukaryotic monoamine oxi-dases strongly suggests that the 6-hlno gene on plasmidpAO1 resulted from a horizontal DNA transfer from aeukaryote, possibly an animal, to bacteria.
The data retrieved from gene or protein banks do notyet allow conclusions about the phylogeny of 6-hydroxy-D-nicotine oxidase. But the comparisons of its base oramino acid sequences with those of 6-hydroxy-L-nicotineoxidase do not argue for a genetic relatedness of theseenantiozymes. This is the more surprising as both en-zymes are expressed in the presence ofDL-nicotine fromthe same plasmid within the same cell and share severalenzymatic properties including similar binding charac-teristics for both enantiomeric substrates. However, theinduction characteristics are different; nicotine dehydro-
genase, 6-hydroxy-L-nicotine oxidase, and ketone dehy-drogenase are induced synchronously during the loga-rithmic phase of growth ofA. nicotinovorans,while6-hydroxy-D-nicotine oxidase is formed mainly in sta-tionary phase cells (Gloger and Decker 1969) and isunder the control of a separate promotor (Mauch et al.1989). Similarly, the induction of 6-hydroxy-D-nicotineoxidase activity is insensitive to molybdate and tung-state, while the synthesis of 6-hydroxy-L-nicotine oxi-dase responds to the presence of these metals (Grether-Beck et al. 1994). It appears that these enantiozymes arethe result of a convergent evolution.
One might speculate about the phylogenetic history ofthese remarkable enzymes. It is unlikely that a selectivepressure to generate nicotine-specific enzymes existedbefore this or some closely related alkaloid was presenton Earth. About 80 million years ago, in the middle ofthe Cretaceous period, the emergence of the herbivorousanimals contributed to the selection of plants with a richspectrum of secondary metabolites, e.g., alkaloids. Afterthis time, the accumulation of the alkaloid generated aselective advantage for microorganisms that acquired theability to develop nicotine-degrading enzymes. The abil-ity to degrade nicotine oxidatively is not confined toA.nicotinovorans,but so far, other bacterial pathways ofnicotine degradation have not been reported. The degra-dation of this alkaloid by mammals follows a differentmetabolic route (Hucker et al. 1960; Decker and Sam-meck 1964); the enzymatic activities involved in the bac-terial pathway do not participate in the animal system.
Acknowledgments. This work was supported by grants from theDeutsche Forschungsgemeinschaft (SFB 206 and De 113/34-1), Bonn,and Fonds der Chemischen Industrie, Frankfurt, Germany.
References
Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basiclocal alignment search tool. J Mol Biol 215:403–410
August PR, Flickinger MC, Sherman DH (1994) Cloning and analysisof a locus (mcr) involved in mytomycin C resistance inStreptomy-ces lavendulae.J Bacteriol 176:4448–4454
Azakami H, Yamashita M, Roh JH, Suzuki H, Kumagai H, Murooka Y(1994) Nucleotide sequence of the gene for monoamine oxidase(maoA) fromEscherichia coli.J Ferment Bioeng 77:315–319
Bach AW, Lan NC, Johnson DL, Abell CW, Bembenek ME, KwanSW, Seeburg PH, Shih JC (1988) cDNA cloning of human livermonoamine oxidase A and B: Molecular basis of differences inenzymatic properties. Proc Natl Acad Sci USA 85:4934–4938
Bockholt R, Masepohl B, Kruft V, Wittmann-Liebold B, Pistorius EK(1995) Partial amino acid sequence of an L-amino acid oxidasefrom the cyanobacteriumSynechococcusPCC6301, cloning andDNA sequence analysis of the aoxA gene. Biochim Biophys Acta1264:289–293
Bork P, Doolittle RF (1992) Proposed acquisition of an animal proteindomain by bacteria. Proc Natl Acad Sci USA 89:8990–8994
Brandsch R, Decker K (1984) Isolation and partial characterization ofplasmid DNA fromArthrobacter oxidans.Arch Micobiol 138:15–17
Brandsch R, Hinkkanen AE, Mauch L, Nagursky H, Decker K (1987)6-Hydroxy-D-nicotine oxidase ofArthrobacter oxidans.Genestructure of the flavoenzyme and its relation to 6-hydroxy-L-nicotine oxidase. Eur J Biochem 167:315–320
Bruhmuller M, Mohler H, Decker K (1972) Covalently bound flavin in6-hydroxy-D-nicotine oxidase ofArthrobacter oxidans.Eur J Bio-chem 29:143–151
Bult CJ, et al. (1996) The complete genome sequence of the methano-genic archaeonMethanococcus jannaschii.Science 273:1058–1073
Chen K, Wu HF, Grimsby J, Shih JC (1994) Cloning of a novel mono-amine oxidase cDNA from trout liver. Mol Pharmacol 46:1226–1233
Choi YH, Matsuzaki R, Fukui T, Shimizu E, Yorifuji T, Sato H, OzakiY, Tanizawa K (1995). Copper/topa quinone-containing histamineoxidase fromArthrobacter globiformis.Molecular cloning and se-quencing, overproduction of precursor enzyme, and generation oftopa quinone cofactor. J Biol Chem 270:4712–4720
Davis BD, Dulbecco R, Eisen HN, Ginsberg HN (1980) Microbiology,3rd ed. Harper International Edition, New York, pp 645–672
Decker K, Bleeg H (1965) Induction and purification of stereospecificnicotine oxidizing enzymes fromArthrobacter oxydans.BiochimBiophys Acta 105:313–334
Decker K, Brandsch R (1991) Flavoproteins with a covalent histi-dyl(N3)-8a-riboflavin linkage. BioFactors 3:69–81
Decker K, Dai VD (1967) Mechanism and specificity of L- and D-6-hydroxynicotine oxidase. Eur J Biochem 3:132–138
Decker K, Sammeck R (1964) Enzymchemische Untersuchungen zumNicotinabbau in der Kaninchenleber. Biochem Z 340:326–336
Decker K, Gries FA, Bru¨hmuller M (1961) Uber den Abbau des Nico-tins durch Bakterienenzyme III. Stoffwechselstudien an zellfreienExtrakten. Hoppe-Seyler Z Physiol Chem 323:249–263
Doolittle RF (1994) Convergent evolution: The need to be explicit.Trends Biol Sci 19:15–18
Doolittle RF (1995) The multiplicity of domains in proteins Annu RevBiochem 64:287–314
Eberwein H, Gries FA, Decker K (1961). U¨ ber den Abbau des Nicotinsdurch Bakterienenzyme II, Isolierung und Charakterisierung einesnicotinabbauenden Bodenbakteriums. Hoppe-Seyler Z PhysiolChem 323:236–248
Gloger M, Decker K (1969). Zum Mechanismus der Induktion nico-tinabbauender Enzyme inArthrobacter oxidans.Z Naturforsch246:1016–1025
Goodfellow M, Cross T (1984) V. Classification. In: Goodfellow M,Mordarski M, Williams ST (eds) The Biology of the Actinomy-cetes. Academic Press, London, pp 7–164
Grether-Beck S, Igloi G, Pust S, Schilz E, Decker K, Brandsch R(1994) Structural analysis and molybdenum dependent expressionof the pAO1 encoded nicotine dehydrogenase genes ofArthrobac-ter nicotinovorans.Mol Microbiol 13:929–936
Grosjean H, Fiers W (1982). Preferential codon usage in prokaryoticgenes: The optimal codon-anticodon interaction energy and theselective codon usage in efficiently expressed genes. Gene 18:199–209
Hinkkanen A, Decker K (1983) Enzymatic determination of D-aminoacids using luminol-mediated chemiluminescence. Hoppe-Seyler ZPhysiol Chem 364:1549–1553
Hochstein LI, Rittenberg CS (1959) The bacterial oxidation of nicotine.II. The isolation of the first oxidative product and its identificationas (1)-6-hydroxynicotine. J Biol Chem 234:156–162
Hsu YP, Powell JF, Sims KB, Breakefield XO (1989) Molecular ge-netics of the monoamine oxidases J Neurochem 53:12–18
Hucker HB, Gilette JR, Brodie BB (1960) Enzymatic pathway for theformation of cotinine, a major metabolite of nicotine in rabbit liver.J Pharmacol Exp Ther 129:94–100
Ikemura T (1981) Correlation between the abundance ofE. coli transferRNAs and the occurrence of the respective codons in its proteingenes: A proposal for a synonymous codon choice that is optimalfor the E. coli translational system. J Mol Biol 151:389–409
Ishizuka H, Horinouchi S, Beppu T (1993) Putrescine oxidase ofMi-crococcus rubens:Primary structure and expression inEscherichiacoli. J Gen Microbiol 139:425–432
Jones D, Keddie RM (1992). The genusArthrobacter.In: Balows A,Truper HG, Dwarkin M, Harder W, Schleifer KH (eds) The pro-karyotes 2. Springer, Berlin, Heidelberg, New York, pp 1283–1299
Kisaki T, Tamaki E (1961) Phytochemical studies on the tobacco al-kaloids. I. Optical rotatory power of nornicotine. Arch BiochemBiophys 92:351–355
Klenk HP, et al. (1997) The complete genome sequence of the hyper-thermophilic, sulphate-reducing archaeonArchaeoglobus fulgidus.Nature 390:364–370
Kodama Y, Yamamoto H, Amano N, Amchi T (1992). Reclassificationof two strains ofArthrobacter oxydansand proposal ofArthrobac-ter nicotinovoranssp. nov. Int J Syst Bacteriol 42:234–239
Kretzer A, Frunzke K, Andreesen JR (1993) Catabolism of isonicotin-ate byMycobacteriumsp INA1. Extended description of the path-way and purification of the molybdoenzyme isonicotinate dehydro-genase. J Gen Microbiol 139:2763–2772
Labedan B, Riley M (1995) Gene products ofEscherichia coli:Se-quence comparisons and common ancestries. Mol Biol Evol 12:980–987
Lichtenstein AV, Moiseev VL, Zaboikin MM (1990) A procedure forDNA and RNA transfer to membrane filters avoiding weight-induced gel flattening. Anal Biochem 191:187–191
Mauch L, Kraub B, Brandsch R (1989) Growth stage-dependent ex-pression of 6-hydroxy-D-nicotine oxidase of the nicotine regulon ofArthrobacter oxidans.Arch Microbiol 152:95–99
Menendez C, Igloi GL, Henninger H, Brandsch R. (1995). A pAO1-encoded molybdopterin cofactor gene (moaA) ofArthrobacternicotinovorans:Characterization and site-directed mutagenesis ofthe encoded protein. Arch Microbiol 164:142–151
Menendez C, Igloi GL, Brandsch R (1997). IS 1473, a putative inser-tion sequence identified in the plasmid pAO1 fromArthrobacternicotinovorans:Isolation, characterization, and distribution amongArthrobacterspecies. Plasmid 37:35–41
Pearson WR (1991) Searching protein sequence libraries: Comparisonof the sensitivity and selectivity of the Smith-Waterman andFASTA algorithms. Genomics 11:635–650
Powell JF (1991) Molecular biological studies of monoamine oxidase:Structure and function. Biochem Soc Trans 19:199–201
Pust S, Vervoort J, Decker K, Bacher A, Mu¨ller F (1989)13C-, 15N- and31P-NMR studies on 6-hydroxy-L-nicotine oxidase fromArthro-bacter oxidans.Biochemistry 28:516–521
Richardson SH, Rittenberg SC (1961) The bacterial oxidation of nico-tine V. Identification of 2,6-dihydroxypseudooxynicotine as thethird oxidative product. J Biol Chem 236:964–967
Sambrook J, Fritsch EF, Maniatis T (1989a) Molecular cloning: Alaboratory manual. Cold Spring Harbor Laboratory Press, ColdSpring Harbor, NY, Section I, pp 1.38–1.39
Tanizawa K, Matsuzaki R, Shimizu E, Yorifuji T, Fukui T (1994)Cloning and sequencing of phenylethylamine oxidase fromArthro-bacter globiformisand implication of Tyr-382 as the precursor toits covalently bound quinone cofactor. Biochem Biophys Res Com-mun 199:1096–1102
Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: Improv-ing the sensitivity of progressive multiple sequence alignmentthrough sequence weighting, position-specific gap penalties andweight matrix choice. Nucleic Acids Res 22:4673–4680
