Cloning, Expression, and Sequence Analysis of the Three Genes Encoding Quinoline 2-Oxidoreductase, a Molybdenum-containing Hydroxylase from Pseudomonas putida 86*

The three genes coding for quinoline 2-oxidoreduc- tase (Qor) of Pseudomonas putida 86 were cloned and sequenced. The qor genes are clustered in the transcrip- tional order medium ( M ) small ( S ), large ( L ) and code for three subunits of 288 (QorM), 168 (QorS), and 788 (QorL) amino acids, respectively. Formation of active quinoline 2-oxidoreductase and degradation of quinoline oc- curred in a recombinant P. putida KT2440 clone. The amino acid sequences of Qor show significant homology to various prokaryotic molybdenum containing hydroxylases and to eukaryotic xanthine dehydrogenases. QorS contains two conserved motifs for [2Fe-2S] clusters. The binding motif for the N-terminal [2Fe-2S] clus- ter corresponds to the binding site of bacterial and chlo-roplast-type [2Fe-2S] ferredoxins, whereas the amino acid pattern of the internal [2Fe-2S] center apparently is a distinct feature of molybdenum-containing hydroxy- lases, showing no homology to any other described [2Fe-2S] binding motif. The medium subunit QorM presum- ably contains the FAD, but no conserved sequence areas or described motifs of FAD, NAD, NADP, or ATP binding were detected. Putative binding sites of the molybdopterin cytosine dinucleotide cofactor were detected in QorL Germany,

Here we report the cloning, expression, sequencing, and comparative sequence analysis of the qor genes encoding quinoline 2-oxidoreductase from P. putida 86. Based on alignments with other molybdenum-containing hydroxylases of both prokaryotic and eukaryotic origin, putative cofactor binding motifs are discussed.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Growth of P. putida 86, which had been isolated from soil of a coal tar refining factory (Ruetgerswerke, Castrop-Rauxel, Germany) by selective enrichment on quinoline as carbon source (Schwarz et al., 1988), was described previously (Tshisuaka et al., 1993). Escherichia coli S17-1 (Simon et al., 1983) was used to multiply the cosmid and for the mobilization of recombinant cosmids. For the construction of the genomic library, E. coli ED 8654 (Borck et al., 1976;Murray et al., 1977;Sambrook et al., 1989) was used as the host strain. Both E. coli strains were cultured in LB medium (Sambrook et al., 1989) at 37°C. E. coli clones containing recombinant cosmid DNA were grown in the presence of tetracycline (15 g/ml). In order to investigate expression of the qor genes, recombinant cosmid DNA in E. coli S17-1 was transferred in P. putida mt-2 KT2440 (Bagdasarian et al., 1981). P. putida mt-2 KT2440 was grown in LB medium at 30°C, and recombinant clones of strain KT2440 were cultured in mineral salt medium (Fetzner et al., 1989) with catechol (1 mM) as a source of carbon and tetracycline (50 g/ml) at 30°C. E. coli TG2 (Benen et al., 1989) was the host of recombinant pUC18 and pUC19 plasmid DNA (Vieira and Messing, 1982;Norrander et al., 1983). Such clones were cultured at 37°C on LB medium with ampicillin (100 g/ml). E. coli MV1190 (Vieira and Messing, 1987) grew on M9 mineral salt medium (Miller, 1972) in the presence of vitamin B1 at 37°C. E. coli MV1190 with recombinant M13mp18 or M13mp19 DNA (Norrander et al., 1983) was cultured in 2 ϫ YT medium (Miller, 1972) in the presence of ampicillin (100 g/ml).
Plasmids and Cosmid-In order to construct a genomic library of P. putida 86, the plasmid pCIB 119 was used. Plasmid pCIB 119 is a double cosmid and was a kind gift of Dr. Stephen T. Lam (Ciba-Geigy, Research Triangle Park, NC). It was constructed by cloning a fragment from plasmid c2XB (Bates and Swift, 1983) containing double cos sites into the broad-host range plasmid pRK 290 (Ditta et al., 1980). Plasmid vectors used for subcloning and sequencing were pUC18 or 19 and M13mp18 or 19 (Norrander et al., 1983;Vieira and Messing, 1987), respectively.
DNA Techniques-Genomic DNA of P. putida 86 was isolated according to Davis et al. (1980). Cosmid DNA was isolated by the method of Clewell and Helinski (1969). Plasmid DNA and recombinant cosmids and plasmids were isolated by the methods of Kieser (1984) or by alkaline lysis (Sambrook et al., 1989). Agarose gel electrophoresis, DNA restriction, DNA ligation, and dephosphorylation of DNA fragments were done according to the instruction manuals of the manufacturers (Boehringer Mannheim; Pharmacia Biotech Inc.) or by standard procedures (Sambrook et al., 1989). DNA fragments were isolated from agarose gels according to Tautz and Renz (1983) or by the instruction manual of the Geneclean-II kit (Dianova GmbH, Hamburg, Germany). Transformation of E. coli TG2 with recombinant pUC18 or pUC19 DNA and transformation of E. coli MV1190 with recombinant cosmid DNA was carried out using the CaCl 2 method of Mandel and Higa (1970). Conjugation of recombinant E. coli S17-1 clones with P. putida mt-2 KT2440 was performed according to Sambrook et al. (1989).
Construction of a Genomic Library-Genomic DNA of P. putida 86 was partially digested with Sau3AI. DNA fragments of about 14 -29 kb in size were isolated from an agarose gel and treated with alkaline phosphatase from shrimp (U. S. Biochemical Corp.). The dephosphorylated DNA fragments were ligated into the cosmid, and the resulting genomic library in E. coli ED8654 was screened using an oligonucleotide designated "corg" as a probe.
Hybridization-Colony blotting was performed as described by Grunstein and Hogness (1975). Colonies on nylon membranes (Hybond-N, Amersham Corp.) were subjected to cell lysis, and the released DNA was denatured with alkali and transferred onto the nylon membranes. Colony hybridization using corg as a DNA probe was performed to identify recombinant clones in the genomic library. Corg was an 8-fold degenerate 17-mer (ATG-ATG-AA(A/G)-CA(T/C)-GA(A/G)-GT) deduced from the N-terminal protein sequence (Met-Met-Lys-His-Glu-Val) of the large subunit , which was labeled with the DIG-3Ј-end-labeling kit as described by the supplier (Boehringer Mannheim). Hybridization according to Sambrook et al. (1989) was carried out at 44°C overnight. The membranes were stringently washed twice for 10 min at 49°C in 1 ϫ SSC containing 0.1% SDS. Immunological detection was performed with the DIG luminescent detection kit (Boehringer Mannheim).
Subcloned gene coding DNA fragments in pUC18 as well as in pUC19 plasmid vectors were identified by agarose gel electrophoresis, Southern blotting, and hybridization (Sambrook et al., 1989) with both the corg probe and a second DNA probe named "cork." The latter was a digoxigenin-labeled 32-fold degenerated 17-mer (ATG-CA(A/G)-GCX-CA(C/T)GA(A/G)-GA) deduced from the N-terminal protein sequence (Met-Gln-Ala-His-Glu-Glu) of the small subunit . The hybridization with corg was done as described above, the hybridization with cork was carried out at 49°C for 18 h, and stringent washes were performed with 1 ϫ SSC and 0.1% SDS at 49°C.
Expression of Quinoline 2-Oxidoreductase-Recombinant P. putida mt-2 KT2440 clones that showed positive hybridization signals were assayed for expression of the quinoline 2-oxidoreductase genes by testing for growth on mineral salt medium with quinoline as the only source of carbon, energy, and nitrogen. In the presence of succinate as an additional source of carbon, cometabolic conversion of quinoline was investigated. Growth was monitored by measuring the optical density at 600 nm. The conversion of quinoline to 2-oxo-1,2-dihydroquinoline was followed by recording UV-visible spectra in the range of 250 to 400 nm. The decrease of absorbance at 299 and 323 nm was indicative for consumption of quinoline and 2-oxo-1,2-dihydroquinoline, respectively (Bauder et al., 1990).
Enzyme Assay-The activity of quinoline 2-oxidoreductase in cell free extracts of recombinant P. putida KT2440 clones was determined spectrophotometrically, measuring the substrate-dependent reduction of p-iodonitrotetrazolium violet as described by Bauder et al. (1990). Cellfree extracts which showed enzymatic activity were separated by nondenaturing PAGE using 10% resolving and 4% stacking gels in the high pH system (Hames, 1990), and gels were immersed in a standard enzyme assay solution. Quinoline 2-oxidoreductase was visible on the polyacrylamide gel as red band due to p-iodonitrotetrazolium violetformazan formation.
DNA Sequencing and Sequence Analysis-DNA fragments to be sequenced were inserted into the vectors M13mp18 as well as in M13mp19, and E. coli MV1190 was transfected with them (Sambrook et al., 1989). DNA sequences were determined using the dideoxy-mediated chain termination procedure (Sanger et al., 1977) on single stranded templates with ␣-35 S-dATP and M13 universal primers. Overlapping single DNA fragments which included all three genes of the quinoline 2-oxidoreductase as well as flanking regions were isolated and sequenced in both directions. The sequencing reaction was performed according to the Deaza-G/A-T7-sequencing kit instruction manual (Pharmacia). Vertical PAGE and analysis of the 6% polyacrylamide gel was performed according to Sambrook et al. (1989). Computer analysis of qor/Qor and sequence comparisons with other molybdenum-containing hydroxylases were performed with the GENMON program (Gesellschaft fü r Biotechnologische Forschung mbH, Braunschweig, Germany) and the HUSAR 4.0 program package (EMBL, Heidelberg, Germany) which includes the GCG software version 7.1 of the University of Wisconsin (Devereux et al., 1984). The programs used were BFASTA to search for similar sequences and CLUSTAL W for the multiple sequence alignments of QorS, M, and L.
Nucleotide Sequence Accession Number-The DNA sequence presented in this report, encoding QorS, QorM, and QorL, has been deposited in the EMBL Nucleotide Sequence Library, Heidelberg, Germany, under accession no. X98131.

Cloning and Expression of Quinoline 2-Oxidoreductase
Genes-792 E. coli ED8654 clones of a genomic DNA library of P. putida 86 in plasmid pCIB 119 were screened for the three genes of quinoline 2-oxidoreductase using a digoxigenin labeled mixed oligonucleotide in colony hybridization. The DNA probe corresponded to six amino acids of the N terminus of the large subunit QorL. Four clones harboring inserts about 30 kb in size showed positive hybridization signals. However, attempts to detect catalytically active quinoline 2-oxidoreductase in the E. coli ED8654 clones failed. The recombinant cosmid DNA of each of the four clones was isolated and transferred to competent P. putida mt-2 KT2440. To check for expression of quinoline 2-oxidoreductase genes in the four recombinant P. putida KT2440 clones, cometabolic conversion of quinoline in the presence of succinate was tested. However, such a conversion only occurred with one clone, designated 13/42. In addition, only clone 13/42 grew on mineral salt medium with quinoline as the only source of carbon, energy and nitrogen. Spectrophotometric analysis of the culture supernatant of clone 13/42 showed that 2-oxo-1,2-dihydroquinoline was formed from quinoline. However, 2-oxo-1,2-dihydroquinoline accumulated only transiently and was consumed further. Since P. putida mt-2 KT2440 does not utilize 2-oxo-1,2-dihydroquinoline, this result indicates that not only the qor genes, but also genes encoding further enzymes of the "coumarin pathway" of quinoline degradation  (Schwarz et al., 1989;Shukla, 1989) are located on the recombinant cosmid of clone 13/42. The specific activity of quinoline 2-oxidoreductase in cell-free extracts of clone 13/42 and of wildtype P. putida 86 was 0.37 unit/mg protein and 0.72 unit/mg protein, respectively. In native PAGE, crude extract of clone 13/42 was separated, and after staining for activity (using p-iodonitrotetrazolium violet as cosubstrate), Qor was visible as a red band. Thus, only clone 13/42 contained the complete three genes of quinoline 2-oxidoreductase and was able to synthesize the active enzyme. The recombinant cosmid DNA of clone 13/42 was digested with EcoRI, followed by hybridization with the two DNA probes corg and cork. A DNA fragment of 11.5 kb in size was isolated from an agarose gel, ligated in pUC19, and transformed in E. coli TG2. Fig. 1 shows the restriction map of a 5-kb area of this 11.5-kb fragment which contains the complete three genes of quinoline 2-oxidoreductase as well as flanking regions.
Quinoline 2-Oxidoreductase Genes qorM, qorS, and qorL and its exact molecular mass is 30,650 Da, which agrees with the molecular mass of 30 kDa determined by SDS-PAGE (Bauder et al., 1990). The N-terminal amino acid sequence of the medium subunit determined by Edman degradation 2 totally matches with the amino acid sequence derived from the nucleotide sequence. The potential ribosome-binding site is located 14 bp upstream of the start codon, from position 671 to 680 (AAGTAGGTGA).
qorS starts at position 1538, 14 bp upstream of the end of qorM and in another reading frame, so there is a 14-bp overlapping region. The stop codon TGA is located at position 2050, and the length of qorS is 513 bp. The corresponding protein consists of 168 aa and has a molecular mass of 18,012 Da, coinciding with the mass of 20 kDa as estimated previously by SDS-PAGE (Bauder et al., 1990). The N-terminal amino acid sequence of the small subunit determined by Edman degradation  totally coincides with the amino acid sequence derived from the nucleotide sequence. The putative ribosome binding site (AAGGAGCT) (position 1526 -1533) is located 12 bp upstream of the start codon ATG.
qorL, the coding region of the large subunit, starts with the codon ATG at position 2044 and ends at position 4410. So the 2367-bp region is translated to a 788-aa peptide with a molecular mass of 84,113 Da, which agrees with the result (85 kDa) of previous analysis by SDS-PAGE (Bauder et al., 1990). The potential ribosome-binding site (AGGAG) (position 2032-2036) is located 12 bp upstream of the start codon ATG. The deduced N-terminal amino acid sequence totally agrees with the 16 aa of the N terminus determined by Edman degradation.
The G ϩ C content of qorS, qorM, and qorL is 62.82%, which matches with the G ϩ C content of 62.5% reported for the genome of P. putida biovar A (Palleroni, 1984).
Multiple Alignments-The amino acid sequences of the three subunits of quinoline 2-oxidoreductase derived from the nucleotide sequence were compared with the corresponding subunits of other molybdenum-containing hydroxylases which possess a monooxo-monosulfido-type molybdenum center (Figs. 3-5).
QorS shows about 25% homology 3 to the protein sequences aligned in Fig. 3. 25 aa (14.9%) are identical and 17 aa are similar 4 (10.1%) in all compared protein sequences. Eight of these 25 conserved aa are cysteines, and always four of them are arranged in two distinct motifs as described previously for several molybdenum-containing hydroxylases (Wootton et al., 1991;Hughes et al., 1992aHughes et al., , 1992bLehmann et al., 1995). The first four cysteines in QorS are in the positions 48, 53, 56, and 68, and the corresponding motif (C-X 4 -C-G-X-C-X n -C) is typical for the consensus binding site of bacterial and plant-type [2Fe-2S] ferredoxins. Instead of n ϭ 11 in all prokaryotic molybdenum-containing hydroxylases, there is n ϭ 21 in all described eukaryotic xanthine dehydrogenases and n ϭ 29 in most bacterial and plant ferredoxins. The [2Fe-2S] cluster coordinated by this motif is assumed to correspond to one of the Fe/S centers observed by EPR spectroscopy (Hughes et al., 1992a(Hughes et al., , 1992bTshisuaka et al., 1993). In all described sequences, the second four cysteines are arranged in the motif (C-G-X-C-X 31 -C-X-C) (cysteines in QorS are in the positions 107, 110, 142, and 144), and this motif presumably is the binding site of the other type of [2Fe-2S] center.
QorM probably harbors the FAD cofactor, but no conserved motif of a putative FAD binding site was found. The comparison of five medium subunits of various prokaryotic molybdenum-containing hydroxylases with corresponding domains of eight eukaryotic xanthine dehydrogenases shows only six identical aa (2.1%) and 26 aa which are similar (9.0%) (Fig. 4). Thus, the degree of homology among all sequences aligned is only 11.1%. Furthermore, none of the potential FAD-, NAD-or ATP-binding motifs known shows any homology to the medium subunits or domains, respectively. However, it may be remarkable to note that four of these six conserved aa are glycine residues, which were reported to be involved in FAD binding (Wierenga et al., 1986;Hanukoglu and Gutfinger, 1989;Eggink et al., 1990).
QorL, the large subunit, probably contains the binding site(s) of the pterin molybdenum cofactor, as proposed for other large subunits of prokaryotic molybdenum-containing hydroxylases and for to the C-terminal domains of the eukaryotic xanthine dehydrogenases (Wootton et al., 1991;Grether-Beck et al., 1994;Pearson et al., 1994;Glatigny and Scazzocchio, 1995;Schü bel et al., 1995). The comparison of nine eukaryotic xanthine dehydrogenases and five prokaryotic hydroxylases (QorL included) shows 47 aa (6.0%) which are absolutely conserved and 50 positions (6.3%) where the aligned amino acids are similar (Fig. 5), corresponding to a homology of 12.3% between the several sequences.

DISCUSSION
The qorM, S, and L genes encoding the first enzyme of the "coumarin pathway" of quinoline degradation by P. putida 86 were cloned and sequenced. Expression of the qorM, S, and L genes and formation of catalytically active quinoline 2-oxidoreductase was achieved in recombinant P. putida mt-2 KT2440 clone 13/42. Attempts to detect the active enzyme in the corresponding E. coli ED8654 clone possessing the same recombinant cosmid failed. It is known that Pseudomonas promotors as a rule are poorly recognized by E. coli RNA polymerase, because they show little homology to consensus sequences of E. coli promotors (Frantz and Chakrabarty, 1986;Jeenes et al., 1986). Another possible cause for the lack of enzyme formation in E. coli may be the difference in the GϩC content of Pseudomonas (67-68%) (Palleroni, 1984) and E. coli (50%) (Brenner, 1984), which may affect the codon-anticodon interaction in the E. coli host (Frantz and Chakrabarty, 1986;Soldati et al., 1987). However, if gene expression in E. coli took place, it might result in formation of inactive Qor that lacks the molybdenum molybdopterin cytosine dinucleotide cofactor. E. coli may be unable to synthesize the cytosine dinucleotide, since up to now, molybdenum enzymes in E. coli have been found to exclusively contain the molybdopterin guanine dinucleotide form of the pterin molybdenum cofactor (Rajagopalan FIG. 5-continued Quinoline 2-Oxidoreductase Genes qorM, qorS, and qorL 23075 and Johnson, 1992). P. putida mt-2 KT2440 13/42 utilized both quinoline and 2-oxo-1,2-dihydroquinoline, indicating that apart from the qorM, S, and L structural genes, further genes of the catabolic pathway are localized on the 30-kb DNA insert.
The sequenced DNA stretch was 4584 bp in size, and it contained three open reading frames. Based on comparisons with N-terminal amino acid sequences known and based on the gene expression studies, these three open reading frames were identified as the structural genes encoding the small (QorS), medium (QorM), and large (QorL) subunit of quinoline 2-oxi-doreductase. The transcriptional order of these three genes, 5Ј-qorM-S-L-3Ј (Fig. 1), corresponded to all other described structural genes of molybdenum-containing prokaryotic hydroxylases with LMS or L 2 M 2 S 2 structure.
We presume that both iron-sulfur centers are localized on the small, the FAD on the medium, and the pterin molybdenum cofactor as well as the substrate binding site on the large subunit, as discussed for the subunits or corresponding domains of other molybdenum-containing hydroxylases (Wootton et al., 1991;Grether-Beck et al., 1994;Pearson et al., 1994;Thoenes et al., 1994;Glatigny and Scazzocchio, 1995;Lehmann FIG. 5-continued Quinoline 2-Oxidoreductase Genes qorM, qorS, and qorL 23076 et al., 1995).
The binding motif for the internal [2Fe-2S] center, since it has not been described in any other iron-sulfur protein, apparently is a distinct feature of the molybdenum containing hydroxylases of pro-and eukaryotic organisms where it is an absolutely conserved motif (CX 2 CX 31 CXC) which contains the four other conserved cysteines. It is part of an area from position 82 to 140 (positions refer to Qor) which also shows significant homology to the other aligned sequences. This consensus sequence is (T(V/I ⅐ )E(G/D)(⌸/⌺)⌽⌽X 3-5 (H/S/N)(P/A)(⌸/⌺)QX 2 - Fig. 3). In both iron-sulfur centers it is presumed that all four conserved cysteines are ligands of the two iron atoms (Wootton et al., 1991;Hughes et al., 1992a). The other homologous amino acids may be functional for the correct folding of the protein to bind the iron-sulfur centers, or they may participate in binding of the pterin molybdenum cofactor or the FAD. In this context it is important to note that it was not feasible to define distinct binding motifs for the pterin molybdenum cofactor or FAD on the other subunits of quinoline 2-oxidoreductase (see below).
In analogy to the medium domain or subunit of other molybdenum-containing hydroxylases described, the medium subunit QorM should contain the binding site for FAD (Amaya et al., 1990;Wootton et al., 1991;Hughes et al., 1992a;Grether-Beck et al., 1994;Pearson et al., 1994). This assumption is based on the fact that all flavin-containing hydroxylases have got a medium subunit and that the homodimeric enzymes which lack FAD do not possess a medium subunit (Fetzner and Lingens, 1993;Lehmann et al., 1994Lehmann et al., , 1995. Moreover, aldehyde oxidoreductase (MOP) from Desulfovibrio gigas, which lacks both FAD and aa sequences corresponding to the medium subunits and which is the only molybdenum containing hydroxylase whose crystal structure is determined, does not correspond in its structural features to any known FAD binding subunits or domains (Romão et al., 1995). The sequence alignment of these medium subunits of various prokaryotic molybdenum-containing hydroxylases and the corresponding domains of eukaryotic xanthine dehydrogenases (Fig. 4) shows a high degree of homology between them, which ranged from 56.18%, with nicotine dehydrogenase of Arthrobacter nicotinovorans (Grether-Beck et al., 1994), to 40.97%, with xanthine dehydrogenase from Rattus norvegicus (Amaya et al., 1990). It is important to note that in 13 out of 14 aligned sequences, only 11 aa residues were absolutely or nearly conserved and that 4 of these 11 aa were glycine residues, which generally are thought to be important in FAD binding. Several different FAD-, NAD-/NADP-, and ADP-binding motifs were described by Rice et al. (1984), Wierenga et al. (1986), Eggink et al. (1990) and Hanukoglu and Gutfinger (1989). All these FAD binding motifs comprise G-rich sequences, such as (GXGX 2 GX 3 A) or (GXGX 2 GX 3 G). A similar motif, (GXGX 2 AX 3 A), is a site for NADP-binding. None of these or other known binding motifs were detected in QorM or in the sequences of nine eukaryotic xanthine dehydrogenases and five prokaryotic molybdenumcontaining hydroxylases. Nevertheless, there are studies on Drosophila melanogaster mutants with xanthine dehydrogen-ase variants suggesting that the medium domain contains the FAD binding site: Hughes et al. (1992a) prepared point mutations that were localized in the area between amino acid residues 348 and 357 (positions refer to xanthine dehydrogenase of D. melanogaster). In these flavin mutants, the electron transfer between the molybdenum center and FAD was blocked (Hughes et al., 1992a), indicating that the altered area somehow is correlated with FAD binding. This area of xanthine dehydrogenase is part of the medium domain and corresponds to the amino acid positions 113-122 in QorM. Its alignment with nine eukaryotic and five prokaryotic molybdenum-containing enzymes shows significant homologies. This consensus sequence described before (bold letters) and the conserved flanking areas (normal letters) are: AX 2 Q(⌸/⌺)(74/⍀)X 2 (⌬/⌽)-X 2 (⌬/⌽)(G⌽⌺X[inb]2(⌬/⌽)⌽(⌬/⌽)X⌽)(⌿/⌽)X⌽ (Fig. 4). However, strong evidence that FAD was present in these FAD mutants of D. melanogaster (Hughes et al., 1992a) suggested that the altered amino acids are not the only residues involved in FAD binding. Data on FAD binding in prokaryotic molybdoiron/sulfur-flavoproteins are not yet available. Thus, the mode of FAD binding in enzymes belonging to the family of molybdenum-containing hydroxylases is still unknown.
The large subunit of quinoline 2-oxidoreductase, QorL, probably contains the molybdenum molybdopterin cytosine dinucleotide and the substrate binding site. The 14 compared aa sequences show significant homology, ranging from 63.2%, with nicotine dehydrogenase from A. nicotinovorans (Grether-Beck et al., 1994), to 41.3%, with xanthine dehydrogenase from R. norvegicus (Amaya et al., 1990). The comparison in Fig. 5 shows 46 positions which were absolutely conserved. 37.0% of these amino acids are glycines, and 15.2% are alanine residues.
Up to now, the only molybdenum-containing hydroxylase whose crystal structure was determined is aldehyde oxidoreductase (MOP) from the sulfate reducer D. gigas (Romão et al., 1995). Each subunit of the homodimeric enzyme contains two different [2Fe-2S] centers and molybdenum molybdopterin cytosine dinucleotide. The C-terminal domain of the 907-aa subunit of MOP shows significant homology to QorL (42.0%) and to all other large subunits or corresponding domains listed in Fig. 5. Romão et al. (1995) detected three molybdopterin binding segments and two dinucleotide binding segments in MOP. The first molybdopterin-contacting segment, which was already discussed in previous reports to be involved in binding the molybdenum cofactor (Grether-Beck et al., 1994;Pearson et al., 1994;Lehmann et al., 1995;Schü bel et al., 1995), is the most conserved sequence of the five segments detected in MOP by Romão et al. (1995). In QorL, the corresponding segment starts at position 252 and ends at position 258. The consensus sequence of all compared sequences is: (GG(G 13 /T 1 )FG(G 9 /N 2 / Q 2 /Y 1 )K) (Fig. 5). Based on its homology to the consensus sequence (GXGXXG), this segment was discussed as binding site for the pyrophosphate moiety of a dinucleotide (Amaya et al., 1990;Pearson et al., 1994;Schü bel et al., 1995), but no prokaryotic enzyme contains the last glycine residue of the (GXGXXG) motif. This agrees to the observation that the substitution of the first glycine residue by glutamic acid in the corresponding region of xanthine dehydrogenase from D. melanogaster did not cause significant diminutions in xanthine dehydrogenase electrontransfer activities, indicating that the structural change only is a subtle one, probably not affecting binding of the molybdenum cofactor (Hughes et al., 1992a). However, Romã o et al. (1995) showed that, in MOP, this segment contacts molybdopterin by use of Phe-421 and Gly-422, which are absolutely conserved in all aligned sequences (Fig. 5).
Using site-directed mutagenesis, we plan to investigate putatively crucial amino acids of these five amino acid segments proposed to be involved in binding the molybdenum molybdopterin cytosine dinucleotide in quinoline 2-oxidoreductase. Further studies also are needed to characterize the substrate binding site of quinoline 2-oxidoreductase and to elucidate the mode of FAD binding in molybdo-iron/sulfur-flavoproteins belonging to the family of molybdenum-containing hydroxylases (oxotransferases).