The flavin-containing monooxygenase enzymes expressed in rabbit liver and lung are products of related but distinctly different genes.

Preparations of mRNA isolated from rabbit lung and liver were used in the construction of libraries that were screened for cDNAs encoding the pulmonary or hepatic isozyme of the flavin-containing monooxygenase. The hepatic library was screened with cDNA encoding the flavin-containing monooxygenase expressed in pig liver, and a clone containing a 2.0-kilobase insert was detected and isolated. This cDNA insert encoded a protein of 535 amino acids with a primary structure 87% identical to that of the pig flavin-containing monooxygenase. The pulmonary library was screened with polyclonal antibodies to the flavin-containing monooxygenase expressed in rabbit lung, and a clone containing a 2.6-kilobase insert was detected and isolated. Although the protein encoded by this insert also contained 535 amino acids, its primary sequence was only 56% identical to that of the liver enzyme. The sequences of several peptides obtained by digestion of the purified rabbit pulmonary flavin-containing monooxygenase with trypsin matched exactly with sequences derived from the cDNA structure. Tissue-specific distribution of mRNA for the hepatic and pulmonary isozymes of the flavin-containing monooxygenase was consistent with the distribution of protein, an indication that expression of flavin-containing monooxygenase is controlled at the level of transcription. Analysis of genomic DNA indicates that both the hepatic and pulmonary enzymes may be products of single genes.


Preparations
of mRNA isolated from rabbit lung and liver were used in the construction of libraries that were screened for cDNAs encoding the pulmonary or hepatic isozyme of the flavin-containing monooxygenase. The hepatic library was screened with cDNA encoding the flavin-containing monooxygenase expressed in pig liver, and a clone containing a 2.0kilobase insert was detected and isolated. This cDNA insert encoded a protein of 535 amino acids with a primary structure 67% identical to that of the pig flavin-containing monooxygenase.
The pulmonary library was screened with polyclonal antibodies to the flavin-containing monooxygenase expressed in rabbit lung, and a clone containing a 2.6-kilobase insert was detected and isolated.
Although the protein encoded by this insert also contained 535 amino acids, its primary sequence was only 56% identical to that of the liver enzyme. The sequences of several peptides obtained by digestion of the purified rabbit pulmonary flavin-containing monooxygenase with trypsin matched exactly with sequences derived from the cDNA structure. Tissue-specific distribution of mRNA for the hepatic and pulmonary isozymes of the flavin-containing monooxygenase was consistent with the distribution of protein, an indication that expression of flavin-containing monooxygenase is controlled at the level of transcription. Analysis of genomic DNA indicates that both the hepatic and pulmonary enzymes may be products of single genes.
The toxic effects of a wide variety of environmental chemicals can be highly selective with respect to species, tissue, and cell type. Most of these chemicals must undergo oxidative metabolism before their reactivities can be expressed, and the distribution of the enzymes catalyzing this "activation" may contribute to the biological localization of toxic responses. The lung is a good example of a tissue that is selectively damaged by a number of toxic agents (1). Within the lung, the nonciliated bronchiolar (Clara) cell is particularly vulnerable (2), due likely to the relative abundance of specific * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 505275.
Y To whom correspondence should be sent. 11 Supported by Public Health Service Training Grant ES-07046.
isozymes of cytochrome P-450 (3). This association has been most clearly demonstrated in the case of the cell-selective toxic effects of certain substituted furans, such as 4-ipomeanol (4). Identical enzymes are also expressed in liver (5, 6) but at concentrations that are apparently low enough to preclude a toxic response in hepatocytes.
An additional enzyme of possible importance in differences between the responses of lung and liver to some toxic agents is the flavin-containing monooxygenase (FMO,' EC 1.14.-13.8), an enzyme that catalyzes the oxidation of numerous nitrogen-, sulfur-, and phosphorus-containing drugs, pesticides, and industrial chemicals (7,8). As with cytochrome P-450, some reactions catalyzed by FM0 result in the formation of reactive intermediates that are potentially toxic (9). The function of the FM0 with respect to endogenous substrates is not known but may involve maintenance of the cellular thiol to disulfide ratio by oxidation of cysteamine to cystamine (10).
The majority of what is known about the molecular characteristics of the FM0 comes from studies of the enzyme purified from pig liver. However, expression of distinct isozymes in liver and lung has been described recently. This possibility was first noted with the discovery that FM0 activity in detergent-solubilized preparations from rabbit lung but not from liver could be stimulated by Hg2+ (11). More substantial evidence was in a subsequent report that imipramine, which is metabolized by the FM0 in liver, does not undergo N-oxidation in rabbit pulmonary microsomal preparations that are active with other FM0 substrates (12). Finally, characterization of the FM0 purified from rabbit lung provided direct evidence of major differences between the "hepatic" and "pulmonary" forms of the enzyme (13, 14). The extent to which the structures of the pulmonary and hepatic forms of the FM0 differ is suggested by their lack of immunochemical cross-reactivity (13,14), the stability of the pulmonary enzyme when subjected to temperatures or concentrations of ionic detergents that denature the hepatic enzyme (13), and the ability of the pulmonary enzyme to metabolize several primary arylamines that are not substrates for the hepatic enzyme (15,16). The molecular bases for these differences are not understood, primarily because of a lack of information about the structures of these proteins. Recently, however, we have derived the primary sequence of the FM0 expressed in pig liver (17), and we now report the primary sequences of the FM0 isozymes expressed in rabbit lung and liver. The results presented show clearly that the rabbit enzymes are products of related but distinct genes. The structures of the FM0 enzymes expressed in rabbit liver and lung are compared with each other and with that of the FM0 expressed in pig liver, and a number of interesting differences and similarities are discussed.

Animals,
Tissues, and Subcellular Preparations-Adult male New Zealand White rabbits (Dutchland Farms, Denver, PA) were used. Tissues (liver, lungs, and kidneys) were removed immediately after the animals were killed by suffocation with CO,. RNA and DNA were isolated from tissues that were frozen in liquid nitrogen and stored at -70 "C, and microsomal fractions were prepared from fresh tissue by standard procedures (18). Construction and Screening of cDNA Libraries-Preparations of hepatic and pulmonary mRNA were isolated by a modification (5) of the methods of Chirgwin et al. (19) and Glisin et al. (20). A liver cDNA library was constructed in Xgtll (6) and screened by plaque hybridization with nick-translated cDNA encoding the FM0 expressed in pig liver (17). Similarly, a pulmonary cDNA library was constructed in XZap II (Stratagene, La Jolla, CA) after selection of cDNA species between 1.3 and 2.6 kb. The pulmonary library was screened with antibodies to the pulmonary FM0 (14). Recombinant clones were identified by detection (21,22) of antigen fi-galactosidase fusion product following transfer of plaques to nitrocellulose (23). Seven clones were purified and isolated as subclones in Bluescript plasmid (PBS; Stratagene).

Nucleotide
Sequencing-Purified inserts from pulmonary (pFMO-1) or hepatic (hFMO-1) clones were fragmented by sonication (24) or restriction (25) and sequenced by the dideoxy chain termination method (26,27). Additional pulmonary clones were sequenced using oligonucleotide primers complementary to the T3 and T7 promoter regions flanking the inserts (28). The sequence data were analyzed with the sequence analysis software package from the University of Wisconsin Computer Group (29), and gap alignments were constructed with the algorithm developed by Wilbur and Lipman (30).

Analysis
of mRNA and Genomic DNA-Messenger RNA purified from liver, lung, and kidney was separated by electrophoresis in agarose (FMC BioProducts, Rockland, ME) gels (1%) containing 0.5% methyl mercury (31) and transferred overnight to nylon membranes (Nytran; Schleicher & Schuell). The nylon sheets were baked for 60 min at 80 "C, treated with prehybridization solution (6 x SSC, 250 pg/ml salmon sperm DNA, 5 x Denhardt's, and 0.1% SDS) for 2 h at 42 "C, and reacted with pFMO-1 or hFMO-1 (10' cpm/rg) in hybridization solution (6 x SSC, 100 pg/ml salmon sperm DNA, 4 x Denhardt's, 0.5% SDS, and 50% formamide) overnight at 42 "C (32). The open star represents a 55-nucleotide region of pFMO-1 which contains three putative polyadenylation signals (AATAAA). The 5' and 3' sequences of the cDNA inserts from the other positive clones (pFMO-2-pFMO-7) are represented by heavy arrows. Check marks indicate the locations of identical base changes found in clones pFMO-2 and -4. The solid stars indicate the position of a single base deletion found in clones pFMO-2, -3, and -6, and the asterisk denotes the location of the unique sequence found at the 3' end of clone pFMO-7.
The cDNA was labeled by the random primer technique (33) with a commercial kit (Boehringer Mannheim). Blots were washed twice for 15 min at room temperature in 0.1 x SSC containing 0.1% SDS and then subjected to autoradiography.
Genomic DNA was isolated from rabbit liver (34,35), digested with various restriction enzymes, electrophoresed in 0.7% agarose, transferred to nylon membranes, and analyzed by the method of Southern (36). Following transfer, the membranes were rinsed for 5 min in 5 X SSC at 65 "C, baked for 1 h at 80 "C, treated with prehybridization solution (6 x SSC, 10 x Denhardt's, 50 rg/ml DNA, and 1% SDS) for 2 h at 42 "C and then reacted with full-length, 5' and 3' pFMO-1 and hFMO-1 probes. The reactions were carried out overnight at 52 'C in hybridization solution (6 x SSC, 50 pg/ml salmon sperm DNA, 1% SDS, and 50% formamide). Hybridized blots were washed twice for 30 min at 65 "C in 0.1 x SSC containing 0.1% SDS and then subjected to autoradiography.

Sequencing of FM0 Purified
from Rabbit Lung-The FM0 expressed in rabbit lung was purified by the method of Tynes et al. (14). Purified protein (2 nmol) in elution buffer (10 mM phosphate, pH 7.6, 10 mM EDTA, 0.1 mM dithiothreitol, 20% glycerol, 0.25% Emulgen 911, and 1 mM NADPH) was placed in a filter-centrifuge tube (Amicon Centricon-30) and washed sequentially with 0.015 M Tris (pH 8.6), 5 M guanidinium HCl, and 0.015 M Tris + 5 M guanidinium HCl. The protein was then treated with 0.5 mg of dithioerythritol, incubated for 1 h at 37 "C, treated with 1.5 mg of iodoacetic acid, and finally incubated in the dark at room temperature for 1 h. The alkylated protein was washed with 0.015 M Tris, pH 8.6, resuspended in 2% NH,HCOa, and digested with trypsin (10 rg) at 37 "C for 1 h, after which additional trypsin (5 rg) was added and the incubation continued for 30 min. The resulting peptides were separated on a 300A C-18 column using the following elution scheme: 90% HPO, 10% acetonitrile for 1 min followed by a linear change to 50% HzO, 50% ACN over 59 min. The flow rate was 0.9 ml/min, and peptide elution was monitored at 214 and 254 nm. Fractions were collected manually, and sequencing was carried out with a gas phase instrument (Applied Biosystems 407A).
Sequences of 200 or more bases were also obtained from the 5' and 3' ends of six additional clones isolated from the pulmonary library (Fig. 2). The 5' sequences of four of these clones (pFMO-3, 5, 6, and 7) were identical to those of the corresponding regions of pFMO-1. In contrast, the 5' ends of both pFMO-2 and pFMO-4 differed from pFMO-1 at two positions; each contained a T in place of G at position 440 of pFMO-1 and a G in place of C at position 485 (Fig. 2). The 3' sequences of pFMO-2 and 4, as well as that of pFMO-6, also differed from that of pFMO-1 by the occurrence of a single deletion in a series of seven consecutive adenosines (bases 2109-2115) present in pFMO-1 (Fig.  2). The 3' sequence of pFMO-7 was also different in that it terminated in a sequence (CAAAA) that may represent part of a poly(A) tail for a transcript shorter than that of pFMO-1; two of the three potential poly(A) signals found 385 nucleotides upstream from the 3' end of pFMO-1 are located 16 and 22 bases from the 3' terminus of pFMO-7. No differences be-tween the 3' sequences of pFMO-4 or 5 and those of the corresponding regions of pFMO-1 were found (Fig. 2).

Isolation of cDNA Clones Encoding the FM0 Expressed in
Rabbit Liver-The cDNA library constructed with mRNA isolated from rabbit liver was screened with cDNA encoding the FM0 expressed in pig liver (17). Six clones were detected, and one insert (hFMO-1) was sequenced by the strategy shown in Fig. 4. The complete sequence of hFMO-1 contains 2046 bases with an open reading frame of 1605 bases, a 5'flanking region of 48 bases, and a 3'-flanking region of 390 bases. The sequence of hFMO-1, aligned with that of pFMO-1, is shown in Fig. 3.
Amino Acid Sequences Derived for the Forms of FM0 Expressed in Rabbit Liver and Lung-The primary sequences derived for the FM0 enzymes expressed in rabbit liver and lung are shown in Fig. 5, along with the sequence of the FM0 expressed in pig liver (17). Both rabbit proteins contain 535 amino acids, compared with 532 for the pig enzyme. The calculated molecular mass of the liver FM0 is 59,841 daltons and of the lung FM0 is 61,145 daltons. The base substitutions seen with pulmonary pFMO-2 and pFMO-4, relative to the sequence of pFMO-1, form codons for different amino acids: serine for alanine at position 120 and glutamine for glutamic acid at position 136.
Common structural features of the FM0 enzymes expressed in rabbit liver and lung and in pig liver are shown in Figs. 5 and 6 and Table I. The sequences of the rabbit enzymes are 56% identical with 12 common peptides of 5 or more amino acids. The largest of these peptides begins at position 325 of the liver sequence and contains 17 amino acids. The two liver enzymes are much more similar (87% identity) and share several large areas of absolute identity, including the first 69 N-terminal amino acids and a 52-amino acid peptide beginning at position 328 of the rabbit liver sequence. Two 14amino acid peptides, beginning at positions 166 and 328 of the rabbit liver sequence, are common to all three proteins. Alignment of the three amino acid sequences also reveals that the pig liver and rabbit lung enzymes share a 3-residue gap corresponding to positions 317, 318, and 319 of the rabbit liver sequence (Fig. 5).
Putative FAD and NADP pyrophosphate-binding domains are located at the same positions in the primary sequences of all three FM0 enzymes ( Table I). The FAD pyrophosphatebinding site contains a highly conserved consensus sequence of GxGxxG beginning at residue 9, and the NADP site contains a similar sequence (GxGxxG/A) beginning at residue 191. Similar binding domains, typically separated by about 175 residues, are present in a number of other enzymes, some of which (38-41) are included in Table I.
The hydropathy indices (42) of the proteins are shown in Fig. 6. As expected from the high identity of their primary sequences, the profiles of the rabbit liver and pig liver proteins are nearly superimposable. Somewhat unexpected, however, is the marked similarity between the profile of the rabbit lung FM0 and that of the liver enzymes. In fact, the only notable differences between the lung and liver profiles are a hydrophilic peptide (centered at residue 418) and a hydrophobic peptide (centered at residue 461) present in the liver but not in the lung proteins. Prediction of membrane associated peptides by several methods (43-45) yields virtually identical results with the sequences of the liver FMOs and very similar results with the sequences of the liver and lung enzymes (Fig.  6).

Peptide Sequences Obtained from Purified
Rabbit Lung FMO-Because attempts to sequence intact rabbit lung FM0 were unsuccessful (due likely to a blocked N terminus), the purified protein was digested with trypsin and the peptides separated by reverse-phase HPLC (Fig. 7). Four peptides (labeled 7, 17, 25, and 45) were selected for sequencing by an automated Edman's procedure. The sequence of each peptide was found to be identical to a sequence present in the primary structure of the protein derived from the pulmonary cDNA clone (pFMO-l), and each sequence was flanked by an appropriate cleavage residue (Fig. 5).   and renal mRNA were separated by electrophoresis in agarose gels (1%) containing methyl mercury (0.5%), transferred to nylon membranes, and hybridized with cDNA encoding hepatic (hFMO-1) and pulmonary (pFMO-1) FMO. Single bands of mRNA (2.6 kb) were detected in the hepatic and renal samples by hybridization with hFMO-1; no pulmonary mRNA was detected with this probe (Fig. 8). In contrast, no hepatic mRNA was detected with pFMO-1, whereas four distinct bands of mRNA (2.4, 2.6, 4.8, and 6.0 kb) were detected in the pulmonary and renal samples (Fig. 8). The relative intensities of these four bands were the same in 12 pulmonary mRNA samples prepared from individual rabbits (not shown). The same relative intensities were also obtained at different hybridization temperatures (42, 52, and 65 "C) and with 5' (bases 1-521) and 3' (bases 1368-2611) probes prepared from pFMO-1 by restriction with PstI (not shown). The 3' and 5' pulmonary pFMO-1 probes described above and 5' (bases l-735) and 3' (bases 736-2046) hepatic probes, prepared by restriction of hFMO-1 with KpnI, along with fulllength pFMO-1 and hFMO-1 were used to analyze genomic DNA isolated from liver ( Fig. 9, a and b). The DNA was restricted by incubation with AJcoI, HindIII, EcoRI, or PstI, and the fragments were separated electrophoretically in agarose gels (0.7%) and transferred to nylon membranes. Hybridization to the 3' probes provided the least complicated results in that single bands were observed in all but one case. The exception was that two bands were obtained when DNA restricted with EcoRI was hybridized to the 3' hepatic probe (Fig. 9b). The larger (-9.0 kb) of these fragments was also detected by the 5' hepatic probe and appeared to be due to cross-hybridization with the 3' end of the gene encoding the pulmonary enzyme (Fig. 9a). The 5' pulmonary probe hybridized with two (NcoI and HindIII) or three (EcoRI and P&I) DNA fragments, some of which were also detected with the 3' probe. The full-length pulmonary pFMO-1 hybridized to the same fragments as the 3' and 5' probes, as well as with several others assumed to represent the area of the gene associated with bases 522-1367 of pFMO-1. With NcoI, for example, fragments of 3.0 (5'), 4.0 (3'), 4.4 (internal), and 5.0 kb (5') were detected. Two internal fragments were detected with Hind111 and two with PstI. As expected, the full-length hepatic hFMO-1, did not detect any fragments in addition to those detected by the 3' and 5' hepatic hFMO-1 probes.

DISCUSSION
Consideration of the substrate specificities, immunochemical properties, and certain physical parameters of the FM0 expressed in liver and lung suggests that these enzymes might be more analogous than homologous. Their catalytic activities do differ (12-16) but remarkably little in view of the disparities between their immunochemical properties (13, 14, 46), stabilities (13), and responses to Hg2+ (11). As a first step in ascertaining the structural bases for the similar overall catalytic activities, specific substrate differences, and distinct physical properties of these enzymes, we have derived their primary structures from the nucleotide sequences of cloned cDNAs. Several structural characteristics of the FMOs expressed in rabbit liver and lung and in pig liver are compared in Fig. 10.
The amino acid sequences derived for the FM0 enzymes expressed in lung and liver are 56% identical, a clear indication of homology. However, the extent of the differences between these primary structures is indicative of an evolutionary divergence that occurred prior to speciation. This is apparent when the much higher identity (87%) between the FM0 enzymes expressed in livers of different species (rabbit and pig) is considered (Fig. 10). In this respect, it is noteworthy that the sequences of the rabbit lung and pig liver proteins (57% identity) have certain similarities not shared with the rabbit liver enzyme. Notwithstanding the marked differences between their sequences, the structures of the FM0 enzymes expressed in liver and lung have a number of common features: their pyrophosphate-binding domains are nearly identical, and they have five putative membrane-associated regions in common, including two in areas that have little primary sequence identity (Fig. 10) pulmonary and hepatic enzymes is likely greater than might be predicted from the percent identity of their primary sequences.
The marked differences between the primary structures of the FM0 proteins expressed in lung and liver are sufficient evidence for concluding that these enzymes are products of distinct genes, a conclusion confirmed by analysis of genomic DNA. Although several restriction enzymes yield DNA fragments associated with both enzymes, these are readily explained by cross-hybridization of the probes with genomic DNA containing one or more of the eight identical nucleotide sequences (11-21 bases in length) present in the coding regions of the lung and liver sequences. The fragmentation patterns observed with genomic DNA indicate that the pulmonary and hepatic enzymes may not be encoded for by more than one gene each. In any case, hybridization of genomic DNA with lung or liver cDNA presents the same general picture and offers no explanation for why the pulmonary, but not the hepatic, enzyme is associated with multiple populations of mRNA.
Initially, detection of multiple bands of pulmonary mRNA upon hybridization with pFMO-1 was thought to be related in some way to the multiple bands of pulmonary microsomal protein detected with antibodies to the FM0 expressed in lung (46). This idea was reinforced by detection of the same mRNA and protein bands in samples from kidney (this work and 46). However, in contrast to expression of the protein bands in three distinct phenotypes (46), identical mRNA bands were present in pulmonary samples from 12 individuals examined.
Also, results of preliminary studies reveal that pFMO-1 hybridizes to a single band of pulmonary mRNA from guinea pig (not shown), a second species for which multiple protein bands are associated with the FM0 expressed in lung (46). Based on studies with 3' and 5' probes and analyses carried out at different temperatures, cross-hybridization of pFMO-I with similar but distinctly different mRNAs was also ruled out as an explanation for the multiple bands seen with pulmonary and renal samples. These bands are likely derived from the same transcript by alternative splicing or variable 3' processing, the latter being consistent with the presence of multiple putative polyadenylation sites (47). The largest band, probably an unusually stable or only partially processed transcript, is not represented in our cDNA library because of the size restrictions we used. On the other hand, the smallest band (about 2.4 kb) is likely represented by clone pFMO-7, which terminates in CAAAA, a sequence that is not present in the other clones and occurs 16 bases downstream from a polyadenylation signal. The existence of allelic variants of the FM0 expressed in lung may explain, at least in part, the multiple protein bands observed. Data from sequencing show at least two populations of cDNAs which vary by less than 1% in the regions examined. However, the two base differences detected in the coding regions of these cDNAs do result in amino acid changes. Such variability between proteins having the same number of residues has little effect on monomeric molecular weight but can alter mobility in polyacrylamides gels in the presence of SDS, as has been observed with rat cytochrome P-450 isozymes b and e (48). The contribution of allelic variants and the potential involvement of post-translational modification to formation of the multiple proteins bands associated with the FM0 expressed in lung are being investigated further. In conclusion, we have derived sequences for the FM0 enzymes expressed in rabbit liver and lung. The relationship between these proteins suggests an evolutionary divergence that took place prior to speciation. The lung, but not the liver, FM0 is associated with multiple species of mRNA. These mRNAs do not appear to be related to the multiple protein bands detected with antibodies to the FM0 expressed in lung; however, their exact nature is not yet understood.
In the next stage of our work, we will attempt to define the structural elements responsible for differences between the catalytic and physical properties of the FMOs expressed in lung and liver.
Note Added in Proof-While this manuscript was in preparation, evidence for two related forms of flavin-containing monooxygenase in rabbit liver was reported by 0~01s (Ozols, J. (1989) Biochem. Biophys.
Res. Commun. 163, 49-55). Sequence data obtained from two purified proteins were given for two regions, one including or near the amino terminus and one internal region of about 40 residues. The peptides from one protein (form 1) are nearly identical with peptides found in the sequence we report for the hepatic FMO. On the other hand, the sequences from the second protein (form 2), which are 80% (amino-terminal) and 60% (internal peptide) identical with those from form 1, do not match with sequences from the pulmonary FMO.
Thus, it appears that the flavin-containing monooxygenase may consist of at least three related but distinct gene products. 1