Toll-like receptor genetic variants are associated with Gram-negative infections in VLBW infants.

OBJECTIVE: To test the hypothesis that single-nucleotide polymorphisms (SNPs) in Toll-like receptor (TLR) genes alter susceptibility to bacterial infections and modulate white blood cell (WBC) counts during infections in very low birth weight (VLBW) infants (birth weight <1500 g). STUDY DESIGN: VLBW infants recruited in a multicenter study were genotyped for nine functional TLR SNPs and associations between SNPs and infection rates examined. WBC counts obtained during infections were compared among infants with and without SNPs. RESULT: In our cohort (n=408), 90 infants developed bacterial infections. Presence of TLR4 (rs4986790 and rs4986791) variants were associated with Gram-negative (G-ve) infections. Female infants heterozygous for the X-linked IRAK1 (rs1059703) SNP had less G-ve infections. In regression models controlling for confounders, the TLR4 (rs4986790) SNP was associated with increased G-ve infections. The TLR5 (rs5744105) variant was associated with elevated WBC counts during infections. CONCLUSION: TLR genetic variants can contribute to increased risk of bacterial infections and altered immune responses in VLBW infants.

Developmental expression of drug metabolizing enzymes: impact on disposition in neonates and young children.

Profound changes in drug metabolizing enzyme expression occurs during development that impacts drug efficacy and the risk of adverse events in the neonate and young child. A review of our current knowledge suggests individual hepatic drug metabolizing enzymes can be categorized into one of three classes based on developmental trajectories. The time frame for the perinatal changes observed for both Class 1 and Class 3 enzymes varies considerably between different enzymes. However, for a given enzyme, significant interindividual variation is observed in the timing of the perinatal changes, creating windows of hypervariability. Genetic variation clearly impacts drug disposition in children. However, developmental factors can dominate pharmacogenetic factors. Thus, a major challenge in applying pharmacogenomics to improve pediatric drug safety is determining at what age functional genetic variants identified in adults become a major determinant of expression in children. Developmental and genetic data on drug metabolizing enzyme ontogeny, as well as age-dependent changes in other physiological factors impacting drug disposition, can be integrated into physiologically-based pharmacokinetic models. Such models have proven useful in predicting the range of expected metabolic capacities at a given age.

Modeling interchild differences in pharmacokinetics on the basis of subject-specific data on physiology and hepatic CYP2E1 levels: a case study with toluene.

The objective of the present study was to evaluate the magnitude of interindividual variability in the internal dose of toluene in children of various age groups, on the basis of subject-specific hepatic CYP2E1 content and physiology. The methodology involved the use of a previously validated physiologically based pharmacokinetic (PBPK) model, in which the intrinsic clearance for hepatic metabolism (CL(int)) was expressed in terms of the CYP2E1 content. The adult toluene PBPK model, with enzyme content-normalized CL(int), facilitated the calculation of child-specific CL(int) based on knowledge of hepatic CYP2E1 protein levels. The child-specific physiological parameters, except liver volume, were computed with knowledge of age and body weight, whereas physicochemical parameters for toluene were kept age-invariant based on available data. The actual individual-specific liver volume (autopsy data) was also included in the model. The resulting model was used to simulate the blood concentration profiles in children exposed by inhalation, to 1 ppm toluene for 24 h. For this exposure scenario, the area under the venous blood concentration vs. time curve (AUC) ranged from 0.30 to 1.01 microg/ml x h in neonates with low CYP2E1 concentration (<3.69 pmol/mg protein). The simulations indicated that neonates with higher levels of CYP2E1 (4.33 to 55.93 pmol/mg protein) as well as older children would have lower AUC (0.16 to 0.43 microg/ml x h). The latter values were closer to those simulated for adults. Similar results were also obtained for 7 h exposure to 17 ppm toluene, a scenario previously evaluated in human volunteers. The interindividual variability factor for each subgroup of children and adults, calculated as the ratio of the 95th and 50th percentile values of AUC, was within a factor of 2. The 95th percentile value of the low metabolizing neonate group, however, was greater than the mean adult AUC by a factor of 3.9. This study demonstrates the feasibility of incorporating subject-specific data on hepatic CYP2E1 content and physiology within PBPK models for evaluating the age, interchild and population variability of internal dose for use in risk assessment of inhaled volatile organics.

Regulation of flavin-containing monooxygenase 1 expression by ying yang 1 and hepatic nuclear factors 1 and 4.

The flavin-containing monooxygenases (FMOs) are important for the oxidation of a variety of environmental toxicants, natural products, and therapeutics. Consisting of six family members (FMO1-5), these enzymes exhibit distinct but broad and overlapping substrate specificity and are expressed in a highly tissue- and species-selective manner. Corresponding to previously identified regulatory domains, a YY1 binding site was identified at the major rabbit FMO1 promoter, position -8 to -2, two overlapping HNF1alpha sites, position -132 to -105, and two HNF4alpha sites, position -467 to -454 and -195 to -182. Cotransfection studies with HNF1alpha and HNF4alpha expression vectors demonstrated a major role for each of these factors in enhancing FMO1 promoter activity. In contrast, YY1 was shown by site-directed mutagenesis to be dispensable for basal promoter activity but suppressed the ability of the upstream domains to enhance transcription. Finally, comparisons between rabbit and human FMO1 demonstrated conservation of each of these regulatory elements. With the exception of the most distal HNF4alpha site, each of the orthologous human sequences also was able to compete with rabbit FMO1 cis-elements for specific protein binding. These data are consistent with these same elements being important for regulating human FMO1 developmental- and tissue-specific expression.

Molecular regulation of genes encoding xenobiotic-metabolizing enzymes: mechanisms involving endogenous factors.

It is widely recognized that xenobiotic-metabolizing enzymes play a fundamental role in the basic processes of carcinogenesis and toxicity on one hand, and chemoprevention and drug efficacy on the other. Realization that different factors can profoundly affect the expression of these enzymes at the genome level has resulted in an enhanced appreciation of the importance these genes play in our modern industrialized age. There continues to be rapid proliferation of studies addressing the molecular regulation of these genes. The discovery of common signal transduction pathways and transcription factors that dictate tissue and developmental-specific expression, as well as variation in expression within a given tissue, suggest that there may be significant interaction among these various regulatory systems. This report is a summary of a symposium that was part of the Structure, Function and Regulation of Cytochromes P450 and Xenobiotic Metabolizing Enzymes satellite meeting of the 2000 joint meeting of the American Society for Biochemistry and Molecular Biology, the American Society for Pharmacology and Experimental Therapeutics, the French Pharmacological Society, and the Pharmacological Society of Canada held in Boston, Massachusetts. This symposium brought together several speakers who addressed specific receptor-mediated signal transduction pathways involved in the regulation of xenobiotic-metabolizing enzymes, as well as other molecular mechanisms whereby endogenous factors are involved in controlling tissue- and developmental-specific expression.

Induction of cytochrome P450 1A1 gene expression, oxidative stress, and genotoxicity by carbaryl and thiabendazole in transfected human HepG2 and lymphoblastoid cells.

Carbaryl and thiabendazole, two widely used pesticides, have been shown to induce cytochrome P450 1A1 (CYP1A1) expression, but neither compound is capable of displacing [3H] 2,3,7,8-tetrachlorodibenzo-P-dioxin from its aryl hydrocarbon receptor binding site. In the present study, we investigated the transcriptional regulation of CYP1A1 as well as other genes in various human hepatoma HepG2 cell lines stably transfected with the chloramphenicol acetyl transferase (CAT) reporter gene and cloned under the control of each of 14 promoters or response elements from relevant stress genes. Carbaryl and thiabendazole were found to activate CYP1A1 at the level of transcription, as demonstrated by the dose-dependent increase in reporter CAT and CYP1A1 mRNAs. Moreover, this effect appeared to be mediated via the xenobiotic responsive element (XRE), because both pesticides specifically activated various fusion constructs containing XRE sequences (CYP1A, glutathione S-transferase, and XRE). Carbaryl and to a lesser extent thiabendazole also activated other stress genes such as c-fos and NF-kappaBRE, HSP70 and GRP78, and GADD153 at a transcriptional level. These data suggest that these molecules induce early alert genes, including those known to be sensitive to oxidative stress. This led us to examine the genotoxic effect of carbaryl and thiabendazole by an in vitro DNA repair solid-phase assay. Both compounds provoked a strong DNA-damaging activity in the human lymphoblastoid cell line that constitutively expresses human CYP1A1 cDNA, but not in the parental line, indicating that CYP1A1 is chiefly implicated in carbaryl and thiabendazole genotoxicity. This effect was confirmed on HepG2 cells. These observations support the notion that intracellular signals leading to CYP1A1 induction, oxidative stress, and genotoxicity are intimately related.

Ethnic differences in human flavin-containing monooxygenase 2 (FMO2) polymorphisms: detection of expressed protein in African-Americans.

The flavin-containing monooxygenases (FMOs) are a family of xenobiotic-metabolizing enzymes that are expressed in a species- and tissue-specific manner. FMO2 expression has been observed in pulmonary tissue from several species, but not human. Two human FMO2 point mutations have been reported: a cytosine to thymidine transition at position 1414 resulting in a premature stop codon and a thymidine insertion at position 1589 resulting in a frameshift. To define the frequency of these sequence variations and explore their significance, unrelated African-American, Caucasian, and Korean individuals were genotyped. In the African-American population tested (n = 180), the 1414C allele occurred at a 13% frequency; however, all of the tested Caucasians (n = 52) and Koreans (n = 100) were homozygous for the 1414T allele. The T1589 allele occurred at frequencies of 6.9 and 13.0% in African-Americans (n = 175) and Caucasians (n = 23), respectively, and appears to segregate with the 1414T allele. Thus, it would have no further impact on FMO2 activity. Western blot analysis of pulmonary microsomes failed to detect immunoreactive protein in 1414T homozygotes. A heterozygotic individual did exhibit a single band of the expected size, but no detectable FMO activity in the corresponding lung microsomes. Sequence analysis, however, was consistent with the 1414C allele encoding an active FMO2 enzyme. FMO2 mRNA expression was observed in most individuals, but failed to correlate with genotype or protein expression. In summary, functional FMO2 is expressed in only a small percentage of the overall population. However, in certain ethnic groups, active pulmonary FMO2 enzyme will be present in a significant number of individuals.

Diflubenzuron, a benzoyl-urea insecticide, is a potent inhibitor of TCDD-induced CYP1A1 expression in HepG2 cells.

Diflubenzuron (DFB) belongs to a group of compounds called benzoyphenyl ureas acting as chitin synthesis inhibitors, which also inhibit growth of B16 murine melanomas. The present study was designed to investigate the effect of this insecticide, on CYP1A1 expression and induction in human hepatoma cells HepG2. Treatment of HepG2 cells over 72 h with noncytotoxic concentrations of DFB resulted in a strong dose-dependent decrease in constitutive ethoxyresorufin-O-deethylase activity. Moreover, DFB significantly decreased CYP1A1 induction by 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) after 24 h exposure, as demonstrated by ethoxyresorufin-O-deethylase (EROD) activity and Northern blot analysis. Additional studies were performed both on parental HepG2 cells and HepG2-241c.1, which were stably transfected with the chloramphenicol acetyltransferase (CAT) reporter gene, cloned under the control of the human CYP1A1 promoter (-1140 to +59). Ribonuclease protection assays (RPA) analysis clearly demonstrated an inhibition of CYP1A1 transcription in both cell lines. Surprisingly, in corresponding experiments using 3-methylcholanthrene (3-MC) as a CYP1A1 inducer, DFB was less effective. Finally, in competitive binding studies using a 9S-enriched fraction of HepG2 cytosol, DFB was capable of displacing [(3)H]-2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) from its Ah receptor binding site. Taken together, these results support the involvement of a transcriptional mechanism in the inhibition of CYP1A1 expression in HepG2 cells by DFB, possibly via an Ah receptor antagonism.

Developmental regulation of the L-type calcium channel alpha1C subunit expression in heart.

We used Northern analyses, RNase protection assays and immunoblot analyses to examine the relationship among developmental age of the heart, abundance of mRNA and L-type calcium channel alpha1C subunit protein, and to establish the size of the native protein in heart. Northern analysis, RNase protection assays, and immunoblots were used to study RNA and protein from rat heart of various ages. In fetal and adult ventricles there was a predominant 8.3-kb transcript for the alpha1C subunit with no change in transcript size during development. RNase protection assays demonstrated a 2-fold increase in abundance of the DHP receptor message during postnatal development. Immunoblots identified a 240 kD protein, corresponding to the predicted molecular mass of the full length alpha1C subunit. No change in size of protein for the alpha1C subunit was observed at any developmental stage and there was no evidence for a truncated isoform. There was an approximate 2-fold increase in alpha1C subunit protein in ventricular homogenates during postnatal development. Thus, in the developing rat heart, alterations in calcium channel properties during development appear to result neither from alternative splicing that produces a smaller transcript for the alpha1C subunit nor from expression of a truncated protein, but at least in part from transcriptionally-regulated expression of the 240 kDa polypepde.

Regulation of DHP receptor expression by elements in the 5'-flanking sequence.

The alpha(1)-subunit of the cardiac/vascular Ca(2+) channel, which is the dihydropyridine (DHP)-binding site (the DHP receptor), provides the pore structure for Ca(2+) entry. It contains the binding sites for multiple classes of drugs collectively known as Ca(2+) antagonists. As an initial step toward understanding the mechanisms controlling transcription of the rat cardiac alpha(1C)-subunit gene, we have cloned a 2.3-kb fragment containing the 5'-flanking sequences and identified the alpha(1C)-subunit gene transcription start site. The rat alpha(1C)-subunit gene promoter belongs to the TATA-less class of such basal elements. Using deletion analysis of alpha(1C)-subunit promoter-luciferase reporter gene constructs, we have characterized the transcriptional modulating activity of the 5'-flanking region and conducted transient transfections in cultured neonatal rat cardiac ventricular myocytes and vascular smooth muscle cells. Sequence scanning identified several potential regulatory elements, including five consensus sequences for the cardiac-specific transcription factor Nkx2.5, an AP-1 site, a cAMP response element, and a hormone response element. Transient transfection experiments with the promoter-luciferase reporter fusion gene demonstrate that the 2-kb 5'-flanking region confers tissue specificity and hormone responsiveness to expression of the Ca(2+) channel alpha(1C)-subunit gene. Electrophoretic mobility shift assays identified a region of the alpha(1C)-subunit gene promoter that can bind transcription factors and appears to be important for gene expression.

Flavin-containing monooxygenase isoform 2: developmental expression in fetal and neonatal rabbit lung.

Mammalian flavin-containing monooxygenase functions in the oxygenation of numerous xenobiotics containing a soft nucleophile, usually a nitrogen or sulfur. A total of five distinct flavin monooxygenase (FMO) isoforms are expressed in mammals. Individual isoforms are expressed in a sex-, age-, and tissue-specific fashion. In this study, we document the early developmental appearance of the major isoform in rabbit lung, FMO2. FMO2 catalytic activity as well as protein and mRNA are not only present in fetal and neonatal lung but, in some instances, approach levels found in the adult. The expression pattern of FMO2 is similar to that of the two major constitutive cytochromes P450 found in rabbit lung, 2B4 and 4B1. The early developmental appearance of these monooxygenases indicate an important role in the protection of the fetus and neonate against toxic insult from foreign chemicals.

A genetic polymorphism in the regulatory sequences of human CYP2E1: association with increased chlorzoxazone hydroxylation in the presence of obesity and ethanol intake.

To pursue the hypothesis that differences in the regulatory region of CYP2E1 are partially responsible for the intersubject variation in in vivo CYP2E1 activity, restriction enzyme digestion and Southern blotting of 36 human DNA samples were performed. The fractionated DNA was hybridized with a genomic probe to the upstream region of CYP2E1 from positions -2710 to -580. After digestion with the restriction enzyme XbaI, most subjects (69%) were homozygous for the expected band representing the XbaI fragment from positions -2270 to -408, whereas 31% had an unexpected, slightly larger band. Analyses of Southern blots of the same DNA samples cut by other restriction enzymes were consistent with the larger band containing an estimated 100-bp insertion and localized the mutation to a region from positions -2270 to -1672. To determine the functional significance of this mutation, in vivo CYP2E1 metabolic ability was determined in the same subjects using the 6-hydroxylation of chlorzoxazone as a probe. The presence of the insertion mutation was associated with greater CYP2E1 metabolic ability, but only among individuals who either were obese or had recently consumed ethanol (p < 0.01, both). These data are consistent with a DNA insertion that is associated with altered CYP2E1 induction. The incidence of the mutation was 31% among 65 African Americans and 6.9% among 58 Caucasians (p < 0.01). Thus, this CYP2E1 regulatory polymorphism not only enhances CYP2E1 metabolic ability, but is sufficiently common to impact susceptibility to CYP2E1-related diseases in at least two ethnic groups.

Functional characterization of the human CYP1A1 negative regulatory element: modulation of Ah receptor mediated transcriptional activity.

The mechanisms that underly the regulation of human CYP1A1 have merited considerable attention because of their association both with toxic outcomes and the etiology of several cancers. Previous work conducted in this laboratory has identified a negative regulatory element (NRE) in the 5' region of this gene that appeared to modulate CYP1A1 transcriptional activity. This NRE is present in two functional copies, a high affinity 21-bp palindrome centered at position -784, and an additional element found within a GC-rich region between position -728 and -558. In this report, the regulatory function of the NREs in the context of the CYP1A1 promoter was evaluated. This was accomplished by substituting mutated elements for the corresponding wild-type element in a vector that contained human CYP1A1 sequences positions -1140 to +59 directing the transcription of the chloramphenicol acetyltransferase reporter gene. Expression vectors containing specific mutations in each or both NREs were characterized. We show that eliminating the binding of the CYP1A1 repressor protein to one or both repressor motifs results in a significant 2- to 3-fold increase in the inducibility of CYP1A1 promoter activity. Although mutation of both sites appeared to result in an increase in inducibility over that observed with only one site mutated, the effect was not additive. Such aberrant transcriptional activity correlates with the highly inducible aryl hydrocarbon hydroxylase phenotype that is a reported marker for individuals predisposed to lung cancer. Mutation of the NRE, or more likely, the cognate repressor protein(s), may provide a genetic basis for this phenotype.

Further characterization of the major and minor rabbit FMO1 promoters and identification of both positive and negative distal regulatory elements.

Previously, two promoters were identified for the rabbit FMO1 gene: a major, upstream promoter (P0) that initiates transcription from exon 0 and a second, minor promoter (P1) located approximately 200 bp downstream and initiating transcription from exon 1. Transcription initiation from the P0 promoter results in elimination of the exon 1 leader sequence from the mature transcript. In this report, we further define the major promoter and identify several positive and negative upstream regulatory domains employing deletion analysis and transient expression in HepG2 cells. Of interest, P0 and P1 were equally active in these assays. A 49-bp fragment spanning position -41 to +8 was found essential for the activity of P0 and also capable of basal transcriptional activity. Interestingly, this same 49-bp region was found necessary for P1 activity. Upstream of P0, three positive regulatory regions (positions -348 to -176, -757 to -584, and -1196 to -829) and two negative regulatory regions (positions -2120 to -1724 and 829 to -757) were identified using deletion mutants. Both P0 and P1 share the most proximate, positive regulatory domain but were regulated differentially by more distal 5' sequences. In addition to the upstream regulatory sequences, a potent negatively acting element was observed within intron 1. Using DNA fragments representing the most potent positive (position -348 to -176) and negative (position -829 to -757) regulatory sequences as probes, we demonstrate the formation of multiple specific DNA/protein complexes with protein factor(s) present in HepG2 nuclear extract.

Variation in induction of human placental CYP2E1: possible role in susceptibility to fetal alcohol syndrome?

Fetal alcohol syndrome occurs in less than 10% of women who drink heavily during pregnancy. One potential mechanism for this intersubject variation is differences in placental alcohol metabolism. Alcohol dehydrogenase is present at low concentrations in the placenta and is not inducible. CYP2E1 has not been found in human placentas at early gestation time points or in random term placentas. Hepatic CYP2E1 is induced by alcohol and other environmental agents, but induction varies among heavy drinkers and may be genetically controlled. To test whether CYP2E1 is induced in placenta by heavy drinking during pregnancy, we performed a Western blot analysis on placental microsomes from women (n = 8) whose periconceptional average daily absolute alcohol intake was greater than 1 ounce. Using anti-human CYP2E1, bands consistent with CYP2E1 were identified in six samples, although considerable variation among individuals was observed. Among drinking mothers, offspring head size was smaller among those with placental CYP2E1 (p = 0.04). The association between the presence of the protein and smaller birth weight and birth length was equivocal (p = 0.09). Our data are consistent with placental CYP2E1 being inducible by drinking, but with induction being variable among heavy drinking women. We speculate intersubject variation in induction may have a genetic basis and may play a role in susceptibility to alcohol related defects.

Identification of multiple rabbit flavin-containing monooxygenase form 1 (FMO1) gene promoters and observation of tissue-specific DNase I hypersensitive sites.

We have cloned and partially characterized the rabbit FMO1 gene as a first approach to understanding mechanisms controlling its tissue-specific expression. The isolated clones contain 14 kb of 5' flanking information and approximately 30 kb of the structural gene, but do not include the 3'-end. Two upstream exons were defined, both encoding 5' leader information. The first exon, termed exon 0, contains information not previously reported. The second exon, termed exon 1, contains information previously reported for the rabbit FMO1 cDNA. Protein coding information begins seven nucleotides from the start of exon 2. A single transcription start site was localized in exon 1, while a cluster of sites were defined in exon 0, consistent with two alternative promoters. Transcripts initiating in exon 0 do not contain exon 1 information due to alternative processing and represent the major FMO1 mRNA. Neither promoter contains a TATA box or GC islands, although the exon 1 promoter does share some sequence identity with initiator-type elements. Homologous sequences to several known transcription factor binding sites were found in the upstream region of the FMO1 promoters. Both promoters were active in directing luciferase expression when transiently transfected into human HepG2 cells, although the data are consistent with both requiring upstream enhancer sequences. Consistent with this observation, DNase I hypersensitive sites were mapped to a 600-bp region immediately upstream of exon 0 using liver nuclei. No such sites were detected with nuclei from lung. Differential DNA methylation also was not observed between these two tissues.

Identification of tissue-specific DNase I hypersensitive sites in the rabbit flavin-containing monooxygenase form 2 gene.

Previous studies demonstrated that the flavin-containing monooxygenases (FMO) are expressed in a tissue-specific manner. To begin an elucidation of mechanisms regulating this expression pattern, genomic clones for rabbit FMO2 were isolated and characterized. Two clones were isolated from a lambda EMBL3 genomic library and shown to span approximately 30 kilobase pairs but to contain only about 800 base pairs of coding information. Primer extension analysis was used to map the transcription start site, extending the previously published cDNA sequence by 17 base pairs. The FMO2 promoter does not utilize a classical TATA box, nor are HTF islands present (DNA domains rich in cleavable sites for 5-methyl-cytidine/guanosine-sensitive restriction enzymes). Rather, homology with promoters controlled by initiator elements is observed. Previous studies demonstrated FMO2 expression in rabbit pulmonary Clara and type II cells [Overby, Nishio, Lawton, Plopper, and Philpot: Exp. Lung Res. 18, 131, (1992)]. In the present study, a highly enriched Clara/type II cell population was prepared, and the FMO2 gene was analyzed for DNase I-hypersensitive and methylated regions. These data were then contrasted with those obtained from a similar analysis of the nonexpressed, hepatocyte FMO2 gene. No difference in the methylation status was observed. However, Clara/type II cell-specific DNase I-hypersensitive sites are located within the promoter region of the FMO2 gene. Thus, tissue-specific transcription factors likely are more prominent than methylation in regulating FMO2 expression. Consistent with this observation, both polyomavirus enhancer activator 3 (E26 transformation specific) and thyroid transcription factor 1 consensus sequences are present within the tissue-specific DNase I-hypersensitive domain.

Differential expression and activity of flavin-containing monooxygenases in euryhaline and stenohaline flatfishes indicates potential osmoregulatory role.

N,N-Dimethylaniline (DMA) N-oxidase activity indicative of flavin-containing monooxygenase (FMO) was examined in four tissues (liver, gill, muscle, and kidney) of the flounder (Platichthys flesus). Gill microsomes had the highest levels of activity (456 +/- 343 pmol/min/mg), while kidney (121 +/- 109) and liver (67 +/- 26) had levels just above detection. A single faint band of a 56 kD protein was observed in liver and gill microsomes following Western blot analyses with polyclonal antibodies to FMO 1. DMA N-oxidase activity in gill and liver directly correlated with the expression of the 56 kD protein recognized by polyclonal antibodies against FMO form 1. Likewise a mRNA band of approximately 2.5 kilobases was higher in gill than a 3.0 kb band in liver following hydridization with an FMO 1 cDNA probe. Gill and liver microsomal DMA N-oxidase from the euryhaline P. flesus was compared with that of the stenohaline turbot (Scophthalmus maximus). DMA oxidase activity, FMO protein and mRNA were significantly greater in the gill of P. flesus, while S. maximus had higher levels of enzyme activity in the liver, but also significant levels in gill. Comparison of the enzymatic properties of the P. flesus gill and S. maximus liver enzymes indicated dramatic differences in Km between gill and liver, but were both inhibited by equimolar concentrations of trimethylamine (TMA). Gill microsomal activity in each species was unaffected by the mammalian FMO 2 substrate (competitive inhibitor) n-octylamine. Differential expression of FMO in tissues from stenohaline and euryhaline fish suggests a functional relationship between FMO and osmoregulation.