Flavin-Containing Monooxygenase-3 and 5: Tissue Distribution, Age-Related Expression and Regulation by Endoplasmic Reticulum Stress

Background Flavin-containing monooxygenases (FMO) constitute a class of oxidative enzymes. FMO3 and FMO5 are two major FMOs that are highly expressed in the liver. The study was performed to determine regulated expression of both enzymes by age, fructose, therapeutic agents and endoplasmic reticulum (ER) stressors. The expression of both enzymes in 33 extrahepatic tissues was determined as well. Methods The tissue distribution of FMO3 and FMO5 was determined by immunohistochemistry. For age-related expression, livers were collected and divided into 5 age groups: I (< 31 days), II (35-70 days), III (89-119 days), IV (123-198 days), and V (>18 years of age). These samples were analyzed for the expression of FMO3 and FMO5 by RT-qPCR and Western blotting. For the regulated expression by other factors, human primary hepatocytes were treated with a chemical and the expression was determined. The reporter activity was determined in response to these chemicals. Results FMO3 and FMO5 were strongly stained in the liver but exhibited overlapping and distinct expression patterns in extrahepatic tissues. Both enzymes showed a neonatal surge in mRNA expression and were correlated well with age during the first 7 months after birth. The mRNA expression of both enzymes was suppressed by ER stressors and induced by the steatotic agent valporic acid. Similar changes of the reporter activities were observed with an exception of fructose, a sugar associated with metabolic diseases. This carbohydrate induced FMO5 mRNA but did not activate the

and FMO5 by RT-qPCR and Western blotting. For the regulated expression by other factors, human primary hepatocytes were treated with a chemical and the expression was determined. The reporter activity was determined in response to these chemicals.
Results: FMO3 and FMO5 were strongly stained in the liver but exhibited overlapping and distinct expression patterns in extrahepatic tissues. Both enzymes showed a neonatal surge in mRNA expression and were correlated well with age during the first 7 months after birth. The mRNA expression of both enzymes was suppressed by ER stressors and induced by the steatotic agent valporic acid. Similar changes of the FMO5 reporter.
Conclusions: The expression of FMO3 and FMO5 varies depending on a tissue, age, ER stress and nutritional status. Of significance are the abundant presence of FMO3 and FMO5 in some endocrine cells and regulated expression by ER-stressors and fructose. These observations conclude that FMO3 and FMO5, in addition to xenobiotic metabolism, are involved in pathogenesis, particularly related to metabolic diseases.

INTRODUCTION
Flavin-containing monooxygenases (FMOs, EC 1.14.13.8) constitute a class of major oxidative enzymes (Petriello et al., 2017;Phillips and Shephard, 2017;Rossner et al., 2017). Traditionally, these enzymes are established to play important roles in the metabolism and detoxification of drugs, pesticides and dietary compounds . Emerging evidence, nonetheless, has linked their functionality to many pathophysiologic processes including metabolic diseases and aging (Petriello et al., 2017;Rossner et al., 2017). Without exceptions, all mammalian species examined express multiple forms Phillips et al., 1995;Hernandez et al., 2004;Lattard et al., 2004;Hao et al., 2009). The human genome encodes a total of 11 FMO genes but only 5 of them produce catalytic proteins, which are commonly referred to as FMO1, 2, 3, 4 and 5, respectively . The first 4 genes are closely clustered in the chromosome 1 Hao et al., 2009). While the FMO5 gene is also located in chromosome 1, but it is 20 Mb away from others. Interestingly, other species have similar genomic arrangement among these FMO genes. For example, the mouse genome has 4 Fmo genes (1 to 4) closely clustered in chromosome 1, whereas the Fmo5 gene is located in chromosome 3 . In addition, many FMO genes produce alternative splicing transcripts, further diversifying the expression species (Lattard et al., 2004;Rossner et al., 2017).
FMO enzymes, like many other drug-metabolizing enzymes, have a broad tissue distribution. Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) transcripts for all 5 FMOs among 23 different tissues with few exceptions (Nishimura and Naito, 2006). No FMO1 transcript is present in leukocytes, nor is FMO2 mRNA in leukocytes and placenta. Among tissues that expression FMO mRNA, the relative abundance on FMO1 mRNA varies by as much as 6000-fold between the liver and bone marrow (Nishimura and Naito, 2006 (Novick et al., 2009). The cellular localization shows species-differences as well. For example, rat FMO3 is predominately expressed in the perivenous region whereas mouse Fmo3 in the periportal region Novick et al., 2009). The cellular localization of human FMOs remains largely unknown.
The expression of FMOs is regulated by xenobiotics and inflammatory stimuli (Chung et al., 1997;Zhang et al., 2009;Celius et al., 2010;Rudraiah et al., 2014). Rifampicin, a prototypical activator of the pregnane X receptor in human, has been shown to induce FMO4 and FMO5 (Rae et al., 2001). Extracts from phellinus baumii, a widely used medicinal plant in Asia, have been shown to induce mouse Fmo2, Fmo3 and Fmo4 but not Fmo1 and Fmo5 (Sainkhuu et al., 2016). Interestingly, the induction of FMO mRNA may not lead to significant induction of FMO protein (Celius et al., 2010;Rudraiah et al., 2014). The environmental chemical 3-methylcholanthrene, for example, induces Fmo3 mRNA by as much as 30-fold but only modest increase occurs at Fmo3 protein (Celius et al., 2010). Furthermore, inflammatory stimuli regulate the expression of FMOs but the regulatory effect varies depending on an inflammatory model and an FMO gene as well as a strain (Zhang et al., 2009). For example, the immunostimulant lipopolysaccharide downregulates fmo1 but not fmo4.
The downregulation in HeoU mice is greater than that in HeJ mice. Consistent with an established role of FMOs in detoxification, overexpression of FMOs protects against endoplasmic reticulum (ER) stress, a common initiator for many pathological processes (Liao et al., 2016).
In addition to xenobiotics and inflammatory stimuli, many other factors have been shown to regulate the expression of FMOs including age, sex and hormones (Dolphin et al., 1996;Cherrington et al., 1998;Ripp et al., 1998;Koukouritaki et al., 2002;Larsen-Su et al., 2002;Zhang and Cashman, 2004;Hines, 2006;). In humans, FMO1 is considered to be a fetal liver enzyme and the expression of FMO1 mRNA and protein decreases with gestational days (Dolphin et al., 1996;Koukouritaki et al., 2002). In contrast, FMO3 is expressed higher in adult livers at both mRNA and protein levels (Dolphin et al., 1996;Koukouritaki et al., 2002). In mice, high levels of FMO1 and FMO5 are present in postnatal and adult animals . In contrast, FMO2, 3 and 4 exhibit age-dependent increases with FMO3 being expressed the highest. Several FMOs exhibit sex-preferable expression . For example, female mice express higher liver Fmo1 than male, but the opposite is true in the kidney. Consistent with sex-preferable expression, sex hormones have been shown to regulate the expression of FMOs. The female sex hormone 17β-estradiol, for example, decreases FMO activity by 56% in cultured rat hepatocytes (Coecke et al., 1998). In male mice, castration dramatically increases FMO3 expression, and testosterone replacement to castrated mice results in ablation of FMO3 expression (Falls et al., 1997).
The opposing effects on FMO expression by male and female hormones are also observed with other pairs of counterbalancing hormones such as glucagon and insulin.
In mice, the glucose-elevating hormone glucagon induces all Fmos with Fmo3 being induced by as many as 14-fold (Miao et al., 2015). In contrast, insulin suppresses the expression of these enzymes by 30-69%. In streptozotocin-induced diabetic rats, insulin treatment reverses diabetic-induced elevation of FMO1 activity (Borbás et al., 2006). An early microarray study reports a 59% decrease of FMO5 mRNA in Type 2 diabetic patients (Takamura et al., 2004). In support of critical roles of Fmos in metabolic homeostasis, mice deficient in one of Fmos through knockout or knockdown exhibit a lean phenotype with decreased body weight and/or reduced adiposity (Shephard and Phillips, 2010;Veeravalli et al., 2014;Schugar et al., 2017; metabolic profiles including lower levels of plasma glucose, cholesterol and triglycerides.
In this study, we used a comprehensive approach and investigated human FMO3 and FMO5 for their subcellular localization, age-related expression and regulated expression by ER stressors, fructose and xenobiotics. Both FMO3 and FMO5 were strongly present in the liver but exhibited overlapping and distinct expression patterns in extrahepatic tissues. Both enzymes showed a neonatal surge in mRNA expression and were correlated well with age during the first 7 months after birth. Both enzymes were suppressed by ER stressors and induced by the steatotic agent valporic acid.
These observations conclude that FMO3 and FMO5, in addition to xenobiotic metabolism, are involved in pathogenesis, particularly related to metabolic diseases.

Chemicals and reagents: Brefeldin A (BFA) and Hanks balanced salt solution were from Sigma (St. Louis, MO). Dulbecco's modified eagle medium (DMEM) and
Taqman probes were from Life Technology (Carlsbad, CA). The goat anti-rabbit IgG conjugated with horseradish peroxidase was from Pierce (Rockford, IL). Thapsigargin (THAP) was from R&D Systems (Minneapolis, MN). Nitrocellulose membranes were from Bio-Rad (Hercules, CA). Expression constructs and the FMO5 reporter were from OriGene Technologies Inc (Rockville, MD). Unless otherwise specified, all other reagents were purchased from ThermoFisher Scientific (Fair Lawn, NJ).
Immunohistochemistry: Formalin-fixed paraffin-embedded (FFPE) array slides with a total of 33 tissues were purchased from Pantomics Inc (Richmond, CA). The slides were warmed up in 60°C oven for 60 min followed by deparaffinization and rehydration, essentially as described previously (Yan et al., 1995). The slides were then incubated in Dako Target Retrieval Solution (Agilent, Santa Clara, CA) in pressure cooker in microwave for 15 min and then cool for 30 min. The slides were washed 3 times (3 min/each) in distilled water and then incubated in Dako quenching reagent for 10 min at room temperature. The quenched slides were washed 3 times again and incubated at 4C with primary antibody (FMO3 or FMO5) or pre-immunized rabbit serum at a dilution of 1:1000 overnight. After 3 washings, the slides were incubated with Dako EnVision + Dual Link System-HRP secondary antibodies (antirabbit) followed by 3 washings. The washed slides were incubated with substratechromogen for 10 min. The reactions were stopped in distilled water and the slides were once again washed 3 times. Thereafter, the slides were dehydrated and mounted in Cytosealtm 60 media.
Liver RNA and S9 fractions: A total of 59 tissue samples were used in this study as total RNA from the liver tissues was described previously (Shi et al., 2011), and the quality was determined by electrophoresis. S9 fractions were prepared by differential centrifugation as described previously (Shi et al., 2011). The use of the human samples was approved by the Institutional Review Board.
Reverse transcription-quantitative polymerase chain reaction (RT-qPCR) and Western

Tissue distribution and cellular localization The focus of this study is on FMO3
and FMO5, because they are major FMO enzymes in human liver ). An early study detected the presence of FMO3 and FMO5 transcripts in a wide range of tissues (Nishimura and Naito, 2006). In this study, we initially tested whether FMO3 and FMO5 proteins have a broad tissue distribution and whether they are localized in certain cell populations in a tissue.
Immunohistochemical staining was performed with tissue arrays (33 tissues). As shown Fig. 1, FOM3 was strongly stained in the liver and adrenal; moderately in the lung, brain cortex, spinal cord, thyroid, urinary bladder and testis; and slightly in the small intestine, pituitary, kidney, and ureter. In the liver, FMO3 was stained the most between central vein and portal triad. All staining was detected in the cytoplasm of hepatocytes. Hepatocytes surrounding the central vein was less stained. No staining was detected in the periportal area including bile ducts. In the adrenal gland, strong staining was restricted to the medullar with no staining of the cortex. In addition, moderate staining was detected in some of the pneumocytes of the lung, the colloid and follicular cells of the thyroid and spermatocytes of the testis. Slight staining was detected in some of the proximal tubes of the kidney. Interestingly, no staining was detected in the duodenum, colon and parathyroid gland. Table I summarizes the immunohistochemical staining of FMO3.
FMO5, like FMO3, was stained strongly in the liver. Strong FMO5 staining was also detected in the parathyroid gland. In addition, FMO5 was stained moderately in the stomach and colon; and slightly in the small intestine, lung, pituitary, brain cortex, kidney, bladder, heart, testis, skeletal muscle and ureter. In contrast to FMO3 in the liver, FMO5 was stained across the entire hepatic acinus. In the parathyroid gland, strong staining was detected in oxyphil cells and moderate staining in chief cells. In the duodenum and colon, slight or moderate staining of FMO5 was detected in the epithelium, contrasting strikingly to no FMO3 staining in the digestive tract (Fig. 1, Table I). In the adrenal, FMO5 was stained in the cortex but not medullar as seen with FMO3. In the kidney, FMO5 was stained in both proximal and distal tubules. Once again, Table I shows the relative staining intensity and cellular localization of immunohistochemistry.
A surge in FMO3 and FMO5 mRNA expression during the neonatal period It has been reported that the level of FMO3 mRNA is relatively low in pediatric livers compared with adult counterparts and the level of FMO3 protein is low in the liver during the first month of life and increases sharply thereafter (Koukouritaki et al., 2002). The level of FMO5 mRNA, on the other hand, is slightly higher in pediatric liver, but the ontogenic expression of FMO5 protein remains to be determined. We previously reported that the expression of CES1 exhibited a neonatal surge in mRNA and protein expression between the first and second month after birth (Shi et al., 2011 . 2B), likewise, showed a surge between group 1 and 2. Group 2, 3 and 4 expressed comparable levels of mRNA, and the adult group expressed slightly lower levels of FMO5 mRNA. As seen with FMO3, FMO5 mRNA was correlated significantly with age. In addition, FMO5 protein expression was consistent with the corresponding mRNA level with an exception of group 1. Once again, the very two samples in this group expressed a much higher FMO5 protein level than other samples in this group.

Regulated expression of FMO3 and FMO5 by ER stressors Both FMO3 and
FMO5 have been implicated in metabolic homeostasis (Gonzalez Malagon et al., 2015;Miao et al., 2015). ER stress is a common pathogenic theme for many diseases including metabolic disorders (Hetz and Saxena, 2017;Sozen and Ozer, 2017;Urra et al., 2017). On the other hand, it was reported that overexpression of FMO3 protected against ER-stress (Liao et al., 2016). We next tested whether ER-stress alters the expression of FMO3 and FMO5. To have a broader implication, valproic acid and fructose were included. Valproic acid is a widely used antiepileptic agents with strong steatotic activity (Bai et al., 2017), whereas fructose consumption is closely associated with the development of metabolic disorders (Bidwell, 2017). Human primary hepatocytes were treated with these chemicals at clinically relevant concentrations and the expression of FMO3 and FMO5 was determined. As shown in Fig. 3A whereas induced FMO5 mRNA expression by 127% (Fig. 3A).
The altered mRNA expression of FMO3 and FMO5 pointed to two levels of regulation: mRNA stability, transcription or both. We therefore tested whether transcriptional regulation is involved. Promoter reporters were tested for their response to these chemicals. As shown in Fig. 3B The cellular localization of FMO3 and FMO5 in the liver may have special implications in pharmacological interactions and zone-specific tissue toxicity. In this study, we have shown that FMO3 was stained the most between central vein and portal triad. Hepatocytes surrounding the central vein was less stained (Fig. 1, Table   I). No staining was detected in the periportal area including bile ducts. This localization differs from its rodent counterparts. Rat FMO3 is present around the perivenous region (centrilobular) gradually decreasing toward the periportal area, but the opposite is true with mouse Fmo3 Novick et al., 2009).
FMO5, on the other hand, was stained across the entire hepatic acinus. Cellular localization of FMOs may alter therapeutic efficacy. For example, tamoxifen, a widely used anticancer agent, undergoes metabolism by FMO3 and CYP enzymes such as CYP3A4 (Gjerde et al., 2010). Importantly, metabolites produced by CYP3A4 but not FMO3 are therapeutically active. Based on the cellular localization, FMO3 is exposed to higher concentrations of tamoxifen upon oral administration than CYP3A4. It is conceivable that such competition leads to reduced efficacy, although the relative affinity toward tamoxifen is a critical factor.
While FMOs are generally involved in xenobiotic metabolism, the cellular localization of FMO3 and FMO5 also points to physiological roles, particularly related to endocrinology. In this study, we have shown that FMO3 and FMO5 were strongly or moderately stained in the adrenal gland, parathyroid gland and/or thyroid gland (Fig.   1). In the adrenal gland, FMO3 was strongly stained in the medullar, the main region that produces the catecholamines adrenaline and noradrenaline (Ehrhart-Bornstein and Bornstein, 2008). It remains to be determined whether FMO3 is involved in the biosynthesis and secretion of these hormones. FMO3 is nevertheless established to metabolize the catecholamine releasing agent tyramine (Lin and Cashman, 1997). The substantial staining of FMO3 in the follicular cells of the thyroid gland linked the functionality of this enzyme to the synthesis and secretion of thyroid hormones.
Interestingly, FMO5 was strongly stained in chief and oxyphil cells, two unique cell populations in the parathyroid gland. Chief cells are the source for parathyroid hormone, whereas the functionality of oxyphil cells remain to be elucidated (Howson et al., 2015).
The precise roles of the abundant presence of FMO3 and/or FMO5 in the endocrine cells remain to be determined. One of the functional features shared by these cells is their high secretory capacity. FMO enzymes are abundant in the ER and present in the ER/Golgi transport vesicles (Hay et al., 1997). It is conceivable that FMO enzymes function as trafficking proteins for facilitated secretion of hormones (e.g., parathyroid hormone). These hormones, potentially regulated by FMOs for secretion, are all established to maintain energy balance and metabolic homeostasis (Hay et al., 1997;Lenders and Eisenhofer, 2014;Hannoush and Weiss, 2017). Interestingly, it has been increasingly recognized that FMO enzymes have broad impact on glucose, lipid metabolism and atherosclerosis. Indeed, FMO5 knockout mice show a lean phenotype with less weight gain and lower plasma glucose and cholesterol concentrations (Shephard and Phillips, 2010;Veeravalli et al., 2014;Schugar et al., 2017;Scott et al., 2017). The metabolic effect of FMO5 is more evident in aged animals (Gonzalez Malagon et al., 2015). Likewise, knockdown of Fmo3 decreases plasma lipids, glucose and insulin (Miao et al., 2015;Shih et al., 2015). Conversely, overexpression of Fmo3 delivers the opposite effect. It has been shown that the overexpression of this enzyme increases the conversion of trimethylamine to trimethylamine-N-oxide, a metabolite linked to metabolic diseases (Oellgaard et al., 2017;Subramaniam and Fletcher, 2017).
The involvement of FMOs in metabolic homeostasis is confounded by metabolic conditions that regulate the expression of these enzymes. It has been reported that diabetic condition induced FMOs and insulin suppressed them (Takamura et al., 2004;Borbás et al., 2006;Miao et al., 2015). It remains to be determined how elevation of blood glucose (i.e., diabetic condition) interplays with insulin resistance (elevated insulin) in terms of regulated FMO expression. In this study, we have shown that valproic acid induced both FMO3 and FMO5 and activated their promoters (Fig.   3). Interestingly, this antiepileptic is known to induce liver steatosis (Bai et al., 2017) and surprisingly reduce insulin resistance in rats (Kan et al., 2016). Mice deficient in Fmo3 or Fmo5 showed lean phenotypes with favorable lipid and sugar profiles, and transgenic expression of Fmo3 had the opposite effect (Shephard and Phillips, 2010;Veeravalli et al., 2014;Shih et al., 2015;Schugar et al., 2017;Scott et al., 2017). It appears that induction of FMO3 and FMO5 by valproic acid is a likely contributing to hepatic steatosis. Interestingly, fructose induced FMO5 mRNA but did not activate the FMO5 reporter (Fig. 3), suggesting that increased stability of FMO5 mRNA is involved in the mRNA induction. Given the fact that valproic acid induced FMO5 mRNA and activated the FMO5 reporter, induced expression of FMO5 by fructose and valproic acid was achieved through distinct mechanisms.
It has been reported that the expression of FMOs is ontogenically regulated and our age-related expression study provides additional important findings regarding developmental regulation (Dolphin et al., 1996;Janmohamed et al., 2002;Koukouritaki et al., 2002). The mRNA levels of both FMO3 an FMO5 showed a postnatal surge (Fig. 2), as seen in many other drug-metabolizing enzymes. The adult group, compared with the pediatric groups (groups 2 to 4), expressed higher levels of   and re-probed with an antibody to CES1. The signal was captured by MyCEL Imager.
The membrane was stained with 0.1% Ponceau S to verify the equal loading. (B) Expression of FMO5 Once again, the expression of FMO5 was determined by RT-qPCR and Western blotting. The data are presented as mean SD, and the levels of FMO5 mRNA were correlated with age (days). An asterisk symbol indicates Statistical significance for the correlation (p < 0.05).