Scaling in vitro activity of CYP3A7 suggests human fetal livers do not clear retinoic acid entering from maternal circulation

All-trans-retinoic acid (atRA), the active metabolite of vitamin A, is a critical signaling molecule during embryonic and fetal development and is necessary for maternal health. Fetal exposure to endogenous atRA is tightly regulated during gestation in a tissue specific manner and maternal exposure to exogenous retinoids during pregnancy is teratogenic. The clearance of atRA is primarily mediated by the cytochrome P450 (CYP) 26 enzymes, which play an essential role in controlling retinoid gradients during organogenesis. We hypothesized that CYP26 enzymes in the human fetal liver also function as a protective barrier to prevent maternal atRA reaching fetal circulation. Using human fetal liver tissue, we found that the mRNA of CYP26A1 and CYP26B1 enzymes is expressed in the human fetal liver. However, based on inhibition studies, metabolite profiles and correlation of atRA metabolism with testosterone hydroxylation, clearance of atRA in the fetal livers was mediated by CYP3A7. Based on in vitro-to-in vivo scaling, atRA clearance in the fetal liver was quantitatively minimal, thus providing an insufficient maternal-fetal barrier for atRA exposure.

human fetal organs is limited, and it is unknown whether the same CYP26 enzymes that appear predominant in metabolizing atRA in adult human liver also regulate atRA clearance in the fetal liver. In fact, clear dichotomy of expression of CYP enzymes between fetal and adult livers has been shown for the CYP3A family of enzymes [20][21][22] . While CYP3A4 and CYP3A5 are the main CYP3A enzymes in adult human liver, they are not well expressed in the fetal liver. Instead, CYP3A7 is the main human fetal liver CYP3A isoform [20][21][22] . CYP3A7 also metabolizes atRA 13,23 and before the identification of CYP26 enzymes, CYP3A7 was suggested as the main fetal liver atRA hydroxylase limiting fetal exposure to atRA 23 . Although CYP26 mRNA has been previously detected in human fetal liver tissues [24][25][26] , the quantitative importance of the CYP26 enzymes and CYP3A7 in modulating maternal-fetal transfer of atRA and fetal atRA clearance is not known. Based on the existing data that show the importance of CYP26 enzymes in atRA metabolism in adult liver and the importance of tight regulation of retinoid gradients in the developing fetus, we hypothesized that CYP26A1 plays an important role in fetal atRA clearance, and that the fetal liver limits maternal-fetal transfer of atRA. The aims of this study were to determine whether CYP26 enzymes are important in atRA clearance in human fetal liver, whether CYP3A7 contributes to atRA clearance in human fetal liver and to determine the efficiency of the fetal liver in eliminating atRA that passes to the fetus from maternal circulation.

Results
Detection of CYP26 mRNA in human fetal livers. The mRNA expression of CYP26A1, CYP26B1 and CYP26C1 together with CYP3A7 was measured in fetal livers from 18 individual donors (Fig. 1) and in five control adult human livers (data not shown). CYP26A1 and CYP26B1 mRNAs were detected and quantified in 14 and 11 of the 18 fetal livers, respectively, while CYP26C1 mRNA was not detected in any of the 18 fetal livers. Considerable inter-individual variability, up to 125-fold, was observed in the expression of CYP26A1 in the fetal livers while the expression of CYP26B1 was less variable (Fig. 1). Two of the fetal livers had no detectable expression of any CYP26 mRNAs and in both of these livers robust CYP3A7 mRNA expression was detected. In fact, CYP3A7 expression was relatively high (C t values [26][27][28][29][30][31][32][33][34] in 17 of the 18 fetal livers. One fetal liver had very low CYP3A7 expression (C t value 38) and this liver showed the highest CYP26A1 expression among all 18 fetal livers. The overall C t values of CYP26A1 were higher in the fetal livers (one with C t value 32 and others [37][38] than in the adult livers analyzed (C t [27][28][29][30][31][32][33][34][35][36] while CYP26B1 C t values were generally lower in the fetal livers (C t [36][37][38][39] than adult livers (CYP26B1 was only detectable in one adult liver). Due to differences in housekeeping gene expression between adult and fetal livers, no quantitative comparisons were made between fetal and adult liver mRNA expression. CYP26C1 mRNA was not detected in any of the adult human livers while CYP3A7 mRNA was detected in 3 of the 5 adult livers (C t values 30-37).
atRA metabolism in human fetal livers. To explore whether the metabolites formed from atRA in fetal liver resembled either the metabolite profile observed from atRA with recombinant CYP3A7 or CYP26 enzymes, the metabolites formed from atRA by CYP3A7, CYP26A1 and CYP26B1 and by fetal liver S9 fractions were characterized (Fig. 2). Consistent with previous data, recombinant CYP26A1 and CYP26B1 hydroxylated atRA at multiple sites, while CYP3A7 only formed 4-OH-atRA and 4-oxo-atRA. The metabolite profile in human fetal livers was similar to that observed with CYP3A7. Of the atRA metabolites, only the formation of 4-OH-RA and its metabolite 4-oxo-RA was observed (Fig. 2). No formation of the CYP26 specific metabolite 16-OH-RA was observed in fetal livers. The fetal liver metabolite profile corresponded to that observed with CYP3A7 and suggests lack of significant CYP26 contribution to atRA metabolism in fetal liver.

Identification of CYPs that metabolize atRA in human fetal livers.
To quantify which CYP enzymes contribute to atRA clearance in fetal liver, selective inhibitors of CYP3A7 and CYP26 were used. Fluconazole has been previously shown not to inhibit CYP26 16 . The inhibition of CYP3A7 mediated 4-OH-atRA formation by fluconazole (300 μM) was confirmed using recombinant CYP3A7. Inhibition of CYP26A1 by talarozole (200 nM) was reproduced with recombinant CYP26A1 (Fig. 4). The specificity of talarozole (200 nM) towards CYP26 was confirmed using recombinant CYP3A7. Talarozole did not significantly inhibit CYP3A7 mediated 4-OH-RA formation (Fig. 4). Together these data show that fluconazole and talarozole can be used as specific inhibitors of CYP3A7 and CYP26 respectively. In human fetal liver S9 fractions, atRA metabolite formation was inhibited 30-60% by fluconazole and 30-40% by talarazole (Fig. 4), suggesting that both CYP26 enzymes and CYP3A7 contribute to atRA clearance in human fetal liver. Ketoconazole inhibits both CYP26 enzymes and CYP3A7 16 . Consistent with this inhibition profile, ketoconazole caused a >95% decrease in atRA metabolite formation. However, these inhibition experiments do not unequivocally define which CYP enzyme is predominant in atRA clearance in human fetal liver. Therefore, to further define the importance of CYP3A7 in atRA clearance in fetal liver, correlation analysis between atRA metabolite formation and the formation of the CYP3A7 specific testosterone metabolite 6βOH-testosterone was conducted (Fig. 3). atRA oxidation (the sum of 4-OH-RA and 4-oxoRA formation) and 6βOH-testosterone formation from testosterone correlated significantly in the fetal livers tested (r 2 = 0.58, p < 0.05) suggesting that CYP3A7 plays a major role in atRA metabolism in human fetal liver.
www.nature.com/scientificreports www.nature.com/scientificreports/ prediction of atRA clearance in the human fetal liver. To define the quantitative role of the fetal liver in atRA clearance and in potentially serving as a barrier to maternal-fetal retinoid transfer, the overall organ clearance and extraction ratio of atRA metabolism was calculated for the fetal liver for the gestational ages studied. The overall intrinsic clearance of atRA by the whole fetal liver (Cl intFL ) increased with gestational age due to the growth of the fetal liver. The intrinsic clearance per mg S9 protein was independent of gestational age (Fig. 3). The predicted fraction of atRA removed from fetal circulation by the fetal liver (extraction ratio) was very low, ranging from 0.01 to 0.05 (Fig. 3F). This suggests minimal extraction of maternal atRA by the fetal liver.

Discussion
atRA is a key developmental morphogen, and distinct concentration gradients of atRA within the developing embryo and fetus are crucial for regulation of cellular differentiation 1,2,7 . Embryonic development requires gestational age and tissue-type specific regulation of atRA concentrations 2,3,9 . Based on this, we hypothesized that a barrier, such as the fetal liver or the placenta, exists between the mother and the fetus to prevent maternal endogenous atRA passing to the fetus and enabling autonomous regulation of fetal atRA concentrations. Previous studies have shown CYP26 mRNA expression in fetal liver [24][25][26] . Hence, we hypothesized that the CYP26 enzymes, which are generally believed to be the main human retinoic acid hydroxylases 14,27 , would constitute a maternal-fetal barrier for maternal atRA. Based on the mRNA analysis of individual fetal livers, however, CYP26 enzymes appeared relatively insignificant in the human fetal liver both in terms of observed mRNA expression and apparent activity. At the same time, CYP3A7 mRNA was abundant in the fetal livers, in agreement with past studies 20,21 . The finding of low but detectable expression of CYP26A1 mRNA in the human fetal livers is similar to prior findings. One study showed low to undetectable CYP26A1 mRNA in the human fetal liver and relatively high CYP26A1 mRNA in human fetal cephalic tissue 25 . A second study in a single donor showed weak detection in a single donor 26 . The detection of CYP26B1 mRNA in a subset of the fetal livers is consistent with the prior detection of CYP26B1 mRNA in a single donor of human fetal liver 26 . The mRNA expression of CYP26A1 and CYP26B1 observed in the adult liver in this study agrees with previous reports showing that CYP26A1 is the predominant CYP26 enzyme in adult liver and CYP26B1 is either undetectable or has very low expression 13,15,26,28 . In contrast to the previous single donor analysis however 26 , CYP26C1 mRNA was not detected in adult or fetal livers. The detection of high CYP26A1 mRNA in one fetal liver that had very low CYP3A7 expression is of particular interest. If CYP3A7 is predominantly responsible for atRA clearance in the fetal liver, low expression or lack of this CYP would result in increased atRA concentrations in fetal liver. These increased atRA concentrations should in turn induce the expression of CYP26A1 in the fetal liver leading to the observed mRNA expression pattern.  Table 1 www.nature.com/scientificreports www.nature.com/scientificreports/ Although this observation is from a single donor, it is consistent with the prevailing notion that CYP26A1 expression is responsive to atRA concentrations and its expression is induced by increased atRA concentrations 17,28 .
Despite the detection of CYP26 isoform mRNA in human fetal livers, all the data presented here (the metabolite profiles, CYP inhibition data and correlation of 4-OH-RA formation with 6βOH-TST formation) support the role of CYP3A7 as the main human fetal liver atRA hydroxylase with only a minor contribution of CYP26 enzymes. This was surprising, as in adult human liver CYP26A1 was previously found to be the main atRA hydroxylase despite the activity of CYP3A4 and CYP3A5 as atRA hydroxylases 13 . Yet, even in the adult human liver, CYP3A4 and CYP3A5 expression did significantly correlate with atRA hydroxylation activity. In particular in livers with low CYP26A1 expression, CYP3A enzymes were likely to be significant atRA hydroxylases 13 . Although CYP3A7 constitutes 30-85% of the fetal liver CYPs 29 we expected that the >1,000 fold higher intrinsic clearance of atRA by the CYP26 enzymes in comparison to CYP3A7 13,15,30 would translate to an important contribution of CYP26s to fetal liver atRA clearance, even if their expression was much lower than CYP3A7. However, the data collected does not support this hypothesis and we conclude that CYP3A7 is the main human fetal liver atRA hydroxylase. This finding is likely to translate to other RA isomers, 13-cisRA and 9-cisRA as well, as they have similar clearance profiles by CYP26s as atRA 30 and are metabolized by CYP3A7 23 . The lack of change in atRA and testosterone oxidation activity with gestational age observed in this study is consistent with previous reports that have shown relatively stable CYP3A7 expression across gestational ages 20,21 . The data is, however, discrepant with the study that showed higher atRA hydroxylation rates in three fetal livers from gestational days 96-109 when compared to livers from days 54-89 of gestation 23 . The previously observed change with gestational age is likely due to the small sample size that did not completely capture inter-individual variability in atRA clearance.
This is the first study to scale the observed microsomal or S9 protein activity to the entire fetal liver. This scaling shows that the metabolic capacity of the fetal liver towards atRA is unlikely to contribute significantly to fetal atRA clearance, and thus this metabolism is not quantitatively sufficient to protect the developing fetus. While the intrinsic clearance of atRA metabolism per mg of fetal liver protein did not change with gestational age, our analysis shows that when the growth of the liver and increase in umbilical blood flow is accounted for, the overall metabolic clearance of atRA increases significantly with gestational age. However, the extraction ratio is unchanged with gestational age, and the predicted extraction ratio (0.01-0.05) suggested that maximum 5% of maternal atRA could be extracted by fetal liver at any gestational age. The scaling methods used here, the analysis www.nature.com/scientificreports www.nature.com/scientificreports/ of the extraction ratio and the prediction of the overall fetal liver metabolic clearance are likely useful for future estimations of the role of the fetal liver in clearance of therapeutic drugs and toxins which are metabolized by fetal liver CYPs or glucuronidation enzymes.
Collectively, the data presented here are in agreement with the current consensus 2,3,9 that fetal tissue exposure to atRA and tissue atRA concentrations and signaling are regulated at the individual organ level and not via a maternal-fetal barrier. It is also important to note that atRA signaling is most significant during early embryogenesis, a window of gestation that cannot be feasibly studied in human tissues. As such, concordance of the findings from this study with animal models will need to be established in studies in model organisms at gestational time www.nature.com/scientificreports www.nature.com/scientificreports/ periods after main sensitivity periods to atRA. During the early gestation, the placenta is likely to play a critical role in protecting the embryo, and atRA metabolism and transport in the human placenta requires further study. Interestingly in our preliminary studies, we did not observe any formation of atRA metabolites in human placental S9 preparations (data not shown) suggesting a lack of placental metabolic barrier. Further work is needed to define potential active transport mechanisms in the placenta that modulate species differences in fetal exposure to teratogenic retinoids 4-6 and maternal-fetal transfer of retinoids during sensitive periods of development. Collection of Human livers. The study was approved by the Institutional Review Board (IRB) at the University of Washington and the studies were conducted in accordance with the guidance of Office of Human Research Protections. Human fetal liver tissues (n = 27) ranging from reported gestational days 67 to 137 were collected by the Birth Defects Laboratory at the University of Washington and flash frozen upon collection using liquid nitrogen and stored at −80 °C until ready for use. The gestational days were based on self-reports and exact dates of conception are not known. Donated tissues were from elective abortions and all donor moms signed informed consent for donating the tissues. Tissues from fetuses whose mothers had known drug use were excluded from this study. Adult liver tissues (n = 5) were from de-identified donors from University of Washington human liver bank.

Analysis of CYP26 mRNA expression.
To explore the presence of CYP26 enzymes in human fetal liver, mRNA was extracted from 18 fetal livers and five adult human livers as previously described 28 . 50-70 mg of liver was homogenized in 2 ml Omni Hard Tissue Homogenizing tubes containing 1.4 mm ceramic beads and 1 ml of TRI reagent (Invitrogen; Grand Island, NY, USA). Homogenization was conducted using Omni Bead Ruptor 24 (Omni International; Kennesaw, GA, USA) and mRNA was extracted with TRI reagent according to the manufacturer's recommendations. Total RNA was quantified using a Nanodrop 2000c Spectrophotometer (Thermo Fisher Scientific, Waltham, MA). cDNA was synthesized from 1 μg mRNA using TaqMan reverse transcription reagents (Applied Biosystems, Carlsbad, CA). The mRNAs of CYP26A1, CYP26B1 and CYP26C1 were quantified as target genes, CYP3A7 as a control gene, and β-actin, GAPDH and GUSB were evaluated as housekeeping genes. Based on the variability in the gene expression, β-actin was chosen as the housekeeping gene. mRNA expression was quantified using StepOnePlus TM q-RT-PCR (Applied Biosystems; Carlsbad, CA, USA). All samples were analyzed in duplicates and the q-RT-PCR repeated on three separate occasions. For samples that were undetected in one of the three experiments (5 samples for CYP26A1, 4 for CYP26B1 and 1 for CYP3A7) a C t value of 40 was assigned to the undetected run and the mean of the three experiments was calculated. For samples that were undetected in two of the three experiments, samples were considered as target gene undetected (2 samples for CYP26A1 and 6 for CYP26B1). The relative abundance of CYP26A1, CYP26B1, CYP26C1 and CYP3A7 mRNA expression was analyzed by the ∆∆Ct method using β-actin as a housekeeping gene and the data are presented as a fold difference in comparison to the mean value for each gene. No comparisons for expression between genes and between adult and fetal livers were done. Human primer and probe pairs for CYP26A1 (Hs01075675_m1,  Alaskan (n = 2) 106 ± 9 14.8 ± 1.5 121 ± 10 0.14 ± 0.00 Asian (n = 6) 76 ± 35 6.4 ± 3.6 82 ± 38 0.09 ± 0.04 Caucasian (n = 6) 73 ± 21 5.7 ± 3.1 79 ± 24 0.08 ± 0.03 Table 2. Descriptive analysis of atRA oxidation in the different groups of fetal livers included in the analysis. There were no significant differences in atRA oxidation rates between gestational age groups, sex, or race. www.nature.com/scientificreports www.nature.com/scientificreports/ 250 mM sucrose, 1 mM EDTA and 1 mM PMSF was added to 0.1-0.3 g of fetal liver sample and the tissue was homogenized in 2 mL Omni Hard Tissue Homogenizing tubes containing 1.4 mm ceramic beads using a 2*20 sec cycles with an Omni Bead Ruptor 24 containing dry ice in acetone (Omni International, Kennesaw, GA). The homogenates were then centrifuged at 9,000 g for 20 min to pellet cell nuclei, large organelles and unbroken cells and the resulting supernatant (S9 fraction) was stored at −80 °C. The overall protein concentration was determined using albumin as the calibration standard and a Pierce BCA Protein Assay (Thermo Fisher Scientific, Inc., Rockford IL). evaluation of atRA metabolism via in vitro Incubations. The metabolism of atRA in human fetal livers and the enzymes responsible for atRA metabolism were first qualitatively evaluated by standard incubation methods as previously described by us 13,15,33 . The specific metabolites formed from atRA by recombinant CYP26A1, CYP26B1 and CYP3A7 in comparison to human fetal liver S9 fractions from representative donors was assessed. 5 pmols of CYP26A1, CYP26B1 and CYP3A7 in 1 mL of 100 mM Potassium phosphate (KPi) buffer (pH 7.4) were incubated for 2 mins (CYP26A1) and 10 mins (CYP26B1 and CYP3A7) with 5 µM atRA at 37 °C, respectively. The incubations were quenched with ethyl acetate, metabolites extracted as previously described 15 and the product formation was then measured by LC-MS/MS as described below. For fetal liver S9 incubations two representative livers were chosen based on the mRNA expression levels of CYP3A7 and CYP26A1. In brief, 0.3 mg S9 protein in 1 mL 100 mM KPi buffer were incubated at 37 °C with 5 µM atRA. After a pre-incubation of 5 min the reactions were initiated with the addition of NADPH (1 mM final concentration) and allowed to proceed for 10 min.
The formation of atRA metabolites was measured in individual human fetal liver S9 fractions from 27 donors as previously described 33  To determine the relative contributions of CYP26 and CYP3A7 enzymes in atRA oxidation in the fetal livers, CYP selective inhibitors were used to inhibit the target enzymes in incubations with S9 fractions from four representative donors. Fluconazole was chosen as the CYP3A7 specific inhibitor as it has been shown to not inhibit CYP26A1 16 . Talarozole was chosen as the CYP26 inhibitor based on previous characterization of its potency towards CYP26A1 and CYP26B1 34 . Ketoconazole was included in the analysis as it is a well characterized pan-CYP inhibitor with high potency both towards CYP3A7 and CYP26s 16,34 . The inhibition of atRA metabolism by fluconazole (with CYP3A7) and by talarozole (with CYP26A1 and CYP3A7) was confirmed as previously described 33 . In brief, for fluconazole inhibition, atRA (10 µM) was incubated for 10 min with CYP3A7 (5 pmol/mL) in the presence and absence of fluconazole (300 µM) and the percent decrease in 4-OH-RA and 4-oxo-RA formation was quantified. For talarozole inhibition, atRA (500 nM) was incubated for 10 mins with CYP3A7 (5 pmol/mL) or 2 mins with CYP26A1 (2 pmol/mL with added 4 pmol/mL P450 reductase) in the presence and absence of talarozole (200 nM). All of the samples were extracted and analyzed by HPLC-MS/MS. Based on the data collected using recombinant enzymes, fluconazole was used at 200 µM to selectively inhibit CYP3A7 and talarozole was used at 200 nM to selectively inhibit CYP26 in human fetal liver S9 fractions from four representative donors. In addition, ketoconazole was tested at 10 µM concentration as a pan-CYP inhibitor but it is also likely more potent inhibitor of CYP3A7 than fluconazole. atRA concentration was 500 nM. Incubations were performed as described above for fetal livers and analyzed for metabolite formation by HPLC-MS/MS. The percent inhibition was calculated by comparing the metabolite formation velocity in the presence of the inhibitor to vehicle control.

LC-MS/MS methods of quantification of retinoid metabolites.
The formation of atRA metabolites was measured using an Agilent 1290 Infinity UHPLC (Agilent Technologies, Santa Clara, CA) with an Agilent Zorbax C18 column (3.5 µm, 2.1 mm × 100 mm) and coupled to an AB Sciex API 5500 Q/LIT mass spectrometer (AB Sciex, Framingham, MA) as previously reported 15 with minor modifications on the chromatography 33 . In brief, analytes were separated using a gradient elution as follows: starting from 10:90 acetonitrile: aqueous to 1:1 acetonitrile: aqueous over 0.

Characterization of CYP3A7 specific activity and Testosterone Metabolism in fetal livers.
Testosterone hydroxylation (formation of 6βOH-testosterone) was used as a CYP3A7 specific probe reaction and analyzed as previously described 33 . All incubations were performed in triplicate. Testosterone (at 100 µM, final concentration) was incubated with 0.2 mg hFL S9 protein/mL of 100 mM KPi buffer (pH 7.4). All incubations were performed using 96-well plates with a total volume of 0.1 mL per well. The mixtures were pre-incubated for 10 minutes and reactions were initiated by the addition of NAPDH (1 mM final concentration). After 10 minutes 80 µL of the incubations were added to 80 µL ice-cold acetonitrile to quench the reaction. The samples were then (2019) 9:4620 | https://doi.org/10.1038/s41598-019-40995-8 www.nature.com/scientificreports www.nature.com/scientificreports/ centrifuged at 4 °C for 20 min at 612 g, and an aliquot of the supernatant was collected for LC-MS/MS analysis. 6βOH-testosterone was quantified based on a standard curve of 6βOH-TST (10 to 500 nM). 6βOH-TST was analyzed using a Shimadzu XR DGU-20A5 UFLC (Shimadzu Scientific Instruments, Columbia, MD) coupled to an AB Sciex 3200 Mass Spectrometer (AB Sciex, Framingham, MA). Chromatography was done using a Zorbax SB-C 18 column (5 µm, 2.1 × 50 mm, Agilent Technologies, Palo Alto, CA) and gradient elution (0.3 mL/min) from initial 5:95 acetonitrile: aqueous 0.1% formic acid held for 2 min and the increased to 100% acetonitrile over 2 min and held for 1.5 min before returning to initial conditions for a re-equilibration period of 3.5 min. 6βOH-TST was detected using a mass transition of m/z 305 → 287 Da and positive ion electrospray at source parameters of 5500 V and 450 °C. Metabolite formation was quantified based on peak height and linear standard curve using Analyst software.
In vitro to in vivo scaling of atRA metabolism and Statistical Analysis. Statistical analyses were performed using Prism v.5 (GraphPad Software, Inc., La Jolla, CA). Correlation between metabolite formation velocity from atRA and testosterone as substrates in human fetal livers was tested using linear regression. Differences between atRA metabolite formation in the presence and absence of inhibitors were tested by one-way analysis of variance. Differences in atRA metabolism between fetal livers collected at gestational weeks 10-12, 12-14, 14-16, and 16-20, between different genders of the fetus, and different races were tested by one-way analyses of variance coupled with Bonferroni's Multiple Comparison Test. A p value < 0.05 was considered significant for all statistical analyses.
The overall clearance of atRA by fetal liver was predicted using standard in vitro-to-in vivo scaling methods 35,36 . First, the overall intrinsic clearance (Cl int ) of atRA metabolism in the fetal liver was calculated by summing 4-OH-RA and 4-oxo-RA formation from atRA in the fetal liver S9 fraction incubations for each individual donor and the Cl int scaled to the whole liver using Eq. 1 as described previously for adult liver 13  in which FL refers to fetal liver and the total g of fetal liver (liver weight) is specified for the specific gestational ages in Table 1 in which fetal liver blood flow is calculated as 0.5 times the blood flow of the umbilical vein (Q umbilical vein ) based on the physiology that half of the umbilical vein flow goes to fetal liver and the rest goes to the fetal heart 39 . The Q umbilical vein for each gestational age is listed in Table 1. The plasma unbound fraction of atRA (f u ) used was 0.01 based on previous report 13 .