Developmental expression of drug transporters and conjugating enzymes involved in enterohepatic recycling: Implication for pediatric drug dosing

Abstract

Around 50% of the drugs used in children are never tested for safety and efficacy in this vulnerable population. Immature drug elimination pathways can lead to drug toxicity when body surface area or weight normalized doses are administered to children. In the absence of clinical data, physiologically-based pharmacokinetic (PBPK) modeling has emerged as a useful tool to predict drug pharmacokinetics in children. These models utilize physiological differences in pediatric populations, including age-dependent differences in the abundance of drug-metabolizing enzymes and transporters (DMET), to mechanistically extrapolate adult pharmacokinetic data to children. The reported abundance data of hepatic DMET proteins in subcellular fractions isolated from frozen tissue are prone to high technical variability. Therefore, we carried out the proteomics-based quantification of hepatic drug transporters and conjugating enzymes in 50 pediatric and 8 adult human hepatocyte samples. Out of the 34 studied proteins, 28 showed significant increase or decrease with age. While MRP6, OAT7, and SULT1E1 were highest in <1 year old samples and reached minimal levels in >1 year age samples, the abundance of P-gp, and UGT1A4 was negligible in <1 year age samples and increased significantly after 1 year of age. Incorporation of the generated age-dependent abundance data in PBPK models can help improve dose prediction, leading to safer drug pharmacology in children.

INTRODUCTION

Age-dependent variability in drug elimination processes such as drug metabolism and transport pose a serious challenge in predicting pediatric pharmacokinetics and dosage. Physiological differences in the pediatric population compared to adults, such as differences in blood flow, organ volumes, glomerular filtration, intestinal absorption, immature drug metabolism, and transport pathways, can significantly elevate the risk of severe drug toxicity when applying dosages normalized to the body surface area or body weight from adult recommendations.^1–3 However, 50% of the drugs administered to children lack age-appropriate dosing guidelines and are used “off-label”. Out of the 399 drugs that were prescribed in neonates and infants from 1997–2010, only 28 of them had been studied in this vulnerable population for safety and efficacy.² In the past, clinical pharmacokinetics or safety studies have been sparse in pediatric population owing to scientific, clinical, ethical, and logistic concerns. However, with the advent of Best Pharmaceuticals for Children Act (BPCA) in 2002 and the Pediatric Research Equity Act (PREA) in 2003, the US Food and Drug Administration (FDA) now motivates and incentivizes sponsors to perform pediatric clinical studies to improve labelling of patented drug products in children. These acts have resulted in increased pediatric clinical studies and >500 pediatric-specific labelling changes 2002.^4,5 As the number of clinical studies in children is increasing, it is pertinent to predict drug pharmacokinetics, safety, and pharmacodynamics in this vulnerable population to ensure appropriate dosing. In this regard, in silico prediction tools such as physiologically-based pharmacokinetic (PBPK) modeling that integrates developmental data including the ontogeny of drug-metabolizing enzymes and transporters (DMET), is emerging as powerful approach to predict drug pharmacokinetics in children.^6,7 Such DMET ontogeny-linked PBPK models have been used to inform first-in-child dosing, design safer pediatric studies, and justify waivers for pediatric studies.^8–12

The liver is a major metabolic organ that contributes to the elimination and detoxification of ~70% of marketed drugs as well as a variety of other xenobiotics and endogenous compounds. Hepatic elimination consists of enzymatic biotransformation and transporter-mediated sinusoidal uptake and efflux, and biliary excretion of a multitude of compounds.^13,14 Drug-metabolizing enzymes are mainly categorized as oxidative, reductive, hydrolytic, and conjugating, based on the metabolic reactions they catalyse.^15,16 Hepatic transporters such as solute carrier (SLC) transporters are localized on the sinusoidal membrane facilitating cellular uptake of drugs, while ATP-binding cassette (ABC) transporters are present on the apical and basolateral membranes effluxing chemicals into the bile and blood, respectively.¹⁷ These enzymes and transporters often work in conjunction to regulate the concentration of parent xenobiotics and endobiotics as well as their metabolites. In particular, the conjugated metabolites such as glucuronide or sulfate conjugates, formed by the uridine 5′-diphospho-glucuronosyltransferase (UGT) or sulfotransferase (SULT) enzymes, respectively, are hydrophilic in nature and require transport processes to cross the plasma membrane.¹³ This transporter-enzyme interplay is an important component of UGT-transporter axis and SULT-transporter axis, which regulate the hepatic accumulation of organic compounds. Many drugs (ethinyl estradiol, etoposide, ezetimibe, gemfibrozil, irinotecan, morphine, mycophenolate mofetil, statins, telmisartan, troglitazone, etc.) and endogenous compounds are substrates of both hepatic enzymes and transporters.¹³ After transport into the hepatocytes, drugs and endogenous compounds are metabolized by enzymes, followed by excretion of the metabolites into bile or blood for biliary or renal elimination, respectively. Many of these drugs are administered to children, underscoring the necessity for ontogeny data of these enzymes and transporter proteins. Similarly, the plasma concentration of endogenous compounds such as bilirubin, steroids, prostaglandins, and fat-soluble vitamins is regulated by the interplay of conjugating metabolism and transport.^13,18,19 For example, bile acids are substrates of hepatic sinusoidal uptake and efflux transporters, conjugative enzymes, and biliary efflux transporters which together modulate their vectorial enterohepatic transport and recycling.^20,21 Similarly, drugs such as irinotecan is activated to SN-38 by carboxylesterases, which is subsequently glucuronidated and transported by multiple efflux transporters into bile or blood, where it undergoes further processing in the gut and liver.^13,22 Intestinal absorption of the biliary excreted metabolites (intact or deconjugated) together forms the enterohepatic recycling process, thus leading to a sustained exposure and longer half-lives of critical endogenous compounds such as bilirubin, hormones, vitamins, and many xenobiotics.²³

In the past three decades, ontogeny patterns of various DMET proteins have been well established across different age groups.^24–29 The age-dependent abundance of hepatic proteins has been characterized in human liver microsomal (UGTs), cytosolic (SULT), and membrane (transporters) fractions using liquid chromatography-tandem mass spectrometry (LC-MS/MS) based proteomics approach.^24,26–28 Although proteomics analysis of the enriched subcellular fractions (microsomes, cytosol, membrane, etc.) improves the analytical sensitivity, protein abundance data derived from the subcellular fractions are prone to high technical variability due to cumbersome manual tissue processing.^30,31 This technical variability can be due to either contamination of the fraction of interest with other organelles or loss of proteins present in the subcellular fraction during early fractionation steps.^30–32 While technical variability is inevitable in protein quantification using subcellular in vitro models, it can be significantly minimized by quantifying proteins in total tissue homogenates.³³ Therefore, we carried out a global proteomics analysis of 50 pediatric and 8 adult human hepatocyte samples to determine the association of age and sex with the abundance of hepatic transporters and conjugating enzymes. The resulting protein abundance data can be incorporated into PBPK models for mechanistic prediction of drug disposition in children. Additionally, we measured hepatic deconjugating enzymes, such as β-glucuronidase (GUSB) and sulfatases, which are crucial for the hydrolysis of glucuronide and sulphate conjugates. Finally, using furosemide as a substrate of UGT enzymes, we demonstrated the association between age and the formation of furosemide glucuronide.

MATERIAL AND METHODS

Materials

Ammonium bicarbonate (98% purity), bovine serum albumin (BSA), dimethyl sulfoxide (DMSO), dithiothreitol, iodoacetamide, mass spectrometry-grade trypsin, and Pierce bicinchoninic acid (BCA) total protein quantification kit were obtained from Thermo Fisher Scientific (Rockford, IL). Acetone and furosemide were procured from Sigma-Aldrich (St. Louis, MO). Formic acid, Optima mass spectrometry-grade acetonitrile, and water were procured from Fisher Chemical (Fair Lawn, NJ). INVITROGRO hepatocyte thawing and KHB medium were provided by BioIVT (Westbury, NY). Auto T4 cellometer was obtained from Nexcelom Bioscience (Lawrence, MA). Phosphate buffered saline (PBS) and ammonium acetate were procured from Gibco (Amarillo, TX) and Merck (Billerica, MA), respectively. Furosemide glucuronide and testosterone glucuronide-d3 were obtained from procured from Toronto Research Chemicals (Toronto, Canada) and Cerilliant Corporation (Round Rock, TX), respectively. Cryopreserved pediatric (n=50) and adult (n=8) human hepatocytes were provided by BioIVT (Westbury, NY). Details pertaining to procurement and preparation of human hepatocyte samples are included in the supplementary text file. Demographic details of hepatocyte samples are included in Table S1.

Protein quantification in hepatocyte samples

The total protein concentration was quantified using a BCA assay kit following the vendor protocol, and the hepatocytes were digested using the protocol described elsewhere³⁴. Detailed protocol of trypsin digestion is provided in the supplementary file. To determine the reproducibility of the proteomics assay, 10 hepatocytes were randomly chosen, and trypsin digestion was carried out on two separate days, followed by sample and data analysis. The obtained data of the 10 replicate hepatocyte samples (Day 2) were compared with the original data (Day 1) to establish inter-day precision of the analysis.

Global proteomics of digested hepatocytes using total protein approach

Global proteomics analysis of the digested samples was carried out using EASY-nLC 1200 series system coupled with Q-Exactive-HF MS instrument (Thermo Scientific, San Jose, CA) in the data-independent acquisition (DIA) and positive ionization modes. The MS data (.RAW files) were analyzed using DIA-NN software (v1.8.1)³⁵ with Homo sapiens proteome library for human hepatocyte samples, followed by abundance determination using the total protein approach (TPA).^36,37 Detailed information about LC and MS instrument settings, DIA-NN parameters, and TPA are given in the supplementary file.

Statistical analysis

Statistical analysis was done using GraphPad Prism (v8.4.3; San Diego, CA), and Microsoft Excel (v2209; Redmond, WA). All the graphs were developed using GraphPad Prism. To study the association of age with protein abundance, the age groups were analyzed using Kruskal-Wallis test followed by Dunn’s multiple comparison test. The continuous age-dependent abundance data of proteins were fitted using a non-linear allosteric sigmoidal model described previously.²⁶ Details of the model equation are provided in supplementary file. Comparison of the abundance of proteins between males and females was carried out using Mann-Whitney test. The sex-dependent differences were compared for the hepatocyte samples >1 yr of age (n: male=22, female=19). Hierarchical clustering of the normalized abundance values (G-scores) was performed using Perseus software (v2.0.7.0; Ontario, Canada). For the protein-protein correlation, protein abundance values were normalized to housekeeping proteins such as sodium-potassium ATPase pump (Na⁺K⁺ ATPase; for plasma membrane transporters), calnexin (CANX; for membrane-bound UGT enzymes), and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; for cytosolic SULT enzymes).

Furosemide glucuronide formation assay in human hepatocyte samples

Furosemide activity assay was performed in 52 cryopreserved hepatocytes (47 pediatric and 5 adult) using 0.1 × 10⁶ cells per well, wherein furosemide was spiked in triplicate (incubation concentration: 50 μM, <1% DMSO) and incubated for 120 min at 37 °C and 5% CO₂. Details of the assay protocol, sample quenching, sample preparation and LC-MS based quantification of furosemide glucuronide are provided in the supplementary text file and Table S2.

Determination of the fraction of furosemide glucuronidated by UGT1A1 and UGT1A9

The fraction of furosemide metabolized by different UGT enzymes was calculated using reported data,^27,38 and detailed information is provided in the supplementary file.

RESULTS

Protein abundance and variability of hepatic drug transporters and conjugating enzymes

Although the overall distribution of transporters and conjugating enzymes was variable across different hepatocyte samples, age-dependent protein abundance was clearly evident (Figure 1, Table S3). Organic anion transporter 7 (OAT7) was the most abundant uptake transporter in the younger age samples (age 0–12 d), whereas monocarboxylate transporter 1 (MCT1) and OCT1 were abundant in hepatocytes of the later age groups (Figures 1a and S1a). Similarly, MRP6 was the most abundant efflux transporter up to 1 year of age, representing ~52% of all the studied efflux transporters (Figures 1b and S1b). P-gp was absent or barely detected in younger age group samples (age 0 d to 1 year), but it was abundant in all the other samples (~10%). UGT2B4 and UGT2B15 were abundant as compared to the other UGTs in the newborns and neonates (age, 0–12 days). The proportion of UGT1A1 rose from 2% at birth to 26% at 1 year of age, attaining ~10% level in the older age groups (Figures 1c and S1c). In adults (>18 years age), UGT2B7 (36%) was the most abundant UGT, followed by UGT2B15 (15%). SULT1E1 was majorly detected in the samples up to 12 d of age (Figures 1d and S1d). Until the age of 12 days, SULT1A1 was the predominant SULT enzyme, whereas SULT2A1 exhibited greater abundance in all subsequent age categories. Overall, OAT7, P-gp, UGT2B17, and SULT1E1 exhibited the highest variability among uptake transporters, efflux transporters, UGTs, and SULTs with a %CV of 95, 80, 280 and 109%, respectively (Table S3). Although variability in the protein abundance can stem from either biological or technical factors, technical variability was ruled out through analysis of housekeeping proteins and repeat analysis of ten randomly chosen hepatocyte samples digested and analyzed on two different days. The abundance of marker proteins (Na⁺K⁺ ATPase, CANX, and GAPDH) was relatively consistent across age groups as compared to the DMET proteins (Figure S2). Further, abundance data of the studied DMET proteins showed strong and significant correlation on repeat digestion and analysis on two different days (r > 0.99, p-value < 0.0001, Figure S3).

Figure 1. Inter-individual variability in the abundance of hepatic drug transporters and conjugating enzymes.

Horizontal stacked bar charts illustrating the relative abundance of hepatic (a) uptake transporters, (b) efflux transporters, (c) UDP-glucurosyltransferases (UGTs), and (d) sulfotransferases (SULTs) in 58 cryopreserved human hepatocyte samples (age, 1 day- 63 years). The left side of the charts displays the data categorized by age, while the right side represents the data categorized by total abundance of all proteins in each donor.

Association of age with the abundance of hepatic uptake and efflux transporters

The abundance of MCT1, OAT2, OATP1B1, OATP2B1, and OCT3 was independent of age (Figure 2). MCT8 abundance was significantly lower in children aged 0–12 d and 26 d-1 yr compared to >1–2 yr (59% lower), and >1–2 yr (63% lower), >2–6 yr (50% lower) and >6–12 yr (55% lower). Similarly, thiamine transporter 2 (THTR2) abundance was 54% and 38% lower in 0–12 d and 26 d-1 yr age groups, respectively, in comparison to the >12–18 yr group. While average OCT1 abundance was ~95% lower in the 0–12 d group compared to the >12–18 yr and >18 yr group, OAT7 abundance was ~4–6-fold higher in the 0–12 d group in comparison to >2–6 yr, >6–12 yr, >12–18 yr and >18 yr groups. The abundance of OATP1B3was 2.6-fold higher in the 26 d-1 yr age group compared to the >2–6 yr age group. The ontogeny equation used to fit the protein abundance levels showed a gradual increase in levels of MCT1, THTR2, and OCT1 (Figure S4a). There was a steep increase in the abundance of MCT8 at ~1 yr and of GDP-Fucose Transporter 1 (FUCT1) at ~0.1 yr. The abundance of OAT7 was higher in early age and decreased to minimum levels in >1 yr age group samples.

Figure 2. Age-dependent abundance (categorical data) of hepatic uptake and efflux transporters measured in cryopreserved human hepatocytes (n=58).

The truncated violin plot indicates the range, median, 25th and 75th percentiles of protein abundance values. Statistical comparisons between different age-groups were conducted using the Kruskal-Wallis test followed by Dunn’s multiple comparison test, with significance levels indicated as follows: p-value < 0.05 (*); < 0.01 (**); < 0.001 (***); and < 0.0001 (****). The age-groups and the number of samples in each age-group are labeled as follows: A (0–12 days, n=7); B (26 days-1 year, n=10); C (>1–2 years, n=4); D (>2–6 years, n=18); E (>6–12 years, n=6); F (>12–18 years, n=5); and G (>18 years, n=8).

The abundance of P-gp, and MRP3 was ~100% and ~50% lower in the 0–12 d group compared to the >18 yr group (Figure 2). While the abundance of BSEP and MRP2 did not show a significant age-dependent pattern, MRP6 was significantly higher in the 0–12 d and 26 d-1 yr groups compared to the >18 yr group. The abundance of P-gp, BSEP, and MRP3 increased with age, with average Age₅₀ values of ~58 d, 1 d, and ~1.5 d, respectively (Table S4). MRP6 abundance decreased with age (Figure S4a).

Association of age with the abundance of hepatic conjugating enzymes

The abundance of UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, UGT2B7, UGT2B15, and UGT2B17 was significantly lower in 0–12 d group compared to the >18 yr group (Figure 3). For example, a 40-, 9-, and 3.5-fold difference in the average abundance of UGT1A4, UGT2B7, and UGT2B15, respectively, was observed between hepatocyte samples representing the 0–12 yr and >18 yr groups. The abundance of UGT1A1 and UGT2B4 was >98% and ~62% lower in the 0–12 d group, respectively, in comparison to the >1–2 yr and >2–6 yr age groups. UGT2A1, UGT2A3, and UGT2B10 abundance showed non-significant age-related changes. UGT1A1, UGT1A3, and UGT2B4 showed a steep increase in abundance, reaching adult values by 10–40 d of age (Figure S4b). The average Age₅₀ values for UGT1A4, UGT1A6, UGT1A9, and UGT2B7 were 125 d, 2.6 yr, ~34 d, and 40 d, respectively. UGT2B17 abundance data did not conform to the ontogeny model due to the confounding effect of gene deletion.

Figure 3. Age-dependent abundance (categorical data) of hepatic UGT enzymes in cryopreserved human hepatocytes (n=58).

The truncated violin plot indicates the range, median, 25th and 75th percentiles. Statistical comparisons between different age-groups were conducted using the Kruskal-Wallis test followed by Dunn’s multiple comparison test, with significance levels indicated as follows: p-value < 0.05 (*); < 0.01 (**); < 0.001 (***); and < 0.0001 (****). The age-groups and the number of samples in each age-group are labeled as follows: A (0–12 days, n=7); B (26 days-1 year, n=10); C (>1–2 years, n=4); D (>2–6 years, n=18); E (>6–12 years, n=6); F (>12–18 years, n=5); and G (>18 years, n=8).

SULT1A1 and SULT2A1 abundance was 51 and 76% lower in the 0–12 d group, respectively, compared to the >2–6 yr group (Figure 4). Further SULT2A1 abundance was also lower in the 0–12 d group compared to the 26 d-1 yr (70% lower) and >1–2 yr (84% lower) age groups. SULT1E1 levels were highest in children aged 0–12 d. In age groups >12 d, SULT1E1 levels were negligible. While SULT1B1 was absent in the 0–12 d group, a significant 2-fold difference was observed in the average abundance between the 26 d-1 yr and >18 yr groups. The Age₅₀ values for SULT1A1, 1B1, and 2A1 were 18 d, ~10 yr, and12 d, respectively (Figure S4c, Table S4).

Figure 4. Age-dependent abundance (categorical data) of hepatic SULT enzymes in cryopreserved human hepatocytes (n=58).

The truncated violin plot indicates the range, median, 25th and 75th percentiles of protein abundance values. Statistical comparisons between different age-groups were conducted using the Kruskal-Wallis test followed by Dunn’s multiple comparison test, with significance levels indicated as follows: p-value < 0.05 (*); < 0.01 (**); < 0.001 (***); and < 0.0001 (****). The age-groups and the number of samples in each age-group are labeled as follows: A (0–12 days, n=7); B (26 days-1 year, n=10); C (>1–2 years, n=4); D (>2–6 years, n=18); E (>6–12 years, n=6); F (>12–18 years, n=5); and G (>18 years, n=8).

Ontogeny-based protein classification

Based on the continuous age-dependent abundance data (Figure S4), all the studied proteins were classified into three categories. The class 1 proteins showed the highest expression after birth and were silenced or reduced to low levels by 2 years of age. MRP6, OAT2, OAT7, and SULT1E1belonged to class 1 (Figure 5). The abundance of the class 2 proteins remained relatively constant throughout the age range and included MRP2, OATP1B1, OATP1B3, OATP2B1, OCT3 and 3`-phosphoadenosine 5`-phosphosulfate transporter 1 (PAPST1). Class 3 included proteins that were expressed at negligible/low levels at birth and significantly increased after birth to reach adult levels. The following 24 proteins belonged to this class: BSEP, GDP-fucose transporter 1 (FUCT1), MCT1, MCT8, MRP3, OCT1, P-gp, SULT1A1, SULT1B1, SULT2A1, THTR2, UGT1A1, UGT1A3, UGT1A4, UGT1A6, UGT1A9, UGT2A1, UGT2A3, UGT2B4, UGT2B7, UGT2B10, UGT2B15, UGT2B17, and UDP-GA/UDP-N-acetylgalactosamine transporter (UGTrel7) (Figure 5).

Figure 5. Pie chart illustrating hepatic drug transporters and conjugating enzymes categorized into different groups based on their developmental trajectories.

Class 1 proteins exhibit the highest protein abundance shortly after birth but are subsequently silenced or reduced by the age of 2 years. Class 2 proteins maintain relatively constant levels across different ages. Class 3 proteins either exhibit minimal or low levels at birth and undergo a significant increase after birth, eventually reaching adult levels.

Protein-protein correlations

Hierarchical clustering of the G-scores of protein abundance values of hepatocytes from donors older than 1 yr of age, showed clustering of various proteins (Figure S5). BSEP, MCT8, and THTR2 formed a cluster, while MRP2, OATP2B1, and SULT2A1 were grouped in another cluster. Similarly, MRP6, and OAT7 were also clustered together, indicating a potential correlation and coregulation of their abundance. Therefore, we determined the protein-protein correlations of the normalized protein abundance values (Figure 6). BSEP correlated positively (r > 0.7; p-value < 0.05) with MCT8, OATP1B3, OCT1, THTR2, and UGT1A4. The results of other protein-protein correlations are discussed in the supplementary file.

Figure 6. Significant positive correlations among hepatic drug transporters and conjugating enzymes in cryopreserved human hepatocytes (>1 yr of age samples).

Correlations are deemed significant if the Pearson correlation coefficient is equal to or greater than 0.5, with a corresponding p-value of less than 0.05.

Ontogeny of furosemide glucuronide formation in human hepatocytes

The furosemide glucuronide formation showed an increasing trend with age (Figure 7a); however, the association was not significant (Figure 7b). For all the individual samples, furosemide glucuronide formation moderately correlated with UGT1A9 (r = 0.41) and UGT1A1 (r = 0.25) (Figure 7b–c). The average furosemide glucuronide formation showed a strong and significant positive correlation with average UGT1A9 abundance (r = 0.93, p-value < 0.05). Based on the average abundance and activity data, UGT1A9 was not involved in the glucuronide formation in 0–12 d age group, whereas f_m,UGT1A1 and f_m,UGT1A9 changed to 0.63 and 0.37, respectively, in children age 26 d-1 yr (Figures 7 and S6). Interestingly, UGT1A1 was only detected in two out of six 0–12 d group samples but furosemide glucuronidation activity was present in all six hepatocytes (Figure S6). This suggests that in the absence of UGT1A1 and 1A9, other UGT enzymes such as UGT1A3, 1A6 and/or 2B7 may catalyze the glucuronidation reaction. The same can be inferred from the y-intercept of Figures 7b and 7d. Thereafter, the contribution of UGT1A9 to glucuronide formation increased to 0.62 in >1–2 yr samples. For age 12 yr and above, >65% of furosemide glucuronide was formed by UGT1A9. A detailed explanation of these results is provided in the supplementary file.

Figure 7. Ontogeny of furosemide glucuronide formation in human hepatocytes.

(a) Continuous and categorical age-dependent data of furosemide glucuronide (FG) formation in cryopreserved human hepatocytes (n=52). Correlation of FG formation with (b) UGT1A9, and (c) UGT1A1. (d) Fractional contribution of UGT1A1 (f_m,UGT1A1) and UGT1A9 (f_m,UGT1A9) in FG formation across different age-groups.

Association of the abundance of hepatic drug transporters and conjugating enzymes with sex and of deconjugating enzymes (GUSB and arylsulfatases) with age and sex is discussed in the supplementary file (Table S5, Figure S7 and S8).

DISCUSSION

A multitude of drugs such as etoposide, ezetimibe, furosemide, irinotecan, mycophenolate mofetil, statins, and telmisartan are given to children and are substrates of hepatic drug transporters and UGT or SULT enzymes.^13,19,39 The abundance of drug transporters and conjugating enzymes shows age-dependent maturation,^24,26–28 and differential ontogeny patterns of these proteins have been implicated in the altered rate-determining step in the hepatic elimination of drugs. Mechanistic prediction of these changes in the hepatic elimination across different age groups using PBPK modeling is crucial for dosing during drug development and in clinic.¹² Indeed, the available DMET protein abundance data across different age groups have been utilized to expand PBPK models for pediatric aplcaitions.^26,40,41

In the last decade, the age-dependent abundance of hepatic DMET proteins such as CYPs, UGTs, SULTs, and transporters has been characterized in human liver microsomal, cytosolic, and membrane fractions.^24,26,27 While the enrichment of subcellular fractions improves the method sensitivity, it can lead to high technical variability. Microsomes and membrane fractions are highly contaminated by other organelles as shown by the concentration of ribosome marker (RPL7A), proteasome marker (PSMD1), mitochondria marker (COX4I1), and Golgi body marker (formimidoyltransferase cyclodeaminase) in enriched microsomes.^30,42 Further, endoplasmic reticulum-associated proteins are also lost during early fractionation steps of microsome preparation, leading to variable enrichment of CYP and UGT enzymes in microsomes.³¹ Cytosolic fractions are frequently contaminated by mitochondria-specific soluble enzymes caused by freezing-related damage of mitochondria, leading to the release of components from the matrix.³² Protein quantification in these subcellular fractions obtained from frozen tissues can lead to high technical variability due to differential enrichment and absence of cell viability information. Although enrichment variability can be minimized by using total tissue homogenates for protein quantification,³³ the effect of variable tissue collection and handling on cell viability can lead to high technical variability in the proteomics data. Therefore, to reduce technical variability, we characterized the abundance of drug transporters and conjugating enzymes in cryopreserved human hepatocytes. In comparison to frozen tissue samples with unknown cell viability, we estimated the cell viability in these hepatocytes before performing proteomics analysis of DMET proteins in the viable cells.

The present study reports the ontogeny profiles of the following proteins in human liver for the first time: MCT1, MCT8, THTR2, OCT3, OAT2, OAT7, PAPST1, FUCT1, UGTrel7, MRP6, UGT1A3, UGT2B4, and UGT2B10. OAT7 is a basolateral uptake transporter specifically involved in the transport of sulfate conjugates of endogenous and exogenous compounds such as, estrone sulfate, dehydroepiandrosterone sulfate, β-estradiol sulfate, and minoxidil sulfate.⁴³ Riedmaier et al. identified pravastatin as a substrate of OAT7 and postulated that OAT7 may contribute to its uptake in conditions of OATP inhibition.⁴⁴ The abundance of OAT7 was highest in 0–12 d age group, reaching minimal levels in 26 d-1 yr and above age group samples. Drug substrates of OAT7 given to newborns can show differential hepatic uptake compared to adults. Similarly, MRP6 abundance was also significantly higher in 0–12 d and 26 d-1 yr age group samples. MRP6 is majorly expressed on the basolateral membrane of hepatocytes and renal proximal tubule cells and is involved in the transport of glutathione S-conjugates such as leukotriene C₄ and N-ethylmaleimide S-glutathione.^45–48 Mutations in MRP6 are associated with a heritable connective tissue disorder called pseudoxanthoma elasticum (PXE) which is marked by the presence of dystrophic elastic fibers in the skin, retina, and large blood vessels, causing loss of vision, appearance of bags in the skin, and calcification of blood vessels.^46,48 While some of the endogenous substrates of OAT7 and MRP6 are known, the drug substrates of these transporters have not been well characterized. Furthermore, physiological basis for the high abundance of these transporters in neonates and potential ontogeny switching with another transporter needs to be studied. We postulate that the high abundance of OAT7 and MRP6 in <1 yr age is critical for the transport of endogenous substrates and inhibition of these transporters by drugs and other xenobiotics can affect their function, leading to pathophysiological consequences in children.

We also reported the ontogeny data of other DMET proteins. The generated data correlated well with the limited available studies conducted previously by us and others.^{24,26–28,49–51} SULT1E1 is highly abundant in the prenatal liver and drastically decreases postnatally to reach negligible levels by 1 yr of age.^50,52 This is corroborated by the current study wherein SULT1E1 levels were highest in 0–12 d age group samples and decreased by 3-fold in 26 d-1 yr samples to finally reach negligible levels in all samples >1 yr of age. During prenatal development, SULT1E1 levels are significantly higher in male fetuses to inactivate estrogen for maximum androgen intensity during gonadal organogenesis.⁵⁰ Ontogeny data of UGT2B4, UGT2B7, and UGT2B17 generated by targeted proteomics of these hepatocyte samples has previously been reported.³⁴ UGT1A3 is involved in the glucuronidation of drugs like statins, losartan, opioids, etc. 25-Hydroxyvitamin D₃ (25OHD₃), a clinical biomarker for the assessment of Vitamin D levels, is metabolized by hepatic UGT1A3 and UGT1A4 to form 25OHD₃-3-O-glucuronide, which is a substrate of MRP2, OATP1B1, and OATP1B3.^53,54 This glucuronidation along with SULT2A1-mediated sulfation, followed by enterohepatic transport is physiologically relevant in maintaining vitamin D homeostasis.^53,54 UGT1A6 catalyzes the glucuronidation of planar phenols and primary amines and includes examples such as serotonin, estradiol, acetaminophen, valproic acid, deferiprine, β-blockers, etc.⁵⁵ The abundance of UGT1A6 was ~5 to 7-fold higher in adults compared to 0–12 d and 26 d-1 yr groups, highlighting potentially large age-dependent effects on the metabolism of UGT1A6 substrates in younger children. Further, it has been reported that during the first week after birth, concentration of uridine diphosphate glucuronic acid (UDP-GA) can be the rate-limiting step in glucuronidation reaction.⁵⁶ For example, lower clearance values of morphine, propofol, and zidovudine during first 7–10 days of age have been reported which can partially be explained by UDP-GA levels.^56–58 While immature UGT abundance can partially justify the decreased clearance, low UDP-GA concentration can become rate-limiting step in the glucuronidation reaction.⁵⁶ Similarly, although phenobarbital-mediated induction of UGT1A1 failed to increase bilirubin clearance, lower UDP-GA levels were consistent with hyperbilirubinemia in the first four days after birth.⁵⁹ Further, UGTrel7 protein, encoded by SLC35D1, is present in endoplasmic reticulum and is involved in the transport of UDP-GA and UDP-N-acetylgalactosamine from cytoplasm to the endoplasmic reticulum lumen where they act as cofactors for glucuronidation reaction. In the current study, we quantified the abundance of hepatic UGTrel7 across different age-groups for the first time. The abundance of UGTrel7 was 25% lower in 0–12 d age-group compared to the other age-groups. Among the marker proteins analyzed in this study, CANX and GAPDH did not show significant age-dependent differences. Na⁺K⁺ATPase levels were also consistent across age-groups, except for a moderate but statistically significant difference (<15%) between 26 d-1yr versus >18 yr age groups. There were no sex-dependent differences in the abundance of these marker proteins.

Deconjugating enzymes, GUSB and sulfatase, deconjugate glucuronide and sulfate conjugates, respectively. While the bacterial deconjugating enzymes contribute majorly to the enterohepatic recirculation of endogenous and exogenous compounds, human forms of GUSB and sulfatases are also present in the liver and other tissues,^60,61 which may play critical roles in regulating intracellular concentrations of parent compounds and conjugated metabolites. Therefore, we determined the abundance of GUSB and sulfatases across different age groups. GUSB is widely distributed in the tissues and body but is majorly present in liver, spleen, endometrium, breast, and adrenal gland.⁶¹ The serum levels of GUSB are elevated in late pregnancy, postpartum period, in postmenopausal women on estrogen therapy and in cancers of breast, cervix, and uterus following estrogen or androgen therapy.^61,62 It plays an important role in bilirubin and steroid metabolism.⁶¹ Drugs like ciprofloxacin can inhibit GUSB leading to decreased enterohepatic recirculation of mycophenolic acid.⁶³ Similarly, sulfatases play a critical role in various cellular functions (hormone regulation, cellular degradation, modulation of signalling pathways), by regulating the sulfation of substrates such as small cytosolic steroids and cell-surface carbohydrates.⁶⁰ Here, we also quantified the abundance of arylsulfatases (ARSD and ARSE), which are localized in the endoplasmic reticulum and Golgi apparatus, respectively.⁶⁴ While the substrates of these enzymes have not been well characterized, ARSD is potentially responsible for iodothyronine sulfatase activity and ARSE mutation has been linked to a congenital disorder called X-linked chondrodysplasia punctata (CDPX1) and its inhibition by warfarin leads to warfarin embryopathy (clinical manifestations similar to CDPX1).^60,64–67

Using furosemide as a substrate of UGT enzymes, we demonstrated the association of age with the formation of furosemide glucuronide. UGT1A9 and UGT1A1 are majorly involved in the glucuronidation of furosemide in adults, with a fractional contribution of 75 and 25%, respectively.³⁸ While the abundance of these enzymes showed an age-dependent pattern, the furosemide glucuronide formation was independent of age. Further, the correlation of abundance vs furosemide glucuronide formation was weak but significant. We hypothesize that the uptake of furosemide, a poorly permeable drug, is the rate-limiting step leading to non-significant changes in glucuronide formation with age and poor activity-abundance correlation. Further, in samples <12 d of age, other UGT enzymes could be contributing to the glucuronidation of furosemide. Taken together, these factors lead to high variability (>70% CV) in the glucuronidation of furosemide across all the samples. Although this study provides comprehensive DMET ontogeny data, there are a few limitations of the study. First, this study is just observational and does not provide a mechanistic information on the ontogeny regulation mechanisms. However, the ontogeny could be partly explained by the age-dependent changes in plasma concentration of bile acids and steroidal hormones that are ligands of nuclear receptors, such as farnesoid X receptor (FXR), pregnane X receptor (PXR), and constitutive androstane receptor (CAR), which are involved in the regulation of hepatic transporters and conjugating enzymes.^18,68–71 Second, the age-dependent differences in the abundance of some of the polymorphic proteins can be confounded due to genetic polymorphisms. Further, the hepatocytes used in this study have not been qualified for transport activity across all age groups. In the current study, the samples were analyzed using total protein approach (TPA) based quantitative proteomics. TPA is a label-free approach for the quantification of large-scale proteomic datasets, wherein the peptide and total MS intensities are used to quantify the abundance of all the detected proteins. The replicate digestion and analysis of 10 randomly chosen hepatocyte samples yielded the abundance data that showed significant consistency with the original data, highlighting the robustness of the TPA-based analysis. Nevertheless, there are a few limitations of global proteomics approach. For example, differences in the sequence coverage of proteins can lead to under- or over-prediction of absolute abundance values of low-abundant proteins.⁷² Further, human proteome databases and software tools such as DIA-NN are continuous evolving, which could yield slightly different absolute protein abundance values when the same data are analysed again. Despite these limitations, we don’t anticipate differences in the relative data, e.g., fold-changes between various age groups. Membrane transporters such as NTCP, BCRP, ENT1, MRP4, OSTα/β were not quantified in this study due to analytical challenges. For example, NTCP is a small transmembrane protein (38 kDa) with limited tryptic digestion sites outside the transmembrane domain, making it difficult to detect and quantify surrogate peptides. Also, for proteins such as OAPT1B3, and PAPST1, the abundance data did not fit to the ontogeny model due to the high interindividual variability and limited sample size. Despite these limitations, the findings presented here offer valuable insights into the ontogeny of DMET proteins, which has practical implications, as it can be leveraged to refine pediatric PBPK models. By incorporating age-specific data on DMET protein expression, these models can yield more accurate predictions of how drug pharmacokinetics vary throughout human development, spanning from neonatal age to adulthood. Ultimately, such improvements in predictive modeling can contribute to safer and more effective medication dosing across different pediatric age groups.

STUDY HIGHLIGHTS.

1. What is the current knowledge on the topic?

Ontogeny data on the abundance of some hepatic drug transporters and enzymes using subcellular fractions (microsomes and cytosol) are available, but similar data are not available using viable hepatocytes. Additionally, ontogeny data for many hepatic drug transporters and enzymes are also limited.

2. What question did this study address?

This study characterized the association of age and sex with various hepatic drug transporters and conjugative enzymes using global proteomics analysis using a large number of human hepatocytes. We also analysed the protein-protein correlations to determine potential co-regulations of these proteins.

3. What does this study add to our knowledge?

A total of 34 hepatic drug transporters and conjugative enzymes were quantified in 50 pediatric and 8 adult human hepatocyte samples. The ontogeny patterns for various proteins (MCT1, MCT8, THTR2, OCT3, OAT2, OAT7, PAPST1, FUCT1, UGTrel7, MRP6, UGT1A3, UGT2B4, and UGT2B10) have been characterized for the first time.

4. How might this change clinical pharmacology or translational science?

With the introduction of legislation recommending pediatric clinical trials, the number of clinical investigations in children has increased globally. The ontogeny data generated in this study constitutes an essential resource for incorporation into PBPK modeling, aiding in the enhancement of drug pharmacokinetics prediction in children.