Discussion

We present from a large cohort of pediatric and adult patients the intestinal protein abundance data of DMEs and DTs that are relevant to intestinal drug metabolism and transport. To our knowledge, this is the first report demonstrating the application of TQP to quantify the DMEs and DTs in the intestine of the pediatric population. With our newly developed TQP method, we were able to quantify all proteins in both children and adults. Somewhat to our surprise, the impact of age on DME and DT abundance was less evident than we expected based on the limited pediatric data from intestine and the more extensive liver, kidney, and blood-brain barrier data on transporter ontogeny (Mooij et al., 2016; van Groen et al., 2018; Cheung et al., 2019; Verscheijden et al., 2020).

Previously reported intestinal pediatric protein level data originate from IHC or Western blot experiments. Gene expression values do not always predict protein abundance, but for certain proteins strong correlations between age and gene expression were described (Drozdzik et al., 2018, 2019). Indeed, an age-related decrease of OATP2B1 was previously observed at the gene expression level but not in our novel proteomics results when arranged into age groups (Fig. 1) (Mooij et al., 2014). In contrast, our measurements support reverse-transcription polymerase chain reaction analyses indicating that infants express lower quantities of MRP2 compared with adults (Mooij et al., 2014). In our study, BCRP abundance correlates positively with age in both the jejunum and ileum. This finding is in contrast with previous IHC analysis showing no difference between fetal and adult samples (Konieczna et al., 2011). Mooij et al. (2016) detected PEPT1 protein in neonates and infants with decreasing IHC staining intensity toward adolescence, whereas their mRNA data showed a significant increase between the neonatal/infant and the older age groups. Our current findings are in good agreement with the latter—but not with the IHC results—and support that PEPT1 increases with PNA in the ileum (Fig. 1; Table 1). Using Western blotting, Johnson et al. (2001) showed that CYP3A4 protein levels displayed significant differences between fetuses and all other age groups as well as between infants and children above 5 years of age. Similarly, Chen et al. (2015) showed an increase in CYP3A4 after birth up to 11 months PNA with the same technique. Our data indicate a significant increase with age in the jejunum, echoing earlier observations. Studies on MDR1 showed no differences among age groups, neither on the mRNA nor on the protein level after 12 weeks of gestation compared with adults (van Kalken et al., 1992; Fakhoury et al., 2005; Konieczna et al., 2011; Mooij et al., 2014). Interestingly, our age group data shows an age-related increase of MDR1 protein abundance in the ileum (Fig. 1), although significance was not retained after FDR correction on a continuous scale (Table 1). UGT1A1 levels were significantly higher in adults than in the youngest patient group in both intestinal segments tested (Fig. 1) and positively correlated with age in the ileum when measured on a continuous scale (Fig. 2; Table 1). A similar developmental trend was seen in mice expressing humanized UGT1A1 (Fujiwara et al., 2010). Higher protein abundances for UGT1A1 were measured in the adult jejunum and ileum compared with adolescent tissues. This difference was probably due to the administration of irinotecan, a frequently administered agent in adult patients with cancer, which is metabolized by UGT1A1 and can induce its expression (Takahara et al., 2013). Unfortunately, we did not have access to the patient information to check for drug exposure.

UGT2B7 abundance was found to decrease with increasing age (Table 1). This result was unexpected because UGT2B7 is a major enzyme whose absence is responsible for the chloramphenicol-induced Gray baby syndrome in neonates (Kauffman et al., 1981; Chen et al., 2010). Interestingly, this finding suggests organ-dependent ontogeny. Additionally, based on adult UGT2B7 values, less than 1% of the total enzyme is found in the intestine, whereas the majority is produced in the liver (Drozdzik et al., 2018).

Our study is the first one indicating that MCT1 protein levels may correlate negatively with age in intestines.

In addition, ASBT is normally only present in the ileum; however, in our study it was detected in some of our jejunal samples, whereas in this segment it was BLQ in previous works with one exception (Gröer et al., 2013; Drozdzik et al., 2014, 2019; Akazawa et al., 2018). This observation can be attributed either to varying detection limits of the different methods or to external factors. Diet, for instance, can modulate the expression of transporters. For example, the expression of Slc10a in the jejunum has been shown to be increased after oat-containing diet in mice (Andersson et al., 2017). Rats exhibited similar upregulation of Slc10a2/ASBT by 1α,25-dihydroxyvitamin D₃ (Chen et al., 2006).

To highlight the developmental changes from birth to adulthood, we have included adult samples in our dataset. Also, adult values allow a comparison of the current dataset with literature values, since this is the first time pediatric intestinal protein abundance levels are reported. The recent article by Couto et al. (2020) provides a summarized overview of most recent publications on TQP of adult human intestinal ADME proteins. All research groups applied their own unique methodology, which makes the comparison of the current dataset to the literature data challenging. When comparing adult protein abundance values from our study with literature values, it becomes obvious that the non-normalized values are higher in comparison with data obtained by other research groups. The underlying reasons can include the fast preservation of the collected tissue samples, the use of the membrane fractionation, application of PI cocktail from the beginning of the sample preparation, thorough sonication, or extensive digestion. Consistency in the digestion was addressed by spiking equal amounts of horse myoglobin into all the samples prior to digestion (Supplemental Fig. 4). Addition of heavy peptides was used also as a quality-control measure to correct for any discrepancies that might occur during sample clean-up and with the LC-MS/MS system.

Protein abundance data in the current study are presented in two ways: 1) relative to villin-1, a constitutively expressed age-independent marker for normalizing for enterocyte levels in intestinal samples, or 2) without normalization. The advantage of normalization is to overcome the issue of uneven sampling in a heterogenous, multilayered tissue like the small intestine. Since in the intestinal tract villin-1 is only expressed in the enterocytes, it is a suitable normalization factor for the number of enterocytes (West et al., 1988). Additionally, it has previously been shown that villin-1 is stable across age groups, including the fetal and neonatal groups (Johnson et al., 2001). Also, according to our own measurements, villin-1 abundance is stable across age groups. However, most published studies describe uncorrected data, and only few research groups applied villin-1 normalization (Vaessen et al., 2017; Zhang et al., 2020). We have chosen to present our data in both ways in an attempt to make cross-comparison with literature results feasible.

A relatively large proportion of pediatric samples had abundance values below the LLOQ. The number of proteins that fall below the LLOQs per tissue correlated with villin-1 abundance (unpublished data). Consequently, LLOQ data may be explained by low villin-1 abundances. Because villin-1 abundance is directly proportional to the number of enterocytes in the sample, we hypothesized that the average yield of all proteins was similarly affected, explaining our finding of relatively frequent protein abundances below the LLOQ, and fortified our assumption that normalization was necessary for the interpretation of the data. We used surrogate values to address the issue and increase statistical power. As we are aware that imputing missing with low values may bias the results, we conducted our statistical analyses with or without including the LLOQ/2 values. Table 1 compares the different outcomes of these calculations. Leaving out BLQ values would be inappropriate because these would then be “missing not at random,” which is poison to most statistical methods.

Although no striking differences were found in most protein abundances, it does not mean that there is no age effect on the intestinal metabolism and pharmacokinetics of drugs. There are multiple reasons that back this observation: 1) The effect is influenced/masked by medications and food the patients received around the time of sampling. These drugs might change the abundance of DTs and DMEs. Children admitted to hospitals usually have serious health conditions and require medication(s). Disease can also affect DT and DME quantities, but collecting tissues from healthy children is not possible. To reduce the impact of disease we only collected tissue from nondiseased areas. For the stoma tissues, we used the nondiseased areas of the tissue as close to the excision site as possible to avoid using externally exposed tissue. 2) The effect of age is not linear but follows a certain developmental trajectory. This could be solved by including higher number of patients divided into more age groups, but sample collection is a main challenge. 3) Despite our efforts to collect leftover tissue material from all age groups, the neonatal age group was underrepresented in our study population. 4) Genetic polymorphisms may affect gene expression and, ultimately, the function of DTs and DMEs. These investigations would be highly valuable but also introduce new variables. In this experimental setup, drawing proper conclusions would require a larger population. 5) Furthermore, TQP quantitates the protein of interest but gathers no information about its activity. 6) Posttranslational modifications could account for discrepancies when the abundance of a protein does not correlate with its activity. 7) DME and DT activity measurements were out of the scope of the current work, but future studies can bridge this gap in the ontogeny of DTs and DMEs.

The possible implications of our findings include the therapeutic dosing of drugs for pediatric patients. In the current clinical practice, accurate dosing is imperative for agents metabolized by CYP3A4 and UGT1A1, which were found to exhibit age-dependent maturation. Furthermore, BCRP, MDR1, and MRP2 are the main transporters affecting the absorption of the majority of therapeutic substances, thus in the future measuring their abundance could provide the basis for precision drug dosing.

In conclusion, this is the first time that TQP was applied to study the impact of age on DT and DME protein abundance in small intestinal tissues. In spite of the limitations, we showed that our approach detected significant age-related differences for major, clinically relevant DTs and DMEs. The importance of these changes in relation to intestinal drug absorption remains to be demonstrated through the incorporation of the current dataset into physiology-based pharmacokinetic models. To assess the overall importance of the current findings, it will be critical to simultaneously consider all system parameters that are subject to age (e.g., intestinal surface area, length of the intestinal segments). In itself, the lack of change in the abundance of the majority of DTs and DMEs is already a key finding. It can provide better confidence to physicians to prescribe oral drugs and pharmaceutical companies to develop oral dosing regimens for pediatric patients. Future research should expand the dataset by including more neonates to determine possible changes during the first weeks of extrauterine life.