Abstract
The default assumption during in vitro in vivo extrapolation (IVIVE) to predict metabolic clearance in physiologically based pharmacokinetics (PBPK) is that protein expression and activity have the same relationship in various tissues. This assumption is examined for uridine 5′-diphosphate glucuronosyltransferases (UGTs), a case example where expression and hence metabolic activity are distributed across various tissues. Our literature analysis presents overwhelming evidence of a greater UGT activity per unit of enzyme (higher kcat) in kidney and intestinal tissues relative to liver (greater than 200-fold for UGT2B7). This analysis is based on application of abundance values reported using similar proteomic techniques and within the same laboratory. Our findings call into question the practice of assuming similar kcat during IVIVE estimations as part of PBPK and call for a systematic assessment of the kcat of various enzymes across different organs. The analysis focused on compiling data for probe substrates that were common for two or more of the studied tissues to allow for reliable comparison of cross-tissue enzyme kinetics; this meant that UGT enzymes included in the study were limited to UGT1A1, 1A3, 1A6, 1A9, and 2B7. Significantly, UGT1A9 (n = 24) and the liver (n = 27) were each found to account for around half of the total dataset; these were found to correlate with hepatic UGT1A9 data found in 15 of the studies, highlighting the need for more research into extrahepatic tissues and other UGT isoforms.
SIGNIFICANCE STATEMENT During physiologically based pharmacokinetic modeling (in vitro in vivo extrapolation) of drug clearance, the default assumption is that the activity per unit of enzyme is the same in all tissues. The analysis provides preliminary evidence that this may not be the case and that renal and intestinal tissues may have almost 250-fold greater uridine 5′-diphosphate glucuronosyltransferase activity per unit of enzyme than liver tissues.
Introduction
Applications of physiologically based pharmacokinetics (PBPK) over the last 20 years have increased exponentially compared with the rest of pharmacokinetics (El‐Khateeb et al., 2021). This has been linked to the ability of PBPK models to extrapolate kinetics beyond the average patient by using fundamental aspects of the biology related to the change of the expression in enzymes between healthy individuals and various patient groups (Howard et al., 2018).
The default assumption during the in vitro in vivo extrapolation (IVIVE) steps of metabolic information for drug clearance during PBPK is that expression mirrors activity regardless of the location of the enzyme. In other words, the activity per unit of enzyme (kcat) is considered to be the same in various tissues. We wished to examine this common assumption for the case example of uridine 5′-diphosphate glucuronosyltransferase (UGT) enzymes. These enzymes are involved in phase II biotransformation of many drugs, and they are currently the second most common route for primary drug metabolism, responsible for the metabolic clearance of 10%–30% of all drugs (Stingl et al., 2014). This proportion is set to increase, as pharmaceutical companies are intentionally designing new drug candidates that go through non–cytochrome P450 (CYP450) pathways to reduce the burden of CYP450-related drug-drug interactions (DDIs) (Achour et al., 2014).
Previous research has identified the liver as the epicenter of xenobiotic metabolic processes, containing the most diverse and abundant population of drug-metabolizing enzymes (Achour et al., 2014). However, some may argue that the contributions of other key metabolic tissues involved in drug disposition have been neglected or underestimated. Studies involving extrahepatic metabolism are very limited compared with hepatic metabolism (Scotcher et al., 2016), and to build a clinically realistic model of the human body, the involvement of enzyme kinetics across extrahepatic tissues must be quantified. This is as true of UGTs as of other enzymes. UGT enzymes quantified in the liver do not have a complete set of corresponding expression values in renal and intestinal tissues (Achour et al., 2014; Couto et al., 2020; Al‐Majdoub et al., 2021). This highlights the need to generate a reliable dataset for absolute enzyme abundances across the key metabolic organs as a starting point for quantifying tissue-specific enzyme kinetics.
To begin quantifying UGT enzyme kinetics per unit of enzyme, absolute abundance for individual UGT isoforms must be determined as amount of enzyme per milligram of microsomal protein (Crewe et al., 2011). Vmax, measurable as amount of isoform-specific probe substrate converted to its metabolite per unit time, must also be determined. The enzyme abundance-activity relationship can then be quantified as kcat, defining the differences in intrinsic activity per unit of UGT. To accurately reflect tissue-specific kinetics, kcat must account for tissue-specific enzyme abundances (Robinson, 2015). The common assumption that kcat is the same across various tissues has not been examined for UGTs, and information on other enzymes is sparse (von Richter et al., 2004; Yang et al., 2004; Galetin and Houston, 2006).
Obviously, the accuracy of clearance predictions will also depend on correct assignment of the extrahepatic metabolism. However, this is not the subject of current exercise. Nonetheless, we make extensive use of the research into the abundance (Achour et al., 2014) and activity of UGT in intestine and kidney even though abundance values are missing for several UGT enzymes (Couto et al., 2020; Al‐Majdoub et al., 2021).
Materials and Methods
Collection of Data
Two electronic databases, Web of Science (https://wok.mimas.ac.uk) and PubMed (https://www.ncbi.nlm.nih.gov/pubmed/), were searched for relevant literature from the years 2000 to 2019 using appropriate keywords (UDP-glucuronosyltransferase, UGT activity). Both UGT abundance and activity studies were searched for glucuronidation data; this involved searching for other key terms in place of ‘activity’ to widen the search scope (glucuronidation, kcat, metabolism, abundance, concentration, content, quantification, measurement, LC-MS, ELISA, Western blotting). Citation lists within the collected studies were also inspected to identify any further relevant literature. Searches were species and tissue-specific for human intestinal and kidney microsomes; keywords included synonyms for these tissues (gut, renal). The search criteria were repeated for human liver microsomes, focusing on literature using the probe substrates identified in renal and intestinal studies, to compile a database of comparable data. All but one publication included data from ‘adult’ populations; hence only data generated from adult tissue samples were included for analysis.
Calculation of Enzyme Activity
The kcat values for individual UGT isoforms were calculated using eq. 1, where Vmax represents the maximal metabolic capacity in pmol/min/mg microsomal protein and UGT abundance is tissue-specific for individual isoforms in pmol UGT/mg protein:
Where Vmax was not specified for activity, Km values (substrate concentration at 1/2 Vmax) were identified for the probe substrates; if the probe concentration (μM) was found to be significantly above the maximum Km value (μM) (i.e., >2-fold), the assumption was made that the reaction was conducted at Vmax and these data were used to calculate kcat. On the other hand, if the probe concentration (μM) was found to be significantly below the minimum Km value (μM) (i.e., <0.5-fold), the assumption was made that the activity value was within the intrinsic clearance range and the clearance was used as a supplementary measurement of enzyme activity. Here, the literature was examined to identify reported probe Km values, and where this information was not available, reference Km values were found (Seo et al., 2014; Miners et al., 2021). We have to acknowledge that our assumption will result in significant errors in the calculation of these parameters depending on how far the substrate concentrations deviate from those related to initial rate [in which case, UGT intrinisc clearance (CLint,UGT) = Vmax/(Km)] and Vmax. However, the error introduced by this approach will be less than that associated with the comparison of kinetic data from studies that used vastly different experimental conditions. Where necessary, intrinsic clearance data were corrected for the microsomal fraction of unbound drug to give unbound intrinsic clearance (CLint,u) a closer estimate for in vivo clearance (Hallifax and Houston, 2006; Gao et al., 2008). Once the corrected clearance values had been determined, CLint,u (μl/min/mg microsomal protein) values were divided by the abundance (pmol/mg protein) and probe concentrations (μM) to give CLint,u per unit enzyme (μl/min/pmol enzyme).
Results
Filtering Data
A total of 19 studies were used in this analysis; 15 of these were relevant for calculating kcat (Soars et al., 2001, 2003; Miles et al., 2005; Picard et al., 2005; Al-Jahdari et al., 2006; Shimizu et al., 2007; Benoit-Biancamano et al., 2009; Rowland et al., 2008; Komura and Iwaki, 2011; Liang et al., 2011; Walsky et al., 2012; Gill et al., 2013; Knights et al., 2016; Achour et al., 2017, 2018; Chen et al., 2018), whereas the remaining four were used for calculating CLint,u (Cubitt et al., 2009; Gill et al., 2012; Scotcher et al., 2017; Bhatt et al., 2019). Of the total dataset, 29% of the data did not meet the search criteria and were excluded on account of the following: no probe specificity, no availability of abundance data, or not falling into the Vmax or intrinsic clearance range. The data useful in this analysis are summarized in Table 1. Probes were selected based on availability of data across two or (preferably) all of the studied tissues; because data were very limited, it was necessary to focus on availability rather than specificity of probes. Data were available for UGT1A1, 1A6, 1A9, and 2B7 for kcat calculations. Only UGT1A6, with probe substrate deferiprone, was comparable across all three tissues (Benoit-Biancamano et al., 2009; Knights et al., 2016); UGT1A1 was comparable across the liver and intestine (Komura and Iwaki, 2011); and data for UGT1A9 and 2B7 were comparable across the liver and kidney (Miles et al., 2005; Komura and Iwaki, 2011; Knights et al., 2016). Intrinsic clearance calculations allowed for comparison of UGT1A1 and 1A3 across all three tissues using probe substrates ezetimibe and telmisartan, respectively (Gill et al., 2012); UGT1A9 data were comparable across the liver and kidney (Gill et al., 2012; Scotcher, et al., 2017; Bhatt et al., 2019); and data for UGT2B7 were comparable across the liver and intestine (Gill et al., 2012).
Summary of the activity data available for analysis (kcat and CLint,u data combined) with a sum total of data for each probe-specific UGT enzyme and a sum total of data for each tissue
Reference Abundance Values
The collated reference abundance values (Achour et al., 2014; Couto et al., 2020; Al‐Majdoub et al., 2021), presented in Table 2, were used to perform kcat and CLint,u calculations, where activity data did not have corresponding abundance values presented in the literature (Soars et al., 2001, 2003; Miles et al., 2005; Picard et al., 2005; Al-Jahdari et al., 2006; Shimizu et al., 2007; Benoit-Biancamano et al., 2009; Rowland et al., 2008; Cubitt et al., 2009; Komura and Iwaki, 2011; Liang et al., 2011; Gill et al., 2012, 2013; Walsky et al., 2012; Scotcher et al., 2017; Chen et al., 2018).
Mean abundance values from the literature for UGT isoforms across human liver, intestine, and kidney microsomes (pmol/mg protein)
UGT isoforms included in the analysis were based on availability of common probes and activity data (at Vmax) across two or more of the tissues (UGT1A1, 1A3, 1A6, 1A9, and 2B7); however, a wider range of enzyme expressions are demonstrated in Table 2 to highlight data availability across tissues.
Correlation of UGT Expression and Activity between Different Tissues
The reported activity values for human liver, intestinal, and kidney microsomes used to calculate kcat and CLint,u are demonstrated in Tables 3 and 4, respectively. Some assumptions for Vmax were made for data used for calculating mean kcat (Table 3), where Vmax was not specified but probe concentration was found to be more than 2-fold greater than the substrate Km (Knights et al., 2016). Similarly, for CLint,u calculations, where activity data were not specified as Vmax, it was assumed to be in the intrinsic clearance range if probe concentration was found to be less than 0.5-fold of the substrate concentration (Bhatt et al., 2019). Where activity was assumed to be at Vmax or intrinsic clearance, probe substrate concentration and Km have also been recorded. Probe concentration was recorded across all CLint,u enzymes to calculate CLint,u per unit enzyme (see Materials and Methods). Key experimental differences that may have influenced the activities seen across the studies are recorded in Table 5.
Vmax (pmol/min/mg microsomal protein) and calculated kcat (pmol/min/pmol enzyme) data for UGT enzymes using probe substrates across human liver, intestine, and kidney tissues
CLint,u (μl/min/mg microsomal protein) and calculated CLint,u per unit enzyme (μl/min/pmol enzyme) data for UGT enzymes using probe substrates across human liver, intestine, and kidney tissues
Interlaboratory differences between the literatures used for Vmax and intrinsic clearance data for kcat and CLint,u calculations, including concentrations of UDPGA, alamethicin, MgCl2, and % of BSA
To demonstrate the differences in the mean relative expressions and activities of UGT enzymes across the tissues, scatter graphs were generated with the y-axis in logarithmic scale (log10), demonstrating the ratio of fold difference of intestinal and renal abundances and activities relative to the liver (Figs. 1 and 2). Data for enzyme activities were segregated for kcat and CLint,u (Fig. 2, A and B). The reference line for the liver crosses the y-axis horizontally at 1; values above or below the line represent greater or fewer enzyme expression, respectively, found in the intestine and kidneys than that found in the liver.
Comparison of abundance values for UGT enzymes in the gut and kidney as a ratio to the liver. The liver has been used as a reference point, crossing the y-axis horizontally at 1. Plotting the y-axis in logarithmic scale (log10) demonstrates the fold difference in abundance for intestinal and renal tissues with respect to the liver (specific fold values labeled on the graph).
Comparison of activity values (kcat) (A) and intrinsic clearance (CLint,u) (B) for UGT enzymes in the gut and kidney as a ratio to the liver. The liver has been used as a reference point, crossing the y-axis horizontally at 1. Plotting the y-axis in logarithmic scale (log10) demonstrates the fold difference in abundance for intestinal and renal tissues with respect to the liver (fold values labeled specifically on the graph). Error bars are included where there is sufficient data (i.e., ≥2 data sets).
Discussion
This analysis uniquely reviews a comprehensive list of all significant UGT activity studies conducted using comparable probe substrates for human liver, intestinal, and kidney microsomal tissues from 2000 to 2019. Nevertheless, it only provides a preliminary database of mean activity values, as kcat and CLint,u, for comparable UGT isoforms for the three major metabolic organs since the functional assays were not conducted in the same laboratory nor under exactly similar conditions. We were able to map a ratio of fold difference for the intestine and kidney relative to liver for UGT enzyme abundance but more importantly activity per unit of enzyme. The intestinal and kidney abundance data, which we used for calculating activity per unit of enzyme, were all taken from a single laboratory (University of Manchester; Table 2). A meta-analysis conducted by Achour et al. (2014) presented weighted average abundance values for liver from data published between 1980 and 2014 that were measured using liquid chromatography–mass spectrometry (LC-MS) proteomics and that were used for calculating kcat values in liver. Mean renal and intestinal (from kidney cortex) abundance data were taken from recent studies (Couto et al., 2020; Al‐Majdoub et al., 2021), respectively, using the same laboratory environments, LC-MS technology, and consistent assay conditions. Consistency in LC-MS–based quantification across the three studies with abundance reported in the microsomal fraction (pmol/mg microsomal protein) provides a reliable source of reference abundance values. This degree of consistency in abundance values used for the calculation of cross-tissue enzyme kinetics has not been accomplished in any other UGT activity study; it also provides a reference point for cross-tissue UGT enzyme expressions for future research.
Our major objective in this analysis was to explore the assumption of similarity in kcat for UGT enzymes across the liver, intestine, and kidneys, as a case example. Because there is limited research into UGT functional assays involving full kinetics (at different concentrations), we could not calculate kcat in many cases and used CLint,u per unit enzyme instead (measured at low concentrations of the probe). We focused on data for probes that were common for functional assays conducted in at least two of the tissues. Accordingly, the analysis was limited to UGT1A1, 1A3, 1A6, 1A9, and 2B7. Among these, UGT1A9 (n = 24) and the liver (n = 27) comprised almost half of the data for the analysis (Soars et al., 2001, 2003; Miles et al., 2005; Picard et al., 2005; Al-Jahdari et al., 2006; Shimizu et al., 2007; Rowland et al., 2008; Komura and Iwaki, 2011; Liang et al., 2011; Gill et al., 2012, 2013; Walsky et al., 2012; Achour et al., 2018; Chen et al., 2018; Bhatt et al., 2019), emphasizing the paucity of functional assays conducted in extrahepatic tissues for majority of UGT isoforms.
Although UGT abundances are unsurprisingly greater within the liver across all measured isoforms (Fig. 1), activity per unit of enzyme appears to be lower in the liver than in the other tissues. There was an overall trend when UGT functional activities were available across all three tissues [UGT1A6 (kcat), 1A1, 1A3, and 2B7 (CLint,u)]. These results therefore suggest that the relative contribution of drug metabolism by liver may have been assigned incorrectly for UGT substrates. This is when PBPK models assume the same metabolic clearance by UGT per unit of enzyme in various tissues. A renal kcat of more than 200-fold greater than the liver (e.g., UGT2B7) can compensate greatly for a 300-fold lower abundance relative to liver (e.g., UGT1A1).
Contribution of any enzyme to overall kinetics also depends on other factors such as the blood flow to the organ and the topological arrangements related to the physiology and anatomy (Nishimura et al., 2007; Pang et al., 2019). In addition, enzyme-specific cofactors [e.g., UDP-glucuronic acid (UDPGA) is a glucuronic donor in glucuronidation reactions] are critical. Lack of UDPGA, which needs to be at least 5 mM for optimum glucuronidation activity, and simplification of kinetic analyses leads to loss of activity in all UGT isoforms. Although specific probe substrates have in fact been identified for UGT enzymes (Miners et al., 2021), the analysis in the current study was limited by the lack of enzyme activity data measured using these specific probe substrates—a common research gap for non-CYP450 enzymes (Argikar et al., 2016). Hence, the data in this study were limited by the nonspecificity of some of the probes used for measuring activity (i.e., ezetimibe and naloxone), as data were selected based on availability of comparable probes across the studied tissues. Nevertheless, most of the isoforms could be assessed with confidence using specific probe data: UGT1A3, 1A6, 1A9, and 2B7 using telmisartan, deferiprone, propofol, mycophenolic acid (MPA), and zidovudine (AZT), respectively (Miners et al., 2021). For the remaining UGT enzymes, there is a clear research gap. For instance, although UGT1A9 activity values were available for hepatic, intestinal, and renal tissues as both Vmax and CLint,u, there was a lack of intestinal abundance data and kcat and CLint,u (per unit enzyme) could not be calculated across all organs (Picard et al., 2005; Shimizu et al., 2007; Komura and Iwaki, 2011; Gill et al., 2012, 2013).
Consistency between the protocols measuring the functional activity is another issue that hampers the robust assessment of the kcat across the tissues based on literature data (Table 5). The requirement for a minimum concentration of UDPGA of 5mM (Miners et al., 2021) was met in only half of the studies (Soars et al., 2001, 2003; Rowland et al., 2008; Cubitt et al., 2009; Liang et al., 2011; Gill et al., 2012, 2013; Walsky et al., 2012; Scotcher et al., 2017; Achour et al., 2018). Factors known to affect Vmax include the concentration of alamethicin in the incubation buffer and the time of preincubation of human tissue microsomes with alamethicin, concentration of magnesium chloride (MgCl2), and choice of organic solvent used for aglycone solubilization in the incubation medium (Miners et al., 2021). Thus, differences in these assay conditions across the studies (Table 5) were expected to have an impact on calculated kcat (Table 3). Bovine serum albumin (BSA) binds free fatty acids that inhibit UGT activity and is therefore frequently included as a component of incubation buffers. Its presence usually results in an increase in measured intrinsic clearance (fraction unbound) for UGT1A9 and UGT2B7 substrates as a result of reduction in the Km (Wu et al., 2013). However, BSA levels were inconsistent across the protocols for assessing functional activity of UGT (Table 5), which may have reduced accuracy in calculated CLint,u (Table 4). Badée et al. (2019) showed that mean CLint,u values are dependent on the nature and concentration of the buffer, with reduced buffer concentration seen to reduce the rate of glucuronidation. Table 5 shows the buffers used in studies cited here. Despite these interlaboratory differences, the trends seen in this analysis are consistent in showing greater glucuronidation activities across the intestinal and kidney tissues than the liver (Fig. 2, A and B).
The abundance measurements suffered much less from these concerns. We were able to use proteomic measurements generated in a single laboratory in this study, so the general lack of interlaboratory consistency in quantifying intestinal and renal UGT enzymes did not apply in the present study (Couto et al., 2020; Al-Majdoub et al., 2021). Moreover, hepatic data taken from the meta-analysis conducted by Achour et al. (2014) uses literature measuring abundance with LC-MS proteomic technology, much like the intestinal and renal studies, maintaining consistency in the standards used for measuring abundance. Going forward, it is imperative to develop and use standardized experimental conditions for future UGT enzyme kinetic research to generate reliable cross-study comparisons.
In conclusion, this preliminary analysis provides a starting point for building tissue-specific IVIVE data. The methods are generalizable to other enzymes involved in drug metabolism. These are important for continuous improvements to PBPK simulations in the development of new drugs. As this case study for UGTs illustrates, accurate estimates of functional enzyme kinetics in various tissues are still limited. It would be desirable to conduct functional assays on the same samples as proteomic measurements to confirm the preliminary findings presented here. Nevertheless, our results suggest that kcat may vary from tissue to tissue, perhaps even within similar tissues depending upon disease state.
Authorship Contributions
Participated in research design: Ahmed, Rostami-Hodjegan, Al-Majdoub.
Performed data analysis: Ahmed.
Wrote or contributed to the writing of the manuscript: Ahmed, Rostami-Hodjegan, Barber, Al-Majdoub.
Footnotes
- Received December 12, 2021.
- Accepted May 3, 2022.
This work received no external funding.
No author has an actual or perceived conflict of interest with the contents of this article.
Abbreviations
- AZT
- zidovudine
- BSA
- bovine serum albumin
- CLint,u
- unbound intrinsic clearance
- CYP450
- cytochrome P450
- IVIVE
- in vitro in vivo extrapolation
- kcat
- activity per unit of enzyme
- Km
- substrate concentration at 1/2 Vmax
- LC-MS
- liquid chromatography–mass spectrometry
- MgCl2
- magnesium chloride
- MPA
- mycophenolic acid
- PBPK
- physiologically based pharmacokinetics
- UDPGA
- UDP-glucuronic acid
- UGT
- uridine 5′-diphosphate glucuronosyltransferase
- Copyright © 2022 by The American Society for Pharmacology and Experimental Therapeutics
