Improvement of protein expression profile in three-dimensional renal proximal tubular epithelial cell spheroids selected based on OAT1 gene expression: a potential in vitro tool for evaluating human renal proximal tubular toxicity and drug disposition

1: Pharmacokinetics and Non-Clinical Safety Department, Nippon Boehringer Ingelheim Co., Ltd., Kobe, Japan 2: R&D Department, Industrial Division, Nikkiso Co., Ltd., Kanazawa, Japan 3: Faculty of Pharmaceutical Sciences, Institute of Medical, Pharmaceutical and Health Sciences, Kanazawa University; Kanazawa, Japan 4: Center for iPS Cell Research and Application (CiRA), Kyoto University, Kyoto, Japan


Introduction (656 words)
The proximal tubule of the nephron-the primary site for the reabsorption and secretion of endogenous and exogenous substances-has a biotransformation capacity. Numerous enzymes are involved in the metabolism and clearance of endogenous and exogenous compounds in the kidney, including cytochrome P450 (CYP) and non-CYP enzymes, such as uridine-diphosphate-glucuronosyltransferases (UGTs), esterases, glutathione-S-transferases, and sulfotransferases (Bajaj et al., 2018). Transporters expressed on the basal membrane transfer xenobiotics from the systemic circulation to proximal tubular cells. Subsequently xenobiotics are secreted into urine via transporters expressed on the brush border membrane of the proximal tubular cells, or vice versa to be reabsorbed from the filtrate.
lysates were centrifuged at 10,700 × g for 20 min at 4°C, post-nuclear supernatants collected and sedimented twice at 100,000 × g for 60 min at 4°C. The resulting microsomal pellet was suspended in buffer B (0.25 M sucrose, 20 mM Tris-HCl, pH 7.4), layered on top of a 38% (w/v) sucrose solution and ultracentrifuged at 100,000 × g for 40 min at 4°C. The turbid interface fraction was recovered, suspended in buffer B, and sedimented at 100,000 × g for 60 min at 4°C. The plasma membrane fraction was obtained by suspending the pellet in MPEX PTS reagent B (GL Science, Tokyo). Protein concentration of total lysate and plasma membrane fraction was determined using the CBQCA Protein Quantitation kit (Thermo Fisher, Waltham, MA).

Sample preparation for tandem mass tag (TMT)-based proteomics.
Proteins in total lysates and plasma membrane fractions for liquid chromatography-tandem mass spectrometry (LC-MS/MS) proteomics analysis were prepared using the single-pot solid-phase-enhanced sample preparation (SP3) method (Sielaff et al., 2017;Hughes et al., 2019) and labeled with the TMTpro reagent (Thermo Fisher Scientific, Waltham, MA) as previously described (Ohtsu et al., 2022). In brief, 9 μg of samples were dissolved in 100 mM triethylammonium bicarbonate (TEAB) (Thermo Fisher Scientific) and 150 mM NaCl solution with MPEX PTS reagent B (GL Science, Tokyo, Japan), and then reduced with tris-(2-carboxyethyl) phosphine (TCEP) (Thermo Fisher Scientific) followed by alkylation with iodoacetamide (Thermo Fisher Scientific). After protein purification using Sera-Mag SpeedBeads Carboxylate-Modified Magnetic Particles (SP3 beads) (GE Healthcare, Chicago, IL), the samples were digested with trypsin/LysC mix (Promega, Madison, WI). The total concentration of the purified peptides was determined using a Fluorometric Peptide Assay Kit (Thermo Fisher Scientific). Then the peptides were labeled with TMTpro reagent, fractionated (into 12 fractions) with Pierce High pH Reversed-Phase Peptide Fractionation Kit (Thermo This article has not been copyedited and formatted. The final version may differ from this version. DMD Fast Forward. Published on June 29, 2023as DOI: 10.1124 at ASPET Journals on July 6, 2023 dmd.aspetjournals.org Downloaded from 13 LSM 800 software.

Transport experiments
Approximately 200 wells of 3D RPTEC spheroids cultured for 12-14 days were collected in one tube. The collected spheroids were washed with an assay buffer (pH 7.4), consisting Hanks' balanced salt solution (HBSS) supplemented with 15 mM HEPES, by repeating the following cycle twice; 1) addition of the assay buffer, 2) centrifugation at 100 × g for 3 min, and 3) aspiration of the supernatant. The pellet of spheroids was re-suspended in the assay buffer and then equilibrated for 15 min at 37°C. After centrifugation (100 × g, 3 min) and removal of the supernatant, uptake was initiated by adding an assay buffer containing either a transporter cocktail (10 μM metformin, 10 μM furosemide, 10 μM digoxin) or [ 3 H]adefovir (80.6 nM) in the presence or absence of each transporter inhibitor. After the designated incubation time, the spheroids were centrifuged at 500 × g for 1 min. The concentrations of compounds in the assay were determined using the supernatant. Uptake was terminated by removing the supernatant, followed by the addition of ice-cold assay buffer (3 mL), centrifugation at 500 × g for 1 min, and removal of the supernatant. The cells were solubilized in NaOH for approximately 1 h at room temperature. The lysates were neutralized with HCl.
The concentration of the drug in cell lysates was measured either by LC-MS analysis or with a liquid scintillation counter (AccuFLEX LSC-7200, Hitachi, Tokyo) after mixing with a scintillation cocktail (Clear-sol I, Nacalai Tasque, Inc., Kyoto). Protein concentrations were determined using the Lowry method (Lowry et al., 1951) or BCA Protein Assay Kit (Fujifilm-Wako Chemicals, Osaka) with bovine serum albumin as the protein standard.

LC-MS/MS analysis for drugs
Cell lysate (20 μL) was mixed with 60 μL of internal standard solution (acetonitrile with 0.1% This article has not been copyedited and formatted. The final version may differ from this version. DMD Fast Forward. Published on June 29, 2023as DOI: 10.1124 at ASPET Journals on July 6, 2023 dmd.aspetjournals.org Downloaded from formic acid containing 200 nmol/L Hesperetin) followed by centrifuging at 10,000 × g for 5 min. The supernatant (45 μL) was mixed with 10 mmol/L ammonium formate (45 μL). After the sample preparation, an aliquot of the samples (5 μL) was injected into the LC-MS/MS system consisting of an ExionLCTM AD LC system coupled to a QTRAP5500 mass spectrometer (AB SCIEX, Framingham, MA). The samples were separated on Kinetex column, 2.6 µm, F5, 100 Å, 2.1mm I.D.×100 mm (Phenomenex, Torrance, CA) heated to 40°C with a constant flow rate of 0.4 mL/min of mobile phases (A) 5 mmol/L ammonium formate and (B) acetonitrile and a gradient elution program. The program started with 10% eluent B and increased to 95% at 2.6 min followed by a wash step with 95% eluent B for 1 min before equilibration for 1.4 min with 10% eluent B. The total LC run time was 5 min.
Metformin, furosemide, and digoxin were monitored using multiple reaction monitoring (MRM). The ionization conditions and m/z transitions are listed in (Supplemental Table 1). The limits of detection for metformin, furosemide, and digoxin were 1 nmol/L. The mean accuracy of the intra-and inter-run Quality Control samples were within 80 -120% (data not shown).

In vitro ATP assay
3D RPTEC spheroids (on days 10-12) in a V-bottom ultra-low attachment 96-well plate (PrimeSurface ® ; Sumitomo Bakelite, Tokyo) were exposed with respective concentrations of cisplatin and adefovir in REGM supplemented with 0.5% FBS. The exposure medium was replaced on days 3 and 5. Intracellular ATP levels were assessed using CellTiter-Glo ® 3D Cell Viability Assay (Promega, Madison, WI) according to the manufacturer's protocol. Briefly, 100 µL of the reagent was plated into each well and incubated for 30 min at room temperature.
Bioluminescence was measured using a Wallac ARVO MX 1420 luminometer (Perkin Elmer, Waltham, MA). The ATP concentration was determined using an ATP standard curve.

Data analysis
Uptake experiment data analysis Uptake clearance was calculated by normalizing the amount of compound inside the cells to that in the initial solution and the protein concentration in each tube using the following equation: where Uptake CL is the uptake clearance (μL/designated time point/mg), X cell is the amount of compound in the cells (pmol/designated time/tube), and C buffer is the concentration of the compound in the initial solution (pmol/μL). Uptake CL was normalized to the amount of total cellular protein (mg/tube).

Proteomics data analysis
Proteins were identified and quantified using Proteome Discoverer (version 2.4; Thermo Fisher Scientific). Briefly, the MS/MS spectra were searched against UniProtKB/Swiss-Prot human data downloaded in March 2020 using the SEQUEST HT search engine. Two missed cleavages were allowed, along with carbamidomethylation of cysteine as a fixed modification.
Variable modifications included oxidation of methionine, deamidation of asparagine/glutamine, acetylation/methionine-loss/methionine-loss + acetylation of the protein N-terminus and phosphorylation of serine, tyrosine and threonine. Mass tolerance for precursor and fragment ions was 10 ppm, and 0.6 Da, respectively, and false discovery rates at the peptide and protein levels were less than 0.01. Relative protein abundances were calculated from the sum of the peptide ion abundances of unique and razor peptides which are peptides unique to a protein group and shared peptides in a protein group with more identified peptides, respectively. The peptide ion abundances were quantified using TMT reporter ion control the channel average, which was set for a common sample over two-batch experiments.
The calculated protein abundances were used for further analysis.

Statistical analysis
All statistical analyses were performed using GraphPad Prism 7.0 (GraphPad Software, San Diego, CA) or MS-Excel 2010 (Microsoft Corporation, USA). Data were presented as mean ± standard deviation (SD) for at least three independent experiments. Unpaired two-tailed Student's t-test was used to compare datasets from independent groups of cells. Multiple comparisons were performed using one-way analysis of variance followed by Dunnett's test.

Data availability
The Microarray dataset generated in the current study is available from the Gene Expression

OAT1 mRNA expression in 3D RPTEC spheroids
When RPTECs were cultured in spherical cellular aggregates at a seeding cell number of 1,000 cells/well, OAT1 mRNA increased in accordance with the culture period ( Fig. 1A). OAT1 mRNA showed the highest level at a seeding cell number of 1,000 cells/well and decreased at a seeding cell number greater than 1,000 cells/well (Fig. 1B). The size of the spheroid showed seeding cell number dependence; the size at a seeding cell number of 1,000 cells/well was approximately 271±19 µm (n=10; data not shown). Based on these results, a seeding cell number of 1,000 cells/well was chosen for further studies.

mRNA expression in 2D RPTECs and 3D RPTEC spheroids in comparison to human renal cortical tissues
The mRNA expression of the renal transporters, OCT2, OAT1, OAT3, MATE1 and Approximately 80% of the detected ABC transporters when cultured in 2D on days 2 and 12 and in 3D on day 5 showed a difference within 5-fold of the average mRNA expression in the human kidney cortices. When RPTECs were cultured under 3D spheroid conditions for more than five days, the % of ABC transporters showing a difference within 5-fold increased to 90%. In 2D RPTECs, 73% and 25% of the detected SLC transporters on day 2 showed a difference within 5-fold and less than 20% of the average mRNA expression in human kidney cortices, respectively, and these values changed to 83% and 16%, respectively, on day 12. In 3D RPTEC spheroids, 82% and 17% of the detected SLC transporters on day 5 showed a difference within 5-fold and less than 20% of the average mRNA expression in human kidney cortices, respectively, and these values changed to 83% and 16%, respectively on day 12. The mRNA expression patterns of SLC transporters did not change largely beyond day 12. Next, we compared the mRNA expression profiles of 3D RPTEC spheroids from three different donors (days 12-14) with those of human kidney cortices from five different donors. The equivalence between 3D RPTEC spheroids and the human kidney cortex was investigated using GMR and their 90% CIs by applying 2-, 3-, and 5-fold differences (Table 2), because the coefficient of variation × 100 (CV%) of mRNA expression of the majority of the detected genes (16,985) in human kidney cortices was > 20 CV% (Supplemental Table 4). When a 5-fold difference was used as a benchmark, 92% of all genes, 87% of enzymes, 77% of SLC transporters and 44% of ABC transporters were judged to be statistically equivalent to those in the human kidney cortex. The percentage of genes decreased when low-fold differences (2and 3-fold) were applied.

Protein expression in 2D RPTECs and 3D RPTEC spheroids in comparison to human renal cortical tissues
This article has not been copyedited and formatted. The final version may differ from this version. Protein expression in the total lysate (TL) of 2D RPTECs, 3D RPTEC spheroids and human kidney cortices was investigated using label-based LC-MS/MS proteomics to examine whether the results based on mRNA, often used as a surrogate for protein expression, were observed in protein expression. A total of 4,848 proteins were quantified and approximately 90% of the detected proteins showed a 0.2-to 5-fold difference in the average protein expression in human kidney cortices; the same % was observed in the detected 1,642 enzymes and 21 ABC transporters (Table 3). In contrast, 70.5% and 25.9% of the 139 SLC transporters detected in 2D RPTECs showed a difference within 5-fold in average protein expression and less than 20% in the human kidney cortices, respectively. When RPTECs were cultured in 3D spheroid conditions, the % of proteins showing a 0.2-to 5-fold difference improved to 78.4%.
The impact of the 3D spheroidal culture condition, selected by monitoring OAT1 gene expression, was further assessed (Supplemental Tables 2 and 3). When a 5-fold difference was used as the criterion for assessing the similarity between human kidney cortices and RPTECs, a total of 203 proteins showed similar protein expression to that in human kidney cortex, compared to 2D RPTECs. In contrast, 85 proteins showed diversified expression in 3D RPTEC spheroids, compared to 2D RPTECs. The protein expression of more than 20 transporters has been investigated using the total membrane fraction of human kidney cortices from 41 different donors by the LC-MS based quantitative targeted proteomics method (Prasad et al., 2016). However, it was difficult to compare the absolute expression levels between the two studies because TL and a different proteomics method were used in our study.
However, it was possible to evaluate the appropriateness of the human cortical samples used in our study by comparing the variability in the protein expression of transporters between the two studies.  et al., (2016), and showed larger variability, the CV% in our study is considered to be similar if outliers are omitted from the results of Prasad et al., (2016), indicating that the samples used in our study were valid ( Fig. 3B-D).

Protein expression of kidney markers and selected SLC and ABC transporters
RPTECs are renal epithelial cells isolated from human proximal tubules. However, they may dedifferentiate during the culture process, and the original function of proximal tubular cells may not be maintained. The properties of the RPTECs were assessed based on the protein expression of podocalyxin (PODXL) and nephrin 1 (NPHS1) as representative glomerular markers, uromodulin (UMOD) and SLC12A1 as representative distal tubule markers, and aquaporin 1 (AQP1) and cadherin 6 (CDH6) as representative markers of the proximal tubules (Tsujimoto et al., 2020). The expression of glomerular and distal tubule markers was considerably lower in both 2D and 3D RPTEC spheroids than in human renal cortices, which contain both glomeruli and distal tubules. In contract, AQP1 expression in 2D RPTECs was as low as 26% of that in human renal cortices, whereas AQP1 and CDH6 were expressed at the same level in 3D RPTEC spheroids as in human renal cortices (Fig. 3A). This observation, based on the proteomics results, was further confirmed by the mRNA expression of additional markers (Supplemental Fig. 1), although there were some exceptions (such as CUBN for proximal tubules, LCN2 for distal tubules and CD2AP for podocytes).
OAT1 protein expression, used as the selection marker of culture condition, in 2D RTPECs and 3D RPTEC spheroids was 33% and approximately 65% of the average protein expression in the human renal cortices, respectively (Fig. 3B), and the protein expression was almost the same as that of mRNA. As is often the case with the down-regulation of proteins  (Fig. 3B). In the 3D RPTEC spheroids, out of the seven SLC transporters, proteins of GLUT2 and OCTN1 were increased to more than 30% of the average protein expression in the human renal cortices (Fig. 3B). The protein expression of the five selected ABC transporters in 3D RPTEC spheroids ranged from 70% to 306%, compared to the respective average protein expression in human renal cortices and 2D RPTECs, which ranged from 47% to 396% (Fig. 3D). No significant change in protein expression was observed in proteins other than PEPT1, even when the culture period of 3D RPTEC spheroids was extended from 12 days to 20 days ( Fig. 3A-D). A heat map of the protein expression profiles of all quantified enzymes and SLC/ABC transporters is shown in Supplemental Fig. 2. In addition to protein expression in the TL, which represents total protein expression in cells, protein expression in the plasma membrane (PM) fraction, an indicator of functional transporters in the PM, was also investigated in 3D RPTEC spheroids (Fig. 4). The PM/TL ratios of cytosolic proteins were below 0.7, and those of Na + /K + -ATPase and P-gp, which were confirmed to be expressed mainly in the PM of the 3D RPTEC spheroids ( Fig. 8A and B), were 7.54 and 7.09, respectively, indicating that the preparation of the PM fraction was carried out without any remarkable contamination. All ABC transporters and some SLC transporters such as PEPT1, GLUT9, OATP4C1, MATE1, OAT4, OCTN2, and OAT1 showed PM/TL ratios similar to those of Na + /K + -ATPase and P-gp, but far different from those of cytosolic proteins, suggesting that these transporters are mainly expressed in the PM. However, the PM/TL ratios for GLUT2, SGLT1, SGLT2, URAT1, OCTN1, OCT2, OAT3 and MATE2-K ranged from 1.28 to 4.00. SGLT1, SGLT2, URAT1 and OAT3 showed very low protein expression in both the TL (Fig. 3) and PM, indicating that these transporters were not abundantly expressed in the 3D RPTEC spheroids. GLUT2 and OCT2 protein expression in the PM was low, although that in the TL was high, indicating that GLUT2 and OCT2 proteins are expressed not only in the PM but also in other parts of the 3D RPTEC spheroids.

Effect of cisplatin and adefovir on intracellular ATP levels of 3D RPTECs
Cisplatin and adefovir are substrates of OCT2/MATE1 and OAT1, respectively (Jongh et al., 2004;Izzedine et al., 2005). We investigated whether cisplatin and adefovir showed difference in toxicity between 2D RPTECs and 3D RPTEC spheroids and exhibited transporter-dependent toxicity in 3D RPTEC spheroids using intracellular ATP levels as an endpoint; because the dose-response to cefalotin and cisplatin on intracellular ATP levels showed a very similar response to those on lactate dehydrogenase release, which is often used to evaluate cell viability (Supplemental Fig. 3). Cisplatin showed a similar dose-response to intracellular ATP levels in 2D RPTECs and 3D RPTEC spheroids (Fig. 5), whereas the in vitro toxicity of adefovir was weaker in 2D RPTECs than in 3D RPTEC spheroids (Fig. 5).
Cisplatin showed a concentration-dependent decrease of intracellular ATP levels and the addition of 1,000 µM cimetidine-an OCT and MATEs inhibitor-significantly suppressed the decrease in ATP compared to that in the absence of cimetidine, while 1,000 µM probenecid-an OAT inhibitor-did not cause a change (Fig. 6A). Adefovir reduced the intracellular ATP levels in a concentration-dependent manner. The decrease of intracellular ATP levels was not weakened by the addition of 1,000 µM cimetidine, while the addition of 1,000 µM probenecid almost completely diminished the adefovir-dependent decrease of intracellular ATP levels (Fig. 6B).

Transport function in 3D RPTEC spheroids using metformin, furosemide, adefovir and digoxin as probe substrates
This article has not been copyedited and formatted. The final version may differ from this version. The transporter function in the 3D RPTEC spheroids was evaluated using the following substrates: metformin for OCT2/MATEs, furosemide and adefovir for OAT1 and digoxin for P-gp (Uwai et al., 2007;Ebner et al., 2015) (Fig. 7). The uptake of metformin by 3D RPTEC spheroids was decreased by the OCT/MATEs inhibitor cimetidine, but only very weakly decreased by the OAT inhibitor probenecid. Probenecid did not significantly decrease the uptake of furosemide or adefovir, nor did it increase the uptake of digoxin. The P-gp inhibitor zosuquidar did not show a statistically significant increase in digoxin uptake; whereas cimetidine significantly increased digoxin uptake.

P-gp and Na + /K + ATPase localization in 3D RPTEC spheroids
P-gp and Na + /K + -ATPase are expressed on the apical and basolateral membrane sides of proximal renal tubular cells, respectively (Rostgaard and Møller, 1980;Launay-Vacher et al., 2006). The structure of 3D RPTEC spheroids was investigated using confocal laser scanning microscopy with the respective antibodies. One 3D RPTEC spheroid was approximately 200-300 µm in size, and the apical membrane marker P-gp was mainly detected on the outside of the 3D RPTEC spheroids and in the non-cell-to-cell contact area ( Fig. 8A; green). P-gp was not highly expressed at the center of 3D RPTEC spheroids. In contrast, Na + /K + -ATPase, a marker of basolateral membranes, was mainly expressed inside 3D RPTEC spheroids and in the area where the cells were in contact with each other (Fig. 8B; red).

Discussion
This article has not been copyedited and formatted. The final version may differ from this version. When protein expression was measured as an endpoint, 90% of the detected 4,848 proteins in 3D RPTEC spheroids were within the 5-fold range of protein expression in the human renal cortices (Table 3). For SLC transporters, the percentage of transporters that were not within the 5-fold range of human renal cortices was approximately 30% in the 2D RPTECs, which was approximately 2-fold higher than that of enzyme and ABC transporters ( post-promoter region, rather than of the promoter region, occurs in the 3D RPTEC spheroids.
It is also possible that factors other than HNF1 and methylation are involved in the regulation of OAT1/3 expression. Further investigations are necessary to better understand the regulation of OAT1/3 expression in RPTECs.
Cisplatin decreases glomerular filtration rate and increases urinary albumin at clinical concentrations of approximately 10 µM unbound cisplatin (Erdlenbruch et al., 2001).
Cisplatin showed a concentration-dependent decrease in intracellular ATP levels in 3D RPTEC spheroids, and a 50% reduction of intracellular ATP was obtained at 10 µM which is the clinical plasma-free cisplatin concentration (Fig. 6A). Intracellular ATP levels in HK-2 cells, which do not express OCT2 or MATE (Sancho-Martínez et al., 2011;Hauwaert et al., 2014), were reduced to about half at 30-100 µM cisplatin. To understand why almost the same concentration decreased intracellular ATP levels by half in HK-2 and 3D RPTEC spheroids although the expression of OCT/MATE transporters was different, two points should be taken into consideration. First, the contribution of passive diffusion of basic drugs is larger than that of acidic drugs because passive diffusion is affected by the internal negative membrane potential. Secondly, longer exposure times are generally used in in vitro toxicity studies than in in vitro transporter studies. The intracellular concentration of cisplatin is unlikely to differ greatly between 3D RPTEC spheroids and HK-2 cells, because the impact of the OCT2/MATE process is not large in in vitro toxicity studies.
Adefovir (120 mg/day) has been reported to cause nephrotoxicity in 20-30% of patients when administered at a clinic for 48 weeks (Kahn et al., 1999;Fisher et al., 2001).
The intracellular ATP level in 3D RPTEC spheroids was reduced by adefovir in a concentration-dependent manner, with the amount of intracellular ATP reduced to about half of the control at concentrations of 100-1,000 µM (Fig. 6B). The addition of the OAT inhibitor probenecid completely suppressed the decrease in cellular ATP, suggesting that adefovir toxicity observed in 3D RPTEC spheroids is a function of the OAT transporter. The in vitro cytotoxicity of adefovir was also evaluated with RPTECs in which OAT1 was transfected using cell viability as an endpoint and the obtained IC 50 value was 230 µM (Nieskens et al., 2016). The clinical unbound C max of adefovir was approximately 1 µM after 125 mg/day (Barditch-Crovo et al., 1997) and nephrotoxicity was seen in 20-30% of patients. The discrepancy may be explained by exposure period with adefovir is not sufficient because the clinical toxicity of adefovir has been observed 48-72 weeks post-dose in clinics, and the in vitro toxicity of cisplatin is reported to be enhanced 100-fold by increasing the duration of drug exposure from 24 h to 14 days (Secker et al., 2019). This study revealed that 3D RPTEC spheroids maintained their morphology over 5-20 days and the gene expression did not significantly change from 12 to 32 days (Fig. 2), thus the toxic effects of long-term exposure can also be examined in 3D RPTEC spheroids.

In addition to the OCT/MATE-dependent in vitro cytotoxicity of cisplatin in 3D
RPTEC spheroids (Fig. 6), cimetidine reduced the uptake of metformin into 3D RPTEC (Fig.   7). This was consistent with the proteomics results showing that the expression levels of OCT2, MATE1, and MATE-2K in 3D RPTEC spheroids were within the range of variation in their expression levels in human renal cortices, and their expression was confirmed in the PM fraction (Figs. 3B and 4A). In contrast, no uptake of adefovir or furosemide by OAT1 was observed, although OAT1 protein expression was within the range of variation in protein expression in the human renal cortices, was observed in PM fractions and the in vitro cytotoxicity of adefovir in 3D RPTEC spheroids was OAT-dependent. Immunostaining revealed that P-gp, a marker of apical membranes, was strongly expressed on the outer side of the 3D RPTEC spheroids and in areas where the RPTECs were not in contact with each other.
However, P-gp expression decreased toward the center of the 3D RPTEC spheroids. Na + /K + -ATPase, a basolateral membrane marker, was strongly expressed in areas where RPTECs This article has not been copyedited and formatted. The final version may differ from this version. were in contact with each other, and its expression in the center of the 3D RPTEC spheroids was not as weak as that of P-gp (Fig. 8). The following two possibilities were considered as reasons why basolateral OAT1 activity was not detected in 3D RPTEC spheroids: 1) apical and basal transporter activities cannot be evaluated separately because both apical transporters are exposed to the buffer as well as in vitro kidney slice systems (Arakawa et al., 2017(Arakawa et al., , 2019, 2) a wash process in the in vitro transporter assay is not sufficient for transporters on the basolateral side because transporters on the basolateral side of RPTECs are expressed in the area where cells are in contact. Genetic polymorphisms can weaken the correlation between activity and protein expression. However, for OAT1, although SNP with a complete loss of function (R454Q) has been reported, the allele frequency is extremely low (0.2%) (Fujita et al., 2004). Therefore, it is unlikely that the SNPs in OAT1 caused a discrepancy between OAT1 protein expression and the undetected OAT activity in the 3D RPTEC spheroids, although unidentified SNPs may be involved. Further efforts are necessary to precisely evaluate transporter activity in the future.
Reproducibility is an important factor in evaluating the usefulness of an in vitro system. The reproducibility of the 3D RPTEC spheroids was evaluated by comparing the CV% of mRNA expression in 3D RPTEC spheroids from three different lots of RPTECs at a culture period of 12-14 days. Furthermore these CV% values were compared with those of human kidney cortex from five different donors (Supplemental Table 4). Of the 16,993 genes detected, 87% showed variation within 50 (CV%), which was also true for 3,193 enzymes, 334 SLC transporters, and 39 ABC transporters in the case of 3D RPTEC spheroids, whereas CV% for all genes, enzymes, SLC transporters, and ABC transporters in the human kidney cortex were 75%, 80%, 61%, and 80%, respectively. This indicates that the 3D culture method used in this study is a highly reproducible culture condition with a small variation in mRNA expression among the three different lots of RPTECs (from three different donors) This article has not been copyedited and formatted. The final version may differ from this version. compared with the variation in mRNA expression in the human kidney cortex from five different donors.
In summary, we have confirmed that a spheroidal culture condition, selected by monitoring OAT1 gene expression, improved the protein expression of other genes to be more similar to that in human kidney cortices, resulting in a similar protein expression level of 90% of more than 4,800 proteins detected over an extended period of up to 20 days in culture. This condition also enabled the evaluation of not only transporter-dependent cisplatin nephrotoxicity, which has been evaluated in 2D culture conditions and other cell systems, but also transporter-dependent adefovir nephrotoxicity more sensitively. In vitro 3D RPTEC spheroids could positively impact preclinical drug discovery/drug development, helping to prevent unwanted failures in late-stage drug development including clinical trials. However, the polarized expression of transporters has not yet been fully evaluated in the current system and genetic polymorphisms may interfere the comparison of mRNA and protein expression between RPTECs and human kidney cortices. Moreover, the current system is a static culture, and the inclusion of laminar flow on the RPTEC could further enhance tubular epithelial cell function.

Acknowledgments
The excellent technical assistance of Kaori Makino, Tokuko Takatsuka

References
This article has not been copyedited and formatted. The final version may differ from this version.

Figure 3 Protein expression of kidney markers (A), SLC transporters (B and C), and ABC transporters (D).
Protein abundance was expressed as % of the human kidney cortex 1. The protein abundance of human kidney cortices represents the mean ± SD of five different donors. Human cortex, human kidney cortex.

Figure 5 Effect of cisplatin (A) and adefovir (B) on cellular ATP contents in 2D
RPTECs and 3D RPTEC spheroids.
This article has not been copyedited and formatted. The final version may differ from this version.   This article has not been copyedited and formatted. The final version may differ from this version. Digoxin uptake (µL/10 min/mg)