Illuminating the function of the orphan transporter, SLC22A10, in humans and other primates

Human SLC22A10 showed no expression on the plasma membrane and no transporter activity of prototypical anionic substrates of SLC22 family members

Human SLC22A10 is in a cluster that includes known organic anion transporters: SLC22A24 is closest, followed by SLC22A9, SLC22A11 and SLC22A12 (Fig. 1A). Phylogenetic analyses reveals that its closest homolog is SLC22A24. The substrates of SLC22A24 are steroid conjugates, bile acids and dicarboxylic acids, which our laboratory has successfully deorphaned³. Overexpression of human SLC22A10 tagged with GFP in the N-terminal resulted in no detection of a GFP-tagged protein on the plasma membrane (Fig. 1B). Furthermore, no uptake of prototypical organic anions was observed in cells expressing SLC22A10 whereas significant uptake was observed in cells expressing known SLC22 organic anion transporters including SLC22A6, SLC22A8 and SLC22A24 (Fig. 1C).

The long isoform of chimpanzee and gorilla SLC22A10 was expressed on the plasma membrane whereas the short isoform was not

The organic anion transporters depicted in Fig. 1A consist of 536 to 563 amino acid proteins. Predictions from reliable sources such as Uniprot, Ensembl, and NCBI Nucleotide databases confirmed that the human SLC22A10 gene produces a 541 amino acid isoform. Conversely, according to reports from Ensembl and UniProt, the orthologs of SLC22A10 found in great apes are predicted to have two isoforms: a short isoform comprising 540 amino acids and a longer isoform containing 552 amino acids. Because there was no detectable expression of the human SLC22A10 on the plasma membrane of cells recombinantly expressing the transporter (Fig. 1B), we inquired whether the direct species orthologs in great apes exhibited a similar lack of plasma membrane expression when expressed recombinantly in cells. In fact, we observed that the long isoforms (552 amino acids) of both chimpanzee and gorilla SLC22A10 were detected on the plasma membrane (Fig. 2A). The shorter isoforms (540 amino acids) of chimpanzee, bonobo and gorilla SLC22A10 showed a similar lack of plasma membrane localization as the human ortholog, which consists of 541 amino acids (Figs. 1A and 2A).

Chimpanzee and gorilla SLC22A10 expressing the long isoform transport estradiol glucuronide but not other anions that are canonical substrates of members in the SLC22A family

Because the long forms of the great ape SLC22A10 showed a plasma membrane localization, we attempted to identify substrates of SLC22A10 using isotopic uptake assays in cells recombinantly expressing the long isoforms of the great ape transporters. Typical anions that are canonical substrates of members in the SLC22A family were screened for accumulation in human, chimpanzee, bonobo and gorilla expressing the long as well as the short isoforms. Significant accumulation of [³H]-estradiol-17β-glucuronide and [³H]-androstanediol-3α-glucuronide were observed in cells expressing chimpanzee and gorilla SLC22A10 encoding the long but not the short isoforms (Fig. 2B, Supplementary Fig. 1A). No significant uptake was detected for other anions that are canonical substrates of members of the SLC22A family, such as estrone sulfate, taurocholic acid, cGMP, uric acid and succinic acid (Supplementary Fig. 1A–F). However, there was a small but significant uptake of [³H]-methotrexate in HEK293 cells expressing chimpanzee and gorilla SLC22A10 long isoforms (Supplementary Fig. 1G).

The SLC22A10 protein in humans consists of 541 amino acids, resulting from a single nucleotide insertion that causes a frameshift in the last exon

The genetic mechanism that led to the formation of the 541 amino acid SLC22A10 protein in humans was investigated. Sequence alignments of the last exon (exon 10) of the SLC22A10 gene was compared between humans and great apes and revealed an insertion of one nucleotide leading to the expression of different isoforms in each species (see Fig. 2C). In particular, humans exhibit an A nucleotide insertion at the first base pair of exon 10, which is highly prevalent with an allele frequency of 98% in all populations in gnomAD¹² (11-63311006-GA-G (GRCh38)/11-63078478-GA-G (GRCh37)). The frequency ranges from 95% in the African population to over 99% in other ethnic groups. In contrast, the adenosine insertion has a 2.5% allele frequency in chimpanzees and is not present in other great apes. The adenosine insertion in the human SLC22A10 gene causes a frameshift and results in a 541 amino acid protein, instead of the predicted 552 amino acids in the SLC22A10 gene of great apes and in the humans who do not harbor the adenosine insertion (Fig. 2C). We utilized our previously generated human and liver RNAseq data¹³ alongside a large human RNA-seq databases (recount3¹⁴) to validate the splicing event in human and chimpanzee SLC22A10. Our analysis confirmed that splicing mostly occurs at the exact orthologous genomic region in both species, utilizing the canonical splice sites in their corresponding genomes. Thus, we found no coordinated splicing alterations that compensate for the A nucleotide insertion. Consequently, the additional A nucleotide remains present in the final transcribed transcript in humans. This additional nucleotide provides evidence that human SLC22A10 protein contains 541 amino acids, while the chimpanzee SLC22A10 protein comprises 552 amino acids.

Additionally, SLC22A10 harbors a prevalent nonsense variant, p.Trp96Ter (rs1790218), which is frequently observed at high allele frequencies in human populations ranging from ~20% in African to 50% in European (reported in the Single Nucleotide Polymorphism Database, dbSNP, version 156). Nonsense-mediated decay (NMD) is known to be triggered by premature stop codons. The Genotype-Tissue Expression (GTEx) project uncovered a significant association between the rs1790218 variant, which encodes the A-allele and p.Trp96Ter, and a considerable decrease in the transcript level of SLC22A10 samples (source: https://gtexportal.org/home/snp/rs1790218). This finding suggests that individuals carrying one copy of the p.Trp96Ter are likely to exhibit an even lower protein signal in the liver in comparison to individuals that harbor p.Trp96.

Mutagenesis of a single amino acid, at position 220 of the human SLC22A10 ortholog, to the respective amino acid in great apes rescues the function of human SLC22A10

The alignment of human SLC22A10 and primate ortholog sequences revealed differences in amino acid positions p.Met18IIe and p.Pro220Leu (Fig. 3A). Interestingly, site-directed mutagenesis experiments demonstrated that the substitution of proline with leucine at position 220 (p.Pro220Leu) restored the plasma membrane localization and function of human SLC22A10, but the substitution of methionine with isoleucine at position 18 (p.Met18Ile) did not have any effect (see Fig. 3B). In contrast, the replacement of leucine with proline at position 220 (p.Leu220Pro) abolished the localization and function of the chimpanzee SLC22A10 (see Figs. 3B and 3C). Additionally, we observed that a human-chimpanzee chimera protein, consisting of a fusion of human SLC22A10 (1-533) with chimpanzee SLC22A10 (534-552), while retaining the proline residue at position 220, showed no function (Figs. 3B and 3C). However, with a leucine substitution at position 220, SLC22A10 remained functional (Fig. 3B).

Human-specific p.Leu220Pro predates the common ancestor of modern humans and Neanderthals

Human-specific SLC22A10 variants likely emerged following the divergence of chimpanzee and human lineages. To better understand when SLC22A10 became non-functional during human evolution, we leveraged genomic data from archaic hominins^15–18 and ancient humans¹⁹. We first considered the fixed human variant at chr11: 63,064,927 (hg19), which results in the p.Leu220Pro amino acid in humans change compared to other apes. All four high-coverage archaic hominin genomes, representing three Neanderthals and a Denisovan, harbor a cytosine at this position, suggesting that this fixed difference evolved before the common ancestor of archaic hominins and modern humans, ~500 to 700 kya^15,20. Next, we evaluated the p.Trp96Ter nonsense polymorphism at chr11: 63,057,925 (hg19). This variant was not assayed in archaic hominin genomes; however, we identified two SNPs in high linkage disequilibrium (LD) with the polymorphism that were also genotyped in ancient human samples. Variants in high LD constitute a haplotype and enable indirect dating of the nonsense variant. We detected both alleles in Eurasians as early as 30,000 years ago (Supplementary Fig. 2). Further, analyses of modern human genomes estimate polymorphism itself likely evolved ~ 120 kya (Supplementary Fig. 3). Third, we assessed the high-frequency insertion largely absent in other apes at chr11: 63,078,478 (hg19). While this variant could not be assessed using ancient DNA, we did not find LD between alleles at this locus and alleles at the p.Trp96Ter polymorphism indicating that these variants occur on separate haplotypes (Supplementary Table 1). This supports the hypothesis that the insertion emerged separately from the polymorphic stop and is also ancient; however, it is challenging to determine which arose first. Taken together, data from ancient DNA reveal that the p.Leu220Pro mutation predates the archaic hominin-modern human common ancestor and that both polymorphisms are also likely ancient.

Human and chimpanzee SLC22A10 with Pro220 exhibit lower protein expression compared to orthologs with Leu220

The objective of this study was to analyze the protein expression of SLC22A10 in HEK293 Flp-In cells that were transfected with either vector only, or the cDNA of human or chimpanzee SLC22A10. This was achieved by quantifying the global proteomes of the cells, with a specific focus on amino acids at position 220 of SLC22A10. Comparable transcript levels of SLC22A10 in HEK293 cells that over-expressed either human or chimpanzee SLC22A10, as well as the respective variants (p.P220L or p.L220P), were observed (Supplementary Fig. 4). However, as illustrated in Table 1, lower protein expression levels for the human SLC22A10 reference (Proline220) and chimpanzee SLC22A10-L220P were observed when compared to SLC22A10 with leucine at the 220 amino acid position. The results showed that protein levels of human SLC22A10 are approximately 10-fold lower in cells expressing human SLC22A10-Pro220 compared to human SLC22A10-Leu220 (Table 1) suggesting that human SLC22A10 is transcribed but the protein is unstable. The lower overall protein expression may explain the lack of detectable expression of human SLC22A10 on the plasma membrane in contrast to orthologs (both human and chimpanzee) of SLC22A10 with Leu220. In particular, human SLC22A10 was not detected on the plasma membrane (Fig. 1B), whereas the mutant, SLC22A10-Leu220 exhibited expression on the plasma membrane (Fig. 3C).

Human SLC22A10 with Pro220 is predicted to have poor stability compared to orthologs with Leu220

We tested the effect of the P220L mutation on protein stability using a prediction pipeline tuned for transmembrane proteins²¹. The Rosetta physics-based score suggested that P220L is a stabilizing mutation with a ΔΔG value of −9.84 Rosetta Energy Units (REU). Using rank-normalized evolutionary-based ΔΔE scores, we calculated values of 0.0 for leucine and 0.16 for proline at this position. This combination of a higher ΔΔE value and a less negative ΔΔG value suggests that the Pro220 creates a potential loss of function protein, likely due to decreased stability and lower cellular abundance²². To test this idea, we explored the relative stability of human SLC22A10 (Pro220) and SLC22A10 (Leu220) using flow cytometry. Specifically, HEK293 Flp-In cells were stably transfected with vector control, human SLC22A10 reference (Pro220), or human SLC22A10 mutant (Leu220). In these cells, we measured both the stable level of GFP fluorescence and the response of that fluorescence to treatment with 30 nM of the 26 S proteosome inhibitor, bortezomib. For the L220 mutant, GFP fluorescence is high in comparison to the empty vector control and increases minimally upon treatment with bortezomib (Supplementary Fig. 5), suggesting that this protein is relatively stable and not actively degraded by the proteosome. In contrast, the GFP fluorescence of SLC22A10 reference (Pro220) occupies a value midway between that of the empty vector control and the SLC22A10 mutant (Leu220), confirming that it is less stable. Upon bortezomib treatment, the GFP fluorescence of SLC22A10 reference (Pro220) increases, such that it is comparable to that of SLC22A10 mutant (Leu220) (Supplementary Fig. 5). Thus, SLC22A10 (Pro220) appears to be relatively unstable and degraded by the proteosome, such that its protein levels can be partially restored by proteasome inhibition.

Analyses of SLC22A10 isoforms in chimpanzee, bonobo, orangutan and gibbon

Upon examining various databases (UniProt, Ensembl, and NCBI nucleotide) that predict SLC22A10 isoforms, we observed that chimpanzee, bonobo, orangutan, and gibbon are predicted to have different isoforms, including (i) chimpanzee (XM_024347215.1, 533 amino acid); (ii) bonobo (XM_034932815.1, 538 amino acid); (iii) gibbon (XM_003274123.1, 533 amino acid); and (iv) orangutan (XM_024255628.1, 533 amino acid). The shorter isoforms are derived from alternative acceptor sites (chimpanzee and bonobo) and exon extensions (orangutan and gibbon) resulting in 533-amino acid proteins (see Fig. 4A, Supplementary Fig. 6). We conducted experiments using HEK293 Flp-In cells that were stably transfected with GFP-conjugated chimpanzee, bonobo, and gorilla SLC22A10 to examine the localization and function of the shorter isoforms. The HEK293 Flp-In cells that were stably transfected with GFP-conjugated chimpanzee SLC22A10, consisting of 533 amino acids, showed a clear plasma membrane localization. However, it appears that the bonobo SLC22A10, which consists of 538 amino acids, exhibited weaker localization, and the orangutan SLC22A10, also comprising 533 amino acids, showed mixed localization (Fig. 4B). Our results indicated that the SLC22A10 533 amino acid isoform from chimpanzee, orangutan and gibbon accumulated [³H]-estradiol-17β-glucuronide and [³H]-folic acid, but not [³H]-estrone sulfate and [³H]-taurocholic acid, similar to the SLC22A10 isoform with 552 amino acids (Fig. 4C, Supplementary Fig. 7). Moreover, we observed that longer isoforms of SLC22A10 created for the bonobo and orangutan, which were not predicted to express these isoforms (552 amino acids), accumulated [³H]-estradiol-17β-glucuronide, and were also expressed on the plasma membrane (Supplementary Fig. 7). However, the uptake by bonobo SLC22A10, comprising 538 amino acids, is not significant, and there is a weaker fold uptake by orangutan SLC22A10, consisting of 533 amino acids, consistent with its mixed localization on the plasma membrane. (Fig. 4B and C). During the revision of this article, it was noted that the NCBI updated the sequence of bonobo SLC22A10 from 538 amino acids to 533 amino acids. This sequence of bonobo SLC22A10, comprising 533 amino acids, exhibited a significant uptake of [³H]-estradiol-17β-glucuronide and demonstrated clear plasma localization (Supplementary Fig. 8).

Uptake of various steroid glucuronides by chimpanzee SLC22A10 reveals that 17β-glucuronides are preferred over 3α-glucuronide conjugates

Our experiments revealed significant accumulation of estradiol-3β-glucuronide, estradiol-17β-glucuronide, estrone-3β-glucuronide and testosterone-17β-glucuronide (Fig. 5A). We also observed weak but significant accumulation of androstanediol-3α-glucuronide, but no significant accumulation of androsterone-3-glucuronide, etiocholanolone-3α-glucuronide, progesterone, and testosterone (Fig. 5A). Although the data are limited, it is intriguing to note the substrate preference of SLC22A10. Specifically, 17β-glucuronides appear to be favored over 3α-glucuronide conjugates.

Chimpanzee SLC22A10-mediated uptake is sodium independent, pH independent, and trans-stimulated by glutaric acid

Transporters in the SLC22 family may be secondary active, in which case they rely on various sources of energy to mediate active flux of their substrates. Accordingly, we examined the transport mechanism for two isoforms (533 and 552 amino acids) of chimpanzee SLC22A10. At the three pH levels evaluated, the uptake of [³H]-estradiol-17β-glucuronide by both isoforms of chimpanzee SLC22A10 were similar (Fig. 5B). While the uptake was not dependent on sodium (Fig. 5C), it was partially reduced in the absence of chloride in the buffer (Fig. 5C). Several organic anion transporters in the SLC22 family are trans-stimulated by dicarboxylic acids^3,23. Similarly, we observed that the uptake of [³H]-estradiol-17β-glucuronide was trans-stimulated by glutaric acid but not significantly by other dicarboxylates (α-ketoglutarate and succinic acid) or the monocarboxylic acid, butyrate (Fig. 5D). The uptake kinetics of [³H]-estradiol-17β-glucuronide exhibited saturable characteristics with virtually identical Km values for the two protein isoforms at 3.28 ± 2.25 µM and 2.16 ± 0.59 µM for the 533 aa and 552 aa isoforms, respectively (Fig. 5E).

Generation of SLC22A10 ortholog cDNA constructs

Sequences of SLC22A10 orthologs from various species, along with their respective sequence IDs, were obtained from ensembl.org (release version 104)⁵⁵ and the National Center for Biotechnology Information (NCBI). The gene’s cDNA was synthesized using GenScript Gene Synthesis and DNA Synthesis Services and subsequently inserted into the multiple cloning sites on the expression vectors pcDNA3.1(+) or pcDNA3.1( + )N-eGFP. The sequence of the constructs was verified through sequencing conducted by GenScript.

Human SLC22A10, NM_001039752, NP_001034841.3 (541 aa)

Chimpanzee SLC22A10, XM_016921102.2, XP_016776591.1 (552 aa) Chimpanzee SLC22A10, ENSPTRT00000087785.1, A0A2I3TWY5 (540 aa)

Chimpanzee SLC22A10, XM_024347215.1, XP_024202983 (533 aa)

Bonobo SLC22A10, XM_055094430.1, XP_054950405.1 (533 aa)

Bonobo SLC22A10, XM_034932815.1, XP_034788706 (538 aa)

Bonobo SLC22A10, ENSPPAT00000058172.1, A0A2R9CA90 (540 aa)

Gorilla SLC22A10, XM_019037494.1, XP_018893039 (552 aa)

Gorilla SLC22A10, ENSGGOT00000014974.3, G3RG24 (540 aa)

Orangutan SLC22A10, XM_024255628.1, XP_024111396.1 (533 aa)

Orangutan SLC22A10, XM_054439153.1, XP_054295128.1 (552 aa)

Gibbon SLC22A10, XM_003274123.2, XP_003274171.2 (533 aa)

Squirrel monkey SLC22A10, ENSSBOT00000021267.1, A0A2K6SBP5 (552 aa)

Site-directed mutagenesis

Site-directed mutagenesis to create Methionine18 to Isoleucine18 in human SLC22A10, Proline220 to Leucine220 or vice versa in chimpanzee SLC22A10 and human SLC22A10, were performed using Q5 Site-Directed Mutagenesis Protocol from NEB (#E0554). NEBaseChanger, https://nebasechanger.neb.com/, is used to assist in the design of primers for the site-directed mutagenesis experiment. The following primers were ordered from IDT™ (Integrated DNA Technologies).

Primers for p.M18I:

mut_hsA10_c54C_F: GATTTCAGATcCTTCATCTGGTTTTTATTCTTC

mut_hsA10_c54C_R: TCCCAAGGCCTCCAACTT

Primers for p.P220L:

mut_hsA10_c659T_F: AATTCTTTGCtCATTACTGAG

mut_hsA10_c659T_R: ATTTGATATAATGATCATGGAAG

Primers for p.L220P:

mut_ptA10_c659C_F: AATTCTTTGCcCATTACTGAGTG

mut_ptA10_c659C_R: ATTTGATATAATGATCATGGAAG

Primers for deletion of A nucleotide in position c1607 and then inserting the tail at the last exon for human to chimpanzee (consist of 34 bp):

mut_hsA10_c1607Adel_F: TCTCAAGGAAAAGGCATAAATG

mut_hsA10_c1607Adel_R: TTTTTTTTTGATTTTCCACATCC

mut_hsA10_c1622ins_F: caaaagtgaccaaatttTAACTCGAGTCTAGAGGG

mut_hsA10_c1622ins_R: tgtagcaatcatttttaTGCCTTTTCCTTGAGATTTTTTTTTG

Cell culture

Human embryonic kidney cell lines (HEK293) containing a Flp-In expression vector (HEK293 Flp-In) obtained from ThermoFisher Scientific (#R75007) were utilized in all experiments in this study. These cells have previously been employed for the generation of stable cell lines for transporter assays^2,3,23,46. HEK293 Flp-In cells have not been authenticated. The HEK293 Flp-In cells were cultured in DMEM, high glucose (#11965118, ThermoFisher Scientific) supplemented with 10% fetal bovine serum (heat inactivate, #10438026, ThermoFisher Scientific). Penicillin-Streptomycin (#15070063, ThermoFisher Scientific) was added to DMEM media (50 unit/500 mL DMEM). During transfection and when cells were plated for transporter studies, media without penicillin/streptomycin supplementation were used. The cells were periodically screened for mycoplasma contamination (MycoProbe Mycoplasma Detection Kit, #CUL001B, Fisher).

Generation of cells transiently or stably expressing cDNAs

Expression vectors of SLC22A10 orthologs were introduced into HEK293 Flp-In cells either through transient transfection or stable transfection using Lipofectamine LTX (Thermo Fisher Scientific). For transfections in a 48-well plate (seeding density: 1.0×10⁵ cells/well), 200 ng of DNA and 0.4 μL of Lipofectamine LTX were utilized, while for transfections in a 100 mm tissue culture plate (seeding density: 4 × 10⁶ cells/well), 10 μg of DNA and 44 μL of Lipofectamine LTX were used. More comprehensive methods for generating transiently or stably transfected cells have been described in our previous work (see reference^2,3). In the case of transient transfection, cells were used for transporter studies (refer to the section titled “Transporter uptake studies”) after 36-48 hours or for protein quantification after 72 hours. To establish stable cell lines, 3000 ng of DNA (SLC22A10 ortholog expression vectors) and 10.5 μL of Lipofectamine LTX were employed to transfect HEK293 Flp-In cells seeded in a 6-well plate (seeding density: 7-8 × 10⁵ cells/well). After 48 hours, cells were transferred to a new 100 mm tissue culture plate and treated with 800 μg/mL Geneticin. Fresh media containing 800 μg/mL Geneticin was replenished every other day for 1 week. Stable cell lines were utilized for confocal imaging to determine the plasma membrane localization of SLC22A10 orthologs and their various isoforms. Unless specified otherwise, stable cell lines were used for transporter assays.

Fluorescence microscopy

For the immunostaining experiments, HEK293 Flp-In stable cell lines expressing different SLC22A10 orthologs were cultured on poly-D-lysine-treated 12-well plates with sterile coverslips at a density of 200,000 cells per well. After two days of seeding when the cells reached 90-100% confluency, the staining procedure was conducted. On the day of staining, the cell culture media was carefully removed, and the cells were washed using cold Hank’s Balanced Salt Solution (HBSS, #14025092, ThermoFisher Scientific). To initiate the staining process, the plasma membrane was first labeled using Wheat Germ Agglutin (WGA) Alexa Fluor 647 conjugate (Invitrogen Life Sciences Corporation) diluted at a ratio of 1:500 in HBSS, followed by a 15-minute incubation at room temperature. Following the staining step, the WGA solution was aspirated, and the cells were washed three times with HBSS. Subsequently, the cells were fixed with a solution of 3.7% formaldehyde in HBSS for 20 minutes. After the fixation step, the cells were washed three times with HBSS. To stain the nucleus, Hoechst solution (ThermoFisher Scientific Inc.) diluted at a ratio of 1:2000 in HBSS was applied to the cells and incubated for 20 minutes at RT, in darkness. After the staining period, the Hoechst solution was aspirated, and the cells were washed twice with HBSS. The coverslips were carefully mounted on Superfrost Plus Microscope Slides (ThermoFisher Scientific) using a small amount of SlowFade^TM Gold Antifade Mountant (#S36940, ThermoFisher Scientific). The mounted slides were left to dry overnight in darkness before being imaged using an inverted Nikon Ti microscope equipped with a CSU-22 spinning disk confocal system available at the Center for Advanced Light Microscopy (CALM) at University of California San Franciso. The image acquisition settings were as follows: DAPI channel with a 300 ms exposure time and 50% laser power, FITC channel with a 300 ms exposure time and 25% laser power, and CY5 channel with a 100 ms exposure time and 5% laser power. Image alignment and merging were performed using Fiji software. This experimental protocol has been previously utilized and described in our published work^56,57.

Transporter uptake studies

HEK293 Flp-In cells expressing SLC22A10 were seeded at a density of 120,000 to 150,000 cells/0.3 mL in poly-D-lysine-coated 48-well plates approximately 16 to 24 hours prior to conducting uptake studies. The uptake studies for transporters, described below, are methods published by our research group^2,3. For transiently expressing SLC22A10, the methods outlined in the previous section pertaining to transient expression in HEK293 Flp-In cells were followed prior to this step. Prior to uptake studies, the culture medium (Dulbecco’s modified Eagle’s medium, DMEM) supplemented with 10% fetal bovine serum was aspirated, and the cells were incubated in 0.8 mL of Hank’s Balanced Salt Solution (HBSS) at 37 °C for 10-20 minutes. For screening radiolabeled compounds as SLC22A10 substrates, minute quantities of radiolabeled compounds (³H or ¹⁴C) were diluted in HBSS (at ratios of 1:2000 or 1:3000) for uptake experiments. Unlabeled compounds were added to obtain specific concentrations, which are described in the Results section or figure legends along with the uptake times. Uptake reactions were terminated by washing the cells twice with 0.8 mL of HBSS buffer, followed by incubation in 750 μL of lysis buffer (0.1 N NaOH, 0.1% v/v SDS). A 690 μL portion of the cell lysate was transferred to scintillation fluid for scintillation counting. For pH dependence experiments, the HBSS buffer was adjusted to different pH levels (5.5, 7.4, and 8.5) using hydrochloric acid or sodium hydroxide. For sodium and chloride dependence studies, three distinct uptake buffers were employed: (1) chloride-free buffer (composed of 125 mM sodium gluconate, 4.8 mM potassium gluconate, 1.2 mM magnesium sulfate, 1.3 mM calcium gluconate, and 5 mM HEPES; adjusted to pH 7.4 with sodium hydroxide); or (2) sodium buffer (composed of 140 mM sodium chloride, 4.73 mM potassium chloride, 1.25 mM calcium chloride, 1.25 mM magnesium sulfate, and 5 mM HEPES, adjusted to pH 7.4 with sodium hydroxide); or (3) sodium-free buffer (composed of 140 mM N-methyl-D-glucamine chloride, 1.25 mM magnesium sulfate, 4.73 mM potassium chloride and 1.25 mM calcium chloride), adjusted to pH 7.4 with potassium hydroxide). For trans-stimulation studies, the experimental conditions described in our previously published methods were followed^2,3. In brief, the SLC22A10 or EV stable cell lines were pre-incubated with either buffer or 2 mM succinic acid, 2 mM α-ketoglutaric acid, 2 mM butyric acid, or 2 mM glutaric acid for 2 hours. Subsequently, the cells were washed twice with HBSS before commencing the uptake of the anions (estradiol-17β-glucuronide). All transporter uptake assays were replicated in at least one additional experiment to confirm the results. All results shown in figures are from a representative experiment with three or four technical replicates.

Kinetic studies of estradiol glucuronide

Kinetic studies of estradiol-17β-glucuronide were conducted in HEK293 Flp-In cells expressing chimpanzee SLC22A10 isoforms (533 amino acid and 552 amino acid) that were stably transfected. The kinetic studies described below followed the methods published by our research group^2,3,23. Initially, we examined the time-dependent uptake of the substrates using trace amounts of the radioactive compound. Concentrations of the non-labeled compounds were varied up to 50 µM. For the kinetic studies, a duration of five minutes at 37 °C was chosen as it fell within the linear range observed in the uptake versus time plot for each substrate. Each data point represents the mean ± standard deviation of uptake in the cells transfected with the transporter, subtracted by that in empty vector cells. The obtained data were fitted to a Michaelis-Menten equation to estimate the kinetic parameters. Plots were generated based on a representative experiment out of three independent studies.

Protein extraction and global proteomics of HEK293 cells expressing SLC22A10 orthologs

HEK293 Flp-In cells were transfected transiently with various SLC22A10 orthologs, including human, chimpanzee, and the mutations to proline or leucine at position 220. After 72 hours of transfection, cell pellets were collected and shipped to Dr. Per Artursson’s laboratory in Uppsala University for protein quantification. The quantification was performed on both HEK293 cells and HEK293 cells expressing the different SLC22A10 orthologs and mutations. HEK293 cell pellets (50–92 mg) were lysed in a lysis buffer containing 50 mM dithiothreitol, 2% sodium dodecyl sulfate in 100 mM Tris/HCl pH 7.8. The lysates were incubated at 95 °C for 5 min and sonicated with 20 pulses of 1 second, 20% amplitude by using a sonicator coupled with a microtip probe. The lysates were centrifuged at 14,000×g for 10 min and supernatants were collected. Using LysC and trypsin, the multi-enzyme digestion filter-aided sample preparation (MED-FASP) approach was performed⁵⁸. C18 stage tips were used to desalt the peptide mixture^59,60 and samples were stored at −20 °C until analysis. Protein and peptide content were determined by using tryptophan fluorescence assay⁶¹. The global proteomics analysis was performed on a Q Exactive HF mass spectrometer (Thermo Fisher Scientific) coupled to a nano–liquid chromatography (nLC). EASY-spray C18-column (50 cm long, 75 µm inner diameter) was used to separate peptides on a ACN/water gradient (with 0.1% formic acid) over 150 min. MS was set to data dependent acquisition with a Top-N method (full MS followed by ddMS2 scans). Proteins were identified using MaxQuant software (version 2.1.0.0)⁶² with the human proteome reference from UniProtKB (October, 2022). Total protein approach was used as the protein quantification method⁶³. The minimum number of unique peptides for quantification is more than 2 unique and razor peptides. The proteomics data generated in this study have been deposited in the PRIDE database under accession code PXD047102.

RNA isolation and quantitative RT-PCR

HEK293 Flp-In cells were cultured in poly-D-lysine coated 24-well plates at a seeding density of 1.5-1.8 ×10⁵ cells per well, allowing them to reach 75-80% confluency. The RT-PCR method for transcript levels determination as detailed bleow, are methods we have previously described³. Once the desired confluency was achieved, the cells were transiently transfected with either the vector alone or the vector containing different SLC22A10 orthologs (in the pcDNA3.1(+) expression vector). For the transfection mixture, 500 ng of plasmid DNA, 1 μL of Lipofectamine LTX (Thermo Fisher Scientific), and 100 μL of Opti-MEM I reduced serum media (Thermo Fisher Scientific) were used. After 36-48 hours of transfection, the media was removed, and RNA Lysis buffer (350 μL) was added to each well. Total RNA was isolated from the cells using the Qiagen RNeasy kit (Qiagen). Subsequently, cDNA was synthesized using the SuperScript VILO cDNA Synthesis Kit (ThermoFisher Scientific). For quantitative RT-PCR (qRT-PCR), Taqman reagents and specific primer and probe sets were used, targeting human SLC22A10 (Assay ID: Hs01397962_m1) and beta actin (Assay ID: Hs99999903_m1) (Applied Biosystems, Foster City, CA). The qRT-PCR reactions were conducted in a 96-well plate, with a reaction volume of 10 μL, using the QuantStudio™ 6 Flex Real-Time PCR System and the default instrument settings. The expression levels were determined using the Ct method, and the data were normalized to the endogenous levels of beta actin. The results are presented as fold-increases in the SLC22A10 transcript levels relative to the cell lines expressing the vector control.

Cloning of SLC22A10 in pooled human liver

For the cloning process, pooled total RNA samples from human liver were obtained from Clontech. Each sample (2 μg) of total RNA was reverse transcribed into cDNA using the SuperScript VILO cDNA Synthesis kit (Thermo Fisher Scientific) following the manufacturer’s instructions. The primers specified below were employed for PCR amplification of the NM_001039752 transcript: Forward primer: ACCGAGCTCGGATCCATGGCCTTTGAGGAGCTC; and reverse primer: CCCTCTAGACTCGAGTTATGCCTTTTCCTTGAGATT. The nucleotide underlined are open reading frame of SLC22A10. The resulting PCR products were cloned into BamHI and XhoI multiple-cloning site of the pcDNA5FRT vector and subsequently subjected to sequencing at MCLAB in South San Francisco to determine the sequence of the transcript. For the cloning of human SLC22A10, the KOD Xtreme Hot Start DNA polymerase kit (Takara) was utilized. The PCR cycling conditions were as follows: (i) initial activation at 94 °C for 2 minutes, (ii) denaturation at 98 °C for 10 seconds, (iii) annealing at 57.5 °C for 30 seconds, and (iv) extension at 68 °C for 1 minute.

Calculating ΔΔG with the PRISM’s rosetta_ddG_pipelne v 0.2.4²¹ using Rosetta v 3.15

In brief, the full-length human SLC22A10-P220 (Uniprot ID Q63ZE4) structure predicted by the AlphaFold⁶⁴ from the AlphaFold DB⁶⁵ was oriented in the membrane with PPM 3.0 web server⁶⁶ with the default settings and relaxed using the Rosetta’s relax protocol⁶⁶. The best of 20 generated structures was used for the ΔΔG calculations with the cartesian_ddG protocol⁶⁷ repeated in 5 replicas. ΔΔE scores were calculated using GEMME⁶⁸ and rank-normalized as in the reference paper²¹. The relaxed structure, ΔΔG, and ΔΔE results are available at zenodo, [10.5281/zenodo.8411757].

Flow cytometry analysis

To determine if human SLC22A10 reference (Pro220) is rapidly degraded upon biosynthesis, HEK293 Flp-In cells stably transfected with SLC22A10 reference (Pro220) or SLC22A10 mutant (Leu220) in the pcDNA3.1( + )-NGFP, or an empty vector control and seeded in a 24-well cell-culture plate. When the cells had reached approximately 70 % confluence, they were treated with 30 nM of the 26 S proteosome inhibitor bortezomib (Millipore Sigma (CAS 179324-69-7)). After 16 hours the cells were washed and analyzed by flow cytometry on an Attune NxT Acoustic Focusing Cytometer (Life Technologies). Data was analyzed using the Flowjo software package and GFP fluorescence used to determine the extent of protein accumulation upon proteosome inhibition.

Transporter uptake studies and LC-MS/MS analysis

The list of steroid conjugates and their sources can be found in the Supplementary Table 2. Each steroid conjugates was dissolved in DMSO to obtain 20 mM stock solution. Compounds were stored in −20 °C freezer. HEK293 Flp-In cells stably transfected with GFP only, chimpanzee SLC22A10 (533 amino acid) and chimpanzee SLC22A10 (552 amino acid) were plated in poly-D-lysine coated 48-well plates at a seeding density of 1.5 ×10⁵ cells per well, allowing them to reach 90-95% confluency after 16-24 hours. Prior to uptake studies, the culture medium (Dulbecco’s modified Eagle’s medium, DMEM) supplemented with 10% fetal bovine serum was aspirated, and the cells were incubated in 0.8 mL of Hank’s Balanced Salt Solution (HBSS) at 37 °C for 10-20 minutes. To screen various steroid and steroid conjugate compounds as SLC22A10 substrates, HEK293 Flp-In cells stably expressing GFP or chimpanzee SLC22A10 were incubated with HBSS buffer containing 10 µM of the respective compounds for 20 minutes. The uptake reactions were terminated by washing the cells twice with 0.8 mL of HBSS buffer, followed by incubation in 400 μL of methanol. After 30 minutes of shaking at room temperature, 300 µL of methanol containing the extracted steroid or steroid conjugates from each well were transferred to a 1.5 mL tube and stored at −80 °C before quantification using LC-MS/MS analysis.

Experimentally, cells in culture were presented with commercially available steroid metabolites and their presence in isolated cells was quantified against authentic standards. Subaliquot of methanol disrupted cellular extracts (90 µL) were spiked with 10 µL of deuterated 17β-estradiol glucuronide, mixed by vortexing, and filtered at 0.2 µm through polyvinyl difluoride membranes (Agilent Technologies, Santa Rosa, CA, USA) by centrifugation and 10,000 g. After filtration, samples were enriched with 25 nM 1-cyclohexyl-3-uriedo-decanoic acid (Sigma-Adlrich, St. Lousi MO) as an internal standard. Metabolites were measured using ultra-performance liquid chromatography–electrospray ionization tandem mass spectrometry (UPLC-ESI-MS/MS) on a API 4500 QQQ (Sciex, Framingham, MA) with a scheduled multiple reaction monitoring (MRM) using methods adapted from Ke et.al, 2015⁶⁹. Analytes were separated on a Waters I-Class UPLC-FTN equipped with a 2.1 × 100 mm i.d., 1.7 μm Acquity BEH C₁₈ column (Waters Co; Milford, MA) held at 50 °C. Analytes in 5 µL injections were separated using water (solvent A) and methanol (solvent B) both containing 2 mM ammonium formate at 400 µL/min with the following gradient: Initial 40% B to 70% B at 2 min, to 98%B at 3 min, held to 4 min, to 40%B at 5.1 min held to 6 min. Data were processed using MultiQuant v. 3.02 (Sciex), with five-point calibration curves bracketing the observed concentration ranges fit with quadradic curves using 1/x weighting for each steroid conjugate (Supplementary Fig. 10). Mass spectrometer acquisition parameters and analyte retention times are described in Supplementary Table 3.

Sequencing data processing and analyses of SLC22A10 in greater apes

Orthologous genome regions of SLC22A10 coding sequence in multiple primate species were obtained using UCSC liftOver (default parameters)⁷⁰ based on hg38 (human), panTro6 (chimpanzee), panPan3 (bonobo), gorGor6 (gorilla) and ponAbe3 (orangutan) assemblies. These regions were aligned using MUSCLE⁷¹ and visualized using MView⁷². SLC22A10 exons and coding sequences are based on RefSeq annotations⁷³ for human, chimpanzee and gorilla. Despite the long SLC22A10 isoform not being annotated in RefSeq for bonobo and orangutan, we included these two species in the alignment considering the plausibility of the protein model in the context of their genomes and the high genomic conservation in comparison to chimpanzee and gorilla.

Allele frequencies from non-human great apes were obtained from available whole-genome sequencing data including 59 chimpanzees, 10 bonobos, 49 gorillas and 16 orangutans^74–77. All samples were mapped to Hg19. We extracted the genotyping information in position chr11:63078478 (Hg19 coordinates) to calculate the allele frequencies per population. We manually curated the genotypes by checking the raw reads overlapping this region in the BAM files. In chimpanzees we report a frequency of the insertion to be 3/108 = 2.5%; in bonobos is 0/20 = 0%; in gorillas is 0/98 = 0%; and in orangutans is 0/32 = 0%. The global human allele frequencies were obtained from 1000 Genomes Project Phase 3⁷⁸ database in Ensembl for the rs562147200 SNP.

SLC22A10 variants in archaic hominins and ancient humans

Archaic hominin genotypes for three Neanderthals and a Denisovan were retrieved from [http://ftp.eva.mpg.de/neandertal/Vindija/VCF/[and [http://ftp.eva.mpg.de/neandertal/Chagyrskaya/VCF/]^15–18. We used the LDproxy tool from LDlink⁷⁹ to identify variants in high LD with rs1790218 in all Thousand Genomes populations and intersected these variants with those assayed in ancient humans from the Allen Ancient DNA Resource (AADR)¹⁹, retrieved from [https://reichdata.hms.harvard.edu/pub/datasets/amh_repo/curated_releases/V54/V54.1.p1/SHARE/public.dir/v54.1.p1_1240K_public.tar]. We identified two such variants: rs1783634 (D’ = 1, r2 = 0.9839) and rs1201559 (D’ = 0.9975, r2 = 0.9775). In the AADR analyses, we excluded samples from archaic hominins and individuals that were not genotyped at both loci. For modern samples present in duplicate in the AADR, we filtered out genotypes called from pseudo-haploid sequences and retained a single genotype per SNP per individual from samples with coverage sufficient for calling diploid genotypes (N = 2,610). Among the remaining duplicated samples, we retained genotypes for each individual where the genotypes from the duplicated samples matched at both proxy SNPs (N = 83). Genotypes that did not match between duplicated samples were removed from consideration (N = 11). This resulted in genotypes from 5,045 individuals (1,375 ancient and 3,670 modern). We calculated allele frequency in 17 time periods, stratifying by five continental groups defined based on sampling locality: Africa, Americas, Europe, Asia, and Oceania. We also retrieved the allele age estimate for rs1790218 using the Human Genome Dating tool⁸⁰ with all default settings. We used the LDhap tool LDlink to quantify haplotypes for rs1790218 and rs562147200 in all Thousand Genomes populations. The archived version of the code used to analyze ancient human genomes and visualize allele frequencies has been deposited in Zenodo [10.5281/zenodo.10823210]. A non-archived version is available on GitHub [https://github.com/brandcm/SLC22A10_Variant_Evolutionary_History].

Statistical analysis

When comparing the significant differences among HEK293 cells transfected with GFP only and various SLC22A10 ortholog species or other transporters, we performed multiple comparisons using one-way analysis of variance followed by Dunnett’s two-tailed test. HEK293 cells transiently transfected with the GFP vector served as the control. The fold uptake of the substrate, relative to the control cells, was plotted based on one representative experiment conducted in triplicate wells (mean ± s.d.). Statistical significance was indicated as ***p < 0.0001, **p < 0.01, *p < 0.05. These findings were further confirmed through at least one or two additional experiments. For specific differences and more detailed information, please refer to the figure legend.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Source data

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.