To test our hypothesis regarding the involvement of Abcc3 in the sinusoidal cycling of bilirubin glucuronides, and to assess a possible interplay with Abcc2, we generated Slco1a/1b–/–Abcc3–/–
), and Slco1a/1b–/–Abcc2–/–Abcc3–/–
) mice by crossbreeding of existing strains. All strains were fertile, with normal life spans and body weights. As previously found for Abcc2–/–
), liver weights of Slco1a/1b;Abcc2–/–
mice were significantly increased (~30% and ~50%, respectively) compared with wild-type mice (data not shown). Quantitative RT-PCR analysis of functionally relevant uptake and efflux transporters in liver, kidney, and intestine of the single and combination knockout strains revealed only some modest expression changes (Supplemental Table 1 and Supplemental Results; supplemental material available online with this article; doi:
). Hepatic UDP-glucuronosyltransferase 1a1 (Ugt1a1
) expression was not significantly altered in any of the strains.
Importantly, the markedly increased plasma bilirubin monoglucuronide (BMG) and bilirubin diglucuronide (BDG) levels observed in Slco1a/1b–/–
mice were substantially reduced in Slco1a/1b;Abcc3–/–
mice, demonstrating that Abcc3 is necessary for most of this increase (Figure , A and B). Plasma BMG levels in Slco1a/1b;Abcc2–/–
mice, even further increased owing to strongly reduced biliary BMG excretion (Figure , A and B), were similarly decreased in Slco1a/1b;Abcc2;Abcc3–/–
mice (Figure , A and B). Thus, Abcc3 secretes bilirubin glucuronides back into blood, and Oatp1a/1b proteins mediate their efficient hepatic reuptake, thereby together establishing a sinusoidal liver-blood shuttling loop. The incomplete reversion of plasma bilirubin glucuronide levels in the Oatp1a/1b/Abcc3-deficient strains (Figure , A and B) suggests that additional sinusoidal exporter(s), e.g., Abcc4 (28
), can partly take over the sinusoidal bilirubin glucuronide extrusion function of Abcc3.
Increased plasma bilirubin glucuronide in Slco1a/1b–/– mice is in part dependent on Abcc3.
In the presence of Oatp1a/1b, but not in its absence, Abcc3 enhances biliary excretion of bilirubin glucuronides.
The biliary output of bilirubin glucuronides in the single and combination knockout mice showed that, as long as Oatp1a/1b was functional, Abcc3 improved the efficiency of biliary bilirubin glucuronide excretion, even though it transports its substrates initially from liver to blood, not bile (Figure , A and B, strains +Oatp1a/1b). This suggests that, within liver lobules, the bilirubin glucuronide extruded by Abcc3 in upstream hepatocytes is efficiently taken up in downstream hepatocytes via Oatp1a/1b and then excreted into bile. The resulting relief of possible saturation of (or competition for) biliary excretion in the upstream hepatocytes may explain why the overall biliary excretion is enhanced by this transfer to downstream hepatocytes. However, when Oatp1a/1b was absent, Abcc3 instead decreased biliary bilirubin glucuronide excretion (Figure , strains –Oatp1a/1b) and redirected excretion toward urine via the increased plasma bilirubin glucuronide levels (Supplemental Figure 1). Obviously, in the absence of Oatp1a/1b-mediated hepatic reuptake, Abcc3 activity can only decrease hepatocyte levels of bilirubin glucuronide in upstream and downstream hepatocytes alike, and will therefore reduce overall biliary excretion. Thus, both components of the Abcc3 and Oatp1a/1b shuttling loop are necessary to improve hepatobiliary excretion efficiency.
Human hepatocytes express only two OATP1A/1B proteins at the sinusoidal membrane, OATP1B1 and OATP1B3 (15
). To test whether these could mediate the identified Oatp1a/1b functions, and in a liver-specific manner, we generated Slco1a/1b–/–
mice with liver-specific expression of either human OATP1B1 or OATP1B3. Liver-specific expression was obtained using an apoE promoter (29
). These strains were viable and fertile, and displayed normal life spans and body weights. Liver levels of transgenic OATP1B1 and OATP1B3 proteins were similar to those seen in pooled human liver samples (data not shown). Both of the transgenic rescue strains displayed a virtually complete reversal of the increases in plasma and urine levels of BMG and BDG seen in Slco1a/1b–/–
mice (Figure , A and B, and Supplemental Figure 2). This indicates that both human OATP1B1 and OATP1B3 effectively reabsorb bilirubin glucuronides from plasma into the liver, in line with their demonstrated in vitro role in bilirubin glucuronide uptake (30
). The modest (~1.8-fold) increase in plasma UCB in Slco1a/1b–/–
mice was also reduced in the rescue strains (Figure C), suggesting an ancillary role of these proteins in hepatic UCB uptake.
Increased plasma bilirubin glucuronide in Slco1a/1b–/– mice is reversed by human OATP1B1 and OATP1B3.
These findings collectively raised the question as to whether humans with a severe deficiency in OATP1B1 and OATP1B3, possibly leading to a conjugated hyperbilirubinemia, might exist. A literature search suggested RS as a candidate inborn metabolic disorder. A search for RS subjects by part of the present group led to collaboration with another team already working on mapping of the RS gene(s).
In an unbiased approach, scanning the whole genome, we mapped the genomic candidate intervals for RS in 11 RS index subjects from 8 different families, 4 Central European (CE1–CE4), 3 Saudi-Arabian (A1–A3), and 1 Filipino (P1) (Figure A and Supplemental Table 2). Homozygosity mapping identified a single genomic region on chromosome 12 for which 8 tested index subjects and no healthy siblings or parents were homozygous (Figure B), suggesting inheritance of both alleles from a common ancestor. Three distinct homozygous haplotypes (R1–R3) segregated with RS: R1 in families CE1, CE2, and CE4; R2 in families CE3, A1, A2, and A3; and R3 in family P1 (Figure B; for genotyping details, see Methods). Intersection of these haplotypes defined a candidate genomic region spanning the SLCO1C1, SLCO1B3, SLCO1B1, SLCO1A2, and IAPP genes (Figure B). A parallel genome-wide copy number analysis detected a homozygous deletion within the SLCO1B3 gene in the R1 haplotype and a homozygous approximately 405-kb deletion encompassing SLCO1B3 and SLCO1B1 and the LST-3TM12 pseudogene in the R2 haplotype (Figure B and Supplemental Figure 3).
RS families display deficiencies in SLCO1B1 and SLCO1B3.
Sequence analysis revealed predictably pathogenic mutations affecting both SLCO1B3 and SLCO1B1 in each of the haplotypes (Figure , B–D, Table , Supplemental Figure 3, and Supplemental Table 3). In the R1 haplotype, a 7.2-kb deletion removes exon 12 of SLCO1B3, encoding amino acids 500–560 of OATP1B3 (702 aa long) and introduces a frameshift and premature stop codon, thus removing the C-terminal 3 transmembrane domains. Furthermore, a nonsense mutation in exon 13, c.1738C→T, introduces a premature stop codon (p.R580X) in R1-linked OATP1B1 (691 aa long), removing the C-terminal one-and-a-half transmembrane domains. The 405-kb R2 deletion encompasses exons 3–15 of SLCO1B3 (sparing only a small N-terminal region) and the whole of SLCO1B1, but not SLCO1A2. The R3 haplotype harbors a splice donor site mutation, c.1747+1G→A, in intron 13 of SLCO1B3. If SLCO1B3 is still yielding functional mRNA, this would truncate OATP1B3 after amino acid 582, deleting the C-terminal one-and-a-half transmembrane domains. A nonsense mutation, c.757C→T, in exon 8 of R3-linked SLCO1B1 introduces a premature stop (p.R253X), truncating OATP1B1 before the C-terminal 7 transmembrane domains. All of these mutations would severely disrupt or annihilate proper protein expression and function. Moreover, they all showed consistent autosomal recessive segregation with the RS phenotype in the investigated families (Table ). No SLCO1A2 sequence variation was found in probands representing the 3 haplotypes, rendering involvement of OATP1A2 in RS unlikely. The severity of the identified mutations affecting SLCO1B3 and SLCO1B1 and their strict cosegregation with the RS phenotype indicate that RS is caused by co-inherited complete functional deficiencies in both OATP1B3 and OATP1B1.
Mutations in SLCO1B genes detected in RS subjects and their family members
The severity of the mutations was independently supported by immunohistochemical studies of the sparse RS liver biopsy material available. Given their sparseness, immunostaining of these liver biopsies was performed using one antibody recognizing the N terminus of both OATP1B1 and OATP1B3 (31
). This revealed absence of detectable staining in probands representing each haplotype (Figure ). In controls, basolateral membranes of centrilobular hepatocytes stained crisply, as previously reported (31
). Thus, the SLCO1B1
mutations in each haplotype result in absence of a detectable signal for OATP1B protein in the liver.
Liver expression of OATP1B proteins in RS subjects and control.
In family A2, a heterozygous splice donor site mutation, c.481+1G→T, in intron 5 of SLCO1B1 would result in dysfunctional RNA or protein. Its co-occurrence with the 405-kb R2 deletion in two asymptomatic family members (Table ) indicates that a single functional SLCO1B3 allele can prevent RS.
A search for copy number variations (CNVs) in existing databases and CNV genotyping of more than 2,300 individuals from various populations (see Supplemental Results) revealed additional heterozygous small and large deletions predicted to disrupt SLCO1B1 or SLCO1B3 function, including several approximately 400-kb deletions similar or identical to the R2 haplotype–linked deletion. One individual without jaundice, heterozygous for the R1 haplotype–associated c.1738C→T (p.R580X) mutation in SLCO1B1, was also homozygous for the R1 haplotype–associated deletion in SLCO1B3. Thus, a single functional SLCO1B1 allele can also prevent RS. Combined with the findings in family A2 described above, this demonstrates that only a complete deficiency of both alleles of SLCO1B1 and SLCO1B3 will result in RS.