|Home | About | Journals | Submit | Contact Us | Français|
Mutations of the orphan transporter ABCC6 (ATP-binding cassette, subfamily C, member 6) cause the connective tissue disorder pseudoxanthoma elasticum. ABCC6 was thought to be located on the plasma membrane of liver and kidney cells.
Mouse systems genetics and bioinformatics suggested that ABCC6 deficiency affects mitochondrial gene expression. We therefore tested whether ABCC6 associates with mitochondria.
We found ABCC6 in crude mitochondrial fractions and subsequently pinpointed its localization to the purified mitochondria-associated membrane fraction. Cell-surface biotinylation in hepatocytes confirmed that ABCC6 is intracellular. Abcc6-knockout mice demonstrated mitochondrial abnormalities and decreased respiration reserve capacity.
Our finding that ABCC6 localizes to the mitochondria-associated membrane has implications for its mechanism of action in normal and diseased states.
ABCC6 (ATP-binding cassette, subfamily C, member 6) is an orphan transporter expressed primarily in liver and kidney.1 Human ABCC6/MRP6 mutations cause pseudoxanthoma elasticum (PXE),2 a heritable recessive disorder characterized by calcification of elastin fibers in connective tissue, including in the heart, vasculature, skin, and eye.3 Mice with Abcc6 mutations recapitulate the calcification patterns.4 Because ABCC6 is absent or minimally expressed in calcified tissues, it is thought that the ABCC6 substrate mediates calcification via the circulation. Accordingly, parabiotic combination of blood circulation between Abcc6−/− and wild-type mice rescued vascular calcification.5
From systems genetics data, we hypothesized that this transporter influenced mitochondrial function. We now report that ABCC6 localizes to the mitochondria-associated membrane (MAM) rather than, as previously thought, the basolateral plasma membrane of liver and kidney. Given the major role that the MAM plays in calcium signaling, lipid metabolism, and secreted protein processing, our findings may lead to new approaches for establishing the connection between ABCC6 and tissue calcification.
Experiments were conducted in accordance with guidelines by the National Research Council and were approved by the University of California, Los Angeles. C57BL/6J and C3H/HeJ mice were purchased from the Jackson Laboratory. Abcc6-knockout (KO) mice were from Dr AAB Bergen.4 Abcc6-Tg mice were generated as described.6 Mitochondria-associated membranes were isolated by ultracentrifugation as described.7 The anti-ABCC6 polyclonal antibody (S-20) (SCBT) recognizes the N-terminus of mouse ABCC6.
We used a Hybrid Mouse Diversity Panel comprised of 100 inbred mouse strains8 to identify Abcc6 gene networks. We found that liver transcripts whose expression significantly correlated with Abcc6 expression (r>0.390) were highly enriched for mitochondrial genes (Online Table I). Likewise, when we compared liver and kidney transcripts from mice with nonfunctional ABCC6 versus functional ABCC6 we, again, found an enrichment for mitochondrial genes (Online Table II). We therefore investigated the subcellular location of the protein. We compared the C3H inbred mouse strain, which has an endogenous Abcc6 splice site mutation resulting in reduced transcripts and nonfunctional proteins,9 to C3H Abcc6 transgenic mice (Abcc6-Tg) and C57BL/6 Abcc6-wildtype (WT) to Abcc6-KO mice on the C57BL/6 background (Online Table III). Both C3H and Abcc6-KO mice are Abcc6-deficient and have extensive vascular calcification and other pathologies resembling PXE,4 which are rescued in Abcc6-Tg mice.6
We first conducted subcellular fractionation using a sucrose gradient and observed strong ABCC6 signals in immunoblots of mitochondrial fractions in both the liver (Figure 1A) and kidney (Online Figure I) of Abcc6-Tg mice. ABCC6 signals were lower in C3H as expected. The mitochondrial isolation was validated by immunoblotting for compartment-specific markers (Figure 1A). Most proteins were enriched in their known subcellular location; however, the endoplasmic reticulum (ER) marker, protein disulfide isomerase, was also present in the mitochondrial fraction, suggesting further purification was needed to eliminate the possibility that ABCC6 was an ER protein. Before proceeding, we confirmed the specificity of the ABCC6/MRP6 (S-20) antibody.10 In our experiments, the antibody recognized a single predominant band of ≈160 kDa. The signal intensity was highest for Abcc6-Tg, present in WT, significantly lower in C3H, and absent in Abcc6-KO mice (Figure 1B). Moreover, the signal in Abcc6+/− heterozygotes was approximately half of that in WT (Online Figure II). The immunoreactive ABCC6 protein was N-glycosylated (Online Figure III), characteristic of full-length ABCC6 expressed in mammalian cells.11 We therefore concluded that the ≈160-kDa band in the mitochondrial fraction corresponded to ABCC6.
To corroborate the subcellular fractionation findings, we investigated purified mitochondria using confocal microscopy. The isolated kidney crude mitochondrial fraction was Percoll-purified to reduce microsomal contamination. ABCC6 immunofluorescence showed clear colocalization with functional mitochondria from WT kidney (Figure 1C). The specificity of immunofluorescence was confirmed by the absence of ABCC6 signal in Abcc6-KO mitochondria. Notably, ABCC6 was detected in a subset of mitochondria, as opposed to the respiratory chain complex subunit COXIV (Online Figure IV). In ultrahigh resolution images, ABCC6 was visible as distinct clusters of ≈20 and ≈40 nm (Figure 1D).
The detection of ABCC6 in a subset of mitochondria could be due to Abcc6 expression in specific cell types or in a subpopulation of cellular mitochondria. To address the latter possibility, we examined the MAM, a specialized cellular compartment that bridges the ER with some but not all mitochondria in the cell.12 The MAM and mitochondrial outer membrane couple through protein-protein interactions but constitute separate lipid layers. Because the MAM was not completely removed in our immunofluorescence experiment, we further separated the MAM and other microsomal contamination by ultracentrifugation.7 Immunoblotting of the resulting fractions revealed that ABCC6 resides in the MAM fraction (Figure 2A). This fraction contained minimal to no contamination from mitochondria, cytosol, and caveolae. Remarkably, ABCC6 was more enriched than known MAM markers including calnexin.7 ABCC6 was undetectable in the highly purified mitochondrial fraction and only faintly visible in pure non-MAM ER.
Previous studies using immunohistochemical staining have suggested that ABCC6 localizes to the plasma membrane.1,10 To rule out the possibility of dual ABCC6 localization to MAM and plasma membrane, we biotin-labeled cell surface proteins of primary hepatocytes from WT mice. Following affinity precipitation with streptavidin, the cell surface marker pan-cadherin was retained, but not ABCC6, which associated entirely with the intracellular fraction (Figure 2B). The result indicates the majority of ABCC6 protein is intracellular.
We previously showed that C3H and Abcc6-KO mice suffer increased cardiac necrosis following cardiac ischemia-reperfusion.13 To investigate the possible mitochondrial contribution to this result, we used electron microscopy to image liver, kidney, and heart tissue. Liver electron microscopy sections did not reveal mitochondrial differences but showed a reduction in ER bordering the mitochondria (Figure 3A). Most mitochondria in Abcc6-KO liver were surrounded by a single strand of ER, but the extensive ER clusters between mitochondria, present in the WT, were absent (Online Figure V). In the Abcc6-KO kidney proximal tubule, >40% of mitochondria had virtually no cristae (Figure 3A and Online Figure VI), compared to ≈10% mitochondria in WT. Heart cross-sections displayed striking abnormalities of the cristae (Figure 3A), along with inconsistent size and numbers of mitochondria per sarcomere (Online Figure VII).
The morphological defects in Abcc6-KO kidney and heart prompted us to measure mitochondrial respiration. Normal mitochondria responded to the uncoupler FCCP with a sharp increase in oxygen consumption rate (respiratory reserve), whereas Abcc6-KO liver, kidney, and heart mitochondria all demonstrated significantly lower response on uncoupling (Figure 3B). These findings corroborate the mitochondrial dysmorphology in the kidney and heart and suggest functional defects in hepatic mitochondria despite inapparent structural abnormalities.
Our microarray results suggested that of the known MAM functions, protein processing and purine nucleotide activity were the most impacted by ABCC6 deficiency (Online Table IV). Calcification pathway proteins, protein C, ENPP1, and GGCX, a key protein involved in vitamin K-mediated calcification, were differentially expressed.
We report evidence from genetics, subcellular fractionation, and cellular morphology that the PXE disease gene Abcc6 encodes an intracellular transporter associated with mitochondrial function. Determining the subcellular localization of ABCC6 is a significant step toward finding the mechanism and substrate of the transporter. We and others have previously searched for the ABCC6 substrate in the plasma. However, ABCC6 in the MAM suggests that mitochondria, ER, or cytosol may be the critical location of the ABCC6 substrate. Vitamin K2 has been described as an electron carrier involved in mitochondrial functions,14 and ABCC6 could contribute to vitamin K2-mediated electron transport in mitochondria along with γ-carboxylation in the ER.
Decreased maximal respiratory capacity is often due to disruption of the mitochondrial proton gradient, which in turn generates increased reactive oxygen species. This model is consistent with increased oxidative stress in Abcc6-deficient animals and humans.15 Oxidative conditions at the subcellular level in PXE patients, because of mitochondrial defects, may disrupt ER processing of lipids and proteins, possibly producing key calcification mediators. Also, proteins not folded or processed properly could trigger the unfolded protein response and initiate apoptotic signaling. Finally, because the ER and mitochondria are both intracellular calcium stores that exchange through the MAM, faulty calcium transport may lead to downstream apoptotic signals in target cells or to calcium efflux into the cytosol. With the finding that ABCC6 is located in the MAM, clues for substrate identity can now be sought in ER and mitochondrial signaling pathways.
Electron microscopy was performed at the Electron Microscopy Services Center of UCLA Brain Research Institute.
Sources of Funding
This work was supported by funding from the NIH (HL30568, S10RR026744, HL088640, and HHSN268201000035C) and American Heart Association (11SDG7230059, 11POST7300060, and 12PRE11610024).
Brief UltraRapid Communications are designed to be a format for manuscripts that are of outstanding interest to the readership, report definitive observations, but have a relatively narrow scope. Less comprehensive than Regular Articles but still scientifically rigorous, BURCs present seminal findings that have the potential to open up new avenues of research. A decision on BURCs is rendered within 7 days of submission.
The online-only Data Supplement is available with this article at http://circres.ahajournals.org/lookup/suppl/doi:10.1161/CIRCRESAHA.112.276667/-/DC1.