ABCC6–Mediated ATP Secretion by the Liver Is the Main Source of the Mineralization Inhibitor Inorganic Pyrophosphate in the Systemic Circulation—Brief Report

Abstract

Objective—

Mutations in ABCC6 underlie the ectopic mineralization disorder pseudoxanthoma elasticum (PXE) and some forms of generalized arterial calcification of infancy, both of which affect the cardiovascular system. Using cultured cells, we recently showed that ATP-binding cassette subfamily C member 6 (ABCC6) mediates the cellular release of ATP, which is extracellularly rapidly converted into AMP and the mineralization inhibitor inorganic pyrophosphate (PP_i). The current study was performed to determine which tissues release ATP in an ABCC6-dependent manner in vivo, where released ATP is converted into AMP and PP_i, and whether human PXE ptients have low plasma PP_i concentrations.

Approach and Results—

Using cultured primary hepatocytes and in vivo liver perfusion experiments, we found that ABCC6 mediates the direct, sinusoidal, release of ATP from the liver. Outside hepatocytes, but still within the liver vasculature, released ATP is converted into AMP and PP_i. The absence of functional ABCC6 in patients with PXE leads to strongly reduced plasma PP_i concentrations.

Conclusions—

Hepatic ABCC6-mediated ATP release is the main source of circulating PR, revealing an unanticipated role of the liver in systemic PP_i homeostasis. Patients with PXE have a strongly reduced plasma PP_i level, explaining their mineralization disorder. Our results indicate that systemic PP_i is relatively stable and that PXE, generalized arterial calcification of infancy, and other ectopic mineralization disorders could be treated with PP_i supplementation therapy.

Pseudoxanthoma elasticum (PXE) is an autosomal recessive disease characterized by progressive ectopic mineralization of the skin, eyes, and arteries.¹ Approximately 150 000 patients with PXE worldwide experience stigmatizing skin lesions, progressive loss of vision, and cardiovascular complications, against which no effective therapy exists.²

In 2000, several groups reported that PXE is caused by inactivating mutations in the ATP-binding cassette subfamily C member 6 (ABCC6) gene,^3–5 and more recently, ABCC6 defects were also found to cause some forms of generalized arterial calcification of infancy (GACI),⁶ a severe form of arterial calcification. ABCC6 (also known as multidrug resistance protein 6) is an ATP-dependent orphan efflux transporter that is primarily expressed in the liver.⁷ Importantly, PXE is not caused by a lack of ABCC6 in the affected tissues but by the absence of an unknown factor in the central circulation requiring active ABCC6.⁸ Despite extensive research, the identity of this factor has long remained a mystery.

We recently showed that overexpression of ABCC6 in human embryonic kidney 293 (HEK293) cells induces the release of nucleoside triphosphates, predominantly ATP, in vitro.⁹ Secreted ATP was extracellularly converted into AMP and the mineralization inhibitor inorganic pyrophosphate (PP_i) by ectonucleotide pyrophosphatase-phosphodiesterase (ENPP)–type ectonucleotidases. The in vivo relevance of these findings was demonstrated in Abcc6^−/− mice, which have plasma PP_i levels <40% of those found in wild-type control animals. ABCC6 is a member of the ABCC (multidrug resistance protein) family, which contains large proteins transporting a variety of organic anions.¹⁰ ABCC6 is mainly present in the sinusoidal membrane of the hepatocytes.¹¹ Because we could not demonstrate direct ABCC6-mediated ATP transport in vitro, we postulated that ABCC6 secretes an organic anion, factor X, into the circulation that induces local ATP release in the periphery.⁹ The alternative possibility that the liver directly releases ATP in an ABCC6-dependent manner seemed unlikely. Secretion of ATP over the sinusoidal membrane of hepatocytes has never been described, and the extremely short half-life of ATP in the blood circulation (<1 seconds)¹² does not allow PP_i formation from liver-derived ATP in the periphery. The current study was performed to show that ABCC6 affects plasma PP_i levels in humans and to assess whether ABCC6 directly affects hepatic ATP release or indirectly induces peripheral ATP release.

Materials and Methods

Materials and Methods are available in the online-only Supplement.

Results

HEK293 and HeLa Cells Release ATP on the Expression of ABCC6

We have previously shown that the introduction of ABCC6 in HEK293 cells results in the release of large amounts of ATP into the culture medium.⁹ To determine whether ABCC6-dependent ATP release is cell type-dependent, we generated HeLa cells in which the expression of rat ABCC6 could be induced by doxycycline. A luciferin/luciferase-based assay was used to follow the appearance of ATP in the cell culture medium in real-time. In the absence of rat ABCC6, cells released almost no ATP (Figure 1A and 1B). However, on induction of rat ABCC6, both 293 and HeLa cells released substantial amounts of ATP into the cell culture medium (Figure 1). These data show that ATP release is a general feature of ABCC6-containing cells and not specific for HEK293 cells.

Figure 1.

HEK293 and HeLa cells overproducing rat ATP-binding cassette subfamily C member 6 (rABCC6) release ATP. A, Flp-In T-REx 293 control (squares) or Flp-In T-REx 293 rABCC6 (circles) cells were grown in the presence (filled symbols) or absence (open symbols) of 1 μg/mL doxycycline to induce rABCC6 expression. Two days later, ATP efflux was followed in real-time for 2 hours using the ATP detection reagent BactiterGlo. B, ATP efflux from Flp-In T-REx HeLa control (squares) or Flp-In T-REx HeLa rABCC6 (circles) cells grown in the presence (filled symbols) or absence (open symbols) of 1 μg/mL doxycycline was followed for 2 hours in real-time. Data (n=12) represent mean±SEM.

Appearance of PP_i in the Culture Medium of Sandwich-Cultured Hepatocytes Depends on ABCC6

ABCC6 is predominantly present in the liver.¹¹ Next, we therefore explored in sandwich-cultured hepatocytes the possibility that hepatocytes directly release ATP over their basolateral membrane in an ABCC6-dependent manner. We were unable to detect ATP release directly in these experiments, presumably because of the high ectonucleotidase activity of hepatocytes. We, therefore, followed the appearance of the ATP metabolite PP_i in the culture medium. PP_i levels clearly increased in culture medium of wild-type hepatocytes over time, with substantially lower levels detected in medium of hepatocytes lacking ABCC6 (Figure 2A). These results indicate that hepatocytes release ATP over their sinusoidal membrane in an ABCC6-dependent manner and are also able to convert it to PP_i. We also detected some PP_i in medium from Abcc6^−/− cells, which we attribute to ATP release unrelated to ABCC6, or leakage from damaged cells.

Figure 2.

Hepatic ATP-binding cassette subfamily C member 6 (ABCC6) raises inorganic pyrophosphate (PP_i) levels via ATP release. Released ATP is rapidly converted into AMP and PP_i within the liver vasculature. A, PP_i levels in culture medium of sandwich-cultured primary wild-type (WT) and Abcc6^−/− hepatocytes (n=3 for WT, n=4 for Abcc6^−/−); total amount of (B) PP_i, (C) AMP, and (D) ATP in mouse liver perfusates collected from WT and Abcc6^−/− livers during 30 minutes (n=5 for WT, n=6 for Abcc6^−/−). E, PP_i levels in platelet-free plasma samples from healthy subjects (n=14) and patients with PXE (n=12). Patient and control characteristics are given in the online-only Data Supplement. Data are presented as mean±SD. **P<0.01, ***P<0.001. Note that AMP and PP_i levels are in nmoles, whereas ATP levels are in pmoles and close to background levels.

Hepatic ABCC6 Mediates the Sinusoidal Release of ATP, Which Is Converted Into AMP and PP_i Within the Liver Vasculature

To assess whether ABCC6 is an important factor in hepatic ATP release in vivo, we performed liver perfusion experiments. PP_i and AMP levels in the liver perfusates strongly depended on the presence of ABCC6 (Figure 2B and 2C). Interestingly, ATP levels did not differ between the 2 genotypes and were extremely low, representing <1% of the PP_i and AMP levels (Figure 1D). The AMP and PP_i that we detect in the liver perfusates must be derived from ATP: Enpp1^−/− mice have pp. levels that are <5% of those found in wild-type mice,¹³ implying that also the PP_i in plasma that depends on ABCC6 must come from ATP. Conversion of released ATP into AMP and PP_i within the liver is fast. We calculated that during our single-pass perfusion experiments the buffer is present in the liver for ≈10 seconds (for the calculation, see the Materials and Methods section in the online-only Data Supplement). During this short period, the substantial amounts of ATP released are almost quantitatively converted into PR and AMP (Figure 2B–2D). This rapid and efficient conversion also explains why we were unable to detect ATP release from cultured wild-type hepatocytes: any released ATP is almost instantaneously converted into AMP and PP_i by hepatic NPP1.

From our perfusion experiments, we calculate that ABCC6 mediates ≈90% of the hepatic nucleotide release. During 24 hours, this corresponds to ≥5% of the total hepatic adenine nucleotide pool (Figure 2B; for the calculation, see the Materials and Methods section in the online-only Data Supplement). The plasma t_1/2 of PP_i has been estimated to be 33 minutes, which requires a hepatic release rate of 6 nmoles pp. per hour to achieve the steady state levels of 2.3 μmol/L that we have reported for mice⁹ (for calculation, see the Materials and Methods section in the online-only Data Supplement). Importantly, the amount of PP_i detected in liver perfusates of wild-type mice is high enough to explain these steady state PP_i levels in mouse plasma.

Patients With PXE Have Strongly Reduced PP_i Plasma Levels

An important question is whether our mouse results translate to human PXE patients. We have, therefore, studied a group of 12 Dutch patients with PXE with known ABCC6 mutations (Table I in the online-only Data Supplement). The plasma PP_i concentrations were ≈2.5-fold lower in patients than in healthy individuals (Figure 2E). This difference did not depend on sex and is in line with the reduced plasma PP_i levels we previously reported for Abcc6^−/− mice.⁹

Discussion

PP_i is a key regulator of ectopic mineralization acting by inhibiting hydroxyapatite crystal growth.¹⁴ As a result, mutations in genes encoding known PP_i-regulating enzymes like ENPP1, ecto-5′-nucleotidase, progressive ankylosis protein homolog, and tissue-nonspecific alkaline phosphatase (TNAP) cause various mineralization disorders.^15–18 The clinical symptoms of the mineralization disorders caused by non-functional ENPP1 (GACI) and ecto-5′-nucleotidase (arterial calcification due to deficiency of CD73) highly overlap those of PXE.¹⁹ The similarity between GACI and PXE is underlined by the recent observations that both GACI and PXE can be caused by mutations in ENPP1, as well as ABCC6.⁶ Our data unexpectedly falsify the factor X-hypothesis⁹ and show that ABCC6-mediated ATP release from the liver is the principal source of plasma PP_i. A factor involved in the local release of PP_i is progressive ankylosis protein homolog, a membrane protein postulated to mediate the direct release of PP_i from cells. Progressive ankylosis protein homolog does, however, not substantially contribute to plasma PP_i levels, which almost exclusively depend on ENPP1 activity and hence ATP release.¹³ Based on the currently available data, we propose the model presented in Figure 3.

Figure 3.

Proposed model for hepatic ATP-binding cassette subfamily C member 6 (ABCC6)–mediated pyrophosphate generation and ectopic mineralization. ATP released from the liver by an ABCC6-dependent mechanism is converted into the mineralization inhibitor pyrophosphate (inorganic pyrophosphate [PP_i]) by hepatic ectonucleotide pyrophosphatase-phosphodiesterase 1 (ENPP1). In the periphery, PP_i is hydrolyzed by tissue-nonspecific alkaline phosphatase (TNAP). Inactive ABCC6 classically causes pseudoxanthoma elasticum (PXE), whereas inactive ENPP1 causes generalized arterial calcification of infancy (GACI). Nonfunctional ecto-5′-nucleotidase results in arterial calcification due to deficiency of CD73 (ACDC), and inactive TNAP causes hypophosphatasia (HOPS). Local PP_i levels also depend on the transmembrane protein progressive ankylosis protein homolog (ANKH), a protein postulated to be a PP_i channel/efflux transporter. Mutations in ANKH can result in chon-drocalcinosis type 2 (CCAL2) or craniometaphyseal dysplasia (CMD).

Our finding that PP_i generated within the liver is able to act in the periphery shows that increased systemic PP_i levels are sufficient to inhibit local ectopic mineralization. Importantly, Lomashvili et al¹³ recently showed in Enpp1^−/− mice that ectopic calcification depends on plasma PP_i levels and not local PP_i production. The crucial role of plasma PP_i in the prevention of ectopic calcification has important therapeutic consequences: raising PP_i levels in the blood circulation of patients with PXE, GACI, and arterial calcification due to deficiency of CD73 should suffice to halt ectopic mineralization. The short plasma half-life and lack of a suitable dosage form do not make PP_i an attractive candidate for supplementation therapy in humans,²⁰ but it might be possible to generate suitable PP_i precursors. Alternatively, bisphosphonates, a class of metabolically stable, synthetic PP_i analogs that have been used in GACI with reasonable success,²¹ may represent an attractive treatment strategy for PXE and arterial calcification due to deficiency of CD73.

The AMP metabolite adenosine is known to inhibit the expression of TNAP (Figure 3).¹⁶ It is, therefore, tempting to speculate that the increased TNAP activity seen in fibroblasts isolated from patients with PXE²² and Abcc6^−/− mice²³ is because of a reduction in the amount of released AMP. Low AMP levels might reduce local formation of adenosine and subsequent TNAP inhibition. AMP-derived adenosine might, therefore, be involved in priming of the periphery for subsequent PP_i influx. This model would imply that both AMP and PP_i are necessary to prevent ectopic mineralization: PP_i by directly inhibiting the formation of calcium phosphate crystals and AMP after being metabolized to adenosine by inhibiting premature degradation of circulating PP_i by TNAP.

In vitro, ABCC6 transports glutathione conjugates and the synthetic cyclic peptide BQ-123, suggesting that ABCC6 is a bona fide transporter.^11,24 We were unable, however, to demonstrate ABCC6-mediated nucleoside triphosphate transport in vesicular transport experiments.⁹ Factors could be missing in vitro, however, that allows ABCC6 to transport ATP in vivo, or ABCC6 could indirectly stimulate ATP release by regulating vesicular transport or ion channels.²⁵

Taken together, we show that ABCC6 mediates the release of ATP directly from the liver into the circulation. Within the liver vasculature, ATP is converted into AMP and PP_i and represents the main source of the mineralization inhibitor PP_i in plasma. This fully explains why absence of ABCC6 results in the ectopic mineralization observed in patients with PXE. Our data indicate that correcting PP_i to normal levels could prevent the ectopic mineralization observed in PXE, GACI, and arterial calcification due to deficiency of CD73.

Significance.

Pseudoxanthoma elasticum is a hereditary ectopic mineralization disorder caused by the absence of functional ATP-binding cassette subfamily C member 6 that affects ≈150 000 patients worldwide. An effective therapy does not exist because the pathology underlying the disease is not well understood. Here, we show that ATP-binding cassette subfamily C member 6-mediated ATP secretion by the liver is the main source of the mineralization inhibitor inorganic pyrophosphate in the systemic circulation, explaining the ectopic calcification observed in patients with pseudoxanthoma elasticum. Our data indicate that correcting inorganic pyrophosphate to normal levels could prevent the ectopic mineralization observed in pseudoxanthoma elasticum and related mineralization disorders.

Nonstandard Abbreviations and Acronyms

ABCC6

ATP-binding cassette subfamily C member 6

ENPP

ectonucleotide pyrophosphatase-phosphodiesterase

GACI

generalized arterial calcification of infancy

PP_i

inorganic pyrophosphate

PXE

pseudoxanthoma elasticum

TNAP

tissue-nonspecific alkaline phosphatase