An Alternative Splice Variant in Abcc6, the Gene Causing Dystrophic Calcification, Leads to Protein Deficiency in C3H/He Mice*

Dystrophic cardiac calcification (DCC) is an autosomal recessive trait characterized by calcium phosphate deposits in myocardial tissue. The Abcc6 gene locus was recently found to mediate DCC; however, at the molecular level the causative variants remain to be determined. Examining the sequences of Abcc6 cDNA in DCC-resistant C57BL/6 and DCC-susceptible C3H/He mice, we identified a missense mutation (Cys to Thr at codon 619, rs32756904) at the 3′-border of exon 14 that creates an additional donor splice site (GT). Accordingly, an alternative transcript variant was detected, lacking the last 5 bp of exon 14 (-AGG(C/T)GCTgtga-) in DCC-susceptible C3H/He mice that carry the Thr allele. The 5-bp deletion was found to result in premature termination at codon 684, in turn leading to protein deficiency in DCC-susceptible mouse tissue as well as in cells transfected with Abcc6 cDNA lacking the last 5 bp of exon 14. All mouse strains that were found to carry the Thr allele, including C3H/He, DBA/2J, and 129S1/SvJ, were also found to be positive for DCC. In summary, we identified a splice variant leading to a 5-bp deletion in the Abcc6 transcript that gives rise to protein deficiency both in vivo and in vitro. The fact that all mouse strains that carry the deletion also develop dystrophic calcifications further suggests that the underlying splice variant affects the biological function of MRP6 protein and is a cause of DCC in mice.

Dystrophic cardiac calcifications (DCC) 2 are calcium phosphate deposits outside osseous tissue that occur independently from calcium and phosphate homeostasis. In western countries such calcification can affect the arterial system in patients with atherosclerosis, diabetes mellitus, and chronic renal failure. Interestingly, similar deposits are also observed in pseudoxan-thoma elasticum (PXE), a heritable disorder of the connective tissue that prominently affects the arteries of skin, eye, and heart.
Based on gene mapping data, recent studies identified Abcc6 as causative gene for both DCC and PXE, in mice and in human, respectively. The genetic basis of DCC was initially facilitated by the identification of a major QTL locus named Dyscalc1 on mouse chromosome 7 that contributes to DCC (12). We ultrafine-mapped this locus to an 80-kb region on proximal mouse chromosome 7 that contains only two known genes, epithelial membrane protein-3 (EMP-3) and the ATP binding cassette C 6 (Abcc6), and one gene of unknown function (BC013491) using an in silico mapping strategy (13). In parallel, the locus for PXE was identified on human chromosome 16p13.1 to a 500-kb region that also includes Abcc6 (14,15). Subsequently, using knock-out and transgenic mouse models, the Abcc6 gene was demonstrated to cause both DCC and PXE (16 -18).
Abcc6 belongs to the large ABC family containing nearly 48 genes. ABC proteins bind and hydrolyze ATP to meet the energy requirement for transportation of various molecules across the plasma membrane. Similar processes may take place at intracellular membranes of the endoplasmic reticulum, peroxisome, and mitochondria (19). ABC proteins are involved in the transport or removal of toxic metabolites using substrates such as glutathione (20). The Abcc6 gene encodes a 165-kDa protein named MRP6. It is predominantly expressed in the liver and to a lesser extent in the kidney. Interestingly, these tissues are not predominantly affected by either DCC or PXE (21,22). Here we present an approach for identification and functional characterization of a primary genetic variant causing DCC in mice.

EXPERIMENTAL PROCEDURES
Animal Housing and Histological Analysis-Animal studies were performed in accordance with the German animal studies committee of Schleswig-Holstein. Female mice from C57BL/6J (C57) and C3H/HeJ (C3H) inbred strains were purchased from Charles River Laboratories.
Five mice in each group were sacrificed at age 6 -8 months under anesthesia by cervical dislocation. Tissues were prepared as previously described (13). Slides were stained using alizarin red S and calcein stains for analysis of calcium phosphate deposits (13,23).
Reverse Transcriptase PCR and mRNA Quantification-Total RNA was extracted from liver and reverse-transcribed into complementary DNA (cDNA) as previously described (13). Changes in mRNA levels were determined using the ⌬⌬Ct method as previously reported (11,13).
Abcc6 cDNA Sequencing-Primer pairs (Abcc6 1-10) covering the 5Ј-UTR, the coding region, and the 3Ј-UTR of the Abcc6 gene (Ensembl Transcript ID ENSMUST00000002850) were designed on-line (Table 1). Direct sequencing was performed as described previously (13) on PCR fragments amplified from liver cDNA in C57 and C3H mice. Sequencing of PCR products was performed on both strands by a commercial sequencing service (Seqlab, Goettingen, Germany).
PCR Amplification and Electrophoresis-An Eppendorf MasterMix TM (2.5ϫ) containing dNTP, buffer, and polymerase was used to amplify Abcc6 cDNA fragments following the instruction of the manufacturer. Amplicons were analyzed after electrophoresis on a 1% QA-Agarose TM gel (Qbiogene) and Sybr Green I staining. For the amplification of the PCR product flanking the 5-bp deletion, we used primer pairs that produce a 156-bp PCR fragment (Table 1, del-5bp-ex14). For the separation of the 5-bp deletion fragment, high resolution electrophoresis was performed overnight using a 4% MetaPhor gel (Biozym Scientific GmbH, Oldendorf, Germany) and ethidium bromide staining.
In Vitro Transcription and Translation-pSG5-Abcc6cDNA, pSG5-Abcc6-5bpdel-cDNA, and pSG5 empty vector constructs were transcribed and translated in vitro in the presence of [ 35 S]methionine (TNT Coupled Reticulocyte Lysate System; Promega) according to the manufacturer's protocol. [ 35 S]Methionine was obtained from Hartmann Analytic (Braunschweig, Germany). After incubation, 1 l of the reaction batch was dissolved in Laemmli buffer and proteins were separated by SDS-PAGE (10%). To detect the [ 35 S]-labeled proteins, the dried gels were autoradiographed (16 h). The gels were analyzed using the software PCBAS 2.09g (Raytest Isotopenmessgeräte GmbH).
Cell Culture and Cell Transfection-Human embryonal kidney cells (HEK-293) were kindly provided by Dr. Stefanie Stoelting (University of Luebeck), and the cells were grown in complete Dulbecco's modified Eagle's medium (1ϫ) with glucose 4.5 g/liter and L-glutamine-pyruvate, containing 15% fetal calf serum and penicillin/streptomycin (1ϫ) at 37°C and 5% CO 2 . One day prior to transfection, cells were plated at a density of 1 ϫ 10 5 into Lab-Tek 4-well glass chamber slides (Nalge Nunc Int.). The cells were transfected with DNA from pSG5, pSG5-Abcc6, and pSG5-Abcc6-5bpdel plasmids using nanofectin transfection reagent and following the instruction of the manufacturer (PAA Laboratorie GmbH). Two days after transfection, cells were fixed in 70% ethanol and analyzed immunohistologically. Similarly, cells were plated into 24-well plates and transfected with the three constructs for Western blot analysis.
Immunohistological Analysis-Immunostaining was performed with a polyclonal anti-mouse MRP6-S20 antibody at 5 g/ml (Santa Cruz Biotechnology, Inc.) as previously described (11). Briefly, bound MRP6 primary antibodies were detected using biotinylated secondary antibodies that were visualized using a streptavidin-horseradish peroxidase complex and diaminobenzidine as supplied with the Cell and Tissue Staining kit (R&D systems). Slides were then counterstained with hematoxylin. Controls were performed without primary antibodies.
Western Blot Analysis-Protein extract was prepared from the liver of mice following standard protocols. Liver tissue samples frozen in liquid nitrogen were homogenized using a mortar. Lysis buffer (1ϫ cell lysis buffer (Cell Signaling), 1ϫ Roche cocktail (Roche Diagnostics GmbH), and 1 mM Phenylmethane- sulfonyl fluoride (Sigma)) containing urea (8 M) was added, and samples were centrifuged. After protein quantification, 30 g were sampled and separated on a 6.5% SDS-PAGE and afterward electrotransferred to an Immobilon-P transfer membrane (Millipore, Bedford, MA). After blocking with 5% lowfat milk, membranes were incubated with the primary antibodies MRP6-S20 (Santa Cruz Biotechnology) and ␣-actin from (Abcam plc). The epitope of the polyclonal MRP6 has a length of ϳ15-25 aa and maps within aa 1-50 at the N terminus of MRP6 of mouse origin (Swiss Prot protein accession number Q9R1S7). For signal detection, the ECL-plus Western blotting detection system (RPN, 2132; GE Healthcare) was used. Chemiluminescence detection was performed with a Molecular Imager ChemiDoc XRS system (Bio-Rad).
Statistical Analysis-Data analyses were performed by Student's t test, and results are expressed as the means Ϯ S.E., with p Ͻ0.05 as significant.

MRP6 Protein Expression in Mice-Previous reports have
shown that the Abcc6 gene product MRP6 is highly expressed in the liver. To find evidence of an association or correlation between the level of MRP6 protein expression and the observed DCC phenotype in mice, we analyzed MRP6 expression in the liver of DCCsusceptible C3H and DCC-resistant C57 mice as well as congenic mice (Cg1). Western blot analysis revealed a dramatic decrease in MRP6 expression in the liver of both DCC-susceptible C3H and Cg1 mice compared with liver tissue from DCC-resistant C57 mice (Fig. 1,  A and B). With the MRP6 protein expression level in the liver of C57 mice defined as 100%, the expression levels were determined to be 34 Ϯ 12 and 33 Ϯ 17% in C3H and Cg1 mice, respectively.
Sequencing of Abcc6 cDNA-To search for novel genetic regulatory elements that might affect the Abcc6 gene expression and thus of the corresponding encoded protein MRP6, we sequenced the 5Ј-UTR, the coding region, and the 3Ј-UTR in the cDNA of Abcc6 from DCC-resistant C57 and DCC-susceptible C3H mice. Comparing the genomic sequences between the two inbred strains of mice, we FIGURE 1. MRP6 expression. Western blot analysis of MRP6, the encoding protein of Abcc6. A, dramatically reduced MRP6 expression is demonstrated in the liver of C3H mice as well as congenic (Cg1) mice in comparison to C57 mice. 30 g of total protein was loaded in each lane. ␣-Actin was used as loading control. B, quantification of the level of MRP6 expression from Western blots (n ϭ 5) in each mouse strain, C57, C3H, and Cg1, after normalization to ␣-actin. The band intensity and size was determined using the software Quantity Oneா (Bio-Rad). The expression of MRP6 in the wild type C57 mice is set as 100%. Error bars indicate S.E.

TABLE 2 Genomic sequence analysis of Abcc6 and its respective SNP genotypes repartition on DCC-resistant C57 and DCC-susceptible C3H mouse strains
SNPs and deletion (del) are represented in parentheses.

Position
SNPs Flanking sequence Amino acid substitution Position in cDNA  (Table 2). Of these, 25 were known and one SNP was not published previously. Among these SNPs nine resulted in amino acid substitutions (p.A28V, p.V95M, p.S138A, p.I151V, p.R619S, p.V706A, p.K767R, p.T927I, p.A1368T). In addition, we identified two novel small deletions within this region. The 3Ј-UTR of DCC-susceptible C3H mice was found to contain a 10-bp deletion (GAGCA(TCACAC-CGAC/del)TCTGA), which is located in a region important for mRNA stability. C3H mice were also found to be heterozygous for another novel 5-bp deletion that was identified at the 3Ј-border of exon 14.
Alternative Splice Variant in DCC-susceptible C3H Mice-To confirm the above mentioned heterozygosity for the 5-bp deletion in exon 14, we designed new primer pairs flanking this region (Table 1, del-5bp-ex14 F/R), producing an ϳ150-bp PCR product, suitable for separating on a high resolution 4% MetaPhor gel. Fig. 2 shows three bands in C3H mice (Fig. 2A, lane C3H, arrows): the 5-bp deletion (Mut), the wild type (Wt), and an extra band (Hd). In contrast, the C57 mice contain only one band ( Fig. 2A, lane C57). To further analyze the bands in C3H mice, we subcloned the PCR product using the  TOPO TA Cloning kit and screened insert clones for differences in insert size. We found inserts with two different sizes ( Fig. 2A, lanes C-1 and C-2), the wild type (C-1) and the 5-bp deletion (C-2). We then sequenced the two inserts after PCR amplification using the M13 universal primers (C-1 and C-2). We did not find clones with an insert that corresponds in size to the third band. So we hypothesized the third band ( Fig. 2A, Hd) to be a heteroduplex of the two fragments, the wild type and the 5-bp deletion variants. To test this hypothesis, we ran a PCR on a mixture of plasmid DNA from the two clones, C-1 and C-2, that carry either the wild type sequence or the sequence with the 5-bp deletion. Fig. 2B shows the extra band (Hd) present in this mixture. p.R619S Mutation Creates an Alternative Splice Variant-To find the genomic mutation creating the alternative splice variant, we further analyzed the genomic sequence variation that had been found between C57 and C3H mice in exon 14 (ex14) and identified a single base pair mutation (C/T) within the 5-bp deletion in ex14 (-AGG(C/T)GCTgtga-). In C3H mice, this mutation leads to an amino acid substitution at position 619 (p.R619S). Interestingly, in pre-mRNA of C3H mice, this mutation creates an additional splice donor site (GU) at codon 619, thus effectively resulting in deletion of the last 5 bp (GUGCU) of exon 14 (Fig. 3). This deletion creates a premature stop codon at codon 685, in turn giving rise to a truncated protein of only 684 aa in C3H mice instead of 1449 aa in C57 mice (Fig. 4).
In Vitro Overexpression of the Truncated Protein from a Construct Carrying the 5-bp Deletion-To examine whether the Abcc6-del5bp mRNA construct encodes an expressed protein, we tested the transcription and translation of pSG5, pSG5-Abcc6, and pSG5-Abcc6-del5bp plasmids in a TNT Coupled Reticulocyte Lysate System. Fig. 5A shows two bands with two different molecular masses, respectively: The predicted wild type protein band (MRP6) of nearly 165 kDa and an additional band (ϳ50 kDa) corresponding to the truncated protein (MRP6 trun). Similar protein bands could be also demonstrated by Western blot analysis using an anti-MRP6 polyclonal antibody that recognizes epitope in the N-terminal region after transfection of HEK-293 cells with both pSG5-Abcc6 and pSG5-Abcc6-del5bp (Fig. 5B). Moreover, we analyzed the expression level of both pSG5-Abcc6 and pSG5-Abcc6-del5bp constructs in HEK-293 cells at RNA and protein level using relative real-time reverse transcription PCR and immunohistological analysis, respectively. In comparison to negative controls (empty pSG5 vector), induction in cells transfected with pSG5-Abcc6-del5bp and pSG5-Abcc6 was 73 Ϯ 12 (p Ͻ 0.001) and 475 Ϯ 85 (p Ͻ 0.001), respectively (Table 3). Thus, using relative real-time reverse transcription PCR, the gene expression of Abcc6 was found to reach only 15% in cells transfected with pSG5-Abcc6-del5bp as compared with cells transfected with the wild type pSG5-Abcc6. Accordingly, immunohistological analysis showed also a low expression level of the truncated protein in cells transfected with pSG5-Abcc6-del5bp compared with cells transfected with the wild type pSG5-Abcc6 (Fig. 6).
Genotyping of the rs32756904 SNP in Laboratory Mouse Strains-To find evidence that the Cys to Thr mutation at codon 619 (rs32756904) is indeed a causative variant, we sequenced the region spanning exon 14 in available DNAs from  11 laboratory inbred strains as well as the B6.C3H Dyscalc1 congenic mice (Cg1), i.e. mice that carry the Abcc6 allele from C3H mice on a C57 genetic background. As shown in Table 3 we tested for genotype-phenotype association (Table 4)  C3H, DBA, and 129S1 mice as well as B6.C3H Dyscalc1 congenic mice were found to carry the Thr allele at the 3Ј-border of exon 14 (at codon 619). All these mice were found to display DCC after freeze-thaw injury (13). In contrast, mice that carry the Cys allele were found either negative (C57, FVB, MRL, A, and CBA) or positive (NZB and Balb/c) for DCC. We suggest that other mutations in Abcc6 might explain the susceptibility of NZB and Balb/c mice to develop DCC.
Predisposition of DCC-susceptible C3H Mice to PXE-We previously reported the predisposition of C3H mice to cardiovascular calcification (DCC). Particularly, these mice demonstrated calcium phosphate deposits some days following freezethaw injury of myocardial tissue as well as of the abdominal aorta (11,13,23). In the Abcc6 knock-out mice, MRP6 disruption led to a rather similar phenotype called PXE that, however, occurs rather late during the aging process. Specifically, these Abcc6 knock-out mice displayed calcium phosphate mineral deposits in small-and medium-sized vessels predominantly in kidney tissue at 6 months of age and progressed to affect more organs, including skin, eye, and adipose tissue as well as aorta, vena cava, and Bruchs membrane, at 17-22 months of age (16,17). The C57 wild type mice studied here displayed no such signs of calcification (data not shown). However, MRP6 deficiency in C3H mice was related in characteristics suggesting PXE in addition to known phenotype in these animals, DCC. Specifically, we sampled different organs from C3H mice (n ϭ 5) at age 6 -8 months, the abdominal and muzzle skin, eye, heart, and kidney. Calcification was detected within the lumina of small vessels and capillaries close to glomeruli and in renal tubuli (Fig. 7, A  and B). Calcification within the walls of medium-sized myocardial arteries was also observed in some mice (3 of 5) (Fig. 7C). In addition, a fiber-rich region (oblique fibers) in the anterior layer of the corneal substantia propria displayed extensive calcification (Fig. 7D). No calcification was observed in agematched C57 mice (n ϭ 3). No clear differences were seen either in the dermal elastic fibers or in retina between the tested C3H and C57 mice.

DISCUSSION
The ATP binding cassette C 6 (Abcc6) gene has been identified to cause both DCC and PXE (16 -18, 24); however, the mutations responsible for DCC remained elusive. In this study, we present a variant occurring in several mouse strains that relates to phenotypic characteristics of both DCC and PXE. Specifically, examining the genomic DNA of DCC-susceptible C3H and DCC-resistant C57 mice, a missense mutation p.R619S (rs32756904) in the Abcc6 gene was found that gives rise to an alternative splice variant in the corresponding mRNA, which in turn leads to protein expression deficiency. The functional implications of this deletion were then tested in some detail.
The Abcc6 gene encodes MRP6, a 1449-aa protein of 165 kDa that consists of three transmembranous domains and two intracellular nucleotide binding folds (25). It is involved in the ATP-driven transport of various molecules across the cell membrane. MRP6 is mainly expressed in the liver and to a lesser extent in kidney. In contrast, very low levels have been found in tissues affected by dystrophic calcification, such as heart and muscle (18,21,22). Thus, the preventive mechanism of Abcc6 on cardiac calcification seems to be mediated by systemic rather than tissue-specific factors. Accordingly, alterations of hepatic expression of MRP6 may explain the development of DCC and PXE.  Using Western blot analysis, we found a dramatic decrease of the MRP6 protein in hepatocytes of both C3H and congenic mice compared with wild type C57 mice. To determine the genetic variant that might lead to protein deficiency, we examined the 5Ј-UTR, the coding region, and 3Ј-UTR of Abcc6 in both C3H and C57 mice. Comparing the cDNA sequence of Abcc6 in these mice, we found two relevant deletions that might affect mRNA stability, a 10-bp deletion at the 3Ј-UTR region and a 5-bp deletion at exon 14. We therefore hypothesized that one of these deletions might contribute to the observed deficiency in MRP6 expression. Meng et al. (18) previously found no effect of the 10-bp deletion on the mRNA stability, therefore excluding this deletion as causal mutation for DCC.
In this study, we focused on the 5-bp deletion at exon 14 that was newly detected. This deletion was related to an alternative splice variant in C3H mice that is caused by a newly uncovered donor splice site (GU) in the Abcc6 pre-mRNA transcript. The variant leads to a deletion that skips the last 5 bp (GUGCU) of exon 14. The translation of the corresponding mRNA results in a premature stop codon at position 685 that in turn gives rise to a dysfunctional protein lacking nearly half of the expected sequence (aa 685-1449). The introduction of such premature stop codons had been shown by Stamm et al. (26) to enhance mRNA degradation by nonsensemediated decay and thus results in protein deficiency. To examine the expression of the truncated protein we used an MRP6 N terminus antibody and analyzed liver tissues from DCC-susceptible C3H and Cg1 mice using Western blot analysis. We first tested whether this antibody recognizes the truncated protein overexpressed by pSG5-Abcc6-del5bp construct in HEK-293 cells. Transfected HEK-293 cells display expression of the truncated protein. By contrast, we could not detect the truncated protein in liver tissue of DCC-susceptible mice. In fact, this finding suggested early nonsense-mediated decay degradation processes resulting in loss of MRP6 immunoreactivity. Furthermore, to examine the expression level of the Abcc6 variants in transfected HEK-293 cells, we employed relative real-time reverse transcription PCR and immunohistological analysis. We observed a decrease in Abcc6 expression both at RNA and protein levels in cells expressing the truncated protein as compared with cells expressing the wild type Abcc6 cDNA.
Moreover, to find an association between the identified rs32756904 missense mutation at codon 619 and DCC phenotype in mice, we genotyped 11 laboratory mouse strains for this mutation and found three mouse strains that carry the Thr allele at the 3Ј-border of exon 14, C3H, DBA, and 129S1. All of these mice were found to display dystrophic cardiac calcification (13).
Finally, we sought to investigate whether the novel alternative splice variant in Abcc6 gene might also be associated with the development of PXE in mice. Interestingly, a more detailed phenotypic characterization of DCC-susceptible mice carrying the 5-bp deletion displayed features of PXE-like phenotype as well, thus suggesting that this variant may also affect this disease phenotype. Further functional analyses are needed to clarify the downstream mechanisms related to this splice variant on the initiation and development of calcification in mice.

CONCLUSION
We identified a missense mutation in the Abcc6 gene, the gene causing both DCC and PXE. Our series of experiments FIGURE 7. Histological analysis of calcium phosphate deposits in C3H strains (carrying the alternative splice variant). Calcium phosphate deposits (red) were detected in small renal vessels (arrows, A) and tubuli (B) (alizarin red S staining). Calcifications were also observed in medium-sized myocardial arteries (C, green, arrows) and in the cornea (D, green) (calcein staining). Nuclei were counterstained with DAPI.

Abcc6 Splice Variant Leads to Protein Deficiency in Mice
shows that this mutation gives rise to an alternative splice site leading to a 5-bp deletion in the Abcc6 transcript that in turn results in protein deficiency. We suggest that the protective effect of MRP6 seems to be mediated by systematic factors and that alterations of hepatic expression of MRP6 may explain the development of DCC and PXE.