ENPP1-Fc prevents mortality and vascular calcifications in rodent model of generalized arterial calcification of infancy

Diseases of ectopic calcification of the vascular wall range from lethal orphan diseases such as generalized arterial calcification of infancy (GACI), to common diseases such as hardening of the arteries associated with aging and calciphylaxis of chronic kidney disease (CKD). GACI is a lethal orphan disease in which infants calcify the internal elastic lamina of their medium and large arteries and expire of cardiac failure as neonates, while calciphylaxis of CKD is a ubiquitous vascular calcification in patients with renal failure. Both disorders are characterized by vascular Mönckeburg's sclerosis accompanied by decreased concentrations of plasma inorganic pyrophosphate (PPi). Here we demonstrate that subcutaneous administration of an ENPP1-Fc fusion protein prevents the mortality, vascular calcifications and sequela of disease in animal models of GACI, and is accompanied by a complete clinical and biomarker response. Our findings have implications for the treatment of rare and common diseases of ectopic vascular calcification.

G eneralized arterial calcification of infancy (GACI) is an ultra-rare neonatal disease characterized by infantile onset of widespread arterial calcifications in large and mediumsized vessels resulting in cardiovascular collapse and death in the neonatal period. The disease presents clinically with heart failure, respiratory distress, hypertension, cyanosis and cardiomegaly. The prognosis is grave, with older reports of an 85% mortality rate at 6 months 1 , while recently intensive treatment with bisphosphonates has lowered mortality to 55% at 6 months 2 . Tempering this apparent progress is the severe skeletal toxicity associated with prolonged use of etridonate in patients with GACI (ref. 3), the observation that the limited available data makes it difficult to determine if bisphosphonate treatment is truly protective or reflects the natural history of the disease in less effected patients, and the ineffectiveness of bisphosphonates to prevent mortality in some patients even when instituted early [4][5][6][7] .
The overall incidence of GACI is rare, with 200 reported cases in the medical literature 8 and a disease frequency of one in 391,000 (ref. 9). Although the disease was first described by Bryant and White in 1901 (ref. 10), it was not until 2000 that Rutsch et al. 11 noted that serum plasma inorganic pyrophosphate (PP i ) levels and ectonucleotide pyrophosphatase/phosphodiesterase 1 (ENPP1) enzymatic activity were significantly impaired in GACI patients. ENPP1 (also known as PC-1) is the founding member of the ENPP or NPP family of enzymes, which are characterized by phosphodiesterase activity 12 , and is a type II extracellular membrane bound glycoprotein located on the mineral-depositing matrix vesicles of osteoblasts and chondrocytes, as well as the vascular surface of cerebral capillaries 13 . ENPP1 is the primary source of extracellular PP i in the body, and hydrolyzes extracellular ATP into AMP and PP i (refs 12,14). PP i acts as a potent inhibitor of mineralization, presumably by occupying some of the P i sites on the surface of nascent or growing hydroxyapatite crystals, thereby creating irregularities that slow or terminate crystal growth 14,15 . Rutsch et al. discovered that inactivating mutations in ENPP1 account for 75% of GACI patients 2,11,16 and have contributed to recent observations that a sizable fraction of the remaining patients result from inactivating mutations in the ATP dependent membrane transporter ABCC6 (refs 17-22). ABCC6 mutations result in decreased extracellular concentrations of nucleoside triphosphates through an unknown mechanism 23 , thereby limiting ENPP1's metabolism of ATP into extracellular PP i . The identification of both ENPP1 and ABCC6 inactivating mutations as genetic etiologies for GACI therefore links the same biochemical pathway to unregulated tissue mineralization in both genetic variants of GACI.
Despite the multiple genetic etiologies and multifactorial nature of the expression, progression, and severity of medial wall vascular calcification, we hypothesized that disruption of (a) Daily weights of enpp1 wt (cyan squares, n ¼ 9 animals) and enpp1 asj/asj mice (green circles, n ¼ 8 animals) on the acceleration diet in utero over a 70 day period. Mean average weights are plotted with s.d's denoted by error bars. A failure to thrive point is noted in the enpp1 asj/asj cohort at day 26, when the weights diverge from enpp1 wt . Death events are marked with red arrows. (b) Mean survival of enpp1 asj/asj was 58 days. No deaths were observed in the enpp1 wt cohort. Analysis by log-rank (Mantle-Cox) test yields a w 2 of 15.73 and P value of o0.0001. (c) enpp1 asj/asj animals displayed dramatic calcifications in heart and aorta visible on micro-CT scans. (d,e) Histology of enpp1 asj/asj mice, aorta (Hematoxylin and Eosin (H&E) and Alizarin red). Aortas of enpp1 asj/asj mouse all possessed near circumferential calcifications that were pervasive in the vascular walls, as illustrated by Alizarin red staining of the aortas. Scale bar, 20 mm (f). Histology of enpp1 asj/asj mice, left ventricle. Extensive calcifications surrounded by scar tissue revealing the presence of repeated, old and healed myocardial infarctions. Scale bar, 10 mm (g). Histology of enpp1 asj/asj Mice, Septum (H&E). More typically, the enpp1 asj/asj mice displayed small foci of calcifications with surrounding scar tissue as seen here in the myocardial septum, also diagnostic of previous myocardial infarctions. Scale bar, 10 mm.
ENPP1's extracellular purinergic metabolism accounts for the pathologic sequela and mortality associated with GACI and that enzyme replacement therapy with ENPP1 is a tractable therapeutic approach. To test this hypothesis, we employed ENPP1-asj mice (Jackson Laboratory), which carry an inactivating mutation in enpp1 such that homozygotes (enpp1 asj/asj ) have markedly reduced ENPP1 function. To enhance the disease phenotype, the mice were maintained on the an 'acceleration diet' in utero 24,25 . Here we demonstrate that daily subcutaneous doses of a biologic agent comprising the extracellular domain of ENPP1 fused to the Fc region of human IgG1 eliminates the vascular calcifications, mortality, myocardial infarctions, and sequela of GACI in enpp1 asj/asj mice, and that this clinical response is accompanied by normalization of serum and tissue PP i concentrations.

Results
Natural history study of GACI in enpp1 asj/asj mice. When fed an acceleration diet, the daily weights of enpp1 asj/asj mice diverged from wild-type (WT) siblings pairs at day 26, when the ENPP1asj/asj mice experienced a 'failure to thrive' event and began to lose weight (Fig. 1a). After day 26 the enpp1 asj/asj animals displayed progressive stiffness and reductions in physical activity. All of the enpp1 asj/asj animals died between days 35-71, with a median lifespan of 58 days (Fig. 1b). The presence of calcifications in enpp1 asj/asj and enpp1 wtj mice was evaluated via postmortem by micro-computed tomography (CT) scans (Fig. 1c) and histologic sections taken from the heart, aorta and kidneys ( Fig. 1d-g). Approximately one-third of the enpp1 asj/asj mice had visible calcifications in their hearts, and two-thirds had calcifications in their aortas visible by micro-CT imaging (Table 1). These percentages increased to 100% on histologic examination, which also showed that many of the animals had dramatic nearly circumferential calcifications in their aortic walls (Fig. 1d,e). Histologic examination also revealed that 100% of the coronary arteries possessed arterial wall calcifications and that 70% of the animals had focal or confluent areas of myocardial necrosis consistent with myocardial infarction (Fig. 1f,g). Conversely, the enpp1 wt mice displayed none of these abnormalities. These findings demonstrate that the animal model recapitulates GACI in humans, which is characterized by prominent calcifications of the large and medium sized arteries and a cardiac demise.
Design and characterization of therapeutic. To produce soluble, recombinant ENPP1 for in vivo use, we modified our previous expression construct 26  substrate by over two orders of magnitude, while also reducing (slowing down) the k cat by a factor of 2-3 ( Fig. 3a-c). The altered Michaelis-Menton constants could reflect either the change in expression system from baculovirus-infected insect cells to mammalian cells or the addition of the Fc tag. To differentiate between these two possibilities, we expressed hENPP1 without an Fc tag in mammalian (HEK293) cell lines to directly compare the kinetics of ENPP1 with and without an Fc tag. We found that the kinetic constants of the hNPP1-Fc and hNPP1 enzymes expressed in mammalian cells were essentially identical (Fig. 3d), suggesting that the alteration in kinetic constants was due to the change in expression systems.
To understand the pharmacokinetics of ENPP1-Fc we determined plasma concentrations of drug following a single subcutaneous injection over time. Fifteen mice were administered a single subcutaneous dose of 10 mg kg À 1 and blood was collected from tail veins to quantitate ENPP1-Fc in the plasma. The concentration ENPP1-Fc reaches a maximal plasma concentration of B300 nM (C max ) within 8 h (t max ) and remains elevated at 100 nM for 72 h after dosing (Fig. 4a). The integral of the plotted concentration-time curve, or area under the curve, is 9 mM h À 1 . Plotting the data as fraction of drug absorbed over time allows for the determination of the elimination (k e ) and absorption (k a ) constants by fitting the data to the equation for the total systemic absorption of a drug administered at a This analysis yield a k e ¼ 0.107 ± 0.016 h À 1 and a k a ¼ 0.048 ± 0.008 h À 1 , which yields an elimination half life (t 1/2 ) ¼ 6.5 h (Fig. 4b).
The activity of ENPP1-Fc was noted to diminish over a 30 day period when stored at 4°C, but the enzyme could be frozen at À 80°C and retain nearly complete activity on thawing (Fig. 4c). The enzyme was therefore stored as a frozen stock solution after purification until needed. The oligomerization state of ENPP1-Fc at 1 mM and 100 nM was determined by size exclusion chromatography coupled with light scattering, refractive index, and absorbance (ultraviolet) detection (SEC-MALLS/RI/UV) 27 , which allows for a determination of the molecular weight (MW) of glycosylated proteins from the relationship of molecular weight and the ratio of laser light scattering, ultraviolet and refractive index (RI) signals in the absence of knowledge of the extend of glycosylation 28 . The results of the SEC-MALLS analysis revealed that the MW of the ENPP1-Fc protein represents a dimer over the concentration range of 100 nM-1 mM, with an experimentally determined MW of 274 ± 38 KDa (Fig. 4d). This agrees well with the calculated MW of 252 KDa for an ENPP1-Fc dimer. The experimental approach also allows for an independent estimation of the glycosylation level of the fusion protein, which was determined to be B5% according to the method of Hayashi et al. 29 Following purification, ENPP1-Fc was dialysed into PBS supplemented with Zn and Ca 2 þ (PBS plus ) concentrated to between 5 and 7 mg ml À 1 and frozen at À 80°C in aliquots of 200-500 ml. Aliquots were thawed immediately before use, and the specific activity of the solution was adjusted to 31.25 a.u. ml À 1 (or E0.7 mg ml À 1 depending on the preparation) by dilution in PBS plus .
Therapeutic proof of concept in enpp1 asj/asj mice. Dosing was performed according to activity units (a.u.) per kg animal weight to account for variations in specific activity in different protein preparations. The specific activity of the enzyme varied with each protein preparation, and because the clinical response was noted to be highly dependent on enzyme specific activity, we rejected protein preparations with specific activities of o40 a.u. mg À 1 . To establish initial dosing levels for the proof of concept study we performed dose escalation trials in limited numbers of animals (1-2 per dose level). While both the human and mouse version ENPP1 were used in the dose escalation trials, the proof of concept study was performed with the mouse isoform of ENPP1-Fc (mENPP1-Fc). enpp1 asj/asj mice were dosed daily on the fourteenth day of life with subcutaneous injections of mENPP1-Fc and weekly with intra-peritoneal injections of GK 1.5, the latter added to minimize immune rejection of recombinant protein 30 . Subcutaneous doses of mEnpp1-Fc at 500 a.u. kg À 1 qD demonstrated a strong early response in weight with an absence of the observed 'failure to thrive' crisis observed in undosed enpp1 asj/asj animals ( Supplementary Fig. 1). This dose corresponded to between 6-10 mg kg À 1 of ENPP1-Fc, depending on the specific activity of the protein preparation.
On the basis of the results of the dose escalation trials we chose to enrol a cohort of 8 enpp1 asj/asj animals dosed with mENPP1-Fc at 500 a.u. kg qD and weekly intra-peritoneal injections with GK  The initial rate of ATP cleavage by ENPP1-Fc from HEK293 cells is essentially the same at [ATP] greater than 7.8 mM, yielding a k cat (the average of the rates Z7.8 mM) of 3.4 ( ± 0.4) s À 1 per enzyme. The initial rate at 2.0 mM ATP concentration is about a half of the k cat value. We therefore estimate a K M B 2 mM for ATP hydrolysis by hNPP1-Fc protein. (d) ATP concentration dependence of the initial hydrolysis rate of ENPP1 (no Fc) purified from HEK293 cells. The reaction displays a k cat of 3.46 ( ± 0.44) s À 1 per enzyme, estimated from the average of all ATP concentrations Z2 mM. At 1 mM initial ATP (the first data point), the hydrolysis rate is slightly less than k cat values. We therefore estimate a K M o2 mM for ATP hydrolysis by ENPP1 (no Fc) purified from HEK293 cells.
1.5 (Fig. 5). We included a control groups (enpp1 wt and enpp1 asj/asj ) dosed daily with vehicle and weekly with GK 1.5 in an identical manner as the dosed cohort, and the study duration was shortened to 55 days. All 8 treated enpp1 asj/asj animals survived the full 55 days of the trial, with a dramatic clinical response observed in treated animals (Supplementary Movie 1), while the median lifespan of the untreated enpp1 asj/asj animals decreased from 58 to 35 days in the therapeutic trial, perhaps resulting from the weekly intra-peritoneal injections of the GK 1.5 immunosuppressant. The untreated enpp1 asj/asj animals also all experienced a failure to thrive crisis at day 26, followed by weight loss and mobility restriction progressing variably to paralysis and death over the next 30 days. All but one untreated enpp1 asj/asj animal expired over the 55 day trial, while in contrast all treated enpp1 asj/asj mice gained weight comparable to the enpp1 wt mice and displayed no signs of reduced mobility or stiffness (Supplementary Movie 1).
At the conclusion of the study, 100% of the enpp1 asj/asj mice treated with vehicle displayed calcifications in their hearts, aortas and coronary arteries, and 77% of the animals displayed histologic evidence of myocardial infarction (Table 2). In most cases this took the form of small areas of myocardial cell necrosis and single cell drop out in the vicinity of the cardiac calcifications (Figs 5d and 6f-h), but in two animals (22%) there were large, full thickness myocardial infarctions in the free wall of the right ventricle (Fig. 6c-e). Myocardial fibrosis in the myocardial tissue adjacent to regions of coronary artery calcification was a common finding (Fig. 6g,h), illustrating that ischaemia from coronary artery calcification likely accounts for a substantial burden of the myocardial disease. Some untreated enpp1 asj/asj animals displayed dramatic calcifications of coronary arteries, heart, ascending and descending aorta (Supplementary Movie 2). In contrast, none of the enpp1 asj/asj animals treated with ENPP1-Fc displayed cardiac, arterial, or aortic calcification on histology or post-mortem micro-CT (Table 2 and Figs 5e, 6b and 7a). Although renal calcifications are not a feature of human GACI, the enpp1 asj/asj mouse model is noted to have calcifications in the kidneys. Similar to previous findings, we noted light calcifications in the kidneys of E60% of the WT animals and 100% of the dosed enpp1 asj/asj mice. The calcifications in these animals were centred in the renal medulla. 100% of the undosed enpp1 asj/asj animals had heavy, extensive calcifications, centred in the outer medulla, with extension into the renal cortex. In comparison to previous reports, the renal calcifications in the dosed animals seem to approximate renal calcifications seen in the heterozygous enpp WT/asj animals.
Biomarker response. In addition to survival, daily animal weights, and terminal histology, treatment response was also assessed via post-mortem high-resolution micro-CT scans to image vascular calcifications, plasma PP i concentrations, and  Table 2). The biochemical and physiologic response was complete as measured by all of these parameters. None of the WT or treated enpp1 asj/asj animals were noted to possess any vascular calcifications via micro-CT, in contrast to the dramatic calcifications noted in the aortas, coronary arteries, and hearts of the untreated enpp1 asj/asj cohort (Fig. 7a). In addition, serum PP i concentrations of treated enpp1 asj/asj animals (E5.2 mM) were elevated to WT levels (4.4 mM) and significantly above untreated enpp1 asj/asj levels (o0.5 mM) (Fig. 7b). 99m PYP, an imaging agent typically employed in cardiac imaging and bone remodelling, was used as a marker for treatment response. It is sensitive to areas of unusually high-bone rebuilding activity since it localizes to the surface of hydroxyapatite and then may be taken up by osteoclasts. One would expect increased 99m PYP uptake in animals lacking functional ENPP1 since they have reduced plasma [PP i ] and correspondingly elevated rates of mineralization. To test this hypothesis, we performed weekly in vivo 99m PYP imaging in enpp1 wt and undosed enpp1 asj/asj animals to detect differences in 99m PYP uptake between the sibling pairs (Fig. 7c,d). We limited analysis of 99m PYP uptake to the head, which is comprised of both enchondral bone (skull) and soft tissue (vibrissae), which are known sites of ectopic calcification in enpp1 asj/asj mice. This also simplified data analysis as the head does not overlap with internal organs showing transient 99m PYP uptake (such as the bladder, heart and diaphragm) during the 180°camera rotation that occurs during data collection.
Weekly serial imaging of enpp1 wt and untreated enpp1 asj/asj animals demonstrated that, as expected, uptake of 99m PYP in the heads as per cent injected dose was greater in enpp1 asj/asj animals than in enpp1 wt animals, and that changes in 99m PYP uptake within experimental groups did not vary significantly over the course of the study (Fig. 7c,d). We therefore chose to measure 99m PYP uptake in treated and untreated enpp1 asj/asj animals at two time points-days 30-35 and at the completion of the study (days 50-65). Comparison of these experimental groups demonstrates that ENPP1-Fc treatment returned 99m PYP uptake in GACI mice to WT levels (Fig. 7e,f), suggesting that ENPP1-Fc treatment is able to abrogate unregulated tissue, vibrissae and skull mineralization in enpp1 asj/asj mice by raising the extracellular PP i concentrations.

Reappearance of calcifications following cessation of dosing.
To address questions regarding the reappearance of calcifications following the cessation of dosing, we enrolled two animals in a limited dosing trial in which enpp1 asj/asj animals were dosed with hENPP1-Fc starting on day 14 and ending on day 27. On day 28 and thereafter the animals were dosed with PBS plus and the appearance of vascular calcifications were followed with weekly in vivo CT scans (Fig. 8a). Both animals were free of calcifications until day 64, when one animal developed calcifications in the heart, which were noted on day 79 to progress to the aorta, spleen, kidney, and liver. The second animal developed renal calcifications on day 79. Surprisingly both animals remained alive past day 84. The limited dosing study demonstrates that calcifications reappear following the cessation of dosing and that treatment in the 14-27 day window significantly extends survival in enpp1 asj/asj animals maintained on the acceleration diet in utero.

Discussion
Diseases of ectopic tissue calcification range from ultra-rare diseases such as GACI to nearly ubiquitous maladies in the aging population such as hardening of the arteries. The genetic aetiology of human GACI suggests that the lethal arterial calcifications result from impairment of extracellular purinergic metabolism, either through loss of function mutations in ENPP1 or upstream reductions in nucleotide triphosphates metabolized by ENPP1 into extracellular PP i . Here we demonstrate that daily subcutaneous injections of ENPP1-Fc fusion protein eliminate the mortality, cardiac and arterial calcifications, and other sequela of disease in rodent models of GACI. ENPP1 is the enzyme   responsible for the generation of extracellular PP i , and we demonstrate that ENPP1-Fc raises plasma pyrophosphate levels from the nearly undetectable levels present in enpp1 asj/asj mice to concentrations comparable to those seen in WT sibling pairs. Our findings suggest that ENPP1-Fc may be effective in other diseases of ectopic calcification in which plasma PP i concentrations are decreased.
Extracellular PP i is a potent inhibitor of mineralization. ENPP1 is the major producer, and tissue-nonspecific alkaline phosphatase (TNAP) is the primary degrader, of extracellular PP i (Fig. 2a) The contribution of AMP generation and purinergic signalling to the suppression of vascular calcification is incompletely understood but has been addressed previous studies of enpp1 asj/asj mice using aortic allografts of WT into enpp1 asj/asj mice and vice versa 31 . These studies demonstrated that normal levels of extracellular pyrophosphate were sufficient to prevent vascular calcification over the entire surface of a transplanted enpp1 asj/asj aortic allograft. The suppression of calcification in aortic allografts was throughout the entire length of the allograft with no gradient of suppression of calcification observed. Because purinergic signalling is believed to be autocrine or paracrine in nature (and not systemic), the authors concluded that purinergic signalling was either not important for vascular calcification suppression or was not substantially altered by ENPP1 deficiency. Our observations that enpp1 asj/asj mice dosed with ENPP1-Fc are free of vascular calcifications and have normal plasma PP i concentrations supports the notion that plasma PPi levels may be primarily responsible for suppression of vascular calcification in ENPP1 deficiency.
Reduced plasma PP i levels are also present in vascular calcification associated with end stage renal disease (ESRD) 31,32 . One in ten adults suffer from chronic kidney disease (CKD), and it is estimated that 80% of patients with CKD on dialysis and 47-83% of patients with CKD not on dialysis possess vascular calcifications compromising their quality of life and endangering their health. Vascular calcifications associated with ESRD contributes to poor outcomes by increasing pulse pressure, causing or exacerbating hypertension, and inducing or intensifying myocardial infarctions and strokes. Most patients with ESRD do not die of renal failure, but from the cardiovascular complications of ESRD, and it is important to note that many very young patients with ESRD on dialysis possess coronary artery calcifications. The histologic subtype of vascular calcification associated with CKD is known as Mönckeburg's sclerosis, which is a form of vessel hardening in which calcium deposits are found in the muscular layers of the medial vascular wall. This form of calcification is histologically distinct from intimal or neo-intimal vascular wall calcification commonly observed in atherosclerosis but identical to the vascular calcifications observed in human GACI patients, and in the rodent models of the disease described herein. The similar pathophysiology of vascular calcification present in GACI and CKD suggests that these disorders may be treated by a common therapeutic.
To this point, the current medical strategies treating ectopic vascular calcification of CKD are typically ineffective and are directed at reducing hyperphosphatemia and minimizing serum calcium concentrations. Continuous intra-peritoneal infusions of exogenous PP i substantially inhibits vascular calcifications in rodent models of ESRD, but the rapid hydrolysis of PP i in vivo prevents translation of this therapy to the clinic 33 . Attempts treating vascular calcifications in rodent models of ESRD with non-hydrolyzable PP i analogues (bisphosphonates) required doses above those known to inhibit bone formation, and the anticipated bone toxicity coupled with a renal mechanism of clearance discourages the use of bisphosphonates as therapeutic agents in CKD 34 . It appears that we are able to circumvent both the poor pharmacokinetics of PP i and the bone toxicity of bisphosphonates by fusing the PP i generating enzyme ENPP1 to the Fc domain of IgG1, and demonstrated that the biologic is effective in GACI. We suspect that it may also be effective in vascular calcification associated with CKD.
Methods ENPP1-asj GACI mouse model. Animal care and maintenance were provided through Yale University Animal Resource Center at Yale University (New Haven). All procedures were approved by the Animal Care and Use Committee of Yale University and complied with the US National Institutes of Health guide for the care and use of laboratory animals. Heterozygous enpp1 asj/ þ (genotype C57BL/6J-Enpp1 asj /GrsrJ, Jackson Laboratory stock number 012810) breeding pairs were maintained on the 'acceleration diet' (TD.00442, Harlan Laboratories, Madison WI) throughout the entire experiment and food and water were delivered ad libitum. The animal colony was housed in pathogen free conditions. All experimental animals were maintained on the acceleration diet in utero through completion of the study. Litters were genotyped on day 8 and weaned at day 21. Following weaning, sibling pairs were sequentially divided into cohorts as described below and enrolled in experimental trials. Animals were consecutively enrolled in experimental trials without regard to gender, and the gender of the experimental animals was not recorded. On the basis of the trial dosing we estimated that statistically significant survival differences could be determined between dosed and undosed animals using a sample size of eight animals in each experimental group. We used identical breeding pairs throughout the study, and all experimental groups were enrolled sequentially beginning with the enpp1 wt and undosed enpp1 asj/asj cohorts and ending with the dosed enpp1 asj/asj cohort. Enrolment of the dosed and undosed enpp1 asj/asj cohorts spanned 4 months. Once the enrolment of an experimental group began both sexes of the appropriate genotype were consecutively enrolled in an experimental cohort with the exclusion of severely runted animals weighing o5.5 g at 14 days of life. Following weaning, all experimental animals were housed with littermates to allow for cooperative grooming and nesting. Experimentalists were not blinded during the study.
ENPP1-Fc design. Human and mouse NPP1 (Human: NCBI accession NP_006199; Mouse: NCBI accession NP_03839) modified to express soluble, recombinant protein as described previously 26 were fused to IgG 1 by subcloning into pFUSE-h IgG 1 -Fc1 or pFUSE-m IgG 1 -Fc1 plasmids (InvivoGen, San Diego, CA), respectively. The final human protein sequence is listed in Supplementary  Fig. 1. ARTICLE Protein production with shaking flasks. Stable transfections of the ENPP1-Fc were established in HEK293 cells under zeocin selection. Briefly, adherent HEK293 cells were adapted for suspension growth by progressive dilution of fetal bovine serum from 10 to 1% over 5-6 passages 35 . Zeocin selection (450 mg ml À 1 ) was maintained throughout the adaptation process, and cultures were supplemented with 1% insulin-transferrin-selenium (Mediatech 25-800-CR), and cells were passaged when they reached confluence in 175 cm 2 tissue culture flasks. Adapted cells were used to seed liquid culture growths in FreeStyle medium (Gibco, Waltham MA) in shaker flasks at 37°F and 5% CO 2 , agitated at 120 r.p.m. with high humidity. Cell density in liquid culure was maintained between 0.5 and 2.5 Â 10 6 cells per ml and zeocin selection was maintained in liquid culture at 450 mg ml À 1 until the cells reached a volume of 4 l in liquid culture, and then reduced to 225 mg ml À 1 thereafter. The culture was gradually expanded to 8 l and then maintained for another 12 days to accumulate extracellular protein. During the maintenance phase, cultures were supplemented with CD EfficientFeed C AGT (Gibco #A13275-05) to enhance protein production.
Protein production with bioreactor. Cells were propagated in a 10 l bioreactor equipped with dissolved oxygen and pH control. Dissolved oxygen was kept at (g) Untreated enpp1 asj/asj mice, coronary arteries (H&E). All untreated enpp1 asj/asj mice had coronary calcifications, with most displaying circumferential calcifications in coronary arteries surrounded by scar tissue, diagnostic of ischaemia and myocardial infarction. Scale bar, 10 mm.
(h) Untreated enpp1 asj/asj mice, coronary arteries (Trichrome). Trichrome stains of coronary artery regions of untreated enpp1 asj/asj mice demonstrates increased fibrosis associated with vascular wall calcifications (blue colour and labelled 'scar'), demonstrating the myocardial injury in the animals. Scale bar, 10 mm. ] in enpp1 wt and treated and untreated enpp1 asj/asj animals revealed that treatment with mENPP1-Fc increased [PP i ] in enpp1 asj/asj mice to WT levels, and well above the nearly undetectable levels present in untreated enpp1 asj/asj mice. *Po0.0015, Students two-tailed t-test. (c,d) Per cent uptake of injected 99m PYP in heads of WT and asj/asj animals. The per cent uptake of 99m PYP in heads of animals in the natural history study were recorded weekly in the WT and asj/asj animals on the acceleration diet, demonstrating that 99m PYP uptake remains nearly constant over an 80 day period following birth, but differs markedly between the two experimental groups. (d) In the natural history study, the average 99m PYP uptake in heads of enpp1 wt animals was around 15% of injected dose over the 80 day period, while the PYP uptake in enpp1 asj/asj animals was around 20% *Po0.001, Students two-tailed t-test.
(e,f) 99m PYP uptake. The per cent 99m PYP uptake in the heads of all experimental groups was recorded in the middle of the study (days 30-35, in e) and at the end of the study (days 50-65, in f). enpp1 wt and treated enpp1 asj/asj animals had per cent uptake in the skulls around 15%, while the untreated enpp1 asj/asj cohort was at or above 20%. *Po0.001, Students two-tailed t-test.
40% air saturation by supplying the culture with mixture of air and oxygen not exceeding 3 l minute at an agitation rate of 80 RPM. pH was controlled at 7.4 by sparging CO2 when the pH was higher than 7.4. Culture growth was followed by measuring cell number, cell viability, glucose and lactate concentrations. Final yields for both methods of production were B5 mg of purified ENPP1-Fc per liter of culture.  (a) To determine if calcifications reappear following cessation of dosing, two enpp1 asj/asj animals were dosed daily with hENPP1-Fc between days 14-27 (green arrows), followed by daily dosing with PBS plus . Rejection of human version of ENPP1-Fc (hENPP1-Fc) was suppressed during the limited dosing period with weekly doses of GK 1.5. The average daily weights are of enpp1 wt (blue triangles, n ¼ 10) and enpp1 asj/asj (red squares, n ¼ 2) mice are plotted with the error bars denoting s.d.'s. The dosed enpp1 asj/asj animals were followed for reappearance of vascular and organ calcifications by weekly in vivo CT scans. Calcifications in both animals were eventually observed (cyan arrows). One animal developed calcifications in the heart on day 64, which progressed to the aorta, liver, kidney and spleen by day 79. The second animal developed renal calcifications on day 79. (b) Negative control experiment. To demonstrate that ENPP1-Fc enzyme activity is essential for therapeutic effect, and that weekly GK 1.5 administration does not alter the natural history disease, enpp1 asj/asj mice were dosed daily with 10 mg kg À 1 inactive hENPP1-Fc, and weekly with GK 1.5 (blue diamonds, n ¼ 3). The average daily weights are plotted compared with enpp1 asj/asj mice dosed with vehicle-daily PBS plus and weekly GK 1.5-(green triangles, n ¼ 9) and the error bars denote s.d.'s. All three enpp1 asj/asj mice dosed with inactive ENPP1-Fc experienced a drop in weight and mortality (deaths denoted by blue arrows) similar to enpp1 asj/asj mice dosed with vehicle (deaths denoted by red arrows), demonstrating that neither inactive ENPP1-Fc nor GK 1.5 extends survival.
Dosing. Animals in the proof of concept study (Figs 5-7) were dosed either with vehicle or with mouse ENPP1-Fc (mENPP1-Fc) formulated in vehicle. Mice were dosed with daily subcutaneous injections starting on day 14 at dose levels of 500 a.u. kg À 1 mENPP1-Fc. Animals in the limited dosing study (Fig. 8a) were dosed with human ENPP1-Fc (hENPP1-Fc) formulated in vehicle beginning on the fourteenth day of life and ending on the twenty-seventh day of life. To tolerize the mice to human protein, the mice were dosed intraperitoneally with 75 mg of GK 1.5 (eBioscience, San Diego, CA) on the thirteenth day of life, and 50 mg of GK 1.5 on the twentieth day of life. Between days 28 and 83 the mice received vehicle. Inactive hENPP1-Fc protein (Fig. 8b) was generated by exposure to EDTA. The specific activity hENPP1-Fc following EDTA exposure was o 20 a.u. mg À 1 , or o50% of the activity of native protein.
Clearance rate of ENPP1-Fc. Briefly, 15 male C57BI/6J mice aged 8 weeks or older were administered a single subcutaneous dose of 10 mg kg À 1 of mENPP1-Fc. Blood was collected from tail veins into heparin-containing tubes before the injection and at different time points post injection. Mice were divided into three groups for blood collection: (1)  measured with pNP-TMP assay in the following buffer: 100 mM Tris Á HCl (pH 9.0), 500 mM NaCl, 5 mM MgCl2, 0.05% (vol/vol) Triton X-100 and 4 mM pNp-TMP. Plasma activity units were converted into protein mass using an empirically determined conversion factor of 0.332 units per mg protein. To derive plasma concentrations and fraction of subcutaneous drug absorbed into plasma, the total blood volume of each mouse was estimated at 1.5 ml. Clearance rate constants were derived as detailed in the text.
Definition of activity unit. Animals were dosed according to total activity of enzyme delivered rather than concentration of enzyme to account for varying activity among batches of enzyme used. One activity unit (1 a.u.) is defined as pM of pNP-TMP substrate hydrolyzed min À 1 mg À 1 enzyme. The activity assay was performed in a buffer consisting of 50 mM Tris pH9, 150 mM NaCl, 0.1 mM ZnCl 2 , 0.1 mM CaCl 2 , 0.1 mM MgCl 2 . The activity of acceptable protein preparations varied between 40 and 43 a.u. mg À 1 , and preparations with o40 a.u. mg À 1 were discarded. A dose of 500 a.u. kg À 1 corresponds to between 6 and 10 mg kg À 1 , depending on the specific activity of the protein preparation.
Quantification of plasma PP i . Animals were terminally bled retro-orbitally using heparinized, micropipets, and the blood was immediately dispensed into heparintreated eppendorf tubes and placed on wet ice. The samples were spun in a 4°C pre-cooled microcentrifuge at 4,000 r.p.m. for 5 min, and plasma was collected and diluted in one volume of 50 mM Tris-Acetate pH ¼ 8.0. Plasma was then filtered through a 300 KDa membrane via ultracentrifugation (NanoSep 300 K, Pall Corp., Ann Arbour, MI) and frozen at À 80°C. Pyrophosphate was quantitated using standard three-step enzymatic assays using uridine 5' diphospho[ 14 C]glucose to record the reaction product, uridine 5' diphospho[ 14 C]gluconic acid 37 . Briefly, a reaction mixture (100 ml) containing 5 mM MgCl 2 , 90 mM KCL, 63 mM Tris-HCL (pH 7.6), 1 nmol NADP þ , 2 nmol glucose 1,6-diphosphate, 400 pmol uridine 5'-diphosphoglucose, 0.02 mCi uridine 5' diphospho[ 14 C]glucose, 0.25 units of uridine 5'-diphosphoglucose pyrophosphorylase, 0.25 units of phosphoglucose mutase, 0.5 units of glucose 6-phosphate dehydrogenase, and inorganic pyrophosphate (50-200 pmol) is incubated for 30 min at 37°C. The reaction is terminated by the addition of 200 ml of 2% charcoal well suspended in water. The mixture is kept on ice, vortexed three times over 15 min and clarified by centrifugation for 5 min. An aliquote of 200 ml of supernatant is then counted in scintillation solution.
In vivo 99m PYP imaging. The bone imaging agent 99m Tc-pyrophosphate (Pharmalucence, Inc) was evaluated in cohorts of animals using a preclinical microSPECT/CT hybrid imaging system with dual 1 mm pinhole collimators (X-SPECT, Gamma Medica-Ideas) 38 . Each animal was injected intraperitoneally with 2-5 mCi of the radiolabelled tracer and imaged 1-1.5 h after injection. A CT scan (512 projections at 50 kVp, 800 uA and a magnification factor of 1.25) was acquired for anatomical co-localization with the SPECT image. The SPECT imaging was acquired with 180°per collimator head in a counter-clockwise rotation, 32 projections, 60 s per projection with an ROR of 7.0 cm, FOV of 8.95 cm and an energy window of 140 keV±20. CT images were reconstructed with the FLEX X-O CT software (Gamma Medica-Ideas) using a filtered back-projection algorithm. SPECT images were reconstructed using the FLEX SPECT software (5 iterations, 4 subsets) and subsequently fused with the CT images and analysed using the AMIRA software and offline in-house script. Data were corrected for decay and injected dose to achieve % injected dose.
Quantification of 99m PYP uptake. For the 99m PYP murine scans, the animals were imaged two hours postinjection. The resulting SPECT scans were imported into NIH's ImageJ image processing software and regions of interest were drawn around each animal's head (target organ) and whole body. Per cent injected activity (PIA), often referred to as 'per cent injected dose' was calculated by comparing the ratio of counts in the head to the counts in the whole body, and expressed as per cent injected dose to give a measure as of the affinity with which the radiotracer is taken up by the region of interest (head). The total counts in each scan were taken as the whole body measure of injected dose.