Novel Inhibitors of Alkaline Phosphatase Suppress Vascular Smooth Muscle Cell Calcification


  • The authors state that they have no conflicts of interest.


We report three novel inhibitors of the physiological pyrophosphatase activity of alkaline phosphatase and show that these compounds are capable of reducing calcification in two models of vascular calcification (i.e., they suppress in vitro calcification by cultured Enpp1−/− VSMCs and they inhibit the increased pyrophosphatase activity in a rat aortic model).

Introduction: Genetic ablation of tissue-nonspecific alkaline phosphatase (TNALP) leads to accumulation of the calcification inhibitor inorganic pyrophosphate (PPi). TNALP deficiency ameliorates the hypermineralization phenotype in Enpp1−/− and ank/ank mice, two models of osteoarthritis and soft tissue calcification. We surmised that the pharmacological inhibition of TNALP pyrophosphatase activity could be used to prevent/suppress vascular calcification.

Materials and Methods: Comprehensive chemical libraries were screened to identify novel drug-like compounds that could inhibit TNALP pyrophosphatase function at physiological pH. We used these novel compounds to block calcification by cultured vascular smooth muscle cells (VSMCs) and to inhibit the upregulated pyrophosphatase activity in a rat aortic calcification model.

Results: Using VSMC cultures, we determined that Enpp1−/− and ank/ank VSMCs express higher TNALP levels and enhanced in vitro calcification compared with wildtype cells. By high-throughput screening, three novel compounds, 5361418, 5923412, and 5804079, were identified that inhibit TNALP pyrophosphatase function through an uncompetitive mechanism, with high affinity and specificity when measured at both pH 9.8 and 7.5. These compounds were shown to reduce the calcification by Enpp1−/− VSMCs. Furthermore, using an ex vivo rat whole aorta PPi hydrolysis assay, we showed that pyrophosphatase activity was inhibited by all three lead compounds, with compound 5804079 being the most potent at pH 7.5.

Conclusions: We conclude that TNALP is a druggable target for the treatment and/or prevention of ectopic calcification. The lead compounds identified in this study will serve as scaffolds for medicinal chemistry efforts to develop drugs for the treatment of soft tissue calcification.


Vascular calcification refers to the deposition of hydroxyapatite in cardiovascular tissues such as arteries and heart valves and is a significant risk factor in the pathogenesis of cardiovascular disease, being associated with myocardial infarction and coronary death.(1) Vascular disease is also common in diabetes, obesity, aging, and renal failure, where it is responsible for much of the morbidity and mortality in end-stage renal disease.(2) Two types of calcification are recognized: the first occurs primarily in the form of intimal calcification, usually associated with atherosclerosis, and the second, also known as Mönckeberg's sclerosis, is defined as calcification restricted to the arterial media.(3) A key regulator of mineralization, both in bone and vessels, is extracellular pyrophosphate (ePPi).(4) This compound is a potent inhibitor of hydroxyapatite formation at concentrations normally found in plasma(5) and prevents calcification of rat aortas in culture.(6) Conversely, ePPi depletion promotes spontaneous arterial calcification.(7) Mice lacking ecto-nucleotide pyrophosphatase/phosphodiesterases-1 (NPP1, a.k.a PC-1), a major generator of ePPi, spontaneously develop articular cartilage, perispinal, and medial aortic calcification at a young age.(8) These NPP1 knockout mice (Enpp1−/−) share phenotypic features with a human disease, idiopathic infantile arterial calcification.(9,10) Interestingly, another mouse model of depressed ePPi levels, this time caused by defective transport function of the transmembrane protein ANK (ank/ank mutant mice), also develops soft tissue calcification, including vascular calcification.(7,11,12)

Alkaline phosphatases (ALPs; E.C. are dimeric enzymes present in most organisms.(13) They catalyze the hydrolysis of phosphomonoesters with release of inorganic phosphate (Pi) and alcohol. In humans, three of the four isozymes are tissue-specific, i.e., the intestinal (IALP), placental (PLALP), and germ cell (GCLP) APs, whereas the fourth ALP is tissue-nonspecific (TNALP) and is expressed in bone, liver, and kidney. Recent studies have provided compelling proof that a major role for TNALP in bone tissue is to hydrolyze ePPi to avoid accumulation of this mineralization inhibitor, thus ensuring normal bone mineralization.(14–16) Normalization of ePPi levels in NPP1 null and ANK-deficient mice improves their calcification abnormalities.(14,15) Crossbreeding either the Enpp1−/− or the ank/ank mice to mice deficient in TNALP (Akp2−/−) mice normalizes ePPi levels and induces a secondary upregulation of osteopontin (OPN) levels, another calcification inhibitor.(16,17) Importantly, these studies have indicated that TNALP may be a useful therapeutic target for the treatment of diseases such as ankylosis and osteoarthritis, but also arterial calcification. The presence of TNALP-enriched matrix vesicles (MVs) in human atherosclerotic lesions suggests an active role in the promotion of the accompanying vascular calcification.(18–22) Increased expression of TNALP accelerates calcification by bovine vascular smooth muscle cells (VSMCs)(23) and macrophages may induce a calcifying phenotype in human VSMCs by activating TNALP in the presence of IFNγ and 1,25(OH)2D3.(24) Calcification of rat aorta in culture and of human valve interstitial cells has been shown to be dependent on TNALP activity.(6,25) Thus, there is ample evidence warranting exploration of the therapeutic potential of TNALP inhibition at sites of arterial calcification to increase local concentrations of ePPi, which is expected to antagonize the deposition of hydroxyapatite. For these reasons, in this study, we have undertaken the screening of large, comprehensive small, drug-like, molecule libraries to identify and characterize novel potent TNALP inhibitors that might be useful for the treatment of vascular calcification. We presently describe three new compounds, one of which acts as a potent uncompetitive TNALP inhibitor, capable of inhibiting TNALP's pyrophosphatase activity at physiological pH, thereby suppressing calcification in primary cultures of VSMCs as well as in aortic explants.



All routine chemicals were of analytical grade from Sigma (St Louis, MO, USA), unless otherwise indicated.

Expression and preparation of test enzymes

Expression plasmids containing a secreted epitope-tagged TNALP, PLALP, and IALP were transfected into COS-1 cells for transient expression using a standard electroporation method. Medium was replaced to Opti-MEM 24 h later, and the serum free media containing secreted proteins were collected 60 h after electroporation. Conditioned medium was dialyzed against TBS containing 1 mM MgCl2 and 20 μM ZnCl2 (to remove phosphate) and filtered with a 2-mm cellulose acetate filter. The TNALP stock solution was obtained from the dialyzed conditioned medium. The PLALP and IALP solutions were produced in the same way. To compare the percentage of inhibition, dilutions of PLALP- and IALP-conditioned media were adjusted to obtain the same value of activity as TNALP-conditioned medium without inhibitor.

High-throughput screening

The TNALP stock solution was diluted 120-fold, and 12 μl of diluted TNALP solution was dispensed into 96-well microtiter plates with half area bottom (Costar, Corning, NY, USA) by an auto dispenser (Matrix, Hudson, NH, USA). The maximum volume in each well was 190 μl (well depth, 10.54 mm; well bottom area, 0.1586 cm2). The library compounds were dissolved in 100% dimethylsulfoxide (DMSO) in the master plates, and our working plates contained 10% DMSO, giving 1% DMSO in the final enzymatic reaction. Whereas 10% DMSO inhibits TNALP by about 30%, the final 1% DMSO concentration does not affect TNALP activity. A robotic liquid handler, Biomek FX (Beckman Coulter, Fullerton, CA, USA) dispensed 2.5 μl of each compound (dissolved in 10% DMSO) from the library plates. Plates were incubated at room temperature for at least 1 h to allow TNALP to interact with each compound before addition of 10.5 μl substrate solution (1.19 mM pNPP). After 30-min incubation, A405 nm was measured with a microtiter plate reader, Analyst HT (Molecular Devices, Sunnyvale, CA, USA). Both the enzyme (TNALP) and substrate (pNPP) solution were made in diethanolamine (DEA) buffers; the final reaction consists of 1 M DEA-HCl buffer, pH 9.8, containing 1 mM MgCl2 and 20 μM ZnCl2. The concentration of TNALP and pNPP (final 0.5 mM) were adjusted to obtain A405 nm ∼0.4, while maintaining good sensitivity to the known inhibitors levamisole and phosphate, used as positive controls. Km obtained with a 1/120 dilution of TNALP and a fixed incubation period of 30 min was 0.58 ± 0.081 mM.

Enzyme kinetic experiments

To determine the inhibition selectivity for inhibitor candidates, human TNALP, PLALP, or IALP was added to microtiter plates followed by addition of the substrate pNPP (0.5 mM), and activity was measured in 1 M DEA-HCl buffer, pH 9.8, or in 1 M Tris-HCl buffer, pH 7.5,(26) containing 1 mM MgCl2 and 20 μM ZnCl2, in the presence of potential inhibitors (0–30 μM). TNALP, PLALP, and IALP activities were adjusted to an approximate ΔA405 nm, equivalent to 1, measured after 30 min. Residual ALP activity in the presence of inhibitors was expressed as percentage of the control activity. To study the mechanism of inhibition, double reciprocal plots of enzyme activity (expressed as mA405 nm min−1) versus substrate concentration were constructed in the presence of various concentrations of added inhibitors (0–30 μM). The y-axis intercepts of the 1/v versus 1/[S] plots were plotted versus [I] to graphically extract Ki values as the x-intercept in this plot. The numerical values from y- and x-intercepts were derived using linear regression analysis, using software Prism 3.02 (GraphPad Software). These analyses were performed, using pNPP as a substrate in 1 M DEA-HCl buffer, pH 9.8, as well as in 1 M Tris-HCl buffer, pH 7.5, to determine Ki at optimal and physiological pH, respectively. The TNALP concentration in those experiments was chosen to generate a ΔA405 nm, equivalent to 0.3, measured after 1 h, to allow a linear increase of A405 nm with time, for the lowest substrate concentration tested ([pNPP] = 100 μM). Inhibitors were further tested and sorted based on their kinetic properties at pH 7.4 using PPi, the relevant natural substrate of TNALP.(27) In this part of the study, pyrophosphate sodium salt (99% ACS reagent; Sigma-Aldrich, St Louis, MO, USA) was used as a substrate. Amounts of released phosphate were measured using the Biomol Green Reagent (Biomol Research Laboratories, Plymouth Meeting, PA, USA). Finally, to document the potency of selected inhibitors in physiological media, TNALP inhibition by compound 5804079 (0–30 μM) was studied at pH 7.4, during catalysis of 0.1 mM pNPP, in the presence of increasing concentrations of Na2HPO4 (0–10 mM) and pyrophosphate (0–40 mM).

Computer docking

Compound docking was performed using the Flexx program, part of the Sybyl package from Trios, as before.(28) Formal charges were used for protein and compound atoms. Heteroatoms (phosphate, zinc, and magnesium) were considered as part of the pocket while docking.

Maintenance of Enpp1−/− and ank/ank mice

The generation and characterization of Enpp1−/− mice has been described earlier.(29) Mice carrying the ank mutation were purchased from The Jackson Laboratory. To determine genotypes, genomic DNA was isolated from tails and analyzed using PCR protocols.(12)

Tissue preparation and morphological analysis

Whole mount skeletal preparations were prepared by removal of skin and viscera of mice followed by a 1-wk immersion in 100% ethanol, followed by 100% acetone. Samples were transferred to a 100% ethanol solution containing 0.01% Alizarin Red S, 0.015% Alcian Blue 8GX, and 0.5% acetic acid for 3 wk. Samples were destained with 1% (vol/vol) KOH/50% glycerol solution. Cleared samples were stored in 100% glycerol.(12)

Isolation and culture of primary calvarial osteoblasts and VSMCs

Mouse calvarial cells were isolated from 3-day-old WT mice through sequential collagenase digestion, as previously described.(12,15) Vascular smooth muscle cells (VSMCs) were isolated from explants using a collagenase digestion method, and the smooth muscle phenotype was confirmed by RT-PCR analysis for smooth muscle α-actin. One mouse aorta provided on average 5 × 105 cells. These cells were cultured (in triplicate) at a density of 3 × 104 cells/cm2 using α-MEM supplemented with β-glycerophosphate (10 mM) and 50 μg/ml ascorbic acid for 3 wk. Media was renewed every third day and inhibitors were freshly prepared and added each time. To quantify calcium deposited in these cultures, either the o-cresolphthalein complexone method(16) or the standard Alizarin Red method(15) was used.

Reverse transcription and quantitative real-time PCR

Total RNA was extracted from the osteoblast and VSMC pellets and 2 μg of RNA used for reverse transcription using the Superscript kit (Invitrogen, Carlsbad, CA, USA). OPN mRNA was quantified by real-time PCR using dual-labeled hydrolysis probes (FAM-TAMRA). The sequences for mouse OPN and 18S primers and probes were as follows: OPN forward 5′-TGAGGTCAAAGTCTAGGAGTTTCC-3′, OPN reverse 5′-TTAGACTCACCGCTCTTCATGTG-3′, OPN Probe 5′-TTCTGATGAACAGTATCCTG-3′, 18S forward 5′-CGGCTACCACATCCAAGGAA-3′, 18S reverse 5′-GCTGGAATTACCGCGGCT-3′ and 18S Probe 5′-TGCTGGCACCAGACTTGCCCTC-3′. For quantitative real-time PCR, 2 μl of the cDNA and the reaction mixture used 12.5 μl of platinum qPCR UDG supermix (Invitrogen). The reaction was performed in a 96-well plate on a Stratagene MX2000P real time machine (Stratagene, La Jolla, CA, USA). The reaction was run at an initial temperature of 95°C for 10 min and at 95°C for 30 s, 55°C for 1 min, followed by 72°C for 30 s for 45 cycles. Ct values were determined by the MX2000P Software according to the optimization of the baseline. For computing the relative amount of OPN in the samples, the average Ct for 18S was subtracted from that of OPN to give changes in Cts (ΔCt). Relative units (log2 ΔCts) were calculated and used as a measure of OPN expression.

PPi hydrolysis by whole aortas ex vivo

Sprague-Dawley rats were killed, and aortas perfused with Hanks salt solution to remove blood. The aortas were removed and, after the adventitia was dissected away, were cut into rings ∼2 mm in length. Four rings were placed in 1 ml of DMEM without serum containing the compounds to be tested. After 90 min at 37°C, sodium PPi (final concentration, 1 μM) and [32P]PPi (final concentration, 1 μCurie/ml) were added, and six samples were removed over 4 h.(6) Pi was separated from PPi by adding 800 μl of 0.028 M ammonium molybdate in 0.75 M H2SO4 to the samples and extracting with 1600 μl of isobutanol and petroleum ether (4:1). 32P was counted in the organic phase by Cerenkov radiation. Hydrolysis of PPi was linear over 4 h, and the rate was determined by linear regression.


Aortic calcification in Enpp1−/− and ank/ank mice is associated with elevations in TNALP expression

Given the coordinated function of NPP1 and ANK in establishing extracellular PPi concentrations and the similarity of the calcification abnormalities found in Enpp1−/− and ank/ank mutant mice,(12) it was to be expected that the similarities would also extend to the arterial calcification sites.(7) We dissected whole mount preparations of Enpp1−/− and ank/ank heart and aorta and stained them with Alizarin red to visualize calcium deposition. Figure 1A clearly shows the presence of multiple foci of aortic calcification in Enpp1−/− mice, whereas none are evident in control mice. The same qualitative results were obtained for the ank/ank mice (data not shown). We quantified the amount of calcium deposited in wildtype (WT), Enpp1−/−, and ank/ank aortas. The data, using mice at 3 mo of age, clearly showed a higher degree of calcification in Enpp1−/− and ank/ank compared with WT control animals. We also found more calcification in Enpp1−/− mice than in ank/ank mice (Fig. 1B) in agreement with the more severe calcification phenotype that we observed in the Enpp1−/− mice.(12)

Figure Figure 1.

Vascular calcification in the Enpp1−/− and ank/ank mouse models. (A) Three-month-old WT and Enpp1−/− mice were dissected free from adherent tissue with the exception of the heart and aorta. The ribcage, heart, and aorta were fixed in 100% ethanol, and unmineralized osteoid was stained with Alcian blue followed by Alizarin red staining for mineralized osteoid. The samples were cleared in 2% KOH and stored in 100% glycerol. The top panels show low-magnification images of the ribcage with heart and aorta attached and outlined in yellow. The arrow points to several foci of calcification as revealed by positive staining (red) for calcium in the aorta in the Enpp1−/− sample. The foci are better observed at higher magnification in the middle panel. In the higher magnification panels, the aorta has been dissected away from the spine and the presence of calcium deposits is clearly seen in the Enpp1−/− specimen. (B) Quantification of the amount of calcium present in aortas or in serum of 3-mo-old WT, Enpp1−/−, and ank/ank mice (*p = 0.0022). (C) VSMCs isolated from WT, Enpp1−/−, and ank/ank mice were cultured in the presence of β-glycerophosphate and ascorbic acid for 3 wk. Cells were stained for TNALP activity (pink), and using von Kossa staining, mineral was detected (black/brown). The area of the culture well in which mineralization was present was quantified by using a point-counting method in which the plate was placed on a grid (divided into 10 × 10-mm squares) and visualized using a dissecting microscope; the percent area occupied by mineral was assessed by counting the occurrences where the presence of mineral coincided with intercepts on the grid (***p < 0.0001). (D) Quantification of the levels of TNALP activity in WT, Enpp1−/−, and ank/ank VSMCs.

Given that arterial calcification is more severe in Enpp1−/− than in ank/ank mice, we chose Enpp1−/− mice for subsequent in vitro experiments to determine the putative involvement of TNALP in the ectopic calcification process. Using a collagenase digestion method, we therefore isolated VSMCs and identified them by immunofluorescence and RT-PCR detection of SMC α-actin. Hence, we obtained a population of cells in which, on average, 89% stained positive for SMC α-actin (data not shown). Using these VSMC cultures, we first determined that WT VSMCs express TNALP activity; second, that WT VSMCs when cultured in the presence of β-glycerophosphate and ascorbic acid can lay down mineral in a manner, and with kinetics, similar to that of osteoblast cultures(30); and third, and most importantly, that VSMCs from Enpp1−/− and ank/ank mutant mice produce significantly more mineral than WT cells (Fig. 1C) and express a higher level of TNALP activity than WT cells (Fig. 1D). We surmised that by inhibiting this upregulated pyrophosphatase TNALP activity, we would be able to restore the normal ePPi levels, which in turn would contribute to suppressing HA deposition in the vasculature. However, the currently available inhibitors of TNALP (i.e., levamisole or theophylline) are weak inhibitors and do not adequately suppress the pyrophosphatase activity of TNALP at physiological pH.(28) To be able to do this efficiently, we undertook the screening of comprehensive chemical libraries to identify and characterize novel lead compounds that could enable the development of potent drug-like inhibitor of TNALP's physiological pyrophosphatase function.

High-throughput screening

To identify such novel small molecule inhibitors of TNALP activity, we optimized an assay to screen chemical libraries containing 53,280 compounds. These included (1) the Spectrum Collection (from MicroSource,, containing 2000 compounds (25 plates, 80 compounds/plate); about one half of the collection contains known bioactive agents, permitting the evaluation of hundreds of marketed drugs and biochemical standards; the other half of the collection includes pure natural products and their derivatives; (2) the LOPAC1280 Collection (, containing 1280 pharmacologically active compounds; this library contains effector molecules for all major target classes and all of the compounds in this collection are available for powder resupply from SIGMA; and (3) the Chembridge DIVERSet Collection (from Chembridge, that contains 50,000 diverse, predesigned compounds (625 plates, 80 compounds per plate); this collection was selected by a “rational” approach based on 3D pharmacophore analysis to cover the broadest part of biologically relevant pharmacophore diversity space.

Screening the chemical libraries was based on a 96-well plate assay using 0.5 mM pNPP as substrate. We used 30 μM of the uncompetitive inhibitor levamisole and 300 μM of the competitive inhibitor Pi in each individual assay plate as positive controls. The concentration of the chemical library compounds in the reaction mixture was ∼10 μM. Even though some of the compounds in the libraries absorb at 405 nm, monitoring p-nitrophenol production proved to be more sensitive in our hands than other methods, detecting liberated Pi through alternative colorimetric assays, such as the Biomol green reagent that absorbs at 620 nm. Instead, any compound with inherent A405 nm absorption was retested manually in the laboratory, at various concentrations, to complement the single point assay of the robotic station. After each daily run of assay, we manually tested all those compounds that had shown >20% inhibition. A total of 11 hits with reproducible inhibition were retested, and finally, four compounds were identified as effective TNALP inhibitors: one was levamisole, a well-known albeit weak ALP inhibitor, contained within the 2000 Spectrum Collection of known drugs and presently used as a positive control during our screening. The other three represented novel structures as shown in Fig. 2. The physicochemical property of these compounds is summarized in Table 1. All three compounds conform to Lipinski's rule of 5 ( they have a molecular weight of <500, have <5 H-bond donors; have <5 H-bond acceptors; have <10 rotational bonds, and have an octanol/water repartition coefficient (LogP) ≤ 5. Their nitrogen content ranges from 3 to 7 N atoms per inhibitor (Fig. 2).

Table Table 1.. Structural Properties of Presently Identified Novel TNALP Inhibitors
original image
Figure Figure 2.

Chemical structure of the three lead compounds that inhibit TNALP activity, identified through high-throughput library screening.

Kinetic properties of the inhibitors

None of the three identified novel TNALP inhibitors seemed to inhibit, either at pH 9.8 or at physiological pH, other relevant human ALPs, such as PLALP or IALP, which share 50% and 52% sequence identity with TNALP, respectively. Figure 3 shows the inhibition of TNALP, PLALP, and IALP for increasing concentrations (0–30 μM) of the inhibitors at physiological pH. Furthermore, none of the inhibitors had any effect on PHOSPHO1, a novel phosphatase proposed to be involved in the initiation of MV-mediated calcification.(31) The double reciprocal plots of 1/v versus 1/[S], for various inhibitor concentrations, showed parallel lines for all three inhibitors, indicating that each TNALP inhibitor acts in an uncompetitive manner, both at pH 9.8 and at physiological pH. Figure 4A shows the Lineweaver-Burk plot of 1/v versus 1/[S] for compound 5804079 at pH 9.8 and pH 7.5. The Eadie-Hofstee plot at pH 9.8 confirms that the mechanism is uncompetitive, because all lines intersect on the y-axis and show intersections on the x-axis, proportional to the concentration of inhibitor (Fig. 4A). For these reasons, all the mathematical analyses were made through double reciprocal plots. Secondary replots of the y-intercepts (Fig. 4B) therefore enabled determining Ki, describing the potency for each inhibitor at pH 7.5. Compound 5804079 had the lowest Ki value at physiological pH (i.e., is ∼6-fold more potent than the frequently used inhibitor levamisole; Table 2). In addition, it is more potent at pH 7.5 than at pH 9.8.

Table Table 2.. Inhibition Constants for Levamisole and the Three Novel TNALP Inhibitors
original image
Figure Figure 3.

Specificity of inhibition of TNALP vs. IALP and PLALP as a function of the concentration of each novel inhibitor, as indicated, measured at pH 7.5. The TNALP stock solution was obtained from the dialyzed conditioned medium. The PLALP and IALP solutions were produced in the same way. To compare the percentage of inhibition, dilutions of PLALP- and IALP-conditioned media were adjusted to obtain the same value of activity as TNALP condition medium without inhibitor.

Figure Figure 4.

Kinetic analysis of TNALP inhibition. (A) Double-reciprocal plot of v vs. substrate concentration at both pH 9.8 and 7.4 and Eadie-Hofstee plot at pH 9.8 for compound 5804079. (B) y-intercept replots and graphical determination of Ki, as indicated, for each of the three inhibitors, derived from 1/v vs. 1/[pNPP] plots made at pH 7.5 for increasing concentrations of the indicated inhibitors.

Figure 5A shows that the potency of compound 5804079 is not affected by the presence of the competitive inhibitor Pi at concentrations largely exceeding those for inhibitor or substrate. Figure 5B shows that the degree of inhibition by compound 5804079 is not affected by high concentrations of PPi, in agreement with the uncompetitive nature of this inhibitor, which does not have to compete with Pi or PPi for binding to the enzyme but only binds to the phospho-enzyme complex, once it is formed.(32)

Figure Figure 5.

Residual TNALP activity at pH 7.5, measured in the presence of 0.1 mM pNPP as a function of the concentration of compound 5804079, in the absence (○) or presence (•) of 10 mM Pi (A) or 40 mM PPi (B).

Docking of the inhibitors in the TNALP active site

We have recently documented the likely positioning of three well-known inhibitors of ALP activity (i.e., l-homorginine, levamisole, and theophylline), in the active site of TNALP.(28) We found two distinct areas in the TNALP active site able to accommodate inhibitors; the first, comprising residues R433 and H434, accommodates hydrophobic ringed structures such as levamisole and theophylline, whereas the second, comprising residues E108/G109, can accommodate more hydrophilic extended inhibitors such a l-homoarginine. Compatible with those data, we found that two of the three newly identified compounds predominantly dock into the R433/H434 region of the binding site (Fig. 6). Interestingly, however, the most potent compound 5804079 seems to dock in a manner that spans both binding areas. This may in part explain the low Ki for this compound and its better performance at pH 7.5.

Figure Figure 6.

Computer docking of the three lead compounds into the active site of the modeled structure of TNALP. The top left panel shows a detail of all the active site residues involved in stabilization of inhibitors as previously determined.(28) The other three panels shows a close up view of each novel inhibitor docked in the active site environment.

TNALP inhibition abrogates calcification and pyrophosphatase activity of VSMCs

To validate the inhibitory potential of all three inhibitors on in vitro calcification, we tested the ability of all three novel compounds, using levamisole as control, to inhibit TNALP activity in primary calvarial osteoblast cultures and in Enpp1−/− VSMCs. As shown in Fig. 7, all four compounds inhibited mineralization to some degree in both culture systems, with compound 5804079 being the most effective in both in vitro assays. As expected,(12,17) the inhibitor treatment induced OPN mRNA expression in primary calvarial osteoblast cultures but had the opposite effect on OPN expression in Enpp1−/− VSMC cultures.

Figure Figure 7.

Effect of the three inhibitors on in vitro calcification by WT calvarial osteoblasts and Enpp1−/− VSMCs. Primary calvarial osteoblasts and Enpp1−/− VSMCs were grown in the presence of β-glycerophosphate, ascorbic acid, and either compound 5361418, 5923412, or 5804079, and compared with the relative effect of levamisole, at 30 μM concentration. Extractions and analyses were done after 3 wk. One-way ANOVA analysis indicated p < 0.0001 for the osteoblast experiment and p = 0.0006 for the VSMC experiment. Parallel cultures of osteoblasts and VSMCs were used to extract mRNA to measure changes in OPN mRNA expression after these treatments. Each histogram plots the mean ± SE of six determinations.

To measure the degree of pyrophosphatase inhibition by the new TNALP inhibitors, we used an ex vivo organ culture system in whole aortas. For this analysis, rat rather than mouse aortas were selected, because they are larger and easier to dissect. This analysis also showed that compound 5804079 was the most effective in suppressing endogenous pyrophosphatase activity at the site of vascular calcification (Table 3) at the maximal concentration of 30 μM (chosen for all these highly aromatic inhibitors to avoid solubility problems), i.e., 40% inhibition. Given that approximately one half of the pyrophosphatase activity found in aortic tissue is attributable to TNALP (WC O'Neill, unpublished data), this ex vivo assay indicates that compound 5804079 is able to pharmacologically ablate ∼80% of TNALP's pyrophosphatase activity in the aortic rings.

Table Table 3.. Reduction of the Rate of Hydrolysis of PPi in Aortic Rat Rings by Levamisole and the Novel TNALP Inhibitors
original image


According to the World Health Organization (WHO) an estimated 17 million people die every year of cardiovascular diseases, particularly heart attacks and strokes. Vascular calcification is a major cause of cardiovascular morbidity and mortality. It is commonly found in atherosclerotic lesions and has been associated with increased accumulation of macrophages in the chronically inflamed atherogenic vessel wall,(33) but also with enhanced vascular TNALP activity. At the cellular level, calcifying vascular cells have the potential to undergo osteoblastic differentiation and mineralization,(34–36) and it has been recently shown that the adult artery wall also contains mesenchymal progenitor cells with myogenic and chondrogenic potential.(37) Matrix vesicles (MVs), the membrane limited chondroblast- and osteoblast-derived structures in which the process of mineralization is initiated,(38,39) have also been documented in calcified atherosclerotic lesions.(22,40,41)

These findings document that murine VSMCs produce TNALP in culture in sufficiently large amounts to play a role in the hydrolysis of PPi, which acts as an inhibitor of mineralization. Accordingly, the elevated levels of TNALP in Enpp1−/− and ank/ank VSMCs, compounded with the inherent deficient production/transport of ePPi in these NPP1- and ANK-deficient cells, respectively, provide the basis to explain excessive arterial calcification in these animal models. We tested whether inhibition of vascular TNALP through potent new inhibitors can raise PPi concentrations to levels capable of blocking ectopic vascular calcification in these genetic models of medial calcification. Whereas a number of ALP inhibitors have been described in the literature, none of them are entirely specific for TNALP, and some of the best, such as levamisole and theophylline, have effects in vivo that are unrelated to ALP function. Levamisole is an FDA-approved drug used in patients as an adjuvant treatment for colon cancer(42,43) and the drug has also been used as an anti-helmintic in animals.(44–46) Theophylline is a bronchodilator approved for the treatment of asthma. Thus, there is a need to develop specific TNALP inhibitors with drug-like properties suitable for further development toward in vivo therapeutics for vascular calcification. We presently report how the use of high-throughput library screening has identified three novel lead compounds able to inhibit TNALP with high affinity and specificity. We validated the usefulness of these compounds kinetically and for their ability to prevent calcification in primary cultures of osteoblasts and Enpp1−/− VSMCs and through their inhibition of TNALP's pyrophosphatase activity in whole rat aortas ex vivo.

The presently selected inhibitors are very aromatic molecules; all three inhibit TNALP through an uncompetitive mechanism, both at pH 9.8 and at physiological pH. The small differences in Ki measured at pH 9.8 and 7.5, for two of the three inhibitors, imply that inhibitor positioning is not ionic, in accordance with the high number of aromatic nitrogen atoms in their backbone structure. The most potent inhibitor (i.e., compound 5804079) is even more potent at pH 7.5 and can be modeled in the active site pocket of TNALP in a manner that spans the large enzyme pocket. The uncompetitive nature of enzyme neutralization has the additional advantage that inhibition of TNALP can be achieved by micromolar concentrations of inhibitor, even on a background of millimolar concentrations of Pi and PPi. However, we expect that future medicinal chemistry efforts on these compounds will enable us to optimize their chemical structure to increase solubility and improve affinity even further, to enable in vivo trials.

When compound 5804079 was used to treat calvarial osteoblasts cultures, we observed the expected increase in OPN mRNA expression that is secondary to the increase in ePPi concentrations resulting from inhibiting TNALP's pyrophosphatase activity. This is entirely in agreement with our previous data, indicating that the expression of these two potent calcification inhibitors is strictly correlated in skeletal tissue.(12,16,17) However, this TNALP inhibitor had the opposite effect on Enpp1−/− VSMCs. The reasons for this apparent misregulation are not clear at this moment, but given that these are genetically abnormal cells where the production of ePPi is affected, it is possible that ePPi and OPN, which normally act as counter-regulatory mechanisms balancing each others concentrations in skeletal cells,(17) may not be operating normally in these knockout cells, which already have reduced levels of OPN.(7) Nevertheless, when added to Enpp1−/− VSMCs, compound 5804079 potently reduced the cell-mediated calcification process.

The possibility that treatment with TNALP inhibitors may have secondary deleterious effects on skeletal tissues needs to be considered. TNALP activity is highest in skeletal tissues undergoing growth spurts (children and adolescents) or during the repair of fractures or in cases of osteoblastic metastasis. An adult individual only has steady-state levels of TNALP activity needed for bone homeostasis. Given that the amount of TNALP expressed by VSMCs at the site of arterial calcification is upregulated compared with the steady-state levels of TNALP in osteoblasts, one would expect VSMC activity to be selectively sensitive to inhibitor treatment. However, in the event that TNALP inhibitor treatments would lead to adverse side effects in bone or other tissues and that this toxicity could not be manipulated by dosage or route of administration, targeted delivery strategies can be contemplated to achieve a high concentration of TNALP inhibitors only at the site of vascular calcification.

In conclusion, this study showed that potent inhibitors of TNALP pyrophosphatase activity can be obtained by systematic chemical library screening efforts. These compounds have the potential to be used to treat and/or prevent ectopic soft tissue calcification. The lead compounds identified in this paper will serve as the foundation for thorough structure-activity relationship studies to optimize these novel inhibitors for optimal solubility and pharmacokinetic properties with the goal of testing them in vivo in diverse experimental models of vascular calcification.


The authors thank Alexey Eroshkin for help with the docking predictions; Steve Vasile for help with the high-throughput chemical library screening; and Soetkin Van kerckhoven for help with kinetic measurements. This work was supported by Grants DE12889, AR47908, and DK06981 from the National Institutes of Health and KULeuven Grant GOA 2004/09.