Medial vascular calcification (MVC), also known as Monckeberg's sclerosis, is characterized by the deposition of hydroxyapatite (HA) crystals within the medial layers of arteries and is increasingly recognized as an important clinical problem.1, 2 For decades, calcification of the cardiovascular system was considered an uncommon condition limited to advanced senile arteriosclerosis. However, its prevalence corresponds roughly with age, with coronary artery calcification occurring in 20% of young adults, 60% in the middle age group, and 90% of the elderly.3 Although vascular calcification is a major contributor to cardiovascular disease (CVD) and death in patients with chronic kidney disease (CKD), there is no established treatment for this morbid condition.4, 5
MVC follows an active endochondral ossification-like process, resulting in the formation of bone marrow and cartilage.6, 7 The process involves chondro-osteoblastic conversion of vascular smooth muscle cells (VSMCs),8, 9 evident from the loss of characteristic smooth muscle cell markers and development of osteoblastic features, such as expression of tissue-nonspecific alkaline phosphatase (TNAP), osteocalcin (BGLAP), and osteopontin (SPP1), and osteocyte markers such as sclerostin and podoplanin.10–14 It has been hypothesized that phosphate accelerates this chondro-osteogenic conversion by inducing expression of runt-related transcription factor 2 (RUNX2), osteocalcin, and TNAP.15 Elevated TNAP further favors vascular calcification by hydrolyzing the calcification suppressor inorganic pyrophosphate (PPi).16–19 The importance of PPi has been illustrated in mouse models, deficient in ectonucleotide pyrophosphatase/phosphodiesterase 1 (Enpp1–/–), or in the PPi-channeling function of ANK (ank/ank). Both models are characterized by decreased levels of extracellular PPi, caused by a combination of deficient output and increased degradation by TNAP, and exhibit a phenotype of hypermineralization.20 We have shown in Enpp1–/– that inhibition of TNAP by potent inhibitors can restore PPi to sufficient levels to maintain normal mineralization.16
The phosphatase PHOSPHO1, first identified in the chicken as a member of the haloacid dehalogenase superfamily of Mg2+-dependent hydrolases,21–24 is expressed in chondrocytes in mineralizing cartilage at levels 120-fold higher than in nonmineralizing tissues.25 PHOSPHO1 shows high phosphohydrolase activity toward phosphoethanolamine (P-Etn) and phosphocholine (P-Cho), both of which are key components of the phospholipids of matrix vesicles (MVs).26 Using lansoprazole as a pharmacological inhibitor of PHOSPHO1, we demonstrated that PHOSPHO1 is present and active inside chondrocyte- and osteoblast-derived MVs.25 However, lansoprazole was identified from the Library of Pharmacologically Active Compounds (LOPAC1280) and, as expected, inhibits a number of proteins other than PHOSPHO1, including TNAP.27, 28 More recently, we showed that PHOSPHO1 and TNAP expression coincide during skeletal mineralization29 and using single- and double-knockout mice, we demonstrated that PHOSPHO1 controls TNAP expression in mineralizing cells and is essential for mechanically competent mineralization.30, 31 Furthermore, PHOSPHO1/TNAP double-knockout mice show complete absence of skeletal mineralization.30
Ultrastructural studies have identified HA-containing MVs in human aorta, which indicates that these structures may provide the nidus for vascular calcification.32, 33 In this study, we provide the first description of the role of PHOSPHO1 in the calcification of VSMCs. We show that inhibition of PHOSPHO1 activity can reduce calcification in hypermineralizing wild-type (WT) VSMCs and that the combined use of selective PHOSPHO1 and TNAP inhibitors considerably reduces calcification in these cells, indicating that “phosphatase inhibition” constitutes a viable approach for the prevention and treatment of MVC.
Materials and Methods
Isolation and culture of primary WT, Phospho1–/– and Enpp1–/– VSMCs
VSMCs isolated from WT, Enpp1–/–, and Phospho1–/– mice were used for in vitro calcification studies. The VSMCs were isolated from excised aortas using a collagenase digestion method and the smooth muscle phenotype was confirmed by RT-PCR analysis for smooth muscle α-actin as reported.16 One mouse aorta provided an average of 5 × 105 VSMCs. These cells were cultured (in triplicate) at a density of 0.25 × 105 cells/mL/well in a 24-well plate using α modified essential medium (α-MEM) supplemented with 50 µg/mL ascorbic acid and 2.5 mmol/L β-glycerophosphate or 3 mmol/L sodium phosphate to induce calcification. Cells were cultivated in the mineralization media for up to 4 weeks and was changed every second day. Inhibitors were freshly dissolved in dimethylsulfoxide (DMSO) and added at each medium change at a final concentration of 30 µmol/L.
Analysis of gene expression
RNA was extracted using an RNAeasy Plus Kit (Qiagen, Valencia, CA, USA), according to the manufacturer's instructions. RNA was reverse-transcribed and specific cDNAs were quantified by real-time PCR using dual-labeled hydrolysis probes (FAM-TAMRA) as reported.14, 30 Primers and probes were obtained from Eurogentec North America (San Diego, CA, USA) and their sequences are provided in the online Data Supplement.
Expression and preparation of test enzymes
A construct for expression of the human bone-specific PHOSPHO1 isoform was generated by ligating a PCR fragment encoding the 40 bone-specific N-terminal amino acids, a partial fragment of cDNA (GenBank accession BC117187) encoding the common isoform, and a C-terminal polyhistidine tag into the pCMV-SCRIPT vector. The expression vector was transfected into HEK293 cells, and recombinant bone isoform PHOSPHO1 protein was purified by a standard procedure.26 Soluble epitope-tagged human TNAP and ENPP1 was produced and purified as described.16
High-throughput screening (HTS) of 55,000 compounds from the Molecular Libraries Small Molecule Repository (MLSMR) compound collection was conducted using a colorimetric assay based on the ability of PHOSPHO1 to liberate phosphate from P-Etn and its reaction with the Biomol Green reagent (Biomol International, Plymouth Meeting, PA, USA). HTS provided approximately 5000 compounds that showed greater than 50% activity in the single point assay, a hit rate of 3%. Subsequent hit follow-up and validation in dose-response identified submicromolar inhibitors of PHOSPHO1 (PubChem BioAssay AID 1666). Initial HTS was performed in duplicate at a concentration of 20 µmol/L with dose-response assays using a 10-point twofold serial dilution of the hit compounds in DMSO. Hit confirmation was performed using the Biomol Green colorimetric assay to verify inhibitory activity against PHOSPHO1 in dose-response mode. Compounds that were active in dose-response mode against PHOSPHO1 and soluble in the range relevant to their potency were prioritized for synthetic chemistry follow-up to increase selectivity against ENPP1 and TNAP and/or potency against PHOSPHO1.
Reactions were measured in triplicate in 96-well plates containing 20 mmol/L MES-NaOH, pH 6.7, 0.01% (wt/vol) bovine serum albumin (BSA), 0.0125% (vol/vol) Tween-20, 2 mmol/L MgCl2, and 62.5 µmol/L P-Etn or P-Cho (Sigma-Aldrich, St. Louis, MO, USA) test compound dissolved in DMSO (ranging from 0.0143 nmol/L to 300 µmol/L) and 0.3 nmol/L of purified recombinant human PHOSPHO1, in a total volume of 50 µL. Each reaction contained a final concentration of 1% (vol/vol) DMSO, which was determined to be insufficient to interfere with PHOSPHO1 activity (data not shown). Enzyme was preincubated with the inhibitors and the reaction was initiated by addition of substrate 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (P-Eth; Sigma-Adrich) or phosphocholine chloride calcium salt tetrahydrate (P-Chol; Sigma-Aldrich) and allowed to proceed for 60 minutes at room temperature. Reactions were stopped with 100 µL Biomol Green reagent and absorbance was measured at 620 nm, after which the inhibitory effect of each compound was calculated as a percentage of control. The one-half maximal inhibitory concentration (IC50) was calculated from plots of residual enzyme activity against inhibitor concentration, using Origin Scientific plotting software (Northampton, MA, USA). Single-point experiments with inhibitors at 30 µmol/L were performed to test selectivity for PHOSPHO1 inhibition versus TNAP inhibition, at those concentrations used during VSMC calcification, as described.16
The inhibition of recombinant human TNAP, and three TNAP variants (Tyr371Ala, Glu108Phe, and His434Glu), with mutations previously implicated in the active site positioning of the classic TNAP inhibitor levamisole,34 by the TNAP inhibitor MLS-003894935 was carried out at pH 7.5, as reported.16 The nature of the inhibition was investigated using double reciprocal plots of velocity (v) versus substrate concentration, at various inhibitor concentrations (0–1 µmol/L) and secondary replots. Likewise, the nature of the inhibition of PHOSPHO1 by the selected inhibitors was determined via double reciprocal plots of v versus the concentration of P-Etn.
Computer modeling and docking
To complement the kinetic studies on PHOSPHO1 inhibition by both classical inhibitors such a lansoprazole25 and the novel inhibitors described here, we modeled the protein and performed computer docking. Several protein models were obtained using a number of homology modeling web servers. The PHOSPHO1 protein sequence was extracted from the NCBI Protein database and submitted for homology modeling using FFAS03,36 ITASSER,37 and SWISSModel.38 Retrieved models were superposed and regions of high structural homology (consensus structure) were defined. The mobile regions were examined and the models were grouped into conformational classes as judged by the root-mean-square deviation (RMSD). The models showed divergence in two areas: two helices and two loops closing the active site. Structural representatives of at least three classes of models were extracted and used for docking using the SwissDock server39 and the program EAdock.40 A single model of PHOSPHO1, produced by ITASSER, was selected because all other classes had hindered access to the active site. Docking was performed using this model with selected PHOSPHO1 inhibitors. A total of 150 hits with the highest pseudo-energy were retained in the search. Subsequently, clustering analysis was performed and the hits were clustered into spatially matching hits. The representatives of these families show different binding modes for individual compounds.
Calcium deposition in cultured VSMCs in the presence or absence of inhibitors was evaluated by staining cell layers with Alizarin Red, as reported.16 Cells were washed twice with both PBS and 70% ethanol, then fixed in 100% ethanol for 5 minutes. Cells were stained with 2% Alizarin Red (pH 4.2) for 30 minutes at room temperature and washed with distilled water. Alizarin Red–stained cultures were extracted with 10% cetylpyridinium chloride for 45 minutes and was measured. The matrix was also decalcified in 0.6 N HCl for 24 hours, and free calcium was determined colorimetrically using a commercially available kit (Randox Laboratories, Crumlin, UK), after correction for total protein concentration. The protein content of the cultures was measured using Bio-Rad protein assay reagent (Bio-Rad Laboratories, Hertfordshire, UK) based on the Bradford dye procedure, and gamma globulin was used as a standard (Fig. 1C, D). For experiments on the effect of inhibitors on mineralization of cultured WT, Enpp1–/–, and Phospho1–/– VSMCs, the cell layers were stained with Alizarin Red, extracted with 10% cetylpyridinium and A570 was measured.
3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide reduction assay for cell viability
Cells were cultivated for 7 days and treated with 30 µmol/L of TNAP and/or PHOSPHO1 inhibitors as described. To determine cell viability, the growth medium was removed and cells were incubated with 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT; Sigma-Aldrich) at a final concentration of 0.5 mg/mL in phenol-red-free DMEM containing 10% heat-inactivated fetal bovine serum (FBS) at 37°C in the dark for 3.5 hours. An equal volume of the dissolving buffer (22% (wt/vol) SDS in 0.02 M HCl) was added to the wells with gentle trituration. Cultures were incubated with the dissolving buffer at 37°C in the dark for 4 hours. After the dye was completely dissolved, 200 µL of the solution was transferred to a 96-well plate and absorbance was recorded at 570 nm.
All data are expressed as the mean ± SEM. Differences in relative RNA expression for Phospho1 and Alpl in WT VSMCs and for calcium deposition in Phospho1–/– versus WT VSMCs were analyzed with an unpaired t test with Welch correction and calculation of a two-tailed p value. Differences in the degree of VSMC calcification and multiple comparisons of the relative RNA expression data were also analyzed by Student's t test. The upregulation of Phospho1 in VSMCs and osteoblasts was analyzed as a function of time by linear regression analysis. Values of p < 0.05 were considered significant. Statistical tests were performed using Instat 3 (GraphPad, San Diego, CA, USA).
Role of PHOSPHO1 in VSMC calcification
Phospho1 mRNA expression was measured in WT VSMC cultures, under calcifying conditions over 28 days and compared to Phospho1 expression in WT calvarial osteoblasts. In osteoblasts, the normalized Phospho1 expression increased in a linear manner (r2 = 0.9, p < 0.0001) until day 28 (Fig. 1A). In VSMCs, Phospho1 expression was upregulated and increased linearly (r2 = 0.881, p < 0.0001) until day 21, then plateaued between days 21 and 28 (Fig. 1B). To confirm that PHOSPHO1 is a key enzyme in VSMC mineralization, we analyzed calcium deposition normalized to the amount of protein in 28-day cultured Phospho1–/– VSMCs, in comparison to WT control cells. Quantitative (Fig. 1C) and qualitative (Fig. 1D) analyses of calcium deposition indicated a reduced ability of Phospho1–/– VSMCs to calcify. We also examined the expression of Phospho1 and Alpl mRNA in Enpp1–/– VSMCs. The initial 18s-normalized Alpl expression was approximately 10-fold higher in Enpp1–/– VSMCs than in WT after 21 days of culture, in agreement with our previous observations,16 whereas Phospho1 expression in Enpp1–/– VSMCs was slightly lower than in WT VSMCs (Fig. 1E).
Selection and properties of PHOSPHO1 and TNAP inhibitors
Because PHOSPHO1 is upregulated in cultures of calcifying WT VSMCs and the known PHOSPHO1 inhibitor lansoprazole lacks target specificity, development of more specific PHOSPHO1 inhibitors is necessary to study the effect of PHOSPHO1 inhibition in VSMC calcification. High-throughput screening for PHOSPHO1 inhibitors revealed several potent hits in the benzoisothiazolone class of compounds, indicating a nascent structure-activity relationship (SAR) that would provide traction for inhibitor optimization. The potency and selectivity of the synthesized compounds were assessed by in vitro enzymatic assays using purified human PHOSPHO1, TNAP,41 and ENPP1 (see Supporting Information, Data Supplement Fig. S1 for the dose-response curves for the inhibition of PHOSPHO1 by the nine new inhibitors using PHOSPHO1 natural substrates P-Etn and P-Cho). The IC50 values derived from these plots are shown in Table 1. MLS-0390838 and MLS-0263839 were the most potent, with IC50 values of 0.13 and 0.14 µmol/L, respectively. The mechanism of PHOSPHO1 inhibition was investigated via double reciprocal plots of PHOSPHO1 activity versus substrate concentration (32–170 µmol/L), in the presence of various concentrations of inhibitor (0–0.9 µmol/L). Figure 2 shows how these lines intersect on the y-axis, confirming the competitive nature of the inhibition for both lansoprazole (Fig. 2A) and MLS-0263839 (Fig. 2B) with inhibition constant (Ki) values of 0.16 ± 0.03 µmol/L and 0.22 ± 0.01 µmol/L, respectively. Likewise, MLS-0390838 was found to inhibit PHOSPHO1 in a competitive manner (not shown). MLS-0390838 and MLS-0263839 were selected as specific PHOSPHO1 inhibitors, as they showed minimal cross-inhibition of TNAP (Fig. 3A, B) and no inhibition of ENPP1 activity (not shown). Figure 3C, D shows the specific inhibition of PHOSPHO1 by MLS-0390838 and MLS-0263839 using P-Etn (Fig. 3C) and P-Chol (Fig. 3D) as substrates.
Table 1. IC50 Values for PHOSPHO1 Inhibitors, Using the Physiological Substrates P-Etn and P-Cho
Inhibition of TNAP has been shown to prevent VSMC calcification.16 To achieve more potent TNAP inhibition, new inhibitors were identified.35 In this study we aimed to determine whether combined inhibition of both TNAP and PHOSPHO1 was more effective than inhibition of either enzyme on its own. To accomplish this, we characterized the previously reported TNAP inhibitor MLS-003894935 to evaluate its suitability for inclusion in combined phosphatase inhibition experiments. MLS-0038949 was able to inhibit the p-nitrophenylphosphatase (pNPPase) (Fig. 3A) and pyrophosphatase (PPiase) activities of TNAP (Fig. 3B). Conversely, Fig. 3C and D shows that this compound selectively inhibits TNAP, but not PHOSPHO1. We also studied the mechanism of TNAP inactivation by MLS-0038949 at pH 9.8 (not shown) and pH 7.5 (Fig. 4) as described.16 The parallel lines (Fig. 4A) indicate that TNAP is inhibited by MLS-0038949 in an uncompetitive manner, both at pH 7.5 and at pH 9.8 (not shown). Secondary re-plots of the y-axis intercepts allow derivation of the corresponding inhibition constant, which was calculated as Ki = 0.32 ± 0.04 µmol/L at pH 7.5 and 0.74 ± 0.01 µmol/L at pH 9.8. The active site positioning of MLS-0038949 (Fig. 4B) was investigated in inhibition studies with TNAP mutants implicated in the active site positioning of the classic TNAP inhibitor levamisole36 (Glu108Phe, Tyr371Ala, and His434Glu) and compared to inhibition by the reference compound levamisole (Fig. 4C). These analyses revealed that, similar to levamisole, MLS-0038949 positioning in the TNAP active site depends primarily on the active site residues Tyr371 and His434, which are important for TNAP function and for the stabilization of uncompetitive inhibitors levamisole and theophylline.34 Correspondingly, the Ki values for the inhibition of TNAP by MLS-0038949 at pH 7.5 were 0.13 ± 0.02 µmol/L (Glu108Phe), 4.1 ± 0.5 µmol/L (Tyr371Ala), and 20 ± 1.7 µmol/L (His434Glu).
In silico docking of inhibitors in the active site
A model of PHOSPHO1 produced by ITASSER displayed the best features for molecular docking and was used for further studies. The model showed an organized tertiary structure with well-packed external helices surrounding the beta-pleated core of the protein. The active site, identified by homology to mouse 5′-nucleotidase III (Protein Data Bank ID 2bdu) and phosphoserine phosphatase from Helicobacter pylori (Protein Data Bank ID 3m1y) was located deep in the protein matrix and was surrounded by two helices with two loops closing the site. The metal ion was modeled into the active site as suggested by the presence of homologous carboxylate-bearing groups (Asp and Glu). Docking carried out in SwissDock with three compounds (lansoprazole, MLS-0263839, and MLS-0390838) produced 150 top scoring hits for each. Clustering analysis identified 19 clusters (docking poses) for lansoprazole, six clusters for MLS-0263839, and 17 clusters for MLS-0390838. Analysis of all clusters showed different binding modes for individual compounds in reference to the three-dimensional location of the active site. Only a single cluster for lansoprazole and MLS-0390838 was not located in the active site pocket. The rest of the clusters were blocking the entry to the metal ion-binding pocket, although none of the compounds had a direct contact with a metal ion, consistent with the competitive nature of these inhibitors. For visualization purposes, a single representative of every compound with the best pseudo-energy score was selected. The pseudo-energy of these representatives corresponds with the physiological effect and affinity of the compounds to PHOSPHO1. The highest pseudo-energy was shown by MLS-0390838 (30 kcal/mol), followed by MLS-0263839 (25 kcal/mol) and lansoprazole (15 kcal/mol). Figure 5A shows a comparison of docking poses for the lowest energy representatives of lansoprazole, MLS-0390838, and MLS-0263839 in the PHOSPHO1 active site. In this model MLS-0390838 and MLS-0263839 do not penetrate deeply into the active site pocket and do not interact with the catalytic site metal ion. Likewise, Fig. 5B shows MLS-0038949 docked in the TNAP active site. Docking of this uncompetitive TNAP inhibitor indicates a preferred interaction with active site residue His434, consistent with the kinetic data.
Inhibition of in vitro VSMC calcification and its effect in relative mRNA expression
The effect of the PHOSPHO1 and TNAP inhibitors on calcification of WT, Enpp1–/–, and Phospho1–/– VSMCs was investigated. Toxicity of the inhibitors in 7-day cultures of WT VSMCs was excluded by MTT test (see Data Supplement Fig. S2). The MTT test was not successful in 21-day cultures because of the more extensive mineral content (not shown). Quantification of Alizarin Red staining of WT VSMCs cultivated for 21 days in the presence of inorganic phosphate (Pi) (Fig. 6A) or β-glycerophosphate (β-GP) (Fig. 6B) shows that MLS-0263839 was able to reduce mineralization to 41.8% ± 2.0% and 53.8% ± 4.1%, respectively, relative to vehicle-treated control cells. The TNAP inhibitor (MLS-0038949) reduced Alizarin Red staining to 54.4% ± 3.6% of the value for vehicle-treated WT VSMCs cultured in the presence of Pi and 62.1% ± 9.9% for WT VSMCs cultivated in the presence of β-GP, reinforcing the involvement of TNAP in vascular calcification.16 Combined inhibition of both enzymes reduced calcification in Pi- and β-GP-containing media significantly further than inhibition of either inhibitor alone. Mineralization was reduced to 20.9% ± 0.74% (Fig. 6B) and 34.5% ± 0.78% (Fig. 6C) of the control by the MLS-0263839/MLS-0038949 combination in VSMCs cultured with Pi and β-GP, respectively. Lansoprazole alone and/or in combination with MLS-0038949 also significantly inhibited mineralization. The combination of the two PHOSPHO1 inhibitors, MLS-0026839 and lansoprazole, was toxic to the cells (see Data Supplement Fig. S2). The effect of the inhibitors MLS-0038949, MLS-0263839, and combination MLS-0038949/MLS-0263839 was also investigated in Enpp1–/– and Phospho1–/– VSMCs. In Enpp1–/– VSMCs, both inhibitors strongly reduced mineralization and the combination MLS-0038949/MLS-0263839 essentially abolished in vitro mineralization in Enpp1–/– (Fig. 6C). As expected, in Phospho1–/– VSMCs, only the TNAP inhibitor MLS-0038949 significantly inhibited mineralization (Fig. 6D).
WT cells cultivated under the same conditions as described in Fig. 6B were analyzed for differences in the mRNA expression of key enzymes involved in mineralization as well as osteogenic and VSMC markers. Figure 7 shows the effect of inhibitor treatment on the expression of osteogenic markers such as Alpl, Runx2, Spp1, Bglap, and Col1a1. Figure 7A shows that the combination of PHOSPHO1 and TNAP inhibitors significantly decreased the expression of Alpl mRNA (p < 0.05) whereas the expression of Enpp1 increased (p < 0.01, Fig. 7B). The expression of Phospho1 mRNA was not significantly different compared to the DMSO control cells in any of the treatments (Fig. 7C). Dual inhibitor treatment did not affect mRNA levels of the osteogenic markers Runx2, Opn, and Ocn (Fig. 7D–F), but the expression of Col1a1 was strongly decreased (p < 0.01, Fig. 7G). In contrast, dual inhibition strongly raised expression of the VSMC marker Acta2 (Fig. 7H).
The pathological mechanisms underlying medial vascular calcification are incompletely understood. It is known that MVC involves the loss of calcification suppressors, such as extracellular inorganic pyrophosphate (ePPi).42, 43 We have shown previously that Enpp1–/– and ank/ank mice display equivalent reductions in plasma ePPi concentrations19, 20 and both animal models develop comparable soft-tissue calcifications, including MVC accompanied by upregulation of TNAP expression in their VSMCs.16 Furthermore, we have documented that MVC in the uremic rat, a model of end-stage renal disease, is accompanied by an increase in the expression of TNAP, a key enzyme in PPi metabolism.17 MVC resulting from NT5E deficiency also appears to be caused by elevated expression of TNAP.44
More recently, the role of another phosphatase, PHOSPHO1, in calcification has been established. This enzyme is crucial for initiation of skeletal mineralization30 and mechanically competent mineralization,31 and is found inside MVs in calcifying tissues.23, 25 PHOSPHO1 has been proposed to scavenge Pi from phospholipids inside the MV to generate a Pi/PPi ratio conducive to calcification, whereas TNAP acts later in the mineralization process by restricting extracellular PPi. Phospho1–/– mice display a distinct phenotype, with growth plate abnormalities, spontaneous fractures, bowed long bones, osteomalacia, and scoliosis.30 Consistent with the nonredundant functions of these two enzymes, transgenic overexpression of TNAP does not correct the bone phenotype in Phospho1–/– mice.30 Furthermore, TNAP/PHOSPHO1 double-knockout mice exhibit a phenotype more severe than those of the respective single-knockouts: complete absence of skeletal mineralization. The importance of PHOSPHO1 in skeletal mineralization combined with the knowledge that MVC also appears to proceed via chondro-osteogenic differentiation of VSMCs and MV-mediated calcification,32, 33 suggest that PHOSPHO1 might have a functional role in MVC and could serve as a therapeutic target. Here, we provide the first evidence that PHOSPHO1 is involved in MVC, as Phospho1 expression increased in calcifying WT VSMCs and calcification was impaired in Phospho1–/– VSMCs.
Inhibitors of PHOSPHO1 identified from the small LOPAC1280 library of pharmacologically active compounds (lansoprazole, ebselen, and SCH-202676), showed that inhibiting PHOSPHO1 function further impaired both the mineralization ability of MVs from TNAP null osteoblasts and skeletal mineralization of developing chick long bones.25, 29 Lansoprazole partially inhibited ENPP1 and TNAP activity. We also observed in the mineralization studies that lansoprazole showed toxicity to smooth muscle cells and that lansoprazole inhibited mineralization in Phospho1–/– chondrocytes (not shown), confirming its off-target effects. The HTS performed in this study provided two compounds (MLS-00390838 and MLS-0263839), superior to lansoprazole. These inhibitors are highly selective for PHOSPHO1, with no detectable cross-inhibition of TNAP or ENPP1. MLS-00390838 and MLS-0263839 appeared to stabilize at several docking points inside the PHOSPHO1 active site, corroborating inhibitor selectivity. Both inhibitors act competitively, with sufficient affinity and selectivity to evaluate them as PHOSPHO1 inhibitors in VSMC calcification.
The finding that TNAP upregulation is a common phenomenon at sites of medial calcification in several disease models16, 17, 44 has important translational implications and has motivated the development of potent inhibitors of the pyrophosphatase activity of TNAP to prevent and reverse MVC.16, 35, 45–48 Selective inhibition of TNAP restores ePPi concentrations to levels high enough to prevent aberrant mineralization.16, 35, 46 To compare the relative merits of PHOSPHO1 and TNAP inhibition, we characterized the TNAP inhibitor, MLS-0038949, for use in mineralization assays. MLS-0038949 is highly specific for TNAP and does not cross-inhibit PHOSPHO1 or ENPP1.35 MLS-0038949 is an uncompetitive TNAP inhibitor, stabilized in the active site by residues H434 and Y371.17
When VSMCs were cultured for 21 days in the presence of the PHOSPHO1 or TNAP inhibitors, PHOSPHO1 inhibition was more effective at preventing calcification than TNAP inhibition. This enhanced effect may be due to the earlier action of PHOSPHO1 that affects the intravesicular generation of HA seed crystals during MV-mediated calcification.30 In addition, in complete agreement with the genetic data in which ablation of Phospho1 gene functions leads to downregulation of Alpl gene expression,30 the pharmacological inhibition of PHOSPHO1 leads to downregulation of Alpl mRNA. Thus, PHOSPHO1 appears to be a good target because not only is initiation of mineralization suppressed by the direct action of the inhibitor on PHOSPHO1 activity, but it also acts downstream by downregulating the amount of TNAP activity. Finally, the combined effect of both types of enzyme inhibitor suppressed VSMC calcification more than each inhibitor separately, both in WT and Enpp1–/– VSMCs. Interestingly, whereas double-inhibition essentially abolished mineralization in Enpp1–/– VSMCs (by 93.2% ± 0.2%, Fig. 6B), the effect was less pronounced in WT cells (65.5% ± 0.7%, Fig. 6C). This discrepancy may be accounted for by an increased Enpp1 expression in WT cells, in response to double inhibition treatment (Fig. 7B), because it has been proposed that ENPP1 can act as a backup pyrophosphatase in the absence of TNAP and maintain the levels of calcification.30 Thus, the enhanced inhibition seen in Enpp1–/– cells may be related to the absence of this compensatory mechanism. This effect was also observed by Villa-Bellosta and colleagues18 using inhibitors in WT aortic ring cultures, who found that TNAP inhibition reduced PPi hydrolysis by approximately 50%, whereas the remaining hydrolysis was inhibited by the ENPP1 inhibitors α,β-methylene ATP and γ,β-α,β-methylene ATP.
MVC involves transdifferentiation of VSMCs into cells and consequent matrix mineralization. Here, we measured the effect of phosphatase inhibitors on the expression of various markers of these processes. We observed no significant difference in the expression of the markers Runx2, osteocalcin (Bglap), and osteopontin (Spp1), although expression of Col1a1 was markedly impaired by the inhibitors, both alone and in combination. This indicates that phosphatase inhibition does not affect the chondro-osteogenic conversion of VSMCs, but prevents deposition of a matrix on which calcification can occur. Although increased TNAP expression is also a marker of VSMC transdifferentiation, it was significantly downregulated in cultures containing the PHOSPHO1 inhibitor. This discrepancy can be explained by our previous work, which has shown that PHOSPHO1 knockout mice have reduced TNAP expression and activity.30 MVC is also traditionally accompanied by loss of VSMC cell markers. Neither class of inhibitor alone affected expression of the VSMC gene Acta2; however, it was significantly higher in the combined inhibitor cultures, indicating that this treatment protected against loss of VSMCs features.
In conclusion, this work shows that at least two phosphatases are implicated in VSMC calcification, and that combined phosphatase inhibition offers a promising approach for developing compounds to prevent or attenuate MVC. However, it is important to note that this strategy may interfere with normal bone formation both by the intravesicular initiation of HA seed crystal deposition mediated by PHOSPHO1 function as well as extravesicular propagation of HA crystal deposition as a result of raising PPi levels resulting from TNAP downregulation and inhibition; future work will need to address this potentially significant concern. Furthermore, study of selective inhibitors in cultured VSMCs also provides us with new insights about how the enzymes TNAP, PHOSPHO1, and ENPP1 interact during VSMC mineralization. We are currently using the scaffolds described here to design drug-like compounds for in vivo testing and to further investigate the potential of PHOSPHO1 inhibitors as a potential treatment for MVC.
All authors state that they have no conflicts of interest.
We thank Soetkin Van Kerckhoven and Yalda Bravo for technical help. This work was supported in part by grant R01AR053102 and R01AR047908 from the National Institute of Arthritis and Musculoskeletal Diseases (NIAMS), and American Recovery and Reinvestment Act (ARRA) Challenge grant RC1HL10899 from the National Heart, Lung and Blood Institute (NHLB), National Institutes of Health (NIH), USA, and Institute Strategic Program Grant Funding from the Biotechnology and Biological Sciences Research Council.
Authors' roles: Study design: CF, MFH, VEM, and JLM. Study conduct: TK-M, MCY, DZ, SN, CS, BS, NDC, and RD. Data collection: AF. Data analysis: TK-M, MCY, DZ, SN, CS, BS, and NDC. Data interpretation: CF, MFH, VEM, and JLM. Drafting manuscript: TK-M, MCY, and JLM. Revising manuscript content: TK-M, MCY, and JLM. Approving final version of manuscript: TK-M, MCY, DZ, SN, CS, BS, NDC, RD, CF, MFH, VEM, and JLM. JLM takes responsibility for the integrity of the data analysis.