Inorganic pyrophosphatase induces type I collagen in osteoblasts

Abstract

Introduction

The physiologic selectivity of calcification in bone tissue reflects selective co-expression by osteoblasts of fibrillar collagen I and of tissue nonspecific alkaline phosphatase (TNAP), which hydrolyzes the calcification inhibitor pyrophosphate (PP_i) and generates phosphate (P_i). Humans and mice deficient in the PP_i-generating ecto-enzyme NPP1 demonstrate soft tissue calcification, occurring at sites of extracellular matrix expansion. Significantly, the function in osteoblasts of cytosolic inorganic pyrophosphatase (abbreviated iPP_iase), which generates P_i via PP_i hydrolysis with neutral pH optimum, remains unknown. We assessed iPP_iase in Enpp1^−/− and wild type (WT) mouse osteoblasts and we tested the hypothesis that iPP_iase regulates collagen I expression.

Methods

We treated mouse calvarial osteoblasts with ascorbate and β-glycerol phosphate to promote calcification, and we assessed cytosolic P_i and PP_i levels, sodium-dependent P_i uptake, Pit-1 P_i co-transporter expression, and iPP_iase and TNAP activity and expression. We also assessed the function of transfected Ppa1 in osteoblasts.

Results

Inorganic pyrophosphatase but not TNAP was elevated in Enpp1^−/− calvariae in situ. Cultured primary Enpp1^−/− calvarial osteoblasts demonstrated increased calcification despite flat TNAP activity rather than physiologic TNAP up-regulation seen in WT osteoblasts. Despite decreased cytosolic PP_i in early culture, Enpp1^−/− osteoblasts maintained cytosolic P_i levels comparable to WT osteoblasts, in association with increased iPP_iase, enhanced sodium-dependent P_i uptake and expression of Pit-1, and markedly increased collagen I synthesis. Suppression of collagen synthesis in Enpp1^−/− osteoblasts using 3,4-dehydroproline markedly suppressed calcification. Last, transfection of Ppa1 in WT osteoblasts increased cytosolic P_i and decreased cytosolic but not extracellular PP_i, and induced both collagen I synthesis and calcification.

Conclusions

Increased iPP_iase is associated with “P_i hunger” and increased calcification by NPP1-deficient osteoblasts. Furthermore, iPP_iase induces collagen I at the levels of mRNA expression and synthesis and, unlike TNAP, stimulates calcification by osteoblasts without reducing the extracellular concentration of the hydroxyapatite crystal inhibitor PP_i.

Introduction

Osteoblasts deposit inorganic and organic components in an extracellular matrix (osteoid) specialized for calcification, with collagen I fibrils making up >90% of osteoid [1]. Moreover, physiologic restriction of calcification to bone tissue reflects selective co-expression of fibrillar collagen I and the ecto-enzyme tissue nonspecific alkaline phosphatase (TNAP) (EC 3.1.3.1) [2]. Extracellular P_i, under the systemic regulation of diet, renal and parathyroid function, and locally generated by TNAP in bone, directly promotes the deposition of the crystalline basic calcium phosphate hydroxyapatite [2]. This occurs in the spaced “hole regions” of triple helical fibrils of collagen I under the influence of non-collagenous proteins [3] as well as in chondrocyte- and osteoblast-derived matrix vesicles [4].

TNAP generates P_i both by ATPase activity and by hydrolyzing PP_i, a potent physiologic inhibitor of hydroxyapatite crystal growth in the extracellular matrix [5]. P_i regulates calcification not simply by uptake in matrix vesicles, but also by effects on gene expression and the differentiation of mineralization-competent cells [6]. P_i uptake is mediated by multiple transporters, which include the sodium-dependent P_i co-transporters Pit-1 (SLC20A1, Glvr-1) and Pit-2 [7,8]. P_i functional effects are modulated by P_i-driven signaling via protein kinase C, Akt, and the MAP Kinase ERK1/ERK2 [9]. For example, direct provision of exogenous P_i, or of the TNAP substrate and organic P_i donor β-glycerophosphate modulates expression and nuclear transport of the osteoblastic master transcription factor Runx2 (cbfa1) [10,11], a promoter of collagen I expression (12) and essential driver of physiologic bone calcification [13]. Exogenous P_i also induces transcription of Pit-1, the hydroxyapatite crystal growth inhibitor and bone remodeling mediator osteopontin (OPN) [14], cyclin D1 [15], as well as certain transcription factors [6], including Nrf2 [16]. Exogenous P_i suppresses expression of mRNAs in osteoblasts that include those for tenascin-C, decorin, and certain collagens [16], and exogenous P_i post-transcriptionally regulates several proteins in osteoblasts including the AP-1 family transcription factor Fra-1 [6].

There has been only limited prior description of activity in bone [17] or in cultured osteoblasts [18] of the cytosolic enzyme inorganic pyrophosphatase, which generates P_i via PP_i hydrolysis, with a neutral pH optimum (EC 3.6.1.1) [19]. Though inorganic pyrophosphatase activity has been reported to be substantial in osteoblasts [18], regulation and function of inorganic pyrophosphatase, unlike the case for TNAP, have not previously been characterized in osteoblasts. Extracellular PP_i, including PP_i associated with matrix vesicles, physiologically functions partly as a critical inhibitor of hydroxyapatite crystal growth [20]. However, intracellular PP_i, like other phosphate esters, also serves as a biochemical intermediate and interacts with certain ATPases, modulates iron transport and mitochondrial function, and can regulate protein synthesis [20,21]. Deficiency in the PP_i-generating ecto-enzyme NPP1 (EC 3.1.4.1) manifests with phenotypic features including significant alterations in bone mineralization in long bones and calvariae, and pathologic, severe soft tissue and arterial calcification, with marked expansion of extracellular matrix in the soft tissues of the tunica media (myointimal hyperplasia) at sites of pathologic calcification of large arteries as well as in perispinal soft tissues [22-26]. Hypophosphatemia corrects the pathologic soft tissue calcification in Enpp1^−/− mice [2,5,24]. Furthermore, increased collagen synthesis has been observed in cultured osteoblasts from ttw/ttw mice, which are homozygous for a truncation mutation in Enpp1 and are phenotypically similar to Enpp1^−/− mice [25]. Hence, we tested the hypothesis that the capacity of inorganic pyrophosphatase to hydrolyze cytosolic PP_i to P_i could modulate both collagen I synthesis and calcification.

Materials and methods

Reagents

All chemical reagents were from Sigma-Aldrich, unless otherwise indicated.

Enpp1^−/− mice and culture of primary osteoblasts

All animal studies were performed humanely by institutionally approved protocol. We employed Enpp1^−/+ breeders backcrossed for >10 generations on C57BL/6 background, genotyping littermate wild type (WT) and Enpp1^−/− newborns as described [23]. Primary calvarial osteoblasts were isolated from 0- to 3 -day-old mice [23], with osteoblasts of the same genotype pooled, and aliquots of 4 × 10⁴ cells placed in a T75 tissue culture flask (BD Falcon) and carried in α-MEM (Cellgro/Mediatech) supplemented with 10% FCS, 2 mM glutamine, penicillin (50 U/ml), and streptomycin (0.5 mg/ml) (“Medium A”) at 37 °C in 5% CO₂. To stimulate calcification, growth medium was supplemented with 50 μg/ml l-ascorbic acid-PO₄ and 10 mM β-glycerophosphate (BGP) (“Medium B”), with medium replacement every 3 days.

Studies of cultured osteoblasts

Quantitative RT-PCR was performed as described [27], with quantification of target gene calculated by plotting crossing points on a Sybr Plus Green generated standard curve normalized to the quantification of a reference gene (GAPDH). PCR primers for murine cDNAs were: mouse collagen I α1 (Col1a1) (5′-TCT TTC CTT ATG AAT CAT CCC GCA-3′ and 5′-AAT TTC AAG CAC CCA TCT GTA G-3′), inorganic pyrophosphatase (Ppa1)(5′-CAA GGA CTT TGC AGT TGA CAT-3′ and 5′-ATG GCT TTG GCA GCA TC-3′), tissue non-specific alkaline phosphatase or TNAP (Akp2) (5′-AGA CAC AAG CAT TCC CAC TAT-3′ and 5′-CAC CAT CTC GGA GAG CG-3′), runt related transcription factor 2 (Runx2) (5′-GAT GCT CTG TTT CTT TCT TTC AGG-3′ and 5′-CTC CAG CAT TTC ATG CTA GT-3′), activating transcription factor 4 (Atf4) (5′-TTG ACC ACG TTG GAT GAC AC-3′ and 5′-CAG AGA TAT CAA CTT CAC TGC CTA-3′), osteocalcin (Bglap2) (5′-GTA GTG AAC AGA CTC CGG C-3′ and 5′-AGT GAT ACC GTA GAT GCG T-3′), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (5′-CAT CCC AGA GCT GAA CG-3′ and 5′-CTG GTC CTC AGT GTA GCC-3′).

To measure collagen synthesis, aliquots of 1.5 × 105 calcifying osteoblasts were incubated for 48 h at 37 °C, then 1.0 μCi ³H-Proline was added for 24 h, conditioned media collected, and collagenase-sensitive ³H incorporated determined [28], with data expressed as collagenase sensitive cpm/μg of protein. To specifically quantify cellular production of type I collagen, the collagen was first solubilized with pepsin under acidic conditions and then further digested with pancreatic elastase at neutral pH to convert polymeric to monomeric collagen, and measured (μg collagen/ml) by Collagen I Detection Kit (Chondrex).

To measure radiolabeled P_i uptake, we adapted an approach described in chondrocytes [29], here using aliquots of 1.5 × 10⁵ osteoblasts. In brief, cells were incubated for 2 h at 37 °C in P_i-free buffer (1 mM CaCl₂, 1.8 mM MgCl₂, and 10 mM HEPES), then assays initiated by adding 200 μl of the P_i-free buffer also containing 3 μCi/ml carrier-free ³³P_i (MP Biomedicals) and 150 mM NaCl. After 5 min, uptake was terminated by medium replacement with ice cold P_i-free buffer [29]. Results for P_i uptake were normalized per cell DNA. We assessed Pit-1 expression by SDS-PAGE/Western blotting as described, using a rabbit polyclonal Pit-1 specific antibody [29].

To assess calcification, osteoblasts (75,000 cells/well) were cultured in Medium B, and von Kossa staining nodules quantified in a minimum of 5 high power fields [26] or Alizarin Red S binding assay performed [23]. For immunocytochemical analyses, cells fixed with 4% paraformaldehyde were studied using the Histomouse-SP kit (Zymed) and employed the DAB chromogen/substrate system to detect antigen/antibody/enzyme complex as intense brown deposits, with hematoxylin QS (Vector Laboratories) as counterstain. Primary antibody (rabbit anti-mouse collagen I (Chemicon) (1:50) was incubated with sections for 1 h. Nonimmune rabbit IgG (Santa Cruz Biotechnology) was the negative control.

Inorganic pyrophosphatase transfection, iPP_iase activity, alkaline phosphatase activity and intracellular P_i and PP_i assays

For measurement of cytosolic P_i, we adapted a described assay [30], in which cells were washed with 10 mM PBS and their cytosolic content extracted using the Subcellular Proteome Extraction Kit (Calbiochem) according to manufacturer protocol. P_i was measured colorimetrically at OD₆₂₀ using the PiBlue Phosphate Assay Kit (BioAssay Systems). Data were expressed as pmoles of P_i per μg protein in each well. Due to the low sample volumes recovered from cytosol, PP_i was measured radiometrically [23] in separate sets of cytosolic extracts. Extracellular PP_i was measured in parallel in conditioned media [23] collected from each well.

We transfected osteoblasts with human iPP_iase cDNA subcloned into pBSSK from Dr. T. Wensel (Baylor) [19]. To do so, aliquots of 3 × 10⁵ primary calvarial osteoblasts were placed in Medium B as above, and on day 10 of culture were transiently transfected using Targefect-Osteoblast (Targeting Systems) according to manufacturer protocol. Control transfection efficiency in each experiment was determined by β-galactosidase transfection and staining, and was consistently >30–40%. To assay iPP_iase activity, cells were resuspended in 50 mM Glycine, 10 mM MgCl₂ and incubated with Taussky-Shor Color Reagent at 50 °C for 10 min, with enzymatic activity assessed colorimetrically at OD 660 nm [31]. Five micrograms of protein was used to colorimetrically measure alkaline phosphatase (AP) specific activity with 5 mM p-nitrophenolphosphate used as a substrate. One unit of AP was defined as 1 μmol substrate hydrolyzed per hour (per microgram protein/sample).

Statistics

Data are expressed as mean ± SD, unless otherwise noted. Statistically significant differences between two groups were determined by 2-tailed unpaired Student's t-test, or by 1-way ANOVA to compare means between >3 experimental sample groups.

Results

iPP_iase is increased in Enpp1^−/− calvariae in situ

We first established specificity of assays for alkaline phosphatase and iPP_iase enzyme activity (Supplemental Fig. 1). Using these assays, we demonstrated that Enpp1^−/− calvariae had elevated iPP_iase activity and Ppa1 mRNA levels, whereas alkaline phosphatase activity and Akp2 mRNA expression were not significantly altered in situ (Figs. 1A–D).

Fig. 1.

Increased Ppa1 expression and iPP_iase activity in situ in Enpp1^−/− calvariae. In freshly isolated mouse calvariae of the indicated genotypes, (A) inorganic pyrophosphatase 1 (Ppa1) and (B) TNAP (Akp2) mRNA expression were quantified by real-time PCR and normalized to the housekeeping gene Gapdh, using the LightCycler FastStart DNA Master Plus SYBR Green system. Six samples per run were averaged for the above results. (C) Inorganic pyrophosphatase 1 (abbreviated as iPP_iase here and in other figure labels) and alkaline phosphatase (D) activity were measured in freshly isolated calvariae, as described in Methods, with data expressed as Units per μg protein (n = 5). *p <0.05.

Cultured Enpp1^−/− osteoblasts maintain cytosolic P_i levels comparable to normal osteoblasts and demonstrate increased calcification despite decreased TNAP

Assessing primary mouse calvarial osteoblasts, stimulated to calcify by treatment with ascorbate and the organic phosphate donor β-glycerophosphate (10 mM), we first confirmed [23] a 2- to 4-fold increase in calcification between 10 and 18 days in Enpp1^−/− osteoblasts (data not shown). Under these conditions, Ppa1 mRNA expression and iPP_iase activity were greater in calcifying Enpp1^−/− than WT osteoblasts in cells at several time points, though elevation of iPP_iase activity in Enpp 1^−/− cells principally occurred in the first few days (Figs. 2A, B). In contrast, WT osteoblasts were confirmed [32] to have significant elevations in both Akp2 mRNA expression and alkaline phosphatase activity (Figs. 2C, D).

Fig. 2.

Comparison of iPP_iase, cytosolic PP_i and P_i levels, and alkaline phosphatase in Enpp1^−/− and WT osteoblasts. (A) Ppa1 expression relative to Gapdh and (B) whole cell iPP_iase activity were measured in calvarial osteoblasts stimulated to calcify using 10 mM BGP and 50 μg/ml ascorbate, as described in Methods (n = 6). (C) Akp2 mRNA (n = 6) and (D) whole cell alkaline phosphatase activity (n = 4) were measured in primary calvarial osteoblasts, as described in Methods. (E) Cytosolic PP_i was measured radiometrically (n = 5), and (F) cytosolic P_i measured colorimetrically at OD₆₂₀ (data pooled for 3 separate experiments run in quadruplicate). *p<0.05.

Since P_i and PP_i levels are modulated by iPP_iase and TNAP, we examined cytosolic PP_i and P_i levels in the calcifying osteoblasts. Enpp1^−/− osteoblasts demonstrated ~3-fold less cytosolic PP_i at day 1 in culture, but the Enpp1^−/− osteoblasts maintained cytosolic P_i levels at comparable levels to those of WT osteoblasts between days 1 and 14 in culture (Figs. 2E, F), under conditions where the hypercalcifying Enpp1^−/− osteoblasts did not mount a rise in either Akp2 mRNA expression or alkaline phosphatase activity (Figs. 2C, D).

In association with maintenance of cytosolic P_i levels comparable to WT osteoblasts in Enpp1^−/− cells, sodium-dependent P_i uptake sensitive to blunting by sodium-dependent P_i co-transporter inhibitor phosphonoformic acid (PFA) [29] was increased several-fold in Enpp1^−/− compared to WT osteoblasts at days 4 and 14 (Fig. 3A). Under these conditions, detectable expression of the sodium-dependent P_i co-transporter Pit-1 (by SDS-PAGE/Western blotting) was more sustained in the Enpp1^−/− osteoblasts (Fig. 3B).

Fig. 3.

Sodium-dependent P_i uptake is increased and the sodium-dependent P_i co-transporter Pit-1 expression more sustained in Enpp1^−/− compared to WT primary calvarial osteoblasts. (A) ³³P_i uptake was quantified in osteoblasts with and without treatment with the sodium-dependent P_i co-transporter inhibitor phosphonoformic acid (PFA) (300 μM), as described in Methods (n = 4). (B) Pit-1 (~85–90 kDa) was assessed by SDS-PAGE/Western blotting of osteoblast cell lysates induced to calcify, using rabbit antibodies to Pit-1 and to tubulin as a control. *p <0.05.

Increased Runx2, ATF4 and collagen I expression in cultured Enpp1^−/− osteoblasts

Elevation of TNAP expression is one marker of osteoblast maturation [32], as is increase in osteocalcin (Bglap2) and the P_i-sensitive osteoblast master transcription factor Runx2. However, there was no significant alteration in the onset or extent of elevation in expression of Bglap2 [33] as Enpp1^−/− osteoblasts carried out calcification in vitro (Fig. 4A). The Enpp1^−/− osteoblasts demonstrated increased expression of Runx2 (Fig. 4B), one of several transcription factors that promote osteoblast collagen I expression [12] as well as expression of ATF4, a key transcription factor regulating osteocalcin expression (Fig. 4C). Type I collagen production, assessed by immunocytochemistry, and total synthesis of collagen were increased in calcifying Enpp1^−/− osteoblasts relative to WT cells (Figs. 5A, B), but differences in Col1a1 mRNA expression did not reach statistical significance during the same course of osteoblast differentiation (not shown). Increased collagen synthesis in Enpp1^−/− osteoblasts was reversed by treatment with exogenous PP_i (Fig. 5B).

Fig. 4.

Comparison of expression of osteocalcin, runx2, and Atf4 in Enpp1^−/− and WT osteoblasts. (A-C) Osteocalcin (Bglap2), Runx2, and ATF4 were quantified and normalized to Gapdh by real-time PCR in calvarial osteoblasts treated with 10 mM BGP and 50 μg/ml ascorbate acid (n = 6).

Fig. 5.

Comparison of expression of type I collagen and collagen synthesis in Enpp1^−/− and WT osteoblasts. (A) Immunocytochemistry of the calcifying osteoblasts using rabbit anti-mouse collagen I-specific antibody and negative staining control (nonimmune rabbit IgG). Magnification, 100×. (B) Total collagen synthesis was measured by ³H-proline incorporation into osteoblasts grown for 72 h days in the same medium as above, with and without exogenous PP_i added, and results expressed as collagenase sensitive cpm/μg DNA. *p<0.05.

Increased iPP_iase induces collagen I expression and calcification in osteoblasts

Treatment of Enpp1^−/− osteoblasts with the collagen synthesis inhibitor 3,4-dehydroproline, which was confirmed [34] to suppress total collagen synthesis (data not shown), inhibited collagen I deposition in the extracellular matrix (Fig. 5A), and inhibited calcification by up to ~50% in both Enpp1^−/− and WT osteoblasts (Fig. 6B). Since increased type I collagen appeared intimately involved in the hypercalcification process in Enpp1^−/− osteoblasts, we next tested the hypothesis that type I collagen was directly regulated by iPP_iase. To do so, we transfected Ppa1 cDNA into normal osteoblasts, and observed association of ~50% more cell-associated iPP_iase activity with ~100% higher cytosolic P_i and with ~6-fold lower cytosolic PP_i (Figs. 7A-C). Under these conditions, iPP_iase did not significantly lower the concentration of extracellular PP_i (Fig. 7D). Nevertheless, Ppa1 transfection induced calcification (visualized as several-fold increased von Kossa positive nodule formation) (Figs. 7E, F).

Fig. 6.

Inhibition of collagen synthesis in Enpp1^−/− osteoblasts markedly suppresses calcification. (A) WT and Enpp1^−/− osteoblasts grown in medium containing 10 mM BGP and 50 μg/ml ascorbate and 1.0 mM dehydro-l-proline for 72 h were studied by immunocytochemistry for type I collagen, using nonimmune rabbit IgG antibody as negative control. Magnification, 100×. Representative scale bar indicates 400 μ. (B) WT and Enpp1^−/− osteoblasts were treated with 10 mM BGP and 50 μg/ml ascorbate, as well as 0.1 mM and 1.0 mM dehydro-l-proline for 72 h. Calcification was quantified by Alizarin Red S binding, determined chromogenically at OD₅₇₀. *p<0.05.

Fig. 7.

Effects of Ppa1 transfection on cytosolic P_i and PP_i and extracellular PP_i levels, and on calcification in WT primary calvarial osteoblasts. WT osteoblasts grown in medium supplemented with 10 mM BGP and 50 μg/ml ascorbate were transfected at day 10 in culture and assayed 72 h later. (A) iPP_iase activity was measured and expressed as Units per microgram protein. (B) Cytosolic P_i, (C) cytosolic PP_i, and (D) extracellular PP_i were measured. (E) Calcification was assessed by von Kossa staining, with counterstaining using nuclear fast red, and representative staining results shown. Calcium salts appear brown-black. Magnification, 200×. Representative scale bar indicates 500 μ. (F) von Kossa positive nodules per high-power field were counted after transfection by a blinded observer. Mean ± SD, representative of 6 individual fields. *p<0.05. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The Ppa1 transfection did not modulate expression of the transcription factor ATF4 (Figs. 8A, B), which critically promotes collagen I expression as well as osteoblast maturation and terminal differentiation in vitro and is essential for promoting accrual of the normal mass of calcified bone in vivo [35]. Nevertheless, Ppa1 transfection induced Col1a1 mRNA (Fig. 8C) in association with increased collagen I synthesis (Figs. 8D, E). Our findings are summarized in the model of Supplementary Fig. 2, and discussed further below.

Fig. 8.

iPP_iase induces type I collagen expression in primary calvarial osteoblasts without significantly modulating ATF4 expression. WT osteoblasts grown in medium supplemented with 10 mM BGP and 50 μg/ml ascorbate acid were transfected with Ppa1 at day 10 and assayed 72 h later. (A, B) ATF4 mRNA and protein expression were assessed by real-time PCR and Western blotting, respectively (n = 4). (C) Col1a1 mRNA was measured by real-time PCR and (D) capture ELISA performed for collagen I, as described in Methods, with results expressed μg collagen I/ml (n = 8). *p<0.05. (E) Immunocytochemistry for collagen I, as described above was performed after iPP_iase transfection. Magnification, 100×. Representative scale bar indicates 400 μ.

Discussion

Osteoblast PP_i and P_i metabolism are subject to regulation coordinated partly via NPP1-catalyzed intracellular and extracellular PP_i generation, predominant cytosolic to extracellular movement of PP_i driven by ANK, and PP_i hydrolysis by TNAP [2,5,20,23,24,30]. Linkage of these events can modulate the effects of these and other regulators of calcification [23,24,36,37]. For example, PP_i inhibits the enzymatic activity of alkaline phosphatase [37], and both exogenous PP_i and P_i induce osteopontin (OPN) expression [23,24,37].

Certain regulators of PP_i and P_i metabolism have been observed to be modulated at the expression level in PP_i deficiency states [24,38,39], and here, we observed that Ppa1 expression and iPP_iase activity were increased in situ in Enpp1^−/− mice calvariae previously documented to have a primary deficiency of PP_i generation [23]. Moreover, there were early increases in Ppa1 expression and activity of iPP_iase in hypercalcifying Enpp1^−/− cultured calvarial osteoblasts relative to WT osteoblasts. Other abnormalities in P_i metabolism observed in calcifying Enpp1^−/− osteoblasts were lack of the typical increase in TNAP at 10 days and greater in culture seen in normal cells, as well as more sustained Pit-1 expression and increased sodium-dependent P_i uptake. Expression of the P_i-regulated osteoblast master transcription factor Runx2 also was increased in cultured Enpp1^−/− osteoblasts.

The maintenance of cytosolic P_i levels comparable to those of WT osteoblasts in Enpp1^−/− osteoblasts, despite lack of increased TNAP P_i-generating PP_i-hydrolyzing and ATPase activities in Enpp1^−/− osteoblasts, suggested contributory effects of both early increases in iPP_iase activity and sustained elevation of sodium-dependent P_i uptake (essentially increased “P_i hunger”). The marked decrease in cytosolic PP_i early in culture in Enpp1^−/− osteoblasts observed here may have partially reflected adaptive increases in iPP_iase activity. We did not test for changes in PP_i transport by Ank [40] or for altered PP_i generation by the NPP3 isoenzyme [41].

Pit-1-driven P_i uptake is known to promote calcification by osteoblasts and vascular smooth muscle cells [7,8]. Here, we demonstrated that increased iPP_iase, which lowered cytosolic PP_i (but not extracellular PP_i) levels in association with elevation of cytosolic P_i, also increased collagen I mRNA and collagen I synthesis, as well as calcification in normal osteoblasts. Pathologic calcification of perispinal ligaments and certain other soft tissues in association with NPP1 deficiency is intimately linked with marked extracellular matrix expansion [22,23]. Here, we observed increased collagen I production by Enpp1^−/− osteoblasts. The induction by iPP_iase of collagen I in osteoblasts was not dependent on modulation of ATF4 expression [35], and therefore may have partly reflected direct effects of cytosolic P_i. In this context, intracellular P_i has been observed to provide signals that affect mitochondrial function and intracellular movement of Runx2 in chondroosseous differentiation [10,42], and P_i regulates expression of multiple genes that modulate calcification [6,9,14-16].

Since cytosolic P_i levels were not elevated in PP_i-depleted Enpp1^−/− osteoblasts compared to wild type cells, we speculate that ambient cytosolic PP_i regulates cytosolic P_i signaling potential and impacts on P_i-sensitive gene expression. Alternatively, the subcellular loci for PP_i and P_i generation and transport may modulate distinct signaling and transcriptional regulatory effects of P_i. Exogenous P_i down-regulates collagen I expression in murine immortalized calvarial osteoblastic MC3T3-E1 cells [16]. As such, P_i effects on osteoblast determinants of osteoid composition may also depend on timing of administration in relation to osteoblast differentiation. However, transfection of Ppa1 did not act by affecting osteoblast extracellular PP_i. It will be of interest to assess the net role increased iPP_iase plays in extracellular matrix expansion at sites of pathologic soft tissue calcification in Enpp1^−/− mice.

Significantly, Ank, which lowers intracellular PP_i [40], promotes osteoblast maturation [43]. However, factors other than cytosolic PP_i and balance may contribute to increasing Runx2, Pit-1, and collagen P_i I expression in Enpp1^−/− osteoblasts that demonstrate decreases in cytosolic PP_i. For example, Enpp1-deficient murine ttw/ttw osteoblasts demonstrate enhanced PTH responsiveness and collagen synthesis in culture [25], and PTH regulates Runx2 expression, and PTH anabolic effects in bone require Runx2 [44]. Cellular sodium-dependent P_i uptake not only is induced by P_i deprivation but also is modulated by IGF-I [45,46]. In this context, NPP1 is known to antagonize receptor signaling by insulin in a PP_i-independent manner [47], and insulin receptor and IGF-I receptor signaling partially overlap.

Limitations of this study included temporal compression of the experimental analyses by adding 4-fold more β-glycerophosphate to stimulate calcification than in our previous study of Enpp1^−/− osteoblasts (10 mM vs. 2.5 mM) [23]. Nevertheless, we have observed a blunting of the normal rise in alkaline phosphatase activity as Enpp1^−/− osteoblasts are stimulated to calcify over 2–3 weeks in culture via addition of as little as 2.5 mM β-glycerophosphate (Johnson KA, et al., unpublished observations). We did not study modulation by iPP_iase of proteins other than collagen that play substantial roles in modulating hydroxyapatite crystal nucleation and growth [1,23,24,48]. Last, PFA effectively inhibits class II sodium-dependent P_i co-transporters but is not a potent inhibitor of class III Pit-1-mediated P_i co-transport [49], and this study did not perform experiments to specifically knock down Pit-1 or other sodium-dependent P_i co-transporters.

In conclusion, pathologic calcification in NPP1 deficiency states may be mediated by increased hunger of mineralizing PP_i-deficient cells for P_i. In turn, increased P_i generation by iPP_iase and increased P_i uptake mediated by Pit-1 function to normalize cytosolic P_i levels in cells deficient in NPP1 and PP_i, and such cells respond with heightened Runx2 and collagen I expression. Our results implicate increased collagen I expression in osteoblastic cells in promoting the marked extracellular matrix expansion documented in association with NPP1 deficiency in soft tissues containing ectopic chondroosseous differentiation and calcification. Increased collagen I expression by Enpp1^−/− osteoblastic cells may do more to promote calcification than simply providing more extracellular matrix loci for hydroxyapatite deposition. For example, the capacity of a type I collagen-rich extracellular matrix to induce integrin-mediated osteoblastic differentiation [50,51] may promote the dysregulated chondro-osseous differentiation seen in soft tissues in association with NPP1 deficiency [39]. Type I collagen supports maintenance of osteoblast phenotype [32], promoting increased expression of several genes that modulate differentiation in osteoblastic cells (e.g., PTH/PTHrP receptor [32,52,53]. Taken together, the results of this study shed new light on the fundamental role in calcification of the interface between extracellular matrix composition and closely linked PP_i and P_i metabolism. Though TNAP appears to be in charge of controlling the steady state P_i:PP_i ratio extracellularly that regulates matrix calcification, iPP_iase could be central to appropriate balance of steady state intracellular P_i relative to its potential inhibitor PP_i,. Functional relationships between imbalances in P_i relative to PP_i in different subcellular compartments merit further investigation.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.bone.2009.08.055.