| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
* Department of Internal Medicine, Erasmus MC, Rotterdam, The Netherlands; and
Department of Bioinformatics, Erasmus MC, Rotterdam, The Netherlands
1Correspondence: Erasmus MC, Department Internal Medicine, Rm. Ee585, P.O Box 2040, 3000 CA, Rotterdam, The Netherlands. E-mail: j.vanleeuwen{at}erasmusmc.nl
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: osteoblasts • vascular smooth muscle cells • microarray • autocrine
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Osteoblasts play a pivotal role in regulating bone quality, as they are the bone-forming cells and the directors of bone resorption by osteoclasts. Osteoblasts are of mesenchymal origin and undergo a complex differentiation process regulated by multiple endocrine, paracrine, and autocrine factors. During differentiation and bone formation, osteoblasts produce a complex ECM that eventually starts to mineralize. Initiation of mineralization requires the precipitation and attachment of calcium phosphate crystals to the ECM, although the exact mechanism is poorly understood (2)
. The process of mineralization is not unique for bone cells demonstrated by ectopic mineralization in pathological conditions. Vascular smooth muscle cells (VSMCs) are thought to be primarily involved in vascular mineralization. These cells undergo an osteoblast-like differentiation process expressing osteoblastic factors, including alkaline phosphatase (ALPL) and runt-related transcription factor 2 (3
4
5)
.
Several members of the transforming growth factor-ß (TGFß) superfamily, including TGFß and bone morphogenic proteins (BMPs), are well-known regulators of bone formation. Both TGFß and BMPs promote bone development by stimulating the differentiation of osteoblast progenitors (6
, 7)
. In addition, TGFß seems to inhibit later phases of osteoblast differentiation and mineralization (8
9
10)
. Activins belong to the TGFß superfamily, but their precise role in bone formation is unknown. The structure of activins is closely related to that of TGFß, and activins act via similar intracellular signaling molecules (11)
. Activins and their relatives, inhibins, were initially purified from gonadal fluids and characterized on the basis of their ability to modulate FSH secretion from pituitary gonadotropes (12
, 13)
. Besides this classical role of activins, activins can affect the function of other cell types and tissues (e.g., the adrenal gland, liver, neurons, pancreas, and bone) (14
15
16
17
18)
.
Activins and inhibins are composed of the inhibit subunits 
, ßA, and ßB. Heterodimerization of the and ßA or ßB subunit forms inhibin A or inhibin B. Homo- and heterodimerization of the ßA and ßB subunits result in formation of activin A, activin AB, or activin B (12)
. Activins need type I and type II activin receptors for signal transduction. Activins bind to the activin type IIA or type IIB receptors, leading to recruitment and phosphorylation of the activin type IB receptor (ALK4). The phosphorylated type I receptor, in turn, phosphorylates intracellular signaling proteins known as Smads. Smad2 or 3 are phosphorylated by activins and TGFß, whereas Smad1, 5, or 8 are phosphorylated by the BMP-like ligands (11)
. Activin signaling is inhibited by the extracellular action of inhibins or follistatin (19)
. Follistatin is a soluble protein that functions as an activin binding protein preventing activin from interacting with its receptor (20)
. Inhibins need the presence of a coreceptor, betaglycan, to inhibit activin signaling via competitive binding to the activin type II receptor, preventing recruitment of the activin type I receptor (21
, 22)
.
Several studies demonstrated a role for activins in bone metabolism. Large quantities of activin are found in the ECM of bovine bone (23)
. Activin A enhances osteoclast-like cell formation in murine bone marrow cultures (24
, 25)
. Activin A promotes osteoblastogenesis in murine bone marrow cultures (25)
and, in vivo, promotes bone formation and fracture healing in rodents (23
, 26
, 27)
. However, other reports demonstrated an inhibitory effect of activin on osteoblast differentiation in rat and murine osteoblasts (28
, 29)
.
The aim of this study was to assess the impact of activin signaling on osteoblast differentiation related to human bone formation and bone quality. The effect of activin A on matrix mineralization was measured together with the endogenous production of activin A and activin antagonists during osteoblast differentiation. Moreover, genome-wide expression profiling in osteoblasts was performed to reveal downstream mediators of activin signaling in relation to matrix mineralization.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Medium was supplemented with freshly added 10 mM ß-glycerophosphate (Sigma, St. Louis, MO, USA), 100 nM dexamethasone (DEX) (Sigma), or other additives (activin and follistatin) and replaced every 2 or 3 days. Follistatin was purchased from PeproTech and activin A from R&D Systems (Minneapolis, MN, USA). Normal human osteoblasts (NHOst) (Cambrex Bio Science CC-2538; East Rutherford, NJ, USA), vascular smooth muscle cells (coronary artery smooth muscle cells; Cambrex Bio Science; CC-2583), and human mesenchymal stem cells (Cambrex Bio Science) were cultured in a manner similar to that of SV-HFO cells, with only the following adjustments. NHOsts and hMSCs were used between passages 3 and 6 and seeded in a density of 5 x 103 and 10 x 103 vital cells per cm2, respectively. NHOst and hMSC cultures were induced to mineralize in a medium similar to SV-HFO, except that 10% charcoal-treated, heat-inactivated FCS (Gibco BRL, Carlsbad, CA, USA) was used. VSMCs were used between passages 3 and 7 and seeded in a density of 5 x 103 vital cells/cm2. Expansion of VSMCs was performed in smooth muscle cell medium (Cambrex Bio Science) supplemented with Clonetics Sm-GM-2 Bulletkit (Cambrex Bio Science). VSMCs were induced to mineralize in DMEM (Gibco BRL; with 4500 mg/l glucose, L-glutamine, and pyruvate) supplemented with 100 nM DEX, 0.1 mM ascorbic acid (Sigma), 10 µg/ml insulin (Sigma), 1 mM CaCl2 (final concentration of 2.8 mM), 10 mM ß-glycerophosphate, and 10% FCS.
DNA, alkaline phosphatase activity, and mineralization assays
ALPL, DNA, and calcium measurements were performed as described previously (31)
.
Quantitative PCR analysis (QPCR)
RNA isolation, cDNA synthesis, and QPCR were performed as described previously (30)
. Primer and probe sequences and concentrations used for QPCR are listed in supplemental Table 1.
Quantification of activin A, follistatin, and inhibin A
At various days during culture, medium was collected for activin A and follistatin measurements. Medium was collected from the cultures after 48 h incubation. Medium was centrifuged (5 min, 500 g) and stored at –20°C for further analysis. Cell lysates were also prepared so as to analyze DNA content of the corresponding cultures. Activin A was measured using 50–100 µl medium and the activin A DuoSet ELISA kit. Follistatin was measured using 100 µl medium and the follistatin quantikine ELISA kit (R&D Systems). Inhibin A was measured using the inhibin A ELISA kit purchased from Serotec (Raleigh, NC, USA).
Immunohistology
Plastified bone sections were deacrylated in a 1:1 mixture of xylene and chloroform for 30 min, then dipped in xylene, rehydrated, and rinsed twice with distilled water. Slides were pretreated with Tris-EDTA buffer, pH 9.0, for 15 min at 100°C and cooled down for 15 min, followed by rinsing in tap water. Endogenous peroxidase activity was inhibited by a 10:1 mixture of PBS and H2O2 for 10 min, followed by two water washes and one Tris-HCl pH 8.0 wash. The following steps were carried out in a humidified chamber at room temperature. Primary antibodies were diluted 1:50 in normal antibody diluent (Skytek Laboratories, Logan UT, USA; code ABB999) and incubated for 60 min. Inhibin-ßA, inhibin-, and IgG2b negative control antibodies were purchased from Serotec (MCA950 ST, MSA951S, MCA691). Slides were rinsed twice in Tris-HCl, pH 8.0. Immunoreactivity was detected with the Dako REALTM EnvisionTM Detection System, peroxidase/DAB+, rabbit/mouse (Dako, Carpinteria, CA, USA; code K5007). After DAB detection, slides were rinsed in tap water and counterstained in Harris hematoxylin for 1 min, rinsed in tap water for 2 min, dehydrated in ascending ethanol steps, rinsed in xylene, and coverslipped with Pertex mounting medium (Histolab, Västra Frölunda, Sweden).
Luciferase reporter assays
On day 5 of culture, cells were transfected with 200 ng reporter plasmids per well (12-well plate) using Fugene6 (Roche, Nutley, NJ, USA). After 24 h, medium was replaced by fresh medium containing low serum (0.2% FCS) and incubated for 3 h. After 3 h the medium was refreshed for the second time with medium containing low serum, but this time supplemented with additives as described in Results. After 24 h, cells were lysed by incubating for 20 min in 100–200 µl lysis buffer (Promega, Madison, WI, USA). Luciferase activity was measured using 25 µl cell lysate and the Steady-Glo Luciferase Assay System (Promega). Activin signaling was measured using pGL3(CAGA)12-lux (CAGA-Luc) (32)
and BMP signaling was measured using pGL3-BRE-luc (BRE-Luc) (33)
.
Devitalization of osteoblast cultures
On day 10 of culture, medium was removed and the cultures were washed once with PBS (Gibco BRL). Cultures were air dried and frozen overnight at –20°C. Next, devitalized cultures were incubated normally as described in the cell culture methods (indicated as day 0 for devitalized cultures).
Affymetrix Genechip-based gene expression
Purity and quality of isolated RNA were assessed by RNA 6000 Nano assay on a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). The RNA of three different biological samples was pooled per condition analyzed. Synthesis of first- and second-strand cDNA from total RNA was performed according to the One-Cycle Target Labeling protocol (Affymetrix, Santa Clara, CA, USA; 701024 Rev. 3). In total, 4.0 µg of total RNA was reverse transcribed using a Superscript ds-cDNA Synthesis Kit according to the manufacturer's description (Invitrogen, Carlsbad, CA, USA). Subsequently, double-stranded cDNA was purified using GeneChip Sample Cleanup Module (Affymetrix) and served as a template in the in vitro transcription reaction using BioArray HighYield RNA Transcript Labeling Kit (Affymetrix). The amplified biotinylated complementary RNA (cRNA) was purified using a GeneChip Sample Cleanup Module. In total, 20 µg of biotin-labeled cRNA was fragmentized by metal-induced hydrolysis at a final concentration of 0.5 µg/µl for 35 min at 94°C. Fragmentation was checked on an Agilent 2100 Bioanalyzer, confirming an average size of 
100 nt. Fifteen migrograms of fragmented biotinylated cRNA was hybridized to GeneChip Human Genome U133 Plus 2.0 oligonucleotide Genechips (Affymetrix) according to the manufacturer's protocol (Affymetrix, 701025 Rev. 5). Staining, washing, and scanning procedures were carried out as described in the GeneChip Expression Analysis technical manual (Affymetrix). Data acquisition was performed using the GeneChip Scanner 3000.
Genechip data analysis
To examine the quality of the various arrays, measured intensity values were analyzed using GeneChip Operating Software (Affymetrix). The percentage of present calls (40%), noise, background, and ratio of GAPDH 3' to 5' (<1.4) all indicated a high quality of samples and an overall comparability. Probe sets that were not present (according to Affymetrix MAS5.0 software) in any of the Genechips were omitted from further analysis. Raw intensities of the remaining probe sets of each chip were log2 transformed and normalized using quantile normalization. After normalization, the data were back-transformed to normal intensity values. Data analysis was carried out using OmniViz software, version 3.6.0. For gene ontology (GO) analysis, the 520 selected affymetrix IDs were analyzed using OntologyTraverser (34)
.
Gene nomenclature
Gene names and gene symbols were used as provided by HUGO Gene Nomenclature Committee (35)
.
Statistics
Data were presented only if multiple independent experiments showed similar results. Experiments were performed at least in triplicate. Values are the means ± SE. Significance was calculated using the Student's t test. The Benjamini and Hochberg false discovery rate was used to calculate significance for the GO analyses (34)
.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Activin A treatment strongly inhibited the mineralization process (Fig. 1
A) (up to 50–100%). Three other mineralization models were treated with activin A to demonstrate that inhibition of matrix mineralization by activin A is not unique for the SV-HFO osteoblast model. Activin A was tested in another human osteoblast model (NHOst) and in human mesenchymal stem cell (MSC) cultures that were induced to differentiate toward osteoblasts (Fig. 1B, C
). Similar to SV-HFO cells, mineralization in both human osteoblast models was inhibited by activin A. In addition, human VSMCs were treated with activin A. VSMCs can be used as a model for vascular mineralization since they can form a mineralized ECM in vitro (36)
. Activin A treatment strongly inhibited matrix mineralization in these VSMC cultures (Fig. 1D
).
|
The effect on matrix mineralization was dose dependent, showing significant inhibition after treatment with activin A concentrations of 5 ng/ml and higher (Fig. 1E
). Together with decreased matrix mineralization, activin A also decreased the activity of the osteoblast differentiation marker ALPL dose-dependently, although the magnitude of inhibition (10–30%) was smaller than that of mineralization (data not shown).
Osteoblasts produce activin A in a differentiation-dependent manner
To reveal whether activin can act in an autocrine manner, we measured mRNA levels of the activin subunits inhibin-ßA (INHBA) and inhibin-ßB (INHBB) together with activin A protein production in human osteoblast cultures. QPCR demonstrated that INHBA was expressed abundantly compared with the INHBB subunit. In contrast, inhibin- (INHA) mRNA was almost undetectable (Fig. 2
A). This suggests that osteoblasts mainly produce activin A. In addition, Fig. 2A
shows mRNA expression of the activin type I receptor (ACVR1B) and the activin type II receptors (ACVR2A/2B) in osteoblasts.
|
Quantification of activin A protein showed that osteoblast cultures secrete biologically relevant amounts of activin A (up to 5 ng/ml), whereas inhibin A protein could not be detected (data not shown). Activin A production corrected for cell number (DNA content) was measured at different stages of differentiation (days 5, 12, and 19). Activin A production decreased during differentiation and was lowest during stages of matrix mineralization (days 12 and 19) (Fig. 2B
, mineralizing osteoblasts). To explore this differentiation-dependent regulation, activin A production was also measured in preosteoblast cultures (SV-HFO) that did not undergo differentiation into fully functional and mineralizing osteoblasts. Nonmineralizing osteoblasts (control osteoblasts) were created by excluding glucocorticoids (DEX) from the culture medium. These osteoblast cultures had low ALP activity and did not show in vitro mineralization (31)
. Control osteoblasts had significantly higher activin A production than their mineralizing counterparts (Fig. 2B
), demonstrating that, in differentiated osteoblasts, activin A production is suppressed during mineralization.
In addition, immunohistology on human bone biopsies showed that human bone tissue contained high levels of activin A. The inhibin-ßA subunit was detected in mineralized bone matrix, but inhibin- subunit could not be detected (Fig. 2C
).
Activin A inhibits mineralization in an autocrine manner
The impact of endogenously produced activins on osteoblast function was measured by neutralizing activin signaling by the addition of the activin binding protein follistatin. First we measured whether follistatin was able to neutralize activin signaling in osteoblast cultures. An activin-signaling luciferase reporter construct (CAGA-Luc) was used to measure activin signaling. Activin A treatment increased CAGA-Luc activity dose-dependently, which could be inhibited by coincubation with follistatin (Fig. 3
A). Follistatin also reduced basal CAGA-Luc signaling, which further supports endogenous production of activin and activation of activin signaling in osteoblasts. It has been suggested that follistatin also neutralizes BMP action (37
, 38)
. To study this in osteoblasts, a BMP-responsive luciferase reporter construct was used (BRE-Luc). Addition of 10 and 100 ng/ml BMP2a strongly induced BRE-Luc activity. In contrast to activin signaling, addition of follistatin had no effect on either basal or BMP2a-induced BMP signaling (Fig. 3B
). This suggests that, under these conditions, follistatin specifically neutralizes activin signaling in osteoblasts.
|
Neutralization of endogenous activins by follistatin treatment (100 and 500 ng/ml) clearly increased matrix mineralization (Fig. 3C
) together with a minor increase in ALPL activity (data not shown). This shows that osteoblasts secrete activin A, which inhibits mineralization and ALPL activity in a paracrine and/or autocrine way. In these experiments, osteoblasts were also treated with activin A to compare the impact of activin A with that of follistatin. These incubations demonstrated a dose-dependent effect of activin signaling on mineralization, where follistatin- (500 ng/ml), vehicle-, and activin A-treated cultures (50 ng/ml) represent neutralized, endogenous, and high activin signaling, respectively (Fig. 3C
).
Activin A-to-follistatin ratio is decreased in mineralizing cultures
Besides activin A production by osteoblasts, we demonstrated that osteoblasts secrete biologically relevant amounts of follistatin. On day 12 of culture, maximum levels of follistatin (up to 1–3 ng/ml) could be measured in the culture supernatants. This indicates that, in osteoblasts, endogenous activin signaling is regulated by the expression of activin A as well as by expression of follistatin. The production of follistatin was measured (corrected for DNA content) throughout culture in mineralizing and control osteoblasts (Fig. 3D
). Follistatin production was higher at the beginning of culture (day 5) in mineralizing osteoblast compared with control osteoblasts and decreased during culture in both conditions. For the eventual action of activin A, it is important to take into account the molar ratio of activin A and follistatin. This demonstrated that mineralizing cultures have a lower activin A-to-follistatin ratio compared with control cultures, indicating suppressed activin signaling in mineralizing conditions (Fig. 3E
).
Activin signaling is most effective prior to the onset of mineralization
We investigated whether activin signaling inhibits mineralization in a specific time window of osteoblast differentiation by treating cultures with activin A during specific periods. First, cultures were treated in the premineralization period (up to day 10 of culture) or during the mineralization period (from day 10 of culture onward) with a moderate dose of activin A (20 ng/ml). Subsequently, mineralization was measured on day 19 (Fig. 4
A). Treatment preceding the mineralization period most effectively inhibited mineralization, whereas treatment in the period when mineralization was ongoing appeared to be ineffective. To zoom in on this premineralization period, more specific incubations were performed, as depicted in Fig. 4B
. Activin A was most effective when present in the final 7 days before the onset of mineralization. Even when activin A was present only during the final 2 days (days 10–12) before mineralization, a significant decrease in mineralization was measured (Fig. 4B
).
|
We hypothesized that activin signaling in this premineralization phase leads to an altered matrix composition, keeping it in an immature state not able to mineralize. To prove this, we exploited the fact that osteoblast cultures produce an ECM within the first 10 days of culture that is mature and can subsequently mineralize independent of the presence of additional living osteoblasts (39)
. This effect was achieved by freezing osteoblast cultures to –20°C at the onset of mineralization (day 10). These cultures containing nonliving cells but an intact ECM were subsequently incubated for an additional 3 or 5 days with culture medium. Mineralization was quantified at the moment of fixation (day 10 living cultures=day 0 devitalized cultures) and after 3 and 5 days of additional incubation (see incubation scheme Fig. 4C
). No mineralization was measured on day 0 of devitalized cultures. However, after 3 and 5 days, mineralization of the ECM could be measured (Fig. 4C
, vehicle). In cultures that had been pretreated with activin A prior to devitalization, no mineralization of the ECM could be detected. These findings support our hypothesis that the major impact of activin A occurs before the onset of mineralization and suggest an effect on matrix composition and maturation.
Gene profile analysis of follistatin- and activin A-treated osteoblast cultures
Gene profile experiments were performed to gain insight into the molecular action of activin A in osteoblasts using Affymetrix HG U133 Plus 2.0 Genechips. Gene expression profiles of follistatin- (500 ng/ml), vehicle-, and activin A- (50 ng/ml) treated osteoblasts were compared. These treatments created cultures having low, moderate (endogenous), and high activin signaling, resulting in three different levels of matrix mineralization as demonstrated in Fig. 3C
.
A phenotype-based query was designed to identify genes regulated by activin signaling in the premineralization period (days 5 and 12), which was shown to be important for inhibition of mineralization by activin A (Fig. 4B
). An important aspect in the selection query was that we searched for genes that were regulated in follistatin- and activin A-treated cells, since both conditions affected mineralization. Genes were selected that were up-regulated in activin A-treated cultures (low mineralization) and down-regulated in follistatin-treated cultures (high mineralization) compared with vehicle cultures, or vice versa, on days 5 and 12. In addition, the difference in gene expression between activin A- and follistatin-treated cultures should be at least 2-fold on one of these 2 days (Fig. 5
A). In total, 520 Affymetrix IDs were selected representing 397 unique genes and 62 nonannotated Affymetrix IDs. The selected Affymetrix IDs are presented as a hierarchical clustering in Fig. 5A
; red and green indicate up- and down-regulation compared with vehicle cultures, respectively.
|
Gene ontology analysis
Next, GO enrichment analysis was used to categorize the 397 selected genes for the GO categories: biological process (Fig. 5B
), molecular function (Fig. 5C
), and cellular component (Fig. 5D
). In total, we identified 47 biological processes, 13 molecular functions, and 4 cellular components as significantly over-represented compared with what would be expected if 397 random genes were analyzed. Several GO terms for biological process shared high similarity and were identified by similar groups of genes. In Fig. 5B
, we reduced the number of biological process terms to 20 by excluding these overlapping GO terms.
QPCR showed that mRNA expression of osteoblast differentiation markers like RUNX2, collagen type-I, secreted phosphoprotein 1 (osteopontin), and bone gamma-carboxyglutamate gla protein (osteocalcin) were unchanged by activin signaling (data not shown), suggesting no direct effect on the osteoblast differentiation process. However, GO biological process analysis demonstrated that differentiation-related processes like development and morphogenesis were strongly enriched by activin signaling (Fig. 5B
), including the osteoblast transcription factor MSX2 (40)
. QPCR confirmed GeneChip analyses that MSX2 mRNA expression was significantly inhibited by activin A in osteoblasts (SV-HFO) as well as in VSMCs (Fig. 5E
).
The GO enrichment analysis for cellular component showed that genes related to localization in the extracellular region, space, or matrix were significantly enriched (Fig. 5D
). This result is in line with our hypothesis that activin A affects extracellular matrix formation and maturation (see Fig. 4C
). Although expression of the well-known and abundant ECM proteins in bone, collagen type-I, osteopontin, and osteocalcin was unchanged by activin signaling, these GeneChip analyses identified numerous ECM genes that were regulated by activin signaling (Fig. 6
A).
|
Activin signaling alters ECM composition in osteoblasts as well in VSMCs
A selection of ECM genes was analyzed in more detail using QPCR in follistatin-, vehicle-, and activin-treated osteoblasts and VSMCs (Fig. 6B
). The expression of ALPL was also quantified as a positive control. This experiment showed that CLEC3B, COL5A3, TIMP4, NID2, and ALPL, which were suppressed by activin A (upper 5 panels) in osteoblasts, were also suppressed by activin A in VSMC cultures. Activin signaling in osteoblasts increased POSTN and MMP2 expression, which was also demonstrated in VSMCs. In contrast, activin A increased FBLN5 and CSPG2 expression in osteoblasts, but the expression of these genes was unchanged in VSMCs.
Follistatin treatment did not alter gene expression in VSMCs. This result agrees with the expression data of INHBA and INHBB (mRNA) in VSMC, both of which had a low level compared with levels in osteoblasts (data not shown). Moreover, activin A production by VSMCs cultures was much lower than the production by osteoblasts (data not shown). In VSMCs, only POSTN mRNA expression was regulated by follistatin (0.5-fold). This gene was strongly induced by activin signaling and is probably already activated by a low concentration of activins (Fig. 6B
).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 41)
. We demonstrated that, in osteoblasts, follistatin acts as a specific inhibitor of activin A and not of BMP2. This is in line with other data showing that follistatin does not inhibit BMP2 (42)
. It is tempting to speculate that the function of the activin-follistatin mechanism is to control the extent of mineralization and, for example, to prevent overmineralization of bone tissue, which can result in brittle bones that are too stiff and unable to reform during loading (1)
.
Suppression of osteogenesis by activins was also found by others using murine and rat osteoblasts (28
, 29)
. Other studies performed in rodent models showed stimulation of osteogenesis, bone formation, and fracture healing by activins (25
26
27
, 43
, 44)
. Mice deficient for the inhibin-ßA subunit and follistatin have been generated and have shown abnormal tooth and craniofacial development after birth, but no other peculiar skeletal phenotypes (45
, 46)
. Unfortunately, both of these available systemic knockout mouse models die within 24 h of birth, making it impossible to study postnatal bone development and bone remodeling. Follistatin-overexpressing transgenic mice are viable, but no skeletal phenotype has yet been described in these mice (47)
. Whether the lack of effects or opposite effects in murine models reflect yet-unexplained species differences is not clear. However, these studies lack detailed analysis of bone tissue (e.g., bone mineral content, bone mineralization rate, and bone microarchitecture).
It has been shown that TGFß, a structural relative of activin, can inhibit matrix mineralization in osteoblast cultures (8
9
10)
. At the signal transduction level, activins and TGFß both stimulate Smad2/3 phosphorylation (11)
, which might explain the overlapping effects of activin and TGFß in vitro. Even though the effects of TGFß on matrix mineralization are similar to the effects we describe for activin A, the physiology might differ. TGFß is present as a latent complex in the bone matrix, and accumulated TGFß is activated only during bone resorption by osteoclasts (48
, 49)
. In contrast, activins are produced and released directly in an active form. Therefore, activin and TGFß differ in bioavailability during osteoblast development in vivo. TGFß may activate osteoblasts (precursors) during bone resorption when inactive TGFß becomes activated, whereas activin and follistatin control osteoblast differentiation and mineralization in a mechanism independent of a direct coupling to resorption.
The current mineralization data demonstrate that, in osteoblasts, activin treatment before the initiation of mineralization is crucial for strong inhibition of mineralization. On the basis of these data, it can be hypothesized that activin causes an altered ECM composition in such a way that the mineralization potential of the ECM is strongly inhibited. This hypothesis is substantiated by gene expression studies. The unbiased gene expression profiling approach demonstrated an altered expression of numerous ECM genes by activin signaling. Little is known about the role of the majority of these ECM proteins in matrix mineralization, which makes it difficult to interpret the full extent and significance of these ECM proteins alone or in combination. Only a limited number of deficient mice have been generated for these ECM genes, of which the majority lack a bone phenotype. Although for functional matrix protein analysis, knocking out or changing the expression of only a single ECM might not be sufficient to alter matrix mineralization. The combination of matrix proteins (i.e., the ECM composition) might be of greater importance for the eventual matrix quality and mineralization capacity. Nevertheless, the observed direction of regulation coupled to the mineralization phenotype may hint at the function of these ECM genes. ECM genes that are up-regulated by activin signaling (low mineralization) may act negatively on mineralization, whereas ECM genes that are down-regulated may act positively. In support of this is the observation that C-type lectin domain family 3 member B (CLEC3B, also known as tetranectin), down-regulated by activin signaling, has been shown to be positively involved in mineralization (50)
, whereas periostin (POSTN) and chondroitin sulfate proteoglycan 2 (CSPG2), up-regulated by activin signaling, are negatively implicated in mineralization (51
52
53)
. The direction of the regulation of genes involved in ECM turnover (MMP2, ADAMTS3, and TIMP4) is also in line with this hypothesis.
To extend the observation beyond skeletal development and mineralization, we showed that in a VSMC-based model for vascular mineralization, matrix mineralization was also inhibited by activin A treatment. This indicates that, besides osteoporosis, these findings can also be of great importance for vascular mineralization, which is a major risk factor for cardiovascular failure (54
, 55)
. Other studies have shown that activin A is expressed in VSMC cultures and in atherosclerotic lesions (56
, 57)
. Moreover, it was demonstrated that activin A promotes the contractile phenotypes of VSMC (56)
. Promotion of this VSMC phenotype might simultaneously inhibit VSMC osteogenesis and matrix mineralization. On the other hand, activin A might inhibit VSMC-induced matrix mineralization via a mechanism similar to that in osteoblasts. This is supported by our observations that several ECM genes were similarly regulated by activin A in VSMCs and osteoblasts. Shao and colleagues showed that Msx2 promotes cardiovascular calcification by activating paracrine Wnt signaling (58)
. We showed that activin signaling suppressed MSX2 expression in VSMCs as well as osteoblasts. According to GeneChip data, paracrine Wnt signaling seemed to be unaffected, with the exception of WNT2B. QPCR analyses demonstrated that activin signaling significantly decreased WNT2B mRNA expression in osteoblast and VSMC cultures (data not shown).
In summary, the key findings are 1) osteoblasts express activin A and its natural inhibitor follistatin to control activin signaling in a differentiation-dependent manner, 2) activin inhibits mineralization in a human bone formation model as well as in a model for vascular mineralization, and 3) activin does so by changing the expression of a wide range of matrix proteins before the onset of mineralization leading to a matrix composition with no or reduced mineralizing capacity. This led to the conclusion that activin signaling is a potent regulator of bone matrix formation and mineralization, and thereby an interesting mechanism in the control and maintenance of bone quality. Mineralization can be controlled in two directions using activin A, follistatin, or analogs of these compounds. As a consequence, activin signaling and activin target genes are important therapeutic targets for controlling matrix mineralization in bone as well as mineralization in pathological conditions.
Received for publication January 10, 2007. Accepted for publication March 15, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. S. Pearsall, E. Canalis, M. Cornwall-Brady, K. W. Underwood, B. Haigis, J. Ucran, R. Kumar, E. Pobre, A. Grinberg, E. D. Werner, et al. A soluble activin Type IIA receptor induces bone formation and improves skeletal integrity PNAS, May 13, 2008; 105(19): 7082 - 7087. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||
