Introduction

Bone is a dynamic tissue with high metabolic demand, which comprises bone-forming and bone-resorbing processes and includes several specialised cells (Grabowski 2009; Buckwalter and Cooper 1987). Amongst them, osteoblasts are the bone-forming cells and their differentiation as well as bone metabolism involves a range of factors, including several regulatory proteins (Blair et al. 2017).

The extracellular matrix (ECM) protein Agrin is a member of the family of heparan sulphate proteoglycans, containing more than 2000 amino acids with approximately 500 kDa of molecular weight. This protein acts at the neuromuscular junction (NMJ) regulating the synapses between presynaptic motor neurons and postsynaptic muscle fibres (Li et al. 2018). In this structure, Agrin is released by motor neuron and interacts with the transmembrane low-density lipoprotein receptor-related protein 4 (Lrp4) activating the co-receptor tyrosine kinase muscle-specific tyrosine kinase (Musk) leading to acetylcholine receptor (AChR) aggregation that is important to the formation of a functional NMJ (Herbst, 2020; Shen et al. 2015; Kim et al. 2008).

Although the main and best-known function of Agrin is the formation and maintenance of NMJ, it is also expressed in several tissues, where it plays distinct roles that include cell adhesion, proliferation, motility, and even regulation of invasiveness in some types of cancer (Anselmo et al. 2016; Chakraborty et al. 2015; Daniels 2012). Previous studies have shown Agrin signalling in the immunological synapse of T cells with antigen presenting cells, whose receptor involved is α-dystroglycan (Dag1), a cell surface receptor with high affinity for ECM proteins (Barresi and Campbell 2006; Zhang et al. 2006). It was also observed that Agrin is able to induce the division of cardiomyocytes in vitro by mechanisms that involve the disruption of dystrophin-glycoprotein complexes and signalling mediated by yes-associated protein (Yap) and kinases regulated by extracellular signals (Erk), in addition to promoting cardiac regeneration in mice (Bassat et al. 2017). High expression of Agrin and its oncogenic role were previously demonstrated in oral squamous cell carcinoma (Kawahara et al. 2015). Despite the presence in various tissues, the functions of Agrin in non-neuronal tissues are still underexplored and poorly understood. Regarding the skeletal system, until now, there are only three studies of Agrin contribution to the cartilage and bone formation processes (Eldridge et al. 2016; Mazzon et al. 2011; Hausser et al. 2007).

Agrin has been detected in chondrocytes and cartilage and its in vitro knockdown resulted in downregulation of the cartilage transcription factor Sry-related high-mobility group box 9 (SOX9) as well as other chondrocyte differentiation markers such as aggrecan and collagen type II, whilst in vivo exogenous Agrin promoted ectopic cartilage formation (Eldridge et al. 2016). Moreover, it was demonstrated that transgenic mice, with reduced expression of Agrin in chondrocytes, present morphological changes in the growth plate and severe retardation of the skeletal development (Hausser et al. 2007). Agrin was also detected in mesenchymal stromal cells (MSC) and osteoblasts of the endosteum presenting a role in the interaction between these cells and hematopoietic stem cells (HSC) to promote the survival of the later ones (Mazzon et al. 2011). In addition, considering the close relationship between cartilage and bone, Agrin could be involved in the substitution of cartilage by bone, which requires multiple steps that directly involve several components of the ECM (Eldridge et al. 2016; Grol and Lee 2018; Gentili and Cancedda 2009; Hausser et al. 2007). Furthermore, Lrp4, the main receptor that triggers Agrin signalling, is involved in the modulation of wingless-related integration site (Wnt) signal transduction, which is an important pathway in the osteogenic differentiation and bone formation (Houschyar et al. 2019; Ahn et al. 2017; Shen et al. 2015).

Based on the findings of the aforementioned studies and considering the lack of studies focused on Agrin function in osteoblastic cells, this study was designed to test the hypothesis that Agrin contributes to osteoblast differentiation. Indeed, the results showed that Agrin is expressed by osteoblasts from three different sources and that Agrin knockdown disrupts whilst its addition increases the osteoblast differentiation. Additionally, as Lrp4 acts as receptor for both Agrin and Wnt and due to the crosstalk between Wnt and bone morphogenetic protein (BMP) signals (Zhang et al. 2013; Itasaki and Hoppler, 2010), these results suggest that the Agrin-Wnt-BMP circuit could be involved in the Agrin-mediated osteoblast differentiation.

Material and methods

Cell cultures

Mouse pre-osteoblastic cell line MC3T3-E1

The MC3T3-E1 pre-osteoblastic cell line, subclone 14, was acquired from the American Type Culture Collection (ATCC, Manassas, VA, USA) and initially cultured in 75-cm2 flasks (Corning Incorporated, New York, NY, USA) in a regular medium composed by alpha-minimum essential medium (α-MEM, Gibco-Invitrogen, Grand Island, NY, USA) supplemented with 10% foetal bovine serum (Gibco-Invitrogen) and 1% penicillin–streptomycin (Gibco-Invitrogen) until reaching 70% of confluence. After that, cells were enzymatically detached, transferred to 24-well plates (Corning Incorporated) in the quantity of 2 × 104 cells per well, containing osteogenic medium to stimulate osteoblast differentiation, and cultured for up to 14 days. The osteogenic medium was regular medium plus 50 µg/mL ascorbic acid (Gibco-Invitrogen) and 7 mM β-glycerophosphate (Sigma-Aldrich, Darmstadt, Germany). The medium was changed every 2–3 days and cells were kept in a controlled environment at 37 °C, 95% air and 5% carbon dioxide (CO2).

Osteoblasts differentiated from bone marrow mesenchymal stromal cells (MSC-OB)

Seven-week-old C57BL/6 mice were used to obtain MSC harvested from bone marrow by flushing both femurs and tibias and characterised (Fig. s1) as previously described (Freitas et al. 2020). Then, cells were initially cultured during 7 days to select the MSC by adherence to the polystyrene of cell culture flask, enzymatically detached and plated in the same conditions, and time described above to MC3T3-E1 cells in order to induce osteoblast differentiation (MSC-OB), excepting the osteogenic medium, in which 10−7 M dexamethasone (Sigma-Aldrich) was added.

Calvaria-derived osteoblasts (CA-OB)

Newborn C57BL/6 mice calvarias (7-day-old) were harvested and osteoblasts isolated as previously described (Helfrich and Ralston 2012). Briefly, mice calvaria were dissected, cleaned, and cut into small pieces. To isolate osteoblasts (CA-OB), bone fragments were submitted to a digestion by using 0.25% trypsin (Gibco-Invitrogen) plus collagenase type II 0.20% (Gibco-Invitrogen). Then, cells were initially cultured, enzymatically detached and plated in the same conditions, and time described above to MC3T3-E1 cells to induce osteoblast differentiation.

Gene expression of Agrin and its receptors during osteoblast differentiation

Quantitative real-time polymerase chain reaction (RT-qPCR) was carried out at 3, 5, 7, 10, and 14 days to evaluate temporal gene expression of Agrin. To investigate a possible correlation between the expression of Agrin and the progression of osteoblast differentiation, bone gamma-carboxyglutamate protein (Bglap, osteocalcin) gene expression was measured at the same time points. In addition, temporal gene expression of the Agrin receptors Lrp4, Musk, and Dag1 was evaluated at 5, 7, 10, and 14 days, as well as a possible correlation between them.

Real-time quantitative polymerase chain reaction (RT-qPCR)

Trizol reagent (Invitrogen, Carlsbad, CA, USA) was added to each sample to extract the total ribonucleic acid (RNA). Then, a reverse transcription reaction (M-MLV, Promega Corporation, Madison, WI, USA) was performed by using 1 µg of RNA to synthesise a complementary deoxyribonucleic acid DNA (cDNA). The RT-qPCR reactions were carried out in the Step One Plus Real-Time PCR System (Thermo Fisher Scientific, Waltham, MA, USA). TaqMan universal PCR master mix AmpErase UNG 2X (Applied Biosystems, Foster City, CA, USA) and TaqMan probes (Table s1, 20 × TaqMan gene expression assay mix) or Fast SybrGreen Master-Mix (Applied Biosystems) and specific primers designed with Primer Express 2.0 (Table s2, Applied Biosystems) were used. The relative gene expression of β-actin was used to normalise all gene expressions (n = 3) by using the 2−ΔΔCt method (Livak and Schmittgen 2001).

AGRIN protein detection by immunofluorescence labelling

Protein expression of AGRIN was detected by immunofluorescence labelling on day 3. Briefly, 4% paraformaldehyde solution (Sigma-Aldrich) in 0.1 M sodium phosphate buffer (PB), pH 7.2 was used to fix the cells for 10 min and 0.5% Triton X-100 (Acros Organics, Geel, Belgium) in PB was used to permeabilise them for 10 min, both at room temperature. Then, the samples were blocked with 5% skimmed milk in PB for 30 min. Primary rabbit polyclonal antibody anti-AGRIN (1:200, ab85174, Abcam, Cambridge, MA, USA) and Alexa Fluor 594 conjugated with goat anti-mouse secondary antibody (1:200, Thermo Fischer Scientific, Waltham, MA,USA) were incubated for 1 h and 50 min, respectively. Alexa Fluor 488 conjugated with phalloidin (1:200, Thermo Fisher Scientific) was incubated for 50 min to visualise the actin cytoskeleton. The cell nuclei (blue fluorescence) were stained with 300 nM 4′,6-diamidino-2-phenylindole,dihydrochloride (DAPI, Thermo Fischer Scientific). After that, an epifluorescence microscope (Axio Imager M2, Carl Zeiss, Inc. Göttingen, GO, GE) coupled to a digital camera (AxionCam MRm, Carl Zeiss, Inc.) was used to observe the cells and the images were acquired, merged, and analysed by the Zen Software (Carl Zeiss, Inc.).

Agrin knockdown by small interfering RNA (siRNA)

Agrin knockdown was performed through the RNA interference technique, using siRNA, which has a specific sequence complementary to the target messenger RNA (mRNA) (Fakhr et al. 2016). After subconfluence, MC3T3-E1 cells were cultured at a density of 1 × 105 cells per well in 6-well plates (Corning Incorporated) containing osteogenic medium. The transfection was performed after 7 days of culture by using siRNA to Agrin (4390771, Life Technologies, Carlsbad, CA, USA) and nonspecific siRNA (Scrambled) (4390843, Life Technologies), at a concentration of 30 ρmol per well and lipofectamine reagent RNAiMax (Invitrogen). Briefly, in a tube, 9 µl of lipofectamine were diluted in 150 µl of reduced-serum medium (Opti-MEM, Gibco-Invitrogen). Then, in another tube, 30 ρmol of each siRNA was also diluted in 150 µl of Opti-MEM. The complexes were formed by mixing the reagents of the two tubes and that solutions were incubated for 5 min at room temperature. Then, the RNAiMax lipofectamine siRNA complexes were added to the cells. Five days post-transfection, the effectiveness of Agrin knockdown was evaluated by analysing the gene and protein expression through RT-qPCR and immunofluorescence labelling, respectively, as described above.

Effect of Agrin knockdown on the expression of its receptors and osteoblast differentiation

The effect of Agrin knockdown on the expression of its receptors and osteoblast differentiation was evaluated in Agrin-silenced (siAgrin) and non-silenced (Scrambled) MC3T3-E1 cells cultured for 5 days post-transfection. Then, the gene expression of the receptors Lrp4 and Dag1, as well as the osteoblast markers runt-related transcription factor 2 (Runx2), osterix (Sp7), alkaline phosphatase (Alpl), integrin-binding sialoprotein (Ibsp, bone sialoprotein), Bglap, and alpha-1 type I collagen (Col1a1) was evaluated by RT-qPCR as described above. In addition, osteoblast phenotype was assessed by RUNX2 protein expression and in situ ALP activity, as described below.

RUNX2 protein expression by western blotting

RUNX2 protein expression was evaluated by western blotting in siAgrin and Scrambled MC3T3-E1 cells. On day 5 post-transfection, 140 µl of buffer constituted by 1 × protease inhibitor mixture (Roche Applied Science, Indianapolis, IN, USA), 1 mM phenylmethanesulfonyl fluoride (Sigma-Aldrich), and 25 mM MG132 proteasome inhibitor (Roche Applied Science) were used for cell lysis. Then, 80 µg of total protein of each sample were denatured, separated by electrophoresis gel, and transferred to a membrane, which was blocked for 2 h at room temperature, with a solution containing Tris-buffered saline 0.1% Tween-20 (TBS-T, Sigma-Aldrich) and 5% skimmed milk prior to antibody incubation. Primary rabbit monoclonal antibody anti-RUNX2 (1:1000, 8486S, D1H7, Cell Signalling Technology, Danvers, MA, USA) was added and incubated overnight at 4 °C and followed by secondary antibody, goat anti-rabbit IgG–horseradish peroxidase (HRP) conjugate (1:2000, Santa Cruz Biotechnology, Santa Cruz, CA, USA) during 1 h at room temperature. As a control, mouse monoclonal GAPDH antibody (1:1000, Santa Cruz Biotechnology) was used, followed by secondary antibody goat anti-rabbit (1:2000, Santa Cruz Biotechnology). Western lightning chemiluminescence reagent (Perkin Elmer Life Sciences, Waltham, MA, USA) and G:Box gel imaging (Syngene, Cambridge, UK) were used to detect proteins and acquire images. Then, the pixels were counted and RUNX2 protein expression (n = 3) was normalised to the expression of GAPDH.

In situ ALP activity

In situ ALP activity of siAgrin or Scrambled MC3T3-E1 cells was detected on day 5 post-transfection and qualitatively analysed by Fast Red staining (Sigma-Aldrich). Briefly, the culture medium was removed, and 1 mL of the following solution was added to each well:1.8 mM Fast Red-TR Salt (Sigma-Aldrich), 0.9 mM Naphtol AS-MX phosphate (Sigma-Aldrich), and 4 mg/mL of dimethylformamide (Sigma-Aldrich). The samples were kept in a controlled environment at 37 °C, 95% oxygen (O2) and 5% carbon dioxide (CO2) for 30 min, and then the solution was removed, the wells dried and photographed using a high-resolution digital camera (Canon EOS Digital Rebel Camera, Canon, Lake Success, NY, USA). The area of each well was used to quantify the pixels by using the ImageJ 1.52 software (National Institute of Mental Health, Bethesda, MD, USA) and the data were expressed as a percentage of area.

Effect of Agrin knockdown on the modulation of genes related to Wnt and BMP signalling pathways

To evaluate the effect of Agrin knockdown on the modulation of genes related to Wnt and BMP signalling pathways, siAgrin and Scrambled MC3T3-E1 cells were cultured for 5 days post-transfection and a RT-qPCR was performed as described above. The expression of Wnt signalling components frizzled class receptor 6 (Fzd6), frizzled class receptor 9 (Fzd9), Lrp4, low-density lipoprotein receptor-related protein 6 (Lrp6), beta-catenin (Ctnnb1), transcription factor 7 (Tcf7), and lymphoid enhancer factor 1 (Lef1) and the BMP signalling components activin A receptor type 1 (Alk2), bone morphogenetic protein receptor type 1A (Bmpr1a), bone morphogenetic protein receptor type 2 (Bmpr2), SMAD family member 1 (Smad1), SMAD family member 4 (Smad4), and activating transcription factor 4 (Atf4) were evaluated.

Agrin treatment

Selection of recombinant Agrin concentration

To select the most osteogenic concentration of Agrin, the MC3T3-E1 cells, MSC-OB, and CA-OB were cultured and plated in the same conditions described above for up to 5 days and treated from the first day of osteoblast differentiation with 10, 50, 100, or 200 ng/mL of recombinant Agrin (R&D systems, Minneapolis, MN, USA). The vehicle was used as a Control and the osteogenic medium was changed every 2 days. On day 5, the gene expression of Runx2, Sp7, Alpl, Bglap, and Col1a1 was evaluated by RT-qPCR as described above.

Effect of Agrin treatment on the expression of Agrin, its receptors, osteoblast differentiation, and modulation of genes related to Wnt and BMP signalling pathways

The effect of Agrin treatment was evaluated in MSC-OB cultured in the same conditions described above and plated in osteogenic medium containing recombinant Agrin in the previously selected concentration for up to 10 days. Then, the gene expression of Agrin, Lrp4, Dag1, and of the bone markers Runx2, Sp7, Alpl, Ibsp, Bglap, and Col1a1 was evaluated on day 5 by RT-qPCR, and the osteoblast phenotype was assessed on day 10 by RUNX2 protein expression and in situ ALP activity, as described above. In addition, the modulation of genes related to Wnt (Fzd6, Fzd9, Lrp6, Ctnnb1, Tcf7, and Lef1) and BMP (Alk2, Bmpr1a, Bmpr2, Smad1, Smad4, and Atf4) signalling pathways was evaluated also on day 5 by RT-qPCR as described above.

Statistical analysis

Analysis of variance (ANOVA) was used to compare the data of temporal gene expression of Agrin, its receptors and Bglap during osteoblast differentiation, as well as of the selection of Agrin concentration, followed by the Tukey test, when appropriated. The correlation between the expression of Agrin and the expression of Bglap, Lrp4, and Dag1 was determined by the Spearman’s rank correlation coefficient. Student’s t-test was used to compare Scrambled and SiAgrin cells, as well as the cells treated with vehicle (Control) or recombinant Agrin. All numerical data are presented in charts as mean ± standard deviation and the level of significance was set at p ≤ 0.05.

Results

Expression of endogenous Agrin and its receptors during osteoblast differentiation

In the first set of experiments, the expression of Agrin and its receptors were investigated in cells from three different sources during osteoblast differentiation, using Bglap gene expression as marker of differentiation (Fig. 1).

Fig. 1
figure 1

Expression of Agrin during osteoblast differentiation. Gene expression of Agrin (a–c) and bone gamma-carboxyglutamate protein (Bglap; d–f) at 3, 5, 7, 10, and 14 days; low-density lipoprotein receptor-related protein 4 (Lrp4; g–i) and α-dystroglycan (Dag1; j–l) at 5, 7, 10, and 14 days; and protein detection of AGRIN (m–o) at 3 days in MC3T3-E1 cells, osteoblasts differentiated from bone marrow mesenchymal stromal cells (MSC-OB), and osteoblasts derived from calvaria (CA-OB), grown in osteogenic medium on polystyrene. AGRIN expression was predominantly perinuclear exhibiting a granular aspect throughout the cytoplasm (arrowheads). Actin filaments were organised as parallel bundles along the major axis of the cells (arrows). Red fluorescence indicates immunolocalisation of AGRIN, green fluorescence indicates actin cytoskeleton, and blue fluorescence indicates cell nuclei. The data of gene expression are presented as mean ± standard deviation (n = 3) and different letters indicate statistically significant differences (ANOVA, p ≤ 0.05). Scale bar (m, n, and o) = 100 µm

Full size image

The gene expression of Agrin in MC3T3-E1 cells (Fig. 1a) increased from day 3 to day 14, but day 7, when a slight drop was observed (p = 0.032 between day 5 and 7, p = 0.005 between day 5 and 10, and p = 0.001 for all other time point comparisons). Regarding Bglap (Fig. 1d), there was no change in the expression from day 3 to day 5, but it increased from day 5 until day 10, when it peaked and then decreased on day 14 (p = 0.208 between day 3 and 5, and p = 0.001 for all other time point comparisons). It was observed a positive correlation between the gene expression of Agrin and Bglap (Table s3; rs = 0.757, p = 0.001), whilst a non-significant positive correlation was observed between the gene expression of Agrin and Lrp4 (Table s4; rs = 0.392, p = 0.208) and Agrin and Dag1 (Table s5; rs = 0.462, p = 0.131). The expression of Lrp4 (Fig. 1g) decreased from day 5 to day 10, and increased on day 14 (p = 0.003 between day 5 and 10, and p = 0.001 for all other time point comparisons). The expression of Dag1 (Fig. 1j) decreased from day 5 to days 7 and 10, and increased on day 14 (p = 0.708 between day 7 and 10, and p = 0.001 for all other time point comparisons). The gene expression of Musk was not detected on any time point.

In MSC-OB, the gene expression of Agrin (Fig. 1b) remained stable until day 5 and from there increased over time until day 14, the last evaluated time point (p = 1.000 between days 3 and 5, and p = 0.001 for all other time point comparisons). The expression of Bglap (Fig. 1e) followed exactly the same pattern of Agrin (p = 0.743 between days 3 and 5, and p = 0.001 for all other time point comparisons). A positive correlation was observed between the gene expression of Agrin and Bglap (Table s3; rs = 0.932, p = 0.001), Agrin and Lrp4 (Table s4; rs = 0.783, p = 0.003), and Agrin and Dag1 (Table s5; rs = 0.923, p = 0.001). The expression of Lrp4 (Fig. 1h) increased from day 5 to days 10 and 14 (p = 0.302 between days 10 and 14, and p = 0.001 for all other time point comparisons). The expression of Dag1 (Fig. 1k) increased over time being higher on day 14 (p = 0.001 for all comparisons between periods). The gene expression of Musk was not detected on any time point.

In CA-OB, the gene expression of Agrin (Fig. 1c) increased over time until peaking on day 7, then decreased on day 10, and increased again on day 14 (p = 0.115 between days 5 and 10, and p = 0.001 for all other time point comparisons). There was no change in the expression of Bglap (Fig. 1f) from day 3 to day 5, then it increased from day 5 to days 7 and 10, being higher on day 14 (p = 1.000 between days 3 and 5, p = 0.021 between days 5 and 7, p = 0.203 between days 7 and 10, p = 0.020 between days 3 and 7, and p = 0.001 for all other time point comparisons). It was observed a non-significant positive correlation between the gene expression of Agrin and Bglap (Table s3; rs = 0.421, p = 0.118), whilst a positive correlation was observed between the gene expression of Agrin and Lrp4 (Table s4; rs = 0.734, p = 0.007) and Agrin and Dag1 (Table s5; rs = 0.769, p = 0.003). The expression of Lrp4 (Fig. 1i) increased from day 5 to day 7, decreased on day 10, and increased again on day 14 (p = 0.085 between days 7 and 14, p = 0.045 between days 10 and 14, and p = 0.001 for all other time point comparisons). The expression of Dag1 (Fig. 1l) increased from day 5 and peaked on day 7, then decreased on day 10, and increased again on day 14 (p = 0.001 for all time point comparisons). The gene expression of Musk was not detected on any time point.

The protein expression of AGRIN was detected by immunofluorescence labelling at 3 days of culture in cells from the three sources. In all cultures, the images showed adhered and spread cells with the actin filaments of cytoskeleton arranged parallel to the long axis of the cells and AGRIN was predominantly expressed in the perinuclear region of the cytoplasm and granularly (Fig. 1m–o). Despite being detected in the three cultures, AGRIN was more expressed in MC3T3-E1 and MSC-OB (Fig. 1m, n), with cells displaying strong red fluorescence in comparison to the weakly labelled CA-OB (Fig. 1o).

After the demonstration that osteoblasts from different sources synthesise Agrin and its gene expression positively correlates to Bglap expression, some experiments were designed to show that Agrin is involved in osteoblast differentiation. For this, Agrin was knockdown in MC3T3-E1 cells by using siRNA approach or Agrin was added to the culture media of MC3T3-E1, MSC-OB, and CA-OB.

Efficiency of cell transfection with siRNA to knockdown Agrin

Five days post-transfection, the gene and protein expression of Agrin were evaluated in MC3T3-E1 cells transfected with either nonspecific siRNA (Scrambled) or siRNA for Agrin (siAgrin) (Fig. 2). The knockdown was validated by a significant decrease (p = 0.001) of approximately 63% in the gene expression of Agrin in siAgrin compared to Scrambled cells (Fig. 2a). At the protein level, Scrambled cells showed intracellular AGRIN labelling mainly perinuclear (Fig. 2d), whilst in SiAgrin cells the AGRIN labelling was undetectable (Fig. 2e), indicating a strong reduction of AGRIN protein expression. Additionally, Agrin knockdown downregulated the gene expression of its receptors Lrp4 (Fig. 2e; p = 0.003) and Dag1 (Fig. 2c; p = 0.019).

Fig. 2
figure 2

Efficiency of cell transfection with siRNA to knockdown Agrin. Gene expression of Agrin (a), low-density lipoprotein receptor-related protein 4 (Lrp4; b), α-dystroglycan (Dag1; c), and protein detection of AGRIN (d and e) in MC3T3-E1 cells transfected with either nonspecific siRNA (Scrambled) or siRNA for Agrin (siAgrin), grown in osteogenic medium on polystyrene, 5 days post-transfection. Red fluorescence indicates immunolocalization of AGRIN, green fluorescence indicates actin cytoskeleton, and blue fluorescence indicates cell nuclei. The data of gene expression are presented as mean ± standard deviation (n = 3), and different letters indicate statistically significant differences (Student’s t-test, p ≤ 0.05). Scale bar (d and e) = 100 µm

Full size image

Effect of Agrin knockdown on osteoblast differentiation

Five days post-transfection, comparison between Scrambled and SiAgrin MC3T3-E1 cells showed that Agrin knockdown disrupts osteoblast differentiation at both genotype and phenotype levels (Fig. 3). The gene expression of Runx2, Sp7, Alpl, Ibsp, Bglap, and Col1a1 was downregulated by Agrin knockdown (Fig. 3a; p = 0.001 for all genes). Moreover, both RUNX2 protein expression (1.3-fold, p = 0.001; Fig. 3b) and the in situ ALP activity (1.5-fold, p = 0.016; Fig. 3c) were also reduced by Agrin knockdown.

Fig. 3
figure 3

Effect of Agrin knockdown on osteoblast differentiation. Gene expression of the bone markers (a) runt-related transcription factor 2 (Runx2), osterix (Sp7), alkaline phosphatase (Alpl), integrin-binding sialoprotein (Ibsp), bone gamma-carboxyglutamate protein (Bglap), and alpha-1 type I collagen (Col1a1), as well as protein expression (b) of runt-related transcription factor 2 (RUNX2) and in situ alkaline phosphatase (ALP) activity (c) in MC3T3-E1 cells transfected with nonspecific either siRNA (Scrambled) or siRNA for Agrin (siAgrin), grown in osteogenic medium on polystyrene, 5 days post-transfection. The data of gene expression (n = 3), RUNX2 protein expression (n = 3), and in situ alkaline phosphatase (ALP) activity (n = 5) are presented as mean ± standard deviation, and different letters indicate statistically significant differences (Student’s t-test, p ≤ 0.05)

Full size image

Effect of Agrin knockdown on modulation of genes related to the Wnt and BMP signalling pathway

As Lrp4 acts as receptor for both Agrin and Wnt and considering the crosstalk between Wnt and BMP signalling pathways (Zhang et al. 2013; Itasaki and Hoppler 2010), both playing significant roles in the osteoblast differentiation and bone formation (Karner et al. 2017; Karner and Long 2017), it is worth to investigate the impact of Agrin knockdown on the gene expression of components of both signalling pathways (Fig. 4). Five days post-transfection, Agrin knockdown downregulated the expression of genes related to Wnt signalling (Fig. 4a) Fzd6 (p = 0.004), Fzd9 (p = 0.001), Lrp6 (p = 0.045), Ctnnb1 (p = 0.001), Tcf7 (p = 0.001), and Lef1 (p = 0.001) as well as of genes related to BMP signalling (Fig. 4b) Alk2 (p = 0.001), Bmpr1a (p = 0.049), Bmpr2 (p = 0.001), Smad1 (p = 0.001), Smad4 (p = 0.001), and Atf4 (p = 0.001).

Fig. 4
figure 4

Effect of Agrin knockdown on modulation of genes related to the Wnt and BMP signalling pathway. Gene expression of Wnt signalling pathway components (a) frizzled class receptor 6 (Fzd6), frizzled class receptor 9 (Fzd9), low-density lipoprotein receptor-related protein 4 (Lrp4), low-density lipoprotein receptor-related protein 6 (Lrp6), beta-catenin (Ctnnb1), transcription factor 7 (Tcf7), and lymphoid enhancer factor 1(Lef1) and of BMP signalling pathway components (b) activin A receptor type 1 (Alk2), bone morphogenetic protein receptor type 1A (Bmpr1a), bone morphogenetic protein receptor type-2 (Bmpr2), SMAD family member 1 (Smad1), SMAD family member 4 (Smad4), and activating transcription factor 4 (Atf4) in MC3T3-E1 cells transfected with either nonspecific siRNA (Scrambled) or siRNA for Agrin (siAgrin), grown in osteogenic medium on polystyrene, 5 days post-transfection. The data are presented as mean ± standard deviation (n = 3), and different letters indicate statistically significant differences (Student’s t-test, p ≤ 0.05)

Full size image

Effect of Agrin treatment on the expression of Agrin, its receptors, and osteoblast differentiation

Due to the impaired osteoblast differentiation paralleled by the reduced expression of Lrp4 and Dag1 induced by Agrin knockdown, the third set of experiments was carried out to evaluate the effect of adding recombinant Agrin to the culture media on osteoblast differentiation of MC3T3-E1, MSC-OB, and CA-OB (Fig. s2).

Amongst all concentrations of recombinant Agrin and cell cultures evaluated, 50 ng/mL of Agrin in MSC-OB was able to induce higher gene expression of most of the bone markers (Fig. s2). For that reason, this concentration was selected to evaluate the osteoblast differentiation of MSC-OB.

At day 5, Agrin treatment downregulated the gene expression of Agrin (Fig. 5a; p = 0.001), upregulated Lrp4 (Fig. 5b; p = 0.028), and did not affect Dag1 (Fig. 5c; p = 0.206). Moreover, it was observed that the addition of Agrin slightly increases osteoblast differentiation at both genotype and phenotype levels (Fig. 5d–f). Whilst the gene expression of Runx2 (p = 0.013), Sp7 (p = 0.050), Ibsp (p = 0.004), and Bglap (p = 0.001) was upregulated by Agrin treatment, it did not affect the gene expression of Alpl (p = 0.065) and Col1a1 (p = 0.551) (Fig. 5d). Additionally, on day 10, RUNX2 protein expression (1.3-fold, p = 0.019; Fig. 5e) and the in situ ALP activity (1.3-fold, p = 0.045; Fig. 5f) were also increased by Agrin treatment.

Fig. 5
figure 5

Effect of Agrin treatment on the expression of Agrin, its receptors, and osteoblast differentiation. Gene expression of Agrin (a), low-density lipoprotein receptor-related protein 4 (Lrp4; b), α-dystroglycan (Dag1; c), and of bone markers (d) runt-related transcription factor 2 (Runx2), osterix (Sp7), alkaline phosphatase (Alpl), integrin-binding sialoprotein (Ibsp), bone gamma-carboxyglutamate protein (Bglap), and alpha-1 type I collagen (Col1a1), as well as protein expression (e) of runt-related transcription factor 2 (RUNX2) and in situ alkaline phosphatase (ALP) activity (f) in MSC-OB treated with vehicle (Control) or 50 ng/mL of recombinant Agrin, grown in osteogenic medium on polystyrene, for up to 10 days. The data of gene expression (n = 3), RUNX2 protein expression (n = 3), and in situ alkaline phosphatase (ALP) activity (n = 5) are presented as mean ± standard deviation, and different letters indicate statistically significant differences (Student’s t-test, p ≤ 0.05)

Full size image

Effect of Agrin treatment on modulation of genes related to the Wnt and BMP signalling pathways

For the reasons already mentioned, the impact of Agrin treatment on the gene expression of components of Wnt and BMP signalling pathways was investigated (Fig. 6). On day 5, Agrin treatment neither affect the expression of genes related to Wnt signalling (Fig. 6a) Fzd6 (p = 0.976), Fzd9 (p = 0.651), Lrp6 (p = 0.163), Ctnnb1 (p = 0.429), Tcf7 (p = 0.545), and Lef1 (p = 0.505) nor most of the genes related to BMP signalling (Fig. 6b), such as Bmpr1a (p = 0.196), Bmpr2 (p = 0.744), Smad4 (p = 0.951), and Atf4 (p = 0.667). However, Agrin treatment upregulated the gene expression of Alk2 (p = 0.017) and Smad1 (p = 0.020).

Fig. 6
figure 6

Effect of Agrin treatment on modulation of genes related to the Wnt and BMP signalling pathway. Gene expression of Wnt signalling pathway components (a) frizzled class receptor 6 (Fzd6), frizzled class receptor 9 (Fzd9), low-density lipoprotein receptor-related protein 4 (Lrp4), low-density lipoprotein receptor-related protein 6 (Lrp6), beta-catenin (Ctnnb1), transcription factor 7 (Tcf7), and lymphoid enhancer factor 1 (Lef1) and of BMP signalling pathway components (b) activin A receptor type 1 (Alk2), bone morphogenetic protein receptor type 1A (Bmpr1a), bone morphogenetic protein receptor type-2 (Bmpr2), SMAD family member 1 (Smad1), SMAD family member 4 (Smad4), and activating transcription factor 4 (Atf4) in MSC-OB treated with vehicle (Control) or 50 ng/mL of recombinant Agrin, grown in osteogenic medium on polystyrene, for up to 10 days. The data are presented as mean ± standard deviation (n = 3), and different letters indicate statistically significant differences (Student’s t-test, p ≤ 0.05)

Full size image

Discussion

Agrin was originally described approximately 30 years ago and its main known function is the maintenance, development, and organisation of the postsynaptic process at the NMJ (Bezakova and Ruegg 2003; McMahan 1990; Nitkin et al. 1987). Also, Agrin has been proposed as a stimulating factor for T cell activation and the formation of the immune synapse (Khan et al. 2001). Although new findings have shown that Agrin is not restricted to these functions, there is no evidence in the literature reporting on the role of Agrin in bone homeostasis, especially in osteoblast differentiation. Here, we have shown that Agrin and its receptors are expressed during osteoblast differentiation. Moreover, Agrin knockdown reduced the expression of its receptors and impaired osteoblast differentiation by negatively regulating components of both Wnt and BMP signalling pathways, whilst treatment with recombinant Agrin slightly increases osteoblast differentiation without affecting these pathways (Fig. 7).

Fig. 7
figure 7

A schematic diagram depicting the suggested mechanism by which the Agrin-Wnt-BMP axis regulates osteoblast differentiation

Full size image

To show if and how the expression of Agrin occurs during osteoblast differentiation, the temporal expression of Agrin and Bglap was evaluated in three different and well-established cultures of osteoblastic cells. Bglap was chosen because this gene encodes osteocalcin, which is the main non-collagen protein secreted by osteoblasts in the extracellular matrix of bone tissue and recognised as a marker of osteoblast differentiation thanks to its role in both bone formation and remodelling (Li et al. 2016; Owen et al. 1990). The expression of Agrin and Bglap was detected in all evaluated cultures, both exhibiting variations over time. The expression of Agrin and osteocalcin was also noticed in vivo in osteoblasts of the endosteum from mouse femurs, although no correlation between both proteins was determined since the evaluation was done in just one time point (Mazzon et al. 2011). Despite slight differences, a positive correlation between Agrin and Bglap expressions during osteoblast differentiation was observed in all evaluated cultures. In general, the expression of Agrin increased over time concomitantly with the increase of Bglap, which is associated with the progression of osteoblast differentiation as previously demonstrated (Owen et al. 1990).

Although the number of known molecules that interact with Agrin is increasing, the function of most of them remains unclear. Therefore, the temporal expression of Lrp4, Musk, and Dag1 complex receptors, which are essential for signalling triggered by Agrin at NMJ, was also evaluated (Burden et al. 2013; Bowe et al. 1994). At the NJM, after released from nerve terminals, Agrin binds to Lrp4 and Musk, forming a complex that activates a signalling cascade resulting in AChR aggregation, which is essential to the neuromuscular synapse development (Guarino et al. 2020; Lazaridis and Tzartos 2020; Shen et al. 2015). Dag1 is the gene that encodes both extracellular and transmembrane subunits of dystroglycan present in a variety of tissues such as cardiac muscle, nervous system, lung, kidney, and bone that interacts with several ECM proteins, including Agrin (Dempsey et al. 2019). Lrp4 and Dag1 were expressed in all three culture models during osteoblast differentiation and in general, the expression of both increased over time as of Agrin and Bglap. However, the gene expression of Musk was not detected in any culture or time point evaluated, suggesting that Agrin-Musk binding is not relevant to osteoblast differentiation as it is in neuromuscular synapse (Lazaridis and Tzartos 2020). Despite a co-expression assay was not done to evaluate the interaction between Agrin and its receptors, it has been reported that Agrin directly binds to the Lrp4 (Kim et al. 2008). Taken together, these findings suggest that the effect of Agrin on osteoblasts could be, at least in part, mediated by Agrin-Lrp4-Dag1 binding.

To determine the influence of Agrin on osteoblast differentiation, Agrin was knocked down in MC3T3-E1cells by transfection with siRNA, and recombinant Agrin was used to treat MSC-OB. We have demonstrated that Agrin knockdown not only reduced the expression of Agrin at the mRNA and protein levels, but also downregulated the expression of its receptors Lrp4 and Dag1, corroborating that Agrin contributes to osteoblast differentiation through Lrp4-Dag1 activation. In the skeletal system, it was demonstrated that the inhibition of either Lrp4 or Dag1 decreases Agrin-induced transcription factor SOX9 expression, indicating that both receptors are important not only for osteoblast differentiation as showed here, but also for chondrocyte differentiation (Eldridge et al. 2016). However, interestingly, the addition of recombinant Agrin in the cell culture downregulated the gene expression of Agrin and upregulated only the receptor Lrp4. There are different types of protein-mRNA correlations, and the amount of mRNA itself is not sufficient to predict the level of protein expression and vice versa (Vogel and Marcotte 2012; Brockmann et al. 2007). Therefore, the lack of correlation between the recombinant Agrin and the gene of expression of Agrin seems to be related to the transcription limited by the abundant amount of protein added, possibly causing saturation of both transcription and translation (Liu et al. 2016).

Amongst several transcription factors, Runx2 and osterix play essential roles in the regulation of osteoblast differentiation and consequently in the bone formation (Almalki and Agrawal 2016; Komori 2006; Nakashima et al. 2002). In this study, Agrin knockdown significantly impaired the progression of osteoblast differentiation as evidenced by the downregulation of these transcription factors as well as of several other genes such as Alpl, Ibsp, Bglap, and Col1a1. The disruption of osteoblast differentiation by Agrin knockdown was confirmed at the phenotype level, by the reduction of both RUNX2 protein expression and ALP activity. Indeed, some in vivo studies have also shown that the absence of Agrin negatively impacts the function of some tissues (Bassat et al. 2017; Hausser et al. 2007). In mice, the lack of Agrin in the cardiac mesoderm impairs heart function and cardiac regeneration, whilst the absence of Agrin outside the NMJ delays mice skeletal development and growth (Bassat et al. 2017; Hausser et al. 2007). In contrast to the findings described here, it was reported that Agrin deficiency does not affect osteoblast differentiation (Mazzon et al. 2011). A possible explanation for these divergent results could be the different approaches of the two studies. Here, siRNA was used to knockdown Agrin in a well-established osteoblast cell line whilst Mazzon et al. (2011) worked with cells from transgenic mice with global knockout of Agrin but carrying a transgene that increases threefold the expression of Musk to avoid mice died at birth. On the other hand, the treatment with recombinant Agrin slightly increased the osteoblast differentiation as confirmed at both genotype and phenotype levels, by upregulation of gene expression of most of the evaluated bone markers, such as Runx2, Sp7, Ibsp, and Bglap, as well as RUNX2 protein expression and ALP activity. The positive effects of recombinant Agrin on cell differentiation were also observed in myoblasts, which enhanced myogenesis of C2C12 cell line (Ebrahimi et al. 2018).

In this study, the possible mechanisms involved in the Agrin-mediated osteoblast differentiation were also explored. Since Lrp4 acts as receptor for both Agrin and Wnt and considering the crosstalk between Wnt and BMP signalling pathways (Zhang et al. 2013; Itasaki and Hoppler 2010), these pathways were investigated. The Wnt signalling participates in bone formation during embryonic development, and in the course of the life continues to play an important role in bone homeostasis, repair, and regeneration, through the regulation of cell proliferation, differentiation, maintenance, and apoptosis (Houschyar et al. 2019). The BMP signalling pathway plays a crucial role from embryogenesis to homeostasis in the adult phase in different systems such as the cardiovascular and gastrointestinal system; however, this class of proteins was named based on its main function in osteogenesis, exerting several functions, including osteoblast differentiation (Rahman et al. 2015; Dimitriou and Giannoudis 2005). These data suggest that Agrin has an important contribution to osteoblast differentiation, at least in part, by modulating Wnt and BMP signalling pathways, since Agrin knockdown downregulated the expression of several genes related to both pathways. It has been shown that Agrin treatment promotes cartilage repair by suppressing Wnt signalling in a process that involves the activation of the transcription factor cyclic adenosine 3′,5′-monophosphate response element-binding protein (Eldridge et al. 2020). In contrast, it was observed that the treatment with recombinant Agrin induced a slight increase of osteoblast differentiation without affecting Wnt and BMP signalling pathways. These findings together with the unchanged gene expression of Dag1 suggest that endogenous Agrin is more relevant than the exogenous recombinant protein in terms of promoting osteoblast differentiation.

In conclusion, to the best of our knowledge, this study is the first to show the key contribution of Agrin to osteoblast differentiation. The time course expression of Agrin exhibited a positive correlation with the progression of osteoblast differentiation and Agrin knockdown negatively impacted osteoblast differentiation, whilst the treatment with recombinant Agrin slightly increases this process. Also, an Agrin-Wnt-BMP circuit could be, at least in part, involved in the cellular mechanism of the Agrin-mediated osteoblast differentiation, mainly the endogenous Agrin. Although additional studies are needed to confirm that the activation of this circuit is triggered by the direct interaction between Agrin and Lpr4, our findings may drive further research focused on Agrin as a target for developing new therapeutic strategies to treat bone-related diseases and injuries.