INTRODUCTION

Collagens are the most abundant proteins of the extracellular matrix. They are essential for cell attachment, migration, proliferation, and differentiation in connective tissue [1, 2]. Type I, II, and III collagens are fibril forming. Type I collagen is the principal collagen of bone, tendons, skin, ligaments, cornea, and many interstitial connective tissues. The type II collagen is the characteristic component of hyaline cartilage. Type III collagen is widely distributed in collagen I containing tissues except for the bone. It is an important component of reticular fibers in the interstitial tissue of the lungs, liver, dermis, spleen, and vessels. It forms mixed fibrils with type I collagen and also localizes in elastic tissues [3–7]. Because in various pathological situations collagen gene expression patterns may change, it is important to understand how these genes are controlled by cell type-speific regulatory factors. For instance, Wu and colleagues detected an increase in Col1A2 gene expression in osteosarcoma and proposed this molecule as a diagnostic marker. Collagenolytic MMPs have critical functions for maintaining invasive features of cancers, including osteosarcoma [8–18]. Some transcription factors are also capable of affecting osteoblast maturation by interacting both collagen and tumor suppressor genes which is related to the molecular pathogenesis of osteosarcoma. The studies of the effects of Upstream Stimulatory Factor 1 (USF1) on collagen gene expression are limited. Rippe and colleagues determined that binding of USF to an E-box in the 3' flanking region stimulates type I collagen gene transcription [19].

In the present study, we focused on the regulation of collagen genes (type I, II, and III) and a collagen processing matrix metalloproteinase ADAMTS-3 by USF1 transcription factor in osteosarcoma models. USF1 interacts with E-boxes and plays an essential role during embryonic development and also involves in the regulation of chondrocyte differentiation [20, 21]. In Saos-2 and MG-63 cells, we elucidated the supressive effect of USF1 protein on the expression patterns of collagen genes (type I, II, and III). In addition, we studied USF1-dependent regulation of ADAMTS-3. As a member of the pNp family, ADAMTS-3 primarily responsible for amino-terminal cleavage of fibrillar type II procollagen. Besides the type II procollagen, ADAMTS-3 also processes N-termini of the type I procollagen molecules. In our previous studies, we have characterized the human ADAMTS-3 promoter and identified SP1-dependent transcriptional regulation of its gene [22]. As the ADAMTS-3 promoter contains potential E-boxes, we further aimed to elucidate the regulatory effect of the USF1 on ADAMTS-3 promoter activity. Luciferase experiments show that USF1 suppresses the ADAMTS-3 promoter activity, resulting in a decrease in the levels of both ADAMTS-3 mRNA and respective protein. The functional binding of USF1 to ADAMTS-3 promoter was shown in wet-lab EMSA experiments. Together, presented data indicate that the transcriptional factor USF1 influences ADAMTS-3 expression and, therefore, contrubutes to deregulation of collagens in osteosarcoma.

EXPERIMENTAL

Cell culture and transient transfection assay. Saos-2 and MG-63 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) containing 10% fetal calf serum (FCS, Gibco) and 2 mM of L-Glutamine (Gibco). All cells were maintained at 37°C in a 5% CO2 and 95% air in a humidified incubator. Truncated ADAMTS-3 promoter-reporter constructs namely; pMET_TS-3[–1340/+40], pMET_TS-3[–879/+40], pMET_TS-3 [–576/+40] and pMET_TS-3 [–131/+40] were prepared as described previously [22]. The USF1expression plasmid was a gift from Dr. Dipak P. Ramji, Cardiff School of Biosciences. The calcium phosphate precipitation method was used for the transient transfection of ADAMTS-3 promoter constructs as described before [23]. For co-transfection assays. USF1 expression plasmid was transiently-transfected with ADAMTS-3 promoter-reporter constructs (1 µg). Secreted human alkaline phosphatase (SEAP, Promega) plasmid was also transfected (0.5 µg) into cells for normalization of the transfection efficiency. Luciferase and SEAP activities were measured from collected media after 48 h and 72 h of transfection using Ready-To-Glow™ Secreted Luciferase Reporter Systems (Clontech) and Fluoroskan Ascent FL Luminometer following the instructions. For normalization, the promoter activities of various deletion fragments were represented as the ratio of firefly luciferase activity to SEAP activity. pMetLuc control and pMetLuc reporter vectors (Clontech) were transfected into different wells for each experimental set up as positive and negative control. Transfection experiments were repeated at least three times [23].

RNA isolation and quantitative RT-PCR (qRT-PCR). RNeasy Kit (Thermo Scientific) was used to extract total RNA from cell pellets which were transiently-transfected with/without USF1 expression plasmid following instructions. An equal amount of total RNA (1 µg) was used for cDNA synthesis from control and overexpressed groups. cDNA synthesis was performed as described before [24]. Specifically, 1 µL of cDNA was added into 5 µL of Light Cycler-FastStart DNA Master SYBR Green I mix (Roche) as a template and 0.5 µL of each pair of primers (50 ng/µL) as shown in Table 1, at 10 µL final volume. q RT-PCR was performed with Light Cycler 485 instrument (Roche Diagnostics) under the following conditions; Initial denaturation of 95°C for 10 min, followed by 35 cycles of 95°C for 15 s and 58°C for 15 s 72°C for 10 s and final extension of 72°C for 1 min. Each sample was studied triplicate, and the Ct value was defined automatically by the instrument. The relative changes in gene expression between control and USF overexpressed groups were calculated according to 2(–Delta Delta C(T)) method using the mean of human β-2 microglobulin (hβ-2) and ribosomal protein L13 (RPL 13A) as internal controls [25].

Table 1. Primer sequences used in qRT-PCR and EMSA experiments
Full size table

Protein extraction and western blotting. Protein extraction with RIPA buffer and western blotting from control and USF1 overexpressing cells were performed according to previously described methods [26]. A fluorimeter (Qubit) was used to determine protein concentration. 50 µg of protein was loaded on SDS-polyacrylamide gels. Blots were incubated with primary antibodies namely; polyclonal ADAMTS3 (3 µg/mL) (Abcam, ab45037), or polyclonal type II collagen (Col2A1) (2.5 µg/mL) (Santa Cruz Biotech sc7764) at 4°C for overnight, or monoclonal β-actin (Santa Cruz Biotech., sc81178) antibody at room temperature for 1 h. Blots were washed and incubated with horseradish peroxidase-conjugated secondary antibodies at room temperature for 1 h. Protein bands were visualized with ECL (Thermo Scientific) and photographed by using Fusion FX Vilber Lourmat. Protein quantification analysis was performed using ImageJ software [27].

Electrophoretic mobility shift assays. Synthetic oligonucleotide probes specific to ADAMTS-3 promoter regions, [–641/–647] and [–973/–937], were biotinylated from 3' ends with Biotin-11-UTP and Terminal Deoxynucleotidyl Transferase (TdT) using Biotin 3' End DNA Labeling Kit (Thermo Scientific) and then annealed with their complementary strands at 95°C for 5 min in dH2O. EMSAs were carried out using a LightShift Chemiluminescent EMSA Kit (Thermo Scientific) according to instructions. Briefly, 4 μg of nuclear extract, 10% binding buffer (340 mM KCl, 50 mM MgCl2, 1 mM dithiothreitol, 0.1 mM EDTA, 40 mM KCl), and 0.05 µg/mL PolydIdc were added to reaction at a 20 μL final volume and incubated on ice for 10 min. Subsequently, 20 pmol of biotinylated double-stranded oligonucleotides were added into the reaction mixture and incubated for 30 min at room temperature. Competitive EMSAs were performed under the same conditions by adding the 10.000 fold amount of the same un-labeled double-stranded oligo or unlabeled consensus USF1 oligonucleotides to examine the functional binding of USF1 (Table 1). Whole cell extracts were prepared from control and USF overexpressing Saos-2 cells as described previously [24]. Electrophoresis was performed in 6% native polyacrylamide gel. Semi-dry transfer system was used to transfer complexes to the nylon membrane. The membrane was subjected to UV for 15 min for cross-linking. Biotin signals were detected using a Chemiluminescent Nucleic Acid Detection Module (Thermofisher Scientific) according to the manufacturer’s instructions.

In silico sequence analysis. Putative transcription factor-binding sites in the promoter region of the ADAMTS3 gene were predicted using MatInspector (Genomatix Software) with a threshold of 0.9 [28, 29].

Statistical analysis. Standard deviations p-values were calculated by using Mini Tab 14 software. One way ANOVA analysis was applied between pairs for statistical significance.

RESULTS

USF1 Decreases Human Collagen Gene Expressions (Type I, II, and III) in Osteosarcoma Models

Because of the critical role of the ADAMTS-3 on procollagen amino-terminal processing, mainly type II collagen, firstly we determined endogenous type I, II, and III collagen expression levels in osteoblastic cell lineages, MG-63 and Saos-2. Steady-state collagen mRNA levels determined by qRT-PCR using specific primer pairs (Table 1). Expression analysis revealed that type I collagen was the most abundant collagen type in both MG-63 and Saos-2 cells. Subsequently, type III and type II were present in these cells (Figs. 1a, 1b). These cells also differ in terms of the endogenous USF expression level (Fig. 1c). To evaluate the effect of the USF1 protein on the expression levels of these collagen types, hUSF1 expression plasmid was transiently-transfected into MG-63 and Saos-2 cells for ectopic expression. USF1 overexpression was confirmed by qRT-PCR using specific primers to USF1 (Table 1) (Figs. 1d, 1g). USF1 overexpression downregulated all collagen types at the mRNA level in MG-63 cells (Fig. 1e). A maximum decrease was observed in type II collagen (to 0.02 fold) subsequently type III (0.2 fold) and type I collagen (0.6 fold) mRNA levels (Fig. 1e). A similar decreasing effect was also observed in the type II collagen protein level in western blot assay (Fig. 1f). The effect of the USF1 on these collagen types was also investigated on another osteoblastic cell model, Saos-2 to better understand the regulation of collagen expression with USF1 in osteosarcoma. Overexpression of USF1 protein leads to a similar decrease for all collagen types at mRNA level with a statistically important manner, in type I, II, and III collagen mRNA with 0.3, 0.4, and 0.3 fold respectively. The decrease in type I and III collagen mRNA levels in Saos-2 was more dominant than MG-63 cells (Fig. 1h).

Fig. 1.
figure 1

(a, b) Steady-state mRNA levels of type I, II and III collagens in MG-63 and Saos-2 cells. (c) Endogenous USF1 mRNA levels in Saos-2 and MG-63 cells. (d) Confirmation of USF ectopic expression at mRNA level in MG-63 cells after 48 h of transient transfection. (e) Type I, II, and III collagen mRNA levels in USF overexpressing MG-63 cells. (f) Type II collagen protein level in USF overexpressing MG-63 cells. (g) Confirmation of USF ectopic expression at mRNA level in Saos-2 cells after 48 h of transient transfection. (h) Type I, II, and III collagen mRNA levels in USF overexpressing Saos-2 cells. The asterisk indicates the significant levels of treated groups (*p ≤ 0.05).

Full size image

USF1 Downregulates ADAMTS-3 Gene Expression in Two Different Osteosarcoma Model

Next, to assess whether USF1 protein regulates ADAMTS-3 gene expression, ADAMTS-3 mRNA levels were measured from cell extracts USF1 overexpressing and control groups by qRT-PCR in MG-63 cells. USF1 overexpression led to a significant decrease in the ADAMTS-3 mRNA level to 0.4 fold when compared to the control group (Fig. 2a). Although USF1 significantly repressed endogenous ADAMTS-3 mRNA expression, it is not necessary that USF1 can decrease its protein level. Therefore, the ADAMTS-3 protein level was also detected by western blotting from USF1 overexpressing or control groups. Similar to mRNA data, USF1 decreased the ADAMTS-3 protein level to 0.56 fold (p ≤ 0.005, Fig. 2b).

Fig. 2.
figure 2

(a, b) ADAMTS-3 mRNA, and protein levels after USF1 overexpression in MG-63 cells. (c, d) D ADAMTS-3 mRNA and protein levels after USF1 overexpression in Saos-2 cells. The asterisk indicates the significant levels of treated groups (*p ≤ 0.05).

Full size image

USF1 protein had a more pronounced effect in another osteosarcoma model, Saos-2, on the repression of the human ADAMTS-3 mRNA and protein levels. Results revealed that USF1 was able to decrease ADAMTS3 mRNA (to 0.24 fold) and protein (to 0.2 fold) levels compared with control groups (Figs. 2c, 2d).

ADAMTS-3 Promoter Activity is Negatively Regulated by USF1 in Saos-2 and MG-63 Cells

USF is a basic-helix-loop-helix family transcription factor that interacts with E-box motifs in the genomic sequences [30]. Sequence analysis indicated potential E-box motifs between –1315/–1331, –1051/–1068, ‒933/–950, and –614/–642 in the ADAMTS-3 promoter region. To examine whether USF1 has any regulatory effect on the ADAMTS-3 promoter activity, the USF1 expression plasmid was co-transfected with four different truncated ADAMTS-3 promoter fragments that were previously constructed, namely pMET_TS3 [–131/+40], pMET_TS3 [–576/+40], pMET_TS3 [–879/+40] and pMET_TS3 [–1340/+40] and dual-luciferase reporter assays were performed [24]. As shown in Fig. 3a, overexpression of the USF1 slightly decreased pMET_TS3 [–131/+40] and pMET_TS3 [–576/+40] relative luciferase activity, which does not contain a USF1 binding motif. The decrease in the relative luciferase activity of two longer promoter constructs pMET_TS3 [–879/+40] and pMET_TS3 [–1340/+40] having E-box motifs were statistically important containing E-box motifs within the region, in MG-63 cells. Because the USF1 had cell-type specific activity its effect on the ADAMTS-3 promoter activity was also determined in another osteosarcoma cell model, Saos-2 [31]. Inconsistent with MG-63 data USF1 repressed ADAMTS-3 promoter activity for all constructs. Repression was statistically important for pMET_TS3 [–1340/+40] promoter construct.

Fig. 3.
figure 3

(a) Relative Luciferase activities of ADAMTS-3 promoter reporter constructs in MG-63 and Saos-2 cells under USF1 overexpression. Schematic representation of the reporter constructs used in the transient transfection experiments was represented on the left side. E-box motifs were shown with a rectangle. (b–d) In vitro binding analysis of USF1 protein to ADAMTS-3 promoter by EMSA. The asterisk indicates the significant levels of treated groups (*p ≤ 0.05).

Full size image

USF1 Functionally Binds E-Box Motifs in the ADAMTS-3 Promoter Region

To characterize the presence of the interactions between USF1 protein and ADAMTS-3 promoter, EMSAs were performed using nuclear extracts prepared from Saos-2 cells and biotin-labeled DNA oligonucleotide probes namely, USF1 Probe (consensus), [–131/–103] (probe 1), [–641/–607] (probe 2) and [–973/–937] (probe 3). When the biotinylated USF1 consensus probe was incubated with Saos-2 nuclear extract, a single complex was detected in the gel (Fig. 3b, lane 2). The sequence specificity of the DNA–protein complex was checked by the addition of the unlabeled USF1 probe in binding reactions. Unlabeled oligonucleotide diminished the complex formation (data not shown). The formation of the complex was stronger when USF1 overexpressing nuclear extract was incubated with a biotinylated USF1 probe, showing USF1-ADAMTS-3 promoter interaction (Fig. 3b, lane 3). To provide more direct evidence USF1 and E box element interactions, EMSAs were performed using probe [–641/–607] and [–973/–937] including E box motifs with Saos-2 nuclear extracts. A single complex was detected in the gel (Fig. 3d, lanes 5 and 8). The intensity of the complexes was increased with the extracts prepared from USF1 overexpressing cells indicating specific interaction between USF1 protein and ADAMTS-3 promoter (Fig. 3d, lanes 6 and 9).

According to bioinformatic analyses, an E box binding motif couldn’t be detected in [–131/–103] bp region of ADAMTS-3 promoter. However, using [–131/–103] probe with Saos-2 nuclear extract created three different complexes in EMSA experiments (Fig. 3c, lane 2). Complex 1 and complex 2 were abolished when [‒131/–103] unlabeled probe was added into binding reactions indicating specific binding of this probe to the ADAMTS-3 promoter (Fig. 3c, lane 3). Further competition of the USF1 consensus probe with [‒131/–103] probe was also eliminated the formation of complex 1 and complex 2. These findings interestingly revealed the functional binding of USF1 protein to between –131/–103 region of ADAMTS-3 promoter.

DISCUSSIONS

Because of the abundance and critical functions of the collagens, many studies have been performed by different groups to characterize regulatory elements present in these genes to better understand molecular mechanisms controlling collagen gene expressions in normal and pathological circumstances. The regulation of the transcriptional activities of collagens depends largely on the cell type and other regulatory factors. For example, Yasuda and colleagues showed that SOX9 had an enhancer function on type II collagen expression [32]. Otero and colleagues found out that ELF3 modulates type II collagen gene transcription in chondrocytes [33]. Another group determined that YY1 may serve as a positive regulator of transcription of the type I collagen gene [34]. USF1 mediated regulation of the collagen genes (type I, II, and III) was not yet investigated. Here we studied the effect of USF1 on the transcription of type II collagen, the main substrate of the ADAMTS-3, and also type I and III collagen in osteosarcoma models. Besides its primary function of the N-terminal processing of the type II collagen, it has known that ADAMTS3 aid maturation of type I collagens [35]. According to our initial studies of MG-63 and Saos-2 cells, steady-state expression levels of the collagen type I, II, and III encoding mRNAs in these lines differ, with the relative abundances of collagen types displaying the patterns compatible with the data obtained in previous studies [36, 37]. In both types of cells, USF1 overexpression led to a statistically significant decrease in expression levels of mRNA encoding type I, II, III collagens. Overexpression of USF1 also led to a decrease in the type II collagen protein level, thus, endorsing the result of mRNA quantitation collected in MG-63 cells. As compared to Saos-2 cells, in MG-63 cells, the effects of USF1 protein on type II and type III collagen gene mRNA expressions were stronger. On the other hand, in Saos-2 cells, overexpression of USF1 led to a profound decrease of type I collagen mRNA levels.

Secondly, we focused on the USF1 mediated transcriptional regulation of the ADAMTS-3 due to its critical function on N terminal processing of type II and I procollagen molecules and also the involvement of E-box motifs in the ADAMTS-3 promoter region. It has been known that USF1 functions by interacting with E-box motifs in the genomic sequences. So, we thought that USF1 probably has a regulatory effect on ADAMTS-3 transcription. Here we determined that ADAMTS-3 mRNA and protein expression levels decreased to 0.6 fold in USF1 overexpressing MG-63 cells. USF1 also decreased ADAMTS-3 mRNA and protein expressions to 0.2 fold in another osteoblastic model, Saos-2. Further, we showed negative a regulatory effect of USF1 protein on ADAMTS-3 promoter activity with co-transfection studies. USF1 downregulated ADAMTS-3 promoter activity maximum with pMET_TS3 [–1340/+40] and pMET_TS3 [–879/+40] promoter construct including E-box motifs, in Saos-2 cells. Although USF1 is a ubiquitously expressed protein, its expression levels are cell type-specific. Differences in expression levels of USF1, its transcriptional co-factors, or other DNA binding proteins may lead to variations in its transcriptional activity. To account for that, the downregulatory effect of the USF1 protein were studied in another osteosarcoma model, MG-63. Similar to that in Saos-2 cells, USF1 also downregulated ADAMTS-3 promoter activity in MG-63 cells. Finally, functional binding of USF1 protein to ADAMTS-3 promoter was shown in EMSA assays. Besides the consensus USF1 probe, E box motif containing probes covering [–973/–937] and [–641/–607] generated complexes in the gel. In addition, these complexes got stronger when USF1 overexpressed cell extracts were used in the experiments. Although the probe covering [–131/–103] bp doesn’t contain an E‑box motif, EMSA confirmed the functional binding of USF1 to this region, probabaly due to functional interactions between USF1 and the other regulatory elements. In conclusion, here we show that USF1 are capable of transcriptionally regulating the pruduction of type I, II, and III collagens in osteosarcoma. USF1 binds E-box motifs in the promoter of ADAMTS-3 gene and negatively regulates it, thereby, affecting N‑terminal processing of type II and I procollagen molecules region. These findings contribute to understanding of the regulation of collagen and ADAMTS-3 encoding genes in osteosarcoma.