Enamel Matrix Derivative and TGF-Beta 1 Target Genes in Human Tongue Carcinoma Cells

Enamel matrix derivative (EMD) can enhance proliferation and migration of different oral cell lines, including malignant oral carcinoma cells, in vitro and in vivo. The composition of EMD is not known, but part of the effects have been postulated to be caused by transforming growth factor-beta-1 (TGF-beta 1). This study aimed to compare target genes of EMD and TGF-beta 1 on highly malignant oral carcinoma HSC-3 cells. Microarrays were used to examine differentially expressed genes in HSC-3 cells after 6h and 24h incubations with EMD (200 μg/ml) or TGF-beta 1 (10 ng/ml). Gene Ontology (GO) enrichment analysis of the regulated genes was also conducted. After 6h and 24h of EMD treatments 42 and 12 genes, respectively, were statistically significantly (P<0.05) upor down-regulated. However, as many as 393 and 346 genes were statistically significantly (P<0.05) upor down-regulated by TGF-beta 1. Among the most up-regulated genes by both of the study reagents were MMP-9 and -10. The expression of MMP-10 by EMD treated carcinoma cells was also verified in protein level. In conclusion, TGF-beta 1 regulates more and mostly different genes compared with EMD, but both regulate the expression of matrix metalloproteinase genes in oral carcinoma cells. © 2020 Matti Mauramo. Hosting by Science Repository. All rights reserved.

The predominant compound of EMD is amelogenin (>90%) [11,12]. However, several studies have shown EMD to be more effective than amelogenin in enhancing regeneration related factors [13][14][15][16][17]. Therefore, it is generally assumed that EMD contains other biologically active factors in addition to enamel proteins [11][12][13]. Results concerning cytokines and growth factors present in EMD or stimulated by EMD have been inconsistent, but a body of evidence suggests that at least some effects of EMD on gene expression require TGF-beta activity [18][19][20][21][22]. TGF-beta, and in particular TGF-beta 1, is a key regulator of epithelial homeostasis. TGF-beta 1 also acts as one of the major cytokines in carcinogenesis by stimulating angiogenesis, invasion and metastasis in oral carcinomas [23][24][25][26]. Recently, TGF-beta receptor inhibitors such as galunisertib have been developed with promising results in treating patients with advanced hepatocellular carcinoma [27]. A very recent review concluded that squamous cell carcinomas are an optimal cancer type to study the effectiveness of TGFβ inhibition due to the high prevalence of dysregulated TGF-beta signaling [28].
Previous studies have shown that TGF-beta 1 can directly induce expression of matrix metalloproteinases (MMPs) MMP-2 and MMP-9 in pre-and malignant epithelial cell lines, also of oral origin, as well as in endothelial cells [29][30][31][32]. Furthermore, it has been shown that TGFbeta 1 stimulates HSC-4 oral squamous cell carcinoma invasion through MMP-10 signaling [33]. In carcinomas, MMPs are important because of their ability to degrade extracellular matrix and basal membrane components and overexpression of MMPs, particularly MMP-2 and -9, are associated with poor prognosis in oral cancer [34][35][36]. We have previously shown EMD to be capable of enhancing MMP-2 and -9 release by HSC-3 carcinoma cells and to stimulate their migration. Furthermore, EMD promoted metastasis formation in vivo [8]. This effect of EMD on MMPs might be due to TGF-beta 1 activity in EMD.
As expected, most in vitro studies on EMD have focused on periodontal fibroblasts or bone cells. In contrast, the effects of EMD on other cell lines, particularly on malignant cells, have not been adequately studied. As EMD is widely used in regenerative dentistry, also patients with premalignant or malignant oral lesions may become subject to EMD treatments. In this study, which is also a part of an academic dissertation of MM, we aimed to find out and compare the target genes of EMD and TGF-beta 1 in HSC-3 oral carcinoma cells using Affymetrix microarrays [37]. Furthermore, the presence of MMP-10 in HSC-3 culture media was analysed by western blotting. Our hypothesis was that EMD and TGFbeta 1 have rather similar effects on the gene expression of HSC-3 oral carcinoma cells.

II Cell Cultures
The experiments were carried out using highly malignant HSC-3 tongue squamous cell carcinoma cells (tongue, Japan Health Science Resources Bank, JRCB 0623), cultured at 37°C in a humidified atmosphere and in their normal media as described earlier [8].

III Microarray Sample Preparation and Hybridization
1x106 HSC-3 cells/flask were let to grow overnight. The culture media was changed, and TGF-beta 1 (10 ng/ml) or EMD (200 µg/ml) was added, and untreated cells acted as a control. TGF-beta 1 and EMD concentrations used in were selected based on previous studies [8,38]. The study cells were incubated for 6 h and 24 h and control cells for 24 h. Total cellular RNA was extracted using Qiagen RNeasy Mini Kit according to the manufacturer's instructions (Qiagen, Oslo, Norway). RNA quality and quantity were determined using the Agilent 2100 Bioanalyser (Agilent Technologies, DE, USA). Microarray experiments were carried out using the Affymetrix Human Genome U133 Plus 2.0 chip for analysis of over 47,000 transcripts. Total RNA (8 µg) from each sample was used for target cDNA synthesis according to the Affymetrix protocol (Link 1). Following the recommendations by Lee and Whitmore, three biological replicates were produced at each time point [39].

IV Data Deposition
The microarray data is publicly available at ArrayExpress (E-MEXP-2645).

V Microarray Data Analysis
The microarray data analysis was performed using open-source Chipster software (Link 2). For normalization, the Robust Multi-Array method was used. Empirical Bayes t-test was used to estimate the statistical significance of individual genes with a cut-off threshold of P<0.05. The fold change (FC) of the statistically significantly regulated genes is given in (Table 1 or in Supplement material S1 & S2).
For the differentially expressed genes, Gene Ontology (GO) enrichment analysis was performed using DAVID annotation tool (Link 3) according to Huang et al. [40]. The analysis was conducted using the list of all statistically significantly up-and down-regulated genes separately. The whole-genome list of the Human Genome U133 Plus 2.0 chip was used as a background. An EASE score (E-score), a modified Fisher Exact P-Value, was used to identify enriched categories. E-score <0.05 was considered as statistically significant. Furthermore, the enriched terms were grouped into functional clusters according to terms having similar biological meanings due to sharing similar gene members.

I 6 h EMD Treatment
A total of 29 genes were statistically significantly (P<0.05) up-regulated and 13 genes down-regulated in HSC-3 carcinoma cells after 6 h of EMD treatment compared to untreated control cells [(List of genes and fold changes (FC) in (Table 1)]. MMP-10 and PI3 were among the upregulated genes. Claudin 1, transforming growth factor beta 2, epithelial membrane protein 1 (EMP1), and endothelin 1 were among the most down-regulated genes.

II 24 h EMD Treatment
After 24 h of EMD incubation, the regulated genes (P<0.05) were highly attenuated, showing only four up-regulated and eight down-regulated genes ( Table 1). MMP-9 was the most significantly up-regulated, whereas surprisingly, MMP-10 was now the most down-regulated gene. Small proline-rich protein, annexin A10, TGF-beta-induced factor homeobox 1, neuregulin 1, and T-box 1 were down-regulated.  Table 1: Differentially expressed genes in HSC-3 cells, 6 h (A) and 24 h (B) after EMD administration. The two columns on the right present the genes, which were also regulated by TGF-1 after 6 h (C) and 24 h (D).

IV 24 h TGF-Beta 1 Treatment
A total of 242 genes were statistically significantly (P<0.05) upregulated and 104 genes down-regulated in HSC-3 carcinoma cells after 24 h of TGF-beta 1 treatment compared to untreated control cells (Supplement material, S2). 178 (74%) of the genes up-regulated by 6 h of TGF-beta 1 treatment were also regulated after 24 h. Of the downregulated genes, 67 (44%) were down-regulated also after 24 h of treatment. No genes showed inverted regulation during the observation time.

V GO Enrichment Analysis
GO enrichment analysis revealed several enriched terms after 6 h EMD treatment (Table 2). Furthermore, the cluster analysis revealed keratinocyte differentiation, inflammatory response and proteolysis to be the most up-regulated. Among the down-regulated genes, the most enriched clusters were cell growth and morphogenesis, circulatory system and cellular developmental processes. Because of the relatively few regulated genes by EMD after 24 h, the GO enrichment analysis was not performed. After 6 h TGF-beta 1 treatment, GO analysis revealed several enriched terms (Table 2), which were grouped into functional clusters. The most enriched clusters among up-regulated genes were system development, chemokine activity/inflammatory response, hemopoiesis and angiogenesis. Those among down-regulated clusters were cell cycle processes, mitosis/ M-phase and cell growth. After 24 h TGF-beta 1 treatment, the most enriched clusters among up-regulated genes were developmental processes, inflammatory/defense response and angiogenesis. Down-regulated clusters were regulation of cellular processes, intracellular organelle and calcium homeostasis.

VI Comparison of the Regulated Genes by TGF-Beta 1 and EMD Treatment
TGF-beta 1 regulated markedly more genes than EMD after 6 h and 24 h of treatments. About half (55%) of the genes (n=16) up-regulated by 6 h of EMD treatment were the same as those up-regulated by the 6 h of TGF-beta 1 treatment. Five (31%) of the genes down-regulated by EMD were the same as those by TGF-beta 1 ( Table 1). Of the 15 most enriched GO terms after 6 h of incubation, EMD and TGF-beta 1 both upregulates five identical GO groups ( Table 2). Of the four up-and eight down-regulated genes by 24 h of EMD treatment, two up-and two downregulated genes were the same as with the TGF-beta 1 treatment (Table  1).

VII Western Blotting
We have previously shown that EMD can enhance the production of MMP-9 protein by HSC-3 cells analysed with ELISA immunosorbent assay and gelatin zymography [8]. In the present study, we wanted to confirm the presence of MMP-10 protein in HSC-3 culture media after 24 h of EMD treatment. Western blot confirms the result obtained with microarray. Untreated HSC-3 cells did not produce MMP-10, but the production was clearly induced by both EMD concentrations (100 and 200 µg/ml) ( Figure 1).

Discussion
In the present study, we used microarrays to compare the effects of EMD and TGF-beta 1 on HSC-3 tongue carcinoma cell gene expression. HSC-3 carcinoma cells were highly more sensitive to TGF-beta 1 than to EMD. After 6 h of TGF-beta 1 treatment, a total of 393 genes were statistically significantly regulated, whereas 346 genes were regulated after 24 h of TGF-beta 1 incubation. However, EMD regulated only 48 genes after 6 h of treatment, and the effect was further attenuated after 24 h of incubation. In previous studies, EMD has regulated the expression of 202 genes in periodontal ligament fibroblasts and 107 genes in MG-63 osteosarcoma cells [5,6]. Although there are some differences between the methods, the results confirm that the regulating effects of EMD on gene expression depend highly on the cell type investigated.
Several studies have suggested EMD to contain TGF-beta 1 as the principal bioactive factor, but the results are conflicting [2,17,18,20,42,43]. Thus, the synergistic effect of various proteins, including amelogenin and unidentified mitogenic factors, is commonly proposed [18,20]. In the present study, the number of regulated genes differed substantially according to study reagents. This indicates that EMD has only limited TGF-beta 1 activity on oral carcinoma cells. Nonetheless, EMD down-regulated the gene expression of TGF-beta 2 and TGF-betainduced factor homeobox 1, as did also the recombinant TGF-beta 1.
Curiously, in this assay, EMD and TGF-beta 1 did not regulate the expression of TGF-beta 1. However, in mammals, the TGF-beta isoforms (TGF-beta 1, TGF-beta 2 and TGF-beta 3] show 70-80% sequence identity, bind to the same cell surface receptors and elicit similar responses in vitro. Furthermore, it has been shown that elevated expression of either TGF-beta 1 or TGF-beta 2 results in more aggressive phenotype in human squamous cell carcinoma cells as well as in stromal fibroblasts of oral carcinoma [44,45]. Consequently, it can be expected that TGF-beta isoforms have an overlapping regulation of their expression. In keeping with this hypothesis, it has been shown that both TGF-beta 1 as well as TGF-beta 2 regulate the expression of TGF-beta 1 gene in trabecular cells of porcine eyes [46]. To our knowledge, this is the first study to suggest that TGB-beta 1 can have effects on the expression of TGF-beta 2 in oral cancer cells. Furthermore, the result may suggest that the down-regulation of TGF-beta family members caused by EMD could be due to the slight amount presence of TGF-beta 1 in EMD.
Both TGF-beta 1 and EMD regulated the expression of MMPs. MMPs are centrally involved in malignant diseases by degrading basement membrane and extracellular matrix components [34]. In particular, MMP-2, -9 and -10 are up-regulated in oral carcinomas and correlates with the invasive potential and poor prognosis of HNSCCs [33,35,36,[47][48][49]. TGF-beta 1 has been shown to induce the expression of several MMPs, particularly MMP-9, in HNSCC cell lines, thereby favouring HNSCC invasion [30,31,50]. In this study, the MMP-9 was the most significantly up-regulated gene after 24 h of EMD treatment as well as after 6 and 24 h of TGF-beta 1 treatments. The results are also consistent with our previous studies, showing that the production of MMP-9 from cultured HSC-3 cells is significantly induced by EMD [8]. In addition, in this study, both EMD and TGF-beta 1 had significant effects on calcium homeostasis related GO terms. EMD up-regulated GO groups related to calcium and calcium-binding after 6 h of treatment, whereas TGF-beta 1 was observed to down-regulate calcium ion homeostasis after 24 h of incubation. Calcium has been observed to increase MMP-9 expression in normal and premalignant oral keratinocytes and to act via ERK1/2 and MAPK signaling pathways [51]. Furthermore, it has been suggested that failures in these calcium-induced signaling pathways accompany the malignant transformation of the oral epithelium [51]. Our results are in keeping with these previous observations, and both EMD and TGF-beta 1 regulated MMP-9 and calcium homeostasis simultaneously, and the regulations of these factors are most likely related.
In addition to MMP-9, TGF-beta 1 can induce the expression of MMP-1 and MMP-10 by carcinoma cells [29,49,52,53]. In line with the previous studies, MMP-1 was statistically significantly up-regulated in HSC-3 oral carcinoma cells after 6 and 24 h of TGF-beta 1 treatments. However, MMP-10, which is widely expressed in oral carcinomas but rarely in adjacent oral tissues, acted highly interestingly [54,55]. MMP-10 was first among the most up-regulated genes after 6 h of TGF-beta 1 treatment but then absent after 24 h. Moreover, MMP-10 was the most up-regulated gene after 6 h of EMD treatment but was then surprisingly the most down-regulated gene after 24 h. MMP-10 was the only gene in this study possessing this kind of a drastic shift in regulation during 6 to 24 h of observation. Thus, the up-and down-regulation of MMP-10 by HSC-3 cells have to be exceptionally responsive to the study reagents and incubation time. A previous study has reported TGF-beta 1 to upregulate MMP-10 production in HSC-4 but not in HSC-3 cells after a minimum of 48h incubation [33]. That study also found TGF-beta 1 to promote invasion through MMP-10 signaling [33]. In this study, untreated HSC-3 cells did not produce MMP-10 but EMD induced the production of MMP-10, which was detected in the culture media after 24 hours of incubation. Taken together, the activations of MMPs by TGF-beta 1, EMD or other MMPs seem to be interrelated and acts in complex cascades [30,31,56]. Also, other studies have suggested that MMPs [-2 and -9] acts co-operatively and are induced by MMP-10 [33]. Thus, further studies are needed to analyse the down-stream target genes and activation cascades of MMPs in oral carcinoma cells.
To further analyse the biological effects of the genes regulated by EMD and TGF-beta 1, a GO enrichment analysis was performed. Furthermore, the enriched terms were grouped into functional clusters according to terms having a similar biological meaning. However, these analyses have some limitations. A gene can be related to several different GO groups, and an extensive list of GO terms can be produced with only a small number of genes. Therefore, we did not conduct the GO enrichment analysis from the few genes regulated by EMD after 24 h. The GO analysis revealed that the most relevant up-regulated GO clusters by 6 h of EMD treatment were inflammatory response and proteolysis. In addition, the most important GO clusters among EMD down-regulated genes were cell growth and morphogenesis. These were almost similar for TGF-beta 1.
After 6 h of TGF-beta 1 treatment, the up-regulated genes in HSC-3 cells were related to inflammatory response and angiogenesis. However, TGF-beta 1 also had a pronounced effect on the growth and proliferation of HSC-3 cells, as the three most down-regulated clusters were all related to these terms. After 24 h of TGF-beta 1 incubation, angiogenesis and inflammatory/defense were still among the three most significantly upregulated clusters. However, cell growth and mitosis were not anymore among the down-regulated clusters. Thus, this result suggests that TGFbeta 1 can suppress carcinoma cell growth and mitosis at first, but then its carcinogenesis promoting effects, for example induction of angiogenesis, become more expressed and may exceed former anti-cancer activity. This observation is in keeping with, and further explains, the previous results showing that TGF-beta 1 can first inhibit and at a later stage, promote carcinogenesis [25,26].
This study was the first microarray to compare the effects of EMD and TGF-beta 1 on invasive oral carcinoma cells. TGF-beta 1 was observed to be more effective and shows a different gene expression pattern compared with EMD. However, both EMD and TGF-beta 1 statistically significantly induced the expression of MMP-9, but their effects on MMP-1 and -10 varied.