Expression and inactivation of glycogen synthase kinase 3 alpha/ beta and their association with the expression of cyclin D1 and p53 in oral squamous cell carcinoma progression

GSK3 immunoreactivity was observed, and different tumors showed the expression of both of the proteins (GSK3α/β) to different extents. GSK3α/β protein expression was observed in the cytoplasmic, nuclear and both the cytoplasmic and nuclear regions of the cancer cells (Figure 1). In most of the samples, intense overexpression of GSK3β compared to GSK3α was observed. The age group >40 ≤ 70 showed expression and overexpression of GSK3β compared to GSK3α, and this observation was statistically significant (p = 0.03 and p = 0.0006, respectively). Similarly, the age group >70 showed more overexpression of GSK3β compared to the GSK3α isoform (p < 0.001). Males and females showed a greater overexpression of GSK3β compared to GSK3α (p < 0.0001 and p = 0.002, respectively). Both expression and overexpression of GSK3β compared to GSK3α was found independently of nodal invasion (p = 0.04 and p < 0.0001). The smaller sized oral tumors showed more expression (p = 0.006) and overexpression (p < 0.0001) of GSK3β compared to GSK3α. Similarly, GSK3β expression and overexpression was significantly higher than GSK3α expression and overexpression in non-metastatic (p = 0.03 & p < 0.0001) and metastatic oral tumors (p = 0.03 & p < 0.002), respectively. All of these independent observations demonstrate a higher expression level of GSK3β than GSK3α in oral tumor tissue samples (Table 1).

Various types of oral tumors and control samples (normal squamous epithelium, hyperplasia, cancer adjacent oral tissue, non-neoplastic oral cavity glands, adenoid cystic carcinoma, mucoepidermoid carcinoma, adamantinoma, basal cell carcinoma, OSCC, and acinic cell carcinoma) were analyzed to determine the expression of GSK3α/β by IHC (summarized in Table 1). Normal oral mucosa, hyperplasia and various oral cancer samples showed immunoreactivity to both GSK3α/β antibodies to different extents according to the extent of differentiation but some samples did not show any reactivity. OSCC tissue samples of the cheek (Figure 1. A (a, b)), gingiva (g, h), lower mandible (i, j), and lip (w, x) showed a more intense expression of GSK3β than GSK3α. Similarly, different oral SCC (b, h, j, x) samples showed maximum immunoreactivity to GSK3β. The tissue samples of most of the mucoepidermoid carcinoma showed expression of both GSK3α and GSK3β in ductal cells (c, d and q, r and u, v). The mandibular benign adamantinomas showed immunoreactivity to both GSK3α and GSK3β (e, f). In tissue samples of adenoid cystic carcinoma, staining was not observed, either in ductal or in myoepithelial cells (k, l). The tissue samples of basal cell carcinoma showed very faint expression of both GSK3α and GSK3β (m, n), and acinic cell carcinoma showed no expression (o, p) of either GSK3α or GSK3β. In the normal salivary gland, staining was observed in the ductal cells only and more expression of GSK3α than GSK3β (s, t) was observed. Meanwhile, in total 33.75% (27/80) and 51.25% (41/80) of various oral cancer tissue samples did not show GSK3β and GSK3α protein expression (Table 1), respectively. In addition to OSCC, no significant correlation was observed for various other types of oral cavity neoplasms. The OSCC tumors of various sites, such as tongue, lip, cheek and gingival, showed overexpression of GSK3β compared to GSK3α and was statistically significant (at p = 0.006, p = 0.0002, p = 0.04, p = 0.004, respectively).

The expression of GSK3α was observed more in OSCC tumors than other types of tumors in the oral cavity (such as mucoepidermoid carcinoma, adenoid cystic carcinoma, basal cell carcinoma, adamantinoma, and acinic cell carcinoma considered together) (p = 0.02). Similarly, a significant difference was observed in the overexpression of GSK3β in OSCC tumors compared with other types of tumors as indicated in Table 1 (p < 0.0001; Figure 1B). These results clearly demonstrate a greater role of GSK3β overexpression (which was later found to be mostly inactive) in OSCC.

The cellular expression and distribution of GSK3 was located within different cellular compartments. GSK3β expression was observed in eleven higher grade (MDSCC and PDSCC) tumors. It was expressed in the nuclear compartment (NC), the nuclear-cytoplasmic compartments (N-CC) (Figure 2A (c)) and in only the cytoplasmic compartment (CC) in 5, 4 and 2 cases, respectively. Alternatively, in the lower grade tumors (WDSCC), 4, 5 and 12 samples were found to have positive expression of GSK3β in NC, N-CC and CC, respectively. The trend of GSK3β expression shifting from the cytoplasmic to the nuclear compartment according to disease severity was observed (p = 0.09, though was not significant due to sample size). Alternatively, among the five (5/14) positive tumors of a higher grade (MDSCC and PDSCC), the expression of GSK3α was observed in the NC, N-CC and CC in 2, 2 and 1 of the tumor samples, respectively. Likewise, in the lower grade tumor (WDSCC) samples, GSK3α expression was observed in various cellular compartments, including 5 samples that express it in only the NC, 3 samples in the N-CC and 9 samples in the CC. The normal samples showed expression of GSK3β in the NC and CC in one sample (Figure 2A (a, b)) and the expression of GSK3α in the NC and CC in two and one of the samples, respectively.

Statistics were gathered to determine the expression of GSK3α and GSK3β in OSCC progression. The results showed that in the normal lip, no immunoreactivity was observed for both the GSK3α and the GSK3β antibody (Figure 3A (a, b)). However, in the normal tongue tissue, mild expression of the GSK3α/β protein was observed in the peripheral epithelial layer in most of the cases (Figure 3B (a, b)), except one sample that showed strong immunoreactivity to the GSK3β antibody. In the tissue samples with mild hyperplasia, the expression of GSK3α and GSK3β was observed, and it was limited to the deeper epithelial zone of both the lip and tongue tissues (Figure 3A (c, d) & B (c, d)). Alternatively, the expression of the GSK3α/β protein was observed in the tumors of the lip and tongue (Figure 3A (e, f) & B (e, f)). In the OTSCC progression model, a total of twenty-two samples were analyzed, and GSK3β overexpression was observed in 14.2% (1/7) of normal samples, in 20.0% (1/5) of hyperplasia and 70.0% (7/10) of cancer samples (p = 0.0396). In the lip cancer progression model, a total of fifteen samples were analyzed and GSK3β overexpression was not observed in normal (n = 2) or hyperplasia (n = 3) samples but was observed in 70.0% (7/10) of cancer samples (p = 0.0376). In the distant metastatic SCC samples, the invasive cancer cells at the new location demonstrated an overexpression of GSK3β in 87.5% (7/8) whereas only 12.5% (1/8) of samples showed an overexpression of GSK3α (p = 0.002) (Table 1 and Figure 3C (a to h)). Similarly, nearby lymph node positive cases were found in 60% (3/5) of the GSK3β overexpressing tumors and in only 20% (1/5) of the GSK3α overexpressing tumors.

Inactivation of the GSK3 proteins was detected by determining the expression of pS²¹GSK3α and pS⁹GSK3β at various stages in OTSCC progression. Cancer adjacent apparently normal tongue samples showed faint immunoreactivity of pS²¹GSK3α and pS⁹GSK3β in 42.8% (3/7) and 28.5% (2/7) of the samples (Figure 4. a, b), respectively. However, the overexpression of pS⁹GSK3β was not observed in 85.7% (6/7) of cancer adjacent normal looking tongue samples. One tumor adjacent tongue sample with mild hyperplasia showed moderate expression of both pS²¹GSK3α and pS⁹GSK3β (Figure 4. c, d). Alternatively, in total 85.96% (49/57) and 66.6% (38/57) of OTSCC tissue samples showed pS⁹GSK3β and pS²¹GSK3α protein expression, respectively (Figure 4 e, f) (Table 2). Moreover, 80% (8/10) of the OTSCC samples showed immunoreactivity to GSK3α and GSK3β antibodies (Table 1). If this statistic remains consistent, then all of the OTSCC samples that express GSK3β may be inactivated and nearly 15.0% of all of the OTSCC samples that express GSK3α may still remain active.

Both male and female patient tissue samples showed overexpression of pS²¹GSK3α (p = 0.0032) and pS⁹GSK3β (p = 0.036). Meanwhile, in the age group >40 ≤ 70, the over-expression of inactive GSK3β compared to GSK3α was observed (p = 0.0002). Similarly, the extent of inactive GSK3β expression was observed more than GSK3α expression in small sized (T1-T2 group) tumors (p = 0.0005). The over-expression of the pS⁹GSK3β protein was observed in 86.4% of WDSCC (32/37), 33.3% of MDSCC (2/6) and 60.0% of PDSCC (3/5) samples (p = 0.01). Alternatively, the expression of the pS⁹GSK3β protein was observed in all of the WDSCC and MDSCC (43/43) and observed in at least 60.0% of the PDSCC (3/5) samples (p = 0.0001).

The distribution of pS⁹GSK3β was observed in the nuclear, cytoplasmic or both cellular compartments in human OTSCC samples (Figure 2B (a, b)). Among the nine pS⁹GSK3β-expressing tumors of a higher grade (MDSCC and PDSCC), pS⁹GSK3β expression was observed in the NC, N-CC and in only CC in 6, 2 and 1 samples, respectively. Alternatively, in the lower grade tumors (WDSCC), 8, 6 and 23 samples showed pS⁹GSK3β expression in the NC, N-CC and in only CC, respectively (Figure 2B (a, b)). This trend of pS⁹GSK3β expression shifting from the cytoplasmic to the nuclear compartment with tumor progression was significant (p = 0.01; Figure 2C). Alternatively, a faint reactivity of pS²¹GSK3α was observed in the OTSCC samples. Among the pS²¹GSK3α-stained five positive tumors of a higher grade (MDSCC and PDSCC), the expression in the NC, N-CC and only CC was observed in 3, 1 and 1 tumor samples, respectively. Similarly, in the lower grade tumors (i.e., WDSCC), 4, 7 and 18 tumor samples were positive for pS²¹GSK3α in the NC, N-CC and in only CC, respectively (p = 0.054).

The expression of GSK3α/β, cyclin D1 and p53 (n = 39; Table 3) was detected using WB analysis. Detectable expression of GSK3α was observed in 83.3% normal (5/6), 66.6% PMLs (4/6), and 51.8% tumor samples (14/27). Similarly, the expression of GSK3β was observed in 83.3% normal (5/6), 83.3% PMLs (5/6) and 59.25% oral tumor (16/27) samples. Moreover, 25.9% (7/27) and 40.7% (11/27) of oral tumor samples showed decreased expression of GSK3α and GSK3β expression compared to normal samples. The expression of pSer²¹GSK3α and pSer⁹GSK3β was not observed in normal samples and less expression was observed in the PMLs samples. Alternatively, 81.4% (22/27) and 85.1% (23/27) of the OSCC samples showed immunoreactivity for pSer²¹GSK3α and pSer⁹GSK3β, respectively. Overexpression of the cyclin D1 protein was observed in 70.3% (19/ 27) of OSCC samples and in 100% (6/6) of PMLs compared to the normal oral mucosa tissue samples. Expression of the p53 protein was observed in 48.1% (13/ 27) of the OSCC samples, 50.0% (3/6) of the PMLs samples and 16.6% (1/6) of the normal oral mucosa samples. β-Actin was used as a loading control in these experiments (Figure 5A).

The expression of pS⁹GSK3β (p = 0.001) and pS²¹GSK3α (p = 0.0001) was significantly different in tumors compared to normal and PMLs (Figure 5B (a and b)). Similarly, cyclin D1 protein expression was observed more in tumor samples and PML samples than in normal samples (p = 0.0001; Figure 5B (c)). The overexpression of the p53 protein was greater in oral tumor samples than in the normal counterpart (p = 0.0001; Figure 5B (d)). Further, the expression of pS²¹GSK3α and pS⁹GSK3β was positively correlated with cyclin D1 expression (p = 0.0001 and p = 0.002, respectively; Figure 5C (a, b)). Similarly, p53 expression was positively correlated with the expression of pS²¹GSK3α and pS⁹GSK3β (p = 0.01 and p = 0.001, respectively; Figure 5C (c, d)).

The interaction of GSK3β and cyclin D1 was observed in various oral tumor samples (Figure 6A). Alternatively, a GSK3α-cyclin D1 interaction was not observed. A GSK3β-cyclin D1 association was observed in 100% (6/6) of the higher stage (T3/T4) tumors compared to 66.6% (4/6) of the initial stage (T1/T2) oral tumors samples. The expression of pS⁹GSK3β, total GSK3β and cyclin D1 was detected in the corresponding WCE (Figure 6B) and was correlated with the extent of the GSK3β-cyclin D1 interaction. The results show no statistically significant correlation between the extent of the GSK3β-cyclin D1 interaction and the level of expression of total GSK3β, pS⁹GSK3β, and cyclin D1. Moreover, no correlation was observed between the extent of the interaction and the active fraction of GSK3β (arbitrary units of the GSK3β reading minus the pS⁹GSK3β reading) as shown in Figure 6C.

RT-PCR analysis was performed to determine the expression of cyclin D1 mRNA in various tumor, PML and normal samples (Figure 7A). PMLs and tumor samples showed increased cyclin D1 mRNA levels compared to normal oral mucosa (Figure 7B). The mRNA expression was correlated with the protein expression in the same tissue samples. The correlation of pS²¹GSK3α expression with cyclin D1 mRNA expression was not significant (n = 28, Pearson’s r = 0.2896, p = 0.135) whereas the correlation of pS⁹GSK3β expression with cyclin D1 mRNA expression was significant (n = 28, Pearson’s r = 0.8624, p < 0.0001; Figure 7C and D).