In the course of our studies, we sought to understand the regulation of certain downstream target genes of the Shh pathway, including PTCH1, Cyclin D2, Plakoglobin, NKX2.2, and PAX6, in a panel of medulloblastoma and astrocytoma cell lines and tumors. We attempted to explore any putative regulation of these genes by the major transcription factor involved in Shh signaling, GLI1, as well as at the epigenetic level.
PTCH1
After siRNA-mediated
GLI1 silencing in Daoy and U87MG cell lines, expression of
PTCH1 decreased, compared with scrambled siRNA-transfected and untransfected cell lines, which may suggest positive regulation of
PTCH1 by
GLI1 in medulloblastomas and astrocytomas. We further assessed
PTCH1 expression in cell lines and samples of both tumor types, and observed that 50% of the samples showed high expression levels of
PTCH1. This "mixed-pattern" of
PTCH1 expression amongst samples suggests that there may be
GLI1 independent regulatory mechanisms, both at the genetic and/or epigenetic level, influencing
PTCH1 expression. It is important to note that reports have shown
PTCH1 promoter hypermethylation in several cancers, including medulloblastomas [
31,
32].
Our previous study demonstrated a correlation between high expression levels of
GLI1 and
PTCH1 [
26]. However, some of the cell lines and samples expressed low levels of
PTCH1 in spite of high
GLI1 levels, which may be suggestive of the fact that those cells are in a different phase of GLI1/PTCH1 interplay. As a putative explanation, high expression of
GLI1 associated with low expression of
PTCH1 may indicate a switch-on of Shh signaling (phase 1), trying to exit from a previous resting point -phase 0- (GLI1 low expression/PTCH1 low expression), to advance towards phase 2, a GLI1 high expression/PTCH1 high expression phase characterized by PTCH1 demethylation and expression, to finally reach phase 3 (GLI1 low expression/PTCH1 high expression), equivalent to switching off the Shh signaling process.
For epigenetic studies, we chose the distal region 1C of the
PTCH1 promoter [
29], and found only 1/6 of the medulloblastoma cell lines bearing methylation at the promoter, although two other cell lines also showed low expression levels of
PTCH1. Among the medulloblastoma samples, 2/8 (25%) showed methylation together with lack of
PTCH1expression. Several reports have illuminated aspects of
PTCH1 epigenetic regulation: no methylation has been reported in the proximal promoter region 1B of the
PTCH1 promoter in primary medulloblastomas, suggesting the possibility of methylation of its distal region, 1C [
31]; a knockout mouse tumor model has documented changes in
PTCH1 expression after treatment with demethylating agents [
32]. To follow up on these reports, we decided to further analyze the hypermethylation of
PTCH1 proximal promoter region, 1B. Our results identified methylation of the promoter region in only 1/8 astrocytoma cell lines and 3/27 (11%) astrocytic tumor samples of high histologic grades.
The
PTCH1 promoter was hypermethylated in mice tumor models as demonstrated by changes in
PTCH1 expression after treatment with demethylating agents [
32]. However, another study suggests that there is no methylation of the proximal region of the
PTCH1 promoter,
PTCH1-1B, although methylation may be concentrated at the distal end of the promoter, or even alternative exon variants, including
PTCH1-1B and
PTCH1-1C [
29].
Cyclin D2
To determine whether
GLI1 regulates
Cyclin D2 in medulloblastomas, we quantified levels of
Cyclin D2 transcript upon silencing of
GLI1 by siRNA in the Daoy medulloblastoma cell line. We observed a decrease in
Cyclin D2 expression in comparison to controls, which suggests that
GLI1 may up-regulate
Cyclin D2, concurring with previous reports showing similar results in
GLI1 transformed epithelial cells [
13]. This evidence is strengthened by the presence of the
GLI1 consensus binding sequence on the
Cyclin D2 promoter [
13]. To complete this study, we determined the expression of
Cyclin D2 in 6 medulloblastoma cell lines and 14 tumor samples, and overall, we observed high expression levels of
Cyclin D2 that correlated with high levels of
GLI1 expression. These results indicate that
Cyclin D2 may be positively regulated by
GLI1 in medulloblastomas.
GLI1 silencing in the astrocytic cell line U87MG led to increased
Cyclin D2 expression in comparison with controls transfected with scrambled siRNA and untransfected cells. Our results suggest that
GLI1 does not up-regulate
Cyclin D2. These results do not concur with a previous report showing up-regulation of
Cyclin D2 in a
GLI1 transformed epithelial cell line [
13]. This discrepancy suggests there may be two regulatory pathways: first, the differential regulatory action of the Shh signal (it may vary in different tissue types); and second, the dual nature of
Cyclin D2, behaving at times as an oncogene and other times as a tumor suppressor gene [
33].
Cyclin D2 expression was either low or absent in 8 astrocytic cell lines in comparison with normal brain tissue, although we detected expression of Cyclin D2 protein in all these cell lines, with the exception of SW1044. However, even protein expression was very low in comparison with the housekeeping gene GAPDH, used as a positive internal control. We detected low expression levels of Cyclin D2 transcript in U87MG, A172, LN405, T98G and SW1088 cell lines, which may correlate with low protein expression. Paradoxically, two astrocytic cell lines (CCF-STTG-1 and GOS-3) did not appear to express Cyclin D2 transcript, however, low levels of protein expression was detected. This suggests two possibilities: First, early degradation of Cyclin D2 mRNA due to a short half-life, and second, the possibility of differential splicing. We failed to detect expression of Cyclin D2 protein in any of the tumor samples.
Exploration of epigenetic regulation of
Cyclin D2 in medulloblastomas and astrocytomas was motivated by previous studies which had revealed
Cyclin D2 silencing in cancers such as breast [
34], lung [
35], and prostate [
36], due to promoter hypermethylation.
Our study revealed, to a certain extent, hypermethylation of the Cyclin D2 promoter, although methylation did not fully correlate with silencing of expression in medulloblastoma cell lines. Interestingly, the methylation of the promoter in primary tumor samples was associated with low or no expression of Cyclin D2.
We treated astrocytic cell lines that did not express
Cyclin D2 with the demethylating drug 5-Aza-2'-deoxycytidine and the HDAC inhibitor TSA. The combination of these two drugs improves epigenetic modulation [
37]. After 72 h of treatment, we were able to observe
Cyclin D2 expression in these cell lines (p = 0.0014). Our results are contrary to another study [
38] that showed high levels of
Cyclin D2 expression in the astrocytic cell lines, U87MG and T98G, and a decrease in
Cyclin D2 expression after treatment with the HDAC inhibitor SAHA. Our results suggest that the demethylating agent rather than TSA, is responsible for
Cyclin D2 re-expression. However, when we treated these two cell lines with only 5-Aza-2'-deoxycytidine, we observed little to no expression of
Cyclin D2. However, after 5-Aza-2'-deoxycytidine and TSA treatment, there was an increase in expression of
Cyclin D2 in T98G, but not in U87MG cell line (accompanied by low expression levels of the Cyclin D2 protein).
Despite hemi-methylation of the
Cyclin D2 promoter in U87MG cells, there was no change in
Cyclin D2 expression after treatment. We assessed the hypermethylation status of the
Cyclin D2 promoter in 8 cell lines and 44 astrocytic tumors by studying two putative CpG islands. SW1783 and CCF-STTG-1 cells appeared to be methylated by MSP but not by MCA-Meth. Neither MCA-Meth nor MSP detected methylation in two other astrocytic cell lines (LN405 and SW1088). Moreover, the A172 cell line did not show methylation or even partial methylation at the two CpG sites in spite of no
Cyclin D2 expression in this cell line. Treatment with 5-Aza-2'-deoxycytidine and TSA induced expression of
Cyclin D2, indicating that a third CpG island that we did not analyze may play an important role in regulating
Cyclin D2 expression in this cell line. Unfortunately, we were unable to obtain any primers to assess hypermethylation at this CpG region. Interestingly, a
GLI1 binding consensus sequence is also located at the third CpG-rich region of
Cyclin D2 promoter, indicating the possibility of "patches hypermethylation" at this promoter [
39] (differential methylation pattern in a promoter region which is supposed to be distributed at CpG rich regions).
Plakloglobin
We attempted to determine
Plakoglobin expression in the Daoy medulloblastoma cell line upon
GLI1 silencing.
Plakoglobin expression was decreased when compared with control-transfected and untransfected cell lines. This indicates that
GLI1 may positively regulate
Plakoglobin expression. However, our results do not concur with a previous report which suggests that
GLI1 down-regulates
Plakoglobin in
GLI1 transformed epithelial cells [
13]. Additionally, we assessed
Plakoglobin expression in 6 medulloblastoma cell lines and 14 tumor samples. A majority of the cell lines and tumor samples displayed high expression levels of
Plakoglobin, while only a few of the tumor samples showed little or no expression of
Plakoglobin. Notably, one report suggests that high expression of
Plakoglobin in medulloblastoma samples is considered to be of high prognostic value [
40]. Therefore, our results may support high expression and up-regulation of
Plakoglobin by
GLI1 in medulloblastomas.
High levels of
Plakoglobin expression after
GLI1 silencing in the U87MG astrocytoma cell line is not indicative of positive regulation of this gene by
GLI1. This result concurs with a previous study on
GLI1-transformed epithelial cells [
13]. We then sought to determine
Plakoglobin expression in 8 astrocytic cell lines and 23 primary astrocytic tumor samples. More than 60% (5/8) of the cell lines and 89% (24/27) of the tumor samples expressed
Plakoglobin at lower levels than normal adult brain tissue. Interestingly, there was a distinct pattern of
Plakoglobin expression amongst astrocytic tumor samples: low-grade samples expressed
Plakoglobin, while a few high-grade samples also showed high expression levels of
Plakoglobin in absence of
GLI1 transcript. However, the remaining samples all showed low levels of
Plakoglobin expression in presence of
GLI1 transcript. These results support that
GLI1 does not appear to up-regulate
Plakoglobin in astrocytomas.
PAX6
Silencing of
GLI1 in Daoy cells indicated that
GLI1 may up-regulate the expression of the homeodomain transcription factor I
PAX6 in medulloblastomas. We also determined the expression of
PAX6 in 6 medulloblastoma cell lines and 14 primary tumor samples and observed all cell lines with the exception of two expressed high levels of
PAX6. Similarly, a majority of the primary tumor samples expressed high levels of
PAX6 transcript compared to normal brain tissue. Previous studies have shown that
GLI1 down-regulates
PAX6 gene expression during normal neuronal development [
14,
41,
42].
PAX6 is a transcription factor which regulates several genes involved in cell fate, proliferation, as well as migration of neuroectodermal precursor cells during development [
43,
44]. Interestingly, this suggests different mechanisms of Shh regulation during normal and malignant tissue development. A few studies report high expression levels of
PAX6 in medulloblastoma samples [
45]. Due to the multifunctional roles of this group of genes, it is entirely possible that other mechanisms regulating
PAX6 in medulloblastomas exist, which further need to be explored.
Subsequent to GLI1 silencing, we observed an increase in PAX6 expression in the transfected astrocytoma cell line U87MG. A majority of the cell lines displayed low levels of PAX6 despite high GLI1 expression, as was similarly seen in primary astrocytic tumor samples. Thus, GLI1 does not appear to up-regulate PAX6 expression in astrocytic tumors.
NKX2.2
GLI1 silencing suggests that GLI1 may up-regulate the homeodomain transcription factor II NKX2.2. in medulloblastomas. We failed to detect expression of NKX2.2 transcript in 3 cell lines, observed low expression in 2, and high expression in only one cell line. This pattern of NKX2.2 transcript expression was recapitulated in tumor samples, and was associated with high levels of PAX6 transcript in cell lines and tumors.
GLI1 silencing did not perturb
NKX2.2 expression in the astrocytic cell line U87MG. A majority of astrocytic cell lines (75%) and astrocytoma samples (70%) showed either low or no expression of
NKX2.2 compared to normal adult brain tissue. Nevertheless, a few high-grade samples expressed
NKX2.2 at very high levels when
GLI1 was expressed at low levels. Overall, a majority of the samples displayed low levels of
NKX2.2 expression in the presence of high
GLI1 expression. Interestingly, however, reports suggest that Shh signaling up-regulates
NKX2.2 expression during normal neuronal development [
42,
46]. Low expression levels of
NKX2.2 seen in our study, despite active Shh signaling, is suggestive of differential Shh signaling during normal development and in astrocytomas. Our study further supports a previous report [
47] which shows that
NKX2.2 is a direct target gene of Shh signaling, and is up-regulated during normal development.
Statistical analysis
Statistical analysis by the Fisher's test revealed significant correlations of GLI1 expression with PAX6 (p = 0.015) and NKX2.2 (p = 0.015) expression in medulloblastoma cell lines. However, we did not find significant correlations in the expression of GLI1 with PTCH1, Cyclin D2 or Plakoglobin in the medulloblastoma cell lines. Similarly, in medulloblastoma primary tumor samples, only expression of NKX2.2 showed significant correlation with GLI1 expression (p = 0.004).
On the contrary, we observed significant correlation of GLI1 expression with downstream target genes PTCH1 (p = 0.07), Cyclin D2 (p = 0.006), Plakoglobin (p = 0.02), PAX6 (p = 0.006) and NKX.2.2 (p = 0.0001) in astrocytoma cell lines. Finally, GLI1 expression correlated significantly with downstream target genes PTCH1 (p = 0.005), Cyclin D2 (p = 0.04), Plakoglobin (p = 0.006), PAX6 (p = 0.002) and NKX2.2 (p = 0.008) in astrocytoma primary tumor samples.