Previously, we have shown that the TSPO
gene is differentially expressed in a panel of human breast cancer cell lines and clinical specimens in a manner correlating negatively with estrogen receptor status and positively with increasing malignant phenotype [13
]. In the present study, we address the hypothesis that differences in transcriptional regulation of the TSPO
gene contribute to the differences in expression observed in these cell lines and tumor biopsies. To model the increasing malignant character of tumor biopsies, we performed functional analysis of the TSPO
gene in MCF-7 cells, a more differentiated, less invasive, ER-positive breast cancer cell line that weakly expresses TSPO, and MDA-MB-231 cells, which are a less differentiated, more invasive, ER-negative breast cancer cell line that strongly expresses TSPO.
The work described in this manuscript provides the first in-depth description of the human TSPO
promoter and its transcriptional regulation. Similar to the mouse Tspo promoter, 5′-RLM-RACE indicated that the human TSPO
promoter is a TATA-less promoter situated within a CpG island in both MCF-7 and MDA-MB-231 cell lines [16
]. The promoter of the human TSPO
gene was shown to direct transcription from multiple sites within a 40- to 50-bp window in a variety of tissue and cellular contexts. Multiple transcription start sites within a loosely defined region are typically characteristic of TATA-less promoters located within CpG islands. Each cell line and tissue examined exhibited an array of common and unique transcription sites. Interestingly, two of these sites were used in all tissues and cell types examined.
To determine whether different regulatory elements mediate the differences in transcription of the TSPO gene observed in MCF-7 and MDA-MB-231 cells, we functionally analyzed the promoter identified by RLM-5′RACE in each cell line. Basal promoter activity of the full length (−3545/+66) promoter roughly correlated with the relative levels of TSPO expression by each cell line, with MDA-MB-231 cells showing up to 6-fold more promoter activity than MCF-7 cells. Deletion mutagenesis indicated the presence of a distal regulatory element between -3545 and -2679 which was strongly activating in MDA-MB-231 cells, but not MCF-7 cells, which may be characterized in a future investigation. For the remainder of this study, we investigated the regulatory mechanisms acting on the −121/+66 construct, designated the TSPO proximal promoter, which was sufficient to reconstitute near-maximal promoter activity in each cell line. Within this region, database analysis identified five putative GC boxes, regulatory elements which are commonly found in the promoters of genes which are ubiquitously expressed. These regulatory elements typically function as high affinity binding sites for Sp1, Sp3, Sp4 proteins, although additional GC-binding transcription factors have been reported.
Alignment of the TSPO promoter with the mouse Tspo proximal promoter revealed areas of conservation and divergence. In the human promoter, two of the GC boxes overlapped proximally (GC.1/2) and two overlapped distally (GC.4/5), with a single canonical GC box (GC.3) located at a position between the overlapping motifs. In contrast, the murine Tspo promoter was organized as central overlapping GC boxes (Sp1.2.3) flanked distally (Sp1.4) and proximally (Sp1.1) by canonical GC boxes. Additional areas of conservation in the two promoters included the regions surrounding the most distal tss (+1) observed human cells and tissues and the region including the common tss at nucleotide +38. The region of conservation upstream of the tss at the +1 nucleotide includes the overlapping GC boxes 1 and 2, which correspond to the Sp2 and Sp3 elements of the mouse promoter. The region corresponding to the Sp1 element of the murine promoter is also conserved; however, this sequence more closely resembles a MZF-1 motif in the human promoter due to the presence of an adenine at position +10. Additional conserved sequences are found in the vicinity of GC.4, although neither the overlapping core motif (GGGCGG) of GC.5 nor the single GC.3 motif is present in the mouse promoter.
Whether these interspecies differences in promoter architecture result in differences in the regulation of transcriptional is not known. Targeted mutation of GC boxes 1 and 2 did not decrease levels of promoter activity in either breast cancer cell line. In contrast, mutation of GC box 3 resulted in substantial loss of activity, with promoter activity decreasing by as much as 60% in MCF-7 cells and 50% in MDA-MB-231 cells. Mutation of GC boxes 4 and 5 decreased promoter activity by 20–35% in each cell line. Our previous analysis of the mouse Tspo
promoter showed a similar dependence on the central motif, despite its overlapping core binding sequences, with minor contributions from the more distal canonical GC box (Sp1.4) [16
]. Together, these results suggest that central and distal GC boxes within the proximal promoter must be intact for near-maximal basal activity in a variety of contexts, including human breast cancer cell lines and mouse steroidogenic cells.
EMSA and supershift experiments demonstrated that Sp1 and Sp3 from MDA-MB-231 and MCF-7 nuclear extracts bind to the isolated GC Box 3 and the overlapping motifs of GC Boxes 1/2 and GC Boxes 4/5 in vitro. In addition, mutations targeting the core motifs of these GC boxes both reduced proximal promoter activity of luciferase reporter constructs and eliminated competition by these elements in gel shift assays. It should be noted that in the EMSA and anti-Sp1 and Sp3 supershift experiments with both MDA-MB-231 and MCF-7 nuclear extracts, there was a residual retarded band that was not supershifted (, band 3). This could be due to the presence of Sp4 binding. Sp4 protein expression has been reported in several breast cancer cell lines [22
], including the MCF-7 and MDA-MB-231 cell lines. We performed supershift analyses using probes corresponding to either GC Box4/5 or GC Box 3 and nuclear extracts from MDA-MB-231 cells. Incubation with an antibody to Sp4 did not result in either the formation of an observable complex of slower mobility nor a decrease in the amount of retarded complex formed, compared to the reaction in which probe was incubated with nuclear extract alone. Since the binding of Sp1, Sp3, and Sp4 to GC-rich oligos has previously been shown to form retarded complexes with overlapping mobilities, these results suggest that any contribution of Sp4 binding to the probes corresponding to GC boxes 4/5 and 3 in supershift experiments is likely to be minimal compared to the binding of Sp1 and Sp3. Alternatively, ChIP was employed to verify the ability of Sp1, Sp3 and Sp4 to bind the endogenous TSPO
promoter and regulate its expression.
Although EMSA and ChIP analyses indicate that Sp1 and Sp3 bind to GC4/5, GC3, and GC1/2 in vitro
and ChIP showed Sp4 binding to the endogenous TSPO
proximal promoter in intact cells, it does not provide information regarding the function of bound proteins. Sp1 and Sp3 are both bifunctional, acting as either an activator or inhibitor of transcription depending on factors such as post-translational modification, isoform expression, and promoter architecture and context [23
]. Expression studies in Drosophila SL2 cells, which are deficient in GC box-binding Sp/KLF family members, showed that expression of Sp1 or Sp3, either alone or in combination, is sufficient to activate the TSPO
promoter, albeit at relatively low levels. The additive effect of Sp1 and Sp3 co-expression contrasts with some promoters, where Sp3 co-expression diminishes the ability of Sp1 to activate promoter activity [14
]. Since alterations in the relative levels of Sp1 and Sp3 expression have been shown to be important in regulating cell proliferation and tumor progression [25
], we also investigated the effects of over-expressing Sp1 and Sp3 in breast cancer cells on TSPO
proximal promoter activity. Titrating increasing amounts of Sp1 and Sp3 had little effect on promoter activity in MDA-MB-231 cells, except when the largest amount of pPacSp3 is used (50ng), although transfecting increasing amounts of either pPacSp1 or pPacSp3 was sufficient to repress proximal promoter activity in MCF-7 cells. Whether these results reflect actual competition for binding to the TSPO
promoter, differential autoregulatory mechanisms, or off-target effects of these transcription factors is not known. Additionally, combined siRNA pools targeting Sp1, Sp3 and Sp4 were able to significantly reduce Sp1, Sp3 and Sp4 protein levels and TSPO expression in both MDA-MB-231 and MCF-7 cells. These results confirmed the role of Sp proteins on TSPO expression regulatory elements.
Since the mutation of GC boxes had proportional effects on TSPO promoter activity in both MCF-7 and MDA-MB-231 cells, we also examined the sequences including and downstream of the transcription initiation window to see if additional regulatory elements are present which influence promoter activity in these cells. Deletion and substitution analyses suggested that two or more activating regulatory elements are present between +39 and +66 which have a greater effect on TSPO promoter activity in MDA-MB-231 cells than MCF-7 cells, and an inhibitory element between +38/+43 which has greater influence on promoter activity in MCF-7 cells. Interestingly, these elements are located within the region of conservation observed in the alignment of the human and mouse promoters. Analysis of the human TSPO promoter 3′ deletion mutants in MA10 cells, which strongly express TSPO, showed a similar requirement for these downstream sequences for maximal promoter activity (data not shown). Together, these results suggest that sequence dependent mechanisms within the region surrounding and downstream of the +38 tss may be required for full promoter activity in cells that highly express TSPO. Whether these mechanisms include the enrichment of transcription initiation at +38 through Inr function or the action of regulatory elements that differentially modulate TSPO promoter activity in different cell types remains to be determined.
Epigenetic modification of chromatin and DNA may provide additional mechanisms by which promoter activity may regulated in MCF-7 and MDA-MB-231 cells. Treatment with 5-azacytidine showed that methylation appears to regulate TSPO promoter activity in MCF-7 cells, but not MDA-MB-231 cells. Partial methylation was observed in 4/11 sequences in the region separating GC.4/5 and GC.3, suggesting that steric interference at these binding sites may influence promoter activity. However, the methylation of more distal CpGs outside of the region considered in this study may also influence endogenous promoter activity, particularly CpG nucleotides that are immediately adjacent in intron 1.
In contrast to the effects of 5-azacytidine, which only induced promoter activity in MCF-7 cells, treatment with a histone deacetylase inhibitor (TSA) strongly activated the TSPO
proximal promoter in both MCF-7 and MDA-MB-231 cells. Inhibiting histone deacetylase activity to increase histone acetylation provides a signature which can be read by other proteins and destabilizes higher chromatin order to provide greater access to transcription machinery. While this mechanism could account for the observed TSA-inducibility, histone acetyltransferases and histone deacetylases also have non-histone substrates and can modulate gene expression by directly acetylating or deacetylating transcription factors and cofactors. Sp1 and Sp3 have both been reported to be acetylated [28
], although the functional implication of this post-translational modification is not clear. Post-translational modifications could alter the transactivating activities of these factors. Alternatively, synergistic interactions that are present in MDA-MB-231 cells, but absent in MCF-7 cells may control cell-specific mechanisms regulating TSPO
gene expression. Sp1 has been reported to have synergistic interactions with numerous transcription factors [23
]. Given the results of our 3′ deletion analysis, we hypothesize that interactions between a protein bound to the downstream region and Sp1 bound to one or more of the GC boxes synergistically activates transcription in cells which highly express TSPO, such as Leydig cells and MDA-MB-231 cells. Similarly, the absence of binding of this protein to the downstream region or the binding of a protein which does not synergistically interact with Sp1 may lock the TSPO
promoter in a less active functional state, such as normal tissues that ubiquitously express low levels of TSPO or in well-differentiated, ER-positive breast cancer cell lines like MCF-7 cells.
Since Sp1, Sp3 and Sp4 did not seem to be the sole regulators of human TSPO expression, we investigated whether TSPO
gene expression is modulated by epigenetic mechanisms. Interestingly, treating MDA-MB-231 cells with TSA for periods as short as 5 min resulted in the formation of the 36-kDa TSPO dimer, which increased in a time-dependent manner. Since this time is not sufficient for TSA to act at the transcriptional level, it remains to be determined how TSA induced TSPO dimer formation. This phenomenon may be due to TSA inducing TSPO
mRNA stability or translation, or TSA-induced acetylation of TSPO at a lysine residue leading to TSPO dimerization. Acetylation is a less common form of posttranslational modification that is rarely looked at, which takes place at the var- epsilon amino group of lysines [30
]. The addition of an acetyl group to lysines has a significant impact on the electrostatic properties of the protein by preventing positive charges from forming on the amino group. Acetylation can also interact with protein phosphorylation and sumoylation, as well as regulating protein stability and protecting proteins against degradation by ubiquitination. This may play a part in forming the 36-kDa TSPO dimer, which is more prominent in cancers and in response to ROS [31
]. Any potential consequence of TSPO dimerization on its function and cellular distribution in breast cancer cells remains elusive. Antisera for distinct TSPO epitopes identified the main 18-kDa TSPO protein in MA-10 Leydig cells, and in the presence of hormones, immunoreactive proteins of 36 kDa were detected [31
]. These antisera recognized mainly the 36-kDa protein and occasionally the 56-kDa protein in human breast cancer cells, and only limited amounts of the 18-kDa TSPO protein were observed. The higher molecular weight proteins may correspond to TSPO polymers, and their presence correlates with the higher levels of reactive oxygen species (ROS) present in breast cancer cells relative to other cells [31
]. Interestingly, a rapid increase in 36-kDa TSPO dimer formation following TSA treatment was observed in MDA-MB-231 aggressive breast cancer cells, but not in MCF-7 cells (data not shown). In previous studies, we demonstrated that TSPO polymer formation is due to the formation of dityrosines as the covalent cross-linker between TSPO monomers leading to increased drug ligand binding and reduced cholesterol-binding capacity of the polymers [31
]. Further addition of TSPO drug ligands to polymers increases the rate of cholesterol binding. These data indicate that ROS induce the formation of covalent TSPO polymers both in vivo and in vitro. We proposed that the TSPO polymer is the functional unit responsible for ligand-activated cholesterol binding, and that TSPO polymerization is a dynamic process modulating the function of this receptor in cholesterol transport and other cell-specific TSPO-mediated functions. In this context, human breast cancer cells appear to express only TSPO dimers, suggesting the presence of a constitutively active receptor in these cells.
The effect of TSA seems to be mediated mainly through the GC3 box, where a mutation of this site significantly reduced TSPO promoter activity in both cell lines. Sp1 and Sp3 appear to potentiate the TSA effect when co-expressed with the −121/+66 TSPO promoter in MDA-MB-231 cells, but act as inhibitors in MCF-7 cells. Sp1, Sp3, and Sp4 remained bound to the TSPO promoter following TSA treatment, indicating that the effect of TSA may not be mediated directly through binding of Sp proteins to the TSPO promoter. How TSA regulates TSPO expression remains to be investigated. TSA was not able to enhance TSPO expression in MCF-7 to levels comparable to those of MDA-MB-231, indicating that acetylation and methylation are not solely responsible for the difference in TSPO expression between the two cell lines. In addition, whether acetylation and methylation regulate TSPO expression through modifications of the intronic sequence is worthy of investigation. A database analysis of the first intron revealed the presence of multiple putative transcription factor binding sites, such as AP1, Ets, Sp1/Sp3, STAT, P300, PPARα, and c/EBP-α, among many others. The elucidation of any probable enhancing or inhibitory roles of these factors in the regulation of TSPO expression will be beneficial to understanding the mechanisms responsible for differential TSPO expression.
Analysis of the regions flanking the tss window indicated that the TSPO
proximal promoter is situated within a CpG island extending approximately 470 bp upstream and 615 bp downstream. Initiation of transcription at multiple sites has been proposed to be regulated as a cassette by MED-1, a putative regulatory element identified by comparative sequence analysis of the region downstream of the transcription initiation window of several genes with TATA-less promoters [32
]. No sequence motifs displaying greater than 60% identity to the MED-1 consensus sequence were found downstream of the TSPO
promoter. In contrast, the flanking sequence around the common start sites at positions +24 and +38 were found to differ from the consensus mammalian initiator sequence (Inr; consensus sequence YCA+1
NT/AYY) by only one and two bases, respectively. It is currently unclear whether either of these elements can reconstitute Inr function. The presence of two such elements in the same transcription window is uncommon, although TATA-less promoters with more than one Inr have been described for some genes [33
]. Deletion of these sequences did diminish TSPO
promoter activity in MCF-7 and MDA-MB-231 cells by 20–35%, as well as in MA-10 cells (data not shown), whereas only the deletion of the +40 tss diminished promoter activity in HepG2 cells (data not shown). Interestingly, deletion of additional sequences in MCF-7 resulted in the recovery of promoter activity to maximum levels, whereas further deletion caused additional loss of activity in MDA-MB-231 cells. Substitution of unrelated sequences for these Inr-elements also affected promoter activity in MCF-7 and MDA-MB-231 cells. Both deletion and substitution mutagenesis suggested a sequence-dependence in the vicinity of the common start sites that affects promoter activity. This sequence harbors STAT1/3/5, Ets, and AhR/ARNT transcription factor binding sites, as well as a transcription start site. Interpretation of this effect is complicated, since it can disrupt the common start site at +40 and the flanking sequence of the STAT1/3/5 - Ets motif and other transcription factor binding sites. Alternatively, this mutation could disrupt an interaction that negatively regulates the TSPO
promoter in MCF-7 cells. If this repressive interaction were specific to the +40 tss, it may not have been detected by deletion analysis, since the −121/+39 mutation also deletes the start site. In summary, the deregulation of factors interacting with this region as part of inducing TSPO
expression is a potential mechanism by which TSPO may be up-regulated in certain cancers. Additional studies are necessary to identify the factors that interact with this region and to determine the mechanism by which these interaction(s) modulate TSPO expression.
Given its downstream location, it is possible that this region does not exert its effect at the level of transcription. Instead, this sequence could interact with trans-acting factors expressed in cells that express high levels of TSPO to alter mRNA stability or translational efficiency. If the downstream element proves to be regulated at the level of transcription, then the modular nature of the TSPO
promoter should be investigated. One possible implication of the architecture of the TSPO
promoter is that transcriptional regulation is directed through upstream and downstream modules that can integrate multiple signals. Activation through the array of GC boxes by Sp1 or a related protein may be sufficient to activate the TSPO
promoter, but only at low to moderate levels. Based on our characterization of the TSPO
promoter, it appears that interactions with GC Box 3 may activate the TSPO
promoter most efficiently. However, it is likely that the other GC boxes integrate additional signals to modulate TSPO
expression as part of maintaining homeostasis, much in the same way the p21 WAF/Cip1 promoter uses multiple GC boxes to integrate signals from Ras, BRCA1, and ionizing radiation [35
]. In this model, regulation of the TSPO
promoter would require additional interactions with regulatory proteins binding to the downstream element. Full promoter activity may require interaction with an activator; however, the overlapping nature of these putative elements suggests that these downstream sequences may also contribute to cytokine-responsiveness, redox homeostasis, and tissue-specific regulation. Increased understanding of the manner in which the putative GC box module interacts with both the basal transcription machinery and the downstream regulatory module will provide important insights into the mechanism by which TSPO levels are altered in cancer and other disease states where it is overexpressed [11
In summary, regulation of TSPO expression in MDA-MB-231 and MCF-7 cells depends on gene amplification, Sp1, Sp3 and Sp4 regulation of constitutive TSPO
expression through the GC3 promoter region, and epigenetic modification of the proximal promoter, mainly at the GC3 site and the first exon, which seems to play a distinct role in mediating TSPO
expression in the rich-in-TSPO MDA-MB-231 cells. Because of numerous recent reports showing that high concentrations of TSPO drug ligands inhibit proliferation of various cancer cells and sensitize cancer cells to chemotherapy [37
], the concept of a TSPO-mediated cancer therapeutic approach has emerged. However, understanding the regulation of expression and function of TSPO in human cancer cells is paramount to any use of this protein as a drug target for cancer therapy.