Id3 induces an Elk-1–caspase-8-dependent apoptotic pathway in squamous carcinoma cells

Inhibitor of differentiation/DNA-binding (Id) proteins are helix–loop–helix (HLH) transcription factors. The Id protein family (Id1–Id4) mediates tissue homeostasis by regulating cellular processes including differentiation, proliferation, and apoptosis. Ids typically function as dominant negative HLH proteins, which bind other HLH proteins and sequester them away from DNA promoter regions. Previously, we have found that Id3 induced apoptosis in immortalized human keratinocytes upon UVB exposure, consistent with its role as a tumor suppressor. To investigate the role of Id3 in malignant squamous cell carcinoma (SCC) cells (A431), a tetracycline-regulated inducible system was used to induce Id3 in cell culture and mouse xenograft models. We found that upon Id3 induction, there was a decrease in cell number under low serum conditions, as well as in soft agar. Microarray, RT-PCR, immunoblot, siRNA, and inhibitor studies revealed that Id3 induced expression of Elk-1, an E-twenty-six (ETS)-domain transcription factor, inducing procaspase-8 expression and activation. Id3 deletion mutants revealed that 80 C-terminal amino acids, including the HLH, are important for Id3-induced apoptosis. In a mouse xenograft model, Id3 induction decreased tumor size by 30%. Using immunofluorescent analysis, we determined that the tumor size decrease was also mediated through apoptosis. Furthermore, we show that Id3 synergizes with 5-FU and cisplatin therapies for nonmelanoma skin cancer cells. Our studies have shown a molecular mechanism by which Id3 induces apoptosis in SCC, and this information can potentially be used to develop new treatments for SCC patients.


Introduction
Squamous cell carcinoma (SCC) is one of the most common types of cancer and patients with advanced SCC have a median survival time of only 6-9 months [1]. Understanding the molecular mechanism by which SCC tumor cells can be eliminated would therefore benefit a great number of patients.
The inhibitor of differentiation/DNA-binding (Id) family of proteins (Id1-Id4) are dominant negative helix-loophelix (HLH) transcription factors, which heterodimerize with other HLH proteins, including E proteins and MyoD [2], and inhibit their transcriptional activity by preventing them from binding to DNA promoter regions.
Genetic aberrations of Id3 suggest that it may function as a tumor suppressor. More than 60% of neuroblastoma patients harbor an Id3 deletion in chromosome 1p36 [13]. In addition, Id3 is identified as a tumor suppressor gene for Burkitt's lymphoma, as inactivating mutations occur in 68% of cases [11,14]. Thus, there is a link between the ability of Id3 to induce apoptosis and function as a tumor suppressor gene.
In addition to bHLH, Id proteins have been shown to bind and regulate other transcription factors, including the ternary complex factors (TCF, Elk-1, Elk-4/SAP-1, and Elk-3/SAP-2) [15]. TCF proteins are involved in apoptosis. For example, Elk-1 mediates sodium arseniteinduced apoptosis in HaCaT keratinocytes [16]. In addition, in response to secreted protein, acidic and rich in cysteine (SPARC) expression, Sp1 and Elk-1 bind to the caspase-8 promoter and induce apoptosis in MIP101 colorectal cancer cells [17]. Elk-1 is also involved in expression of death receptor 5 (DR5) in response to celecoxib in lung cancer cells [18].
Previously, we found that Id3 induced apoptosis in immortalized human keratinocytes [9]. To investigate the role of Id3 in malignant SCC cells, a tetracycline (Tet)inducible system was used to study effects of Id3 in cell culture and mouse xenograft models. This inducible system was chosen to avoid selection of clones that may acquire genetic alteration to tolerate constitutive Id3 expression.
We found that Id3 induction reduced A431 cell numbers in culture under low serum conditions accompanied by an increase in the sub-G1 population. We conclude that this Id3-mediated apoptosis pathway is Elk-1-and caspase-8-dependent for several reasons. First, Id3 induces elk-1 mRNA and protein as determined by microarray, RT-PCR, and immunoblot analysis. Second, siRNA inhibition of Elk-1 blocks both procaspase-8 and active caspase-8. Third, when A431 cells were treated with caspase-8 and pan-caspase inhibitors, Id3 no longer decreased cell number in low serum or in soft agar assay.
Using Id3 deletion mutants, we found that Id3-induced apoptosis is mediated through its HLH and C-terminal domain (80 amino acids from the C-terminus). Further, Id3 sensitized SCC cells to chemotherapeutic agents including cisplatin and 5-FU (5-fluorouracil). Our studies show that Id3 mediates apoptosis in A431 cells through an Elk-1-caspase-8-dependent pathway. This research can potentially help the development of targeted therapy for SCC patients.

Growth curves
In six-well plates, 3 9 10 4 cells were plated per well, and growth curves for AETet cells determined by Trypan Blue exclusion assays. For cocultures, 1. (National Institutes of Health, Bethesda, MD). The same procedure was employed to determine growth curves with caspase inhibitors (20 lmol/L of Z-IETD-fmk or Z-VADfmk).

Cell cycle analysis
AETet cells were collected, fixed in ethanol, stained with propidium iodide to determine DNA content, and analyzed by Flow Cytometry (FACStar; BD BioSciences, San Jose, CA).

Microarray analysis
AETet cDNA in both A431/Id3 and A431/Vc cells were hybridized to Affymetrix HG-U133A 2.0 gene chips. Gene expression raw data was normalized using dChip software developed by Wong and Lin [19], and analyzed with Meta-Core TM software (Thomson Reuters, Washington, DC). Microarray data have been uploaded to Gene Expression Omnibus database (Accession GSE64535).

Tumor xenografts
In athymic nude mice, 1 9 10 6 GFP-labeled A431/Id3 cells were injected subcutaneously along with 0.5 9 10 6 Ds-Red-labeled A431/Vc cells. Animals fed with AEDoxycline (Dox) diet (200 mg/kg; Bio-Serv, Flemington, NJ) were monitored for tumor sizes using the Maestro TM In-Vivo Imaging System (CRi, Woburn, MA), as described in the approved Georgetown University IACUC protocol. To monitor xenograft sizes, we use Maestro TM software to track GFP fluorescent signal (peaks at 515 nm), and tumor sizes are measured by the software using an area-defined measurement system.

Immunostaining
Paraffin-embedded tumor sections were deparaffinized with xylene followed by antigen retrieval in 6.5 mmol/L sodium citrate at 95°C, pH 6.0. Sections were incubated with antisera specific for active caspase-3 (Asp174; Cell Signaling Technology, Danvers, MA) or Ki-67 (Novus Biologicals, Littleton, CO), overnight at 4°C. Sections were then stained with Alexa 594-conjugated secondary antibodies (Invitrogen TM ) for 1 h and visualized by fluorescence microscopy. Cell scoring was performed using MetaMorph software (Molecular Devices, Sunnyvale, CA).

Real Time quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR)
One step RT-PCR (Applied Biosystems, Foster City, CA) was performed using 1 lg of total RNA. The two-step qRT-PCR was performed using 100 ng of total RNA. Relative Id3 expression levels were normalized to GAPDH levels using 2 ÀDDCt quantification method.

5-FU and cisplatin treatment
In six-well plates, 3 9 10 4 cells per well were plated. AETet cells were treated with 10 lmol/L of 5-FU (Sigma-Aldrich, St. Louis, MO) or 20 lmol/L of cisplatin (Sigma). Seventy-two hours after drug treatment, cell viability was quantified by Trypan Blue exclusion assay. Mice bearing A431 xenografts were also treated twice with 5-FU (100 mg/kg) and cisplatin (6 mg/kg) by intraperitoneal injection. 5-FU treatments were administrated on days 11 and 18, and cisplatin on days 16 and 23. Tumor sizes were monitored using the CRi Maestro TM In-Vivo Imaging System.

Statistical analysis
Student's t-tests were performed and P ≤ 0.05 are considered statistically significant (represented by one asterisk).  based on the Chou-Talalay drug synergy quantification model [20], using Id3/CDDP or Id3/5-FU ratios ranging from 0.01 to 2000.

Establishment of A431/Id3 and A431/Vc cell lines
To investigate the effects of Id3 in malignant SCC cells, a Tet-inducible system was used to induce Id3 expression. First, stable cell lines were verified for Id3 protein induction by Tet (Fig. 1A) and a robust induction of Id3 mRNA (seven-to eightfold) was confirmed by semiquantitative reverse transcriptase-mediated PCR (RT-PCR) as well as by qRT-PCR (Fig. S1). A Tet-induction time course was performed in A431/Id3 clone 5 and A431 vector control (A431/Vc) cell lines. The induction of Id3 in the A431/Id3 cl5 cell line can be seen as early as 2 h after addition of Tet, and a decrease observed after 16 h. There was no detectable Id3 protein in A431/Vc cells throughout the time period tested (Fig. S2). Normalized Id3 protein expression levels relative to GAPDH loading control are shown in Figure S2.

Cells with induced Id3 expression showed decreased cell numbers in low serum
Since Id proteins are associated with cell cycle progression and survival [6,21], we first determined whether induction of Id3 alters cell growth. In normal serum, no differences were observed between AETet cells. A431/Id3 cells were then cultured in low serum (0.1% FBS) conditions and we observed that Id3 induction resulted in significantly fewer viable cells compared to uninduced cells (Fig. 1B). We then examined if the decrease in cell number was concomitant with changes in the cell cycle. We observed an increase in the sub-G1 cell population in Id3-induced cells (Fig. 1D) but no changes in S phase (Fig. 1F) or in other phases of the cell cycle ( Fig. 1E and G). The growth curve of A431 Vc cells was also assayed in the same manner and no differences in cell number were observed, indicating that tetracycline did not alter cell cycle or cell death (Fig. 1C). We next cocultured GFP-expressing A431/Id3 and Ds-Red-expressing A431/Vc cells and examined cell growth AETet in low serum conditions (Materials and Methods). Student's t-tests revealed that coculture with induced A431/Id3 cells did not alter proliferation of A431/Vc cells (Fig. 1H), indicating that changes in cell numbers were not due to cell-cell communication or secreted factors.
To identify the form of cell death induced by Id3, we determined whether caspase-3, the converging point for most apoptotic pathways, was proteolytically processed from procaspase-3 into its smaller, active form. Upon Id3 induction, active caspase-3 subunits were clearly observed (Fig. 1I). This indicates that Id3 induction is sufficient to induce apoptosis in A431 cells by a pathway most likely involving caspase-3.

Microarray analysis revealed Elk-1 as a downstream target of Id3
To investigate the molecular pathways by which Id3 mediates caspase-3 activation and apoptosis in A431 cells, microarray analysis was performed. A431/Id3 and A431/ Vc AETet cells were subjected to microarray analysis. Id3 altered expression levels of 16 genes greater than 1.5-fold, when induced A431/Id3 was compared with the other three groups ( Fig. 2A). We verified expression levels of candidate genes and showed that TBK-BP1, Elk-1, and MXD3 were upregulated at the mRNA (Fig. 2B) and protein (Fig. 2C) levels by Id3.

Caspase inhibitors abolished Id3-induced decreases in cell numbers in low serum, and colony formation in soft agar
To further investigate the involvement of Id3 in apoptosis, we treated cells with caspase-8 and pan-caspase inhibitors. As before, Tet-induction of Id3 reduced cell numbers in the presence of vehicle (Fig. 3B), while the caspase inhibitors attenuated effect of Id3 on cell number decrease in low serum ( Fig. 3C and D). This indicates that Id3-induced decrease in cell number under low serum is caspase dependent. We next investigated the ability of A431/Id3 cells to form colonies in agarose. We observed a significant decrease in cell colonies when Id3 was induced (Fig. 3E). However, in the presence of caspase inhibitors, the decrease in colony formation due to Id3 induction was eliminated (Fig. 3E), indicating that Id3 inhibition of soft agar colonies is also caspase dependent.

Id3-induced apoptosis is C-terminus and HLH dependent
To investigate the domains of Id3 that are responsible for apoptosis, four inducible cell lines expressing Id3 deletion mutants were created and verified (Fig. 4A-B). We observed that deleting the N-terminus of Id3 (DN39) did not abolish Id3-induced apoptosis (Fig. 4D). Other Id3 mutants all lost their ability to reduce cell numbers (Fig. 4C, E-F). Further, Id3 DN39 showed a drastic increase in sub-G1 population (Fig. 4H), while other mutants showed no significant difference in sub-G1 population (Fig. 4G, I-J). These results suggest that the HLH and C-terminal domains of Id3 gene are important for induction of apoptosis in A431 cells.

Id3 induction decreased xenograft sizes in vivo
To investigate the ability of Id3 to suppress tumor growth in vivo, we injected A431 cells subcutaneously into mice and tracked tumor growth via in vivo imaging of  ; Fig. 5A). Induction of Id3 in vivo was achieved through doxycycline (Dox)-containing feed (Fig. 5B). We found that Id3 significantly reduces tumor size (%33%, P = 0.002; Fig. 5C). When A431/Vc were injected into nude mice, no differences were observed in the sizes of AEDox tumors (Fig. 5D). Further, immunoblot analysis showed that Elk-1 is induced in tumors with high levels of Id3 induced by Dox (Fig. 5B). Immunofluorescent staining was then performed to investigate if reduced Id3 xenograft sizes are due to decreased proliferation or increased apoptosis. We observed a 50% increase in active caspase-3 levels in tumors expressing high levels of Id3 (P = 0.01; Fig. 5E). On the other hand, expression of the proliferation marker Ki-67 was unaltered (Fig. 5E). Our data suggest that Id3 decreases xenograft sizes via apoptosis.

Id3 sensitizes SCC cells to chemotherapeutic drugs
Id3 has been shown to mediate cisplatin-induced apoptosis in sarcoma cells [12]. We therefore investigated the effect of Id3 on cell death in A431 cells in response to 5-FU and CDDP. For both drugs, the CI values for 90% and 50% population death are all less than 1, which is an indication of a synergistic effect (Table 1). Thus, Id3 sensitizes A431 cells to CDDP and 5-FU treatments (Fig. 6A-B). We further investigated the efficacy of CDDP or 5-FU in the presence or absence of Id3 in mouse xenografts derived from A431 cells. Our data indicate that CDDP or 5-FU alone decrease xenograft sizes by 10% and 35%, respectively, on the last measured day of growth (day 26). However, in Dox-induced Id3 tumors, CDDP or 5-FU decreases xenograft sizes by 55% (Fig. 6C-D). These results suggest that Id3 levels may be a predictor of drug response or that Id3 itself could serve as a therapeutic target.

Discussion
We demonstrate that Id3 induced apoptosis in malignant A431 SCC cells in culture and in vivo. In cell culture, we show that Id3 reduced cell numbers in low serum and in soft agar, accompanied by an increase in sub-G1 population. In the mouse xenograft model, we show that Id3 significantly decreases tumor size by >30% via apoptosis. Furthermore, we have shown that Id3 induced Elk-1 in vitro and in vivo. This signaling cascade results in the proteolytic activation of caspases-8 and -3, leading to cell death. As revealed by our Id3 mutant analyses, this pathway is likely dependent on the C-terminal and HLH domains of Id3 (Fig. 4C-F).
The mechanism by which Id3 modulates apoptosis in malignant keratinocytes appears to be different from that of nontransformed keratinocytes. In premalignant cells, Id3 induces bax expression at the promoter level [9,22]. On the other hand, ectopic expression of Id3 alone did not elevate Bax protein levels in A431 cells. However, Id3 does enhance Bax protein in response to UVB exposure (480 J/m 2 , data not shown). Deletion of N-terminal 39 amino acids of Id3 renders it a more potent apoptosis inducer ( Fig. 4D and H). The mechanism remains to be investigated, although this might be due to the loss of the Cdk2 S5 phosphorylation site. Phosphorylation of Id3 by CyclinA/E-Cdk2 complexes abolishes interaction between Id and TCF proteins [23]. Deleting this phosphorylation site from Id3 protein may allow Id3 to escape Cdk2 regulation, and thus interact with Elk-1 with greater affinity.
We have demonstrated that Id3 sensitizes A431 cells to cisplatin and 5-FU. Id3 has also been shown to sensitize sarcoma cells to cisplatin [12]. However, the Id1/Id3 aptamer (Id1/3-PA7) has been shown to induce cell cycle arrest and apoptosis in breast cancer cells [32]. Further, Id3 levels in cancer have been shown to be study-and/or cell type-specific. Analysis of the Oncomine â database revealed that Id3 was significantly upregulated in 83 studies and downregulated in 84 studies in which cancers were compared to normal samples (threshold P = 0.05, fold change = 1.5, accessed date 18 June 2014). Thus, employing Id3 as a therapeutic target requires careful evaluation, and context-specific effects should be used to inform design of therapeutics when making decisions to treat cancer patients.

Supporting Information
Additional Supporting Information may be found in the online version of this article: Figure S1. A431/Id3 clone 5 cells were induced by Tet (1 lg/mL) and assayed for Id3 RNA expression by RT-PCR as described in Materials and Methods. GAPDH served as a loading control. qRT-PCR (right) shows Id3 mRNA levels at 0, 4, 8, 12, and 24 h after Tet-induction. Relative Id3 levels were normalized to GAPDH levels in each time point using Ct values. Figure S2.