Transcriptional Regulation of Serine/Threonine Protein Kinase (AKT) Genes by Glioma-associated Oncogene Homolog 1

Background: Little is known regarding the transcriptional regulation of AKT.

Results: GLI1 contributes to the survival of DLBC cells by promoting transcription of AKT genes.

Conclusion: AKT1 is a novel direct downstream target of the Hedgehog transcriptional factor GLI1.

Significance: Identifying target genes of GLI1 provides insights into the contribution of Hedgehog signaling in the pathobiology of malignant tumors.

Abstract

Aberrant activation of Hedgehog signaling has been described in a growing number of cancers, including malignant lymphomas. Here, we report that canonical Hedgehog signaling modulates the transcriptional expression of AKT genes and that AKT1 is a direct transcriptional target of GLI1. We identified two putative binding sites for GLI1 in the AKT1 promoter region and confirmed their functionality using chromatin immunoprecipitation, luciferase reporter, and site-directed mutagenesis assays. Moreover, we provide evidence that GLI1 contributes to the survival of diffuse large B-cell lymphoma (DLBCL) cells and that this effect occurs in part through promotion of the transcription of AKT genes. This finding is of interest as constitutive activation of AKT has been described in DLBCL, but causative factors that explain AKT expression in this lymphoma type are not completely known. In summary, we demonstrated the existence of a novel cross-talk at the transcriptional level between Hedgehog signaling and AKT with biological significance in DLBCL.

Introduction

Diffuse large B-cell lymphoma (DLBCL)³ is one of the most common non-Hodgkin lymphomas in adults (1). Despite overall improvements in outcomes of DLBCL, ∼30–40% of patients have disease that is either refractory or relapses after standard therapy (2). Therefore, new advances in our understanding of the molecular pathobiology of DLBCL are needed. These advances are expected to contribute to the development of new therapeutic approaches.

Hedgehog (Hh) signaling is an evolutionary conserved signaling pathway that serves several physiological and development processes (3) and is deregulated in several cancers, including DLBCL (4, 5). In fact, inhibition of Hh signaling has been proposed as a useful therapeutic target for some cancers (6–9).

The Hh family of proteins comprises three distinct ligands, Sonic Hh, Indian Hh, and Desert Hh, which appear to be processed intracellularly by autoproteolysis from a precursor peptide (∼45 kDa) to generate an N-terminal secreted peptide that is retained at or near the cell surface (10). Secreted N-terminal Hh peptides interact with a receptor complex composed of two major proteins, a 12-transmembrane domain protein, patched 1, and a seven-transmembrane domain protein, smoothened (SMO). In the absence of Hh ligands, patched 1 inhibits SMO. Once Hh ligands bind to patched 1, this inhibition is released, allowing SMO to transduce the Hh signal mediated by the five-zinc finger transcription factors GLI1, GLI2, and GLI3 (11–13).

The GLI family members contain a conserved C₂H₂ zinc finger DNA-binding domain that can specifically interact with target DNA sequences encompassing a GACCACCCA-like motif (14, 15). Because the full-length GLI1 transcription factor does not contain a repressor domain, it consequently acts as a strong transcriptional activator and participates in the regulation of the expression of numerous Hh target genes, including itself (16–20). Although inappropriate activation of GLI1 has been shown in many cancers, the assessment of the contribution of GLI1 has not been thoroughly examined in hematological malignancies.

We showed previously that GLI1-mediated canonical Hh signaling is active in DLBCL (5, 21). This activation of GLI1 is not due to amplifications of GLI1 (22) or mutations of Hh-related genes (23, 24). Our data support that the Hh pathway in DLBCL is aberrantly activated as its activation is in part mediated by external Hh ligands (autocrine and paracrine Hh signaling loops) (5, 16) but also intrinsically by cross-talks with other oncogenic pathways (22).

AKT (protein kinase B) is a serine/threonine kinase involved in the regulation of cell survival signals in response to growth factors or cytokine stimulation. AKT is one of most frequently hyperactivated kinase in cancer, and it has been shown to play critical roles in the tumorigenesis of many neoplasms (25–27). In mammalian cells, three major isoforms of AKT, termed AKT1, AKT2, and AKT3, encoded by three separate genes have been identified (28). Among the three isoforms, AKT1 is ubiquitously expressed and constitutively activated in several cancers (29, 30). Whereas post-translational regulation of AKT signaling is being extensively studied, there are few data available regarding the transcriptional regulation of AKT1 (31, 32), and its transcriptional regulation remains largely unknown.

It has been reported that activation of AKT predicts poor outcome in patients with DLBCL (33). Multiple mechanisms have been proposed for the activation of AKT in cancer such as mutations of PTEN, amplification of AKT genes, and mutations of genes coding the regulatory and catalytic subunits of PI3K (e.g. PIK3CA) (34). In DLBCL, however, causative factors that might explain the constitutive activation of AKT are not completely known. Functional loss in the p110 catalytic subunit-α (PIK3CA) and inactivation or deletion of PTEN have been reported in a small subset of DLBCL (35–37).

In this report, we provide evidence that canonical Hh signaling regulates the transcription of AKT genes and that AKT1 is a novel direct downstream target of the transcriptional factor GLI1. We also provide evidence that GLI1 contribute to the survival of DLBCL cells by promoting the transcription of AKT genes. Moreover, by finding a strong correlation between AKT1 and GLI1 in DLBCL patient samples, our in vitro data may be extrapolated to DLBCL tumor samples.

EXPERIMENTAL PROCEDURES

Cell Lines, Cell Culture, and Patient Samples

DOHH2 and OCI-LY19 cell lines were purchased from DSMZ (Braunschweig, Germany). HBL1 and 293T cell lines were obtained from ATCC (Manassas, VA). LP cells were established from a diagnostic specimen from a DLBCL patient (38) and characterized as a DLBCL cell line of activated B-cell type (39). LP cells were a kind gift from Dr. Richard J. Ford (Department of Hematopathology, The University of Texas M. D. Anderson Cancer Center, Houston, TX). DLBCL cell lines were exclusively maintained at 37 °C in RPMI 1640 medium (ATCC) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Sigma-Aldrich), 1% l-glutamine, and 1% penicillin-streptomycin in a humidified atmosphere containing 5% CO₂. 293T cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% FBS, 1% l-glutamine, and 1% penicillin-streptomycin at 37 °C in an atmosphere of 5% CO₂. When mentioned, cells were treated with recombinant Shh N-terminal peptide and cyclopamine-KAAD for the indicated time periods. All frozen and paraffin-embedded DLBCL patient specimens, reactive lymph nodes, and DLBCL cells from pleural fluids were provided by the Hematopathology Tissue Bank of The University of Texas M. D. Anderson Cancer Center, Houston, TX.

RNA Extraction and Quantitative Real Time PCR Analysis

Quantitative (q) real time PCR analysis was performed according to the described protocol (16). The primers for GLI1 (Hs01110766), SMO (Hs01090243), AKT1 (Hs00178289), AKT2 (Hs01086102), AKT3 (Hs00178533), BCL2 (Hs00608023), and 18 S RNA (Hs03928985) were obtained from Applied Biosystems (Carlsbad, CA). Each target was amplified in duplicate, and data analyses were done using 2^−ΔΔCT method (40).

Cell Lysates and Immunoblotting

Cells were rinsed with ice-cold phosphate-buffered saline and lysed in buffer containing 40 mm HEPES (pH 7.5), 120 mm NaCl, 1 mm EDTA, 10 mm sodium pyrophosphate, 10 mm sodium glycerophosphate, 50 mm NaF, 1% Triton X-100, and protease inhibitor mixture (Roche Applied Science). The cell lysates were incubated for 20 min at 4 °C for complete lysis and processed for immunoblotting as described (41). The following antibodies were used: histone H3, GLI1 (C6H83), AKT, and phospho-AKT (Ser-473) (Cell Signaling Technology, Danvers, MA); GLI1 (H70) (Santa Cruz Biotechnology, Santa Cruz, CA); SMO (Abcam, Cambridge, MA); and β-actin-HRP (Sigma-Aldrich).

Chromatin Immunoprecipitation (ChIP) Assay

A ChIP assay was conducted using the SimpleChIP Enzymatic Chromatin IP kit according to the manufacturer's protocol (Cell Signaling Technology). GLI1 (C6H83; Cell Signaling Technology) and GLI1 (H300; Santa Cruz Biotechnology) antibodies were used for ChIP analysis. The sequences of primers used in ChIP assays are as follows: binding site 1 (BS1): forward (−4293/+1), 5′-GTACCTAGGTGAATGGTTGACTCC-3′; reverse, 5′-GTGGCTTAGGTTGACTTTCAGG-3′; binding site 2 (BS2): forward (−325/+1888), 5′-GATCAATGGATAAAGTGTGCTCAG-3′; reverse, 5′-ACAAAGAGAGGTTCAGACAAGTCC-3′. PCR product was purified using a PCR purification kit (Invitrogen) and sequenced at the gene sequencing core facility of The University of Texas M. D. Anderson Cancer Center, Houston, TX.

Luciferase Reporter Gene Assay

Luciferase reporter plasmids with luciferase gene under transcriptional control of AKT1 gene regulatory chromatin were obtained from Dr. Jin Q. Cheng (Departments of Pathology and Interdisciplinary Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL) as described previously (32). FLAG-tagged full-length GLI1 plasmid was a kind gift from Dr. Peter Zaphiropoulos (Department of Bioscience and Nutrition, Karolinska Institutet, Sweden). The AKT1-driven luciferase reporter assay was performed as described previously (16).

Site-directed Mutagenesis

Site-directed mutagenesis was performed as described previously (16). The sequences of primers used for mutagenesis are as follows: BS1: forward (−4293/+1), 5′-GCCCCCTCCACGGGGCCCAGAGTGGGG-3′; reverse, 5′-CCCCACTCTGGGCCCCGTGGAGGGGGC-3′; BS2: forward (−325/+1888), 5′-GTAATTATGGGTCTGTAACCAGGGTGGACTGGGTGCTCCT-3′; reverse, 5′-AGGAGCACCCAGTCCACCCTGGTTACAGACCCATAATTAC-3′.

Transfection, Lentiviral Particle Formation, and Infections

Lentiviral human GLI1 shRNAs GLI1-1 (TRCN0000020484) and GLI1-2 (TRCN0000020488) and luciferase (control) plasmids (puromycin selection) were obtained from Sigma Mission shRNA (Sigma-Aldrich). Isopropyl β-d-1-thiogalactopyranoside (IPTG)-inducible control and GLI1 shRNA plasmids were custom made by Sigma Mission shRNA (Sigma-Aldrich Corp.). Precision Lenti AKT1 (blasticidin selection) ORF clone (PLOHS-100067520) was obtained from the shRNA core facility of The University of Texas M. D. Anderson Cancer Center, Houston, TX. Transfections, lentiviral particle production, and infections were done as described previously (42). Briefly, 293T cells (1.2 × 10⁶) were cotransfected with control or GLI1 shRNA together with the ΔVPR and vesicular stomatitis virus G protein plasmids into actively growing cells. Lentiviral particles were harvested 48 h after transfection and centrifuged at 3000 × g for 15 min to eliminate any remaining HEK 293T cells. DLBCL cells were transduced with the collected viral supernatant in the presence of Polybrene (8 μg/ml) and incubated for 48 h. After 48 h, infected cells were selected with puromycin (2–5 μg/ml) or blasticidin (10 μg/ml) for 7 days.

For IPTG-inducible GLI1 knockdown experiments, puromycin-selected control and GLI1 OCI-LY19 cells were treated with 1 mm IPTG for 3–4 days. Upon removal of IPTG from the culture media, the GLI1 expression could be recovered. To recover GLI1 expression, OCI-LY19 cells were washed at least thrice with PBS, centrifuged, and cultured in new RPMI 1640 medium (without IPTG) for 7 days.

Cell Viability, Cell Cycle Arrest, and Apoptosis Analysis

Cell viability of DLBCL cell lines was determined using the Vi-CELL cell viability analyzer (Beckman Coulter, Miami, FL). Cell cycle analysis and Annexin V and propidium iodide (PI) staining (Pharmingen) were performed by flow cytometry as described previously (5).

Double Immunofluorescence Labeling

Immunofluorescence assays were performed as described previously (43). Routine histologic tissue sections of DLBCL involving lymph nodes and reactive lymph nodes were used. Briefly, tissue sections were deparaffinized and hydrated and then underwent heat-induced epitope antigen retrieval in a steamer for 20 min as described previously (21). Tissue sections were permeabilized in PBS containing 0.3% Triton X-100 and 0.3% sodium deoxycholate for 30 min, blocked with Image IT FX signal enhancer (Invitrogen), and incubated with mouse monoclonal GLI1 and rabbit AKT primary antibodies (Cell Signaling Technology) overnight at 4 °C. Unbound antibodies from tissue specimens were removed by washing three times with PBS for 5 min. The secondary antibodies, Alexa Fluor 544 donkey anti-rabbit IgG and Alexa Fluor 488 donkey anti-mouse IgG (Invitrogen), were applied at 1:1000 in PBS for 45 min followed by three PBS washes for 5 min. The sections were incubated in 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen) at 1:2000 in PBS for 5 min for nuclear staining and washed twice with PBS for 5 min. ProLong Gold antifade reagent (Invitrogen) was used for mounting the slides. Images were acquired with a deconvolution microscope using 40×/1.30 objectives with oil immersion. The dot profile analysis of the immunofluorescence images was performed with ImageJ software (National Institutes of Health, Bethesda, MD) as described previously (44).

Statistical Analysis

To calculate the statistical significance of the changes in the response of DLBCL cell lines to drugs, paired Student's t test was used. Student's t test was performed using GraphPad Prism version 5.00 for Windows (GraphPad Software, San Diego CA). To evaluate the significance of AKT1 and GLI1 mRNA expression in tumor specimens and circulating DLBCL tumor cells, the Spearmen correlation coefficient was used.

RESULTS

Activation Status of Hh Signaling Modulates the Transcriptional Expression of AKT Genes

We observed previously that silencing of SMO in lymphoma cell lines resulted in a decrease of total protein levels of AKT. Because of this finding, we decided to explore whether the canonical Hh signaling pathway has a role in the regulation of the expression of AKT genes.

We first inhibited Hh signaling using the classical SMO inhibitor cyclopamine-KAAD (45). Treating two DLBCL cell lines (DOHH2 and LP) that are sensitive to Hh signaling inhibition with cyclopamine-KAAD (2.5 μm for 24 h) resulted in a significant decrease in the mRNA levels of AKT1, AKT2, and AKT3 as compared with controls (Fig. 1A). Cyclopamine-KAAD (2.5 μm) treatments did not affect significantly the number of viable cells as indicated by a cell viability assay (data not shown). A concomitant decrease of GLI1 mRNA confirmed the inhibition of the canonical Hh signaling pathway (Fig. 1A). Expression levels of BCL2 gene, a known direct target of GLI1, were used as an additional positive control (46). Similarly, we observed a decrease in AKT phosphorylation at Ser-473 and total AKT protein levels from lysates obtained from cyclopamine-KAAD (2.5 μm for 48 h)-treated DOHH2 and LP cells as compared with those collected from the control cells (Fig. 1A, right panels).

FIGURE 1.

Hh signaling activity regulates the expression of AKT genes. A, DLBCL cells (DOHH2 and LP) were treated with or without the classical SMO inhibitor cyclopamine-KAAD (CY) 2 μm for 24 h. Treated cells were subjected to qRT-PCR to analyze the mRNA expression of GLI1, BCL2, AKT1, AKT2, and AKT3. As a control (C), we used DMSO (vehicle for cyclopamine-KAAD). DLBCL cells as described in A were used for immunoblotting to detect the phosphorylation levels of AKT (Ser-473) or total AKT after treatment with 2 μm cyclopamine-KAAD for 48 h. B, DOHH2 cells were treated with or without recombinant Shh N-terminal peptide (250 nm) for 24 h. Treated cells were subjected to qRT-PCR to analyze the mRNA expression of GLI1, BCL2, AKT1, AKT2, and AKT3. As a control, we used PBS (vehicle for recombinant Shh N-terminal peptide). C, DOHH2 cells were exposed to CCM obtained from HS5 cells (24 h). Cells were then subjected to qRT-PCR to analyze the mRNA expression of GLI1, BCL2, AKT1, AKT2, and AKT3. As a control, we used DOHH2 cells cultured in 2% FBS medium. DOHH2 cells as described in B and C were used for immunoblotting to detect the phosphorylation levels of AKT (Ser-473) or total AKT after treatment with 250 nm recombinant Shh N-terminal peptide for 48 h or CCM. Results shown in A–C are normalized to the 18 S mRNA level and expressed as -fold change in mRNA expression compared with control. Error bars represent the mean and S.D. of three independent experiments. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

We have shown previously that DLBCL cells are responsive to recombinant Shh N-terminal peptide (R&D Biosystems, Minneapolis, MN) or conditioned culture medium (CCM) obtained from HS5 cells (human stromal bone marrow cells), resulting in activation of the canonical Hh signaling pathway (5, 16). To determine whether AKT mRNA expression is induced with the activation of Hh signaling pathway, we treated DOHH2 cells with recombinant Shh N-terminal peptide or CCM for 24 h and analyzed the expression levels of AKT genes by quantitative RT-PCR. Expression levels of AKT1, AKT2, and AKT3 mRNA were significantly increased in DOHH2 cells after treatment with recombinant Shh N-terminal peptide or CCM (Fig. 1, B and C). Increased mRNA levels of GLI1 and BCL2 in response to recombinant Shh N-terminal peptide or CCM confirmed activation of the Hh signaling pathway (Fig. 1, B and C). To address whether the phosphorylation status of AKT at Ser-473 and total protein levels of AKT are modulated with recombinant Shh N-terminal peptide or CCM, we performed immunoblotting analysis. We observed that phosphorylation of AKT at Ser-473 and total AKT protein levels were increased in recombinant Shh N-terminal peptide (250 nm for 48 h)- or CCM-treated cell lysates as compared with control (untreated) cell lysates (Fig. 1, B and C, right panels).

To further confirm the transcriptional regulation of AKT genes by canonical Hh signaling, we established LP and OCI-LY19 cells with stable knocked down expression of SMO gene using a lentiviral shRNA system (Fig. 2). The mRNA levels of AKT1, AKT2, and AKT3 were significantly decreased in SMO-depleted LP cells in comparison with the control cells harboring luciferase shRNA (Fig. 2A). The decreased level of SMO and GLI1 mRNA confirmed the suppression of the Hh signaling pathway (Fig. 2A). As expected, we observed decreased AKT phosphorylation at Ser-473 and total AKT protein levels from lysates collected from SMO knockdown LP cells as compared with those collected from the control cells (Fig. 2B). Because GLI1 is constitutively activated in DLBCL cells, these experiments were done without Hh stimulation.

FIGURE 2.

SMO participates in the expression of AKT genes. A, LP cells were infected with lentiviruses expressing shRNAs (sh) targeting luciferase (Luci.) (control) and SMO. The transduced cells were selected with puromycin, and the expression levels of SMO, GLI1, AKT1, AKT2, and AKT3 mRNA were analyzed by qRT-PCR. B, the same cells as described in A were used for immunoblotting to detect the phosphorylation levels of AKT (Ser-473) or total AKT after silencing of SMO. Cont., control. C, similarly, OCI-LY19 cells were transduced with lentiviruses expressing shRNAs targeting luciferase (control) and SMO and treated with or without recombinant Shh N-terminal peptide (250 nm) for 24 h. Treated cells were subjected to qRT-PCR to analyze the mRNA expression of SMO, GLI1, AKT1, and AKT2. As a control, we used PBS (vehicle for recombinant Shh N-terminal peptide). Results shown in A and C are normalized to the 18 S mRNA level and expressed as -fold change in mRNA expression compared with control. Error bars represent the mean and S.D. of three independent experiments. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

Similar results were obtained in SMO-depleted OCI-LY19 cells (Fig. 2C). Although GLI1 is constitutively activated in DLBCL cells, adding recombinant Shh N-terminal peptide resulted in further increased expression levels of AKT1 and AKT2 in cells harboring luciferase shRNA. However, a significant decrease in the expression of AKT1 and AKT2 was noticed in those cells with silenced SMO despite the presence of Shh N-terminal peptide (Fig. 2C), further supporting a role of SMO in the expression of AKT genes.

These findings indicate that canonical Hh signaling modulates the expression of AKT genes. However, these data do not clarify whether the modulation of AKT expression is due to GLI1, the main transcriptional activator of Hh signaling.

AKT1 Gene Expression Is Transcriptionally Regulated by GLI1

Having shown that SMO played a role in the transcriptional regulation of AKT genes, we decided to investigate whether these changes are mediated by its downstream transcriptional factor, GLI1. We knocked down GLI1 in OCI-LY19 cells using an RNAi system (Fig. 3) and performed qRT-PCR. Silencing of GLI1 resulted in a decrease of GLI1 expression associated with decreased levels of BCL2, AKT1, and AKT2 (Fig. 3A). We could not detect mRNA levels of AKT3 in OCI-LY19, indicating very low expression of AKT3 in these cells. Similarly, we observed decreased phosphorylation of AKT at Ser-473 in parallel with decreased levels of total AKT protein in GLI1 knockdown OCI-LY19 cells as compared with the control cells harboring luciferase shRNA (Fig. 3B).

FIGURE 3.

Silencing of GLI1 decreases the expression of AKT genes. A, qRT-PCR analysis of GLI1, BCL2, AKT1, and AKT2 mRNA expression levels in OCI-LY19 cells lentivirally transduced with shRNAs (sh) targeting GLI1 or luciferase (Luci.) (control). B, the same cells as described in A were used for immunoblotting to detect the phosphorylation levels of AKT (Ser-473) and total AKT after silencing of GLI1. C, OCI-LY19 cells were transduced with lentiviruses expressing IPTG-inducible shRNAs targeting luciferase (control (Cont.)) and GLI1. The transduced cells were selected with puromycin and treated with 1 mm IPTG for 3–4 days to suppress the expression of GLI1. qRT-PCR analysis of GLI1, BCL2, AKT1, and AKT2 mRNA expression levels in IPTG-inducible control, GLI1-depleted, and GLI1-recovered OCI-LY19 cells (IPTG-GLI1-R) were performed as described under “Experimental Procedures.” D, the same cells as described in C were used for immunoblotting to detect the phosphorylation levels of AKT (Ser-473) and total AKT. E, the same cells as described in C were treated with or without recombinant Shh N-terminal peptide (250 nm) for 24 h. Thereafter cells were subjected to qRT-PCR to analyze the mRNA expression of GLI1, BCL2, AKT1, and AKT2. As a control, we used PBS (vehicle for recombinant Shh N-terminal peptide). Results shown in A, C, and E are normalized to the 18 S mRNA level and expressed as -fold change in mRNA expression compared with control. Error bars represent the mean and S.D. of three independent experiments. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

To confirm the regulatory role of GLI1 in the transcription of AKT genes, we established an inducible GLI1 knockdown lymphoma cell line (OCI-LY19) using a lentiviral IPTG-inducible system. Initial characterization confirmed that, upon IPTG addition, OCI-LY19 cells showed decreased expression of GLI1, BCL2, AKT1, and AKT2 genes (Fig. 3C). Removal of IPTG from the culture medium resulted in recovery of the expression levels of AKT1 and AKT2 (Fig. 3C). Similarly, we observed a decrease in AKT phosphorylation at Ser-473 and total protein levels of AKT from lysates obtained from IPTG-inducible GLI1 knockdown OCI-LY19 cells as compared with IPTG-inducible control cells. Removal of IPTG from the culture medium resulted in recovery of the expression levels of phospho-AKT and total AKT (Fig. 3D).

These experiments were also performed in the presence of recombinant Shh N-terminal peptide (Fig. 3E). Adding Shh N-terminal peptide resulted in increased expression levels of AKT1 and AKT2 genes in cells harboring luciferase shRNA but not in those with silenced GLI1, further confirming a role for GLI1 in the transcriptional regulation of AKT genes. Altogether, these data confirm that GLI1 regulates the transcription of AKT genes.

GLI1 Binds to the AKT1 Promoter

AKT is one of the hyperactivated kinases in cancer and plays an essential role in tumorigenesis (25–27). In mammalian cells, three major isoforms of AKT (AKT1, AKT2, and AKT3) have been identified (28). Using MatInspector professional version 7.2 (47), we identified several potential GLI1 binding sites (GLI1 binding motif, 5′-GACCACCCA-3′) in the gene promoters of the three AKT genes (Fig. 4A and Table 1). We identified two putative GLI1 binding sites (BS1 and BS2) located upstream of the transcriptional start site of AKT1 gene (Fig. 4A), and three and two potential GLI1 binding sites were identified in the promoters of AKT2 and AKT3, respectively (Table 1). The homology of each GLI1 binding site to the consensus sequence was 67% for BS1 and 78% for BS2.

FIGURE 4.

GLI1 transactivates AKT1 promoter activity. A, schematic diagram of the luciferase constructs that include two potential GLI1 binding sites (BS1 and BS2) in the AKT1 promoter. The 9-base pair sequence of the GLI binding site is shown along with the sequence of two closely spaced GLI1 binding sites identified in the AKT1 promoter. B, ChIP assays were performed with two independent control or GLI1-specific antibodies. These studies resulted in the precipitation of AKT1 promoter chromatin containing the GLI1 binding sites in DOHH2 cells. The human RPL30 promoter region serves as a positive control, and IgG serum monoclonal (Mono) and polyclonal (Poly) antibodies were used as controls. Neg., negative. C, a series of AKT1-luciferase constructs were transfected into 293T cells together with or without full-length GLI1 plasmid and subjected to a luciferase reporter assay. Vect., vector. D, AKT1 luciferase constructs as indicated were transfected into DOHH2 cells. After 24 h, cells were treated with or without recombinant Shh N-terminal peptide for 24 h and subjected to a luciferase reporter assay. As a control, we used PBS (vehicle for recombinant Shh N-terminal peptide). E, DOHH2 cells were also treated with or without cyclopamine-KAAD (CY) (2.5 μm for 24 h) and subsequently subjected to luciferase reporter assays. As a control, we used DMSO (vehicle for cyclopamine-KAAD). Results shown in C–E are normalized to Renilla luciferase and expressed as -fold change in relative luciferase activity compared with control. Error bars represent the mean and S.D. of three independent experiments. Expression of GLI1 was confirmed using qRT-PCR assays. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

TABLE 1.

Prediction of potential GLI1 binding sites (GLI1 binding motif, 5′-GACCACCCA-3′) on AKT2 and AKT3 promoters using MatInspector professional version 7.2

Opt., optimized threshold; sim., similarity.

Gene name	Matrix family	Detailed family information	Matrix	Opt.	Start position	End position	Anchor position	Strand	Core sim.	Matrix sim.	Sequence
AKT2	V$GLIF	GLI zinc finger family	V$GLI1.01	0.87	730	744	737	−	1	0.876	cagtcctCCCAaaat
AKT2	V$GLIF	GLI zinc finger family	V$GLI1.01	0.87	379	393	386	−	1	0.904	gccacctCCCAgcca
AKT2	V$GLIF	GLI zinc finger family	V$GLI1.01	0.87	137	151	144	+	1	0.944	ttgacctCCCAaagt
AKT3	V$GLIF	GLI zinc finger family	V$GLI1.01	0.87	3	17	10	−	1	0.911	tccacctCCCAggtt
AKT3	V$GLIF	GLI zinc finger family	V$GLI1.01	0.87	336	350	343	−	1	0.911	tccacctCCCAggtt

To confirm a direct interaction between GLI1 and the AKT1 promoter in DLBCL cells, we performed ChIP studies in DOHH2 cells using two independent control and GLI1-specific antibodies (Fig. 4B). These studies demonstrated the precipitation of AKT1 promoter encompassing BS1 and BS2 regions and therefore the direct binding of GLI1 with the AKT1 promoter.

To validate the functionality of both BSs, a luciferase reporter assay was performed in 293T cells cotransfected with vector or full-length GLI1, Renilla luciferase, and AKT1 promoter-luciferase constructs (pGL3−4293/+1888) containing BS1 and BS2. As shown in Fig. 4C, expression of GLI1 significantly stimulated the luciferase activity of AKT1 promoter (13-fold), implying that the AKT1 response elements reside in this region. To test the functionality of the individual BSs, we performed the same luciferase reporter experiments in two deleted constructs (pGL3-AKT1−4293/+1 and pGL3-AKT1−325/+1888) of the AKT1 promoter, each containing one putative GLI1 BS. The luciferase assay showed that GLI1 significantly increased the luciferase activity of both AKT1-deleted constructs as compared with the empty vector control (Fig. 4C). Similarly, activation of GLI1 using recombinant Shh N-terminal peptide increased the luciferase activity in DOHH2 (Fig. 4D) and 293T cells (data not shown) transfected with the AKT1 constructs as compared with untreated cells. On the other hand, inhibition of GLI1 activity with cyclopamine-KAAD (2.5 μm for 24 h) considerably decreased the AKT1 luciferase activity as compared with untreated DOHH2 (Fig. 4E) and 293T cells (data not shown).

To further confirm the functionality of these BSs, we mutated three cytosines (C) to guanines (G) in BS1 and BS2 (Fig. 5A) to see whether these mutations could abrogate the ability of GLI1 to increase AKT1 luciferase activity. When we cotransfected mutated BS1 and BS2, Renilla luciferase constructs, and full-length GLI1 in 293T cells, we observed ∼4-fold less luciferase activity in comparison with the wild-type constructs, indicating a decreased binding of GLI1 to both mutated BSs (Fig. 5B).

FIGURE 5.

Mutation analysis of GLI1 coding sequences on AKT1 promoter region. A, three cytosines (C) were mutated to three guanines (G) in the BS1 and BS2 regions. B, mutated (mut.) and wild-type (wt) AKT1-luciferase constructs were transfected into 293T cells together with or without full-length GLI1 plasmid and subjected to luciferase reporter assays. Results are normalized to Renilla luciferase and expressed as -fold change in relative luciferase activity compared with control. Error bars represent the mean and S.D. of three independent experiments. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

GLI1 Contributes to DLBCL Cell Survival through Transcription of the AKT1 Gene

It has been reported previously that AKT1 knock-out mouse embryonic fibroblasts are more susceptible to apoptosis upon exposure to genotoxic stress, suggesting an important role for AKT1 in cell survival (48). To address whether GLI1 is important for DLBCL cell survival, we established stable GLI1 knockdown DLBCL cell lines (OCI-LY19 and HBL1) using a lentiviral shRNA system (Figs. 2A and 6A) and performed cell viability (trypan blue) assays. Cell viability assays demonstrated that GLI1 knockdown DLBCL cells experienced a statistically significant decrease in the number of viable cells in comparison with control cells harboring luciferase shRNA (Fig. 6B).

FIGURE 6.

AKT1 mediates GLI1 survival signals in DLBCL. A, two DLBCL cell lines (OCI-LY19 and HBL1) were infected with lentiviruses expressing shRNAs (sh) targeting luciferase (Luci.) (control) and GLI1. The transduced cells were selected with puromycin, and the expression level of GLI1 mRNA was analyzed by qRT-PCR (results for OCI-LY19 are presented in Fig. 2A, left panel). Results are normalized to the 18 S mRNA level and expressed as -fold change in mRNA expression compared with control. B, the same cells as described in A were seeded onto 6-well plates and incubated for the indicated time periods. Cells were harvested, and viable cell numbers were counted by trypan blue exclusion. Relative cell viability (-fold change) was calculated as follows: Relative cell viability = Viable cell numbers in control or cyclopamine-KAAD (CY)-treated at the indicated time periods/Control at 0 h. C, control and GLI1 knockdown HBL1 and OCI-LY19 cells were seeded onto 6-well plates and incubated for 24 h. Annexin V and PI staining was analyzed by flow cytometry. The percentage of Annexin V/PI-positive cells was calculated as follows: Percentage of Annexin V/PI-positive cells = (Annexin V/PI-positive cells/Total cell number) × 100. Error bars shown in A–C represent the mean and S.D. of three independent experiments. D, full-length AKT1 was expressed in control (harboring luciferase) and GLI1-2 knockdown HBL1 cells using a lentiviral expression system. The AKT1-transduced cells were selected with blasticidin and subjected to qRT-PCR to analyze the expression levels of GLI1 and AKT1 mRNA. Results are normalized to the 18 S mRNA level and expressed as -fold change in mRNA expression compared with control. E, the same cells as described in D were seeded onto 6-well plates and incubated for the indicated time periods. Cells were harvested, and viable cell numbers were counted by trypan blue exclusion. The percentage of relative cell viability was calculated as follows: Percentage of relative cell viability = (Viable cell number in control or GLI1 shRNA group/Control at 0 h) × 100. *, p < 0.05; **, p < 0.005; ***, p < 0.0005.

To examine how depletion of GLI1 suppresses the number of viable DLBCL cells, we performed apoptosis and cell cycle analysis. Annexin V and PI assays demonstrated a marked increase of apoptosis in GLI1-depleted DLBCL cells versus controls (Fig. 6C). In addition, cell cycle analysis demonstrated that DLBCL cells with GLI1 knockdown experienced a mild to modest decrease in cell proliferation due to G₁ cell cycle arrest in comparison with the control cells harboring luciferase shRNA (data not shown).

To address whether decreased cell survival in GLI1-depleted lymphoma cells is associated with AKT1 expression, full-length AKT1 was expressed in control (harboring luciferase) and GLI1-2 knockdown HBL1 cells using a lentiviral expression system (Fig. 6D). Constitutive expression of AKT1 in GLI1 knockdown cells partially rescued the cell viability of lymphoma cells in comparison with GLI1 knockdown cells without transduced AKT1 (Fig. 6E). Altogether, these data suggest that GLI1 contributes to cell survival in part through the transcription of AKT1 gene.

GLI1 and AKT1 Levels Are Positively Correlated in Human DLBCL Tumors

To investigate whether there is any correlation between GLI1 and AKT protein levels in human DLBCL tumors, we performed double immunofluorescence analyses in paraffin-embedded DLBCL tumor and reactive lymph node (non-neoplastic) specimens. We found higher nuclear expression of GLI1 and cytoplasmic AKT expression in DLBCL specimens as compared with reactive lymph nodes (control; Fig. 7, lower panels). There was a correlation between GLI1 and AKT protein levels as shown in the plot profile analysis (Fig. 7, upper panels).

FIGURE 7.

Expression levels of AKT and GLI1 in non-neoplastic lymph nodes and DLBCL tumors. Reactive lymph node (control) and DLBCL tissue sections were stained with anti-mouse GLI1 (green) and anti-rabbit AKT (red) antibodies. Nuclear staining was done with DAPI (blue). Plot profile analyses of GLI1 and AKT fluorescence signal intensities were done using ImageJ software.

Next, we performed quantitative real time PCR analyses in 11 frozen DLBCL tumor specimens. The real time PCR analysis revealed a strong Spearmen correlation coefficient (R² = 0.88) between GLI1 and AKT1 mRNA expression (Fig. 8A). We additionally analyzed expression of AKT1 and GLI1 in four DLBCL samples collected from apheresis samples from pleural effusions (samples lacking a stromal component). In the apheresis samples, a positive correlation between the mRNA expression of GLI1 and AKT was also noted (R² = 0.90) (Fig. 8B). Similarly, we also found a significant Spearmen correlation coefficient (R² = 0.74) between AKT1 and GLI1 protein expression in 50 neoplastic non-lymphoma cell lines available in the Human Protein Atlas database (Fig. 8C).

FIGURE 8.

Correlation between AKT and GLI1 expression levels in cancer cells. A, qRT-PCR analysis of GLI1 and AKT1 mRNA expression levels in DLBCL tumor specimens. Results are normalized to the 18 S mRNA level and expressed as the percentage of change in mRNA expression compared with control. B, qRT-PCR analysis of GLI1 and AKT1 expression in primary DLBCL cells obtained from apheresis samples from pleural effusions from four patients with DLBCL. Results are normalized to the 18 S mRNA level and expressed as the percentage of change in mRNA expression compared with control. Error bars represent the mean and S.D. of two independent experiments. C, protein expression levels of AKT1 and GLI1 in 50 cell lines available in the Human Protein Atlas database.

DISCUSSION

GLI1, a full-length transcriptional activator, has been shown to be involved in the intracellular signal transduction controlled by the Hh family of secreted ligands (49). Although a comprehensive analysis of direct transcriptional targets of GLI1 at the genomic level is lacking, a number of studies have shown that GLI1 directly regulates the transcription of various genes that are known to be involved in cell proliferation, survival, and chemotolerance (16–20).

We have shown previously that canonical Hh signaling is a key factor behind high ABCG2 expression in DLBCL through direct up-regulation of ABCG2 gene transcription (16). Now we provide evidence of the contribution of GLI1 in the survival of DLBCL cells and demonstrate for the first time that the GLI1-mediated canonical Hh signaling pathway modulates the transcriptional expression of AKT1, AKT2, and AKT3 genes. In particular, we demonstrated that AKT1 promoter possesses two GLI1 binding sites and that the expression of AKT1 is regulated at the transcriptional level by GLI1. These findings are important because they contribute to the understanding of the transcriptional regulation of AKT1, which is known to provide survival signals to the cells (48).

The AKT signaling pathway is activated in DLBCL cell lines as well as in a subset of DLBCL primary tumor samples independently of the molecular subtype (27, 33). Several mechanisms have been proposed for the activation of AKT in DLBCL such as inactivation or deletion of PTEN and mutations of PIK3CA, the gene coding for the catalytic subunit, p110α, of PI3K (35–37). Some DLBCLs are also characterized by the overexpression and secretion of cytokines such as IL-6 and IL-10 that result in the activation of the AKT pathway among other pathways (50). Although the AKT signaling has been extensively studied at the kinase activity levels, there are few data available regarding AKT transcriptional regulation, and the overall transcriptional regulation of AKT remains largely unknown. Transcription factors that have been found to contribute to the transcription of AKT genes are β-catenin and signal transducer and activator of transcription (STAT3) (31, 32).

In our study, we observed that the total levels of AKT are parallel with its phosphorylation levels at Ser-473. A similar correlation between total AKT and its phosphorylation levels was also seen by Dihlmann et al. (31) using colon cancer cells. These authors found that the phosphorylation levels of AKT at Ser-473 correlated with the total levels of AKT, supporting that up-regulation or down-regulation of AKT genes at the transcriptional level results in changes of AKT activity. It is known that the activity of a kinase is not necessarily a reflection of its total protein level as extra- and intracellular factors such as cytokines, growth factors, and activated oncogenic signals contribute to modulate the kinase activity independently of its transcriptional level. However, the presence of high kinase levels likely contributes to oncogenesis by establishing the basis for an enhanced kinase signaling activity.

Bidirectional interconnectivity between Hh signaling and the PI3K/AKT pathway at the post-transcriptional level has been well documented (41, 51, 52). For example, Riobó et al. (51) have reported that AKT activation is important for activation of Hh signaling effects that were mediated by AKT control of PKA-mediated GLI proteasomal degradation. Several other groups have also reported that adenoviral delivery of activated AKT in non-neoplastic and neoplastic cells is associated with activation of Hh signaling and increased expression of GLI1 (41, 53). Recently, Wang et al. (53) showed that activation of GLI1 by AKT requires S6K1 and that S6K1 phosphorylates GLI1 and enhances GLI1 activity. Moreover, studies in mice have also found a synergistic effect between PI3K/AKT and Hh signaling in medulloblastoma tumorigenesis (52). This view of interconnectivity between Hh and PI3K/AKT at multiple levels, including at the transcriptional level as reported here, has potential clinical implications as inhibitors of both pathways are currently available, and the use of a combination of inhibitors of both pathways may result in synergistic cytotoxicity effects or in decreased chemotolerance of the tumor cells to current chemotherapeutic protocols.

In conclusion, we have demonstrated that the canonical Hh signaling pathway regulates the transcription of AKT genes and that AKT1 is a novel direct downstream target of the transcriptional factor GLI1. We also provide evidence that GLI1 contributes to cell survival of DLBCL cells through the expression of AKT in DLBCL and likely in other malignant tumors.