|Home | About | Journals | Submit | Contact Us | Français|
We used a xenograft model to investigate whether the aryl hydrocarbon receptor deletion construct CΔ553 suppresses tumor growth. HeLa cells that were infected with CΔ553 expressing adenovirus (Ad553) formed very small tumors whereas the control adenovirus-infected cells formed large tumors at day 15. CΔ553 inhibited the formation of the HIF-1 DNA complex and suppressed the induction of the HIF-1α target proteins CAIX and GLUT1. The Ad553 tumors had less HIF-1 function since they showed reduced microvessel formation and lesser amounts of HIF-1α, Arnt, phospho-Akt, CAIX, and GLUT1. Proteasome-mediated Arnt degradation was enhanced in Ad553-infected HeLa cells and tumors.
Hypoxia inducible factor-1 (HIF-1) plays a pivotal role in our compensatory response to the low cellular oxygen content in localized tissues. HIF-1 is a heterodimeric transcription factor consisting of HIF-1α and Arnt (HIF-1β) that regulates the expression of a battery of target genes related to glucose metabolism and angiogenesis . In the case of solid tumors, the hypoxic core stabilizes the HIF-1α protein which is responsible for more efficient glucose metabolism in cancers. Other than control of the HIF-1α function at the protein level, HIF-1α transcription is regulated via the Akt pathway  and phosphorylation of HIF-1α by MAPKs p42/p44  and ERK  appears to affect its transcriptional activity, revealing other means cancer cells may control the HIF-1α function under normoxia. Cancers that overexpressed HIF-1α have a poor prognosis [5–6] and inhibition of HIF-1α after ionizing irradiation appears to be a rational approach for chemotherapy .
The aryl hydrocarbon receptor (AhR) is a well known xenobiotic sensing transcription factor which regulates a battery of gene transcription most notably the phase I and II drug metabolizing enzymes . Since Arnt is a mutual partner of AhR and HIF-1α, we had proposed that there may be crosstalk between these two pathways and had demonstrated that indeed crosstalk exists between AhR and HIF-1α . In principle, if there is an AhR-like molecule interacting with most of the nuclear Arnt protein, HIF-1α signaling may be inhibited. We tested out this hypothesis by utilizing the human AhR construct CΔ553 (amino acid 1–295) which contains the DRE and Arnt binding domains but not the ligand binding domain because 553 amino acids are deleted from the C-terminus. We reported that CΔ553 suppresses the CoCl2-driven luciferase and HIF-1 target gene expression in MCF-7 cells . In this paper, we showed that CΔ553 inhibits xenograft tumor growth and the mechanism of inhibition is consistent with the inhibition of the HIF-1 signaling in xenograft tumors.
HeLa cells were cultured in DMEM (Sigma) supplemented with 10% FBS (Tissue Biologicals), 100 units/ml of penicillin, 0.1mg/ml of streptomycin and 2mM glutaMAX. Gel shift probes were purchased from Invitrogen as follows: HRE (OL105, 5′-GCCCTACGTGCTGTCTCA-3′; OL106, TGAGACAGCACGTAGGGC-3′); mutated HRE (OL107, 5′-GCCCTAATTGCTGTCTCA-3′; OL108, TGAGACAGCAATTAGGGC-3′). Adenovirus Ad553 and AdLacZ generation is described under the supplementary data section.
Animal experiments were carried out in accordance with the guidelines of NIH animal use (NIH publication number 85–23) and of the University of the Pacific. When HeLa cells were 80% confluent in a 75cm2 flask, 100 μl of high titer (3×107 pfu/ml) stock of Ad553 or AdLacZ or PBS was added to the cells (see supplementary data for virus generation). After 72 h at 37 °C, cells were harvested, washed with PBS, and subjected to centrifugation at 1,000g for 5 min at 4 °C. Cells were resuspended in cold PBS and 5 × 106 cells (100 μl) were injected subcutaneously to each of two lower back sites of a 6–8 week-old female homozygous athymic nude mouse (Simonsen Lab). Tumors were measured two to three times a week, and the tumor volume was calculated as a × b2 × 0.5, where a and b were large and small diameters, respectively. Mice were sacrificed after 2–3 weeks when tumors were close to 2 cm3 and the tumor weights were then measured.
After SDS-PAGE, proteins transfer was performed using Bio-Rad mini Trans-Blot cell for 1.5 h at 4 °C. Primary antibody was incubated overnight at 4 °C. SuperSignal West Pico HRP chemiluminescent substrate (Thermo Scientific) was used to detect the signals. Antibodies used: anti-HIF-1α IgG 07-628 (Millipore), anti-Arnt IgG H172 (Santa Cruz), anti-CAIX IgG H-120 (Santa Cruz), anti-GLUT1 IgG ab652 (Abcam), anti-phospho-Akt S473 IgG 9277 (Cell Signaling), and anti-GAPDH IgG G9549 (Sigma), anti-AhR IgG H-211 to detect AhR (Santa Cruz), and anti-AhR IgG N19 to detect CΔ553 (Santa Cruz). N19 was not used to detect AhR because it shows a nonspecific band very close to AhR and ten times higher antibody amount is necessary to detect AhR when compared with H-211. H-211 cannot detect CΔ553 since H-211 was raised against the C-terminal 211 amino acids of AhR which is absent in CΔ553.
HeLa cells were grown under various conditions in a 75 cm2 flask. Nuclear extracts were obtained (see supplementary data) and were immediately diluted to a final concentration of 0.1 M KCl. Three μg of nuclear extract was incubated with 2 μg of poly-dIdC, 100 ng of calf thymus DNA, and 40 ng of mutated HRE for 15 min at room temperature. 32P-Labelled HRE probe (100,000 cpm) was added to each sample. After 15 min at room temperature, samples were separated on a 4% native acrylamide gel at 4 °C and the dried gel was analyzed.
Unpaired two-tailed t test was performed using the Prism 5 software.
When we infected HeLa cells with the high titer Ad553 virus, an abundant amount of CΔ553 was expressed 3 days after infection (Fig. 1A). We examined whether expression of CΔ553 would have any direct effect on the HeLa cell growth. We seeded 1.5×104 cells/well in a 96-well plate in the presence or absence of adenovirus and then followed the cell growth for 3 days. Wild type cells showed proportionate growth from 12 to 60 h and then leveled off up to 72 h because cells were approaching 100% confluency at 60 h (Fig. 1B). When cells were infected with adenovirus, regardless of Ad553 or AdLacZ, cells proliferated faster from 12 to 36 h, leveled off from 36 to 60 h, and then cell death occurred from 60 to 72 h. The increased cell growth and the eventual cell death were caused by adenoviral infection but not a CΔ553 effect, since Ad553 and AdLacZ showed the same pattern. We observed a slight increase of cell proliferation at 36 h when compared the Ad553-infected cells with the AdLacZ-treated cells.
We examined whether HeLa cells expressing CΔ553 would inhibit tumor formation in nude mice. To rule out the general adenoviral effect, AdLacZ-infected HeLa cells were used to develop tumor. Uninfected and AdLacZ-infected HeLa cells formed tumor readily in nude mice; however, there was substantial difference in tumor size at day 14 when cells were infected with Ad553 prior to injection (Fig. 2A). The wild type cells showed significant tumor growth at day 7 (548 mm3 ± 92) and the tumor continued to grow up to 1,572 mm3 ± 403 at day 14 (Fig. 2B). But the Ad553-infected HeLa cells formed only a small tumor: the tumor size was < 5% of the AdLacZ tumor at day 14 (57 mm3 ± 28 and 1,208 mm3 ± 484 for Ad553 and AdLacZ tumors, respectively). At the time of harvest (day 15), the weights of wild type, Ad553, and AdLacZ tumors were 580 mg ± 68, 32 mg ± 13, and 459 mg ±150, respectively (Fig. 2C).
Most of the Ad553 tumors were too small to be excised for analysis. Nonetheless, we were able to obtain a few Ad553 tumors to perform Western analysis, which showed that protein levels of HIF-1α and Arnt were significantly less in Ad553 tumors but were present in the control (wild type and AdLacZ) tumors (Fig. 3). In addition, the HIF-1α target proteins CAIX and GLUT1 were also significantly less in Ad553 tumors when compared with the control tumors, suggesting that there was less HIF-1function in Ad553 tumors. Immunohistochemical staining data showed that CAIX and GLUT1 were expressed in control tumors but minimal, if any, in Ad553 tumors (Fig. 4). Staining using CD31 antibody showed that microvessels were more apparent in the control tumors than in Ad553 tumors. Interestingly, Akt was phosphorylated in control tumors but was noticeably less phosphorylated in Ad553 tumors (Fig. 3 and and44).
Protein levels of both HIF-1α and AhR were less in CoCl2-treated HeLa cells in the presence of Ad553 or AdLacZ, suggesting that these reductions were caused by adenoviral infection, not a CΔ553 effect (Fig. 5A and B). Interestingly, the Arnt content was pronouncedly reduced at the Ad553 dose we used to infect HeLa cells before injection (Fig. 5A, lane 10). This reduction of the cellular Arnt level was inversely proportional to the amount of the CΔ553 protein present. AdLacZ did not cause any change of the Arnt content, suggesting that CΔ553 reduced the nuclear Arnt content in HeLa cells. We further explored whether CΔ553 might affect the Arnt degradation using the proteasome inhibitor MG132. We observed that MG132 prevented the lowering of the Arnt content, suggesting that the proteasome-mediated Arnt degradation was triggered by CΔ553 (Fig. 5C). Additionally, the CΔ553 protein was also degraded by the 26S proteasome. Next, we examined whether the amount of HIF-1 formation in the HeLa nuclear extract could be reduced in the presence of CΔ553 by gel shift assay. We observed that the intensities of the HIF-1 complexes were reduced by CΔ553 in a dose-dependent manner (Fig. 5D) and this reduction was not caused by the AdLacZ control. The identities of the HIF-1 complexes were confirmed by using cold oligonucleotides: cold HRE abolished the HIF-1 complexes whereas an equal amount of mutated HRE did not. Interestingly, we observed two HIF-1 complexes which were shown by another group as well . The upper complex appeared to be more sensitive to wild type HRE than the lower complex.
Many researchers have successfully blocked the HIF-1α function based on our current knowledge of how HIF-1 signals. A dominant-negative HIF-1α construct presumably competes with HIF-1α for Arnt binding and in turn inhibits the HIF-1 transcriptional activity  and suppresses tumor growth . Similarly, an HIF-1α splice variant inhibits the HIF-1α function by interacting with Arnt . The interaction surface between HIF-1α and its coactivator p300 appears to be a good target for inhibiting the HIF-1α activity . Although the Arnt content in cell lines seems abundant , the Arnt-dependent function should be blocked if we can somehow sequester the nuclear Arnt protein. We predicted that when adequate (large) amount of CΔ553 is present in the cell nucleus, it should dimerize with Arnt and physically limit Arnt for any function. We proved that expression of an abundant amount of CΔ553 indeed inhibits the HIF-1 function.
There are opposing arguments on the effect of HIF-1α inhibition on tumor growth. In most solid tumors, overexpression of HIF-1α is a hallmark for tumor development; however, some researchers observed the contradicting results in the case of head and neck squamous cell carcinoma  and lung cancer . HIF-1α is essential for tumor growth in nude mice because the inhibition of the xenograft tumor growth was observed when (1) the HIF-1α content was suppressed by RNA interference  and (2) the HIF-1 function was inhibited by an N-terminal HIF-1α dominant negative construct . Nonetheless, it is widely known that tumors typically rely more on glycolysis to produce ATP for survival and HIF-1α promotes higher amount of ATP production through glycolysis . Many of the glycolytic enzymes and proteins involved in glucose transport are induced by HIF-1α; therefore, blockade of the HIF-1α function should limit ATP production and suppress tumor growth. Results from our xenograft experiments clearly showed that HeLa cells that expressed an abundant amount of CΔ553 lost their capability of developing into a tumor. CAIX and GLUT1 were expressed in much lesser extent in Ad553 tumors, which was consistent with our prediction that HIF-1α function was suppressed in the presence of CΔ553.
Since CΔ553 interferes with the HIF-1α and AhR function , we examine whether the protein levels of Arnt, HIF-1α and AhR could be altered in response to the CΔ553/Arnt interaction. Surprisingly the Arnt content, which is normally quite stable, was drastically reduced in the Ad553 tumors, as compared with the wild type and AdLacZ tumors. In contrast, the lesser amount of AhR in Ad553 tumors appeared to be an adenoviral effect since AdLacZ caused a similar reduction. The Arnt content in Ad553-infected HeLa cells at the time of injection was already reduced and the Arnt content remained low in Ad553 tumors at day 15. We were only able to detect CΔ553 in HeLa cells before injection but not in Ad553 tumors. Therefore, the effect of CΔ553 on the suppression of tumor growth is likely to occur at the onset of tumor growth and the sustained presence of CΔ553 is not required for the low Arnt content observed in the Ad553 tumors. Our data clearly showed that HIF-1α protein accumulation was inhibited in Ad553 tumors, although this inhibition was not observed in Ad553-infected HeLa cells. It is unclear how HIF-1α is induced in HeLa xenografts and whether the sustained low Arnt content in Ad553 tumors is a cause or an effect of the tumor suppression. Regardless, the proteasome-mediated degradation of Arnt is clearly triggered in Ad553-infected HeLa cells. More studies are necessary to explore the mechanism of the CΔ553 effect at play in these tumors.
Reduction of the phospho-Akt content by Ad553 is intriguing. Activation of the PI3K/Akt pathway increases the HIF-1α protein content via mechanisms involving transcriptional control  and Hsp70/Hsp90-dependent HIF-1α protein accumulation . Sphingosine kinase 1 activates phosphorylation of Akt at serine 473 and in turn phosphorylates GSK3β, which is responsible for HIF-1α protein accumulation via a VHL-mediated mechanism . Interestingly, phospho-Akt is reduced in AhR null mouse mammary fibroblasts which cause xenograft tumor growth . Although suppression of phospho-Akt content by Ad553 might contribute to the disappearance of HIF-1α and relate to the inhibition of AhR function in Ad553 tumors, the picture seems to be much more complex. Highest phospho-Akt content was consistently observed in uninfected wild type HeLa cells where there was no HIF-1α expression, suggesting that accumulation of the HIF-1α protein in HeLa cells could not be explained solely by the phospho-Akt action. Certainly CΔ553 might reduce the potential AhR action in angiogenesis during tumor growth . More studies are necessary to elucidate the mechanisms for CΔ553-mediated tumor suppression.
This work is supported by the National Institutes of Health (R01 ES014050).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.