Hypoxia signals autophagy in tumor cells via AMPK activity, independent of HIF-1, BNIP3, and BNIP3L.

Macroautophagy (called autophagy hereafter) is a catabolic process activated by various types of stress, most notably by nutrient deprivation. The autophagic degradation of intracellular macromolecules provides metabolic support for the cell; however, this physiological process can also initiate a form of cell death (type 2 programmed cell death). Here we report that oxygen deprivation can activate the autophagic pathway in human cancer cell lines. We observed that hypoxia induced distinct cellular changes characteristic of autophagy such as an increase in cytoplasmic acidic vesicles, and processing and cellular localization of microtubule-associated protein-1 light chain 3. Oxygen deprivation-induced autophagy did not require nutrient deprivation, hypoxia-inducible factor-1 (HIF-1) activity, or expression of the HIF-1 target gene BNIP3 (Bcl-2 adenovirus E1a nineteen kilodalton interacting protein 3) or BNIP3L (BNIP3 like protein). Hypoxia-induced autophagy involved the activity of 5'-AMP-activated protein kinase (AMPK). Finally, we determined that cells lacking the autophagy gene ATG5 were unable to activate the autophagic machinery in hypoxia, had decreased oxygen consumption and increased glucose uptake under hypoxia, had increased survival in hypoxic environments, and exhibited accelerated growth as xenografted tumors. Together, these findings suggest that the autophagic degradation of cellular macromolecules contributes to the energetic balance governed by AMPK, and that suppression of autophagy in transformed cells can increase both resistance to hypoxic stress and tumorigenicity.

Induction of activating transcription factor 3 by anoxia is independent of p53 and the hypoxic HIF signalling pathway.

Solid tumors often have an inadequate blood supply, which results in large regions that are subjected to hypoxic or anoxic stress. Hypoxia-inducible factor-1 (HIF-1) is a transcription factor that regulates much of the transcriptional response of cells to hypoxia. Activating transcription factor 3 (ATF3) is another transcription factor that responds to a variety of stresses and is often upregulated in cancer. We investigated the regulation of ATF3 by oxygen deprivation. ATF3 induction occurred most robustly under anoxia, is common, and it is not dependent on presence of HIF-1 or p53, but is sensitive to the inhibition of c-Jun NH2-terminal kinase activation and the antioxidant N-acetylcystein. ATF3 could also be induced by desferrioxamine but not by the mitochondrial poison cyanide or the nonspecific 2-oxoglutarate dioxygenase inhibitor dimethyloxalylglycine. We also show that anoxic ATF3 mRNA is more stable than normoxic mRNA providing a mechanism for this induction. Thus, this study demonstrates that the regulation of ATF3 under anoxia is independent of 2-oxoglutarate dioxygenase, HIF-1 and p53, presumably involving multiple regulatory pathways.

Genetic determinants that influence hypoxia-induced apoptosis.

The p53 tumour suppressor gene is capable of activating both death receptor and mitochondrial-signalled forms of apoptotic cell death in response to diverse stimuli. Studies have suggested that impairment of the mitochondrial-signalled Apaf/caspase 9 pathway and not the death receptor Fas pathway results in almost complete resistance to apoptotic cell death induced by a low oxygen environment. However, it is unclear how p53 signals the activation of this pathway and whether it is through already identified p53 effector genes such as the pro-apoptotic gene bax, or through novel effectors such as BNIP-3/BNIP-3L. Comparison of cell lines genetically matched at the bax, cytochrome c, apaf, caspase 9 and caspase 3 loci indicated that except for bax, all of these genes were essential for hypoxia induced apoptosis both in cell culture and in transplanted tumours. These data imply that cytochrome c plays a pivotal role in signalling cell death by apoptosis under hypoxic conditions, and that the release of cytochrome c is independent of both Bax and p53. In contrast to cytochrome c, p53 modulates the magnitude of apoptosis under hypoxic conditions, but in itself is not required for the activation of the caspase cascade.

Opposing effects of hypoxia on expression of the angiogenic inhibitor thrombospondin 1 and the angiogenic inducer vascular endothelial growth factor.

Tumor angiogenesis, the development of new blood vessels during malignant progression, is a regulated process that has both genetic and physiological controls. Physiologically, angiogenesis is stimulated by decreases in tissue oxygenation (i.e., hypoxia). We investigated the effect of hypoxia on the expression of two angiogenic factors reported to be genetically regulated by the p53 tumor suppressor gene: (a) the angiogenic inhibitor thrombospondin 1 (TSP-1); and (b) the angiogenic inducer vascular endothelial growth factor (VEGF). Analysis of rodent cells that differ in their p53 genotype (p53+/+ or p53-/-) indicated that in vitro exposure to hypoxia simultaneously suppressed TSP-1 and induced VEGF expression, regardless of the p53 genotype. On transformation of these cells with E1A and oncogenic H-ras, the basal level of TSP-1 expression was strongly diminished, whereas that of VEGF could still be induced by hypoxia. Consistent with these in vitro findings, sections of tumors derived from the transformed p53+/+ and p53-/- cells showed that VEGF protein overlapped with regions of hypoxia, whereas TSP-1 protein was below the limits of detection in tumor tissue. Using a panel of normal/immortalized and transformed human cells, it was found that the ability of hypoxia to inhibit TSP-1 expression depends on the cell type and/or the degree of transformation. In contrast, VEGF expression was induced by hypoxia in all of the human cell types examined. Together, these findings suggest that hypoxic and oncogenic signals could interact in the tumor microenvironment to inhibit TSP-1 and induce VEGF expression, promoting the switch to the angiogenic phenotype.

p53 checkpoint-defective cells are sensitive to X rays, but not hypoxia.

X-ray-induced damage leads to cell-cycle "checkpoint" arrest by p53-dependent induction of the cyclin-dependent kinase inhibitor p21 (Waf1/Cip1/Sdi1). Human tumor cells that lack this response fail to arrest after exposure to DNA-damaging agents, undergo multiple rounds of endoreduplicative DNA synthesis, and eventually commit to an apoptotic cell death. Since low oxygen tension can also induce p53 protein accumulation, and can lead to cell-cycle arrest or apoptosis, we examined the expression of p21 in tumor cells under normoxic and hypoxic conditions. In a survey of cells, mRNA for the p21 gene was induced two- to threefold in response to hypoxia in a seemingly p53-independent manner. We therefore examined genetically matched cells that differ in their p21 and p53 status for response to ionizing radiation and hypoxia. We found that both p21-deficient and p53-deficient cells exhibit an increase in chromosome instability, an increased level of apoptosis, and a failure to arrest after exposure to ionizing radiation. However, cells that lack either p21 or p53 exhibit no increase in chromosome instability or elevated apoptosis and still arrest in response to hypoxia. Thus, the mechanism responsible for the differential response to either hypoxia or X rays presumably lies in the control of cell-cycle progression in response to stress and its dependence on p21. Since the loss of a DNA-damage-dependent checkpoint does not sensitize cells to killing by stresses that elicit a DNA-damage-independent checkpoint, targeting the function of p21 pharmacologically will not kill tumor cells in situ in the absence of a DNA damage signal.

Candidate genes for the hypoxic tumor phenotype.

In this study, we have analyzed changes induced by hypoxia at the transcriptional level of genes that could be responsible for a more aggressive phenotype. Using a series of DNA array membranes, we identified a group of hypoxia-induced genes that included plasminogen activator inhibitor-1 (PAI-1), insulin-like growth factor-binding protein 3 (IGFBP-3), endothelin-2, low-density lipoprotein receptor-related protein (LRP), BCL2-interacting killer (BIK), migration-inhibitory factor (MIF), matrix metalloproteinase-13 (MMP-13), fibroblast growth factor-3 (FGF-3), GADD45, and vascular endothelial growth factor (VEGF). The induction of each gene was confirmed by Northern blot analysis in two different squamous cell carcinoma-derived cell lines. We also analyzed the kinetics of PAI-1 induction by hypoxia in more detail because it is a secreted protein that may serve as a useful molecular marker of hypoxia. On exposure to hypoxia, there was a gradual increase in PAI-1 mRNA between 2 and 24 h of hypoxia followed by a rapid decay after 2 h of reoxygenation. PAI-1 levels were also measured in the serum of a small group of head and neck cancer patients and were found to correlate with the degree of tumor hypoxia found in these patients.

Epigenetic regulation of gene expression in cervical cancer cells by the tumor microenvironment.

Evidence is accumulating that the adverse tumor microenvironment both modifies the malignant progression of tumor cells and contributes to chemotherapy and radiation resistance. We hypothesized that some of the effects on malignant progression are mediated through the transcriptional regulation of genes responsive to the stresses of the microenvironment, such as low oxygen or low glucose conditions. To determine epigenetic changes in gene expression that were consistent with that hypothesis, we used an in vitro subtractive hybridization method, representational difference analysis, to identify hypoxia-induced cDNAs from cultured human cervical epithelial cells. We identified 12 induced genes: two novel genes (HIG1 and HIG2), three genes known to be hypoxia-inducible (tissue factor, GAPDH, thioredoxin), and seven genes not previously identified as hypoxia-inducible [HNRNP(a1), ribosomal L7, annexin V, lipocortin 2, Ku(70), PRPP synthase, and acetoacetyl-CoA thiolase]. In cultured cells, HIG1 and HIG2 expression is induced by hypoxia and by glucose deprivation, but their expression is not induced by serum deprivation, UV, or ionizing radiation. The putative HIG1 and HIG2 open reading frames are expressed in cells, as confirmed by epitope tagging. In addition, tumor xenografts derived from human cervical cancer cells display increased expression of HIG1 and HIG2 when they are deprived of oxygen. Taken together, these data suggest a coordinated transcriptional response of eukaryotic cells to microenvironmental stresses found in the solid tumor.

A p53 and apoptotic independent role for p21waf1 in tumour response to radiation therapy.

Loss of p21 in human cancer cells results in checkpoint failure, induction of polyploidy and subsequent apoptosis following DNA damage. Tumours in immunodeficient mice derived from cells lacking p21 are also more sensitive to ionizing radiation than their wild-type counterparts. Abrogation of p53 in the p21+/+ parental cells results in an in vitro phenotype that is indistinguishable from that of the p21 knockout cells. Thus, the in vitro phenotype resulting from loss of p21 is consistent with its well-established role in the p53/p21 damage response pathway. However, despite the similar in vitro phenotype, p21+/+ cells with abrogated p53 show no evidence of the sensitivity observed in the p21-/- cells when grown as tumours in immunodeficient mice. The increased radio-sensitization stabilization of p21-/- tumours is also unrelated to the increase in apoptosis observed in these tumours following radiation treatment. Apoptosis in the p21-/- tumours was significantly reduced by expression of bcl-2 without any corresponding change in the overall response of the tumour. Similarly, abrogation of p53 in the p21+/+ tumours substantially increased radiation-induced apoptosis within the tumours without increasing their radiation sensitivity. Dissociation of these in vivo and in vitro phenotypes indicates that p21 participates in a novel in vivo specific damage response pathway that is distinct from its role in the p53 pathway, and therefore that it may be an effective therapeutic target for cancer therapy.

Highly conserved RNA sequences that are sensors of environmental stress.

The putative function of highly conserved regions (HCRs) within 3' untranslated regions (3'UTRs) as regulatory RNA sequences was efficiently and quantitatively assessed by using modular retroviral vectors. This strategy led to the identification of HCRs that alter gene expression in response to oxidative or mitogenic stress. Databases were screened for UTR sequences of >100 nucleotides that had retained 70% identity over more than 300 million years of evolution. The effects of 10 such HCRs on a standard reporter mRNA or protein were studied. To this end, we developed a modular retroviral vector that can allow for a direct comparison of the effects of different HCRs on gene expression independent of their gene-intrinsic 5'UTR, promoter, protein coding region, or poly(A) sequence. Five of the HCRs tested decreased mRNA steady-state levels 2- to 10-fold relative to controls, presumably by altering mRNA stability. One HCR increased translation, and one decreased translation. Elevated mitogen levels caused four HCRs to increase protein levels twofold. One HCR increased protein levels fourfold in response to hypoxia. Although nonconserved UTR sequences may also have a role, these results provide evidence that sequences that are highly conserved during evolution are good candidates for RNA motifs with posttranscriptional regulatory functions in gene expression.

Loss of p21Waf1/Cip1 sensitizes tumors to radiation by an apoptosis-independent mechanism.

Cellular checkpoints are important mediators of the response of normal cells following genotoxic damage, and interruption of these checkpoints is a common feature of many solid tumors. Although the effects of loss in checkpoint function in tumor cells are well understood in terms of cell cycle control, there is little information on their role in determining treatment efficacy in vivo. We have examined both the in vitro and in vivo responses of isogenic lines differing only in the p53-transactivated checkpoint gene, p21Waf1/Cip1. When assayed in vitro, loss of p21 in human colon tumor cells results in a selective induction of apoptosis [Waldman, T., et al., Nature (Lond.), 381: 713-716, 1996.] but no difference in the clonogenic survival. However, when grown as xenografts and irradiated in situ, p21-deficient tumors were significantly more sensitive to radiation as assessed both by clonogenic survival and by regrowth of the tumors following treatment. These data indicate that loss of p21 results in increased sensitivity to killing by ionizing radiation that is independent of the induction of apoptosis and cell cycle arrest but that is specific to cells when they are grown as a solid tumor. These results have important implications for assessing both the genetic determinants of sensitivity to anticancer agents and efficacy of anticancer agents.

Uncoupling of S-phase and mitosis by recombinant cytotoxic necrotizing factor 2 (CNF2).

Cytotoxic necrotizing factor 2 (CNF2) is an exotoxin identified from virulent clinical isolates of Escherichia coli. It has been characterized in adherent cell lines as an inducer of cellular death, hyperploidy (multinucleation), and cytoskeletal reorganization. The molecular mechanism of these actions is unclear. Two cellular mechanisms can be hypothesized to explain the DNA content increase (hyperploidy) induced by the toxin. The first is that the toxin interferes with cytoplasmic division without interfering with normal nuclear cycling, such that DNA is replicated in the absence of cell division. The second is that the toxin drives the nuclear machinery to replicate the DNA multiple times within one cell cycle, without interfering with cytoplasmic division. In order to investigate these phenomena, we have constructed a recombinant CNF2 gene that expresses a toxin with both an epitope tag and a polyhistidine tag. Extracts made from E. coli that express this gene have a high multinucleating activity that colocalizes with the recombinant 115-kDa protein. To distinguish between these hypotheses, we used recombinant CNF2 and several growth conditions (time, partial differentiation, and stage of growth) to establish a relationship between cellular divisions and generation of hyperploidy. It was also determined that the toxin had no effect upon in vitro DNA replication using a Xenopus egg extract system. In aggregate, these data are consistent with the hypothesis that CNF2 is affecting cytoplasmic division and thereby removing the requirement for a completed mitosis before the initiation of another S-phase. These data are discussed in relation to the generation of polyploid cells during megakaryopoeisis and the generation of aneuploid cells during tumorigenesis.

Expression of Rap 1 suppresses genomic instability of H-ras transformed mouse fibroblasts.

Among the multiple genetic changes that occur during cancer progression are the activation of proto-oncogenes and the inactivation or loss of genes encoding tumor suppressors. The potential roles for these genes in the perturbation of genome stability continues to be of major interest. We have previously shown that conditional expression of H-ras in NIH3T3 cells increases genetic instability in these cells, rendering them more permissive to gene amplification and to the generation of chromosome aberrations which can be induced within a single cell cycle. In the present study we show that genetic instability induced by H-ras expression can be suppressed by co-expression of Rap 1, a Ras-related tumor suppressor gene. An NIH3T3 cell line transformed with activated human H-ras was transfected with Rap 1. Expression of the Rap 1 gene reverted the transformed cells to a flat morphology. The reverted cells reestablished contact inhibition of growth and lost the capacity to form colonies in soft agar. These cells were subsequently studied for the role of Rap 1 on the suppression of genomic instability induced by oncogenic H-ras. Cells transformed with H-ras manifest an increase in methotrexate resistance as measured by an increase in Dhfr gene amplification. Cells which concommitantly express Rap 1 showed reduced levels of methotrexate resistance as well as reduction of gene amplification capacity. Furthermore fluorescent-in-situ hybridization (FISH) with a pancentromeric mouse probe showed that elevated levels of chromosome aberrations in cells expressing H-ras were also suppressed after co-expression of Rap 1.

Protease inhibitor TPCK represses Ha-ras (Val12) transformation and nuclear factor-kappa B activation.

Certain chymotrypsin-like protease inhibitors such as TPCK exhibit a well described anti-tumorigenic activity by an as yet undescribed mechanism. One potential cellular target for TPCK in transformed cells is the ms-inducible NF-kappa B family of transcription factors. We therefore used TPCK to examine the physiologic role of NF-kappa B during Ha-ras induced transformation, independent of another major downstream effector of Ha-ras, AP-1. Using a conditionally transformed NIH3T3 cell line, we found that TPCK (but not the control inhibitor TLME) inhibited the anchorage-independent growth of Ha-ras transformed cells, but not their anchorage-dependent growth on plastic tissue culture dishes. Likewise, TPCK reduced the ability of Ha-ras to stimulate DNA synthesis in growth factor depleted cells, but not the ability of serum to stimulate DNA synthesis in the same growth factor depleted cells. Gel shift analysis and reporter gene expression indicated that TPCK blocked Ha-ras-induced NF-kappa B activity, while only having minimal effects on Ha-ras-induced AP-1 activity. TPCK is therefore able to Inhibit the Ha-ras transformed phenotype of cells by inhibiting the transcriptional activity of NF-kappa B, while having little effect upon transcriptional activity of AP-1.

Megakaryocytic cell line-specific hyperploidy by cytotoxic necrotizing factor bacterial toxins.

Cytotoxic necrotizing factor (CNF) toxins, isolated from certain Escherichia coli strains known to cause intestinal and extra intestinal infections, induce reorganization of the actin cytoskeleton and generate hyperploidy in adherent cell lines. We have examined the effect of CNF toxin on one of the few cell types that naturally increase nuclear DNA content, megakaryocytes. Our studies show that only hematopoietic cells capable of differentiating along the megakaryocyte lineage responded to the CNF2 toxin by becoming polyploid and by reorganizing actin. The K562, HEL, and CHRF-288-11 cell lines can be induced with phorbol ester to differentiate along the megakaryocyte lineage, and these cells also respond to the toxin with increased DNA content and actin cytoskeletal rearrangements. Interestingly, treatment of the K562 and HEL cell lines with CNF2 does not result in an increase in production of the megakaryocytic marker glycoprotein IIIa, unlike phorbol ester treatment. Conversely, two T-cell leukemic cell lines, CEM and Molt4, and the promyelocytic HL-60 cell line, which do not differentiate along the megakaryocyte lineage in response to phorbol myristate acetate, do not respond to CNF2, by increased expression of gpIIIa, increased nuclear DNA content, or actin reorganization. A potential target of these toxins, RhoA, is expressed by both megakaryocytic and nonmegakaryocytic cell lines, as shown by reverse transcription-polymerase chain reaction and Western blot. Although it is clear that the CNF toxins can affect a wide variety of adherent nonhematopoietic cell lines, we propose that the response to CNF, in terms of reorganizing actin structure and increase in DNA content in hematologic suspension cells, correlates with the capability of these target cells to differentiate along the megakaryocytic lineage.

Mitotic and post mitotic consequences of genomic instability induced by oncogenic Ha-ras.

Induced expression of a mutant human Ha-ras oncogene in NIH3T3 cells leads to the rapid production of multicentric chromosomes, acentric chromosome fragments, double minute chromosomes, increased heteroploidy, and increased capacity to undergo gene amplification. In this study we have used fluorescent-in-situ hybridization (FISH) to demonstrate that induction of the Ha-ras oncogene also leads to disruption of the mitotic machinery, resulting in aberrant mitoses and abnormal daughter cells. Cells induced to express an oncogenic Ha-ras transgene accumulate chromosomes that lag outside of the rest of the chromosomal architecture, chromosomes that form bridges between daughter nuclei at anaphase, and that form micronuclei. Many of these mitotic aberrations contain structurally abnormal chromosomes. These ras-induced changes were suppressed by the introduction of a gene encoding the dominant negative effector of ras, raf 301. Expression of raf301 in cells induced to express Ha-ras reduced the level of growth in soft agar, chromosome aberrations, mitotic aberrations, and frequency of gene amplification. These data provide evidence for an association between Ha-ras induced transformation, chromosome aberrations and gene amplification. Furthermore they offer insight into how the cell responds to the formation of aberrant chromosomes, and how disrupting chromosomal architecture could lead to further imbalances in the distribution of genetic material between daughter cells.

Hypoxic activation of nuclear factor-kappa B is mediated by a Ras and Raf signaling pathway and does not involve MAP kinase (ERK1 or ERK2).

We have previously shown that hypoxia causes the activation of nuclear factor-kappa B (NF-kappa B), and the phosphorylation of its inhibitory subunit, I kappa B alpha, on tyrosine residues. With the use of dominant negative mutants of Ha-Ras and Raf-1, we investigated some of the early signaling events leading to the activation of NF-kappa B by hypoxia. Both dominant negative alleles of Ha-Ras and Raf-1 inhibited NF-kappa B induction by hypoxia, suggesting that the hypoxia-induced pathway of NF-kappa B induction is dependent on Ras and Raf-1 kinase activity. Furthermore, although conditions of low oxygen can also activate mitogen-activated protein kinases (ERK1 and ERK2), these kinases do not appear to be involved in regulating NF-kappa B by low oxygen conditions, as dominant negative mutants of mitogen-activated protein kinase do not inhibit NF-kappa B activation by hypoxia. Since Ras and Raf-1 have been previously shown to work downstream from membrane-associated tyrosine kinases such as Src, we determined if the Src membrane-associated kinase was also activated by low oxygen conditions. We detected an increase in Src proto-oncogene activity within 15-30 min of cellular exposure to hypoxia. We postulate that Src activation by hypoxia may be one of the earliest events that precedes Ras activation in the signaling cascade which ultimately leads to the phosphorylation and dissociation of the inhibitory subunit of NF-kappa B, I kappa B alpha.

The human Ha-ras oncogene induces genomic instability in murine fibroblasts within one cell cycle.

Many human tumors contain an activating mutation in one of the ras protooncogenes. Additionally, these tumor cells are often heteroploid and characterized by chromosome breaks and rearrangements that are consequences of the genomic instability that is thought to contribute to tumor progression. The concurrence of ras mutations and genomic instability in tumors prompted us to ask whether selective induction of an activated Ha-ras gene could render a genome unstable. The NIH 3T3 cells used in this study contained mutant p53 genes and carried a selectively inducible activated (EJ) Ha-ras transgene under the control of bacterial lactose regulatory elements. When stably transfected cells were induced to express activated Ha-ras by isopropyl beta-D-thiogalactoside administration, there was a marked increase in the number of gross chromosomal aberrations including acentric fragments, multicentric chromosomes, and double minutes, which occurred within the time frame of a single cell cycle from the time of induction. To confirm that these aberrations occurred within the first cell cycle after mutant Ha-ras induction, the cells were arrested in G1 phase by serum depletion and, subsequently, released by administration of isopropyl beta-D-thiogalactoside or serum. The mitoses from cells released with isopropyl beta-D-thiogalactoside contained a 3-fold elevation in the fraction of chromosomes containing aberrations compared to mitoses from parallel cell cultures that were released with serum. Thus, the induction of activated Ha-ras gene expression in these cells results in genomic instability that can be detected as aberrant chromosomes at the next mitosis.

A 55 kDa antigen of Pneumocystis carinii: analysis of the cellular immune response and characterization of the gene.

Rat-derived Pneumocystis carinii contains a major antigen complex of 45-55 kDa. The fusion protein of a cDNA encoding the 3' portion of the 55kDa antigen, which had previously been shown to be recognized by serum antibodies of exposed subjects, was investigated for its ability to stimulate a cellular immune response. Rats exposed to P. carinii via the environment exhibited a vigorous proliferative response to the antigen whereas unexposed rats did not. The full-length cDNA for a 55kDa antigen was cloned and found to contain a 1245bp open reading frame capable of encoding a 414-amino-acid peptide. The gene encoding this protein contained a single 39bp intron and transcribed a 1.45kb RNA message. The cloning and characterization of the 55kDa antigen gene will allow production of the specific immunological reagents necessary to characterize this molecule and study its role in the biology and pathogenesis of P. carinii.