BackgroundAccumulating evidence indicates that the long non-coding RNA HOTAIR plays a critical role in cancer progression and metastasis. However, the overall biological role and clinical significance of HOTAIR in gastric carcinogenesis remains largely unknown.MethodsHOTAIR expression was measured in 78 paired cancerous and noncancerous tissue samples by real-time PCR. The effects of HOTAIR on gastric cancer cells were studied by overexpression and RNA interference approaches in vitro and in vivo. Insights of the mechanism of competitive endogenous RNAs (ceRNAs) were gained from bioinformatic analysis, luciferase assays and RNA binding protein immunoprecipitation (RIP). The positive HOTAIR/HER2 interaction was identified and verified by immunohistochemistry assay and bivariate correlation analysis.ResultsHOTAIR upregulation was associated with larger tumor size, advanced pathological stage and extensive metastasis, and also correlated with shorter overall survival of gastric cancer patients. Furthermore, HOTAIR overexpression promoted the proliferation, migration and invasion of gastric carcinoma cells, while HOTAIR depletion inhibited both cell invasion and cell viability, and induced growth arrest in vitro and in vivo. In particular, HOTAIR may act as a ceRNA, effectively becoming a sink for miR-331-3p, thereby modulating the derepression of HER2 and imposing an additional level of post-transcriptional regulation. Finally, the positive HOTAIR/HER2 correlation was significantly associated with advanced gastric cancers.ConclusionsHOTAIR overexpression represents a biomarker of poor prognosis in gastric cancer, and may confer malignant phenotype to tumor cells. The ceRNA regulatory network involving HOTAIR and the positive interaction between HOTAIR and HER2 may contribute to a better understanding of gastric cancer pathogenesis and facilitate the development of lncRNA-directed diagnostics and therapeutics against this disease.

Recent evidence highlights long noncoding RNAs (lncRNA) as crucial regulators of cancer biology that contribute to essential cancer cell functions such as cell proliferation, apoptosis, and metastasis. In non-small cell lung cancer (NSCLC), several lncRNAs' expressions are misregulated and have been nominated as critical actors in NSCLC tumorigenesis. LncRNA ANRIL was first found to be required for the PRC2 recruitment to and silencing of p15 INK4B , the expression of which is induced by the ATM-E2F1 signaling pathway. Our previous study showed that ANRIL was significantly upregulated in gastric cancer, and it could promote cell proliferation and inhibit cell apoptosis by silencing of miR99a and miR449a transcription. However, its clinical significance and potential role in NSCLC is still not documented. In this study, we reported that ANRIL expression was increased in NSCLC tissues, and its expression level was significantly correlated with tumor-node-metastasis stages and tumor size. Moreover, patients with high levels of ANRIL expression had a relatively poor prognosis. In addition, taking advantage of loss-of-function experiments in NSCLC cells, we found that knockdown of ANRIL expression could impair cell proliferation and induce cell apoptosis both in vitro and vivo. Furthermore, we uncover that ANRIL could not repress p15 expression in PC9 cells, but through silencing of KLF2 and P21 transcription. Thus, we conclusively demonstrate that lncRNA ANRIL plays a key role in NSCLC development by associating its expression with survival in patients with NSCLC, providing novel insights on the function of lncRNA-driven tumorigenesis.

Recently, the non-protein-coding functional elements in the human genome have been identified as key regulators in postgenomic biology, and a large number of pseudogenes as well as long non-coding RNAs (lncRNAs) have been found to be transcribed in multiple human cancers. However, only a small proportion of these pseudogenes has been functionally characterized. In this study, we screened for pseudogenes associated with human non-small-cell lung cancer (NSCLC) by comparative analysis of several independent datasets from the GEO. We identified a transcribed pseudogene named DUXAP8 that is upregulated in tumor tissues. Patients with higher DUXAP8 expression exhibited shorter survival, suggesting DUXAP8 as a new candidate prognostic marker for NSCLC patients. Knockdown of DUXAP8 impairs cell growth, migration, and invasion, and induces apoptosis both in vitro and in vivo. Mechanistically, DUXAP8 represses the tumor suppressors EGR1 and RHOB by recruiting histone demethylase LSD1 and histone methyltransferase EZH2, thereby promoting cell proliferation, migration, and invasion. These findings indicate that the pseudogene DUXAP8 may act as an oncogene in NSCLC by silencing EGR1 and RHOB transcription by binding with EZH2 and LSD1, which may offer a novel therapeutic target for this disease.

The present study examined the relationship between an important energy-generating enzyme (cytochrome oxidase; CO), a key energy-consuming enzyme (Na+ K+ ATPase) and neurochemicals associated with excitatory glutamatergic synapses (NMDAR1 and neuronal nitric oxide synthase, nNOS) in the adult macaque retina. Polyclonal antibodies against neuronal nitric oxide synthase and N-methyl-D-aspartate receptor subunit I were generated for immunohistochemical examination and labeled sites not previously reported were found. We have also isolated cDNAs for cytochrome oxidase subunits III (mitochondrial-encoded) and IV (nuclear-encoded), as well as for a fragment of neuronal nitric oxide synthase, from a human cDNA library. The distributions of mRNAs of these genes were analyzed by in situ hybridization. We found that three or more of the markers examined coexisted in a number of sites: (a) In the inner segments of photoreceptors, high energy demand for maintaining the dark current was placed by Na+ K+ ATPase. This was partially met by ATP-generating enzymes such as CO. Neuronal NOS was also present there for the synthesis of NO and the cascading event leading to the generation of cGMP and the gating of channels for visual transduction. (b) Both the outer and inner plexiform layers had detectable amounts of all four markers, although the levels varied among them. This was most likely due to the presence of depolarizing glutamatergic synapses arising from photoreceptors and bipolar cells and such synaptic events were energy-demanding. The involvement of NMDA receptors and nNOS in these synaptic layers is strongly implicated in the present study. (c) All four markers were present in the majority of retinal ganglion cells, with some inherent heterogeneity related to intensity and size. Retinal ganglion cells are known to receive excitatory synapses from glutamatergic bipolar cells and are themselves highly active. The presence of both NMDAR1 and nNOS in these cells were verified in the present study and the energy demands related to these synaptic activities were necessarily high. Thus, active ion transporting functions related to synaptic or non-synaptically induced repolarization from the basis for an interrelationship between the neurochemicals/enzymes studied. Finally, (d) all four markers and the gene expression of CO and nNOS in the macaque retina were regulated by neuronal activity.

Previous studies have shown that a transcription factor of the Ets family, nuclear respiratory factor 2 (NRF‐2), can activate in vitro the gene expression of cytochrome oxidase (CO), a mitochondrial enzyme of oxidative metabolism. The goals of our present study were to determine whether the distribution of NRF‐2 α subunit proteins correlated with that of CO activity in the macaque monkey visual cortex and whether the level could be perturbed by visual deprivation. We generated polyclonal antibodies specifically against human NRF‐2 α subunit. In normal monkeys, patterns of NRF‐2 α distribution resembled closely that of CO activity: 1) NRF‐2 α immunoreactivity was localized in both nuclei and cytoplasm of neurons, but the levels differed among various laminae; 2) layers IVA, IVC, and VI, which had high CO activity, were labeled more densely by NRF‐2 α than layers I, IVB, and V, which contained lower levels of both NRF‐2 α and CO activity; and 3) CO‐rich puffs in layers II and III contained a higher level of NRF‐2 α than CO‐poor interpuffs. From 1 day to 7 days after monocular impulse blockade with tetrodotoxin, there was a progressive reduction of NRF‐2 α in deprived ocular dominance columns, in parallel with decreases in CO activity. These results suggest that local levels of NRF‐2 in the monkey visual cortex closely reflect neuronal physiological and metabolic levels revealed by CO activity and that the expression of NRF‐2 α, like that of CO, is regulated tightly by neural functional activity. J. Comp. Neurol. 404:310–320, 1999. © 1999 Wiley‐Liss, Inc.

In the primate striate cortex, cytochrome oxidase (CO)-rich puffs differ from CO-poor interpuffs in their metabolic levels and physiological properties. The neurochemical basis for their metabolic and physiological differences is not well understood. The goal of the present study was to examine the relationship between the distribution of gamma aminobutyric acid (GABA)/non-GABA synapses and CO levels in postsynaptic neuronal profiles and to determine whether or not a difference existed between puffs and interpuffs. By combining CO histochemistry and postembedding GABA immunocytochemistry on the same ultrathin sections, the simultaneous distribution of the two markers in individual neuronal profiles was quantitatively analyzed. In both puffs and interpuffs, GABA-immunoreactive (GABA-IR) neurons were the only cell type that received both non-GABA-IR (presumed excitatory) and GABA-IR (presumed inhibitory) axosomatic synapses, and they had three times as many mitochondria darkly reactive for CO than non-GABA-IR neurons, which received only GABA-IR axosomatic synapses. GABA-IR neurons and terminals in puffs had a larger mean size, about twice as many darkly reactive mitochondria, and a higher ratio of non-GABA-IR to GABA-IR axosomatic synapses than those in interpuffs (2.3:1 vs. 1.6:1; P < 0.01). There were significantly more synapses of both non-GABA-IR and GABA-IR types in the neuropil of puffs than of interpuffs; however, the ratio of non-GABA-IR to GABA-IR synapses was significantly higher in puffs (2.86:1) than in interpuffs (2.08:1; P < 0.01). Our results are consistent with the hypothesis that the level of oxidative metabolism in postsynaptic neurons and neuronal processes is tightly governed by the strength and proportion of excitatory over inhibitory synapses. Thus, the present results suggest that (1) GABA-IR neurons in the macaque striate cortex have a higher level of oxidative metabolism than non-GABA ones because their somata receive direct excitatory synapses and their terminals are more tonically active; (2) the higher proportion of presumed excitatory synapses in puffs imposes a greater energy demand there than in interpuffs; and (3) excitatory synaptic activity may be more prominent in puffs than in interpuffs because puffs receive a greater proportion of excitatory synapses from multiple sources including the lateral geniculate nucleus, which is not known to project to the interpuffs.

One of the hallmarks of the primate striate cortex is the presence of cytochrome oxidase (CO)-rich puffs and CO-poor interpuffs in its supragranular layers. However, the neurochemical basis for their differences in metabolic activity and physiological properties is not well understood. The goals of the present study were to determine whether CO levels in postsynaptic neuronal compartments were correlated with the proportion of excitatory glutamate-immunoreactive (Glu-IR) synapses they received and if Glu-IR terminals and synapses in puffs differed from those in interpuffs. By combining CO histochemistry and postembedding Glu immunocytochemistry on the same ultrathin sections, the simultaneous distribution of the two markers in individual neuronal profiles was quantitatively analyzed. As a comparison, adjacent sections were identically processed for the double labeling of CO and GABA, an inhibitory neurotransmitter. In both puffs and interpuffs, most axon terminals forming asymmetric synapses (84%)--but not symmetric ones, which were GABA-IR--were intensely immunoreactive for Glu. GABA-IR neurons received mainly Glu-IR synapses on their cell bodies, and they had three times as many mitochondria darkly reactive for CO than Glu-rich neurons, which received only GABA-IR axosomatic synapses. In puffs, GABA-IR neurons received a significantly higher ratio of Glu-IR to GABA-IR axosomatic synapses and contained about twice as many darkly CO-reactive mitochondria than those in interpuffs. There were significantly more Glu-IR synapses and a higher ratio of Glu- to GABA-IR synapses in the neuropil of puffs than of interpuffs. Moreover, Glu-IR axon terminals in puffs contained approximately three times more darkly CO-reactive mitochondria than those in interpuffs, suggesting that the former may be synaptically more active. Thus, the present results are consistent with our hypothesis that the levels of oxidative metabolism in postsynaptic neurons and neuropil are positively correlated with the proportion of excitatory synapses they receive. Our findings also suggest that excitatory synaptic activity may be more prominent in puffs than in interpuffs, and that the neurochemical and synaptic differences may constitute one of the bases for physiological and functional diversities between the two regions.

Cytochrome oxidase (CO), a mitochondrial energy-generating enzyme, contains both mitochondrial- and nuclear-encoded subunits. In neurons, local levels of CO activity vary among different neuronal compartments, reflecting local demands for energy. The goals of the present study were to determine if compartmental distribution of CO subunit proteins from the two genomes was correlated with local CO activity, and if their expression was regulated proportionately in neurons. The subcellular distributions of mitochondrial-encoded CO III and nuclear-encoded CO Vb proteins were quantitatively analyzed in mouse cerebellar sections subjected to postembedding immunocytochemistry. Local levels of subunit proteins were also compared to local CO activity, as revealed by CO cytochemistry. In order to study the regulation of subunit protein expression, we assessed changes in immunoreactivity of the two CO subunits as well as changes in CO activity in mouse superior colliculus after 1 to 7 days of monocular enucleation. We found that immunoreaction product for both CO III and CO Vb existed almost exclusively in mitochondria, but their compartmental distributions were different. CO III was nonhomogeneously distributed among different neuronal compartments, where its local level was positively correlated with that of CO activity. In contrast, the subcellular distribution of CO Vb was relatively uniform and did not bear a direct relationship with that of CO activity. Moreover, the two subunit proteins were disproportionately regulated by neuronal activity. CO III and CO activity exhibited parallel decreases after the deprivation of afferent input, and their changes were earlier and to a greater degree than that of CO Vb proteins. Thus, the present findings indicate that the local expression and/or distribution of CO subunit proteins from the two genomes may involve different regulatory mechanisms in neurons. Our data also suggest that the activity-dependent regulation of mitochondrial-encoded CO subunits is likely to play a major role in controlling the local levels of CO content and its activity.