Long non-coding RNAs: evolution of new epigenetic and post-transcriptional functions

Staufen1-mediated mRNA decay (SMD), which occurs when translation terminates sufficiently upstream of a STAU-binding site (SBS), is important to developmental and homeostatic pathways. An SBS can be created by intramolecular base-pairing within an mRNA 3'UTR or by intermolecular base-pairing between a 3'UTR and one or more lncRNAs. Intermolecular base-pairing in humans involves Alu elements, which are a type of short interspersed repetitive element (SINE), whereas intermolecular base-pairing in rodents involves B and identifier SINEs. The STAU1 paralog STAU2, and STAU1 and STAU2 homodimerization and heterodimerization also promote SMD. In addition to mRNA−lncRNA interactions, they have found that mRNAs crosstalk in a way that involves direct mRNA−mRNA interactions between 3'UTR Alu elements in each mRNA, uncovering a new role for mammalian-cell mRNAs. This unexpected function, together with them discovering how STAU1 binding to inverted repeated 3'UTR Alu elements (IRAlus) (i) competes with nuclear (i.e., paraspeckle) retention mediated by p54^nrb binding to 3'UTR IRAlus and also (ii) the repression of cytoplasmic translation mediated by PKR binding to 3'UTR IRAlus add new layers of complexity to the network of post-transcriptional interactions that regulate gene expression and involve ncRNA. They are in the process of purifying individual Alu element-containing lncRNAs together with the associated transcripts (both mRNAs and other lncRNAs) and proteins to gain insight into lncRNA structure and function.

To gain insight into the complex cellular roles performed by long non-coding RNAs, Dr Valadkhan analyzed the function of an intergenic lncRNA, BORG, which is almost exclusively expressed in neuronal cells in both mouse and human. Study of the function of BORG indicated that it is required for neuronal differentiation in mouse cell culture-based systems. In addition, they show that BORG plays a critical role in the stress response that was distinct from its neuronal differentiation function. Knock down and overexpression studies indicated a strong, positive correlation between the expression level of BORG and post-stress survival in cultured mammalian cells. In order to determine if BORG plays a role similar to what we have observed in cultured cells in vivo, they developed a transgenic mouse model that overexpresses BORG from a tet-regulated promoter. Induction of expression of BORG from the transgene in postnatal mice resulted in potentiation of the stress response, similar to what we had observed in cell culture-based systems. Despite inducing an increase in pro-survival factors, overexpression of BORG in vivo does not seem to be tumorigenic. Their observations indicate that both in vivo and in cultured cells, BORG RNA is a strong regulator of the stress response and suggest that the pro-survival function of BORG can have a protective effect against neurodegeneration and other conditions characterized by stress-mediated cell death.

Many genome-wide studies of long non-coding RNAs (lncRNAs) have been performed; however, few structural studies of individual lncRNAs exist. A clear consensus has not been achieved on the level of structure occurring in lncRNA systems. Structures are necessary not only to establish a mechanistic framework for lncRNAs, but also to establish the level of conservation of each lncRNA, since this cannot be achieved from sequence alone. Historically, secondary structures of the 16S and 23S rRNA provided an important framework for biochemical investigations and for structural studies of the ribosome, which is a 5-10 kB non-coding RNA, depending on species. Dr Sanbonmatsu establishes similar structural frameworks by determining secondary structures of individual lncRNA systems. In the case of smaller RNA systems, chemical probing has proved an instrumental methodology for directly determining secondary structures. In the case of larger systems, chemical probing alone has significant limitations due to the exponentially large number of possible secondary folds. To address this challenge, they developed Shotgun Secondary Structure (3S), a divide-and-conquer experimental strategy to derive secondary structures of lncRNAs. The 3S experimental strategy allows us to produce the secondary fold with minimal need for computational predictions. Here, they chemically probe the entire lncRNA transcript. Next, they repeat the probing on several overlapping fragments on the transcript. If the SHAPE profile for the fragment matches the profile for the same region of RNA in the context of the full transcript, this strongly suggests the fragment folds into a modular sub-domain or modular secondary motif, eliminating a large number of other possible folds. They applied this method to the Coolair lncRNA in A. thaliana and Braveheart in mouse, both of which possess clear phenotypes. They also applied the method to the human steroid receptor RNA activator (SRA-1). Coolair, Braveheart and SRA-1 each show highly structured sub-domains and share a unique expansive internal loop (‘right hand turn’ motif). They then use the secondary structure to find examples of the same lncRNA in other species, demonstrating conservation of structure.

The majority of the human genome is transcribed into non-protein-coding RNA. Hence, RNA is also the primary product of the cancer genome. Dr Diederichs lab have defined the ncRNA expression landscape of lung, breast and liver cancer providing a comprehensive expression map of over 17000 long ncRNAs and discovering new lncRNAs associated with cancer whose molecular and cellular functions they are currently elucidating. The nuclear lncRNA MALAT1 was one of the first lncRNAs associated with cancer: it is associated with metastasis development in lung cancer. However, its high abundance and nuclear localization have hampered its functional analysis. To uncover its functional importance, they developed a MALAT1 knockout model in human lung tumor cells by genomically integrating RNA destabilizing elements site-specifically into the MALAT1 locus. This approach yielded a 1000-fold silencing of MALAT1 providing a unique loss-of-function model. Proposed mechanisms of MALAT1 action include regulation of splicing or gene expression. In lung cancer, MALAT1 does not alter alternative splicing but regulates gene expression inducing a signature of metastasis-associated genes. Consequently, MALAT1-deficient cells are impaired in migration and form fewer tumor nodules in a mouse xenograft model. Encouraged by this discovery of the essential function of MALAT1 in lung cancer metastasis, they analyzed whether MALAT1 could serve as therapeutic target for an Antisense oligonucleotide (ASO). Notably, MALAT1-ASO treatment prevented metastasis formation after tumor implantation. Thus, targeting MALAT1 with ASOs provided a therapeutic approach to prevent lung cancer metastasis with MALAT1 serving as both, predictive marker and therapeutic target. In summary, ten years after the discovery of the lncRNA MALAT1 as a biomarker for lung cancer metastasis, their loss-of-function model unraveled the active function of MALAT1 as a regulator of gene expression governing hallmarks of lung cancer metastasis.

A crucial role of mitochondria is to provide the universal currency for energy. This task forces them to be in tight communication with the rest of the cell machinery, and more specifically with the nucleus, which represents the control center. Connections take place not only through the cytoskeleton but also through intense genomic cross-talk with the nucleus, which provides mitochondria with nuclear-encoded proteins as well as non-coding RNAs that are required for mitochondrial homeostasis and function. Conversely, both strands of the mitochondrial DNA are entirely transcribed that generate also some non-coding sequences. Although information about mitochondrial long non-coding RNAs (lncRNAs) per se is still limited, candidates non-coding RNAs of various sizes recovered in mitochondrial fraction open up a rich field of information and investigation. Some insights as to mitochondrial lncRNAs as regulators of gene expression and as biomarkers will be presented. Some of these biological insights will hopefully find application in the steadily increasing number of pathologies that are related to dysfunctions of mitochondria, in conditions where pathogenicity and phenotypic expression are not readily understandable.

In phyla ranging from arthropods to mammals, lncRNAs are expressed in spatially restricted patterns – in one or a few tissues on average. Temporal expression patterns are similarly restricted; in Drosophila, expression during more than two hours of development is predominantly limited to neural tissues and the germ line. Environmental challenges reveal a host of largely non-coding genes that are not otherwise at detectable levels in any tissue or time point. In a diverse set of environmental challenges surveyed in Drosophila and Daphnia, lncRNAs are among the strongest and earliest responsive genes. Daphnia is an ecologically critical clade and forms the basis of the metazoan food web in over 90% of fresh water bodies on earth. Daphnia magna responds to microcystin (toxic algae; blooms cost the US >$2B per year) with the over-expression of a single gene, in which the longest ORF is 57aa and is not conserved. The modENCODE Consortium and the Consortium for Environmental Omics and Toxicology have generated large datasets interrogating the patterns of expression and interaction of lncRNAs. Dr Brown will discuss the developing picture of lncRNAs as rapidly evolving, adaptive molecules that help to shape the emergence of new ecological niches, as well as cell types and tissues, and hence, species.

Efficient enzymatic activity of the mitochondrial electron transport chain depends on replenishment of its subunits whose availability is determined by their rates of transcription, translation, import, and protein and mRNA turnover. We have shown that Cerox1, an unusually abundant and conserved long noncoding RNA (lncRNA), substantially alters complex I enzymatic activity, reactive oxygen species production, and levels of transcripts encoding 12 complex I subunits. These enzymatic changes are similar to those observed in disorders with mitochondrial dysfunction, such as Parkinson’s and Alzheimer’s diseases. Increased Cerox1 levels reduce cell division, protect cells against the cytotoxic effects of the complex I inhibitor rotenone and decoy microRNAs which otherwise reduce the abundance of complex I proteins. MicroRNA-488 overexpression reduces Cerox1 levels 13-fold, and mutation of its single response element in Cerox1 alters complex I transcripts’ levels. Cerox1 is the first lncRNA known to regulate mitochondrial energy metabolism.

Long non-coding RNA (lncRNA) genes are at least as abundant as protein-coding genes in humans. LncRNA genes have low interspecies conservation: over 60% are not conserved beyond primates. Diverse regulatory roles – positive and negative, epigenetic and post-transcriptional, nuclear and cytoplasmic – have been identified for over a hundred human lncRNA genes to date, still a small number in contrast to the approximately 20,000 lncRNA genes discovered in the first one and a half decades of post-genomic biology. Our integration of the Gencode lncRNA resource with the NHGRI catalog of significant disease-associated SNPs from Genome-Wide Association Studies (GWAS) revealed specific diseases where a highly statistically significant portion of GWAS hits occurs in lncRNA exons, including obesity and several cancers. Notably, one obesity-associated lncRNA, LOC1572723, lacking conservation beyond primates, was independently pinpointed by 19 published GWAS datasets. Professor Lipovich pursued functional lncRNAomics in another lncRNA-associated disease – breast cancer – using the human estrogen receptor α positive cell line MCF7. Overexpression and knockdown of 26 estrogen-responsive lncRNAs shifted breast cancer cells along the apoptosis-proliferation axis. Reduced cell growth and increased cell death were consistently observed upon knockdown of estrogen-induced, and overexpression of estrogen-repressed, lncRNAs. 12 of these 26 lncRNAs originated after the prosimian split. Overexpression of BC041455, a primate-specific estrogen-repressed lncRNA, reduced ERK phosphorylation uniquely in ERα-positive cells, indicating modulation of the conserved MAP kinase pathway. Professor Lipovich shows by ribosome profiling that BC041455's polysome occupancy is distinct from that of highly translated mRNAs, with implications for hormone dependence of ectopic lncRNA translation in disease.