RNA-Based Gene Regulations

Non-coding RNAs are used in all walks of life to regulate gene expression (Carthew and Sontheimer, 2009). Our group studies the mechanisms and functions of RNA-based gene regulation.

Genomes contain thousands of genes and their expression is controlled by transcription. Although it was thought that transcription fidelity ensures that genes do not make double-stranded transcripts (dsRNA), this is not entirely true. Convergent transcription of neighboring genes can generate dsRNA where they overlap. In humans, the 3'UTRs of 2000 genes overlap in such a manner. Transcripts can also foldback into extended hairpins. Collectively, these dsRNAs generate endogenous siRNAs (endo-siRNAs) due to the action of the Dicer ribonuclease. The endo-siRNAs guide RNAi silencing of homologous RNAs such that hundreds or thousands of protein-coding genes are negatively regulated in this way.

What functions do endo-siRNAs have? We know of at least one function for them during embryogenesis. If an embryo is subjected to a temperature gradient over its body, the warmer regions will develop faster than the cooler regions but only up to a certain point. Thereafter, something compensates for the difference and the regions synchronize their development (Figure 1). This synchronization requires endo-siRNAs to be made (Lucchetta et al., 2009 ). If not, the embryo remains out of sync and dies. We speculate that endo-siRNAs are working like molecular thermostats, slowing gene expression down if it goes too fast because of higher temperature.

MicroRNAs

Certain transcripts make hairpin-loops that generate microRNAs (miRNAs) due to the action of Dicer. There are hundreds of miRNA genes encoded by animal and human genomes. MicroRNAs silence protein-coding genes whose transcripts bear complementary sequences in their 3'UTRs. Basepairing does not need to be perfect for silencing; in fact, binding is typically imperfect between miRNA and transcript. Silencing occurs by transcript destabilization and inhibition of protein translation. How miRNAs inhibit translation is not clear and is quite controversial (Carthew and Sontheimer, 2009).

We do not exactly know how many genes are regulated by a miRNA. Experimental and computational methods predict hundreds of genes are regulated by each miRNA because of the imperfect nature of miRNA binding. Some of the genes that are regulated do very important things for the organism, and their silencing is paradoxically important to the organism as well. For example, one such gene makes a protein that prevents germ stem cells from dividing (Figure 2). However, the stem cells produce miRNAs that silence the cell cycle inhibitor, thereby allowing the stem cells to divide and produce eggs (Hatfield et al., 2005 ) There are many examples of this kind of silencing, and many aspects of physiology and development depend on miRNAs.

MicroRNA genes arose at the dawn of animal evolution, and about 30 of these genes have remained conserved across the animal kingdom. The regions of the body where each of these genes is expressed has also changed very little across the animal kingdom. For example, the miRNA miR-7 is specifically expressed in neurosecretory cells of animals as diverse as annelid worms, insects, and mammals (Li et al., 2009 ). We still do not know whether these ancient miRNAs have conserved functions within the cell types in which they are expressed. MicroRNA evolution did not stop with the Cambrian explosion. It turns out that miRNA genes are the fastest evolving type of animal gene known. They are born at rates about ten-fold faster than protein-coding genes, and they seem to change very rapidly (Lu et al., 2008). At last count, there are 1400 miRNAs identified in humans and 240 miRNAs in Drosophila. Clearly, they are a rapidly growing gene family.

Stress Responses and miRNAs

MicroRNAs seem to play major roles in stress responses, when individuals are challenged by an environmental or physiological perturbation. Some have argued that this is because miRNAs function in biochemical networks (Figure 3). Most biochemical networks are robust; things happen reproducibly and uniformly even in the face of variability that can be induced by the environment, genome variation, and random chance (Pelaez and Carthew, 2012). Biological processes, particularly irreversible ones such as differentiation, are strongly robust to ensure a minimal impact of error. We think that miRNAs help to generate resistance of biological processes to perturbation (Figure 3).

Focused experimentation on individual miRNAs corroborate our idea. In Drosophila, the transcription repressor Yan binds and represses the miR-7 gene. In turn, miR-7 represses expression of Yan. Yan and miR-7 are part of a network that regulates the transition of multipotent progenitor cells to become differentiated photoreceptors (Li and Carthew, 2005). YAN and miR-7 act within several feedback and feedforward loops in the network (Figure 4). When the environment is perturbed in a miR-7 mutant by oscillating growth temperature, the network switches states less robustly, and errors in cell differentiation are observed (Li et al., 2009 ). These errors are undetectable under uniform temperature conditions. If miR-7 is not mutated, the network is robust under either uniform or oscillating temperature growth conditions. This miRNA makes differentiation resistant to mild environmental perturbation.

Gene expression also has to withstand genomic perturbations. For example, 0.5% of the human genome is variable between any two unrelated individuals, and other species show even greater diversity. This diversity can potentially create problematic variability in the activity of gene regulatory networks, and ultimately, in cell behaviors. We have found that in the proneural network, regulation of senseless gene expression by microRNA miR-9a dampens the effects of genomic diversity (Cassidy et al., unpublished). Reducing miR-9a levels, or its ability to regulate senseless, frees the genomic landscape to exert greater influence on cell behavior. We have used whole-genome sequencing to identify the genome landmarks or loci that potentially exert such effects. A larger set of sequence variants, including variants within network genes, exhibit these characteristics when miR-9a concentration is reduced. These findings reveal that microRNA-target interactions may be a key mechanism by which the impact of genomic diversity on cell behavior is dampened. As our approach can be generalized, it should aid in understanding how genomic diversity is buffered in other organisms for traits that include disease risk and prognosis.

Disease and MicroRNAs

Given the sheer number of miRNAs and their fundamental function in gene regulation, it is not surprising to find them linked to or causative of many human diseases. These are

•Cancer - leukemia, liver, lung, glioma, renal, pancreatic, colon, breast, prostate


•Cardiac hypertrophy and ischemia


•Muscular dystrophy


•Lou Gehrig’s Disease


•Alzheimers Disease


•Parkinsons Disease


•Tourettes Syndrome


•Psoriasis


•Diabetic nephropathy

Anti-miRNA oligonucleotides show promise as a new class of therapeutic treatment for some of these diseases.

The Future

Our objective is to understand the principles of gene regulation by endo-siRNAs and miRNAs in animals as a means to understand and treat human disease. One research aim is to use systems-level approaches to gain a complete picture about how miR-7 provides robustness to the YAN network. This aim utilizes gene recombineering, live-time-lapse imaging of fluorescently tagged proteins, next-generation sequencing, and quantitative computer modeling, with the help of Luis Amaral’s group at Northwestern. Another aim is to explore the role of miRNAs in adaptation during selection. Robustness operates on the evolutionary scale as well, and miRNAs might inhibit adaptation when populations are selected for certain traits. We are using artificial selection paradigms coupled with next-generation sequencing. A third aim is to determine whether a miRNA elicits robustness by regulating a handful of target genes or by its overall effect on a dynamic transcriptome. An intriguing idea is that the level of miRNA regulation for any given transcript depends on its transcriptome environment - if other transcripts that bind the miRNA are more abundant, then less miRNA is available to regulate the transcript. A shifting transcriptome responding to a variable could be buffered by a counteracting shift in miRNA partitioning between transcripts.