Abstract
Abstract Chapter 1: Circular RNA regulation at the single cell and tissue specific levelExonic circular RNAs (circRNAs) are highly abundant RNAs generated mostly from exons of protein-coding genes. Assaying the functions of circRNAs is not straightforward as common approaches for circRNA depletion tend to also alter the levels of mRNAs generated from the hosting gene. In this work we describe an shRNA approach to knock down (at a post- transcriptional level) circular RNAs (circRNAs) without affecting their linear counterparts in vivo. We also describe a computational platform for determining the potential off-target effects as well as for verifying the obtained phenotypes.
Using this approach, we studied in vivo the role of circMbl, the most abundant circular RNA in the fly brain. We found that knocking it down leads to specific changes in the brain transcriptome. In particular, we found changes in genes enriched in some glial clusters, peptidergic neurons, clock neurons, Lawf2 cells, and photoreceptors. To understand where circMbl is expressed we developed a new bioinformatic approach that allowed us to map circMbl in a single cell dataset from the fly brain. We found circMbl highly enriched in Lawf2 neurons, photoreceptor, peptidergic and clock neurons. We found that circMbl levels correlate well with the signal from the linear mRNA. This can be explained as MBL promotes circMbl synthesis.
We could validate these results using total RNA sequencing from sorted cells. Moreover, this data allowed us also to map different mbl linear and circular isoforms. Different mbl mRNA isoforms are expressed in a mutually exclusive manner and are co-expressed with different circMbl isoforms: MBL-OP is expressed with circMbl2-4 in photoreceptors, while MBL-C is expressed along with circMbl in muscles and brain cells. By doing knockdown and overexpression experiments we could corroborate that this co-expression pattern is actually co-regulatory: MBL promotes circMbl back-splicing; while circMbl synthesis competes with the formation of the linear mRNA. In photoreceptors MBL-OP, not only promotes circMbl formation but also inhibits splicing of its first two introns.
We then explored the functions of both MBL-C and MBL-OP and we found that ubiquitous knockdown (KD) of both isoforms leads to major changes in the brain transcriptomes. Moreover, KD of both mbl isoforms lead to changes in similar sets of genes. We do not see these major effects when we knock down each isoform only in neurons by using the neuronal specific driver elav-gal4 driver. As mbl-C and mbl-OP are expressed in different cell types in the adult head, this indicates that MBL-OP and MBL-C might share some functions over development in elav negative cells. Moreover, we found that in heads MBL-C lead to changes mainly in muscle related genes while in photoreceptors MBL-OP KD leads to changes in alternative splicing of genes related to photoreceptor survival to light.
Abstract Chapter 2: Circadian and ultradian RNA oscillations from a single cell perspectivePhysiology and behavior of most organisms oscillate in cycles that follow daily environmental changes. These cycles of 24 hours periodicity include general behavior patterns as activity levels, feeding, and sleep-wake cycles; but also include less conspicuous oscillations such as metabolic or immune function. Circadian rhythms are driven by a transcription–translation feedback loop (TTFL) that generates oscillation at the RNA and protein levels of the so-called ́core-clock genes ́. In Drosophila melanogaster, a set of 300 neurons (clock neurons) control the circadian rhythm of locomotor activity and directly or indirectly control rhythms in other behavioral processes as sleep and feeding. In addition, peripheral tissues in Drosophila also display 24-hour rhythms at the gene expression and physiology levels. These peripheral oscillations can be either completely self- sustained by the peripheral tissue, as well as entirely dependent or partially regulated by the central brain through neurohormones like PDF. To this date it is unknow which cells in peripheral tissues harbor a molecular clock. Moreover, it is not known if the cycling genes in these tissues are coming directly from cells with a functional molecular clock or, or from signaling from neighboring cells or the central brain. To answer these questions, we mapped the main core-clock genes, and PDF receptor (Pdfr) in all cells in the fly body using publicly available single-cell data from the fly atlas. We also mapped cycling genes coming from 3 tissues with robust oscillations: fat body, gut and Malpighian tubule. Strikingly, we found that in peripheral tissues, only some cells express core-clock genes. Expression of Pdfr is highly expressed in cell with no core-clock expression particularly high in lamina neurons in the optic lobe. Additionally, we found Lawf2 and Poxn neurons as new putative clock neurons in the optic lobe and the brain respectively.
Circadian rhythms are present in most living animals but rhythms with shorter than 24-hours period (ultradian) are also prevalent. In mammals, some examples or ultradian rhythms are body temperature, hormone release and sleep. In mice and flies, ultradian locomotor activity patterns arise with the disruption of the circadian clock. However, there is no knowledge on ultradian gene expression in the fly brain or the molecular mechanism governing these rhythms. To study this, we surveyed RNA expression from fly brains every two hours. We found genes cycling in an ultradian fashion with two and three peaks of expression over the day (period of around 8 and 12 hours). Among the ultradian cyclers, we found various genes implicated in different functions such as Wnt signaling, Serotonin receptor signaling, and response to bacterial infection. Interestingly, a proportion of transcripts cycle in an ultradian fashion in the Drosophila melanogaster brain and are enriched in glial cells.