RNA binding proteins (RBPs) bind to RNA molecules in order to define the mature spliced sequence, direct nuclear export and sub-cellular localization, control translation initiation and efficiency, and regulate stability and ultimate decay. RNA processing plays critical roles in regulation of early development and differentiation of stem cells into mature lineages, and mis-regulation of RNA processing plays critical roles in cancer, neurodegenerative and other heritable diseases, and in both viral replication and anti-viral immune responses. Indeed, the emergence of genomics techniques have rapidly advanced our ability to identify genetic and transcriptomic causes of disease, and indeed there is now an ever-growing list of mutations in RBPs and RNA processing events causally linked to human diseases. However, it remains challenging to rapidly convert this genetic knowledge into the mechanistic understanding of the physiologically relevant mis-regulation required to develop therapeutic interventions, and to understand the complex regulatory networks controlled by the more than 1500 RBPs in the human genome. We utilize a mix of experimental and computational approaches to map RNA binding protein interaction networks and the regulatory roles of RBPs, in order to develop a global understanding of how RNA processing regulatory networks drive human physiology.
The first step to understanding an RBP's roles is to identify the RNAs it interacts with. The development of HITS-CLIP and iCLIP methods revolutionized our ability to map RBP binding sites in vivo by enabling pulldown of an RBP of interest, followed by high-throughput sequencing of crosslinked RNA fragments. However, low efficiency of converting RNA into sequencing library limited their scalability in building large-scale networks incorporating many RBPs.
To address this limitation, we and others recently described improved CLIP methodologies that dramatically improve the efficiency of converting immunoprecipitated RNA into high-throughput sequencing libraries, decreasing experimental failures and costs. The improved efficiency in our eCLIP approach enabled quantitative normalization against paired inputs to distinguish true binding events from false positive artifacts, and empowered the generation of 223 eCLIP datasets profiling 150 RBPs in K562 and HepG2 cells.
We are continuing to build upon our eCLIP work to enable profiling of additional RBPs (including those lacking suitable antibodies), as well as in low-input or complex tissue samples. We are also using the eCLIP framework to develop improved targeted methods that deeply explore individual aspects of RNA processing, including deep experimental mapping of microRNA and snoRNA targets and simplified approaches to quantify translation efficiency transcriptome-wide.
Genes are often thought of as fixed instructions, but in reality cells can “edit” these instructions through a process called alternative splicing, allowing a single gene to produce many different proteins. This process is essential for normal human development and is frequently disrupted in diseases such as cancer, neurodegenerative disorders, and heart disease. While scientists have identified many factors that influence splicing, it remains difficult to pinpoint what actually drives these changes in disease. We are exploring how cells control this process in two ways:
RNA splicing controls both sequence and expression, and it is well-established that RNA binding proteins (RBPs) bind to cis-regulatory elements in exons or flanking introns to drive exon inclusion in cell-type or condition-specific ways. However, although the “arrows” visualizing enhanced exon inclusion by RBPs are common in review articles, the mechanism of this induction is understood for surprisingly few examples; for most we do not know the stage of splicing they act on, proteins or snRNAs they interact with, or whether they act to recruit, stabilize, or enhance catalytic activity or transition steps of the spliceosome. A handful of known examples involve the first recognition steps of splicing, where RBPs can enhance recruitment of U1 or U2 complexes to the 5’ or 3’ splice sites respectively. However, in recent work we uncovered a novel mechanism through which the RBP QKI directly interacts with the U6 snRNA and tri-snRNP proteins, enabling tissue-specific inclusion of exons with strong U1 but weak U6 complementarity. We are exploring whether U6:5'SS duplex formation is a highly-regulated fidelity checkpoint, with QKI as a prototype ‘U6 engagement module’ that directly interfaces with U6 and tri-snRNP to exert specific control of U6-limited 5' splice sites in the right place and at the right time.
In a second project, we are exploring a largely overlooked idea: that the cellular machinery responsible for splicing, called the spliceosome, can itself be dynamically regulated. In particular, we focus on small RNA components of the spliceosome (snRNAs), which carry chemical modifications that were once thought to be static but are now emerging as flexible regulators of gene expression. Recent preliminary work suggests that these modifications may act as a hidden layer of control that helps determine how genes are spliced in different tissues and disease states, but this remains poorly understood. By revealing how this previously hidden layer of gene regulation operates, this work will not only uncover a novel basic mechanism that helps explain how different cells can achieve dramatically different functions, but also may open the door to novel diagnostic and therapeutic strategies that manipulate this layer to alleviate disease.