Everyone in the world is connected in some way or another through networks. We network in the workplace to expand our professional connections. We network on Facebook to keep in touch with our high school friends. Bustling networks between biomolecules like nucleic acids, proteins, sugars, and lipids, in our own cells, make sure that we get to live another day. Proteins are the primary workhorses of life and participate in some of the most important biochemical networks. An essential function of proteins is their direct involvement in transporting cargo from the extracellular environment into the cell, a process that involves intricate networks between participating proteins.
Clathrin-mediated endocytosis (CME) is one of the many modes of transport that bring cargo from one place to another in the cell. CME is a dynamic process that involves dozens of proteins, including clathrin, to form vesicles on the cellular membrane. These vesicles are portals connecting the extracellular and intracellular spaces, serving as sites for clathrin and its protein partners to form a protective “coat” around the cargo to bring it into the cell. Analogous to pinching off a ball of dough that will eventually be used to make dinner rolls, vesicles sequestered within “clathrin-cages” pinch off the membrane and get transported to their eventual destination for further processing. How well the system works depends on the assembly of available proteins and their competition for binding sites on the cell surface. A mathematical representation for their contacts is represented by a model called a protein-protein interaction network (PPIN). Conventional PPINs are characterized by nodes to represent proteins and edges for their one-to-one (or pairwise) interactions.
Like any model, PPINs are only as good as the data they represent. Currently, extensive biochemical information, including the complete roles of different domains (smaller parts of a protein) and strengths of the interactions with their targets, cannot be extracted from PPINs. Unless we map these pairwise PPIs in detail, we will not acquire the complete picture. Take human interactions as an example. If we only knew that two people met and became friends, questions such as “How did these two people meet? When did these two people meet? Are they work acquaintances or best friends?” cannot be answered. As in the case for proteins, understanding when and how strongly a protein interacts with another in a complex cellular process is impossible with traditional PPINs. A traditional PPIN that maps out pairwise interactions at the protein-protein level makes further critical analyses such as modeling the dynamics of any macromolecular system challenging. Identification of a stretch of residues within a protein that interacts with a domain of another might illuminate or possibly uncover target areas for designing drugs. These drug molecules would inhibit potentially harmful interactions at these binding sites to delay the advancement of disease, as in schizophrenia that arises from complications in a central CME protein.
In 2013, Margaret Johnson, then a Postdoctoral Fellow at the National Institutes of Health and now an Assistant Professor of Biophysics at the Johns Hopkins University, tackled this problem. While studying the process of protein assembly, Johnson realized that database-curated PPIs lacked the biochemical specificity needed for a physics-based model to simulate CME interactions. After an extensive but unsuccessful search for PPINs that could provide relevant biochemical information, she sought to create her own network.
Using baker’s yeast (Saccharomyces cerevisiae) as a model system, she combined PPI results acquired from high-throughput screens (experiments that generate large amounts of data using little time and material) and those discovered from more rigorous biochemical experiments to account for all possible PPIs amongst CME proteins. Users of the ‘yeast network’ would not only know which players interact with each other, but acquire novel insight into which parts of proteins interact with each other and how these parts organize themselves into active structures. With nodes not as proteins but as parts of proteins, her yeast PPIN had substantially greater network resolution than any standard PPIN. Margaret termed this Interface-Resolved PPIN, an IIN.
In 2018, as a senior undergraduate in the Johnson Lab, I realized how important a resource an IIN could be for understanding human pathology. I undertook a project to construct the mammalian version of CME IIN, so that we and other biologists could make predictions about dysregulated protein interactions that give rise to disease. Dr. Johnson and I combined our literature search with information extracted from three widely-used PPI databases—BioGrid, IntAct, and mentha, for network construction. Partially automated, the finalized mammalian CME IIN contained 82 proteins, 617 interface-resolved edges, 28 unique interface types, and 11 major interface-interaction types. But what now? So far, we have been able to demonstrate the utility of the mammalian CME IIN within systems biology by analyzing changes to protein copy numbers in various cell types. Knowledge of an approximate amount of protein in a given cell type will allow us to identify deviations in copy number, which could indicate a diseased state. Some proteins are preferentially expressed in certain types of cells. Combining our knowledge of protein expression levels in epithelial and neuronal cells, experimentally derived numbers of proteins, and their relationships extracted from the IIN, we could simulate CME events in various cellular environments.
We can simulate diseases and their progression within the body when proteins are removed or altered in quantities. Depending on structure and interaction characteristics, a variety of potential drug molecules can be deduced and applied in practice to treat CME-associated neural disorders. For example, the mixture of proteins that normally make up the Arp2/3 complex, a key regulator of the cytoskeleton that provides every living cell its integrity, is “off-balanced” in neuronal cells of schizophrenia patients. We can simulate this diseased cellular environment using an algorithm developed by David Holland, a previous member of the Johnson Lab. By implementing this algorithm, we can glean novel insights into the downregulation or upregulation of Arp2/3 partners, and how differences in their populations give rise to schizophrenia. Discovery of protein drug targets that inhibit interface interactions can be validated experimentally, preventing CME-associated neurodegenerative diseases.
Providing a clearer picture of which protein interfaces participate in vital interactions will allow scientists to, not only model knockdowns, as typically done in an experiment, but predict how irregular protein expression levels can affect molecular interactions, and hence cause disease. Perhaps one day, IINs will become just as accessible as Facebook for maintaining close connections. The mammalian IIN will soon become an open-access resource, helping scientists all over the world untangle the web of molecular handshakes that holds life together.
Author
Daisy Duan is a second year Molecular Biophysics & Biochemistry Ph.D. student at Yale University. She is currently studying how a non-receptor tyrosine kinase, Abl2, regulates microtubule dynamics in the Koleske Lab. Outside of the research, Daisy serves as the STEM Education and Outreach Director for Científico Latino, a network of undergraduates, post-baccs, graduate students, post-docs, and faculty in STEM. She is currently one of the organizers for the Graduate Student Mentorship Initiative Program, helping underrepresented and historically disadvantaged students successfully navigate the STEM Ph.D. application process with open-access guides and writing resources.
Editors
Saurja Dasgupta is originally from Kolkata, India. He obtained his Ph.D. at the University of Chicago, where he studied the structure, function, and evolution of catalytic RNA. He is currently doing his postdoctoral research at Massachusetts General Hospital, Boston, where he is trying to understand the biochemical milieu that could have given birth to life on earth (and elsewhere) and reconstruct primitive cells. One of his scientific dreams is to observe the spontaneous emergence of Darwinian evolution in a chemical system. When not thinking about science, Saurja pursues his love for the written word through poetry and song-writing (and meditating on Leonard Cohen’s music). His other passions are trying to make science easier to understand, and fighting unreason and pseudoscientific thinking with a mixture of calm compassion and swashbuckling spirit.
Amrita Anand is in her 4th year of Ph.D. in Genetics and Genomics at the Baylor College of Medicine, Houston. She studies the reprogramming potential of certain key factors in the regeneration of mouse inner ear hair cells. She has been actively pursuing Science communication over the last three years as she enjoys bridging the gap between scientists and non-experts. As an editor, she wants to make science more accessible to the public and also hopes the hard work behind the science gets due credit.
Illustrators
Disha Chauhan did her Ph.D. in IRBLLEIDA, University of Lleida, Spain in Molecular and Developmental Neurobiology. She has post-doctoral experience in Cell Biology of Neurodegenerative diseases and is actively seeking a challenging research position in academia/industry. Apart from Developmental Neurobiology, she is also interested in Oncology. She is passionate about visual art (Illustration, painting, and photography) and storytelling through it. She enjoys reading, traveling, hiking, and is also dedicated to raising scientific awareness about Cancer. Follow her on Instagram.
Saurabh Gayali recently completed his Ph.D. in Plant Molecular Biology from the National Institute of Plant Genome Research (JNU, New Delhi). Currently, he is DBT RA at IGIB (New Delhi), and his research focuses on finding binding associations of Indian plant metabolites with human pathogen proteins, creating a platform for future plant extract-based drug discovery. He has a keen interest in data analysis, visualization, and database management. He is a skilled 2D/3D designer with a specific interest in scientific illustration. In leisure, Saurabh plays guitar and composes music, does photography, or practices programming. Follow him on Instagram.
This article was initially written as part of ComSciCon‘s write-a-thon, and refined by Club SciWri’s editorial team.
The contents of Club SciWri are the copyright of Ph.D. Career Support Group for STEM PhDs (A US Non-Profit 501(c)3, PhDCSG is an initiative of the alumni of the Indian Institute of Science, Bangalore. The primary aim of this group is to build a NETWORK among scientists, engineers, and entrepreneurs).
This work by Club SciWri is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
