Structure and Specificity of Clustered Protocadherin Interactions That Mediate Neuronal Self-Avoidance

Abstract

This dissertation is presented in two parts: the first on understanding the determinants of clustered protocadherin interaction specificity and the second on the reconstitution of the bacterial periplasmic methionine sulfoxide reductase system MsrPQ. Part I: Clustered protocadherin interaction specificity Brain development is orchestrated through synaptic interaction of neighboring neuronal cells. Remarkably, neuronal dendrites are able to distinguish ‘self’ from ‘other,’ and subsequently avoid forming synapses with dendrites originating from the same cell. This self- avoidance is a key feature in the patterning of neuronal networks and requires neurons to have a unique identity that is provided by the more than 53 isoforms of the clustered protocadherin gene locus. Clustered protocadherins are surface-expressed adhesion proteins that are members of the Ca2+-dependent cadherin superfamily. Each neuron expresses a collection of these isoforms and homophilic interactions of clustered protocadherin between cells in trans dictates self/non-self discrimination. I have used x-ray crystallography to determine the architecture of the trans homodimer that is formed by clustered protocadherin extracellular cadherin-repeat (EC) domains. The first four (of six) ECs engage in an antiparallel two-fold symmetric dimer where EC1 interacts with EC4 and EC2 with EC3. We found that this dimer is conserved throughout the clustered protocadherins and is also in non-clustered protocadherins that are important for the development and maintenance of the nervous system. The function of clustered protocadherins requires the homophilic interactions to be exquisitely specific. I used bioinformatics, in collaboration with the laboratory of Debora Marks (Harvard Medical School), to determine how specificity between isoforms arose. Analysis of isoform-specific conservation and sequence coevolution in combination with structural comparison of distinct isoforms led us to the observation structural differences between isoforms and chemical properties of amino acids contribute to this specificity between subfamilies and within subfamilies, respectively. Our bioinformatics work, including a coevolution-based statistical interaction energy model, also identified the EC2/EC3 interaction as the primary source of specificity. In aggregate, my results explain how these proteins encode specificity to ensure self- avoidance. These results provide a framework to explore the role of clustered protocadherins in brain development and to understand why clustered protocadherin mutations are implicated in complex brain disorders such as autism, bipolar disorder and schizophrenia. Furthermore, we have shown that the clustered protocadherin proteins are a valuable system to study the specificity of protein-protein interactions and to develop statistical models for evaluating the role of individual mutations on interaction specificity. Part II: Biochemical reconstitution of MsrPQ Gram-negative bacteria that encounter reactive oxygen species (ROS) stress, such as the ROS generated by the innate immune system by macrophage, are susceptible to methionine oxidation, which can unfold and inactivate proteins. To repair this damage in the periplasm, bacteria rely on MsrPQ, an operon that produces two proteins: MsrP, a periplasmic methionine sulfoxide reductase molybdoenzyme; and MsrQ, a transmembrane cytochrome that transfers electrons to MsrP. Our understanding of MsrPQ function and mechanism is limited by a lack of knowledge about the biochemical requirements for MsrPQ function. I purified and characterized MsrQ to understand how MsrP and MsrQ work together to repair damaged periplasmic proteins. I identified that MsrQ copurifies with flavin mononucleotide (FMN), suggesting, in agreement with other work, that reduced FMN, generated by NADPH-dependent flavin reductase Fre, can reduce MsrQ. Based on electrochemical studies demonstrating that MsrP requires a reduction potential of -250 mV to support catalysis, I proposed that FMN, which has a similar reduction potential, is capable of serving as the electron donor while membrane-bound quinones, which have much less negative reduction potentials, would be unable to support MsrP catalysis. Reconstitution of MsrPQ with the flavin reductase Fre supports methionine sulfoxide reduction, confirming that reduced FMN is sufficient to support MsrPQ activity. The work presented in this part of my dissertation resolves some of the questions about MsrPQ activity but still more questions remain. The biochemical assay developed here can be used to further elucidate the mechanism of MsrPQ. This assay can also be used to identify inhibitors of MsrPQ that would prevent methionine sulfoxide repair and lower survival of gram-negative bacteria during infection.