2004 Summer Scholar Participants-HHMI Undergraduate Program - Carnegie Mellon University

2004 HHMI Summer Scholar Participants

David Hill

David Hill, Carnegie Mellon University
(Mentor: Dr. Elizabeth Jones)

Analyzing Distance Functions for Clustering Protein Subcellular Location Features

Proteomics is the term used to describe large scale documentation and characterization of many or all proteins expressed in a given cell type. Knowledge of protein subcellular location is critical to the understanding of its function. Previous research in Murphy Lab has shown that automated classifier and clustering methods could be trained to recognize and organize protein location patterns. The core of these approaches is the implementation and optimization of the Subcellular Location Features (SLFs), which are numerical descriptors of subcellular location patterns. The purpose of the current research is to compare different approaches to constructing optimal cluster sturctures for a set of proteins based on their location patterns. One of the ways that this will be accomplished is by comparing different distance functions and the effects that they have on the structure of a cluster. For this research nine different distance functions were implemented and a ratio of the distances between clusters and within clusters was obtained using each distance function. The ones that yielded the highest ratio were implemented in a k-means clustering algorithm, and the generated clusters were compared. Another goal of this research is to analyze the clusters that have been generated in an attempt to determine how well they fit the data set that was used to create them.

Catherine Hofler

Catherine Hofler, Carnegie Mellon University
(Mentor: Dr. John Woolford)

Dominant-Negative Mutants of Ytm1p, an Essential Non-Ribosomal Protein in Saccharomyces cerevisae

Ribosome assembly is a complicated process involving the maturation of rRNA and it's association with ribosomal proteins. Non-ribosomal proteins are also involved in coordinating the assembly of ribosomes by associating with the pre-rRNPs (ribonuclear protein particles) and then later dissociating. Ytm1p, an essential and conserved gene, was found to be involved in the maturation of the 66S pre-rRNP in both classic genetic studies and state of the art proteomic studies. Ytm1p is one of 17 known proteins involved in ribosome assembly that contains WD40 repeat structures. However, none of these proteins have been studied in great detail. Thus, studying the structure and function of Ytm1p can provide a model for the behavior of the WD40 repeat protein in multi-molecular complexes. Ytm1p, with its seven WD40 repeats, may be used as a scaffolding protein to nucleate and coordinate the assembly of protein-protein complexes or protein-RNA complexes used in the synthesis and maturation of the 66S pre-ribosomes. Specific sites on the protein may be involved in the scaffolding activity of Ytm1p. By mutating specific amino acids in the Ytm1p protein and studying the specific effects of the mutations, one can infer information on the function of the protein. Single point mutations were created in YTM1 using error-prone PCR and the genes were introduced into plasmids via gap repair. Mutant alleles of ytm1 were placed under the control of the inducible GAL1 promoter in order to screen for dominant-negative mutations. Seven mutants were identified in this screen: four strong dominant-negative mutants which inhibited growth of the yeast cells on media containing galactose, and three weaker mutants which only slowed growth on galactose media. These mutants were screened for defects in the synthesis of rRNA and will be screened for defects in polysome profiles and cell cycle progression. Characterizing the specific phenotypes of these dominant-negative mutations and their effects on the synthesis of the 60S ribosomal subunit will help to identify relationships between the structure of the Ytm1p and its function.

Sara Kapner

Sara Kapner, Carnegie Mellon University
(Mentor: Dr. Justin Crowley)

β-Catenin as a Molecular Correlate of Neural Circuitry Development

The brain transmits information from one area to another through specific neural circuits. The mechanisms behind the formation of these precise patterns of neural circuitry have been suggested to be either activity-dependent or governed by patterned gene expression. Primary visual cortex has many stereotyped patterns of neural circuitry specific to feature "maps", e.g. visual space, ocular dominance (or eye specific columns) and orientation preference. Although it was once widely accepted that the development of these maps was activity dependent, it has recently been shown that these columns form even when the inputs are modified. We hypothesize that the stereotyped patterns of circuitry found in visual cortex are generated by patterns of gene expression. In order to identify the genes that could be responsible for the formation of these circuits we identified potential candidate genes from the literature. β-catenin, a candidate gene, is a known cell adhesion and signal transduction molecule that is responsible for segregating cells and directing their fate. I used fluorescence microscopy and a CCD camera to image immuno-labeled ferret brain slices in order to determine the pattern of β-catenin expression in tangential sections of developing visual cortex. I then created a mosaic that joined the many images of one histological slice together using blood vessels and other reference points as registration marks. Once an entire slice was completed, the β-catenin expression "splotches" were segmented according to high-reliability and low-reliability. We are still in the process of acquiring data, but through preliminary analysis we have uncovered certain facts about the splotches. The developmental stage of the animal does not seem to correlate with splotch size. Across all ages, the average splotch size that does not intersect the section edge is approximately a square millimeter. As we collect more data we will determine correlations between β-catenin expression and the specific patterning of the neural circuits. Future experiments will test if β-catenin is not only correlated with but actually causing the specific development of neural circuitry by over and under expression.
Daniel Kleinbaum

Daniel Kleinbaum, Carnegie Mellon University
(Mentor: Dr. William Brown)

Building Biosensors from Single Chain Variable Fragments

Single Chain Variable Fragments (ScFvs) are engineered antibodies that bind to a particular target. This specificity makes these molecules good candidates to develop as components of biosensors. In order to do this, it is necessary to identify sites on the antibody that can be used as attachment points for a signal transducer molecule. These sites must be close enough to the binding pocket for the transducer to detect binding, but cannot interfere with the molecule's affinity for the target. Since antibodies have a conserved structure, it is thought that such sites will be generic and therefore may be applicable to other antibodies. In the summer of 2003 "The Seeker" computer program was written to identify sites on the ScFv, based on the molecule's structure, to which a signal transducer should be attached. This was done by analyzing the structure of the ScFv and searching for sites based on criteria like distance from the binding pocket, solvent accessibility, and the C -C angle. When the program was run on several ScFvs, four consensus locations for placement of the signal transducer were found. Site-directed mutagenesis was performed on these sites to convert the wild-type amino acid to Cysteine, which is necessary for attachment of the transducer. These mutant ScFvs were then expressed in cells and extracted. Preliminary binding assays were performed on these extracts. The antibodies were then purified from the extract and standardized so that the concentrations of the mutants were equal. All that remains is to perform binding assays with and without an attached transducer molecule to determine whether or not its presence affects the binding.

Meng Lu

Meng Lu, Carnegie Mellon University
(Mentor: Dr. Elizabeth Jones

Genetic Interactions of PBN1 in Saccharomyces cerevisiae

PBN1 is an essential gene that encodes the endoplasmic reticulum (ER) integral membrane protein Pbn1p in the budding yeast, Saccharomyces cerevisiae . The only known function of Pbn1p is to facilitate the autocatalytic cleavage of the protease B precursor in the ER. However we know this is not the essential role of Pbn1p since cells lacking protease B (PRB1) can survive while deletion of PBN1 is lethal. The purpose of this project is to understand the essential role of PBN1 in S. cerevisiae by looking for synthetic interactions between pbn1 and mutant alleles of cne1, mpd1, mpd2, eug1, and eps1. These genes, encoding proteins responsible for proper protein folding and quality control in the ER were chosen because we believe that Pbn1p might play a similar role. Our belief is strengthened by the previous observation that pbn1-1 shows synthetic lethal interactions with a mutant allele of the gene that encodes Ero1p. Ero1p helps catalyze oxidation and disulfide bond formation during protein folding in the ER. Targeted deletions were made of the non-essential genes MPD1, MPD2, EUG1, EPS1 , and CNE1 by replacing the wild type allele with a HIS3 marker gene. The strains containing the null mutations were each crossed with strains carrying pbn1-1. These diploids were then induced to sporulate. Tetrad dissections were performed and double mutant spores containing pbn1-1 along with one of the deletions for MPD1, MPD2, EUG1, EPS1, or CNE1 were obtained. They were identified by looking for Prb - haploids viable in histidine lacking medium. These double mutants are currently being tested for growth defects and in other ER protein folding assays such as sensitivity to dithiothreitol (DTT) and tunicamycin. This approach will give us further insight into the essential role of PBN1.

Carolyn Mallozzi

Carolyn Mallozzi, Carnegie Mellon University
(Mentor: Dr. Brooke McCartney)

The Role of Adenomatous Polyposis Coli Proteins in Cytoskeletal Organization

Mutations in the tumor suppressor Adenomatous polyposis coli (APC) are linked to the development of colon cancer. APC is a well-known component of the destruction complex that negatively regulates Wnt/Wg signaling by targeting the key effector b -catenin for destruction. APC is also associated with actin and microtubules, although its exact cytoskeletal functions are not well understood. Our lab has observed that cells in the Drosophilia wing blade mutant for both fly APC1 and APC2 (APC1/2) are associated with blisters, but the cellular biological basis for this phenotype is not known. Blisters can be caused by defects in integrin-based adhesion and cytoskeletal organization. Because APC proteins can mediate cytoskeleton interactions, we predict that the blisters observed in APC1/2 mutant clones are due to cytoskeletal defects. To test this hypothesis, we marked APC2 g10 APC1 Q8 mutant clones with yellow to determine the boundaries of the clones in the wing. To determine the relationship between APC2, actin, and microtubules in wild type pupal wing epithelia, we used antibody staining of fixed pupal wings during different stages of development, and observations of APC2-GFP in live pupal wings. Our analyses have determined that APC2 is found at the bases and along pupal wing hairs and is enriched at the apical surface of wing epithelial cells, consistent with the localization of actin. The colocalization of APC2 with microtubules is unknown, but is currently being investigated. As a complement to these analyses, we asked whether APC2 genetically interacts with the centrosome component c entrosomin ; previous studies suggested a link between these proteins in the organization of microtubules. Our preliminary genetic studies in the embryo suggest that APC2 and c entrosomin do not interact, however, we will pursue additional genetic assays to confirm and extend our preliminary finding.

Tara Marsh

Tara Marsh, Carnegie Mellon University
(Mentor: Dr. Nathan Urban)

Inducing Neurogenesis in the Olfactory Bulbs of Mice

The long standing dogma of neuroscience was that at birth, the brain contains all the neurons it will ever have. More recent research has revealed signs of neurogenesis in several areas of the brain, including the hippocampus and the subventricular zone. Cells generated in the subventricular zone of the lateral ventricle migrate to the olfactory bulb, a brain region involved with processing odor information. Little is known about the mechanism by which the generation, migration or survival of these neurons are regulated. My research this summer focused on determining if neurogenesis can be induced through activity. The hypothesis was that by forcing the brain to learn new tasks, the generation, migration or survival of new neurons will be altered. To test this hypothesis, mice were engaged in a series of odor recognition and discrimination tasks. Mice were first taught to dig in a dish to find a Froot Loop, and were then trained to locate the Froot Loop by way of a particular odor. At first, mice had to discriminate only between two odors, such as peppermint and spearmint. They then had to discriminate between increasingly more difficult odor mixtures, such as a 70% to 30% peppermint to spearmint mixture versus a 30% to 70% peppermint to spearmint mixture. After reaching the threshold of discrimination, which appears to be a 60%/40% odor mixture, the mice were injected with bromodeoxyuridine (BrdU), a marker of cell proliferation. The next step in the process is to sacrifice the mice, excise the brains, and perform a series of immunohistochemistry steps to label the BrdU positive cells so that we will be able to compare the generation and migration of new neurons in trained and untrained animals. Further experiments using markers for apoptotic neurons will allow us to determine whether the rate of apoptosis was altered. We hypothesize that the number of new neurons found in the trained mice will be greater than the number of neurons generated in the brains of the control group mice, which will receive odor exposure but not be trained to associate any particular odor with food reward. If novel, challenging activities or tasks can in fact induce neurogenesis, this experiment could lead to mechanisms that may be useful in treating a variety of human neurodegenerative conditions including brain damage, strokes and Alzheimer's disease.

Jessica McGillen

Jessica McGillen, Carnegie Mellon University
(Mentor: Dr. William Brown)

Kinetic Study of Polychlorinated Biphenyl Dechlorination in Hudson River Sediments

Poly-chlorinated biphenyls (PCBs), a type of carcinogenic contaminant, were once widely used in industry. Their stable, lipophilic nature has allowed them to persist in the environment and damage river ecosystems through wastewater and spills. Fortunately, some microorganisms in river sediment, including bacteria and archaea, have developed an ability to anaerobically dechlorinate PCBs and form products that are both less toxic and more biodegradable. As a model system, microbes can dechlorinate one PCB congener, 2,4,5-trichlorobiphenyl (BZ-29), along two pathways to form the daughter product 2-monochlorobiphenyl (BZ-1). Among the many PCB-contaminated rivers in the US, the Hudson River (NY) is of particular interest. Microorganisms in its sediments dechlorinate PCBs up to three times faster than microbes in the Grasse River (NY), another contaminated site. The Hudson River's location in farming country may mean that certain nutrients, deposited in the sediment as part of agriculture waste, affect the growth of PCB dechlorinating microbes. A kinetic study is being carried out to explore the impact of nutrients on the rate and pathway of PCB dechlorination in Hudson River sediment. The experiment comprises samples of 140 sediment microcosms in serum bottles, with each bottle containing combined live and PCB-spiked river sediment and varying nutrients, including NH4 Cl (N); NH4 OH; NaCl; phosphate (P); and formate and acetate (FA). The samples are anaerobically and statically incubated at room temperature. Gas chromatography has been used at intervals to monitor the samples' biological activity. After five weeks of incubation, most of the headspace gas was nitrogen, with minimal amounts of carbon dioxide (under 2.1%), only traces of oxygen (under 0.7%), and no detectable methane. These results are consistent with the anaerobic, nonmethanogenic nature of the Hudson River's PCB-dechlorinating microorganisms. Periodic PCB extractions are being used to measure changing amounts of BZ-29 and its dechlorination products as time progresses. The presence of BZ-29 exclusively in each subset on day 10 indicated successful PCB extraction and no significant dechlorination. When dechlorination occurs at later time points, its path and rate within each subset will be determined and compared. The results will reveal important information about the impact of sediment composition on the growth and activity of PCB dechlorinating microbes, and ultimately will help improve bioremediation of river ecosystems.
Mayur Parepally

Mayur Parepally, Carnegie Mellon University
(Mentor: Dr. David Hackney)

Determining the Effects of Microtubule Binding Domain of Kinesin's Tail in the Processive Movement of the Kinesin Motor Protein

Kinesin is a motor protein found in cells capable of sliding processive movement of cargo along microtubules in a step-wise "hand over hand" motion of its dimer motor head groups. In its folded state, full length kinesin is inactive as its tail binds to the motor head groups blocking its microtubule stimulated ATPase activity. Kinesin reaches an active state when unfolded, as kinesin's tail releases the dimeric head groups facilitating processive movement along a surface. The tail domain also contains a secondary microtubule binding region that is unmasked when kinesin exists in the unfolded state. It is known that the microtubule binding region increases the affinity of kinesin for microtubules, and can in fact independently bind to microtubules. Theoretically the unmasked microtubule binding region could act as an anchor facilitating continuous processive movement. Evidence suggesting this include aggregation of microtubules due to motors sticking and cross linking. We hypothesize that the microtubule binding region of kinesin's tail domain can also contribute to the increase in the processive movement of the motor when in the active unfolded conformation and can also lead to accumulation of the motor at the + end of the microtubule. To test if the tail's microtubule binding region plays a role in the motility of kinesin, truncated versions of the motor protein (at the c-terminus) were created. Constructs were derived from the drosophila kinesin vector, which included PCR inserts of varying tail lengths. The series of truncations were made so kinesin's tail had amino acid lengths of 841, 892, 927, 941, and 960. The constructs were expressed in transformed E. Coli cells and confirmed using IPTG screens and DNA sequencing. The longer truncations contained the microtubule binding region whereas the shorter versions had it partially present or completely removed, giving a method to establish a relationship between the presence of microtubule domain and processive movement. The kinesin constructs also contained the gelsolin tag able to link kinesin to fluorescent actin filaments. This offered a way to visually measure the single sliding velocity of individual kinesin motor proteins through fluorescent microscopy. Significant differences in the movement of the series of truncations of kinesin would indicate that the microtubule binding region of kinesin's unmasked tail is responsible for the increase in the motor's processive movement. Tests are now underway measuring the sliding velocity of each kinesin construct, and in the future kinesin constructs unfolded by cargo binding will be investigated to see if similar results would be derived in a more physiological set-up.

Abigail Rives

Abigail Rives, Carnegie Mellon University
(Mentor: Dr. Fred Lanni)

Production of GTPase Binding Domain from p21-Activated Kinase and Wiskott Aldrich Syndrome Protein for Analyzing Cdc42 and Rac1 Activity in Live Cells

The purpose of this project is to optimize the production and purification of the p-21 Binding Domain (PBD) and Cdc42 Binding Domain (CBD) and to use these proteins in binding and uncaging experiments. PBD is the domain in p-21 Activated Kinase 1 (PAK1) which binds the active form of the small GTPases Cdc42 and Rac1. Similarly, CBD, from Wiskott Aldrich Syndrome protein (WASp), is the domain which binds active Cdc42. With GTP bound, Cdc42 and Rac1 activate signaling pathways promoting actin assembly and the formation of filopods and lamellipods, respectively, at the plasma membrane. The recombinant domains were produced in expression strain Escherichia coli (BL21-AI) by transfection with pET23 plasmid coding for wild type and modified PBD and CBD that have been hexa-histidine tagged. In both cases the production of these proteins resulted in the formation of inclusion bodies. We tested various methods of dealing with this, such as growing the bacteria at various temperatures and solublizing the proteins in urea before purifying them. The recombinant protein was purified through cobalt chelate affinity chromatography. Since the PBD coding sequence codes for 86 amino acids, with 11 lysines and the CBD 121 amino acids and 8 lysines, we have tested these purified proteins in caging and uncaging experiments. To cage a molecule, the photoremoveable protecting group, 6-nitroveratryl chloroformate (NVOC-Cl) is added specifically to the lysine side chains of the purified protein, where it can block binding of target proteins. Exposure to UV light (340-360nm) breaks off the caging groups from the peptide and regenerates the lysine side chains, allowing native protein binding. Experiments in uncaging proteins that have been caged were performed, using native protein gels to evaluate success. Various applications of UV light were tested and analyzed. Once native protein interactions are restored, the PBD and CBD should resume competitive binding of active GTPases, inhibiting their downstream activity in the cell. Work continues on using caged proteins to observe how inhibition of these GTPases alters acto-myosin structure and behavior in live cultured mammalian cells.

Audra Siegel

Audra Siegel, George Washington University
(Mentor: Dr. Alison Barth)

Examination of the Difference Between "Green" (FosGFP+) and "Non-green" (FosGFP-) Pyramidal Cells in Layer II/III of the Barrel Cortex in Mice

The immediate-early gene c-fos is frequently used as a marker for neural activity (Sheng et al., 1990). The green fluorescent protein (GFP) yields an effective way of identifying neural configurations, when linked with the c-fos promoter (Barth et al., 2004). The basis of this experiment involves anatomical comparisons between "green" pyramidal cells (neurons expressing the fosGFP gene) and "non-green" pyramidal cells (neurons that do not express the fosGFP gene) in the superficial layers of the cortical tissue in mice. Preliminary data reveals that "green" cells tend to have larger soma than "non-green" cells, and thus are likely to have longer axons as well. Furthermore, the fact that both c-fos and fosGFP expression is found in active neurons introduces the prospect that the axons of "green" cells being highly active contain callosally projecting axons that cross to the opposing hemisphere of the brain by way of the corpus callosum (a vital chiasm that connects the two hemispheres of the brain). The primary objective of this experiment was to detect the projection targets of "green" pyramidal cells in layer II/III of the barrel field. In order to observe projection patterns within fixed tissue, tracer injection surgeries were performed on both wild type (mice without fosGFP expression) and transgenic (mice with fosGFP expression) mice from 20 to 30 days old using predominantly 5-10% Dextran tetramethylrhodamine (a fluorescent red hydrophilic polysaccharide) as the tracer dye (Vercelli et al., 2000). A hole was drilled into the skull and the dye was injected under pressure through an electrode, while the mouse's head was held tightly in a stereotaxic apparatus. The injections were targeted around the D1 barrel. Our intention was to see labeled cells around the injection site and on the opposing hemisphere in the symmetrical position. The retrograde labeling (labeling progression in which the tracer travels along the axons into the soma of neurons) of Dextran tetramethylrhodamine enabled us to discover whether or not the long axons of "green" cells do in fact project callosally. Following the surgical procedure, the mice were allowed to recover, anesthetized, and perfused. Once the brains were removed they were fixed in 4% Paraformaldehyde (PFA), sectioned using a cryostat, and mounted on slides for imaging. It was determined that in no animal were labeled cells evident on the opposing hemisphere and symmetrical position to the injection site. We concluded that the proportion of "green" neurons that were retrogradely labeled at the injection site was low compared with "non-green" cells that were labeled at the injection site. This indicates that perhaps the processes of the "green" pyramidal cells do not project up to the surface of the brain (where the injection was made). In addition to the surgical procedure, several brains were removed from the mice following perfusion and live slices were made using a vibratome; 1, 1'-dioctadecyl-3, 3, 3', 3'-tetramethyl-indocarbocyanine perchlorate (DiI, a fluorescent lipophilic carbocyanine dye that diffuses through the membranes of cells and in doing so fills extended processes) was incorporated into the slices, and they were then wet-mounted on slides and imaged. The processes observed from the light microscopy imaging were then analyzed and measured to provide data for comparison purposes. Several brains were also extracted from mice following perfusion and then fixed in 4% PFA; after fixation, the brains were swabbed along the midline using an electrode with either DiI or Dextran tetramethylrhodamine in an attempt to clearly observe callosally projecting processes in both wild type and transgenic mice. In morphologically defining "green" cells, we are attempting to conclude that they are more active than "non-green" cells. A future implication of this finding is that levels of neural activity can be predicted by neural structure. Moreover, the detailed examination of the anatomy of the "green" cell will be useful in future experiments as it could lead to the identification of neural networks employed during learning and memory.