2006 Summer Scholar Participants-HHMI Undergraduate Program - Carnegie Mellon University

2006 HHMI Summer Scholar Participants

Image of Victoria Aveson

Victoria Aveson, Carnegie Mellon University
Mentor: Dr. Tina Lee

Anionic Phospholipid-binding and ER Network Formation Functions of NDKB Are Required for Its COPII Assembly Function

The export of proteins from the ER occurs by vesicle formation, which is mediated by cytosolic coat protein complex II (COPII). The minimal machinery for COPII mediated vesicle formation is known, but there are more components of the system that are not yet fully understood. One such component is Nucleoside Diphosphate Kinase B (NDKB). Previous experiments have shown that NDKB is required for COPII assembly and ER network morphogenesis. A lipid-binding assay has shown that NDKB binds directly to anionic phospholipids, and the amino acids responsible for lipid-binding have been mapped to Lys 56 and Arg 58. Mutation of these residues greatly reduced anionic lipid binding, localization of NDKB to the ER membrane, ER network formation and COPII assembly. Mutation of these residues was not, however, enough to abolish COPII assembly. This raised the question of whether the ER network formation function of NDKB accounts for its COPII assembly function, or whether NDKB has an additional function that does not depend on ER network morphology. To address this question, two positively charged residues near the lipid-binding site, Lysine 49 and Lysine 66, were introduced into the K56E,R58E double mutant and the resulting protein tested for COPII assembly activity. The results indicate that the quadruple mutant version of NDKB (K49A,K56E,R58E,K66D) is completely defective in COPII assembly, suggesting that the lipid binding and ER network functions of NDKB are likely to account for its COPII assembly activity.

Image of Simon Hucko

Simon Hucko, Carnegie Mellon University
Mentor: Dr. Newell Washburn

Analysis of Aptameric Structure and Binding to a Pro-Inflammatory Cytokine

Cytokines are proteins that act as signaling molecules in the human inflammation pathway. Interleukin1ß (IL1B) is a central member of this class of proteins. Cytokines are active in a wide range of disorders including arthritis, chronic inflammation and wound healing, and have even been linked to cancer development. In order to regulate these inflammatory responses, it is desirable to develop a molecule that will bind IL1B tightly. Candidates for binding molecules can be developed from libraries of single-stranded DNA (ssDNA), known as aptamers. The purpose of this project was to develop and characterize an aptamer with a high binding affinity for IL1B. Aptamers were first selected from a large (10 13 individual molecules) randomized library of ssDNA. This was accomplished using a high-throughput technique, in which molecules with high binding affinities for the target protein were separated out from the rest of the library. Once isolated and purified, the top candidates were sequenced. A cursory structural analysis of each aptamer was carried out using software to come up with a theoretical secondary structure. Using these structures, an aptamer of interest was selected based on structural stability and conservation of folding between different structures. Aptamer analysis, however, has proven to be more difficult than first hypothesized. The first round of aptamer selection was carried out using primers that were later discovered to have a large self-complementary region. This lead to self-dimerization, and other structural oddities that made analysis nearly impossible. While we are continuing to troubleshoot our analysis tools, a new round of SELEX has been started using non-complementary primers.

Image of Christina Onorato

Christina Onorato, Carnegie Mellon University
Mentor: Dr. Justin Crowley

Development of Columnar Architecture in Visual Cortex: Seeking Accord Between Anatomical and Proteomic Level Data

The goal of this project is to investigate the mechanisms underlying the formation of two functionally defined columns in primary visual cortex: ocular dominance columns and orientation columns. Two approaches were taken to investigate the normal course of events that occur during the formation of these structures. First, anatomical reconstructions of horizontal axonal connections of layer 2/3 pyramidal cells in primary visual cortex of tree shrews were completed using Neurolucida tracing software on previously fixed tissue collected at different developmental stages. In the normal adult, these axons form connections that have both axial and modular specificity consistent with the orientation preference of the column in which their cell bodies are found. Analyses of axonal bouton distributions indicate that axial specificity develops by P18, prior to eye opening and the onset of visual experience.

As a compliment to this work, we are collaborating on a method to analyze changes in protein levels between different cortical columns during development. A fluorescence based two-dimensional difference gel electrophoresis (DIGE) approach was taken to identify differences in protein expression between cortical columns. DIGE analysis compares two samples of fluorescently labeled proteins based on mass and isoelelectric focusing point. In these studies, cortical samples are prepared by identifying either ocular dominance columns or orientation columns by in vivo intrinsic signal optical imaging, labeling physiologically identified columns with fluorescent latex microspheres and micro-dissecting labeled columns for DIGE analysis. In primary visual cortex we perform pairwise comparisons of our investigation of ocular dominance and orientation columns at opposite tuning preferences. In support of ocular dominance development we are also comparing protein level differences between eye specific lateral geniculate layers and between nasal and temporal retina, areas that we hypothesize will have protein distribution patterns correlated to those of ocular dominance columns.

Image of Kimberly Parks

Kimberly Parks, Carnegie Mellon University
Mentor: Dr. Lynne Reder

Trying to Understand the Other-Race Effect in Face Recognition

The Other-Race Effect (ORE) is defined as difficulty remembering or recognizing faces of other races. One possible explanation for ORE is lack of expertise in encoding other-race faces. Even with much research on face recognition and ORE, it is still debated why performance worsens when recognizing faces of another race. Our hypothesis is experience affects encoding ease and encoding of unfamiliar other-race faces consumes more working memory. The experimental methods include stimuli collection, experimental design, forming an IRB protocol, subject running, data analysis and results. The design is a 2x2x2x2 quasi-factorial with a study session and test portion.   There are three face pairs categories in study: Caucasian-Caucasian (C-C), Caucasian-Asian (C-A), and Asian-Asian (A-A). During study, subjects will decide which face, right or left, is more tired. Test session performance will be a function of a face's race and the pair the face was originally seen in. At test, subjects view single pictures and decide, on a confidence level, how sure he/she is the picture was seen during study. We hypothesize that for a Caucasian participant, a Caucasian face should be harder to remember if seen in a C-A pair while an Asian face seen in a C-A pair should be better remembered than those in an A-A pair.   Evidence of ORE was shown using the dependent measure (d') of performance. Additionally, the race of the face seen at study with a tested face affects the ability to recognize the face tested. This suggests the familiarity of race affects how much working memory is required to encode a face.   We assert greater working memory load for an unfamiliar race can decrease performance across the face pairs.   This experiment bolsters the claim that familiarity affects ease of encoding (Diana & Reder, in press).

Image of Daniel Smith

Daniel Smith, Carnegie Mellon University
Mentor: Dr. Justin Crowley

Optimization of In VivoTwo-Photon Laser Scanning Microscopy by Physiological Motion Compensation

Two-photon laser scanning microscopy (2PLSM) applied to the brain in vivo allows fine anatomical features such as dendritic spines and axonal boutons to be visualized as much as 500 m m from the surface of the neocortex in time-lapse imaging experiments. However, averaging of multiple scans at one time point and comparison of features across multiple time points are complicated by normal physiological events, primarily heartbeat and respiration. While this problem is present in any in vivo experiment, it is especially significant with laser scanning microscopy because of its high resolution and the fact that the pixels in an image volume are acquired sequentially. In fact, one can calculate the heart rate and respiration rate of an animal by examining their signatures in 2PLSM image data. To address these issues, we developed a real-time hardware and software control system to coordinate the image acquisition of a two-photon laser-scanning microscope in synchrony with animal physiology. In addition, the system we designed concurrently collects high-resolution, time-stamped physiological and laser scanning data for post hoc analysis. To compensate for cardiac motion, initiation of planar scans was triggered from the peak of the q-r-s complex of the electrocardiogram (EKG) recording of the animal being imaged. This ensures that cardiac variation between sequentially acquired images of the same focal plane is minimized. Initial analyses suggest that the cardiac triggering does indeed decrease variation between images of the same plane. These preliminary studies also suggest that if artificial respiration of the imaging subject is paused during image acquisition, variation in image data due to respiration would be eliminated. Synchronization or elimination of these sources of physiological motion would ensure that the same volume of tissue could be imaged in a consistent way in time-lapse experiments. Finally, the effect of different anesthetic regimens on physiologically-driven brain motion is also being assessed. Anesthesia induction with urethane appears to result in a significant cardiac artifact with little respiratory artifact present. This is likely due to urethane's minimal effect on blood pressure and respiration. Anesthesia induction with ketamine or ketamine/xylazine followed by maintenance with 1-2% isoflurane and nitrous oxide has been previously been found to decrease mean arterial blood pressure. This lowers the magnitude of the cardiac artifact to less than 2 m m in our 2PLSM data, however respiration is slowed and breath volume is increased in order to compensate, contributing to a greater respiratory artifact. Our ongoing work is intended to decisively quantify the improvement in data quality associated with cardiac triggering, respiratory pause and anesthetic regimens.

Image of Adam Suhy

Adam Suhy, Carnegie Mellon University
Mentor: Dr. Bruce Armitage

Selection, Evolution and Characterization of Peptide Nucleic Acid Dependent Deoxyribozyme

It has been determined that RNA and DNA can form secondary structures that are capable of catalyzing chemical reactions. In previous work by Santoro and Joyce, a deoxyribozyme (DNAzyme) that cleaves a specific RNA sequence from the transcription initiation sequence of an HIV-1 strain was selected from a random pool. Using similar methods, this project will attempt to create a more active, more effective DNAzyme by employing a peptide nucleic acid (PNA) complementary to a constant region of the DNA pool, and then characterize that DNAzyme. PNA is an analog of DNA that uses a neutral peptide-like backbone in place of the charged sugar-phosphate backbone of DNA. Synthesis of PNA allows for the incorporation of nucleotide bases as well as a wide variety of other functional groups, such as amino acids. First, a PNA that incorporates two lysine residues and one histidine residue was synthesized. The random DNA pool was then bound to streptavidin beads in a column through an RNA linker. This linker acts as the substrate to the DNAzyme that may form. A catalytic DNA then cleaves itself from the column and is collected in the eluant. These random sequences that are able to cleave themselves from the column without PNA present are considered "losers". After the losers were removed from the column, a reaction buffer containing the synthesized PNA will be added. The molecules that cleave themselves at this stage are considered "winners" because they depend on the PNA cofactor for their activity, and they are collected. Polymerase chain reaction (PCR) is used to amplify the winners, which is then reattached to streptavidin beads through the RNA substrate linker and is selected for again. Through several rounds of this selection and amplification random mutations in the sequence accrued during PCR will eventually lead to a faster acting, more effective enzyme that can be selected for using various selective pressures, such as shorter reaction times. Following the selection, a characterization study will be conducted to determine the kinetics of the selected enzyme. Comparisons of other enzymes selected for in the same manner, but without PNA, will be done to determine if addition of the PNA is significant. The ability to create a DNAzyme that can cleave a specific product can create a new method of treating pathogenic disease. By cleaving RNA of a virus or bacteria, higher efficiency can potentially be obtained compared to antisense methods of silencing their activity. Selecting for an efficient DNAzyme will also help to shed light on what makes a good enzyme. This knowledge can be useful in designing a new enzyme for a similar purpose. The use of PNA can also be expanded to assist in other types of reactions requiring catalysis.

Image of Blake Sweeney

Blake Sweeney, Carnegie Mellon University
Mentor: Dr. Russell Schwartz

The Effects of Hierarchical Assembly Pathways on Capsid Formation

Virus self-assembly is an imperfectly understood, but important topic. Studying this process may lead to antiviral drugs as well as an ability to design self-assembly systems for use in nanotechnology. Virus self-assembly is a complex process where many identical monomers must spontaneously form into a complex shell, known as a capsid. Capsids must assemble quickly and accurately. In order to understand the complexities of this process a discrete-event simulator using the Local Rules model has been constructed. This model explains the global behavior of a virus shell through simple binding rules for each monomer. Recent incorporation of a spring-force model into this simulator enables the study of malformation in capsids. Hierarchical assembly is a possible mechanism in capsid assembly, where certain well-defined structures form quickly and then assemble into a complete shell. However, little is known about how prominent this mechanism is. The purpose of this work is to study how the hierarchical assembly pathway would affect the virus shell's ability to form efficiently. First, the effect of hierarchical assembly upon kinetics of normal assembly was studied. This was accomplished in hypothetical Cube and Icosahedron models through the comparison between assembly systems with identical binding affinities for all binding sites versus those with varied binding affinities. Results thus far agreed with the hypothesis that in certain parameter domains hierarchical assembly would increase assembly yields at equilibrium. However, experiments with more complex shells are needed to investigate the generality across all capsid self-assembly model systems. Initial experiments have also been done using the simulator with spring force model to study the role of hierarchical assembly mechanism in the robustness of viruses against malformation. Future experiments will test the accuracy of hierarchical pathways in capsid self-assembly despite the initial distortion in the binding sites of subunits.

Benjamin Williams

Benjamin Williams, Carnegie Mellon University
Mentor: Dr. Javier López

Mechanisms of recursive splice site recognition inDrosophila melanogaster

Expression of genes in multicellular animals requires excision of introns from the pre-messenger RNA and splicing together of the exons, which contain the protein-coding information. It was assumed that all introns in eukaryotic pre-mRNAs are removed as single entities by splicing directly from one exon to the next. Recently, it was dicovered that many large introns in Drosophila are actually removed co-transcriptionally in a stepwise fashion by a process called recursive splicing. Computational analysis and preliminary experimental evidence suggests that this process also operates in mammals, and the patterns of phylogenetic conservation indicate that it plays an important role in the expression of genes with introns longer than 10 kb. Recursive splice sites (also known as ratcheting points: RPs) consist of sequences that function sequentially as 3' and 5' splice sites. The active 5' splice site is regenerated at the precise junction that results from use as a 3' splice site, so the use of recursive splice sites leaves no trace in the final mRNA. Little is known about how recursive splice sites are recognized by the splicing machinery or how their sequential activity is coordinated. A computational analysis was used to identify candidate splicing enhancers in the vicinity of recursive splice sites, and the function of such enhancers has been verified using site RP3 from the Ultrabithorax ( Ubx ) gene as a model. In addition, intronic silencers have been identified that suppress the use of cryptic splice sites in the vicinity of recursive splice sites. All of the auxiliary sequences identified so far using the Ubx model are required to activate the regenerated 5' splice site and to avoid use of incorrect competing 5' splice sites. These observations suggest that the RP is recognized as a 3' splice site by default, but that a special mechanism is required to activate the regenerated 5' splice site after the first round of splicing. This hypothesis is consistent with a statistical analysis of recursive splice sites in Drosophila , which reveals distinctive sequence features of the 3' splice site component, including an extended polypyrimidine tract with an enhanced bias for U residues; the consequence is a doubling of the information content for the 3'ss component of RPs compared to regular exonic splice sites. In contrast, the 5' splice sites of RPs are similar to those of exons. My project has been to test these ideas by weakening the 3' splice site component of Ubx RP3 and examining the consequences for RP recognition and coordination. For this purpose I introduced mutations that delete the normal branchpoint, reduce the U content of the polypyrimidine tract, or replace the preferred pyrimidine at -3 with an A, which should also strengthen the 5'ss component. I then used a minigene transfection system to test whether these mutations led to failure to recognize the RP, to initial use of the RP as a 5' splice site, or to other aberrations in RP function. My results show that even complete inactivation of the 3'ss component does not induce use of RP3 as a 5'ss, arguing against simple competition and implying the involvement of still unidentified auxiliary elements. This is also consistent with my finding that removal of the branchpoint and polypyrimidine tract of RP3 led to aberrations in choice of the upstream 5'splice site, suggesting that some mechanism still pairs the normal exonic 5' splice with the mutant RP3 sequence and prevents use of the former. I also tested the effect of deleting a conserved downstream element. My preliminary results suggest that this deletion stimulates inappropriate use of RP3 as a 5'ss.

Image of Elizabeth Wiltrout

Elizabeth Wiltrout, Carnegie Mellon University
Mentor: Dr. Danith Ly

Determination of the Cellular Uptake of Guanidinium Peptide Nucleic Acid

Guanidinium Peptide Nucleic Acid (GPNA) has been found to bind sequence-specifically to DNA and RNA and be taken up by human somatic and embryonic stem cells. Our preliminary experimental results have shown that GPNA diffuse into zebrafish and are non-toxic at low concentrations. We plan to determine the mechanism of cellular uptake of GPNA in the zebrafish organism Danio rerio . We will do this by studying the efficiency and intracellular location of various GPNA oligomers and the effect of removing the positive charge from the guanidinium group inside the cell. We have synthesized the necessary monomers for the oligomers, and we are currently synthesizing the oligomers. In collaboration with the Bahary zebrafish laboratory at the University of Pittsburgh Medical Center (UPMC), we will study the uptake of each oligomer in vivo . Danio rerio is often used to study many diseases common to humans, and knowing the mechanism of GPNA cellular uptake will further the development of GPNA applications for correcting genetic diseases, addressing the cellular origin of cancer, cancer treatment, and gene regulation.