The lcrF gene of Yersinia pestis, the etiological agent of plague, encodes a transcription activator responsible for inducing expression of several virulence-related proteins (Yops) in response to temperature. The mechanism of this thermoregulation was investigated. Using a yopE::lacZ reporter fusion, lcrF-mediated thermal regulation was observed in Y. pestis and Escherichia coli. The lcrF gene was sequenced, the 30.8 kDa. LcrF protein identified and purified, and LcrF-dependent yopE-specific DNA binding activity was detected. A sequence similarity search revealed that LcrF exhibits 98% homology to VirF of Yersinia enterocolitica and significant homology to the carboxy termini of other members of the AraC family of transcription activators. During localization studies, a significant proportion of LcrF was found associated with the membrane fraction in E. coli. However, pulse-chase experiments indicated that this result is an artifact of fractionation. lcrF-mediated thermal induction of the yopE::lacZ reporter fusion remains intact in a Shigella flexneri virR mutant. The virR mutation is known to affect thermal induction of Shigellavirulence genes, which are also controlled by an activator in the AraC family. As a first step toward identifying the temperature-sensitive step in the regulation of yop expression, lcrF::lacZ transcriptional fusions were constructed and analyzed in Y. pestis and E. coli. The activity of the fusions was not affected by the native pCD1 virulence plasmid, an intact lcrF gene, or temperature. Thus, induction of lcrF transcription is not essential for temperature-dependent activation of yopE transcription. To confirm these results, attempts were made to identify both the native lcrF message in Y. pestis, and a lcrF-lacZ hybrid message in Y. pestis and E. coli. These attempts were unsuccessful. Examination of LcrF protein production revealed temperature-dependent expression in Y. pestis. Surprisingly, high-level T7 polymerase-directed transcription of the lcrF gene in Escherichia coli also resulted in temperature-dependent production of the LcrF protein. Pulse-chase experiments showed that the LcrF protein was stable at both 26 and 37°C, suggesting that translation rate or message degradation is thermally controlled. Comparison of the amount of LcrF protein produced per unit of message at 26 and 37°C in E. coli indicated that the efficiency of translation of lcrF message increased with temperature. mRNA secondary structure predictions suggest that the lcrF Shine-Dalgarno sequence is sequestered in a stem-loop. A model in which decreased stability of this stem-loop with increasing temperature leads to increased efficiency of translation initiation of lcrF message is presented.

The Dam-dependent mismatch repair system of Escherichia coli is part of a large network of DNA surveillance and error avoidance systems that identify and repair DNA damage. In this thesis, I have investigated the Dam-dependent mismatch repair system of E. coliand its role in the recognition and repair of DNA substrate molecules containing small insertion/deletion heterologies. This investigation was divided into two parts: the first part utilized genetic techniques to evaluate the specificity of repair and the second part utilized biochemical approaches into the recognition of insertion/deletion heterologies. I have developed a sensitive in vivo transformation system to rapidly evaluate the repair of small insertion/deletion heterologies by Dam-dependent mismatch repair. Heteroduplexes were constructed, for each state of methylation of d(GATC) sequences, by annealing single strand DNA to the linearized complementary strand of duplex DNA. The unmethylated single strand DNA was isolated from f1 phage (R408) propagated on a strain of E. coli containing the dam-16 allele (Chapter 2) to eliminate the possibility of residual Dam-methylation of d(GATC) sequences. Tranformation of E. coli indicator strains with heteroduplexes containing 1, 2, 3, 4 and 5 base insertion/deletion heterologies were scored for repair based on colony color. The results of these experiments show that the Dam-dependent mismatch repair system can recognize and repair 1, 2 and 3 base heterologies as well as repairing G/T mispairs (Chapters 3 and 4). The repair of 4 base heterologies was marginal, while no repair was observed with 5 base heterologies (Chapters 3 and 4). Repair of the 1, 2, 3 and 4 base heterologies proceeded in a Dam-dependent process that required the gene products of mutL, mutS, and I have demonstrated that MutS protein from both Salmonella typhimurium and E. coli can recognize and bind in vitro to the same 1, 2, 3 and 4 base heterologies used for the genetic studies above (Chapters 4 and 5). In fact, MutS protein binds to 1, 2 and 3 base heterologies with greater affinity than it binds to a G/T mismatch. The in vitro observation that MutS does not bind to 5 base heterologies is consistent with the in vivoobservation that 5 base heterologies are not subject to repair. I have also shown that MutS protein specifically binds to 1, 2 and 3 base heterologies since MutS protects about 25 base pairs of DNA flanking the site of the heterology from DNaseI digestion. The results of the genetic and biochemical experiments described in this thesis (and summarized above) serve to re-emphasize the importance of the role that methyl-directed mismatch repair plays in mutation avoidance, and hence in the preservation of genetic integrity.

The cell division of Saccharomyces cerevisiae is controlled by the action of pheromones at the G1 phase of the cell cycle. A general method was developed for the isolation of constitutive mutants in the pheromone response pathway. Recessive alleles of the SCG1 gene (encoding the α subunit of a G protein) were isolated as well as a dominant mutation in the STE4 gene (encoding the β subunit of a G protein). Analysis of double mutants suggested that the STE4 gene product functions after the SCG1 product but before the STE5 gene product. Double mutants carrying either scg1 or STE4Hp1 constitutive alleles together with the temperature-sensitive unresponsive mutation, ste5-3ts, showed arrest and recovery when shifted from 34° C to 22° C. Recovery from the constitutive signal was independent of the receptor. The STE4Hp1 sst2 ste5ts triple mutant was not able to recover from arrest, suggesting that an SST2-dependent mechanism is involved in recovery of the STE4Hp1 mutant from constitutive arrest. In contrast, the scg1-7 sst2 ste5ts triple mutant recovered only partially suggesting that even though SST2 gene product is probably involved in recovery of the scg1-7 mutant, this mutant can recover by an SST2-independent mechanism. This implies existence of another, SST2-independent postreceptor recovery mechanism. The scg1-null mutant do not recover from constitutive arrest (J. Hirschman, personal communication). Both recovery mechanisms probably operate at the G protein step. Isolation of a constitutive allele of STE5 allowed the definition of its site of action as being after the STE4-controlled step. In addition, constitutive activation of the pheromone pathway by STE5Hp1 mutation was found to be partially dependent on the STE4 and STE18 gene products, the β and γ subunits of a G protein. A comprehensive genetic model is presented to explain the mechanisms of signal transduction and recovery.

Chromatin structure plays an essential role in the regulation of many nuclear processes such as transcription, replication, recombination, and repair. It is generally accepted that chromatin remodeling is a prerequisite step in gene activation. Over recent years, large multisubunit enzymes that regulate the accessibility of nucleosomal DNA have emerged as key regulators of eukaryotic transcription. It seems likely that similar enzymes contribute to the efficiency of DNA replication, recombination, and repair. These chromatin remodeling complexes can be classified into two broad groups: (1) the ATP-dependent enzymes, which utilize the energy of ATP hydrolysis to increase the accessibility of nucleosomal DNA; and (2) histone modifying enzymes that phosphorylate, acetylate, methylate, ubiquitinate, or ADP-ribosylate the nucleosomal histones (for review see Kingston and Narlikar, 1999; Muchardt and Yaniv, 1999; Brown et al., 2000; Vignali et al., 2000; Strahl and Allis, 2000). The mechanism by which these two groups of large, multi-subunit enzymes function to alter chromatin structure is enigmatic. Studies suggest that ATP-dependent and histone acetyltransferase chromatin remodeling enzymes have widespread roles in gene expression and perform both independent and overlapping functions. Interestingly, although both groups of enzymes appear to be distinct, several features of these enzymes have been conserved from yeast to man. Thus, understanding the role of these similar features will be essential in order to elucidate the function of remodeling enzymes, their functional interrelationships, and may uncover the fundamental principals of chromatin remodeling. In this study, we use a combination of yeast molecular genetics and biochemistry to dissect out the function of individual parts of these chromatin remodeling machines and to understand how these large macromolecular assemblies are put together. In addition, we also investigate the mechanism by which the ATP-dependent enzymes exert their regulatory effects on chromatin structure. Structure/function analysis of Saccharomyces cerevisiaeSwi3p (conserved in SWI/SNF complexes across all eukaryotic phyla) reveals a unique scaffolding role for this protein as it is essential for assembly of SWI/SNF subunits. We have also characterized a novel motif that has homology to the Myb DNA binding domain, the SANT domain, and that is shared among transcriptional regulatory proteins implicated in chromatin remodeling. Mutational analysis of this domain in yeast Swi3p (SWI/SNF), Rsc8/Swh3p (RSC), and Ada2p (GCN5 HATs) reveals an essential function for the SANT domain in chromatin remodeling. Moreover, our studies suggest that this novel motif may be directly involved in mediating a functional interaction with chromatin components (i.e. histone amino terminal domains). We have also directly compared the activities of several members of the ATP-dependent chromatin remodeling enzymes. Surprisingly, we find that these enzymes utilize similar amounts of ATP to increase nucleosomal DNA accessibility. In as much, we show that changes in histone octamer comformation or composition is not a requirement or consequence of chromatin remodeling by SWI/SNF. Taken together, these data suggest a similar mechanism for ATP-utilizing chromatin remodeling enzymes in which disruption of histone-DNA contacts occur without consequence to the structure of the histone octamer. These data have striking implications for how we view the mechanism of chromatin remodeling.

CP1 is a sequence specific DNA-binding protein of the yeast Saccharomyces cerevisiae which recognizes the highly conserved centromere DNA element I (CDEI) of yeast centromeres. The gene encoding CP1, which was designated CEP1 for centromere protein 1, was cloned and sequenced. CEP1 encodes a highly acidic protein of molecular weight 39,400. CEP1 was mapped to a position 4.6 centiMorgans centromere distal to SUP4 on the right arm of chromosome X. Phenotypic analysis of cep1 mutants demonstrated that yeast strains lacking CP1 are viable but have a 35% increase in cell doubling time, a ninefold increase in the rate of mitotic chromosome loss, and are methionine auxotrophs. Detailed analysis of the mitotic chromosome-loss phenotype showed that the loss is primarily due to chromosome nondisjunction (2:0 segregation). During meiosis cep1 null mutants exhibited aberrant segregation of centromere containing plasmids, chromosome fragments, and chromosomes. The predominant missegregation event observed was precocious sister segregation. The mutants also displayed a nonrandom 20% decrease in spore viability. Missegregation of chromosomes accounted for some but not all of this decreased spore viability, the remainder of which is presumed to be related to the pleiotropic consequences of the cep1 mutation. Together with the observed mitotic missegregation phenotype the results are interpreted as suggesting that CP1 promotes sister chromatid-kinetochore adhesion. The following conclusions are based on my mutational analysis of CP1: (1) CP1 is normally present in functional excess, (2) the C-terminal 143 amino acids are sufficient for full CP1 function in chromosome segregation and methionine metabolism, and (3) while DNA binding is apparently necessary for function, DNA binding per se is not sufficient. All of the mutations which caused an observable phenotype affected both centromere function and methionine metabolism. In addition, a direct correlation was observed in the degree to which both phenotypes were affected by different mutations. None of the mutant proteins displayed trans-dominant effects in a wild type background; however, two nonfunctional DNA binding-competent mutants exerted a dominant negative effect on the ability of PHO4 to suppress cep1 methionine auxotrophy. The data are consistent with a model in which CP1 performs a similar function at centromeres and promoters.

The work presented here describes novel functions for centrosome proteins, specifically for pericentrin and centriolin. The first chapter describes the involvement of pericentrin in ciliogenesis. Cells with reduced pericentrin levels were unable to form primary cilia in response to serum starvation. In addition we showed novel interactions between pericentrin, intraflagellar transport (IFT) proteins and polycystin 2 (PC2). Pericentrin was co-localized with IFT proteins and PC2 to the base of primary cilia and motile cilia. Ciliary function defects have been shown to be involved in many human diseases and IFT proteins and PC2 have been implicated in these diseases. We conclude that pericentrin is required for assembly of primary cilia possibly as an anchor for other proteins involved in primary cilia assembly. The second chapter describes identification of centriolin, a novel centriolar protein that localizes to subdistal appendages and is involved in cytokinesis and cell cycle progression. Depletion of centriolin leads to defects in the final stages of cytokinesis, where cells remain connected by thin intercellular bridges and are unable to complete abscission. The cytokinesis defects seemed to precede the G0/G1 p53 dependant cell cycle arrest. Finally, the third chapter is a continuation of the cytokinesis study and it identifies pericentrin as an interacting partner for centriolin. Like centriolin, pericentrin knockdown induces defects in the final stages of cytokinesis and leads to G0/G1 arrest. Moreover, pericentrin and centriolin interact biochemically and show codependency in their centrosome localization. We conclude that pericentrin and centriolin are members of the same pathway and are necessary for the final stages of cytokinesis.

Nonsense-mediated mRNA decay (NMD) is an important mRNA surveillance mechanism conserved in eukaryotes. This thesis explores several interesting aspects of the NMD pathway. One important aspect of NMD which is presently the subject of intense controversy is the subcellular localization of NMD. In one set of experiments, the decay kinetics of the ade2-1 and pgk1 nonsense mRNAs (substrates for NMD) were investigated in response to activating the NMD pathway to determine if cytoplasmic nonsense mRNAs are immune to NMD in the yeast system. The results of these studies demonstrated that activation of NMD caused rapid and immediate degradation of both the ade2-1 and the early nonsense pgk1 steady state mRNA populations. The half lives of the steady state mRNA populations for both ade2-1 and pgk1 (early nonsense) were shortened from >30 minutes to approximately 7 minutes. This was not observed for pgk1mRNAs that contained a late nonsense codon demonstrating that activation of NMD specifically targeted the proper substrates in these experiments. Therefore, in yeast, nonsense mRNAs residing in the cytoplasm are susceptible to NMD. While these findings are consistent with NMD occurring in the cytoplasm, they do not completely rule out the possibility of a nuclear-associated decay mechanism. To investigate the involvement of the nucleus in NMD, the putative nuclear targeting sequence identified in Nmd2p (one of the trans-acting factors essential for NMD) was characterized. Subcellular fractionation experiments demonstrated that the majority of Nmd2p localized to the cytoplasm with a small proportion detected in the nucleus. Specific mutations in the putative nuclear localization signal (NLS) of Nmd2p were found to have adverse effects on the protein's decay function. These effects on decay function, however, could not be attributed to a failure in nuclear localization. Therefore, the residues that comprise the putative NLS of Nmd2p are important for decay function but do not appear to be required for targeting the protein to the nucleus. These results are in accordance with the findings above which implicate the cytoplasm as an important cellular compartment for NMD. This thesis then investigates the regulatory roles of the trans-acting factors involved in NMD (Upf1p, Nmd2p, and Upf3p) using a novel quantitative assay for translational suppression, based on a nonsense allele of the CAN1 gene (can1-100). Deletion of UPF1, NMD2, or UPF3 stabilized the can1-100 transcript and promoted can1-100 nonsense suppression. Changes in mRNA levels were not the basis of suppression, however, since deletion of DCP1 or XRN1 or high-copy can1-100 expression in wild-type cells caused mRNA stabilization similar to that obtained in upf/nmd cells but did not result in comparable suppression. can1-100 suppression was highest in cells harboring a deletion of UPF1, and overexpression of UPF1 in cells with individual or multiple upf/nmd mutations lowered the level of nonsense suppression without affecting the abundance of the can1-100 mRNA. These findings indicate that Nmd2p and Upf3p regulate Upf1p activity and that Upf1p plays a critical role in promoting termination fidelity that is independent of its role in regulating mRNA decay.

The overall emphasis in this thesis is the study of the regulation of virus infections by natural killer (NK) cells. In initial analyses, vaccinia virus (VV)-infected cells were found to be more sensitive to NK cell-mediated lysis during a discrete period of time post-infection. This enhanced susceptibility to lysis correlated with enhanced triggering (but not binding) of the effector cells and a concomitant decrease in target cell H-2 class I antigen expression. Furthermore, VV-infected cells became resistant to lysis by allospecific cytotoxic T lymphocytes (CTL) at a time when they were very sensitive to killing by NK cells or VV-specific CTL. This suggested that alterations in class I MHC antigens may affect target cell sensitivity to lysis by NK cells. The hypothesis that viral peptide charging of H-2 class I molecules can modulate target cell sensitivity to NK cell-mediated lysis was tested by treating target cells with synthetic viral peptides corresponding to the natural or minimal immunodominant epitopes defined for virus-specific CTL, and then target cell susceptibility to NK cell-mediated lysis was assessed. None of the 12 synthetic viral peptides used were able to significantly alter target cell lysis by NK cells under any of the conditions tested. In order to determine if H-2 class I molecules were required in the regulation of a virus infection by NK cells in vivo, intact or NK depleted (treated with anti-asialo GM1 antiserum) β2-microglobulin-deficient [β2m (-/-)] mice, which possess a defect in H-2 class I antigen expression, were infected with the prototypic NK-sensitive virus, murine cytomegalovirus (MCMV). In anti-asialo GM1-treated β2m (-/-) mice, as well as in β2m + (H-2 class I normal) control mice also treated with anti-asialo GM1 a significant enhancement in splenic MCMV titers as compared to NK-intact animals, was observed. When thymocyte expression of H-2 class I molecules (H-2Db) in normal mice was analyzed, it was found that following MCMV infection, H-2Db expression was significantly greater than the low level of expression found in uninfected thymocytes. In marked contrast, thymocytes from β2m (-/-) mice did not display any detectable H-2Db before or after infection. These in vivoresults demonstrate that NK cells can regulate a virus infection, at least in the case of MCMV, independent of H-2 class I molecule expression. Thymocytes from uninfected normal mice were found to be very sensitive to NK cell-mediated lysis, whereas those from MCMV-infected animals were completely resistant, presumably due to the protective effects of MCMV-induced interferon (IFN). However, thymocytes from MCMV-infected β2m (-/-) mice were only slightly protected from lysis by NK cells, consistent with the inverse correlation between MHC class I antigen expression and sensitivity to NK cell-mediated lysis. These results provide in vivoevidence suggesting a requirement for MHC class I molecules in IFN-mediated protection from lysis by NK cells. In addition to the analysis of H-2 class I molecules on target cells, the identity of a molecule present on the surface of all NK cells and other cytotoxic effector cells, which is recognized by a monoclonal antibody (mAb) generated in this laboratory designated CZ-1, and can also modulate NK cell triggering, was also of interest. This laboratory has previously reported that this antigen is upregulated on cytotoxic (and other) lymphocytes following a virus infection in vivo, or upon activation in vitro. Using competitive FACS analysis and fibroblasts transfected with various isoforms of CD45, it was found that mAb CZ-1 recognizes a sialic acid-dependent epitope associated with a subpopulation of CD45RB molecules.

P22 lysozyme, encoded by gene 19, is an essential phage protein responsible for hydrolyzing the bacterial cell wall during lytic infection. P22 lysozyme is related to T4 lysozymein its mode of action, substrate specificities, and in its structure. Gene 19 was located on the phage genome, subcloned, and then sequenced. lysozyme was produced in large quantities and purified for biochemical characterization and for crystallograpic studies. Gene 19consists of 146 codons, and encodes a protein with a molecular weight of 16,117. Amber mutations were created in gene 19 by in vitro primer-directed mutagenesis. The mutations were crossed by homologous recombination onto the phage genome. The phages bearing the amber mutations in gene 19 were screened for the ability to grow on six different amber suppressor strains. Amino acid substitutions that resulted in nonfunctional or less functional lysozyme were determined. Of 60 possible amino acid substitutions at 11 different sites in P22 lysozyme, 20 are deleterious. The phage bearing amber mutations in gene 19that failed to grow on given suppressor strains were reverted and second site intragenic revertants were obtained. The mutations were sequenced. A substitution of serine for glutamine at residue 82 is compensated for by changing residue 46 from serine to leucine. This single change enables the phage to form a plaque at 300C but not at 400C. When the triple change asn42->lys; ser46->leu; and ser43->pro is present the lysozyme produced is no longer temperature sensitive. The crystal structure of P22 lysozyme is not yet solved. Assuming that the structures of T4 lysozyme and P22 lysozyme are similar, one can examine the positions of equivalent residues in the T4 lysozyme structure. The spatial arrangement of the residues changed by the secondary site mutations and the original substitution can then be visualized. The mutations discussed above all map far from the original mutation on the T4 three dimensional model. A substitution of leucine for tyrosine at position 22 is compensated for by the double mutation of arg18->ser and ser23->lys. When the equivalent residues are mapped on the T4 three dimensional model the changes map in close proximity to the original mutation.

The human erythrocyte glucose transport protein (GLUT1) interacts with, and is regulated by, cytosolic ATP. This study asks the following questions concerning ATP modulation of GLUT1 mediated sugar transport. 1) Which region(s) of GLUT1 form the adenine nucleotide-binding domain? 2) What factors influence ATP modulation of sugar transport? 3) Is ATP interaction with GLUT1 sufficient for sugar transport regulation? The first question was addressed through peptide mapping, n-terminal sequencing, and alanine scanning mutagenesis of GLUT1 using [32P]-azidoATP, a photoactivatable ATP analog. We then used a combination of transport measurements and photolabeling strategies to examine how glycolytic intermediates, pH, and transporter oligomeric structure affect ATP regulation of sugar transport. Finally, GLUT1 was reconstituted into proteoliposomes to determine whether ATP is sufficient for the modulation of GLUT1 function in-vitro. This thesis presents data supporting the hypothesis that residues 332-335 contribute to the efficiency of adenine nucleotide binding to GLUT1. In addition, we show that AMP, acidification, and conversion of the transporter to its dimeric form antagonize ATP regulation of sugar transport. Finally, we present results that support the proposal that ATP interaction with GLUT1 is sufficient for transport modulation.