Heritable trait variation and differential fitness among trait variants are conditions required for pollinator-mediated natural selection on attractive traits like floral scent. However, previous studies of floral scent have focused on assessing evolution through stereotypical pollination syndromes and often fail to evaluate the conditions of natural selection. I assess the potential for pollinator-mediated natural selection on the floral scent of color polymorphic Hesperis matronalis (Brassicaceae). A study that assessed the importance of shared biochemistry between floral scent and color found significant diurnal variation in scent emission and a population-specific effect of floral color on floral scent composition. Specifically, purple morphs tended to be similar, while white morphs tended to differ significantly. A survey of five wild populations across part of H. matronaliss introduced range supported this trend, particularly for aromatic composition; both scent composition and overall emission rates varied among populations. An experiment comparing scent profiles of plants grown in a common garden environment suggested both environmental and genetic causes of among-population variation. A three-part study assessed the relationship between scent and fitness. Experimental augmentation of floral targets with color-specific floral scent revealed increased syrphid fly visitation in response to increased scent emission rate, predicting a positive linear relationship between plant fitness and emission rate. An experiment limiting pollinator access to plants showed this expected relationship for plants exposed to diurnal pollinators, but no relationship for plants exposed to night pollinators. In contrast, I found a negative quadratic relationship between daytime emission rate and fitness across plants in four large wild populations, suggesting possible costs of scent production under wild conditions, i.e., attraction of herbivores or energetic expenditures. Overall, this dissertation suggests strong potential for pollinator-mediated natural selection on H. matronalis floral scent. Additionally, the results illustrate the importance of assessing all conditions necessary for natural selection of floral scent rather than relying on the observational pollination syndrome framework to describe the evolutionary trajectory of a species.

The genomes of root-nodulating, nitrogen-fixing rhizobia that have been completely sequenced contain multiple copies of genes that encode the heat-shock transcription factor RpoH and the chaperone GroEL-GroES. Sinorhizobium meliloti maintains two rpoH genes, four groESL operons, and a single groEL gene. Mutations in some of these genes result in symbiotic defects: an rpoH1 mutant cannot fix nitrogen in nodules, an rpoH1 rpoH2 double mutant cannot form nodules, and a groEL1 mutant cannot fix nitrogen in nodules. My work has sought to further characterize the roles of multiple rpoH and groEL genes during growth and symbiosis. In E. coli, groESL is the key target of RpoH. However, I showed that S. meliloti rpoH suppressor mutants do not overproduce GroEL, and overexpression of groESL does not bypass the rpoH mutant defects. In addition, RpoH1 controls expression of only groEL5, which is not required for symbiosis, and RpoH2 does not control expression of any of the groEL genes. Therefore, the requirements for RpoH1 and RpoH2 during symbiosis cannot be explained solely by loss of GroEL-GroES production, and there must be other crucial targets. To determine what genes are controlled by RpoH1 and RpoH2, I performed microarray experiments to compare global gene expression profiles between wild-type and rpoH mutant cells. Although the regulon of RpoH1 is incomplete, the results indicate that the RpoH1 and RpoH2 regulons at least partially overlap with each other and with the E. coli RpoH regulon. To uncover functional redundancies among the groE genes during growth and symbiosis, I constructed strains containing all possible combinations of groEL mutations. Although a groEL1 groEL2 double mutant could not be constructed, the 1-3-4-5- and the 2-3-4-5- quadruple mutants are viable, demonstrating that like other bacteria S. meliloti requires one groEL for growth. Analysis of the quadruple mutants during symbiosis indicates that only groEL1 is necessary and sufficient for symbiosis. The groEL1 groESL5 double mutant is temperature sensitive unlike either single mutant, suggesting overlapping roles during stress response. I conclude that groESL1 encodes the housekeeping GroEL-GroES and that groESL5 is specialized for stress response.

Establishing patterns of differentiation is an important theme in developmental biology. A key mechanism involved in creating these patterns of differentiation is the establishment and interpretation of transcription factor gradients. The Drosophila transcriptional repressor Brinker (Brk) is expressed in lateral-to-medial gradients across the anterioposterior axis of the wing imaginal disc where it negatively regulates the spatial patterns of expression of genes including spalt (sal) and optomotor-blind (omb); the precise pattern of expression of these targets is determined by their sensitivity to repression by Brk so that the sal domain is narrower than that of omb largely because it is repressed by lower levels of Brk than omb. The brk gradient is established by an inverse gradient of BMP signaling through the secreted BMP homolog Decapentaplegic (Dpp): the intracellular Smad effectors of Dpp signaling, pMad and Medea, bind together with the repressor protein Schnurri (Shn) to silencer elements at the brk locus and repress the activity of a constitutive enhancer. My studies have revealed that the generation of the brk gradient is not simply a precise negative read-out of the dpp gradient; Brk must also negatively autoregulate its own expression by interacting with the pMad/Medea/Shn repressor complex. Additionally, I have demonstrated that this Brk/pMad/Medea/Shn repressor complex alone cannot establish the graded profile of brk but that an additional positive cis-regulatory element that is activated by pMad is required. This may provide the first example of a transcription factor both activating and repressing the same gene, brk, in the same cells at the same time via two different response elements. In conclusion, generating the brk gradient requires at least two positive and two negative inputs: constitutive activation by an activator, Brk negative autoregulation, and both activation and repression by pMad. Generating a stable expression gradient appears to be much more complex than previously thought and may reflect the importance of multiple inputs in generating intermediate levels of gene expression rather than a simple on/off threshold response.

A key observation from natural communities is that different species vary widely in their abundance and distribution. Understanding what factors are most important in explaining this variation is a fundamental goal of ecology. Here I take a comparative phylogenetic approach to address this problem. Using two clades of diverse tropical understory plants, I use information garnered from species' evolutionary relationships to test hypotheses about why some species are common while other species are rare. In a study of geographic range size variation of Neotropical Piper (Piperaceae) species, I used published DNA sequences to infer species' divergence times and herbarium collection records to infer their range sizes. I found that younger species have significantly smaller range sizes than older species. I examined a similar question using Mesoamerican Psychotria subgenus Psychotria (Rubiaceae) species. To infer the evolutionary relationships of species, I sequenced DNA from two loci of > 60 species in this clade. I concurrently inferred the phylogenetic relationships and absolute divergence times of species using a Bayesian relaxed-molecular clock method. I calculated two metrics of geographic range size using herbarium collection records, and predicted species' potential ranges using species distribution modeling. I found that Mesoamerican Psychotria subgenus Psychotria species have diversified primarily over the past 17 million years (Mya), and species largely fall into two clades that diverged approximately 15 Mya. In one clade, younger species have colonized a significantly smaller proportion of their potential range extent than older species. Finally, using two genera in the clade Psychotrieae (Rubiaceae), I examined the impact of phylogenetic relatedness on the co-occurrence and variation in abundance among these species in Costa Rica, Central America. Using data collected on 240 transects nested in seven assemblages across Costa Rica and a phylogenetic hypothesis of species relationships based on DNA sequences, I found that Psychotrieae assemblages are significantly phylogenetically overdispersed, indicating that co-occurring species are less related than expected by chance. Within one heavily sampled assemblage, I found an inverse relationship between species' phylogenetic relatedness and their variation in abundance. The opposite trend was found across assemblages, where phylogenetic relatedness and variation in abundance were positively correlated.

Genetic manipulations of Mycobacterium tuberculosis are complicated by its slow growth, inefficient DNA uptake, and relatively high levels of illegitimate recombination. Most methods for construction of gene replacement mutants are lengthy and complicated, and the lack of generalized transducing phages that infect M. tuberculosis prevents simple construction of isogenic mutant strains. Characterization and genomic analysis of mycobacteriophages has provided numerous molecular and genetic tools for the mycobacteria. Recently, genes encoding homologues of the Escherichia coli Rac prophage RecET proteins were revealed in the genome of mycobacteriophage Chec9c. RecE and RecT are functional analogues of the phage &#x3BB; Red recombination proteins, Exo (exonuclease) and Beta (recombinase), respectively. These recombination enzymes act coordinately to promote high levels of recombination in vivo in E. coli and related bacteria using short regions of homology, facilitating the development of a powerful genetic technique called 'recombineering.' Biochemical characterization of Che9c gp60 and gp61 demonstrated that they possess exonuclease and DNA binding activities, respectively, similar to RecET and &#x3BB; Exo/Beta. Expression of gp60/gp61 in M. smegmatis and M. tuberculosis substantially increases homologous recombination such that 90% of recovered colonies are the desired gene replacement mutants. Further development of this system demonstrated that Che9c gp61 facilitates introduction of selectable and non-selectable point mutations on mycobacterial genomes at high frequencies using short (<50 nt) ssDNA substrates. The mycobacterial recombineering system provides a simple and efficient method for mutagenesis with minimal DNA manipulation. While it is clear that similar phage-encoded recombinase homologues are rare, they can be readily identified by genomic studies and by in vivo characterization. Several putative recombination systems have been identified in mycobacteriophages Halo, BPs, and Giles, and recombineering of drug-resistance point mutations provides an easy assay for recombinase activity. Analysis of recombinases from various phages – including &#x3BB; Beta and E. coli RecT – indicates that these proteins function best in their native bacteria. The mycobacteriophage-encoded proteins exhibited varying levels of activity, suggesting that analysis of multiple proteins is required to achieve optimal recombination frequencies. The apparent species-specific nature of these recombinases suggests the recombineering technology could likely be extended to any bacterial system through characterization of host-specific bacteriophages.

In order to inject their DNA into the bacterial cytoplasm and establish infection, bacteriophages must ensure their genetic material successfully traverses both the bacterial membrane(s) and the layer of peptidoglycan surrounding the host cell. Phages accomplish this in a variety of ways, and some have virion-associated murein hydrolase enzymes that facilitate this process, particularly in conditions where the peptidoglycan is highly cross-linked. Phages that infect the mycobacteria must also contend with these barriers to infection, as well an impermeable layer of mycolic acids that decorates the cell surface; however, the mechanisms by which they do this are mostly unknown. In this regard, three small sequence motifs have been identified within mycobacteriophage tape measure proteins (TMPs) − extended molecules that span the tail lumen and determine its length − at least two of which have similarity to host proteins with muralytic activity. This suggests that phages may utilize regions of the TMP, which because of its location within the tail might be uniquely primed for host interaction, to facilitate localized peptidoglycan hydrolysis and DNA injection. The focus of this study is the motif found in the TMPs of mycobacteriophages Barnyard and Giles that has identity to a group of bacterial proteins known as resuscitation promoting factors (Rpfs). These Rpf proteins stimulate growth of non-growing bacteria and seem to exert their activity by cleaving inert peptidoglycan in the cell wall. Notably, the Barnyard Rpf Motif is contained within a 70 kDa C-terminal cleavage product of TMP, which appears to be cell wall- and/or membrane-associated during infection. Further, mycobacteria expressing TMP fragments containing this motif show aberrant behavior in culture and on solid media, and hybrid proteins in which the Rpf domain of the Micrococcus luteus Rpf protein is replaced with either of the phage motifs have muralytic activity in vivo. A recombineering-based method for generating mutations on lytically replicating mycobacteriophages has been developed and utilized to make multiple mutations in the Giles TMP motifs. Mutant phages infect host cells in late-stationary phase with a reduced efficiency, an observation that further supports a role for these motifs in cell wall hydrolysis during infection.

Replicative helicases are essential enzymes in DNA replication that separate duplex DNA into single strands to be used as templates by polymerases. The identity of these helicases in prokaryotes and viruses has been known for some time, but there is controversy in defining the eukaryotic replicative helicase. Fifteen years of in vivo work and comparison to other helicases suggests that the minichromosome maintenance proteins two through seven (Mcm2-7 complex) play this role, but in vitro DNA unwinding has never been demonstrated. In order to address this dilemma, I analyzed the biochemical activities of the Saccharomyces cerevisiae Mcm2-7 complex and the contributions of its conserved ATPase motifs to these activities. My findings indicate that Mcm2-7: 1) binds single-stranded DNA with high affinity in an ATP-dependent manner, 2) has six distinct active sites that differentially contribute to its biochemistry, and 3) is sensitive to anionic conditions in vitro. Elucidation of these characteristics led to the discovery that Mcm2-7 helicase activity is stimulated by large anions, and I was able to demonstrate bona fide in vitro DNA unwinding for the first time. The culmination of this work is the view that the Mcm2-7 complex is the true eukaryotic replicative helicase and the emerging view that Mcm2-7 ATPase active sites have evolved specialized functions within the complex.

DNA bending by DNA binding proteins is required to facilitate a myriad of essential cellular processes, including genome packaging, the formation of multi-protein complexes necessary for the expression of genetic information, the regulation of gene expression, and the correct positioning of recognition and catalytic elements required for site-specific hydrolysis of DNA. One mechanism proposed for protein induced DNA bending is that protein mediated neutralization of negatively charged phosphates on one face of the DNA produces a collapse of the DNA toward the neutralized surface. In order to further understand the role of phosphate neutralization in DNA bending, I have manipulated the energy required for achieving the 50o axial bend in the EcoRV endonuclease-DNA complex by either removing cationic protein side chains that contact DNA phosphates and/or replacing the charged phosphate with an uncharged methylphosphonate. I present evidence that neutralization of particular phosphates, positioned on the concave face of the bound DNA, can contribute favorably not only to the formation of the EcoRV-recognition complex but also to cleavage of the GATATC site. In addition, synergistic effects are observed when particular combinations of phosphates are neutralized. Fluorescence resonance energy transfer studies show that there is no significant difference in the degree of DNA bending in unmodified and modified complexes, implying that phosphate neutralization modulates the energetic cost of bending rather than the extent of bending. Further, vant Hoff analyses indicate that removal of interphosphate repulsion by phosphate neutralization contributes favorable enthalpy to EcoRV-DNA complex formation, and molecular dynamics simulations show that this favorable enthalpy does not derive from the formation of new contacts between the introduced methyl group and the protein. Taken together, these results support the model that asymmetric phosphate neutralization by a site-specific protein promotes DNA bending. My work marks the first in depth thermodynamic analysis of the impact of phosphate neutralization on protein-induced bending by a DNA bending protein. Further, it shows that this is an energetic strategy employed by proteins to overcome the energetic cost of DNA bending.

The evolution of environment-specific trait expression (i.e., phenotypic plasticity) represents a seemingly unbeatable evolutionary strategy because a plastic organism may be able to maximize fitness in multiple environments. Traditionally, studies of adaptive plasticity have examined a single type of environment, but organisms in nature may simultaneously adjust their phenotypes to multiple environments. In a series of experiments, I examined whether predation risk and mate availability interact to affect morphology and life history in a hermaphroditic snail (Physa acuta). Predation risk was expected to induce an investment in defense at the expense of reproduction. Mate availability was expected to affect the age at first reproduction where isolated snails are expected to delay selfing because this snail is a preferential outcrosser with the potential for self-fertilization at the cost of inbreeding depression. To establish the adaptive benefit of the predator-induced changes, I induced snails by rearing them in the presence and absence of chemical cues from predatory crayfish and exposed both phenotypes to selection by lethal crayfish. Crayfish induced an increase in mass and shell thickness, and snails with these traits experienced higher survival when exposed to a lethal predator. Therefore, predator-induced plasticity was favored by selection. To establish the adaptive benefit of delayed selfing, I quantified inbreeding depression by comparing the fitness of selfed and outcrossed snails reared in predator and no-predator environments. Inbreeding depression occurred in both environments and therefore, delayed selfing is favored by selection. I went on to demonstrate that inbreeding depression exists for two types of adaptive plasticity (i.e., delayed selfing and an inducible defense). Both types of inbreeding depression in plasticity may act as important constraints on the evolution of self-fertilization. In general, my results highlight the role of enemies in mating-system evolution and the role of mate availability in the evolution of inducible defenses as well as novel forms of constraint on the evolution of plasticity, including the existence of inbreeding depression in adaptive plasticity.

Transcription in eukaryotes takes place in the context of a repressive chromatin template. Access to the DNA is facilitated by histone modifying enzymes and ATP-dependent chromatin remodeling complexes, which modify chromatin structure. The activities of chromatin modifying proteins are often coordinated by nonenzymatic accessory factors that interact with actively transcribing RNA Polymerase II (Pol II). One such factor is the Saccharomyces cerevisiae Paf1 transcription elongation complex. This complex, which is minimally composed of Paf1, Ctr9, Rtf1, Cdc73, and Leo1, physically interacts with Pol II and localizes to the coding regions of active genes. The Rtf1 subunit of the Paf1 complex performs several cotranscriptional functions: it facilitates recruitment of the chromatin remodeling enzyme Chd1, promotes covalent modification of specific lysine residues in histones H2B and H3, and mediates association of other Paf1 complex subunits. Using a collection of internal deletion mutations that remove 20 to 50 amino acid segments across the length of Rtf1, I demonstrated that Rtf1's known functions are mediated by nonoverlapping regions, implying that the multiple functions of this protein are not completely interrelated. Deletion of the regions of Rtf1 that are required for promoting histone modification or its association with active genes resulted in the strongest transcription-related phenotypes, which suggested that promoting cotranscriptional histone modification is a critical means by which Rtf1 exerts its effects on transcription. Detailed analysis of the region of Rtf1 required for histone modification determined that it is sufficient to promote Rtf1-dependent histone modifications and that this function is dependent on several highly conserved residues. Additionally, a screen for factors that become essential in the absence of Rtf1 uncovered mutations in the first two enzymes of the inositol polyphosphate (IP) signaling pathway: Plc1 and Arg82. The IP signaling pathway has been linked to the function of several chromatin remodeling complexes. I uncovered strong genetic interactions between Arg82, Paf1, and mutations in the SWI/SNF and INO80 chromatin remodeling complexes and demonstrated that the expression of several target genes was strongly impaired by mutations in these factors. Together, these data suggest that transcription elongation, IP signaling, and chromatin remodeling cooperate to coordinate proper gene expression.