The structure of the sc D-maltose-binding protein, an essential component of the Escherichia coli high affinity osmotic-shock sensitive transport and chemotaxis systems for 1-$\alpha$-4 linked glucose oligosaccharides, has been determined at a resolution of 2.3A by x-ray crystallography. The R-factor is 25% for 15,162 reflections by a restrained least squares method. The maltose-binding protein is ellipsoidal with dimensions of 30 A $\times$ 40 A $\times$ 65 A. The secondary structure is folded from a single polypeptide chain of 370 residues into two domains connected by three segments. The N-domain is made up of a five strand $\beta$ sheet (with the fourth strand antiparallel) flanked by two $\alpha$ helices on one side and three on the other. The C-domain is arranged similarly with the addition of a pair of antiparallel $\alpha$ helices that span the cleft. These helices act to extend the length of the cleft. The antiparallel strand is the first element after the initial crossover into either domain. The three crossovers are located in close physical proximity, although widely separated in sequence. The first two crossovers form a short $\beta$ sheet. The crossovers serve as the base for the cleft. The maltose-binding protein consists of 40% $\alpha$ helix and 20% $\beta$ sheet. The principle folding pattern is of alternating $\beta$ sheet and $\alpha$ helix. The binding site(s) for maltose and maltotriose were determined to lie in the cleft formed by the two domains. The reducing end of both sugars was found to occupy the same site in MBP. Hydrogen bonds formed between side chain residues of MBP and hydroxyl groups of the sugar are the main stabilizing force in the binding of substrate. Maltose is almost entirely buried by the protein. The binding site is rich in aromatic residues. The location of site specific mutations in MBP that are defective in chemotaxis, but not in transport have been found to lie in exterior portions of MBP in different domains. Other mutations affecting a region on the opposite side of the $\beta$ sheet from the chemotactic mutant in the C-domain eliminate transport but do not effect substrate binding or chemotaxis. These findings support the theory of recognition of ligand-bound binding protein by differential separation of two sites on opposite sides of the cleft due to conformational change upon ligand binding. The structural evidence also supports the existence of separate recognition sites for transport and chemotaxis.

We have measured the magnetic circular dichroism of cytochrome c peroxidase and some of its derivative from 250-350 nm. Comparison of the changes observed on conversion to the catalytic intermediate (cytochrome c peroxidase-I) with spectra obtained from horseradish peroxidase and its derivatives and model compounds of protoheme leads us to the conclusion that the observed changes in the magnetic circular dichroism spectra reflect conversion of the heme to the ferryl state. No evidence was found for the modification of tryptophan in cytochrome c peroxidase I. Five programs for the investigation of steady-state kinetics of catalyzed reactions have been developed. Two provide the solution to the steady-state rate equation: the first of these is a straightforward implementation of the rules developed by Chou, and the second is a very efficient procedure for evaluating King-Altman diagrams and can be used for quite large mechanisms. The third is a program which accepts the output of the second and simplifies the equation to one in identical powers of substrate and products. The fourth program provides the numeric solution to a specific mechanism and set of initial conditions; it is well suited to extremely large, complex mechanisms. The fifth program combines the fourth method and a least squares fitting routine to optimize a mechanism to selected data sets. Mathematical criteria, deduced from empirical observations, have been developed that strictly limit the type of steady-state cytochrome c oxidase mechanisms that are practicable. Two current models (Brzezinski and Malmstrom (1986) Proc. Natl. Acad. Sci. USA 83, 4282-4286; Speck et al. (1984) Proc. Nat. Acad. Sci. USA 81, 347-351) are shown to fail these criteria and thus cannot represent the steady-state reaction of oxidase. Models that do satisfy these criteria are discussed.

An observed phenomenon of cell mediated immunosuppression in mice caused by the lack of preformed purine and pyrimidines in the diet was characterized and the mechanism of action explored utilizing three different approaches. The three approaches employed were as follows: (1) measurement of T cell mediated immune response in vitro and in vivo; (2) analysis of lymphocyte phenotype involving surface marker immunofluorescence, measurement of the levels of purine and pyrimidine enzymes as predictors of lymphocyte maturation and evaluation of putative G$\sb1$ phase characteristics; and (3) examination of diet fed host effects on the in vivo out growth of syngeneic lymphoid tumors. Balb/c mice fed a nucleotide free (NF) diet exhibited significantly decreased T helper/inducer cell function and number relative to mice fed normal rodent chow (F) or NF plus RNA (NFR). Lymphoproliferative response in vitro and in vivo was significantly lower in mice fed the NF diet than the response exhibited by mice fed the control diets or NF plus uracil (NFU). This lack of proliferative response was accompanied by decreased induction of two purine enzymes important to the immune response: adenosine deaminase and purine nucleoside phosphorylase. Bone marrow, thymus and spleen obtained from NF diet fed mice contained significantly more T cells of an immature phenotype with high terminal deoxynucleotidyl transferase and adenosine deaminase levels and low purine nucleoside phosphorylase level enzyme profiles relative to that of similar lymphoid tissues obtained from mice fed the control diets. Levels of ecto 5$\sp\prime$ nucleotidase, another enzyme linked to optimal B cell function were not affected by the NF diet. Lymphocyte nucleotide pools were altered and the number of putative G$\sb1$ phase T lymphocytes were increased in mice fed the NF diet relative to mice fed the purine and pyrimidine supplemented diets. In vivo outgrowth of the syngeneic T cell lymphoma, 5F4, in Balb/c hosts fed the NF diet was significantly decreased compared to the 5F4 tumor growth observed in mice fed the NF diet supplemented with RNA, adenine or uracil. (Abstract shortened with permission of author.)

The sequence and biochemical requirements of pre-mRNA polyadenylation were investigated using the simian virus 40 (SV40) early polyadenylatin site. This signal directs 3$\sp\prime$ processing of two mRNA molecules coding for the large and small T-antigens of the DNA tumor virus SV40. The processing of this site was studied using a nuclear extract capable of accurately and efficiently polyadenylating exogenously provided RNA molecules. The sequence requirements of polyadenylation were examined by point mutagenesis of conserved sequence elements within the SV40 early site followed by assay of their effect on in vitro polyadenylation. Alteration of the internal pyrimidine of the highly conserved sequence AAUAAA from U to an A abolishes cleavage at the normal polyadenylation site and activates a minor site downstream of the distal, usually inactive, hexanucleotide. Point mutation of the conserved CAYUG element decreases the efficiency of cleavage without changing its specificity. Complete removal of this sequence abolishes cleavage of the SV40 early polyadenylation site. These results suggest that multiple sequence elements are involved in generating accurately polyadenylated RNA from the SV40 early genes. The ATP requirement of the polyadenylation cleavage reaction was examined by removal of ATP from in vitro reactions using the SV40 early and the adenovirus L3 substrate RNAs. In vitro polyadenylation of the SV40 early site requires 1mM ATP to meet the energy requirements of endonucleolytic cleavage as well as to provide substrate for polymerization. At ATP concentrations below 0.1 mM, the SV40 early polyadenylation site is subject to alternate cleavage approximately 21 nucleotides downstream of the correct cleavage site. Adenovirus L3 RNA is not an efficient substrate for alternate cleavage in the absence of ATP. This alternate activity has many characteristics in common with the correct cleavage. The behavior of the ATP-independent cleavage product is inconsistent with its being an intermediate in the polyadenylation reaction. The ATP-independent cleavage activity requires an intact polyadenylation consensus signal, a 3$\sp\prime$ terminus located within 100 nucleotides of the polyadenylation site, and extract U-snRNPs. These requirements suggest that ATP-independent cleavage is an alternate activity of the normal polyadenylation machinery observations suggest that SV40 early and adenovirus L3 RNAs are differentially recognized by the polyadenylation machinery. (Abstract shortened with permission of author.)

Nearly every aspect of sterol biosynthesis has been studied or exploited as means of regulating pathway flux, inhibiting specific enzymatic reactions, elucidating product cyclization or determining the evolutionary nature of the pathway itself. Despite enormous efforts in this area pathway regulation, product formation and the evolutionary origins of sterol biosynthesis remain unknown. Described herein is the use of molecular, synthetic and phylogenetic techniques to examine how oxidosqualene cyclases (OSC) (a specific enzyme within the sterol biosynthetic pathway) have evolved over time and how this class of enzymes controls the cyclization of oxidosqualene to a variety of triterpenes. OSCs are a unique family of enzymes producing over 100 naturally occurring cyclization products ranging from mono to hexacyclic compounds. These enzymes display high sequence similarity and it is predicted that OSCs share a similar three-dimensional structure, while subtle residue changes in the active site alter catalysis. There are two subclasses of OSCs, protosteryl and dammarenyl, and they are defined by the tetracyclic intermediate cation formed during cyclization. Residues that are differentially conserved between protosteryl and dammarenyl type cyclases were identified as candidates imparting a specific catalytic function. Using DNA synthesis, libraries of chimeric enzymes were generated that blended the sequences of one protosteryl-type cyclase (cycloartenol synthase from Arabidopsis thaliana) and a dammarenyl-type cyclase (lupeol synthase from Olea europaea) and screened for the ability to produce protosteryl-type products. Using this approach we were able to narrow the number of residues imparting specific catalytic function from 759 to 15 candidate residues within the Arabidopsis thaliana cycloartenol synthase. This information prompted direct mutagenesis of several single residues to better our understanding of the active site environment. To complement the DNA synthesis experiments, the second half of this thesis describes the cloning and characterization of the Methylococcus capsulatus lanosterol synthase, the Gemmata obscuriglobus parkeol synthase and several other bacterial/ancient cyclases. Sterols, once thought to be only eukaryotic in nature, are now being isolated from many prokaryotic organisms. Whether OSCs evolved in prokaryotes or were acquired later via horizontal gene transfer remains unknown. By studying both the residue conservation patterns and the product profiles of these bacterial cyclases, we have been able to formulate a working hypothesis that supports the evolution of sterols in prokaryotes. These primitive cyclases also retain residue conservation patterns that vary from eukaryotic enzymes indicating they direct product formation differently than more modern cyclases.

Recent clinical trials involving administration of a cell-free hemoglobin based blood substitute have shown that several detrimental side effects are the result of reactions with nitric oxide produced by endothelial cells. We have studied key reactions between NO and hemoglobin to determine which active site amino acids influence the reaction rates, to construct mutants that have reduced rates of NO scavenging and to identify any potentially toxic reaction intermediates. A library of over 100 recombinant myoglobins and hemoglobins was used to examine: (1) reversible binding of NO to reduced Fe(II) forms, (2) reversible binding to ferric Fe(III) forms; (3) NO dioxygenation by the Fe(II)-O 2 complex, (4) oxidation of Fe(II)NO complexes in the presence of oxygen, and (5) anaerobic autoreduction of Fe(III) complexes. Results from these studies will be crucial not only for identifying problems with current blood substitutes, but also for designing new recombinant hemoglobins for the next generation of blood substitutes. The rates of reversible NO binding to ferrous and ferric myoglobin are governed largely by the free energy required to replace non-covalent and coordinated distal pocket water, respectively. Steric constraints governing the rate of ligand entry into the distal pocket play a smaller role in dictating binding rates. NO dioxygenation by oxyhemoglobin is the major cause of the hypertensive effect of extracellular hemoglobin. This reaction generates a high-spin intermediate, which quickly isomerizes to nitrate to be released from the heme pocket. Studies with H64Q mutant myoglobin suggest that a ferryl intermediate may also occur during the internal peroxynitrite isomerization reaction. Autooxidation of ferrous nitrosyl complexes is directly governed by the rate of NO dissociation. This reaction proceeds by a rapid binding of oxygen to the vacant binding site, and the displaced NO undergoes dioxygenation by the newly formed oxymyoglobin. Autoreduction of ferric nitrosyl complexes proceeds by nucleophilic addition of a hydroxide ion to the iron-bound NO, generating reduced iron and nitrous acid. The rate of autoreduction shows little dependence on distal pocket structure. In contrast, the geometry of the proximal histidine plays a key role in regulating reduction of ferric nitrosyl complexes.

Adenosine deaminase (ADA) catalyzes the irreversible deamination of (2$\sp\prime$-deoxy)adenosine to (2$\sp\prime$-deoxy)inosine. It is an indispensable enzyme, with a role in purine catabolism and in the development of a competent immune system. This work focuses on the study of the catalytic mechanism employed by the murine enzyme through the use of site-directed mutagenesis. A glutamate mutation at the conserved active site Asp 295 shows that this residue is necessary for the proper orientation and placement of the catalytic hydroxylate. An alanine and an asparagine mutant of Asp 296 show that this residue functions mainly by anchoring the substrate in the active site via hydrogen bonding and thus reducing the aromaticity of the purine ring. Alanine, glutamate and arginine mutations at the proposed base for the reaction, His 238, clearly show that it does not abstract the proton from the zinc-bound water, but rather promotes the formation of the hydroxylate through charge stabilization. Replacements of the conserved Cys 262 by alanine and serine clearly demonstrate that it is not directly involved in the reaction mechanism. Structural studies with the ADA apoenzyme reveal that chelation of zinc does not result in structural rearrangements of either the active site or the secondary and tertiary structures of the enzyme. Loss of zinc is accompanied by loss of activity, which can be restored upon stoichiometric re-addition of zinc or cobalt. A transition-state analog such as deoxycoformycin can bind the apoenzyme by inducing the same type of conformational change as it does when it binds the holoenzyme. Mutants such as D296A and D296N denature more slowly compared to the wild-type, probably due to the better packing of an Ala or Asn side chain compared to the native Asp in the part of the enzyme surrounding residue 296. By contrast, mutants such as D295E, H238A, and H238E destabilize the holoenzyme, and mutants H238R, C262A, and C262S destabilize the holoenzyme and may also impede the in vivo folding pathway. Removal of the metal cofactor from wild-type or mutant ADA generally increases the enzyme's rate of denaturation.

The crystal structures of adenylate kinases from the mesophile Methanococcus voltae (37°C) and the thermophile M. thermolithotrophicus (65°C) have been solved to 2.5 A resolution using molecular replacement methods. These adenylate kinases share 78% primary sequence identity, yet exhibit significantly different thermal stabilities and optimal activity ranges. Analyses of these archaeal structures reveal possible details regarding their disparate thermostabilities. In this study, we perform a comparative structural analysis of the mesophilic and thermophilic adenylate kinases and draw four general conclusions. First, we find correlation between thermostability and ionic interactions and identify a unique ionic network in the thermophilic enzyme. Second, we find beta-branched residues incorporated within alpha-helices of the thermophilic enzyme with significantly greater frequency. Third, we find examples of tighter packing within the CORE domain of the thermophilic enzyme. Last, most of the mutations in the thermophile occur very near the surface resulting in greater negative surface potential. In additional to the comparative thermostability study, we also analyze these two methanococcal structures with respect to their apparent lack of an essential lysine residue, which is present within the P-loop of previously characterized adenylate kinases. Previous modeling experiments proposed that, if protonated, His92 could participate in a similar manner as the lysine residue present in homologous adenylate kinases. From our investigation, we conclude that, as in the case of the homologous Sulfolobus acidocaldarius adenylate kinase structure, His92 is involved in interactions with the terminal phosphate group of AMP. Structural alignments of Methanococcus and homologous enzymes demonstrate that Gly14 is structurally equivalent to Lys14, indicating that an amino acid outside of the canonical P-loop must compensate for the essential lysine deficiency. We propose that this compensatory role is filled either by Arg138 or Arg140, each of which are LID domain residues conserved among the archaeal enzymes.

This thesis describes mutagenesis experiments in cycloartenol synthase and lanosterol synthase that allowed the identification of important catalytic residues necessary to broaden and narrow product profiles. These results provided insights into factors critical for redesigning enzyme function. Mutagenesis in Arabidopsis thaliana cycloartenol synthase revealed that His477 is an essential component of the catalytic distinction between cycloartenol synthase and lanosterol synthase. Mutations at position 477 abolish cycloartenol biosynthesis, and subtle structural changes at this position dramatically alter product profile. The His477Asn mutant produces lanosterol as its major product (88%) whereas His477Gln produces primarily parkeol (73%). His477 influences deprotonation more strongly than any of the previously studied catalytic residues, but changes at this position are catalytically irrelevant in the presence of Tyr410Thr and Ile481Val mutations. The His477Asn Tyr410Thr Ile481Val and His477Gln Tyr410Thr Ile481Val triple mutants have the same product profile as the Tyr410Thr Ile481Val double mutant. Homology modeling studies established that His477 is a second-sphere residue that affects catalysis indirectly through its interactions with the active-site residue Tyr410. Changes at His477 strongly affect the location, orientation, and electronics of the Tyr410 side chain. Efforts to modify the catalytic specificity of enzymes consistently show that it is easier to broaden the substrate or product specificity of an accurate enzyme than to restrict the selectivity of one that is promiscuous. Careful examination of the homology model allowed the identification of a combination of mutations necessary to redesign cycloartenol synthase into a highly accurate lanosterol synthase. A double mutant was constructed and characterized and was shown to cyclize oxidosqualene accurately to lanosterol (99%). This catalytic change entailed both relocating polarity with a His477Asn mutation and modifying steric constraints with an Ile481Val mutation, and is among the best examples of redesigning an enzyme to accurately generate a new product. Known lanosterol synthase mutants make monocyclic or tetracyclic byproducts from oxidosqualene. Mutation of Saccharomyces cerevisiae lanosterol synthase at Tyr510 caused partial substrate misfolding and generated a tricyclic byproduct. This novel triterpene, (13alphaH)-isomalabarica-14(27),17 E,21-trien-3beta-ol, is the putative biosynthetic precursor of the isomalabaricane triterpenoids in sponges. The results suggest the facile evolution of tricyclic terpenoids as secondary metabolites in sponges.

A study of the lac repressor hinge region was undertaken to decipher its role in the function of DNA binding and communication between binding sites within the protein. Specific mutations were designed, and the proteins purified from these constructs were used in various biochemical experiments. These mutants revealed alterations in DNA binding as well as in the response to inducer. The first set of mutations involved insertion of glycines to the hinge region in an attempt to disrupt communication between the binding domains. The placement of the insertions was important for the maintenance of DNA binding. As the number of insertions increased, the protein's affinity for the natural operator decreased, approaching activity for non-specific DNA. Another mutation introduced within the hinge region involved the addition of a disulfide bond across partner hinges within the dimer of LacI. This alteration apparently stabilizes the N-termini to a significant degree and prevents operator dissociation in the presence of inducer for the oxidized protein. This mutant protein in both oxidized and reduced forms also demonstrated high affinity for the natural operator as well as several variant constructs. These mutants offered the potential to study the effect of operator sequence on the protein's interaction with DNA. Variant operators were constructed that altered the spacing and/or the symmetry of the half-sites. A broad range of responses are reported, some of which were not predicted based on the wild-type operator binding or inducibility for the particular protein under examination. In composite, the results reported in this thesis confirm a key role for sequence and structure in the hinge region of this protein in determining both affinity for operator DNA (and variants) and the allosteric response to inducer. Perhaps more importantly, however, the binding properties of these mutant lactose repressors with operator DNA variants illuminate a less well understood and appreciated phenomenon---the role of non-contacted DNA sequence and therefore presumably conformation in this regulatory interaction. These observations provide a strong base for detailed exploration of the mechanisms by which DNA sequence variation influences binding parameters for regulatory proteins.