Expression of a highly toxic protein, Bax, in Escherichia coli by attachment of a leader peptide derived from the GroES cochaperone.

Expression of the human apoptosis modulator protein Bax in Escherichia coli is highly toxic, resulting in cell lysis at very low concentrations (Asoh, S., et al., J. Biol. Chem. 273, 11384-11391, 1998). Attempts to express a truncated form of murine Bax in the periplasm by using an expression vector that attached the OmpA signal sequence to the protein failed to alleviate this toxicity. In contrast, attachment of a peptide based on a portion of the E. coli cochaperone GroES reduced Bax's toxicity significantly and allowed good expression. The peptide, which was attached to the N-terminus, included the amino acid sequence of the mobile loop of GroES that has been demonstrated to interact with the chaperonin, GroEL. Under normal growth conditions, expression of this construct was still toxic, but generated a small amount of detectable recombinant Bax. However, when cells were grown in the presence of 2% ethanol, which stimulated overproduction of the molecular chaperones GroEL and DnaK, toxicity was reduced and good overexpression occurred. Two-dimensional gel electrophoresis analysis showed that approximately 15-fold more GroES-loop-Bax was produced under these conditions than under standard conditions and that GroEL and DnaK were elevated approximately 3-fold.

A domain flip as a result of a single amino-acid substitution.

BACKGROUND: The self-assembly properties of beta domains are important features of diverse classes of proteins that include cell-adhesion molecules, surface receptors and the immunoglobulin superfamily. Immunoglobulin light-chain variable domains are well suited to the study of structural factors that determine dimerization, including how residues at the interface influence the preferred dimer arrangement. RESULTS: Single-site mutants of light-chain variable domain Len, designated LenQ38E and LenK30T, formed 'flipped' dimers in which one domain was rotated by about 180 degrees compared with the native protein. The dimer in the native protein is similar to that found between variable domains in Fab immunoglobulin fragments. When compared to the native dimer, more surface area is buried, and more hydrogen bonds and salt bridges are formed between the monomers in the flipped conformation. CONCLUSIONS: Immunoglobulin light-chain variable domains can form a minimum of two distinct quaternary structures. Single-site mutations resulting from changes of one base, such as the exchange of Gln38 to Glu or Lys30 to Thr, change the 'conventional' dimer of protein Len to a flipped arrangement. Native Len is not found in the flipped-domain dimer conformation because it would have excess positive electrostatic potential at the dimer interface that is not compensated by other forces. Excess negative or positive electrostatic potential at the dimer interface can have a determining effect on the mode of dimerization.

Reengineering immunoglobulin domain interactions by introduction of charged residues.

The formation of the antibody variable domain binding unit (Fv) is the net result of three competing assembly reactions. The affinities of concurrent homologous interactions of heavy and light chain variable domains limits the heterologous interaction leading to productive formation of the Fv. To address the possible role of light chain dimerization in this phenomenon, the Gln38 residue at the dimer interface of an immunoglobulin light chain variable domain (VL) was replaced by charged amino acids. The effects of these mutations on VL homodimer formation were monitored by small-zone size exclusion HPLC and the affinities of interaction were determined by computer simulation. Reduced VL homodimerization was observed in three of the four mutants, Q38R, Q38D and Q38K. The association constants for the Q38R and Q38D homodimers were 1.2 x 10(4) and 3.2 x 10(3) M(-1), respectively. This corresponded to a 20-75-fold reduction in the homodimer association constant relative to the wild-type VL, which had an association constant of 2.4 x 10(5) M(-1). Surprisingly, the fourth charge mutant, Q38E, had a higher association constant than the wild-type VL. The potential for charged residues to facilitate heterodimeric assembly of immunoglobulin domains was also tested. Heterodimerization was observed between the Q38D and Q38R V(L)s, but with an association constant of 4.7 x 10(4) M(-1), approximately fivefold lower than that obtained for homodimerization of the native V(L). In addition, replacement of the neutral, solvent-accessible Gln38 residue with either Asp or Arg was found to be significantly destabilizing. These results suggest that charged residues could be introduced at immunoglobulin domain interfaces to guide heterodimer formation and to minimize unfavorable competing homologous associations. Nonetheless, these apparently simple modifications may also result in unintended consequences that are likely to depend upon structural features of particular variable domains.

Stringency of substrate specificity of Escherichia coli malate dehydrogenase.

Malate dehydrogenase and lactate dehydrogenase are members of the structurally and functionally homologous family of 2-ketoacid dehydrogenases. Both enzymes display high specificity for their respective keto substrates, oxaloacetate and pyruvate. Closer analysis of their specificity, however, reveals that the specificity of malate dehydrogenase is much stricter and less malleable than that of lactate dehydrogenase. Site-specific mutagenesis of the two enzymes in an attempt to reverse their specificity has met with contrary results. Conversion of a specific active-site glutamine to arginine in lactate dehydrogenase from Bacillus stearothermophilus generated an enzyme that displayed activity toward oxaloacetate equal to that of the native enzyme toward pyruvate (H. M. Wilks et al. (1988) Science 242, 1541-1544). We have constructed a series of mutants in the mobile, active site loop of the Escherichia coli malate dehydrogenase that incorporate the complementary change, conversion of arginine 81 to glutamine, to evaluate the role of charge distribution and conformational flexibility within this loop in defining the substrate specificity of these enzymes. Mutants incorporating the change R81Q all had reversed specificity, displaying much higher activity toward pyruvate than to the natural substrate, oxaloacetate. In contrast to the mutated lactate dehydrogenase, these reversed-specificity mutants were much less active than the native enzyme. Secondary mutations within the loop of the E. coli enzyme (A80N, A80P, A80P/M85E/D86T) had either no or only moderately beneficial effects on the activity of the mutant enzyme toward pyruvate. The mutation A80P, which can be expected to reduce the overall flexibility of the loop, modestly improved activity toward pyruvate. The possible physiological relevance of the stringent specificity of malate dehydrogenase was investigated. In normal strains of E. coli, fermentative metabolism was not affected by expression of the mutant malate dehydrogenase. However, when expressed in a strain of E. coli unable to ferment glucose, the mutant enzyme restored growth and produced lactic acid as the sole fermentation product.

Assessment of adsorption and adhesion of proteins to polystyrene microwells by sequential enzyme-linked immunosorbent assay analysis.

Originally developed as a method for estimating the solid-phase concentration of an adsorbed protein, sequential enzyme-linked immunosorbent assay (ELISA) analysis (P. A. Underwood and J. G. Steele (1991) J. Immunol. Methods 142, 83-94) is also useful for assessing other aspects of protein-surface interactions without radioactively labeled reagents. The method provides a way to compare protein adsorption on various microwell surfaces and to track adhesion of the protein throughout an assay's wash and incubation steps. In cases where adsorption complies with Langmuir-type binding and protein desorption is minimal, sequential ELISA data can be analyzed with a mathematical model to quantitatively estimate the microwell's protein-binding capacity. Even for protein-surface interactions where the mathematical model is not appropriate, characteristics of sequential ELISA binding data qualitatively differentiate between initial adsorption with subsequent desorption and poor initial adsorption of the protein. Using this method, we analyzed the binding of three proteins (rabbit immunoglobulin G, bovine serum albumin, and horse spleen ferritin) to four types of polystyrene microwell surfaces (Immulons 1, 2, 3, and 4). In addition, we examined differences in protein adhesion to the four surfaces when Tween 20 was omitted from the assay's wash buffer or when fewer washes were used.

Recombinant immunoglobulin variable domains generated from synthetic genes provide a system for in vitro characterization of light-chain amyloid proteins.

The primary structural features that render human monoclonal light chains amyloidogenic are presently unknown. To gain further insight into the physical and biochemical factors that result in the pathologic deposition of these proteins as amyloid fibrils, we have selected for detailed study three closely homologous protein products of the light-chain variable-region single-gene family VkIV. Two of these proteins, REC and SMA, formed amyloid fibrils in vivo. The third protein, LEN, was excreted by the patient at levels of 50 g/day with no indication of amyloid deposits. Sequences of amyloidogenic proteins REC and SMA differed from the sequence of the nonpathogenic protein LEN at 14 and 8 amino acid positions, respectively, and these amino acid differences have been analyzed in terms of the three-dimensional structure of the LEN dimer. To provide a replenishable source of these human proteins, we constructed synthetic genes coding for the REC, SMA, and LEN variable domains and expressed these genes in Escherichia coli. Immunochemical and biophysical comparisons demonstrated that the recombinant VkIV products have tertiary structural features comparable to those of the patient-derived proteins. This well-defined set of three clinically characterized human kIV light chains, together with the capability to produce these kIV proteins recombinantly, provide a system for biophysical and structural comparisons of two different amyloidogenic light-chain proteins and a nonamyloidogenic protein of the same subgroup. This work lays the foundation for future investigations of the structural basis of light-chain amyloidogenicity.

Estimation of the protein-binding capacity of microplate wells using sequential ELISAs.

A non-linear mathematical method for estimating the protein binding capacity of a microwell is developed by applying Langmuir-type binding principles to data obtained from sequential ELISAs (Underwood and Steele (1991) J. Immunol. Methods 142, 83). Experimental data from sequential ELISAs measuring the binding of rabbit and mouse immunoglobulin G and of BSA to Immulon 2 microwells agree well with the binding curves predicted by our simple mathematical model. Estimates of Immulon 2 microwell capacity for these proteins when coated in a 100 microliters/well volume are between 250 and 325 ng/well. The method provides a convenient, quantitative method for estimating microwell capacity without the use of radioisotopes. A spreadsheet form of the method is also presented, so that the mathematical calculations for any set of experimental data may be performed readily using the spreadsheet's SOLVER function. The method is not applicable to data from experiments where significant desorption occurs or where binding to the solid phase deviates significantly from the simple Langmuir model.

Monoclonal antibodies to HIV-1 p24 core protein include pairs which exhibit synergistic binding.

Gamma and kappa chain cDNAs from four mouse monoclonal antibodies (mAbs) which bind three different sites on the core antigen (p24) of HIV-1 have been cloned and their V-region sequences determined. These mAbs are part of a larger group of seven anti-p24 mAbs analyzed in simultaneous competition assays with HIV-1 lysate as antigen and in protein blotting experiments using 10 carboxy-terminal truncations of a p24 fusion protein. One mAb, BB128, recognizes the p24 loop sequence EAAEWDRVHP and enhances the binding of two other mAbs (BI1777 and BI1279) when tested pairwise in simultaneous competition assays. The two monoclonals enhanced by BB128 recognize different antigenic sites on p24, with BI1777 binding to carboxy-terminal sequences and BI1279 to amino-terminal residues. In the pairwise assays, mAb BI1279 also acts as enhancing antibody for BI1777, as does mAb BB328, which recognizes residues in the central region of p24. Since aggregated p24 monomers form the HIV-1 capsid, p24 is a multivalent antigen in HIV-1 lysate. It seems likely, therefore, that synergistic binding of mAb pairs to p24 is effected by bivalent binding of the enhancing mAb stabilizing a conformation favorable for bivalent binding of the enhanced mAb.

Structure and expression of the chicken ferritin H-subunit gene.

Although the genomes of many species contain multiple copies of ferritin heavy (H)- and light (L)-chain sequences, the chicken genome contains only a single copy of the H-subunit gene. The primary transcription unit of this gene is 4.6 kilobase pairs and contains four exons which are posttranscriptionally spliced to generate a mature transcript of 869 nucleotides. Chicken and human ferritin H-subunit genomic loci are organized with similar exon-intron boundaries. They exhibit approximately 85% nucleotide identity in coding regions, which yield proteins 93% identical in amino acid sequence. We have identified a sequence of 22 highly conserved nucleotides in the 5' untranslated sequences of chicken, human, and tadpole ferritin H-subunit genes and propose that this conserved sequence may regulate iron-modulated translation of ferritin H-subunit mRNAs.

Replacement variant histone genes contain intervening sequences.

The nucleotide sequences of two chicken histone genes encoding replacement variant H3.3 polypeptides are described. Unlike the replication variant genes of chickens (and almost all other organisms), these genes contain intervening sequences; introns are present in both genes in the 5' noncoding and coding sequences. Furthermore, the replacement variant histone mRNAs are post-transcriptionally polyadenylated. The locations, but not the sizes, of the two introns within the coding segments of the two genes have been exactly conserved, whereas the intron positions in their respective 5' flanking regions differ. Although both H3.3 genes predict the identical histone polypeptide sequence, they are as different from one another as each of them is from a more common replication variant H3.2 gene in silent base substitutions within the coding sequences. Thus, the H3.3 polypeptide sequence has been precisely maintained over a great evolutionary period, suggesting that this class of histones performs a strongly selected biological function. Although replacement variant histones can account for more than 50% of the total H3 protein in the nuclei of specific chicken tissues, the steady-state level of H3.3 mRNA is nearly the same (and is quite low) in all tissues and ages of animals examined. These properties suggest novel mechanisms for the control of the basal histone biosynthesis which takes place outside of the S phase of the cell cycle.

Some properties of D-mannose isomerase from Escherichia coli K12.

A second-stage mutant of Escherichia coli K12 designated as strain 806 grew faster on D-lyxose than the mutant strain 805 previously described. Both mutants produced constitutively a novel enzyme, D-mannose isomerase, but strain 806 produced twice as much as strain 805. The enzyme could fortuitously convert D-lyxose to D-xylulose, which is a normal intermediate in the D-xylose catabolic pathway. The purified enzyme consisted of four subunits each with a molecular weight of about 40 000. In 0.14 M-Na2SO4, the tetramer dissociated completely into dimers. While the tetramer Km values for D-mannose and D-lyxose were 80 mM and 300 mM, respectively, the dimer Km values for these two sugars were both 300 mM. The amino acid composition of the enzyme was also determined.