Designing proteins that work using recombinant technologies.

Therapeutic proteins have been engineered for a variety of purposes including reduced antigenicity, longer half-life, simplified process development, and increased affinity. Fusion proteins bring together functions from two different molecules creating therapeutics with completely novel activities. Protein engineering technologies have relied on rational design, directed evolution, DNA shuffling, RNA-peptide fusion, phage and ribosomal display methods to select out candidate protein forms with the desired therapeutic properties. Engineered site-specific pegylation and glycosylation strategies have improved circulation half-life, reduced immunogenicity and increased protein therapeutic stability. In this review we describe how protein engineering techniques have been used to select out, improve stability and clinical efficacy of protein therapeutics.

Structure-function analysis of FLT3 ligand-FLT3 receptor interactions using a rapid functional screen.

FLT3 ligand (FLT3L) stimulates primitive hematopoietic cells by binding to and activating the FLT3 receptor (FLT3R). We carried out a structure-activity study of human FLT3L in order to define the residues involved in receptor binding. We developed a rapid method to screen randomly mutagenized FLT3L using a FLT3R-Fc fusion protein to probe the relative binding activities of mutated ligand. Approximately 60,000 potential mutants were screened, and the DNA from 59 clones was sequenced. Thirty-one single amino acid substitutions at 24 positions of FLT3L either enhanced or reduced activity in receptor binding and cell proliferation assays. Eleven representative proteins were purified and analyzed for receptor affinity, specific activity, and physical properties. Receptor affinity and bioactivity were highly correlated. FLT3L affinity for receptor improved when four individual mutations that enhance FLT3L receptor affinity were combined in a single molecule. A model of FLT3L three-dimensional structure was generated based on sequence alignment and x-ray structure of macrophage colony-stimulating factor. Most residues implicated in receptor binding are widely dispersed in the primary structure of FLT3L, yet they localize to a surface patch in the tertiary model. A mutation that maps to and is predicted to disrupt the proposed dimerization interface between FLT3L monomers exhibits a Stokes radius that is concentration-dependent, suggesting that this mutation disrupts the FLT3L dimer.

Relationship between proliferation and susceptibility to CD95- and CD2-mediated apoptosis in stimulated primary T lymphocytes: T cells manifesting proliferative unresponsiveness are preferentially susceptible to CD95-mediated apoptosis.

We examined the relationship between proliferation and susceptibility to Fas- and CD2-mediated apoptosis of human peripheral T lymphocytes that had been exposed in primary culture to CD3- or CD2-derived mitogenic stimuli in the presence of monocytes and exogenous IL-2. After 5 days, activated T cells were fractionated into large (F2) and small (F6) cells on Percoll density gradients and analyzed for their susceptibility to apoptosis and for their position in the cell cycle. Most F6 cells displayed a CD45RA+, CD25-, CD2R- phenotype and were unable to incorporate bromodeoxyuridine (BrdUrd) during the entire culture period. However, they were activated to express Fas Ag and some cell cycle regulatory proteins specific to late G1 phase. T cells with proliferative unresponsiveness were sensitive to Fas-mediated apoptosis whether it was triggered by anti-Fas mAb or by Fas ligand, but were almost completely resistant to CD2 apoptotic signaling. In contrast, F2 cells exhibited classical activation markers (CD45RO, CD25, and CD2R), had crossed S phase at least once, and were sensitive to both Fas and CD2 apoptotic signals. In large cells harvested earlier (on day 3), the signals were operative in both BrdUrd+ and BrdUrd- cells. Thus, S phase entry is not required for Fas- and CD2-mediated apoptosis. The profound proliferative unresponsiveness of F6 cells to CD3 and CD2 stimuli (bypassed by ionomycin plus PMA) and the CD2R- conformation of their CD2 molecules suggest that they may be in vivo anergized cells whose elimination, upon restimulation, is highly dependent on the Fas death pathway.

Expression of trimeric CD40 ligand in Pichia pastoris: use of a rapid method to detect high-level expressing transformants.

Pichia pastoris is a yeast capable of expressing large amounts of some proteins. When expression vectors are introduced into P. pastoris, individual transformants typically express widely varying amounts of protein. Because clones expressing the highest level of protein occur infrequently during the transformation process, finding them can be very labor-intensive. We developed an immunological based filter screening method that rapidly detects transformants secreting large amounts of a heterologous protein. We have applied this method to the expression of a soluble trimeric form of CD40L, a molecule that regulates B-cell responses. Using this method, we identified transformants with one to 13 copies of the CD40L expression cassette. Maximum expression was obtained with clones containing eight or more copies of the expression cassette, and a clone with eight copies was selected for further analysis. High cell density fermentation of this clone using a mixed glycerol:methanol feed yielded 255 mg CD40L per liter of supernatant.

Ligands for the receptor tyrosine kinases hek and elk: isolation of cDNAs encoding a family of proteins.

Hek is a member of the eph subfamily of receptor tyrosine kinases whose members include elk, hek2, sek, eph and eck among others. Using a soluble form of hek consisting of the extracellular region of the receptor fused to the Fc domain of human IgG1 and an expression cloning strategy, we have isolated two different but related cDNAs from the human T-lymphoma line HSB-2 that encode ligands for hek. The cDNAs encode proteins of 238 and 201 amino acids (44% amino acid identity) that are anchored to the membrane by glycosylphosphatidylinositol (GPI)-linkage. The proteins encoded by these cDNAs are bound by hek with affinity constants of 2 x 10(8) M-1. These proteins also bind the elk tyrosine kinase receptor. These cDNAs are related to other cDNAs that we have recently isolated from a human placental library that encode ligands for both hek and elk and define an emerging family of ligands for eph-related kinases (LERKs).

CSE1 and CSE2, two new genes required for accurate mitotic chromosome segregation in Saccharomyces cerevisiae.

By monitoring the mitotic transmission of a marked chromosome bearing a defective centromere, we have identified conditional alleles of two genes involved in chromosome segregation (cse). Mutations in CSE1 and CSE2 have a greater effect on the segregation of chromosomes carrying mutant centromeres than on the segregation of chromosomes with wild-type centromeres. In addition, the cse mutations cause predominantly nondisjunction rather than loss events but do not cause a detectable increase in mitotic recombination. At the restrictive temperature, cse1 and cse2 mutants accumulate large-budded cells, with a significant fraction exhibiting aberrant binucleate morphologies. We cloned the CSE1 and CSE2 genes by complementation of the cold-sensitive phenotypes. Physical and genetic mapping data indicate that CSE1 is linked to HAP2 on the left arm of chromosome VII and CSE2 is adjacent to PRP2 on chromosome XIV. CSE1 is essential and encodes a novel 109-kDa protein. CSE2 encodes a 17-kDa protein with a putative basic-region leucine zipper motif. Disruption of CSE2 causes chromosome missegregation, conditional lethality, and slow growth at the permissive temperature.

Requirement for ESP1 in the nuclear division of Saccharomyces cerevisiae.

Mutations in the ESP1 gene of Saccharomyces cerevisiae disrupt normal cell-cycle control and cause many cells in a mutant population to accumulate extra spindle pole bodies. To determine the stage at which the esp1 gene product becomes essential for normal cell-cycle progression, synchronous cultures of ESP1 mutant cells were exposed to the nonpermissive temperature for various periods of time. The mutant cells retained viability until the onset of mitosis, when their viability dropped markedly. Examination of these cells by fluorescence and electron microscopy showed the first detectable defect to be a structural failure in the spindle. Additionally, flow cytometric analysis of DNA content demonstrated that massive chromosome missegregation accompanied this failure of spindle function. Cytokinesis occurred despite the aberrant nuclear division, which often resulted in segregation of both spindle poles to the same cell. At later times, the missegregated spindle pole bodies entered a new cycle of duplication, thereby leading to the accumulation of extra spindle pole bodies within a single nucleus. The DNA sequence predicts a protein product similar to those of two other genes that are also required for nuclear division: the cut1 gene of Schizosaccharomyces pombe and the bimB gene of Aspergillus nidulans.

Stimulation of human Jurkat cells by monoclonal antibody crosslinking of transfected-Ly-6A.2 (TAP) molecules.

Monoclonal antibody crosslinking of phosphatidylinositol-anchored Ly-6A.2 molecules on the surface of murine T lymphocytes leads to cell activation and secretion of IL-2. To examine the potential activity of these molecules in human T cells we transfected the Ly-6A.2 gene into Jurkat cells. Transfection of Jurkat cells with genomic Ly-6A.2 sequences results in low levels of Ly-6A.2 on the cell surface. However, linking the Ly-6A.2 sequences to the enhancer from the human CD2 gene results in greatly increased expression of Ly-6A.2. These molecules are anchored to the membrane via a phosphatidylinositol linkage. Crosslinking of Ly-6A.2 molecules with soluble mAb stimulates the transfected Jurkat cells to produce IL-2. This stimulation is abrogated by treatment with phosphatidylinositol-specific phospholipase C. The transfected human T cells displayed the same unusual crosslinking requirements for stimulation with anti-Ly-6A.2 mAbs as previously observed for murine T cells. Crosslinking of Ly-6A.2 with soluble antibodies is stimulatory, whereas immobilized antibodies are inactive. The crosslinking requirements for antiCD3 mAb stimulation display a reciprocal pattern. These data demonstrate that the Ly-6A.2 pathway for T cell activation is conserved between human and murine T cells.

Isolation, expression, and sequence of the TAP/Ly-6A.2 chromosomal gene.

The murine Ly-6 locus controls the expression of a number of genes. One of the products of the Ly-6 locus, Ly-6A.2, has been implicated in the process of T cell activation. We have identified the chromosomal sequences encoding the Ly-6A.2 molecule using very stringent hybridization and washing conditions. We confirmed that this gene encoded the Ly-6A.2 molecule by transfection studies using a cell line genetically negative for the Ly-6A.2 gene as a DNA recipient. Sequence analysis showed that the Ly-6A.2 gene is made up of four exons. The start site of transcription was determined by primer extension analysis. The first exon does not contain protein coding sequences. The structure of the Ly-6A.2 gene supports previous speculation that various Ly-6 RNA can be generated by alternate splicing events. The Ly-6A.2 chromosomal gene is closely related to the previously characterized Ly-6C.1 chromosomal gene in the intron, exon, and 5' flanking regions. This analysis indicates that these genes have arisen as a consequence of gene duplication. Although endogenous Ly-6A.2 and Ly-6C genes are IFN responsive, only the latter contains a clearly identifiable IFN responsive element. A transfected Ly-6A.2 chromosomal gene that contains 3.9 kb of 5'-untranslated sequence is IFN responsive. However, a shorter chromosomal clone containing only 940 bp of 5' sequence is constitutively expressed in transfectants but does not respond to IFN.

Saccharomyces cerevisiae mutants defective in chromosome segregation.

We have devised a genetic screen to identify trans-acting factors involved in chromosome transmission in yeast. This approach was designed to potentially identify a subset of genes encoding proteins that interact with centromere DNA. It has been shown that mutations in yeast centromere DNA cause aberrant chromosome segregation during mitosis and meiosis. We reasoned that the function of an altered centromere should be particularly sensitive to changes in factors with which it interacts. We constructed a disomic strain containing one copy of chromosome III with a wild-type centromere and one copy of chromosome III bearing the SUP11 gene and a mutant CEN3. This strain forms white colonies with red sectors due to nondisjunction of the chromosome bearing the mutant centromere. After mutagenesis we picked colonies that exhibited increased nondisjunction of the mutant chromosome as evidenced by increased red-white sectoring. Using this approach, we have isolated three trans-acting chromosome nondisjunction (cnd) mutants that are defective in maintaining chromosomes during mitosis in yeast.