Libraries of random peptide sequences were constructed and screened to identify peptides that specifically bind to proteins. In one of these about 2 X 10(7) different 15-residue peptide sequences were expressed on the surface of the coliphage M13. Each phage encoded a single random sequence and expressed it as a fusion complex with pIII, a minor coat protein present at five molecules per phage. Phage encoding nine different streptavidin-binding peptide sequences were isolated from this library. The core consensus sequence was His-Pro-Gln and binding of these phage to streptavidin was inhibited by biotin. This type of library makes it possible to identify peptides that bind to proteins (or other macromolecules) that have no previously known affinity for peptides.

The isolated protease domain of single chain urokinase–type plasminogen activator (scu–PA) preferentially activates fibrin–associated plasminogen in plasma, yet scu–PA binds poorly to forming fibrin clots. We are investigating the effect of combining this protease domain with protein domains that are reported to be involved in the binding of other plasma proteins to forming fibrin clots; the chimeric proteins consist of the protease domain and either the plasmin kringle 1 domain or the tissue plasminogen activator kringle 2 domain. Rapid construction of expression vectors for these chimeras was possible because much of the coding region required was synthesized either by polymerase chain reaction using RNA as the initial template or by the ligation of multiple long synthetic oligonucleotides. We also used an artificial signal peptide and baculovirus expression vectors to direct the secretion of these chimeras from insect cells. The signal peptide was designed by selecting those amino acid residues that occur more frequently than expected (based on amino acid composition) at positions −13 to +2 relative to the signal peptidase cleavage site in eukaryotic proteins. The chimeric proteins expressed by insect cells were enzymatically active, but their binding to forming fibrin clots was similar to that of the poorly binding scu–PA. In an in vitro clot lysis system, these chimeras displayed a specificity for fibrin–associated plasminogen greater than the specificity of two chain urokinase–type plasminogen activator and similar to the specificity of tissue plasminogen activator and scu–PA.

We have improved the expression of recombinant human granulocyte-colony-stimulating factor (G-CSF), produced by either pL or trpP expression vectors in Escherichia coli, by altering the sequence at the 5' end of the G-CSF-coding region. Initial attempts to express G-CSF resulted in neither detectable G-CSF mRNA nor protein in the trpP system, and only G-CSF mRNA was detectable in the pL system. We modified both expression vectors to decrease the G + C content of the 5' end of the coding region without altering the predicted amino acid sequence. This resulted in expression of detectable G-CSF mRNA and protein in both systems. Expression reached 17% and 6.5% of the total soluble cellular protein in the pL and trpP expression systems, respectively. The N-terminal sequence of the recombinant G-CSF from the pL system was Met-Thr-Pro-Leu-Gly-Pro-. G-CSF isolated from several human cell lines (including the LD-1 cell line reported here), does not have an N-terminal methionyl residue. Deletion of the threonine codon at the beginning of the coding region for the mature G-CSF resulted in efficient removal of the N-terminal methionine residue during expression in E. coli.

We have optimized conditions for genomic Southern analysis of eukaryotic DNA with both unique and degenerate oligonucleotide probes. Using the human Factor IX gene, optimal probe concentrations and hybridization times were determined, and washing conditions with sodium chloride and tetramethylammonium chloride (TMA-Cl) were compared. With TMA-Cl, the washing temperature was independent of GC content. With these conditions, the Factor IX gene was detected in genomic DNA using a 128-fold degenerate 20-mer and a 32-fold degenerate 17-mer. This permits the detection of a gene prior to cloning and the reduction of probe degeneracy, and facilitates the isolation of that gene. We apply this method of probe degeneracy reduction using probes for the human Factor VIII gene.

We have isolated and expressed a cDNA clone that encodes a human granulocyte colony-stimulating factor from the MIA PaCa-2 cell line. A genomic clone of this factor has been isolated from the CHU-2 cell line and is reported to encode two alternative transcripts [The EMBO J. 5,575, 1986]; one transcript predicts an amino acid sequence identical to that predicted by our MIA PaCa-2 cDNA clone; the other transcript predicts a similar protein containing a three amino acid residue insertion. To investigate which types of this colony-stimulating factor are produced by other cell lines, we used specific oligonucleotides to determine which types of transcripts were present in MIA PaCa-2, 5637, and LD-1 cells, all of which have been reported to produce a factor that can stimulate the growth of predominantly granulocyte colonies in human bone marrow cell cultures. Northern analysis with these probes revealed MIA PaCa-2-like transcripts in all of these cell lines and failed to detect transcripts that would encode the colony-stimulating factor that contained the three-amino-acid-residue insertion.

We have isolated a human melanoma line (LD-1) from a patient with melanoma and unexplained leukocytosis. The LD-1 cells produced a colony-stimulating factor (CSF) which stimulated primarily granulocytic colonies in human and murine bone marrow cultures. Erythroid burst and mixed colony-stimulating activity was not detected. A single CSF species with a molecular weight of 21,000 was detected in LD-1-conditioned media by G-200 chromatography. Nude mice transplanted with LD-1 tumors developed granulocytosis and had increased blood CSF levels. Messenger RNA from LD-1 cells directed the synthesis of CSF by Xenopus oocytes. Northern blots of LD-1 RNA hybridized strongly with oligonucleotide probes based on the published sequences for human G-CSF, but not with a probe based on the human GM-CSF sequence. Northern blots hybridized with an oligonucleotide probe based on the CSF-1 sequence showed a high-molecular weight band; however, low-molecular weight CSF-1 mRNAs, which are present in the CSF-1-producing cell line MIA-PaCa-2, were not detected in the LD-1 mRNA. The CSF activity of LD-1 cells is best described as human granulocyte CSF.

In Bacillus licheniformis, alpha-amylase production varied more than 100-fold depending on the presence or absence of a catabolite-repressing carbon source in the growth medium. alpha-Amylase was produced during the growth phase and not at the onset of the stationary phase. Induction of alpha-amylase correlated with synthesis of mRNA initiating at the promoter of the alpha-amylase gene.

We have cloned lines of IL 2-dependent human T cells derived from alloantigen, soluble antigen (tetanus toxoid), mitogen, or IL 2-stimulated peripheral blood lymphocytes and characterized their surface marker expression and cytolytic activity. The surface phenotype and cytolytic function was compared with the ability of these T cell clones to release cytotoxic lymphokines in response to mitogenic lectins. The cytotoxins released by these CTL clones were detected on the murine L929 target cells in a 16-hr assay. All of the T cell clones, whether stimulated by HLA alloantigens, tetanus toxoid, or mitogens, exhibited killer cell activity and the capacity to secrete a soluble cytotoxin(s). Specific polyclonal antisera to recombinant human tumor necrosis factor (rTNF) and human alpha-lymphotoxin (alpha LT) were unable to neutralize the cytotoxic activity released by most of these CTL clones. These results indicate that human CTL produce a novel antigenic form(s) of cytotoxin that we have termed CTL-toxin. Supernatants from several CTL clones yielded a cytotoxic activity that was partially neutralized (10 to 40%) by saturating levels of anti-TNF (but not anti-alpha LT) indicating that human CTL may be capable of producing a TNF-like molecule. Only two out of 60 CTL clones studied thus far produced a cytotoxic activity that was partially neutralized by anti-alpha LT (20 to 40%). Collectively, these results suggest that although both the CD4 and the CD8 subpopulations of human cytotoxic T cells may be capable of releasing several types of cytotoxins in response to mitogenic signals, the predominant cytotoxin is distinct from alpha LT and TNF.