Characterization of human and mouse granulocyte-macrophage-colony-stimulating factors derived from Escherichia coli

Heparin-binding Sites in Granulocyte-Macrophage Colony-stimulating Factor

The biological activity of granulocyte-macrophage colony-stimulating factor (GM-CSF) is modulated by the sulfated glycosaminoglycans (GAGs) heparan sulfate and heparin. However, the molecular mechanisms involved in such interactions are still not completely understood. We have proposed previously that helix C, one of the four alpha-helices of human GM-CSF (hGM-CSF), contains a GAG-binding site in which positively charged residues are spatially positioned for interaction with the sulfate moieties of the GAGs (Wettreich, A., Sebollela, A., Carvalho, M. A., Azevedo, S. P., Borojevic, R., Ferreira, S. T., and Coelho-Sampaio, T. (1999) J. Biol. Chem. 274, 31468-31475). Protonation of two histidine residues (His83 and His87) in helix C of hGM-CSF appears to act as a pH-dependent molecular switch to control the interaction with GAGs. Based on these findings, we have now generated a triple mutant form of murine GM-CSF (mGM-CSF) in which three noncharged residues in helix C of the murine factor (Tyr83, Gln85, and Tyr87) were replaced by the corresponding basic residues present in hGM-CSF (His83, Lys85, and His87). Binding assays on heparin-Sepharose showed that, at acidic pH, the triple mutant mGM-CSF binds to immobilized heparin with significantly higher affinity than wild type (WT) mGM-CSF and that neither protein binds to the column at neutral pH. The fact that even WT mGM-CSF binds to heparin at acidic pH indicates the existence of a distinct, lower affinity heparin-binding site in the protein. Chemical modification of the single histidine residue (His15) located in helix A of WT mGM-CSF with diethyl pyrocarbonate totally abolished binding to immobilized heparin. Moreover, replacement of His15 for an alanine residue significantly reduced the affinity of mGM-CSF for heparin at pH 5.0 and completely blocked heparin binding to a synthetic peptide corresponding to helix A of GM-CSF. These results indicate a major role of histidine residues in the regulation of the binding of GM-CSF to GAGs, supporting the notion that an acidic microenvironment is required for GM-CSF-dependent regulation of target cells. In addition, our results provide insight into the molecular basis of the strict species specificity of the biological activity of GM-CSF.

Isolation, expression and bioactivity of feline granulocyte–macrophage colony-stimulating factor

A cDNA encoding feline granulocyte-macrophage colony stimulating factor was cloned from alveolar macrophages using the reverse transcriptase-polymerase chain reaction (RT-PCR). The cDNA is 426 bp in length and encodes a predicted mature protein of 127 amino acids and the majority of the signal peptide. The recombinant protein (rfGM-CSF) was expressed in both Escherichia coli, as a calmodulin fusion protein, and mammalian cells. Biological activity of both recombinant proteins was demonstrated using the human erythroleukaemic cell line, TF-1. In a soft agar clonogenic assay, rfGM-CSF supported the development of granulocyte, macrophage and granulocyte-macrophage colonies. In combination with phytohaemagglutin (PHA) lymphocyte-conditioned medium, the number and size of such colonies were increased. Culture of feline bone marrow cells with rfGM-CSF was an efficient method for producing cells with morphology typical of dendritic cells (DC). The availability of the recombinant cytokine will permit further studies, in particular, the evaluation of the role of dendritic cells in feline immunopathology and its potential as a vaccine adjuvant.

A bicistronic expression system for bacterial production of authentic human interleukin-18

Interleukin-18 (IL-18) is activated and released from immune effector cells to stimulate acquired and innate immune responses involving T and natural killer (NK) cells. The release of IL-18 from mammalian cells is linked to its proteolytic activation by caspases including interleukin 1 converting enzyme (ICE). The absence of a signal peptide sequence and the requirement for coupled activation and cellular release have presented challenges for the large-scale recombinant production of IL-18. In this study, we have explored methods for the direct production of authentic human IL-18 toward the development of a large-scale production system. Expression of mature IL-18 directly in Escherichia coli with a methionine initiating codon leads to the production of MetIL-18 that is dramatically less potent in bioassays than IL-18 produced as a pro-peptide and activated in vitro. To produce an authentic IL-18, we have devised a bicistronic expression system for the coupled transcription and translation of ProIL-18 with caspase-1 (ICE) or caspase-4 (ICE-rel II, TX, ICH-2). Mature IL-18 with an authentic N-terminus was produced and has a biological activity and potency comparable to that of in vitro processed mature IL-18. Optimization of this system for the maximal production yields can be accomplished by modulating the temperature, to affect the rate of caspase activation and to favor the accumulation of ProIL-18, prior to its proteolytic processing by activated caspase. The effect of temperature is particularly profound for the caspase-4 co-expression process, enabling optimized production levels of over 150 mg/L in shake flasks at 25 degrees C. An alternative bicistronic expression design utilizing a precise ubiquitin IL-18 fusion, processed by co-expressed ubiquitinase, was also successfully used to generate fully active IL-18, thereby demonstrating that the pro-sequence of IL-18 is not required for recombinant IL-18 production.

Molecular cloning and expression of gerbil granulocyte/macrophage colony-stimulating factor

Using a combination of cross species reverse transcriptase-polymerase chain reaction and 3' rapid amplification of cDNA ends techniques, we cloned the cDNA encoding gerbil granulocyte/macrophage colony-stimulating factor (GM-CSF). The open reading frame had 81% nucleotide identity with its mouse counterpart, while the mature protein had 80% homology with mature mouse GM-CSF. COS-7 cells transfected with gerbil GM-CSF cDNA secreted high levels of bioactive GM-CSF, as their supernatant stimulated gerbil bone-marrow cell proliferation and colony formation in semi-solid medium.

Determining the identity and structure of recombinant proteins

In this unit peptide mapping protocols with separation of the constituent peptides by high-performance liquid chromatography (HPLC) analysis and by high-resolution SDS-PAGE are presented. Peptide mapping is ideally suited for comparative purposes--for example, combined analysis of the recombinant protein and its natural counterpart (or some other well-characterized standard). This unit also outlines the general strategy used to determine the linkage pattern of a monomeric recombinant protein containing two intramolecular disulfide bonds. The approach is an extension of peptide mapping, where the aim is to isolate and characterize peptides containing only a single disulfide bond. A two-dimensional electrophoretic method is also described in which the protein isoelectric point is displayed as a function of pH to yield an electrophoretic titration curve. This method is especially useful for checking for deamidation (e.g., of Asn to Asp) in which additional negative charge is introduced into the modified protein.

FTIR studies of recombinant human granulocyte-macrophage colony-stimulating factor in aqueous solutions: secondary structure, disulfide reduction and thermal behavior

Fourier transform infrared spectroscopy (FTIR) has been used to investigate the secondary structure, disulfide reduction and thermal behavior of recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF) in aqueous solutions. The contributions of amino-acid side-chain groups to the amide I bands of rhGM-CSF in H2O and in D2O solutions were carefully scrutinized, as 40% of the total 127 amino-acid residues of rhGM-CSF is side-chain absorptive (asparagine, glutamine, etc.). The FTIR results indicated that rhGM-CSF is composed of 46% alpha-helix, 7% beta-sheet, 23% turn and 24% loop/irregular structures which are in good agreement with the X-ray diffractional data. Reduction of rhGM-CSF with dithiothreitol caused apparent unfolding of the native conformation followed by the time-dependent increase of beta-aggregation bands which arose at 1622 and 1693 cm(-1) in H2O, 1613 and 1684 cm(-1) in D2O solutions. The result also showed that tertiary structure can change independently of the secondary structure. Thermal denaturation of rhGM-CSF took place at 55 to 70 degrees C and the denatured protein adopted an irregular structure as revealed by the FTIR spectra. The thermal denaturation did not show the formation of intermolecular beta-aggregates which is typical of most thermal denatured proteins. Moreover, it is partly reversible, indicating a special thermal stability of rhGM-CSF.

Production and characterisation of ovine GM-CSF expressed in mammalian and bacterial cells

A cDNA encoding ovine granulocyte-macrophage colony-stimulating factor (GM-CSF) was isolated and two forms of recombinant ovine GM-CSF were produced. A glycosylated form was produced in mammalian cells infected with a recombinant vaccinia virus encoding ovine GM-CSF. Recombinant ovine GM-CSF was also produced in Escherichia coli and purified by affinity chromatography. Both forms of the protein were detected by ovine GM-CSF-specific monoclonal antibodies, and exhibited activity on ovine bone marrow haemopoetic progenitor cells.

Purification of recombinant human granulocyte-macrophage colony-stimulating factor from the inclusion bodies produced by transformed Escherichia coli cells

Recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF), produced as inclusion bodies in genetically transformed Escherichia coli cells was purified to homogeneity by a three-step chromatographic procedure involving hydrophobic interaction, ion exchange and gel filtration. Each purification step is reproducible and well suited for process-scale operations. The purification process also leads to a significant decrease in DNA and endotoxin levels in the final product. Of the three gel media used, Phenyl Sepharose 6 FF (high sub) was most effective in reducing the DNA content (by a factor of ca. 2000) while Superdex 75 prep grade was more effective for removing endotoxins (reduction factor ca. 15). The recovery of purified rhGM-CSF was 35% by enzyme-linked immunosorbent assay and 70% by a biological assay method. The overall purification factor obtained was about 4.6, which is in the range of those reported for recombinant proteins produced in E. coli as inclusion bodies. The purified rhGM-CSF is an acidic protein (pI = 5.4) and has a specific activity of ca. 3.3 x 10(7) units/mg, which is in excellent agreement with that reported for its natural counterpart. Its monomer molecular mass of 14,605, as determined by electrospray mass spectrometry, corresponds exactly to the mass calculated from its cDNA sequence. Its amino acid composition and partial NH2-terminal sequence (up to seventeen residues) are also identical with those reported for this protein. These and other results confirm the identity of the purified rhGM-CSF with its natural counterpart. However, the results also showed that it is apparently heterogeneous from its NH2-terminal side as it is composed of three polypeptides having Met, Ala and Pro as the NH2-terminal residues in which the intact Met analogue accounts for 60% for the mixture. This heterogeneity does not seem to have any biological significance since the specific activity of the purified rhGM-CSF is identical with that of its natural counterpart.

Granulocyte-macrophage colony-stimulating factor mimicry and receptor interactions

Development of small molecular mimics of larger polypeptide ligands is an important approach to pharmacophore design. One strategy for the development of such mimics is analysis of alternative ligands that bind to the same site as the native ligand. These allow examination of the structural and chemical constraints for binding within the setting of diverse backbone geometries. The use of antireceptor antibodies as alternative ligands has allowed the development of biologically active peptides in several ligand-receptor systems. This technology has been applied to the study of interactions between human granulocyte-macrophage colony-stimulating factor (GM-CSF) and its receptor (GM-CSFR). GM-CSF is one of a family of signal-transducing cytokines and growth factors characterized by a four-helix bundle core structure. The GM-CSFR is comprised of an alpha-chain (GM-CSFR alpha) specific for GM-CSF, and a beta-chain (beta c) shared with the interleukin-3 and interleukin-5 receptors. At least two sites on GM-CSF have been implicated in the GM-CSF-GM-CSFR alpha/beta c ternary complex. In studies summarized here, synthetic peptide analogs of GM-CSF sequences were designed and used to map neutralizing epitopes. One neutralizing epitope corresponded to the A helix of GM-CSF, and a synthetic analog displayed biological activity as a GM-CSF antagonist in vitro, suggesting interaction with the GM-CSFR alpha/beta c complex. A second peptide comprising the B and C helices was recognized by monoclonal neutralizing antibodies and similarly displayed antagonist activity. Recombinant antibody (rAb) technology was also employed. An expression library of rAbs from mice immunized with neutralizing anti-GM-CSF antibodies was developed and screened with a neutralizing anti-GM-CSF monoclonal antibody. One clone which displayed receptor binding activity exhibited structural similarity with epitopes on GM-CSF previously implicated as interaction sites with the neutralizing monoclonal antibody. A synthetic peptide analog of the rAb inhibited GM-CSF bioactivity. Critical contact residues were predicted on the basis of structural similarity of the rAb peptide and GM-CSF. These studies indicate the feasibility of using rAbs in bioactive peptide design, providing lead compounds and information regarding contact residues for drug design.

An enzyme-linked immunosorbent assay (ELISA) for quantitation of adducts of granulocyte-macrophage colony stimulating factor (GM-CSF) and human serum albumin (HSA) in stressed solution mixtures

HPLC analyses of GM-CSF in solution mixtures containing both GM-CSF and HSA showed losses of GM-CSF which could not be accounted for using conventional electrophoretic and/or RP-HPLC techniques. Further investigation of these mixtures by immunoblotting and by immunoaffinity chromatography demonstrated the presence of high molecular weight (> 67,000 GM-CSF related species. No such species was detectable in solutions of GM-CSF alone. This experiment pointed to the formation of an adduct between GM-CSF and HSA in the solution mixtures. To probe further the hypothesis of a GM-CSF/HSA adduct, an immunologically based test was conceived which could react only with this type of hybrid molecule. A sandwich enzyme-linked immunosorbent assay (ELISA) was developed using two antibodies, anti-GM-CSF (capture antibody) and anti-HSA (detection antibody), as part of the quantitation of GM-CSF/HSA adducts. After confirming its existence by ELISA, a GM-CSF/HSA adduct was isolated from the solution mixture containing both GM-CSF and HSA. This isolate served as a primary reference standard in the ELISA assay. The immunoassay has a subnanogram sensitivity and is highly specific for GM-CSF/HSA adducts in the presence of either free GM-CSF or free HSA. As a verification, conjugates of GM-CSF/HSA were synthesized using a cross-linking reagent. These covalent conjugates reacted positively in the ELISA and are employed as a convenient alternative reference standard.

Isolation and characterization of a resistant core peptide of recombinant human granulocyte-macrophage colony-stimulating factor (GM-CSF); confirmation of the GM-CSF amino acid sequence by mass spectrometry

A trypsin-resistant core peptide of recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF) was isolated and analyzed by high-energy Cs+ liquid secondary-ion (LSI) mass spectrometric analysis. This analysis provided successful detection of the high-mass disulfide-linked core peptide as well as information confirming the existence of disulfide pairing. Similarly, LSI mass spectrometric analysis of the peptide fragments isolated chromatographically from a Staphylococcus aureus V8 protease digest of rhGM-CSF provided rapid confirmation of the cDNA-derived sequence and determination of the existing disulfide bonds between cysteine residues 54-96 and 88-121. Electrospray ionization mass spectrometry was employed to measure the molecular weight of the intact protein and to determine the number of the disulfide bonds in the protein molecule by comparative analysis of the protein before and after reduction with beta-mercaptoethanol.

Characterization of a saporin mitotoxin specifically cytotoxic to cells bearing the granulocyte-macrophage colony-stimulating factor receptor

When granulocyte-macrophage colony-stimulating factor (GM-CSF) is chemically conjugated to the ribosome-inactivating protein saporin, the resulting protein conjugate is highly toxic for cells expressing the GM-CSF receptor. Structural and Western blot analyses of the purified conjugate establish that it contains equimolar amounts of the starting materials and is free of any contamination by the non-conjugated components. The resulting bifunctional reagent is specifically cytotoxic to cells expressing the GM-CSF receptor, but is ineffective to cells that do not express the receptor. The cytotoxic activity is inhibited in a dose-dependent manner by GM-CSF, but not by any one of five other peptide growth factors. This is the first report of a mitotoxin for cells that express the GM-CSF receptor and which promises to be a valuable tool to study the expression of the GM-CSF receptor in normal and pathological states.

Purification and characterization of biologically active human recombinant 37 kDa soluble CD23 (sFc epsilon RII) expressed in insect cells

Human recombinant soluble 37 kDa CD23 has been expressed in insect cells and secreted into the culture medium using the IL-2 leader sequence. The 37 kDa CD23 was purified 600-fold to homogeneity by monoclonal antibody affinity chromatography and gel filtration. The pure protein is monomeric, glycosylated, depleted of one N terminal amino acid and contains four disulphide bonds. It degrades into smaller fragments of 33, 29 and 25 kDa if purified in the absence of protease inhibitors. The same pattern of proteolytic fragments is observed when the pure preparation is incubated at room temperature for 3 weeks. Physical characterization of the 37 kDa CD23 by circular dichroism indicates that the protein contains mainly beta sheet and 20% of alpha helical structures. Specific binding of IgE to natural CD23 (low affinity IgE receptor) was inhibited by purified recombinant 37 kDa CD23. Moreover, purified recombinant 37kDa CD23 and interleukin-1 promoted the survival of germinal centre B cells.

Three-dimensional structure of recombinant human granulocyte-macrophage colony-stimulating factor

The crystal structure of recombinant human granulocyte-macrophage colony-stimulating factor (rhGM-CSF) has been determined at 2.8 A resolution using multiple isomorphous replacement techniques. There are two molecules in the crystallographic asymmetric unit, which are related by an approximate non-crystallographic 2-fold axis. The overall structure is highly compact and globular with a predominantly hydrophobic core. The main structural feature of rhGM-CSF is a four alpha-helix bundle, which represents approximately 42% of the structure. The helices are arranged in a left-handed antiparallel bundle with two overhand connections. Within the connections is a two-stranded antiparallel beta-sheet. The tertiary structure of rhGM-CSF has a topology similar to that of porcine growth factor and interferon-beta. Most of the proposed critical regions for receptor binding are located on a continuous surface at one end of the molecule that includes the C terminus.

Molecular structure and granulocyte/macrophage colony-stimulating factor activity

Granulocyte/macrophage colony-stimulating factor (GM-CSF) plays a critical role in myeloid differentiation and in several immune and inflammatory processes. GM-CSF binds to specific cellular receptors (GM-CSFR) which belong to a recently described supergene family. These receptors are potential targets for pharmacologic design and such design depends on a molecular understanding of ligand-receptor interactions. We present our initial studies evaluating the potential active sites of the molecule. The sites on the GM-CSF molecule that were studied represent two alpha-helices predicted to be critical for GM-CSF activity, as implicated by human-murine chimeric molecule studies. These helices are predicted to be adjacent in native GM-CSF. Peptides corresponding to amino acids 17-31 and 78-99 of GM-CSF were synthesized and cross-linked to one another in two different orientations. The ability of anti-GM-CSF to bind the individual and complexed peptides was evaluated by both ELISA and radioimmunoassay. Significant binding to all peptides was demonstrated. A preferred orientation of the two peptides was apparent, and this agreed with the predicted model structures. Antibodies were developed against the coupled peptides, and these demonstrated significant cross-reactivity with recombinant human GM-CSF. Additionally, analyses of anti-peptide antisera binding studies predict these two amino acid sequences to lie in parallel planes to one another in the native human GM-CSF molecule.

Evaluation of human N-linked glycosylation sites in murine granulocyte-macrophage colony-stimulating factor

Nonglycosylated murine and human granulocyte-macrophage colony-stimulating factor have a molecular mass of approximately 14.5 kDa predicted from the primary amino acid sequence. The expression of both proteins in COS cells leads to a heterogeneous population of molecules that differ in the degree of glycosylation. Both human and murine molecules contain two N-linked glycosylation sites that are situated in nonhomologous locations along the linear sequence. Despite this difference both proteins show a similar size distribution among the glycosylation variants. These studies analyze the effects of introducing in the murine protein novel N-linked glycosylation sites corresponding to those sites found in the human molecule. A panel of molecules composed of various combinations of human N-linked glycosylation sites in either the presence or the absence of murine N-linked glycosylation was compared. Substitution of a proper human N-linked glycosylation consensus sequence at Asn 24 did not result in N-linked glycosylation, nor was there any considerable effect on bioactivity. Replacement of the N-linked glycosylation consensus sequence at Asn 34 results in glycosylation similar to that found in the human molecule and causes a significant decrease in bioactivity. These data suggest that the position of N-linked glycosylation is critical for maximal bioactivity in a particular species and that the changes in position of these sites in different species probably occurred during evolution in response to changes in their receptors.

Structure-function relationships of the hematopoietic growth factors

The hematopoietic growth factors are a family of glycoproteins involved in the production of blood cells from their bone marrow precursors and in the activation of mature blood cells. Much has been learned about the structural features of these molecules responsible for their characteristic biological activities. Most studies have been based upon mutagenesis strategies of intact polypeptides and on epitope mapping of informative monoclonal antibodies to the growth factors. A more limited amount of physical data is available. This review will summarize these findings, highlight the growing body of evidence suggesting that many of these proteins share common evolutionary origins and structural elements, and hopefully point to the directions being taken for further investigations of these scientifically informative and clinically useful group of proteins.

Low-resolution structure of recombinant human granulocyte-macrophage colony stimulating factor

A recombinant form of human granulocyte-macrophage colony stimulating factor (GM-CSF) which contains no carbohydrate has been crystallized. Multiple isomorphous replacement analysis using five heavy-atom derivatives has yielded an image of the structure at 6 A resolution that showed two molecules per asymmetric unit and allowed determination of the non-crystallographic symmetry transformation. The 6 A resolution result shows that the core of GM-CSF consists of four helices. The angles at which the helices pack together distinguishes this structure from known antiparallel four-helix bundle proteins. Consideration of the amino acid sequence properties and previous structural characterizations of GM-CSF leads to an assignment of the probable protein segments that form the helices.

Enhanced hematopoietic activity of a human granulocyte/macrophage colony-stimulating factor-interleukin 3 fusion protein

Granulocyte/macrophage colony-stimulating factor-interleukin 3 (GM-CSF-IL-3) fusion proteins were generated by construction of a plasmid in which the coding regions of human GM-CSF and IL-3 cDNAs were connected by a synthetic linker sequence followed by subsequent expression in yeast. Both GM-CSF-IL-3 and IL-3-GM-CSF fusion proteins were purified to homogeneity and shown to bind to cell-surface receptors through either their GM-CSF or IL-3 domains. The fusion proteins exhibited enhanced receptor affinity, proliferative activity, and hematopoietic colony-stimulating activity compared with either IL-3 and/or GM-CSF alone. This suggests that GM-CSF-IL-3 fusion proteins may hold future promise as therapeutic agents.

Preparation and characterization of human interleukin-5 expressed in recombinant Escherichia coli

The gene coding for human interleukin-5 was synthesized and expressed in Escherichia coli under control of a heat-inducible promoter. High-level expression, 10-15% of total cellular protein, was achieved in E. coli. The protein was produced in an insoluble state. A simple extraction, renaturation and purification scheme is described. The recombinant protein was found to be a homodimer, similar to the natural murine-derived protein. Despite the lack of glycosylation, high specific activities were obtained in three 'in vitro' biological assays. Physical characterization of the protein showed it to be mostly alpha-helical, supporting the hypothesis that a conformational similarity exists among certain cytokines.

Deletion of carboxy-terminal residues of murine granulocyte-macrophage colony-stimulating factor results in a loss of biologic activity and altered glycosylation

A deletion mutant of murine granulocyte-macrophage colony-stimulating factor (GM-CSF) which differs in primary structure from native GM-CSF in the carboxy-terminal 11 amino acids was prepared. Four amino acid residues are mutated and the seven terminal residues including Cys-118 are deleted. Supernatants from COS-1 cells transfected with this deletion mutant (GM-CSF(del] showed a 3000-fold decrease in the ability to stimulate bone marrow stem cells to proliferate and differentiate into granulocytes and macrophages in vitro. Northern blot analysis using poly(A)+ RNA extracted from the transfected cells showed equal accumulations of GM-CSF and GM-CSF(del). Transfection with full-length GM-CSF followed by immunoprecipitation of metabolically labeled supernatant proteins with rabbit anti-rGM-CSF antiserum yielded predominantly the 23-kDa, fully glycosylated form and small amounts of both a 29-kDa form and the 18-kDa non-N-glycosylated form. Transfection of the GM-CSF(del) mutant and immunoprecipitation revealed a large, diffuse band on sodium dodecyl sulfate--polyacrylamide gel electrophoresis with a molecular weight of about 29 kDa. Digestion of the immunoprecipitated 29-kDa species with N-glycanase converted the 29-kDa form into two forms of about 23 and 18 kDa, suggesting that the increase in molecular weight of the deletion mutant protein resulted from hyperglycosylation. Adding tunicamycin to the culture medium of cells transfected with GM-CSF(del) also yielded a single non-N-glycosylated species of about 18 kDa, but secretion was at a significantly lower level than either the 29-kDa hyperglycosylated GM-CSF(del) protein from non-tunicamycin-treated cells or the 18-kDa non-N-glycosylated full-length GM-CSF from tunicamycin-treated cells. Since very recent scanning-deletion analysis indicates that there is a critical region for activity near Cys-118 and that Cys-118 is necessary for maximal activity, we conclude that the Cys-118 residue is necessary for proper glycosylation and maximal biologic activity of GM-CSF.

Identification of critical regions in mouse granulocyte-macrophage colony-stimulating factor by scanning-deletion analysis

Structure-function relationships for mouse granulocyte-macrophage colony-stimulating factor were examined by generating a series of small deletions scanning the entire length of the molecule. Deletions of three amino acids were introduced at intervals of five amino acids by site-directed mutagenesis of the mature mouse granulocyte-macrophage colony-stimulating factor gene. The mutant proteins were expressed in Escherichia coli and assayed for biological activity. This procedure identified four regions critical to activity. These critical regions were further delineated by additional three-amino acid deletion mutants. Larger deletions at each terminus were also made, as well as changes of specific amino acid residues. The four critical regions span amino acid residues 18-22, 34-41, 52-61, and 94-115. The disulfide bridge between Cys-51 and Cys-93 was also shown to be essential for activity, whereas that between Cys-85 and Cys-118 could be removed without loss of activity. The possible structural and/or functional roles of the critical regions are discussed.

Crystallization and preliminary X-ray studies of recombinant human granulocyte-macrophage colony stimulating factor

Human granulocyte-macrophage colony stimulating factor (hGM-CSF) is an important regulator of growth and differentiation for mononuclear and polymorphonuclear phagocytes. Here we report the crystallization and preliminary X-ray studies of Escherichia coli-expressed hGM-CSF. The crystals are orthorhombic, with the space group P212121, and have unit cell dimensions a = 46.62 A, b = 58.73 A and c = 126.42 A. Recombinant hGM-CSF crystals diffract X-rays to 2.4 A resolution and are thus suitable for X-ray structural studies.

Characterization of recombinant-derived granulocyte-colony stimulating factor (G-CSF)

Human granulocyte colony-stimulating factor (G-CSF), and a mutant having a Ser for Cys substitution at residue 18 were produced in Escherichia coli strain W3110. About 60 mg of pure protein was obtained from 50 g of wet cells with a recovery of about 20%. The proteins were characterized physically and chemically, including determination of disulphide bonds, which were found to exist between residues 37-43 and 65-75. Cys-18 is not involved in disulphide bond formation and was substituted by Ser with no effects on gross protein conformation or biological activity. Both the wild-type and the mutant recombinant-derived proteins, although not glycosylated, possess colony-stimulating activities. In a bioassay using the murine myelomonocytic leukaemic cell line WEH1 3B D+, activities were obtained which were similar to those of natural G-CSF and of a glycosylated recombinant-derived human G-CSF produced in monkey cells.

The conformation and stability of recombinant-derived granulocyte-macrophage colony stimulating factors

The conformation and stability of recombinant-derived human and murine granulocyte-macrophage colony stimulating factors produced in Escherichia coli have been investigated by analytical ultracentrifugation, urea-gradient polyacrylamide gel electrophoresis and several spectroscopic methods. The proteins were demonstrated to be physically homogeneous monomeric proteins with compact globular shapes and shown to have similar secondary structures containing both alpha-helix and beta-sheet structure. The intramolecular disulphide linkages of both proteins were shown to be essential for maintaining native conformation as reduction with dithiothreitol resulted in protein unfolding. Comparison of the human E. coli-derived (non-glycosylated) and mammalian cell culture-derived (glycosylated) proteins by urea-gradient electrophoresis indicated that glycosylation had no major effect on the conformational stability and kinetics of urea induced unfolding and refolding.

C-terminal peptide identification by fast atom bombardment mass spectrometry

A previously described technique [Rose, Simona, Offord, Prior, Otto & Thatcher (1983) Biochem. J. 215, 273-277] permits the identification of the C-terminal peptide of a protein as the only peptide that does not incorporate any 18O upon partial enzymic hydrolysis in 18O-labelled water. Formation of chemical derivatives followed by combined g.l.c.-m.s. was used in this earlier work. We now describe the isolation from protein digests, by reversed-phase h.p.l.c., of labelled and unlabelled polypeptides and their direct analysis by fast atom bombardment mass spectrometry. Under the conditions used, the 18O label is retained throughout the separation and analysis, thus permitting assignments of C-terminal peptides to be made. Enzyme-catalysed exchange of label into the terminal carboxy group was found to occur in some cases without hydrolysis of a peptide bond. This effect, which may be exploited to prepare labelled peptides, does not prevent application of the method (two separate digests must then be used). We have applied our method to the analysis of enzymic partial hydrolysates of glucagon, insulin and of several proteins produced by expression of recombinant DNA.

Native cytokines do not bind to uromodulin (Tamm-Horsfall glycoprotein)

Uromodulin bound with high affinity to human tumour necrosis factor (TNF) coated on microtitre plates. This interaction was not competitively inhibited by native TNF in solution. No interaction was observed between immobilized uromodulin and TNF in the liquid phase unless conditions were chosen which denatured the latter protein. Recombinant interleukin-1 alpha adsorbed on microtitre plates also interacted with uromodulin. However, gel filtration experiments demonstrated no interaction between the proteins in the liquid phase. These and additional results indicate that uromodulin interacts with denatured cytokines, but not with native, soluble cytokines.