WO1998004589A2 - Production of recombinant plasma gelsolin containing a disulfide bond - Google Patents

Abstract

A method is provided for producing recombinant gelsolin containing a disulfide bond. The oxidized gelsolin may be produced by treating intracellular recombinant gelsolin with an oxidant, or by denaturing gelsolin and then refolding gelsolin.

Description

PRODUCTTON OF RECOMBINANT PLASMA GELSOLIN CONTAINING A DISULFIDE BOND

Field of the Invention

This application relates to methods of production of recombinant plasma gelsolin.

Background of the Invention

Gelsolin is an actin binding protein that regulates the length of actin filaments. As indicated by recent data from mice lacking the gene for gelsolin, this protein has an important role in hemostasis, inflammatory responses, and fibroblast responses (Witke et al., Cell 81:41-51, 1995). Gelsolin has promise as a potential therapeutic agent for the treatment of patients with cystic fibrosis (Vasconcellos et al., Science 263:969-971, 1994).

Human plasma gelsolin is a nonglycosylated protein consisting of 755 amino acid residues. It is made up of six repeated domains each containing approximately 110 amino acids. The putative domain boundaries are residues 30-149, 150-266, 267-389, 407-528, 529-634, and 635-755 (numbering based on the mature plasma gelsolin sequence). The six domains make up three distinct actin binding sites. The gelsolin sequence contains five cysteine residues at positions 93, 188, 201, 304, and 645. Natural gelsolin exists in two forms, plasma and cytoplasmic, which are respectively extracellular or intracellular. The two gelsolin forms are generated from a single gene product by alternative splicing. Both forms contain the 730 amino acid core structure made up of six repeated domains, and differ in sequence only by the presence of an N-terminal extension in the plasma form (Kwiatkowski et al., Nature 323:455-458, 1986). This N-terminal extension comprises 25 amino acids of unknown function, as well as a signal sequence that is removed during secretion. The disulfide structure of human gelsolin is unknown, but in a recently published NMR structure for domain 2 from severin, the authors noted the presence of a disulfide bond (Schnuchel et al., J. Mol. Biol. 247:21-27,1995). The presence of a single disulfide in bovine plasma gelsolin, which contains seven Cys residues, has been inferred from studies with Ellman's reagent (Kilhoffer and Gerard, Biochemistry 24:5653-5660, 1985). Cytoplasmic gelsolin is folded differently from plasma gelsolin. Cytoplasmic and plasma gelsolin appear to have different regulation and function. In early development of Drosophila embryos, both forms are expressed throughout the embryo initially, but in later development the two forms of gelsolin are expressed in specific tissues (Stella et al., J. Cell Biol. 125:607-616, 1994).

The reason for the two forms of gelsolin is unknown. The scarcity of the cytoplasmic form and difficulty in isolating it free from actin have hampered a rigorous comparison of potential differences between the proteins. Typical purification methods rely on denaturants or chaotropes as part of the method, which may alter the properties of the protein. In one study where actin severing properties of cytoplasmic and plasma forms of human gelsolin were compared by microinjection into cells, the proteins displayed different activities (Huckriede et al., Cell Motil. Cytoskel.16:229-238, 1990), raising the possibility that structural differences might also exist. Source of gelsolin may also affect activity: gelsolin isolated from human plasma readily degraded stress fibers from tissue culture cells, while "human plasma gelsolin" expressed in E. coli lacked the ability to cleave the stress fibers (Huckriede et al.. Cell Motil. Cytoskel.16:229-238, 1990). Gelsolin is a potential therapeutic for diseases in which modulating levels of actin filaments is desirable. For therapeutic and regulatory purposes, a pure preparation of one form of a compound is required. Plasma gelsolin is the form normally present in vertebrate extracellular fluids, so is the preferred form of gelsolin for use as a pharmaceutical product. In the past, plasma gelsolin has been isolated from human plasma, a method which is undesirable because of low yield, need for large volumes of human blood, lack of purity, and risk of blood-borne infectious disease. Gelsolin has also been produced by expression of the human plasma gelsolin gene in recombinant hosts, such as CHO cells, NS/0 mouse myeloma cells, yeast such as Pichia, or E. coli. In tests by the inventors, gelsolin expression levels have been inadequate for CHO cells and Pichia. Furthermore, gelsolin produced in Pichia and NS/0 is unstable, being very susceptible to proteoiysis. Harvest of gelsolin produced by secretion in recombinant E. coli requires osmotic shock to break the cell wall and release the secreted product; this procedure results in contamination of the product gelsolin with reduced gelsolin. It is an object of the invention to provide a method of producing pure, recombinant plasma gelsolin containing a Cys - Cys disulfide bond, such as is found in native plasma gelsolin. It is a further object of the invention to provide a method for producing properly folded plasma gelsolin in E. coli. Summarv of the Invention The invention provides a method for producing recombinant gelsolin or a recombinant gelsolin fragment containing a disulfide bond, including contacting a sample containing reduced gelsolin or reduced gelsolin fragment with an oxidant. The oxidant may be oxygen, peroxide, a second disulfide compound, radiation, a biological oxidant or a nonbiological oxidant. Where the oxidant is oxygen or peroxide, the sample may be further contacted with a catalyst. The catalyst may be a free metal, a metal ion, a metal complex, or selenium. Where the oxidant is a second disulfide compound, it may be an oxidized form of dithiothreitol, dithioerythritol, cystamine, glutathione, glutathione derivatives , thioredoxin, ferredoxin, cystine, or di-β- hydroxyethyl disulfide. A suitable biological oxidant may be a flavin, a cytochrome, or an ascorbate. A nonbiological oxidant of the invention can include a halogen, a redox dye, a nitrate or nitro compound, an azo compound, or a sulphoxide.

The sample may be contacted with both an oxidant and with calcium. The invention also provides a method of producing a recombinant gelsolin molecule or gelsolin fragment containing a disulfide bond; the method includes expressing a gelsolin molecule or gelsolin fragment in a cell transfected with a transgene encoding gelsolin or a gelsolin fragment, disrupting the cell, and inducing the formation of a disulfide bond within the recombinant gelsolin molecule or gelsolin fragment. The disulfide bond may be induced by contacting the gelsolin with a denaturing compound, then removing the denaturing compound under conditions which allow refolding of the gelsolin.

Another method of the invention includes producing recombinant gelsolin or a gelsolin fragment containing a disulfide bond using steps including expressing gelsolin or gelsolin fragment in a cell transfected with a gene for gelsolin, and contacting the gelsolin or gelsolin fragment with an oxidant. The gelsolin may be plasma gelsolin or a fragment of plasma gelsolin, or human gelsolin or a fragment of human gelsolin. The gelsolin or gelsolin fragment may be contacted with an oxidant either before the gelsolin or gelsolin fragment is purified, or after the gelsolin or gelsolin fragment is purified. The cell may be a bacterial, yeast, insect, or mammalian cell.

A further embodiment of the invention provides recombinant gelsolin containing a disulfide bond, produced by lysis of a transgenic cell containing the gelsolin, followed by treatment of the gelsolin with an oxidant. A further method of the invention provides a recombinant gelsolin fragment containing a disulfide bond, produced by proteolytic cleavage of full-length gelsolin or of a longer gelsolin fragment which had previously been treated with one of the methods described above. In a variant of this method, the recombinant gelsolin fragment containing a disulfide bond is produced by first proteolytically cleaving reduced full-length gelsolin or a longer reduced gelsolin fragment, followed by treating the resultant gelsolin fragment with an oxidant.

Brief Description of the Drawings

FIGURE 1 is a graph showing the dependence of the actin severing activity of gelsolin on calcium. In the rhodamine phalloidin assay, fluorescence decreases as actin is severed by gelsolin. Both natural and recombinant gelsolin show increased severing activity with increased calcium concentration. The natural protein is more active than recombinant at the low calcium concentrations.

FIGURE 2 is an SDS-PAGE gel showing an analysis of the proteolytic susceptibility of plasma gelsolin. Preparations of human plasma gelsolin alone (lanes a-d) or after treatment with plasmin (lanes e-p) are analyzed by SDS-PAGE. The proteins (3μg/lane) are stained with Coomassie blue. Lanes a, e, i, and m are natural plasma gelsolin purified as described in the materials and methods. Lanes b, f, j, and n are natural plasma gelsolin purified by the one step elution method. Lanes c, g, k, and o are recombinant plasma gelsolin from E.coli. Lanes d, h, 1, and p are recombinant plasma gelsolin from Cos cells. The positions of molecular weight standards are indicated at the left of the figure.

FIGURE 3 is a plot showing an identification of PE-Cys peptides in endo Asp-N digests of recombinant gelsolin. Recombinant gelsolin from E.coli is treated with full vinylpyridine, then digested with endo Asp-N and subjected to peptide mapping by reversed phase HPLC on a Cj₈ column. Elution profiles at 214, 254 and 280 nm are shown. PE-Cys containing peptides are denoted as peaks A-D. UV-absorption spectra for the peaks are shown at the right of the figure. The presence of PE-Cys and of Trp are indicated by respective absorption maxima at 254 and 280 nm. FIGURE 4 is a plot comparing endo Asp-N digests of recombinant and natural plasma gelsolin. Partial HPLC elution profiles of endo Asp-N digests of unreduced, pyridylethylated preparations of natural (lower panel) and recombinant gelsolin from E.coli (upper panel) are monitored at 254 nm. The four PE-Cys containing peaks are indicated (A-D). UV spectra of PE-Cys-containing peaks B-D from natural gelsolin are essentially identical to those in Figure 2 for recombinant gelsolin. Inset: UV-absorption spectrum of peak E, which contains the Cys¹⁸⁸- Cys²⁰¹ disulfide.

FIGURE 5 is a plot showing identification of PE-Cys peptides in endo Lys-C digests of recombinant plasma gelsolin. The HPLC elution profiles from an endo Lys-C digest of unreduced, pyridylethylated, recombinant gelsolin from E.coli at 214, 254 and 280 nm are shown. The lettered peaks denote PE-Cys containing peptides as indicated by UV maxima at 254 nm (data not shown).

FIGURE 6 is a plot showing comparisons of endo Lys-C digests of natural plasma and cytoplasmic gelsolin. Partial HPLC elution profiles from endo Lys-C digests of reduced, pyridylethylated natural plasma gelsolin (upper panel), unreduced, pyridylethylated natural plasma gelsolin (middle panel), and pyridylethylated natural cytoplasmic gelsolin (lower panel) monitored at 214 nm are shown. Peaks F, Gl and G2 are as described in Figure 5. Asterisks denote N-terminal endo Lys-C peptides from cytoplasmic and plasma gelsolin.

FIGURE 7: Assessing the proteolytic susceptibility of glutathione-oxidized recombinant gelsolin and platelet-derived cytoplasmic gelsolin to digestion with plasmin. The samples indicated below are treated with plasmin at an enzyme to gelsolin ratio of 1 : 10 in calcium containing buffer as described in the experimental section. Cleavage products are subjected to SDS-PAGE on a 4-20% gradient gel and visualized with Coomassie blue. Recombinant human plasma gelsolin from E.coli is plasmin digested directly (lane d), or after the following treatments: an eight fold molar excess of f-actin (lane e), a six fold molar excess of 2C4 antibody in solution (lane f)» immobilization on 2C4-Sepharose (lane h). Recombinant human plasma gelsolin produced in E.coli that had been oxidized with glutathione has been analyzed directly (lane b), digested with plasmin in solution and analyzed (lane c), or immobilized on 2C4 CNBr Sepharose and digested with plasmin (lane g). Natural human cytoplasmic gelsolin from platelets is immobilized on 2C4-Sepharose and then analyzed with (lane j) or without digestion with plasmin (lane i). Lane a, prestained molecular weight markers from GibcoBRL. The positions of molecular weight standards are indicated at the left of the figure. The arrowhead at the right marks the position of the 65 kDa cleavage fragment. Prominent bands at 42 kDa in lanes e, i , and j, at 55 kDa in lane f, and at 25 kDa in lanes f-i correspond to actin, IgG heavy chain, and IgG light chain, respectively.

FIGURE 8 is an SDS-PAGE gel showing plasmin digestion of natural, recombinant and oxidized recombinant gelsolin after an overnight treatment of the gelsolin forms with 5 mM glutathione and 2 mM CaCl₂. Samples are tested for the suceptability to digestion with plasmin either in the presence of EDTA (first three lanes) or 2 mM calcium (last three lanes). The molecular weight markers are as in Figure 2. EDTA treatments are used as a control to show that the oxidation process does not impact structure. Digestion in the presence of calcium is used for changes in the disulfide structure. In the presence of calcium, gelsolin that has the correct disulfide bond will be cleaved to a 70Kd band, while gelsolin that has all sulfhydryls as free thiols will digest to a 65Kd band.

FIGURE 9 is an SDS-PAGE gel showing the effect of pH on disulfide formation by oxidized glutathione. After treatment with oxidant, the samples indicated are analyzed for susceptability to digestion with plasmin either in the presence of EDTA (first four lanes) or calcium (last four lanes). Both the pH 7.5 and 7.0 samples have decreased levels of the oxidized product.

FIGURE 10 is an SDS-PAGE gel showing the effect of calcium and temperature on oxidation. After the treatment indicated in the figure, samples were analyzed for susceptibility to digestion with plasmin. Only data for digestions performed in the presence of calcium is shown. Oxidation at 37 °C is the best, followed by oxidation at room temperature. Oxidation at 4°C is the slowest. Oxidation only occurs in the presence of calcium; EDTA inhibits the oxidation process..

FIGURE 11 is an SDS-PAGE gel showing extent of oxidation with varying glutathione concentrations that range from 0.6 mM to 20 mM. After the treatments indicated, samples are analyzed for susceptability to digestion with plasmin (only data for digestions performed in the presence of Ca are shown). 5 mM is the minimum amount of oxidant needed to achieve the highest level of oxidation.

FIGURE 12: Apparent features of domain 2 of gelsolin based on the villin structure. The previously published NMR structure of domain 1 from villin (Markus et al., Protein Sci. 3:70-81 , 1994) is used as a model for domain 2 of gelsolin. Residue numbers 145 and 264 indicate the sequence orientation. The positions of the relevant amino acids Cys¹⁸⁸, Cys²⁰¹, His¹⁵¹, and Ala²²⁹ are indicated. Apparent distances between His¹⁵¹ and Cys residues 188 or 201, and Ala²²⁹ and Cys residues 188 or 201 range from 17-20 A. The apparent distance between His¹⁵¹ and Ala²²⁹ is 16 A. Green circles denote the relative positions of the Cys residues from domain 2 of severin that are disulfide linked.

Detailed Description of the Invention We have developed a peptide mapping/mass spectrometric strategy to analyze the disulfide structure of gelsolin. We have shown that all five of the Cys residues in cytoplasmic gelsolin are in the free thiol form, while only three of the five thiols in plasma gelsolin are free. In plasma gelsolin, an intradomain disulfide links Cys¹⁸⁸ and Cys²⁰¹. The mapping results provide the first definitive evidence that the cytoplasmic and plasma forms of gelsolin are structurally distinct. The changes in disulfide structure track with differences in proteolytic susceptibility and provide the first definitive evidence for structural differences between plasma and cytoplasmic gelsolin. We have also found that recombinant plasma gelsolin expressed in E.coli lacks the disulfide, but can be converted to the plasma-like structure with mild oxidation. We have used this insight to develop a method of production for recombinant gelsolin which includes an oxidation step. This method yields gelsolin with the disulfide bond characteristic of native human plasma gelsolin. When "intracellular" gelsolin is recovered from a preparation of disrupted transgenic cells in which recombinant gelsolin has been expressed, it is at least partially reduced. By "reduced", we mean that at least 5% of the gelsolin in the preparation contains no disulfide bond. When treated with the methods of the invention, such gelsolin becomes fully oxidized, meaning that at least 80% of the gelsolin in the preparation contains a disulfide. Preferred methods of the invention yield preparations of gelsolin in which at least about 90% of the gelsolin in the preparation contains a disulfide.

The methods of the invention are intended for use not only on full length cytoplasmic and plasma gelsolin, but also on fragments of these molecules. By fragments, we mean any portion of the native plasma or cytoplasmic gelsolin which contains at least domains 1 and 2 of the full length molecule; this would specifically include gelsolin fragments encompassing domains 1-2, 1-3, 1-4, 1-5 and 1-6. Fragments of gelsolin can be produced using any available means of proteoiysis, as are well known to protein chemists. In one embodiment, proteolytic enzymes can be used to cleave the gelsolin into fragments for use in the methods of the invention. Specific examples of useful proteolytic enzymes include chymostrypsin (Kwiatkowski et al., J. Biol. Chem. 260: 15232-15238, 1985), thermolysin (Chaponnier et al., J. Cell Biol. 103: 1473-1481, 1986), and subtilisin (using the methods described for the gelsolin-like protein brevin, as described by Bryan and Hwo, J. Cell Biol. 102: 1439-1446, 1986). The proteoiysis can be done either before or after treatment with an oxidant. The gelsolin useful in the methods of the invention may be obtained by culturing and lysing any cell containing a transgene for gelsolin, including bacterial cells such as E. coli, yeast cells, insect cells, or mammalian cell lines. Suitable yeast cells include Saccharomyces cerevisiae, Hansenula polymorpha, Pichia pastoris, Kluyveromyces lactis, Schisosaccharomyces pombe, Schwanniomyces occidentalis, or Yarrowia lipolytica. Examples of suitable mammalian cell lines include monkey kidney CVI line transformed by SV 40 (COS- 7, ATCC CRL 1651); human embryonic kidney line 293S (Graham et al., J. Gen. Virol. 36:59, 1977); baby hamster kidney cells (BHK, ATCC CCL 10 ); mouse sertoli cells (TM4, Mather, Biol. Reprod. 23:243, 1980); monkey kidney cells (CVI-76, ATCC CCL 70); African green monkey kidney cells (VERO-76, ATCC CRL-1587); human cervical carcinoma cells (HELA, ATCC CCL 2);canine kidney cells (MDCK, ATCC CCL 34); buffalo rat liver cells (BRL 3A, ATCC CRL 1442); human lung cells (W138, ATCC CCL 75); human liver cells (Hep G2, HB 8065); mouse mammary tumor cells (MMT 060562, ATCC CCL 51); rat hepatoma cells (HTC, MI 54, Baumann et al., J. Cell Biol. 85: 1, 1980); and TR-1 cells (Mather et al., Ann. N.Y. Acad. Sci. 383:44, 1982) and hybridoma cell lines.

EXPERIMENTAL PROCEDURES A. Oxidation of Reduced Gelsolin

Reduced gelsolin may be oxidized to plasma gelsolin using any oxidant, including oxygen, peroxide, a second disulfide compound, radiation, biological oxidants and nonbiological oxidants. Where the oxidant is oxygen or peroxide, oxidation also requires the presence of a catalyst.

Suitable catalysts include free metals such as Hg, Pt and Au, as well as traces of metal ions, metal complexes, and selenium as selenite. Suitable metal ions include Cu⁺⁺, Fe⁺⁺, Co⁺⁺, Mn⁺⁺, and selenite. The rate of oxidation may be influenced by temperature, pH, buffer, type of catalyst, oxygen tension, and concentration of gelsolin. Disulfide oxidants include any disulfide compound, such as oxidized forms of dithiothreitol, dithioerythritol, cystamine, glutathione or its derivatives, thioredoxin, ferredoxin, cystine, and di-β-hydroxyethyl disulfide.

Reduced gelsolin can also be oxidized by exposure to radiation, such as X-rays, β-rays, or ultraviolet radiation. Radiation-induced oxidation increases proportionately with increase in radiation intensity.

Biological oxidants include flavins such as riboflavin, flavin mononucleotide, and flavin dinucleotide; cytochromes such as cytochromes a, b or c; and ascorbates such as dehydroascorbate.

Nonbiological oxidants include halogens such as iodine or bromine; redox dyes such as methylene blue, tetrazolium salts, chloranil, or dichloφhenol indophenol; nitrate and nitro compounds such as nitroxide, nitroglycerol, peroxyacetylnitrate or tetranitromethane; azo compounds such as acetylphenylhydrazine or other azo esters or amides; and sulphoxides such as dimethylsulphoxide.

B. Isolation. Oxidation, Recovery and Analysis of Fully Oxidized Plasma Gelsolin Many procedural substitutions can be made in the following methods for producing oxidized gelsolin, as are known to those of skill in the art.

In one procedure, 3 g of E.coli cell paste producing human plasma gelsolin intracellularly is suspended in 21 ml of 50 mM Tris (pH 8.3) with 1 mM EDTA, and sonicated for 3x30 seconds on ice. Other disruption methods may include breaking cells in a Manton Gaulon, a French press, or by detergent lysis. The lysate is centrifuged at 12k φm (SA600 rotor) for 30 min and the supernatant (23 ml) is collected. Floculating reagents such as polyimine can be used to reduce the centrifugation speed needed to clarify the lysate. The pellet is washed with 12 ml of the lysis buffer, centrifuged, and the supernatant is combined with the original supernatant. The pooled supernatant is diluted with 1/2 vol of Dl-water, and filtered through a 0.2 μ membrane.

Gelsolin is then isolated by anion exchange chromatography. While the procedure is described using DEAE-sepharose, any anion exchange resin could be utilized. The lysate is loaded on a 4 ml DEAE-sepharose column. The column is washed with 5 column volumes (CN.) of 25 mM Tris HCl, 1 mM EGTA, 45 mM ΝaCl, pH 8.3, and eluted with 7 CN. of 25 mM Tris HCl, 1 mM EGTA, 100 mM ΝaCl, pH 8.3 ( 1 CN. fraction is collected). The absorbance at 280nm of the fractions is read, and the protein peak is pooled. The elution pool is diluted with 1 vol of 25 mM Tris HCl, 1 mM EGTA, 45 mM ΝaCl, pH 8.3, and filtered through a 0.2 μ membrane. The DEAE eluate is next subjected to purification on Q-sepharose on a 2.5 ml column. Once loaded, the column is washed with 7 CN. of 25 mM Tris HCl, 0.1 mM EDTA, 45 mM ΝaCl pH 8.3 and eluted with 5 C.V.of 25 mM Tris, 45 mM ΝaCl, 2 mM CaCl₂ , pH 8.3. 0.5 CN. fractions are collected and analyzed for absorbance at 280nm.

To oxidize the gelsolin, the Q eluate is treated with 5 mM Glutathione (Oxidized form) and incubated at 4°C for 18-24 hr. The gelsolin sample is acidified with 0.1 vol of 0.5 M sodium acetate pH 5.0 and the pH is further adjusted to 6.0 with HCl. In the next step, the gelsolin is concentrated and glutathione is removed. The gelsolin is loaded onto a 0.7 ml SP-Sepharose column. The column is washed with 5 CN. of 10 mM sodium acetate, 150 mM ΝaCl, 0.5 mM CaCl₂, pH 5.5 (pyrogen free) and the gelsolin eluted with 7 CN. of 10 mM ΝaOAC, 350 mM ΝaCl, 0.5 mM CaCl₂, pH 5.5 (pyrogen free). 0.5 CN. fractions are collected and read at 280 nm. The peak is pooled, and neutralized with 1/20 vol of 0.2 M HEPES pH 7.5. The final yield is approximately 4 mg gelsolin per gram of cell paste. At this stage the product can be used as is or buffer exchanged into a desired formulation using any of a variety of methods common to one skilled in the art. Samples are aliquoted and stored at -

The product is evaluated for purity, extent of oxidization, and activity. These data are summarized in Table 1. By SDS-PAGE the product contains a single band with apparent mass of 85 kDa. There is no evidence of aggregation by SEC on a Superose 6™ column, and the absorbance spectrum from 240-340 nm looks normal with an absorbance maximum at 282 nm and a minimum at 250 nm. The percentage of gelsolin having the correct disulfide bonding was assessed by the protease assay using limited proteoiysis with plasmin to determine whether it looked like the natural plasma or recombinant product. By proteoiysis, gelsolin that has the correct disulfide bond will be cleaved to a 70Kd band and gelsolin that has all free sulfhydryls will digest to a 65Kd band. If only a fraction of the recombinant gelsolin is oxidized the percentage can be estimated by the amount of 70kd band verses the 65Kd band. The extent of oxidation is also assessed by peptide mapping. Both methods reveal that oxidation is complete. The product is tested for function in the rhodamine phalloidin severing assay (Allen and Janmey, J. Biol. Chem. 269:32916-32923, 1994) where we determined it is fully active. This assay measures the fluorescence emitted by rhodamine palloidin. Gelsolin activity is associated with a decrease in rhodamine fluorescence in this assay, due to the dissociation of the phalloidin from actin filaments. As shown in Figure 1, natural gelsolin has greater severing activity than recombinantly produced gelsolin, particularly at lower calcium concentrations. Increased calcium concentration is associated with greater severing activity, up to a maximal level.

Finally, the product is tested for structural integrity by ES-MS and by peptide mapping. For ES-MS the sample is treated with thermolysin which converts the product into two-40 kDa fragments. The masses are consistent with data previously observed for recombinant gelsolin. For peptide mapping the product is digested both with trypsin and with endo Lys-C. Parallel digestions of CHO gelsolin (run as a control) have been analyzed. The peptide maps generated with the E.coli product and CHO gelsolin are indistinguishable.

TABLE 1. STRUCTURE/ FUNCTION DATA FOR NATURAL, E.COLI DERIVED RECOMBINANT, OXIDIZED E.COLI DERIVED RECOMBINANT, AND MAMMALIAN RECOMBINANT GELSOLIN

Property Natural E.coli Oxidized CHO

Structural studies

Proteolytic susceptibility plasmin + EDTA 70K 70K 70K 70K plasmin + Ca 70K 65K 70K 70K

Response in ELISA +EDTA ++ + ++ ++

Disulfide structure 3-SH 5-SH 3-SH 3-SH

N-terminus 30%des1-5 intact intact intact

C-terminus (C-1 ) 35% <5% 5% 15%

SEC <100K <100K <100K <100K

CD similar structures n d n d fluorescence spectra similar structures n d n d

DSC similar structures n d n d

Functional studies

Severing activity ++ ++ ++ ++

Ca dependence of sevenng ++ + ++ ++ Published microinjection study-Stress fiber severing n d n d

Nucleation activity +/ - n d n d

Mucolytic, visco-elasticity, adhesiveness assays C. Production of Oxidized Gelsolin by Refolding of Denatured Gelsolin

Fully oxidized gelsolin may also be produced by denaturing reduced gelsolin and refolding it. In this procedure, the gelsolin can be obtained from any recombinant source, and is denatured with any standard process. The following is a list of compounds which either denature, or result in local perturbations due to disruption of hydrogen bonds, ionic interactions or hydrophobic interactions. Also given are standard concentrations of use, which can vary widely. These include denaturants such as urea (8M) and guanidine (6M); solubilizing acids such as formic acid (70%) or acetic acid (10%); chaotropes such as thiocyanates (2-3 M), magnesium chloride (up to 5 M) or lithium chloride (up to 6 M); detergents such as octaglucoside (2%), cholic acid (2%) or Mega 8™ (2%, CalBiochem); or any other compound that perturbs the secondary structure of the gelsolin molecule. When the denaturing compound is removed from the gelsolin, as by dialyzing or dilutions, the gelsolin forms a disulfide bond characteristic of mature plasma gelsolin.

In one example of this process, gelsolin is first denatured by a 15 minute incubation at room temperature in a solution of 1 ml 3.6 mg/ml gelsolin, 0.666g urea (ultra pure, BRL, enzyme grade), and 7 μl 1 M DTT. The solution is diluted by adding 1.87 ml 100 mg/ml bovine serum albumin (Sigma, A-9647), 0.375 ml 0.2 N sodium phosphate pH 8.0 and 1.5 M NaCl, and 0.5 ml water, and incubating at room temperature overnight. The solution is dialyzed for 4 hours against 2 liters of 25 mM Tris-HCl pH 8.0 with 50 mM NaCl. The buffer is then changed and the dialysis continued overnight. Finally, the gelsolin is purified on a Q-Sepharose column. A 6.2 ml column is packed, equilibrated with buffer B (25 mM Tris-HCl pH 8.0, 50 mM NaCl, 1 mM EGTA), the sample loaded, the column washed with 2X bed volumes of buffer B, and the gelsolin eluted with buffer C (25 mM Tris-HCl pH 8.0, 75 mM NaCl, 2 mM Ca(Cl)₂.

D. Characterization of Differences between Cytoplasmic and Plasma Gelsolin

The following abbreviations are used in the description of our gelsolin production method: Endo Asp-N, endoprotease Asp-N; HPLC, high performance liquid chromatography; Endo Lys- C, endoprotease Lys-C; PE-Cys, 2-(4-pyridyl)ethylcysteine; MES, 4-moφholineethanesulfonic acid; and PAGE, polyacrylamide gel electrophoresis. Purification of human plasma gelsolin.

Gelsolin is purified from human plasma using a modified version of the published method for bovine gelsolin (Kurokawa et al., Biochem. Biopohys. Res. Commun. 169:451-457, 1990). The gelsolin is precipitated from plasma using a 35-50% ammonium sulfate cut, dialyzed and purified by DE52 anion exchange chromatography as described. Gelsolin containing fractions are identified by SDS-PAGE and pooled. The product is further purified by cation-exchange chromatography on SP-Sepharose (Pharmacia). The pH of the eluate pool is lowered to 6.0 with MES and loaded onto a SP-Sepharose column that is equilibrated with 50 mM MES, pH 6.0. The column is washed with 50 mM MES, pH 6.0 and the gelsolin is eluted with 25 mM Tris HCl, pH 7.5, 180 mM NaCl. Typically about 1 mg of gelsolin is obtained from 50 mL of human plasma. The gelsolin is aliquoted and stored at -70 °C.

Purification of human cytoplasmic gelsolin.

Cytoplasmic gelsolin for disulfide analysis is purified from fresh platelets obtained from the American Red Cross within five days of isolation. Gelsolin-actin complexes are purified from the platelets by DNase I affinity chromatography. The platelets are washed as described by Kurth and Bryan (J. Biol. Chem. 259:7473-7479, 1984), lysed by sonication, and then applied to a DNase I affinity column following the method developed by Wang and Bryan (Cell 25:637- 649, 1981). The eluate is subjected to reversed phase HPLC on a Vydac C column. The column is developed with a 30 min 0-70% gradient of acetonitrile in 0.1 % trifluoroacetic acid at a flow rate of 1.4 mL/min. The column effluent is monitored at 280 nm and 0.5-min fractions are collected. Gelsolin containing fractions of >90% purity are identified by SDS-PAGE.

Purification of recombinant human plasma gelsolin from E.coli in its Fully Reduced State.

The plasma form of gelsolin is expressed in E. coli behind the P_L promotor. Cells in 50 mM Tris HCl, pH 8.0, 30 mM NaCl, 1 mM EDTA (1 part wet cell weight to 9 parts buffer) are lysed in a Manton gaulin and subjected to centrifugation for 1 h at 14000 g. The supernatant is loaded onto a Q Sepharose column equilibrated in the same buffer. The column is washed with 4 column volumes of the load buffer then with 3 column volumes of the same buffer containing 100 mM NaCl. Gelsolin is eluted with 25 mM Tris HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA. The gelsolin is diluted 1: 1 with 25 mM Tris HCl, pH 8.0, 1 mM EDTA, and loaded onto a second Q Sepharose column. The column is subjected to the same wash strategy except the second wash contained only 30 mM NaCl. The gelsolin is eluted with 25 mM Tris HCl, pH 8.0, 75 mM NaCl, 2 mM CaCl₂ The gelsolin is further purified on a SP-Sepharose column. The column is washed with 50 mM MES, pH 6.0, and then the gelsolin eluted with 15 mM sodium phosphate, pH 7.5, 300 mM NaCl. All columns are routinely loaded at 10-20 mg total protein per mL resin with an overall recovery of approximately 1 g of gelsolin per 1 kg of cells (wet weight).

In one method, E.coli derived recombinant gelsolin in 25 mM Tris HCl, pH 8.0, 2 mM CaCl , 50 mM NaCl is incubated overnight at 4°C in the presence or absence of 2 mM oxidized glutathione. The samples are treated with 4-vinyl pyridine, and subjected to peptide mapping both with endo Asp-N and endo Lys-C.

Purification of recombinant human plasma gelsolin from Cos 7 cells.

The human plasma gelsolin gene in a CDM8 expression plasmid (Kwiatkowski & Yin, Cell Motil. Cytoskel. 14:21-25, 1989) is transfected into Cos 7 cells by electroporation. After 24 h, the cells are transferred to serum free media. After an additional 72 h, the conditioned medium is filtered (0.2 μm), concentrated by ultrafiltration in an Amicon stirred cell, and dialyzed overnight against 25 mM Tris HCl, pH 8.0, 45 mM NaCl, 1 mM EDTA. The dialyzed material is subjected to the same DE52/ SP-Sepharose purification protocol described above for natural plasma gelsolin. Four mg of gelsolin is purified from 2 liters of cell culture media.

Alkylation of Gelsolin.

Samples containing about 60 μg of gelsolin in 0.1 mL of 6 M guanidine HCl in the presence or absence of 50 mM dithiothreitol are treated with 500 mM 4-vinylpyridine for 2-3 h at room temperature. Alkylated gelsolin is recovered by precipitation with 40 volumes of cooled ethanol (Pepinsky, Anal. Biochem. 195:177-191, 1991). The solution is stored at -20 °C for 1 h and then centrifuged at 14,000 x g for 8 min at 4 °C. The supernatant is discarded and the protein is stored at -20 °C.

Peptide Mapping. Alkylated gelsolin (0.5 mg/mL in 20 mM Tris HCl, pH 7.5) is digested with endo Asp-N (Calbiochem) at an enzyme to gelsolin ratio of 1 :60 (w/w) or with endo Lys-C (Wako Pure Chemical Industries, Ltd.) at a 1 :25 ratio. Digests are conducted at room temperature for 8-h with endo Asp-N or for 24-h with endo Lys-C. The reactions are stopped by acidification with 10 μL of 25% trifluoroacetic acid and analyzed by reversed phase HPLC on a 2.1 mm x 25 cm Vydac C|₈ column. The column is developed with a 180-min gradient (0-90% acetonitrile) in 0.1 % trifluoroacetic acid at a flow rate of 0.2 mL/min. The absorbance of the eluate is monitored using a Waters model 991M photodiode array detector. Individual peaks are manually collected for mass and/or sequence analysis. Prior to injection, solid guanidinium chloride is added into the peptide solutions to a concentration of 6 M to dissolve insoluble material.

Titration of Thiol Groups.

Thiol groups of native and denatured (in 6M guanidine HCl) gelsolin at 1-3 mg/mL are assayed with Ellman's reagent (5,5'-dithiobis(2-nitrobenzoic acid)) by monitoring absorbance at 412 nm (Creighton, in Protein structure: a practical approach (Creighton, T. E., Ed.) pp. 157- 158, IRL Press, Oxford. 1990). For denatured samples, absorbance maxima are reached after 4 min. Moles of free -SH are calculated either against a standard curve of reduced glutathione or from the molar extinction coefficient of thionitrobenzoate (e ι₂=13700/M cm in 6M guanidine HCl and 14150 in its absence).

Mass Determination The molecular masses of peptides are determined by matrix assisted laser desoφtion mass spectroscopy on a Finnigan LaserMat mass spectrometer using α-cyano-4-hydroxycinnamic acid as the matrix. All spectra are calibrated against internal standards.

Limited proteoiysis of gelsolin with plasmin

Human plasmin is obtained from Sigma (Cat. No. P-4895). The protease is suspended at 1 mg/mL in water, aliquoted and stored at -20 °C. Typically 10 μg of gelsolin in 20 μL of 25 mM Tris HCl, pH 8.0, 100 mM NaCl is incubated for 60 min at 37 °C with 0.6 μg of plasmin. Samples are treated with electrophoresis sample buffer and subjected to SDS-PAGE on a 10- 20% gradient gel from Integrated Separation systems. Each test sample is analyzed in the presence of 5 mM EDTA, with 2 mM CaCl or with no addition.

For analysis of cytoplasmic gelsolin, a modified version of the method is developed in which the gelsolin is cleaved with plasmin while immobilized on 2C4-Sepharose. The anti- gelsolin 2C4 monoclonal antibody (Sigma) (Chaponnier et al., J. Cell Biol. 103: 1473-1481 ,

1986) is purified on Protein A Sepharose and conjugated to CNBr Sepharose 4B at 2mg antibody per mL resin. Gelsolin samples (0.5 mL, 40 μg/ml) in 50 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% Tween-20 are incubated with 20 μL of 2C4-Sepharose with continuous agitation at 4^°C for 1 h. The beads are collected by centrifugation, washed with 0.5 mL of 20 mM Tris pH 7.1, 150 mM NaCl, 5 mM MgCl₂, 5 mM ATP, 1 mM EGTA ( 1 h at ambient temperature with agitation), and treated with plasmin (1 h at 37°C with agitation in 0.5 mL of 100 mM Tris HCl pH 7.5, 0.5 mM CaCl₂). Gelsolin is released from the beads by incubation with electrophoresis sample buffer for 10 min at room temperature. Solution controls are run in which the gelsolin is treated with excesses of antibody or actin. For these tests, gelsolin (0.1 mg/mL) in 5 mM triethanolamine pH 7.5, 100 mM KCl, 2mM MgCl₂, 0.2 mM CaCl₂, 0.5 mM ATP, is incubated at ambient temperature for 1 h alone or in the presence of actin or 2C4 antibody and then subjected to plasmin digestion. Rabbit skeletal muscle actin is polymerized to F-actin immediately prior to use.

RESULTS

Determination of the Cys oxidation state of human plasma gelsolin.

Human plasma gelsolin is purified from natural and recombinant sources using a modification of the calcium dependent elution protocol published by Kurokawa et al. (Biochem. Biophys. Res. Commun. 168:451-457, 1990) for bovine plasma gelsolin. The final product is >95% pure by SDS-PAGE and contained a single major band with apparent mass of 85 kDa (Figure 2, lanes a-d). Free thiol content is determined by reaction with 5,5'-difhio-bis(2- nitrobenzoic acid) (Ellman's reagent) under native and denaturing conditions. The appearance of thionitrobenzoate is monitored at 412 nm. Recombinant gelsolin from E.coli contained 5.0 ± 0.1 moles of SH groups/mole of protein under denaturing conditions while natural and the Cos derived product contained only 3.0 ± 0.2 moles of SH group/mole protein. Since a theoretical value of 5 would be expected if all the Cys residues are free, this result indicated the presence of a disulfide in the natural product. The same samples contained less than 0.1 moles SH/mole protein in the native, undenatured state.

Since many groups have worked with recombinant human plasma gelsolin and the methods for expression and purification are similar to what we employed (Way et al., J. Cell Biol. 109:593-605, 1989), it is unlikely that the product would be inactive for actin severing activity. Indeed the recombinant and natural preparations of plasma gelsolin displayed similar levels of actin severing assay when assayed with 0.3 mM calcium present, reflecting concentrations which are routinely used to evaluate activity. Interestingly, when the products are tested side by side at low calcium over a range of calcium concentrations, the natural plasma gelsolin is more active than the recombinant, suggesting that differences in the disulfide structure impact function (Figure 1 ). We have biochemically defined the structural differences between the natural and recombinant products and have extended these analysis to evaluation of cytoplasmic gelsolin.

Analysis of the disulfide structure of plasma gelsolin by peptide mapping.

The plasma gelsolin sequence contains Cys residues at positions 93, 188, 201, 304, and 645. A peptide mapping strategy is used to assess the disulfide structure of plasma gelsolin. Recombinant gelsolin from E.coli is pyridylethylated with 4-vinyl pyridine, and subjected to peptide mapping with endo Asp-N. Digests are analyzed by reversed phase HPLC on a C|₈ column. Endo Asp-N is selected as the cleavage enzyme because it is expected to generate peptides that contain only one Cys residue. Figure 3 (left panel) shows the peptide map from an 8 h digest at 214, 254, and 280 nm. A time course study of the digestion showed no major changes in the HPLC profile between 2 and 16 h (data not shown). The gelsolin sequence contains 46 Asp residues and, after optimization of the gradient, around 40 major peaks could be detected from the digest.

Pyridylethyl-Cys (PE-Cys) has a maximum absoφtion at 254 nm, which can be used to identify PE-Cys containing peptides directly from spectral data. Endo Asp-N digests of reduced and alkylated gelsolin should contain five Cys-containing peptides; however, since two of them have identical sequences, only four peaks are expected. Indeed four PE-Cys peptide peaks are identified (peaks A-D, Figure 3). Two of these (peaks A and B) also have absorbance maxima at 280 nm, indicating the presence of aromatic residues. PE-Cys-containing peptides are further characterized by mass spectrometry. The 254-nm absorbance for peak C (Figure 3) is twice that of the other PE-Cys containing peptides (see peak A and D) due to the presence of two copies of the peptide per mole of gelsolin. The height of peak B is variable because of partial cleavage at Glu⁹². The peptide maps generated from samples with or without reduction are essentially identical, verifying that the five Cys residues in the recombinant gelsolin from E.coli are fully reduced (only data for unreduced sample are shown).

When natural plasma gelsolin is subjected to the same analysis, the endo Asp-N peptide map is very similar to the profile seen with recombinant gelsolin. The major differences occurred in the region of the disulfide-containing peptides. The lower panel of Figure 4 shows an expanded region of the map from 65-110 min in order to better highlight the differences. The corresponding region from the map of recombinant gelsolin from E.coli is shown in the upper panel. Peak A is absent from the cleavage profile of natural plasma gelsolin and peak C is reduced in height relative to peaks B and D. The peptide map of natural gelsolin also contains an extra peak (Peak E). The observed molecular mass (MH⁺ = 2589) of Peak E is consistent with that of a disulfide linked peptide which is made up of the missing peak A and peak C components. Such a product would have a calculated MH⁺ of 2589 and an absoφtion maximum at 280 nm due to aromatic residues. N-terminal sequence analysis of peak E confirmed the presence of the two expected peptides.

Since the DCFIL peptide occurs twice in the natural plasma gelsolin sequence (residues 187-191 and 303-307), two possibilities existed for the disulfide in plasma gelsolin; an intradomain 2 linkage between Cys¹⁸⁸ and Cys²⁰¹, or an interdomain linkage between Cys²⁰¹ in domain 2 with Cys ⁴ in domain 3. To distinguish between these possibilities, gelsolin is also subjected to peptide mapping with endo Lys-C in which Cys¹⁸⁸ and Cys³⁰⁴ are located in unique peptides. The gelsolin sequence has 45 potential endo Lys-C cleavage sites and should generate only four Cys-con taining peptides, since Cys¹⁸⁸ and Cys²⁰¹ are in the same endo Lys-C fragment. Figure 5 shows the peptide map for pyridylethylated recombinant gelsolin from E.coli after 24 h of digestion. Shorter digestion times are tested but are inadequate due to incomplete cleavage (data not shown). Five PE-Cys containing peptides are detected (labeled peaks F, Gl, G2, HI and H2 in Figure 5). Two of the peptides (Gl and H2) are partials resulting from incomplete cleavage at Lys residues. The three other peptides are as expected. The large Cys⁹¹ containing- peptide corresponding to residues 73-135 is not detected, presumably due to irreversible binding to the C]₈ column. All of the relevant peaks have been characterized by mass spectrometry.

A comparison of the endo Lys-C peptide maps of unreduced, PE-labeled, natural gelsolin (Figure 6, middle panel) and reduced, PE-labeled, natural plasma gelsolin (Figure 6, upper panel), reveals the loss of peaks HI and H2 and the presence of a new peak (Peak J) in the unreduced product. The absorbance spectrum of Peak J contains a single absorbance maxima at 280 nm indicating the presence of aromatic groups (data not shown). Mass spectral data for peak J (MH⁺ = 5340) agrees with the theoretical mass (MH⁺ = 5338) of the disulfide linked peptide resulting from the intradomain 2 disulfide. The identity of Peak J is further confirmed by N- terminal sequencing. As expected, only one peptide sequence is obtained. The alternative interdomain disulfide structure would have resulted in a complex series of products due to partial proteolytic cleavage products with masses of 7160, 7032, 9117, and 9246 Da, and two peptide sequences would have been seen in N-terminal sequencing. Since the presence of peak HI or J is a direct indication of the oxidation state of gelsolin, we used the peak J/ peak J + HI ratio as a measure of the amount of gelsolin containing the Cys -Cys disulfide, which proved to be a very sensitive assay for evaluating its oxidation state (see below).

Analysis of the disulfide structure of cytoplasmic gelsolin.

The disulfide structure of cytoplasmic gelsolin is also assessed by peptide mapping. Results from this study are shown in the lower panel of Figure 6. Table 2 gives the percentage of the gelsolin molecules in each preparation that has a Cys¹⁸⁸-Cys²⁰¹ disulfide bond. The pattern of PE-Cys-containing peptides resembles that of recombinant gelsolin from E.coli in that all of the expected Cys-containing peptides are detected, indicating that the five Cys residues in cytoplasmic gelsolin exist as free thiols. A small Peak J component, representing about 20% of the product, is observed in the map of cytoplasmic gelsolin; however, we presume that this is generated during the purification by oxidation.

Table 2. Quantitation of peptide mapping data. The relative amounts of Cys -Cys -201 i •n the various preparations of gelsolin indicated below were quantified from endo Lys-C peptide maps. Values indicated reflect amounts of peak J relative to Peak J + Peak HI based on peak height.

Peak J/Peaks J+Hl (%)

Sample

Natural plasma gelsolin >95

Natural plasma gelsolin (reduced) <5

Cytoplasmic gelsolin 24

Cytoplasmic gelsolin (+24 h dialysis) 40

E.cø/.^'-derived gelsolin <5

E. coli-άeήved gelsolin (oxidized) >95

E. cσ/i^'-derived gelsolin (+24 h dialysis) 42

E. co/j-derived gelsolin (refolded) >92

Cos derived gelsolin >92

The relevant sequences from different species of gelsolin, and other actin-severing proteins such as severin, villin and fragmin that are homologous with domain 2 are listed elsewhere (Wen et al., Biochemistry 30:9700-9707, 1996). Only mouse and pig gelsolin have the two Cys residues seen in human gelsolin. Lobster, fruit fly, and frog gelsolin as well as villin, severin, and fragmin are missing one or both of the Cys residues. Thus these related proteins already exist in forms that evolved with and without the disulfide.

Detection of structural differences between natural and recombinant forms of human plasma gelsolin by limited proteoiysis. The differences in the disulfide structure of natural and recombinant forms of plasma gelsolin raised the possibility that more significant structural differences might exist. To test this possibility, samples have been evaluated by CD and limited proteolysis. The CD spectra are very similar indicating that there are no gross changes in structure. In contrast, the samples differed dramatically in their susceptibility to proteoiysis. Many proteases could distinguish between the natural and recombinant products. Figure 2 shows the results from a study with plasmin. Under limiting digestion conditions, the major cleavage product of natural plasma gelsolin is a 70 kDa fragment (Figure 2, lane e) while the major cleavage product for recombinant gelsolin from E.coli is a 65 kDa fragment (lane g).

Since calcium is known to induce a structural. change in gelsolin (Kilhoffer & Gerard, Biochemistry 24:5653-5660, 1985; Reid et al., Arch. Biochem. Biophys. 302:31-36, 1993), we next tested if calcium affected plasmin susceptibility. In the presence of calcium, we obtained the same pattern of results that are obtained without added calcium. Natural plasma gelsolin is cleaved into a 70 kDa fragment and recombinant into a 65 kDa fragment (Figure 2, lanes i and k). In contrast, both forms are converted into a 70 kDa form by plasmin in the presence of EDTA (compare lanes m and o). The difference in proteolytic susceptibility in the presence of calcium or EDTA indicates that the 65 kDa cleavage site is only exposed in the calcium dependent conformation. The calcium dependence is further evaluated by first treating the samples with EDTA and then adding excess calcium and testing them for proteolytic susceptibility. Under these conditions, natural gelsolin generated the 70 kDa fragment and recombinant the 65 kDa fragment (data not shown), supporting this notion. The formation of the 65 kDa adduct in the absence of added calcium presumably reflects the fact that the gelsolin has been treated with calcium at the later stage of purification and therefore is already in the appropriate state.

The 65 and 70 kDa bands are further characterized by N-terminal sequencing. The sequence for the 70 kDa band starts at position His¹⁵¹ of the plasma gelsolin sequence, while the sequence for the 65 kDa band starts at position Ala²²⁹. The spacing between the cleavage sites and the relevant Cys residues spans 78 amino acids, indicating that a substantial segment of domain 2 is affected by the formation of the disulfide. Other proteases that have been tested, including V8 protease, trypsin and endo Lys-C, produce a similarly sized 65 kDa fragment for recombinant gelsolin under limiting digestion conditions, that is not detected with natural plasma gelsolin, supporting the notion that the structural perturbation to domain 2 is substantial. The results with plasmin are particularly striking, since plasmin seems to target a single site albeit different in the two gelsolin preparations.

Similar studies are performed on recombinant human plasma gelsolin that has been produced in Cos cells. The secreted gelsolin is indistinguishable from natural gelsolin. Based on peptide mapping analysis, over 92% of the product contained the Cys^l88-Cys²⁰¹ disulfide. By limited proteoiysis with plasmin, the Cos derived and natural gelsolin generated the same pattern of cleavage products i.e. the 70 kDa band formed under all conditions (see Figure 2).

Platelet-derived cytoplasmic gelsolin is also tested for susceptibility to proteoiysis with plasmin. As shown in lane j of Figure 7, only the 65 kDa cleavage product is observed after treatment with plasmin, indicating that this feature of cytoplasmic gelsolin structure resembles recombinant gelsolin produced in E.coli.

While the cleavage data for cytoplasmic gelsolin clearly reveal the 65 kDa fragment, the analysis is complicated by the presence of actin in the preparation. Preliminary tests reveal that the actin-gelsolin complex is resistent to proteoiysis presumably because actin binding blocked the sites that are susceptible to digestion when gelsolin is in the free state (see Figure 7, lane e). A method for cleavage was developed using 2C4 Sepharose to collect and wash the gelsolin prior to digestion. Immobilization on the 2C4 antibody has no impact on the ability to cleave gelsolin with plasmin (see Figure 7). Even with these modifications, only about 30% of the cytoplasmic gelsolin is cleaved. The three dimensional structures of gelsolin domain 1 and villin domain 1 are highly homologous (Markus et al., Protein Sci. 3:70-81, 1994). By assuming that the tertiary structures of the other gelsolin domains also are homologous, and using domain 1 of villin as a model for their structure, we have found that Cys¹⁸⁸ and Cys²⁰¹ in domain 2 and Cys³⁰⁴ in domain 3 are buried and Cys⁹³ in domain 1 and Cys⁶⁴⁵ in domain 6 are in loop structures. Based on the model, the distance between the α-carbons of Cys¹⁸⁸ and Cys²⁰¹ is 4-5 A, which is close enough to accommodate the disulfide without invoking the need for a large conformational change (Figure 12). The spacial orientation of the plasmin cleavage sites was also investigated. Both the His¹⁵¹ and Ala²²⁹ cleavage sites are far removed from the relevant Cys residues (Figure 12). The fact that the His and Ala cleavage sites are influenced by oxidation despite the relatively large distance between the relevant amino acids indicates that a large segment of domain 2 is affected by the formation of the disulfide (see Figure 12).

Formation of the correct disulfide after treatment with oxidized glutathione.

Since recombinant gelsolin from E.coli lacked the intradomain 2 disulfide, we tested if the disulfide could be induced to form by mild oxidation. Using peptide mapping to monitor disulfide formation, we achieved the desired result. After an 18 h incubation with 2 mM oxidized glutathione, over 95% of the product contained the Cys¹⁸⁸-Cys²⁰¹ disulfide. The other Cys residues are not affected by the oxidation step. The glutathione oxidized gelsolin produced the 70 kDa fragment in the proteoiysis assay with plasmin when run in the presence of calcium (Figure 8), indicating that the absence of the disulfide and the structural differences detected by limited proteoiysis are related. Forty two percent of the disulfide form in parallel samples that have been incubated without the addition of glutathione. While the free thiols are readily susceptible to oxidation, we are unable to reverse the process by reduction. Treatment of natural plasma gelsolin with 0.2 mM DTT have no effect on the oxidation state. We also have examined the effects of several critical variables on oxidation of gelsolin.

Figures 9, 10 and 11 show the effects of pH, calcium, and oxidized glutathione concentration on the extent of oxidation.

In summary, the disulfide structures of two recombinant versions of human plasma gelsolin that had been expressed intracellularly in E.coli and as a secreted protein from Cos cells have been characterized. The E.coli derived gelsolin lacks the disulfide, while the mammalian product contains the correct disulfide. Disulfide formation in the E.coli product can be induced by mild oxidation. Without added oxidant, disulfide formation occurs at a reduced rate. Calcium is needed for the oxidation to occur (data not shown). The present data indicate that E.coli derived gelsolin product is not a true mimic of natural plasma gelsolin but instead more closely resembles cytoplasmic gelsolin, and can be converted to the plasmin-like conformation through oxidation.

Although the foregoing invention has been described in some detail by way of illustration and example for puφoses of clarity of understanding, it will be obvious to one skilled in the art that certain changes and modifications may be practiced within the scope of the invention, as limited only by the scope of the appended claims.

Claims

What is claimed is:

1. A method of producing recombinant gelsolin or gelsolin fragment containing a disulfide bond, comprising contacting a sample containing reduced gelsolin or reduced gelsolin fragment with an oxidant.

2. The method of claim 1 , wherein the oxidant is oxygen, peroxide, a second disulfide compound, radiation, biological oxidants or nonbiological oxidants.

3. The method of claim 2, wherein the oxidant is oxygen or peroxide, and wherein the sample is further contacted with a catalyst.

4. The method of claim 3, wherein the catalyst is a free metal, a metal ion, a metal complex, or selenium.

5. The method of claim 2, wherein the second disulfide compound is an oxidized form of dithiothreitol, dithioerythritol, cystamine, glutathione, a glutathione derivative, thioredoxin, ferredoxin, cystine, or di-β -hydroxyethyl disulfide.

6. The method of claim 5, wherein the second disulfide compound is oxidized glutathione.

7. The method of claim 2, wherein the biological oxidant is a flavin, a cytochrome, or an ascorbate.

8. The method of claim 2, wherein the nonbiological oxidant is a halogen, a redox dye, a nitrate or nitro compound, an azo compound, or a sulphoxide.

9. The method of claim 1, wherein the sample is contacted with both an oxidant and with calcium.

10. A method of producing a recombinant gelsolin molecule or gelsolin fragment containing a disulfide bond, comprising expressing a gelsolin molecule or gelsolin fragment in a cell transfected with a transgene encoding gelsolin or a gelsolin fragment, disrupting the cell, and inducing the formation of a disulfide bond within the recombinant gelsolin molecule or gelsolin fragment.

1 1. The method of claim 10, wherein the disulfide bond is induced by contacting the gelsolin with a denaturing compound, then removing the denaturing compound under conditions which allow refolding of the gelsolin.

12. A method of producing recombinant gelsolin or a gelsolin fragment containing a disulfide bond, comprising expressing gelsolin or gelsolin fragment in a cell transfected with a gene for gelsolin, and contacting the gelsolin or gelsolin fragment with an oxidant.

13. The method of claim 12, wherein the gelsolin is plasma gelsolin or a fragment of plasma gelsolin.

14. The method of claim 12, wherein the gelsolin is human gelsolin or a fragment of human gelsolin.

15. The method of claim 12, wherein the gelsolin or gelsolin fragment is contacted with an oxidant before the gelsolin or gelsolin fragment is purified.

16. The method of claim 12, wherein the gelsolin or gelsolin fragment is contacted with an oxidant after the gelsolin or gelsolin fragment is purified.

17. The method of claim 12, wherein the cell is a bacterial cell.

18. The method of claim 17, wherein the bacterial cell is E. coli, Bacilis subtilis, or Streptomyces.

19. The method of claim 12, wherein the cell is a yeast cell.

20. The method of claim 19, wherein the yeast cell is Saccharomyces cerevisiae, Hansenula polymorpha, Pichia pastoris, Kluyveromyces lactis, Schisosaccharomyces pombe,

Schwanniomyces occidentalis, or Yarrowia lipolytica.

21. The method of claim 12, wherein the cell is a mammalian cell.

22. The method of claim 21, wherein the mammalian cell is a myeloma, CHO, COS-7, NS/0, BHK, mouse Sertoli, monkey kidney, HELA, MDCK, BRL, W138, Hep G2, MMT, HTC, TR- 1 , or EBNA 293 cell.

23. The method of claim 12, wherein the cell is an insect cell, plant cell or fungal cell.

24. The method of claim 23, wherein the fungal cell is Aspergillus or Dictyostelium.

25. Recombinant gelsolin containing a disulfide bond, produced by lysis of a transgenic cell containing the gelsolin, followed by treatment of the gelsolin with an oxidant.

26. The gelsolin of claim 25, wherein the gelsolin is plasma gelsolin.

27. The gelsolin of claim 25, wherein the gelsolin is human gelsolin.

28. A pharmaceutical composition containing recombinant gelsolin or gelsolin fragment containing a disulfide bond, said gelsolin or gelsolin fragment produced by contacting a sample containing reduced gelsolin or reduced gelsolin fragment with an oxidant.

29. A method of producing a recombinant gelsolin fragment containing a disulfide bond, comprising contacting a gelsolin molecule or a portion of a gelsolin molecule with a proteolytic enzyme, and then contacting the gelsolin fragment with an oxidant.

30. The method of claim 29, wherein the enzyme is chymotrypsin, thermolysis, or subtilisin.

31. A method of producing a recombinant gelsolin fragment containing a disulfide bond, comprising contacting a gelsolin molecule or a portion of a gelsolin molecule with an oxidant, and then contacting the gelsolin molecule or the portion of a gelsolin molecule with a proteolytic enzyme.

32. The method of claim 29, wherein the enzyme is chymotrypsin, thermolysis, or subtilisin.