Evaluating secretion and surface attachment of SapA, an S-layer-associated metalloprotease of Caulobacter crescentus

Caulobacter crescentus is used to display foreign peptides at high density as insertions into the surface (S)-layer protein (RsaA). Many recombinant RsaA proteins, however, are cleaved by SapA, a 71-kDa metalloprotease, suggesting a role in maintaining S-layer integrity. When overexpressed on a multicopy plasmid SapA was detected on the surface by fluorescent antibody only if RsaA and the O-side chain of LPS that mediates S-layer attachment were removed by mutation, indicating an outer membrane location beneath the S-layer. Secretion was mediated by the RsaA type 1 transporter since secretion was eliminated in transporter deficient strains or by C-terminal deletions in SapA (the presumed location of type 1 secretion signals). Secretion was required to become an active protease; mass spectrometry suggested this might be due to N-terminal processing during secretion, a feature shared with other type 1-secreted proteases. Overexpression leads to additional processing C-terminal to the protease domain, producing a 45-kDa protein. This was demonstrated to be self-processing. Deletion analysis revealed the C-terminal 100 amino acids were sufficient for anchoring and secretion. When protein G was fused to the last 238 amino acids of SapA it was secreted, surface attached and bound immunoglobulin, indicating potential for foreign protein display.     

The characterization of the first anti-mouse Muc6 antibody shows an increased expression of the mucin in pancreatic tissue of Cftr-knockout mice

Gel-forming mucins are large high-molecular weight secreted O-glycoproteins responsible for the gel-properties of the mucus blanket. Five orthologous gel-forming mucins have been cloned in human and mouse. Among them, the mucin MUC6 has been less studied, particularly in rodents and no anti rodent-Muc6 antibody has been reported yet. In order to further study Muc6 in mice, our aims were to obtain a specific Muc6 antibody, to validate it and to test it in Cftr deficient mice. A polyclonal serum named CP4 was isolated from a rabbit immunized by a mouse Muc6 peptide. In Western blot experiments, the antibody detected a high-molecular weight molecule secreted by the gastric tissue. Using immunohistochemistry, we showed that the antibody reacted strongly with deep glands of duodenum and ileum and mucous neck cells of gastric body. CP4 also recognized Muc6 protein secreted at the surface of the stomach and renal collecting tubules. The centroacinar cells of pancreatic tissue also reacted with the antibody. Cftr−/− mice showed a higher expression of Muc6 at both protein and RNA levels compared with their control Cftr+/+ littermates suggesting that as in the human disease, Muc6 may contribute to the formation of materials that block pancreatic acini and ducts in mouse models of cystic fibrosis. The rabbit anti-mouse Muc6 polyclonal antibody seems highly specific to the mouse mucin and will be useful to study pancreatic pathology in cystic fibrosis.     

High-level expression of human glycosyltransferases in insect cells as biochemically active form

cDNAs, encoding human beta1,4-galactosyltransferase (hGalT I, EC 2.4.1.22), human Galbeta1,3(4)-GlcNAc alpha2,3-sialyltransferase (hST3GalIII, EC 2.4.99), and human Galbeta1,4-GlcNAc alpha2,6-sialyltransferase (hST6Gal I, EC 2.4.99.1), were cloned from human cell lines. In order to express these glycosyltransferases as secreted form in insect cells, cDNAs were inserted into a novel baculovirus transfer vector equipped with the mouse IgM signal peptide and IgG binding domain of the Staphylococcus aureus protein A as an N-terminal fusion partner. About 14 mg hGalT I, 8 mg hST3GalIII, and 6.4 mg hST6Gal I were purified from 1 liter of recombinant baculovirus infected insect cell culture media. The specific activities of recombinant hGalT I and hST6Gal I were determined as 0.65 and 1.6 U/mg protein, respectively. These results indicated that the recombinant hGalT I and hST6Gal I retained enzyme activities at similar level to those of the authentic one although they were fused with the IgG binding domain at the N-terminus. Taken together, the mouse IgM signal peptide and IgG binding domain of the protein A could be efficiently used as an N-terminus fusion partner for the over-expression of heterologous proteins in insect cells.     

Absence of CFTR is associated with pleiotropic effects on mucins in mouse gallbladder epithelial cells

Mucus of cystic fibrosis patients exhibits altered biochemical composition and biophysical behavior, but the causal relationships between altered cystic fibrosis transmembrane conductance regulator (CFTR) function and the abnormal mucus seen in various organ systems remain unclear. We used cultured gallbladder epithelial cells (GBEC) from wild-type and Cftr((-/-)) mice to investigate mucin gene and protein expression, kinetics of postexocytotic mucous granule content expansion, and biochemical and ionic compositions of secreted mucins. Muc1, Muc3, Muc4, Muc5ac, and Muc5b mRNA levels were significantly lower in Cftr((-/-)) GBEC compared with wild-type cells, whereas Muc2 mRNA levels were higher in Cftr((-/-)) cells. Quantitative immunoblotting demonstrated a trend toward lower MUC1, MUC2, MUC3, MUC5AC, and MUC5B mucin levels in Cftr((-/-)) cells compared with cells from wild-type mice. In contrast, the levels of secreted MUC1, MUC3, MUC5B, and MUC6 mucins were significantly higher from Cftr((-/-)) cells; a trend toward higher levels of secreted MUC2 and MUC5AC was also noted from Cftr((-/-)) cells. Cftr((-/-)) cells demonstrated slower postexocytotic mucous granule content expansion. Calcium concentration was significantly elevated in the mucous gel secreted by Cftr((-/-)) cells compared with wild-type cells. Secreted mucins from Cftr((-/-)) cells contained higher sulfate concentrations. Thus absence of CFTR is associated with pleiotropic effects on mucins in murine GBEC.     

Analysis and isolation of human transferrin receptor using the OKT-9 monoclonal antibody covalently crosslinked to magnetic beads.

A method is described for the use of magnetic beads as a solid phase for the immunoprecipitation of labeled proteins. The anti-human transferrin receptor monoclonal antibody OKT-9 has been coupled to sheep anti-mouse IgG1-coated magnetic beads using the crosslinking agent dimethyl pimelimidate. The transferrin receptor is readily detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and autoradiography following immunoprecipitation from 35S-labeled cell lysates. When compared with precipitations using OKT-9 coupled to protein G Sepharose the magnetic beads result in fewer nonspecific bands. The protocol described is generally applicable to the identification of labeled proteins. In addition, because magnetic beads are amenable to covalent crosslinking procedures they can be used for the purification of proteins from complex mixtures. Covalently crosslinked OKT-9 sheep anti-mouse IgG1-coated magnetic beads have been used to affinity purify unlabeled transferrin receptor from cell lysates giving comparable purity and yield to transferrin Sepharose isolated transferrin receptor. The major advantages offered by magnetic beads compared to conventional affinity matrices are low nonspecific binding and the rapidity with which the purification can be performed.     

Streptococcal protein G-gold complex: comparison with staphylococcal protein A-gold complex for spot blotting and immunolabeling

Protein G, a cell wall protein isolated from human group G streptococci strain G148, binds in a similar manner as protein A from Staphylococcus aureus to the Fc portion of IgG molecules. Indeed, protein G has been proposed as a superior Fc binding protein due to its broader species reactivity. Thus, we have prepared a complex of protein G with particles of colloidal gold and determined its applicability for spot-blot analysis and postembedding immunolabeling by comparing it with protein A-gold complex. By spot-blot analysis no difference in binding of protein G-gold or protein A-gold to IgG molecules from a whole spectrum of animal species was observed. Moreover, using rabbit, sheep, or goat anti-rat albumin antibodies to detect nitrocellulose-immobilized rat albumin or antigenic sites in paraffin and Lowicryl K4M thin sections from rat liver, no difference was found with protein G-gold or protein A-gold. Similarly, no difference in binding to protein G-gold or protein A-gold was observed with a battery of monoclonal antibodies. However, in contrast to expectations, protein A-gold reacted well with both sheep and goat IgG molecules; indeed, for the light and electron microscopic localization of albumin with sheep or goat antibodies it was as efficient as protein G-gold. These results demonstrate, therefore, that both protein G-gold and protein A-gold are useful second step reagents for immunolabeling and that protein G-gold was not a superior probe in the systems tested.     

Protein G: a powerful tool for binding and detection of monoclonal and polyclonal antibodies

Protein G is an immunoglobulin (IgG)-binding bacterial cell wall protein recently isolated from group G streptococci. We have investigated the avidity of protein G for various monoclonal and polyclonal Ig of the IgG class, and compared it with the binding properties of protein A, the staphylococcal Fc-binding protein. Radiolabeled Ig were mixed with Sepharose-coupled protein G or protein A, and the amounts of radioactivity bound to the matrix-coupled bacterial proteins were determined. The avidity was found to be greater for protein G than for protein A for all examined Ig. Protein G bound all tested monoclonal IgG from mouse IgG1, IgG2a, and IgG3, and rat IgG2a, IgG2b, and IgG2c. In addition, polyclonal IgG from man, cow, rabbit, goat, rat, and mouse bound to protein G, whereas chicken IgG did not. The binding property of protein G was additionally exploited in the Western blot assay, in which iodine-labeled protein G was used successfully for the detection of a rat monoclonal antibody against ovalbumin, and for the detection of rabbit and goat polyclonal whole antisera against human urinary proteins. In these experimental situations, protein G was found to be a powerful reagent for the detection of IgG, and consequently the antigen against which these antibodies are directed.     

Cell-surface display of a Pseudomonas aeruginosa strain K pilin peptide within the paracrystalline S-layer of Caulobacter crescentus

The paracrystalline surface (S)-layer of Caulobacter crescentus is composed of a single secreted protein (RsaA) that interlocks in a hexagonal pattern to completely envelop the bacterium. Using a genetic approach, we inserted a 12 amino acid peptide from Pseudomonas aeruginosa strain K pilin at numerous semirandom positions in RsaA. We then used an immunological screen to identify those sites that presented the inserted pilin peptide on the C. crescentus cell surface as a part of the S-layer. Eleven such sites (widely separated in the primary sequence) were identified, demonstrating for the first time that S-layers can be readily exploited as carrier proteins to display ‘epitope-size’ heterologous peptides on bacterial cell surfaces. Whereas intact RsaA molecules carrying a pilin peptide could always be found on the surface of C. crescentus regardless of the particular insertion site, introduction of the pilin peptide at 9 of the 11 sites resulted in some proteolytic cleavage of RsaA. Two types of proteolytic phenomena were observed. The first was characterized by a single cleavage within the pilin peptide insert with both fragments of the S-layer protein remaining anchored to the outer membrane. The other proteolytic phenomenon was characterized by cleavage of the S-layer protein at a point distant from the site of the pilin peptide insertion. This cleavage always occurred at the same location in RsaA regardless of the particular insertion site, yielding a surface-anchored 26 kDa proteolytic fragment bearing the RsaA N-terminus; the C-terminal cleavage product carrying the pilin peptide was released into the growth medium. When the results of this work were combined with the results of a previous study, the RsaA primary sequence could be divided into three regions with respect to the location of a peptide insertion and its effect on S-layer biogenesis: (i) insertions in the extreme N-terminus of RsaA either produce no apparent effect on S-layer biogenesis or disrupt surface-anchoring of the protein; (ii) insertions in the extreme C-terminus either produce no apparent effect on S-layer biogenesis or disrupt protein secretion; and (iii) insertions more centrally located in the protein either have no apparent effect on S-layer biogenesis or result in proteolytic cleavage of RsaA. These data are discussed in relation to our previous assignment of the RsaA N- and C-terminus as regions that are important for surface anchoring and secretion respectively.     

Protein AG-gold complex: an alternative probe in immunocytochemistry.

The purpose of this study was to evaluate the use of protein AG tagged with colloidal gold as a reliable immunocytochemical reagent. Protein AG is a recombinant of 47.3 KD molecular weight and pI = 4.3, which displays immunoglobulin Fc binding sites for both staphylococcal protein A and streptococcal protein G. It adsorbs to 10-nm colloidal gold particles with a lower affinity than does protein A, and is saturable. A maximal number of 12 protein AG molecules could be accommodated on the gold particle surface. Protein AG-gold conjugates yielded positive signals in post-embedding immunocytochemical assays when used as a secondary reagent in conjunction with several species and classes of polyclonal (rabbit, goat, sheep, guinea pig) and mouse monoclonal immunoglobulins (IgG1, IgG2, and IgG3). In addition, protein AG-gold was found to be a useful reagent in immunoblot analysis because of its ability to bind and identify nitrocellulose-immobilized IgGs (rabbit, mouse, goat, sheep, rat, and cow). Its spectrum of specificity towards various types of antibodies combines those of the parental protein A and protein G molecules. The protein AG-gold complex therefore appears to be a highly versatile and convenient alternative probe for immunochemical and immunocytochemical studies.     

Cloning of the gene and immunochemical specificity of recombinant immunoglobulin binding protein V7 from Streptococcus valente (group G)]

The fragment of the structural gene coding for the Fc-receptor of Streptococcus Valente (G group) has been cloned. The resulting recombinant plasmid pPGSV1 contains the O, kb HindIII fragment of streptococcal chromosomal DNA inserted into the vector plasmid pUC19 and determines the expression of the 31 kD protein in Escherichia coli cells. The protein binds the immunoglobulins of human, rabbit, guinea pig origin, but in contrast to the G protein of another G group streptococcus it is nonreactive with mouse, pig and sheep IgG.     

The C-terminus of the transmembrane mucin MUC17 binds to the scaffold protein PDZK1 that stably localizes it to the enterocyte apical membrane in the small intestine

The membrane-bound mucins have a heavily O-glycosylated extracellular domain, a single-pass membrane domain and a short cytoplasmic tail. Three of the membrane-bound mucins,MUC3, MUC12 and MUC17, are clustered on chromosome 7 and found in the gastrointestinal tract. These mucins have C-terminal sequences typical of PDZ-domain-binding proteins. To identify PDZ proteins that are able to interact with the mucins,we screened PDZ domain arrays using YFP (yellow fluorescent protein)-tagged proteins. MUC17 exhibited a strong binding to PDZK1 (PDZ domain containing 1), whereas the binding toNHERF1 (Na+/H+-exchanger regulatory factor 1) was weak.Furthermore, we showed weak binding of MUC12 to PDZK1, NHERF1 and NHERF2. GST (glutathione transferase) pull-down experiments confirmed that the C-terminal tail of MUC17 coprecipitates with the scaffold protein PDZK1 as identified byMS. This was mediated through the C-terminal PDZ-interaction site in MUC17, which was capable of binding to three of the four PDZ domains in PDZK1. Immunostaining of wild-type or Pdzk1-/- mouse jejunum with an antiserum against Muc3(17),the mouse orthologue of human MUC17, revealed strong brushborder membrane staining in the wild-type mice compared with an intracellular Muc3(17) staining in the Pdzk1-/- mice. This suggests that Pdzk1 plays a specific role in stabilizing Muc3(17)in the apical membrane of small intestinal enterocytes.     

Immunoselection of stably transformed MDCK cells expressing the vesicular stomatitis virus G-protein at the basolateral surface

We have used immunoisolation on a magnetic solid support for the positive selection of stable MDCK transformants which express VSV-G protein via genomic integration of a cloned cDNA. This method is a simple, inexpensive alternative to selection with a fluorescence-activated cell sorter. The G-protein is synthesized in the absence of other viral proteins and is transported to the plasma membrane. The G-positive cells were enriched by immunoselection during normal passage of the transformed population. Using sterile conditions, antibodies to G were incubated with a suspension of transformed cells at 4 degrees C, unbound antibodies were then removed, and the cells were incubated with the immunoabsorbent (3 micron magnetic beads; J. Ugelstad et al. (1983) Nature (London) 303, 95) containing bound IgG molecules against the Fc portion of rabbit IgG. The magnetic properties of the beads were used to retrieve and further wash the immunoselected population. The cells are then removed from the beads by the same trypsinization conditions used for routine passaging and returned to culture. Using this selection scheme we have been able to increase the number of G-expressing cells five- to sevenfold per round; with repeated rounds enrichment from 2 to 74% was obtained. When grown on filters the immunoselected cells were shown to have the same morphology and electrical resistance (150-200 ohm.cm2) as untransformed MDCK II cells. Indirect immunofluorescence staining and [125I]protein A binding assays carried out on these cells demonstrated that G protein was localized exclusively to the basolateral surface as is observed with viral infection.     

Mechanisms and kinetics of monoclonal antibody synthesis and secretion in synchronous and asynchronous hybridoma cell cultures.

The kinetics of monoclonal antibody synthesis and secretion have been studied in synchronous and asynchronous mouse hybridoma cell cultures. Pulse-labelling of IgG followed by immunoprecipitation and quantitation of synthesized and secreted IgG in synchronous cultures show maximum production during G1/S phases. Secretion takes place through exocytotic release of vesicle contents. Pulse-chase experiments show that 71% of the synthesized IgG is secreted within 8 h of the pulsing period and only a further 4% is secreted by 22 h. Higher specific antibody production (QA) is obtained if (a) cells are arrested and then maintained in G1/S phases, (b) viability is decreased during the death phase of batch culture, (c) the dilution rate is decreased in continuous culture or (d) cells are subjected to hydrodynamically induced stress. The increase in QA in all these cases is mainly due to the passive release of the accumulated intracellular antibody. DNA and protein synthetic activity peak during the early exponential phase and decline rapidly during mid and late exponential and death phases. Metabolic activity however peaks up to 20 h after the peak in DNA synthesis, and declines similarly during the death phase. The data are consistent with the idea that slow growth and higher death rates increase QA and that Ig secretion is probably subject to complex intracellular control.     

Stable expression of a secreted form of the mouse IFN-gamma receptor by rat cells.

The biologic effects of IFN-gamma are mediated through a receptor that is expressed in relatively low abundance on normal mammalian cells. As a consequence, investigations of the physicochemical and ligand-binding properties of the purified receptor have been limited. The work reported here characterizes a secreted form of the receptor for mouse IFN-gamma, made by deletion of the nucleotides that code for the anchoring domain from a cDNA that encodes the receptor binding protein and its related signal peptide. When transfected into rat XC cells, this construct produced up to approximately 1 mg/liter of a secreted protein that had the characteristics of the native receptor. Both the secreted protein and its mRNA were of sizes that were consistent with loss of the transmembrane region. The protein was detectable by a mAb that is specific for an epitope that is found in the ligand binding site of the receptor for mouse IFN-gamma, as well as by a goat polyclonal IgG that is monospecific for the mouse IFN-gamma R. Supernates that contained the secreted protein blocked binding of IFN-gamma to mouse IFN-gamma R and inhibited in a dose-dependent manner the IFN-gamma-mediated priming of mouse bone marrow culture-derived macrophages for tumor cell killing. Availability of relatively large amounts of a secreted protein that retains ligand-binding activity should facilitate purification and basic studies of the receptor binding protein and could provide new approaches to the treatment/prevention of diseases that arise due to inappropriate response of cells to IFN-gamma. In addition, because this secreted receptor, unlike others, consists of both the extracellular and intracellular domains, it is likely that it will be useful in determining how the cytoplasmic portion of the receptor is involved in receptor function.     

Synthesis and secretion of IgG in synchronized mouse myeloma cells

A mutant of the MPC-11 mouse myeloma cell line which grows as a monolayer has been used to study the synthesis and secretion of IgG in relation to the cell cycle. The mitotic detachment method has been used to obtain a pure population of mitotic cells which were then allowed to progress through the G1, S, and G2 phases of the cell cycle. The synthesis and the rate of secretion of IgG have been studied in each phase of the cycle by incubation of cells with ¹⁴C-amino acids, followed by immunoprecipitation and quantitation of synthesized and secreted IgG_2b. The data are consistent with the idea that synthesis and secretion of Ig are not a cell cycle dependent event in myeloma cells.     

金黄色葡萄球菌重组GapC蛋白的GAPDH活性及免疫原性分析

为研究金黄色葡萄球菌(Staphylococcus aureus)表面GapC蛋白的GAPDH活性、免疫原性及免疫保护作用, 应用PCR方法扩增出S. aureus的gapC基因, 插入到pQE-30载体相应位点, 构建重组质粒pQE/gapC。将其导入宿主菌E.coli M15(pREP4)后, IPTG诱导表达。重组蛋白纯化后进行GAPDH活性检测, 并与灭活全菌体分别免疫健康家兔。然后, 应用ELISA方法检测血清中IgG抗体水平及IFN-g、IL-4细胞因子浓度, 并用1.0×108CFU/mL S. aureus菌株Wood46对免疫家兔攻毒。SDS-PAGE结果显示, GapC蛋白在E. coli M15(pREP4)中获得表达; 经GAPDH活性检测及Western Blot检测, 重组蛋白具有较高的GAPDH活性和抗原特异性; 经ELISA检测, GapC蛋白及全菌体组兔血清中IgG抗体水平迅速升高, 并在加强免疫后第28天达到最高(1:64 000), 加强免疫后第14 d, 血清中细胞因子IFN-g和IL-4浓度与对照组相比, 显著升高(P<0.05), 而全菌体免疫组升高不明显(P>0.05); 攻毒结果为蛋白免疫组家兔获得一定的免疫保护(4/5)。以上结果表明, 表达的重组GapC蛋白具有GAPDH活性、较好的免疫原性及免疫保护力, 可作为深入研究S. aureus基因工程疫苗的良好靶向。     

金黄色葡萄球菌重组GapC蛋白的GAPDH活性及免疫原性分析

为研究金黄色葡萄球菌(Staphylococcus aureus)表面GapC蛋白的GAPDH活性、免疫原性及免疫保护作用, 应用PCR方法扩增出S. aureus的gapC基因, 插入到pQE-30载体相应位点, 构建重组质粒pQE/gapC。将其导入宿主菌E.coli M15(pREP4)后, IPTG诱导表达。重组蛋白纯化后进行GAPDH活性检测, 并与灭活全菌体分别免疫健康家兔。然后, 应用ELISA方法检测血清中IgG抗体水平及IFN-g、IL-4细胞因子浓度, 并用1.0×108CFU/mL S. aureus菌株Wood46对免疫家兔攻毒。SDS-PAGE结果显示, GapC蛋白在E. coli M15(pREP4)中获得表达; 经GAPDH活性检测及Western Blot检测, 重组蛋白具有较高的GAPDH活性和抗原特异性; 经ELISA检测, GapC蛋白及全菌体组兔血清中IgG抗体水平迅速升高, 并在加强免疫后第28天达到最高(1:64 000), 加强免疫后第14 d, 血清中细胞因子IFN-g和IL-4浓度与对照组相比, 显著升高(P<0.05), 而全菌体免疫组升高不明显(P>0.05); 攻毒结果为蛋白免疫组家兔获得一定的免疫保护(4/5)。以上结果表明, 表达的重组GapC蛋白具有GAPDH活性、较好的免疫原性及免疫保护力, 可作为深入研究S. aureus基因工程疫苗的良好靶向。     

A physicochemical study of protein G, a molecule with unique immunoglobulin G-binding properties

Protein G, an IgG-binding molecule, was prepared from the cell walls of a group G streptococcal strain, G-148. The protein could be extracted from the cells by papain digestion and purified by the sequential use of ion-exchange chromatography on DEAE-Sephadex, affinity chromatography on Sepharose-coupled human IgG, and gel chromatography on Sephadex G-200. Two protein bands with similar molecular weight, 34,000 and 36,000, were obtained when analyzing the pure protein G on sodium dodecyl sulfate-polyacrylamide gel electrophoresis. The yield using this purification scheme was 27% of the protein G solubilized from the cells or 70 micrograms/ml packed bacteria. The Stokes radius and frictional ratio of protein G were determined to 3.53 nm and 1.64, respectively, suggesting an elongated fibrous molecule. The protein did not contain any intrachain disulfide bonds. The amino acid composition of protein G was determined and was found to be different from that of protein A, the well known staphylococcal IgG-binding protein. The equilibrium constants of the reactions between protein G and human, rabbit, mouse, and goat polyclonal IgG, determined by Scatchard plots, ranged between 1 X 10(10) and 7 X 10(10), for rat polyclonal IgG 1.4 X 10(9), and human monoclonal IgG1, IgG2, IgG3, and IgG4 between 2 X 10(9) and 6 X 10(9). These affinity constants were always greater than the corresponding values for protein A. The binding between protein G and various polyclonal and monoclonal IgG was pH dependent between 2.8 and 10, strongest at pH 4 and 5, and weakest at pH 10.     

Purification and some properties of streptococcal protein G, a novel IgG-binding reagent

Protein G, a bacterial cell wall protein with affinity for immunoglobulin G (IgG), has been isolated from a human group G streptococcal strain (G148). Bacterial surface proteins were solubilized by enzymatic digestion with papain. Protein G was isolated by sequential use of ion-exchange chromatography on DEAE-cellulose, gel filtration on Sephadex G-100, and affinity chromatography on Sepharose 4B-coupled IgG. The presence of protein G in various pools and fractions during the isolation was followed by their ability to inhibit the binding of radio-labeled IgG to G148 bacteria. A highly purified protein G was obtained. On polyacrylamide gel electrophoresis in sodium dodecyl sulfate, the apparent m.w. was 30,000, and on agarose gel electrophoresis the purified protein gave rise to a single band in the alpha 1-region. Protein G was found to bind all human IgG subclasses and also rabbit, mouse, and goat IgG. On the IgG molecule, the Fc part appears mainly responsible for the interaction with protein G, although a low degree interaction was also recorded for Fab fragments. IgM, IgA, and IgD, however, showed no binding to protein G. This novel IgG-binding reagent promises to be of theoretical and practical interest in immunologic research.