Mapping of multiple B cell epitopes on the 70-kilodalton autoantigen of the U1 ribonucleoprotein complex.

High titer IgG autoantibodies to the 70-kDa polypeptide component (p70) of the U1 ribonucleoprotein (RNP) complex occur in the sera of patients with mixed connective tissue disease, SLE, and related rheumatic diseases. To gain insight into the pathogenesis and diversity of this antibody response we have used recombinant DNA technology to map the linear B cell epitopes on p70. A full length 1.7-kb cDNA clone encoding p70 was isolated from a human placental library and restriction fragments or polymerase chain reaction-generated fragments of the gene subcloned into the bacterial expression vector pGEX. Purified fusion proteins representing specific regions of p70 were immunoblotted with a panel of 70 anti-(U1)RNP+ sera containing anti-p70 antibodies. Six epitopes, four major (A, B, C, and F) and two minor (D and E) were mapped and were located throughout the molecule. The anti-(U1)RNP sera displayed heterogeneity in their pattern of reactivity to the six epitopes although reactivity to epitope C was more frequently associated with SLE rather than mixed connective tissue disease. The identification of multiple B cell epitopes on p70 is consistent with the concept that this self Ag drives the autoantibody response.     

Characterization of immunoreactive epitopes of the HIV-1 p41 envelope protein using fusion proteins synthesized in Escherichia coli

We have identified several immunoreactive epitopes on the human immunodeficiency virus (HIV) type 1 transmembrane envelope protein by synthesizing various regions of the protein as fusions to the trpE gene in Escherichia coli. Ten fusion clones which expressed overlapping peptides were found to contain epitopes reactive with antibodies in sera of North American (NAm) and West African (WAf) patients with acquired immune deficiency syndrome (AIDS). An immunodominant epitope which reacted with all HIV-infected patients' sera was mapped to a 51-amino acid sequence in the N terminus of p41. A novel epitope was also identified in the C terminus of p41 which was reactive with 41% and 48% of the sera tested from NAm and WAf, respectively. In addition, several minor epitopes were identified. We observed that sera from WAf reacted more strongly to minor HIV-1 p41 epitopes than did sera from similarly infected individuals in NAm. We also report on the detection of antibodies from patients with HIV-2 infection in WAf which cross react with HIV-1 p41 recombinant envelope antigens.     

Primary structure of a human small nuclear ribonucleoprotein polypeptide as deduced by cDNA analysis

Anti-Sm is an antibody specificity often associated with the autoimmune disease systemic lupus erythematosus. The polypeptides Sm-B'/B (estimated molecular mass 27 and 26 kDa, respectively) are primary targets of Sm antibodies. Sm-B'/B are part of the core polypeptides of small ribonucleoprotein particles (snRNP) involved in pre-mRNA splicing. Sm-B'/B share the same amino-terminal sequence as we determined by microsequence analyses of the purified polypeptides. Oligonucleotide probes based on that sequence were used to isolate seven clones from a human lymphoblastoid cDNA library in lambda gt10. The clones contained a single coding region for a protein of approximately 25 kDa. The predicted amino-terminal sequence was identical to that of the isolated Sm-B'/B polypeptides. In vitro translation experiments produced a protein immunoreactive with human polyclonal anti-Sm antibodies. The isolation of only one unique cDNA sequence suggests that Sm-B'/B may be post-translational variants encoded by a single message. The specific structural features which distinguish Sm-B' from Sm-B have yet to be determined. Northern blot analysis confirmed the diverse tissue and species distribution expected for these immunologically conserved polypeptides. The Sm-B'/B primary sequence is rich in proline (20%) and glycine (15%) residues. The prolines are concentrated in the carboxyl-terminal half of the protein and display a repetitive unit that is shared with other snRNP and nucleic acid binding proteins. Analysis of these arrays suggests an eight residue proline-rich consensus sequence with potential as either an RNA binding domain, or as a site of protein/protein interaction.     

Antibodies that are cross-reactive for human immunodeficiency virus type 1 clade a and clade B v3 domains are common in patient sera from Cameroon, but their neutralization activity is usually restricted by epitope masking

Sera from human immunodeficiency virus type 1 (HIV-1)-infected North American patients recognized a fusion protein expressing a V3 loop from a clade B primary isolate virus (JR-CSF) but not from a clade A primary isolate virus (92UG037.8), while most sera from Cameroonian patients recognized both fusion proteins. Competition studies of consensus V3 peptides demonstrated that the majority of the cross-reactive Cameroonian sera contained cross-reactive antibodies that reacted strongly with both V3 sequences. V3-specific antibodies purified from all six cross-reactive sera examined had potent neutralizing activity for virus pseudotyped with envelope proteins (Env) from SF162, a neutralization-sensitive clade B primary isolate. For four of these samples, neutralization of SF162 pseudotypes was blocked by both the clade A and clade B V3 fusion proteins, indicating that this activity was mediated by cross-reactive antibodies. In contrast, the V3-reactive antibodies from only one of these six sera had significant neutralizing activity against viruses pseudotyped with Envs from typically resistant clade B (JR-FL) or clade A (92UG037.8) primary isolates. However, the V3-reactive antibodies from these cross-reactive Cameroonian sera did neutralize virus pseudotyped with chimeric Envs containing the 92UG037.8 or JR-FL V3 sequence in Env backbones that did not express V1/V2 domain masking of V3 epitopes. These data indicated that Cameroonian sera frequently contain cross-clade reactive V3-directed antibodies and indicated that the typical inability of such antibodies to neutralize typical, resistant primary isolate Env pseudotypes was primarily due to indirect masking effects rather than to the absence of the target epitopes.     

Autoantibodies to ribonucleoprotein particles containing U2 small nuclear RNA.

Autoantibodies exclusively precipitating U1 and U2 small nuclear ribonucleoprotein (snRNP) particles [anti-(U1,U2)RNP] were detected in sera from four patients with autoimmune disorders. When tested by immunoblotting, these sera recognized up to four different protein antigens in purified mixtures of U1-U6 RNP particles. With purified antibody fractions eluted from individual antigen bands on nitrocellulose blots, each anti-(U1,U2)RNP serum precipitated U2 RNP by virtue of the recognition of a U2 RNP-specific B" antigen (mol. wt. 28 500). Antibodies to the U2 RNP-specific A' protein (mol. wt. 31 000) were found in only one serum. The B" antigen differs slightly in mol. wt. from the U1-U6 RNA-associated B/B' antigens and can be separated from this doublet by two-dimensional gel electrophoresis, due to its more acidic pI. In immunoprecipitation assays, the purified anti-B" antibody specificity also reacts with U1 RNPs which is due to cross-reactivity of the antibody with the U1 RNA-specific A protein, as demonstrated by immunoblotting using proteins from isolated U1 RNPs as antigenic material. Thus the A antigen not only bears unique antigenic sites for anti-A antibodies contained in anti-(U1)RNP sera, it also shares epitopes with the U2 RNP-specific B" antigen.     

Epitope mapping with a recombinant human 68-kDa (U1) ribonucleoprotein antigen reveals heterogeneous autoantibody profiles in human autoimmune sera

Several cDNA fragments encoding parts of the (U1)RNP specific 68-kDa autoantigen were expressed in Escherichia coli and the fusion proteins were used as substrate for localization of the autoreactive epitopes. We have identified a region of approximately 30 amino acids reacting with more than 90% (16 of 17) of all human anti-p68 sera tested, regions which carry only a few and a region with no autoepitopes. Comparative analysis of epitopes recognized on partially degraded fusion proteins indicated that the anti-p68 autoimmune response is polyclonal. It involves generation of antibodies to several epitopes including one in a region with retroviral gag protein homology speculated to play a role in the initiation of the autoimmune response. Each of the 17 sera tested contained a different set of autoantibody specificities. These data are not consistent with random mutation as a sole mechanism of anti-p68 autoantibody induction and argue for an Ag-driven autoimmune response.     

Mapping of B cell epitopes in an immunodominant antigen of Trypanosoma cruzi using fusions to the Escherichia coli LamB protein

The JL8 protein antigen from Trypanosoma cruzi, a dominant immunogen in man, has been characterized as containing tandem amino acid repeats. Here, we describe the use of the LamB protein of Escherichia coli as a carrier of JL8 derived sequences in order to map the immunodominant B cell epitopes in this antigen. Five different sequences of JL8 were inserted in the LamB protein and the JL8-LamB fusion proteins were tested by ELISA with human chronic chagasic sera. The fusion carrying the sequence AEKQKAAEATKVAE was recognized by most sera. This protein was also capable of inhibiting the binding of human chagasic antibodies to GST-JL8 in competitive ELISA suggesting that it contains an immunodominant B cell epitope of JL8.     

Mapping of B cell epitopes on small nuclear ribonucleoproteins that react with human autoantibodies as well as with experimentally-induced mouse monoclonal antibodies

Small nuclear ribonucleoprotein (snRNP) particles are a class of RNA-containing particles in the nucleus of eukaryotic cells. They consist of uridylate-rich small nuclear RNA complexed with several proteins. snRNP particles U1, U2, U4/U6, and U5 all contain a common protein core consisting of proteins B'/B, D, D', E, F, and G. In addition to this core, U1 snRNP particles contain proteins 70K, A, and C, whereas U2 snRNP particles contain proteins A' and B". Almost any of the small nuclear RNA-associated polypeptides is targeted by autoantibodies in the sera from patients with SLE or related connective tissue diseases. We immunized a genetically non-autoimmune mouse with recombinant human B" protein and obtained three mAb reactive with native U2 snRNP particles. Two of these mAb particles cross-reacted with U1 snRNP, 9A9 and 11A1, via epitopes present on the U2 snRNP B" protein as well as on the U1 snRNP-specific A protein. A third mAb 4g3, reacted exclusively with U2 snRNP via a unique epitope on protein B". Two epitopes mapped at the carboxy-terminal region of the B" protein, whereas binding of the third mAb involved both amino- and carboxy-terminal amino acids of the B" protein. Epitope mapping, employing a DNAse I fragment library of the B" cDNA, revealed that the three mAb-reactive sites were discontinuous. Autoantibodies in sera from patients with SLE and other connective tissue diseases competed for binding with the mAb, implying that the mAb define a major autoantibody-reactive region on protein B".     

Identification of a Domain Containing B-Cell Epitopes in Hepatitis C Virus E2 Glycoprotein by Using Mouse Monoclonal Antibodies

Evidence from clinical and experimental studies of human and chimpanzees suggests that hepatitis C virus (HCV) envelope glycoprotein E2 is a key antigen for developing a vaccine against HCV infection. To identify B-cell epitopes in HCV E2, six murine monoclonal antibodies (MAbs), CET-1 to -6, specific for HCV E2 protein were generated by using recombinant proteins containing E2t (a C-terminally truncated domain of HCV E2 [amino acids 386 to 693] fused to human growth hormone and glycoprotein D). We tested whether HCV-infected sera were able to inhibit the binding of CET MAbs to the former fusion protein. Inhibitory activity was observed in most sera tested, which indicated that CET-1 to -6 were similar to anti-E2 antibodies in human sera with respect to the epitope specificity. The spacial relationship of epitopes on E2 recognized by CET MAbs was determined by surface plasmon resonance analysis and competitive enzyme-linked immunosorbent assay. The data indicated that three overlapping epitopes were recognized by CET-1 to -6. For mapping the epitopes recognized by CET MAbs, we analyzed the reactivities of CET MAbs to six truncated forms and two chimeric forms of recombinant E2 proteins. The data suggest that the epitopes recognized by CET-1 to -6 are located in a small domain of E2 spanning amino acid residues 528 to 546.     

Cloning and expression of antigenic epitopes of the human 68-kDa (U1) ribonucleoprotein antigen in Escherichia coli.

Autoantibodies directed against the 68-kDa (U1) ribonucleoprotein antigen are mainly found in sera of patients with mixed connective tissue disease. The corresponding cDNA was fragmented into four regions coding for the major antigenic epitopes A', B', C' and D'. All the epitopes were subcloned and expressed as fusion proteins with the glutathione S-transferase in Escherichia coli using the novel expression system pGEX that allows very high yields of recombinant proteins after a single-step purification. The sera of patients with the autoimmune disease were analyzed for the expressed recombinant proteins by an immunoblotting technique. All positive sera showed a patient-specific behavior and could be divided into four groups regarding recognition of the four antigenic epitopes of the 68-kDa (U1) ribonucleoprotein antigen. The epitope B' was reactive to all patient sera positively tested and classified as the marker antigenic epitope for the mixed connective tissue disease.     

Engineering of papaya mosaic virus (PapMV) nanoparticles through fusion of the HA11 peptide to several putative surface-exposed sites

Papaya mosaic virus has been shown to be an efficient adjuvant and vaccine platform in the design and improvement of innovative flu vaccines. So far, all fusions based on the PapMV platform have been located at the C-terminus of the PapMV coat protein. Considering that some epitopes might interfere with the self-assembly of PapMV CP when fused at the C-terminus, we evaluated other possible sites of fusion using the influenza HA11 peptide antigen. Two out of the six new fusion sites tested led to the production of recombinant proteins capable of self assembly into PapMV nanoparticles; the two functional sites are located after amino acids 12 and 187. Immunoprecipitation of each of the successful fusions demonstrated that the HA11 epitope was located at the surface of the nanoparticles. The stability and immunogenicity of the PapMV-HA11 nanoparticles were evaluated, and we could show that there is a direct correlation between the stability of the nanoparticles at 37°C (mammalian body temperature) and the ability of the nanoparticles to trigger an efficient immune response directed towards the HA11 epitope. This strong correlation between nanoparticle stability and immunogenicity in animals suggests that the stability of any nanoparticle harbouring the fusion of a new peptide should be an important criterion in the design of a new vaccine.     

Identification of a trpG-related glutamine amide transfer domain in Escherichia coli GMP synthetase

An improved method was developed to align related protein sequences and search for homology. A glutamine amide transfer domain was identified in an NH2-terminal segment of GMP synthetase from Escherichia coli. Amino acid residues 1-198 in GMP synthetase are homologous with the glutamine amide transfer domain in trpG X D-encoded anthranilate synthase component II-anthranilate phosphoribosyltransferase and the related pabA-encoded p-aminobenzoate synthase component II. This result supports a model for gene fusion in which a trpG-related glutamine amide transfer domain was recruited to augment the function of a primitive NH3-dependent GMP synthetase. Sequence analyses emphasize that glutamine amide transfer domains are thus far found only at the NH2 terminus of fused proteins. Two rules are formulated to explain trpG and trpG-related fusions. (i) trpG and trpG-related genes must have translocated immediately up-stream of genes destined for fusion in order to position a glutamine amide transfer domain at the NH2 terminus after fusion. (ii) trpG and trpG-related genes could not translocate adjacent to a regulatory region at the 5' end of an operon. These rules explain known trpG-like fusions and explain why trpG and pabA are not fused to trpE and pabB, respectively. Alignment searches of GMP synthetase with two other enzymes that bind GMP, E. coli amidophosphoribosyltransferase and human hypoxanthine-guanine phosphoribosyltransferase, suggest a structurally homologous segment which may constitute a GMP binding site.     

Translational fusions with fragments of the trpE gene improve the expression of a poorly expressed heterologous gene in Escherichia coli

A series of plasmids expressing fusions between the trpE gene product, anthranilate synthase component I and the major immunogen (VP1) of foot and mouth disease virus were constructed such that increasing amounts of the 3' end of trpE were deleted. Deletions removing up to 70% of trpE had little effect on the quantity of fusion protein expressed, while the number of molecules appeared to increase. Larger deletions led to a steady decrease in both the quantity of fusion protein produced and in the number of molecules. This is consistent with trpE being responsible for the high levels of expression of VP1 by its gene product stabilizing VP1 against proteolytic degradation. Some out-of-frame deletion mutants were also produced. All deletion mutants in one wrong reading frame expressed low levels of two VP1-containing polypeptides. This observation is interpreted as being due to re-initiation of translation at a site inside the VP1 sequence which is activated by local termination of translation.     

Homologous and Hybrid Complexes of Anthranilate Synthase from Bacillus Species

The subunits of anthranilate synthase were separated and partially purified by Sephadex G-100 gel filtration from the following six species of Bacillus: Bacillus subtilis, Bacillus licheniformis, Bacillus alvei, Bacillus coagulans, Bacillus pumilus, and Bacillus mascerans. Our data suggest that the enzyme from B. alvei is unique among these species. First, the anthranilate synthase complexes are readily dissociated during gel filtration in the absence of glutamine into a large component (aminotransferase), subunit E, and a small component subunit X (glutamine-binding protein), whereas a higher salt concentration is required to dissociate the complex from B. alvei. Second, the aminotransferase activity from all six species is stimulated by glycerol and inhibited by tryptophan; however, only the large component from B. alvei is stimulated by 2-mercaptoethanol. Finally, the large component can be titrated with the small component to yield a complex which can utilize glutamine as a substrate (amidotransferase). The homologous complexes have an amidotransferase to aminotransferase ratio of 1.4 to 2.3, but the B. alvei complex has a ratio of 0.9. Except for complexes that involve the large component from B. alvei, hybrid complexes can be formed which have ratios as good as the homologous complexes. These data are consistent with the hypothesis that B. alvei is unique among the bacilli with respect to some enzymes in the aromatic amino acid biosynthetic pathway.     

Expression of cloned cDNA for a major surface antigen of Plasmodium falciparum merozoites

A cDNA library of P. falciparum was constructed. Using size-selected mRNA as a probe several clones were isolated which hybridized to mRNAs larger than 5 kilobases (kb). The cDNA insert of pFC 17, which hybridizes to 5.6-kb mRNA was expressed by fusion to anthranilate synthetase I in a plasmid expression vector. The expressed fusion protein was shown to contain epitopes of a 195-kDa protein which is the precursor to 3 major surface antigens of P. falciparum merozoites.     

Presentation of native epitopes in the V1/V2 and V3 regions of human immunodeficiency virus type 1 gp120 by fusion glycoproteins containing isolated gp120 domains.

The immune response to viral glycoproteins is often directed against conformation- and/or glycosylation-dependent structures; synthetic peptides and bacterially expressed proteins are inadequate probes for the mapping of such epitopes. This report describes a retroviral vector system that presents such native epitopes on chimeric glycoproteins in which protein fragments of interest are fused to the C terminus of the N-terminal domain of the murine leukemia virus surface protein, gp70. The system was used to express two disulfide-bonded domains from gp120, the surface protein of human immunodeficiency virus type 1 (HIV-1), that include potent neutralization epitopes. The resulting fusion glycoproteins were synthesized at high levels and were efficiently transported and secreted. A fusion protein containing the HXB2 V1/V2 domain was recognized by an HIVIIIB-infected patient serum as well as by 17 of 36 HIV-1 seropositive hemophiliac, homosexual male and intravenous drug user patient sera. Many of these HIV+ human sera reacted with V1/V2 domains from several HIV-1 clones expressed in fusion glycoproteins, indicating the presence of cross-reactive antibodies against epitopes in the V1/V2 domain. Recognition of gp(1-263):V1/V2HXB2 by the HIVIIIB-infected human patient serum was largely blocked by synthetic peptides matching V1 but not V2 sequences, while recognition of this construct by a broadly cross-reactive hemophiliac patient serum was not blocked by individual V1 or V2 peptides or by mixtures of these peptides. A construct containing the V3 domain of the IIIB strain of HIV-1, gp(1-263):V3HXB2, was recognized by sera from a human and a chimpanzee that had been infected by HIVIIIB but not by sera from hemophiliac patients who had been infected with HIV-1 of MN-like V3 serotype. The reactive sera had significantly higher titers when assayed against gp(1-263):V3HXB2 than when assayed against matching V3 peptides. Immunoprecipitation of this fusion glycoprotein by the human serum was only partially blocked by V3 peptide, indicating that this infected individual produced antibodies against epitopes in V3 that were expressed on the fusion glycoprotein but not by synthetic peptides. These data demonstrated that the chimeric glycoproteins described here effectively present native epitopes present in the V1/V2 and V3 domains of gp120 and provide efficient methods for detection of antibodies directed against native epitopes in these regions and for characterization of such epitopes.     

B cells induce tolerance by presenting endogenous peptide-IgG on MHC class II molecules via an IFN-gamma-inducible lysosomal thiol reductase-dependent pathway

We have previously demonstrated that splenic B cells, transduced with peptide-IgG fusion proteins, are efficient tolerogenic APCs in vivo. Specific hyporesponsiveness to epitopes encoded in the peptide-IgG fusion protein has been achieved to over one dozen Ags, and clinical efficacy has been established in animal models for several autoimmune diseases and hemophilia. Previous studies also demonstrated that tolerance in this system requires MHC class II expression by the transduced B cells. Yet, the mechanisms of this B cell tolerogenic processing pathway remain unclear. In this study, we show that MHC class II molecules on tolerogenic B cells present epitopes derived from endogenous, but not exogenous (secreted), peptide-IgG fusion protein. These class II epitopes from the IgG fusion protein are processed in lysosomes/endosomes in an IFN-gamma-inducible lysosomal thiol reductase-dependent manner. We suggest that the MHC class II presentation of endogenously produced fusion protein epitopes represents a novel mechanism for tolerance induced by peptide-IgG-transduced B cells. An understanding of this process might provide insights into central and peripheral tolerance induced by other professional and nonprofessional APCs.     

Murine graft vs host disease. A model for study of mechanisms that generate autoantibodies to ribonucleoproteins

We established chronic graft vs host disease in (BALB/c x A/J) F1 mice with the injection of lymphoid cells from the parental A/J strain. These animals developed glomerulonephritis, forefoot edema, alopecia, splenomegaly, and lymphadenopathy to various degrees, and all developed antinuclear antibodies. To determine whether these antibodies were directed against the small nuclear ribonucleoprotein (snRNP) particles that are characteristic targets for autoimmune responses in human rheumatic diseases, sera were studied in the 32P immunoprecipitation and immunoblotting assays. Among 20 mice, antibodies to snRNP developed in 10. These antibodies usually reached maximal levels about 4 wk after induction of graft vs host disease and generally fell thereafter. However, two mice developed antibodies to snRNP between the 10th and 20th wk of follow-up. Sera from six mice strongly recognized the U1 snRNP and an additional serum strongly bound both the U1 and U3 particles. Several sera contained lower levels of antibodies specific for the U3 and possibly pre-U2 snRNP particles. In immunoblots, sera that immunoprecipitated the U1 snRNP bound epitopes located on its 70,000 Da, A, B'/B, and/or C polypeptides. Sera that immunoprecipitated the U3 snRNP recognized a 34,000-Da polypeptide. These polypeptides are known to bear the autoantigenic epitopes that are recognized by human sera containing anti-U1 RNP and anti-U3 RNP autoantibodies. We conclude that chronic graft vs host disease in mice provides a model for the study of the autoimmune responses that characterize human diseases such as mixed connective tissue disease, scleroderma, and SLE.     

Epitope mapping and characterization of the infectious hematopoietic necrosis virus glycoprotein, using fusion proteins synthesized in Escherichia coli.

A characterization of the antigenic determinants (epitopes) of the glycoprotein (G) of infectious hematopoietic necrosis virus was made by expressing different regions of the G gene in Escherichia coli. A cDNA copy of the G gene was divided into four fragments by TaqI digestion, and the fragments were subcloned into pATH vectors, placing the expression of each G gene fragment under control of the trpE promoter. The resulting plasmids, pXL2, pXL3, and pXL7, encoded trpE-G fusion proteins subsequently detected with anti-infectious hematopoietic necrosis virus sera by Western immunoblots. A comparison of reactivities of the fusion proteins encoded by these plasmids was made by Western immunoblot and radioimmunoassay with a number of anti-G specific monoclonal antibodies (MAbs). The nonneutralizing MAb 136J reacted with the trpE-G fusion protein encoded by pXL3 and fusion proteins encoded by plasmids p52G and p618G, which were described in previous studies (R. D. Gilmore, Jr., H. M. Engelking, D. S. Manning, and J. C. Leong. Bio/Technology 6:295-300, 1988). Another nonneutralizing MAb, 2F, bound to the pXL3 fusion protein, and the neutralizing MAb RB/B5 recognized the pXL7 fusion protein. All fusion proteins were tested as vaccines in rainbow trout fry. Although significant protection was induced by all fusion proteins, the pXL3 fusion protein was most effective as a vaccine.     

Rapid screening for improved solubility of small human proteins produced as fusion proteins in Escherichia coli

A prerequisite for structural genomics and related projects is to standardize the process of gene overexpression and protein solubility screening to enable automation for higher throughput. We have tested a methodology to rapidly subclone a large number of human genes and screen these for expression and protein solubility in Escherichia coli. The methodology, which can be partly automated, was used to compare the effect of six different N-terminal fusion proteins and an N-terminal 6*His tag. As a realistic test set we selected 32 potentially interesting human proteins with unknown structures and sizes suitable for NMR studies. The genes were transferred from cDNA to expression vectors using subcloning by recombination. The subcloning yield was 100% for 27 (of 32) genes for which a PCR fragment of correct size could be obtained. Of these, 26 genes (96%) could be overexpressed at detectable levels and 23 (85%) are detected in the soluble fraction with at least one fusion tag. We find large differences in the effects of fusion protein or tag on expression and solubility. In short, four of seven fusions perform very well, and much better than the 6*His tag, but individual differences motivate the inclusion of several fusions in expression and solubility screening. We also conclude that our methodology and expression vectors can be used for screening of genes for structural studies, and that it should be possible to obtain a large fraction of all NMR-sized and nonmembrane human proteins as soluble fusion proteins in E. coli.