Secretion and in vivo folding of the Fab fragment of the antibody McPC603 in Escherichia coli: influence of disulphides and cis-prolines

Using the well-characterized antibody McPC603 as a model, wehad found that the F_v fragment can be isolated from Escherichiacoli as a functional protein in good yields, whereas the amountof the correctly folded F_ab fragment of the same antibody producedunder identical conditions is significantly lower. In this paper,we analyse the reasons for this difference. We found that avariety of signal sequences function in the secretion of theisolated chains of the F_ab fragment or in the co-secretion ofboth chains in E.coli. The low yield of functional F_ab fragmentis not caused by inefficient expression or secretion in E.coli,but by inefficient folding and/or assembly in the periplasm.We compared the folding yields for the F_v and the F_ab fragmentin the periplasm under various conditions. Several diagnosticframework variants were constructed and their folding yieldsmeasured. The results show that substitutions affecting cis-prolineresidues and those affecting various disulphide bonds in theprotein are by themselves insufficient to dramatically changethe partitioning of the folding pathway to the native structure,and the cause must lie in a facile aggregation of folding intermediatescommon to all structural variants. However, all structural variantscould be obtained in native form, demonstrating the generalutility of the secretory expression strategy.     

Engineering a soluble extracellular erythropoietin receptor (EPObp) in Pichia pastoris to eliminate microheterogeneity, and its complex with erythropoietin

The extracellular ligand-binding domain (EPObp) of the humanEPO receptor (EPOR) was expressed both in CHO (Chinese HamsterOvary) cells and in Pichia pastoris. The CHO and yeast expressedreceptors showed identical affinity for EPO binding. Expressionlevels in P.pastoris were significantly higher, favoring itsuse as an expression and scale-up production system. Incubationof EPO with a fourfold molar excess of receptor at high proteinconcentrations yielded stable EPO–EPObp complexes. Quantificationof EPO and EPObp in the complex yielded a molar ratio of oneEPO molecule to two receptor molecules. Residues that are responsiblefor EPOR glycosylation and isomerization in Pichia were identifiedand eliminated by site-specific mutagenesis. A thiol modificationwas identified and a method was developed to remove the modifiedspecies from EPObp. EPObp was complexed with erythropoietin(EPO) and purified. The complex crystallized in two crystalforms that diffracted to 2.8 and 1.9 Å respectively. (Form1 and form 2 crystals were independently obtained at AxyS Pharmaceuticals,Inc. and Amgen, Inc. respectively.) Both contained one complexper asymmetric unit with a stoichiometry of two EPObps to oneEPO.     

An internal cellulose-binding domain meidates adsorption of an engineered bifunctional xylanase/cellulase

A chimeric xylanase/endoglucanase (XynCenA) with an internalcellulose-binding domain was constructed by fusing the Bacillussubtilis xyn gene fragment to the 5'-end of the Cellulomonasfimi cenA. A polyhistidine-encoding sequence was also fusedto the 5'-end of the xyn gene. The gene fusion was overexpressedin Escherichia coli and the fusion poly-peptide purified fromthe cell extracts using the polyhistldine tail. The hybrid proteinbehaved like the parental endo-glucanase or xylanase when assayedon a number of soluble and insoluble cellulosic substrates orxylans. The presence of two distinct active sites and the internalcellulose-binding domain did not significantly affect the hydrolysisof any of these substrates. However, the fusion protein exhibiteda strong affinity for both mkrocrystaUine cellulose (Avicel)and regenerated chitin. Like the parental endoglucanase, boundXynCenA could not be duted from these porysaccharides with eitherlow or high salt buffer or distilled water. More stringent conditions,such as 1% SDS or 8 M guanidinium hydro-chloride, fully desorbedthe protein. The fusion protein did not adsorb significantlyto insoluble xylan     

A combinatorial library of an {alpha}-helical bacterial receptor domain

The construction and characterization of a combinatorial libraryof a solvent-exposed surface of an -helical domain derived froma bacterial receptor is described. Using a novel solid-phaseapproach, the library was assembled in a directed and successivemanner utilizing single-stranded oligonucleotides containingmultiple random substitutions for the variegated segments ofthe gene fragment The simultaneous substitution of 13 residuesto all 20 possible amino acids was carried out in a region spanning81 nucleotides. The randomization was made in codons for aminoacids that were modelled to be solvent accessible at a surfacemade up from two of the three a-helices of a monovalent Fc-bindingdomain of staphylococcal protein A. After cloning of the PCR-amplifiedlibrary into a phagemid vector adapted for phage display ofthe mutants, DNA sequencing analysis suggested a random distributionof codons in the mutagenized positions. Four members of thelibrary with multiple substitutions were produced in Escherichiacoli as fusions to an albumin-binding affinity tag derived fromstreptococcal protein G. The fusion proteins were purified byhuman serum albumin affinity chromatography and subsequentlycharacterized by SDSelectrophoresis, CD spectroscopy and biosensoranalysis. The analyses showed that the mutant protein A derivativescould all be secreted as soluble full-length proteins. Furthermore,the CD analysis showed that all mutants, except one with a prolineintroduced into helix 2, have secondary structures in closeagreement with the wild-type domain. These results proved thatmembers of this -helical receptor library with multiple substitutionsin the solvent-exposed surface remain stable and soluble inE.coli. The possibility of using this library for a phenotypicselection strategy to obtain artificial antibodies with novelfunctions is discussed.     

Insertion of a disulfide-containing neurotoxin into E.coil alkaline phosphatase: the hybrid retains both biological activities

We have inserted a disulfide-containing snake neurotoxin intothe N-terminal end of Escherichia coli alkaline phosphatase,between residues +6 and +7 of the mature enzyme. For this purpose,we have designed a cloning and expression vector which allowsinsertion of foreign DNA between the corresponding codons, andvisual selection of the desired recombinant clones upon recoveryof phosphatase activity. The hybrid protein is exported to thebacterial periplasm, the alkaline phosphatase signal peptideis correctly processed, and both domains are functionally conformed.The phosphatase domain displays catalytic activity, and theinserted toxin is able to bind to its biological target, thenicotinic acetylcholine receptor. The hybrid molecule is remarkablystable and resistant to proteolysis. Crude periplasmic extractcontaining the hybrid can be used as a tracer-containing reagentin competitive enzymo-immuno and enzymo-receptor assays. Wepropose to use the system described in this paper for fast preparationof properly folded disulfide-containing enzymatic probes.     

A dual-affinity gene fusion system to express small recombinant proteins in a soluble form: expression and characterization of protein A deletion mutants

A novel gene fusion system to express and purify small recombinantproteins in Escherichia coli has been constructed. The conceptallows for affinity purification of soluble gene products bysequential albumin- and Zn²+-affinity chromatography. The dual-affinitysystem is well suited for expression of unstable proteins asonly full-length protein is obtained after purification andproteins gain proteolytic stability in the fusion protein. Herewe show that the dual-affinity approach can be used for theexpression of various unstable derivatives of a single IgG-bindingdomain based on staphylococcal protein A. Analysis of the proteolyticstabilities and the IgG-binding properties of the differentmutant proteins suggest that the model for the structure ofan IgG-binding domain must be re-evaluated.     

Analysis of the structure of Pseudomonas glumae lipase

The lipase produced by Pseudomonas glumae is monomeric in thecrystalline state and has a serine protease-like catalytic triad;Ser87-His285-Asp263. The largest domain of the protein resemblesclosely a subset of the frequently observed /ß-hydrolasefold and contains a well-defined calcium site. This paper describesstructural analysis of this protein, focusing on (i) structuralcomparison with the lipase from Geotrichum candidum, (ii) theprobable nature of the conformational change involved in substratebinding and (iii) structural variations amongst the family ofPseudomonas Upases. This analysis reveals similarities betweenP.glumae lipase and G.candidum lipase involving secondary structuralelements of the hydrolase core and the loops carrying the catalyticserine and histidine residues. A possible functional equivalencehas also been identified between parts of the two moleculesthought to be involved in a confonmational change. In addition,determination of the structure of P.glumae lipase has allowedrationalization of previously reported protein engineering experiments,which succeeded in improving the stability of the enzyme withrespect to proteolysis.     

The predicted secondary structure of the G-type glutamineamidotransferase is compatible with TEM-barrel topology

Glutamine amidotransferase (GAT) subunits or domains catalyzean important partial reaction in many complex biosynthetic reactions.The structure of one member of the F-type GATs is known, butthe structure of the unrelated G-type is still unknown. Becausemany protein sequences are available for anthranilate synthasecomponent II (product of the trpG gene), we have predicted itsaverage secondary structure by a joint prediction method [Niermannand Kirschner (1991a) Protein Engng, 4, 359–370]. Thepredicted eight ß-strands and seven -helices followan 8-fold cyclic repetition of a ß-strand-loop--helix-loopmodule with helix 7 missing. This pattern of secondary structuresuggests that the G-type GAT domain has an 8-fold ß-barreltopology, as found first in triose phosphate isomerase (TIM-barrel).This model is supported by the location of known catalyticallyessential residues in loops between (ß-strands and-helices. Evidence from published sequencing and mutationalstudies on selected members of the GAT superfamily (carbamoylphosphate, imidazoleglycerol phosphate, GMP and CTP synthases)support both the secondary structure prediction and the TIM-barreltopology.     

High level expression of a synthetic gene coding for IgG-binding domain B of Staphylococcal protein A

A gene coding for one of the IgG-binding domains of Staphylococcalprotein A, designated domain B, was chemically synthesized.This gene was tandemly repeated to give dimeric and tetramericdomain B genes by the use of two restriction enzymes which gaveblunt ends. The genes were highly expressed in Escherichia colito afford a large amount of dimeric and tetrameric domain Bproteins. The single domain B protein was efficiently producedas a fusion protein with a salmon growth hormone fragment. Thefusion protein was converted to monomeric domain B by cyanogenbromide cleavage. The CD spectra of the monomeric, dimeric andtetrameric domain B proteins were essentially the same as thatof native form protein A, showing that their secondary structureswere very similar. The dimeric and tetrameric domain B proteinsformed precipitates with IgG as protein A. This system permitsthe efficient production of mutated single and multiple IgG-bindingdomains which can be used to study structural changes and proteinA–immunoglobulin interactions.     

Testing the procedure of simulated annealing by refining homologous immunoglobulin light-chain dimers

The refinement of antigen-binding fragment structures by themethod of simulated annealing was tested. Using the programX-PLOR, we refined the structure of one immunoglobulin light-chaindimer against 2.8 Å diffraction data collected for a homologouslight-chain dimer. The refinement proceeded smoothly; -carbonsof the conserved segments of the domain moved to the positionsin the reference structure solved independently. An averagemovement of –1.5 Å for atoms in the variable domains(half of the molecule) was observed. Though the final R-factorsand energy terms of the reference and test structures were verysimilar, some of the chain segments of the hypervariable loops(HVR3s₃) and the ends of some side chains did not converge tothe positions in the reference structure. Therefore, althoughglobally the refinement worked very well, positions of the loopsand the side chains that are critical for immunoglobulin functionhave to be carefully examined by difference Fourier techniques.     

Prediction of the antigenic sites of the cystic fibrosis transmembrane conductance regulator protein by molecular modelling

Antibodies are powerful tools for studying the in situ localizationand physiology of proteins. The prediction of epitopes by molecularmodelling has been used successfully for the papilloma virus,and valuable antibodies have been raised [Muller et aL (1990)J. Gen. Virol, 71, 2709–2717]. We have improved the modellingapproach to allow us to predict epitopes from the primary sequencesof the cystic fibrosis transmembrane conductance regulator.The procedure involves searching for fragments of primary sequenceslikely to make amphipathic secondary structures, which are hydrophilicenough to be at the surface of the folded protein and thus accessibleto antibodies. Amphipathic helices were predicted using themethods of Berzofsky, Eisenberg and Jahnig. Their hydrophobichydrophilicinterface was calculated and drawn, and used to predict theorientation of the helices at the surface of the native protein.Amino acids involved in turns were selected using the algorithmof Eisenberg. Tertiary structures were calculated using ‘FOLDING’,a software developed by R.Brasseur for the prediction of smallprotein structures [Brasseur (1995) J. MoL Graphics, in press].We selected sequences that folded as turns with at least fiveprotruding polar residues. One important property of antibodiesis selectivity. To optimize the selectivity of the raised antibodies,each sequence was screened for similarity (FASTA) to the proteinsequences from several databanks. Ubiquitous sequences werediscarded. This approach led to the identification of 13 potentialepitopes in the cystic fibrosis transmembrane conductance regulator:seven helices and six loops.     

In vivo copper- and cadmium-binding ability of mammalian metallothionein domain

The ß domain of mouse metallothionein 1 (ßMT) wassynthesized in Escherichia coli cells grown in the presenceof copper or cadmium. Homogenous preparations of Cu–ßMTand Cd–ßMT were used to characterize the correspondingin vivo-conformed metal-clusters, and to compare them with thespecies obtained in vitro by metal replacement to a canonicalZn₃–ßMT structure. The copper-containing ßMTclusters formed inside the cells were very stable. In contrast,the nascent ß peptide, although it showed cadmium bindingability, produced a highly unstable species, whose stoichiometrydepended upon culture conditions. The absence of ßMT proteinin E.coli protease-proficient hosts grown in cadmium-supplementedmedium pointed to drastic proteolysis of a poorly folded ßpeptide, somehow enhanced by the presence of cadmium. Possiblefunctional and evolutionary implications of the bioactivityof mammalian ßMT in the presence of monovalent and divalentmetal ions are discussed.     

Molecular modeling of the amyloid-{beta}-peptide using the homology to a fragment of triosephosphate isomerase that forms amyloid in vitro

The main component of the amyloid senile plaques found in Alzheimer'sbrain is the amyloid-ß-peptide (Aß), a proteolyticproduct of a membrane precursor protein. Previous structuralstudies have found different conformations for the Aßpeptide depending on the solvent and pH used. In general, theyhave suggested an -helix conformation at the N-terminal domainand a ß-sheet conformation for the C-terminal domain.The structure of the complete Aß peptide (residues 1–40)solved by NMR has revealed that only helical structure is presentin Aß. However, this result cannot explain the large ß-sheetAß aggregates known to form amyloid under physiologicalconditions. Therefore, we investigated the structure of Aßby molecular modeling based on extensive homology using theSmith and Waterman algorithm implemented in the MPsrch program(Blitz server). The results showed a mean value of 23% identitywith selected sequences. Since these values do not allow a clearhomology to be established with a reference structure in orderto perform molecular modeling studies, we searched for detailedhomology. A 28% identity with an /ß segment of a triosephosphateisomerase (TIM) from Culex tarralis with an unsolved three-dimensionalstructure was obtained. Then, multiple sequence alignment wasperformed considering Aß, TIM from C.tarralis and anotherfive TIM sequences with known three-dimensional structures.We found a TIM segment with secondary structure elements inagreement with previous experimental data for Aß. Moreover,when a synthetic peptide from this TIM segment was studied invitro, it was able to aggregate and to form amyloid fibrils,as established by Congo red binding and electron microscopy.The Aß model obtained was optimized by molecular dynamicsconsidering ionizable side chains in order to simulate Aßin a neutral pH environment. We report here the structural implicationsof this study.     

Structure-function relationships in the catalytic and starch binding domains of glucoamylase

Sixteen primary sequences from five sub-families of fungal,yeast and bacterial glucoamylases were related to structuralinformation from the model of the catalytic domain of Aspergillusawamori var. X100 glucoamylase obtained by protein crystallography.This domain is composed of thirteen -belices, with five conservedregions defining the active site. Interactions between methyl-maltoside and active site residues were modelled, and the importanceof these residues on the catalytic action of different glucoamylaseswas shown by their presence in each primary sequence. The overallstructure of the starch binding domain of some fungal glucoamylaseswas determined based on homology to the Cterminal domains ofBacillus cyclodextrin glucosyltransferases. Crystallographyindicated that this domain contains 6–8 ß-strandsand homology allowed the attribution of a disulfide bridge inthe glucoamylase starch binding domain. Glucoamylase residuesThr525, Asn530 and Trp560, homologous to Bacillus stearothermophiluscyclodextrin glucosyltransferase residues binding to maltosein the Cterminal domain, could be involved in raw-starch binding.The structure and length of the linker region between the catalyticand starch binding domains in fungal glucoamylases can varysubstantially, a further indication of the functional independenceof the two domains.     

Characterization of the DNA binding domain of the mouse IRF-2 protein

The DNA binding domain of the interferon regulatory factor-2protein (IRF-2) has been produced and characterized, -chymotrypsindigestion of the purified IRF-2 protein bound to a syntheticbinding site yields a peptide fragment of 14 K in molecularweight. N-terminal analysis of this peptide fragment showedthat its sequence is the same as that of the intact IRF-2. Apeptide fragment of {small tilde} 14 K, IRF-2(113), which correspondsto the N-terminal 113 amino acids of the intact IRF-2 protein,has been expressed in a functional form in Escherichia coli.The first methionine was processed during the expression andthe purified IRF-2(113) thus contains 112 amino acids. DNaseI footprinting and gel retardation assaying showed that IRF-2(113)binds to a synthetic DNA having the consensus binding site andto the upstream regulatory sequence of the IFN-ß geneas intact IRF-2 does. These results showed that this peptidefragment, IRF-2(113), may be a good material for investigationof the DNA binding domain of IRF-2 and of the DNA–proteininteraction.     

Quantitative analysis of protein far UV circular dichroism spectra by neural networks

A new method based on neural network theory is presented toanalyze and quantify the information content of far UV circulardichroism spectra. Using a backpropagation network model witha single hidden layer between input and output, it was possibleto deduce five different secondary structure fractions (helix,parallel and antiparallel ß-sheet, ß-turnand random coil) with satisfactory correlations between calculatedand measured secondary structure data. We demonstrate that foreach wavelength interval a specific network is suitable. Theremaining discrepancy between the secondary structure data fromneural network prediction and crystallography may be attributedto errors in the determination of protein concentration andrandom noise in the CD signal, as indicated by simulations.     

Native-like {beta}-hairpin structure in an isolated fragment from ferredoxin: NMR and CD studies of solvent effects on the N-terminal 20 residues

The conformational properties of protein fragments have beenwidely studied as models of the earliest initiation events inprotein folding. While native-like -helices and ß-turnshave been identified, less is known about the factors that underlyß-sheet formation, in particular ß-hairpins,where considerably greater long-range order is required. TheN-terminal 20 residue sequence of native ferredoxin I (fromthe blue-green alga Aphanothece sacrum ) forms a ß-hairpinin the native structure and has been studied in isolation byNMR and CD spectroscopy. Local native-like interactions aloneare unable to stabilize significantly a folded conformationof the 20-residue fragment in purely aqueous solution. However,we show that the addition of low levels of organic co-solventspromotes formation of native-like ß-hairpin structure.The results suggest an intrinsic propensity of the peptide toform a native-like ß-hairpin structure, and that theorganic co-solvent acts in lieu of the stabilizing influenceof tertiary interactions (probably hydrophobic contacts) whichoccur in the folding of the complete ferredoxin sequence. Thestructure of the isolated hairpin, including the native-likeregister of interstrand hydrogen bonding interactions, appearsto be determined entirely by the amino acid sequence. The solventconditions employed have enabled this intrinsic property tobe established.     

Structural similarities in glucoamylases by hydrophobic cluster analysis

The model of the catalytic domain of Aspergillus awamori var.X100 glucoamylase was related to 14 other glucoamylase proteinsequences belonging to five subfamilies. Structural featuresof the different sequences were revealed by multisequence alignmentfollowing hydrophobic cluster analysis. The alignment agreedwith the hydrophobic microdomains, normally conserved throughoutevolution, evaluated from the 3-D model. Saccharomyces and Clostridiumglucoamylases lack the -helix exterior to the catalytic domain.A different catalytic base was found in the Saccharomyces glucoamylasesubfamily. The starch binding domain of fungal glucoamylaseshas identical structural features and substrate interactingresidues as the C-terminal domain of models of Bacillus circulanscyclodextrin glucosyltransferases. Three putative N-glycosylationsites were found in the same turns in glucoamylases of differentsubfamilies. O-Glycosylation is present at different levelsin the catalytic domain and in the linker between the catalyticand starch binding domains.     

Bacterial secretion of the Fab fragment of a mouse monoclonal IgM that reacts with IgG variable regions

In this report, we describe the expression system that enabledus to produce in Escherichia coli the Fab fragment of a mouseIgM that has previously been shown to inhibit the binding ofIgG to autoantigens by interacting with their variable regions.In our system, both light chain and heavy chain fragments wereput under the control of the malE promoter. The light chainwas fused to the MalE signal sequence, while the heavy chainvariable and first constant region were fused to the alkalinephosphatase signal sequence. In this system, after inductionof the promoter with maltose, the Fab fragment could be detectedin a periplasmic extract of the bacteria by Western blottingand also by ELISA. This Fab fragment was purified on a goatanti-mouse immunoglobulin immunoadsorbent and biotinylated.The Fab fragment produced by E.coli reacted with the trinitrophenyl(TNP) hapten and F(ab')₂ fragments of mouse IgG and these reactivitiescould be specifically inhibited by the corresponding solubleantigens. The dissociation constants of this Fab were 1.65 x10^–6 M for TNP and 5 x 10^–6 M for IgG F(ab')₂ fragments,indicating that the affinity of the Fab fragment compared withthat of the whole IgM molecule was similar for TNP but was lowerfor IgG F(ab')₂ fragments     

An approach to protein homology modelling based on an ensemble of NMR structures: application to the Sox-5 HMG-box protein

A new approach has been developed to reduce multiple proteinstructures obtained from NMR structure analysis to a smallernumber of representative structures which still reflect thestructural diversity of the data sets. The method, based onthe clustering of similar structures, has been tested in thehomology model building of the structure of Sox-5, a sequence-specificDNA-binding protein belonging to the high mobility group (HMG)nuclear proteins family. Sox (SRY box) genes are the autosomalgenes related to the sex-determining SRY, Y chromosomal gene.The Sox-5 protein, encoded by one of the SRY-related genes,displays a 29% sequence identity with the HMG1 B-box domainwhose structure, determined previously by NMR, has been usedin our study to predict the structure of Sox-5. Two independentensembles of HMG1 structures, each represented by closely relatedcoordinate sets, were used. Nine representative structures forHMG1 were subsequently selected as starting points for the modellingof Sox-5. The model of the protein shows close similarity tothe HMG1 fold, with differences at the secondary structure levellocated mainly in a-helices 1 and 3. A left-handed, three residueper turn polyproline II helix, forming a conserved polyprolineII/-helix supersecondary motif, was identified in the N-terminalregion of Sox-5 and other HMG boxes.