Jonathan M. Boulter, Meir Glick, Penio T. Todorov, Emma Baston, Malkit Sami, Pierre Rizkallah, Bent Okay. Jakobsen, Secure, soluble T‐cell receptor molecules for crystallization and therapeutics, Protein Engineering, Design and Choice, Quantity 16, Situation 9, September 2003, Pages 707–711, https://doi.org/10.1093/protein/gzg087
Summary
Introduction
T‐cell receptors (TCRs) are specialised antigen recognition molecules expressed on the floor of T cells and able to particularly recognizing peptide–main histocompatibility advanced (MHC) antigens (Davis et al., 1998; Garcia et al., 1999). TCRs are extremely variable, enabling T cells to acknowledge an enormous variety of peptide–MHC antigens, though particular person T cells usually categorical just one sort of TCR. The TCR consists of an α/β heterodimer, every chain of which contains a variable area linked to a relentless area.
Many methods have been proposed for making soluble variations of the α/β TCR. Most of those work for a really restricted variety of TCRs, e.g. single‐chain TCR designs (Chung et al., 1994; Schodin et al., 1996; Khandekar et al., 1997). A extra usually relevant technique for producing soluble TCRs has just lately been developed, which entails stabilizing the α/β heterodimer by fusion of the TCR extracellular domains to the jun/fos coiled‐coil domains (Willcox et al., 1999a,b). Though this technique is extremely profitable at producing soluble TCRs which retain particular ligand binding exercise, these soluble TCRs have proved troublesome to crystallize. Moreover, no technique has but been reported that enables TCRs for use for therapeutic antigen concentrating on.
The native TCR heterodimer is stabilized by a membrane‐proximal disulphide bridge, however methods to provide soluble TCRs incorporating this bond haven’t been profitable (Garboczi et al., 1996). Subsequently, we sought to design a generic technique for producing soluble human TCRs, stabilized by a non‐native disulphide bond between the extracellular α and β fixed domains. Molecular modelling was used to find out the optimum website for the disulphide bond, and three recombinant soluble TCRs have been produced in Escherichia coli and refolded with excessive effectivity. The disulphide‐linked TCRs (dsTCRs) have been extremely secure and displayed genuine binding exercise demonstrated by BIAcore™ floor plasmon resonance (SPR) experiments.
The design of the dsTCRs makes them extremely amenable to crystallization and we report the preliminary crystallization of 1 dsTCR together with crystallographic evaluation confirming the inter‐chain disulphide bond place. These dsTCRs are more likely to be amenable for in vivo therapeutic functions, and candidates are at present being developed for medical functions. The design of the dsTCR is topic to a patent utility (Jakobsen and Glick, 2001).
Supplies and strategies
Molecular modelling
The crystal construction of the B7 human α/β TCR, particular for human leukocyte antigen (HLA)‐A*0201 in advanced with the human T‐trophic lymphocyte virus sort‐1 (HTLV‐1) tax 11‐19 peptide, solved at 2.5 Å decision and deposited within the Protein Knowledge Financial institution (Berman et al., 2000) beneath the title 1BD2 (Ding et al., 1998), was used for the disulphide bond predictions. Residues forming the interface between the fixed domains have been examined, and aspect chains whose β‐carbons have been nearer than 7 Å have been mutated insilico to cysteine. The χ1 dihedral angle (of the cysteine aspect chain) was then rotated as a way to affirm that the space between the sulphur atoms was optimum for a disulphide bond and didn’t perturb the tertiary construction of the protein.
Cloning of TCR chains
TCR chains have been cloned into the bacterial expression vector pGMT7 by polymerase chain response (PCR) cloning utilizing a 5′ primer containing an NdeI website which includes the ATG begin codon, and a 3′ primer containing a HindIII website and a TAA cease codon which replaces the native inter‐chain cysteine codon. A free cysteine within the fixed area of the β‐chain was mutated to alanine as a way to facilitate invitro refolding.
Cysteine codons have been launched by PCR mutagenesis. Complementary primers have been designed which annealed 5′ and three′ of the specified mutation, however contained an altered codon encoding a cysteine. These have been utilized in a pfu‐pushed PCR utilizing a template encoding the related TCR chain produced as above. After 16 rounds of PCR with an extension time of 8 min, the response was digested with DpnI to digest methylated DNA. After 1 h digestion at 37°C, 10 µl of the response was reworked into E.coli XL1‐blue cells which have been plated out onto LB/100 µg/ml ampicillin plates. Plates have been incubated in a single day at 37°C, then single colonies have been grown to stationary section in 10 ml of LB containing 100 µg/ml ampicillin. A Qiagen miniprep equipment was used to purify DNA from these clones. The sequence of the insert was verified by automated sequencing.
The β‐chain fixed area has a pure BglII website 5′ of the engineered cysteine codon which permits cloning of different β‐chains into the mutated assemble. So as to have the ability to clone totally different α‐chains into the mutated α‐chain assemble, we launched a silent BamHI website by PCR mutagenesis at place TRAC P5D6P7 cctgaccct→ccggatcct. This website might then be used to PCR clone new α‐chains into the assemble containing the mutant cysteine codon.
TCR expression and refolding (Willcox et al., 1999b)
TCR chains have been expressed individually as inclusion our bodies within the E.coli pressure BL21‐DE3(pLysS) by induction in mid‐log section with 0.5 mM IPTG. Inclusion our bodies have been remoted by sonication, adopted by successive wash and centrifugation steps utilizing 0.5% Triton X‐100. Lastly, the inclusion our bodies have been dissolved in 6 M guanidine, 10 mM dithiothreitol (DTT), 10 mM ethylenediaminetetra‐acetate (EDTA), buffered with 50 mM Tris pH 8.1 and saved at –80°C.
Soluble TCR was refolded by fast dilution of a combination of the dissolved α‐ and β‐chain inclusion our bodies into 5 M urea, 0.4 M l‐arginine, 100 mM Tris pH 8.1, 3.7 mM cystamine, 6.6 mM β‐mercapoethylamine (4°C) to a closing focus of 60 mg/l.
Purification of the soluble TCR
The refold combination was dialysed for twenty-four h in opposition to 10 vol of demineralized water, then in opposition to 10 vol of 10 mM Tris pH 8.1 at 4°C. The refolded protein was then filtered and loaded onto a POROS 50HQ column (Utilized Biosystems). The column was washed with 10 mM Tris pH 8.1 previous to elution with a 0–500 mM NaCl gradient in the identical buffer. Fractions have been analysed by Coomassie‐stained sodium docecyl sulphate (SDS)–10% NuPAGE (Novagen, WI), and TCR‐containing fractions have been pooled and additional purified by gel filtration on a Superdex 75PG 26/60 column (Amersham Biosciences, Uppsala, Sweden) pre‐equilibrated in phosphate‐buffered saline. Fractions comprising the principle peak have been pooled and analysed additional. The ultimate purified 1G4 dsTCR was analysed by Coomassie‐stained SDS–10% NuPAGE beneath lowering and non‐lowering situations, and an aliquot of protein was buffer exchanged into HBSE (HEPES pH 7.4, 150 mM NaCl, 3 mM EDTA) and concentrated previous to exercise willpower by BIAcore™ SPR.
Preparation of soluble biotinylated HLA complexes
Peptide–HLA‐A*0201 complexes have been ready by in vitro refolding from bacterially expressed inclusion our bodies and artificial peptide (Garboczi et al., 1992), adopted by enzymatic biotinylation with BirA enzyme (O’Callaghan et al., 1999).
Willpower of binding exercise by BIAcore™ SPR
CM5 BIAcore™ chips (BIAcore AB, St Albans, UK) have been coated with streptavidin utilizing amine coupling, and pHLA complexes have been flowed over particular person movement cells at a focus of ∼50 µg/ml till the response measured ∼1000 response models (RU). Soluble TCR was concentrated to >10 mg/ml and the focus was decided by measurement of OD280 utilizing an extinction coefficient calculated from the sequence. Serial dilutions of the TCR have been flowed over the totally different pMHC complexes and the response values at equilibrium have been decided for every focus. Dissociation constants (KD) have been decided by plotting the response over background in opposition to the protein concentrations adopted by a least‐squares match to the Langmuir binding equation, assuming a 1:1 interplay (Willcox et al., 1999a).
Analysis of stability
Aliquots of sterile 1G4 dsTCR in phosphate‐buffered saline have been incubated at –65 and 25°C and analysed utilizing Coomassie‐stained SDS–polyacrylamide gel electrophoresis (PAGE), isoelectric focussing, and ion‐trade and dimension‐exclusion excessive‐stress liquid chromatography.
Crystallization and structural evaluation of dsTCRs
1G4 dsTCR was concentrated to 10 mg/ml and crystallized by the hanging drop technique. Two microlitre aliquots of the concentrated protein have been combined with an equal quantity of a precipitating resolution, and equilibrated in opposition to 500 µl of the precipitant. The precipitant resolution was made up of 0.2 M ammonium acetate, 0.1 M tri‐sodium citrate dihydrate pH 5.6, 30% (v/v) polyethylene glycol (PEG) 4000 in water.
Outcomes
Molecular modelling
Desk I exhibits the β‐carbon aspect chain distances for chosen pairs of residues within the α/β fixed area interface. TRAC threonine 48 and TRBC serine 57 [residue numbering according to the International Immunogenetics Database (IMGT) (Lefranc and Lefranc, 2001)] have been predicted to be optimum due to their shut proximity, the course of their aspect chains, that are oriented in direction of one another, and the relative chemical similarity of their aspect chains to cysteine. Certainly, mutating these residues to cysteine in silico and orienting the sulphur atoms in direction of one another by rotating the χ1 dihedral angle yields a disulphide distance as shut as 1.5 Å with out perturbing the tertiary construction of the protein.
Manufacturing of dsTCRs
A6 TCRs (Utz et al., 1996) (particular for HLA‐A*0201 in advanced with the HTLV‐1 tax 11‐19 peptide) with engineered inter‐chain disulphide linkages in varied positions, described in Desk I, have been produced utilizing PCR mutagenesis, adopted by expression as E.coli inclusion our bodies, and in vitro refolding. Of those, the TCR with an engineered disulphide linkage between cysteines launched at positions TRAC 48 and TRBC 57 produced the most effective yields of soluble TCR with clear BIAcore™ binding information (information not proven).
TCR chains from human T‐cell clones JM22 (Lehner et al., 1995) (particular for HLA‐A*0201‐influenza matrix peptide) and 1G4 (Chen et al., 2000) (particular for HLA‐A*0201‐NY‐ESO peptide) have been additionally cloned into the α‐ and β‐chain constructs containing engineered cysteines at positions TRAC 48 and TRBC 57, respectively. The A6, JM22 and 1G4 dsTCRs refolded with >40% effectivity to provide protein preparations which have been >95% pure after ion‐trade and gel‐filtration chromatography, as judged by Coomassie‐stained SDS–PAGE evaluation.
Determine 1 exhibits a typical purification for the 1G4 dsTCR on (i) anion‐trade and (ii) gel‐filtration columns. The purified protein was analysed by Coomassie‐stained SDS–PAGE beneath lowering and non‐lowering situations (Determine 2). The dsTCR α‐ and β‐chains run individually beneath lowering situations, however the launched disulphide bond holds the chains collectively beneath non‐lowering situations they usually run as a single band of upper molecular weight.
BIAcore™ SPR binding evaluation
BIAcore™ SPR equilibrium binding evaluation for 1G4 dsTCR binding to the HLA‐A*0201‐NY‐ESO peptide advanced is proven in Determine 3. The estimate for the dissociation fixed (KD) of 15.5 µM compares with a worth of 14 µM obtained for a similar TCR refolded as a jun/fos fusion (N.Lissin, unpublished outcomes).
KD estimates for the dsTCRs are in contrast with the same jun/fos fusions in Desk II. In all instances, the binding affinity is comparable with the jun/fos fusions indicating that the binding domains of the dsTCRs are usually not considerably affected by the fixed area cysteine linkage.
Stability evaluation of the 1G4 dsTCR
The 1G4 dsTCR was extremely secure, exhibiting no vital chemical or bodily degradation at –65°C over a interval of seven months and at 25°C over a interval of 4 weeks, utilizing the analytical strategies described (information not proven). The protein was additionally secure over the course of 4 freeze–thaw cycles.
Crystallization and structural evaluation of the 1G4 dsTCR
Single crystal X‐ray diffraction information of 1G4 dsTCR have been collected at Station 9.6 of the SRS Daresbury Laboratory, UK. The efficient decision of the information was 2.5 Å. The construction was solved by molecular alternative utilizing the 1BD2 construction of the B7 TCR (Ding et al., 1998) because the beginning mannequin. The construction of the fixed domains was refined at 2.5 Å decision. Determine 4 exhibits that the construction of the engineered disulphide bond matches a lot better into the experimentally decided electron density map than the native construction, indicating that in in vitro refolding this bond kinds as predicted. The bond distance between the sulphur atoms of the engineered disulphide bond is 2.03 Å, and the space between the β‐carbon atoms is 4.61 Å in contrast with 4.73 Å within the authentic 1BD2 construction.
The foundation‐imply‐sq. deviation (r.m.s.d.) between the fixed domains of the 1BD2 construction and the 1G4 dsTCR construction for the spine carbon atoms, was calculated as a way to point out how a lot the native construction of the TCR had been disturbed by the introduction of the engineered disulphide bond (Desk III). Though there are vital variations between the 2 buildings, a lot of this distinction might be as a result of pure flexibility of the TCR molecule. This interpretation is borne out by the remark that the area across the launched disulphide bond exhibits a decrease r.m.s.d. than the fixed domains as a complete (Desk III). This suggests that there’s little or no native disturbance of the pure construction across the launched disulphide bond.
Dialogue – “t protein”
The investigation of TCRs has been hampered by difficulties in making soluble variations of those cell‐floor molecules. The intrinsic instability of the α/β heterodimer has made it troublesome to provide in a soluble truncated kind, besides in a really restricted variety of instances. So as to stabilize the α/β heterodimeric pairing, quite a few approaches have been reported. Formation of single‐chain TCRs, analogous to scFv antibody fragments, has had restricted success regardless of quite a few makes an attempt (Chung et al., 1994; Schodin et al., 1996; Khandekar et al., 1997; Plaksin et al., 1997). For instance, the Vα‐linker‐VβCβ single‐chain TCR assemble yields totally useful A6 TCR however not 1G4 TCR, when produced by in vitro refolding from E.coli inclusion our bodies (P.Molloy, unpublished outcomes).
A extra profitable strategy has been to provide soluble TCR through which the extracellular domains are stabilized within the heterodimer by a jun/fos coiled‐coil area (Willcox et al., 1999b). Though this assemble permits manufacturing of all kinds of antigen‐particular soluble TCRs (Avidex, unpublished outcomes), the versatile fusion to the coiled‐coil domains makes the assemble usually troublesome to crystallize for structural evaluation. The presence of a non‐native fusion area would additionally make this assemble unsuitable for in vivo therapeutic functions.
The dsTCR described right here offers an answer to each of those issues. It’s a utterly globular construction which favours formation of properly ordered protein crystals. The distinction between the dsTCR and the pure TCR is minimal, consisting of two mutations to cysteine which might be buried within the inaccessible interface between the α‐ and β‐chain fixed domains. The dsTCR ought to due to this fact additionally current a minimal danger of antigenicity and immunogenicity in vivo.
This technique of engineering soluble TCRs has been examined in quite a few totally different TCRs particular for various peptide–HLAs. Right here we current information on three totally different dsTCRs particular for sophistication I HLA–peptide antigens. All have proven particular binding exercise to their cognate peptide–HLA advanced in a BIAcore™ SPR assay. The antigen binding domains of the jun/fos fusion TCRs and the dsTCRs are unlikely to have been affected by the engineering concerned in producing the soluble proteins. Certainly, the affinity measurements for dsTCRs evaluate intently with these obtained with jun/fos fusion TCRs (Desk II).
The globular construction of the dsTCRs makes them usually amenable for crystallization research. The 1G4 dsTCR was readily crystallized which enabled us to substantiate the presence of the engineered disulphide bond. Additional dsTCRs and dsTCR–pHLA complexes have been crystallized, and we’re at present within the means of gathering information for these crystals. dsTCRs ought to present a dependable software for producing a basic understanding of the molecular particulars of antigen recognition by T cells. Moreover, dsTCRs might present the premise for producing focused therapeutic brokers retaining the specificity of human T cells for peptide–HLA antigens. This might make a lot of new antigen targets, significantly these derived from intracellular proteins, accessible to ‘monoclonal’ protein remedy.
Acknowledgements
We wish to thank Andrew Johnson for conducting the analysis of stability, Corneli van der Walt for technical help working SDS–PAGE, and Nikolai Lissin and Peter Molloy for permitting inclusion of unpublished information.
Be aware added in proof
We have now just lately engineered increased affinity TCRs, based mostly on the disulphide‐linked TCRs reported herein, with KDs as little as 2.5 nM, that are being developed as therapeutic concentrating on brokers.
“t protein”