Peptide exchange–competent class I MHC molecules produced in eukaryotic cells for rapid production of MHC multimers

Abstract

Class I MHC peptide (MHC-Ip) multimers are well-established reagents that detect antigen-specific T cells. The classical method for production of MHC-Ip multimers begins with the expression of MHC heavy chains (HCs) and β₂-microglobulin (β₂m) subunits as inclusion bodies in Escherichia coli and is followed by denaturant solubilization, in vitro folding in the presence of a defined peptide ligand, and purification by size exclusion chromatography. This protocol is labor intensive, difficult to scale, and represents a significant bottleneck in application of the technology. Herein, we present a novel method for the expression in eukaryotic cells of secreted peptide exchange–competent class I MHC proteins in their native conformation. In this method, expression constructs are engineered as bimolecular complexes composed of an MHC HC and a β₂m molecule covalently linked at its amino terminus to an MHC-binding peptide through a flexible peptide linker containing a defined protease site. Upon proteolysis, the original peptide occupant of the MHC binding site dissociates and is easily replaced with a synthetic peptide. When leucine zippers are added to the carboxyl terminus of each subunit, protease cleavage of the linker results in a stable HC/β₂m complex that can be isolated and stored for subsequent peptide loading. Using this method, we have produced homogeneous MHC-Ip complexes for 25 class I MHC alleles and demonstrated that tetramers produced in this way are equivalent to conventionally produced tetramers for T-cell staining.

Introduction

CD8 T cells express antigen receptors that recognize class I MHC (MHC-I)/peptide complexes, initiating a signaling cascade that results in effector functions such as cytokine production or target-cell cytolysis mediated by the perforin/granzyme pathway. CD8 T-cell responses are now customarily assayed at the single-cell level, either using functional assays such as cytokine flow cytometry, the activation-induced markers assay, or the cytokine enzyme-linked immunospot assay, or by antigen-binding methods using fluorescently tagged MHC-I peptide (MHC‑Ip) multimers, most often in the form of MHC tetramers arrayed around a streptavidin core.¹ In the tetramer assay, the production of MHC molecules containing a single, homogeneous peptide ligand is essential to achieve the avidity effect necessary for stable binding of the reagents to T cells. The classical method for production of MHC tetramers starts with denaturant solubilized MHC heavy chains (HCs) and β₂-microglobulin (β₂m) expressed in the form of inclusion bodies in Escherichia coli that are then refolded in the presence of a single synthetic peptide.² Separate folding reactions (and subsequent purifications) must be performed for each individual reagent, and this constitutes a significant bottleneck in application of the technology.

Methods for MHC-Ip exchange

The original method for production of MHC-Ip complexes for tetramers required separate in vitro folding reactions and subsequent processing and purification of each complex. Because this process is laborious, there has been sustained interest in addressing this production bottleneck, mostly through development of methods to promote peptide exchange. These methods are summarized in Table 1.^3–10 All these methods still use in vitro folding of MHC-I HCs with β₂m (and sometimes a “placeholder” ligand), and these methods remain difficult to perform at larger scales.

Table 1.

Summary of methods for production of MHC-Ip complexes by peptide exchange.

Approach	Year	Reference	Tetramer test
Photo-cleavable placeholder ligand	2006	Toebes et al.3	Yes
One-pot, mix-and-read	2008	Leisner et al.4	No
Periodate-cleavable ligand	2009	Rodenko et al.5	Yes
Dipeptide-catalyzed exchange	2015	Saini et al.6	Yes
Thermolabile complexes	2018	Luimstra et al.7	Yes
Disulfide-stabilized empty MHC-I	2019	Saini et al.8	Yes
Acid-catalyzed exchange from disulfide stabilized empty MHC-I	2022	Saikia et al.9	No
Tapasin-catalyzed exchange	2023	Papadaki et al.10	No

Single-chain trimers

In a separate series of developments, a number of laboratories have demonstrated it is possible to tether MHC-I binding peptides to the amino terminus of β₂m and that bimolecular complexes (or single-chain constructs) can be assembled both in vitro and in vivo into complexes that have native structures that are recognized by cognate T cells.^11–18 This approach seems to violate the rule that peptides binding to wild-type MHC-I cannot generally be extended at either the amino- or carboxy-termini, but the energetic cost associated with extending a peptide at its carboxy terminus is apparently compensated by the advantage of covalent attachment to β₂m. Covalent attachment of peptides to MHC molecules was first described, in fact, for class II MHC (MHC-II)¹⁹ and, in a subsequent development, protease sites were introduced in the tethers to binding peptide ligands, enabling peptide exchange after proteolysis.^20,21

The Epitope-linked β₂m MHC-I proteins design

Here, we demonstrate that proteolysis of sites engineered into the tether region of MHC-I complexes with epitopes linked to β₂m can be used to efficiently trigger peptide exchange in a manner analogous to the photolytic and chemical exchange triggers in Schumacher’s conditional peptide ligands (Table 1). In our reagents—which we refer to as epitope-linked β₂m MHC-I proteins (ELBMs)—proteolysis-initiated peptide exchange occurs because the cleaved peptide fragment is significantly longer than an optimal binding core sequence, but it no longer retains the compensatory advantage of covalent linkage to β₂m and therefore dissociates from the MHC molecule (Fig. 1). In the original MHC-Ip exchange protocols, triggered dissociation of the placeholder ligand was performed in the presence of so-called rescue peptides because the newly empty MHC proteins are unstable and aggregate in the absence of a peptide ligand.²² Here, we show that addition of leucine zippers to the class I protein subunits—an approach that is nearly universal for production of MHC-IIp multimer reagents but never before, to our knowledge, used for classical MHC-I proteins—stabilizes the protease-cleaved intermediates, permitting isolation and storage of intermediates for loading at a future time. Using this method, homogenous MHC‑Ip complexes can be produced using efficient eukaryotic protein expression systems. MHC tetramers prepared by this protocol are nearly indistinguishable from corresponding conventionally prepared tetramers in T-cell staining assays.

Figure 1.

The ELBM reaction scheme. The method starts with a purified bimolecular complex consisting of an MHC-I HC and a β2m light chain with an appropriate placeholder peptide ligand fused to the amino terminus of β2m; in this case, each subunit is fused to a leucine zipper that has charge complementarity to the zipper on the alternate subunit (the zippers are referred to as aZip for the acidic leucine zipper and bZip for the basic leucine zipper). In step A, the 3C20 linker is cleaved by the Human rhinovirus-3C site-specific protease in the presence of a synthetic exchange peptide ligand. In step B, the formerly tethered peptide dissociates from the peptide binding site, leaving a transiently empty MHC protein (dissociation is promoted by the C-terminal extension on the formerly tethered peptide relative to the optimal binding ligand). In step C, the synthetic peptide ligand, usually with an optimal binding sequence, binds to the empty MHC molecule, forming a stable target complex.

Materials and methods

Materials

The hybridoma for the production of the anti-Velcro leucine zipper monoclonal antibody (mAb) 2H11(23) was a gift from Dr. Ellis Reinherz (Dana-Farber Cancer Institute). The 3C protease was produced in-house as a glutathione-S-transferase (GST) fusion using a plasmid that was a gift from Kai Wucherpfennig (Dana-Farber Cancer Institute). The PreScission Protease from Cytiva (catalog no. 27084301) is available as an equivalent product. The protease cleavage buffer was 150 mM sodium chloride, 50 mM Tris HCl pH 7.5. For the media designations, D1b, D10b, and D20b refer to DMEM supplemented with the indicated amount of FBS (1%, 10%, and 20%) and with 50 µM biotin. Tetramers were prepared with streptavidin APC (Agilent, PJ27S-1) or Streptavidin PE (Agilent, PJRS25) as described elsewhere.² Reagents used in flow cytometry panels are listed in Table 2.

Table 2.

Antibodies used in flow cytometry experiments.

Table of antibodies used in flow cytometry
Specificity	Supplier	Catalog number
Mouse CD4 BUV395	BD	563790
Mouse CD8 FITC	BD	553031
Mouse CD8 APC	eBioscience	17-0081-82
Mouse CD19 BB700	BD	566411
Mouse CD11b BB700	BD	566416
Mouse CD44 PE-Cy7	BD	560569
Human CD3 BV421	BD	563798
Human CD3 BV510	BioLegend	317332
Human CD4 BUV805	BD	612887
Human CD8 APC-H7	BD	560179
Human CD8 BV570	BioLegend	301038
Human CD14 PE-Cy7	BD	562698
Human CD19 PE-Cy7	BioLegend	302216
Human CD45RA Alexa 700	Biolegend	304120

Peptides

All synthetic peptides were obtained from Genscript.

E. coli expression vectors

The design of the expression cassette is shown in Fig. 2A. The coding sequences were cloned into pET24a⁺ (Novagen/EMD Millipore) and vectors were transformed into BL21-Gold (DE3) Gold competent cells. Production of inclusion bodies and refolding reactions were performed as described.²

Figure 2.

Design of constructs for expression of ELBM proteins in E. coli and in eukaryotic cells. (A) For expression in E. coli, the constructs follow the design of classical MHC-I tetramers.² The β₂m subunit has an N-terminal extension consisting of an optimal peptide ligand (relative to the HC allele), followed by a flexible linker sequence containing a cleavage site for the 3C protease (GSGGSGGSGLEVLFQGPGSG); the 3C recognition site in the linker sequence is underlined. (B) For expression in 293T cells, the constructs include a secretion signal sequence on each subunit. The HC is extended at the carboxy terminus by a pair of short, flexible linkers (labeled 1 and 2) flanking a basic leucine zipper,²⁸ followed by the BSP85 peptide,²³ followed by either a His₆ tag or a tandem (His₆)₂ tag.²⁸

Lentiviral vectors and cell lines

All steps for production of stable lines for expression of ELBM proteins—including construction of lentiviral vectors, production of lentivirus stocks, transduction of the 293 T.TPM expression cell line, and optional sorting for high-level expression based upon vector-encoded marker proteins—were as described by Willis et al.²³ All constructs contain a modified signal sequence (SS) from human ceruloplasmin with the penultimate 2 amino acids modified by the incorporation of an SbfI site (MKILILGIFLFLCSTPAWA → MKILILGIFLFLCSTPLQA). This provides a 5′ restriction site, incorporates a very short SS, has no predicted effect on processing by signal peptidase (SignalP 4.1 Server), and has no noticeable effect on protein production (our own observations). The 3′ restriction site for the HCs is SpeI (encoding “TS” as part of a short linker at the end of the extracellular domain). The 3′ restriction site for the light chain is BamHI (encoding the terminal “GS” at the end of the linker 1 (GGGGSGGGGSGGGGS in Fig. 2B). The various alleles described were generated either by polymerase chain reaction from cDNA or as synthetic fragments (Integrated DNA Technologies), and all constructs were verified by direct sequencing (Macrogen USA). The overall design of the expression cassette is shown in Fig. 2B.

Production culture

For protein production, cell lines were expanded in T225 flasks to confluency and loaded into the lower chamber of a CELLine 1000 flask (DWK Life Sciences) in 25 mL of D20b medium. The upper chamber was loaded with 1 L of D1b medium and the flasks were then cultured at 37 °C in a 5% CO₂ incubator. After 7 d, the upper chamber was emptied and ∼35 mL were harvested from the lower chamber, leaving 5 mL of cell suspension to seed a second round of production. This cycle was repeated typically for 5 to 15 wk. In some cases, expression levels increased when production cultures were grown at 30 °C to 32 °C instead of the customary 37 °C, though the reduced temperature did limit the achievable number of passages.

Enzyme-linked immunosorbent assays

Expression levels were measured by enzyme-linked immunosorbent assay (ELISA) of tissue culture supernatants as described previously.²³ The sandwich ELISA assay used the 2H11 anti-leucine zipper antibody for capture and commercially available anti–His6-HRP conjugates (Abcam ab1187) as a readout.

Purification

MHC proteins from culture supernatants were purified on either 2H11 mAb columns or immobilized metal affinity chromatography columns as described elsewhere.²³

Fluorescence polarization

Fluorescence polarization (FP) assays were performed on a BioTek Synergy H1 hybrid reader with filter settings suitable for FITC excitation at 485 nm and emission at 528 nm. Each well was to set to read at 6-min intervals with automatic instrument measurements of 8 readings/well. For a standard assay, each well of a black 384-well microplate (Corning 3575) was loaded with 40 µL of master mix containing labeled peptide specific for each MHC allele (final concentration 32 nM), protease cleavage buffer, 3C protease, and water. For measurements in the absence of protease, 3C protease was omitted. All reactions were performed in duplicate wells. Prior to adding individual MHC proteins to the reaction mix, microplates were prewarmed at 37 °C and preread for 5 min to reduce temperature-induced fluctuations. Peptide exchange was initiated by adding 10 µL of each of the diluted MHC monomers (150–300 nM) and microplates were then sealed with ThermalSeal RTS sealing film. Plates containing a 50 µL final volume were read on kinetic mode for a specified duration at 37 °C. For each assay, control samples (protein only, labeled peptide only, and buffer only) were also included for background corrections. Assays were also performed with negative control labeled peptides that do not bind to the specific MHC allele. The selection of fluorophore-labeled peptide ligands was generally based upon examination of published X-ray crystal structures in which a residue from the peptide ligand with minimal contact with the MHC protein was replaced with a lysine that was modified with FITC during peptide synthesis.

Exchange protocols

Preparative peptide exchange reactions were performed in a volume of 1 mL. First, a master mix containing 100 µL of 10X protease cleavage buffer (1.5 M sodium chloride, 0.5 M Tris HCl, pH 7.5), 5 µL of 3C protease at 2 mg/mL, and 775 µL of water was added to a 1.5 mL Eppendorf tube. Then 20 µL of a specific peptide (10 mg/mL stocks in 100% dimethylsulfoxide) was added to the master mix and vortexed gently. Peptide exchange was initiated by adding 100 µL of each MHC protein to the peptide mix, gently pipetted up and down to mix the reagents, and the reactions were incubated at 25 °C for 12 to 15 h. For mock exchange reactions, peptide was omitted and equal volume of dimethylsulfoxide was added. After the specified incubation time, reactions were spun down at for 15 min at maximum speed in a refrigerated microcentrifuge to remove aggregates, and the supernatants were concentrated in 10k MWCO Amicon ultracentrifugal filters, followed by buffer exchange into phosphate-buffered saline (PBS) buffer with 2 mM ethylenediamine tetra-acetic acid and 1 µg/mL pepstatin and leupeptin. The ultraviolet absorbances of the samples were then measured at 280 nm using a Nanodrop spectrophotometer, and protein concentrations were estimated based on the extinction coefficient of each MHC protein. The concentrations were then adjusted to 2 mg/mL and samples were snap frozen in liquid nitrogen prior to storage at −80°C. For later experiments, ELBM proteins were cleaved with 3C protease in the absence of exchange peptides, followed by removal of the GST-tagged 3C protease, as described by Willis et al.²³

Isoelectric focusing gels

Analysis of peptide exchange by vertical native isoelectric focusing gel electrophoresis (IEF) was as described by Willis et al.²³

Analysis of peptide loading by matrix-assisted laser desorption/ionization-time-of-flight

Loading of the MHC proteins with synthetic peptides was analyzed by matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF), as described previously.²³

Protein stability assays

Protein melting and aggregation temperature measurements were performed on the Uncle instrument from Unchained Labs, following manufacturer’s instructions. Briefly, full-spectrum autofluorescence and side light scattering were measured over temperatures ranging from 15 °C to 95 °C at a ramp rate of 0.5 °C/min at 0.3 mg/mL protein concentration in PBS buffer with protease inhibitors. Using the Uncle software, data were transformed to the barycentric mean and plotted as a function of temperature before calculation of the melting temperature (Tm) and aggregation temperature.

Tetramer staining and flow cytometry

Cell staining and data acquisition were performed as described elsewhere² and reiterated here; antibody panels for staining are summarized in Table 2. Single-cell suspensions of mouse spleen cells were prepared using nylon screens and red blood cells were removed with BD PharmLyse. Cells were resuspended in staining buffer (PBS, 2% bovine serum albumin) and stained with tetramer for 1 h at room temperature. Thermo Fisher LIVE/DEAD Fixable Aqua (catalog no. L34957) and BD Biosciences Brilliant Stain Buffer Plus (catalog no. 566385) were used according to the manufacturers’ instructions. Other surface stains (described in the legend for Fig. 8) were added without washing and cells were incubated 20 min on ice. Samples were fixed with BD Stabilizing Fixative (catalog no. 664907) (10 min, room temperature), washed, and resuspended in staining buffer. For human samples, fresh peripheral blood from an HLA-A*02:01⁺ CMV⁺ donor was stained with tetramer for 30 min at room temperature, followed by an additional 30 min with other surface stains. Cells were fixed and red blood cells lysed with BD FACSLyse according to the provided instructions. Mouse and human samples were analyzed on a BD FACS Symphony cytometer and data were plotted using FlowJo version 10. Directly conjugated mAbs were from BD Biosciences or BioLegend.

Figure 8.

Comparison of staining by tetramers compared by the ELBM and conventional methods. Mouse or human cells were stained with tetramers produced by conventional E. coli refolding techniques or ELBM exchange, as described in Materials and Methods. Uncut or mock-exchanged (i.e., no peptide) ELBM tetramer staining is also shown for each allele. Gate statistics on the dot plots show percentages of cells (black on the first line in each panel) and mean fluorescence intensity (red on the second line in each panel). (A) Kb LCMV gp 34–41 (AVYNFATM) APC tetramers were used to stain spleen cells from lymphocytic choriomeningitis virus (LCMV) Armstrong–infected B6 mice (day 20). Cells were gated on live CD8 FITC⁺ CD4 BUV395⁻ CD19 BB700⁻ CD11b BB700⁻ lymphocytes. CD44 PE-Cy7 is plotted against tetramer. (B) Db LCMV gp 33–41 (KAVYNFATM) APC tetramers were used to stain spleen cells from LCMV Armstrong infected B6 mice (day 110). Cells are gated on live CD8 FITC+ CD4 BUV395⁻ CD19 BB700⁻ CD11b BB700⁻ lymphocytes. CD44 PE-Cy7 is plotted against tetramer. (C) Kd respiratory syncytial virus (RSV) M2 82–90 (SYIGSINNI) APC tetramers were used to stain spleen cells from RSV-infected BALB/c mice (day 8). Cells were gated on live CD8 APC⁺ CD4 BUV395⁻ CD19 BB700⁻ CD11b BB700⁻ lymphocytes. CD44 PE-Cy7 is plotted against tetramer. (D) HLA-A*02:01 CMV pp65 495–503 (NLVPMVATV) phycoerythrin (PE) tetramers were used to stain human peripheral blood from an HLA-A*02:01⁺ CMV⁺ donor. Cells were gated on CD3 BV421⁺ CD4 BUV805⁻ CD14 PE-Cy7⁻ CD19 PE-Cy7⁻ lymphocytes. CD8 APC-H7 is plotted against tetramer.

Results

Prototypes

At the outset of our development of the ELBM system, we built a small set of prototypes using the conventional E. coli–based expression system for MHC-I proteins.^2,24 The design of the ELBM subunit begins with an allele-specific peptide ligand, followed by a linker sequence containing a protease cleavage site, followed by the sequence of mature β₂m (Fig. 2A). For expression of subunits in inclusion bodies in the bacterial cytoplasm, this required us to apply and verify the rules governing removal of N-terminal methionine residues by E. coli, which are largely determined by the size of the peptide residue at position 2 of the translated protein.²⁵ For example, for HLA-A*02:01, we chose the peptide MLQEKPFKL from cottontail rabbit papilloma virus,²⁶ and N-terminal sequencing by Edman degradation confirmed that its initiator methionine was intact (data not shown). In contrast, for H-2Kb we chose the SIINFEKL peptide as a candidate for our ELBM construct, and even though the coding sequence began as MSIINFEKL; Edman degradation confirmed efficient removal of the initiator methionine. During prototype development, we explored a variety of linker lengths and compared thrombin and HRV-3C protease cleavage sites, concluding that the sequence GGSGLEVLFQ↓GPGGM, which we refer to as 3C-15, provided satisfactory results, where the underlined sequence corresponds to the recognition site for the human rhinovirus 3C protease with cleavage between the Q and G amino acids as indicated.²⁷ Each ELBM subunit (peptide-b2m) was folded with its corresponding HC as described.² The full set of constructs for E. coli expression that we made and successfully tested are listed in Table S1. We carried many of these constructs through coupled protease cleavage/peptide exchange and tetramer testing, but for further development, we switched focus to a eukaryotic expression system described in the remainder of this work.

Design: 293T-expressed ELBM proteins

With 2 recent exceptions using single-chain trimer constructs,^17,28 eukaryotic cell expression systems have generally not been used in MHC-I tetramer production pipelines because class I molecules expressed in mammalian cells are generally loaded with a diverse spectrum of endogenous peptides with a wide range of dissociation rates, preventing production of homogeneous complexes required for MHC multimer applications. It occurred to us that expression of MHC-I proteins with ELBM constructs might provide homogeneous complexes by out-competing the endogenous peptides and that these could be used for exchange reactions that would, themselves, yield homogeneous MHC‑Ip complexes. For this work, we adopted a lentiviral expression system in a derivative of 293 T cells that we routinely use for MHC-II expression.²³ For optimal system performance, we engineered a 293 T-cell derivative with inactivated genes coding for TAP2 and MGAT1 and that were transduced with an expression construct for a secreted version of BirA. We chose to knockout TAP2 to minimize the availability of endogenous peptides available in the endoplasmic reticulum that might compete for binding to the class I peptide binding site with the desired linked peptide from the ELBM construct. In addition, we chose to knockout MGAT1,²⁹ an enzyme required to produce complex N-glycans, to reduce the charge heterogeneity of the proteins produced in the knockout cells, permitting the use of native IEFs to monitor peptide exchange. The secreted version of BirA typically provides greater than 98% biotinylation of the desired product when the cell culture medium is supplemented with 50 µM biotin.³⁰ We refer to this expression line as 293 T.TPM.

Design: lentiviral vector expression cassettes

The lentiviral vector expression cassettes for the HC and ELBM subunits are modified from those we use in the E. coli system and are shown in Fig. 2B. To optimize the linker between the peptide epitope and the β₂m subunit for expression, cleavability, and exchange, we performed a more extensive series of experiments than for the E. coli–expressed prototypes described at the beginning of Results. Because we had seen off-target cleavage of the HC of some alleles when we used a thrombin-cleavable linker, we used a 3C-cleavable site in all current generation constructs (we have never seen off-target cleavage with 3C). For 2 alleles, HLA‑A*02:01 and HLA‑B*57:01, we tested a variety of linker lengths from 14 to 22 amino acids and found that a 20 amino acid linker we refer to as 3C20 gave optimal performance (Fig. S1). Complementary leucine zippers,³¹ commonly used for expression of MHC-II proteins, were added to the carboxy termini of the HC and β₂m subunits to ensure the purified product contained the recombinant but not the endogenous or serum-derived β₂m and to add stability to the protein during purification and peptide exchange. In addition, the zippers provide an affinity handle for ELISA assays and for purification. (For a few alleles, we also prepared versions without zippers, but the expression levels for these constructs was significantly lower, and we abandoned the zipperless constructs.) Flexible linkers were used to permit alignment of the heterodimeric zippers with the termini of the MHC subunits. A His₆ tag and, in some later constructs, a tandem His₆ tag³² was added as an option for affinity purification. Finally, all expression constructs use an optimized hybrid secretion signal sequence that takes the first 15 amino acids from the human ceruloplasmin protein followed by the amino acids leucine, glutamine, and alanine “LQA,” permitting inclusion of a unique restriction site for Sbf I without changing the SignalP-predicted efficiency of signal sequence cleavage.³³

All sequences were produced using synthetic DNA with codons optimized for expression in human cells. The HC and β₂m subunits were subcloned into separate lentiviral vectors, each with a different marker gene downstream of an internal ribosome entry site, allowing sorting of cell lines transduced with both vectors. We have generally found the dual-vector approach gives superior protein yields compared with a single-vector approach in which the subunits are separated by a 2A ribosomal skip sequence³⁴ (data not shown).

Design: source of peptides

The linked peptides for ELBM constructs for each HC allele were chosen in a variety of ways. Many correspond to well-known T-cell epitopes. Some were picked somewhat at random from searches of the Immune Epitope Database (IEDB) (http://www.iedb.org)³⁵ for peptides binding to their MHC restricting allele with relatively high affinity, which we assumed would correlate with expression levels. Once stable-expression cell lines were established, expression levels varied from allele to allele (Table 3) and within the same allele across a spectrum of tethered peptides (data not shown). After purification of the HC/ELBM heterodimers by affinity chromatography as described in Materials and Methods, exchange competence was tested as we describe shortly (with and without 3C protease cleavage of the linker). A subset of the complexes produced by exchange were tested in tetramer staining assays.

Table 3.

Class I MHC alleles currently provided by ELBM production.

Allele	Tethered peptide	Expression level (mg/L)	Positive control exchange peptide (by IEF)	FP-labeled peptide a
A*01:01	LTEEFYHSY	1.8	WMDSTKATRY
A*02:01	MLQEKPFQL	11.0	GLSRYVARL	TLAxFVLAAV
A*03:01	AIFQSSMTK	4.5	RLRPGGKKK
A*11:01	AIFQSSMTK	5.2	RQLANAIFK
A*24:02	SWPKFAVPNL	20.0	KYKLKHVW
A*68:01	TTYMDTFFR	18.0	HNIRGRWTGR
A*68:02	FVFLVLAGR	4.0	FVFLVLAGR
B*07:02	APRTLVYLL	9.5	IPRRIRQGL
B*08:01	YMRERFEPM	7.8	RAKFKQLL
B*15:01	FMYALSRAF	1.4	FSFKKCLVY
B*18:01	FEYGGFPPA	3.0	DEFKPIVQY
B*27:05	ARAFIESHY	14.0	KRWIILGLNK	KRWIIxGLNK
B*35:01	LPYPQPQPF	25.0	IPIILKALY
B*35:02	VPLRPMTY	3.8	FPVRPQVPL
B*40:01	LEDGIYGIF	2.2	YENAFLPFTL
B*44:02	EEILGTVSW	1.7	SENEVKLTI
B*55:02	KPWDVLPMV	11.0	QPEKGGRKPA
B*57:01	LSSPVTKSF	5.0	ISPRTLNAW	LSSPVTKxF
B*57:03	ISPRTLNAW	32.0	KAFSPEVIPMF	KAFSPxVIPMF
B*73:01	NRYGVAYVY	4.2	RRFWFRLRP
Kb	SIYQYVRL	19.0	RSYRYKLL	xAPGNYPAL
Kd	TYQRTRALV	7.0	TYQRTRALV	TYQxTRALV
Db	ASNENMETM	3.4	KCSRNRQYL	SSLxNFRAYV
Mamu-A*01	STPRPLRL	11.0	STPESANL
Mamu-B*08	TRALVRTGM	3.4	RRDNRRGL

^aThe letter x in the peptide symbols stands for fluorescein-labeled lysine.

FP assay

For many alleles, we adopted an FP assay³⁶ to measure peptide exchange with and without cleavage of the linker by the 3C protease (cleavage was verified by a shift of the β₂m subunit band on sodium dodecylsulfate–polyacrylamide gel electrophoresis [SDS-PAGE]). Our interests were almost wholly in qualitative evidence for peptide binding and whether binding is relatively fast. Furthermore, the change in FP over time is due to a complex combination of the kinetics of cleavage of the ELBM linker by the 3C protease, the dissociation rate of the formerly tethered peptide, and the association and dissociation rates of the labeled peptide to the newly emptied MHC peptide binding site.

Representative data are shown for 4 alleles in Fig. 3. Figure 3A, which shows the exchange rates for 3 versions of HLA-A*02:01 with 3 different peptide ligands tethered to β₂m, demonstrates the basic patterns that are seen across all alleles. For 2 of the ligands (FL9 and YI9), exchange is only seen in the presence of the 3C protease. For 1 of the ligands (ML9), exchange is seen even in the absence of the 3C protease, but the exchange rate is much faster in the presence of the 3C protease. This suggests that for the tethered ML9 ligand, there is a significant dynamic motion in which the ligand dissociates from the peptide binding pocket while remaining tethered to β₂m, and that the binding rate of the labeled peptide is competitive with the re-association of the still-tethered ligand. In the presence of 3C protease, the apparent exchange rates vary with the peptide sequence, suggesting differences in the dissociation rates of the different peptides, which is likely due to some combination of the breathing motion of the tethered peptides and variation in post-cleavage dissociation rates; it is important to note that after cleavage, the indicated peptides all have C-terminal extensions from the portion of the 3C20 linker that is N-terminal to the 3C cleavage site, as indicated in Fig. 1. Cleavage of the linker does not require the presence of a competitor peptide, nor does it appear to be significantly accelerated by it (data not shown). The same basic patterns seen for the HLA-A*02:01 constructs are seen for all alleles tested, with representative data shown for individual ELBM constructs in association with H2-K^d, H2-K^b, and H2-D^b (Fig. 3B–D).

Figure 3.

Kinetic analysis of combined linker cleavage and peptide binding by FP. The combined kinetics of 3C-mediated linker cleavage and peptide binding were analyzed by FP assays as described in Materials and Methods. In each panel, samples including the 3C protease are shown as a solid line; those omitting the 3C protease are shown as dotted lines. In the following, “x” indicates fluorescein-modified lysine. (A) Binding of TLAxFVLAAV to 3 HLA-A*02:01 proteins with β₂m tethered to MLQEKPFQL (ML9), FIALWIPDL (FL9), or YLMPYSVYI (YI9). (B) Binding of TYQxTRALV to H2-K^d with β₂m tethered to TYQRTRALV (TV9) or KYLPLDKGI (KI9). (C) Binding of SSLxNFRAYV to H2-K^b tethered to SIYQYVRL (SL8). (D) Binding of SSLxNFRAYV to H2-D^b tethered to ASNENMETM (AM9).

Native IEF for peptide exchange

Although FP assays are useful for establishing the exchange competence of a particular MHC-ELBM combination, they are not useful for monitoring the loading of unlabeled T-cell epitopes suitable for MHC tetramer preparation. For this, we turned to native IEF, which can resolve single charge differences in MHC/peptide complexes.³⁷ In contrast to the FP assays, which are performed with an excess of MHC to the labeled peptide,³⁸ preparative loading reactions are performed with a 15- to 20-fold molar excess of exchange peptide to HC-ELBM. In general, loading reactions are performed overnight at 25°C at near-neutral pH.

For Fig. 4A and B, we used data in the IEDB to choose a panel of H2-K^b and HLA-A*02:01 peptide ligands that were all high affinity but that had distinct overall net charges. We compared uncleaved, mock exchanged (3C cleaved in the absence of an exchange peptide), and exchanged complexes. For each sample, multiple bands are seen even though the proteins were expressed in MGAT1^−/− cells (the mutation limits charge heterogeneity). The same banding pattern is seen in each of the exchanged complexes, but the entire pattern shifts according to the net charge of the exchange peptide. This suggests that the multiple bands seen for each sample are not due to binding of endogenous peptides with a range of net charges and, more importantly, the shifts of the entire set of bands suggest, in each case, that replacement of the formerly tethered peptide with the indicated synthetic peptide is essentially complete. Furthermore, the HC-ELBM construct used for these assays in Fig. 4B was the HLA-A*02:01 binary complex with the ML9-ELBM subunit that, in FP assays, showed peptide exchange even in the absence of the 3C protease (Fig. 3A). The fact that complete exchange was seen with this protein suggests it did not bind to high-affinity endogenous peptides during expression cultures; this could be, in part, a benefit of using a TAP2^−/− line for expression. Finally, it is important to note that the IEF method has a blind spot when the exchange peptide has a net charge of 0, and the exchanged complex is indistinguishable on IEF from the mock cleaved sample.

Figure 4.

Native IEF gels document peptide exchange into 3C-protease cleaved ELBM constructs. Protein bands shift according to the net charge of the exchange peptides for H2-K^b (A) and HLA-A*02:01 (B). (A) The peptides are RSYRYKLL (+3), SSYRRPVGI (+2), RGYVYQGL (+1), SIINFEKL (0), and EIINFEKL (−1). (B) The peptides are YLLPRRGPRL (+3), GLSRYVARL (+2), FLLTRILTI (+1), FLLPLTSLV (0), and FLPSDFFPSV (−1). More details on the peptides are found in Table S2. (C) IEF gel shifts for a panel of 6 MHC-I alleles upon exchange with allele-specific peptides, each with a net charge of +3. The peptides are A*02:01-KLVSQGIRKV, A*03:01-RVLSFIKGTK, D^b -KCSRNRQYL, K^b -RSYRYKLL, B*07:02-IPRRIRQGL, and B*35:01-HPGAGKTKRY.

Figure 4C shows a panel of 6 class I alleles, each exchanged with a peptide ligand for that allele chosen from IEDB that has a net +3 charge. For some of the alleles, there is a notable difference in the native IEF mobility between the uncleaved and the mock-exchanged samples, with the largest difference seen for H2-D^b. The H2-D^b expression construct used for these experiments has β₂m tethered to the influenza A NP366–374 peptide ligand, ASNENMETM, and, after cleavage, retains the linker sequence GSGGSGGSGLEVLFQ, such that the overall net charge of the formerly tethered peptide is −3. The observed shift of the mock-exchanged sample is consistent with loss of the negatively charged, formerly tethered peptide, suggesting that the empty protein is stable under conditions of the assay.

Beyond the 6 alleles shown in Fig. 4C, we have designed, expressed, and verified peptide exchange by native IEF (and, in some cases, by FP) for a total of 25 alleles, which are listed in Table 3. Table 3 also indicates the expression levels for each of the established production lines, using an ELISA assay (described in Materials and Methods) to measure products in the supernatants of cells cultured at confluence in 6-well plates.

MALDI-TOF analysis of peptide exchange

One major drawback of the native IEF protocol for analysis of peptide exchange is that it is uninformative for peptide ligands that have a net zero charge. To address this deficiency (and to develop a potentially simpler and higher-throughput assay), we used MALDI-TOF mass spectrometry to characterize bound peptides before and after exchange. Because we did not expect our MHC proteins, at any stage of the protocol, to contain more than 1 or 2 distinct peptides, we did not include the standard liquid chromatography separation of peptides and instead opted for a simple protocol in which we eluted peptides at acidic pH, desalted the samples on C18 tips with a 10 µL bed volume, and eluted the bound peptides in methanol prior to spotting on the MALDI matrix. In Fig. 5, we show the analysis of both cleaved and peptide-exchanged products for 4 ELBM constructs: HLA-A*02:01, H-2K^b, H-2D^b, and H-2K^d. In all cases, the formerly tethered peptide is seen in the “cleaved-only” samples, whereas only the exchange peptide is seen in the cleaved-and-exchanged samples.

Figure 5.

MALDI-TOF analysis of peptide exchange. MALDI-TOF analyses were performed, as described in Materials and Methods, for ELBM constructs for HLA-A*02:01, H2-K^b, H2- D^b, and H2-K^d after 3C protease cleavage alone or after 3C-cleavage in the presence of allele-specific peptide ligands and buffer exchange as indicated. M, mock; X, exchange. In some cases, the indicated peptides were detected as sodium ion adducts (A*02:01 M, Kb X, Db M and X, and Kd X). Otherwise, the peptides were detected as H⁺ ions.

Tm analysis of peptide exchange

Ligand binding often increases protein stability. Accordingly, we measured intrinsic fluorescence as a function of temperature for 5 forms of each of 4 MHC alleles: (1) uncut ELBM proteins, (2) linker-cleaved proteins in the absence of exchange peptides, and (3) complexes loaded with 3 distinct allele-specific peptides. The results are presented in Fig. 6. As expected, cleavage of the linker in the absence of an exchange peptides leads to a decrease in Tm for all alleles, with a ΔTM ranging from −3.5 °C for the K^b construct to −12.2 °C for the D^b construct. Also as expected, peptide binding led to an increase in Tm compared with the cleaved form with ΔTm ranging from 3.5 °C for the K^b/TSINFVKI complex to 27.8 °C for the D^b/SSLENFRAYV complex.

Figure 6.

Binding of peptides to cleaved ELBM proteins increases their thermal stability. The intrinsic fluorescence of MHC peptide complexes as indicated was measured as a function of temperature and transformed to the barycentric mean (BCM) using Uncle software, which was also used to calculate the Tm, indicated by the vertical lines in each panel.

These data provide a third independent validation of the efficacy of peptide exchange using the ELBM approach. Some notable limitations of the data are (1) melting of the proteins is not reversible and, above the Tm, the proteins aggregate (measurable by side light scattering; data not shown); (2) although different peptides stabilize an isoform to different extents (e.g., the Tms for the 3 D^b peptides span a range of 21.2 °C), we have not made an effort to interpret the Tms in more than a semiquantitative way, because the specific values are likely to be due to idiosyncrasies of specific constructs; and (3) although the shape of the melting curves gives the impression of complete loading, we have not attempted to model the data to prove this (and we are skeptical that we could do so). Finally, the specific peptide complexes tested were not chosen to permit correlation to binding data available in the IEDB, but more extensive and carefully designed comparisons could be done in the future.

Exchange reactions with precleaved monomers

During the initial stages of development of the ELBM platform, we followed the convention of performing the triggered placeholder ligand dissociation with replacement peptide ligand binding in a single step,²² thereby minimizing the exposure time of transiently empty class I molecules (empty MHC molecules are, at best, metastable and prone to unfolding or aggregation). However, the presence of leucine zippers in our MHC constructs—added in part because they should enhance protein stability during expression—raised the possibility of decoupling cleavage of the linked surrogate ligand from addition of exchange peptides. Uncoupling cleavage from exchange peptide binding has a number of advantages. Most importantly, it enables streamlined removal of the 3C protease from a single precursor reaction, yielding a product that can be used to produce multiple MHC‑Ip complexes that are free of the 3C protease. In Fig. 7, we compare exchange reactions for 2 MHC alleles using (1) a 1-step protocol in which the exchange peptide was included in the protease cleavage reaction; (2) a 2-step protocol in which the ELBM construct was first cleaved by the 3C protease and the exchange peptide was added in a subsequent step; or (3) a modified 2-step protocol in which the cleaved product was frozen at −80 °C and then thawed prior to addition of the exchange peptide. All protocols were equivalent by the criteria of native IEF gel analysis of peptide binding and by equivalent yields (data not shown). The 2-step protocol has worked for all ELBM constructs so far tested; results from a total of 10 alleles are shown in Fig. S2.

Figure 7.

Cleavage of the ELBM linker can be decoupled from the addition of exchange peptides. ELBM constructs for H2-K^b and HLA-B*57:03 were incubated with high-affinity binding (P) or negative control (N) peptides and analyzed by native IEF gel electrophoresis. For the 1 step reactions, ELBM constructs were incubated overnight with peptides in the presence of 3C protease at 25 °C. For the 2-step protocols, ELBM constructs were precleaved with 3C protease in overnight reactions at 4 °C and either incubated overnight with peptides at 25 °C (2 step) or aliquots were snap frozen in liquid nitrogen and stored at −80 °C for 2 d, followed by overnight peptide loading at 25 °C (2-step-freeze/thaw). Positive-binding peptides were RSYRYKLL for K^b and ISPRTLNAW for B*57:03, whereas negative control peptides were KYKLKHIVW for K^b and IPIILKALY for B*57:03. Peptides used for K^b had a net charge of +3, and peptides used for B*57:03 had a net charge of +1. Recoveries from each set of protocols were 85% to 90%.

Tetramer staining

Our primary goal for the ELBM method was to develop a more efficient method for preparation of MHC tetramers. In Fig. 8, we compare the staining efficacy of 4 MHC‑Ip tetramers containing subunits prepared by conventional in vitro folding methods with the ELBM exchange methods described here, including negative controls in which the ELBM linker is uncut and in which it is cut but the MHC is not loaded with the peptide of interest (mock). For these tests, we chose 3 distinct mouse class I alleles and 1 human allele, each loaded with a distinct peptide epitope (or negative controls). For testing of the H‑2D^b and H‑2K^b reagents, we chose immunodominant epitopes from the lymphocytic choriomeningitis virus glycoprotein signal peptide³⁹; for testing of H‑2K^d, we chose an immunodominant epitope from respiratory syncytial virus⁴⁰; and for HLA‑A*02:01, we chose an immunodominant epitope from human cytomegalovirus.⁴¹ In all cases, the CD8⁺ T-cell populations identified by the tetramers prepared by conventional and by ELBM methods were well resolved from the bulk, nonspecific population, and the frequencies of the tetramer-positive populations were comparable. In contrast, the mean fluorescence intensities of tetramer-positive populations defined by the ELBM tetramers were 13% to 38% lower than the conventional tetramers, but baseline separation between the tetramer-positive and tetramer-negative cells was still seen.

As expected, there was no evidence for staining with the uncut and mock tetramers, though we did observe an increase in background staining with the H-2Db mock tetramer. We speculate that this was due to a decrease in the stability of the H-2Db ELBM protein after cleavage of the linker that increases the nonspecific stickiness of the reagent, and we note that this effect is expected to be an idiosyncratic feature of the allele/tethered peptide combination. We also found that the conventional H-2Kb tetramer had higher background staining than the ELBM version. We frequently see higher background stains with conventional H-2Kb reagents that vary from lot to lot, appear to be CD8 mediated, and can be reduced by careful cross-titrations with specific anti–mouse CD8 antibodies,² but we have yet to see this with H-2Kb ELBM reagents.

Finally, to demonstrate the efficacy of ELBM-exchange reagents containing the HLA-B*57:01 protein (an allele that is normally difficult to refold), we obtained peripheral blood mononuclear cells from a donor who had previously received the yellow fever virus 17D vaccine and stained them with a panel of ELBM-exchange reagents containing peptides from the yellow fever virus vaccine or from the influenza A virus NP protein (as well as control reagents), as indicated in Fig. S3. These results demonstrate the ability of ELBM-exchange reagents with an additional human class I allele to identify even low-frequency CD8⁺ T cells and suggest that the approach is highly likely to be completely generalizable.

Discussion

Summary

Herein, we have demonstrated production of peptide exchange–competent MHC-I proteins from stably transduced eukaryotic expression lines; these proteins are suitable for high-throughput production of MHC multimers that are commonly used to detect antigen-specific CD8⁺ T cells. Our method, which we refer to as epitope-linked β₂m, has several advantages compared with previously described approaches for production of exchange-competent MHC-I proteins. First, the ELBM method does not require specialty equipment and chemicals, such as the conditional MHC ligands using photo-cleavable amino acid analogs that have been used to enable peptide exchange.³ Second, once expression cell lines are established, production of secreted, soluble, and native MHC-I proteins in tissue culture is less labor intensive and more accessible to many laboratories than the classical method for production of MHC-I peptide complexes via in vitro folding of subunits produced in E. coli.² Third, our expression system uses a constitutive promoter that enables production from serially passaged cells, obviating the need to grow seed cultures, which require subsequent chemical induction of protein expression or infection with nonintegrating agents such as baculovirus. Fourth, our method provides high yields across a large panel of MHC-I alleles, including alleles for which the in vitro folding methods have proven difficult, such as HLA-B*57:01. Finally, we have found it easy to design and produce ELBM constructs for new MHC-I alleles without having to go through extensive rounds of new designs and empirical testing.

Choosing among available options

The design of practical ELBM constructs for class I alleles was considerably simplified by optimization of the length of the linker connecting an optimal peptide ligand and the amino terminus of β₂m. In early iterations of the ELBM approach, where we used 13 to 15 amino acid linkers, we found that protease cleavage of the linker and peptide exchange was highly dependent upon the sequence of the fused peptide ligand, and that some combinations could be expressed, but proteolysis or exchange did not work at all. After optimization of the linker length and development of the 3C20 linker (Fig. S1), we still found that cleavage/exchange rates varied from 1 peptide ligand to another (as shown in Fig. 3A), but for achieving the primary practical goal of enabling complete peptide exchange in reactions no longer than 12 to 16 h, all constructs we have produced have proved suitable. With the 3C20 linker, it has proven more important to choose the peptide ligand that provides the highest expression yields, and failure of the formerly tethered peptide to dissociate from the peptide binding site has not been a significant problem. In other words, the carboxy-terminal extension to an optimal peptide ligand that remains after cleavage of the 3C20 linker is generally sufficient to destabilize the MHC peptide complex, as envisioned in the model depicted in Fig. 1.

Zippers and stability

Another novel feature of our ELBM design is the inclusion of leucine zippers. Leucine zippers have long been used to promote heterodimerization and stability of soluble T-cell receptors³¹ and MHC-II proteins⁴²; however, to our knowledge, they have not been used previously for expression of soluble classical MHC-I proteins, presumably because there had been no perceived need or because alternatives such as the single-chain trimer approach were available.^12,18 We had 2 primary rationales for use of leucine zippers in the ELBM method: (1) suppression of association of our soluble HCs with wild-type β₂m subunits produced by the 293 T cell–based expression line or bovine β₂m from the serum-supplemented medium and (2) stabilization of products we expected to be inherently metastable by design due to extension of MHC-I peptide ligands at their carboxy termini. Both of these could also have been met by using a single-chain trimer approach. The inclusion of zippers did increase production yields from our 293 T cell–based expression system (data not shown), consistent with the stabilization hypothesis. Furthermore, the use of zippers has a practical advantage over the single-chain trimer approach: It provides an additional, universal affinity handle recognized by the 2H11 mAb³¹ we have used for both purification and ELISAs to monitor expression yields. All the ELBM proteins we have tested have been stable at the pH of the buffer we use to elute them from the anti-zipper affinity column (pH 11.3, followed by immediate neutralization).

In most tested cases, the inclusion of leucine zippers appears to have imparted substantial stability to verifiably empty constructs that have clearly lost their formerly tethered peptides after cleavage of the 3C20 linker. The clearest example of this is for the mock exchange reactions for our H2-D^b construct depicted in Fig. 4C. In this construct, the peptide ligand is the influenza A NP 366–374 epitope ASNENMETM, and the portion of the 3C20 linker that remains attached to it after proteolysis is GSGGSGGSGLEVLFQ; the only charged side chains in the resulting 24-mer are the 3 glutamic acid residues, imparting a net charge of −3. For the H2-D^b sample that was cleaved in the absence of an exchange peptide, the bands on native IEF gels shifted substantially toward the cathode, consistent with an increase in the net charge of the remaining HC/β₂m heterodimer of +3, strongly suggesting that the bands on the gel correspond to a stable, empty class I heterodimer (stabilized by the leucine zipper). This stands in contrast to the zipperless exchange-competent molecules described by Schumacher et al.,⁴³ where photo-cleavage of the class I complexes with conditional MHC ligands in the absence of suitable exchange peptides (also called “rescue” peptides) resulted in unstable, empty class I molecules that unfolded, aggregated, and were easily removed by centrifugation. On rare occasions, we see some evidence for instability in mock exchange reactions in some alleles, either as visible precipitates or as bands that do not enter either SDS-PAGE or IEF gels, but it is rarely as prevalent as described for the zipperless constructs. One important consequence of this is that we cannot use ELISA assays as evidence for the extent of binding of exchange peptides.

IEF comments

To monitor binding of exchange peptides, we first turned to native IEF gels, previously shown to be able to resolve MHC‑Ip complexes differing by a single charge.³⁷ We were surprised to discover that the uncleaved ELBM proteins, purified to homogeneity by SDS-PAGE and chromatographic criteria, separated into multiple bands on native IEF, even though the proteins were 100% biotinylated (data not shown) and were expressed in an MGAT1^−/− cell line, a mutation that prevents the synthesis of charged, complex N-linked carbohydrates (a substantially greater number of bands was seen when the proteins were expressed in an MGAT1⁺ cell line; data not shown). We attempted multiple approaches to both understand and reduce the baseline IEF heterogeneity, including orthogonal purification techniques, endoglycosidase H treatment, and alternative construct designs, but each of these failed to yield conclusive results. From the data in Fig. 4, we can state with certainty that each of the species resolved by IEF is fully active as a peptide receptor, because all bands shift in concert upon binding of a synthetic peptide with a defined net charge. Furthermore, when using high-affinity exchange peptides, essentially 100% of the protein appears to bind to the exchange peptide, because all the bands seen in the exchange reactions are shifted relative to the bands in the mock exchange reactions. If there is a small amount of protein that does not bind the exchange peptide, it is highly unlikely to affect downstream assays such as multimer staining. The native IEF method for peptide binding does have 1 significant drawback: when the exchange peptide has a net charge of 0, it is impossible to distinguish a potential exchange complex from the mock exchange samples.

Mass spectrometry is insensitive to characteristics such as net charge in solution of short-peptide MHC ligands, and it has become an essential component of the assays we use for MHC‑Ip reagent validation. The examples shown in Fig. 5 demonstrate complete replacement of formerly tethered peptides with synthetic exchange peptides. This is seen in most cases with high-affinity peptide ligands, but in some cases, mass spectrometry analysis shows mixtures of the formerly tethered peptide and exchange peptides (data not shown). In the absence of appropriate standards, the size of the peaks in the MALDI-TOF spectra are, at best, semi-quantitative. Furthermore, although partial exchange can be expected to reduce the performance of tetramer stains (perhaps only seen in staining titration curves), it does not necessarily abrogate staining.

Tetramer staining

Tetramer staining efficacy using reagents prepared by the ELBM method has invariably been comparable to staining using reagents prepared by the conventional folding method, using population frequencies as a primary criterion. We have consistently observed that the staining intensity of the populations identified by the ELBM reagents tends to be slightly lower than when the conventional reagents are used; in most cases, this should not make a difference in assay results, but it could have effects when attempting to detect low-affinity T-cell populations such as those specific for self- or tumor-derived peptides. We have tried numerous approaches to both understand and address this—reagent titration, endoglycosidase H treatment, removal of the 3C protease after exchange, and preparation of ELBM constructs lacking leucine zippers (data not shown)—but none have produced significantly better results. We do note that comparisons of fluorescence intensity for tetramers prepared by conventional and exchange methods rarely have been reported.³

Other uses of ELBM proteins

Finally, although our primary interest was in developing a high throughput system for the production of MHC tetramers, the ELBM proteins, like previously described exchange-competent complexes, can be used for other assays such as competitive inhibition peptide binding assays using FP or other readouts.^6,43 In addition, the unique zipper-induced stabilization of apparently empty MHC molecules potentially opens new areas of investigation such as immunoregulation by empty conformers⁴⁴ or biochemical studies of components of the MHC-I peptide loading complex.

Author contributions

Vasanthi Ramachandiran (Conceptualization [Supporting], Methodology [Lead], Validation [Lead], Visualization [Lead], Writing—review & editing [Supporting]), John C. Shires (Conceptualization [Equal], Investigation [Supporting], Methodology [Supporting]), Richard A. Willis (Conceptualization [Supporting], Data curation [Supporting], Formal analysis [Supporting], Investigation [Supporting], Methodology [Supporting], Project administration [Supporting], Supervision [Supporting], Validation [Supporting], Visualization [Supporting], Writing—review & editing [Supporting]), Ge Bai (Investigation [Supporting], Methodology [Supporting]), Donielle L. Bell (Investigation [Supporting], Methodology [Supporting]), Project administration [Supporting], Supervision [Supporting]), Lixia Han (Investigation [Supporting], Methodology [Supporting], Project administration [Supporting], Tamara R. Kazarian (Investigation [Supporting], Methodology [Supporting]), Fred O. Ede (Investigation [Supporting], Methodology [Supporting]), Oskar Laur (Resources [Supporting]), Sujin Lee (Investigation [Supporting], Resources [Supporting]), Dale L. Long (Data curation [Supporting], Project administration [Supporting], Software [Lead]), and John D. Altman (Conceptualization [Lead], Funding acquisition [Lead], Investigation [Lead],Methodology [Equal], Project administration [Lead], Resources [Lead], Supervision [Lead], Writing—original draft [Lead], Writing—review & editing [Lead])

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This work was supported by the National Institutes of Health (grants HHSN272201300006C, 75N93020D00005, P30AI050409, and P51OD011132).

Conflicts of interest

Authors declare that they have no competing interests.

Data availability

All data are available in the main text or the supplementary materials.