Production of recombinant biotinylated MHC-I heavy chains
To assure one single specific biotinylation of each recombinant MHC molecule, a biotinylation substrate peptide (BSP) sequence specific for the E.coli
biotin protein ligase, BirA, enzyme was attached to the C-terminal end of the recombinant MHC-I HC. This enables specific biotinylation of MHC-I HC by BirA. For MHC tetramer production, biotinylation of the BSP site has mostly been based on biochemical in vitro biotinylation of the folded MHC-I complex. This needs to be followed by chromatographic purification steps, which are time consuming and potentially associated with significant loss of material. Induction of efficient in vivo biotinylation of the HC would therefore save both time and material. The endogenous BirA expression in E. coli is insufficient for an efficient biotinylation of over-expressed proteins. However, co-expression of the BirA enzyme either in the E. coli or insect cell expression systems has been shown to increase the biotinylation efficiency
. Therefore, the BL21 (DE3) E. coli strain was co-transformed with a pET28a+ vector encoding the MHC-I HC-BSP and a pASYC vector encoding the BirA enzyme. Production of HC and BirA were induced by addition of IPTG for the last 3 h of culture (). Biotin was added to the bacterial culture immediately before induction of expression of HC and BirA. The biotinylation efficiency of the HC was evaluated in a SDS-PAGE gel-shift assay. Samples of HC were incubated with increasing amount of Streptavidin (SA), and subsequently separated by SDS-PAGE. Biotinylated HC will bind to SA and thereby significantly increase the molecular weight of the complex. This resulted in disappearance of the HC-band and appearance of bands of higher molecular weights (). Using densitometry, the residual HC-band representing the non-biotinylated HC were quantified and compared to the total amount of HC (lane without SA added). We have produced and analyzed 27 HLA-A alleles and 14 HLA-B alleles this way, and in all cases obtained biotinylation efficiencies in the range of 85–100% (). More than ¾ of these preparations were more than 95% biotinylated the rest were between 82 and 94% biotinylated. Thus, the in vivo biotinylation procedure was highly efficient.
Production of MHC-I heavy chains.
Biological activity of biotinylated MHC-I
To ensure that the in vivo biotinylation did not affect the specificity and subsequent folding of the MHC class I complex, we compared the peptide binding affinities of HLA-HLA-A*0201-HAT-BSP with and without biotinylation to the unmodified recombinant HLA-A*0201. A panel of small pox peptides representing good binders of KD
<10 nM (YLDYDTIYV, FLRDNLYHV, and YLSDSAINI), intermediate binders of KD
10-100 nM (YLSTERDHV and FLETDAGRV), and non binding peptides of KD
>1 µM (ALSDACKKI), were analyzed for binding affinity to each of HLA-A*0201 HC constructs. The HLA-A*0201 constructs were folded in the presence of ß2
m and a titration of the peptides and the KD
of each combination was determined in a quantitative ELISA assay 
(). Each peptide, strong as well as poor binders, gave essentially the same KD
value for all three HLA-A*0201 constructs. Thus, neither the addition of the BSP site nor the biotinylation prior to folding affected the binding specificity of the HLA-A*0201 molecule.
Peptide binding affinity of biotinylated HLA-A*0201 versus non-biotinylated.
In addition to a high biotinylation frequency, our approach also depends upon a high refolding efficiency. To this end we have exploited two previous observations; the first demonstrating that denatured class I MHC heavy chains with pre-oxidized disulfide bonds efficiently refold (when provided with β2
m and an appropriate peptide)
, and the second demonstrating that the efficiency of this process can be further improved by purification of denatured heavy chain molecules with correct disulfide conformations
. Thus, we have previously shown that subspecies of correctly oxidized recombinant denatured MHC-I HC can be purified from the inclusion bodies, and that these upon dilution folds into functional complex with a >90% efficiency
. In this paper, we have used 4 different MHC-I HC preparations for tetramer production: HLA-A*0101, HLA-A*0201, HLA-B*0702, and HLA-B*4402. These HC's were in vivo biotinylated, purified, and subsequently folded in the presence of a relevant peptide, pp65363–373
, and EBNA3b657–666
respectively. These complexes were titrated and measured in a quantitative ELISA. The folding efficiency was determined by comparison to a standard of known complex concentration. As seen in , we obtain 95–100% correctly folded complex and these HC were 97–100% biotinylated. Note, that although we don't know where and how these disulfide bonds are generated, we have at this point successfully generated tetramers from HLA preparations representing 14 different HLA alleles, and in no case have we failed. We conclude that these molecules in principle are ready to be tetramerized with SA without further purification.
Tetramerization of MHC and streptavidin
In order to follow the stoichiometric distribution of SA-MHC-I complexes during the formation of peptide-MHC tetramers, the folding reaction of biotinylated HLA-A*0201-HAT-BSP was spiked with radioactively labeled peptide. MHC-tetramers were the generated at different molar ratios of MHC-I and SA; and the formations of various SA-MHC-I multimers were followed by size-exclusion chromatography (SEC) through detection of radioactivity (). SA binds MHC-I at a 1 to 4 molar ratio. Gradually increasing the SA to MHC-I ratio from MHC-I in excess to SA in excess, saturated and unsaturated SA molecules could be generated at different stoichiometric ratios; SA:MHC4, SA:MHC3, SA:MHC2, SA:MHC, and eventually MHC monomers. Quantification of molecular weights according to standard marker proteins showed agreement between theoretical and empirical molecular sizes (). All four SA-MHC-I multimers were generated dependent upon the ratio between SA and MHC-I. At SA to MHC-I ratio of 1:4, the tetramer form was almost quantitatively produced, generating none of the other multimer species and only leaving a small fraction of non-reacted monomers. This presumably represents a small fraction of non-biotinylated HC.
Binding of SA and biotinylated HLA-A*0201-peptide complexes.
Staining specific T cells using “one pot, mix and read” tetramers
The efficiency of the biotinylation and folding processes suggested to us that it should be possible to generate MHC tetramers in a “one-pot” reaction omitting most, if not all, of the preparative biochemical work normally involved in tetramer production. To analyze whether our “one-pot” tetramers were able to stain specific T cells in a “mix-and-read” mode, we initially selected the HLA-A*0201 restricted dominant CMV T cell epitopes, CMV pp65495–503 peptide as model system. CMV pp65495–503 specific T cells from healthy donors were generated and HLA-A*0201/pp65495–503-SA-phycoerythrin (PE) tetramers were produced as described above. The tetramers were not concentrated nor further purified, but used directly to stain the specific T cells.
The specificity and sensitivity of the tetramer staining were compared to that of the commercially available HLA-A*0201/pp65495–503
pentamers (®ProImmune, www.proimmune.com
) (). The tetramers stain about 60% of the CD8 T cells with a baseline separation from the negative population. The same frequency of specific T cells was detected using the commercial pentamers. The staining intensity of the tetramers were slightly lower than that of the pentamers (FI 1480 vs. 2042), however, the background staining of the negative population was also lower (FI 4 vs. 10) resulting in a higher signal to noise ratio for the in-house generated tetramers (370 vs. 202). Thus the “one-pot, mix-and-read” strategy for the HLA-A*0201/pp65495–503
tetramer production appeared to work as well as the MHC pentamer of the same specificity.
Tetramer staining comparison to pentamer.
The MHC-I monomers were produced at a rather low concentration compared to conventional tetramer production. These monomers were mixed with streptavidin directly, without any concentration, to generate the tetramers. The stability, yield, and robustness of this production strategy were of concern. We found that the tetramers could be stored for more that 9 months at 4°C with no change in activity (data not shown). Another concern was whether there was a lower concentration limit of monomers needed for production of tetramers in order to give an optimal staining of the specific CD8 T cells. To analyze this HLA-A*0201/pp65495–503 monomers were folded at 200 nM, 100 nM and 50 nM and each mixed with streptavidin to produce tetramers. Thus, the maximum tetramer concentration of the 200 nM monomers was 50 nM tetramer, of the 100 nM monomers 25 nM tetramer, and of the 50 nM monomers 12.5 nM tetramer. Two fold titrations were made of each of these three productions and used to stain pp65495–503 specific T cells. Baseline separation between the positive and negative cell population was seen for all the concentrations used. The relative fluorescence intensity is depicted as a function of the actual tetramer concentration used to stain the cells (). The curves for the three-tetramer productions were super imposable showing considerable robustness in set-up and staining reactions. The actual tetramer concentration used in the staining reaction, and not the concentration of monomers used for the tetramer production, was important for the staining intensity. The staining intensity reached a plateau and the T cell labeling was saturated at tetramer concentrations above 12.5 nM. Therefore, saturable T cell staining can be achieved using tetramers produced from concentration of monomers higher than 50 nM.
Titration of complex concentration for tetramer production.
Unreacted monomers do not inhibit HLA tetramer staining
It is inherent to the “mix-and-read” tetramer strategy that the staining reaction is done with the entire tetramerization reaction mixture including any unreacted monomer such as that of any un-biotinylated HLA molecule. A priori, the tetramerized peptide-HLA complexes would have a much stronger reaction with the appropriate T cell receptor than any unreacted monomers would have. Addressing whether the monomers could inhibit tetramer staining, we added increasing concentrations (up to 180 Nm) of monomers to specific T cells, incubated the cells at room temperature for 5 minutes, and then added a fixed concentration of tetramer (5 nM tetramer corresponding to 20 nM monomer). In no case was any inhibition of tetramer staining observed () suggesting that unreacted monomers do not compromise staining with our one-pot, mix-and-read tetramers. Note, that we have not addressed any physiological effect of peptide-HLA monomers such as the partial activation demonstrated by Malissen and co-workers 
Unreacted monomers do not inhibit tetramer staining.
Extending to other HLA-A*0201-peptide epitopes
To validate the generality of the “one-pot, mix-and-read” strategy for tetramer production, a different set of tetramers were produced from 3 different HLA-A*0201 EBV peptide complexes: HLA-A*0201/LMP1125–133, HLA-A*0201/LMP2356–364 and HLA-A*0201/LMP2416–424. PBMC's from four different HLA-A2 positive donors were restimulated with the three different peptides for 8 days and then stained with the tetramers (). PBMC's from all four donors could be stained by one or more of the three tetramers. In all cases, the specific staining showed baseline separation between positive and negative populations. Three of the donors had LMP2 356–364 specific T cells varying from about 3 to 11% of the CD8 T cells being labeled with HLA-A*0201/LMP2 356–364. Two of the donors were positive for LMP2 416–424 with about 3 and 4% specific CD8 T cells. Only one out of four donors was positive for LMP1125–133 with about 3% specific CD8 T cells. The same results were obtained when the frequency of peptide specific T cells producing IFNγ was evaluated by intracellular staining (data not shown). Thus this “one-pot, mix-and-read” strategy can be used to produce several different HLA-A*0201 tetramers and can directly be used for epitope screening analysis.
HLA-A*0201-tetramer screening of EBV reactive donors.
Extending to other HLA-alleles
We wanted to extend this application to other HLA and peptide-HLA combinations. Complexes of two dominant CMV pp65 T cell epitopes, HLA-A*0101/pp65363–373 and HLA-B*0702/pp65417–426, were folded. As for the HLA-A*0201, the folding efficiency of these complexes were >96% (). Tetramers of these complexes and the HLA-A*0201/pp65495–503 were produced and used to analyze the specificity of a CMV response. PBMC's from four different healthy donors were stimulated for 8 days with a mixture of 134 overlapping 15-mer peptides representing the entire pp65 CMV protein and subsequently analyzed for pp65 specific responses using intracellular IFNγ (IC-IFNγ) staining. All four donors showed a pp65 specific response, having between 16% and 46% pp65 specific CD8 T cells (, top panel). Using the HLA-A*0101/pp65363–373, HLA-A*0201/pp65495–503, and HLA-B*0702/pp65417–426 tetramers these donors were analyzed for the specificity distribution against these three dominant pp65 T cell epitopes (). All three tetramers showed specific staining and we were clearly able to differentiate the donors as responder and non-responder to the given epitope. Two of the four donors (Donor 1 and 4; ) stained positive for the HLA-A*0101/pp65363–373 tetramer; one having a relative high frequency of positive CD8 T cells and the other a very low frequency (17.6% versus 0.06%). Two donors (Donor 1 and 2; ) stained positive for the HLA-A*0201/pp65495–503, tetramer, both with relative high frequencies (28.5 and 15.6%). Only donor 4 stained positive for HLA-B*0702/pp65417–426 with 36% of the CD8 T cells being specific for this epitope. Except for one donor, the frequencies of tetramer positive CD8 T cells for each donor fully covered the entire pp65 response measured by intracellular staining for that donor. Donor 3 was not stained by any of the three tetramers. Therefore, the pp65 specific CD8 T cells response, must be represented by pp65 epitopes different from the three used for tetramers.
Screening for CMV specific T cells with three different Class I-tetramers.
Correlating staining with functionality
Finally, we correlated tetramer staining with functionality. Though the frequency of tetramer positive cells and IFNγ producing cells correlated nicely we wished to assure that the cells, which stained tetramer positive, were in fact the IFNγ producing cells. Three different tetramers were produced: HLA-A*0101/pp65363–373, HLA-A*0201/pp65495–503, and B*4401/EBNA3b657–666. Three different donors were restimulated for 8 day with the three peptides, CMV pp65363–373, CMV pp65495–503, and EBV EBNA3b657–666, respectively. T cells from each donor were double labeled for tetramer and IC-IFNγ (). For all combinations, the specific IFNγ producing CD8 T cells were also stained with the relevant tetramers. For the two donors specific for pp65363–373 and pp65495–503, a few percentage of the tetramer labeled cells did not produce IFNγ. Since, the majority of the CMV specific CD8 T cells produced both IFNγ and TNFα and only a few percentage produced either one or the other alone, these tetramer+, IFNγ- cells were most likely TNFα producing. Thus, the CD8 T cells stained with these tetramers were in fact the specific CD8 T cells that produce IFNγ in response to the relevant peptide epitope.
Double staining with tetramer and IC-IFNγ.
In summary, we have developed a rapid “one-pot, mix-and-read” strategy for tetramer production. Due to highly efficient biotinylation, folding, and tetramerization processes, all the preparative biochemical work normally needed during MHC tetramer generation can be omitted and the production can run as a fast sequential addition of the relevant reagents. Others have also attempted to generate a high-throughput production of MHC class I tetramers
. Their strategy is based on peptide exchange of MHC complexes folded with an UV cleavable peptide. These degradable complexes are produced in the conventional way being time consuming and associated with significant loss of material. At this point, this peptide cleavable approach has been applied to one human MHC class I allele (HLA-A*0201) and a mouse allele (Db
). The broad application of their strategy is limited by the need to define a unique UV-labile epitope for each MHC class I allele. In contrast, our approach is based on the ability to produce fully in vivo biotinylated and readily foldable MHC I HC. Ready-to-use tetramers with the desired specificity and scale can be produced within 2 days in any lab independent of special biochemical skills or equipment. We have demonstrated the application of our “one-pot, mix-and-read” strategy for 7 different MHC I restricted T cells epitopes spanning 4 different HLA class I alleles. We have produced an additional 37 fully biotinylated HLA-class I HC using this strategy (). This procedure is suitable for large-scale production of a single peptide-MHC tetramer for repeated analysis or purification of specific T cells as well as for a small-scale production of a larger number of different class I-peptide complexes e.g. for a comprehensive T cell screening program.