Plasma samples from repeated healthy donors were initially screened for IgG and IgM against tetanus toxoid, rhCD152 and mouse IgG2a by ELISA. It was found that no donors (0/17) had any significant IgM or IgG responses to CD152; however, anti-tetanus toxoid IgG (15/17) and some, albeit low, levels of specific IgG antibody against mouse IgG2a (2/17) were found. After primary and secondary in vitro immunization of five PBMC's from individual donors, the respective EBV-activated lymphoblastoid cultures were screened for human Igs against CD152. When assayed by ELISA, frequencies of specific cultures detected in the lymphoblastoid cells after primary in vitro immunization ranged from 0% to above 3% and were significantly elevated when CD8⁺ and CD56⁺cells were simultaneously removed before primary peptide stimulation (Fig. 1A).

As shown in Additional File 1, although there was not a statistically significant increase, the removal of IL-10⁺ cells before secondary peptide stimulation yielded the highest frequency of specific IgG-producing cells.

Wells containing anti-CD152 IgG level that was five times higher than anti-murine IgG2a were subsequently pooled for electrofusion. One fusion yielded eleven clones secreting anti-CD152 Abs. Three of them were subcloned and found to be of the IgG4 isotype. To select a mAb for further characterization, two parameters were taken into account: the secretory capacity of the trioma, estimated from the ELISA titer and the relative specificity, reflected by the reactivities to other unrelated antigens such as tetanus toxoid, bovine serum albumin, and uncoated wells. Figure 1B also demonstrates that the selected monoclonal had a near-background reaction with unrelated Ags. Long-term (over a period of 60 months of continuous culture) stable Ab production in a concentration above 4 μg/ml/10⁷ trioma cells in spent medium was consistently observed.

To investigate whether the specificity observed in ELISA was also applicable for the activated human T cells in situ, we used the mitogen-dependent stimulation system, where peripheral T cells were activated with anti-CD3, PHA, ConA, or the combination of PMA/ionomycin or PHA/PMA. CD152-expressing T cells were numerated in CD3⁺ population by flow cytometry. After a 72-h culture period, few CD3⁺ T cells without stimulation were stained by our human mAb whilst a large number of CD152⁺CD3⁺ T cells were constantly observed after activation with PMA and PHA (Fig. 1B). Control human IgG4λ myeloma protein failed to distinguish among the activation statuses and the present mAb did not react with activated murine CD3⁺ T cells (data not shown).

To further characterize the nature of mAb binding, epitope mapping was performed by the Western blotting method with arrays containing the overlapping pendecapeptides, encompassing CDR1-like (the B-C loop), CDR3-like (the F-G loop) and the Met 55-cored sequence localized between the C' and D strands of the CD152 extracellular portion. Figure 2 depicts that only the peptide corresponding to the C-terminus of the Met 55-cored sequence (⁵⁴YMMGNELTFLDDSIC⁶⁸) was best recognized by the mAb while neither the promiscuous T-cell epitope, nor CDR1-like or CDR3-like region contributes to the binding. From this result, it can be concluded that the Ala 51, Ala 52, Thr 53 and Thr 69 are not essential for mAb recognition.

The mAb was isotyped and subtyped by solid phase ELISA, utilizing its reactivity with human CD152 and appropriate immobilized typing Abs. The binding depicts the simultaneous presence of γ4 and λ chains in the mAb while other Ig chains are absent (Fig. 3A), thus the mAb was termed as γ4λhuCD152. Additionally, to compare the clonal nature with existing human monoclonal IgG1λ and IgG4λ derived from purified myeloma proteins, the isoelectric focusing (IEF) patterns were subsequently visualized. The resolvable bands, as shown in Figure 3B, indicate that γ4λhuCD152 has a slightly lower yet basic pI similar to the IgG1λ, but in contrast to the acidic IgG4λ myeloma protein or to the anionic (pI 4.5–5.0) species of proteins commonly described for polyclonal IgG4 24. Western blot confirmed the purity of the Ab samples as illustrated by the anti-λ staining configurations. As shown in Figures 3B, the corresponding pI of the myeloid IgG1λ, IgG4λ and γ4λhuCD152 obtained with linear regression of pH gradient were 7.92–8.79, 5.76–6.52 and 7.87–8.41, respectively.

The equilibrium dissociation constant (Kd) for the purified intact γ4λhuCD152 was determined by an IAsys analysis. The rate constant was evaluated directly from the sensogram using five cycles of soluble mAb binding to the immobilized CD152-muIg. Figure 3D reveals that, with the analysis of extent and association in single phase, the Kd was deduced to be 4 × 10^-9 M.

Following elelectrofusion and subsequent clonings, stable trioma cells producing IgG4λ were selected and the cDNA encoding the Ig variable regions were cloned and sequenced. By comparing the VH with the available Ig sequences, it was concluded that the VH, being 89.80% (88/98) amino acid identity (Fig. 4A), is associated with three human VH3 germline segments [GenBank: AB019439, VH3–30 and VH3–33]. Alignments have also disclosed homology of VL to three existing human Vλ germline genes [GenBank: BAC01778, S78058 and CAA38313] with a measure of 92.13% similarity (Fig. 4B). High similarity to accessible V germline genes of human but not other origins is indicative of complete human Ab. Moreover, both VH (CDR2) and VL (CDR1 and CDR2) regions contained evidence of hyper mutations away from the germline.

Although the CDR2-containing epitope does not seem to be involved in the binding of endogenous cognate ligands (CD80 and CD86) and mAb engagement would not be antagonistic, the epitope might present an allosteric site for non-competitive inhibition. To investigate this possibility, CTLA-4-muIg was immobilized onto wells and either biotinylated CD80-muIg or CD86-muIg was used as a binding ligand in the presence of γ4λhuCD152 or BNI3, i.e., a CD152 antagonistic mouse IgG2a mAb 25. Figure 5A shows, in contrast with the expected dose-dependent inhibition of specific receptor binding by the antagonistic BNI3, γ4λhuCD152 could not compete binding significantly with either CD80-muIg or CD86-muIg ex situ. No clear difference in binding was found when competing CD80-muIg or CD86-muIg was used in the binding of γ4λhuCD152 to CTLA-4-muIg (data not shown). Surprisingly, high doses of γ4λhuCD152 mAb display synergism with the natural ligand CD80 but not CD86 with a consequence up to 50% enrichment of CD80 binding to CTLA-4-muIg.

In order to investigate the binding effects more thoroughly, and to provide further information on their possible in situ relevance, it was assessed whether labeling of endogenous CD152, a known rare Ag expressed only on activated T cells, by different mAbs, influences the profile of detection. A well-contrasted cytometric picture of total CD152-binding and a clearly distinguishable concentrated pattern of labeling were evidenced on PHA/PMA-activated T cells by using γ4λhuCD152, indicating a lower nonspecific binding over the antagonistic BNI3 (Fig. 5B). Hence, non-ligand competitive γ4λhuCD152 could be used as a new and refined probe which should be useful for sensitive assay and localization of CD152 both in situ and ex situ.

The culture medium used was RPMI-l640 (HyClone, Logan, UT), supplemented with 1 × non-essential amino acids (Life Technologies, Gaithersburg, MD), 10% fetal bovine serum (FBS; Life Technologies) and 50 μg/ml of gentamycin and kanamycin (Sinton Chemical & Pharmaceutical, Hsinchu, Taiwan). Purified and biotinylated human CD152-murine Ig fusion protein (CD152-muIg), CD80-muIg and CD86-muIg (Ancell, Bayport, MN) were used in Ag-specific and competing enzyme-linked immunosorbent assay (ELISA), together with peroxidase-labeled goat antibodies against human IgG and IgM (Zymed Laboratories, South San Francisco, CA) or avidin horseradish peroxidase (eBioscience, San Diego, CA) as the reporting system. The fluorochrome-conjugated mouse mAb against human IgGs and human CD3 (UCHT1; mouse IgG1), together with rat mAb against mouse IgG2a were commercially available from Becton Dickinson Immunocytometry Systems (San Jose, CA) and Abcam (Cambridge, UK). The anti-CD3 (OKT3; mouse IgG2a) used for T cell activation and the antagonistic anti-CD152 (BNI3; mouse IgG2a) were purchased from eBioscience and Abcam, respectively.

Plasma and buffy coat samples from healthy routine blood donors, screened negative for HIV-1/2, HTLV-I/II, HCV, HBsAg and containing normal levels of alanine transferase (ALT), were obtained from the Tainan and Hualien Blood Centers, Taiwan Blood Services Foundation. Written informed consents were obtained from five repeatedly-healthy regular blood donors after an explanation of the nature, purpose, and potential risks of the study and then 230 ml of whole blood was used for the purpose of site-directed in vitro immunization. PBMC's were isolated by density centrifugation on Ficoll-Paque (GE Healthcare Bio-Sciences, Uppsala, Sweden) as described elsewhere.

PBMC's were magnetically labeled with CD45RO MACS^® microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany) then separated by a VarioMACS™ (Miltenyi) instrument according to the manufacturer's instructions. The purified CD45RO⁺ T cells were cultured at a density of 2 × 10⁶ cells/ml in the culture medium supplemented with 50 μM 2-mercaptoethanol and 10 μg/ml pokeweed mitogen (PWM; Sigma, St. Louis, MO). After 24 h, cells were removed by 400 × g centrifugation to collect CD45RO⁺ T cell replacing factor. Removal of cytotoxic cell populations, which inhibit in vitro immunization 23, was similarly performed by using colloidal super-paramagnetic microbeads conjugated to monoclonal anti-human CD8 and anti-CD56 antibodies (Miltenyi). Removal of IL-10-producing cells was achieved by using rat anti-human IL-10 (SouthernBiotech, Birmingham, AL) and goat anti-rat IgG microbeads (Miltenyi).

Cytotoxic cell-depleted PBMCs were immunized in vitro based on a previously described two-step principle 6. Primary immunization was performed by incubating the cells for 6 days in a medium containing 10 nM of the heterotopic peptide Ag (QYIKANSKFIGITELAATYMMGNELTFLDDSICT; Fine Research Biochem, Taoyuan, Taiwan), 50 μM 2-mercaptoethanol, 10% heat-inactivated human serum, 0.05 ng/ml recombinant human (rh) IL-2 (eBioscience), and 25% (v/v) CD45RO⁺ T cell replacing factor. For secondary immunization, 3 × 10⁷ primary-immunized cells were mixed with the peptide in a flask that had been immobilized overnight with 5 mg/ml of CD40L (CD154; eBioscience) together with 1 × 10⁷ QYIKANSKFIGITEL (Fine Research Biochem)-stimulated CD4⁺ T cells and 5 ng/ml rh IL-15 (eBioscience). The cells were cultured for 3–5 days in a medium supplemented with 5% human serum, 50 mM 2-mercaptoethanol and 10 nM heterotopic peptide Ag. The significance of differences between treated and control cultures was established by using Student's t test. A P value of less than 0.05 was considered statistically significant.

The in vitro immunized cells were infected with EBV by virus-containing supernatant derived from the EBV-producing marmoset cell line B95-8 (American Type Culture Collection, ATCC CRL 1612; kindly provided by Dr. L.-F. Sheu, Tri-Service General Hospital, Taipei). The infected cells were seeded at 10⁵/well in 96-well plates together with mytomycin (Kyowa Hakko Kogyo, Tokyo, Japan)-treated PBMC as feeder cells (10⁴/well) for the establishment of lymphoblastoid cells and screened for Abs by ELISA.

Ag-specific ELISA was performed by coating 0.25 μg/ml purified rhCD152-muIg, 0.5 μg/ml monoclonal mouse IgG2a (mIgG2a; Ancell), 1 μg/ml bovine serum albumin (BSA; Sigma) or 1 μg/ml tetanus toxoid (TT; ADImmune, Taichung, Taiwan) onto microtitre plates overnight at 4°C. Culture supernatants were diluted to the desired level in 10 mM sodium phosphate buffer (pH 8.0), containing 0 5 M sodium chloride and 0.1% Tween-20. Coated plates were incubated with diluted culture supernatants, washed, incubated with peroxidase-labeled goat antibodies against human IgG and IgM and developed (15 min) by addition of 100 μl of the chromogenic substrate o-phenylaenediamine (OPD) (Sigma). The reaction was stopped after 30 min by adding 1 M sulphuric acid, and the absorbances were read at 490 nm.

Somatic cell hybridization was generated by electrofusion. Briefly, Ag-specific EBV-infected lymphoblastoid cells were fused with heteromyeloma cells 22 in an isotonic medium (280 mM sorbitol, 0.5 mM magnesium acetate, 0.1 mM calcium acetate and 1 mg/ml BSA; pH6.9–7.1). Cell fusion was induced by high-voltage pulses using a BTX Electro Cell Manipulator ECM 2001 (Harvard Apparatus, Holliston, MA). Ag-specific hybrids were selected and cloned by limiting dilution.

To define the specific epitope of human CD152 recognized by the mAb, we used peptide arrays (Genesis Biotech, Taipei, Taiwan and Fine Research Biochem,) containing in-situ synthesized peptides immobilized on special membrane. In brief, 1 μg/mL of protein A (Proteus MIDI kit, Pro-Chem, Littleton, MA)-purified mAb was incubated by shaking in room temperature for 2 h. After washing, the membrane-bound mAb was then visualized by diluted anti-human IgG conjugated with peroxidase (Jackson ImmunoResearch Laboratories, West Grove, PA) and FAST™ DAB (Sigma). The amount of bound mAb was calculated by Image-Pro Plus 4.5 software (Media Cybernetics, Silver Spring, MD) on the scanned images.

The surface expression of the CD152 epitope recognized by the present mAb was analyzed using two-color flow cytometry on mitogen-stimulated PBMC's by a FACSCaliber™ flow cytometer and CellQuest™ software (Becton Dickinson Immunocytometry Systems). 2 × 10⁶ isolated PBMC's were resuspended in supplemented culture medium and treated with either anti-human CD3 (OKT3; final concentrations in culture 10 μg/ml, eBioscience), concanavalin A (Con A; final concentrations in culture 10 μg/ml, Sigma), 10 μg/ml phytohemagglutinin (PHA; final concentrations in culture 1 μg/ml, GE Healthcare Bio-Sciences), a combination of phorbol 12-myristate acetate (PMA; 50 ng/ml, Sigma) + ionomycin (1 μM, Sigma) or a combination of PHA (1 μg/ml) + PMA (50 ng/ml). Logarithmically amplified fluorescence data were collected on 10,000 CD3⁺ cells. All flow cytometry staining procedures were performed at 4°C in cytometry buffer. For extracellular detection of CD152, activated cells were first surface stained with the mAb or isotype control at 4°C, followed by anti-human IgG-FITC and anti-CD3-PE (Becton Dickinson) staining.

The Ab primary structures were deduced by cDNA sequencing from cloned trioma cells. Briefly, poly(A)⁺ RNA was isolated from by using Dynabeads^® mRNA DIRECT™ Kit (Invitrogen, Carlsbad, CA). Purified mRNA was then employed as the reaction template in reverse transcription polymerase chain reactions (RT-PCR). The RT-PCR was carried out with Titan One Tube RT-PCR System™ (Roche Diagnostics Corporation, Indianapolis, IN). PCR primers (1 μM) used to amplify human VH and VL were the HuVH-JH set (forward: 5'-caggt caact taagg gagtc tgg-3' and reverse: 5'-tgaga gacgg tgacc gtggt ccc-3') and the HuVλ set (forward: 5'-tccta tgtgc tgact cagcc acc-3' and reverse: 5'-accta ggacg gtgac cttgg tccc-3'), respectively. The 37 temperature cycles include: one 2-min denature cycle of 94°C; 35 cycles of 3-min denaturation at 94°C, 1-min annealing and extension at 68°C; and a final 10-min extension cycle of 68°C. Single banded PCR fragments were seperated by 2% agarose gel electrophoresis. The DNA fragments were purified from gel by Wizard PCR Preps DNA purification system (Promega, Madison, WI). The purified products were subjected to nucleotide sequencing. Sequences were verified (Molecular Clinical Diagnostic Laboratory, Dr. Chip Biotechnology, Inc., Taipei, Taiwan) and converted to corresponding amino acids.

The isoelectric point of secreted IgG was examined by Novex IEF Gels (Invitrogen). Desalted and dialyzed protein samples were mixed with IEF sample buffer at 1:1 (v/v) ratio. Electrophoresis was performed at following condition: 100 V for 1 h, 200 V for 1 h and 500 V for 30 mins. After electrophoresis, gels were removed from gel cassette and fixed in 10% trichloroacetic acid for 30 mins. Fixed gel was developed by Coomassie brilliant blue staining or silver staining. The broad-range calibration kit for pI determinations (#17-0471-01, pH 3–10; GE Healthcare Bio-Sciences) was included as the standard. The isoelectric point of interested proteins was calculated by Phoretix 2D Elite software (Nonlinear Dynamics, Durham, NC).

The affinity of the mAb was determined against CD152-muIg with an IAsys optical biosensor (Affinity Sensors, Cambridge, UK) according to the manufacturer's instructions. Briefly, 200 μg/ml dialyzed and diluted CD152-muIg was immobilized on the activated surface of carboxymethyl dextran cuvettes in 10 mM of sodium acetate buffer at pH 3.8. After conditioning with 10 mM HCl, immobilization of 2 mg/mL CD152-muIg resulted in a response of 1100 arc sec. This represents the highest immobilization response for CD152 and gives a ligate binding capacity (R_max) of 300 arc sec. Serial dilutions of the mAb in PBS, i.e. 1.34 × 10^-9 M, 6.70 × 10^-9 M, 1.34 × 10^-8 M, 2.68 × 10^-8 M and 5.36 × 10^-8 M, were added to the CD152-coated cuvettes (final volume, 50 μl). Affinity constants (Kd) were calculated from these measurements as k_diss/k_ass by using the FASTFIT^® program provided by the manufacturer.