Antigen-specific CD4+ T cells can be best visualized with an antibody specific for activated β2-integrins
We have previously shown that antigen-specific CD8+ T cells are successfully detected by staining of activated β2-integrins with fluorescent mICAM-1 [5]. We therefore investigated whether mICAM-1 binding could also be used as a marker for detection of antigen-specific CD4+ T cells. We selected a healthy donor with a high frequency of ex vivo detectable CD4+ T cells specific for the HPTFTSQYRIQGKLE epitope derived from the CMV pp65 protein (CMV/HPT). WB cells were incubated with the CMV/HPT peptide, with Staphylococcus enterotoxin B (SEB), or remained unstimulated for 4 h, and mICAM-1 was added for the final 4 min of activation. Blood cells were further processed for detection of intracellular TNF production (gating strategy is shown in Fig. 1A). CMV/HPT-, as well as SEB- stimulated cells, showed only a modest staining for both TNF and activated integrins (Fig. 1B, top). This was unexpected since the frequency of CMV/HPT reactive CD4+ T cells was found to be approximately 1.2% in intracellular cytokine staining prescreening experiments in this donor. To check whether the antigen-specific CD4+ T cells had aggregated during the activation, and therefore were excluded from the lymphocyte gate, we treated WB with 4 mM EDTA after the mICAM-1 staining (Fig. 1B, bottom). Then, much higher frequencies of CMV/HPT-specific or SEB-reacting TNF+ CD4+ T cells were detected (approximately 1.2% and 9%, respectively). However, these were mostly mICAM-1neg, most probably because EDTA reversed mICAM-1 binding to activated β2-integrins by chelating Mg2+ and Ca2+ ions required for β2-integrin conformational change and stable interaction with mICAM-1. To confirm that the stimulation leads to aggregation of antigen-specific CD4+ T cells, which can be disrupted by EDTA treatment, we also analyzed the distribution of TNF+ CD4+ T cells within the lymphocyte gate (FSC-A/SSC-A) or within singlet cells (FSC-A/FSC-H). Without EDTA treatment, 15% of the CMV/HPT-specific and 42% of the SEB-responding TNF+ cells were within the lymphocyte or singlet gates, whereas EDTA treatment increased these numbers to 84-88% (Fig. S1 A-D).
Because mICAM-1 staining was not suitable for labelling activated β2-integrins on antigen-specific CD4+ T lymphocytes, we next tested staining of the cells with a monoclonal Ab that binds specifically to the extended/open high-affinity, but not to the resting, un-activated, conformation of β2-integrins (clone m24, referred thereafter as m24 Ab). WB cells from the same donor were stimulated with the CMV/HPT peptide, with SEB, or remained unstimulated for 4 h, then stained with m24 Ab for 15 min, followed by detection of intracellular TNF. Unstimulated cells showed a very low staining with m24 or anti-TNF Ab. Significant numbers of TNF+ CD4+ T cells were detected after stimulation with CMV/HPT (0.2%) and SEB (6.1%), most of them being m24+ (Fig. 1C, top), however, these were much less than when mICAM-1 and EDTA were used. Addition of EDTA after m24 Ab staining improved TNF detection to expected frequencies (0.94% and 10.5% for CMV/HPT and SEB, respectively, Fig. 1C, bottom). The TNF-producing cells were essentially m24+, indicating that once the Ab is bound, EDTA does not reverse the binding, as it does for mICAM-1. Without EDTA treatment, 18% of the CMV/HPT-specific and 50% of the SEB-stimulated TNF+ cells were within the lymphocyte or singlet gates, whereas EDTA treatment increased these numbers to 74-86% (Fig. S1 E-H). When EDTA was added before, or at the time of mICAM-1 or m24 Ab staining, the cells did not stain with either of the reagents (≤ 0.001%). Hence, m24 Ab staining of activated β2-integrins followed by EDTA treatment can be used to detect antigen-specific CD4+ T cells.
Antigen-specific β2-integrin activation on CD8+ T cells can be visualized by either mICAM-1 or m24 Ab staining
In our previous work, antigen-specific CD8+ T cells were detected with mICAM-1 complexes (without EDTA) [5]. We therefore tested if staining of activated CD8+ T cells with m24 Ab would be comparable to that obtained with mICAM-1, and to which extent EDTA influences detection. We selected an HLA-A*02+ donor with a high frequency of CD8+ T cells specific for the immunodominant epitope CMV pp65 NLVPMVATV (NLV). WB cells were stimulated with the CMV/NLV peptide for 16 min, then mICAM-1 and HLA-A*02 (referred thereafter as A*02) tetramers refolded with the CMV/NLV peptide were added for the final 4 min of the activation, or m24 Ab and A*02/NLV tetramers for 15 min after the activation. The combination of A*02/NLV and mICAM-1 identified 0.92% A*02/NLV+ CD8+ T cells (from which nearly all - 0.88% - were mICAM-1+, Fig. 1D, top left), while the combination of A*02/NLV and m24 Ab identified 0.99% A*02/NLV+ CD8+ T cells (0.95% were m24+, Fig. 1D, top right). Treatment with EDTA after mICAM-1 or m24 Ab staining only marginally increased the frequency of tetramer+ cells (1.1%, Fig. 1D, bottom). Similar to what we had observed for CD4+ T cells, staining of activated integrins on CD8+ T cells was largely lost when EDTA was used in combination with mICAM-1, but not with m24 Ab. To confirm that the stimulation did not aggregate the antigen-specific CD8+ T cells even without EDTA treatment, we analyzed the distribution of A*02/NLV+ CD8+ T cells on FSC-A/SSC-A and FSC-A/FSC-H plots (Fig. S2 B and C, top). Without EDTA treatment, 80-81% of the CMV/NLV+ cells were within the lymphocyte or singlet gates, whereas EDTA treatment increased slightly these numbers to 91-93% (Fig. S2 B and C, bottom).
Next, we assessed whether m24 Ab staining indeed identifies the same antigen-specific T cells as mICAM-1 staining. WB cells from the same donor were stimulated with the CMV/NLV peptide, with SEB, or remained unstimulated for 16 min, then the cells were stained with mICAM-1 and m24 Ab. Double stainings confirmed that both mICAM-1 and m24 Ab essentially identify the same cells (Fig. 1E). Hence, antigen-specific β2-integrin activation on singlet CD8+ T lymphocytes can be visualized by either mICAM-1 (without EDTA) or by m24 Ab with or without EDTA treatment, but for simultaneous detection of CD4+ and CD8+ T cells, m24 Ab and EDTA treatment should be combined.
Kinetics of β2-integrin activation and m24 Ab binding is different for CD4+ and CD8+ T cells
To determine the optimal duration for cell stimulation, we performed time courses of β2-integrin activation and m24 Ab binding on various antigen-specific CD4+ and CD8+ T cells. WB from one to four selected donors was stimulated with various antigens as indicated or remained unstimulated. Blood cells were harvested after different incubation times, followed by m24 Ab staining plus EDTA treatment (kinetics are shown as mean±SEM % m24+ in Fig. 2A and representative examples in Fig. 2 B-E). As previously observed with mICAM-1 staining [5], responding CMV/NLV+ CD8+ T cells were very rapidly detected (immediate peak response of m24 Ab staining within only 4-16 min, Fig. 2A). Following prolonged activation with the CMV/NLV epitope (>1 h), percentage of m24+ cells diminished only slightly (1.45% vs. 1.37% after 16 min or 4 h of stimulation, respectively, with a more substantial decrease found in the intensity, Fig. 2D), in contrast to the strong decline if using mICAM-1 [5]. Antigen-specific CD4+ T lymphocytes showed different kinetics of β2-integrin activation. A short-term activation with the CMV/HPT epitope in one donor induced only a weak m24 Ab staining on CD4+ T cells, with a peak response after 4-6 h of stimulation (Fig. 2 A and B). Reactivity to overlapping 15mer peptides (HBV/Env) in two HBV-vaccinated donors also reached maximum between 4 and 6 h (Fig. 2 A and C). Finally, the percentage of m24+ CD4+ T cells strongly increased and peaked after 4-6 h of SEB stimulation for all 4 donors tested (Fig. 2 A-C). For CD8+ T cells, m24 Ab staining peaked at 4 min in one case, while for the other two donors, it reached maximum at 4 h, but with decreased intensity (Fig. 2 A and D). All unstimulated cells showed a weak staining with m24 Ab (0.01 to 0.05%, Fig. 1 B-D).
To confirm that m24 Ab selectively identifies antigen-specific T cells, we co-stained cells from the CMV-reactive HLA-A*02+ donor with A*02/NLV tetramers and m24 Ab. The majority of NLV-specific CD8+ T cells could be detected by m24 Ab binding, with maximal staining achieved within 4-16 min of activation (Fig. 2 A and E, 97% of double-stained cells after 16 min of stimulation). For the HLA-DRB1*11+ donor, co-staining with CMV/HPT tetramers and m24 Ab was not feasible because of strong TCR downregulation after peptide stimulation, however, we found comparable frequencies of CMV-specific cells with both methods (2% DRB1*11/HPT+ (Fig. 2F) vs. 1.76% m24+ CD4+ T cells at the max. of response after 6 h of stimulation with the CMV/HPT peptide). Hence, m24 Ab identifies antigen-responding CD4+ T cells, but in contrast to CD8+ T cells, several hours of stimulation are required to activate β2-integrins.
m24 Ab binding identifies functional antigen-specific CD4+ T cells
We then tested whether m24 Ab binding correlates with functionality of antigen-specific CD4+ T cells. WB from the DRB1*11+ donor was stimulated with the CMV/HPT peptide, with SEB, or remained unstimulated. Blood cells were harvested after 1, 2, 3, 4 or 6 h, followed by incubation with m24 Ab plus EDTA treatment, and staining for intracellular CD154 and cytokines (TNF, IFN-γ, and IL-2). Results of the CMV/HPT stimulation are shown in Fig. 3A, and density plots after 6 h incubation without or with CMV/HPT in Fig. 3B top and middle, respectively. m24 Ab binding was the earliest event to be detected (after 1-2 h of stimulation, Fig. 3A opened bars), and upregulation of CD154 and cytokines was exclusively confined to the m24+ cell subset (positive (marker+ m24+ cells) vs negative (marker+ m24neg cells) colored bars, Fig. 3A). Also, almost all m24+ cells (85%) expressed CD154, TNF or IFN-g at 6 h. Moreover, cytokine producing cells were mainly m24+ CD154+. Only 15%, 12%, and 1% of the TNF, IFN-g or IL-2 producing cells did not express CD154, and virtually none were m24neg (Fig. 3B bottom). Some single stained m24+ or CD154+ events, but no double stained m24+ CD154+ were observed in unstimulated cells (Fig. 3B top, left). Comparable results were obtained for the SEB-stimulated cells in this individual, with a strong association between m24 Ab staining and TNF or IFN-g production; still, a fraction of the CD154+ or IL-2+ cells were m24neg, however, we found that CD154+ m24neg cells were predominantly cytokineneg (Fig. S3 A and B).
To address specificities against further viruses, WB from two HBV-vaccinated donors were stimulated with HBV/Env overlapping peptides, with SEB (one donor only), or remained unstimulated. Cells were harvested after 1, 2, 3, 4, or 6 h, followed by m24 Ab staining, EDTA treatment, and staining for intracellular CD154 and cytokines. For the first donor, time course of HBV/Env stimulation is shown in Fig. 3C and density plots after 6 h of incubation without or with HBV peptides are displayed in Fig. 3D top and middle, respectively. Frequencies of HBV-specific cells were low (approximately 0.1% within the CD4+ T cell subset), but readily detectable by m24 Ab staining. Similar to what we observed with CMV-specific cells, CD154 expression and cytokine production were essentially confined to the m24+ cell fraction (positive (marker+ m24+ cells) vs negative (marker+ m24neg cells) colored bars in Fig. 3C). However, a significant proportion of m24+ cells (about 50%) did not upregulate CD154 or cytokines even after 6 h. Some m24+ cells negative for the functional markers were also present in the unstimulated control. When we analyzed single and double m24+ and CD154+ cells, we found that cytokine-producing cells were mainly m24+ CD154+. Only 6%, 0% and 0% of the TNF, IFN-g, or IL-2 producing CD4+ T cells, respectively, were CD154neg, while 6%, 9% and 7% were m24neg (Fig. 3D bottom). Some single stained m24+ or CD154+, but no double stained m24+ CD154+ events were observed in unstimulated cells (Fig. 3D top, left). After SEB stimulation, a strong association between m24 Ab staining and TNF, IFN-g, and IL-2 production was observed. Again, a significant proportion of the CD154+ were m24neg, however, those were predominantly also cytokineneg (Fig. S3 C and D).
Similar observations were made for a second HBV vaccinee (Fig. S4). Although some background staining with m24 or CD154 Abs was observed in the unstimulated test, the combination of both markers (0.001% of m24+ CD154+ cells) allowed to distinctly detect HBV-specific cells even at very low frequencies (Fig. S4B: 0.011%, 0.008%, 0.001%, and 0.004% for CD154, TNF, IFN-g, and IL-2 dot plots, respectively).
In conclusion, m24 Ab staining is suitable for detecting functional antigen-specific CD4+ T cells, and the combination of this marker with CD154 can be used for identification of extremely low frequencies of functional antigen-specific T cells.
Monitoring of SARS-CoV-2 specific T cell immunity
Next, we examined the feasibility of our assay to monitor SARS-CoV-2-specific CD4+ and CD8+ T cells. Blood was obtained from CoV-2 convalescents (n=3) and unexposed healthy donors (n=4), as confirmed by antibody ELISA on the day of blood withdrawal. In accordance with our previous results for CMV- and HBV-specific CD4+ T cells, there were no m24+ cells producing CD154 or cytokines without peptide-specific stimulation (0-0.001%, exemplary donor CoV-2 3 is shown in Fig. 4A top, pink frames). Addition of overlapping peptides derived from the membrane (M), nucleocapsid (N), or spike (S) proteins induced a clear CD4+ T cell response in previously exposed individuals, detectable by co-staining with m24 and CD154 Ab, even at very low frequencies (down to 0.005-0.006%, M-, N-, and S-specific CD4+ T cell responses from donor CoV-2 3 occurring at frequencies of 0.050%, 0.006%, and 0.014% are shown in Fig. 4A, pink frames). All three volunteers responded to the three proteins, albeit at a different intensity. Strongest responses were directed at the protein M (up to approximately 0.05% of the CD4+ T cells in CoV-2 2 and 3, Fig. 4B). CoV-2 1 responded weakly to protein M (0.005%), but strongly to proteins N and S (both 0.009%). Cytokine production (TNF, IFN-g, and IL-2) was observed within m24+ cells. T cell activation was virtually not detected in the four unexposed donors (0-0.003%), except for one of them (UD 2) who interestingly showed a distinct, but very low (possibly cross-reactive) response to the protein S only (0.005% m24+ cells expressing either CD154, TNF or IFN-γ, Fig. 4B).
Anti-SARS-CoV-2 CD8+ T cell reactivity was also assessed, either simultaneously to the CD4+ T cell measurement (m24 Ab staining after 4 h of stimulation) or in an independent test after 1 h of stimulation only, followed by surface detection of activated integrins with mICAM-1. There were no m24+ cells producing cytokines without antigen-specific stimulation (0-0.001%, shown for CD8+ T cells from CoV-2 1 in Fig. 5A, top left). Interestingly, only protein N was recognized in two out of three convalescents, and m24+ cells produced TNF (0.017% and 0.006% for CoV-2 1 and CoV-2 3, respectively) and IFN-γ (0.023% and 0.011%) (Fig. 5B, left), but no IL-2 (≤ 0.001%). None of the four unexposed volunteers reacted to any of the CoV-2 peptides (0-0.002%), except again for UD 2 who had a response to the N-derived peptide pool (0.010% m24+ CD8+ T cells expressing either TNF or IFN-γ). When mICAM-1 staining was applied for detection, the background was stronger (0-0.006%, CoV-2 1 is shown in Fig. 5A, right), which did not allow for a sensitive assessment of specific CD8+ T cells. As a result, only the response of CoV-2 1 directed at protein N was clearly detected (0.011%; Fig. 5B, right).
We next tested CD4+ T cell reactivity in the blood of a healthy volunteer who had been immunized approximately 6 weeks before with a cocktail of synthetic peptides containing HLA-class II SARS-CoV-2 sequences derived from the nucleocapsid (N) protein (IGYYRRATRRIRGGD, IGY and ASAFFGMSRIGMEVT, ASA) and from the envelope (Env) protein (FYVYSRVKNLNSSRV, FYV), as well as one recall CMV pp65 peptide (YQEFFWDANDIYRIF, YQE) [9]. Virtually no m24+ cells producing CD154 or cytokines were detected when cells were left unstimulated for 6 h (0-0.001%, Fig. 6A). CMV/YQE stimulation caused a strong expression of CD154 (0.418%), TNF (0.312%), IFN-γ (0.382%), and IL-2 (0.253%) within m24+ cells (Fig. 6B). The three SARS-CoV-2 peptides used for vaccination induced variable responses, with 0.094%, 0.060% and 0.012% of m24+ CD154+ CD4+ T cells, for SARS-CoV-2 N/IGY, N/ASA and Env/FYV, respectively, as well as production of all three cytokines (Fig. 6C-E). These results are coherent with those obtained by IFN-γ ELISpot approximately 3 weeks after immunization [9].
m24 Ab stainings of antigen-specific CD4+ and CD8+ T cells in frozen/thawed PBMCs
We have also tested our method for detection of T cells within cryopreserved PBMC samples. All PBMCs were thawed and rested overnight at 37 °C and 7.5% CO2. Then 2x106 cells were stimulated in 1 ml TCM. Two donors, with high (approximately 2% of DRB1*11/HPT tetramer+) and low (approximately 0.05% CD154+ against HBV/Env) frequencies of specific CD4+ T cells were selected (experiments with WB of the same donors were reported in Fig. 2 and 3). PBMCs were stimulated with the CMV/HPT peptide (first donor), with HBV/Env overlapping peptides (second donor) or remained unstimulated for 6 h, followed by incubation with m24 Ab plus EDTA treatment, and staining for intracellular CD154 and cytokines (TNF, IFN-γ, and IL-2).
For the first donor, b2-integrin activation (m24 Ab binding) and functionality (CD154, TNF, IFN-g and IL-2 production) were stronger in PBMCs than in WB (1.90% vs 1.36% of m24+ CD154+ cells, respectively; compare Fig. S5B and Fig. 3B). Similar to what we observed with WB (Fig. 3A), upregulation of CD154 and cytokines was exclusively confined to the m24+ cell subset and almost all m24+ cells (93%) expressed CD154, TNF or IFN-g (Fig. S5B). Moreover, cytokine producing cells were mainly m24+ CD154+. Only 17%, 12%, and 1% of the TNF, IFN-g or IL-2 producing cells did not express CD154, and virtually none were m24neg (Fig. S5B bottom). Similar observations were made for the second donor, with a more robust response in PBMCs than in WB (0.058% vs 0.041% of m24+ CD154+ cells for PBMCs and WB, respectively; compare Fig. S5C and Fig. 3D). Again, some single stained m24+ or CD154+ events were observed in unstimulated cells, but virtually no double-stained m24+ CD154+ events were detected (Fig. S5C top, left). Cytokine-producing cells were mainly m24+ CD154+. Virtually none of the TNF, IFN-g, or IL-2 producing CD4+ T cells, were CD154neg, while only 1%, 0% and 8%, were m24neg, respectively (Fig. S5C bottom). Thus, m24 Ab staining together with the expression of CD154 allows the reliable identification of low frequency, polyfunctional CD4+ T cells in cryopreserved PBMCs.
We applied the m24 Ab assay for simultaneous detection of antigen-specific CD4+ and CD8+ T cells within PBMCs. Cells of one pre-screened donor (with a known low CD4+ T cell reactivity against a HLA-class II peptide mix containing CMV, EBV and Flu-derived epitopes, and a low frequency of CD8+ T cells against one HLA-A*02 restricted EBV-derived epitope) were thawed, rested, then stimulated with all peptides in pool for 4 h. Activation was then measured with m24 Ab combined to intracellular Abs. A background staining was observed with single m24 or single CD154 Abs on unstimulated CD4+ T cells, however the combination of both markers allowed to readily detect approximately 0.05% of reactive cells (Fig. 7A). Combination of m24 Ab with TNF and IFN-g (but not with CD154 or IL-2) identified sensitively EBV-specific CD8+ T cells (Fig. 7B).
Optimization of m24 Ab and ICAM-1 stainings of CD8+ T cells within frozen/thawed PBMCs after short-term stimulation
The short-term stimulation in contrast to the 4-6 h protocol allows faster and simpler evaluation of functional CD8+ T cells by surface detection of activated β2-integrins alone or together with tetramers [5]. Hence, we optimized the detection of activated β2-integrins on antigen-stimulated cryopreserved CD8+ T cells after short-term stimulation. PBMCs from a HLA-A*02+ donor with a high frequency (approximately 1.2%) of A*02/NLV tetramer+ CD8+ T cells were thawed and rested overnight at 37 °C and 7.5% CO2. 2x106 cells (in 1 ml or 0.2 ml TCM) were subsequently stimulated for 16 min with the CMV/NLV peptide and stained with either m24 Ab (with EDTA) or mICAM-1 (without EDTA); some stains were also performed in combination with CMV/NLV tetramers. Activation of specific CD8+ T cells was much stronger for the 2x106 cells/ml concentration, as seen for m24 Ab and mICAM-1 stainings (m24 Ab fluorescence intensity was weaker than that of mICAM-1 multimers, Fig. S6 and S8). Indeed, antigen-stimulated CD8+ T cells aggregated heavily at the higher cell concentration (1x107 cells/ml) and were consequently lost from the FSC-A/SSC-A lymphocyte gate (Fig. S7 and S9). This aggregation depended on the presence of the CMV/NLV peptide during the stimulation, but not on that of m24 Ab or mICAM-1 (Figs. S 6E-9E).
Finally, we assessed the ability of m24 Ab or mICAM-1 to detect rare antigen-specific CD8+ T cells using the optimal concentration of PBMCs (2x106 cells in 1 ml). An HLA-A*02+ donor with approximately 0.04% of A*02/GLC tetramer+ CD8+ T cells was chosen. Staining with m24 Ab detected 0.022% while staining with mICAM-1 detected 0.017% EBV/GLC-specific cells (Fig. S10). However, m24 Ab staining had a lower intensity than that of mICAM-1, and more cells were observed in the control (0.012% m24+ vs 0.004% mICAM-1+), making mICAM-1 a superior reagent when CD8+ T cells alone shall be detected without additional markers.