Biomimetic Nanoparticles Enabled by Cascade Cell Membrane Coating for Direct Cross-Priming of T Cells.

Despite the activation of T lymphocytes by antigen-presenting cells being responsible for eliciting antigen-specific immune responses, their crosstalking suffers from temporospatial limitations and endogenous influencing factors, which restrict the generation of a strong antitumor immunity. Here, cascade cell membrane coating is reported to prepare biomimetic nanoparticles (BNs) that can manipulate the cross-priming of T cells. BNs are obtained from coating nanoparticulate substrates with cell membranes extracted from dendritic cells (DCs) that are pre-pulsed with cancer cell membrane-coated nanoparticles. With a DC membrane that presents an array of cancer cell membrane antigen epitopes, BNs inherit the intrinsic membrane function of DCs, which can directly cross-prime T cells and provoke robust yet antigen-specific antitumor responses in multiple mouse models. Combination with clinical anti-programmed death-1 antibodies demonstrates a robust way of BNs to achieve desirable tumor regression and survival rate. This work spotlights the impact of nanoparticles on direct cross-priming of T cells and supports a unique yet modulate platform for boosting an effective adaptive immunity for immunotherapy.


Introduction
Increasing evidence has shown that immunotherapy is able to successfully fight various classes of cancer by properly regulating the immune system 1,2 .Training T lymphocytes with appropriate antigen specificities stands for a central route to promote tumor regression 3 .To elicit an antigen-specific antitumor immunity, the stimulation of cytotoxic T cell populations by professional antigen-presenting cells (APCs) is critical 4- 6 .Delivering antigen and adjuvant into APCs and reinfusing engineered antigenpresenting APCs are two main strategies to stimulate tumor-specific T cell responses 7,8 .
However, these approaches inevitably depend on the crosstalking between T cells and APCs, as cross-priming of T cells relies on surface markers of APCs 9,10 .Unfortunately, numerous endogenous factors including susceptibility, restricted migration, immune suppression, and metabolic constraints can influence the inherent function of APCs 11- 13 .In addition, the communication between APCs and T cells is often inadequate due to temporal and spatial limitations, resulting in suboptimal antigen presentation and insufficient activation of T cells [14][15][16] .
To tackle these difficulties, various artificial APCs that mimic the function of endogenous counterparts have been engineered to express proper biological signals, such as major histocompatibility complex (MHC) proteins and costimulatory markers [17][18][19] .These artificial cells have been successfully exploited for engaging with and activating T cells 20 .In addition to cell-based artificial APCs, particulate APCs anchoring with requisite ligands including CD28 agonists and epitope-loaded MHC on their surface have emerged as elegant analogues to facilitate antigen presentation [21][22][23][24] .
Despite the potential of artificial APCs for adaptive cell-based therapy and the advantage of synthetic particles to penetrate lymph nodes after subcutaneous administration, the construction of artificial systems is complex and time consuming 25- 27 .Furthermore, epitope-embedded MHC multimer complexes can be engineered only individual epitope of interest has been identified beforehand 28 .Additionally, introducing certain membrane proteins of APCs can solely confer with partial function of the interaction with T cells, largely limiting their application 29 .Therefore, alternative methods that are facile and versatile yet capable of directly cross-priming T cells even with an array of unidentified tumor-associated antigen epitopes are highly desirable.
Here, we report the use of biomimetic nanoparticles (BNs) enabled by cascade cell membrane coating for directly manipulating cross-priming of T cells.Cancer cell membranes with an array of membrane antigens are extracted and coated onto a nanoparticulate substrate that can be easily devoured by dendritic cells (DCs) for antigen processing and presentation.DC membranes presenting related antigen epitopes are subsequently acquired and fused onto a nanoparticulate core to form BNs.
Exceptionally, BNs enabled by the cascade coating retain the entire membrane composition and intrinsic function of DC membranes.Different from endogenous DCs, direct cross-priming of T cells by BNs that bypass the need for regular antigen processing and presentation elicits strong antigen-specific T cell responses.We demonstrate the ability of BNs to boost an extraordinary tumor-specific immunity in multiple tumor models, including ovalbumin expressing B16 (B16-OVA), HPV E6 and E7-expressing TC-1, and Hepa 1-6 tumor-bearing mice.In combination with a clinical immune checkpoint inhibitor, anti-programmed cell death-1 (αPD-1) antibodies, BNs evidence a practical path to enhance tumor growth inhibition and improve survival.Our work suggests a simple yet universal platform for the manipulation of the interaction with T cells and supports the potential of engineered materials to directly cross-priming T cells for immunotherapy.

Preparation of BNs by cascade cell membrane coating
Cell membrane coating, which fuses synthetic cores with natural cell membranes, has emerged as a useful approach to endow coated substrates with foreign functions that are intrinsic unachievable [30][31][32] .For instance, various nanoparticles coated with erythrocyte membranes exhibit prolonged blood circulation due to the presence of a "don't eat me" signal of CD47-SIRPα on the surface 33 .Furthermore, coated nanoparticles can play as decoys to bind and neutralize toxic molecules since the maintenance of the antigenic characteristics of the source cells, such as macrophage and neutrophils [34][35][36][37] .More recently, coating with engineered cancer cell membranes expressing MHC and costimulatory marker CD80 has been explored to mimic DCs for antigen presentation 26 .As primary APCs, DCs execute a vital role for antigen processing and cross-priming of antigen-specific T cells, which can provoke an adaptive immunity for cancer treatment 38,39 .However, DC membranes presenting identified antigen epitopes have rarely utilized to coat synthetic cores for stimulating tumor-specific immune responses.To build DC membrane-coated nanoparticles, we developed a technique of cascade cell membrane coating, which could display a set of cancer cell associated peptide epitopes on the surface (Figure 1a).First, cell membranes with multiple antigens were acquired from tumor cells 40 and subsequently fused onto a polymeric nanoparticulate core to form cancer cell membrane-coated nanoparticles (CCNPs).Then, DCs derived from bone marrow (BMDCs) were incubated with CCNPs for antigen processing and presentation.Lastly, membranes displaying a cluster of exclusive tumor-associated antigen epitopes were extracted from CCNPs-pulsed BMDCs and coated onto a polymeric nanoparticulate substrate to generate DC membrane-coated nanoparticles.
Transmission electron microscopy (TEM) images showed a clear core-shell structure for CCNPs, in which each poly (lactic-co-glycolic acid) (PLGA) nanoparticulate core was visualized surrounding with a cell membrane coating (Figure 1d and Figure S1a).Compared with bare PLGA nanoparticles (Ns), dynamic light scattering (DLS) measurement indicated that the hydrodynamic diameter and surface zeta potential of CCNPs separately increased by ~35 nm (Figure 1b) and ~6 mV (Figure 1c).It was worth noting that the membrane proteins were successfully retained after coating onto PLGA nanoparticles (Figure 1e), which was in good agreement with previous studies 1 .After adding into BMDCs, CCNPs could be internalized rapidly, as confirmed by confocal laser scanning microscopy (CLSM) images (Figure 1f and Figure S1b).Fluorescent colocalization analysis indicated that CCNPs were devoured in its entirety by DCs, as verified by a highly overlapping signals from the cores and membrane coatings of the coated nanoparticles.As expected, typical membrane antigens including the overexpressed OVA and main melanoma-associated transmembrane protein glycoprotein 100 (gp100) of B16-OVA cells were processed and presented by BMDCs (Figure 1g-i).To quantify epitopes presented on the cell membrane, CCNPs with cell membrane proteins (CMP) of 40, 60, 80, 120, or 160 µg were added into 1 × 10 6 DCs.An amount of 40 µg OVA complexed with polyetherimide (PEI-OVA) was cultured as a control 41 .Importantly, OVA epitope presented by CCNPs-matured DCs increased with the quantity of membrane proteins (Figure 1h and i).Even pulsed with CCNPs that had a total protein amount of 40 µg, BMDCs exhibited near 5-times higher OVA epitope level on the surface, in contrast to cells primed with equivalent PEI-complexed OVA.In consideration of the fact that actual quantity of OVA delivered by CCNPs was lower than that of PEI-OVA, enhanced presentation suggested that rather than complexed OVA, antigens embedded in cell membranes could be processed more favorably by DCs.It was noted that incubation with free OVA showed negligible corresponding epitope level on the surface of BMDCs (Figure S1c).These results illustrated that coating nanoparticles with cancer cell membranes represented an efficient way to deliver tumor-associated antigens.
Similarly, PLGA nanoparticles coated with DC membranes that were extracted from B16-OVA cell membrane-coated nanoparticles (termed as BDCNs) appeared a typical core-shell nanostructure (Figure 1j and Figure S3a).The hydrodynamic diameter and surface zeta potential of BDCNs approximated to those of DC membrane vesicles (Figure 1k and l).Similar to other cell membrane-coated nanoparticles 42 , the colloidal stabilities of cancer membrane-coated and dendritic cell membrane-coated nanoparticles could maintain up to 14 days in both PBS and pure water (Figure S2).
The physicochemical properties of the nanoparticles were summarized in Table S1.
Taken together, through translocation of B16-OVA cell membranes onto nanoparticles, a biomimetic structure of cancer cells was occurred for interaction with DCs and subsequent antigen processing and presentation.Take advantage of cascade cell membrane coating, nanoparticles displaying autologous epitopes that were derived from membrane antigens of cancer cells could be accomplished by coating with mature DC membranes.

BNs mediated cross-priming of T cells
We investigated whether BDCNs could execute the function of directly activating T cells in vitro.To study the interaction with T cells, the binding of BNs with the surface receptors of T cells was analyzed by coculture.CD8 + T cells were isolated from the spleen of C57BL/6 mouse and incubated with BDCNs.To examine the specific adhesion with T cells, both Ns and Ns coated with red blood cell membranes (RBC-Ns) or cell membranes extracted from DCs without pre-pulsing with CCNPs were used as controls.As viewed by confocal imaging, T cells were surrounded with numerous BDCNs, as shown by fluorescence overlap between the cell membranes and nanoparticles (Figure 2a and Figure S3b).In comparison with limited non-specific binding of the control nanoparticles, the enhanced fluorescence intensity reflected the specificity of BDCNs to efficiently bind with T cells, which was further supported by flow cytometric analysis (Figure 2b).The activation of T cells could be explained by tightly regulated T cell receptor recognition of antigenic peptides in complex with MHC glycoproteins presented on DC membranes.Then, carboxy fluorescein succinimidyl ester (CFSE) dilution was performed to detect T cell proliferation.CD8 + T cells were incubated individually with Ns, BMDCs and BDCNs.PLGA nanoparticles coated with membranes derived from DCs without pretreatment (DCNs) or with PEI-OVA treatment (ODCNs) were prepared as controls.In striking contrast, BDCNs induced significantly increased proliferation of CD8 + T cells than any other groups.As given in Figure 2c, BDCNs showed ~2-3 times higher activation of CD8 + T cell proliferation in comparison with those of Ns, BMDCs and DCNs, which confirmed the preferential ability of BDCNs to activate CD8 + T cells.To further validate whether BDCNs could initiate activation of antigen-specific T cells, OVA positive CD8 + T cells were labelled using OVA (SIINFEKL) tetramer (tet) antibody 29 .Other than nanoparticles coated with membranes that were extracted from DCs without antigen pulsing, BDCNs could directly prime remarkable proliferation of OVA-specific T cells, which was near 10fold higher than that of DCNs (Figure 2d and e).We then employed antigen-specific epitopes based on OVA and gp100 to assess in vivo T cell production after intravenous injection of BDCNs.It was worth mentioning that comparable biodistribution to other commonly used nanoparticles was observed after intravenous injection of BDCNs 43 .
As expected, these nanoparticles accumulated in the spleen, a major organ for DCs residence.Moreover, the accumulation of BDCNs in tumor increased with time postinjection extending from 0 to 6 hours, as demonstrated by the increase of fluorescence intensity in the sectioned tumor tissue (Figure S4).Mice were intravenously administered with 3 doses of phosphate buffered saline (PBS), Ns, DCNs, ODCNs, or BDCNs on day 0, 5, and 10, respectively (Figure 2f).Spleen tissues were collected and homogenized on day 15 to analyze the percentage of antigen-specific T cells by flow cytometry.Expectedly, the induction of OVA (Figure 2g and i) and gp100 (Figure 2h and j) positive T cells by BDCNs was far exceeded all control groups.In addition, significantly elevated levels of Ki67 expression and interferon-gamma (IFN-γ) production in CD3 + T cells further confirmed that BDCNs promoted the total proliferation and activation of T cells (Figure 2k-m).Interestingly, the similar expression of costimulatory markers CD80 and CD86 on DCs indicated that BDCNs had limited influence on DC maturation (Figure S3e and f).

The comparison of activation efficacy between DCs pulsed with PEI-OVA and
ODCNs was also studied.As shown in Figure S3g and h, mice vaccinated with ODCNs showed significantly elevated levels of CD3 + CD8 + and OVA-specific T cells, which separately increased by near 2-times compared with mice administrated with DCs.In order to measure their susceptibility to environmental conditions, BDCNs were incubated with CD8 + T cells in presence of 10 mM lactic acid, a tumor-derived metabolite that can influence dendritic cell activation and antigen expression 44 .
Interestingly, BDCNs showed 1.7-fold higher proliferation of CD8 + T cells than that of DCs after incubation for 3 days, demonstrating their less susceptible to the insult of lactic acid (Figure S3c).The insusceptibility of BDCNs was further validated by near 1.4-fold higher activation efficacy compared to DCs after 24 hours storage at 4 ℃ (Figure S3d).These data implied the superiorities of BNs over living DCs particularly in activation of T cells, insusceptibility to the tumor microenvironment, and storage condition.

Antigen-specific antitumor responses
We next examined the function of BDCNs in tumor-bearing mice, which were subcutaneously inoculated with B16-OVA in the right hind leg.Mice were daily administered with BDCNs at predetermined time points [45][46][47] (Figure 3a).Experimental endpoint was defined as the tumor size of PBS group exceeded 2000 mm 3 .Peripheral blood, spleen and tumor were sampled for further analysis.Mice treated with PBS and equivalent nanoparticles of Ns, DCNs, or ODCNs were individually used as controls.
Encouragingly, mice administered with BDCNs emerged significantly increased percentages of CD8 + and Ki67 + in CD3 + T cells in the spleen, which disclosed systemic facilitation of T cell proliferation (Figure 3b and c).Moreover, the higher levels of CD80, CD86 and MHC-II in splenic DCs represented a more mature phenotype induced by BDCNs (Figure 3d-f).On the contrary, decreased CD3 + CD4 + T cells and increased CD8 + /Foxp3 + CD25 + CD4 + ratio in comparison to the control groups of PBS, Ns, DCNs and ODCNs might indicate that immunosuppressive environment was relieved (Figure 3g and S5a).Furthermore, BDCNs could elevate IFN-γ production from T cells in the spleen tissue (Figure 3h).Particularly, the increase of OVA tet + T cells demonstrated that BDCNs elicited immune responses in an antigen-specific manner (Figure 3i).
Consistently, appropriate priming of tumor-infiltrating lymphocytes mediated by BDCNs was significantly higher than all the controls (Figure 3j-p).Expectedly, negligible CD4 + CD3 + T cells were found in tumor site after vaccination with BDCNs (Figure S5b).The higher concentrations of IFN-γ (Figure 3q) and TNF-α (Figure 3r) in serum sampled from BDCNs-treated mice further revealed the elicitation of a robust systemic immune response.Additionally, in contrast to other groups, immunofluorescence staining images depicted more infiltrated CD8 + T cells and upgraded cytokine production (Figure 3u and Figure S5c) in the tumor tissue of BDCNs-treated mice.Moreover, immunohistochemistry staining images implied improved accumulation of CD3 + T cells and CD8 + T cells in tumor site after treatment with BDCNs (Figure S6).The promotion of potent immune responses was further supported by the most efficient inhibition of tumor growth after treatment with BDCNs (Figure 3s, t and Figure S5d).Briefly, BNs could generate a strong antigen-specific antitumor immunity in an aggressive yet OVA-expressing melanoma model.
A common tumor model of Hepa 1-6 hepatocellular carcinoma was further developed to verify the generality of BNs for utilization in other cancers without identified antigen epitopes.BNs were prepared by coating PLGA nanoparticles with cell membranes extracted from mature DCs that were pre-pulsed with Hepa 1-6 cell membrane-coated nanoparticles and defined as HDCNs.The tumor-bearing mice were similarly treated, as illustrated in Figure 4a.Consistent with results obtained in B16-OVA tumor model, the percentages of CD8 + and Ki67 + in CD3 + T cells (Figure 4b and    c) and the levels of costimulatory signals CD80, CD86 and MHC-II in DCs were remarkably upgraded in the spleen sampled from mice injected with HDCNs, in comparison with the controls of PBS, Ns, and DCNs (Figure 4d-f).The production of IFN-γ from T cells remained the highest level (Figure 4g), while CD3 + CD4 + T cells were downregulated and CD8 + /Foxp3 + CD25 + CD4 + ratio was upregulated after treatment with HDCNs (Figure 4h and i).Together with the upgraded secretion of TNFα (Figure 4j) and IFN-γ (Figure 4k) in the serum, these data claimed that HDCNs could initiate a stronger systemic tumor-specific cytotoxic T lymphocyte (CTL) response in contrast to the controls.In parallel, the proportions of tumor-infiltrating CTLs (Figure 4l), Ki67 + (Figure 4m), DC mature markers (Figure 4n-p), CD8 + /Foxp3 + CD25 + CD4 + ratio (Figure 4q) and inflammatory cytokines (Figure 4t and Figure S7b) were the highest in the HDCNs treatment group, whereas the population of CD3 + CD4 + T cells maintained the lowest in all experimental groups (Figure S7a).The tumor growth after treatment with HDCNs retarded notably, with part of the tumors eradicated completely, suggesting the stimulation of a potent antitumor response (Figure 4r and s).Further supported by haematoxylin eosin (H&E) staining (Figure S7c), the results found in Hepa 1-6 tumor model reasoned that BNs could be exploited to treat tumors in absence of identified antigens.

Cross-validation of BNs-leveraged antitumor immunity
We further cross-validated the antitumor immunity in B16-OVA and HPV E6 and E7expressing TC-1 tumor models to identify the specific and autologous characteristic of BNs.Similarly, BNs were fabricated by fusing PLGA nanoparticles with cell membranes sampled from BMDCs that were pre-treated with TC-1 cell membranecoated nanoparticles and abbreviated as TDCNs.Mice inoculated with B16-OVA cells were treated with TDCNs, while TC-1 tumor-bearing mice were inversely injected with BDCNs (Figure 5a).Both PBS and the corresponding BNs were used as controls.In B16-OVA tumor model, the application of TDCNs displayed negligible tumor growth inhibition, which was comparable to that of PBS control (Figure 5b and c).On the contrary, BDCNs showed notable tumor regression, which resulted in an extension in median survival time by 12 days (Figure 5d).Rather than TDCNs, the distinct potency of BDCNs to inhibit B16-OVA tumor growth explained that the elicitation of an effective tumor-specific immunity depended on the presence of autologous peptide epitopes on the surface.In TC-1 tumor model, BDCNs lost their ability to suppress tumor growth and provided neglectable survival benefit over PBS control (Figure 5e and f).As expected, TDCNs slowed the growth of TC-1 tumor and prolonged the survival over the controls of PBS and BDCNs (Figure 5g).Meanwhile, the body weight of the treated mice had no evident fluctuations during the experiment, clarifying the limited side effects of these BNs (Figure S8).Treatment with BDCNs and TDCNs in two cross animal models demonstrated the specificities and personalities of BNs in the initiation of antitumor immune responses.

Combination with clinical αPD-1 antibody
Immune checkpoint blockade by preventing PD-1 on T cells from binding with programmed death ligand 1 (PD-L1) can activate CTLs to attack cancer cells [48][49][50] .The use of such inhibitors has elicited durable clinical response and long-term remissions in a fraction of cancer patients 51 .However, a large fraction of patients has failed to respond to these agents, showing a low objective response rate of anti-PD therapy 52 .
We speculated that the largely augmented immune responses by direct cross-priming of T cells could address the problem of anti-PD therapy, particularly in solid tumors.Moreover, the splenic and intratumoral CD3 + CD4 + T (Figure 6p and Figure S9a) decreased tremendously, implying the lowered negative effect in immune stimulation.
In addition to the priming assay, the strikingly upgraded expressions of IFN-γ and TNFα in the serum demonstrated the robust systemic immune responses induced by the combination treatment (Figure 6q and r).Immunofluorescence images including TNFα together with IFN-γ (Figure 6s) and CD8 (Figure S9b) crowded in the tumor site further validated the promotion of a strong antitumor immunity (Figure S9c).In contrast to the limited beneficial effect from αPD-1 therapy alone, the combination therapy delayed the tumor growth, which accompanied with an extension of median survival period from 25 to 40 days (Figure 6t-v).Hepa 1-6 carcinoma model was further used to evaluate the broad applicability of the combination of BNs with αPD-1 for implementation in other cancers in absence of identified antigen epitopes.As shown in Figure 7, the stimulation of T cells (Figure 7a-d), DC maturation (Figure 7e-g), the secretion of cytokines IFN-γ and TNF-α (Figure 7i and j) were greatly improved by the combination of HDCNs and αPD-1, which reversely suppressed the CD3 + CD4 + T cells (Figure 7h).Different from continuous tumor growth in mice treated with αPD-1, the combination treatment was able to abolish all the tumors (Figure 7k and l), benefiting from the ability of HDCNs to elicit a powerful antitumor immunity.
In summary, we have described the use of BNs enabled by cascade cell membrane coating to manipulate the cross-priming of T cells.BNs have been fabricated through coating PLGA nanoparticles with cell membranes extracted from DCs that have been matured with cancer cell membrane-coated nanoparticles.Due to the maintenance of an intact composition of DC membranes, BNs presenting an array of cancer cell membrane antigen epitopes on the surface can inherit intrinsic membrane function of DCs to cross-prime T cells.Significantly, BNs can directly interact with T cells and have elicited a robust antigen-specific antitumor immunity in multiple mouse models including B16-OVA, HPV E6 and E7-expressing TC-1, and Hepa 1-6 tumors.In combination with clinical αPD-1, BNs have demonstrated a practical way to accomplish desirable tumor regression and survival rate.To this end, treatment efficacy and injection frequency of BNs need to be rationally optimized by using immunoadjuvants or adjusting dosage, particle size, and the loading of antigens for future translation.
Beyond the specific use of BNs for cross-priming of T cells, the direct interaction with other immune cells may inspire new methods that can manipulate immune responses for treating other diseases, such as autoimmune disease, infections and inflammations, studies that are currently ongoing.
Therefore, we combined BNs with clinical αPD-1 and evaluated the antitumor responses in a more clinically relevant therapeutic setting.The combined modality was first tested in B16-OVA tumor model.In contrast to αPD-1 alone, indicators of upgraded immune responses associated with T cells and APCs together with inflammatory cytokine IFN-γ both in the spleen (Figure 6a-h) and tumor (Figure 6i-o) warranted extremely positive function of the combination of αPD-1 with BDCNs.

Figure 1 .
Figure 1.Preparation and characterization of BNs.(a) Schematic illustration of BNs enabled by cascade cell membrane coating for direct cross-priming of T cells.(b) Sizes and (c) Zeta potentials of Ns, B16-OVA cell membrane vesicles (CMV), and CCNPs, respectively.CCNPs were prepared by extruding PLGA nanoparticles with cell membranes at a cell membrane protein/nanoparticle weight ratio of 1:1.(d) Typical TEM images of Ns (left) and CCNPs (right) negatively stained with uranyl acetate.Scale bar: 100 nm.(e) Western blot analysis of membrane-specific protein marker OVA on B16-OVA cells and CCNPs, respectively.(f) Colocalization of PLGA cores (DiD, red channel) and cancer cell membranes (Neuro-DiO, green channel) upon BMDC uptake by confocal imaging.Cell nucleus was stained with DAPI (blue channel).Scale bar: 10 μm.(g) Processing and presentation of antigen gp100 associated with B16-OVA cell membranes by BMDCs.(h) Flow cytometric profiles and (i) mean fluorescence intensities of BMDCs pulsed by CCNPs with different amounts of membrane proteins.BMDCs were stained with OVAp-H-2k b -APC.BMDCs, isotype, and PEI-complexed OVA were used as controls.(j) Representative TEM images of BDCNs and DC cell membrane vesicles (DCCMV) negatively stained with uranyl acetate.Scale bar: 100 nm.(k) Sizes and (l) zeta potentials of DC cell membrane vesicles and BDCNs, respectively.BDCNs were prepared by extruding through a polycarbonate membrane with a pore size of 200 nm.Error bars represent the standard deviation (n = 3).

Figure 2 .
Figure 2. Direct cross-priming of T cells by BNs.(a) Representative confocal images and (b) flow cytometric analysis of primary CD8 + T lymphocytes after incubation with Ns, RBC-Ns or BDCNs