Proper Regrafting of Ig-Like Transcript 2 after Trogocytosis Allows a Functional Cell–Cell Transfer of Sensitivity

Trogocytosis is a mechanism of fast cell-to-cell contact-dependent uptake of membranes and associated molecules from one cell by another (1). Intercellular transfers of membrane patches are observed during interactions between immune cells (reviewed in Ref. 2). Trogocytosis of ligands is well established, in vitro in humans and also in vivo in mice (3–8), in which MHC class I, MHC class II, and costimulatory molecules can transfer from APC to T cells. The ligands acquired through trogocytosis may temporarily transfer functions of the donor cells to the acceptor cells. For example, 1) CD8⁺ T cells that acquired their MHC class I ligands became the targets of fratricide Ag-specific cytolysis (3); 2) T cells that acquired HLA-DR and CD80 could stimulate resting T cells (2); and 3) CD4⁺ T cells that acquired HLA-G behaved as suppressor cells (9). Hence, acquisition of exogenous ligands by T cells endows them with new outward signaling properties and enables them to influence the functions of surrounding immune cells.

Transfers of receptors and inward signaling through acquired receptors have been much less studied, even though work has started in this direction. For instance, it has been shown that constitutively activated molecules that were acquired by intercellular transfers may signal to their acquirer cells. However, in these reports involving cell lines, receptor–ligand interactions played no role because of the constitutive activation of the transferred molecules (10, 11). By contrast, a recent report investigated the functional consequences of the trogocytic transfer of FcRs from murine APC to murine T cells. In this report, it was shown that FcRs could bind with their ligands after transfer, but could not transduce signals to their new host cells (12). Hence, the capability of a transferred receptor to bind to its ligand, transduce a signal, and function remains to be demonstrated, especially for nontransformed human cells. This requires the demonstration of: 1) the proper insertion of the acquired receptor within the plasma membrane of the new host cell; 2) the capability of the transferred receptor to access the intracellular machinery of the new host cell; and 3) the functional consequences of its binding to a ligand.

In this study, we investigated these issues using the transfer of the Ig-like transcript 2 (ILT2) inhibitory receptor (LILRB1/CD85j) from monocytes to ILT2-negative autologous T cells. The T cells we used were ILT2-negative, but nevertheless, it has been shown that in cases in which ILT2 is expressed by T cells, ILT2 engagement inhibits their proliferation and cytolytic functions (13). In these cases, ILT2 inhibitory functions are mediated by tyrosine phosphorylation of ITIM-like sequences in its cytoplasmic tail and involve the p56lck kinase and recruitment of Src homology region 2 domain-containing phosphatase-1 and -2 phosphatases (14). In ILT2-transfected cells, ILT2 phosphorylation could be induced by cross-linking or by the protein tyrosine phosphatase inhibitor pervanadate (PV). ILT2 signaling depends on binding to HLA class I molecules, and ILT2’s highest affinity ligand is HLA-G (for a review on HLA-G, see Ref. 15).

In this study, we demonstrate the ILT2 transfers from monocytes to autologous activated T cells by trogocytosis and the proper integration of transferred ILT2 within the plasma membrane of a new host cell. Furthermore, we show that transferred ILT2 behaved as an endogenously produced molecule, capable of using the intracellular biochemical machinery of its new cell and send inward-oriented signals, which are then responsible for the acquirer T cell functional inhibition. These data are a formal demonstration that a transferred molecule may send signals to its new host cell after ligand engagement. They imply that sensitivity to modulatory molecules can be acquired from other cells.

Blood was obtained from healthy volunteers from French Blood Bank under a protocol approved by the Institutional Review Board of the Saint-Louis Hospital, Paris, France.

For all purification steps, FcRs on PBMC were blocked using 20 μg/ml human IgG (Sigma-Aldrich) for 30 min.

For monocyte isolation, PBMCs were labeled with 20 μg/ml anti-CD14 and then positively separated using goat anti-mouse (GAM)-coated magnetic beads according to the manufacturer’s specifications (Ademtech). To obtain highly purified CD4⁺ or CD8⁺ T cells, PBMCs were labeled with anti-CD4 or anti-CD8 and then positively separated using GAM-coated magnetic beads. Positively isolated cells were then incubated overnight at 37°C and repeatedly washed to increase the purity. To obtain negatively purified CD4⁺ or CD8⁺ T cells, PBMCs were labeled with a mixture of anti-CD14, anti-CD16, anti-CD19, and anti-CD8 or anti-CD4 followed by immunomagnetic depletion using GAM-coated magnetic beads. ILT2-positive cells were removed using 20 μg/ml purified blocking Ab against ILT2 (GHI/75) and then positively separated using GAM-coated magnetic beads. ILT2 expression was evaluated by flow cytometry.

Cell lines used were previously described (9).

Purified monocytes were activated with 100 ng/ml LPS (Sigma-Aldrich) for 3–5 d. PBMCs or purified T cells were activated for 48 h with 4 μg/ml leucoagglutinin (PHA-L; Sigma-Aldrich) and then cultured in IL-2–supplemented medium (100 U/ml; Sigma-Aldrich) for 2–4 more d prior to use. Cells were not used beyond these times.

The following Abs were used from Exbio (Prague, Czech Republic): purified anti-CD4, -CD8, -CD14, and -CD19, blocking anti–HLA-G 87G Fab, nonblocking anti–HLA-G Mem-G/9, and FITC-conjugated anti-CD4, -CD8, and -CD14; from Beckman Coulter: purified anti-CD28, anti-CD16, PC5-conjugated anti-CD3, and anti-CD25; from BD Biosciences: purified and PE-conjugated anti-ILT2 (clone GHI/75); from Santa Cruz Biotechnology: purified anti-ILT2 (clone VMP55); from Invitrogen: Alexa Fluor 488 GAM IgG1; and from Upstate Biotechnology: anti-phosphotyrosine 4G10. Purified anti-CD3 (OKT3) was provided by Janssen-Cilag.

For flow cytometry experiments, FcRs were blocked in 25% human serum supplemented with 20 μg/ml human IgG serum prior to staining. All FcR blocking conditions were maintained during the entire duration of the experiments. Appropriate isotypic controls were systematically used to evaluate nonspecific binding. PKH67 or PKH26 dyes (Sigma-Aldrich) were used for fluorescent labeling of cell membranes and the sNHS-Biotin kit (Uptima) was used for cell-surface biotinylation according to the manufacturers’ specifications.

T cells were conjugated with PKH26-labeled monocytes at 37°C and left adhered on poly-l-lysine–coated slides for 5 min at 37°C. The cells were then fixed for 10 min with 3% paraformaldehyde, stained using an anti-ILT2 (VMP55) followed by GAM Alexa Fluor 488, and analyzed using a Carl Zeiss LSM 510 confocal microscope (Carl Zeiss).

Trogocytosis assays were performed as previously described (9). Briefly, highly purified CD4⁺ and CD8⁺ T cells were cocultured for 30 min with purified autologous monocytes at a 1:1 ratio, to a total concentration of 10⁶ to 10 × 10⁶ cells/ml, at 37°C in a 5% CO₂ humidified incubator. All further steps were performed on ice.

For acid wash, cells were washed twice in PBS and resuspended for 4 min at 20°C in citrate buffer (0.133 M citric acid and 0.066 M Na₂HPO₄ [pH 3.3]) at a density of 5 × 10⁶ cells/ml. The treatment was stopped by addition of 10% FCS in RPMI 1640 and 10 mM HEPES.

Prior to the coincubation with autologous T cells, the monocytes’ cell-surface proteins were biotinylated using sNHS-Biotin (Uptima). At the end of the coincubation, T cells were purified, lysed, and the presence of biotinylated ILT2 was investigated by immunoprecipitation of biotinylated proteins with streptavidin-agarose beads (Sigma-Aldrich) followed by electrophoretic separation on a 10% SDS-PAGE acrylamide gel, transfer to nitrocellulose membranes, and immunoblotting with anti-ILT2 (VMP55 clone).

Cells treated with the phosphatase inhibitor PV (200 μM sodium orthovanadate and 200 μM H₂O₂ at 37°C for 10 min) were lysed in 1% Triton X-100, 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA, 0.2 mM sodium orthovanadate, 0.2 mM PMSF, 0.5% Nonidet P-40, protease inhibitors (Sigma-Aldrich), and phosphatase inhibitors cocktails I and II (Sigma-Aldrich). Lysates were precleared with protein G-Sepharose beads (Amersham Biosciences) and subjected to immunoprecipitation with 2 μg purified anti-ILT2 (clone GHI-75; BD Biosciences) conjugated to protein G-Sepharose beads. Immunoprecipitates were separated by standard SDS-PAGE and transferred to nitrocellulose membranes (Amersham Biosciences). Immunoblotting was performed with anti-phosphotyrosine Ab (4G10; Upstate Biotechnology) or anti-ILT2 followed by GAM-HRP (Sigma-Aldrich) or with streptavidin-HRP (Sigma-Aldrich). The membrane was developed using ECL⁺ Western blotting detection reagents (Amersham Biosciences).

For blocking HLA-G–ILT2 interactions, FcRs were first blocked by incubation of the cells 30 min in 25% human serum supplemented with 20 μg/ml human IgG. Cells were then incubated with 10 μg/ml blocking Ab anti–HLA-G 87G or nonblocking anti–HLA-G MEM-G/9 or blocking Ab anti-ILT2 GHI/75, or their isotypic controls, prior to use. All blocking conditions were maintained during the subsequent experiments.

Proliferation was measured by tritiated thymidine ([³H]thymidine) incorporation (1 μCi/well; Amersham Biosciences) on a β-counter (Wallac 1450; Amersham Biosciences). All samples were run in triplicate. Appropriate proliferation and isotypic controls were included for each experiment on each plate.

A total of 5 × 10⁴ proliferating cells (CD4⁺ILT2^{Acq +}or CD4⁺ T cells) was cocultured with 2.5 × 10⁴ cells gamma-irradiated HLA-G–expressing cells lymphoblastoid cell line (LCL)–HLA-G1 (75 Gy) and their negative counterparts LCL-RSV and plated in a final volume of 200 μl. [³H]thymidine was added at the time of incubation and proliferation was measured 18 h later.

A total of 5 × 10⁴ responder naive PBMCs or highly purified CD4⁺ T cells from these PBMCs or highly purified CD4⁺ T cells incubated with autologous monocytes at a 1:1 ratio (CD4⁺ T cells plus monocytes) were stimulated with 2.5 × 10⁴ gamma-irradiated LCL-RSV cells or LCL–HLA-G1 cells (75 Gy). Proliferation was measured from day 1–6 every 24 h. Separate plates were set up for each time point, and [³H]thymidine was added 18 h before collection.

Cytolytic activity of CD8⁺ T cells was assessed in a standard 4 h [⁵¹Cr]release assay against ⁵¹Cr-labeled HLA-G–negative melanoma M8 cells transfected by a control vector (M8-pcDNA) or the same vector containing HLA-G1 cDNA (M8–HLA-G1). (M8 cells and the inhibition of CTL killing are detailed in Refs. 16, 17). Briefly, effector cells were mixed with 5 × 10^{3 51}Cr-labeled target cells (Amersham) at different E:T ratios in U-bottom microtiter plates. After 4 h of coincubation at 37°C in a humidified 5% CO₂ incubator, supernatant was collected for liquid scintillation counting (Wallac 1450 Microbeta; Pharmacia). Percentage of specific lysis was determined as (cpm experimental well − cpm spontaneous release)/(cpm maximum release − cpm spontaneous release) × 100. Spontaneous release was determined by incubation of labeled target cells with medium. Maximum release was determined by solubilizing target cells in 1% Triton X-100. For each experiment, triplicate samples were used.

Data are presented as means ± SD. Student t test was used, and a p value <0.05 was taken to be significant. For figures showing representative experiments, error bars represent SD of triplicates.

ILT2 can be expressed at the surface of a minor fraction (0–25%) of resting CD4⁺ and CD8⁺ T cells in healthy individuals (13, 18), and this proportion is increased in activated and Ag-specific T cells (18–20). In the experiment described in Fig. 1, we used freshly isolated PBMCs for which no ILT2 was detected at the surface of CD4⁺ and CD8⁺ T cells. In agreement with the data described above, PHA plus IL-2 activation significantly increased ILT2 expression by CD4⁺ and CD8⁺ T cells (Fig. 1A). However, this upregulation did not occur if ILT2⁺ cells had been removed from PBMC prior to PHA plus IL-2 activation (Fig. 1B). In the reverse experiment, when highly purified (>99%, no monocytes detected) CD4⁺ or CD8⁺ T cells were activated by anti-CD3 plus anti-CD28, no ILT2 cell-surface upregulation was observed (Fig. 1C), whereas an additional 30-min coincubation of these purified and activated T cells with autologous ILT2⁺ monocytes was sufficient to restore the original activation-dependent ILT2 cell-surface upregulation. These data show that ILT2 upregulation might not be a feature of T cell activation per se but dependent on the presence of ILT2⁺ cells such as autologous monocytes. The absence of ILT2 expression by T cells purified prior to activation was confirmed by intracellular flow cytometry (Supplemental Fig. 1A) and confocal microscopy using another anti-ILT2 Ab, VMP55 (Supplemental Fig. 1B).

Out of all of the possible explanations of the fact that monocytes are required for activation-dependent T cell surface expression of ILT2, we show in this study that ILT2 is actually not upregulated by T cells themselves, but acquired from monocytes by T cells through the transfer of ILT2-containing membrane patches, a mechanism known as trogocytosis.

The hallmarks of trogocytosis are kinetics in the order of minutes, transfer of membrane patches and not of individual molecules, cell-cell contact dependence, and a limited lifespan of the acquired molecules at the surface of the acquirer cell (1, 2). Fig. 2A shows that when monocytes for which membranes had been labeled with a lipophilic dye (PKH67) were coincubated with activated autologous CD4⁺ T cells for 30 min or less, a correlation between T cell positivity for PKH67 and ILT2 was observed. This shows that ILT2-positive T cells had acquired membranes from monocytes. Supplemental Fig. 2A shows that ILT2 acquisition was dependent on cell-to-cell contact, and Supplemental Fig. 2B shows the limited lifespan of acquired ILT2 at the T cell surface, which indicates that ILT2 was not endogenously produced by T cells. Real-time quantitative PCR experiments also indicated that ILT2-negative activated CD4⁺ T cells expressed very low amounts of ILT2 mRNA prior to use and that these levels were not significantly increased after trogocytosis experiments (data not shown). Moreover, in accordance with previously published data (21), Supplemental Fig. 3 shows that CD4⁺ and CD8⁺ T cells acquire ∼1% of membranes of monocytic origin and between 1 and 0.5% of monocytic ILT2 proteins after a 30-min coincubation with autologous monocytes. Fig. 2B illustrates by confocal microscopy that the only ILT2 expressed by CD4⁺ T cells after coincubation with monocytes colocalizes with membranes acquired from monocytes.

To confirm the exogenous origin of the ILT2 receptors displayed on T cells, surface-biotinylated monocytes were coincubated with T cells. After coincubation, the presence of biotinylated ILT2 was sought using streptavidin-agarose immunoprecipitation and Western blotting with anti-ILT2. Biotinylated ILT2 receptors were easily detected in T cells, demonstrating their monocytic origin (Fig. 2C, lower panel). Immunoprecipitation of ILT2 on control T cells (Fig. 2C, upper panel) demonstrated that the only ILT2 detectable on T cells was transferred ILT2. These data demonstrate that T cells can acquire ILT2 receptors from autologous monocytes by trogocytosis.

Membranes transferred from a cell to another may not integrate, but may remain affixed onto the acceptor cell (i.e., outside). Obviously such a transfer is unlikely to allow signaling of the acquired molecules to the acquirer cell (2, 22). Membranes that are affixed on acceptor cells can be removed by a mild acid wash treatment (pH 3.3) (23). A pH 3.3 treatment (Fig. 3A) was efficient and removed β₂-microglobulin, leading to a loss of staining for HLA class I Ags using an Ab directed against conformed HLA class I molecules (W6/32). However, this treatment did not remove the PKH-labeled membranes of monocytic origin, or the ILT2 they contained, from the CD4⁺ T cell surface. This ruled out the possibility that the acquired membranes were just affixed onto the surface of T cells.

If properly inserted, transferred membranes and molecules should spatially behave as endogenously produced ones, and move, or diffuse within the acceptor cell’s plasma membrane. Fig. 3B shows by confocal microscopy a representative example of our observations. Early within the first 30 min of coincubation, most of the transferred ILT2 at the surface of CD4⁺ T cells remained within patches of PKH-labeled monocytic membranes. At later times, ILT2 receptors were also observed mainly within PKH-labeled monocytic membrane patches. However, an increasing proportion of cells that showed ILT2 receptors outside PKH-labeled membranes of monocytic origin was observed when incubation time increased. Representative images are shown in Fig. 3B. Three-dimensional images of T cells with PKH-free ILT2 are provided in Supplemental Video 1.

The independent movements of dyed lipids and ILT2 show that at least some ILT2⁺ monocytic membranes integrated properly, even though we cannot know whether ILT2 molecules exited the original membrane patch, or whether labeled lipids diffused out of the area beneath ILT2.

It was shown that ILT2 phosphorylation is induced in T cells by the protein tyrosine phosphatase inhibitor PV (14). Thus, we coincubated activated CD4⁺ T cells and biotinylated monocytes, CD4⁺ T cells that had acquired ILT2 receptors (CD4⁺ILT2^BiotinAcq+) were then sorted, treated with PV, lysed, and the phosphorylation of acquired ILT2 receptors was determined by Western blotting with streptavidin-HRP or antiphosphotyrosine. Fig. 4 shows once again that CD4⁺ T cells did not express endogenous ILT2 prior to coincubation with monocytes and acquired biotinylated ILT2 during the coincubation with biotinylated autologous monocytes. Fig. 4 also shows that PV treatment had no effect on the level of acquired ILT2 in T cells, but was able to induce ILT2 phosphorylation in CD4⁺ILT2^BiotinAcq+ T cells. This demonstrates that acquired ILT2 receptors were accessible to the biochemical machinery of the acceptor T cells. Purity of the T cell population and contamination controls ensured that the observed ILT2 phosphorylation was not related to the presence of contaminating monocytes.

Thus, these data demonstrate that acquired ILT2 receptors are properly inserted within the plasma membrane of T cells, can diffuse within the plasma membrane, and that transferred ILT2 was accessible to components of the T cell intracellular machinery.

It was shown that engagement of T cell ILT2 with HLA-G induces inhibition of CD4⁺ T cell proliferation (9, 24) and CD8⁺ T cell cytolytic function (25). We investigated whether engagement of acquired ILT2 with HLA-G was able to cause T cell functional inhibition.

We first showed that transferred ILT2 receptors were able to mediate the inhibition of CD4⁺ T cell ongoing proliferation upon engagement with HLA-G. For this, we generated purified, polyclonally activated ILT2-negative CD4⁺ T cells and had them acquire or not ILT2 from autologous monocytes (CD4⁺, CD4⁺ILT2^Acq+ T cells, respectively). These T cells were already proliferating at the start of the experiment, and their original lack of ILT2 was assessed by flow cytometry. HLA-G–expressing cells were then added and T cell proliferation measured. Fig. 5A shows that the proliferation of ILT2-negative CD4⁺ T cells was not inhibited by HLA-G, but that of CD4⁺ILT2^Acq+ T cells was completely stopped in the presence of HLA-G. The inhibition of CD4⁺ILT2^Acq+ T cells was directly due to the interaction between acquired ILT2 receptors and HLA-G. Indeed, anti–HLA-G (87G) and anti-ILT2 (GHI/75) blocking Abs, but not nonblocking anti–HLA-G Ab (Mem-G/9), restored the ongoing proliferation of CD4⁺ILT2^Acq+ T cells. Furthermore, proliferation of CD4⁺ILT2^Acq+ T cells was unaffected when HLA-G-negative control cells were used in the coincubation.

We next showed that acquired ILT2 receptors could inhibit CD8⁺ T cell cytotoxic function. For this, we set up cytotoxicity assays between purified activated CD8⁺ T cells that originally did not express ILT2 and had acquired ILT2 or not from autologous monocytes (CD8⁺ and CD8⁺ILT2^Acq+ T cells, respectively) and HLA-G–positive targets. Fig. 5B shows that ILT2-negative CD8⁺ T cells lysed their targets regardless of HLA-G. CD8⁺ILT2^Acq+ T cells lysed HLA-G–negative targets but their cytotoxic functions were blocked by HLA-G–expressing target cells. This inhibition was directly due to HLA-G–ILT2 interaction, as shown by the use of blocking Abs against HLA-G (87G) or ILT2. Isotypic controls or the nonblocking anti–HLA-G Mem-G/09 did not restore the cytotoxic activity of CD8⁺ T cells against HLA-G–positive target cells.

Thus, these data show that activated ILT2-negative CD4⁺ and CD8⁺ T cells are intrinsically insensitive to ILT2 ligands (e.g., HLA-G) until they can borrow ILT2 from ILT2-expressing cells such as monocytes by trogocytosis of ILT2-containing membranes. Given that resting T cells have little or no trogocytic capability (9), these data also indicate that T cell inhibition through ILT2 engagement should concern only fully activated T cells.

We next demonstrated that trogocytosis-dependent ILT2-mediated inhibition held true in situations in which resting T cells were involved. For this demonstration, we studied ILT2-mediated inhibition of resting T cell responsiveness to allogeneic stimulation using HLA-G as ILT2 ligand, because it is known that HLA-G inhibits the alloproliferative response of PBMC, as measured by thymidine incorporation at day 6 or 7 after allostimulation (9, 26, 27).

Thus, we set up alloproliferative experiments between irradiated HLA-G–positive or HLA-G–negative LCL as stimulator cells and freshly isolated PBMC or highly purified CD4⁺ T cells from these PBMC (CD4⁺ T cells), or highly purified CD4⁺ T cells from these PBMC mixed with autologous monocytes (CD4⁺ T cells plus monocytes), as responder cells. PBMC used in these experiments were chosen for their lack of ILT2 expression on T cells by flow cytometry performed at the time of isolation. In this system, allostimulation was provided by control or HLA-G–transfected LCL cells, and ILT2 was provided by autologous monocytes. To correlate ILT2 acquisition with alloproliferation inhibition, the ILT2 expression and the proliferation of the responder cells were measured every 24 h for 6 d.

For the first 3 d of the experiment, no difference was observed in the phenotype and proliferation of CD4⁺ T cells. Thus, in Fig. 6, time points until day 3 are shown as a unique 0→3 d time point. For illustration, day 3 data were arbitrarily chosen but do not differ from day 0, showing in particular no ILT2 expression on CD4⁺ T cells.

As shown by the first two plots in Fig. 6A, CD4⁺ T cells within a responder population of total PBMC remained negative for cell-surface ILT2 for 4 d. Consistently, as shown by the first two bar graphs, these T cells remained insensitive to inhibition by HLA-G during the first 4 d of allostimulation. As a consequence, the daily levels of proliferation of these PBMCs were identical regardless of whether HLA-G was present in their environment or not. These data on the first days of PBMC allostimulation clearly show that ILT2 does not inhibit the Ag selection and the initial activation steps of resting T cells or even possibly their first proliferation cycles. After day 4, a proportion of CD4⁺ T cells displayed cell-surface ILT2, and this proportion increased through day 6 (rightmost two plots of Fig. 6A). This means that a proportion of CD4⁺ T cells became trogocytic as of day 4 and that this proportion increased through day 6. The proportion of ILT2-positive CD4⁺ T cells was nevertheless much lower than in Figs. 1–5, but this can be explained by the fact that allogeneic activation such as in Fig. 6 concerns only a fraction of resting T cells, whereas polyclonal activation as in Figs. 1–5 concerns all. The ILT2-positive T cells in Fig. 6 would then correspond to effectors generated through Ag-specific activation. This also means that it took 4 d from a resting T cell to become trogocytic upon allostimulation, which matches the kinetics of trogocytic capability induction through allostimulation and polyclonal activation described in this study and elsewhere (9). As can be seen in the corresponding last two bar charts of Fig. 6A, it is from the point when T cells started acquiring ILT2 that they reacted to the presence of HLA-G and that their proliferation decreased compared with controls. At day 6, the HLA-G–mediated alloproliferation inhibition that could be observed was typical of what has been reported elsewhere (26–28). The observed alloproliferation inhibition was directly due to HLA-G–ILT2 interaction and therefore to ILT2 transfer from monocytes to activated T cells, because it could be prevented by blocking Abs directed against HLA-G (87G) or ILT2, respectively, whereas a nonblocking anti–HLA-G Ab (MEM-G/9) had no effect.

By contrast with the results obtained when total PBMCs were allostimulated, highly purified CD4⁺ responder T cells, which had no access to ILT2 exogenously produced by monocytes, remained cell-surface ILT2 negative throughout the 6 d of the experiment (Fig. 6B, plots) and also remained mostly insensitive to HLA-G–mediated proliferation inhibition (Fig. 6B, bar graphs). However, sensitivity to HLA-G could be restored if the very same highly purified ILT2-negative CD4⁺ T cells were coincubated with autologous ILT2⁺ monocytes at a 1:1 ratio (Fig. 6C). In this case, the kinetics of ILT2 display by CD4⁺ T cells and those of alloproliferation inhibition matched those of control PBMC.

Thus, these data show that membrane transfers are required for the in vitro ILT2-mediated inhibition of T cell alloproliferative responses and that the parameters of trogocytosis dictate which cell is concerned as well as the kinetics of ILT2-mediated inhibition.

A minor proportion of T cells have been shown to be able to express ILT2 at the resting stage, a proportion that was shown to increase with activation and even more so in cloned T cells (29). Because endogenous expression of ILT2 by T cells would preclude any investigation on transferred ILT2, we used ILT2-negative resting CD4⁺ and CD8⁺ T cells and autologous monocytes purified from resting PBMCs of healthy donors as the only ILT2-expressing cells. The absence of ILT2 cell-surface and intracellular expression was systematically verified for the T cells we used, prior to each experiment, before activation and again after activation, right before use (Fig. 1A, Supplemental Fig. 1).

Trogocytosis differs from other mechanisms of intercellular Ag transfers, because it concerns entire portions of a donor cell’s plasma membrane and not individual molecules. Thus, trogocytosis has the potential of being a transfer of functional units from one cell to another. Thus far, we and others have demonstrated that acquired molecules can carry on their outward-oriented function (i.e., can act as a ligand and signal from the acquirer cell to another cell), but the possibility of a transferred receptor signaling inward has not been investigated in humans. In theory, this has always been a possibility on the condition that the originally functional membrane patch integrates properly within the plasma membrane of the acceptor cell and finds within its new cell a biochemical environment that it can reach and that is compatible with its functional requirements (e.g., signaling pathways). The data we presented in this study demonstrate that this is not only a possibility, but also a requirement for the well-described in vitro function of the inhibitory receptor ILT2.

The key findings of this article, as well as the trogocytosis-based mode of action of ILT2, are summarized in Fig. 7, which is only a schematic representation based on the data presented in Fig. 6. As depicted, during the first phase of the immune response, resting T cells do not express ILT2 and cannot acquire it from surrounding autologous monocytes for lack of Ag-independent trogocytic capability. It is possible that trogocytosis occurs during this phase, as reported in other systems. However, trogocytosis by resting cells is dependent on Ag recognition (5) and thus concerns molecules expressed by the presenting APC, not molecules expressed by bystander environmental cells, which do not present the selecting Ag. Immune regulation through molecules transferred from the stimulating APC may also happen at this time point but cannot be detected in our experimental system. At one point (day 4 in our case of allostimulation), activated Ag-specific T cells become capable of performing Ag nonspecific trogocytosis and acquire membranes from surrounding cells, including ILT2-containing membranes from autologous monocytes. The set of molecules involved in this type of acquisition is still unknown but clearly different from those used in Ag-specific trogocytosis by resting cells (9). Thus, through trogocytosis, activated T cells acquire multiple sensitivities from their microenvironment, which include sensitivity to inhibition by ILT2 ligands, as shown by our data. In the second phase of the response, T cells have become sensitive to a new array of immunomodulatory molecules. If these immunomodulatory molecules are present within the local environment, they may now signal to their transferred receptors and alter T cell behavior. In the case presented in this study, T cells became sensitive to ILT2 ligands, and because HLA-G was present within the local environment, these T cells stopped acting as effector cells. Thus, according to this model, the requirements for inhibition of T cells through ILT2 are: 1) Ag-selection followed by activation; 2) presence of ILT2-expressing monocytes; and 3) presence of an ILT2 ligand, HLA-G. To the best of our knowledge, it is the first time that transfers of sensitivity to immune modulatory molecules are shown and that this mechanism is proven crucial to an already acknowledged immune-regulation mechanism.

On a purely mechanistic standpoint, the data presented in this study show that membrane patches and molecules acquired by trogocytosis do not remain foreign patches, but integrate and, if possible, behave as would endogenously produced molecules. In our case, exogenously produced ILT2 was able to send an inhibitory signal strong enough to block the functions of the acceptor T cells, as would have endogenously produced ILT2. In our experimental configuration, donor and acceptor cells are both known to be able to signal through ILT2. It is therefore reasonable to postulate that the acquired patch integrated properly within the T cell plasma membrane and could use the compatible biochemical machinery of the acceptor cell to function. However, for other configurations in which donor and acquirer cells are biochemically very different, some signaling pathways might not be enabled, possibly leading to a partial or a lack of inward-oriented function of the acquired patch, even after integration of the transferred membrane. This limitation might be considered as a regulation mechanism for sensitivity transfers, allowing some but not all acquirer cells to acquire some but maybe not all sensitivities of the donor cell. Transfers of sensitivities would then be regulated at the cell-type level (trogocytic cell versus non trogocytic cell), molecular level (transferred versus not transferred molecules, integrated versus nonintegrated membrane patch), microenvironmental level (presence versus absence of a triggering ligand), and biochemical level (enabled versus disabled function of the transferred molecules). The possibility of inward signaling by acquired molecules through compatible biochemical pathways also raises two important but yet unresolved points: 1) if a signal is sent through a given membrane patch, will the consequences of this signal be different for the donor and the acceptor cells?; and 2) when considering only one signaling pathway and its consequences for a given cell, is it possible to broaden the range of triggering events for this signaling pathway through acquisition of new sensibilities?

From an immunological standpoint, our data show that trogocytosis is involved in a mechanism of in vitro immune regulation that is well known and may be important should it be confirmed in vivo. This is a key point, because if an immune regulation mechanism has been studied, agreed upon, and validated, so that it is regarded as a reliable fact, and if it is now proven that trogocytosis is required for this immune regulation mechanism, it means that trogocytosis might very well be involved in other immune mechanisms and thus may constitute a normal step and/or a general mechanism of immunity. This is first a problem: indeed, 1) if trogocytosis is a general mechanism of immune responses, the direct link among gene expression, phenotype, and function might be too much of a shortcut, and origin and traceability of molecules might become parameters (difficult ones) to seriously consider. Furthermore, 2) if trogocytosis is involved in or responsible for immunological phenomena currently studied or relied upon (such as ILT2-mediated inhibition of T cells), it means once more that the simplification of experimental systems by removal of bystander cells to obtain clean models might sometimes be just the wrong thing to do.

Yet, if intercellular transfers of cell-surface proteins, sensitivities, and functions constitute a generic mechanism in immunology, it means that there exists a level of powerful immune regulation that we hope is now clearly evidenced that is still vastly ignored and completely unexploited. This level of immune regulation would constitute one of the last ones in a sequence of many. Its purpose could be the adaptation of one given immune response to various local contexts. This would be a very flexible and cheap way for the immune system to check the fitness of a generic (centrally generated) response against particular (local) environmental constrains. Because trogocytosis-mediated regulation might dramatically alter the outcome of immune responses (9, 28, and current study) it should be exploitable for the benefit of the patients. Acting on trogocytosis might include enabling it, stopping it, or using it by acting on microenvironmental cellular parameters to endow immune cells with new desirable sensitivities and functions that would be location and time restricted.

We thank Niclas Setterbald and the Confocal Imagery Department (Institut de Formation et de Recherche 105) of the Institut Universitaire d’Hematologie, Hopital Saint-Louis, Paris, France, for technical help and Agnès Lemas, Maria Loustau, and Emilie Lesport from Commissariat a l’Energie Atomique-Service de Recherches en Hemato-Immunologie for technical contributions.

The authors have no financial conflicts of interest.