Antibody-Dependent Cellular Cytotoxicity RIIIa and Mediate g Express Fc T Lymphocytes Can b a Effector Memory

Human memory T cells are comprised of distinct populations with different homing potential and effector functions: central memory T cells that mount recall responses to Ags in secondary lymphoid organs, and effector memory T cells that confer immediate protection in peripheral tissues. In the present study we demonstrate that a proportion of effector memory T cells express Fc (cid:3) RIIIa (CD16), are perforin positive, and directly mediate Ab-dependent cytotoxicity ex vivo. This particular (cid:1)(cid:2) T lymphocyte subset has the morphology of large granular lymphocytes, increases proportionately in vivo during reactive lympho-cytosis, and can be detected in vitro among EBV-speciﬁc T lymphocytes after stimulation with EBV Ags. Consequently, during a normal immune response, ampliﬁcation of these effector memory T lymphocytes that are capable of Ab-dependent cytotoxicity may have beneﬁcial or harmful consequences depending on the presence of pathogen- or tissue-speciﬁc Abs, respectively. The Journal of Immunology, 2008, 180: 5327–5334.

T he vast majority of cells that express the IgG FcR responsible for Ab-dependent cellular cytotoxicity (ADCC 3 ; Fc␥RIIIa or CD16) belong to the innate immune system (1). These include monocytes/macrophages and NK cells. Among PBMC, it is assumed that essentially all ADCC activity is mediated by NK cells. Nevertheless, the existence of T lymphocytes bearing FcRs for Ig has long been recognized, although data concerning ADCC mediated by T lymphocytes are rare.
Indeed, in the past, initial attempts to discriminate subsets among T lymphocytes relied on the use of either antisera or differential FcR binding. Using the latter approach, two human T cell subsets could be separated: those capable of binding the Fc portion of IgM (T) and those capable of binding the Fc portion of IgG (T␥) (2)(3)(4)(5)(6). However, at that time, discrimination between different lymphocyte subsets was limited because the separation technique relied on density-gradient centrifugation of the subpopulations of lymphocytes that form rosettes with erythrocytes coated with rabbit IgG or IgM. Using this technique, the so-called "T␥" and "T" populations were found to display different morphological, histochemical, and functional characteristics (7,8). In particular, T␥ cells have been shown to mediate ADCC (9). Later on, when mAb targeting lymphoid populations became available, it became apparent that the T population was similar to the unfrac-tionated T cell population with respect to reactivity with anti-CD3, -CD4, and -CD5 mAb (OKT3, OKT4, and OKT5, respectively) while the T␥-enriched population contained, in fact, only 10 -20% of T cells as defined by OKT3 (10 -12). The authors of these studies concluded appropriately that the T␥ cells included a small population reactive with OKT3 and a large population reactive with OKM1, most likely corresponding to NK cells (10). Based on these findings, it was unclear whether ADCC attributed to the T␥ subset was mediated by the CD3-(non-T) NK fraction or the small proportion of CD3 ϩ T cells that express FcR for IgG. Later on, as stated above, essentially all NK activity present among PBMC was attributed to the NK population. In the early 1980s, several mAb were produced (anti-Leu-11, VEP 13, B73.1, and 3G8) that specifically reacted with the IgG FcR responsible for ADCC (Fc␥RIIIa, CD16) (13)(14)(15)(16) and the presence in most individuals of cells with the CD3 ϩ CD16 ϩ phenotype was confirmed (14,17,18). These cells usually comprise Ͻ2% of total PBL, with rare exceptions (18). In addition, it was specified that human T lymphocytes expressing the ␥␦ TCR can express CD16 and mediate ADCC (19 -21). In contrast, the link between the expression of CD16 and ␣␤ TCR on T lymphocytes has remained unclear. Although cells from large granular lymphocyte leukemia have been shown to coexpress CD16 and an ␣␤ TCR (22,23), the existence of "normal" T cells harboring this phenotype has been suggested in only rare instances (24 -27) and these cells have not been characterized. The present study was initiated to directly assess the proportion, the phenotype, and the functional capabilities of CD16 ϩ ␣␤ TCR T cells. We demonstrate that CD16 ϩ ␣␤ TCR T cells, which are present in small numbers in all individuals 1) belong to the memory effector subset, 2) are perforin positive, 3) are capable of mediating ADCC immediately ex vivo, 4) increase proportionately in vivo during lymphocytosis, and 5) appear in vitro in all cultures of EBV-specific CTL. Consequently, because of their ADCC potential and their presence among virus-specific T cells during a specific T cell response, amplification of these CD16 ϩ ␣␤ TCR effector memory T lymphocytes during a regular immune response may be involved in beneficial or harmful consequences depending on the presence of pathogen-or tissue-specific Abs.

Samples and cell lines
PBMCs were prepared by Ficoll (PAA Laboratories) gradient centrifugation of blood obtained from adult volunteers or patients with hyperlymphocytosis that were recruited nonselectively from the Department of Infectious Diseases (Centre Hospitalier de l'Université de Nantes, Nantes, France). All individuals gave informed consent. EBV B lymphoblastoid cell lines (BLCL) were derived from PBMCs by in vitro infection using EBV-containing culture supernatant from the Marmoset B95.8 cell line purchased from the American Type Culture Collection in the presence of 1 g/ml cyclosporine A. The K562 cell line was cultured in complete medium consisting of RPMI 1640 (Sigma-Aldrich), 10% heat-inactivated FCS, 2 mM glutamine (Sigma-Aldrich), 100 U/ml penicillin, and 10 g/ml streptomycin (Sigma-Aldrich).

Cell cultures
Sorted CD16-expressing ␣␤ T lymphocytes were cloned using a nonspecific amplification procedure: T cells were seeded at 3.0, 1.0, and 0.3 cells/well in 96-well U-bottom plates together with irradiated (35 Gy) pooled allogeneic feeder cells (1 ϫ 10 5 PBMC and 1 ϫ 10 4 B cells from a BLCL), 1 g/ml leukoagglutinin PHA-L (Sigma-Aldrich), and 300 IU/ml rIL-2 (Roussel-Uclaff) in a final volume of 200 l. After cloning, T cells were further expanded: first, in 24-well culture plates, then in culture flasks using the same culture conditions (feeder cell concentration, rIL-2, and PHA concentration) that allow for maximal proliferation of the T cell clones for several weeks.

Generation and expansion of EBV-specific cytotoxic cell lines
Donor PBMCs were plated in 24-well culture plates at 2 ϫ 10 6 cells/well in RPMI 1640 with glutamax (Invitrogen Life Technologies) culture medium supplemented with 8% pooled human serum, and stimulated with 5 ϫ 10 4 40 Gray-irradiated autologous BLCL (PBMC:BLCL ratio of 40:1). After 10 days, T cells were collected and restimulated at a T:B ratio of 4:1 (5 ϫ 10 5 T cells and 1.25 ϫ 10 5 BLCL/well). IL-2 was added 4 days after the second stimulation (40 IU/ml). A third and a fourth stimulation were performed every 7 days in the presence of IL-2 and at the same T:B cell ratio (4:1).

Cytotoxicity assay
Cytotoxic activity was assessed using a standard 51 Cr-release assay. Target cells were labeled with 100 Ci 51 Cr for 1 h at 37°C, washed four times with culture medium, and then plated at the indicated E:T cell ratio in a 96-well flat or U-bottom plate. An autologous BLCL was used as a model of an autologous tumor and the chimeric anti-CD20 mAb rituximab (Roche) was used (at 2 g/ml) to induce ADCC. In some experiments, the anti-Her2/neu mAb trastuzumab (Roche) was used (at 10 g/ml) as a con-trol. For ADCC assays, the indicated mAb was incubated with target cells for 20 min before addition of effector cells. After a 4-h incubation at 37°C, 25 l of supernatant were removed from each well, mixed with 100 l of scintillation fluid, and 51 Cr activity was counted in a scintillation counter. Each test was performed in triplicate. The results are expressed as the percentage of lysis, which is calculated according to the following equation: (experimental release Ϫ spontaneous release)/(maximal release Ϫ spontaneous release) ϫ 100, where experimental release represents the mean cpm for the target cells in the presence of effector cells, spontaneous release represents the mean cpm for target cells incubated without effector cells, and maximal release represents the mean cpm for target cells incubated with 1% Triton X-100. For the assessment of ex vivo ADCC by CD16 ϩ ␣␤ T lymphocytes, cells positively sorted by the 3G8-PC5 Ab were washed and incubated overnight in complete medium (without any cytokines) before the functional ADCC assay.

Statistical analysis
Differences between subjects with hyperlymphocytosis and healthy controls were analyzed using the t test. A p value of Ͻ0.05 was considered significant.
The CD16 ϩ ␣␤ T lymphocyte subset was further characterized for CD4 or CD8 expression and with a panel of mAb that discriminates between different memory T lymphocyte subsets (CD27, CD28, CD45RO, CD45RA, CD57, CD62L, and CCR7). Staining for CD32a/CD32b (anti-Fc␥RIIa and -Fc␥RIIb), CD64 (anti-Fc␥RI), and three killer Ig-like receptors (CD158a,h, CD158b, and CD158e) was also performed. Data from 15 donors are reported in Table I. The vast majority of CD16 ϩ ␣␤ T lymphocytes (that were all negative for CD32 and CD64, data not shown) were CD8 ϩ (89%) and mainly CD45RA ϩ CCR7 Ϫ , with only 42% of cells expressing CD62L, suggesting that they belonged to the effector memory T lymphocyte population. In addition, 74% of these cells expressed CD57 and only 12.9% expressed CD28. Because a lack of CD28 and presence of CD57 on CD8 ϩ T cells are generally associated with a status of replicative senescence, altogether these results suggest that CD8 ϩ CD16 ϩ ␣␤ T lymphocytes belong to a small population that has been previously described as TEMRA lymphocytes for terminally T effector memory CD45RA ϩ lymphocytes (28,29).

The CD16 ϩ ␣␤ T lymphocyte subset is capable of mediating ADCC ex vivo
The ADCC activity of CD16 ϩ ␣␤ T lymphocytes from three healthy donors was tested and compared with that of autologous NK cells. CD16 ϩ ␣␤ T lymphocytes, CD16 ϩ NK cells, and CD16-negative ␣␤ T lymphocytes from the same donor were sorted by FACS and their cytotoxic activity assessed using a 4-h 51 Cr-release assay against the NK-sensitive K562 cell line and the autologous BLCL in the presence of absence of anti-CD20 or anti-Her2/neu humanized mAb (BLCL were all positive for CD20 and negative for Her2/neu Ags). Fig. 2 shows representative data obtained from one healthy donor. Only NK cells were able to kill the K562 cell line. In the absence of mAb, NK cells, CD16 ϩ ␣␤, and CD16-negative ␣␤ T lymphocytes did not recognize the autologous BLCL. In contrast, both CD16 ϩ NK cells and CD16 ϩ ␣␤ T lymphocytes killed the BLCL incubated with anti-CD20 mAb. This cytotoxicity was not observed in the presence of anti-Her2/neu mAb. Together, these results demonstrate that despite expressing low levels of CD16, CD16 ϩ ␣␤ T lymphocytes all express perforin (one example of six donors tested is presented in Fig. 2B) and are able to mediate ADCC ex vivo.
Limiting dilution analysis of the CD16 ϩ ␣␤ T lymphocytes CD16 ϩ ␣␤ T lymphocytes from two healthy donors were sorted and cloned by limiting dilution using nonspecific stimulation (lectin plus feeder plus IL-2). Cloning efficiency was 0.75 and 0.30 according to the Poisson distribution (Fig. 3A). Nineteen days after cloning, 47 of 52 clones from donor 1 and 18 of 19 clones from donor 2 were CD16 positive. Next, 4 clones from donor 1 and 6 from donor 2 were selected according to their CD16 expression levels and tested for changes in CD16 expression and ADCC activity over a 3-mo culture period. One CD16-negative ␣␤ T cell clone from each donor was included in these experiments as a negative control for ADCC assays as well as a control for assessment of CD16 expression during culture. After stimulation (lectin plus feeder plus IL-2) CD16 on CD16-positive clones was downregulated for 10 -15 days but was eventually re-expressed and maintained for several weeks. Fig. 3B shows one example of CD16 expression and ADCC activity monitoring for clone 2 from donor 1. CD16 expression is shown at day 28 after cloning (Fig. 3Ba), at days 27 (Fig. 3Bb) and 52 (Fig. 3Bc) after the first restimulation and then after freezing and thawing, 38 days after stimulation. Under the same conditions, the control CD16-negative ␣␤ T cell clone never expressed CD16. These results indicate that T cells programmed to express CD16 maintained CD16 after TCR stimulation at a level depending on the state of activation. ADCC activity of CD16 ϩ ␣␤ T cell clones was also assessed using a 4-h 51 Cr-release assay performed against the autologous target BLCL. Fig. 3B shows the results obtained at different times points with the CD8 ϩ CD16 ϩ ␣␤ T cell clone 2 that was derived by limiting dilution from the CD16 ϩ ␣␤ T cell fraction of a healthy donor and whose TCR specificity was unknown. In the absence of mAb,  clone 2 did not recognize the autologous BLCL. In contrast, addition of the chimeric anti-CD20, but not the anti-Her2/neu mAb, induced lysis of the autologous BLCL, thus demonstrating that clone 2 was able to mediate ADCC (the same results were observed with the four clones tested). One can note that despite the suggestion that anti-CD20 might be directly cytotoxic against target cells, this was never observed in our experiments using BLCL, even at high anti-CD20 concentrations, in the presence or absence of CD16-negative CTL (data not shown).

Thymidine incorporation and TNF-␣ production by CD16 ϩ ␣␤ T lymphocyte clones
To test whether T cell responses other than cytotoxic activity could be initiated in CD16 ϩ ␣␤ T lymphocytes, several T cell clones were tested for their ability to incorporate thymidine and to produce TNF-␣ after exposure to Ab-coated cells. To exclude the possibility that the clones were activated by soluble Ab, mAb concentrations of up to 1000 g/ml were tested in the absence of target cells. Examples of the results are presented in Fig. 3, C and D. In Fig. 3C, the CD16 ϩ ␣␤ T cell clones 3 and 21 and the CD16-negative ␣␤ T cell clone 16 obtained by limiting dilution of FACS-sorted CD16 ϩ ␣␤ T cells from donor 2 were tested in a 72-h proliferation assay in the presence of IL-2 (40 IU/ml). The assays were performed in the presence of either soluble anti-CD20 mAb (1-1000 g/ml) or against the autologous BLCL in the absence or presence of rituximab (0.02 and 2 g/ml). As shown, thymidine incorporation was observed in the presence of rituximab and the target cell only. In the presence of rituximab only, no thymidine incorporation was detected at any concentration tested, demonstrating that cross-linking was required to induce proliferation. The same conclusion could be drawn for cytokine production, as presented in Fig. 3D. Essentially all cells from CD16 ϩ / CD8 ϩ T cell clones 14 (Fig. 3D, D1) and 21 (Fig. 3D, D2) were able to produce TNF-␣ when stimulated with PMA and calcium ionophore. Following cross-linking to target BLCL, 14.4 and 14.6% of cells from clones 14 and 21 became positive for TNF-␣, respectively. In contrast, in the absence of target cells, the soluble mAb was unable to induce significant TNF-␣ production by the clones, even at concentrations of up to 1000 g/ml.

CD16 ϩ ␣␤ T lymphocytes appear in vivo during hyperlymphocytosis
Because CD16 was found to be expressed by memory ␣␤ T lymphocytes in normal, healthy donors, we anticipated that this particular subset would appear and be amplified during the T cell response. To test whether this is indeed the case, we analyzed CD16 expression by ␣␤ T cells in blood samples from patients with hyperlymphocytosis that were retrieved nonselectively from the Department of Infectious Diseases (above 4000 lymphocytes/ l). Twenty-six healthy controls and 15 subjects with hyperlymphocytosis were tested. The results shown in Fig. 4   autologous EBV B lymphoblastoid cell line allowed for the amplification of the EBV-specific memory T cell repertoire. In this situation, it becomes possible to analyze Ag expression directly on virus-specific T cells and at a documented time point after Ag exposure. We analyzed EBV-specific cytotoxic T cell lines from six healthy donors. The results shown in Fig. 5 demonstrate that CD16 ϩ ␣␤ T lymphocytes were detected in all EBV-specific cytotoxic T cell lines. CD16 expression became detectable from days 15 to 20 after restimulation with the autologous BLCL. The frequency of CD16 ϩ EBV-CTL ranged from 1 to 27% (Fig. 5A). Finally, the ADCC activity of these EBV-CTL containing significant numbers of CD16 ϩ T cells was revealed by their cytotoxic activity against an allogeneic BLCL in the presence of an anti-CD20 mAb (Fig. 5B).

Discussion
In this study, we demonstrated that T lymphocytes coexpressing the ␣␤ TCR and Fc␥RIIIa (CD16) are present in significant numbers in all individuals tested. The mean numbers of these cells were similar to those of the previously described "innate-like" CD16 ϩ ␥␦ TCR T cells. Together with the latter cells, these CD16 ϩ ␣␤ TCR T cells represent a pool of 70 lymphocytes/ml of blood, while ϳ194 CD3 Ϫ CD16 ϩ NK cells/ml were detected in the same population. Thus, these two T cell subsets make up onefourth of all peripheral lymphocytes expressing CD16 and capable of mediating ADCC. Phenotypic analysis revealed that the vast majority of CD16 ϩ ␣␤ TCR T cells do not express CCR7 but do express CD45RA and belong to the T effector memory CD45RA ϩ or TEMRA lymphocyte subset (28,29). We additionally showed here that CD16 ϩ ␣␤ TCR T cells are perforin positive, directly mediate ADCC ex vivo, and increase in vivo during hyperlymphocytosis. Finally, we showed that CD16 ϩ ␣␤ TCR T cells appeared in vitro in all cultures of EBV-specific CTL.
Because the main effector cell type involved in ADCC is of a non-T and non-B phenotype the finding that a significant proportion of Ag-specific T lymphocytes from the adaptative immune system can have two specificities, one through their TCR and the other through FcR recognition, may have as yet unrecognized physiopathological consequences. For example, given that gp350 has been shown to be a target Ag for EBV-specific ADCC during EBV infection (30), our results suggest that the same T lymphocytes able to recognize an EBV-infected target cell after TCR mediated recognition of EBV-peptide/HLA-complex could potentially recognize the same EBV-infected target cell after CD16 mediated ADCC in the presence of gp350 Abs. In the case of HIV-1-seropositive individuals, the gp120-specific cell-mediated cytotoxicity was attributed to the presence of circulating CD16 ϩ effector cells armed in vivo with cytophilic Abs. In this latter work, the cells responsible for this activity were identified as NK cells. Nevertheless, it remains possible that under certain circumstances, such as EBV or CMV reactivation in HIV-infected patients, the effector memory cells that become CD16 ϩ engage this mechanism against HIV-infected target cells (31). In this case, the mechanism would be of benefit to the host. In contrast, CD16 has also been involved as a route of virus entry into mononuclear cells and if this mechanism is also efficient for T cells then in the presence of anti-HIV Abs, the CD16 ϩ fraction of T lymphocytes reacting against other pathogens may also become infected (32).
If Abs are directed against healthy tissues, in some circumstances the CD16 ϩ fraction of a pathogen-reactive T cell population might be involved in tissue damage. Because the ADCC process requires three components (target cells expressing Ags on their surface, Abs of the IgG isotype directed against the target Ag and effector cell bearing Fc␥R), its occurrence in vivo will depend on the local concentration of Ags, Abs, and CD16 ϩ effector cells. Accordingly, large variations in the concentration of each of these components may have an impact on the risk of autoimmune manifestation. Because autoantibodies, autoantigens, and CD16 ϩ effector cells capable of ADCC are present in all healthy individuals, only future longitudinal exploration of the quantitative variation in specific CD16 ϩ ␣␤ TCR T cells will enable an exploration into the potential participation of these cells in autoimmune processes. Nevertheless, an already well-known clinical condition exists where an abnormal proliferation of ␣␤ TCR T lymphocytes expressing CD16 is observed. This condition is referred to as LGL leukemia (22,23) and one of the unique features of this malignancy is its frequent association with autoimmune disease.  . EBV-specific cytotoxic T cell lines (CTL) contain CD16 ϩ ␣␤ T cells that mediate ADCC. A, EBV-specific CTLs were selected against the autologous BLCL and stained with PE-anti-␣␤ Ab and PC5anti-CD16 Ab. B, The ADCC activity of two EBV-specific CTLs was evaluated against 51 Cr-labeled allogeneic BLCL (not recognized by the TCR of the CTL) in the presence of either rituximab (anti-CD20, 2 g/ml) or herceptin (anti-HER-2, 10 g/ml) as a negative control. Results are expressed as percentage of specific lysis (E:T ratio ϭ 30:1, mean of triplicate wells).
Although the etiology of LGL leukemia is unknown, chronic activation of T cells with auto or viral Ag has been suggested as an initial stimulus leading to an expansion of LGL (33). Moreover, Baesso et al. (34) recently demonstrated that LGL cells display a phenotypic pattern typical of effector cytotoxic cells and the central hypothesis put forward by Loughran, Jr. and colleagues (33) is that leukemic LGL are Ag-driven CTL. Indeed, the usual immunophenotype of the malignant T cells in LGL leukemia resembles that of typical effector memory T cells (CD3 ϩ CD4 Ϫ CD8 ϩ CD16 ϩ CD27 Ϫ CD45RO Ϫ CD57 ϩ ) and in the vast majority of cases, the TCR is of the ␣␤ subtype and the cells are capable of ADCC (35)(36)(37). Although the exact mechanism involved in LGL-associated autoimmune disease is not completely understood, the efficacy of immunosuppression or depletion in improving this condition strongly suggests a direct implication of the leukemic LGL in the clinical symptoms (33). Data presented in this study fit well with the hypothesis of Loughran, Jr. and colleagues (33) because the characteristics of the T cell subset we describe correspond exactly to those of LGL leukemia cells. We reasoned that if the chronic amplification of LGL leukemia cells is directly involved in autoimmune diseases, then the transient amplification of their normal counterparts may also be directly responsible for transient autoimmunity. This phenomenon would rely on an increase in the ADCC potential of a patient through amplification of this particular memory subset above the threshold that determines the appearance of autoimmune symptoms. Such an increase may be either transitory as is the case during a normal immune response, or abnormal, as suggested in many examples of autoimmune disease where activation-induced cell death has been found to be disturbed (38 -42).
Along these lines, in the presence of tissue-specific Abs, the possible influence of CD16 ϩ ␣␤ TCR T cells may theoretically provide a simple link between microbial infections and autoimmune exacerbation. This could explain how a single autoimmune disease can be exacerbated by many different infections, or alternatively, how different infections can be associated with exacerbation of the same autoimmune disease (see the recent review by Fujinami et al. (43)).
Beyond the analysis of CD16 ϩ ␣␤ TCR T lymphocyte participation in different physiopathological processes, manipulation of a patient's "ADCC potential" through amplification of this specific memory subset could be worth considering in the context of Abbased immunotherapy. Although CD16 expression by CD16 ϩ ␣␤ TCR T lymphocytes appeared lower than that observed in NK lymphocytes, this T lymphocyte subset does demonstrate ADCC that can be amplified in vitro, as shown in the present work. Because amplification, freezing, and thawing is far easier with T than with NK lymphocytes, CD16 ϩ ␣␤ TCR T lymphocytes are good candidates for ex vivo amplification before reinjection into a patient with a view to improving their ADCC potential. One can also note that if EBV-CTL can express CD16 some time after Ag exposure, then patients treated with EBV-CTL for post-transplant lymphoproliferative disease after a first line of treatment involving anti-CD20 injection may also benefit from the ADCC activity of the EBV-CTL they receive.
In conclusion, the present demonstration that Ag-experienced T cells can express CD16 and mediate ADCC will likely stimulate many investigations in different fields of immunology 1) to precisely assess the potential role of these populations during a normal immune response; 2) to directly confirm their relationship with LGL leukemia cells; 3) to test their precise implication in different autoimmune diseases where a role for autoantibodies has already been established; and 4) to consider their potential in the context of immunotherapy.