In Vitro Generation of Allospecific Human CD8⁺ T Cells of Tc1 and Tc2 Phenotype

IT HAS RECENTLY been recognized that murine CD4⁺¹ and CD8^+2,3 T cells can be subdivided based on specific patterns of cytokine secretion. Type I T cells secrete primarily interleukin-2 (IL-2) and interferon gamma (IFN-γ), whereas type II cells secrete IL-4, IL-5, and IL-10. Importantly, in murine systems, both type I and type II cytokine-secreting subsets of CD8⁺ T cells demonstrate cytolytic function (Tc1/Tc2 subsets).2 Although the in vivo significance and cross-regulatory nature of CD4⁺ Th1 and Th2 subsets has been established in murine models,4 the in vivo role of CD8⁺ Tc1/Tc2 subsets is only now being investigated. We have recently demonstrated that allospecific murine CD8⁺ T cells of Tc1 and Tc2 phenotype mediate graft-versus-leukemia (GVL) effects with reduced graft-versus-host disease (GVHD).5 The allospecific Tc2 subset, which lysed allogeneic targets and secreted type II cytokines, mediated GVL without inducing ongoing GVHD; others have also demonstrated that allospecific murine CD8⁺ T cells secreting type II cytokines fail to induce acute GVHD.6 In comparison, the allospecific Tc1 population, which lysed allogeneic targets and secreted type I cytokines, mediated a more potent GVL effect and increased GVHD. These observations may relate both to the contribution of cytolytic pathways to the GVL effect7 and to the role of type I cytokines in the pathogenesis of GVHD.8 Because the Tc1 population mediated much less GVHD than noncultured donor CD8 cells, our study indicated that both Tc1 and Tc2 populations may improve the therapeutic index of allogeneic T-cell administration for the treatment of leukemia. In light of these findings, we investigated whether analogous allospecific CD8⁺ Tc1 and Tc2 populations can be generated in humans.

The evaluation of T-cell clones generated from patients in various disease states has demonstrated that CD4⁺ and CD8⁺ T cells of the type I and type II cytokine profile exist in humans (reviewed in Romagnani9 ). Importantly, the relative balance of type I and type II cytokine responses appears to influence the natural history and clinical course of human infectious diseases, including leprosy,10 tuberculosis,11 leishmania,12 and HIV infection.13 In general, type I CD4⁺ and CD8⁺ T-cell responses are associated with resistance to infection, whereas type II responses are associated with disease progression. These studies indicate that in vivo immune responses are amenable to type I/type II regulation in both murine models and humans.

Although these studies demonstrate that cytokine-secreting subsets of human CD8⁺ T cells exist in human disease states, there is limited information regarding the ability or undifferentiated human CD8 cells to develop into Tc1 of Tc2 subsets. In vitro culture of human neonatal CD4⁺¹⁴ and CD8⁺¹⁵ T cells after polyclonal stimulation with anti-CD3 antibody has demonstrated that human T cells can differentiate toward type I and type II patterns of cytokine secretion. However, it has not been shown that primary in vitro culture can generate antigen-specific cytolytic human CD8⁺ cells of disparate type I and type II cytokine production (Tc1 and Tc2 subsets).

In this report, we demonstrate that allospecific human CD8⁺ T cells of Tc1 and Tc2 phenotype can be generated by primary in vitro culture. This ability to skew antigen-specific human CD8⁺ cytolytic responses toward polarized type I and type II cytokine profiles indicates that the potential for human Tc1 and Tc2 subset differentiation exists. In addition, in light of results from our murine studies, these findings may have implications for attempts to regulate allogeneic T-cell responses in the setting of bone marrow transplantation.

Flow cytometric analysis.Cells for flow cytometric analysis were washed, suspended in Hanks' balanced salt solution (GIBCO-BRL, Grand Island, NY) containing 1% bovine serum albumin (Sigma Chemical, St Louis, MO) and 0.1% sodium azide (Sigma), blocked with 10 μL human IgG (derived from human AB serum; Sigma), and stained with the following antibodies (or isotype controls): B7.1 (CD80)-FITC or B7.2 (CD86)-FITC (mouse IgM-FITC) (PharMingen Corp, San Diego, CA); HLA-DR-PE (mouse IgG-2b-PE), CD4-PE, CD8-FITC, and CD3-TC (mouse IgG-2a-PE/FITC/TC) (Caltag Laboratories, San Francisco, CA). Flow cytometry was performed on a FACSort (Becton Dickinson Immunocytometry Systems, Mountain View, CA) using Lysis II software. Five thousand to 10,000 live events were acquired per experimental sample; dead cells were gated out by propidium iodide staining.

Purification of human CD8⁺ T cells.Peripheral blood lymphocytes (PBLs) were obtained from volunteer normal donors after provision of appropriate informed consent as part of a protocol approved by our Institutional Review Board. PBLs were obtained by a single 10-L leukopheresis (Fenwal CS-3000; Fenwal Corp, Los Angeles, CA) and were further purified by countercurrent centrifugal elutriation (Beckman Ultracentrifuge XL-100 with elutriation rotor; Becton Dickinson) as previously described.16 This methodology produced a population that was greater than 95% lymphocytes. The lymphocytes were cryopreserved in 7.5% dimethylsulfoxide (Sigma) and 40% fetal calf serum (Hyclone, Logan, UT) and stored under liquid nitrogen until the time of use. Thawed lymphocytes were enriched for CD8⁺ cells by a negative-selection column (R&D Systems, Minneapolis, MN) that consistently produced greater than 95% CD8⁺ purity as assessed by flow cytometry. If significant numbers of CD4⁺ T cells existed in culture (>1%), they were removed before enzyme-linked immunosorbent assay (ELISA) evaluation by immunomagnetic bead positive selection (PerSeptive Diagnostics, Cambridge, MA).

Generation of bone marrow–derived stimulator cells.Bone marrow was obtained from volunteer normal donors after provision of appropriate informed consent. A unilateral iliac crest aspiration was performed removing 2 to 3 mL marrow into preservative-free heparin (Fujisawa, New York, NY). Low-density bone marrow cells were enriched by centrifugation with lymphocyte separation media (LSM; Organon Teknika Corp, Durham, NC); 2.5 to 7.5 × 10⁷ low-density marrow cells were placed into complete media consisting of RPMI 1640 (GIBCO-BRL) that contained 10% fetal calf serum, 1% sodium pyruvate, 1% nonessential amino acids, 0.5% L-glutamine, 2-ME (5 × 10⁻⁵ mol/L), 0.5% penicillin, and 0.5% streptomycin (NIH Media Unit, Bethesda, MD). Marrow culture media were supplemented with rhuIL-4 (1,000 U/mL; Peprotech, Rocky Hill, NJ), rhuGM-CSF (1,000 U/mL; Peprotech), rhuIL-7 (20 ng/mL Peprotech), and N-acetylcysteine ([NAC]-10 mmol/L pH adjusted to 7.2; Sigma Chemical). Cells were cultured at 37°C and 5% CO₂ in 75-cm² flasks (Costar, Cambridge, MA) at a cell density of 0.5 to 1.0 × 10⁶/mL. The media and flask were changed every 72 to 96 hours; the resultant bone marrow–derived stimulator cells were harvested on day 10 and cryopreserved until their use in coculture.

Bulk culture of CD8⁺ T cells and bone marrow–derived stimulator cells under Tc1 and Tc2 conditions.Purified human CD8⁺ T cells were cultured with bone marrow–derived stimulatory cells from an unrelated donor at a T cell to stimulator cell ratio of 2:1; stimulator cells were irradiated (600 cGy) prior to culture. For T-cell culture, NAC (10 mmol/L) was added to complete media (as already described). Tc1 cultures were additionally supplemented on day 0 with rhuIL-12 (2.5 ng/mL, R&D Systems) and recombinant human transforming growth factor-beta ([TFG-β] 5 ng/mL; R&D Systems), whereas Tc2 cultures were initiated in the presence of rhuIL-4 (1,000 U/mL; Peprotech). All cultures received rhuIL-2 (40 Cetus U/mL; Chiron Therapeutics, Emeryville, CA) and rhuIL-7 (20 ng/mL) on days 0 and 2; fresh media containing IL-2 (20 U/mL), IL-7 (20 ng/mL), and NAC (10 mmol/L) were then added every 3 to 4 days.

Cytokine secretion profiles by ELISA.On day 13 of culture, aliquots of cells from Tc1 and Tc2 cultures were harvested, brought to a final concentration of 0.5 × 10⁶ cells/mL, and stimulated in separate wells of a 24-well plate with either allogeneic or autologous bone marrow–derived stimulator cells or a combination of phorbol ester (PMA, 5 ng/mL; Sigma) and calcium ionophore (A23187, 375 ng/mL; Sigma). Culture supernatants were harvested after 24 hours and tested in two-site ELISAs using commercially available reagents (IL-2, IL-4, IL-5, IL-10, and IFN-γ paired antibodies; PharMingen). Cytokine levels were calculated by reference to standard curves constructed on supernatants containing known amounts of recombinant cytokine.

Evaluation of allospecificity of Tc1/Tc2 cytolytic function.On day 13 of culture, Tc1 and Tc2 effector cell populations were harvested and evaluated for the ability to mediate target lysis. Lytic function was determined by standard 4-hour ⁵¹Cr-release assays17 at known effector to target ratios, with calculation of percent specific lysis. To evaluate allospecific Tc1/Tc2 subset lytic function, the bone marrow–derived stimulator cell population used in the primary in vitro allosensitization was used as the target; autologous bone marrow–derived stimulator cell populations were used as control targets.

Heteroconjugate assay for determination of T-cell receptor–mediated cytolytic function.The mouse leukemia/lymphoma cell line L1210 (courtesy of Dr Pierre Henkart, NIH) was haptenized with trinitrobenzenesulfonic acid (TNBS) and incubated (30 minutes at 37°C) with a CD3-activating bispecific antibody specific for TNBS and the human T cell receptor (TCR) (T3X-DNP¹⁸; courtesy of Dr David Siegel, NIH). To evaluate the Tc1 and Tc2 populations (day 13 of culture) for lytic activity mediated through TCR activation, TNBS-haptenized or unmodified L1210 cells were incubated with the bispecific antibody (final concentration, 0.1 μg/mL) and used as targets in standard 4-hour ⁵¹Cr-release assays.

Evaluation of Fas-based Tc1/Tc2 cytolytic function.To evaluate the Tc1/Tc2 subsets for fas-mediated cytolytic function, the fas-transfected murine lymphoma line (L1210-fas)19 and the untransfected line (L1210) were used as targets in chromium-release assays. Tc1 and Tc2 subsets were analyzed on day 13 of culture without further stimulation, or after a 24-hour period of stimulation with either alloantigen (marrow-derived allostimulatory cells) or PMA (5 ng/mL) and calcium ionophore (375 ng/mL).

In vitro generation of allospecific human CD8⁺ T cells secreting type I and type II cytokines.To generate allospecific Tc1 and Tc2 populations, CD8⁺ T cells were cocultured with bone marrow–derived stimulator cells from unrelated volunteers. The marrow-derived stimulator population was 90% to 95% positive for the T-cell marker CD3, with an equal proportion of CD4⁺ and CD8⁺ cells; such T cells expressed HLA-DR and also the costimulatory molecules B7.1 and B7.2. Using this marrow-derived population as stimulators, we evaluated various culture variables for the ability to propagate CD8⁺ T cells under Tc1 and Tc2 conditions. We have previously shown that the antioxidant NAC facilitates in vitro generation of allospecific murine CD8⁺ T cells of Tc1 and Tc2 phenotype.5 In the human CD8 culture system detailed in this report, yields for Tc1 and Tc2 cultures were greater if NAC was added to the culture media; in some experiments, Tc1 and Tc2 expansion was only observed in conditions that included NAC (data not shown). In addition to this beneficial role of NAC, we found that addition of IL-7 (for both Tc1 and Tc2 cultures) and TGF-β (for Tc1 cultures) was required for optimal T-cell expansion.

After 13 days of culture, CD8⁺ T cells cultured under Tc1 or Tc2 conditions were characterized by flow cytometry for cell-surface phenotype and by ELISA for cytokine secretion. All cultures were greater than 95% CD8⁺ at the time of ELISA supernatant generation, with less than 1% contaminating CD4⁺ cells. In murine systems, downregulation of CD8 expression has been described in type II cytokine-secreting populations.20 In our experiments, there was no detectable difference in surface CD8 expression between allospecific human Tc1 and Tc2 populations.

Figure 1A and B shows a representative ELISA result illustrating the marked polarization of cytokine secretion observed in Tc1 and Tc2 cultures. Cytokine production for both Tc1 and Tc2 cultures required specific alloantigenic restimulation, as no cytokines were detected when autologous bone marrow–derived stimulator cells were used as the stimulator for supernatant generation. In response to allogeneic restimulation, CD8⁺ T cells generated in the presence of IL-12 and TGF-β (Tc1 culture) secreted significant levels of the type I cytokines IL-2 and IFN-γ, but did not produce detectable levels of the type II cytokines IL-4, IL-5, or IL-10. In contrast, allogeneic restimulation of CD8⁺ T cells generated in the presence of IL-4 (Tc2 culture) produced significant levels of IL-5 and IL-10 and either undetectable (IL-2) or markedly reduced (IFN-γ) levels of type I cytokines. Allogeneic restimulation of the Tc2 population did not result in measurable secretion of the type II cytokine IL-4 (Fig 1B); however, significant IL-4 production was observed in the Tc2 population after pharmacologic stimulation with PMA and calcium ionophore.

Both Tc1 and Tc2 populations demonstrate potent allospecific cytotoxic function.Allospecific cytolytic function of Tc1- and Tc2-type cells was tested on day 13 of culture. To evaluate allospecificity of the generated CD8⁺ populations, cells from each donor (A and B) were used as responder (CD8 cells), stimulator (marrow-derived cells), and target (marrow-derived cells) populations. Figure 2 shows cytotoxicity results for representative Tc1 and Tc2 populations generated from two normal donors. Tc1 and Tc2 populations from both donors were polarized for cytokine secretion (similar to that presented in Fig 1; data not shown). Both Tc1 and Tc2 populations showed potent allospecific cytolytic activity, as demonstrated by the lysis of allogeneic but not of autologous targets. The ability of the bone marrow–derived populations to effectively stimulate cytotoxic responses was confirmed by the experimental design. In all experiments performed (n = 6), Tc1 and Tc2 populations have been comparable in the ability to lyse allogeneic targets.

Both Tc1 and Tc2 populations mediate cytotoxicity via TCR activation.To directly measure cytolytic activity mediated through the TCR, Tc1 and Tc2 populations were evaluated for the ability to lyse the TNBS-modified murine L1210 tumor line after incubation with a bispecific anti-TNBS/anti-TCR antibody; unmodified L1210 cells were used as the control target. Figure 3 shows that both Tc1 and Tc2 populations demonstrated potent cytotoxicity against the TNBS-modified murine target. Because no cytolytic activity was seen for any of these populations against the unmodified L1210 control target, these results demonstrate that killing was mediated via TCR activation. The figure also demonstrates that the cytolytic function of both Tc1 and Tc2 populations was markedly increased when CD8⁺ cells were restimulated with alloantigen 72 hours before the assay (Fig 3).

Both Tc1 and Tc2 populations mediate cytotoxicity via the Fas pathway after pharmacologic activation.To evaluate Tc1 and Tc2 populations for the ability to mediate cytolysis via the fas pathway, the fas-transfected murine line L1210 was used in chromium-release assays. Figure 4 demonstrates that both Tc1- and Tc2-type CD8⁺ T cells lysed the fas-transfected L1210 line, but not the nontransfected L1210 control. Lysis of the fas-transfected cell line was only observed if Tc1 and Tc2 populations (day 13 of culture) were activated for 24 hours with PMA and calcium ionophore prior to use as effectors in the lytic assay; assays performed with Tc1 and Tc2 cells restimulated with alloantigen (or not restimulated) did not show lysis of the L1210-fas target (not shown). These results thus demonstrate that both allospecific Tc1 and Tc2 populations have the potential for functional fas ligand expression upon pharmacologic stimulation, and that the acquisition of fas-based cytolytic function required activation events not provided by allogeneic restimulation.

The existence of cytokine-secreting subsets of cytolytic CD8⁺ T cells has been demonstrated in multiple murine systems.2,3,21 Importantly, we have shown that allospecific murine Tc1 and Tc2 subsets have differential effects in the setting of allogeneic bone marrow transplantation, and that these functionally defined CD8 populations may offer therapeutic advantages compared with noncultured donor CD8⁺ T cells.5 In light of this murine evidence, we evaluated whether analogous Tc1 and Tc2 populations can be generated in humans. In this report, we have defined in vitro culture conditions that allow for the generation of allospecific human CD8⁺ cytolytic T cells that display dichotomous type I and type II cytokine production. Our findings provide evidence that human Tc1-and Tc2-type populations exist, and define strategies for in vitro generation of these subsets to evaluate their in vivo effects.

Similar to results in murine systems, the type I/type II cytokine restriction of Tc1 and Tc2 populations was relatively absolute, except for the reduced (but still significant) IFN-γ production by Tc2 cells. In addition to secreting high levels of IFN-γ, the CD8⁺ Tc1 population secreted significant amounts of the type I cytokine IL-2. We observed that the propagation of allospecific human Tc1 cells was enhanced by TGF-β; in light of our prior demonstration that TGF-β facilitates the generation of IL-2–secreting murine CD8⁺ Tc1 cells,22 we thus conclude that TGF-β promotes the development of type I CD8⁺ T cells. Interestingly, others have observed that TGF-β promotes the development of murine CD4⁺ Th1 cells.23,24 

The CD8⁺ Tc2-type population secreted significant amounts of the type II cytokines IL-4, IL-5, and IL-10. It is interesting that the Tc2 population secreted IL-5 and IL-10 after allogeneic restimulation, whereas IL-4 production was only observed after pharmacologic stimulation. These results indicate that antigen-specific IL-4 production in human Tc2 cells may require distinct activation events. The lack of antigen-specific IL-4 production in our system is consistent with previous findings that human neonatal CD4⁺ T cells secreted IL-4 after pharmacologic stimulation but not after TCR stimulation.14 The high level of IL-10 secretion by the Tc2 population may have negative implications for their clinical use in the allogeneic bone marrow transplantation setting: we have previously shown in murine models that the GVL effect of a given Tc2 population was inversely related to the magnitude of its IL-10 secretion.22 Importantly, we have not observed IL-10 secretion by allospecific Tc1 cells.

The culture methodology we used incorporates two factors that facilitated the generation of type I and type II cytokine-secreting subsets of allospecific CD8 cells. We found that the antioxidant NAC increases allospecific CD8⁺ T-cell expansion; this facilitating effect of NAC may relate to its ability to block T-cell apoptosis.25-27 In addition, we found that culture of CD8 cells in the presence of IL-7 increased cell yields for both Tc1 and Tc2 cultures; IL-7 has been identified as a cytokine that supports T-cell generation both in vitro28-30 and in vivo.31,32 

Because both cytokine-secreting subsets of allospecific CD8⁺ T cells possessed potent cytolytic function, we conclude that Tc1 and Tc2 subsets exist in both murine and human systems. Using bone marrow–derived cells as both allostimulators and targets, we demonstrated that the Tc1 and Tc2 populations mediated lysis in an allospecific manner. As a result, we conclude that the Tc1 and Tc2 populations were allospecific in terms of both cytokine secretion and cytolytic function, and that the human Tc1 and Tc2 populations are analogous to the murine CD8⁺ T-cell subsets we have used in our transplantation models.

Our observation that the Tc1 and Tc2 subsets lysed the modified xenogeneic target after incubation with the TCR-activating heteroconjugate antibody illustrates that both CD8 subsets are capable of potent cytotoxicity when stimulated through the TCR. Because cytolytic function mediated through the TCR likely reflects perforin- and granzyme-mediated killing,33 our results suggest that human Tc1 and Tc2 populations have equivalent utilization of this cytolytic pathway. It is important to note that murine Tc1 and Tc2 subsets appear to equally use the perforin/granzyme cytolytic pathway.21 

Murine studies have suggested that a second major pathway of cytolysis, the fas pathway,34 may be differentially used by CD8⁺ Tc1 and Tc2 populations: we5 and others21 have shown that murine Tc1-type CD8⁺ cells preferentially express functional fas ligand. In contrast to our murine studies, where functional fas ligand was detected within the Tc1 subset after allogeneic restimulation, we were unable to detect fas-based cytolytic function in human Tc1 and Tc2 cells after allogeneic restimulation. However, because high levels of fas-based killing were observed in Tc1 and Tc2 populations after pharmacologic stimulation, the potential for functional fas ligand expression clearly exists within both Tc1 and Tc2 subsets of human CD8⁺ T cells. Given the emerging role of the fas pathway in immune regulation, its known role in the pathogenesis of GVHD,35 and its potential role in the mediation of antitumor responses,36 further investigation into Tc1 and Tc2 utilization of this pathway is important.

In summary, we have demonstrated that allospecific T cells of classic Tc1 and Tc2 phenotype exist in cultured human CD8⁺ T-cell populations, and have defined an in vitro culture methodology for their rapid generation. Our findings indicate that clinical studies evaluating the in vivo role of Tc1 and Tc2 populations in the allogeneic bone marrow transplantation setting are possible.