Large-Scale Production and Characterization of Novel CD4⁺ Cytotoxic T Cells with Broad Tumor Specificity for Immunotherapy Free

One of the most promising anticancer therapies is to use the patient's own immune system to eradicate tumor cells. Based on natural killer (NK) cells (1, 2), dendritic cells (3), hematopoietic stem cells (4), T cells (5) with focus on CTLs (6, 7), as well as genetically engineered T cells (8), multiple therapeutic concepts were developed for adoptive immunotherapy of a large variety of cancers like breast (9), prostate (10), and ovarian (11) cancers, as well as against melanoma (12), childhood malignancies (13), medullary thyroid carcinoma (14), or renal cell cancer (15).

The idea of adoptive immunotherapy goes back to results in the early 1980s showing that exposure of peripheral blood mononuclear cells (PBMC) to high concentrations of interleukin (IL)-2 led to the generation of lymphokine activated killer cells, which consisted mainly of NK cells and showed significant lysis of tumor cells in vitro sparing normal cells. In humans, however, lymphokine activated killer cells in combination with high-dose IL-2 gave response rates of only up to 20% (16). Further development of adoptive immunotherapy led to the use of tumor-infiltrating lymphocytes (TIL) with very promising results in mouse models (17). Since then, a huge effort has been undertaken to establish adoptive immunotherapy using TILs in humans. TIL cultures were expanded from tumors of different epithelial origin but rarely showed specific antitumor activity. On the contrary, melanoma TIL cultures could be generated more reliably, and patients who had received TILs expanded from subcutaneous tumor deposits showed response rates of up to 49% (18). Nowadays, patients with metastatic melanoma can be treated quite effectively by adoptive cell therapy using TILs, which mediate cancer regression in ∼50% of patients. Furthermore, this approach has been extended by genetic engineering of human lymphocytes to yield even higher response rates (6). However, the attempt to prolong the in vivo survival of transferred TILs by insertion of the IL-2 gene could not improve in vivo persistence or clinical effectiveness, in contrast to results from in vitro experiments (19). Nevertheless, the expansion of TILs from tumor material is labor- and time-intensive and an additional strain on the patient because tumor tissue has to be excised in a biopsy (20).

Antigen-independent activation of PBMCs derived from cancer patients [e.g., by using CD3 monoclonal antibodies (mAb; refs. 6, 21)] generated activated T cells, which exerted a certain antitumor effect when reinfused into the patient, whereby the clinical outcome was mostly poor. Furthermore, a limiting step in the cultivation of all these effector cells and, in particular, cells of T-cell origin is the demand of high amounts of IL-2 and specific culturing skills.

Among the CTLs, most are of the CD8⁺ phenotype that recognize antigen presented on MHC class I; however, ∼10% of human CTLs are CD4⁺ and recognize antigen presented via MHC class II (19, 22). Additionally, some CD4⁺ CTLs restricted to MHC class I have been reported (23-25). It is important to note that CD4⁺ CTLs are not, as previously postulated, an in vitro artifact (26) produced by prolonged activation under proliferative in vitro conditions, but increased in viral infection (27), autoimmune disorders (28, 29), or cancer (25). CD4⁺ CTLs seem to be different from CD4⁺ T cells and exhibit the phenotype of memory cells (CD11a^high, CCR7⁻). In addition, losses of CD27 and CD28 on the cell surface are a sign of an advanced stage of cellular differentiation (30).

Unlike conventional CD4⁺ T cells functioning as immune mediators by the production of cytokines, CD4⁺ CTLs induce apoptosis in target cells. Results from studies investigating their death mechanisms show the role of perforin as a major effector molecule (23, 29, 31-34). Furthermore, Fas ligand (32) and tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) are involved. The latter is a powerful inducer of apoptosis that acts through an unusually diverse system of receptors (32, 35, 36). TRAIL is important for immune surveillance, especially in the context of cancer (32, 35, 37, 38). It is not displayed on the surface of resting peripheral blood T cells but is significantly induced by stimulation with CD3 mAbs in the presence of IFNs (39). CD4⁺ T-cell clones were found to constitutively express TRAIL, which was fully or at least partially responsible for cytotoxicity against certain tumor cells (39). Sato et al. (40) reported that CD4⁺ CTLs acting via the TRAIL pathway were able to induce apoptosis in vascular smooth muscle cells present in atherosclerotic plaques of patients with acute coronary syndromes. These cells originated from a CD4⁺ CD28⁻ subset that seemed to be enriched in the patients' blood (40). TRAIL as well as agonistic antibodies specific to TRAIL receptors, death receptors DR4 and DR5, have progressed to phase I and II clinical trials as potential therapeutic agents for a variety of malignancies (41, 42). At present, the classification of CD4⁺ CTLs can neither be assigned to distinct types of cancer targets nor are they always restricted to MHC (23, 25, 43-45). Their cytokine pattern varies as does their way of exerting cytotoxicity (23, 25, 43-48).

Because the variety of adoptive immunotherapy against cancer is as large as the promises it holds, in-depth understanding and improvements are of high relevance. We present here a novel method that allows the generation of a subset of CD4⁺ cytotoxic T cells, called tumor killer T cells (TKTC). These cells induce apoptosis in tumor cells of various tissue origins and leave normal cells unaffected. The cytotoxicity could be attributed, to a significant extent, to a TRAIL-dependent mechanism. The generation procedure is relatively easy to perform and does not require autologous tumor material but only small amounts of peripheral blood. The cultivation under serum-free conditions meets the requirements of their use as a therapeutic agent. Thus, this technology together with the special features of TKTCs might be the basis by which a novel type of adoptive immunotherapy of cancer may evolve.

We found that among PBMCs, a T-cell population strongly proliferated on stimulation with the EBV-transformed B cell line HB617, and that depletion of cytotoxic cells using l-leucyl-l-leucine methyl ester improved expansion of these cells. l-Leucyl-l-leucine methyl ester is a lysosomatropic compound that is converted by dipeptidyl peptidase I to metabolites that are membranolytic for cytotoxic T cells, NK cells, and lymphokine activated killer cells (47).

To prepare the culture for adoptive transfer purposes, we established serum-free conditions. A stimulating pulse with HB617 for 5 days was sufficient, and the optimal growth factors were a combination of IL-2 for continuous growth plus the immunosuppressive drug cyclosporin A. The culture was adjusted to 3 × 10⁶ cells/mL and readjusted every 2nd day. When the culture reached a volume of 50 mL, it was transferred into a 490-cm² roller bottle or a Roux flask to compare TKTC growth under two different conditions. The cells propagated in the Roux flasks ceased to grow after 5 weeks (data not shown). However, cultures grown in the roller bottles could be propagated up to 10¹⁰ to 10¹² cells in a period of about 8 to 10 weeks (Fig. 1A shows four representative growth curves). The enormous expansion of the culture in roller bottles is probably due to an improved oxygen supply and the increased cell density at the bottom of the bottle. The growth in the roller bottle was very consistent and allowed a 1.5-fold expansion every 2nd day. After 8 to 10 weeks in exponential growth phase, the cells stopped dividing and persisted in a nonproliferating state, maintaining a relatively high viability (≥80%). After a further 2 weeks, the viability rapidly decreased and the culture had to be abandoned. In total, we generated using this method eight different TKTC cultures: five from PBMCs of healthy donors (KT-HHD1 to KT-HHD5) and three from PBMCs of cancer patients (KT-PC1, KT-PC2, and KT-HCC1).

To exclude any risk of virus transmission (hypothetical minute amounts of EBV or other unknown viruses), we performed a virus safety study to evaluate virus depletion and inactivation during the preparation of the stimulating cell line HB617, although it does not produce any detectable EBV virus (data not shown). Therefore, we mixed HB617 with two model viruses, polio and influenza. When we analyzed the virus titer after glutaric aldehyde fixation and the various washing steps, all viruses analyzed were reduced to the limit of detection of the assays. The reduction factors were 9.8 log/mL for polio and 12.8 log/mL for influenza (data not shown). Thus, we conclude that hypothetical viruses produced by HB617 cells are significantly removed and inactivated during the preparation process.

In all TKTC lines shown in this article, the percentage of CD4⁺ cells increased during cultivation, whereas the percentage of CD8⁺ cells decreased. After 8 weeks in culture, three of the eight TKTC lines contained basically only CD4⁺ cells (KT-PC1, 100%; KT-HCC1, 100%; KT-HHD2, 99%). Two lines contained ∼90% CD4⁺ cells (KT-HHD4, 91%; KT-PC2, 94%) and three contained ∼80% CD4⁺ cells (KT-HHD1, 80%; KT-HHD3, 82%; KT-HHD5, 83%); the residual cells in the latter five cultures were CD8⁺. The cells further expressed CD3, CD25, CD28, CD45RA, CD45RO, CD69, CD86, and MHC-II (Fig. 1B). Between the different TKTC cultures, the expression of surface markers was rather consistent. Only MHC class II displayed variation, and KT-HHD2 cells showed an extraordinarily high expression of the activation and memory marker CD45RO. In the populations containing a residual amount of CD8⁺ cells (KT-HHD1, KT-HHD3, KT-HHD4, and KT-HHD5), CD31 was solely expressed by the CD8⁺ subset (data not shown). The memory marker CD27 and chemokine receptor 7 (CCR7) were never detected. All TKTC cell lines analyzed were negative for CD56, CD57, CD161, and NKG2D at all times tested (i.e., after 4, 6, and 8 weeks in culture; Fig. 1B).

To evaluate Foxp3 expression, we performed real-time quantitative PCR. Conventional CD4⁺CD25⁻ and CD4⁺CD25⁺ T regulatory lymphocytes (Treg) served as negative and positive control cells, respectively. TKTC lines expressed Foxp3 mRNA, however, at variable degrees (Fig. 1C). The highest Foxp3 mRNA expression (1.8× more than in CD4⁺CD25⁺ T lymphocytes) was found in the prostate tumor patient derived line KT-PC1, the lowest in line KT-HHD5, 16× less than in CD4⁺CD25⁺ T lymphocytes but 5× more than in CD4⁺CD25⁻ T lymphocytes.

Because of its high FoxP3 mRNA expression level, we subjected the KT-PC1 line to a T-cell suppression assay using CD4⁺CD25⁺ Tregs and CD4⁺CD25⁻ conventional T cells freshly isolated from human peripheral blood. Whereas the Tregs suppressed the conventional T cells in a dose-dependent manner, we did not observe any suppressive effect of the KT-PC1 line (Fig. 1E).

We then analyzed Foxp3 expression in KT-PC1 at the protein level using immunofluorescence staining and flow cytometry of permeabilized cells. In contrast to the mRNA level, the protein staining was low (Fig. 1F). We obtained the same results with TKTCs KT-HHD3 and KT-HCC1 (data not shown), indicating posttranscriptional control of Foxp3 expression in TKTCs.

As determined by reverse transcriptase PCR, all TKTC cultures contained exclusively α/β T cells. The expression of the T-cell receptor (TCR) chains was assessed from transcripts of each TKTC culture. Younger cultures of 5 weeks of age displayed rearrangement of all 47 TCR chains tested (Fig. 1D, top), similar to that of freshly isolated PBMCs (data not shown). After 6 to 8 weeks in culture, TKTCs lost expression of some TCR chains (Fig. 1D, bottom). Older cultures of 10 weeks of age became oligoclonal, displaying only two to five TCR chains. However, the TCR expression in oligoclonal TKTC cultures varied. In general, the faster a TKTC culture grew, the earlier the development of oligoclonality appeared.

After a cultivation period of 6 weeks, we tested the supernatants of the TKTC lines for cytokine content by ELISA. As depicted in Table 1, the TKTC lines produced IL-10 and IFN-γ at high variation. IL-4 was produced by four TKTC lines (KT-HHD1, KT-HHD4, KT-HHD5, and KT-PC1) in low concentrations (3.6-37.3 pg/mL) and by others (KT-HHD2, KT-HHD4, and KT-HCC1) below the detection limit of the assay.

To test the effector function of the generated cells, we analyzed the cytotoxicity of TKTCs toward a panel of tumor and normal cells and quantified the killing efficiency by flow cytometry. To distinguish the target cells in the coculture from TKTCs, we labeled them before the assay with carboxyfluorescein diacetate N-succinimidyl ester (CFSE). After the cocultivation, we stained the sample with propidium iodide (PI) to assess the percentage of dead target cells. Coincubation of fibrosarcoma Saos cells with a 10-fold excess of TKTCs led to an extensive increase in the percentage of PI⁺ tumor cells (Fig. 2A): 50% when incubated with the healthy donor TKTC culture KT-HHD5, and 64% by incubation with the cancer patient TKTC culture KT-PC1. The basal level of dead Saos cells was 4%. In contrast, when we exposed normal umbilical vein endothelial cells IE 104.4 to KT-HHD5 or KT-PC1, the basal level of 3% did not significantly increase but stayed low at 4% in both cases.

This discrimination potential of the TKTCs was confirmed when we analyzed other normal and tumor-derived cells. As summarized in Table 2, all tumor-derived cell lines were killed by TKTCs, whereas normal cell types and lines were unaffected. There were, however, two exceptions: Burkitt's lymphoma Daudi cells and chronic myelogenous leukemia K562 cells, both classic NK cell targets, were not killed by TKTCs. Among the susceptible cells, there was a variation in the killing kinetics (Fig. 2B). The fibrosarcoma cell line Saos as well as melanoma and medullary thyroid carcinoma cell lines were faster killed by TKTC (50-70% dead cells after 20 hours) compared with HT-29 or Me180 cells (∼30% dead cells after 20 hours). These results show that TKTCs have a high capability to discriminate between normal and transformed cells.

To uncover the mechanisms underlying killing by TKTCs, we watched the induction of cell death by microscopy. Depending on the type of target cell, the process took place between 24 and 72 hours until all targets were eradicated. In particular, TKTCs required a longer time to exert their cytotoxicity when the targets grew in clusters, such as the HT-29 cells. The visual investigations showed a very tight contact between effector and target cells, suggesting that TKTCs kill their targets in a direct cell-to-cell contact manner (Fig. 3A). This assumption was confirmed by a transwell chamber setting, which did not lead to killing of the tumor cells (data not shown).

Next we assessed whether TKTC-mediated cytotoxicity in target cells is mediated by induction of apoptosis. For that purpose, we modified our flow cytometric cell death assay. Target cells were labeled with the membrane dye PKH-26 before exposure to effector cells. After 20 hours of exposure, cocultures were harvested and stained with Annexin V for detection of early apoptotic cells and with PI, which penetrates late apoptotic and necrotic cells. In Fig. 3B, we give an example of such an assay done on Saos cells, which were identified according to PKH-26 staining and analyzed for different stages of cell death. The analysis revealed an increase in early apoptosis to 12% at an effector/target ratio (E/T) ratio of 10:1 and 18% at an E/T of 20:1. Under the influence of TKTCs for 20 hours, the late apoptotic/necrotic Saos population reached a level of 45% at an E/T of 10:1 and 50% at an E/T of 20:1, whereas only 13% of Saos cells underwent apoptosis spontaneously.

As mentioned above, TKTC cultures do not have the capacity to grow indefinitely but stop proliferating after an exponential growth phase over ∼8 to 10 weeks. During the following 2 weeks, persisting in a nonproliferative (potentially senescent; refs. 49, 50) state, the cells still possess cytotoxic activity. We show this function in Fig. 3C with the MelLa melanoma cell line as target and KT-HHD1 as TKTC effector. The latter had been cultured for 11 weeks before it stopped proliferation and then was kept for another week in culture. A 10-fold excess of the nonproliferating TKTCs led to 27% and a 20-fold excess to 54% PI⁺ MelLa cells, in contrast to a spontaneous cell death of 4% in the control assay without effector cells.

To determine whether TKTC cytotoxicity is MHC restricted, we first assessed expression of MHC-I and MHC-II molecules on target cells using a panel of mAbs to different MHC class I and II epitopes (data not shown). Next, we incubated the tumor cells with those mAbs that positively reacted with the cells to block the MHC molecules and then subjected the treated cells to the flow cytometric cytotoxicity assay. The fibrosarcoma Saos cells were positive for MHC class I; however, none of the anti–MHC class I mAbs had an influence on killing of these cells by the TKTC culture KT-HHD2. In contrast, the cytotoxicity of a positive control CD8⁺ CTL line toward Saos cells was greatly impaired by the anti–MHC class I mAbs (Table 3). Furthermore, KT-HHD2 killing of MHC class II⁺ MelE melanoma cells was not influenced by three mAbs directed to different epitopes of MHC class II (data not shown). Because none of the mAbs of both the MHC class I and MHC class II panels inhibited killing, it is very likely (but not 100% proved) that TKTC killing is not restricted to MHC.

Next we analyzed, whether TKTCs use perforin-, Fas-, or TRAIL-based cytolytic pathways for cytotoxicity. Flow cytometry revealed a low expression of perforin in the TKTC lines (Fig. 4A). Experiments using concanamycin A to block perforin-mediated cytotoxicity mechanisms indicated that TKTC killing did not involve perforin (data not shown). Further, mAb SM1/23, the blocking mAb of Fas, was not able to reduce the cytotoxic effect of TKTCs, whereas it significantly reduced cell death induced by anti-Fas mAb CH-11 in the control cell line LN-18 (Fig. 4B). However, inhibition of TRAIL binding to its receptor resulted in a remarkable reduction of the amount of dead tumor targets. This effect was dependent on the concentration of the blocking anti-TRAIL mAb RIK-2 (Fig. 4C). Depending on the tumor target, killing was reduced by 17.4% to 46.8% (Table 4).

To analyze the TRAIL-mediated apoptosis induction in tumor cells further, we analyzed surface expression of the TRAIL receptors TRAIL-R1 and TRAIL-R2. Because TRAIL receptors were reported to be expressed in some tumor cells only on stimulation (51, 52), we also tested for intracellular TRAIL receptors. As shown in Fig. 5, we found a correlation between expression of the TRAIL receptors on tumor cells and susceptibility to TKTC killing. All five susceptible tumor cell lines tested (Saos, MelE, MelLa, MelJuso, and HT-29—staining pattern identical to that of Saos cells; not shown) expressed TRAIL receptors either on the surface or intracellularly.

On the contrary, the TKTC-resistant normal human renal proximal tubular epithelial cells NHK (53) and the normal human skin fibroblasts HDF-5 (54) did not express significant levels of TRAIL receptor. Normal human umbilical vein endothelial cells IE 104.4 as well as the classic NK target cell line K562, although not susceptible to TKTC killing, expressed TRAIL receptors. However, the latter finding is in agreement with recent reports, which show that TRAIL does not promote apoptosis but rather survival and proliferation in endothelial cells (55) and promotes apoptosis in K562 cells only on sensitization by specific agents (56-58). Thus, our results indicate that the TRAIL pathway is at least partially involved in the TKTC-mediated induction of apoptosis in tumor cells.

It is extremely difficult to obtain tumor as well as normal tissues from tumor patients, especially from those suffering from late-stage disease. Nevertheless, we were able to grow normal fibroblasts from the prostate cancer patient from whom TKTC line KT-PC2 was derived. Because normal cells have very long population doubling times, we were limited in target cell material and could only conduct microscopic cytotoxicity assays. Figure 6 shows the experiment conducted over a period of 1 week. Over that period, KT-PC2 cells, which had been added to the patient's normal fibroblasts in an E/T ratio of 10:1, vanished, whereas the autologous fibroblasts proliferated and almost formed a confluent layer after 7 days. Thus, the TKTCs were not able to induce cell death in the autologous fibroblasts. The establishment of tumor cell lines from primary tumors is not trivial and thus hampered our attempts to perform cytotoxicity experiments on autologous tumor cell lines.

We show here a novel method to generate large amounts of TKTCs, a CD4⁺ population of cytotoxic T cells from PBMCs of peripheral blood, which induce apoptosis involving TRAIL in many transformed cells of different tissue origins but leave normal cells unaffected.

We found partial inhibition of TKTC killing of tumor cells by an anti-TRAIL mAb and correlation between expression of TRAIL receptors on target cells and susceptibility to TKTC killing. Factors that influence TRAIL sensitivity in multiple cancer cells are not well defined. A new study suggests that O-glycosylation of death receptors expressed in cancer cells modulates the sensitivity to TRAIL (36). In addition to the general lack of knowledge on the TRAIL mechanism, TKTCs provide another puzzle: Although anti-TRAIL mAbs reduced significantly the cytotoxic effect of TKTCs, we were unable to show the expression of TRAIL on the surface of TKTCs by flow cytometry (data not shown). TRAIL is also not found on the surface of peripheral blood T cells but expressed by costimulation with CD3 and type I IFNs (39). Further, on the surface of T-cell lines, TRAIL was shown to be constitutively expressed at various levels, and some of these lines initially down-regulate it before reexpression on CD3 activation (59, 60). This shows specific requirements and a high dynamic of TRAIL expression in T cells on differentiation/activation. Thus, we speculate that TKTCs might need a specific interaction with the susceptible target cells for TRAIL expression.

Differences in target cell recognition involving regulation of TRAIL expression may be as well a possible explanation for the delayed kinetics of TKTC cytotoxicity and the extent of killing dependent on the type of the target cell. The latter may be also influenced by the growth characteristics of the targets such as growth in clusters, which might hamper target cell recognition. Indeed, HT-29 cells that grow in clusters are less susceptible to TKTC killing. The IFN-γ produced by TKTCs may also be involved because it was reported that IFN-γ mediates susceptibility to TRAIL killing (61, 62). Furthermore, susceptibility of target cells may be dependent on their expression and exposure of TRAIL receptors. In two tumor targets, HT-29 and Saos, TRAIL-R1 and TRAIL-R2 could only be detected intracellularly. Because TKTCs need longer time to kill HT-29 than Saos cells, one could speculate that TRAIL receptors are probably differently transported to the surface of these cells because intracellular TRAIL receptors are only transported to the cell surface on contact.

Based on expression profiling and functional characteristics, TKTCs cannot be assigned to any known subtype of T cells. Although the expression profile of TKTCs (i.e., up-regulation of CD45RO, down-regulation of CD28 together with the lack of CD27 expression) resembles an effector memory phenotype (CD45RO⁺ CD27⁻ CD28⁻ CCR7⁻; refs. 63, 64), they coexpress also the CD45RA isoform of the tyrosine phosphatase CD45, which is a feature of continuously cycling cells (63) retaining their functional program after activation (65). CD45RA⁺ CD45RO⁺ double-positive T cells can be isolated from human peripheral blood, where they show an intermediate phenotype for CD31, CD62L, CD58, and CD95, which are known to be differentially expressed on unprimed versus primed cells (66). However, TKTCs are negative for CD31 and CD62L.

CD4⁺ TKTC cultures do express on their surface at low density CD25 molecules, the IL-2 receptor α chain and one of the hallmarks of Tregs. Tregs can be subdivided into CD25^high and CD25^low cells. CD25^high cells represent about 2% to 3% of the total CD4⁺ T-cell population, and at least in in vitro settings exhibit the characteristics postulated for Tregs initially identified in the mouse (67). Whereas the CD25^high cells seem to be rather homogeneous in terms of their surface antigen expression, CD25^low cells are more heterogeneous, which we could also somehow detect for TKTCs, in particular for MHC class II and CD45RO. However, CD69 (an antigen indicating recent activation) is not expressed on Tregs, but TKTCs were determined to be CD69⁺. Therefore, TKTCs seem to belong to a different cell subset. This assumption is further supported by the fact that Tregs can only be expanded to high numbers on antigenic stimulation in combination with IL-15, compared with activation in combination with IL-2 (68). Moreover, Tregs preferentially kill targets using a perforin-dependent mechanism (69), whereas TKTCs barely express perforin (Fig. 4A) and TKTC killing was not blocked by concanamycin A, which should block perforin-mediated cytotoxicity. We also evaluated Foxp3 expression. In contrast to mice, human Foxp3 is not restricted to Tregs but rather a T-cell activation marker probably counterbalancing the stimulation process for maintaining homeostasis (70). All TKTC lines tested expressed Foxp3 mRNA, but at variable degrees from very low levels to high expression (16× less up to 1.8× more than CD4⁺CD25⁺ Treg lymphocytes that served as positive control). Strikingly, in the mRNA high lines, Foxp3 expression was low at the protein level, suggesting posttranscriptional control of Foxp3 in these TKTCs. Finally, we tested the TKTC line KT-PC1 in a T-cell suppression assay, because it showed the highest Foxp3 mRNA expression. As shown in Fig. 1E, this line did not show any suppressive capacity.

TKTCs can also be clearly distinguished from natural killer T (NKT) cells. Although killing of TKTC, similar to that of CD4⁺ NKT cells, is MHC independent (71), in contrast to NKT and NKT-like cells, TKTCs do not express the typical NK cell markers CD56, CD57, CD161, and NKG2D. Furthermore, classic NKT cells are characterized by the expression of distinct TCR chains, namely Vα24 and Jα18 coexpressed with Vβ11. However, the TKTC lines are not restricted to the NKT TCRs but vary in their TCR expressions.

The fact that TKTCs secrete IFN-γ rather than IL-4 leads to the assumption that this cell type might be related rather to T cells of the Th1 type than to Th2 or Tregs. In addition, TKTCs also produce IL-10. During many infections, IL-10–producing effector Th1 cells are generated and thought to represent a regulatory mechanism to dampen exaggerated inflammation (72). Thus, it will be a future goal to investigate whether the IFN-γ/IL-10 double-producing TKTCs could both exert antitumor capacity and, via secretion of IL-10, locally control the defense mechanism thereby preventing systemic inflammation.

During the first few weeks of a TKTC culture, the permanent presence of IL-2 as well as components of HB617 might lead to an unspecific activation and proliferation of all T cells in the culture. However, to cause distinct T-cell clones to proliferate for 2 months and beyond, a stronger activation signal must be provided. Whether this signal is due to specific antigen stimulation or not, we tested the TKTC lines for TCR clonality by PCR. TKTC cultures of 4 to 6 weeks of age showed polyclonality. Because TCR diversity diminished in the course of cultivation (e.g., KT-PC1 turned out to be oligoclonal and KT-HCC1 was even monoclonal), it is tempting to speculate that a specific kind of signal leads to the activation and proliferation of TKTCs. However, the V segment usage varied between the TKTC cultures: Oligoclonal cultures expressed different TCR-α and TCR-β chains.

It should be noted that the stimulating cell line HB617 is EBV⁺. It is well documented that CD4⁺ CTLs are increased in EBV infection (27). In fact, patients experience a primary burst of CD4⁺ effector T cells, which are then down-regulated. EBV-specific CD4⁺ effector T cells stay enriched in the CD27⁺CD28⁺ compartment during the primary and persistent phases of infection (30, 73). Furthermore, many of these T cells coexpress CCR7 as opposed to TKTCs, which clearly were confirmed to be CD27⁻ and CCR7^-. Unlike EBV-associated CD4⁺ CTLs described in the literature, which gain cytotoxic potential by the acquisition of lytic granules containing granzymes as they lose CD27 expression and acquisition of perforin as soon as they are CD28⁻ (30), TKTCs exert cytotoxicity, at least in part, by the TRAIL pathway. Moreover, EBV-specific CD4⁺ and CD8⁺ effector T cells are MHC restricted (30, 74), whereas our data indicate that TKTC cytotoxicity is rather MHC independent.

In summary, we provide a relatively simple and cheap protocol to generate large amounts of a novel type of CD4⁺ CTLs, termed TKTCs, with broad tumor specificity. Whether this population exists in vivo and/or is expanded under certain pathologic conditions remains to be elucidated in next studies. However, we cannot grow TKTCs in vitro without depletion of monocytes, NK cells, and suppressor T cells by using l-leucyl-l-leucine methyl ester. Thus, based on these in vitro studies, we suggest that in vivo the generation of this subset is likely to be impaired and controlled by other cells of the immune system. The charm of this particular T-cell subset for potential tumor patient therapy is given by the possibility to generate it in large amounts from small samples of peripheral blood and its broad killing of tumor cells involving the TRAIL pathway. Many tumors manage to develop resistance to conventional therapy on treatment. However, TRAIL is able to bypass this antiapoptotic effect in some tumor systems (75). Therefore, research groups investigating antitumor immunotherapies started to focus on TRAIL (76). In addition, the fact that TKTC killing is probably MHC unrestricted might be an additional advantage in patients whose tumor has down-regulated MHC expression to escape immune surveillance. These characteristics of TKTCs together with the technology to generate them serum-free under good manufacturing practice conditions prove this novel cell type as a potential new tool for adoptive immunotherapy against cancer.

Heparinized human whole blood donations were purchased from the Austrian Red Cross for the preparation of five TKTC cultures from healthy human donors, designated KT-HHD1 to KT-HHD5. Heparinized whole blood from one colorectal carcinoma patient as well as from two prostate adenocarcinoma patients was obtained using sodium heparin–coated Vacutainer vials (Becton Dickinson) for the establishment of tumor patient TKTC cultures designated KT-HCC1 (killer T-human colorectal carcinoma patient 1) and KT-PC1 and KT-PC2 (killer T-prostate carcinoma patients 1 and 2). PBMCs were isolated by Ficoll-Paque (GE Healthcare) gradient centrifugation. Monocytes, NK cells, and cytotoxic T cells were eliminated by incubation with l-leucyl-l-leucine methyl ester hydrochloride (NovaBiochem) for 20 min at room temperature as described (47). Remaining cells were stimulated with the lymphoblastoid cell line HB617 (fixed with 0.02% glutaraldehyde for 10 min at room temperature) in vitro for 5 d and then propagated in serum-free medium (Aim-V, Life Technologies, Inc.) supplemented with 360 IU/mL IL-2 (Proleukin, Chiron Corp.) and 12.5 ng/mL cyclosporine A (Novartis Pharma). The cell number was adjusted to 3 × 10⁶/mL every other day, and the population doubling time approached 3 d. TKTCs were propagated in Roux flasks until they had reached a volume of 50 mL. The cell number was upscaled in roller bottles rotating at 1 rpm at 37°C and 5% CO₂.

The stimulator cell line HB617 and the target cell lines Daudi (American Type Culture Collection), K562, and Saos were cultivated in X-Vivo 20 medium (Cambrex Bio Science) and passaged every 3 to 4 d. All melanoma cell lines—MelE, MelLa, and MelJuso—were cultivated in RPMI 1640 (Biochrom AG)/4 mmol/L l-glutamine/10% FCS (Hyclone) and passaged every 3 to 4 d at a split ratio of 1:10. The colon carcinoma cell line HT-29 (American Type Culture Collection) was cultivated in D-MEM/Ham's F-12 1:1 (Biochrom)/4 mmol/L l-glutamine/10% FCS and passaged every 3 to 4 d. The uterine cervix carcinoma cell line Me180 (American Type Culture Collection) was propagated in RPMI 1640/4 mmol/L l-glutamine/10% FCS and passaged every 3 to 4 d. LN-18 were grown in D-MEM (Biochrom)/4 mmol/L l-glutamine/5% FCS and passaged every 3 to 4 d, 1:5. The medullary thyroid carcinoma cell line (kindly provided by R. Pfragner, Institute of Pathophysiology, Medical University of Graz, Graz, Austria) was cultivated in Ham's F-12 (Biochrom)/4 mmol/L l-glutamine/10% FCS and passaged every 3 to 4 d at a split ratio of 1:1 or 1:2 as described (77). Target cells were split at a ratio of 1:5, unless indicated otherwise.

Normal human umbilical vein endothelial cells IE 104.4 were isolated using standard procedures and propagated in M199 (Biochrom)/15% FCS/10% endothelial cell growth supplement with a subculture ratio of 1:2 every 3 to 4 d (78). Normal human kidney epithelial cells NHK-13, NHK-27, and NHK-16 were cultivated in serum-free medium and passaged every 3 to 4 d at a split ratio of 1:2 as described (79). Normal human fibroblasts HDF-5 and HDF-6 were isolated from skin biopsies (28) and cultivated in D-MEM/Ham's F12 (1:1)/4 mmol/L l-glutamine/10% FCS and passaged at a split ratio of 1:2 every 3 to 4 d (54). Concanavalin A–stimulated blast cells were generated by culture of freshly isolated PBMCs in RPMI 1640/4 mmol/L l-glutamine/10% FCS supplemented with 5 μg/mL concanavalin A (Sigma-Aldrich; ref. 80) and 10 ng/mL IL-2 for 5 d. All cells were cultivated in a humidified atmosphere at 37°C and 5% CO₂.

Polio as a small, nonenveloped, robust virus and influenza representing medium to large, enveloped viruses were chosen as model viruses based on the recommendation of the Committee for Proprietary Medicinal Products (81). To show virus removal and inactivation in the course of the fixation and washing steps of HB617, the protocol for TKTC preparation was done as described above using HB617 cell cultures mixed with polio or influenza virus, respectively, before the centrifugation/washing step, before fixation, and before washing after fixation. After addition of fixed HB617 to PBMCs on day 1, virus was measured and virus reduction factors were calculated for each individual step as described earlier (81), which resulted in the cumulated reduction factor.

The following mouse anti-human mAbs were kindly provided by Vaclav Horejsi (Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Prague, Czech Republic): MEM-57 (CD3ε; IgG2a), MEM-115 (CD4; IgG2a), MEM-31 (CD8; IgG2a), MEM-174 (CD11b; IgG2a), MEM-18 (CD14; IgG1), MEM-168 (CD16; IgM), WIN-19 (CD19), MEM-97 (CD20; IgG1), MEM-181 (CD25; IgG1), MEM-56 (CD45RA; IgG2b), MEM-28 (CD45; IgG1), MEM-102 (CD48; IgG1), MEM-188 (CD56; IgG2a), MEM-43/5 (CD59; IgG2b), MEM-75 (CD71; IgG1), MEM-119 (MHC-I; IgM), MEM-123 (MHC-I; IgG3), MEM-135 (MHC-I; IgG1), MEM-147 (MHC-I α chain; IgG1), MEM-149 (MHC-I; IgG1), MEM-32B (MHC-II DR dimer; IgG1), MEM-136 (MHC-II DR, DP, DQ β chain; IgG1), and MEM-137 (MHC-II DR α chain; IgG1). The mouse anti-human mAb 248.23.2 (CD28; IgM) was a kind gift from U. Moebius (University of Heidelberg, Heidelberg, Germany), mouse anti-human mAb UCHL-1 (CD45RO; IgG2a) from Peter Beverley (Jenner Institute, Compton, United Kingdom), and mouse anti-human mAb FN50 (CD69; IgG1) from Steinar Funderud (Norwegian Radium Hospital, Oslo, Norway). Other mAbs were obtained from the 6th Human Leukocyte Differentiation Antigen workshop: Dreg56 (CD62L; IgG1) and BU63 (CD86; IgG1). Mouse anti-human mAbs to TRAP1 (CD154; IgG1), RIK-2 (TRAIL; IgG1), TRAIL-R1 (DR4), TRAIL-R2 (DR5), perforin (δG9; IgG2b, phycoerythrin labeled), and Foxp3 (259D/C7; IgG1, AlexaFluor647 labeled) were from Becton Dickinson Biosciences Pharmingen; mAb DX2 (CD95; IgG1) was from Sigma-Aldrich. The blocking mAb to human Apo-1/Fas (SM1/23) was purchased from Bender MedSystems. The apoptosis-inducing mAb to human Fas (CH-11) was purchased from Immunotech. The CD3 mAb OKT3 was from Orthoclone.

For cell surface immunofluorescence staining, cells were pelleted and resuspended in staining buffer (PBS/10% FCS/0.02% NaN₃). After blocking (10% human AB serum) for 10 min on ice, cells were seeded into a 96-well round-bottomed plate and titrated concentrations of primary mAbs were added. After a 45-min incubation on ice, cells were washed twice with staining buffer and resuspended in 10 μg/mL FITC-conjugated goat anti-mouse F(ab′)₂ antibody fragments (Sigma-Aldrich) diluted in staining buffer. The cells were left for 30 min on ice and then washed twice with buffer. Pellets were resuspended in PBS. Before measurement, PI (Sigma-Aldrich) was added to a final concentration of 250 ng/mL for exclusion of dead cells.

For intracellular immunofluorescence staining and flow cytometric analysis, the cells were permeabilized with 0.3% saponin for 10 min at room temperature and fixed with 1% paraformaldehyde in PBS for 15 min. Then the cells were incubated for 20 min with the primary mAbs. In case of indirect staining, the cells were washed and incubated for another 20 min with the fluorophore-conjugated secondary antibody.

Samples were measured on a FACSCalibur (Becton Dickinson) in a standard three-color setup. Analysis was done using the CellQuest and FlowJo softwares.

For quantitative real-time PCR analysis, RNA was extracted using TRI Reagent (Sigma-Aldrich) according to the manufacturer's instructions. cDNA was prepared with oligo-dT primers (VBC) and StrataScript reverse transcriptase (Stratagene). mRNA levels were quantified via the LightCycler instrument (Roche Diagnostics GmbH). Amplification was done in a total volume of 10 μL for 50 cycles of 5 s at 95°C and 15 s at 72°C. The product was detected using SYBR Green I dye (Roche Diagnostics). The relative expression of the samples was determined by normalizing the expression of each target to elongation factor 1α (EF1a) and then comparing this normalized value to the normalized expression in a reference sample (CD4⁺CD25⁻ cells) to calculate the fold change value. Primers were designed so that amplicons spanned intron/exon boundaries to minimize amplification of genomic DNA. Primer sequences were as follows: EF1a, 5′-GTGCTAACATGCCTTGGTTC-3′ and 5′-AGAACACCAGTCTCCACTCG-3′; Foxp3, 5′-GAAACAGCACATTCCCAGAGTTC-3′ and 5′-ATGGCCCAGCGGATGAG-3′.

PBMCs were isolated from blood of healthy donors by standard density-gradient centrifugation using Lymphoprep (Nycomed). CD4⁺ T cells were purified from the PBMCs by depletion of CD8-, CD14-, CD16-, CD19-, CD20-, and CD56-positive cells using mouse anti-human mAbs (CD8 mAb MEM-87, CD14 mAb MEM-18, CD16 mAb MEM-154, CD19 mAb WIJ09, CD20 mAb MEM-97, and CD56 mAb MEM-188) and magnetic sorting using goat anti-mouse IgG–coated microbeads (Miltenyi Biotec). The resulting CD4⁺ T-cell population was separated into conventional CD4⁺CD25⁻ T cells and CD4⁺CD25⁺ Tregs by using anti-CD25 magnetic microbeads (Miltenyi Biotec).

For the suppression assay, we incubated either autologous CD4⁺CD25⁺ Tregs or TKTCs together with CD4⁺CD25⁻ conventional T cells (5 × 10⁴ per well) in 96-well flat-bottomed tissue culture plates (Nunc) at the E/T ratios 1:1, 1:2, or 1:4. We induced the proliferation of the CD4⁺CD25⁻ T cells by adding irradiated non–T cells (1 × 10⁵ per well) and the CD3 mAb OKT3 at a final concentration of 1 μg/mL. The stimulatory allogeneic non–T cells were prepared by depleting T cells from PBMCs using rosetting with neuraminidase (Sigma-Aldrich)–treated sheep erythrocytes (Dade Behring) and irradiation of the obtained cell population twice at 3,000 rad. The assay was done in triplicates at a final volume of 200 μL. After 6 d of incubation, proliferation was determined by quantifying [³H]thymidine (1 μCi/well) incorporation during the last 18 h of culture. Results are presented as mean ± SD. The medium used throughout was RPMI 1640 (Sigma-Aldrich) supplemented with 5% FCS (Life Technologies), 2 mmol/L glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin (all Invitrogen).

Six weeks after isolation from whole blood, TKTC cultures were adjusted to 3 × 10⁶ cells/mL and supernatants were collected 48 h later by centrifugation of cells at 170 × g for 10 min. Samples were stored at −80°C until analyzed. Concentrations of IL-4, IL-10, and IFN-γ were determined by ELISA using 96-well microplates (Corning 9018) coated with catching antibody (diluted in PBS to 5 μg/mL, 100 μL/well) at 4°C over night. Plates were washed and blocked with PBS/2% bovine serum albumin (150 μL/well) for 1 h at 37°C. Standards as well as samples were diluted in four steps at a ratio of 1:4 in RPMI 1640/10% FCS and pipetted into the precoated plates (70 μL/well). After 90 min at 37°C, plates were washed again and biotinylated secondary antibody was applied. After 60 min at 37°C, the plates were washed, the streptavidin-alkaline phosphatase conjugate (Chemicon) was added (1:2,000), and incubation was continued for 25 min at room temperature. After another washing step, substrate was added and incubated for about 30 min at room temperature until a yellow staining of all standard dilutions appeared. Plates were measured at 405 nm (reference wavelength, 650 nm). Values were assessed in duplicates. All standards, coating, and secondary antibodies were purchased from R&D Systems.

Total RNA was isolated with Trizol reagent (Invitrogen) and 2.5 μg were reverse transcribed with Superscript II (Invitrogen) following the manufacturer's instructions. One hundredth of the resulting cDNA was used as template in a standard 50-μL PCR reaction. 5′ primers specific for 22 Vα and 25 Vβ region genes (all of which were amplified with the same 3′ Cα and Cβ region primers) were designed according to sequences of T-Cell Receptor Typing Amplimer Kits (Clontech).

For microscopic evaluation of TKTC cytotoxicity, target cells were seeded into 24-well plates. After adherence, effector cells were added at an E/T ratio of 10:1.

For the flow cytometric cytotoxicity assays, the target cells were labeled with CFSE (Sigma-Aldrich) according to the manufacturer's instructions. Briefly, target cells were washed twice in HBSS, and then 10⁶ cells were pelleted and resuspended in HBSS/10 μmol/L CFSE and incubated at 37° C for 20 min in the dark. The staining reaction was stopped by addition of 5 mL Aim-V/10% FCS. After exposure to TKTCs for 20 h in cell culture plates at various E/T ratios, the coculture was harvested, pelleted, resuspended in PBS, and stained with PI at 250 ng/mL for 5 min before acquisition. For analysis, CFSE-positive target cells were gated and evaluated for percent PI-positive cells, which were regarded as late apoptotic or necrotic. All samples were analyzed using a FACSCalibur. All assays using tumor cell targets were done in AIM-V (for serum-free target cells) or AIM-V supplemented with 10% FCS and, if necessary, growth supplements (for targets growing in serum-containing medium). Normal cell targets were cultured in the corresponding medium.

Induction of apoptosis was evaluated by a flow cytometric assay adapted from Fischer et al. (82). Briefly, target cells were labeled with the membrane dye PKH-26 (Sigma-Aldrich) according to the manufacturer's instructions for distinction between target and effector cells. After exposure to TKTCs for 20 h at various E/T ratios, the coculture was harvested. For that purpose, coculture supernatants (containing TKTCs and cell debris) were transferred into a 10-mL vial; remaining adherent cells were trypsinized and added to the supernatants. The mixture was centrifuged, washed in cold PBS, and costained with Annexin V-FITC (Roche) and PI for discrimination between viable, apoptotic, and late apoptotic/necrotic cells. Media were used as described above (cytotoxicity assays). For analysis, PKH-26⁺ target cells were gated and evaluated for percent Annexin V⁺PI⁻ cells, which were in the early apoptotic state, as well as percent Annexin V⁺PI⁺ cells, which were regarded as late apoptotic or necrotic. All samples were analyzed using a FACSCalibur.

These assays are based on the flow cytometric cytotoxicity assay described above using CFSE and PI. TKTCs were preincubated with anti-Fas mAb SM1/23 or anti-TRAIL mAb RIK-2 at a concentration of 1, 10, or 100 μg/mL for 1 h at 37°C before their addition to the targets. Blocking antibodies were maintained in cultures throughout the whole experiment. As positive control of the FAS blocking mAb, 2 × 10⁵ LN-18 cells were preincubated for 1 h at 37°C in the absence and in the presence of 1 μg/mL blocking anti-FAS SM1/23 mAb. Thereafter, the cells were treated with or without 100 ng/mL of cytotoxic anti-Fas CH-11 mAb for 48 h. After detachment and centrifugation, the percentages of dead cells were evaluated by PI staining and flow cytometric analysis on a FACSCalibur. To test for MHC restriction, targets were seeded into tissue culture plates. After adherence, MHC-I or MHC-II blocking mAbs were added according to the expression of their corresponding epitopes on the target cells at a concentration of 5 or 50 μg/mL and preincubated for 1 h before TKTC addition. After 20 h of cocultivation, all cultures were harvested and the percentage of PI positive (PI⁺) cells was assessed by flow cytometry. Results are presented as percent killing reduction [%PI⁺ cells of the control assay (without blocking agent) were taken as 100% killing] in the presence of blocking agent. Each experiment was done in duplicate; results are presented as mean ± SD.

Dr. Hermann Katinger is the founder and CEO of Polymun Scientific Immunbiologische Forschung GmbH. Thomas Hemetsberger is employed by Polymun Scientific Immunbiologische Forschung GmbH.

We thank Thomas Sterovsky for excellent technical support; Vaclav Horejsi for providing a huge panel of invaluable mAbs and Gerhard Zlabinger (Institute of Immunology, Medical University of Vienna, Vienna, Austria) for assisting and measuring the cytokines; and David Bernhard (Institute of Experimental Pathophysiology and Immunology, Medical University Innsbruck, Innsbruck, Austria) for providing CH-11 mAb.