Interferon-α Suppresses cAMP to Disarm Human Regulatory T Cells

IFN-α is a well-known antineoplastic drug that exhibits both antitumoral and immune-stimulating activities (1, 2). Among IFNs, IFN-α2 has been most broadly clinically evaluated and established as a treatment in a multitude of malignancies, including several hematologic malignancies and solid tumors (3–5). Direct effects on tumor cells include inhibition of growth-promoting cytokines, induction of apoptosis, inhibition of cell proliferation, and enhancement of tumor cell immunogenicity. Immune-stimulating actions of IFN-α are promotion of dendritic cell (DC) maturation, facilitation of T-cell activation, and stimulation of natural killer (NK) cell activity. Therapeutically applied, IFN-α also increases the risk for development of autoimmune symptoms as frequently observed in tumor patients upon prolonged treatment (6, 7).

Induction of autoimmunity by IFN-α has been ascribed to several mechanisms including amplification of T-cell activation and improved survival of activated T cells (6, 7). We recently observed that IFN-α also interferes with the induction of T-cell tolerance by reversing human tolerogenic DC function (8). However, the strong immune-activating and autoimmune-stimulating activity of IFN-α suggests that additional immune regulatory pathways could be affected.

CD4⁺CD25^highFoxp3⁺ regulatory T cells (Treg) represent a specialized T-cell subset skewed toward autoantigen recognition that plays a key role in the maintenance of self-tolerance by repressing immune responses against self-antigens (9, 10). Owing to their role in preserving the immune privilege of self-antigens, Treg confine the expansion and differentiation of T cells against tumor-associated self-antigens (11). In line, elevated numbers of Treg in the tumor and tumor-draining lymphoid tissues of patients with cancer are associated with poor prognosis (12), and the efficiency of cancer vaccination could be improved by concomitant Treg depletion or blockade (13, 14). As a consequence, emerging therapies for cancer aim at the ablation or inhibition of CD4⁺Foxp3⁺ Treg. Unfortunately, due to their phenotypic similarity, current strategies such as anti-CD25 mAb used to deplete or inhibit Treg also target activated T cells. Remarkably, clinical and immunologic features observed upon adjuvant IFN-α treatment resemble immune dysregulation in Treg-deficient individuals. Surprisingly, despite these indications pointing to an influence of IFN-α on Treg activity, the effect of IFN-α on human Treg function has not been investigated.

Here, we show that IFN-α inhibits the suppressive function of human Treg by repressing their cAMP production. IFN-α–mediated cAMP repression resulted from MEK/ERK-mediated phosphodiesterase 4 (PDE4) activation by IFN-α. Notably, IFN-α abolished the suppressive function of Treg without affecting their differentiation program. In regard to tumor immunity, functional inhibition of Treg suppression by IFN-α increased both CD4⁺ effector T-cell function and NK cell tumor cytotoxicity. In summary, these results discover Treg inactivation as an important new IFN-α function in the context of cancer therapy.

Rag2^−/−γc^−/− mice were obtained 1 to 4 days after birth from the central animal facility of the Johannes Gutenberg-University Mainz (Mainz, Germany). All animal procedures were in accordance with relevant laws (authorization G09-1-040), current institutional guidelines and conducted according to the Helsinki convention for the use and care of animals.

Leukapheresis products and buffy coats were obtained from adult healthy volunteers with approval of the local ethics committee of Rhineland-Palatinate.

CD4 (Beckman Coulter), CD25 (Miltenyi Biotec), AnnexinV, CD25, CD127, CTLA-4, CD147, CD45RO, CD45RA, HLA-DR, CD39 (all BD Pharmingen), Foxp3 (eBioscience), CD62L (Immunotools), TNFR-2, CCR7, IFNAR-1 (R&D Systems), IL-1R1 (Antigenix America), ICOS (kindly provided by Richard Kroczek, Robert Koch Institute, Berlin, Germany), and IFN receptor chain (IFNAR)-2 (PBL InterferonSource).

Amounts of IFN-γ, interleukin (IL)-10, IL-2, and TGF-β were assessed by cytometric bead array (CBA; BD Biosciences) according to the manufacturer's instructions.

STAT phosphorylation was measured in isolated Treg and effector T cells (Teff) upon incubation with 10⁴ U/mL IFN-α. Cells were fixed and permeabilized (BD Cytofix Buffer, BD PermBuffer III) and stained with anti-CD25-PE-Cy5, anti-STAT1-PE (pY701), anti-STAT3-PE (pY705), anti-STAT5-PE (pY694), anti STAT1-PE (BD Pharmingen), anti-STAT3-PE (all from BD Pharmingen), or anti-STAT5-PE (R&D Systems).

ξ-associated protein-70 (ZAP-70) phosphorylation was determined in Treg preincubated with 10⁴ U/mL IFN-α overnight and subsequent cross-linking by anti-CD3 (OKT-3) antibody. Cells were fixed with BD Cytofix Buffer and BD PermBuffer III, stained with anti-CD25-PE-Cy5/ZAP-70-PE (pY319, BD Pharmingen), and analyzed by flow cytometry (BD FACSCalibur). In some experiments, cAMP production was repressed in Treg by pretreatment with the adenylate cyclase inhibitor MDL-12 (Calbiochem; ref. 15).

CD4⁺ CD25⁻ T cells (Teff) and CD4⁺CD25⁺ (Treg) were isolated as described previously (8). For suppressor assays, 1 × 10⁵ Teff and 1 × 10⁵ Treg were cocultured in X-VIVO 20 and stimulated with 0.5 μg/mL anti-CD3 mAb (clone OKT-3) in the presence of irradiated allogeneic peripheral blood mononuclear cell (PBMC; 90 Gy). Alternatively, T cells were stimulated with 0.5 μg/mL anti-CD3 mAb and 0.25 μg/mL anti-CD28 mAb. T-cell proliferation was assessed at day 4 by [³H]-thymidine incorporation. Relative suppressor activity is presented as mean ± SD (proliferation of Teff + Treg normalized to proliferation of Teff = 100%).

For blocking of IFNAR2, Treg were preincubated with either 5 μg/mL mouse IgG2a isotype control (BD Pharmingen) or blocking anti-IFNAR-2 mAb (MMHAR-2; PBL InterferonSource) overnight before subjection to suppressor assays.

For inhibition of PDEs, the PDE inhibitor 3-isobutyl-1-xanthine (IBMX, Sigma-Aldrich; 25 μmol/L) or the PDE4-specific inhibitor rolipram (Cayman Chemicals; 5 μmol/L), respectively, were added.

Teff and Treg were incubated overnight with or without 10⁴ U/mL IFN-α, subsequently stimulated with 0.5 μg/mL anti-CD3 mAb and 0.25 μg/mL anti-CD28 mAb for 6 hours, and cAMP was quantified using a cAMP-specific ELISA (Direct cAMP ELISA, 900-066, AssayDesign). In some experiments, T cells were incubated with the ERK1/2-specific inhibitor (5 μmol/L; FR180204, Calbiochem) or the PDE4-specific inhibitor rolipram (5 μmol/L; Cayman Chemicals), respectively, 1 hour before IFN-α treatment.

Human NK cells were separated using the CD56⁺CD16⁺ NK cell isolation kit (Miltenyi Biotec), resulting in a purity of NK cells between 90% and 95% (Supplementary Fig. S1). Teff and Treg were cultured for 2 hours at 37°C with or without 10⁴ U/mL IFN-α in the presence of 0.5 μg/mL anti-human CD3 (clone OKT-3) and 1 μg/mL anti-human CD28 (BD Pharmingen). After washing, Teff or Treg were cocultured with isolated NK cells (T cell/NK cell ratio: 2:1) for 4 hours. Subsequently, equal numbers of CFSE (1 μmol/L)-labeled K562 tumor cells (16) and carboxyfluorescein succinimidyl ester (CFSE) (0.1 μmol/L)-labeled PBMC (from the NK cell donor) were added (NK/K562 ratio 10:1) to the cultured cells for further 4 hours at 37°C. Afterwards, cells were subjected to flow cytometry (BD FACSCalibur), data were analyzed with Summit v4.0 (Dako), and killing efficiency was determined as the ratio of propidium iodide-negative CFSE^low control cells (syngeneic PBMC) to CFSE^high K562 target cells normalized to background cell death.

Genomic DNA was isolated from 5 to 6 × 10⁵ T cells. Male donors were used to avoid possible artefacts due to random x chromosome inactivation in females. Bisulfite treatment of genomic DNA was conducted using the EpiTect Bisulfite Kit (Qiagen). The methylation status of the FOXP3 Treg-specific demethylated region (TSDR) was analyzed by bisulfite pyrosequencing, which was conducted on a PSQ96MA Pyrosequencing System (Biotage) with the PyroGold SQA reagent kit (Biotage; ref. 17). PCR and sequencing primers for bisulfite pyrosequencing were designed using the Pyrosequencing Assay Design Software (Biotage) to cover 8 CpG Sites in the FOXP3 TSDR: forward 5′-TTGTATTTGGGTTTTGTTGTTATAGTT-3′, biotinylated reverse 5′-AAATATCTACCCTCTTCTCTTCC-3′, sequencing 5′-ATAGTTTTAGATTTGTTTAGATTTT-3′. Pyro Q-CpG software (Biotage) was used for pyrosequencing data analysis.

Treg were incubated overnight with or without 10⁴ IU IFN-α/mL. For graft-versus-host disease (GvHD) induction, newborn Rag2^−/−γc^−/− mice were injected intraperitoneally with 5 × 10⁶ human PBMC, which resulted in a total mortality of more than 95% of all treated animals, with a median time to death of 40 days (range, 35–55 days). Development of xenogeneic GvHD was characterized by decelerated growth and weight loss and was accompanied by massive inflammatory infiltration of human T cells into all murine organs (15). Cotransfer of resting human Treg at a 4:1 ratio (PBMC/Treg) decelerated GvHD onset and increased the median time to death by 10 days (range, 5–15 days) (15). Untreated mice served as controls.

Statistical significances of differences between experimental groups were evaluated using the paired or unpaired Student t test, the one-way and two-way ANOVA and Tukey Multiple Comparison test, and the GraphPadPrism 5 software package (Graphpad). P values of 0.05 or less were considered significant. Survival distributions were analyzed with the log-rank (Mantel–Cox) test and GraphPad Prism 5.

To determine whether IFN-α influences the suppressive activity of human Treg, suppressor assays with cocultures of human CD4⁺CD25⁻ Teff and Treg were set up in the presence of irradiated costimulatory PBMC with or without IFN-α. In this setting, Treg inhibited Teff proliferation up to 88%, dependent on the Treg/Teff ratio (Fig. 1A and B). Addition of IFN-α caused a considerable loss in Treg-mediated suppression independent of Treg/Teff ratios (Fig. 1A and B).

In mice, IFN-α modulates the suppressive activity of Treg by affecting the function of antigen-presenting cells (APC; ref. 18). We therefore compared the effect of IFN-α on human Treg function in the presence and absence of APC (after stimulation with anti-CD3/anti-CD28 mAb). However, in contrast to the murine system, IFN-α abrogated the suppressive function of human Treg both in the presence (Fig. 1A and B) and absence of APC (Fig. 1C), showing APC independency of the IFN-α–mediated Treg inactivation.

Treg promote tolerance to tumors in mouse models and Treg infiltration in human tumors is associated with worse clinical outcomes (19, 20). Previously, it was shown that preactivated Treg efficiently suppress the cytotoxic activity of NK cells against tumor cells (Fig. 2; ref. 21). We confirmed these results using a CFSE-based system of NK-mediated cytotoxicity against K562 tumor cells. Here, activated Treg were capable of abrogating tumor cell lysis by NK cells (Fig. 2 and Supplementary Fig. S1). Intriguingly, preincubation of Treg with IFN-α restored NK tumor killing in the presence of Treg showing that IFN-α overcomes Treg-mediated immune privilege for tumor cells. In contrast with Treg, conventional CD4⁺ T cells slightly increased NK tumor killing and were not affected by IFN-α pretreatment (data not shown).

Analysis of IFNAR-2 expression revealed that Treg express slightly higher levels of the receptor compared with Teff (Fig. 3A; control T cells without IFN-α: gray lines and bars). In line, preincubation of Treg with anti-IFNAR-2 mAb significantly counteracted IFN-α–mediated Treg inactivation as compared with treatment with control mAb (Fig. 3B). However, incubation with IFN-α also induced a significant reduction in IFNAR-2 expression of both, Teff and Treg (Fig. 3A, black lines and bars: + IFN-α), at least suggesting that IFN-α limits its effect by receptor downregulation in a self-controlling process.

Binding of IFN-α to the IFNAR complex initiates a signaling cascade comprising phosphorylation and dimerization of STAT molecules followed by their translocation to the nucleus, where they regulate the expression of IFN-stimulated genes (22). Comparison of STAT3 and STAT5 phosphorylation in Treg and Teff in response to IFN-α revealed no differences (Supplementary Fig. S2). However, these data are supposedly in line with previous results showing preferential cAMP-mediated STAT1 inhibition in T cells (23). IFN-α–mediated STAT1 phosphorylation was significantly reduced in Treg compared with Teff (Supplementary Fig. S2). The altered STAT1 activation in Treg was not regulated by upstream Tyk2 phosphorylation (data not shown).

It has been previously suggested that plasmacytoid DCs impair the regulatory function of Treg by inducing Treg apoptosis through IFN-α (24). To determine whether IFN-α induces apoptosis in Treg, Teff and Treg were incubated with IFN-α, and the frequency of viable or dead cells was subsequently evaluated by Trypan blue (data not shown) or Annexin-V staining. Apoptosis rates of Teff versus Teff + IFN-α (15.3 ± 8.9 vs. 13.6 ± 8.3) and of Treg versus Treg + IFN-α (25.6 ± 9.7 vs. 25.9 ± 6.7) did not significantly differ, excluding increased cell death as a cause for IFN-α diminished suppressor activity.

A couple of Treg-associated surface molecules seem to be important for function, activation, and homeostasis of Treg populations (9, 10, 12). However, expression of Treg differentiation molecules CTLA-4, CD127, CD39, CD62L, CD147, ICOS, IL1-R, TNFR2, CCR7, HLA-DR, and CD45RO/RA remained unaffected by IFN-α treatment (data not shown).

As previously reported, IL-10 contributes to the regulatory activity of Treg (10) and its production is increased in CD4⁺ T cells after IFN-α treatment (25). IFN-α neither affected IL-10 nor IFN-γ production of Treg (Fig. 4A). In Treg/Teff cocultures, both IFN-γ and IL-10 levels were slightly increased by IFN-α reflecting reduced Treg activity (Fig. 4A), whereas no alterations in IL-2 and TGF-β amounts were observed (data not shown).

It is widely accepted that the suppressive function of Treg is linked to their anergic state, associated with low IL-2 production in vitro (9, 10). However, IFN-α did not induce Treg proliferation as assessed by [³H] thymidine incorporation (Fig. 1A and B) and by flow cytometry of Treg in suppressor assays (data not shown). Thus, abrogation of Treg suppressive activity by IFN-α does not involve reversion of Treg anergy.

Compelling evidence suggests that the transcription factor Foxp3 acts as a master switch governing the development and function of Treg (9, 10). Recently, an evolutionary conserved CpG-rich element within the FOXP3 locus was found to be demethylated in Treg (26). This TSDR was associated with transcriptional activity and a stable imprinted phenotype of Treg (26). We therefore additionally analyzed methylation of the TSDR in Treg in response to IFN-α treatment. Teff with high methylation state of the TSDR served as controls (Fig. 4B). However, bisulfite pyrosequencing of the TSDR revealed unchanged methylation levels, both in resting and activated Treg upon IFN-α incubation compared with untreated control Treg (Fig. 4B), ruling out epigenetic modulation of Foxp3 as a target of IFN-α–induced Treg inactivation. These data are in line with our finding that IFN-α–mediated Treg inactivation, including cAMP repression, did not result in an alteration of the Foxp3 expression in Treg (Supplementary Fig. S3).

As reported recently, upregulation of endogenous cAMP is required for the suppressor function of murine and human Treg (27–29). Consistent with previous reports, assessment of cytosolic cAMP concentrations revealed that Teff produced substantially lower amounts of cytosolic cAMP compared with Treg (Fig. 5A; ref. 27). IFN-α–mediated functional impairment of human Treg was associated with significantly repressed cAMP upregulation in activated Treg, corroborating the previously identified cAMP dependency of Treg-suppressive function (Fig. 5A).

Intracellular cAMP levels can be regulated by alterations in the rate of synthesis by adenylate cyclases, degradation by PDEs, or excretion. Human T cells predominantly express the PDE isoforms, PDE4B and PDE4D (30). Blocking PDE activity by the PDE inhibitor IBMX or by the PDE4-specific inhibitor rolipram, respectively, abrogated IFN-α–induced cAMP deprivation and restored the regulatory activity signifying a role of cAMP degradation in IFN-α–mediated Treg inhibition (Fig. 5B and C). Furthermore, specific inhibition of PDE4 activity in IFN-α–treated Treg resulted in recovery of cAMP levels, confirming PDE4-mediated cAMP degradation as key mechanism in IFN-α–mediated Treg inactivation (Fig. 5D).

PDE4 isoforms are functionally regulated by the ERK2 MAP kinase, and activation of the MEK/ERK pathway in turn belongs to the prominent actions of IFN-α in human CD4⁺ T cells (22, 31). We therefore additionally examined whether IFN-α–induced cAMP deprivation in Treg is affected by an ERK inhibitor. Blockade of the ERK signaling pathway prevented IFN-α–induced cAMP degradation in Treg (Fig. 5E), confirming ERK-driven PDE4 activation as a molecular mechanism in IFN-α–mediated cAMP downregulation in Treg.

Because Treg upregulate their cAMP content in response to T-cell receptor (TCR) engagement (27–29) and cAMP has been shown to participate in regulation of TCR-mediated signaling (32, 33), we analyzed the impact of IFN-α on Src tyrosine kinase p56lck-mediated ZAP-70 phosphorylation at the Tyr³¹⁹ residue as a direct TCR-related downstream signaling event. Importantly, preincubation of Treg with IFN-α resulted in decreased ZAP-70 activation in response to TCR stimulation (Fig. 6), revealing modulation of TCR signaling in human Treg by IFN-α. However, reduction of the cAMP level in Treg by an adenylate cyclase inhibitor (MDL-12) did not affect ZAP-70 phosphorylation, suggesting that the IFN-α–mediated modulation of TCR-mediated signaling in Treg may not be directly linked to the cAMP level (Supplementary Fig. S4; ref. 15).

To investigate the effect of IFN-α treatment on Treg-suppressive activity in vivo, we resorted to a previously established xenogeneic GvHD model induced by human T cells in immunodeficient mice that is applicable for the functional analysis of human Treg (15, 34). Upon intraperitoneal engraftment with human PBMC, newborn Rag2^−/−γc^−/− mice developed lethal GvHD as assessed by loss of body weight (Fig. 7A) and death (Fig. 7B) within 2 months (Fig. 7B). In this model, preincubation of Treg with IFN-α abrogated their ability to protect from disease onset as compared with untreated Treg (Fig. 7A and B), without affecting their viability (Fig. 7C), signifying a regulatory role of IFN-α in Treg-suppressive activity in vivo. Human Treg survived only for limited time in vivo, regardless of their pretreatment with IFN-α (Fig. 7C, right). However, in agreement with previous reports showing that suppression by Treg is confined to the very early phase of GvHD (35, 36), limited survival of human Treg in vivo did not affect their suppressive activity in GvHD (Fig. 7). As compared with Treg-injected animals, the xenoreactive population of human effector T cells in IFN-α-Treg–treated mice expanded over time and ultimately induced GvHD (Fig. 7C, left).

IFN-α exhibits significant effects in promoting tumorgenicity through activation of immune cells and repression of tumor growth. Herein, we show that IFN-α significantly inactivates the suppressive function of human Treg and thereby releases CD4⁺ T cells and NK cells from Treg-mediated suppression. Thus, abrogation of Treg activity by IFN-α adds a novel and conceivable explanation for the immune-promoting effect of IFN-α in malignancies (3–5), including frequently observed autoimmune symptoms upon continuous therapeutic application (6, 7). In line with this assertion, IFN-α treatment has been previously observed to reduce Treg numbers in patients with melanoma or renal cell carcinoma (37, 38), whereas, conversely, declining IFN-α levels and impaired IFN-α signaling in patients with cancer seem to be correlated with an increase in the number of Treg and a higher risk of cancer progression (39–41). Recent reports on improved immune responses upon co-administration of IFN-α with peptide vaccines in renal (42), pancreatic (43), and colorectal cancer (44) represent additional evidence.

Numerous investigations have firmly established a key role of cAMP in the suppressive function of Treg (27–30, 45–47). Importantly, within the CD4⁺ T-cell population, Treg exclusively accumulate cAMP. By interfering with the function of cAMP-producing adenylate cyclases or cAMP-degrading PDEs, both the suppressive activity and functional state of Treg can be selectively manipulated (15, 27, 29). It is currently not clear whether differential cAMP upregulation in human Treg results from differential enzyme endowment or activity; however, it is at least known that human T cells predominantly express the short isoforms PDE4B and PDE4D, which are functionally regulated by ERK2 MAP kinase, and that engagement of the TCR leads to recruitment of a β-arrestin/PDE4 complex to lipid rafts that serves to degrade the local, TCR-induced production of cAMP (30, 31, 48). Activation of the MEK/ERK pathway again belongs to the prominent actions of IFN-α in human CD4⁺ T cells (22). In our study, both blockade of the MAK kinase ERK1/2-regulated pathway and specific inhibition of PDE4 activity completely restored the suppressor activity and accumulation of cAMP in IFN-α–treated Treg. Thus, we show here that IFN-α turns down the cAMP level in Treg by stimulation of the MEK/ERK pathway (22) leading to ERK-mediated PDE4 activation. This observation is in line with the previous finding that both, ERK inhibition and blockade of cAMP degradation by PDE inhibitors, enhances the Treg-suppressive function (29, 49).

IFN-α signals primarily through the Janus-activated kinase (JAK)/STAT pathway (22). Comparing IFN-α–mediated STAT signaling in Treg and Teff, we observed similar activation of STAT3 and STAT5 but significantly lower IFN-α–induced STAT1 activation in Treg. The reason for altered STAT1 phosphorylation in Treg is unclear, however, may be consistent with a study reporting that cAMP differentially reduces STAT1 activation in T cells (23).

In T cells, the IFN-α receptor not only recruits the JAK–STAT pathway, but also additionally associates with signaling molecules of the TCR signalosome curtailing their availability in TCR signal transduction (50). Investigating ZAP-70 phosphorylation in place of TCR signaling in Treg, we observed partly decreased ZAP-70 phosphorylation in response to TCR stimulation in presence of IFN-α. However, decreased ZAP-70 phosphorylation did not seem to be directly related to the cAMP level because pharmacologic repression of the second messenger by adenylate cyclase blockade did not lead to an altered ZAP-70 activity.

It is widely accepted that the suppressive function of Treg is linked to their anergic state, expression of Treg function-associated surface molecules, and production of immunosuppressive cytokines like IL-10 (9, 10). Our study revealed that IFN-α did not affect the anergic state, the cytokine profile, or expression of surface molecules of human Treg. In addition, the methylation state of the TSDR within the FOXP3 locus, which is associated with transcriptional activity and a stable imprinted phenotype of Treg, remained unaltered by IFN-α. Thus, our results indicate that IFN-α deactivates the suppressive function of Treg without interfering with their lineage program.

In conclusion, repression of Treg activity by IFN-α may significantly contribute to its antineoplastic effects in cancer and provide a rational basis to consider IFN-α as a potent Treg-inactivating drug in the context of therapeutic vaccination against tumor antigens.

No potential conflicts of interest were disclosed.

Conception and design: C. Becker, K. Steinbrink

Development of methodology: T. Bopp, U. Zechner, C. Becker, K. Steinbrink

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Hofmann, E. Graulich, M. Schwenk, T. Bopp, U. Zechner, L. Merten, K. Steinbrink

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Bacher, V. Raker, C. Becker, K. Steinbrink

Writing, review, and/or revision of the manuscript: N. Bacher, V. Raker, U. Zechner, C. Becker, K. Steinbrink

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): R. Baumgrass, U. Zechner, C. Becker, K. Steinbrink

Study supervision: K. Steinbrink

The authors thank P. Hoelter for expert technical assistance and E. von Stebut and H. Jonuleit for critical reading of the manuscript.

This work was supported by the German Research Foundation (DFG German Research Foundation: Transregio 52/A7, STE791/6-1) initiative CRC 1066/B6 (K. Steinbrink), by the initiative CRC 1066/B8 (C. Becker and T. Bopp), by the German Cancer Aid (110631, K. Steinbrink), and by intramural grants to K. Steinbrink, C. Becker, and N. Bacher (MAIFOR, NMFZ).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.