The Cytoplasmic Tyrosine Motifs in Full-Length Glycoprotein 130 Have Different Roles in IL-6 Signal Transduction¹

Interleukin-6 was first described as a B cell-stimulating factor (1) which induces the differentiation and proliferation of B and T cells. Furthermore, IL-6 is the major mediator of acute phase protein (APP)³ expression in rat hepatocytes in vitro (2, 3) and in rats in vivo (4). IL-6 signals through the glycoprotein 130 (gp130)/Janus kinase (Jak)/STAT pathway (5, 6). Upon ligand binding to the α-receptor subunit gp80, the signal-transducing receptor subunit gp130 dimerizes, and activation of constitutively gp130-associated tyrosine kinases of the Jak family Jak1, Jak2, and Tyk2 occurs. Subsequently, the receptor is tyrosine phosphorylated, and STAT factors (5, 6) and the Src homology domain 2 protein tyrosine phosphatase (SHP2) (7) are recruited via their Src homology 2 domains (8, 9), resulting in STAT and SHP2 phosphorylation on tyrosine residues. The tyrosine-phosphorylated STAT factors homo- or heterodimerize (10) and translocate to the nucleus, where they bind to enhancer elements of respective IL-6 target genes (11). In contrast, SHP2 acts negatively on acute phase protein induction (12, 13, 14) and is involved in the activation of the mitogen-activated protein kinase (MAPK) pathway by IL-6 (15).

Previous work of Stahl et al. (7) and Gerhartz et al. (16) has shown that the four distal tyrosine modules of gp130 are involved in STAT activation. Both groups have used in their experiments the individual tyrosine modules linked to the box 1 and 2 region of gp130. It was found that the last four tyrosine modules stimulate STAT3 phosphorylation, whereas STAT1 was activated by the two distal tyrosine modules. The consensus sequences YXPQ and YXXQ were proposed for STAT1 (16) and STAT3 (7, 16) activation, respectively. Furthermore, Tyr⁷⁵⁹ was identified to be crucial for the phosphorylation of the tyrosine phosphatase SHP2 (7).

However, the contributions of the individual tyrosines within the context of the whole cytoplasmic tail of gp130 in IL-6 signaling have not been studied yet. The purpose of this work was to determine whether the entire cytoplasmic part of gp130 affects the potential of the different tyrosine residues to activate IL-6 signal transduction. In the present paper, we show that the tyrosine residues 767, 814, 905, and 915 are not equivalent with respect to the induction of STAT3 tyrosine phosphorylation, DNA binding, reporter gene activation, and stimulation of proliferation when investigated in the context of the full-length gp130 receptor protein. Furthermore, our data are contrary to the observation previously published by Fukada et al. (15) who showed, using a cytoplasmically truncated G-CSF/gp130 chimeric receptor, that a single STAT recruitment site in gp130 (Tyr⁷⁶⁷) is not sufficient for IL-6-mediated proliferation in gp130-transfected Ba/F3 cells. From our results, it must be concluded that the full-length cytoplasmic part of gp130 modulates the functionality of the individual tyrosine motifs in IL-6 signal transduction.

Restriction enzymes were purchased from Boehringer Mannheim (Mannheim, Germany) and AGS (Heidelberg, Germany), oligonucleotides were synthesized by MWG-Biotech (Ebersberg, Germany). Vent polymerase and Abs to the Tyr⁷⁰⁵-phosphorylated STAT3 were obtained from New England BioLabs (Beverly, MA). Abs to gp130 were gifts from Dr. J. Wijdenes (Besançon, France). Recombinant erythropoietin (Epo) was a generous gift of Drs. J. Burg and K. H. Sellinger (Boehringer Mannheim). Recombinant IL-6 and sIL-6R were prepared as described earlier (17, 18). The specific activity of IL-6 was 2 × 10⁶ B cell-stimulatory factor-2 U/mg of protein.

Constructions were conducted by standard procedures (19). pGL3α2 M-215Luc contains the promoter region −215 to +8 of the rat α₂-macroglobulin (α2M) gene fused to the luciferase-encoding sequence and was described previously (12). pGL3-hACT-359LUC was cloned by inserting the promoter region (−379 to +25) of the human α₁-antichymotrypsin into pGL3 (Promega, Madison, WI) and kindly provided by F. Horn (Leipzig, Germany).

The expression vector pSVL-EG encoding the chimeric Epo/gp130 receptor (16) was modified by PCR mutagenesis to introduce recognition sites for restriction enzymes between the cytoplasmic tyrosines (pSVL-EG(YYYYYY) (tyrosine residues in the wild-type cytoplasmic part of gp130 are boldfaced). These mutations were designed not to change the encoding amino acid sequence. The introduced restriction sites were the following: SacI between Tyr⁶⁸³ and Tyr⁷⁵⁹; BspEI between Tyr⁷⁵⁹ and Tyr⁷⁶⁷; AspI between Tyr⁷⁶⁷ and Tyr⁸¹⁴; Bsp119I between Tyr⁸¹⁴ and Tyr⁹⁰⁵; and RsrII between Tyr⁹⁰⁵ and Tyr⁹¹⁵. pSVL-EG(FFFFFF) was constructed similarly to pSVL-EG(YYYYYY) and contains additional Y→F substitutions in the six cytoplasmic tyrosine motifs.

These two vectors were used to construct add-back mutants by combining appropriate DNA fragments of the cytoplasmic part of gp130, resulting in pSVL-EG(YFFFFF), pSVL-EG(FYFFFF), pSVL-EG(FFYFFF), pSVL-EG(FFFYFF), pSVL-EG(FFFFYF), and pSVL-EG(FFFFFY). Also add-back mutants containing two tyrosines were prepared: pSVL-EG(YYFFFF); pSVL-EG(FYYFFF); pSVL-EG(FYFYFF); pSVL-EG(FYFFYF); and pSVL-EG(FYFFFY).

The DNA fragments encoding the Epo/gp130 chimeric receptors mutants were also transferred into pRc/CMV-EG (16) and used for expression in HepG2 cells (e.g., pRc/CMV-EG(YFFFFF)).

The DNA fragment coding for the transmembrane and cytoplasmic domains of the chimeric receptor mutants was also introduced into pSVL-gp130 (20) (e.g., pSVL-gp130 (YFFFFF)). The latter constructs were used for the generation of stably transfected Ba/F3 cells.

The sequences of all constructs have been verified by fluorescence sequencing.

Human hepatoma HepG2 cells were cultivated and transient transfections by the calcium phosphate coprecipitation method were performed as described previously (21). Cell lysis and luciferase assays were conducted with the Promega luciferase kit according to the manufacturer’s instructions. Three independent transfections have been performed. Luciferase activity values were normalized to transfection efficiency monitored by the cotransfected β-galactosidase expression vector (pCR3lacZ, Pharmacia, Uppsala, Sweden). The error bars are SD.

Ba/F3 cells were cultivated and stably transfected as described previously (20). Equal expression levels of gp130 protein were verified by FACS analysis with the B-P4 Ab specific for the extracellular domain of gp130 (20, 22). For each receptor mutant, two independent stable transfections were performed and found to give similar results in the experiments executed (data not shown).

Nuclear extracts of Ba/F3 cells were prepared, protein concentration was measured by the Bio-Rad protein assay, and EMSA was performed as previously described (20). We used a STAT1- and STAT3-specific double-stranded ³²P-labeled probe: a mutated serum-inducible element (SIE) oligonucleotide of the c-fos promoter (m67 SIE: 5′-GATCCGGGAGGGATTTACGGGAAATGCTG-3′). Protein-DNA complexes were separated on 4.5% polyacrylamide gels containing 7.5% glycerol in 0.25× Tris-borated EDTA buffer at 20 V/cm for 4 h. Gels were fixed in 10% methanol, 10% acetic acid, and 80% water for 30 min; dried; and autoradiographed.

Equal amounts of nuclear extracts were separated by SDS-PAGE (7% gel) and transferred to a polyvinylidene fluoride membrane. Ags were detected by incubation with the appropriate primary Ab (1:1000) and HRP-coupled secondary Abs (1:1000) (Dako, Hamburg, Germany). The membranes were developed with an enhanced chemiluminescence kit (Amersham-Buchler, Braunschweig, Germany). To verify application of equal amounts of protein, blots were stripped and reprobed.

Ba/F3 cells (2 × 10⁴) expressing gp130 receptor mutants were cultured in DMEM containing the indicated concentrations of IL-6 and 1 μg/ml soluble (s) IL-6R or conditioned medium from X63Ag-653 BPV-mIL-3 myeloma cells as a source of IL-3. After 72 h of incubation, viable and metabolically active cells were quantified by using the colorimetric Cell Proliferation Kit II (XTT) (Boehringer Mannheim) as described by the manufacturer. Values represent the average of three independent experiments ± SD.

A crucial step in the activation of SHP2 and STAT factors in IL-6 signal transduction is their recruitment to phosphotyrosine residues of the signal transducer gp130. To examine the role of the individual tyrosine motifs in the context of the full-length gp130, a cloning strategy has been established that allowed us to generate all the possible gp130 mutants with Y→F substitutions in its cytoplasmic part. Therefore, two gp130 DNA constructs have been generated, containing either all six cytoplasmic tyrosine codons or a substitution of all tyrosine to phenylalanine codons. Additionally, these two constructs contain new unique restriction sites, which do not alter the amino acid sequence of the receptors, between all tyrosine or phenylalanine positions. These restriction sites enabled us to exchange single or combinations of tyrosine motifs of gp130 between both receptor mutants (for details, see Materials and Methods).

The gp130 mutants used in this work contain either all, none, or only a single tyrosine (add back mutants) of the six cytoplasmic tyrosine residues present in the wild-type human gp130 receptor protein (Fig. 1,A). All constructs were stably transfected into IL-3-dependent Ba/F3 cells, which do not express endogenous gp130 (23). Receptor surface expression of the transfected Ba/F3 cells was monitored by FACS analysis with the use of an Ab raised against the extracellular domain of gp130 (22) and found to be comparable for all transfectants (Fig. 1 B).

To compare the capabilities of the single tyrosine motifs with respect to STAT activation, the stably transfected Ba/F3 cells, expressing the add-back mutants of gp130 (Fig. 1), were stimulated with IL-6/sIL-6R complexes. Nuclear extracts from these cells were analyzed for DNA-binding activities in an EMSA with DNA probes providing binding sites for STAT3 and STAT1 (Fig. 2). No DNA-binding activity was found in nuclear extracts from untransfected Ba/F3 cells. Stimulation of cells expressing the wild-type gp130 receptor protein (gp130(YYYYYY)) led to the formation of protein-DNA complexes containing STAT3 homodimers and to a lower extent STAT1 homodimers and STAT1/STAT3 heterodimers. No protein-DNA complex formation was detected after stimulation of Ba/F3 cells transfected with the receptor mutant gp130(FFFFFF).

Stimulation of stably transfected Ba/F3 cells expressing the six add-back receptor mutants with only one cytoplasmic tyrosine residue showed remarkable differences in STAT activation. The add-back mutants containing the membrane-proximal tyrosines Tyr⁶⁸³ or Tyr⁷⁵⁹ were not able to mediate protein-DNA complex formation. In contrast, an induction of STAT-DNA binding could be achieved by all four individual distal tyrosines (Tyr⁷⁶⁷, Tyr⁸¹⁴, Tyr⁹⁰⁵, Tyr⁹¹⁵) in the full-length gp130 receptor protein. Interestingly, induction of STAT-DNA binding via gp130(FFFFYF) or gp130(FFFFFY) was more pronounced than through gp130(FFYFFF) or gp130(FFFYFF).

A time course experiment was elaborated to investigate the kinetics of STAT3 activation via the individual tyrosines in the context of the whole gp130 protein. Ba/F3 cells stably expressing the gp130 receptor mutants described in Fig. 1 were stimulated with IL-6/sIL-6R complexes up to 60 min and analyzed for STAT3-DNA binding in EMSAs (Fig. 3 A). Stimulation of Ba/F3 cells expressing the wild-type gp130 receptor resulted in maximal STAT3-DNA binding between 20 and 30 min after stimulation. The add-back receptor mutants (gp130(FFYFFF), gp130(FFFYFF), gp130(FFFFYF), and gp130(FFFFFY)) which were found to mediate STAT3-DNA binding showed a weaker stimulation than gp130(YYYYYY) but similar kinetics.

DNA binding of STAT factors requires their tyrosine phosphorylation leading to subsequent homo- or heterodimerization. Tyrosine phosphorylation of STAT3 activated via the add-back gp130 receptor mutants expressed in Ba/F3 cells was analyzed by Western blotting performed with an Ab specific to the Tyr⁷⁰⁵-phosphorylated STAT3 (Fig. 3 B). The capability of the individual receptor mutants to exert tyrosine phosphorylation coincides with their capability to mediate DNA binding of STAT3. A strong STAT3 tyrosine phosphorylation was found in Ba/F3 cells expressing the wild-type receptor. Ba/F3 cells expressing the receptor mutants gp130(FFYFFF) or gp130(FFFYFF) showed a weaker tyrosine phosphorylation of STAT3 after stimulation than cells expressing gp130(FFFFYF) or gp130(FFFFFY). These experiments clearly demonstrate that the four distal tyrosines (Tyr⁷⁶⁷, Tyr⁸¹⁴, Tyr⁹⁰⁵, and Tyr⁹¹⁵) in the full-length gp130 receptor protein differ in their potency to activate STAT3.

STAT3 is the most important transcription factor for the induction of the APP genes (24). It exerts its action by binding to the enhancer sequences of APP gene promoters (11). In reporter assays the capability of single tyrosines in the cytoplasmic part of gp130 to mediate activation of APP gene promoters in the human hepatoma cell line HepG2 was analyzed. Chimeric receptors containing the extracellular domain of the EpoR and the transmembrane and the cytoplasmic domains of the various gp130 mutants (Fig. 1,A) allowed us to induce IL-6 signal transduction in HepG2 cells independent from endogenous gp130. Vectors encoding these chimeric receptor mutants were cotransfected with a construct harboring the promoter of the α2M gene linked to the luciferase reporter gene (pGL3α2M-215Luc) (Fig. 4 A).

Stimulation of cells expressing chimeric Epo/gp130 receptors containing the cytoplasmic domain of wild-type gp130 (EG(YYYYYY)) with Epo led to an 18-fold induction of the reporter gene. The α2M gene promoter activation in cells expressing the EG(FFYFFF) or the EG(FFFYFF) receptor was slightly reduced. The add-back mutants that were unable to mediate STAT3 DNA-binding activity also failed to induce the reporter gene. Surprisingly, reporter gene activation via the receptors containing only Tyr⁹⁰⁵ or Tyr⁹¹⁵ led to a 40-fold induction.

In addition, similar reporter gene assays were performed using the promoter of the APP gene α₁-antichymotrypsin linked to the reporter gene luciferase (Fig. 4,B). The results were very similar to those obtained with the α2M promoter reporter gene construct (Fig. 4 A). We conclude from these data that the add-back mutants containing one of the four distal tyrosines differ in their potency to activate APP gene promoters.

To elucidate the influence of the SHP2 recruitment site in gp130 on STAT-mediated gene induction, chimeric receptor add-back mutants containing Tyr⁷⁵⁹ plus one of the four membrane-distal cytoplasmic tyrosine residues (Fig. 5,A) were also analyzed in the reporter gene assay with the α2M gene promoter/luciferase construct (Fig. 5,B). Epo stimulation of HepG2 cells expressing the chimeric receptors EG(FYFFYF) or EG(FYFFFY) led essentially to the same luciferase induction compared with the wild-type receptor (EG(YYYYYY)) indicating an influence of Tyr⁷⁵⁹ on the gene induction through the distal tyrosine motifs (compare EG(FFFFYF) in Fig. 4,A with EG(FYFFYF) in Fig. 5,B or EG(FFFFFY) in Fig. 4,A with EG(FYFFFY) in Fig. 5 B). Activation of receptors containing no STAT-binding sites did not result in significant reporter gene induction. Interestingly, the α2M promoter-luciferase reporter activation in cells expressing the receptors EG(FYYFFF) or EG(FYFYFF) was strongly reduced when compared with the EG(YYYYYY) receptor demonstrating the importance of the most distal tyrosine motifs in gp130.

Proliferation of transfected Ba/F3 cells expressing gp130 can be induced by stimulation with IL-6/sIL-6R (23). Up to now there are only scarce data regarding the role of individual tyrosines in the full-length gp130 for IL-6-induced proliferation. Therefore, we investigated the proliferation of the Ba/F3 transfectants expressing the gp130 add-back mutants (Fig. 1). The cells were stimulated with increasing amounts of IL-6/sIL-6R complexes for 72 h. Subsequently, the viable and metabolically active cells were quantified (Fig. 6,A). No proliferation of untransfected cells was observed. Cells expressing a gp130 mutant lacking all cytoplasmic tyrosine residues (gp130(FFFFFF)) and cells expressing the gp130 mutants gp130(YFFFFF) or gp130(FYFFFF) did not proliferate even at the highest concentration of IL-6/sIL-6R. In contrast, a dose-dependent proliferative response was observed for gp130(YYYYYY) transfectants (Fig. 6,A, top). The stably transfected Ba/F3 cells expressing the gp130 add-back mutants containing only one of the four distal tyrosines, which were found to mediate STAT3 activation (Fig. 2), show also an IL-6 dose-dependent growth stimulation. In several independent experiments, we found that the receptors gp130(FFFYFF) and gp130(FFFFFY) are weaker signal transducers with respect to the IL-6/sIL-6R-induced proliferation (Fig. 6,A, bottom) than the wild-type gp130 receptor protein (Fig. 6,A, top) indicating that the level of STAT3 activation did not correspond to the proliferative activity. For control, cells of the different Ba/F3 cell lines were incubated with a conditioned medium of IL-3-expressing cells (Fig. 6 B). No differences were found for the stably transfected Ba/F3 cells with regard to their IL-3-dependent proliferative response.

IL-6 exerts all its activities by activating the IL-6-receptor complex composed of the IL-6Rα subunit gp80 and the signal-transducing subunit gp130. The membrane-proximal boxes 1 and 2 in gp130 have been identified to bind the Jaks (23, 25). Thus far, the binding sites of STAT1 and STAT3 on gp130 were studied only by using isolated tyrosine motifs of gp130 fused to a carboxyl-terminal deletion mutant of gp130 containing box 1 and box 2. With these fusion proteins, it has been demonstrated that receptors containing a YXXQ motif are capable of activating STAT3 (7, 16), whereas a proline in this consensus sequence at position +2 (YXPQ) is required for STAT1 activation (16). Recently, we confirmed these results in an in vitro binding assay with isolated STAT3-Src homology 2 domains (26). A mutation of Tyr⁷⁵⁹ to phenylalanine was demonstrated to impair SHP2 phosphorylation (7, 12).

In addition to these data on the modular organization of gp130, information about the tertiary structure of the cytoplasmic tail is not available. Undoubtedly, the three-dimensional structure of the cytoplasmic part of the gp130 molecule may also influence the activation of the signaling components such as Jaks, SHP2, and STATs. Generation of add-back mutants allowed us to examine the potential of the six individual cytoplasmic tyrosine motifs in the context of the full-length receptor (Fig. 1). Indeed, we found that these receptor mutants are not equivalent with regard to STAT activation, acute phase protein induction, and stimulation of cell proliferation. We confirmed that STAT1 and STAT3 are activated only via the four most distal tyrosine motifs (Tyr⁷⁶⁷, Tyr⁸¹⁴, Tyr⁹⁰⁵, and Tyr⁹¹⁵) (Fig. 2). Interestingly however, add-back receptor mutants with only one of the two most distal tyrosines (Tyr⁹⁰⁵ or Tyr⁹¹⁵) were more potent to mediate STAT activation than the receptors containing the proximal tyrosines. These differences were not simply due to altered kinetics of STAT activation (Fig. 3). Therefore, we propose that the individual tyrosine motifs of gp130 are not equivalent in their capacity to activate STAT factors in the context of the full-length cytoplasmic part of gp130.

Thus far, these differences in the potential of the individual STAT recruitment sites of gp130 have not yet been taken into consideration. To examine, whether they affect biological activities of IL-6, such as acute phase protein expression in liver cells, we analyzed the potential of the add-back mutants to activate promoters of the two APP genes α2M and α₁-antichymotrypsin. As expected, we found that only receptor mutants that activate STAT factors are able to mediate APP gene induction. Similar to the pattern of STAT activation, the add-back receptors containing Tyr⁹⁰⁵ or Tyr⁹¹⁵ were the most efficient ones to activate APP gene promoters. Surprisingly, these two add-back mutants, containing only a single cytoplasmic tyrosine, were much more potent than the wild-type receptor subunit containing all six tyrosine motifs. This implies an inhibitory function on APP gene induction through one of the four membrane-proximal cytoplasmic tyrosine residues. We were able to localize the inhibitory tyrosine at position 759: the addition of Tyr⁷⁵⁹ in the add-back mutants reduced the efficiency of the individual receptors to mediate α2M promoter activation. None of these add-back mutants containing Tyr⁷⁵⁹ plus one of the other five tyrosines was more potent in α2M promoter activation than the receptor containing all six tyrosine motifs. This is consistent with the observation that SHP2, which binds to Tyr⁷⁵⁹, negatively regulates STAT activation and APP gene induction (12, 13, 14). Thus, SHP2 does not affect individual tyrosine motifs in gp130 but acts in a more general manner. Recently, it has been published that activation of the MAPK pathway results in a reduction of STAT phosphorylation (27), which is in line with the fact that SHP2 phosphorylation after IL-6 stimulation leads to the induction of the MAPK pathway (15).

In addition to STAT and APP promoter activation, we examined the potency of all gp130 add-back mutants with respect to the proliferation of Ba/F3 cells. Ba/F3 cells proliferate in response to IL-3 or after transfection of gp130 in response to IL-6/sIL-6R (23). We found that all receptors that mediate STAT activation also stimulate proliferation. However, the strength of STAT and APP gene promoter induction via the individual add-back receptor mutants did not correlate with their ability to stimulate proliferation. Ba/F3 cells expressing receptor mutants containing only a single STAT recruitment site are less sensitive to IL-6 (Fig. 6,A, bottom) than cells expressing the wild-type receptor (Fig. 6 A, top). This finding implies that STAT activation is not the only prerequisite for maximal IL-6-induced Ba/F3 cell proliferation. The lower efficiency of the receptor mutants in mediating proliferation might reflect the presence of only one STAT recruitment site or the lack of a protein activated by the wild-type receptor but not by these mutants. None of the two proximal tyrosine motifs is essential for proliferation, because receptor mutants with a Y→F substitutions at amino acid positions Tyr⁶⁸³ and Tyr⁷⁵⁹ still mediate IL-6-induced Ba/F3 cell proliferation. Nevertheless, we cannot exclude an influence of these tyrosines on the IL-6-induced proliferation of gp130-transfected Ba/F3 cells.

Our observation that a gp130 add-back mutant containing a single STAT-binding tyrosine is sufficient for IL-6 induced proliferation is in contrast to the data of Fukada et al. (15). Their studies have been performed with chimeric receptors composed of the extracellular domain of the G-CSF-R, the transmembrane domain of gp130, and a truncated cytoplasmic part of gp130 lacking the three distal tyrosine motifs. Thus, their receptor contains only one STAT recruitment site (Tyr⁷⁶⁷), the SHP2 recruitment site (Tyr⁷⁵⁹) and Tyr⁶⁸³. The requirement of both STAT3 and SHP2 for Ba/F3 cell proliferation was derived from the results with Y→F substitutions within the truncated part of gp130. Only receptors containing the STAT as well as the SHP2-recruiting tyrosine were able to mediate cell proliferation. In summary, the authors found STAT3 to inhibit apoptosis and SHP2 to exert a mitogenic signal. The requirement of Tyr⁷⁵⁹ in gp130 for proliferation was confirmed in Ba/F3 cells expressing a G-CSF-R/gp130 chimeric receptor mutant, which contains the full-length cytoplasmic part of gp130 but a mutation of the SHP2 recruitment site (15). These cells do not proliferate but survive in response to stimulation with G-CSF.

Fukada et al. (15) supposed that the deleted region, including the three distal STAT-recruitment sites, in their chimeric G-CSF-R/gp130 receptors is negligible for IL-6 signal transduction. In contrast, we performed our experiments with a full-length gp130 receptor with point mutations of the individual cytoplasmic tyrosine residues. Thus, the mutation of the individual tyrosines in gp130 compared with a deletion of the whole C-terminal region led to different biological responses. One can assume that the differences in the cytoplasmic parts of the receptor mutants used in both studies cause the different requirements for SHP2 with regard to stimulation of proliferation.

In conclusion, we suggest that the role of the entire cytoplasmic part of gp130 must be taken into account, and the data obtained with truncated chimeric receptor mutants should be considered with caution and be confirmed by the use of the respective full-length native receptors.

We thank Iris Behrmann and Lutz Graeve for critical reading of the manuscript and Sonja Linnemann for excellent technical assistance.