Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
This report was presented at The Journal of Physiology Symposium on TRP channels: physiological genomics and proteomics, San Diego, CA, USA, 5 April 2005. It was commissioned by the Editorial Board and reflects the views of the authors.
Corresponding author M. Freichel: Experimentelle und Klinische Pharmakologie und Toxikologie, Universität des Saarlandes, D 66421 Homburg, Germany. Email: firstname.lastname@example.org
Available data on transient receptor potential channel (TRPC) protein functions indicate that these proteins represent essential constituents of agonist-activated and phospholipase C-dependent cation entry pathways in primary cells which contribute to the elevation of cytosolic Ca2+. In addition, a striking number of biological functions have already been assigned to the various TRPC proteins, including mechanosensing activity (TRPC1), chemotropic axon guidance (TRPC1 and TRPC3), pheromone sensing and the regulation of sexual and social behaviour (TRPC2), endothelial-dependent regulation of vascular tone, endothelial permeability and neurotransmitter release (TRPC4), axonal growth (TRPC5), modulation of smooth muscle tone in blood vessels and lung and regulation of podocyte structure and function in the kidney (TRPC6). The lack of compounds which specifically block or activate TRPC proteins impairs the analysis of TRPC function in primary cells. We therefore concentrate in this contribution on (i) studies of TRPC-deficient mouse lines, (ii) data obtained by gene-silencing approaches using antisense oligonucleotides or RNA interference, (iii) expression experiments employing dominant negative TRPC constructs, and (iv) recent data correlating mutations of TRPC genes associated with human disease.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
TRP channels may couple plasma membrane receptors and the activation of voltage-dependent channels
Stimulation of cells with extracellular agonists leads to an increase of [Ca2+]i both from Ca2+ release from intracellular organelles and Ca2+ entry across the plasma membrane (Fig. 1). This process is initiated upon the binding of a hormone or a growth factor to its receptor and activation of phospholipase C (PLC). PLC in turn converts phosphatidylinositol-4,5-bisphosphate (PIP2) into inositol 1,4,5-trisphosphate (IP3) and diacylglycerol (DAG). Plasma membrane channels, including TRPC channels, are activated and extracellular cations, including Ca2+, flow into the cell. The activation of plasma membrane channels might involve a reduction of the PIP2 concentration or an increase of DAG or its metabolites such as polyunsaturated fatty acids (Hardie, 2003). By reduction of PIP2 tonic PIP2-dependent inhibition might be abolished whereas DAG or polyunsaturated fatty acids might activate TRPC channels directly. In addition IP3 will bind to its receptor in the endoplasmic reticulum membrane which results in the release of intracellular Ca2+. This process might be accompanied by the opening of Ca2+ release-activated or store-operated channels in the plasma membrane which allow additional cations, including Ca2+, to enter the cell. When [Ca2+]i reaches a threshold level Ca2+-activated non-selective (CAN) channels will be activated. The TRPC-related TRPM4 and TRPM5 proteins might underlie CAN channel activity and allow Na+ to enter the cell. Thereby, the cell is depolarized and voltage-dependent Ca2+ channels do open. Ca2+ influx mediated by voltage-gated Ca2+ channels then further contributes to the increase of [Ca2+]i and the spatio-temporally controlled elevations of [Ca2+]i in defined cell compartments as key signals for numerous cellular functions. Available data support the engagement of this signalling cascade in T-lymphocytes (Badou et al. 2005) and, probably, in certain smooth muscle cells. Concerning TRPC channels there is evidence that they can be activated by DAG, by direct interaction with the Ca2+ release channels of the endoplasmatic reticulum, the IP3 receptor and the ryanodine receptors, by translocation to the plasma membrane via an exocytotic process and by store-operated mechanisms (Parekh & Putney, 2005). Whereas the molecular composition of the TRPC channels in native tissues is still elusive the subunit structure of voltage-gated Ca2+ channels (Cav) has been known for years. They represent heteromultimeric protein complexes consisting of the pore-forming α1 subunit and the ancillary subunits β, γ, α2 and δ, which modulate the expression, targeting, gating and activity of the α1 subunits. Recent studies indicate that individual Cavβ subunits not only constitute a portion of the Ca2+ channel but also regulate intracellular IP3 levels (Berggren et al. 2004) and gene transcription (Hibino et al. 2003).
TRPCs are members of the TRP family of cation channels and, to date, at least 28 TRP homologues have been identified in mammalian genomes. According to their primary structure, mammalian TRP proteins are classified into six subgroups, in addition to the TRPC subgroup, TRPV, TRPM, TRPP, TRPML and TRPA (Montell et al. 2002; Montell, 2005). The seven members of the TRPC subfamily (classical or canonical TRPs) are structurally related to Drosophila TRP and to each other (> 30% amino acid sequence identity within the amino-terminal 750–900 amino acid residues) and differ mainly within their carboxy-terminal regions. There is evidence that TRPC1, TRPC4 and TRPC5 or TRPC3, TRPC6 and TRPC7, respectively, do specifically interact and form heteromultimeric protein complexes (Strubing et al. 2001, 2003; Hofmann et al. 2002; Goel et al. 2002). Accordingly coexpression of TRPC1, TRPC4 and TRPC5 can be readily detected in hippocampal neurones (Fig. 1) although in most other tissues and cells expression patterns of the three genes differ substantially (Freichel et al. 2004).
Concerning the possible functions of TRP proteins there is ample evidence that phospholipase C-dependent processes are involved in TRPC channel gating (see above). However, the data available have been mainly obtained after overexpression of the TRPC cDNAs in common cell culture systems such as HEK 293 cells (human embryonic kidney cell line). These cells were introduced originally because they readily overexpress foreign proteins (Pritchett et al. 1988; Flockerzi et al. 2005). Moreover mRNA isolated from HEK 293 cells have been used as a source to clone the full length cDNAs of the human TRPC1 (Zhu et al. 1995) and TRPC3 (Zhu et al. 1996). Accordingly it is not unlikely to assume that endogenous TRPC1 and TRPC3 proteins are already expressed in these cells at decent concentrations. These proteins can build endogenous homo- and heteromultimeric TRPC1, TRPC3 and TRPC1/TRPC3 channels which may contribute to currents recorded from these cells. It is not difficult to foresee that the expression of additional foreign TRPC proteins in these HEK cells will further aggravate the interpretation of whole-cell current recordings and to define the contribution of the same TRPC protein to these currents. Unfortunately, reliable anti-TRPC antibodies which might allow the monitoring of endogenous TRPC protein expression are scarce (Ong et al. 2002; Flockerzi et al. 2005).
TRPC-deficient mouse models and gene silencing of TRPC genes in primary cells
Few methods overcome the restraints of overexpression systems and the lack of compounds known to block or to activate TRPC-mediated cation entry into cells with sufficient potency and specificity. To characterize the functional relevance of individual TRPC proteins in their native environment for cation entry into cells, for isolated organ physiology and for complex systemic functions in a whole organism we and others therefore used (i) targeted gene inactivation in embryonic stem cells to generate mice carrying null alleles of TRPC genes, (ii) silencing of TRPC transcripts in primary isolated cells using antisense oligonucleotides or RNA interference, or (iii) expression of dominant negative TRPC variants. In addition, for the first time spontaneous mutations within one TRPC gene were linked to a hereditary human disease. Based on segregation studies in several families and direct identification of missense mutations in affected individuals two groups described independently an essential role for the TRPC6 gene in glomerular slit diaphragm-associated channel activity in the kidney.
TRPC1 TRPC1 is expressed in brain, heart, kidney, endothelium, ovaries, testis and in smooth muscle cells (Freichel et al. 2004; Inoue et al. 2004; Montell, 2005). The mode of activation of TRPC1-mediated cation entry is still under discussion but may accompany PLC-dependent processes and depletion of intracellular Ca2+ stores (Zitt et al. 1996; Montell, 2005). So far no results have been published on TRPC1-deficient mice. However, the TRPC1 gene was inactivated by somatic cell gene targeting in the DT-40 B lymphocyte cell line (Mori et al. 2002). These DT-40 cells lack expression of TRPC1 transcripts and protein as shown by Northern blot and immunocytochemical studies. Stimulation of the B cell receptor or application of thapsigargin leads to an elevation of [Ca2+]i and this elevation was significantly impaired in the TRPC1-deficient cells. In addition, targeting of the TRPC1 gene resulted in a reduction of the occurrence of Ca2+ currents activated by dialysis of these cells with IP3 by 80%, of the number of Ca2+ oscillations evoked upon B cell receptor stimulation, and of the nuclear factor of activated T cell (NF-AT)-dependent gene expression (Mori et al. 2002). In another study, antisense oligonucleotides directed against TRPC1 mRNA were injected into endothelial cells isolated from human pulmonary artery. Upon stimulation with thapsigargin the injected cells responded with a smaller elevation of the intracellular Ca2+ concentration compared to non-injected cells. The extent of reduction of TRPC1 protein in the antisense injected endothelial cells has not been shown in this study. (Brough et al. 2001). To study TRPC1 function in neurones, Kim et al. (2003) generated a mutated TRPC1 protein by replacing the phenylalanine residue at position 561 of the primary structure by an alanine residue (TRPC1/F561A). In contrast to wild-type TRPC1, this mutated TRPC1 did not induce significant inward currents when overexpressed in the Chinese hamster ovary cell line (CHO cells) constitutively expressing the metabotropic glutamate receptor type 1 (mGluR1). Transfection of the TRPC1/F561A mutant into cerebellar Purkinje cells resulted in a 49% reduction of mGluR-evoked slow excitatory postsynaptic currents (EPSCs) whereas fast transmission mediated by AMPA-type glutamate receptors remained unaffected. The mGluR1 receptor is coupled to PLC via Gq and the results obtained are in line with PLC-mediated gating of TRPC1. Recently, TRPC1 has been implicated as a component of a mechanosensitive cation channel in Xenopus oocytes, which transduces membrane stretch into cation flux across the cell membrane (Maroto et al. 2005). However, in primary cells from mammals there is so far no evidence for any TRPC being activated by mechanical stimuli. Singh et al. (2001) introduced TRPC1 into the salivary gland of rats by adenovirus-mediated gene transfer. They demonstrated that after TRPC1 overexpression saliva secretion increased upon stimulation with a muscarinic receptor agonist. In parallel the muscarinic agonist-induced Ca2+ entry was increased in cells isolated from infected salivary glands (Singh et al. 2001).
TRPC2 TRPC2 is a pseudogene in humans containing six mutations that generate premature stop codons (Liman & Innan, 2003). In rodents, a functional TRPC2 gene is expressed in the vomeronasal organ (VNO) and possibly in the testis (Vannier et al. 1999; Jungnickel et al. 2001). The VNO neurones and their associated projections represent a set of olfactory neurones which is anatomically distinct to the main olfactory epithelium. The VNO is found in most terrestrial vertebrates but is likely to be vestigial in humans. In rodents, the VNO gains access to water-soluble chemical cues carried by the nasal mucus and sends fibres to the accessory olfactory bulb, which in turn projects to discrete loci of the ventromedial hypothalamus via the mediocortical amygdala. The VNO pathway was previously suggested to play a primary role in the detection of non-volatile pheromones and in eliciting behavioural and physiological responses to conspecifics and thereby in mating and intraspecies aggressive behaviour. Results from two independently generated mouse lines (Stowers et al. 2002; Leypold, 2002) show a loss of both sex discrimination and male–male aggression in TRPC2-deficient males. Additionally, lactating females are docile and fail to initiate aggressive attacks on intruder males, and TRPC2−/− males vigorously mount other males. TRPC2−/− males mate normally with females indicating that male–female sexual behaviour appears normal. TRPC2-deficient males are fertile since matings with TRPC2−/− males produced normal litter sizes (Stowers et al. 2002; Leypold et al. 2002) although the acrosome reaction of murine sperms was significantly inhibited in the presence of an anti-TRPC2 antibody in vitro (Jungnickel et al. 2001). VNO neurones isolated from TRPC2−/− mice do not show the specific increase in spiking rate upon application of urine pheromones, although they are electrically active under basal conditions. The transduction mechanisms of mammalian pheromone signalling by VNO neurones involves phospholipase C (PLC) activation, and recently, the search for ion conductances activated by pheromone signals that are regulated by PLC activity lead to the identification of a DAG-activated cation channel in the dendritic tips of these neurones. This DAG-activated cation channel is defective in TRPC2−/− VNO neurones indicating that in mice TRPC2 proteins are principal subunits of this DAG-gated channel (Lucas et al. 2003).
TRPC3 TRPC3 RNA has been shown to be enriched in the adult human brain (Zhu et al. 1996) but another study showed that in rats the TRPC3 protein is predominantly expressed in the brain during a relatively narrow developmental period before and after birth (Montell, 2005). Several modes of activation of TRPC3-mediated Ca2+ entry were reported including direct activation by DAG analogues (Hofmann et al. 1999; Montell, 2005). Very recently, TRPC3 has been shown to be involved in the development of the highly specialized neuronal network and movements of axonal growth cones. The chemotropic axon guidance is determined by the sensing of extracellular cues that are translated into intracellular signals including activation of PLC and elevation of the intracellular Ca2+ concentration mediated at least in part through the activation of agonist-activated cation channels but also of voltage-gated Ca2+ channels (Henley & Poo, 2004). Li et al. showed that the rise of [Ca2+]i and growth cone turning induced by brain-derived neurotrophic factor (BDNF) was abolished by transfection of rat cerebellar granule cells with short interfering RNAs (siRNAs) directed against TRPC3 mRNA as well as dominant negative TRPC3 and TRPC6 constructs (Li et al. 2005). In contrast, Ca2+ elevations or chemoattraction of axonal growth cones were not altered by siRNA-based strategies aimed at impairing TRPC1 expression. However, RNAi-mediated gene silencing of the TRPC1 homologue in Xenopus TRP-1 (also referred to as xTRPC1) leads to an almost 70% reduction of xTRPC1 protein expression in spinal neurones from Xenopus embryos, and both growth cone turning and elevation of [Ca2+]i following stimulation with the agonist netrin-1 were abolished (Wang & Poo, 2005).
Based on results from RT-PCR studies TRPC3 might also be expressed in smooth muscle cells and endothelial cells from blood vessels (Freichel et al. 1999; Inoue et al. 2001, 2004; Beech et al. 2004). In line with TRPC3 expression in vascular smooth muscle cells treatment of rat cerebral arteries with oligonucleotides directed against TRPC3 mRNA reduces the depolarization of smooth muscle cells and the degree of constriction of isolated vessel rings by 61% and 37% upon application of 10−6m and 10−5m UTP, respectively, whereas myogenic responses were not affected (Reading et al. 2005).
TRPC4 TRPC4 is expressed in brain, endothelium, kidney, retina, testis and adrenal gland and the mode of activation may differ in these cell systems (Freichel et al. 2004; Montell, 2005). The endothelium that lines the inner side of blood vessels plays a critical role in the preservation of normal vessel functions. Agonist-activated Ca2+ entry is involved in many of these functions, and transcripts of several TRP genes, including TRPC1, TRPC2, TRPC3, TRPC4, TRPC6, TRPV4, TRPM4 and TRPM7, were readily detected in primary isolated endothelial cells from mouse aorta (MAECs) (Freichel et al. 1999, 2001, 2004; Wissenbach et al. 2000; Nilius et al. 2003). Based on experiments in isolated cells from wild-type and TRPC4−/− mice, TRPC4 proteins were identified as indispensable components of store-operated Ca2+ channels in MAECs acting either as channel-forming subunits or as constituents of the channel complexes that are essential for channel activation (Freichel et al. 2001). Additionally, agonist-activated Ca2+ entry into TRPC4−/− endothelial cells is drastically decreased leading to a markedly impaired regulation of vascular tone. In addition, following stimulation with thrombin, which activates G protein-coupled proteinase-activated receptor-1 (PAR-1), the Ca2+ influx in endothelial cells isolated from the lung of neonatal TRPC4−/− mice was also markedly reduced (Tiruppathi et al. 2002). Thrombin-induced Ca2+ entry is a critical determinant of endothelial permeability and, hence, thrombin-induced decreases of transendothelial resistance of cultured endothelial monolayers as a measure of endothelial cell retraction efficiency were significantly smaller in TRPC4−/− confluent monolayers than in wild-type controls. Similar results were observed when changes of microvascular permeability was studied by perfusing the intact circulation of isolated lungs. In these ex vivo preparations the agonist-activated permeability was reduced by about 50% (Tiruppathi et al. 2002). These results clearly support the conclusion that TRPC4 is, or is part of, a Ca2+ entry pathway in macrovascular endothelium such as in MAECs. The expression of various other TRP genes additional to TRPC4 in MAECs and other types of endothelial cells raises the possibility that Ca2+-entry channels in these cells may be formed by TRP heteromers (Strubing et al. 2001, 2003; Hofmann et al. 2002; Goel et al. 2002).
Within the central nervous system, a functional role for TRPC4 proteins was recently revealed within unique types of synaptic terminals of the thalamic network, the so-called F2 terminals. F2 terminals derive from dendrites of local GABAergic interneurones and are situated in a triadic synaptic arrangement within a glomerular neuropil of the dorsal lateral geniculate nucleus (Pape et al. 2004). The release of γ-aminobutyric acid (GABA) from F2 terminals is modulated by extrathalamic input systems acting on metabotropic receptors and it was found that this release occurs in a Ca2+-dependent manner. In TRPC4−/− mice, the 5-hydroxytryptamine (5-HT)-induced increase of GABA release from the thalamic interneurones is largely reduced. In contrast, the increase of GABA release upon stimulation with an agonist of metabotropic glutamate receptors or inhibition of GABA release by acetyl-β-methylcholine is unchanged, indicating that F2 terminals are still functional in TRPC4−/− mice (Munsch et al. 2003). The regulation of the GABAergic component within these local circuit interactions by extrathalamic input systems may be critical for the gating of signals through the thalamus depending on the sleep–wake cycle. Therefore, by regulating F2-mediated signals through these systems, TRPC4 might contribute to the state-dependent processing of visual information, such as visual contrast sensitivity, or might be operative as a state-dependent gain control of retinogeniculate transmission in general (Pape et al. 2004).
In addition to these thalamic interneurones, TRPC4 is expressed in a variety of other cells within the central nervous system, such as in the olfactory bulb, septal nuclei, hippocampus, cortex and cerebellum. Additionally, TRPC4 transcripts were readily identified in uterus, ovary and kidney by Northern blot analysis and could be detected in pancreatic islets using RT-PCR. Results of glucose tolerance tests do not differ between wild-type and TRPC4-deficient mice and ongoing studies are aimed at revealing the possible functions of TRPC4 in pancreatic islets.
TRPC5 The predominant site of expression of TRPC5 is the central nervous system (Freichel et al. 2004; Montell, 2005). Here, TRPC5 transcripts (Fig. 2) and proteins (Greka et al. 2003) are abundant in hippocampal neurones. To address the function of TRPC5 in a native system, a TRPC5 mutant (DN-TRPC5) harbouring an exchange of three amino acids between the fifth and sixth predicted transmembrane domain was expressed in cultured hippocampal neurones. Co-transfection of this construct together with wild-type TRPC5 was reported to abolish carbachol-induced whole-cell currents that are obtained after TRPC5 cDNA transfection alone. Transfection of DN-TRPC5 in dissociated hippocampal neurones induced an increased length of their filopodia, the finger-like projections characteristic of growth cones. Furthermore, the neurite length and also the filopodia perimeter were increased under these conditions. Based on experiments in the same cellular system it was recently concluded that the translocation of TRPC5 proteins from intracellular vesicles to the plasma membrane, which was shown to depend on the activity of phosphatidylinositol 4-phosphate 5-kinase (PIP(5)Kα), is a key step for TRPC5-mediated regulation of neurite extension rates (Bezzerides et al. 2004). Based on these results TRPC5 and potentially the TRPC5-mediated Ca2+ entry can be regarded as an important determinant of axonal growth and growth cone morphology. Notably, no results for a TRPC5-related role in growth cone turning upon chemotropic stimulation were reported although this process was shown to be related to the expression of other TRPC proteins such as TRPC3 and TRPC6 or TRPC1 depending on the neuronal system analysed. Apparently, the available results obtained from studies in isolated neurones may indicate that TRPC proteins show a specific subcellular localization pattern and are assigned to distinct functional tasks within separate compartments of a particular cell type.
TRPC6 TRPC6 proteins are expressed in the lung, the brain (Montell, 2005) and in smooth muscle cells, and their expression levels were reported to correlate with cell proliferation of pulmonary artery smooth muscle cells (PASMCs) (Yu et al. 2004). Furthermore, expression of TRPC6 was shown to be increased in lung tissues and PASMCs from patients with idiopathic pulmonary hypertension and, hence, siRNA-mediated knockdown of TRPC6 expression markedly attenuated proliferation of PASMCs from these patients (Yu et al. 2004).
Recently, initial results were presented that describe the phenotype of TRPC6-deficient mice obtained by targeting the TRPC6 gene. In TRPC6−/− mice, a higher contractility in isolated tracheal and aortic rings was observed following application of the agonists methacholine and phenylephrine, respectively, as well as a reduced agonist-induced expiration rate (Dietrich et al. 2003). The density of inward currents that were induced by perfusion of the cells with the DAG derivate 1-stearoyl-2-arachidonoyl-sn-glycerol (SAG) and that were carried to a large extent by Na+, were significantly higher in cerebrovascular artery smooth muscle cells isolated from TRPC6−/− mice. The increased current densities were also observed in non-stimulated cells indicating a higher basal and agonist-induced cation entry in TRPC6−/− smooth muscle cells (Mederos y Schnitzler et al. 2004). Accordingly, an increase in the responsiveness of peripheral resistance arteries to vasoconstrictors and in the mean arterial blood pressure was described, and the intravascular pressures necessary to induce myogenic constriction due to the Bayliss effect were shifted to the left towards lower values (Dietrich et al. 2004). The observed increased smooth muscle tone both under basal conditions and after agonist stimulation might be due to the formation of TRPC3 homo-oligomeric channel complexes in TRPC6-deficient smooth muscle cells, because mRNA expression of TRPC3 appears to be up-regulated two- to threefold in TRPC6−/− mice.
In an earlier study, TRPC6 expression was knocked down using an antisense oligodesoxynucleotide approach in cultured rat cerebral arteries. In this study, vasoconstriction induced by elevating intravascular pressure was inhibited by 70% to 80% in TRPC6 antisense-treated arteries compared with control arteries which were treated with sense oligonucleotides (Welsh et al. 2002). Differences in the results from the different approaches might be due to the varying degrees of reduction of TRPC6 protein expression, which was in the range of 60% (Welsh et al. 2002) in the knock-down approach. Antisense oligonucleotides were also used to down-regulate TRPC6 expression in cultured rabbit portal vein myocytes (Inoue et al. 2001). Here, the density of a cation current activated by the α-adrenoceptor agonist phenylephrine was markedly decreased after treatment with the TRPC6 antisense oligonucleotide compared with cells treated with the control TRPC6 sense oligonucleotides.
Expression of the TRPC6 protein in glomerular podocytes was demonstrated recently by immunostaining. These podocytes represent a central component of the renal filtration barrier (Pavenstadt et al. 2003). Furthermore, the TRPC6 gene was mapped as a candidate gene for focal and segmental glomerulosclerosis (FSGS), a hereditary kidney disorder leading to proteinuria and renal failure (Winn et al. 2005; Reiser et al. 2005). Sequencing of the TRPC6 gene of FSGS-affected individuals yielded six missense mutations in independent families; these mutations were located in the amino- or carboxyterminal parts of the protein. After expression in cell culture systems, some but not all of the TRPC6 mutants induced inward currents with larger densities than those induced by non-mutant TRPC6. Apparently the causative mutation of FSGS occurs in the TRPC6 gene raising the possibility that TRPC6 may be a useful therapeutic target in chronic kidney disease.
The use of gene ablation and gene silencing approaches has already enabled tremendous progress towards the identification of functional roles for TRPC proteins in their native environment, in primary cell systems, in isolated organs and whole organisms. It is apparent that TRPCs like other members of the TRP superfamily of cation channels play decisive physiological roles and if mutated could lead to human disease.
The concepts demonstrated by the genetic approaches summarized above will certainly further encourage the efforts to screen for potent and specific agonists and/or antagonists of these TRPC cation channels which might be useful not only as experimental pharmacological tools but also as potential therapeutics.
We thank Stefanie Buchholz and Kerstin Fischer for their excellent technical assistance, and P. Wollenberg for help during preparation of the figures. This work was supported in part by the Deutsche Forschungsgemeinschaft and HOMFOR.