Association of the NADPH:protochlorophyllide oxidoreductase (POR) with isolated etioplast inner membranes from wheat


For correspondence (fax +46 35 167 500; e-mail


Membrane association of NADPH:protochlorophyllide oxidoreductase (POR, EC: with isolated prolamellar bodies (PLBs) and prothylakoids (PTs) from wheat etioplasts was investigated. In vitro-expressed radiolabelled POR, with or without transit peptide, was used to characterize membrane association conditions. Proper association of POR with PLBs and PTs did not require the presequence, whereas NADPH and hydrolysable ATP were vital for the process. After treating the membranes with thermolysin, sodium hydroxide or carbonate, a firm attachment of the POR protein to the membrane was found. Although the PLBs and PTs differ significantly in their relative amount of POR in vivo, no major differences in POR association capacity could be observed between the two membrane systems when exogenous NADPH was added. Experiments run with only an endogenous NADPH source almost abolished association of POR with both PLBs and PTs. In addition, POR protein carrying a mutation in the putative nucleotide-binding site (ALA06) was unable to bind to the inner membranes in the presence of NADPH, which further demonstrates that the co-factor is essential for proper membrane association. POR protein carrying a mutation in the substrate-binding site (ALA24) showed less binding to the membranes as compared to the wild type. The results presented here introduce studies of a novel area of protein–membrane interaction, namely the association of proteins with a paracrystalline membrane structure, the PLB.


Plants grown in the absence of light lack plastids with photosynthetically competent thylakoid membranes, and thus also lack several of the components required for light capture and charge separation. Plastid development in light involves the differentiation of proplastids or etioplasts to chloroplasts. Etiolated angiosperms posses photoreceptors and catalytic proteins that, upon illumination, rapidly signal the re-organization of the plastid from a non-photosynthetic state to a photosynthetically active chloroplast. The differentiation of etioplasts to chloroplasts involves the transformation of the regular prolamellar bodies (PLBs) into lamellar membrane structures, synthesis of chlorophyll from the precursor protochlorophyllide (Pchlide), and a massive accumulation of chloroplast proteins. The majority of the chloroplast proteins are encoded by nuclear genes, synthesized in the cytosol and imported into the plastid. The enzyme responsible for the light-dependent step in chlorophyll biosynthesis is the NADPH:protochlorophyllide oxidoreductase (POR; Griffiths, 1978). This is a nuclear-encoded protein which, together with Pchlide and NADPH, constitutes a minimal functional unit. Aggregates of such units, connected to the formation of PLBs, can be detected spectroscopically due to their low-temperature (77 K) fluorescence at 657 nm (Pchlide657). Following light-induced transformation of Pchlide to chlorophyllide (Chlide), the large aggregates of the ternary Chlide–POR–NADPH complexes disaggregate into smaller entities.

Two different POR proteins (PORA and PORB) were demonstrated in barley by Holtorf et al. (1995) and in Arabidopsis by Armstrong et al. (1995), and recently a third isoform (PORC) was found in Arabidopsis (Oosawa et al., 2000). The PORA protein is present in dark-grown plants, but is inactivated and degraded during greening (Reinbothe et al., 1995a). The PORB protein is also present in dark-grown plants, but persists during greening in light. The discovery of three PORs, one light-labile (PORA), one light-stable (PORB) and one light-inducible (PORC), could partly explain the paradox of varying levels of POR during massive chlorophyll synthesis in greening etiolated plants (Holtorf et al., 1995; Runge et al., 1996).

Nuclear-encoded proteins destined for the plastid interior, such as the inner membranes, are generally transported by a default pathway over the envelope to the stroma and then subsequently translocated to their final suborganelle location. Import of PORA has been claimed to require strictly plastid-localized Pchlide, and thus to deviate from the common import mechanism (Reinbothe et al., 1995b). Under certain conditions it has been shown that, concomitant with decreasing amounts of Pchlide, etioplasts rapidly lost their ability efficiently to import PORA during greening. On the other hand, import of PORB was equally efficient in chloroplasts and etioplasts (Reinbothe et al., 1997). The substrate requirement of PORA for import suggested that PORA binds Pchlide during import, and that this complex, possibly in the presence of NADPH, is transported to the PLB of etioplasts. However, recent data have not supported the requirement of Pchlide for import of pea POR (Aronsson et al., 2001) or barley PORA (Aronsson et al., 2000; Dahlin et al., 2000).

Although our biochemical and physical knowledge of POR has increased significantly during recent years, the exact nature of binding of the POR to the plastid inner membranes has not been satisfactorily defined. It is possible to discern two steps in the assembly pathway. First, an attachment of POR to the membrane; and second, a firmer and more protected physiological association of POR with the membrane. After attachment POR is not protected from thermolysin-catalysed breakdown, but after the physiological membrane association, or binding, thermolysin causes only limited proteolysis of POR. Given the very strong membrane association of POR without a typical membrane-spanning region in the amino acid sequence, the final membrane-association reactions of POR with the etioplast inner membranes merits further investigation. We present here the first detailed analysis of the membrane-targeting mechanism of POR, leading to a physiological membrane association with isolated etioplast inner membranes of wheat.


Membrane association of radiolabelled POR with or without the transit peptide in plastid lysates

Purified etioplasts and chloroplasts from wheat were lysed in 10 HK buffer (see Experimental procedures). The lysates were used directly for membrane-association assays, or centrifuged further to obtain stroma-free inner membranes (EPIMs). Initially, the assays were performed in the presence of 10 mm MgATP and 1 mm NADPH with radiolabelled POR, either with (pPOR) or without (Δ-TP) the transit peptide. Thermolysin post-treatment of EPIMs, with or without stroma, resulted in a protease degradation product (POR-DP) of approximately 4 kDa smaller than the mature size of ≈36 kDa (Figure 1). Similar treatment of chloroplast inner membranes resulted in a POR-DP of approximately 34 kDa. This is the same approximate size of POR-DP as observed in green pea thylakoids under physiological conditions (Dahlin et al., 1995). We therefore conclude that the POR-DP also represents correctly assembled POR in EPIMs, but in this case the paracrystalline structure of the PLBs has a slight effect on the accessibility of the protease. The appearance of the 32–34 kDa protease product is not dependent on, nor inhibited by, the transit peptide. The absence of stroma caused no change on the membrane targeting of POR (Figure 1).

Figure 1.

Membrane association of radiolabelled POR with inner membranes of etioplasts (E) and chloroplasts (C).

The assays were performed in the presence of 1 mm NADPH with (+S) or without (–S) stroma. All samples included 10 mm MgATP and 20 mm MgCl2, and were treated with thermolysin after the assays, giving a POR-DP of approximately 32–34 kDa. pPOR, precursor POR protein containing the transit peptide; Δ-TP, mature POR protein without the transit peptide; molecular weight in kDa of POR with (44) or without (36) the transit peptide; TM, translation mixture. Left, pPOR; right, Δ-TP.

POR associates with both PLBs and PTs

Prolamellar body and prothylakoid (PT) fractions were isolated and characterized using SDS–PAGE and fluorescence spectroscopy. The protein composition of the membrane fractions revealed a qualitatively similar pattern, but with higher content of POR in PLBs compared to PTs (Figure 2, inset). Low-temperature fluorescence spectra of the isolated PLBs and PTs revealed major peaks at 657 and 633 nm, respectively (Figure 2). The strong dominance of fluorescence at 657 nm of the PLB fraction indicates highly purified PLBs with an intact regular structure. Together, these methods confirmed a proper separation of PLBs and PTs into highly purified lamellar PT and paracrystalline PLB membrane fractions (Lindsten et al., 1988). Fluorescence measurements were chosen for routine evaluation of the purity of PLB- and PT-enriched fractions before each experiment.

Figure 2.

Characterization of purified PLB and PT preparations from etiolated wheat.

Isolated membrane fractions were used for SDS–PAGE analysis (inset) or low-temperature (77 K) fluorescence emission spectra. Excitation wavelength 440 nm. Spectra were normalized by multiplying the PT spectrum 7.8 times to give the same maximum peak height. Arrow indicates position of POR.

When inner membrane fractions were used in the association assays together with radiolabelled pPOR or Δ-TP, we found that both membrane types were capable of binding (Figure 3). After thermolysin treatment, the purified PLB and PT membranes retained a protease-resistant form (POR-DP), similar to inner membranes in lysates. PLB, which in vivo contain most POR (Ryberg and Dehesh, 1986), also had a great capacity to bind POR in vitro in our experiments. Surprisingly, we also found an almost similar amount of thermolysin-resistant POR in PTs after the assay, even if the PTs in vivo contain only very low amounts of POR.

Figure 3.

Association of radiolabelled POR with or without the transit peptide with PLBs and PTs.

(a) Protein association assays with PLBs and PTs were conducted with the mature POR (Δ) lacking the transit peptide, or with the precursor POR (p) containing the transit peptide, in the presence of 1 mm NADPH, 10 mm MgATP, and 20 mm MgCl2. After assays, samples were treated with (+th) or without (–th) thermolysin.

(b) Fluorograms from membrane-association assays of the precursor pPOR protein containing the transit peptide. Assays were performed in the presence of 1 mm NADPH, 10 mm MgATP and 20 mm MgCl2. After assays, samples were treated with 50 mm NaOH (OH), 50 mm Na2CO3 (CO3), thermolysin (+th) or import buffer only (–th). Samples treated with thermolysin (+th) show a POR-DP with a molecular weight approximately 2–4 kDa lower than the mature POR of 36 kDa. Molecular weight in kDa of POR with (44) or without (36) the transit peptide. TM, translation mixture.

The nature of the membrane association of POR was assayed with NaOH and Na2CO3. Following binding reactions of pPOR to isolated PLBs and PTs, the samples were post-treated with sodium hydroxide (50 mm) or carbonate (50 mm). After this treatment of the samples we found pPOR still attached to the membrane, indicating that pPOR is tightly associated with the membrane (Figure 3b). As can be seen in Figure 3(b), a substantial amount of the pPOR used in the assay is unable to attain the mature size in the absence of the stroma processing peptidase.

Exogenous NADPH and ATP stimulate in vitro association with isolated PLBs and PTs

Analyses were performed to elucidate if NADPH is required for a physiological membrane–POR association, and to find out if there is a difference between PLBs and PTs in this respect. During the purification procedure of PLB and PT membranes, NADPH is present to stabilize the membranes and avoid a decay of the spectral properties (Lindsten et al., 1988; Ryberg and Sundqvist, 1988). In a concentration series of 0.25–3 mm NADPH, we found that the association of pPOR with PLBs and PTs occurred even at the lowest concentration of 0.25 mm NADPH added (data not shown). In Figure 4(a) the membrane association of POR is compared with and without the addition of 1 mm exogenous NADPH. POR attached both to PLBs and PTs when thermolysin was not present. However, after thermolysin post-treatment the POR-DP was found only in PLBs and not in PTs when no NADPH was added. Due to the structural differences between the two types of membrane, it was suspected that the difference could be due to a washing effect: NADPH could prevail longer in the three-dimensional network of PLBs than in the plain PTs. Repeated washing of PLBs was performed to verify if the binding was an intrinsic property of the membrane. Washing of PLBs with NADPH-free buffer abolished the thermolysin-resistant association in the PLBs after a single washing step, but had no effect on POR attachment even after repeated washing steps (Figure 4b).

Figure 4.

Membrane association of radiolabelled pPOR with PLBs and PTs requires NADPH and ATP.

(a) Assays were performed in the presence (+N) or absence (–N) of 1 mm NADPH.

(b) PLBs were prewashed repeatedly in NADPH-free buffer prior to the association assay. Numbers indicate sequential washing steps. Assays in (a) and (b) contained 10 mm MgATP and 20 mm MgCl2.

(c) Association assays were performed in the absence of ATP. Translation mixtures were filtered through Sephadex G-25 to remove excess nucleoside triphosphates. Import buffer containing 10 mm MgATP and 20 mm MgCl2 (+A) or 20 mm MgCl2 only (–A) was added prior to assembly reaction. Membranes were post-treated with (+th) or without (–th) thermolysin. Samples treated with thermolysin (+th) show a POR-DP with a molecular weight approximately 2–4 kDa lower than the mature POR of 36 kDa. pPOR, radiolabelled precursor POR containing the transit peptide. Molecular weight in kDa of POR with (44) or without (36) the transit peptide. TM, translation mixture.

To investigate the possible ATP requirement for the membrane-association reaction of POR, the assays were performed using samples with or without ATP. As seen in Figure 4(c), the addition of MgATP to ATP-depleted assembly mixtures clearly stimulated the proper association of POR with both PLBs and PTs.

Mutations in the substrate-binding and co-factor binding sites reduce membrane association

POR protein, altered in the amino acid sequence either in the predicted nucleotide-binding or active site (ALA06 and ALA24, respectively; cf. Dahlin et al., 1999), were used for membrane association studies with isolated PLBs and PTs. As seen in Figure 5(a), almost no labelling was found on the fluorograms following thermolysin post-treatment when a protein mutated in the NADPH-binding site was used. That is, replacement of charged amino acids to alanine in the NADPH-binding site abolished assembly. Charges in these regions of POR are of crucial importance for the association with the membrane. When the assay was performed with POR mutated in the active site, only small amounts of the POR-DP could be found (Figure 5b). Control experiments were performed with ALA06, ALA24 Δ-TP and pPOR proteins treated with thermolysin in the absence of membranes. The results showed that the mutated POR proteins under these conditions had the same protease resistance as the wild-type POR (data not shown). That is, those mutations per se in POR did not affect the protease susceptibility.

Figure 5.

Association of mutated radiolabelled POR protein with isolated PLBs and PTs.

POR protein carrying mutation in (a) the NADPH binding site (06) or (b) the predicted active site (24) compared to wild-type mature POR (Δ) lacking the transit peptide. Assays were performed in the presence of 1 mm NADPH, 10 mm MgATP and 20 mm MgCl2. Samples were post-treated with (+th) or without (–th) thermolysin. Samples treated with thermolysin (+th) show a POR-DP with a molecular weight approximately 2–4 kDa lower than the mature POR of 36 kDa. Molecular weight in kDa of mature POR (36) is given. TM, translation mixtures.


During recent years, large efforts have been made to elucidate the assembly mechanism of lumenal and integral thylakoid proteins into functional structures in chloroplasts. Current data suggest the existence of several specific pathways for the insertion or trans-thylakoid transfer of thylakoid proteins. Some of these can be traced back to mechanisms utilized in the export of bacterial proteins or co-translational transport of ER proteins. Our data on the assembly of pea POR to PLB- and PT-enriched fractions from wheat imply that POR does not utilize any of the pathways outlined for thylakoid proteins (cf. Keegstra and Cline, 1999; Robinson et al., 1998). No distinction between the assembly conditions of POR into thylakoids or inner etioplast membranes from wheat has been found. For instance, proper binding to both membrane systems includes an absolute requirement for NADPH, stimulation by ATP, but no requirement for stroma. Similar observations have been made during the association of pea pPOR with thylakoid membranes from light-grown pea (Dahlin et al., 1995).

In the present paper, we have used a heterologous system with POR having its origin in pea and the purified PLBs and PTs originating from wheat. The pea POR has characteristics of both PORA and PORB, and is therefore suitable for an exploratory study on etioplast inner membrane binding. However, PLBs and PTs are difficult to purify from pea, as the production of etiolated leaf material is low and methods for purification have not been developed. Furthermore, isolated pea PLBs have not been characterized, although it can be supposed that the major features are similar to those of wheat.

To our knowledge, this is the first time the association reaction of a protein has been reconstructed in vitro with an isolated and purified paracrystalline membrane structure such as the PLB. These results thus have a general importance, in addition to outlining the assembly of POR to etiolated inner membranes. It may be anticipated that the tubular structure of the PLB may pose a problem for detecting proper membrane association. However, the 32 kDa protease-degradation product of radiolabelled POR following thermolysin post-treatment (Figure 1) could also be observed when isolated etioplast inner membranes were treated with thermolysin (Lütz and Tönissen, 1984). In the same work, it was also shown that thermolysin has only a limited effect on the proper PLB structure.

POR has been shown to associate specifically with the plastid inner membranes and, for example, the envelope membranes of pea chloroplasts do not bind POR (Dahlin et al., 1995). This strongly indicates that the binding of POR to PLBs and PTs is a specific process. The membrane-association assays of POR with isolated PLBs and PTs led to several interesting observations. The POR expressed from cDNA clones from etiolated plants associated with inner membranes of chloroplasts (Figure 1; Dahlin et al., 1995) and etioplasts (Figure 1). This suggests that identification mechanisms (receptors) and/or other components of the assembly apparatus for POR may be functionally similar between etioplast inner membranes and thylakoid membranes of chloroplasts. Logically, this reflects the in vivo situation, which requires that POR is present in inner membranes of both etioplasts and chloroplasts (Sundqvist and Dahlin, 1997).

The basis for the formation of PLBs in etioplasts is not well understood. PLBs contain POR as the dominating protein, but we do not know if the association of POR occurs directly with growing PLBs or takes place primarily in PTs, which may then transform into PLBs. A comparison can be made between the attachment of POR to EPIMs, and the integration or thylakoid transfer of several other chloroplast proteins. Insertion of LHCP (Yalovsky et al., 1990) and transfer across the thylakoid membranes of OE33 (Hashimoto et al., 1997) and newly synthesized D1 protein (Mattoo and Edelman, 1987) occur primarily in the single stroma lamellae. The proteins thereafter translocate to the granal region. Alternatively, an enrichment, for example LHCP molecules, may induce the formation of stacked regions. A similar mechanism may operate during the development of proplastids to etioplasts in dark-grown seedlings. Proplastids in dark- or light-grown plants contain invaginations from the inner envelope and a few lamellar membranes, but generally lack PLBs (Whatley, 1977). Following transport of POR into the proplastids, the protein associates with the inner lamellar membranes. A massive accumulation of POR, together with a relatively high content of MGDG (Leese and Leech, 1976; Ryberg et al., 1983), stimulates the formation of PLBs (Murakami and Ikeuchi, 1982). The inner membranes then efficiently incorporate POR and the small PLBs grow into a large PLB structure.

The physical nature of the interaction of POR with inner membranes is confusing. Sequencing data (Schulz et al., 1989; Spano et al., 1992) and in vitro assembly studies (Dahlin et al., 1995; Teakle and Griffiths, 1993) in chloroplasts have suggested that POR is a peripheral protein located on the stromal side of the thylakoid membrane. Despite this, it is also a protein tightly bound to the membrane, as previously shown by salt and detergent washes of immobilized PLBs (Grevby et al., 1989). This is further confirmed in Figure 3(b), where post-treatment with NaOH and Na2CO3 following association, in the presence of NADPH, had a very limited effect on the removal of assembled pPOR. The location of the protein to the stromal side of the inner membranes is clearly seen. POR must be accessible to thermolysin, as the action of this protease results in a 2 or 4 kDa smaller POR-DP product.

The assembly of pPOR into PTs appears to be strictly dependent on the presence of exogenous NADPH (Figure 4a). The assembly of pPOR into PLBs could occur even when the samples were deprived of exogenous NADPH. However, as the preservation of the regular ultrastructure of isolated PLBs has been shown to depend on the presence of NADPH (Ryberg et al., 1988), the PLBs (and PTs) were isolated in the presence of NADPH. During the study of NADPH dependence, NADPH was excluded from the last centrifugation step. The loss of the assembly after an extra washing step indicates that the retained binding was not an intrinsic property of the PLB membrane, but more probably an effect of retained NADPH in the tubular PLBs from the isolation procedure. This retention did not occur with PTs. However, repeated washing did not eliminate the capacity for POR to associate with PLB membranes, but only abolished the proper binding as revealed by the thermolysin post-treatment (Figure 4b). Attachment, but no binding, occurred in either PLBs or PTs of the POR protein ALA06 mutated in the NADPH-binding site, not even in the presence of 1 mm NADPH (Figure 5a). This strengthens the interpretation that the PLB can retain a substantial amount of NADPH, which suffices for binding of POR, even in the absence of exogenous NADPH. This suggests a strict NADPH requirement for proper association of POR with both PLBs and PTs. A requirement of NADPH for correct folding of POR is possible. POR mutated in the active site (ALA24) had a severely hampered capacity to bind to both PLBs and PTs, but small amounts of the protein still managed to assemble in the membranes (Figure 5b). This suggests that membrane association of POR can take place, albeit at a significantly lower efficiency, in the absence of bound pigment. Similar observations have been made in chloroplasts where membrane-bound pigments stimulated the integration of LHCP into the thylakoids (Dahlin and Timko, 1994).

In conclusion, both pPOR and POR (as assayed with the Δ-TP translation product) appear properly to associate with the two types of inner membranes – the PTs and PLBs – present in etioplasts. The binding process is dependent on the presence of NADPH and ATP, but does not require that the ternary complex of Pchlide, POR and NADPH has formed before membrane association. In addition, the existence of PLB as a cubic-phase membrane system does not present any hindrance for physiological membrane association of POR.

Experimental procedures

Plant material

Wheat grains (Triticum aestivum L. cv. Kosack, from Weibull, Landskrona, Sweden) were imbibed in tap water overnight and sown in commercially available soil (Yrkesplantjord, Hammenhög, Sweden). The seedlings were grown in darkness or in a greenhouse at 24°C for 6 days.

Isolation of plastids and etioplast inner membranes

Etioplasts and chloroplasts were isolated by a combination of differential centrifugation and Percoll gradient centrifugation (Cline, 1986). Purified plastids were resuspended in a small volume of import buffer (0.33 m sorbitol, 50 mm Hepes–KOH pH 8.0) and counted in a haemocytometer. Lysates were prepared by resuspending plastids in a suitable volume of a buffer of 10 mm Hepes–KOH and 10 mm MgCl2 pH 8.0 (10 HK buffer) for 5 min. Stroma-free EPIMs and thylakoids were obtained by a 4200 g centrifugation of the lysates for 8 min. The membranes were resuspended in import buffer, centrifuged as above, and again resuspended in a small volume of import buffer.

Isolation and purification of PLBs and PTs were performed in the presence of 0.3 mm NADPH according to Lindsten et al. (1988). To preserve the PLB structure and the Pchlide657 form after isolation, PLBs and PTs were resuspended in import buffer containing 0.3 mm NADPH. The purity of fractions was assayed with Coomassie Blue-stained SDS–PAGE gradient gels (10–15%) and low-temperature fluorescence measurements using an SLM Aminco 8100C spectrofluorometer (SLM Instruments, Urbana, IL, USA).

Preparation of radiolabelled precursor protein

Radiolabelled proteins were produced in vitro from the Bluescript vector containing the coding sequence for the pPOR (Spano et al., 1992) and different pAlter vectors, containing the coding sequences for the wild-type mature POR (Δ-TP), and the mutant proteins ALA06 and ALA24 (carrying, respectively, the charge-to-alanine mutations RD120,121A and DK270,273A; Dahlin et al., 1999; Wilks and Timko, 1995) from pea. The linearized plasmids were used in a coupled transcription/translation wheatgerm-extract system (TNT) from Promega (Madison, WI, USA), using [3H]leucine (5.8 PBq mol−1) and T7 RNA polymerase. Before use the translation mixture was diluted threefold with import buffer and adjusted to 30 mm unlabelled leucine.

Protein analysis

The total amount of protein was measured (Bradford, 1976) prior to association and SDS–PAGE. Bovine serum albumin (Sigma, St Louis, MO, USA) was used as standard protein. Samples were solubilized in CHAPS (3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate, 2.5% w/v, final concentration) before addition of the colour reagent (B. Wiktorsson, Department of Plant Physiology, Göteborg University, Sweden, personal communication).

Membrane association assays

Isolated PLB and PT membranes or EPIMs and thylakoids were assayed for proper membrane association during incubation with the radiolabelled POR constructs. Exogenous MgATP (10 mm), containing 20 mm MgCl2, and NADPH (1 mm) were present during the incubation. In samples with no MgATP added, the translation mixture containing radiolabelled protein was filtered through a Sephadex G-25 column in order to remove excess nucleoside triphosphates. All handling and membrane association assays were performed in weak green light or in darkness. If not stated otherwise, samples were post-treated with or without thermolysin (100 µg ml−1) while kept on ice for 30 min. The post-treatment was inhibited by EDTA (final concentration 25 mm), the samples were centrifuged, and the pellets (consisting of membrane-bound radiolabelled POR) were dissolved in a small volume of import buffer containing EDTA (final concentration 25 mm). The nature of the membrane association of POR was assayed with post-treatment with NaOH (final concentration 50 mm) and Na2CO3 (final concentration 50 mm). Proteins were separated by 12.5% SDS–PAGE using the buffer system of Laemmli (1970) and prepared for fluorography and exposed to X-ray film at −80°C.


The authors thank Ann Marie Månqvist and Nancy Ilenius for skilful technical assistance, and Margareta Ryberg and Agneta Lindsten for valuable discussions. Financial support from the Adlerbertska foundation to H.A., the Hvitfeltds stipend foundation and Helge Axelssons foundation to S.E., and grants from the Swedish Natural Science Research Council to C.S., The Carl Trygger Foundation for Scientific Research to C.D., and the US DoE to M.P.T. are gratefully acknowledged.