Nearest-Neighbor Analysis of Higher-Plant … Physiol. (1 996) 11 2: 409-420 Nearest-Neighbor...

Plant Physiol. (1 996) 1 1 2: 409-420

Nearest-Neighbor Analysis of Higher-Plant Photosystem I Holocomplex’

Stefan Jansson*, Birgitte Andersen, and Henrik Vibe Scheller

Plant Biochemistry Laboratory, Department of Plant Biology, Royal Veterinary and Agricultural University, Thorvaldsenvej 40, 1871 Frederiksberg C, Denmark (S.J., B.A., H.V.S.); and Department of Plant Physiology,

University of UmeA, 901 87 UmeA, Sweden (S.J.)

Photosystem I (PSI) preparations from barley (Hordeum vulgare) and spinach (Spinacia oleracea) were subjected to chemical cross- linking using the cleavable homobifunctional cross-linkers dithiobis(succinimidy1propionate) and 3,3’-dithiobis(sulfosuccinimidyl- propionate). The overall pattern of cross-linked products was analyzed by the simple but powerful technique of diagonal electrophoresis, in which the disulfide bond in the cross-linker was cleaved between the first and second dimensions of the gel, and immunoblotting. A large number of cross-linked products were identified. Together with preexisting data on the structure of PSI, it was deduced that the subunits PSI-D, PSI-H, PSI-I, and PSI-L occupy one side of the complex, whereas PSI-E, PSI-F, and PSI-J occupy the other. PSI-K and PSI-G appear to be adjacent to Lhca3 and Lhca2, respectively, and not close to the other small subunits. Experiments with isolated light-harvesting complex I preparations indicate that the subunits are organized as dimers, which seem to associate to the PSI-A/PSI-B proteins independent of each other. We sug- gest which PSI subunit corresponds to each membrane-spanning helix in the cyanobacterial PSI structure, and present a model for higher-plant PSI.

In the photosynthetic light reactions of higher plants, electrons are transferred from water to NADP+ via the two photosystems, PSI and PSII, and a number of electron carriers. Details of the molecular structure of PSII are suggested from the crystal structure of the homologous reaction center of photosynthetic purple bacteria (Deisenhofer et al., 1985), but structural information on PSI at the atomic leve1 is lacking. PSI prepared from the thermophilic cyanobacterium Synechococcus elongatus has been crystallized and the structure determined at a resolution of 6 A (Krauss et al., 1993). The resolution subsequently has been im- proved, and a 4.5-A map was recently presented (Schubert et al., 1995). The determination of the structure of higher- plant PSI has been less successful, most likely because it is a much larger complex than its cyanobacterial counterpart. The cyanobacterial PSI complex consists of 11 subunits, PSI-A to PSI-F, and PSI-I to PSI-M (Golbeck, 1993; Chitnis et al., 1995). In higher plants PSI-M has not been reported,

’ This work was financially supported by the Swedish Forestry and Agricultural Research Council (S.J.), the Nordic Energy Re- search Programme (B.A.), and the Danish Center for Plant Biotech- nology.

* Corresponding author; e-mail stefan.janssonQp1antphys.umu.se; fax 46-90-16 -66-76.

409

but seven additional proteins are found, PSI-G, PSI-H, PSI-N, and the four LHC I polypeptides, which are the gene products of the Lhcal, Lhca2, Lhca3, and Lhca4 genes (Jansson et al., 1992). FNR (the PetH gene product) and the LHC I1 polypeptides Lhcbl and Lhcb2 also associate with higher-plant PSI and are found in varying amounts in PSI preparations. In cyanobacteria PSI can assemble into trimers (Kruip et al., 1994), but no compelling evidence for a trimeric structure of higher-plant PSI has been presented.

Twenty-nine transmembrane (Y helices have been assigned in the 4.5-i% cyanobacterial PSI structure, of which 22 are oriented in a 2-fold symmetrical pattern. The symmetry is not surprising, since the ancestor of PSI probably was a homodimeric reaction center, similar to what has been retained by green sulfur bacteria (Büttner et al., 1992) and heliobacteria (Liebl et al., 1993). Duplication of the ancestral gene for the reaction center polypeptides has occurred over time, and the two halves of the reaction center (the PSI-A and PSI-B subunits) have diverged (Ver- maas, 1994). At the resolution of 4.5 A, it is not possible to directly assign helices to specific parts of polypeptide chains. Although a pair-wise similarity between some of the smaller subunits can be observed, it has been assumed that the 22 symmetrical helices (a-h and a’-h’) are parts of the two large subunits, PSI-A and PSI-B, and the remaining 7 (j, k, 1, m, p , y, and w) correspond to the smaller subunits (Schubert et al., 1995). Helices i, k, I , p, and w are localized on the side of the complex forming the domain for monomer interaction in trimers, whereas helices m and y are located on the opposite ”outer” side of the complex. Little is known about the functions and positions of the smaller PSI subunits. Located on the stromal side of the complex (Oh-oka et al., 1989) are PSI-C, carrying the iron-sulfur centers F, and F, (Hsj et al., 1987); PSI-D, required for electron transfer to Fd (Chitnis et al., 1989) and binding of PSI-C (Li et al., 1991; Naver et al., 1995); and PSI-E, which is important for Fd reduction (Rousseau et al., 1993) and cyclic electron transport around PSI (Yu et al., 1993). PSI-C is positioned at the PSI-A/ PSI-B interface directly adjacent

Abbreviations: CPI, chlorophyll a-protein I; DM, n-decyl-f.3-D- maltopyranoside; DM-PSI, barley PSI prepared with DM; DTSP, dithiobis(succinimidy1propionate); DTSSP, 3,3‘-dithiobis(sulfosuc- cinimidylpropionate); EDC, ethyl-3-[3-(dimethylamino)propyl] carbodiimide; FNR, Fd:NADP+ oxidoreductase; LHC I, light- harvesting complex of PSI; LHC 11, light-harvesting complex of PSII; TX-PSI, barley PSI prepared with Triton X-100.

www.plantphysiol.orgon June 30, 2018 - Published by Downloaded from Copyright © 1996 American Society of Plant Biologists. All rights reserved.

http://www.plantphysiol.org

41 O Jansson et al. Plant Physiol. Vol. 11 2, 1996

to these proteins, apparently covered by PSI-D and PSI-E (Oh-oka et al., 1989; Andersen et al., 1992a). The PSI-H subunit, which is absent in cyanobacteria, may also be an extrinsic protein on the stromal side. Although PSI-H of higher plants is predicted to be membrane-spanning, that of Chlamydomonas reinhardtii is not (Andersen and Scheller, 1993). The remaining subunits seem to be intrinsic membrane proteins, except for PSI-N, which is an extrinsic protein at the lumenal side (Nielsen et al., 1994).

It is hoped that crystallizing PSI and solving its structure by x-ray diffraction or electron crystallography (Kiihl- brandt et al., 1994) will provide detailed information on the structure of PSI. Electron microscopy of single particles (Boekema et al., 1990) and chemical cross-linking can also be used to obtain structural information, albeit not at the atomic level. Chemical cross-linking has been used to dem- onstrate protein-protein contacts between PSI-C and PSI-D, PSI-C and PSI-E, PSI-D and PSI-E (Oh-oka et al., 1989; Andersen et al., 1992a), PSI-D and PSI-H (Andersen et al., 1992a), PSI-D and Fd (Zanetti and Merati 1987; Zilber and Malkin, 1988), PSI-F and plastocyanin (Wynn and Malkin, 1988; Hippler et al., 1989), PSI-E and FNR (Andersen et al., 1992b), PSI-D and PSI-L (Chitnis and Chitnis, 1993), and PSI-E and PSI-F (D. Bryant, personal communication). In- teraction between PSI-C and PSI-D and/or PSI-E has been observed in studies of mutants (Chitnis et al., 1989; Man- nan et al., 1994) and in reconstitution experiments (Li et al., 1991; Naver et al., 1995, 1996). Interactions between PSI-F and PSI-J (Xu et al., 1994) and PSI-L and PSI-I (Xu et al., 1995) have also been suggested. Hydrophilic cross-linkers such as EDC and DTSSP cross-link parts of neighboring membrane proteins exposed on the outside. Hydrophobic cross-linkers, on the other hand, have a greater penetrating capacity and produce a more complex pattern of products. When a hydrophobic cross-linker such as DTSP is used to cross-link a multiprotein complex such as PSI, a large number of different products are observed and analysis of the nearest-neighbor relations is arduous.

We have previously generated a nearly complete set of monospecific antibodies against the different PSI subunits. This collection allows the products to be identified when the higher-plant PSI holocomplex is cross-linked with DTSP. A more important tool, however, was to combine the immunoblotting with diagonal electrophoresis, thereby exploiting the ability to cleave DTSP and DTSSP under reducing conditions. We also used PSI preparations with different polypeptide compositions isolated from barley (Hordeum vulgare) and spinach (Spinacia oleracea). With this experimental setup, the composition of many cross-linked products not previously reported has been elucidated. The structural information obtained has been interpreted and now presented in a model for the PSI of higher plants.

MATERIALS A N D METHODS

PSI Preparations

DM-PSI was prepared essentially as described previously (Andersen et al., 1992b). Barley (Hordeum vulgare) thylakoids were solubilized in 1.5% DM for 30 min at 4°C

and the PSI holocomplex was purified over two successive anion-exchange columns (DEAE fast flow, Pharmacia) eluted with a 60 to 500 mM NaCl gradient in 20 mM Tricine, pH 7.5, 0.3% DM, and a gel-filtration column (Ultrogel AcA34, IBF Biotechnics, Villeneuve-la-Garenne, France) equilibrated in 25 mM Mes, pH 6.5, 250 mM NaC1, 0.3% DM. TX-PSI was prepared according to H0j et al. (1987) and spinach (Spinacia oleracea) PSI and LHC I were prepared according to PBlsson et al. (1995). A barley LHCI-730 preparation was obtained by separating barley thylakoids solubilized in 3% DM in an anion-exchange column as above.

NADP+Photoreduction Assay

Naver et al. (1995). NADP' photoreduction was measured according to

Protein Sequencing

N-terminal sequencing of proteins immobilized on membranes (Problot, Applied Biosystems) was carried out as described previously (Kjaerulff et al., 1993), except that the transfer buffer was 50 mM Tris, 190 mM Gly, and 0.1% SDS. The membrane was washed severa1 times with water to remove excess Gly that would otherwise interfere with amino acid sequencing.

Production of PSI-K Antibody

Antibodies against PSI-K were produced by coupling a synthetic peptide with the sequence EARESGLQTGDPAC to soybean trypsin inhibitor using N-succinimidyl-3-(2- pyridy1dithio)-propionate according to Gordon et al. (1987) and immunizing a rabbit with the conjugate. The peptide was synthesized at the Department of Immunology, Bio- medica1 Center, Uppsala University, Sweden.

Chemical Cross-Linking and Analysis of the Products

For cross-linking with DTSP and DTSSP, the preparations were diluted to 0.2 mg chlorophyll/mL in 20 mM Tricine, pH 7.5, 0.3% DM. DTSP was dissolved in dimeth- ylformamide and added to a final concentration of 0.015%. DTSSP was dissolved in water and added to a concentration of 0.05%. After incubation at room temperature for 30 min, the cross-linking reactions were stopped by addition of one-seventh volume of 10 mM Tris, pH 7.4,l mM EDTA. One volume of nonreducing sample buffer (50 mM Na,CO,, 15% SUC, 2.5% SDS) was added and the reactions were stored at -20°C. Diagonal electrophoresis was carried out using a modification of the method of Wang and Richards (1975). Samples were solubilized at room temperature for 20 min and electrophoresed on 8 to 25% polyacrylamide gels prepared according to Fling and Gregerson (1986). After electrophoresis, the lanes were cut out and incubated for 30 min in sample buffer containing 50 mM DTT (reducing sample buffer) to obtain complete reductive cleavage of the introduced cross-links. The gel slice was placed on top of a second 8 to 25% gel and re-electrophoresed. The gels were first stained with Coomassie brilliant blue



Cross-Linking Studies of PSI 41 1

R-250 and subsequently, after thorough destaining, with silver (Wray et al., 1981) to detect minor cross-linked products.

For immunoblotting, electrophoretically separated proteins (solubilized in nonreducing sample buffer) were transferred to nitrocellulose membranes and incubated with antibodies. Protein-antibody conjugates were visual- ized either with alkaline phosphatase-linked secondary antibody and a dye reaction or with the ECL system (Amer- sham). Table I lists the antibodies that were used for detection of the different proteins, together with the molecular masses of the PSI subunits and the predicted number of membrane-spanning regions.

A11 cross-linking reactions were analyzed by diagonal SDS-PAGE. Proteins that have not been cross-linked migrate with the same apparent molecular mass in both dimensions, and will thus form a diagonal in the final gel. Cross-linked products migrate in the first dimension approximately according to their combined apparent molecular masses. After cleavage of the disulfide bond in the cross-linker, the previously linked proteins will migrate according to their normal apparent molecular masses in the second dimension and thus form two vertically aligned spots on the gel. For example, a cross-linked product of a 15- and a 10-kD protein would migrate along with an un-cross-linked 25-kD protein in the first dimension, but after the second dimension it would appear as two spots at 15 and 10 kD, located beneath the 25-kD protein on the diagonal.

There are two major limitations to this analysis. First, two cross-linked products with the same gel mobility will yield four spots on the same vertical line after cleavage and it cannot be deduced which of these proteins constitute the cross-linked partners. Second, proteins with the same apparent molecular mass (such as Lhcal and Lhca4) cannot be distinguished from each other. To overcome the first problem, we analyzed PSI preparations with a less complex

polypeptide composition (barley TX-PSI and a DM-PSI preparation lacking Lhca3 and PSI-N). A more simple cross-linking pattern was also obtained using DTSSP, which forms fewer products than DTSP. In this way we could distinguish between two co-migrating products. To overcome the second problem and to positively identify the individual protein components of the cross-linked products, we used immunoblotting. In general, the pattern obtained in immunoblots matched the stained pattern for the protein in the diagonal electrophoresis. Some antibodies, however, were too weak to identify less-abundant cross- linked products, particularly with the spinach samples. In this paper we show only a few of the western blots; a11 cross-linked products could be seen on the diagonal gels, and when the pattern from the immunoblots differs from the pattern on the gels, it is mentioned in the text.

RESULTS

The Polypeptide Composition of the PSI Preparations

The polypeptide composition of the barley DM-PSI and TX-PSI and spinach PSI is shown in Figure 1. For the barley PSI complex the protein bands in the gel were assigned to specific subunits by amino acid sequencing and antibody reactions. We have not previously been able to identify the PSI-J subunit in our preparations. Sequencing of a protein migrating as a weakly stained band in the front of the free pigment band (approximately 1 kD) resulted in the sequence MRDIKTYLSVAP(V)LV, which is almost identical to the N-terminal part of the amino acid sequence obtained from the psaJ gene from rice (Hiratsuka et al., 1988), and is in agreement with the deduced amino acid sequence of the barley psaJ gene (U. Karnal, unpublished data). We can thus conclude that the barley DM-PSI preparations contain a11 known higher-plant PSI subunits. A small, substoichio-

Table 1. The protein subunits of higher-plant PSI holocomplex and antibodies used in this study to detect the proteins

Antibody (ref.) Gene Molecular Apparent Molecular Apparent Molecular Transmembrane Present in Product M a s Mass in Barley Mass in Spinach Helices (predicted) Cyanobacteria

PsaA PsaB PsaC PsaD PsaE PsaF PsaG PsaH Psal PsaJ PsaK PsaL PsaN Lhcal Lhca2

Lhca3 Lhca4

83 83 9

18 11 16 11 10 4 4 7

18 10 22 23

25 22

70 70 9

18 15 15 9

10 3 2 7

14 10 20 22

24 20

kD

70 70 8

18 13 15 9

10 3 2 7

14 9

20 22

24 20

11 11

O O O 1

1-2 0-1

1 1

1-2 2 O 3 3

3 3

Yes Y es Yes Yes Y es Yes N o N o Yes Y es Yes Yes No N o N o

N o N o

a -

Andersen et al. (1992a) Andersen et al. (1 992a) Andersen et al. (1 992a) Scheller (unpublished data) Scheller (unpublished data) Andersen et al. (1 992a) Andersen et al. (1992a) Andersen et al. (1 992a) (see text) This work Andersen et al. (1992a) Scheller (unpublished data) Kr61 et al. (1995) Sigrist and Staehelin (1 994) H ~ y e r Hansen et al. (1 988) Kr61 et al. (1995) Kr61 et ai. (1995)

a -, N o antibodies against these proteins.



412 Jansson et al. Plant Physiol. Vol. 112, 1996

Barley SpinachDM-PSI TX-PSI LHCI LHCI PSI

A+B —

A,B —

FNR

.iif

[HI— A+B

Chl -J -

Figure 1. Polypeptide composition of barley DM-PSI, TX-PSI, andLHCI-730 (LHCI), spinach LHC I, and PSI. The proteins were solu-bilized, separated in 8 to 25% gels, and stained with Coomassiebrilliant blue. A + B, CPI; A,B, PSI-A and PSI-B; C, D, E. . . , PSI-C,PSI-D, PSI-E. . . , respectively; 1, 2, 3, 4, Lhcal, Lhca2, Lhca3, andLhca4, respectively; Chl, free chlorophyll.

metric amount of FNR was also present in the preparation.The barley TX-PSI preparation lacks the four LHC I pro-teins PSI-F, PSI-N, PSI-K, and PSI-J (Andersen et al., 1992a;Kjaerulff et al., 1993). We found that the antibody raisedagainst electroeluted PSI-I (Andersen et al., 1992a) recog-nized both the PSI-I and PSI-J subunits. This did not con-stitute a major problem, since the PSI-I and PSI-J polypep-tides could be distinguished by their different elec-trophoretic mobility.

The polypeptide profile of spinach PSI was deducedfrom immunoblots. With the exception of the PSI-I/PSI-Jantibody, all antibodies recognized their spinach counter-parts, although the reactions were generally weaker thanwith barley proteins. Some minor differences in the mobil-ity of the individual proteins were detected between spin-ach and barley. In barley, PSI-E and PSI-F co-migrate, andso do PSI-H and PSI-N and also PSI-C and PSI-G. Inspinach, these proteins separate from each other, whereasPSI-N and PSI-G co-migrate (Fig. 1).

One particular barley DM-PSI preparation containedvery little PSI-N, LhcaS, and PSI-E but instead contained anadditional protein of 12 kD. N-terminal sequencing of the12-kD protein gave the sequence TKEPAKAKPPPRGP,which matches the sequence of PSI-E without the first 25amino acid residues. Thus, the 12-kD band is a proteolyticbreakdown product of PSI-E. This preparation turned outto be useful for the characterization of the cross-linkedproducts. The absence of these three proteins made it pos-sible to distinguish between some otherwise co-migratingcross-linked products by comparing the immunoblotting

pattern from this preparation with the "native" pattern (seebelow). This preparation was almost as active in reducingNADP+ (approximately 1800 jumol NADPH mg"1 chloro-phyll h"1) as the normal preparations (approximately 2200/xmol NADPH mg"1 chlorophyll h"1).

The barley LHCI-730 preparation contained Lhcal andLhca4, with a small contamination of Lhca2 and LhcaS (Fig.1). The spinach LHC I preparation contained all four Lhcapolypeptides. The protein visible as a band between Lhca2and LhcaS in the spinach LHC I preparation has beenobserved occasionally in LHC I preparations and is recog-nized by our Lhca2 antibodies and probably represents amodified form of this protein. In this particular gel (Fig. 1)separation of barley Lhcal and Lhca4 was achieved, butthis could not be done reproducibly.

Epitope Loss upon Cross-Linking

DTSP and DTSSP are homobifunctional succinimide es-ters that react with the amino groups of Lys residues andthe amino termini of proteins. Lys residues are often sur-face exposed and are involved in forming the majorepitopes of the protein. Attachment of a cross-linker tothese epitopes hampers antibody binding and can result inreduced affinity of the antibody toward the cross-linkedprotein. This effect was observed with several of the anti-bodies and sometimes resulted in difficulties in detectingcross-linking products, as illustrated in Figure 2. Identicalfilters with barley DM-PSI were incubated with two differ-ent antibodies specific for Lhca2. The polyclonal Lhca2antibody, raised against a synthetic peptide with a se-quence derived from the gene sequence, efficiently recog-nized a DTSP cross-linked product of approximately 44 kDalong with the un-cross-linked Lhca2 protein of 22 kD;however, DTSSP did not efficiently cross-link Lhca2 (Fig.2A, lane DTSSP). The monoclonal CMpLHCI:! antibody(Lhca2 Carlsberg), on the other hand, detected only the22-kD band (Fig. 2B).

Lhca2BCMpLHCI :1

44kD- 44 kD-

22 kD-

PSI-L

«^ c^ <^

Figure 2. Cross-linker-induced epitope loss of PSI polypeptides.Equal amounts of barley PSI proteins un-cross-linked (Cntrl) or cross-linked with DTSP or DTSSP were loaded on a gel, blotted, andincubated with the polyclonal Lhca2 antibody (A), the monoclonalLhca2 antibody CMpLHCI:! (B), and the PSI-L antibody (C).



Cross-Linking Studies of PSI 413

A,BFigure 3. Diagonal electrophoresis of spinachPSI cross-linked with DTSP. The first dimensionof the gel was run from left to right and thesecond dimension (after cleavage of the cross-linker) was run from top to bottom. The gel (left)was first stained with Coomassie brilliant blueand subsequently with silver. The apparent mo-lecular weight in the first dimension is indicatedat the bottom and the migration of the differentproteins, abbreviated as in Figure 1, in the sec-ond dimension is shown on the right. A pictorialmap of the gel is shown (right) with the identi-fied protein spots derived from the differentcross-linking products labeled as in Table II.

Another illustration of epitope loss is the behavior of thePSI-L antibody, shown in Figure 2C. Although the sameamount of PSI and thus PSI-L was loaded in all three lanes,the reaction against PSI-L cross-linked with DTSP is muchweaker than against PSI-L un-cross-linked and DTSSP-cross-linked. A weak band of 37 kD is seen in the DTSPlane, which turned out to be a highly abundant cross-linked species between PSI-L and PSI-D (see below). It hasto be pointed out, however, that the epitopes for most ofthe antibodies were not altered by the cross-linking.

Identification of Cross-Linked Products

We used the diagonal electrophoresis to analyze thepattern of cross-linking products in the different prepara-tions. One example of the results is presented in Figure 3,where the pattern of products in the spinach DTSP reac-tion, which produced the clearest result, is shown. If non-abundant products of small proteins (e.g. PSI-C) should bevisualized on the silver-stained gel, abundant proteins thatreadily cross-linked (such as the LHC I proteins) wereseverely overstained and obscured the pattern of products.This was exaggerated by a tendency to increase streakingin the first dimension of the gel, where no DTT was presentin the sample buffer. Therefore, we will show subsectionsof the gels that have been adequately stained to visualizethe different cross-linking products discussed in the textand listed in Table II. The interpretation of the cross-linking pattern is shown in line drawings next to the pho-tographs of the gels. The numbers of the protein spotscorrespond to the numbers of the cross-linking products inTable II. In a single gel not all products could easily bedetected. With the few exceptions mentioned below, noattempts were made to identify products containing threeor more polypeptide chains, since we assume that the twocross-links that are keeping the trimeric product togethercould also be formed separately.

The gel in Figure 3 shows a product of 13 kD (no. 1),which is clearly resolved into PSI-H and PSI-I after reduc-tive cleavage. At 19 kD there is a product (no. 2) composedof PSI-L and PSI-I, at 20 kD a product (no. 3) of PSI-F andPSI-J, and at 21 kD a product (no. 4) of PSI-E and PSI-C. The

latter product is only faintly stained. At 19.5 kD there is asingle spot (?) that corresponds to PSI-G or PSI-N. In thespinach samples these two proteins co-migrate and ourantibodies were not able to react with cross-linked prod-ucts of these proteins. The combined molecular mass ofPSI-G and PSI-N would, however, correspond well to 19.5kD and the single spot could represent a dimer of the twoin any combination. At 25 to 30 kD there are several prod-ucts and, to aid in identifying them, a magnified subsectionof a gel with spinach PSI cross-linked with DTSSP is shownin Figure 4A. PSI-L makes a product with PSI-H (no. 6) anda trimeric product (no. 9) with PSI-H and PSI-I (the latter

Table II. Molecular masses of cross-linked products of spinachand barley PSI holocomplexes

Product

1 PsaH + Psal2 PsaL + Psal3 PsaF + PsaJ4 PsaE + PsaC5 PsaF + PsaNa

6 PsaL + PsaH7 PsaD + PsaC8 PsaF + PsaE9 PsaL + PsaH + Psal

10 Lhca3 + PsaK11 PsaF+?12 Lhca2 + PsaCa

13 Lhca3 + PsaK14 PsaD + PsaE15 PsaD + PsaL16 Lhca1/4 + Lhca1/417 Lhca4 + Lhca2/3a

18 Lhca2 + Lhca2/3a

19 Lhca3 + Lhca3a

20 (Lhca)3

ApparentMolecular

Mass inSpinach

1318192026272829303031313234

n.d.=40=42=44-46=70

ApparentMolecular

Mass inBarley

kD

13n.d.b

1824

n.d.2627332830

n.d.n.d.

323536

=40=42=44=45=70

Confirmed byImmunoblotting

XC

X--X_XXX__

XXXXXXXX

a Uncertain identification,confirmed by immunoblotting.

b n.d., Not detected. c-, Not www.plantphysiol.orgon June 30, 2018 - Published by Downloaded from

Copyright © 1996 American Society of Plant Biologists. All rights reserved.



only in DTSP; Fig. 3), and PSI-D cross-links to both PSI-E(no. 14) and PSI-C (no. 7). PSI-F makes products with PSI-E(no. 8), with PSI-G or PSI-N (no. 6), and a product of about31 kD (no. 11) with an unknown partner, perhaps repre-senting a homodimeric product of PSI-F resulting fromintercomplex cross-linking. The partner in the 31-kD prod-uct was neither Lhca2 nor LhcaS, which are both compo-nents of products of similar sizes, since the Lhca2 andLhca3 products were abundant only after reaction withDTSP, whereas the PSI-F product also formed readily withDTSSP (cf. Figs. 3 and 4A). LhcaS forms two products (30and 32 kD, nos. 10 and 13, respectively) with PSI-K. Theexistence of a Lhca3-PSI-K cross-link was confirmed by theabsence of PSI-K products of that size when the barleypreparation without LhcaS was analyzed. We find no evi-dence for the involvement of a third polypeptide in thelarger of the PSI-K/LhcaS cross-linking products. Since wehave found that two LhcaS polypeptides probably are indirect contact with each other (see below), these two pro-teins could be adjacent to different parts of the PSI-Kprotein.

Spinach Lhca2 also forms a product (no. 12) of similarsize, although we have not been able to clearly identify thepartner. There is a weak spot corresponding to the size ofPSI-G or PSI-N in this region. Since no 31-kD protein wasdetected by the PSI-N antibodies, this spot might corre-spond to PSI-G, although a positive identification was notpossible. In the absence of a better suggestion, we thinkthat PsaG could be the partner to Lhca2, but this issue hasto be addressed in future studies. The abundant dimericLHC I products will be discussed below.

For most of the proteins (PSI-E, PSI-H, PSI-K, Lhcal,Lhca2, LhcaS, and Lhca4) our antibodies were reactiveenough against the cross-linked spinach proteins and per-mitted confirmation of the cross-linking pattern by westernblotting. For the others (PSI-C, PSI-D, PSI-F, PSI-G, PSI-L,and PSI-N) such confirmation was not possible.

The pattern on the diagonal gels of barley PSI proteinscross-linked with DTSSP is shown in Figure 4B. In barley,three protein pairs co-migrated and the barley proteins hada higher tendency for streaking compared with their spin-ach counterparts, probably a consequence of differences inthe preparation methods. On the other hand, the antibodiesrecognized the barley proteins better. In most cases theproducts in spinach and barley were the same (Table II);the only unique product in barley was between PSI-L andPSI-D (no. 15). The protein migrating between PSI-H/PSI-N and PSI-L in Figure 4B is the truncated form of PSI-Ediscussed above, which was present in small amounts inthis particular preparation as well. There is a product la-beled E/F + ? (8?) in Figure 4B that probably consists ofPSI-E + PSI-F. In the preparation containing truncatedPSI-E, cross-linking between truncated PSI-E and PSI-F wasvery efficient (data not shown) and we believe that theseproteins should cross-link to each other in the "intact"preparation as they do in spinach. A weakly stained, dif-fuse PSI-F + truncated PSI-E product seen in Figure 4Baligned vertically with the PSI-L + PSI-H product. Trun-cated PSI-E also cross-linked to PSI-D (data not shown),

showing that the N terminus of PSI-E is not important forthe interaction with either PSI-D or PSI-F. Immunoblottingconfirmed the pattern from the diagonal gels, although thePSI-C and PSI-G antibodies did not detect any cross-linkingproducts and the PSI-L antibody did so only very ineffi-ciently (Fig. 2).

The TX-PSI preparation was also subjected to cross-linking with DTSP and that pattern was much simpler,due to the absence of many subunits. The products thatcould be detected on the diagonal gels and in the immu-noblots were PSI-H + PSI-I, PSI-H + PSI-L, PSI-C +PSI-E, PSI-C + PSI-D, PSI-D + PSI-E, and PSI-D + PSI-L(data not shown).

The reaction center polypeptides of PSI, PSI-A and PSI-B,co-migrated in our gels and we had no antibodies thatdistinguished these proteins. When samples prepared us-ing DM were electrophoresed, PSI-A and PSI-B werepresent as a pigmented dimer (CPI), but in the TX-PSIsample they migrated as apoproteins (Fig. 1). Apparently,DM stabilizes CPI even in the presence of sample buffercontaining 2.5% SDS. Using DTSP, most if not all PSIpolypeptides formed large cross-linked products. Simi-larly, immunoblotting with most antibodies produced a.diffuse reaction in the >100-kD region consisting of PSI-Aand/or PSI-B and one or more smaller PSI proteins. Thegeneration of such products is not surprising because it islikely that all PSI polypeptides have direct contact withPSI-A or PSI-B. No valuable information could be obtainedfrom these large products.

A Spinach DTSSP

13

11 8

24 20/ / A \\ /

UJ cj Ul o I g O

321/4DFLEHN/GC

B Barley DTSSP12

40 35

19 -• _17 j»»-10

^ £ -717 14 15

15-* 8?

7^.*

I I I / / /50 5 LU C'- o x

4

0

• 3• 2• 1/4• D• E,F• L

• H,N-C,G

Figure 4. A, Subsection of a gel of spinach PSI cross-linked withDTSSP; B, subsection of a gel of barley DM-PSI cross-linked withDTSSP. Abbreviations are as in Figure 3. www.plantphysiol.orgon June 30, 2018 - Published by Downloaded from



Cross-Linking Studies of PSI 415

Cross-Linked Products with LHC I

The four LHC I polypeptides cross-linked efficiently toCPI and formed, in both spinach (Fig. 3; a subsection of alesser stained gel is shown in Fig. 5) and barley, abundantproducts in the 40- to 45-kD range with DTSP, apparentlyconsisting of Lhca dimers. Most products were also formedusing DTSSP, but at lower yield. It is evident that LhcaS,Lhca2, and Lhcal / Lhca4 form short horizontal lines in the40- to 45-kD region rather than single spots, which showsthat they each form at least two different dimeric cross-linking products. Lhcal /Lhca4 seem to make products ofboth 42 and 40 kD (nos. 16 and 17), but immunoblottingshowed that the 42-kD product consisted of only Lhca4.The Lhca4 antibody decorates a diffuse band correspond-ing to a closely spaced, poorly resolved doublet of whichthe Lhcal antibody recognized only the lower band (Fig. 6).Additional evidence for this came from immunoblots of thebarley PSI cross-linked with DTSSP, where only the 42- andnot the 40-kD Lhcal/Lhca4 product was formed (Fig. 4B).We failed to detect dimeric Lhcal but a dimeric productwas clearly recognized by the Lhca4 antibody (data notshown), indicating that Lhca4, but nor Lhcal, formed a42-kD product. The 40-kD product (no. 16) could be aLhcal/Lhca4 heterodimer or homodimers with identicalmolecular masses.

Above 40 kD, the diagonal gels indicated the presence ofproducts of the LhcaS, Lhca2, and the Lhca4 proteins, butour data do not allow us to make firm conclusions aboutthe identities of these products. Immunoblotting with theLhca antibodies did not yield distinct bands, but ratherdiffuse zones similar to the pattern on the stained gels.Lhca4 does not form products larger than 42 kD, and itappears that LhcaS forms a slightly larger product (no. 19)than Lhca2, suggesting that LhcaS forms homodimers (45kD). Lhca4 could cross-link to both Lhca2 and LhcaS, form-ing 42-kD products (no. 17), and there could be a ho-modimeric Lhca2 product of 44 kD. When the preparationlargely devoid of LhcaS was cross-linked with DTSP andcompared with the normal preparation, the largest of theLhca2 products disappeared, whereas the Lhca4 productpattern was not affected, indicating that Lhca2 but notLhca4 cross-linked to LhcaS. One possible interpretation ofour results is that there are products consisting of LhcaS +LhcaS, Lhca2 + LhcaS, Lhca2 + Lhca2, Lhca2 + Lhca4, and,as mentioned above, Lhcal / Lhca4 homo- or heterodimers,but alternative interpretations of the cross-linking patterncould also be made.

- Dimer

- Monomer

CM i-

Figure 5. Subsection of a diagonal gel of spinach PSI cross-linkedwith DTSP. Abbreviations are as in Figure 3.

I \cts coo o

Figure 6. Immunoblotting of spinach PSI cross-linked with DTSPusing antibodies against Lhcal and Lhca4. A nitrocellulose mem-brane with spinach PSI cross-linked with DTSP was incubated withantibodies against Lhcal, the antibodies were washed off, and themembrane was subsequently incubated with antibodies againstLhca4.

We did not examine products that consisted of more thantwo polypeptide chains, but special attention was paid tothe cross-linking pattern of the LHC I polypeptides. Thiswas important because the stoichiometry of the LHC Iproteins in the native PSI complex remains an open ques-tion. Pigment stoichiometry arguments have been used tosuggest that two copies of each of the four Lhca polypep-tides are associated with each PSI center (Jansson, 1994),but it was suggested recently that each Lhca protein formsa homotrimer (Dreyfuss and Thornber, 1994), resemblingthe structure of the related Lhcbl/Lhcb2 proteins (Kiihl-brandt et al., 1994). LhcaS, Lhca2, and Lhcal/Lhca4 formedtrimeric complexes (no. 20) using DTSP (Fig. 3), althoughbarley LhcaS formed them only with low efficiency. Theappearance of trimeric cross-linked products shows onlythat one Lhca protein is in contact with two others; it doesnot indicate whether the in vivo structure of the protein isa trimer. Even monomeric proteins cross-link efficiently tomore than one protein, as shown with PSI-L, PSI-H, andPSI-I.

To obtain information about the configuration of theLHC I polypeptides, cross-linking of isolated LHC I wasperformed. A barley LHCI-730 preparation, consistingmainly of the two polypeptides Lhcal and Lhca4 (Fig. 1),was cross-linked with a higher ratio of DTSP to protein andthe cross-linked pattern was studied by diagonal electro-phoresis (Fig. 7). In spite of a very high yield of dimericproducts, no trimeric cross-linked products could be de-tected. We also analyzed the pattern of a cross-linked spin-ach LHC I preparation consisting of all four LHC I proteins(Fig. 1). As in the case of barley LHCI-730, only dimericcross-linking products were obtained without a trace oftrimeric complexes (Fig. 7). This strongly indicates that theLHC I polypeptides in these two preparations exist asdimers, and that higher-ordered structures are absent.

DISCUSSION

A Working Model for the Structure of Higher-Plant PSI

We present a series of experiments using two chemicalcross-linkers with several PSI preparations containing dif- www.plantphysiol.orgon June 30, 2018 - Published by Downloaded from




Figure 7. Diagonal electrophoresis of barley LHCI-730 (top) andspinach LHC I (bottom) cross-linked with DTSP. The migration of thedifferent proteins in the second dimension of the gel is indicated onthe right.

ferent polypeptides in order to study the molecular orga-nization of the PSI holocomplex. In contrast to earlier ex-periments of this type, we have used the hydrophobiccross-linker DTSP that reacts very efficiently with membrane-bound polypeptides in addition to DTSSP, and we havealso attempted to analyze the total pattern of the products.The number of cross-linked products was large, and iden-tifying their components would not have been possiblewithout our antibody collection. The patterns on the diag-onal gels and immunoblots are complex and in a few cases(e.g. spinach PSI-F) there are cross-linked products inwhich one of the participating proteins has not been iden-tified. Nevertheless, several previously undetected cross-linked products were observed and these are likely toreflect nearest-neighbor relationships between the differentprotein subunits of the higher-plant PSI holocomplex. Byconsidering this in the framework of the three-dimensionalstructure of cyanobacterial PSI (Krauss et al., 1993; Schu-bert et al., 1995) and by using positional information ob-tained with electron microscopy (Kruip et al., 1995), amodel for higher-plant PSI is suggested (Fig. 8).

PSI, like PSII, is a multiprotein complex with an internalpseudo-2-fold symmetry (Krauss et al., 1993). This is notsurprising, considering the sequence similarity betweenPSI-A and PSI-B. The symmetry is partially reflectedamong the smaller PSI subunits; PSI-G / PSI-K are homolo-gous and PSI-I/PSI-J have similarities (Andersen andScheller, 1993). Thus, it is possible that these proteins oc-

cupy corresponding positions on opposite sides of the com-plex. PSI-D and PSI-E may not be truly homologous, butthe barley polypeptides have similarities in the N-terminalregion (Scheller et al., 1988). In the structure of the cya-nobacterial PSI, seven membrane-spanning a helices (/, fc, /,m, p, q, and w in Schubert et al., 1995) have been assignedto the minor PSI proteins. Helices / and k at the side closeto the contact site between the monomers in the PSI trimerare linked by electron density (Krauss et al., 1993), showingthat they belong to the same polypeptide chain. The onlysmaller proteins in cyanobacteria that are predicted to spanthe membrane twice are PSI-K and PSI-L. The proteinoccupying this position should be important for trimeriza-tion because this is the only contact site between the mono-mers in the trimeric structure (Krauss et al., 1993). SincePSI-L, but not PSI-K, is essential for trimerization (Chitnisand Chitnis, 1993), it is likely that helices / and k are part ofPSI-L. Another helix, p, is in close contact with helices / andk (Schubert et al., 1995). Our data indicate that this helix ispart of PSI-I, which efficiently cross-links to PSI-L, and thisassignment is in agreement with the observation that psal~mutants of Synechocystis sp. PCC6803 have loosely boundPSI-L (Xu et al., 1995). Based on a weak sequence similaritybetween PSI-I and the D2 protein of PSII, it has beensuggested that PSI-I is involved in binding of electronacceptors (Scheller et al., 1989a). However, the position ofPSI-I away from the center axis of the PSI complex clearlydemonstrates that PSI-I cannot have such a function. This

Figure 8. A tentative model for the higher-plant PSI holocomplex. www.plantphysiol.orgon June 30, 2018 - Published by Downloaded from Copyright © 1996 American Society of Plant Biologists. All rights reserved.



conclusion is in agreement with the observation that psal- mutants have functional PSI (Xu et al., 1995). A relatively complex cross-linking pattern has been obtained for PSI-H. We have demonstrated cross-linking to PSI-L and PSI-I, and since a PSI-D + PSI-H contact has been demonstrated previously (Andersen et al., 1992a), the most probable location for PSI-H is on top of PSI-L and PSI-I on the stromal side. Our data give no direct information concerning the function of PSI-H, but the suggested involvement in binding of LHC I (Scheller and Msller, 1990) seems unlikely.

At the opposite location to PSI-I is helix m (Krauss et al., 1993), which we believe belongs to PSI-J. This has also been suggested from single-particle, high-resolution electron microscopy studies of wild-type and a psaF- / p s a J - strain of Synechocystis sp. PCC6803; these proteins constitute the outer edge of the complex, opposite to the site of contact between the PSI units in a trimer (Kruip et al., 1995). It has been speculated previously that PSI-F does not span the membrane, but this cannot be true, since PSI-F cross-links both to plastocyanin (Wynn and Malkin, 1988; Hippler et al., 1989), located in the lumen of the thylakoids, and to PSI-E, which is at the stromal side (this work; D. Bryant, personal communication). We believe that PSI-F corresponds to helix q, in agreement with the data presented by Kruip et al. (1995).

We speculate that helix 1 in the 4.5-A cyanobacterial PSI structure (Schubert et al., 1995) is part of PSI-M, which is not found in higher plants. Only two cyanobacterial proteins, PSI-K and PSI-M, remain unpositioned, and we think that our cross-linking data rules out the possibility that it should be PSI-K. A protein found at this position would probably cross-link to PSI-L and/ or PSI-D, but PSI-K cross- linked only to PSI-A 1 PSI-B and Lhca3. Instead, helix w on the same side of the complex is located quite far from the PSI-A/ PSI-B interface (Schubert et al., 1995). We think that this helix belongs to cyanobacterial PSI-K, which does not necessarily mean that the higher-plant PSI-K is found at the same position, since higher-plant PSI-G and PSI-K are equally related to cyanobacterial PSI-K (Kjzrulff et al., 1993). However, these homologous proteins presumably occupy symmetrical positions in the complex, implying that Lhca2 and Lhca3 also are located at opposite sides, provided that PSI-G is a cross-linking partner to Lhca2. We speculate that PSI-K occupies the position of cyanobacterial PSI-K and PSI-G takes the opposite position (Fig. 8), although it could also be the other way around. Another possibility is that PSI-G and PSI-K are not positioned symmetrically, which is more compatible with the simplest interpretation of the Lhca protein cross-linking pattern (see below). In any case, an association of Lhca polypeptides to the position of cyanobacterial PSI-K would interfere with protein-protein (PSI-L-PSI-L) contacts between the PSI monomers in a trimer. In Iight of this finding, it is not surprising that PSI from higher plants has not been isolated in a trimeric state.

A 31-kD product formed in spinach appeared to contain only PSI-F. Since PSI-F is present in only one copy per PSI (Scheller et al., 1989b), it is possible that this reflects an intercomplex cross-linking. A specific PSI-PSI interaction

is plausible, but then higher-plant PSI centers should attach to each other at the PSI-F side, not the PSI-L side, as in cyanobacteria. A nonspecific interaction, perhaps at the lumena1 side, cannot be ruled out, but it seems unlikely because the cross-linking experiments were carried out in a fairly diluted sample. No intercomplex products such as PSI-D or PSI-E were found in this study or in previous studies using other cross-linkers. Thus, we offer no expla- nation for the origin of the 31-kD cross-linking product.

PSI-N is known to be an extrinsic protein located on the lumenal side of PSI. In the spinach complex we were unable to differentiate between PSI-G and PSI-N, but found that either PSI-G or PSI-N cross-links to PSI-F. It is likely that PSI-N is the partner to PSI-F, which has a large portion of its polypeptide chain in the lumen. Even so, this does not give very detailed positional information, since PSI-F is known to cover a large lumenal area (Kruip et al., 1995).

The truncated 12-kD form of barley PSI-E is likely to be an experimental artifact, but its presence could nevertheless lend insight into the function of PSI-E. We have observed the 12-kD band in other PSI preparations and, in- terestingly, it seems to be correlated to the FNR content of the preparation. Preparations containing large amounts of FNR lack the 12-kD band, whereas preparations containing the 12-kD protein have drastically reduced amounts of FNR (B. Andersen, unpublished data). The truncated form of PSI-E cross-links efficiently to PSI-D and PSI-F, and our PSI-E antibody recognizes the 12-kD protein only weakly. Our interpretation of these results is that the N terminus of PSI-E carries the major epitopes and seems to mediate FNR binding, whereas protein-protein contacts with PSI-D and PSI-F involve the other parts of the PSI-E protein. Cya- nobacterial PSI-E Iacks a region corresponding to the N- terminal part of the higher-plant PSI-E (Andersen and Scheller, 1993) and resembles our truncated PSI-E. Based on our results, cyanobacterial PSI-E should not be able to bind FNR, and to our knowledge, no PSI-FNR preparation has been obtained from a cyanobacterium. The N terminus of PSI-E as well as PSI-N might also be dispensable for electron transport from plastocyanin to Fd, since the preparation with truncated PSI-E and reduced amounts of PSI-N was almost as active in photoreduction of NADP+ as our other preparations.

Dimeric Organization of LHC I

From the present data we cannot localize the Lhca proteins in the PSI complex with confidence, but we provide evidence that LHC I is organized as dimers. In both the spinach LHC I and barley LHCI-730 preparations, only dimers were found as cross-linked products and no trimers were detected. This suggests that the preparation consists of the four Lhca polypeptides in a dimeric state, not an ”LHC I holocomplex.” The finding that pigment-protein complexes in an identical LHC I preparation could not transfer energy to each other (Pilsson et al., 1995) agrees with this conclusion.

In contrast to the Lhca polypeptides, Lhcbl and Lhcb2 (and probably Lhcb3) proteins are found in trimeric complexes, whereas Lhcb4, Lhcb5, and Lhcb6 seem to be mo-



41 8 Jansson et al. Plant Physiol. Vol. 112, 1996

nomeric (Peter and Thornber, 1991; Jansson, 1994). Lhcbl and Lhcb2 (commonly designated LHC 11) can form partially solubilized, dimeric complexes during gel electrophoresis, but the yield of dimers is low. When gentle solubilization conditions are applied, the dimers disappear and the LHC I1 proteins migrate exclusively as trimers during electrophoresis (Peter and Thornber, 1991). In a recent publication (Dreyfuss and Thornber, 1994) oligo- meric LHC I subfractions were separated by electrophoresis, but it was assumed that the oligomers consisted of three, not two, polypeptides. The LHC I oligomers migrated considerably faster than Lhcbl / Lhcb2 trimers, but it was argued that the difference in size between the Lhca and Lhcb polypeptides could cause the difference in gel mobility. A size difference exists between Lhcal /Lhca4 and Lhcbl/Lhcb2, but the largest of the Lhca polypeptides, Lhca3, does in fact almost co-migrate with Lhcbl and Lhcb2 in denaturing gels. Instead, the mobility of the LHC I "trimers" seems to be similar to the mobility of our dimers. We think our data indicate that the Lhca oligomers obtained after mild solubilization actually are dimers, suggesting that the Lhca polypeptides also in vivo are organized as dimers. Dimeric organization is supported by the pigment stoichiometries of PSI and LHC I (Jansson, 1994), and also by the observation that dimers, not trimers, of Lhca polypeptides are obtained after gel electrophoresis if solubilization is incomplete (S. Jansson, unpublished data). Lhca3 might form homodimers, but we were not able to determine whether Lhca2, Lhcal, and Lhca4 form homo- or heterodimers. Biochemical evidence indicates that the Lhcal and Lhca4 polypeptides are tightly linked in the LHCI-730 complex (Knoetzel et al., 1992), thus a heterodimeric nature of Lhcal /Lhca4 is feasible. However, we have recently analyzed a barley mutant that lacks Lhca4 but has wild- type levels of Lhcal, which favors the idea that Lhcal and Lhca4 are not intimately associated (B. Bossman, J. Knoetzel, and S. Jansson, unpublished data).

The Lhca proteins appear to bind to the PSI-A/PSI-B proteins independently of each other. A11 four Lhca proteins cross-link to the PSI-A/PSI-B proteins, and close as- sociations between CPI and the Lhca proteins are necessary to ensure efficient energy transfer between the pigment molecules. As in the preparation analyzed here, Lhca3 is easily lost during isolation of barley PSI (B. Andersen, unpublished data), but the presence of the other Lhca proteins in such preparations shows that their binding to PSI is independent of Lhca3. Besides the barley mutant lacking Lhca4, we have also characterized a mutant lacking Lhcal and Lhca4 but retaining Lhca2 and Lhca3 (B. Boss- man, J. Knoetzel, and S. Jansson, unpublished data). Taken together, these data indicate that the postulated Lhca dimers associate directly with PSI-A/ PSI-B without the requirement for each other.

Concerning the positions of the Lhca proteins, it seems clear that they are not found close to the interface between PSI-A and PSI-B, where most of the other subunits are clustered. Instead, the four Lhca and the PSI-G and PSI-K proteins are probably located toward the ends of the com-

plex. There is a contradiction in our interpretation of the Lhca cross-linking pattern and our speculation that PSI-G and PSI-K are localized symmetrically: since Lhca3 cross- links to PSI-K and Lhca2 perhaps cross-links to PSI-G, Lhca2 and Lhca3 could not cross-link to each other if they are present on opposite sides of the complex. In our model we have paid more attention to the homology between PSI-G and PSI-K than to the Lhca cross-linking pattern, which we cannot interpret with confidence, but we want to stress that this model is a working model when it comes to the positions of the Lhca proteins. An alternative model could place a11 Lhca dimers next to each other on one side of the complex, perhaps with Lhca3 in the same position as shown in Figure 8, followed by Lhca2, Lhca4, and Lhcal. Future cross-linking experiments on PSI preparations from the barley mutants mentioned above should perhaps give more attention to this problem.

Diagonal Electrophoresis: A Powerful Technique to Study Nearest-Neighbor Relationships in Complex Systems

In this paper we have shown that the simple diagonal electrophoresis technique can be used to analyze the total pattern of cross-linking products in large multiprotein complexes, in our case consisting of 17 polypeptides. Tra- ditionally, nearest-neighbor relationships are analyzed by immunoblotting, but blots of cross-linked preparations of membrane proteins must be interpreted with great care, bearing in mind that cross-linked proteins could escape detection by antibodies even under conditions in which non-cross-linked proteins are strongly recognized. Diago- na1 electrophoresis, complemented with immunoblotting, represents a more powerful method to study nearest- neighbor relationships in multiprotein complexes, and we believe that the method may be applied successfully to systems other than PSI. Even though we could not un- equivocally determine the pattern of Lhca protein cross- links, a large number of cross-linking partners were identified, some that have already been found using other techniques and some that have not been reported previously. It is likely that epitope loss is more pronounced when membrane-bound proteins are studied. We imagine that hydrophobic proteins have relatively few antigenic sites, and these sites are likely to coincide with amino acids that are modified by chemical cross-linkers.

In conclusion, we have used diagonal electrophoresis to perform a nearest-neighbor analysis of PSI and we have presented a model for the physical organization of the subunits in the higher-plant PSI holocomplex. Some as- pects of the model are still speculative, but we hope that the model can stimulate further research to elucidate the functions of the different PSI polypeptides.

ACKNOWLEDCMENTS

We thank Professor Birger Lindberg Mdler for his support and valuable comments, Hanne Linde Nielsen and Inga Olsen for technical assistance, Staffan Tjus for his generous gift of the spinach I'S1 preparation, Roberto Bassi for the initial impetus to do diagonal electrophoresis, Matthias Rogner, Petra Fromme, and Don Bryant for valuable discussions and communication of un-




published data, Ib Svendsen for protein sequencing, and David Simpson for the gift of antibody.

Received May 6, 1996; accepted June 18, 1996. Copyright Clearance Center: 0032-0889/96/112/0409/12.

LITERATURE CITED

Andersen B, Koch B, Scheller HV (1992a) Structural and functional analysis of the reducing side of photosystem I. Physiol Plant 8 4 154-161

Andersen B, Scheller HV (1993) Structure, function and assembly of photosystem I. In C Sundqvist, ed, Pigment-Protein Complexes in Plastids: Synthesis and Assembly. Cell Biology: A Series of Monographs. Academic Press, San Diego, CA, pp 383-418

Andersen B, Scheller HV, Moller BL (199213) The PSI-E subunit of photosystem I binds ferredoxin:NADP+-oxidoreductase. FEBS Lett 311: 169-174

Boekema EJ, Wynn RM, Malkin R (1990) The structure of spinach photosystem I studied by electron microscopy. Biochim Biophys Acta 1017: 49-56

Biittner M, Xie D-L, Nelson H, Pinther W, Hauska G, Nelson N (1992) Photosynthetic reaction center genes in green sulfur bacteria and in photosystem I are related. Proc Natl Acad Sci USA

Chitnis PR, Reilly PA, Nelson N (1989) Insertional inactivation of the gene encoding subunit I1 of photosystem I from the cyanobacterium Synechocystis sp. PCC6803. J Biol Chem 264: 18381- 18385

Chitnis PR, Xu Q, Chitnis VP, Nechustai R (1995) Function and organization of photosystem I polypeptides. Photosynth Res 4 4 23-40

Chitnis VI', Chitnis PR (1993) PsaL subunit is required for the formation of photosystem I trimers in the cyanobacterium Syn- eckocystis sp. PCC 6803. FEBS Lett 336: 330-334

Deisenhofer J, Epp O, Miki K, Huber R, Michel H (1985) Struc- ture of the protein subunits in the photosynthetic reaction centre of Rhodopseudomonas viridis. Nature 318: 618-624

Dreyfuss BW, Thornber JP (1994) Organization of the light- harvesting complex of photosystem I and its assembly during plastid development. Plant Physiol 106 841-848

Fling SP, Gregerson DS (1986) Peptide and protein molecular weight determination by electrophoresis using high-molarity Tris buffer system without urea. Ana1 Biochem 155: 83-88

Golbeck JH (1993) The structure of photosystem I. Current Opin- ion in Structural Biology 3: 508-514

Gordon RD, Fieles WE, Schotland DL, Hogue-Angeletti R, Bar- chi RL (1987) Topological localization of the C-terminal region of the voltage-dependent sodium channel from Electrophorus electricus using antibodies raised against a synthetic peptide. Proc Natl Acad Sci USA 84: 308-312

Hippler M, Ratajczak R, Haenel W (1989) Identification of the plastocyanin binding subunit of photosystem I. FEBS Lett 250

Hiratsuka J, Shimada H, Whittier R, Ishibashi T, Sakamoto M, Mori M, Kondo C, Honji Y, Sun C-R, Meng B-Y, Li Y-Q, Kanno A, Nishizaea Y, Hirai A, Shinozaki K, Sugiura M (1988) The complete sequence of the rice (Oryza sativa) chloroplast genome: intermolecular recombination between distinct tRNA genes ac- count for a major plastid DNA inversion during the evolution of cereals. Mo1 Gen Genet 217: 185-194

Hej PB, Svendsen I, Scheller HV, Moller BL (1987) Identification of a chloroplast-encoded 9-kDa polypeptide as a 2[4Fe-4S] protein carrying centres A and B of photosystem l. J Biol Chem 262:

Heyer-Hansen G, Bassi R, Henberg LS, Simpson DJ (1988) Im- munological characterization of chlorophyll a / b-binding proteins of barley thylakoids. Planta 173: 13-21

Jansson S (1994) The light-harvesting chlorophyll a / b-binding proteins. Biochim Biophys Acta 1184 1-19

89: 8135-8139

280-284

12676-12684

Jansson S, Pichersky E, Bassi R, Green BR, Ikeuchi M, Melis A, Simpson DJ, Spangfort M, Staehelin LA, Thornber JP (1992) A nomenclature for the genes encoding the chlorophyll alb- binding proteins of higher plants. Plant Mo1 Biol Rep 10: 242-253

Kjaerulff S, Andersen B, Skovgaard Nielsen V, Meller BL, Okkels JS (1993) The PSI-K subunit of photosystem I from barley (Hordeum vulgare L.). Evidence for a gene duplication of an ancestral PSI-G/K gene. J Biol Chem 268 18912-18916

Knoetzel J, Svendsen I, Simpson DJ (1992) Identification of the photosystem I antenna polypeptides in barley isolation of three pigment binding antenna complexes. Eur J Biochem 206: 209-215

Krauss N, Hinrichs W, Witt I, Fromme P, Pritzkow W, Dauter Z, Betzel C, Wilson KS, Witt HT, Saenger W (1993) Three- dimensional structure of system I of photosynthesis at 6 A resolution. Nature 361: 326-331

KrÓ1 M, Spangfort MD, Huner NPA, Oquist G, Gustafsson P, Jansson S (1995) Chlorophyll a / b-binding proteins, pigment conversions, and early light-induced proteins in a chlorophyll b-less barley mutant. Plant Physiol 107: 873-883

Kruip J, Bald D, Boekema E, Rogner M (1994) Evidence for the existence of trimeric and monomeric photosystem I complexes in thylakoid membranes from cyanobacteria. Photosynth Res 40:

Kruip J, Bald D, Hankamer B, Nield J, Boonstra AF, Barber J, Boekema E, Rogner M (1995) Localization of subunits in PS1, PS2 and in a PS2/light-harvesting-supercomplex. In P Mathis, ed, Photosynthesis: From Light to Biosphaere, Vol 111. Kluwer Academic Publishers, Dordrecht, The Netherlands,

Kiihlbrandt W, Wang DN, Fujiyoshi Y (1994) Atomic model of plant light-harvesting complex. Nature 367: 614-621

Li N, Warren PV, Bryant DA, Golbeck JH (1991) PsaD is required for the stable binding of PsaC to the photosystem I core protein of Syneckococcus sp. PCC 6301. Biochemistry 30: 7863-7872

Liebl U, Mockensturm-Wilson M, Trost JF, Brune DC, Blanken- ship RE, Vermaas W (1993) Single core polypeptide in the reaction centre of the photosynthetic bacterium Heliobacillus mo- bilis: structural implications and relations to other photosystems. Proc Natl Acad Sci USA 90: 7124-7128

Mannan RM, Pakrasi HB, Sonoike K (1994) The PsaC protein is necessary for stable association of the PsaD, PsaE and PsaL proteins in the photosystem I complex: analysis of a cyanobac- teria1 mutant strain. Arch Biochem Biophys 315: 68-73

Naver H, Scott MP, Andersen B, Moller BL, Scheller HV (1995) Reconstitution of barley photosystem I reveals that the N-terminus of the PSI-D subunit is essential for tight binding of PSI-C. Physiol Plant 9 5 19-26

Naver H, Scott MP, Golbeck JH, Meller BL, Scheller HV (1996) Reconstitution of barley photosystem I with modified PSI-C allows identification of domains interacting with PSI-D and PSI-A/B. J Biol Chem 271: 8996-9001

Nielsen VS, Mant A, Knoetzel J, Moller BL, Robinson C (1994) Import of barley photosystem I subunit N into the thylakoid lumen is mediated by a bipartite presequence lacking an inter- mediate processing site. J Biol Chem 269: 3762-3766

Oh-oka H, Takahashi Y, Matsubara H (1989) Topological con- siderations of the 9-kDa polypeptide which contains centers A and B, associated with the 14- and 19-kDa polypeptides in the photosystem I complex of spinach. Plant Cell Physiol 30:

Pilsson LO, Tjus SE, Andersson B, Gillbro T (1995) Ultrafast energy transfer dynamics resolved in isolated spinach light- harvesting complex I and the LHC 1-730 subpopulation. Biochim Biophys Acta 1230: 1-9

Peter GF, Thornber JP (1991) Biochemical composition and organization of higher plant photosystem I1 light-harvesting pigment proteins. J Biol Chem 266: 16745-16754

Rousseau F, Setif P, Lagoutte B (1993) Evidence for the involvement of PSI-E subunit in the reduction of ferredoxin by photosystem I. EMBO J 12: 1755-1765

Scheller HV, Hej PB, Svendsen I, Meller BL (1988) Partia1 amino acid sequences of two nuclear-encoded photosystem I polypeptides from barley. Biochim Biophys Acta 922: 501-505

279-286

pp 405-408

869-875



420 Jansson e t al. Plant Physiol. Vol. 11 2 , 1996

Scheller HV, Meller BL (1990) Photosystem I polypeptides. Physiol Plant 78: 484-494

Scheller HV, Okkels JS, H0j PB, Svendsen I, Roepstorff P, Meller BL (1989a) The primary structure of a 4.0-kDa photosystem I polypeptide encoded by the chloroplast psaI gene. J Biol Chem 264 18402-18406

Scheller HV, Svendsen I, Msller BL (1989b) Subunit composition of photosystem I and identification of center X as a [4Fe-4S] iron-sulfur cluster. J Biol Chem 264 6929-6934

Schubert WD, Klukas O, Krauss N, Saenger W, Fromme E', Witt HT (1995) Present state of the crystal structure analysis of photosystem I at 4.5 A resolution. In P Mathis, ed, Photosynthesis: From Light to Biosphaere, Vol 11. Kluwer Academic Publishers, Dordrecht, The Netherlands, pp 3-10

Sigrist M, Staehelin LA (1994) Appearance of type 1, 2 and 3 LHCII and LHCI proteins during light-induced greening of barley (Hordeum vulgare) etioplasts. Plant Physiol 104: 135-145

Vermaas WFJ (1994) Evolution of heliobacteria: implications for photosynthetic reaction centre complexes. Photosynth Res 41:

Wang K, Richards FM (1975) Reaction of dimethyL3,3'-dithiobis- propionimidate with intact human erythrocytes. J Biol Chem

285-294

150: 6622-6626

Wray W, Boulikas T, Wray VP, Hancock R (1981) Silver staining of proteins in polyacrylamide gels. Ana1 Biochem 155: 83-88

Wynn RM, Malkin R (1988) Interaction of plastocyanin with photosystem I: a chemical cross-linking study of the polypeptide that binds plastocyanin. Biochemistry 27: 5863-5869

Xu Q, Armbrust TS, Guikema JA, Chitnis PR (1994) Organization of photosystem I polypeptides. A structural interaction between the PsaD and PsaL subunits. Plant Physiol 106: 1057-1063

Xu Q, Hoppe D, Chitnis VP, Odom WR, Guikema JA, Chitnis PR (1995) Mutational analysis of photosystem I polypeptides in the cyanobacterium Synechocystis sp. PCC 6803: targeted inactivation of psal reveals the function of PsaI in the structural organization of PsaL. J Biol Chem 270: 16243-16250

Yu L, Zhao J, Miihlenhoff U, Bryant DA, Golbeck JH (1993) PsaE is required for in vivo cyclic electron flow around photosystem I in the cyanobacterium Synechococcus sp. PCC 7002. Plant Physiol103 171-180

Zanetti G, Merati G (1987) lnteraction between photosystem'1 and ferredoxin. Identification by chemical cross-linking of the polypeptide which binds ferredoxin. Eur J Biochem 169:

Zilber AL, Malkin R (1988) Ferredoxin cross-links to a 22 kD 143-146

subunit of photosystem I. Plant Physiol 88: 810-814



Nearest-Neighbor Analysis of Higher-Plant … Physiol. (1 996) 11 2: 409-420 Nearest-Neighbor...

Documents

Transcript of Nearest-Neighbor Analysis of Higher-Plant … Physiol. (1 996) 11 2: 409-420 Nearest-Neighbor...