Subunit Stoichiometry of the Chloroplast Photosystem 11 Antenna System and Aggregation State of the Component Chlorophyll u/b Binding Proteins*

Photosystem (PS) I1 membranes, obtained by the method of Berthold et al. (Berthold, D. A., Babcock, G. T., and Yocum, C. F. (1981) FEBS Lett. 134, 231- 234), have been fractionated by a sucrose gradient ultracentrifugation method which allows the quantitative separation of the three major chlorophyll bind- ing complexes in these membranes: the chlorophyll (chl) a binding PSII reaction center core, the major light-harvesting complex 11, and the minor chl alb proteins called CP26, CP29, and CP24. Each fraction has been analyzed for its subunit stoichiometry by quanti- tative sodium dodeeyl sulfate-polyacrylamide gel elec-trophoresis methods. The results show that 12 mol of light-harvesting complex I1 and 1.5 mol of each of the minor chl alb proteins are present per mol of the PSII reaction center complex in PSII membranes. These data suggest a dimeric organization of PSII, in agree- ment with a recent crystallographic study (Bassi, R., Ghiretti Magaldi, A,, Tognon, G., Giacometti, G. M., and Miller, K. (1989) Eur. J. Cell Biol. 50,84-93) and imply that such a dimeric complex is served by antenna chl a/b proteins whose minimal aggregation state includes three polypeptides. This was confirmed by

In oxygenic photosynthesis, two photochemical systems cooperate in the transfer of electrons from water to NADP+. In both cases the light reactions are driven by the excitation energy absorbed by the antenna pigments and transferred to the reaction centers (RC).' Although the primary chromophore responsible for light absorption in higher plants and algae is chlorophyll a, accessory pigments such as chlorophyll b and carotenoids, which are bound to antenna proteins, extend the spectral range of light absorption and transfer energy to chl a.
The PSII core complex is composed of a pigment-protein complex which contains 4-5 chlorophyll a molecules, including the photoactive pigment P680, bound to the Dl and D2 polypeptides (3) and of two partially homologous chlorophyllproteins, CP47 and CP43, each binding 25 chl a molecules (4). Although these two proteins, mainly CP47, have been indicated as the site of P680, they are now thought to have an inner antenna function (for a review see Ref. 5). Although the above mentioned proteins are coded by the chloroplast genome, the remaining 180 chlorophyll molecules are bound to a number of chl alb binding proteins coded by an extended multigene family in the nuclear genome (6)(7)(8). Three of them, called CP29, CP26, and CP24, are present in small amounts in PSII membranes, whereas a fourth component, the major light-harvesting complex I1 (LHCII), is the most abundant thylakoid protein (for a review see Ref. 9).
Previous work with higher plants has shown that CP43 is the PSII core subunit which mediates the binding of both LHCII and the minor chl a/b proteins and topological models have been proposed for the organization of chlorophyll a/b proteins (10,11). The PSII antenna system is an highly complex structure capable of many physiological mechanisms for the adaptation of the light-harvesting function to the environmental conditions such as state I-state I1 transitions (12), heat stress (13), and cold stress (14). The understanding of the molecular mechanisms underlying these physiological adaptations requires more information about the composition and supramolecular organization of the antenna system.
Several basic aspects of the system are, however, unknown. As an example, information about the relative abundance of the components of the antenna system as well as their aggregation state and their pigment complement i s very imprecise. In this study, we have determined the stoichiometry of the different chl a/b proteins in PSII membranes, as well as the number of chlorophyll molecules they bind and their aggregation state. On the basis of these data and consistent with information obtained from the analysis of two-dimensional crystals of the PSII complex (15), we propose a dimeric organization for PSII reaction center. Our data indicate that each dimer is served by a common antenna system which is composed of trimeric chl a/b proteins.

MATERIALS AND METHODS
Preparation of Thylakoid Membranes-Zea mays seedlings (cv Dekalb DF280) were grown for 2-3 weeks in a growth chamber at 28/ 21 "C day/night at a light intensity of 10,000 lux and 80% humidity.
Leaves from 2-to 3-week-old plants were harvested at the end of a 12 h dark period and thylakoids from mesophyll chloroplasts were obtained as described previously (10). PSII membranes were obtained according to the method of Herthold et al. (16) using the modifications described in (17). Aliquots were resuspended in 25 mM Hepes, pH 7.6. 5 mM Mg CI,, 10 mM NaCI, 0.2 M sorbitol, and frozen at -80 "c until required.
Sucrose Cradient Ultrarrntrifugntion-PSI1 memhranes were washed twice in 1 mM EDTA pH 8.0, then resuspended in 1% DM. Solubilized membranes were spun 2' a t 15,000 X g at 4 "C and rapidly loaded onto a 0.1-1 M sucrose gradient containing 10 mM Hepes, pH 7.6, and 0.06% DM. The gradient was then spun on a Reckman SW41 rotor a t 39,000 rpm for 23 h at 4 "C. For quantitative determinations, the gradient was fractionated from the top into 250-pl aliquots that were analyzed hy SDS-PAGE, absorption spectra, and for their protein and pigment content. Alternatively individual green bands were harvested with a syringe.
SDS-PAGEand Immunohlotting-Analytical SDS-PAGE was performed with gradient gels (12-18rh acrylamide, 350 X 350 X 1 mm) containing 6 M urea and run at 10 mA for 3 days using the huffer system described previously (10). Alternatively, a high Tris buffer system without urea (12-18% acrylamide gradient) was used (18). For immunohlot assay, samples were separated hy one of the gel systems described ahove and transferred to a nitrocellulose filter (Millipore). The filters were then assayed with antihodies and antibody binding was detected by using alkaline phosphatase-coupled anti-rabbit IgG (Sigma). Antibodies were prepared in rahbits and characterized as described previouly (19).
Purification of I'roteins-Chlorophyll a/h proteins were purified from DM-solubilized PSI1 memhranes by preparative IEF in the pH range 3.5-5 as described previously (20). OEEl and OEE2 were also purified by preparative IEF hut a pH gradient from 3.5 to 10 was used.
Amino Acid Ana1.vsi.q-Solutions containing purified proteins were analyzed hy quantitative aminoacid assay. Proteins were hydrolyzed in 6 M HCI for 24 h a t 110 "C. Samples were then vacuum-dried and treated with phenylisothiocyanate to obtain phenylthiocarhamoyl amino acids following the Pico-Tag procedure (Millipore-Waters) as suggested by the manufacturer. Phenylthiocarbamoyl amino acids were then fractionated by HPLC by using a reverse phase C18 column (Pico-Tag, Waters) a t 38 "C. Eluents were A: 100% sodium acetate. 140 mM, pH 4.6, containing 0.5 ml of triethilamine and 40 ml of acetonitrile/liter and R: 60% acetonitrile. Eluents A and H were used in a linear gradient from 100% A to 35% R. Phenylthiocarhamoyl amino acids were revealed and quantified by their absorbance at 269 nm. Values of threonine and serine were corrected assuming, respectively, 13 and 17% decomposition during the hydrolysis. A known amount (1000 pmol) of norleucine was added to the protein solution as concentration reference.
Quontitation of Coomassie-stained Protrins in SI1.S-PAGE Gc1.v-Samples from sucrose gradient bands were analyzed hy both the SDS-PAGE methods descrihed ahove. After running, the gel was fixed in 5 5 1 methano1:water:acetic acid including 10rE trichloroacetic acid, stained in 0.125% Coomassie Brilliant Blue R-250 in 40% methanol overnight. The gels were destained twice in 10% methanol and 10% acetic acid for 8 h and then in 7.55 acetic acid. Gel pieces containing a stained protein band were excised with a razor hlade. weighted. and placed in an Eppendorf tuhe with 1 ml of 3% SDS in 50% isopropanol. Gel pieces with the same weight, cut from the interlane areas, were used for background suhtraction. The eluted dye was quantified spectrophotometrically at 595 nm (21). The specific hinding of Coomassie to individual proteins (CP29, CP26, C1'24. LHCII, OEE1, and OEE2) was estimated by loading increasing am0unt.s of purified proteins on a SDS-PAGE gel and determining the bound dye as descrihed ahove. Alternatively, Coomassie binding was measured by densitometry hy using a Shimadzu chromatoscanner with a program allowing base-line correction. The two methods gave consistent results, although we found it necessary to stain the gels overnight rather than for 2 h as reported in Hall (21).
Cross-linking-For cross-linking experiments, chlorophyll-proteins were isolated as reported ahove ancl tested for their aggregation state hy sucrose gradient ultracentrifugation in 0.06% DM. Green hands from the sucrnse gradient were diluted to 10 pg cbl/ml in 10 mM Hepes, pH 7.6, 0.03"; DM and treated with lr'; glutaraldehyde for 5' at 20 "C. The reaction was blocked by addition of 0.025 volumes of 2 M NnRH, dissolved in 0.1 M NaOH; this reduces and stabilizes the cross-linked products, and unreacted glutaraldehyde is inactivated by reduction. The samples were analyzed for their migration in sucrose gradient orrsus unreacted samples to ensure that higher aggregation states were not induced hy the treatment. Cold acetone was then added to 80rE and the protein precipitated hy centrifugation at 15,000 X R for 5'. The pellet was dried and redissolved in 10"; SDS. 1"; mercaptoethanol. 0.125 M Tris sulfate, pH 9.0. and analyzed by SDS-PA(;E (22). A hetter resolution of the high molecular mass cross-linking products was obtained by using long runs in SI)S-urea-PAGE as shown in Fig. 6.
Othrr Mrfhods--Chlorophvlls and carotenoid determination was made in 80r; aretone hy using the equations described hy IVellhurn and Lichtenthaler ( 2 3 ) . See this reference for limitation in the determination of carotenoids. Many experiments were also made t~y using the equations of l'orra r t al. (24). The two methods yielded very similar results for rhl a to chl h ratios ranging from I to 3.
Ahsorption spertra of chlorophyll a / h prnteins were taken in 1 0 mM Hepes, pH 7.6, 0 . O ( j r ; I)M hv using a Perkin-Elmer I,amhda S spectrophotometer. Slit width was 1 nm. Protein determination in solution was hy the hirinchoninir acid method (2SJ.

RESlILTS
Fractionation of P S I 1 Mrmhranrs by Sucrosr (;mdicwt 1'1-tracrn~rifujiation-Photosystem I1 rnemhranes ohtained hy the procedure of Rerthold rt nl. (16) were soluhilized with I r ; DM and fractionated in a 0.1-1 M sucrose gradient containing 0.06% DM. The separation pattern is shown in Fig. I A , whereas the ahsorption spectra of the major hands are shown in Fig. 1H. The polypeptide composition of the green hands is shown in Fig. 1C.'. Since the apparent molecular mass of the chlorophyll a / h binding apoproteins is very similar, we have assayed blots with antihodies raised against purified complexes as described previously (Dainese rt 01.. 1990) to identify

FIG. 2. Polypeptide composition (A) and immunoblot analysis ( B ) of chl a containing sucrose gradient bands (5 to 7).
apoproteins (Fig. 1D). Sucrose band 1 (yellowish-green from free carotenoids) contained only the colorless OEE 1 (33 kDa) polypeptide. The chl a/b proteins were found migrated in bands 2 to 4 as follows: the minor chl a/b proteins CP29, CP26, and CP24 were found in sucrose band 2, whereas the major LHCII complex migrated in band 3. The antibody against CP29 recognizes a polypeptide band in sucrose bands 2 and 4. Band 4 also contained CP24 and LHCII polypeptides with molecular mass of 29.5,29, and 26 kDa. These appear to form a supramolecular complex whose mobility in the sucrose gradient is greater than each of the component chlorophyllproteins. Interestingly, only a subset of the LHCII polypeptides (molecular masses 26, 29, and 29.5) was present in band 4, suggesting that a specific LHCII subpopulation is bound to CP29 and CP24. A more detailed analysis of this supramolecular complex will be reported in another paper.* We will restrict ourselves to the quantitation of its constituent components.
In the best separations, sucrose band 2 is almost free from LHCII contamination as shown in Fig. lC, whereas in other cases variable amounts of LHCII were found as indicated in Table I. Sucrose bands 5-7 contained chl a and no chl b, associated with the PSII core complex as shown by the polypeptide composition (Fig. 2 A ) and immunoblot analysis (Fig.  2B). The six intrinsic membrane polypeptides of 47, 43, 34, 32,9, and 4 kDa were found in sucrose bands 5-7, whereas all the three extrinsic OEE proteins were present only in the minor band (band 6). Minor contamination by the PSI-LHCI complex, if present in the PSII membrane preparation, comigrated in sucrose band 6. The PSII preparation in sucrose band 5 contained OEEl(33 kDa) and OEE2 (23 kDa), whereas the complex in band 7, which was the most abundant form of the PSII core complex (approximately 70%), bound OEE3 and OEE2 but not OEE1.
The characteristics of the green bands above described are summarized in Table I, which shows that 23% of the chlorophyll and 31% of the protein content of grana membranes are due to the chl Q reaction center core complex, whereas the remaining 77% (69% in protein) is due to chl a/b antenna complexes.
Stoichiometry of Chlorophyll a/b Binding Proteins-To determine the quantitative relations between chl a/b binding proteins, we analyzed sucrose gradient bands 2-4 for their polypeptide composition by using two different SDS-PAGE gel systems to avoid comigration. The individual Coomassiestained bands were identified by immunoblot with specific antibodies and quantified either by densitometry or by excising the bands and measuring the eluted dye (21). As an example, the results obtained by the analysis of sucrose band 4 are shown in Fig. 3. The specific binding of Coomassie R-250 to individual chl a/b apoproteins as well as the two major extrinsic OEE polypeptide which are often contaminants of sucrose bands 1-4 was evaluated by loading different amounts of purified proteins obtained by preparative isoelectrofocusing on gels and determining the amount of bound dye. The results obtained by the analysis of sucrose band 4 are shown in Fig.  3 as an example. There were differences in the stainability of Cab proteins as shown in Fig. 4, and the data were corrected according to the specific binding of Coomassie to isolated proteins. The same samples were also analyzed for their chlorophyll and carotenoid content and the pigment to protein ratios are reported in Table 11. These values were obtained with complexes purified by a single IEF step. Alternative procedures including several purification steps or mild SDS-PAGE consistently yielded lower chlorophyll to protein ratios while chl a/b ratios were higher for CP29 and CP26 and lower for CP24 and LHCII. The results of quantifying Cab apoproteins are summarized in Table I11 4. Quantification of the Coomassie binding to individual chl a/b apoproteins. Different amounts of purified chl alb proteins were run on a SDS-urea gel that was stained with Coomassie R-250. Individual bands were excised from the gel and the bound dye was eluted and spectrophotometrically quantified. branes in a ratio approaching 1:l:l and together constitute 25% of the chl a/b proteins. The data in Tables I and I11 can be used to calculate stoichiometric data for all chlorophyll binding proteins of PSII membranes (Table IV). In these calculations we have considered the PSII-RC core complex as a single chlorophyll binding protein having a molecular mass of 240 kDa based on the sum of the component apoproteins. On the basis of the determined chl to protein ratio, the numbers of chlorophyll molecules were also calculated for individual chlorophyll-proteins. The values obtained were consistent with the experimental data for the chl to protein ratio within the individual sucrose gradient bands (data not, shown). We calculate that 53 chl a molecules are bound to the PSII-RC core complex, 131 to the major LHCII, and 15, 14, and 8, respectively, to CP29, CP26, and CP24. In terms of protein stoichiometry, we obtain 12 LHCII moles and approx- , . . , , t J u u u t r " " " Aggregation State of Chlorophyll a/b Proteins-From the stoichiometric determinations reported above, the minor chl a/b proteins are present in PSII membranes in a fractional ratio with respect to the PSII-RC complex. It is therefore important to measure the aggregation state of chl a/b proteins in the antenna system. With this in mind, we compared the mobility of isolated chlorophyll a/b proteins in sucrose gradient. Consistent with the separation shown in Fig. 1 and with previous studies, the major LHCII complex yielded two bands: the major one corresponding to sucrose band 3 (Fig. 1) and a variable amount of the chlorophyll, depending on the detergent concentrations, migrated more slowly, corresponding in mobility to sucrose band 2 (Fig. 5). The purified minor chl a/b proteins migrated as single bands in sucrose gradients corresponding in mobility to sucrose band 2. Diluted aliquots (10 Kg chl/ml) of CP24, CP26, CP29, and of the two aggregation states of LHCII were treated with the cross-linking agent glutaraldehyde and analyzed by SDS-PAGE. Before SDS-PAGE, aliquots were analyzed by sucrose gradient ultracentrifugation to verify that no changes in sedimentation behavior were induced by the treatment. The results are shown in Fig. 6; besides the uncross-linked material migrating at around 30 kDa, two major bands were obtained at approximately 50 and 65 kDa in all the samples of LHCII (lane 2 in Fig. 6, A and B ) . The minor chl a/b proteins CP29, CP26, and CP24 yielded identical results as the LHCII aggregation form with lower mobility in sucrose gradient, i.e. two crosslinking products of 50 and 65 kDa were obtained. In addition to the 50-and 65-kDa bands, the higher aggregation state of LHCII (but not its lower aggregation form or the minor chl a/b proteins) also produced multiple fainter bands at higher M , after cross-linking with glutaraldehyde. We interpret the two major bands with molecular mass higher than that of the control sample as cross-linking products, although their apparent molecular mass is not exact multiple of 30 kDa. In fact it has been shown that the migration of cross-linked proteins may be perturbed by the increased compacteness of crosslinked products and decreased SDS binding (22). These re- sults are a clear demonstration that the lower aggregat.ion state of chl n/b proteins includes three polypeptides. This is common to the minor chl a/h proteins and to the lower aggregation state of LHCII, whereas a higher number of high molecular mass LHCII cross-linking products is present in the case of the LHCII preparation migrating in sucrose band 3. It is not possible from our data to determine the exact number of polypeptides in LHCII due to the increasingly diffuse appearence of the high molecular mass bands produced hy cross-linking. DISCUSSION We report here on the stoichiometry of chlorophyll binding proteins within the grana1 membranes of maize chloroplasts and on the aggregation state of polypeptides within purified chl a/b proteins.
Three methods have been particularly important in this work: (i) the development of a sucrose gradient fractionation method which allows the separation of the minor chl a/b proteins, the major LHCII complex and of the chl a hinding PSII-RC core complex; (ii) the use of specific antibodies to identify closely migrating polypeptides in SDS-PAGE gels; and (iii) the use of a recently developed nondenaturing IEF method that allows the purification of undenatured chl binding proteins (20). When PSI1 membranes are separated by SDS-PAGE, the apparent molecular mass of chl a / b apoproteins is very similar, ranging from 20 and 31 kDa. I t has therefore been necessary to fractionate the DM-soluhilized membranes by sucrose gradient ultracentrifugation. This is a very mild and effective procedure, since less than 2% of the pigments are released and allows the separation of the major LHCII and t.he minor chl a/b proteins by a simple procedure. The conditions of solubilization are very critical and can be used to modulate the results of the separation; short incubation time with 1% D M yields a very clean preparation of the minor chl a/h proteins in sucrose band 2, the same procedure with 0.4% DM yields essentially CI'26 in hand 2, whereas CP29 and CP24 migrate in band 4 with a fraction of LHCII. Longer incuha-tions in the detergent 0 1 5 ' ) decrease the amount of chl associated with sucrose band 4 but also increase the contamination of sucrose hand 2 with dissociated 1,HCII. We have determined the distrihution of the chl among chlorophyllproteins in each of the ahove conditions obtaining consistent results.
The determination o f suhunit stoichiometry has been mostly carried out with aquatic organisms such as ('hlarnydomonas rpinhnrdtii ( 2 6 ) or Ixmnn minor (27) in which uniformly radioactively laheled proteins can he ohtained by supplying ',IC substrates in the growing solution. This is practical since the method does not require the isolation o f each protein suhunit. Since this procedure is not feasihle with X. rnoys, we have used the Coomassie hinding for protein quantitation. The hinding of dyes to proteins is variahle and, in principle, the possibility that the stainahility o f different apoproteins is very different cannot he ruled out. We have assayed purified proteins and determined in each case the specific binding o f the dye, whereas absolute protein concentration in the purified protein solution was assyed hy quantitative amino acid analysis. The measured values are very similar for the different chl a/b hinding proteins within either the 1,HCII group and the minor chl a / h protein group as could be expected from their belonging to the same multigene family anti sharing of epitopes (19,28). Although the presence of common epitopes could be a prohlem in the identification of the apoproteins hy immunohlot analysis, we have shown that only minor crossreaction could he detected when native chlorophyll-proteins rather than denatured apoproteins are used as antigens (19). At the dilution of antisera used in this work, the apoproteins could he easily distinguished. Only the antihody against CP26 cross-reacted with CP29, but the apoproteins could be easily resolved by SDS-urea-PAGE. The isolation of pure chlorophyll-proteins allowed the determination of the stoichiometry of pigment hinding to the different chl a / b proteins. The value we obtained for LHCII (12 chl and 2 carotenoids for 28.5-kDa protein) is in close agreement with the most recent determinations (29, 30). whereas for CP29, CP26. and C1'24 we obtained a higher pigment content ranging from 9 to 5 chl molecules for each polypeptide. Chlorophyll has been shown to he bound to the hydrophohic portion of LHCII huried in the thylakoid memhrane (29). Although CP26 and CP29 genes have not been cloned yet, the deduced amino acid sequence for CP24 (31) predicts two trans-memhrane tr-helices rather than the three indicated for LHCII, consistent with the lower pigment content. To our knowledge, quantitative determination of pigment hinding to CP24 has not been reported hefore, whereas CP29 and CP26 have been reported to hind 4 and 5 chl molecules with a chl n to h ratio of 4.8 and 2.7, respectively (32). We have obtained lower pigment content than those reported in Tahle I1 in the case of preparations obtained hv multistep procedures (see "Sucrose Gradient Ultracentrifugation" under "Materials and Methods"). We therefore suggest that the procedure employed hy these authors. which includes solubilization with octyl /j-D-glucoside in high salt conditions, results in partial removal of pigments from CI '26 and CP29. To generate stoichiometry data, the values for protein cont.ent (Tahle 111) have heen divided hy the molecular mass of each chl n/h binding suhunit and corrected for specific Coomassie binding. The correction for molecular mass requires several assumptions relating to the molecular mass o f these proteins. Molecular masses of membrane proteins hased on SDS-PAGE can he very anomalous. In an extreme case, the L suhunit of the reaction center of Rhodr'p.scud"rn"nas capsulatn, sequence data has shown the SI)S-I'A(;E molecular mass value to be underestimated by over 50% (33). However, no molecular mass values obtained by deduced amino acid sequences are available for maize chl a/b proteins with the exception of a single c-DNA clone (34) which it is not known to correspond to a particular LHCII apoprotein. We have therefore used for the correction molecular mass values obtained from SDS, 6 M urea-PAGE (lo), since the more complete denaturation obtained in these conditions is more likely to favor stoichiometric SDS binding to the unfolded polypeptide and minimize anomalies in migration patterns. In any case, it is likely that such anomalies are similar for all chl a/ b proteins, since they belong to the same multigene family and share sequence stretches and secondary structure patterns. It should be remembered, moreover, that quantitation of apoproteins from SDS-PAGE has been used in this work only to assess the relative contributions of the different proteins within each sucrose gradient band, while the absolute protein content was determined in solution by the bicinchoninic acid method (25) which does not have the limitations of the SDS-PAGE/Coomassie stain method and, when required, confirmed by quantitative aminoacid analysis.
A more critical assumption had to be made in the generation of Table IV in which a molecular mass value of 240 kDa has been attributed to the chl a binding PSII-RC core complex by the summation of the molecular masses of the component subunits CP47, CP43, D l , D2, cytochrome b559, OEEl,OEE2, OEE3, and 9-kDa phosphoprotein. This value is commonly accepted (35, 36), and when employed in our calculations yields a pigment complement of 53 chlorophyll molecules for the PSII core complex, a value that is consistent with reported data (3,4,35).
To our knowledge there has been no attempt to determine the stoichiometry of chlorophyll binding proteins in the PSII antenna system. We have obtained a value of 12 mol of the major LHCII complex and 1.5 mol of each of the three minor chl a/b proteins CP29, CP26, and CP24/mol of PSII-RC core complex. If we attribute to each chlorophyll-protein the pigment complement determined (Table 11), then a value of 232 chl molecules/PSII reaction center is obtained, which compares favorably with the spectrophotometric determination of 230 chlorophylls for the PSII antenna size in isolated grana membranes (2). Since proteins of the antenna system bind chl a and chl b in ratios ranging from 1.4 to m, a comparison can be made with chl a/b ratio values determined in isolated complexes. A calculated chl a/b ratio of 2.28 is thus obtained for PSII membranes uersus a measured value of 2.3 k 0.05 by using the same determination method for the membranes and isolated complexes (23).
We assume here that all the PSII complexes in the membrane preparation used have all of the component polypeptides and polypeptide-chlorophyll complexes in the same ratio. This cannot be proven at the present; however, several data support the assumption: ( a ) PSII has been shown to be heterogeneous with respect to the antenna size and the characteristics of the donor side, but the two PSII types (cy and (I) were found to be located in different membrane compartments. Thus grana membranes, that we have used in this work, contain mainly PSIIa, whereas PSIIP was found into stroma exposed thylakoids (37); ( b ) the chlorophyll a PSII core complex appears to be very homogeneous, since 80% of it is contained in sucrose band 7, whereas sucrose band 5 is likely to be a dissociation product of the complex in band 7; (c) the chlorophyll a/b proteins CP24, CP29, and a LHCII subset can be isolated in conditions which preserve their complete association in the supramolecular complex of su-crose band 4,3 thus showing that the organization of these antenna complexes is homogeneous in PSII membranes. The molar ratio of 1.5 obtained for each of the three minor chl a/b proteins and the PSII-RC complex is intriguing, since both whole numbers 1 and 2 are too far from the experimental values to be preferred with respect to 1.5. On this basis, we suggest that PSII-RC is organized in dimers which are served by a common antenna system composed of 24 LHCII polypeptides and three each of the CP29, CP26, and CP24 proteins. Three lines of evidence support our hypothesis: (i) in our sucrose gradient fractionation (Fig. 1) two bands (5 and 7) have PSII-RC complexes containing all the intrinsic membrane subunits of PSII. The complex in band 7 could be an oligomeric (probably dimeric) form of the one in band 5. In most conditions the putative dimeric form is predominant (by 80%) and increases to 100% when the solubilization is carried on in 5 mM MgC12 (37).
(ii) Image analysis of two-dimensional crystals of PSII-RC (15) show a dimeric structure for the complex. (iii) Electrophoretic and ultracentrifugation studies indicate that LHCII can be present mainly in two aggregation states whose apparent molecular masses differ by a factor of three (30, 38). Minor chl a/b proteins show the same sedimentation rate as the lower molecular mass form of LHCII (20). Although some authors (30,38) have suggested that the two aggregation states correspond, respectively, to monomer and trimer, in our opinion this has not been proven. In this paper we show by covalent cross-linking that the lower aggregation state of LHCII as well as the three minor chl a/b proteins, CP29, CP26, and CP24, are in fact trimers, whereas the higher aggregation form of LHCII exhibits an even higher oligomeric state. Although our experiments with a cross-linking agent can hardly indicate the number of LHCII polypeptides present in the latter case, we propose that this number is nine based on ultracentrifugation (30), electrophoretic, and spectroscopical data (39). Our results are consistent with previous structural studies with light-harvesting proteins that have been shown to have a trimeric symmetry such as the bacteriochlorophyll-protein complex of Prosthecochlork aestuarii (40), the c-phycocyanine of Mastigocladus laminosus (41), and the purple membrane of Halobacterium halobium (42). The trimeric symmetry appears to be essential for excitation energy delocalization within the antenna protein and therefore for the efficiency of light-harvesting function (43).