Full Subunit Coverage Liquid Chromatography Electrospray Ionization Mass Spectrometry (LCMS+) of an Oligomeric Membrane Protein

Highly active cytochrome b6f complexes from spinach and the cyanobacterium Mastigocladus laminosus have been analyzed by liquid chromatography with electrospray ionization mass spectrometry (LCMS+). Both size-exclusion and reverse-phase separations were used to separate protein subunits allowing measurement of their molecular masses to an accuracy exceeding 0.01% (±3 Da at 30,000 Da). The products of petA, petB, petC, petD, petG, petL, petM, and petN were detected in complexes from both spinach and M. laminosus, while the spinach complex also contained ferredoxin-NADP+ oxidoreductase (Zhang, H., Whitelegge, J. P., and Cramer, W. A. (2001) Flavonucleotide:ferredoxin reductase is a subunit of the plant cytochrome b6f complex. J. Biol. Chem. 276, 38159–38165). While the measured masses of PetC and PetD (18935.8 and 17311.8 Da, respectively) from spinach are consistent with the published primary structure, the measured masses of cytochrome f (31934.7 Da, PetA) and cytochrome b (24886.9 Da, PetB) modestly deviate from values calculated based upon genomic sequence and known post-translational modifications. The low molecular weight protein subunits have been sequenced using tandem mass spectrometry (MSMS) without prior cleavage. Sequences derived from the MSMS spectra of these intact membrane proteins in the range of 3.2–4.2 kDa were compared with translations of genomic DNA sequence where available. Products of the spinach chloroplast genome, PetG, PetL, and PetN, all retained their initiating formylmethionine, while the nuclear encoded PetM was cleaved after import from the cytoplasm. While the sequences of PetG and PetN revealed no discrepancy with translations of the spinach chloroplast genome, Phe was detected at position 2 of PetL. The spinach chloroplast genome reports a codon for Ser at position 2 implying the presence of a DNA sequencing error or a previously undiscovered RNA editing event. Clearly, complete annotation of genomic data requires detailed expression measurements of primary structure by mass spectrometry. Full subunit coverage of an oligomeric intrinsic membrane protein complex by LCMS+ presents a new facet to intact mass proteomics.

ing the presence of a DNA sequencing error or a previously undiscovered RNA editing event. Clearly, complete annotation of genomic data requires detailed expression measurements of primary structure by mass spectrome-

try. Full subunit coverage of an oligomeric intrinsic membrane protein complex by LCMS؉ presents a new facet to intact mass proteomics. Molecular & Cellular Proteomics 1:816 -827, 2002.
Mass spectrometry (MS) 1 has revolutionized the biological sciences since the development of matrix-assisted laser desorption ionization (MALDI) (1) and electrospray ionization (ESI) (2) in the late eighties (3), fertilizing the emergence of a new discipline called proteomics. Biological macromolecules are now mass-measured with great accuracy, and highly resolved spectra reveal subtle molecular heterogeneity. In fact, a mass spectrum of an intact protein is an essential piece of proteomic information that defines the native covalent profile of the product of a gene and its associated heterogeneity (4). For intact proteins, ESI-MS provides superior accuracy (within 0.01% error) and resolution, while MALDI is more sensitive and more tolerant of extreme heterogeneity and complex mixtures. Although intrinsic membrane proteins have been traditionally regarded as problematic, a suite of techniques has been developed that allows routine ESI-MS of numerous examples with up to 15 transmembrane helices (4 -8). Although proteomics has largely been fueled by utilization of mass spectrometry to identify proteins, based upon mass and sequence analysis of small peptides derived from the parent protein, it is now possible to include intact mass measurements as an integral part of proteomics (intact mass proteomics).
An isolated fully active protein complex is especially useful in proteomics because it defines the functional unit and the structural genes that interact to provide this functionality, although other gene products may of course be involved for successful assembly and regulation. Consequently, intrinsic membrane protein complexes have been previously targeted for analysis by mass spectrometry with some success. MALDI was used successfully for analysis of cytochrome bo 3 (4 subunits) and cytochrome bd (2 subunits) from Escherichia coli as well as the bc 1 complex (3 subunits) and cytochrome c oxidase (3 subunits) from Rhodobacter sphaeroides (9). However, lower accuracy and resolution in MALDI of larger proteins lessens the value of these analyses compared with ESI, although it is noted that some membrane proteins have recently been analyzed with improved accuracy by MALDI (10). ESI interfaced to HPLC (LCMS) was applied to bovine cytochrome bc 1 complex, but not all subunits were detected (11). Photosystem 2 (PS2) reaction-center subcomplexes have been successfully analyzed by ESI (4,(12)(13)(14) as have appressed thylakoid membrane subfractions that include the PS2 dimer and associated light-harvesting complex polypeptides (8), although a systematic, comprehensive examination of the native PS2 dimer has yet to be described. Toward the goal of complete description of an oligomeric intrinsic membrane protein complex, a number of techniques were used to fully characterize highly active preparations of cytochrome b 6 f complex, including the cytochrome b 6 (PetB) and subunit IV (PetD) subunits that have a total of seven transmembrane helices.
The cytochrome b 6 f complex provides the electronic connection between the two reaction centers, photosystem 2 and 1, of oxygenic photosynthesis and, by oxidizing the lipophilic plastoquinol and transferring the resulting protons to the electrochemically positive side of the membrane, also contributes to the generation of the transmembrane proton electrochemical potential that drives ATP synthesis (15). This function is analogous to that carried out by the cytochrome bc 1 complex of bacterial photosynthesis and the respiratory chain with which there are many closely similar features, including the redox cofactors and the cytochrome b polypeptide in the membrane core of the complexes (16). The b 6 f complex contains at least four redox cofactors, the two b hemes, the heme of cytochrome f, and the high potential [2Fe-2S] cluster of the Rieske iron-sulfur protein and a fifth, the FAD moiety of the ferredoxin:NADP ϩ reductase in the b 6 f complex from spinach (17) and presumably other plant sources. The latter cofactor has no analogue in the cytochrome bc 1 complex. The b 6 f complex has been purified as a very active (Ͼ300 electrons/ cytochrome f/s) dimer with eight and nine polypeptide subunits and a molecular weight of ϳ210,000 and 280,000, respectively, in the cyanobacterium Mastigocladus laminosus and spinach chloroplasts (with FNR) (17). Twenty-six transmembrane helices are predicted in the dimeric complex.
Two-dimensional crystals of the b 6 f complex from the green alga Chlamydomonas reinhardtii have been obtained that provide maps to 8 -9 Å (18, 19), and three-dimensional crystals of the complex from the thermophilic cyanobacterium M. laminosus (20) have been obtained that diffract to Ͻ4 Å. The space group and unit cell dimensions have been determined. 2 High resolution (ϳ1.8 -1.9 Å) structures, which will facilitate the solving the structure of the complex, have previously been obtained of the lumen-side soluble domains of cytochrome f (21) and the Rieske iron-sulfur protein (22,23) that constitute ϳ40% of its total mass. The high resolution structure of the soluble form of cytochrome f, which turns out to be completely different from cytochrome c 1 and a unique c-type cytochrome (24), has been obtained from plant chloroplasts (21), cyanobacteria (23), and green algae (26).
As part of an effort to better understand the structure and evolution of the b 6 f complex a complete mass spectroscopic characterization of the b 6 f complex isolated from spinach thylakoids and M. laminosus has been performed. This is the first complete mass spectroscopic characterization of an active, multiply oligomeric, integral membrane protein complex from more than one phylum, providing precise information about the detailed nature of the complex that has helped to catalyze the successful crystallization of the b 6 f complex.

EXPERIMENTAL PROCEDURES
Preparation of the cytochrome b 6 f complex from spinach chloroplasts and the thermophilic cyanobacterium M. laminosus and quantitation of polypeptide, heme content, and activity of the complex have been described (17). The electron transfer activity, from reduced decyl-plastoquinol to plastocyanin oxidized by ferricyanide, of the complex isolated from both sources was the same within experimental error, 250 -350 electrons/cytochrome f/s, and the same as the activities measured in situ or in vivo (17). As noted previously, the absence of FNR in the cyanobacterial complex did not affect linear electron transfer (17).
Samples of isolated cytochrome b 6 f complex were analyzed by liquid chromatography with mass spectrometry and fraction collection (LCMSϩ). Protein (200 g) was precipitated with either chloroform/methanol or cold acetone. Precipitation at the interface of an aqueous chloroform/methanol phase separation (27) was as described previously (5). Precipitated proteins were recovered after removal of the aqueous phase and addition of methanol. Proteolipids (proteins with lipid-like properties) were isolated by recovery of the lower phase. Precipitated samples were dried at atmospheric pressure for 2 min (25°C) and dissolved in 60% formic acid (100 l) immediately prior to HPLC. Acetone precipitation was achieved by addition of 1 ml of 80% aqueous acetone (Ϫ20°C), vigorous mixing for 1 min, and incubation for 30 min (Ϫ20°C). Material was recovered by centrifugation (30 s at 10,000 ϫ g in a microcentrifuge).
LCMSϩ-Fractions were collected concomitant with ESI-MS using a liquid-flow splitter inserted between the HPLC detector and mass spectrometer. A fused silica capillary was used to transfer liquid to the ESI source (50 cm) or fraction collector (25 cm). Fractions were collected into microcentrifuge tubes at 1-min intervals. For cyanogen bromide (CNBr) cleavage one-tenth volume of CNBr solution (1 g/ml in acetonitrile) was added to each fraction, and, after a 3-h incubation in the dark at room temperature, the sample was dried by centrifugal evaporation (SpeedVac, Savant). Peptides were dissolved in small volumes (5 l) of 70% acetic acid for MALDI-TOF.
Electrospray Ionization Mass Spectrometry-ESI-MS was performed as described previously (4) using a triple quadrupole instrument (API III, Applied Biosystems). Data were processed using MacSpec 3.3, Hypermass, or BioMultiview 1.3.1 software (Applied Biosystems). Tandem mass spectrometry (MSMS) was performed under standard conditions with orifice voltage and collision gas (99.999% argon) thickness to achieve 20 -90% fragmentation of the parent ion. Spinach PetL and M. laminosus PetG and PetL sequences were obtained by analysis of fractions collected during reverse-phase LCMSϩ using nanospray ESI-MSMS on a quadrupole time-of-flight mass spectrometer (Q-Star Pulsar, Applied Biosystems, equipped with Protana source).
MALDI time-of-flight (MALDI-TOF) was performed using a high performace UV laser time-of-flight mass spectrometer (Voyager DE-STR, Applied Biosystems) operated in reflector mode with internal calibration where possible. 0.5 l of a matrix solution (10 mg/ml ␣-cyano-4-hydroxycinnamic acid in 0.1% trifluoroacetic acid, 70% acetonitrile) was mixed with 0.3 l of an HPLC fraction collected during LCMSϩ (stored at Ϫ20°C).

RESULTS
Organic solvents were used to extract lipid and pigment molecules and precipitate the polypeptide subunits of the complex. Precipitation at the interface of a chloroform/methanol/water phase separation (27) was effective, although PetN and PetL partitioned into the lower chloroform-enriched phase and can thus be regarded as proteolipids. Consequently, more recent analyses used precipitation in 80% acetone at Ϫ20°C. After delipidation with organic solvents, a high concentration of organic acid (70% acetic or 60% formic acid) was used to solubilize the proteins for immediate HPLC analysis.
Size-exclusion chromatography mass spectrometry provides a rapid means with which to analyze purified membrane proteins such as the 12 transmembrane domain lactose permease (5). Fig. 1 depicts the results when this technique was applied to the components of the cytochrome b 6 f complex. The absorbance and total ion chromatogram from mass spectrometry are shown for both spinach and M. laminosus (Fig. 1, A and B, respectively). The technique is very effective at separating protein components from more highly retained small molecule contaminants. Thus proteins elute first, providing a rapid overview (Ͻ15 min) of the components of the complex. The main limitation to the technique is chromato-graphic resolution such that larger components (15-35 kDa) largely co-elute, generating complex overlapping ESI spectra that must be reconstructed to a zero-charge mass spectrum by appropriate software, effectively limiting the number of individual components that may be resolved by SEC-MS. The products of petA-D, as well as FNR in the spinach preparation, were detected in the reconstructed zero-charge spectrum, and their presence was confirmed by identifying genuine, unique ions for individual polypeptides in the original mass spectrum (Fig. 1C). The small subunits are quite well chromatographically resolved from the larger ones, although residual larger subunits can still be observed in the mass spectrum ( Fig. 1D) and contribute to noise in the zero-charge reconstruction. Nevertheless, the four most prominent peaks in the reconstruction correspond to the four small subunits, confirmed by careful examination of the mass spectrum for the corresponding multiply charged ions (labeled in Fig. 1D m/z spectrum) and MALDI-TOF data (not shown). It is important to note that relative peak intensities are dependent upon abundance and ionization efficiency such that quantitation by this measurement is unreliable unless the latter is known. Cytochrome b has a quite low ionization efficiency and probably would have been regarded as insignificant in a blind analysis based upon SEC-MS alone. Consequently, analysis by a second technique, with greater chromatographic resolving power, was used for further definition of the complex.
Reverse-phase chromatography was combined with MS for a high resolution analysis of the complex (Fig. 2). While the chromatographic separation takes significantly longer, nearly all subunits are resolved in the experiment, allowing consideration of absorbance measurements for quantitation, although it should be noted that chromatographic elution efficiency 3 is not 100% for all subunits. The addition of isopropanol to buffer B dramatically enhances the elution efficiency of many integral membrane proteins as reported originally (28), providing a viable alternative, in the case of the b 6 f complex, to the solvent system containing 60% formic acid described previously (4). Most important was the ability to elute significant quantities of the cytochrome b polypeptide that transferred to the mobile phase at higher organic solvent concentrations close to the proteolipid PetN. The mass spectrum of cytochrome b illustrates the signal to noise (Fig. 2C). Zero-charge reconstructions are shown for all spinach subunits and display levels of heterogeneity typical for material stored for short periods at Ϫ80°C (Fig. 2D). M. laminosus material also showed similar levels of heterogeneity (cytochrome b is shown) and no other subunits of significant abundance (see total ion chromatogram shown in Fig. 2B). The masses of the nine chloroplast and eight cyanobacterial subunits measured with ESI-MS are shown in Table I. In some FIG. 1. Electrospray ionization mass spectra after size-exclusion chromatography (LCMSϩ) in aqueous organic solvents of subunits of cytochrome b 6 f complex from spinach (A) and M. laminosus (B). Subunits (100 g of protein) precipitated with organic solvent (see "Experimental Procedures") were dissolved in 60% formic acid prior to immediate injection onto a Super SW2000 size-exclusion column (4.6 ϫ 300 mm, Tosoh Biosep) equilibrated in chloroform/methanol/1% aqueous formic acid (4:4:1, v/v). Both A 280 nm (upper panel) and total ion chromatogram (lower panel) are shown with full scale representing 100% relative intensity in each case. Peak absorbance for A was 750 milliarbitrary units and for B was 630 milliarbitrary units. The sum of ion intensities over the complete mass range (600 -2300) is integrated in each scan (6 s) and plotted against time for the total ion chromatogram. Elution of protein components is complete by 15 min, and the later part of chromatogram B is omitted. Boxed C, spectra from region 1 correspond to large subunits. Boxed D, spectra from region 2 are dominated by small subunits. The y axes in all cases have arbitrary units. C, large subunits. The m/z spectrum is complex due to co-elution of the larger subunits (upper panel). Deconvolution of the m/z spectrum to generate the zero-charged molecular mass spectrum (BioMultiView 1.3.1) allowed identification of five large subunits, including PetD, PetC, cytochrome b, cytochrome f, and FNR (from left to right, lower panel). The M. laminosus preparation showed the presence of only the first four subunits (not shown). It should be noted that while strong signals were observed for cytochrome f, PetC, PetD, and FNR, where present, the relative ionization efficiency of cytochrome b was low such that in the zero-charge molecular mass spectrum shown it had ϳ1% of the intensity of that of cytochrome f. To distinguish the true cytochrome b signal from noise it was necessary to perform further experiments with reverse-phase chromatography to more thoroughly resolve cytochrome b (Fig.  2). D, small subunits. The SEC separation did allow for partial separation of small subunits with the spectrum still showing signal from larger proteins. While the four small subunits were the four strongest peaks in the reconstructed zero-charge spectrum (lower panel), the signals from larger proteins resulted in a quite noisy base line requiring careful examination of the appropriate ions in the mass spectrum, labeled in the upper panel (the letter used abbreviates the name of the subunit, and the number indicates the number of charges). cases the measured masses agree very well with masses calculated based upon annotated data base entries (Swiss-Prot). In spinach, FNR, PetC, and PetD are in good agreement (Ͻ0.015% deviation of measured versus calculated mass) with current entries, while greater uncertainties surround cytochrome f and cytochrome b as well as the cyanobacterial subunits that are less well characterized with respect to genomic sequence. In both species, cytochrome b showed the largest mass discrepancy even with the inclusion of a heme molecule in the theoretical mass calculation. Since it is thought that the two hemes associated with cytochrome b are non-covalently bound it is a possibility that one remains noncovalently associated during the ESI process, although this a rare observation under the conditions of sample preparation/ ionization used. Alternatively a labile covalent modification should be considered, and we note the presence of a hemestaining band at 22 kDa that was observed in an original b 6 f preparation providing a precedent for a strongly bound heme on cytochrome b that is not dissociated during SDS-PAGE (29). Of course other modifications are possible, and the presence of approximately one chlorophyll a and one echinenone carotenoid per cytochrome b 6 subunit were reported recently (30). While the identities of the four large subunits were confirmed by peptide mass fingerprinting of CNBrtreated fractions from LCMSϩ (see Supplemental Data), further analysis is underway to fully describe the primary structures of cytochromes f and b.
Toward the complete description of the subunits of the cytochrome b 6 f complex, the small subunits were subjected to MSMS to generate sequence data. PetL (Fig. 3A) a doubly charged parent was fragmented to produce singly charged daughters that corresponded to y3-y20 and b2-b13 providing overlapping sequence information for comparison with the sequence translated from the spinach chloroplast genome, in this case agreeing well except for the presence of Phe at position 2 (A CNBr experiment confirmed the presence of formyl-Met at position 1 (see Fig. 3

legend).).
In the case of M. laminosus PetG (Fig. 3B) a triply charged parent ion was fragmented generating a wide range of fragments with one to three charges. The inset shows a triply charged daughter ion enlarged from the same mass spectrum illustrating the resolution obtained with hybrid quadrupole time-of-flight instruments allowing assignment of charge state, a great asset in interpretation of MSMS spectra from larger, multiply charged parent ions. In the absence of genomic data for M. laminosus, the spectra of the small subunits of this species were interpreted de novo with comparison to available sequences from other species. While the PetN, PetL, and PetM proteins could be fully assigned, the PetG sequence could be read only to position 29. Products of the spinach chloroplast genome, petG, petL, and petN, all retained their initiating formylmethionine, while the nuclear encoded petM product was cleaved after import from the cytoplasm (Table II). Interestingly, PetM from M. laminosus was the only small subunit from the cyanobacterium whose N terminus was not formylated despite being coded in the same genome as the other subunits (Table III). The sequences of the chloroplast-encoded subunits were compared with translations of the recently available sequence of the spinach chloroplast genome (31). While the sequences of PetG and PetN revealed no discrepancy, Phe was detected at position 2 of PetL (raw data, MS-Product (prospector.ucsf.edu/) outputs for all eight small subunits and an annotated assignment of the spinach PetL data are provided in Supplemental Data). The spinach chloroplast genome reports a codon for Ser at position 2 implying the presence of either a DNA sequencing error or a previously undiscovered RNA editing event. The spinach PetM sequence shows a number of mainly conserved alterations when compared with the translated Arabidopsis thaliana petM sequence, and it is noted that while ChloroP (www.cbs.dtu.dk/services/ChloroP/) correctly identified the chloroplast location of PetM, the predicted N terminus was incorrect.

DISCUSSION
Obtaining full coverage of all protein classes is an important requirement for true proteomic studies, and some integral membrane proteins, especially larger ones with multiple transmembrane helices, present a unique challenge in this respect. While improvements in two-dimensional gel technology are addressing the problem (32,33), the low accuracy and resolution of SDS-PAGE-determined molecular weights limit the value of this technique in intact mass proteomics. Intact mass measurements by ESI-MS provide mass accuracy, of- c Modifications include proteolytic removal of transit peptides (36 -320 denotes removal of 1-35) or initiating methionine, N-formylation of the N terminus (retention of initiating formylmethionine), a discrepancy between published genomic translation and experimentally determined sequence and presence of heme. ten exceeding 0.01% (100 ppm), and resolution sufficient to observe the first oxidation adduct (16 Da) on proteins as large as 100 kDa, providing an attractive alternative technology with which to monitor the subtle changes that often accompany physiological adaptations (4). Importantly, HPLC separations, including both reverse-phase and size-exclusion protocols, have been developed that allow for chromatography of large intact membrane proteins and on-line ESI-MS (4,5). While early studies relied upon quite highly purified protein preparations, it is desirable that these techniques can be applied to more complex mixtures, and a recent analysis of the thylakoid granal proteome demonstrated that high performance reverse-phase separations can resolve between 50 and 100 intact protein masses (intact mass tags) in a single dimensional analysis (8). The same study also emphasized that, while increasing the complexity of the mixture provided The ion inset is from the MSMS spectrum of the triply charged parent ion illustrating the resolution achieved with a modern quadrupole time-of-flight instrument (Q-Star Pulsar, Applied Biosystems; nanospray source, Protana). This ion corresponds to y34 3ϩ , assuming the protein is 37 amino acids long. Interpretation of MSMS spectra was accomplished manually, aided by comparison of experimental data to all publicly available homologues aligned using ClustalW (npsa-pbil.ibcp.fr/). greater proteome coverage, a return to simplified, purified protein complexes afforded improved detection of minor isoforms associated with dominant components. Thus, for detailed analysis of functional complexes, it is reasonable to subject them to individual analysis.

Size-exclusion and Reverse-phase Chromatography of Integral Membrane Proteins
A combination of size-exclusion and reverse-phase chromatographic separations allowed full subunit coverage analysis of the cytochrome b 6 f complex from photosynthetic membranes of eukaryotes and prokaryotes, emphasizing both the advantages and disadvantages of these approaches. SEC provides a rapid overview of complex components, although chromatographic resolution is limited. Reverse-phase chromatography provides a more detailed look at individual components over the longer time course of the separation, although it is limited by reduced elution efficiency of specific components in 0.1% trifluoroacetic acid-containing buffers, in this case cytochrome b 6 . This problem can be overcome by eluting the column a second time with 60% formic acid/ isopropanol, as was necessary for efficient elution of the larger PS2 subunits PsbA-D (8), and indeed such analyses confirmed the incomplete elution of cytochrome b 6 in the primary separation (data not shown). The latter point is of note when UV measurements are used in quantitation.

Documentation/Discovery by ESI-MS of Additional Subunits in the Cytochrome b 6 f Complex: FNR and Small Subunits
Before definitive mass spectral analysis, the presence of a fifth "large subunit" of the b 6 f complex was controversial.
FNR (PetH)-The possibility of an additional FNR subunit in the spinach b 6 f complex was first raised by Hind and coworkers (34) and was also noted by Hauska (35). However, gel electrophoresis provided definitive evidence for the presence in the complex of an active FNR in the spinach complex at apparently stoichiometric levels with important implications for the regulation of cyclic electron transport and the location in the photosynthetic membrane of the active site for NADP ϩ reduction mediated by photosystem I (17). FNR has not yet been detected in the complex isolated from the cyanobacte-  MDIVSLAWAA LMVVFTFSLS LVVWGRSGL a Chloroplast-encoded subunits were compared to translations from the spinach chloroplast genome (31). PetM was compared to its A. thaliana homologue. Sequence disagreements are in bold. MSMS does not distinguish Ile and Leu so these are assigned by homology where possible.
b Monoisotopic mass with no 13 C. c High-resolution MALDI-TOF of first isotopic peak (zero 13 C). conditions, an FNR protein with a 9 kDa N-terminal extension is bound to the phycobilisomes (36). It has been suggested that in response to growth under high salt (ϳ0.55 M) conditions, FNR binds to a different site in the thylakoid membrane (37). It has been shown that FNR is functionally associated with the purified spinach cytochrome b 6 f complex in vitro (17). 4 Cytochrome f reduction by reduced ferredoxin/NADPH is inhibited by 2,5-dibromo-6-isopropyl-3-methyl-1,4-benzoquinone, and the half-time of the reduction is limited by the time (1-2 s) of conventional mixing in the spectrometer. Thus, the presence of FNR in the plant b 6 f complex implies a significant physiological role of FNR in cyclic electron transport. The role of the b cytochromes in this pathway is not clear as their reduction by NADPH/ferredoxin is much slower than that of cytochrome f (38 -40). This may be a result of the use of conditions that are not strictly anaerobic. It is not clear whether FNR is stoichiometric in all chloroplast b 6 f preparations (35). Substoichiometric recovery of FNR might result due to loss of the subunit from the complex during isolation or substoichiometric association in vivo. While a role for FNR in cyclic electron transport is reasonable for b 6 f in the stromal lamellae, this is not the case for b 6 f in appressed membrane regions.
Another aspect of higher plant b 6 f function that might be impacted by bound FNR is that of protein kinase regulation. Current models of kinase activation involve transmission of information regarding conformational status of the Rieske protein on the luminal side of the complex to the active site of the kinase on the stromal side via a conformational shift in the dimeric form of the b 6 f complex (41). It has been proposed that peripheral proteins might modulate the behavior of the complex (42), and FNR would certainly be a candidate. Wor-thy of discussion is the potential for FNR to provide an interface for communication of stromal reductant status to the b 6 f complex and the kinase-modulated regulatory system. Such a regulatory role does not require that all complexes bind FNR but just that stromal ferredoxin/NADPH concentrations exert appropriate regulatory pressure at this control point. This contribution could be modulated by the amount of FNR expressed under different growth conditions and the abundance or status of other subunit proteins.
Small Subunits (PetG, PetL, PetM, and PetN)-After the original report of one to three acidic acetone-extractable small proteins Ͻ6 kDa (29) and the identification of the first small hydrophobic subunit in the b 6 f complex of maize (43), four such small (molecular mass ϭ 3.2-4.2 kDa) polypeptides, PetG, PetM, PetN, and PetL, have been found in the complex from higher plant (44) and cyanobacterial sources (44,45) and more recently from Chlamydomonas (46). The function of these small subunits is unclear, although it is likely that the structural motif of each is a single transmembrane-spanning ␣-helix. Polypeptide subunits of this nature are not present in the cytochrome bc 1 complex. Many such subunits have been found in the reaction center complexes of PS1 and PS2, and they may serve as a unique kind of integral membrane "hydrophobic filler" in photosynthetic membrane protein complexes (15) with critical roles in assembly and/or stability of the complex. It was reported that in C. reinhardtii photoautotrophic growth was impaired in the deletion mutant of petG (47) or petL (48) due to the dramatic decrease of the cytochrome b 6 f level. A similar effect was observed in a petN deletion mutant in tobacco (49). Unlike the other three small subunits, petM is not essential for the assembly and function of the cytochrome b 6 f complex in Synechocystis sp. PCC 6803 (50).
PetG, PetL, and PetN are chloroplast-encoded in higher plants and algae. However, PetM is a nuclear encoded gene product in the green alga C. reinhardtii (51) Table II).
plants (52). Sequence comparison shows a similarity of 52% (30% identity and 22% pseudo-identity) for PetG proteins (based on 29 sequences) and 56% (28% identity and 28% pseudo-identity) for PetN (based on 20 sequences). PetM is less conserved with a similarity of 29% (6% identity and 23% pseudo-identity based on 12 sequences). In PetL sequences, no conserved residue has been found, and there is only 19% pseudo-identity (based on 22 sequences). The sequences of the four small subunits from the M. laminosus b 6 f complex have high similarity to those from the cyanobacterium Nostoc sp. PCC 7120. Fig. 4 shows the alignment of available PetL sequences including the M. laminosus sequence described here. The most conserved features appear to be a tyrosine residue at approximately position 6 and a basic residue approximately 4 residues from the C terminus. It is noted that two sequences display Phe in position 2 comparable to our observation of this residue at position 2 of the spinach protein.
There is not enough genome data accumulated so far to allow firm conclusions concerning conservation of the small polypeptide complement of the cyanobacterial b 6 f complex. In the two cyanobacterial genomes completed thus far, Nostoc sp. PCC 7120 and Synechocystis sp. PCC 6803 contain genes for only three small subunits, petG, petM, and petN as annotated. However, the presence of PetL has been reported in Synechocystis, along with the other three small subunits (45), based upon mass spectrometric analysis of purified complexes. Blast searches have not yet allowed us to identify a petL homologue in Synechocystis sp. PCC 6803. Partial genome data for Synechococcus sp. PCC 7002 and 7942 show the presence of petG and petM, while the presence of petL and petN is not yet clear.
It is of significant interest that mass spectroscopic analysis has shown that the purified b 6 f complex from spinach and the cyanobacterium M. laminosus contains all four of the small subunits including PetL. One may speculate that the presence of the PetL subunit in the M. laminosus complex is the reason that it is thus far the only cyanobacterium from which it has been possible to purify an active dimeric b 6 f complex. The basis for this speculation is the finding that the absence of the PetL subunit in a mutant of C. reinhardtii destabilizes the active dimeric form of the complex in favor of the inactive monomer (53). A phosphorylated 15.2-kDa PetO subunit has been reported in a "gentle" preparation of the C. reinhardtii b 6 f complex that did not include hydroxyapatite chromatography in the purification (54). PetO was not seen in another active preparation of active C. reinhardtii complex that includes this step (55) and has not been seen in any preparations of the complex from other sources including those from spinach or cyanobacteria. A 15.2-kDa component, as yet unidentified, was found in LCMS analysis of appressed thylakoid fractions (8), but homology searches in A. thaliana provide no evidence for a higher plant homologue of petO. No direct evidence for the presence of PetO in the preparations studied here has yet been obtained, but minor components, potentially represent-ing weakly bound and non-stoichiometric subunits, are being investigated. A fifth small subunit of the cyanobacterial b 6 f complex, PetP, has been proposed based upon MALDI-TOF analysis of purified complexes from Synechocystis sp. PCC 6803 that revealed an 8-kDa component (45 (25), we did not find any evidence for this species based upon mass spectral evidence. Either the isoforms are very similar in mass/retention, or the second isoform is expressed to a low, undetectable level. It is further noted that a single spontaneous asparagine to aspartate/isoaspartate conversion would alter the pI significantly while changing the mass by ϩ1 Da only.
The difficulties discussed above highlight the need for improved quantitation in proteomics. A technique for more accurate identification and quantitation of minor isoforms based upon downstream analysis of fractions collected during LCMSϩ is under development.