Protein diffusion in and around the photosynthetic membrane must play a crucial role in photosynthetic functions including electron transport, regulation of light-harvesting, and biogenesis, turnover and repair of membrane components. Protein mobility is controlled by a complex web of specific interactions, plus the viscosity of the environment and the extent of macromolecular crowding. I discuss the techniques that can be used to measure protein mobility in photosynthetic membranes. I then summarize what we know about the constraints on protein mobility imposed by macromolecular aggregation and crowding in and around the thylakoid membranes of green plants and cyanobacteria, with particular reference to the fluidity of the thylakoid membrane and the aqueous phases on either side of the membrane (the stroma/cytoplasm and the thylakoid lumen). Current indications are that the stroma/cytoplasm is a relatively fluid environment, whereas protein mobility in the lumen may be extremely restricted. The thylakoid membrane itself has an intermediate fluidity: some protein complexes are virtually immobile, probably due to their incorporation into large, stable macromolecular aggregates. However, there is sufficient free space to allow the long-range diffusion of some complexes. Finally, I discuss some future directions for research in this area.
Photosynthetic membranes tend to be densely packed with protein. Two well-characterized examples are the chromatophore membranes of purple photosynthetic bacteria and the grana thylakoid membranes of green plant chloroplasts. In both these cases the organization of proteins in the membrane has been directly visualized by atomic force microscopy (AFM) under near-native conditions, giving a graphic impression of the dense packing of protein (1,2). In grana membranes about 75% of the membrane area is occupied by protein, making this one of the most densely-packed biological membranes known (2,3). Freeze-fracture electron microscopy (4) and an indirect estimate based on measurement of the number of photosynthetic complexes per cell (5) indicate that the thylakoid membranes of phycobilisome-containing organisms (cyanobacteria, red algae, etc.) are also densely packed with protein.
The dense packing of protein probably has two physiological advantages. First, it obviously allows a greater concentration of chromophores and electron transport complexes in a limited membrane area, permitting a higher rate of photosynthetic activity in the cell or organelle. Secondly, it seems to enhance the efficiency of light-harvesting by promoting the interaction of peripheral light-harvesting antennae with reaction center core complexes. The clearest experimental test for this idea comes from grana thylakoid membranes, where “dilution” of the membrane with additional lipid results in a reduction in the efficiency of energy transfer from LHCII to PSII (6,7). It is also very likely that a dense packing of photosynthetic membrane proteins facilitates energy migration from the light-harvesting complexes LH2 and LH1 to the reaction center in the chromatophore membrane of purple bacteria (1). In cyanobacteria the situation is rather different because the phycobilisome light-harvesting antennae are not integral membrane proteins; instead they sit on the cytoplasmic surface of the membrane. However, it is clear that phycobilisome-reaction center coupling is rather weak and unstable in vivo. Therefore, efficient phycobilisome-reaction center energy transfer probably again depends on a dense packing of reaction centers in the thylakoid membrane, providing a high concentration of phycobilisome-binding sites at the membrane surface (8).
The dense packing of protein in and around the photosynthetic membrane must clearly be advantageous on balance, but it may also cause problems. In general, the more densely packed a cell environment is with macromolecular obstructions, the slower is the diffusion of proteins (9). Beyond a critical packing density, proteins become almost immobile. The problem is strikingly illustrated by Monte Carlo simulations of protein movement is densely packed environments. For example, a simulation based on realistic estimates of the packing density in grana thylakoid membranes suggests that a protein complex might take as long as 1 h to diffuse out of a granal disk of 500 nm diameter (10). Such a drastic restriction on protein mobility would have major physiological implications (10). It is in stark contrast to the more usual situation in cells, where most proteins are in a state of rapid, random motion unless they are anchored by specific, strong interactions with immobile cell structures. For example, a typical eukaryotic plasma membrane protein has a lateral diffusion coefficient of around 0.1 μm2 s−1 (11) and a similar value was measured for a protein in the Escherichia coli plasma membrane (12). The mean displacement from the starting position can be predicted by the Einstein diffusion equation in two dimensions (Δx2 = 4Dt, where x is the mean displacement, D is the diffusion coefficient and t is time). This means that a protein diffusing at the same rate as a typical plasma membrane protein would be able to diffuse out of a 500 nm granum within a fraction of a second, around 20 000 times faster than the predicted behavior of a Photosystem II complex in a granum (10).
The diffusion of proteins in and around photosynthetic membranes must be vital for a whole range of processes. Electron transport seems to depend on the diffusion of mobile electron carriers (for example plastocyanin associated with the lumenal face of the thylakoid membrane in plants) (13–15). Regulation of light-harvesting depends on the movement of light-harvesting complexes, probably over significant distances (for example the exchange of LHCII between the grana and the stroma lamellae in plants, induced by LHCII phosphorylation ). See Dekker and Boekema (17) for a recent review of the organization and dynamics of thylakoid membranes in plants. Photosystem II frequently sustains photodamage and must be repaired, a process that, in green plants, requires the diffusion of damaged complexes out of the grana for repair in the stroma lamellae (18). The Photosystem II repair cycle takes place over a timescale of about an hour (19), so it must require significant fluidity in the membrane system. The supramolecular aggregates and arrays that are so characteristic of photosynthetic membranes must be able to self-assemble during membrane biogenesis, and this too must require lateral mobility in the membrane.
Understanding photosynthetic membrane function therefore requires knowledge of protein mobility. In practice, how much does the dense packing of proteins restrict mobility and impede membrane function? What are the main factors that control the mobility of proteins in the membrane? Are the rates of processes like electron transport and the Photosystem II repair cycle limited by diffusion, or are other factors more important? This article will outline the different experimental approaches to measuring protein mobility in intact photosynthetic membranes. Our knowledge of protein mobility is still rather fragmentary, but I will summarize what we currently know about protein mobility in and around the thylakoid membranes of cyanobacteria and green plants, and outline research directions for the future.
Experimental approaches to measuring lateral diffusion of thylakoid membrane components
Decades of multidisciplinary photosynthesis research have provided us with a wealth of non-invasive techniques that provide real-time data on photosynthetic processes in intact systems. Examples include a whole range of techniques for monitoring photosynthetic light-harvesting and electron transport by measuring chlorophyll fluorescence, and measurements of electron transport by flash photolysis or oxygen electrode. In general these measurements are carried out with a cuvette containing a suspension of cells or chloroplasts, or by monitoring several mm2 of leaf surface. They report on the averaged behavior of many millions of membrane systems (cells or chloroplasts) and thousands of billions of photosynthetic complexes. Clearly they provide no direct information on the microscopic behavior and movement of individual complexes. However, the kinetic information obtained from such techniques can be combined with structural data (usually from electron micrographs) to infer the mobility of complexes. The first estimates for diffusion rates of photosynthetic components all came from such indirect estimates. Examples include estimates of the diffusion of plastoquinol and plastocyanin based on measurement of electron transfer kinetics (13,20) and estimates of the diffusion coefficient for phospho-LHCII based on its redistribution between the grana and stroma lamellae following a temperature jump (21). In the latter case, the kinetic data came from a biochemical approach: thylakoid membranes were fractionated into grana and stroma lamellae at various times after the temperature jump, and the amount of phospho-LHCII in each fraction was measured.
The more sophisticated indirect methods for inferring protein mobility involve a combination experimental data with Monte Carlo simulations (10,21). Although these approaches are often very ingenious, my experience is that, wherever the estimates can be checked using more direct measurements, the estimates tend to be wrong, sometimes by many orders of magnitude. Usually they are wrong for interesting reasons. Inferring protein mobility indirectly always involves a set of (sometimes implicit) assumptions, and usually some of these prove to be wrong. Comparing the indirect estimate with a direct measurement can often be illuminating, as it reveals incorrect assumptions about the nature of the system. An example is the mobility of chlorophyll–protein complexes in grana thylakoids, where a Monte Carlo simulation suggested very severely restricted mobility leading to an escape time from the grana of around 1 h (10). By contrast, a more direct measurement using fluorescence recovery after photobleaching (FRAP) showed a rather mobile sub-population of chlorophyll–protein complexes probably able to escape from the grana within a few seconds (22). Here the “error” in the simulation was probably the assumption of a random distribution of complexes: in reality, organization of the complexes into large supramolecular aggregates probably facilitates the rapid diffusion of the mobile population (7). This case highlights the need for direct measurements of protein mobility in photosynthetic membranes.
Fluorescence recovery after photobleaching
Fluorescence recovery after photobleaching is a classic method for measuring diffusion in biological systems. The component of interest is first labeled with a fluorescent tag. The distribution of the component within the biological system (a cell or an isolated membrane, for example) can then be visualized using fluorescence microscopy. Generally FRAP is used in cases where the labeled component is too densely packed to allow the visualization of individual molecules at optical resolution: instead fluorescence microscopy reveals the overall distribution of a population of many thousands of molecules. To see whether the component of interest is mobile, an intense, highly-focused laser spot is used to bleach fluorescence in a small region of the sample. The bleached area can then be imaged. If the labeled component is immobile, the bleach will not change during subsequent imaging. By contrast, if the labeled component can diffuse, this will result in a characteristic “blurring” of the bleached area, as unbleached molecules diffuse into the bleached area and bleached molecules diffuse out. The kinetics of the process can often be analyzed to yield an estimate for the diffusion coefficient (23,24). FRAP measurements are often carried out using standard laser-scanning confocal microscopes, equipment now readily available in most biology departments.
A particular advantage for FRAP in photosynthetic systems is that most photosynthetic pigment–protein complexes are naturally fluorescent; sometimes there is no need to perturb the system by adding an artificial fluorescent tag (24). However, the technique has its limitations. It has very limited spatial resolution: FRAP measurements only really report on long-range diffusion on scales of about a micron or more. Much of the significant action in photosynthetic membranes must take place on smaller scales than this. For example, models for diffusion of photosynthetic electron carriers often invoke very spatially restricted diffusion within small “corrals” formed within supramolecular aggregates (13). Movement of this sort may be completely undetectable by FRAP because there is no long-range diffusion. Fully-quantitative FRAP requires a predictable membrane conformation and a membrane environment that is uniform throughout an area of at least 1 μm2. These conditions are rarely met in photosynthetic membranes, which are notorious for their intricate architecture and lateral heterogeneity (24).
Despite its limitations, FRAP has provided some novel and unexpected insights into the dynamics of photosynthetic membranes. Some examples will be discussed later in this article.
Single particle tracking
Single particle tracking (SPT) is a more advanced technique for measuring the molecular dynamics of biological systems (25,26). Like FRAP, it employs fluorescence microscopy. In contrast to FRAP (which measures the ensemble behavior of thousands of molecules), SPT observes the dynamic behavior of an individual particle (for example a single membrane protein complex). This requires more sophisticated instrumentation than FRAP, as the fluorescence microscope must be sensitive enough to detect the fluorescent tag attached to a single complex, and the rate of image capture must be fast enough to follow the motion of the single complex, which often proves to be very rapid and chaotic. SPT can provide better spatial resolution than FRAP, despite the limited resolution of an optical microscope. When a single particle is imaged, the result is a characteristic blurred point-spread function. However, if it is known that the image is of a single small particle, it can be deconvoluted to obtain the position of the particle to within a few nanometers. Thus SPT offers the promise of being able to visualize molecular movement at very fine scales. However, as with FRAP, photosynthetic systems present particular challenges for SPT. Unlike FRAP, the native fluorescence from the photosynthetic pigments cannot be used for SPT in intact systems. This is because the photosynthetic pigments are simply too tightly packed to allow individual complexes to be visualized separately. A condition for SPT is that the fluorescent label is very sparse, so that only a single particle within an area of several μm2 is tagged. This clearly does not apply to the photosynthetic pigments. Therefore, an artificial fluorescent tag must be used instead, and this may perturb the system. One method is to use an antibody to link a specific membrane protein to a fluorescent tag. However, there may be real problems in densely-packed photosynthetic membranes, where the antibody may considerably modify the behavior of the particle (for example by preventing its movement into the appressed grana membrane regions). To date I am aware of only one instance of the use of SPT in an intact photosynthetic membrane. This was an ingenious work by Consoli et al., who used an antibody to link LHCII to a fluorescent bead (26). It is clear that the bead must have excluded the LHCII from the appressed grana membranes, and the relatively large size of the bead may also have perturbed LHCII movement and interaction. Nevertheless, single LHCII complexes could be observed in rapid random motion, with a mean diffusion coefficients of about 0.03 μm2 s−1 for phospho-LHCII and 0.008 μm2 s−1 for unphosphorylated LHCII. It is suggested that the faster diffusion of phospho-LHCII is due to weaker LHCII–LHCII interactions (26).
What factors control the mobility of photosynthetic proteins?
The mobility of photosynthetic proteins must be controlled by a combination of their specific interactions with other membrane components, and the overall viscosity and crowding of the medium within which they move. Examples of specific interactions include the docking of electron donors and acceptors, and the association between light-harvesting antennae and reaction centers. When observing the movement of a photosynthetic protein it can be very hard to distinguish the specific and nonspecific factors that influence its mobility. One way to disentangle these two influences is by comparison with the movement within the system of a nonindigenous protein whose mobility is likely to be influenced only by the nonspecific factors of viscosity and crowding. This approach has been employed in E. coli for example (12) but has yet to be systematically applied to photosynthetic organisms. Nevertheless, there is now a body of data on the mobility of photosynthetic proteins within the thylakoid membrane and on the lumenal and stromal/cytoplasmic sides of the membrane, and this can be used to infer some things about the general dynamic properties of these three environments. These data will be discussed below: they are beginning to reveal the constraints imposed on the mobility of photosynthetic proteins by molecular crowding.
Protein mobility in the cytoplasm or stroma
Generally, diffusion in the bacterial cytoplasm is considerably faster than in the membrane (12). The diffusion of GFP in the chloroplast stroma appears rather rapid (27), suggesting that the stroma is also a relatively fluid environment.
There is a considerable body of data from FRAP measurements on the diffusion of phycobilisomes in cyanobacteria (28–31). The phycobilisomes are large complexes associated with the cytoplasmic surface of the thylakoid membrane, but consisting mainly of hydrophilic subunits and probably without any membrane-integral domain (29). Thus phycobilisome mobility gives an indication of the fluidity of the cytoplasm in the vicinity of the thylakoid membrane. Although phycobilisomes appear densely packed on the membrane surface (32) they are remarkably mobile in vivo, with diffusion coefficients of around 0.03 μm2 s−1 (29) despite the drag that must be imposed by interaction with the membrane surface and the reaction centers. A mutation that reduces phycobilisome size leads to faster diffusion (D ∼ 0.07 μm2 s−1) (29), suggesting that obstruction in the cytoplasm is one of the factors that slows phycobilisome diffusion. It is clear that interaction with the membrane also has an effect, as there is a marked change in diffusion coefficient in a mutant with altered lipid composition (29) and phycobilisome diffusion is drastically slowed when the water activity is lowered, apparently due to stabilization of the interaction between phycobilisomes and reaction centers (30). This work also suggests that phycobilisome mobility is essential for rapid regulation of light-harvesting in cyanobacteria (30).
The rapid diffusion of phycobilisomes indicates that crowding on the cytoplasmic side of the cyanobacterial thylakoid membrane is most unlikely to limit the rate of any photosynthetic process. The long-range diffusion coefficient measured by FRAP indicates that a phycobilisome will diffuse 100 nm away from its starting position within 80 ms. However, the diffusion coefficient will be considerably decreased by binding to the reaction centers: it is clear from the efficiency of light-harvesting that most phycobilisomes are bound to reaction centers most of the time. So phycobilisomes probably move much faster during the short time-intervals when they are not bound to reaction centers (5,8) and smaller protein complexes will probably move even faster. It is likely that the stromal surface of the stroma lamellae in chloroplasts is equally fluid.
In summary, our current picture of the cytoplasm/stroma is a rather fluid environment where soluble proteins (probably including large complexes like ribosomes) can diffuse rather rapidly, allowing them to interact efficiently with the complexes in the thylakoid membrane ( Fig. 1).
Protein mobility in the thylakoid membrane
Fluorescence recovery after photobleaching has been used to probe the mobility of various chlorophyll–protein complexes in the cyanobacterial thylakoid membrane in vivo (28,29,33,34) and chloroplast grana membranes in vitro (22). The picture that emerges is rather different from the free diffusion of proteins seen on the stromal/cytoplasmic side of the membrane. In isolated grana membranes about 75% of the chlorophyll fluorescence appears completely immobile (22). In this case it is clear that interactions within the membrane were responsible for preventing diffusion, as the stromal and lumenal proteins had been washed away and the membranes were mounted by adsorption onto a fluid artificial bilayer (22). Is protein immobility due to general crowding of the membrane, or more specific interactions leading to the formation of large, stable aggregates? Further work suggests that both factors have an influence. “Dilution” of the membranes with additional lipid leads to an increase in the mobile fraction and the diffusion coefficient, presumably because macromolecular crowding is reduced (22). However, incubating the membranes in “low-salt” buffer also increases mobility, probably because specific protein–protein interactions are weakened (22). Mobility of chlorophyll-proteins is clearly heterogeneous. In “native” conditions about 25% of the chlorophyll fluorescence is mobile, and this population of chlorophyll-protein can diffuse surprisingly fast (D ∼ 0.005 μm2 s−1) (22). We are not yet sure of the identity of the mobile population, but it is rather likely to be a subpopulation of LHCII. The conclusion is that much of the protein in these membranes is aggregated into large, immobile assemblages. However, despite the crowding of the membrane there is enough space to leave open channels for diffusion of smaller protein complexes (22). It remains to be seen whether protein mobility in grana changes under different physiologic conditions. For example, does protein mobility increase under photoinhibitory conditions, to allow Photosystem II to leave the grana for repair in the stroma lamellae?
A system where protein diffusion has been measured in vivo is the thylakoid membranes of cyanobacteria. Here the situation is more complex than in granal membranes in vitro. In this intact system membrane protein mobility may also be influenced by the interaction of protruding hydrophilic domains with the cytoplasm and the thylakoid lumen ( Fig. 1). The cytoplasm appears to be a rather fluid environment (see above) but the lumen may be a different matter (see below). The experimental finding is that Photosystem II is normally completely immobile (D below 0.00002 μm2 s−1 ). We cannot be sure whether the immobility is due to interactions within the membrane, or the lumen, or both. However, interestingly, another chlorophyll protein, IsiA, is mobile in the membrane, albeit rather slowly (D ∼ 0.003 μm2 s−1) (33). Unlike Photosystem II, IsiA has no significant lumenal domain. So, as in granal membranes, there must be enough space in the membrane to allow the diffusion of proteins that are not strongly complexed into large assemblages. It was also shown that up to about 50% of Photosystem II can become mobile under specific conditions, following exposure to intense red light (D ∼ 0.02 μm2 s−1) (34). This is considerably faster than the movement of IsiA, and it suggests a large-scale mobilization of the thylakoid membrane, perhaps due to the breakdown of supramolecular assemblages of Photosystem II. This eventually leads to a large-scale redistribution of Photosystem II in the cell, which may be required for the Photosystem II repair cycle (34). We cannot be sure whether the lumenal domain of Photosystem II travels with the rest of the complex—it is possible that it is detached under these conditions.
A further factor that must be considered is the fluidity of the lipid bilayer, which is strongly temperature-dependent and also influenced by factors such as lipid desaturation. Both temperature and desaturation were shown to have strong effects on lipid mobility, as judged from FRAP measurements of the diffusion of a lipophilic fluorescence probe in cyanobacterial thylakoid membranes (35).
In summary, protein mobility in thylakoid membranes is controlled by a complex combination of specific protein–protein interactions and general membrane viscosity and crowding. However, some proteins are always mobile in both the systems that have been examined in detail (chloroplast grana membranes in vitro and cyanobacterial thylakoids in vivo).
Protein mobility in the thylakoid lumen
To date, we have very little direct information on the fluidity of this compartment. It would be very interesting to probe the fluidity of the thylakoid lumen directly, perhaps by introducing GFP into the lumen and then using FRAP to measure its diffusion. However, to date it has not been possible to produce a transformant that translocates GFP into the lumen. One attempt to do this (by expressing in a cyanobacterium GFP coupled to a leader sequence recognized by the twin-arginine translocation pathway) resulted in export of GFP to the periplasm rather than the lumen (36). The most direct indication of lumen dynamics that we have comes from recent work on a cryptophyte alga. Cryptophytes use phycobilins as light-harvesting proteins. In cryptophytes the phycobilins are in the thylakoid lumen (unlike cyanobacteria and red algae where they are assembled into phycobilisomes on the cytoplasmic side of the membrane). As the phycobilins are highly fluorescent they can be used as natural probes for lumen dynamics in FRAP experiments. FRAP measurements on phycoerythrin in the cryptophyte Rhodomonas salina show that it is completely immobile (R. Kaňa, O. Prášil and C.W. Mullineaux, unpublished). This is in very sharp contrast to the highly mobile phycobilins on the cytoplasmic side of the thylakoid membrane in cyanobacteria (see above). The difference suggests that protein mobility in the lumen may be extremely restricted. Perhaps the lumenal proteins are so tightly packed that they form an almost crystalline gel. Movement may only be possible within small microdomains. Indirect approaches provide some support for this picture, as some electron transport measurements suggest highly restricted diffusion of plastocyanin (13,37). However, there are also indications for long-range movement of plastocyanin under some conditions (14). The dimensions of the lumenal space appear quite flexible, changing under illumination and osmotic stress (15,38), so it is likely that protein mobility is also strongly dependent on such conditions (15).
In summary, we still have much to learn about the dynamics of the thylakoid lumen. However, the very limited information that we have suggests that protein mobility in this compartment may be extremely restricted, perhaps more so than in the membrane itself (Fig. 1).
Future directions for research
So far our knowledge of the dynamics of photosynthetic membranes is rather fragmentary. There is plenty of work still to be done. Some obvious directions for future research include:
1 Tracking the mobility of a wider range of photosynthetic complexes. This could be done by using FRAP or related techniques, in conjunction with GFP tagging in vivo or fluorescent antibody labeling in vitro. We have some information already about Photosystem II and light-harvesting antennae (phycobilisomes, LHCII) but what about Photosystem I, plastocyanin and the cytochrome b6f complex?
2 Further exploration of the effect of different physiological conditions on the dynamics of thylakoid membranes. One piece of work provides evidence for a large-scale mobilization of the cyanobacterial thylakoid membrane as a physiological response to high-light stress (34). Many other physiological responses remain to be explored.
3 Extension of FRAP and similar measurements to other photosynthetic membrane systems. To date the technique has been applied mainly to only a small subset of photosynthetic membranes that have particularly convenient geometry (chloroplast grana membranes in vitro and the thylakoid membranes of some cyanobacteria in vivo). However, with some limitations, the technique can be applied to more complex membrane systems. For example, preliminary work on intact green plant chloroplasts suggests that there is a pool of chlorophyll–protein complexes that can exchange between grana on a timescale of a few minutes (T.K. Goral and C.W. Mullineaux, unpublished).
4 Use of nonindigenous fluorescent probes in or near the thylakoid membrane. The dynamic behavior of these probes will give the best indication of the physical environment, without the complications caused by the specific interactions of indigenous protein complexes.
5 Use of more sophisticated techniques to provide more detailed information on the mobility of photosynthetic proteins. Techniques with higher spatial resolution than FRAP could provide direct information about the mobility of those proteins that may be confined to small localized domains. This sort of localized diffusion may play a crucial role in electron transport, for example, SPT is one promising approach. Scanning-probe techniques (AFM and scanning near-field optical microscopy) may also be useful here.
Acknowledgements— I would particularly like to thank Paul Falkowski, Tomasz Goral, Silvia Haferkamp, Radek Kaňa, Helmut Kirchhoff, Ondrej Prášil and Alexander Ruban for useful discussion and their input into current collaborations. Financial support for work in this area in C.W.M.’s laboratory has come from the Biotechnology and Biological Sciences Research Council, The Royal Society and the Wellcome Trust.