Organization and Flexibility of Cyanobacterial Thylakoid Membranes Examined by Neutron Scattering

Background: In cyanobacteria, light harvesting and photosynthesis occur in the thylakoid membranes. Results: The distances between thylakoid membranes are correlated with the size of the phycobilisome antenna and change reversibly and rapidly upon illumination. Conclusion: Thylakoid membranes have a structural plasticity tied to the regulation of photosynthesis. Significance: Characterizing the structural changes in photosynthetic membranes is crucial for understanding light harvesting and photosynthetic productivity. Cyanobacteria are prokaryotes that can use photosynthesis to convert sunlight into cellular fuel. Knowledge of the organization of the membrane systems in cyanobacteria is critical to understanding the metabolic processes in these organisms. We examined the wild-type strain of Synechocystis sp. PCC 6803 and a series of mutants with altered light-harvesting phycobilisome antenna systems for changes in thylakoid membrane architecture under different conditions. Using small-angle neutron scattering, it was possible to resolve correlation distances of subcellular structures in live cells on the nanometer scale and capture dynamic light-induced changes to these distances. Measurements made from samples with varied scattering contrasts confirmed that these distances could be attributed to the thylakoid lamellar system. We found that the changes to the thylakoid system were reversible between light- and dark-adapted states, demonstrating a robust structural flexibility in the architecture of cyanobacterial cells. Chemical disruption of photosynthetic electron transfer diminished these changes, confirming the involvement of the photosynthetic apparatus. We have correlated these findings with electron microscopy data to understand the origin of the changes in the membranes and found that light induces an expansion in the center-to-center distances between the thylakoid membrane layers. These combined data lend a dynamic dimension to the intracellular organization in cyanobacterial cells.

Cyanobacteria are prokaryotes that can use photosynthesis to convert sunlight into cellular fuel. Knowledge of the organization of the membrane systems in cyanobacteria is critical to understanding the metabolic processes in these organisms. We examined the wild-type strain of Synechocystis sp. PCC 6803 and a series of mutants with altered light-harvesting phycobilisome antenna systems for changes in thylakoid membrane architecture under different conditions. Using small-angle neutron scattering, it was possible to resolve correlation distances of subcellular structures in live cells on the nanometer scale and capture dynamic light-induced changes to these distances. Measurements made from samples with varied scattering contrasts confirmed that these distances could be attributed to the thylakoid lamellar system. We found that the changes to the thylakoid system were reversible between light-and dark-adapted states, demonstrating a robust structural flexibility in the architecture of cyanobacterial cells. Chemical disruption of photosynthetic electron transfer diminished these changes, confirming the involvement of the photosynthetic apparatus. We have correlated these findings with electron microscopy data to understand the origin of the changes in the membranes and found that light induces an expansion in the center-to-center distances between the thylakoid membrane layers. These combined data lend a dynamic dimension to the intracellular organization in cyanobacterial cells.
In photosynthetic organisms like plants, algae, and cyanobacteria, the light reactions occur in the highly specialized thylakoid membranes that house the components of the photosynthetic electron transfer chain. Although plant thylakoids are arranged in stacked grana and unstacked stroma membranes, and algal thylakoids can be appressed or separated depending on the strain (1), cyanobacterial thylakoids typically do not stack or appress and are more uniformly sheet-like (2). Cyano-bacterial strains display differences in intracellular organization, in particular having various arrangements of the thylakoid membranes within the cell interior (2). The organization of the thylakoid membrane system in cyanobacterial cells has been explored by different techniques, including recent work using electron tomography to show the three-dimensional arrangement of the membrane system (3,4). Although these studies have begun to show the complexity and spatial detail of these membrane systems, such work provides static views of what are undoubtedly dynamic structures. To date, there is little information exploring the rearrangements of the thylakoid membrane system in response to environmental stimuli such as light or temperature in live cyanobacterial cells. To thoroughly understand the form and function of cyanobacterial thylakoid membranes, studies are needed that incorporate high-resolution measurements of membrane structure with information on how the organization of these membranes changes in response to environmental inputs.
Probing the structural responses of the thylakoid membrane systems of photosynthetic organisms to light is of particular interest. In Synechocystis 6803, growth in low light (0.5 mol of photons m Ϫ2 s Ϫ1 ) with glucose as a carbon source resulted in inflated thylakoid lumens measuring up to 300 nm (5). Kirchhoff et al. (6) used cryo-electron microscopy to examine vitreous sections and found that the grana lumen in Arabidopsis thaliana expands in the light, thereby increasing the lumenal volume and presumably facilitating protein diffusion. Earlier studies of the thylakoid membranes in the marine algae Ulva and Porphyra demonstrated a reproducible and reversible light-induced flattening of the thylakoid lumenal space as measured by electron microscopy of chemically fixed samples and microdensitometry (7). Recently, Nagy et al. (8) used smallangle neutron scattering (SANS) 3 to examine the Synechocystis 6803 PAL mutant and found a light-induced increase in repeat distances that they attributed to periodic spacing between adja-cent thylakoid membrane pairs. These data demonstrate the need for additional studies to understand the effect of light on the organization of thylakoids in detail and to clarify whether illumination results in expansion or contraction of the membrane system.
Obtaining structural information from photosynthetic organisms under changing conditions requires nondestructive techniques capable of probing the membrane systems of live cells. Studies using light microscopy approaches such as fluorescence recovery after photobleaching and hyperspectral confocal fluorescence microscopy have led to intriguing findings regarding the mobility of photosynthetic components and the organization of photosynthetic pigments within cyanobacterial cells (9,10). SANS can also be used to characterize periodically organized biological membrane systems on smaller length scales by observing diffraction from regularly spaced structural features separated by ϳ1-200 nm (11). Both the accessible length scales and the ability to examine individual components of biological systems using contrast matching have led to the application of SANS in studies of cyanobacteria, eukaryotic algae, plant thylakoids (8,12), and mammalian mitochondria (13).
To examine thylakoid membrane organization in cyanobacterial cells under different conditions, we used a correlative approach incorporating both transmission electron microscopy (TEM) and SANS. These techniques both provide complementary data at the length scales relevant to the study of biological structures, including membranes. We compared the thylakoid membrane organization in wild-type Synechocystis 6803 cells with that of mutants in which the phycobilisomes have been truncated by varying degrees. These mutants are CB (containing one phycocyanin hexamer per rod), CK (lacking phycocyanin rods but still possessing the allophycocyanin core), and PAL (lacking functionally assembled phycobilisomes) (14 -16). Our results show that the distances between thylakoid membrane layers is correlated with the size of the light-harvesting phycobilisome antenna complexes and that these distances change reversibly and rapidly upon illumination, with strain-specific differences. Furthermore, our data indicate that the photosynthetic membranes in Synechocystis 6803 have a structural plasticity that is tied to the function of the photosynthetic electron transfer chain.

EXPERIMENTAL PROCEDURES
Electron Microscopy-Cells were prepared for electron microscopy as described previously (3). Cultures of WT, CB, CK, and PAL cells were grown for 5 days in liquid BG11 medium (17) with appropriate antibiotics, harvested by centrifugation, resuspended in a small volume, and loaded into planchettes for high-pressure freezing (BalTec). Thin sections (ϳ80 nm) were cut from resin-embedded material and imaged using an LEO 912AB electron microscope equipped with a ProScan digital camera. Measurements were taken using iTEM software (Soft Imaging System) and are given as the results of 20 -30 measurements per cell type.
Small-angle Neutron Scattering-SANS measurements were performed using the Bio-SANS instrument (18) at the High Flux Isotope Reactor, Oak Ridge National Laboratory. Cyano-bacterial suspensions were in BG11 medium prepared with H 2 O (H-medium), D 2 O (D-medium), or a mixture of the two. Cells grown in standard H-medium were harvested by centrifugation, washed twice with 10 ml of fresh medium (with D 2 O depending on the experiment), resuspended in 5 ml of medium, and equilibrated overnight at 30°C under white fluorescent light at ϳ25 mol of photons m Ϫ2 s Ϫ1 with shaking at 150 rpm. All cultures were then adjusted to the same optical density (measured at ϭ 730 nm). SANS measurements were carried out in 1-mm path cylindrical cells at 30°C. The sample enclosure was equipped with a lighting apparatus that illuminated each sample position with cool white LED light at ϳ20 mol of photons m Ϫ2 s Ϫ1 . A dark enclosure was also installed to allow for dark adaptation of individual samples. Three different configurations of the Bio-SANS instrument were employed to collect data over the scattering vector range of 0.003 Ͻ q Ͻ 0.7 Å Ϫ1 with sample-to-detector distances of 1.1, 6.8, and 15.3 m and a neutron wavelength of 6 Ϯ 0.15 Å.
Data Analysis-SANS data were analyzed using routines implemented in IGOR Pro software (WaveMetrics). The scattering intensity curves, I(q), were obtained by azimuthally averaging the processed two-dimensional detector images, which were normalized to incident-beam monitor counts and corrected for detector dark current, pixel sensitivity, and solidangle geometry. Scattering from the quartz cell and appropriate media was subtracted with a volume fraction of 0.95 used to scale the intensity of the buffer scattering before subtraction. The resulting intensity profiles are presented in absolute scattering cross-sections per unit volume and per solid angle. Individual diffraction peak q positions were determined by fitting each scattering profile with a variable number of pseudo-Voigt curves and a single underlying power-law function with exponent Ϸ2.5. Diffraction peak positions were equated to corresponding repeat distances through the relation q ϭ 2/d, where q is the peak center in reciprocal space (Å Ϫ1 ) and d is the repeat distance (Å).
Quasi-elastic Neutron Scattering-Quasi-elastic neutron scattering experiments were performed on the backscattering spectrometer (BASIS) at the Spallation Neutron Source, Oak Ridge National Laboratory (19). The samples were prepared as described above and placed in rectangular aluminum cells with colorless sapphire windows to allow for illumination with white light (ϳ280 mol of photons m Ϫ2 s Ϫ1 ). The raw data were converted from time of flight to energy transfer at selected q values using standard BASIS data reduction software, and data analysis was performed using the DAVE software package (20). Data fitting was carried out using the jump diffusion model, G(q) ϭ បDq2/(1 ϩ Dq2), where ប is the reduced Planck's constant, D is the translational diffusion coefficient, and is the residence time between diffusion jumps.

RESULTS
We have used TEM to examine the thylakoid membrane systems in wild-type Synechocystis 6803 and the phycobilisome mutant strains CB, CK, and PAL (10). Thin-section electron micrographs of high-pressure frozen cells showed that the distances between thylakoid membranes were greatest in WT cells with intact phycobilisomes and smallest in the PAL mutant, in which phycobilisomes were absent (Fig. 1). In all strains, we measured the thickness of the membrane pairs that enclose the lumenal space and form a single thylakoid and the interthylakoidal distances between the membrane pairs. The thickness of the thylakoid membrane pair was consistently ϳ150 Å in all strains, but the distances between membrane pairs varied considerably. We report the distance between membrane pairs in two ways: as the membrane-to-membrane boundary distance across the width of the interthylakoidal space and as the centerto-center repeat distance between the thylakoid layers. The membrane-to-membrane distance is relevant for discussing the space available to the phycobilisome antenna in the cytoplasmic region, whereas the center-to-center repeat distance is the measurable quantity directly obtained in SANS.
Wild-type cells (Fig. 1A) showed the characteristic intracellular arrangement of Synechocystis 6803: layers of approximately three to six concentric membrane pairs generally con-forming to the curvature of the cell envelope, with a central cytoplasmic region largely devoid of thylakoid membranes but in which other cellular components were found. We measured the interthylakoidal distances between membrane pairs as averaging 400 Ϯ 80 Å in WT cells, consistent with reported measurements of 300 -500 Å (21,22). The center-to-center thylakoid membrane distance was 550 Å for WT cells ( Fig. 1B and Table 1). In terms of overall cell morphology and thylakoid membrane organization, the CB strain (which contains only one phycocyanin hexamer per rod) appears very similar compared with WT (Fig. 1C). Measurements of interthylakoidal distances averaged 330 Ϯ 60 Å, and the center-to-center membrane distance was 480 Å for CB ( Fig. 1D and Table 1). In the CK strain, which possesses only the allophycocyanin core, distances between thylakoid membrane layers averaged 320 Ϯ 20 Å (Fig. 1E), with center-to-center distances of 470 Å ( Fig. 1F and Table 1). In the PAL mutant (Fig. 1G), thylakoid membranes consistently formed straight bands with little curvature (10). The interthylakoidal distances in PAL averaged 190 Ϯ 30 Å, with center-to-center membrane distances of 340 Å ( Fig. 1H and Table 1).
Periodic cellular structures were indicated in the SANS profiles obtained from samples suspended in 100% D-medium by the appearance of diffraction peaks above power-law and incoherent scattering backgrounds as shown for WT in Fig. 2A. A series of samples equilibrated in 0, 20, 40, 60, 80, and 100% D-medium were measured to determine the scattering contrast match points(s) of the diffracting structures and to identify their general molecular composition (viz. protein, lipid, etc.). Fig. 2B shows that the contrast for the structure diffracting at the lowest q was "matched" (i.e. disappeared) at ϳ17% D-medium, a value known to correspond to lipid structures. The scattering profile of WT in 20% D-medium ( Fig. 2A) demonstrated that all diffraction peaks were largely matched at this media scattering length density and pointed to the ordered thylakoid membranes observed by TEM as the source of all of the diffraction measured from these samples.
Including the thylakoid repeat features, up to five distinct peaks, depending upon the strain, could be observed with small strain-to-strain variations in both the q positions and relative intensities (Table 1). A putative assignment of all of the diffrac-  tion features observed from the WT sample is presented in the following discussion. Furthermore, to ensure that the observed peaks were originating from physiological phenomena in live cyanobacterial cells, we performed plating experiments comparing cells exposed to neutrons for 2 h with cells treated identically but not exposed to neutrons. We found that exposing cells to neutrons for this amount of time did not decrease cell viability, and we therefore concluded that our results show the responses of live cyanobacterial cells to our experimental conditions. The small-angle scattering observed for all strains equilibrated in 100% D-medium and incubated in light is shown in Fig. 3A. The scattering profiles from all strains showed significant differences in the observed diffraction at the lowest q (Peak 1). On the basis of the length scale of both this periodicity and the periodic thylakoid arrangements from TEM, we attributed this diffraction feature to the center-to-center thylakoid membrane period ( Table 1). The systematic shortening of this period with antenna truncation can be seen in the shift to higher q values for this feature.
To determine whether changes to the SANS profiles would occur if cells were incubated in the dark, cells were darkadapted for 1 h, and data were collected in an identical manner as for cells in the light. SANS data from cells in the dark showed changes in the q positions and relative intensities, but the overall pattern of the profiles remained fairly consistent for each strain (Fig. 3B). All strains have a first peak at 0.010 -0.017 Å Ϫ1 and a third peak centered at ϳ0.042 Å Ϫ1 . However, Peaks 2, 4, and 5 are present only in a subset of the strains (Table 1). Peak 2 is clearly observed in WT in the dark at ϳ0.022 Å Ϫ1 . In the light, this peak is strongly reduced but still discernible as a shoulder (Fig. 3A), and the curve shape suggests yet another peak at ϳ0.016 Å Ϫ1 in WT in the light. Peak 2 is weakly present in CB but not in the other mutant strains. There is a fourth peak in all strains except WT in the dark from 0.045 to 0.071 Å Ϫ1 . A fifth peak is present in WT and CK ( Table 1).
The trend in changes between light and dark was not consistent among the strains ( Table 1). The values that corresponded to the largest scale (Peak 1; Table 1) showed a small dark-induced expansion in the CB mutant, but CK and PAL showed a contraction in this distance in the dark, with changes of 37.4 Å for CK and 25.8 Å for PAL. No change in the position of Peak 1 was discernible for WT, but its intensity was clearly reduced in response to illumination. Peaks 3-5, observed at larger q, which corresponds to smaller distances, showed mostly expansion in light. This effect was clearest for Peak 3, with ϳ8% expansion in WT and 4 -5% expansion in the mutants. All four strains showed a reduction in scattering intensity for the peaks at smaller q values of Ͻ0.025 Å Ϫ1 (distances of Ͼ250 Å) when exposed to light. In contrast, scattering intensity for the higher q peaks clearly increased for WT. Similarly, in all three mutants, the intensity of Peak 4 at q Ϸ 0.045 Å Ϫ1 increased when cells were exposed to light.
Time-series measurements were made to characterize the time scale of the transition between the dark-and light-adapted states. Samples of WT, CK, and PAL strains were adapted to The contrast match point for the major peak is 17.9% in the light and 16.5% in the dark, within the range of a peak that is composed primarily of lipids. In A, intensities for WT and PAL were vertically offset by Ϫ12 and 6 cm Ϫ1 , respectively. In B, intensities for WT, CB, and PAL were vertically offset by Ϫ12, Ϫ3, and 6 cm Ϫ1 , respectively. darkness for 1 h before a series of 5-min measurements during illumination began. SANS from the WT strain (Fig. 4A) showed that the dark-to-light changes occurred rapidly, with the majority happening within the first 10 min of illumination. In contrast, however, structural changes in the PAL mutant began to appear only after 30 min of illumination (Fig. 4B), with change occurring up to 2 h. Reversed light-to-dark timeseries measurements were conducted for the CK strain. The dark-adapted structures were largely recovered within 20 min (Fig. 4, C and D).
To determine whether these changes depended upon functional photosynthetic electron transport, we incubated darkadapted cells with 3-(3,4-dichlorophenyl)-1,1-dimethyl urea (DCMU), an inhibitor of electron transport from photosystem II. DCMU treatment did not perturb the dark-adapted structure. However, the SANS measured from these samples after 30 min of illumination (Fig. 5) showed little change from the corresponding dark state and demonstrated the dependence of the light-induced changes on an intact electron transport chain.
It was of interest to investigate if the light-induced structural rearrangements observed by SANS that are attributed to a swelling of the thylakoid lumen resulted in changes in the intracellular dynamics of the cyanobacteria. Quasi-elastic neutron scattering is ideally suited to investigate dynamics on a pico-to nanosecond time frame. Measurements on the Spallation Neutron Source backscattering spectrometer (BASIS) were carried out to assess the effect of light on the microscopic dynamics of the cyanobacteria and the intracellular water. Both the faster water dynamics and the slower cellular dynamics could be measured simultaneously due to the combination of a highenergy resolution and a wide dynamic range of BASIS. It was possible to analyze the translational dynamics of the intracellu-lar water up to q ϭ 0.7 Å Ϫ1 and the internal dynamic motions of cellular components up to q ϭ 0.9 Å Ϫ1 . Fig. 6 shows the q dependence of the quasi-elastic signal broadening (half-width at half-maximum) for WT Synechocystis 6803 cells measured with and without light. Data for Escherichia coli are presented for comparison. The broader q-dependent component originates from the intracellular water, whereas the narrower q-independent component originates from the cellular components of the bacteria. Unlike the measured dynamics of water, which are clearly translational, the internal dynamics of the bacteria are of localized character, resulting in q-independent broadening of the scattering signal. The q dependence of the water data shown in Fig. 6 was fitted with the jump diffusion model described above. The results of the fitting are shown in Table 2. Within the accuracy of the experiment, the fit values obtained are essentially the same for cells in the light-and darkadapted states and also similar to the translational diffusion coefficient of bulk water (29.8 ϫ 10 Ϫ10 m 2 s Ϫ1 ) (23). Thus, the space accessible to the intracellular water in cyanobacteria is not sufficiently restricted to affect its translational dynamics, despite the profound structural changes in the membranes observed in the SANS experiment. Similarly, the measured internal dynamics of the cyanobacteria are not affected by light either; the relaxation times calculated from the q-independent broadening component as ϭ ប/⌫(q) are (157 Ϯ 21) ϫ 10 Ϫ12 s and (153 Ϯ 21) ϫ 10 Ϫ12 s with and without light, respectively. In the case of E. coli, the diffusion coefficient of water was similar to the values obtained with the cyanobacteria ( Table 2), but the dynamics of the cellular components exhibited a significantly slower relaxation time ((175 Ϯ 15) ϫ 10 Ϫ12 s), as is evident in Fig. 6. This possibly reflects the notable differences in the cellular architectures of the two bacterial species.

DISCUSSION
Cyanobacteria such as Synechocystis 6803 have a highly organized thylakoid membrane architecture, and it has been proposed that rearrangements in the membrane system and the spacing between membrane layers play a role in the cellular regulation of photosynthesis (10). Ideally, investigation of membrane architecture in cyanobacteria requires real-time in situ studies that can follow the time course of reorganization in response to external stimuli such as light or other environmental conditions. However, techniques that combine the required spatial resolution with the capability to resolve changes over time are rare. TEM is the method of choice for determining cell physiology at the nano-to meso-scale and has been used extensively to determine thylakoid architecture (4); however, it can rely heavily on fixation of the specimen and does not provide a dynamic picture. On the other hand, solution scattering techniques can, in principle, provide this missing link to the understanding of photosynthesis but have hardly been used in this field of research until recently (24).
The length scales of interest in this context range from the thickness of individual lipid bilayers to the spacing between thylakoid membrane layers (a few nanometers to just under 100 nm), which are covered by small-angle scattering techniques using x-rays (small-angle x-ray scattering) or neutrons (SANS). However, well defined peaks are not commonly observed in the small-angle range of scattering data obtained from complex  black circles) profiles. In A, intensities were vertically offset by 0.4 cm Ϫ1 for treated and untreated samples exposed to light. In B, intensities were vertically offset by 0.7 cm Ϫ1 for untreated samples and by 0.3 cm Ϫ1 for treated samples exposed to light. In C and D, intensities for treated and untreated samples exposed to light were vertically offset by 0.3 cm Ϫ1 .

TABLE 2 Translational diffusion coefficients (D) and mean residence times () obtained from the jump diffusion model of the quasi-elastic neutron scattering data
Error is reported as S.D.

Changes in Cyanobacterial Cells Examined by Neutron Scattering
FEBRUARY 1, 2013 • VOLUME 288 • NUMBER 5 biological systems. To observe diffraction-like peaks in biological small-angle scattering, well defined discrete repeat distances must be present in the system, and there must be good contrast between the scattering length densities of the ordered cellular component and the surrounding medium. Both of these prerequisites are fulfilled in cyanobacteria that are equilibrated in D 2 O media: the membrane system has been shown to exhibit well defined spacings, and the D 2 O-rich cytoplasm and lumen have strong contrast relative to hydrogen-rich lipid membranes. For a system that is as complex as entire cells, however, the convoluted scattering contributions from a mixture of structural features generally pose great challenges to the interpretation of small-angle scattering data. Fortunately, this situation can be greatly improved for SANS by contrast enhancement through selective replacement of hydrogen atoms with the heavier isotope deuterium ( 2 H or D). In cyanobacterial cells, well organized arrangements of phycobilisomes have been reported (21), and a number of other structural features could, in principle, produce peaks. By using contrast variation, we were able to conclude that the observed peaks are indeed due to membrane repeat distances (Fig. 2). We attributed these distances to the thylakoid membrane system due to the repeating nature of the concentric thylakoid membrane layers and considered it unlikely that the membrane components of the cell envelope are able to contribute to the observed peaks. Recent work by Nagy et al. (8,12) has successfully applied SANS to photosynthetic systems. However, their pioneering study has left many open questions. In our study, we have developed a more comprehensive picture through correlation of SANS data with TEM images of the same cyanobacterial strains, observation and interpretation of additional diffraction peaks, and application of scattering contrast variation. SANS data from the four cyanobacterial strains investigated under both light and dark conditions show qualitative agreement (Fig. 3). All profiles can be described by sets of correlation peaks above a power-law base line (exponent approximately Ϫ2.5). The largest repeat distance observed in our studies was found in WT cells at 627 Å (Peak 1). This peak has been predicted previously (8), but this is the first time that it was observed experimentally. The repeat distance agrees well with the expected sum of ϳ150 Å for thylakoid membrane thickness and an interthylakoidal spacing of ϳ450 Å (21) or 300 -500 Å (22) as has been reported from TEM data, as well as with our own measured interthylakoidal distance of ϳ400 Å (Fig. 1). The differences seen here among the TEM data are likely due to the inherent biological variability of the membrane spacing when measured within individual cells, and sample preparation and variations in growth conditions may also contribute.
Studies have shown that modulation of phycobilisome antenna size results in changes to the interthylakoidal distances (21). The largest SANS repeat distance (position of Peak 1) systematically decreases with successive reduction of the lightharvesting antenna in the order WT, CB, CK, PAL, giving further strong evidence that the size of the phycobilisome is closely related to membrane spacing as reported previously (10,21). In particular, removal of the phycocyanin rods in CK leads to a reduction of SANS-derived membrane spacing by 170 Å (in the light) or 207 Å (in the dark), and the latter number agrees per-fectly with the result reported for the phycocyanin-less strain PMB11 by Olive et al. (21). In the mutant strain P⌬E, which lacks assembled phycobilisomes, a small interthylakoidal distance of 40 Å was measured (21,22). However, our TEM data did not show as severe a decrease in the interthylakoidal distance in the PAL mutant, which lacks phycobilisomes, where the interthylakoidal distances averaged 190 Å. The origin of the differences found in the interthylakoidal distances measured by TEM between these two strains lacking phycobilisomes, PAL and P⌬E, is unknown but may be due in part to fixation methods (high-pressure freezing versus chemical fixation). It is interesting that our SANS data showed a repeat distance of 383 Å (in the light) or 357 Å (in the dark) for the PAL strain (Peak 1), and this value is similar to our center-to-center membrane distance from PAL TEM images (340 Å) ( Table 1).
For a system with appressed membranes, a repeat distance of ϳ190 Å would be expected based on the thickness of a thylakoid pair (ϳ150 Å) and an interthylakoidal distance of ϳ40 Å, as seen in the P⌬E strain and as has been measured for appressed thylakoids in other systems (1). This value was observed by Nagy et al. (8) and also in our SANS data for the PAL mutant as Peak 3. However, a value very close to this is also present in WT, CB, and CK, and this repeat distance varies relatively little among the different strains. These data suggest that all of the strains may contain some regions of nearly appressed thylakoids with interthylakoidal distances of 20 -40 Å. We did not see evidence for this in our TEM data; however, SANS gives data averaged over a large population. Moreover, if such appressed membranes form extended stacks, then the diffraction signal would be strongly amplified, and even a small concentration of such regions could be visible in the diffraction pattern. In all strains, this repeat distance increases in response to illumination. This may be interpreted as an increase in interthylakoidal spacing from 20 -30 Å in the dark to 30 -40 Å in the light. However, an alternative interpretation can be given in that the thickness of the lumen also increases in response to light.
SANS data from WT in the dark exhibited a repeat distance of 300 Å (Peak 2 at 0.022 Å Ϫ1 ). This corresponds to an interthylakoidal spacing too narrow to accommodate phycobilisomes. In light, this peak is attenuated and accompanied by an additional repeat distance of 393 Å (at ϳ0.016 Å Ϫ1 ). The latter value suggests that such thylakoids are separated by a single layer of phycobilisomes. This peak is also weakly present in the CB mutant.
The repeat distances obtained from Peaks 4 and 5 are too small to encompass the width of a thylakoid membrane plus interthylakoidal spacing. Instead, these peaks correspond to distances that are similar to the width of a single thylakoid membrane. Indeed, the repeat distance of the pair of lipid bilayers that enclose the width of the lumen is expected to be ϳ100 -110 Å, and Peak 5 matches this distance. However, the reason for the absence of Peak 5 in CB and PAL is unknown at this time. With values slightly larger than Peak 5, Peak 4 might suggest a swollen thylakoid membrane in which the lumen width is increased by ϳ30 Å. Peak 4 is absent in WT in the dark but rises in response to light. Thus, SANS data suggest a swelling of the lumen in response to light exposure in WT. The significance of lumen swelling in response to light is unknown at present, but this is consistent with the findings of Kirchhoff et al. (6).
In summary, SANS data of WT and mutant Synechocystis 6803 cells revealed at least five distinct peaks that could be correlated with TEM images of these same strains. These findings are summarized in Fig. 7, which shows a schematic drawing of our interpretation of these data. Peak 1 is the center-to-center thylakoid membrane repeat distance that may accommodate phycobilisomes, whereas Peak 2 is a center-tocenter repeat distance that is too small for phycobilisomes. Peak 3 originates from a population of appressed membranes in all strains, and Peaks 4 and 5 originate from single thylakoid membrane layers with variable lumenal widths. Although raising further questions, these descriptions are the most comprehensive analysis of SANS/TEM data sets to date and may serve as a basis for future work in other photosynthetic membrane systems.
Our findings show that cyanobacterial cells undergo reversible structural changes in response to light and that these changes are significantly different among the phycobilisome antenna mutants. These intracellular rearrangements within the thylakoid system were reproducible between biological and technical replicates in Synechocystis cells in both the wild-type and mutant strains, suggesting that these responses are robust. Future measurements under different light qualities and intensities will probe these intracellular reorganizations further. In Synechocystis 6803, it is known that many components, including photoreceptors, can play a role in the response of cells to light, and probing the membrane changes in mutant strains lacking such components will be particularly informative. The ability to directly observe membrane rearrangement in response to light exposure and its dependence on antenna mutations point toward a rich arsenal of interdependent membrane regulation processes that allow cyanobacteria a rapid and efficient adaptation to changes in light intensity.