Due to the high degree of autofluorescence it is difficult to study subcellular structures in cyanobacteria by fluorescence microscopy. However, we found that there is very little fluorescence in the blue fluorescence channel (UV excitation/blue emission, see material and methods) (Fig. 1A, right panel is scaled like the DAPI panel in Fig. 1E), and in the green channel (blue excitation/green emission) (Fig. 1B, right panel is scaled like the FM1-43 panel in Fig. 1C and like Fig. 1D). Thus, Synechocystis genomic DNA can be efficiently stained with DAPI (Fig. 1E, middle panel), and the green membrane stains FM1-43 as well as mitotracker can be used in the green filter (Fig. 1C, right panel and 1D). FM1-43 is unable to cross the cell membrane in eukaryotic cells and in Gram-positive bacteria, while mitotracker green can traverse membranes, and thus stains mitochondria, or engulfed forespores in Bacillus subtilis cells 14. Fig. 1C and 1D clearly show that FM1-43 and mitotracker green stain different parts of Synechocystis cells. While FM1-43 shows a uniform staining of the periphery of the cell (Fig. 1C, right panel), mitotracker stains the thylakoids, which have also strong autofluorescence in the red filter (green excitation/red emission) (Fig. 1C, left panel). To investigate if FM1-43 stains only the outer membrane, or both outer and cytoplasmic membranes in Synechocystis, several control experiments were performed. Firstly, cells were put to hyperosmotic conditions for 15 min. Because only the inner membrane shows plasmolysis bays in Gram negative bacteria, the presence of membrane-proximal fluorescent patches in Fig. 1G clearly shows that FM1-43 also stains the inner membrane. Secondly, inner membranes could clearly be seen as septa in the filamentous cyanobacterium Fischerella sp. (Fig. 1H). Along the lateral side of filaments, outer and inner membranes are seen as bright single membrane (they are spaced less than 250 nm apart and thus appear as single line), whereas between cells, only a weak membrane staining is visible, corresponding to the inner membrane that partitions single cells (Fig. 1H). Thirdly, in some Synechocystis cells, the thylakoid membranes have already separated between daughter cells, indicating the presence of a septum between daughter cells, while the outer membrane has not yet completely separated both cells (compare cells in Fig. 1F, indicated by white triangles, where the membrane signal between daughter cells is about twice as strong as on the outer surface of the cells, with cells in Fig. 1E). In rare cases, separation of thylakoids occurs when the outer membranes have not deeply invaginated between daughter cells (Fig. 1E), and the cell membrane between daughter cells is weakly but clearly visible, while staining of the thylakoids is similar to background fluorescence (compare Fig. 1E, upper and lower panels). Because proteins diffuse through membranes on a time scale of few μm per second (0.13 μm² s^-1) 15, diffusion of a membrane stain is extremely fast and cannot account for the differential staining of cell and thylakoid membranes. It was, however, possible that the FM1-43 fluorescence was not observed in thylakoid membranes due to an efficient quenching of the dye fluorescence by thylakoid membrane attached phycobilisomes. To exclude this possibility, we have isolated membranes from Synechocystis cells under high salt conditions, under which a large portion of the phycobilisomes remain attached to the thylakoid membranes. A fraction of these membranes was extensively washed with a low salt buffer with 0.5% of the detergent Triton X-100. With this buffer the phycobilisomes were almost completely removed from the thylakoid membranes, as proven by absorbance spectroscopy. When the dye FM1-43 was added to these membranes and the fluorescence emission of the membrane solution was measured in an Aminco Bowman Series II fluorimeter after excitation of the dye at 470 nm, we did not observe any differences in fluorescence emission and thus no quenching of the FM1-43 fluorescence by phycobilisomes (data not shown). Therefore, FM1-43 and mitotracker can distinguish between the cytoplasmic and intracellular membrane compartments, and these observations suggest that there is no direct link between cytoplasmic and thylakoid membranes, or else a weak staining of the thylakoids would be expected after addition of FM1-43.

The absence of a direct connection between the cytoplasmic membrane and thylakoids in Synechocystis suggests that parts of the cytoplasmic membrane may be invaginated and fused to thylakoid membranes for their continued synthesis during cell growth or for transfer of proteins between the two membranes. To test this idea, we added FM1-43 to the agarose pads containing growth medium, onto which cells were added to fix them in one plane. Continued supply of FM1-43 resulted in the formation of 1 to 4 (on average 2.3, with 65 cells analysed) fluorescent patches or dots close to the cytoplasmic membrane (Fig. 2A). In very rare cases, 5 patches were observed. These structures were only seen 20 min after addition of cells to the slides, and on average only after 30 min (Fig. 2A). The structures persisted for at least 2 hours, and did not move much within the cells, although different patterns between 10 min intervals were observed in 8 out of 55 cells analysed (see grey triangle in Fig. 2A). Fluorescence intensity of cells with continuous supply of FM1-43 increased over time, for at least 2 hours, and FM1-43 signals became apparent within thylakoid membranes in many (but not all) cells (Fig. 2C). However, care must be taken with the interpretation of the latter finding, as we have no means to investigate if the cells are still in their normal physiological state after 2 hours, although there is no strong argument against their well-being.

To obtain more information about the nature of the observed fluorescent bodies close to the cytoplasmic membrane, we obtained Z-series through the cells. Fig. 2B shows a representative 200 nm increment series through two cells. In the single round cell, two fluorescent patches are apparent, one below and one around the central plane. In the larger, almost fully divided cell, at least 4 patches can be seen, two below and two above the central plane. In general, larger cells contained more fluorescent patches than smaller cells, in agreement with their larger volume.

Labelling of cells with the fluorescent dyes DAPI or Hoechst 333 showed that the DNA occupies central intracellular spaces devoid of thylakoid membranes (in 92% of all cells, with >250 cells analysed), whereas the thylakoid membranes occupy the periphery of the cells (Fig. 3A). In the remaining 8% of the cells, DNA stain was present directly adjacent to the cell membranes. Usually, a single (occasionally irregularly shaped) nucleoid was present in round cells (Fig. 3A and first 2 panels of 2B), and a bilobed nucleoid in dividing cells (Fig. 3B, 3^rd and 4^th panel, Fig. 3C). Infrequently, single round cells contained 2 up to 4 distinct DAPI-stained regions (7.5% of all cells analysed), and were thus more complex than the majority of the cells. The morphologies of thylakoid membranes were highly heterogenic, thylakoid membranes occupied the cell periphery at every possible pattern (Fig. 3A–D).

We wished to obtain a quantitative view on the cell cycle in Synechocystis. To this end, cell morphologies were grouped into 5 different patterns, according to the progression through the cell cycle. Group 1 comprised entirely round cells, which must be early in the cell cycle. Group 2 was defined as cells that are elongated, and thus clearly growing but have no visible membrane constriction (because elongated cells can look round when they are vertical to the plane of the view, this group may be slightly underrepresented). Group 3 comprises cells that have a clear membrane invagination, and group 4 those cells with deep invagiations that are not yet closed. Cells in which outer membranes have completely surrounded both daughter cells, but which are still connected, were scored as group 5. Figure 3B shows representatives of each group, and reveals that most cells in an exponentially growing culture belong to group 1. 16% of the cells were clearly elongated, and thus initiating to divide, while 18% had clear membrane invaginations at the division plane, and were thus actively dividing. 3% of the cells had deep invaginations, while less than 2% were fully divided but still connected (with 437 cells analysed). These data indicate that Synechocystis cells grow and gradually divide, to quickly separate once outer membranes have fully separated daughter cells.

Only 1 out of 80 cells from group 3 had completely separated thylakoids, and none had 2 completely separated nucleoids. About 50% of the cells of group 4 had separated thylakoids and exhibited two distinct nucleoids, while all cells from group 5 had separated thylakoids and 2 nucleoids. All cells from group 4 that still contained non-separated thylakoids (i.e. 50%, Fig. 3C, middle panel) showed nucleoids that were still connected by thin DNA threads. These observations show that nucleoids are separated very late in the cell cycle, just before the division septum closes (as seen by separated versus non-separated thylakoids). This is in contrast to E. coli and B. subtilis cells, in which nucleoids are generally well separated before the septum closes 3. Fig. 3E shows an example of two B. subtilis cells in which nucleoids are completely separated well before a division septum is detectable.

Having several copies of the chromosome, Synechocystis cells may not require an elaborate chromosome segregation machinery as is operative in many bacteria that contain only a single chromosome (or two before cell division). DAPI can be used to quantify the amount of DNA in a cell; therefore, we sought to investigate if fluorescence intensity of DAPI can shed light on the question if chromosome segregation occurs in a tightly regulated fashion in Synechocystis. Because nucleoids occupy only the central portion of Synechocystis cells (while the outer parts are occupied by the thylakoid membranes, see above), their width (or diameter, which is equal to their depth because the cells are round) is comparable to the width and depth of nucleoids in B. subtilis cells. On average, Synechocystis nucleoids were 1.12 μm wide, while B. subtilis nucleoids were 0.96 μm wide, showing that there is no considerable difference in focal depth between nucleoids of the two organisms. Size and intensity of completely separated nucleoids (or those connected by a thin thread, see above) was frequently strikingly different between Synechocystis daughter cells (Fig. 3D, indicated by white triangles), in contrast to cells of B. subtilis (Fig. 3E). Intensities of separated nucleoids were measured in more than 100 cells using Metamorph 6 software. In contrast to B. subtilis cells, in which the scatter of staining intensity was relatively low, fluorescence intensity of Synechocystis nucleoids varied to a much higher degree (Fig. 4). In B. subtilis, the relative ratios of nucleoids per daughter cells varied between 0.76–1 with a mean of 0.94 and a standard deviation of 0.05, while in Synechocystis daughter cells, the relative ratio scattered from 0.4–1 with a mean of 0.85 and a large standard deviation of 0.13. These results prove that chromosome segregation occurs less stringently in Synechocystis than in B. subtilis, and suggest that possibly, segregation operates by a passive mechanism. This idea is supported by our finding that nucleoids are only completely separated when thylakoid membranes are also separated by the septum.

A glucose tolerant Synechocystis sp. PCC 6803 wild type strain was grown photomixotrophically in liquid BG11 medium 32 supplemented with 10 mM glucose at 34°C under 40 μE/m²s of cold white light. Cells were harvested in the mid-log growth phase for subsequent analyses. Fischerella sp. were grown in liquid BG11 without glucose at 28°C under 40 μE/m²s white light. For induction of plasmolysis, an exponentially growing cell culture was diluted 1:1 with a 1 M sucrose solution, yielding 0.5% sucrose final concentration.

Exponentially growing cells were mounted on 1% agarose pads containing growth medium for microscopy. FM1-43FX and Mitotracker green (Molecular Probes, Netherlands) were added to cells at 2.5 μg/ml or 5 μg/ml, respectively, DAPI (blue DNA-specific membrane permeable stain) was added at 200 ng/ml, and FM1-43FX was added at 2.5 μg/ml to agarose pads for continued supply of membrane stain. The red variant of FM1-43FX, FM4-64, stains the cell membrane of the Gram positive bacterium Bacillus subtilis, but not fully engulfed forespore, whereas Mitotracker green stains the membrane of fully engulfed spores 14, as well as mitochondria in eukaryotic cells, which are not stained by FM1-43FX. Thus, Mitotracker green is a membrane permeant stain, while FM1-43FX can not cross (inner) cell membranes. Images were acquired on an Olympus AX70 upright microscopy using a 100× objective (A = 1.45), a MicroMax CCD camera, and Metamorph 6.0 software (Universal Imaging, USA). For acquisition of Z-stacks, 20 planes with a spacing of 0.2 μm were taken from bottom to top, or reverse. Signal intensities were measured in Metamorph program. Filter specificities were as follows: "blue" ex360-370, dc400, em420-460, "green" ex460-495, dc505, em510-550, "red" ex480-550, dc570, em590. Usually, 100 ms exposure times were used.

1 l of Synechocystis cells were grown to an OD₇₀₀ of about 1. After centrifugation at 3000 rpm the cells were resuspended in 2 ml 750 mM K-Phosphate, 150 mM Tricine (pH 7.2) buffer and the cell were subsequently broken with a Mini BeadBeater (Biospec Products Inc.). Unbroken cells and cellular debris was removed by centrifugation at 3000 rpm (4°C), and the supernatant was further separated into membranes and soluble proteins by an ultracentrifugation step (40000 rpm, 30 min, 4°C). The sedimented membranes were either resuspended in high salt buffer (as above) or in a low salt buffer (50 mM Na-Phosphate, pH 7.2) to which 300 mM glycerol and 0.5% Triton X-100 were added. The solubilized membranes were sedimented again as described above. This washing step was repeated three times and the membranes were afterwards washed two times with low salt buffer in the absence of glycerol and detergent. By these washing steps most of the membrane attached phycobilisomes were removed. After determination of the chlorophyll concentrations of phycobilisome containing and phycobilisome-less membranes, UV/Vis spectra were recorded on a Perkin Elmer Lambda 35 UV/Vis spectrometer to proof removal of the phycobilisomes. To 1 ml membranes with a chlorophyll concentration of 10 μg/ml 1 μl FM1-43 was added and fluorescence emission spectra were recorded in the range between 480 – 750 nm after excitation of the dye at 470 nm. Fluorescence spectra were recorder at room temperature with an Aminco Bowman Series II fluorescence spectrometer.