Dynamic localization of membrane proteins in Bacillus subtilis

A. S. Johnson{dagger}, S. van Horck and P. J. Lewis

School of Environmental and Life Sciences, Biological Sciences, University of Newcastle, Callaghan, NSW 2308, Australia

Correspondence
P. J. Lewis
Peter.Lewis{at}newcastle.edu.au


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The subcellular localization of membrane proteins in Bacillus subtilis was examined by using fluorescent protein fusions. ATP synthase and succinate dehydrogenase were found to localize within discrete domains on the membrane rather than being homogeneously distributed around the cell periphery as expected. Dual labelling of cells indicated partial colocalization of ATP synthase and succinate dehydrogenase. Further analysis using an ectopically expressed phage protein gave the same localization patterns as ATP synthase and succinate dehydrogenase, implying that membrane proteins are restricted to domains within the membrane. 3D reconstruction of images of the localization of ATP synthase showed that domains were not regular and there was no bias for localization to cell poles or any other positions. Further analysis revealed that this localization was highly dynamic, but random, implying that integral membrane proteins are free to diffuse two-dimensionally around the cytoplasmic membrane.


Abbreviations: FRAP, fluorescence recovery after photobleaching

The online version of this paper at http://mic.sgmjournals.org has supplementary movie files and a supplementary figure.

{dagger}Present address: Life Therapeutics, PO Box 6126, Frenchs Forest, NSW 2086, Australia.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Biological membranes are not homogeneous structures, but consist of domains enriched for specific lipids and proteins often referred to as lipid rafts (Edidin, 2003). These rafts appear to be highly dynamic structures, in terms of both lipid/protein content and size (Edidin, 2003). Although lipid rafts and membrane heterogeneity have been intensively studied in eukaryotic systems, relatively little work has been carried out with microbial systems, probably in part because of their small size. In recent years, however, convincing evidence has been presented indicating the presence of lipid domains in microbial cytoplasmic membranes (Fishov & Woldringh, 1999; Mileykovskaya & Dowhan, 2000; Kawai et al., 2004). In particular, the use of specific fluorescent probes has indicated that the phospholipid cardiolipin localizes preferentially to the cell poles and division sites (mid-cell regions) of both Gram-positive and Gram-negative bacteria (Mileykovskaya & Dowhan, 2000; Kawai et al., 2004). Further experiments using 2-pyrene derivatives of the phospholipids phosphatidylethanolamine (PE) and phosphatidyleglycerol (PG) indicated that the two lipids segregate into distinct domains in both Gram-positive and Gram-negative cell membranes, and that the PE domains were enriched for protein content (Vanounou et al., 2003). Despite the similarity of results obtained with Gram-positive (Bacillus subtilis) and Gram-negative (Escherichia coli) model systems, the two organisms have quite different proportions of common phospholipids in their cytoplasmic membranes. In B. subtilis PE accounts for approximately 50 % of phospholipid, PG 15 %, the lysine ester of PG 2·4 %, cardiolipin (CL) 0·8 %, and mono- di- and triglucosyldiacylglycerols 30 % (de Mendoza et al., 2002), whereas in E. coli PE accounts for about 70–80 % of phospholipid, PG 15–25 % and CL 5–10 % (Kadner, 1996).

Do integral membrane proteins also segregate into domains within the cytoplasmic membrane? The presence of chemoreceptors at cell poles is well established (Maddock & Shapiro, 1993) and this may reflect a preference of these proteins for acidic lipids such as cardiolipin. Penicillin-binding proteins involved in cell wall synthesis have also been shown to localize in several patterns, including to division septa and discrete foci along the long axis of the cell, reflecting the role of these proteins in specific stages of cell wall synthesis (Scheffers et al., 2004). However, analysis of the localization of the E. coli Sec protein secretion machinery, BglF sugar sensor and B. subtilis phage {pi}29 DNA replication protein p16.7 indicates that these proteins are homogeneously distributed around the cytoplasmic membrane, with no reported preference for cell poles or other observable domains (Brandon et al., 2003; Lopian et al., 2003; Meijer et al., 2001). Certainly, the images presented in these papers indicate that the localization pattern of these proteins creates a clear outline of the cytoplasmic membrane.

We have examined the localization of ATP synthase and succinate dehydrogenase, which carries out steps in both the tricarboxylic acid cycle and the electron-transport chain (as complex II), and re-examined the localization of p16.7 using a series of high-resolution fluorescence microscopy techniques with fluorescent protein fusions in live B. subtilis cells. We found that all of the proteins localized around the cytoplasmic membrane heterogeneously, and appeared to be free to move randomly throughout the membrane. We propose that such localization to domains is a general feature of integral membrane proteins.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Strains and growth conditions.
Bacterial strains and plasmids used in this study are listed in Table 1. All cloning was performed using E. coli DH5{alpha} (Gibco-BRL). Unless otherwise stated B. subtilis strains were cultured in CH medium (Sharpe et al., 1998) supplemented with 0·5 % (w/v) xylose at 37 °C. Transformation of B. subtilis was carried out by the method of Anagnostopoulos & Spizizen (1961) as modified by Jenkinson (1983) on nutrient agar plates supplemented with the appropriate antibiotics and inducers: chloramphenicol, 5 µg ml–1; spectinomycin, 50 µg ml–1; xylose 0·5 % (w/v); IPTG 0·5 mM. The functionality of atpA– and sdhA–gfp fusions was checked on succinate minimal media plates (SMM: Santana et al., 1994) supplemented with the appropriate antibiotics and inducers, and grown at 37 °C.


View this table:
[in this window]
[in a new window]
 
Table 1. Strains and plasmids

 
The vital membrane stain FM4-64 (Molecular Probes) was added to cultures at a final concentration of 30 ng ml–1 30 min prior to visualization.

DNA manipulations.
The complete atpA gene was PCR amplified using the following primers: forward 5'-AGAAGGGGTGAACTCGAGGTGAGCATCAAA-3' and reverse 5'-TCATCAGCATTCGAATTCGTTGCTTGGTGC-3'. The resulting product was digested with MunI and EcoRI and the 840 bp 3' fragment inserted into EcoRI-digested pSG1164. One recombinant with the fragment inserted in the correct orientation was named pNG24 (Table 1).

The 3' 1073 bp of sdhA were PCR amplified using the following primers: forward 5'-CGTATTATGGGTACCGAGAATTCATTC-3' and reverse 5'-GTTCACTCATGACTCGAGCGCCACCTTCTT-3'. The PCR product was digested with Acc651 and XhoI and inserted into similarly cut pSG1164, and the recombinant plasmid was named pNG107 (Table 1).

DNA manipulations for the construction of CFP, YFP and IPTG-inducible fusions to atpA and sdhA are indicated in Table 1.

Microscopy and image processing
Image acquisition.
Images were obtained using a Zeiss Axioscop 2 epifluorescence microscope fitted with a Quantix 1401ECCD camera (Photometrics), 100 W Hg lamp source and a 100x ApoPlan NA 1.4 lens. GFP fluorescence was visualized with set 41018, CFP with set 31044V2, YFP with set 41029, and FM4-64 was visualized using set 61000v2SBX with a 61560 TRITC exciter (all sets from Chroma Technology). Cells were mounted onto 1·2 % agarose pads as described by Glaser et al. (1997) or within Gene Frames (Advanced Biotechnologies) as described by Lewis et al. (2000) for microscopy. Image acquisition was performed using MetaMorph version 5.0 (UIC); backgrounds were subtracted, and out-of-focus light removed, using the Nearest Neighbours deconvolution drop-in. Linescans were performed in MetaMorph and data transferred to Microsoft Excel for further analysis. Final figures were prepared for publication using Adobe Photoshop version 7.0.

3D imaging and deconvolution.
Image stacks at 20 nm incremental steps were collected using a Pifoc PI P-721.10 microscope focus drive (Physik Instrumente) controlled via MetaMorph. Background subtractions and image stack alignments were carried out in MetaMorph and image stack deconvolution in AutoDeblur version 6.0 (AutoQuant Imaging). 3D reconstruction was performed in MetaMorph.

Time lapse.
For time-lapse microscopy, exponentially growing cells were placed onto 1·2 % agarose pads in Gene Frames and a suitable field of cells was located by phase-contrast microscopy. Cells were then imaged using the ACQUIRE TIMELAPSE drop-in from MetaMorph with a sampling interval of 1 min. Stacks were aligned and images processed as detailed above prior to analysis.

Fluorescence recovery after photobleaching (FRAP).
FRAP was carried out using a Zeiss LSM 510 confocal microscope fitted with an Orca ER CCD camera (Hammamatsu) using a 100x PlanApo NA1.3 objective to view exponentially growing cells mounted on an inverted agarose pad. GFP fluorescence was viewed using the 488 nm laser line at 10 % full power. A region of interest was bleached using 40 ms pulses at 40 % power over a 2 s exposure period. One-second exposures were obtained at the intervals specified in the text during photobleaching recovery. Twelve-bit images were analysed in MetaMorph.


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
atpA– and sdhA–gfp fusions are functional
It is important to ascertain whether a fluorescent protein fusion is functional, partially functional or non-functional when making conclusions about observed localization patterns. This is particularly important when considering the localization pattern of fusions involved in the formation of multi-protein complexes such as ATP synthase and succinate dehydrogenase. Both atpA and sdhA, to which fusions were made, lie within operons, encoding ATP synthase and succinate dehydrogenase, respectively, and so on single crossover of fusions into the chromosome it was important to ensure the expression of genes downstream of the insertion through use of either xylose- or IPTG-inducible promoters (Lewis & Marston, 1999). The growth of strains BS23 (atpA–gfp), BS112 (sdhA–gfp), BS121 (sdhA–yfp) and BS133 (atpA–cfp sdhA–yfp) (Table 1) was tested on SMM agar plates. With succinate as the only carbon source, these strains should be dependent on a functional electron-transport chain and ATP synthesis via ATP synthase. Growth of the strains on SMM supplemented with either xylose or IPTG is shown in Fig. 1(A, B). Xylose was required for induction of expression of genes downstream of the fluorescent protein fusions in strains BS23, BS112 and BS133, whereas IPTG was required in strain BS121. In the presence of xylose, extensive growth of strains BS23, BS112 and BS133 was observed (Fig. 1A), indicating expression of the fusion, and the genes downstream in both the atp and sdh operons led to formation of functional protein complexes. These results were confirmed by growing strain BS23 in liquid SMM (Fig. 1C). In the presence of xylose, the fusion strain grew with very similar kinetics to the control strain (168; td approximately 85 min; not shown), whereas very little growth was observed in the absence of xylose. In addition, Western blots of whole-cell extracts together with cytoplasmic and membrane fractions showed that there was no detectable degradation of the fusions, which localized exclusively to the membrane fraction (data not shown). Virtually no growth of BS121 containing an IPTG-inducible promoter downstream of an sdhA–yfp insertion was observed (Fig. 1A), indicating the dependence of these strains on the formation of functional enzyme complexes for growth. Conversely, when these strains were grown on SMM supplemented with IPTG instead of xylose, growth was only observed for strain BS121 (Fig. 1B).



View larger version (51K):
[in this window]
[in a new window]
 
Fig. 1. Construction of functional GFP fusions to atpA and sdhA. (A) and (B) show the clockwise arrangement of strains BS23 (atpA–gfp), BS112 (sdhA–gfp), BS121 (sdhA–yfp) and BS133 (atpA–cfp and sdhA–yfp) grown on SMM plates supplemented with xylose (A) or IPTG (B), grown at 37 °C. Strains BS23, BS112 and BS133 show xylose-dependent growth whereas strain BS121 shows IPTG-dependent growth. (C) Growth curve of strain BS23 grown in SMM liquid medium in the presence ({blacksquare}) or absence ({blacktriangleup}) of 0·5 % (w/v) xylose.

 
Together, these results indicate that functional fusions to both AtpA and SdhA were constructed. It was possible that xylose could be used as an alternative carbon source in these strains and so permit growth by bypassing the requirement for functional electron transport and ATP synthase. However, the lack of growth of strain BS121 on SMM supplemented with xylose, and growth of the same strain on SMM supplemented with the non-metabolizable galactose analogue IPTG, indicate that this does not occur.

Heterogeneous localization of integral membrane proteins
The vital membrane stain FM4-64 has been used for many years as an indicator of the cell boundaries and cell membrane in cell biology and shows the membrane to be a smooth homogeneous structure enveloping the cytoplasm (e.g. Pogliano et al., 1999; Lewis et al., 2000). When exponentially growing cells of strain 168 (Table 1) were stained with FM4-64 a homogeneous band outlining the cells was observed, and this was confirmed when a linescan through the top border of the cells was performed (Fig. 2A). The peaks in the trace correspond to the division septa that contain two membranes and so appear more heavily stained. When exponentially growing cells of strains BS24 (atpA–gfp) and BS112 (sdhA–gfp) were visualized using the same image processing techniques a much more heterogeneous staining pattern was observed (Fig. 2B, C). To test the prevalence of this phenomenon, the localization of a completely unrelated integral membrane protein, p16.7 from the B. subtilis phage {pi}29, was tested. In the absence of other phage proteins, p16.7 is simply a small integral membrane protein with no known role in B. subtilis physiology. The localization of a p16.7–GFP fusion has been previously reported and shown to be confined to the cytoplasmic membrane in a pattern very similar to FM4-64 (Meijer et al., 2001). However, we found that the p16.7–GFP localization pattern was indistinguishable from that of either SdhA–GFP or AtpA–GFP (Fig. 2D), suggesting that this punctate distribution pattern was most likely a general feature for the localization of integral membrane proteins. An explanation for the apparently different localization pattern for p16.7 found by us and by Meijer et al. (2001) is given in the Discussion.



View larger version (45K):
[in this window]
[in a new window]
 
Fig. 2. Subcellular localization of membrane proteins. The left-hand panels show fluorescence micrographs and the right-hand panels show linescans through the uppermost membrane margin of chains of four cells from the corresponding micrograph. Arrows indicate the start points for the linescans. (A) Cells stained with the membrane dye FM4-64; (B) cells expressing the atpA–gfp fusion; (C) cells expressing the sdhA–gfp fusion; (D) cells expressing the p16.7–gfp fusion.

 
Although the above results indicate that membrane proteins preferentially localize to submembranous domains, they do not tell us whether this localization is to the same or different domains. The rather punctate pattern of distribution was intriguing, and it was possible that electron transport (succinate dehydrogenase) and ATP synthesis were closely juxtaposed within the cell due to their connected roles in energy generation. There is some circumstantial evidence for close juxtapositioning of these enzymes, as it is possible to isolate respiratory chain supercomplexes containing various combinations of components of the electron-transport chain and ATP synthase (Schägger, 2002). In order to address this issue, dual-labelled cells were constructed containing atpA–cfp and sdhA–yfp fusions (strain BS133; Table 1). In this strain the fusions were inserted by single crossover into the chromosomal locus, similar to their respective single-labelled strains, with expression of genes downstream of both atpA and sdhA dependent on the presence of xylose (see Fig. 1A, B). For reasons we cannot explain, this appeared to be the only combination of fusions that gave reproducibly bright signals, despite numerous attempts to construct strains in which one fusion was expressed ectopically from the amyE locus or when one of the fusions was placed under the control of the IPTG-inducible Pspac promoter, which we know produces functional fusions (Fig. 1A, B) and gives bright signals in single-labelled strains (not shown). Nevertheless, chromosomal PCR and fluorescence microscopy checks indicated that strain BS133 contained the correct constructs in the right chromosomal context, and growth on SMM supplemented with xylose indicated that the strain produced functional fluorescent protein fusions (Fig. 1A).

Fluorescence micrographs of strain BS133 grown in CH medium supplemented with 0·5 % (w/v) xylose at 37 °C are shown in Fig. 3. Fluorescence crossover experiments confirmed that there was no detectable crossover of CFP signal into the YFP channel and vice versa (not shown). The fluorescence patterns of both CFP-labelled ATP synthase and YFP-labelled succinate dehydrogenase appear very similar to that observed for the single-labelled strains (compare Fig. 2B and C with Fig. 3A and B). The ATP synthase signal was pseudocoloured red, while the succinate dehydrogenase was pseudocoloured green in the overlay shown in Fig. 3(C). Usually, when red and green signals of approximately equal intensity coincide in overlays, a yellow colour is observed. While yellow can be seen in the overlay, there are clearly regions of either green or red signal, implying that there was relatively little colocalization of the signals. However, careful examination of the linescan shown in Fig. 3(D) indicates that there were substantial regions of signal overlap, but that there was considerable variation in the intensity of red or green signal at some regions of overlap that may have given rise to the appearance of a red or green region in the overlay in Fig. 3(C) (see regions arrowed in Fig. 3D). However, the linescan also shows that the signals are not perfectly coincident, with considerable regions of signal heterogeneity. Analysis of multiple fields of cells indicated a mean level of 62 % signal overlap with respect to the number of AtpA-CFP peaks. Overall, we feel these results are consistent with the localization of both proteins to submembranous regions, with both enzymes present in similar, at least partially overlapping, domains.



View larger version (33K):
[in this window]
[in a new window]
 
Fig. 3. Colocalization of AtpA–CFP and SdhA–YFP. (A) AtpA–CFP fluorescence; (B) SdhA–YFP fluorescence; (C) an overlay of the two signals. AtpA–CFP has been pseudocoloured red and SdhA–YFP pseudocoloured green. A linescan around the pair of rightmost cells indicated with the arrow using the same colour assignment is shown in (D); the arrows above the linescan indicate regions of signal colocalization.

 
Integral membrane proteins localize in large amorphous patches
A more detailed analysis of the localization of integral membrane proteins was carried out using image stack deconvolution and 3D reconstruction. Results are presented for the AtpA–GFP fusion, but SdhA and p16.7 fusions have similar localization patterns.

A 3D reconstruction of ATP synthase localization in a pair of cells in a series of orientations is shown in Fig. 4(A) and as an animation in supplementary movie 1 with the online version of this paper at http://mic.sgmjournals.org. The 3D rendition clearly shows heterogeneously distributed ATP synthase over the membranes. There is no clearly ordered organization of the enzyme; rather it appears to be distributed in a series of discrete foci and larger heterogeneous patches. A more detailed analysis of the image stacks was used to confirm this observation. Three image slices taken from positions indicated in the top right panel corners are shown in Fig. 4(B). A slice from towards the top of the cell is shown in red, from the middle in green, and towards the bottom in blue. These three slices were overlaid (Fig. 4B bottom right panel) and a linescan made around the periphery of the pair of cells (Fig. 4C). The arrows in Fig. 4(B) and (C) indicate the position where the linescan starts. A region we interpret as representing a large patch of ATP synthase that extends rightwards from the top of the cell to the bottom is indicated by the ‘a’ in Fig. 4(B, C). A more defined region that we interpret as representing a focus is indicated by the ‘b’ in Fig. 4(B, C). This analysis confirmed the heterogeneity of localization of ATP synthase and there was no evidence for a specific pattern of protein localization, such as aggregation at cell poles/mid-cells or in spirals that could reflect the distribution of recently described actin-like cytoskeletal filaments (see Discussion; Jones et al., 2001; Kruse et al., 2003).



View larger version (28K):
[in this window]
[in a new window]
 
Fig. 4. 3D localization of ATP synthase. (A) A contrast-inverted 3D rendering of ATP synthase localization. The numbers indicate the degrees of rotation of the image. (B) Three processed image slices taken from near the top (red), middle (green) and bottom (blue) of the cell. An overlay of the three slices is shown in the fourth panel. (C) A linescan around the margin of the overlaid image; the arrow indicates the start point of the scan. Colour coding is the same as in (B). In (B) and (C), ‘a’ and ‘b’ refer to a diffuse domain and discrete focus, discussed in more detail in the text.

 
Protein mobility within the membrane
Given the heterogeneous localization of proteins in the membrane, we wanted to know whether the observed localization represented protein restricted to specific domains within the cell membrane, or whether proteins were free to diffuse around the membrane. Time-lapse movies of the AtpA–GFP fusion strain were obtained at 1 min intervals over 30 min. A full movie sequence is provided in supplementary movie 2 with the online version of this paper (http://mic.sgm.journals.org), and a more detailed analysis of a 4 min period of the time-course is shown in Fig. 5. It is clear from both the movie and Fig. 5 that the localization of AtpA–GFP is highly dynamic within the membrane. Images taken at 0, 2 and 4 min into the time-lapse are shown in red, green and blue, respectively in Fig. 5(A) along with an overlay of the three images. The lack of coincidence of the signals in the overlay panel presents a graphic illustration of the movement of AtpA–GFP over the imaging period. A linescan around the cell, starting at the point indicated by the arrow, is shown in Fig. 5(B). The movement of a small focus of AtpA–GFP, marked ‘a’, ‘b’ and ‘c’ in the red, green and blue panels, respectively, was mapped for further analysis (Fig. 5A, B). This particular focus appears to start near the cell pole (a), diffuse slightly and move away from the pole (b) then move back towards the pole (c), suggesting that movement was due to random diffusion through the membrane. However, interpretation of these images is complicated, with unequivocal assignment of foci in each panel difficult. It may be that the focus mapped in ‘b’ and ‘c’ starts at ‘a'’ rather than ‘a’ (Fig. 5A, B). If this is the case, then the time-lapse images suggest that movement is unidirectional within the membrane. However, it is not possible to draw such conclusions from this experiment as it is unlikely that all the images were obtained in exactly the same focal plane. As indicated by the 3D reconstruction images in Fig. 4, a small change in focal plane could result in focusing on a very different distribution pattern for AtpA–GFP. In order to resolve the issue of whether movement within the membrane was random or directed, we performed fluorescence recovery after photobleaching (FRAP) experiments using a confocal microscope system. Following bleaching of a portion of a cell, the recovery of fluorescence could be monitored to determine if it is uni- or bi-directional. Unidirectional recovery would imply that Atp–GFP moved around the observed plane of the cell in one direction, whereas bi-directional recovery implies random movement of Atp–GFP.



View larger version (29K):
[in this window]
[in a new window]
 
Fig. 5. ATP synthase is highly mobile within the membrane. (A) Images of ATP synthase localization taken at the start (red), and 2 min (green) and 4 min (blue) into a time-lapse series. An overlay of the three images is shown in the fourth panel. (B) A linescan around the margin of the overlaid image, starting at the point indicated by the arrow. Colour coding is the same as in (A). Letters ‘a’, ‘a'’, ‘b’ and ‘c’ are referred to in more detail in the text.

 
The results of a FRAP experiment are shown in Fig. 6. Images of cells before, immediately following, 3 and 6 min after photobleaching are shown in Fig. 6(A). Linescans from the top right cell in Fig. 6(A) are shown in Fig. 6(B), with the arrow indicating the start of the linescan. Linescans were made down the left-hand (L) and right-hand (R) sides of the cell and are shown in Fig. 6(B). The colour scheme used in Fig. 6(A) is retained in Fig. 6(B) for comparative purposes with the region that was bleached lying to the right of the dashed line in the linescans. The cells indicated by ‘a’ in Fig. 6(A) were totally bleached, and the lack of fluorescence recovery in these cells indicates that there was no detectable de novo synthesis of AtpA–GFP over the course of this experiment.



View larger version (50K):
[in this window]
[in a new window]
 
Fig. 6. FRAP of the AtpA–GFP fusion. (A) A group of cells analysed in a FRAP experiment. Cells prior to bleaching are shown in red, immediately after bleaching in green, 3 min post-bleach in blue, and 6 min post-bleach in cyan. (B) Linescans of the cell indicated by an arrow in (A) down the left and right hand margins, indicated by L and R, respectively in (A). The colour coding is the same as in (A). The letter ‘a’ indicates a pair of cells that were totally photobleached and failed to redevelop fluorescence during the experiment (see text for more details). The dotted line in the linescans in (B) indicates the approximate rightmost limit of photobleaching.

 
Following photobleaching, the intensity of fluorescence in the bottom half of the analysed cell dropped considerably (compare the red and green panels in Fig. 6A and the red and green linescans in Fig. 6B). At 6 min following photobleaching, fluorescence recovery throughout the cell appeared to be largely complete (cyan panel and linescan in Fig. 6A and B). The image taken 3 min following photobleaching represents an intermediate stage between bleaching and recovery (blue panel and linescan Fig. 6A and B). Comparison of the linescans through the left- and right-hand side of the cell indicated that the rate of fluorescence recovery was approximately the same on both sides (Fig. 6B), which was observed in all FRAP experiments. This result suggests that AtpA–GFP movement within the membrane is most likely due to random diffusion.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
We have shown in this study that microbial membrane proteins appear to be restricted to domains rather than being homogeneously distributed within the cytoplasmic membrane. This appears to be a general property, which was analysed with GFP-tagged subunits of ATP synthase (AtpA) and succinate dehydrogenase (SdhA) as well as the B. subtilis phage {pi}29 protein p16.7. All the proteins appeared to have a similar heterogeneous pattern of localization that was strikingly different from the distribution of total membrane when observed with the membrane stain FM4-64. The localization pattern for p16.7 has previously been reported and described as being similar to that of FM4-64 (Meijer et al., 2001). Indeed, in this work we used the same strain, 110WA, to observe p16.7 localization. However, we do not feel that our results contradict those previously reported; rather, our analysis and interpretation has been more focused on the signal heterogeneity. In addition, our images were processed to remove out-of-focus light (see Methods) whereas Meijer et al. (2001) used unprocessed images. Both sets of results indicate that p16.7 is distributed within the cytoplasmic membrane, and it is possible to observe signal heterogeneity in the insert box of Fig. 5 in Meijer et al. (2001), indicating that strain 110WA gives similar results in our hands to those of Meijer et al. (2001).

Dual-labelling experiments with AtpA–CFP and SdhA–YFP fusions suggested that membrane proteins localize to approximately similar submembranous domains, as observed by the partial colocalization of AtpA-CFP and ShdA-YFP signals (Fig. 3). However, the degree of colocalization of these signals also indicated that it is very unlikely that ATP synthase and succinate dehydrogenase aggregate together in specific ‘energy-generating’ domains, but rather that membrane proteins segregate to domains that are approximately similar. Heterogeneities observed between the two signals could be due to differences in phospholipid preferences for the membrane complexes and/or the levels of expression of the genes. ATP synthase fusions were extremely bright in comparison to the other fusions, possibly indicating a higher level of expression of this operon, and it has been reported that the level of expression of the atp operon is extremely high (Santana et al., 1994).

3D reconstruction of ATP synthase distribution indicated that protein domains varied between small concentrated foci to rather large diffuse domains that spread over a significant proportion of the cell (Fig. 4 and supplementary movie 1). These imaging experiments were performed in live, non-fixed cells and so it is possible that the domains were moving during the image acquisition process. This did probably occur, although we feel that movement was not large during the acquisition process. At 200 ms exposures, 60 images could be acquired in little over 12 s. As seen in Figs 5 and 6, movement was monitored over a period of minutes, not seconds. Time-lapse movies taken with short (5 s) gaps showed little observable movement of protein domains over a 10–15 s period (see supplementary Fig. S1 at http://mic.sgm.journals.org). Therefore, we feel the results presented in Fig. 4 and supplementary movie 1 do provide a reasonable snapshot of ATP synthase and other integral membrane protein distribution around the membrane.

We also analysed the dynamic distribution of ATP synthase in more detail, and time-lapse microscopy experiments showed that the irregular protein domains were highly mobile within the membrane (Fig. 5). It was not possible to conclude from these experiments whether movement within the membrane was unidirectional (specific) or bidirectional (random). Further experiments (Fig. 6) monitoring the recovery of fluorescence around the membrane in cells where half the membrane was bleached indicated that movement of the protein domains was most likely by random diffusion.

In eukaryotic cells, lipid rafts are known to have contacts with the actin cytoskeleton via actin-binding proteins such as annexins (Edidin, 2003), and many proteins are known to interact with actin filaments, including enzymes involved in glycolysis (Minaschek et al., 1992). It has recently become clear that bacteria also contain filamentous structures composed of proteins that belong to the actin superfamily, and that these cytoskeletal filaments form regular helical structures that are probably closely juxtaposed with the cytoplasmic membrane (Jones et al., 2001; Kruse et al., 2003). If there were discrete ‘energy-generating domains' within the cell, it is possible that glycolysis, the tricarboxylic acid cycle, electron transport and ATP synthase could be closely juxtaposed. However, we found no evidence for the proteins examined in this study localizing in a pattern consistent with a connection with cytoskeletal filaments. In addition, we found our protein patches to be highly mobile within the membrane, with substantial movements observed over a period of a few minutes. FRAP of the Mbl cytoskeleton in B. subtilis occurred over a period of tens of minutes (Carballido-López & Errington, 2003), indicating that the dynamics of protein movement in membranes and cytoskeletons occurs on different timescales. Most likely, integral membrane proteins are simply inserted into the membrane directly via translation from ribosomes with little or no further connection with the cytoplasm. We do know that ribosomes preferentially localize towards the cell poles (Lewis et al., 2000), and membrane protein localization does not reflect a similar polar bias. However, once inserted into the membrane, it is clear that most proteins (or those that do not have a specific function dependent on their subcellular localization) are free to diffuse within the membrane and so would have a localization pattern unrelated to their site of synthesis.

In conclusion, integral membrane protein localization is probably dependent on the lipid distribution within the membrane, and is highly dynamic and random. Given the apparent preference of these proteins for submembranous domains, it will be interesting to determine whether formation of these protein-rich domains is correlated with lipid raft distribution.


   ACKNOWLEDGEMENTS
 
This work was supported by the Australian Research Council through grants A00105184 and DP0449482 and University of Newcastle RMC grants to P. L. P. L. acknowledges the help of Scott Merrington and Eileen McLaughlin with confocal imaging and FRAP experiments.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Anagnostopoulos, C. & Spizizen, J. (1961). Requirements for transformation in Bacillus subtilis. J Bacteriol 81, 741–746.

Brandon, L. D., Goehring, N., Janakirman, A., Yan, A. W., Wu, T., Beckwith, J. & Goldberg, M. B. (2003). IcsA, a polarly localized autotransporter with an atypical signal peptide, uses the Sec apparatus for secretion, although the Sec apparatus is circumferentially distributed. Mol Microbiol 50, 45–60.[CrossRef][Medline]

Carballido-López, R. & Errington, J. (2003). The bacterial cytoskeleton: in vivo dynamics of the actin-like protein Mbl of Bacillus subtilis. Dev Cell 4, 19–28.[Medline]

de Mendoza, D., Schujman, G. E. & Aguilar, P. S. (2002). Biosynthesis and function of membrane lipids. In Bacillus subtilis and its Closest Relatives: from Genes to Cells, pp. 43–56. Edited by A. L. Sonenshine, J. A. Hoch & R. Losick. Washington, DC: American Society for Microbiology Press.

Edidin, M. (2003). The state of lipid rafts: from model membranes to cells. Annu Rev Biophys Biomol Struct 32, 257–283.[CrossRef]

Feucht, A. & Lewis, P. J. (2001). Improved plasmid vectors for the production of multiple fluorescent protein fusions in Bacillus subtilis. Gene 264, 289–297.[CrossRef][Medline]

Fishov, I. & Woldringh, C. (1999). Visualization of membrane domains in Escherichia coli. Mol Microbiol 32, 1166–1172.[CrossRef][Medline]

Glaser, P., Sharpe, M. E., Raether, B., Perego, M., Ohlsen, K. & Errington, J. (1997). Dynamic, mitotic-like behaviour of a bacterial protein required for accurate chromosome partitioning. Genes Dev 11, 1160–1168.[Abstract]

Jenkinson, H. (1983). Altered arrangement of proteins in the spore coat of a germination mutant of Bacillus subtilis. J Gen Microbiol 129, 1945–1958.[Medline]

Jones, L. J. F., Carballido-López, R. & Errington, J. (2001). Control of cell shape in bacteria: helical, actin-like filaments in Bacillus subtilis. Cell 104, 913–922.[CrossRef][Medline]

Kadner, R. J. (1996). Cytoplasmic membrane. In Escherichia coli and Salmonella: Cellular and Molecular Biology 2nd edn, pp 58–87. Edited by F. C. Neidhardt and others. Washington DC: American Society for Microbiology.

Kawai, F., Shoda, M., Harashima, R., Sadaie, Y., Hara, H. & Matsumoto, K. (2004). Cardiolipin domains in Bacillus subtilis Marburg membranes. J Bacteriol 186, 1475–1483.[Abstract/Free Full Text]

Kruse, T., Møller-Jensen, J., Løbner-Oelsen, A. & Gerdes, K. (2003). Dysfunctional MreB inhibits chromosome segregation in Escherichia coli. EMBO J 22, 5283–5292.[Abstract/Free Full Text]

Lewis, P. J. & Marston, A. L. (1999). GFP vectors for controlled expression and dual labelling of protein fusions in Bacillus subtilis. Gene 227, 101–109.[CrossRef][Medline]

Lewis, P. J., Thaker, S. D. & Errington, J. (2000). Compartmentalization of transcription and translation in Bacillus subtilis. EMBO J 19, 710–718.[Abstract/Free Full Text]

Lopian, L., Nussbaum-Schochat, A., O'Day-Kerstein, K., Wright, A. & Amster-Choder, O. (2003). The BglF sensor recruits the BglG transcription regulator to the membrane and releases it on stimulation. Proc Natl Acad Sci U S A 100, 7099–7104.[Abstract/Free Full Text]

Maddock, J. R. & Shapiro, L. (1993). Polar location of the chemoreceptor complex in the Escherichia coli cell. Science 19, 1717–1723.

Meijer, W. J. J., Serna-Rico, A. & Salas, M. (2001). Characterisation of the bacteriophage {pi}29-encoded protein p16.7: a membrane protein involved in phage DNA replication. Mol Microbiol 39, 731–746.[CrossRef][Medline]

Mileykovskaya, E. & Dowhan, W. (2000). Visualization of phospholipid domains in Escherichia coli by using the cardiolipin-specific fluorescent dye 10-N-nonyl acridine orange. J Bacteriol 182, 1172–1175.[Abstract/Free Full Text]

Minaschek, G., Groschel-Stewart, U., Blum, S. & Bereiter-Hahn, J. (1992). Microcompartmentation of glycolytic enzymes in cultured cells. Eur J Cell Biol 58, 418–428.[Medline]

Pogliano, J., Osborne, N., Sharp, M. D., Abanes-De Mello, A., Perez, A., Sun, Y. L. & Pogliano, K. (1999). A vital stain for studying membrane dynamics in bacteria: a novel mechanism controlling septation during Bacillus subtilis sporulation. Mol Microbiol 31, 1149–1159.[CrossRef][Medline]

Santana, M., Ionescu, M. S., Vertes, A., Longin, R., Kunst, F., Danchin, A. & Glaser, P. (1994). Bacillus subtilis F0F1 ATPase: DNA sequence of the atp operon and characterization of atp mutants. J Bacteriol 176, 6802–6811.[Abstract]

Schägger, H. (2002). Respiratory chain supercomplexes of mitochondria and bacteria. Biochim Biophys Acta 155, 154–159.

Scheffers, D.-J., Jones, L. J. F. & Errington, J. (2004). Several distinct localization patterns for penicillin-binding proteins in Bacillus subtilis. Mol Microbiol 51, 749–764.[Medline]

Sharpe, M. E., Hauser, P. M., Sharpe, R. G. & Errington, J. (1998). Bacillus subtilis cell cycle as studied by microscopy: constancy of cell length at initiation of DNA replication and evidence for active nucleoid partitioning. J Bacteriol 180, 547–555.[Abstract/Free Full Text]

Vanounou, S., Parola, A. H. & Fishov, I. (2003). Phosphatidylethanolamine and phosphatidylglycerol are segregated into different domains in bacterial membrane. A study with pyrene-labelled phospholipids. Mol Microbiol 49, 1067–1079.[CrossRef][Medline]

Received 8 April 2004; revised 27 May 2004; accepted 27 May 2004.



This Article
Abstract
Full Text (PDF)
Supplementary movies
Alert me when this article is cited
Alert me if a correction is posted
Citation Map
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Download to citation manager
Google Scholar
Articles by Johnson, A. S.
Articles by Lewis, P. J.
Articles citing this Article
PubMed
PubMed Citation
Articles by Johnson, A. S.
Articles by Lewis, P. J.
Agricola
Articles by Johnson, A. S.
Articles by Lewis, P. J.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
INT J SYST EVOL MICROBIOL MICROBIOLOGY J GEN VIROL
J MED MICROBIOL ALL SGM JOURNALS
Copyright © 2004 Society for General Microbiology.