Species-dependent protoplast enlargement involves different types of vacuole generation in bacteria

Vacuole generation occurs frequently during the enlargement of bacterial protoplasts and spheroplasts. Gram-positive Enterococcus faecalis protoplasts and gram-negative Lelliottia amnigena spheroplasts had large and small vacuoles inside the cytoplasm, respectively. Although no vacuoles were found at the early stage of cell enlargement, all enlarged cells used in the microinjection procedures had vacuoles. The plasma membrane of L. amnigena was more flexible than that of E. faecalis. In addition, E. faecalis protoplasts had unique discoidal structures as well as spherical structures in the cytoplasm. Our findings showed that the number of vacuoles increased as the L. amnigena plasma membrane expanded and that the size of vacuoles increased as the E. faecalis plasma membrane expanded, suggesting that bacterial cell enlargement involved vacuole generation. Thus, biosynthesis of the plasma and vacuolar membranes was synchronous with the bacterial cell enlargement. Differences in the plasma membrane flexibility might influence the different types of vacuole generation.

The aim goal of our research was to generate a novel organism with a defined genome. We enlarged different bacterial cells for microinjection of various substances, for example, heterogeneous or designed genomic DNA. In order to generate the enlarged cells suitable for microinjection, it was essential to regulate cell enlargement and maintain the enlarged size of the cells. Although bacterial cells cannot enlarge due to the presence of cell wall (peptidoglycan), the spheroplasts or protoplasts could enlarge under suitable conditions [1][2][3][4][5][6][7][8] . In this study, the cells lacking peptidoglycan with an outer membrane are called spheroplasts, and those without an outer membrane are called protoplasts. Generally, bacterial cells are only a few micrometres in diameter and do not contain a vacuole. However, all enlarged cells used in the microinjection procedures had vacuoles. Thus, to achieve microinjection of various substances directly into the cell cytoplasm it is necessary to avoid microinjections into the vacuoles. We had a question whether vacuole generation can be regulated or not in the cell enlargement process. In this study, we investigated whether the bacterial cell enlargement involved vacuole generation or not.
Electron microscope observations revealed that the vacuoles in the E. coli and B. subtilis protoplasts were surrounded by single membranes, which did not contain DNA 2,3 . These vacuoles could be isolated from the protoplasts by the removal of the plasma membrane 2, 3 . This suggested that the mature vacuoles were not connected to the plasma membrane. The vacuolar membrane of the E. coli protoplasts had the plasma membrane proteins, but outer membrane proteins were not detected 2 . In addition, the respiratory chain and F 0 F 1 -ATPase were detected in the vacuolar membrane of the E. coli protoplast 2 and the respiratory chain was also detected in the B. subtilis protoplast 3 . These results showed that the vacuolar membrane components were same or very similar to the components of the plasma membrane. Furthermore, patch clamp analyses showed that the E. coli and B. subtilis vacuoles have everted membrane 2,3 . These results strongly suggested that the vacuolar membrane was generated by endocytosis of the plasma membrane 2,3 . Thus, it can be concluded that the vacuoles were generated from the plasma membrane and the vacuolar membrane was connected to the plasma membrane at the early stage of cell enlargement, following which the vacuoles grew independently of the plasma membrane.
In our laboratory, we incubated and enlarged bacterial protoplasts and spheroplasts in Difco Marine Broth 2216 (DMB) or modified DMB containing penicillin [4][5][6][7][8] . DMB contains four major metal salts, CaCl 2 , KCl, MgCl 2 , and NaCl. Thorough investigation of the compositions of medium metal salts showed that the presence of Ca 2+ or (the second vacuole) was faster than that of the senior vacuole (the first vacuole) and they eventually became the same size (Fig. 1).
In L. amnigena, the plasma membrane expansion was activated in the modified Marine Broth MMB3CaKMg, which led to inhibit a large periplasm generation (Supplementary Figs. 2b and 5). Surprisingly, the rate of enlargement was very rapid, and the collapse of the enlarged cell was also fast ( Supplementary Fig. 6a). Considering that Ca 2+ was more effective than Mg 2+ in D. grandis spheroplast enlargement 7 , we prepared a new broth, eMMB3CaKMg (62 mM CaCl 2 , 7.4 mM KCl, and 16.2 mM MgCl 2 ), which was a modification of MMB3CaKMg (Supplementary Table 1). As a result, the enlargement effect was higher in eMMB3CaKMg than MMB3CaKMg ( Supplementary Fig. 6). Although in MMB3CaKMg the spheroplasts began to collapse within 24 h of incubation, their enlargement continued after 24 h in eMMB3CaKMg ( Supplementary Fig. 6). Thus, we used enlarged cells incubated in MMB3CaKMg and eMMB3CaKMg for microinjection. The enlarged spheroplasts in eMMB-3CaKMg had the largest plasma membrane size and formed the highest number of vacuoles, which had various sizes ( Supplementary Fig. 6b). These results showed that metal salt composition affects plasma membrane expansion and generation of vacuoles.

Morphology comparison of E. faecalis and L. amnigena enlarged cells. An optical microscope
observation revealed that the vacuoles observed in the cytoplasm of the E. faecalis and L. amnigena cells had diameters larger than approximately 10 μm ( Supplementary Fig. 7). To elucidate the characteristics of E. faecalis and L. amnigena enlarged cells, DNA and membrane-staining experiments were performed (Fig. 2). Spherical vacuolar membranes in the cytoplasm were stained as well as the plasma membranes by FM4-64 in both E. faecalis and L. amnigena enlarged cells (Fig. 2). In addition, spherical vacuole membranes were also stained by Bocillin FL penicillin in E. faecalis, indicating vacuolar membranes contain a penicillin binding protein 13 as well as the plasma membrane (Fig. 3). We confirmed that no vacuoles from E. faecalis and L. amnigena contained any DNA (Fig. 2). Nucleoids were located in the cytoplasm outside of the vacuoles (Fig. 2). These results are consistent with the characteristics of the vacuoles in the enlarged E. coli spheroplasts 2 and B. subtilis protoplasts 3 . Thus, the vacuolar membranes of E. faecalis and L. amnigena probably have the same or similar components to the plasma membranes.
Based on the transmission electron microscopy images of the E. faecalis protoplasts at 65 h post-incubation and the L. amnigena spheroplasts at 24 h post-incubation, we measured the diameters of cells and vacuoles and counted the number of vacuoles in each cytoplasm (Fig. 4). These results were represented by scatter plot and histogram (Fig. 5). The diameter of cells and the number of vacuoles had a higher correlation in L. amnigena (r = 0.62, n = 154) than E. faecalis (r = 0.46, n = 48) (Fig. 5a,b). On the other hand, the diameter of cells and the www.nature.com/scientificreports www.nature.com/scientificreports/ diameter of vacuoles had a higher correlation in E. faecalis (r = 0.59, cell; n = 48, vacuole; n = 82) than L. amnigena (r = 0.2, cell; n = 19, vacuole; n = 506) (Fig. 5c,d). These results showed that in the enlargement process of the spheroplasts and protoplasts, the E. faecalis vacuoles increased the size and the L. amnigena vacuoles increased in number.
The transmission electron microscopy images showed many discoidal structures in enlarged E. faecalis protoplasts (Fig. 4a). The discoidal structures could not be observed by phase contrast microscopy ( Fig. 2a and Supplementary Fig. 7a). Such structures have not been observed in enlarged L. amnigena cells (Fig. 4b) or in any bacterial cells 2,3,7 .  www.nature.com/scientificreports www.nature.com/scientificreports/ permeability of vacuolar membrane. In order to evaluate whether the enlarged protoplasts and spheroplasts can be used for microinjection, a blue fluorescent protein (BFP) solution was microinjected into the cytoplasm of the E. faecalis protoplasts and L. amnigena spheroplasts using an Eppendorf TransferMan 4r micromanipulator ( Supplementary Fig. 8).
First, we used the enlarged E. faecalis protoplasts incubated in DMB. The protoplasts were flexible and viscous; therefore, they adhered to the glass slide to perform the microinjection without use of a holding pipette ( Fig. 6a and Supplementary Movie 1). The BFP solution was successfully microinjected into the cytoplasm of the remaining cell, as confirmed by fluorescence microscopy (Fig. 6b and Supplementary Fig. 9). In addition, fluorescence microscopy showed that BFP was contained in the cytoplasm but not in the vacuoles ( Fig. 6b and Supplementary  Fig. 9).
Next, we used enlarged L. amnigena spheroplasts incubated in eMMB3CaKMg. Because the spheroplasts were floating cells, a holding pipette was used ( Supplementary Fig. 10). The outer membranes of the enlarged spheroplasts were removed using a microinjection-needle ( Supplementary Fig. 10). The protoplasts were flexible and viscous; they adhered to the glass slide and microinjection could be performed without the use of a holding pipette ( Fig. 7a and Supplementary Fig. 10 and Movie 2). The BFP solution was successfully microinjected into www.nature.com/scientificreports www.nature.com/scientificreports/ the cytoplasm, as confirmed by fluorescence microscopy (Fig. 7b). Fluorescence microscopy showed that BFP was contained in the cytoplasm but not in the vacuoles, and no fluorescence was observed in the debris of cells broken during the injection (Fig. 7b). These results showed that both E. faecalis and L. amnigena enlarged cells could be used for microinjection and the vacuolar membranes were devoid of the fluorescent protein.
Flexibility of plasma membrane. Endocytosis is a system that takes up extracellular material by invaginating the plasma membrane. In our microinjection experiments, an artificial vacuole was generated by releasing BFP solution into the plasma membrane of enlarged L. amnigena spheroplasts ( Fig. 8

and Supplementary
Movie 3). We used enlarged L. amnigena spheroplasts incubated in MMB3CaKMg. After removing the outer membrane by inserting the microinjection-needle, we brought the microinjection-needle close to the protoplast while releasing the BFP solution (Fig. 8a,b). When the microinjection-needle contacted the protoplast, the plasma membrane expanded inside like a balloon and subsequently a vacuole was generated in the cell (Fig. 8c). The vacuole enlarged from 13.9 µm in diameter to 18.8 µm in diameter during a 26 second interval (Fig. 8c). Thus, BFP solution was being released at a rate of approximately 0.08 pl s −1 from the tip of the microinjection-needle. This vacuole existed after the microinjection needle was pulled out (Fig. 8d). Finally, the protoplast was collapsed (Fig. 8e). Surprisingly, after the protoplast collapse, the vacuole was maintained in the solution, indicating that this vacuole was not connected to the plasma membrane (Fig. 8f). The fluorescent microscope observation showed that the BFP existed in this vacuole (Fig. 8g). Thus, the plasma membrane of the enlarged L. amnigena spheroplast incubated in MMB3CaKMg has such significant flexibility and viscosity that the vacuole could be created using micromanipulator.
On the other hand, no artificial vacuoles were generated in the microinjection experiments using the enlarged E. faecalis protoplasts. This result showed that the high plasma membrane flexibility of L. amnigena allowed for artificial vacuole formation.

Discussion
For microinjection into cells, the cell size should be more than 15 μm in diameter. All of the enlarged spheroplasts and protoplasts used in microinjection have vacuoles in the cytoplasm of bacterial cells. Our findings indicated that vacuole synthesis continued during cell enlargement. These results imply that vacuole generation and synthesis are essential for bacterial spheroplast and protoplast enlargement and that plasma membrane expansion is associated with vacuolar membrane synthesis in bacterial protoplast enlargement. On the other hands, vacuoles Differential interference contrast microscopy images were captured using an Olympus IX73 microscope. Fluorescent microscopy images were captured using a Keyence BZ-X710 microscope. Scale bar = 50 µm.
are always observed in the enlarged bacterial spheroplasts or protoplasts (cell size > approximately 10 μm in diameter), but their role is uncertain. Most bacteria produce ATP using ATP synthase on the plasma membrane. Thus, the volume of cytoplasm may be limited due to the plasma membrane size. Our hypothesis is that vacuole enlargement may inhibit the increase of the cytoplasm volume.
Results of the nucleoid, phospholipid and the penicillin binding protein localization using DAPI, FM4-64, and Bocillin FL penicillin staining, respectively confirmed that the fundamental characteristics of the E. faecalis and L. amnigena vacuoles were consistent with those of E. coli and B. subtilis enlarged cells. On the other hand, Listeria L-forms form intracellular vesicles as reproduction elements by local phospholipid accumulation or membrane invagination 14,15 . The vacuoles of E. faecalis and L. amnigena are not intracellular vesicles in the same way that reproduction elements are because they lack DNA and phospholipid accumulation.
In the enlarged protoplasts of E. faecalis, spherical and discoidal structures were observed within 65 h of incubation (Fig. 4a). Time-lapse observation using bright field microscopy showed that a newly formed small vacuole has grows more quickly than the previously formed vacuole(s), strongly suggesting that the expansion speed of the smaller vacuole is faster than that of larger vacuoles in the enlargement process of E. faecalis protoplasts until those vacuoles reach the same size (Fig. 1). If the discoidal structures are precursors to spherical vacuoles, the vacuole generation is continuously occurring even after 65 h of incubation. We observed multiple small vacuoles when a microinjection needle was inserted into the E. faecalis protoplast (Supplementary Fig. 11). The discoidal structure may absorb the water of the injected solution and then transform into a spherical structure, we think.
We have shown that the enlarged cells suitable for microinjection could be produced by manipulating the metal salts amounts in the medium (Supplementary Figs. 3 and 5). In microinjection process, the L. amnigena enlarged cells incubated in DMB could not be used but the cells incubated in eMMB3CaKMg could be used, indicating that the metal ions changed the membrane characteristics. On the other hand, E. faecalis enlarged cells incubated in DMB could be used for microinjection, indicating that metal salts composition of DMB is suitable for forming micro-injectable membrane in E. faecalis. E. faecalis and L. amnigena showed that the fluorescent protein did not pass through the vacuolar membrane from the cytoplasm, which indicates that the vacuolar membrane has a substance-selecting function. In addition, the vacuolar membranes were stained by Bocillin FL penicillin, indicating that a penicillin binding protein exists in the vacuolar membrane as well as the plasma membrane. This suggests that the vacuolar membrane has similar components to the plasma membrane in E. faecalis and L. amnigena. However, the characteristics of E. faecalis and L. amnigena plasma membranes differed. In L. amnigena, when the microinjection needle was brought close to the plasma membrane while releasing the www.nature.com/scientificreports www.nature.com/scientificreports/ solution, the membrane invaded to the cytoplasm and endocytosis occurred (Fig. 8). By contrast, in E. faecalis, membrane invagination and endocytosis were never observed in the microinjection experiment. These results indicated that the membrane flexibility of E. faecalis is higher than that of L. amnigena. From the electron micrograph, the surface layer of E. faecalis is sharp, while that of L. amnigena is flaccid (Fig. 4). This supports our postulation that the L. amnigena plasma membrane is more flexible than the E. faecalis plasma membrane.
Medium components affect membrane properties in the bacterial cell enlargement process. In D. grandis, the metal salts and osmotic stabilizers affected the lipid composition of the membrane and the frequency of outer membrane fusion in the enlargement process 7,16,17 . Based on the previous studies and this study, the composition of metal salts in the incubation media affects the membrane synthesis during the bacterial spheroplast or protoplast enlargement. However, the metal salts that play an important role in cell enlargement differ among the bacterial species. For example, although enlargement of Deinococcus spheroplasts requires calcium ion or magnesium ion 7 , that of Lelliottia does not. Thus, the type of the metal salt varies depending on the bacterial species. Interestingly, the phase contrast microscope observation showed that D. grandis enlarged spheroplasts, which inhibited the plasma membrane expansion, do not generate vacuoles in the cytoplasm 7,16,17 . Thus, outer membrane expansion is not associated with vacuole generation. In contrast to this, our findings showed that composition and concentration of metal salts affect not only outer membrane but also plasma membrane expansion during the enlargement of L. amnigena spheroplasts. In addition, L. amnigena spheroplasts with larger plasma membranes had more vacuoles in the cytoplasm (Fig. 5b). This indicates that the L. amnigena plasma membrane synthesis is accompanied by vacuolar membrane synthesis in the process of the spheroplast enlargement. However, in the process of E. faecalis protoplast enlargement, each vacuole also became larger (Fig. 5c). This was confirmed by time-lapse observation using the bright field microscope (Fig. 1). Thus, the plasma membrane synthesis is also accompanied by vacuolar membrane synthesis in the E. faecalis protoplast enlargement.
The E. faecalis protoplasts and L. amnigena spheroplasts exhibit different types of vacuole generation during the cell enlargement. However, the biosynthesis of vacuolar and plasma membranes occurs at the same time during enlargement. At the early stage of cell enlargement, no vacuoles are generated 8 . In addition, spheroplasts of the purple bacteria Erythrobacter and Rhodospirillum could not generate vacuole were enlarged to a maximum of (f) Isolation of the vacuole containing BFP. (g) Fluorescent microscopy images of BFP. Differential interference contrast microscopy images were captured using an Olympus IX73 microscope. Fluorescent microscopy images were captured using a Keyence BZ-X710 microscope. OM, outer membrane; PM, plasma membrane. Scale bar = 50 µm. 7 μm in diameter 5,6 . In the E. faecalis protoplast enlargement, inhibition of the plasma membrane expansion leads to inhibition of vacuole generation 8 .
Moreover, vacuole of E. coli is morphologically similar to that of L. amnigena, and that of B. subtilis is similar to that of E. faecalis. The difference between gram-negative and gram-positive may affect vacuole generation. On the other hands, the number of vacuoles of E. coli is smaller than that of L. amnigena, and the vacuole of B. subtilis has less force to push the cell membrane than that of E. faecalis. As mentioned above, the enlarged gram-negative D. grandis has larger outer membrane than plasma membrane and does not generate vacuoles, which completely differs from E. coli and L. amnigena 7 . Thus, the vacuole type has a morphological variation among different species of bacteria. The vacuole type may depend on the plasma membrane character. The vacuoles may function in the enlarged cells, since the vacuolar membrane has the plasma membrane proteins 2,3 . In addition, in nature, the largest bacterium, Thiomargarita namibiensis also generates an unusually large vacuole and accumulates nitrate in the vacuole 18,19 . The bacteria contained sulfur in the cytoplasm 18,19 . The vacuole function is the separation of nitrate and sulfide 18 . Our incubation medium for enlargement do not contain sulfur, therefore, the physiological function differs between the enlarged protoplasts and T. namibiensis large cells. In general, vacuoles are not observed in normal bacterial cells, suggesting that vacuole is not needed for bacterial normal growth. On the other hand, vacuole may be essential for maintaining giant bacterial cells.
Methods preparation and cultivation of spheroplast or protoplast. A single colony of L. amnigena