Dynamin-Like Protein B of Dictyostelium Contributes to Cytokinesis Cooperatively with Other Dynamins

Dynamin is a large GTPase responsible for diverse cellular processes, such as endocytosis, division of organelles, and cytokinesis. The social amoebozoan, Dictyostelium discoideum, has five dynamin-like proteins: dymA, dymB, dlpA, dlpB, and dlpC. DymA, dlpA, or dlpB-deficient cells exhibited defects in cytokinesis. DlpA and dlpB were found to colocalize at cleavage furrows from the early phase, and dymA localized at the intercellular bridge connecting the two daughter cells, indicating that these dynamins contribute to cytokinesis at distinct dividing stages. Total internal reflection fluorescence microscopy revealed that dlpA and dlpB colocalized at individual dots at the furrow cortex. However, dlpA and dlpB did not colocalize with clathrin, suggesting that they are not involved in clathrin-mediated endocytosis. The fact that dlpA did not localize at the furrow in dlpB null cells and vice versa, as well as other several lines of evidence, suggests that hetero-oligomerization of dlpA and dlpB is required for them to bind to the furrow. The hetero-oligomers directly or indirectly associate with actin filaments, stabilizing them in the contractile rings. Interestingly, dlpA, but not dlpB, accumulated at the phagocytic cups independently of dlpB. Our results suggest that the hetero-oligomers of dlpA and dlpB contribute to cytokinesis cooperatively with dymA.


Introduction
The dynamin superfamily is a large GTPase family that is responsible for diverse cellular processes, including various membrane-remodeling events, such as the fusion and fission of intracellular trafficking vesicles and fusion and fission of large organelles, including mitochondria, chloroplasts, and peroxisomes. Other dynamin superfamily proteins are involved in membrane independent viral resistance of host cells [1,2]. Recent studies have shown that dynamin also contributes to cytokinesis. Mammals have three classical dynamins and several dynamin-like proteins. In HeLa cells, dynamin

Antibodies
Polyclonal antibodies against dlpA and dymA were described previously [17]. Polyclonal antibodies against dlpB were newly generated. A synthetic polypeptide (aa 790-808) of dlpB (CNYKKYSQSFSHPFPSAVRN) was used for immunizing rabbits, by a custom service (Sigma-Aldrich, Tokyo, Japan). These antibodies were absorbed with fixed dlpBcells, as previously described [28], and used after 100-fold dilution for immunostaining and 1000-fold dilution for western blot.
For immunostaining with anti-dlpB antibodies, cells were fixed by agar-overlay method as described previously [30]. The fixed cells were then incubated with anti-dlpB antibodies and later with Alexa 488-conjugated secondary antibodies (Thermo Fisher Scientific, Tokyo, Japan).

Microscopy
The DAPI-stained cells were observed using a fluorescence microscope (TE 300, Nikon, Japan) equipped with a regular UV filter set. Fluorescence images of live cells expressing GFP-dynamins and GFP-clathrin were acquired by a confocal microscope (LSM510, Zeiss, Germany) or a custom-made total internal reflection fluorescence (TIRF) microscope [31].
Traction force exerted by dividing cells was measured as previously described [22]. Briefly, cells were placed on an elastic silicone substratum with fluorescent red beads and observed using a DeltaVision microscope system (GE Healthcare, Little Chalfont, UK). To acquire the initial position-image of the beads, 10% sodium azide was added to kill the cells after the observation.

Fluorescence Recovery after Photobleaching (FRAP) Analysis
For FRAP experiments, wild-type and mutant cells expressing GFP-lifeact were agar-overlaid. Live imaging and photobleaching experiments were performed using the confocal microscope. The half-time of fluorescence recovery was calculated as described previously [32].

Immunoblotting and Co-Immunoprecipitation
For immunoblotting of the whole cell lysates, cells were directly lysed in 2× SDS sample buffer (0.125 M Tris-HCl, 4% sodium dodecyl sulfate, 20% glycerol, 0.2 M DTT, 0.02% bromophenol blue, pH 6.8). Alternatively, the cells were extracted with a buffer containing Triton X-100 and subjected to western blot, as described previously [17]. Briefly, the cells were extracted with Triton buffer (5 mM PIPES, 50 mM NaCl, 5 mM EGTA, 5 mM EDTA, 2 mM NaF, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM DTT, 10 mM benzamidine, and 1% Triton X-100, pH 7.5) on ice for 15 min; the insoluble and soluble fractions were separated by centrifugation, and were subjected to western blot. The quantitative analysis was performed using the ImageJ software.

Statistical Analysis
Statistical analysis was performed using GraphPad Prism 7 (GraphPad Software Inc., San Diego, CA, USA). Data were analyzed using Student's t-test for comparison between two groups, or one-way ANOVA with Tukey's multiple comparison test, and are presented as the mean ± standard deviation (SD).

DlpA, DlpB, and DymA Contribute to Cytokinesis in Different Manners
Dictyostelium discoideum has five genes coding for dynamin-like proteins: dymA, dymB, dlpA, dlpB, and dlpC. For simplicity, here we will refer to dynamin-like proteins as dynamins. Previously, we and other groups have suggested that among these five dynamins, dlpA, dlpB, and dymA may contribute to cytokinesis [16,17]. We observed the nuclei in each knockout mutant (dlpA -, dlpB -, and dymA -) after the culture in suspension condition ( Figure 1A). The three mutant cells became much larger than wild-type (A×2) cells and contained multiple nuclei, suggesting that these mutant cells have a defect in cytokinesis. A double knockout mutant in both dlpA and dlpB (dlpA -/dlpB -) also showed multinucleation. We tried to generate other double and triple mutants; however, the efforts were unsuccessful, suggesting that they could be lethal.

DlpA, DlpB, and DymA Contribute to Cytokinesis in Different Manners
Dictyostelium discoideum has five genes coding for dynamin-like proteins: dymA, dymB, dlpA, dlpB, and dlpC. For simplicity, here we will refer to dynamin-like proteins as dynamins. Previously, we and other groups have suggested that among these five dynamins, dlpA, dlpB, and dymA may contribute to cytokinesis [16,17]. We observed the nuclei in each knockout mutant (dlpA -, dlpB -, and dymA -) after the culture in suspension condition ( Figure 1A). The three mutant cells became much larger than wild-type (A×2) cells and contained multiple nuclei, suggesting that these mutant cells have a defect in cytokinesis. A double knockout mutant in both dlpA and dlpB (dlpA -/dlpB -) also showed multinucleation. We tried to generate other double and triple mutants; however, the efforts were unsuccessful, suggesting that they could be lethal. Figure 1B shows a summary of multinucleation of each mutant when cultured in suspension and on a surface. In both conditions, these mutants showed more multinucleation as compared to the wild-type cells; the multinucleation in the suspension culture was much more severe than that observed in adherent culture conditions. Interestingly, the double mutant dlpA -/dlpBcells showed a similar level of multinucleation as that of single mutants, suggesting that dlpA and dlpB might cooperatively contribute to the cytokinesis.   Figure 1B shows a summary of multinucleation of each mutant when cultured in suspension and on a surface. In both conditions, these mutants showed more multinucleation as compared to the wild-type cells; the multinucleation in the suspension culture was much more severe than that observed in adherent culture conditions. Interestingly, the double mutant dlpA -/dlpBcells showed a similar level of multinucleation as that of single mutants, suggesting that dlpA and dlpB might cooperatively contribute to the cytokinesis.
We observed each cell line during cell division on coverslips under phase contrast microscopy ( Figure 1C). Typically, when Dictyostelium cells enter the mitotic phase, they stop migration, become round, elongate, and constrict the cleavage furrow to separate into two daughter cells. All mutant cells showed a similar morphological process as the wild-type cells; however, dlpAand dlpBcells appeared darker under the phase contrast microscopy. Remarkably, all mutants took a longer time to complete the final separation.
To determine the stage of cytokinesis at which the cells were delayed, the time required from the round stage to the initiation of furrowing, that from the initiation of furrowing to the final separation, and the total time for the cell division were examined in each mutant ( Figure 1D). The time required from round stage to the initiation of furrowing of the mutants was not significantly different from that of wild-type cells. However, dlpA -, dlpB -, and dlpA -/dlpBcells took a significantly longer time for constricting the furrow. On the other hand, dymAcells exhibited a much longer intercellular bridge and took a longer time to separate (arrows in Figure 1C) (sometimes longer than 20 min).

Three Dynamins Localize at the Cleavage Furrow
Next, we examined the localization of the three dynamins in dividing cells. We have previously shown that dlpA and dymA localize at the furrow region [17]. Here, we observed the localization of dlpB and compared it with that of other dynamins in live cells. Each GFP-tagged dynamin was expressed in individual knockout mutant cells. Figure 1B shows the frequencies of multinucleation in these cells, indicating that the expression of the GFP-tagged protein rescued the defects in cytokinesis in both suspension and adherent culture conditions. Figure 2A shows typical time course of fluorescence images of GFP-dlpA, GFP-dlpB, and GFP-dymA in the respective knock-out cell line during the cell division. Both GFP-dlpA and GFP-dlpB localized at the cleavage furrow from the initial stage of furrowing to the final separation. However, a small amount of GFP-dymA localized at the intercellular bridge only during the final separation.
We generated antibodies against dlpB and confirmed their specificity by western blot ( Figure 2B). There was no detectable band corresponding to dlpB in dlpBcells. The anti-dlpA antibodies, which have been previously described [17], found a similar expression level of dlpA in dlpBcells to that of wild-type (A×2) cells. Similarly, dlpB was found to be expressed in a similar level in dlpAcells to that of wild type cells, suggesting that dlpA and dlpB were stable in each mutant cell.
Next, we immuno-stained wild-type and dlpBcells with the anti-dlpB antibody ( Figure 2C). The fluorescence images showed that dlpB localized at the cleavage furrow of wild-type cells, although there was no detectable staining in dlpBcells, suggesting that the endogenous dlpB also localize at the furrow in wild-type cells. We have already confirmed that the endogenous dlpA and dymA show the same localization as the GFP-tagged proteins in the live cells [17].
Collectively, dlpA and dlpB might contribute to the furrowing and dymA might contribute to the final separation.

DlpA and DlpB Colocalize at the Cleavage Furrow
GFP-dlpA and mCherry-dlpB were simultaneously observed in dlpA -/dlpBcells. Figure 3A shows a typical time course of fluorescence images. The merged images (Merge) indicate that both proteins simultaneously accumulate at the same place.
Next, the cells expressing GFP-dlpA were observed using total internal reflection fluorescence (TIRF) microscopy, which enables the selective visualization of about 100 nm (depth) above the coverslip, covering the thickness of the cell cortex. Figure 3B shows a typical TIRF image of GFP-dlpA in a dividing cell. Many individual dots consisting of dlpA were observed at the furrow cortex.
The scission of newly formed vesicles from the membrane, such as in endocytosis, by the dynamin is well established. To examine the possibility that dlpA is also involved in endocytosis, we compared the localization of dlpA with that of clathrin, which plays a major role in the formation of coated vesicles. Although GFP-clathrin also appeared as dots in the cell membrane when visualized by the TIRF microscope, it did not accumulate at the furrow ( Figure 3B). Thus, it is unlikely that dlpA is involved in the clathrin-mediated endocytosis.
Next, GFP-dlpA and mCherry-dlpB were simultaneously observed using TIRF microscopy ( Figure 3C). The fluorescence intensity profile across the single dot (white line in TIRF images) indicates that both proteins were found to colocalize at the same dots. As dynamin is generally known to form oligomers [33], dlpA and dlpB may form a hetero-oligomer as a functional unit.
To investigate this possibility, co-immunoprecipitation assay was performed. The anti-dlpB

DlpA and DlpB Colocalize at the Cleavage Furrow
GFP-dlpA and mCherry-dlpB were simultaneously observed in dlpA -/dlpBcells. Figure 3A shows a typical time course of fluorescence images. The merged images (Merge) indicate that both proteins simultaneously accumulate at the same place.
Next, the cells expressing GFP-dlpA were observed using total internal reflection fluorescence (TIRF) microscopy, which enables the selective visualization of about 100 nm (depth) above the coverslip, covering the thickness of the cell cortex. Figure 3B shows a typical TIRF image of GFP-dlpA in a dividing cell. Many individual dots consisting of dlpA were observed at the furrow cortex.
The scission of newly formed vesicles from the membrane, such as in endocytosis, by the dynamin is well established. To examine the possibility that dlpA is also involved in endocytosis, we compared the localization of dlpA with that of clathrin, which plays a major role in the formation of coated vesicles. Although GFP-clathrin also appeared as dots in the cell membrane when visualized by the TIRF microscope, it did not accumulate at the furrow ( Figure 3B). Thus, it is unlikely that dlpA is involved in the clathrin-mediated endocytosis.
Next, GFP-dlpA and mCherry-dlpB were simultaneously observed using TIRF microscopy ( Figure 3C). The fluorescence intensity profile across the single dot (white line in TIRF images) indicates that both proteins were found to colocalize at the same dots. As dynamin is generally known to form oligomers [33], dlpA and dlpB may form a hetero-oligomer as a functional unit.
To investigate this possibility, co-immunoprecipitation assay was performed. The anti-dlpB antibody-bound magnetic beads were mixed with the cell lysate of the partially-synchronized mitotic cells and the co-sedimented proteins were subjected to western blot analysis using anti-dlpA and anti-dlpB antibodies ( Figure 3D). A substantial amount of dlpA was detected in the sedimented fraction. However, only a slight amount of dlpA was detected when using non-synchronized (interphase) cell lysate (if the total cell division time is 5 min and the total doubling time is 8 hr, mitotic cells account for 1.0% in the non-synchronized cells). These results indicate that dlpA and dlpB directly or indirectly bind to each other, depending on the cell cycle. antibody-bound magnetic beads were mixed with the cell lysate of the partially-synchronized mitotic cells and the co-sedimented proteins were subjected to western blot analysis using anti-dlpA and anti-dlpB antibodies ( Figure 3D). A substantial amount of dlpA was detected in the sedimented fraction. However, only a slight amount of dlpA was detected when using non-synchronized (interphase) cell lysate (if the total cell division time is 5 min and the total doubling time is 8 hr, mitotic cells account for 1.0% in the non-synchronized cells). These results indicate that dlpA and dlpB directly or indirectly bind to each other, depending on the cell cycle.

Both DlpA and DlpB are Required for their Localization to the Furrow
To examine which part of the dlpB molecule is required for its localization to the furrow, domain analysis was performed. DlpB contains a GTPase domain, which is a common characteristic domain of the dynamin family. A GFP-GTPase domain (aa 1-340), a GFP-fragment without the GTPase domain (aa 341-808), and a GFP-fragment without the C terminal domain (aa 1-600) were expressed in dlpBcells. None of these fragments showed any localization at the furrow ( Figure 4A). Therefore, full length of dlpB is indispensable for its localization at the furrow.

Both DlpA and DlpB are Required for their Localization to the Furrow
To examine which part of the dlpB molecule is required for its localization to the furrow, domain analysis was performed. DlpB contains a GTPase domain, which is a common characteristic domain of the dynamin family. A GFP-GTPase domain (aa 1-340), a GFP-fragment without the GTPase domain (aa 341-808), and a GFP-fragment without the C terminal domain (aa 1-600) were expressed in dlpBcells. None of these fragments showed any localization at the furrow ( Figure 4A). Therefore, full length of dlpB is indispensable for its localization at the furrow.
Next, GFP-full length dlpB was expressed in dlpAcells. Interestingly, dlpB did not localize at the furrow. Conversely, GFP-dlpA did not localize at the furrow in dlpBcells ( Figure 4B). Therefore, both, dlpA and dlpB are required for their localization at the furrow.
Given these observations, we concluded that dlpA and dlpB accumulate to the furrow as a complex, presumably in a form of hetero-oligomer.

DlpA and DlpB Associate with Actin Filaments at the Cleavage Furrow
Previously, we have suggested that dlpA may associate with actin filaments as the lack of dlpA caused severe fragmentation of actin filaments in the contractile ring [17]. Hence, we examined whether dlpB associates with actin filaments in the contractile ring as well. When latrunculin A, a depolymerizer of actin filaments, was added to dividing cells expressing GFP-lifeact, the actin filaments disappeared, which caused loosing of the furrow and failure of cytokinesis ( Figure 5A, left). GFP-dlpB also delocalized from the cell membrane after latrunculin A treatment, suggesting that dlpB associates with actin filaments (Figure 5A, right). Our previous observations showed that GFP-dlpA was also delocalized from the cleavage furrow upon latrunculin A treatment [17].
To confirm the association between dlpB and actin filaments, the synchronized cells were extracted with a buffer containing Triton X-100. The insoluble actin cytoskeletons were subjected to western blot analysis using an anti-dlpB antibody. After the extraction, a substantial amount of dlpB was still present in the cytoskeleton (63.0 ± 12%, Figure 5B and C). DlpA also remained in the actin cytoskeleton, as described previously [17]. Next, GFP-full length dlpB was expressed in dlpAcells. Interestingly, dlpB did not localize at the furrow. Conversely, GFP-dlpA did not localize at the furrow in dlpBcells ( Figure 4B). Therefore, both, dlpA and dlpB are required for their localization at the furrow.
Given these observations, we concluded that dlpA and dlpB accumulate to the furrow as a complex, presumably in a form of hetero-oligomer.

DlpA and DlpB Associate with Actin Filaments at the Cleavage Furrow
Previously, we have suggested that dlpA may associate with actin filaments as the lack of dlpA caused severe fragmentation of actin filaments in the contractile ring [17]. Hence, we examined whether dlpB associates with actin filaments in the contractile ring as well. When latrunculin A, a depolymerizer of actin filaments, was added to dividing cells expressing GFP-lifeact, the actin filaments disappeared, which caused loosing of the furrow and failure of cytokinesis ( Figure 5A, left). GFP-dlpB also delocalized from the cell membrane after latrunculin A treatment, suggesting that dlpB associates with actin filaments (Figure 5A, right). Our previous observations showed that GFP-dlpA was also delocalized from the cleavage furrow upon latrunculin A treatment [17].
Interestingly, TRITC-DNase I mainly stained the furrow regions, whereas FITC-phalloidin stained the cell cortex as well as the furrow ( Figure 5D). The fluorescence intensity of TRITC-DNase I relative to that of FITC-phalloidin at the furrow was significantly higher (p ≤ 0.001) in dlpA and dlpB mutant cells than in wild-type cells ( Figure 5E), which suggests that actin filaments are fragmented at the furrow region of the mutant cells. Therefore, these dynamins may contribute to the stabilization of actin filaments in the contractile rings.  To confirm the association between dlpB and actin filaments, the synchronized cells were extracted with a buffer containing Triton X-100. The insoluble actin cytoskeletons were subjected to western blot analysis using an anti-dlpB antibody. After the extraction, a substantial amount of dlpB was still present in the cytoskeleton (63.0 ± 12%, Figure 5B,C). DlpA also remained in the actin cytoskeleton, as described previously [17].
Collectively, both, dlpA and dlpB directly or indirectly associate with actin filaments at the cleavage furrow.

DlpA and dlpB Stabilize the Actin Filaments at the Cleavage Furrow
To examine the fragmentation of the actin filaments in the contractile rings of mutant cells, we carried out TRITC-DNase I staining. DNase I binds to the pointed ends of actin filaments as well as the subdomains II and IV of monomeric actin [34,35]. If actin filaments are fragmented at the furrow, there would be more free ends of actin filaments available for binding to TRITC-DNase I. After lysing the cells with Triton X-100, the insoluble cytoskeleton was simultaneously stained with TRITC-DNase I and FITC-phalloidin. Monomeric actin is substantially extracted under these conditions. Interestingly, TRITC-DNase I mainly stained the furrow regions, whereas FITC-phalloidin stained the cell cortex as well as the furrow ( Figure 5D). The fluorescence intensity of TRITC-DNase I relative to that of FITC-phalloidin at the furrow was significantly higher (p ≤ 0.001) in dlpA and dlpB mutant cells than in wild-type cells ( Figure 5E), which suggests that actin filaments are fragmented at the furrow region of the mutant cells. Therefore, these dynamins may contribute to the stabilization of actin filaments in the contractile rings.
To examine this possibility, fluorescence recovery after photobleaching (FRAP) was carried out at the furrow of each mutant cell expressing GFP-lifeact ( Figure 5F). Previously, we have used GFP-lifeact for the estimation of the turnover of actin filaments and proved that the turnover of GFP-lifeact reflects the turnover of actin filaments [36]. However, a recent report has described that GFP-lifeact can modify the turnover of actin filaments depending on the expression levels [37]. We compared the expression level of each mutant from western blot using anti-GFP antibodies and found that there were no significant differences (A×2:dlpA -:dlpB -= 1.00:1.09 ± 0.10:0.98 ± 0.11, p > 0.05, three independent experiments). Therefore, we considered ourselves to be able to compare the half-time of recovery among mutants. The half-time of fluorescence recovery was significantly shorter in the mutant cells than in the wild-type cells ( Figure 5G-K).
Taken together, these results suggest that dlpA and dlpB stabilize the actin filaments by suppressing the extent of filament fragmentation in the contractile rings. The actin turnover rates were almost the same in the double mutant as in the single mutants, again suggesting that dlpA and dlpB function cooperatively.
Myosin II also accumulates to the cleavage furrow and contributes to the constriction of the furrow. Myosin II can cut actin filaments and enhance their turnover in the contractile rings [31]. Thus, we examined whether dlpA and dlpB localize at the cleavage furrow in myosin II null cells (HS1). Figure 5L shows typical immunofluorescence images of HS1 cells stained with anti-dlpA and anti-dlpB antibodies, indicating that dlpA and dlpB localize at the cleavage furrow independently of myosin II. In addition, this observation suggests that these dynamins contribute to cytokinesis B (traction-mediated cytokinesis) as well as cytokinesis A (contractile ring-dependent cytokinesis).

Dynamins Contribute to Cell-Substratum Adhesion and Traction Force
Under the phase-contrast microscope, dividing dlpAand dlpBcells appeared significantly darker than the wild-type cells ( Figure 1C), suggesting that they are flatter and more adherent to the substratum. Dictyostelium cells have a traction force-dependent cytokinesis mode (cytokinesis B) [38]. HS1 cells divide by traction forces exerted by the two daughter cells on the surface [18]. Dividing HS1 cells appear darker under phase-contrast microscopy and exert a larger traction force than the wild-type cells [22]. We speculated that the defects in the organization of contractile rings in dynamin mutant cells might affect the cell-substratum adhesion, traction force, and cytokinesis mode. Therefore, we simultaneously observed the cell-substratum adhesion and traction force using reflection interference (IR) and traction force (TF) microscopy. Figure 6A shows representative differential interferential contrast (DIC) images, IR images, traction maps, and vector maps. the polar regions in all dividing cells, which suggests that both the daughter halves migrate in the opposite directions, thereby exerting traction stress against the substratum towards the cell body. The mean traction stress of dymAcells was significantly lower than that of the wild-type cells.
Together, these results suggest that these dynamins contribute to the regulation of the cellsubstratum adhesion and traction force.

DlpA Localizes at the Phagocytic Cup Independently of DlpB
DlpA and dlpB may not always function cooperatively with each other. Interestingly, dlpA localized at phagocytic cups when cells internalized bacteria. Figure 7A shows typical sequential events during phagocytosis of a cell expressing GFP-dlpA. When the cell extended pseudopods around a bacterium and encircled and internalized it, GFP-dlpA localized around the bacterium. In contrast, neither dlpB nor dymA showed any localization at the phagocytic cups ( Figure 7B, arrows). The dlpA was also observed at the phagocytic cups in the dividing cells ( Figure 7C), suggesting that dlpA accumulates there independently of the cell cycle. Interestingly, GFP-dlpA localized at the phagocytic cups in dlpBcells ( Figure 7D). Therefore, dlpA accumulation at the phagocytic cups is independent of dlpB. The dark area in the IR micrograph was quantified as the cell-substratum adhesion area in each mutant cell ( Figure 6B). The averaged adhesion areas of dlpAand dlpBcells were significantly larger than that of the wild-type cells, whereas the averaged adhesion area of dymAcells was significantly smaller than that of the wild-type cells.
Mean traction stresses of dlpAand dlpBcells were much higher than those of wild-type cells ( Figure 6C). The vector map indicates that most of the traction stress was directed inward from both the polar regions in all dividing cells, which suggests that both the daughter halves migrate in the opposite directions, thereby exerting traction stress against the substratum towards the cell body. The mean traction stress of dymAcells was significantly lower than that of the wild-type cells.
Together, these results suggest that these dynamins contribute to the regulation of the cell-substratum adhesion and traction force.

DlpA Localizes at the Phagocytic Cup Independently of DlpB
DlpA and dlpB may not always function cooperatively with each other. Interestingly, dlpA localized at phagocytic cups when cells internalized bacteria. Figure 7A shows typical sequential events during phagocytosis of a cell expressing GFP-dlpA. When the cell extended pseudopods around a bacterium and encircled and internalized it, GFP-dlpA localized around the bacterium. In contrast, neither dlpB nor dymA showed any localization at the phagocytic cups ( Figure 7B, arrows). The dlpA was also observed at the phagocytic cups in the dividing cells ( Figure 7C), suggesting that dlpA accumulates there independently of the cell cycle. Interestingly, GFP-dlpA localized at the phagocytic cups in dlpBcells ( Figure 7D). Therefore, dlpA accumulation at the phagocytic cups is independent of dlpB.

Discussion
The present study provides the first report about the role of dlpB in Dictyostelium cells. We found that dlpB, along with dlpA and dymA, contributes to cytokinesis. The mutants deficient in these dynamins exhibited defects of cytokinesis in both suspension and adherent culture conditions. The cytokinesis defects were much more severe in suspension culture conditions, which is reminiscent of defects of myosin II null cells [39]. Myosin II also regulates the dynamics of the contractile ring, and the deletion of myosin II results in the failure of cytokinesis [40]. Dysfunction of the dynamic organization of the actin structure in the contractile ring of the dynamin mutant cells may also result in cytokinesis defects. The cytokinesis defects were found to be of similar extent in dlpA -, dlpB -, and dlpA -/dlpBdouble knockout cells. In addition, dlpA and dlpB were found to colocalize at the cleavage furrow from the early phase of furrowing till the final separation. TIRF microscopy revealed dlpB as small dots, where dlpA colocalized. Furthermore, dlpA and dlpB were co-precipitated in the co-immunoprecipitation experiments; dlpA did not localize at the cleavage furrow in dlpBcells and vice versa. DlpA and dlpB colocalized with actin filaments in the contractile ring, and latrunculin A treatment lost their localization. All these results strongly suggest that the hetero-oligomerization is required for these molecules to accumulate at the cleavage furrow. The hetero-oligomerization of dynamins has been recently identified in mitochondrion fission in Entamoeba histolytica [41].
The role of dymA in cytokinesis seems to be different from that of the dlpA and dlpB complex. Dividing dymAcells frequently formed a long intercellular bridge, and took much longer time for the final separation. Taken together with its localization at the intercellular bridge, dymA might contribute to the final separation.

Discussion
The present study provides the first report about the role of dlpB in Dictyostelium cells. We found that dlpB, along with dlpA and dymA, contributes to cytokinesis. The mutants deficient in these dynamins exhibited defects of cytokinesis in both suspension and adherent culture conditions. The cytokinesis defects were much more severe in suspension culture conditions, which is reminiscent of defects of myosin II null cells [39]. Myosin II also regulates the dynamics of the contractile ring, and the deletion of myosin II results in the failure of cytokinesis [40]. Dysfunction of the dynamic organization of the actin structure in the contractile ring of the dynamin mutant cells may also result in cytokinesis defects. The cytokinesis defects were found to be of similar extent in dlpA -, dlpB -, and dlpA -/dlpBdouble knockout cells. In addition, dlpA and dlpB were found to colocalize at the cleavage furrow from the early phase of furrowing till the final separation. TIRF microscopy revealed dlpB as small dots, where dlpA colocalized. Furthermore, dlpA and dlpB were co-precipitated in the co-immunoprecipitation experiments; dlpA did not localize at the cleavage furrow in dlpBcells and vice versa. DlpA and dlpB colocalized with actin filaments in the contractile ring, and latrunculin A treatment lost their localization. All these results strongly suggest that the hetero-oligomerization is required for these molecules to accumulate at the cleavage furrow. The hetero-oligomerization of dynamins has been recently identified in mitochondrion fission in Entamoeba histolytica [41].
The role of dymA in cytokinesis seems to be different from that of the dlpA and dlpB complex. Dividing dymAcells frequently formed a long intercellular bridge, and took much longer time for the final separation. Taken together with its localization at the intercellular bridge, dymA might contribute to the final separation.
Endocytosis has been implicated in cytokinesis. Continuous endocytosis is crucial for cytokinesis of zebrafish blastomeres; endocytosis inhibitors are known to block the separation of the daughter cells [5]. Clathrin null Dictyostelium cells have defects in cytokinesis [42]. Dictyostelium lvsA, which is involved in membrane trafficking, is required for cytokinesis [43]. Mutations in clathrin or dynamin also lead to cytokinesis failure in other organisms, such as C. elegans, Drosophila, and mammalian and plant cells [6,7,10,11,44,45]. These observations suggest that dynamin and clathrin may cooperatively participate in the endocytosis or membrane trafficking pathways required for cytokinesis. However, as dlpA and dlpB did not colocalize with clathrin, it is unlikely that they are directly involved in the clathrin-mediated endocytosis; they might be involved in other types of endocytosis. In zebrafish blastomeres, caveolae-mediated endocytosis, as well as clathrin-mediated endocytosis, contribute to cytokinesis [5]. However, we could not find any gene homologous to caveolin in the Dictyostelium genome.
Another possible role of the dlpA and dlpB complex in cytokinesis is the regulation of actin filaments in the contractile ring. The present results suggest that the hetero-oligomers of dlpA and dlpB directly or indirectly bind to the actin filaments and stabilize them. Dynamin I has been reported to directly bind to the actin filaments and remove the actin-capping protein, gelsolin, from barbed ends in vitro, thus allowing the elongation of the actin filaments [46]. Conversely, direct binding of short actin filaments to dynamin 1 enhances the oligomerization of dynamin 1 in vitro as well as in vivo [47]. A cross talk between actin and dynamin has also been reported in phagosome formation and closure [48]. The actin-binding site of dynamin 1 is located at amino acids 399-444 of the middle domain [46]. DymA and dymB have approximately 45% and 28% homology to this sequence in their middle domains, respectively; however, dlpA and dlpB have no middle domain equivalent to that of dynamin 1. The direct interaction between purified actin and Dictyostelium dynamins in vitro remains to be examined in the future.
Dynamins can also indirectly interact with actin filaments by binding to several actin-binding proteins, which is mediated by the proline, arginine-rich domain (PRD) at dynamin's C-terminus [49]; nonetheless, the PRD domain is absent in Dictyostelium dynamins. Mammalian dynamin 1 binds to profilin I and II [50], which are essential for actin assembly. The complex of dynamin 1 and cortactin has been reported to stabilize actin filaments in the growth cones of the neuronal cells [51]. Dynamin 2 influences actin nucleation by the Arp2/3 complex and cortactin in vitro in a biphasic manner; low concentration of dynamin 2 enhances actin nucleation by Arp2/3 complex and cortactin, whereas a high concentration is inhibitory [52]. Dynamin 2 also modulates localization of Rac, a small GTPase that regulates actin cytoskeletons [53].
The fact that dlpAand dlpBcells showed a faster turnover of actin filaments than the wild-type cells indicates that these dynamins are likely to be responsible for antagonizing the filament severing activities, thereby helping to stabilize actin filaments in the contractile rings. Alternatively, dynamin may promote actin polymerization as described above [46] and maintain a constant length of actin filaments in the contractile rings. Myosin II is a candidate protein responsible for severing actin filaments in the contractile ring [31]. Cofilin and formin, in addition to myosin II, are known as the proteins responsible for regulating the length of actin filaments in the contractile rings in other organisms [54][55][56]. The cooperation between Dictyostelium dynamins and these proteins needs to be examined.
Another role of dlpA and dlpB is their involvement in the cell-substratum adhesion and traction force. It was previously reported that dymBcells are flatter on the substratum than the wild-type cells [14]. In the present study, we found that dlpAand dlpBcells have a wider adhesion area and higher traction force than the wild-type cells. However, dymAcells showed the opposite features. These dynamins may be involved in the formation of the focal adhesions, which anchor the actin filaments [57]. Although these Dictyostelium dynamins do not localize at the focal adhesions, dynamins have been reported to localize at the focal adhesions in mammalian cells [58,59]. Aberrant cell-substratum adhesion may result in cytokinesis defects of Dictyostelium dynamin null cells on a surface. How dynamins contribute to the cell adhesion and traction force remains to be clarified in the future.
Although dlpA and dlpB were mutually dependent on each other for their accumulation at the cleavage furrow ( Figure 4B), dlpA accumulated at the phagocytic cup independently of dlpB. DlpA may function independently of dlpB in phagocytosis. DymA has been reported to localize at the phagosomes during their maturation but, unlike dlpA, dymA did not localize at the phagocytic cup at an early stage of phagocytosis [60]. The regulatory mechanisms underlying the localization of dlpA and dymA at the cleavage furrow or the phagocytic apparatus remain elusive. Such questions also need further attention in the future. Figure 8 summarizes the localization of the three dynamins. DlpA and dlpB form hetero-oligomers and associate with and stabilize the actin filaments in the contractile ring. DymA localizes at the intercellular bridge in the last stage of cytokinesis. Therefore, we conclude that dlpB contributes to cytokinesis cooperatively with dlpA and dymA. surface. How dynamins contribute to the cell adhesion and traction force remains to be clarified in the future. Although dlpA and dlpB were mutually dependent on each other for their accumulation at the cleavage furrow ( Figure 4B), dlpA accumulated at the phagocytic cup independently of dlpB. DlpA may function independently of dlpB in phagocytosis. DymA has been reported to localize at the phagosomes during their maturation but, unlike dlpA, dymA did not localize at the phagocytic cup at an early stage of phagocytosis [60]. The regulatory mechanisms underlying the localization of dlpA and dymA at the cleavage furrow or the phagocytic apparatus remain elusive. Such questions also need further attention in the future. Figure 8 summarizes the localization of the three dynamins. DlpA and dlpB form heterooligomers and associate with and stabilize the actin filaments in the contractile ring. DymA localizes at the intercellular bridge in the last stage of cytokinesis. Therefore, we conclude that dlpB contributes to cytokinesis cooperatively with dlpA and dymA.