The fission yeast SPB component Dms1 is required to initiate forespore membrane formation and maintain meiotic SPB components

The spindle pole body (SPB) plays a central role in spore plasma membrane formation in addition to its recognized role in microtubule organization. During meiosis, a biomembrane called the forespore membrane (FSM) is newly formed at the SPB. Although several SPB proteins essential for the initiation of FSM formation (meiotic SPB components) have been identified, the molecular mechanism is still unknown. Here, we report the isolation and functional characterization of Dms1 as a component of the SPB. We show that FSM formation does not initiate in dms1Δ cells. Dms1 protein is constitutively expressed throughout the life cycle and localizes to the SPB and the nuclear envelope. The predicted Dms1 protein has a transmembrane domain, which is required for correct localization at the SPB. Dms1 is essential for the proper localization of three meiotic SPB components, Spo15, Spo2, and Spo13, but these components do not affect localization of Dms1. Collectively, these results suggest that Dms1 anchors these meiotic SPB components to the SPB, thereby facilitating the initiation of FSM formation.


Introduction
The fission yeast Schizosaccharomyces pombe cells enter sporulation under nutrient starvation, particularly when nitrogen is the limiting nutrient [1][2][3]. Sporulation consists of two overlapping processes, meiotic nuclear division and spore morphogenesis. The main event of the latter process is the formation of a double-layered intracellular membrane, called the forespore membrane (FSM). Formation of the FSM initiates during meiosis II. As the nucleus divides, the FSM expands by vesicle fusion and eventually encapsulates each of the four nuclei generated. Subsequently, the inner layer of the FSM becomes the spore plasma membrane, while the outer layer is degraded by autolysis [4,5].
An interesting aspect of FSM formation is its initiation because, unlike other biological membranes, the FSM is newly synthesized within the cytoplasm of the parent cell [4][5][6][7]. FSM formation takes place at the spindle pole body (SPB), a proteinaceous structure composed of PLOS

Yeast strains and media
The S. pombe and plasmids used in this study are listed in Tables 1 and 2, respectively. For vegetative cultures, complete (YE) or synthetic (SD or MM+N) media were used. For sporulation, cells precultured in YE, or SD/MM+N were incubated in malt extract medium (ME) or synthetic sporulation medium (SSA/MM-N), respectively [22,23]. Unless stated otherwise, S. pombe strains were incubated at 30˚C for growth or 28˚C for mating and sporulation. Synchronous meiosis was induced by a temperature shift using strains carrying the pat1-114 allele as previously described [24].

Fluorescence microscopy
For fluorescence microscopy, proteins were visualized by fusion to GFP, 3 X mRFP or mCherry. The nuclear chromatin region was stained with bisbenzimide H33342 fluorochrome trihydrochroride (Hoechst 33342). Living cells were observed under a fluorescence microscope (model  [31]. Images were processed with Image J.

Western blotting
Proteins were resolved by SDS-PAGE and then transferred to a polyvinylidene difluoride membrane (Immobilon-P; Merck Millipore, Darmstadt, Germany). Blots were probed with a rabbit (Medical and Biological Laboratories, Nagoya, Japan) or rat anti-GFP antibody (a gift from S. Fujita, Mitsubishi Kagaku Institute of Life Sciences), and the anti-α-tubulin antibody TAT-1 [32]. Immunoreactive bands were revealed by ECL select chemiluminescence (GE Healthcare, Buckinghamshire, UK) with horseradish peroxidase-conjugated goat anti-rabbit IgG (Bio-Rad Laboratories, Hercules, USA) or goat anti-mouse IgG (BIOSOURCE, Camarillo, USA).

The novel gene dms1 + is involved in FSM formation
Genome-wide reverse genetic studies have shown that many S. pombe genes are involved in sporulation [33,34]. Among them, we selected 13 deletion mutants that have been reported to show defects in sporulation but have not been uncharacterized (mug123, mug165, mug166, mug168, csn1, csn2, erp2, erp5, spe2, spe3, tpp1, SPAC6C3.06c, and SPCC1739.04c). We introduced GFP-tagged Psy1, an FSM marker, into each mutant and observed formation of the FSM by fluorescence microscopy (S1 Fig reported a functional genomics screen of genes important in S. pombe meiosis in which they identified SPCC1739.04c as dms1 [35]. Therefore, SPCC1739.04c is designated as dms1 + hereafter. The dms1 + gene was predicted to encode a 31-kDa protein composed of 274 amino acids. The deduced amino acid sequence was found to be conserved only in the Schizosaccharomyces genus. Hydropathic profiling and prediction of secondary structure indicated that Dms1 contains one potential membrane-spanning domain in its carboxy terminus ( Fig 1A), but no other known functional motifs.

Dms1 is essential for initiating FSM formation
To examine how the dms1 mutation impairs sporulation, assembly of the FSM in the dms1 mutant was observed in detail by using GFP-Psy1. In wild-type cells at metaphase II, the GFP-Psy1 signal was observed as two cup-like structures at each end of the spindle microtubules, which were Table 2. Plasmids used in this study.

S. pombe
pBR(leu1) leu1 + in pBR322 [28] pREP41 ars1, LEU2-based expression vector carrying nmt41 promoter [29] pREP42 ars1, ura4 + -based expression vector carrying nmt41 promoter [30] pAL-KS ars1, LEU2-based multicopy shuttle vector [30] pREP42(GST) pREP42, Glutathione-S-transferase (GST) [18] This study This study  visualized by an mCherry-labeled α-tubulin, Atb2 (Fig 1B). In contrast, no GFP-Psy1 signal was observed at the spindle poles in about 90% of dms1Δ cells at the same stage (Fig 1B), a result that was confirmed by live-cell imaging (Fig 1C; S1 and S2 Movies). This phenotype is similar to that of mutants of meiotic SPB components such as Spo13 and Spo15 [16,18]. Taken together, these data indicate that Dms1 is essential for the initiation of FSM formation. In complete medium, the dms1 deletion mutant did not differ from the wild-type strain in growth rate, cell size, or shape (data not shown), indicating that dms1 + is not essential for vegetative growth. Next, we examined the progression of meiotic nuclear division in dms1Δ cells.
To synchronize meiosis effectively, we used the pat1-114 temperature-sensitive strain, which enters meiosis in a highly synchronous manner when it is shifted to its restrictive temperature, 34˚C [24]. The first and second meiotic divisions of the pat1dms1Δ cells were found to proceed with kinetics similar to that observed for pat1 cells, with a final yield of tetranucleate cells of 90% ( Fig 1D). Therefore, the dms1Δ mutant seems to undergo meiosis normally, but is defective in ascospore formation.

Dms1 localizes at the SPB and the nuclear envelope
A comprehensive mRNA expression study previously revealed that the dms1 + gene is expressed throughout the life cycle, but is upregulated slightly during meiosis and/or sporulation [36]; however, the levels of Dms1 protein during meiosis/sporulation have not been reported. To ascertain whether the levels of Dms1 protein mirror those of dms1 mRNA, we constructed a strain in which GFP-tagged dms1 + was expressed from its endogenous locus. As expected, GFP-Dms1 was detected during vegetative growth and its abundance essentially remained constant after the induction of sporulation (Fig 1E).
The subcellular localization of Dms1 was observed by fluorescence microscopy. Consistent with the western blot analysis, GFP-Dms1 signal was observed in both vegetative and in meiotic cells as one or two dots in the periphery of nuclei (Fig 2A). These dots were detected at both ends of the spindle microtubules at meiosis II (Fig 2B), indicating that Dms1 localizes to the SPB. In addition, a weak signal surrounding the nucleus was also observed in vegetative cells. This signal became stronger after meiosis II completed and overlapped with signals from RFP-tagged Cut11 (Fig 2B), a marker of the nuclear envelope [37]. We therefore concluded that Dms1 is also localized to the nuclear envelope.
As described above, the predicted Dms1 protein has a transmembrane region. To assess the role of this transmembrane domain, we constructed a dms1ΔTM mutant lacking this domain ( Fig 3A). Spore formation was severely compromised in the dms1ΔTM mutant ( Fig 3B). Consistent with this, GFP-Dms1ΔTM was not observed at the SPB but was dispersed throughout the cytoplasm (Fig 3C). The levels of GFP-Dms1ΔTM protein were comparable to those of wild-type GFP-Dms1 (Fig 3D), demonstrating that the sporulation defect of dms1ΔTM was not due to degradation of the mutant protein.
To determine whether the transmembrane domain alone is sufficient for the localization of Dms1, we expressed the GFP-fused transmembrane domain of Dms1 in the dms1Δ strain (dms1TM). As shown in Fig 3C, although GFP signal was clearly detected at the periphery of the nucleus and partially at the SPB, the dms1TM mutant did not undergo sporulation ( Fig  3B). Thus, these data suggest that, via its transmembrane domain, Dms1 localizes at the nuclear envelope and the SPB, where it carries out its sporulation function.

Dms1 is essential for cellular localization of Spo15, Spo2, and Spo13
To examine the relationship between Dms1 and other meiotic SPB components, we simultaneously expressed GFP-Dms1 and either Spo13-mCherry or Spo15-mCherry. As shown in Fig 2B, GFP-Dms1 was in close contact with both proteins at the SPB. Interestingly, the GFP-Dms1 signal was positioned within the Spo13-mCherry signal at the SPB, but overlapped with the Spo15-GFP signal.
Recruitment of Spo2, Spo13, and Spo15 to the SPB is known to be strictly controlled: localization of Spo13 depends on both Spo15 and Spo2, whereas that of Spo2 depends on only Spo15 [18]. We therefore examined whether the dms1Δ mutation affects the localization of these proteins. In wild-type vegetative cells, Spo15-GFP localized to the SPB as previously reported [16]. By contrast, the Spo15-GFP signal was not restricted to the SPB but was diffused throughout the cytoplasm in dms1Δ cells during both vegetative growth and sporulation ( Fig  4A and 4B), indicating that the localization of Spo15 to the SPB requires Dms1 function. In addition, Spo2-GFP and Spo13-GFP signals were not detected at the SPB, but were dispersed in the nucleus in dms1Δ cells (Fig 4A). These observations are similar to our previous finding that Spo2 and Spo13 localize in the nucleus in the spo15 deletion strain [18]. By contrast, GFP-Dms1 localized to the SPB in spo2Δ, spo13Δ, and spo15Δ cells (Fig 5). Moreover, Spo15-GFP did not localize at the SPB in dms1ΔTM or dms1TM cells (Fig 4C), supporting our findings that the transmembrane region of Dms1 is essential for its localization and function. As described above, Dms1 also localizes to the nuclear envelope (Fig 2A). The signal at the nuclear envelope became stronger when Dms1 was overexpressed (Fig 4D). Moreover, GFP-Spo45 signal was also observed at the cell periphery. Even in these cells, the distribution of Spo15-mCherry signals was essentially identical to that of wild strain (Fig 4D). Consistent with these results, Dms1 overexpression did not affect the sporulation frequency (Fig 4D). Taken together, these data suggest the possibility that Dms1 anchors Spo15 via other SPB component(s), and thereby Spo2 and Spo13, at the SPB. To further examine this possibility, we constructed a plasmid for expressing a Spo15-GFP-Dms1TM fusion protein, in which Spo15 was fused to the transmembrane region of Dms1, and introduced it into the dms1Δ strain.  However, Spo15-GFP-Dms1TM signal was not detected at the SPB and the fusion gene did not suppress the sporulation defect of dms1Δ cells. Western analysis revealed that the amount of the fusion protein was remarkably decreased compared to the wild-type Spo15 (data not shown). Therefore, we presume that the defect of suppression by Spo15-GFP-Dms1TM is due to degradation of the mutant protein.
Previously, we showed that the calmodulin Cam1 regulates the initiation of FSM formation by recruiting the Spo15-Spo2-Spo13 complex to the SPB [17]. Next, therefore, we examined the behavior of Cam1 in dms1Δ cells. GFP-Cam1 signal was detected at the SPB in dms1Δ cells as in wild-type cells. Reciprocally, GFP-Dms1 was localized at the SPB in cam1 mutant cells (Figs 4A and 5), suggesting that Dms1 does not interact with Cam1.
Lastly, we examined the meiotic SPB component Spo7, which is known to localize to the SPB independently of Spo2, Spo13, Spo15, and Cam1 [19]. Similar to Cam1, localization of Spo7 at the SPB was not affected by the dms1Δ mutation (Fig 4A), and GFP-Dms1 was localized at the SPB in spo7 mutant cells (Fig 5).

Discussion
In S. pombe, the SPB functions as an origin of FSM formation. Although several meiotic SPB components have been identified, the full picture of how these proteins are recruited to the SPB remains unclear. In this study, we have examined the SPB component Dms1, showing that it has a transmembrane domain and locates at the innermost position of known meiotic components on the SPB. Our findings suggest the possibility that Dms1 anchors other meiotic SPB components to the SPB, thereby regulating the initiation of FSM formation.
Although FSM formation initiates in meiosis, Dms1 is expressed even in vegetative cells, as revealed previously by DNA microarray [36] and here in western blot analyses (Figs 1 and 2). Fluorescence microscopy also showed that GFP-Dms1 localizes at the SPB throughout the mitotic cell cycles and meiotic nuclear divisions, confirming the constitutive expression of dms1 + . The structure and behavior of SPBs and spindles appeared normal in dms1Δ cells, Role of the SPB in sporulation consistent with the fact that dms1 null mutants exhibited no apparent defects in either mitosis or meiotic nuclear division. Interestingly, expression of dms1 + was previously found to be upregulated under various environmental stresses such as heat shock, heavy metal, oxygen, and osmotic shock [36], suggesting the possibility that Dms1 is involved in the stress response. We found, however, that the growth of dms1Δ cells under these stresses was indistinguishable from that of wild-type cells (S2 Fig). The possibility that Dms1 plays a dispensable role in mitotically growing cells also remains to be tested.
In Saccharomyces cerevisiae, sporulation proceeds in a similar manner to that in S. pombe [38]. Assembly of the prospore membrane, corresponding to the S. pombe FSM, initiates with modification of the SPB during meiosis II. The prospore membrane elongates by membrane vesicle fusion, and encapsulates the haploid nucleus generated by two rounds of meiotic nuclear division. Despite these similarities, the structure of known meiotic SPB components are poorly conserved between S. cerevisiae and S. pombe. Consistent with this, to our knowledge, no meiotic SPB component with a transmembrane domain has been identified in S. cerevisiae. Duplication of the SPB also occurs slightly differently in these two yeasts: the S. cerevisiae SPB remains embedded in the nuclear membrane throughout the cell cycle, whereas the S. pombe SPB spends most of interphase in the cytoplasm adjacent to the nuclear envelope and duplicates there. After duplication, the S. pombe SPB is associated with the cytoplasmic face of the nuclear envelope. As the cell enters M phase, the nuclear envelope becomes fenestrated and the SPBs separate and enter the nucleus, forming the spindle microtubule [9,39,40]. We therefore presume that structural differences in the SPB components of the two yeasts may be due to the behavior of the SPB.
As mentioned above, Blyth et al. recently reported that Dms1 has a role in sporulation [35]. Via an immunoprecipitation assay and mass spectrometry, they showed that Dms1 interacts with Spo15; therefore, they suggested that Dms1 has a role in FSM formation via Spo15, although they did not examine this experimentally. To confirm whether Dms1 binds to Spo15 directly, we performed a yeast two-hybrid assay and pull-down assay. However, no positive interaction between these two proteins was detected in either assay (S3 Fig), suggesting that Dms1 may bind to Spo15 via unknown protein. Blyth et al, also reported that dms1Δ cells show abnormal behavior of spindle microtubules in meiosis II and suggested that Dms1 regulates the spacing of nuclei produced by meiosis [35]. We consider that the defect of sporulation in dms1Δ cells is mainly due to a fault in the recruitment of SPB components including Spo15, Spo2, and Spo13 which are essential for initiation of FSM formation. In addition, we did not observe a defect of the meiotic nuclear division in dms1Δ cells ( Fig 1D); however, because both spindle formation and FSM formation initiate at the SPB, it is possible that Dms1 coordinates the progression of meiotic nuclear division and FSM formation. Fig 6 illustrates our working hypothesis in which Dms1 is involved in construction of the meiotic SPB. During vegetative growth, Dms1 localizes to the SPB through its transmembrane domain. Dms1 also shows a hierarchical arrangement, localizing at the most nuclear side of the SPB among the known meiotic SPB components (i.e., Cam1, Spo2, Spo7, Spo13, Spo15, Ypt2, and Ypt3). Subsequently, Spo15 is recruited to the SPB and may bind to Dms1 via an unknown protein, X. Our previous study showed that the calmodulin Cam1 binds to Spo15, and this interaction is essential for SPB localization of Spo15. However, Dms1 and Cam1 localize to the SPB independently (Fig 4A and 4B). Consistent with this, no interaction was detected between Dms1 and Cam1 by a yeast two-hybrid or pull-down assay (data not shown). During meiosis I, Spo2 and Spo13 are expressed and localized to the SPB, constructing the MOP. Spo13 then recruits and anchors the Rab-family GTPase Ypt2 [15]. When cells proceed to meiosis II, the MOP is differentiated ("SPB modification") by a currently unknown mechanism, and the FSM, which is continuously associated with the MOP, expands by membrane vesicle fusion.
How is Dms1 connected to the SPB? The transmembrane region of Dms1 is sufficient for its localization at the nuclear envelope but not at the SPB (Fig 3C). One possibility is that the N-terminal region of Dms1 interacts with core components of the SPB and stabilizes its localization. Several SPB components have been identified. We examined the physical interaction of Dms1 with seven of them (Sad1, Pcp1, Cdc11, Sid4, Ppc89, Kms1 and Cut11), but did not obtain positive results (data not shown). A comprehensive analysis of Dms1 interaction with SPB components will be necessary in future studies.
As described above, FSM formation shares similarities with mammalian primary ciliogenesis, in which the plasma membrane is newly synthesized in the vicinity of the centrosome. The distal appendages (DAPs) of centrioles are thought to anchor cilia to the plasma membrane, suggesting that DAPs might play a role similar to the MOP. Recently, several DAP components have been identified. Similar to meiotic SPB components, they localize to the DAP in a hierarchical manner that is essential for ciliogenesis [41,42]. However, these DAP components share very low structural similarity to those of the S. pombe MOP. On the other hand, Rab family proteins and their regulators are involved in both formation of the ciliary membrane and formation of the FSM [15,[43][44][45][46]. Thus, the mechanism by which the MOP (DAP) activates Rab proteins, thereby initiating membrane formation, should be elucidated in future studies.
In conclusion, we have identified and characterized Dms1 as a SPB component that is crucial for FSM formation, and shown that it maintains meiotic SPB components at the SPB, thereby regulating the initiation of FSM formation. Our findings will help to better understand not only the molecular mechanisms of FSM formation but also the function of the SPB as an origin of membrane organization.  Fig. dms1Δ is not sensitive to various stress. Homothallic haploid wild-type (L968, KI175, and TN105) and dms1Δ (CAL72, CAL92, and CAL93) strains were serially diluted from 10 2 to 10 7 cells/ml, spotted onto YEA, YEA + 1mM H 2 O 2 , YEA + 40 μM CdSO 4 or YEA + 1.2 M sorbitol plates, and incubated at 30˚C for 2 days. To test heat shock stress, cells were spotted onto YEA and treated at 50˚C for 30 min before incubation at 30˚C for 2 days.  prepared from vegetative cells expressing the tagged proteins, GST or GST-Dms1, were subjected to pull down with anti-GFP antibody. Precipitates were analyzed by western blotting using anti-GFP, or anti-Spo15 antibody. (TIF) S1 Movie. Behavior of GFP-Psy1 and mCherry-Atb2 in wild type. The movie essentially corresponds to the images shown in Fig 1C. The movie plays at 10 frames per second. (AVI) S2 Movie. Behavior of GFP-Psy1 and mCherry-Atb2 in dms1Δ. The movie essentially corresponds to the images shown in Fig 1C. The movie plays at 10 frames per second. (AVI)