The spindle pole body of Aspergillus nidulans is asymmetrical and contains changing numbers of γ-tubulin complexes

Microtubules (MTs) play essential roles in eukaryotic cells in many cellular processes, such as mitosis, protein and organelle transportation and cell polarity establishment. They are nucleated from large protein complexes called microtubule-organizing centers (MTOCs). The centrosome is a prominent MTOC in animal cells. Centrosomes organize the mitotic spindle and also polymerize cytoplasmic MTs in certain interphase cells. However, many – if not all – cell types employ non-centrosomal MTOCs (ncMTOCs) after differentiation to produce cell-type-specific MT arrays (Tillery et al., 2018). Whereas centrosomes are studied very well, the structure and function of ncMTOCs is in its infancy. The filamentous fungus Aspergillus nidulans has spindle pole bodies (SPBs), the functional analogs of centrosomes, and septum-associated MTOCs (sMTOCs) as ncMTOCs (Konzack et al., 2005; Oakley, 2004; Oakley et al., 2015; Oakley and Oakley, 1989; Zekert et al., 2010; Zhang et al., 2017). Thus, A. nidulans is a great model to study different MTOC types in differentiated cells.

The key component of MTOCs is γ-tubulin, which was first identified in A. nidulans (Oakley and Oakley, 1989), and later in fission yeast, animals, human and plants (Horio and Oakley, 2003; Horio et al., 1991; Julian et al., 1993; Liu et al., 1994). Together with the γ-tubulin complex proteins (GCPs, also known as TUBGCPs in mammals), GCP2–GCP6, it functions as an MT nucleation template. This large complex is termed the γ-tubulin ring complex (γ-TuRC) based on its highly ordered three-dimensional ring-like structure (Moritz et al., 2000; Zheng et al., 1995).

Three components, γ-tubulin, GCP2 and GCP3, constitute the core structure of the γ-TuRC and are termed the γ-tubulin small complex (γ-TuSC) (Kollman et al., 2010). The γ-TuSC proteins are essential in all eukaryotic cells, and Saccharomyces cerevisiae only contains γ-TuSCs (Colombié et al., 2006; Knop and Schiebel, 1997; Martin et al., 1998; Oegema et al., 1999; Vardy and Toda, 2000; Xiong and Oakley, 2009). γ-TuSC itself can self-assemble into a ring structure facilitating MT nucleation, although with lower efficiency than the γ-TuRC (Oegema et al., 1999). To distinguish GCP4–GCP6 from the core components of the γ-TuSC, they are also called non-core components. After high-salt treatment, the γ-TuRC dissociates into the γ-TuSC (Oegema et al., 1999; Zhang et al., 2000). In most animal and plant cells, the non-core components of the γ-TuRC play essential roles in γ-tubulin-related MT organization (Kong et al., 2010; Murphy et al., 2001). In contrast, these components are not essential in Drosophila melanogaster, Schizosaccharomyces pombe and A. nidulans (Anders et al., 2006; Vérollet et al., 2006; Xiong and Oakley, 2009). In A. nidulans, any single deletion of the γ-TuRC-specific components GcpD, GcpE and GcpF and double and triple mutants are viable (Xiong and Oakley, 2009).

MTOCs are present in different forms in animals, plants and fungi. The prominent MTOC in higher eukaryotes is the centrosome and its functional ortholog in fungi are the SPBs. Different from what is seen in animal cells, plants lack centrosomes but employ an acentrosome system (Binarová et al., 2006; Murata et al., 2005; Stoppin et al., 1994). The simplest, but best characterized MTOC is the SPB of S. cerevisiae. It consists of a three-layered structure embedded into the nuclear envelope. The outer plaque of the SPB polymerizes cytoplasmic MTs in interphase and astral MTs during mitosis, while the inner plaque produces the mitotic spindles (Jaspersen and Winey, 2004). Both plaques share the γ-TuSC. However, both plaques apply different γ-tubulin complex receptors. Spc72 (named Mto1 in S. pombe and ApsB in A. nidulans) recruits the γ-tubulin complexes to the outer plaque, while Spc110 (named Pcp1 in S. pombe and PcpA in A. nidulans) recruits them to the inner plaque (Chen et al., 2012; Knop and Schiebel, 1997; Knop and Schiebel, 1998; Lin et al., 2015; Nguyen et al., 1998; Samejima et al., 2008, 2010; Soues and Adams, 1998).

One interesting aspect of A. nidulans, in comparison to, for example, S. cerevisiae, is that two types of MTOCs have been described, SPBs and non-centrosomal septal MTOCs, and their structures and compositions appear to be different (Konzack et al., 2005; Zekert et al., 2010; Zhang et al., 2017). This situation is similar to S. pombe where two non-centrosomal MTOCs are found, the interphase (i)MTOCs and equatorial (e)MTOCs (Piel and Tran, 2009). However, in S. pombe those MTOCs are only transient structures, whereas sMTOCs of A. nidulans are formed during septation and remain at septa permanently (Heitz et al., 2001; Zhang et al., 2017). The SPBs in A. nidulans generate spindles and astral MTs during mitosis and cytoplasmic MTs during interphase (Manck et al., 2015; Morris and Enos, 1992; Suelmann et al., 1997). Several components of SPBs, like the γ-TuRC and the outer plaque receptors ApsB and Spa18 (Mto2 in S. pombe and absent in S. cerevisiae) are found at sMTOCs, while the inner plaque receptor PcpA (Pcp1 in S. pombe and Spc110 in S. cereviase) and proteins required for SPB duplication are missing at septa. As a further difference, an intrinsically disordered protein, Spa10, has been identified as an anchor of sMTOCs, and is targeted to septa earlier than ApsB or Spa18 and concentrated as a central disk structure at mature septa (Zhang et al., 2017). This protein is not present at SPBs. Moreover, A. nidulans has both large and small γ-tubulin complexes (Xiong and Oakley, 2009). However, it remained open to what extent the two complexes exist, or whether the small complexes are the result of a salt-dependent dissociation of GCPs from the γ-TuRC.

Recently, novel MTOC proteins were discovered in human and named mitotic-spindle organizing protein associated with a ring of γ-tubulin or short, MOZART proteins. Three isoforms were described, MOZART1, MOZART2A and MOZART2B (also known as MZT1, MZT2A and MZT2B, respectively) (Cota et al., 2017; Hutchins et al., 2010; Teixidó-Travesa et al., 2010). MOZART2A/2B, also both called GCP8, are involved in interphase MT organization. MOZART2A and 2B are only conserved in the deuterostome lineage, while MOZART1 is conserved in most eukaryotes including in Xenopus laevis, D. melanogaster, Arabidopsis thaliana [GIP1a,b or GIP2/GIP1 (Janski et al., 2012; Nakamura et al., 2012), S. pombe (Tam4/Mzt1; Dhani et al., 2013; Masuda et al., 2013)] and Candida albicans (CaMztA; Lin et al., 2016), but not in S. cerevisiae. MOZART1 is a novel component of the γ-TuRC. In S. pombe, Mzt1 is essential for MT nucleation and γ-TuRC recruitment to MTOCs (Dhani et al., 2013; Masuda et al., 2013). In C. albicans, MztA interacts with the receptor proteins Spc72 and Spc110 to turn the γ-TuSC into an active MT nucleation template (Lin et al., 2016). Hence, MTOCs in fungi appear to be unexpectedly variable and can consist of γ-TuSCs without MOZART1 (S. cerevisiae), γ-TuSC with MOZART1 (C. albicans) and γ-TuRC with MOZART1 (S. pombe). In D. melanogaster MOZART has also been characterized and also revealed heterogenity of different MTOCs (Tovey et al., 2018). In A. nidulans, it was reported that MztA is involved in the regulation of septation through suppressing the phosphatase ParA (Jiang et al., 2018). However, nothing is known about the function of MztA in MTOCs and MT polymerization. In this study, we show that the outer plaque of the SPB of A. nidulans consists of γ-TuSCs without MztA and the inner plaque of γ-TuRCs including MztA. The non-centrosomal septal MTOCs contain MztA in combination with components from the outer plaque of the SPBs. In addition, we present evidence that the number of γ-TuRCs in MTOCs during interphase is dynamic.

We identified gene AN1361 in A. nidulans whose translation product showed high similarity to H. sapiens MOZART1 (56% identity), S. pombe Mzt1 (52%) and C. albicans Mzt1 (40%) (Fig. 1A). It had been previously been named mztA (Jiang et al., 2018). The gene is 379 bp in length and is located on chromosome VIII. In S. pombe Mzt1 was originally described as a 94-amino-acid-long protein, later corrected to being 64 amino acids long (Dhani et al., 2013; Hutchins et al., 2010; Masuda et al., 2013). In A. nidulans two versions were also predicted, of 64 and 74 amino acids in length. RNAseq data from our laboratory confirmed two introns (the ATG of the shorter version lies in the intron region), and re-complementation of a deletion mutant (see below) was only possible with the long version.

To characterize MztA in A. nidulans, we first asked whether MztA is a component of SPBs and septal MTOCs (Fig. 1B). All γ-TuRC components (γ-tubulin, GcpB, GcpC, GcpD, GcpE and GcpF) located to both types of MTOCs (Xiong and Oakley, 2009). Endogenous mztA was fused with GFP at its 3′ end and visualized in a strain where nuclei were labeled with DsRed (Toews et al., 2004). All nuclei showed one bright green spot, suggesting that MztA was a SPB-associated protein (data not shown). To co-visualize MztA and mitotic spindles, the MztA–GFP construct was transformed into a strain expressing mCherry-tagged α-tubulin. The green fluorescent signals clearly localized to the poles of the mitotic spindles (Fig. 1C, left panels). In addition to the SPB localization, MztA localized as several dots at septa, as was previously described for other sMTOC components (Zhang et al., 2017) (Fig. 1B,C, right panels). These localization patterns are reminiscent of the localization of GcpC or γ-tubulin. In order to exclude a possible interference of GFP with MztA function, we showed that the MztA–GFP construct fully rescued the mztA deletion phenotype (data not shown). By contrast, N-terminally tagged MztA only localized to SPBs but not to septa, indicating that the N-terminus is important for sMTOC recruitment (Fig. S1A). In order to prove the MztA localization with an alternative method, we tagged MztA C-terminally with the small HA tag and performed immunofluorescence analysis with the transgenic strains. MztA–HA clearly localized to SPBs (Fig. S1B), and also at septa (data not shown). Signals at septa were very weak. In addition, immunostaining always leads to some spot-like, unspecific signals in the cytoplasm, making it difficult to distinguish septal signals from those unspecific signals. We excluded the possibility of MztA locating at kinetochores, which also appear as a single spot at nuclei, by showing that MztA–GFP and mRFP–KatA did not colocalize (Fig. S1C).

To gain insights into MztA functions, we deleted the entire open reading frame (ORF) of mztA by homologous recombination. The deletion event was confirmed by diagnostic PCR and southern blotting. In contrast to the case in S. pombe, mztA is not essential in A. nidulans (Dhani et al., 2013; Jiang et al., 2018; Masuda et al., 2013) (Fig. 2A–D). The sporulation of the ΔmztA strain was decreased as compared to wild type but not as much as in apsB deletion mutants. In apsB mutants, the primary defect is a nuclear migration defect due to reduced astral MT formation (Suelmann et al., 1998). Conidiophores of the mztA deletion strain, however, resembled wild-type conidiophores and nuclear migration appeared not to be affected (data not shown). The developmental phenotype was completely rescued after transformation with the wild-type mztA or the MztA–GFP construct (Fig. 2D). A putative shorter MztA variant (MztA-S) did not complement (Fig. 2E).

As a difference to what was seen in wild type, we observed more elongated nuclei in hyphae, suggesting a delay in mitosis. To test for a role for MztA in mitosis, we introduced GFP-tagged TubA into the ΔmztA strain and into wild type. Mitosis required ∼10 min in the mztA mutant and only ∼5 min in wild type (Fig. 3A,C). A total of 60% of the spindles in the ΔmztA strain showed abnormal morphologies (bent spindles, loose bipolar spindles and loose monopolar spindles; Fig. 3B,C). In wild type, defective spindles are very rare. Thus, MztA plays an important role in spindle assembly and stability, and reduced spore formation appears to be a consequence of retarded mitoses and not of astral MT malfunction.

Because MztA appears to play an important role for the functioning of the MT cytoskeleton, we did initial experiments to test whether deletion of mztA affects MT stability. We did not observe differences in the sensitivity towards the MT-destabilizing drug benomyl in comparison to wild type (data not shown).

To investigate the exact time for recruitment of MztA during septum formation, we constructed a strain in which mCherry-tagged tropomyosin TpmA was introduced into a strain with GFP-tagged MztA. TpmA binds specifically to F-actin and localizes very early at the septation site and follows the constricting ring (Bergs et al., 2016). We did not observe any colocalization of MztA and TpmA even at nearly mature septa (Fig. 4A). Thus, the time of recruitment of MztA was similar to ApsB and Spa18, which only appear at mature septa (Zhang et al., 2017).

Next, we asked whether MztA recruitment at septa requires ApsB and/or Spa18 (the orthologs of S. pombe Mto1 and Mto2) or the anchoring protein Spa10. MztA–GFP was introduced into apsB, spa18 and spa10 deletion mutants (Fig. 4B). MztA was not detected at septa in any of the gene deletion strains, while targeting to SPBs appeared to be unaffected. On the other hand, ApsB localization was not affected at SPBs nor at sMTOCs in the mztA deletion mutant (Fig. S2). These localization dependences are similar to that of the γ-TuRC component GcpC, which also requires Spa10, ApsB and Spa18 at sMTOCs (Zhang et al., 2017).

Next, we tested whether MztA interacts with the γ-TuRC component GcpC. By using bimolecular fluorescence complementation (BiFC), we detected interaction of MztA and GcpC at SPBs (Fig. 5A, left upper panels). We confirmed this interaction via co-immunoprecipitation (co-IP) assays. In a strain containing MztA–GFP and 3HA–GcpC, we detected GcpC in the precipitate of MztA. In a strain containing only MztA–GFP or only 3HA–GcpC, no GcpC band was detected (Fig. 5A, lower panel). Thus, MztA binds directly to the γ-TuRC component GcpC. No interaction signal was detected at septa, indicating different arrangements of γ-TuRCs at nuclei and at septa (Fig. 5A, right upper panels).

We next asked whether MztA plays a role in GcpC recruitment to SPBs or to sMTOCs. We introduced GcpC–GFP into an mztA deletion strain and compared the GFP signal intensity to the one in a corresponding wild-type strain. At septa, GcpC signals were observed in wild type but not in the mztA deletion strain (Fig. 5B). Therefore, we analyzed the sMTOC activity by using the MT plus-end-tracking kinesin 7 motor protein KipA (Konzack et al., 2005). In wild type, KipA signals continuously emerged from septa whereas in the mztA mutant the sMTOCs were inactive (Movies 1 and 2). At SPBs, the GcpC signal was detected in the mztA deletion strain with ∼70% of the intensity of that in wild type (Fig. 5C). Hence, MztA appears to be essential for MT polymerization at sMTOCs but not at SPBs.

The function of MztA in A. nidulans appeared to be different to the essential function in S. pombe or C. albicans (Dhani et al., 2013; Lin et al., 2016; Masuda et al., 2013). The different phenotypes of an mztA and an apsB mutant in A. nidulans suggested that MztA plays a role in the organization of mitotic spindles but not in nuclear distribution, which requires proper astral MT organization and thus proper function of the outer plaque of the SPBs (Zhang et al., 2017). This prompted us to speculate that MztA locates and functions only at the inner plaque and not at the outer plaque of SPBs. To test this hypothesis, we assessed the colocalization of MztA with PcpA or ApsB. In a strain in which MztA was tagged with GFP and PcpA with mCherry, we found colocalization at nuclei in 82% of the observed SPBs (n=77) (Fig. 6A, upper panels). However, in a similar experiment with tagged MztA and ApsB, we observed a clear gap between the two fluorescent spots in 79% of the cases (n=63) (Fig. 6A, lower panels). The fact that in 21% of the observed SPBs, the MztA and ApsB signal overlapped is due to different observation angles (top view or side view). To further confirm this observation, we studied interaction of MztA with ApsB and/or PcpA. Indeed, there was interaction between MztA and PcpA as shown by BiFC, but not between MztA and ApsB as shown by BiFC and co-IP (Fig. 6B,C). Co-IP experiments with PcpA were unsuccessful, because PcpA was very unstable in our experiments. Taken together, we conclude that MztA resides only at the inner plaque of SPBs.

A previous study revealed that the γ-TuRC-specific components GcpD, GcpE and GcpF are not essential for cell viability, spindle formation or sexual reproduction, and their localization at SPBs depends on the γ-TuSC (Xiong and Oakley, 2009). Here, we analyzed the localization of GcpD in the presence or absence of MztA. GcpD was tagged with mEosFP at the C-terminus. In wild type, GcpD–mEosFP dots were observed at nuclei. In the mztA deletion strain, the GcpD signal was completely absent. In comparison, GcpD SPB localization was independent of ApsB (Fig. S3A). To confirm that GcpD resides only at the inner plaque of SPBs, we did similar experiments to those for MztA. In a strain in which GcpD was tagged with GFP and PcpA fused with mCherry, we observed colocalization in 80% of SPBs (n=50). In contrast, GcpD and ApsB signals did not overlap at ∼80% of the nuclei (n=50) (Fig. S3B). We conclude that MztA recruits GcpD to the inner plaque of SPBs to assemble the γ-TuRC. Furthermore, this suggests that the outer plaque consists only of γ-TuSCs.

To get more information on the organization of the different MTOCs in A. nidulans, we used photo-activated localization microscopy (PALM) to quantify the numbers of single molecules (Fig. 7A). In previous studies, the stoichiometry of MTOC components in yeast was determined biochemically (Masuda et al., 2013). However, these methods did not allow SPBs and non-nuclear MTOCs to be distinguished in A. nidulans. We tagged the three γ-TuRC components GcpC, GcpD, and MztA, and the outer plaque receptor ApsB with mEoSFPthermo C-terminally and determined their numbers at SPBs and sMTOCs. The proteins were tagged at the native gene locus. We counted 22–60 SPBs and found ∼20 molecules of ApsB, ∼32 of GcpC, ∼10 of GcpD and ∼31 molecules of MztA at each interphase SPB (Table S4). The estimated molecular ratio of ApsB, GcpC, GcpD and MztA at SPBs in interphase was about 2:3:1:3 (Table S4). We also determined the number of proteins at sMTOCs. For each protein, 10–20 septa were analyzed. On average ∼5 molecules of GcpC, ∼3 molecules of GcpD and ∼6 molecules of MztA were found. Thus, the estimated molecular ratio of GcpC, GcpD and MztA at sMTOCs in interphase is about 3:2:4 (Table S4). We did not quantify ApsB at sMTOCs because C-terminal tagging of ApsB interfered with its localization. The molecular ratio of γ-TuRC at sMTOCs is different from that at SPBs, which might indicate a new structure of γ-TuRC assembly. We did not quantify ApsB at sMTOCs because C-terminal tagging of ApsB interfered with its localization.

Further interpretation of the molecular ratios of the analyzed proteins at SPBs appears to be difficult, because we noticed large fluctuations of the numbers of molecules during the cell cycle at individual SPBs. In the case of GcpC, numbers varied from 4 to 31 in interphase. When we plotted the numbers, we realized there were stepwise increases with multiples of 6 or 7 (Fig. 7B, left panel). Assuming that each γ-TuRC contains 13–14 γ-tubulin and 6–7 GcpC molecules, the number of γ-TuRCs ranges from 1 to 5. In addition, but rarely, GcpC numbers ranging up to 105 molecules were detected. The average increase of MTOC components from interphase to mitosis was between 50% and 100% (Table S4). The increase of MztA and GcpC numbers from interphase to mitosis was also obvious in intensity measurements undertaken with standard epifluorescence microscopy (data not shown). GcpC numbers (∼60) appeared to be constant during metaphase (Fig. 7B, right panel). With 5–6 GcpC molecules per γ-TuRCs, this number suggests the presence of ∼10 γ-TuRCs at each SPB during mitosis. The number of ApsB molecules at SPBs varied between 20 and 50 during mitosis (Fig. S4A). Assuming that each γ-TuSC contains between 14 and 20 ApsB molecules, as was calculated in S. pombe (Lynch et al., 2014), the number of γ-TuRCs would be between 1 and 3–4. This corresponds well to the small number of astral MTs polymerized from outer plaques. We also analyzed Spa18 at SPBs by time-lapse epifluorescence microscopy. At the beginning of mitosis, no Spa18 signals were detected at the very short mitotic spindles. As mitosis progressed, the signal intensity increased (Fig. S4B).

The composition and functioning of MTOCs is a fascinating field of research, where the study of lower eukaryotes such as S. cerevisiae and S. pombe revealed many important insights (Cavanaugh and Jaspersen, 2017; Ruthnick and Schiebel, 2018; Wu and Akhmanova, 2017). Here, we studied A. nidulans protein MztA, the homolog of the MTOC protein MOZART1, relatively recently discovered in humans, and found that it is not required for viability but is required for development. We discovered that the SPB is composed of γ-TuSCs at the outer and γ-TuRCs at the inner plaque. MztA was only found at the inner plaque. The septal MTOCs also contain MztA, but the anchorage of the γ-TuRCs resembles the anchorage of the γ-TuSCs at the outer plaque of SPBs. Thus, sMTOCs combine properties of both, the outer and the inner plaque of SPBs (Fig. 8A). Our data add to the increasing complexity of MTOCs in eukaryotes, with different compositions, architectures and functions (Fig. 8A). Hence, in S. cerevisiae, C. albicans, S. pombe and A. nidulans four different γ-tubulin complexes are described, γ-TuSC, γ-TuSC with MOZART, γ-TuRC with MOZART and γ-TuRC with MOZART but alternative receptors (Cavanaugh and Jaspersen, 2017; Lin et al., 2016; Xiong and Oakley, 2009). A. nidulans, in addition offers the opportunity to study different MTOCs in a cell. This resembles the situation in other eukaryotes, such as D. melanogaster, where centrosomes and several non-centrosomal MTOCs have been described (Tillery et al., 2018; Tovey et al., 2018).

Our results will be discussed mainly as three points: (1) the role of MOZART in MTOC functioning, (2) the organization of the A. nidulans SPBs and sMTOCs, and (3) the dynamics of the composition of the SPB.

In S. pombe and C. albicans, Mzt1 plays crucial roles for the recruitment of γ-TuSCs to SPBs and activation of MT nucleation (Dhani et al., 2013; Masuda et al., 2013). In C. albicans, Mzt1 together with the receptor proteins Spc110 and Spc72 promote γ-TuSC oligomerization into active MT nucleation rings (Lin et al., 2016). Hence, Mzt1 plays a dual role in C. albicans in the recruitment of γ-TuSCs to the SPBs and in their oligomerization and activation. However, in A. nidulans, the GcpC signal at SPBs was only reduced by 30% in the mztA mutant suggesting that MztA has lost the essential role for the recruitment of γ-TuSCs to SPBs. Still, it appears to improve the assembly. MztA has also lost its essential function for the activation of the γ-TuRC. Hence, the γ-TuRC of A. nidulans appears to be more similar to the γ-TuSC in S. cerevisiae, where MOZART is absent. However, MztA was essential for MTOC activity at septa. To explain this, it has to be considered that sMTOCs are comprised of γ-TuRCs, like the inner plaque of the A. nidulans SPB, but that the receptor proteins are the outer plaque proteins ApsB and Spa18. Taken together, we find that A. nidulans MztA plays an important role at the inner plaque of SPBs during mitosis and an essential role at sMTOCs during interphase. In human, the mitotic and the interphase functions of MOZART appear to be separated into two distinct proteins. Whereas MOZART1 (also called GCP9) is involved in γ-tubulin recruitment to centrosomes in mitotic cells, and its depletion leads to mitotic spindle defects (Hutchins et al., 2010), MOZART2 (also called GCP8) plays a role in γ-TuRC recruitment to interphase centrosomes (Teixidó-Travesa et al., 2010). Likewise, plant cells contain two homologous MOZART proteins, GIP1 and GIP2, or GIP1a and GIP1b. GIPs play roles in γ-tubulin complex localization and spindle stability during mitosis (Janski et al., 2012). Moreover, GIPs represent a subset of the Arabidopsis γ-tubulin complexes, which is consistent with the roles we have found for MztA (Nakamura et al., 2012). We therefore speculate that MztA in A. nidulans is more related to the proteins found in human or plant cells than in fission yeast, since MztA functions during mitosis as does MOZART1 in human or plant cells. The MOZART2 homolog, which should specifically function in interphase SPBs, is absent in A. nidulans. In contrast, Mzt1 in S. pombe or C. albicans combines the functions of MOZART1 and MOZART2. A. nidulans appears to be a link between γ-TuSCs functioning without the MOZART protein and fungi where MOZART is essential.

MztA specifically localizes at the inner plaque of A. nidulans SPBs and recruits γ-TuRC-specific components, whereas the outer plaque only contains γ-TuSCs (Fig. 8A). This asymmetric architecture is quite different from S. cerevisiae and S. pombe. In budding yeast, both plaques harbor only γ-TuSCs while in fission yeast both layers contain γ-TuRCs (Cavanaugh and Jaspersen, 2017). In A. nidulans, the outer plaque hence resembles the composition found in budding yeast, while the inner plaque resembles the composition found in fission yeast. The large γ-TuRC is composed of γ-tubulin, GcpB, GcpC, GcpD, GcpE, GcpF and MztA, and the latter four components are not essential for viability. A previous study described the hierarchy of the recruitment of the components with the order GcpF, GcpD and GcpE (Xiong and Oakley, 2009). Here, we showed that after γ-TuSC recruitment, MztA binds before GcpF, GcpD and GcpE.

The asymmetry of the SPB raises the question for reasons for this arrangement or the advantage of γ-TuRC being at the inner plaque in comparison to γ-TuSC. One explanation could be enhanced efficiency of MT nucleation during mitosis. However, we do not have evidence yet that the outer plaque of the SPB is less efficient in MT polymerization than sMTOCs, which contain γ-TuRCs and MztA. Another possibility could be that MztA might be involved in the regulation of spindle formation through interaction with specific kinases, such as the polo-like kinase PlkA. PlkA plays important roles in spindle assembly and corresponding mutants exhibit similar spindle defects to those seen in mztA mutants (Bachewich et al., 2005; Mogilevsky et al., 2012).

Thus, sMTOCs represent a novel class of γ-TuRC-containing MTOCs (Fig. 8A). The composition of sMTOCs is different from both the outer and the inner plaque of SPBs. sMTOCs consist of γ-TuRCs including MztA, GcpD, GcpE and GcpF, but also the outer plaque receptors ApsB and Spa18. In addition, a septal-pore associated protein Spa10 exclusively locates at sMTOCs (Zhang et al., 2017). The inner plaque receptor PcpA is missing in sMTOCs. The presence of MztA at the inner plaque and at sMTOCs may suggest some common regulation or coordination of the activities in a cell-cycle-dependent manner. However, the role of MztA appears to be different at sMTOCs and the inner plaque of SPBs. Whereas MztA is essential at sMTOCs, it is not in SPBs. Comparable to the situation in A. nidulans, γ-TuRC heterogeneity has been reported for D. melanogaster (Tovey, et al., 2018)

The SPB is a highly dynamic structure given that it divides prior to mitosis once during each cell cycle. Likewise, we found that the number of several γ-TuRC proteins increases during mitosis. This holds true for inner and for outer plaque proteins. However, more astonishing was the observation that the numbers of γ-TuRCs at different SPBs varied drastically during interphase. Our calculations suggest the presence of 1 to 5 γ-TuRCs. The increase of the number of GcpC molecules occurred in steps of ∼6 molecules. This suggests the addition of preassembled γ-TuRCs rather than an assembly at the inner plaque. We cannot exclude the second possibility completely, because it could be that the assembly is extremely quick once initiated. There was already some evidence for such dynamics. For example, ultrastructural analyses of SPB in S. cerevisiae had previously revealed different sizes of the structure, suggesting different protein compositions (Byers and Goetsch, 1974). Later studies analyzing the fluorescent intensity of RFP-labeled PcpA suggested that the γ-TuRC receptor protein Spc110 (PcpA) was also dynamic during the cell cycle (Viswanath et al., 2017; Yoder et al., 2003). The authors found that the fluorescent intensity increased at the G1/S border during the cell cycle, when the daughter SPB was generated. By using FRAP experiments, they also found that ∼50% of Spc110 is exchanged in the old SPB. However, in this paper, we present the first direct evidence for changing numbers of γ-TuRC during interphase of the cell cycle. This observation raises the question of whether there could be an advantage of such dynamic behavior as compared to there being a stable structure.

The number of GcpC molecules at SPBs during mitosis was more stable and reached ∼60. Assuming that each γ-TuRC contains 5–6 GcpC molecules, this result suggests that there are ∼10 γ-TuRCs at each SPB, enough to connect each of the eight chromosomes of A. nidulans. However, in an electron microscopy study the number of MTs in mitotic spindles was determined to be 35–50 (Oakley and Morris, 1983), suggesting some release of MTs from the SPBs after their polymerization. This is in agreement with studies in haploid yeast cells using electron micrographs and tomographic reconstructions (Byers and Goetsch, 1974; Viswanath et al., 2017). It was shown that, in metaphase, that there are 16 kinetochore MTs, one for each of its 16 chromosomes, and 3–4 interpolar MTs (Winey and O'Toole, 2001).

The numbers of the outer plaque proteins ApsB and Spa18 gradually increased during mitosis while in the beginning of mitosis, no Spa18 was detected. This confirms that the outer plaque is inactive before anaphase because in the absence of ApsB or Spa18 γ-TuSCs cannot be recruited (Fig. 8B). In comparison to the step-wise increase of the number of GcpC molecules at the inner plaque, the gradual increase of ApsB at the outer plaque suggests assembly of the complex at the outer plaque rather than addition of pre-assembled γ-TuSC.

Supplemented minimal medium (MM) and complete medium (YAG) for A. nidulans were prepared as previously described, and standard strain construction procedures were used (Hill and Käfer, 2001). Expression of tagged genes under the control of the alcA-promoter was regulated by the carbon source: repression on glucose, de-repression on glycerol (2%), and induction on threonine (2%) (Waring et al., 1989). A list of A. nidulans strains used in this study is given in Table S1. Standard laboratory Escherichia coli strains (Top 10 F′) were used. Plasmids are listed in Table S2.

Standard DNA transformation procedures were used for A. nidulans (Yelton et al., 1984) and Escherichia coli (Sambrook and Russel, 1999). For PCR experiments, standard protocols were applied using a Biometra Personal Cycler (Biometra, Göttingen, Germany) for the reaction cycles. Phusion polymerase was used. Denaturation was achieved at 98°C, annealing temperatures were chosen according to the corresponding DNA oligonucleotides, and the polymerization temperature was 72°C. DNA oligonucleotides used in this study are listed in Table S3. DNA sequencing was undertaken commercially (MWG Biotech, Ebersberg, Germany). Total DNA was extracted from A. nidulans according to Zekert et al. (2010). Southern hybridizations were performed according to the DIG Application Manual for Filter Hybridization (Roche Applied Science, Technical Resources, Roche Diagnostics GmbH, Mannheim, Germany). The first-strand cDNA synthesis was carried out with a SuperScript III Reverse Transcriptase kit (Invitrogen).

The mztA deletion strain was created by protoplast transformation and homologous integration of a fusion PCR-derived knockout cassette. The flanking regions of mztA were amplified by PCR with genomic DNA as template, and primers mztA-LB_fwd/mztA-LB-pyroA linker_rev (see Table S3) for the upstream region of mztA and mztA-RB-pyroA linker_fwd/mztA-RB_rev for the downstream region. pyroA was amplified with pyroA_fwd/pyroA_rev. Then the left border (LB), pyroA and right border (RB) were fused together by fusion PCR with nested primers mztA-LB-N_fwd and mztA-RB-N_rev. The construct was ligated to pJET1.2/blunt, yielding vector pFY2. After transformation of pFY2 to TN02A3 and selection by diagnostic PCR (Table S3 primers; mztA-dele_check_fwd/mztA-RB_rev) and Southern blotting, the mztA deletion strain SXL49 was selected (Fig. S2).

We constructed two plasmids to check whether the 64 amino acid or 74 amino acid form of MztA could re-complement the mutant phenotype. The whole expressing frame, including promoter, ORF and AfpyrG, was fused into pJET with NEBuilder HiFi DNA Assembly Master Mix (NEB) in one step. Here we describe the 74 amino acid version. The gpdA promoter was amplified using primers: gpdA builder_fwd/gpdA_rev; ORF of MztA was amplified using primers: mztA379_gpdA linker_fwd/ mztA_pyrG linker_rev and pyrG was amplified using primers: pyrG_fwd/ pyrG builder_rev. The three fragments were ligated to pJET1.2/blunt resulting in plasmid pXL69. After transformation into the mztA deletion strain SXL49, the re-complementation strain SXL103 showed 100% complementation of the phenotype.

We tagged MztA with GFP at the C-terminus and expressed it from the natural promoter. One kb from the 3′-end of the mztA gene was amplified with the primer pair MztA-C-termi_fwd/MztA-C-termi_GA linker_rev, and 1 kb of the terminator region of the gene was amplified with the primer pair MztA-RB-pyrG linker_fwd/MztA-RB_rev. The fragment of GFP::pyrG cassette was amplified from pFNO3 (Yang et al., 2004) using primers GA linker_fwd/pyrG overhang_rev. Subsequently, the three fragments were fused together by fusion PCR with nested primers MztA-C-termi-N_fwd/MztA-RB-N_rev (Table S3). The construct was ligated to pJET1.2/blunt, resulting in plasmid pXL23. The plasmid was transformed into TN02A3, and the strain SMS4 where the construct replaced the endogenous 3′-end of the gene, was checked with primers pyrG check fwd/MztA C-termi check rev. The strain expresses MztA-GFP from the natural promoter and MztA-GFP is the only source of MztA. The same strategy was used for C-terminal tagging of mEoSFPthermo for MztA, GcpC, GcpD, ApsB. The primers were the same as the GFP-tagging cassette (Table S3). The fragment of mEoSFP::pyrG cassette was amplified from pRM35. The resulting plasmids are pXL60 (mztA::mEoSFP::pyrG), pXL63 (gcpD::mEoSFP::pyrG), pXL89 (gcpD::GFP::pyrG), pXL64 (apsB::mEoSFP::pyrG) and pXL66 (gcpC::mEoSFP::pyrG) (Table S2). After transformation into the wild-type TN02A3 strain, strains SXL85, SXL98, SXL99 and SXL100 were constructed (Table S1).

To check GcpC localization in an mztA deletion strain, we crossed the two strains SXL49 (ΔmztA) and SNZ-SH80 (GcpC–GFP) (Todd et al., 2007). After screening with microscopy and diagnostic PCR (primers: pyrG check fwd/gcpC C-termi check rev), the recombinant strain SXL53 was selected.

To investigate the timing of the recruitment of MztA at septa, colocalization with the actin ring marker TpmA (tropomyosin) was analyzed. TpmA was tagged with mCherry at the N-terminus in strains where MztA was C-terminally tagged with GFP. The plasmid pXL9 pCMB17apx-mCherry-tmpA was transformed into SMS4 (MztA–GFP) yielding strain SXL37. The homologous integration event for TpmA was confirmed with primers alcA check fwd/tpmA 1 kb Int check rev (Table S3).

To investigate colocalization of MztA and kinetochore protein KatA, KatA was tagged with mCherry at the N-terminus in the MztA–GFP-expressing strain SMS4. The plasmid pSH33 pMCB17apx-mRFP-katA was transformed into SMS4 resulting in strain SXL68.

To investigate the colocalization of MztA with PcpA or ApsB, the C-terminal tagging cassette of MztA in pXL23 was transformed into strain SXL80 or SXL7, where PcpA or ApsB was N-terminally fused with mCherry. The resulting strains were SXL96 and SXL88, respectively. Homologous integration events were checked with primers pyrG check fwd/MztA C-termi check rev (Table S3). The same method was used for determining the colocalization of GcpD with PcpA or ApsB. Plasmid pXL89, which contains the C-terminal GFP cassette of GcpD, was transformed into strain SXL80 or SXL7, yielding strains SXL149 and SXL148. Homologous integration events were checked with primers pyrG check fwd/GcpD C-termi check rev (Table S3).

To analyze the interaction of MztA with GcpC, ApsB or PcpA, a bimolecular fluorescence complementation assay (BiFC or Split YFP) was performed. In these analyses, MztA was tagged with YFP^C at the C-terminus under the control of the natural promoter. The C-terminal YFP^C tagging cassette was constructed similar to the GFP-tagging cassette via fusion PCR. The same primers were used as for creating the GFP tagging cassette. YFP^C::pyrG cassette was amplified. After fusion, the construct was ligated to pJET1.2/blunt, resulting in plasmid pXL31. We also constructed an YFP^N tagging cassette in pJET1.2 pXL30 using the same strategy. For the analysis of the interaction of MztA and GcpC, 1 kb of the gcpC open reading frame was amplified with primers gcpC_AscI_fwd and gcpC_PacI_rev, cloned into pMCB17apx-YFP^N vector, which contained YFP^N instead of GFP, yielding pXL33. The C-terminal YFP^C tagging cassette of GcpC was constructed with the same strategy as for MztA. For primers, see Table S3. The resulting plasmid is pXL50. Then pXL31 and pXL33 were co-transformed into TN02A3. After screening, the strain SXL5,1 where MztA was C-terminally tagged with YFP^C and GcpC was N-terminally tagged with YFP^N, was examined for YFP signals. In addition, pXL31 and pXL50 were also co-transformed into TN02A3. Strain SXL78, where MztA and GcpC were C-terminally tagged with YFP^C and YFP^N, was examined for YFP signals. All the YFP strains were checked for homologous integration events by PCR.

For the analysis of the interaction of MztA and PcpA, 1 kb of the pcpA open reading frame was amplified with primers pcpA_AscI_fwd and pcpA_PacI_rev, and cloned into pMCB17apx-YFP^N vector, giving pXL58. Then pXL31 and pXL58 were co-transformed into TN02A3. After screening, the resulting strain SXL82 was examined for YFP signals. The strain was checked for homologous integration by PCR. For the analysis of the interaction of MztA and ApsB, plasmids pXL59 (full-length apsB in pMCB17apx-YFP^N) and pXL31 were co-transformed into TN02A3. After three rounds of screening, no transformant showed YFP signals, suggesting no interaction between ApsB and MztA.

To confirm the interaction results of the BIFC assay, we performed co-IP assays. We constructed the Co-IP strains SXL108 and SXL109, where GcpC or ApsB were tagged with triple HA at the N-terminus of the protein, while MztA was tagged with GFP C-terminally. The control strains are SXL106, SXL107 and SMS4, where only one tag (HA or GFP) was present. The corresponding plasmids for the strains are pXL71 and pXL72. The homologous integration events were checked by PCR (Table S3). The Co-IP and control strains were pulled down with anti-GFP agarose beads and tagged proteins detected with anti-HA antibodies. GFP-Trap agarose beads were from Chromotek. HA antibodies were produced in rabbit and were from Sigma-Aldrich (H6908).

For live-cell imaging, fresh spores were inoculated in 0.5 ml MM (2% glycerol for alcA promoter induction) with appropriate selection markers on 18×18 mm coverslips (Roth, Karlsruhe). The samples were incubated at 28°C overnight, followed by 2 h incubation at room temperature before microscopy. VECTASHIELD Mounting Medium with DAPI (Vector Laboratories) was used for staining nuclei in cells. Light and fluorescence images were taken with the Zeiss Microscope AxioImager Z1 (Carl Zeiss, Jena, Germany) and the software ZEN pro 2012, using the Planapochromatic 63× or 100× oil immersion objective lens, and the Zeiss AxioCam MR camera. Image and video processing was undertaken with ZEN pro 2012, Adobe Photoshop or ImageJ (National Institutes of Health, MD, USA).

Alternatively, for in vivo time-lapse microscopy, cells were incubated in u-slide 8-well glass-bottom dishes from Ibidi (cells in focus) in 0.5 ml MM+2% glycerol and appropriate supplements, and an additional 2 ml of medium was added after overnight incubation. Movies were taken at appropriate intervals for time-lapse and maximum projections of a deconvolved z-stack images were used. For quantification of fluorescence intensity, images of 15–20 sections were taken along the z-axis at 0.27 µm increments. After deconvolution, projection images of maximum intensity were obtained, and maximum fluorescence intensities over the background intensity were used for statistical data analysis. A 5×5 pixel region of interest (ROI) was selected and two 20×20 pixel regions around the ROI were used for background subtraction.

All super-resolution cell images (PALM) were obtained with an inverted microscope (ELYRA 3D-PALM; Carl Zeiss AG) equipped with a high numerical aperture oil immersion objective (α-plan-Apochromat, 100×, numerical aperture, 1.46; Carl Zeiss AG), multiple excitation laser lines (405, 473, and 561 nm), an image splitter Optosplit (Cairn Research Ltd, Faversham, UK) and an electron-multiplying charge-coupled device (EMCCD) camera (iXon Ultra 897; Andor Technology, Ltd., Belfast, Northern Ireland). The fluorescent proteins were photo-activated by using low intensity 405 nm light and excited by 561 nm simultaneous illumination. After passing through the excitation dichroic, fluorescence emission was filtered by a 607/50 band-pass filter and recorded with a back-illuminated EMCCD camera (iXon Ultra 897; Andor Technology, Ltd.) at 50 ms/frame. All PALM images were constructed from 2000 images using Zen software (Carl Zeiss AG) with the PALM method and quantified with IMARIS 9 software (BITPLANE).

Significant differences were calculated with the two-tailed Mann–Whitney U-test with GraphPad 7. P-values are *P<0.05; **P<0.01; ***P<0.001 and ns, not significant.

We thank E. Wohlmann for excellent technical assistance.