Cyclins in aspergilli: Phylogenetic and functional analyses of group I cyclins.

We have identified the cyclin domain-containing proteins encoded by the genomes of 17 species of Aspergillus as well as 15 members of other genera of filamentous ascomycetes. Phylogenetic analyses reveal that the cyclins fall into three groups, as in other eukaryotic phyla, and, more significantly, that they are remarkably conserved in these fungi. All 32 species examined, for example, have three group I cyclins, cyclins that are particularly important because they regulate the cell cycle, and these are highly conserved. Within the group I cyclins there are three distinct clades, and each fungus has a single member of each clade. These findings are in marked contrast to the yeasts Saccharomyces cerevisiae, Schizosaccharomyces pombe, and Candida albicans, which have more numerous group I cyclins. These results indicate that findings on cyclin function made with a model Aspergillus species, such as A. nidulans, are likely to apply to other Aspergilli and be informative for a broad range of filamentous ascomycetes. In this regard, we note that the functions of only one Aspergillus group I cyclin have been analysed (NimECyclin B of A. nidulans). We have consequently carried out an analysis of the members of the other two clades using A. nidulans as our model. We have found that one of these cyclins, PucA, is essential, but deletion of PucA in a strain carrying a deletion of CdhA, an activator of the anaphase promoting complex/cyclosome (APC/C), is not lethal. These data, coupled with data from heterokaryon rescue experiments, indicate that PucA is an essential G1/S cyclin that is required for the inactivation of the APC/C-CdhA, which, in turn, allows the initiation of the S phase of the cell cycle. Our data also reveal that PucA has additional, non-essential, roles in the cell cycle in interphase. The A. nidulans member of the third clade (AN2137) has not previously been named or analyzed. We designate this gene clbA. ClbA localizes to kinetochores from mid G2 until just prior to chromosomal condensation. Deletion of clbA does not affect viability. However, by using a regulatable promoter system new to Aspergillus, we have found that expression of a version of ClbA in which the destruction box sequences have been removed is lethal and causes a mitotic arrest and a high frequency of non-disjunction. Thus, although ClbA is not essential, its timely destruction is essential for viability, chromosomal disjunction, and successful completion of mitosis.

Introduction

The eukaryotic cell cycle is driven by the periodic synthesis and destruction of <span class="Gene">cyclins that bind and activate a class of serine/<span class="Chemical">threonine kinases called cyclin-dependent kinases (CDKs). Distinct cyclin/CDK complexes form and become active at different stages of the cell cycle. Once a cyclin has completed its function, it is targeted for destruction, inactivating its partner CDK. The formation of cyclin/CDK complexes and subsequent cyclin destruction drives the cell through G1, S, G2 and mitosis in a sequential and irreversible manner. In addition to cell cycle regulation, cyclin/CDK complexes function in a variety of cellular processes, such as transcriptional regulation, the ubiquitin-proteasome pathway, epigenetic events, metabolism, stem cell self-renewal, neuronal function, spermatogenesis, and the DNA damage response [reviewed in Lim & Kaldis (2013)]. Interestingly, there is also evidence that some of these non-canonical functions are carried out by cyclins independent of kinase activity [reviewed in Hydbring ].

Recent phylogenetic analyses have revealed that <span class="Gene">cyclins from a variety of different eukaryotes, including <span class="Species">yeasts, plants, and humans, can be divided into three major groups (Ma et al., 2013, Cao et al., 2014). Group I is composed of the “classic” cell cycle regulatory cyclins, while group III cyclins tend to function primarily in transcriptional regulation. Group II cyclins have varied functions that include, but are not limited to, roles in cell cycle regulation (Davidson et al., 2009, Jimenez et al., 2013) and development (Wu et al., 2004, Liu and Finley, 2010, Mikolcevic et al., 2012, Zi et al., 2015). However, it is important to note that (1) these phylogenetic analyses incorporated cyclins from a small number of eukaryotes, (2) our understanding of cyclin function is derived from only a handful of model species, and (3) there has been relatively little study of cyclins in filamentous fungi.

Filamentous fungi are, of course, of huge ecologi<span class="Species">cal importance and have an enormous impact on <span class="Species">human and animal health, agricultural crop production, and industrial production of important commodities (May and Adams, 1997, Bok et al., 2006, Pagano et al., 2006, Nalley et al., 2016). An excellent recent commentary addressed our limited understanding of filamentous fungal biology, the threat filamentous fungi pose to the food supply and to human and animal health, and the critical need to advance fungal biotechnology (Meyer ). There is, thus, every reason to investigate cyclin function and cell cycle regulation in filamentous fungi, to help us facilitate their growth when appropriate and inhibit their growth when needed. In addition, it will be of benefit to carry out phylogenetic and functional analyses of cyclins from additional organisms to further our understanding of eukaryotic cell cycle regulation and how these control systems evolved.

The budding <span class="Species">yeast <span class="Species">Saccharomyces cerevisiae and the fission yeast Schizosaccharomyces pombe are popular models for studying the cell cycle, and cyclins have been studied extensively in these organisms. While one might imagine that these yeasts can be used as a “reference system” for filamentous fungi, and they are often cited as such in the literature, this is not the case [reviewed in Meyer ].

In this work we have investi<span class="Chemical">gated the <span class="Gene">cyclin repertoires in the filamentous fungi analysed in a recent comparative genomics study (de Vries ). This group contains 17 species of the genus Aspergillus and 15 species from other, phylogenetically diverse, genera. [Note, Aspergillus zonatus which was included in the de Vries study does not belong in the genus Aspergillus and has been reclassified as Penicilliopsis zonata (Kocsube )]. The list includes 23 species from the order Eurotiales (including Aspergillus nidulans), seven species from Onygenales, Trichoderma reesi from Hypocreales, and Neurospora crassa from Sordariales. The group is well chosen because it allows one to determine if there are commonalities within the aspergilli and if these commonalities are shared with other, diverse, filamentous ascomycetes. All species in this group fall within the subdivision Pezizomycotina that includes most ascomycetes. This group does not contain members of the Saccharomycotina which includes yeasts such as S. cerevisiae, dimorphic fungi such as Candida albicans and filamentous fungi such as Ashbya gossypii that are closely related to, and probably derived from yeasts, nor does our study group contain any member of the Taphrinomycotina such as S. pombe (Wang ). Although basidiomycetes and oomycetes are of great interest, they are also outside of this study. Finally, to avoid repeated use of awkwardly long phrases, we will use the term “filamentous ascomycetes” to refer to members of the Pezizomycotina, which are by far the most numerous filamentous ascomycetes, realizing that there are some ascomycetes from outside of this subdivision that have a filamentous growth form.

Our phylogenetic analyses of <span class="Gene">cyclins in these 32 species of filamentous ascomycetes reveal that, as in other organisms that have been studied (Ma et al., 2013, Cao et al., 2014), cyclins fall into three groups. Strikingly, the complement of cyclins is remarkably conserved. For example, each species surveyed has three group I cyclins. These are particularly important because they typically play critical roles in the regulation of the cell cycle. The group I cyclins fall into three clades with each fungus having a single member of each clade. It follows that the cyclins in these phylogenetically diverse filamentous ascomycetes are likely to be functionally conserved. Lessons learnt from studying cyclin function in the model organism A. nidulans, are likely to apply across the genus Aspergillus and are likely to be informative with respect to many other filamentous ascomycetes, including pathogenic and industrially important species. Our analysis also reveals that the complement of cyclins found in these fungi is significantly different from that of yeast models. For example, S. cerevisiae has nine group I cyclins while S. pombe and C. albicans each have five. The yeasts are, thus, likely to be of limited value in understanding cyclin function in filamentous fungi.

Our analyses also highlight how under-studied the <span class="Gene">cyclins of aspan>ergilli and other Pezizomycotina are. Only a single group I <span class="Gene">cyclin has been analysed thoroughly, NimECyclin B of A. nidulans. NimECyclin B (O'Connell ) is a binding partner of the Cdk1 homolog NimX (Osmani ). NimECyclin B is required for both S-phase and G2/M and its destruction during mitosis is required for mitotic exit (De Souza et al., 2009, Nayak et al., 2010). Up to now, NimECyclin B is the only cyclin that has been shown to be essential in A. nidulans and the only cyclin shown to have clear, cell cycle-related functions. The other two A. nidulans group I cyclins are a named but uncharacterized cyclin, PucA, and a previously unstudied cyclin encoded by gene AN2137 [using the AspGD (http://aspgd.org/) and FungiDB (http://fungidb.org/fungidb/) gene designation].

Given the importance of group I <span class="Gene">cyclins and the dearth of data on the functions of these proteins in filamentous ascomycetes in general and aspan>ergilli in particular, we have <span class="Species">carried out a functional analysis of PucA and AN2137. Our genetic and live imaging data reveal that PucA is an essential G1/S cyclin that is required for the inactivation of the anaphase promoting complex/cyclosome (APC/C) complexed with its binding partner CdhA. Deletion of the pucA gene is lethal, but viability is restored if the cdhA gene is also deleted. We designate AN2137, the previously unnamed B-type cyclin gene, clbA. ClbA localizes to kinetochores from mid-G2 until just before chromosomal condensation. Deletion of clbA does not affect viability. Timely destruction of cyclins is critical for their function, and cyclins are normally targeted for destruction by APC/C binding to one or both of two conserved sequences called destruction boxes (d-boxes) or KEN boxes. Using a regulatable promoter system new to A. nidulans, we have found that expression of a version of ClbA in which the d-boxes have been removed results in a mitotic block, a high frequency of chromosomal non-disjunction, and consequent lethality. Using the same promoter system, we have found that expression of d-box-deleted NimECyclin B results in non-disjunction in addition to its previously reported phenotypes. Our functional characterization of A. nidulans group I cyclins provide new insights into cell cycle control mechanisms that are likely to be generally applicable to filamentous ascomycetes.

Materials and methods

Cyclin identification and phylogenetic analyses

To identify putative A. nidulans <span class="Gene">cyclins, we performed BLASTP searches using <span class="Species">human, S. cerevisiae, S. pombe, and C. albicans cyclin protein sequences as queries against the A. nidulans FGSC4 protein database (AspGD; www.aspgd.org). We also identified all A. nidulans proteins that contained a predicted N-terminal cyclin domain (IPR006671), a C-terminal cyclin domain (IPR004367), a cyclin Pho80-like domain (IPR013922), and/or a cyclin-like domain (IPR013763). Domain predictions were made using InterProScan software from the European Bioinformatics Institute (EBI) and could be found on AspGD's “Domains/Motifs” pages. We then conducted a second round of BLASTP searches, using the putative A. nidulans cyclins as queries against 31 filamentous ascomycetes (see Table S2 for a list of the fungi analysed) using NCBI's non-redundant protein database.

Protein sequences were aligned with MAFFT using the L-INS-i method and the Blosum62 matrix (Katoh ). Maximum likelihood analysis was done in RAxML v8.2.10 (Stamatakis 2014) using the rapid bootstrapping (-f a) algorithm with 1 000 bootstrap repli<span class="Species">cates under the PROT<span class="Species">CAT + Auto (automatically chooses the best protein substitution matrix with respect to the likelihood). RAxML analyses were performed using the free computational resource CIPRES Science Gateway (Miller ). Phylogenetic trees were visualized in iTOL (Letunic & Bork 2016).

Strains and media

A list of strains used in this study along with their genotypes is given in Table S1. YG (5 g/L <span class="Species">yeast extract, 20 g/L <span class="Chemical">D-glucose), supplemented with 400 μL/L of trace element solution (Vishniac & Santer 1957) was used as a liquid complete medium and YAG (YG plus 15 g/L agar) was used as a solid complete medium. As yeast extract does not provide enough riboflavin to fully supplement the riboB2 mutation or enough pyrimidines to supplement the pyrG89 mutation, the following were added to YAG if needed: riboflavin (2.5 μg/mL), uridine (2.442 mg/mL), and uracil (1 mg/mL). Liquid minimal medium (MM) for imaging consisted of 6 g/L NaNO3, 0.52 g/L KCl, 0.52 g/L MgSO4·7H2O, 1.52 g/L KH2PO4, 10 g/L D-glucose, 400 μL/L of trace element solution (Vishniac & Santer 1957) and any additional nutrients required to supplement mutations. For solid MM, the same components as liquid MM plus 15 g/L agar was used. Solid medium utilized for plating protoplasts after transformation required either 342.3 g/L sucrose or 44.7 g/L KCl. pH was adjusted to 6.0–6.5.

The al<span class="Species">cA promoter [al<span class="Species">cA(p)] is repressed by glucose and induced by several compounds, one of which is threonine. For our experiments, we repressed the alcA(p) with complete medium (e.g. YAG) and for induction we used MM in which the D-glucose was replaced with 9 g/L fructose plus 6.25 mM threonine. The nmtA promoter [nmtA(p)] is strongly repressed by thiamine, which is present in complete medium (e.g. YAG). Repression experiments with the nmtA promoter were, thus, carried out in minimal media with the concentrations of thiamine specified in the text.

Gene targeting and transformation

Linear DNA constructs for transformation of A. nidulans were generated by fusion PCR as previously described (Yang et al., 2004, Yu et al., 2004, Zarrin et al., 2005, Nayak et al., 2006, Szewczyk et al., 2006, Oakley et al., 2012). Generation of a fusion PCR product for deleting pu<span class="Species">cA or <span class="Chemical">clbA followed the procedure given in Szewczyk as modified by Oakley . Q5 Hot Start High Fidelity 2X Master Mix (New England Biolabs) or Phusion (New England Biolabs) DNA polymerases were used for initial amplifications and for fusion PCR.

To create C-terminal fluorescent fusion proteins, the transforming molecules consisted of ∼1 000-bp of the C-terminal coding sequence of the target protein (using primers to remove the stop codon) fused in frame to a 30-bp <span class="Chemical">glycine-alanine (<span class="Chemical">GA) linker (Yang ), which was, in turn, fused in frame with the fluorescent protein coding sequence. The fluorescent protein coding sequence was followed by a 3′ untranslated region from Aspergillus fumigatus and a selectable marker [the A. fumigatus pyrG gene (AfpyrG), riboB gene (AfriboB) or pyroA gene (AfpyroA)] and a ∼1 000-bp sequence downstream of the target gene. The GFP variant we used in our experiments was a plant-adapted GFP (Fernandez-Abalos ). The mCherry sequence was the original version described by Shaner , and the mRFP variant was as previously described (Campbell et al., 2002, Toews et al., 2004). The mCherry and tdTomato clones were a gift from Dr. Roger Tsien. pBAD24-sfGFPx1 was a gift from Sankar Adhya & Francisco Malagon (Addgene plasmid # 51558) and pBAD-mTagBFP2 was a gift from Vladislav Verkhusha (Addgene plasmid # 34632). We found that mTagBFP2 was superior to T-Sapphire because it was brighter and the emission spectrum did not overlap with GFP. This allowed it to be distinguished easily from GFP in strains expressing proteins tagged with both fluorescent proteins. A strain carrying An-Nup49-mCherry was provided by C. De Souza and S. Osmani (The Ohio State University, Columbus, OH). The A. fumigatus riboB gene used for the mTagBFP2 construct was amplified by PCR from genomic DNA provided by S. Earl Kang, Jr. and Dr. Michelle Momany (The University of Georgia, Athens, GA). Aspergillus terreus pyrG and riboB genes (AtpyrG and AtriboB) were amplified by PCR from genomic DNA supplied by Dr. Kenneth Bruno (Pacific Northwest National Laboratory) and further details on those selectable markers were recently published (Dohn ) and they are available from the Fungal Genetics Stock Center and from Addgene.

The dbΔ-<span class="Chemical">NimECyclin B construct was created as previously described (Nayak ). The two trun<span class="Species">cated forms of ClbA were generated as follows. A fragment of the clbA gene extending from nucleotide 342 to the end of the gene (nucleotide 2156) was amplified by PCR to generate db1Δ-ClbA. This fragment excludes the N-terminal 60 amino acids that contain the first putative destruction box (RAAFGDVSN). A second fragment of the clbA gene extending from nucleotide 603 to the end of the gene (nucleotide 2156) was amplified by PCR to generate db2Δ-ClbA. This fragment excludes the N-terminal 147 amino acids that contain both the first and second (RKTLNKRAT) putative destruction boxes. The primers used to amplify these fragments were designed with a tail at the 5′ end that would anneal to either the alcA or the nmtA promoter, including a start codon, and a tail at the 3′ end that would anneal to either the GA-GFP cassette or to the selectable marker. The complete transforming molecules of d-box-deleted ClbA or NimECyclin B under the control of either alcA(p) or nmtA(p) at the wA locus consisted of the following: ∼1 000-bp of 5′UTR from the wA gene, 400-bp of alcA(p) or 761-bp of nmtA(p), d-box-deleted ClbA or NimECyclin B, GA linker + GFP if desired, a 3′ untranslated region from A. fumigatus, a selectable marker (either AfpyrG or AfpyroA), and ∼1 000-bp of 3′UTR from the wA gene. A similar strategy was used to generate full length ClbA and NimECyclin B (with and without a C-terminal fusion to GA-GFP) under control of the alcA(p) or nmtA(p) at the wA locus.

PCR products were purified using the QIAQuick PCR Purifi<span class="Species">cation Kit (Qiagen) or the Monarch PCR & DNA Cleanup Kit (NEB) and analysed by <span class="Chemical">agarose gel electrophoresis. Protoplast formation, purification, and transformation were carried out as described in Szewczyk and Oakley . Recipient strains carried nkuAΔ to minimize non-homologous end-joining (Nayak ).

Verification of transformants

Genomic DNA was prepared from transformants using a miniprep procedure (Edgerton-Mor<span class="Chemical">gan & Oakley 2012). Conidia were collected from the surface of a colony using a toothpick or loop and suspan>ended in 50 μL TE buffer (10 mM <span class="Chemical">Tris-HCl, pH 7.5, and 1 mM EDTA) in a microcentrifuge tube. Approximately 50 μL of acid-washed, 425–600 μm glass beads (Sigma) were added and the tube vortexed for 2 min. 2 μL of this solution was immediately removed and added to 18 μL of TE. 2 μL of this 1:10 solution was used for diagnostic PCR using OneTaq Hot Start Quick-Load 2X Master Mix with Standard Buffer (New England Biolabs, Inc.) according to the manufacturer's instructions. Correct gene integration in each transformant was verified by at least two diagnostic PCR amplifications using different primer pairs: (i) primers outside the target region and (ii) one outside primer and one primer inside the selectable marker.

Microscopy

For imaging, conidia were cultured in 400 μL of liquid MM in eight-chambered cover glasses (Lab-Tek; Thermo Fisher Scientific) with necessary supplements to complement nutritional markers. In some experiments, germlings were fixed in the wells by removing the liquid medium and adding 400 μL of fixative solution [8 % formaldehyde in 50 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES), pH 6.7; 25 mM EGTA, pH 7.0; 5 mM MgSO4; and 5 % DMSO, pre-warmed to the culture temperature] for 20 min at 37 °C or 30 min at 30 °C. The fixative solution was then removed and replaced with 400 μL <span class="Species">Calcofluor White staining solution. For <span class="Chemical">DAPI experiments, 5 × 107 conidia were grown in 10 mL liquid media plus 0.1 % agar (to minimize clumping of germlings) at 30 °C at 140 rpm for 12 h. Samples (900 μL) were added to 100 μL of 10 % glutaraldehyde (Electron Microscopy Sciences, Fort Washington, PA), which was previously equilibrated to the culture temperature. They were fixed for 10 min at the culture temperature, spun down for 3 min at 15,500 × g, and then washed 2 × 10 min in double distilled water at room temperature. Samples were resuspended in 0.015 μg/mL DAPI solution.

Two systems, both with environmental chambers to maintain stable temperatures, were used for imaging. The first system was an Olympus IX71 inverted microscope equipped with a <span class="Chemical">mercury illumination source along with Prior shutters, filter wheels, Z-axis drives, and an OR<span class="Species">CA ERAG camera (Hamamatsu Photonics). Images were collected with an Olympus 60X/1.42 numerical aperture Plan Apo objective. Filter sets used were a GFP/DsRed2X2M-B dual-band Sedat filter set (Semrock) with a 459–481 nm bandpass excitation filter for GFP, a 546–566 nm excitation filter for mCherry and mRFP, a dual reflection band dichroic (457–480 nm and 542–565 nm reflection bands, 500–529 and 584–679 nm transmission bands), a 499–529 nm emission filter for GFP and a 580–654 nm emission filter for mCherry/mRFP. The second system was an UltraView VoX spinning disk confocal system (PerkinElmer) mounted on an Olympus IX71 inverted microscope. This system was equipped with a software-controlled piezoelectric stage for rapid Z-axis movement. Images were collected using a 60X/1.42 numerical aperture Plan Apo objective (some images taken with a 1.6× Optovar) and an ORCA ERAG camera (Hamamatsu Photonics). Solid state 405-, 488-, and 561-nm lasers were used for excitation and fluorochrome-specific emission filters were used to prevent emission bleed through between fluorochromes. Both systems were controlled by Volocity software (PerkinElmer) running on Power Mac computers (Apple). Magnifications were calibrated with a stage micrometer. Images were exported directly from Volocity after adjustment of minimum and maximum intensity levels (black and white levels) for each channel. Figures were prepared from exported images using Pages (Apple) with no further adjustments.

Results

Phylogenetic analyses of cyclins in aspergilli and other filamentous ascomycetes

Identification of the cyclin repertoire of A. nidulans

Phylogenetic analyses of <span class="Gene">cyclins have recently been <span class="Species">carried out on a variety of different organisms including the model yeasts S. cerevisiae and S. pombe (Ma et al., 2013, Cao et al., 2014). However, these evolutionary studies focused on cyclins with an N-terminal cyclin domain, which omits several fungal cyclins. These fungal cyclins do not have an N-terminal cyclin domain but instead have either a cyclin-like or cyclin pho80-like domain (Supplementary Spreadsheet, Tab 1 titled “Cyclins in Anid & Yeasts”). To ensure that we identified all putative cyclins in A. nidulans, we utilized two methods: (1) BLASTP searches of the Aspergillus genome database (AspGD) using the published amino acid sequences of human, S. cerevisiae, S. pombe, and C. albicans cyclins and (2) using the InterProScan protein motif search tool to identify all A. nidulans proteins that contain an N-terminal cyclin domain (IPR006671), a C-terminal cyclin domain (IPR004367), a cyclin Pho80-like domain (IPR013922), and/or a cyclin-like domain (IPR013763). The cyclin-like domain is also present in the transcription factors TFIIB and TFIIB-related Brf1 which are conserved in fungi.

We determined that the A. nidulans genome encodes fifteen putative <span class="Gene">cyclins (Fig. 1A) of which only six had been previously characterized. Two additional proteins that contain a <span class="Gene">cyclin-like domain were identified, but they are strong homologs of human TFIIB (AN4928, E-value of 9e-54) and TFIIB-related Brf1 (AN3116, E-value of 4e-68) and are unlikely to be cyclins. Based on both sequence similarity and our phylogenetic analyses, cyclins in fungi fall into three major groups, consistent with previous cyclin family analyses (Ma et al., 2013, Cao et al., 2014). These results are summarized in Fig. 1A, Table S2, and in the Supplementary Spreadsheet (particularly Tab 1, Table 1 of the Supplementary Spreadsheet).

Fig. 1

Cyclin complements of model yeasts and Table summarizing the number of cyclins in group I, group II, and group III of three model yeasts (S. cerevisiae, S. pombe, and C. albicans) and in the filamentous fungus A. nidulans (bold, red font). Cyclin subfamilies within each group are also shown. The total number of cyclins are displayed at the far right. *S. pombe has two additional proteins with cyclin-like domains that do not have sequence similarity to other fungal cyclins. See the Supplementary Spreadsheet (Tab 1 and 2) for additional details. B. Midpoint-rooted maximum-likelihood phylogenetic tree of group I cyclins in A. nidulans (An), S. cerevisiae (Sc), S. pombe (Sp), and C. albicans (Ca). Fungal group I cyclins cluster into two subfamilies, which are indicated in blue font. The abbreviated name of the fungal organism (e.g. An) is followed by the protein name. Cyclins that have been published as essential, or that we have determined are essential in this work, have green dots to the right of their name. Branch support values (70–100 % bootstrap support) are displayed on the branches.

<span class="Gene">Cyclin complements of model <span class="Species">yeasts and Table summarizing the number of cyclins in group I, group II, and group III of three model yeasts (S. cerevisiae, S. pombe, and C. albicans) and in the filamentous fungus A. nidulans (bold, red font). Cyclin subfamilies within each group are also shown. The total number of cyclins are displayed at the far right. *S. pombe has two additional proteins with cyclin-like domains that do not have sequence similarity to other fungal cyclins. See the Supplementary Spreadsheet (Tab 1 and 2) for additional details. B. Midpoint-rooted maximum-likelihood phylogenetic tree of group I cyclins in A. nidulans (An), S. cerevisiae (Sc), S. pombe (Sp), and C. albicans (Ca). Fungal group I cyclins cluster into two subfamilies, which are indicated in blue font. The abbreviated name of the fungal organism (e.g. An) is followed by the protein name. Cyclins that have been published as essential, or that we have determined are essential in this work, have green dots to the right of their name. Branch support values (70–100 % bootstrap support) are displayed on the branches.

The complement of cyclins is highly conserved in filamentous ascomycetes

We conducted BLASTP searches of the NCBI protein database for <span class="Gene">cyclins in the 31 filamentous ascomycetes analysed in de Vries using the amino acid sequences of A. nidulans <span class="Gene">cyclins as a query. The most striking (and perhaps surprising) result of these analyses was the degree of conservation among cyclins, not just in the Aspergillus species, but also in the other species which are phylogenetically diverse. In most of the fungi analysed, we identified a single, strong homolog of each of the fifteen cyclins present in A. nidulans. A few absences, which may, in part, be due to annotation errors and incidents of gene duplication were also observed (discussed below). The results of the BLASTP searches are summarized in Table S2.

Consistent with previous work, our phylogenetic analyses of all putative <span class="Gene">cyclins from the 32 species, including A. nidulans, indicate that the cyclins fall into three major groups (Supplementary Spreadsheet, Tabs 2–6). The cyclins in group I are the most highly conserved of the three groups based on E-value, query coverage, and the lack of any gene duplications or absences. All species we analysed have three identifiable group I cyclins. They fall into two groups that we designate Cln-like and B-type. The B-type cyclins further fall into two well-separated clades. The group I cyclins from filamentous ascomycetes, thus, fall into three clades with each fungus having a single member of each clade. The A. nidulans members of the three clades are PucA, NimECyclin B, and ClbA (Supplementary Spreadsheet, Tab 4 titled “FilFungi Group I Cyclins”).

The A. nidulans group II <span class="Gene">cyclins do not share strong sequence homology with either group I or group III <span class="Gene">cyclins, and they only appeared as “hits” when we conducted BLASTP searches using the model yeasts group II cyclins as queries (Supplementary Spreadsheet, Tab 2 titled “Anid vs Model Yeasts Cyclins”). The majority of the species in our study have seven group II cyclins that are strong homologs of the seven A. nidulans group II cyclins (Supplementary Spreadsheet, Tab 5 titled “FilFungi Group II Cyclins”). We also observed two subfamilies (Pcl1,2 and Pho80 subfamilies) similar to those observed in S. cerevisiae. However, some evidence of gene duplications was observed with group II cyclins. For example, we found that both Penicillium chrysogenus and P. rubens have three PclA-like proteins and three Pho80-like proteins while P. digitatum has only one copy of each. There are three species with only six group II cyclins, all of which were lacking one member of the Pho80 subfamily. However, the missing cyclins could be due to missing or inaccurate annotations.

The group III <span class="Gene">cyclins in A. nidulans also do not share strong sequence similarities with either group I or group II <span class="Gene">cyclins (Supplementary Spreadsheet, Tab 2 titled “Anid vs Model Yeasts Cyclins”). They are easily detected in BLASTP searches, however, using group III cyclins from yeasts as queries. Group III cyclins are well-conserved in the species in our study. The majority of the species analysed have only five group III cyclins and they are strong homologs of the A. nidulans group III cyclins (Supplementary Spreadsheet, Tab 6 titled “FilFungi Group III Cyclins”). Only four species have four instead of five cyclins, but in at least three of the four cases, another sequenced strain of the same species has all five cyclins so the apparent absence may simply be due to sequencing or annotation error. We did not observe any gene duplications in this group.

Model yeasts and filamentous ascomycetes contain different complements of cyclins

The <span class="Species">yeasts S. cerevisiae and S. pombe, have been instrumental in revealing the general principles of cell cycle regulation. The pathogenic fungus C. albi<span class="Species">cans has also been studied extensively because of its significance as a pathogen. We will refer to C. albicans as a yeast, although it is dimorphic and has both yeast and filamentous phases. Since these are well-studied ascomycetes, one might expect that they would be good models for understanding cyclin function in aspergilli and other filamentous ascomycetes. These model yeasts are quite phylogenetically distant, however, from the aspergilli and other members of the Pezizomycotina which make up the great majority of filamentous ascomycetes. The sole filamentous ascomycete in which the functions of group I cyclins have been studied at some depth is Ashbya gossypii. This species is a member of the Saccharomycotina, however, and is thus closely related to S. cerevisiae and related yeasts and quite distant, phylogenetically, from the great majority of filamentous ascomycetes (Wang ).

Our analyses reveal that the complement of <span class="Gene">cyclins in the fungi in our study is signifi<span class="Species">cantly different from those of S. cerevisiae, S. pombe and C. albicans. For simplicity, we will present our comparison of the cyclins of A. nidulans to those of these model yeasts, but, as mentioned, A. nidulans fairly reflects the complement of cyclins in the other fungi in our study. The sequence homology and phylogenetic relationships between A. nidulans and the three model yeasts is shown in the Supplementary Spreadsheet (Tab 2 titled “Anid vs Model Yeasts Cyclins”).

With respect to group I <span class="Gene">cyclins, the most obvious difference between A. nidulans and the model <span class="Species">yeasts is that whereas A. nidulans has only three group one cyclins, S. cerevisiae, S. pombe and C. albicans have nine, five and five group I cyclins respectively (Fig. 1B). S. cerevisiae and S. pombe have six and four B-type cyclins respectively whereas A. nidulans and C. albicans have only two. S. cerevisiae and C. albicans each have three cln-like cyclins, while S. pombe and A. nidulans have only a single cln-like cyclin. The multiplicity of cyclins in S. cerevisiae may reflect an ancient whole genome duplication (Wolfe & Shields 1997). This may have resulted in functional redundancy accounting for the fact that none of the cyclins are individually essential in this organism. S. pombe and C. albicans are not thought to have undergone whole genome duplications, and the extra cyclins likely have arisen through individual gene duplications. A. gossypii has five group I cyclins, including two cln-like cyclins and three B-type cyclins. One of the two cln-type cyclins is essential and two of the B-type cyclins are essential (Hungerbuehler ).

The number of group II <span class="Gene">cyclins varies signifi<span class="Species">cantly among model yeasts and differs between the model yeasts and A. nidulans. Group II cyclins in the three model yeasts are called Pho85 cyclins (PCLs) after their interacting CDK partner, Pho85. PCLs were first identified and characterized in S. cerevisiae. S. cerevisiae has ten PCLs and they separate into two sub-families: the Pcl1,2 subfamily (Pcl1, Pcl2, Pcl5, Pcl9, and Clg1) and the Pho80 subfamily (Pho80, Pcl6, Pcl7, Pcl8, and Pcl10) (Measday ). S. pombe in contrast has only three group II cyclins. A. nidulans has seven cyclins, PclA, AN9500, ClgA, AN4984, AN3755, PclB, and Pho80, that cluster with group II cyclins (Fig. 1A, Supplementary Spreadsheet Tab 2) as does C. albicans. Although the number is the same, the complement of cyclins in C. albicans and A. nidulans is significantly different. For example, C. albicans has an uncharacterized cyclin (C3_02720W) that has no obvious homolog in A. nidulans. In addition, A. nidulans has two clg1-like genes (clgA and an uncharacterized gene AN4984), while C. albicans has only one clg1 gene. Based on the sequence homology and phylogenetic analyses of these four fungi, we observe six major clades of PCL cyclins which we designate as the following, based on the S. cerevisiae PCL naming scheme: Pcl1,2,9-like, Pcl5-like, Clg1-like, Pho80-like, Pcl6,7-like, and Pcl8,10-like. The sequence homology and phylogenetic relationship between A. nidulans and model yeasts group II cyclins is shown in the Supplementary Spreadsheet (Tab 2 titled “Anid vs Model Yeasts Cyclins”).

A. nidulans has five group III <span class="Gene">cyclins: AN2172 (<span class="Gene">cyclin C-like), AN2211 (cyclin H-like), AN7719 (cyclin L-like), and PchA and AN10640 (cyclin T/K-like) (Supplementary Spreadsheet, Tabs 1–3). The cyclin C-like and cyclin H-like families are well-conserved with the three model yeasts. However, a cyclin L-like homolog is present only in A. nidulans and S. pombe.

Functional analysis of Group I cyclins in A. nidulans

While it will ultimately be important to characterize all the groups of <span class="Gene">cyclins in filamentous ascomycetes, the group I <span class="Gene">cyclins are particularly important because they are key cell cycle regulators. As mentioned above, in the aspergilli and the other fungi in our study they fall into three clades with each fungus having one member of each clade (Supplementary Spreadsheet, Tab 4 titled “FilFungi Group I Cyclins”). This suggests that the group I cyclins are functionally conserved, and that functional information gleaned from one fungus is likely to carry over to many filamentous ascomycetes. Perhaps surprisingly, given the importance of filamentous ascomycetes and the importance of group I cyclins, they have been studied relatively little. The functions of only a single group I cyclin from a member of the aspergilli, NimECyclin B from A. nidulans, has been analysed extensively. A few other group I cyclins from the organisms in our study have been named, but functional analyses have been not been carried out in depth.

<span class="Chemical">NimECyclin B is an essential B-type <span class="Gene">cyclin required for both S-phase and mitosis (Bergen et al., 1984, O'Connell et al., 1992, Nayak et al., 2010). Members of the other two clades of group I cyclins have hardly been studied at all in filamentous ascomycetes. We have, consequently, chosen to study the functions of the other two group I cyclins of A. nidulans, PucA and the unnamed cyclin encoded by AN2137. PucA was named by De Souza on the basis of similarity to S. pombe Puc1 (Pombe unidentified cyclin 1) (E-value of 5e-37), a cyclin that associates with Cdc2Cdk1 to regulate the length of G1 (Martin-Castellanos ). PucA was found to co-purify with a dual localization-affinity purification (DLAP) tagged version of NimXCdk1. AN2137 has not been characterized at all, nor has it been given a standard A. nidulans gene designation. We now designate it clbA.

PucA is essential for viability

In order to determine if Pu<span class="Species">cA is essential, we attempted to delete the pu<span class="Species">cA gene (1 312-bp) by transforming the strain LO1516 (genotype listed in Table S1) with a fragment carrying a selectable marker, (AfpyrG), flanked by ∼1 000-bp pucA flanking sequences. When essential genes are deleted in A. nidulans, nuclei carrying the null allele are often maintained in heterokaryons spontaneously formed during transformation. Such heterokaryons carry two types of nuclei, parental nuclei and transformed nuclei. Parental nuclei carry a functional copy of the gene under study but do not carry the selectable marker. In transformed nuclei the gene under study has been deleted by replacement with a selectable marker that supports growth on selective media. Such heterokaryons are able to grow on selective media. Importantly, A. nidulans conidia are uninculeate, so each conidium carries a parental nucleus or deletant nucleus but not both. The conidia from a heterokaryon carrying a deletion of an essential gene will, thus, not be viable on selective media. By streaking out conidia from the primary transformants on nutritionally selective media, we can determine if the deleted gene is essential by the presence or absence of growth and colony formation (Osmani et al., 1988, Martin et al., 1997, Osmani et al., 2006). In the case of PucA, multinucleate mycelia transferred to selective media from the primary transformants grew, but conidia from transformants did not grow to form colonies (Fig. 2). This indicates that pucA is essential. Diagnostic PCR on DNA prepared from conidia revealed bands diagnostic for both pucAΔ and pucA+ (data not shown), confirming that the colonies are, indeed, heterokaryons. The fact that the deletion of pucA does not block growth in heterokaryons reveals that it is recessive.

Fig. 2

PucA is an essential cyclin in We used the heterokaryon rescue technique to determine if PucA is essential. A parental, pyrG89 auxotrophic strain (LO1516) was transformed with a fragment designed to delete pucA (pucAΔ) by replacing it with AfpyrG. If PucA is essential, conidia carrying pucAΔ will not support growth. However, if a heterokaryon is generated during transformation that carries both pucAΔ and parental nuclei, hyphae will grow on selective medium (medium lacking uridine and uracil) because parental nuclei provide PucA and transformed nuclei complement pyrG89. The conidia produced by the heterokaryon are uninucleate, so they will have either parental nuclei or pucAΔ nuclei, and neither will grow on the selective medium. Squares of agar from the parental strain (A, C) and a pucAΔ transformant heterokaryon (B, D) have been placed on the selective medium (left plate) and the nonselective medium (right plate). On the selective medium, hyphae do not grow from the parental square (A) but do grow from the transformant, creating a colony with rough edges as is typical for a heterokaryon (B). Conidia have been streaked below each colony. As expected, parental conidia do not grow (A). Crucially, conidia from the transformant colony also do not grow (B), revealing that the transformant is a heterokaryon carrying the lethal deletion pucAΔ. The fact that the heterokaryon hyphae grow reveals that pucAΔ is recessive. As expected, both hyphae and conidia from the parental strain and the heterokaryon grow on the nonselective medium (C, D).

Pu<span class="Species">cA is an essential <span class="Gene">cyclin in We used the heterokaryon rescue technique to determine if PucA is essential. A parental, pyrG89 auxotrophic strain (LO1516) was transformed with a fragment designed to delete pucA (pucAΔ) by replacing it with AfpyrG. If PucA is essential, conidia carrying pucAΔ will not support growth. However, if a heterokaryon is generated during transformation that carries both pucAΔ and parental nuclei, hyphae will grow on selective medium (medium lacking uridine and uracil) because parental nuclei provide PucA and transformed nuclei complement pyrG89. The conidia produced by the heterokaryon are uninucleate, so they will have either parental nuclei or pucAΔ nuclei, and neither will grow on the selective medium. Squares of agar from the parental strain (A, C) and a pucAΔ transformant heterokaryon (B, D) have been placed on the selective medium (left plate) and the nonselective medium (right plate). On the selective medium, hyphae do not grow from the parental square (A) but do grow from the transformant, creating a colony with rough edges as is typical for a heterokaryon (B). Conidia have been streaked below each colony. As expected, parental conidia do not grow (A). Crucially, conidia from the transformant colony also do not grow (B), revealing that the transformant is a heterokaryon carrying the lethal deletion pucAΔ. The fact that the heterokaryon hyphae grow reveals that pucAΔ is recessive. As expected, both hyphae and conidia from the parental strain and the heterokaryon grow on the nonselective medium (C, D).

PucA is a G1/S cyclin

Pu<span class="Species">cA has homology to G1/S <span class="Gene">cyclins in other organisms, and we hypothesized that the essential function of PucA is to regulate the G1/S transition. In G1, the APC/C bound to its activator Cdh1 functions to ubiquitinate S-phase cyclins and other substrates and, thus, targets them for destruction. In order for nuclei to enter S-phase the APC/C-Cdh1 complex must be inactivated by cyclin/CDK complexes (Lukas et al., 1999, Sorensen et al., 2001, Fukushima et al., 2013). CdhA, the A. nidulans Cdh1 homolog, has been shown to function similarly to Cdh1 in other organisms by preventing NimECyclin B accumulation in G1 (Edgerton-Morgan & Oakley 2012). Previously, we found that the deletion of cdhA was not lethal and had little effect on growth, but the length of G1 was shortened and NimECyclin B-GFP fluorescence started to become visible shortly after mitosis (Edgerton-Morgan & Oakley 2012). If PucA is a G1/S cyclin involved in the inactivation of CdhA in A. nidulans, then the lethality of pucAΔ is predicted to be due to failure of inactivation of the APC/C-CdhA complex and consequent blockage of the cell cycle in G1. If this were the case, deletion of CdhA should allow progression through the cell cycle and enable growth of strains carrying pucAΔ. We tested this possibility by attempting to delete pucA in a cdhAΔ strain (LO2019). We obtained numerous transformants that were viable and could be streaked to single colony (Fig. 5A). Diagnostic PCR revealed that pucA was deleted in the transformants (data not shown). Deletion of pucA is, thus, not lethal in a cdhAΔ background, and this strongly supports the hypothesis that the essential function of pucA is to inactivate CdhA at the end of G1 allowing the accumulation of NimECyclin B and progression into S phase. Note, however, that the pucAΔ, cdhAΔ strains grew more slowly than WT or cdhAΔ strains and sporulated poorly. This result strongly suggests that pucA has non-essential functions, in addition to its essential function in inactivating CdhA, that are important for growth and sporulation.

Fig. 5

Deletion of (A) Strains FGSC4/WT, LO2019, and LO8743 were stabbed on complete medium and incubated at 37 °C for two days. The pucAΔ, cdhAΔ strain [LO8743 (cdhA::AfpyrG, pucA::AfpyroA)] displays reduced growth and sporulation compared to cdhAΔ [LO2019 (cdhA::AfpyrG)] and WT (FGSC4) strains at all temperatures tested (other temperatures not shown). (B) Cell cycle duration was calculated using time-lapse images collected at 10-min intervals of a control (LO1806), cdhAΔ (LO2019), and pucAΔ, cdhAΔ (LO8743) strains, all of which contained histone H1-mRFP. Cell cycle duration was calculated as the time from the end of one mitosis to the end of the next mitosis. Differences in cell cycle duration were statistically significant (* = p value of 0.005, ** = p value of 8e-08, *** = p value of 6e-17). Values are means and error bars indicate standard deviation. n = number of nuclei.

Deletion of Bright field images are from a single fo<span class="Species">cal plane (A, E) and images B–D and F–H are maximum intensity projections from Z-series stacks. Images are of a representative control (A–D) and pu<span class="Species">cAΔ (E–H) germling fixed after 8 h of growth at 37 °C. Three rounds of nuclear division have occurred in the control strain resulting in 8 nuclei. A single nucleus is present in the pucAΔ germling, and NimECyclin B-GFP fluorescence is not apparent. (I) Very few pucAΔ germlings have undergone nuclear division at the 4-, 6-, and 8-h time points, while the control strain has undergone 2–3 rounds of nuclear division by the 8-h time point. (J) pucAΔ germlings do not accumulate NimECyclin B. Over 100 fixed germlings were imaged and scored at each time point for each of two experiments for both control (strain LO10795) and pucAΔ (2 transformants). Error bars indicate mean ± standard deviation of two experiments.

Control pu<span class="Species">cA+ (<span class="CellLine">LO9537) (A–D) and pucAΔ (E–H) conidia were incubated at 30 °C for 16 h and imaged. Bright field images (A and E) are single focal plane images, and other images are maximum intensity projections from Z-series stacks. An-Nup49-mCherry allows visualization of the nuclear envelope. The control strain has gone through several cell cycles resulting in many nuclei. The nuclei are normal sized. In the pucAΔ germling, however, there is only one large nucleus, and it is extremely stretched as shown by An-Nup49-mCherry fluorescence. There is also faint mCherry and T-Sapphire fluorescence in the vacuole in the conidial swelling of the germling (blue arrowhead in F–H). The histone H1 fluorescence in the pucAΔ strain is very faint. Note there are four ungerminated parental conidia below the conidial swelling that show bright histone H1-T-Sapphire fluorescence. (I–L) DAPI staining of control and pucAΔ germlings incubated for 14 h at 30 °C and fixed. I and K are single focal plane images, and J and L are maximum intensity projections of Z-series stacks. Nuclear fluorescence is barely visible in L even though software was used to increase the brightness level 3.5X in L relative to J. The arrowhead in L points to a possible nucleus. The punctate DAPI staining is mitochondrial DNA. (M–N) Control (LO9537) and pucAΔ (pucAΔ heterokaryon made in parent strain LO9481) conidia containing histone H1-T-Sapphire and An-Nup49-mCherry were grown for 12, 14, and 16 h at 30 °C and then live imaged for one hour at 30 °C. The percentage of germlings with stretched nuclei was scored (M); error bars indicate mean ± standard deviation of three separate experiments. n = the number of germlings. Similarly, germlings were scored for the presence of obvious nuclear DAPI staining (N); error bars indicate mean ± standard deviation of two experiments (n = 100 germlings per time point). All germlings scored displayed mitochondrial DAPI staining, indicating DAPI staining was successful.

Deletion of (A) Strains FGSC4/WT, <span class="CellLine">LO2019, and LO8743 were stabbed on complete medium and incubated at 37 °C for two days. The pu<span class="Species">cAΔ, cdhAΔ strain [LO8743 (cdhA::AfpyrG, pucA::AfpyroA)] displays reduced growth and sporulation compared to cdhAΔ [LO2019 (cdhA::AfpyrG)] and WT (FGSC4) strains at all temperatures tested (other temperatures not shown). (B) Cell cycle duration was calculated using time-lapse images collected at 10-min intervals of a control (LO1806), cdhAΔ (LO2019), and pucAΔ, cdhAΔ (LO8743) strains, all of which contained histone H1-mRFP. Cell cycle duration was calculated as the time from the end of one mitosis to the end of the next mitosis. Differences in cell cycle duration were statistically significant (* = p value of 0.005, ** = p value of 8e-08, *** = p value of 6e-17). Values are means and error bars indicate standard deviation. n = number of nuclei.

We used heterokaryon rescue to determine the phenotype of pu<span class="Species">cAΔ in a <span class="Chemical">cdhA+ strain. As mentioned, although pucAΔ is lethal in a cdhA+ background, nuclei carrying pucAΔ can be maintained in heterokaryons along with nuclei carrying pucA+ and a selectable nutritional marker. The phenotype of pucAΔ can then be examined in germinating conidia from the heterokaryon. In the past, we have often used pyrG89 as a selectable mutation for heterokaryon rescue. Untransformed spores carrying pyrG89 barely germinate on selective media, allowing easy identification of germlings carrying the deletion of interest. We have found, however, that riboB2 is an even better selectable marker for heterokaryon rescue. Conidia carrying riboB2 swell, but do not extend a germ tube and there is no nuclear division, whereas nuclear division occasionally occurs in pyrG89 conidia. It is, thus, easy to distinguish untransformed riboB2 conidia among the conidia produced by a heterokaryon. We deleted pucA by replacing it with AtriboB in strain LO10761, which carries NimECyclin B-GFP as well as An-Nup49-mCherry. Nup49 is a nucleoporin and An-Nup49-mCherry allows visualization of the nuclear envelope. NimECyclin B accumulates in S phase and if pucAΔ prevents inactivation of CdhA, nuclei should be blocked at G1/S and NimECyclin B should not accumulate.

We found that An-Nup49-mCherry and <span class="Chemical">NimECyclin B fluorescence was preserved by a fixation procedure we routinely use and was, in fact, stable for days after fixation. This allowed us to score large numbers of germlings from particular time points easily. Although the fluorescence was stable, we scored germlings within 24 h of fixation. We separately incubated conidia at 37 °C from two heterokaryons as well as a control strain (LO10795), which was constructed by inserting AtriboB at the yA locus. We fixed the germlings at 4, 6, and 8 h after inoculation. We collected through-focus Z-series image stacks of random fields and scored germinated conidia for number of nuclei and presence of <span class="Chemical">NimECyclin B. At the 4-h time point, 35.5 ± 0.8 % (2 experiments) of spores had germinated in the control strain. We scored pucAΔ germlings at the same time point, realizing that some pucAΔ conidia would not have germinated at that time. As Fig. 3 shows, nuclear division was almost completely blocked in the pucAΔ germlings. NimECyclin B accumulation was also almost completely blocked. No NimECyclin B nuclei were seen at the 4-h time point and less than 1 % of pucAΔ nuclei contained NimECyclin B at the 6-h and 8-h time points. In the control strain 40.6 ± 3.3 % of nuclei were NimECyclin B positive at the 4-h time point, 64.4 ± 0.4 % were NimECyclin B positive at the 6-h time point, and 64.7 ± 4.5 % were NimECyclin B positive at the 8-h time point (Fig. 3).

Fig. 3

Deletion of Bright field images are from a single focal plane (A, E) and images B–D and F–H are maximum intensity projections from Z-series stacks. Images are of a representative control (A–D) and pucAΔ (E–H) germling fixed after 8 h of growth at 37 °C. Three rounds of nuclear division have occurred in the control strain resulting in 8 nuclei. A single nucleus is present in the pucAΔ germling, and NimECyclin B-GFP fluorescence is not apparent. (I) Very few pucAΔ germlings have undergone nuclear division at the 4-, 6-, and 8-h time points, while the control strain has undergone 2–3 rounds of nuclear division by the 8-h time point. (J) pucAΔ germlings do not accumulate NimECyclin B. Over 100 fixed germlings were imaged and scored at each time point for each of two experiments for both control (strain LO10795) and pucAΔ (2 transformants). Error bars indicate mean ± standard deviation of two experiments.

Interestingly, the <span class="Chemical">NimECyclin B positive nuclei in the control strain at the 4-h time point tended to be in germlings with a single nucleus whereas nuclei in binucleate germlings tended to be <span class="Chemical">NimECyclin B negative. Our interpretation is that the single nuclei were in the first S or G2 phase after germination and that the binucleate nuclei tended to be in G1 following mitosis. The percentages of NimECyclin B positive nuclei at the 6-h and 8-h time points are consistent with our previous observations (Nayak et al., 2010, Edgerton-Morgan and Oakley, 2012).

These data, in aggre<span class="Chemical">gate, reveal that Pu<span class="Species">cA is an essential cyclin that is required for the G1/S transition and that it functions by inactivating CdhA, and thus the APC/C, and by doing so allows the accumulation of NimECyclin B. Furthermore, this indicates that there are no additional cyclins in A. nidulans that are sufficient to inactivate CdhA at the G1/S boundary and that there are no redundant pathways for CdhA inactivation as has been observed in humans and in S. cerevisiae [reviewed in Sivakumar & Gorbsky (2015)].

Deletion of PucA allows nuclear growth but inhibits the DNA replication cycle resulting in nuclei with very diffuse chromatin

To observe nuclear behaviour in a pu<span class="Species">cAΔ strain, we deleted pu<span class="Species">cA in several cdhA+ strains carrying different fluorescent proteins, in each case replacing it with AfriboB. Since pucA is essential in cdhA+ strains, we used the heterokaryon rescue technique in all cases and observed the phenotype of pucAΔ in germlings that grew from conidia produced by the heterokaryon. We first deleted pucA in LO9481, a strain expressing histone H1-T-Sapphire and An-Nup49-mCherry. A control pucA+, riboB+ strain (LO9537) was constructed by inserting AfriboB at the wA locus in the parent strain LO9481. Conidia from the control strain and from the pucAΔ heterokaryon were incubated separately at 30 °C for 12 h at which time both control conidia and pucAΔ conidia had germinated. Z-series stacks of random fields were then captured over the next hour. Separate, identically grown samples were also imaged at the 14–15 h and 16–17 h periods. Representative control and pucAΔ images at 16–17 h are shown in Fig. 4. Surprisingly, pucAΔ germlings at these time points contained either faint or no obvious histone H1-T-Sapphire fluorescence (Fig. 4F). We consistently observed this phenotype in pucAΔ germlings from several additional strains with different fluorescent proteins (mRFP and GFP) fused to histone H1 (data not shown). The loss of histone fluorescence is, thus, not due to the T-Sapphire fluorescent protein. We DAPI stained pucAΔ and control germlings to determine if there was a loss of DNA fluorescence that parallelled the loss of histone fluorescence. In DAPI stained control germlings, nuclear and mitochondrial genomes are visible and easily distinguished from each other. Mitochondrial DNA is visible as faint punctate fluorescence and nuclear DNA is much brighter and occupies a larger volume, and the nucleus has a characteristic shape. DAPI staining of pucAΔ germlings (deletion made in parent strain LO9560) revealed a loss of nuclear DNA fluorescence over time, with over 50 % of pucAΔ germlings lacking obvious nuclear DAPI staining at the 16-h time point (incubation at 30 °C, 2 experiments, n = 50 germlings per experiment) (Fig. 4I–L, N). Mitochondrial DNA staining was still apparent in all the pucAΔ germlings (Fig. 4L), assuring us that the lack of identifiable nuclear DNA staining was not due to failed DAPI staining. Additionally, we DAPI stained a pucA+ control strain (LO9732) and pucAΔ germlings side-by-side with identical conditions and timing and imaged them on the same day, and all control germlings had visible nuclear and mitochondrial fluorescence. Thus, loss of identifiable nuclear DNA fluorescence over time parallels the loss of histone H1 fluorescence. We should note, however, that as nuclear DNA becomes stretched and fragmented in pucAΔ germlings, it may be present, but too faint to image or it might be indistinguishable from mitochondrial DNA.

Fig. 4

Control pucA+ (LO9537) (A–D) and pucAΔ (E–H) conidia were incubated at 30 °C for 16 h and imaged. Bright field images (A and E) are single focal plane images, and other images are maximum intensity projections from Z-series stacks. An-Nup49-mCherry allows visualization of the nuclear envelope. The control strain has gone through several cell cycles resulting in many nuclei. The nuclei are normal sized. In the pucAΔ germling, however, there is only one large nucleus, and it is extremely stretched as shown by An-Nup49-mCherry fluorescence. There is also faint mCherry and T-Sapphire fluorescence in the vacuole in the conidial swelling of the germling (blue arrowhead in F–H). The histone H1 fluorescence in the pucAΔ strain is very faint. Note there are four ungerminated parental conidia below the conidial swelling that show bright histone H1-T-Sapphire fluorescence. (I–L) DAPI staining of control and pucAΔ germlings incubated for 14 h at 30 °C and fixed. I and K are single focal plane images, and J and L are maximum intensity projections of Z-series stacks. Nuclear fluorescence is barely visible in L even though software was used to increase the brightness level 3.5X in L relative to J. The arrowhead in L points to a possible nucleus. The punctate DAPI staining is mitochondrial DNA. (M–N) Control (LO9537) and pucAΔ (pucAΔ heterokaryon made in parent strain LO9481) conidia containing histone H1-T-Sapphire and An-Nup49-mCherry were grown for 12, 14, and 16 h at 30 °C and then live imaged for one hour at 30 °C. The percentage of germlings with stretched nuclei was scored (M); error bars indicate mean ± standard deviation of three separate experiments. n = the number of germlings. Similarly, germlings were scored for the presence of obvious nuclear DAPI staining (N); error bars indicate mean ± standard deviation of two experiments (n = 100 germlings per time point). All germlings scored displayed mitochondrial DAPI staining, indicating DAPI staining was successful.

To help understand this phenomenon we deleted pu<span class="Species">cA in strain LO10761 (which <span class="Species">carries An-Nup49-mCherry). Live imaging revealed that pucAΔ resulted in nuclei growing to be much larger than control nuclei (Fig. 3B vs F). Upon longer incubation periods, the nuclear envelopes of pucAΔ nuclei exhibited varying degrees of invagination and many became stretched (Fig. S1A–F). Two or more spindle-pole bodies (SPBs) attaching to cytoplasmic microtubules might result in forces being applied to nuclei from different directions. This would, in turn, result in nuclear stretching. We were, consequently, curious as to whether deletion of pucA blocked nuclei in G1 but allowed for SPB duplication. We deleted pucA in strain LO10066 which carries GFP-alpha-tubulin, HH1-T-Sapphire, and SepK-tdTomato (a SPB marker). pucAΔ germlings were incubated at 37 °C for 8 h and then z-series stacks of random fields were captured over the next hour (two separate experiments). At this time point and temperature, control strains have undergone 2–3 rounds of nuclear division. Most pucAΔ germlings (35/49) had either a single nucleus or no obvious histone H1 fluorescence and, of these germlings, most (33/35) had a single SPB despite the fact that many of the nuclei were severely stretched (14/33). Scoring germlings with two nuclei was challenging as the faint histone H1 signal made it difficult to determine whether the nuclei were completely separate or still connected. However, of the germlings that we scored as having two nuclei (14/49 germlings), most (11/14) had a single SPB (Fig. S1J). These data indicate that abnormal SPB duplication is not the cause of the stretching we observe. These results also indicate that forces are applied to the nucleus other than through the SPB.

Long-term time-lapse imaging of pu<span class="Species">cAΔ germlings <span class="Species">carrying An-Nup49-mCherry and histone H1-GFP (pucAΔ heterokaryon made in parent strain LO9775) revealed that, given sufficiently long incubation, some pucAΔ germlings would eventually undergo what appeared to be nuclear division (14/23 germlings imaged for 16–18 h at 37 °C). However, this was difficult to evaluate because determining whether nuclei underwent nuclear division or were simply pulled apart was sometimes challenging. Interestingly, in those pucAΔ germlings that did undergo obvious nuclear division, the histone H1 fluorescence would change rapidly from not apparent to brightly visible as the chromatin condensed (Movie S1). These data demonstrate that chromosomes are present in the large pucAΔ nuclei even though they are not visible in interphase. Our interpretation is that DNA replication is strongly inhibited by pucAΔ but nuclear growth proceeds, and this results in very large nuclei with very diffuse chromatin. Eventually, the pucAΔ-induced G1 block is overcome in some germlings resulting in a form of mitosis, but, even in those germlings it is not clear that the cell cycle proceeded normally from G1 to M or that mitosis was normal.

Supplementary video related to this article <span class="Species">can be found at https://doi.org/10.1016/j.simyco.2018.06.002.

The following are the supplementary video related to this article:

Movie S1

Histone H1 fluorescence becomes visible during and immediately after mitosis in Images are projections of Z-series stacks <span class="Species">captured at 10-min intervals and follow two pu<span class="Species">cAΔ germlings (parent strain LO9775) carrying histone H1-GFP (left panel) and An-Nup49-mCherry (middle panel). GFP and mCherry channels are merged in the right panel. A bright field image of the two pucAΔ germlings can be found in Fig. S6. This movie begins ∼6.5 h after pucAΔ conidia were incubated at 37 °C. Each pucAΔ germling has a single, large nucleus at the beginning of this movie (15:51) with faint histone H1-GFP fluorescence. In the lower pucAΔ germling, histone H1-GFP becomes apparent at 18:11, and a likely nuclear division event occurs at 18:31, which is ∼9 h after incubation at 37 °C. The histone H1-GFP signal remains bright for ∼40 min after mitosis before decreasing in intensity, and both nuclei are stretched and move around rapidly for the duration of the movie. In the upper germling, histone H1-GFP becomes apparent at 21:01, and a likely nuclear division event occurs at 21:31, which is ∼12 h after incubation at 37 °C.

Notably, this diffuse nuclear histone H1 fluorescence phenotype has been described previously for the deletion of the nimXCdk1 gene (De Souza ). NimXCdk1 has been shown to physi<span class="Species">cally interact with both Pu<span class="Species">cA and NimECyclin B in A. nidulans (De Souza ). It is, thus, possible that the phenotype we observed in pucAΔ germlings is caused by a lack of NimX kinase activity due to the absence of its binding partner PucA.

Deletion of pucA in cdhAΔ strains results in an interphase cell cycle delay

The deletion of pu<span class="Species">cA in a <span class="Chemical">cdhAΔ strain results in a decrease in growth and sporulation compared to that of the cdhAΔ parent (Fig. 5A). This indicates that PucA has non-essential functions in addition to its essential function of inactivating APC/C-CdhA. To determine if the cell cycle time was altered in the double mutant, we generated a strain carrying pucAΔ and cdhAΔ and expressing histone H1-mRFP (LO8743, LO8744) to allow us to observe nuclear division. The cell cycle duration was calculated using time-lapse images collected at 10-min intervals at 25 °C. The cell cycle durations for control strains (199 ± 49 min) and cdhAΔ strains (178 ± 39 min) at 25 °C have been determined previously (Edgerton-Morgan & Oakley 2012). We scored a different control strain (LO10327) and found it has an essentially identical cell cycle time (198 ± 38 min, n = 72 nuclei). Compared to control and cdhAΔ strains, the length of the cell cycle was increased in pucAΔ, cdhAΔ strains (248 ± 63 min, n = 100 nuclei) (Fig. 5B). The increase was statistically significant. The p-value for pucAΔ, cdhAΔ vs the control strain was 7.76e-8 and for pucAΔ, cdhAΔ vs cdhAΔ the p-value was 5.74e-17 (unpaired Student's t-test). The duration of mitosis in the pucAΔ, cdhAΔ strain at 25 °C (10 ± 2 min, n = 36) was not significantly different from the control strain (9 ± 1 min, n = 40). It follows that the increased cell cycle time in the pucA, cdhA double deletant was due to a delay in interphase.

Hyphal growth outpaced nuclear repli<span class="Species">cation in pu<span class="Species">cAΔ, cdhAΔ germlings, and this resulted in nuclei being spaced far apart. To quantify this, we fixed control (LO10327), cdhAΔ (LO2019), and pucAΔ, cdhAΔ (LO8743) germlings after 20 h of incubation at 25 °C and used Calcofluor White to visualize the hyphal walls. We scored the distances between interphase nuclei in tip cells that could be seen on the same XY plane using Volocity 6.3 software (Perkin-Elmer). We found that nuclei were 14.5 ± 4.4 μm apart in the control strain (n = 100), nuclei were 12.3 ± 4.3 μm apart in the cdhAΔ strain (n = 97), whereas nuclei were 22.5 ± 7.1 μm apart in the pucAΔ, cdhAΔ strain (n = 89). The fact that longer cell cycle times correlate with increased nuclear spacing is expected. However, the degree of increase in nuclear spacing was disproportionate to the increase in the cell cycle time. Comparing the WT to the cdhAΔ, pucAΔ double deletant, there was a 25 % increase in cell cycle time in the double deletant relative to the WT, but a 55 % increase in internuclear spacing. The double deletant had a 39 % increase in cell cycle time relative to the cdhAΔ strain, but an 83 % increase in nuclear spacing. These data suggest that PucA may have a role, as yet undefined, in internuclear spacing in germlings, beyond its effects on the cell cycle. It is also worth noting that inhibition of colonial growth of the pucAΔ, cdhAΔ strain (Fig. 5A) was disproportionate to the lengthening of the cell cycle in this strain. PucA may, thus, have roles in colonial growth beyond its cell cycle effects and beyond those observable in germlings.

<span class="Chemical">NimECyclin B is required for both S-phase and entry into mitosis (Morris, 1975, Bergen et al., 1984). We were curious if Pu<span class="Species">cA had a role in regulating NimECyclin B accumulation in S and/or G2 and, if so, whether that was the cause of the interphase delay we observe in pucAΔ, cdhAΔ mutants. We generated strains containing pucAΔ and cdhAΔ and expressing NimECyclin B-GFP and histone H1-mRFP (LO10750, LO10751). NimECyclin B-GFP is absent in G1, becomes visible in early S phase, and remains apparent in G2. The length of G1 is proportional to the percentage of interphase nuclei with no visible NimECyclin B-GFP. Previous work from our lab revealed that the mean percentage of tip cells in cdhA+ strains that contained visible NimECyclin B-GFP was 60.0 ± 5.4 % compared to 87.2 ± 3.5 % in cdhAΔ strains (Edgerton-Morgan and Oakley, 2012). In the pucAΔ, cdhAΔ double mutant strains (LO10750 and LO10751), we found that 89.2 ± 4.8 % of tip cells contained visible NimECyclin B-GFP (4 experiments, n = 120 tip cells), and this is essentially identical to pucA+, cdhAΔ strains. Therefore, the lengthened cell cycle in the double mutant does not appear to be due to a delay in NimECyclin B accumulation. NimECyclin B localization appeared normal with a clear enrichment of NimECyclin B at SPBs.

Deletion of pucA in cdhAΔ strains does not increase mitotic defects but causes interphase nuclear abnormalities

To determine if pu<span class="Species">cA has functions in mitosis, we incubated spores from control (<span class="CellLine">LO1806), cdhAΔ (LO2019), and pucAΔ, cdhAΔ (LO8743) strains at 25 °C, captured Z-series image stacks at 3-min intervals and scored the following mitotic errors: 1) failure of nuclei to divide, 2) mitotic delay (abnormally long periods with condensed chromosomes), 3) apparent nuclear division followed by separated chromatin masses coming back together to reform a single nucleus, and 4) abnormal chromosome segregation (Fig. S2). cdhAΔ mutants showed an increase in the rate of total mitotic errors (8.5 % of nuclei, n = 177) compared to control (0.94 % of nuclei, n = 106). The percentage, however, of mitotic errors observed in the pucAΔ, cdhAΔ strain (10 % of nuclei, n = 160) was not significantly different from that observed in cdhAΔ. As presented above, we found no significant difference in the length of mitosis in the pucAΔ, cdhAΔ strain vs control strains. Taken together, our data indicate that pucAΔ does not add significantly to the mitotic abnormalities seen with cdhAΔ strains.

In the course of looking for <span class="Disease">mitotic abnormalities, we observed a number of abnormalities in nuclear appearance and behaviour that were rarely or never seen in control strains. Most of these were also seen in <span class="Chemical">cdhAΔ strains, but their frequency was enhanced in the cdhAΔ, pucAΔ strain. These abnormalities included abnormally large nuclei, nuclei that were pulled or stretched visibly (Fig. S2 D, E), and, most strangely, nuclei in which the chromatin partially condensed and decondensed without mitosis occurring (Movie S2). This phenotype of chromatin partially condensing and decondensing happened to individual nuclei within a cell while the chromatin in other nuclei remained decondensed (Fig. S2F, Movie S2). Altogether, these abnormalities were significant in number with 33.6 % of cdhAΔ, pucAΔ nuclei exhibiting at least one of these abnormalities in the 30 min prior to mitosis and 22.4 % exhibiting them in the 30 min after mitosis. These abnormalities are quantified in Fig. S2. The abnormally large and stretched nuclear phenotypes are notable in that they are also seen in pucAΔ, cdhA+ germlings from heterokaryons. We looked for microtubule abnormalities that might account for the nuclear stretching, but none were evident (results not shown). The untimely chromosomal condensation and decondensation phenotype was never seen in control cells. These data suggest that PucA, possibly in combination with CdhA, has a role in regulating chromosomal condensation, and, more specifically, in preventing untimely chromosomal condensation.

Supplementary video related to this article <span class="Species">can be found at https://doi.org/10.1016/j.simyco.2018.06.002.

The following are the supplementary video related to this article:

Movie S2

. Z-series stacks of a <span class="Chemical">cdhAΔ, pu<span class="Species">cAΔ germling (strain LO8743) were captured at 3-min intervals at 25 °C, and each frame in the movie is a maximum intensity projection of a Z-series stack. In this movie, five interphase nuclei are visible. The third nucleus from the right partially condenses at 13:26, whereas all other nuclei remain decondensed. This nucleus partially condenses and decondenses repeatedly for the duration of the movie. The nucleus at the lower left also slightly condenses (14:11) and decondenses repeatedly, although not as severely.

Attempts to localize PucA

For clarifying the functions of Pu<span class="Species">cA it would obviously be advantageous to observe its localization. We have not been successful in doing so, but we will summarise our efforts here because we believe they are informative. We generated an N-terminal GFP fusion to PucA but our transformation frequency with the fragment was lower than our positive control. Nevertheless, we were able to isolate 40 transformants from two transformations and carry out diagnostic PCR on them. None of the transformants, however, had a correct GFP-pucA integration. These data appear to indicate that transformants carrying the correct N-terminal fusion are not viable. We did not, however, obtain heterokaryon transformants as we would have expected if the GFP-pucA construct was recessive lethal. Transformants carrying a C-terminal GFP fusion to PucA were viable but no fluorescence signal was observed. It is likely that PucA is present at levels too low to detect and/or that its half-life is too short to allow GFP to fold and become fluorescent before the protein is destroyed (a particular problem with short lived proteins such as cyclins). We tested a fast-folding fluorescent protein, Superfolder GFP (sfGFP) (Malagon 2013), fusing it to PucA and to NimECyclin B. We did not see a fluorescent signal for the C-terminal tagged version of PucA with sfGFP. Interestingly, the fluorescent signal for C-terminal tagged NimECyclin B with sfGFP was weaker than the signal from the GFP variant we currently utilize in the lab (Fernandez-Abalos ), suggesting that the variant we use is superior to sfGFP.

Localization of ClbA (AN2137)

We generated strains that <span class="Species">carried <span class="Chemical">ClbA-GFP and histone H1-mRFP (LO10126 and LO10127), and these strains grew as well as WT strains at all temperatures tested (data not shown). Long-term live time-lapse imaging at 37 °C revealed that ClbA-GFP was first detectable as a single dot in the nucleus 23 ± 10 min before mitosis (n = 73). This correlated to 23 % of the cell cycle. G2 of A. nidulans lasts approximately 40 % of the cell cycle at 37 °C (Bergen & Morris 1983), and, thus, ClbA becomes detectable in mid-G2. Imaging at 1-min intervals revealed that ClbA-GFP began to disappear from most nuclei just prior to chromatin condensation (51/58 nuclei) (Fig. 6A–O'). ClbA-GFP was rarely observed in nuclei during anaphase or telophase, but occasionally we observed faint ClbA-GFP between the separating chromatin masses in anaphase.

Fig. 6

(A–O') ClbA localizes to kinetochores in Gand disappears at mitotic entry. All images were obtained with a spinning disk confocal microscope. T = 0 is in late G2 and the nucleus transits into mitosis in the following 2 min. Anaphase is underway by 4 min and nuclear division is largely completed by 6 min. The A–M and B–N columns show maximum intensity projections of Z-series stacks. The C–O column shows single focal plane images taken from the same data set, and the C'–O' column shows intensity traces over the lines shown in the C–O column with ClbA-GFP fluorescence shown in green and histone H1-mRFP shown in red. ClbA-GFP localizes to a dot in the nucleus in G2 (arrows in A, D) and disappears during mitotic entry (G). ClbA-GFP is not observed as a dot in the nucleus again until the next G2. (P–R') ClbA-GFP localizes adjacent to the SPB marker SepK. Histone H1-mTagBFP2 is shown in gray, SepK-tdTomato in red, and Clb-GFP in green. P and Q are three dimensional projections, while R is a single focal plane image from the Z-series stack. The minimal overlap is verified by intensity traces in R', which are along the line shown in R. The tdTomato trace is shown in red and the GFP trace in green. (S–U') ClbA-GFP co-localizes with a kinetochore marker Ndc80. S and T are three dimensional projections from a Z-series stack with HH1-mTagBFP2 shown in gray, An-Ndc80-mCherry in red and ClbA-GFP in green. U is a single focal plane image from the same series. The overlap is verified by intensity traces in the mCherry (red) and GFP (green) channels in U' along the line shown in U. In P and S the position of the nucleolus is shown by the absence of histone fluorescence (blue arrows).

(A–O') <span class="Chemical">ClbA localizes to kinetochores in Gand disappears at mitotic entry. All images were obtained with a spinning disk confocal microscope. T = 0 is in late G2 and the nucleus transits into mitosis in the following 2 min. Anaphase is underway by 4 min and nuclear division is largely completed by 6 min. The A–M and B–N columns show maximum intensity projections of Z-series stacks. The C–O column shows single focal plane images taken from the same data set, and the C'–O' column shows intensity traces over the lines shown in the C–O column with ClbA-GFP fluorescence shown in green and histone H1-mRFP shown in red. ClbA-GFP localizes to a dot in the nucleus in G2 (arrows in A, D) and disappears during mitotic entry (G). ClbA-GFP is not observed as a dot in the nucleus again until the next G2. (P–R') ClbA-GFP localizes adjacent to the SPB marker SepK. Histone H1-mTagBFP2 is shown in gray, SepK-tdTomato in red, and Clb-GFP in green. P and Q are three dimensional projections, while R is a single focal plane image from the Z-series stack. The minimal overlap is verified by intensity traces in R', which are along the line shown in R. The tdTomato trace is shown in red and the GFP trace in green. (S–U') ClbA-GFP co-localizes with a kinetochore marker Ndc80. S and T are three dimensional projections from a Z-series stack with HH1-mTagBFP2 shown in gray, An-Ndc80-mCherry in red and ClbA-GFP in green. U is a single focal plane image from the same series. The overlap is verified by intensity traces in the mCherry (red) and GFP (green) channels in U' along the line shown in U. In P and S the position of the nucleolus is shown by the absence of histone fluorescence (blue arrows).

The “dot” lo<span class="Species">calization pattern is consistent with localization to the SPB and/or kinetochores (KTs). In A. nidulans, the KTs are immediately adjacent to the SPB during interphase, and they do not separate from the SPB except in prometaphase through early anaphase (Yang et al., 2004, De Souza et al., 2009). To confirm that ClbA-GFP localizes to the SPB and/or KTs, we generated two additional strains that carried ClbA-GFP. Both carried histone H1-mTagBFP2 (Subach ), but one carried the SPB marker SepK-tdTomato (Xiong & Oakley 2009), while the other carried the KT marker An-Ndc80-mCherry. Interestingly, we noticed that ClbA-GFP and SepK-tdTomato did not fully overlap and instead localized immediately adjacent to each other (Fig. 6 P–R') while ClbA-GFP and An-Ndc80-mCherry did overlap precisely (Fig. 6 S–U').

This result suggests that <span class="Chemical">ClbA localizes to KTs rather than the SPBs. We performed colocalization analysis with Volocity 6.3 software (Perkin-Elmer) using the Costes Pearson's Correlation algorithm (Manders et al., 1993, Costes et al., 2004, Barlow et al., 2010). With this algorithm, higher correlation values indicate more precise 3D co-localization (1.0 = 100 % overlap of signal). In some cases, optical systems can cause an apparent lateral displacement of objects imaged at different wavelengths. To control for this, we imaged microspheres that fluoresce at both GFP and tdTomato/mCherry wavelengths (Molecular Probes' PS-Speck Microscope Point Source Kit) and found a correlation value of 0.905 ± 0.044 (n = 22 microspheres) when using the 488 nm and 561 nm lasers. The high correlation value indicates that the two wavelengths are not shifted significantly in the XY dimension in our system. We found a correlation value of 0.253 ± 0.234 of ClbA-GFP to SepK-tdTomato (n = 86 nuclei, 2 strains LO11067 and LO11068) and a correlation value of 0.761 ± 0.100 of ClbA-GFP to An-Ndc80-mCherry (n = 89 nuclei, 2 strains LO11253 and LO11254). (Note that the same filter set was used to image SepK-tdTomato and An-Ndc80-mCherry.) These data indicate strongly that ClbA localizes to KTs.

ClbA is not essential but destruction of ClbA is required for viability

To determine if <span class="Chemical">clbA is essential we deleted it by replacing it with AfpyrG. We found that <span class="Chemical">clbAΔ strains (LO10129-LO10131) grew like WT strains at all temperatures tested (data not shown).

An important characteristic of group I <span class="Gene">cyclins is that their destruction is important for their roles in the cell cycle. In this re<span class="Chemical">gard, group I cyclins, particularly the B-type cyclins, have motifs called destruction boxes (d-boxes, RXXLXXXXN) that are recognized by the APC/C. They are then ubiquitinated by the APC/C, and this leads to their destruction by the proteasome. The d-boxes are typically near the N-terminus of these cyclins, and truncation of the N-terminus to delete the d-box or boxes results in a cyclin that drives the cell cycle but arrests the cell cycle at a point at which destruction of the cyclin is required (Murray ). Removal of the d-box of NimECyclin B, for example, causes a mitotic exit arrest and consequent growth inhibition (De Souza et al., 2009, Nayak et al., 2010).

<span class="Chemical">ClbA contains two putative d-boxes in its N-terminal region (Fig. 7A). The first putative d-box in <span class="Chemical">ClbA (RAAFGDVSN) is not canonical, but it is almost identical to the NimECyclin B d-box (RAALGDVSN). The second putative d-box in ClbA (RKTLNKRAT) contains the conserved “RXXL” but not the conserved asparagine at the 9th position. No other identifiable d-box motifs are present in the N-terminus of ClbA. We wished to determine if ClbA d-box motifs play a role in ClbA destruction and, if so, whether destruction of ClbA is important for cell cycle progression.

Fig. 7

Inducing expression of full-length and truncated ClbA via the regulatable Diagram of full-length (FL) ClbA with two putative destruction box motifs, db1 (RAAFGDVSN) and db2 (RKTLNKRAT), and two truncated versions of ClbA. CBOX1 designates the first cyclin box fold, also called the N-terminal cyclin domain, and CBOX2 designates the C-terminal cyclin box domain. Constructs shown in A were fused to the nmtA regulatable promoter and placed at the wA locus. B. The nmtA promoter [nmtA(p)] was fused to the N-terminus of both FL-ClbA fused to GFP and truncated (db2Δ) ClbA fused to GFP. These fusion fragments were then placed at the wA locus. The endogenous clbA gene was not altered in any of the above strains. Three nmtA(p)-FL-clbA-GFP strains (LO11119-LO11121) grow as well as the control and WT strains (LO10327 and FGSC4) under both repressing (high thiamine) and non-repressing (low or no thiamine) conditions. Three nmtA(p)-db2Δ-clbA-GFP strains (LO11122-LO11124) grow as well as the WT under repressing conditions (YAG and 1.0 μM thiamine), and grow nearly normally at a thiamine concentration of 0.1 μM. They are extremely sick under partially-repressing conditions (10 nM thiamine) and dead under non-repressing conditions (1.0 nM thiamine or less). Although GFP-tagged strains are shown, identical results were obtained with strains expressing non-GFP-tagged FL ClbA and db2Δ-ClbA.

Inducing <span class="Species">expression of full-length and truncated ClbA via the regulatable Diagram of full-length (FL) ClbA with two putative destruction box motifs, db1 (RAAFGDVSN) and db2 (RKTLNKRAT), and two truncated versions of ClbA. CBOX1 designates the first cyclin box fold, also called the N-terminal cyclin domain, and CBOX2 designates the C-terminal cyclin box domain. Constructs shown in A were fused to the nmtA regulatable promoter and placed at the wA locus. B. The nmtA promoter [nmtA(p)] was fused to the N-terminus of both FL-ClbA fused to GFP and truncated (db2Δ) ClbA fused to GFP. These fusion fragments were then placed at the wA locus. The endogenous clbA gene was not altered in any of the above strains. Three nmtA(p)-FL-clbA-GFP strains (LO11119-LO11121) grow as well as the control and WT strains (LO10327 and FGSC4) under both repressing (high thiamine) and non-repressing (low or no thiamine) conditions. Three nmtA(p)-db2Δ-clbA-GFP strains (LO11122-LO11124) grow as well as the WT under repressing conditions (YAG and 1.0 μM thiamine), and grow nearly normally at a thiamine concentration of 0.1 μM. They are extremely sick under partially-repressing conditions (10 nM thiamine) and dead under non-repressing conditions (1.0 nM thiamine or less). Although GFP-tagged strains are shown, identical results were obtained with strains expressing non-GFP-tagged FL ClbA and db2Δ-ClbA.

We created a version of <span class="Chemical">ClbA in which we removed the first d-box (db1Δ<span class="Chemical">-ClbA, eliminating amino acids 2–60) and another version in which we deleted the N-terminal region that contained both putative d-boxes (db2Δ-ClbA, eliminating amino acids 2–147) (Fig. 7A). We also created versions of the d-box-deleted ClbA and two d-box-deleted ClbA in which GFP was fused to the C-terminus (db1Δ-ClbA-GFP and db2Δ-ClbA-GFP). We initially placed db1Δ-ClbA, db1Δ-ClbA-GFP, db2Δ-ClbA, db2Δ-ClbA-GFP, full length (FL) ClbA, and FL-ClbA-GFP (separately) under the control of the regulatable alcA promoter [alcA(p)] at the white (wA) locus in a strain expressing histone H1-mRFP (parental strain LO1516). In these strains, the endogenous clbA gene was not altered. On repressing media, all six strains grew like WT and control strains (Fig. S3). On inducing media, the expression of db1Δ-ClbA or db1Δ-ClbA-GFP significantly decreased growth while the expression of db2Δ-ClbA or db2Δ-ClbA-GFP almost completely inhibited growth and sporulation. Induction of expression of FL-ClbA or FL-ClbA-GFP did not alter growth compared to WT strains. Since the endogenous copy of clbA was not altered in these strains, these data indicate that destruction of ClbA is essential for viability and that d-box-deleted versions of ClbA act in a dominant negative fashion.

We also wished to determine the phenotype <span class="Species">caused by failure of destruction of <span class="Chemical">ClbA. Since the growth phenotypes caused by db2Δ-ClbA expression and db2Δ-ClbA-GFP expression appear to be the same, we have used GFP fusions for the bulk of our experiments, because the GFP allows us to monitor the location of the molecules. We encountered difficulty with the alcA(p) due to its “leakiness”. While colony growth was normal on alcA repressing media (Fig. S3), spore viability was very low in strains carrying the alcA(p) driving expression of db2Δ-ClbA or db2Δ-ClbA-GFP even on alcA(p) repressing media, making it difficult to determine the phenotype of db2Δ-ClbA-GFP via microscopy. We reasoned that this was due to leakiness of the alcA(p) therefore, we searched for a more strongly repressible promoter.

The nmtA promoter can be regulated more effectively than the alcA promoter

In S. pombe, nmt1 has been shown to be highly transcribed in minimal medium and strongly repressed by the addition of <span class="Chemical">thiamine (Maundrell, 1990, Tamm, 2012). Our lab had previously analysed the promoter activity of a 761-bp sequence immediately upstream of the A. nidulans nmt1 homolog <span class="Chemical">nmtA (An8009). This 761 bp nmtA promoter [nmtA(p)] sequence, fused to GFP was placed at the yA locus and found to repress GFP fluorescence in the presence of thiamine and allow constitutive GFP expression in the absence of thiamine.

We placed <span class="Chemical">FL-ClbA-GFP and db2Δ<span class="Chemical">-ClbA-GFP under control of nmtA(p) at the wA locus in a strain that carries histone H1-mRFP (parental strain LO1516). The endogenous copy of clbA was not altered. We found that a thiamine concentration of 0.1 μM repressed nmtA(p)-driven expression of db2Δ-ClbA-GFP sufficiently to allow nearly normal growth and 1.0 μM or more was sufficient to allow growth like the WT on solid (Fig.7B) and in liquid medium. These results suggested that nmtA(p) can be more strongly repressed than alcA(p) and this, in fact, proved to be the case (see results below).

Failure to degrade ClbA results in a mitotic arrest as well as nondisjunction

To determine the phenotype of db2Δ<span class="Chemical">-ClbA-GFP, we collected conidia from strains <span class="Species">carrying double d-box-deleted, GFP tagged ClbA, [LO11122 (nmtA(p)-db2Δ-clbA-GFP)], a strain carrying clbA+ fused to GFP and under control of nmtA(p) [LO11119 (nmtA(p)-FL-clbA-GFP)], and a control clbA+ strain (LO10327). The conidia were collected from material grown on solid medium containing a concentration of 0.1 μM thiamine, a concentration that partially represses expression (Fig. 7B). To allow expression from nmtA(p), we inoculated these conidia into liquid media without thiamine. We incubated them at 30 °C for ∼6 h, and then captured Z-axis image stacks at 10-min intervals for 12 h. We found that LO11122 conidia (nmtA(p)-db2Δ-clbA-GFP) generated short germ tubes, and, once the cells entered mitosis, they became blocked in mitosis. The mitotically arrested nuclei displayed one or more of the following defects: failure of anaphase A, failure of anaphase B, nondisjunction, or chromatin collapsing back into a single mass after initially separating.

To better quantify these observations, we fixed LO11122, LO11119, and LO10327 germlings after 12 h of growth at 30 °C. At this time and temperature, control germlings (LO10327) had undergone 2.0 ± 0.1 nuclear divisions and contained ∼ 4 nuclei (206 germlings scored from 2 separate experiments). <span class="Species">Expression of <span class="Chemical">FL-ClbA-GFP from nmtA(p) (strain LO11119), in the absence of thiamine, slightly decreased the number of nuclear divisions to 1.8 ± 0.1 (207 germlings scored in 2 separate experiments). FL-ClbA-GFP is predicted to be over expressed in these strains, and this was confirmed by microscopy. This overexpression, however, had a modest effect on the cell cycle, if any. Expression of db2Δ-ClbA-GFP (LO11122), however, drastically reduced the number of nuclear divisions to 0.3 ± 0.1 (180 germlings scored from 2 separate experiments). This indicates that destruction of ClbA is required for nuclear replication. Additionally, we found that chromatin was condensed in 53.8 % ± 3.1 % of germlings in the db2Δ-ClbA-GFP expressing strain (LO11122) nuclei compared to 5.9 % ± 2.1 % in the FL-ClbA-GFP expressing strain (LO11119) and 4.5 ± 1.6 % in the clbA+ strain (LO10327). These data reveal that expression of db2Δ-ClbA-GFP results in a strong mitotic block.

To determine the nature of the <span class="Disease">mitotic arrest, we needed to be able to collect Z-series image stacks at short intervals. We found that if we collected LO11122 conidia from solid medium with a concentration of 1.0 μM <span class="Chemical">thiamine and then grew those conidia at 30 °C in liquid media without thiamine, germination and initial growth were normal. Germlings underwent 2–3 rounds of normal nuclear divisions and then began to show mitotic abnormalities and mitotic blockage. This approach allowed us to find and image cells entering mitosis and displaying the phenotypes caused by expression of db2Δ-ClbA-GFP.

Imaging revealed that, as anticipated, little or no of db2Δ<span class="Chemical">-ClbA-GFP was present initially. The absence of db2Δ<span class="Chemical">-ClbA-GFP is almost certainly due to the conidia carrying over enough thiamine to inhibit expression from nmtA(p). Upon continued incubation without thiamine, db2Δ-ClbA-GFP began to be visible at KTs. The exact timing of its appearance varied to some extent among germlings, but it generally became detectable between the second and third nuclear divisions. Nuclei often went through one round of mitosis after db2Δ-ClbA-GFP became visible. In these cases, we found that db2Δ-ClbA-GFP was removed from the KTs, and this strongly correlated with mitotic entry. In most cases mitosis appeared normal. Immediately after mitosis, however, db2Δ-ClbA-GFP became visible at KTs (in marked contrast to controls), and db2Δ-ClbA-GFP fluorescence increased markedly through interphase. At the end of this interphase db2Δ-ClbA-GFP left the KTs, nuclei entered mitosis, and a catastrophic mitosis followed (Movie S3, Fig. 8). Full condensation of chromatin was never observed until db2Δ-ClbA-GFP left the KTs (n = 46/46 nuclei, strains LO11122 and LO11123). These data indicate that ClbA is removed from KTs by a mechanism other than simple destruction, and they are consistent with the hypothesis that ClbA must be removed from KTs in order for mitotic entry to occur.

Fig. 8

Expression of db2Δ-ClbA-GFP results in nondisjunction and mitotic arrest in anaphase. Conidia carrying nmtA(p)-db2Δ-clbA-GFP and histone H1-mRFP (strains LO11122-LO11124) were collected from hyphae growing on media containing 1.0 μM thiamine, incubated at 30 °C for 10–12 h in non-repressing liquid media (i.e. lacking thiamine), and then imaged in 2-min intervals. Images are maximum intensity projections from Z-series stacks. In G2, db2Δ-ClbA-GFP localizes strongly to the kinetochores (KTs) and faintly to the nucleoplasm (A–D). At mitotic entry, db2Δ-ClbA-GFP leaves the KTs but can be seen faintly in the nucleoplasm (E–H). Mitosis is stalled or blocked in anaphase, and db2Δ-ClbA-GFP can be seen faintly along the spindle and at a dot in the separating chromatin (M–T). White arrows designate db2Δ-ClbA-GFP at the KTs and/or nucleoplasm (B, F, J). Blue arrows point to chromosomes that failed to disjoin properly (M, Q).

<span class="Species">Expression of db2Δ<span class="Chemical">-ClbA-GFP results in nondisjunction and mitotic arrest in anaphase. Conidia carrying nmtA(p)-db2Δ-clbA-GFP and histone H1-mRFP (strains LO11122-LO11124) were collected from hyphae growing on media containing 1.0 μM thiamine, incubated at 30 °C for 10–12 h in non-repressing liquid media (i.e. lacking thiamine), and then imaged in 2-min intervals. Images are maximum intensity projections from Z-series stacks. In G2, db2Δ-ClbA-GFP localizes strongly to the kinetochores (KTs) and faintly to the nucleoplasm (A–D). At mitotic entry, db2Δ-ClbA-GFP leaves the KTs but can be seen faintly in the nucleoplasm (E–H). Mitosis is stalled or blocked in anaphase, and db2Δ-ClbA-GFP can be seen faintly along the spindle and at a dot in the separating chromatin (M–T). White arrows designate db2Δ-ClbA-GFP at the KTs and/or nucleoplasm (B, F, J). Blue arrows point to chromosomes that failed to disjoin properly (M, Q).

During these <span class="Species">catastrophic mitoses, although db2Δ<span class="Chemical">-ClbA-GFP left the KTs at mitotic onset, it was present at higher levels in the nucleoplasm than in the cytoplasm in the majority of nuclei (45/46 nuclei; Fig. 8 F, J). Overexpressed ClbA-GFP or ClbA-GFP driven by its endogenous promoter rapidly disappeared from the nucleoplasm of mitotic nuclei. The vast majority of nuclei in db2Δ-ClbA-GFP germlings entered anaphase (43/46 nuclei; Fig. 8 M–P) relatively quickly after chromosomal condensation but displayed obvious nondisjunction (42/43 nuclei; Fig. 8 M–T). The chromatin in many nuclei eventually collapsed back into one mass (34/43 nuclei) after initially separating in anaphase. This phenotype is consistent with force being exerted on the chromosomes by the mitotic apparatus, causing chromosomal separation but daughter chromatids failing to disjoin from each other and the elasticity of chromatin pulling the chromatids back into a single mass. During these failed anaphase attempts, db2Δ-ClbA-GFP could be seen faintly on the spindle, in the nuclei, and/or at dot(s) in the separating chromatin (46/46 nuclei) (Fig. 8 N–P, R–T). These data, in aggregate indicate that destruction of ClbA is required for chromosomal disjunction.

Supplementary video related to this article <span class="Species">can be found at https://doi.org/10.1016/j.simyco.2018.06.002.

The following are the supplementary video related to this article:

Movie S3 (corresponds to Figure 8)

<span class="Species">Expression of db2Δ<span class="Chemical">-ClbA-GFP causes severe non-disjunction and mitotic arrest in anaphase. This movie follows a single germling expressing db2Δ-ClbA-GFP (left panel) and histone H1-mRFP (middle panel). GFP and mRFP channels are merged in the right-most panel. Images are maximum intensity projections from Z-series stacks taken at 1-min intervals. db2Δ-ClbA-GFP begins to leave the kinetochores at 7:46 and is gone from all nuclei at 7:50. The top left nucleus attempts anaphase at 7:53 but immediately collapses back to one mass. The other three nuclei enter anaphase and display severe non-disjunction for the duration of the movie. After mitosis begins, db2Δ-ClbA-GFP can be seen in the nucleoplasm, along the spindle, and faintly at the poles.

Interestingly, we observed that <span class="Chemical">nmtA(p)-driven <span class="Species">expression of either ClbA-GFP or db2Δ-ClbA-GFP resulted in a GFP signal at septa. This was not observed in strains in which expression of ClbA-GFP was driven by its endogenous promoter, even when the laser and exposure settings were increased to allow imaging of very faint signals (data not shown).

Destruction of NimECyclin B is important for chromosomal disjunction

<span class="Species">Expression of destruction box-deleted <span class="Chemical">NimECyclin B (dbΔ-NimECyclin B) also results in a strong mitotic block. dbΔ-NimECyclin B-GFP expressing cells enter mitosis and progress through anaphase before becoming blocked in telophase with dbΔ-NimECyclin B-GFP remaining at the spindle poles (De Souza et al., 2009, Nayak et al., 2010). We utilized the same destruction box-deleted cyclin B construct used in previous work (De Souza et al., 2009, Nayak et al., 2010) fused to GFP (dbΔ-NimECyclin B-GFP) but placed it under the control of the A. nidulans nmtA promoter at the wA locus (LO11357-LO11359). Our results were consistent with the previously mentioned studies (data not shown). However, when we followed nuclei through mitosis by capturing time-lapse Z-series stacks at one-minute intervals, we also observed a high frequency of nondisjunction in dbΔ-NimECyclin B-GFP expressing cells (19/32 nuclei) (Movie S4), which was not previously reported. Thus, destruction of both ClbA and NimECyclin B during mitosis is required for proper chromosomal disjunction.

Supplementary video related to this article <span class="Species">can be found at https://doi.org/10.1016/j.simyco.2018.06.002.

The following are the supplementary video related to this article:

Movie S4

<span class="Species">Expression of dbΔ-NimE-GFP results in non-disjunction and <span class="Disease">mitotic arrest in anaphase. Images are of a strain expressing dbΔ-NimECyclin B-GFP (left panel) and histone H1-mRFP (middle panel) (LO11359). GFP and mRFP channels are merged in the right most panel. This movie follows two nuclei, and images are projections of Z-series stack captured at 1-min intervals. Nuclei enter mitosis at 11:10, and dbΔ-NimECyclin B-GFP can be seen faintly in the nucleoplasm and at the SPBs. dbΔ-NimECyclin B-GFP remains at the poles as the duplicated SPBs move apart. Both nuclei enter anaphase (11:16-11:18), and non-disjunction occurs for the duration of the movie. The faint GFP fluorescence at the right side of the movie is an out-of-focus hypha.

Cohesin is removed from chromosomes in db2Δ-ClbA and dbΔ-NimECyclin B expressing strains

The cohesin complex holds sister chromatids together, and its removal is required for chromosomal disjunction. Cohesin complex removal is regulated by mitotic <span class="Gene">cyclins and the <span class="Disease">APC/C [reviewed in Wong (2010)]. Failure of cohesin complex removal is, thus, a plausible cause for the high frequencies of non-disjunction caused by db2Δ-ClbA and dbΔ-NimECyclin B. Although the cohesin complex has been studied extensively in other organisms, there is very little information on it in filamentous fungi. To determine if db2Δ-ClbA and dbΔ-NimECyclin B altered the removal of the cohesin complex, we first had to investigate the localization and removal of the complex in control cells.

The core cohesin complex is composed of <span class="Gene">Scc1 (also known as <span class="Gene">Rad21 in humans and Mcd1 in S. cerevisiae), Scc3, and two SMC (Structural Maintenance of Chromosomes) proteins called SMC1 and SMC3. The A. nidulans homolog of human Rad21 and S. cerevisiae Mcd1 is AN7465 (E-value of 2E-29 and 3E-17, respectively), which we now designate sccA. Strains carrying SccA-GFP and HH1-mRFP (LO3231) had previously been constructed in our lab. We found that SccA-GFP localizes to nuclei but not nucleoli throughout interphase (arrows in Fig. S4A–C), and, as expected, it leaves the chromatin during mitosis. To narrow down the timing of disappearance of SccA-GFP from chromosomes during mitosis, we captured Z-series image stacks at 1-min intervals (n = 42 nuclei). SccA-GFP localized to condensed chromatin (42/42 nuclei, Fig. S4D–F) early in mitosis but, as expected, disappeared at the beginning of anaphase. In some cases (10/42 nuclei), SccA-GFP was observed between the separating chromatin in early anaphase (Fig. S4G–I, see blue arrow), but this signal always disappeared in late anaphase. After mitosis, SccA-GFP slowly accumulated in nuclei in early G1, and this signal intensified as the cell cycle progressed.

We next wanted to determine if Sc<span class="Species">cA was removed from chromosomes during mitosis in db2Δ<span class="Chemical">-ClbA and dbΔ-NimECyclin B expressing cells. We generated strains that carried SccA-GFP, histone H1-mRFP, and either nmtA(p)-db2Δ-clbA (LO11211) or nmtA(p)-dbΔ-nimE (LO11217) inserted at the wA locus. A control strain (LO11202) that carried SccA-GFP and histone H1-mRFP, was constructed by inserting AfpyroA at the wA locus. The endogenous clbA and nimE genes were not altered in any strain. We harvested conidia from each of these strains grown on media with 1.0 μM thiamine. We inoculated conidia into liquid media lacking thiamine (allowing expression of the d-box-deleted cyclins). incubated the cultures at 30 °C for ∼6 h, and then captured images of Z-axis stacks at 10-min intervals for 12 h.

<span class="Species">Expression of db2Δ<span class="Chemical">-ClbA (Fig. 9) or dbΔ-NimECyclin B (Fig. S5) led to a mitotic arrest in the second or third nuclear division cycle, as seen in previous experiments. Perhaps surprisingly, we found that SccA-GFP disappeared in all nuclei at anaphase onset in both LO11211 (db2Δ-clbA, n = 31/31 nuclei) and LO11217 (dbΔ-NimECyclin B, n = 38/38 nuclei). We followed these nuclei over long periods of time and found that they stayed in mitosis for a long (although variable) time after cohesin removal/anaphase (the range was 30–300+ min but most were in mitosis more than 1 h). However, eventually SccA-GFP would begin to re-accumulate faintly in condensed nuclei, and the chromosomes of some nuclei would decondense and accumulate SccA-GFP more brightly. Interestingly, we found that mitotic exit did not occur for all nuclei at the same time in the same cell, and some nuclei would remain condensed even if SccA-GFP accumulated (Figs 9P–U, S5P–U). These cells continued to cycle and attempt additional mitoses, and the number of nuclear fragments, likely due to repeated nondisjunction, increased (Figs 9, S5). We also imaged these strains in short, 2-min, intervals and obtained similar results. These data indicate that SccA-GFP is removed from chromosomes during mitosis in cells expressing either db2Δ-ClbA or dbΔ-NimECyclin B and, thus, failure of cohesin complex removal is not the cause of the failure of chromosomal disjunction we observe.

Fig. 9

SccA-GFP is removed from chromatin in db2Δ-ClbA expressing strains. Conidia from strain LO11211 (which carries nmtA(p)-db2Δ-clbA, SccA-GFP, histone H1-mRFP) were collected from 1.0 μM thiamine plates and incubated in liquid medium without thiamine at 30 °C. Images are projections from a time-lapse data set collected at 10-min intervals at 30 °C. At T = 0 min, two nuclei are in G2 with SccA-GFP present in the nucleoplasm, but not in the nucleoli (A–C). Nuclei enter mitosis with SccA-GFP still present in the nucleoplasm (D–F). SccA-GFP leaves chromatin during mitosis (G–I) despite obvious nondisjunction (arrows in G–H). SccA-GFP remains absent from nuclei during a lengthy mitotic block (D–R) (60+ min) until nuclei begin to exit mitosis (M–R). Not all nuclei exit mitosis at the same time (blue arrow in P–Q indicates a nucleus that is slow in exiting). Nuclear fragments and nuclei of different sizes are apparent during interphase (S–U). Fragmented chromatin enters another aberrant mitosis (V–X), and SccA leaves chromatin once more (X).

Sc<span class="Species">cA-GFP is removed from chromatin in db2Δ<span class="Chemical">-ClbA expressing strains. Conidia from strain LO11211 (which carries nmtA(p)-db2Δ-clbA, SccA-GFP, histone H1-mRFP) were collected from 1.0 μM thiamine plates and incubated in liquid medium without thiamine at 30 °C. Images are projections from a time-lapse data set collected at 10-min intervals at 30 °C. At T = 0 min, two nuclei are in G2 with SccA-GFP present in the nucleoplasm, but not in the nucleoli (A–C). Nuclei enter mitosis with SccA-GFP still present in the nucleoplasm (D–F). SccA-GFP leaves chromatin during mitosis (G–I) despite obvious nondisjunction (arrows in G–H). SccA-GFP remains absent from nuclei during a lengthy mitotic block (D–R) (60+ min) until nuclei begin to exit mitosis (M–R). Not all nuclei exit mitosis at the same time (blue arrow in P–Q indicates a nucleus that is slow in exiting). Nuclear fragments and nuclei of different sizes are apparent during interphase (S–U). Fragmented chromatin enters another aberrant mitosis (V–X), and SccA leaves chromatin once more (X).

Expression of db2Δ-ClbA, but not dbΔ-NimECyclin B, extends interphase

Although the mitotic phenotypes <span class="Species">caused by <span class="Species">expression of db2Δ-ClbA and dbΔ-NimECyclin B are similar, we found the effects of their expression on cell cycle timing were very different. To examine the effects of db2Δ-ClbA and dbΔ-NimECyclin B on the cell cycle, we grew strains LO11211 [nmtA(p)-db2Δ-clbA] and LO11217 [nmtA(p)-dbΔ-nimE] on 1.0 μM thiamine, harvested spores, and inoculated them into media with no thiamine. As mentioned above, the thiamine is gradually depleted, and the d-box-deleted cyclins become expressed, resulting in a mitotic block. The presence of a mitotic block indicates that the d-box-deleted cyclin was expressed in the previous cell cycle. By timing the cell cycle prior to a mitotic block, we can determine if the d-box-deleted cyclin alters the length of the cell cycle. In LO11202 (control), we found the average cell cycle duration (time in min from the beginning of one mitosis to the beginning of the next mitosis) at 30 °C for the first 3 nuclear divisions was 150 ± 34 min (n = 63 nuclei). In LO11217 [nmtA(p)-dbΔ-nimE], we found the cell cycle duration was 150 ± 36 (n = 21 nuclei) when followed by a normal mitosis and 155 ± 39 min (n = 35) when followed by an abnormal mitosis. These values were not significantly different from the control. Expression of dbΔ-NimECyclin B, thus, does not significantly alter the length of the cell cycle. LO11211 which expresses db2Δ-ClbA, however, had an average cell cycle time of 183 ± 28 min (n = 49 nuclei) when followed by a normal mitosis and an average cell cycle time of 209 ± 49 min (n = 31 nuclei) when followed by an abnormal mitosis. These were both significantly different from the control (p < 0.001). Expression of db2Δ-ClbA, thus, does increase the length of interphase. We attribute the lengthened cell cycle before normal mitosis to the accumulation of a small amount of db2Δ-ClbA at levels insufficient to arrest mitosis as noted above. These data strongly indicate that ClbA has an inhibitory role in cell cycle progression in interphase.

Discussion

Phylogenetic analyses

Given the signifi<span class="Species">cance of <span class="Gene">cyclins in cell cycle regulation and cell growth, there has been surprisingly little study of cyclins in filamentous ascomycetes. We report the identification and phylogenetic analysis of all cyclin domain-containing proteins in the model filamentous fungus A. nidulans as well as 31 diverse filamentous ascomycetes. Our analyses reveal that cyclins in these species fall into three distinct groups as is the case for other eukaryotes that have been studied (Ma et al., 2013, Cao et al., 2014). Our analyses also revealed that the cyclin repertoires of Aspergilli and the other species in our study are remarkably similar. These species include phylogenetically diverse members of the subdivision Pezizomycotina, which includes most filamentous ascomycetes. It follows that the complement of cyclins is highly conserved throughout filamentous ascomycetes. It is worth noting, however, that there may be functional specializations of cyclins, particularly non-essential cyclins, that do not change the number or phylogenetic relationships of the cyclins.

Group I cyclins

All of the species analysed in this study have three group I <span class="Gene">cyclins that fall into three distinct clades. Two clades cluster more strongly together, and the <span class="Gene">cyclins in these two clades are B-type cyclins. The cyclins in the third clade are more similar to yeast “Cln-like” cyclins than to B-type cyclins. Importantly, each species has only a single member of each of these three clades. Group I cyclins in filamentous ascomycetes are less numerous than in the model yeasts, S. cerevisiae, S. pombe and C. albicans. In particular, S. cerevisiae has nine group I cyclins, none of which are essential for cell cycle progression. The multiplicity of cyclins in S. cerevisiae is likely the consequence of a whole genome duplication (Wolfe & Shields 1997), resulting in functional redundancy.

Group II cyclins

Most of the species we have analysed have seven group II <span class="Gene">cyclins. Four of the seven A. nidulans group II <span class="Gene">cyclins have been functionally characterized to some extent. As their functions may be instructive with respect to group II cyclins in other filamentous ascomycetes, we will briefly summarise the known functions of the A. nidulans group II cyclins.

None of the four group II <span class="Gene">cyclins in A. nidulans (An-<span class="Gene">Pho80, PclA, PclB, and ClgA) are essential, and all appear to play roles in development. An-Pho80 is involved in the negative regulation of phosphate acquisition enzymes and in promoting the switch from asexual to sexual development (Wu ). PclA expression is cell cycle regulated and peaks during S-phase (Schier ). PclA is upregulated during the late stages of sporulation (Schier et al., 2001, Bathe et al., 2010), while PclB is upregulated at both early and late stages of sporulation (Kempf ). Deletion of pclA reduces the number of conidia (Schier ), and the effects of pclAΔ on sporulation are additive with the effects of pclBΔ (Kempf ). PclA has also been shown to bind and activate NimXCdk1 (Schier & Fischer 2002), and it is predicted to play a role in the rapid cell divisions that occur during sporulation (Schier ). Finally, deletion of clgA reduces vegetative growth and sporulation at higher temperatures, while also delaying and repressing the development of cleistothecia (Yu ).

Interestingly, three filamentous ascomycetes have gone through gene dupli<span class="Species">cation events of one or two group II <span class="Gene">cyclin members. Aspergillus carbonarius has two PclB-like proteins. Both P. chrysogenum and P. rubens have three PclA-like proteins and three Pho80-like proteins, while P. digitatum has only one copy of each. We did not find evidence for duplications of group I or group III cyclins in the species analyzed.

Group III cyclins

The majority of the species we have analysed have five group III <span class="Gene">cyclins. Only one group III <span class="Gene">cyclin, PchA, has been functionally characterized in A. nidulans. Deletion of pchA is not lethal, although pchAΔ reduces vegetative growth and production of conidia. Deletion of pchA is synthetically lethal with pclAΔ (Bathe et al., 2010, Kempf et al., 2013).

Functional analyses of group I cyclins

Cln-like cyclin PucA

We have now found that Pu<span class="Species">cA is an essential G1/S <span class="Gene">cyclin in A. nidulans that is required for cells to accumulate NimECyclin B and enter S-phase. We have determined that PucA promotes entry into S phase by inactivating CdhA, the activator of the APC/C, during late mitosis and G1. Although pucAΔ is lethal, deletion of pucA in a cdhAΔ background is viable. This result indicates that PucA is necessary and sufficient to inactivate APC/C-CdhA at the G1/S transition. It also reveals that no other proteins, including other cyclins, can sufficiently inactivate CdhA at the G1/S boundary to allow cell cycle progression. If there were other sufficient mechanisms for CdhA inactivation in A. nidulans, deletion of pucA would not have been lethal. This result is different from results in other organisms that have been studied, as CdhA is typically regulated by several redundant mechanisms. In both humans and S. cerevisiae, multiple cyclin/CDK complexes (e.g. Cln1, Cln2, Clb5, and Clb6 in S. cerevisiae) phosphorylate and inactivate Cdh1 at G1/S. APC/C-Cdh1 can also be inactivated through binding of inhibitors (e.g. EMI1 in vertebrates, Acm1 in S. cerevisiae, and Rca1 in Drosophila melanogaster), but none of these inhibitors have an obvious homolog in A. nidulans. Thus, although Cdh1 inactivation is regulated by multiple proteins in other organisms [reviewed in Sivakumar & Gorbsky (2015)], APC/C-CdhA is essentially regulated at G1/S in A. nidulans by PucA.

Although our evidence indi<span class="Species">cates that deletion of pu<span class="Species">cA causes a G1 arrest, it allows continuous nuclear growth. As pucAΔ nuclei increased in size, we observed a concurrent loss of both nuclear DAPI fluorescence and histone H1 fluorescence. Some pucAΔ nuclei would eventually overcome the G1 block and undergo mitosis, and, in those cases, the histone H1 signal would suddenly become visible as the chromosomes condensed. This indicates that the chromatin is present in interphase nuclei, but it is very diffuse. Interestingly, this diffuse chromatin phenotype has also been observed with a deletion of nimX (De Souza ), and NimXCdk1 has been shown to interact with PucA (De Souza ). It is also important to note that we have worked with other mutants that result in very large, stretched nuclei (e.g. the ɣ-tubulin mutant mipAD159), but these nuclei have very bright histone H1 fluorescence and are obviously polyploid. It is likely that deletion of pucA inhibits DNA replication but not nuclear growth, resulting in large nuclei with diffuse chromatin. It is also possible that this phenotype is caused by a lack of NimX activity, as PucA is a binding partner of NimX and both pucAΔ and nimXΔ result in similar phenotypes.

Pu<span class="Species">cA also has non-essential functions in interphase in addition to inactivating <span class="Disease">APC/C-CdhA at the G1/S transition. In pucAΔ, cdhAΔ double mutant strains we observed nuclear abnormalities such as severe stretching and partial condensation and decondensation of chromatin in a nuclear autonomous fashion. These interesting interphase phenotypes are not due to obvious microtubule abnormalities, inhibition of polarized growth, or failure to accumulate NimECyclin B in the nucleus or at the SPB.

B-type cyclins ClbA and NimECyclin B

Prior to this study, the B-type <span class="Gene">cyclin NimE was the only <span class="Gene">cyclin shown to be essential in A. nidulans and the only cyclin with clear cell cycle regulatory functions. The nimE gene was identified more than 40 years ago (Morris 1975) and found to be required for both S-phase and entry into mitosis (Morris, 1975, Bergen et al., 1984). We have now analysed the functions of a second B-type cyclin, which we have named ClbA.

We have found that <span class="Chemical">ClbA localizes to KTs from mid-G2 until mitotic onset. clbA is not essential, but expression of a version of ClbA with two d-boxes deleted (db2Δ-ClbA) completely blocked colonial growth. To carry out this work, we needed to develop a more tractable repressible promoter system than the alcA system that has been used extensively in A. nidulans. We found that the nmtA promoter was strongly repressible and regulatable with thiamine and this provided us with the ability to determine the phenotypes of db2Δ-ClbA expression via microscopy. We found that expressing db2Δ-ClbA resulted in mitotic catastrophe at anaphase with a high frequency of chromosomal non-disjunction as well as a mitotic arrest. Expression of db2Δ-ClbA-GFP resulted in abundant GFP signals at KTs and nuclei throughout interphase, but, interestingly, we observed that db2Δ-ClbA-GFP was still removed from KTs at mitotic entry. This indicates that ClbA is removed from kinetochores by a mechanism other than destruction. However, db2Δ-ClbA-GFP remained present in the nucleoplasm, and when anaphase occurred, db2Δ-ClbA-GFP could be seen along the spindle and at the poles of nuclei in aberrant anaphases. This was in contrast to full-length ClbA-GFP which was never seen in nuclei or at the poles during anaphase/telophase, whether overexpressed or expressed only from its native promoter. In addition, we found that expression of db2Δ-ClbA-GFP caused a lengthening of interphase. The facts that ClbA-GFP is removed from kinetochores at mitotic entry and that expression of db2Δ-ClbA-GFP causes a lengthening of interphase suggest ClbA might have a role in the G2/M checkpoint in addition to its role mitotic progression.

<span class="Species">Expression of dbΔ-<span class="Chemical">NimECyclin B-GFP has been shown previously to cause a mitotic block in telophase with dbΔ-NimECyclin B-GFP remaining at the poles (De Souza et al., 2009, Nayak et al., 2010). Our use of the nmtA promoter to regulate expression of dbΔ-NimECyclin B-GFP has allowed us to determine that its expression also causes a high frequency of non-disjunction. Destruction of both ClbA and NimECyclin B are, thus, required for chromosomal disjunction in mitosis.

We also determined that the high frequency of chromosomal nondisjunction observed in both db2Δ<span class="Chemical">-ClbA-GFP and dbΔ-<span class="Chemical">NimECyclin B-GFP cells was not caused by a failure of removal of the cohesin complex. For disjunction to occur successfully, two processes must occur. The cohesin complex must be removed and catenanes (DNA intertwinings) of the daughter chromatids must be resolved by topoisomerase II. Since the cohesin complex is removed in strains expressing db2Δ-ClbA or dbΔ-NimECyclin B, it follows that destruction of both ClbA and NimECyclin B are required for DNA decatenation, presumably through regulation of topoisomerase II. Indeed, the phenotypes we observe are consistent with a failure of decatenation. Unlike the cohesin complex, catenanes cannot resist the pulling forces of microtubules [reviewed in Haarhuis ]. A failure to resolve catenanes would not block anaphase but is predicted to lead to chromosome stretching and failure of disjunction, as we have observed.

Summary of cell cycle regulation by group I cyclins in filamentous ascomycetes

Now that the three group I <span class="Gene">cyclins have been substantially characterized in A. nidulans, it is clear that they have distinct, non-redundant functions in cell cycle regulation. Our current findings, and previous findings from other labs, greatly clarify the functions of group I <span class="Gene">cyclins in cell cycle regulation in A. nidulans, and we will now summarise what we know about those functions. We anticipate, given the conservation of group I cyclins, that many of these findings will apply to other filamentous ascomycetes.

Starting our summary at the G1/S transition, Pu<span class="Species">cA activity is required to inactivate <span class="Disease">APC/C-CdhA and allow accumulation of NimECyclin B and consequent entry into S-phase. NimECyclin B is required for S-phase and for entry into mitosis. Non-degradable ClbA, but not non-degradable NimECyclin B, causes a delay in interphase, so destruction of ClbA is important for normal progression through interphase. This fact, along with the fact that ClbA is removed from kinetochores at mitotic onset, leads us to speculate that ClbA may play an inhibitory role in the G2-to-M checkpoint. Destruction of both ClbA and NimECyclin B during mitosis is required for chromosomal disjunction (likely through topoisomerase II resolution of catenanes) and removal of NimECyclin B from the SPB (which is normally brought about by destruction of NimECyclin B earlier in mitosis) is required for the M-to-G1 transition.