Fission Yeast CSL Transcription Factors: Mapping Their Target Genes and Biological Roles

Abstract

Background: Cbf11 and Cbf12, the fission yeast CSL transcription factors, have been implicated in the regulation of cell-cycle progression, but no specific roles have been described and their target genes have been only partially mapped.

Methodology/principal findings: Using a combination of transcriptome profiling under various conditions and genome-wide analysis of CSL-DNA interactions, we identify genes regulated directly and indirectly by CSL proteins in fission yeast. We show that the expression of stress-response genes and genes that are expressed periodically during the cell cycle is deregulated upon genetic manipulation of cbf11 and/or cbf12. Accordingly, the coordination of mitosis and cytokinesis is perturbed in cells with genetically manipulated CSL protein levels, together with other specific defects in cell-cycle progression. Cbf11 activity is nutrient-dependent and Δcbf11-associated defects are mitigated by inactivation of the protein kinase A (Pka1) and stress-activated MAP kinase (Sty1p38) pathways. Furthermore, Cbf11 directly regulates a set of lipid metabolism genes and Δcbf11 cells feature a stark decrease in the number of storage lipid droplets.

Conclusions/significance: Our results provide a framework for a more detailed understanding of the role of CSL proteins in the regulation of cell-cycle progression in fission yeast.

Fig 1. Genes regulated by Cbf11 and Cbf12.

(A) Heatmap of expression ratios of DEGs in CSL knock-outs under several conditions (‘LOG’–exponential growth; ‘STAT’–stationary phase) and during CSL overexpression (‘OE’; 12 or 18 hours post induction). Different data columns under the same condition represent independent biological repeats. The mRNA levels at each condition relative to the levels in wild-type control cells are colour-coded as indicated at bottom, with missing data in grey. CSL binding to the upstream intergenic regions of the respective DEGs, as detected by ChIP-seq from cells growing exponentially in YES, and periodicity of gene expression during cell cycle [47] are indicated at right (dark bars signify CSL binding/periodic expression). Six major DEG clusters (‘1’-‘5a/b’) are indicated by the right-most colour bars. (B) Functional enrichment analysis of the DEG clusters from (A). Numbers in the matrix represent fold enrichment in the cluster compared to all other genes, and the enrichment significance is denoted by the colour of the cell background. ‘N starvation’–genes induced upon nitrogen removal [48]; ‘late meiotic’–genes induced after meiotic divisions [48]; ‘CESR’–core environmental stress response genes [49]; ‘top 500 periodic’–top-ranking 500 genes expressed periodically during cell cycle [47]; ‘transmembrane’–genes coding for transmembrane proteins [50]; ‘GPI anchor’–genes coding for GPI-anchored proteins [51]. (C) Venn diagrams showing overlaps between all CSL DEGs and genes with CSL binding in their upstream intergenic regions, as detected by ChIP-seq from cells growing exponentially in YES. The p-values (one-sided Fisher's exact test) for significance of overlap are indicated.

Fig 2. Subset of cell cycle-regulated genes show altered expression upon CSL manipulation.

Heatmap of expression ratios of DEGs (as in Fig 1A) that belong to top-ranking 500 genes expressed periodically during the cell cycle [47]. CSL binding to promoters of respective DEGs, as detected by ChIP-seq from cells growing exponentially in YES, is indicated at right (dark bars signify CSL binding). Cluster membership is indicated by right-most colour bars (colour coding as in Fig 1A).

Fig 3. Cbf11 and Cbf12 affect multiple aspects of cell division cycle.

(A) CSL knock-out cells grown in YES were fixed and stained with DAPI (nuclei). Overall, 10–20% cells lacking cbf11 underwent catastrophic mitosis (‘cut’ phenotype, denoted by asterisks). (B) CSL knock-out cells (YES medium) or (C) CSL-overexpressing cells (EMM medium) were fixed and stained with calcofluor (B; septum and cell wall) or calcofluor and DAPI (C; septum, cell wall and nucleus). Note the multiple septa in Δcbf11 and Δcbf11 Δcbf12 cells (B), and septum formation in absence of nuclear division in cells overexpressing cbf11 or cbf12 (‘cut’ phenotype; asterisks) (C). The fractions of dividing cells (septation index; %) were increased upon cbf11 and cbf12 overexpression (n >1000 cells). (D) The length of fully septated cells from (B, C) was measured (orange, red). Each dot represents a single cell measurement; median length and n values are indicated above each distribution; WT/control median values are indicated as black horizontal lines. Deletion of cbf11 resulted in decreased cell length at division. Severely shortened cell length at division was observed in a fraction of cells overexpressing cbf11. Cells overexpressing cbf12 displayed gross deregulation of cell size at division.

Fig 4. cbf11 interacts genetically with sty1 and pka1 pathways.

(A) 10-fold serial dilutions of cultures with the indicated genotypes were spotted on YES plates and grown for 2 days. The slow growth phenotype of Δcbf11 cells is strongly and moderately suppressed by the deletion of sty1 and pka1, respectively. (B) Cells growing exponentially in YES were fixed and stained with DAPI. Nuclear integrity defects of Δcbf11 cells (marked with '#') are diminished by deletion of sty1 or pka1. The DAPI signal was overlaid with the corresponding DIC image to visualize cell contours. Scale bar 10 μm. (C) Calcofluor staining documents that the occurrence of Δcbf11-associated septation defects (e.g., single cells with multiple septa) is decreased in the double mutants with sty1 and pka1. Multicellular filaments are marked by arrows. Scale bar 10 μm. (D) Flow cytometry analysis of DNA content in fixed, propidium iodide-stained cells grown to the exponential phase in YES. Deletion of cbf11 results in aberrant DNA content distribution, which is corrected by deletion of sty1 and, in part, pka1. Fractions of cells with <2C, 2C, and >2C DNA content are indicated in the histograms. (E) The length of fully septated cells from (C) was measured. Deletions of both sty1 and pka1 have marked influence on the length of Δcbf11 cells. Each dot represents a single cell measurement; median length and n values are indicated above each distribution; WT median value is indicated as a black horizontal line. Data points corresponding to short, multicellular filaments and their n values are shown in blue.

Fig 5. Cbf11 protein expression and defects of Δcbf11 cells are diminished in minimal medium.

(A) Growth curves of WT and Δcbf11 cultures in different liquid media show that the slow growth of Δcbf11 cells observed in YES is partially suppressed in EMM (i.e., curve slope is increased in EMM). The noise appearing during the Δcbf11 lag phase in EMM was caused by cell flocculation. (B) Doubling times of exponentially growing WT and Δcbf11 cultures in the indicated media. The addition of EMM to YES causes dose-dependent decrease in doubling time of Δcbf11 cells. (C) Microscopy of fixed, DAPI-stained wild-type and Δcbf11 cells grown to exponential phase in YES or EMM media. Cells lacking cbf11 display heterogeneous morphology, cell separation defects and the ‘cut’ phenotype when grown in YES. These mutant phenotypes are largely absent from cells grown in EMM. (D) Quantification of the occurrence of the ‘cut’ phenotype in cells from panel (C). Mean values ± SD from three independent biological repeats are shown (n > 200 cells). (E) Flow cytometry analysis of DNA content in wild-type and Δcbf11 cells growing exponentially in YES or EMM. The broad signal distribution in Δcbf11 cells from rich medium is narrowed towards wild-type values when cells were grown in EMM (n > 15,000 cells). (F) EMSA assay of fission yeast cell extracts. When cells were grown in EMM, the DNA binding activity of Cbf11 decreased by ~40–80% compared to YES. Lane 1: no cell extract added; lane 2: cell extract from Δcbf11 Δcbf12 cells; lanes 3–4: wild-type cells; lanes 5–6, samples from cells expressing a chromosomally TAP-tagged version of Cbf11. ‘w’ and ‘p’ denote the position of wells and free probe, respectively. The arrowheads mark bands corresponding to the DNA binding activity of Cbf11. The bar chart at bottom shows the quantification of bandshift intensities in the respective lanes. Irrelevant gel lanes were omitted for clarity. (G) Western blot detection of Cbf11-TAP in cell extracts from panel F (lanes 5–6, 3–4) showing decreased Cbf11 protein amounts in cells grown in EMM as compared to YES. As a loading control, the blots were probed with an anti-PSTAIRE (Cdc2) antibody. Representative examples of 3 biological repeats are shown in (F, G).

Fig 6. Putative Cbf11-regulated lipid metabolism genes.

(A) The mRNA levels at each condition relative to the levels in wild-type control cells are colour-coded as indicated at top right, with missing data in grey; for description of cultivation conditions see legend to Fig 1A. (B) The sequence motif identified by MEME-ChIP [65] in the Cbf11-bound promoter regions of genes from (A) resembles closely the canonical metazoan CSL response element, ‘Su(H)’. (C) Previous transcriptome profiling of the genes from (A) over two cell cycles of elutriation-synchronized wild-type cultures suggest moderate periodic oscillations in the expression of some of these genes. Data taken from [2]. (D) Probes used for EMSA experiments (sense strand shown); CSL response elements are in bold and introduced mutations are in lower-case and underlined. ‘RBP’, ‘KSHV’, ‘MUT’ and ‘DEL’ probes were derived from metazoan/viral CSL-responsive genes [22]. (E) Representative competitive EMSAs with wild-type cell extracts and radioactively labelled RBP probe; unlabelled cut6, ptl1 and RBP probes were used as competitors (5× and 20× excess). The decrease in band intensity reflects strength of binding of unlabelled competitor probes. (F) Quantification of competitive EMSA experiments for 5× and 20× competitor excess. Mean values ± SD are shown. (G) Representative examples of live WT and Δcbf11 cells stained with Nile red to visualize neutral lipid droplets. Scale bar 5 μm. (H) Numbers of lipid droplets per cell were normalized to cell volume. Each dot corresponds to one cell (n ≥ 200); data for three independent repeats are shown.