Cyclin-Dependent Kinase Inhibits Reinitiation of a Normal S-Phase Program during G₂ in Fission Yeast

Strains and growth conditions.Standard growth conditions and methods were used (30); all experiments were performed in EMM4S (Edinburgh minimal medium with supplements) unless otherwise stated. 5-Bromo-2- deoxyuridine (BrdU) pulse-labeling was performed as described previously (40) using 10-min pulses of 300 μM BrdU. SPAP14E8.02 was tagged with green fluorescent protein (GFP) (S65T) at the C terminus using a PCR-based method (4) and is referred to as tos4-GFP. tos4-GFP strains were grown in EMM4S with 750 mg/liter adenine to reduce autofluorescence (P. Perez, personal communication). cdc25-22^tsade6-M210 tos4-GFP cells grown to 1.2 × 10⁶/ml were synchronized by a shift to 36.5°C to block cells in G₂, with cells then released synchronously into the cell cycle at 25°C. To induce rereplication in cdc13 switch-off (s/o) for pulse-labeling, Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 leu1 adh1-hENT his adh1-tk (from Atanas Kaykov) cells grown at 25°C to 3.6 × 10⁶/ml were filtered, washed four times with Edinburgh minimal medium without NH₄Cl (EMM-N), and resuspended at 2.4 × 10⁶/ml in EMM-N. After 12 h, 5 μg/ml thiamine was added to the culture; 8 h later, NH₄Cl was added back to the culture with all supplements and the culture was shifted to 32°C to begin the time course. The same protocol was used for Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 tos4-GFP cells. To impose a cdc10 block and release in endoreduplicating cells, rereplication was induced as described above in cdc10-v50 Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 tos4-GFP cells at 25°C. Once cells had begun endoreduplication, cells were shifted to 36.5°C at 6 h; they were then released synchronously into endoreduplication to 25°C at 9 h. For a cdc10 block without rereplication, a culture in the absence of thiamine was grown to 1.2 × 10⁶/ml and shifted to 36.5°C. For origin mapping studies, rereplication was induced in Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 leu1-32ura4-D18 cells by 5 h of nitrogen starvation at 25°C in the presence of thiamine (5 μg/ml), followed by a shift to 32°C with or without 11 mM hydroxyurea (HU).

Microscopy and cell measurements.Microscopy was performed on a Zeiss Axioplan 2 microscope using a 63× objective, Photometrics Coolsnap HQ camera, and Metamorph software, and cell length and width measurements were taken in the ImageJ software program. Cell volume (fl) was calculated by estimating the shape of the cell as a cylinder with a half-sphere at each end. Cells from the 0- to 3-h time points in the cdc13 s/o pulse-labeling experiment were excluded from analysis, since the BrdU signal was not reliably detectable in those cells, potentially due to prior nitrogen starvation. To find the cell volume per nucleus, half of the cell volume for binucleate and septated cells was associated with each nucleus. All cell volumes were binned in discrete intervals, and the percentage of nuclei which incorporated BrdU in each volume cohort was determined. Experiments were performed at least in triplicate unless otherwise noted; representative data are shown.

Microarray experiments.Microarray design, DNA preparation, microarray hybridizations, and data acquisition and analysis were as described for HU experiments in reference 19. Data analysis was based on the genome sequence from June 2006. This and current sequence data can be obtained from the Sanger Institute ftp server at ftp://ftp.sanger.ac.uk/pub/yeast/pombe/Chromosome_contigs/ . At least two biological repeats were carried out for each microarray experiment. In some cases, signal ratios for wild-type mitotic S phase have been reassigned in mapping the 904 origins, so efficiencies for 3.8% (34/904) of origins deviate from those in references 19) and 25 (31/904). The cdc13 experiment without HU was performed twice using open reading frame arrays and once in conjunction with tiling arrays; the reference DNA from G₁-arrested cells was also used for the self-self control. The reference DNA for all other microarray experiments was from cells blocked in G₂ with 2C DNA content. Data from two repeats were averaged for origin mapping in cdc13 s/o. Percent AT content was defined as the highest percent AT in a 500-bp window (50-bp step) in the intergenic region with at least 65% AT richness that maps to or close to the origin. Intergenic length refers to the number of base pairs in the intergenic region to which the AT-rich island of the origin maps. Efficiency of origins was determined as in reference 19.

Analysis of prereplicative complex (pre-RC) colocalization with origins.A custom-made Perl script was used to map signal log₂ ratios from genome-wide chromatin immunoprecipitation-on-chip analysis by Hayashi et al. (16) to each origin in our study. Briefly, among probes with P values which conform to the criteria described by Hayashi et al. (>0.025), the script detects the peak signal ratio in the vicinity of the origins by performing iterative searches for a defined threshold within an increasing range from the origin. The data from this analysis using a cutoff signal log₂ ratio of 0.5 and a range of 4 kb about the origin were pooled (from highest to lowest threshold). Our second, more stringent analysis followed a cutoff of signal log₂ ratio greater than 0.8, as in the work of Hayashi et al. Further details are available upon request.

Microarray data accession number.Microarray data are available at Array Express under accession no. E-MTAB-105.

Single-cell assays for DNA synthesis and G₁/S gene expression.A normal mitotic S phase in fission yeast is characterized by the following: (i) a burst of G₁/S Mlu1 cell cycle box binding factor (MBF)-dependent gene expression, (ii) the initiation of DNA synthesis at a critical cell size, (iii) the use of a specific set of replication origins, and (iv) equal replication across the genome. To investigate whether the DNA synthesis induced when G₂/M CDK is depleted from G₂ cells exhibits these characteristics, we have used single-cell assays for DNA synthesis and G₁/S gene expression.

DNA synthesis was monitored in single cells by BrdU pulse-labeling using a strain that expresses the human nucleoside transporter (hENT) and herpes simplex thymidine kinase (tk). This allows the uptake and conversion of a nucleoside analog label to a nucleotide label and its incorporation into cells undergoing DNA synthesis. Visualization of newly replicated DNA is monitored after fixation by immunofluorescence (40). To validate this method, an asynchronous wild-type culture was pulsed with BrdU for 10 min and signal was detected in septated and recently divided cells which were in S phase (Fig. 1A). Since cell cycle progression is coupled to growth, we measured the cell volume per nucleus of labeled fixed cells, considering binucleate and septated cells as two nuclei within a single cell (see Materials and Methods). The percentage of BrdU-positive cells reached a maximum at about 44 fl (cells shrink due to fixation and processing for immunofluorescence and so are smaller than live cells), and the peak was asymmetrical (Fig. 1A and Table 1). The smallest cell volume per nucleus at which cells were labeled was observed for the cohort of 37 fl, where 50% of the population already was labeled. The 37-fl cohort consisted of binucleate cells which had just completed mitosis. This result is consistent with the fact that there is only a very short G₁ phase in wild-type fission yeast (28) and S phase is initiated soon after completion of mitosis in binucleate cells.

• Open in new tab
• Download powerpoint

FIG. 1.

Single-cell methods for studying DNA synthesis and MBF-dependent gene expression. (A) BrdU pulse-labeling of wild-type and wee1-50 cells. Left panels: a leu1-32 his7-366 pJL218 (his7 adh1-tk) pFS181 (leu1 adh1-hENT1) culture at 32°C was pulse-labeled with BrdU. BrdU was detected by immunofluorescence and DNA visualized by 4′,6′-diamidino-2-phenylindole (DAPI) staining. The scale bar is 5 μm long. The percentage of BrdU-positive nuclei was plotted as a function of fixed cell volume per nucleus. Right panel: a wee1-50 leu1-32 his7-366 pJL218 (his7 adh1-tk) pFS181 (leu1 adh1-hENT1) culture at 25°C was shifted to 32°C for 6 h. Cells were pulse-labeled with BrdU. (B) MBF-dependent gene expression in wild-type and wee1-50 cells. The cell volume per nucleus with peak Tos4-GFP signal was determined. Left panels: an ade6-M210 tos4-GFP::KanMx6 culture at 32°C. The percentage of nuclei with a strong GFP signal was plotted as a function of cell volume per nucleus. Right panel: a wee1-50 ade6-M210 tos4-GFP::KanMx6 culture at 25°C was shifted to 32°C for 6 h. The percentage of nuclei with a strong Tos4-GFP signal was plotted as a function of cell volume per nucleus. (C) FACS analysis of wild-type cells at 32°C and wee1-50 cells shifted from 25°C to 32°C for 6 h. Wee1-50 cells show a 1C peak. (D) Experiment schematic: a cdc25-22 ade6-M210 tos4-GFP::KanMx6 culture at 25°C was shifted to 36.5°C for 4 h and released at 25°C. (E) FACS analysis: cdc25-22 tos4-GFP cycled similarly to cdc25-22 alone. (F) The percentage of nuclei with strong Tos4-GFP signal was plotted as a function of time, peaking at the end of S phase. Tos4-GFP localizes to the nucleus periodically during the cell cycle. (G) A cycling cdc10-v50 Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 tos4-GFP::KanMx6 culture in the absence of thiamine at 25°C was shifted to 36.5°C for 4 h. The percentage of nuclei in the culture with a strong Tos4-GFP signal was assessed every hour. (H) Rereplication was induced in a cdc10-v50 Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 tos4-GFP::KanMx6 cul- ture. At 6 h after refeeding nitrogen, half of the culture was shifted to 36.5°C for 3 h (gray line, block) while half continued to rereplicate at a permissive temperature (dashed line, control). At 9 h, half of the blocked culture was released to 25°C (black line, block release). The percentage of nuclei with Tos4-GFP was determined at each time point.

In the normal wild-type mitotic cell cycle, the critical cell size threshold for entry into S phase is cryptic, since newly divided cells are above the size threshold (31), so the critical size is revealed only in a wee1 mutant which reduces cell size at division. In cells lacking Wee1, newly divided cells are small and G₁ is lengthened, since cells must grow to reach the minimum size required for S phase. Therefore, we used temperature-sensitive wee1-50 cells which are advanced into mitosis to determine if the single-cell assay could detect the reduced cell size for S phase in these cells (31). Wee1-50 cells growing exponentially at 32°C have a longer G₁ phase than wild-type cells (data not shown and Fig. 1C) and undergo S phase at a reduced cell volume (Fig. 1A), and as expected, the cell volume cohort with maximal labeling was smaller than wild-type cells at about 28 fl (Fig. 1A and Table 1). We conclude that this assay can be used to monitor DNA synthesis and to determine the cell volume at which cells undergo DNA synthesis.

We next developed a single-cell assay to monitor G₁/S MBF-dependent gene expression in live cells. Factors required for DNA replication expressed at G₁/S include Cdc18, Cdt1, Cig2, and Cdc22, under transcriptional control of MBF, which contains Cdc10, Res1, Res2, and Rep2 (reviewed in reference 3). Since these MBF-dependent gene products are thought not to give an easily detectable signal in live cells when fluorescently labeled (S. Kearsey, personal communication), we used the nonessential gene SPAP14E8.02, related to TOS4 of budding yeast, as a marker for G₁-S gene expression. This gene is periodically expressed in the cell cycle, peaking at G₁ (36). It has MCB1 and MCB2 (Mlu1 cell cycle box) motifs, bound by MBF (36). We constructed a strain in which endogenous SPAP14E8.02 was tagged with a single GFP (S65T) at its C terminus, and we refer to the construct here as tos4-GFP. To determine when Tos4-GFP signal peaks in the cell cycle, a culture was synchronized using the cdc25-22 temperature-sensitive mutation (13) (Fig. 1D). Cells were blocked in G₂ at the restrictive temperature of 36.5°C and then released at 25°C, allowing the population to progress synchronously through the cell cycle. The percentage of cells with GFP signal was monitored over two cell cycles, and DNA content was assessed by FACS analysis (Fig. 1E). Fluorescent protein was seen as a high-intensity signal localized to the nucleus in S phase which decreased in G₂ (Fig. 1F). This was consistent with a peak in message at G₁ (36) and the ∼25-min time period needed for GFP (S65T) maturation, which delays the appearance of the protein until later in the cell cycle (19). To verify that tos4-GFP is MBF dependent, an exponentially growing culture of cdc10-v50 tos4-GFP cells at 25°C was shifted to 36.5°C to inactivate the temperature-sensitive mutant Cdc10 (26, 32). As cells accumulated in early G₁ over 4 h, the Tos4-GFP signal disappeared, indicating that tos4 expression depends on Cdc10 and thus the MBF complex (Fig. 1G). In contrast, the Tos4-GFP signal was readily observed in control cells with active MBF at 36.5°C (data not shown). A wild-type strain with tos4-GFP had a normal cell size, generation time, and DNA content (data not shown). The fluorescent signal was seen in the nuclei of septated cells which are in S phase and recently divided cells which are in S phase and early G₂ (Fig. 1B). The smallest cell volume (per nucleus) was represented by the 90-fl cohort, which was already 60% labeled (Fig. 1B and Table 1). Thus, Tos4-GFP levels oscillate in the normal cell cycle, and the appearance of the protein depends upon MBF. Labeling then gradually declined, in agreement with peak tos4-GFP transcription occurring immediately after mitosis. There was a peak in wee1-50 cells at 75 fl; this smaller size was as expected, given that wee1-50 cells replicate at a smaller size than wild-type cells (Fig. 1B and Table 1). We conclude that the Tos4-GFP fusion can be used to monitor MBF-dependent gene expression and the cell size at which it occurs.

DNA synthesis and MBF-dependent gene expression in the cdc13 switch-off strain.Cdc13 is the G₂/M cyclin B which associates with the protein kinase Cdc2 to form the G₂/M CDK in fission yeast cells. In the cdc13 s/o strain, the cdc13 gene is under the control of the thiamine-repressible nmt41 promoter. Cdc2-Cdc13 CDK is depleted after thiamine addition, leading to DNA synthesis in G₂ cells (14, 17). Previous experiments with the cdc13 s/o strain established that continued DNA synthesis in the absence of mitosis occurs most efficiently in cells which have been nitrogen starved, treated with thiamine, and then returned to growth medium in the presence of thiamine so that cells lack Cdc13 when they reach G₂ (J. Hayles, personal communication) (14) (see Materials and Methods). During the subsequent 11 h of growth, the DNA content in the majority of cells increased from 2C to 32C. To determine if DNA synthesis occurred in discrete intervals that correlated with cell volume increase, samples were taken from the cdc13 s/o tk hENT culture every hour and pulsed with BrdU. Cells from each time point were examined for BrdU incorporation into the nucleus (Fig. 2A). The percentage of labeled cells remained essentially constant over time (Fig. 2B) (Materials and Methods). When cells from all time points were ordered according to volume, similar to the assay shown in Fig. 1A, three peaks of DNA synthesis and a partial fourth peak were revealed (Fig. 2C). DNA synthesis peaked at fixed cell volumes of 50, 100, and 200 fl, with the partial fourth peak at 400 fl (Fig. 2C and Table 1). These data indicate that the cdc13 s/o cells undergo discrete intervals of DNA synthesis in the absence of mitosis and do so periodically approximately with each cell volume doubling. Comparison of the cell size data and FACS profile from each time point showed, respectively, that the peak at 50 fl corresponded to the transition from 2C to 4C, 100 fl to the 4C-to-8C transition, 200 fl to the 8C-to-16C transition, and 400 fl to the 16C-to-32C transition (Fig. 2 and Table 1). These volumes are multiples of about 25 fl, roughly the size at which cells lacking Wee1 activity undergo S phase. We conclude that DNA synthesis is periodic and is correlated to cell size doubling in the cdc13 switch-off strain. The fact that reinitiation of DNA synthesis occurred roughly at multiples of the normal minimal size for S phase suggests that a dependence of initiation upon cell size increase is maintained.

• Open in new tab
• Download powerpoint

FIG. 2.

Periods of DNA synthesis correspond to cell volume doublings in cdc13 switch-off. (A) BrdU pulse-labeling of cells lacking Cdc13. An exponentially growing Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 leu1 adh1-hENT his adh1-tk culture was filtered and resuspended in medium lacking nitrogen. After 8 h, thiamine was added, and after an additional 12 h, nitrogen was replenished and the culture shifted to 32°C. Cells depleted of the Cdc13 cyclin were pulse-labeled with BrdU during a time course as cells underwent DNA doublings from 1C to 32C. DNA was visualized by 4′,6′-diamidino-2-phenylindole (DAPI) staining, and BrdU was detected by indirect immunofluorescence. Cells from 13 h after replenishment of nitrogen are shown. The scale bar is 5 μm long. (B) The percentage of BrdU-positive cells was plotted as a function of time. (C) Cells from all time points were binned according to fixed cell volume, and the percentage of cells with incorporated BrdU in each bin was plotted as a function of cell volume. Peaks in the percentage of BrdU-positive cells corresponded to DNA doublings.

To determine whether G₁-to-S-phase MBF-dependent gene expression was also periodic and accompanied the periodic DNA synthesis, we used the cdc13 s/o strain containing the tos4-GFP marker. We first confirmed that Tos4-GFP was MBF dependent in the absence of Cdc13 by inducing rereplication in a cdc13 s/o tos4-GFP strain containing the cdc10-v50 mutation. A rereplicating culture was shifted to 36.5°C, and the Tos4-GFP signal was observed to disappear as cells blocked in G₁; upon release to 25°C, Tos4-GFP reappeared, demonstrating its dependence on MBF in rereplicating cells (Fig. 1H). During the experimental time course, nuclear Tos4-GFP appeared periodically during successive DNA doublings and four periods of nuclear Tos4-GFP signal coincided with the four DNA doublings (Fig. 3 and Table 1). Therefore, there are periodic rounds of both DNA replication and G₁/S transcription during the time course. These data also indicate that MBF-dependent gene expression does not require passage through mitosis. We conclude that in the cdc13 s/o strain, G₁/S-type MBF-dependent gene expression is periodic and each round of DNA replication is associated with a period of gene expression. These characteristics indicate that a normal S-phase replication program is induced from G₂ when the G₂/M CDK is removed.

• Open in new tab
• Download powerpoint

FIG. 3.

MBF-dependent gene expression is periodic in cdc13 switch-off. Nuclear signal of Tos4-GFP in the absence of cdc13. Left panel: Δcdc13::ura4⁺pREP45 cdc13⁺ade6-704 tos4-GFP::KanMx6 cells depleted of Cdc13. DNA content increased to 16C. Right panel: the percentage of cells with strong nuclear GFP as a function of cell volume.

DNA replication across the genome and replication origins.In a normal mitotic S phase, the genome is uniformly replicated with no areas amplified relative to other regions, so there is an equal copy number of each portion of the genome. To determine whether equal rounds of replication or local amplification occurred in the cdc13 s/o strain, we assayed genomic DNA from cells which had increased their DNA content to 16C to 32C. Using microarrays, we measured the relative DNA content across the genome, using DNA from control G₁- arrested cells as a reference. This experiment revealed that replication was essentially equal across the genome, with no region becoming significantly amplified to a higher copy number than any other (Fig. 4). The only possible exceptions are regions very close to the telomeres, where there are slight deviations from the self-self control (Fig. 4). We conclude that although S phase occurs in the absence of mitosis, replication is restricted to discrete and essentially full rounds of genome doubling. This establishes cdc13 s/o as a system of endoreduplication wherein the genome is duplicated in complete rounds without mitosis as opposed to rereplication or amplification in which aberrant refiring of origins leads to partial reduplication of the genome (reviewed in references 1 and 11).

• Open in new tab
• Download powerpoint

FIG. 4.

Equal replication across the genome in a cdc13 switch-off strain. DNA from rereplicating cells which had attained 16C and 32C DNA contents in the absence of HU was hybridized against a reference to microarrays. Signal was normalized to the rest of the genome.

Since the genome was found to be essentially evenly replicated, as would be expected if each round of replication corresponded to a normal S phase, we investigated whether this was the result of a normal replication program at the level of origin firing. We asked whether S-phase origins are used to reduplicate the genome and whether origins are used with the same efficiency as in wild-type cells. The replication origin profiles from the fission yeast wild-type cell cycle were identified in our previous study (18) using synchronous cell populations treated with HU, which confines DNA synthesis to the vicinity of fired origins (see Materials and Methods). We remapped the 904 mitotic S-phase origins (401 “strong” origins and 503 “putative weaker” origins described by Heichinger et al.), regarding origins within clusters as distinct origins, and assigned each origin the highest signal ratio value around the AT-rich island to which it maps (see Table SI in the supplemental material; also see Materials and Methods). We identified the origins utilized in the first endoreduplication cycle of cdc13 s/o using microarray analyses of cells treated with 11 mM HU. Samples were taken at 5 h (Fig. 5A), when most cells of the culture not treated with HU had undergone a doubling in DNA content. Examples of the microarray profiles are given in Fig. 5B for two regions of the genome, where peaks in the signal ratio can be observed which mark the locations of fired origins. These data are compared to data from wild-type cells undergoing replication in a normal cell cycle to determine if the same origins are used.

• Open in new tab
• Download powerpoint

FIG. 5.

Replication origin usage in a cdc13 cyclin switch-off. Mapping origins in endoreduplication of a cdc13 cyclin switch-off mutant. (A) Left panel: schematic of the cdc13 cyclin switch-off time course experiment. Cells were nitrogen starved for 5 h at 25°C in the presence of thiamine. Nitrogen was replenished and the culture shifted to 32°C, and 11 mM HU was added. Right panel: FACS analysis of the cdc13 switch-off mutant, induced into endoreduplication with or without the addition of 11 mM HU. In the presence of HU, cells block at the onset of the first round of endoreduplication. In the absence of HU, cells undergo repeated rounds of replication without intervening mitoses. (B) Comparison of wild-type and cdc13 switch-off replication profiles from the average for two HU experiments. Examples of origins: arrows mark origins which are used both in a normal S phase and in the cdc13 s/o strain, gray circles mark normal S-phase origins not used in cdc13 s/o, and asterisks mark new origins which are used only in the cdc13 s/o strain. The regions were selected for illustration because they contain all three types of origins.

A total of 799 origins were identified, a number roughly similar to the 904 identified in a normal S phase (Table 2; see Table SI in the supplemental material). Of these 799 origins, 697 are active in a normal S phase and 102 are not used in a normal S phase, while 207 normal S-phase origins are not fired in cdc13 s/o (Table 3). The percentage of AT content (see Materials and Methods) and the intergenic length, both of which are known to be correlated with origin strength (9), did not differ on average among these groups (Tables 2 and 3). The mean AT contents were similar, between 72.1% and 74.0% (intergenic regions in the genome contain ∼70% AT on average [9, 38]), and the mean AT content for all origins activated in cdc13 s/o was identical to that in S phase (73.8%) (Table 2). The lengths of the intergenic regions in which the origins are embedded were also similar in inactive and active mitotic origins, with average lengths ranging from 1,349 to 1,776 bp (average intergenic distance in the genome overall is 960 bp). As previously described, AT-rich sequences acting as replication origins are preferentially located in intergenic regions of divergent transcription (38), which are longer than the average S. pombe intergene region (42), and a similar bias was reflected in the wild-type and cdc13 s/o origins. We also found that there was no correlation of activated or suppressed origins with loci of cell cycle-regulated genes. Although many of the same origins were used, their efficiency in cdc13 s/o was not well correlated to wild-type efficiency (r² value of 0.3). We conclude that origins used in cdc13 s/o are mostly similar to those of a normal S phase, although with changes in efficiency.

We examined whether the new origins activated in cdc13 s/o were sites on the chromosome where pre-RCs assemble. The new origins mapped were compared with the previously reported genome-wide localization of Orc4, a component of the pre-RC, in G₁ (16) (see Table SI in the supplemental material). We devised a script to search the data set of Hayashi et al. for Orc4 localization in the vicinity of our origins using two signal ratio thresholds (see Materials and Methods). With the lower threshold, 96% of the 401 strong origins reported in the work of Heichinger et al. (18) had Orc4 localization, and 69% of the 503 weaker origins reported here were colocalized with Orc4. These values should be compared with the 74% of the 102 cdc13 s/o-activated origins which were colocalized with Orc4. Applying a higher threshold similar to that of Hayashi et al. (16) (see Materials and Methods), 92% of the strong wild-type origins colocalized with Orc4 while 50% of the activated cdc13 s/o origins and 50% of weaker wild-type origins showed Orc4 binding. We conclude that the extent of Orc4 binding (reflecting pre-RC formation) to the new origins activated in cdc13 s/o was similar to the weaker normal S-phase origins but less than the stronger normal S-phase origins.