pol30 mutations lead to transient de-repression of URA3 at the VIIL sub-telomere

The irreversible 5-FOA resistance and CRASH assays [12, 20] do not distinguish between events of transient destabilization of gene silencing or a complete Silent→Active (S→A) conversion of a gene. Recently, we produced two dual adh4-URA3-HTB1→yEGFP-tel and adh4-URA3-yEGFP←HTB1-tel reporters for the analyses of gene silencing at the VIIL telomere [18, 19]. These reporters contain the URA3 gene followed by fusion HTB1-yEGFP reporter driven by the Histone H2B promoter (HTB1) in the orientation towards or away from the telomere, respectively (Additional file 1: Fig. S1). The proportion of cells with silenced URA3 or HTB1-yEGFP can be determined as the percentage of 5-FOA resistant cells (FOAR, high sensitivity) or GFP-negative cells (low sensitivity), respectively [19]. Loss of gene silencing is revealed by the decrease of the percentage of 5-FOAR or the proportion of GFP-negative cells in the mutant relative to wild-type isogenic strains. These two reporters had reproduced previously documented effects of various mutants in replication factors or histone chaperones [18, 19].

Using the 5-FOA resistance assay, we compared the impact of the pol30-6, pol30-8 and pol30-79 alleles on the expression of URA3 by the adh4-URA3-tel reporter [20], which had been used to characterize the pol30 mutants in prior publications, and our dual reporters (Fig. 1). Two concentrations of 5-FOA were applied. We found that the proportions of FOAR in the cells harboring the dual reporters were about 5 times lower that in the cells harboring adh4-URA3-tel (Fig. 1A–C). We attribute these differences to the longer distance between the telomere and URA3 and of the presence of the HTB1 promoter in-between (Additional file 1: Fig. S1). Importantly, all three reporters produced similar reductions of the percentage of FOAR cells in the various pol30 mutants (Fig. 1D, E). These results were in agreement with previous analyses of the pol30 mutants with the FOAR or the CRASH assays [9, 10, 12]. Taken together, we conclude that the dual reporters faithfully capture the effects of the pol30 mutations.

Fig. 1

Proportions of cells with silenced URA3 (% FOAR) in strains harboring adh4-URA3-tel (A), adh4-URA3-HTB1→yEGFP-tel (B) and adh4-URA3-yEGFP←HTB1-tel reporters. The analyses were performed with 0.5 × and 1 × concentrations of 5-FOA as described in the text. The pol30 mutants are listed on the horizontal axis. The calculations in A–C represent average values and standard deviations s of three experiments. Asterisks represent statistical significance compared to the wild type within each concentration. *p value 0.05, **p value < 0.001. In D and E the ratio between % FOAR cells in the POL30 versus pol30 mutant strains was calculated and plotted from the values in A–C. The logarithmic scale of the vertical axis in D and E was chosen to better capture the range of differences at the two concentrations of 5-FOA

The higher concentration of 5-FOA revealed a larger difference in the percentage of FOAR-cells between the mutants and the POL30 strain (Fig. 1C). This outcome suggested that the two concentrations of 5-FOA detect a gradient of cellular URA3 expression and not a distinct bi-modal Active/Silent state of this gene. While it was apparent that the gradient was shifted towards higher expression of URA3 in the pol30 mutants, the mechanism that causes this shift was not clear. We attempted to resolve this issue by flow cytometry analyses of the expression of the juxtaposed HTB1-yEGFP reporters.

Lack of bi-modal active/silent expression of yEGFP←HTB1-tel

We used an isogenic POL30 strain with no yEGFP reporter to set the upper threshold for the GFP-negative cells, which are the functional equivalent of the 5-FOAR cells. Based on this criterium, the calculated percentage of GFP-negative cells harboring the adh4-URA3-yEGFP←HTB1-tel construct in the pol30 mutants displayed results consistent with the results with this construct in the FOAR assay; however, the proportions of GFP-negative cells were in the 75–98% range compared to 0.2–6% range of FOAR cells (compare Figs. 1C and 2A). Consequently, the calculated differences between the POL30 and the mutant pol30 strains followed a similar trend but were significantly smaller compared to the differences observed with the FOAR assay (compare Figs. 1D, E and 2B). Hence, there was a notable discrepancy in the detection of silenced reporters by the high sensitivity and low sensitivity assays. Importantly, the flow cytometry analyses revealed little effect of the pol30 mutations on the silencing of the yEGFP←HTB1-tel reporter (Fig. 2C). Equally important, there was no evidence for a bi-modal Active/Silent state of HTB1-yEGFP expression in the POL30-0, pol30-6 and pol30-79 strains, and a tiny population of cells with elevated GFP signal in the pol30-8 strain (Fig. 2C). These results supported the idea that the pol30 mutations cause a transient de-repression and not a conversion to a distinct active state of yEGFP←HTB1-tel reporter. This transient de-repression cannot be detected by a low sensitivity assay.

Fig. 2

Flow cytometry analysis of cells harboring the adh4-URA3-yEGFP←HTB1-tel (A) and adh4-URA3-HTB1→yEGFP-tel (B) reporters. A, B Flow cytometry density plots (top) and GFP signal distribution graphs (bottom) with the indicated strains are shown. C Percentage of GFP-negative cells from three independent experiments are plotted. Asterisks represent statistical significance compared to the wild-type harboring each fragment. *p value 0.05. D The ratio between the proportions of GFP-negative cells in POL30 versus the pol30 mutants was calculated and plotted in an exponential graph

Bi-modal active/silent expression of HTB1→yEGFP-tel

We readily observed two distinct populations of GFP-positive and GFP-negative cells when the flow cytometry assays were performed with the adh4-URA3-HTB1→yEGFP-tel reporter (Fig. 2D). In the POL30 strain, the percentage of the GFP-negative cells was 60% and decreased to about 40% in the mutants. The differences between POL30 and the mutants were far smaller when compared to the differences detected by the FOAR assay with URA3, again supporting the idea that the FOAR assay is detecting transient de-repression and not Silent→Active epigenetic conversions. Importantly, the comparison between the adh4-URA3-HTB1→yEGFP-tel and adh4-URA3-yEGFP←HTB1-tel reporters pointed to a difference in the expression of HTB1-yEGFP depending on the position and orientation of the reporter and independent of the effect of the pol30 mutations (Fig. 2A, and B).

We noticed a higher spread of the GFP-positive population across the FSC-A axis. We performed a Pearson correlation analysis between FSC-A and EGFP-A compensated values for both orientations of the fragment. A linear positive correlation was observed in all strains, with strains harboring the HTB1→yEGFP-tel fragment showing stronger correlation (Additional file 1: Fig. S2A). Using microscopy, we followed up by measuring the size of the cells harboring no GFP or the HTB1-yEGFP fragments in both orientations. This analysis showed that all pol30 mutants have an increased cell size as compared to the wild-type counterparts (Additional file 1: Fig. S2B). However, we did not observe any statistically significant difference between the strains with and without HTB1-yEGFP. These analyses suggest that although higher GFP signal and FSC-A are associated, the cell size and the GFP signal statistical associations are not caused by the HTB1 promoter. Equally importantly, the single or bi-modal modes of expression of the reporters are not related to the size of the cells.

Analyses of the expression of HTB1-yEGFP by fluorescent microscopy

We followed with analyses of the expression of HTB1-yEGFP using fluorescent microscopy (Fig. 3A, B). The measurement of the intensity of GFP fluorescence in individual cells demonstrated that the adh4-URA3-HTB1→yEGFP-tel construct produced, on average, 3.5 times higher signals compared to the adh4-URA3-yEGFP←HTB1-tel construct across all strains (Fig. 3C, D). Both constructs produced higher signals in the pol30 mutants relative to the isogenic POL30-0 strain with pol30-8 cells showing the strongest effect (Fig. 3C, D). The percentage of GFP-positive cells was determined as follows. We averaged the intensity of signals in ROI (Region of Interest) with no cells and postulated a threshold of 1.25% higher signal for GFP-positive cells. By these criteria, in the POL30-0 strain the two constructs produced 60% and 94% GFP-negative cells, respectively (Fig. 3E). The reduction of GFP-negative cells in the pol30 mutant strains mirrored the effects observed by flow cytometry (Figs. 2C and 3E). Again, the pol30-8 mutation showed the strongest statistically significant effect regardless of the construct used, while only the adh4-URA3-HTB1→yEGFP-tel construct produced statistically significant reduction of silencing in the pol30-6 and pol30-79 strains (Fig. 3D). These analyses were in good agreement with the data obtained by flow cytometry and 5-FOA-resistance assays (Figs. 1 and 2).

Fig. 3

Fluorescent microscopy analysis of cells harboring the adh4-URA3-HTB1→yEGFP-tel (A) and adh4-URA3-yEGFP←HTB1-tel (B) reporters. A, B Images of indicated strains captured at ×40 resolution. While the signal ratios in the two panels were kept constant, the brightness/contrast of the GFP and MERGE images in the righthand and lefthand panels were altered to better represent the on/off state of yEGFP expression and the differences in intensity of the yEGFP signal between the two constructs. C GFP intensity distribution of strains harboring adh4-URA3-HTB1→yEGFP-tel fragment compared against background intensity. D GFP intensity distribution of strains harboring adh4-URA3-yEGFP←HTB1-tel fragment compared against background intensity. E Percentage of GFP-negative cells from three independent experiments are plotted. Asterisks represent statistical significance compared to the wild-type harboring each fragment. *p value 0.05, **p value < 0.001

Genetic interactions of pol30 mutants with CAC1 and ASF1

Previous studies have shown that the deletion of CAC1 has a synergistic negative effect on the gene silencing at the VIIL sub-telomere in pol30-6 and pol30-79, but not pol30-8 mutants [9, 10]. Conversely, in pol30-8 cells, the deletion of the histone chaperone ASF1 further reduced silencing at the sub-telomeres while having little effect in pol30-6 and pol30-79 [10]. These studies were performed with the 5-FOA-resistance assay and did not distinguish between transient de-repression and/or epigenetic conversions of URA3. Therefore, we asked if the deletions of ASF1 and CAC1 in the pol30 mutants would present evidence for transient de-repression or a bi-modal Active/Silent state of the HTB1-yEGFP reporter.

First, we confirmed that URA3 in these reporters can also reproduce the previously reported observations of the pol30 mutations in cac1∆ and asf1∆ genetic backgrounds. The analyses were performed with the adh4-URA3-HTB1←yEGFP-tel construct, which produced slightly higher percentage of FOAR cells (Fig. 1C) and can more reliably detect the reduction of silencing in the cac1∆ and asf1∆ strains. We found that at 0.5 × 5-FOA the pol30 mutations had no statistically significant effect in the cac1∆ background (Fig. 4A, B). At the higher 1 × 5-FOA concentration, pol30-6 and pol30-79, but not pol30-8, exacerbated the silencing defects at URA3 in cac1∆ cell (Fig. 4C, D). In the asf1∆ background, the pol30-8 mutation markedly reduced the silencing of URA3 at both concentrations of 5-FOA (Fig. 4A, C), while pol30-6 and pol30-79 showed statistically significant loss of silencing only in 1 × concentration (Fig. 4C, D). Hence, our construct can reproduce the effects of cac1∆ and asf1∆ in the context of the three pol30 mutations [9, 10, 14]. Notably, the effects of cac1∆ and asf1∆ have been reproduced at different concentrations of 5-FOA, thus reiterating that subtle differences in the concentrations of 5-FOA could lead to different interpretations in different studies.

Fig. 4

Analysis of URA3 gene expression in mutant pol30 strains harboring deletions of CAC1 and ASF1. The analyses were performed using adh4-URA3-yEGFP←HTB1-tel fragment at 0.5 × and 1 × concentrations of 5-FOA as described in the text. The pol30 mutants are listed on the horizontal axis. The graphs in A, C represent average values and standard deviations of three experiments at the indicated concentrations. Asterisks represent statistical significance compared to the wild type within each gene deletion background. *p value 0.05, **p value < .001. B, D The ratio between % FOAR cells in the POL30 versus the pol30 mutant strains were calculated and plotted from the values in A and C

Next, we tested how the deletions of CAC1 and ASF1 affect the expression of HTB1-yEGFP in the two reporters. As previously shown in Fig. 2, the adh4-URA3-HTB1→yEGFP-tel reporter displayed a distinct bi-modal Active/Silent state of the expression of the HTB1-yEGFP (Fig. 5A, Additional file 1: Fig. S3A, B). The deletion of CAC1 decreased the proportion of GFP-negative cells in all four strains, with stronger effects in the pol30-6 and pol30-79 strains (Fig. 5C). The deletion of ASF1 had a stronger effect in the pol30-8 mutants (Fig. 5C).

Fig. 5

Genetic interactions of the pol30 mutants with CAC1 and ASF1. A Density plots of the pol30 strains with deletions of CAC1 and ASF1 analyzed by the adh4-URA3-HTB1→yEGFP-tel construct. B Density plots of the pol30 strains with deletions of CAC1 and ASF1 analyzed by the adh4-URA3-yEGFP←HTB1-tel construct. C The percentage of GFP-negative cells from three independent experiments with the indicated strains are plotted. Asterisks represent statistical significance compared to the wild-type harboring each fragment in each genetic background. *p value 0.05, **p value < 0.001

The analyses with the adh4-URA3-yEGFP←HTB1-tel construct revealed a different picture. In Fig. 5B, we show that the deletion of CAC1 in the context of POL30 and the pol30 mutant backgrounds decreased the overall proportions of GFP-negative cells as demonstrated by the upward shift of the signals in the density plots. However, a clear bi-modal distribution of GFP-positive and GFP-negative cells was observed only in the pol30-8 mutant (Fig. 5B, Additional file 1: Fig. S3A, B). This, along with the data in Fig. 4, suggests that the deletion of CAC1 only leads to transient de-repression of both genes in this dual reporter. In contrast, the deletion of ASF1 produced a small but distinct proportion of GFP-positive cells in POL30 cells, which was increased in the pol30 mutants with pol30-8 having a more significant effect (Fig. 5B, C, Additional file 1: Fig. S3C, D). We conclude that at this position of the HTB1-yEGFP reporter the loss CAF1 function is leading to transient de-repression and that this effect is similar to the one caused by the deficiency of POL30 function. Conversely, the deficiency of Asf1p leads to a bi-modal Active/Silent state of the reporter. In both genetic backgrounds, the mutations in POL30 quantitatively exacerbate these effects, but have distinct qualitative effects.

Genetic interactions of pol30 mutants with RRM3

The replication forks frequently pause in the sub-telomeric regions of the chromosomes. The events of pausing are exacerbated in rrm3∆ mutants, but the exact positions of the pausing are not known [21, 22]. RRM3 encodes a DNA helicase necessary for the restart of paused replication forks [17]. We considered the possibility that the positional effects observed with our dual reporters are linked to the pausing of the fork. If this assumption is correct, the deletions of RRM3 would have different effects on the expression of URA3 and HTB1-yEGFP in the two reporters.

In Fig. 6A, B, we show that the deletion of RRM3 in the POL30-0 and pol30-79 strains reduced the percentage of FOAR cell threefold to tenfold at 0.5 × or 1 × 5-FOA concentration, respectively. The deletion of RRM3 in the pol30-6 and pol30-8 strains did not show a significant additive effect (Fig. 6A, B, Additional file 1: Fig. S4A). We followed up by flow cytometry to measure the expression of HTB1-yEGFP. In both constructs, deletion of RRM3 produced minor, but statistically significant decreases in the percentage of GFP-negative cells in POL30-0 and pol30-79 strains and not in pol30-6 and pol30-8 strains (Fig. 6C, D Additional file 1: Fig. S4B, C). Hence, it seems unlikely that the observed positional effects in the expression of HTB1-yEGFP are caused by the pausing of the fork. At the same time, it is noteworthy that RRM3 synergistically interacts with POL30-0 and pol30-79, but not with the pol30-6 and pol30-8 alleles.

Fig. 6

Analysis of URA3 gene expression and genetic interactions of the pol30 mutants with RRM3. URA3 gene expression analyses were performed at ×0.5 (A) and ×1 (B) 5-FOA concentrations. The pol30 mutants in W303 and rrm3Δ backgrounds are listed on the horizontal axis. The calculations in A and B represent average values and standard deviations of three experiments. C The percentage of GFP-negative cells from three independent experiments with the indicated strains are plotted. D Density plots of the pol30 strains with deletions of RRM3 analyzed by adh4-URA3-HTB1→yEGFP-tel and adh4-URA3-yEGFP←HTB1-tel constructs. Asterisks represent statistical significance between connected strains. *p value 0.05, **p value < 0.001

Physical interactions of the pol30 mutants with Cac1p and Rrm3p

Cac1p and Rrm3p contain a PIP (PCNA Interacting Peptide) and both are known to directly interact with Pol30p (PCNA) proteins in vitro [7, 8, 23,24,25]. We tested if these two proteins differentially interact with the mutant Pol30p proteins by a two-hybrid interaction assay as described previously [25]. Cac1p and Rrm3p were used as baits and the wild type and mutant Pol30p as prey. In this assay, the three pol30 mutations reduced the binding to Cac1p to the levels observed with the negative cac1PIP∆ control in which the PCNA-Interacting-Peptide (PIP) sequence was destroyed (Fig. 7A). The interaction with Rrm3p was reduced by the pol30-6 and pol30-8 mutations, but not by the pol30-79 mutation.

Fig. 7

Physical interactions of the pol30 mutant proteins with Cac1p and Rrm3p. A Yeast two-hybrid interaction measured as units of β-galactosidase produced per milligram of protein (U/mg of protein). Average values and standard deviations from three independent experiments are plotted on Y-axis as a factor of wild-type protein interactions. The bait and prey plasmids are indicated on the graph. B Co-immunoprecipitation of Pol30p and mutant pol30p by Cac1p-FLAG. Western blot was performed using α-PCNA and α-FLAG antibodies. C Co-immunoprecipitation of Pol30p and mutant pol30p by Rrm3p-Myc. Western blot was performed using α-PCNA and α-Myc antibodies. D Pixels of western blot band images were quantified, and the Pol30p bands were normalized against corresponding Cac1p/Rrm3p band intensities. Average values and standard deviations from two independent experiments with two biological replicates were plotted. Asterisks represent statistical significance compared to the wild-type Pol30p in each IP. *p value 0.05, **p value < 0.001

We followed up by co-immunoprecipitation assays with Cac1p-FLAG and Rrm3p-Myc tagged proteins (Fig. 7B, C). These were expressed from low copy plasmids in the POL30-0 and the mutant pol30 strains and the immuno-precipitates were tested for the presence of Pol30p. In agreement with the two-hybrid assay and previous reports [9], the three Pol30p mutant proteins were found to bind poorly to Cac1p-FLAG (Fig. 7B, D). The interaction of Rrm3p was reduced fivefold by the pol30-6 mutation, less than twofold by pol30-8 and was not affected by pol30-79 (Fig. 7C, D). We conclude that all three mutations in POL30 affect the association with Cac1p, while pol30-6 and pol30-8 reduce the association with Rrm3p.