The establishment of gene silencing at single-cell resolution

Abstract

The establishment of silencing in Saccharomyces cerevisiae is similar to heterochromatin formation in multicellular eukaryotes. Previous batch culture studies determined that the de novo establishment of silencing initiates during S phase and continues for up to five cell divisions for completion. To track silencing phenotypically, we developed an assay that introduces Sir3 protein into individual sir3Delta mutant cells synchronously and then detects the onset of silencing with single-cell resolution. Silencing was completed within the first one to two cell divisions in most cells queried. Moreover, we uncovered unexpected complexity in the contributions of a histone acetyltransferase (Sas2), two histone methytransferases (Dot1 and Set1) and one histone demethylase (Jhd2) to the dynamics of silencing. Our findings showed that removal of methyl modifications at H3K4 and H3K79 were important steps in silent chromatin formation and that Jhd2 and Set1 had competing roles in the process.

Figure 1. A pedigree assay to measure the establishment of silencing as a function of cell divisions

(a) Wild-type cells contain cryptic copies of α1 and α2 transcription factor genes at HML whereas copies of these same genes at MAT are transcribed. (b) In the pedigree assay, Strain 1 (JRY8828) containing a wild-type copy of SIR3 was mated to a sir3-deficient Strain 2 (JRY8829). (Shown prior to mating in e, and after mating in f). (c) Using a micromanipulator, zygotes were moved to an α-factor source where they divided (pictured in g) until HML α1 and α2 transcription factor genes were functionally silenced. (d) Upon silencing of HML, cells became sensitive to α-factor and arrested as shmoon (pictured in h).

Figure 2. Strain 1 (JRY88828) and Strain 2 (JRY8829) fusion products established silencing after 1 – 3 cell divisions

(a) Barplot of pedigree pattern counts. Zygotes and their daughters were tracked by microscopy to determine their pattern of arrest, and by inference, HML silencing. Upon silencing, cells arrested division as shmoon at different points in their lineage, producing five possible patterns of silencing. No zygote ever silenced HML prior to cell division (Pattern 0). Pattern 1 – Pattern 5 represent silencing events produced between 1 – 3 cell divisions. In Pattern 5, the extra division could have occurred in any of the four grand-daughter cells, not necessarily the D1-1 cell as depicted. The data represented 643 zygotes and all their descendants. (b) Proportional stacked plot. This figure depicts the same pedigree pattern counts as in Fig. 2a as a stacked plot. (c) Pedigree notation and silencing probability. The names of cells are given here as they are produced in a dividing pedigree lineage. The probability that a given cell of this type was silenced is shown adjacent (computed from the data in Fig. 2a).

Figure 3. Sir3 over-expression studies

(a) Expression of SIR3 by qRT-PCR. An additional copy of Sir3 on either a CEN-ARS (single-copy) or 2 μ (multi-copy) plasmid in Strain 1 caused an over-expression of SIR3 transcript (JRY8847 – JRY8850, using JRY8828 and JRY8829 as controls). (b) Silencing HML using an over-abundance of Sir3. Pedigree profiles of silencing establishment using Sir3 over-expression strains are shown compared to strains silenced with the native SIR3 and empty vectors. From left to right, the strains were JRY8847- JRY8850 × JRY8829. There was no significant association between pedigree pattern and Sir3 expression levels. That is, the likelihood ratio test nominal p-values of 0.247 (CEN-ARS) and 0.545 (2μ) were not statistically significant. (c) Variation between replicates. Technical experimental replicates of Strain1 (JRY8828) and Strain 2 (JRY8829) in the pedigree assay were performed on zygotes from temporally coincident assays. Several replicates are depicted to illustrate the variation between Strain1 and Strain2 in the pedigree assay. The likelihood ratio test was performed for the pairwise comparison of the 9 groups of wild-type control pedigree assays: the 36 nominal p-values ranged from 0.181 to 0.999 with a mean of 0.626 suggesting that the differences between the profiles of SIR3 over-expression lines to their corresponding control pedigrees are similar to the variation within wild-type assays (Supplemental Fig. 2a). (d) CEN-ARS and 2μ plasmid loss rates. The plasmid loss per cell division of the two SIR3 over-expression plasmids is shown.

Figure 4. Silencing of cells with mutations in chromatin modifying enzymes

Yeast strains isogenic to Strain 1 (JRY8828) and Strain 2 (JRY8829) and lacking either dot1Δ, sas2Δ, set1Δ or jhd2Δ were assayed for their kinetics of silencing using the pedigree assay. They were compared to silencing in zygotes from wild-type Strain 1 and Strain 2 silencing on the same plates. Pedigree patterns generated from these strains are displayed using barplots. The likelihood ratio test was applied to detect associations between pedigree pattern and genotype. All four mutants were significantly different from wild type: p-value_dot1Δ = 4.59 E-10; p-value_sas2Δ < E-16; p-value_set1Δ= 9.80 E-5; p-value_jhd2Δ = 4.22 E-3. As a benchmark, pairwise comparisons between the four groups of wild-type assays yielded six nominal p-values ranging from 0.179 to 0.900, with a mean of 0.610 representing the low variability amongst wild-type samples (Supplemental Fig. 2). The number of pedigrees tabulated for each comparison is indicated beneath the genotypes.

Figure 5. Loss of DOT1 enhanced the sir1Δ loss-of-silencing phenotype

(a) Mating ability of sir1Δ dot1Δ double mutants.sir1Δ and dot1Δ single and double mutant cells (JRY8873, JRY4621, JRY8874, JRY8875, JRY8957, JRY8958) were tested for their ability to mate with tester strains (JRY2726, JRY2728) by quantitative mating efficiency assay and compared to wild-type (W303-1a, W303-1b) and sir4Δ (JRY3411, JRY3841) strains. (b) Mating ability in jhd2Δ and JHD2 over-expressing cells.jhd2Δ strains (JRY8843, JRY8844) and JHD2 over-expressing yeast (JRY8884, JRY8885) were tested as in (a) for their ability to mate. (c) Silencing of a URA3 reporter in cells lacking SIR1 and DOT1. A strain replacing the Kluyveromyces lactis URA3 open reading frame for a1 at HMR was tested for HMR expression on CSM – ura plates and for growth on 5-fluoro-orotic acid (5-FOA) plates to which strains expressing URA3 are sensitive. Isogenic sir1Δ, dot1Δ, and sir1Δdot1Δ double mutants were diluted to 1 OD and 1:10 serial dilutions, spotted onto appropriate plates, and grown at 30° C for 2 days (JRY8876 – JRY8833). (d) Expression of α2 and a1 in cells lacking or over-expressing JHD2 by qRT-PCR. Transcript levels of a1 from HMR were measured in cells that lacked or over-expressed Jhd2 (MATα background, JRY8844 and JRY8885) using qRT-PCR. set1Δ strains (JRY8889) and wild-type strains (W303-1a, W303-1b) served as controls. HML α2-gene expression in cells in a MATa cells (JRY8843, JRY8884, JRY8888, W303-1a, W303-1b) was also assessed. Results were expressed as the average fold-expression over actin relative to MAT expression in biological triplicates.