Analysis of epigenetic stability and conversions in Saccharomyces cerevisiae reveals a novel role of CAF-I in position-effect variegation

Position-effect variegation (PEV) phenotypes are characterized by the robust multigenerational repression of a gene located at a certain locus (often called gene silencing) and occasional conversions to fully active state. Consequently, the active state then persists with occasional conversions to the repressed state. These effects are mediated by the establishment and maintenance of heterochromatin or euchromatin structures, respectively. In this study, we have addressed an important but often neglected aspect of PEV: the frequency of conversions at such loci. We have developed a model and have projected various PEV scenarios based on various rates of conversions. We have also enhanced two existing assays for gene silencing in Saccharomyces cerevisiae to measure the rate of switches from repressed to active state and vice versa. We tested the validity of our methodology in Δsir1 cells and in several mutants with defects in gene silencing. The assays have revealed that the histone chaperone Chromatin Assembly Factor I is involved in the control of epigenetic conversions. Together, our model and assays provide a comprehensive methodology for further investigation of epigenetic stability and position effects.


Figure 1. Toxicity of 5-FOA depends on the expression of URA3 and is not reversed by Hydroxyurea.
It has been shown that Hydroxyurea (HU) has different effects on DNA replication at different temperatures (1) and that HU can modulate the sensitivity to 5-FOA (2,3). It is possible that under certain conditions (different temperatures, different concentrations of Hydroxyurea and 5-FOA) URA3-expressing cells can survive in the presence of 5-FOA. We tested if under our experimental conditions toxicity to 5-FOA depends on the expression of URA3.
Wild type wine yeast (URA3), W303 (ura3-52 URA3::VIILtel) and W303Δpol30 cells carrying POL30 on a plasmid (pBL230-POL30 URA3) were streaked on SC, SC-ura and SC/FOA plates supplemented with 10 mM Hydroxyurea and grown at 23°C for 2 days on SC and SC-ura plates and for 6 days on SC/FOA plates. The variegating URA3 at the VIIL telomere of W303 cells confers growth on medium without uracil and on medium containing 2 μg/ml 5-FOA. In contrast, the plasmid-borne URA3 confers growth on SC-ura. However, its loss renders the cells non-viable because of the concomitant loss of POL30. Wild type wine yeast cells are also nonviable indicating complete sensitivity to 5-FOA. The addition of 10 mM Hydroxyurea does not reverse the toxicity of 5-FOA after 6 days of incubation. We conclude that at the conditions used the toxicity of 5-FOA strictly depends on the expression of URA3 and not on the activity of RNR. and HU. CAN1 encodes for an Arginine transporter. Can1p also transports the toxin canavanine and confers sensitivity to the drug in the absence of arginine. Gain of canavanine resistance is frequently used as a measure for spontaneous mutations in S. cerevisiae. In wild type cells the yields of forward CAN1 mutations is in the range of 3x10 -7 (4,5). A genome-wide screen has identified non-essential genes whose loss moderately (3x10 -6 ) or severely (up to 1.67x10 -5 ) increase these mutation rates (5). In Δcac1 and pol30 mutants these rates are in the range of 1-3x10 -6 (4,6,7).
We tested if 5-FOA or HU increase mutation rates in Δcac1 cells and if the mutation rates change depending on whether the cells contain telomeric URA3. Liquid cultures of Δcac1 or isogenic BY4742 cells with or without URA3 at the VIIL telomere, respectively, were grown in YPD and washed once in sterile water. 2-3 x10 7 cells were spread on SC-arg plates containing 60 μg/ml canavanine, 2 μg/ml 5-FOA or 10 mM HU as indicated. Plates were grown for 4 days and photographed. As demonstrated earlier (4,7), there is a moderate increase in the forward CAN1 mutations in Δcac1 cells (see the SC+Can plates). Such spontaneous mutation rates (less than 10 -6 ) could not account for the incidence of FOA-resistant cells (see SC/FOA plates) in Δcac1 with telomeric URA3 (10 -2 to10 -3 ). We also show that 5-FOA, HU or the combination of both did not increase the mutation yields in cells without URA3 and that the mutation rates in the presence of the drugs were not enhanced by the insertion of URA3 in the telomeres. Hence, 5-FOA derivatives that are produced by the URA3-encoded Orotidine-5'-phosphate-decarboxylase do not increase mutation rates. We do not exclude that at significantly lower concentration of 5-FOA such derivatives could show mutagenic activity. However, in TPE assays the toxicity of 5-FOA seems to significantly exceed its potential mutagenicity.

Appendix 1. Calculations on a recurrence relation of the type Y (A)n = Y (A)n-1 -Y (A)n-1 C (A→S) + (1-Y (A)n-1 ) C (S→A)
where Y (A)n is the proportion of cells with Active gene at generation n (n 0 … n -1 , n, n +1 … etc), C (A→S) is the coefficient of conversions from Active to Silent state and C (S→A) is the coefficient of conversions from Silent to Active, or simply presented as Y n = Y n-1 (1 -C (A→S) ) + (1-Y n-1 ) C (S→A) or Y n = Y n-1 (1 -C (A→S) -C (S→A) ) + C (S→A) I. The basic problem is to solve a first order recurrence relation, for (1) Where X 0 , a and b are parameters that we specify in the model. a = 1 -C (A→S) -C (S→A) b = C (S→A) In the initial condition (growth on selection medium) X 0 could be X 0 = 1 or X 0 = 0.
A second order recurrence relation would be X n =aX n-1 + bX n-2 + c. We will not be concerned with second order recurrence relations other than to say that the method we describe will apply to these and higher order recursions as well. We then apply the method of telescoping with the idea to write

II.
We keep the first equation, we multiply the second equation through by r, the third equation through by r 2 … and so on, so that the last equation is multiplied through by r n-1 . The system of equations now looks this way: M 1 X appears once on the left hand side of the system, and once, as 1 raX , on the right hand side of the system. If we want 1 X to cancel from the system, we must choose r such that  (4), and then multiplying through both sides of (4) by 1 − n a , we obtain the solution for n X : With (5) in hand, we return to our initial recursion: Putting these values in (5), we obtain ( ) The last item to consider is what happens for large values of n, i.e., what is the convergence behavior as ∞ → n . A limit will only exist if As long as we have this condition, we may take a ratio of the parameters.