Where does mediator bind in vivo?

Abstract

Background: The Mediator complex associates with RNA polymerase (Pol) II, and it is recruited to enhancer regions by activator proteins under appropriate environmental conditions. However, the issue of Mediator association in yeast cells is controversial. Under optimal growth conditions (YPD medium), we were unable to detect Mediator at essentially any S. cerevisiae promoter region, including those supporting very high levels of transcription. In contrast, whole genome microarray experiments in synthetic complete (SC) medium reported that Mediator associates with many genes at both promoter and coding regions.

Principal findings: As assayed by chromatin immunoprecipitation, we show that there are a small number of Mediator targets in SC medium that are not observed in YPD medium. However, most Mediator targets identified in the genome-wide analysis are false positives that arose for several interrelated reasons: the use of overly lenient cut-offs; artifactual differences in apparent IP efficiencies among different genomic regions in the untagged strain; low fold-enrichments making it difficult to distinguish true Mediator targets from false positives that occur in the absence of the tagged Mediator protein. Lastly, apparent Mediator association in highly active coding regions is due to a non-specific effect on accessibility due to the lack of nucleosomes, not to a specific association of Mediator.

Conclusions: These results indicate that Mediator does not bind to numerous sites in the yeast genome, but rather selectively associates with a limited number of upstream promoter regions in an activator- and stress-specific manner.

Figure 1. Mediator association at putative target and non-target regions in SC and YPD medium.

(A) Association of Med15-(HA)₃ was determined at the indicated genomic regions as described previously . Relative occupancy values were calculated by determining the apparent immunoprecipitation efficiency (ratio of immunoprecipitated to input DNA) and normalized to the average value of 10 non-target regions (underlined with dashes), which was defined as 1 and served as the internal control. All values represent the mean of at least three independent experiments and the standard deviation is indicated. The indicated ranks of genomic regions are taken from Andrau et al. (2006) by sorting by Mediator.BR descending of the 1241 targets (gene names: p, promoter regions; o, ORF regions) and the rest of non-targets separately. (B) Control experiment with an isogenic yeast strain expressing untagged Med15. Relative occupancy values were normalized to the average of 10 non-target regions.

Figure 2. Highly transcribed regions give higher signals in ChIP experiments.

(A) Mediator association at the indicated GAL regions in cells grown in galactose (gal) or glucose (glu). Primer pair GAL1+976 (underlined) is identical to that used by Zhu et al. (2006). Values were normalized to an ORF-free region on chromosome V, which was defined as 1. (B) TBP and (C) Hsf1 association at the indicated GAL regions in cells grown in galactose (Gal) or a mixture of glucose and galactose (GalGlu). (D) TBP and (E) Gal4 association with the indicated GAL and SSA4 regions in cells grown in absence (NHS) or presence of an 8 min heat shock (HS) at 39°C. (F) Mediator association with genes highly transcribed by Pol I (RDN18 and RDN37) and Pol III (RDN5).