Nucleosomes Are Essential for Proper Regulation of a Multigated Promoter in Saccharomyces cerevisiae

Abstract

Nucleosome-depleted regions (NDRs) are present immediately adjacent to the transcription start site in most eukaryotic promoters. Here we show that NDRs in the upstream promoter region can profoundly affect gene regulation. Chromatin at the yeast HO promoter is highly repressive and numerous coactivators are required for expression. We modified the HO promoter with segments from the well-studied CLN2 NDR, creating chimeric promoters differing in nucleosome occupancy but with binding sites for the same activator, SBF. Nucleosome depletion resulted in substantial increases in both factor binding and gene expression and allowed activation from a much longer distance, probably by allowing recruited coactivators to act further downstream. Nucleosome depletion also affected sequential activation of the HO promoter; HO activation typically requires the ordered recruitment of activators first to URS1, second to the left-half of URS2 (URS2-L), and finally to the right-half of URS2 (URS2-R), with each region representing distinct gates that must be unlocked to achieve activation. The absence of nucleosomes at URS2-L resulted in promoters no longer requiring both the URS1 and URS2-L gates, as either gate alone is now sufficient to promote binding of the SBF factor to URS2-R. Furthermore, nucleosome depletion at URS2 altered the timing of HO expression and bypassed the regulation that restricts expression to mother cells. Our results reveal insight into how nucleosomes can create a requirement for ordered recruitment of factors to facilitate complex transcriptional regulation.

CHROMATIN is a key regulator of gene expression, influencing factor binding, transcription initiation, and elongation (Li et al. 2007). Nucleosomes can occlude transcription factor binding sites, and at some promoters nucleosomes must be disassembled by transcriptional coactivators prior to effective protein binding (Verdone et al. 2002; Adkins et al. 2004; Schwabish and Struhl 2007; Biddick et al. 2008; Takahata et al. 2009). At most promoters, however, transcription factor binding is much less complex due to natural depletion of nucleosomes from their binding sites. Nucleosome-depleted regions (NDRs) exist at ∼95% of yeast promoters (Mavrich et al. 2008a). NDRs are usually found immediately adjacent to the transcription start site (TSS) and span ∼140–150 nt (Yuan et al. 2005; Mavrich et al. 2008a). These NDRs overlap with most identified binding motifs for transcription factors, facilitating the recruitment of factors necessary for gene expression (Yuan et al. 2005; Morse 2007).

The rules governing NDR formation are unclear. It has been postulated that NDR formation is guided by DNA sequences that influence DNA–nucleosome interactions (Segal et al. 2006). In support of this model, many NDRs contain long rigid poly A/T stretches that are known to disfavor nucleosome binding (Nelson et al. 1987; Kaplan et al. 2009). However, comparisons of nucleosomes formed in vitro and in vivo argue against positioning being determined solely by DNA sequence (Zhang et al. 2009). Certain DNA-binding factors have also been implicated in NDR formation, including Abf1, Mcm1, Rap1, Reb1, and Rsc3 in yeast (De Winde et al. 1993; Angermayr et al. 2003; Yarragudi et al. 2004; Badis et al. 2008; Hartley and Madhani 2009; Bai et al. 2011). Depletion of these factors individually typically reduced the size of the NDR but did not eliminate it (Hartley and Madhani 2009). Consistent with this, a recent study of the CLN2 NDR concludes that nucleosome depletion can result from the concerted actions of several DNA binding factors (Reb1, Mcm1, and Rsc3) with each factor being necessary but none sufficient for full NDR formation (Bai et al. 2011).

The CLN2 NDR deviates from the typical TSS NDR in both location and size. The CLN2 NDR is centered ∼400 bp upstream of the TSS and extends >300 bp in length. It includes sites for Reb1, Mcm1, and Rsc3 as well as three binding sites for the SBF (Swi4/6) complex required for CLN2 activation. The factors required for NDR formation are entirely contained within this region, as demonstrated by the portability of this element to other promoters. Furthermore, mutating all of the binding sites within the NDR for the known nucleosome-depleting factors results in restored nucleosome occupancy within this region (Bai et al. 2011). Significantly, the absence of nucleosomes at these SBF sites at CLN2 is critical for reliable gene expression in every cell cycle (Bai et al. 2010, 2011).

SBF is also required for activation of the HO promoter, but the steps leading to binding contrast sharply with those observed at CLN2. The yeast HO promoter has an unusually long promoter that can be subdivided into two distinct regulatory regions, URS1 and URS2 (Nasmyth 1985). The URS2 region of the HO promoter contains nine CRCGAAA consensus sequences required for SBF binding, but despite having three times the number of SBF sites, chromatin immunoprecipitation (ChIP) experiments show sharply lower SBF enrichment at the HO promoter than at CLN2 (Takahata et al. 2011). Furthermore, in vivo SBF binding across the nine HO SBF sites is not equivalent, but instead is predominantly on the left side of URS2; in vitro SBF binds equivalently to the various HO sites (Takahata et al. 2011; Yarrington et al. 2015). The differential binding of SBF appears to depend on accessibility, as the occupied sites at the CLN2 promoter are within an NDR, while the less occupied sites at HO URS2 are within a nucleosome dense region (Bai et al. 2010; Yarrington et al. 2015).

HO activation is mother-cell-specific and occurs sequentially, with ordered recruitment of transcription factors and coactivators first to URS1, second to the left-half of URS2 (URS2-L), and finally to the right-half of URS2 (URS2-R) (Cosma et al. 1999; Bhoite et al. 2001; Yarrington et al. 2015). Accompanying this ordered recruitment, waves of chromatin disassembly occur across the HO promoter, likely revealing factor binding sites necessary for the next step of activation (Takahata et al. 2009). The nucleosomes present at different regions of the HO promoter may therefore act as gates that must be opened in a defined order, and we have previously postulated that nucleosomes present at URS2-L must be remodeled by coactivators recruited upstream at URS1 prior to effective SBF binding (Takahata et al. 2009; Yarrington et al. 2015). This requirement, however, has not been directly demonstrated prior to this study.

Here we show that nucleosomes form barriers that are required for establishing multiple gates within a single promoter. We describe the construction of four chimeric HO promoters in which regions of URS2 are replaced with either the native CLN2 NDR or a mutant version of this sequence that has nucleosomes present. We demonstrate that nucleosome depletion at URS2 alters sequential activation, allowing for independent activation of the HO promoter by either URS1 or URS2. We further show that nucleosomes act as a barrier to long-range transcriptional activation and that nucleosome depletion may extend the range of coactivator-mediated chromatin disassembly. Finally, we demonstrate that nucleosome depletion affects mother–daughter HO regulation, indicating that nucleosomal barriers are required for the proper timing of gene expression.

Materials and Methods

All yeast strains used in this study are listed in Table 1 and are isogenic in the W303 background (Thomas and Rothstein 1989). Standard genetic methods were used for strain construction (Rothstein 1991; Sherman 1991). Promoter manipulations in the study were constructed using pCORE-UH and the delitto perfetto method (Storici et al. 2001), by transformation with the appropriate PCR products. Oligos used in strain construction are available upon request. Plasmids pLB202 and CLN2pr-allmut (Bai et al. 2011), generously provided by Lu Bai, were used as templates for construction of CLN2-wtNDR and CLN2-mutNDR, respectively. We synthesized alternative versions of the CLN2-wtNDR and CLN2-mutNDR segments, but lacking SBF sites, using GenScript (plasmids M5628 and M5629).

Table 1. Strain list.

Strain	Genotype
Figure 2
DY150	MATa ade2 can1 his3 leu2 trp1 ura3
DY17231	MATa CLN2[−764 to −435 deleted]:CLN2[−764 to −435, mutNDR] ade2 can1 his3 leu2 trp1 ura3
DY17533	MATa CLN2[−764 to −435 deleted]:CLN1[−850 to −450] HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17536	MATa CLN2[−764 to −435 deleted]:CLN1[−850 to −450] HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17527	MATa CLN2[−764 to −435 deleted]:CLN1[−850 to −450] HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17530	MATa CLN2[−764 to −435 deleted]:CLN1[−850 to −450] HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
Figure 3
DY150	MATa ade2 can1 his3 leu2 trp1 ura3
DY15856	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY15857	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR] ade2 can1 his3 leu2 trp1 ura3
DY16240	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16241	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
Figure 4
DY16942	MATa SWI4−V5::His3MX ade2 can1 his3 leu2 trp1 ura3
DY17366	MATa SWI4−V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY17368	MATa SWI4−V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY17362	MATa SWI4−V5::His3MX HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17364	MATa SWI4−V5::His3MX HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
Figure 5
DY150	MATa ade2 can1 his3 leu2 trp1 ura3
DY161	MATa swi5::LEU2 ade2 can1 his3 leu2 trp1 ura3
DY15856	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16077	MATa swi5::LEU2 HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY15857	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16078	MATa swi5::LEU2 HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY16240	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16493	MATa swi5::LEU2 HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16241	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16491	MATa swi5::LEU2 HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY15855	MATa HO(LX4−sbf)::KanMX ade2 can1 his3 leu2 trp1 ura3
DY18401	MATa swi5::TRP1 HO(LX4−sbf)::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY16732	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16737	MATα swi5::LEU2 HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16733	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16743	MATa swi5::LEU2 HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16734	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, mutSBF-wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16774	MATa swi5::LEU2 HO[−953 to −120 deleted]:CLN2[−764 to −435, mutSBF-wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16735	MATa HO[−953 to −120 deleted]:CLN2[−764 to −435, mutSBF-mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16778	MATa swi5::LEU2 HO[−953 to −120 deleted]:CLN2[−764 to −435, mutSBF-mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
Figure 6
DY17405	MATa SWI4-V5::His3MX HO::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17366	MATa SWI4-V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY17368	MATa SWI4-V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 lys2 trp1 ura3
DY17503	MATa SWI4-V5::His3MX HO(LX4-sbf)::KanMX ade2 can1 his3 leu2 trp1 ura3
DY18314	MATa SWI4-V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY18317	MATa SWI4-V5::His3MX HO[−953 to −624 deleted]:CLN2[−764 to −435, mutSBF-mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
Figure 7
DY150	MATa ade2 can1 his3 leu2 trp1 ura3
DY17653	MATa HO(RX5 sbf)::KanMX SWI4-V5::His3MX ade2 can1 his3 leu2 trp1 ura3
DY15856	MATa HO[−953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY18165	MATa HO[RX5 sbf; −953 to −624 deleted]:CLN2[−764 to −435, wtNDR]::KanMX SWI4-V5::His3MX ade2 can1 his3 leu2 trp1 ura3
Figure 8
DY6669	MATa GALp:CDC20:ADE2 ade2 can1 his3 leu2 trp1 ura3
DY16745	MATa GALp:CDC20:ADE2 HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY17517	MATa GALp:CDC20:ADE2 HO[−953 to −120 deleted]:CLN2[−764 to −435, mutNDR]::KanMX ade2 can1 his3 leu2 trp1 ura3
DY16322	MATa HO-GFP-NLS-PEST::NatMX4 MYO1-MCherry::HIS3MX can1 his3 leu2 lys2 met15 trp1 ura3
DY17911	MATα HO[−953 to −120 deleted]:CLN2[−764 to −435, wtNDR]-GFP-NLS-PEST::NatMX4 MYO1-MCherry::HIS3MX can1 his3 leu2 trp1 ura3

Cell cycle synchronization was performed by galactose withdrawal and readdition with a GALp::CDC20 strain grown at 25° in YP medium containing 2% galactose and 2% raffinose (Bhoite et al. 2001). A high degree of synchrony was confirmed by examination of budding indices and analysis of cycle-regulated messenger RNAs (mRNAs) (data not shown). In all other experiments, cells were grown at 30° in YPAD medium (Sherman 1991).

ChIPs were performed as described (Bhoite et al. 2001; Voth et al. 2007) using mouse monoclonal antibody to the V5 epitope (SV5-Pk1, Abcam) or antihistone H3 (07-690, Upstate), and antibody-coated magnetic beads (rabbit and Pan Mouse IgG beads, Life Technologies). Samples prepared for ChIPs were cross-linked in 1% formaldehyde overnight on ice. ChIP assays were analyzed by real-time qPCR as described (Eriksson et al. 2004). ChIP qPCR primers are available upon request. H3 samples were first normalized to the ChIP signal at the IGR-I gene-free reference region on chromosome I (Mason and Struhl 2005), while Swi4-V5 ChIP samples were first normalized to the CLN1 promoter, and then both types of ChIPs were normalized to their respective input DNA sample. Error bars in the H3 ChIP assays reflect the standard deviation of four biological samples. Error bars in V5 ChIP assays reflect the standard deviation of three or more biological samples. P-values were calculated by paired t-tests.

RNA was isolated from either synchronized or logarithmically growing cells, and HO mRNA levels were measured by RT-qPCR as described (Voth et al. 2007). For all logarithmically grown strains, RNA expression was normalized to RPR1 expression and graphed relative to wild-type expression. For the synchrony experiment, RNA expression was normalized to RPR1 expression and graphed relative to the peak WT expression. HO is expressed at very low levels, about one to two transcripts per cell (Holland 2002; Bon et al. 2006; Miura et al. 2008). RPR1 encodes the RNA component of RNase P; it is a good internal control because it is transcribed by RNA pol III and it is not affected by most genetic manipulations that affect pol II transcription. Based on the abundance of the stoichiometric protein subunits of RNase P (Chong et al. 1995; Newman et al. 2006), RPR1 RNA should be at ∼50 molecules per cell. Unless otherwise noted, error bars reflect the standard deviation of three biological samples. P-values were calculated by paired t-tests. RT-qPCR primers are available upon request.

Strains for GFP fluorescence contained the yEGFP-NLS-PEST reporter driven by either the native HO promoter (Mitra et al. 2006) or by the urs2:CLN2-wtNDR version of HO constructed by the delitto perfetto method (Storici et al. 2001). A Myo1-mCherry protein fusion (Skotheim et al. 2008) allowed monitoring of cell-cycle progression. Cells were logarithmically grown and visualized on slides using a Zeiss Axio Observer microscope. Cells expressing GFP were identified by the presence of localized GFP in the nucleus. For quantification of daughter cell GFP expression, 47 mother–daughter pairs were examined for the WT promoter and 201 mother–daughter pairs were examined for the urs2:CLN2-wtNDR fusion promoter.

Data availability and reagent sharing

Strains are listed in Table 1 and are available upon request.

Results

Construction of HO urs2:CLN2 chimeras

The HO promoter experiences waves of nucleosome eviction during the cell cycle leading to promoter activation (Takahata et al. 2009; Stillman 2013). The Swi5 activator enters the nucleus during M phase, binds to sites at −1800 and −1300 in URS1 (Figure 1), and recruits coactivators leading to chromatin changes. We have previously proposed that different stretches of chromatin along the HO promoter serve as padlocks or gates that must be unlocked sequentially for ultimate gene expression (Takahata et al. 2009). We believe there are three gates, because there are three activation functions. There is one gate in URS1 that requires Swi5, and there are two gates in URS2 that each require SBF. One such URS2 gate is the URS2-L that must be unlocked by Swi5-dependent remodeling activities at URS1 prior to effective SBF binding at URS2-L. We propose that nucleosomes at URS2 constitute this gate, and that these nucleosomes restrict SBF binding prior to the chromatin disassembly that represents gate opening.

Figure 1.

Construction of HO/CLN2 chimeras. The diagrams show the general strategy used for construction of HO/CLN2 chimeric promoters. For each set of promoters, the CLN2 promoter is pictured above the HO promoter. CLN2 sequence is indicated in blue, and HO sequence, in black. CLN2 nucleosomes are represented by blue ovals, HO nucleosomes, by brown ovals, Swi5 sites, by blue boxes, and SBF sites, by red boxes. The transcription start site for each promoter is specified by an arrow. Red lines diagram the region of HO URS2 replaced with CLN2 sequence. (Top) The 329-bp CLN2-wtNDR (Left) or CLN2-mutNDR (Right) replaces a 329-bp region of URS2-L. (Bottom) The 329-bp CLN2-wtNDR (Left) or CLN2-mutNDR (Right) replaces the entire 833-bp URS2 region.

To investigate the role of nucleosomes at a multigated promoter, we made use of promoter sequences from another SBF-activated gene, CLN2. Importantly, unlike HO URS2, the SBF sites at the CLN2 promoter are in an ∼329-bp NDR that has been shown previously to be portable to HO (Bai et al. 2010). Furthermore, the same study constructed a CLN2 promoter mutant, CLN2pr-allmut, that eliminated binding sites for nucleosome-depleting factors, thus creating an alternative 329-bp CLN2 promoter region in which these SBF sites are occupied by nucleosomes. Using the mapped positions of nucleosomes (Yarrington et al. 2015) as a guide, we replaced a two nucleosome 329-bp region of HO URS2-L with either the native 329-bp CLN2 NDR sequence (urs2L:CLN2-wtNDR) or the mutated 329-bp nucleosome-containing CLN2pr-allmut sequence (urs2L:CLN2-mutNDR; Figure 1, Top). These two constructs compare the effects of nucleosomes over the major SBF binding sites in a full-length HO promoter context.

We have previously demonstrated that URS2 consists of two separate subgates, URS2-L and URS2-R (Yarrington et al. 2015). To examine the role of nucleosome occupancy on URS2 activation in a simplified setting, we created two additional constructs: urs2:CLN2-wtNDR and urs2:CLN2-mutNDR (Figure 1, Bottom). These constructs involved replacing a five nucleosome, 833-bp region corresponding to the entire URS2 with either the native 329-bp CLN2 NDR sequence (urs2:CLN2-wtNDR) or the mutated 329-bp nucleosome containing CLN2pr-allmut sequence (urs2:CLN2-mutNDR), respectively. These chimeras should allow us to explore HO activation in a single-gate URS2 setting and also allow us to probe the effects of URS2 nucleosome occupancy on HO activation without the complication of additional essential SBF sites downstream that may exist in a different nucleosomal context. Significantly, these two URS2 replacements also alter the positional context of URS1, bringing this element ∼500-bp closer to the TSS.

HO urs2(L):CLN2-wtNDR and urs2(L):CLN2-mutNDR demonstrate differing nucleosome occupancy

Transplantation of a functional CLN2 NDR into the HO promoter has been described previously (Bai et al. 2010). That study, however, inserted the CLN2 NDR region immediately downstream of HO URS2 rather than replacing URS2 sequence. Furthermore, only the native CLN2 NDR, and not the mutant CLN2pr-allmut, sequence was tested. It was therefore necessary to determine whether these two sequences were able to affect nucleosome occupancy in novel settings at URS2, especially in the very repressive environment of the HO promoter.

Nucleosome occupancy at the four HO/CLN2 chimeras was measured by H3 ChIP using primers corresponding to the CLN2 NDR (Figure 2). To avoid amplifying sequence from the native CLN2 promoter, we deleted this region from the CLN2 promoter. This deletion, however, resulted in a growth defect, presumably due to a lack of CLN2 expression. We therefore replaced the CLN2 promoter sequence with a similar region from the CLN1 promoter. This replacement suppressed the growth defect caused by deletion of the CLN2 promoter and had little effect on HO expression (data not shown). In the urs2L:CLN2-wtNDR and urs2:CLN2-wtNDR chimeras, H3 nucleosome occupancy was very low, comparable to, if not lower than, the occupancy observed at the native CLN2 nucleosome-depleted region (Figure 2, A, C, and E). Nucleosome occupancy at those same regions using the mutant CLN2-mutNDR replacements, however, produced H3 ChIP signals approximately two- to three-fold higher than their native NDR counterparts (Figure 2, B, D, and F). These results demonstrate that the CLN2 nucleosome-depleting signals are portable to HO URS2-L and URS2 and that disruption of these signals results in repopulation of these sequences by nucleosomes.

Figure 2.

H3 occupancy at HO/CLN2 chimeras validates nucleosome-depleted regions inserted at HO. (Left) Diagrams of native and mutNDR CLN2 promoters and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. H3 enrichment to various promoters was determined by ChIP followed by qPCR: (A) native CLN2 NDR, (B) CLN2-mutNDR, (C) urs2L:CLN2-wtNDR, (D) urs2L:CLN2-mutNDR, (E) urs2:CLN2-wtNDR, and (F) urs2:CLN2-mutNDR. Enrichment values relative to the native CLN2 NDR are indicated on the right. Location of ChIP primers is indicated in the diagram. Error bars reflect the standard deviation of four biological samples. **P < 0.01.

HO urs2:CLN2 chimeras show elevated HO expression

We next asked whether altered nucleosome occupancy in different regions of the HO promoter affected HO transcription. Notably, all chimeric URS2-L and URS2 replacements displayed levels of HO expression higher than WT (Figure 3). Greater levels of HO expression were observed at the HO/CLN2 chimeras with NDRs over their nucleosome-occupied mutant counterparts (Figure 3, compare B and C and D and E). This result suggests that nucleosome occupancy at URS2 represses HO expression, likely via blocking SBF binding and preventing premature gate opening. Furthermore, the highest levels of HO expression were observed with the simplified URS2 replacements that still have a normal URS1 (Figure 3, D and E). The significant increase in expression observed for these chimeras over their full-length URS2 counterparts may be attributed to two nonmutually-exclusive possibilities. First, truncation of URS2 has decreased the distance from URS1 to the TSS. We have previously proposed that the length of URS2 has been evolutionarily conserved to prevent inappropriate URS1 long-range transcriptional activation, independent of the URS2 gate (Yarrington et al. 2015). It is therefore possible that the URS2 chimeras have URS1-dependent gene activation prior to SBF binding at URS2, thereby increasing the length of time HO is activated. Second, we have previously demonstrated that although SBF binds primarily to URS2-L (Takahata et al. 2011), it is the poorly occupied SBF sites at URS2-R that ultimately activate gene expression (Yarrington et al. 2015). The simplified URS2 of the urs2:CLN2 chimeras likely allows predominant SBF binding closer to the TSS, resulting in elevated HO gene expression.

Figure 3.

Nucleosome depletion at URS2 results in elevated HO expression. (Left) Diagrams of the HO and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. HO mRNA levels were measured for various promoters: (A) native HO, (B) urs2L:CLN2-wtNDR, (C) urs2L:CLN2-mutNDR, (D) urs2:CLN2-wtNDR, and (E) urs2:CLN2-mutNDR. HO expression values relative to the native promoter are indicated on the right. Error bars reflect the standard deviation of three biological samples. *P < 0.05, **P < 0.01.

The >50% increase in HO expression observed in our urs2L:CLN2-mutNDR chimera relative to WT (Figure 3, compare A and C) was puzzling as we felt this particular construct should best mimic the WT HO promoter. It is possible that there are differences in repressive capabilities of the nucleosomes that exist at the native HO promoter and those that occur in the urs2L:CLN2-mutNDR chimera. One possible explanation for these differences would be the presence of negative repressive elements in URS2-L, which would have been replaced in the chimera, consistent with our previous results (Yarrington et al. 2015).

Nucleosome occupancy influences SBF binding at HO URS2

We predict that loss of nucleosomes at the HO/CLN2 chimeras with NDRs would result in enhanced SBF binding, and to test this, we performed ChIPs using strains where the Swi4 component of SBF had a V5 epitope tag. The Swi4-V5 tag did not affect HO expression (data not shown). The HO/CLN2 chimeras have different DNA sequences near the SBF binding sites, limiting choices for PCR primers for ChIP that would allow direct comparison to the native HO promoter. Our solution was to use primers immediately upstream of URS2 (Figure 4, Top).

Figure 4.

Nucleosome depletion affects SBF binding at HO/CLN2 chimeras. The HO promoter is portrayed, with rectangles indicating URS1, URS2, and TATA. Predominant SBF binding is represented by yellow ovals, with the location of SBF sites indicated by numbers and ChIP primers by red arrows. ChIP primers were chosen immediately upstream of URS2 to interrogate SBF enrichment at both the native HO SBF sites (A) as well as the introduced CLN2 SBF sites (B–E). (Left) Diagrams of the HO and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. Enrichment of the Swi4-V5 subunit of SBF to various promoters was determined by ChIP followed by qPCR: (A) native, (B) urs2L:CLN2-wtNDR, (C) urs2L:CLN2-mutNDR, (D) urs2:CLN2-wtNDR, and (E) urs2:CLN2-mutNDR. Enrichment values relative to the native promoter are indicated on the right. Error bars reflect the standard deviation of four biological samples. **P < 0.01.

ChIP analysis of URS2-L and URS2 replaced with native CLN2-wtNDR sequences revealed an approximate threefold increase in SBF enrichment over WT (Figure 4, compare A, B, and D) and a similar threefold increase in SBF enrichment of the CLN2-wtNDR constructs over their mutant NDR counterparts (Figure 4, compare B and C and D and E). Unsurprisingly, these chimeras also demonstrated a threefold decrease in H3 enrichment (Figure 2), suggesting an inverse linear relationship between transcription factor binding and nucleosome occupancy. The higher SBF ChIP signals observed at the HO/CLN2 chimeras with NDRs are likely due to increased SBF binding, probably via an increase in time of occupancy. This is reasonable as SBF binding normally requires Swi5-dependent remodeling activities at URS1 prior to effective SBF binding at URS2, but this would not be required when nucleosomes are absent (see below).

Nucleosome depletion at URS2 allows for HO expression independent of Swi5 and SBF

HO activation requires Swi5 action at URS1, which then leads to nucleosome eviction at URS2. It is possible that depletion of nucleosomes at URS2 in the HO/CLN2 chimeras may make the promoter somewhat independent of SWI5. Alternatively, Swi5 can activate HO expression independent of SBF when the distance between URS1 and TATA is decreased (Yarrington et al. 2015), and this could be a factor, at least with the shortened promoters of the urs2:CLN2 chimeras. It is also possible that nucleosomes may act as a barrier to long-range transcriptional activity, and, if so, nucleosome depletion at URS2-L might allow Swi5 to activate over a longer distance. To determine which of these two DNA-binding activators are required for activation of these promoters, we tested them with either a swi5 mutation (Figure 5, swi5 column) or with mutations in the SBF binding sites (Figure 5, F–J). For the two urs2L:CLN2 constructs (Figure 5, G and H), the HO SBF sites in the right half of URS2 are not mutated; mutating only the SBF sites in the left half of URS2 in the native promoter largely eliminates HO expression (Figure 5F, Yarrington et al. 2015).

Figure 5.

Nucleosome depletion affects URS1 and URS2 dependency of HO/CLN2 chimeras. Diagrams of HO and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. HO mRNA levels were measured for the indicated versions of HO or HO/CLN2 promoters in SWI5 (WT column) or swi5 (swi5 column) strains. Indicated HO mRNA values are normalized to WT. The relative standard deviation (RSD) for HO expression of our control and chimeric promoters was generally <10%. Three promoters (LX4 sbf SWI5, LX4-sbf swi5, and urs2:CLN2-mutSBF-mutNDR SWI5) had higher RSDs of 18, 13, and 12%, respectively.

Combining our chimeric promoters with a swi5 mutation revealed that nucleosomes at URS2 are indeed required to make Swi5-mediated events at URS1 relevant for activation (Figure 5, compare SWI5⁺ and swi5 columns). Both HO/CLN2 chimeras with NDRs showed high levels of HO expression even in the absence of Swi5 (Figure 5, B and D). SBF presumably binds to the nucleosome-depleted URS2 in the absence of Swi5-dependent activities at URS1, thus altering HO activation from a sequential URS1→URS2 two-step activation system to a simpler promoter that only requires SBF bound at URS2. In agreement with this model, loss of Swi5 severely reduced HO expression in the two nucleosome-occupied HO/CLN2 chimeras with mutated NDRs (Figure 5, C and E). Due to the presence of nucleosomes at the mutant NDR, these two chimeric promoters still require Swi5-dependent activities at URS1 to open the URS2-L gate for SBF binding and maximal promoter activation. These two chimeras have mutated NDRs and thus possess nucleosomes, but they show higher residual HO expression in the swi5 mutant compared to the native promoter. This is consistent with the loss of repressive promoter elements at URS2-L, as seen previously (Yarrington et al. 2015).

We next investigated whether HO expression in our chimeric promoters was affected by SBF site mutations that should eliminate subsequent downstream events dependent on SBF bound at URS2-L (Yarrington et al. 2015). Significantly, nucleosome-occupied chimeras (mutNDR) demonstrated a 3 to fourfold drop in HO expression relative to their nucleosome-depleted counterparts when SBF sites were absent from the introduced CLN2 sequence (compare Figure 5, G and H and I and J). The swi5 mutation sharply reduced expression from all of these chimeras, and we can conclude that they largely require Swi5 bound at URS1 for robust gene activation. There is significant residual activation from the urs2:CLN2-mutSBF-mutNDR chimera in the absence of Swi5 (Figure 5I); this could be due to the Reb1, Mcm1, and Rsc3 binding sites in the NDR but absent from the mutNDR (Figure 5J). However, mutating these binding sites individually had no significant effect on CLN2 expression (Bai et al. 2011), and these authors argued the direct influence of Reb1, Mcm1, and Rsc3 on transcription is very minor in the presence of the NDR. Alternatively, this activation could be due to Ace2, the Swi5 paralog that is normally unable to activate HO (Dohrmann et al. 1992; Voth et al. 2007). Although Ace2 is unable to activate native HO, it may be able to activate the construct in Figure 5I; the presence of nucleosomes would thus block Ace2 from activating the Figure 5J construct. For constructs differing only in the presence of nucleosomes, the sharp differences in gene activation show that nucleosomes act as barriers to long-range transcriptional activation by Swi5. Interestingly, even two nucleosomes in the minimal 329-bp urs2:CLN2-mutSBF-mutNDR chimera were sufficient to greatly reduce HO expression (Figure 5, compare I and J). Nucleosome depletion weakens this activation barrier, as seen with the chimeras lacking SBF sites but with NDRs (Figure 5, G and I), allowing bypass of the URS2 gate. URS2 nucleosome occupancy is therefore essential for both the URS1 and URS2 gates to function normally at the native HO promoter, and when these nucleosomes are compromised either URS1 or URS2 is independently capable of activating HO.

Nucleosome depletion at the left-half of URS2 facilitates SBF binding at the right-half of URS2

The levels of SBF binding in Figure 4 do not correlate fully with HO expression in Figure 3, where urs2:CLN2-wtNDR and urs2:CLN2-mutNDR show higher HO expression than expected based upon SBF binding. A possible explanation for this result is based on our previous demonstration that URS2 contains two separate subgates, URS2-L and URS2-R (Yarrington et al. 2015). Here, SBF binding at URS2-L acts as a required relay from URS1 for the subsequent SBF binding at URS2-R that ultimately activates HO expression. We have proposed that nucleosome depletion at the HO/CLN2 chimeras allows them to be activated by Swi5, despite SBF site mutations, particularly for the urs2:CLN2-mutSBF-wtNDR, with a short 329-bp URS2 region (Figure 5I). However, similar activation is harder to explain for the urs2L:CLN2-mutSBF-wtNDR construct, which has 1200 bp between the Swi5 binding site and the TSS (Figure 5G). Transcriptional activation in Saccharomyces cerevisiae does not normally occur at such great distances (Guarente and Hoar 1984; Struhl 1984). Although it is possible that nucleosome depletion allows for such activation from a very long distance, a more likely explanation is that nucleosome depletion at URS2-L in the urs2L:CLN2-mutSBF-wtNDR chimera allows for expanded Swi5 function. Thus in the absence of nucleosomes, the Swi5-dependent chromatin remodeling originating in URS1 might penetrate further into URS2 than normal and uncover SBF sites at URS2-R, despite the mutations in the normally required left-half SBF sites.

To examine whether the Swi5-dependent chromatin remodeling does indeed extend into URS2-R, we performed ChIP experiments to examine SBF binding to URS2-R of the HO/CLN2 chimeras with full-sized URS2s (Figure 6, blue bars). As described previously (Yarrington et al. 2015), mutation of left-half SBF sites in the native promoter largely eliminated SBF binding to the right-half of URS2 (Figure 6D). Nucleosome depletion at URS2-L, however, allowed for abundant SBF binding at URS2-R irrespective of functional left-half SBF sites (Figure 6, B and E). This result suggests that Swi5-dependent chromatin remodeling from URS1 into the right-half of URS2 is enhanced by nucleosome depletion at URS2-L, allowing for normally nucleosome-impeded sites in the right-half of URS2 to become highly SBF occupied. Importantly, SBF binding at URS2-R was reduced more than twofold at the nucleosome-occupied HO/CLN2 chimeras with mutant NDRs (Figure 6, C and F). These results support the model that in the absence of nucleosomes at URS2-L, coactivators recruited by Swi5 are able to affect nucleosome changes at URS2-R, allowing SBF to bind. Significantly, the two promoters lacking SBF sites in URS2-L but containing nucleosomes in this region, LX4 SBF (Figure 6D) and urs2L:CLN2-mutSBF-mutNDR (Figure 6F), were similar in terms of SBF binding. These results suggest that, if nucleosomes are present, the chimeric replacements do not affect the URS2-L→URS2-R sequential activation at URS2.

Figure 6.

Nucleosome depletion at URS2-L facilitates URS2-R SBF binding. (Left) Diagrams of the HO and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. Logarithmically grown cells were harvested and processed for both RNA and chromatin. (Blue bars) Enrichment of the Swi4-V5 subunit of SBF to various promoters was determined by ChIP followed by qPCR: (A) native, (B) urs2L:CLN2-wtNDR, (C) urs2L:CLN2-mutNDR, (D) LX4 sbf, (E) urs2L:CLN2-mutSBF-mutNDR, and (F) urs2L:CLN2-mutSBF-mutNDR. (Red bars) HO mRNA levels were measured for the same promoters. Enrichment (blue) and expression (red) values relative to the native promoter are indicated on the right. Error bars reflect the standard deviation of three biological samples. *P < 0.05, **P < 0.01.

It remained possible that SBF binding to URS2-R and consequential HO expression was independent of both long-range Swi5 activation and SBF binding to URS2-L. In this scenario, nucleosome depletion at URS2-L somehow allows SBF binding to these URS2-R sites independent of upstream remodeling activities originating from either URS1 or URS2-L. However, the left-half SBF-site mutant chimeric promoter urs2L:CLN2-mutSBF-wtNDR (Figure 5G) is largely SWI5 dependent. Thus for robust activation, this promoter requires either Swi5- or SBF-dependent activities at URS2-L, arguing that the URS2-R SBF sites are not readily accessible, even when nucleosomes are depleted at URS2-L.

Lastly, we measured HO expression from the same cultures of cells that were used to measure SBF binding by ChIP. HO gene expression (Figure 6, red bars) shows a tight correlation with SBF binding at URS2-R (Figure 6, blue bars). Such a strong correlation was not seen when comparing SBF binding to URS2-L (using a probe immediately upstream of URS2, Figure 4) to HO expression (Figure 3). These results support our previous data, suggesting that it is the SBF bound at URS2-R rather than at URS2-L that ultimately activates HO expression (Yarrington et al. 2015).

Nucleosome depletion at the left-half of URS2 does not bypass the requirement for SBF bound at the right-half of URS2

We have shown here that depletion of nucleosomes at URS2-L results in increased SBF enrichment at both URS2-L and URS2-R (Figure 4B and Figure 6B). Additionally, SBF binding to URS2-R and consequent HO expression does not require functional SBF sites at the URS2-L gate (Figure 6E). Thus, it is possible that nucleosome depletion at URS2-L and the consequent increase in SBF binding there could overcome the requirement for SBF sites at URS2-R. However, mutation of the SBF sites in URS2-R in the native promoter largely eliminates HO expression, demonstrating that these URS2-R sites are required in the native nucleosome-filled promoter (Figure 7B and Yarrington et al. 2015).

Figure 7.

Nucleosome depletion at URS2-L does not bypass the requirement for SBF bound at URS2-R for HO expression. Diagrams of the HO and HO/CLN2 chimeric promoters are shown, with features as in Figure 1. The positions of the nucleosomes are presumptive, based on the parents. HO mRNA levels were measured for various promoters: (A) native HO, (B) RX5 sbf, (C) urs2L:CLN2-wtNDR, and (D) RX5 sbf urs2L:CLN2-wtNDR. HO expression values relative to the native HO promoter are indicated on the right. Error bars reflect the standard deviation of three biological samples. *P < 0.05, **P < 0.01.

To investigate whether nucleosome depletion at URS2-L abrogated the need for functional SBF sites at the URS2-R gate, we mutated the last five SBF sites in the urs2L:CLN2-wtNDR chimera and examined this new strain for HO expression. This new chimeric promoter with mutated URS2-R SBF sites, named RX5 sbf urs2L:CLN2-wtNDR, displayed nearly a 10-fold decrease in HO expression (Figure 7, compare C and D). This result indicates that nucleosome depletion at URS2-L does not bypass the requirement of SBF binding at the URS2-R gate and strongly supports the spatiotemporal cascade model of SBF binding for promoter activation (Yarrington et al. 2015).

Nucleosome depletion at URS2 alters the timing of HO expression

We have demonstrated here that depletion of URS2 nucleosomes alters the sequential activation cascade seen at the native HO promoter, allowing for URS1 or URS2 to activate HO expression independent of each other. Factor binding at the native HO promoter is sequential, with Swi5 binding to URS1 occurring ∼10–15 min earlier than SBF binding to URS2 (Cosma et al. 1999). If nucleosome depletion at URS2 in our urs2:CLN2-wtNDR chimera has indeed generated a promoter independent of the normally required URS2 gate, we should be able to observe long-range, Swi5-dependent promoter activation, and this activation should occur earlier in the cell cycle compared to WT. For this experiment, we focused on the urs2:CLN2-wtNDR chimera, with a minimal, single-gated URS2 that should simplify interpretation of expression kinetics, and the two-gated urs2:CLN2-mutNDR chimera with nucleosomes (Figure 8). We refer to the urs2:CLN2-wtNDR promoter as single gated because only a single activation event, either by Swi5 or SBF, is required, while the native promoter requires one activation event by Swi5 and two by SBF. We also examined activation of the urs2:CLN2-mutNDR chimera with the expectation that this two-gated promoter, requiring activation events by both Swi5 and SBF, would show activation kinetics more similar to that of the native HO promoter.

Figure 8.

Nucleosomes are required for the proper timing of HO expression. (Top) Diagrams of the HO and urs2:CLN2-wtNDR promoters, with features as in Figure 1. (A) Cells containing the GALp::CDC20 allele were synchronized by galactose withdrawal and readdition. The t = 0 min time point represents the G2/M arrest, before release. Cells were harvested at the indicated time points and samples were processed for RNA analysis. HO mRNA levels were measured for the native (blue) and urs2:CLN2-wtNDR (red) promoters. Error bars for the native promoter and urs2:CLN2-wtNDR represent the standard deviation of six independent experiments. Error bars for urs2:CLN2-mutNDR represent the standard deviation of two independent experiments. (B, Top) Overlaid differential interference contrast (DIC) and GFP images of cells with the WT and urs2:CLN2-wtNDR promoters driving GFP expression. (Bottom) GFP-only images of the same cells. Mother–daughter expression pairs are indicated by red boxes. (C) Percentage of GFP-expressing daughter cells for the two GFP fusions. A total of 4.3% of daughters cells with the WT promoter express GFP, and 65.7% of daughter cells with urs2:CLN2-wtNDR chimeric promoter express GFP.

To examine expression kinetics, we synchronized cells by GALp:CDC20 arrest and release, which typically results in peak HO expression at 40 min after release from G2/M (Takahata et al. 2009). As predicted, HO expression in our urs2:CLN2-wtNDR chimera demonstrated early HO activation, with transcription initiating ∼10 min earlier than the WT promoter (Figure 8A). The early activation observed here is presumably Swi5 dependent but SBF independent and is normally inhibited by the nucleosomes present at URS2 in the WT promoter (Figure 5, compare D and I). In agreement with this model, the nucleosome-containing urs2:CLN2-mutNDR chimera demonstrates initial expression kinetics similar to that of the native promoter. Interestingly, our chimeric promoters appear to have a prolonged window of HO expression that is consistent with greater SBF occupancy at sites close to the TSS.

URS1-activating factor Swi5 enters the nucleus of both mother and daughter cells shortly after anaphase (Nasmyth et al. 1990). HO expression in daughter cells, however, is normally strongly repressed by the asymmetrically distributed inhibitor Ash1 factor, which binds to the HO promoter ∼30 min after GALp:CDC20 arrest and release (Bobola et al. 1996; Sil and Herskowitz 1996; Takahata et al. 2011). As our urs2:CLN2-wtNDR chimeric promoter is expressed earlier in the cell cycle than WT, it was possible that this chimeric promoter may experience a kinetic bypass of the Ash1 inhibitor in daughter cells, allowing for HO expression in both mother and daughter cells. Significantly, deletion of URS2 from the HO promoter is known to permit HO expression in both mother and daughter cells (Sil and Herskowitz 1996).

To address whether the urs2:CLN2-wtNDR chimera experiences daughter cell HO expression, we replaced the HO open reading frame with a destabilized GFP reporter containing a nuclear localization sequence (Mitra et al. 2006). Using fluorescence microscopy to monitor gene activation, we found that nucleosome depletion at URS2 did indeed result in daughter cell expression, with approximately two-thirds of daughter cells expressing GFP (Figure 8, B and C). This result demonstrates that nucleosomes are not only an essential component of a multigated promoter, but are also required for both the proper regulation and timing of promoter activation.

Discussion

HO regulatory regions URS1 and URS2 can be thought of as gates that must be unlocked sequentially for proper promoter activation. Using chimeric promoters that allow for control over nucleosome occupancy, we show here that nucleosomes are required to form the conditional barriers that comprise these gates. We showed that nucleosome depletion at the left half of URS2 increases both SBF binding and HO expression. HO expression normally requires Swi5 to bind and recruit coactivators that promote SBF binding to URS2-L; however, in the absence of nucleosomes at URS2-L, the URS1 and URS2-L gates are no longer both required for gene activation. Importantly, nucleosome depletion at URS2-L allows the coactivators recruited by Swi5 to act at a greater distance and facilitate SBF binding at URS2-R. Thus nucleosomes act as barriers to long-range transcription factor activation, and NDRs can extend the range of coactivator-mediated effects on chromatin. It is not surprising that chromatin is repressive, but the studies presented here suggest specific functional roles for individual nucleosomes in vivo and provide mechanistically distinct roles for nucleosomes in a complex, multigated promoter. Lastly, we demonstrate that nucleosomes delay the timing of gene expression at HO and are essential for the normal asymmetric expression of HO between mothers and daughters.

The ordered recruitment of factors to a promoter was first demonstrated at HO (Cosma et al. 1999); however, sequential activation is not unique to this gene. Indeed, sequential factor recruitment was later described for several mammalian genes, including interferon beta (INF-β), α1 antitrypsin, cathepsin D, collagenase, and PPARγ2 (Agalioti et al. 2000; Shang et al. 2000; Soutoglou and Talianidis 2002; Martens et al. 2003; Salma et al. 2004). In the above examples, chromatin appeared to be important in the regulation, as the sequential activation involved chromatin-modifying coactivators that allowed for subsequent recruitment of additional factors (Cosma 2002). The question remained, however, whether chromatin was the major barrier to be overcome during sequential activation and whether alterations in the normal chromatin structure would affect the timing and factor requirements for gene activation.

Signals that result in promoter activation can originate from multiple sources, and individual promoters can be responsive to more than one stimulus. A significant goal of systems biology is understanding how a requirement for multiple stimuli can be built into promoter architecture (Kim et al. 2009; Keung et al. 2015). At HO, sequential activation is a way to build a promoter that requires two DNA-binding proteins, Swi5 and SBF, that are activated at different times in the cell cycle. Swi5 binds at −1800 and −1300, at sites that are in NDRs (Takahata et al. 2009; Yarrington et al. 2015). The closest Swi5 binding site is quite far from the TATA, and it is unlikely that transcriptional activation in S. cerevisiae can occur at such great distances (Guarente and Hoar 1984; Struhl 1984). Instead, the coactivators recruited by Swi5 act by facilitating nucleosome eviction and SBF binding at URS2-L. SBF at URS2-L facilitates the SBF binding at URS2-R, which activates transcription (Yarrington et al. 2015). Here we show that nucleosomes must be present at URS2 to inhibit SBF binding, and this inhibition is what creates SWI5 dependence. Nucleosome depletion at URS2 therefore eliminates the requirement for multiple stimuli to achieve HO activation and instead creates a promoter that requires only one of the two activators, disrupting the timing and regulation of gene expression.

HO has the unusual feature of being expressed exclusively in only one of the two progeny following mitotic division. At HO, this asymmetric expression depends in part upon the inhibitory Ash1 factor that is expressed in M phase and is asymmetrically localized to daughter cells (Bobola et al. 1996; Sil and Herskowitz 1996; Takahata et al. 2011; Stillman 2013). However, a multigated promoter is also required for mother-specific expression, since removal of nucleosomes from URS2-L allows HO expression in daughters. The HO promoter is set up to require Swi5 to act before cell separation and SBF to act after separation, and this sequential activation is required to block expression in daughter cells. This asymmetric expression of HO has been compared to maintenance and differentiation of stem cells (Herskowitz 1989), where asymmetry in gene expression is crucial for allowing one product of mitosis to maintain stemness while the other takes on a different fate (Knoblich 2008; Inaba and Yamashita 2012). Our results indicate how multigated promoters can produce this form of differential regulation.

Nucleosomes have long been known to be repressive, and removing a nucleosome at a core promoter can allow for coactivator-independent activation (Zhang and Reese 2007). The mechanism by which nucleosomes interfere with long-range transcriptional activation is unclear and remains an important area for future study. Swi5 and its coactivators do not directly activate transcription at the TATA because the distance between them is too great, but instead they bind to URS1 and act at nearby sites in URS2-L. In the native promoter, SBF sites at URS2-L serve as an essential relay from upstream activation, promoting SBF binding at URS2-R, and mutation of these URS2-L SBF sites results in loss of both SBF binding to URS2-R and gene expression (Yarrington et al. 2015). However, in the nucleosome-depleted urs2L:CLN2-mutSBF-wtNDR chimera, mutation of the URS2-L SBF sites does not decrease transcription, and increased SBF binding is seen (Figure 6E). Chromatin remodeling originating from URS1 normally induces chromatin changes that facilitate strong SBF binding for a limited distance, extending only to URS2-L (Takahata et al. 2011; Yarrington et al. 2015). We speculate that when the intervening nucleosomes normally present at URS2-L are absent, coactivators that would usually disassemble chromatin only close by at URS2-L are able to affect nucleosomes further downstream; this allows URS1 events to promote chromatin disassembly at URS2-R. In this model, nucleosome depletion at URS2-L in our chimeric promoter expands the limited chromatin neighborhood conducive for SBF binding further downstream. Significantly, this neighborhood expansion circumvents the SBF transcription factor cascade from URS2-L to URS2-R that is normally required at URS2 for HO expression (Yarrington et al. 2015).

NDRs are present at nearly all eukaryotic promoters, but usually at sites immediately adjacent to the TSS (Yuan et al. 2005; Lee et al. 2007; Mavrich et al. 2008a,b; Schones et al. 2008; Kaplan et al. 2009). At some genes, such as CLN2, NDRs are present upstream in the promoter region and play a role in facilitating transcription factor recruitment (Morse 2007; Bai et al. 2010). The rules governing NDR formation are presently unclear but, significantly, our results suggest that NDRs in the promoter can have profound effects on transcriptional regulation. Complex promoters require individual conditions to be satisfied independently, and our results show that nucleosomes form the barriers that compartmentalize these separate steps to ensure appropriate timing and activation of expression. Sequential promoter activation has been seen at important promoters in metazoans (Cosma 2002), and knowledge of how nucleosomes can create a requirement for ordered recruitment of factors will be important for studying these challenging regulatory mechanisms.