Poly(dA:dT) Tracts Differentially Modulate Nucleosome Remodeling Activity of RSC and ISW1a Complexes, Exerting Tract Orientation-Dependent and -Independent Effects

Abstract

The establishment and maintenance of nucleosome-free regions (NFRs) are prominent processes within chromatin dynamics. Transcription factors, ATP-dependent chromatin remodeling complexes (CRCs) and DNA sequences are the main factors involved. In Saccharomyces cerevisiae, CRCs such as RSC contribute to chromatin opening at NFRs, while other complexes, including ISW1a, contribute to NFR shrinking. Regarding DNA sequences, growing evidence points to poly(dA:dT) tracts as playing a direct role in active processes involved in nucleosome positioning dynamics. Intriguingly, poly(dA:dT)-tract-containing NFRs span asymmetrically relative to the location of the tract by a currently unknown mechanism. In order to obtain insight into the role of poly(dA:dT) tracts in nucleosome remodeling, we performed a systematic analysis of their influence on the activity of ISW1a and RSC complexes. Our results show that poly(dA:dT) tracts differentially affect the activity of these CRCs. Moreover, we found differences between the effects exerted by the two alternative tract orientations. Remarkably, tract-containing linker DNA is taken as exit DNA for nucleosome sliding catalyzed by RSC. Our findings show that defined DNA sequences, when present in linker DNA, can dictate in which direction a remodeling complex has to slide nucleosomes and shed light into the mechanisms underlying asymmetrical chromatin opening around poly(dA:dT) tracts.

1. Introduction

Regulation of gene expression at the level of transcription requires the coordinated contribution of a large number of factors. Among these are ATP-dependent chromatin remodeling complexes (CRCs), which alter the chromatin landscape at defined genomic regions by evicting histone octamers, altering nucleosome positioning and/or altering their composition [1]. In this context, the position of nucleosomes at gene regulatory regions plays a key role in transcriptional regulation. Genome-wide analyses of nucleosome positioning have uncovered a common chromatin landscape, consisting of a well-positioned nucleosome, called the +1 nucleosome, covering the transcription start site (TSS) or a few base pairs downstream of this site, followed by a nucleosome array that covers part of the gene body. A long stretch of DNA of low nucleosome occupancy is commonly observed upstream of the +1 nucleosome, called the nucleosome-depleted region (NDR) or the nucleosome-free region (NFR). These terms have been commonly used as equivalent terms, and we do so in the current report, although it has to be noted that differences between them have been described [2]. The typical chromatin landscape in the boundaries of TSSs is also completed by a further well-positioned nucleosome located upstream of the NFR, called the −1 nucleosome [1,2].

It is currently understood that the players involved in the dynamics of nucleosome positioning are transcription factors, CRCs and DNA sequences [1,2,3,4]. Within this context, it has been observed that defined CRCs play opposite roles in gene-promoter NFRs. Particularly in the yeast Saccharomyces cerevisiae, the RSC complex plays a role in the establishment and maintenance of the NFR, while the ISW1a complex performs NFR shrinking [5]. In addition, other CRCs have been highlighted as relevant players in NFR maintenance and positioning of the +1 nucleosome in this species, such as Ino80 and ISW2 [6,7]. Regarding the involvement of DNA sequences in nucleosome positioning, early genome-wide analyses assigned them a major role. It was then suggested that this role relies essentially on variations in the flexibility of different DNA sequences, which drive histone octamers to preferential translational locations [8,9]. Further studies using in vitro reconstitution of the whole S. cerevisiae genome into nucleosomes determined that the obtaining of an in vivo-like nucleosome positioning pattern depends essentially on active processes. Thereafter, these studies pointed to transcription factors and CRCs, notably RSC, as the main players involved in the processes determining nucleosome positioning [10,11,12,13]. A more marginal and passive role was then assigned to DNA sequences in nucleosome positioning. This passive role of DNA sequences would include mechanical (intrinsic) properties, which determine variations in the flexibility of different sequences, and the presence of binding sites for transcription factors that participate in nucleosome remodeling processes [12,13]. Afterward, however, a growing number of studies have pointed to DNA sequences as relevant players in the active processes involved in the dynamics of nucleosome positioning. Until now, the participation of DNA sequences in these active processes has been essentially postulated based on the observed effects of poly(dA:dT) tracts on ATP-dependent nucleosome remodeling activity [4,14,15,16].

The presence of poly(dA:dT) tracts at gene regulatory regions in S. cerevisiae was reported a few years after the onset of DNA sequencing. These studies and further analyses demonstrated the involvement of these sequences mainly in transcriptional regulation of constitutive genes, but also in inducible transcription [17,18]. Afterward, an abundant number of in vivo and in vitro genome-wide studies have shown the involvement of these tracts in NFR formation and maintenance [1,4]. It has been determined that these sequences enhance nucleosome disassembly mediated by the RSC complex [14] and influence the nucleosome remodeling activity of CHD1 [16]. Intriguingly, it has been observed that NFRs harboring a single poly(dA:dT) tract span a region that is asymmetrically distributed around the tract, with a longer stretch of the NFR located 5′ from the polyA sequence [19]. In the same context, there is a higher frequency of polyT in the sense strand upstream of the midpoint of gene-promoter NFRs, while there is an enrichment of polyA in the sense strand downstream of this midpoint, with the presence of both patterns at individual gene promoters being less frequent than expected [20]. Although it is known that RSC has the ability to asymmetrically clear chromatin around poly(dA:dT) tracts [15], the means by which this biochemical outcome is obtained are currently unknown. In addition, there are no studies comparing the effects of the two alternative orientations of poly(dA:dT) tracts on ATP-dependent nucleosome remodeling activity.

In order to obtain insight into the role played by poly(dA:dT) tracts in the active processes involved in nucleosome positioning, we performed a systematic analysis of the effect of these sequences on ATP-dependent nucleosome remodeling activity. Considering their stimulatory effect on RSC remodeling activity, we aimed to obtain further insight into the effects of these tracts on the activity of this complex, specifically RSC2 [21]. In addition, taking into account the opposite roles played by ISW1a and RSC at NFRs, we also aimed to determine whether these tracts affect the activity of ISW1a and, if so, whether they differentially affect the activity of RSC and ISW1a complexes. For this, we used a large variety of nucleosome probes harboring these tracts in different orientations and positions relative to the nucleosome core, in addition to probes harboring different lengths of the tracts. Our analyses were mainly focused on the effect displayed by these tracts when located on linker DNA considering the following criteria: Firstly, poly(dA:dT) tracts of the S. cerevisiae genome are mainly located at NFR or linker regions, not only in vivo, but also upon in vitro reconstitution of this genome into nucleosomes [9,15]. Secondly, interaction with linker DNA has been described for both complexes, including mechanistic implications of these interactions in the remodeling process. For ISW1a, as for other complexes of the ISWI subfamily, the interaction with linker DNA stimulates its remodeling activity and makes this linker the entry DNA in the sliding process [1,22,23]. For the RSC complex, recent structural analyses have characterized a linker DNA interaction module (DIM) in the complex, with this linker DNA being the exit DNA during nucleosome remodeling [24,25]. Our findings revealed that poly(dA:dT) tracts differentially affect the activity of ISW1a and RSC complexes. Moreover, we found that the two alternative orientations of poly(dA:dT) tracts exert differential effects on nucleosome remodeling activity. In the case of RSC, this differential effect implies different outcomes of RSC’s activity. In addition, we found that poly(dA:dT) tracts orient RSC to select the tract-containing linker as the exit DNA during remodeling. These new properties of poly(dA:dT) tracts might constitute the underlying mechanism of the asymmetric chromatin opening that occurs around these sequences.

2. Results

2.1. Poly(dA:dT) Tracts Exert a Differential Effect on the Activity of ISW1a and RSC Complexes

In this study we used a large number of nucleosome probes that harbor different designs, mainly in terms of location and orientation of poly(dA:dT) tracts (see Table S1 for details). To clearly state the features of every probe, including the location and orientation of tracts, we use a set of nomenclature guidelines throughout the whole text. One strand of Widom’s 601 pronucleosomal sequence is arbitrarily defined as the “upper strand” (Figure S1); in some sections, particularly in the Discussion, the upper strand is considered equivalent to the sense strand of genes. The location of a poly(dA:dT) tract is defined as upstream or downstream of the nucleosome core (or dyad) relative to the upper strand when referring to a tract as a whole double-strand entity and also when referring to only one of the tract sequences (polyA or polyT). The orientation of the poly(dA:dT) tract is defined by indicating the presence of polyA (or polyT) in the upper or lower strand. Tract orientation is also defined in terms of the location of the nucleosome core or dyad relative to the 5′ or 3′ end of the polyA sequence (see a summary of these denominations in Figure S1). For ease, extra-core DNA (also known as extranucleosomal DNA) is referred to as “linker DNA” or “linker” throughout the text. Nucleosome probes are named including their main features in the names, such as the location of nucleosome core, the presence and length of linker DNA, and the presence, length and orientation of poly(dA:dT) tracts. These elements are named relative to the 601 upper strand, following its 5′ to 3′ direction (i.e., from left to right). The nucleosome core is designated as “NC”.

In order to ascertain whether poly(dA:dT) tracts differentially affect the activity of RSC and ISW1a complexes, together with testing the two alternative orientations of these tracts, we first performed remodeling assays comparing a set of three probes reconstituted into mononucleosomes. These 217 bp probes harbor the 147 bp positioning region of the 601 sequence (or the first 145 bp of this sequence) to position the histone octamer upon nucleosome reconstitution, plus 70 (or 72) bp downstream of this sequence. These 70 bp become linker DNA after reconstitution (0-NC-70 probes, Figure 1A, left panel). Two of these probes contain a 15 bp poly(dA:dT) tract located immediately downstream of the 145 bp 601 sequence; one of them harbors the polyA sequence in the upper strand and the other one harbors this sequence in the lower strand. Under this design, most of the tract becomes part of linker DNA upon nucleosome reconstitution (Figure 1A). The sliding activity of these complexes was visualized by electrophoresis in non-denaturing gels, followed by quantification of the slid-over total mononucleosome signal ratio (Figure 1B). In the case of RSC, its sliding activity involves the generation of a fast-migrating nucleosomal band that has been characterized as entailing mobilization of the histone octamer up to 51 bp beyond a DNA end [26]. Interestingly, the presence of these tracts hindered the sliding activity of ISW1a (Figure 1C, compare lane 2 to 4 and 6, and the corresponding probes in the graph). Moreover, we found that there is a difference between the extent of inhibition given by the orientation of the homopolymeric tract, with the strongest inhibition corresponding to the presence of polyA in the lower strand (Figure 1C). Conversely, in the case of RSC—as expected from previous studies [14]—the presence of the tracts enhanced its sliding activity (Figure 1D, compare lane 4 to 6 and 8, and the corresponding probes in the graph). Here, we also found a differential effect given by the orientation of the tract, with the strongest stimulation observed in the probe harboring polyA in the lower strand (Figure 1D). This difference in the stimulatory effect on RSC’s activity only applies to its sliding activity, as determined by the ratio of slid-over total mononucleosome signal, and not to other biochemical outcomes that are addressed below. Thus, the orientation that exerted the strongest stimulatory effect on the sliding activity of RSC was the same which resulted in the strongest inhibitory effect on the sliding activity of ISW1a (Figure 1C and Figure 1D, respectively).

In order to confirm these effects of poly(dA:dT) tracts, we performed the same analysis on probes harboring linker DNA and the poly(dA:dT) tract on the other side of the 601 sequence (70-NC-0 probes, Figure 1A, right panel). Here, it is worth remembering that the nucleosome has a two-fold symmetry axis [27,28]. Hence, opposite tract orientations—in terms of the presence of a homopolymeric sequence in one strand or the other—correspond to equivalent orientations in terms of nucleosome structure when one tract is upstream of the dyad and the other is downstream of this axis. Considering this, a more structurally comprehensive way to describe the two possible orientations of poly(dA:dT) tracts corresponds to “core (or dyad) located on the 5′ side of polyA” and “core (or dyad) located on the 3′ side of polyA” (Figure S1). Thus, we reasoned that, in this new set of probes, the strongest stimulatory effect on RSC’s sliding activity (and the strongest inhibitory effect on ISW1a’s activity) should be given by polyA in the upper strand rather than in the lower strand. Of note, the 147 bp 601 sequence is not palindromic; in addition, in Widom’s original 601 plasmid vector as well as in our derivatives, the DNA sequence upstream of the 601 positioning region is different from that present downstream of this region. In this respect, it has been observed that the 601 sequence biases the directionality of RSC’s sliding activity [29]. Thus, the analysis using probes harboring linker DNA upstream of the 601 sequence was also designed to ascertain whether poly(dA:dT) tracts exert their influence on RSC and ISW1a’s activity, overcoming these differences between the upstream and downstream DNA sequences relative to the 601 axis. The results of the assays performed with this new set of probes confirmed the stimulatory effect of poly(dA:dT) tracts on RSC’s sliding activity, as well as the inhibitory effect on ISW1a’s sliding activity. Importantly, as expected from the two-fold symmetry axis of the nucleosome, the strongest effect (stimulatory for RSC and inhibitory for ISW1a) was given by the presence of polyA in the upper strand (Figure 1C,D). Thus, considering both sets of probes, the tract orientation that exerts the strongest stimulatory effect on RSC’s sliding activity (and the strongest inhibitory effect on ISW1a’s activity) corresponds to the core located 3′ from polyA. Although differences in the extent of sliding activity were observed when comparing the control probe harboring linker DNA downstream of the core to the control harboring linker upstream of the core, it did not affect the differential pattern given by tract orientation.

It has been previously shown that poly(dA:dT) tracts located within the nucleosome core stimulate the remodeling activity of RSC [14]. Considering this earlier finding, we tested the effect of a 15 bp tract whose midpoint is placed on superhelix location (SHL) 5.5 (Figure S2A). By placing a poly(dA:dT) tract on this location, we observed no effect on nucleosome stability, relative to the original 601 sequence, in terms of the extent of nucleosome reconstitution (see below). In this set of probes, the sliding activity of ISW1a remained mostly unaltered (Figure S2B). On the other hand, a stimulatory effect on RSC’s sliding activity was observed, although this effect was less pronounced than that found on probes harboring the tract on linker DNA, and there were no differences between the two alternative tract orientations (Figure S2C). Further analyses, using different sets of probes, showed additional effects on ISW1a and RSC’s activity given by tracts located within the nucleosome core (see below).

We next searched for a minimal tract length still having an impact on RSC and/or ISW1a’s sliding activity, considering that an inverse correlation between the length of the poly(dA:dT) tracts and nucleosome occupancy has been reported in previous studies [30]. In addition, in vivo and in vitro studies have found a nucleosome depletion effect of poly(dA:dT) tracts when testing lengths starting from 5 bp [15,19]. First, 15, 10 and 5 bp lengths were tested (Figure 2A). Our results showed that the inhibitory effect on ISW1a’s sliding activity, as well as the stimulatory effect on RSC’s activity, were the same when comparing 15 bp to 10 bp length. However, when reducing the length of the tract to 5 bp, the effect was lost for both complexes (Figure 2B,C). We then tested the effect of a 7 bp tract, finding an inhibitory effect on ISW1a’s activity, although weaker than that given by 15 bp length. In the case of RSC, a slight, non-statistically significant, stimulatory effect was observed for 7 bp tract length over the control probe (Figure S3). Next, we analyzed the effect of placing progressively longer DNA stretches (0, 19 and 40 bp) between the nucleosome core and a 15 bp poly (dA:dT) tract. In this respect, interactions with linker DNA have been described for both ISW1a and RSC complexes, affecting their interaction with the nucleosome and their remodeling activity [22,24,25]. As observed in Figure 2E, in the probes harboring a separation of 19 and 40 bp between the core and the tract, the inhibitory effect on ISW1a’s activity was lost. Indeed, even a slight statistically significant stimulatory effect was observed for the probe harboring a 40 bp separation. The analyses performed for RSC showed no variation in the stimulatory effect when comparing 0 and 19 bp separations. In the case of the probe harboring a 40 bp separation between the core and the tract, there was still stimulation of RSC’s activity, although weaker than that observed for shorter separations (Figure 2F).

2.2. Poly(dA:dT) Tracts Exert Stronger Enhancement of RSC Nucleosome Disassembling Activity When the Nucleosome Core Is Located 5′ of PolyA

In order to obtain insight into the mechanisms underlying the differential stimulatory effect given by the two alternative orientations of poly(dA:dT) tracts, we analyzed the influence of these sequences on the strength of RSC–nucleosome interaction. For this, we performed electrophoretic mobility shift assays (EMSA) using 0-NC-70 probes, previously used to test RSC and ISW1a’s activity (Figure 1). The effect on the strength of ISW1a–nucleosome interaction was also tested. For these assays, complex concentrations higher than those used in previous remodeling assays were used in order to obtain observable binding levels. Binding strength was tested in the presence of ATP or in the presence of the non-hydrolyzable analog ATP-γ-S. As expected from previous studies [31,32], the presence of ATP enhanced the binding strength of both complexes. This enhancement was observed in the absence and in the presence of poly(dA:dT) tracts (Figure 3A,B). The binding of ISW1a to the mononucleosome probe was only detected in the presence of ATP, and no major differences were observed between the presence and absence of tracts (Figure 3A). On the other hand, the binding of RSC to nucleosome probes harboring a tract was significantly stronger than that observed for the control probe under both conditions—in the presence of ATP-γ-S (Figure 3B, compare lane 2 to 5 and 8) and in the presence of ATP (Figure 3B, compare lane 3 to 6 and 9). Interestingly, RSC displayed a higher interaction strength for the probe harboring the core 5′ of polyA than for the probe harboring the core 3′ of polyA (Figure 3B, compare lane 6 to 9). Binding assays performed using the same probes in the form of naked DNA showed stronger binding of RSC to probes harboring a tract as compared to the absence of a tract, with no differences in binding strength between the two tract orientations (Figure S4), indicating that the orientation-dependent differential binding strength arises in the context of the nucleosome structure. At first glance, the stronger interaction of RSC with the probe harboring core 5′ of polyA was not consistent with the stimulation of its sliding activity given by the tracts, which was stronger for the probes harboring core 3′ of polyA than for the other orientation (Figure 1). However, electrophoretic analysis in native gels performed in several nucleosome remodeling studies has shown that incubation of mononucleosome probes with RSC or other complexes of the SWI/SNF subfamily results in a reduction in the total signal of the mononucleosomal species, accompanied by the appearance of smear or discrete bands reflecting biochemical outcomes such as nucleosome dimers [33,34,35,36]. It has been proposed that the appearance of these remodeled species corresponds to nucleosomes trapped in structures corresponding to unfinished octamer transfer [35,37]. In this context, we reproducibly observed the generation of smear and the appearance of a signal in the loading well to a higher extent for probes harboring the orientation corresponding to core 5′ of polyA. Indeed, when considering the smear signal as part of the remodeled species generated by RSC’s action, the two alternative tract orientations displayed stimulation of RSC’s activity at more similar levels (Figure S5). Taking these aspects into account, we reasoned that the stronger binding of RSC to probes harboring core 5′ of polyA might be connected to a stronger octamer transfer activity of this complex, which could result in nucleosome probe disassembly (also known as nucleosome eviction). This outcome would be reflected by an increment in the free DNA probe. To test this possibility, we performed remodeling assays using the same high concentrations of RSC used in the binding assays, analyzing the level of nucleosome disassembly obtained for the two alternative tract orientations. Two sets of probes harboring linker DNA downstream of the core were tested; one set harbors a 70 bp linker DNA (0-NC-70 probes) and the other a 30 bp linker (0-NC-30 probes). For both probe sets, a higher level of nucleosome disassembly was observed for the core located 5′ from polyA, compared to probes harboring no tract and to probes harboring the other tract orientation (Figure 3C,D). In 0-NC-70 probes, the orientation corresponding to core 3′ of polyA, resulted in a level of nucleosome disassembly similar to that found in the absence of a tract, while in 0-NC-30 probes, this orientation resulted in a level of disassembly higher than that observed in the absence of a tract. However, this disassembly level was still lower than that found for the orientation corresponding to core 5′ of polyA (Figure 3D, compare lane 7 to 10). Taken together, these results show that a poly(dA:dT) tract exerts a stimulatory effect that is stronger under the orientation corresponding to the core located 5′ from polyA than under the opposite orientation in terms of stimulation of RSC binding to the nucleosome and eviction activity.

2.3. In a Nucleosome Harboring Linker DNA on Both Sides of the Core but a Poly(dA:dT) Tract on Only One Linker, RSC Preferentially Selects This Linker as Exit DNA for Sliding

As mentioned in the Introduction, detailed structural analyses have recently characterized a module in RSC responsible for interaction with linker DNA (DIM) upon binding of this complex to the nucleosome. Thus, the presence of linker DNA on only one side of the nucleosome core drives directional binding of RSC, making this DNA entry/exit site of the nucleosome the exit DNA in the sliding process catalyzed by the complex. In light of this property of RSC, it has been proposed that this complex interacts with linker DNA that is turned into an NFR at gene promoters, allowing the complex to push the −1 or +1 nucleosomes away from the center of the NFR being generated or expanded [24,25]. However, these studies did not analyze what differential properties present in NFR’s DNA make it more likely for the DIM to preferentially bind to this linker in detriment of the linker DNA located on the other side of the nearest nucleosome core. Properties such as the presence of binding sequences characterized for RSC subunits [38] and a longer length of the linker/NFR to be extended, relative to upstream and downstream linkers before RSC arrival, would be among differential properties driving directional binding of RSC. In this context, we asked whether the presence of a poly(dA:dT) tract in only one linker of a mononucleosome harboring the same length of linker DNA on both sides could drive the directional sliding activity of RSC. To answer this question, we performed sliding assays using probes harboring 35 bp of linker DNA on both sides of the nucleosome core (35-NC-35) and a poly(dA:dT) tract located right upstream or downstream of the core, assessing the directionality of the sliding activity by digestion with restriction enzymes after the remodeling step and after removal of the complex (Figure 4A). The two alternative tract orientations were tested for probes harboring a tract upstream of the core as well as for probes harboring a tract downstream of the core (Figure 4B). As in our previous analyses performed on probes harboring linker DNA on only one side of the core, the presence of tracts stimulated the sliding activity of RSC, with probes harboring core 3′ of polyA giving the strongest stimulation in terms of slid-over total mononucleosome signal (Figure S6).

We first tested a probe harboring the tract downstream of the core, with polyA in the lower strand (35-NC-35T15). As observed in Figure 4C, on this probe, RSC mainly mobilized the histone octamer upstream, reflected by minimal digestion of the slid nucleosome by PmlI (cut probe signal mainly coming from the small fraction of naked DNA present in the probe) and strong digestion by MfeI (Figure 4C, compare lane 8 to 9). We then tested a probe harboring the tract upstream of the core, with polyA in the upper strand (35A15-NC-35). Using this probe, RSC mainly mobilized the histone octamer downstream, reflected now by strong digestion of the slid nucleosome by PmlI and minimal digestion by MfeI (Figure 4D, compare lane 8 to 9). These results demonstrate that the presence of the poly(dA:dT) tract determines the selection of the exit DNA for the sliding activity catalyzed by RSC, ensuring mobilization of the histone octamer away from the tract. Importantly, the same results were obtained when testing the opposite tract orientation (core 5′ of polyA, Figure 4E,F), indicating that sliding of the nucleosome core away from the tract by RSC is independent of tract orientation. This effect of poly(dA:dT) tracts on RSC’s sliding activity was also confirmed using mononucleosome probes harboring linker DNA on only one side of the core (Figure S7).

Regarding the effects of poly(dA:dT) tracts on the directionality of sliding activity, we asked whether these tracts hinder ISW1a’s activity when they are not located in the way of the histone octamer movement during the sliding reaction catalyzed by this complex. In this regard, it has been demonstrated that ISW1a is highly sensitive to the presence and length of linker DNA and that this complex mobilizes the histone octamer towards the linker DNA to which it is bound [22]. In the case of in vitro assays performed using mononucleosome probes harboring linker DNA on only one side of the nucleosome core, the complex mobilizes the histone octamer towards a central position in the DNA segment. Thus, in this type of probe, ISW1a attempts to mobilize the histone octamer towards the poly(dA:dT) tract if this sequence is present in the linker DNA, as was the case of our initial analyses performed using nucleosome probes harboring a 70 bp linker DNA on only one side of the core (Figure 1, Figure 2 and Figure S3). In order to assess whether poly(dA:dT) tracts still inhibit ISW1a’s activity when located on linker DNA that becomes exit DNA upon ISW1a engagement, we tested probes harboring a long linker upstream of the core and a tract-containing short linker downstream of the core (70-NC-30 probes, Figure 5A). Under this design, poly(dA:dT) tracts did not hinder ISW1a’s sliding activity; in fact, there was a slight stimulation of its activity (Figure 5B,C). The same result was obtained using a probe harboring the shorter linker and a tract upstream of the nucleosome core (Figure S8).

2.4. A Design That Forces RSC to Mobilize a Tract across the Nucleosome Core Results in Blocking of Its Sliding Activity by One Tract Orientation

Our finding that RSC’s sliding activity always proceeds by mobilizing the nucleosome core away from the poly(dA:dT) tract, independently of tract orientation, led us to ask what would be the effect of a tract on the sliding activity of RSC if the complex is forced to mobilize the tract across the nucleosome core. To test this scenario, we generated mononucleosome probes harboring linker DNA upstream and downstream of the core, with both linkers being of the same length and both harboring a 15 bp poly(dA:dT) tract (35-NC-35 probes, Figure 6A and Figure S9A). We first analyzed a probe harboring the tract orientation that results in the strongest stimulatory effect on RSC’s sliding activity, corresponding to the core located 3′ from polyA (35A15-NC-35T15, Figure S9A). As expected, the presence of tracts on both sides of the core resulted in a stronger stimulatory effect on RSC’s sliding activity, as compared to the presence of only one tract (Figure S9). Intriguingly, RSC’s action on this probe did not result in the generation of the fastest migrating nucleosomal band. However, its action did result in the generation of this remodeled species in centrally positioned nucleosomes harboring either one tract (Figure S9) or two tracts in the opposite orientation (core 5′ of polyA, 35T15-NC-35A15, Figure 6A), and even in the control probe under longer incubations (Figure S9D). As mentioned above, it has been previously determined that the remodeled state present in this fastest migrating band corresponds to a nucleosome harboring a histone octamer that has slid up to 51 bp beyond a DNA end, resulting from the action of RSC; this outcome extends to other complexes of the SWI/SNF subfamily [26,39]. Other studies have also found the fastest migrating band reflecting the loss of one H2A-H2B heterodimer [40,41], an effect that would also involve mobilization of the histone octamer beyond a DNA end [40]. Thus, in light of the result obtained in the sliding assays testing the probes harboring poly(dA:dT) in both linkers, we reasoned that there might be a region inside the nucleosome core where the presence of a tract in one specific orientation (dyad 3′ of polyA) could make the sliding activity of this complex more difficult. Thereafter, in a probe harboring longer tract-containing linkers, with both tracts in the orientation core 3′ of polyA, the sliding reaction would not only result in the absence of the fastest migrating band, but even in the absence of bands reflecting relocation of the histone octamer from the starting central position to positions reaching DNA ends. To test this prediction, we performed sliding assays using 70-NC-70 probes (Figure 6B). Under this configuration, where each tract is 2 bp inside the core after nucleosome reconstitution and there are 70 bp for histone octamer mobilization in cis until it reaches a DNA end, octamer mobilization reaching these ends forces one of the tracts to travel all the way up to the nucleosome dyad. As predicted, RSC’s action on the probe harboring the configuration consisting of core located 3′ from each polyA sequence resulted in the mobilization of the histone octamer to translational positions that do not reach DNA ends (Figure 6B, lane 4). On the other hand, RSC’s action on the probe harboring the opposite orientation in both tracts did result in the relocation of the core to translational positions reaching DNA ends (Figure 6B, lane 6). For this probe, at least two faster-migrating bands are obtained, relative to the main band obtained for the probe harboring the orientation core 3′ of polyA (Figure 6B, compare lane 4 to 6). An equivalent effect was found for ISW1a using a different set of probes (Figure S10).

Considering that poly(dA:dT) tracts have been described as stiff DNA sequences that negatively affect nucleosome stability [42,43], we asked whether this property of the tracts would be the factor underlying this particular effect on RSC’s sliding activity. If so, one tract orientation should display a deeper negative impact on nucleosome stability. To test this possible mechanism, we performed competitive nucleosome reconstitution analyses [44] on 147 bp DNA segments (Figure 6C) harboring a 15 bp tract in one of the two alternative orientations and one of two alternative locations: the midpoint of the tract 15 bp downstream of the 601′s dyad (A15-SHL1.5 and T15-SHL1.5) and the midpoint 54 bp downstream of the 601′s dyad (A15-SHL5.5 and T15-SHL5.5). Regarding this, it has been determined that the regions of higher distortion of DNA in the nucleosome structure might vary with changes in DNA sequence [45,46,47]. In the case of the 601 sequence, the regions of higher distortion of DNA correspond to SHL1.5 and SHL4.5 [45]. Therefore, in one set of probes, the tract covers a region that, upon reconstitution, corresponds to one of the points of higher distortion for the 601 sequence, while in the other set, the tract is located near to but does not overlap with the other region of higher distortion. Our results showed that locating a tract in SHL5.5 has no effect on nucleosome stability, but placement on SHL1.5 resulted in a marked reduction in nucleosome stability, with an 8- to 10-fold reduction in the relative equilibrium constant of nucleosome formation (Figure 6C). Importantly, no differences in reduction in nucleosome stability were found when comparing the two alternative tract orientations in any of the locations tested (Figure 6C), discarding the effects of poly(dA:dT) tracts on nucleosome stability as the mechanism underlying this differential effect exerted by the two alternative tract orientations on RSC’s sliding activity.

2.5. Poly(dA:dT) Tracts Globally Influence the Directionality of ISW1a Sliding Activity

The results of our analyses testing the effect of poly(dA:dT) on RSC’s activity are in agreement with the general trend observed in vivo for single-tract-containing NFRs, which shows a longer stretch of the NFR located 5′ from polyA [19], and with in vitro genome-wide analyses performed by Krietenstein and co-workers [15], which show that RSC displaces nucleosomes asymmetrically around these tracts (see the Discussion for further details). In addition, this study shows that ISW1a displaces nucleosomes symmetrically around poly(dA:dT) tracts, with displacement of surrounding nucleosomes away from the tract. However, an analysis of the directionality of ISW1a’s sliding activity in terms of the presence versus the absence of poly(dA:dT) tracts was not performed. On the other hand, it has been previously demonstrated that the in vivo action of ISW1a results in nucleosomes populating gene-promoter NFRs [5]. In this context, the results of our assays testing ISW1a’s activity predict that ISW1a would act by reducing the length of non-tract-containing NFRs and expanding the length of tract-containing NFRs. To test this prediction, we analyzed publicly available data generated by Krietenstein and co-workers, who performed genome-wide reconstitution of the S. cerevisiae genome into nucleosomes and determined nucleosome repositioning mediated by several factors and complexes, including ISW1a [15]. From these data, we determined the frequency of polyA and polyT tracts in the sense strand of gene promoters that displayed mobilization of the −1 nucleosome by ISW1a in at least 20 bp. Three clusters were generated: “all genes”, corresponding to the full gene promoter coverage of the study (4064 genes); “upstream mobilization”, corresponding to gene promoters displaying ISW1a-mediated upstream mobilization of the −1 nucleosome (1793 genes); and “downstream mobilization”, corresponding to gene promoters displaying ISW1a-mediated downstream mobilization of the −1 nucleosome (854 genes). As the analysis searched for the presence of poly(dA:dT) tracts that might influence ISW1a’s activity, the frequency of poly(dA:dT) tracts was determined relative to the pre-existing position of the −1 nucleosome found in the absence of ISW1a (termed SGD condition). The same analysis was performed for the +1 nucleosome; in this case, 890 genes displayed ISW1a-mediated upstream mobilization and 1543 displayed downstream mobilization of this nucleosome. As expected from previous studies, there is a high frequency of poly(dA:dT) tracts right downstream of the −1 nucleosome and right upstream of the +1 nucleosome under the SGD condition, together with a low frequency near the −1 and +1 dyads, which reflects the translational location passively adopted by histone octamers upon reconstitution (Figure 7A,B, respectively). Interestingly, in line with our prediction, for those gene promoters that displayed upstream mobilization of the −1 nucleosome by ISW1a, the frequency under the SGD condition is higher right downstream of the −1 nucleosome core (Figure 7A). Conversely, for the cluster that displayed downstream mobilization of the −1 nucleosome, the frequency of poly(dA:dT) tracts under the SGD condition is lower right downstream of the −1 nucleosome core relative to the other clusters (Figure 7A). The equivalent result was obtained for the +1 nucleosome, where a low frequency of poly(dA:dT) tracts—relative to the other clusters—was found in the cluster of genes that displayed upstream mobilization of the nucleosome by ISW1a in the region right upstream of the +1 nucleosome core (Figure 7B). Interestingly, a higher frequency of poly(dA:dT) tracts downstream of this nucleosome can be observed for this cluster, relative to the frequency observed for the other clusters in that region (Figure 7B). Taken together, our bioinformatics analyses show that in those genes displaying ISW1a-mediated nucleosome mobilization towards the NFR, for both −1 and +1 nucleosomes, there is a lower frequency of poly(dA:dT) tracts in this region.

3. Discussion

In this work, we performed a systematic analysis addressing the effect of poly(dA:dT) tracts on the nucleosome remodeling activity of ISW1a and RSC complexes, finding new properties of these DNA sequences in terms of their influence on ATP-dependent nucleosome remodeling. These properties are as follows: (i) tracts enhance RSC binding to nucleosomes and dictate what linker will be the exit DNA in the sliding process, resulting in nucleosome core mobilization away from the tracts; (ii) although by different means, these sequences also impose sliding of the nucleosome core away from them to the ISW1a complex; (iii) tracts enhance RSC eviction activity when being in one of the two possible orientations relative to a neighboring nucleosome core; and (iv) in a scenario where RSC is forced to mobilize a poly(dA:dT) tract inside the nucleosome core, one tract orientation blocks the sliding activity of this complex.

3.1. Poly(dA:dT) Tracts Exert Orientation-Dependent and -Independent Effects on RSC Activity

Our findings show that, when a poly(dA:dT) tract is present in linker DNA, one of its orientations (core 5′ of polyA) exerts a stronger stimulation on RSC binding to the nucleosome and, consistently, a stronger stimulation of its nucleosome disassembly activity. On the other hand, the alternative orientation (core 3′ of polyA) exerts a strong stimulation of RSC’s sliding activity. These effects correspond to differential orientation-dependent effects exerted by poly(dA:dT) tracts on RSC’s activity. The non-differential effect is entailed in the stimulation of RSC’s sliding activity by the tracts. We found that both orientations stimulate sliding activity and that RSC slides the histone octamer away from the tracts, independently of their orientations. This result also implies that RSC selects the tract-containing linker as the exit DNA during the remodeling process. In this context, as mentioned in the Introduction, recent structural analyses have characterized a linker DNA interaction module (DIM) in the complex, with this linker DNA being the exit DNA during nucleosome remodeling [24,25]. Considering this property of RSC, our findings suggest that both tract orientations act in the same way on RSC, consisting of a stimulatory effect on DIM binding to linker DNA. Then, the stronger binding stimulation given by one tract orientation (core 5′ of polyA) would make a nucleosome more prone to being evicted. In this respect, single-molecule experiments have observed the generation of intranucleosomal loops of DNA during the remodeling process catalyzed by SWI/SNF and RSC [48]. We speculate that these loops might be generated by the accumulation of exit DNA between this border of the nucleosome core and the point where the DIM binds to the exit linker DNA [24,25]. Then, the presence of a poly(dA:dT) tract on linker DNA in the orientation “core 5′ of polyA” would facilitate the generation of this loop and, in turn, histone octamer eviction as a remodeling outcome. Alternatively, or being part of the process, this enhanced binding to linker DNA might result in improving the coupling between DNA translocation and ATP hydrolysis [49].

Another differential effect exerted by poly(dA:dT) tracts on RSC’s activity comes from a design that forces this complex to mobilize a tract across the nucleosome core. Under this special design, which involves the presence of tracts on both borders of the nucleosome core, the sliding activity of RSC is initially stimulated and then blocked by the orientation “core 3′ of polyA”. In this respect, our competitive nucleosome reconstitution analyses showed that the two alternative orientations of poly(dA:dT) tracts affect nucleosome stability to the same extent, discarding this property of tracts as the mechanism underlying this effect. In this context, it has been described that the tracking lobes of the catalytic subunit of most remodeling complexes, including RSC, engage at the SHL2 position and perform tracking in one strand, in the 3′ to 5′ direction [1,50]. This property of RSC and the tract orientation that blocks its sliding activity suggests that for the catalytic subunit of the complex, it is more difficult to track on polyA than on polyT and other sequences. Nevertheless, it is worth recalling that this effect results from a special design. In the presence of a single tract in one of the linkers of a nucleosome probe, RSC selects this linker as the exit DNA to exert its remodeling activity, implying that there is no entrance of the tract into the nucleosome core during the remodeling process. Thus, RSC does not require its motor to perform tracking on poly(dA:dT) sequences for stimulation of its sliding and eviction activities.

3.2. Poly(dA:dT) Tracts Hinder ISW1a Activity When Located in the Way of Histone Octamer Mobilization

As mentioned in the Results section, ISW1a is highly sensitive to the presence and length of linker DNA. Upon binding to the nucleosome, ISW1a preferentially binds to the longer linker and makes this linker (proximal linker) the entry DNA in the sliding process [22]. Our analyses show that when located on this linker, poly(dA:dT) tracts hinder the sliding activity of this complex. Conversely, when placed on the distal linker, there was even a slight enhancement of its activity. The differential effect given by the two alternative tract orientations was essentially observed in the case of hindering of its sliding activity, obtained when the tracts were located in the proximal linker, where the stronger hindering was given by the orientation “core 3′ of polyA”. According to our results, this inhibition of ISW1a’s sliding activity would not be owing to the intrinsic antinucleosomal properties of poly(dA:dT) tracts. Our competitive nucleosome reconstitution analyses showed that poly(dA:dT) tracts do not reduce nucleosome stability when placed near the border of the core and display no orientation-dependent effects on nucleosome stability. In addition, we did not detect an effect of tracts on ISW1a binding to the nucleosome. These results suggest that poly(dA:dT) tracts, when located in the proximal linker DNA, might hinder the productive engagement of ISW1a on the nucleosome, with the orientation “core 3′ of polyA” being more efficient in generating this effect. We also found that the same tract orientation exerts an additional inhibitory effect on the sliding activity of this complex when approaching the nucleosome dyad.

3.3. Effects of Poly(dA:dT) Tracts on ISW1a and RSC at NFRs

Our findings are in accordance with the known role played by poly(dA:dT) tracts at gene-promoter NFRs, consisting of the establishment and maintenance of these regions [4,13], and with the opposite roles played by RSC and ISW1a at these genomic regions, where these complexes aim to widen and shrink these regions, respectively [5]. In addition, our results support a model for the asymmetric chromatin opening that takes place around poly(dA:dT) tracts (Figure 8).

Regarding RSC and asymmetrical chromatin opening around poly(dA:dT) tracts, we propose the following mechanism based on our findings and the known properties of the complex (Figure 8A). First, the DIM of RSC binds to a tract-containing linker, making this linker the exit DNA in the remodeling process. This binding ensures that this linker will be expanded, leading to NFR formation, independent of the adjacent core bound by the ATPase motor of the complex. Now, depending on the core bound by the motor, RSC will mainly expand this linker by sliding (if the core bound is 3′ of polyA) or by eviction of the histone octamer (if the core bound is 5′ of polyA). Chromatin opening would be of a short range in the case of sliding, as the stimulatory effect of the tract decreases as its distance from the core increases. On the other hand, eviction of the histone octamer clears a longer stretch of DNA. A higher processivity of RSC’s sliding activity under the orientation “core 5′ of polyA”, in light of the higher affinity of this complex to the nucleosome under this tract orientation, could also account for a longer range of chromatin opening relative to the other tract orientation. Thereafter, this differential effect on the nucleosome cores located beside the tract would lead to the asymmetrical opening of this linker DNA region (Figure 8A). The same mechanism would explain why it is more common to find polyT upstream of the midpoint of NFRs and polyA downstream of this point at gene-promoter NFRs, as well as low rates of RSC-mediated eviction of −1 and +1 nucleosomes (Figure 8B). Additionally, histone acetylation could also prevent the eviction of −1 and +1 nucleosomes by RSC [51]. Future studies will be required to confirm whether the differential effects given by the two alternative tract orientations lead to differential biochemical outcomes of RSC in vivo (sliding versus eviction), in order to confirm whether this duality is involved in asymmetrical chromatin opening catalyzed by this complex around poly(dA:dT) tracts.

In the case of ISW1a, in vivo studies have found both shrinking and widening of the NFR mediated by this complex [5,52,53]. Regarding these apparently contradictory findings, our remodeling assays and bioinformatics analyses support a view where ISW1a acts by shrinking tract-lacking NFRs and broadening tract-containing NFRs. Interestingly, in those genes that displayed upstream mobilization of the +1 nucleosome by ISW1a (NFR shrinking), there is a higher frequency of poly(dA:dT) tracts right downstream of this nucleosome. This pattern suggests that there might be genes where poly(dA:dT) tracts, located downstream of the +1 nucleosome, could assist ISW1a-mediated mobilization of this nucleosome towards the NFR, with the caveat that positioning of +1 nucleosomes obtained upon in vitro reconstitution of the S. cerevisiae genome does not necessarily match their in vivo positioning. Regarding the differential effect given by the two alternative tract orientations, the higher frequency of polyA in the sense strand closely upstream of the +1 nucleosome and of polyT closely downstream of the −1 nucleosome fits well with the stronger hindering of ISW1a given by the orientation “core 3′ of polyA” observed in our analyses (Figure 8B). In this context, recent studies have pointed out that other CRCs play a more relevant role in the positioning of the +1 nucleosome, such as Ino80 and ISW2 [6,54], making it interesting to study the effects of poly(dA:dT) tracts on nucleosome remodeling activity of these and other CRCs. Regarding this, it has been recently shown that a poly(dA:dT) tract placed near the border of a nucleosome core or on linker DNA exerts an inhibitory effect on Ino80 activity, although the effect of tract orientation was not tested [55]. Finally, it is worth mentioning that other factors could also affect the action of ISW1a at gene-promoter NFRs, such as transcription factors, including general regulatory factors [15,56].

Several relevant questions are raised by our findings. One is whether other DNA sequences exert effects on RSC in a way similar to this found for poly(dA:dT) tracts. Another question refers to the existence of two forms of the RSC complex, RSC1 and RSC2 [21]. Our assays were performed using the RSC2 complex. Thus, it will be of interest to compare the effect of poly(dA:dT) sequences on the RSC1 and RSC2 complexes. An additional relevant question is whether poly(dA:dT) tracts and/or other DNA sequences exert similar effects on CRCs of higher eukaryotes. Interestingly, in the case of mammals, NFRs are also asymmetrically positioned around poly(dA:dT) tracts, but display a longer stretch of NFR on the 3′ side of polyA [19].

Taken together, our findings give insight into the mechanisms by which poly(dA:dT) tracts affect ATP-dependent nucleosome remodeling activity and also support the view that defined DNA sequences play roles in both passive and active processes that govern nucleosome positioning, which in turn affect transcriptional activity and other processes involving access to DNA. In addition, our findings suggest that defined DNA sequences are strategically located at gene promoters and other genomic regions, determining in which direction nucleosome sliding should proceed and even determining the biochemical outcome of the nucleosome remodeling process (sliding versus histone octamer ejection).

4. Materials and Methods

4.1. Protein Complexes

The S. cerevisiae ISW1a and RSC complexes were obtained by tandem affinity purification from Ioc3-TAP and Rsc2-TAP strains, respectively (Table S2), as previously described [57]. Purified complexes were analyzed as detailed in our previous report [58]. For both complexes, a portion of a purification was extensively concentrated (Microcon Ultracel YM-10, Amicon-Millipore, Burlington, MA, USA), quantified by SDS–PAGE/Coomassie staining and then used as standard for Western blot quantification of the purified complex in that and further purifications.

4.2. Plasmids, Probes and Nucleosome Reconstitution

DNA probes of different lengths and harboring the 601 nucleosome positioning sequence [59,60] located at different positions were generated by PCR using distinct plasmids as templates. The 147 bp positioning region of the 601 sequence was defined as previously described [59]. The plasmids and primer sets used are depicted in Table S1. Before PCR amplification, one of the primers used in each reaction was labeled on its 5′ end using [γ-³²P]-ATP (Perkin-Elmer NEG035C or ARC ARP0102B, Perkin-Elmer, Waltham, MA, USA). Nucleosome reconstitution was carried out by the octamer transfer method, as previously described [61] (see detailed protocol in the Supplementary Information). Oligonucleosomes used as histone donors for reconstitution were obtained from HeLa cells as described elsewhere [62]. All the reconstitution reactions used for this study were carried out using 0.5 pmol of probe and 1.5 µg of oligonucleosomes. Once reconstituted, the nucleosome probes (and mock-reconstituted probes) contain 4 fmol/μL of probe and 12 ng/μL of non-labeled oligonucleosomes (the latter value in terms of DNA content).

4.3. Competitive Nucleosome Reconstitution

Nucleosome reconstitution for these assays was performed as stated above, with the following changes: donor oligonucleosomes were obtained from chicken erythrocytes as previously described [63]. The number of oligonucleosomes for the reconstitution reaction was set as the amount required to obtain 30% reconstitution of the 147 bp 5 s probe. This probe corresponds to the dominant positioning region of the Lytechinus variegatus 5 s rDNA when reconstituted into mononucleosomes [64,65]. All the different probes were reconstituted simultaneously in each independent assay. Three independent assays were performed (two in triplicate, one in duplicate). In each assay, the components used for reconstitution were taken from the same master mix. Relative equilibrium constant values (relative Keq) and relative free energies (ΔΔG°) were calculated as described by Thastrom and co-workers [44].

4.4. Binding Assays

In each binding reaction, a mix containing 6.9 μL of Remodeling buffer, 0.6 μL of ATP (Roche, 11140965001, Roche, Basel, Switzerland) or Adenosine-5′-O-(3-thiotriphosphate) (Roche, 11162306001), 1.5 μL of TF buffer, 0.5 μL of CADS buffer, 3 μL of CRC or CRC buffer and 2.5 μL of probe was incubated for 45 min at 30 °C (see the composition of each solution used in the mixes in the Supplementary Information). Final concentration of the different components of this mixture was as follows: 10.4 mM HEPES-KOH (pH 7.4–7.9), 3.9 mM Tris-Cl (pH 7.4–8.0), 83.6 mM NaCl, 36.4 mM KCl, 5.1 mM MgCl₂, 2 mM ATP (or ATP-γ-S), 0.2 mM Mg(CH₃COO)₂, 0.05% NP-40, 10% glycerol, 1 μM ZnCl₂, 0.2 mM EDTA (pH 8.0), 0.4 mM EGTA, 73 μg/mL BSA, 1.2 mM imidazole, 2.1 mM DTT, and 0.45 mM PMSF. The samples were then subjected to electrophoresis in non-denaturing polyacrylamide gel (200 V, 0.3× TBE, 3.5% acrylamide, 60:1 AA:Bis proportion) in a cold room. Afterwards, the gel was dried and autoradiographed on film or scanned using a phosphor screen and Molecular Imager FX (BioRad, Hercules, CA, USA). Densitometric analyses were performed using Quantity One software, v4.1.1 (for phosphor imager files) or UN-SCAN-IT software, v6.1 (films). The extent of binding was calculated as the ratio of the signal of bound nucleosome band over the combined signal of bound and unbound nucleosomes in the lane. The resulting values were made relative to the average of all binding reactions present within an individual assay before combining data with the other independent assays.

4.5. Nucleosome Remodeling Assays

For remodeling reactions, a mix containing the same composition described for binding assays was used. Each reaction was incubated for 45 min at 30 °C, except for assays where the time is specifically indicated in a figure. After this incubation, a mix (0.7 μL) containing 750 ng salmon sperm DNA and 500 ng long oligonucleosomes was added and incubated for 20 additional minutes at 30 °C. The samples were then subjected to electrophoresis in non-denaturing polyacrylamide gel (200 V, 0.3× TBE, 5% acrylamide, 40:1 AA:Bis proportion) in a cold room. Afterwards, the gel was dried and then scanned using a phosphor screen and Molecular Imager FX (BioRad). Densitometric analyses were performed using Quantity One software, v4.1.1 (BioRad). Film autoradiography was also performed. The extent of nucleosome sliding was calculated as the ratio of the signal of the slid mononucleosome band over the signal of all mononucleosome bands in the lane. The extent of nucleosome eviction was calculated by subtracting the free DNA signal of the probe present in the absence of RSC from the free DNA signal obtained in the presence of this complex. The resulting value was divided by the nucleosomal DNA signal present in the absence of a complex, obtaining the fraction of nucleosomal DNA that is converted to free DNA. Values of sliding and eviction extents were made relative to the average of all reactions present within an individual assay before combining data with the other independent assays. In time-course assays, after the remodeling and removing steps, the mixes were placed on ice until the end of all reactions. Afterward, the samples were subjected to gel electrophoresis and subsequent analysis as described above.

4.6. Determination of Nucleosome Sliding Directionality

Nucleosome remodeling reactions were performed as described above. Once removing reactions were carried out, 10 U of a specific restriction enzyme were added and an additional incubation was performed for 30 min at 30 °C. Afterward, the samples were subjected to gel electrophoresis and subsequent analysis as described for remodeling assays.

4.7. Bioinformatics Analyses

For determination of nucleosome positioning and its change in the presence of ISW1a, datasets obtained from a study of Krietenstein and co-workers (GEO dataset GSE72106) were used [15]. The datasets used were MNase-ChIP-H3-seq on S. cerevisiae genome reconstituted into nucleosomes (SGD), replicate 1, replicate 2, and replicate 3 (Sequence Read Archive accession number SRR2164369, SRR2164396, and SRR2164397, respectively) and MNase-seq on reconstituted genome in the presence of purified ISW1a, replicate 1 and replicate 2 (Sequence Read Archive accession number SRR2164386 and SRR2164392, respectively). These datasets were processed as detailed in the Supplementary Information. Mitochondrial genome sequences and ribosomal genes were excluded from the analysis. To obtain nucleosome positions, the nf-core/mnaseseq pipeline from the NextFlow hub was executed through Docker. The +1 nucleosome was defined as the nucleosome dyad peak that was closest to a TSS in a window of −80 bp to +140 bp. The −1 nucleosome was defined as the upstream nucleosome dyad peak that was closest to the +1 nucleosome. All conditions were processed using these criteria. Changes in the translational location of nucleosomes were defined as the difference in nucleosome position between the absence (SGD) and the presence of purified ISW1a for +1 and −1 nucleosomes. The obtained changes in nucleosome positioning were classified into two groups: more than 20 bp upstream and more than 20 bp downstream of the original position (SGD), for both +1 and −1 nucleosomes. For the analysis of poly(dA:dT) tracts frequency, windows of 250 bp around nucleosome dyad peaks on SGD were defined and the corresponding sequences were obtained in FASTA format (500 bp window). PolyA and polyT tracts were defined to be at least 6 nucleotides (5′-AAAAAA-3′ and 5′-TTTTTT-3′). Sequences around nucleosome dyad were binned in 25 bp intervals and polyA or polyT tract frequency was plotted for each bin. Additional details regarding each step and tools [66,67,68,69,70,71,72,73] used can be found in the Supplementary Information.