At the outset of this study we hypothesized that the recruitment of hSWI/SNF to GR binding sites and GR-regulated promoters might result in discrete changes in nucleosome positions that would activate or repress transcription by allowing or blocking the binding of transcription factors to DNA (as suggested by our recent biochemical studies 
, and for review see 
). By contrast, in UL3 cells we found almost no clear cases of nucleosomes shifting from one position to another after addition of Dex and/or after knock down of the hSWI/SNF ATPase, BRG1. Instead, the most frequently observed effect, after 1 hr of Dex treatment, was an increase in the apparent occupancy at already existing nucleosome peaks within ~2 kb of transcription start sites. This effect could be seen for specific nucleosomes on individual promoters, as well as on average across all Dex repressed or Dex activated promoters. Surprisingly, a strong increase in average nucleosome occupancy was also seen over the promoters of genes that were not regulated by Dex, an effect that was reproducible in independent samples and could not be explained by differences in MNase digestion or any systematic bias in the microarray analysis. Accordingly, these results suggest that the most prominent effect of GR and Dex on chromatin is to rapidly increase measured nucleosome occupancy on a large fraction of Pol II promoters, apparently genomewide.
Given that this effect is seen after only 1 hour of dex treatment, it seems unlikely that it would be due to GR-directed transcriptional activation or repression of a second wave of transcription factors. This is also consistent with studies showing that inhibition of translation via cyclohexamide does not alter the distribution of genes that are upregulated, unregulated and downregulated by a two or four hour incubation with Dex 
. For Dex-unregulated genes, while there are no known or expected GR binding sites near their promoters, the increased nucleosome occupancy that is observed could potentially be mediated by very long range influences, in cis
or even in trans
, of GR bound to chromatin. Indeed, recent studies have indicated that more than half of all functional GRBSes are located over 10 kb away from the start site of genes they regulate 
, and GR and Dex can mediate dramatic unfolding of large chromatin domains in fluorescence microscopy studies 
. Furthermore, all but one of the genes on our array had a start site within 100 kb of one or more of the over 15 thousand GR binding sites recently identified in A549 cells (an average of one per 180 kb 
). On the other hand, there is growing evidence that hormone bound GR can regulate a variety of cellular kinases and other signaling molecules, independent of its direct transcriptional regulatory functions (e.g. 
). These altered signal transduction cascades might then regulate common transcription factors, co-activators, basal factors or chromatin modifying complexes that might have a broad, GRBS-site-independent effect on promoter chromatin structure in response to Dex. Interestingly, global as well as gene-specific dephosphorylation of the linker histone H1 has been observed after prolonged Dex exposure 
. While the slower timing of this effect appears to rule out its direct involvement in the chromatin changes we observe, it does establish a precedent for Dex and GR-dependent changes in fundamental chromatin structure cell-wide.
Interestingly, the increase in apparent promoter nucleosome occupancy that we see at 1 hr of Dex treatment is largely lost after 4 hrs of Dex treatment. This may be most consistent with the possible mechanism described above, in which increased promoter nucleosome occupancy results from activation of cellular kinases by Dex-bound GR, an effect that could potentially be transient, and attenuated by long-duration Dex exposures. This observation is also broadly consistent with other studies indicating that chromatin remodeling effects associated with the activation of specific genes by GR and other nuclear hormone receptors can change over time. For instance, a greater than 24 hr exposure to Dex has been shown to silence MMTV
transcription and eliminate restriction enzyme accessibility at Nuc B 
. In addition, studies examining the effect of estradiol-bound estrogen receptor (ER) on the human PS2
gene showed that, under some circumstances, ER activation could lead to ~2 hr long periodic cycles of transcription factor binding and release together with promoter nucleosome alterations at the promoter 
. The possibility of this type of cycling effect was also suggested for GR by a set of in vitro
transcription studies on chromatin 
Using RNAi, we showed that BRG1-containing hSWI/SNF was important for the high nucleosome occupancy after 1 hr dex treatment at GR-repressed and GR-activated promoters, suggesting that part of the increase in nucleosome occupancy at these promoters may be due to GR-dependent recruitment of hSWI/SNF. We also found that hSWI/SNF was essential for the low nucleosome occupancy in the absence of Dex at both GR-regulated and GR-independent genes. This surprising effect might potentially mean that hSWI/SNF is continuously present, and remodeling chromatin, near the TSSes of most genes on the array. Basal levels of hSWI/SNF recruitment might be possible through the dozens of activators, repressors and basal factors that it has been shown to bind to (e.g. 
). Alternatively, early biochemical characterization of hSWI/SNF indicated that the complex is present at ~10,000 copies per cell 
, raising the possibility that it might be sufficiently abundant to have significant non-targeted effects on genomic chromatin. The seemingly more likely possibility, however, is that the BRG1 hSWI/SNF complex may be essential for the transcriptional activation or repression of some other factor (be it a histone modifying enzyme, histone chaperone or other remodeling complex) which is required to promote low occupancy over TSSes in cycling cells.
In addition to the effects of Dex in U2OS cells, we also found striking increases in measured nucleosome occupancy near TSSes of both regulated and non-regulated genes in human HL60 cells induced to differentiate to granulocytes (as assayed by tiling microarray), and CD4+ T-cells activated by addition of anti-CD3 and anti-CD28 antibodies (as assayed by Solexa/Illumina multiparallel sequencing, 
). These results indicate that genomewide alterations in promoter nucleosome occupancy may be a common cellular response to a variety of stimuli.
The simplest interpretation of this effect is that that the fractional occupancy of promoter DNA by histone octamers (e.g. the fraction of gene copies with a nucleosome covering each position on DNA) increases in response to these stimuli, perhaps as the result of new deposition of nucleosomes using S-phase independent chaperones. Intriguingly, one recent study revealed that, unlike the case for yeast promoters, human Pol II promoters have sequence characteristics which are expected to promote higher than average nucleosome occupancy 
. This suggests the interesting possibility that high promoter nucleosome occupancy is the default state, and that low occupancy must be actively maintained. If so, the stimuli we have examined here might inhibit these active processes, causing promoters to revert to an intrinsic sequence-encoded high occupancy state.
It must be emphasized, however, that apparently increased occupancy could also be caused by other effects that might alter the abundance of nucleosomal MNase fragments from promoter regions in our samples. For instance, apparently low occupancy could result if association with nuclear matrix proteins prevented the release of mononucleosomes after MNase digestion, or if association with heterochromatin proteins, HMGs or variant linker histones blocked digestion between adjacent promoter nucleosomes (resulting in dinucleosome-sized fragments that would be lost when ~146 bp products mononucleosomal MNase products were isolated (Fig. S1
). It is also possible that differences in histone tail modifications, linker histones or core histone variants might change the MNase sensitivity of promoter mononucleosomes. In the most extreme case this might result in the complete hydrolysis of octamer-covered DNA by MNase (reducing signal in both assays). Alternatively, if these effects changed the MNase protected footprint size to more than ~155 bp or less than ~135 bp, the range we isolated by PAGE (or to a size greatly different from the ~150 bp band isolated in 
), these larger or smaller fragments would not be detected. Nevertheless, whether the observed occupancy increase is due to increased histone octamer abundance or to one of these other effects, the observations described here provide evidence for a striking, unanticipated change in chromatin structure associated with Pol II promoter DNA, apparently genomewide, which can be caused by at least three different inducing conditions.
What possible function might be ascribed to genomewide increases in promoter nucleosome occupancy in response to environmental stimuli? HL60 cell differentiation and T-cell activation are long-term processes that result in dramatic changes in cellular functioning. In cases like these, the system-wide reduction in promoter accessibility by increased promoter nucleosome occupancy might function to globally slow the production of proteins that promote and regulate default cellular processes involved in undifferentiated growth. At the same time, condition-specific transcription factors and signaling cascades would be expected to be able to contend with this general effect at promoters, allowing the production of new proteins specific for the differentiated or activated cell's functions.
For GR-regulated genes, a global increase in promoter nucleosome density could potentially act to suppress plieotropic responses that might result from GR and Dex activation of non-genomic signal transduction cascades (such as phosphorylation of transcription factors by src kinase, as reviewed in 
). Even though this effect is transient, and lost after 4 hours of dex treatment, it might be sufficient to limit transcriptional responses to glucocorticoid hormone more specifically to genes containing GRBSes or containing DNA binding sites for second-wave transcription factors regulated by GR. In this regard, it was quite interesting to note that the rise in nucleosome occupancy was of lower intensity and lesser range around TSSes within 500 bp of a GRBS. This suggests that nearby GR binding may be able to suppress the increase in promoter nucleosome occupancy that is otherwise stimulated, genomewide, by Dex. If so, the maintenance of these promoters in a state of low nucleosome occupancy during this initial period after Dex addition may be an essential aspect of their regulation by Dex. It is unclear, as yet, whether the transient increase in promoter chromatin density after Dex addition quantitatively alters transcription rate, and it will be interesting, in future studies, to see whether this is the case, using techniques capable of directly measuring transcription rate such as nuclear run on, GRO-seq or ChIP microarray analysis of Pol II occupancy