Molecular Cell 24, 469–479, November 3, 2006 ª2006 Elsevier Inc. DOI 10.1016/j.molcel.2006.09.009

Nucleosomes Can Form a Polar Barrier

to Transcript Elongation by RNA Polymerase II

Vladimir A. Bondarenko,

1

Louise M. Steele,

2

Andrea U

´

jva

´

ri,

2

Daria A. Gaykalova,

Yury S. Polikanov,

Department of Pharmacology

University of Medicine and Dentistry of New Jersey

Robert Wood Johnson Medical School

675 Hoes Lane, Room 405

Piscataway, New Jersey 08854

2

Department of Molecular Genetics

Lerner Research Institute

Cleveland Clinic

Cleveland, Ohio 44195

entry into the H3/H4 tetramer that can be recapitulated

without H2A/H2B dimers. The same nucleosomes tran-

scribed in the opposite orientation form a weaker, more

diffuse barrier that is largely relieved by higher salt,

TFIIS, or FACT. Barrier properties are therefore dic-

tated by both the local nucleosome structure (influ-

enced by the strength of the histone-DNA interactions)

and the location of the high-affinity DNA region within

the nucleosome. Pol II transcribes DNA sequences

at the entry into the tetramer much less efficiently

somes. Within each nucleosome core, a central H3/H4

tetramer is flanked on both sides by H2A/H2B dimers.

Nucleosomes can be transiently displaced from eukary-

otic genes when transcription levels are very high (Krist-

juhan and Svejstrup, 2004; Schwabish and Struhl, 2004),

but the majority of transcribed genes retain nucleosomal

organization and thus each Pol II complex must encoun-

ter nucleosomes during elongation. Although Pol II can

clearly traverse nucleosomes efficiently in vivo, nucleo-

somes form a high barrier to transcribing Pol II in vitro

II passage in vivo (Kimura and Cook, 2001; Thiriet and

Hayes, 2005). The elongation factor FACT can act as

an H2A/H2B chaperone during nucleosome assembly,

and this activity most likely mediates its stimulating

effect on transcription of nucleosomal templates (Belot-

serkovskaya et al., 2003). TFIIS, which facilitates tran-

script cleavage and restoration of catalytic activity of

Pol II in arrested transcription complexes, can also facil-

itate transcript elongation on nucleosomal templates

(Kireeva et al., 2005).

tamer. These DNAs direct assembly of nucleosomes to

single, well-defined locations. Our analysis indicates

that there is considerable sequence and orientation-

dependent variation in both overall height and location of

the nucleosomal transcription barrier. Nucleosomes as-

sembled on strong nucleosome-positioning sequences

(NPS) can present very strong salt- and factor-insensi-

tive blockades to Pol II, but these exceptional barriers

occur in only one transcriptional orientation. A substan-

tial part of these very strong nucleosomal barriers can

Results

A Nucleosome Formed on a Strong

Nucleosome-Positioning Sequence Can

Present an Insurmountable Barrier to Pol II

In our previous studies, transcription complexes were

directly assembled with yeast Pol II, template oligonu-

cleotides, and short initial RNAs. DNA fragments bear-

ing single nucleosomes assembled in multiple positions

were then ligated downstream of the yeast transcription

complexes (Kireeva et al., 2002). A major goal of the

mononucleosomes on the templates were verified by

analysis in native gels (Figure S1 in the Supplemental

Data available with this article online), restriction diges-

tion analysis (Figure S2), and DNase I and hydroxyl rad-

ical footprinting (data not shown). Yeast Pol II elongation

complexes containing 9 nt RNA (EC9) were ligated to

these uniquely positioned nucleosomes and advanced

to produce 45 nt EC45 complexes as described earlier

(Kireeva et al., 2002; Figure 1A). We also wished to com-

pare the elongation competence of the assembled yeast

Thus, for the human polymerase, templates containing

both a promoter and an NPS were employed. After

nucleosomes were assembled on these longer DNA

fragments, human Pol II preinitiation complexes were

assembled with purified RNA polymerase and purified

or recombinant general transcription factors (Pal et al.,

2005; Figure 1B). Human Pol II complexes were initially

advanced with a subset of the NTPs to produce stalled

EC21 elongation complexes (Figure 1B). The yeast

EC45 and human EC21 complexes were extended in

were washed, and transcription was resumed by addition of all unlabeled NTPs.

(B) Experimental system for analysis of transcription through a nucleosome by human Pol II. DNA fragments preassembled into nucleosomes

were attached to beads and then used for assembly of Pol II preinitiation complexes, which were advanced to EC21 and then chased with

four NTPs. All templates for yeast and human Pol II were constructed in a similar way; in particular, strong nucleosome-positioning sequences

(NPS) were always located on the downstream DNA end and overall length of template DNA was constant in most experiments.

(C) Nucleosome cores assembled on the 601 DNA sequence form a strong barrier to yeast Pol II. Histone-free 601 DNA or nucleosomes were

transcribed in the presence of the indicated concentrations of NTPs and KCl, in the presence or absence of FACT. Arrows indicate positions

of the labeled transcripts in the gel. Position of the nucleosome on the template is indicated by the oval (the nucleosomal dyad is shown as a black

the tetramer. If this were true and all DNA-histone inter-

actions are equally strong along the tetramer-bound

DNA region, one would expect that (1) the 601 sequence

would provide an equally strong barrier regardless of its

orientation on the template and (2) nucleosomes assem-

bled on other DNA sequences with comparable affinities

for histones would provide comparably strong barriers

to transcript elongation in analogous locations.

template with the reverse orientation of the 601 position-

trom et al., 1999). Yeast and human Pol II elongation

complexes were assembled on these nucleosomal tem-

plates as described in Figures 1A and 1B. Transcripts

were extended at different concentrations of KCl in the

presence or absence of TFIIS (human Pol II) or FACT.

The results are shown in Figure 3 and Figure S3 and

summarized in Figure 4 and Figure S5.

height and location of the transcriptional barrier formed

on these nucleosomes. Polymerases paused primarily

(B) FACT can relieve the 601R nucleosomal barrier to yeast Pol II. Transcription of the 601R nucleosomes was conducted as described above; the

45-mer was extended in the presence or in the absence of FACT.

(C) The height and location of the nucleosomal barrier to transcription by human Pol II is variable and depends on the underlying DNA sequence

and its orientation. Pol II transcription complexes assembled on nucleosomal templates bearing the 601R, 603, or 603R sequences were

extended as in Figure 1D.

Molecular Cell

472

contrast, the 601R, 603, and 605 barriers were less local-

ized at 40 or 150 mM KCl and significantly relieved by 300

plates (42% on 601 and 30% on 603R) as compared to

traversal levels for the more permissive template group

(67%, 82%, and 57% on 601R, 603, and 605, respec-

tively). On the 601R, 603, and 605 templates, a significant

fraction of yeast and human Pol II complexes transcribed

past the +45 pausing area even at 150 mM KCl. The bar-

rier on the nucleosomal 603 template is lower than the

barrier formed by nucleosomes bound to the 5S DNA

sequence (Kireeva et al., 2002), although 5S DNA has at

least 100-fold lower affinity for the octamer than the

pause locations between roughly +10 to +20 and +35 to

+55 were seen for all nucleosomes, the details of the

pausing patterns were quite distinct for each template

(Figure 4 and Figure S5). For a given salt concentration,

the human enzyme always produced a greater propor-

tion of run-off transcripts in comparison with yeast

Pol II. Comparison of the nucleosomal pause locations

with patterns of transient pausing on the corresponding

histone-free DNA at lower concentrations of NTPs dem-

onstrates that only a subset of the nucleosomal pauses

templates occurred at or beyond the nucleosomal dyad

(601R, 603, and 605 templates; see Figure 4 and Fig-

ure S5), suggesting that nucleosomes remain bound to

the DNA template after Pol II has crossed the dyad.

Figure 4. Summary of the Positions and Intensities of Nucleosome-Specific Pauses Formed during Transcription of 601 and 603 Templates by

Yeast or Human Pol II

The extent of pausing at different sites was quantified. The location of each pause within the given nucleosome is shown on the x axis (RO, run

off). Narrow-width bars show pausing at a single location, whereas wide bars indicate pausing in closely spaced groups. Bar height shows the

amount of pausing at the indicated location, with the amount of run off at 1 M KCl (yeast) or 1% sarkosyl (human) set to 100%. Dark bars show

results at 40 mM KCl, gray bars show results at 150 mM KCl, and white bars show results at 300 mM KCl.

to those seen with 300 mM KCl. With human Pol II, FACT

essentially eliminated pausing in the promoter-proximal

segment of the 601R nucleosome, whereas pausing in

the analogous region of the 601 nucleosome was only

slightly reduced by FACT (Figure S3). TFIIS stimulated

elongation by human Pol II on the 601R, 603, and 605

nucleosomal templates (Figure 3C and Figure S4), in-

creasing run-off levels in 40 mM salt to those seen with

150 mM salt in the absence of TFIIS. As we observed

with the 601 template, run-off transcription on 601R

of TFIIS. Thus, in agreement with earlier results (Kireeva

et al., 2005) with yeast Pol II, stimulation of human Pol II

by TFIIS on nucleosomal templates was easily evident

at low-salt concentrations, but at physiological ionic

strength, TFIIS increased run-off transcription beyond

the 50% level on only one of our nucleosomal templates.

Nucleosome Organization Does Not Necessarily

Determine the Location of the Barrier

to Transcription by Pol II

The variability of the nucleosome-specific pausing pat-

NTP concentrations resulted in transient, apparently

sequence-specific pauses at positions +4 and +5 and

at several locations near the dyad (Figure 5B). These

pauses occurred within the defined-sequence linkers

used to assemble the otherwise random-sequence

DNA templates (Figure 5A).

cleosome at 40 mM KCl occurred at a discrete set of

nucleosome-specific positions near +15 and a broader

range of sites centered within the promoter-proximal

of no more than 2–3 bp. The mean distribution of the

pausing pattern progressively shifted into the nucleo-

some as the salt concentration was increased to 150

and 300 mM KCl. The only discrete pauses seen at

150–300 mM KCl (downstream of the +15 pause) were

very faint bands within the proximal half of the nucleo-

some (marked by dots in Figure 5B), which lacked the

10 bp periodicity observed during transcription through

a nucleosome by yeast Pol III (Studitsky et al., 1997).

Very little run-off RNA was obtained until the salt con-

tetramer and were never detected within the dyad-distal

part of the histone tetramer.

Transcription of the high-strength barrier templates in

the reverse orientation did not reveal high-strength bar-

riers. Traversal of the 601R, 603, and 605 nucleosomes

by Pol II resulted in a diffuse collection of pause sites

that could be largely relieved by higher salt or the elon-

gation factors TFIIS or FACT. The characteristics of the

nucleosomal barrier are therefore dictated both by the

local nucleosome structure (which is likely to be a func-

mer, Pol II facilitates its own progression through the

nucleosome. A likely explanation for this effect would

be a partial unfolding of the histone tetramer from DNA

as a result of the entry of Pol II into the tetramer.

The Nature of the Nucleosomal Barrier

to Transcribing Pol II

None of the nucleosomes in our study could be effi-

ciently traversed by either human or yeast Pol II at (or

below) physiological ionic strength in the absence of

elongation factors. For one orientation of each of the

any condition tested. The location of the most prominent

pause sites on these three templates, w45 bp into the

nucleosome (Figure 7), suggested that the H3/H4 tetra-

mer alone can impose the major transcriptional barrier

on these sequences. This hypothesis was confirmed by

demonstrating that the pausing at +45 directed by a 601

DNA-bound histone tetramer is similar to the pausing

produced at this location by the complete 601 nucleo-

some (Figures 2C and 2D). Consistent with this observa-

tion, the H2A/H2B chaperone FACT was not able to sig-

man Pol II. These results indicate that traffic of the H2A/

H2B dimers does not always control nucleosome tra-

versal by Pol II. The other factor known to facilitate tran-

scription through nucleosomes, TFIIS (Guermah et al.,

2006; Kireeva et al., 2005), could only partially stimulate

readthrough on the nucleosomal 601 and 603R tem-

plates. Thus, nucleosomes can form a barrier to tran-

script elongation by Pol II that cannot be significantly

relieved by 300 mM salt or any known elongation factor.

and elongation factors, exceptional barrier strength

Figure 7. Positions of Nucleosome-Specific Pauses within the Nucleosomal Structure

The structure of the nucleosome core (Luger et al., 1997) is shown on the left and the path of nucleosomal DNA on the right. The backbone of

nucleosomal DNA is shown in white and gray, the H3/H4 tetramer in purple, and the H2A/H2B dimers in green and blue. The regions of histone

interactions with DNA are colored correspondingly (on the right). The locations of ten base intervals on the template DNA strand (numbered from

the upper end of the nucleosome) are indicated along the DNA backbone. The major positions of strong Pol II pausing that occur in the 601, 603R,

and 605R nucleosomes are shown by white rectangles.

Molecular Cell

476

must require high affinity of DNA for the histone core.

high affinity of the 601, 603, and 605 assembly se-

quences for the H3/H4 tetramer does not extend over

the entire central segment of the elements but instead

can be localized to only one side of the H3/H4 tetra-

mer-bound DNA region. This model can explain the

sharp blockade to entry into the tetramer provided

by nucleosomes assembled with the high-affinity se-

quences on the promoter-proximal side of the nucleoso-

mal dyad (i.e., 601, 603R, and 605R). However, if local-

ized high affinity of DNA for the underlying H3/H4

scription through the nucleosome. If polymerase entry

into the tetramer reduces downstream barriers, then

high-affinity elements located downstream of the dyad

would not be expected to provide an exceptionally

strong blockade to Pol II after it passes the +45 region.

on the 601 and 601R nucleosomes (Figure 6) occurred at

the entry into the H3/H4 tetramer, suggesting that this

is a particularly difficult step for any transcriptional

machinery. At this stage of our analysis, we cannot iden-

1990) and up the base preceding a psoralen crosslink

in the template (Shi et al., 1988). Thus, because the lead-

ing edge of Pol II halted at the +45 barrier should be

located w20 nt downstream of +45 (Samkurashvili and

Luse, 1996), the major barrier in the 601, 603R, and

605R nucleosomes could actually be located within the

range from 1 to 20 bp downstream of the +45 region.

difficulties in transcribing the underlying DNA sequence

that are made more severe by the association of the

results obtained with nucleosomes formed on random-

sequence DNA (Figure 5). These data show that nucleo-

some organization does not necessarily determine the

location of the barrier to transcription by Pol II. Thus,

given the tight association of DNA with histones on

the nucleosome surface, sequence-specific DNA paus-

ing is likely to be generally increased by the presence

of nucleosomes. However, within some nucleosomes,

particular free DNA pauses can be very selectively am-

plified. A striking example is the strong +45 pause on

nucleosome or from the tendency of Pol II to pause on

that sequence when it is transcribed as free DNA.

Polymerase Specificity of the Nucleosomal Barrier

to Transcription

A major goal in these studies was the comparison of

nucleosomal transcription by assembled yeast Pol II

elongation complexes and promoter-initiated mamma-

lian Pol II complexes. Although yeast Pol II was uni-

formly less effective in elongating transcripts at a given

salt concentration, the two polymerases showed similar