2. RESULTS 24
2.1. Refining the in vitro reconstitution system 24
Central to this study was the in vitro
reconstitution assay that was previously shown
to establish in vivo‐like nucleosome positioning
at the PHO5 and PHO8 loci [175, 176] (Fig. 7).
First, supercoiled plasmid DNA is assembled into
chromatin by salt gradient dialysis using purified Drosophila embryo histone octamers. This does
not generate in vivo‐like positioning at the PHO5
and PHO8 loci [175, 176] and, on linear DNA
fragments, also not genome‐wide [119, 120].
Second, in the actual positioning assay, the
improperly positioned nucleosomes are moved
to their in vivo positions by incubation with an S.
cerevisiae WCE. In the absence of ATP,
nucleosome positions remain unchanged but in
the presence of ATP some unknown activity
within the extract alters the positioning at the PHO5 and PHO8 promoters such that it closely
resembles the in vivo positioning [175, 176].
Furthermore, any components alone or in
addition to the S. cerevisiae WCE can be tested
with this system for their influence on
positioning.
I refined and extended the original protocol
[175, 176] of this in vitro reconstitution system.
To counteract chromatin aggregation, the
concentration of histones and DNA during salt
assembly was lowered and the total reaction
volume increased to reduce the effect of the
small volume changes that occurred during
dialysis. Originally, the assay buffer contained
1.5 mM MgCl2 on top of 3 mM ATP/MgCl2 [175], which promoted chromatin aggregation in many
instances (data not shown). This extra 1.5 mM
MgCl2 was omitted from the buffer, which greatly reduced aggregation without impact on
the nucleosome positioning activity (data not
shown). Finally, as we wished to include a
number of new loci, we constructed a series of
new plasmid templates for the in vitro system.
Originally, the shuttle vectors C‐leu (PHO5) and Figure 7. The in vitro reconstitution system
Schematic overview. Individual plasmids or a plasmid based
genomic library are assembled into chromatin with purified Drosophila embryo histone octamers by salt gradient
dialysis. This yields positions that are specified by the
intrinsic DNA preferences under these conditions. Next, the
preassembled plasmids are incubated with WCE, WCE sub‐
fractions and/or purified components in the presence or
absence of ATP for 2 hours at 30°C. In the presence of ATP,
components may alter nucleosome positions according to
their own preferences. Nucleosome positions can be
analyzed by DNaseI indirect endlabelling, by restriction
enzyme accessibilities or by MNase digestion followed by
pP8apin (PHO8) were used as templates to allow in vitro versus in vivo comparisons on the very same
template. Since in vivo positioning looked virtually indistinguishable on these plasmids and on the
chromosomal locus we chose to make further in vivo comparisons directly with the corresponding
chromosomal loci. So we did not need the rather larger shuttle vectors (~ 10kb) but could use a
smaller backbone, and make more efficient use of material in assembly reactions with template
mixtures as less DNA was required to include the same number of templates. All new vectors
contained an about 3.5 kb yeast PCR fragment cloned into the ~2.6 kb pUC19 vector. For the new PHO8 locus template, we initially sub‐cloned a pP8apin fragment into pUC19 resulting in plasmid
pUC19‐PHO8‐short (Fig. 8A). Sequencing of this insert revealed a number of sequence differences
relative to the SGD data base that were already present in the original pP8apain plasmid (data not
shown). Apparently, the PHO8 sequence in pP8apain was based on a different allele. In order to use
yeast DNA templates that were all based on the reference genome sequence, we used genomic DNA
of strain BY4741 to also create plasmid pUC19‐PHO8‐long (Fig. 8A), as well as to create all other new
plasmids (chapter 4.4.1.). Reconstitution of in vivo‐like positioning worked equally well with these
three PHO8 templates (Fig. 8B).
Figure 8. Nucleosome positioning at the PHO8 promoter could be reconstituted independent of the
vector backbone, insert size or the PHO8 allele.
(A) Schematic overview of different inserts in the pUC19‐PHO8‐short, pUC19‐PHO8‐long and pP8apin plasmids. Numbers
indicate positions relative to the PHO8 ATG (in bp). (B) In vitro reconstitution of nucleosome positioning at the PHO8
promoter on plasmids described in (A). Pre‐assembled plasmids pUC19‐PHO8‐short, pUC19‐PHO8‐long and pP8apin were
incubated with ("+WCE +ATP") or without ("‐WCE ‐ATP") WCE and ATP, and analyzed by DNaseI indirect endlabelling.
Asterisks mark the position of a hypersensitive site that is generated at a site within the lacZ ORF of the bacterial vector
backbone. The horizontal line indicates the approximate location of the insert‐backbone border. Schematics of the PHO8
promoter are shown on the left. Ramps indicate increasing DNaseI concentrations. The numbers above the marker bands
refer to the position (in base pairs) relative to the PHO8 ATG.
Successful reconstitution of nucleosome positioning at the PHO5 promoter critically depends on near
maximal assembly degrees (histone:DNA mass ratio of ~1)[175]. We controlled for high assembly
purified from E. coli are heavily negatively supercoiled. Nucleosome formation itself introduces about
one negative supercoil [191], or, in the case of in vitro assembly of supercoiled templates, stabilizes
one pre‐existing negative supercoil against relaxation by topoisomerase I. As pre‐existing supercoils
might help nucleosome formation, salt assembly works better with negatively supercoiled templates
[192]. The number of negative supercoils in E. coli‐purified plasmids roughly matches the number
seen in fully assembled plasmids [193]. To identify the number of nucleosome‐constrained
supercoils, the assembled plasmids were treated with topoisomerase I (Topo I).
Figure 9. Salt gradient dialysis yielded a homogenous chromatin populations even for
underassembled plasmids.
(A) Supercoil analysis in pUC19‐PHO8‐long plasmids assembled at various degrees and treated with topoisomerase I .
pUC19‐PHO8‐long plasmids were assembled at the indicated histone:DNA mass ratios and incubated with (+) or without (‐)
topoisomerase I. pUC19‐PHO8‐long was linearized with PstI. Number of topoisomerase I resistant supercoils were resolved
by gel electrophoresis in the presence or absence of 3.3 µM chloroquine and detected by ethidium bromide. (B) Equimolar
mixtures of plasmids pUC19‐PHO8‐long, pUC19‐PHO5, pUC19‐RNR3 and pUC19‐ADH2 were assembled by salt gradient
dialysis at different histone:DNA mass ratios as indicated, separated on a native 0.35X TBE 0.9% agarose gel and analyzed
by Southern blotting and hybridization with probes specific for the respective locus.
Assembly with a histone:DNA ratio of about 1:1 indeed fully protected the same number of
supercoils from removal by Topo I as the number generated in E. coli, i.e. the plasmids migrated at a
similar position as the untreated plasmids (Fig. 9A "no chloroquine" gel ‐ Note that plasmids with
more supercoils migrate faster). Moreover, in samples with a histone:DNA ratio of 0.3, plasmids
contained much fewer supercoils than the samples with a 1:1 ratio. To further resolve the
topoisomers of the highly assembled plasmids we electrophoresed the same samples in the presence
of chloroquine, which intercalates into the DNA, introduces positive supercoils, and thereby reduces
the number negative supercoils towards the resolution range of the gel. Plasmids assembled with a
1:1 ratio had a very similar supercoil distribution as the untreated plasmids, confirming the high
assembly degree (Fig. 9A ‐ panel on the right). Noteworthy, most of the plasmid population was
nicked due to the sheering forces (pipetting etc.) during handling of the large (~6.1 kb) assembled
templates. Such nicking was previously reported for even smaller plasmids [194], is promoted by
nucleosome assembly (Fig. 9A ‐ compare unassembled with assembled plasmids), and precludes
direct analysis of the majority of templates by this technique. However, the un‐nicked subpopulation
As described above, we cloned ~3.5 kb fragments each of another ten loci of interest (ADH2, CHA1, GCY1, HIS3, HO, PHO84, POT1, RNR3, SNT1, SUC2) into pUC19 vectors. Positioned nucleosomes at
these loci had been described and mapped, and for several loci positioning factors were previously
identified or implicated (e.g. RSC for CHA1 [149] or Reb1/Abf1 for SNT1 [42]). For direct comparisons,
we assembled several plasmids together in one tube as previously done for the PHO5/PHO8 loci
[175]. In case of limiting histones per total DNA, different histone affinities of the plasmid inserts
could lead to uneven assembly degrees within such plasmid mixtures. Even with only one plasmid
type it was unclear to what extent assembly with limiting amounts of histones could give rise to sub‐
populations of different assembly degrees. Such uneven assembly could lead to biased results when
comparing the reconstitution at different loci. To check for such uneven assemblies, we assembled
an equimolar mixture of pUC19‐PHO8‐long/pUC19‐PHO5/pUC19‐ADH2/pUC19‐RNR3 with varying
amounts of histones (approximate histone:DNA ratio from 0.3 to 1.15) followed by native agarose
gel, Southern blot and differential hybridization (Fig. 9B). For each of the six histone:DNA mass ratios
there was only one major band, but its position varied between assembly degrees, even between
rather similar ratios (Fig. 9B, compare 0.3 with 0.35 or 0.6 with 0.7). Thus, all of each of the different
plasmids were assembled into rather homogeneous chromatin populations, i.e. acquired a similar
number of nucleosomes. Furthermore, the highly similar migration behaviour of the PHO8, PHO5 and RNR3 containing plasmids for each assembly degree also suggested that even different plasmids
were assembled to the same degree. The uniform assembly degree across different plasmids was
later also confirmed by an MNase protection assay (data not shown). It was later realised that
plasmid pUC19‐ADH2 was actually a dimer of ~12 kb (data not shown), which likely explains its
different migration behaviour.