J. Mol. Biol. (1997) 268, 37±53
Effects of Neutral Salts on RNA Chain Elongation and
Pausing by Escherichia coli RNA Polymerase
Cathleen L. Chan1 and Robert Landick2*
1
Department of Biology and
Biomedical Sciences
Washington University
St. Louis, MO 63130, USA
2
Department of Bacteriology
University of WisconsinMadison, Madison, WI 53705
USA
We examined the effects of neutral salts and the non-ionic solute
2-methyl,-2,4-pentanediol (MPD) on transcript elongation by Escherichia
coli RNA polymerase and on pausing induced by the multipartite his leader pause signal. All solutes tested slowed the overall rate of elongation,
with anions showing the dominant effects in the order:
ÿ
ÿ
ÿ
ÿ
ÿ
HPO24 > OAcÿ > SO24 > ClOÿ
4 > I NO3 > Br Cl MPD
…most inhibitory†
…least inhibitory†
ÿ
Although the protein structure-stabilizing anions HPO2ÿ
4 , OAc , and
2ÿ
SO4 also increased the pause half-life at the his leader pause site, the
remaining solutes accelerated escape from pause site in the order:
ÿ
ÿ
ÿ
ÿ
NOÿ
3 > ClO4 > I > Br > Cl > MPD
…greatest acceleration†
…least acceleration†
Clÿ-induced acceleration of escape from the pause site also occurred on
mutant templates altered for the 30 -proximal region, RNA 30 end, or
downstream DNA. The effect was eliminated, however, by base substitutions that destabilize the pause RNA hairpin or that extend it toward
the 30 end. This ``perfect hairpin'' itself reduced the pause half-life by a
factor of 3. We suggest that the pause RNA hairpin stabilizes a paused
conformation of the transcription complex through an interaction with an
easily disordered region of RNA polymerase. Extending the stem of the
pause hairpin may disrupt the interaction by altering the position of the
hairpin in the transcription complex. Anions may either compete for the
interaction directly or disorder the site of hairpin interaction by chaotropic effects. We suggest that the negative effect of structure-stabilizing
anions like OAcÿ and SO2ÿ
may re¯ect passage of RNA polymerase
4
through signi®cantly different conformations during rapid elongation,
some of which may expose hydrophobic surface.
# 1997 Academic Press Limited
*Corresponding author
Keywords: RNA polymerase; transcription complex; transcriptional
pausing; RNA secondary structure; salt
Introduction
Regulation of RNA chain elongation depends in
part on intrinsic signals that are recognized by RNA
polymerase during transcription. These signals can
Present address: C. L. Chan, Department of
Stomatology, University of California, San Francisco,
CA 94143, USA.
Abbreviations used: ApU, adenylyl (30 ! 50 )uridine;
PCR, polymerase chain reaction(s); NTP, nucleoside
triphosphate; KOAc, potassium acetate; KGlu,
potassium glutamate; MPD, 2-methyl-2,4-pentaediol.
0022±2836/97/160037±17 $25.00/0/mb970934
pause polymerase temporarily, arrest it, or release it
from the RNA and DNA chains (i.e. termination).
Interactions among RNA, DNA and RNA polymerase mediate recognition of these signals, but precisely how the contacts change during signal
recognition (or normal elongation) is poorly understood.
One intrinsic regulatory site that offers a well-de®ned set of interactions to study is the pause site in
the leader region of the his biosynthetic operon.
Pausing in the his leader is mediated by a multipartite signal involving a nascent RNA hairpin (the his
pause hairpin), an 11 nt 30 -proximal region of RNA
# 1997 Academic Press Limited
38
or DNA, bases in the active site, and duplex DNA
downstream from the active site. Pausing halts RNA
polymerase until a ribosome initiates leader peptide
synthesis and dissociates the pause hairpin, thereby
synchronizing transcription and translation in the
his attenuation control region.
To learn more about the protein-nucleic acid interactions that mediate pausing and chain elongation,
we examined the effects of neutral salts on transcription through the his pause signal. Elongating RNA
polymerase is well suited for solute studies because
it is resistant to dissociation by high concentrations
of neutral salts (Arndt & Chamberlin, 1990;
Reynolds et al., 1992). Thus, solute-induced changes
in protein-nucleic acid interactions can be detected
by changes in pausing or chain elongation. Most protein-nucleic acid interactions are salt-sensitive,
although several different effects have been described (Record et al., 1978; Lohman, 1986; Harrison
& Agarwal, 1990). Counterions that interact with the
protein or nucleic acid may compete for key ionic
contacts in the complex. Preferential exclusion or
binding of ions to the protein or nucleic acid may affect protein-nucleic acid interaction entropically
through redistribution of ions and water (Leirmo
et al., 1987). The electrostatic potentials of the ion
spheres surrounding the bound and free macromolecules also may differ (Misra et al., 1994). Finally, at
high concentration, ions can increase or decrease
protein hydration and thus stabilize or destabilize
folded protein structures through effects that often
follow the well-characterized Hofmeister, or lyotropic, series (von Hippel & Schleich, 1969a).
Although all interactions in the transcription complex are possible targets for solute effects, the potential interactions of nascent RNA hairpins that are
components of both hairpin-dependent pause sites
and r-independent terminators are of particular interest. An interaction of nascent hairpins with RNA
polymerase appears to be competed by Clÿ, both at
r-independent
terminators
(Telesnitsky
&
Chamberlin, 1989; Reynolds et al., 1992; Reynolds &
Chamberlin, 1992) and at other sites where complexes that contained a nascent RNA hairpin dissociate rapidly at elevated KCl (Arndt &
Chamberlin, 1990). The his and trp leader paused
transcription complexes are resistant to dissociation
by KCl (unpublished results), but this difference
may re¯ect an intrinsic propensity to dissociation
conferred by the shorter, U-rich 30 -proximal region
present after terminator hairpins (7 to 9 nt for terminators, 10 or 11 nt for the pause; see Chan &
Landick, 1993). In fact, there are good reasons to anticipate similarities in nucleic acid-RNA polymerase
interactions at hairpin-dependent pause sites and rindependent terminators. Both intrinsic signals are
multipartite and can be affected by changes in the
hairpin, 30 -proximal, active site, and downstream
DNA components (Telesnitsky & Chamberlin, 1989;
Lee et al., 1990; Reynolds et al., 1992; Reynolds &
Chamberlin, 1992; Chan & Landick, 1993). Both
events occur via a structural isomerization of the
transcription complex that blocks elongation and
Salt Effects on RNA Chain Elongation and Pausing
whose rate relative to elongation past the site determines the ef®ciency of pausing or termination; the
paused complex, once formed, escapes slowly back
to the elongation pathway, whereas the termination
complex dissociates (Kassavetis & Chamberlin,
1981; Chan & Landick, 1994; Wang et al., 1995;
Landick et al., 1996). Further, when RNA polymerase
is arti®cially halted by NTP deprivation, its apparent
DNA footprint ®rst shrinks preceding arrival at both
types of sites and then expands at the sites (Nudler
et al., 1995; Wang et al., 1995). These changes in DNA
contacts appear to play some role in formation of
paused or termination complexes based on correlation of the altered footprints with changes in the ef®ciency of pausing or termination, but their role is
not understood.
Given these previous ®ndings and the likelihood
that a study of salt effects on pausing and elongation
could yield generalizable results, we undertook the
study of solute effects reported here. However, the
complexity of the transcription reaction, which involves multiple conformational intermediates that
vary signi®cantly at different template positions
(Krummel & Chamberlin, 1992b; Borukhov et al.,
1993; Erie et al., 1993; Chan & Landick, 1994; Nudler
et al., 1994, 1995; Chamberlin, 1995; Wang et al., 1995;
Zaychikov et al., 1995), currently precludes the type
of theoretical analyses that have been possible for
simpler protein-DNA interactions at equilibrium. Instead, we surveyed the effects of different neutral solutes on transcript elongation and pausing to ®nd
the most signi®cant effects and to learn if more detailed study would be informative. We also compared effects of KCl on the wild-type template and
mutant templates that reduce the effects of individual pause signal components. One template
encoded an altered RNA hairpin that itself appeared
to reduce interaction with RNA polymerase. In this
altered structure, which we termed the his perfect
hairpin, the stem of the pause hairpin can extend closer to the transcript 30 end. Our results suggest that
the pause RNA hairpin interacts with an easily disordered charged region of RNA polymerase.
Results
Experimental approach
To test how elevated concentrations of neutral
salts in¯uence RNA chain elongation and pausing,
we needed to examine their effects at both pause and
non-pause positions. Although it would be desirable
to measure effects of neutral salts on both ef®ciency
and half-life of pausing, ef®ciency measurements are
dif®cult to make accurately (Landick et al., 1996),
whereas the pause half-life can be easily obtained
from the concentration of pause RNA sampled over
time, provided RNA polymerase molecules reach
the pause site with reasonable synchrony (Winkler
& Yanofsky, 1981; Landick & Yanofsky, 1984; Chan
& Landick, 1989). To achieve this synchrony and to
circumvent the destabilizing effects of salts on open
complexes (even the relatively non-disruptive salt
Salt Effects on RNA Chain Elongation and Pausing
KCl blocks initiation at >150 mM), we ®rst formed
stable G16 elongation complexes in optimal ionic
conditions by omitting UTP (Figure 1; see Materials
and Methods). We then adjusted the concentrations
of salt and NTPs to make the desired measurements,
so that each measurement began with nearly all of
the elongating RNA polymerase molecules at G16
(Figure 2(a), ®rst lane). To discern effects on chain
elongation and pausing, we measured two different
parameters at increasing concentrations of a variety
of neutral solutes: (1) the time required for RNA
polymerase to elongate between G16 and the his leader pause (a region devoid of strong pause sites);
and (2) the half-life of pausing at the his leader site.
We chose a set of ten solutes to test for effects on
transcript elongation and pausing based on several
criteria. We varied the anion component of the solute
because we were principally interested in identifying anions that might perturb protein± nucleic acid
39
interaction and because some cations (Zn2‡ and
Mg2‡) are known co-factors for transcription. Selection of cations was dictated principally by solubility
of salts (e.g., Na2SO4, but not K2SO4, is relatively soluble in water). We saw only minor differences in
transcription at elevated concentrations of NaCl versus KCl (data not shown), suggesting that presence
of Na‡ versus K‡ was relatively unimportant for the
comparisons we wished to make. To see how anions
with dramatically different properties affected transcription, we chose anions that spanned the range of
Hofmeister effects on protein structure, from those
that stabilize protein structures, like SO2ÿ
4 , to chaotropic anions that denature proteins, like ClOÿ
4.
Finally, to distinguish effects on the stability of protein structures from ionic effects, we tested elevated
concentrations of 2-methyl-2,4-pentanediol (MPD),
which destabilizes proteins by altering hydration,
but is not charged.
Figure 1. DNA templates and transcribed RNA sequence. PCR primers 1 and 2 were used to amplify DNA for in vitro
transcription (see Materials and Methods). To simplify comparison with other studies, nucleotides in the his leader
portion of the transcript are numbered according to their positions in the wild-type his leader RNA, and are 8 and
31 nt greater than the actual sizes of the G16 and A29 T7 A1-his leader fusion RNAs, respectively. The boxed U bases
indicate where polymerase stops (G16, A20 or A29) when transcription is initiated with ApU, GTP, CTP and ATP,
but withholding UTP. Position 17 is G in the A20 and A29 templates. (*) The 30 ends of RNAs whose elongation was
slow enough to yield a discrete intermediate RNA in transcription experiments (Figure 2). (*) The 30 ends of RNAs
whose elongation rate was accelerated by KCl.
40
Salt Effects on RNA Chain Elongation and Pausing
Figure 2. Effects of neutral solutes on transcript elongation on a DNA template carrying the wild-type his leader
pause site. G16 complexes were prepared as described in Materials and Methods and then stored on ice. Solutions
were adjusted to the solute concentrations listed below, equilibrated to 37 C, and then adjusted to 10 mM GTP,
150 mM ATP, CTP and UTP to restart chain elongation. The solutions were adjusted to the following solute concentrations: (a) no addition; (b) 1 M MPD; (c) 1 M KCl; (d) 1 M KBr; (e) 0.5 M KNO3; (f) 0.5 M KI; (g) 0.5 M NaClO4; (h)
0.5 M Na2SO4; (i) 0.5 M KOAc; (j) 0.25 M Na2HPO4. Samples were removed and combined with stop solution at the
times (in seconds) indicated above each lane and electrophoresed through denaturing 10% polyacrylamide gels (see
Materials and Methods). Lanes labeled C contain samples collected after the time-course was completed and the reaction had been incubated for an additional ®ve minutes with all four NTPs adjusted to 300 mM each. P, 94 nt his pause
RNA. RO, 141 nt run-off RNA formed when RNA polymerase reached the end of the template. Panels are arranged
in the approximate order that the different solutes inhibit the rate of transcript elongation. The horizontal line in (d)
is an artifact introduced by the Phosphorimager software. To compare (a) to (j), note that the samples were removed
from the ten reactions at different times.
Transcription at elevated concentrations of
neutral solutes alters the rate of chain
elongation, but has little effect on the pattern
of the major intermediate transcripts
To simplify comparison of different neutral solutes, we assembled a panel of similar time-courses
of transcript elongation between G16 and the end
of the DNA template, using only a few samples
from among the large number of time-points examined in the multiple experiments performed for
each salt (Figure 2). Concentrations of solutes (1 M
or less) that gave signi®cant effects, but that did
not prevent RNA polymerase from reaching the
end of the template were chosen for this comparison (see the legend to Figure 2).
{ To facilitate comparison to other studies, the
nucleotide positions in the RNA transcripts are assigned
using the corresponding positions in the wild-type his
leader transcript (see the legend to Figure 1). Thus, G28
occurs 2 before G39.
None of the solutes we examined dramatically
altered the pattern of RNA intermediates produced
when RNA polymerase transcribed from position
16 to the end of the template (Figure 1), although
the relative persistence of several intermediates did
change (Figure 2). Two major effects were evident:
(1) elevated concentrations of all solutes slowed
the rate of elongation, but some had much greater
effects than others; and (2) some solutes decreased
the apparent half-life of the paused complex,
whereas others increased it. The reduction in chain
elongation rate was evident both in the persistence
of several intermediate RNAs and in the increased
time required for transcription complexes to move
from G16 to the pause (Figure 2). The template
positions where elongation slowed signi®cantly
were determined by comparison to RNA sequencing ladders generated by incorporation of chainterminating 30 -deoxynucleotides (data not shown;
see Materials and Methods). Elongation was most
obviously reduced prior to addition of G28{, G39,
G41, G77, and G100 (see Figure 1; *, positions
41
Salt Effects on RNA Chain Elongation and Pausing
where solutes increased the persistence of discrete
transcripts). We note that similar effects may occur
at positions where elongation is much faster, but
would escape detection on the time-scale of our experiments.
Na2SO4 and KOAc at 1 M and NaHPO4 at 0.5 M
slowed chain elongation to the point that pause
half-lives could not be estimated reliably. KNO3,
NaClO4, and KI at 1 M caused prominent transcription intermediates, including at the his pause
RNA, to resist elongation even when the concentration of each NTP was raised to 300 mM.
Complexes stopped prior to addition of G28 and
G39 were especially prone to dissociation. To determine if these persistent RNAs were released
from the transcription complex, we examined
complexes that were immobilized on streptavidin
beads using a biotinated RNA polymerase (see
Materials and Methods). When the immobilized
G16 transcription complexes were elongated in
0.5 M NaClO4, these refractory transcripts were
found in the supernatant after the beads were pelleted (data not shown). It is likely that persistent
RNA species formed in the other conditions also
re¯ect sites where transcription complexes dissociated.
All salts slowed the rate of transcript
elongation at most sites on the DNA template
Visual inspection of the data in Figure 2
suggested the following order of inhibition of transcript elongation at most positions other than the
his leader pause site:
ÿ
2ÿ
ÿ
HPO2ÿ
4 > OAc > SO4 > ClO4
…most inhibitory†
ÿ
> Iÿ NOÿ
3 > Br Cl MPD
…least inhibitory†
To compare the magnitude of these effects, we
wished to estimate the average time required for
RNA polymerase to reach the pause site under the
different conditions. Since the mechanism of transcript elongation involves multiple intermediates,
which even at a single template position may
sometimes exist in alternative states with different
intrinsic elongation rates (Erie et al., 1993;
Matsuzaki et al., 1994), there is no simple equation
to which data can be ®t to obtain this estimate.
Therefore, we estimated the average time for chain
elongation from position 16 to the pause site by
numerical integration of idealized progress curves
for arrival of polymerase at the pause site
(Figure 3(a)). In other words, we ignored chain extension past the pause site by treating all RNAs
larger than or the same size as pause RNA as a
single species produced as RNA polymerase arrived at the pause site.
To generate idealized progress curves, we
tested several different kinetic mechanisms using
the KINSIM and FITSIM computer programs
(Frieden, 1993). We selected a minimal mechanism that gave reasonable ®t to the experimental
data (see Materials and Methods and Figure 3(a)
and (b)). These measurements revealed modest
decreases (by a factor of 1.4) in the average rate
of nucleotide addition at 1 M KCl, KBr or MPD
(Figure 3(c)). KOAc or Na2SO4 at 0.5 M slowed
the rate of transcript elongation by a factor of
about 4, whereas 0.25 M Na2HPO4 slowed it by a
factor of about 10. C. Altmann and M. J. Chamberlin (personal communication) have observed
similar reductions in the overall rate of transcription using a much different assay on phage T7
DNA. They observed increasing inhibition in the
order Gluÿ > OAcÿ > Clÿ.
Halides, chaotropic anions and MPD
accelerated escape from the pause site
To determine the effects of solutes on pausing at
the his leader pause site, we measured the half-life
of the complexes formed under the different conditions. Concentrations of pause RNAs were
measured (e. g., for low salt and for 1 M KCl
(Figure 4(a) and (b)); see Materials and Methods)
and then used for non-linear regression as a function of time (Figure 4(c)). In the example shown,
1 M KCl decreased the pause half-life from 84
seconds at low salt (Chan & Landick, 1993) to 40
seconds at 1 M KCl.
When we examined the effect of different
anions on pause half-life using this approach,
we found that some anions increased the pause
half-life, whereas others decreased it. Halides,
chaotropic anions and MPD accelerated escape
from the pause site in the following order
(Figure 4(d)):
ÿ
ÿ
ÿ
ÿ
NOÿ
3 > ClO4 > I > Br > Cl > MPD
…greatest acceleration†
…least acceleration†
SO2ÿ
OAcÿ and HPO2ÿ
anions normally
4 ,
4 ,
thought to stabilize protein structure, slowed escape from the pause site in the same order that
they inhibited elongation at most template pos2ÿ
ÿ
itions (HPO2ÿ
4 > OAc > SO4 ).
Halides and chaotropic anions also appeared to
accelerate transcription at a set of sites between
the pause and the end of the template (Figure 1).
These sites correspond to natural pause sites that
we noted previously may be correlated with formation of the C:D hairpin in the nascent his leader RNA (Chan & Landick, 1993). They also
could arise because RNA polymerase is approaching the end of the DNA template. Comparison to 30 -deoxy NTP sequencing ladders
revealed that they re¯ect pausing prior to addition of G129, G133, G137 and G141 (data not
shown, see Figure 1). Interestingly, MPD did not
appear to reduce pausing at the set of pause sites
following the C:D hairpin.
42
Salt Effects on RNA Chain Elongation and Pausing
Effects of Clÿ and OAcÿ on pausing were not
mutually exclusive
Since anions like Clÿ and OAcÿ caused opposite
effects on pause half-life, we wondered whether
one effect or the other might dominate if Clÿ and
OAcÿ were both present at elevated concentration.
To test this, we measured the pause half-life at
20 mM KOAc or 250 mM KOAc in combination
with four different concentrations of KCl between
60 and 500 mM. Interestingly, elevated Clÿ partially reversed the effects of 250 mM OAcÿ on
pause half-life (from 1.6 to 1.8-fold increase at low
Clÿ to 1.2-fold at 500 mM Clÿ; Figure 5) even
though it did not reverse the overall inhibition of
chain elongation by OAcÿ (data not shown).
This result might be explained if Clÿ and OAcÿ
compete for binding to the transcription complex,
but cause opposite effects on pause half-life. However, it seems more likely that they exert effects on
pausing via distinct interactions because this more
simply accounts for their different effects at pause
and non-pause positions (see Discussion).
KCl increased pausing when pause hairpinpolymerase interaction was compromised, but
not when another pause signal component
was altered
Figure 3. Effect of neutral solutes on arrival time. (a)
Relative concentration of RNA species of pause RNA
length or greater. RNA concentrations in experiments
such as shown in Figure 2(a) and (c) are plotted versus
time. (*) No added KCl; (*) 1 M KCl. Continuous
lines represent the best ®t predictions of kinetic models
with seven steps preceding the pause (mechanism C; see
Materials and Methods, and Results) as determined
with the computer programs KINSIM and FITSIM (Barshop et al., 1983; Frieden, 1993). The broken line shows
the best ®t to the 1 M KCl data possible with a kinetic
model with only one step preceding the pause, but
allowing for isomerization into and out of an unreactive
conformation (mechanism B, see Results). (b) Distribution of arrival times predicted by mechanism C for no
KCl and 1 M KCl (a numerically determined ®rst derivative of the continuous lines in the upper portion of the
chart). The average arrival times are calculated from
these distributions by equation (2) given in Materials
and Methods. They are equivalent to the points in the
reaction progress curves in (a) where vertical lines
would produce equal areas to the left below the curve
and to the right above the curve (with 1 as the upper
bound). (c) Average arrival time for transcription complexes at the pause site in the presence of different neutral solutes. The average length of time required for the
transcription complexes to reach the pause site from
G16 (arrival time) was calculated as illustrated in (a)
and (b) and described in Materials and Methods, and
Since all the solutes we tested inhibited the overall rate of chain extension, whereas halides, MPD
and chaotropes accelerated transcription at certain
pause sites, the structure of the paused complexes
must differ in a way that allows these solutes to increase their propensity for nucleotide addition.
Given that the sites where solute-dependent stimulation occurred all could be correlated with a nascent RNA secondary structure, one possibility is
that these solutes disrupt an interaction of pause
RNA structures with RNA polymerase that itself
slows transcription at the pause sites.
To test this idea, we ®rst used a derivative of the
his leader pause site that increased the stem length
of the pause hairpin. The so-called perfect hairpin
corrects mismatches in a bulge that separates the
wild-type pause hairpin from seven potential basepairs that could extend its stem to within 2 nt of
the transcript 30 end (Figure 6(a)), but that normally form only after polymerase has escaped the
pause site (Chan & Landick, 1993). Polymerase
paused on the perfect hairpin template before addition of G103, like wild-type (Figure 6(b)), but escaped three times faster (Figures 6(b) and 7).
Although arrival at the perfect hairpin pause was
Results. The solution composition of the reactions are
standard transcription buffer (see Materials and
Methods) plus the additional neutral solutes indicated
in the list at the right of the graph. Based on cases
where error could be calculated from multiple data sets,
we estimate the arrival time calculations are precise to
10%.
Salt Effects on RNA Chain Elongation and Pausing
43
Figure 4. Effect of neutral solutes on pause half-life. (a)
Pause and run-off RNAs at low salt. The relative concentrations of pause (*) and run-off (*) RNAs as a
function of time were determined from Phosphorimager
scans of gels such as shown in Figure 2(a) and (c). Total
RNA concentration was set equal to 10 nM based on the
amount of template in reactions. (b) Pause and run-off
RNAs at 1 M KCl. As described for (a). (c) Pause halflife determination. Pause half-lives were calculated by
non-linear regression of the pause half-life data as
shown here (Winkler & Yanofsky, 1981; Landick &
Yanofsky, 1984; Chan & Landick, 1989, 1993). To avoid
overestimating the true pause half-life because transcription complexes still arriving at the pause site in¯ate the
measurements of pause RNA concentration, only times
after the arrival progress curves such as shown in
Figure 3(a) plateaued were used in the regression analysis. (d) Effect of various neutral solutes on relative
pause half-life. Half-lives were determined from the
pseudo-®rst order disappearance of pause RNA during
the reactions by non-linear regression analysis (see Materials and Methods and (c)). The solution composition
of the reactions are standard transcription buffer (see
Materials and Methods) plus the additional neutral
solutes indicated in the list at the right of the graph. For
most conditions, the values given are averages determined from multiple experiments. Based on these cases
and many similar measurements we have performed (e.
g. see Figure 7), we estimate the standard deviations of
such results to be at most 20%, and probably 10 to
15%, but lack suf®cient repetitions for rigorous error calculation in all cases.
delayed by KCl or KOAc as for wild type (compare the ®rst two lanes of Figure 6(b), (c) and (d))
both KCl and KOAc slowed escape, in contrast to
their opposing effects for wild-type (Figures 6(c)
and (d), and 7). Further, at 1 M KCl some transcription complexes failed to leave the pause site
even when all four NTPs were present at 300 mM
for ®ve minutes (Figure 6(b), lane C). To test
whether these refractory transcripts resulted from
termination, we assayed their retention on nitrocellulose ®lters. RNAs associated with the transcription complex are retained on nitrocellulose ®lters
and released RNAs ¯ow through (Morgan et al.,
1984). Interestingly, the refractory pause RNA was
found in the ¯ow-through fraction (data not
shown), demonstrating that the perfect hairpin
causes low-ef®ciency (15%) termination in 1 M
KCl.
The results with the perfect hairpin are consistent with the idea that solute-mediated acceleration
of transcription at the his leader pause site occurs
by disrupting a RNA hairpin-polymerase interaction, since changing the structure of the his
pause hairpin in a way that itself accelerated transcription eliminated the special effect of Clÿ at the
pause site. To test this idea further, we examined
the effect of 1 M KCl on the duration of pausing
when substitutions were introduced into each
pause signal component (Figure 8). We selected
mutations in the pause hairpin (C74 ! A,
44
Salt Effects on RNA Chain Elongation and Pausing
was quanti®ed by kinetic approximation (see Materials and Methods).
Taken together, these results argue strongly that
Clÿ competes for a pause hairpin interaction that
slows escape from the pause site, and that the
structure of the pause hairpin is critical for establishing this interaction. When the stem of the hairpin reaches closer to the transcript 30 end, it must
eliminate this interaction and both reduce pausing
and destabilize the transcription complex.
Discussion
Figure 5. Effect on pausing at the his leader pause site
of KCl and KOAc in combination. Transcription reactions were conducted as described in Materials and
Methods and the legend to Figure 2. Relative pause
half-lives (at 10 mM GTP; 1 ˆ 84 s) plotted for several
concentrations of KCl added to reactions containing
either 20 mM KOAc or 250 mM KOAc. Error in these
measurements should be similar to that described in the
legend to Figure 4(d).
C75 ! U, G77 ! A, ``multisubstituted pause hairpin''; Figure 8(a) and (b); Chan & Landick, 1993;
Wang et al., 1995), the 30 -proximal region
(G100 ! A; Figure 8(c) and (d); Chan & Landick,
1993), and the 30 -terminal nucleotide (C102 ! U;
Figure 8(e) and (f); Chan & Landick, 1993) that previously were shown to reduce pause half-life by a
factor of at least 4. To examine the downstream
component of the pause signal, we placed a sequence (non-template strand 50 -GCGAGAGTAGGGAAC; Figure 8(g) and (h)) after the his pause site
that previously was shown to reduce the half-life
of pausing at the trp leader pause by a factor of 3
(Lee et al., 1990). This sequence had approximately
the same effect on the his pause signal (Figure 8(g)),
suggesting that effects of the downstream sequence
at the his and trp pause signals are similar. None of
these substitutions abolished pausing, but the 30
end and downstream mutants did enhance pausing at other sites near the wild-type site (Figure 8(e)
to (h)).
Strikingly, however, only substitutions in the
pause RNA hairpin compromised the Clÿ effect at
the pause site (Figures 8 and 9). This is particularly
evident for the 30 end and downstream mutants,
where nucleotide addition at the sites near the
pause is slowed by Clÿ in parallel with accelerated
transcription at the wild-type pause position (compare positions 99 and 102 (wild-type pause) in
Figure 8(e) and (f) and (g) and (h)). This comparison is particularly important for the 30 -end mutant
since it validates the presence of a Clÿ effect that
Our study produced several conclusions about
the effects of neutral solutes on RNA chain
elongation and pausing. All solutes reduced the
overall rate of chain extension, though to greatly
variable extent. Protein-stabilizing anions like SO2ÿ
4
were most inhibitory, followed by chaotropic
ÿ
anions like ClOÿ
4 , then by inert anions like Cl and
the non-ionic solute MPD. In contrast, halides,
chaotropes, and MPD accelerated elongation at the
pause site. Finally, the ability of Clÿ to reduce
pausing depended on the pause hairpin, but not
the other components of the pause signal. When
the pause hairpin stem was extended closer to the
30 end of the nascent RNA, pausing was weakened,
Clÿ increased rather than reduced pausing, and
low-ef®ciency termination occurred.
Below we consider three ideas to explain these
different effects. First, conformational changes in
RNA polymerase necessary to achieve rapid chain
elongation may expose hydrophobic surfaces and
thus be inhibited by anions that increase protein
surface hydration. Second, the pause hairpin may
slow escape from the pause site by interacting with
an easily disordered surface of RNA polymerase in
a way that can be competed by anions like Clÿ or
disrupted by chaotropic solutes. Third, extension
of the stem of the pause hairpin closer to the transcript 30 end also disrupts this interaction and allows low-ef®ciency termination. To explain these
ideas, we will ®rst describe possible ways that neutral salts might affect transcription and then discuss each point in light of these general
considerations.
Inhibition of transcript elongation by neutral
salts may reflect anion-mediated changes in
protein structure or protein-nucleic
acid interaction
In principle, the effects we observed could re¯ect
anion or cation-mediated changes in the structures
of RNA polymerase, the RNA transcript, or double
or single-stranded DNA in the transcription complex, or they could re¯ect changes in interactions
between these macromolecules. We will consider
each possibility.
45
Salt Effects on RNA Chain Elongation and Pausing
Figure 6. Effect of the his perfect hairpin on pausing at low and high concentrations of KCl and KOAc. (a) Predicted
structure of the his perfect hairpin. Base substitutions relative to the his leader A:B RNA structure (Figure 1) are indicated by bold italic font. The his perfect hairpin has the potential to pair to within 2 nt of the pause RNA 30 end.
However, it is unlikely that all these potential base-pairs form in the paused transcription complex (see the text). (b)
Transcription of the his leader perfect hairpin pause site with no added solute. Only the region of the gels around the
pause RNAs are shown, but the experiments otherwise are comparable to those shown in Figure 2. P, Pause RNA;
RO, run-off RNA. Times when the samples were removed after transcription of the halted G16 complexes was
restarted are indicated at the top of the lanes. C is a sample removed from reactions ®ve minutes after adjusting all
four NTPs to 200 mM each following the last time-point shown. (c) Transcription of the his leader perfect hairpin
pause site at 1 M KCl. (d) Transcription of the his leader perfect hairpin pause site at 0.5 M KOAc. Note that a different time-course is shown from that in (a) and (b).
Effect of salts on nucleic acid structure
Cations interact with the negatively charged
phosphate backbone of nucleic acids and stabilize
double-stranded DNA, RNA:DNA hybrids or
RNA structures (Schildkraut & Lifson, 1965). Cations (principally divalent) also can bind speci®cally
to some RNA structures (e.g. tRNA). However, cation-mediated changes in nucleic acid structure do
not easily explain our results. First, the principal effects were attributable to anions. Second, the divalent cation Mg2‡, which binds tightly to nucleic
acids and was present at 10 mM, should decrease
electrostatic repulsion of the negative phosphate
backbone as effectively as 1 M Na‡ (Record et al.,
1978; Reynolds et al., 1992).
Anions can alter nucleic acid structure (Hamaguchi & Geiduschek, 1962; Robinson & Grant, 1966;
Tunis & Hearst, 1968), for instance by binding to
the bases. However, these effects require concentrations well above the 0.25 M to 1 M range in
which we observed dramatic changes in pausing
and transcript elongation. Thus, effects of solutes
on nucleic acid structure are unlikely to explain
our results.
Effect of salts on protein structure
The extent to which anions interact with proteins
versus water is thought to account for their effects
on protein structure and activity (Jencks, 1969; von
Hippel & Schleich, 1969a,b; Record et al., 1978;
Collins & Washabaugh, 1985; Timasheff, 1993). Stabilizing anions like SO2ÿ
preferentially associate
4
with water, rather than protein, and thus increase
hydration of the protein's surface. This orders the
water molecules surrounding the protein and increases the thermal stability of its folded conformation. As this effect increases, proteins aggregate
because their increased interaction decreases the
solvent-exposed surface area and increases the entropy of the water shell.
Chaotropic anions, such as ClOÿ
4 , displace water
from the protein surface and bind to the amide
backbone. These interactions both neutralize the
unfavorable electrostatic effects of unfolding the
46
Salt Effects on RNA Chain Elongation and Pausing
Figure 7. Effects of KCl and KOAc on pause half-lives
on the wild-type and perfect hairpin templates. Pause
half-lives were determined and plotted as described in
the legend to Figure 4. The solution compositions of the
reactions were standard transcription buffer (see Materials and Methods) plus up to 0.5 M KOAc or up to
1 M KCl, as indicated by the list at the right of the
graph.
protein and allow greater disorder in the surrounding water shell, which increases the solubility of
the protein.
The ability of different anions to affect the
structure and activity of most proteins by interacting with the protein surface and displacing
water is re¯ected in the empirically derived
``Hofmeister series'', which is based on the ability
of salts to increase or decrease protein solubility
in the order:
ÿ
2
ÿ
ÿ
ÿ
ÿ
SOÿ
4 < HPO4 < F ; OAc < Cl < Br
Figure 8. Effects of KCl on pausing on wild-type and
pause mutant templates. Base changes in the mutants
are indicated in bold italic font. For each template tested
the sequence around the pause site is shown above a
panel of gel lanes that show the regions from just below
the pause to the run-off transcript as function of time (in
seconds). All reactions were performed at 20 mM GTP,
150 mM each other NTPs, except (e) to (h) for which the
GTP concentration was 5 mM. The lanes labeled C contain a sample removed ®ve minutes after the concentrations of all four NTPs were raised to 300 mM each. P,
Position of wild-type pause (U102).(a) Multisubstituted
hairpin, low salt. (b) Multisubstituted hairpin, 1 M KCl.
(c) The 30 -proximal region mutant, low salt. (d) The 30 proximal region mutant, 1 M KCl. (e) The 30 -end
mutant, low salt. (f) The 30 -end mutant, 1 M KCl. (g)
Downstream DNA mutant, low salt. (h) Downstream
DNA mutant, 1 M KCl.
ÿ
ÿ
ÿ
ÿ
< NOÿ
3 < I < CCl3 COO < ClO4 < SCN
Anions at the extreme left of the series decrease
protein solubility or ``salt-out'' proteins and
stabilize the folded tertiary structure of proteins,
whereas anions at the right of the series increase
protein solubility or ``salt-in'' proteins, and destabilize the folded tertiary structure of proteins.
Clÿ, in the middle of the series, is relatively inert
and, as such, is often used in buffers for in vitro
studies (Leirmo et al., 1987).
The ability of different anions to increase or decrease the pause half-life at the his leader pause
site correlates with the Hofmeister series
(Figure 4(d)), suggesting that anion effects on pro-
47
Salt Effects on RNA Chain Elongation and Pausing
plex formation at the l Pr promoter more than 30fold (Leirmo et al., 1987).
Thus, there is ample precedent for the idea that
certain anions may compete with the pause hairpin
for binding to RNA polymerase. However, the
effects of KCl and KGlu on open complex formation differ from those we observe on elongation
and pausing. Both KOAc and KGlu inhibited
elongation to a roughly equal and signi®cant extent, both at the pause site and other template positions (C.L.C., unpublished results), whereas KCl is
a modest elongation inhibitor at most sites, but
stimulates NTP addition at the pause site.
Figure 9. Effects of KCl on pause half-life on wild-type
and pause mutant templates. Pause half-lives were
measured from time-courses more extensive than shown
in Figure 8 either by non-linear regression of the concentration of pause RNA versus time or by kinetic approximation (indicated by asterisks; see Materials and
Methods). Pause half-lives are plotted relative to wildtype at low salt and appropriate [GTP], which is arbitrarily set equal to 1 and should be accurate to 10 to 15%
(see the legend to Figure 4(d)).
tein structure play an important role in altering
pausing and elongation. Stabilizing anions increased the pause half-life, whereas halides and
chaotropes decreased it. Anions at either end of the
Hofmeister series strongly inhibited the overall
rate of transcript elongation, whereas intermediate
anions like Clÿ had the least effect.
Effect of salts on protein-nucleic acid interaction
The strongest effects of salts on protein-nucleic
acid interaction arise from release of ions bound to
the DNA or protein upon interaction (Lohman,
1986). At low salt concentrations, the entropically
favored release of cations bound to DNA is a predominant driving force for DNA binding by at
least some well-studied proteins, such as the lac repressor (deHaseth et al., 1977; Revzin & von
Hippel, 1977; Barkley, 1981). However, the effects
on transcription reported here do not appear to be
caused by cations (see above).
Signi®cant differences in the ability of anions to
inhibit protein-nucleic acid interactions also have
been reported. These effects have been attributed
to speci®c binding of certain anions to nucleic acidbinding domains in proteins. For instance, both
speci®c and non-speci®c binding of lac repressor to
DNA are decreased signi®cantly in KCl (or potassium acetate) but not potassium glutamate, leading
to the suggestion that ``Clÿ and Acÿ compete with
phosphate groups in binding to lac R,'' whereas
glutamate binds lac repressor poorly (Ha et al.,
1992). Replacing KCl in buffers with equivalent
concentrations of potassium glutamate (KGlu) dramatically increases the activity of some restriction
endonucleases and increases the rate of open com-
Stabilizing anions may destabilize some
conformations of RNA polymerase that form
during RNA chain elongation
Although chaotropic ions almost certainly inhibit
elongation by binding to polymerase, the surprisingly greater inhibition by stabilizing ions like
2ÿ
ÿ
SO2ÿ
4 , OAc , and HPO4 might re¯ect either binding or a Hofmeister effect on protein hydration.
These anions could bind in RNA polymerase's active site, or in channels that position the downstream DNA duplex, the separated template and
non-template strands, or the nascent RNA. Alternatively, they might destabilize conformations of
RNA polymerase that are formed during transcript
elongation and that expose more hydrophobic surface, which would be increasingly disfavored as
protein hydration is increased.
We favor the second explanation for three
reasons. First, OAcÿ is least similar to a phosphodiester in geometry and is monovalent, but inhibits
transcription as effectively as SO2ÿ
4 . It seems unli2
ÿ
kely SO2ÿ
4 , OAc , and HPO4 would bind to polymerase equivalently. Second, elongation is slowed
to an equivalent extent by the protein-stabilizing
cation N(CH3)‡
4 (I. Artsimovitch & R. L., unpublished results). Being positively charged, N(CH3)‡
4
certainly does not bind to the same sites as the
anions, but could increase protein hydration by a
Hofmeister effect (von Hippel & Schleich, 1969a).
Finally, Erijman & Clegg (1995) reported that
pressures above 200 KPa reversibly halt elongation
complexes. Both high pressure and stabilizing
anions can affect protein structure by disordering
the normal structure of water and increasing protein hydration (Leberman & Soper, 1995), although
pressure might also favor protein conformations
with lower overall volume. In either case, like stabilizing anions, high pressure may restrict the conformational ¯exibility of RNA polymerase.
The degree to which RNA polymerase ¯exibility is required for elongation is currently in
dispute. Inchworm-like movements of RNA polymerase have been proposed to explain changes in
the size of halted elongation complex footprints
on DNA and RNA, in some cases as components
of pausing or termination mechanisms (Krummel
& Chamberlin, 1992a; Chan & Landick, 1994;
Nudler et al., 1994, 1995; Chamberlin, 1995; Wang
48
et al., 1995; Zaychikov et al., 1995). More recently,
the same footprint changes have been explained
by sliding of a rigid RNA polymerase along the
RNA and DNA chains (Reeder & Hawley, 1996;
Komissarova & Kashlev, 1997). Although our results do not directly distinguish these models,
they appear at odds with the extreme case of a
completely rigid RNA polymerase that undergoes
no signi®cant conformational change during
elongation and suggest that some degree of conformational ¯exibility is required for rapid chain
elongation.
The effects of neutral solutes on pausing
suggest an easily disordered region of RNA
polymerase interacts with the pause hairpin
The ability of some solutes to reduce pausing at
the his leader site most likely results from disruption of an interaction between the pause hairpin
and RNA polymerase, but by two distinct mechanisms: direct competition for hairpin binding or disordering of the structure of the binding site.
We favor the idea that these solutes interfere
with a pause hairpin interaction for four reasons.
First, the pause hairpin is a feature that distinguishes the pause from other positions on the
template where anions inhibited nucleotide addition. Second, only substitutions in the pause hairpin make the pause site respond to Clÿ like other
template positions. Third, increasing the stem
length of the pause hairpin seems to mimic the Clÿ
effect by precluding a hairpin ± RNA polymerase.
This idea is explored further in the accompanying
paper (Chan et al., 1997). Finally, we observed an
increased rate of nucleotide addition at a few other
sites on the his leader template at elevated concentrations of Clÿ and these sites mapped to locations
where the C:D RNA secondary structure conceivably could form and interact with RNA polymerase
(see Results).
The idea that either disordering the region of
RNA polymerase that the pause hairpin contacts
or directly competing for the interaction by ionbinding stems primarily from comparison of the
properties of Clÿ and MPD. The chaotropes Iÿ,
ÿ
NOÿ
3 , and ClO4 , which decrease pausing most dramatically, could disrupt an interaction between the
pause hairpin and RNA polymerase by either
mechanism. Clÿ, on the other hand, has minimal
effects on the stability of most protein tertiary
structures and probably interferes with pause hairpin interaction by simple competition. In fact, Clÿ
is known to bind polymerase and inhibit its dimerization (Record et al., 1978; Shaner et al., 1982, 1983;
Leirmo et al., 1987).
MPD, however, is uncharged and thus is unlikely to bind to the same sites as anions. MPD decreases the stability of the folded state of proteins
because it interacts favorably with non-polar
amino acid side-chains, thus tending to unfold the
protein and disperse the charges on the surface of
the protein (Timasheff, 1993). Although it acts by a
Salt Effects on RNA Chain Elongation and Pausing
different mechanism, MPD, like chaotropic anions,
destabilizes the folded conformation of proteins
and promotes denaturation. Thus, its ability to accelerate transcription at the pause site is most easily explained if it disorders a region of protein
that ordinarily interacts with the pause hairpin,
rather than competes directly for binding.
The susceptibility of this putative hairpin-binding site to unfolding, the ability of Clÿ or Brÿ to
compete with the pause hairpin for binding, the
similarity of the effect of the pause hairpin to that
of stabilizing anions like OAcÿ (both slow
elongation, probably by slowing conformational
changes), and the ®nding that the pause hairpin
stabilizes an extended DNA footprint at the pause
site (Wang et al., 1995) together lead to the following idea. The pause hairpin may inhibit elongation
by interacting with an easily disordered surface on
RNA polymerase, ¯exibility of which is required
for translocation and rapid transcription. In support of this conclusion, the pause hairpin can be
crosslinked to a region in the b subunit directly adjacent to a known surface-exposed segment near
amino acid residue 1000 (Wang & Landick, unpublished results). The elucidation of the exact nature
of pause hairpin-polymerase interaction must
await more detailed structural studies.
Prospects for further study of solute effects on
transcript elongation and pausing
This preliminary survey of solute effects on transcript elongation and pausing revealed two mechanistically important effects: strong inhibition of
transcript elongation by protein-stabilizing anions
and reduction of pausing by halides, chaotropes
and MPD. These ®ndings provide ample justi®cation for future, more detailed studies to extend understanding of solute effects on transcription. A
survey of cation effects, for instance, is needed.
The ability to examine individual transcription
complexes poised at any template position (Nudler
et al., 1994, 1995; Wang et al., 1995) may make it
possible to apply quantitative approaches to solute
effects on transcript elongation. It also would be interesting to examine the denaturing and stabilizing
solutes we tested in more detail as well as others
like urea, guanidinium HCl, glycerol and NaGlu.
By combining study of solute effects with mutant
RNA polymerases it may be possible to discern
separate effects on individual domains within the
transcription complex.
Materials and Methods
Enzymes and reagents
RNA polymerase was puri®ed from Escherichia coli
strain MRE600 (Midwest Grain Processing Co., Muscatine, IA) using the method of Burgess & Jendrisak (1975)
with modi®cations described by Hager et al. (1990). Restriction endonucleases were purchased from New England Biolabs and used according to the manufacturer's
instructions. Synthetic oligonucleotides were prepared
49
Salt Effects on RNA Chain Elongation and Pausing
on an Applied Biosystems model 380B DNA synthesizer
using cyanoethyl phosphoramidites and following procedures from the manufacturer. Salts, deoxyribonucleotides and ApU dinucleotide were obtained from Sigma;
high-pressure liquid chromatography-puri®ed nucleoside triphosphates (NTPs), from Pharmacia LKB Biotechnology Inc.; Taq DNA polymerase, from Promega; and
[a-32P]GTP and [a-35S]dATP, from Amersham Corp.
DNA manipulations
DNA templates for transcription were ampli®ed by
PCR from 200 ng of plasmid DNA in a 100 ml reaction
as described (Chan & Landick, 1993), using primers hybridizing upstream from the promoter and downstream
from the pause site. The ampli®ed DNA products were
recovered as previously (Chan & Landick, 1993), except
that precipitation with spermine was used to remove unincorporated primers and dNTPs (Hoopes & McClure,
1981).
Plasmid DNAs were isolated by the alkaline lysis procedure (Birnboim & Doly, 1979) after transformation into
strain JM109 and used directly for preparation of DNA
fragments for in vitro transcription reactions by PCR
(Higuchi et al., 1988) or, following treatment with RNase,
for thermal cycle sequencing.
DNA sequencing was performed using Taq DNA
polymerase, dideoxyNTPs, [a-35S]dATP and oligonucleotide primers following a thermal cycling protocol from
Promega. All other DNA manipulations were performed
by standard methods (Maniatis et al., 1982; Ausubel et al.,
1989)
Plasmid construction and DNA templates
For most experiments, we used templates generated
from plasmid pCL185 or mutant derivatives of pCL185
(Feng et al., 1994). Mutant templates were constructed by
transferring the MluI-BglII fragment from the pCL208
derivatives described by Chan & Landick (1993) to
pCL185. These templates formed initial transcription
complexes halted at G16 in the absence of UTP (see
Figure 1). These templates and the perfect hairpin template were generated using the pUC-plasmid lacZ0 -20
universal primer (primer 1; Figure 1) and 5 0 -d[GGCTTTCTCATGC-GTTCATGC]-30 (primer 2; Figure 1; 50
end corresponds to C-149 of the his leader). This produced a 427 bp DNA fragment that did not include the
his leader terminator (Chan & Landick, 1993).
The perfect hairpin pause signal (Figures 6 and 7) was
constructed by ligating complementary oligonucleotides
containing the appropriate base substitutions between
the unique MluI and SpeI restriction sites in the his leader
of pCL103 (Chan & Landick, 1989). pCL103 contains a
promoter-proximal fragment from the his operon of Salmonella typhimurium. The plasmid pCL186 encoding the
perfect hairpin pause site downstream from the phage
T7 A1 promoter was then constructed by ligating the
MluI-SphI fragment from the pCL103 derivative between
the same two sites in pCL172 (Chan & Landick, 1993).
The perfect hairpin template was ampli®ed with primers
1 and 2, but formed complexes halted at A20 (Figure 1).
The downstream DNA mutant and its wild-type control plasmid (pDW309 and pCL102b, respectively;
Figure 8(g) and (h)) were constructed as follows.
pCL102b was produced from pCL185 (Feng et al., 1994)
by cleavage with BamHI, treatment with mung bean nuclease, cleavage with HindIII, and ligation of the large
vector fragment to a DraI-HindIII fragment from
pRL671. pRL671 itself was a derivative of pCL185 in
which the sequence downstream from the pause site was
changed by PCR mutagenesis to 50 -GGAAGACATTCAGAGCTGCACTT-CGAAGATCTT (®rst G corresponds
to base 103 in the wild-type his leader and the last T to
base 120; Figure 1). pDW309 was created similarly from
plasmid pRL695, where the corresponding sequence was
50 -GCGAGAGTAGGGAACCTGCACTTGCAAGTGCAGGTTCCCTGACTCTCGAGAGTAGGGACTGCACTTGCAGTGCAGGGTTACTCTCGAGATGTT. This altered
the downstream DNA sequence at the pause site to one
previously shown to reduce pausing at the trp leader
pause site (Lee et al., 1990; and see Results). This template and the wild-type control (from pCL102b; results
not shown here) formed initial transcription complexes
halted at A29 in the absence of UTP (Wang et al., 1995;
Figure 1) and eliminated weak pause sites around position ‡30 that make measurement of short half-lives dif®cult (see Wang et al., 1995; Chan et al., 1997). They were
generated using primer 2 (Figure 1) and an upstream primer (50 -d[GGAGAGACAACTTAAAGAGA]-30 ) that hybridizes closer to the T7 A1 transcription start site,
between positions ÿ92 and ÿ73.
In vitro transcription reactions
Transcription reactions were initiated in transcription
buffer (20 mM Tris-acetate (pH 8.0), 10 mM MgCl2,
20 mM NaCl, 14 mM û-mercaptoethanol, 0.1 mM EDTA,
4% (v/v) glycerol, 10 mg acetylated bovine serum albumin/ml) as described (Chan & Landick, 1993). The concentrations of neutral salts reported for experiments at
elevated salt concentrations do not include contributions
from this buffer.
G16, A20 or A29 complexes were formed by incubating 40 nM DNA, 250 mM ApU dinucleotide, 2.5 mM ATP
and CTP, 2 mM [a-32P]GTP (speci®c activity 200 Ci/
mmol) and 50 nM RNA polymerase at 37 C for ten minutes. Solutions of these halted complexes retained the capacity to resume chain elongation for greater than four
hours when stored on ice. Complexes were diluted on
ice with 0.5 vol. of a solution that produced concentrations of transcription buffer, rifampicin, and GTP
1.3 those given below. The diluted G16 or A20 complexes were then combined with 1/3 vol. of an appropriate salt solution (or water) approximately one minute
prior to addition of heparin to 150 mg/ml and ATP, UTP
and CTP to 150 mM each. The ®nal concentration in transcription buffer was rifampicin 80 mg/ml, and GTP 10 or
20 mM (see Figure legends). To ensure that different transcription reactions contained identical concentrations of
buffer components, NTPs, and transcription complexes,
portions of the same solution of halted complexes were
added to solutions of the different neutral salts.
After transcript elongation was initiated by addition
of UTP, samples (5 ml) were removed from the reactions
at the times indicated in the Figure legends and added
to an equal volume of stop mix (178 mM Tris-borate (pH
8.3), 5 mM Na2EDTA, 0.5% (w/v) bromophenol blue
and xylene cyanol, saturated with urea). The samples
were heated to 65 C for ®ve minutes and then electrophoresed through an 0.4 mm thick, 10% (w/v) polyacrylamide gel (19:1 (w/w) acrylamide to bis-acrylamide)
containing 7 M urea and either 1 or 0.5 TBE (TBE is
88 mM Tris-borate, 2.5 mM EDTA). 1 TBE gels were
used to separate RNAs in samples containing high con-
50
Salt Effects on RNA Chain Elongation and Pausing
centrations of salt, which greatly reduced electrophoretic
artifacts caused by the salt front. Radioactivity in the
separated RNA bands was measured using a Molecular
Dynamics Phosphorimager and software supplied by the
manufacturer.
Estimation of average arrival time at the his leader
pause site
The average arrival time at the pause site in different
salt conditions was estimated using the programs KINSIM and FITSIM (Barshop et al., 1983; Frieden, 1993), to
model the movement of transcription complexes from
G16 to the pause site. The rate of transcription complex
arrival at the pause site was measured using the rate of
increase in the Phosphorimager signal (minus background) from all RNA species at or beyond the pause
site in experiments such as depicted in Figure 2, except
with more samples for each condition.
We tested the ability of four types of mechanisms to
®t these data. All four mechanisms assumed pseudo-®rst
order chain extension, since the concentration of NTPs
was in substantial excess and pyrophosphorolysis (chain
shortening by chemical reverse of NTP addition) should
be negligible. Mechanism A was a simple sequential
mechanism with one intermediate (I) between G16 and
the paused complex (PTC, used here to designate all
transcription complexes at or past the pause site).
Mechanism A:
G16 ! I ! PTC
…1†
Mechanism B added an isomerization step similar to that
found necessary to ®t chain elongation data in the studies by Erie et al. (1993) and Matsuzaki et al. (1994).
Mechanism B:
G16 ! I ! PTC
"#
I
…2†
Not surprisingly, given the minimal number of steps involved (76), neither mechanism A nor mechanism B produced acceptable ®ts to the data (Figure 3(a)). However,
these experiments were conducted at limiting GTP and
there are only 13 G bases between G16 and the pause.
Thus, the rate of movement likely is determined principally by at most 13 slow steps. To see if an even simpler
model would be adequate, we tested mechanism C with
seven sequential intermediates.
Mechanism C:
G16 ! I1 ! I2 ! I3 ! I4 ! I5 ! I6 ! I7 ! PTC …3†
This mechanism ®t the observed data acceptably at both
low salt and 1 M KCl. Interestingly, addition of an isomerization step (mechanism D) did not improve the ®t
to the G16 ! PTC arrival data and so was not used for
the analysis.
Mechanism D:
G16 ! I1 ! I2 ! I3 ! I4 ! I5 ! I6 ! I7 ! PTC
"#
I4 …4†
To estimate the average time for transit of transcription
complexes between G16 and the pause under the different solute conditions tested, we used FITSIM to ®nd
values for the pseudo-®rst-order rate constants k1 to k8
implicit in mechanism C (see scheme 5) that gave a least-
squares ®t to the experimental data, and then computed
the average arrival time by numerical integration
(Figure 3(b)).
k1
k2
k3
k4
k5
k6
k7
k8
G16 ! I1 ! I2 ! I3 ! I4 ! I5 ! I6 ! I7 ! PTC
…5†
k1 to k8 were held equal to each other but FITSIM was
allowed to vary them in concert to ®nd the best ®t to
the observed concentrations of RNA species of length
92 nt (pause RNA) and greater (e.g. Figure 3(a) and
(b)). Average arrival times for the different transcription
conditions were then computed from these model progress curves (see Results and Figure 3(a) and (b)) using
equation (6):
P
…a t †
Pnn ;
…6†
…an †
where an is the amount of RNA predicted to arrive at the
pause in time interval n when kinetic simulation is divided into n intervals of length t. See Landick et al.
(1996) for a complete description of this approach to
kinetic modeling.
Pause half-life measurements
For most cases, pause half-lives were determined from
the pseudo-®rst order rates of pause RNA disappearance, as calculated by non-linear regression analysis of
the pause RNA concentration versus time, after subtraction of appropriate background values (see Results and
Figure 4). Times were carefully selected to ensure that
complexes arriving at the pause site slowly did not bias
the measurement. In the case of the 30 -end substitution
(Figures 8 and 9), appearance of a new pause site after
incorporation of U99 and short half-lives precluded accurate measurement by this method. For these experiments,
individual half-lives were estimated by ®rst measuring
the total delay caused by the paused species
(U99 ‡ C102). The contribution of each to the overall
delay was then estimated by ®rst obtaining their relative
half-lives (or rate constants) from time-points in which
transient steady-state conditions were maintained. When
the relative concentrations of RNA species do not change
during a transient steady-state (e.g. the number of polymerases arriving at the sites is equal to the number leaving), the ratio of their rate constants for nucleotide
addition is equal to the ratio of the concentrations of the
RNA species (see Matzusaki et al., 1994). These ratios
were 1:5 at low salt and 1:1 at 1 M KCl. We then determined the approximate half-life of polymerases paused
at the individual sites assuming that the overall delay
would be the sum of the individual delays (or written in
terms of pseudo-®rst order rate constants, equation (7):
1
1
1
ˆ ‡
kobs k1 k2
…7†
We also used this method to estimate pause half-lives for
the 30 -end mutant plus multisubstituted hairpin and
downstream DNA mutant plus multisubstituted hairpin
combination templates in the accompanying paper
(Chan et al., 1997). It yields reasonable approximations of
pause half-lives, but depends on an assumption that
RNA polymerase molecules traversing the region in
question stop at all successive slow steps if they stop at
all. An alternative possibility is that the populations of
polymerases pausing at different successive sites are distinct. For the 30 -end mutants considered here and in the
accompanying paper, independent lines of reasoning
51
Salt Effects on RNA Chain Elongation and Pausing
that are described in the respective Results sections verify the conclusions based on these approximated halflives (and thus make it likely the assumption is correct).
Filter-binding experiments
Nitrocellulose ®lter-binding was used to distinguish
between ternary transcription complexes and free RNA,
following the procedure of Morgan et al. (1984). Nitrocellulose ®lters (BA85, 13 mm, 0.45 mm, Schleicher and
Schuell) were prepared by wetting with distilled H2O,
placed in boiling 20 mM Tris-HCl (pH 8), 0.5 M KCl, and
allowed to cool to room temperature. Filters were rinsed
immediately prior to use with ®lter-rinse solution
(20 mM Tris-HCl (pH 8), 0.5 M KCl, 50 mg heparin/ml,
10 mg tRNA/ml) to block sites where nucleic acids nonspeci®cally stick to the ®lters. Samples (10 ml) from
single-round transcription reactions were diluted tenfold with ®lter ± rinse solution. Half the sample was applied to the ®lter and the ®lter was rinsed with 200 ml of
®lter rinse solution. The ¯ow-through and rinse, which
together contained the released RNA transcripts, were
combined. RNA polymerase-associated transcripts were
eluted from the ®lter at 95 C for ten minutes with
300 ml of ®lter-elution buffer (20 mM Tris-HCl (pH 8),
1% (w/v) SDS, 0.3 M NaOAc, 10 mg tRNA/ml) and precipitated by addition of 750 ml of ethanol. Filter-rinse solution (200 ml) was added to the remaining half of the
aliquot, which served as a total RNA control. RNAs
from the control and ¯ow-through samples were precipitated by addition of 50 ml of ®lter-elution buffer and
750 ml of ethanol. These samples and a control un®ltered
sample were electrophoresed through a denaturing 10%
polyacrylamide gel and the amounts of RNA associated
with the transcription complex and released from it
were determined by comparison of the amounts of the
RNA in each lane.
Mapping RNA 30 ends
RNA transcript 30 ends were mapped by comparison
to ladders generated using 30 deoxyNTPs as described
(Chan & Landick, 1993).
Immobilized transcription complexes
G16 complexes were assembled with RNA polymerase
that had the biotinated segment of protein from the
E. coli biotin carrier protein fused genetically to the C terminus of b0 . This enzyme will be described elsewhere.
G16 complexes were isolated from the excluded volume
of a Sephadex G50 (Sigma) gel-®ltration column equilibrated with transcription buffer by low-speed centrifugation (Morgan et al., 1984; Levin et al., 1987). Complexes
were immobilized on streptavidin-coated magnetic
beads (Dynal Inc.) by incubating at room temperature in
transcription buffer plus 0.15 M KCl on a rotating wheel
for 45 to 60 minutes. The immobilized complexes were
collected by centrifuging brie¯y and then rinsed three
times with transcription buffer plus 1 M KCl to remove
any complexes that were not stably bound to the beads.
Complexes were resuspended in the appropriate buffer
and elongation at 37 C was resumed by addition of
NTPs and heparin. Aliquots (10 ml) were removed approximately 30 seconds prior to each time indicated and
spun brie¯y to pellet the beads. At the times indicated, a
5 ml portion of the supernatant was removed and combined with 5 ml of stop solution and 5 ml of stop solution
was added to beads and supernatant remaining in the
tube. The fractions of the RNA released from the transcription complexes were determined by comparison of
the pause RNAs in the supernatant samples to the total
pause RNAs after electrophoresis of the samples through
a denaturing 10% polyacrylamide gel.
Acknowledgments
We thank Curtis Altmann and Michael Chamberlin
for discussions and sharing of data prior to publication,
Peter von Hippel, Tim Lohman and Tom Record for
helpful suggestions during the course of this work, and
Irina Artsimovitch, Carol Gross and Rachel Mooney for
critical reading of the manuscript. This study was supported by the NIH (GM-38660) and by the NSF through
a Presidential Young Investigator Award (DMB8957331).
References
Arndt, K. M. & Chamberlin, M. J. (1990). RNA chain
elongation by Escherichia coli RNA polymerase. Factors affecting the stability of elongating ternary
complexes. J. Mol. Biol. 213, 79± 108.
Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D.,
Siedman, J. G., Smith, J. A. & Struhl, K. (1989). In
Current Protocols in Molecular Biology. John Wiley &
Sons, New York.
Barkley, M. D. (1981). Salt dependence of the kinetics of
the lac repressor-operator interaction: role of nonoperator deoxyribonucleic acid in the association
reaction. Biochemistry, 20, 3833± 3842.
Barshop, B. A., Wrenn, R. F. & Frieden, C. (1983).
Analysis of numerical methods for computer
simulation of kinetic processes: development of
KINSIM± a ¯exible, portable system. Anal. Biochem.
130, 134± 145.
Birnboim, H. C. & Doly, J. (1979). A rapid alkaline
extraction procedure for screening recombinant
plasmid DNA. Nucl. Acids Res. 7, 1513± 1523.
Borukhov, S., Sagitov, V. & Goldfarb, A. (1993). Transcript cleavage factors from E. coli. Cell, 72, 459 ±
466.
Burgess, R. R. & Jendrisak, J. J. (1975). A procedure for
the rapid, large-scale puri®cation of Escherichia coli
DNA-dependent RNA polymerase involving
polymin P precipitation and DNA-cellulose
chromatography. Biochemistry, 14, 4634± 4638.
Chamberlin, M. J. (1995). New models for the mechanism of transcription elongation and its regulation.
In The Harvey Lectures, pp. 1 ±21. Wiley-Liss, New
York.
Chan, C. & Landick, R. (1989). The Salmonella typhimurium his operon leader region contains an RNA hairpin-dependent transcription pause site. J. Biol.
Chem. 264, 20796± 20804.
Chan, C. L. & Landick, R. (1993). Dissection of the his
leader pause site by base substitution reveals a multipartite signal that includes a pause RNA hairpin.
J. Mol. Biol. 233, 25± 42.
Chan, C. L. & Landick, R. (1994). New perspectives on
RNA chain elongation and termination by E. coli
RNA polymerase. In Transcription: Mechanisms and
Regulation (Conaway, R. C. & Conaway, J. W., eds),
pp. 297± 321. Raven Press, New York.
52
Chan, C. L., Wang, D. & Landick, R. (1997). Multiple
interactions stabilize a single paused intermediate
in which hairpin to 30 end spacing distinguishes
pause and termination pathways. J. Mol. Biol. 268,
54± 68.
Collins, K. & Washabaugh, M. (1985). The Hofmeister
effect and the behaviour of water at interfaces.
Quart. Rev. Biophys. 18, 323 ± 422.
deHaseth, P. L., Lohman, T. M. & Record, M. T.
Jr (1977). Nonspeci®c interaction of lac repressor
with DNA: An association reaction driven by counterion release. Biochemistry, 16, 4783± 4790.
Erie, D. A., Hajiseyedjavadi, O., Young, M. C. & von
Hippel, P. H. (1993). Multiple RNA polymerase
conformations and GreA: control of ®delity of
transcription. Science, 262, 867± 873.
Erijman, L. & Clegg, M. (1995). Heterogeneity of E. coli
RNA polymerase revealed by high pressure. J. Mol.
Biol. 253, 259 ± 265.
Feng, G., Lee, D. N., Wang, D., Chan, C. L. & Landick,
R. (1994). GreA-induced transcript cleavage in transcription complexes containing Escherichia coli RNA
polymerase is controlled by multiple factors, including nascent transcript location and structure. J. Biol.
Chem. 269, 22282± 22294.
Frieden, C. (1993). Numerical integration of rate equations by computer. Trends Biochem. Sci. 18, 58± 60.
Ha, J.-H., Capp, M. W., Hohenwalter, M. D., Baskerville,
M. & Record, M. T. Jr (1992). Thermodynamic stoichiometries of participation of water, cations and
anions in speci®c and non-speci®c binding of lac
repressor to DNA: Possible thermodynamic origins
of the ``glutamate effect'' on protein-DNA
interactions. J. Mol. Biol. 228, 252± 264.
Hager, D. A., Jin, D. J. & Burgess, R. R. (1990). Use of
Mono Q high-resolution ion-exchange chromatography to obtain highly pure and active Escherichia
coli RNA polymerase. Biochemistry, 29, 7890± 7894.
Hamaguchi, K. & Geiduschek, E. P. (1962). The effect of
electrolytes on the stability of the deoxyribonucleate
helix. J. Am. Chem. Soc. 84, 1329± 1338.
Harrison, S. & Agarwal, A. (1990). DNA recognition by
proteins with the helix-turn-helix motif. Annu. Rev.
Biochem. 59, 933 ± 969.
Higuchi, R., Krummel, B. & Saiki, R. K. (1988). A general method of in vitro preparation and speci®c
mutagenesis of DNA fragments: study of protein
and DNA interactions. Nucl. Acids Res. 16, 7351±
7367.
Hoopes, B. C. & McClure, W. R. (1981). Studies on the
selectivity of DNA precipitation by spermine. Nucl.
Acids Res. 9, 5493± 5504.
Jencks, W. P. (1969). Catalysis in Chemistry and Enzymology. McGraw-Hill, New York.
Kassavetis, G. A. & Chamberlin, M. J. (1981). Pausing
and termination of transcription within the early
region of bacteriophage T7 DNA in vitro. J. Biol.
Chem. 256, 2777± 2786.
Komissarova, N. & Kashlev, M. (1997). Arrest of transcription: E. coli RNA polymerase translocates backward leaving the 30 end of the RNA intact and
extruded. Proc. Natl Acad. Sci. USA, in the press.
Krummel, B. & Chamberlin, M. J. (1992a). Structural
analysis of ternary complexes of Escherichia coli
RNA polymerase. Deoxyribonuclease I footprinting
of de®ned complexes. J. Mol. Biol. 225, 239± 250.
Krummel, B. & Chamberlin, M. J. (1992b). Structural
analysis of ternary complexes of Escherichia coli
RNA polymerase. Individual complexes halted
Salt Effects on RNA Chain Elongation and Pausing
along different transcription units have distinct and
unexpected biochemical properties. J. Mol. Biol. 225,
221± 237.
Landick, R. & Yanofsky, C. (1984). Stability of an RNA
secondary structure affects in vitro transcription
pausing in the trp operon leader region. J. Biol.
Chem. 259, 11550± 11555.
Landick, R., Wang, D. & Chan, C. (1996). Quantitative
analysis of transcriptional pausing by RNA polymerase: the his leader pause site as a paradigm.
Methods Enzymol. 274, 334±352.
Leberman, R. & Soper, A. (1995). Effect of high salt concentrations on water structure. Nature, 378, 364 ±
366.
Lee, D. N., Phung, L., Stewart, J. & Landick, R. (1990).
Transcription pausing by Escherichia coli RNA polymerase is modulated by downstream DNA
sequences. J. Biol. Chem. 265, 15145± 15153.
Leirmo, S., Harrison, C., Cayley, D. S., Burgess, R. R. &
Record, M. T. (1987). Replacement of potassium
chloride by potassium glutamate dramatically
enhances protein-DNA interactions in vitro. Biochemistry, 26, 2095± 2101.
Levin, J. R., Krummel, B. & Chamberlin, M. J. (1987).
Isolation and properties of transcribing ternary
complexes of Escherichia coli RNA polymerase positioned at a single template base. J. Mol. Biol. 196,
85±100.
Lohman, T. M. (1986). Kinetics of protein-nucleic acid
interactions: Use of salt effects to probe mechanisms
of interaction. CRC Crit. Rev. Biochem. 19, 191 ± 245.
Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982). In Molecular Cloning: A Laboratory Manual. Cold Spring
Harbor Laboratory Press, Cold Spring Harbor, NY.
Matsuzaki, H., Kassavetis, G. A. & Geiduschek, E. P.
(1994). Analysis of RNA chain elongation and termination by Saccharomyces cerevisiae RNA polymerase III. J. Mol. Biol. 235, 1173± 1192.
Misra, V., Sharp, K., Friedman, R. & Honig, B. (1994).
Salt effects on protein-DNA interactions. The lcI
repressor and EcoRI endonuclease. J. Mol. Biol. 238,
264 ± 280.
Morgan, W. D., Bear, D. G. & von Hippel, P. H. (1984).
Speci®city of release by Escherichia coli transcription
termination factor rho of nascent mRNA transcripts
initiated at the l PR promoter. J. Biol. Chem. 259,
8664± 8671.
Nudler, E., Goldfarb, A. & Kashlev, M. (1994). Discontinuous mechanism of transcription elongation.
Science, 265, 793 ± 796.
Nudler, E., Kashlev, M., Nikiforov, V. & Goldfarb, A.
(1995). Coupling between transcription termination
and RNA polymerase inchworming. Cell, 81, 351 ±
357.
Record, M. T. Jr, Anderson, C. F. & Lohman, T. M.
(1978). Thermodynamic analysis of ion effects on
the binding and confromational equilibria of proteins and nucleic acids: the roles of ions association
or release, screening, and ion effects on water
activity. Quart. Rev. Biophys. 11, 103± 178.
Reeder, T. & Hawley, D. (1996). Promoter proximal
sequences modulate RNA polymerase II elongation
by a novel mechanism. Cell, 87, 767± 777.
Revzin, A. & von Hippel, P. H. (1977). Direct measurement of association constants for the binding of
Escherichia coli lac repressor to non-operator DNA.
Biochemistry, 16, 4769± 4776.
Reynolds, R. & Chamberlin, M. J. (1992). Parameters
affecting transcription termination by Escherichia coli
Salt Effects on RNA Chain Elongation and Pausing
RNA polymerase. II. Construction and analysis of
hybrid terminators. J. Mol. Biol. 224, 53 ± 63.
Reynolds, R., BermuÂdez-Cruz, R. M. & Chamberlin, M.
J. (1992). Parameters affecting transcription termination by Escherichia coli RNA polymerase. Analysis
of 13 rho-independent terminators. J. Mol. Biol. 224,
31± 51.
Robinson, D. R. & Grant, M. E. (1966). The effects of
aqueous salt solutions on the activity coef®cients of
purine and pyrimidine bases and their relation to
the denaturation of deoxyribonucleic acid by salts.
J. Biol. Chem. 241, 4030± 4042.
Schildkraut, C. & Lifson, S. (1965). Dependence of the
melting temperature of DNA on salt concentration.
Biopolymers, 3, 195± 208.
Shaner, S. L., Piatt, D. M., Wensley, C. G., Yu, H.,
Burgess, R. R. & Record, M. T. Jr (1982). Aggregation
equilibria of Escherichia coli RNA polymerase: evidence for anion-linked conformational transitions in
the protomers of core and holoenzyme. Biochemistry,
21, 5539± 5551.
Shaner, S. L., MelancËon, P., Lee, K. S., Burgess, R. R. &
Record, M. T. Jr (1983). Ion effects on the aggregation and DNA-binding reactions of E. coli RNA
polymerase. Cold Spring Harbor Symp. Quant. Biol.
47, 463± 472.
Telesnitsky, A. & Chamberlin, M. J. (1989). Terminatordistal sequences determine the in vitro ef®ciency of
the early terminators of bacteriophages T3 and T7.
Biochemistry, 28, 5210± 5218.
Timasheff, S. N. (1993). The control of protein stability
and association by weak interactions with water:
53
how do solvents affect these processes?. Annu. Rev.
Biophys. Biomol. Struct. 22, 67± 97.
Tunis, M.-J. B. & Hearst, J. E. (1968). On the hydration
of DNA. I. Preferential hydration and stability of
DNA in concentrated tri¯uoroacetate solution. Biopolymers, 6, 1325± 1344.
von Hippel, P. H. & Schleich, T. (1969a). The effects of
neutral salts on the structure and conformational
stability of macromolecules in solution. In Structure
and Stability of Biological Macromolecules (Timasheff,
S. N. & Fasman, G. D., eds), pp. 417± 574. Marcel
Dekker, New York.
von Hippel, P. H. & Schleich, T. (1969b). Ion effects on
the solution structure of biological macromolecules.
Acc. Chem. Res. 2, 257± 265.
Wang, D., Meier, T., Chan, C., Feng, G., Lee, D. &
Landick, R. (1995). Discontinuous movements of
DNA and RNA in E. coli RNA polymerase accompany formation of a paused transcription complex.
Cell, 81, 341 ± 350.
Winkler, M. E. & Yanofsky, C. (1981). Pausing of RNA
polymerase during in vitro transcription of the
tryptophan operon leader region. Biochemistry, 20,
3738± 3744.
Zaychikov, E., Denissova, L. & Heumann, H. (1995).
Translocation of the Escherichia coli transcription
complex observed in the registers 11 to 20: ``jumping'' of RNA polymerase and asymmetric expansion and contraction of the transcription bubble.
Proc. Natl Acad. Sci. USA, 92, 1739± 1743.
Edited by P. E. Wright
(Received 18 June 1996; received in revised form 16 January 1997; accepted 17 January 1997)