Nucleic Acids Research, Vol. 20, No. I
A CT promoter element binding protein: definition of a
double-strand and a novel single-strand DNA binding motif
Rukmini Kolluri, Ted Albert Torrey and Alan J.Kinniburgh*
Department of Human Genetics, Roswell Park Cancer Institute, Elm and Carlton Streets, Buffalo,
NY 14263, USA
Received September 10, 1991; Revised and Accepted November 22, 1991
Numerous genes contain promoter elements that are
nuclease hypersensitive. These elements frequently
possess polypurine/polypyrimidine stretches and are
usually associated with altered chromatin structure. We
have previously isolated a clone that binds a class of
CT-rich promoter elements. We have further
characterized this clone, termed the nuclease-sensitive
element protein-1, or NSEP-1. NSEP-1 binds both
duplex CT elements and the CT-rich strand of these
elements in a 'generic' sequence specific manner and
has overlapping but distinct single-and double-strand
DNA binding domains. The minimal peptide region
sufficient for both duplex and single-strand DNA
binding includes two regions rich in basic amino acids
flanking an RNP-CS-1 like octapeptide motif. Deletion
analysis shows that the single-strand DNA binding
activity is mediated by the RNP-CS-1 like octapeptide
motif and is the key peptide region necessary for
single-strand binding. NSEP-1's affinity for CT rich
promoter elements with strand asymmetry in addition
to its double- and single-strand DNA binding properties
suggests that it may be a member of a class of DNA
binding proteins that modulate gene expression by
their ability to recognize DNA with unusual secondary
Numerous genes contain cis-acting promoter elements that are
nuclease hypersensitive. These regions are nucleosome free and
are therefore hypersensitive to DNase I (1-5). Many of these
elements are also sensitive to single-strand-specific nucleases.
These dually sensitive elements have a strong purine/pyrimidine
strand asymmetry and have been termed CT elements. The basis
of these elements' nuclease sensitivities has been ascribed to an
unusual DNA conformer. This conformer, termed the H-DNA
form, is an intramolecular triplex/single-stranded structure. One
of us (6) has previously shown that a CT promoter element of
the c-myc gene can assume the H-DNA conformation under
certain conditions in vitro. These conditions include low pH
and/or high levels of supercoiling which make it unlikely that
H-DNA forms spontaneously in vitro. There is also an alternative
intramolecular, triplex that may form with a G0G *C structure
(1,lA). We have therefore worked under the hypothesis that
cellular factors may induce these changes and produce a nuclease-
sensitive conformer (H-DNA or G *G *C triplex) in vivo. We have
previously isolated a clone that binds to a related set of CT
promoter elements (7). Here we define the binding specificity
of this factor, termed nuclease-sensitive element protein-I or
NSEP-1. We show that NSEP-1 binds both duplex CT elements
and the CT-rich single strand of these elements in a 'generic'
sequence-specific manner. That these are separable binding
activities is demonstrated by the creation of mutations that disrupt
single-strand binding but do not disrupt duplex CT element
MATERIALS AND METHODS
Expression of NSEP-1 proteins and deletion derivatives
A full length cDNA clone was constructed in the pET expression
vectors system (8,9) from sub-clones of the EcoRI fragments of
the NSEP-1 phage clone in pUC13. Expression of this construct
should produce a fusion protein containing 14 amino acids derived
from the phage T7 gene 10 product of the expression vector, 6
amino acids from the pUC13 multiple cloning site lacZ coding
sequence and 22 amino acids derived from the NSEP-1 5'
untranslated region which is in frame with the methionine initiator
codon (amino acid 1) of the 322 amino acid NSEP-1 protein.
Deletion derivatives of this NSEP-1 expression fusion protein were
engineered by restriction fragment deletion where convenient, or
by polymerase chain reaction employing as primers synthetic
oligonucleotides complementary to specific portions of the NSEP-1
recognition sites for sub-cloning into the expression vectors.
Expression of the recombinant proteins in E. coli was performed
as per Studier et al. (9). All deletion constructs refer to the amino
acids ofthe mature proteins. All amino terminaly deleted NSEP-1
derivatives lack the 22 amino acids ofthe 5' untranslated NSEP-l
region but retain the d10 gene and linker amino acids. Production
of extracts of the recombinant protein-producing bacteria wasby
the procedure of Landschulz et al. (10).
*To whom correspondence should be addressed
k.. 1992Oxford UniversityPress
Nucleic Acids Research, Vol. 20, No. 1
Gel Retardation Assay
Recombinant proteins from lysates of bacterially expressed
proteins were purified by fractionation on heparin-agarose
columns (Figure 1) as described by Lichtsteiner et al. (1 1) except
that the load and wash buffers contained IX PBS with SM urea.
The procedure used was essentially that described previously (12)
except that poly (dI-dC) * poly (dI-dC) was used at 1 ytg/sample
in place of the calfthymus DNA. All binding reactions employed
-1 ng of 5'-labeled oligonucleotide probe. Specific competition
was tested by the addition of unlabeled probe prior to the addition
ofextract so that the recombinant NSEP- 1 proteins were exposed
to both target and competitor simultaneously. The amount of
recombinant NSEP-1 extract was chosen so that the probe was
in excess to the protein.
Cells from 1.5 ml aliquots of induced cultures were rapidly
pelleted and resuspended with 100 pl aliquots of SDS-PAGE
loading buffer. After heating at 100°C for five minutes, proteins
in each sample (25 pl) were resolved by 7.5% SDS-PAGE.
Proteins were transferred electrophoretically onto a nitrocellulose
filter (0.45i)using 25 mM Tris, 190 mM glycine. After transfer,
the filters were blocked with blotto overnight at 4°C and then
washed twice at 22'C with TNE-50 (50 mM Tris, 50 mM NaCl,
1 mM EDTA, 1 mM DTT). To assay for DNA binding activity,
the transfers were probed with 32P-labeled oligonucleotides
corresponding to the C-rich strand and duplex monomer of the
c-myc NSE as per the procedure of Singh et al. (13).
Isolation of Nuclease-Sensitive Element Binding Protein cDNA
We isolated several related cDNA clones by ligand screening
with an oligonucleotide that is homologous to a c-myc promoter
element. We had shown previously that this element is responsible
for approximately 80% of the c-myc transcription in c-mycICAT
fusion constructs (12). We have reported the sequence for this
cDNA and protein and discovered that our cDNA was 99%
homologous to a relatively uncharacterized clone. This clone had
been isolated using a restriction fragment that contained a
nuclease-sensitive element of the EGF receptor gene promoter
(14). We attribute this difference to sequencing errors and discuss
our rationale for this elsewhere (7). Our first objective was to
delineate the specificity of the interaction of NSEP-1 and CT
elements in various genes.
NSEP-1 Binds to Related CT Promoter Elements in a
'Generic' Sequence-specific Manner
We first asked to which CT elements does NSEP-1 bind? Several
end labeled oligonucleotides with different CT element sequences
or unrelated but GC rich sequences were prepared and binding
of NSEP-1 was assayed by the gel retardation assay (12). It can
be seen that NSEP-1 bound the c-myc and EGFR nuclease-
sensitive elements with similar avidities (Figure 2). One element
with more divergent CT elements, Neu-1, binds NSEP-1 less
well (Figure 2). Finally, a muscle-specific promoter element
(MCAT), a GC-rich region of the Neu proto-oncogene (Neu-2)
and an HLA gene promoter element (Y-box) do not bind NSEP-1
(Figure 2). The lack of binding to the Y-box element is of interest
since a previous report (15) claims that NSEP-1 will bind the
Figure. 1. Purification of recombinant NSEP- 1 protein from E. coli extracts by
heparin-agarose chromatography. Equivalent portions of the bacterial extract (load)
and the fractions eluting from the column were resolved by electrophoresis on
a 7.5% SDS-PAGE and stained with Coomassie Blue. Samples are as follows:
lane 1, molecular weight markers (Mws); lanes 2 & 3, W.T. NSEP-1 (load);
lane 4, flowthrough (F.T.); lane 5, wash; lane 6, 0.2 M Kcl fraction; lane 7,
0.6 M Kcl fraction; lane 8, 1.0 M Kcl fraction.
Y-box element. We will also show that Y-box sequences cannot
compete for the binding of the c-myc nuclease-sensitive element
to NSEP-1 (see below).
To gather further information on the binding specificity of
NSEP-1, we performed competition experiments. We utilized
oligonucleotide (labeled on the G-rich strand only) as a probe,
and various unlabeled oligonucleotides as competitors for binding
of NSEP-1 in an electrophoretic shift assay. As expected from
the direct binding assay, the c-myc NSE and the EGFR CT
element compete well for the binding of the labeled c-myc NSE
probe (Figure 3). A CT element (Ki-ras NSE) with homology
to these two elements competes, partially for the binding of the
c-myc NSE probe (Figure 3). Likewise, the poorly binding Neu-l
element competes less well for c-myc NSE binding (Figure 3).
Competition by oligonucleotides with mutated NSE sequences
compete less well than the wild-type c-myc NSE oligonucleotide
(Figure 3). Finally, two elements show very low levels of
competition even at a 500-fold excess of unlabeled competitor
(Neu-2 and Y-box) and two other elements (MCAT and a random
oligonucleotide with the same base composition as the c-myc NSE
but lacking any purine/ pyrimidine strand asymmetry) show no
competition (Figure 3). These results are summarized in Table 1.
An inspection of the CT elements that bind NSEP-1 indicate
that NSEP- 1 does have a limited specificity for CT-rich,
asymmetric DNA sequences (Table 1). The asymmetry is an
important aspect of this. For example, the random sequence
oligonucleotide has the same percentage of C and T bases but
has no purine/ pyrimidine strand asymmetry and
compete for DNA binding with the c-myc NSE oligonucleotide
(Table 1, Figure 3). The purine/pyrimidine strand asymmetry
Nucleic Acids Research, Vol. 20, No. 1113
+ - + -+
2 3 4 5 6 7 89 101 li
Figure. 2. NSEP-1 recognizes CT rich elements. Synthetic oligonucleotides
corresponding to the c-myc NSE, EGFR-NSE, and various other divergent CT
elements (Table 1) were tested for their ability to form complexes with recombinant
NSEP-1 protein. The assays were performed in the absence (odd numbered lanes)
or presence of extracts (even numbered lanes).
blading coo"titlon blnding coptitlioo
[i-ras MU-C -CCACG
T box Bottom
neu.-2 X Y-box
1 0' 2 0.
!EGFCJFK2rasneJ MCAT NQCAT ranranI
2 3 4 5 6 7 8 9 lOll1 12
Figure. 3. Specificity of recombinant NSEP-l-dsDNA interactions. Binding of
the recombinant NSEP- 1 protein to the c-myc NSE oligonucleotide was competed
for with increasing amounts (5, 20, 50, and 200-fold molar excess relative to
the labeled target) ofthe wild-type and mutant c-myc NSE, EGFR-NSE and other
divergent CT element oligonucleotides (Table 1).
is also important for H-DNA formation due to the nature of the
triplex base-pairing (Hoogsteen base pairing occurs between
pyrimidines in the third strand and purines in the Watson-Crick
'duplex'). The structure of all these CT-rich elements consists
of perfect or near perfect mirror image or direct repeats. The
c-myc NSE repeat is TCCCCACCC whereas the EGFR NSE
repeat is CCTCCTCCT. We have, in fact, made randomers of
T/C and have converted these to duplex T/C A/G. We have
found a significant proportion of these will bind to NSEP-1 (data
not shown). Therefore, we conclude that NSEP-1 will bind C-
T-rich duplexes with a high degree of purine/pyrimidine strand
1 2 3 4 5 6 7 8
Figures. 4A,B. NSEP-1 recognizes single-stranded DNA probes. Gel retardation
assay performed with recombinant NSEP-1 protein and various synthetic single-
strand oligonucleotides (Table 1). The assay was performed in the absence (last
lane in Figure 3A) or in the presence of extracts.
specificity we refer to as a generic sequence specificity in that
NSEP-1 binds all of one type of DNA species.
Single Strand Binding Properties of NSEP-1
In the formation of intramolecular DNA triplexes, there is a
portion of both DNA strands that is single stranded. The large
kink caused by formation of the intramolecular triplex creates
a steric hinderance to precise reformation of the duplex DNA
after the region of intramolecular triplex. In the case of the c-
myc NSE element, a large single strand region exists on both
the C-rich strand and the G-rich strand when H-DNA forms. In
Because B DNA is not likely to spontaneously form non-B DNA
conformers under intracellular conditions, these structures are
likely to be stabilized or induced by cellular factors. Therefore,
in preliminary experiments we examined NSEP-1 's ability to bind
-G *C triplex, the displaced C-rich strand is single-stranded.
&..-. -A.A AL
Nucleic Acids Research, Vol. 20, No. I
Wild Type and Truncated NSEP-1 Constructs
|.HR I I
Nucleic Acid ainding
ss DNA dsDNA
-o (AE a a 309 322!
(A a-a-201 322)
A o135 (.a-a-1 23)
]NH,(A a-a-1 48)
] 258(83 a-a
288 (A a-a-
AB2(-)a-a3l 23 3141 32)
M1 (Aaa-1l48 A158-322)
2M2(a-a-1 57 A158322)
ARNPCS (A a-a-31-48
Figure 6. Summary of NSEP-
properties. A schematic of NSEP-1 showing peptide motifs. Wild-type and deletion
mutants of the NSEP- 1 protein are shown. Each constructs name is followed by
deleted amino acids in parentheses. Nucleic acid binding is summarized at the
fight. + binding observed; - no binding observed; +/- weak binding observed.
mutants single and double-strand DNA binding
Figure 5. Specificity of recombinant NSEP-1-ssDNA interactions. Binding of
the recombinant NSEP-1 protein to the C-rich strand of the c-mycNSE was
competed for with increasing amounts (2, 5, 20 and 50-fold molar excess relative
to the labeled target) of oligonucleotides corresponding to the C-rich strands of
the c-myc and EGFR NSE regions, various CT elements and individual strands
corresponding to both strands of elements that did not bind in their duplex form
each of the single strands of the c-myc nuclease-sensitive element.
We found that NSEP-1 bound to the C-rich strand of the c-myc
NSE oligonucleotide but not to the G-rich strand (data not shown).
To examine the
interaction we performed binding experiments with various
single-stranded probes. Purified recombinant NSEP-1 binds to
the C-rich strands of the EGFR, Neu-1 and Neu-2 elements but
not to that of the Y-box element (Figure 4A). The higher band
shift complexes are not observed at lower NSEP- 1 concentrations
and we therefore believe that NSEP-1 binds to multiple sites on
these oligonucleotides. We will present evidence below to confirm
this conclusion. NSEP-1 also binds the C-rich strand of the Ki-
c-ras NSE oligonucleotide (data not shown and Table 1). We
next examined G-rich strands and both strands of elements that
did not bind in their duplex form (Figure 4B). We find no binding
for the EGFR, Ki-c-ras, Neu-1 G-rich strand or for either strand
of the MCAT or random oligonucleotides (Figure 4B). In
addition NSEP-1 does not bind the G-rich strand of the Neu-2
or either strands ofthe Y-box element (data not shown and Table
1). We conclude that NSEP-1 does not bind single strand DNA
sequence specificity of this single-strand
indiscriminately but rather has a marked preference for CT-rich
DNA. To further define this single-strand binding specificity,
we have performed competition experiments with various
competitors. As in the direct binding assay, only CT-rich strands
compete for the binding of purified, recombinant NSEP- 1
(Figure 5). The c-myc, Ki-c-ras, and EGFR nuclease-sensitive
elements C-rich strands compete equally well for c-myc NSE
binding to NSEP-1 (Figure SA). The Neu-1, CT randomer, and
two monomer c-myc NSE C-rich strands compete approximately
1/5 as well as the above group (Figure 5). The monomer
embedded in the 25 bp random oligonucleotide (Table 1). This
demonstrates that the monomer sequences contain
information for single strand DNA binding. In experiments with
direct binding of NSEP-1 to the monomer sequences, little or
no multimer is observed (data not shown). We believe this
indicates that the multimer is formed by multiple NSEP-l proteins
binding to multiple sites on the nucleic acid and not by protein-
protein association of two NSEP-1 molecules. The Neu-2 C-rich
strand shows a 50-fold lower level of competition than the
parental c-myc NSE C-rich strand (Figure 4B). Finally, the
MCAT bottom strand, the Neu-2 top strand, either strands of
the Y-box and random oligonucleotides do not compete for
binding at > 200-fold molar excess (Figure 5; Table 1). We have
also used poly(dT) and poly(dC) as competitors and do not
observe competition with either of these DNAs, even at a 50-fold
molar excess (data not shown). We therefore conclude that
binding protein as well as having duplex DNA binding properties.
NSEP-1 binds single-strand sequence elements that are CT-rich
but not pure dT or dC (Table 1).
are two permutations of the c-myc NSE repeat
is a generic, sequence specific single-strand DNA
Defining the Single-strand and Double-strand Binding Motifs
of the Nuclease-Sensitive Element Binding Protein-1
NSEP-l binds both duplex and single-strand DNA in a 'generic'
sequence-specific manner. There are 7 regions of distinctive
(Figure 6). There are 3 regions rich in basic amino acids, a region
rich in acidic amino acids, a glutamine-rich region which overlaps
the acidic region, a proline/serine/threonine-rich region and an
the NSEP- 1
amino acid no.
Nucleic Acids Research, Vol. 20, No. 1
N ~~ ~~
Figure 7. Deletion analysis with bacterially expressed proteins for identification
of the double-strand and single-strand DNA binding domains. A. Wild-type and
mutated forms of NSEP-
T7 expression system as per the procedure of Studier et al. (9). The bacterially
expressed proteins were size fractionated on a 7.5% SDS-PAGE after boiling
in SDS-PAGE loading buffer. The fractionated proteins were visualized by
Coomassie Blue staining (top row). Samples are as follows: lane 11, vector control;
lane 10, wild-type NSEP-1; lane 9, 1.0 (A a.a. 309-322); lane 8, A 135 (A
a.a. 1-24); lane 7, A Smal (A a.a. 1-38); lane 6, A NH2 (A a.a. 1-48); lane
5, A 258 (A a.a. 1-67); lane 4, A 288 (A a.a. 1-75); lane 3, A Sall (A a.a.
201-322); lane 2, A RNPCS (A a.a. 1-48: A 70-75: A 309-322); lane 1,
A NH2-2 (A a.a. 1-57). Southwestern analysis was performed on lanes 1-11
following electrotransfer to nitrocellulose. The filter was probed with 32P-labeled
double-strand oligonucleotide corresponding to the c-myc NSE region (middle
row) and 32P-labeled single-strand oligonucleotide corresponding to the C-rich
strand of the c-myc NSE (bottom row). B. Southwestern analysis of bacterially
expressed proteins corresponding to lanes 2, 6, 10 and 11 in Figure A. Top row:
Coomassie Blue staining; Middle row: Southwestern analysis with ds c-myc NSE;
Bottom row: Southwestern analysis with C-rich strand of c-myc NSE.
were synthesized in bacteria with the use of the phage
octapeptide region with homology to the ribonucleoprotein
consensus sequence 1 (RNP-CS-1) motif. We wished to examine
the protein motif(s) that specify both the double-strand and single-
strand binding activities. To this end we prepared wild-type,
polymerase, T7 010 gene promoter system of Studier (8,9). The
NSEP-l constructs were expressed in the appropriatehost bacteria
and cell extracts were analyzed by SDS-PAGE and Southwestern
blotting. Ten constructs were analyzed using the c-myc NSE
duplex probe. Blots were strippedofduplex probe bydenaturation
in guanidine-HCl and re-probed with the C-rich strand of the
c-myc NSE. In these experiments, the films are purposely
overexposed to ensure that binding and/or the lack ofbinding
Southwestern blots, we detect minor, proteolytic cleavage
products generated in vivo. The overexposure of the autoradio-
graph, however, exaggerates their levels.
The results of these Southwestern blots define two sets of
overlapping yet distinct peptide motifs that are necessary for
duplex and single-strand binding (Figure 6). We find that the
Pro/Ser/Thr-rich region, the acidic region, the gln-rich region
and the basic region 3 are not necessary for either double- or
single-strand DNA binding (Figures 6 and 7A). On the other
hand, basic-1, basic-2 and the intervening amino acids are
apparently necessary for both duplex and single-strand binding
(Figures 6 and 7). A few amino acids may be removed from
the basic-I region and still maintain duplex binding but not single-
strand binding (see NH2-2 construct; Figures 6 and 7). This
small deletion of basic-i allows double strand binding only if
there is no deletion of COOH amino acids (Figure 6; compare
the NH2-2 construct with the M2 construct). Perhaps the basic-3
region near the COOH-terminus can replace a basic-I region
function such as stabilizing NSEP-1 phosphate interactions. These
differences not withstanding, the key peptide region in single-
strand binding is the RNP-CS-1 region. When the RNP-CS-1
region is deleted, duplex NSE binding is maintained but single-
strand DNA binding is not observed (Figures 6 and 7A,B). The
minimal peptide region sufficient for both duplex and single-
strand binding (construct Ml) includes the basic-1, RNP-CS-1,
amino acids 75-140 and the basic-2 region (Figures 6 and 7).
DNA binding properties ofnuclease-sensitive element-binding
We have characterized a CT element or nuclease-sensitive
element-binding protein (NSEP-1) utilizing recombinant wild-
type protein as well as mutant proteins. We have found that
NSEP-1 binds to a family ofCT-rich elements having the property
of strong purine/pyrimidine strand asymmetry. The binding of
NSEP-1 is therefore 'specific' in that it binds a 'type' of DNA
element to the exclusion of other sequences. We term this DNA
binding specificity 'generic' sequence specificity.
NSEP-l also binds the C-rich strand ofCT or nuclease-sensitive
elements. As in the case of double-strand binding, related
sequence elements that are rich in CT show specific binding but
not the G-rich strand ofthese elements or other non-C-rich single-
strand oligonucleotides. One ofthe single-strand oligonucleotides
tested, MCAT, has been shown to bind to other transcription
factors (16). We therefore conclude that there are at least two
Substituting A for C or T bases reduces NSEP-l binding to these
elements. Poly(dT) and poly(dC) do not compete for the binding
of the c-myc nuclease-sensitive element, further defining the
nucleotide specificity to CT rich but not pure poly(dC) or
poly(dT) elements. Therefore, we conclude for the single-strand
binding, as we have for the double-strand binding that the single-
strand binding has generic sequence specificity.
Possible Relationships between the Binding Properties ofthe
Nuclease-Sensitive Element-Binding Protein and Alternate
DNA Topologies of CT-rich Sequence Elements
DNA sequence elements that are CT-rich and have a strong
purine/pyrimidine strand asymmetry can assume alternate DNA
conformations under certain conditions (6,17-20). One of these
alternate, non-B DNA conformation has been termed H-DNA.
H-DNA is a intramolecular triplex, singlestranded structure. The
116 Nucleic Acids Research, Vol. 20, No. I
purine-rich strand ofthe Watson-Crick base-pairs can Hoogsteen
base pair with a portion ofthe pyrimidine-rich strand thus forming
a triplex. The triplex forms in vitro in supercoiled DNA at low
pH. The dependency on low pH is due to the need for a proton
in the formation of the C+G Hoogsteen base pair. This can be
overcome by high levels of supercoiling. Alternatively, these
same elements can form a G *G C triplex at neutral pH in the
presence of Mg+2. We have previously predicted that the
proteins which bind these CT rich promoter elements that have
a high degree ofpurine/pyrimidine strand asymmetry can either
induce alternate, non-B DNA structures or stabilize them. That
nuclease-sensitive elements should be considered as in vivo
structures comes from several pieces of information. 1. Some
of these elements are DNase I sensitive in isolated nuclei
indicating that these elements exclude nucleosomes. 2. DNase
I sensitivity can be eliminated by changes in gene activity. 3.
In the case of the c-myc NSE element, the factors that bind this
element disappear coincidentally with the change in nuclease
sensitivity. Taken together, we hypothesize that the factors that
bind these nuclease-sensitive, CT-rich structures help form non-B
DNA structure(s) in vivo. As stated above, the c-myc nuclease-
sensitive element's, DNase I sensitive structure is correlated with
c-myc activity and the nuclease-resistant structure is found to
correlate with c-myc down-regulation (12).
We believe that the single-strand DNA binding properties of
NSEP-1 are consistent with the c-myc nuclease-sensitive element's
ability to form a non-B DNA conformer. We envision that an
NSEP-1 molecule can bind duplex NSE and 'capture' or bind
the C-rich strand ofthe NSE as it 'breathes' or is denatured under
torsional stress. The non-B DNA conformer (whichever it may
be) may then form spontaneously or other factors may bind and
further stabilize/induce triplex formation.
Ifthe senario we discuss above occurs, then one may ask why
elements with poor H-DNA forming potential bind more
vigorously to NSEP-1 than elements that are more likely to form
H-DNA (for example the EGFR NSE has a more stable H-DNA
structure than the c-myc NSE). We believe there are several
reasons to explain these data. First, the ability to form an H-
DNA structure under conditions of low pH in vitro may have
no bearing on in vivo H-DNA formation, especially if sequence-
specific DNA binding proteins are necessary for H-DNA
formation or stability. Second, one may speculate that DNA
elements with poor H-DNA forming potential would need a
greater affinity for DNA binding proteins that would induce or
stabilize the H-DNA structure because of the poor energetics.
On the otherhand, elements that could form H-DNA more readily
might not need the stabilizing effects of a DNA binding protein
and therefore would need to bind these proteins only weakly or
not at all. In any case, we may find that there is a family ofCT
element binding proteins with overlapping affinities for these
elements, and that H-DNA formation at any one element is more
a matter of cell physiology than chemical stability.
Cooney,M., Czemuszewicz,G., Postel,E.H., Flint,S.J. and Hogan,M.E.
(1988) Science, 241, 456-459.
Kohwi,Y. and Kohwi-Shigematsu, T. (1988) Proc. Natl. Acad. Sci. USA,
Johnson,A.C., Junno,Y. and Merlino,G.T. (1988) Mol. Cell. Biol. 8,
Larsen,A. and Weintraub,H. (1982) CELL, 29, 609-622.
McGhee,J.D., Wood,W.T., Dolan,M., Engel,J. and Felsenfield, G. (1981)
CELL, 27, 45-55.
Wu,C., Bingham,P.M., Livak,K.J., Holmgren,R. and Elgin, S.C.R. (1979)
CELL 16, 797-806.
Kinniburgh,A.J. (1989) Nucl. Acids. Res., 17, 7771-7778.
Kolluri,R.V. and Kinniburgh,A.J. (1991) Nucl. Acids. Res. (In Press).
Rosenberg,A.H., Lade,B.N., Chui,D.-S., Lin,S.-W., Bunn, J.J. and
Studier,F.W. (1987) Gene, 56, 125-135.
Studier,F.W., Rosenberg,A.H., Dunn,J.J. and Dubendorff,J.W. (1990)
Methods in Enzymology, 185, 60-89.
Landschulz,W.H., Johnson,P.F. and Mcknight,S.L. (1989) Science, 243,
Lichtsteiner,S., Wuarin,J. and Schibler,U. (1987) CELL, 51, 963-973.
Davis,T.L., Firulli,A.B. and Kinniburgh,A.J. (1989) Proc. Natl. Acad.
Sci. USA, 86, 9682-9686.
Singh,H., LeBowitz,J.H., Baldwin,A.S. and Sharp,P. (1988) CELL, 52,
Sakura,H., Maekawa,T., Imamoto,F., Yasuda,K. and Ishii,S. (1988) Gene,
Schwartz,B.D. (1988) Proc. Natl. Acad. Sci. USA, 85, 7322-7326.
Santoro,I.M., Yi,T.M. and Walsh,K. (1991) Mol. Cell. Biol., 11,
Lyamichev,V.I., Mirkin,S.M. and Frank-Kamenetskii,M.D. (1985) J.
Biomol. Struc. Dyn., 3, 327-338.
Hanvey,J., Schimizu,M. and Wells,R.D. (1988) Proc. Nati. Acad. Sci.
USA, 85, 6292-6296.
Johnston,B.H. (1988) Science 241, 1791-1795.
Htun,H. and Dahlberg,J.E. (1989) Science 243, 1571-1575.
We thank Drs. J. Ross and W. Held for many helpful comments
during the course of this work. We also thank Dr. Simon
Williams for materials and advice on their use. We thank Donna
Rypinski for technical assistance and Nancy Frame for secretarial
assistance. This work was supported by a grant from the NIH,
CA 43661, and a Cardiovascular Discovery Award from Glaxo,