ABSTRACT
The use of complementary RNA or DNA sequences to selectively interfere with the
utilization of mRNA of a target gene is an attractive therapeutic strategy. Two
well-studied targets for oligonucleotide therapy are the c-
myc
and c-
myb
proto-oncogenes. It has been reported that sequences which contain four contiguous Gs can elicit a non-antisense response, due to the formation of a homotetrameric G
quartet structure. Therefore, it was of interest to determine whether anti-c-
myc
and anti-c-
myb
phosphorothioate DNAs including tetraguanylate form higher order structures
under physiologically relevant salt conditions and temperature. First, the
identity of the higher order structure was established and was found to be a
tetraplex. Employing intracellular (high K
+
), extracellular (low K
+
) and normal saline (no K
+
) salt mixtures, native gel electrophoresis revealed no tetraplex formation at
37
o
C, the physiologically relevant temperature. On the other hand, tetraplex
structure formation was observed at 4 and 23
o
C. Hence, the potential for these sequences to form tetraplex structures at
lower temperatures may not be relevant for their activity in cells and animals
at physiological temperature.
The potential of oligonucleotides to achieve inhibition of gene expression
selectively is being studied intensively (
1
,
2
). To be active as an antisense agent, an oligonucleotide has to enter and
travel within a cell to reach RNA in the cytoplasm or nucleus and form a
specific duplex with the target mRNA. In early experiments, this laboratory
reported that targeting the human c-
myc
proto-oncogene with complementary DNA phosphodiesters inhibited proliferation of
PMA-stimulated human peripheral blood lymphocytes (
3
) and human promyelocytic leukemia cells (
4
). The same sequence, as a DNA methylphosphonate, inhibited MYC protein levels
in peripheral and splenic lymphocytes of an E[mu]-
myc
transgenic mouse model for Burkitt's lymphoma (
5
). Similarly, the anti-c-
myc
sequence, as a DNA phosphorothioate, inhibited arterial restenosis by smooth
muscle cells in a porcine model (
6
). In parallel studies, tumorigenesis by human leukemia cells (
7
) and human melanoma cells in SCID mice (
8
) was inhibited by targeting the c-
myb
proto-oncogene with complementary DNA phosphorothioates. Recently, this
laboratory found that targeting murine c-
myc
with complementary DNA phosphorothioates in the E[mu]-
myc
mouse lymphoma model led to a prophylactic effect against lymphoma development
(
9
).
The antisense oligonucleotide sequences used in the experiments above (Table
1
) contained contiguous tetraguanosine tracts, which have been found capable of
sequence-independent antiproliferative effects on cultured human epithelial cells (
10
). Similarly, it was observed that high concentrations (30-50 [mu]M) of antisense phosphorothioate DNAs directed against c-
myc
and c-
myb
showed non-antisense activity in primary cultures of smooth muscle cells
in vitro
and in arteries
ex vivo
(
11
). The antiproliferative effects were ascribed to the contiguous tetraguanylates
present in those antisense sequences. Rows of guanosines are necessary for
telomere function and are known to form a G quadruplex or tetraplex structure (
12
). Formation and dissociation rates of G tetraplex structures are deceptively
slow and vary widely depending upon the 5' and 3' flanking sequences (
13
-
15
). For example, full reassociation of dTGGGT phosphodiester single strands in 1
M KCl, 0.1 mM EDTA, 10 mM K
2
HPO
4
, pH 7, at 0oC, required 20 h (
13
). For dTTGGGGTT phosphorothioate single strands in phosphate-buffered saline (PBS) at 37oC, the fourth order association rate constant was 6 * 10
4
/M
3
s (
15
).
Thus it was suggested that since the c-
myc
and c-
myb
antisense sequences contain G4 tracts, they could form tetraplex structures,
interfering with base pairing with the target mRNA, exclusively imparting their
therapeutic effect through a non-antisense mechanism (
11
). On the other hand, it has not been demonstrated that these sequences act via
a non-antisense, G quartet mechanism at the low concentrations (0.1-10 [mu]M) used in many other systems where these sequences were found
to be effective (
3
-
9
). Furthermore, it has been found that tetraguanylates do not always give rise
to a non-antisense effect (
16
). Therefore, it was of interest to explore the formation of higher order
structures by the c-
myc
and c-
myb
phosphorothioate DNA antisense sequences
in vitro
under intracellular and extracellular salt conditions at a variety of
temperatures, especially at the physiologically relevant 37oC, by native gel electrophoresis. At the onset the identity of the
tetraplex band was unambiguously established. Although these tetraguanylate-containing sequences form tetraplexes and another structure with the
mobility of a duplex in certain salts at 4 and 23oC, the tetraplex structures were not observed at 37oC under physiologically relevant conditions.
Sequences synthesized (Table
1
) included human (HMYC) (
3
) and mouse (MMYC) (
9
) c-
myc
antisense oligonucleotides against codons 1-5, reversed (MREV) and scrambled (MSCR) (
9
) mouse antisense c-
myc
sequences, a human c-
myb
antisense oligonucleotide (
8
) against codons 3-7 (HMYB) and a sequence (QG14) with two tetraguanylate tracts previously
shown to form a tetraplex structure (
17
), used as a positive control sequence. A sequence complementary to the MSCR
sequence was synthesized to allow formation of a duplex control. Also, MYC10, a
3' decamer fragment of MMYC, was synthesized for heterogeneous tetraplex
formation with MMYC.
Table 1
Oligodeoxynucleotides were synthesized by standard phosphoramidite coupling to
produce normal phosphodiesters (
18
) or phosphorothioates (
19
). When desired, oligonucleotides were 5'-radiolabeled with [[gamma]-
33
P]ATP (2000 Ci/mmol; no. NEG-602H; DuPont/New England Nuclear, Boston, MA) using T4 polynucleotide
kinase (no. M410A; Promega, Madison, WI) (
4
). Full-length oligonucleotides were purified from failure sequences by
electrophoresis on denaturing 20% polyacrylamide gels in 89 mM Tris-borate, pH 8.3, 1 mM EDTA (TBE) plus 7.0 M urea, then extracted by
crushing and soaking. The purified DNAs were desalted using C
18
Sep-Pak cartridges (no. 51910; Waters, Milford, MA). Oligonucleotides were
quantitated by absorbance at 260 nm as described (
20
). Oligonucleotide purities were re-analyzed, following purification and desalting, by 20% polyacrylamide-7 M urea denaturing gel electrophoresis. Oligonucleotide bands were
visualized with Stains-All (no. 19171; BioRad, Richmond, CA) following the standard protocol
supplied by the manufacturer. A calibration experiment for linearity of
tetraplex staining with 0.1, 0.3, 0.5, 0.7 and 0.9 nmol MMYC pre-incubated in solution A and analyzed at 4oC displayed linear densitometry results with a correlation
coefficient of 0.964. Gels with [5'-
33
P]oligonucleotides were quantitated with a PhosphorImager 445 SI (Molecular
Dynamics, Palo Alto, CA).
Samples were incubated in three different salt mixtures: intracellular salt, 150
mM KCl, 10 mM Na
2
HPO
4
, 1 mM EDTA, pH 7.0 (solution A); extracellular salt, 150 mM NaCl, 10 mM K
2
HPO
4
, 1 mM EDTA, pH 7.0 (solution B); control lithium, 150 mM LiCl, 10 mM Li
2
HPO
4
, 1 mM EDTA, pH 7.0 (solution L); normal saline, 137 mM NaCl, 1.5 mM K
2
HPO
4
and 8 mM Na
2
HPO
4,
pH 7.0 (solution S). In each case, the 1 nmol samples (2.0 [mu]l at 0.5 mM, tightly sealed to prevent evaporation) were denatured at 90oC for 4 min, then incubated in the given salt solution for 24 h, unless
otherwise specified, at 4, 23 or 37oC. Formation of higher order structures was studied by non-denaturing electrophoresis on 20% polyacrylamide gels, 10 * 8 * 0.1 cm, in TBE in a Protean II electrophoresis chamber
(BioRad, Hercules, CA). In this apparatus, the glass plates containing the gel
are immersed on both sides in running buffer, which serves as a heat sink to
dissipate the heat generated by electrophoresis and equilibrate the temperature
of the gel with that of the buffer. Electrophoretic analyses were carried out
in a cold room at 4oC for 12 h or on a laboratory bench at 23oC for 8 h or in a warm room at 37oC for 5 h, because oligonucleotide mobility increases with
temperature. Native gels were also electrophoresed at 50oC in a water bath for 3 h after incubating the samples at 4oC for 48 h. The gels and the running buffers were pre-run and pre-equilibrated at the corresponding temperature for at least
4 h, until current, voltage and temperature were equilibrated, before the
samples were loaded onto the gels. All gels were run with samples at a constant
electric field of 6 V/cm, at a current of 20 mA, dissipating power of 1.0 W
(0.24 cal/s) for an 8 cm gel. At this heating rate, the 800 ml of upper and
lower reservoir buffer would be predicted to increase in temperature by 0.0003oC/s, or 1.1oC/h, unless cooled by the surrounding air. The temperatures of the
electrophoresis buffers, which are presumed to be in equilibrium with the
immersed gels, were monitored during electrophoretic analyses and were never
found to vary by >1oC from the 4, 23 or 37oC temperatures of their environments.
To demonstrate tetraplex formation by the anti-c-
myc
phosphorothioate pentadecamer, MMYC was mixed in 1:1 molar ratio with a
decamer, MYC10, lacking the five 5' residues, then heat denatured at 90oC for 4 min and annealed at 4oC for 48 h in the presence of 150 mM KCl, 10 mM Na
2
HPO
4
, 1 mM EDTA, pH 7.0 (solution A). The mixture was analyzed by electrophoresis on
a 20% polyacrylamide native gel at 4oC.
Alzet micro-osmotic pumps (no. 2002; Alza, Palo Alto, CA) are utilized in this
laboratory as a controlled slow release device for the delivery of
oligonucleotides in transgenic mice (
9
). The pumps were removed from subjects under anesthesia 2 weeks after s.c.
implantation. Immediately after removal, the pumps were placed in a 37oC room. Residual oligonucleotides were taken out of the pumps and the
concentrations measured by UV absorption. The concentration of oligonucleotides
remaining inside the pumps after 2 weeks varied from 4.5 to 5.0 mM. Samples
were analyzed on native polyacrylamide gels at 37oC. Aliquots of the collected samples were also incubated at 4oC for 24 h and analyzed on native polyacrylamide gels at 4oC.
Aliquots of oligonucleotides collected from micro-osmotic pumps were incubated in 50% heat-inactivated fetal bovine serum, 50% PBS for 1, 4 or 24 h at 37oC. The samples were then analyzed by electrophoresis under
native conditions at 37oC. These samples were also incubated at 4oC for 24 h, then incubated for 1, 4 and 24 h at 37oC and subjected to electrophoresis at 37oC.
Oligonucleotides were incubated in the three salt mixtures (A, B and S) at 4oC for 7 days to maximize higher order structure formation and then analyzed
by electrophoresis at 4, 23, 37 and 50oC, to observe any transition of structure from the complexes formed after 7
days incubation.
To confirm that the anti-c-
myc
sequence MMYC forms a tetraplex structure, MMYC was mixed with MYC10, which
lacks the first five 5' nucleotides of MMYC but includes the tetraguanylate tract, and incubated
for 48 h at 4oC in solution A, mimicking intracellular salt. The interstrand tetraplex
structures if formed would yield five different combinations, (MMYC)
4
, (MMYC)
3
(MYC10)
1
, (MMYC)
2
(MYC10)
2
, (MMYC)
1
(MYC10)
3
and (MYC10)
4
, resulting in five distinct bands. Similar patterns have been observed before
with other tetraguanylate sequences, such as QG14 (
17
,
21
). The five bands predicted for a tetraplex were observed upon native gel
electrophoresis at 4oC (Fig.
1
). This result is consistent with the model that the slowest moving band formed
by the anti-c-
myc
pentadecamer MMYC is indeed due to a tetraplex structure.
Formation of higher order structures, such as tetraplexes, is often studied
using high salt conditions (
12
), which may not be relevant to salt conditions
in vivo
. One tetraplex hypothesis holds that ion fluctuations (especially K
+
) play a role in the control of nucleic acid-dependent physiological functions via their effect on the structure of G-rich sequences (
22
). Therefore, physiologically relevant salt conditions were employed in order to
determine whether these particular DNA sequences form higher order structures
under salt conditions encountered
in vivo
. The sequences may exist in three forms: tetraplex structure (Tx), which
migrates on a native gel similarly to a tetraplex formed by QG14 (
17
); an intermediate structure (Dx), which migrates similarly to a control 15 bp
duplex of the MSCR scrambled control and a complement of that sequence; a
single strand (SS), which migrates similarly to MSCR, which has not displayed
any higher order structures under any of the temperature and salt conditions
used. We found that the tetraguanylate sequences QG14, MMYC, MREV, HMYC and
HMYB all formed higher order structures (Tx) after 24 h incubation at 4oC under solution conditions mimicking intracellular salt composition
(solution A) and native gel electrophoresis at 4oC (Fig.
2
). The same tetraplexes were also observed in extracellular salt (solution B)
(not shown) and normal saline (solution S) (not shown) after 24 h incubation at
4oC. These results establish that these sequences have the ability to form
tetraplex structures, as would be expected from their sequences. Next, the
incubation and electrophoretic temperatures were varied to examine the role of
temperature in higher order structure formation.
After incubation of the candidate sequences in solution A at 37oC for 24 h, native gel electrophoresis at 37oC revealed only the intermediate structure (Dx) and single strand (SS)
(Fig.
3
). The same pattern was observed after incubation at 37oC in extracellular salt (solution B) (not shown) and normal saline (solution S) (not shown).
To extend the dynamic range of detection below that possible with Stains-All detection, the 37oC incubation in solution A was repeated with [5'-
33
P]MMYC and [5'-
33
P]QG14. In each case, labeled oligonucleotide was added to unlabeled
oligonucleotide in order to maintain an incubation concentration of 0.5 mM.
Parallel incubations were carried out in 0.15 M LiCl (solution L), because Li
+
disfavors tetraplex formation (
12
).
Alzet micro-osmotic pumps are utilized in this laboratory for long-term delivery of oligonucleotides to E[mu]-
myc
transgenic mice (
9
). These micropumps are implanted s.c. and are designed to deliver
oligonucleotides at a steady rate for 14 days. After the end of one such
experimental period the micropumps were removed from anesthetized mice, immediately put into a 37oC warm room and residual oligonucleotides were collected by syringe. In
this case, the micropumps mimicked the situation of incubating the
oligonucleotides at 37oC for 14 days in PBS at 5 mM concentration. Aliquots of the
oligonucleotides were analyzed on a 20% polyacrylamide-7 M urea denaturing gel and found to be essentially intact. These
samples, when analyzed on a native gel at 37oC, did not show any evidence of tetraplex formation (Fig.
7
). The results were identical to the situation where the samples were incubated
and analyzed at 37oC. However, when samples of these collected oligonucleotides were incubated
at 4oC for 24 h and the gel run at 4oC, there was evidence of substantial tetraplex formation (data not
shown). This result demonstrates that although the MMYC anti-c-
myc
sequence has a propensity for tetraplex formation at lower temperatures,
tetraplexes were not observed at 37oC, even after 2 weeks in micropumps in mice. This result is significant
because the anti-c-
myc
sequence may be free to act as an antisense agent if it is not involved in
tetraplex structure, but is primarily single stranded. These
in vitro
situations may not always be comparable with
in vivo
conditions, because proteins and other factors present in serum may facilitate
higher order structure formation, as it is known that proteins often mediate
the formation of structures by nucleic acids.
In mammalian cell culture or
in vivo
the temperature is usually near 37oC and the oligonucleotides are exposed to a mixture of proteins that might
mediate or stabilize a preformed higher order structure. It would be difficult
to test this hypothesis
in vivo
, as extraction of the structured oligonucleotides from whole cells or tissues
without denaturation, i.e. disruption of the higher order structure(s), may not
be possible. Hence, to mimic a similar condition, we incubated the
oligonucleotides in the presence of serum. The oligonucleotides collected from
the micropumps were incubated in 50% fetal bovine serum, 50% solution S at 37oC for different time periods and then analyzed on native polyacrylamide
gels at 37oC. Again, no tetraplex formation was detected (data not shown). Identical
bands were observed with or without incubation in serum. Therefore, the serum
did not facilitate tetraplex formation. However, in this case the
oligonucleotides were not initially in tetraplex form; it was also necessary to
examine the case where the tetraplex structure was preformed and then incubated
with serum at physiological temperature. Oligonucleotides were first incubated
in high K
+
intracellular salt (solution A) at 4oC for >48 h to allow substantial tetraplex formation. This incubation was
followed by a second incubation at 4oC in 10% fetal bovine serum in RPMI medium. The samples were then analyzed
by electrophoresis at either 4 or 37oC. Again, the results were similar to those in the absence of serum (Figs
2
and
6
). The samples analyzed at 4oC showed almost complete tetraplex formation, but those analyzed at 37oC did not show any tetraplex, implying that serum proteins under the
conditions employed did not stabilize tetraplex formation at 37oC.
In order to determine the stability of the intermediate structure with the
mobility of a pentadecamer duplex formed by all the tetraguanylate-containing sequences, the native gels were also analyzed at 50oC, after preforming the tetraplex structure by incubating in high K
+
salt (solution A) for 48 h at 4oC. We found that even at 50oC the tetraplex structure redistributed to the intermediate structure,
so that both the single-stranded and the intermediate structure co-existed at 50oC, with a predominance of the latter (data not shown).
Short complementary oligonucleotide sequences targeting c-
myc
and c-
myb
mRNA have been employed as gene-specific inhibitory agents both in cell culture and in animal experiments
(
3
-
9
). The sequences used for this purpose contain four guanylates in a row, which
have the potential to form tetraplex or higher order structures (
12
). In this work we investigated tetraplex formation by these sequences and
control sequences containing tetraguanylates at physiologically relevant
temperatures and salt concentrations. Most of the previous studies of tetraplex
formation were done with telomeric repeat sequences from
Oxytricha
to human (
23
,
24
). Evidence exists that tetraplex formation by a tetraguanylate-containing phosphorothioate sequence, selected by a combinatorial
approach, is a potent inhibitor of HIV envelope-mediated cell fusion (
25
).
Factors that influence the equilibrium of formation of G tetraplex structures
are different ions, DNA and ion concentrations, length of the G-rich strand, Watson-Crick complementarity and sequence differences (
14
,
22
). The findings from telomeric repeat sequences cannot necessarily be applied to
the sequences we investigated, as the antisense oligomers have only one
tetraguanylate in a heterogeneous sequence context and sequence plays a major
role in tetraplex structure formation (
14
,
22
). In particular, tetraguanylates at the 5'-end of octadecamers were reported to favor higher order structures,
while tetraguanylates at the 3'-end disfavored higher order structures (
14
).
The tetraplex structures in our system are presumed to be formed by four
separate strands, thereby forming an interstrand G quartet structure, as all
the sequences tested contain only one tetraguanylate tract. To test the
molecularity of the complex, two oligonucleotide sequences of different
lengths, the anti-c-
myc
pentadecamer MMYC and the homologous decamer MYC10, both containing the
essential tetraguanylate tract, were incubated together at 4oC then analyzed by electrophoresis at 4oC. Five bands were observed, as predicted, due to all the combinations
possible for interstrand tetraplex formation. Such a strategy has been utilized
previously to establish the identity of a tetraplex band (
17
,
21
).
In the present study, native gel electrophoresis revealed that all the
tetraguanylate-containing sequences studied above are capable of forming tetraplex
structures, depending on the temperature. The intracellular (high K
+
), extracellular (low K
+
) and normal saline (absent K
+
) conditions used for the experiments all permitted tetraplex formation when the
oligonucleotides (0.5 mM) were incubated at 4oC for 24 h and the gels were run at 4oC. When the same experiments were carried out at 23oC, three structures were observed: the tetraplex, the
intermediate structure (Dx), which moves with a 15 nt control duplex, and a
single strand. The intermediate structure with mobility similar to a control
duplex has been observed in earlier tetraplex studies (
17
), but the nature of these structures has not yet been elucidated. When the
incubations and electrophoresis were carried out at 37oC, only two structures were seen for the antisense sequences, the
intermediate (Dx) and the single strand. Under the same conditions, the strong
tetraplex control sequence QG14 displayed 56% tetraplex. Hence, higher
temperature seems to disfavor the formation of tetraplex. As a negative
control, pre-incubation at 37oC in the presence of Li
+
, without Na
+
or K
+
, revealed no MMYC tetraplex bands and only 2.5% QG14 tetraplex bands after
electrophoresis.
Further insight into tetraplex dependence upon temperature was obtained when the
samples were incubated at 4oC but analyzed by gel electrophoresis at either 23 or 37oC. In these experiments, similar band patterns were observed as were
seen after incubation at 23 or 37oC respectively, indicating that the tetraplex structures rapidly
redistributed to other structures when the temperature was raised, as shown by
the clear presence of the relevant bands without smeared transitional forms.
It may be argued that the complex array of molecules present inside cells and in
serum may stabilize a preformed higher order structure or may mediate formation
of it, as evidence exists for a tetraplex binding protein observed to have a
stabilizing effect on tetraplex structure (
26
). To test this hypothesis, the oligonucleotides would have to be retrieved from
tissues without denaturation. As an alternative, we incubated the
oligonucleotide samples in serum to expose them to serum proteins, to determine
whether serum proteins stabilize a preformed higher order structure or mediate
the formation of higher order structures. Under these conditions higher order
structures were neither stabilized nor mediated by serum proteins and other
factors present, though it was found previously that in telomeric repeats,
tetraplex structures are clearly involved in specific interactions with protein
components and proteins promoting the formation of tetraplex structures (
27
,
28
).
These findings are significant with respect to the question of the availability
of tetraguanylate-containing antisense sequences in a single-stranded form. The electrophoretic analyses demonstrated that at 37oC at physiological ionic strength, under intracellular or
extracellular ionic conditions, there were no detectable tetraplex structures.
Even in the case where tetraplex structures already existed, due to
equilibration at lower temperatures, they quickly dissociated at physiological
temperature to lower order structures. One would therefore expect the same
situation to exist under clinical conditions, comparable with the MMYC samples
removed from a s.c. micropump after 14 days in mice, when such oligonucleotides
are administered for genetic therapy or gene-specific inhibition experiments.
The structures present at physiological and higher temperatures include both the
intermediate Dx and the single strand. Since the Dx structure is quite stable,
even at higher temperatures, the possibility exists for non-antisense effects by the Dx fraction, in addition to antisense or non-antisense effects by the single strand fraction. The exact nature of
the structure Dx needs to be elucidated in further detail, in order to help
elucidate non-antisense effects observed with these sequences at high concentrations.
We thank Dr Michael Kligshteyn for synthesizing the oligonucleotides and Dr
Jason Rife for initiating the tetraplex analyses in our laboratory. This work
was supported by NIH grant CA42960 to E.W.
*To whom correspondence should be addressed. Tel: +1 215 955 4578; Fax: +1 215
955 4580; Email: ewick@lac.jci.tju.edu
+
Present address: Department of Molecular Biophysics and Biochemistry, Yale
University School of Medicine, New Haven, CT 06520, USA
Description
Label
Sequence
Mouse c-
myc
(antisense)
MMYC
5'-dCACGTTGA
Mouse c-
myc
(reverse)
MREV
5'-dTAC
Mouse c-
myc
(scrambled)
MSCR
5'-dCTGCTGAGAGTCGAG
Human c-
myc
(antisense)
HMYC
5'-dAACGTTGA
Human c-
myb
(antisense)
HMYB
5'-dGTGCC
Decamer c-
myc
(antisense)
MYC10
5'-dTGA
Decamer c-
myc
(scrambled)
SCR10
5'-dCTGCTGAGAG
Tetraplex control
a
QG14
5'-dT
REFERENCES
Return