Oligonucleotide N3
' ->
P5
' phosphoramidates as antisense agents
Oligonucleotide N3 ' -> P5 ' phosphoramidates as antisense agents
Sergei
Gryaznov*
,
Tomasz
Skorski
1
,
Carla
Cucco
1
,
Malgorzata
Nieborowska-Skorska
1
,
Choi Ying
Chiu
,
David
Lloyd
,
Jer-Kang
Chen
,
Maria
Koziolkiewicz
2
and
Bruno
Calabretta
1
Lynx Therapeutics, Inc., 3832 Bay Center Place,
Hayward
, CA 94545,
USA
,
1
Thomas Jefferson University, 233 South 10th Street,
Philadelphia
, PA 19107,
USA
and
2
Centre of Molecular and Macromolecular Studies, Sienkiewicza 112,
Lodz
90-363,
Poland
Received December 12, 1995;
Revised and Accepted February 28, 1996
ABSTRACT
Uniformly modified oligonucleotide N3
' ->
P5
'
phosphoramidates, where every 3
'
-oxygen is replaced by a 3
'
-amino group, were synthesized. These compounds have very high affinity to
single-stranded RNAs and thus have potential utility as antisense agents. As was
shown in this study, the oligonucleotide phosphoramidates are resistant to
digestion with snake venom phosphodiesterase, to nuclease activity in a HeLa
cell nuclear extract, or to nuclease activity in 50% human plasma, where no
significant hydrolysis was observed after 8 h. These compounds were used in
various
in vitro
cellular systems as antisense compounds addressed to different targeted regions
of c-
myb
, c-
myc
and
bcr-abl
mRNAs. C-myb antisense phosphoramidates at 5
[mu]
M caused sequence and dose-dependent inhibition of HL-60 cell proliferation and a 75% reduction in c-
myb
protein and RNA levels, as determined by Western blot and RT-PCR analysis. Analogous results were observed for anti-
c-myc
phosphoramidates, where a complete cytostatic effect for HL-60 cells was observed at 1
[mu]
M concentration for fully complementary, but not for mismatched compounds, which
were indistinguishable from untreated controls. This was correlated with a 93%
reduction in c-
myc
protein level. Moreover, colony formation by the primary CML cells was also
inhibited 75-95% and up to 99% by anti-
c-myc
and anti-
bcr-abl
phosphoramidate oligonucleotides, respectively, in a sequence- and dose-dependent manner within a 0.5 nM-5
[mu]
M dose range. At these concentrations the colony-forming ability of normal bone marrow cells was not affected. The
presented
in vitro
data indicate that oligonucleotide N3
' ->
P5' phosphoramidates could be used as specific and efficient antisense agents.
INTRODUCTION
Synthetic oligonucleotides may become a new generation of rationally designed
therapeutic agents, offering potentially general applicability and high target
selectivity in action. Currently oligonucleotide phosphorothioates are the
leading candidates for the first generation of antisense compounds and several
of them are in phase I/II clinical trials (
1
-
3
). Unfortunately, phosphorothioates possess properties that may not always be
suitable for antisense and/or antigene agents: mainly, low (relative to natural
phosphodiesters) binding affinity to single-stranded (ss) RNA and especially to double-stranded (ds) DNA targets (
4
,
5
) and nucleotide sequence independent binding to a variety of intra- and extracellular proteins in mono- or multimeric forms (
6
-
8
). Thus, the search for better oligonucleotide analogues continues and new
classes of compounds have been introduced. Among these are 2'-modified oligomers (
9
-
11
), chimeric methylphosphonate-phosphodiesters (
12
), 5-propynyl-pyrimidine containing compounds (
13
), phosphodithioates (
14
) and peptide nucleic acids (
15
).
Recently, a new type of oligonucleotide analogue, uniformly modified N3' -> P5' phosphoramidates, where a 3'-amino group was substituted for the 3'-oxygen of the 2'-deoxyribose ring, was
synthesized (
16
). Physico-chemical studies of these compounds demonstrated that they form very
stable duplexes with complementary ssRNA, as well as triplexes with dsDNA
polypurine targets, and that they are resistant to nuclease catalyzed
hydrolysis (
16
-
18
). These features of the oligonucleotide N3' -> P5' phosphoramidates prompted us to evaluate these compounds as
potential antisense agents in different
in vitro
cell culture systems well characterized with other oligonucleotide analogs,
using
bcr-abl
, c-
myc
and c-
myb
proteins as model targets. These three systems were chosen and studied to
demonstrate the generality of antisense activity of the oligonucleotide
phosphoramidates not in one but in several
in vitro
cell cultures. Results presented here demonstrate that N3' -> P5' phosphoramidates do act as sequence-specific and efficient antisense agents in the model
systems studied.
MATERIALS AND METHODS
General methods
Oligonucleotide N3' -> P5' phosphoramidates were synthesized on an ABI 394 automated
synthesizer according to the described procedure (
18
). Compounds were analyzed and purified by ion exchange (IE) HPLC, on a Dionex
DX300 chromatograph using Pharmacia MonoQ 5/5 or 10/10 columns, desalted by gel
filtration on Pharmacia NAP-5 or NAP-10 columns and lyophilized
in vacuo
from sterile water according to ref.
18
. Sequences of oligonucleotides synthesized and used in this study are presented
in Table
1
.
Enzymatic digestion experiments
Oligonucleotide
14
(Table
1
) 0.2 OD units with phosphodiester or N3' -> P5' phosphoramidate internucleoside linkages was treated with
0.02 U snake venom phosphodiesterase and 0.8 U alkaline phosphatase (both from
Sigma) in 0.2 ml 20 mM Tris-HCl buffer, pH 8.9. Reaction mixtures were analyzed by reversed phase
(RP) HPLC on a Hypersil ODS 4.6 * 200 mm column using a 0.5%/min gradient of acetonitrile in 0.1 M
triethylammonium acetate (TEAA) buffer, pH 7.0; time points were 0, 10 and 50
min, 2.5, 4.5 and 22 h for the phosphoramidate and only 0 and 10 min for the
phosphodiester oligomers, since the latter compound was completely digested by
that time.
To determine the stability of oligonucleotide phosphoramidates in human plasma
or HeLa cell nuclear extract, 0.1 OD units compounds
12
and
13
(Table
1
) were each 5'-
32
P-labeled with [[gamma]-
32
P]ATP, 2 pmol in 20 [mu]l 10 mM Tris-HCl, 10 mM MgCl
2
, 7 mM [beta]-mercaptoethanol, pH 8.5 buffer by 5 U T4 polynucleotide kinase
(Amersham) for 3 h at 37oC. Then ~0.01 OD units labeled compound was incubated in 10 [mu]l 50% human plasma or in 10 [mu]l 50% HeLa nuclear extract at 37oC. Reaction mixtures were analyzed by electrophoresis in
denaturing 20% polyacrylamide gel at time points 0, 1, 2, 4 and 8 h for human
plasma and 0, 0.5, 1 and 2 h for HeLa nuclear extract.
Experiments to determine the binding affinity of oligonucleotide phosphodiesters
and phosphoramidates to nuclear proteins derived from
tax
-activated fibroblasts were performed as reported before (
7
).
Bcr-abl
antisense experiments
All
in vitro
antisense experiments were carried out in triplicate. Standard deviations for
cell counts were usually within 10-15%. Averaged results are presented in all figures, and error bars are
omitted for clarity.
The experiments were conducted in BV173, HL-60 or K562 cells as described (
19
). Clonogenic assays with primary CML patient cells and normal bone marrow cells
were conducted as follows. Marrow cells were obtained by aspiration from the
illiac crest of healthy individuals and CML patients after informed consent.
Light density mononuclear cells were separated using a Histopaque-1077 (Sigma) density gradient. Marrow cells from healthy individuals were
enriched for hematopoietic progenitors after removing adherent cells and T
lymphocytes (A-T-NBMC) as described (
20
). Cells were incubated with ODNs as described (
21
). Briefly, 10
5
cells in 0.4 ml of Iscove's modified Dulbecco medium (IMDM) supplemented with
10% heat-inactivated fetal bovine serum (Sigma), l-glutamine (Gibco-BRL) and penicillin/streptomycin (Gibco-BRL), were cultured in 24-well plates with the indicated concentrations of
phosphorothioate or phosphoramidate oligodeoxynucleotides (see legend to Fig.
5
) and 10 U/ml of recombinant human interleukin-3 (IL-3, Genetics Institute). The second and the third dose of ODNs (50%
of the first dose) were added after 24 and 48 h of culture. After 120 h in
culture, cells were plated in MethodCult H4230 (Stem Cell Technologies)
semisolid medium as described (
20
), and colonies were counted 9-12 days later.
RT-PCR amplification of
bcr-abl
mRNA and hybridization with a specific
32
P-labeled 18mer 5'-GAAGGGCTTCTTCCTTAT-3' probe was carried out as described (
19
). The
bcr-abl
specific primers were located at 3273-3294 (
bcr
portion) and 431-452 (c-
abl
portion) nucleotide positions for 5'- and 3'-primers, respectively. The amplified fragment was 257
bp long.
C-
myb
and c-
myc
antisense experiments
The HL-60 acute promyelocytic leukemia cell line used for antisense experiments
was maintained in RPMI 1640 medium supplemented with 10% fetal bovine serum
(FBS, Sigma), l-glutamine and penicillin/streptomycin (Gibco-BRL). During the
in vitro
treatment with the ODNs, the cells were cultured in the same complete medium
supplemented with 10% heat-inactivated fetal bovine serum.
To assess the
in vitro
dose-response of HL-60 cells to anti-
c-myb
ODNs, exponentially growing cells were seeded in 96-well microtiter plates (Falcon) at an initial concentration of 1 * 10
4
/ well/100 [mu]l in complete medium. At the beginning of the culture (on day 0) ODNs were
added at the following concentrations: 3, 6 or 12 [mu]M for the phosphorothioates and 0.5, 1, 2 or 5 [mu]M for the 15mer phosphoramidates, Table
1
. Number of cells and viability (trypan blue exclusion test) were determined
daily until day 6.
The effect of c-
myb
ODNs on c-
myb
expression was studied at both the mRNA and protein level. For mRNA level
studies, HL-60 cells were plated in 48-well multiplates (Falcon) at an initial concentration of 2 * 10
5
/ml in complete medium. On the same day, the cells were treated with c-
myb
ODNs at 12 [mu]M for the 24mer phosphorothioates and at 5 [mu]M for the 15mer phosphoramidates. After 48 and 72 h from the beginning
of the treatment, samples were collected by centrifugation and washed with PBS.
Total RNA was extracted in the presence of 10 [mu]g
Escherichia coli
ribosomal RNA by the acid guanidinium thiocyanate-phenol-chloroform method. The RNA from each sample was then analyzed by
RT-PCR. For protein level studies, the cells were plated in 6-well multiplates (Falcon) at 2.5 * 10
5
/ml in complete medium. On the same day, the cells were treated with anti-
c-myb
24mer phosphorothioates at 15 [mu]M and 15mer phosphoramidate at 2 or 5 [mu]M. After 48 and 72 h from the beginning of the treatment, the samples
were collected by centrifugation, washed with ice-cold PBS and solubilized in lysis buffer (10 mM HEPES, pH 7.5; 150 mM
NaCl; 1% NP-40; 10% glycerol; 10 [mu]g/ml each leupeptin and aprotinin: 1 mM phenylmethylsulfonyl fluoride;
1 mM NaVO
4
; 5 mM EDTA). Total lysates were frozen at -80oC, thawed at 37oC, centrifuged at 11 000 r.p.m. at 4oC for 15 min, and postnuclear supernatants were then
fractionated by SDS-PAGE.
RT-PCR analysis of c-
myb
mRNA was carried out as described (
22
).
.
Oligonucleotide N3' -> P5' phosphoramidate sequences and their targets
No.
Oligonucleotide
a
Target
Type
b
1
CTTCTTCCTTA
bcr-abl
AS
2
T
T
CTC
TC
TC
TA
bcr-abl
MM
3
AACGTTGAGGGGCAT
c-
myc
AS
4
AACG
AGTT
GGGGCAT
c-
myc
MM
5
TTTCATTGTTTTCCA
c-
myc
AS
6
TTTCT
A
T
TG
TTTCCA
c-
myc
MM
7
GTGCCGGGGTCTTCG
c-
myb
AS
8
GT
C
C
T
GGGGTC
G
TCG
c-
myb
MM
9
GTGC
G
GG
T
G
C
CTTCG
c-
myb
MM
10
TATGCTGTGCCGGGGTCTTCGGGC
c
c-
myb
AS
11
GCCCGAAGACCCCGGCACAGCATA
c
c-
myb
S
12
CCACCGGGTCCAC
bcr
AS
13
TTGGGGTT
d
-
SCR
14
TTTTTTTTTT
-
SCR
15
TGTGGGATTTTCCCAT
e
-
SCR
16
ATGGGAAAATCCCACA
-
SCR
a
Oligonucleotide sequences are 5' to 3'.
b
AS, S, MM and SCR are abbreviations for antisense, sense, mismatched and
scrambled oligomers, respectively.
c
These oligonucleotides were used only as phosphorothioates.
d
This compound binds to gp120 HIV protein, see refs 35 and 36.
e
Oligomers
15
and
16
are complementary to each other and contain the recognition site for NF-kB transcription factor, see ref. 7.
SDS-PAGE and Western blotting
The protein lysates from each sample were fractionated by 8.5% SDS-PAGE and transferred onto a nitrocellulose membrane (Schleicher & Schuell). The membrane was saturated in PBS containing 5% non-fat dry-milk/0.5% Tween-20 at room temperature for 2 h and incubated for 12 h
at 4oC with a monoclonal anti-mouse c-
myb
antibody (UBI) used at a concentration of 1 [mu]g/ml in TBS (Tris-buffered saline) containing 0.1% gelatin and 0.01% NaN
3
. After washing five times with 0.125% Tween-20, 0.125% NP-40 in TBS, the filter was incubated for 1 h at room temperature with
sheep anti-mouse IgG conjugated to horseradish peroxidase (Amersham) and then washed
again five times as described above. Bound proteins were detected by using the
ECL Western blotting detection system (Amersham) according to the
manufacturer's instructions. After stripping, the filter was blotted with human
heat shock protein (HSP 72/73) monoclonal antibody (Oncogene Science Inc.) as a
control for loaded protein amounts.
Antisense experiments in HL-60 cells with c-
myc
as a target and subsequent Western blot analysis were conducted analogously to
those for c-
myb
; experiments in primary CML patient cells were done as described above for the
anti-
bcr-abl
oligomers.
RESULTS AND DISCUSSION
Hydrolytic stability and protein binding of the oligonucleotide N3
' ->
P5
'
phosphoramidates
Oligonucleotide N3' -> P5' phosphoramidates used in this study and their targets are
listed in Table
1
.
First, stability of oligonucleotide phosphoramidates toward hydrolysis with
phosphodiesterases was evaluated. The decathymidilic acid
14
containing natural phosphodiester and N3' -> P5' phosphoramidate linkages were treated with a mixture of
snake venom phosphodiesterase and alkaline phosphatase and the course of
hydrolysis was monitored by RP-HPLC (see Materials and Methods). Chromatographic analysis showed that
the phosphodiester compound was completely hydrolyzed in 10 min and thymidine
was formed as the sole product of digestion. In contrast, phosphoramidate
14
(Table
1
) was practically intact after 10 min of reaction. After 4.5 h, ~50% of the phosphoramidate still remained intact and mainly the 9mer with a
terminal 3'-amino group and thymidine were formed. Complete transformation of
the starting 10mer to shorter products, mainly 7-, 8- and 9mers was observed after 22 h (data not shown). Additionally, 5'-
32
P-labeled phosphoramidates
12
and
13
(Table
1
) were completely resistant to 3'-exonuclease hydrolytic activity in 50% human plasma after 8 h of
exposure, or in 50% HeLa cell nuclear extract after 2 h, as judged by
electrophoretic analysis of the reaction mixture in 20% polyacrylamide gel
(data not shown). These experiments demonstrate that oligonucleotide N3' -> P5' phosphoramidates are stable toward hydrolysis with
nucleases and can be used for
in vitro
and
in vivo
experiments.
One of the main features of the oligonucleotide phosphorothioates and to some
extent of phosphodiesters is non-sequence specific binding to a variety of proteins, which is determined by
the nature of their sugar-phosphate backbones (
6
-
8
). This binding may lead to unexpected effects of antisense oligonucleotides as
well as to a reduction of the effective concentrations of oligonucleotides
available to hybridize with the intended mRNA target. We studied the ability of
oligonucleotide phosphoramidates to associate with proteins in comparison with
isosequential phosphodiesters using a gel shift analysis as described before (
7
). Nuclear extract from
tax
activated fibroblasts was incubated with 5'-
32
P-labeled phosphoramidate oligonucleotides
15
and
16
and their phosphodiester counterparts (Table
1
) and then the products of nucleic acid-protein interactions were analyzed by gel electrophoresis. As shown in
Figure
1
, under identical conditions single-stranded phosphodiester oligonucleotides do bind to some nuclear proteins,
whereas isosequential phosphoramidates do not. Similar results were previously
reported (
23
), where phosphoramidates
3
and
4
(Table
1
) but not isosequential phosphorothioates did not seemingly associate with
extracellular proteins in U87 human glioblastoma cells. The observed lack of
binding to nuclear proteins, which may render oligonucleotide phosphoramidates
more available for binding with the mRNA target, could be caused by several
important differences between phosphodiester and phosphoramidate compounds:
substitution of the 3'-oxygen atom by a 3'-amino group leads to a change in the sugar-phosphate
backbone rigidity (
18
); geometry of the nucleoside sugar rings, where phosphodiester and
phosphoramidate linked compounds exist predominantly in S- and N- conformations, respectively (
17
,
24
); a different formal negative charge on the phosphate groups resulting in a
change in their hydration pattern (
18
).
Bcr-abl
as an antisense target
To evaluate the antisense properties of oligonucleotide phosphoramidates,
several
in vitro
model systems were selected. One of them was designed to study inhibition of
expression of p210 tyrosine kinase, resulting from the 9/22 chromosomal
translocation and associated with chronic myelogenous leukemia (CML) (
25
). Oligonucleotides
1
and
2
(Table
1
) were synthesized to be fully or partially complementary to the junction point
of the
bcr
(b2) and
abl
(a2) fragments of the
bcr/abl
mRNA. Thus, BV173 cells containing the b2/a2 translocation and expressing p210
tyrosine kinase were treated with oligonucleotide
1
, and a dose-dependent inhibition of cellular proliferation was observed (Fig.
2
). In contrast, proliferation of K562 cells, containing the partially
complementary (8/11 bp) b3/a2 junction, or HL-60 cells without any
bcr/abl
translocation, were not affected by this oligonucleotide (data not shown). The
study was extended to experiments with primary bone marrow cells taken from CML
patients (chronic phase) which contained translocated b2/a2 mRNA. In this
clonogenic assay, oligonucleotide phosphoramidate
1
(Table
1
) showed good target-specific inhibitory activity against colony formation by CML cells with an
IC
50
value 0.18 [mu]M, but not against normal bone marrow cells, IC
50
3.0 [mu]M, thus demonstrating a therapeutic index equal to 17 (Fig.
3
C). In addition to the colony formation assay, the antisense effects of
oligonucleotides were monitored by RT-PCR analysis of the b2/a2 mRNA level and were normalized to the internal
reference [beta]-actin mRNA; oligonucleotide phosphorothioates were also used as
reference compounds. As demonstrated in Figure
3
A and B, a significant reduction of the mRNA level correlated with the effective
inhibition of colony formation observed in cells treated with antisense
oligonucleotide
1
, but not with the mismatched control
2
.
C-
myc
as an antisense target
Oligonucleotide phosphorothioates have been extensively used as antisense
compounds for regulation of c-
myc
protein expression in different cellular and
in vivo
systems (
28
-
31
). To compare the activity of phosphoramidates with the well-characterized behavior of phosphorothioates, several oligonucleotide
phosphoramidates
3-6
(Table
1
) were synthesized for antisense inhibition experiments of c-
myc
protein expression. Thus HL-60 cells, expressing c-
myc
, were treated with oligomer
3
complementary to the start site and the following four codons of c-
myc
mRNA, or with compound
5
, complementary to the 1709-1724 nt region of c-
myc
mRNA. Mismatched oligomers
4
and
6
were used as negative controls. A profound dose-dependent inhibition of HL-60 cell proliferation was observed for the antisense oligomers
3
and
5
, but not for the mismatched compounds
4
and
6
(Fig.
4
) (data for
5
and
6
not shown). The reductions of c-
myc
protein level in HL-60 cells after treatment with antisense phosphoramidates at 1 [mu]M were in good agreement with the cellular proliferation data, as
determined by Western blot analysis (Table
2
).
The
in vitro
experiments with
anti-c
-
myc
phosphoramidates were extended to primary cells from CML patients (eight
patients tested). In a representative experiment a dose-dependent reduction of colony formation in the 0.4 nM-4 [mu]M dose range was observed after treatment of primary cells with
antisense oligomers
3
and
5
, but not with mismatched controls
4
and
6
, with oligomer
3
being 10-100 times more active than
5
, indicating a sequence specificity in the action of oligonucleotide
phosphoramidates (Fig.
6
). In this dose range the growth of normal bone marrow cells was not affected by
the mismatched oligomers
4
and
6
, and reduced by 40-50% with antisense oligomer
3
and
5
only at the highest 4 [mu]M concentration (data not shown). The observed high
in vitro
activity of the phosphoramidates is probably due to their high binding affinity
to the mRNA targets and low adhesion to proteins. Additionally, one may
speculate that an as yet unidentified concentration mechanism might elevate the
intracellular concentration of phosphoramidates above extracellular levels and
consequently lead to the high apparent antisense activity.
.
Inhibition of c-
myc
protein expression by antisense oligonucleotides in HL-60 cells
Expt
Oligonucleotide
No.
a
Type and
Concentration
Relative
backbone
b
([mu]M)
inhibition (%)
c
1
AACGTTGAGGGGCAT
3
AS, s
10
74
2
ATGCCCCTCAATGTT
17
S, s
10
41
3
AACGTTGAGGGGCAT
3
AS, np
1
70
4
AACG
AGTT
GGGGCAT
4
MM, np
1
41
5
TTTCATTGTTTTCCA
5
AS, np
1
93
6
TTTC
TA
T
TG
TTTCCA
6
MM, np
1
33
a
Numbers correspond to those from Table 1 and mismatched nucleotides are
underlined.
b
Abbreviations for oligonucleotides type are the same as in Table 1, and s and np
correspond to the uniformly modified phosphorothioate and N3' -> P5' phosphoramidate oligonucleotides, respectively;
c
Amount of the c-
myc
determined densitometrically from Western blot was normalized to amounts of
72/73 heat shock proteins, and for the untreated cells relative inhibition is
0%.
C-
myb
as an antisense target
CONCLUSIONS
A series of uniformly modified oligonucleotide N3' -> P5' phosphoramidates were synthesized. These compounds are very
stable toward the hydrolytic activity of nucleases from snake venom, HeLa cell
nuclear extract and human plasma, yet apparently lack non-sequence specific protein binding. Moreover, they act as efficient and
sequence-specific antisense agents regulating
bcr-abl
,
c-myc
and c-
myb
protein expression in different
in vitro
cellular systems. The presented data indicate that oligonucleotide N3' -> P5' phosphoramidates have good potential for development as
antisense-based therapeutic agents.
ACKNOWLEDGEMENTS
We wish to thank C. Chavany and L. Neckers for conducting experiments in U87
glioblastoma cells and P. Dal Bozzo for expert assistance in preparation of
this manuscript. We are grateful to Olaf Heidenreich and Michael Nerenberg for
conducting the gel-shift experiments and helpful discussions.
