(Received for publication, November 19, 1996)
From Isis Pharmaceuticals, Inc., Carlsbad, California 92008
Little is known about the mechanisms that account
for inhibition of gene expression by antisense oligonucleotides at the
level of molecular cell biology. For this purpose, we have selected potent 2-O-(2-methoxy)ethyl antisense oligonucleotides
(IC50 = 2 and 6 nM) that target the 5
cap
region of the human intercellular adhesion molecule 1 (ICAM-1)
transcript to determine their effects upon individual processes of
mRNA metabolism in HUVECs. Given the functions of the 5
cap
structure throughout mRNA metabolism, antisense oligonucleotides
that target the 5
cap region of a target transcript have the potential
to modulate one or more metabolic stages of the message inside the
cell. In this study we found that inhibition of protein expression by
these RNase H independent antisense oligonucleotides was not due to
effects on splicing or transport of the ICAM-1 transcript, but due
instead to selective interference with the formation of the 80 S
translation initiation complex. Interestingly, these antisense
oligonucleotides also caused an increase in ICAM-1 mRNA abundance
in the cytoplasm. These results imply that ICAM-1 mRNA turnover is
coupled in part to translation.
Antisense oligonucleotides have been shown to be effective agents
for inhibition of gene expression at the mRNA level (1-3). They
may be described as exogenous regulators of mRNA metabolism intended to sterically interfere with one or more metabolic processes upon hybridization, such as initiation of translation, or to promote enzyme-mediated mRNA degradation by formation or exposure of a region for nuclease activity, such as RNase H. The mode of action of an
antisense oligonucleotide in cells is dependent upon its composition
(sugar, backbone, and base residues) and mRNA binding site location
(5-UTR, coding region, 3
-UTR).1 Although
several types of antisense oligonucleotides, which differ in
composition and target site, have been found to be effective agents for
sequence-specific inhibition of gene expression in mammalian cells,
direct or detailed evidence of their mode(s) of action remains
limited (4-9).
Intercellular adhesion molecule 1 (ICAM-1) is one of several cell
adhesion molecules expressed on the cell surface of vascular endothelium that participates in a broad range of immune and
inflammatory responses (10). ICAM-1 is also expressed on nonendothelial
cells, such as keratinocytes, monocytes, and fibroblasts in response to
inflammatory mediators. Elevated levels of ICAM-1 expression have been
observed in a number of immune-related human diseases (11, 12),
e.g. rheumatoid arthritis, psoriasis, and asthma. Thus,
regulation of ICAM-1 gene expression is of therapeutic interest (13-15). The ICAM-1 gene has been sequenced, and the transcription initiation site has been characterized for several cell lines following
induction by a variety of cytokines (16, 17), including human umbilical
vein endothelial cells (HUVECs) with induction by TNF- (18).
Previous research has demonstrated that elevated expression of ICAM-1
may be controlled in cells by phosphorothioate-modified antisense
oligonucleotides (4, 5). At that time the most effective
oligonucleotides were those that were compatible with RNase H and
targeted the 3-UTR of the transcript. Since then advances in chemical
synthesis have brought forth a number of oligonucleotide modifications
at the 2
-sugar position which give significant increases in duplex
stability and nuclease resistance but do not support RNase H activity
(19). Antisense oligonucleotides that bind more tightly to the target
mRNA are expected to be more effective at interfering with the
processes of metabolism when bound at suitable locations. Bulky
substituents at this position also have been shown to provide a high
degree of nuclease resistance.
Biophysical and biological analysis of a set of these uniformly
2-modified oligonucleotides (2
-O-methyl (20),
2
-O-allyl (20), 2
-O-(2-methoxy)ethyl (21), and
2
-fluoro (22)) that target the 5
terminus of the ICAM-1 transcript
led to our selection of the exceptionally active
2
-O-(2-methoxy)ethyl-modified oligonucleotides, ISIS 11158 and 11159, for an investigation of their intracellular mode of action
in HUVECs (Fig. 1 and Table I). The 5
cap of eukaryotic
mRNA has been shown to be a structural element that functions
throughout mRNA metabolism (24-36). Therefore, antisense oligonucleotides which target the 5
cap region of a designated transcript have the potential to modulate one or more metabolic stages
of the message inside the cell (37). In this study the antisense mode
of action was determined by evaluation of the target transcript's
metabolic processes (splicing, transport, translation, and stability)
following antisense treatment and induction of gene expression.
|
HUVECs were purchased from Clonetics Corp. (San Diego, CA) and cultivated in the designated EBM medium supplemented with 10% fetal bovine serum from HyClone (Logan, UT). Cells were used for experiments from passages two to ten at 80-90% confluency.
Oligonucleotide SynthesisOligonucleotides were synthesized
utilizing conventional solid-phase triester chemistry (38). The
2-deoxy (Perseptive Biosystems), 2
-O-methyl (ChemGene),
and 2
-O-allyl (Boehringer Mannheim) phosphoramidites were
purchased from commercial sources. 2
-O-(2-Methoxy)ethyl and
2
-fluoro phosphoramidites were manufactured either in house (Dr. Bruce
Ross) or under contract (R.I. Chemicals).
Cells were washed three
times with Opti-MEM (Life Technologies, Inc.) pre-warmed to 37 °C.
Oligonucleotides were premixed with 10 µg/ml Lipofectin (Life
Technologies, Inc.) in Opti-MEM, serially diluted to the desired
concentrations, and applied to washed cells. Basal and untreated (no
oligonucleotide) control cells were also treated with Lipofectin. Cells
were incubated for 4 h at 37 °C, at which time the medium was
removed and replaced with standard growth medium with or without 5 ng/ml TNF- (R & D Systems). Incubation at 37 °C was continued
until the indicated times.
Cells were removed from plate
surfaces by brief trypsinization with 0.25% trypsin in PBS. Trypsin
activity was quenched with a solution of 2% bovine serum albumin and
0.2% sodium azide in PBS (+Mg/Ca). Cells were pelleted by
centrifugation (1000 rpm, Beckman GPR centrifuge), resuspended in PBS,
and stained with 3 µl/105 cells of the ICAM-1 specific
antibody, CD54-PE (Becton Dickinson) and 0.1 µg of the control
antibody, IgG2b-PE (Pharmingin). Antibodies were incubated with the
cells for 30 min at 4 °C in the dark, under gentle agitation. Cells
were washed by centrifugation procedures and then resuspended in 0.3 ml
of FacsFlow buffer (Becton Dickinson) with 0.5% formaldehyde
(Polysciences). Expression of cell surface ICAM-1 was then determined
by flow cytometry using a Becton Dickinson FACScan. Percentage of the
control ICAM-1 expression was calculated as follows:
[(oligonucleotide-treated ICAM-1 value) (basal ICAM-1 value)/(nontreated ICAM-1 value)
(basal ICAM-1 value)].
Cells were treated
with oligonucleotide at a concentration of 50 nM for 4 h. Cells were harvested 2 h post TNF- induction. Preparation of
nuclei and measurement of gene transcription were based upon published
procedures (39). Equal counts/min of 32P-labeled RNA were
hybridized to slot-blot membranes loaded with cDNA fragments for
ICAM-1 and G3PDH. The
X 174 DNA HaeIII digest was
included as a control.
Total cellular RNA was isolated from HUVECs by lysis and precipitation with Catrimox-14 (Iowa Biotechnology Corp.), followed by extraction of the DNA from the precipitate with lithium chloride. Isolated RNA was separated on a 1.0% agarose gel containing 1.1% formaldehyde, then transferred and UV-crosslinked (Stratalinker 2400, Stratagene) to a Hybond N+ nylon membrane (Amersham). Blots were hybridized with Prime-a-Gene cDNA probes (Promega) using RapidHyb (Amersham). Probes were generated from the following human cDNA restriction or PCR fragments: a 1.88-kilobase ICAM-1 fragment (BBG 58, R & D Systems), a 1.1-kilobase G3PDH fragment (pHcGAP, American Tissue Culture Collection), and a 2-kilobase E-selectin fragment (4). A Molecular Dynamics PhosphorImager was utilized to quantitate Northern blot probe signals.
Nuclear and Cytoplasmic RNA FractionationHarvested cells
were incubated in mild lysis buffer (0.5% Nonidet P-40, 10 mM Tris-Cl (pH 7.4), 140 mM KCl, 5 mM MgCl2, and 1 mM DTT) for 5 min
at 4 °C (40). Nuclei were separated from the cytosol by
centrifugation at 1000 × g for 5 min at 4 °C. The cytosol was transferred to a sterile tube containing 3 volumes of a
denaturing solution (5.3 M guanidinium isothiocyanate, 37.5 M sodium citrate (pH 7.0), 0.75% Sarkosyl, and 0.15 M -mercaptoethanol), phenol-extracted under acidic
conditions (pH 4.0), and isopropyl alcohol-precipitated (41). The
cytosol precipitate was redissolved in 300 µl of the denaturing
solution plus 30 µl of 2 M sodium acetate (pH 4.0) and
reprecipitated with isopropyl alcohol. A second lysis step was
performed on the collected nuclei to ensure removal of the cytosol
fraction. Washed nuclei were lysed at room temperature by the addition
of 1 ml of Catrimox-14 surfactant. Nuclear RNA was isolated by the LiCl
extraction procedure.
Approximately 106 oligonucleotide-treated cells were pelleted, washed with PBS, then mixed into 0.3 ml of cold lysis buffer (0.5% Nonidet P-40, 10 mM Tris-Cl (pH 7.4), 140 mM KCl, 5 mM MgCl2, 1 mM DTT, 100 µg/ml cycloheximide, and RNase inhibitor (5 Prime 3 Prime)) and incubated for 5 min at 4 °C. Nuclei were pelleted at 1000 × g, and the resulting supernatant was layered on a 10-35% (w/v) linear sucrose gradient (4 ml), with a 50% cushion (0.75 ml), in gradient buffer (10 mM Tris (pH 8.0), 50 mM potassium acetate, 1 mM magnesium acetate, 1 mM DTT). Gradients were centrifuged at 35,000 rpm for 3 h at 5 °C with a Beckman SW55 Ti rotor. 250-µl fractions were collected with an Isco model 185 density gradient fractionator connected to a Pharmacia UV monitor and fraction collector. Collected fractions were treated with proteinase K (0.2 mg/ml) in 0.2% SDS at 42 °C for 20 min, phenol-extracted, and ethanol-precipitated. 5 to 10 µg of tRNA was added to each fraction prior to precipitation. Precipitated RNA was applied to a 1.0% denaturing agarose gel and analyzed by standard ethidium bromide staining and Northern blotting techniques. Fractions 1 and 2 and 3 and 4 were combined for gel analysis.
Scrambled control oligonucleotides were tested in a dose-response
analysis to verify that inhibition of ICAM-1 protein expression by the
2-O-(2-methoxy)ethyl-modified oligonucleotides, ISIS 11158 and 11159, was sequence-specific. The respective scrambled control oligonucleotides, ISIS 12344 and 12345, showed negligible effects on
ICAM-1 protein expression (Fig. 2). As indicated in
Table I, both the phosphodiester, ISIS 11158, and the
phosphorothioate, ISIS 11159, 2
-O-(2-methoxy)ethyl-modified oligonucleotides were more potent inhibitors of ICAM-1 expression in HUVECs than the analogous RNase H-compatible phosphorothioate oligodeoxynucleotide, ISIS 3067.
Intracellular distribution of the
2-O-(2-methoxy)ethyl-modified oligonucleotides was
determined by fluorescence microscopy, using fluorescein-labeled
oligonucleotides, to further compare and delineate the basis of their
antisense activity with respect to the first generation
2
-deoxyoligonucleotides (Fig. 3). As reported
previously (42) treatment of HUVECs with the fluorescein-labeled 2
-deoxy phosphorothioate oligonucleotide, in the presence of the
cationic lipid formulation (Lipofectin, Life Technologies, Inc.),
resulted in a heterogeneous accumulation of the oligonucleotide in the
cell nucleus as well as in punctate cytoplasmic vesicles (Fig.
3A). Treatment of HUVECs with the 2
-deoxy phosphodiester analog showed a diffuse distribution of label in the cytoplasm (Fig.
3B), attributed to degradation of the oligomer by nucleases (43, 44). In comparison, the fluorescein-labeled
2
-O-(2-methoxy)ethyl-modified phosphorothioate
oligonucleotide yielded a distribution pattern (Fig. 3C)
similar to the 2
-deoxy analog (Fig. 3A), with a high degree
of localization in the nucleus. However in striking contrast to the
2
-deoxy analog (Fig. 3B), the
2
-O-(methoxy)ethyl phosphodiester showed a homogeneous
nuclear localization (Fig. 3D), attributed to its greater
resistance to nucleases, the absence of "nonspecific" phosphorothioate-protein interactions (45, 46), and possibly more
compatible interactions with the lipid formulation for uptake and
delivery.
Total cellular RNA was isolated and analyzed to determine whether
inhibition of ICAM-1 protein expression by ISIS 11158 and 11159 resulted from antisense-promoted degradation of the target transcript,
an end point observed following treatment with most active RNase
H-compatible antisense oligonucleotides (4, 5). Total RNA was harvested
at 4 and 20 h following a 1-h TNF- induction period for
untreated and oligonucleotide-treated HUVECs. Interestingly, Northern
blot analysis showed a significant increase in relative abundance of
the ICAM-1 transcript in those cells treated with the anti-ICAM-1
oligonucleotides at both time points (Fig. 4). Nuclear
runoff experiments showed that this increase in transcript abundance
was not due to an increase in the rate of transcription of the ICAM-1
gene (data not shown).
To determine whether the antisense effect on transcript abundance was
specific to ICAM-1, blots were probed for the E-selectin transcript
whose expression is also transiently induced by TNF- in HUVECs. As
with ICAM-1 an increase in abundance of the E-selectin transcript was
observed only in those cells treated with the
2
-O-(2-methoxy)ethyl-modified oligonucleotide, ISIS 11929, that targets the 5
-terminal region of the E-selectin transcript (Fig.
5). The increase in target transcript abundance
following antisense treatment may be an attribute of these transiently
expressed transcripts in combination with the antisense mode of
action.
The 5 cap structure of mRNA has been shown to facilitate splicing
of the first intron of pre-mRNA constructs in several different systems (48, 49). The ICAM-1 gene consists of seven exons separated by
six introns, with intron 1 approximately 4000 nucleotides in length
(16). Evaluation of the Northern blots showed that this set of modified
antisense oligonucleotides had no effect on splicing of the ICAM-1
pre-mRNA to the mature transcript, as evidenced by lack of ICAM-1
mRNA species or intermediates of longer lengths (data not
shown).
Nuclear and cytosolic fractionation was utilized to determine if
antisense inhibition of ICAM-1 protein expression resulted from
inhibition of transport of the mature transcript out of the nucleus to
the cytoplasm. Fractionated mRNA was evaluated by Northern analysis
2 h post TNF- induction for 2
-O-(2-methoxy)ethyl
oligonucleotide-treated (phosphorothioate and phosphodiester; antisense
and control each at 50 nM) and untreated cells (Fig.
6). At this time point no substantial alteration in the
abundance of the ICAM-1 transcript was observed in the nuclear
fractions of antisense treated cells (110-114%) versus
scrambled control treated (113-118%) and untreated (100%). In
contrast, a significant increase in the abundance of the ICAM-1
transcript was observed in the cytosolic fraction from the antisense
treated cells, ISIS 11158 (230%) and ISIS 11159 (181%), in comparison
to the scrambled control treated, ISIS 12344 (133%) and ISIS 12345 (108%), and untreated cells (100%). Relative abundance of the ICAM-1
transcript in each compartment was also determined 4 h post 1-h
TNF-
induction. Under these conditions, the relative abundance of
the ICAM-1 transcript was 451 and 425% in the cytosolic fraction and
128 and 126% in the nuclear fractions of the antisense treated cells,
ISIS 11158 and 11159, respectively, relative to the untreated cells.
The significant increase in abundance of the transcript in the
cytoplasm of the antisense treated cells suggested a decrease in the
rate at which the transcript is normally degraded. The lack of a
substantial change in ICAM-1 mRNA abundance in the nuclear fraction
indicated that the antisense oligonucleotides did not significantly
affect the nucleocytoplasmic transport rate of the mature ICAM-1
transcript.
Polysome profiles were utilized to determine the effect of antisense
oligonucleotide treatment upon the translation process of the target
ICAM-1 transcript (Fig. 7). ICAM-1 protein and mRNA were evaluated 4 h after a 1-h TNF- induction from cells
treated with antisense oligonucleotides ISIS 11158 and 11159, and their respective scrambled controls ISIS 12344 and 12345. Cytosolic extracts
were sedimented by linear sucrose gradient centrifugation (10-35%).
The ethidium bromide-stained gel of the fractionated RNA showed a
respectable separation of the subpolysomal and polysomal pools (Fig.
7A). Assignment of the fractions were verified by UV
absorbance plots obtained during fractionation. Northern blots showed a
significant difference in the polysomal distribution of the ICAM-1
transcript in cells treated with ISIS 11158 and 11159, in comparison to
those of the controls, ISIS 12344 and 12345 (Fig. 7B). The
polysome profiles for the ISIS 11158 and 11159 treated cells showed the
majority (71 and 65%, respectively) of the full-length target
transcript localized in the subpolysome fractions, e.g.
40 s and 60 s, whereas the ISIS 12344 and 12345 polysome
profiles showed the majority (77 and 86%, respectively) of the target
transcript in the monosome and polysome fractions.
The polysome profile data for ISIS 11158 and 11159 indicate that
inhibition of ICAM-1 protein expression occurs through interference with translation initiation and specifically ribosomal assembly, as
indicated by the dramatic redistribution of transcript into the
subpolysome fractions. The formation of a stable antisense-mRNA duplex (or secondary structure) in the 5 cap region is likely the
basis of this effect (see Table I). The increase in abundance of the
ICAM-1 mRNA in the cytosolic fraction of the antisense treated
cells in conjunction with the change in the polysome distribution patterns indicates that one of the target transcript's decay pathways is coupled to translation. These data are consistent with observations of transcripts that contain stability determinants in the coding region, e.g. c-fos and c-myc (50,
51).
Regulation of gene expression may occur at one or more stages of
mRNA metabolism. The most well known examples of regulation through
mRNA metabolism have been found at the stages of translation (52)
and degradation (53) of the mature transcript, where certain mRNA
sequences and structural elements have been found to be key regulatory
determinants. Of particular relevance, it has been shown that stable
secondary and tertiary structures located in the 5-terminal region
may regulate initiation of translation (54-56). The
2
-O-(2-methoxy)ethyl-modified antisense
oligonucleotides, complementary to the 5
-terminal region of the target
transcript (nucleotides 1-20), mimic this endogenous mode of
regulation in cells by inhibiting formation of the 80 S translation
initiation complex. We believe that this event in turn affects the
turnover rate of the transiently expressed ICAM-1 transcript.
We thank John Brugger and Pierre Villiet for oligonucleotide synthesis, Tracy Reigle for graphic illustrations, Lex Cowsert for technical advice, and Stan and Rosanne Crooke for their comments on the manuscript.