Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Although the intestinal tract is bombarded with a daily stream of foreign antigens in the form of luminal bacteria and food particles, gut inflammation is the exception, rather than the rule. Cytokines secreted by the intestinal immune system are probably one of the key factors in maintaining the gut in quiescent homeostasis. In particular, the balance between proinflammatory cytokines like TNF-, IL-1, IL-12, and IFN-, and regulatory cytokines like transforming growth factor- and IL-10, may ultimately determine whether an immune response to a gut antigen is detrimental or innocuous (Strober et al., 1997). An increasingly wide body of evidence suggests that one of the pivotal cytokines in this process is TNF-. This is seen in both animal models of gut inflammation and human inflammatory bowel disorders like Crohn's disease.

Animal models of inflammatory bowel disease (IBD) may occur naturally, or they may be induced by exposure to chemicals or by genetic manipulation of an animal's immune system. TNF- neutralization has proven effective at blocking gut inflammation in all three scenarios. Antibodies to TNF- have prevented the colitis that develops spontaneously in cotton top tamarins (Watkins et al., 1997), and they have blocked DSS- (Kojouharoff et al., 1997; Murthy et al., 1999) and trinitrobenzene sulfonic acid-induced colitis (Neurath et al., 1997; Videla et al., 1998). TNF- antibodies have also proven effective in treating the colitis that develops when T cell- and natural killer cell-deficient mice are reconstituted with normal bone marrow (BMTg26 mice; Mackay et al., 1998), or when SCID mice are reconstituted with CD45 Rb^hi CD4+ T cells (Powrie et al., 1994).

Tissue from patients with active IBD express increased levels of both TNF- mRNA and protein (Murch et al., 1993; Funakoshi et al., 1998). A chimeric monoclonal antibody to human TNF- has shown efficacy in the treatment of patients with steroid-resistant Crohn's disease (Stack et al., 1997; Present et al., 1999). Data have also suggested that Crohn's patients who are at high risk of a relapse can be identified by higher levels of TNF- and IL-1 secretion from their lamina propria cells (Schreiber et al., 1999).

Although short-term TNF- blockade by mAbs appears to be relatively safe, there are anecdotal accounts of the antibodies causing severe side effects in a few patients. Antibody efficacy may also be diminished over time by the development of anti-antibody responses, even with antibodies that have been "humanized".

As an alternative approach to TNF- blockade, we have developed an antisense oligonucleotide (ISIS 25302) that binds to mouse TNF- mRNA, causing its degradation and blocking its translation into protein. ISIS 25302 is a second-generation phosphorothioate oligonucleotide containing methoxyethyl (MOE)-modified nucleosides on its 5' and 3' ends. The methoxyethyl modification both increases ASO affinity for targeted mRNA and provides heightened resistance to nucleases that can prematurely degrade an ASO (Altmann et al., 1996; Freier and Altmann, 1997). We tested ISIS 25302 in several murine models of colitis, all of which involve TNF- secretion during their pathogenesis. ISIS 25302 proved effective at ameliorating acute and chronic models of dextran sodium sulfate-induced colitis as well as the colitis that occurs spontaneously in IL-10-deficient mice.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Antisense Oligonucleotide Design and Synthesis

Modified oligonucleotides were synthesized on a Milligen model 8800 DNA synthesizer (Millipore Inc., Bedford, MA). A crude product of approximately 70% purity was purified and desalted by column chromatography using a Millipore HC18-HA column. Final purity was assessed by capillary gel electrophoresis and electrospray mass spectrometry. Oligonucleotides were found to be greater than 92% full-length material. 20-mer phosphorothioate oligonucleotides containing 2'-methoxyethyl (2'-MOE) modifications (Henry et al., 2000) were used in all experiments. Cytosine residues were methylated at their 5' position to reduce potential immunostimulatory properties (Krieg et al., 1995). ISIS 25302 (base sequence 5'-AACCCATCGGCTGGCACCAC-3', with 2'-MOE modifications indicated by underlining) is complementary to mouse TNF- mRNA and hybridizes to a segment in the coding region of TNF- (nucleotides 5891-5910, GenBank accession #Y00647). ISIS 30799, 30780, 30781, and 30782 are 2-, 4-, 6-, and 8-base mismatch control oligonucleotides, respectively, for ISIS 25302. Base sequences for these oligonucleotides are ISIS 30799: 5'-AACCCATCTGCGGGCACCAC-3'; ISIS 30780: 5'-AACCCAGCTGCGGTCACCAC-3'; ISIS 30781: 5'-AACCAAGCTGCGGTCCCCAC-3'; and ISIS 30782: 5'-CACCAAGCTGCGGTCCCCAA-3' (mismatched bases indicated by bold type). ISIS 15931 is the full 2'-deoxy analog of ISIS 25302 with identical base sequence but containing a 2'-deoxy, phosphorothioate chemistry. ISIS 18154 (5'-TCAAGCAGTGCCACCGATCC-3') is a 2'-MOE phosphorothioate control oligonucleotide with a similar, but scrambled, base composition compared with ISIS 25302. ISIS 13920 is a 2'-MOE phosphorothioate oligonucleotide, (5'-TCCGTCATCGCTCCTCAGGG-3'), which is recognized by an oligonucleotide-specific monoclonal antibody and was used for immunohistochemical staining as detailed below. Cytosines in the MOE-modified "wings" of ISIS 13920 were methylated, but cytosines in the "gap" were not.

Cell Culture and Oligonucleotide Transfection

P388D₁-(IL1) cells were purchased from the American Type Culture Collection (Manassas, VA) and grown in RPMI-1640 medium (Invitrogen, Carlsbad, CA) supplemented with 15% fetal bovine serum. Cells were plated 1 day before transfection. Oligonucleotides at the indicated concentrations were then mixed with 3 µg of Lipofectin (Invitrogen) per 100 nM oligonucleotide in OptiMEM medium (Invitrogen) and incubated for 4 h at 37°C and 5% CO₂. Transfection medium was replaced with normal growth medium containing 10 ng/ml LPS from Salmonella typhosa (Difco, Detroit, MI) for 2 h at 37°C prior to RNA isolation.

TNF- mRNA Analysis

In Vitro. Total RNA from mouse P388D₁-(IL1) cells was isolated using an RNeasy Mini Kit (QIAGEN, Valencia CA). RNA was separated on a 1% agarose formaldehyde gel and transferred to a Hybond N⁺ membrane (Amersham Biosciences Inc., Piscataway NJ) for Northern blot analysis. The RNA was UV-crosslinked to the membrane and hybridized with a mouse TNF- probe generated by random priming of a 0.5-kilobase mouse TNF- cDNA restriction fragment or polymerase chain reaction fragment derived from the BBG 56 expression plasmid (R & D Systems, Minneapolis MN). RNA band intensities were measured using a PhosphorImager (Amersham Biosciences Inc.). The membrane was then stripped, probed, and analyzed for mouse G3PDH mRNA. TNF- mRNA levels were normalized to the corresponding G3PDH mRNA levels to correct for gel loading differences between samples.

In Vivo. For analysis of TNF- levels in adipose tissue, db/db mice were dosed every other day i.p. for 2 weeks with the indicated quantities of oligonucleotides. To analyze TNF levels in isolated macrophages, db/db mice were dosed daily i.p. for four consecutive days. Mice were euthanized by CO₂ inhalation, and adipose tissue from inguinal fat pads was removed for each experiment. A 1-g section of adipose tissue was homogenized in 3 ml of guanidinium isothiocyanate solution to isolate RNA directly from the whole tissue. Homogenate was layered onto 1.5 ml of cesium chloride (5.7 M, pH 5.6) and centrifuged at 35,000 rpm for 18 h. The RNA pellet was resuspended in 350 µl of RLT buffer (QIAGEN) and then further purified using the RNeasy Mini Kit (QIAGEN). Macrophages were isolated by digesting the adipose tissue with 2 mg/ml collagenase in Ringer's solution containing 5 mM glucose and 4% albumin for 2 h, followed by centrifugation at 700 rpm for 5 min. Pelleted macrophages were subsequently lysed with 350 µl of RLT buffer from which total RNA was again purified using the RNeasy Mini Kit. TNF- mRNA levels were measured by ribonuclease protection assay using a Riboquant kit (BD Biosciences PharMingen, San Diego CA), with 15 µg of RNA per sample loaded onto the gel. Probes were generated from the mCK-3b Multi-Probe Template set (BD Biosciences PharMingen), and labeled with [-³²P] UTP (Amersham Biosciences Inc.). Undigested probes were separated by polyacrylamide gel electrophoresis and analyzed with a PhosphorImager (Amersham Biosciences Inc.).

Mice

Female db/db mice (C57BLKS/J-m +/+Leprdb) and age-matched lean littermates (C57BLKS/J heterozygotes) were purchased from The Jackson Laboratory (Bar Harbor, ME) at 8 to 12 weeks of age and used to validate ISIS 25302 activity in vivo. For the DSS colitis studies, female Swiss-Webster mice were purchased from Taconic Farms (Germantown, NY) or The Jackson Laboratory and used at 7 to 8 weeks of age. Homozygous IL-10 gene-deficient mice, generated on a 129 Sv/Ev genetic background, and 129 Sv/Ev controls were housed under specific pathogen-free conditions until weaning, when they were moved to conventional animal housing. These mice were bred and raised in the animal facility at the University of Alberta. All animal procedures were conducted in accordance with protocols approved by institutional Animal Care and Use Committees.

Localization of Oligonucleotide in Colon

Female Swiss-Webster mice (n = 2) were maintained on either normal drinking water or dextran sodium sulfate-supplemented water, and then injected i.v. with 20 mg/kg ISIS 13920 or saline on days 1, 3, and 5 of the acute DSS colitis protocol (described below). ISIS 13920 is a 20-mer oligonucleotide specifically recognized by the 2E1 mAb (Butler et al., 1997). Mice were euthanized on day 7, and their colons were removed, trimmed longitudinally, and fixed in 10% neutral buffered formalin. Then, 4-µm sections were cut from paraffin-embedded tissues and deparaffinized by standard histological procedures. Endogenous tissue peroxidase activity was quenched with Peroxidase Blocking Reagent (DAKO, Carpinteria, CA) for 10 min. Tissue was treated with proteinase K (DAKO) for 10 min to make it permeable for staining. After blocking with normal donkey serum, the sections were incubated for 45 min with 2E1-B5 anti-oligonucleotide mAb. Sections were rinsed with phosphate-buffered saline and then incubated for 30 min with peroxidase-conjugated rabbit anti-mouse IgG1 (Zymed Laboratories, South San Francisco, CA), diluted 1:200. The 2E1 mAb was omitted from control sections, which received peroxidase-conjugated secondary antibody only. The immunostaining was developed with 3,3'-diaminobenzidine (DAKO) for 5 min, and slides were then counterstained with hematoxylin. All blocking and staining steps were conducted at room temperature.

Acute Dextran Sulfate Colitis Model

Disease Induction/Animal Dosing. Female Swiss-Webster mice were randomized into treatment groups of seven to eight animals and maintained on drinking water containing 4% dextran sodium sulfate, mol. wt. 40,000 (ICN Pharmaceuticals Biomedicals Division, Aurora OH) from day 0 until day 5. Mice received normal tap water from day 5 until sacrifice on day 7. Animals were administered the indicated doses of oligonucleotide or saline vehicle every other day by i.v. injection from day 2 to day 6. Anti-mouse TNF- mAb, clone XT22 (Pierce Endogen, Rockford, IL), was administered as a single i.v. dose on day 0.

Disease Scoring. A disease activity index was calculated on day 7 based on the summation of separate scores evaluating weight loss, blood in stool, and stool consistency. Each parameter was graded on a scale of 0 to 4 as previously described (Cooper et al., 1993), and the combined scores were then divided by 3 to obtain the final disease activity index. This method has been shown to correlate well with histological measures of inflammation and crypt damage. The presence of blood in the stool was evaluated with Hemoccult test strips (Quest Diagnostics, Inc., Teterboro, NJ). Colon lengths (from ileocecal junction to the anal verge) were also measured when mice were sacrificed.

Chronic Dextran Sulfate Colitis Model

Disease Induction/Animal Dosing. Female Swiss-Webster mice were weighed and randomized into treatment groups of 8 to 10 animals. Chronic colitis was induced by giving the mice 4% DSS in their drinking water for two cycles. For each cycle, DSS was administered until the disease activity index (DAI) reached a score of 2.0 to 2.5 in at least one group, at which time the 4% DSS was replaced with plain drinking water. The first cycle of DSS administration was followed by 14 days of plain drinking water. The second cycle of DSS was followed by 8 to 10 days of plain drinking water, after which the mice were sacrificed.

Disease Scoring. Disease progression was determined by daily measurement of a DAI, calculated as outlined above in the acute DSS model.

Histopathology. Hematoxylin-and-eosin-stained sections were obtained from proximal and distal sections of the colon. Crypt loss and inflammation (acute or chronic) were separately scored for each section, and the final crypt or inflammation score represents the sum of scores for both proximal and distal sections. Crypt loss and inflammation scores are the product of a grading scheme that analyzes both the severity of the lesion and the extent of tissue damage, as previously described (Cooper et al., 1993). The colon sections were scored by a pathologist in a blinded fashion.

IL-10 / Colitis Model

Animal Dosing. In the prophylactic treatment regimen, 2-week-old IL-10 / mice were treated for a total of 6 weeks by s.c. injection every other day with the indicated doses of oligonucleotides. In the therapeutic treatment regimen, 8-week-old mice were treated every other day for a total of 3 weeks with s.c. injection of the indicated doses of oligonucleotides. At 8 weeks of age, nearly 100% of IL-10 / mice have histological signs of colitis, whereas 2-week-old mice are largely disease-free. An additional treatment group in the therapeutic regimen consisted of mice receiving weekly i.p. injections of 2 mg/kg anti-mouse TNF- mAb (Pierce Endogen clone XT22).

Histopathology. Animals were sacrificed by sodium pentobarbitol injection (160 mg/kg). Whole colons were harvested, cut lengthwise, fixed in formalin, paraffin-embedded, sectioned at 4 µm, and stained with hematoxylin and eosin for light microscopic examination. The slides were reviewed independently by a pathologist in a blinded fashion and assigned a histological score for intestinal pathology (Table 1). The total histological score represents the numerical sum of the four scoring criteria: mucosal ulceration, epithelial hyperplasia, lamina propria mononuclear cell infiltration, and lamina propria neutrophilic infiltration.

View this table: [in this window] [in a new window]   TABLE 1 Grading scheme for histological scoring of intestinal injury to IL-10 (/) mice

Organ Culture Cytokine Measurement. Colon organ cultures were prepared at the termination of the prophylactic and therapeutic treatment regimens. Due to the patchy nature of colitis in IL-10 gene-deficient mice, whole colons were removed, cut lengthwise, flushed with phosphate-buffered saline, and placed in tissue culture plates containing RPMI-1640 medium supplemented with 10% fetal calf serum, 200 µM 2-mercaptoethanol, penicillin (100 U/ml), and streptomycin (100 U/ml). Cultures were incubated at 37°C in 5% CO₂. After 24 h in the presence or absence of 10 µg/ml Escherichia coli strain 0111:B4 LPS (Sigma-Aldrich, St. Louis, MO), supernatants were harvested and stored at 70°C for analysis of cytokine levels. TNF- and IFN- levels in cell supernatants were measured using enzyme-linked immunosorbent assay kits purchased from Biosource Cytoscreen (Montreal, QC, Canada).

Statistics

Differences between treatment groups were evaluated by analysis of variance. Specific differences were evaluated using the Student-Newman-Keuls test (IL-10 / studies) or Dunnett's test (DSS studies).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of Active Antisense Oligonucleotide

The activity of a panel of MOE-modified phosphorothioate oligonucleotides hybridizing to discrete sites throughout murine TNF- mRNA was evaluated in mouse P388D1-(IL1) cells. This is a macrophage-like cell line with constitutive TNF- production that was further increased by LPS treatment. From this screen, ISIS 25302 was identified as producing a dose-dependent decrease in TNF- mRNA expression (Fig. 1). This oligonucleotide reduced TNF- mRNA by 70% at a dose of 100 nM and had an IC₅₀ value of approximately 45 nM. The decrease in TNF- mRNA was sequence-specific, since a series of oligonucleotides containing 2-, 4-, 6-, or 8-base mismatches had either greatly reduced activity (2-base mismatch), or no activity whatsoever (4-, 6-, and 8-base mismatch oligonucleotides).

View larger version (44K): [in this window] [in a new window]   Fig. 1.   Activity and sequence specificity of the murine anti-TNF- oligonucleotide, ISIS 25302. Relative levels of TNF- mRNA in LPS-induced murine P388D1 cells are shown 2 h after oligonucleotide treatment. A, Northern blot showing phosphoimage of TNF- and G3PDH mRNAs. B, corresponding graph of phosphoimage data. ISIS 25302 is specific for mouse TNF-, whereas MM indicates the number of mismatched bases relative to ISIS 25302 contained in control oligonucleotides. mRNA levels are depicted as percentages relative to the level of TNF- mRNA found in LPS-induced cells.

The activity of ISIS 25302 was confirmed in vivo by the systemic dosing of db/db mice, a strain known to express TNF- in adipose tissue (Hotamisligil et al., 1993). db/db mice dosed every other day for 2 weeks with 10 mg/kg ISIS 25302 had a 64% reduction in TNF- mRNA in their inguinal fat pads, a level approaching that found in mice without the db mutation (Fig. 2). A scrambled control oligonucleotide had no effect on TNF- mRNA levels in db/db mice. An oligonucleotide of identical base sequence to ISIS 25302 but without the 2'-methoxyethyl chemical modification (ISIS 15931) was less effective than ISIS 25302, reducing TNF- mRNA by 42%. This is consistent with reports that the 2'-MOE modification increases antisense potency (Freier and Altmann, 1997).

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Reduction of TNF- mRNA levels in adipose tissue of db/db mice treated with TNF- antisense oligonucleotides. Oligonucleotides were administered i.p. at 10 mg/kg every other day for 2 weeks. Results are expressed as the mean + S.D., where n = 3 (except for control oligo where n = 2). ISIS 25302 is the second-generation TNF--specific oligonucleotide (2'-MOE gapmer chemistry), and ISIS 15931 is a first-generation deoxy phosphorothioate analog of ISIS 25302. The control oligonucleotide is ISIS 18154, a 2'-MOE gapmer with a similar, but scrambled sequence of bases relative to ISIS 25302. Asterisk indicates significant difference from saline-treated db mice (p < 0.05).

The adipose tissue of db/db mice contains increased numbers of macrophages compared with those found in fat from wild-type mice. ISIS 25302 also reduced TNF- mRNA levels in macrophages isolated from the db/db fat pads (Fig. 3). In this experiment, animals were dosed daily for just 4 days with 2 to 5 mg/kg ISIS 25302 or 5 mg/kg of a 6-base mismatch control oligonucleotide. ISIS 25302 reduced TNF- mRNA levels by more than 60%, whereas the mismatch control oligonucleotide had no statistically significant effect.

View larger version (27K): [in this window] [in a new window]   Fig. 3.   Sequence-specific reduction of TNF- mRNA levels in macrophages isolated from adipose tissue of db/db mice treated with ISIS 25302. Oligonucleotides were administered daily i.p. for 4 days. TNF- mRNA levels were determined by ribonuclease protection assay. A, ribonuclease protection assay showing phosphoimage of undigested TNF- probe. B, corresponding graph of phosphoimage data. TNF- mRNA levels are expressed relative to those found in mice treated with a scrambled control oligonucleotide (ISIS 18154). ISIS 25302 is the mouse TNF--specific oligonucleotide. The mismatch control oligonucleotide (ISIS 30781) contains six base mismatches relative to ISIS 25302. Results are expressed as mean ± S.D., where n = 3. Asterisk indicates significant difference compared with scrambled control (p < 0.05).

Oligonucleotide Distribution to Normal and Inflamed Colons

First-generation phosphorothioate oligonucleotides are known to localize in both normal and inflamed colon tissue (Bennett et al., 1997). Tissue distribution patterns of 2'-MOE-modified oligonucleotides are quite similar to those of first-generation oligonucleotides in normal animals (Bennett et al., 2000), but distribution patterns into inflamed tissue are largely unknown. To evaluate 2'-MOE oligonucleotide uptake in the inflamed mouse colons, a monoclonal antibody (2E1) that specifically recognizes phosphorothioate oligonucleotides (Butler et al., 1997) was used for immunohistochemical staining.

Mice were given 4% dextran sodium sulfate in their drinking water to induce colon inflammation. These animals also received three injections of ISIS 13920 during their 5-day exposure to DSS-supplemented water. ISIS 13920 is a 2'-MOE oligonucleotide that is recognized by the 2E1 monoclonal antibody. It has the same pattern of methoxyethyl modification and cytosine methylation as ISIS 25302, but a different nucleotide sequence. It is more strongly recognized by the 2E1 mAb than ISIS 25302 (unpublished data), facilitating interpretation of immunohistochemical staining patterns.

As seen in Fig. 4, 2'-MOE oligonucleotides are localized in cells scattered throughout the lamina propria and muscularis mucosa of the normal colon, with increased staining intensity present in the serosal layer (Fig. 4A). Mice exposed to DSS in their drinking water exhibit crypt pathology, including crypt shortening, erosion of the epithelial layer, goblet cell hypertrophy, and an inflammatory cell infiltrate in the lamina propria (Fig. 4D). Oligonucleotide distribution is greatly increased in these inflamed colons, and especially notable in the lamina propria. Staining for oligonucleotide is particularly intense in areas where inflammatory cells are present (Fig. 4B).

View larger version (101K): [in this window] [in a new window]   Fig. 4.   Localization of 2'-methoxyethyl-modified oligonucleotide, ISIS 13920, in colon tissue from normal and DSS-treated mice. Four-micrometer formalin-fixed, paraffin-embedded sections of mouse large intestine were stained with the 2E1 mAb, which specifically recognizes phosphorothioate oligonucleotides (including ISIS 13920). A and C, colon sections from normal mice given plain drinking water. B and D, colons from mice given DSS-supplemented drinking water. Sections shown in panels A and B were immunostained with the 2E1 mAb, whereas the sections in panels C and D received peroxidase-conjugated secondary Ab only. Original magnification, 200×.

TNF- Antisense Activity in Acute DSS Colitis Model

The activity of ISIS 25302 was evaluated in a model of acute murine colitis elicited by the substitution of 4% dextran sodium sulfate for normal drinking water for 5 days. Mice on this regimen develop inflammation and crypt pathology in the large intestine (Fig. 4), which peaks approximately a week after the initial exposure to DSS-water and then gradually resolves. A DAI, which measures weight loss, diarrhea, and the presence of blood in the stool, has been found to correlate well with tissue pathology in this model (Cooper et al., 1993). Colon length decreases with DSS exposure, and large intestine length is inversely correlated with increases in DAI.

Mice treated with ISIS 25302 every other day at a dose of 1 mg/kg in the acute DSS model had a 44% lower disease activity index than mice given saline (1.4 ± 0.2 versus 2.6 ± 0.2, respectively; Fig. 5A). In comparison, mice given a single 25-µg injection of a monoclonal antibody specific for mouse TNF- had a 57% reduction in DAI. In both cases, the reductions in DAI were statistically significant (p < 0.05) when compared with saline-treated controls. Paradoxically, mice given a higher dose of the TNF- antibody (50 µg) failed to show improvement in their DAI.

View larger version (43K): [in this window] [in a new window]   Fig. 5.   Efficacy of the anti-mouse TNF- oligonucleotide, ISIS 25302, versus anti-TNF- monoclonal antibody in the acute model of DSS-induced colitis. Animals were dosed every other day for a total of five injections with ISIS 25302 or given a single injection of the TNF- mAb. A, disease activity index. B, colon length. Results are expressed as the mean ± S.E.M., where n = 7. Asterisk indicates significant difference compared with saline-treated group (p < 0.05).

Improvements in DAI correlated with increases in colon length (Fig. 5B). The colons of saline-treated DSS mice had shortened to 57% of the length of those in mice given normal drinking water. Colons from mice treated with ISIS 25302 had increased to 76% of the length of control colons, a statistically significant effect. This increase in colon length was similar to that induced by the 25-µg dose of TNF- mAb. The 50-µg dose of antibody had no significant effect on colon length.

The effect of ISIS 25302 on the development of acute DSS colitis was sequence- and dose-dependent, although the dose-response relationship was not linear (Fig. 6). A reduction in clinical symptoms occurred with doses of 0.04 mg/kg, 0.2 mg/kg, and 1 mg/kg ISIS 25302. DAI was not reduced when the dose of ISIS 25302 was increased to 5 mg/kg, however. DAI was not significantly changed when mice were dosed with either 1 or 5 mg/kg of the 8-base mismatch control oligonucleotide, suggesting that the effects of ISIS 25302 on DAI were sequence-dependent.

View larger version (25K): [in this window] [in a new window]   Fig. 6.   Sequence-dependent prevention of acute DSS colitis by antisense oligonucleotide to mouse TNF-. A, disease activity index. B, colon length. Results are expressed as the mean ± S.E.M., where n = 7. ISIS 25302 is the anti-murine TNF- oligonucleotide. The mismatch control oligonucleotide is ISIS 30782, which contains eight mismatched bases compared with ISIS 25302. Asterisk indicates significant differences from saline and mismatch control groups (p < 0.05).

The improvements in DAI with 0.04 to 1.0 mg/kg doses of ISIS 25302 were mirrored by changes (increases) in colon length. The 5.0 mg/kg dose of ISIS 25302 also resulted in an increase in colon length compared with saline-treated mice, yet this same dose had no effect on DAI. Neither dose of mismatch control oligonucleotide had a significant effect on colon length.

TNF- Antisense Activity in Chronic DSS Colitis Model

A chronic form of colitis may be induced when DSS is given repeatedly, in two cycles of DSS-supplemented water followed by normal drinking water. Clinical symptoms (DAI) and colon histopathology wax and wane with the first administration of DSS-supplemented water, as they do in the acute DSS model, but they remain elevated/abnormal for a prolonged period following a second cycle of 4% DSS (Cooper et al., 1993).

In this model, mice were dosed for approximately 3 weeks with ISIS 25302, beginning on day 2 of the first DSS cycle. A dose-dependent reduction in DAI with ISIS 25302 treatment was observed (Fig. 7, A and B). By the end of the second cycle of DSS administration, DAI was reduced 49% in mice given the lowest dose (0.25 mg/kg) of ISIS 25302 when compared with saline-treated controls. DAI was reduced 87 and 86%, respectively, in mice treated with 2.5 and 12.5 mg/kg ISIS 25302. In comparison, animals given two injections of TNF- mAb had a 61% reduction in their DAI.

View larger version (34K): [in this window] [in a new window]   Fig. 7.   Efficacy of ISIS 25302 in chronic model of DSS colitis. Animals were dosed s.c. for four consecutive days at the beginning of the second day of the first cycle of DSS administration, and then every other day thereafter at the indicated doses of ISIS 25302, the murine TNF--specific oligonucleotide. Injections (15 µg) of TNF- mAb were given at the beginning of each of the two DSS cycles. A, mean DAI for each group over the course of two cycles of DSS exposure. B, mean DAI at representative cycle times. C, histopathology scores at termination of study. Acute inflammatory infiltrates contain granulocytes, lymphocytes, and plasma cells, whereas chronic inflammatory infiltrates contain monocytes and macrophages in addition to the previous cell types. Scoring protocol is detailed under Materials and Methods. D, Northern blot analysis of TNF- mRNA levels in colons at time of sacrifice. Results in B-D are expressed as the mean ± S.E.M. Asterisk indicates statistical significance compared with the vehicle-treated group (p < 0.05).

Clinical improvements were mirrored by improvements at the histological level (Fig. 7C). Both inflammatory infiltrates and crypt damage scores were ameliorated with ISIS 25302 treatment. For example, mice treated with 2.5 and 12.5 mg/kg ISIS 25302 had, respectively, 43% and 52% reductions in total inflammatory infiltrates, and 43% and 48% reductions in total crypt damage relative to vehicle-treated animals. This represented a slight improvement over that seen with TNF antibody treatment. The proximal region of the colon was more responsive to treatment with ISIS 25302 than the distal region (data not shown). However, the severity of the disease was greater in the distal colon.

By the conclusion of the study, a modest reduction in TNF- mRNA was seen in colon tissue from mice treated with the two highest doses of antisense oligonucleotide (Fig. 7D). Although not statistically significant, the reduced levels of TNF- mRNA observed for mice treated with the two higher doses of ISIS 25302 support the reduction in DAI seen at these doses.

TNF- Antisense Activity in IL-10 / Mice

IL-10 gene-deficient mice develop a patchy distribution of transmural inflammation, extensive mucosal ulceration, and epithelial hyperplasia reminiscent of human Crohn's disease (Kuhn et al., 1993). The mice are born disease-free but begin to develop histological signs of colitis by 3 weeks of age when raised in conventional animal housing. Mice raised under germ-free or specific pathogen-free conditions, by contrast, display much less severe disease (Sellon et al., 1998). Mice in our colony do not tend to show clinical symptoms of colitis (i.e., weight loss, diarrhea, bloody stools), but by 8 weeks of age, most animals exhibit colitis at the histological level. Disease can then be reliably scored according to the criteria listed in Table 1.

IL-10 / mice underwent both prophylactic and therapeutic treatment regimens with ISIS 25302. In the prophylactic regimen, mice were dosed every other day with oligonucleotides from 2 to 8 weeks of age. The therapeutic regimen lasted half as long as the prophylactic regimen, with dosing from 8 to 11 weeks (i.e., when most of the animals already have histological signs of colitis). One group of mice in the therapeutic regimen received weekly doses of the same anti-mouse TNF- mAb used in the acute and chronic DSS colitis experiments.

Prophylactic Regimen. Over the 6-week treatment period, all groups of IL-10-deficient mice gained weight at a similar rate (data not shown). By the end of the study, mice treated with vehicle displayed a patchy distribution of transmural acute and chronic inflammation, mucosal ulceration, and epithelial hyperplasia in the large intestine. Mice receiving the 0.1 mg/kg dose of the anti-TNF- oligonucleotide, ISIS 25302, demonstrated a marked and statistically significant improvement in their mucosal architecture (Fig. 8A). The mismatch control oligonucleotide did not significantly change intestinal pathology, and neither did any of the other doses of ISIS 25302. 

View larger version (35K): [in this window] [in a new window]   Fig. 8.   Prophylactic treatment regimen in IL-10 / mice. Animals were dosed every other day from 2 to 8 weeks of age with the indicated doses of the murine TNF--specific oligonucleotide, ISIS 25302, or with an 8-base mismatch control oligonucleotide. A, histological scores (see Table 1) in large intestines at termination of study. B-C, cytokine levels in colon organ cultures. Whole colons were obtained from mice at the conclusion of the 6-week study, and levels of cytokines secreted into the culture medium under basal or LPS-stimulated conditions were measured by enzyme-linked immunosorbent assay. B, basal TNF-; C, induced TNF-. Cytokines are expressed as picograms per gram of tissue. Results are represented as mean ± S.D. (n = 6-9). Asterisk indicates a significant difference (p < 0.05) relative to vehicle-treated IL-10 / mice.

Therapeutic Regimen. With a 3-week treatment period begun when mice were 8 weeks old, the 0.1, 1.0, and 10 mg/kg doses of ISIS 25302 all produced statistically significant improvements in histology scores compared with control IL-10 / mice injected with saline (Fig. 9A). None of the groups receiving the mismatch control oligonucleotide had improved histology scores, nor did the animals receiving the TNF- mAb.

View larger version (37K): [in this window] [in a new window]   Fig. 9.   Therapeutic treatment regimen in IL-10 / mice. Animals were dosed every other day from 8 to 11 weeks of age with the indicated doses of the murine TNF--specific oligonucleotide, ISIS 25302, or with an 8-base mismatch control oligonucleotide. An additional treatment group received weekly i.p. injections of 2 mg/kg anti-mouse TNF-. A, histological scores (see Table 1) in large intestines at termination of study. B-E, cytokine levels in colonic organ cultures, measured as discussed for the prophylactic study (see legend to Fig. 8). B, basal TNF-; C, LPS-induced TNF-; D, basal IFN-; E, LPS-induced IFN-. Asterisk indicates a significant difference (p < 0.05) relative to vehicle-treated IL-10 / mice.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

ISIS 25302, the TNF- antisense oligonucleotide selected for these studies, displayed dose-dependent, sequence-specific activity both in vitro and in vivo. Activity disappeared when a series of mismatch control oligonucleotides or a scrambled oligonucleotide were tested, suggesting that the ability of ISIS 25302 to reduce TNF- mRNA occurs through a hybridization-dependent, antisense mechanism.

Systemically administered 2'-methoxyethyl-modified oligonucleotides such as ISIS 25302 are broadly distributed to tissues and organs, including the small and large intestines (Geary et al., 2001). Distribution patterns under pathological conditions such as intestinal inflammation could vary, however, so we examined oligonucleotide uptake in the large intestines of mice with DSS-induced colitis. Immunohistochemical staining showed a greatly enhanced accumulation of oligonucleotide during colitis, especially in areas of the lamina propria populated by an inflammatory cell infiltrate. A previous study, which examined the distribution of first-generation oligonucleotides (non-methoxyethyl-modified) to DSS-inflamed colons also noted enhanced uptake when colons were inflamed (Bennett et al., 1997). This enhanced distribution of oligonucleotides during colitis suggests that antisense therapy could be a reasonable approach to the treatment of IBD.

ISIS 25302 was evaluated in murine colitis models of distinct etiology: DSS-induced colitis and the colitis that spontaneously occurs in IL-10 / mice. DSS colitis involves weight loss, diarrhea, and rectal bleeding in mice, and is characterized by epithelial cell damage, ulceration, and multifocal inflammatory lesions in the colonic mucosa and submucosa (Okayasu et al., 1990; Cooper et al., 1993). Entire crypts may drop out, leaving behind islands of relatively normal-appearing mucosa. This colitis can be induced in both acute and chronic forms, dependent on the number of cycles in which mice receive DSS in their drinking water. It is likely that distinct molecular mechanisms are at work in the acute and chronic forms of DSS colitis; therefore, ISIS 25302 was evaluated in both models.

The acute DSS model may be mediated by local macrophage activation in the colonic mucosa and direct toxic effects of DSS on colonic epithelial cells (Okayasu et al., 1990; Dieleman et al., 1994; Ni et al., 1996). Lymphocytes are not required for induction of acute DSS colitis, since the disease can be induced in SCID mice (Dieleman et al., 1994). The chronic model, however, may be governed by interactions of macrophages with T cells (Dieleman et al., 1998; Shintani et al., 1998; Leung et al., 2000). It is intriguing to note that the immunosuppressant cyclosporin exacerbated acute DSS colitis when administered prophylactically, yet was beneficial if given therapeutically or during the chronic phase of DSS colitis.(Murthy et al., 1993). This suggests that the beneficial effects of cyclosporin may not begin to accrue until lymphocytes become involved in the disease process.

TNF- blockade with ISIS 25302 was effective at ameliorating both the acute and chronic forms of DSS colitis, but the most effective dose differed between the two models. In the acute colitis model, optimal efficacy was obtained with a dose of 1 mg/kg, and efficacy was lost when a higher dose (5 mg/kg) was used. Similarly, a 25-µg dose of TNF- antibody lowered the disease activity index to the same extent as the 1 mg/kg dose of ISIS 25302, whereas a higher dose of antibody had no effect at lowering DAI.

The lack of efficacy at higher ASO doses was seen in an earlier antisense study involving ICAM-1 inhibition in the acute DSS model. In work by Bennett et al. (1997), a 1 mg/kg daily dose of ICAM-1 ASO significantly reduced DAI, whereas higher doses (5-10 mg/kg) lost their efficacy. It is interesting to note that TNF- is well known for its role in inducing ICAM-1 expression on leukocytes and endothelial cells during inflammation (Pober et al., 1987).

In contrast to the acute DSS model, ISIS 25302 displayed true dose-dependent efficacy in the chronic DSS model, with the highest (12.5 mg/kg) dose showing equivalent efficacy to TNF- mAb. The lowest dose of ISIS 25302 used in the chronic study, 0.25 mg/kg, had no efficacy early in the disease, but began lowering DAI toward the end of the study, perhaps as a result of drug accumulation in diseased tissue (Yu et al., 2001). In the chronic study, no doses of ISIS 25302 were effective at reducing DAI during the first cycle of DSS administration, a phase corresponding to the acute form of the disease. However, dosing in the acute DSS studies depicted in Figs. 5 and 6 began 2 days before mice first received DSS in their drinking water, whereas in the chronic study shown in Fig. 7, dosing did not begin until day +2. This suggests that there may be a critical window for preventing the acute form of DSS colitis with TNF- antisense and that TNF- mRNA needs to be down-regulated early in acute DSS colitis for therapy to be effective.

Antibodies targeting TNF- have also been effective in ameliorating DSS colitis, but timing appears to be a critical issue as to whether they are effective. Several studies have shown that TNF- ablation by mAb treatment is effective in the chronic model of DSS colitis (Kojouharoff et al., 1997; Obermeier et al., 1999). TNF- mAb in combination with the TNF release inhibitor pentoxifylline also significantly lowered DAI in the chronic model when given during either one or two cycles of DSS administration, but these agents showed proinflammatory effects if administered during the third cycle of DSS exposure (Murthy et al., 1999).

In several studies involving acute DSS colitis, however (Olson et al., 1995; Kojouharoff et al., 1997), mAb to TNF failed to block or even exacerbated disease. Although we did not see an exacerbation of disease with TNF antibody treatment, the higher (50 µg) dose of antibody we tested failed to lower the disease activity index, whereas the lower (25 µg) dose lowered DAI to about the same extent as the TNF ASO. This suggests that disease protection in the acute form of DSS colitis may be dependent on the degree of TNF inhibition attained. A small amount of TNF may actually be protective in the early stages of disease development. Antisense (or low dose mAb treatment) may not be totally blocking all TNF production and hence could have efficacy in the acute model.

It should be noted that the phosphodiesterase IV inhibitor rolipram has successfully mitigated symptoms in an acute model of DSS colitis. Rolipram is known to be a potent suppressor of TNF- synthesis, and it was shown to suppress colonic TNF levels in mice with DSS colitis (Hartmann et al., 2000).

ISIS 25302 was also tested for its ability to prevent or reverse the colitis that develops spontaneously in IL-10-deficient mice. IL-10 is a potent regulatory cytokine that can suppress both macrophage and T cell activity. It can down-regulate the ability of macrophages to produce proinflammatory cytokines like TNF-, IL-1, IL-6, and IL-12 (Fiorentino et al., 1991), decrease costimulatory molecule and major histocompatibility complex class II expression on antigen-presenting cells, and inhibit T cell proliferation and their secretion of Th1 cytokines like IFN- (Moore et al., 1993). IL-10 knockout mice develop a chronic enterocolitis similar in many respects to human Crohn's disease, with patchy, transmural inflammation, mucosal ulceration, and a cytokine profile that is characteristic of a Th1 immune response (Kuhn et al., 1993; Davidson et al., 2000). A dysregulated immune response to intestinal bacteria is probably involved in the initiation and progression of IL-10 / colitis (Kuhn et al., 1993; Madsen et al., 1999, 2000; Kullberg et al., 2001).

Whereas a single dose of ISIS 25302 (0.1 mg/kg) lowered colitis histology scores in the 6-week prophylactic study in IL-10 / mice, multiple doses (0.1-10 mg/kg) proved effective in the 3-week therapeutic treatment regimen. The difference in effective doses in the prophylactic versus therapeutic studies could reflect not only the length of treatment, but the pathology and immune cell profile present in colons at the time of treatment.

Doses of ISIS 25302 effective at improving colon histology also tended to be effective at reducing cytokine secretion from treated IL-10 / colons placed into organ culture. Effective doses of ISIS 25302 lowered not only basal and LPS-stimulated TNF- secretion, but IFN- secretion as well. The IFN- effects probably occurred through a reduction in recruitment and activation of IFN--producing CD4+ T cells to inflamed colons. TNF- is known to activate expression of key inflammatory intermediates that promote this process, including expression of cell adhesion molecules, chemokines, and other proinflammatory cytokines (Van Deventer, 1997).

It is worth noting that the highest dose of the control oligonucleotide had nonspecific effects on lowering basal and LPS-induced IFN- secretion, whereas it had no significant effect on histology scores or basal or LPS-induced TNF- secretion. The reason for this discrepancy is unclear but may be related to nonspecific immune-stimulatory effects of high doses of oligonucleotides. Oligonucleotides may increase IL-12 secretion by macrophages, and this, in turn, could result in IFN- induction in gut-infiltrating T lymphocytes (Chan et al., 1991; Zhao et al., 1997).

TNF inhibition has proven effective in several other models of colitis (Powrie et al., 1994; Neurath et al., 1997; Watkins et al., 1997; Mackay et al., 1998; Videla et al., 1998), but to our knowledge, this is the first study to uncover a beneficial role for TNF--targeted therapy in the IL-10 / colitis model. In fact, two previous studies that examined the effect of mAbs to TNF- in this model failed to show any amelioration of disease (Davidson et al., 1998; Kullberg et al., 2001), although they did indicate a role for IFN- in the inductive phase of the disease and a role for IL-12 in both inducing and maintaining colitis. In support of these previous studies, we also evaluated a mAb to TNF- in our therapeutic treatment protocol and found it to be without effect on histology scores, although it was able to reduce LPS-stimulated (but not basal) TNF- and IFN- secretion in colon organ cultures.

The reasons for the discrepancy between antisense and antibody therapies targeting TNF- in this model are unclear but could have to do with the degree to which TNF is inhibited by the different therapies. Perhaps there is a critical degree of TNF ablation in this model that is achieved by antisense- but not antibody-based treatments. Alternatively, a regulatory cell required for amelioration of colitis could be deleted by a mechanism involving antibody-dependent cellular cytotoxicity and thus could render antibody therapy ineffective.

The IL-10 and DSS colitis models have a distinct etiology, yet the work reported here indicates that TNF- may be critical to the pathogenesis of each. Similarly, for human inflammatory bowel diseases, it could be envisioned that although the initial insult to the gut may vary, there is a final common immunopathologic pathway in which TNF- plays a critical role. The recent success of the TNF- mAb infliximab in clinical trials in patients with Crohn's disease certainly lends weight to this hypothesis (Stack et al., 1997; Present et al., 1999).

An antisense-based approach to TNF- blockade is an intriguing alternative to antibody-based therapies. Clinical experience with MOE gapmer oligonucleotides such as the one used in the current studies has so far revealed few dose-limiting toxicities (Henry et al., 2000). Neutralizing antibody responses are not mounted against antisense oligonucleotides, so they can be repeatedly administered. An antisense oligonucleotide targeting the adhesion molecule ICAM-1 has already been tested clinically in patients with Crohn's disease, where it proved to be both well tolerated and capable of producing long-lasting disease remissions (Yacyshyn et al., 1998). The 2'-MOE gapmer chemistry used in the current studies provides both increased potency and duration of action over the chemistry used in the ICAM-1 antisense trials, and it also has fewer nonspecific immunostimulatory side effects than the first-generation chemistry (Henry et al., 2000). The efficacy of ISIS 25302 in diverse murine models of colitis suggests that antisense therapy targeting TNF- mRNA could be a viable approach to the treatment of human inflammatory bowel disease.