Toxicology Of Oligonucleotides

Phosphorothioate oligodeoxynucleotides have been examined extensively in a full range of acute, chronic, and reproductive studies in rodents, lagomorphs, and primates. At high doses, there is a distinctive pattern of toxicity that is common to all phosphorothioate oligodeoxynucleotides (146-149). The remarkable similarity in toxicity with different phosphorothio-ate oligodeoxynucleotides suggests that, for this class of antisense compounds, toxicity is independent of sequence and is the result of nonantisense-mediated mechanisms. The most probable mechanism of the observed toxicities is the binding of oligodeoxynucleotides to proteins. These nonantisense-me-diated pathways are believed to be responsible for most, if not all, of the toxicities associated with the administration of these compounds to laboratory animals. This conclusion is strengthened by studies in which little or no differences in toxicity are observed between pharmacologically active and inactive sequences. Different patterns of toxicity exist between rodents and primates. Understanding the mechanisms behind these differences is crucial to understanding which species best predicts the potential human effects. A comparison of the toxicological profiles of phosphorothioate oligode-oxynucleotides with that of the next generation of phosphoro-thioate oligonucleotides suggests that some of the chemical class-related toxicities of phosphorothioate oligodeoxynucle-otides can be ameliorated by chemical modification.

A number of phosphorothioate oligodeoxynucleotides have been examined in 1 or more of the following battery of genotoxicity assays: Ames test, in vitro chromosomal aberrations, in vitro mammalian mutation (HGPRT locus and mouse lymphoma), in vitro unscheduled DNA synthesis tests, and in vivo mouse micronucleus. In all these assays, the results were negative and there was no evidence of mutagenicity or clasto-genicity of these compounds (150).

A. Acute Toxicities

In rodents, the acute toxicity of phosphorothioate oligodeoxynucleotides has been characterized as part of an effort to determine the maximum tolerated dose for in vivo genotoxicity assays. The doses of 3 phosphorothioate oligodeoxynucleotides required to produce 50% lethality (LD50) were estimated to be approximately 750 mg/kg (150).

In primates, the acute dose-limiting toxicities are a transient inhibition of the clotting cascade and the activation of the complement cascade (146,151,152). Both of these toxici-ties are believed to be related to the polyanionic nature of the molecules and the binding of these compounds to specific protein factors in plasma.

Prolongation of clotting times following administration of different phosphorothioate oligodeoxynucleotides is charac-

terized by a concentration-dependent prolongation of activated partial thromboplastin times (aPTT) (149,153-155). The prolongation of aPTT is highly transient and directly proportional to plasma concentrations of oligodeoxynucleotide and therefore parallels the plasma drug concentration curves with various dose regimens. As drug is cleared from plasma, the inhibition diminishes such that there is complete reversal within hours of dosing. With repeated administration, there is no evidence of residual inhibition. Prolongation of aPTT has been observed in all species examined to date, including human, monkey, and rat. The mechanism of prolongation of aPTT by phosphorothioate oligodeoxynucleotides is believed to be a result of the interaction of the oligonucleotides with proteins. It is well known that polyanions are inhibitors of clotting, and phosphorothioate oligodeoxynucleotides may act through similar mechanisms. If these oligonucleotides inhibit the clotting cascade as a result of their polyanionic properties, then binding and inhibition of thrombin would be a likely mechanism of action. However, the greater sensitivity of the intrinsic pathway to inhibition by phosphorothioate oligode-oxynucleotides suggests that there are other clotting factors specific to this pathway that may also be inhibited. Recent data suggest that there is a specific allosteric inhibition of the tenase complex as well as binding to thrombin (152,156).

In clinical trials with ISIS 2302, normal volunteers and patients were dosed with 2 mg/kg infused over 2 h. This regimen produced total oligonucleotide concentrations of 10 to 15 ^g/mL and a concomitant increase in aPTT of approximately 50% (130), which correlates well with in vitro human and animal data. The transient and reversible nature of aPTT prolongation, combined with the relatively small magnitude of the change, makes these effects clinically insignificant for the current treatment doses and regimens.

Activation of the complement cascade by phosphorothioate oligodeoxynucleotides has the potential to produce the most profound acute toxicological effects. In primates, treatment with high doses over short infusion times resulted in marked hematological effects and marked hemodynamic changes that are believed to be secondary to complement activation. Hema-tological changes are characterized by transient reduction in neutrophil counts, presumably due to margination, followed by neutrophilia with abundant immature, nonsegmented neu-trophils (147,151). In a small fraction of monkeys, complement activation was accompanied by marked reductions in heart rate, blood pressure, and subsequently cardiac output. In some animals, these hemodynamic changes were lethal (146,151,157).

There is an association between cardiovascular collapse and complement activation. That is, all monkeys demonstrating some degree of cardiovascular collapse or hemodynamic changes had markedly elevated levels of complement split products. However, the converse is not true, in that only a fraction of the animals with activated complement had cardiovascular functional changes (150). Thus, this observation suggests that there may be sensitive subpopulations or predisposing factors within individual animals that make them susceptible to the physiological sequelae of complement acti vation. Because of these observed hemodynamic changes, primate studies to monitor for these effects have become part of the normal evaluation of these compounds (158,159). Although complement activation at high doses is consistent and predictable between animals, there is currently little appreciation for the variability in the severity of the associated hemo-dynamic changes. Although the split product Bb can be used to monitor complement activation, it is C5a (complement split product) that is the most biologically active split product. Preliminary data obtained relating response to complement split product levels indicate that C5a levels are elevated more significantly in some of the more affected animals (150).

The goal of toxicity studies is to characterize the toxicity of compounds and to establish a framework upon which clinical safety studies can be designed. In this regard, it is useful to examine the relationship between plasma concentrations of oligonucleotides and the activation of complement. When Bb concentrations were plotted against the concurrent plasma concentrations of oligodeoxynucleotides in primates, it was apparent that complement was only activated at concentrations of phosphorothioate oligodeoxynucleotides that exceed a threshold value of 40 to 50 ^g/mL (151). Bb levels remained unchanged from control values at plasma concentrations below the threshold. Remarkably, this threshold concentration is similar for three 20-mer phosphorothioate oligodeoxynucle-otides and for an 8-mer phosphorothioate oligodeoxynucleotide that forms a tetrad complex (160,161). Recent data demonstrate that human serum may be less sensitive to activation than monkey serum, suggesting a species difference in sensitivity. Regardless of small differences, it is clear that clinical dose regimens should be designed to avoid plasma oligodeox-ynucleotide concentrations that exceed 40 to 50 ^g/mL. To this end, the similarities in plasma pharmacokinetics between monkeys and humans have allowed the design of dose regimens that achieve desired plasma concentration profiles.

The most direct approach for staying below the plasma thresholds for complement activation is to reduce the dose rate by substituting prolonged infusions for bolus injections. In clinical trials with phosphorothioate oligodeoxynucleo-tides, the drugs are administered either as 2-h infusions or as constant 24-h infusions. At a rate of infusion of 2 mg/kg over 2 h, the Cmax was 8 to 15 ^g/mL, still well below the threshold for complement activation (130). Phosphorothioate oligo-deoxynucleotides have been administered by intravenous infusion to more than 3000 patients and volunteers without any significant indication of activation of the alternative complement cascade.

1. Modified Oligonucleotides

Chemical modifications to phosphorothioate oligodeoxynu-cleotides may reduce the potential to activate complement. In one study, cynomolgus monkeys were administered an intravenous infusion over a 10-min period with a 5, 20, or 50 mg/ kg dose of a 17-mer phosphodiester oligodeoxynucleotide, Ar177, that had phosphorothioate caps on the 3' and 5' termini (154,162). This oligonucleotide is known to have a complex secondary structure. In this experiment, although there was a dose-related increase in plasma concentrations of Bb, the magnitude of the increases were small in comparison to the known activity of full-phosphorothioate oligodeoxynucleotides (162). Whether this diminished potential to activate the complement cascade is related to the reduction of phosphoro-thioate linkages or whether it is due to the complex secondary structure of this particular oligodeoxynucleotide was not established by these experiments. Some insight into this question was obtained in a second series of experiments performed with oligonucleotides that contained 2'-O-methoxyethyl modifications of the ribose sugar in 12 of the 20 nucleotides (149,150). Cynomolgus monkeys were treated by 10-min intravenous infusion with single doses of 1, 5, or 20 mg/kg of this 20-mer oligonucleotide that was either fully modified phosphorothioate linkages (ISIS 13650) or had phosphodiester wings and a central region of phosphorothioate linkages (9 linkages, ISIS 12854). The termini of both compounds contained six 2'-modified nucleotides. A third unmodified phosphorothioate oligodeoxynucleotide, ISIS 1082, was included as a positive control. The unmodified compound produced marked increases in Bb and severe cardiovascular effect at the dose of 5 mg/kg (30- to 60-fold over baseline). At 5 mg/ kg, the aPTT values were 41 and 33 sec for the fully phospho-rothioate and partially phosphorothioate 2 -modified oligonu-cleotides, respectively. In contrast, the unmodified phosphorothioate oligodeoxynucleotide produced an aPTT of 72 sec at the same dose. These data suggest that reduction in the number of phosphorothioate linkages reduced the inhibitory effects on aPTT and the activation of the complement cascade. However, the more important difference was that both 2'-O-methoxy-ethyl compounds were markedly less potent in activating complement than an unmodified oligodeoxynucleotide (D.K. Monteith, P.L. Nicklin, and A.A. Levin, unpublished observations, 1997). Although the safety profile of phosphorothioate oligodeoxynucleotides has proven satisfactory, the acute safety profile of the next generation of oligonucleotides may be improved by modification of the 2 -position of the ribose sugar with an alkoxy such as 2'-O-methyl or 2'-O-methoxy-ethyl and by reductions in phosphorothioate linkages.

B. Toxicological Effects Associated with Chronic Exposure

One of the characteristic toxicities observed with repeated exposure of rodents to phosphorothioate oligodeoxynucleo-tides is a profile of effects that can be described as immune stimulation. The profile is characterized by splenomegaly, lymphoid hyperplasia, and diffuse multiorgan mixed mononu-clear cell infiltrates (149). The severity of these changes is dose dependent and most notable at doses equal to or exceeding 10 mg/kg. The mixed mononuclear cell infiltrates consisted of monocytes, lymphocytes, and fibroblasts and were particularly notable in liver, kidney, heart, lung, thymus, pancreas, and periadrenal tissues (148,163-165).

Although immune stimulation in rodents is believed to be a class effect of phosphorothioate oligodeoxynucleotides and not dependent on hybridization, sequence is an important fac tor in determining immunostimulatory potential (166-169). Immunostimulatory motifs have been described in the literature and involve palindromic sequences and CpG (cytosine-guanosine) motifs (169).

Among the most remarkable features of oligodeoxynucleo-tide-induced immune stimulation are the species differences. Rodents are highly susceptible to this generalized immune stimulation, whereas primates appear to be relatively insensitive to the effect at equivalent doses. Even 6 months of treatment of cynomolgus monkeys with 10 mg/kg of a 20-mer oligodeoxynucleotide, ISIS 2302, given every other day produced only a relatively mild increase in B cell numbers in spleen and lymph nodes of the primates with no change in organ weights. The mixed mononuclear cellular infiltrates in liver and other organs that are so characteristic of the response in rodents are absent even after long-term exposure in monkeys (149). It is known that rodents are more susceptible to the stimulatory effects of lipopolysaccharides, and much of the immune stimulation produced by oligodeoxynucleotides shares characteristics with lipopolysaccharide stimulation. Assuming results obtained in monkeys can be used to predict stimulation in humans, then the immunostimulatory effects may not be a prominent adverse effect in humans.

It is evident that there are both species and sequence differences involved in immune stimulation and that specific sequences should, if possible, be excluded from oligodeoxynu-cleotides. In long-term toxicity studies in rodents, the constant cell proliferation associated with immune stimulation may have promoter-like effects and may thus complicate the interpretation of rodent carcinogenicity studies. At this time, there are no reports of toxicity studies longer than 6 months, and the long-term sequelae of immune stimulation in rodents are at present merely speculation. More important, immune stimulation following systemic administration of phosphorothioate oligodeoxynucleotides does not appear to be clinically relevant.

Morphologic changes in the bone marrow of mice were observed after 2 weeks of treatment (3 doses/week) with 100 to 150 mg/kg phosphorothioate oligodeoxynucleotide There was reduction in number of megakaryocytes that was accompanied by a reduction of approximately 50% in circulating platelet counts (164). Reductions in platelets have been observed in rats treated with 21.7 mg/kg ISIS 2105 given every other day (148), but were not observed in primates administered 10 mg/kg. Similarly, a reduction in platelets was observed in mice, but not in monkeys treated for 4 weeks with ISIS 2302 at doses of 100 and 50 mg/kg every other day, respectively. Similar observations were made for ISIS 5132 with reductions in platelets at 20 and 100 mg/kg in mice and no observed effect in monkeys up to 10 mg/kg (163). These data suggest that the mouse may be more sensitive to these subchronic effects on platelets than nonhuman primates. However, in acute studies in primates, transient reductions in platelets are occasionally observed. These transient reductions in platelets occur acutely during 2-h infusions at doses of 10 mg/ kg, reverse after completion of the infusion, and have not been associated with any measurable change in platelet number 24

to 48 h after subchronic or chronic treatment regimens (150). Thrombocytopenia has been reported in AIDS patients treated with GEM 91, a 27-mer phosphorothioate oligodeoxynucleo-tide (170).

Tissue distribution studies have shown that the liver and kidney are major sites of deposition of phosphorothioate oli-godeoxynucleotide. In toxicity studies with phosphorothioate oligodeoxynucleotides, a variety of hepatic changes have been observed. The immune-mediated cellular infiltrates in rodent livers were discussed above. With high-dose administration of oligodeoxynucleotides in all species examined, there was a hypertrophic change in Kupffer cells accompanied by inclusions of basophilic material that was observed with hematoxylin and eosin staining. These basophilic granules have been identified as inclusions of oligodeoxynucleotide (114). Furthermore, it was demonstrated that the presence of these inclusions was related to dose.

Hepatocellular changes were not a prominent feature of toxicity in primates. In cynomolgus monkeys, 50 mg ISIS 2302 per kg administered every other day for 4 weeks by intravenous injection produced no morphologic indication of liver toxicity, although there was a slight (1.5-fold) increase in AST in this group (171). Following subcutaneous doses of ISIS 3521 and ISIS 5132 of up to 80 mg/kg every other day for 4 doses, there was Kupffer cell hypertrophy and periportal cell vacuolation, but no indication of necrosis and only a very slight increase in ALT (150). After 4 weeks of alternate-day dosing with 10 mg/kg via 2-h intravenous infusion of either ISIS 3521 or ISIS 5132, there were no alterations in AST or ALT, suggesting that at clinically relevant doses of these compounds, there was no evidence for hepatic pathology or tansaminemia. In clinical trials with ISIS 2302, ISIS 3521, and ISIS 5132 at doses of 2 mg/kg administered by 2-h infusion on alternate days for 3 to 4 weeks, there was no indication of hepatic dysfunction, nor was there any evidence of trans-aminemia.

Like Kupffer cells in the liver, renal proximal tubule epithelial cells take up oligodeoxynucleotide, as demonstrated by autoradiographic studies and immunohistochemistry as discussed previously (114,118,172,173) and by the use of special histologic stains (147). The appearance of basophilic inclusions is dose dependent in proximal tubule cells. Significant renal toxicity can be induced by extremely high doses. Doses of 80 mg/kg in rats and monkeys have induced both histologic and serum chemistry changes in the kidney (174). At clinically relevant doses, however, there was no indication of renal dysfunction. In 4-week or 6-month toxicity studies with phospho-rothioate oligodeoxynucleotides, we observed a much more subtle type of morphologic change in the kidney. At a dose of 10 mg/kg on alternate days, here was a decrease in the height of the brush border and enlarged nuclei in some proximal tubule cells. These changes have been characterized as minimal to mild tubular atrophic and regenerative changes. At a dose of 3 mg/kg and below, these changes were only infrequently observed, if at all.

An important aspect of dose-dependent effects is characterization of exposure concentrations and their relationship to morphological changes. To assess exposure, concentrations of oligodeoxynucleotides have been measured in the renal cortex obtained in subchronic and chronic toxicity studies. Renal concentrations increase with increasing doses. The concentration of total oligodeoxynucleotide in the renal cortex associated with minimal to mild (although not clinically relevant) renal tubular atrophy or regenerative changes is approximately 1000 ^g/g of tissue. The cortex concentrations of total oligodeoxynucleotide that are associated with moderate degenerative changes after subcutaneous doses of 40 to 80 mg/ kg are greater than 2000 ^g/g. At a clinically relevant dose of 3 mg/kg every other day, the steady-state concentration of total oligodeoxynucleotide in the kidney is in the range of 400 to 500 ^g/g, thus demonstrating a significant margin of safety between the clinical doses and those doses associated with even the most minimal morphologic renal changes. Application of clearance and steady- state pharmacokinetic models suggests that continued administration of oligodeoxynucleo-tide at this dose should never achieve the renal concentrations associated with dysfunction (129). These models have been confirmed in 6-month chronic toxicology studies, where tissue concentrations measured at the end of 6 months of every-other-day dosing was no different than levels observed after 4 weeks of dosing at a similar or equivalent dose.

C. Chemical Modification of Oligodeoxynucleotides

Chemical modifications of oligodeoxynucleotides have been shown to reduce the potency of immune stimulation. The simplest modification with remarkable activity for reducing the immunostimulatory effects of oligodeoxynucleotides is the replacement of cytosine with 5-methyl cytosine. The methyla-tion of a single cytosine residue in a CpG motif reduced [3H]uridine incorporation and IgM secretion by mouse spleno-cytes. Methylation of a cytosine not in a CpG motif did not reduce the immunostimulatory potential (175). In our experience with mice, when sequences with 5-methyl cytosine are compared with the same sequence without methylation, the methylated sequence has a lower potency for inducing immune stimulation, as determined by spleen weights and immune cell activation (176,177).

Substitution of methylphosphonate linkages for phospho-rothioate linkages on each of the 3' and 5' termini have also been reported to reduce the proliferative effects and the secretion of IgG and IgM compared 2 with the full phosphorothio-ate analog (178). This suggests that that this modification can also be used to ameliorate immune stimulation. The addition of 2'-0-methyl substituents also reduced immunostimulatory potential (178). The relative contribution of the uridine substitution and the 2'-methoxy substitution could not be differentiated in this experiment. The effect of 2'-alkoxy modifications on immunostimulatory potential needs further investigation. Finally, the effects of chemical modifications of phosphoro-thioate oligonucleotides on renal and hepatotoxicity are currently being investigated.

0 0

Post a comment