The Role of Ribavi~n-Induced Mutagenesis in HCV Therapy: A Concept or a Fact?

I See Article on Page 869 I ibavirin is currently approved by the FDA to treat infections by respiratory syncytial virus and hep- R atitis C virus (HCV). Ironically, ribavirin remains

the single most promising broad-spectrum antiviral agent more than 30 years after its discovery. For many life- threatening viral infections, such as Lassa fever and hem- orrhagic fever (caused by members of the Bunyaviridae or Arenaviridae family), compassionate use of intravenous ribavirin is the only choice between life and death. Prior to 1998, the dominant theory of the antiviral efficacy of ribavirin was based on the depletion of nucleotide pools in infected cells via inhibition of the host enzyme inosine monophosphate dehydrogenase.’ This theory was re- cently challenged in view of its poor correlation with the antiviral properties of ribavirin against HCV and related vir~ses.~,3 Clinical observations suggested that ribavirin has a unique antiviral profile in combination with inter- feron, acting via a reduction in relapse rates to interferon therapy.*,5 This fact gave rise to newer theories, one of which postulated an enhanced cellular immune response as a result of the drugs effects.’ While there is compelling evidence suggesting that ribavirin in combination with interferon helps to restore the immune response,’ the fo- cus of this editorial is the assessment ofyet another theory, namely that lethal mutagenesis induced by ribavirin that leads to the “genetic meltdown” of RNA viruses.6

The initial finding of a promiscuous incorporation of ribavirin triphosphate via a viral RNA polymerase led to the hypothesis that ribavirin may be a mutagen to HCV and related RNA viruses.* Craig Cameron and colleagues later confirmed that once incorporated into the nascent RNA chain, the ribavirin base could template the incor- poration of both uridine monophosphate and cytidine monophosphate, thereby inducing muta t i~ns .~J Indeed, ribavirin was found to increase the mutation rate, as scored by the plaque-forming ability of polioviruses (PV)

Abbreviations: HCI.: bepatitir C virus; PI.: poliovirus. From Ribapbarm Inc., Costa Mesa, CA. Aah!ress reprint requests to: Zbi Hong, Pb.D., Ribapbarm Inc., 3300 Hyland

Ave., Costa Mesa, CA 92626. E-mail: zbibong@icnpbarm.com; fm:

Copyright 0 2003 by the American Association for the Study of Liver Diseases.

doi: 10.1053/jhep.2003.50464

714-688-3141.

0270-9139/03l3804-0005$30.00l0

in the presence of 2 mmol/L guanidine.6 The observation that ribavirin is a mutagen of the PV genome provided an experimental approach to test the mutagen theory di- rectly. By growing PV in the presence of increasing con- centrations of ribavirin, a finite number of random mutations were directly determined by sequencing several genome equivalents of capsid-coding sequences. A 4-fold increase in mutation frequency reduces viral fitness 20- fold.8 These data provide convincing evidence for a direct lethal mutagenic effect of ribavirin on the poliovirus ge- nome.

Subsequently, several publications supported the mu- tagen theory based on in vitro studies using HCV repli- cons and other related viruses.3.9-’3 Through large-scale sequencing or functional selection of mutant viruses in the presence of ribavirin, mutation frequencies seen by different groups are summarized in Table 1. At first glance, methods to determine these mutation frequencies vary significantly between investigators. The viral system, ribavirin dose, time of exposure, and level of selection pressure all contribute to the determination of mutation frequency. To allow a comparison, reported mutation fre- quencies are converted to mutation rates, which factor in the time component for ribavirin exposure. The unit of mutation rate is standardized as mutation per site per year, thereby allowing a direct comparison. However, as a note of caution, several assumptions are made in this conver- sion: (1) mutations in untreated viral genomes represent a baseline frequency to allow the calculation of mutation rates based on 2 time points (time zero and end of treat- ment); (2) the mutation frequency during treatment is linear with time; and (3) the extrapolation to per-year basis does not consider longer-term selection and fixation of a mutation. Clearly, these assumptions can be prob- lematic. For example, the mutation rate based on short- term treatment (10 hours) probably reflects a “true” mutagenic potential without the confounding effects of the elimination of defective mutations or reversions. As shown in Table 1, ribavirin-induced mutagenesis is statis- tically significant in some studies”1° but not in other^.^ In general, the mutation rates range from 8 X 1 0-2 to 1 .O/ site/y and agree with those reported for other RNAviruses such as Sendai virus (3 X 10-5/site/replication cycle) and bacteriophage QP (3 X /site/replication cycle) on a per-year basis (assuming > 1,000 replication ~ycles/y).’~

However, extrapolation of in vitro ribavirin-induced mutagenesis to efficacy in HCV therapy is not obvious.

807

808 HONG HEPATOLOGY. October 2003

Table 1. Summary of Ribavirin-Induced Mutagenesis Reported by Various Investigators Young et aI.17 Zhou et aL3 Pfelffer et aL9 Lanford et al.11-12 Severson et aI.l3 Contreras et al.IO Crow et aL6J

System

Methods of analysis

Exposure to ribavirin

Ribavirin doses Cell lines Mutation

frequency” Mutation ratet

Fold increase in mutation rate

Nonsynonymous vs. synonymous mutations

Ribavirin resistance Comments

Ribavirin monotherapy Replicon replicon

Sequencing Sequencing

36 weeks 3-10 days

10 pmol/L in patients Huh7 Huh7 2.4-2.8 X 10-3/site

6.08 X lW3/si te/y 1.8 X

2-3 2

10-500 pmol/L

5 X 10-3/site

site/y

0.29 NR

F4 15Y NR Polymorphism at codon Mutation by

415 among various ribavirin HCV genotypes increased with

other IMPDH inhibitors

Poilovirus-plaque RepliconjGBV-B assays

Sequencing Functional test functional test (infectivity)

Nine single cycle 24-72 hours passages (9 X 6.5 h)

100-400 pmol/L 50-400 pmol/L HeLa Huh7 NR NR

NR NR

NR >3

<1 NR

G64S NR Mixture of mutant Cumulative effect

and wild-type of ribavirin- viruses in induced resistant pools mutagenesis

Hantaan virus Vaccina/T7 Poliovirus plaque assays

Sequencing Sequencing Sequencing functional test

72 hours 24 hours 10 hours

100 pmol/L 50-400 pmol/L 400 pmol/L Vero E6 CV-l/HepG2 HeLa 8.4 X l V 3 / s i t e 2.2 x 10-4-1.5 7.2 X W4/

l.O/site/y 0.8-5.5 X 6.3 X lo-’/ x 10-3/site site

site/y site/y 8.6 2.2-3.5 5-10

>1 2.2-2.7 NR

NR NR NR lnsettions > Mutation rate not First direct

substitutions dose-dependent; demonstration IFN induces of ribavirin mutations as a

mutagen

Abbreviation: NR, not reported. “The mutation frequency here is ribavirin induced and represents the difference between end-of-treatment frequency minus that of placebo or no drug control. tThe rate is based on 2 time points at the initiation and end of ribavirin treatment. The mutation frequency at time zero or treatment initiation is assumed to be

the same as that of placebo or no drug control.

First, all in vitro studies used high concentrations of riba- virin (100-500 pmol/L) that may not be achieved in the therapeutic setting, where a steady state plasma concen- tration of approximately 10 pmol/L is seen. Second, clin- ical trials using ribavirin monotherapy for up to 24 months have failed to result in significant or lasting anti- viral effects. 15,16 This observation contradicts the cumu- lative nature of ribavirin-induced mutagenesis observed in vitro.8.’ 1 ~ 1 2 Perhaps ribavirin-induced mutations reach a steady state in vivo, counter-balanced by host factors that preclude the triggering of an “error catastrophe.” Ad- ditional synergistic activity from interferon may help to tip the balance and push the virus over the edge to lethal “genetic meltdown. ”

In this issue of HEPATOLOGY, Young et al.17 addressed the mutagen theory by direct sequencing of HCV RNA from patients enrolled in an earlier ribavirin monotherapy multicenter trial. l 5 They compared the mutation fre- quency in the NS5B region at 3 time points: before the initiation of ribavirin treatment, at the end of the 36-week course, and 16 weeks following therapy. Multiple reverse transcription polymerase chain reaction clones obtained from 5 patients infected by genotype l a HCV were se- quenced. Mutations at both the nucleic acid and the

amino acid level were identified. Controls included clones from 2 patients in the placebo group and 8 patients on interferon monotherapy. The investigators concluded that ribavirin monotherapy increased the apparent muta- tion rate by approximately 3-fold (9.13 2 3.09 X site/y vs. 3.05 ? 2.93 X lO-’/site/y); however, the difference was not statistically significant. Interestingly, the mutation rate estimated at 6.08 X lO-’/site/y (after removing the background mutation rate) is substantially lower than those determined in the in vitro studies (Table 1) but rather consistent with the one reported by Ina et al. through phylogenetic relationship analysis (0.22-7.5 1 X 10-’/site/y).’8 The lower mutation rate is likely the result of the selection and elimination of defective mutations during the 36-week treatment period.

Therefore, mutation rates based on the long-term ex- posure in vivo to a mutagen do not reflect the true muta- genic potential. To asses the mutation rate for ribavirin- containing therapies, HCV RNA after short-term exposure to ribavirin should be used. Given the high pro- duction rate and short serum half-life (2.7 hours) for HCV, serum HCV RNA should be isolated shortly after the initiation of ribavirin therapy, preferably at multiple time points. The exact timing should also be a factor in

HEPATOLOGY, Vol. 38, No. 4, 2003 HONG 809

the pharmacokinetic and pharmacodynamic profiles of oral ribavirin.

Perhaps, the most surprising result in the report by Young et al. is the identification of a mutation in NS5B that the investigators claimed to cause resistance to riba- virin. This is a semiconservative mutation in the thumb subdomain that changes phenylalanine at amino acid po- sition 4 15 to tyrosine (F4 15Y). From a structural point of view, this mutation may affect viral replication by steric interference with the formation of nascent RNA product. Young et al. further tested their finding in the replicon system and showed that a replicon bearing a phenylala- nine at position 415 is more susceptible to ribavirin inhi- bition than that with a tyrosine at the same position. Given that many have tried without success to isolate ribavirin-resistant viruses, this claim may garner its own criticisms, the foremost being that polymorphism exists at this mutation site. Almost all HCV genotypes encode a tyrosine at this position except for isolates in genotype l a and 6a, which have a phenylalanine at this site. If the tyrosine at position 415 is responsible for resistance to ribavirin, HCV with genotypes other than l a and 6a would be less susceptible to ribavirin therapy, which has not been shown to be the case in clinical studies. Coinci- dentally, another group at Stanford University reported the identification of a ribavirin-resistant mutation in po- liovirus 3D polymerase at amino acid position 64 (from glycine to serine, G64S, refer to Table 1). They propose that the mechanism of ribavirin resistance is likely due to an increase in fidelity of the polymerase that reduces the frequency of ribavirin misincorporation.9 Interestingly, the G64S mutation is located in the fingers subdomain according to the computer model for poliovirus 3D poly- merase. This is structurally distinct from the correspond- ing F415Y in HCV NS5B. Although there is a lack of structural correlation between the 2 reports for ribavirin resistance, a more general mechanism may explain the apparent ribavirin resistance. If mutation F415Y or G64S somehow reduces the rate of the nucleotidyl transfer re- action and increases the residence time of a nucleotide at the active site, this may allow the polymerase extra time to select for the cognate substrates and reduce the frequency of misincorporation, thereby increasing the fidelity of the polymerase. Future biochemical and molecular studies may support or disprove the role of these mutations in relation to ribavirin resistance.

One intriguing observation’’ was that interferon could also increase the mutation frequency in the HCV ge- nome. Young et al. showed a numerical increase in muta- tion frequency in the interferon treatment group (5.87 ? 4.29 X 10-3/site/y), but it was not statistically significant compared with that in the placebo group (3.05 * 2.93 X

1 OP3/site/y). A brief literature review reveals that inter- feron has pleiotropic activities affecting many cellular processes, including RNA editing through the induction of RNA-specific adenosine deaminase (ADARl). ‘9 This enzyme converts the adenosine base in the RNA strand to that of inosine, thereby inducing A to G transition muta- tions. While further studies are required to confirm the potential of interferon to increase the mutation frequency in the replicating HCV genome, it is possible that inter- feron in combination with ribavirin may additively or synergistically increase the mutation frequency beyond the sublethal level induced by ribavirin alone. Ribavirin alone is unable to drive the virus to extinction and re- quires interferon to further compromise viral replication and break through a “protective barrier” to lethal mu- tagenesis.

The fundamental question is whether the primary mode of action of ribavirin is through the induction of lethal mutagenesis. The answer at present is that we don’t know. Young et al. suggest that ribavirin may be a weak mutagen based on their limited studies with 5 patients infected with HCV of genotype la. This argument may have been complicated by the observation of a ribavirin- resistant mutation (F415Y) in NSSB, which, in theory, should result in no increase in mutation frequency at week 36. Although this report fails to provide convincing evi- dence to indicate the lethal mutagenesis as a therapeutic mechanism of ribavirin, it clearly beckons for additional studies. Many investigators believe in nucleotide-based mutagenesis as a potential effective antiviral strategy against RNA viruses as these organisms exist so near their error thresholds.* Substantial therapeutic selectivity may be achieved because the mutagenic effect on host RNA polymerase is likely to be terminal and noncumulative. The tolerance of host RNA to mutagens is likely to be greater than the error thresholds of RNA viruses. The traditional drug discovery paradigm has excluded drugs with mutagenic potential based on their effect on DNA/ chromosome replication and repair. This certainly has its merit in that heritable genetic alterations must be avoided. It is therefore imperative that an RNA mutagen not be converted to a DNA mutagen in the cell.

A viral population with a spectrum of sequences rather than a uniform sequence has the advantage ofviral adapt- ability and fitness? Survival as a population at the ex- pense of individual viral variants is a strategy mastered by RNA viruses throughout evolution. Mutations that in- crease polymerase fidelity and consequently decrease qua- sispecies diversity should decrease viral fitness. A potent RNA mutagen will select for individual viruses that ac- quire resistance to the lethal mutagen by deploying a more faithful polymerase but concomitantly reduce the viral

810 HONG HEPATOLOGY, October 2003

quasispecies and population fitness. As a result, the viruses 8. Crotty S, Cameron CE, Andino R. RNA virus error catastrophe: direct molecular test by using ribavirin. Proc Natl Acad Sci U S A 2001;98:6895- will be in double jeopardy, facing the assault of a direct 6900,

antiviral agent and the hostile host environment. Future studies on the mechanisms of ribavirin resistance, similar to those reported by Young et al., may provide helpful insight for the development of lethal mutagens as curative antiviral agents.

9,

10.

11. ZHI HONG, PH.D. Ribapharm Inc, Costa Mesa, CA

References 12.

1. Lau JY, Tam RC, Liang TJ, Hong 2. Mechanism of action of ribavirin in the combination treatment of chronic HCV infection. HEPATOLOGY 2002; 3.

2.

3.

4.

5.

6.

7.

35:1002-1009. Hong Z, Ferrari E, Wright-Minogue J, Skelton A, Glue P, Zhong W, Lau JYN. Direct antiviral activity of ribavirin: hepatitis C virus NS5B polymer- ase incorporates ribavirin triphosphate into nascent RNA products [Ab- stract]. HEPATOLOGY 1999;30(Part 2):354A. Zhou S, Liu R, Baroudy BM, Malcolm BA, Reyes GR. The effect of ribavirin and IMPDH inhibitors on hepatitis C virus subgenomic replicon RNA. Virology 2003;3 10:333-342. McHutchison JG, Gordon SC, Schiff ER, Shiffman ML, Lee WM, Rustgi VK, Goodman ZD, et al. Interferon alfa-2b alone or in combination with ribavirin as initial treatment for chronic hepatitis C. Heparitis Interven- tional Therapy Group. N Engl J Med 1998;339:1485-1492. DiBisceglie A, Hoofnagle J. Optional therapy of hepatitis C. HEPATOLOGY 2002;36:121-127. Crotty S, Maag D, Arnold JJ, Zhong W, Lau JY, Hong 2, Andino R, et al. The broad-spectrum antiviral ribonucleoside ribavirin is an RNA virus mutagen. Nat Med 2000;6:1375-1379. Maag D, Castro C, Hong 2, Cameron CE. Hepatitis C virus RNA-depen- dent RNA polymerase (NS5B) as a mediator of the antiviral activity of ribavirin. J Biol Chem 2001 ;276:46094-46098.

14.

15.

16.

17.

18.

19.

20.

Pfeiffer JK, Kirkegaard K. A single mutation in poliovirus RNA-dependent RNA polymerase confers resistance to mutagenic nucleotide andogs via increased fidelity. Proc Natl Acad Sci U S A 2003;100:7289-7294. Contreras AM, Hiasa Y, He W, Terella A, Schmidt EV, Chung RT. Viral RNA mutations are region specific and increased by ribavirin in a full- length hepatitis C virus replication system. J Virol2002;76:8505-8517. Lanford RE, Guerra B, Lee H, Averett DR, Pfeiffer B, Chavez D, Notvall L, et al. Antiviral effect and virus-host interactions in response to alpha interferon, gamma interferon, poly(i)-poly(c), rumor necrosis factor alpha, and ribavirin in hepatitis C virus subgenomic replicons. J Virol 2003;77: 1092-1 104. Lanford RE, Chavez D, Guerra B, Lau JY, Hong 2, Brasky KM, Beames B. Ribavirin induces error-prone replication ofGB virus B in primary tamarin hepatocytes. J Virol 2001;75:8074-808 I . Severson WE, Schmaljohn CS, Javadian A, Jonsson CB. Ribavirin causes error catastrophe during Hantaan virus replication. J Virol2003;77:481-488. Eigen M. The origin of genetic information: viruses as models. Gene

Bodenheimer HC, Jr., Lindsay KL, Davis GL, Lewis JH, Thung SN, Seeff LB. Tolerance and efficacy of oral ribavirin treatment of chronic hepatitis C: a multicenter trial. HEPATOLOGY 1997;26:473-477. Hoofnagle JH, Lau D, Conjeevaram H , Kleiner D, Di Bisceglie AM. Prolonged therapy of chronic hepatitis C with ribavirin. J Viral Hepat

Young K-C, Lindsay KL, Lee K-J, Liu W-C, He J-W, Milstein SL, Lai MMC. Identification of a ribavirin-resistant NSSB mutation of hepatitis C virus during ribavirin monotherapy. HEPATOLOGY 2003;38:869-878. Ina Y, Mizokami M, Ohba K, Gojobori T. Reduction of synonymous substitutions in the core protein gene of hepatitis C virus. J Mol Evol

Samuel CE. Antiviral actions of interferons. Clin Microbiol Rev 2001;14: 778-809. Doming0 E, Escarmis C, Sevilla N, Moya A, Elena SF, Quer J, Novella IS, et al. Basic concepts in RNA virus evolution. Faseb J 1996;10:859-864.

1993; 135:37-47.

1996;3:247-252.

1994;38:50-56.