Antiviral Drugs and RNA Viruses

The need for antiviral drugs

The majority of known viruses are RNA viruses. It is therefor not surprising that they cause the majority of human, animal and plant viral diseases. The Table shows some of the more familiar diseases caused by RNA viruses.

Examples of diseases caused by RNA viruses
Human Animal Plants
common cold
yellow fever
various hemorrhagic fevers
Newcastle disease
fowl plague
foot-and-mouth disease
many viruses that cause mottling, spotting, striping or wilting of the foliage

As with all infectious disease, the control of RNA virus infection should logically include three strategies:

  1. Public health measures to minimize the risk of infection.

  2. Vaccination protects individuals who are exposed to infectious agents despite the public health measures.

  3. Anti-infective drugs for individuals who are infected despite public health measures and vaccination.

Public health measures are arguably the single most important factor contributing to the improved health and increased longevity of humans. For example,

  • safe drinking water helps to prevent infection by enteric viruses (poliovirus, rotavirus)
  • pest control helps to prevent infection by arthropod-borne or rodent-borne viruses (e.g., yellow fever, dengue, and encephalitis viruses; hantaviruses and arenaviruses)

The major current challenge is to bring these benefits to more than just humans living in the developed world. A second challenge is to discover means by which to control airborne respiratory infections.

As public health measures are never totally effective, vaccination is a second strategy to control infectious diseases. Indeed, vaccines have dramatically decreased the incidence of several hitherto common diseases. Diseases caused by RNA viruses that are controlled by vaccination include:

  • poliomyelitis, yellow fever, rabies, measles, mumps, rubella, hepatitis B, rabies

Research and development of vaccines continue, and a number of potential new vaccines are in the pipeline. Vaccination does have limitations, however. Many diseases are not so amenable to control by vaccination, as the benefits of vaccines need to be balanced by the technical difficulties and social and financial costs of developing safe vaccines and instituting appropriate vaccination programs. For example, the common cold can be caused by any of the more than 100 known strains of rhinoviruses, and it will be quite challenging to develop and test a vaccine that protects against all of them. Many viruses, including the alphaviruses (e.g. O'nyong-nyong virus) and the filoviruses (e.g. Ebola virus), have the potential of causing devastating epidemics, but the outbreaks are sporadic and unpredictable, making it difficult to justify vaccinating entire populations as a precautionary measure, even when effective and safe vaccines can be developed and are available.

Thus, despite our best preventive measures, human and animal infection can and will occur. Bacterial infections are reasonably well controlled by antibiotics. Granted, antibiotic-resistance has become a significant problem, in part because of their misuse. However, it remains true that when a patient with a bacterial infection shows up in a clinic, the physician has any number of very effective antibiotics to prescribe, that in the vast majority of cases take care of the infection. In fact, the success due to the availability of effective vaccines and antibiotics led to the illusion that infectious diseases have been conquered.

Contrast this with a patient with a viral infection. In almost all cases, the physician can offer no better help than age-old nostrums (e.g., get plenty of rest, drink lots of fluids, eat mother's chicken soup), or symptomatic support (from cough suppressants to iron lungs). This has to remain an acute and continuing embarrassment to the biomedical community. More than this, we have little to offer if (when) a virulent epidemic strikes. The developed world has been fortunate for the absence of major viral epidemics (other than HIV infection) during the last 40 years or so. It is not widely known that yellow fever killed about 10% of the population of Philadelphia in 4 months in 1793, and that influenza killed 500,000 Americans in 1917-18. Only with the emergence of the human immunodeficiency virus was our continued vulnerability clearly exposed for all to see. Furthermore, most citizens of the developed countries are not aware of the epidemics still besetting the developing countries. These include the continuing spread of e.g., dengue hemorraghic fever, yellow fever, endemic food-borne or water-borne infections (e.g., rotaviruses), and the explosive alphavirus / encephalitis virus outbreaks in Africa and South America.

Few would argue that we do not need antiviral drugs. How antiviral drugs are best developed is not so clear. It is therefor useful to consider what an ideal drug might look like.

The ideal antiviral drug

Features of an ideal antiviral drug might include the following:

  • Effective inhibition of some essential viral process
  • Drug-resistant viruses do not appear
  • Broad spectrum activity (e.g., a single drug effective against any of the 100+ common cold viruses)
  • No effect on host processes

The first feature is obvious.
The remaining features depend strongly on the viral process being targeted, as will be discussed below.

The viral target, and drug resistance

The virus infection cycle presents many potential targets for inhibition [Sim, 1990]. The viral polymerase is an obvious and the most commonly chosen target. Many different polymerase inhibitors, both nucleoside analogs and non-nucleoside compounds, have been tested, a few are in clinical use. Structural information from X ray crystallography or NMR studies has been quite useful in antiviral drug design, for example, to block viral protein processing by inhibiting the protease of human immunodeficiency virus [e.g., Kaldor, et al., 1997]. These approaches are sometimes difficult, as the aim is to exploit the subtle differences between viral and host enzymes, to inhibit the viral enzyme preferentially. Host toxicity presents a constant challenge in antiviral drug development. Another approach is to block virus invasion of host cells, e.g., by inhibiting the rhinovirus surface protein [Badger, et al., 1988], or the hemagglutinin-neuraminidase of influenza virus [e.g. Gubareva, et al., 1996; Gubareva, et al., 1997; McKimm-Breschkin, et al., 1996; Wade, 1997].

A major problem with current antiviral drugs is the high frequencies of drug-resistant mutants. There are experimental and clinically approved drugs that are very effective, dramatically reducing virus yields, or providing good relief of symptoms in patients, but that ultimately have limited usefulness, because viruses that escape the initial inhibition are frequently drug-resistant. The experience with rimantadine/amantadine in controlling influenza A virus is instructive: drug-resistant viruses arose at high frequencies, were pathogenic, and were transmitted in human and chickens [Hayden and Hay, 1992]. Similarly, human immunodeficiency virus readily develops resistance to a variety of nucleoside and non-nucleoside agents targeted against its reverse transcriptase and protease [Richman, 1993], and there is even evidence of resistance to multiple drug treatment [Gunthard, et al., 1998]. The available evidence show that the respective sites on the viral proteins targeted by the drugs can mutate to drug resistance by one or a few changes. To make matters worse, at least some drug-resistant virus grow almost as well as, and are as pathogenic as the parental, drug-sensitive viruses [e.g., Rayner, et al., 1997].

The rapid development of drug resistance by retroviruses and RNA viruses is due to their high mutation rates. Indeed, it is not unreasonable to propose that the large diversity of RNA viruses is in large part due to their high mutation rates, that allow them to mutate and adapt to new niches very rapidly.

High mutation rates of RNA viruses

RNA viruses and retroviruses generally have very high mutation rates, about 10-3 to 10-5 per nucleotide incorporated [Eigen and Biebricher, 1988; Holland, et al., 1992; Steinhauer and Holland, 1987]. This is up to a million-fold higher than that of organisms with DNA genomes. A priori calculations show that a population of RNA viruses descended from a single parent contains an astonishingly large number of mutants [Eigen and Biebricher, 1988]. For example, 1 ml of media from cells infected with Sindbis virus typically contains 109 plaque forming units of the virus. For a mutation rate of 10-4 per nucleotide and a genome length of 11.7 kb, it can be calculated that the population contains over 300 million different sequences; and every possible point mutation, over 35000 different species in all, will be present in that 1 ml of media. Indeed, there are over 10000 copies of each and every one of the single-base mutants in the same 1 ml of media.

For HIV infections, up to 1010 viruses are produced per day in the patient. As all possible single- and double-mutants will be present in such a large population, drug-resistant viruses are guaranteed to be present even before exposure to the drug, if resistance can be acquired by one or two base changes, [e.g., Coffin, 1995; Tucker, et al., 1998]. Given that there are about 30 million HIV-infected individuals, each producing up to 1010 viruses a day, the global daily HIV load is about 1017. All possible single-, double- and triple-mutants are present in such a population.

Of equal importance is the fitness of the drug-resistant mutants. For example, the fitness of drug-resistant HIV mutants (in the absence of the drug) may be higher than, comparable to, or lower than the wildtype (e.g., Kosalaraksa et al., 1999). Even when the drug-resistant mutants have low fitness (e.g., Verhofstede et al. 1999), they may still grow well enough to continue to be a problem. The initial decrease in virus load during drug treatment simply reflects the disappearance of the drug-sensitive viruses. The drug resistant viruses, although initially at relatively low abundances, may be able to multiply despite the presence of the drug and despite the host responses. Clearly, sustained replication by even low-fitness resistant viruses can result in the acquisition of additional, fitness-improving mutations (also, see Berkhout, 1999 for a discussion).

The implication for antiviral drug design is that the mutability of the viral target, and the fitness of the spectrum of resistant mutants should be prime considerations if the problem of drug-resistance is to be minimized.

Conserved features of viruses as ideal drug targets

Given the rapid emergence of drug resistance due to the high mutation rates of the RNA viruses, it is useful to identify targets that do not or cannot mutate rapidly. There are several potential advantages of targeting the most conserved features of the virus genomes:

  1. Effective inhibition of viral growth.
    Conserved features are likely to play essential roles in the virus lifecycle.

  2. Decreased rates of drug resistance.
    The slow rate of change of the conserved features suggest that they will be slow to change when subjected to drug inhibition, resulting in slow development of drug resistance.

  3. Drug-resistant viruses are 'attenuated'.
    Even if drug-resistant variants arise, they might have low fitness, as the changes leading to drug resistance might not be most consistent with virus growth. Low fitness of the mutants should result in lower virus loads, and thus decreasing the chances for additional fitness-improving mutations to arise and be selected for. A sufficiently low fitness of the resistant mutants may also allow the host responses to be able to effectively control the infection.

  4. Fewer drugs needed.
    Decreased rates of drug resistance suggest that fewer drugs will be needed for effective control of infection.

  5. Broad spectrum drugs.
    Drugs that inhibit the conserved feature of one virus might be effective against related viruses with minimal modification.

Ideally, we should also focus on those conserved targets that allow us to design drugs with minimal effects on host processes, i.e., we should identify virus-specific processes.

Conserved features of RNA viruses

A survey of RNA viruses identified 3 broad classes of highly conserved features:

  1. Conserved protein sequences consisting of the familiar enzyme 'motifs'
    • e.g., polymerase, NTP-binding, helicase, protease, methyltransferase motifs

    The motifs per se are not good targets, as the host has the same motifs. Drugs targeting these motifs are likely to present toxicity problems. Targeting the (typically not well conserved) binding site encompassing the motifs is also non-ideal, as we already know that it mutates too easily, and at least some of the resulting resistant mutants have high fitness.

  2. Conserved protein sequences, frequently of unknown function.
    e.g., conserved sequences among viruses in the same genus; also, those that help define the Sindbis-like, picorna-like, flavi-like superfamilies among the plus-strand viruses [Ahlquist, et al., 1985; Koonin and Dolja, 1993; Strauss and Strauss, 1994].
    Examination of the available structures suggest that their functions include
    • Internal residues essential for , e.g. tight packing, facilitation of confromational changes; or directing proper folding of the proteins
    • Surface residues involved in essential protein-protein or protein-RNA interactions

    Typically, the conserved residues outnumber the variable residues, so there is no lack of potential targets for inhibition. Information on the tertiary structure of the viral proteins is vitally needed, to begin to assess the conserved sequences as targets for antiviral drugs!

    Detailed analyses of the conserved sequences of many virus genera are yet unpublished, but are available on request. Once published, the results will be made available here.

  3. cis-acting sequences
    These are discussed in the next section.

Conserved cis-acting sequences of RNA viruses

The cis-acting sequences of RNA viruses are essential RNA signals

  • at the termini of the viral RNAs, recognized for initiating viral RNA synthesis
  • at internal locations, recognized for initiating and terminating the transcription of viral mRNAs

They typically do not overlap with protein-encoding sequences, but are as well conserved as the amino acid sequence of the polymerases.

Conservation of the cis-acting sequences is seen for many RNA virus families relevant to human and animal disease:

More detailed discussion of the pattern of conservation is on a separate page.

Mechanisms for maintaining cis-acting sequence conservation

A number of models can be envisioned to explain the conservation of the cis-acting sequences, and that are consistent with the available data. For example, one possibility is that the cis-acting sequence plays several, pleiotrophic roles in the virus lifecycle. Waxman and Peck [1998] predict that in such a situation, a single optimal genetic sequence may become common. Our working model is different, but is not inconsistent with this theoretical derivation.

The model

The cis-acting sequences are probably recognized by viral or host proteins during the initiation of viral RNA synthesis. Suppose that the specificity of recognition is determined by a host protein. Since the host evolves at very much slower rates, the host protein remains unchanged for relatively long time spans. The virus, however, mutates at high rates, such that the cis-acting sequence is rapidly selected to achieve an optimal interaction with the host protein. Once this occurs, mutations in the cis-acting sequence will usually be sub-optimal, and be selected against. Thus, the cis-acting sequence will be able to change only at a rate comparable to the cognate host protein.

If, instead, recognition of the cis-acting sequence is mediated by a viral protein, the interaction between the two should also be rapidly optimized [Steinhauer and Holland, 1987]. Once this occurs, the cis-acting sequence and the cognate protein become mutually constrained: neither can change independently without disturbing the optimized interaction. Change is possible only when both mutate coincidentally, and in an exactly compensatory fashion. This will be quite rare.

    Predictions of the model
  1. The recognition of the cis-acting sequence is functionally conserved. The cis-acting sequences of a group of closely related viruses show little divergence, despite divergence in many other properties of the viruses. We therefor expect that the mechanism for recognizing the cis-acting sequence should not have diverged much. The prediction is that the cis-acting sequence of a virus should be recognized efficiently by closely related viruses.

  2. The conserved 'wildtype' sequence is optimal. The model assumes that mutation of the cis acting sequence or its cognate protein is deleterious. This is the same as postulating that the conserved sequence is better than most or all mutants that might arise. Indeed, for the cis-acting sequence to be conserved, it must have been superior to all the mutants that were generated during the evolution of the viruses.

The available evidence are consistent with both predictions.

Inhibiting the recognition of the cis-acting sequences

Inhibition of the recognition of the cis-acting sequence clearly involves two questions: what to target, and how to target it.

  1. Attack the cis-acting sequence itself

      Suggestive evidence
    • Cis-acting sequence is required for initiating viral RNA synthesis
    • Mutations in the cis-acting sequence are deleterious for virus growth

      Potential methods:
    • Ribozymes specific for the cis-acting sequence
    • Conventional nucleic acids (DNA or RNA) as antisense inhibitors
    • Nucleic acid analogs (e.g., peptide nucleic acids) as antisense inhibitors

    • How to deliver the inhibitor to the interior of the infected cell.
      • Introduce transgene into animals to make them virus-resistant?

  2. Attack the cognate proteins

      Suggestive evidence
    • Excess cis-acting sequence competitors inhibit virus growth

      Potential methods:
    • Nucleic acid decoys (DNA or RNA) as competitive inhibitors
    • Synthetic competitive inhibitors

    • Is the cognate protein(s) of viral or host origin?
    • How to screen for, identify potential inhibitors?
    • How to deliver the inhibitor to the interior of the infected cell.

These are discussed in more detail on a separate page.



Those that are not available through PubMed of the National Library of Medicine, USA.

  • Eigen, M., and C. K. Biebricher. 1988. Sequence space and quasispecies distribution, p. 211-245. In E. Domingo and J. J. Holland and P. Ahlquist (ed.), RNA Genetics: Variability of RNA Genomes, vol. 3. CRC Press Inc., Boca Raton, LA.
  • Sim, I. S. 1990. Virus replication: Target functions and events for virus-specific inhibitors, p. 1-47. In G. J. Galasso and R. J. Whitley and T. C. Merigan (ed.), Antiviral Agents and Viral Diseases of Man, 3rd ed. Raven Press, Ltd., New York.