Antimicrobial Chemotherapy, 4th Edition

General properties of antimicrobial agents


Antiviral agents

  1. L. Irving

Viruses are almost as versatile as bacteria in the range of diseases they can cause. Vertebrates, insects, plants, and even bacteria are all open to attack. Some viruses of vertebrates (arboviruses) develop in and are transmitted by mosquitoes or other arthropods; others—rabies is a good example—can infect a wide range of mammalian hosts. In general, however, viruses are highly specific in their host range.

All viruses are obligate intracellular parasites; that is, they replicate only within living cells and cannot usually survive for long outside the host cell. Selectivity usually extends not only to the host, but also to the type of cell within the host, as viruses only infect cells that express appropriate receptors on their surface. The preference of a virus for certain types of cell is known as the tropism of the virus, and this often accounts for the characteristic clinical manifestations of particular viral infections. Thus, some viruses preferentially infect liver cells, giving rise to hepatitis.

Intact skin is impermeable to viruses and access to sites within the body can occur only through mucous membranes, damaged skin, insect bites, or direct inoculation. Most viruses that infect man gain entry to the body by adsorption to superficial cells of the mucous membranes of the respiratory, intestinal, and genital tracts, or of the conjunctivae.

The principal types of virus causing human disease are listed in Table 6.1.


Table 6.1 Principal types of virus causing human disease

Properties of viruses

Viruses are deceptively simple. Sizes range from about 20 nm (parvovirus) to 300 nm (poxvirus); consequently, even the biggest viruses fall barely within the limits of resolution of conventional light microscopy and the electron microscope must be used to visualize them.

Complete virus particles (virions) consist of a nucleic acid core (the viral genome), surrounded by a few proteins, and possibly a lipid envelope. The nucleic






acid may be DNA or RNA (never both), single or double stranded, circular or linear, continuous or segmented. This provides all the information needed for viral replication once it is released within the host cell. The proteins serve a number of functions. The capsid, or protein coat surrounding the nucleic acid consists of repeating structural units made up of 1–3 different protein molecules that are generally arranged in helical or icosahedral symmetry. The nucleic acid surrounded by its capsid is referred to as the nucleocapsid of the virus. Many viral proteins have enzymatic properties. These include polymerases and proteases that are necessary for replication and assembly of viral particles. Proteins protruding from the surface coat of the virus act as ligands which will bind to cellular receptors during the first stage of infection of a cell. The lipid envelope possessed by some viruses is derived from membranes of the host cell.

Targets for antiviral drugs

Virus infection of and replication within cells proceeds via a number of distinct steps, each of which, theoretically, provides a possible target for attack (Fig. 6.1). Although the exact details of these steps vary between different viruses, they can be broadly summarized as follows:

  • Adsorption to the cell surface. This involves a specific interaction between proteins (ligands) on the surface of the virus, with receptors on the surface of the cell. The nature of the viral ligands and cellular receptors is known in great detail for some viruses (e.g. the gp120 of human immunodeficiency virus and the CD4 molecule on T lymphocytes), but not at all for others.
  • Uptake into the host cell. Some viruses are able to enter cells by fusion of their own outer lipid membrane with the plasma membrane of the cell, resulting in release of viral nucleocapsid into the cell cytoplasm. In other cases, the process of translocation of the viral particle from the outside of the cell to the inside is very poorly understood.
  • Uncoating of the viral genome. The viral nucleic acid must be released from its capsid before replication. This process may be mediated by cellular lyso-somal enzymes.
  • Macromolecular synthesis.Within the cell, multiple copies of the viral genome are made, and mRNA derived from the virus is translated into multiple copies of the proteins encoded by the viral genome. Many viruses use enzymes present within the host cell to perform these activities, but some carry their own enzymes for certain synthetic processes which are not present within the host cell. A good example is HIV, which copies its RNA genome into a DNA intermediate. Since host cells are not able to convert RNA into DNA, the necessary reverse transcriptase must be encoded by the HIV genome itself.
  • Assembly. Before final assembly into new viral particles, viral proteins may be modified by host cell processes such as glycosylation or phosphorylation.



  • Release. The infected cell has by now become little more than a viral factory and complete viral particles may be released by destruction of the cell, or by continuous export through the cell membrane.

Fig. 6.1 Schematic representation of the virus replication cycle within the host cell, showing the stages that are theoretically open to inhibition by antiviral agents.

Virus–cell interactions

The viral replication cycle described above is representative of an acute viral infection: virus enters the cell, replicates, disrupts normal cellular function, and is released, often resulting in cell death. However, viruses may interact with cells in other ways. Some viruses can undergo latency within cells: the viral genome is present within the cell, and may even become incorporated into the host cell chromosomes, but no replication of the genome occurs, and few, if any, viral proteins are synthesized. The latent virus may cause the cell no harm, but it has the propensity, under certain conditions, to become reactivated, with consequent viral replication and damage to the cell.

Another form of virus–cell interaction is chronic or persistent infection. In this, there is a steady but low-level production and release of virus, but the host


cell is able to survive. However, host cell function may be impaired, and the expression of viral antigens on the cell surface can lead to a chronic inflammatory state as in chronic infection of hepatocytes with hepatitis B virus leading to chronic active hepatitis.

A further potential consequence of virus-cell interaction is transformation, whereby virus infection leads to uncontrolled cell division, resulting in an immortal cell line. For example, Epstein–Barr virus (EBV) infection of B lymphocytes in vitro results in stimulation of cell division, and the establishment of a continuous lymphoblastoid cell line. The molecular mechanisms underlying this process, and possible means of interfering with them, are of great interest, since several viruses, including EBV, have been implicated in initiating malignant change in vivo.

Limitations of antiviral therapy

Although all stages of the cycle of virus replication are potential targets for antiviral drugs, the intimate relationship between the virus and its host cell means that the design or discovery of compounds that are selectively toxic for the virus is beset with considerable problems. Moreover, antiviral therapy may be at a disadvantage for quite a different reason: in many viral diseases initial infection and spread is commonly asymptomatic and the onset of illness, often at the peak of viral multiplication, occurs when the host defences have been fully mobilized. Unless the patient is immunodeficient in some way, these defences are usually quite able to deal with the infection unassisted. Consequently, initiation of antiviral therapy at the onset of symptoms may have little influence on the course of the disease.

Viruses that undergo latency pose a further problem. Unless the latently infected cells can be killed or removed, the virus cannot be eliminated from the patient. Achieving such elimination by therapeutic intervention is a daunting task, as there are no virus-specific metabolic processes occuring in those cells, and there is no expression of viral antigens on the cell surface against which an immune response could be mounted. Despite these discouraging considerations, some success has been achieved in the development of antiviral compounds and a handful of antiviral agents are now in regular use (Table 6.2). Most presently available agents are nucleoside analogues that interfere with viral replication. Most clinical success has been achieved with agents active against viruses of the herpes group, although considerable progress has been made in developing drugs which act to inhibit the replication of HIV (see Chapter 7).

Table 6.2 Principal antiviral agents (other than anti-HIV agents) in present use



Mode of action

Route of administration



Herpes simplex; varicella zoster

Nucleoside analogue

Oral; topical; intravenous


Influenza A

Uncoating of virus




Nucleotide analogue



Herpes simplex

Nucleoside analogue




Varicella zoster






Antisense oligonucleoside





DNA polymerase inhibitor





Nucleoside analogue




Chronic hepatitis





Chronic hepatitis

Nucleoside analogue; reverse transcriptase inhibitor




Respiratory syncytial virus; chronic hepatitis

Nucleoside analogue




Herpes simplex; varicella zoster

Nucleoside analogue





Neuraminidase inhibitor





Properties of antiviral agents


Aciclovir was the first antiviral drug with selective toxicity: it inhibits the replication of certain viruses while exhibiting virtually no toxic side-effects on host cells. Structurally, it is acycloguanosine (Fig. 6.2), an analogue of the purine nucleoside, guanosine, in which the ribose moiety has lost its cyclic configura-tion. Aciclovir itself is inactive; in order to achieve its antiviral effect, it must first be phosphorylated to the triphosphate form within the infected cell. Although the second and third phosphate groups are added by cellular enzymes, the first phosphorylation step is accomplished by a viral thymidine kinase, specified by herpes simplex and varicella zoster viruses (HSV and VZV); the cellular form of this enzyme is much less efficient in producing aciclovir monophosphate. This unique feature is the basis for the selective toxicity of aciclovir, for two reasons: firstly, it means that the active form of the drug is produced only in virally infected cells; secondly, by the law of mass action, once the equilibrium

         aciclovir [left harpoon over right harpoon] aciclovir monophosphate

is shifted to the right within an infected cell, more free aciclovir will enter the


cell from the extracellular space, thus resulting in concentration of the drug precisely where it is needed: in the infected cell.


Fig. 6.2 Structures of deoxyguanosine and three antiviral agents that act as analogues of the nucleoside: aciclovir, ganciclovir and penciclovir.

Mode of action

As an analogue of guanosine triphosphate, aciclovir triphosphate is incorporated into the growing DNA chain and causes chain termination, since it lacks the 3′-hydroxyl group necessary to form the 5′–3′phosphodiester linkage with the next base. Aciclovir triphosphate is also a direct inhibitor of viral DNA polymerase.


It is theoretically possible for a virus to become resistant to aciclovir in several ways:

  • Strains of virus which lack the thymidine kinase enzyme, known as TK–mutants, are inherently resistant to the drug, as they cannot generate the active form.
  • Mutations in the TKgene may alter the thymidine kinase molecule, such that it becomes unable to perform the first phosphylation step.



  • Alterations in viral DNA polymerase may reduce the ability of the enzyme to bind aciclovir triphosphate, and thereby to escape the inhibitory properties of the drug.

All these mechanisms occur in the laboratory and in nature. However, TK-strains, and those with altered thymidine kinases appear to exhibit reduced virulence, and DNA polymerase mutants are the ones that cause most clinical difficulties. These emerge particularly in immunocompromised patients including those with HIV infection, in whom recurrent HSV infections cause frequent and extensive disease necessitating prolonged treatment with aciclovir.

Aciclovir represents a prime example of what a good antiviral drug should be. It has potent antiviral activity, but is virtually free of toxic side-effects. It can be life-saving in certain infections, and in others can significantly decrease morbidity (see Chapter 28). However, it is poorly absorbed when given orally, and in life-threatening infection it must be given intravenously.

Analogues of aciclovir

Several structural analogues of aciclovir have been developed. Valaciclovir, the L-valyl ester of aciclovir, is an oral pro-drug that is well absorbed when given by mouth to release aciclovir into the bloodstream. A similar relationship exists between famciclovir and penciclovir; the former is metabolized into the latter after oral administration. Penciclovir (Fig. 6.2) exhibits antiviral activity very similar to that of aciclovir, but the half-life within cells is considerably longer, so fewer doses are necessary to achieve an antiviral effect.

Although aciclovir and its analogues are undoubtedly successful antiviral drugs, their clinical usefulness is limited by their narrow spectrum of activity. They are only active against a subgroup of herpesviruses—HSV and VZV—that can perform the first phosphorylation step. It is particularly disappointing that these otherwise excellent drugs have no activity against cytomegalovirus (CMV), a herpesvirus which does not encode its own thymidine kinase enzyme. CMV is an opportunist pathogen that causes severe and even life-threatening disease in the ever-increasing population of immunosuppressed patients such as transplant recipients and HIV-infected individuals. The four drugs described next were specifically developed with the aim of finding an effective anti-CMV agent.


Ganciclovir (dihydroxypropoxymethylguanine, DHPG, Fig. 6.2) is a derivative of aciclovir that shows useful activity against CMV. Ganciclovir must be activated by phosphorylation, but in contrast to aciclovir, it can be phosphorylated by CMV, which uses a different virally encoded phosphotransferase, the product of the CMV gene known as UL97. Subsequent phosphorylation to the triphos-phate generates a compound which acts as a viral DNA polymerase inhibitor. Unfortunately, cellular enzymes can also phosphorylate ganciclovir, so that active


drug is generated in uninfected cells. Thus the toxicity of ganciclovir is considerably greater than that of aciclovir. The most frequent unwanted effects, leucopenia and thrombocytopenia, arise from a toxic effect on the function of the bone marrow, particularly affecting the production of neutrophils. Ganciclovir treatment can also result in a rise in serum creatinine. It is poorly absorbed (<10 per cent) when given orally, and has to be administered intravenously. A valyl ester of ganciclovir, valganciclovir, is much better absorbed orally (40 per cent), and serum concentrations seem to be very close to those arising from intravenous ganciclovir therapy. Despite its drawbacks, the anti-CMV activity of ganciclovir can be sight- or life-saving in immunosuppressed patients with severe CMV infections (see Chapter 28).

Resistance to ganciclovir can arise through point mutations in the UL97 gene resulting in an inability of the virus to phosphorylate the drug. Such mutants arise fairly commonly in clinical practice, but fortunately they remain sensitive to foscarnet (see below). Mutations in the CMV DNA polymerase gene may also confer resistance to ganciclovir.


The difficulty in achieving the first phosphorylation step of aciclovir-like compounds is avoided by acyclic nucleoside phosphonates, a class of compounds of which cidofovir (Fig. 6.3) is an example. The phosphonate group acts as a phosphate mimetic, and is attached to the acyclic nucleoside moiety through a stable P–C bond which cannot be cleaved by cellular hydrolases. These agents need only two phosphorylation steps to reach the active triphosphate form, and these steps are performed by cellular enzymes. Thus, active drug may arise in both infected and uninfected cells, but a selective antiviral activity is maintained, since the triphosphate exhibits a higher affinity for viral DNA polymerases than


for the corresponding cellular enzymes. Like aciclovir triphosphate, the drugs act as chain terminators and competitive inhibitors.


Fig. 6.3 Structure of cidofovir.

Cidofovir, the first such compound to be licensed for use, is an acyclic cytosine analogue. An acyclic adenine analogue, adefovir, is in development. Given the mechanism of action of these drugs they would be expected to have a broad antiviral spectrum against all viruses with a DNA polymerase. Cidofovir is indeed active against all herpesviruses, as well as adeno-, polyoma-, papilloma-, and poxviruses. In addition, adefovir has activity against reverse transcriptase (which, after all, is a modified form of a DNA polymerase enzyme), and inhibits replication of both HIV and hepatitis B virus. This class of drugs thus offers great promise for the treatment of a wide range of DNA virus and retrovirus infections.

Cidofovir is administered by intravenous infusion. It has a prolonged antiviral effect such that dosing is required only once a week or less. It is, however, nephrotoxic in a dose-dependent manner, and must be administered with probenecid to prevent irreversible renal damage. In theory, resistance to cido-fovir and its congeners may arise through mutations in the viral DNA polymerases and time will tell how much of a problem this will become.


Unlike the drugs discussed so far, foscarnet (Fig. 6.4) is not a nucleoside analogue; it is trisodium phosphonoformate, a derivative of phosphonoacetic acid, and is therefore a pyrophosphate analogue. It does not require phosphorylation. It forms complexes with DNA polymerases, and prevents cleavage of pyrophosphate from nucleoside triphosphates, resulting in inhibition of further DNA synthesis. It shows some selective toxicity for viral rather than host cell enzymes, and is active against all the herpesviruses, including CMV. It is used as an alternative to ganciclovir in the treatment of serious CMV infection, as well as in the treatment of aciclovir-resistant HSV infection. It also has some antiretroviral activity, but the clinical significance of this in the treatment of HIV infection is unclear. Like ganciclovir, oral bioavailability is poor, and it has to be administered by intravenous injection. It is nephrotoxic, and can cause acute renal failure. Other side-effects include symptomatic hypocalcaemia, and penile ulceration.


Fig. 6.4 Structure of foscarnet.




Fomivirsen is an oligonucleotide, 21 bases in length available in some countries for the treatment of CMV retinitis by intravitreous injection. It is an antisense molecule: the mirror image of a section of CMV-derived mRNA encoding regulatory proteins. Binding of the drug to this region prevents translation of the RNA into protein.


This nucleoside analogue, 3-thiacytidine, is an inhibitor of reverse transcriptase developed for the treatment of HIV infection (see Chapter 7). However, it is also active against hepatitis B virus. Although this is not a retrovirus (it has a DNA genome), there is a step in the viral replication cycle in which a viral DNA polymerase converts RNA into a DNA copy by reverse transcription. Lamivudine is triphosphorylated in the cell by cellular enzymes, and the triphos-phate acts as a chain terminator and DNA polymerase inhibitor.

Ribavirin (tribavirin)

Tribavirin (Fig. 6.5) is the British Approved Name for the substance known elsewhere as ribavirin. It is a triazole compound structurally related to guanosine. Like other nucleoside analogues it has to be activated intracellularly by phos-phorylation. In the triphosphate form it inhibits viral protein synthesis, apparently by interfering with ‘cap’ formation at the 5′ end of mRNA.


Fig. 6.5 Structure of ribavirin (tribavirin).

Ribavirin has an unusually broad spectrum of activity, against both RNA and DNA viruses, at least in vitro. Its main use is in the treatment of severe lower respiratory tract infection in young children, caused by respiratory syncytial virus. The compound is administered by inhalation of an aerosolized solution. Oral ribavirin has also been used succesfully in the treatment of Lassa fever.



In the treatment of chronic hepatitis C, combination therapy with ribavirin plus interferon has led to a considerable improvement in response rates compared to the use of interferon alone. However, hopes that this drug might be useful in the treatment of HIV-infected patients have not been realized.

The most commonly encountered serious side-effect of prolonged ribavirin use is macrocytic anaemia. It may also cause a rise in serum bilirubin levels.

Older nucleoside analogues

Nucleoside analogues have been used for many years as antiviral agents, but the older ones are very toxic and have been eclipsed by later developments. Idoxuridine and trifluridine are still available in many countries for topical application in recurrent herpes simplex infections, but are no longer recommended. Cytarabine (cytosine arabinoside; ara-C) and vidarabine (adenine arabinoside; ara-A) are cytotoxic drugs of limited availability that are sometimes used as agents of last resort in life-threatening and otherwise untreatable viral infections.

Amantadine and rimantadine

Amantadine (Fig. 6.6) is a tricyclic amine derivative of adamantane, a compound that originally aroused interest because of its symmetrical three-dimensional structure which is composed entirely of carbon atoms and is thus related to the crystalline array of natural diamonds. The antiviral activity of amantadine was first described in 1964. Rimantadine is a closely-related substance that is available in some countries.


Fig. 6.6 Structure of amantadine.

The activity of both amantadine and rimantadine is restricted to influenza A virus. Other influenza viruses are virtually unaffected at therapeutically achievable concentrations. These compounds block a viral matrix protein ion channel, thereby interfering with uncoating of the viral nucleic acid within the infected cell. Resistance readily arises by mutation in the matrix protein, which is then unable to bind the drugs.

Amantadine binds to N-methyl-D-aspartate receptors in the brain, and therefore has dopaminergic effects. These effects can be therapeutically useful (amantadine


is also licensed in the UK as an anti-Parkinsonian drug), but also account for the side-effects of the drug: restlessness, insomnia, agitation, and confusion. Unfortunately, amantadine is poorly tolerated, especially by the elderly, one of the groups who stand to benefit most from an effective anti-influenza drug. Rimantadine is said to give rise to fewer side-effects, but experience is limited.

As with ribavirin, combination therapy of amantadine plus interferon is under trial for the treatement of patients with chronic hepatitis C virus infection. Amantadine is also licensed in the UK for use in shingles, but this has few advocates.


Zanamivir is the first of a new class of anti-influenza compounds with a novel mode of action: it is a selective inhibitor of viral neuraminidase, one of the two proteins (the other being haemagglutinin) present on the surface of all types of influenza viruses. Viral neuraminidase enables budding influenza virus particles to break away from an infected cell. It may also facilitate the passage of virus particles through mucus to reach epithelial cell surfaces, thereby aiding infection of new cells. Zanamivir acts extracellularly to inhibit the release and propagation of infectious influenza viruses from the epithelial cells of the respiratory tract. Unlike amantadine, it acts against both influenza A and B viruses, and inhibits all the known neuraminidase subtypes of influenza A viruses. It is administered by oral inhalation. Early indications are that zanamivir is generally safe, but there is a risk of bronchospasm in patients with underlying lung disease

The site of neuraminidase to which zanamivir binds is highly conserved among different strains of influenza virus, and, being the active site of the enzyme, is not a region of the molecule that undergoes significant antigenic drift. It is hoped, therefore, that resistance to the drug through mutation at its target site will not occur. No evidence of resistance has yet been detected, but experience is limited.

A similar neuraminidase inhibitor, oseltamivir, which is administered orally, is also available in some countries.


Interferon was described in 1957 as an antiviral compound in chick embryo cells. It turns out that the activity is associated with a family of species-specific glycosylated proteins produced in vivo in response to viral or antigenic challenge. There are three types of human interferon: interferon-α (IFN-α), produced by many cell types; interferon-β (IFN-β), produced by fibroblasts; and inter-feron-γ (IFN-γ, sometimes referred to as immune interferon) produced by T lymphocytes. IFN-α and IFN-β share 30 per cent structural homology, but they are quite distinct from IFN-γ, which shares only about 10 per cent homology in its amino acid sequence. Moreover, there are over 15 different forms of IFN-α, each differing by a few amino acids, and, possibly, two forms of IFN-β.



Interferons have a wide range of biological effects. They activate several different biochemical pathways within a cell, with the result that the cell is rendered resistant to virus infection. The relative importance of each of these pathways differs between different interferons, and indeed between different cells stimulated by the same interferon. One pathway results in the activation of a ribonuclease that digests viral RNA. Another results in phosphorylation of a protein known as initiation factor; phosphorylation of this factor, the normal function of which is to assist in the initiation of transcription of mRNA into protein, effectively prevents production of viral proteins.

Interferons also have various effects on cells of the immune system. As with their antiviral properties, these immunomodulatory effects vary in detail between different interferons, but they include stimulation of natural killer cells, induction or suppression of antibody production, stimulation of T cell activity, and stimulation of expression of HLA class I and II molecules on the surface of cells. Finally, interferons affect cell proliferation, which has led to their successful use in the management of certain malignant tumours.

When interferons were discovered, they were hailed as the antiviral equivalent of penicillin. As these substances were produced by cells as a natural defence in response to a wide range of virus infections, it seemed reasonable to imagine that when used as therapeutic agents, they would exhibit a broad antiviral spectrum, and that their toxic effects on host cells would be minimal. Early studies were hampered by difficulties in obtaining sufficient quantities of interferons to conduct clinical trials. This problem has been solved by recombinant DNA technology, which has allowed cloning and expression of the relevant genes, and by the use of Sendai virus to induce interferon production by a lymphoblastoid cell line (so-called lymphoblastoid interferon). Unexpectedly, clinical trials revealed that patients receiving interferon experienced ‘flu-like side-effects: fever, headache, and myalgia. This led to the realization that individuals suffering from influenza complain of ‘flu-like symptoms precisely because of the induction of interferons by the virus. Most patients become tolerant to these effects after the first few doses.

Clinical experience with interferons has been disappointing. Interferon-α is succesfully used in the management of patients with chronic viral hepatitis, but even here, the therapeutic effect may be due to its immunomodulatory rather than antiviral properties.

Prevention of virus infections

Despite the progress that has been made in antiviral therapy, the major impact on virus infections is achieved by prevention rather than by treatment. The successful eradication of smallpox by appropriate use of vaccinia vaccine is ample testament to this statement, and has lent credence to the notion that other viral infections may be similarly eradicated. The World Health Organization


campaign to eliminate polio is progressing towards a successful conclusion, and in many developed countries measles, mumps and rubella are becoming rare.

Immunization against infectious disease may be passive or active.

Passive immunization

The transfer of preformed antibodies from one individual to another can be achieved with normal human gammaglobulin derived from blood donors, provided that the antibody in question is present in a sufficiently large number of randomly selected donors, as in the case of anti-hepatitis A virus. If not, then hyperimmune globulin can be prepared specifically from individuals known to have high antibody titres. Examples of preparations in current use include hyper-immune hepatitis B immunoglobulin (HBIg), varicella-zoster immune globulin (VZIg), and human rabies immunoglobulin (HRIg).

Recently, a monoclonal antibody directed against respiratory syncytial virus (RSV), palivizumab, has been introduced as prophylaxis against severe RSV disease in certain at-risk babies.

Active immunization

The host can be stimulated to produce a protective immune response by vaccination with a form of the infectious agent that does not cause disease. Vaccines can be alive or dead. Live vaccines consist of attenuated forms of the infectious agent. Examples include measles, mumps, rubella, and live poliovirus (Sabin) vaccines. Difficulties arise in ensuring that the attenuated variant is unable to revert to virulence: the few cases of paralytic polio that have occurred in Europe and North America in recent years are nearly all caused by vaccine strains that have reverted to virulence.

Dead vaccines may consist of the whole agent, grown in the laboratory and subsequently killed by some means, or of a subunit of the agent, usually prepared by recombinant DNA technology. Dead vaccines have the advantage that there is no risk of reversion to virulence, although there is at least one infamous example of an inadequately killed poliovirus vaccine that led to an outbreak of paralytic polio. However, they are less immunogenic than live vaccines, and therefore more doses need to be given to achieve a satisfactory response. Examples of such vaccines include poliovirus (Salk), rabies virus, hepatitis A and hepatitis B virus vaccines; the latter is a subunit vaccine, consisting only of the surface protein (HBsAg), prepared by cloning and expressing the appropriate gene into yeast cells.