A is for adaptation

Studies of a bacterial virus have revealed an unexpected weapon that helps it to overcome its host's rapidly changing defences. A look at other organisms hints that the mechanism might be widespread.

Adapt or die. This axiom has been used so many times, in so many different contexts, that its origin is difficult to trace. But it aptly describes an intriguing mechanism exploited by viruses that infect Bordetella bacteria. This mechanism guarantees the viruses' survival in the face of host adaptations that would otherwise severely limit their ability to infect and multiply. On page 476 of this issue, Doulatov et al.1 describe how these viruses use a form of the enzyme reverse transcriptase (RT) to generate variability very specifically in the gene encoding a protein required for binding to the bacterial surface. RTs use RNA sequences as templates to generate complementary DNA (cDNA). In this case, it is proposed that the cDNA product directs RT-induced mutations to just the right spot in the viral genome. Natural selection then acts on the mutated progeny viruses to select for those ‘winners’ that can infect bacteria efficiently. This work sheds new light on the pervasive presence of RT genes in many genomes, suggesting another way in which such genes can benefit the organisms in which they reside.

The Bordetella genus includes the causative agents of human whooping cough and canine kennel cough. These bacteria have an elaborate system for evading the immune system of the organism they infect — the nature of their surface proteins is constantly changing. This presents a formidable challenge to any virus (bacteriophage) attempting to infect Bordetella cells, because the first step of viral multiplication is attachment to the bacterial cell surface. Such viruses must evolve exceedingly rapidly to keep pace with a dynamic surface structure as their bacterial host undergoes its own infectious cycle. This is evolution on a very fast track indeed.

The bacteriophage BPP-1 has evolved just such a diversity-generating system, which spawns variant progeny bacteriophages possessing altered surface-binding properties at very high rates (at least 0.1% of the progeny can have significantly altered surface properties). Previous work found that this process is facilitated by interactions between the bacteriophage-encoded RT and two closely related sequences, 134 base pairs long, that lie respectively within and near the gene encoding the surface-binding protein (Fig. 1)(Ref.2). These sequences are called the variable repeat (VR) and the tandem repeat (TR). The TR is an unchanging sequence that provides the raw material for generating variability. On the basis of limited information, it was proposed that the TR is transcribed into an as-yet-undefined RNA, which could in turn be copied by the RT in a highly error-prone manner, generating a hypervariable cDNA product. This would replace a segment of the VR in the genome, producing a swarm of viruses with altered VR sequences in the next generation. So, this VR–TR–RT ‘cassette’ would maintain TR in a pristine state, while generating diversity in VR.

Figure 1: Generating diversity with reverse transcriptase.

a, Bacteriophage BPP-1, a virus that attacks bacteria of the Bordetella genus, contains a reverse transcriptase (RT) enzyme, which, unlike other viral RTs, is not needed for replication. Instead, its role seems to be to create vast diversity within a strictly circumscribed region of the bacteriophage genome, namely the region that encodes the tail-fibre protein, which is needed for the virus to bind to bacteria. This region includes, in addition to the tail-fibre gene, two nearly identical sequences called the variable repeat (VR) and the tandem repeat (TR). TR provides an invariant master source of sequence information. It is copied by RT, during which, as Doulatov et al.1 show, mutations (dots) occur specifically at adenine residues by a process that is not yet understood. The product then replaces a segment of the VR, creating a slightly altered tail-fibre gene. b, Doulatov et al. also discovered similar VR–TR–RT cassettes in various bacteria, as well as the one shown, from Nostoc species (blue-green algae).

Doulatov and colleagues1 now provide more support for this model, and fill in some of the gaps. They show that, remarkably, the diversity-generating system mutates only adenine (A) bases in the TR: by replacing one of the three possible non-A bases in TR with A, they found that the previously invariant base is converted to a hypervariable position in VR. The 23 A residues in the TR sequence can change to any of the other three bases, generating about 1012 different potential versions of VR. The biochemical basis of the specificity for mutation at A is unknown, but presumably the bacteriophage RT could be ‘sloppy by design’ whenever an A is copied.

Insightful genetic analyses by Doulatov et al. also demonstrate that various genetic-selection protocols can produce bacteriophages that have undergone the diversity-generation process. The selection criteria for the altered bacteriophages include natural ones, such as being able to infect a host cell with an altered surface, and artificial ones, such as surviving DNA cleavage by a restriction enzyme. Depending on the nature of the selection, different patterns of survivor mutations were observed.

Restriction enzymes cleave DNA in strictly defined, short sequences. So it is easier to identify and interpret the specific mutations that enable bacteriophages to survive this threat (that is, mutations that prevent DNA cleavage) than to identify those that allow infection of bacteria with altered surface properties. In addition to mutations that occurred directly in the specific restriction-enzyme-recognition site, each variant typically contained a ‘patch’ of additional mutations flanking the site. However, the patterns of inherited mutation differed dramatically in the survivors of two different restriction-site selections, with patches centred precisely on the enzyme-recognition sites. Thus, it is the type of selection imposed that apparently determines the pattern of mutations in survivors. These results are consistent with a mechanism in which segments of the variable cDNA are randomly directed to replace the existing VR segment (but not the TR segment).

Reverse transcriptases are central to many RNA viruses, or retroviruses, where they are key to the viral replication process, through an RNA→DNA→RNA mechanism. Being error-prone is an intrinsic property of RTs, which lack the ability to ‘proofread’ mistakes made in the copying reaction3 — so the RTs also generate genetic diversity in the viral sequences. Such diversity enables the virus to evolve resistance to host defence mechanisms and drugs. However, there is a crucial and noteworthy difference between the Bordetella bacteriophage RT and that of all true retroviruses. The Bordetella bacteriophage has a large DNA genome, and it is predicted to replicate by a DNA→DNA mechanism. The bacteriophage RT is therefore completely dispensable for replication — mutants lacking it are merely unable to generate variants. Thus, it appears that the sole ‘purpose’ of the bacteriophage RT gene is to create sequence diversity. Perhaps similar logic can explain the long-sought ‘function’ of another class of RTs, the retrons4, long known to generate peculiar but very specific nucleic-acid species through reverse transcription5.

The success of this diversity-generating system is underlined by Doulatov and colleagues' discovery that similar cassettes occur in various bacteria and blue-green algae1. These cassettes all consist of an RT gene and suitably oriented TR and VR sequences, one of which is mutated at the A residues relative to the other. This widespread structural conservation suggests that RT sequences lacking any apparent replicative capability can provide a strong selective advantage, based simply on their ability to generate high levels of DNA diversity within specific genes.


  1. 1

    Doulatov, S. et al. Nature 431, 476–481 (2004).

  2. 2

    Liu, M. et al. Science 295, 2091–2094 (2002).

  3. 3

    Gabriel, A., Willems, M., Mules, E. H. & Boeke, J. D. Proc. Natl Acad. Sci. USA 93, 7767–7771 (1996).

  4. 4

    Rest, J. S. & Mindell, D. P. Mol. Biol. Evol. 20, 1134–1142 (2003).

  5. 5

    Lampson, B., Inouye, M. & Inouye, S. Prog. Nucleic Acid Res. Mol. Biol. 67, 65–91 (2001).

Download references

Author information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Boeke, J. A is for adaptation. Nature 431, 408–409 (2004). https://doi.org/10.1038/431408a

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.