Gene regulation

Reviving the message

Article metrics

Studies of developmental regulators in worms and cell-cycle regulators in yeast have revealed a new family of enzymes, which may affect the fate of specific messenger RNA molecules.

Messenger RNA molecules are crucial intermediates between genes and their encoded proteins. When a gene is activated, enzymes produce an mRNA copy of it; this mRNA in turn provides a template for the production of proteins, which carry out specific tasks in the body. mRNAs consist of strings of nucleotides, and in our cells most mRNA molecules have a long tract of 'A's (adenosine nucleotides) at one end — the so-called 3′ end. Such poly(A) tails seem to be required for every step in an mRNA's life, including its export from its site of production in the cell nucleus, translation into protein, and stability1. On page 312 of this issue, Wang and colleagues2 describe an unusual enzyme, important in the development of the nematode worm Caenorhabditis elegans, that they propose lengthens the poly(A) tails on certain mRNAs. What's interesting is that this work apparently defines an entirely new class of such enzymes, and has implications for developmental and cell biology.

Messenger RNA precursors first become 'polyadenylated' in the nucleus, during or shortly after gene transcription, in a reaction involving two coupled steps: cleavage of the RNA to form a new 3′ end, then poly(A) addition by a poly(A) polymerase enzyme. Polyadenylation can also, however, occur outside the nucleus, in the cell cytoplasm. During the early embryonic development of many animals and the maturation of germ cells (eggs and sperm), transcription is largely switched off. Elongating the short poly(A) tail of dormant cytoplasmic mRNAs provides a rapid way to stabilize and activate them3, allowing proteins to be produced without transcription. (mRNAs need a long poly(A) tail to function as efficient templates in translation.)

In C. elegans, the gld-2 and gld-3 genes control various aspects of germline development, including the mitosis/meiosis decision — simply put, whether germline cells multiply to produce other such cells, or generate eggs and sperm4. But exactly how the proteins encoded by these genes regulate developmental decisions has been unclear. This is what Wang et al.2 set out to investigate, and they have found that these proteins form a cytoplasmic poly(A) polymerase with a difference.

Wang et al. started by inspecting the predicted amino-acid sequence of the GLD-2 protein, and discovered an immediate clue to its biochemical function. The protein contains a domain that is similar to a region in certain nucleotidyltransferase enzymes (Fig. 1). These enzymes form a protein superfamily to which all eukaryotic poly(A) polymerases belong5. The authors also found that GLD-2 is located in the cytoplasm of germline and early embryonic cells, and interacts specifically with GLD-3, which itself belongs to a family of RNA-binding proteins (J. Kimble et al., personal communication).

Figure 1: A new type of RNA-modifying enzyme?

Wang et al.2 have identified the GLD-2 protein as a poly(A) polymerase enzyme in vitro; they propose that it extends the tract of adenosine nucleotides at the 3′ end of certain messenger RNAs in vivo. Unlike many such enzymes, GLD-2 can work only with another protein, GLD-3, which binds RNA. Other such unusual poly(A) polymerases may exist, as can be seen in this comparison of GLD-2 with other proteins. A nuclear poly(A) polymerase, PAP-α, which can bind RNA without a helper protein, is also shown. The proteins are represented as lines, with conserved regions shown as cylinders (orange, catalytic domain; blue, central domain; red, RNA-binding domain; yellow, protein-interaction domains called zinc fingers). The polymerase activities are A, adenylylating, and U, uridylylating; the asterisk indicates that poly(A) polymerase and DNA polymerase activities have been reported for this protein.

Given this cytoplasmic location of GLD-2, and the slight similarity of its amino-acid sequence to those of nuclear poly(A) polymerases, Wang et al. decided to test the protein for RNA-dependent poly(A) polymerase activity in vitro. They found that GLD-2 alone had low poly(A) polymerase activity, but was stimulated by GLD-3, which by itself was completely inactive. Analysis of the reaction products showed that GLD-2 alone extended an RNA 'primer' by only a few adenosines, whereas GLD-2 and GLD-3 together made tails of up to 30 adenosines. Two GLD-2 mutant proteins, one designed to abolish its predicted catalytic centre and the other to disrupt its binding to GLD-3, were inactive when tested either alone or with GLD-3. All known poly(A) polymerases comprise a single protein. So Wang et al. have found a new type of cytoplasmic poly(A) polymerase, in which GLD-2 provides the catalytic subunit and GLD-3 contributes the RNA-binding function.

These findings raise several questions. Which mRNAs are the physiological substrates of the newly discovered enzyme during early development? Can GLD-2 interact with other RNA-binding proteins to expand its substrate repertoire — an idea proposed by Wang et al.? Are GLD-2 and GLD-3 enough to carry out the reaction in vivo, or are other factors involved? And does this newly discovered process share any components with previously described cytoplasmic poly(A) polymerases involved in development, or with the nuclear 3′-end-processing apparatus?

Wang and colleagues' work also has ramifications that go beyond the control of early development. The polyadenylation of cytoplasmic mRNAs appears to be widespread in eukaryotes, and GLD-2 may represent a new family of bipartite cytoplasmic poly(A) polymerases. For example, the fission-yeast proteins Cid1 and Cid13 are cytoplasmically located relatives of GLD-2, and have poly(A) polymerase activity in vitro6,7. Like GLD-2, Cid1 is involved in controlling the cell-division cycle. Cid13 has been proposed to increase the pools of nucleotides needed for DNA replication, by extending the poly(A) tail of the mRNA encoding Suc22 — part of an enzyme involved in nucleotide synthesis. Moreover, it has been reported6 that Trf4, a relative of the Cid proteins that occurs in budding yeast8, has in vitro poly(A) polymerase activity (although this is controversial, and previous in vitro tests identified Trf4 as a DNA polymerase8). Many more members of the GLD-2 family may exist, as inferred from sequence comparisons (Fig. 1).

Further insights into GLD-2 can be gleaned from a look at its amino-acid sequence. Mammalian and yeast nuclear poly(A) polymerases have a three-domain structure consisting of a catalytic portion, a central linker and an RNA-binding domain. GLD-2, like most of its close relatives, lacks an RNA-binding domain. Nonetheless, Wang et al.'s comparison of the sequences of GLD-2 and its relatives with that of mammalian nuclear poly(A) polymerase suggests that the structure of the catalytic domain and part of the central domain is similar, despite considerable sequence divergence (Fig. 2). The most prominent conserved features are three aspartate amino acids — which chelate divalent metal ions — in the active site, and several amino acids in the catalytic and central domains that help to bind ATP (see Fig. 2 on page 313).

Figure 2: Structure of the mammalian nuclear poly(A) polymerase, with regions of similarity to GLD-2 and its relatives highlighted (colour code as in Fig. 1).

The conserved loop might be a hallmark of enzymes that elongate single-stranded RNA substrates. ATP in the active site is shown in red. (Figure modified from ref. 10.)

Moreover, analysis of mutant forms of mammalian poly(A) polymerases suggests that amino acids in a loop near the ATP-binding pocket are needed to bind the 3′ end of mRNAs (our unpublished results). The sequence of this loop is also seen in GLD-2 and its relatives, and in 3′-terminal uridylyltransferase — an enzyme in trypanosomes that catalyses the addition of uridine nucleotides to the 3′ ends of intermediates of RNA editing9. So this sequence seems to be a hallmark of enzymes that elongate single-stranded RNA substrates.

Lengthening the poly(A) tails of selected mRNAs at specific times can both counteract normal mRNA turnover and stimulate translation. This is useful, because it bypasses the requirement for transcription and RNA processing, for example in times of metabolic stress or when the genome is damaged or inactive. Moreover, this mechanism — like others that act on cytoplasmic mRNAs — is much more rapid than controlling transcription, particularly at large genes. Cytoplasmic polyadenylation would gain great versatility if, as Wang et al. suggest2, different mRNAs could be targeted by using different RNA-binding proteins to recruit a poly(A) polymerase. Whether that happens remains to be seen, but this and other questions will keep many of us busy and excited for a long time.


  1. 1

    Shatkin, A. J. & Manley, J. L. Nature Struct. Biol. 7, 838–842 (2000).

  2. 2

    Wang, L., Eckmann, C. R., Kadyk, L. C., Wickens, M. & Kimble, J. Nature 419, 312–316 (2002).

  3. 3

    Richter, J. D. Microbiol. Mol. Biol. Rev. 63, 446–456 (1999).

  4. 4

    Kadyk, L. C. & Kimble, J. Development 125, 1803–1813 (1998).

  5. 5

    Aravind, L. & Koonin, E. V. Nucleic Acids Res. 27, 1609–1618 (1999).

  6. 6

    Saitoh, S. et al. Cell 109, 563–573 (2002).

  7. 7

    Read, R. L. et al. Proc. Natl Acad. Sci. USA (in the press).

  8. 8

    Castaño, I. B., Heath-Pagliuso, S., Sadoff, B. U., Fitzhugh, D. J. & Christman, M. F Nucleic Acids Res. 24, 2404–2410 (1996).

  9. 9

    Aphasizhev, R. et al. Cell 108, 637–648 (2002).

  10. 10

    Martin, G., Keller, W. & Doublié, S. EMBO J. 19, 4193–4203 (2000).

Download references

Author information

Correspondence to Walter Keller.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Keller, W., Martin, G. Reviving the message. Nature 419, 267–268 (2002) doi:10.1038/419267a

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.