Translational control is a key genetic regulatory mechanism underlying the initial establishment of the major spatial axes of the Drosophila embryo. Many translational control mechanisms target eIF4E (eukaryotic initiation factor 4E), an initiation factor that recognizes the 5′-cap structure of the mRNA. Cap recognition by eIF4E, in complex with eIF4G, is essential for recruitment of the mRNA to the small ribosomal subunit. One established mechanism for repressing translation involves eIF4E-binding proteins, which competitively inhibit the eIF4E–eIF4G interaction. Our group has uncovered a novel mechanism for repression in which an eIF4E cognate protein called d4EHP, which cannot bind eIF4G, binds to the 5′-cap structure of cad mRNA thus rendering it translationally inactive. These two related, but distinct, mechanisms are discussed and contrasted in this review.
Translational control is a key mechanism underlying the initial establishment of the major spatial axes of the Drosophila embryo. Proteins encoded by several mRNAs, including oskar (osk), nanos (nos) and caudal (cad), are essential for anterior–posterior patterning. In order to carry out their functions in development, Osk, Nos and Cad proteins must be restricted in space to the future posterior of the embryo. To achieve asymmetric distribution of the proteins, translation of all these mRNAs is repressed in regions of the embryo from where the relevant protein needs to be excluded. Mechanisms also exist to localize and enrich the concentrations of osk and nos mRNAs at the posterior pole, where the proteins they encode are required. Unlike osk and nos, cad mRNA is uniformly distributed. Translational repression of osk, nos and cad mRNAs occurs at the mRNA 5′-cap structure recognition step, but two distinct mechanisms that target different components of the cap-binding complex are involved. For osk and nos, regulatory proteins that bind eIF4E (eukaryotic initiation factor 4E) are recruited to the mRNA. These proteins competitively inhibit the interaction between eIF4E and eIF4G, which is an essential step of cap-dependent translation initiation. In contrast, our group has uncovered a different mechanism that regulates cad mRNA, whereby an eIF4E-related cap-binding protein is recruited to the 5′-cap structure . This protein, called 4EHP (eIF4E-homologous protein), cannot bind eIF4G and therefore acts in a dominant-negative manner by associating with the cap and blocking its association with the cap-binding complex. These two related but distinct mechanisms of translational repression will be discussed in detail and contrasted below.
Translational repression of osk through recruitment of the eIF4E-binding protein Cup
osk mRNA is translationally repressed outside of the pole plasm, and this repression is alleviated for the small proportion  of osk RNA that is localized. The interaction of Bruno (Bru) and Apontic (Apt) proteins with specific sequences [BREs (Bru-response elements)] in the osk 3′-UTR (3′-untranslated region) contributes to translational repression of unlocalized osk mRNA ([3–5], reviewed in ). Bru-mediated translational repression of osk involves its association with Cup, an eIF4E-binding protein [7,8]. As such, Cup blocks the recruitment of eIF4G to eIF4F, the complex that binds the 7-methylguanosine cap present at the 5′-end of most eukaryotic mRNAs and is required for their translation. Thus an association between osk and Cup through Bru renders osk translationally inactive.
In the posterior pole, repression of osk translation is alleviated and translation is activated; these may be distinct processes . Several molecules have been implicated in these processes and several models have been proposed. Staufen is required for osk localization [10,11] and for its translational activation  through an unknown mechanism. Perhaps more relevant to alleviating Cup-mediated translational repression is the activity of Orb, the Drosophila homologue of CPEB (cytoplasmic polyadenylation element-binding protein). Orb and the 3′-UTR of osk RNA interact, and in orb mutants osk translation is reduced, suggesting that cytoplasmic polyadenylation might underlie activation of osk translation . Although experiments in extracts that support Bru-mediated regulation indicate that changes in poly(A) tail length are not involved in translational control of osk , a more recent report indicates that Orb and a poly(A) tail of at least 150 residues are necessary for Osk accumulation in vivo .
nos translation is also repressed by Cup
Like osk, translation of nos is regulated through the interaction of proteins with cis-acting elements present in its 3′-UTR. Also similarly to osk, unlocalized nos, which makes up over 90% of the total  is translationally repressed, and this repression is alleviated in the pole plasm. In early embryos, nos is translationally repressed outside the pole plasm through an interaction between Smaug (Smg) protein and two specific sequences in the 3′-UTR called SREs (Smg-response elements) [15–17]. Recent evidence indicates that Smg, like Bru, functions by interacting with the eIF4E-binding protein Cup . Thus Smg specifically recruits nos mRNA into a translationally inactive state, blocking translation by interacting with eIF4E and sequestering it away from the cap structure at the 5′-end of the mRNA.
In addition to its translational repressor activity, which requires eIF4E binding, Smg has an important function in the degradation of unlocalized maternally derived mRNAs in the early embryo . This phenomenon has been studied by examining Hsp83 mRNA, which is concentrated to the pole plasm by a mechanism that involves its stabilization at the posterior and its degradation elsewhere. In smg mutant embryos, Hsp83 mRNA is not degraded outside the pole plasm. The function of Smg in this process is to recruit the CCR4–POP2–NOT deadenylase complex (where CCR4 stands for carbon catabolite repressor 4) to Hsp83 mRNA. Interestingly, Smg has no effect on Hsp83 (heat-shock protein 83) translation, nor are Cup or the SRE involved in Smg-mediated effects on RNA stability.
cad translation is repressed by a novel d4EHP-dependent mechanism, not by sequestering eIF4E
Our work has identified a novel mode of mRNA-specific translational inhibition, which does not involve competition between eIF4G and an eIF4E-binding protein such as Cup, for eIF4E. These two modes of translational inhibition are compared graphically in Figure 1. The competition in this case is for binding to the 5′-cap structure itself. It has long been established that a posterior-to-anterior gradient of Cad protein is established in early embryogenesis from uniformly distributed maternal cad mRNA, and this gradient is essential for posterior patterning . Establishment of this Cad gradient requires Bicoid (Bcd), which mediates cap-dependent translational repression of cad mRNA dependent on the BBR (Bcd-binding region), an element in its 3′-UTR [21–24]. Thus as for eIF4E-binding protein-mediated translational repression, specificity for a particular mRNA is conferred by an RNA-binding protein (in this case, Bcd) that interacts with an element located within its 3′-UTR.
Two contrasting mechanisms of cap-dependent translational regulation of specific mRNAs
An eIF4E-related protein called human 4EHP was described previously . While 4EHP has 5′-cap structure-binding activity, it does not interact with eIF4G, suggesting that it could function as a negative regulator of translation [25,26] as it cannot function in ribosome recruitment.
The Drosophila genome encodes an orthologue of 4EHP (d4EHP), which we showed has biochemical activities similar to human 4EHP. To understand the function of 4EHP in development, a hypomorphic allele was created by imprecise excision of a nearby P-element. This produced a small deletion which removed part of the d4EHP gene but which still allowed low-level expression from a cryptic AUG codon located within the first intron. Homozygotes for this d4EHP allele are viable, but approximately half of the embryos produced by homozygous females fail to hatch into larvae and exhibit patterning defects reminiscent of the progeny of bcd females. This suggested that d4EHP has a role in Bcd-mediated developmental processes.
We observed no effect of the d4EHP mutation on the transcriptional activation function of Bcd, as measured by monitoring zygotic hb expression. However, we observed a clear defect in translational regulation of cad mRNA, in that Cad protein, which in wild-type embryos is only detectable in the posterior half of the embryo, was apparent throughout the entire length of many embryos produced by d4EHP mutant females.
By immunoprecipitation from embryonic extracts and from cultured mammalian cells transfected with constructs expressing epitope-tagged proteins, a physical interaction between Bcd and d4EHP was demonstrated. Conversely, Bcd and deIF4E did not show detectable interaction in these assays. The co-transfection assay was used to map residues in d4EHP required to bind the 5′-cap structure and to bind Bcd. Mutant forms of d4EHP that were abrogated either for Bcd binding or for cap binding could not confer translational regulation of cad in embryos produced by d4EHP mutant mothers, whereas wild-type d4EHP expressed from a transgene in the same way fully rescued the Cad gradient and embryonic viability. Thus d4EHP must interact with both the 5′-cap structure and Bcd to carry out its developmental function. In a similar manner, individual residues in Bcd were identified that are essential for d4EHP binding. While Bcd contains a sequence motif related to the consensus eIF4E-binding domain , we determined that a critical residue for d4EHP binding (Tyr66) is distinct from this domain and a tyrosine residue that is invariant and essential in eIF4E-binding domains (Tyr68) is dispensable for d4EHP binding. Furthermore, mutant forms of Bcd that cannot interact with d4EHP are unable to confer translational repression of cad mRNA in vivo, while wild-type or mutant forms of Bcd that do not affect d4EHP binding can rescue this phenotype of embryos produced by bcd females.
In a final experiment to demonstrate that the d4EHP–Bcd interaction is required for the BBR-mediated inhibition of cad translation, capped reporter mRNAs containing BBR sequences in their 3′-UTRs were used as templates for in vitro translation reactions. These assays were performed with mouse Krebs-2 cell-free translation extracts, because their activity is highly cap-dependent  and they do not express endogenous Bcd. We found that Bcd or d4EHP has no significant effect on translation of a BBR-containing mRNA when added to these extracts individually, but when added together they repress translational activity by approx. 60%, depending on a BBR in the sense strand.
Inhibition of cad mRNA translation by the d4EHP–Bcd complex demonstrates for the first time the involvement of a cellular cap-binding protein other than eIF4E in cap-dependent translational control. Furthermore, it provides a new molecular mechanism governing the formation of morphogenetic gradients in the early Drosophila embryo. 4EHP-mediated translational repression is potentially more potent than that mediated by eIF4E-binding proteins such as Cup, which have to compete with eIF4G for binding to eIF4E. This is because the eIF4E–eIF4G interaction is very stable .
In summary, we have uncovered a novel mechanism for achieving spatially regulated expression of a protein, which in turn is essential for embryonic patterning and development. We expect that future research will uncover numerous additional examples of 4EHP-mediated translational repression, in Drosophila and in other systems.
Stem Cells and Development: A Focus Topic at BioScience2005, held at SECC Glasgow, U.K., 17–21 July 2005. Edited by T. Kouzarides (Cambridge, U.K.), S. Newbury (Newcastle upon Tyne, U.K.), B. Richardson (University College London, U.K.), R. Sablowski (John Innes Centre, Norwich, U.K.), D. Tosh (Bath, U.K.), M. Welham (Bath, U.K.) and A. Willis (Nottingham, U.K.).
Present address: Department of Integrative Biology, University of California, Berkeley, CA 94720-3140, U.S.A.