An AUG codon conserved for protein function rather than translational initiation: the story of the protein sElk1

Preinitiation Complex Loading onto mRNAs with Long versus Short 5' TLs

The first step in translation initiation consists in the recruitment of the small ribosome onto the mRNA. This preinitiation complex (PIC) loads via interactions with eIF4F that has assembled on the 5' cap. It then scans the 5' TL (transcript leader) to locate a start site. The molecular architecture of the PIC-mRNA complex over the cap is beginning to be resolved. As part of this, we have been examining the role of the 5' TL length. We observed in vivo initiation events on AUG codons positioned within 3 nts of the 5' cap and robust initiation in vitro at start sites immediately downstream of the 5' end. Ribosomal toe-printing confirmed the positioning of these codons within the P site, indicating that the ribosome reads from the +1 position. To explore differences in the eIF4E-5' cap interaction in the context of long versus short TL, we followed the fate of the eIF4E-cap interaction using a novel solid phase in vitro expression assay. We observed that ribosome recruitment onto a short TL disrupts the eIF4E-cap contact releasing all the mRNA from the solid phase, whereas with a long the mRNA distributes between both phases. These results are discussed in the context of current recruitment models.

What Is the Impact of mRNA 5' TL Heterogeneity on Translational Start Site Selection and the Mammalian Cellular Phenotype?

A major determinant in the efficiency of ribosome loading onto mRNAs is the 5′ TL (transcript leader or 5′ UTR). In addition, elements within this region also impact on start site selection demonstrating that it can modulate the protein readout at both quantitative and qualitative levels. With the increasing wealth of data generated by the mining of the mammalian transcriptome, it has become evident that a genes 5′ TL is not homogeneous but actually exhibits significant heterogeneity. This arises due to the utilization of alternative promoters, and is further compounded by significant variability with regards to the precise transcriptional start sites of each (not to mention alternative splicing). Consequently, the transcript for a protein coding gene is not a unique mRNA, but in-fact a complexed quasi-species of variants whose composition may respond to the changing physiological environment of the cell. Here we examine the potential impact of these events with regards to the protein readout.

The effect of heterogeneous Transcription Start Sites (TSS) on the translatome: implications for the mammalian cellular phenotype

BACKGROUND

The genetic program, as manifested as the cellular phenotype, is in large part dictated by the cell's protein composition. Since characterisation of the proteome remains technically laborious it is attractive to define the genetic expression profile using the transcriptome. However, the transcriptional landscape is complex and it is unclear as to what extent it reflects the ribosome associated mRNA population (the translatome). This is particularly pertinent for genes using multiple transcriptional start sites (TSS) generating mRNAs with heterogeneous 5' transcript leaders (5'TL). Furthermore, the relative abundance of the TSS gene variants is frequently cell-type specific. Indeed, promoter switches have been reported in pathologies such as cancer. The consequences of this 5'TL heterogeneity within the transcriptome for the translatome remain unresolved. This is not a moot point because the 5'TL plays a key role in regulating mRNA recruitment onto polysomes.

RESULTS

In this article, we have characterised both the transcriptome and translatome of the MCF7 (tumoural) and MCF10A (non-tumoural) cell lines. We identified ~550 genes exhibiting differential translation efficiency (TE). In itself, this is maybe not surprising. However, by focusing on genes exhibiting TSS heterogeneity we observed distinct differential promoter usage patterns in both the transcriptome and translatome. Only a minor fraction of these genes belonged to those exhibiting differential TE. Nonetheless, reporter assays demonstrated that the TSS variants impacted on the translational readout both quantitatively (the overall amount of protein expressed) and qualitatively (the nature of the proteins expressed).

CONCLUSIONS

The results point to considerable and distinct cell-specific 5'TL heterogeneity within both the transcriptome and translatome of the two cell lines analysed. This observation is in-line with the ribosome filter hypothesis which posits that the ribosomal machine can selectively filter information from within the transcriptome. As such it cautions against the simple extrapolation transcriptome → proteome. Furthermore, polysomal occupancy of specific gene 5'TL variants may also serve as novel disease biomarkers.

The impact of the phosphomimetic eIF2αS/D on global translation, reinitiation and the integrated stress response is attenuated in N2a cells

A plethora of stresses trigger a rapid downregulation of protein synthesis. However, a fraction of mRNAs continue to be recruited onto polysomes and their protein products play a key role in deciding cell fate. These transcripts are characterized by the presence of uORFs within their 5' TL coupling protein expression to reinitiation. The translational brake arises due to the activation of a family of kinases targeting the α subunit of the trimolecular eIF2(αβγ) initiation factor. Phosphorylation of eIF2αSer51 inhibits ternary complex regeneration reducing the pool of 43S ribosomes. It is popular to mimic this event, and hence the integrated stress response (ISR), by the expression of the phosphomimetic eIF2αS51D. However, we report that whereas the ISR is reproduced by eIF2αS51D expression in human HEK293T cells this is not the case in N2a mouse neuroblastoma cells. With regards to translational downregulation, this arises due to the failure of the phosphomimetic protein to assemble an eIF2 complex with endogenous eIF2β/γ. This can be compensated for by the transient co-expression of all three subunits. Curiously, these conditions do not modulate reinitiation and consequently fail to trigger the ISR. This is the first demonstration that the inhibitory and reinitiation functions of eIF2αS/D can be separated.