Efficient eukaryotic translation is initiated upon recognition of the mRNA 5' cap structure by initiation factors and binding of the small ribosomal subunit to the 5' end of the mRNA. However, many viral mRNAs have alternatives to the cap and poly(A) tail that ensure efficient viral translation at the expense of host protein synthesis. Specifically, plant viral RNAs in the Luteovirus genus, including that of Barley yellow dwarf virus (BYDV) harbor a cap-independent translation element (CITE) located in the 3' untranslated region (UTR) of viral genome. The BYDV-like translation element (BTE) is defined by the presence of a conserved 17 nt sequence (17 nt CS) GAUCCCUGGGAAACAGG that adopts a stem-loop with paired underlined bases and a loop that can base pair to a complementary bases in the 5'UTR. The BTE promotes translation initiation at the 5'-proximal AUG on viral RNAs by recruiting translation machinery and delivering it to the 5' end via a kissing stem-loop.

The key to understanding how the BTE outcompetes the host mRNA for protein synthesis machinery is to determine the structure of the BTE at atomic resolution. Thus, we conducted phylogenetic and mutagenesis analysis of BYDV and BTEs from diverse viruses to identify the most structurally homogeneous sequences before subjecting them to extensive crystallization screening. In this process, we discovered that Groundnut rosette virus (GRV) in genus Umbravirus also harbors a functional BTE in its genome despite violating 17 nucleotide (nt) consensus. We also demonstrated that BTE sequences are capable of crystallization under a variety of conditions suggesting that they all adopt a compact fold. One of the screened BTEs, the 87c BTE RNA gave crystals that diffracted initially to 30 y.

Upon crystallization condition optimization, we obtained crystals that diffract below 5y, with a complete data set collected to 6.9 y. This crystal form indexes with an Rmerge of 0.094 in the monoclinic space group C2 with unit-cell parameters a=316.6 y, b=54.2 y, c=114.5 y, alpha=beta=90y, gamma =105.1y.

Despite screening thousands of conditions and dozens of sequence variants, I have been unable to obtain BTE crystals that diffract at less than 4.5y. Thus, I switched to an entirely different class of structure, the Pea enation mosaic virus RNA 2-like translation enhancer (PTE). The PTE binds and requires smaller translation initiation factor eIF4E for which crystal structure is known. Although the PTE alone did not crystallize, a 1:1 mixture of eIF4E with PTE gave large crystals that diffracted at 2.3y resolution. Data analysis revealed that PTE-eIF4E RNA crystals belong to the same space group and have near identical unit cell parameters as eIF4E crystals suggesting that eIF4E simply crystallized free of the RNA despite being presence of the RNA in the mother liquor.

All these elements utilize unique magnesium dependent folds to bind the rate-limiting eIF4F complex, but the binding site for most CITE has not been mapped out yet. Thus, we used chemical and enzymatic probing to map the eIF4G-binding site on the BTE. Footprinting experiments revealed that eIF4G alone is capable of protecting most of the BYDV BTE structure from chemical and enzymatic modification. This BTE protection pattern is even more exaggerated in the presence of heterodimeric eIF4F complex suggesting binding to a very large area of the BTE with the exception of loop three. This loop known to base pair to the 5' UTR remained solvent accessible even at high (1 micromolar) eIF4G/4F concentrations. This data fit the proposed model where the eIF4F complex binds directly to the BTE and once recruited, it is delivered to the 5'UTR via a long distance kissing interaction