Mitigation of cassava mosaic disease (CMD) focuses on the introgression of resistance imparted by the polygenic recessive (CMD1), dominant monogenic (CMD2) and CMD3 loci. The mechanism(s) of resistance they impart, however, remain unknown. Two CMD susceptible and nine CMD resistant cassava genotypes were inoculated by microparticle bombardment with infectious clones of African cassava mosaic virus Cameroon strain (ACMV-CM) and the Kenyan strain K201 of East African cassava mosaic virus (EACMV KE2 [K201]). Genotypes carrying the CMD1 (TMS 30572), CMD2 (TME 3, TME 204 and Oko-iyawo) and CMD3 (TMS 97/0505) resistance mechanisms showed high levels of resistance to ACMV-CM, with viral DNA undetectable by PCR beyond 7days post inoculation (dpi). In contrast, all genotypes initially developed severe CMD symptoms and accumulated high virus titers after inoculation with EACMV KE2 (K201). Resistant genotypes recovered to become asymptomatic by 65dpi with no detectable virus in newly formed leaves. Genotype TMS 97/2205 showed highest resistance to EACMV KE2 (K201) with <30% of inoculated plants developing symptoms followed by complete recovery by 35dpi. Deep sequencing of small RNAs confirmed production of 21-24 nt virus derived small RNAs (vsRNA) that mapped to cover the entire ACMV-CM and EACMV KE2 (K201) viral genomes in both polarities, with hotspots seen within gene coding regions. In resistant genotypes, total vsRNAs were most abundant at 20 and 35dpi but reduced significantly upon recovery from CMD. In contrast, CMD susceptible genotypes displayed abundant vsRNAs throughout the experimental period. The percentage of vsRNAs reads ranked by class size were 21nt (45%), 22 nt (28%) and 24 nt (18%) in all genotypes studied. The number of vsRNA reads directly correlated with virus titer and CMD symptoms.

The cap-independent translation of plus-strand RNA plant viruses frequently depends on 3′ structures to attract translation initiation factors that bind ribosomal subunits or bind directly to ribosomes. Umbraviruses are excellent models for studying 3′ cap-independent translation enhancers (3′CITEs), as umbraviruses can have different 3′CITEs in the central region of their lengthy 3′UTRs, and most also have a particular 3′CITE (the T-shaped structure or 3′TSS) near their 3′ ends. We discovered a novel hairpin just upstream of the centrally located (known or putative) 3′CITEs in all 14 umbraviruses. These CITE-associated structures (CASs) have conserved sequences in their apical loops and at the stem base and adjacent positions. In 11 umbraviruses, CASs are preceded by two small hairpins joined by a putative kissing loop interaction (KL). Converting the conserved 6-nt apical loop to a GNRA tetraloop in opium poppy mosaic virus (OPMV) and pea enation mosaic virus 2 (PEMV2) enhanced translation of genomic (g)RNA, but not subgenomic (sg)RNA reporter constructs, and significantly repressed virus accumulation in Nicotiana benthamiana. Other alterations throughout OPMV CAS also repressed virus accumulation and only enhanced sgRNA reporter translation, while mutations in the lower stem repressed gRNA reporter translation. Similar mutations in the PEMV2 CAS also repressed accumulation but did not significantly affect gRNA or sgRNA reporter translation, with the exception of deletion of the entire hairpin, which only reduced translation of the gRNA reporter. OPMV CAS mutations had little effect on the downstream BTE 3′CITE or upstream KL element, while PEMV2 CAS mutations significantly altered KL structures. These results introduce an additional element associated with different 3′CITEs that differentially affect the structure and translation of different umbraviruses.