ANALISIS DE RESULTADOS

As RABV delivered shRNA precursors could not be processed by nuclear enzymes of the miRNA or shRNA pathway we aimed at producing RABV transcripts resembling Dicer substrates. In the miRNA pathway, mostly Pol-II transcribed pri-miRNAs are processed by the nuclear Drosha/DGCR8 complex into about 60 nts long pre-miRNA hairpins that contain now a 3’- 2 nts overhang at the cleavage site. Exported into the cytoplasm by Exportin 5 (Lund et al., 2004), they are further processed by Dicer into RNA duplexes from which one strand, the passenger strand, that has the same sequence as the target, is removed, whereas the so- called guide strand being antisense to the target site (which is located primarily in the 3’-UTR of mRNAs) is loaded into the RISC to repress translation from this mRNA (reviewed in Bartel, 2004; Cullen, 2004; Li and Liu, 2011).

Short hairpin RNAs have either been made synthetically (Paddison et al., 2004; Siolas et al., 2005) or expressed from Pol-II (Zhou et al., 2005) and Pol-III (Paddison et al., 2004) promoters. Depending on their origin they enter into the processing pathways at different stages. The Pol-II derived shRNAs for example have a 5’-cap structure and a 3’-poly(A) tail and therefore, like pri-miRNAs, must be cleaved by Drosha/DGCR8. Pol-III transcribed shRNAs however start with a 5’-ppp and are designed to terminate exactly in such a way that

they have a 2 nt overhang at their 3’-end upon hairpin formation. Therefore, they resemble already the Drosha/DGCR8 products that only differ in having a monophospate at their 5’- end, and only have to be exported by Exportin 5 from the nucleus. Synthetic shRNA have a 5’-monophosphate and are delivered i.e. by lipofection. They, once in the cytoplasm, directly can be processed by Dicer.

The first approach to achieve RABV delivered Dicer substrates was to cleave off the 5’-cap and the 3’-poly(A) tail by cis-active ribozymes. For removal of the 5’-cap an HHRz was chosen that has been shown to produce an exact 5’-end for rescue of measles virus and BDV (Martin et al., 2006). HHRzs were first described as autocatalytic RNAs in plant viroids (Buzayan et al., 1986; Hutchins et al., 1986; Prody et al., 1986). All HHRz share structural similarities like the presence of three stems and a catalytic pocket. Cleavage produces a cyclic 2’, 3’- monophosphate at the cleaved off ribozyme end and a 5’-OH at the processed RNA (Birikh et al., 1997; Uhlenbeck, 1987; van Tol et al., 1990).

In vitro-tests conducted in this work with the HHRz revealed its functionality and efficiency as we observed about 90 % of the RNA to be processed. The HHRz was placed in an orientation that allowed its correct folding and cleavage in RABV mRNAs (or in the antigenome). Although during rescue of the virus the antigenome-like RNA transcribed by T7-pol initially lies naked in the cytoplasm before its encapsidation by N protein, it was possible to generate recombinant RABV containing the HHRz sequence. The rescue of this critical virus did not seem to be impaired significantly in comparison with wild-type virus. This might indicate a lower in vivo-activity of the ribozyme. However in the second part of this thesis, it is shown that an HHRz processing the 5’-end of the antigenome-like RNA in vivo significantly increases the rescue efficiency and minigenome activity (Ghanem et al., 2012). Also as shown in part 3 of this thesis, IRES mediated translation from a Pol-II transcript decreased significantly, dependent on the ribozymes. These findings emphasize the in vivo- activity of the ribozyme although not allowing a direct quantitative comparison with the in vitro-activity.

Analogously to the 5’-cap structure the 3’-poly(A) tail should be cleaved off from the RABV derived mRNA by a ribozyme. Therefore an HDVagRz was applied. The HDVagRz was discovered as a self-cleaving RNA sequence in the Hepatitis Delta virus antigenome. The

cleavage reaction it produces a cyclic 2’, 3’-monophosphate at the processed RNA and a 5’- OH at the cleaved off ribozyme end (Sharmeen et al., 1988).

The HDVagRz I tested first was the 84 nt “core” sequence that has been used to process the 3’-end of the antigenome-like RNA in the first rescue system for RABV developed in our lab (Schnell et al., 1994). This “core” sequence initially was reported to be the HDVagRz (Perrotta and Been, 1991), however in more recent publications longer sequences are considered to constitute the HDVagRz (Perrotta and Been, 1998; Perrotta et al., 1999). The initial “core” sequence was abbreviated in this work as HDV and the two newer versions we chose, “supercut 1” (SC1) and “supercut 2” (SC2) as we found both to cleave significantly better in vitro.

The HDVagRzs again were inserted in an orientation that allowed cleavage to occur (either in the antigenome or) in the viral mRNAs generated from the extra transcription unit between the G and the L gene. Interestingly, RABV containing internal SC1 or SC2 could only be rescued after the significant improvement of the rescue system (Ghanem et al., 2012) shown in the second part of this thesis. This strongly indicated the in vivo-activity of SC1 and SC2. A further proof was the significantly increased rescue efficiency itself, which depended on the replacement of HDV with SC1 to process the 3’-end of the RABV antigenome-like RNA. In summary, evidence was provided that it is possible to generate recombinant RABV containing diverse ribozymes within their genome. These ribozymes may interfere with virus rescue, when they cleave the naked full-length RABV RNA. Once the full-length RNAs are encapsidated by N protein and rescued into virus, however, the ribozymes can only fold in the virally transcribed mRNAs, but not in the tightly packaged RABV N-RNA genome or antigenome.