For the recovery of recombinant RABV from cDNA, one has to provide the full-length cDNA from which antigenome-like RNA is transcribed into the cytoplasm. This newly made RNA is, however, not infectious per se. Since neither plus-sense nor minus-sense RNA of RABV can be translated to provide the “helper” proteins N, P and L, extra expression plasmids are necessary. Only upon encapsidation by the N protein into NCs and association of these NCs with the RABV polymerase P-L into a vRNP is the viral RNA infectious. Therefore at least 4 plasmids have to be introduced into the cell. Transcription of all 4 plasmids is usually dependent on T7-pol that can be either transiently provided by an additional plasmid or viral vectors like vaccinia virus, or be stably expressed by certain cell lines (for review see (Ghanem and Conzelmann, in press)). As a first step towards a genetically encoded RABV this work addressed the possibility of simplifying the rescue system by reducing the numbers of plasmids needed for rescue.
For influenza virus, a segmented negative stranded RNA virus from the family Orthomyxoviridae such a simplification in terms of a single plasmid system recently has been demonstrated (Zhang et al., 2009). To reconstitute an infectious influenza virus cycle, the 8 genome segments have to be delivered together with 4 proteins of the vRNP, the nucleoprotein NP and the tripartite polymerase PB1, PB2 and PA. In the initial rescue systems this was achieved by simultaneous delivery of 12 plasmids, 4 expressing the proteins and 8 delivering the segmented genome RNA (Fodor et al., 1999; Neumann et al., 1999). By using ambisense strategies, where a Pol-I promoter transcribes the negative orientated genome segment and a Pol-II promoter from the other direction transcribes an mRNA to generate the proteins, this was reduced to 8 plasmids (Hoffmann et al., 2000; Hoffmann and Webster, 2000). Zhang et al. fused these 8 plasmids to a single plasmid of approximately 24 kb from which they were able to rescue recombinant influenza virus.
For RABV rescue, the antigenome must be transcribed as naked RNA to reconstitute the virus together with N, P and L. This means the antigenome-like RNA has already the orientation of mRNAs, thus an ambisense strategy is not helpful. The fusion of the three “helper” plasmids, coding for N, P and L, in one plasmid together with the cDNA for the full- length RNA would result in a single plasmid of at least 24 kb, the same size as the influenza single rescue plasmid. The genome comprises about 12 kb, the L gene about 6.5 kb, the N gene about 1.5 kb and the P gene about 1 kb. Additionally promoter and termination sequences together with sequences for amplification and selection in bacteria would be necessary. The major difficulty, however, would not be the size, but rather the duplication of about 75 % of the sequences, as there would be two N genes, two P genes and two L genes, one of each as part of the genome and the other as expression unit for the “helper” proteins. This is expected to result in extensive recombination between the plasmid sequences in bacteria. Therefore, another approach using regulative elements on RNA basis, the IRES element was explored to overcome this problem.
Recently, the successful replacement of RABV gene borders by diverse picornaviral IRES elements has been demonstrated (Marschalek et al., 2009). There it was shown first that these internal IRES elements were tolerated in the viral genome. Further, as the gene border and thereby the transcriptional stop/restart signals were replaced by IRES elements, the translation of the downstream gene was dependent on the respective IRES element. Different IRES elements could therefore direct the relative levels of protein translated from the downstream ORF. In regard to expression of “helper” proteins these IRES elements were expected to also translate proteins from the downstream ORF from the naked full-length antigenome-like RNA directly after transcription, as this RNA has the same orientation as mRNAs. Once the genome is packaged into helical N-RNA, this should not be the case anymore, due to the loss of IRES secondary structures.
The translation of protein from an ORF with the support of an upstream IRES element indeed was confirmed. Transfection of only pSAD T7-HH_PVP(bi)_SC, the full-length cDNA for a RABV in which the NP gene border is replaced by the poliovirus IRES, resulted in expression of low levels of P protein as indicated in Western blot experiments. The replacement of the NP gene border in RABV results in translational regulation of gene expression instead of transcriptional regulation. As we did not aim primarily to influence the natural transcription
gradient, the construct pSAD T7-HH_PVP(mono)_SC was created. In contrast to pSAD T7- HH_PVP(bi)_SC, in this construct, the poliovirus IRES does not replace the NP gene border but is located in the 5’-UTR of the P gene. As this does not lead to a bicistronic N-PV-P mRNA but to a monocistronic PV-P mRNA, the abbreviations “bi” or “mono” were chosen. Like pSAD T7-HH_PVP(bi)_SC, pSAD T7-HH_PVP(mono)_SC was able to translate P protein at low levels when transfected into BSR-T7/5 cells. As the P levels still were low, other IRES
elements were tested. The EMCV IRES was found to result in the highest levels of P protein translated directly from the full-length transcript.
More interesting in regard to a simpler rescue system was the question if these P levels were sufficient to rescue the virus. And indeed they were. Transfection of pSAD T7- HH_EP(mono)_SC, together with only pTIT-N and pTIT-L, but omitting pTIT-P was sufficient to rescue the RABV SAD EP(mono) from cDNA. This demonstrated that a cis-active sequence, namely the EMCV IRES, can direct the translation of P protein in amounts sufficient for rescue and therefore supersedes the use of an extra P plasmid. The full-length plasmids pSAD T7-HH_PVP(bi)_SC and pSAD T7-HH_PVP(mono)_SC, however, comprising the poliovirus IRES, did not provide similar amounts of P protein and therefore could not be rescued without pTIT-P. When pTIT-P was co-transfected, recombinant RABV successfully could be recovered from these cDNAs.
The next step was to transfer this principle to the other two genes of RABV that need to be expressed during rescue, N and L. For the L gene this was working readily, by introducing an EMCV IRES downstream of the GL gene border into the 5’-UTR of L. The resulting cDNA construct, pSAD T7-HH_EL(mono)_SC could be rescued by co-transfection of only pTIT-N and pTIT-P, but without pTIT-L. Due to a lack of L antibodies, the levels of L protein translated directly from the full-length RNA could not be determined. As L is at the most promoter distal (5’-end) position in the NNSV context, the least mRNA is made. Moreover, viruses have been made containing two additional transcription units between the G and the L gene, probably reducing the already low amounts of L mRNA, without significantly attenuating the virus (unpublished data). Thus it is thought that already low amounts of L protein are sufficient for supporting virus replication and transcription.
When an EMCV IRES element was introduced upstream of the N gene, directly between the RABV antigenome leader and the AUG of the N gene (pSAD T7-HH_EN_SC), it was able to - 122 –
mediate translation of significant amounts of N protein from the full-length RNA. Strikingly, however, no RABV could be rescued from pSAD T7-HH_EN_SC. In pSAD T7-HH_EN_SC the EMCV IRES with its intensive secondary structure is in close proximity to the antigenomic leader sequence at the 5’-end of the newly transcribed RNA. To exclude that the leader- proximal IRES has a negative effect on virus replication, an eGFP ORF was inserted as a spacer between the leader and the EMCV IRES. As no gene border was inserted, this would result in a bicistronic GFP-IRES-N mRNA transcribed by the virus. The construct pSAD T7- HH_GFPEN_SC still was able to provide sufficient amounts of N protein translated from the N-ORF on the full-length RNA, but again could not be rescued, neither with nor without providing N in trans. The same was true when instead of the eGFP, an extra RABV P ORF was inserted (data not shown). As indicated by these results and confirmed by later experiments, cis-active sequences at the leader-N gene junction, as well as within the N coding region play a critical role for virus viablitiy. This is discussed below (4.3.5). In order to retain an authentic leader-N gene junction, a virus with a bicistronic N-IRES-N gene (SAD NEN) was constructed. This construct has the first N gene (N1) directly downstream of the leader, followed by the EMCV IRES and a second N gene (N2). In contrast to pSAD T7-HH_EN_SC, pSAD T7-HH_NEN_SC could be rescued readily into viable virus. Moreover, this rescue was possible by co-transfecting only pTIT-P and pTIT-L.
The major implication from these results is that it was demonstrated that each “helper” protein can be omitted by the use of cis-active sequences in the antigenome-like RNAs.