A panel of 15 chimeric viruses were developed here to study specific elements of the ONNV genome and to determine which of these regions are necessary for ONNV to infect An. gambiae
mosquitoes. As CHIKV virus primarily infects Aedes
species and ONNV primarily infects Anopheles
species, these two closely related viruses provide an ideal opportunity to study these viral genetic determinants of infection. This study is the first to look at the importance of ONNV non-structural proteins in An. gambiae
infection. Of the ten CHIKV-backbone chimeras constructed and tested, only the one containing ONNV nsP3 produced infection rates closer to parental ONNV than to the parental CHIKV. The ability of ONNV nsP3 to up-regulate infection rates so substantially shows that ONNV nsP3 is the main determinant of ONNV vector specificity for An. gambiae
. Interestingly, the reciprocal chimeric virus (full length-ONNV with the CHIKV nsP3) was not able to be rescued from cDNA in either mammalian or insect cells. This would further suggest that nsP3 plays a critical role in viral replication that is distinct in these two closely related viruses that exhibit 81% and 72% amino acid and nucleotide identity respectively in nsP3. That nsP3 should be found to be essential to infection is especially interesting given the fact that the precise functions of this protein are not fully defined. It is required for the correct formation and localization of replication complexes and does provide essential functions in both minus strand and subgenomic RNA synthesis, but specific mechanisms are not yet resolved 
To further add to the intrigue of this protein, it has been shown that some members of the alphavirus family actually contain inserts of foreign genetic material within nsP3. An eight amino acid sequence from the carboxyl-terminus of CHIKV nsP3 maps to a putative zinc finger protein in Ae. aegypti
, the main vertebrate vector for that virus 
. In Semliki Forest virus, a 7 amino acid sequence corresponds to elements found in a wide-range of cellular proteins 
. Numerous other examples of what may be inserts of foreign genetic material been shown by sequencing nsP3 from the following alphaviruses: CHIKV, eastern equine encephalitis virus, Semliki Forest virus, and Venezuelan equine encephalitis virus 
Alphavirus nsP3 can be clearly divided into two distinct domains. The macro domain, or amino-terminal region, is highly conserved, not just among alphaviruses but also among coronaviruses, hepatitis E virus, rubella virus and even cellular proteins 
. The carboxyl-terminus domain of the alphavirus nsP3 is highly variable in size and sequence and is devoid of any predicted secondary structure 
. Chimeric viruses were constructed using the natural division between the conserved and non-conserved regions of nsP3 to engineer two additional chimeric viruses to determine if the region conferring specificity to An. gambiae
could be attributed solely to either of the distinct domains within nsP3. Interestingly, the addition of just the carboxyl-terminus of ONNV nsP3 did produce a small, although not a statistically significant, increase in infection rates as compared with parental CHIKV in An. gambiae
The carboxyl-terminus of nsP3, which has been subject to rapid alteration during alphavirus evolution, may also be involved in the optimization of replication in diverse host cell types 
. Studies with Sindbis showed that deletions in the carboxyl-terminus rendered mutants defective at initiating a productive infection, generating plaques in mosquito cells at only 1–2% the efficiency of the parental virus 
. A recent study noted a carboxyl-terminus, proline-rich sequence motif, the PIPPPR motif, shared by many alphavirus nsP3 proteins and demonstrated that even a single mutation in this region of Semliki Forest virus or Sindbis virus greatly impaired RNA synthesis by disrupting binding with host cell amphiphysins 
. It is possible that this motif also modulates ONNV vector specificity since ONNV and CHIKV do differ from one another by one amino acid in this PIPPPR region. Attenuated virulence and reduced rates of RNA synthesis and virus replication were also seen in vertebrate cells with Semliki Forest virus mutants lacking some portion of the carboxyl-terminus of nsP3 
. Yet, studies in mammalian cell lines showed that a 34 amino acid deletion in this region of nsP3 in Venezuelan equine encephalitis had no detectable effect on replication 
. Collectively, these studies support the current finding that nsP3 can be vital for productive infection, but in a manner that is host and virus specific.
Another interesting characteristic of the carboxyl-terminus of nsP3 is that it is phosphorylated at multiple serine and threonine residues 
. The role of this phosphorylation is not exactly clear, except that it does seem to modulate the efficiency of minus-strand RNA synthesis 
. Determination of the exact mechanisms of this modulation and the mechanisms for the host-specific effects seen with nsP3 mutants in this and other studies would be extremely valuable information and allow for design of further studies. Furthermore, our studies show that an intact ONNV nsP3 is required for ONNV-like infection rates, and that dividing the region either disrupts a vital interaction between the two or removes an element necessary for An. gambiae
infection. The former seems more probable since substituting CHIKV for either half of ONNV nsP3 results in infection rates not significantly different from rates with parental CHIKV.
While molecular determinants residing in nsP3 did turn out to be the most dramatic finding of our study, we did also examine the structural regions of the genome. Previously published studies by another group had suggested that all of the viral structural proteins are necessary for ONNV to infect An. gambiae
. Our study was able to provide critical fine tuning to this conclusion. In our experiments with CHIKV-backbone chimeras containing various regions of the ONNV structural proteins, each maintained parental CHIKV-like infection profiles despite containing portions of the ONNV genome. In fact, even an intact ONNV structural region was not sufficient for infection of An. gambiae
, as shown with the chimera CHIK/ONN eSTR (). The reciprocal chimeras, substituting sections of CHIKV structural regions for the like section of ONNV structural genes, in most cases, did not greatly reduce mosquito infection rates. The notable exceptions were in the chimeras that divided the ONNV structural region in half. Both ONN/CHIK 5′STR and ONN/CHIK 3′STR were significantly less infectious to An. gambiae
than was parental ONNV. However, since the reciprocal chimeras, CHIK/ONN 5′STR and CHIK/ONN 3′STR, did not show up-regulated infection rates, the drop in infection with the chimeric viruses is likely due to disruption of one or more virus-virus or virus-host interactions.
In alphaviruses, the extreme 3′ terminus of the genome, just preceding the poly(A) tail, has a sequence which is highly conserved among all alphaviruses and which is absolutely required for efficient virus replication 
. This 19-nucleotide sequence is identical in CHIKV and ONNV so this could not have played a role in the decreased infection rates seen with ONN/CHIK 3′STR. However, studies using Sindbis mutants with large deletions in the 3′non-translated region (NTR) have shown that the rest of the 3′ NTR is also important for virus replication in a host-specific manner 
. ONNV is 156 additional nucleotides shorter in the 3′NTR when compared to CHIKV; this size difference alone could result in conformational changes resulting in the inability of the virus to interact with itself or with host proteins. Of note is the design of our eSTR and 3′STR structural clones, which contain the indicated structural region as well as the 3′ NTR from the non-parental virus (); this design is different from those previously described 
and may suggest the possibility of multiple interactions within the proteins or gene sequences of the virus itself that may have a minor role in the overall ability of a chimeric virus to replicate within the mosquito. It is further possible that the differences in CHIKV and ONNV conserved sequence elements 
are sufficient to undercut RNA stability, resulting in greatly reduced mosquito infection patterns.
Studies with chimeric viruses must be viewed in the overall context of the virus' life cycle. When substitutions are made to construct chimeric viruses, numerous aspects of the virus-host interactions and virus-virus regulatory functions can be disrupted resulting in reduced infection rates. Reduced infection rates may be a direct consequence of missing the essential genomic region or may be an indirect result of disrupting an essential regulatory interaction. Conversely, when the addition of a specific region increases mosquito infection rates, we must conclude the region itself to be essential for infection. Interestingly, because there was such a low dissemination rate of all viruses within this study, elements involved in dissemination throughout the mosquito may be distinct from those important in initial infection. However, this study has shown that ONNV nsP3 is directly responsible for ONNV infection of An. gambiae. There are also numerous interactions within nsP3 itself, within the two halves of the structural region, and possibly the 3′ NTR which, when disrupted, can eliminate mosquito infection.