A synthetic defective interfering SARS-CoV-2
Abstract
Viruses thrive by exploiting the cells they infect, but in order to replicate and infect other cells they must produce viral proteins. As a result, viruses are also susceptible to exploitation by defective versions of themselves that do not produce such proteins. A defective viral genome with deletions in protein-coding genes could still replicate in cells coinfected with full-length viruses. Such a defective genome could even replicate faster due to its shorter size, interfering with the replication of the virus. We have created a synthetic defective interfering version of SARS-CoV-2, the virus causing the Covid-19 pandemic, assembling parts of the viral genome that do not code for any functional protein but enable the genome to be replicated and packaged. This synthetic defective genome replicates three times faster than SARS-CoV-2 in coinfected cells, and interferes with it, reducing the viral load of infected cells by half in 24 hours. The synthetic genome is transmitted as efficiently as the full-length genome, suggesting the location of the putative packaging signal of SARS-CoV-2. A version of such a synthetic construct could be used as a self-promoting antiviral therapy: by enabling replication of the synthetic genome, the virus would promote its own demise.
Discussion
DI particles have long been known to virologists (Gard et al., 1952; Huang & Baltimore, 1970) and their use in unravelling the location of functional elements of a genome is well known. Our synthetic DIs suggest that a disputed (Masters, 2019) putative packaging sequence of SARS-CoV-2 could indeed enable packaging of our synthetic defective genome –and therefore presumably acts as a packaging signal for the WT genome. However, because the difference between our DI1 and DI0 synthetic constructs is not limited to the portion with the putative packaging signal (part of nsp15), we cannot rule out that packaging signals reside in the other parts of the DI1 genome that DI0 lacks, most notably a conserved region (28554–28569) with a SL5 motifs in the N partial sequence included in the DI1 genome but not in the DI0 genome. It is also possible that DI0 can be packaged but because it does not replicate efficiently, it is rapidly degraded after transfection and the amount of packaging does not meet the threshold for detection.
The interference with the WT virus is the most remarkable effect of our DI1 construct. As we have shown, while DI0 does not interfere significantly with WT, DI1 induces a reduction of about 50% in the amount of WT virus in coinfections compared to infections with WT alone, and this is likely due to the faster replication of the DI1 genome. DI particles are often described as by-products of inaccurate replication or as having a regulatory function for a viral quasi-species. However, DIs can also be seen as defectors in the sense of evolutionary game theory (Szathmáry, 1994; Turner & Chao, 1999, Brown, 2001): ultra-selfish replicators, able to freeride as parasites of the full-length genome. As such, DI particles need not serve any purpose for the WT virus.
Indeed, DIs could be used as antivirals: by virtue of their faster replication in cells coinfected with the WT virus, DI genomes can interfere with the virus. Potentially, as the DI genomes increase in frequency among the virus particles pool, the process becomes more and more effective, until the decline in the amount of WT virus leads to the demise of both virus and DI. A similar therapeutic approach has been proposed for bacteria (Brown et al., 2009) and cancer (Archetti, 2013). The potential of DIs as antivirals has been suggested before (Marriott & Dimmock, 2010; Dimmock & Easton, 2014; Vignuzzi & López, 2019), and a synthetic DI particle could perhaps be immune to the evolution of resistance (although coevolution of viruses and DIs has been shown in Rhabdoviridae (Horodyski, Nichol & Spindler, 1983). Unlike, for example, HIV and influenza, which are perhaps not ideal candidates because of their short genome, multiple genomic fragments and complex replication process, coronaviruses may be more amenable to DI therapy because of their long single fragment genome and relatively simple life cycles. While the immediate 50% reduction in virus load we observed is arguably not enough for therapeutic purposes, efficacy would compound over time (as the DIs increase in frequency) and a higher initial efficacy could be obtained using a delivery vector and an improved version of the DI genome.
Conclusions
We have established a proof of principle that a synthetic defective interfering SARS-CoV-2 can replicate in cells infected with the virus and interfere with its replication. Further experiments are needed to verify the potential of SARS-CoV-2 DIs as antivirals. Our experiments should be repeated in human lung cell lines, against other variants of SARS-CoV-2 and by transfecting DI RNA after infection, a more realistic simulation of therapy, which will, however, ultimately require in vivo experiments. An efficient delivery method should be devised to increase the initial amount of DI RNA and to deliver it in vivo. It would also be interesting to measure how the fraction of DI and WT genomes changes over time to test whether the DI genome drives the WT genome to extinction, or they coexist at a mixed equilibrium. Finally, it would be useful to analyse the long-term evolution of coinfections to test how SARS-CoV-2 and its DIs coevolve and whether resistant mutants can arise