A potential mechanism for Wolbachia pathogen blocking

Wolbachia was dragged out of the obscurity of phenomenology literature and into the limelight of medical relevance by two high impact publications in 2008 (here and here). Both revealed that Wolbachia can protect insects (in this case, Drosophila melanogaster) from RNA viruses, resulting in a reduction of viral titers during the course of infection or increasing host survival.  This trait has obvious vector control applications and indeed, Wolbachia-infected mosquitos are being released in different parts of the world to control the spread of Dengue, among other human pathogens.  Although we are currently using Wolbachia to control arboviruses, we do not yet understand the mechanism by which Wolbachia confers pathogen protection.  This week, Rainey et al published a potential mechanism in our favorite journal, PLoS Pathogens. I was excited to read it and share the figures with y'all (the actual figures, not the mixed up ones originally published, and thanks for Alain for sending them along).

In Rainey et al., the authors use a Wolbachia-infected Drosophila cell line (Jw18 of Serbus and Sullivan fame), infected with Semliki Forest virus (SFV), a positive stranded, RNA virus - part of the alpha virus family.  They first showed that Jw18 cells are competent for SFV infection. To do this they are using a reporter construct, which expresses luciferase. One very nice thing about the SFV system is the suite of genetic tools that you can use to query which part of the virus is being transcribed and translated. Also, because they are using Drosophila, they can introduce a replicase system into the flies and expressed using the Drosophila actin promotor. But I am getting ahead of myself! Let’s look at Figure 1 below.

Fig 1. Virus, replicon and transreplicase systems used in this study.

(A) Schematic representation of genome of SFV4(3H)-RLuc, carrying the RLuc reporter gene flanked by duplicated nsP2-protease cleavage sites at the nsP3/4 junction. Note that the genome is split into two major ORFs, 1 and 2, encoding non-structural and structural proteins respectively. (B) Schematic representation of the genome of viral replicon pSFV1(3F)RLuc-SG-FFLuc, where RLuc is fused to the region encoding for nsP3 and the structural genes have been replaced by the reporter gene firefly luciferase (FFLuc). Expression of FFLuc occurs only from subgenomic RNA produced from the subgenomic promoter; hence detection of this marker is dependent on the active replication of transfected RNA. (C-D) Schematic representation of the SFV-derived transreplicase constructs used in this study. Expression of the replicase proteins is under the control of the Drosophila Actin promoter (C). Expression of SFV template RNA is also under the control of the Actin promoter (D). When the replicase proteins are expressed this leads to active replication of the template RNA. FFLuc expression is therefore under both the control of the Actin promoter and the SFV genomic promoter. Whereas Gluc is exclusively under the control of the subgenomic promoter and therefore requires active replication of the template RNA in order for expression to occur. Two replicase constructs were used in this study: one functional, and one non-functional due to the insertion of a GDD-GAA mutation in nsP4 as indicated in (C).

They then went on to use these constructs in Jw18 cells with and without Wolbachia using the construct in Figure 1A, where luciferase is a proxy for virus replication (or infectivity).  In Figure 2 they present their data - let's pass by 2A&B for now (which just show that this arbovirus is not affecting Jw18 growth - as expected) and focus on 2C.  

 

Fig 2. Analysis of cell growth and infection dynamics of Jw18Wol and Jw18Free cells infected with SFV.

(A) Jw18Free cells were infected with SFV(3H)-RLuc at an MOI of 20. Cells were lysed 4, 8, 12 and 24 hpi and RLuc activity determined. The figure represents data from three independent experiments, where each treatment was carried out in triplicate. (B) The density of Jw18Wol and Jw18Free cells 7 and 24 hpi with SFV(3H)-RLuc at an MOI of 20; 0 (seed) is 24 h prior to infection. Data represents three independent experiments carried out in duplicate. (C) Jw18Wol and Jw18Free cells were infected at an MOI of 20 with SFV4(3H)-RLuc and RLuc activity was measured at 7 and 24 hpi. The graph indicates the mean ratio of RLuc activity in Jw18Wol and Jw18Free cells, where Jw18Wol at 7hpi and 24 hpi is equal to one. The data represents five independent experiments carried out in duplicate. Error bars represent the standard error of mean in all figures. Stars indicate significance P = <0.05 in T-Test analysis.

It looks like the Jw18 with Wolbachia was able to repress RNA virus replication - by 2-3 fold, is what they say in text.  In and of itself, that is not surprising as the pathogen blocking effect has been observed in cell lines before.  However, but look at the time course. At only 7 hpi, they see an effect of Wolbachia.  This means that Wolbachia is blocking early in the infection.

As I said before, one of the cool things about their model system is that they can then use the genetic tools available to examine where the block in replication occurs. To disentangle effects of viral entry or exit from replication within the cell, they use a clever transfection of a SFV replication construct (Figure 1C).   Because they will transfect the construct in, defects observed in their readout (in this case, the luciferase reporter) can be interpreted as problems with translation and/or transcription, not entry.  

Fig 3. The effect of Wolbachia on SFV replicon and transreplicase activity.

(A, B) Jw18Wol and Jw18Free cells were transfected with in vitro transcribed SFV1(3F)RLuc-SG-FFLuc RNA, and both RLuc (A) and FFLuc (B) activity was measured 24 h post transfection (hpt). Graphs indicate mean fold change of measurements of luciferase activity where activity in Jw18Wol cells is taken as 1 and represent three independent experiments carried out in triplicate. RLuc activity represents translation of genome RNAs whereas FFLuc indicates translation of the subgenomic mRNA. Error bars represent standard error of mean. Stars indicate significance where P = <0.05 in T-Test analysis. (C, D) Jw18Wol and Jw18Free cells were transfected with two plasmids: one expressing SFV replicase proteins (wt or mutant, under the control of the D. melanogaster Actin promoter) and one expressing viral template and both FFluc (C) and Gluc (D) activity were measured 24 hpt. FFLucactivity represents translation of RNA produced from the Actin promoter (and to some extent the genomic promoter) and Gluc activity represents translation of RNA produced from the subgenomic promoter following replication. Graphs represent relative luciferase activity and represent three independent experiments carried out in duplicate. Stars indicate significance where P = <0.01 in T-Test analysis.

In Figure 3AB you can see a reduction in the amount of luciferase produced from each construct with Wolbachia present. Either RLuc (representing translation of genome RNAs) and FFLuc (translation of subgenomic mRNA) are affected by Wolbachia.  Interestingly, though, the amount of luciferase activity suppressed in the subgenomic mRNA is MUCH more than expected based on their result for the genomic promotor.  This makes it seem like something is happening to the mRNA between the translation of the first open reading frame and translation of the subgenomic promoter's mRNA (using the construct as proxy).

Then they decouple viral replicase production from translation of these two open reading frames.  They introduce the viral replicase on a separate plasmid, in addition to the constructs tagged with luciferase, and ask, if we express this replicase in trans, does that rescue the defect in the presence of Wolbachia? The answer? Nope. Wolbachia can still block virus replication, and strongly when the subgenome is the source of the reporter, even when the replicase is provided and driven with actin (Figure 3D). Cool! 

They next explore two often touted hypotheses for the mechanism of Wolbachia pathogen blocking. First, the manipulation of small, interfering RNAs by the Wolbachia infection.  Indeed, previous work has suggested that these small RNAs, upregulated during a Wolbachia infection, may specifically antagonize viruses (although see here, where they show a functional siRNA pathway is not necessary for pathogen blocking in Drosophila).  Rainey et al. sequence the small RNAs from the Jw18 cell lines with and without Wolbachia during a virus infection with SFV.  They find no evidence for the induction of small interfering RNAs in the presence of Wolbachia, perhaps because of the reduced replication overall. As a side note, they also looked at overall miRNA patterns in the cell lines with and without Wolbachia and saw no significant difference, again, suggesting that this is not the mechanism. Finally, they also looked at Wolbachia gene expression when challenged with SFV. One might hypothesize that Wolbachia somehow senses the virus and alters its transcription to deal with the invading organism. Again, a negative result here - in this model system, Wolbachia does not seem to be changing its behavior in response to virus.  

Fig 4. The effect of Wolbachia on the production of virus-derived small interfering RNAs in Jw18 cells.

(A-B) The length and first nucleotide distribution of small RNAs mapping to genome (upper bars, 5′-3′ orientation) or antigenome (lower bars, 3’-5’ orientation) of the SFV genome in the (A) absence (Jw18Free) or (B) (Jw18Wol) presence of Wolbachia. A = red, C = green, G = blue and T = pink. Data are from 24 hpi with SFV4(3H)-RLuc, of Jw18 D. melanogaster cells. Concatenated data from 5 independent infections are shown in all panels.

OK, so what can we take away from this manuscript?  It was done in a completely artificial system, true, but one in which the pathogen blocking phenomenon works.  Caveats aside (cell lines, non-Drosophila virus, etc), they can recapitulate the phenotype in this system.  Therefore, they conclusively show that pathogen blocking by Wolbachia acts early during infection.  The response they observe in this system is perhaps not dependent on changes in the siRNAs produced by the host or by a Wolbachia response (although one could argue that a more complicated picture would emerge in whole animals or tissues, or dare I say it, mosquitos).

Which leads them to their final conclusions - pathogen blocking is an intrinsic result of the Wolbachia infection. This is something you might have suspected anyway, given the tissue specific data from the O'Neill group and the Wolbachia titer associated results from Teixeira.  We still don't know exactly how Wolbachia is blocking virus replication early but this is a great system in which to ask these questions.