Escape from RNA-guided targeting is set by whether a virus can recombine, and where it cannot, a host editing enzyme can supply the escape
lab-led
Replication in Cells of Hematopoietic Origin Is Necessary for Dengue Virus Dissemination
lab-ledDiscovery Discovery
Inserting hematopoietic-specific miR-142 target sites into the dengue virus 3-prime untranslated region excludes replication from macrophages and dendritic cells in vivo and abolishes spread to other tissues, identifying hematopoietic cells as the dominant amplification compartment.
- Senior authors
- Benjamin R. tenOever
- Correspondence
- Benjamin R. tenOever
Research areas & themes
Programmable VirologyMicroRNA-Mediated Viral AttenuationCell-Type Restriction as an Experimental Tool
Citation
Pham AM, Langlois RA, tenOever BR. Replication in Cells of Hematopoietic Origin Is Necessary for Dengue Virus Dissemination. PLoS Pathogens. 2012. Volume 8, Issue 1, article e1002465.
DOI 10.1371/journal.ppat.1002465. PMID 22241991. PMCID PMC3252368.
One-sentence contribution
Inserting hematopoietic-specific miR-142 target sites into the dengue virus 3-prime untranslated region excludes replication from macrophages and dendritic cells in vivo and abolishes spread to other tissues, identifying hematopoietic cells as the dominant amplification compartment.
Executive summary
Dengue virus is known to replicate in dendritic cells and macrophages, but whether other cell types serve as productive reservoirs during infection has been hard to settle. The difficulty is a detection problem. Viral antigen or RNA found in a tissue can reflect active replication, or entry without replication, or uptake of virus or infected debris by that cell. Studies reporting endothelial cell infection have been questioned on exactly these grounds.
The study takes a genetic rather than a detection-based approach. Four tandem target sites for miR-142, a microRNA abundant in hematopoietic cells and absent elsewhere, were inserted into the variable region of the dengue virus 3-prime untranslated region downstream of the NS5 coding frame. The resulting virus should be silenced wherever miR-142 is present and should replicate normally everywhere else, so the hematopoietic compartment can be subtracted from an otherwise intact infection.
The targeted virus was attenuated in miR-142-expressing cells and unaffected in cells lacking the microRNA. In interferon receptor deficient mice it showed reduced replication in sorted CD11b positive, CD11c positive and CD45 positive cells, and reduced titers in spleen and liver by three routes of inoculation. Sequencing of the residual virus recovered from spleen showed that every recoverable genome had excised the entire targeting cassette, which the authors read as evidence that the hematopoietic compartment is where the virus is predominantly amplified.
Scientific context
Dengue virus is a mosquito-borne flavivirus with four serotypes and a large global burden, and at the time of the study no vaccine or specific therapeutic was available. Infection is thought to begin in resident cutaneous Langerhans dendritic cells at the site of mosquito inoculation, followed by migration through the lymphatic system and recruitment of monocytes and macrophages, which are themselves targets. The paper states that although hematopoietic cells are understood to be major sites of replication, it remains uncertain whether non-hematopoietic cells are permissive during natural infection.
The paper frames the unresolved point as an interpretive one. A wide range of cell lines from several host species are permissive in culture, and the virus engages many candidate receptors including heparan sulfate, DC-SIGN, CD14 and heat shock proteins 70 and 90, which has led some to propose endothelial cells as targets in vivo. The paper notes that this interpretation has been questioned because viral antigen in endothelial cells could reflect entry rather than replication, and cites work finding undetectable viral RNA in endothelial cells.
Separately, inserting fully complementary microRNA target sites into a viral genome had already been used to restrict tropism for influenza A virus, poliovirus, vesicular stomatitis virus and measles virus. For dengue specifically, the paper notes that the approach had previously been applied only to a replication-incompetent replicon or to a chimera carrying tick-borne encephalitis virus structural proteins, so a fully replication-competent targeted dengue virus had not been available.
Central question
If dengue virus replication is selectively excluded from hematopoietic cells in an otherwise intact animal, can the virus still be amplified and disseminated from a non-hematopoietic compartment?
Experimental strategy
The strategy converts a tropism question into a genetic subtraction. Rather than detecting virus in candidate cell types and arguing about whether the signal means replication, the design removes one compartment from the infection and asks what remains.
The tool is the endogenous microRNA machinery. Perfectly complementary target sites for a cell-type-restricted microRNA make that microRNA function as a virus-specific silencing guide, so the virus is attenuated only where the microRNA is expressed. The choice of miR-142 is what confers the desired restriction, since it is among the most abundant hematopoietic-specific microRNAs and is absent from the fibroblast, hamster kidney and mosquito cells used as permissive controls.
Placement matters because the flavivirus 3-prime untranslated region carries structures required for translation and RNA synthesis. The insert was therefore placed in the variable region just downstream of the NS5 open reading frame, a position previously reported to tolerate nucleotide insertion. A control virus carrying the same 157 nucleotide cassette in reverse orientation controls for the insertion itself rather than for its sequence, and an unmodified wild-type clone controls for the presence of any insert.
Before using the tool, the study checks that dengue infection does not itself disable microRNA biogenesis or silencing, since the entire approach depends on that machinery functioning during infection.
In vivo, mice lacking receptors for type I and type III interferon were used because human dengue isolates lack an adequate immunocompetent animal model. Three inoculation routes address whether the result is an artifact of delivery. Magnetic sorting of splenocytes into CD11b positive, CD11c positive, CD45 positive and CD45 negative fractions separates the targeted compartment from the untargeted one within the same animal. Finally, sequencing the 3-prime untranslated region of virus recovered from infected animals asks whether residual virus represents a genuine non-hematopoietic reservoir or escape from targeting, a distinction that quantitative RT-PCR for NS5 cannot make.
Key findings
-
In human fibroblasts carrying a miR-142-targeted green fluorescent protein reporter, supplying miR-142 reduced fluorescence by about 75 percent, and dengue virus infection at a multiplicity of one did not alter that silencing despite confirmed high-level replication (Figure 1). Northern blots for an exogenous microRNA and an endogenous microRNA likewise showed no change during infection (Figure S1). The interpretation offered is that dengue does not interfere with microRNA biogenesis or silencing, which licenses the rest of the approach.
-
The targeted and reverse-orientation control viruses grew to comparable levels in mosquito C6/36 cells and in BHK cells, neither of which express miR-142, while unmodified wild-type virus exceeded both by roughly one log by quantitative RT-PCR and by plaque assay (Figure 2C, Figure 2D, Figure S3). The observation is a fitness cost attributable to the 157 nucleotide insertion itself and independent of its sequence, since the reverse-orientation control behaves the same way.
-
Electroporation of in vitro transcribed genomes into BHK cells expressing miR-142 abolished NS5 protein from the targeted construct alone and left wild-type and control constructs unaffected (Figure 3B). Infection of the same cells gave the same result (Figure 3C), with residual NS5 for the targeted strain attributed by the authors to cells not successfully transfected with the miR-142 plasmid.
-
A variant with the nucleotides complementary to positions 3 and 10 of miR-142 mismatched, intended to destroy seed pairing and the Argonaute 2 cleavage site, was nonetheless still repressed in a miR-142-dependent manner at the protein level (Figure 3D), while repression at the RNA level was significantly weaker than for the fully complementary virus with a reported p value of 0.0178 (Figure 3E). The observation is a dissociation between protein-level and RNA-level repression. The authors interpret it as a shift away from cleavage toward some form of translational repression or steric interference, and they state explicitly that the underlying molecular biology is unclear and that future studies will be required.
-
Infection of two hematopoietic lineages, a human B cell line and murine bone marrow derived macrophages, showed selective attenuation of the targeted virus, whereas HEK293 cells showed no difference between targeted and control virus (Figure 3F). This is the endogenous counterpart of the ectopic experiments and establishes cell-type-specific attenuation without supplied miR-142.
-
In interferon alpha receptor and interleukin 28 receptor double knockout mice inoculated intravenously, CD11b positive and CD11c positive splenocytes showed minimal NS5 transcript from the targeted virus and a significant decrease relative to control virus (Figure 4A).
-
Sorting splenocytes into CD45 positive and CD45 negative fractions showed reduced targeted virus in the hematopoietic fraction as expected, and also reduced growth in the non-hematopoietic fraction, while the ratio of non-hematopoietic to hematopoietic signal was higher for the targeted virus (Figure 4B, Figure S4A). The authors interpret this as showing that the targeted virus is not intrinsically attenuated in non-hematopoietic cells and that its lower titer there is a consequence of the loss of hematopoietic amplification.
-
Across intraperitoneal, intravenous and subcutaneous routes, spleen viral transcript was reduced for the targeted virus, with about a 1.5 log reduction after intraperitoneal delivery and greater attenuation by the intravenous and subcutaneous routes that the authors describe as better simulating natural inoculation (Figure 4C). Titers from spleen and liver showed approximately a three log reduction (Figure 4D, Figure S4B, Figure S4C). Heart, lung and brain titers were undetectable, identifying spleen and liver as the primary replication sites in this model.
-
Reduced targeted virus was also seen in liver, where the paper notes miR-142 expression is lower than in spleen. The authors read this as evidence that liver virus resides predominantly in resident macrophages and dendritic cells, and they raise as a possibility that these cells are needed to sustain basal replication in non-hematopoietic cells. The second part is an interpretation rather than a measurement.
-
Sequencing the 3-prime untranslated region of virus recovered from infected spleens found a complete absence of intact targeted genomes. Every recovered species had lost all four target sites, either by complete excision of the cassette or by replacement with a small host RNA fragment (Figure 5). In cell culture, by contrast, the mutation frequency was unbiased and comparable between cohorts, consistent with polymerase error rather than selection (Figure S5B). The authors note this explains the low residual signal in the NS5 quantitative RT-PCR, which cannot distinguish escape mutants from intact targeted virus, and they argue it makes a pre-existing quasispecies in the stock an unlikely explanation.
Mechanistic model
The study does not establish a mechanism for dengue tissue tropism at the molecular level, and it does not resolve the mechanism of the silencing it uses as a tool. What it establishes is a compartment requirement.
The model supported by the data is that hematopoietic cells, specifically macrophages and dendritic cells, are the compartment in which dengue virus is predominantly amplified in this animal model, and that replication in non-hematopoietic cells is insufficient to sustain dissemination on its own. The strongest evidence for the compartment claim is the escape mutant result. Under conditions where the hematopoietic compartment is closed to the virus, the entire recoverable population has reverted by excising the targeting cassette, which is difficult to reconcile with a productive alternative reservoir in which the targeted virus could have replicated unmodified. The authors state this conclusion as an inference from the completeness of the conversion.
For the silencing mechanism itself the paper is explicit that the picture is incomplete. Attenuation of the fully complementary virus is presumed by the authors to involve Argonaute 2 cleavage, and they describe the repression of the seed-mismatched variant as an enigma. They offer two alternative hypotheses, that residual extensive complementarity supports non-canonical silencing, or that loaded RISC binding provides steric hindrance that disrupts the long-range 5-prime to 3-prime base pairing required for flavivirus genome cyclization. They note that the second is supported by the nature of the escape mutants, since the only escapes recovered were complete cassette excisions rather than point mutations in seed or central positions. This remains an interpretation, not a demonstration.
Conceptual or technical advance
The work supplies a replication-competent dengue virus whose tropism can be edited genetically, which turns a question about which cells matter into an experiment that can be run in an intact animal. Where prior microRNA targeting of dengue had been limited to a replicon or to a chimeric virus, this construct replicates and disseminates, so compartment subtraction can be tested against dissemination rather than only against single-cycle gene expression.
The design also shows why sequencing the recovered population is necessary rather than optional. Quantitative RT-PCR for a viral gene registers escape mutants and intact targeted virus alike, and in this study that difference was the difference between apparent residual replication and no surviving targeted genomes at all. The authors present the approach as generalizable to other viruses and other cell subsets.
Relationship to the broader research program
The study sits in a line of work from this laboratory on engineering RNA viruses to interact with the host small RNA machinery, following directly from the use of microRNA target sites to attenuate influenza A virus in a species-specific manner and from work on engineered RNA viral synthesis of microRNAs, both cited here and both from the same group. The recurring idea is that the host small RNA system can be recruited as a programmable constraint on where a virus is permitted to replicate.
It also connects to a broader question that runs through the corpus, namely what small RNA pathways actually do during mammalian virus infection as opposed to what they can be made to do. This paper uses the machinery instrumentally and reports that dengue infection leaves it intact. Placing that observation beside the laboratory's other work on small RNA silencing during infection is category 3 synthesis and would need those records to support it.
Related publications
- Perez, Pham, Lorini, Chua, Steel and colleagues 2009, Nature Biotechnology, microRNA-mediated species-specific attenuation of influenza A virus. Relationship methodological foundation. Cited here as the precedent for microRNA-mediated viral attenuation and as the source of the extraction and Northern blot protocols used.
- Varble, Chua, Perez, Manicassamy, Garcia-Sastre and colleagues 2010, Proceedings of the National Academy of Sciences, engineered RNA viral synthesis of microRNAs. Relationship methodological foundation. Cited as the source of the miR-142 target sites used in the construct.
- Kinney and colleagues 1997, Virology, construction of infectious cDNA clones for dengue 2 virus. Relationship methodological foundation from another laboratory. The pD2/IC-30P-A clone was the backbone for all recombinant viruses here.
- Mordstein and colleagues 2008, Journal of Virology, on lambda interferon and epithelial resistance. Relationship methodological foundation from another laboratory. Source of the interferon receptor double knockout mice used for the in vivo work.
Limitations and boundaries
The animal model lacks receptors for both type I and type III interferon, which the authors chose because human dengue isolates have no adequate immunocompetent model. Interferon signaling shapes which cells support replication and how far a virus spreads, so conclusions about compartment requirements are bounded by that genetic background and should not be transferred directly to immunocompetent hosts or to human infection.
Only one serotype and one strain were used, dengue virus 2 strain 16681, so serotype-specific differences in tropism are not addressed.
The insertion itself costs fitness. Both the targeted and the reverse-orientation control virus grew about one log below unmodified virus in the permissive cell lines, so all comparisons are between two insertion-bearing viruses rather than against an unmodified baseline.
The in vivo analysis rests on spleen and liver, with heart, lung and brain below detection. Skin, the natural site of mosquito inoculation, and the Langerhans cells the introduction identifies as the first target were not sampled. The cell sorting used CD11b, CD11c and CD45 markers on splenocytes, which does not enumerate every hematopoietic subset and does not identify which non-hematopoietic cell types are represented in the CD45 negative fraction.
The claim that non-hematopoietic cells are not productively infected is stated by the authors as one of two possibilities, the other being that macrophages and dendritic cells are required for spread. The data constrain dissemination, not intrinsic permissiveness of every other cell type, and the paper phrases the closing conclusion in that conditional form.
The mechanism of attenuation for the seed-mismatched variant is unresolved and the authors say so. The steric interference model for disruption of genome cyclization is a hypothesis supported indirectly by the character of the escape mutants.
Escape mutant sequencing used fifteen randomly selected clones per condition after PCR amplification, so the estimate of population composition has the depth limits of clonal sequencing rather than deep sequencing.
Audience summaries
25 words
Dengue virus was engineered so immune cells silence it. Without replication in those cells the virus stopped spreading, and every surviving genome had deleted the silencing sites.
75 words
Dengue virus infects dendritic cells and macrophages, but whether other cells sustain infection has been unclear. Researchers inserted target sites for a microRNA found only in blood-lineage cells into the dengue genome, so the virus was silenced in those cells alone. In mice the modified virus lost replication in macrophages and dendritic cells and failed to spread to spleen and liver. All virus recovered from animals had deleted the inserted sites.
150 words
Determining which cells actually support dengue virus replication in a host is difficult because viral antigen can reflect entry or uptake rather than productive infection. This study takes a genetic approach instead, inserting four tandem target sites for the hematopoietic-restricted microRNA miR-142 into the variable region of the dengue virus 3-prime untranslated region. The resulting virus replicates normally in cells lacking that microRNA and is silenced in cells expressing it. In interferon receptor deficient mice the targeted virus showed reduced replication in sorted CD11b positive, CD11c positive and CD45 positive cells, and reduced spleen and liver titers by intraperitoneal, intravenous and subcutaneous routes. Sequencing the residual virus recovered from spleen found no intact targeted genomes, only variants that had excised the whole cassette. The authors read the completeness of that conversion as evidence that hematopoietic cells are the predominant site of amplification and are required for dissemination.
Discoveries supported by this paper
Encoding a perturbation in a virus makes tropism, host restriction and viral output experimental variables inside an intact animal
lab-led for Perez 2009, Varble 2010, Langlois 2012 in PNAS, Langlois 2012 in Molecular Therapy, Pham 2012, Varble 2013, Schmid 2014, Benitez 2015 on in vivo screening, Møller 2018 and the tenOever 2019 review, and co-led for Langlois 2013, whose ferret transmission work sits with the Perez and García-Sastre groups. Daniloski 2021 in Cell is co-led with the Sanjana laboratory
Documented publication relationships
- Degradation of Host MicroRNAs by Poxvirus Poly(A) Polymerase Reveals Terminal RNA Methylation as a Protective Antiviral Mechanism — conceptual extension.
- Hematopoietic-specific targeting of influenza A virus reveals replication requirements for induction of antiviral immune responses — companion.
- MicroRNA-mediated species-specific attenuation of influenza A virus — methodological foundation.
- Engineered RNA viral synthesis of microRNAs — methodological foundation.
- MicroRNA-based strategy to mitigate the risk of gain-of-function influenza studies — application.
- RNA viruses and the host microRNA machinery — application.
- The Mammalian Response to Virus Infection Is Independent of Small RNA Silencing — methodological foundation.
- The Mammalian Response to Virus Infection Is Independent of Small RNA Silencing — predecessor.
- Engineered Mammalian RNAi Can Elicit Antiviral Protection that Negates the Requirement for the Interferon Response — predecessor.
- Homologous recombination is an intrinsic defense against antiviral RNA interference — application.