Prokaryotic viperins produce diverse antiviral molecules

Abstract

Viperin is an interferon-induced cellular protein that is conserved in animals¹. It has previously been shown to inhibit the replication of multiple viruses by producing the ribonucleotide 3'-deoxy-3',4'-didehydro (ddh)-cytidine triphosphate (ddhCTP), which acts as a chain terminator for viral RNA polymerase². Here we show that eukaryotic viperin originated from a clade of bacterial and archaeal proteins that protect against phage infection. Prokaryotic viperins produce a set of modified ribonucleotides that include ddhCTP, ddh-guanosine triphosphate (ddhGTP) and ddh-uridine triphosphate (ddhUTP). We further show that prokaryotic viperins protect against T7 phage infection by inhibiting viral polymerase-dependent transcription, suggesting that it has an antiviral mechanism of action similar to that of animal viperin. Our results reveal a class of potential natural antiviral compounds produced by bacterial immune systems.

Extended Data Figure 1. pVips protect against phage infection.

Bacteria expressing pVips, GFP or MoaA (negative controls), or the human viperin gene were grown on agar plates and tenfold serial dilutions of the phage lysate were dropped on the plates. a - h. Efficiency of plating (EOP) data, representing plaque-forming units per millilitre; each bar graph represents average of three replicates, with individual data points overlaid.

Extended Data Figure 2. T7 infection in liquid culture in the presence of pVips.

a. For each pVip, growth curves of liquid cultures infected by phage T7 (MOI 0.001) are shown. Light and dark grey are uninfected and infected controls (strain expressing GFP), respectively. Light and dark red are uninfected and infected strains expressing pVips, respectively. Two technical replicates are presented as individual curves; representative of three biological replicates. The negative controls (GFP uninfected, GFP infected) are the same for pVips 6, 7, 8, 10, 15, 27, 37, 39, 42, 50, 54, MoaA, and for pVip12, 19, 32, 44, 46, 47, 48, 57, 58, 60, 61, 62, 63. b. The catalytic activity of pVips is required for defense against T7 phage. For each pVip and its respective mutant (mutation of three cysteines in the active site), growth curves of liquid cultures infected by phage T7 (MOI 0.001) are presented. Light and dark grey are uninfected and infected controls (strain expressing MoaA), respectively. Light and dark red are uninfected and infected strains expressing viperins, respectively. Light and dark blue are uninfected and infected strains expressing catalytically inactive mutants. Two technical replicates are presented as individual curves; representative of three biological replicates.

Extended Data Figure 3. Detection of ddhCTP and ddhCTP derivatives in cell lysates from an E. coli strain expressing the human viperin.

a.Extracted ion chromatogram of the ddhC standard. b-d. Extracted ion chromatogram for singly charged masses that are predicted to correspond to ddhC (m/z 226.08223, retention time (RT) of 2.2 minutes)(b), ddhCMP (m/z 306.04856, RT 9.7)(c), ddhCTP (m/z 465.98122, RT 10.7)(d) in cell lysates from an E. coli strain expressing the human viperin. Representative of three replicates.

Extended Data Figure 4. Detection of ddh-ribonucleotides in lysates of cells that express pVips.

a. Quantification of ddh-cytidine (ddhC) in lysates of cells expressing pVips. Detection and quantification of ddhCwas performed using LC-MS with a synthesized chemical standard (Methods). For MoaA, the measurement was under the limit of detection (LOD 0.0003 uM). Bar graph represents average of three replicates, with individual data points overlaid. b-h.Relative abundance for singly charged masses that are predicted to correspond to ddhC (m/z 226.08223, retention time (RT) of 2.2 minutes)(b), ddhCMP (m/z 306.04856, RT 9.7)(c), ddhCTP (m/z 465.98122, RT 10.7)(d), ddhUMP (m/z 307.03258, RT 8.7)(e), ddhUTP (m/z 466.96524, RT 9.9)(f), ddhGMP (m/z 346.05471, RT 9.8)(g), and ddhGTP (m/z 505.98737, RT 10.7)(h). Average relative abundance is presented as bar graph, with individual data points from three biological replicates overlaid. Limit of detection (LOD) is indicated by a dashed grey line. A compound was defined as present, in Figure 3, if all three replicated were above the LOD.

Extended Data Figure 5. MS/MS fragmentation spectra for predicted compounds.

MS/MS data were acquired in positive ionization mode for a synthesized chemical standard ddhC (a) as well as for masses from the human viperin cell lysate predicted to correspond to ddhC (b), and ddhCMP (c). Similar data were obtained for masses from the pVip21 cell lysate predicted to correspond to ddhGMP (d), and ddhGTP (e). MS/MS data were acquired, in negative ionization mode, from the pVip47 cell lysate for masses predicted to correspond to ddhUMP (f), and ddhUTP (g). In all panels, assignment of hypothetical structures is indicated for informative fragment ions. The ddhC molecule is annotated to level 1, and all other molecules are annotated to level 2b, per the Metabolomics Standards Initiative nomenclature.

Extended Data Figure 6. MS/MS fragmentation spectra for predicted compounds from in vitro reactions with purified pVips.

(a-b) MS/MS data were acquired in positive ionization mode for the product detected in reaction samples using purified pVip6 or purified pVip56 and CTP and GTP as nucleotide substrates respectively; the resulting products are predicted to correspond to ddhCTP (a) and ddhGTP (b). (c) MS/MS data were acquired in negative ionization mode for product detected in reaction samples using purified pVip8 UTP as substrate; the resulting product is predicted to correspond to ddhUTP (c).

Extended Data Figure 7. Transcription during induction of WT and mutant pVips.

a. The catalytic activity of pVips is required for defense against T7 phage and repression of viral transcription. Application of the reporter assay (same as presented in Figure 4a) for strains expressing the human viperin, pVips and their cognate catalytically inactive mutants. Strains are first induced with arabinose for 45 minutes to express the pVip. At t=0, IPTG is added to express the GFP. Fluorescence/OD over time curves are presented for each strain. Dark and light red correspond to induced and non-induced wild type viperins, respectively; Dark and light blue correspond to induced and non-induced mutant viperins, respectively. Grey curve corresponds to negative control (WT viperin, no addition of IPTG). Two technical replicates are presented by individual curves. Representative of two biological replicates. b.T7 RNAP expression as measured by RNA-seq. The expression (RPKM) of T7 RNAP in cells expressing viperins was compared to that in cells expressing the MoaA negative control. Bar graphs represent average of two replicates, with individual data points overlaid.

Extended Data Figure 8. Heterologous expression of pVips is not toxic in E. coli.

Expression of pVips, human viperin or negative controls (GFP, MoaA) was induced at 45min by addition of arabinose (final concentration 0.2%). CFU were measured right after dilution from overnight culture (t=0), before induction (t=45), and 45 and 90 minutes after induction (t=90, t=135).

Extended Data Figure 9. Putative multi-gene defense systems that include pVips.

Representative instances of pVips and their genomic neighborhood. Genes predicted to be part of the pVip-containing defense system are highlighted. Genes known to be involved in defense are in yellow. Genes of mobile genetic elements are in dark grey. RM, restriction-modification; TA, toxin-antitoxin. The name of bacterial species, and the accession of the relevant genomic scaffold in the IMG database are indicated on the left. Panels a-d represent four common configurations of putative pVip-containing systems found in bacterial and archaeal genomes.

Extended Data Figure 10. Phylogenetic tree of pVips and putative eukaryotic viperins.

MoaA sequences were used as an outgroup (grey). pVips are depicted in red and putative eukaryotic viperins selected for the phylogenetic tree presented in Figure 2 are depicted in blue.

Figure 1. pVips and the human viperin have antiviral activity in bacteria.

a. Representative instances of pVips and their genomic neighborhood. Homologs of the human viperin are in red, genes annotated as nucleotide kinase in brown, genes known to be involved in defense in yellow, and genes of mobile genetic elements in dark grey. RM, restriction-modification; TA, toxin-antitoxin; Gabija is a recently described defense system. The name of the bacterial species, and the accession of the relevant genomic scaffold in the IMG database are indicated on the left. b. Plaque sizes of phage T7 infecting E. coli strains that express viperins. Bacteria expressing pVips, negative controls (GFP, MoaA), or the human viperin gene were grown on agar plates and phage lysate was dropped on top of them. Bar graph represents average of three replicates, with individual data points overlaid. Star represents statistically significant difference compared to negative control (GFP) (two tailed t-test, p-value<0.01). c. Growth curves of E. coli strains expressing viperins that were infected by phage T7. Light and dark grey are uninfected and infected controls (strain expressing GFP), respectively. Blue and red are uninfected and infected strains expressing viperins, respectively. The negative control (GFP uninfected, GFP infected) is the same in all four graphs. Curve corresponds to the mean of three biological replicates, each with an average of two technical replicates, and the shade corresponds to a confidence interval (CI) of 95%.

Figure 2. Phylogenetic tree of the viperin family.

Branches are colored according to major clades. Bootstrap values (derived from the ultrafast bootstrap function in the IQtree software) are indicated for major nodes. The presence of a nucleotide kinase in the genomic vicinity of the pVip is shown by a brown rectangle in the surrounding ring (or a dark grey rectangle, in case the kinase is fused to the pVip gene). Triangles correspond to the type of ddh-nucleotide derivatives produced by a specific pVip, as measured by mass spectrometry analysis. The phylogenetic tree was generated using a set of 205 non-redundant pVip sequences.

Figure 3. pVips produce a variety of modified ribonucleotides.

a. Extracted ion chromatograms for selected pVip lysates analyzed via LC-MS. Presented are chromatograms of singly charged masses with a precision +/- 5 ppm corresponding to ddhC (m/z 226.08223, retention time (RT) of 2.2 minutes), ddhCMP (m/z 306.04856, RT 9.7), ddhCTP (m/z 465.98122, RT 10.7), ddhUMP (m/z 307.03258, RT 8.7), ddhUTP (m/z 466.96524, RT 9.9), ddhGMP (m/z 346.05471, RT 9.8), and ddhGTP (m/z 505.98737, RT 10.7). X-axis depicts RT in minutes, y-axis depicts normalized ion intensity (A.U, arbitrary units). Normalization was performed on all pVips and MoaA (negative control) samples, with maximal values set to 1.0. In black, peak assigned to ddh nucleotides. In grey, peaks that appear in the negative controls and are not assigned to ddh nucleotides. Representative of three replicates. b. Production of ddh-nucleotide derivatives by pVips. Colored boxes depict detected compounds. Colored rectangles on the left and associated numbers represent the clade of pVips as described in Figure 2. c. Chromatograms of ddh-nucleotides detected in reaction samples performed in vitro with purified pVips. The presence of a product corresponding to ddhCTP, ddhUTP, and ddhGTP is observed in samples where a pVip was incubated with SAM, dithionite, and the respective nucleotide substrate.

Figure 4. pVips inhibit T7 polymerase-dependent transcription.

a. Schematic representation of the reporter system for T7 polymerase-dependent transcription. E. coli BL21-DE3 encodes a chromosomal T7 RNA polymerase (T7 RNAP) under the control of an IPTG-inducible promoter. A reporter plasmid encodes GFP under the control of a T7 promoter. Upon IPTG induction, the T7 RNA polymerase is expressed and drives the expression of GFP. The pVip (or MoaA control) is encoded on a second plasmid under the control of an arabinose promoter. b-f. Application of the reporter assay for strains expressing MoaA (negative control), the human viperin, and pVips. Strains are first induced with arabinose for 45 minutes to express the pVip. At t=0, IPTG is added to express the GFP. Fluorescence/OD over time curves are presented for each strain. Grey lines correspond to no induction (no arabinose, no IPTG), green to IPTG only (GFP expressed, viperin not expressed), and red to induction with both IPTG and arabinose. Curve corresponds to the mean of two technical replicates and the shade to a confidence interval (CI) of 95%. Representative of two biological replicates. b. Strain expressing MoaA (negative control). c. Strain expressing the human viperin. c-f. Strains expressing prokaryotic viperins. g. GFP expression as measured by RNA-seq. GFP expression (RPKM) in cells expressing viperins was compared to that in cells expressing the MoaA negative control. Bar graph represents average of two replicates, with individual data points overlaid.