Poliovirus tropism and attenuation are determined after internal ribosome entry (original) (raw)
Expression of bicistronic mRNAs from recombinant human adenovirus. A system was developed to measure IRES activity in murine organs. Recombinant human adenovirus vectors were produced that express bicistronic mRNAs that encode two reporter proteins separated by an IRES (Figure 1A). These bicistronic mRNAs possess a 5′ cap structure, and translation of the first open reading frame occurs by 5′ end–dependent initiation. Efficient translation of the second open reading frame requires an IRES to mediate internal binding of ribosomes.
IRES-mediated translation in A549 cells infected with recombinant adenoviruses. (A) Schematic of bicistronic reporter DNA. The arrow indicates the transcription-initiation site of the murine cytomegalovirus immediate early promoter. Firefly luciferase and Renilla luciferase have independent translation initiation and termination codons. SV40 An, simian virus 40 polyadenylation signal. (B) IRES activity in cells infected with adenovirus encoding bicistronic mRNA. The x axis indicates IRES encoded in the recombinant adenovirus. PV 5′NC-X472, poliovirus IRES with _Xho_I linker substitution at nucleotide 472; HCV Ø28_69, HCV IRES with deletion of nucleotides 28_69. (C) Northern blot hybridization of A549 cells. Cells were infected with recombinant adenoviruses indicated at the top. RNA size markers are given at left. Lane 1, monocistronic firefly luciferase; lane 2, monocistronic Renilla luciferase; lane 3, bicistronic mRNA, no IRES; lane 4, bicistronic mRNA with P3/119/70 IRES; lane 5, bicistronic mRNA with P3/Sabin IRES; lane 6, bicistronic mRNA with HCV IRES; lane 7, bicistronic mRNA with HCV IRES lacking nucleotides 28_69. First two lanes are from the same gel as lanes 3_7 but were exposed for a longer period of time. (D) Luciferase expression in adenovirus-infected cells. Upper panel: Firefly luciferase (white bars) and Renilla luciferase expression (black bars). Luciferase was normalized to the P3/119/70 IRES value for each reporter protein, and also to the quantity of bicistronic mRNA determined by Northern blot hybridization. Lower panels: Representative Northern blot hybridizations of RNA from the same infected cells. IRES encoded by bicistronic mRNA is indicated at the bottom. Data points in B and D, the results of separate experiments, are the mean of three infections, and error bars indicate SD.
To demonstrate that an IRES-dependent mRNA can be translated when expressed from an adenovirus vector, A549 cells were infected with adenoviruses that encode bicistronic mRNAs with the IRES of poliovirus, HCV, or CVB3. Inclusion of the IRES within a bicistronic mRNA results in significantly higher levels of Renilla luciferase activity compared with a short, unstructured sequence (Figure 1B). These results agree with those of previous studies that showed that the inclusion of the IRES of poliovirus, HCV, or CVB3 in an mRNA results in increased expression of the downstream reporter gene (8, 31, 32). The effects of mutations previously shown to abrogate IRES-mediated translation were determined. Substitution of an _Xho_I linker for nucleotides 472–479 of the poliovirus IRES, or deletion of nucleotides 28–69 from the HCV IRES, has been shown to impair translation (33, 34), and these mutations reduce Renilla luciferase activity in adenovirus vector–infected cells (Figure 1B). These findings provide genetic evidence that Renilla luciferase activity observed in adenovirus vector–infected cells is a consequence of internal initiation.
To confirm that Renilla luciferase is present only in full-length mRNAs, Northern blot hybridization of RNA from adenovirus vector–infected cells was performed (Figure 1C). No species smaller than the full-length bicistronic mRNA was detected in cells infected with these adenoviruses. These findings demonstrate that Renilla luciferase is translated by internal ribosome entry on full-length bicistronic mRNAs, and not by 5′ end–dependent translation of smaller RNAs produced by RNA degradation, splicing, or aberrant transcription. An additional RNA approximately 500 nucleotides larger than the bicistronic mRNA was also detected (Figure 1C). This RNA likely results from transcription termination at the adenovirus E1b polyadenylation signal 500 nucleotides downstream of the simian virus 40 polyadenylation signal (35).
Because IRES activity is based on standardization of Renilla to firefly activity, any effect of the IRES on translation of the upstream reporter protein would prevent a reliable quantitation of translation initiated by the IRES. To address this possibility, firefly and Renilla luciferase activities were measured in infected cells and normalized to the amount of bicistronic RNA as determined by Northern blot hybridization. Normalization compensates for differences in luciferase activity that result from variation in infection efficiency or transcription. The findings demonstrate that the IRES has no effect on firefly luciferase translation (Figure 1D).
IRES-mediated initiation of translation in murine organs. In animals, IRES-mediated internal initiation of translation could be restricted to specific organs. To address this hypothesis, mice were infected with recombinant adenovirus vectors that encode bicistronic mRNAs. The IRES of poliovirus, HCV, and CVB3 is functional in all organs examined, including brain, spinal cord, skeletal muscle, heart, lung, liver, kidney, and ileum (Figure 2, A–C).
IRES-mediated translation in murine organs infected with recombinant adenoviruses. (A) Poliovirus type 3 119/70 IRES (black bars) and poliovirus type 1 5′NC-X472 IRES (white bars). (B) HCV 1b IRES (black bars) and HCV 1a IRES with nucleotides 28_69 deleted (white bars). (C) CVB3 IRES. Data points are the mean of at least five mice, and error bars indicate SD.
To provide genetic evidence that Renilla luciferase translation in mice is mediated by the viral IRES, the effect of mutations in the IRES was assessed. As was observed in cultured cells (Figure 1B), substitution of an _Xho_I linker for nucleotides 472–479 of the poliovirus IRES, or deletion of nucleotides 28–69 from the HCV IRES, nearly eliminates IRES activity in mouse organs (Figure 2, A and B). No firefly or Renilla luciferase activity was detected when mice were infected with an adenovirus lacking the promoter necessary for production of bicistronic mRNA with the poliovirus IRES (data not shown). This result shows that translatable Renilla luciferase mRNAs are not produced from a promoter within the poliovirus IRES.
IRES activity should be independent of the quantity of infecting virus. When mice were inoculated with amounts of virus ranging from 106 to 1010 PFUs, levels of firefly luciferase and Renilla luciferase increased nearly 10,000 times (Figure 3A), but activity of the poliovirus IRES remained constant (Figure 3B). Therefore IRES activity in an organ is not affected by the quantity of infecting vector virus.
Relationship between dose of recombinant adenovirus and IRES activity in mouse brain. (A) Firefly luciferase (squares) and Renilla luciferase (triangles) expression in mouse brain. (B) Activity of the poliovirus type 3 119/70 IRES in the mouse brain, normalized to activity at a dose of 106 PFUs. The x axis indicates adenovirus dose in log10 PFUs. Data points are the mean of five infections, and error bars indicate SD.
Does the IRES determine sites of viral replication and disease? Studied in isolation from the viral genome, the IRES of poliovirus, HCV, or CVB3 mediates translation in many organs. In the genome, organ specificity of an IRES could be influenced by viral RNA sequences or proteins produced during viral replication. To address this possibility, the poliovirus IRES was substituted with the cognate sequence from viruses that infect different organs (Figure 4A). The IRES of human poliovirus type 1 was replaced with that of HCV, which is hepatotropic (36), or with the IRES of CVB3, which causes myocarditis (37). These recombinant viruses were named P1/HCV and P1/CVB3. Single-step growth analysis of both recombinant viruses in HeLa cells reveals a defect in replication, comprising an early delay in virus production (Figure 4B). While the final yield of P1/CVB3 approached that of poliovirus type 1, the yield of P1/HCV was significantly lower. Previous results have demonstrated that the poliovirus IRES contains determinants of viral RNA replication (38, 39). It is possible that exchange of the poliovirus IRES has removed _cis_-acting sequences important for RNA replication.
Replication and virulence of recombinant poliovirus strains. (A) Genome structure of poliovirus type 1 strain Mahoney, recombinant strain P1/CVB3, and recombinant strain P1/HCV. IRES, predicted AUG initiation codons, and poliovirus polyprotein (open box) are indicated. Translation of P1/HCV is predicted to initiate at the HCV AUG initiation codon, which is followed by 369 nucleotides of HCV polyprotein sequence (lined box). Sequence encoding the recognition site for poliovirus protease 2Apro (triangle) separates HCV sequence and nucleotide 745 of the poliovirus genome. (B) Single-step replication analysis in HeLa cells of poliovirus type 1 strain Mahoney (squares), P1/CVB3 (triangles), and P1/HCV (circles). Data points are the mean of two infections.
Virulence and tropism of recombinant viruses were determined in mice transgenic for the human poliovirus receptor gene (40). After intraperitoneal inoculation of adult transgenic poliovirus receptor (TgPVR) mice with poliovirus type 1, virus replicates to high titers in the brain and spinal cord (Figure 5A). In contrast, virus titers in the heart steadily decline after infection, and virus titers in the pancreas increase by day 3 and subsequently decline (Figure 5A). In newborn TgPVR mice, there is a large increase in virus titer in the brain and spinal cord, and a moderate increase in the liver (Figure 5C). Infection of adult and newborn TgPVR mice leads to flaccid limb paralysis and death (40). Viral replication and pathogenesis are not observed in nontransgenic mice (40).
Poliovirus infection of mice. (A_D) Poliovirus replication in murine organs. The y axis indicates virus titer at the indicated days after infection. (A) Virus titers in pancreas (filled inverted triangles), spinal cord (filled triangles), brain (filled diamonds), and heart (filled squares) of adult TgPVR mice infected with poliovirus. (B) Virus titers in spinal cord (filled and open triangles), brain (filled and open diamonds), pancreas (filled inverted and open inverted triangles), and heart (filled and open squares) of adult TgPVR (solid lines) and nontransgenic (dashed lines) mice infected with P1/CVB3. (C) Virus titers in spinal cord (filled triangles), brain (filled diamonds), and liver (filled squares) of newborn TgPVR mice infected with poliovirus. (D) Virus titers in spinal cord (filled and open triangles), brain (filled and open diamonds), and liver (filled and open squares) of newborn TgPVR (solid lines) or nontransgenic (dashed lines) mice infected with P1/HCV. Data points are the geometric mean titer in organs from at least three mice. (E) Virulence of recombinant poliovirus strains. The y axis indicates the percentage of surviving mice at different times after infection. TgPVR mice were infected with 107 PFUs P1/CVB3 (squares), 2 ∞ 106 PFUs P1/HCV (triangles), or 109 PFUs P1/HCV (circles); nontransgenic mice were infected with 2 ∞ 106 PFUs P1/HCV (diamonds).
To determine the effect of IRES replacement on poliovirus pathogenesis, TgPVR mice were infected with recombinant viruses. P1/CVB3 replicates and causes disease in adult TgPVR mice (Figure 5, B and E). Virus titers in the brain and spinal cord increase 10,000-fold during the course of infection (Figure 5B). Some infected mice develop flaccid hind-limb paralysis or die (Figure 5E). CVB3 replication and pathogenesis occur in heart and pancreas of adult mice (41, 42), yet the pattern of P1/CVB3 replication in these organs is unchanged from that of poliovirus type 1 (Figure 5, A and B). When adult TgPVR mice were inoculated with P1/HCV, viral replication and disease did not occur (data not shown). However, P1/HCV replicates and causes disease in newborn TgPVR mice (Figure 5, D and E). Virus titers in the brain and spinal cord increase 500- and 1,000-fold, respectively, during the course of infection (Figure 5D). Some infected mice develop flaccid hind-limb paralysis or die (Figure 5E). HCV replicates and causes disease in primate liver, yet in TgPVR mice P1/HCV is cleared from this organ (Figure 5D). Virulence of virus recovered from paralyzed newborn TgPVR mice was unchanged in newborn and adult mice (data not shown). Replication of poliovirus type 1 in the brain and spinal cord of mice and development of paralytic disease are dependent on the human poliovirus receptor (40). P1/HCV and P1/CVB3 are cleared from the brain and spinal cord of nontransgenic mice (Figure 5, B and D), and paralysis does not occur (Figure 5E and data not shown). The human poliovirus receptor is required for paralytic disease and increases in viral titers after inoculation, further evidence that P1/HCV and P1/CVB3 replicate in the murine brain and spinal cord.
Do mutations in the IRES of poliovirus vaccine strains have an organ-specific effect on translation? It has been suggested that neuroattenuation caused by a C472U mutation in the IRES of poliovirus vaccine strains is a consequence of reduced translation of poliovirus RNA in the brain and spinal cord. The effect of the C472U mutation on translation in murine organs was therefore determined. Recombinant adenoviruses were produced that encode bicistronic mRNAs with the poliovirus type 3 IRES and either a C or a U at nucleotide 472. The C472U mutation decreases IRES-dependent translation in continuous cell lines of both neuronal and non-neuronal origin (Figure 6A). No species smaller than the full-length bicistronic mRNA was detected in cells infected with either adenovirus (Figure 6A). The C472U mutation also decreases IRES-dependent translation in murine brain, spinal cord, heart, lung, liver, kidney, ileum, and muscle (Figure 6B).
Effect of C472U mutation on translation of bicistronic mRNAs expressed by recombinant adenovirus. (A) Upper panel: Activity of P3/Sabin IRES (white bars) relative to P3/119/70 IRES (black bars) in SY5Y, HeLa, and A549 cells. Lower panel: Representative Northern blot hybridization of RNA from the same infected cells. (B) Activity of P3/Sabin IRES (white bars) relative to P3/119/70 IRES (black bars) in murine organs. Data points are the mean of three infections, and error bars indicate SD.
It has previously been shown that poliovirus strains with the C472U mutation are cleared from the brain and spinal cord of adult mice and fail to cause paralysis (26). To determine whether virus strains with the C472U mutation have lost the ability to replicate within the murine brain and spinal cord, newborn TgPVR mice were infected with polioviruses with either C or U at nucleotide 472. The neurovirulence of these viruses was then determined as a measure of their ability to replicate in the murine brain and spinal cord (Table 1). Poliovirus strain PRV7.3, with U at nucleotide 472, is nearly as neurovirulent in newborn mice as virus strain PRV8.4, which is identical except for a C at nucleotide 472. In contrast, the neurovirulence of virus strain PRV7.3 is attenuated in adult TgPVR mice. The poliovirus type 3 vaccine strain P3/Sabin, which is neuroattenuated in adult TgPVR mice, is virulent in newborn TgPVR mice. As expected, virus strains P3/119/70 and PRV8.4, both with C at nucleotide 472, are neurovirulent in adult TgPVR mice, and virus strain P3/119/70 is highly neurovirulent in newborn mice. These findings demonstrate that the neuroattenuating mutation at nucleotide 472 of the poliovirus genome does not eliminate viral replication in the murine brain and spinal cord.
Virulence of poliovirus strains in TgPVR mice
Viruses with a reversion of the C472U mutation may accumulate during replication of PRV7.3 and P3/Sabin in cell culture or in animals (23, 43). The fraction of viral revertants can be readily determined by restriction enzyme cleavage of a DNA copy of the IRES (44). To confirm that stocks of PRV7.3 and P3/Sabin are free of such revertants, the proportion of U at nucleotide 472 was determined by cleavage of a DNA copy of the IRES with _Mbo_I (Figure 7). According to a standard curve produced with plasmid DNA (data not shown), more than 99% of the viral RNA in stocks of PRV7.3 and P3/Sabin has a U at nucleotide 472. Similarly, nearly 99% of the PRV7.3 or P3/Sabin viral RNA recovered from the brains of paralyzed mice has a U at nucleotide 472. When plasmid DNA that encodes a C at this position was assayed, cleavage of 100% of _Mbo_I sites was never observed, most likely because the DNA is damaged during PCR amplification (Figure 7, lane 2). However, the proportion of base C detected at nucleotide 472 in viral RNA from stocks of PRV8.4 and P3/119/70 and in RNA recovered from the brains of mice infected with these viruses was similar to that in plasmid DNA that encodes a C at this position. These findings show that paralysis in newborn mice inoculated with neuroattenuated viruses is not caused by a reversion of the C472U mutation.
Base at nucleotide 472 in poliovirus recovered from paralyzed mice. Upper panel: Representative Southern blot hybridization of _Mbo_I-cleaved PCR products. The 93-bp DNA (472U) and 61-bp DNA (472C) are indicated. Lower panel: Proportion of 472U in each sample, determined by quantitative analysis of Southern blots. For mouse brain, data points are the mean of at least three mice except for P3/Sabin, for which one mouse was analyzed. Error bars indicate SD. The x axis indicates template for RT-PCR. Lane 1, plasmid DNA encoding P3/Sabin IRES; lane 2, plasmid DNA encoding P3/119/70 IRES; lanes 3_6, virus stocks used for infection of TgPVR mice; lanes 7_10, RNA from brain of TgPVR mice infected with the indicated virus stock.
In cell culture, the C472U mutation confers a temperature-sensitive phenotype that can be suppressed by mutations in the 2Apro coding region (45). Furthermore, mutation to A at IRES nucleotide 537 might restore neurovirulence by improving base pairing between nucleotides 472 and 537 (46). To ensure that the neurovirulence of PRV7.3 in newborn mice is not a consequence of suppressor mutations, the nucleotide sequence was determined in these regions of viral RNA from the brains of four paralyzed mice. No sequence changes were found in the 2Apro coding region. In viral RNA from two mice, the parental base, G, was found at nucleotide 537. In viral RNA from two other mice, a mixed population of the parental nucleotide 537G and the mutation 537A was found. These results indicate that mutations within the 2Apro coding region or at nucleotide 537 are not essential for neurovirulence.