Differential Activation of Influenza A Virus Endonuclease Activity Is Dependent on Multiple Sequence Differences between the Virion RNA and cRNA Promoters

Abstract

Influenza virus endonuclease activity was studied in vitro with model virion RNA (vRNA) and cRNA molecules. We show that endonuclease activity can be partially rescued by transplanting vRNA-like promoter features into the model cRNA promoter. This study defines three distinctive features within the vRNA promoter—absent in the cRNA promoter—that are required for endonuclease cleavage.

Influenza virus is a segmented negative-sense RNA virus. Eight virion RNA (vRNA) segments are packaged within the virion together with the nucleoprotein and an RNA-dependent RNA polymerase (4, 10, 15). During viral replication, the negative-sense vRNA acts as a template for the synthesis of two RNA products. cRNA is a full-length copy of vRNA and is used as a template for the synthesis of new vRNA, whereas mRNA is capped at its 5" end and polyadenylated (10). The mRNA cap is generated by endonuclease cleavage (cap snatching) of cap 1 structures from host cell pre-mRNA, a process carried out by the influenza virus RNA polymerase complex (3, 9, 14).

The influenza virus polymerase complex (a heterotrimeric complex of three subunits, PB1, PB2, and PA) cleaves capped substrates very poorly unless synthetic vRNA molecules are added (1, 2, 8, 12, 13). In addition, cRNA does not activate endonuclease activity (3). This phenomenon is termed differential activation.

The binding of a capped mRNA by the polymerase is known to be dependent on the prior binding of the vRNA 5" promoter arm (6, 17). Subsequent binding of the 3" vRNA promoter arm to PB1 leads to cleavage of the mRNA (3, 13). Recent data suggest that endonuclease activity is catalyzed by PB1 (14) and that the stem-loop structure near the 3" end of the vRNA is needed for cleavage (11). As vRNA and cRNA promoters may bind to different sites within the PB1 polymerase subunit (7), this may control the differential levels of activation of mRNA cleavage. However, the specific features of vRNA compared to those of cRNA that lead to differential levels of endonuclease activation are unknown.

The cRNA promoter is complementary to the vRNA promoter and is superficially very similar in secondary structure, since both cRNA and vRNA can be drawn in a corkscrew configuration (Fig. 1). Specific features of the corkscrew secondary structure model (5, 11) are the two short hairpin loops, each with a stem of 2 bp and a tetraloop. However, there are obvious sequence differences between the vRNA and cRNA promoters. Two of the sequence differences occur in the stems of the short hairpin loops and are unlikely to alter the secondary structure. Specifically, a U-A base pair at position 3", 8" in the 5" hairpin loop of vRNA becomes a C-G base pair at the corresponding position of cRNA. In the 3" hairpin loop, a G-C base pair at position 3, 8 of vRNA is replaced by an A-U base pair in cRNA. Previously, we have shown that the endonuclease activity of vRNA is strictly dependent on the presence of these stem base pairs in both the 3" and 5" hairpin loops, although the different base pairs can affect endonuclease activity quantitatively (11, 12). For example, replacing the G-C base pair at position 3, 8 in the stem of the 3" hairpin loop of vRNA with an A-U base pair and thereby generating the exact sequence present in the stem of the 3" hairpin loop of cRNA reduced endonuclease activity by about 50% (11).

FIG. 1.

Promoter constructs used. (A) cRNA promoter (49 nt) showing the mutated residues. 10′ I indicates the location of an A residue inserted to give the 10" I construct. 10 D indicates the U residue deleted during construction of the 10 D mutant cRNA. Residue 5 (A) of the 5" arm and residue 3 (C) of the 3" arm differ from those of the vRNA construct. The boxed residues indicate differences between cRNA and vRNA in the base-paired stems of the hairpin loops. (B) vRNA promoter (49 nt). The dashes indicate the remaining 18 residues (see Fig. 1A and B of reference 12).

The four remaining sequence differences between the vRNA and cRNA promoters occur in bases that are unpaired in the corkscrew model. The first difference is that the 5" arm of the vRNA promoter (Fig. 1B) is 1 nucleotide (nt) longer than the 5" arm of the cRNA promoter because of a “hinge” A residue located between the base-paired panhandle nucleotides and the nucleotides forming the stem of the 5" hairpin loop. The second difference is that the 3" arm of the vRNA promoter (Fig. 1B) is 1 nt shorter than the 3" arm of the cRNA promoter, because the vRNA promoter lacks the hinge U residue present in the 3" arm of the cRNA promoter. These sequence differences are predicted to alter the spatial arrangements between the base-paired regions of the vRNA and cRNA promoters, which could influence the binding of the promoters to the influenza virus RNA polymerase, thereby controlling endonuclease cleavage. A third difference is the identity of residue 5 of the 3" hairpin loop (i.e., the fifth residue of the 3" arm of the promoter) of the vRNA promoter, which is a U in vRNA but a C in cRNA. A fourth difference between the vRNA and cRNA promoters is the identity of residue 5" (i.e., the fifth residue of the hairpin loop of the 5" arm), which is a G in vRNA but an A in cRNA.

Our approach to understanding the sequence and features required for the activation of endonuclease cleavage was to transplant sequence elements of the vRNA promoter that differed from those of cRNA into a modified cRNA promoter (11, 12). However, we excluded from our study the two sequence differences between the vRNA and cRNA promoters in the stems of the base-paired loops (see above), since previous work had shown that these two differences were required simply for the secondary structure of the hairpin loop. They influenced endonuclease cleavage quantitatively rather than qualitatively (11). Thus, these differences could not be the critical feature controlling differential levels of activation. Therefore, we limited our studies to the four unpaired mutations that differed between the cRNA and vRNA promoters. We constructed 13 modified mutant cRNAs with one, two, three, or all four of the aforementioned sequence characteristics of vRNA by site-specific PCR-based mutagenesis, making a model 49-nt cRNA plasmid construct, as described previously (11) (Fig. 1B). By attempting to rescue endonuclease cleavage in these modified cRNA promoters, we hoped to identify which of the four sequences characteristic of vRNA were crucial for endonuclease cleavage. We did not expect to rescue more than about 25% of wild-type vRNA cleavage activity in our modified cRNA promoter constructs. This prediction was based on the fact that each of the two base pair differences between vRNA and cRNA in the stems of the hairpin loops decreased the level of activity by about 50%.

We used a standard endonuclease cleavage assay (8, 11, 12, 16), incubating the excess 49-nt-long mutant cRNAs (about 1 pmol in a 5-μl reaction mixture) with a recombinant influenza virus RNA polymerase in the presence of a capped ³²P-labeled 67-nt RNA substrate. The capped 67-nt substrate was partially cleaved in the presence of the 49-nt wild-type vRNA to give a 12-nt labeled product, which was analyzed by 15% polyacrylamide gel electrophoresis and sensitive autoradiography. The yields of the substrate and product (corrected for any background product in the presence of control cRNA) were measured by phosphorimage analysis for each mutant cRNA, and the level of endonuclease activity was expressed as a percentage of the amount of the substrate cleaved in the presence of wild-type vRNA. All mutant cRNAs, wild-type vRNA, and cRNA were prepared by T7 RNA polymerase runoff transcription of the mutant plasmids, as described previously (11). The recombinant influenza virus RNA polymerase was prepared by coinfection of HeLa cells with recombinant vaccinia virus vectors expressing the PB1, PB2, and PA subunits of the influenza virus RNA polymerase, as described previously (11). The capped, ³²P-labeled 67-nt substrate was prepared by SP6 RNA polymerase runoff transcription of SmaI-digested pGEM-7Zf(+) plasmid. After purification of the RNA, the RNA was capped to give a cap 0 structure by labeling with [α-³²P]GTP (3,000 Ci/mmol; Amersham) by means of the vaccinia virus capping enzyme (guanylyltransferase; Gibco BRL), as described previously (11).

Figure 2 shows the results of a typical experiment with 10 of the 13 mutant cRNA constructs (lanes 1 to 10) compared with wild-type vRNA (lane b) as a positive control and with wild-type cRNA (lane d) and no added RNA (lane c) as negative controls. Figure 3 shows the quantitative endonuclease cleavage results for all 13 mutant cRNAs studied, expressed as percentages of the activity level in the presence of wild-type vRNA. Quantitative experiments were performed in duplicate for 3 mutant cRNAs and in triplicate for the other 10 mutant cRNAs, and standard deviations were calculated.

FIG. 2.

Endonuclease cleavage results of selections of the mutated cRNA promoter constructs. Lane a, 15-nt RNA marker labeled at its 5" end with ³²P (M); lane b, wild-type (wt) 49-nt vRNA control; lane c, no added RNA; lanes 1 to 10, 49-nt cRNAs with indicated vRNA-specific mutations (Fig. 1); lane d, 49-nt cRNA control. Quantitative analysis (Fig. 3) detected low yields of products (<5% of the level of wild-type activity) in lanes 1 to 3, 6, 8, and 10, which were not visible in the figure. S, a 67-nt substrate plus a slower-moving, unidentified band; P, a 12-nt cleavage product. Other faint bands are nonspecific RNase degradation products. Analysis was performed by 15% polyacrylamide gel electrophoresis in 7 M urea followed by autoradiography with Kodak BioMax Trans-Screen HE intensifying screens and Kodak BioMax MS film at −70°C.

FIG. 3.

Quantitation of the endonuclease cleavage activity levels of 13 mutant cRNA promoter constructs, expressed as percentages of that of wild-type vRNA. Lanes 1 to 10 show results for the same mutant constructs in the same order as in Fig. 2. Lanes 11 to 13 show results for three additional mutant constructs not shown in Fig. 2.

To determine whether promoter arm lengths (Fig. 1) direct the differential levels of activity of cap snatching by the two RNA species, we introduced vRNA-like mutations into a cRNA construct to change its promoter arm lengths to those characteristic of vRNA. First, an insertion was made between cRNA residues 9" and 10" (mutation 10" I). Second, a deletion of nucleotide 10 was made (mutation 10 D). Finally, a mutant cRNA with an insertion between residues 9" and 10", combined with a deletion of nucleotide 10, was synthesized (mutant 10" I plus 10 D). It can be seen (Fig. 3, lane 11) that the double mutation, with which the cRNA possessed vRNA-length promoter arms, raised the level of endonuclease cleavage to 8% of the wild-type activity level. The insertion or deletion alone barely raised the level of activity above that of cRNA (Fig. 3, lanes 1 and 2) (P value [Student t test comparing results for lanes 1 and 11] = 0.021).

The third difference between the cRNA and vRNA promoters is the identity of residue 5 of the 3" arm, which is a U in vRNA but a C in cRNA. When cRNA residue 5 of the 3" arm was changed from C to U, the level of activity was barely higher than background levels (Fig. 3, lane 3). Moreover, when this mutation was combined with other upregulatory mutations (see below), no further significant stimulation of endonuclease cleavage was recorded (Fig. 3, compare lanes 1 and 8, 2 and 10, 4 and 12, and 7 and 13), suggesting that residue 5 of the 3" arm is unimportant for the activation of cap snatching.

The fourth difference between the vRNA and cRNA promoters is the identity of residue 5 of the 5" arm, which is an A in cRNA but a G in vRNA. It is known that this G in the 5" arm of vRNA is required for efficient cap snatching (12). Therefore, we investigated the effect of replacing the cRNA residue A with a G (5" A→G). When this substitution was made in isolation, the mutant cRNA activated endonuclease cleavage to 5% of the wild-type level. (Fig. 2, lane 4, and Fig. 3, lane 4). When the substitution was combined with the 10" I mutation, activity increased to 13% of the wild-type level (Fig. 2 and 3, compare lanes 4 and 5), but when the substitution was combined with the 10 D mutation, activity was reduced to 2.5% of the wild-type level (Fig. 2 and 3, compare lanes 4 and 6). However, in both cases, the quantitative changes were not statistically significant. To study the effect of making two complete vRNA-like changes, the 5′ A→G substitution was combined with a swapping of promoter arm lengths (mutation 5" A→G plus 10" I plus 10 D). This increased endonuclease activity significantly, to 31% of the wild-type level (Fig. 2 and 3, lanes 7). When this triple mutation was combined with the 5 C→U mutation in the 3" arm to make a quadruple mutation, activity was increased to 35% of the wild-type level, but this increase was not statistically significant, providing further evidence that residue 5 of the 3" arm is not involved in the differential activation of cap snatching (Fig. 2 and 3, lane 13).

Although four changes were made to the cRNA promoter, there remained the four nucleotide differences in the base-paired stems of the hairpin loops between the chimeric cRNA promoter and the true vRNA promoter. We have previously shown for vRNA that endonuclease activity is maintained as long as the base-paired residues involved in the formation of the 3" and 5" hairpin loops are conserved. However, some loss of activity occurred, suggesting that the sequences within the stems of the hairpin loops as well as the secondary structures were required for full activity. The fact that we could not rescue more than 35% of vRNA endonuclease activity in mutant cRNAs is fully consistent with this previous data (11, 12) and with our prediction (see above) that we should have been able to rescue 25% % of the activity with the mutant cRNAs studied in this paper.

In summary, it appears that the most important elements of vRNA for endonuclease activation are the 3" and 5" promoter arm lengths and the identity of nucleotide 5 in the 5" hairpin loop. The identity of residue 5 of the 3" promoter arm is unimportant for endonuclease cleavage.