|Home | About | Journals | Submit | Contact Us | Français|
Coronavirus nonstructural proteins 1 to 3 are processed by one or two papain-like proteases (PLP1 and PLP2) at specific cleavage sites (CS1 to -3). Murine hepatitis virus (MHV) PLP2 and orthologs recognize and cleave at a position following a p4-Leu-X-Gly-Gly-p1 tetrapeptide, but it is unknown whether these residues are sufficient to result in processing by PLP2 at sites normally cleaved by PLP1. We demonstrate that exchange of CS1 and/or CS2 with the CS3 p4-p1 amino acids in engineered MHV mutants switches specificity from PLP1 to PLP2 at CS2, but not at CS1, and results in altered protein processing and virus replication. Thus, the p4-p1 residues are necessary for PLP2 processing but require a specific protein or cleavage site context for optimal PLP recognition and cleavage.
Coronaviruses are positive-strand RNA viruses that translate their first open reading frames (ORF1a and ORF1b) into polyproteins that are processed by viral proteases into intermediate and mature nonstructural proteins (nsp1 to -16) (Fig. (Fig.11 A) (4, 7, 17, 20). nsp1, -2, and -3 are liberated at cleavage sites (CSs) between nsp1-2 (CS1), nsp2-3 (CS2), and nsp3-4 (CS3) by one or two papain-like protease (PLP) activities encoded within nsp3 (1, 2, 12, 13, 15) (Fig. (Fig.1B).1B). Murine hepatitis virus (MHV) and human coronavirus 229E (HCoV-229E) use two PLPs (PLP1 and PLP2) to process at CS1 to -3, while severe acute respiratory syndrome coronavirus (SARS-CoV) and avian infectious bronchitis virus (IBV) use a single PLP each (PLpro and PLP2, respectively) (10, 20, 25, 26). The factors determining the evolution and use of one versus two PLPs by different coronaviruses for processing of nsp1, -2, and -3 are unknown. Mutations at MHV CSs or within PLP1 alter replication and protein processing in surprising ways (8, 13). Loss of processing at MHV CS1 and CS2 by CS deletion or mutation results in changes in the timing and extent of virus replication. Inactivation of MHV PLP1 is more detrimental for virus replication than deletion of CS1 and CS2 or than inactivation of PLP1 combined with the CS deletions, even though not all of the mutant viruses process at CS1 or CS2 or display similar protein processing phenotypes. In contrast to MHV results, the HCoV-229E PLP1 and PLP2 have both been shown to process at CS1 and CS2, albeit at different efficiencies (Fig. (Fig.1B)1B) (24). Finally, the single SARS-CoV PLP2 homolog (PLpro) mediates efficient processing at CS1 to -3, each of which has an upstream position 4-Leu-X-Gly-Gly-position 1 (p4-LXGG-p1) amino acid motif implicated in PLpro processing (10, 16, 18). MHV possesses a p4-LXGG-p1 sequence only at CS3 and is cleaved by PLP2. These results suggest that p4-LXGG-p1 may be the critical determinant of recognition by PLP2/PLpro, but this hypothesis has not been tested in studies of replicating virus. Thus, it remains unknown whether the differences in PLP/CS recognition and processing are determined by the proximal p4-p1 residues (22).
In this study, we used MHV as a model to test whether PLP/CS specificities could be switched by an exchange of CS amino acid sequences and to determine the impact of CS exchange on protein processing and virus replication. Replacement of the CS3 p4-LKGG-p1 at CS2, but not at CS1, was sufficient for a switch in protease specificity from PLP1 to PLP2. Some combinations of CS exchange could not be recovered with inactive PLP1, and recovered mutant viruses had altered protein processing and/or impaired growth compared to the wild type (WT). The results confirm that p4-LXGG-p1 amino acid sequences are necessary determinants of cleavage by PLP2 but also indicate that a larger cleavage site or a different protein context is required for efficient recognition and processing. Finally, the results support the conclusion that complex relationships with respect to the timing and extent of PLP/CS interactions are essential for successful replication and, likely, for virus fitness.
To determine the effects of alterations in PLP/CS interactions, mutations were engineered in the MHV genome that resulted in replacements of CS1 and/or CS2 by the CS3 p4-LKGG-p1 amino acid sequence in the presence of active PLP1 or inactive PLP1 (P1ko) (Fig. (Fig.1C)1C) (6, 8, 23). Infectious virus was recovered from mutant genomes CS1(3), CS2(3), CS1/2(3), and CS1/2(3)+P1ko. Virus stocks of all recovered mutant viruses at passage 1 retained the engineered mutations, and no other mutations were identified within 300 nucleotides flanking each side of the cleavage site. The CS1/2(3)+P1ko mutant virus also retained the introduced P1ko C1121A/T1122A substitutions (8). In contrast, CS1(3)+P1ko and CS2(3)+P1ko mutant viruses could not be recovered following multiple attempts. Thus, the P1ko could be recovered only with both CS1(3) and CS2(3) exchange. This outcome is consistent with the results of our previous study, in which P1ko was recovered in combination with CS1 and CS2 deletion (ΔCS1/ΔCS2) but the presence of P1ko was highly detrimental in the presence of intact CS1 and CS2, and supports a model proposing that interaction of inactive P1ko with intact CS1 and/or CS2 alters overall protein folding and impairs functions of nsp1, -2, and -3 during virus replication (8).
To determine the effects of the CS1(3) or CS2(3) replacement on processing of nsp1, -2, and -3, cytoplasmic lysates of radiolabeled mock-, WT-, or mutant virus-infected cells were immunoprecipitated with antibodies against nsp1, nsp2, and nsp3 as well as nsp5 (3Clpro)-processed nsp8 (Fig. 2A and B) (3, 5, 9, 19, 21). The CS1(3) and CS2(3) mutants were compared with the previously characterized ΔCS1 and ΔCS2 viruses that lack the p2-p1′ CS residues and consequently have no processing at the deleted sites (8, 9). For WT and the CS2(3) mutant viruses, nsp8 was detected at equivalent levels, suggesting similar replicase polyprotein translation processes and demonstrating that the nsp5-protease function is intact (6). nsp1, nsp2, and nsp3 were also detected as mature proteins, as was the known nsp2-3 precursor (Fig. (Fig.2A2A and and2B)2B) (8, 9). However, compared to WT virus, the CS2(3) mutant virus demonstrated increased levels of uncleaved nsp2-3 and decreased mature nsp2. The processing pattern of CS2(3) also differed from that of the ΔCS2 virus in which no mature nsp2 or nsp3 was detected, as was expected based on ablated processing between nsp2 and nsp3 (8). Thus, the CS2(3) virus appeared to have intact processing at the native CS1 and also at the substituted CS2(3) p4-LKGG-p1 cleavage site, albeit at a reduced efficiency compared to that seen with WT CS2. In contrast, the processing pattern of the CS1(3) virus was similar to that of the ΔCS1 virus but distinct from that of WT, with mature nsp3 but no nsp1 or nsp2 detected and with detectable nsp1-2 and nsp1-2-3 precursors. Therefore, processing appeared to be intact at the native CS2 but absent at CS1(3), suggesting that the substituted p4-LKGG-p1 amino acid sequence was not sufficient for processing of CS1(3) by PLP1 or PLP2.
We next determined whether alterations in protein processing by the CS1(3) and CS2(3) mutant viruses were associated with changes in virus replication (Fig. 2C to F). The CS2(3) mutant virus grew with kinetics and a peak titer indistinguishable from those of WT during both single-cycle and multiple-cycle infections of DBT cells, indicating that replacement of the CS2 p4-FPCA-p1 by CS3 p4-LKGG-p1 supported WT-like growth in culture, even though processing at CS2(3) took place at a lower level than at the native CS2 (Fig. 2E and F) (8). The CS1(3) virus demonstrated growth kinetics intermediate between those of the WT and ΔCS1 mutant viruses (Fig. 2C and D). This result was surprising, since there was no detectable processing at CS1(3), and we therefore predicted that the mutant virus would demonstrate the same impaired growth phenotype as ΔCS1. The growth of CS1(3) virus suggests the possibility that there is processing at CS1(3) below the level of detection by immunoprecipitation or, alternatively, that substitution of p4-LKGG-p1 independently compensates for loss of cleavage at that site.
Since both PLP1 and PLP2 were present in the CS1(3) and CS2(3) mutant viruses, the details of protease processing at the altered cleavage sites could not be determined. To test whether PLP1 or PLP2 or both were active at altered cleavage sites, we compared protein processing of the CS1/2(3) and CS1/2(3)+P1ko mutant viruses with that of the WT, ΔCS1/2, and ΔCS1/2+P1ko viruses (Fig. (Fig.3)3) (8). Infection with WT virus resulted in detectable mature nsp1, -2, -3, and -8, in addition to the known nsp2-3 precursor. During infections with all mutant viruses, nsp1, -2, and -3 proteins and precursors were either not detected or detected in much lower abundance in relation to the nsp5-processed nsp8, even with extended metabolic labeling and exposure times for increased detection (Fig. 3A and B). The ΔCS1/2 and ΔCS1/2+P1ko viruses had no detectable nsp1, -2, -3, or -2-3 but did have detectable nsp1-2-3, as expected based on the abolishment of CS1 and CS2. In contrast, the CS1/2(3) and CS1/2(3)+P1ko viruses had identical patterns, with detectable nsp1-2-3 and nsp3 but no detectable nsp1, -2, or -2-3. The results demonstrate that cleavage of CS2(3), but not CS1(3), can occur in the setting of inactive PLP1, thereby implicating PLP2 in processing at the engineered CS2(3).
We next compared the growth characteristics of WT, CS1/2(3), CS1/2(3)+P1ko, ΔCS1/2, and ΔCS1/2+P1ko viruses (Fig. 3C and D). The CS1/2(3) and CS1/2(3)+P1ko mutant viruses exhibited identical levels of growth, indicating that inactivation of PLP1 was not important in the growth of the viruses and supporting the conclusion that PLP2 is the active protease at the CS2(3) site. It was interesting that both CS1/2(3) and CS1/2(3)+P1ko viruses showed a 4-h delay in exponential growth, a result similar to that seen with the ΔCS2 virus, while the CS2(3) virus alone did not (Fig. (Fig.2F),2F), suggesting that the processing at CS2(3) may have differed in timing in the setting of altered CS1, resulting in the characteristic delay of exponential growth in the setting of impaired or abolished nsp2-3 processing. On the other hand, both CS1/2(3) and CS1/2(3)+P1ko viruses had peak virus titers that approached WT titers and were higher than the titers seen with the ΔCS1/2 and ΔCS1/2+P1ko mutants. The titer was similar to the CS1(3) virus titers and greater than the ΔCS1 virus titers, supporting the conclusion that the p4-LKGG-p1 sequence of CS1(3) either compensates for the growth defect of loss of CS1(3) processing or, alternatively, allows processing at levels below those permitting detection by immunoprecipitation.
In this report, we have demonstrated that introduction of the CS3 p4-LKGG-p1 at CS2 results in a change of the effector protease from PLP1 to PLP2, whereas the same exchange at CS1 results in no processing by PLP2 as well as in loss of processing by PLP1. Thus, it appears that p4-p1 residues are necessary but not sufficient for recognition by PLP1 and PLP2. The demonstration that viable mutants were recovered with P1ko in combination with both CS1(3) and CS2(3), but not with P1ko and either CS1(3) or CS2(3) alone, extends our previous observation that P1ko was much more debilitating when introduced alone than when introduced in combination with deletion of both CS1 and CS2 (8). Further, the growth phenotypes of the CS1(3) and CS2(3) mutants were not predictable on the basis of the results seen with ΔCS1 and ΔCS2 viruses, suggesting that the presence or loss of cleavage alone was not the sole determinant of protein or precursor functions. The presence of two PLPs in some coronaviruses has been proposed to be the result of a paralogous duplication and subsequent evolution of the PLP2-like protease (26). Our results suggest that this involved more than simple protease duplication and cleavage site modification but rather may have required evolution of the entire nsp1, -2, and -3 domains and the included proteases and cleavage sites as a cassette or network of highly linked proteins and functions. If so, it is possible that nsp1, -2, and -3 may be resistant to recombination with distantly related viruses or that any viable recombination event might require exchange of the entire nsp1, -2, and -3 sequence. The independent evolutions of nsp1, -2, and -3 might also explain the significant variations in sequence and organization within nsp1, -2, and -3 between coronaviruses, in contrast to the much greater degree of conservation of sequence and organization of nsp4 to -16. Finally, the results would suggest the possibility that nsp1, -2, and -3 are more adaptable for change over time, a possibility supported by the increased mutation frequency in nsp1, -2, and -3 during the SARS-CoV epidemic compared to nsp4 to -16 (11). The panel of viruses generated in this study will provide powerful tools to study the evolution of coronavirus nsp1, -2, and -3.
We thank Michelle Becker, Lance Eckerle, Rachel Graham, Xiaotao Lu, and Jennifer Sparks for technical assistance and critical reviews of the manuscript.
This work was supported by NIH grant AI26603 (M.R.D.) from the National Institute of Allergy and Infectious Diseases. M.J.G. was supported by the Training Grant in Mechanisms of Vascular Disease through the Vanderbilt University School of Medicine (T32 HL007751). This work was also supported by the Elizabeth B. Lamb Center for Pediatric Research.
Published ahead of print on 28 April 2010.