Journal of General Virology (2000),
81, 853-879.
© 2000 Society for General Microbiology
Virus-encoded proteinases and proteolytic processing in the Nidovirales
John Ziebuhr1,
Eric J. Snijder2 and
Alexander E. Gorbalenya3
Institute of Virology and Immunology, University of Würzburg, Versbacher Str. 7, 97078 Würzburg, Germany1
Department of Virology, Leiden University Medical Center, LUMC P4-26, PO Box 9600, 2300 RC Leiden, The Netherlands2
Advanced Biomedical Computing Center, 430 Miller Dr., Rm 235, SAIC/NCI-FCRDC, Frederick, MD 21702-1201, USA3
Author for correspondence: John Ziebuhr. Fax +49 931 2013934. e-mail ziebuhr{at}vim.uni-wuerzburg.de
 |
Introduction
|
|---|
On the basis of similarities in their genome organization and replication strategy, RNA viruses can now be classified into ‘supergroups’ that often include both animal and plant viruses (Goldbach & Wellink, 1988
; Strauss & Strauss, 1988 ). This concept is also increasingly reflected in the taxonomy of viruses; in particular by the introduction of the taxon ‘order’, which combines virus families for which a common ancestry seems highly probable (Mayo & Pringle, 1998 ). For the positive-stranded, enveloped coronaviruses and arteriviruses, which have recently been unified in the order Nidovirales (Cavanagh, 1997 ), a close phylogenetic relationship has been established on the basis of their similar polycistronic genome organization, the use of common transcriptional and (post)-translational strategies and the conservation of an array of homologous replicase domains (den Boon et al., 1991 ). Thus, it is possible to draw a common outline of the nidovirus life-cycle (Fig. 1) (for reviews see Lai & Cavanagh, 1997 ; de Vries et al., 1997 ; Snijder & Meulenberg, 1998 ). In some respects, however, the two virus families differ significantly from each other. For example, the largest coronavirus genome, that of mouse hepatitis virus (MHV) with 31·5 kb, is about two and a half times the size of the smallest arterivirus genome, the 12·7 kb RNA of equine arteritis virus (EAV). Also, the structural proteins of both virus families are not evidently related, resulting in important differences in virion size and structure (den Boon et al., 1991 ; Snijder & Spaan, 1995 ; de Vries et al., 1997 ).
View larger version (39K):
[in this window]
[in a new window]
|
Fig. 1. Outline of the nidovirus life-cycle showing the most important similarities between coronaviruses and arteriviruses. ORFs in the polycistronic genome are indicated as boxes. ORFs that are translated from specific mRNAs are shown in green, while the downstream, non-translated ORFs are shown in red. The replicase gene, encompassing ORFs 1a and 1b, envelope protein genes (E1 to E3) that may vary in number and the genes for the triple-spanning membrane (M) protein and nucleocapsid (N) protein are shown. It should be stressed that this outline is a generalization. For example, the M and N genes are usually, but not always, the two most 3'-proximal ORFs in the genome. Many nidovirus genomes contain a variable number of additional (structural and non-structural) genes and, consequently, more than five subgenomic mRNAs are often produced. Furthermore, subgenomic mRNA transcription and genome replication involve the synthesis of minus-stranded intermediates that are not depicted here. The circle (L) at the 5' end of the genome represents the common leader sequence that is also present at the 5' end of the subgenomic mRNAs that are shown below the genome. The two replicase polyproteins (1a and 1ab) are depicted as solid black lines. The processing products are depicted as interrupted lines. The conserved domains/functions encoded by the replicase gene are abbreviated as follows: AP, accessory proteinase; hd, hydrophobic domain; MP, main proteinase; RdRp, RNA-dependent RNA polymerase; Z, putative zinc finger; HEL, NTPase/RNA helicase; C, conserved domain specific for nidoviruses. The amino-terminal parts of the nidovirus replicase polyproteins are processed by accessory proteinases, whereas the remainders of the polyproteins are cleaved by the main proteinase.
|
|
Most major groups of positive-stranded RNA viruses of animals produce either a single polyprotein or separate non-structural and structural precursor polypeptides that are subsequently cleaved by virus-encoded or host-encoded proteinases to produce functional subunits (Dougherty & Semler, 1993 ). In contrast, the nidovirus structural proteins, which are encoded in the 3'-proximal region of the genome, are individually expressed from a nested set of subgenomic mRNAs (Fig. 1
) generated by a unique discontinuous transcription mechanism (Spaan et al., 1983 ; Lai et al., 1984 ; van Marle et al., 1999a ). Apparently, virus-encoded proteinases are not involved in the maturation of the structural proteins; although a number of nidovirus envelope proteins undergo proteolytic processing by cellular proteinases as they follow the exocytotic pathway. Thus, for example, signal sequences are co-translationally removed and the coronavirus spike glycoprotein is subject, in some cases, to a host-directed maturation cleavage (Cavanagh, 1995 ).
In contrast to the expression of the structural genes, nidovirus-encoded proteinases play a prominent role in the expression of the replicase gene. The replicase proteins are encoded by two large, 5'-proximal open reading frames (ORFs) that occupy approximately two-thirds to three-quarters of the genome (Fig. 1). The division of the replicase gene into ORFs 1a and 1b, which are connected by a ribosomal frameshift site (Brierley et al., 1987 ), is one of the hallmarks of the nidoviruses. It results in the translation of an ORF1a protein and a carboxyl-extended ORF1ab frameshift protein; also known as replicase polyproteins 1a and 1ab (pp1a and pp1ab)
b. The size of the frameshift protein ranges from 3175 amino acids for the arterivirus EAV to about 7200 amino acids for the coronavirus MHV. The nidovirus ORF1a and ORF1ab translation products are polyprotein precursors which are cleaved by viral proteinases at a minimum of 10 (arteriviruses) or 13 (coronaviruses) sites.
Comparative sequence analyses (Boursnell et al., 1987 ; Gorbalenya et al., 1989b ; Bredenbeek et al., 1990a ; Snijder et al., 1990 ; den Boon et al., 1991 ; Lee et al., 1991 ; Godeny et al., 1993 ; Herold et al., 1993 ; Meulenberg et al., 1993 ; Eleouet et al., 1995 ) and recent experimental data (van Dinten et al., 1997 , 1999 ) suggest that several ORF1b-encoded replicase subunits are directly involved in viral RNA synthesis. The translational downregulation of one of these functions, the (putative) viral RNA-dependent RNA polymerase (RdRp), can also be found in a number of other virus systems of which the alphaviruses have been characterized in most detail (for a review, see Strauss & Strauss, 1994 ).
With respect to the ORF1a-encoded subunits of the nidovirus replicase, two main functions have emerged so far. First, hydrophobic domains in the arterivirus ORF1a protein have been shown to mediate the membrane association of the replication complex and to be able to dramatically alter the architecture of host cell membranes (van der Meer et al., 1998 ; Pedersen et al., 1999 ). A similar role can also be expected for the corresponding hydrophobic segments of the coronavirus replicase polyproteins (Shi et al., 1999 ; van der Meer et al., 1999 ). Second, the ORF1a-encoded regions of the nidovirus replicase polyproteins harbour a variety of proteolytic activities, which will be the topic of this review. Although our knowledge of the biochemical and structural properties of the nidovirus proteinases is still very limited, the available data underline the idea that, as for many other positive-stranded RNA viruses, these enzymes fulfil a crucial role in the regulation of the virus life-cycle.
This review article is organized into five main sections. In the first section, we introduce nidovirus proteinases and classify them into main and accessory proteinases. In the next two sections, we present a brief overview of the two classes of the nidovirus proteinases and, in this context, the coronavirus and arterivirus enzymes are compared to each other and to the prototypic proteinases. This is followed by a detailed description of the nidovirus proteinases themselves. The article is concluded by two sections that describe the regulatory role of proteinases during virus replication and then give an outline of future perspectives concerning nidovirus proteinases. The reader will note that the in vitro characterization of the proteinases of coronaviruses is more advanced than that of arteriviruses. In contrast, current knowledge on nidovirus proteolytic regulation in vivo is essentially derived from research on arteriviruses.
|
‘Main’ and ‘accessory’ proteinases of nidoviruses
|
|---|
A discussion of the diverse group of proteolytic enzymes found in the Nidovirales requires the introduction of a standardized nomenclature. A consensus on this matter has not yet been reached. Indeed, different names are used to describe the same proteinase (Table 1). In most cases, these names allude to the relationships between the nidovirus proteinases and other proteolytic enzymes of viral or cellular origin, the so-called prototypic proteinases. However, they may also refer to major structural or biochemical properties of the proteinase (e.g. fold, catalytic system or substrate specificity; Table 1) or to its position in the virus-encoded polyprotein (a characteristic feature of positive-stranded RNA virus proteinases). Thus, the current names can become both complex and confusing. This is particularly true if combinations of different, sometimes even incompatible, properties are used, or if the specific property to which the name refers is not clear. For example, the designation ‘chymotrypsin-like (CHL) proteinase’ is meaningless, unless it is clearly stated whether this name refers to (i) the catalytic system, (ii) the substrate specificity, (iii) the fold of the proteinase, or (iv) to an additional property in which the enzyme resembles chymotrypsin. Throughout this review, we will use the names given in the second column of Table 1. Depending on the topic discussed, these names may further be elaborated to underscore a specified property.
In this review, we will discuss the nidovirus proteinases in the context of their role as either ‘main’ or ‘accessory’ (leader) proteinases (Gorbalenya et al., 1991 ). The main proteinase is defined as the enzyme that directly mediates the expression of the two most conserved replicase domains, RdRp and RNA helicase (HEL). We believe that this classification can be universally applied to the proteinases of all positive-stranded RNA virus families and, furthermore, it will facilitate their comparison and characterization.
|
Nidovirus main proteinases
|
|---|
Overview
All arteriviruses and coronaviruses encode one main proteinase. For the reasons specified below, this main proteinase is commonly referred to as ‘3C-like’ (after the 3C proteinases of the Picornaviridae
c). Thus, the coronavirus enzyme will be called coronavirus 3C-like proteinase, 3CLpro. Since the nucleophilic, catalytic residue of the arterivirus main proteinase is a serine and differs from that of other 3C-like proteinases, i.e. a cysteine, we have decided to stress this specific property by using the designation 3C-like serine proteinase, 3CLSP, for the arterivirus enzyme.
The 3C-like proteinases of arteriviruses and coronaviruses occupy comparable positions in the replicase polyproteins (Fig. 2). They reside upstream of the ribosomal frameshift site and domains, including RdRp and HEL, which belong to the most conserved domains in this virus order. The 3C-like proteinases are autocatalytically processed at flanking sites§d (Fig. 3A, B) and direct the proteolytic processing of all downstream domains of the replicase polyproteins, in both cases at similarly positioned sites (Fig. 2). This central role in the expression of the major replicative proteins justifies the designation of the 3C-like proteinases as the ‘main’ proteinase of nidoviruses.
View larger version (80K):
[in this window]
[in a new window]
|
Fig. 3. Multiple alignments of nidovirus main proteinases. The alignments of coronavirus 3CLpro domains (A) and arterivirus 3CLSPs (B) were generated using the ClustalX program (Thompson et al., 1997 ). The positions of the amino-terminal and carboxyl-terminal residues of the alignment are indicated at the left and right sides, respectively, and follow the ORF1a polyprotein numbering. Coloured background: red, (putative) catalytic residues; pink, residues thought to be equivalent to the catalytic acidic residue (see text); violet, (putative) substrate-binding residues; blue, invariant residues; black, dark and light grey, residues conserved in 100%, 75% or 50%, respectively, of the sequences. Conservation groups: I, V, L, M; F, Y; K, R; D, N; E, Q; S, T. Yellow colouring, along with a vertical line marks the position of cleavage sites. The carboxyl-terminal domain (extension) is underlined but its border with the amino-terminal enzymatic domain is provisional. LDVC and LDVP, lactate dehydrogenase-elevating virus neurovirulent type C (Godeny et al., 1993 ) and strain Plagemann (Palmer et al., 1995 ), respectively; PRRSVLV and PRRSVVR, porcine reproductive and respiratory syndrome virus strain Lelystad (Meulenberg et al., 1993 ) and strain ATCC VR-2332 (Nelsen et al., 1999 ), respectively; MHVA, mouse hepatitis virus strain A59 (Bonilla et al., 1994 ); IBVB, avian infectious bronchitis virus strain Beaudette (Boursnell et al., 1987 ); for the other viruses see text and Tables. The National Center for Biological Information sequence ID: EAV, 133455; LDVC, 293038; LDVP, 564003; PRRSVLV, 482963; PRRSVVR, 4580009; IBVB, 138147; HCoV, 464694; TGEV, 872319; MHVA, 453423 (nucleotide).
|
|
Nidovirus 3C-like proteinases and the prototypic 3C proteinases of the Picornaviridae possess similar P1|P1' substrate specificity (Fig. 4), which is determined, in part, by the conserved substrate-pocket His residue. They are also likely to employ similar folds and because of these similarities the nidovirus proteinases are ‘3C-like’. However, nidovirus 3C-like enzymes differ in at least two significant respects from the picornavirus 3C proteinases and their relatives, the 3C-like proteinases of the picornavirus-like supergroup (see also Ryan & Flint, 1997 ). First, the nidovirus 3C-like proteinases use catalytic residues that are similar but not identical to the triad His, Asp (Glu) and Cys (hereafter, the linear sequence order is given) found in 3C/3C-like proteinases of the picornavirus-like supergroup. Second, both the nidovirus and the picornavirus-supergroup 3C/3C-like proteinases belong to (and process) a highly conserved array of replicative domains whose linear orders are conserved and supergroup (order)-specific. In the picornavirus-like supergroup, this conserved arrangement of replicative domains can be described as HEL-3C(L)-RdRp, which differs from the elaborated formula of the Nidovirales: PLpro-HD1–3CL-HD2-RdRp-Zn-HEL-C; where 3C(L) is 3C/3C-like proteinase, PLpro is papain-like proteinase, HD1 and HD2 are hydrophobic domains 1 and 2, Zn is a putative zinc-binding domain, and C is a conserved, nidovirus-specific domain.
View larger version (30K):
[in this window]
[in a new window]
|
Fig. 4. Conservation of sites cleaved by nidovirus main proteinases. Two separate multiple, gap-free alignments around the P1|P1' positions of the sites cleaved, or predicted to be cleaved, by arterivirus 3CLSPs and coronavirus 3CLpro domains, respectively, were converted to logos presentations (Schneider & Stephens, 1990 ) in which the size of an amino acid is proportional to its conservation at the specific position and the sampling size. The amino acid conservation is measured in bits of information plotted on a vertical axis whose upper limit is determined by the natural diversity of amino acids (twenty) expressed as a logarithm of two. One standard deviation of the information content at each position is indicated by vertical bars.
|
|
The catalytic residues of 3C and 3C-like proteinases are grafted upon a two-
-barrel structure consisting of 12 antiparallel -strands (Bazan & Fletterick, 1988 ; Gorbalenya et al., 1989a ; Allaire et al., 1994 ; Matthews et al., 1994 ; Mosimann et al., 1997 ). This structure was originally identified in a subset of cellular serine proteinases of which chymotrypsin (or trypsin) is the prototype. Therefore, nidovirus 3C-like proteinases may also be called ‘chymotrypsin-like’. In 3C and 3C-like proteinases, Cys replaces the nucleophilic Ser and, in a subset of viruses, Glu replaces the Asp of the catalytic triad found in cellular proteinases (Bazan & Fletterick, 1988 ; Gorbalenya et al., 1989a ; Matthews et al., 1994 ). The 3C-like proteinases of both corona- and arteriviruses deviate from the prototypic 3C enzymes, but, surprisingly, each of them in a different direction. Whilst the arterivirus 3CLSP employs the canonical His-Asp-Ser triad that is usually found in cellular proteinases, the coronavirus 3CLpro uses Cys as its catalytic nucleophile. However, the coronavirus 3CLpro seems to lack a conserved acidic residue that would be equivalent to the catalytic Asp (Glu) of 3C proteinases. Furthermore, coronavirus and arterivirus 3C proteinases display a very low overall sequence similarity. Basically, the similarity is limited to the regions of active-site residues and, even there, it is barely detectable above the background in sensitive profile-versus-profile dot-plot comparisons (Fig. 5). As a result, any general alignment of the primary structures of these proteinases is largely arbitrary. The same is evident upon comparison of coronavirus 3CLpro domains with the prototypic picornavirus 3C proteinases (Fig. 5). In contrast, arterivirus 3CLSPs and picornavirus 3C proteinases are clearly similar in at least one region, which includes the conserved catalytic nucleophile (Ser or Cys, respectively) and the substrate-binding His. An even more pronounced, albeit still limited, similarity can be traced between arterivirus 3CLSPs and the putative proteolytic enzymes (also called 3C-like proteinases) of sobemoviruses and luteoviruses; two families of positive-stranded RNA plant viruses (Gorbalenya et al., 1988 ; Snijder & Gorbalenya, 1996 ). This similarity is correlated with (and probably related to) the conservation of the same set of active-site residues in these enzymes. Thus, despite their similar name and function, the coronavirus and arterivirus 3C-like proteinases have diverged pronouncedly from each other and from the prototypic proteinases. While the coronavirus enzymes are likely to form a separate, highly diverged branch of the two--barrel proteinases, the arterivirus 3CLSPs may group together with non-nidovirus proteinases. These and other striking differences should be kept in mind when the general features of nidovirus proteinases and nidovirus replicase processing are discussed.
View larger version (24K):
[in this window]
[in a new window]
|
Fig. 5. Dot-plot cross-comparisons of arterivirus 3CLSPs and coronavirus 3CLpro domains with other 3C/3C-like proteinases. The profiles were generated using multiple alignments of different, closely related 3C/3C-like proteinases indicated by their numbers and origins. They were compared in pairs in the dot-plot fashion using the Proplot program (Thompson et al., 1994 ). Two profiles were compared by sliding a window of 21, 23 or 25 amino acids along each possible register and matches between two profiles that were within the top 0·5% were marked by dots. The projected positions of the catalytic residues (H*, D* or E*, C* or S*), as well as the substrate-binding H, are shown at each axis. Those dots, which lay at any of the four possible crosses of projections of two functionally equivalent residues (e.g. H* and H*) or close to a non-visible diagonal passing these crosses, belong, or may belong, to the true matches between two profiles. The rest of dots are background hits (false positives). These data are from an unpublished work of A. E. Gorbalenya.
|
|
Main proteinase-mediated processing of the coronavirus replicase polyproteins
The existence of a nidovirus 3C-like proteinase was initially predicted through computer-assisted, comparative sequence analysis of the complete replicase polyprotein of an avian coronavirus, infectious bronchitis virus (IBV), and the picornavirus 3C proteinases (Gorbalenya et al., 1989b ). In the same study, the processing of a dozen putative cleavage sites by the IBV 3CLpro was predicted to yield the mature components of the replicase complex. All of these sites contained Q|S,G dipeptides, which are typical substrates of 3C/3C-like proteinases. Also, the putative cleavage sites were found to have bulky hydrophobic residues at their P2 positions. These predictions were subsequently extended by sequence analyses of the replicase genes of three other coronaviruses, MHV, human coronavirus 229E (HCoV) and porcine transmissible gastroenteritis virus (TGEV) (Lee et al., 1991 ; Herold et al., 1993 ; Eleouet et al., 1995 ). Specifically, both the 3CLpro domain (~300 amino acids) (Fig. 3A) and the vast majority of the predicted cleavage sites were found to be conserved in coronaviruses (Figs 4 and 3A). Collectively, these studies have produced a model of the 3CLpro-mediated processing of the coronavirus replicase polyprotein(s) which forms the basis of many current experimental studies.
The first experimental evidence for a coronavirus proteinase activity encoded in the 3'-proximal ORF1a sequence was reported for IBV (Liu et al., 1994 ). In this study, the expression of the putative RdRp domain was shown to involve a virus-encoded proteinase activity that mapped to a region previously predicted to contain a 3CLpro domain (Gorbalenya et al., 1989b ). It was concluded that the proteolytic activity observed was 3CLpro-mediated. Evidence supporting this hypothesis was obtained from three studies on the 3CLpro domains of IBV, MHV and HCoV 229E (Liu & Brown, 1995 ; Lu et al., 1995 ; Ziebuhr et al., 1995 ). It is worth noting that, in these initial studies, the coronavirus 3CLpro proved to be active in quite different expression systems. The MHV 3CLpro was expressed in vitro in a rabbit reticulocyte lysate, the HCoV 229E 3CLpro was expressed in Escherichia coli, and the IBV 3CLpro was expressed in Vero cells using the recombinant vaccinia virus/T7 system. To further analyse the 3CLpro-mediated processing of the coronavirus replicase proteins, two of the systems mentioned above were exploited. In the case of IBV, the 3CLpro was co-expressed with substrate proteins derived from different portions of the replicase polyprotein(s) and 3CLpro-mediated proteolysis of substrates containing specific, mostly predicted, cleavage sites was observed (Liu et al., 1994 , 1997 , 1998 ; Liu & Brown, 1995 ; Ng & Liu, 1998 ). The cleavage products were identified by their apparent molecular mass in SDS–PAGE, and the cleavage sites were mapped by site-directed mutagenesis. For HCoV and MHV, the assay systems used were based upon bacterially expressed 3CLpro domains (Ziebuhr et al., 1995 ; Herold et al., 1996 ; Seybert et al., 1997 ). This approach greatly facilitated the identification and N-terminal sequence analysis of 3CLpro cleavage sites by using both in vitro-translated and recombinant protein substrates (Ziebuhr et al., 1995 ; Grötzinger et al., 1996 ). Furthermore, it allowed for the determination of kinetic parameters using synthetic peptides combined with quantitative analysis of the substrate conversion (Ziebuhr & Siddell, 1999 ). Finally, in the MHV system, it has even proved possible to perform amino-terminal microsequence analysis of metabolically labelled 3CLpro cleavage products isolated by immunoprecipitation from virus-infected cells (Lu et al., 1998 ).
The information collected during studies with different coronaviruses can now be used to map the 3CLpro processing sites in the coronavirus replicase polyprotein (Fig. 2A). Taken together, at least 12 processing end-products (including the 3CLpro itself) are generated by 3CLpro-mediated cleavage. The processing products of HCoV, IBV and MHV that have been identified so far are summarized in Table 2. Although the proteolytic processing of TGEV, another coronavirus with a known genome sequence (Eleouet et al., 1995 ), has not yet been characterized experimentally, its close relationship with HCoV allows for reliable functional predictions.
View this table:
[in this window]
[in a new window]
|
Table 2. End-products of the 3CLpro-mediated proteolytic processing of the coronavirus replicase polyproteins pp1a and pp1ab
|
|
Catalytic centre and substrate specificity of the coronavirus main proteinase
Based on the analysis of the IBV sequence (Boursnell et al., 1987 ), the catalytic system of the coronavirus 3CLpro was proposed to resemble that of other viral 3C and 3C-like proteinases (Gorbalenya et al., 1989b ) and to involve a His-Asp(Glu)-Cys catalytic triad. In accordance with these predictions, mutagenesis studies with three different coronavirus 3CLpro domains provided experimental evidence that, for HCoV, His-3006 and Cys-3109, and the equivalent residues in IBV and MHV (Fig. 3A), are indispensable for activity (Liu & Brown, 1995 ; Lu et al., 1995 ; Ziebuhr et al., 1995 , 1997 ; Tibbles et al., 1996 ; Seybert et al., 1997 ). Interestingly, the replacement of the putative active-site Cys by Ser in the IBV 3CLpro was reported to produce an enzyme with residual activity in a cis-cleavage assay, supporting the relatedness of the coronavirus 3CLpro with serine proteinases of the chymotrypsin family (Tibbles et al., 1996 ). In contrast with this finding, no activity was detected for the corresponding Cys-to-Ser mutants of the MHV and HCoV 3CLpro domains using trans-cleavage assays (Seybert et al., 1997 ; Ziebuhr et al., 1997 ). The reasons for this discrepancy remain to be investigated, but may reside in the different sensitivities of trans- versus cis-cleavage assays. In conclusion, the available data strongly suggest that the conserved His and Cys residues represent the general base and nucleophile, respectively, of a charge-relay system similar to that of serine proteinases.
Special efforts have been made to identify the coronavirus 3CLpro counterpart of the third (acidic) residue present in the catalytic triad of CHL enzymes. In the original analysis of the IBV replicase sequence, Glu-2843 was aligned with the catalytic acidic residue of 3C and 3C-like proteinases (Gorbalenya et al., 1989b ). However, subsequent sequence analyses of other coronavirus replicases revealed that Glu-2841 is more conserved than Glu-2843. Also, Glu-2841 is the only residue in the region delimited by the catalytic His and Cys (Fig. 3A; cf. Fig. 3B) whose variability in coronaviruses (Asp in MHV, Glu in IBV, and Asn in HcoV and TGEV) could somehow be reconciled with the role of the third catalytic residue (Gorbalenya & Koonin, 1993 ; Gorbalenya & Snijder, 1996 ). Consequently, this position was probed by site-directed mutagenesis. However, variable results were obtained. For example, the substitution of Glu-2841 by Gln in IBV resulted in an active enzyme (Liu & Brown, 1995 ). Likewise, the replacement of Asp-3398, the IBV Glu-2841 equivalent, by either Ala or Pro did not significantly alter the activity of the MHV 3CLpro (Lu & Denison, 1997 ). In contrast, the replacement of the equivalently positioned Asn by Gly or Pro in the HCoV 3CLpro had different effects. The replacement of Asn with Gly did not change the rate of substrate conversion as compared to the wild-type enzyme in a peptide cleavage assay (Ziebuhr et al., 1997 ). However, the replacement of Asn with Pro substantially reduced the rate of substrate conversion. It is important to note, however, that, in these experiments, the enzymatic activity was not rigorously tested (e.g. by comparison of purified wild-type and mutant 3CLpro domains in quantitative assays using a range of different substrates). Therefore, it would still be premature to draw definitive conclusions concerning the role of this (conserved) residue in the function of the coronavirus 3CLpro. However, keeping in mind that all other 3C/3C-like proteinases tested so far, including those of arteriviruses, only tolerated an exchange of Glu and Asp at this position, the data seem to indicate that the coronavirus 3CLpro lacks a corresponding acidic, catalytic residue in its sequence. It remains to be seen whether an alternative conserved acidic (or even non-acidic) residue outside the region delimited by the catalytic His and Cys assumes the catalytic role and occupies a position equivalent in space to that of the catalytic acidic residue of other 3C/3C-like proteinases.
The coronavirus 3CLpro displays additional features that clearly separate it from other virus-encoded 3C-like proteinases, including the arterivirus main proteinase. For example, it employs a novel version of the substrate-binding pocket ‘core’ motif, which is characteristically Gly-X-His for most other 3C/3C-like proteinases. Thus, the Gly residue of this motif (Bazan & Fletterick, 1988 ; Gorbalenya et al., 1989a , b ) is conserved in the vast majority of serine and cysteine proteinases with CHL folds and only very few proteinases tolerate substitutions with small amino acids (Ala or Cys) at this position. This conservation pattern indicates a strong selection pressure with regard to the space that this specific residue occupies. In contrast, in all coronavirus 3CLpro domains studied so far, Gly appears to be replaced by Tyr (Gorbalenya et al., 1989b ; Lee et al., 1991 ; Herold et al., 1993 ; Eleouet et al., 1995 ) (Fig. 3A). Given the unusual nature of this replacement and the very low level of overall similarity between the coronaviral and the other CHL proteinases (Fig. 5), additional support for this theoretical assignment is needed. The Tyr residue of the Tyr-X-His motif has not yet been probed by mutagenesis. However, the replacement of the His residue (His-3127) by Ser completely abolished the proteolytic activity of the HCoV 3CLpro (Ziebuhr et al., 1997 ). This inactivation was selective since a similar replacement of His-3136, another conserved His residue in this region, was not so detrimental (Ziebuhr et al., 1997 ). Thus, the importance of the Tyr-X-His motif has been confirmed, implying that coronaviruses may indeed have accepted a Gly-to-Tyr replacement during evolution. It can be expected that this replacement is coupled to other substitution(s) in the active site to accommodate the bulky side chain of Tyr. The above data are also compatible with a model, originally developed and substantiated for other 3C/3C-like proteinases (Bazan & Fletterick, 1988 ; Gorbalenya et al., 1989a ; Allaire et al., 1994 ; Matthews et al., 1994 ; Mosimann et al., 1997 ), that implicates His-3127 (and its counterparts in other coronaviruses) in the formation of hydrogen bonds to the P1 glutamine side chain of 3CLpro substrates (Gorbalenya et al., 1989b ). The high degree of conservation of coronavirus cleavage sites upstream of the P1 position (Fig. 4) suggests that the substrate-binding pocket of the coronavirus 3CLpro may have numerous additional contacts with its substrates. The determinants of these interactions remain to be elucidated.
The substrate specificity of 3CLpro resembles that of many other 3C/3C-like proteinases (Kräusslich & Wimmer, 1988 ; Dougherty & Semler, 1993 ; Blom et al., 1996 ) in so far as the P1 position of the substrate is exclusively occupied by Gln and small, aliphatic residues (Ser, Ala, Asn, Gly and Cys) are found at the P1' position (Fig. 4). However, Asn and Cys are most uncommon as P1' residues outside of the coronaviruses, although a P1' Asn is found in rhinoviruses (Blom et al., 1996 ) and, in a mutagenesis study, Cys proved to be a tolerable substitution in one of the encephalomyocarditis virus (EMCV) 3Cpro sites (Parks et al., 1989 ). In four different coronaviruses, one 3CLpro cleavage site consistently contains Asn at the P1' position (Liu et al., 1997 ; Lu et al., 1998 ; Ziebuhr & Siddell, 1999 ) and, for MHV, a P1' Cys residue was predicted for another site (Lee et al., 1991 ). A peptide that mimicked an HCoV 3CLpro site with the P1' Asn residue was processed relatively poorly in vitro, which additionally points to the exceptional nature of Asn at this position (Ziebuhr & Siddell, 1999 ).
Upon comparison of a large number of cleavage sites, most of which have experimentally been confirmed for at least one coronavirus (Fig. 2), it is evident that in addition to P1 and P1', the P2, P3, P4, P2' and P3' positions have a restricted variability (Fig. 4). Among these, the P2 and P4 positions are most conserved with bulky hydrophobic residues (mainly Leu) at P2 and Val, Thr, Ser (and Pro) at P4 being clearly favoured (Fig. 4). A similar complexity was previously described for the primary cleavage site determinants of the potyvirus 3C-like proteinase (NIa protein). The potyvirus sites could be transferred into an alien protein where they promoted selective cleavage of the protein by the cognate 3C-like proteinase in a reaction superficially resembling the cleavage of DNA by restriction endonucleases (Carrington & Dougherty, 1988 ). The coronavirus 3CLpro cleavage sites can be predicted to possess similar properties. The efficiency of cleavage at specific sites is likely to be determined by the exact composition of the sites, since synthetic peptides mimicking different cleavage sites were processed in competition experiments at significantly different rates by the HCoV 3CLpro (Ziebuhr & Siddell, 1999 ). In view of these data, it seems likely that together with the accessibility of potential cleavage sites in the context of the polyprotein the properties of the cleavage sites themselves contribute significantly to the coordinated, temporal release of specific polypeptides from the replicase polyproteins. This might lead to the (irreversible) activation or inactivation of specific functions in the course of the virus life-cycle, as has been demonstrated for a number of other positive-stranded RNA viruses.
Structural aspects of the coronavirus main proteinase
The coronavirus 3CLpro domains are the largest proteinases of their type. They consist of 302–307 amino acids, whereas the prototypic poliovirus 3C proteinase contains only 182 residues. This size difference is due to the presence of a unique, carboxyl-terminal region of approximately 110 amino acids which appears to be required for proteolytic activity. Thus, a large number of different carboxyl-terminally truncated versions of the HCoV 3CLpro are inactive in assays using synthetic peptides (Ziebuhr et al., 1997 ; J. Ziebuhr, unpublished data). Also, the removal of 28 carboxyl-terminal amino acids from the MHV 3CLpro abolishes its activity in an in vitro translation system (Lu & Denison, 1997 ). Recently, in apparent contrast to the HCoV and MHV data it was shown that a recombinant form of the IBV 3CLpro tolerated the introduction of six consecutive His residues near its carboxyl terminus without loss of activity (Tibbles et al., 1999 ).
In the absence of a structural model for the coronavirus 3CLpro, we can only speculate on the function of the carboxyl-terminal region. Obviously, several, not mutually exclusive, functions could be related to this domain, e.g. (i) maintenance of the overall folding of the enzyme, (ii) involvement in catalysis or (iii) substrate recognition, and (iv) a non-proteolytic function. It should be noted that two other groups of 3C-like proteinases, those of arteriviruses (Fig. 3B) and potyviruses (reviewed in Ryan & Flint, 1997 ), also have carboxyl-terminal extensions, albeit of smaller sizes. Again no specific function(s) could be attributed to these domains.
The IBV 3CLpro appears to contain structural determinants that, in reticulocyte lysates, prime this proteinase for degradation by the concerted action of ubiquitin and the 26S ATP-dependent proteinase (Tibbles et al., 1995 ). The relevance of this observation to the turnover of the coronavirus proteinase in vivo has not yet been studied. For another distantly related proteinase that carries a protein destruction signal, the EMCV 3Cpro (Lawson et al., 1999 ), a correlation between the kinetics of proteinase degradation in vitro and in vivo has been reported (Lawson et al., 1994 ). Importantly, the 3C proteinase of another picornavirus, poliovirus, was shown to be stable (Lawson et al., 1999 ). Thus, the proteinase degradation signal is a virus-specific structural feature in picornaviruses and, possibly, in coronaviruses.
Sequence comparisons have revealed that the coronavirus 3CLpro is flanked by two hydrophobic domains, HD1 and HD2 (Gorbalenya et al., 1989b ; Lee et al., 1991 ; Herold et al., 1993 ; Eleouet et al., 1995 ), that are also conserved in arteriviruses (Fig. 2). Recent data from in vitro translation experiments have shown that microsomal membranes are required for the efficient autoproteolytic processing of the 3CLpro from HD1 and HD2, most likely by assisting in the proper folding of these proteins (Tibbles et al., 1996 ; Piñón et al., 1997 ; Schiller et al., 1998 ). However, after being released from the polyprotein, the 3CLpro activity does not depend on membranes, or any other cofactor(s), at least for its proteolytic activity in vitro (Ziebuhr et al., 1995 ). It has been suggested that HD1 and HD2 may contribute to the intracellular localization of the 3CLpro itself and, possibly, of the virus replication complex in general (Gorbalenya et al., 1989b ). Recent data, obtained by using immunofluorescence and electron microscopy, strongly support this hypothesis (Heusipp et al., 1997a ; Bi et al., 1998 ; Schiller et al., 1998 ; Ziebuhr et al., 1998 ; Denison et al., 1999 ; Shi et al., 1999 ; van der Meer et al., 1999 ; Ziebuhr & Siddell, 1999 ). Specifically, it has been found that the coronavirus nucleocapsid protein, numerous replicase gene-derived proteins and newly synthesized RNA co-localize to intracellular (mainly late endosomal) membranes (van der Meer et al., 1999 ). However, there are also reports that favour a Golgi localization for the MHV replication complexes, at least in specific cell types (Bi et al., 1998 ; Shi et al., 1999 ). From the combined data, it can be concluded that coronavirus replication takes place at intracellular membranes and that a large number of non-structural, replicase gene-encoded proteins contribute to the formation and function of the coronavirus replication complex.
As outlined above, the coronavirus 3CLpro has a number of unique properties that remain poorly understood due to the lack of structural information about any of these enzymes. The currently available structures of three picornavirus 3C proteinases (Allaire et al., 1994 ; Matthews et al., 1994 ; Mosimann et al., 1997 ) and the results of inhibitor analyses (Tibbles et al., 1996 ; Ziebuhr et al., 1997 ) support the classification of the coronavirus 3CLpro as a two--barrel-fold protein. However, they are of limited use in understanding the unique features of these very distant relatives. Recently developed expression and purification systems (Ziebuhr et al., 1995 , 1997 ; Seybert et al., 1997 ) could provide a suitable basis for the crystallization of coronavirus 3CLpro domains and the elucidation of their structure.
Arterivirus main proteinase
The arterivirus 3C-like serine proteinase (3CLSP) was first identified by comparative sequence analysis of the ORF1a protein of EAV, the arterivirus prototype (den Boon et al., 1991 ). Subsequently, the 3CLSP domain was shown to reside in a 21 kDa cleavage product (nsp4; 204 residues in the case of EAV) derived from the central region of the ORF1a protein (Fig. 2B) (Snijder et al., 1994 ). In addition to this fully cleaved product, a number of 3CLSP-containing processing intermediates were identified in EAV-infected cells, the most abundant ones being nsp3–12, nsp3–8 and nsp3–4 (Snijder et al., 1994 ; van Dinten et al., 1996 ). The proteolytic activity of the 3CLSP was demonstrated in the recombinant vaccinia virus/T7 expression system (Snijder et al., 1996 ), in which studies to characterize this proteinase (e.g. site-directed mutagenesis) were also carried out. The 3CLSP was shown to mediate (at least) eight cleavages in the EAV replicase; five in the carboxyl-terminal half of the ORF1a protein and three in the ORF1b-encoded polypeptide (Snijder et al., 1996 ; Wassenaar et al., 1997 ; van Dinten et al., 1999 ). These cleavage sites were identified by comparison with other arterivirus sequences (Godeny et al., 1993 ; Meulenberg et al., 1993 ; Snijder et al., 1996 ; van Dinten et al., 1996 ) and, subsequently, confirmed by site-directed mutagenesis and expression studies (Snijder et al., 1996 ; Wassenaar et al., 1997 ; van Dinten et al., 1999 ). Two sites (nsp4|5 and nsp6|7) were also confirmed by direct amino-terminal sequence analysis of cleavage products derived in an alphavirus expression system (Wassenaar et al., 1997 ). Four of the known EAV 3CLSP cleavages (Fig. 2B) occur at Glu|Gly sites, three at Glu|Ser dipeptides, and one (nsp9|10) between Gln and Ser. All of these sites are conserved in the known arterivirus sequences (Wassenaar et al., 1997 ; van Dinten et al., 1999 ) although, for a number of them, Ala (and in one case Lys) is predicted to be the P1' residue (Fig. 4). In addition to the P1|P1' positions, some degree of conservation (although, less than in coronaviruses) is evident for the P2 and P4 positions (Fig. 4). This appears to suggest that the residues at these positions contribute to substrate recognition but experiments to verify this hypothesis have not yet been reported. The sizes of the processing end-products generated by the 3CLSP (nsp3–12) are constant among different arteriviruses (Table 3), which is in contrast to the virus-specific size heterogeneity of proteins carrying accessory proteolytic activities (nsp1–2 region; see below).
The arterivirus 3CLSP was the first experimentally characterized 3C-like proteinase using the His-Asp-Ser catalytic triad typical of cellular CHL proteinases. Replacements of residues in the predicted catalytic triad (His-1103, Asp-1129 and Ser-1184 in EAV), using site-directed mutagenesis, confirmed that these residues are indispensable for proteolysis (Snijder et al., 1996 ). The substitution of Asp-1129 by Glu, but not the substitution of Ser-1184 by Cys [which are the replacements identified in the catalytic centres of (some) 3C/3C-like proteinases], was partially tolerated by the EAV 3CLSP. The putative substrate-binding region of this proteinase was predicted to include the conserved Thr-1179 and His-1199 residues (Fig. 3B), whose counterparts were originally recognized in viral 3C-like cysteine proteinases (Bazan & Fletterick, 1988 ; Gorbalenya et al., 1989a ; Allaire et al., 1994 ; Matthews et al., 1994 ). These residues are assumed to determine the P1 specificity of the 3CLSP for Glu (Gln). Accordingly, the EAV 3CLSP showed a pattern of tolerance to replacements at these positions that is typical of 3C/3C-like cysteine proteinases (Snijder et al., 1996 ). Thus, the EAV 3CLSP can be considered to be the prototype of a distinct CHL proteinase subgroup that has provisionally been named ‘3C-like serine proteinases’ (Snijder et al., 1996 ). Similar proteinases were predicted to be encoded by plant sobemoviruses, luteoviruses and pea enation mottle virus, as well as animal astroviruses (Gorbalenya et al., 1988 ; Snijder et al., 1996 , and references therein).
The ability of the EAV 3CLSP to cleave specific sites in cis or in trans has not been studied in detail. It is likely that the two sites flanking the 3CLSP domain in the replicase polyproteins (nsp3|nsp4 and nsp4|nsp5) are cleaved in cis. However, the EAV 3CLSP has also been shown to be active in trans. For example, it has been found that, in the recombinant vaccinia virus/T7 expression system, an nsp4 expression product was able to cleave the nsp9|10 and nsp10|11 sites in a separately expressed, ORF1b-encoded polyprotein (van Dinten et al., 1999 ).
|
Nidovirus accessory proteinases
|
|---|
Overview
All arteriviruses and coronaviruses encode between one and three ‘accessory’ proteinases, which are very distantly related. Thus, in the course of nidovirus evolution, duplications of accessory proteinases may have occurred (Lee et al., 1991 ; den Boon et al., 1995 ; Snijder et al., 1995 ). It remains unclear whether these duplications happened in the ancestral lineage or independently (and repeatedly) in different lineages. All of the ‘accessory’ proteinases (i) recognize one or two sites that are located in the amino-terminal half of the replicase polyproteins, (far) upstream of the major conserved domains, (ii) cleave peptides that all contain at least one small residue at the scissile bond, (iii) have a catalytic dyad consisting of Cys and a downstream His (an arrangement that is typical of cellular proteinases related to papain), and (iv) may employ variants of the 
+ fold that is conserved in this class of proteinases. For the last two reasons, the nidovirus accessory proteinases are often called ‘papain-like’. Some of the conserved properties mentioned above are also reflected in the names of the accessory proteinases (Table 1). However, there are notable differences between the arterivirus and coronavirus enzymes. The catalytic Cys and His residues of arterivirus accessory proteinases are separated by rather short regions whose sizes are within a range typical of many other viral papain-like proteinases. In coronaviruses, however, these regions are almost twice as long, making the coronavirus proteinases the largest in this class of RNA virus proteins. These proteinases are separated by a region of not less than ~1000 amino acid residues. In contrast, arterivirus accessory proteinases occupy a much more amino-proximal position in the replicase polyproteins. Also, the arterivirus enzymes cleave downstream of the catalytic domain, whereas the characterized coronavirus proteinases cleave upstream. In this review article, two similar but not identical nomenclatures will be used to designate the arterivirus and coronavirus accessory proteinases. Each of these nomenclatures is specifically designed to accommodate the unique features of specific sets of proteinases.
The arteriviruses EAV, lactate dehydrogenase-elevating virus (LDV) and porcine reproductive and respiratory syndrome virus (PRRSV) encode a cassette of three adjacent proteolytic domains that are known (from amino terminus to carboxyl terminus) as papain-like cysteine proteinase 1 (PCP1), papain-like cysteine proteinase 1 (PCP1) and cysteine proteinase of nsp2 (CP2). The digits in the names stand for the non-structural proteins (nsp; numbered from amino to carboxyl terminus) in which the proteolytic domains reside. PCP1 and PCP1 may reside in one protein (nsp1; EAV) or in two separate proteins (nsp1 and nsp1; LDV and PRRSV) that precede nsp2, which itself contains CP2. The EAV PCP1 domain is an inactivated enzyme and, therefore, the cleavage between the PCP1 and PCP1 domains does not occur, resulting in an amino-terminal cleavage product (nsp1) that contains both PCP1 domains. In contrast, due to the activity of PCP1 and PCP1, the equivalent LDV and PRRSV proteins are cleaved into nsp1 and nsp1. In both PCP1 and PCP1, but not in CP2, the catalytic Cys (Fig. 6B, C) is immediately followed by a conserved aromatic residue. This sequence signature is a hallmark of cellular papain-like proteinases and was initially used to characterize viral proteinases as being ‘papain-like’. Therefore, in the paper describing the identification of CP2 (Snijder et al., 1995 ), the enzyme was distinguished from PCP1 and PCP1. Subsequently, however, a distant variant of the papain-like fold was identified in a human ubiquitin carboxyl-terminal hydrolase (Johnston et al., 1997 ) in which, as in the arterivirus CP2 (Fig. 6C), the conserved aromatic residue is replaced by Gly. Hence, the original reasons for discriminating between CP2 and PCP1/PCP1 no longer exist. CP2 may adopt a papain-like fold and, if this is confirmed, its name should be modified accordingly.
Each of the three groups of arterivirus cysteine proteinases has a characteristic set of conserved residues that are scarcely overlapping (Fig. 6B, C). For example, the comparison of the most closely related pair of arterivirus cysteine proteinase groups (PCP1/PCP1) revealed that only one residue, the catalytic His, is absolutely conserved (Fig. 6B, C). (This number increases to four residues if the defective EAV PCP1 is not taken into account). All arterivirus accessory proteinases, with the exception of the inactivated EAV PCP1, process (or are predicted to process) a single, downstream cleavage site (Fig. 2).
The coronaviruses MHV, HCoV and TGEV encode two accessory proteinases (Fig. 2). They are called coronavirus papain-like proteinases 1 and 2, PL1pro and PL2pro, respectively. IBV encodes only one accessory proteinase. It is called coronavirus papain-like proteinase, PLpro (Fig. 2). Despite the relationship implied in the names, only a marginal similarity between the coronaviral and prototypic cellular proteinases is evident in an alignment based on the comparison of (predicted) secondary structures (Herold et al., 1999 ; Fig. 6A). However, a statistically reliable, albeit local, primary structure similarity was detected between coronavirus PLpro domains and the leader proteinase (Lpro) of foot-and-mouth disease virus (FMDV), a picornavirus (Gorbalenya et al., 1991 ; A. E. Gorbalenya, unpublished data). This similarity was used to identify a papain-like catalytic centre in the FMDV Lpro, which was subsequently confirmed by mutagenesis experiments (reviewed in Ryan & Flint, 1997 ) and X-ray crystallography (Guarné et al., 1998 ). Obviously, this relationship suggests that FMDV Lpro and coronavirus PLpro domains have evolved under a similar selective pressure that separates them from other RNA virus papain-like proteinases including those of arteriviruses (for a list of these proteinases see Gorbalenya & Snijder, 1996 ). It is conceivable that common phenotypic features exist that are conserved in Lpro and PLpro domains. It is also worth noting that the functional separation of coronavirus PLpro domains and 3CLpro into accessory and main proteinases also applies to the FMDV Lpro and 3Cpro, respectively. Thus, it is justified to treat the FMDV Lpro as the prototypic viral proteinase for coronavirus PLpro domains (Table 1).
Comparative sequence analyses of the coronavirus accessory proteinases do not provide definitive support for the clustering of the PL1pro and PL2pro domains into two groups (as their names appear to imply) or for the association of the IBV PLpro with one of these groups (A. E. Gorbalenya, unpublished observations). Indeed, the alignment of PL1pro and PL2pro (Fig. 6A) shows that only very few conserved residues are present exclusively in one of the two groups. Instead, and despite the low overall level of similarity in pairwise comparisons (13–32% identical residues), coronavirus accessory proteinases, including the IBV PLpro, have eight absolutely conserved residues (Herold et al., 1999 ). Among these are the catalytic Cys-His dyad as well as three Cys residues that are involved in the formation of a zinc-binding finger. The activity or activities of this zinc finger must be both essential and compatible with the different functions commonly found in related, but paralogous proteinases (i.e. enzymes that evolved by duplication rather than speciation). The conservation of this structural element embedded in the central region of these enzymes clearly discriminates the accessory proteinases of coronaviruses from their arterivirus counterparts.
The presence of only one PLpro domain in IBV is most intriguing and can be interpreted in different ways. For example, in the course of coronavirus evolution, a duplication of the PLpro domain might have occurred after the divergence of IBV from the rest of the coronavirus family. This scenario would imply that the PLpro duplications have occurred independently in the arterivirus and coronavirus lineages. Alternatively, a (single) duplication could have taken place in a common nidovirus progenitor. In this case, the second PLpro domain in IBV must have been deleted or diverged beyond recognition. Thus, the IBV PLpro would be orthologous with either the PL1pro or the PL2pro group (i.e. these enzymes would have diverged from a common ancestor by speciation rather than by duplication). It has been noted before (Gorbalenya et al., 1991 ) that the IBV PLpro and the PL2pro domains of the other coronaviruses are collinear in the replicase polyproteins (Fig. 2A), which favours the hypothesis that they might be orthologous proteins.
Based on a limited sequence similarity with a streptococcal cysteine proteinase, which belongs to a prokaryotic subset of papain-like proteinases, the existence of another accessory proteinase in IBV was initially postulated (named Streptococcus pneumoniae-like proteinase, SPL) (Gorbalenya et al., 1989b ). SPL was predicted in a region of pp1a/pp1ab that partly overlaps with the PLpro domain identified 2 years later (Gorbalenya et al., 1991 ; Lee et al., 1991 ). The predicted catalytic Cys and His residues of SPL are not conserved in other coronavirus replicase polyproteins and, furthermore, the relevant Cys residue immediately follows the catalytic His residue of PLpro. To our knowledge, such an overlapping organization of the active sites of two different enzymes has not been observed elsewhere and, clearly, this raises doubts about the correct identification of SPL and PLpro. Given the conservation of PLpro in all coronaviruses and the solid experimental support for its existence, it is safe to assume that the SPL domain is not functional.
Processing of the amino-proximal region of the coronavirus replicase polyproteins by accessory papain-like cysteine proteinases
The first data on the processing of coronavirus replicase polyproteins were obtained by translation of genomic MHV RNA in rabbit reticulocyte lysates. In pulse–chase experiments, a 28 kDa protein, p28, which is initially synthesized as part of a larger precursor protein, was identified (Denison & Perlman, 1986 ). This protein was also detected in virus-infected cells (Denison & Perlman, 1987 ). The genesis of p28 has been analysed in detail. First, in vitro translation experiments, combined with peptide mapping, were used to show that p28 represents the amino-terminal polypeptide (Soe et al., 1987 ). Second, it was shown that the proteolytic activity involved is virus-encoded and maps to the predicted PL1pro domain (Baker et al., 1989 , 1993 ; Gorbalenya et al., 1991 ). Third, the scissile bond that is cleaved to release the carboxyl terminus of p28 was found to be located between Gly-247 and Val-248 (Dong & Baker, 1994 ; Hughes et al., 1995 ).
In subsequent studies, immunoprecipitation experiments with region-specific antisera revealed that additional proteolytic cleavages within the amino-proximal region of the MHV-A59 replicase polyproteins generate polypeptides of 65, 50, 240 and 290 kDa (p65, p50, p240 and p290) (Denison et al., 1992
TOP
Introduction
‘Main’ and...
