N-Terminal Protease of Pestiviruses: Identification of Putative Catalytic Residues by Site-Directed Mutagenesis

Abstract

Pestiviruses are the only members of the Flaviviridae that encode a nonstructural protease at the N terminus of their polyproteins. This N-terminal protease (N^pro) cleaves itself off of the nascent polyprotein autocatalytically and thereby generates the N terminus of the adjacent viral capsid protein C. In previous reports, sequence similarities between N^pro and the catalytic residues of papain-like cysteine proteases were put forward. To test this hypothesis, substitutions of cysteine and histidine residues within N^pro were carried out by site-directed mutagenesis. Translation of the mutagenized N^pro-C proteins in cell-free lysates confirmed that only the predicted Cys₆₉ was an essential amino acid for proteolysis, not His₁₃₀. Further essential residues were identified with His₄₉ and Glu₂₂. While it remains speculative whether Glu₂₂-His₄₉-Cys₆₉ actually build a catalytic triad, these results invalidate the assumption that N^pro is a papain-like cysteine protease.

Pestiviruses are small enveloped RNA viruses which constitute one genus within the Flaviviridae (23). Currently three pestivirus species are recognized, namely bovine viral diarrhea virus (BVDV), classical swine fever virus (CSFV), and border disease virus of sheep. The genome of pestiviruses is a positive-stranded RNA, usually with a size of about 12,300 nucleotides, which encodes a single polyprotein of almost 4,000 amino acids. Currently, 12 mature pestivirus proteins (N^pro, C, E^rns, E1, E2, p7, NS2, NS3, NS4A, NS4B, NS5A, and NS5B) have been identified as products of polyprotein processing, which occurs co- and posttranslationally due to virus- and host-cell-encoded proteases (20). All members of the virus family, including the genera Flavivirus, Hepacivirus, and Pestivirus have similar genome organizations (16). A divergent evolution of flaviviruses on one side and hepaciviruses and pestiviruses on the other is indicated by a fundamental difference between the two sides concerning translation initiation. The former require a cap structure at the 5′ ends of their genomic RNAs (16), while the latter are able to undergo cap-independent translation initiation due to an internal ribosomal entry site in their genomes (12, 17, 22).

A puzzling pestiviral protein is the N-terminal protease N^pro, which has no counterpart among the other members of the Flaviviridae. While it was initially suggested that N^pro represents the structural capsid protein (4, 24), it has since been demonstrated that N^pro is a nonstructural protein (18, 21). The only known function of N^pro is its proteolytic activity which leads to cleavage between Cys₁₆₈ and Ser₁₆₉ in the polyprotein; Ser₁₆₉ has been determined to be the N terminus of the capsid protein (18). Proteases at the N termini of polyproteins are common in a variety of positive-stranded RNA viruses (6). All known members of the so-called “viral accessory leader proteases” (6) are cysteine proteases which resemble the active-site organization of the type protease papain (4, 7). N^pro was also suggested to belong to this protease family; the catalytic site was proposed to consist of Cys₆₉ and His₁₃₀ (4, 18).

The N^pro-C cleavage occurs efficiently in procaryotic and eucaryotic cells and also during translation in cell-free lysates (18, 24). To study the autoproteolytic properties of N^pro, wheat germ lysate (Promega), together with metabolic [³⁵S]methionine labeling, was chosen for in vitro translation experiments. Rabbit reticulocyte lysate was not used, because endogenous proteins in the 14- to 20-kDa range (e.g., globin) interfered with the separation of N^pro and C during sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE).

M13mp18 DNA served as a vector for a transcription cassette which consisted of an SP6 promoter sequence, the 5′ nontranslated region (NTR) (56 nucleotides) of Sindbis virus and the cDNA encoding amino acids 1 to 246 (N^pro-C) of CSFV Alfort (Fig. 1). The 5′ NTR of Sindbis virus has been shown previously to efficiently promote translation initiation (3). Transcripts were either synthesized from replicative form M13 DNA linearized with XhoI as a template or from fragments generated by PCR with M13 universal and reverse primers which flank the polylinker sequence. SP6 transcription was performed with the MaxiScript kit (Ambion) without the addition of a cap analog. After in vitro translation, the transcripts gave rise to three protein species, N^pro-C, N^pro, and C. In neither in vitro system did cleavage of N^pro-C proceed to completion; pulse-chase experiments indicated that the remaining N^pro-C molecules were not subject to further cleavage (data not shown). Preliminary data suggested that cleavage occurs during synthesis, probably in a monomolecular fashion (cis cleavage), of the nascent polypeptide chain (not shown). In a pilot experiment, we wished to determine whether N^pro was sensitive to protease inhibitors. Prior to the translation reaction a number of reagents which specifically inhibit a variety of serine, cysteine, and aspartic acid proteases (antipain dihydrochloride, 1 mg/ml; aprotinin, 0.5 mg/ml; 4-(2-aminoethyl)benzenesulfonyl fluoride, 2 mg/ml; TLCK (Nα-p-tosyl-l-lysine chloromethyl ketone), 9 mg/ml; TPCK (N-tosyl-l-phenylalanine chloromethyl ketone), 8.5 mg/ml; chymostatin, 1 mg/ml; E64, 1 mg/ml; leupeptin, 0.1 mg/ml; pepstatin, 0.5 mg/ml; phosphoramidon, 1 mg/ml; bestatin, 0.25 mg/ml) were added to the wheat germ lysate. Although they were used at high concentrations, none of the substances showed significant inhibition of N^pro-C cleavage (not shown). Thus, the use of inhibitors did not offer any clue about the type of protease.

FIG. 1.

(Upper line) Structure of the cDNA construct used for in vitro transcription and translation. The cassette including the SP6 promoter, the NTR of Sindbis virus, and the N^pro-C gene was cloned into M13mp18, which served for site-directed mutagenesis. For transcription, the sequence was amplified with M13 reverse and forward primers, and the PCR product was directly transcribed with SP6 RNA polymerase. (Lower line) Cleavage products of the N^pro-C polyprotein. As the precise C terminus of C has not been established, it was tentatively located at amino acid 246 of the CSFV polyprotein.

We began to introduce specific mutations into the N^pro gene by using oligonucleotide-directed mutagenesis according to the modified Kunkel method (8). To test for the suggested catalytic dyad (Cys₆₉, His₁₃₀) the amino acid substitutions Cys₆₉Ser, Cys₆₉Ala, and His₁₃₀Leu were generated, and the respective mutant N^pro-C polyproteins were assayed for cleavage by in vitro translation and SDS-PAGE. While the His₁₃₀Leu mutation displayed a normal rate of cleavage (Fig. 2b), the Cys₆₉Ser and Cys₆₉Ala substitutions had deleterious effects on the cleavage of N^pro-C (Fig. 2a). To support that Cys₆₉ was actually involved in catalysis and not merely a determinant of the secondary structure of N^pro, the other four conserved cysteine residues (Cys₁₁₂, Cys₁₃₄, Cys₁₃₈, and Cys₁₆₁) were replaced by serine and then alanine residues (Fig. 2a). None of these substitutions affected N^pro-C cleavage. The catalytic mechanism of cysteine proteases is often based on the formation of a thiolate-imidazolium ion pair with cysteine as the nucleophile and histidine as the general base (19). Because the His₁₃₀Leu substitution had no effect, the remaining three conserved histidine residues (His₄₀, His₄₉, and His₉₉) were also changed to leucine residues; the substitution His₄₉Leu abrogated the cleavage, which suggested a participation of this residue in catalysis (Fig. 2b). The involvement of His₄₉ in proteolysis was proposed earlier (24), but no additional catalytic residues were revealed in this study.

FIG. 2.

Identification of putative catalytic residues Cys₆₉ and His₄₉ by site-directed mutagenesis of N^pro. Wheat germ lysate was used for translation of uncapped transcripts; metabolic labeling was done with [³⁵S]methionine. (a) The conserved cysteine residues of N^pro were replaced with serine and then alanine residues, and the effect on cleavage of N^pro-C was assayed in wheat germ lysate. Only replacement of Cys₆₉ affected cleavage. wt, unsubstituted N^pro-C. (b) The four conserved histidine residues of N^pro were replaced with leucine residues. Only His₄₉Leu resulted in inactivation of N^pro. wt, unsubstituted N^pro-C.

The identification of the putative catalytic amino acid Cys₆₉ supported the working hypothesis that N^pro very likely is a cysteine protease. However, the position of the catalytic His₄₉ was incompatible with the catalytic domain of papain-like cysteine proteases. In papain-like cysteine proteases, the catalytic histidine is positioned C terminally of the catalytic cysteine (Table 1). The inverted orientation of these residues, as determined for N^pro, is found in a group of cysteine proteases which, according to a proposed novel taxonomy for proteases, encompass clan CB (13). The type protease of one family (C3) of clan CB is the picornavirus 3C protease, also known as picornain (14). All cysteine proteases of this clan are encoded by viruses and are related to the chymotrypsin-like serine proteases (14). Comparison of the amino acid sequences of N^pro and picornain resulted in weak yet significant homology of the amino acid sequence surrounding Cys₆₉ of N^pro and the active-site Cys₁₅₉ of encephalomyocarditis virus 3C protease (Fig. 3). Close inspection of the residual sequences revealed no further homologies. Predictions for poliovirus 3C protease indicated the presence of a third catalytic residue, an acidic amino acid which is positioned between the catalytic histidine and cysteine residues (5) (Table 1). A conserved acidic amino acid at an equivalent position is not found in N^pro. The only conserved acidic residue in the respective region of N^pro is Asp₆₈, which could be replaced by Asn₆₈ without affecting cleavage (not shown). Interestingly, the acidic residue as part of the catalytic triad of picornain was recently put into question. Crystal structure analysis of 3C protease of hepatitis A virus revealed a catalytic dyad consisting of histidine and cysteine residues rather than a triad including an aspartic or glutamic acid residue (1).

TABLE 1.

Comparison of the array of catalytic amino acids of serine and cysteine proteases with the putative catalytic triad of pestivirus N^pro

Protease (type; database code)	Clana	Catalytic residuesb
Papain-like cysteine (papain; PAPA_CARPA)	CA	C25H159N175
Chymotrypsin-like serine proteases (bovine chymotrypsin; CTRA_BOVIN)	SA	H57D102gdS195g
Chymotrypsin-like cysteine proteases (poliovirus 3C; POLG_POL1M)	CB	H40E71gdC147g
Subtilisin-like serine proteases (subtilisin; SUBE_BACSU)	SB	D32H62gtS215xa/s
Npro (CSFV; HCVCGSA)		E22H49gdC69rs

^aClassification into clans is according to Rawlings and Barrett (14, 15). 

^bNumbers in subscript refer to the positions of the respective catalytic residues (in boldface type) in the type proteases. The numbering for papain accounts for the mature protease after removal of the activation peptide (133 amino acids). Note the relatively close spacing of the putative catalytic residues in N^pro compared to the other proteases. 

FIG. 3.

Local alignment of amino acid sequences surrounding the catalytic cysteine residue (arrow) of N^pro and members of the chymotrypsin-like cysteine proteases (picornain). The bottom line shows the sequence of the serine protease bovine chymotrypsin. Abbreviations: EMCV, encephalomyocarditis virus (database code, POLG_EMCV); Polio, poliovirus (database code, POLH_POL1M); bov.chym., bovine chymotrypsin (database code, CTRA_BOVIN).

N^pro invariably consists of 168 amino acids in all sequenced pestivirus isolates. There are a few cases in which N^pro is encoded twice in the viral genome; due to a complex genetic duplication and recombination event, N^pro is then fused to the N terminus of the NS3 protein of cytopathogenic BVDVs (i.e., BVDV strain PE515cp [9]), and there is evidence that N^pro mediates the release of NS3 autocatalytically. In BVDV strain PE515cp, the N^pro fused to NS3 lacks the N-terminal 15 amino acids (9). This finding indicates that the N-terminal portion of N^pro is probably not involved in catalysis. To determine the exact number of dispensable N-terminal amino acids, N^pro deletion mutants that lacked amino acids 2 to 10, 2 to 15, 2 to 19, 2 to 20, 2 to 21, and 2 to 22 were generated (Fig. 4 and data not shown). While the stepwise deletions of N-terminal amino acids including N^pro 2 to 21 had no effect on proteolysis (Fig. 4b, lanes 2, 4, and 5), the deletion of amino acids 2 to 22 resulted in an inactive protease (Fig. 4b, lane 6). Replacement of Glu₂₂, a conserved residue in pestiviruses, with valine heavily affected cleavage in the context of N-terminally truncated N^pro as well as full-length N^pro (Fig. 4b, lanes 3 and 8). The substitution Glu₂₂Asp also rendered the protease inactive, yet some residual activity was observed (Fig. 4b, lane 7). While it is reasonable to assume that His₄₉ and Cys₆₉ are constituents of the active site, participation of Glu₂₂ in catalysis is uncertain. A catalytic triad consisting of Glu₂₂-His₄₉-Cys₆₉ would be unusual, and there is no reported precedence among the cysteine proteases. However, such an array is somewhat reminiscent of the catalytic triad of the subtilisin-like serine proteases (Table 1). Subtilisin-like serine proteases are widespread among procaryotes and eukaryotes and include several prominent members, e.g., proteinase K, kexin, and furin (15). Interestingly, virus-encoded subtilisin-like serine proteases have not been identified so far. On the basis of sequence homology, the ORF 47 gene product of channel catfish herpesvirus was suggested to be a subtilisin-like protease (2). This assumption is questionable, since there is no direct proof for catalysis and especially because the reported homologies do not include the catalytic residues of the subtilisin-like serine proteases. Subtilisin-like serine proteases are believed to be structurally unrelated to the chymotrypsin-like serine proteases, and there is evidence that both classes have evolved separately (15). The vast majority of subtilisin-like proteases contain an active-site serine, and there is only one precedent for a subtilisin-like protease, from Bacillus smithii, which has a cysteine residue at this position (11). Subtilisin-like proteases usually are characterized by patterns of conserved amino acids surrounding the catalytic residues. N^pro, however, has no apparent sequence homology with the subtilisin proteases.

FIG. 4.

(a) Structures of N-terminal deletion mutants of N^pro and amino acid substitutions of Glu₂₂. (b) SDS-PAGE analysis of in vitro-translated mutants of N^pro-C, as shown in panel a. wt, unsubstituted N^pro-C.

Due to the inherent limitations of the mutagenesis approach, it could not be proven that Glu₂₂, His₄₉, and Cys₆₉ are involved in catalysis. Because of this caveat and only minute sequence homologies to other proteases, it is not possible to state whether N^pro is a member of one of the established protease families. A solution to this problem has probably to await the determination of the three-dimensional structure of N^pro.

Linked to the issue of protease classification is the question of the origin of N^pro. One obvious explanation is that pestiviruses acquired N^pro from a host cellular source. Pestiviruses have been shown to frequently integrate parts of host cellular mRNAs into their genomes (for a review, see reference 10). In this case, there should be a good chance to detect remnants of the ancestral sequence. So far, database searches using the nucleotide and amino acid sequences of N^pro have not revealed significant homology to any deposited sequence. It is also possible that pestiviruses “invented” the protease de novo by adapting a protein of a different function to their own needs. This implies the possibility that the protease domain is only one facet of a multifunctional protein, where the other one(s) remains to be elucidated.

As a first step toward understanding the function of N^pro, the application of reverse genetics, together with the infectious pestivirus clones, will provide evidence whether N^pro is essential for virus replication.