Foot-and-Mouth Disease Virus Leader Proteinase

The leader proteinase (Lpro) of foot-and-mouth disease virus frees itself from the nascent polyprotein, cleaving between its own C terminus and the N terminus of VP4 at the sequence Lys-Leu-Lys-↓-Gly-Ala-Gly. Subsequently, the Lpro impairs protein synthesis from capped mRNAs in the infected cell by processing a host protein, eukaryotic initiation factor 4GI, at the sequence Asn-Leu-Gly-↓-Arg-Thr-Thr. A rabbit reticulocyte lysate system was used to examine the substrate specificity of Lpro and the relationship of the two cleavage reactions. We show that Lpro requires a basic residue at one side of the scissile bond to carry out efficient self-processing. This reaction is abrogated when leucine and lysine prior to the cleavage site are substituted by serine and glutamine, respectively. However, the cleavage of eIF4GI is unaffected by the inhibition of self-processing. Removal of the 18-amino acid C-terminal extension of Lpro slowed eIF4GI cleavage; replacement of the C-terminal extension by unrelated amino acid sequences further delayed this cleavage. Surprisingly, wild-type Lpro and the C-terminal variants all processed the polyprotein cleavage site in an intermolecular reaction at the same rate. However, when the polyprotein cleavage site was part of the same polypeptide chain as the wild-type Lbpro, the rate of processing was much more rapid. These experiments strongly suggest that self-processing is an intramolecular reaction.

Specific proteolysis is an essential component of the life cycle of many viruses. Virally encoded proteinases are required not only to process viral protein precursors into the mature proteins but also to cleave host proteins to modulate the physiology of the infected cell (1,2). A classic example is the inhibition of host protein synthesis from capped cellular mRNA that occurs during the replication of most picornaviruses (3). These viruses mediate this through the cleavage of the two homologues of eukaryotic initiation factor 4G (eIF4GI 1 and eIF4GII) by either the L pro of FMDV or by the 2A proteinase of human rhinoviruses, Coxsackie virus, and polioviruses (4 -8). As a consequence of the cleavage of the eIF4G homologues, the domain of eIF4G that binds the cap-binding protein eIF4E is severed from the domain of eIF4G which binds eIF3, so that the infected cell is unable to recruit its own capped mRNA to the 40 S ribosome (Fig. 1A (9, 10)). Translation of viral mRNA is unaffected as it initiates internally via an IRES (11,12).
The exact mechanism of the cleavage of eIF4GI and eIF4GII by the picornaviral proteinases is still unresolved. As this cleavage occurs before measurable amounts of viral proteins can be detected in vivo (13), it has been proposed that the viral proteinases induce cleavage of the eIF4G homologues by activating an as yet unidentified cellular proteinase (5,14). This idea was supported by the inability to imitate in vitro the efficiency of eIF4GI cleavage using purified recombinant proteinases (15)(16)(17). However, we have recently shown that eIF4GI can be efficiently cleaved in RRLs by nascently translated proteinases at concentrations close to those calculated in vivo (18).
The L pro of FMDV, which carries out the cleavage of the eIF4G homologues, is a papain-like cysteine proteinase (19,20); it is the first protein encoded on the FMDV polyprotein ( Fig. 1, B and C). L pro frees itself by cleavage between its own C terminus and the N terminus of VP4. As the initiation of protein synthesis on the FMDV RNA can occur at one of two AUG codons lying 84 nucleotides apart, two forms of the L pro are synthesized (designated Lab pro and Lb pro ). The reason for this is not clear, as both forms appear to have the same enzymatic properties (21). All work described here was carried out with the Lb pro form.
Recognition and cleavage of the sites on the polyprotein and on the eIF4G homologues present a number of questions. Cleavage at the junction of Lb pro and VP4 occurs at Lys-Leu-Lys 201 -2-Gly 202 -Ala-Gly, whereas that on eIF4GI takes place at the sequence Asn-Leu-Gly 634 -2-Arg 635 -Thr-Thr (13). The sequence of Lb pro cleavage on eIF4GII has not yet been determined, but it would appear by comparison to be between Asn-Phe-Gly 685 -2-Arg 686 -Gln-Thr (22). To achieve this unusual specificity, the Lb pro has evolved to be able to bind basic residues at P1 or P1Ј, while maintaining the requirement of papain-like enzymes for a hydrophobic residue at P2 (20,23). Nevertheless, the exact determinants of specificity are still not clear.
A further open question is the mechanism of self-processing (20,21,24). X-ray structural data provided evidence for an intermolecular mechanism, as the CTE of one Lb pro molecule was bound by the active site of the neighboring molecule (20). However, modeling studies using this structure suggested that a CTE would be able to reach into the active site of the same molecule, hinting that an intramolecular reaction was possible. Finally, it is not known whether self-processing is a prerequisite for cleavage of the eIF4G homologues.
In this paper, we examine these questions by expressing various mutants of Lb pro in RRLs and investigating the kinetics of cleavage of the polyprotein and eIF4GI.

EXPERIMENTAL PROCEDURES
Plasmids-The plasmid pet8cFMDV Lb pro (encoding the mature Lb pro that consists of amino acids 29 -201 of the FMDV polyprotein) contains the FMDV nucleotides 892-1413 of the FMDV O1 k cDNA (25) followed by two stop codons, cloned into the NcoI and BamHI restriction sites of the T7 polymerase expression vector pET8c (Novagen). The plasmid pet8cFMDV Lb pro VP4VP2 (encoding the mature Lb pro , all 85 amino acids of VP4 and 78 amino acids of VP2) contains the FMDV nucleotides 892-1896 followed by two stop codons similarly cloned into pet8c (13). In pet8cFMDV Lb pro C51A VP4VP2, the active site cysteine 51 is replaced by an alanine.
The plasmid pet8c ⌬Lb pro VP was constructed by using PCR amplification to delete the amino acids 30 -75 of Lb pro in pet8cFMDV Lb pro VP4VP2 to give pet8cFMDV ⌬Lb pro VP4VP2. To introduce a fragment extending the number of VP2 amino acids encoded to 140, the required fragment was amplified from the original FMDV cDNA subclone p735 (25). Two stop codons followed by a BamHI site were also encoded by the 3Ј PCR primer. This sequence was then introduced into pet8cFMDV ⌬Lb pro VP4VP2 as an RsrII BamHI fragment to give pet8c ⌬Lb pro VP. Following linearization with BamHI, transcription of this DNA gives an RNA encoding an inactive L pro lacking 46 amino acids at the N terminus followed by all 85 amino acids of VP4 and 140 amino acids of VP2.
To construct pCITE Lb pro VP4VP2, the Lb pro VP4VP2 expression block was excised from pet8cFMDV Lb pro VP4VP2 using the XbaI site of pET8c and the BamHI site at the 3Ј end of the expression block and subcloned into pBluescript. The Lb pro VP4VP2 was then introduced as an NcoI PstI fragment into pCITE1 which had been cleaved by NcoI and PstI. In this construction, translation initiates at an AUG codon three codons upstream from the first Lb pro codon. On linearization with SalI and transcription with T7 polymerase, an RNA is produced containing the encephalomyocarditis virus IRES which subsequently encodes the Lb pro with three extra amino acids at the N terminus, all 85 amino acids of VP4 and 23 amino acids of VP2.
In Vitro Mutagenesis-Restriction sites for Bpu10I and SacI were introduced into pCITE Lb pro VP4VP2 at the positions indicated in Fig.  2A using standard methods of PCR mutagenesis; the mutations do not change the amino acid sequence. Replacement of the 38-base pair Bpu10I SacI fragment with synthetic oligonucleotides enables the amino acid substitutions indicated in Table I to be introduced at the L pro VP4 junction. Lb pro C-terminal variants were generated in pet8cLb pro by replacing the 47-base pair C-terminal BsiWI BamHI fragment with the required synthetic oligonucleotides as described previously (23). All mutations were confirmed by sequencing.
In Vitro Transcription and Translation-Plasmids were linearized with the indicated restriction enzyme and transcribed as described (18). Electrophoresis and Immunoblotting-The PAGE system of Dasso and Jackson (26) was used for separation of translation products (gels contained 15% acrylamide) and for monitoring the state of eIF4GI (gels contained 6% acrylamide). Translation products were detected by fluorography; the state of eIF4GI was determined by immunoblotting using the anti-eIF4GI peptide 7 antiserum (27) as described (18).
Quantification of Protein Synthesis in RRLs-Quantification of protein synthesis using an Instant Imager (Canberra Packard) was as described previously, except that calculations were adjusted for the use of [ 35 S]methionine alone (18). Briefly, radioactivity in a particular band in the dried polyacrylamide gel was counted. By taking into account the counting efficiency (0.8%), the specific activity of the methionine in the assay (163 dpm/fmol), the number of methionines in Lb pro , and the molecular weight of Lb pro (19.8 kDa), the amount of protein present in a particular band can be estimated.

RESULTS
What Is the Relationship between Self-processing of Lb pro and eIF4GI Cleavage?-To investigate whether self-processing of the Lb pro from the growing polypeptide chain was a prerequisite for the cleavage of eIF4GI, it was necessary to produce mutant enzymes that could not process themselves from the growing polypeptide chain but that still possessed an intact active site and a correctly folded structure. Therefore, we introduced specific mutations into the cleavage site between Lb pro and VP4VP2. To simplify this, we noted that restriction sites for Bpu10I and SacI could be introduced before and after the site of Lb pro cleavage without affecting the amino acid sequence ( Fig. 2A). However, the previously employed transcription vector pet8c also contained a Bpu10I site, making it necessary to move the Lb pro VP4VP2 expression cassette into pCITE1 which has itself no restriction sites for Bpu10I and SacI. The ensuing plasmid was designated pCITE Lb pro VP4VP2 ( Fig. 2A).
To examine Lb pro self-processing and eIF4GI cleavage using pCITE Lb pro VP4VP2, a standard translation reaction was performed ( Fig. 2, B and C, left panels). The cleavage products Lb pro and VP4VP2 are visible on this fluorogram after 8 min. Lb pro encoded by pCITE Lb pro VP4VP2 has four methionines, whereas VP4VP2 has only two; thus, the intensity of the Lb pro band is about twice that of VP4VP2. The high efficiency of self-processing is indicated by the low amount of the uncleaved precursor (Lb pro VP4VP2). Table I shows the extent of selfprocessing at 12 min.
The cleavage of the endogenous eIF4GI in the RRL under these conditions occurs rapidly (Fig. 2C, left panel); 50% cleavage was observed between 4 and 8 min. The concentration of Lb pro at 8 min was determined to be 15 pg/l by counting the band in the Instant Imager. These results are almost identical to those previously reported, using pet8c Lb pro VP4VP2 driven from an uncapped RNA that did not contain an IRES (18).
The mutations indicated in Table I were then introduced at the Lb pro VP4 cleavage site using synthetic oligonucleotides. First, the P1 lysine was replaced by glutamine (Lb pro VP4VP2-(K201Q)), as the Lb pro three-dimensional structure (20) had revealed that the P1 lysine residue is bound within a deep groove consisting of three glutamate residues. Their aliphatic side chains bind the lysine side chain, whereas the amino group of the lysine interacts with the carboxyl groups of the glutamates. We reasoned that the uncharged glutamine residue would lack the electrostatic interaction and that the shorter side chain would further cause the amide group to clash with the glutamate side chains. Indeed, the pattern of proteins synthesized from this RNA is different from that of the wild-type RNA (Fig. 2B, middle panel). After 12 min, self-processing drops from over 85% in the wild-type to 37.5% in the mutant protein (Table I). Thus, although the self-processing reaction of Lb pro is impaired by this mutation, it is not completely inhibited.
The activity of this mutant protein on the cleavage of endogenous eIF4GI in the RRL was then examined by immunoblotting (Fig. 2C, middle panel). In contrast to the effect on selfprocessing, no effect on the processing of eIF4GI was observed. Reduction of the amount of RNA in the RRL also failed to reveal any differences in the rate of eIF4GI processing between FIG. 2. Effect of mutations at the Lb pro VP4 junction on self-processing and eIF4GI cleavage. The structure of the expression block Lb pro VP4VP2 is shown in A. Restriction enzyme sites mentioned in the text are shown. The sites Bpu10I and SacI were introduced by site-directed mutagenesis without changing the amino acid sequence. The arrowhead depicts the cleavage site of Lb pro . Oligonucleotide cassettes were then used to mutate the amino acids shown in bold. RRLs were incubated with or without the indicated mRNAs (10 ng/l; transcribed in vitro following linearization with SalI) as described under "Experimental Procedures"; protein synthesis was terminated at the times given by placing the samples on ice followed by the addition of unlabeled methionine and cysteine to 2 mM and Laemmli sample buffer. Aliquots were analyzed on 15% polyacrylamide gels followed by fluorography to detect the synthesis of Lb pro VP4VP2 (B) and on 6% polyacrylamide gels followed by immunoblotting for the status of eIF4GI (C). Fluorographs were exposed for 15 h for Lb pro VP4VP2 and Lb pro VP4VP2(L200S, K201Q) and 20 h for Lb pro VP4VP2-(K201Q) to enable self-processing at 8 min to be seen. The positions of uncleaved Lb pro VP4VP2 and the cleavage products Lb pro and VP4VP2 are marked in B; C, intact eIF4GI and the N-terminal cleavage product cp N are marked. Protein standards (in kDa) are indicated in both panels.

TABLE I Effect of amino acid substitutions at the Lb pro VP4
junction on self-processing Substituted amino acids are shown in bold. % self-processing was determined as the amount of Lb pro as a percentage of the total amount of Lb pro and Lb pro VP4VP2 synthesized. the wild-type and the mutant protein (data not shown).
As the K201Q mutant protein still possessed some self-processing activity, we wished to examine whether a complete inhibition of self-processing would eliminate the mature Lb pro from the reaction and thus affect the kinetics of eIF4GI processing. Therefore, we substituted the P2 leucine with serine in addition to the P1 lysine to glutamine substitution; the encoded mutant protein was designated Lb pro VP4VP2(L200S,K201Q). As both the Lb pro and eIF4GI cleavage sites have leucine at the P2 position (13) and the Lb pro structure shows a deep hydrophobic pocket for the binding of the P2 residue, we reasoned that a P2 serine residue would not be accepted and that its presence in addition to the P1 mutation would inhibit any self-processing activity. Fig. 2B (right panel) shows that this is indeed the case. After 12 min, only the uncleaved product was visible. Remarkably, however, no effect on the cleavage of eIF4GI was observed (Fig. 2C, right panel); the kinetics of cleavage are almost identical to those of the wild-type enzyme.
The above changes in the amino acid sequence at the cleavage junction clearly affected the self-processing reaction but not the cleavage of eIF4GI by Lb pro . To produce a second mutant protein that was handicapped in self-processing, we chose to replace the P1 lysine with glycine (Lb pro VP4VP2-(K201G); Table I) and then examined the effect on eIF4GI cleavage. Fig. 3A (compare left and center panels) shows that the absence of a basic group before the scissile bond reduced self-processing to about the level seen with the K201Q mutant protein (Table I). Once again, however, no reduction in the activity of the Lb pro on eIF4GI was observed (Fig. 3B, left and  center panels).
Finally, to confirm the hypothesis that the presence of a basic residue at either P1 or P1Ј was sufficient to allow cleavage by the Lb pro , we used the oligonucleotide cassette to introduce an arginine residue into the K201G mutant protein at the P1Ј position (i.e. to replace residue glycine 1 of VP4 with arginine) to give the mutant protein Lb pro VP4VP2(K201G, G202R) ( Table I). As can be seen in Fig. 3A (right panel), the presence of P1Ј arginine in the K201G,G202R mutant protein restored the cleavage to almost wild-type levels (compare Fig. 3A, left and right panels and Table I). Once again, no effect on the cleavage of eIF4GI could be observed (Fig. 3B, right panel).
It might be argued that the enzyme is recognizing the newly introduced Arg 202 as the P1 residue, with cleavage occurring between Arg 202 and Ala 203 . However, we believe this to be unlikely as this would imply that Leu 200 would be the P3 residue and Gly 201 the P2 residue. As the three-dimensional structure of Lb pro shows that a basic residue is preferred at P3 and a hydrophobic residue at P2, it is improbable that leucine and glycine could provide sufficient binding energy at these positions.
In summary, these results support the predictions of the three-dimensional structure about the substrate specificity of Lb pro and demonstrate clearly that Lb pro which is still covalently connected to the capsid proteins can efficiently cleave eIF4GI. Thus, self-processing of the Lb pro is not a prerequisite for the efficient cleavage of eIF4GI.
Role of the C Terminus on the Activity of the Lb pro -The above experiments confirmed the high efficiency of eIF4GI cleavage in RRLs by Lb pro , regardless of whether self-processing had occurred or not. What are the features of the Lb pro that enable it to cleave eIF4GI so efficiently? One unique feature of Lb pro , which is not found in papain, is the presence of the 18-amino acid CTE in Lb pro . We wished to examine whether this feature was involved in some way in eIF4GI cleavage.
To investigate this, we used a second construction pet8cLb pro that contains a synthetic stop codon after the Lys 201 residue, so that a mature Lb pro is expressed without the need for selfprocessing (Fig. 4A). The kinetics of eIF4GI cleavage are the same as those with pet8c Lb pro VP4VP2 (18). C-terminal deletions were introduced into pet8cLb pro by replacing the wildtype BsiWI BamHI fragment with synthetic oligonucleotides. Initially, two deletions in Lb pro were made; the first (designated Lb pro -6; Table II) lacked the six most C-terminal amino acids, i.e. just those which had been found in the active site of the neighboring molecule in the crystal structure. In the second (Lb pro -18), the entire 18 amino acids of the CTE were removed. RNAs were prepared from both deletion mutants and were used to drive protein synthesis in RRLs; synthesis of Lb pro (Fig.  4B) and cleavage of eIF4GI (Fig. 4C) were analyzed as before. The mutant protein Lb pro -6 cleaved eIF4GI with kinetics sim -FIG. 3. A basic residue before or after the cleavage site is essential for efficient self-processing. The indicated mRNAs (10 ng/l; transcribed in vitro following linearization with SalI) were used to program RRLs. Analysis of protein synthesis (A) and the state of eIF4GI (B) was as in Fig. 2. Fluorographs were exposed for15hforLb pro VP4VP2andLb pro VP4VP2-(K201G,G202R) and 20 h for Lb pro VP4VP2-(K201G) to enable self-processing at 8 min to be seen. ilar to that of the wild-type enzyme (Fig. 4, B and C, left and middle panels); however, the mutant protein Lb pro -18 had a reduced ability to cleave eIF4GI, cleavage was still incomplete after 12 min (Fig. 4, B and C, left and right panels). This was despite the fact that about twice as much Lb pro -18 had been synthesized than in the wild-type reaction. Fig. 4B shows that the deletion Lb pro -6 migrates more slowly than the wild-type Lb pro . This is reminiscent of an observation of Sangar et al. (28) who demonstrated that prolonged incubation (i.e. longer than 1 h) of Lb pro in RRLs led to a modification of the protein which caused it to migrate more slowly on SDS-PAGE. This modified form appeared to be deleted at its C terminus because it could also be generated by carboxypeptidase A digestion (28). Thus, removal of a small number of amino acids decreases the mobility of Lb pro on SDS-PAGE. We believe that the missing 6 amino acids of the Lb pro -6 deletion are responsible for the reduced mobility of this protein.
The above experiment suggested that the presence of the CTE was required for full enzymatic activity. This notion was further strengthened by a serendipitous observation made during the synthesis of the above deletion mutants. A variant plasmid was obtained in which an incorrect, truncated oligo-nucleotide was inserted between the BsiWI site and the BamHI site at the 3Ј end of the Lb pro ( Fig. 4A and Table II). As a consequence, the reading frame is shifted from amino acid 183 onwards and there are only 8 amino acids before the BamHI site. Thus, translation of an RNA transcribed from a template linearized with BamHI produces a variant encoding a C-terminal extension of 8 amino acids, all differing from the wild-type (Table II). This variant was designated Lb pro Ϫ 11*, as it lacks 11 amino acids compared with the wild-type Lb pro ; the asterisk indicates the aberrant amino acid sequence of the CTE. In addition, RNA encoding a second variant was produced by linearizing the same plasmid with HindIII, which cleaves 481 nucleotides downstream of the BamHI site; due to the position of the first stop codon encountered, the CTE in this case has 28 amino acids, 9 longer than the wild-type. Once again, they are of different sequence (Table II), so that the variant was designated Lb pro ϩ 9*. Although the mutant sequences have a relatively high alanine content and are somewhat hydrophobic, they are not otherwise unusual.
Nevertheless, examination of these variants in the RRL system (Fig. 5) showed that both Lb pro Ϫ 11* and Lb pro ϩ 9* had an impaired ability to cleave eIF4GI. With the Lb pro Ϫ 11* variant, about 90% eIF4GI cleavage is observed after 30 min of incubation, even though after 4 and 8 min similar protein concentrations to those of the wild-type experiment shown in Fig. 4B are present. Processing with the variant Lb pro ϩ 9* (Fig. 5, right panels) was reduced even more dramatically, as 50% eIF4GI cleavage did not occur until 30 min after cleavage, even though the RNA concentration had been adjusted to ensure that twice the amount of protein was synthesized in this experiment compared with the wild-type Lb pro in Fig. 4.
The Lb pro C-terminal Variants Are Not Affected in Their Ability to Process a Polyprotein Substrate-In the above exper-  iments, the activity of the Lb pro C-terminal variants was monitored only on the endogenous eIF4GI present in the RRL extract. We wished to investigate whether these variants were also impaired in their ability to recognize the polyprotein cleavage site. To generate a suitable substrate, the construction ⌬Lb pro VP was prepared (Fig. 6A). Following linearization with BamHI, transcription generates an RNA encoding an inactive Lb pro with a 46-amino acid deletion, all 85 amino acids of VP4, and 140 amino acids of VP2. The deletion in the Lb pro part and the extension of the VP2 part make the cleavage products of Lb pro VP4VP2 and ⌬Lb pro VP distinguishable from each other on PAGE.
The protein produced in RRL from the pet8c⌬Lb pro VP construction is of apparent molecular mass of 46 kDa (Fig. 6B, far right lane) and is proteolytically inactive (Fig. 6, B and C, far right lanes). The ability of ⌬Lb pro VP to serve as substrate was examined by simultaneously translating its RNA with the pet8cLb pro RNA (Fig. 6B, left panel). Initially, after 8 min of incubation, two products of translation were observed, the mature Lb pro and the uncleaved substrate ⌬Lb pro VP (Fig. 6B, left  panel). Subsequently, a third product of apparent molecular mass of 34 kDa becomes visible, representing the VP part of the substrate following cleavage at the junction between ⌬Lb pro and VP. 50% cleavage of ⌬Lb pro VP is achieved between 15 and 30 min. Eventually, all ⌬Lb pro VP is converted to product, as sufficient proteinase is presumably present to cleave rapidly any newly synthesized substrate so that it is no longer observed. The ⌬Lb pro part of the substrate has a molecular mass of 14.2 kDa; despite this, however, it is not resolved by this gel system.
Cleavage of eIF4GI also takes place under these conditions. Examination of the fate of the eIF4GI in the RRLs shows that over 50% is cleaved using Lb pro before 8 min (Fig. 6C, left panel; Table III); this is comparable with the experiments in which only one RNA was translated. The Lb pro concentration was estimated to be about 20 pg/l by counting the band. This agrees with previous observations in which 50% cleavage of eIF4GI was achieved when the Lb pro concentration had reached 15 pg/l (18).
This experiment was then repeated using mRNAs encoding Lb pro Ϫ 18 and Lb pro ϩ 9*. The kinetics of cleavage of ⌬Lb pro VP by these two variants are very similar to those of the wild type  Table III), indicating that the absence of the CTE or the presence of an aberrant one do not affect the ability of the variants to process the polyprotein substrate in an intermolecular reaction. In contrast, cleavage of eIF4GI by the C-terminal variants was once again delayed (Fig. 6C, middle and right panels; Table III). For the variant lacking the CTE, Lb pro Ϫ 18, 50% cleavage was achieved between 8 and 15 min at a proteinase concentration between 20 and 52 pg/l; with the variant with the aberrant C terminus, Lb pro ϩ 9*, 50% cleavage was obtained after about 45 min at a concentration of 75 pg/l.
Comparison of Intra-and Intermolecular Lb pro Cleavage-In the experiment shown in Fig. 6, Lb pro can only be cleaving ⌬Lb pro VP and eIF4GI intermolecularly. It has been a longstanding question, however, as to whether the initial cleavage of the Lb pro from the growing polyprotein (i.e. from Lb pro VP4VP2) is also an intra-or an intermolecular event (20,21,24). To investigate this, we translated the ⌬Lb pro VP RNA together with that of Lb pro VP4VP2 to see whether the cleavage kinetics of ⌬Lb pro VP or eIF4GI were changed (Fig. 7). This can be achieved because, as stated above, all the cleavage products from Lb pro VP4VP2 and ⌬Lb pro VP can be separated from one another by SDS-PAGE.
The result is shown in Fig. 7A. Four products are visible during the time course. These are the uncleaved ⌬Lb pro VP (46 kDa) and the VP cleavage products (34 kDa) (the 14.2-kDa ⌬Lb pro cleavage product is once again not resolved) as well as the mature Lb pro (19.8 kDa) and the VP4VP2 (15 kDa) cleavage product. However, the uncleaved protein Lb pro VP4VP2 is not a major product of the reaction; the uncleaved precursor is essentially only visible with the inactive leader proteinase mutant Lb pro C51A VP4VP2 (Fig. 7A, far right lane). Therefore, the processing of the Lb pro from the growing Lb pro VP4VP2 polypeptide chain must be extremely efficient, strongly implying that this reaction is an intramolecular one.
The processing of the endogenous eIF4GI by Lb pro expressed as Lb pro VP4VP2 is comparable with that observed when the Lb pro is translated as a mature protein (compare Fig. 6C with Fig. 7B). The cleavage of ⌬Lb pro VP with Lb pro expressed from Lb pro VP4VP2 is also comparable to that with Lb pro (compare Fig. 6B with Fig. 7A) and is therefore not delayed by selfprocessing. Furthermore, the self-processing of Lb pro VP4VP2 FIG. 5. The presence of heterologous sequences at the C terminus of Lb pro severely disrupts eIF4GI cleavage. The indicated mRNAs (transcribed in vitro following linearization with BamHI for Lb pro Ϫ 11* (10 ng/l) and HindIII for Lb pro ϩ 9* (20 ng/l)) were used to program RRLs. Analysis of protein synthesis (A) and the state of eIF4GI (B) was as in Fig. 2. Fluorograms were exposed for 15 h. occurs much more efficiently than the cleavage of ⌬Lb pro VP, even though the sequence of the cleavage site is the same. This strongly implies that Lb pro frees itself from the growing polypeptide chain by an intramolecular cleavage.
In summary, the reactions in order of efficiency catalyzed by the Lb pro are the intramolecular processing at the Lb pro VP4 junction and the cleavage of eIF4GI as part of the eIF4F complex followed by the much less efficient intermolecular cleavage at the Lb pro VP4 junction. DISCUSSION The experiments described here examine the relationship of the self-processing reaction of the FMDV Lb pro to the cleavage of the cellular substrate eIF4GI. By using mutagenesis to vary the cleavage site at the Lb pro VP4 junction and to produce C-terminal variants, factors affecting the rate of both reactions could be ascertained. As the polyprotein and eIF4GI substrates examined are those that are cleaved during replication of the virus in vivo, these findings are directly relevant to the biological activity of the Lb pro .
We began by investigating whether the inhibition of selfprocessing affected the rate of eIF4GI cleavage. For this, we mutated the Lb pro to prevent self-processing without affecting either the catalytic residues or the structure of the enzyme. Initially, mutations were introduced at the P1 position (Table  I). The substitution of the P1 lysine with either glycine or glutamine impaired self-processing but did not inhibit it completely. Replacement of the P1Ј glycine with arginine (thus mimicking the P1Ј residue in the eIF4GI cleavage site) in the mutant protein containing glycine at P1 restored self-processing to wild-type levels, thus confirming the unusual requirement of this enzyme for a basic residue at either the P1 or P1Ј positions.
Complete inhibition of the self-processing reaction was obtained by the introduction of a second mutation, namely leucine to serine at P2, in the P1 lysine to glutamine mutant protein. This stresses the importance of the presence of a hydrophobic residue at P2, a property that Lb pro shares with most other papain-like enzymes (30). In all of the above self-processing mutant proteins, however, a reduction in the rate or extent of eIF4GI cleavage was not observed. Thus, cleavage at the Lb pro VP4 junction is not an essential prerequisite for cleavage of eIF4GI. This implies that the enzyme is active as part of the growing polypeptide chain and that it may be active while the polypeptide chain is still bound to the ribosome.
Why is the cleavage of eIF4GI so efficient, despite the lack of self-processing? To begin, it can be assumed that the substrate eIF4GI is in the optimal conformation, as the binding of eIF4E to eIF4GI stimulates Lb pro cleavage of eIF4GI (31), and most of the eIF4GI in RRLs is complexed to eIF4E (29). For the enzyme, the C-terminal deletion experiments indicate a role for the C terminus. Thus, the mutant protein lacking the entire CTE cleaved eIF4GI at a significantly reduced rate; however, a much more drastic reduction in the reaction rate could be seen with the variants possessing an aberrant CTE. Nevertheless, all C-terminal variants recognized the Lb pro VP4 cleavage site in the intermolecular cleavage at the same rate. Thus, the CTE is required for efficient intermolecular cleavage of eIF4GI but not for that on the polyprotein substrate.
Three explanations for this observation appear plausible. First, the C terminus may be involved in a direct interaction with eIF4GI, perhaps fitting specifically into a pocket, or even eIF4E; removal of the CTE would prevent this interaction. FIG. 6. Intermolecular cleavage of the L pro VP4 junction by Lb pro wildtype and C-terminal variants. The structure of the expression block for the intermolecular substrate ⌬Lb pro VP is shown in A. Restriction enzyme sites mentioned in the text are shown. RNA was synthesized from ⌬Lb pro VP after cleavage with BamHI. RRLs were programmed simultaneously with the indicated mRNAs (10 ng/l each) and incubated for the indicated times. Analysis of protein synthesis (B) and the state of eIF4GI (C) was as in Fig. 2. The positions of uncleaved ⌬Lb pro VP, the cleavage product VP, and the L pro variants are marked in B; C, intact eIF4GI and the cleavage product cp N are marked. However, it is not clear from this theory why the presence of aberrant CTEs would have a further effect on eIF4GI processing than the variants lacking a CTE. Second, the CTE could stabilize the Lb pro in a conformation that favors eIF4GI cleavage but that is not required for intermolecular polyprotein processing. The lack of the CTE would obviate the stabilizing effect; however, the presence of the aberrant CTE might destabilize this conformation or even prevent it from being adopted. The third possibility would require that binding of the Lb pro CTE to eIF4GI induces a conformational change in this protein that exposes the cleavage site. In this case, proteolysis would be the rate-limiting step; in the CTE deletion mutants, the ratelimiting step would be the exposure of the active site in the absence of the CTE. It is interesting to note that, compared with papain, the Lb pro achieves its active conformation without a pro-domain and without activation at low pH (32). Indeed the experiments with the CTE variants appear to suggest a much more important role for the C terminus, a region of Lb pro that has no equivalent in papain (20).
Finally, cleavage reactions of the Lb pro on intra-and intermolecular polyprotein substrates were examined simultaneously. Thus, RNAs were translated so that the rate at which both intra-and intermolecular self-processing occurred could be measured (Fig. 7). These experiments were compared with one in which intramolecular self-processing was not required because the Lb pro contained a stop codon at its C terminus (Fig.  6). No difference in the rates of cleavage of eIF4GI or the polyprotein intermolecular substrate were observed. In contrast, the kinetics of cleavage of the Lb pro VP4 junction were different, depending on whether the enzyme and substrate were part of the same polypeptide chain or not. The cleavage of the substrate when part of the same chain was much more rapid, suggesting that this reaction is an intramolecular one, with the C terminus of Lb pro folding into the active site of the same molecule, as proposed by modeling studies based on the crystal structure (20). This also implies that the ⌬Lb pro VP substrate adopts a conformation competent for the intramolecular reaction and must unfold so that it can be processed in trans.
In summary, we have shown that self-processing is not a prerequisite for cleavage of eIFGI by Lb pro and have established the hierarchy of the Lb pro cleavage reactions. The first two are the extremely efficient intramolecular self-processing and the intermolecular cleavage of eIF4GI followed by the much less efficient intermolecular cleavage of the polyprotein sequence. FIG. 7. Comparison of intermolecular and intramolecular Lb pro cleavage. RRLs were programmed simultaneously with the indicated mRNAs (10 ng/l each) and incubated for the times shown. As a control for the mobility of unprocessed Lb pro VP4VP2, the mRNA encoding the inactive Lb pro C51A VP4VP2 was also translated simultaneously with ⌬Lb pro VP (far right lane). Analysis of protein synthesis (A) and the state of eIF4GI (B) was as in Fig. 2. A, the positions of the uncleaved intermolecular substrate ⌬Lb pro VP and the cleavage product VP are marked as are the positions of intramolecular cleavage products Lb pro and VP4VP2. B, intact eIF4GI and the cleavage product cp N are marked.