NS2B/3 proteolysis at the C-prM junction of the tick-borne encephalitis virus polyprotein is highly membrane dependent.

The replication of tick-borne encephalitis virus (TBEV), like that of all flaviviruses, is absolutely dependent on proteolytic processing. Production of the mature proteins C and prM from their common precursor requires the activity of the viral NS2B/3 protease (NS2B/3(pro)) at the C-terminus of protein C and the host signal peptidase I (SPaseI) at the N-terminus of protein prM. Recently, we have shown in cell culture that the cleavage of protein C and the subsequent production of TBEV particles can be made dependent on the activity of the foot-and-mouth disease virus 3C protease, but not on the activity of the HIV-1 protease (HIV1(pro)) (Schrauf et al., 2012). To investigate this failure, we developed an in vitro cleavage assay to assess the two cleavage reactions performed on the C-prM precursor. Accordingly, a recombinant modular NS2B/3(pro), consisting of the protease domain of NS3 linked to the core-domain of cofactor NS2B, was expressed in E. coli and purified to homogeneity. This enzyme could cleave a C-prM protein synthesised in rabbit reticulocyte lysates. However, cleavage was only specific when protein synthesis was performed in the presence of canine pancreatic microsomal membranes and required the prevention of signal peptidase I (SPaseI) activity by lengthening the h-region of the signal peptide. The presence of membranes allowed the concentration of NS2B/3(pro) used to be reduced by 10-20 fold. Substitution of the NS2B/3(pro) cleavage motif in C-prM by a HIV-1(pro) motif inhibited NS2B/3(pro) processing in the presence of microsomal membranes but allowed cleavage by HIV-1(pro) at the C-prM junction. This system shows that processing at the C-terminus of protein C by the TBEV NS2B/3(pro) is highly membrane dependent and will allow the examination of how the membrane topology of protein C affects both SPaseI and NS2B/3(pro) processing.

Introduction

Tick-borne encephalitis virus (TBEV), a member of the family of Flaviviridae in the genus Flavivirus (Lindenbach et al., 2007) is a small (∼50 nm) enveloped virus with a single stranded, positive sense RNA genome. In addition to the lipid envelope and the RNA, three structural proteins (capsid (C), membrane (M, derived from a precursor prM) and envelope (E)) are present in the virion. The genetic information in the viral RNA is expressed as a single polyprotein that meanders in and out of the endoplasmic reticulum (ER) membrane. The structural proteins are located in the amino-terminal part of the polyprotein; the rest of the polyprotein comprises the non-structural proteins NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5 (Lindenbach et al., 2007). NS2B and NS3 (designated here NS2B/3pro) comprise the viral protease that is required for polyprotein processing, along with furin and signal peptidase I (SPaseI), two host cell proteases. These cleavages occur co- and post-translationally. In addition, during translation and processing, the surface proteins prM and E are glycosylated in the lumen of the ER. The production of these structural proteins is vital as protein C is the initiator molecule for viral assembly; multiple copies of this protein encapsidate a newly synthesised RNA molecule to produce the nucleocapsid (NC). Furthermore, prM is required to ensure that protein E is manufactured in a viable conformation. Correctly processed prM and E are added to the growing particle by budding of the NC through the ER. The resulting immature particles cross the trans-Golgi network (TGN), permitting furin cleavage of prM; this reaction transforms the immature particles into infectious virions (Stadler et al., 1997).

Processing is thus a prerequisite for the assembly of the TBEV particle; not surprisingly, the two processes are closely co-ordinated (Lobigs, 1993). For instance, the C-terminus of the protein C contains an internal hydrophobic signal sequence responsible for translocating the prM protein into the lumen of the ER. On the cytosolic side of the ER membrane, protein C is cleaved off the signal sequence by NS2B/3pro (Amberg et al., 1994; Yamshchikov and Compans, 1994). This cleavage is a prerequisite to allow the host cell SPaseI cleavage at the signal sequence on the luminal side of the ER membrane to generate the N-terminus of protein prM. Initiation of virion assembly has been proposed to depend upon the timing of these cleavage events at the termini of the signal sequence separating proteins C and prM (Amberg and Rice, 1999; Lee et al., 2000; Lobigs and Lee, 2004; Stocks and Lobigs, 1998). Protein prM is in its turn a prerequisite to ensure correct synthesis and transport of protein E (Konishi and Mason, 1993; Lorenz et al., 2002).

In TBEV, investigation of the cleavage by NS2B/3pro in the C-terminal region of protein C is complicated by the presence of two potential cleavage sites (KR*G and RR*S, underlined in Fig. 2; Mandl et al., 1991). Schrauf et al. (2009) showed that, in vivo, the downstream cleavage site is used and that the upstream cleavage motif cannot be accepted when the downstream cleavage motif had been removed by mutation.

Fig. 2

(A) Variants of recombinant proteases. The 49 amino acid residue-long, central portion of NS2B was linked with the NS3pro sequence via a GASPGGSGA or GGGGSGGGAG linker, respectively. All protease constructs were C-terminally tagged with a (his)×6 tag. The arrow indicates the position of the first amino acid found in the 28 kDa band examined by mass spectroscopy. (B) Schematic diagram of the TBEV polyprotein with cleavage sites for the viral protease (black arrowhead), host signal peptidase I (blue arrowhead) and furin (diamond). (C) Variants of substrates used in this study. Schematic drawing of the C-terminal region of the capsid protein and the signal sequence for prM (not drawn to scale). Engineered mutations are shown together with the corresponding designations. The potential but unused upstream NS2B/3pro cleavage site (KRG) as well as the actual site of cleavage (RRS) are underlined. The downstream cleavage site of NS2B/3 (black), and cleavage sites of SPaseI (blue) and HIV-1 protease (red) are indicated by arrowheads. Numbers at the top refer to the amino acid positions within protein C or prM. The transmembrane region is highlighted in blue. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Schrauf et al. (2009) showed further that, in the absence of the downstream NS2B/3pro cleavage site, the C-terminus of protein C could be released from the polypeptide chain by the introduction of the foot-and-mouth disease virus (FMDV) 2A sequence. The presence of the sequence Asn-Pro-Gly-Pro at the C-terminus of this 20 amino acid protein causes the ribosome to pause, release the first protein and then continue translation of the mRNA. In contrast, upon removal of the residues Pro-Gly-Pro from the 20 amino acid sequence, the separation of C and prM was inhibited, leading to a replication deficient virus. Virion production could be rescued by the introduction of an FMDV 3Cpro cleavage site at the N-terminus of the 2A sequence and the expression of the 3Cpro either in cis from a second ORF or in trans from a TBEV replicon (Schrauf et al., 2012). However, we were unable to achieve success in such a system with the HIV-1pro (Schrauf et al., unpublished).

To enable a more predictable extension of this system to other viral proteinases, we decided to first establish an in vitro biochemical system that would allow the investigation of the parameters required to ensure accurate proteolysis at the C-prM junction of TBEV. In this report, we describe such a system and use it to analyse the membrane dependency of cleavage by the TBEV NS2/3pro.

Materials and methods

Oligonucleotides and plasmids

Oligonucleotide sequences can be found in Supplementary Table 1. Constructs encoding TBEV European subtype strain Neudörfl (EMBL: U27495) were derived from pBR-TBEV-ΔME-E-EGFPo (kind gift of Dr. P. Schlick). Plasmids coding for the co-factor and the protease were constructed in two steps. First, the hydrophilic cofactor region of NS2B (Fig. 1A), corresponding to amino acids 1404–1453 of the TBEV polyprotein (UniProt: P14336) was amplified by primers NS2B-f and NS2B-r-3′ linker. These primers, introducing an NcoI restriction site at the 5′ end and the first half of the synthetic linker containing an XmaI recognition sequence at the 3′ end, were used to clone the fragment into pCite-A1, designated as pCite-A1-NS2B. Then, a fragment encoding the 190 N-terminal residues of NS3 (Fig. 1B), the protease domain (UniProt: P14336, residues 1490–1679) was amplified with primers 5′-Linker-NS3-F and NS3-R-6xHis. Thereby, the second half of the synthetic linker containing an XmaI recognition site was introduced at the 5′-end of the protease region. Furthermore, six histidine codons and a BamHI site were added to the 3′-end. These enzymes were used to ligate the protease domain to pCite-A1-NS2B, finally named pCite-A1-NS2B/3, coding for the tethered NS2B/3 protease (Fig. 2A).

Fig. 1

Alignment of tick-borne encephalitis virus (TBEV) and West Nile virus (WNV) amino acid sequences. (A) Complete sequences of NS2B. A conserved region involved in the formation of the active site in WNV is underlined. (B) N-terminal portions of NS3. Catalytic triad residues (His 54, Asp 78 and Ser 138) are indicated in blue. The parts of NS2B and NS3 used for constructing the recombinant TBEV NS2B/3 are green and red, respectively. Identical, very similar and similar residues are indicated by asterisks, colons and dots, respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

NS2B/3 S135A with alanine substitutions at S135 in the catalytic triad of NS3 were made by site-directed mutagenesis with primers NS3 S135A-f and NS3 S135A-r.

To generate the stabilised proteases with the new linker GGGGSGGGG, a cleavage site for BglII was introduced downstream of the linker with primers NS3-BglI-f and NS3-BglI-r. Cassettes coding for the desired mutations K90A and R93A of NS2B as well as the new linker were designed and introduced by use of the BglII and XhoI. The mutation G to A in the linker region of construction NS2B/3 R93A occurred serendipitously during mutagenesis. As the ensuing protease proved to be stable and active, it was used in all subsequent experiments.

For expression in E. coli, sequences coding for the desired tethered proteases were cloned from pCite-A1-NS2B/3 into the expression vector pET11d by use of NcoI and BamHI.

To generate the plasmids coding for the substrates (Fig. 2B and C), plasmid pTNd/5 which contains the 5′-part of the genome of TBEV European subtype strain Neudörfl, was used as template (kind gift of Dr. R. Kofler). Using primers C-prM-f and C-prM-r, we produced PCR fragments encoding the capsid protein, transmembrane region and pre-membrane protein prM. Thus, C-prM corresponds to residue 1–280 of the TBEV polyprotein. The plasmid was named pCite-C-prM.

To generate pCite-C-2AΔ3-prM and pCite-C-HIV-prM, we used plasmid pBR-C-2AΔ3-prM and pBR-C-HIV-prM, both kind gifts of Dr. S. Schrauf. By use of AgeI and MluI, we isolated the desired fragment and cloned it into pCite-C-prM using the same cleavage sites.

In order to inhibit signal peptidase cleavage, nine leucine residues were introduced into the transmembrane regions of substrates by use of primers TM-9L-f and TM-9L-r.

In vitro transcription and translation

All pCite plasmids were linearised with BamHI. In vitro transcription with T7 RNA polymerase and in vitro translation were as described in Schlick et al. (Schlick and Skern, 2002). In vitro translation reactions (typically 20 μl) contained 70% RRL (Promega), 20 μCi of 35S-methionine (1000 Ci/mmol, American Research Company) and amino acids (except methionine) at 20 μM. After pre-incubation for 1 min at 30 °C, translation was started by addition of RNA. The reaction was stopped at designated time points by immediate transfer to ice and addition of unlabelled methionine and cysteine to a final concentration of 2 mM. Where indicated, translation reactions were supplemented with 4–8% pancreatic microsomal membranes (Promega).

Protease assay

For self-cleavage reactions, translation products with or without microsomal membranes were incubated in assay buffer (20 mM Tris/HCl, pH 7, 50 mM NaCl) at 30 °C and quenched by addition of an SDS–PAGE loading buffer to a final concentration of 2% SDS.

Trans-cleavage reactions were performed at 30 °C in a final volume of 10 μl containing substrate and purified NS2B/3pro at the indicated final concentrations and assay buffer. After the indicated times, the reaction was stopped by addition of SDS sample buffer.

Electrophoresis and immunoblotting

The polyacrylamide gel electrophoresis system of Dasso and Jackson (1989), containing 15% polyacrylamide, was used to separate proteins. 35S-containing proteins were detected by fluorography.

Expression and purification

The plasmids pET 11d-NS2B/3, pET 11d-NS2B/3 S135A, pEt11d-NS2B/3 R93A were used for high-level, inducible expression of hexa-histidine-tagged recombinant proteins. Cultures of E. coli strain BL21(DE3)pLysS transformed with the expression plasmids were grown in 10 ml of LB medium containing 34 μg/ml chloramphenicol and ampicillin at 37 °C overnight. This overnight culture was diluted 1:100 in fresh LB medium containing chloramphenicol and ampicillin at the concentrations stated above and incubated at 37 °C, 225 rpm until an OD600 of 0.5–0.6 had been reached. The expression of the recombinant proteins was induced by addition of isopropyl-β-d-thiogalactopyranose (IPTG) to a final concentration of 0.1 mM. Cultures were incubated for additional 4 h, and cells were harvested by centrifugation. Cell pellets were resuspended in 30 ml resuspension buffer containing 50 mM Tris–HCl, 50 mM NaCl, pH 8 and lysed with a Bandelin Sonoplus sonicator (4 cycles, 40%, 30 s and once continuous for 30 s) and centrifuged at 15,000 rpm for 30 min at 4 °C. The supernatant was loaded onto a 5 ml HiTrap chelating column (GE Healthcare) loaded with nickel and pre-equilibrated with lysis buffer. The protein was eluted with a linear gradient of 0–500 mM imidazole in lysis buffer. Fractions containing NS2B-NS3 proteins, determined by 12.5% SDS–PAGE, were pooled and concentrated using Amicon Ultra-4 Centrifugal filter Device 10 kDa cut-off (Millipore). Subsequently, the protein was loaded on to a HiLoad TM 26/60 Superdex 75 column (GE Healthcare) equilibrated with resuspension buffer and run according to the manufacturer's instructions. Fractions were again analysed by SDS-PAGE. Aliquots were stored until use at −80 °C.

Expression and purification of the HIV-1pro

Penta-stabilised HIV-1pro Q7K, L33I, L63I C67 C95A was a kind gift of Dr. J. Tözsér (Louis et al., 1999). Briefly, the protease was expressed in E. coli and refolded by dialysis against 25 mM formic acid for 2 h and then against 100 mM Na-acetate-trihydrate, 1 mM DTT, 1 mM EDTA, 10% glycerol and 0,03% Triton X-100, pH 5 five times for 2 h. The final concentration of the enzyme was 0.4 U/μl. This buffer was also used in HIV-1pro trans-cleavage experiments.

Results

Generation of an active TBEV NS2B/3pro

The proteolytic domain of the multifunctional flaviviral NS3 protein serine protease resides in the amino-terminal domain. The protein folds into two β-barrels with the hydrophilic core being stabilised by the co-factor NS2B which also plays a role in substrate binding (Erbel et al., 2006). Without the NS2B domain, the NS3 protease is proteolytically inactive. It has been shown by Leung et al. (2001) that the cofactor activity of the NS2B core region (residue 35–48) is comparable to that of the entire NS2B sequence. This knowledge allowed single chain NS2B/3pro to be generated by genetically fusing the NS2B core region to the NS3 protease domain by a short linker region. Indeed, NS2B/3pro of several flaviviruses have been expressed in this way (Bessaud et al., 2005; Leung et al., 2001; Nall et al., 2004; Shiryaev et al., 2006; Yusof et al., 2000). To generate a similar construction for TBEV, the pertinent regions of NS2B and NS3 in the TBEV polyprotein were analysed by sequence alignment with WNV (Fig. 1) and related flaviviruses (data not shown). Residues 46–94 of TBEV NS2B (in green in Fig. 1A) and residues 1–190 (in red in Fig. 1B) of NS3 were identified by similarity to the regions employed in WNV (Nall et al., 2004) as the central hydrophilic co-factor region of NS2B and the protease domain of NS3, respectively. These regions were amplified by PCR as indicated in Section 2.1 and linked by a DNA sequence encoding the flexible linker indicated in Fig. 2A. To prevent the loss of the NS2B activating peptide through autocatalytic cleavage, we attempted to ensure that no cleavage site for the NS3 protease was present in the flexible linker. A C-terminal hexa-histidine tag was introduced at the C-terminus for purification. A similar construct containing the inactivating mutation S135A was also prepared. Both wild-type and the inactivated mutant could be produced either by in vitro transcription and translation assay in rabbit reticulocyte lysates (RRLs) or by expression in E. coli.

To test whether the NS2B/3 hybrid protein was indeed stable, the NS2B/3pro and NS2B/3pro S135A RNAs were translated in RRL in the presence of 35S methionine for 30 min at 30 °C followed by addition of unlabelled methionine and incubation at 30 °C for the indicated times. Synthesised proteins were separated by SDS-PAGE and detected by fluorography. Both NS2B/3pro and NS2B/3pro S135A RNAs were translated into a protein of 33 kDa, slightly higher than the calculated Mr of 28 kDa. During the incubation of the active protease, the 33 kDa band slowly disappeared overnight and was replaced by bands of apparent molecular weight 28 kDa and 12 kDa. However, as the smaller band was actually running with the buffer front of the gel, its molecular weight cannot be given with any confidence. In contrast, the 33 kDa band of the inactive NS2B/3pro S135A remained stable (Fig. 3), suggesting that autolysis of the active protease was responsible for the conversion of the 33 kDa species into the two smaller ones.

Fig. 3

Analysis of autocatalytic activity of TBEV NS2B/3pro. In vitro transcribed RNAs encoding the indicated proteins were translated in RRLs in the presence of 35S-methionine for 30 min at 30 °C. Following the addition of unlabelled methionine, samples were incubated for the indicated times in buffer containing 50 mM NaCl, 50 mM Tris/HCl, pH 7. Proteins were separated by SDS-PAGE and 35S-containing proteins detected by fluorography.

Similar results were obtained using the bacterially expressed NS2B/3pro (data not shown). Mass spectroscopy of the purified 28 kDa band from SDS-PAGE revealed that it lacked the N-terminal 47 amino acids of the NS2B/3 active proteinase (see Fig. 2). Thus, despite the lack of a corresponding cleavage site, the active NS2B/3pro was still able to perform autolysis between residues K90 and E91 (Figs. 2 and 3), giving rise to fragments of calculated Mr 22 and 6 kDa, respectively. To prevent such autolysis, we constructed two modified NS2B/3 proteins in which either one (R93A) or two (K90A R93A) basic residues at the C-terminus of 2B were replaced with alanine (Fig. 2A). In addition, the linker sequences were also modified to more closely resemble that used in other flaviviral NS2B/3 constructs (Fig. 2A; Leung et al., 2001; Nall et al., 2004). RNA from the resulting constructs NS2B/3pro R93A and NS2B/3pro K90A R93A was translated in RRLs; both synthesised proteins were stable over the time-course of the experiment (Fig. 3). NS2B/3pro R93A was selected for use in all subsequent experiments.

In vitro processing of the structural protein precursor C-prM

The precursor (C-prM) of C and prM is connected by a short hydrophobic domain that spans the membrane of the ER (Fig. 4A). C is cleaved into its mature form by the NS2B/3pro at the cytosolic side whereas the SPaseI cleaves prM to its mature form in the ER (Fig. 4A as well as 2B and 2C). SPaseI only gains access to its cleavage site once NS2B/3pro has cleaved at the C-terminus of protein C. To investigate this cleavage reaction in vitro, we incubated purified recombinant NS2B/3pro R93A with C-prM synthesised in RRL. Control experiments were performed with buffer or with recombinant inactive NS2B/3 S135Apro (final concentration 1 μg/μl). Proteins were separated by SDS-PAGE and detected by fluorography.

Fig. 4

Processing of TBEV polyprotein subfragments containing the C and prM proteins by purified recombinant NS2B/3pro. (A) Topology of the C-prM-E region. (B) C-prM. (C) C-2AΔ3-prM. In vitro translated substrates were incubated with buffer or NS2B/3pro for the indicated times. Proteins were separated by SDS-PAGE and 35S-containing proteins detected by fluorography.

A single specific band corresponding to unprocessed C-prM running at 35 kDa was detected when the substrate was incubated with buffer or inactive NS2B/3pro (Fig. 4B, lanes 2–4 and data not shown). When incubated with active NS2B/3pro at a final concentration of 1 μg/μl (Fig. 4B, lanes 5–8), the substrate was converted into two bands running at 27 kDa, corresponding to prM* (prM plus amino-terminal transmembrane region), and about 10 kDa, corresponding to protein C. We tested several assay conditions to optimise the cleavage efficiency but were not able to find conditions in which amounts of protein less than 1 μg/μl led to cleavage. Thus, the cleavage efficiency of the purified protease remained rather low in comparison with recombinant proteases from related viruses (Bera et al., 2007; Chappell et al., 2008; Shiryaev et al., 2007).

NS2B/3pro cleavage of prM in vitro does not depend on the presence of its cleavage site

The canonical NS2B/3pro recognition site in the C-terminal region of protein C of TBEV was shown previously to be at R-R*S (Schrauf et al., 2008; Fig. 2C). To test whether the cleavage of C-prM by NS2B/3pro was dependent on the presence of its cleavage site, we generated a C-prM mutant (C-2AΔ3-prM) in which the downstream NS2B/3pro cleavage site was replaced by a version of the 2A sequence of foot and mouth disease virus (FMDV) lacking the characteristic Pro-Gly-Pro motif at its C-terminus (see Fig. 2C). This form is unable to allow the ribosomes to skip the formation of a peptide bond so that protein C cannot be released from the polyprotein (Donnelly et al., 2001). A TBEV mutant bearing this truncated 2A (C-2AΔ3-prM) was replication deficient, as protein C could not be processed by the NS2B/3pro (Schrauf et al., 2009). This provides further evidence that the upstream cleavage site is not used. Nevertheless, in this in vitro system described here, we observed, in the absence of the downstream cleavage site at the C-terminus of protein C, cleavage of the C-2AΔ3-prM with purified NS2B/3pro (Fig. 4C, lane 3). The most likely interpretation is that the upstream cleavage was substituting for the downstream site.

It should be noted that, despite repeated attempts, we were not able to detect the cleavage product for protein C. We interpret this failure to mean that the modified protein C containing all or part of the 2A protein of FMDV is unstable in RRLs. We have observed this phenomena previously when expressing other viral proteins in this system (Sousa et al., 2006).

Signal peptidase cleavage of C-prM in the presence of microsomal membranes occurs in the absence of NS2B/3pro cleavage

We postulated that the specificity of the NS2/3pro reaction could be increased by the addition of canine pancreatic microsomal membranes (subsequently referred to as membranes) to allow the signal peptide that separates the two proteins to insert into a membrane. Upon translation of C-prM in the presence of the above-mentioned membranes for 30 min, a decrease in its electrophoretic mobility could clearly be seen after a further 60 min of incubation at 30 °C (Fig. 5A, compare lane 2 with lane 3), suggesting that N-glycosylation had occurred at residue Asp 144 of prM (Fig. 5A, lane 3). After 120 min, however, the amount of the C-prM species had decreased; instead, a new band had appeared at 28 kDa. Based on the molecular weight, we suggest that this corresponds to glycosylated prM lacking the trans-membrane region (Fig. 5A, lane 4).

Fig. 5

Effect on proteolytic processing of the synthesis of C-prM subfragments in the presence of microsomal membranes. (A) C-prM. Translation products were incubated at 30 °C for the indicated times. (B) C-9L-prM was synthesised in the presence or absence of microsomal membranes and then incubated in buffer or with NS2B/3pro at the indicated concentrations and times at 30 °C. Proteins were separated by SDS-PAGE and 35S-containing proteins detected by fluorography.

The most likely explanation for the above result is that the membrane bound signal peptidase (SPaseI) has processed its luminal cleavage site in the absence of the viral protease. According to previous studies, however, the SPaseI cleavage should not occur until the NS2B/3pro has performed processing at the C-terminus of protein C (Lobigs, 1993; Lobigs and Lee, 2004; Stocks and Lobigs, 1995). As we wished primarily to investigate NS2B/3pro cleavage, we looked for an approach to inhibit the SPaseI reaction whilst permitting the cleavage by NS2B/3pro. To this end, we first used a known inhibitor of SPaseI, N-methoxysuccinyl-Ala-Ala-Pro-Val-chloromethyl ketone (Lundin et al., 2008; Nilsson et al., 2002). Indeed, this compound, at concentrations of 750 μM or higher, did inhibit the proposed SPaseI cleavage reaction (data not shown). However, this concentration also inhibited the activity of NS2B/3pro (data not shown), leading to the abandonment of this approach.

We therefore took an alternative approach to eliminate SPaseI cleavage. Signal peptides have been shown to possess a common structure, namely a short, positively charged amino-terminal n-region, a central hydrophobic h-region (blue in Fig. 2B and C) and a polar, mostly uncharged c-region containing the signal peptidase cleavage site. In addition to certain other differences, one appreciable difference between cleaved signal peptides and non-cleaved signal anchors is the length of the h-region. Consequently, it is possible to convert a signal peptide into a signal anchor by lengthening its h-region (Nilsson et al., 1994). To this end, we introduced a stretch of nine leucine residues into the h-region of the signal peptide between C and prM (Fig. 2C). RNA from this construct, termed C-9L-prM, was then incubated in RRLs in the presence of microsomal membranes and the stability of the translated protein examined. Fig. 5B shows that the protein remained intact in both the absence (lanes 2 and 3) and presence of membranes (lanes 7 and 8), indicating that extension of the h-region conveyed resistance to SPaseI cleavage.

It is worth noting that, in this experiment, the glycosylation was complete during the translation phase of the reaction (compare Fig. 5A, lane 2 with Fig. 5B, lane 7) and that there was hardly any detectable difference between the protein in the glycosylated and non-glycosylated form (compare left and right panels of Fig. 5B).

NS2B/3pro shows higher activity and specificity in the presence of microsomal membranes

Having generated a form of C-prM that was resistant to cleavage by SPaseI, we could now examine its ability to act as substrate for NS2B/3pro. We therefore incubated C-9L-prM that had been translated in the absence or presence of microsomal membranes with different concentrations of NS2B/3pro. Interestingly, the concentration of NS2B/3pro could be reduced by 10–20-fold (compare Fig. 5B, lanes 4 and 5 with lanes 9 and 10) when microsomal membranes were used. NS2B/3pro at a final concentration of 500 ng/μl (Fig. 5B, lane 4) was required to cleave C-9L-prM in the absence of membranes, whereas 25 ng/μl (Fig. 5B, lane 10) were sufficient to obtain cleavage of this substrate when membranes were used.

To confirm that the NS2B/3pro enzyme was indeed recognising the correct cleavage site, we designed a C-9L-prM mutant in which the downstream NS2B/3pro cleavage site was replaced by a cleavage site of the HIV-1pro (C-HIV-9L-prM). Fig. 6 (lanes 4 and 5) shows that the protein is not cleaved by NS2B/3pro in the presence of membranes, even though the upstream cleavage site was still available. However, once again cleavage was observed in the absence of membranes (Fig. 6, lane 8).

Fig. 6

Effect of the replacement of the NS2B/3 cleavage site in C-9L-prM with that of HIV-1pro on NS2B/3pro processing. C-HIV-9L-prM was translated in the presence (lanes 2–5) or absence (lanes 6–8) of microsomal membranes and incubated with buffer or NS2B/3pro at the indicated final concentrations and for the indicated times at 30 °C. Proteins were separated by SDS-PAGE and 35S-containing proteins detected by fluorography.

Thus, we concluded from the above set of experiments that the NS2B/3pro was indeed recognising its true cleavage site but only in the presence of membranes.

The cleavage between C and prM can be performed by HIV-1pro

The presence of the HIV-1pro cleavage site in C-HIV-9L-M allowed us to investigate whether C-HIV-9L-prM was cleaved by HIV-1pro in the presence or absence of membranes. As a control, we used the construction C-9L-prM lacking the HIV-1pro cleavage site. In the absence of microsomal membranes, both constructs were processed by the HIV-1pro (Fig. 7A, lanes 2–5), showing a similar pattern of two cleavage products of about 28 kDa and 20 kDa. The 28 kDa band presumably arises through cleavage at the C-prM junction to give prM*, whereas the 20 kDa band indicates cleavage at a position within protein C or prM. To verify that HIV-1pro is responsible for the cleavages, we added increasing concentrations of pepstatin, a known HIV-1pro inhibitor (Wondrak et al., 1991). Fig. 7B indeed shows a concentration dependent inhibition of cleavage, indicating that the cleavage is indeed due to HIV-1pro processing.

Fig. 7

Cleavage at the TBEV C-prM junction by HIV-1pro. (A) HIV-1pro cleavage in the absence (lanes 2–5) and presence (lanes 6–8) of microsomal membranes. C-9L-prM (upper panel) and C-HIV-9L-prM (lower panel) were translated with or without microsomal membranes and incubated with purified HIV-1pro (lanes 4 and 7, 5.7 mU/μl; lanes 5 and 8, 2.8 mU/μl) for 30 min at 30 °C. (B) C-HIV-9L-prM was translated in the absence of microsomal membranes and incubated for 60 min at 37 °C with HIV-1pro in the absence or presence of pepstatin at the indicated concentrations. Proteins were separated by SDS-PAGE and 35S-containing proteins detected by fluorography.

Next we investigated HIV-1pro cleavage in the presence of microsomal membranes (Fig. 7A, lanes 6–8). Cleavage of the C-9L-prM construct was no longer observed (upper panel); however, with the C-HIV-9L-prM construct, the same cleavage products were observed as in the absence of membranes (lower panel). Thus, the presence of the membranes eliminated any cleavage of the C-prM construct lacking the HIV-1pro site, suggesting that the protein was correctly inserted in the membrane. In addition, the cleavage products obtained with the C-HIV-9L-prM indicate that correct processing at the HIV-1pro sequence was occurring, but suggested that a second cleavage site was present within the prM protein that was still accessible even in the presence of membranes. Nevertheless, this experiment shows clearly that it is possible, at least in vitro, to make the cleavage between proteins C and prM dependent on HIV-1pro.

Discussion

We present here a system in which the cleavage in the C-terminal region of protein C of TBEV can be specifically performed by purified recombinant viral NS2B/3 protease. Such cleavage required the presence of microsomal membranes to prevent non-specific proteolysis by the NS2B/3pro as well as introduction of nine leucine residues at the C-terminus of the transmembrane segment of protein C to prevent incorrect SPaseI cleavage. Cleavage of the protein C-9L-prM in the presence of membranes was efficient and specific, requiring only 25 ng/μl to initiate cleavage (Fig. 5B, lane 10). When the NS2B/3pro cleavage site was replaced by that of the HIV-1pro cleavage site, the C-9L-prM protein was still cleaved by NS2B/3pro in the absence of microsomal membranes. However, in the presence of microsomal membranes, no cleavage with NS2B/3pro was observed, demonstrating the dependence of NS2B/3pro specificity on the presence of membranes. Instead, cleavage between C and prM could now be carried out by the HIV-1pro. However, a second cleavage was observed with the HIV-1pro both in the presence and absence of membranes. This may explain why it was not possible to make a TBEV that was dependent on cleavage with HIV-1pro as was the case for FMDV 3Cpro (Schrauf et al., 2012).

To establish this system, a number of difficulties had to be overcome. Firstly, the modular NS2B/3 protein that we constructed was capable of undergoing autocatalytic cleavage, even in the absence of a dibasic sequence in the proximity of the linker. This reaction could be prevented by the substitution of an arginine residue (R93) two amino acids upstream of the synthetic linker. This suggests that the presence of this basic residue, coupled with the proximity of the linker to the active site, allowed slow autoproteolytic cleavage to take place on a non-canonical cleavage sequence.

This recombinant NS2B/3pro was purified to homogeneity and was able to cleave specifically in the C-terminal region of protein C, provided that translation took place in the presence of microsomal membranes and that SPaseI cleavage had been prevented by extension of the trans-membrane helix. This is in contrast to experiments that examined the cleavage of West Nile virus C-prM using infected cells as a source of the active WNV NS2B/3pro (Yamshchikov and Compans, 1994) and suggests a fundamental difference in the cleavage of two viral polyproteins.

Substitution of the NS2B/3pro cleavage motif in C-prM by that of HIV-1pro allowed us to examine cleavage of C-HIV-9L-prM by the HIV-1pro in the presence of membranes. In addition to a band corresponding to prM*, indicating cleavage at the HIV-1pro sequence, an additional cleavage product with a size of about 20 kDa was observed. Where might the HIV-1pro additional HIV cleavage sequence be located? Sequence analysis via an online HIV-1pro cleavage site prediction program (Chou, 1996; Chou et al., 1993; Shen and Chou, 2008) revealed a TVIR/AEGK motif spanning amino acids 134–143 with a high probability of cleavage by HIV-1pro. Cleavage at this site in prM would result in the formation of two bands of 20 and 6 kDa, respectively, in good agreement with the observed 20 kDa band; the 6 kDa product would not be retained on the polyacrylamide gel. However, this site on the prM protein should be inside the membrane vesicles and not available for the protease. In contrast, the construction of C-9L-prM without an HIV-1 site was not cleaved by HIV-1pro in the presence of membranes. This implies that there is a difference in the orientation of the two proteins in the membrane that can only derive from the presence of the HIV-1pro cleavage site in the C-terminal region of protein C. This mis-orientation of the polyprotein containing the HIV-1pro cleavage site may explain why it was not possible to obtain virions when we tried to make the C-prM cleavage dependent on the HIV-1pro in cell culture using a bi-cistronic virus (Schrauf et al., unpublished). As the cleavage at the C-terminus was possible with the FMDV 3Cpro (Schrauf et al., 2012), the failure with the HIV-1pro must lie with this enzyme and/or with its cleavage site and not with the system per se.

This system has allowed us to demonstrate the membrane dependency of cleavage specificity of TBEV NS2B/3pro at the C-prM junction. In future, this system should also serve as a model to examine the topology of the transmembrane anchor of the TBEV protein C and the requirements for correct SPaseI processing.