Signal Peptide Peptidase Cleavage of GB Virus B Core Protein Is Required for Productive Infection in Vivo*

Chronic infection by hepatitis C virus (HCV) is a leading cause of liver disease for which better therapies are urgently needed. Because a clearer understanding of the viral life cycle may suggest novel anti-viral approaches, we studied the role of host signal peptide peptidase (SPP) in viral infection. This intramembrane protease cleaves within a C-terminal signal sequence in the viral core protein, but the molecular determinants of cleavage and whether it is required for infection in vivo are unknown. To answer these questions, we studied SPP processing in GB virus B (GBV-B) infection. GBV-B is the closest phylogenetic relative of HCV and offers an accurate surrogate model for HCV infection. We demonstrate that SPP also processes GBV-B core protein and that a serine residue in the hydrophobic region of the signal sequence (present also in HCV) is critical for efficient SPP cleavage. The small size of the serine side chain combined with its ability to form intra- and interhelical hydrogen bonds likely contributes to recognition of the signal sequence as a substrate for SPP. By introducing mutations with differing effects on SPP processing into an infectious GBV-B molecular clone, we demonstrate that SPP processing of the core protein is required for productive infection in primates. These results broaden our understanding of the mechanism and requirements for SPP cleavage and reveal a functional role in vivo for intramembrane proteolysis in host-pathogen interactions. Moreover, they identify SPP as a potential therapeutic target for reducing the impact of HCV infection.

Intramembrane-cleaving proteases are a family of enzymes that target transmembrane domain sequences of proteins and thereby modulate cellular signaling, lipid biosynthesis, and the unfolded protein response (1)(2)(3). Signal peptide peptidase (SPP) 2 is an intramembrane-cleaving protease that cleaves cer-tain signal sequences following proteolysis by signal peptidase (4,5). Studies on the biological functions associated with SPP cleavage are very limited. In higher eukaryotes, cell-based analyses have indicated that disrupting the SPP processing of HLA-E epitopes blocks their presentation on the cell surface (6,7). In the lower eukaryotes, such as Caenorhabditis elegans (8) and Drosophila melanogaster (9), inhibition of SPP activity by either RNAi or mutation impairs embryonic and larval development.
SPP is necessary for maturation of the hepatitis C virus (HCV) core protein (10,11). Core protein forms the virus capsid and is released from the viral polyprotein by signal peptidase and SPP through two coordinated cleavage events at a signal sequence that separates core from the E1 glycoprotein (4). Cleavage by SPP is essential for transfer of core from the endoplasmic reticulum (ER) membrane to lipid droplets, which are cytosolic storage organelles (10). However, because of the limited availability of animal models, it has not been possible to demonstrate that SPP cleavage of core is necessary for the production of virus progeny in vivo.
GB virus B (GBV-B) can act as a surrogate model of HCV infection, because the two viruses have identical genome organizations and share significant sequence similarity (12)(13)(14)(15)(16)(17)(18)(19). GBV-B establishes an acute infection in several species of New World monkeys including tamarins (Saguinus sp.) (12, 14, 16, 17, 20 -22), although its natural host has not been identified unequivocally. Intrahepatic injection of in vitro transcribed RNA from full-length cDNA of GBV-B gives a pattern of infection that is identical to that following inoculation with infectious material. Such a system enables assessment of GBV-B mutant or GBV-B/HCV chimeric genomes to initiate infection (18,19,23).
In this report, we demonstrate that SPP cleavage is required for maturation of GBV-B core. We characterize the molecular determinants of SPP cleavage and show that SPP processing is also required for productive viral infection in nonhuman primates. These observations reveal that host SPP has been coopted by a surrogate for HCV for functions that are essential to its life cycle, thus suggesting novel therapeutic approaches for treatment of HCV infection. The novel role played by SPP in this host-pathogen interaction, which has not been reported previously for other intramembrane-cleaving proteases, offers unique possibilities for unraveling the mechanisms and processes engaged in intramembrane proteolysis.

EXPERIMENTAL PROCEDURES
Construction of Plasmids-pSFV1-GBV-B/WT was created by extending the coding region expressed by pSFV/GB1-398 (24). pSFV/GB1-398 expresses the GBV-B core and E1 proteins and contains an epitope tag following amino acid residue 85 to enable detection of core protein (24). Extended sequences encoding E2, p13, and the N terminus of NS2 were generated by PCR amplification of pGBB (kindly provided by J. Bukh), which contains the entire cDNA of the GBV-B genome (12). Thus, pSFV1-GBV-B/WT contained nucleotides 426 -2762 of the GBV-B genome with an insert of 42 nucleotides following position 700, which encoded the epitope tag. Specific mutations in the core-E1 signal sequence were engineered into this plasmid to generate the mutants described in Table 1. To produce plasmids encoding GBV-B reference polypeptides, nucleotide sequences encoding the N-terminal 144, 146, 148, 150 and 152 amino acids of core were amplified from pSFV1-GBV-B/WT by PCR and inserted into pSP64 poly(A) (Promega). For tamarin inoculations, mutations in the core-E1 signal sequence contained in pSFV1-GBV-B/Mut4, -7, and -9 were inserted into pGBV-B/2, an infectious molecular clone of GBV-B (16), giving rise to pGBV-B/Mut4, -7, and -9.
In Vitro Transcription of RNA and Electroporation of Cells-RNA was transcribed in vitro from recombinant pSFV1 plasmids linearized with SpeI and introduced into BHK C13 or Huh-7 cells by electroporation as described previously (25,26). Where necessary, cells were incubated with 100 M (Z-LL) 2 ketone immediately following electroporation until the cells were harvested 12-14 h after electroporation (27).
Indirect Immunofluorescence-Cells were fixed in 4% paraformaldehyde and probed with CMV-018-48151 (Capricorn Products Ltd.), which is specific for the pp65 epitope tag, and an anti-calnexin antibody (Sigma). Bound antibody was visualized with appropriate secondary antibodies, and cells were imaged using a Zeiss LSM 510 Meta confocal microscope. Oil red O staining was performed as described previously (26).
In Vitro Translation of GBV-B Reference Polypeptides-Plasmids encoding GBV-B core reference polypeptides were linearized with EcoRI prior to SP6 polymerase-mediated transcription in vitro. RNA (1 g) was used to program separate translation reactions using rabbit reticulocyte lysate (Promega) according to the manufacturer's instructions.
Western Blot Analysis-Proteins from cell extracts and in vitro translation reactions were separated by SDS-PAGE, transferred to nitrocellulose membrane, and probed with anti-pp65 antibody as described previously (24). After incubation with secondary antibody, proteins were detected by enhanced chemiluminescence.
Infectivity Assay in Tamarins-Plasmids encoding the fulllength GBV-B nucleotide sequence containing specific mutations in the core-E1 signal peptide region were linearized by digestion with XhoI prior to T7 RNA polymerase-mediated transcription in vitro using the T7 MegaScript system (Ambion). The integrity and abundance of the RNA were evaluated by agarose gel electrophoresis. To assess the infectivity of transcribed RNA in tamarins (Saguinus mystax), ϳ100 g of RNA was inoculated directly into the liver, and at 7-14-day intervals post-inoculation, serum was collected and assayed for alanine aminotransferase (ALT) activity and viral RNA levels. In the latter case, RNA was isolated using a QiaAmp viral RNA extraction kit (Qiagen) and quantitated by real-time, 5Ј-exonuclease, reverse transcription-PCR (TaqMan) using a one-step procedure with the TaqMan EZ core reagent kit (PerkinElmer Life Sciences) and primers located within the GBV-B NS5A sequence as described previously (16). Animals were housed at the Southwest National Primate Center at the Southwest Foundation for Biomedical Research (San Antonio, TX) and cared for in accordance with the Guide for the Care and Use of Laboratory Animals. Experimental protocols were approved by the Institutional Animal Care and Use Committee of the Southwest Foundation for Biomedical Research.

GBV-B Core Protein Is
Processed by SPP-The signal sequences located at the C terminus of GBV-B and HCV core are substrates for cleavage by signal peptidase between amino acid residues 156/157 and 191/192, respectively (28,29). HCV core is processed further by SPP within the h-region of the signal sequence to generate a mature product, which is considered to participate in virion morphogenesis (30). SPP cuts also in an analogous position in classical swine fever virus (CSFV), a pestivirus that is related to HCV (31). To determine whether GBV-B core is a substrate for SPP, a polyprotein encompassing GBV-B core to the N-terminal region of NS2 and including an epitope tag inserted within the core sequence ( Fig. 1A) was expressed in BHK cells in the absence or presence of (Z-LL) 2 ketone, a SPP inhibitor (27). In the absence of (Z-LL) 2 ketone, a core species of about 18.5 kDa was observed, which increased in molecular weight upon treatment of cells with the compound ( Fig. 2A). We deduced that core protein detected in the presence of (Z-LL) 2 ketone represented the signal peptidasecleaved product terminating at Gly 156 , which had been trimmed in untreated cells by SPP to give a smaller species. To identify the approximate position of SPP cleavage, the electrophoretic mobility of mature GBV-B core was compared with that of a panel of GBV-B reference peptides that had been translated in vitro and differed in length by 2 amino acids (Fig. 2B). GBV-B core co-migrated with peptides extending to Val 144 and Cys 146 (Fig. 1B), placing the SPP cleavage site within the h-region of the GBV-B core-E1 signal sequence, which is consistent with the reported sites in the corresponding HCV and CSFV signal sequences (Fig. 1B).
GBV-B core localizes to the surface of lipid droplets within the cytoplasm of cells, and only small amounts of protein are detected at the ER membrane (24) (Fig. 2C). To determine whether SPP cleavage is needed for lipid droplet association, the distribution of GBV-B core was examined in cells treated with (Z-LL) 2 ketone. Under these conditions, no lipid droplet localization was observed and the protein adopted a reticular cytoplasmic distribution (Fig. 2D, compare panels i and ii). Taken together, these findings demonstrate that, analogous to its HCV counterpart, GBV-B core is cleaved by SPP and this process is required for lipid droplet localization.
Efficient SPP Cleavage Requires Ser 149 in the h-region of the GBV-B C-E1 Signal Sequence-Having established that GBV-B core was a substrate for SPP, and that the cleavage status of core could be reliably assayed by Western blot analysis and lipid droplet association, the next goal was to identify residues within the core-E1 signal sequence required for SPP proteolysis. From previous in vitro studies, it has been determined that small, helix-destabilizing amino acids within the h-region of signal sequences are critical for SPP cleavage (4,10). Therefore, we targeted Cys 146 , Ser 149 , Ala 151 , and Cys 152 within the h-region of the core-E1 signal sequence (Fig. 1B) for mutagenesis to either leucine or valine, because these residues have a high propensity to form ␣-helices in a hydrophobic environment (Table  1 and Ref. 32). The cleavage profile of mutated core proteins was examined in BHK and Huh-7 cells by Western blot analysis (Fig. 3A). Treatment of BHK cells with (Z-LL) 2 ketone was B, GBV-B core reference peptides, ending at the indicated amino acid residues, were synthesized in vitro and ran alongside a BHK cell extract expressing the GBV-B core-⌬NS2 polyprotein from pSFV1-GBV-B/WT. In A and B, extracts were examined for the presence of GBV-B core by Western blot analysis. C and D, the intracellular localization of GBV-B core was examined in the absence (C, panels i-vi; D, panels i and iii ) or presence (D, panels ii and iv) of (Z-LL) 2 ketone. Cells were stained for core (C, panels i and iv; D, panels i and ii ) and either counterstained with oil red O (C, panel ii ) or calnexin (C, panel v), and merged panels are shown on the right (C, panels iii and iv). Lipid droplets in cells examined by phase contrast (D, panels iii and iv) are indicated by arrows.
included to control for any variations in protein mobility that might result from modifying the amino acid composition of the core-E1 signal sequence (10,33). Because (Z-LL) 2 ketone is less effective at inhibiting SPP activity in Huh-7 cells, 3 the compound was not used to treat this cell type. However, as an additional indicator of SPP cleavage, the ability of the mutated core proteins to localize to lipid droplets was assessed in both cell types (Table 1). Typically, lipid droplet association was more restricted for mutants expressed in BHK as compared with Huh-7 cells (for example, compare the values for pSFV1-GBV-B/Mut1 in both cell types). This discrepancy results from the greater number of lipid droplets in Huh-7 cells (34). Nonetheless, for mutants with reduced or no lipid droplet association in BHK cells, the amount of protein attached to droplets in Huh-7 cells also was lowered, and there was a corresponding increase in diffuse intracellular staining.
Mutation of three amino acids in the h-region (Cys 146 , Ser 149 , and Ala 151 in pSFV1-GBV-B/Mut1) gave a core species that co-migrated with the protein generated in the presence of (Z-LL) 2 ketone (Fig. 3A, BHK) and abolished lipid droplet association in BHK cells (Table 1). In Huh-7 cells, there was some evidence of attachment of core to lipid droplets in 31% of cells, but most of the protein was distributed throughout the cytoplasm (Table 1). Therefore, only a minor proportion of core expressed by pSFV1-GBV-B/Mut1 retains susceptibility to cleavage by SPP. By contrast, mutation at Cys 152 gave levels of SPP-cleaved core and lipid droplet association that could not be distinguished from those for the wild-type protein (pSFV1-GBV-B/Mut2) (Fig. 3A, BHK, and Table 1). By progressively introducing additional mutations into pSFV1-GBV-B/Mut2, it was found that Cys 146 also did not contribute to SPP proteolysis (pSFV1-GBV-B/Mut3) (Fig. 3A, BHK, and Table 1), whereas there was some contribution from Ala 151 (pSFV1-GBV-B/ Mut4; Table 1), because a small amount of the signal peptidase processed species was detected (Fig. 3A, BHK). An identical profile for processed species was observed in Huh-7 cells (Fig.  3A, Huh-7). Introduction of a further mutation at Ser 149 in pSFV1-GBV-B/Mut4 abolished both SPP proteolysis and lipid droplet association (pSFV1-GBV-B/Mut5) (Fig. 3A and Table   1), suggesting that this amino acid was critical for processing. To determine the impact of mutation only at Ser 149 , we generated pSFV1-GBV-B/Mut6 in which serine was replaced with leucine, but no other mutations were present in the core-E1 signal sequence. For this mutant, neither SPP-cleaved protein nor attachment to lipid droplets could be detected in BHK cells (Fig. 3A, BHK, and Table 1). In Huh-7 cells, only a small amount of protein processed by SPP was found (Fig. 3A, Huh-7), and its association with lipid droplets was reduced to levels detected for pSFV1-GBV-B/Mut1 (Table 1). Therefore, we concluded that Ser 149 is crucial for efficient SPP cleavage in the GBV-B core-E1 signal sequence.
Characteristics of Serine That Promote SPP Cleavage-To test whether SPP cleavage is specified by the small size of the side chain on serine or its propensity to destabilize ␣-helices, we replaced it with glycine (pSFV1-GBV-B/Mut7), alanine (pSFV1-GBV-B/Mut8), methionine (pSFV1-GBV-B/Mut9), and tyrosine (pSFV1-GBV-B/Mut10). Both alanine and glycine have small side chains, in common with serine, whereas tyrosine and methionine are bulkier residues. In terms of ␣-helix propensity in an apolar environment, alanine, tyrosine, and methionine are well tolerated, but glycine destabilizes ␣-helices, because it increases conformational freedom (32). All of the mutants were generated from pSFV1-GBV-B/Mut4. The replacement of Ser 149 with either tyrosine or methionine abolished lipid droplet localization (Fig. 3B, panels vii-ix, data not shown, and Table 1) and production of SPP-cleaved core in both BHK and Huh-7 cells (Fig. 3A). Staining of cells with calnexin indicated that the core protein processed solely by signal peptidase was located on the ER membrane (Fig. 3C, panels x-xii, and data not shown). By contrast, glycine and alanine generated core that was located either on lipid droplets (Fig. 3, B and C, panels iv-vi) or the ER membrane (Fig. 3C, panels vii-ix; data not shown). Introduction of these amino acids led to mature core protein that was cleaved by SPP although to a lesser extent as compared with serine in either BHK or Huh-7 cells (Fig. 3A). Based on these results, particularly on the different characteristics shown by comparing alanine, tyrosine, and methionine, we deduce that the small size of the side chain on serine makes a greater contribution to SPP cleavage than its ability to destabilize ␣-helices. Moreover, serine gave the highest amount of SPP-processed product, and thus its polar group is likely to contribute also to cleavage efficiency.
Maturation of Core by SPP Is Essential for Productive GBV-B Infection in Vivo-To determine whether SPP cleavage is required for genome infectivity in vivo, the mutated core-E1 signal sequences present in pSFV1-GBV-B/Mut4, Mut7, and Mut9 (Table 1) were introduced into pGBV-B/2 (16), an infectious clone of GBV-B. The epitope tag present in GBV-B core in the pSFV constructs was not included in the GBV-B clones to prevent any detrimental effects on introducing additional sequences into the polyprotein. This strategy gave three clones that were predicted to allow 1) efficient SPP cleavage (pGBV-B/Mut4), 2) processing by signal peptidase but not SPP (pGBV-B/Mut9), or 3) comparable amounts of signal peptidase-and SPP-cleaved core (pGBV-B/Mut7).
Two tamarins (T16446 and T16458) were inoculated with RNA prepared from pGBV-B/Mut9, and serum was analyzed TABLE 1 Effect of mutation at the core-E1 signal sequence on lipid droplet localization of core Amino acids that differ from the wild-type sequence are underlined. LD, cells with GBV-B core localized to lipid droplets.

pSFV1-GBV-B Core-E1 signal sequence (Phe 143 -Gly 156 ) BHK LD Huh-7 LD
for GBV-B RNA and ALT activity for up to 22 weeks postinoculation (Fig. 4, A and B). Neither of the animals developed detectable viremia or gave elevated ALT levels during the monitoring period. These findings suggested that SPP cleavage of GBV-B core is required to establish a productive infection in vivo. To demonstrate that modification of the GBV-B core-E1 signal sequence did not abolish the capability of genome-length RNA to produce virus, one of the previously inoculated tamarins (T16458) was re-inoculated with RNA from pGBV-B/ Mut4 (Fig. 4C). The polyprotein encoded by this construct differed from that encoded by pGBV-B/Mut9 only at amino acid 149 where serine replaced methionine. GBV-B/Mut4 initiated a robust infection in T16458 from week 4 post-inoculation, such that by 12 weeks, virus titers had reached Ͼ10 9 genome equivalents (ge)/ml serum with an accompanying rise in ALT levels (Fig. 4C). These titers are comparable with those obtained for the wild-type infectious GBV-B clone, which gave values rang-ing from 1-3 ϫ 10 9 ge/ml for data collected from four tamarins (15,17). 4 At weeks 6, 12, 14, and 16 post-inoculation, the genomic sequence was determined from viral RNA isolated from serum. At each time point, the nucleotide sequence of the core-E1 signal sequence was identical to that of the RNA inoculated, confirming that GBV-B/Mut4 RNA was infectious and did not require adaptive mutations in this region to produce viral progeny. A final tamarin, T16472, was inoculated with pGBV-B/Mut7 RNA, which encoded a core species that was cleaved less efficiently compared with Mut4 in tissue culture cells. The virus titer peaked at 3 ϫ 10 7 ge/ml serum by 4 weeks post-inoculation (Fig. 4D), which was 2 orders of magnitude lower than that produced by GBV-B/Mut4 in T16458 (Fig. 4C), and then started to decline by 12 weeks post-inoculation. ALT values did not rise substantially above the enzyme activity determined at the time of inoculation (Fig. 4D). At weeks 4, 8, and 10 post-inoculation, the nucleotide sequence of viral RNA did not contain any mutations in the core-E1 signal sequence compared with the input RNA. Because a glycine residue at position 149 gave reduced amounts of SPP-cleaved core in tissue culture cells, we concluded that the partial defect in SPP processing was likely to be responsible for the lower viral titers produced by this construct. Taken together, our findings provide in vivo evidence that SPP is essential for productive GBV-B infection of tamarins.

DISCUSSION
Studies on HCV have been hampered by the limited availability of animal models and, until recently, tissue culture systems to define requirements for virus infection (35)(36)(37)(38). GBV-B has been a paradigm for HCV infection, given the close similarity of the two viruses and the ability of GBV-B to readily infect New World monkeys (12,14,16,17). Our aim was to determine whether SPP, which is required for maturation of HCV core protein (10,11), also processed the corresponding GBV-B protein and thereafter to examine the need for SPP cleavage in productive GBV-B infection in animals.
The signal sequence between the GBV-B core and E1 proteins is cleaved by signal peptidase between amino acids 156/ 157 in the viral polyprotein to generate the N terminus of the E1 glycoprotein (28). Here, we have demonstrated that SPP also cuts at the h-region in the core-E1 signal sequence to generate the mature form of core. Thus, an identical combination of cellular proteases directs maturation of both GBV-B and HCV core proteins. Moreover, GBV-B core is targeted to lipid droplets, and this localization requires prior cleavage by SPP. Targeting to lipid droplets relies on a domain (termed D2) in the C-terminal half of GBV-B core that has 41% sequence identity with the corresponding region in the HCV protein (24). These characteristics illustrate the similarities between GBV-B and HCV core proteins and validate the use of GBV-B as a surrogate system to examine the role of SPP processing in HCV infection.
A consensus sequence for SPP cleavage has not been established but the presence of helix-breaking or -destabilizing residues within the h-region of signal sequences is thought necessary to generate a substrate that is recognized by the protease (4,6,39). Because the hydrophobic amino acids valine and leucine are typically present in ␣-helical transmembrane regions, we tested other residues in the h-region for their influence on SPP cleavage. Mutation of cysteine residues at positions 146 and 152 had little effect on processing, but Ala 151 did contribute to SPP cleavage. However, a single mutation at Ser 149 was sufficient to eliminate almost any detectable processing, highlighting the critical role played by this amino acid. Based on this evidence, we examined the characteristics of serine that promoted cleavage. Our results indicated that glycine and alanine enabled processing, whereas tyrosine and methionine abolished SPP cleavage. We conclude that a small amino acid side chain is the major contributory factor, because alanine has a similar propensity to form ␣-helices as tyrosine and methionine (32). Nevertheless, the highest levels of processing were achieved with serine, indicating that it contributes additional properties to favor SPP cleavage. In transmembrane domains, the bonding requirements for the hydroxyl group on serine is accommodated through intrahelical hydrogen bonding with the carbonyl oxygen of residues on the preceding turn of the helix (40). This additional hydrogen bonding can induce a bend in the helix, thereby altering its conformation (41). Moreover, the O␥ in serine can be accessible also for interhelical hydrogen bonding (42). Hence, serine may play multiple roles in the mechanism and specificity of SPP processing. Firstly, intrahelical hydrogen bonding of serine may distort the helical h-region in the signal sequence. Secondly, a distortion in the signal sequence, combined with the small size of the side chain on serine, may allow closer apposition of SPP. Finally, serine could form interhelical hydrogen bonds with SPP and thus stabilize interactions between the protease and its substrate. We note that three other signal sequences (HCV core-E1, bovine prolactin, and human MHC class 1 molecule HLA-A*0301), contain serine residues in the h-region. In each case, mutations that included alteration of serine reduced cleavage efficiency (4,6,10). At the position equivalent to serine in GBV-B and HCV, the CSFV signal sequence contains a threonine residue (Fig. 1B, Ref. 31), which also could form intra-and interhelical bonds (42). Therefore, the available data support the model proposed above, although additional studies on the structure of SPP and its substrates are necessary to test directly our hypothesis.
In our previous studies, SPP cleavage was needed for the association of HCV core with lipid droplets (10, 43), but we could not test whether processing was required for virus production. Even with the recent availability of a tissue culture system for propagation of HCV (35)(36)(37), addressing the question of dependence on SPP processing to establish a robust infection in vivo remains difficult, because the only reliable animal model requires inoculation of chimpanzees. It has been demonstrated recently that growth of CSFV in tissue culture cells relies on SPP proteolysis to generate the core protein that forms the virus capsid (31), but again no studies have been conducted in animal systems. Here, we have tested mutants in the GBV-B core-E1 signal sequence with different capacities for SPP processing for their ability to establish infection in tamarins. The patterns of infectivity observed in each animal matched the cleavage efficiencies for each of the mutants in tissue culture systems, and therefore we conclude that productive infection requires maturation of GBV-B core by SPP. It has been suggested that compounds that inhibit SPP activity could be employed to combat HCV infection (44), because there would be reduced likelihood of developing resistance, a common feature with drugs that are directed against viral proteins. Our results validate SPP as a possible target for anti-viral therapy and suggest that compounds capable of inhibiting this host enzyme may have therapeutic efficacy in chronic hepatitis C.