Hepatitis C virus NS4B carboxy terminal domain is a membrane binding domain

A well-conserved part of the HCV NS4B protein is the carboxyl terminal domain (NS4B-CTD), which is also conserved within the hepacivirus genus [23]. Along with the expected cytosolic localization it proposes a separate function of the NS4B-CTD. To study the localization of NS4B-CTD various constructs were made. To each construct a Myc-epitope-tag was fused at the C-terminus as a detection epitope. These constructs were transfected into Huh7 human hepatoma cells and analyzed using immunofluorescence. Localization of the constructs was first compared to the endoplasmatic reticulum (ER), using Protein Disulphide Isomerase (PDI) as a marker. In Figure 1a, top left panel, Huh7 cells expressing full length NS4B (NS4B-FL, aa 1–261) are shown. NS4B-FL has a perinuclear and reticular staining, typical for ER. Additionally, the pattern of NS4B-FL largely overlaps with PDI. This ER-like staining confirms the previously described localization of native FL NS4B [20, 32]. To our surprise, expression of the NS4B-CTD alone (aa 188–261) does not show a cytosolic staining, but displays small punctate or dot like structures throughout the cells (Fig. 1a, CTD left panel). In the overlay of NS4B-CTD and PDI some co-localization is seen between the two (Fig. 1a, CTD right panel). Together with the small punctate staining, this suggests that the NS4B-CTD might be associated to membranes.

Since the NS4B-CTD shows a dot like pattern, it might have an effect on the attachment to membranes or even localization of NS4B-FL. Therefore, an NS4B lacking the CTD (NS4B-deltaCTD, aa 1–192) was constructed and examined in immunofluorescence. As shown in Figure 1a, NS4B-deltaCTD has a perinuclear and reticular staining, like NS4B-FL and PDI, indicating an ER-like localization (Fig. 1a). Also co-transfections of NS4B-FL and NS4B-deltaCTD show similar localization (data not shown). Together this implies that the absence of CTD does not seem to alter the localization of NS4B.

Two potential lipid modification sites for palmitoylation on cysteines, suggested by Yu and colleagues [24], might render the NS4B-CTD to membranes. We therefore investigated this possibility and mutated the two cysteines (cysteines 256 and 260) of the NS4B-CTD into serines (NS4B-CTD sub-Cys) (Fig. 1a) and expressed this mutant in Huh7 cells. Localization of the NS4B-CTD sub-Cys mutant was very similar to NS4B-CTD, exhibiting small punctate structures in the cells (Fig. 1a). It shows that the dot-like membrane localization of NS4B-CTD is caused by characteristics in the domain other than the cysteines at positions 256 and 260.

Membrane association of proteins can be investigated in a membrane floatation assay. In such an assay, a continuous-density gradient is loaded on top of a cell extract and subjected to centrifugation. Membranes and associated proteins float into the gradient, while cytosolic proteins stay in the loaded bottom fraction. To examine the suggested membrane association characteristics of the NS4B-CTD a membrane floatation assay was performed. Figure 2 shows the results of that assay, in which a cell lysate of Huh7 cells transfected with NS4B-CTD was used. Fractions were collected from the top (10%) to the bottom (80%) of the gradient and the odd fractions were analyzed by western blotting. As a control for cytosolic proteins, glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used. As expected, GAPDH was retained in the bottom fractions 21 and 23 of the density gradient, where the cell extract was loaded (Fig. 2). Calnexin, Transferrin receptor (TfR) and Cytochrome C oxidase subunit IV (COX-IV) are transmembrane proteins and float into the gradient, they are mainly observed in fractions 9 and 11 (Fig. 2). Since calnexin, TfR and COX-IV reside on different membranes in the cell (ER, the endocytic pathway and mitochondria), their distribution differs slightly (Fig. 2). The NS4B-CTD is detected in fractions 7 to 13 and 21 and 23 with its highest signal in fraction 11 (Fig. 2). In conclusion, similar to membrane proteins the NS4B-CTD floats into the gradient, implying membrane association of the NS4B-CTD. Together, the punctate structures in immunofluorescence and the floatation into the membrane floatation gradient, suggest association of the CTD of NS4B to membranes.

Since the CTD of NS4B only partially overlaps with the ER-marker, PDI (Fig. 1a), we were interested in knowing on which other membranes the NS4B-CTD resides. Therefore, co-localization studies with different organelle markers in Huh7 cells transfected with NS4B-CTD were performed. From the exocytic pathway we examined the Golgi (Giantin) and the ER-Golgi intermediate compartment (ERGIC) and found no substantial co-localization (data not shown). Similar results were obtained from co-localization studies with markers from the endocytic pathway, such as Rab5 from early endosomes, mannose-6-phosphate receptor and LAMP1, proteins that resides in late endosomes and lysosomes (data not shown). Recently, lipid droplets were demonstrated to play an important role in the HCV lifecycle [33]. However, no co-localization of lipid droplets and the NS4B-CTD was observed (data not shown). HCV proteins, Core, NS3 and NS4A are suggested to localize to or close to mitochondria [34, 35]. For that reason, co-localization of mitochondria and NS4B-CTD was investigated. We could observe considerable similarity in patterns between COX-IV, a mitochondrial protein marker and the NS4B-CTD (Fig. 1b). However, the overlap is not complete. Even though we did not specifically preserve the plasma membrane during immunofluorescence, we could occasionally see a fraction of NS4B-CTD at the plasma membrane (Fig. 1b. Inset). Taken together the CTD of NS4B seems to be mainly targeted to mitochondria, ER membranes and the plasma membrane.

The importance of the NS4B-CTD might be reflected by the sequence conservation within hepaciviruses. Its sequence conservation may also provide a clue to its function. Identification of potentially remote protein homologues can help to predict protein properties, like folding, structure and most importantly function. Similarity between distantly related proteins can be effectively established using profile based searches of databases of proteins families. In order to elucidate a possible function of the CTD of NS4B, we generated a multiple sequence alignment (profile) of the NS4B-CTD including all genotypes of HCV, Hepatitis GB virus A, B and C (Additional file 1), which we manually refined. Programs for profile-profile comparisons have been developed and are available as web-based tools. We used three different tools for profile-profile comparison with our HCV NS4B-CTD query profile, namely PRC [36], HHpred [37] and COMPASS [38] because each is sensitive to a different set of algorithms and the combination of the three tools reinforces independently detected relationships. This allows us to construct a consensus result with hits found by all tools. Using similar search parameters (see Materials and Methods) twelve, eight and five hits were found by PRC, HHpred and COMPASS respectively. Interestingly, only one protein family Lact-deh-memb (PF09330) was found by all three methods. This was the highest scoring profile for the three methods, next to the NS4B profile (E-values 0.057, 0.086 and 0.35 for the respective searches). According to HHpred the probability (which also includes the contribution from the secondary structure score) that lact-deh-memb is significant similar to NS4-CTD is 44.8%. Members of this Lact-deh-memb family are predominantly found in prokaryotic D-lactate dehydrogenase (d-LDH), which is a peripheral membrane respiratory enzyme located on the cytosolic site of the inner membrane [39]. Comparison of the sequence similarity between HCV NS4B-CTD and d-LDH from E. coli, of which the crystal structure has been resolved [39], revealed that the common region lies in the membrane binding domain (MBD) of d-LDH (Fig. 3a and Additional file 1) [40]. Thus besides apparent sequence similarity, both domains seem to perform similar functions, that is, they allow for membrane association. The MBD of d-LDH was suggested to bind nonspecifically to the membrane through (the positively charged) basic residues (Lys, Arg), interacting with the negatively charged phospholipids of the membrane, rather than penetrating the lipid bilayer [39‐41]. Part of the d-LDH MBD corresponding to the CTD of NS4B is disordered in the crystal structure and is thought to form a defined structure upon binding to the membrane [39]. The central alpha helix of the d-LDH MBD (Fig. 3b) corresponds to the extreme carboxy-terminal end of NS4B-CTD. The amino-acids on the membrane interface of this alpha helix and the corresponding residues of NS4B-CTD are indicated in Figure 3a.

Profile-profile comparison E-values in the range of 0.1–0.001 can indicate a true relationship, but require additional evidence to conclude that there is a functional parallel between NS4B-CTD and d-LDH-MBD in membrane binding [42]. Mutational studies could reveal functional similarity and were accordingly performed. D-LDH is a general membrane binding protein in E. coli located on the cytosolic side of the inner membrane. The position of such a protein in eukaryotic cells is unknown. Therefore, we first investigated localization of d-LDH MBD in Huh7 cells. As shown in Figure 4a d-LDH MBD (aa 319 to 390) mainly overlaps with COX-IV illustrating that when the d-LDH MBD is expressed separate from the enzyme part of the d-LDH protein, functionality of membrane binding is maintained. Moreover, co-transfection of NS4B-CTD and d-LDH MBD showed nearly complete overlap of the two patterns (Fig. 4b). Furthermore these immunofluorescence assays indicate that d-LDH MBD, a general membrane binding domain, has a preference for mitochondrial membranes in eukaryotic cells, which is comparable to the localization of NS4B-CTD (Fig. 4a), implying analogous membrane targeting of the two domains.

To test the hypothesis that the CTD of NS4B associates with the membranes in a way similar to the d-LDH MBD, we introduced mutations designed to disrupt the positive residues postulated to interact with the negative head groups of lipids [39, 40] (Fig. 3a). The side chains of the d-LDH MBD pointing away from the protein, facing the membrane surface are indicated in Figure 3b. Three positively charged amino acids (Lys 247, Arg 248 and His 250) in NS4B corresponding to the structured alpha-helix in d-LDH were simultaneously replaced with a negatively charged glutamic acid (K247E/R248E/H250E; NS4B-CTD tripleE), which should not be able to bind to phospholipid heads. The NS4B-CTD tripleE mutant was expressed in Huh7 cells and membrane association was investigated using immunofluorescence and a membrane floatation assay. Mutation of all three positively charged residues results in a dramatic change of localization of the NS4B-CTD, from punctate structures in the perinuclear region to a diffuse distribution throughout the cell, possibly cytosolic (Fig. 4a, compare NS4B-CTD to NS4B-CTD tripleE). Loss of membrane association was also shown in a continuous-density gradient, in which NS4B-CTD tripleE was detected in the same fractions as the cytosolic marker GAPDH (Fig. 2).

A functional parallel can also be examined by swopping part of the membrane binding domains of two proteins. A mutant was constructed, in which we exchanged the putative membrane contacting helix of NS4B-CTD for the corresponding membrane contacting helix of the d-LDH MBD (NS4B-CTD helix-swop) (Fig. 4a). Huh7 cells expressing NS4B-CTD helix-swop display punctate structures in immunofluorescence, though the staining has a slightly more diffuse localization compared to NS4B-CTD (Fig. 4a). Similarity in patterns with COX-IV also indicated that the NS4B-CTD helix-swop is targeted to membranes while the NS4B-CTD tripleE mutant has lost membrane binding. Furthermore a membrane floatation assay showed that NS4B-CTD helix-swop is membrane associated (Fig. 2), although compared to NS4B-CTD more was observed in the non-floating fractions. Altogether these results illustrate that the CTD of NS4B can interact with membranes via the positively charged residues, comparable to d-LDH MBD.

To examine the importance of the NS4B-CTD positively charged residues for RNA replication, we exchanged these amino acids involved in membrane association for negatively charged glutamic acids in selectable subgenomic replicons [9]. Huh7 cells transfected with replicon RNA that carry the three negatively charged residues (NS4B-CTD tripleE) did not yield any viable colonies (Fig. 5). Moreover the single mutations K247E and R248E were replication defective and gave no colonies (Fig. 5). Thus the positive residues are clearly indispensible for viral RNA replication in cell culture, suggesting that loss in membrane association leads to a replication defect. These results, together with the possible functional parallel between d-LDH MBD and the NS4B-CTD, prompted us to swop the membrane binding helix from d-LDH MBD (PPRMKNWRDK) into replicons (helix-swop, Fig. 3a) and determine colony formation. These replicons in which eight amino acids are exchanged indeed formed several viable colonies (40 colony forming units per ug (CFU) of transfected replicon RNA) (Fig. 5). Clearly, far less colonies were formed relative to wild type (~10.000 CFU), but the replication defect from the helix swop is less than the negative charged mutations, where no colonies were formed. Two separate replicon colonies derived from the NS4B-CTD helix-swop were expanded, RNA isolated and sequenced to analyze whether they still contained the original mutations. Interestingly, the complete introduced helix was retained, confirming the importance of this membrane contacting helix.

The following antibodies were used anti-PDI (Stressgen), anti-Myc (mouse) (Invitrogen), anti-Myc (rabbit) (Roche), anti-GAPDH (SantaCruz), anti-Transferrin receptor, clone H68.4 (Zymed Laboratories Inc), anti-COX-IV (Abcam), anti-Calnexin (BD) and anti-HA (Abcam).

Human hepatoma cell line Huh7 was grown in Dulbecco's Modified Eagle's Medium supplemented with Non-essential amino acids, L-glutamate, Penicillin and Streptavadin. Cells were subcultured using Trypsin and transfected using Fugene6 (Roche) at a DNA/reagent ratio of 1/3, according to manufacturers' instructions.

To construct Myc-epitope-tagged expression plasmids, the sequence was amplified by PCR from pFK5.1Neo [51] or E.coli DNA using specific primers, see Table 1. The PCR products were digested with Kpn I and Xba I and ligated into pCDNA3.1mychisB (Invitrogen) similarly digested with Kpn I and Xba I. This resulted in the construction of expression vectors containing a 10-residue Myc-epitope-tag at its C-terminus. In order to construct HA-epitope-tagged NS4B expression constructs, the Myc-epitope sequence was Xba I – Pme I cut and replaced with an Xba I – Pme I fragment coding for the HA-epitope.

In vitro transcription, electroporation and selection of G418-resistant cell lines was done as described previously [52].

24 h post transfection cells were fixed with 3% paraformaldehyde (PFA) in PBS (154 mM NaCl, 1.4 mM Phosphate, pH 7.5). PFA was quenched using 50 mM NH4Cl in blockbuffer, which contained 5% fetal calf serum (FCS) in PBS. The cells were permeabilized with 0.1% TritonX-100 in blockbuffer and stained with primary antibodies diluted in blockbuffer for 1 h. Next the coverslips were washed with glycinebuffer, 10 mM glycine in PBS, and incubated with secondary antibody diluted in blockbuffer for 1 h. After washing with glycinebuffer, PBS and water, the coverslips were mounted with Prolong (Invitrogen) mounting medium. Fluorescence images were captured using a Zeiss Axioskop 2 fluorescence microscope equipped with the appropriate filter sets, a digital Axiocam HRc camera and Zeiss Axiovision 4.4 software. Images were optimized with Adobe Photoshop CS2.

Transfected Huh7 cells were lysed after 24 h in buffer that contained 20 mM Tris pH 7, 1 mM MgCl₂, 15 mM NaCl and 240 mM sucrose using a ball bearing homogenizer (Isobiotec, Heidelberg Germany). Whole cells and cell debris was spun down at 500 × g for 5 min and supernatant was collected. Cell extracts were mixed with sucrose to 80% w/v and overlaid with a linear sucrose gradient (80%–10% w/v sucrose, 50 mM Tris pH 7, 1 mM MgCl₂, 15 mM NaCl). After centrifugation in a SW41 tube for 15 h at 100,000 × g (Beckmann ultracentrifuge), 500 μl fractions were collected from the top. The odd fractions were analyzed by western blotting, either directly or subsequent to concentration. 200 μl of each fraction was concentrated using 9 volumes of ethanol and incubated overnight at -20°C, followed by centrifugation at max in an Eppendorf 5417R for 1 h. The protein pellets were dissolved in 1× Laemmli.

After separation on SDS-PAGE gels, proteins were transferred to PVDF membranes (HydrobondP, GE-Healthcare) using a Semi-Dry blot apparatus (Biorad). Membrane blocking and antibody incubations were performed using 0.5% Tween-20, 5% non-fat, dry milk (Campina) in PBS. Since all secondary antibodies were conjugated to horseradish peroxidase, the proteins were visualized using enzyme-catalyzed chemoluminescence (ECL+, GE-Healthcare) and Fuji Super RX medical X-ray film.

COMPASS http://​prodata.​swmed.​edu/​compass/​, database pfam21.0, 0 PSI-blast iterations, E-value threshold was set at 10. Profile comparer (PRC; http://​supfam.​org/​PRC), database pfam22.0, E-value threshold was set at 10. HHpred http://​toolkit.​tuebingen.​mpg.​de/​hhpred, selected database pfamA_22.0, 0 PSI-blast iterations.