ABSTRACT
Viral interleukin-6 (vIL-6) encoded by human herpesvirus 8 (HHV-8) is believed to contribute via mitogenic, survival, and angiogenic activities to HHV-8-associated Kaposi's sarcoma, primary effusion lymphoma (PEL), and multicentric Castleman's disease through autocrine or paracrine mechanisms during latency or productive replication. There is direct evidence that vIL-6 promotes latently infected PEL cell viability and proliferation and also viral productive replication in PEL and endothelial cells. These activities are mediated largely through endoplasmic reticulum (ER)-localized vIL-6, which can induce signal transduction via the gp130 signaling receptor, activating mitogen-activated protein kinase and signal transducer and activator of transcription signaling, and interactions of vIL-6 with the ER membrane protein vitamin K epoxide reductase complex subunit 1 variant 2 (VKORC1v2). The latter functional axis involves suppression of proapoptotic lysosomal protein cathepsin D by promotion of the ER-associated degradation of ER-transiting, preproteolytically processed procathepsin D. Other interactions of VKORC1v2 and activities of vIL-6 via the receptor have not been reported. We show here that both vIL-6 and VKORC1v2 interact with calnexin cycle proteins UDP-glucose:glycoprotein glucosyltransferase 1 (UGGT1), which catalyzes monoglucosylation of N-glycans, and oppositely acting glucosidase II (GlucII), and that vIL-6 can promote protein folding. This activity was found to require VKORC1v2 and UGGT1, to involve vIL-6 associations with VKORC1v2, UGGT1, and GlucII, and to operate in the context of productively infected cells. These findings document new VKORC1v2-associated interactions and activities of vIL-6, revealing novel mechanisms of vIL-6 function within the ER compartment.
IMPORTANCE HHV-8 vIL-6 prosurvival (latent) and proreplication functions are mediated from the ER compartment through both gp130 receptor-mediated signal transduction and interaction of vIL-6 with the ER membrane protein VKORC1v2. This report identifies interactions of vIL-6 and VKORC1v2 with calnexin cycle enzymes GlucII and UGGT1, which are involved in glycan processing and nascent protein folding. The presented data show that vIL-6 and VKORC1v2 can cocomplex with GlucII and UGGT1, that vIL-6 promotes protein folding, and that VKORC1v2, UGGT1, and vIL-6 interactions with GlucII and UGGT1 are important for the profolding activity of vIL-6, which can be detected in the context of infected cells. This newly identified ER activity of vIL-6 involving VKORC1v2 may promote viral latency (in PEL cells) and productive replication by limiting the damaging effects of unfolded protein response signaling in addition to enhancing viral protein folding. This is the first report of such a function for a cytokine.
INTRODUCTION
Viral interleukin-6 (vIL-6) specified by human herpesvirus 8 (HHV-8) is unique among IL-6 homologues in that it can signal through the signaling membrane receptor gp130 in the absence of the nonsignaling coreceptor, gp80 (1–3). This may in part explain the ability of vIL-6 to signal from the endoplasmic reticulum (ER); such signaling occurs exclusively through vIL-62:gp1302 tetrameric complexes, in contrast to gp80-containing hexameric complexes (IL-62:gp1302:gp802) formed by vIL-6 and other IL-6 proteins at the cell surface (4–8). The inefficient secretion and predominant ER localization of vIL-6 is another feature that is particular to the viral cytokine and may contribute to biologically significant autocrine activities of vIL-6 (7, 8). In primary effusion lymphoma (PEL) cells, vIL-6 is expressed in the latent phase and acts through ER-localized autocrine mechanisms, including gp130-activated mitogen-activated protein kinase and signal transducer and activator of transcription 3 (STAT3) signaling, to maintain cell proliferation and viability (8, 9). ER-localized interactions of vIL-6 with gp130 and STAT3 signaling are also critical for virus productive replication in PEL cells (10).
In addition to the established ER-localized activities of vIL-6 via gp130 signaling, vIL-6 interacts with other ER proteins that may contribute to functions of the viral cytokine. The first non-gp130 interactor of vIL-6 to be identified was the ER-resident lectin-type chaperone protein calnexin, which is critical for the folding of many diverse glycoproteins entering the secretory pathway (11, 12). This interaction was found to be more robust and durable than typical, transient interactions of calnexin with folding intermediates, and potentially may contribute to the ER localization of vIL-6 (12). However, the possible function of this interaction was not addressed. Yeast two-hybrid screening identified another interaction of vIL-6, namely, with vitamin K epoxide reductase complex subunit 1 variant 2 (VKORC1v2), encoded by a splice-variant of VKORC1 (variant 1), the target of warfarin (13, 14). The region of VKORC1v2 responsible for interaction with vIL-6 was subsequently mapped and shown to be important for progrowth and proviability functions of VKORC1v2 in latently infected PEL cells (13). Subsequent studies identified the ER-transiting proenzyme procathepsin D (pCatD) as a cointeraction partner of VKORC1v2 and vIL-6 and vIL-6-VKORC1v2 complexing to be necessary for the ER retention and destabilization of pCatD via physical and functional association with components of the ER-associated degradation (ERAD) machinery (15, 16). In the context of both latently and lytically infected PEL cells, vIL-6 was found to suppress mature (lysosomal) CatD levels and associated proapoptotic activity, with consequent promotion of virus production in lytically reactivated cells; peptide-mediated disruption of vIL-6-VKORC1v2 interaction had converse effects (16). Thus, one function of VKORC1v2 and a mechanism of vIL-6 activity via the novel ER membrane protein appears to be the regulation of ERAD processing of pCatD. Other coassociating proteins may be similarly regulated.
Another reported vIL-6-interacting ER protein is hypoxia-upregulated protein 1 (HYOU1), also known as 150-kDa oxygen-regulated protein (ORP150) and 170-kDa glucose-regulated protein (GRP170), which was identified by mass spectrometry as a vIL-6-precipitated protein coaffinity-purified from transfected HEK293T cells and subsequently verified to interact with vIL-6 in PEL cells (17). Interestingly, vIL-6 levels in both transfected HEK293T cells and in PEL (BCBL-1) cells were influenced negatively by HYOU1 depletion, demonstrating an important role of HYOU1 in regulating expression of the viral cytokine. Independently of this effect, HYOU1 depletion also inhibited gp130-activated STAT3 signaling induced by vIL-6 in transfected cells, and this correlated with reduced vIL-6-gp130 interaction. Furthermore, HYOU1, a known promoter of angiogenesis through vascular endothelial growth factor processing (18, 19), was found to be required for vIL-6-induced endothelial cell migration, again showing the importance of the cellular protein for vIL-6 activity. HYOU1 was also shown to be important for vIL-6 induction of the angiogenic chemokine CCL2 in endothelial cells, implicating this factor, as well as VEGF, in vIL-6- and HYOU1-promoted endothelial cell migration (17). Collectively, these data implicate HYOU1 as a functionally significant interaction partner of vIL-6 (via direct or indirect association) and of potential significance in Kaposi's sarcoma.
Our previous identification of calnexin as a vIL-6 interaction partner and identification in this paper of calnexin cycle proteins UDP-glucose:glycoprotein glucosyltransferase 1 (UGGT1) and glucosidase II (subunit α) as vIL-6-associating proteins indicate that vIL-6 may influence protein folding via calnexin cycle targeting, or itself be processed by it. Furthermore, HYOU1 is known to coassociate with a subset of chaperones, protein disulfide isomerases, and other proteins, including UGGT1, involved in protein folding within the ER (20). ER glycoprotein folding involves association of nascent proteins with lectin chaperones calnexin or calreticulin that recognize, with different specificities, unfolded proteins having monoglucosylated Man9GlcNAc2 glycans (11, 21). Unfolded proteins are released upon glucosidase II (GlucII)-mediated glycan deglucosylation and can either reenter the calnexin/calreticulin cycle following UGGT-mediated reglucosylation and undergo further folding attempts or exit the cycle as misfolded proteins and undergo degradation via ERAD (22, 23). We have previously identified the promotion of pCatD degradation through ERAD and (direct or indirect) associations of vIL-6 and VKORC1v2 with ERAD proteins (chaperones and translocon components), implicating roles of vIL-6 and VKORC1v2 in ERAD degradation (15). The interactions of vIL-6 with calnexin and HYOU1 (12, 17) raise the possibility of vIL-6 involvement in additional aspects of ER quality control via protein folding. Our identification here of vIL-6 and VKORC1v2 interactions with calnexin cycle enzymes GlucII and UGGT1 lends further support to this hypothesis, which is addressed in this report.
RESULTS
Identification of new interaction partners of vIL-6 and VKORC1v2.To further assess VKORC1v2 and ER-localized vIL-6 functions, StrepII+Flag (SF) affinity-tagged vIL-6 or VKORC1v2 were expressed in transfected HEK293T cells and affinity purified and trypsin-digested proteins analyzed by mass spectrometry (MS) for identification of vIL-6 and VKORC1v2 interactors. These screening experiments identified the glucosidase II α-subunit (GlucIIα; also known as glucosidase, alpha, neutral AB [GANAB] [24, 25]) as the most highly represented protein for vIL-6. 352 protein-specific peptides of GlucIIα were detected, compared to no GlucIIα peptides detected in trypsin-digested precipitates of SF-tagged VKORC1v1 (variant 1), used as a control. GlucIIβ was also identified in this screen (66 protein-specific peptides for vIL-6 versus 1 for VKORC1v1). It is notable that GlucII α and β subunits were also detected as interaction partners of vIL-6 in similar experiments undertaken in HEK293T cells by Davis et al. (26). For VKORC1v2, UDP-glucose:glycoprotein glucosyltransferase 1 (UGGT1 [27, 28]) was identified in coprecipitates (43 protein-specific peptides for VKORC1v2-SF relative to 0 for SF-VKORC1v1). Notably, GlucIIα was also identified as a coprecipitant of VKORC1v2. These initial findings (summarized in Table 1) prompted us to test the potential interactions in standard coprecipitation experiments.
Summary of MS-identified GlucIIα/β and UGGT1 tryptic peptides derived from proteins precipitated from cell extracts by StrepII+Flag (SF)-tagged, tandem-affinity-purified vIL-6 and VKORC1v2
Testing of vIL-6 and VKORC1v2 interactions with their respective potential targets, GlucIIα and UGGT1, was done by expressing chitin-binding domain (CBD)-fused vIL-6 and VKORC1v2 with S-peptide- or StrepII-tagged GlucIIα or UGGT1, respectively, in HEK293T cells transfected with appropriate expression vectors (see Materials and Methods). CBD-fused versions of human IL-6 (hIL-6), terminally tagged with a KDEL ER retention motif, and VKORC1v1 were used as negative controls. The candidate vIL-6 and VKORC1v2 targets were observed in the respective precipitates, but not in the controls, as determined by S-peptide and StrepII immunoblotting for detection of epitope-tagged GlucIIα and UGGT1, respectively (Fig. 1A). In view of the interaction of vIL-6 with VKORC1v2 (13), associated functions of GlucII and UGGT1 (29), and mass spectrometry (MS) identification of GlucIIα, in addition to UGGT1, in VKORC1v2 coprecipitates (Table 1), we undertook an additional experiment to test for “reciprocal” vIL-6 and VKORC1v2 interactions with UGGT1 and GlucIIα, respectively. Indeed, each “bait” protein was able to precipitate each of the two target proteins (Fig. 1B). These data verified the initial interactions suggested by MS and indicated possible cocomplexing between these proteins. Interactions of endogenous vIL-6 with both GlucIIα and UGGT1 was confirmed by vIL-6-based immunoprecipitation from BCBL-1 PEL cell lysates (Fig. 1C).
Interactions of vIL-6 and VKORC1v2 with GlucIIα and UGGT1. (A) Coprecipitation assays were carried out using lysates of cells cotransfected with expression plasmids for CBD-fused vIL-6 and S-tag-fused GlucIIα or CBD-fused VKORC1v2 (v2-CBD) and StrepII-tagged UGGT1 to test the interactions between these protein pairs. CBD- and KDEL ER retention motif-linked hIL-6 (hIL-6-CBD.K) and CBD-tagged VKORC1v1 (v1-CBD) were used as negative controls in the respective experiments. Protein complexes were precipitated with CBD-binding chitin beads, and SDS-PAGE-fractionated and membrane-transferred proteins in precipitates and cell lysates were identified by immunoblotting with CBD antibody to visualize “baits” vIL-6 and VKORC1v2 or with S-peptide or StrepII antibody for detection of their candidate binding partners. β-actin probing of cell lysates provided a loading control. (B) Further precipitations were undertaken to test whether vIL-6 and VKORC1v2 could interact with UGGT1 and GlucIIα, respectively. In this experiment, GlucIIα-S and StrepII-UGGT1 were coexpressed with CBD-tagged vIL-6 or VKORC1v2, or negative controls hIL-6-CBD.K (hIL-6-C′.K) or VKORC1v1-CBD (v1-CBD). Chitin bead precipitates and cell lysates were analyzed as described previously. (C) Immunoprecipitation (IP) of vIL-6 from BCBL-1 PEL cell lysates and subsequent immunoblotting for GlucIIα and UGGT1. A sample of the lysate (10% of amount used for IP) was analyzed alongside material precipitated with vIL-6 antiserum (vIL-6) or control rabbit serum.
Complexing between vIL-6, VKORC1v2, GlucIIα, and UGGT1.Density gradient sedimentation (see Materials and Methods) was used to size-fractionate and distinguish complexes containing VKORC1v2. Flag-tagged VKORC1v2 was coexpressed in transfected HEK293T cells with differently epitope-linked vIL-6, GlucIIα, and UGGT1. The transfected culture lysate was used for Flag-based immunoprecipitation, and the Flag peptide-released precipitated material was loaded onto the density gradient; following ultracentrifugation, fractions were analyzed by immunoblotting. Data from this experiment revealed that vIL-6, GlucIIα, and UGGT1 each could be coprecipitated with VKORC1v2, as expected, and that a proportion of these proteins cofractionated in lower regions of the gradient (corresponding with high molecular weight), most notable in fraction 10 but also detectable in fractions 5, 7, and 8 (Fig. 2). This is consistent with potential cocomplexing of the four input proteins. Notable is that VKORC1v2 was the only protein detectable in fraction 1, likely representing free VKORC1v2, and that a very high proportion of vIL-6 was detected in light fractions (2 to 4) toward the top of the gradient and corresponding very well with the profile of VKORC1v2. These data are consistent with association of vIL-6 directly with VKORC1v2 (13) or in simple cocomplexes with low-molecular-weight proteins. The relatively weak detection of GlucIIα and UGGT1 relative to vIL-6 suggests that GlucIIα/UGGT1-VKORC1v2 association, in contrast to vIL-6-VKORC1v2 binding, may be dependent on limiting amounts of one or more cellular cointeraction partner (although the S-peptide antibody-probed GlucIIα blot cannot be compared directly to the StrepII antibody-probed vIL-6 and UGGT1 blots). The profiles for GlucIIα (fractions 5 and 7 to 11) and UGGT1 (fractions 3 to 5, 7, 8, and 10) are notably distinct from those of VKORC1v2 and vIL-6; the unique profiles for GlucIIα and UGGT1 suggest that they can form multiple different complexes with VKORC1v2, possibly involving distinct cellular proteins. Nonetheless, a peak for each of the four proteins in fraction 10 is suggestive of at least one cocomplex containing all four proteins.
Identification of large protein complexes containing VKORC1v2, vIL-6, GlucIIα, and UGGT1. Flag-tagged VKORC1v2, coexpressed with epitope-tagged vIL-6 (twin-StrepII, TS), GlucIIα (S-peptide), and UGGT1 (StrepII), was precipitated from cotransfected HEK293T cell lysates using Flag-immunobeads, and the washed precipitate was layered onto a 5 to 50% iodixanol gradient and ultracentrifuged at 107,000 × g for 3 h. Fractions of the gradient were extracted sequentially from the top of the column (the “light” fraction, fraction 1) to the bottom (fraction 14). Samples of each fraction were then subjected to SDS-PAGE, membrane transfer, and immunoblotting for the detection of each of the vector-expressed proteins. Horizontal arrows indicate continuous fractions in which VKORC1v2 and vIL-6 were detected; GlucIIα and UGGT1 were detected in particular fractions (+); copeaks (*) of all four proteins were observed in fraction 10 (vertical arrow).
Cocomplexing between vIL-6, VKORC1v2, GlucIIα, and UGGT1 was tested directly by serial precipitation of epitope/affinity-tagged VKORC1v2 (Flag, and also StrepII), vIL-6 (HA), and GlucIIα (S-peptide) and final detection by immunoblotting for UGGT1 (StrepII-tagged). All proteins were expressed together in HEK293T cells by cotransfection of the respective expression plasmids. To provide controls, parallel transfections were set up with each of the expression plasmids individually excluded. The lysates from each of the transfected cultures were subjected to initial precipitation with Flag-antibody beads (for VKORC1v2 precipitation), secondary precipitation of Flag peptide-released material with HA-antibody beads (vIL-6 sedimentation), and tertiary precipitation of HA peptide-released coprecipitates with S-protein beads (targeting GlucIIα). At each step, precipitated material was analyzed by immunoblotting for detection of each of the epitope-tagged proteins, to both confirm appropriate precipitation of “bead-targeted” proteins and monitor coprecipitated proteins, including any nonspecific binding evident in the control lysates (lacking the directly precipitated protein). UGGT1 (StrepII-tagged) was detected in the final precipitate (lane 1 of the “Precip. 3” panel in Fig. 3), thereby demonstrating the existence of complexes containing all four input proteins. The StrepII-UGGT1 protein band on the final immunoblot was specific for the lysate containing all four proteins, thereby ruling out potential nonspecific binding of the protein to the affinity-beads. Similarly, specificity of binding in the first and second precipitation steps was apparent from the controls lacking the respective “bait” proteins. From these data, we conclude that VKORC1v2, vIL-6, GlucIIα, and UGGT1 can cocomplex.
Cocomplexing between VKORC1v2, vIL-6, GlucIIα and UGGT1. Flag- and StrepII-tagged VKORC1v2 HA-tagged vIL-6, and S-tagged GlucIIα were coexpressed with StrepII-UGGT1 in expression vector-cotransfected HEK293T cells. The first three proteins were targeted for serial precipitation with Flag-antibody, HA-antibody, and S-protein beads, respectively. After the first and second precipitations, 3×Flag and HA peptides were used for competitive release of proteins from the beads (see Materials and Methods). Samples from each precipitation and input lysates were denatured, fractionated by SDS-PAGE, blotted onto membranes, and probed with appropriate epitope-tag antibodies (as indicated on the blots) for detection of the respective proteins. The test (all proteins) and various control transfections are tabulated (top left), and the precipitation sequence and targeted proteins are illustrated above the associated blots. Immunoblot band identities are indicated by arrows, asterisks and associated labels; note that detection of S-peptide (*GlucIIα) preceded probing for StrepII (UGGT1 and VKORC1v2), accounting for codetection on the upper blots. The detected StrepII-UGGT1 band in the final precipitate is indicated by an arrowhead (lane 1, rightmost blot).
VKORC1v2-independence of vIL-6 association with GlucIIα and UGGT1.Although the VKORC1v2-based coprecipitation and density gradient experiments (Fig. 1 to 3) indicated VKORC1v2 involvement in vIL-6 complexing with GlucIIα and UGGT1, the initial MS analysis of vIL-6-precipitated versus VKORC1v2-precipitated proteins indicated that GlucIIα, at least, may be a VKORC1v2-independent target of the viral cytokine. To test the requirement for VKORC1v2 for vIL-6 interactions with GlucIIα and UGGT1, we constructed a cell line deleted of VKORC1v2 (and VKORC1v1) using CRISPR-targeting of the VKORC1 gene in HEK293T cells (see Materials and Methods). Dual-allele frameshift mutations were confirmed by sequencing of PCR products spanning the short-guide RNA-targeted locus; also, wild-type sequence-specific PCR primers failed to amplify product from the knockout cell line (data not shown), confirming the absence of wild-type alleles. Using this cell line and also native HEK293T cells for comparison, cotransfection-based coprecipitation experiments were carried out to detect interaction of GlucIIα-S and StrepII-UGGT1 with CBD-tagged vIL-6, which was precipitated from transfected cell lysates with chitin beads. The data from these experiments (Fig. 4A) demonstrated associations of GlucIIα and UGGT1 with vIL-6 in the lysates of both wild-type and VKORC1v2-knockout HEK293T cells and the lack of detectable cell type-specific difference in binding efficiency. A follow-up serial precipitation experiment was carried out to examine cocomplexing between vIL-6, GlucIIα and UGGT1 in wild-type versus VKORC1v2-knockout HEK293T cells. In both cell types, GlucIIα-S-containing vIL-6-Flag coprecipitates, isolated by primary Flag-antibody and secondary S-protein precipitation, also contained StrepII-UGGT1, identified by immunoblotting (Fig. 4B). No coprecipitated StrepII-UGGT1 protein was detected from control lysates isolated from transfectants lacking GlucIIα-S or containing Flag-tagged VKORC1v1 in place of vIL-6-Flag, demonstrating interaction specificity. The results show that cointeraction of vIL-6, GlucIIα, and UGGT1 can occur independently of VKORC1v2.
VKORC1v2-independence of vIL-6 interactions with GlucIIα and UGGT1. (A) Coprecipitation assays were carried out essentially as outlined in the legend to Fig. 1, but genetically engineered VKORC1v2-deficient (KO) cells were included along with native HEK293T cells. Chitin beads were used to precipitate vIL-6-CBD from the lysates of both cell types; coprecipitated GlucIIα-S and StrepII-UGGT1 (left and right panels, respectively) were detected by immunoblotting. Empty vector (vec) was used as a negative control to check for nonspecific matrix binding by the proteins. (B) Serial precipitations of vIL-6 (Flag-tagged) and GlucIIα (S-tagged) from appropriately transfected wild-type and VKORC1v2-KO cell lysates was carried out, followed by immunoblotting to detect coprecipitated UGGT1 (StrepII-tagged). StrepII/Flag-VKORC1v1 (S/F-VKORC1v1) was included in some of the transfection mixes to provide a negative control.
Influence of vIL-6 on protein folding.In view of the interaction of vIL-6 with GlucIIα and UGGT1, proteins integral to the calnexin cycle, we hypothesized that vIL-6 might promote protein folding. In the context of lytic replication, such activity may serve to optimize virus production through promotion of viral protein synthesis and suppression of the unfolded protein response (UPR), a stress response with the potential to negatively impact virus replication (30). To test the influence of vIL-6 on protein folding, we took advantage of a variant of α1-antitrypsin (AAT) known as null Hong Kong (NHK) variant. NHK-AAT folds very inefficiently and accumulates largely as insoluble aggregates within the ER, but its solubility can be increased by UGGT1 and calnexin engagement (31–33). NHK-AAT was expressed as a green fluorescent protein (GFP)-fused protein (NHK-GFP) in transfected HEK293T cells either with or without vIL-6 coexpression. Cell lysates were prepared by membrane disruption with NP-40-containing buffer, followed by high-speed centrifugation to segregate “soluble” (supernatant) from “insoluble” (pellet) fractions of NHK-GFP. These fractions were analyzed by Western blotting. This experiment revealed the expected high proportion of misfolded insoluble NHK-GFP in the control empty vector-transfected cells (Fig. 5A). However, vIL-6 coexpression with NHK-GFP increased the relative amounts of soluble to insoluble protein, more than doubling the ratio (Fig. 5A chart). As an independent approach, we assessed secretion of HHV-8 glycoprotein L (gL) as a marker of successful folding. Since gL is dependent on chaperone functions of gH for its secretion (34), the two glycoproteins were coexpressed, as CBD fusions, in transfected HEK293T cells either with or without vIL-6 coexpression. Cells and culture media were harvested 48 h posttransfection for CBD-based immunoblotting analysis of cellular and secreted gL, the latter precipitated from culture media with chitin beads. Comparisons of intracellular and secreted gL in the absence (empty vector-transfected cells) and presence of vIL-6 showed an increase (>3-fold) in the proportion of secreted gL for the latter (Fig. 5B). These data, together with those of Fig. 5A, indicate that vIL-6 promotes protein folding.
Profolding activity of vIL-6. (A) HEK293T cells were cotransfected with vectors expressing a GFP-fused version of the NHK folding variant of α1-antitrypsin (NHK-GFP) and vIL-6 or with NHK-GFP expression plasmid and empty vector (vec). At 48 h posttransfection, cell lysates were generated using NP-40-containing buffer, material microcentrifuged, and pellet (insoluble [insol.]) and supernatant (soluble [sol.]) fractions were denatured, SDS-PAGE fractionated, and immunoblotted for detection of GFP and also vIL-6 (to confirm expression) and ER luminal soluble protein BiP (to ensure appropriate separation of soluble and pellet fractions). The relative amounts of soluble to insoluble NHK-GFP are shown in the chart. (B) An independent assay was undertaken using secreted HHV-8 glycoprotein L (gL), fused to CBD, as the readout for folding. CBD-tagged HHV-8 gH was coexpressed with gL-CBD to enable secretory trafficking, and either vIL-6 expression plasmid or empty vector (vec) was cotransfected into HEK293T cells with the glycoprotein expression plasmids. After 48 h, cells and media were harvested for preparation of lysates and chitin bead precipitates (for concentration of secreted gL-CBD). Cell lysates (1% of total) and media precipitates (10%) were analyzed by immunoblotting for detection of gL-CBD; gH, vIL-6, and β-actin (loading control) were also analyzed in the lysate fractions. (C) An experiment equivalent to that of panel A was carried out in native HEK293T cells or VKORC1v2- or UGGT1-knockout (KO) derivatives. CBD- and KDEL motif-tagged hIL-6 (h6-C′-K) was included as a negative control along with empty vector (vec) for comparison with vIL-6 (CBD-fused, v6-CBD). Calculated ratios of NHK-GFP in the soluble and insoluble fractions of cell lysates are shown below each corresponding set of immunoblots. Digitally captured data from all immunoblots (panels A to C) were quantified using GeneTools (Syngene) analysis software.
To test the involvement of VKORC1v2 and UGGT1 in this activity of vIL-6, the NHK-AAT-based experiment was repeated in wild-type and VKORC1v2- and UGGT1-knockout HEK293T cells (see Materials and Methods). The data (Fig. 5C) again showed a significant rescue by vIL-6, relative to empty vector-transfected or hIL-6-CBD-KDEL-expressing cells, of NHK-GFP from aggregation into the insoluble fraction in native HEK293T cells, but this activity was completely absent in the VKORC1v2- and UGGT1-negative cells. Therefore, VKORC1v2 is essential for the “profolding” activity of vIL-6, although it is not required for association of vIL-6 with GlucIIα and UGGT1 (Fig. 4), and UGGT1, which mediates glycan reglucosylation and glycoprotein retention in the calnexin cycle, is also critical for this function.
Regions of vIL-6 involved in GlucIIα, VKORC1v2 and UGGT1 interactions.To try to identify vIL-6 variants abrogated for interactions with GlucIIα and VKORC1v2, we utilized a series of domain substitution variants of vIL-6 in which vIL-6 sequences had been replaced with equivalent regions of hIL-6 (35) (Fig. 6A). Expression vectors for each of these proteins, wild-type vIL-6, or KDEL-tagged (ER-retained) hIL-6 control, all fused to CBD, were cotransfected with expression plasmids for epitope-tagged GlucIIα or VKORC1v2 into HEK293T cells. The IL-6-CBD proteins were precipitated from cell lysates with chitin beads, and these proteins and coprecipitated GlucIIα and VKORC1v2 were detected by immunoblotting. Data from these experiments identified vIL-6.hSP (hIL-6 “SP+N” sequences [amino acids 1 to 45, from initiator methionine], substituted for vIL-6 “SS” sequences [residues 1 to 27], Fig. 6A) as abrogated for interactions with both proteins (Fig. 6B and C). The vIL-6 variant substituted with hIL-6 sequences comprising the N-terminal half of α-helix D (vIL-6.hD1) showed weakened interaction with GlucIIα, specifically; this was observed in several repeat experiments (e.g., Fig. 6B, right). Further substitution mutagenesis of the D1 region was carried out, which comprised generation of grouped viral-to-human IL-6 scanning substitutions of 4 to 9 residues in D1. None of these more refined variants was significantly affected with respect to GlucIIα interaction (data not shown). However, one of the variants, designated vIL-6.D1m11 (RLQGLKY166 → KLQAKNQ, with the underlined residues corresponding to colinear hIL-6 amino acids), was abrogated for gp130 interaction (Fig. 7A), contrasting with unaffected gp130 associations and activities noted previously for vIL-6.hSP, but possibly equivalent to the partially altered (gp80-dependent) activities of vIL-6.D1 and most of the other Fig. 6A-depicted domain-substitution variants of vIL-6 (35). Testing of hSP, hD1, and D1m11 variants for UGGT1 binding revealed weakened interactions of vIL-6.hSP and vIL-6.D1m11 but unaffected binding by vIL-6.hD1 (Fig. 7B). In analogous assays, vIL-6.D1m11 was competent for interaction with VKORC1v2 (Fig. 7C); as before (Fig. 6C), VKORC1v2 interacted with vIL-6.hD1 but not with vIL-6.hSP. Taken together, our substitution-binding experiments identified vIL-6.hSP, vIL-6.hD1, and vIL-6.D1m11 as abrogated, respectively, for interactions with GlucIIα, VKORC1v2, and UGGT1, binding to GlucIIα, or association with UGGT1 and gp130. Functional studies have identified altered gp130 association (lack of induced dimerization) by vIL-6.hD1 in the absence of gp80 but wild-type signaling properties of vIL-6.hSP (35).
GlucIIα and VKORC1v2 binding properties of vIL-6 domain substitution variants. (A) Diagrammatic representation of vIL-6 variants containing domain-based substitutions of hIL-6 residues for colinear vIL-6 sequences; plasmid vectors expressing each protein fused C-terminally to CBD were reported previously (35). SS, signal sequence; SP+N, signal peptide plus N-terminal residues; D1 and D2, N- and C-terminal halves of the helix-D region, respectively. (B and C) Coprecipitation-based binding assays equivalent to those of Fig. 1 were carried out to determine the GlucIIα (S-tagged) and VKORC1v2 (SF-tagged) binding activities of each of the CBD-fused vIL-6 variants. Negative controls were provided by KDEL motif-linked hIL-6-CBD (hIL-6-KDEL), empty vector (vec) in place of GlucIIα-S (B) and SF-VKORC1v1 (Flag-v1) as a substitute for VKORC1v2-SF (VKORC1v2-Flag) (C). The results from repeat experiments for vIL-6.hSP (B and C) and vIL-6.hD1 (B) are shown to the right of the “screening” blots. In panel C, arrowheads “1” and “2” indicate Flag-tagged VKORC1v1 and VKORC1v2 protein bands.
Interactions of selected vIL-6 variants. (A) Coprecipitation binding assays were carried out essentially as outlined for Fig. 6 to test the effects of grouped point mutations within the D1 region of vIL-6 for their effects on GlucIIα and gp130 interactions. Immunoblotting analysis of coprecipitates of variant vIL-6.D1m11 is shown alongside data for binding-competent vIL-6. Dotted lines indicate the juxtaposition of lanes that were discontiguous on the original blot (i.e., deletion of intervening lanes). (B) Similar coprecipitation assays were carried out to test the binding of vIL-6.hSP, vIL-6.hD1, and vIL-6.D1m11 to StrepII-tagged UGGT1 (U1). (C) Immunoblotting analysis of coprecipitates from an analogous experiment to test the binding of vIL-6.D1m11, along with vIL-6.hSP and vIL-6.hD1, to VKORC1v2 (v2, Flag-tagged).
Assessment of profolding activities of vIL-6 variants.We next wanted to test the activities of the vIL-6 variants with respect to NHK-AAT folding. To provide a simpler and more readily quantifiable and robust assay, we generated a plasmid vector expressing NHK-AAT fused to firefly luciferase (NHK-FLuc) in addition to Renilla luciferase (RLuc), which provided a normalization control (Fig. 8A). For validation, this was cotransfected into HEK293T cells, along with the NHK-GFP expression vector, to test and compare the respective responses to vIL-6, relative to empty vector and hIL-6-KDEL negative controls. Cell lysates were processed into “soluble” and “insoluble” fractions, as before, and media were harvested for detection of secreted (fully folded) NHK-FLuc. As seen previously (Fig. 6), vIL-6 coexpression with NHK-GFP led to an increase in the proportion of the protein in the soluble fraction (Fig. 8B, left, and associated chart). Essentially the same result was obtained by measuring FLuc activities in the soluble lysate fractions (Fig. 8B, right). In contrast to the barely detectable levels of secreted NHK-GFP seen by immunoblotting (data not shown), NHK-FLuc in culture media was readily detectable by luminescence assay; it was markedly higher in vIL-6 expressing cultures than in those transfected with empty vector or hIL-6-KDEL expression plasmid (Fig. 8B, right), reflecting the ability of vIL-6 to enhance the correct folding and subsequent secretion of NHK-FLuc. These data provided validation of this assay for assessment of vIL-6 profolding activity.
Analysis of profolding activities of binding-abrogated vIL-6 variants. (A) Diagram of firefly luciferase (FLuc)-based reporter vector constructed to enable quantitation of profolding activity of vIL-6 in transfected cells. FLuc was coexpressed with Renilla luciferase (RLuc, for normalization) via internal ribosome entry site (IRES)-mediated internal translational initiation of the former. CAG, CMV enhancer-chicken β-actin promoter; H, hairpin-loop preventing ribosome read-through. (B) Validation of the reporter vector in transfected HEK293T cells, comparing profolding effects of vIL-6 as assessed by immunoblotting-determined levels of soluble versus insoluble NHK-GFP (analogous to assays shown in Fig. 5) with coexpressed FLuc activities in the soluble fraction (FLuc lysate). FLuc activities in the culture media, reflecting levels of properly folded and secreted NHK-FLuc, were also measured. FLuc activities are shown relative to those in vIL-6-expressing cells (set at 100%). CBD-fused vIL-6 and hIL-6 (KDEL-tagged) are indicated by v6-CBD/v6 and h6-CBD-K/h6, respectively. (C) Determination of vIL-6 variant relative to wild-type vIL-6 profolding activities, as determined by analysis of secreted NHK-FLuc activities in transfected HEK293T cultures. (D) Assessment of vIL-6 influence on NHK-FLuc folding (relative expression of secreted protein) as a function of gp130 depletion, mediated via gp130 mRNA-directed shRNA vector cotransfection with vIL-6-encoding or empty (vec) expression vectors. Nonsilencing (NS), untargeted shRNA expression vector was used to transfect parallel cultures. Depletion of gp130 was confirmed by immunoblotting, to detect both gp130 and vIL-6/gp130-activated STAT3 (pSTAT3), relative to total STAT3 and actin (middle panel). Functional gp130 depletion was also verified by assaying for activity of secreted Cypridina luciferase expressed from a cotransfected gp130 signaling-responsive reporter (see Materials and Methods) (chart, right). For all reporter experiments (panels B to D), transfections were conducted in duplicate, and the FLuc activities (normalized to lysate-derived RLuc activities) from these samples were averaged. Error bars show deviations of these values from the means.
The assay system was then used to quantify the relative folding activities of the binding-altered vIL-6 domain- and point-substituted variants of vIL-6 (Fig. 6 and 7). Expression and reporter vector-cotransfected HEK293T culture media and total cell lysates were harvested to measure FLuc and RLuc (normalization control), respectively. Secreted FLuc activity was assayed to provide unambiguous quantitation only of fully folded NHK-FLuc protein. The data from this experiment revealed that while vIL-6 greatly enhanced levels of secreted NHK-FLuc (>10-fold relative to empty vector and hIL-6-KDEL negative controls), the vIL-6 variants had little or no effect (Fig. 8C). Therefore, interactions of vIL-6 with GlucIIα (from vIL-6.hD1), UGGT1 (from vIL-6.D1m11) and/or gp130 (both proteins) are important for the profolding activity of vIL-6. Although vIL-6.hSP is abrogated for interactions with VKORC1v2, GlucIIα, and UGGT1, it is fully functional with respect to gp130 interaction and signaling (35), indicating that gp130 is insufficient to support this activity.
To determine whether gp130 could contribute, indirectly, to the profolding effects of vIL-6, we carried out a further NHK-FLuc-based reporter experiment in HEK293T cells pretransduced with lentiviral vectors expressing either nonsilencing (NS) or gp130-specific shRNA, previously validated in this laboratory (9). Along with the NHK-FLuc reporter and either vIL-6-expressing or empty vectors, a Cypridina luciferase (CLuc)-based STAT3 reporter (see Materials and Methods) was also cotransfected. Viral IL-6 enhanced levels of secreted NHK-FLuc in both NS shRNA-transduced and gp130-depleted cultures, to a similar degree (Fig. 8D, left). Physical and functional depletion of gp130 was evidenced by reduced levels of gp130 and vIL-6-activated STAT3 (Y705-phosphorylated, pSTAT3) and downstream CLuc expression in gp130-shRNA transduced cells (Fig. 8D, middle and right). These data indicate that gp130 signaling does not contribute significantly to the profolding activity of vIL-6; further, they implicate the functional significance of vIL-6 interactions with GlucIIα and UGGT1, abrogated along with gp130 binding and profolding activity by the hD1 and D1m11 mutations in vIL-6.
Profolding activity of vIL-6 in the context of infection.To examine whether profolding activity of virus-produced vIL-6 can be detected in infected cells, we utilized the HHV-8 BAC16 genome and recombineering to generate a vIL-6-null mutant of HHV-8, BAC16-vIL-6.TTG, in addition to a repaired version, BAC16-vIL-6.ATGrep, as a control (see Materials and Methods). Each mutation (ATG>TTG and TTG>ATG) was verified by sequencing (data not shown) and the restriction profiles of the WT, vIL-6-null and repaired bacmids were examined to verify gross integrity of the genomes (Fig. 9A). Each DNA was transfected into iSLK cells, doxycycline-inducible for immediate early RTA expression (36), cells selected in hygromycin for BAC16-containing cells, and latently infected cultures used to derive infectious virus (see Materials and Methods). Lysates of these reactivated cultures were checked for vIL-6 expression, which verified absence of vIL-6 in the ATG→TTG mutant and recovery of vIL-6 expression in the repaired (TTG→ATG, Rep) virus (Fig. 9B). Naive iSLK cells were infected with equivalent infections titers of the BAC16 viruses to derive parallel wild-type, BAC16-vIL-6.TTG, and BAC16-vIL-6.ATGrep cultures, which were subsequently transfected with the RLuc/NHK-FLuc reporter vector and then induced into lytic replication by treatment with doxycycline and sodium butyrate. After 48 h, FLuc activity was measured in the culture media, and RLuc was measured in cell lysates to normalize for any variations in transfection efficiencies. These experiments revealed that media-derived FLuc activities from cultures infected with vIL-6-null virus were greatly reduced relative to both the wild-type and repaired viruses (Fig. 9C). These data provide evidence that vIL-6 profolding activity is operative and of likely biologically significant in the context of virus-infected cells.
Effects of vIL-6 on protein folding in the context of infection. (A) The BAC16 HHV-8 genome was mutated at the vIL-6 ORF translational initiation codon (ATG→TTG) to generate a vIL-6-knockout virus. Reversion of this mutation to ATG (TTG→ATG) was done to generate a repaired virus (Rep) for use as a control in phenotypic analyses. The restriction profiles of the wild-type (WT) BAC16 and engineered genomes were indistinguishable, shown here for BamHI-HindIII digestions, demonstrating overall genome integrity following targeted recombination. The arrow indicates the position of the ORFK2-containing restriction fragment. The sizes of marker (m) bands are indicated. (B) Verification of knockout and repair of the vIL-6 ORF in BAC16-vIL-6.TTG (TTG) and BAC16-vIL-6.ATGrep (Rep), respectively. Cultures of iSLK cells infected with wild-type (WT) or each of the engineered viruses were harvested 48 h after lytic induction with doxycycline (Dox; 1.9 nM) and sodium butyrate (NaB; 1 mM) for preparation of cell extracts for immunoblotting. (C) BAC16-infected iSLK cells were transfected with the RLuc/NHK-FLuc reporter vector (see Fig. 8A), and the cultures were treated with doxycycline and sodium butyrate to induce lytic replication. Cells and media were harvested after 48 h for measurement of secreted (correctly folded) HNK-FLuc (measured from 20 μl, 1%, of media) and lysate-contained RLuc (using 20 μl, 10%, of lysates). Average FLuc activities (normalized to RLuc values) from duplicate cultures infected with each engineered virus are shown relative to average FLuc activity (set at 100%) from duplicate wild-type BAC16-infected cultures. Error bars indicate deviations of the duplicate values from the means. (D) Analogous experiments were carried out using BAC16-vIL-6.hSP (hSP) and BAC16-vIL-6.hSPrep (hSPrep) in addition to BAC16 (WT) and BAC16-vIL-6.TTG (TTG). Data, derived from biological duplicates, are presented as outlined for panel C.
To determine the involvement of vIL-6 interactions with VKORC1v2, GlucIIα, and/or UGGT1 in this activity, we generated a BAC16 genome expressing vIL-6.hSP in place of native vIL-6; vIL-6.hSP is refractory for interactions with VKORC1v2, GlucIIα, and UGGT1 (Fig. 6 and 7) but unaffected with respect to gp130 signaling (35). Wild-type BAC16, BAC16-vIL-6.hSP, and BAC16-vIL-6.TTG were used to infect iSLK cells, which were then transfected with the RLuc/NHK-FLuc reporter vector prior to lytic reactivation and measurement secreted FLuc activity, as before. Relative to FLuc activity in wild-type BAC16-infected cultures, secreted FLuc in BAC16-vIL-6.hSP-infected cultures was reduced, to levels comparable with those measured for BAC16-vIL-6.TTG (Fig. 9D, left). A follow-up experiment incorporating “repaired” BAC16-vIL-6.hSP (BAC16-vIL-6.hSPrep) as a control, along with BAC16 and BAC16-vIL-6.hSP, demonstrated hSP mutation-specific effects on secreted FLuc (Fig. 9D, right). Although the effects of the mutations on secreted FLuc in these experiments were reduced relative to that observed in Fig. 9C (for reasons unknown, but possibly due to difference in viral genome copy number and/or reactivation efficiency), the data are qualitatively consistent and demonstrate the involvement of VKORC1v2, GlucIIα, and/or UGGT1 binding by vIL-6 in promotion of protein folding in infected cells.
DISCUSSION
While HHV-8 shares properties of gp130 signal transduction with cellular IL-6 proteins that have been investigated, it clearly differs with respect to its gp80 (α-subunit) independence and its inefficient secretion and ER retention and function, including via vIL-62:gp1302 complex-mediated signaling (3, 7, 8). These properties of vIL-6 are highly unusual for a cytokine and provided the rationale for further investigations of vIL-6 interactions and activities within the ER. The discovery by yeast-two hybrid assay of direct interaction of vIL-6 with the novel ER membrane protein VKORC1v2, about which nothing was known previously, and the subsequent determination of the biological importance of VKORC1v2 and VKORC1v2-vIL-6 interaction in PEL cell biology and virus productive replication (13, 16) has provided evidence of a second pathway of vIL-6 function. One consequence of vIL-6 interaction with VKORC1v2 is the suppression of proapoptotic cathepsin D by promotion of its degradation through ERAD, implicating VKORC1v2 in this cellular process, likely via direct interactions with the ERAD machinery (16).
What we demonstrate in this report is that the activities of vIL-6 involving VKORC1v2 extend beyond these initial reports to interactions and functions involving other aspects of ER quality control. In particular, the finding of vIL-6 and VKORC1v2 complexing with calnexin cycle components GlucIIα and UGGT1 and promotion of protein folding, dependent on these interactions, identifies a novel and potentially highly significant function of the viral cytokine. Indeed, we have demonstrated that this activity is detectable in the context of virus productive replication, suggesting that it could play a role in virus biology. One likely positive role is facilitating protein folding and limiting excessive UPR in productively infected cells, thereby enhancing the production of viral proteins and limiting UPR-activated proapoptotic signaling (37, 38). However, there is evidence that viruses can utilize particular aspects of UPR signaling for their own benefit while suppressing others. For example, human and mouse cytomegaloviruses utilize PERK (PKR-like ER kinase)-activated ATF4 (activating transcription factor 4) but inhibit IRE1 (inositol-requiring kinase 1) and ATF6 pathways to prevent induction of the full repertoire of UPR-regulated genes (39, 40). Both viruses inhibit IRE1 through direct interactions of their UL50 gene products with the UPR signaling protein (41). The unfolded protein sensor and UPR regulator BiP/GRP78, which is sequestered away from inhibitory interactions with ATF6, IRE1, and PERK by binding to unfolded proteins, is critical for virion assembly and egress (42). In the case of herpes simplex virus (HSV), UPR is blocked at early stages of infection, and while activated ATF6 is present at later stages, associated target gene induction is lacking (43). Interestingly, the HSV immediate early protein ICP0 is induced by UPR signaling, indicating that this may be a sensing mechanism to allow for virus reactivation in response to ER stress. Similarly, the ZTA and RTA immediate early proteins of Epstein-Barr virus and HHV-8, respectively, also have been reported to be induced by ER stress signaling, lytic reactivation from latency being triggered by UPR-inducer thapsigargin (44, 45). It is likely that in HHV-8 and other viruses, UPR signaling and other stress responses serve both positive and negative roles with respect to virus productive replication but that limiting excessive activation of these pathways is important for successful replication (30). It is noteworthy that although activation of the DNA-damage response can lead to apoptosis, this pathway has been shown to be induced by EBV, HHV-8, and other herpesviruses and to be important for virus productive replication and the establishment and maintenance of latency (46–48). While we expect that the same is true of UPR, one consequence of which is to induce the expression of profolding chaperones in addition to translational inhibition and, if unchecked, induction of proapoptotic pathways, further investigation is required to tease out the pro- and antiviral effects of such signaling and the viral mechanisms that help limit the latter. The profolding activity of vIL-6 would be expected to provide one mechanism of UPR suppression by limiting the unfolded protein load in infected cells. However, how significant this is in the context of lytic replication (or latency, in vIL-6-expressing PEL cells) remains to be determined.
The data reported here have identified new interaction partners of vIL-6 and VKORC1v2, but the precise nature of these interactions has not been determined. In view of the fact that both vIL-6 and VKORC1v2 can interact with GlucIIα and UGGT1, that complexes between all four proteins are evident (from serial coprecipitation), that large complexes containing each of these proteins were detected (by density-gradient sedimentation), and that hIL-6 signal peptide domain-substituted vIL-6 was compromised for binding to VKORC1v2, GlucIIα, and UGGT1 (but not gp130 [35], demonstrating overall structural integrity), it is possible that vIL-6 and VKORC1v2 do not interact directly with GlucIIα and UGGT1 and that other cellular proteins are involved. The only direct interaction that has been identified, via yeast two-hybrid assay, is that between vIL-6 and VKORC1v2. Whether this direct binding occurs in the context of GlucIIα and UGGT1 cointeractions has not been determined, however. Further, we do not know how vIL-6 modulates UGGT1- and GlucIIα-containing complexes to enhance protein folding, although it is clear that both VKORC1v2 and UGGT1 are required functionally (from gene knockout experiments) and that vIL-6 interaction with GlucIIα is also likely to be involved (since GlucIIα binding-refractory vIL-6.D1m11 is functionally defective). We speculate that the net effect of vIL-6 is to promote monoglucosylation of nascent-protein N-glycans, following GlucII-mediated deglucosylation in calnexin cycle processing, resulting in enhanced retention of folding intermediates within folding complexes and prevention of their exit to ERAD. If this is correct, we would expect vIL-6 activation of UGGT1 activity and/or inhibition of GlucII activity within cocomplexes of vIL-6, VKORC1v2, and these enzymes. Future studies are planned to test this model and to identify other components of the functional complex(es). Such complexes would be expected to form naturally, in the absence of vIL-6, but to be regulated by the viral cytokine.
In summary, the presented data have identified new interactions of vIL-6 and VKOCR1v2 with calnexin cycle enzymes GlucII and UGGT1 and the involvement of these proteins and interactions in newly identified profolding activity of vIL-6. Our previous identification of vIL-6 interaction with calnexin may be related to the present findings, indicating that vIL-6 and VKORC1v2 associate with calnexin cycle-related complexes to promote protein folding. This activity of vIL-6 is operative in infected cells; however, further investigation is required to determine its particular contribution to virus biology.
MATERIALS AND METHODS
Cell culture and gene transduction.HEK293T cells and iSLK cells (36) were grown in Dulbecco modified Eagle medium supplemented with 10% fetal bovine serum (FBS). For transfection of HEK293T cells, cultures were passaged 12 to 24 h prior to transfection to produce ∼80% confluent monolayers, to which were added polyplexes formed by plasmid DNA and the cationic polymer linear-polyethylenimine (PEI; molecular weight 25,000; Polysciences, catalog number 593215), according to published procedures (49). FuGENE HD (Promega, catalog number E2311) was used for transfection of iSLK cells. For transduction of HEK293T cell with lentiviral vectors for gene deletion (CRISPR-mediated) or depletion (shRNA-mediated), 106 cells were mixed with concentrated lentivirus in the presence of 5 μg/ml Polybrene, followed by incubation at 37°C for 5 h; the inoculum was replaced with fresh medium, and the cells were incubated for 2 days before puromycin selection or experimental use. Cas9/gRNA-transduced cells were selected in medium containing puromycin (1 μg/ml) for 7 days, prior to limiting-dilution cloning to obtain VKORC1v2 and UGGT1 gene-deleted cell lines. PCR amplicons of the former were sequenced to identify altered alleles (which contained 17- and 53-nucleotide deletions at the gRNA target site). For the latter, cell lines were screened by immunoblotting for UGGT1; one validated UGGT1-null cell line was selected for experimental use.
Plasmids and lentiviral vectors.pSG5-based eukaryotic expression vectors for chitin-binding domain (CBD) fusions of wild-type and domain substitution variants of vIL-6, VKORC1v1, VKORC1v2, and KDEL ER-retention motif-linked hIL-6 have been described previously (8, 13, 35). Twin-StrepII-tagged vIL-6 (vIL-6-TS) in pcDNA3.1 also has been reported (16). PCR-amplified coding sequences of vIL-6, vIL-6.hSP, vIL-6.hD1, VKORC1v1, and VKORC1v2 were also cloned into pCEBZ-RFP-CBD(C) vector (15) between the AgeI and BamHI restriction sites to generate CBD fusion proteins driven by the cytomegalovirus (CMV) promoter. hIL-6-CBD-KDEL coding sequences were PCR amplified from pSG5-hIL-6-CBD-KDEL and used to replace the red fluorescent protein (RFP) ORF in the pCEBZ-RFP vector (15), using the AgeI and EcoRI sites for RFP excision and cloning the PCR product. A vector expressing vIL-6-HA was generated by substituting the RFP coding sequence in pCEBZ-RFP-HA, generated by insertion of HA-encoding double-stranded (ds) oligonucleotides between the BamHI and EcoRI sites downstream of RFP, with the vIL-6 open reading frame (ORF), utilizing the AgeI and BamHI in pCEBZ-RFP-HA. Coding sequences for the tandem affinity tag Twin-StrepII+Flag (SF) were cloned into pCEBZ as a ds-oligonucleotide between the BamHI and XhoI sites downstream of RFP ORF sequences to generate pCEBZ-SF(C); vIL-6 and VKORC1v2 coding sequences were cloned between the AgeI and BamHI of this vector to generate expression vectors for the respective SF-tagged proteins. For construction of an expression vector for SF-VKORC1v1, SF-coding sequences were PCR amplified with an added ATG translation initiation codon, the PCR product cleaved at the 3′ end with BamHI, this ligated to VKORC1v1-coding sequences via a 5′-located BglII site, and SF-VKORC1v1 ORF sequences PCR amplified from the ligation mixture and cloned between the AgeI and EcoRI sites of pCEBZ-RFP (replacing the RFP ORF). Vector pCEBZ-S(C) (15) was used to construct S-tag-fused GlucIIα expression plasmid pCEBZ-GlucIIα-S via ORF cloning between the AgeI and BamHI sites of pCEBZ-S(C). To generate an expression vector for StrepII-tagged UGGT1, cDNA sequences for mature (signal-sequence-cleaved) UGGT1 were cloned downstream of sequences encoding the hIL-6 signal peptide (hSP) and StrepII tag, to maintain the functional integrity of the C-terminal ER retention signal of UGGT1. First, hSP sequences, encoding the first 46 amino acids of hIL-6, were cloned between the AgeI and BamHI sites of pCEBZ. Then, sequences encoding a single StrepII tag flanked 3′ by a NheI site were cloned as a ds-oligonucleotide between the BamHI and XhoI sites of pCEBZ-hSP to generate pCEBZ-hSP-StrepII. Finally, the CEBZ-hSP-StrepII-UGGT1 vector was generated by cloning PCR-amplified mature UGGT1 cDNA sequences downstream of StrepII through utilization of the NheI and XhoI restriction sites. For expression of CBD-tagged HHV-8 gH and gL, the respective glycoprotein ORF sequences were amplified from HHV-8 BAC36 (50) and cloned into pCEBZ-CBD(C) (15); gH and gL ORFs were inserted between the NotI and BamHI and AgeI and BamHI sites, respectively. The NHK-GFP expression plasmid (51) was kindly provided by Ron R. Kapito, Stanford University. To generate an expression vector for firefly luciferase (FLuc)-fused NHK, we utilized plasmid pFLAG-XBP1u-FLuc (Addgene plasmid 31239 [52]), as a template for PCR amplification of Renilla luciferase-IRES-FLAG-XBP1u-FLuc sequences, which were cloned into the EcoRI site of pCAGEBZ (pCEBZ derivative containing the CAG promoter [53] in place of the CMV promoter, between XbaI and EcoRI sites) using Gibson assembly (New England BioLabs, catalog number E5510S) to generate pCAGEBZ-RHi-Xu-FLuc. Then, XBP1u sequences in pCAGEBZ-RHi-Xu-FLuc were replaced by the NHK ORF, between XhoI and EcoRI, resulting in pCAGEBZ-RHi-NHK-FLuc, the NHK-luciferase reporter vector. Lentiviral vectors pYNC352-NS and pYNC352-gp130sh1 expressing nonsilencing (NS, control) and gp130 mRNA-directed shRNAs were described previously (8, 9). To generate lentiviral vectors for VKORC1v2 and UGGT1 gene deletion, annealed oligonucleotides for VKORC1v2-specific CRISPR gRNA (5′-GCGGCTCGCTCTTTGCCTGA-3′) and UGGT1 gRNA (5′-TTTTGAGTCGGCCTTTACTG-3′) were cloned between the two BsmBI sites of the lentiCRISPR-v2 vector (Addgene plasmid 52961 [54]). STAT3-responsive rat α2-macroglobulin and minimal thymidine kinase promoter sequences in the plasmid pα2M-CAT (55) were released by SalI and BglII digestion and cloned between the XhoI and BamHI sites of pMCS-Cypridina Luc (Thermo Scientific, catalog number 16149) to generate IL-6 signaling reporter vector pα2M-Cluc.
Lentivirus production.Lentiviral vectors for shRNA-mediated mRNA depletion or CRISPR-mediated mutagenesis (12 μg) were cotransfected with gag-pol and vesicular stomatitis virus G protein expression plasmids (9 and 3 μg, respectively) into 80% confluent monolayers of HEK293T cells in 75-cm2 flasks. At 2 days posttransfection, virus was collected from the medium by centrifugation in an SW28 rotor at 25,000 rpm for 2 h at 4°C. Virus pellets were resuspended in 10 ml of RPMI 1640 supplemented with 10% FBS. Infectious titers of shRNA lentiviruses were established by visualizing GFP expression in HEK293T cells infected with different doses of lentivirus inoculum prior to experimental use of the viral vectors.
Affinity precipitation.For CBD affinity precipitation experiments, generally 10 μg of pSG5-based CBD-tagged protein expression vector or 5 μg of equivalent pCEBZ-based expression vector was cotransfected with 5 ug of each “prey” protein expression vector into HEK293T cells in 10-cm dishes. Cells were harvested at 2 days posttransfection for preparation of cell lysates via suspension in lysis buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, and 0.1% DDM [n-dodecyl β-d-maltoside; Sigma, catalog number 4641]), supplemented with protease inhibitor cocktail (Sigma, catalog number P8340). Lysates were incubated with chitin beads (New England BioLabs, catalog number S6651) at 4°C overnight. Then, beads and associated proteins were pelleted by microcentrifugation and washed in lysate buffer by repeated resuspension and pelleting prior to heat/denaturing-effected release of proteins and analysis by SDS-PAGE and Western blotting. For triple precipitation experiments, plasmid vectors encoding VKORC1v2-SF, vIL-6-HA, GlucIIα-S, and StrepII-UGGT1 (5 μg of DNA for each vector per 10-cm dish) were cotransfected into HEK293T cells using the PEI method. Omission of individual vectors provided controls; the total amount of DNA was maintained using empty vector in place of the omitted expression plasmid. At 2 days posttransfection, the cells were harvested, and lysates were prepared as described above. For the first precipitation, of VKORC1v2-SF, the cell lysates were incubated with anti-FLAG M2 magnetic beads (Sigma, catalog number M8823) with gentle rotation at 4°C overnight. After three washes in lysis buffer, VKORC1v2-SF and associated proteins were released from the magnetic beads with two applications of 100 μl of lysis buffer containing 3×FLAG peptide (150 ng/μl; Sigma, catalog number F4799). The combined eluates were diluted with lysis buffer to a total volume of 1 ml and incubated with 30 μl of anti-HA magnetic beads (Pierce, catalog number 88837) at 4°C overnight to pull down vIL-6-HA. The precipitates were washed three times with lysis buffer, and magnet-captured, bead-bound vIL-6-HA was eluted twice with 100 μl of HA peptide (Thermo Fisher, catalog number 26184) at 1 μg/μl in lysis buffer. After dilution to a final volume of 1 ml with lysis buffer, the eluates were subjected to a third precipitation with S-protein agarose (EMD-Millipore, catalog number 69704) to sediment GlucIIα-S. After incubation at 4°C overnight, the beads were washed three times with lysis buffer, and bound proteins were denatured at 95°C in SDS sample buffer (50 mM Tris, 4% SDS, 4% β-mercaptoethanol, 12% glycerol) and then analyzed by immunoblotting. Similar procedures were used for dual precipitation of vIL-6-Flag and GlucIIα-S and subsequent analysis of coprecipitated StrepII-UGGT1. To prepare samples for mass spectrometry, 1 μg of vIL-6-SF, SF-VKORC1v1, or VKORC1v2-SF expression plasmid was mixed with 5 μg of carrier DNA (pBluescript) and transfected into HEK293T cells in a 10-cm dish. At 2 days posttransfection, the cells were lysed in lysis buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 1% digitonin) supplemented with protease inhibitor cocktail (Sigma, catalog number P8340). The supernatants were incubated with Strep-Tactin Superflow Plus beads (Qiagen, catalog number 1057979) at 4°C overnight. After precipitation and three rinses in washing buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 0.1% digitonin), the proteins were released with elution buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 0.1% digitonin, 2.5 mM desthiobiotin). The Strep-Tactin bead eluates were then incubated with 20 μl of anti-FLAG M2 magnetic bead suspension (Sigma, catalog number M8823) at 4°C overnight. The beads were washed three times with washing buffer, followed by two washes in 50 mM ammonium bicarbonate (pH 8.0). Binding proteins were released with 40 μl of elution buffer (100 μg/ml FLAG peptide, 0.05% RapiGest SF in 50 mM ammonium bicarbonate [pH 8.0]), applied for 20 to 30 min at room temperature with gentle shaking. The protein samples were then subjected to trypsin digestion and analysis by MS.
Density gradient ultracentrifugation analysis of VKORC1v2 complexes.VKORC1v2-SF, vIL-6-TS, StrepII-UGGT1, and GlucIIα-S expression plasmids (5 μg of each) were cotransfected into HEK293T cells in 10-cm dishes. At 48 h posttransfection, the cells were harvested and lysed in lysis buffer (50 mM Tris HCl [pH 7.5], 150 mM NaCl, 5 mM EDTA, 0.1% DDM), supplemented with protease inhibitors. VKORC1v2-Flag was precipitated with FLAG M2 magnetic beads, bead-protein complexes were washed, and bound proteins were released essentially as outlined above. Released proteins were loaded on top of an iodixanol discontinuous gradient, which was prepared using OptiPrep (Sigma, catalog number D1556) layered successively at 50, 40, 30, 20, 10, and 5%. Centrifugation was carried out at 25,000 rpm in an SW41 rotor for 3 h at 4°C. Fractions from the gradient were recovered as 14 equal fractions from the top to the bottom of the gradient for analysis by SDS-PAGE and immunoblotting for the epitope-tagged VKORC1v2-interaction partners.
Protein folding assays and reporters.NHK-GFP plasmid vector (0.2 μg/well of a 6-well plate) was cotransfected with expression vectors (0.5 μg/well) for vIL-6-CBD or hIL-6-CBD-KDEL, or empty vector, into HEK293T cells. Cells were lysed with NP-40-containing lysis buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA, and 0.2% NP-40, supplemented with protease inhibitors) for 1 h on ice. Lysates were then centrifuged at 13,500 × g for 15 min at 4°C to derive “insoluble” (pellet) and “soluble” (supernatant) fractions; the former were washed twice with lysis buffer containing 0.1% NP-40 and material solubilized by 95°C incubation for 10 min in SDS-containing sample buffer (50 mM Tris, 4% SDS, 4% β-mercaptoethanol, 12% glycerol) prior to SDS-PAGE and immunoblotting. For experiments involving coexpression of HHV-8 gL and gH, per well of a 6-well tissue culture plate, 0.5 μg of gL-CBD and gH-CBD plasmids were cotransfected with 1 μg of empty vector or vIL-6 expression vector into HEK293T cells. At 48 h posttransfection, media and cells were collected. Cell lysates were made as described above. For media, 30-μl portions of chitin beads were added to each medium sample, followed by incubation at 4°C overnight. Chitin bead-captured gL-CBD was washed three times with washing buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5 mM EDTA, 0.1% NP-40) and denatured and released from chitin beads by incubation at 95°C with SDS-containing sample buffer prior to SDS-PAGE and immunoblotting. For folding assays involving NHK-luciferase (firefly, FLuc), 0.5 μg of NHK-FLuc expression plasmid was cotransfected with 1.5 μg of hIL-6-CBD-K, vIL-6-CBD or its variants, or empty vector into HEK293T cells in 6-well plates using PEI. Cells and media were harvested at 24 h posttransfection; the cells were lysed in NP-40 lysis buffer (see above), and samples of lysates and media were analyzed by luciferase assays for Renilla (normalization control, intracellular) and firefly luciferase (NHK-linked, intracellular and secreted) using the Promega Dual-Luciferase reporter assay system (catalog number E1910) and a Promega GloMax multidetection system. For experiments including the IL-6 signaling reporter, pα2M-CLuc, culture media were analyzed for secreted CLuc activity using a Cypridina Luciferase Glow assay kit (Thermo Scientific, catalog number 16170).
SDS-PAGE, immunoblotting, and antibodies.Tricine-SDS-PAGE and electrophoretic transfer to nitrocellulose membranes was carried out according to standard procedures. Membranes were blocked in phosphate-buffered saline (PBS)-Tween 20 buffer (PBS-T; 137 mM NaCl, 2.7 mM KCl, 4.3 mM Na2HPO4, 1.4 mM KH2PO4, 0.1% Tween 20) containing 5% nonfat milk prior to the addition of primary antibody (0.1 to 1.0 μg/ml), followed by incubation at 4°C overnight. PBS-T-washed membranes were incubated with horseradish peroxidase-conjugated anti-rabbit, or anti-mouse secondary antibodies in PBS-T containing 5% nonfat milk and washed with PBS-T; membrane-bound antibody was detected by chemiluminescence assay and visualized using a GeneGnome imager (Syngene). Commercially acquired antibodies used for immunoblotting and immunoprecipitation were as follows: CBD (New England BioLabs, catalog no. E8034S), S-tag (Cell Signaling Technology, catalog no. 8476S), GFP (Santa Cruz Biotechnology, catalog no. sc-8334), Strep II (Qiagen, catalog no. 34850), Flag (Sigma, catalog no. F7425), hemagglutinin (HA; Sigma, catalog no. H6908), β-actin (Sigma, catalog no. A5316), BiP (BD Biosciences, catalog number 610978), STAT3 (Santa Cruz Biotechnology, catalog number sc-482), pSTAT3 (Cell Signaling Technologies, catalog number 9131S), and gp130 (Santa Cruz Biotechnology, catalog number sc-655). Rabbit antiserum to vIL-6, generated in this laboratory, has been described previously (2).
Generation of HHV-8 recombinants.Genetically engineered viral genomes were generated using the BAC16 HHV-8 genome (36) as a backbone for Red recombinase-mediated recombination in the bacterial strain GS1783 (56), expressing Red and (single-cutting) I-SceI restriction enzyme in separately inducible fashion. Two-step recombination involving a homologous-end, site-specific insertion of a kanamycin resistance (Kanr) gene, along with mutation-containing sequences, and then intragenic recombination of Kanr gene-flanking duplicated sequences to remove the Kanr gene was carried out using previously established techniques (56, 57). Each step was checked by PCR-mediated detection of inserted and excised sequences, respectively, in BAC16 DNA from kanamycin-resistant (step 1) and kanamycin-sensitive (step 2) bacterial clones. To generate vIL-6-null HHV-8 genomes, the Kanr gene was PCR amplified from pEPkan-S (57) using initiator-ATG mutagenic (ATG→TTG) primers containing internally common and flanking unique sequences spanning the ATG codon, designed as outlined by Tischer et al. (57). Forward and reverse primers were as follows: F1 (5′-gctcgggcaccaccggctagtgtatgccgcgttagcagccTtgtgctggttcaagttgtggAGGATGACGACGATAAGTAGGG-3′; underlining and boldfacing indicate an introduced mutation; capital letters indicate sequences homologous to the Kanr gene-flanking sequence) and R1 (5′-gaaccgaccagcaagagagaccacaacttgaaccagcacaAggctgctaacgcggcatacactCAACCAATTAACCAATTCTGATTAG-3′). To control for potential functionally significant genetic alterations introduced elsewhere in the mutated BAC16 genome (BAC16-vIL-6.TTG), the vIL-6-knockout viral genome was repaired by reintroduction of the wild-type initiator codon (TTG→ATG). The PCR primers were as follows: F2 (5′-gctcgggcaccaccggctagtgtatgccgcgttagcagccAtgtgctggttcaagttgtggAGGATGACGACGATAAGTAGGG-3′) and R2 (5′-gaaccgaccagcaagagagaccacaacttgaaccagcacaTggctgctaacgcggcatacactCAACCAATTAACCAATTCTGATTAG-3′). To generate the vIL-6 domain substitution variant BAC16-vIL-6.hSP and its repair, Kanr gene-containing vIL-6.hSP and vIL-6 ORF transfer vectors were generated for whole-ORF substitutions, essentially as described previously (57). Mutagenesis in all BAC16 recombinants was confirmed by sequencing. Overall integrity of the altered BAC16 genomes was checked by restriction digestion and comparison of the agarose gel-fractionated restriction profiles against those of native BAC16. BAC16 genomes were introduced into iSLK cells by using FuGENE HD-mediated transfection, and transduced cells were selected by treatment of the cultures with hygromycin (2.3 mM) for 3 weeks. Virus was induced from these culture by dual, continuous treatment with doxycycline (1.9 nM) and sodium butyrate (1 mM), and virus-containing medium was harvested after 5 days to provide virus stocks. Virus titers were established by inoculation of naive iSLK cells with different dilutions of the virus stocks to establish appropriate and equivalent infectious doses to use in subsequent experiments.
ACKNOWLEDGMENTS
We thank R. R. Kapito for providing the NHK-GFP expression vector, J. U. Jung for providing HHV-8 BAC16 and iSLK cells, and G. Smith and N. Osterrieder for providing the recombineering bacterial strain GS1783. Plasmids pFLAG-XBP1u-FLuc, plentiCRISPR-v2, and pEPkan-S were generated by the laboratories of J. Horng, F. Zhang, and N. Osterrieder, respectively, and were obtained from Addgene.
This study was supported by NIH grant R01-CA76445 to J.N.
FOOTNOTES
- Received 9 June 2017.
- Accepted 31 August 2017.
- Accepted manuscript posted online 6 September 2017.
- Copyright © 2017 American Society for Microbiology.
