|Home | About | Journals | Submit | Contact Us | Français|
Epstein-Barr virus (EBV), etiologically linked with human B-cell malignancies and nasopharyngeal carcinoma (NPC), establishes three types of latency that facilitate its episomal genome persistence and evasion of host immune responses. The innate inflammasome responses recognize the pathogen-associated molecular patterns which lead into the association of a cytoplasmic sensor such as NLRP3 and AIM2 proteins or nuclear interferon-inducible protein 16 (IFI16) with adaptor ASC protein (apoptosis-associated speck-like protein with a caspase recruitment domain) and effector procaspase-1, resulting in active caspase-1 formation which cleaves the proforms of inflammatory interleukin-1β (IL-1β), IL-18, and IL-33 cytokines. Whether inflammasome responses recognize and respond to EBV genome in the nuclei was not known. We observed evidence of inflammasome activation, such as the activation of caspase-1 and cleavage of pro-IL-1β, -IL-18, and -IL-33, in EBV latency I Raji cells, latency II NPC C666-1 cells, and latency III lymphoblastoid cell lines (LCL). Interaction between ASC with IFI16 but not with AIM2 or NLRP3 was detected in all three latencies and during EBV infection of primary human B cells. IFI16 and cleaved caspase-1, IL-1β, IL-18, and IL-33 were detected in the exosomes from Raji cells and LCL. Though EBV nuclear antigen 1 (EBNA1) and EBV-encoded small RNAs (EBERs) are common to all forms of EBV latency, caspase-1 cleavage was not detected in cells expressing EBNA1 alone, and blocking EBER transcription did not inhibit caspase-1 cleavage. In fluorescence in situ hybridization (FISH) analysis, IFI16 colocalized with the EBV genome in LCL and Raji cell nuclei. These studies demonstrated that constant sensing of latent EBV genome by IFI16 in all types of latency results in the constitutive induction of the inflammasome and IL-1β, IL-18, and IL-33 maturation.
Epstein-Barr Virus (EBV; HHV-4), a gamma-1 human herpesvirus, is a successful pathogen that infects more than 95% of individuals worldwide by adulthood. Human B lymphocytes and epithelial cells are two major targets of EBV although it can also infect a variety of cell types, such as T cells, NK cells, smooth muscle cells, and follicular dendritic cells (1–3). EBV is etiologically associated with a number of human diseases which include (i) benign self-limiting lymphoproliferative infectious mononucleosis, (ii) B-cell lymphoproliferative Burkitt's lymphoma (BL), Hodgkin and non-Hodgkin lymphomas (HLs), posttransplant lympho-proliferative disorders (PTLD), (iii) nasopharyngeal carcinoma (NPC), and some forms of gastric carcinoma (1).
Like other herpesviruses, EBV establishes a lifelong infection in the host by establishing a latent infection in the infected cell nuclei, with periodic reactivation leading into the lytic cycle and progeny virus formation (4). EBV possesses a 175-kb double-stranded linear DNA genome which circularizes after entry into the infected cell nuclei. EBV in vitro infection of human B cells leads into cellular activation, proliferation, and outgrowth of transformed lymphoblastoid cell lines (LCLs). EBV expresses several of its genes during latency. EBV nuclear antigens (EBNAs) are encoded during latency from several alternatively spliced primary transcripts to form EBNA1, EBNA2, EBNA3A, EBNA3B, EBNA3C, and EBNA leader protein (EBNA-LP). The latent membrane proteins (LMPs), LMP1, LMP2A, and LMP2B, are expressed from individual promoters. EBV also expresses noncoding RNAs such as the abundant nonpolyadenylated 167- and 173-bp EBER-1 and EBER-2, respectively, as well as a number of viral microRNAs (miRNAs) during latency. These gene products mediate several functions, such as the maintenance and replication of latent episomal genome and methods to overcome apoptosis, autophagy, transcriptional restriction, and lytic cycle, as well as host intrinsic, innate, and adaptive immune responses.
Three types of latency programs, known as latency I, II, and III, are exhibited in EBV-infected cells, and each latency program leads to the production of a limited, distinct set of viral proteins and viral RNAs depending upon promoter usage (5). All three latency programs are evident in B cells, and only latency II is shown in epithelial cells (5–7). Following initial infection of a naive B cell, 10 latent transcripts encoding EBNA1, EBNA2, EBNA3A, EBNA3B, EBNA3C, EBNA-LP, LMP1, LMP2A, LMP2B, and EBV-encoded small RNAs (EBERs) are expressed in latency III to induce the proliferation of the latently infected cell (5). As the latently infected cells move through the germinal center (from centroblasts to centrocytes) and are subjected to increased immune selection, only the EBNA1, EBNA-LP, LMP1, LMP2A, LMP2B, and EBERs (latency II) are expressed (5). As the infected cell differentiates into a memory B cell, only EBNA1 and EBERs (latency I) are expressed. EBV latency 0, defined as the lack of viral gene expression, is found in nondividing B cells, while latency I is observed in BL and BL-derived cell lines, as well as in memory B cells in a healthy host (5). In contrast, latency II is detected in undifferentiated NPC, EBV-associated gastric carcinoma, HL, and T-cell lymphomas, while latency III is seen in B-cell lymphomas associated with immunosuppression and in vitro immortalized lymphoblastoid cell lines (8).
Latent EBV infection is believed to be controlled by humoral immunity, NK cells, cytotoxic T cells, and the interferon (IFN) responses. Hence, it is not surprising that deterioration of the host immune system (immune suppression, HIV-1 infection, etc.) leads into unchecked proliferation of EBV latently infected cells (3, 9). Signature cytokines have been reported for different latencies of EBV (10), and the EBV-positive (EBV+) BL cell line and LCLs produce an array of cytokines at various levels, including interleukin-6 (IL-6), IL-8, IL-10, monocyte chemoattractant protein 1 (MCP-1), macrophage-derived chemokine (MDC), tumor necrosis factor alpha (TNF-α), TNF-β, and others (10). Some of these cytokines can be detected in sera from patients of EBV-associated malignancies (10). Chronic inflammation triggered by EBV probably contributes to tumor cell proliferation, progression, and inhibition of apoptosis (11). A better understanding of the regime of inflammatory cytokines and their regulation by EBV latent infection will likely reveal viral strategies to subvert immune response pathways and provide avenues to control EBV infection and pathogenesis.
From entry to establishment of latency, EBV must overcome germ line-encoded innate immunity sensing mechanisms such as pathogen recognition receptors (PRRs) that include Toll-like receptors (TLRs), Nod-like receptors (NLRs), retinoic acid-inducible gene-I (RIG-I)-like receptors (RLRs), scavenger receptors (SRs), DNA sensing AIM2-like receptors (ALRs), and DNA-dependent activator of IFN regulatory factors (DAI) triggering type I IFN responses (12). Stimulation of virus-sensing pathways also leads to the expression of proinflammatory cytokines, including IL-1β and IL-18, which contribute to the clearance of viruses at multiple levels. Recently, IL-33 was also found to play an important role in several chronic inflammatory disorders, including asthma, rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), and anaphylactic shock (13).
IL-1β, IL-18, and IL-33 are synthesized as biologically inactive proforms (13, 14) whose maturation requires caspase-1-mediated proteolytic processing. Caspase-1 is in turn activated by an intracellular multiprotein complex called the inflammasome, which assembles in the cytosolic compartment during microbial infection as well as under a variety of other conditions (15). This complex involves sensors and adaptor and executioner molecules. The sensors identify and interact with a variety of pathogen-associated molecular patterns (PAMPs) including bacteria, bacterial products, DNA, RNA, and RNA/DNA (pox) viruses replicating in the cytoplasm. They also identify and interact with danger-associated molecular patterns (DAMPs) such as reactive oxygen species (ROS), cation flux, uric acid, silica, aluminum salts, and asbestos (15). These sensor molecules then bind to an adaptor molecule known as ASC (apoptosis-associated speck-like protein containing a caspase recruitment domain [CARD]) via homotypic interaction of their pyrin domains. Formation of this complex then recruits the executioner procaspase-1 via homotypic interaction of the adaptor molecule CARD leading to autocleavage of caspase-1, which in turn cleaves many proproteins such as pro-IL-1β, pro-IL-18, and pro-IL-33 into their active forms of inflammatory cytokines.
Various inflammasome-related sensors have been reported, including members of the Nod-like receptor (NLR) family such as NLRP1 (Nod-like receptor containing LRR and pyrin domain), NLRP3, NLRP6, NLRP12, and NLRC 4 (Nod-like receptor containing LRR and caspase recruitment domain). Among these, NLRP3 is the most well studied molecule (15). In addition, RLRs or ALRs have also been found to play a role as sensors for viral RNA or DNA, respectively. Infection of vesicular stomatitis virus (VSV; RNA virus) triggers the RIG-I inflammasome where double-stranded RNA (dsRNA) or 5′ triphosphate RNA binds to RLRs and triggers signaling via CARD-CARD interactions between the helicase RIG-I/MDA5 and the adaptor protein IPS-1 (16). The ALR member AIM2 is activated and forms the inflammasome complex by the transfected DNA and by the DNA of pox viruses replicating their DNA in the cytoplasm (17–19). The other member of the ALR family, interferon-inducible protein 16 (IFI16), was reported to sense a 60-mer transfected DNA of herpes simplex virus 1 (HSV-1), leading to type I interferon production via the STING (stimulator of IFN genes) pathway (20).
Previously, the PAMPs and DAMPs were largely known to be sensed by innate immunity only in the cytoplasm. Recently, we have shown that IFI16 was activated upon sensing of EBV-related gamma-2 herpesvirus Kaposi's sarcoma-associated herpesvirus (KSHV) DNA inside the nucleus of infected endothelial and B cells (21, 22). The 729-amino-acid IFI16 protein is constitutively expressed in the nuclei of several cell types. It binds to single-stranded DNA (ssDNA), double-stranded DNA (dsDNA), and RNA; interacts with p53, retinoblastoma protein (RB), and transcription factors; binds to the G-rich DNA in promoters; and functions as a transcriptional repressor (23–25). Though EBV's ability to establish different types of latency has been known for many years, whether the persistent EBV genome is recognized by the host innate sensing mechanism and whether such recognition induces an innate immune response are not known. Our studies here demonstrate for the first time that EBV latency I-, II-, and III-containing cells constantly recognize the presence of EBV genome and induce a constitutive IFI16-mediated inflammasome and cleavage of procaspase-1, IL-1β, IL-18, and IL-33.
An EBV latency III-positive (III+) lymphoblastoid cell line (LCL; a gift from Lindsey Hutt-Fletcher, University of Louisiana, Shreveport, LA), EBV latency I+ Raji cell line (Burkitt's lymphoma cells), and EBV-negative Burkitt's lymphoma Ramos and BJAB cell lines were grown in RPMI 1640 medium-Glutamax (Gibco Life Technologies, Grand island, NY) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Atlanta Biologicals, Inc., Lawrenceville, GA) and 1% PenStrep antibiotics (Gibco Life Technologies). BJAB cells expressing only EBV EBNA1 (a gift from Bill Sugden, McArdle Laboratory for Cancer Research, University of Wisconsin—Madison, Madison, WI) were grown in a similar medium with 1 μg/ml puromycin (Santa Cruz Biotechnology, Santa Cruz, CA). The EBV latency II+ epithelial C666-1 cell line is a subclone of its parental cell line C666 that was derived from an undifferentiated nasopharyngeal carcinoma (NPC) xenograft of southern Chinese origin (26). The C666-1 cells were cultured in RPMI 1640 medium containing 10% heat-inactivated FBS and 1% PenStrep antibiotics. The control epithelial 184B5 cell line (ATCC CRL-8799) was established from normal mammary tissue, obtained from a reduction mammoplasty, which was grown in advanced Dulbecco's modified Eagle's medium (DMEM)-F12 medium (Gibco Life Technologies) with 5% FBS and antibiotics. All the cell types used in the present studies were routinely tested for mycoplasma by a Lonza MycoAlert kit (LT37-618; Lonza, Walkersville, MD), as per the manufacturer's instructions, and were found to be negative.
Rabbit polyclonal anti-human IL-1β and caspase-1 antibodies were from Millipore, Billerica, MA, and Invitrogen Corporation, Carlsbad, CA, respectively. Rabbit polyclonal anti-human IL-1β antibody specific to processed form of IL-1β was from Cell Signaling Technology, Beverly, MA. Goat polyclonal antibodies against human ASC and TMS1 were from Ray Biotech, Norcross, GA. Mouse monoclonal antibodies against ASC and human IL-18 were from MBL International, Woburn, MA. Mouse anti-Alix antibody was from Cell Signaling Technology, Beverly, MA. Rabbit and mouse antibodies against human IFI16 were from Santa Cruz Biotechnology, Inc. Rabbit anti-AIM2 and anti-Rab27a and mouse anti-NLRP3, anti-Tsg101, anti-TATA-binding protein (TBP), and anti-EBV nuclear antigen 1 (EBNA1) antibodies were from Abcam Inc., Cambridge, MA. Mouse anti-AIM2 was from Abnova Taipei City, Taiwan. Anti-rabbit, anti-goat, and anti-mouse antibodies linked to horseradish peroxidase, Alexa Fluor-488, and Alexa Fluor-594 were from KPL, Inc., Gaithersburg, MD, or Molecular Probes, Eugene, OR. Protein G-Sepharose 4 Fast Flow beads were from GE Healthcare BioSciences Corp., Piscataway, NJ. A cytoplasmic nuclear extract kit was from Active Motif, Carlsbad, CA. RNA polymerase III (Pol III) inhibitor [557404 InSolution; N-(1-(3-(5-vhloro-3-methylbenzo[b]thiophen-2-yl-1-methyl-1H-pyrazol-5-yl))-2-chlorobenzenesulfonamide,2-chloro-N-(3-(5-chloro-3 methylbenzo[b]thiophen-2-yl)-1-methyl-1H-pyrazol-5-yl)benzene sulfonamide] was from EMD Millipore, Billerica, MA.
The EBV lytic cycle in LCL cells was induced with n-butyric acid at a final concentration of 3 mM for 4 days. Cells were centrifuged at 1,800 × g for 20 min, and the supernatant was filtered through a 0.45-mm-pore-size cellulose acetate filter. EBV viral particles were concentrated by ultracentrifugation at 70,000 × g at 4°C, and virus pellets were suspended in RPMI medium without serum in 1/100 volume of original medium.
Deidentified peripheral blood mononuclear cells (PBMCs) were obtained from the University of Pennsylvania CFAR Immunology Core, and 1 × 107 PBMCs were infected by EBV as previously described (27, 28). Briefly, PBMCs were infected with EBV in 1 ml of RPMI 1640 medium with 10% FBS and 5 ng/ml of Polybrene (Sigma, Marlborough, MA). After 4 h at 37οC (time point 0), infected and uninfected cells were centrifuged for 5 min at 1,200 × g; the pellet was washed twice with fresh RPMI medium, resuspended in fresh RPMI medium with 10% FBS, and cultured in six-well plates at 37οC. These cells were collected at 2, 8, 24, 48, 72, and 96 h postinfection (p.i.) and washed twice with 1× phosphate-buffered saline (PBS) before being spotted on slides.
Equal numbers of washed suspension cells of uninfected and EBV-infected lymphocytic cells were spotted on 10-well glass slides which were fixed/permeabilized with prechilled acetone. The adherent C666-1 and 184B5 cells were grown in fibronectin-coated eight-well chamber slides for 48 h prior to fixation by 4% paraformaldehyde and permeabilization with 0.2% Triton X-100. These cells were washed, blocked with Image-iT FX signal enhancer (Invitrogen) for 20 min, and incubated with goat anti-ASC, rabbit anti-caspase-1, mouse anti-AIM2, or mouse anti-IFI16 antibody for 2 h at 37°C, followed by anti-goat Alexa Fluor-488-, anti-rabbit Alexa Fluor-594-, or anti-mouse-Alexa Fluor-594-labeled antibody, for 1 h at 37°C. The Nikon Eclipse 80i fluorescence microscope was used for imaging, and analysis was performed using Metamorph imaging software. Mean pixel intensities were analyzed for three different fields with a minimum of 15 cells each with the Metamorph pixel intensity calculator. These results are shown as percent distribution and colocalization in comparison to control cell values.
The LCL, Raji, and Ramos cells fixed with acetone were incubated with mouse anti-IFI16 antibodies followed by donkey anti-mouse Alexa Fluor-488 secondary antibodies. These cells were fixed with 2% paraformaldehyde for 10 min, permeabilized in 0.2% Triton X-100 for 5 min, and treated with 0.1 M Tris-HCl (pH 7.0) for 2 min and 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) twice for 2 min. These cells were further treated with denaturing solution (70% formamide in 2× SSC) at 70°C for 2 min, washed, dried, and subjected to in situ hybridization. Biotinylated EBV fluorescence in situ hybridization (FISH) probes with fragment sizes ranging from 100 to 1,000 bp (Enzo Lifesciences) were diluted in DenHyb solution (Insitus Biotechnologies, Albuquerque, NM) and incubated with cells for hybridization in a humidified chamber at 37°C overnight. The slides were sequentially washed in 2× SSC (10 min) and deionized water (1 min) and stained with streptavidin-conjugated Alexa Fluor-594 (1:1,000); nuclei were counterstained with 4′,6′-diamidino-2-phenylindole (DAPI) and examined. Colocalizations of mean pixel intensities were analyzed for three different fields with a minimum of 15 cells each with the Metamorph pixel intensity calculator. The percent colocalization of nuclear IFI16 of Ramos, LCL, and Raji cells with the EBV-specific FISH probe was analyzed, and the data are presented as bar graphs.
Gene expression of various inflammasome proteins (IL-1β, caspase-1, ASC, AIM2, IFI16, and EBER) was examined by real-time reverse transcription-PCR (RT-PCR) using a SYBR green detection system (21). The primers are listed in Table 1.
The expression levels of these genes were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expression. The final mRNA levels of the genes studied were normalized using the comparative cycle threshold method.
Whole-cell protein from various cell lines was prepared using radioimmunoprecipitation assay (RIPA) lysis buffer containing a protease inhibitor cocktail. Equal amounts of protein samples were resolved by 10 to 20% SDS-PAGE and subjected to Western blotting. Processing of caspase-1 and IL-1β was examined by specific antibodies detecting the proforms and mature forms of each protein. To examine the intracellular distribution of ASC, AIM2, NLRP3, caspase-1, and IFI16, nuclear and cytoplasmic extracts were prepared from cells as per the manufacturer's instruction and as described previously (21). The nuclear and cytoplasmic fractions were subjected to immunoblotting with anti-ASC, anti-caspase-1, and anti-IFI16 antibodies. These membranes were stripped and immunoblotted with anti-tubulin and anti-TBP antibodies to confirm the purity and equal loading of cytoplasmic and nuclear lysates, respectively. The immunoreactive bands were developed by enhanced chemiluminescence reaction (NEN Life Sciences Products, Boston, MA) and quantified by standard protocols.
EBV infection-induced protein-protein interactions were examined by coimmunoprecipitation (co-IP) experiments. Cells were washed, lysed in RIPA lysis buffer supplemented with protease inhibitor cocktail, and clarified by centrifugation for 15 min at 4°C. Equal amounts of total protein lysates were used for co-IP experiments. The lysates were incubated for 2 h with immunoprecipitating antibody (anti-ASC or anti-caspase-1 antibody) at 4°C, and immune complexes were captured using 10 μl of protein G-Sepharose. The samples were washed three times with PBS, boiled with SDS-PAGE sample buffer, resolved by 10% SDS-PAGE, and subjected to Western blotting.
Exosome-depleted cell culture RPMI 1640 medium with 10% FBS was prepared by ultracentrifugation of the medium at 100,000 × g for 16 h to remove naturally occurring exosomes from fetal bovine serum (29). LCL, Ramos, and Raji cells grown in RPMI 1640 medium with 10% FBS were harvested, washed with Hanks balanced salt solution, and resuspended at the same cell density in exosome-depleted RPMI 1640 medium. These cultures were incubated for 3 days, and released exosomes were harvested by stepwise high-speed and ultracentrifugation methods (29). The purity of exosomes obtained was assessed by Western blotting for the presence of the multivesicular body proteins Alix and Tsg101 and the absence of the endoplasmic reticulum (ER) protein calnexin (29).
Data were analyzed, and two groups were compared employing the Student t test. P values of <0.05 were considered significant for analysis of the data.
The inflammasome platform provides the molecular scaffolds required for autocleavage of inactive procaspase-1 into active caspase-1 (p20), which then cleaves pro-IL-1β, IL-18, and IL-33 to produce functional inflammatory cytokines. To determine whether the latent EBV genome is recognized by the host innate sensing mechanism and whether such recognition induces an inflammasome response, EBV+ cells with various types of latency, such as B-cell LCL (latency III), Raji (latency I), and NPC epithelial C666-1 cells (latency II), were tested. Cleaved caspase-1 (p20) was detected in the whole-cell lysates of LCL and Raji cells but not in the control EBV-negative Burkitt's lymphoma (BL) cell line Ramos (Fig. 1A, lanes 1 to 4). Similarly, we detected cleaved caspase-1 in NPC C666-1 cells but not in control normal breast epithelial 184B5 cells (Fig. 1B, lanes 1 and 2). Investigations for caspase-1 target proteins revealed the presence of processed IL-1β (p17.5) in LCL, Raji, and NPC cells (Fig. 1C, lanes 2, 3, and 5, respectively) but not in the control cells (Fig. 1C, lanes 1 and 4). We also detected processed IL-18 (p18) and IL-33 (p18) in LCL and Raji cells (Fig. 1D, lanes 1 to 3), as well as in NPC C666-1 cells (Fig. 1E, lanes 1 and 2), while control Ramos and 184B5 cells did not show evidence of cleavage despite the presence of pro-IL-18 and pro-IL-33 (Fig. 1D and andE,E, lanes 1). These results confirmed the presence of functional caspase-1 in cells with EBV latency I, II, and III.
To further confirm the inflammasome induction, cytoplasmic and nuclear fractions of LCL, Raji and C666-1 cells were prepared, and purities of the fractions were verified by the absence of tubulin and presence of TBP in the nuclear fractions and by the presence of tubulin and absence of TBP in the cytoplasmic fractions (Fig. 1F, middle and bottom blots, lanes 1 to 6). Analysis of the nuclear and cytoplasmic fractions revealed the presence of active caspase-1 (p20) in the cytoplasmic fractions of LCL, Raji, and C666-1 cells (Fig. 1F, top blots, lanes 1 to 6), which suggested that cleavage of procaspase-1 must be occurring in the cytoplasm. When we measured the inflammasome-associated gene expression, expression of caspase-1 mRNA in LCL cells was 4-fold higher than in Ramos cells, while the expression of IL-1β was unaltered (data not shown).
Collectively, these results showing the cleavage of caspase-1, and production of mature IL-1β, IL-18, and IL-33 cytokines demonstrated that inflammasomes are activated constitutively in all three types of EBV latency cells without any external stimuli.
Since activation of caspase-1 needs the sensor component of the inflammasome complex and the adaptor ASC molecule, we next determined the identity of the sensor molecule(s) in EBV latency III LCL cells. Analysis of LCL and Ramos whole-cell lysates revealed equal levels of the NLR protein NLRP3 in LCL and Ramos cells. However, the protein levels of IFI16, AIM2, and ASC were much higher in LCL cells than in Ramos cells (Fig. 2A, lanes 1 and 2). For active inflammasome complex formation, an initial homotypic interaction of the pyrin domains of sensor (IFI16, AIM2, and NLRP3) and adaptor (ASC) molecules is needed, followed by homotypic interaction of the CARD of ASC with the CARD of caspase-1 (15). Hence, we next sought to identify the ASC-interacting sensor protein(s) involved in inflammasome activation during latent EBV infection.
Equal amounts of LCL and Ramos whole-cell lysates were immunoprecipitated with anti-ASC antibodies and subjected to Western blotting for NLRP3, IFI16, and AIM2. Only IFI16 was detected in LCL cells immunoprecipitated with a modest interaction in Ramos cells; no interaction of ASC with NLRP3 and AIM2 was detected in either LCL or Ramos cells (Fig. 2B, lanes 1 and 2). Reverse co-IP studies also demonstrated the presence of IFI16 and ASC but not NLRP3 in caspase-1 immunoprecipitates from LCL cells and to some extent also in Ramos cells (Fig. 2C, lanes 1 and 2). These results suggested that IFI16, ASC, and caspase-1 are in the same complex in LCL cells.
Though IFI16 is a predominately nuclear protein, we have shown previously that during primary infection of KSHV in endothelial cells, after recognizing KSHV DNA, IFI16 forms an inflammasome complex in the infected cell nuclei which is subsequently redistributed into the cytoplasm (21). Hence, we next investigated the distribution of the inflammasome components in the cytoplasmic and nuclear fractions of LCL and Ramos cells. Purities of the fractions were verified by the absence of tubulin and presence of TBP in the nuclear fractions and by the presence of tubulin and absence of TBP in the cytoplasmic fractions (Fig. 2D, fourth and fifth blots, lanes 1 and 2). In LCL cells, an almost equal level of IFI16 was detected in the nucleus and in the cytoplasm, while in Ramos cells IFI16 was predominantly in the nucleus with only a trace in the cytoplasm (Fig. 2D, top blot, lanes 1 to 4). The AIM2 protein was predominantly nuclear in LCL cells with a low level in the cytoplasm; in contrast, it was detected only in the cytoplasm of Ramos cells (Fig. 2D, second blot, lanes 1 to 4). In LCL cells, although ASC was detected in the nucleus, a >4-fold increase was detected in the cytoplasm (Fig. 2D, third blot, lanes 2 and 4), while in Ramos cells it was solely present in the cytoplasm (Fig. 2D, third blot, lanes 1 and 3).
When we utilized the cytoplasmic and nuclear fractions from LCL cells in co-IP experiments, IFI16 was detected in ASC immunoprecipitates of both nuclear and cytoplasmic fractions (Fig. 2E, top blot, lanes 1 and 2). In reverse co-IP experiments, IFI16 and ASC were detected in caspase-1 immunoprecipitates both in the cytoplasmic and nuclear fractions (Fig. 2F, first and second blots, lanes 1 and 2). Furthermore, in reverse co-IP reactions, ASC was detected in IFI16 immunoprecipitates of nuclear and cytoplasmic fractions (Fig. 2G, top blot, lanes 1 and 2). Specificities of caspase-1 (rabbit) and IFI16 (mouse) antibodies were evaluated by immunoprecipitation with corresponding control IgG using whole-cell lysates from LCL cells and subjected to Western blotting for corresponding caspase-1 and IFI16 (Fig. 2H and andI).I). These results clearly demonstrated the presence and interaction of the IFI16–ASC–caspase-1 inflammasome in both the cytoplasm and nucleus of latency III LCL cells. When gene expression levels of IFI16, AIM-2, and ASC were compared, IFI16 and ASC levels were slightly higher in LCL cells than in Ramos cells, with very a high level of AIM2 in LCL cells relative to Ramos cell levels (data not shown).
Taken together, these results suggested that the IFI16-, ASC-, and caspase-1-mediated inflammasome is present in the nucleus as well as in the cytoplasm of latency III LCL cells. Though IFI16 and ASC coimmunoprecipitated with caspase-1 (presumably, procaspase-1), detection of active caspase-1 (p20) in the cytoplasmic fractions of LCL, Raji, and C666-1 cells (Fig. 1F) clearly demonstrated that activation of procaspase-1 occurs in the cytoplasm. Since IFI16 is a nuclear protein, its abundant presence in the cytoplasm depicts its movement from the nucleus, perhaps due to constant recognition of the presence of latent EBV, continuous inflammasome activation in the nucleus, and subsequent subcellular redistribution into the cytoplasm.
To examine the identity of the sensor molecule(s) involved in the inflammasome complex and its distributions in EBV latency I cells, first we examined the expression of the inflammasome components in these cells. NLRP3 and AIM2 protein levels were similar in Raji and Ramos cells, and, in contrast, the levels of IFI16 and ASC molecules were higher in Raji cells than in Ramos cells (Fig. 3A, lanes 1 and 2). From whole-cell lysates, IFI16 and ASC coimmunoprecipitated with caspase-1, and, in contrast, no interaction of NLRP3 and AIM2 with caspase-1 was detected (Fig. 3B, lane 2). Similar to the results shown in Fig. 2C, comparatively lower levels of IFI16 and ASC were coimmunoprecipitated with caspase-1 in whole-cell lysates of Ramos cells (Fig. 3B, lane 1).
When the subcellular distribution of inflammasome components was investigated, IFI16 was present in the nucleus as well as in the cytoplasm of Raji cells (Fig. 3C, lanes 1 and 2), which is similar to findings in LCL cells. Compared to the levels in the nuclear fractions, relatively greater quantities of AIM2 and ASC were detected in the cytoplasm (Fig. 3C, lanes 1 and 2). These results demonstrated the interaction of IFI16 and ASC with caspase-1 and suggested the induction of an IFI16-mediated inflammasome assembly in latency I cells and subcellular redistribution.
The mature inflammasome is a complex of sensor, adaptor, and executioner molecules which can be visualized by fluorescence microscopy (21). Since inflammasome activation needs homotypic interaction between ASC and caspase-1, we visualized this interaction in EBV latent B cells such as LCL and Raji cells and compared the results with those in EBV-negative Ramos cells. Similar to the Western blot findings (Fig. 2D), both ASC and caspase-1 were present in Ramos cell cytoplasm and not in the nucleus (Fig. 4A). We observed a moderate level of colocalization in the cytoplasm (Fig. 4A), which supported the detection of ASC in caspase-1 immunoprecipitations (Fig. 2C). In contrast, in LCL and Raji cells we observed prominent, readily visible caspase-1- and ASC-colocalizing spots in the perinuclear area of the cytoplasm (Fig. 4B and andC,C, red arrows), with some colocalization detected in the nucleus (Fig. 4B and andC,C, yellow arrows). These results corroborated very well the Western blot data of IFI16 and ASC distribution in the nuclear and cytoplasmic fractions and their interactions in co-IP reactions, as shown in Fig. 2D to toGG and Fig. 3B and andCC.
Since AIM2 and NLRP3 did not interact with ASC in EBV+ cells, we selected only IFI16 to visualize the interaction between sensors and adaptor ASC by fluorescence microscopy. Similar to the Western blot findings (Fig. 2D), IFI16 was predominantly present in the nucleus of Ramos cells, while ASC was observed strongly in the cytoplasm, with a moderate level in the nucleus (Fig. 4D). Very few ASC-IFI16 colocalization spots were detected in both the cytoplasm and the nucleus (Fig. 4D), which also supported the detection of IFI16 in caspase-1 immunoprecipitates of Ramos cells (Fig. 2C). In contrast, immunofluorescence imaging showed a nuclear-to-cytosolic redistribution of IFI16 and a strong cytosolic and nuclear ASC-IFI16 colocalization in LCL and Raji cells (Fig. 4E and andF,F, red and yellow arrows). Percent distribution of IFI16 and its percent colocalization with ASC in Ramos, LCL, and Raji cells were enumerated utilizing mean pixel intensities from the images in Fig. 4D to toF.F. These results clearly demonstrated that IFI16 is more abundant in the cytoplasm of LCL and Raji cells than that of Ramos cells (Fig. 4G), with about ≥70 and ≥65% of IFI16-ASC colocalization in LCL and Raji cells, respectively (Fig. 4H). These results validated our biochemical data demonstrating the interaction of IFI16 and ASC in the nucleus and the cytoplasm of LCL cells (Fig. 2B, ,D,D, ,E,E, and andG)G) as well as the distribution of IFI16 and ASC in Raji cells (Fig. 3C). These results clearly demonstrated that sensor IFI16 is interacting with adaptor ASC in latency III and latency I EBV+ cells and that latent EBV infection induces the constitutive formation of an inflammasome protein complex involving IFI16, ASC, and procaspase-1 with a distinct redistribution in latent cells.
We next investigated the presence of inflammasome components in the latency II NPC C666-1 cells compared with normal epithelial (breast) 184B5 cells. Interestingly, the NLRP3 and IFI16 levels were higher in control 184B5 cells than in NPC cells; in contrast, AIM2 and ASC levels were higher in NPC cells than in 184B5 cells (Fig. 5A, lanes 1 and 2). To identify the sensor molecule(s) involved in the inflammasome complex in latency II cells, whole-cell lysates were immunoprecipitated with anti-caspase-1 antibodies and subjected to Western blotting for NLRP3, IFI16, AIM2, and ASC. Caspase-1 interacted with ASC and IFI16 in NPC cells to a much higher degree than in 184B5 control cells (Fig. 5B, lanes 1 and 2). Furthermore, IFI16 was detected in both the nucleus and the cytoplasm of NPC cells but not in 184B5 cells, where it was detected only in the nucleus (Fig. 5C and andD,D, lanes 1 and 2). The adaptor molecule ASC was predominately detected in the cytoplasm of NPC cells, with much lower levels in the nucleus (Fig. 5C, lanes 1 and 2), while it was exclusively cytoplasmic in 184B5 cells (Fig. 5D, lanes 1 and 2). Taken together, these results suggested the constitutive induction of the IFI16, ASC, and caspase-1 inflammasome assembly in EBV latency II cells.
Next, we visualized the IFI16-mediated inflammasome complex in latency II NPC C666-1 epithelial cells. In control normal breast epithelial 184B5 cells, we did not observe any colocalization between ASC and caspase-1 molecules (Fig. 6A). In contrast, we observed readily visible ASC–caspase-1 colocalization spots in both the nucleus (Fig. 6B, yellow arrows) and the perinuclear area of the cytoplasm of NPC cells (Fig. 6B, red arrows). As seen in the Western blot data shown in Fig. 5D, IFI16 was localized in the nucleus of the control 184B5 cells while ASC was predominantly detected in the cytoplasm, with no colocalization between IFI16 and ASC (Fig. 6C). In the NPC cells, IFI16 was detected in both the nucleus and the cytoplasm (Fig. 6D), and we observed prominent IFI16 and ASC colocalization spots in the perinuclear cytoplasmic area (Fig. 6D, red arrows) as well as in the nucleus (Fig. 6D, yellow arrows). We did not observe any colocalization between AIM2 and ASC molecules in either control or NPC cells though they were present predominantly in the cytoplasm (Fig. 6E and andF).F). Pixel density analysis of the images shown in Fig. 6C and andDD also revealed that IFI16 is mostly nuclear in 184B5 cells (Fig. 6G). In contrast, ≥55% and 45% of IFI16 was detected in the cytoplasm and nucleus, respectively, of NPC cells (Fig. 6.G). These analyses also showed that ≥60% of IFI16 colocalized with ASC in the NPC cells (Fig. 6H). Taken together, these results corroborated the Western blot data (Fig. 5A to toD)D) and demonstrated that the IFI16, ASC, and procaspase-1 inflammasome protein complex is constitutively induced in NPC latency program II of EBV, which redistributes in both the nucleus and cytoplasm of these infected cells.
After establishing the presence of an IFI16-, ASC-, and caspase-1-mediated inflammasome complex in latency I, II, and III, we investigated inflammasome formation and redistribution during EBV de novo infection of human PBMC-derived primary B cells. In uninfected cells, IFI16 was mostly detected in the nucleus while ASC could be observed in the cytoplasm without any significant colocalization with IFI16 (Fig. 7A). Eight hours post-EBV infection, we observed ASC and IFI16 colocalization in the nuclear periphery (Fig. 7B). After 24 h p.i., the number of colocalization spots increased and remained within the nucleus (Fig. 7C, red arrows). However, by 48 h p.i., we observed the redistribution of the IFI16 and ASC complex in the cytoplasm (Fig. 7D, red arrows), which was consistently observed in the perinuclear area of the cytoplasm at later time points (Fig. 7E and andF,F, red arrows). In our earlier studies with KSHV, we have shown that IFI16 recognizes the KSHV genome and forms an inflammasome complex with ASC and caspase-1 within the nucleus of infected human dermal microvascular endothelial cells (HMVEC-d) and is subsequently distributed into the cytoplasm (21). Our results here suggested that, similar to KSHV, during primary EBV infection IFI16-ASC inflammasome complex formation takes place in the nucleus and then redistributes into the perinuclear area.
Since we detected cleaved IL-1β, IL-18, and IL-33 in whole-cell lysates of LCL and Raji cells, we next investigated their presence in the culture supernatant of these cells and compared them to the Ramos cells. To our surprise, we could not detect appreciable secretion of these cytokines by enzyme-linked immunosorbent assay (ELISA), which could be due to (i) less sensitivity of the detection kit, (ii) reduced level of secretion, or (iii) secretion in an undetectable form. EBV proteins, such as LMP1 and BARF-1, and host/viral miRNA have been known to be secreted in exosomes that were isolated from the serum and saliva of NPC patients (30). Exosomes are 40- to 100-nm-diameter small-membrane vesicles that correspond to the internal vesicles present in multivesicular endosomes (MVEs). Exosomes contain endosome-associated proteins (Rab GTPases, annexins, flotillin, Alix, and TSG101), integrins, and tetraspanins (CD63, CD9, CD81, and CD82) (31). To investigate whether EBV-induced inflammasome-generated cytokines are part of the exosome, we harvested exosomes from LCL, Raji, and Ramos cells. The purities of the exosomes were shown by the presence of multivesicular body proteins Alix and Tsg101 and the absence of the ER protein calnexin (Fig. 8A, lanes 1 to 3) (29). Furthermore, the presence of LMP1 in exosomes derived from LCL and Raji cells confirmed their origin from EBV+ cells (Fig. 8A, lanes 1 to 3). Interestingly, we detected IFI16 in the exosomes of LCL, Raji, and Ramos cells (Fig. 8B, top blot, lanes 1 to 3). More interestingly, cleaved caspase-1, IL-1β, IL-18, and IL-33 were detected in the exosomes from LCL and Raji cells but not in the exosomes from Ramos cells (Fig. 8B, lanes 1 to 3). Since IL-1β encapsulated in the exosomes will not be detected in the capture ELISA, as exosomes will be washed away after incubation of the supernatants, these results suggested a plausible reason for the low level of IL-1β detected in the supernatants of EBV+ cells.
The above studies demonstrated the secretion of inflammasome components and inflammatory IL-1β, IL-18, and IL-33 cytokines in exosomes of LCL and Raji cells. Since Rab27a, a member of the Rab GTPases, has been implicated in exosome formation and release, we next determined whether Rab27a is part of the exosomes containing the inflammasome products detected in LCL and Raji cells. We detected Rab27a and caspase-1 in the cytoplasm of Ramos cells, but no colocalization was observed (Fig. 8C). In contrast, Rab27a prominently colocalized with caspase-1 in the cytoplasm of EBV+ LCL (Fig. 8D, red arrows) and Raji (Fig. 8E, red arrows) cells. Furthermore, Rab27a also predominantly colocalized with IL-1β in the cytoplasm of LCL (Fig. 8G, red arrow) and Raji (Fig. 8H, red arrow) cells but not in Ramos cells (Fig. 8F). These results clearly demonstrated that Rab27a-associated secretory exosomes contain IL-1β and caspase-1 and suggest that continuous encapsulation and secretion of caspase-1 and cleaved inflammatory cytokines IL-1β, IL-18, and IL-33 in exosomes could be a potential way by which EBV subverts their functions.
Our earlier studies with UV-inactivated KSHV suggested that KSHV latent gene expression and/or the viral genome is crucial for inducing the IFI16-inflammasome (21). Transduction of endothelial cells with lentivirus vectors carrying KSHV latent ORF71, ORF72, ORF73, and K12 genes and lytic K8 and ORF74 genes did not induce caspase-1 activation (21). This suggested that KSHV genes individually do not play a role in KSHV-induced inflammasome activation and that IFI16 does not recognize linear integrated (retro) foreign DNA. The current studies demonstrated that IFI16 responds to EBV infection and constitutively induces the inflammasome complex in all EBV+ latency cells, which suggested that IFI16 must be sensing a common factor present in all three types of EBV latency.
To determine what is sensed by IFI16 to induce the inflammasome, we considered the following three common factors that are present in all three types of EBV latency: (i) EBV genome, (ii) EBNA1, and (iii) EBER-1 and -2. EBNA1 plays key roles in EBV biology and mediates tethering of the episomal EBV genome to the host chromosome, replication of the viral genome once per cell cycle, segregation of the duplicated viral genome once per cell cycle, and stable segregation of the duplicated viral genome to the daughter cells. To determine whether EBNA1 protein alone is sensed by IFI16 to induce the inflammasome, we utilized BJAB cells expressing only EBNA1 protein (32). When whole-cell lysates of control BJAB (EBV-negative BL) cells, EBNA1-expressing BJAB cells, and latency I Raji cells were examined for inflammasome activation, only in Raji cells we did detect cleaved IL-1β (Fig. 9A, fourth blot, lanes 1 to 3). Though IFI16 and ASC were expressed in equal levels in EBNA1+ BJAB and Raji cells (Fig. 9A, top three blots, lanes 1 to 3), we did not detect any evidence of inflammasome induction (Fig. 9A). Co-IP investigations also did not reveal any interaction between IFI16 and caspase-1 in EBNA1-expressing BJAB cells, whereas interactions were readily witnessed in the Raji cells (Fig. 9B). These results clearly demonstrated that EBNA1 alone does not play a role in the IFI16-mediated inflammasome activation observed in the EBV latency I, II, and III cells.
EBV also expresses abundant nonpolyadenylated and noncoding EBER-1 and EBER-2 RNAs (167 and 173 bases, respectively) during latency I, II, and III which are transcribed by host RNA polymerase III. Though IFI16 has been shown to bind predominately to double-stranded DNA and, to some extent, to single-stranded DNA, there is a possibility that it may be recognizing EBERs in the infected cell nuclei. Moreover, RIG-I recognizing RNA in the cytoplasm can form an ASC- and caspase-1-mediated inflammasome (33), and transfection of EBER plasmids in the NPC cell line has been shown to induce IL-1β production via RIG-I and caspase-1 (34). To determine whether the observed inflammasome induction in EBV I, II, and III latency is via the recognition of EBER, Raji cells were treated with dimethyl sulfoxide (DMSO) alone or with RNA Pol III inhibitor reconstituted in DMSO at 12.5 μM and 25 μM concentrations for 24 h. When whole-cell lysates were analyzed by Western blotting, cleaved caspase-1 was detected in both untreated and treated cells but not in control BJAB cells (Fig. 9C, top blot, lanes 1 to 4). Analysis of EBER gene expression by real-time RT-PCR confirmed the inhibition of EBER transcription by the inhibitor treatments (Fig. 9D). These results demonstrated that EBERs or EBER–RIG-I are not involved in the inflammasome activation exhibited in EBV latency I cells. Collectively, these studies suggested that cleavage of caspase-1 was most likely due to an IFI16-mediated inflammasome and not due to other sensors such as NLRP3, AIM-2, and RIG-1 or to viral EBNA1 and EBERs.
Since we ruled out two of the three common factors present in all three types of EBV latency, such as EBNA1 and EBER-1 and -2, as the inducer of IFI16 inflammasomes, we next considered the other common factor, namely, the EBV genome as inducer of the inflammasome. Previously we have shown that IFI16 colocalizes with the KSHV genome in the nucleus during primary infection, which initiates inflammasome complex formation with ASC and procaspase-1 (21). To determine whether IFI16 recognizes the presence of episomal EBV DNA, we utilized a biotinylated EBV FISH probe to detect the EBV genome in these cells. In the combined FISH-IFA colocalization experiments, IFI16 was found predominately in the nucleus of EBV-negative Ramos cells, with no appreciable levels in the cytoplasm and no signal for EBV (Fig. 9E). We observed that the majority of the EBV genome in the LCL and Raji cell nuclei colocalized with nuclear IFI16 (Fig. 9F and andG,G, yellow spots in IFI16/EBV and white spots in IFI16/EBV/DAPI). Analysis of colocalized pixel intensities shown in Fig. 9E to toGG revealed that ≥35 and ≥40% nuclear IFI16 colocalized with EBV genome probe in LCL and Raji cells, respectively (Fig. 9H). As demonstrated in Fig. 2 and and3,3, IFI16 was also detected in the cytoplasm of LCL and Raji cells without any colocalization with viral genome probe. These results indicated the specificity of the FISH probes, cytoplasmic translocation of IFI16 in EBV+ cells, and the absence of viral genome in the cytoplasm. Since the LCL and Raji cell nuclei carry numerous episomal DNA copies, the multiple IFI16 colocalization spots indicate IFI16 interactions with several of these viral genomes. These findings convincingly indicate that IFI16 must be sensing the EBV genome continuously in the latently infected cells, which results in constitutive inflammasome activation.
Viruses possess several structurally diverse PAMPs, including surface proteins, DNA, and RNA species. It is also well established that herpesviruses like EBV modulate the immune system for their own survival, and studies, including ours, have established cross talk between host innate immunity and herpesviruses (13, 14, 21). EBV is unique among human herpesviruses and establishes three types of latency (I, II, and III) with distinct latent protein expression patterns (6). EBV's persistence is indicative of its interplay with the host immune system and its capability to subvert host immune surveillance (35). However, EBV's interplay with the inflammasome-mediated innate immunity has not been investigated. Our comprehensive studies presented here are the first to demonstrate (i) the sensing of EBV genome by the nuclear resident IFI16 protein, (ii) constitutive activation of the IFI16–ASC–caspase-1 inflammasome in all three EBV latency cells, resulting in the production of the cleaved form of proinflammatory IL-1β, IL-18, and IL-33 cytokines, and (iii) the release of IFI16, caspase-1, IL-1β, IL-18, and IL-33 in the exosomes of the culture supernatant of cells harboring the EBV genome (Fig. 10).
Recently solved crystal structures of HIN domains of AIM2 and IFI16 in complex with dsDNA demonstrated that IFI16 and AIM2 recognize DNA in a non-sequence-specific manner through electrostatic attraction between the positively charged HIN domain residues and the sugar-phosphate backbone of the dsDNA (36). In addition, pyrin and HIN domains of AIM2 were shown to be in an autoinhibited intramolecular complex state which is liberated by DNA binding, thus facilitating the assembly of inflammasomes along the DNA staircase (36). As a true pattern recognition receptor, non-sequence-specific direct recognition of EBV genome by IFI16 may be facilitating DNA-mediated assembly of inflammasome signaling complexes. Multiple IFI16-colocalizing spots in cells carrying multiple EBV genomes (Fig. 9), representing the sensing of individual genomes by IFI16, suggest that IFI16 must be sensing the EBV genome continuously in the latently infected cells and thereby contributing to constitutive inflammasome activation (Fig. 10). IFI16 interaction with the EBV genome probably triggers changes that allow it to interact with ASC, leading to the formation of the inflammasome complex with ASC and caspase-1 and rapid translocation to the cytoplasm while a new IFI16 molecule probably comes in contact with viral genome, thus resulting in a continuum of the above processes (Fig. 10).
Studies using short DNA oligonucleotides have demonstrated that IFI16 can directly bind to cruciform structures, with a strong preference for four-way junction DNA (37). IFI16 was demonstrated to bind strongly to superhelical plasmid DNA, resulting in one or more retarded DNA bands in the gels, and removal of IFI16 resulted in the recovery of the original mobility (37). In contrast, IFI16 was found to bind very weakly to the same DNA in a linear state. Absence of caspase-1 activation in endothelial cells transduced with KSHV latent genes in lentivirus vectors also demonstrated that IFI16 does not recognize linear or linear integrated (retro) foreign DNA (21). Cruciform structures of DNA are formed by inverted repeats, and their stability is enhanced by DNA supercoiling; they are important for DNA replication, regulation of gene expression, nucleosome structure, and recombination and are targets for regulatory proteins such as histones, HMG, p53, 14-3-3, etc. (38, 39). The ability of IFI16 to recognize local DNA structures suggests that it may play important roles in recognizing episomal EBV DNA that may be forming cruciform structures during latent gene transcription and latent genome replication. However, how IFI16 differentiates EBV genome from host genome and whether it recognizes the foreign EBV genome alone or in conjunction with other protein(s) or sensor(s) are not known and are under active investigation.
A recent study showed that in vitro transfection of fragmented EBV genomic DNA and poly(dA-dT) into NPC cells (NPC-HK1) leads to activation of the AIM2-inflammasome (34). Transfection of EBV-encoded EBER-1/EBER-2 also activated a RIG-I-mediated inflammasome in cultured NPC cells (34). Additionally, extracellular ATP and ROS-induced NLRP3 activation were also reported to mediate inflammasomes in NPC cells (34). However, these studies overlooked the well-established fact that introduction of DNA by transfection methods results in DNA recognition by the cytosolic sensor AIM2, which leads to activation of the AIM2-inflammasome. Moreover, these studies did not examine the induction of inflammasomes by IFI16 in NPC cells or other endogenous inflammasome induction without external stimuli. Hence, the observed responses in this study do not reflect an endogenous activation of inflammasomes under physiological conditions in latently infected NPC cells with episomal EBV genome and its gene products.
In contrast, our studies were conducted to determine the endogenous inflammasome induction in cells carrying different types of EBV latency and demonstrate IFI16-inflammasome activation and noninvolvement of AIM2 and NLRP3. All cells with latent EBV are positive for the viral genome, EBNA1, EBER, and EBV miRNAs. However, reduction in EBER production by RNA Pol III inhibitor did not hamper the cleavage of caspase-1 into its active form, which clearly ruled out involvement of the RIG-I inflammasome in caspase-1 production in EBV+ latent cells (Fig. 9C). Moreover, studies with cells carrying EBNA1 alone demonstrated that EBNA1 has no role in inflammasome induction by IFI16 (Fig. 9B). These observations together with the colocalization of IFI16 with the EBV genome and the fact that IFI16 does not recognize RNA clearly suggested that the EBV genome is the foreign material sensed by IFI16 in the nucleus, leading into inflammasome formation.
Since AIM2 and NLRP3 are cytosolic DNA sensors, they probably do not have access to sense the latent EBV genome in the nucleus. The 3′ untranslated region (UTR) of NLRP3 mRNA has been shown to be targeted by cellular microRNA-233 (miR-233) and EBV-encoded miR-BART15, which may be an EBV-induced strategy to block NLRP3-mediated inflammasome activation in the cytosol during latency (40). Irradiation treatment of NPC patients has been reported to result in increased levels of DNA in the cytosol from the nucleus and mitochondria, leading to AIM2-inflammasome induction (34). Similarly, the cisplatin-based chemotherapy known to produce the ROS-activating NLRP3 inflammasome was reported in NPC patients (34). These studies suggest that AIM2- and NLRP3-mediated inflammasomes may only be activated upon external stimuli in EBV latent cells. In contrast, our studies suggest that cleaved caspase-1, IL-1β, IL-18, and IL-33 are generated in all EBV+ latent cells by IFI16 sensing and not by AIM2, NLRP3, or RIG-I sensors and without any external stimuli.
Formation of a functional inflammasome platform leads to cleavage of procaspase-1 into active caspase-1, which in turn produces the inflammatory IL-1β and IL-18 cytokines by cleavage of their proproteins. In all latency programs of EBV+ cells, cleaved caspase-1 was detected in the cytoplasm and not in the nucleus (Fig. 1F), despite the presence of inflammasome complex in the nucleus (Fig. 2E to toG).G). This suggested that though the inflammasome complex is formed in the nucleus after sensing the EBV genome, it rapidly moves to the cytoplasm, where autocleavage of caspase-1 occurs. The primary infection of PBMCs by EBV also validates this notion as IFI16 and ASC colocalization, although seen in the nucleus at 8 h p.i., was observed predominantly in the cytoplasm at 48 to 96 h p.i. (Fig. 7). Activation of caspase-1 in the cytoplasm coupled with nuclear-to-cytoplasmic redistribution of IFI16, ASC, and caspase-1 suggests a possibility of additional factors involved in redistribution, probably to regulate caspase-1 activation and likely to prevent unwanted caspase-1 activity in the nucleus.
Li et al. (41) have shown that the sensing of herpes simplex virus 1 (HSV-1) DNA in the nucleus of U2OS cells by overexpressed IFI16 is influenced by the subcellular localization of IFI16, which was determined by the acetylation status of IFI16. Acetylation within the nuclear localization signal (NLS) of IFI16 promoted cytoplasmic localization and inhibited its nuclear import. In addition, inhibition of deacetylase activity by deacetylase inhibitors prevented nuclear accumulation of IFI16, suggesting the role of histone deacetylases (HDAC) in regulating IFI16 localization. Questions such as whether IFI16 undergoes modifications after recognizing the EBV genome and whether IFI16 gets acetylated during primary EBV infection and during latency are currently under study.
Studies suggest that in order to neutralize inflammasome onslaught, viruses have evolved several strategies. For example, cytoplasmic replicating DNA genome containing poxviruses, such as cowpox virus (CPV) and vaccinia viruses, produce IL-1β and IL-18 binding proteins that block the downstream signaling of these cytokines (42). Similarly, ORF63 of KSHV inhibits NLRP1 oligomerization and inflammasome activation (43). Additionally, CPV-encoded cytokine response modifier A (CrmA) inhibits caspase-1 activity (44). Active forms of immunomodulatory IL-1β, IL-18, and IL-33 cytokines could lead into the activation of a type I IFN response and thereby induce apoptosis of infected cells. Studies have demonstrated that EBV has evolved strategies to evade IFN-induced apoptosis during latency; for example, counteraction of EBER-induced IFN-mediated apoptosis through RIG-I activation occurs by binding to PKR and inhibition of apoptosis signals (45). LMP1-induced expression of anti-apoptotic proteins such as survivin, A20, and Bcl-2 has also been reported to prevent IFN-induced apoptosis.
Sequestration of IL-1β, IL-18, and IL-33 in exosomes might be an additional strategy of EBV to escape host immunity. Sequestration of inflammatory cytokines in the exosomes will lead to their unavailability to bind their cognate receptors, which would prevent activation of a cytokine-induced IFN response. Furthermore, it has been reported that chemo-/radiotherapy-induced inflammasome-derived IL-1β in tumor cells recruits neutrophils, which results in recurrence-free survival of NPC patients (34). Thus, the presence of IL-1β in the exosomes, as shown by our data, may render it unavailable to recruit neutrophils, thereby facilitating EBV persistence. Exosomes are probably being exploited by EBV to subvert detrimental effects of the host immune-mediated inflammatory response on itself. Further investigations are required to comprehend (i) whether cleavage of IL-1β, IL-18, and IL-33 occurs in the exosomes by caspase-1, (ii) how EBV is exploiting exosomes to subvert innate immunity, (iii) how exosome secretion is regulated, and (iv) what effect EBV-induced exosomes have on neighboring cells during latency. These studies are beyond the scope of the present study.
EBNA1 and EBNA2 proteins were demonstrated to act as self-antigens and thereby were reported to play a potential role in systemic lupus erythematosus (SLE) (6). IFI16 was also observed in the cytoplasm of SLE patients' skin sections which was implied to contribute to the detection of auto-antibodies against IFI16 in the SLE patient's serum (46). Recently, IL-33 has been shown to play an important role in several autoimmune disorders, including asthma, rheumatoid arthritis, and SLE (13). Our data demonstrating the presence of IFI16 and cleaved IL-33 during EBV latency suggest that they may contribute, along with EBNA1 and EBNA2, in SLE. Antigen-presenting cells (APCs) were reported to engulf exosomes, process their proteins in lysosomes, and present on major histocompatibility complex (MHC) class II to activate CD4+ T cells against exosome-entrapped antigens (47). Hence, the presence of IFI16, caspase-1, and other unknown self-antigens in exosomes released from cells latently infected with EBV may lead to its presentation to elicit an autoimmune response. The potential connections between autoimmune disorders and the EBV latency-induced inflammasome warrant further investigation. Similarly, the fate of exosomes released from EBV+ latent cells and a determination of whether inflammasome activation and exosomal association in EBV latency could be exploited for prognostic/diagnostic biomarkers and therapeutic targets in EBV-associated cancers need to be evaluated further.
Studies suggest the association of IFI16 with DNA damage response (DDR) signaling. IFI16 and DDR signaling components BRCA1, Mre/Rad50/NBS1, and ATM were reported to interact at the genomic site of DNA damage (48). Histone deacetylases (HDAC) and histone acetyltransferases (HATs) reported to modulate cellular distribution of IFI16 have also been shown to mediate the DDR through acetylation of important DNA repair and checkpoint proteins (49, 50). Furthermore, UV-B irradiation of human keratinocytes leads to the distribution of IFI16 from the nucleus to the cytoplasm and in the supernatants (46). These findings suggest that IFI16-mediated innate immunity could occur downstream to DDR and that IFI16 could be sensing the presence of several copies of EBV episomal genomes in the host cell nucleus as DNA damage, leading to IFI16-mediated inflammasome activation. Interestingly, we also observed modest interaction of IFI16 with ASC in EBV-negative Ramos and 184B5 cells in the form of some visible spots in IFA and in co-IP experiments, which could be due to DDR in these cells. Further studies are in progress to define the role of IFI16 in DDR and its connection to inflammasome activation.
The EBV-positive BL and LCL cell lines are known to produce cytokines such as IL-6, IL-8, IL-10, MCP-1, TNF-α, and TNF-β at various levels (10). Latent infection in vivo by related human herpesviruses such as KSHV, human cytomegalovirus (HCMV), and HSV-1 has been shown to be associated with markers for chronic inflammation, such as the inflammatory response, lymphocytic cell infiltration, and elevated cytokine/chemokine expression (51, 52). KSHV-associated Kaposi's sarcoma (KS) and primary effusion lymphoma (PEL) are angiogenic tumors, and the microenvironment is enriched with inflammatory cytokines, angiogenic factors, and chemokines (53) which play roles in the pathogenesis of KS and PEL. Elevated levels of IL-18 have been observed in the sera of patients with latent CMV infection compared to CMV-seronegative patients (51). Lymphocytic cell infiltration and elevated cytokine expression have been observed in trigeminal ganglia of patients with latent HSV-1 infection (52). Similar to our current findings in EBV-infected cells, our recent studies have shown IFI16-inflammasome induction in human B cells and endothelial cells latently infected with KSHV (22). In addition, our studies show that IFI16 recognizes HSV-1 early during primary infection of human foreskin fibroblast cells, leading into the lytic cycle, and induces ASC-IFI16 colocalization into the cytoplasm and caspase-1/IL-1β cleavage (54). Together with the present study, these results reinforce the notion that IFI16 acts as a nuclear pathogen sensor and detects the circular genome of EBV, KSHV, HSV-1, and possibly other viruses and could be one of the factors behind the inflammatory responses observed in EBV latency as well as in latency of other herpesviruses.
IFI16 silencing in human embryonic lung fibroblasts (HELFs) has been recently shown to enhance lytic replication of HSV-1 and HCMV, and IFI16 overexpression strongly inhibited HCMV lytic replication by blocking early and late viral gene expression (55). During primary HSV-1 infection of human fibroblast cells, HSV-1 immediate-early ICP0 protein targets IFI16 for degradation (54, 56). In contrast, in cells latently infected with EBV and KSHV, the levels of IFI16 were not altered, and this suggests that though the IFI16 inflammasome is a host innate response to infection, in appropriate cell types, EBV and KSHV might have evolved to utilize IFI16 for their latency survival advantage. More studies are essential to determine the role of IFI16 in EBV and other herpesvirus latency. Overall, our studies demonstrate for the first time the constitutive innate inflammatory responses in cells latently infected with EBV (Fig. 10) and open up several new avenues of study to understand the dynamic links between host innate responses and EBV latency.
This study was supported, in part, by Public Health Service Grants CA 075911 and CA 168472 to B.C. and the Rosalind Franklin University of Medicine and Science (RFUMS)-H. M. Bligh Cancer Research Fund to B.C.
We thank Keith Philibert for critically reading the manuscript. We thank Lindsey Hutt-Fletcher, University of Louisiana, Shreveport, LA, for the gift of the EBV latency III+ lymphoblastoid cell line (LCL). We thank Bill Sugden, McArdle Laboratory for Cancer Research, University of Wisconsin—Madison, for the gift of BJAB cells expressing only EBV-EBNA1. We thank Neelam-Sharma Walia, RFUMS, for providing us with the EBER primers.
Published ahead of print 29 May 2013