Sec22b Regulates Inflammatory Responses by Controlling the Nuclear Translocation of NF-κB and the Secretion of Inflammatory Mediators

The innate immune function of professional phagocytes is channeled through their capacity to ingest and degrade foreign particles and microbes (1). Following particle binding and uptake, phagocytes release a series of small molecules such as NO and cytokines that dictate the nature and intensity of the immune response. Particularly in dendritic cells, the phagosome also serves as a platform for Ag processing and cross-presentation (2). The aforementioned processes exert a membrane demand that is actively supplied by endosomes and the secretory pathway, a dynamic group of organelles through which proteins and lipids are synthesized, processed, and transported (3, 4). Although recycling/early endosomes and the endoplasmic reticulum (ER) supply membrane and protein effectors to nascent phagosomes (5, 6), the latter also serve as cytokine secretion sites (7). Through sequential membrane exchanges with organelles such as the lysosome, maturing phagosomes acidify and become highly degradative.

Interorganellar vesicular transport is regulated by proteins that determine cargo selectivity and vesicle destination (3). Vesicle docking and fusion rely on organelle-specific members of the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) family (4, 8). For instance, the syntaxin (Stx) 4/SNAP-23/VAMP3 complex facilitates the passage of TNF from recycling endosomes to the cell membrane (7, 9). On the other hand, trafficking within the ER, ER-Golgi intermediate compartment (ERGIC), Golgi apparatus, and phagosome circuitry is orchestrated by SNARE complexes involving Sec22b, Stx5, and Stx4 (10, 11). Notably, their absence hinders phagosome maturation and Ag cross-presentation (10, 11), which translates into phenotypes such as impaired bacterial and tumor clearance (12–14). Moreover, a pan-mouse knockout of Sec22b is embryonically lethal (15), highlighting the key importance of this ERGIC-resident SNARE in immunity and development.

The inducible NO synthase (iNOS) bridges homeostatic regulation and immunity via the production of NO from l-arginine (16). NO modulates inflammation, vasodilation, and neurotransmission. Importantly, this reactive molecule contributes to the generation of highly microbicidal reactive nitrogen species within the phagosome (17, 18) that kill internalized bacteria and Leishmania parasites (19). iNOS expression is mainly regulated at the transcriptional level by NF-κB, and enzymatic activity is regulated at the translational and posttranslational levels. Indeed, iNOS has to assemble into homodimers to transform l-arginine into reactive nitrogen species (20). Although its intracellular trafficking remains uncharacterized, iNOS-positive vesicles have been observed to reside in ER-associated aggresomes and the trans Golgi (21–23), as well as the phagosome surface (17, 22). Therefore, the question emerges as to whether secretory pathway–resident SNAREs regulate iNOS trafficking and function. We showed that iNOS is mainly localized to the ERGIC/Golgi both in the cytoplasm and at the phagosome. Notably, Sec22b knockdown hindered NF-κB nuclear translocation, thereby decreasing iNOS and TNF/IL-6 cytokine production. We also observed that Sec22b and NF-κB co-occur in the cytoplasm and the nucleus. These findings highlight a new role for this SNARE in the control of inflammatory responses.

Mouse work was done as stipulated by protocol 1706-07, which was approved by the Comité Institutionnel de Protection des Animaux of the INRS–Centre Armand-Frappier Santé Biotechnologie. This protocol respects procedures on animal practice promulgated by the Canadian Council on Animal Care, described in the Guide to the Care and Use of Experimental Animals.

Rabbit and mouse Abs anti-SAPK/JNK, -p-SAPK/JNK (Thr183/Tyr185), -p38, -p-p38 (Thr180/Tyr182), -p44/42 MAPK, -p-p44/42 MAPK (Thr202/Tyr204), -IκBα, -NF-κB, -iNOS, and -TBP were purchased from Cell Signaling (Cat. Nos. 9252, 9251, 9212, 9211, 9102, 9101, 9242, 8242, 6956, 13120, 8515); rabbit Abs anti-Sec22b and -Stx5 from Synaptic Systems (186003, 110053); rabbit anti-ERGIC53/p58 and mouse anti-β-actin from Sigma (E1031, A1978); rabbit anti-PDI and -CNX from Enzo Life Sciences (ADISPA890, ADISPA865); anti-iNOS, -fibrillarin, and -TOMM20 from Abcam (ab49999, ab186735, ab4566); and rabbit anti-P115 from Proteintech (135091AP). Pharmacological inhibitors brefeldin A (diluted to 0.5 µg/ml), monensin (5 µM), and retro-2 (100 µM) were obtained from Molecular Probes (B7450), Sigma (M5273), and Calbiochem (554715), respectively, and were reconstituted in DMSO from Bioshop (DMS555).

Bone marrow-derived dendritic cells (BMDCs) were differentiated from the bone marrow of 6–8-wk-old female C57BL/6 mice. BMDCs were differentiated for 7 d in RPMI 1640 (Life Technologies) containing 10% v/v heat-inactivated FBS (Wisent), 10 mM HEPES (Bioshop) at pH 7.4, penicillin-streptomycin (Life Technologies), and 10% v/v X63 cell-conditioned medium as a source of GM-CSF (24, 25). Experiments were carried out in complete medium containing 5% X63 cell-conditioned medium. JAWS-II dendritic cell–like lines stably transduced with short hairpin RNA (shRNA) against Sec22b or a scrambled sequence (10) were a gift from D. S. Amigorena (Institut Curie). JAWS-II cells were cultured in RPMI 1640 containing 20% heat-inactivated FBS, 10 mM HEPES at pH 7.4, penicillin-streptomycin, 10% X63 cell-conditioned medium and 40 μg/ml puromycin (Bioshop). The mouse macrophage cell line RAW264.7 clone D3 was cultured in DMEM containing l-glutamine (Life Technologies), 10% heat-inactivated FBS, 20 mM HEPES at pH 7.4, and penicillin-streptomycin. For microscopy experiments, BMDCs and JAWS-II cells were seeded onto poly-l-lysine–coated coverslips (BD Biosciences, 354085) and RAW264.7 cells onto uncoated glass coverslips (Fisher, 1254581). All mammalian cells were kept in a humidified 37°C incubator with 5% CO₂.

Leishmania major NIH S (MHOM/SN/74/Seidman) clone A2 (A2WF) promastigotes were cultured in Leishmania medium (M199-1× [Sigma] with 10% heat-inactivated FBS, 10 mM HEPES at pH 7.4, 100 μM hypoxanthine, 5 μM hemin, 3 μM biopterin, 1 μM biotin, and penicillin-streptomycin) in a 26°C incubator. Freshly differentiated promastigotes in late stationary phase (5-d cultures at > 50 × 10⁶ promastigotes/ml) were used for infections.

RAW264.7 macrophages in the second passage were seeded onto 24-well plates and reverse-transfected with Lipofectamine RNAiMAX (Thermo Fisher Scientific, 13778075) as per the manufacturer’s recommendations. Cells were transfected with small interfering RNAs (siRNAs) targeting a scrambled sequence, Sec22b (Thermo Fisher Scientific, D0018101005 and 4390815) or Stx5 (Dharmacon, L063346010005) at a final concentration of 25 nM in a 600-µl volume of complete DMEM without antibiotics for 48 h (26). Prior to experiments, macrophages were replenished with complete medium for 4 h before LPS (diluted to 100 ng/ml; Sigma, L3129) stimulation. Cells were subsequently prepared for microscopy, and their culture supernatants used for NO measurements.

Cells were incubated at 4°C for 5 min with mouse serum-opsonized zymosan (Sigma, Z4250) or L. major promastigotes, followed by a 2-min spin at 298.2 g in a Sorvall RT7 centrifuge (27). Particle internalization was triggered by transferring cells to 37°C, and after 15 min (zymosan) or 2 h (promastigotes), excess particles were washed thrice with warm medium. Cells were then lysed or prepared for confocal microscopy (27).

For phagosome isolation assays, BMDCs were fed with LPS-coated (28) 3-μm magnetic beads (Spherotech, PM-30-10). After a 24-h incubation, cells were washed, resuspended and lysed in filtered purification buffer (2 mM EDTA in PBS at pH 7.2 and 4°C) containing 1× protease inhibitors (Roche) (26). Homogenates were clarified and phagosomes were isolated with a magnet. This was followed by six washes in purification buffer with protease inhibitors, and two washes in filtered PBS (4°C) with no additives. Subsequently, phagosomes were incubated 0.5–1 h (at 4°C) in PBS or in 150 μg/ml proteinase K (Bioshop, PRK403) ± 10% Triton-X100 (27). Reactions were stopped by adding protease inhibitors, 1 μM PMSF and Laemmli sample buffer, followed by immediate boiling at 100°C for 6 min. Proteins were then separated by SDS-PAGE.

Cells were pretreated for 3 h with DMSO or the aforementioned chemical inhibitors and subsequently stimulated (in the presence or absence of inhibitors) with LPS, peptidoglycan (PGN) (40 µg/ml; Sigma, 77140), imiquimod (12.5 µg/ml; Invivogen, tlrl-imq), CpG (4 µg/ml; Invivogen, tlrl-1585), zymosan, or L. major promastigotes. Cell culture supernatants were spun to remove debris, and nitrite concentrations were assayed via the Griess reaction (29). ELISA kits were used to quantify secreted TNF (Ready-SET-Go! Mouse TNFα Kit, eBiosciences, 88732488) and IL-6 (BD OptiEIA, Becton-Dickinson Biosciences, 555240), as per the manufacturers’ instructions.

Cells on coverslips were fixed with 2% paraformaldehyde (Thermo Scientific, 28906) for 20 min and blocked and permeabilized for 17 min with a solution of 0.1% Triton X-100, 1% BSA, 6% nonfat milk, 2% goat serum, and 50% FBS. This was followed by a 2-h incubation with primary Abs diluted in PBS. Then, cells were incubated for 35 min in a solution containing suitable combinations of Alexa Fluor–linked secondary Abs (diluted 1:500; Molecular Probes, A11008, A11011, A11031, A11001, A21247) and DAPI (1:40,000; Molecular Probes, D3571). Coverslips were washed three times with PBS after every step. Coverslips were mounted onto glass slides (Fisher) with Fluoromount-G (Southern Biotechnology Associates, 010001) and sealed with nail hardener (Sally Hansen, 45077). Cells were imaged with the 63× objective of an LSM780 confocal microscope (Carl Zeiss Microimaging). Control stainings revealed no cross-reactivity or background (Supplemental Fig. 5). Images were taken in sequential scanning mode via the ZEN 2012 software and mounted with Photoshop (Adobe). Images were analyzed using Icy Software (Institut Pasteur) (30). Colocalization (31) was evaluated using the Pearson R coefficient calculated from regions of interest defined by active contours (32) on the cytoplasm (differential intensity contrast channel) or on the nucleus (DAPI). Nuclear translocation was evaluated by dividing the mean fluorescence intensity of an entire cell by that of its nucleus. Recruitment was evaluated by scoring the presence of a given protein on the phagosome membrane (33, 34).

Imaging flow cytometry experiments were performed as previously described (35). Briefly, cells were washed with cold PBS containing 1% horse serum (Sigma), 0.1% NaN₃, and 5 mM EDTA, and fixed, permeabilized, and stained. DNA was stained with DAPI immediately prior to running the samples through an ImageStreamX MKII flow cytometer running IDEAS software (Amnis). Nuclear translocation of NF-κB was evaluated using a score that quantifies the correlation of NF-κB and DAPI pixels on a per cell basis; a score of > 1 indicated nuclear translocation (35).

LPS-treated or control BMDCs were fixed overnight at 4°C in 0.1% glutaraldehyde + 4% paraformaldehyde in a cocodylate buffer at pH 7.2. After washing, cells were treated with 1.3% osmium tetroxide in collidine buffer and dehydrated (27). Pelleted cells were embedded (Spurr, Ted Pella) and placed in BEEM capsules (Pelco Int, 130). After resin polymerization, samples were sectioned using an ultramicrotome (LKB Bromma 2128, Ultratome). Sections were placed on nickel grids, treated with sodium metaperiodate, and blocked with 1% BSA in PBS. Grids were then incubated with primary Abs for 90 min (diluted 1:30 with 0.1% BSA in PBS), washed, and incubated in suitable 10 nm (anti-mouse) or 20 nm (anti-rabbit) gold particle–conjugated secondary Abs (Abcam, ab39619 and ab27237) for 60 min (diluted 1:10 with 0.1% BSA in PBS). Washed samples were treated with uranyl acetate and lead citrate for contrast and imaged via a Hitachi 7100 electron microscope mounted with an AMT XR-111 camera. Control stainings revealed no background or nonspecificity (Supplemental Fig. 5).

Cells were washed with cold PBS containing 1 mM Na₃VO₄ and lysed in a solution containing 1% NP-40, 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA (pH 8), 1.5 mM EGTA, 1 mM Na₃VO₄, 50 mM NaF, 10 mM Na₄P₂O₇, and complete protease inhibitors (27). To increase lysis efficiency, whole-cell extracts were passed several times through a 27G needle (for coimmunoprecipitation experiments) or incubated at −70°C. Insoluble material was removed by centrifugation for 15 min at 21,000 g, upon which protein concentrations were determined using the bicinchoninic acid protein assay kit (Pierce). For immunoprecipitation (IP), whole-cell extracts (1 mg) were cleared for 1 h with Sepharose-Protein A beads (Life Technologies, 101041) previously blocked with TBS containing 5% BSA, and then incubated for 2 h with anti-NF-κB (Cell Signaling, 8242) preadsorbed onto BSA-blocked Sepharose-Protein A beads. Immune complexes were collected by brief centrifugation and washed thrice with lysis buffer containing protease and phosphatase inhibitors. Whole-cell extracts and IP samples were boiled (100°C) for 6 min in SDS sample buffer and migrated in SDS-PAGE gels. Proteins were transferred onto Hybond-ECL membranes (Amersham Biosciences, 10600003), blocked for 2 h in TBS-0.1% Tween containing 5% BSA, incubated with primary Abs (diluted in TBS-0.1% Tween containing 5% BSA) overnight at 4°C, and thence with HRP-conjugated anti-rabbit or anti-mouse IgG Abs from GE Healthcare (NA931V and NA934V) for 1 h at room temperature. For IP experiments, HRP-linked Abs specific to the Fc portion (Life Technologies, 31463) or to native rabbit IgG (Cell Signaling, L27A9) were used. Washed membranes were incubated in ECL (GE Healthcare, RPN2106) and immunodetection was achieved via chemiluminescence (34). Densitometric analysis of Western blot bands was done using the AlphaEase FC software (α Innotech) with β-actin used as loading control.

RNA from BMDCs and JAWS-II cells was extracted using TRIzol (Life Technologies, 15596026) as per the manufacturer’s protocol. Purified RNA was reverse-transcribed (36) and cDNA was submitted to quantitative RT-PCR analysis using the iTaq Universal SYBR Green Supermix (Bio-Rad, 1725121). Samples were amplified in duplicate on an MX3000P instrument (Stratagene), and the following primers were used. Inos: 5'-CTGCTGGTGGTGACAAGCACATTT-3' (AD529F) and 5'-ATGTCATGAGCAAAGGCGCAGAAC-3' (AD530R); Tnf: 5'-GACGTGGAAGTGGCAGAAGAG-3' (AD537F) and 5'-TGCCACAAGCAGGAATGAGA-3' (AD538R); Il6: 5'-ACAACCACGGCCTTCCCTACTT-3' (AD539F) and 5'-CACGATTTCCCAGAGAACATGTG-3' (AD540R); and Hprt: 5'-GTTGGATACAGGCCAGACTTTGTTG-3' (AD55F) and 5'-GATTCAACTTGCGCTCATCTTAGGC-3' (AD56R). MxPro-MX2000P software (Stratagene) was used to record cycle threshold (C_T) values, which were analyzed by the 2^-ΔΔCT method. Data were normalized to Hprt and expressed in terms of fold change differences relative to nonstimulated cells.

JAWS-II cells expressing either shScr or shSec22b were seeded onto 24-well plates in complete JAWS-II medium. After 24 h, the medium was replaced with Opti-MEM I reduced serum medium (Life Technologies) supplemented with 10% heat-inactivated FBS, without antibiotics. Cells were transfected with 750 ng of the NF-κB-Firefly luciferase reporter construct pGL2-NF-κB-Luc (provided by Dr. P. Duplay, INRS) and 30 ng of the Renilla luciferase control vector pRL-TK (Promega). Plasmids were premixed for 10 min at room temperature and then incubated for 30 min at room temperature with TransIT-TKO reagent (MJS BioLynx MIR2152) (37) in Opti-MEM I reduced serum medium, in the absence of serum or antibiotics. Lipid:DNA complexes were subsequently deposited dropwise on different areas of the wells, followed by gentle rocking from side to side. At 24 h posttransfection, complexes were removed, and cells were stimulated with 100 ng/ml LPS for 6 h, collected in microtubes, washed with PBS, and lysed in Passive Lysis Buffer (Promega, E1941) for 15 min. Finally, Firefly and Renilla luciferase activities were measured by analyzing 20 μl of extracts with the Dual-Luciferase Reporter Assay System (Promega, E1910), according to the manufacturer’s instructions, using a Lumat LB 9507 luminometer (EG & G Berthold).

Statistical differences between two or multiple groups were assessed via the Mann–Whitney U test or one/two-way ANOVA followed by Bonferroni post hoc tests, respectively. Two-way ANOVA was used to evaluate time-dependent differences in inflammatory mediator release. Differences were considered significant when p < 0.05, and graphs were constructed using GraphPad Prism 5.0 (GraphPad Software).

All data are contained within the article and its supporting information.

In BMDCs, the release of NO is induced by both soluble and particulate stimuli (Fig. 1A). To elucidate the underlying mechanisms and pathways, we first assessed the localization of iNOS in LPS-stimulated BMDCs, using immunofluorescence confocal microscopy. (Fig. 1B shows that iNOS was present in vesicular structures situated in the cytoplasm and plasmalemma, consistent with its role in NO secretion to the extracellular milieu (16, 38). To further determine the subcellular localization of iNOS in BMDCs, we performed colocalization analyses by immunofluorescence confocal microscopy using markers of the ER (CNX, PDI), the ERGIC (ERGIC53, Sec22b, Stx5), the Golgi apparatus (P115), the nucleolus (fibrillarin), and mitochondria (TOMM20) (Fig. 1C, 1D and Supplemental Fig. 1A). These analyses revealed that iNOS was widely present in the ERGIC and the Golgi, and absent from nucleoli and mitochondria (Fig. 1C, 1D and Supplemental Fig. 1A). Moreover, following phagocytosis of zymosan particles, we observed that iNOS colocalized with Sec22b, Stx5, and LAMP1 on phagosomes (Fig. 2A, 2B and Supplemental Fig. 1B). These findings suggested a role for the secretory pathway in NO secretion. Unlike endothelial NO synthase (39), iNOS lacks a transmembrane domain but binds to membranes (40) via interactions between its C-terminal domain and ezrin-binding adaptor protein EBP50 (38). With this in mind, we asked whether iNOS was present on the phagosome membrane. To this end, we isolated phagosomes from BMDCs that were fed with LPS-coated magnetic beads. (Fig. 2C). Western blot analysis of the phagosomes confirmed the presence of iNOS. To test whether iNOS was in the cytosolic or luminal side of the phagosome, we incubated intact purified phagosomes with proteinase K. Because iNOS was digested (Fig. 2C), we concluded that it faces the cytoplasm. Collectively, these data demonstrate that iNOS is on the cytoplasmic surface of the organelles that it colocalized with.

The ER-ERGIC-Golgi circuitry is critical for the biogenesis and secretion of proinflammatory cytokines such as TNF and IL-6 (4). Here, we assessed the potential role of ER-Golgi trafficking in the control of iNOS activity and NO release by BMDCs, and we included TNF and IL-6 in our analysis because their release is mediated by the secretory pathway (4). To this end, we treated BMDCs with DMSO or with chemical inhibitors that affect the secretory pathway at different steps (brefeldin A, monensin, and retro-2), prior to stimulation with LPS. Brefeldin A targets GBF1 thereby preventing COP-I recruitment (41). This induces movement of proteins from the Golgi back to the ER, thereby disrupting Golgi structure and blocking protein secretion (42). This can be observed by the dispersal of Golgi protein P115 (Supplemental Fig. 2) (43). On the other hand, monensin is a Na⁺/H⁺ ionophore that blocks endosomal acidification (44), intra-Golgi trafficking from the medial to the trans Golgi complex (45, 46). Monensin treatment also leads to Golgi dispersal (47), as verified by P115 staining in Supplemental Fig. 2. Retro-2 has a more selective effect: it blocks retrograde transport to the ER (48) and did not induce Golgi dispersal (Supplemental Fig. 2).

Although monensin and brefeldin A inhibited NO, TNF, and IL-6 release, retro-2 only inhibited TNF release, highlighting individual differences in the intracellular trafficking of those molecules (Fig. 3A–C). Of note, both monensin and brefeldin A inhibited iNOS protein expression (Fig. 3D, 3E), suggesting that ER-Golgi trafficking is part of the pathway leading to iNOS expression. Inhibition of LPS-induced iNOS protein expression in monensin-treated cells was the consequence of reduced iNOS mRNA levels, as determined by quantitative RT-PCR (Fig. 3F), suggesting that the secretory pathway plays a role in the function of LPS-mediated signaling cascades leading to iNOS gene expression. To understand how perturbation of the secretory pathway lead to the inhibition of NO and cytokine release, we first assessed the impact of monensin on the integrity of LPS-induced signal cascades associated with the expression of iNOS, TNF, and IL-6 (49, 50). As shown in (Fig. 4A and Supplemental Fig. 3A, pretreatment with monensin altered the phosphorylation kinetics of JNK (p54) but had no effect on LPS-induced phosphorylation of p44/p42 MAPK (ERK1/2). Monensin-treated BMDCs also exhibited lower IκBα degradation. One of the key downstream events of inducible IκBα degradation is the nuclear translocation of the transcription factor NF-κB, which mediates the expression of a large number of LPS-inducible genes including iNOS, TNF, and IL-6 (49, 50). As shown in (Fig. 4B and 4C, pretreatment of BMDCs with monensin impaired LPS-induced nuclear translocation of NF-κB. Collectively, these findings are consistent with the notion that the integrity of the secretory pathway is necessary for optimal NF-κB–mediated LPS responses in BMDCs.

The SNARE Sec22b plays a role in ER-Golgi trafficking (51) and in the unconventional secretion of IL-1β in cells treated with autophagy inducers, a process known as secretory autophagy (52). To assess the potential role of Sec22b in the release of NO, we stimulated with LPS dendritic cell–like JAWS-II cells transduced with an shRNA to Sec22b (shSec22b) or a scrambled (shScr) sequence (10). As shown in (Fig. 5A–C, Sec22b knockdown diminished the release of NO, as well as TNF and IL-6. To confirm the role of Sec22b in NO secretion and to test whether this mechanism operates in macrophages, we stimulated with LPS RAW264.7 macrophages pretreated with siRNAs to Sec22b, its binding partner Stx5, or both. In that regard, knockdown of one or both SNAREs resulted in the inhibition of NO release (Fig. 5D). Akin to the effect of pharmacological inhibitors on iNOS expression, Sec22b knockdown diminished iNOS protein levels (Fig. 5E), as well as Inos, Tnf, and Il6 expression, as assessed by quantitative RT-PCR (Fig. 5F).

To ensure that the function of Sec22b on inflammatory mediator production is not contingent on LPS stimulation, we treated Sec22b knockdown cells with agonists to TLR2 (PGN), dectin-1/TLR2 (zymosan), TLR2 (L. major promastigotes), TLR7 (imiquimod), or TLR9 (CpG). Twenty-four hours poststimulation, silencing Sec22b abrogated the induction of NO, TNF, and IL-6 by all the tested ligands (Fig. 6). Hence, the phenotype is not exclusively dependent on TLR4.

To understand how Sec22b regulates LPS-induced responses, we evaluated the impact of Sec22b knockdown on LPS-induced signal transduction cascades leading to the expression of iNOS. Apart from an early but temporary impairment of JNK (p54) phosphorylation, and IκBα degradation, silencing of Sec22b in JAWS-II dendritic cells had no significant effect on LPS-induced signaling events and on NF-κB expression (Fig. 7A and Supplemental Fig. 3B). To determine whether Sec22b is involved in the nuclear translocation of NF-κB, we stimulated JAWS-II cells expressing either an shScr or an shSec22b with LPS and quantified NF-κB nuclear translocation using imaging flow cytometry (Fig. 7B, 7C) and immunofluorescence confocal microscopy (Fig. 7D, 7E). Our results show that knockdown of Sec22b hampered LPS-induced NF-κB nuclear translocation. We made a similar observation in LPS-stimulated RAW264.7 macrophages treated with siRNAs to Sec22b (Fig. 7F, 7G). To further characterize the impact of Sec22b knockdown on the inhibition of NF-κB translocation, we transfected JAWS-II cells expressing either shScr or shSec22b with an NF-κB luciferase reporter construct prior to stimulation with LPS. As shown in (Fig. 7H, activation of the NF-κB reporter was significantly impaired in Sec22b knockdown cells. These data indicate that the secretory pathway–resident SNARE Sec22b modulates proinflammatory effector production by participating in the nuclear translocation of NF-κB.

Because Sec22b is present in the ER-ERGIC-Golgi circuit (8, 53), we postulated that NF-κB and Sec22b are found in the same compartments. To test this, we used immunofluorescence to examine the colocalization of NF-κB with Sec22b, as well as other ERGIC and Golgi proteins. In control and LPS-stimulated BMDCs, we found partial colocalization under steady-state conditions, which disappeared following LPS-induced migration of NF-κB to the nucleus (Fig. 8A, 8B, Supplemental Figs. 4 and 5). We also observed that Sec22b was present in the nucleus; LPS slightly, but significantly, increased the proportion of Sec22b in the nucleus compared with the entire cell (Fig. 8C). Similarly, immunogold electron microscopy revealed that NF-κB and Sec22b co-occur in membranous structures in the cytoplasm, and increasingly in the nucleus after LPS stimulation (Fig. 8D and Supplemental Fig. 5). Both NF-κB and Sec22b were found in nuclear regions that do not costain with DAPI. Because nucleoli also exhibit this characteristic (54), we asked whether Sec22b colocalizes with nucleolar marker fibrillarin (55), and found that Sec22b was not found in these structures (Fig. 8E, 8F and Supplemental Videos 1 and 2). Our observation that Sec22b and NF-κB partially colocalize led us to determine whether these proteins physically interact. To this end, we performed coimmunoprecipitation experiments in BMDCs under steady-state conditions. As shown in (Fig. 8G, NF-κB and Sec22b did not coimmunoprecipitate, in contrast to NF-κB and its inhibitor IκBα that are known to exist as a complex in steady-state cells. This observation indicates that NF-κB and Sec22b are present in the same compartments but do not form complexes. Collectively, these findings revealed that NF-κB is an ER-Golgi–associated protein that necessitates Sec22b to translocate to the nucleus.

In dendritic cells, microbial-derived molecules induce the expression of genes encoding proinflammatory mediators and anti-microbial molecules through complex signaling cascades that culminate in the activation of specific transcription factors (56). In the current study, we aimed at further elucidating the mechanisms and pathways associated with the production of NO and inflammatory mediators in BMDCs exposed to LPS. Our main finding is that the secretory pathway, through the action of the SNARE Sec22b, plays a central role in LPS-induced responses by regulating the nuclear translocation of the transcription factor NF-κB (see Visual Abstract).

Whereas the signaling pathways leading to the expression of iNOS and the roles of NO have been explored in great detail in various immune cell types (16, 18), much less is known on the subcellular localization of iNOS. Using electron microscopy and biochemical approaches, Vodovotz and colleagues (22) reported that over 40% of the iNOS activity in whole sonicates was particulate in LPS-activated primary macrophages. They also observed that iNOS was associated with 50–80 nm vesicles of undefined nature, which did not correspond to endosomes, lysosomes, or peroxisomes. Interestingly, the most intense iNOS labeling observed by immunoelectron microscopy was perinuclear, on the trans side of the trans Golgi network (22). Upon phagocytosis, iNOS was found to associate to phagosomes through an actin-dependent mechanism (17). Our results demonstrated that iNOS co-occurs, to varying extents, with proteins found in the ER, ERGIC, and Golgi apparatus. Indeed, iNOS colocalized with ER-Golgi SNAREs Sec22b and Stx5, raising the possibility that the previously observed iNOS-positive vesicular structures originate from these organelles. Because iNOS does not have a signal sequence directing it to the ER lumen (http://www.cbs.dtu.dk/services/TargetP (57)), iNOS is likely to be found on the cytoplasmic side of these secretory pathway organelles via interactions with other proteins. We showed that this topology is conserved in phagosomes, because iNOS was degraded when phagosomes were treated with proteinase K.

Given the large number of genes controlled by NF-κB, this ubiquitous transcription factor plays a central role in the regulation of inflammation, and innate and adaptive immune responses, including LPS-induced responses in dendritic cells (56, 58). In resting cells, NF-κB dimers are sequestrated in an inactive state in the cytoplasm by a member of the inhibitory proteins known as IκBα (59), preventing NF-κB from being recognized by the nuclear import machinery (60). Cell stimulation triggers signaling cascades leading to the phosphorylation of IκB, its ubiquitination, and degradation by the proteasome (61, 62), enabling the nuclear import of NF-κB dimers, their binding to consensus DNA sequences, and the activation of target gene expression (49). Nucleocytoplasmic trafficking of macromolecules such as transcription factors is tightly controlled by importins and exportins, through the recognition of nuclear localization signals and nuclear export signals (63). Using pharmacological inhibitors, we obtained evidence that the ER-Golgi circuitry contributes to optimal LPS-induced responses in BMDCs, through the regulation of MAPK phosphorylation and NF-κB nuclear translocation. The fact that ERK1/2 and p38 are involved in LPS-induced NO release (64) led us to the finding that secretory pathway blockade deregulated JNK (p54) phosphorylation and precluded IκBα degradation. This may explain in part why NO secretion decreases. Consistent with our findings, previous studies with RAW264.7 macrophages and with primary mouse astrocytes revealed that ER stress induced by either tunicamycin or brefeldin A inhibited LPS-induced iNOS expression (65, 66). These findings highlight a previously unappreciated role for the ER-Golgi secretory pathway in the translocation of NF-κB to the nucleus.

Through their role in membrane fusion, the ER/ERGIC-resident SNAREs Sec22b and Stx5 participate in various cellular processes including secretory autophagy (52), embryonic development (15), and plasma membrane expansion (67). These two SNAREs also play a key role in phagosome maturation and function by regulating the delivery of ER- and ERGIC-resident proteins to phagosomes in macrophages and dendritic cells. Remarkably, we found that Sec22b was needed for the expression, at the mRNA and protein levels, of iNOS, TNF, and IL-6. In immune cells, LPS-induced responses are largely dependent on NF-κB, as it regulates the expression of several genes including iNOS, IL-6, and TNF (49). Sec22b knockdown hindered MAPK phosphorylation and IκBα degradation, which implicates this SNARE in the MAPK-mediated regulation of NO release (64). Aside from the effect of Sec22b on IκBα degradation and NF-κB translocation, we do not exclude the possibility that Sec22b knockdown alters ER-Golgi physiology to the extent that it may alter MAPK phosphorylation via an indirect mechanism. Because of their amenability to genetic manipulation (10, 37), we used the immortalized dendritic cell line JAWS-II to study the impact of Sec22b knockdown on NF-κB dynamics. Although different (68), JAWS-II cells are derived from BMDCs of C57BL/6 mice and behave very similarly in a variety of contexts ranging from infection to Ag presentation (10, 69–71). Our findings were also corroborated with the RAW264.7 cell line. Whether Sec22b controls NF-κB translocation in other leukocytes remains to be determined.

The role of Sec22b on NF-κB translocation was surprising but not without precedent. For one, ER- and Golgi-resident proteins such as HSP90 and golgin-97 have been implicated in the translocation of NF-κB to the nucleus (72, 73). Because we observed that NF-κB was present in the ERGIC and Golgi, it is reasonable to postulate that Sec22b plays an active role in the cytoplasmic–nuclear shuttling of NF-κB. Second, SNAREs and their chaperones are required for nuclear pore assembly and sealing (75). Our finding that Sec22b and NF-κB co-occurred in the cytoplasm and nucleus further supports the notion that Sec22b is part of the circuitry that, with help from importins (60, 63), shepherds NF-κB translocation.

There is growing evidence that SNAREs and alternatively spliced isoforms access the nucleus where they perform unsuspected functions (74–79). In a study aimed at determining the subcellular localization of Stx17, it was found that this SNARE is present in both the cytoplasm and nucleus of several cell types (79). The absence of homology with known nuclear localization signal consensus sequences led the authors to speculate about the existence of a yet uncharacterized nuclear transport mechanism for Stx17. The meaning of the presence of Stx17 in the nucleus remains to be elucidated. Similarly, the SNARE SNAP-47 was found in the ER, ERGIC, and nucleus of HeLa cells (77). Further analyses revealed the presence of several putative nuclear localization signals and two nuclear export signals in SNAP-47. Treatment of HeLa cells with leptomycin B, which inhibits CRM-1/exportin1-dependent nuclear export led to nuclear accumulation of SNAP-47, suggesting a functional nuclear export signal in SNAP-47 (77). SNAP-47 was previously shown to interact with Stx3 (76). Interestingly, a splice variant of Stx3 was found to undergo nuclear translocation, to bind to several transcription factors, and to act as a transcriptional activator (74). An isoform of Stx1, lacking a transmembrane domain, was also found to localize in the nucleus (78). The observation that Sec22b was present in nonnucleolar compartments of the nucleus raises the interesting possibility that Sec22b partakes in nuclear dynamics and gene regulation. Because the a, b, and c isoforms of Sec22 possess transmembrane domains (53), we do not exclude the possibility that Sec22b undergoes cleavage prior to its nuclear translocation. Hence, how Sec22b migrates to the nucleus is the subject of future research.

In sum, our findings demonstrated that the SNARE Sec22b controls inflammatory effector release by facilitating the translocation of NF-κB from the secretory pathway to the nucleus.

We are grateful to D. S. Amigorena for the kind gift of shRNA-transduced JAWS-II cells, J. Tremblay for assistance in immunofluorescence experiments, and A. Nakamura for assistance in electron microscopy. The visual abstract was created with BioRender.com.

The authors have no financial conflicts of interest.