ALG2 regulates type I interferon responses by inhibiting STING trafficking

The innate immune system, the first line of host defense, is critical for a host to combat microbial invasion including by viruses and bacteria (Zhang et al., 2020b). In the host cell, pattern recognition receptors (PRRs) are responsible for the detection of microbial pathogens via combining with unique pathogen-associated molecular patterns (PAMPs) (Konno et al., 2013; Tsuchida et al., 2010). The microbial components, such as viral DNA, viral RNA or lipopolysaccharide, serve as a PAMP that is recognized by distinct PRRs, including cyclic GMP–AMP (cGAMP) synthase (cGAS), Toll-like receptors (TLRs) or RIG-I-like receptors (RLRs) (Sun et al., 2013; Zhong et al., 2009). After being activated by PAMPs, PRRs initiate a series of intracellular signaling cascades, causing the production of type I interferons, pro-inflammatory cytokines and other downstream proteins that all facilitate the elimination of infected cells and microbial pathogens, so as to maintain the homeostasis of the organism (Liu et al., 2015).

For eukaryotic cells, cytosolic exposure to double-stranded DNA from microbial pathogens or damaged cells is a danger signal and can trigger the host cell's innate immune response (Ablasser et al., 2013; Wang et al., 2016). Several proteins have been identified as PRRs to recognize cytosolic DNA, including IFI16, AIM2, DDX41, DAI (also known as ZBP1), TLR9 and LSm14A (Zhou et al., 2014). Although various evidence has demonstrated that these proteins are important for cytosolic DNA-induced innate immune responses, they are all not universally engaged in detecting cytosolic DNA in distinct cell types (Zhou et al., 2014). cGAS, a nucleotidyltransferase family member, has been identified as the PRR to detect cytosolic DNA in multiple cell types, especially immune cells (Sun et al., 2013). Upon binding to cytosolic DNA, cGAS catalyzes the synthesis of the secondary messenger cGAMP, which binds to and activates the downstream adaptor stimulator of IFN genes (STING) (Wu et al., 2013).

STING (also known as STING1, MITA, ERIS, TMEM173 or MPYS) is a transmembrane (TM) protein that is localized to the endoplasmic reticulum (ER) in the resting state (Ishikawa and Barber, 2008; Jin et al., 2008; Sun et al., 2009; Zhong et al., 2008). After being activated, STING translocates from the ER to perinuclear compartments, including the Golgi and endosomes, where STING recruits and activates TANK-binding kinase 1 (TBK1) and the transcription factor IFN regulatory factor 3 (IRF3), ultimately leading to the production of type I interferons and various other cytokines (Ishikawa et al., 2009). Blockade of ER–Golgi traffic using brefeldin A abolishes the STING-mediated signaling cascades, resulting in the inability to produce type I interferons (Dobbs et al., 2015; Ishikawa et al., 2009). In addition, gain-of-function STING mutations, which are found in patients with the autoinflammatory disease known as STING-associated vasculopathy with onset in infancy (SAVI), localize to the Golgi and are constitutively activated without cGAMP (Liu et al., 2014). These findings indicate that the correct location of STING is essential for its proper function. However, why STING must reside on the ER and traffic to the perinuclear vesicles to activate TBK1 and IRF3 has yet to be elucidated.

One of the proteins involved in ER–Golgi trafficking is apoptosis-linked gene 2 (ALG2), also known as programmed cell death 6 (PDCD6), which has been identified as a pro-apoptotic protein in T-cell apoptosis (Vito et al., 1996). ALG2 is composed of a short Ala/Gly/Pro-rich N-terminal tail and five consecutive EF hands, of which EF hands 1 and 3 can coordinate two Ca²⁺ ions (Jia et al., 2001). When coordinated by Ca²⁺, ALG2 changes conformation and exposes the hydrophobic pockets on its surface, leading to the binding of downstream effector partners. Thus, ALG2 can function as a Ca²⁺ sensor in the cytosol (Maki, 2020; Maki et al., 2011). For instance, the Ca²⁺-dependent interaction between ALG2 and SEC31A, a component of the outer layer of COPII, is critical for ER–Golgi vesicle transport (Shibata et al., 2007; Yamasaki et al., 2006). ALG2 can also translate a transient rise in cytosolic Ca²⁺ levels into more persistent SEC31 ubiquitylation, leading to the formation of large COPII coats and promoting collagen secretion (McGourty et al., 2016). Furthermore, by interacting with TSG101 and ALIX (also known as PDCD6IP), ALG2 can influence the endosome sorting complex required for transportation (ESCRT) system in multiple steps (Missotten et al., 1999; Okumura et al., 2009). Thus, a large amount of evidence has shown that ALG2 plays an important role in multiple intracellular cargo trafficking processes.

In this study, we identified ALG2 as a negative regulator of STING. Loss of ALG2 facilitated the cytosolic DNA-induced and STING-dependent innate immune responses and promoted the expression of type I interferons upon cGAMP stimulation or HSV-1 infection. By interacting with the C-terminal tail (CTT) of STING, ALG2 inhibited STING trafficking to perinuclear compartments. In addition, we observed that the Ca²⁺ coordination ability of ALG2 was essential for its function in STING-mediated innate immune responses. These findings suggest, for the first time, that ALG2 plays an important role in regulation of cytosolic DNA-induced innate immune responses and expand our knowledge on the molecular mechanisms of STING trafficking.

To identify new molecules involved in STING-dependent innate immune responses, STING was precipitated with glutathione–sepharose resin from lysates of THP-1 cell lines stably expressing STING–GST, and co-captured proteins were identified using mass spectrometry. From a set of STING-interacting proteins, ALG2, a Ca²⁺-binding protein, was identified as a new interacting protein and was subjected to further analysis (Table S1).

To gain insights into the influence of ALG2 on cytosolic DNA-induced innate immune responses, two different single-guide RNAs (sgRNAs) targeting the human ALG2 (sgALG2) were designed (Table S2), and ALG2 knockout THP-1 cell lines were generated using the CRISPR-Cas9 system. Wild-type (WT) and ALG2 knockout THP-1 cells were permeabilized with perfringolysin O (PFO) and incubated with cGAMP, then the cells were harvested for RNA extraction and quantitative reverse transcription PCR (qRT-PCR) analysis. Compared with WT cells, ALG2 knockout cells exhibited significantly enhanced expression of IFN-β (also known as IFNB1), ISG56 (also known as IFIT1) and IP10 (also known as CXCL10) after stimulation with cGAMP (Fig. 1A). The protein levels of ALG2 in WT and ALG2 knockout THP-1 cells were detected by western blotting, showing that sgALG2 was highly efficient at depleting endogenous ALG2 (Fig. 1B). To further confirm the effect of ALG2 on cytosolic DNA-induced innate immune responses, herpes simplex virus-1 (HSV-1), a human pathogenic DNA virus, was used to infect THP-1 cell lines. Consistent with cGAMP treatment, depletion of ALG2 could also augment the expression of IFN-β, ISG56 and IP10 after infection with HSV-1 (Fig. 1C). In order to eliminate the influence of sgRNA missing its target in the experiment, ALG2, in these knockout THP-1 cell lines, was reconstituted by exogenous expression and innate immune responses were evaluated. As expected, exogenous expression of ALG2 in these knockout THP-1 cell lines rescued the phenotype caused by ALG2 deficiency, i.e. the expression levels of IFN-β, ISG56 and IP10 after HSV-1 infection were restored to those of WT cells (Fig. 1D–F). Taken together, these data strongly suggest that ALG2 is a negative regulator of cytosolic DNA-induced innate immune responses.

In Homo sapiens, STING is a 379-amino acid protein, which contains four TM segments (residues 1–134) in its N terminus and the cytoplasmic region [C-terminal domain (CTD), 135–379]. The CTT (343–379) of STING, part of the CTD domain (Srikanth et al., 2019), is indispensable for its function in innate immune response (Fig. 2A). When activated, the CTT of STING is responsible for recruiting and activating TBK1 and IRF3, which leads to the production of type I interferons and various other cytokines.

To further confirm the interaction between ALG2 and STING, which was identified by mass spectrometry, GFP-Trap co-immunoprecipitation experiments were performed in HEK293T cells. Compared with GFP, GFP–ALG2 showed a significant interaction with STING (Fig. 2B). Next, the interaction domains of ALG2 and STING were further investigated. Because ALG2 was distributed in the cytoplasm or nucleus, the interaction between the two proteins should be attributed to the CTD of STING. The residues 135–379 (CTD), 135–342 or 343–379 (CTT) of STING were fused with GST and co-expressed with ALG2 in HEK293T cells for GFP-Trap co-immunoprecipitation experiments. The results showed that the CTT of STING was responsible for ALG2 interaction, while the region binding cGAMP (135–342) was barely detectable after GFP-Trap (Fig. 2C).

As a Ca²⁺-binding protein, Ca²⁺ coordination is necessary for ALG2 to interact with its partners in most instances (Maki et al., 2011). Therefore, we tested the interaction between ALG2 and STING in the presence of EGTA, the Ca²⁺ chelator, or Ca²⁺. Unexpectedly, both EGTA and Ca²⁺ had little influence on their interaction (Fig. 2D,E). To further confirm this, the interactions between STING and ALG2 mutants, E47D or E114D, which lose Ca²⁺ coordination ability, were evaluated. Consistent with previous results, the interaction between the two proteins was not abolished by the mutations in ALG2 (Fig. 2F,G).

Because ALG2 interacted with STING, as confirmed by co-immunoprecipitation, we examined whether ALG2 and STING colocalized within cells. Given that the transient overexpression of exogenous proteins could lead to their mislocalization, we established HeLa cell lines stably expressing STING–GFP. The endogenous ALG2 of HeLa STING–GFP cells was immunostained by anti-ALG2 antibody and TRITC-conjugated secondary antibody, followed by observation with confocal microscopy. The results showed that endogenous ALG2 indeed colocalized with STING (Fig. 3A). Colocalization between endogenous ALG2 and STING was also verified in THP-1 cell lines stably expressing STING–GFP (Fig. 3B). Additionally, in THP-1 GFP–ALG2 cell lines, we observed that endogenous STING colocalized with ALG2 (Fig. 3C). These data confirmed that ALG2 and STING interact and colocalize.

Although the detailed underlying mechanisms have yet to be elucidated, it is well known that activated STING must traffic from the ER to perinuclear vesicles to activate TBK1 and IRF3 (Ishikawa et al., 2009). By interacting with SEC31A, TSG101 and ALIX, ALG2 plays an important role in intracellular cargo trafficking (Maki et al., 2016). These connections prompted us to investigate whether ALG2 plays a role in STING translocation.

To verify the influence of ALG2 on STING translocation, endogenous STING in WT and ALG2 knockout THP-1 cells, stimulated by cGAMP, was immunostained by anti-STING antibody and TRITC-conjugated secondary antibody, followed by detection with confocal microscopy. As shown in Fig. 4A, ALG2 knockout strongly promoted STING translocation to the perinuclear region (Fig. 4A,B). Knockdown of ALG2 expression in HeLa STING–GFP cells by short hairpin RNA (shRNA) could also promote STING translocation to the perinuclear region (Fig. 4C–G). Furthermore, the influence of ALG2 overexpression on STING translocation was assessed. HeLa STING–GFP cells were transfected with RFP–ALG2 for 2 days before permeabilization with digitonin and incubation with cGAMP. We deliberately captured RFP–ALG2 overexpression cells and normal cells in one field of view, and found that RFP–ALG2 could markedly inhibit STING translocation to the perinuclear region (Fig. 4H,I). After activation and trafficking to the perinuclear region, STING could be phosphorylated, which was important for it to stimulate downstream signaling. Therefore, we examined the effect of ALG2 on STING phosphorylation. Consistent with increased perinuclear location of STING in ALG2 knockout cells, phosphorylation of STING was also increased in ALG2 knockout THP-1 cells after stimulation with cGAMP (Fig. S1A,B). It has been shown that STING is rapidly degraded after stimulation (Gonugunta et al., 2017). We wondered whether ALG2 would affect the degradation of STING. As shown in Fig. S1C,D, ALG2 knockout facilitated the phosphorylation of STING but did not affect its degradation. Collectively, these results indicate that, when STING was activated, ALG2 could inhibit its trafficking from the ER to the perinuclear region.

Although ALG2 colocalizes with STING in resting state (Fig. 3) and reduces STING translocation to the perinuclear region following cGAMP treatment (Fig. 4), we found that the perinuclear-localized STING did not colocalize with ALG2 anymore, i.e. ALG2 did not traffic with STING, as shown in Fig. 4H. To further address this, we examined the association between ALG2 and activated STING by cGAMP, using GFP-Trap co-immunoprecipitation. Colocalization between endogenous ALG2 and activated STING was also evaluated in HeLa STING–GFP cells. Consistent with the observations shown in Fig. 4H, both the biochemical association and colocalization between ALG2 and STING were strongly attenuated by stimulation of STING (Fig. 5A–C). In addition, as shown in Fig. 3A, it seems that ALG2 and STING were both present at ER exit sites (ERES), which were the regions for assembling vesicles for export. To confirm this, the colocalization among STING, ALG2 and SEC13, an independent marker for ERES, was assayed. The results indicated that ALG2 and STING were indeed at ERES at steady state (Fig. 5D). Interestingly, SEC13, rather than ALG2, was trafficked to the perinuclear region with STING after stimulation by cGAMP (Fig. 5E). Additionally, we also observed that the dotted ERES-localized ALG2 became dispersed after cGAMP treatment, suggesting that the localization of ALG2 was disturbed during STING translocation (Fig. 5D,E).

As shown in Fig. 2C, results indicated that the CTT of STING was responsible for its interaction with ALG2, thus STING(1–342), lacking the CTT, should lose its interaction with ALG2. To confirm this, we obtained HeLa and THP-1 cell lines stably expressing STING(1–342)–GFP and examined the colocalization between STING(1–342)–GFP and endogenous ALG2. As expected, there was no colocalization between STING(1–342)–GFP and endogenous ALG2 in HeLa and THP-1 cells (Fig. 6A,B).

Previous study indicates that although STING(1–342) loses the ability to recruit and activate TBK1 and IRF3, it can still traffic from ER to the perinuclear region (Gui et al., 2019; Wu et al., 2019). In view of this, we evaluated the influence of ALG2 on STING(1–342) trafficking using HeLa STING(1–342)–GFP cells, which were infected with sh-CT (negative control shRNA) or sh-ALG2 (shRNA targeting ALG2) viruses for 3 days and stimulated by cGAMP. As expected, owing to the loss of interaction and colocalization between the two proteins, STING(1–342) trafficking to the perinuclear region was no longer affected by ALG2 knockdown (Fig. 6C–G). In addition, ALG2 overexpression had no influence on STING(1–342) translocation (Fig. 6H,I). These data demonstrate that the effect of ALG2 on regulation of STING trafficking relies on the presence of the CTT in STING.

The point mutations E47D and E114D, impairing Ca²⁺ coordination of ALG2 and abolishing the interaction between ALG2 and most of its binding partners, lead to dysfunction of ALG2 (He et al., 2020). With these features in mind, we investigated the impact of these ALG2 mutations on regulation of STING trafficking and DNA-induced innate immune responses. First, consistent with the GFP-Trap co-immunoprecipitation results in HEK293T cells (Fig. 2F,G), ALG2 E47D and ALG2 E114D colocalized with STING in HeLa STING–GFP cells (Fig. 7A,B). Moreover, the ALG2 mutants also colocalized with SEC13 and STING at steady state, and did not traffic to the perinuclear region with STING after stimulation by cGAMP (Fig. S2). Then, HeLa STING–GFP cells were transfected with RFP–ALG2 E47D or RFP–ALG2 E114D for 2 days, stimulated with cGAMP and subsequently imaged by confocal microscopy. Surprisingly, in contrast to ALG2, both ALG2 E47D and ALG2 E114D had no influence on STING trafficking to the perinuclear region (Fig. 7C,D). Furthermore, as shown in Fig. 5D and E, we again observed the dotted ERES-localized ALG2 mutants disperse after cGAMP treatment (Fig. 7A–D; Fig. S2), implying that STING is also critical for the localization of ALG2 mutants in these cells. To further demonstrate the effect of ALG2 E47D and ALG2 E114D on DNA-induced innate immune responses, rescuing experiments were performed by exogenously expressing ALG2, ALG2 E47D and ALG2 E114D in ALG2 knockout THP-1 cell lines, then the cells were infected by HSV-1 and analyzed by qRT-PCR. ALG2, but not ALG2 E47D and ALG2 E114D, restored the expression levels of IFN-β, ISG56 and IP10 as in WT cells, indicating a non-function of these mutants in regulating the expression of IFN-β, ISG56 and IP10 (Fig. 7E,F). Thus, the Ca²⁺-binding capability of ALG2 is important for its regulation of STING trafficking and DNA-induced innate immune responses.

Considering the fact that ALG2 E47D and ALG2 E114D displayed loss of function in regulating STING trafficking and DNA-induced innate immune responses but could still interact and colocalize with STING, we speculated that a third molecule might be involved in the ALG2–STING interaction, which might interact with ALG2 in a Ca²⁺-dependent manner. To explore this, two well-known ALG2-binding partners involved in vesicle trafficking, SEC31A and ALIX, were selected and tested for their effects on DNA-induced innate immune responses (Table S2). SEC31A deficiency slightly inhibited the expression of IFN-β, ISG56 and IP10 after stimulation (Fig. S3A–C), but did not affect the function of ALG2 in regulating DNA-induced innate immune responses (Fig. S3D–F). Furthermore, ALIX had no effect on DNA-induced innate immune responses (Fig. S4). Thus, both SEC31A and ALIX are not involved in ALG2 regulation of STING trafficking.

The cGAS–cGAMP–STING pathway is a major pathway to mediate DNA-induced innate immune responses, which defend against various pathogens that contain DNA or produce DNA, such as DNA viruses, retroviruses, bacteria and parasites (Zhang et al., 2020b; Zhou et al., 2014). Abnormal activation of cGAS and STING can cause autoimmune diseases such as systemic lupus erythematosus, rheumatoid arthritis, Aicardi–Goutieres syndrome and SAVI (Liu et al., 2014; Srikanth et al., 2019; Zhang et al., 2020a). Therefore, it is important to control STING signaling and to avoid undesired activation. The translocation of STING from the ER to perinuclear vesicles is a critical step in STING signaling, and has also been demonstrated to be a rate-limiting step in the activation of downstream signaling pathways (Dobbs et al., 2015). Although several proteins, including STEEP, TMED2, UBXN3B, SNX8 and iRhom2 (also known as RHBDF2), have been shown to be involved in the regulation of STING trafficking, the molecular mechanism underlying this process is still not fully understood (Sun et al., 2018; Zhang et al., 2020a). In this study, we found that ALG2 regulates DNA-induced innate immune responses via affecting the translocation of STING from the ER to perinuclear vesicles. The interaction between ALG2 and STING was confirmed using GFP-Trap co-immunoprecipitation and colocalization analysis, and the CTT of STING was determined to be responsible for binding to ALG2. Furthermore, our data show that the influence of ALG2 on STING trafficking relies on the interaction between the two proteins, because the trafficking of STING(1–342), lacking CTT, was not affected by ALG2.

STING undergoes a series of conformational changes during activation and translocation, accompanied by different oligomerization states. Certain proteins interact with STING in a conformation-dependent manner, thereby regulating the function of STING at distinct steps. For instance, STIM1 mainly associates with STING in resting state to retain it in the ER membrane, and the biochemical association between the two proteins can be reduced by stimulation of STING (Srikanth et al., 2019). EGFR only interacts with activated STING, and thereby determines its trafficking route and cellular innate immunity functions (Wang et al., 2020). Moreover, the CTT of STING, the binding region of ALG2, also undergoes dramatic conformational change upon activation. After cGAMP stimulation, the CTT of STING is relocated from the molecule and free to interact with TBK1 and IRF3, and thus is available for the induction of type I interferons and other cytokines. These results prompted us to investigate whether the interaction between ALG2 and STING is affected by STING activation. Our GFP-Trap co-immunoprecipitation experiments showed that the biochemical association between the two proteins was markedly attenuated by stimulation of STING. Additionally, the colocalization experiments supported these results. Moreover, we demonstrated that ALG2 and STING colocalized with SEC13 at ERES at steady state, and SEC13, instead of ALG2, trafficked to the perinuclear region with STING after stimulation by cGAMP. These observations suggest that ALG2 plays its role in the early stage of STING translocation, prior to or in concert with STING egress from the ER.

One interesting observation in our study is that ALG2 colocalizes with STING on ERES in resting state, and it seems that ALG2 ERES localization is STING dependent in HeLa STING–GFP cells. After STING egress from ERES upon activation, ALG2 becomes dispersed within cells and the dotted ERES localization cannot be observed, despite no co-translocation with STING (Figs. 5D,E and 7A–D; Fig. S2). Several proteins that colocalize with STING on ER or ERES have been identified, such as STIM1, iRhom2 and STEEP (also known as STEEP1) (Luo et al., 2016; Srikanth et al., 2019; Zhang et al., 2020a). Therefore, a complicated protein interaction network might exist on ERES, which modulates STING localization in resting state and influences STING translocation upon activation. Because canonical ALG2 mutations (E47D and E114D) do not affect its colocalization with STING, but lose the ability to affect STING translocation, we speculate that ALG2 might affect STING translocation through this protein network.

ALG2 was originally identified as a pro-apoptotic protein in T-cell apoptosis (Jang et al., 2002; Vito et al., 1996). Subsequently, increasing evidence has shown that ALG2 is a critical molecule in multiple processes, such as ESCRT-related vesicle transportation (Katoh et al., 2005), collagen secretion (McGourty et al., 2016), cell plasma membrane repair (Scheffer et al., 2014), lysosome motility (Li et al., 2016) and inhibition of human immunodeficiency virus (HIV) infection (Ma et al., 2016). By interacting with TSG101, ALIX and SEC31A, ALG2 plays an important role in intracellular cargo trafficking and protein localization, processes that are also crucial for the proper function of STING. Our results indicate that ALG2 can play a role in the early stage of STING translocation. By interacting with STING, ALG2 might tether STING in ER or interfere with the egress of STING from the ER. Further detailed molecular mechanisms remain to be investigated in the future.

Collective evidence suggests that Ca²⁺ signaling is intimately involved in STING-mediated innate immune responses. STIM1, a TM ER Ca²⁺ sensor, regulates STING by anchoring it to the ER under resting conditions, which prevents a spontaneous IFN response under normal conditions (Srikanth et al., 2019). Other proteins in Ca²⁺ signaling, such as calmodulin, CAMKII and CAMKIV, can also regulate the function of STING (Mathavarajah et al., 2019). It has also been reported that both BAPTA-AM-mediated Ca²⁺ depletion and ionomycin-induced Ca²⁺ elevation suppress STING-mediated innate immune responses (Kwon et al., 2018). The detailed molecular mechanisms underlying the association between Ca²⁺ signaling and STING-mediated innate immune responses remain unclear, and further studies are needed to clarify their relationship. At the same time, in most cases, the interaction between ALG2 and its partners is Ca²⁺ dependent, which is important for the function of ALG2. However, Ca²⁺-independent interaction between ALG2 and its partners, such as FASLG (Ji et al., 2020) and ASK1 (Hwang et al., 2002), has also been described. The interaction between ALG2 and STING is Ca²⁺ independent, because both EGTA and Ca²⁺ had no influence on the biochemical association between the two proteins. Moreover, the interaction and colocalization between ALG2 mutants and STING further confirmed this result. Consistent with ALG2, ALG2 mutants colocalized with SEC13 and STING at steady state, and did not traffic to the perinuclear region after stimulation by cGAMP. Surprisingly, despite retaining the interaction and colocalization with STING, ALG2 E47D and ALG2 E114D lost the ability to affect STING translocation or DNA-induced innate immune responses. One explanation for this observation is that there may be a third protein, which binds to ALG2 in a Ca²⁺-dependent manner, involved in the regulation of STING trafficking and DNA-induced innate immune responses. To investigate this, SEC31A and ALIX were selected as candidates and related experiments performed. Unfortunately, both of the proteins were found not to be the potential candidate protein. Further experiments are needed to elucidate the relationships among ALG2, Ca²⁺ signaling and STING.

In summary, our study identifies ALG2 as a new regulating factor for STING trafficking from the ER to perinuclear compartments, to influence DNA-induced innate immune responses. Moreover, Ca²⁺-binding ability is crucial for the function of ALG2 in this process. Further studies on the precise steps of ALG2 effects on STING translocation, and the mechanism of Ca²⁺ signaling regulation in this process, may inform the design of therapeutic strategies for STING-related autoimmune diseases.

ALG2 cDNA was obtained from the HEK293T cDNA library. For transient expression in mammalian cells, ALG2, ALG2 E47D and ALG2 E114D were cloned into pCMV-3HA. For stable expression in mammalian cells, ALG2, ALG2 E47D and ALG2 E114D were fused with RFP-tag or GFP-tag at the N-terminus by overlap PCR and cloned into pQCXIP2.0. ALG2 E47D and ALG2 E114D were generated by site-directed mutagenesis. The shRNAs were designed to target the human ALG2 and cloned into pLKO.1. STING cDNA was a gift from Dr Liangguo Xu (Jiangxi Normal University, Nanchang, China). For fluorescence location analysis, STING or STING(1–342) was fused with GFP-tag at the C-terminus by overlap PCR and cloned into pQCXIP2.0. For co-immunoprecipitation, STING was fused with Flag-tag at the N-terminus and cloned into pQCXIP2.0. STING, STING(135–379), STING(135–342) and STING(343–379) were fused with GST-tag at the N-terminus and cloned into pQCXIP2.0.

sgRNAs were designed to target human ALG2, SEC31A and ALIX. The related sgRNA sequence was cloned into lentiCRISPR vector. The sgRNA-resistant ALG2, ALG2 E47D and ALG2 E114D were generated by overlap PCR. The pVSV-G plasmid was a gift from Dr David Levy (University of Alabama at Birmingham, Birmingham, AL). The pCMV-promoted Gag-pol expression plasmid for MLV retrovirus packaging was a gift from Dr Wentao Qiao (Nankai University, Tianjin, China). The lentiCRISPR and psPAX2 vector were a gift from Dr Lin Liu (Nankai University).

HEK293T, HeLa and THP-1 cells were obtained from the American Type Culture Collection (ATCC). Vero cells were kindly provided by Dr Youjia Cao (Nankai University). All cells were cultured at 37°C, under 5% CO₂. Vero, HEK293T and HeLa cells were cultured in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS). THP-1 cells were cultured in Roswell Park Memorial Institute (RPMI)-1640 medium with 10% FBS. Cells were seeded for 24 h and reached a confluence of 60% before transfection with plasmids using polyethylenimine (PEI).

To produce lentivirus, HEK293T cells were seeded on 10 cm plates for 24 h before transfection. Then, 6 µg retroviral vector, 6 µg pCMV-MLV-gag-pol or psPAX2 and 3 µg pCMV-VSV-G were transfected together using PEI. Virus released into the supernatant was harvested, filtered through 0.22 µm filters and concentrated by ViraTrap™ Lentivirus Concentration Reagent (Biomiga). The concentrated virus was used to infect THP-1 or HeLa cells for 6 h, then the medium was replaced with fresh medium. After 2–3 days, the infected cells were used in subsequent experiments. HSV-1 was propagated and titered by plague assays on Vero cells.

Lentiviruses for cell line establishment were produced using HEK293T cells, as mentioned above. HeLa or THP-1 cells were infected by the corresponding lentivirus for 2 days. Infected cells were selected with 2 µg/ml puromycin for 2 weeks before single colonies were chosen and tested by western blotting.

To analyze protein levels, cells were treated with RIPA buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) for 30 min at 4°C. The cell lysates were boiled at 100°C for 10 min with SDS loading buffer before analysis by SDS-PAGE. The protein bands were transferred to 0.45 μm PVDF membranes, blocked with 5% skim milk and probed with primary antibodies, followed by the corresponding horseradish peroxidase (HRP)-conjugated secondary antibodies. Immunoreactive proteins were detected via Tannon-5500 gel imager. Primary antibodies were as follows: anti-GAPDH (Santa Cruz Biotechnology, sc-32233, western blotting 1:2000), anti-ALG2 (ABclonal, AB-2767269, western blotting 1:2000, immunofluorescence 1:100), anti-HA (Sigma-Aldrich, H3663, western blotting 1:5000), anti-Flag (Sigma-Aldrich, F1804, western blotting 1:5000), anti-GST (Utibody, UM3005, western blotting 1:1000), anti-GFP (Santa Cruz Biotechnology, sc-9996, western blotting 1:2000), anti-p-STING (Cell Signaling Technology, 50907, western blotting 1:1000). Rabbit antibodies against human STING, SEC31A and ALIX were from Proteintech: anti-STING (19851-1-AP, western blotting 1:2000, immunofluorescence 1:100); anti-SEC31A (17913-1-AP, western blotting 1:1000); anti-ALIX (12422-1-AP, western blotting 1:1000). Secondary antibodies used were goat anti-mouse IgG-HRP (Sungene Biotech, western blotting 1:10,000) and goat anti-rabbit IgG-HRP (Sungene Biotech, western blotting 1:10,000).

HEK293T cells were seeded on 10 cm plates for 24 h before transfection. Indicated plasmids were transfected for 48 h, then cells were harvested and lysed by sonication in NP-40 buffer (1% nonidet P-40, 20 mM Tris-HCl, pH 8.0, 137 mM NaCl, 10% glycerol). The cell lysates were clarified by centrifugation (10,000 g, 30 min, 4°C), then the supernatant was incubated with GFP-Trap sepharose resin for 4 h at 4°C before washing with NP-40 buffer five times. The precipitated proteins were treated with the SDS loading buffer at 100°C for 10 min, and detected by western blotting as described previously.

HeLa cells (4×10⁴/well) were seeded on glass coverslips in 12-well plates for 24 h before being transfected with plasmids or infected with lentivirus. Cells were fixed with 4% paraformaldehyde for 20 min, then permeabilized with 0.2% Triton X-100 for 10 min. After blocking in 3% bovine serum albumin (BSA) for 2 h, cells were incubated with primary antibody for 2 h, then probed with TRITC-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories, 1:100) for 2 h. THP-1 cells were also treated as described above. Images were captured with a Leica TCS SP5 confocal laser scanning microscope. ImageJ was used to quantify fluorescence intensity.

Cells were rinsed with PBS and lysed in TRIzol (Invitrogen) for 15 min. Lysed cells were mixed with chloroform and shaken for 15 s before being centrifuged (12,000 g, 15 min 4°C). The aqueous upper phase was transferred into a new tube, then isopropyl alcohol was added for RNA precipitation. The RNA precipitate was dissolved in RNase-free water. Reverse transcription was performed according to the manufacturer instructions of the RT-PCR kit (Abm, Canada). qRT-PCR was completed using a two-step real-time PCR kit (SYBR Green, GenStar).

cGAMP (InvivoGen) was delivered into HeLa or HEK293T cells by permeabilization with digitonin (10 μg/ml) for 30 min at 37°C in buffer A (50 mM HEPES, pH 7.2, 100 mM KCl, 3 mM MgCl₂, 0.1 mM DTT, 85 mM sucrose, 0.2% BSA, 1 mM ATP). cGAMP was delivered into THP-1 cells by permeabilization with PFO (0.01 μg/ml) for 30 min at 37°C in buffer A. The concentration of cGAMP used in stimulating cells was 200 nM or 400 nM.

All data were obtained from at least three independent biological replicates, and a representative result is shown. Data are shown as mean±s.d. Prism 8 software (GraphPad) was used for statistical analysis. Student's t-test (unpaired and two-tailed) was used to determine the significance of differences between two groups.

We thank Dr David Levy (University of Alabama at Birmingham), Dr Wentao Qiao (Nankai University), Dr Liangguo Xu (Jiangxi Normal University) and Dr Lin Liu (Nankai University) for providing plasmids. We thank Dr Youjia Cao (Nankai University) for providing Vero cells and HSV-1.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.259060