STING recruits WIPI2 for autophagosome formation

ABSTRACT

Induction of autophagy is a primordial function of the cGAS-STING pathway. However, the molecular mechanisms regulating autophagosome formation during STING-induced autophagy remain largely unknown. Recently, we reported that STING directly interacts with WIPI2 to recruit WIPI2 onto STING-positive vesicles for LC3 lipidation and autophagosome formation. We found that STING and PtdIns3P competitively bind to the FRRG motif of WIPI2, resulting in a mutual inhibition between STING-induced and PtdIns3P-dependent autophagy. We also showed that STING-WIPI2 interaction is necessary for cells to clear cytoplasmic DNA and attenuate activated cGAS-STING signaling. In summary, by identifying the interaction between STING and WIPI2, our study revealed a mechanism that allows STING to bypass the canonical upstream machinery to induce autophagosome formation.

Abbreviations: ATG: autophagy-related; MAP1LC3/LC3: microtubule associated protein 1 light chain 3; cGAMP: cyclic GMP-AMP; cGAS: cyclic GMP-AMP synthase; ER: endoplasmic reticulum; ERGIC: ER-Golgi intermediate compartment; IRF3: interferon regulatory factor 3; PtdIns3P: phosphatidylinositol-3-phosphate; SQSTM1: sequestosome 1; STING: stimulator of interferon genes; TBK1: TANK-binding kinase 1; ULK1: unc-51 like autophagy activating kinase 1; WIPI2: WD repeat domain, phosphoinositide interacting 2

As a cellular response to microbial infection, macroautophagy (hereafter called autophagy) can be induced by the cGAS (cyclic GMP-AMP synthase)-STING (stimulator of interferon genes) pathway. STING-induced autophagy not only eliminates the cytoplasmic pathogen DNA, but also selectively degrades some signaling molecules, such as STING, to attenuate activated cGAS-STING signaling and prevent the continued production of interferons. However, unlike autophagy induced by nutrient starvation, the canonical upstream machinery required for autophagosome biogenesis, including ULK1 (unc-51 like autophagy activating kinase 1) kinase and phosphatidylinositol 3-kinase complex, are unnecessary for STING-induced autophagy, suggesting that there are different mechanisms regulating autophagosome formation.

During STING-induced autophagy, STING-positive vesicles derived from the ER-Golgi intermediate compartment (ERGIC) are the primary membrane source for autophagosome formation. A simple hypothesis is that proteins that regulate autophagosome formation may be recruited onto STING-positive vesicles. Therefore, we performed mass spectrometry analysis of STING-positive vesicles isolated from cells treated with cyclic GMP-AMP (cGAMP), a specific activator of STING [1]. The phosphatidylinositol-3-phosphate (PtdIns3P) binding protein WIPI2 (WD repeat domain, phosphoinositide interacting 2) and its interacting partner ATG16L1 (autophagy-related protein 16–1) were among the proteins with high abundance. By immunostaining experiments, we verified the localization of WIPI2 and ATG16L1 to STING-positive vesicles and found that WIPI1, another PtdIns3P binding protein of the WIPI family, was not present on these vesicles, indicating that WIPI2 is recruited onto STING-positive vesicles in a PtdIns3P-independent manner. Using a STING mutant that cannot bind to cGAMP, we proved that the redistribution of WIPI2 in cells depends on the activation and trafficking of STING.

To investigate the mechanism by which WIPI2 is recruited onto STING-positive vesicles, we conducted co-immunoprecipitation experiments and found that WIPI2, rather than WIPI1, interacts with STING. However, STING-WIPI2 interaction in cells was not affected by cGAMP treatment. STING^R238A, which cannot bind to cGAMP, and STING^S366A, which cannot activate the transcription factor IRF3, displayed similar affinity to WIPI2 as wild-type (WT) STING, suggesting a constitutive interaction between STING and WIPI2. In order to characterize the membranes on which STING interacts with WIPI2, we fractionated the membranes from cells treated with or without cGAMP, and found that STING-WIPI2 interaction mainly occurs in the ERGIC and the ER, respectively. Then, we constructed truncated WIPI2 mutants to identify the key region in WIPI2 that is responsible for STING-WIPI2 interaction. Surprisingly, the FRRG motif, that mediates the binding to PtdIns3P for WIPI2, is also necessary for the interaction between STING and WIPI2. In addition, although neither WIPI2^FKKG nor WIPI2^FTTG binds to PtdIns3P, only WIPI2^FKKG can interact with STING, indicating that the positively charged basic amino acid residues in the FRRG motif of WIPI2 are crucial for STING-WIPI2 interaction. Using a similar method, we also located the key region within STING, containing negatively charged acidic amino acid residues, that mediates the interaction with WIPI2.

We investigated the role of STING-WIPI2 interaction in STING-induced autophagy. We found that WIPI2^FKKG but not WIPI2^FTTG formed puncta in cells treated with cGAMP. When cGAMP treatment cannot strongly induce MAP1LC3/LC3 (microtubule associated protein 1 light chain 3) lipidation in wipi2^-/- cells, re-introduction of WIPI2^FKKG but not WIPI2^FTTG can markedly restore it. In addition, cGAMP treatment cannot strongly stimulate LC3 lipidation, autophagosome formation and SQSTM1 (sequestosome 1) degradation in cells expressing the STING mutant that cannot bind to WIPI2. Due to the fact that STING can induce LC3 lipidation onto single-membrane vesicles, in order to confirm that the observed LC3 lipidation indeed contributes to the formation of autophagosomes, we first determined the co-localization of syntaxin 17 with LC3 on STING-positive puncta. Using protease protection assays, we also showed that SQSTM1 can be protected from proteolysis in cGAMP-treated cells, confirming that it was sequestered within double-membrane autophagosomes. Using the same method, we found that re-introducing WIPI2^FKKG but not WIPI2^FTTG into wipi2^-/- cells inhibited the proteolysis of SQSTM1 after cGAMP stimulation, indicating that STING-WIPI2 interaction is necessary for STING-induced autophagosome formation.

The FRRG motif of WIPI2 mediates the binding of WIPI2 to STING and PtdIns3P, indicating that STING and PtdIns3P may compete for WIPI2. In fact, manipulation of intracellular PtdIns3P level pharmacologically or genetically provided evidence showing that intracellular PtdIns3P negatively regulates STING-WIPI2 interaction. Furthermore, treatment of cells with VPS34-IN1, a specific inhibitor of the phosphatidylinositol 3-kinase complex, significantly increased the number of autophagosomes stimulated by cGAMP. Interestingly, when we tested the potential role of STING in PtdIns3P-dependent autophagy, we found that knockout of sting increased autophagosomes in cells under nutrient-rich conditions and further promoted the degradation of SQSTM1 in cells that were starved or treated with Torin1. These data suggested a mutual inhibition between STING-induced and PtdIns3P-dependent autophagy.

To evaluate the physiological effects of STING-WIPI2 interaction, we tested its role in cytoplasmic DNA clearance. First, we found that WIPI2 and ATG16L1, rather than WIPI1, were co-localized with cytoplasmic Cy3 labeled-interferon stimulatory DNA (Cy3-ISD) in cells. Using arabinofuranosyl cytidine, a chemical that causes DNA damage and nuclear DNA leakage into the cytoplasm, we observed that WIPI2 wraps DNA in the cytoplasm. Moreover, the co-localization between WIPI2 and Cy3-ISD in the cytoplasm required STING but not TBK1 (TANK-binding kinase 1). Consistent with this, WIPI2^FKKG rather than WIPI2^FTTG was co-localized with cytoplasmic Cy3-ISD. When knockout of wipi2 inhibited cytoplasmic DNA clearance triggered by cGAMP, re-introduction of WIPI2^FKKG but not WIPI2^FTTG restored the clearance. Interestingly, addition of VPS34-IN1 further promoted the clearance of cytoplasmic DNA in cGAMP-treated cells. These data suggested that STING-WIPI2 interaction is necessary for STING-induced clearance of cytoplasmic DNA.

Finally, we investigated the role of STING-WIPI2 interaction in the attenuation of cGAS-STING signaling. By examining TBK1 phosphorylation and measuring IFNB1 expression, we found that knockout of wipi2 blocked the attenuation of cGAS-STING signaling in arabinofuranosyl cytidine-treated cells. Consistent results were obtained using human monocyte-derived macrophages. Furthermore, our results showed that transfection with WT WIPI2 or WIPI2^FKKG, but not WIPI2^FTTG or WIPI2^R108E/R125E, which cannot interact with ATG16L1, reversed arabinofuranosyl cytidine-stimulated increase in the level of IFNB1 mRNA in wipi2^-/- cells. Consistent with this, re-introduction of STING mutant, unable to bind to WIPI2, but not WT STING, further enhanced IFNB1 expression in sting^-/- cells. Thus, these results suggested that STING-WIPI2 interaction is required for the attenuation of activated cGAS-STING signaling.

In summary, our study reveals the mechanism that mediates WIPI2 recruitment for LC3 lipidation and autophagosome formation during STING-induced autophagy. Our findings also explain the dispensability of the upstream autophagy-related machinery, in particular the ULK1 kinase and phosphatidylinositol 3-kinase complex, in STING-induced autophagosome formation. In addition, by elucidating the direct interaction between STING and WIPI2, we uncover a mutual inhibition between STING-induced and PtdIns3P-dependent autophagy.

Funding Statement

This study was supported by the Key R&D Plan of the Ministry of Science and Technology of China [2021YFA1300303], the National Natural Science Foundation of China [32230023, 92057203 and 31970694], the Fundamental Research Funds for the Central Universities [2020×ZZX002-16], the Young Elite Scientists Sponsorship Program by China Association for Science and Technology (CAST) [2019QNRC001], and the Chao Kuang Piu High-tech Development Fund [2020QN024].

Disclosure statement

No potential conflict of interest was reported by the author(s).