Attenuation of cGAS/STING Activity During Mitosis

The innate immune system recognizes cytosolic DNA associated with microbial infections and cellular stress via the cGAS/STING pathway, leading to activation of phospho-IRF3 and downstream IFN-I and senescence responses. To prevent hyperactivation, cGAS/STING is presumed to be non-responsive to chromosomal self DNA during open mitosis, though specific regulatory mechanisms are lacking. Given a role for the Golgi in STING activation, we investigated the state of the cGAS/STING pathway in interphase cells with artificially vesiculated Golgi and in cells arrested in mitosis. We find that while cGAS activity is impaired through interaction with mitotic chromosomes, Golgi integrity has little effect on the enzyme’s production of cGAMP. In contrast, STING activation in response to either foreign DNA (cGAS-dependent) or exogenous cGAMP is impaired by a vesiculated Golgi. Overall our data suggest a secondary means for cells to limit potentially harmful cGAS/STING responses during open mitosis via natural Golgi vesiculation.


29
In addition to innate defense against microbial infections, cGAS/STING is involved in cellular responses studies have revealed that i) chromosome-bound cGAS is tightly tethered to chromatin, potentially via interactions with H2a/H2b dimers, ii) chromatin interaction does not involve the DNA-binding domains of cGAS required for "typical" activation by dsDNA, and iii) that chromosome binding results only in weak activation of 3A), in agreement with prior literature [75].

96
To determine if the requirement for intact Golgi was specific for cGAS/STING or involved a more broad 97 inhibition of IRF3 phosphorylation, we investigated activation of the RIG-I-like receptors (RLRs). The dsRNA-98 mimic polyinosinic-polycytidylic acid (pIC) stimulates RLR-family members which recognize intracellular viral 99 RNA products through the mitochondria-resident adaptor protein MAVS to elicit NFκB inflammatory and 00 IRF3/7-dependent IFN-I responses [76,77]. pGL3-dependent pSTING, pTBK1, and pIRF3 responses were 01 abolished by Golgi dispersal (Figure 3B, C). In contrast, transfection of pIC elicited pTBK1 and pIRF3 02 responses regardless of GCA treatment ( Figure 3B, D), suggesting that the cGAS/STING pathway, but not the production in response to pGL3 transfection (Fig. 3F), indicating that Golgi dispersal blocked the pathway downstream of cGAS.
The GCA-induced repression of cGAS/STING activation was also reflected when measuring pGL3-08 dependent downstream transcriptional responses. RT-qPCR at 4 hr and 8hr post pGL3 transfection revealed 09 that induction of IFN-I (IFNB1), ISGs (Viperin, IFI6, HERC5, IFIT2, IFIT3), and chemokines (CXCL10 and 10 CXCL11) was significantly dampened in the presence of GCA ( Figure 4). Overall, these data show that 11 cGAS/STING activity is blunted at the level of STING upon Golgi vesiculation, similar to what may occur during 12 mitosis.

15
We next investigated the impact of natural mitotic Golgi vesiculation on the ability of cGAS/STING to 16 sense and respond to exogenous DNA. Secretory ER to Golgi traffic is blocked during mitosis [78][79][80][81]. Golgi 17 integrity is dependent on cargo transport from ERES, and mitotic arrest of COPII-dependent ERES traffic 18 causes Golgi dispersal [82,83]. As assembly of the activated STING/TBK1/IRF3 complex requires STING 19 transport from ERES to the Golgi [14], we hypothesized that mitotic Golgi dispersal and inactivation of ERES 20 would blunt cGAS/STING responses.

21
We devised a method to synchronize cells at prometaphase ( Figure 5A). Briefly, cells were cultured at 100% confluence for 48 hr in 1% serum, leading to quiescent arrest in G 0 . Cells were released by replating in 23 10% serum, allowing for G 1 re-entry and progression to S. Twenty-four hours post-G 0 release, cells were 24 synchronized at prometaphase with low dose NOC for 12 hr. Upon NOC washout, synchronized cells 25 progressed through mitosis, returning to G 1 within 3 hr. Propidium iodide staining showed cells enriched at 26 G 2 /M after NOC treatment, and the majority of cells back in G 1 by 3 hr post-NOC release ( Figure 5A).

29
We used this scheme to assess mitotic cGAS/STING responses to pGL3 transfection. Cells were 30 transfected while in prometaphase following the 12 hr NOC sync (non-released group, NR), or at 0, 60, 120, or 31 180 min post-release, and cGAS/STING activity was assessed 90 min post-transfection. Cells arrested at 32 prometaphase mounted a very weak pIRF3 response to pGL3 transfection compared to cells which had returned to G 1 after a 3 hr release ( Figure 5B, compare lanes 6 to 10). These arrested cells had condensed chromosomes with dispersed Golgi (Figure 5D, NR). STING had a cytosolic but granular distribution in 35 arrested cells, which did not change upon pGL3 transfection. In contrast, STING clearly localized to p230-   H2a/H2b dimers [53]. However this binding is not via the DNA-binding domains of cGAS that underlie DNA-44 dependent activation, resulting in a relatively low production of cGAMP. Thus, chromatin appears to blunt 45 cGAS activation [34,53], suggesting a means for the cell to avoid cGAS-driven IFN responses to self

49
We observed only low activation of IRF3 in response to pGL3 transfection in NOC-arrested cells 50 ( Figure 5B,C and 6A). To determine if the dampening of the pathway in these arrested cells was at the level of 51 cGAS or STING, we measured cGAMP production in response to pGL3 transfection and examined cGAS 52 subcellular distribution. Similar to pIRF3, cGAMP production was blunted in transfected arrested cells (NR) 53 compared to asynchronous interphase cells ( Figure 6B). Confocal microscopy of asynchronous cells revealed 54 that cGAS was mostly nuclear although some signal was evident within the cytosol ( Figure 6C), in agreement 55 with a recent report [54]. Within arrested cells, the vast majority of cGAS was chromatin-bound and Golgi were 56 well-dispersed ( Figure 6D). These results agree with recent work showing that cGAS is predominantely 57 nuclear, regardless of cell cycle phase or activation status [54]. Combined, our data suggest that chromatin-58 bound cGAS is unable to produce a robust cGAMP response to either chromosomes or transfected DNA.

59
Although we detected slightly elevated levels of cGAMP in unstimulated arrested versus asynchronous generates only low levels of cGAMP upon activation by chromatin [34,53]. Mitotic cells could potentially take up exogenous cGAMP via SLC19A1 [84,85] or LRRC8 [86] transporters, or directly from neighboring cells [87, autophagosomes promote clearance of cytosolic DNA and incoming DNA viruses like HSV1 [75]. Although

83
Given that cGAS/STING-dependent elevation of pIRF3 during mitosis (particularly during prolonged 84 mitosis) can induce apoptosis [53] and activated STING can induce a potentially harmful autophagic response 85 [75], a parallel dampening mechanism like Golgi dispersal likely serves an important role to limit potentially

32
Cells were transfected with pGL3 as described above, and either harvested for western blotting or prepared for 33 immunofluorescence or cell cycle analysis as described.

60
Following the preparation of immunofluorescence slides, confocal microscopy was performed using a Zeiss

61
LSM880 system with 405 nm, 488 nm, and 543 nm lasers. Samples were examined using an oiled 63x 62 objective, and Z-stacks with a 0.32 μ m depth per plane were taken of each image. Representative single-plane 63 images and Z-stacks were processed with the Zen Blue software.

66
Manders' overlap coefficients [102] for a STING:p230 and IRF3:DAPI channels within individual Z-stacks were 67 determined using the JACoP plugin [103] on ImageJ [104]. Manual thresholds were set below saturation. HaCaTs were plated at 60,000 asynchronous cells or 130,000 synchronous cells per well in 24-well plates.