ABSTRACT
Recognition of foreign nucleic acids is an evolutionarily conserved mechanism by which the host detects microbial threats. Whereas some intracellular bacterial pathogens trigger DNA surveillance pathways following phagosomal membrane perturbation, mechanisms by which extracellular bacteria activate cytosolic nucleic acid reconnaissance systems remain unresolved. Here, we demonstrate that Helicobacter pylori exploits cag type IV secretion system (cag T4SS) activity to provoke STING signaling in gastric epithelial cells. We provide direct evidence that chromosomal fragments delivered to the host cell cytoplasm via trans-kingdom conjugation bind and activate the key DNA sensor cGMP-AMP synthase. To enable paracrine-like signal amplification, translocated H. pylori DNA is sorted into exosomes that stimulate DNA-sensing pathways in uninfected bystander cells. We show that DNA cargo is loaded into the cag T4SS apparatus in the absence of host cell contact to establish a ‘ready-to-fire’ nanomachine and provide evidence that cag T4SS-dependent DNA translocation is mechanistically coupled to chromosomal replication and replichore decatenation. Collectively, these studies suggest that H. pylori evolved mechanisms to stimulate nucleic acid surveillance pathways that regulate both pro- and anti-inflammatory programs to facilitate chronic persistence in the gastric niche.
INTRODUCTION
Innate recognition of invariant pathogen-associated molecular signatures by cellular pattern recognition receptors (PRRs) is the first line of defense against microbial adversaries. Within the gastrointestinal tract, epithelial cells express proximal endosomal and cytosolic nucleic acid-sensing PRRs that rapidly respond to aberrant microbial DNA and RNA to trigger innate defense mechanisms and coordinate adaptative immunity. Localization of foreign DNA to the host cell cytosol activates multiple nucleic acid reconnaissance systems including the key DNA sensor nucleotidyltransferase cyclic GMP-AMP synthase (cGAS) (Cai et al., 2014; Diner et al., 2013; Sun et al., 2013; Wu et al., 2013). Upon binding DNA originating from either extrinsic or intrinsic sources, cGAS catalyzes the formation of the non-canonical cyclic di-nucleotide 2’3’-cGAMP using ATP and GTP as substrates (Diner et al., 2013; Gao et al., 2013a; Gao et al., 2013b; Sun et al., 2013; Zhang et al., 2013). In turn, 2’3’-cGAMP stimulates the endoplasmic reticulum receptor STING (stimulator of interferon genes) to elicit interferon (IFN) signaling and the production of multifarious inflammatory cytokines (Ishii et al., 2006; Ishikawa and Barber, 2008; Ishikawa et al., 2009; Stetson and Medzhitov, 2006; Sun et al., 2009; Zhong et al., 2008). Dysregulation of mucosal STING signaling can disrupt gut homeostasis and generate pro-tumorigenic inflammatory microenvironments (Ke et al., 2022); however, the outcomes of STING-dependent immune surveillance within the context of gastric inflammation and infection-associated carcinogenesis remain unresolved.
Of the known infection-associated cancers, the most significant carcinogenic microbe is the gastric bacterium Helicobacter pylori, which chronically colonizes the stomach of over half of the global population and directly contributes to the development of more than one million new cases of cancer per year (Sung et al., 2021). H. pylori harboring the cancer-associated cag type IV secretion system (cag T4SS) significantly augment disease risk via translocation of pro-inflammatory molecular cargo into gastric epithelial cells. In addition to facilitating the delivery of the bacterial oncoprotein CagA, the cag T4SS translocates a diverse repertoire of immunostimulatory lipid, nucleic acid, and polysaccharide substrates directly into the gastric epithelium (Amieva and Peek, 2016; Cover et al., 2020). Recent work reported that H. pylori cag T4SS activity activates the endosomal DNA-sensing PRR Toll-like Receptor 9 (TLR9), leading to immune suppression conferring tolerance (Varga et al., 2016a; Varga et al., 2016b) as well as other potential inflammation-independent carcinogenic phenotypes (Castano-Rodriguez et al., 2014). In addition to TLR9, multiple cellular nucleic acid sensors, including cGAS (Ablasser et al., 2013; Cai et al., 2014; Diner et al., 2013; Gao et al., 2015; Nandakumar et al., 2019; Storek et al., 2015; Watson et al., 2015; Zhang et al., 2014), RIG-I (Chow et al., 2015; Dixit and Kagan, 2013; Onomoto et al., 2021; Rad et al., 2009), MDA5 (Dixit and Kagan, 2013; Wu et al., 2020), AIM2 (Rathinam et al., 2010), ZBP1/DAI (Kuriakose et al., 2016), IFI-16 (Almine et al., 2017; Unterholzner et al., 2010), and RNA pol III (Ablasser et al., 2009; Chiu et al., 2009) are expressed in the human gastric epithelium and associated mucosal dendritic cells (Rad et al., 2009), raising the hypothesis that H. pylori stimulates additional innate nucleic acid surveillance pathways.
While intracellular bacterial pathogens elicit cGAS-STING signaling following phagosomal membrane destabilization or rupture achieved in a type III, IV, VI, or VII secretion system-dependent manner (Ku et al., 2020; Marinho et al., 2017; Nandakumar et al., 2019; Storek et al., 2015; Watson et al., 2015; Zhang et al., 2014), the mechanisms by which extracellular bacteria stimulate DNA reconnaissance systems remain unclear. STING signaling has been implicated in gastric carcinogenesis and H. pylori has been shown to activate STING in vivo (Song et al., 2017), but whether H. pylori-driven STING activation requires cytosolic nucleic acid immunosurvellience is unknown. Here, we demonstrate that H. pylori-induced STING signaling is a direct consequence of cag T4SS-dependent DNA translocation. We show that H. pylori chromosomal fragments delivered to the gastric epithelium via trans-kingdom conjugation directly bind and activate cGAS to stimulate STING signaling. We discovered that upon translocation into primary gastric epithelial cells, fragmented H. pylori DNA is sorted into exosomes that are released to amplify foreign nucleic acid immune surveillance in uninfected bystander cells. We provide direct evidence that cag T4SS-mediated DNA translocation is mechanistically coupled to chromosomal replication and demonstrate that eukaryotic-optimized constructs greater than 1.5 kb are delivered to the gastric epithelium via cag T4SS mechanisms. Our results highlight how H. pylori exploits the versatile cag T4SS to tip the delicate STING signaling balance towards inflammatory responses that may stimulate carcinogenesis and enable chronic colonization of the gastric niche.
RESULTS
H. pylori provokes multiple DNA surveillance systems in a cag T4SS-dependent manner
Previous studies demonstrate the capacity of H. pylori to stimulate DNA-sensing pattern recognition receptors, including TLR9 (Rad et al., 2009; Varga et al., 2016b), raising the hypothesis that microbial nucleic acids are actively translocated into host cells. In agreement with previous reports (Rad et al., 2009; Varga et al., 2016b), H. pylori challenge of HEK293 reporter cell lines stably transfected with TLR9 demonstrated that cag T4SS activity is required for TLR9 stimulation (Fig. 1A,B). Consistent with the observation that CagA is not translocated into HEK293 cells (Kumar Pachathundikandi et al., 2011; Varga et al., 2016b), disruption of cagA did not diminish levels of H. pylori-induced TLR9 activation (Fig. 1A and Fig. S1A). To confirm that a functional cag T4SS is required to activate TLR9, we co-cultured TLR9 reporter cells with a cagL isogenic mutant or the corresponding genetically complemented strain. Whereas inactivation of cagL abrogated TLR9 activation, complementation in a heterologous chromosomal locus rescued TLR9 stimulation to levels indistinguishable from the parental WT strain (Fig. 1B). In concert with previous investigations, these data demonstrate that cag T4SS activity, but not CagA delivery, is required for robust TLR9 signaling.
In addition to endosomal DNA pattern recognition receptors such as TLR9, gastric epithelial cells harbor cytosolic DNA surveillance proteins including the nucleotidyltransferase cyclic GMP-AMP (cGAMP) synthase (Cai et al., 2014; Diner et al., 2013; Sun et al., 2013; Wu et al., 2013). To test whether H. pylori DNA is trafficked into the host cell cytoplasm to activate cytosolic DNA surveillance sensors, we challenged 293T cells transfected with constructs to enable cGAS and STING expression and quantified levels of IFN-β promoter-driven luciferase produced in response to H. pylori. Compared to mock infected cells, WT H. pylori stimulated high levels of cGAS-STING signaling at 18 h post-infection. Similar to TLR9 activation assays, the cagL mutant was unable to stimulate robust cGAS-STING signaling, a phenotype that was restored by genetic complementation in a secondary chromosomal locus (Fig. 1C). In contrast to HEK293-hTLR9 cells, moderate levels of translocated CagA were detected in 293T cells co-cultured with WT H. pylori but not in corresponding cagE-challenged monolayers (Fig. S1A). Although H. pylori has the capacity to deliver CagA into 293T cells, disruption of cagA did not significantly alter levels of cag T4SS-dependent IFN-β promoter activity in cGAS-STING reporter cells (Fig. S1B). To exclude the possibility that cGAS-STING activation results from bacterial endocytosis or increased bacterial interaction with host cell surfaces, we performed gentamicin protection assays. In contrast to marked differences in cGAS-STING activation elicited by WT and cagX, equivalent levels of adherent and intracellular bacteria were recovered from 293T co-cultures, confirming that cGAS-STING signaling is not an artifact of non-specific bacterial internalization or spontaneous bacterial lysis (Fig. S1C).
Previous studies demonstrate that in addition to detecting invading microbial threats, cGAS senses and responds to DNA damage and genomic instability (Cai et al., 2014; Ke et al., 2022). H. pylori cag T4SS activity is directly linked to nuclear double-stranded DNA (dsDNA) breaks introduced in response to ALPK1/TIFA signaling stimulated by D-glycero-beta-D-manno-heptose 1,7-bisphosphate (HBP) or ADP-beta-D-manno-heptose (β-ADP-heptose) translocation (Bauer et al., 2020; Gall et al., 2017; Zimmermann et al., 2017). We therefore addressed the possibility that cag T4SS-dependent nuclear DNA damage stimulates cGAS-STING signaling elicited by WT H. pylori. Disruption of rfaE, the enzyme responsible for β-ADP-heptose production (Bauer et al., 2020; Gall et al., 2017; Stein et al., 2017; Zimmermann et al., 2017), did not impact the level of cGAS-STING signaling achieved by H. pylori (Fig. 1D), indicating that ALPK1/TIFA signaling-induced DNA damage does not significantly contribute to H. pylori-driven cGAS activation in vitro. In addition to ALPK1/TIFA-mediated DNA modifications (Bauer et al., 2020; Gall et al., 2017; Stein et al., 2017; Zimmermann et al., 2017), H. pylori has the capacity to induce production of DNA-damaging reactive oxygen species (ROS) by triggering NF-κB activation and additional mechanisms involving inducible nitric oxide synthase (iNOS) and associated inflammatory enzymes (Bauer et al., 2020; Kidane, 2018). In support of the observation that cag T4SS-induced dsDNA breaks are not a significant cGAS activating factor in the context of H. pylori infection, cGAS-STING activation was achieved by H. pylori co-cultured in the presence of the antioxidant N-acetyl-cysteine at concentrations that abrogate ROS production (Bauer et al., 2020) (Fig. S1D).
Following infection, damaged mitochondria release DNA (mtDNA) and other constituents into the cytosol to act as potent danger-associated molecular patterns (DAMPs) that engage TLR9 (Garcia-Martinez et al., 2016; Oka et al., 2012; Zhang et al., 2010) and cGAS-STING signaling axes to initiate type I IFN production (Rongvaux et al., 2014; West et al., 2015; White et al., 2014). To test the hypothesis that H. pylori cag T4SS activity modulates mitochondrial integrity resulting in the release of mtDNA and activation of cytosolic DNA-sensing PRRs, we assayed cGAS-STING activation in the presence of BAX/BAK macropore inhibitory peptides that prevent permeabilization of the mitochondrial outer membrane and herniation of mtDNA into the cytosol (McArthur et al., 2018; White et al., 2014). H. pylori induced similar levels of cGAS-STING activation in the presence of BAX inhibitory peptide or non-inhibitory peptide control co-cultures (Fig. S1E) suggesting that ruptured mitochondria are not the primary source of cGAS-activating DNA. Likewise, treatment of co-cultures with a mitochondria-targeted antioxidant did not significantly alter levels of cGAS-STING signaling (Fig. S1D), demonstrating that H. pylori-induced mtDNA damage is not a predominant DAMP within the context of cag T4SS-dependent cGAS activation. To exclude the possibility that cGAS-STING activation is dependent upon the import of released H. pylori DNA via host cell mechanisms, we quantified IFN-β promoter activity produced by cGAS-STING reporter cells that were physically separated from H. pylori. In support of the hypothesis that cGAS-STING activation requires direct H. pylori-host cell interaction, cGAS stimulation was achieved by H. pylori that were in direct contact with reporter cells, but not in samples in which H. pylori and reporter cells were physically separated by a 0.4 µM-pore polycarbonate insert (Fig. 1E). Collectively, these studies demonstrate that cGAS stimulation is a consequence of cag T4SS-dependent DNA delivery into the host cell cytosol.
To define the role of cGAS in H. pylori-driven IFN-β signaling, we next determined whether cag T4SS activity could transactivate STING through direct cGAMP transfer via gap junction-mediated diffusion. To assess cGAS signaling in trans, 293T cells transfected with cGAS constructs were co-cultured with 293T cells harboring STING and IFN-β reporter constructs, and co-cultures were challenged by H. pylori for 18 h. High levels of STING signaling were observed in cells challenged by WT H. pylori, but not a cagX mutant, suggesting that cag T4SS-dependent cGAS stimulation generates sufficient cGAMP for migration into bystander cells (Fig. 1F). To examine the requirement of cGAS protein domains in H. pylori-induced STING signaling, we next monitored IFN-β activation in 293T cells transfected with constructs to express either cGAS alone or cGAS variants in combination with STING. In comparison to cells transfected with only cGASWT, WT H. pylori stimulated high levels of IFN-β transcription when cGASWT and STING were co-expressed (Fig. 1G). In contrast, cag T4SS-dependent IFN-β transcription was markedly reduced in 293T cells transfected with STING and cGAS variants lacking the DNA-binding domain (cGASΔDBD) or harboring point mutations within nucleotidyltransferase catalytic residues that abolish cGAMP production (cGASΔNTase) (Fig. 1G). Collectively, these results demonstrate that cGAS senses and responds to cag T4SS activity.
A previous report suggests that decreased STING signaling is associated with adverse outcomes in gastric cancer patients (Song et al., 2017). We therefore monitored cGAS-STING signaling induced by several H. pylori clinical isolates obtained from patients with gastric diseases of varying severity. In contrast to strains isolated from patients exhibiting gastritis, H. pylori isolated from individuals with duodenal ulcers or gastric cancer elicited lower levels of cGAS-STING signaling (Fig. 1H), suggesting that H. pylori may modulate the capacity to induce cag T4SS-dependent STING activation during chronic stomach colonization.
Microbial DNA is delivered to the gastric epithelial cell cytoplasm via cag T4SS activity
We next assessed the consequence of H. pylori trans-kingdom DNA conjugation within the context of biologically-relevant interactions with gastric epithelial cells. To monitor DNA injection into gastric epithelial cells, AGS cells were challenged by either WT or the cagX isogenic mutant at a MOI of 50. After 6 h, gastric epithelial cell co-cultures were treated with DNaseI to remove extracellular DNA and eukaryotic cells were fractionated using digitonin to selectively permeabilize the plasma membrane, leaving the nuclear envelope and bacterial cells intact. PCR analysis of fractionated infected AGS cells revealed the presence of H. pylori DNA in cytoplasmic fractions (Fig. 2A) with significantly more bacterial DNA present in cytosolic fractions obtained from WT-challenged cells. When comparing the ratio of cytoplasmic bacterial DNA to cytoplasmic-localized mitochondrial DNA by qPCR, significantly more bacterial DNA was present in the cytosol of gastric epithelial cells infected by the WT strain compared to corresponding cells infected by the cagX mutant (Fig. 2B). Levels of mitochondrial DNA did not significantly differ in cytosolic extracts obtained from infected and uninfected cells (Fig. 2A). To exclude the possibility that cytosolic localization of H. pylori DNA resulted from non-specific bacterial lysis or endocytosis, we performed studies to analyze the level of adherent and internalized H. pylori in infected gastric epithelial cells. Using gentamicin-protection assays, we determined that similar levels of adherent and intracellular WT and cagX bacteria were recovered from AGS co-cultures (data not shown). Multiple cancer cell lines, including cell lines derived from gastric adenocarcinoma, continuously export low levels of extracellular cGAMP that serves as a potent immunotransmitter (Carozza et al., 2020). We thus measured levels of extracellular cGAMP secreted by AGS cells in response to H. pylori challenge. Consistent with the hypothesis that cag T4SS-mediated DNA delivery stimulates cytosolic cGAS and the subsequent production of cGAMP, the level of extracellular cGAMP was significantly higher in supernatants obtained from WT-challenged co-cultures compared to mock infected or corresponding cagX-challenged monolayers (Fig. 2C). Collectively, these results suggest that the presence of bacterial DNA in the host cell cytoplasm is a consequence of cag T4SS activity and leads to the production of cGAS-generated cGAMP.
Although gastric adenocarcinoma cell lines (including AGS and MKN45) produce detectable levels of cGAS and other nucleic acid-sensing PRRs, STING expression is absent (Qiao et al., 2020). To determine whether translocated H. pylori DNA elicits cGAS-STING responses in normal gastric epithelia, we challenged primary adult gastric epithelial cells with H. pylori and monitored the formation of peri-nuclear STING polymers that aggregate in response to cGAMP binding. Compared to mock infected or cagX-challenged cells, WT H. pylori induced the formation of large STING aggregates that could be visualized by confocal microscopy (Fig. 2D). Consistent with activation of cGAS-STING signaling, quantification of the average STING polymer size in H. pylori-gastric cell co-cultures revealed significantly larger STING aggregates in WT-challenged cells (Fig. 2E), suggesting cag T4SS-dependent stimulation of cGAS surveillance. In addition to cGAS, epithelial cells harbor other nucleic acid reconnaissance systems that signal through STING. To determine whether H. pylori stimulates additional STING-dependent signaling pathways, we analyzed IFN-β activity in 293T-STING cells. Compared to control 293T cells harboring empty vector, H. pylori stimulated STING signaling in a cag T4SS-dependent manner (Fig. 2F). STING signaling was also elicited by transfection of purified, fragmented H. pylori chromosomal DNA into IFN-β reporter cells (Fig. 2F), suggesting that endogenous 293T cytosolic DNA-sensing pathways respond to H. pylori DNA. Together, these studies demonstrate that H. pylori delivers chromosomal DNA fragments to gastric epithelial cells via cag T4SS mechanisms and reveal that endogenous STING-dependent nucleic acid surveillance systems are activated by translocated H. pylori DNA.
DNA is a specific cag T4SS nucleic acid substrate
We reasoned that in addition to DNA substrates, the cag T4SS may translocate RNA or DNA:RNA hybrids into gastric cells to stimulate STING-dependent signaling. Thus, we sought to characterize the innate inflammatory signature elicited by H. pylori. Primary adult gastric epithelial cells were mock infected or challenged by either WT or the corresponding cagX isogenic mutant prior to isolation of total RNA. To characterize epithelial innate inflammatory responses, gene expression patterns were analyzed at 6 h post-infection using the NanoString Host Response Panel. Compared to mock infected and cagX-challenged primary gastric epithelial cell co-cultures, cag T4SS activity induced a significant increase in transcripts associated with anti-microbial defense pathways and interferon signaling (Fig. 3A and Fig. S2). Hierarchical clustering performed on genes that were differentially expressed in response to H. pylori challenge revealed distinct clustering of mock infected and cagX-infected co-cultures in comparison to WT challenge (Fig. S2C). Comparison of transcripts differentially produced in response to WT versus cagX H. pylori identified 134 genes that were upregulated via cag T4SS activity, including genes encoding pro-inflammatory chemokines and cytokines (e.g., CXCL10, CXCL8/IL-8, TNF, IL1B, CCL5), IFN-α/β signaling (e.g., ISG15, IRF1, IFI35, STAT1, IFNAR2, SAMHD1), and IFN-γ immunoregulatory programs (e.g., IFNGR2, TRIM5, GBP4, VCAM1) (Fig. 3A,B). Consistent with the observed gene expression patterns, pathway analysis of genes induced by cag T4SS activity demonstrated a significant enrichment in transcripts associated with interferon-regulated nucleic acid reconnaissance programs (Fig. 3C). In addition to cag T4SS-dependent regulation of the DNA sensor ZBP1, we observed significantly increased transcript levels of several cytosolic RNA-sensing surveillance systems including DDX58/RIG-I, RIG-I-like Receptor IFIH1/MDA5 (Chow et al., 2015; Dixit and Kagan, 2013), and 2′-5′-oligoadenylate synthetases OAS1 and OAS2 (Schwartz et al., 2020) in WT-infected primary gastric epithelial cells (Fig. 3A,B).
A previous study reported the capacity of gastric mucosa-associated dendritic cells to sense and respond to purified H. pylori RNA through RIG-I and TLR7/8, leading to the production of type I IFN (Rad et al., 2009; Salama et al., 2013). However, whether the cag T4SS actively translocates RNA to stimulate TLR7/8 or RIG-I signaling axes has not been elucidated. We thus sought to determine whether DNA is a specific cag T4SS nucleic acid effector. Challenge of HEK293 reporter cell lines expressing either TLR7 (Fig. 3D) or RIG-1 (Fig. 3E) revealed that in contrast to robust activation of DNA-sensing PRRs, H. pylori was unable to activate either endosomal ssRNA (TLR7) or cytoplasmic ssRNA/dsRNA (RIG-I) sensors. Collectively, these data suggest that chromosomally-derived DNA is a specific cag T4SS substrate, and provide further evidence that cag T4SS-dependent DNA translocation stimulates STING-dependent signaling.
H. pylori cag T4SS activity stimulates DNA immunosurveillance in bystander cells
Prior work demonstrates that foreign intracellular bacterial DNA can be delivered to adjacent cells via extracellular vesicles as a mechanism to amplify IFN signaling (Nandakumar et al., 2019). We therefore investigated whether translocated H. pylori DNA could be sorted into extracellular vesicles to enable paracrine-like DNA signaling by primary gastric epithelial cells. To test the hypothesis that infected gastric epithelial cells release extracellular vesicles containing H. pylori DNA, cell-free supernatants collected from infected primary gastric epithelial donor cells were used to challenge recipient TLR9 reporter cells. Compared to supernatants obtained from either mock infected or cagX-challenged co-cultures, supernatants harvested from WT-challenged gastric cells robustly activated TLR9 signaling (Fig. 4A). The capacity of infected cell supernatants to activate TLR9 was time dependent, as levels of TLR9 activation increased when reporter cells were treated with supernatants obtained from gastric epithelial cells co-cultured with H. pylori for 12 h compared to 6 h (Fig. 4A). To exclude the possibility that TLR9 activation resulted from contaminate DNA arising from cell lysis, we quantified the level of cell-free DNA in donor supernatants. TLR9 activation elicited by gastric cell supernatants was not correlated to the level of donor supernatant cell-free DNA (Fig. 4B), suggesting that the TLR9-stimulating agonist was enclosed within a host cell-derived delivery mechanism.
To confirm the role of translocated DNA in transferrable nucleic acid immunosurveillance in bystander cells, donor gastric cell supernatants were treated with DNase alone or in combination with heat prior to co-culture with TLR9 reporter cells. Whereas DNase treatment had a negligible effect on TLR9 stimulating capacity, heat treatment modesty reduced TLR9 responses, which were further reduced when donor supernatants were treated with both heat and DNase prior to reporter cell challenge (Fig. 4C). These observations led to the hypothesis that translocated microbial DNA packaged within gastric cell-derived extracellular vesicles, such as exosomes, stimulates nucleic acid reconnaissance systems in bystander cells. To test whether exosome biogenesis is required for the delivery of translocated DNA to uninfected cells, we challenged recipient TLR9 reporter cells with donor supernatants obtained from infected primary gastric epithelial cells cultured in the presence or absence of a neutral sphingomyelinase (nSmase2) inhibitor that prevents exosome release. Compared to untreated and mock infected cells, nSmase2 inhibition led to a significant reduction in TLR9 activation levels elicited by WT-challenged donor cell supernatants (Fig. 4D), supporting a role of exosomes in stimulating DNA-sensing pathways in bystander cells. To further investigate whether translocated H. pylori DNA was packaged within exosomes released from infected gastric epithelial cells, we isolated CD9, CD63, and CD81-positive exosomes from cell culture supernatants via immunopurification. Analysis by qPCR revealed that exosomes purified from WT-challenged gastric cell supernatants were significantly enriched in H. pylori DNA compared to corresponding exosomes obtained from cagX-challenged co-cultures (Fig. 4E), suggesting that foreign bacterial DNA is sorted into exosomes for paracrine-like DAMP signal amplification (Fig. 4F).
Random chromosomal fragments are delivered to target cells via cag T4SS-dependent mechanisms
To identify whether a specific DNA sequence is excised and transferred to host cells by cag T4SS activity, we purified cGAS-DNA complexes using modified ChIP-seq workflows to capture bacterial DNA that physically binds and activates cGAS. 293T-cGAS cells were challenged by H. pylori for 6 h, followed by chemical cross-linking and immunopurification of cGAS. Deep sequencing analysis confirmed that co-purifying H. pylori DNA isolated from WT and cagX-infected cells mapped across the entire H. pylori chromosome, with significantly more bacterial DNA reads associated with cGAS purifications obtained from WT-challenged monolayers (Fig. 5A). Normalized sequencing reads and differential peak calling approaches identified more than three hundred H. pylori chromosomal regions specifically associated with cGAS purified from WT-infected cells compared to eight bacterial DNA peaks associated with corresponding preparations obtained from cagX co-cultures (P=0.01). When enriched peak centers were mapped to the coordinate position across the H. pylori chromosome, differentially enriched peaks heavily clustered around the oriC region, with few peaks mapping to the chromosomal region diametrically opposed to oriC (Fig. 5B), suggesting that DNA translocation is linked to bi-directional DNA replication.
To test the hypothesis that chromosomal regions near oriC are translocated to gastric epithelial cells at a high frequency, we cloned a CMV promoter-driven monomeric nanoluciferase construct into the ureA locus (hp0073) adjacent to oriC (Fig. 5C). Gastric epithelial cells and 293T cells were challenged by WT or cagX H. pylori strains harboring the eukaryotic-optimized nanolucifersase expression construct (WTNL and cagXNL, respectively), and luciferase activity in infected cell lysates was monitored by bioluminescence. Compared to mock infected and cagXNL-infected cells, AGS and 293T cells challenged by WTNL produced high levels of bioluminescence (Fig. 5D and 5E), indicating that nanoluciferase production by epithelial cells is linked to cag T4SS-dependent DNA translocation.
To confirm that targeted DNA fragments proximal to oriC can be transferred to host cells via cag T4SS mechanisms, we replaced the nanoluciferase gene with a eukaryotic-optimized construct designed to express monomeric mScarlet tethered to the LifeAct N-terminal peptide to enable live cell visualization of F-actin. Microscopy analysis of AGS (Fig. 5F) and 293T (Fig. 5G) monolayers challenged by WTmScarlet revealed numerous mScarlet-positive cells compared to corresponding cagXmScarlet-infected co-cultures. Flow cytometry analysis confirmed our observation that significantly more mScarlet-positive epithelial cells were generated when monolayers were challenged by WTmScarlet compared to T4SS-deficient controls (Fig. 5H-K). Together, these data provide direct evidence that targeted chromosomal fragments greater than 1.5 kb in length are excised and delivered to host cells via cag T4SS-dependent mechanisms.
H. pylori trans-kingdom conjugation is mechanistically coupled to chromosomal decatenation
We next sought to understand the mechanism by which effector DNA is coupled to the cag T4SS for transfer to target host cells. The observation that fragments of translocated bacterial DNA map predominantly to the oriC region led to the hypothesis that H. pylori trans-kingdom conjugation is linked to chromosomal replication. In support of this hypothesis, cag T4SS-dependent TLR9 activation was significantly impaired in the presence of the DNA gyrase inhibitor ciprofloxacin in a dose-dependent manner (Fig. 6A). To analyze the contribution of DNA segregation proteins to coupling transfer DNA to the cag T4SS apparatus, we employed targeted mutagenesis to delete individual genes known to be involved in DNA partitioning. Whereas isogeneic mutants deficient in genes encoding the DNA partitioning proteins ParA or ParB induced WT levels of cag T4SS-depenedent TLR9 activation, isogenic mutants lacking genes encoding the DNA translocase ftsK (hp1090) or the recombinase xerH (hp0675) exhibited marked defects in TLR9 stimulation (Fig. 6B), suggesting that chromosomal dimer resolution is required for transfer DNA coupling to the cag T4SS apparatus. TLR9 stimulation defects exhibited by the ftsK mutant could be rescued by genetic complementation with full length FtsK, but not a truncated FtsK harboring only the translocase DNA binding domain (FtsK-γ), which is required for interaction with XerH and XerH-mediated DNA recombination (Debowski et al., 2012) (Fig. 6B). Loss of xerH or ftsK was not associated with defects in cag T4SS-dependent induction of IL-8 synthesis by gastric epithelial cells (Fig. 6C), indicating that chromosomal segregation defects specifically impair cag T4SS phenotypes associated with DNA translocation. Consistent with the observed defects in TLR9 activation, ftsK mutants stimulated significantly reduced levels of cGAS-STING signaling compared to WT and the corresponding FtsK complemented strain (Fig. 6D).
To investigate the role of chromosomal segregation in loading effector DNA into the cag T4SS apparatus, we developed a ‘transfer DNA’ immunopurification assay to capture chromosomal fragments contained within the cag T4SS translocation channel. We reasoned that effector DNA trafficked into the cag T4SS apparatus would physically interact with components of the outer membrane complex that comprise the secretion chamber (Chung et al., 2019; Frick-Cheng et al., 2016). To test this hypothesis, we treated H. pylori with formaldehyde to chemically cross-link DNA-protein complexes, lysed the cells by sonication, and purified the cag T4SS core complex via immunopurification targeting CagY, which forms the central cap-like structure within the inner ring of the outer membrane complex (Chung et al., 2019). PCR analysis targeting a 795 bp chromosomal fragment revealed the presence of DNA in reverse cross-linked CagY complexes purified from WT but not in mock preparations obtained from cagY isogenic mutants (Fig. 6E). DNA co-purification with CagY was not dependent on the presence of either the effector protein CagA or CagT, a component localized to the periphery of the outer membrane ring complex (Chung et al., 2019; Frick-Cheng et al., 2016); however, DNA loading into the cag T4SS apparatus was significantly impaired in the absence of FtsK (Fig. 6E). Together, these data demonstrate that effector DNA is loaded into the cag T4SS machinery prior to encountering host cells to establish a ‘ready-to-fire’ nanomachine and demonstrate that transfer DNA loading is mechanistically coupled to chromosomal replication and replichore decatenation. We thus propose a model in which FtsK-XerH complexes resolve imbalanced replichores arising from overlapping rounds of chromosomal replication, resulting in the rare excision of DNA that is subsequently shuttled through the cag T4SS apparatus via unknown mechanisms (Fig. 6F).
DISCUSSION
STING-dependent immunosurveillance plays a critical role in maintaining gastric mucosal homeostasis and regulating inflammatory responses with carcinogenic potential (Ke et al., 2022). STING signaling has been implicated in the development and progression of stomach cancer, but whether STING activation promotes or restricts gastric carcinogenesis remains unresolved. For example, a previous study demonstrated that STING downregulation in primary gastric tumors was associated with increased malignancy and the progression of gastric cancer (Song et al., 2017), while a conflicting report determined that high STING expression in malignant tissues and tumor-associated macrophages was predictive of poor prognosis in stomach cancer patients (Miao et al., 2020). Thus, the molecular role of cGAS-STING signaling in chronic gastric inflammation and pre-malignant lesion development remains unresolved. Consistent with the observation that chronic H. pylori colonization elicits STING signaling in the murine gastric mucosa (Song et al., 2017), our study establishes a critical role of cag T4SS-mediated trans-kingdom DNA conjugation in stimulating STING-dependent outcomes that underscore infection-associated carcinogenesis.
In the context of bacterial infection, type I IFN responses have been associated with both protective and detrimental outcomes (Boxx and Cheng, 2016; Peignier and Parker, 2021). Previous work demonstrates that the H. pylori cag T4SS-dependent delivery of muropeptide fragments activates Nod1 sensing and IRF7-mediated type I IFN responses in epithelial cells that restrict bacterial proliferation via CXCL10-mediated immune cell recruitment (Viala et al., 2004; Watanabe et al., 2010). Our work demonstrates that during acute infection, CXCL10 is highly upregulated by cag T4SS activity in primary gastric epithelial cells (Fig. 3). In addition to increased transcript levels associated with interferon stimulated genes, we observed augmented production of transcripts associated with SMHD1 [a suppressor of type I IFN and NF-κB signaling (Chen et al., 2018)], IFI35 [a negative regulator of RIG-I signaling (Das et al., 2014)], and ISG15 [an IFN-α/β-inducible ubiquitin-like modifier that is a key negative regulator of IFN-α/β immunity (Zhang et al., 2015)] in WT-infected primary gastric cells, suggesting that H. pylori has evolved mechanisms to counteract or suppress nucleic acid signaling pathways induced by cag T4SS activity. We speculate that similar to evasion strategies employed by numerous viruses, H. pylori counteracts type I IFN responses through the cag T4SS-dependent translocation of as yet unidentified, evolutionarily-conserved protein effectors that target and neutralize components of DNA-sensing or STING signaling pathways. Alternatively, pro-inflammatory STING signaling may balance immunosuppressive responses stimulated by cag T4SS-dependent TLR9 activation (Varga et al., 2016a) as a mechanism to sustain gastric homeostasis during acute colonization.
Epithelial DNA damage, genomic alterations, and chromosomal instability are hallmarks of H. pylori-induced gastric cancer (Chaturvedi et al., 2014). While H. pylori cag T4SS activity has been shown to induce the formation of both double-stranded DNA breaks (DSBs) and micronuclei that potentially contribute to cGAS-STING signaling in gastric epithelial cells (Bauer et al., 2020; Hanada et al., 2014; Koeppel et al., 2015), our data suggest that translocated chromosomal DNA fragments serve as the predominant cytosolic cGAS trigger in the context of H. pylori infection (Fig. 1 and Fig. S1). In support of our data indicating that translocated chromosomal fragments are the key agonist that alerts cytosolic nucleic acid surveillance, we found that H. pylori effector DNA is packaged into exosomes released by gastric epithelial cells to enable paracrine-like signal amplification (Fig. 4). Consistent with a previous study demonstrating that intracellular Listeria monocytogenes infection stimulates the production of extracellular vesicles with IFN-inducing potential (Nandakumar et al., 2019), the delivery of pathogen-derived DNA via exosome release may represent a conserved mechanism by which epithelial cells potentiate nucleic acid surveillance-dependent danger signaling to uninfected tissues. We propose that H. pylori has evolved to exploit this innate defense mechanism to shape a tolerogenic immune response that favors persistent gastric colonization (Varga et al., 2016a).
Mechanistically, our data demonstrate that cag T4SS-dependent DNA delivery is coupled to chromosomal replication and replichore decatenation (Fig. 5 and Fig. 6). Similar to Neisseria gonorrhoeae ssDNA secretion mechanisms (Callaghan et al., 2017), chromosomal partitioning influences cag T4SS-mediated DNA translocation. Indeed, our studies reveal that effector DNA coupling to the cag T4SS apparatus requires both FtsK membrane tethering and DNA translocase activity (Fig. 6). In support of the critical role of replichore decatenation in cag T4SS-dependent DNA translocation, H. pylori lacking the DNA recombinase XerH, which requires direct interaction with FtsK to resolve chromosome dimers and catenated chromosomes (Bebel et al., 2016; Debowski et al., 2012), exhibited significantly reduced levels of TLR9 stimulation. Our studies suggest that while xerH mutants harbor increased DNA content per cell compared to the parental WT strain (Debowski et al., 2012), cag T4SS DNA translocation phenotypes are significantly impaired when chromosomal DNA topological isomers cannot be efficiently resolved (Fig. 6). Additionally, our findings establish that DNA effector molecules are pre-loaded into the cag T4SS apparatus to establish a ‘ready-to-fire’ nanomachine that can be rapidly deployed during acute infection, and demonstrate that FtsK is required for partitioning DNA substrates into the cag T4SS (Fig. 6). Although H. pylori harbors several orthologs to paradigmatic DNA conjugation systems, endonucleases, phage integrases, and canonical A. tumefaciens VirD2 relaxases, the mechanism by which chromosomally-derived effector DNA is precisely excised and shuttled through the cag T4SS while maintaining faithful genome inheritance remains unresolved. Thus, further studies are warranted to explore the intriguing possibility that the cag T4SS has co-opted orphaned components of remnant tfs3 and tfs4 conjugation systems to enable trans-kingdom DNA delivery (Fernandez-Gonzalez and Backert, 2014).
Our results provide the first direct evidence that H. pylori delivers immunostimulatory chromosomal DNA fragments into target epithelial cells and demonstrate that translocated effector DNA physically binds and activates cytosolic DNA-sensing reconnaissance systems. The observation that chromosomal fragments encompassing full-length eukaryotic genes can be excised and directed to the cag T4SS for delivery into target cells underscores the potential to exploit H. pylori trans-kingdom conjugation for targeted DNA cargo delivery. Notably, our work provides the essential experimental framework required to harness the cag T4SS as a mucosal-targeted DNA vaccine or therapeutic delivery device that can be pharmacologically controlled in vivo. Studies to determine the maximal DNA fragment length that can be efficiently transported to gastric cells via cag T4SS mechanisms are currently underway.
Our work highlights the importance of understanding host-pathogen conflicts that engage cellular PRR signaling axes to drive chronic inflammation and stimulate the development of infection-related malignancies. While we propose that translocated microbial DNA plays a critical and underappreciated role in the development of inflammation-associated carcinogenesis, further investigations are required to understand STING-dependent outcomes within the context of gastric cancer. Our findings identify a central role of cag T4SS activity in eliciting cGAS-STING immunosurveillance in the gastric epithelium; however, additional studies are necessary to understand how STING activation shapes the gastric immune landscape to enable persistent H. pylori colonization and to elucidate whether STING signaling influences the development of pre-malignant lesions. Finally, our work more broadly points towards therapeutic STING modulation as a promising intervention strategy to reduce the incidence and severity of infection-associated malignancies that arise as a consequence of chronic inflammation.
METHODS
Bacterial strains and culture conditions
Helicobacter pylori strain 26695, strain G27, isogenic derivatives, and clinical isolates were propagated on trypticase soy agar plates supplemented with 5% sheep blood (BD) as previously described (Shaffer et al., 2011). Overnight cultures of H. pylori were grown in Brucella broth supplemented with 5% fetal bovine serum (FBS) at 37°C with 5% CO2. For cloning purposes, E. coli DH5α (New England Biolabs) were grown on lysogeny broth (LB) agar plates or in LB liquid medium with appropriate antibiotics required for plasmid maintenance.
Human cell culture
The AGS human gastric adenocarcinoma epithelial cell line (ATCC CRL-1739) was cultured in RPMI 1640 medium supplemented with 10% FBS, 2 mM L-glutamine, and 10 mM HEPES in the presence of 5% CO2 at 37°C. Primary adult human gastric epithelial cells (Cell Biologics H-6039) were grown in human epithelial cell medium supplemented with ITS, EGF, hydrocortisone, L-glutamine, and 5% FBS (Cell Biologics H6621) in collagen-coated cell culture flasks at 5% CO2 and 37°C. Prior to assays, wells of tissue culture plates were coated in collagen following manufacturer’s protocol. HEK293-hTLR9 (Invivogen hkb-htlr9), the corresponding parental HEK293 null1 (InvivoGen hkb-null1), HEK293-hTLR7 (Invivogen hkb-htlr7), HEK-Lucia™ hRIG-I (Invivogen hkl-hrigi), and 293T (ATCC CRL-3216) were grown in DMEM supplemented with 10% heat-inactivated FBS and 1X GlutaMAX (Life Technologies) in the presence of 5% CO2 at 37°C. For cell treatments prior to bacterial challenge, the following compounds were added at the indicated final concentration one hour prior to bacterial challenge: MitoTempo (Sigma-Aldrich, 100 μM); N-acetylcysteine (Sigma-Aldrich, 1 mM); BAX-inhibiting peptide, negative control (Sigma-Aldrich, 200 μM); BAX-inhibiting peptide V5 (Sigma-Aldrich, 200 μM); or sphingomyelinase inhibitor GW4869 (Sigma-Aldrich, 10 μM).
H. pylori mutagenesis
Isogenic mutant derivatives of H. pylori 26695 and G27 were generated essentially as previously described (Johnson et al., 2014; Shaffer et al., 2011; Varga et al., 2021). Briefly, H. pylori was transformed with a suicide plasmid in which the coding region of the target gene was replaced by either a kanamycin resistance cassette or a chloramphenicol resistance cassette and homologous flanking DNA sequences 500 base pairs (bp) up-and downstream of the target locus. Colonies resistant to kanamycin (12.5 μg/ml) or chloramphenicol (10 μg/ml) were PCR-verified to confirm insertion of the resistance cassette into the appropriate locus. To complement mutants in cis at the ureA locus, plasmids derived from pAD1 (Shaffer et al., 2011) were constructed to express either the native gene or a hemagglutinin (HA) epitope-tagged protein as previously described (Shaffer et al., 2011; Varga et al., 2021). Plasmid sequences were confirmed by sequencing, and constructs were used to transform isogenic mutant strains. Insertion of complementation constructs into the ureA locus was confirmed by PCR amplification and anti-HA immunoblotting, when appropriate.
To construct H. pylori strains harboring NanoLuc luciferase expression constructs, a 1654 bp fragment encompassing the CMV enhancer element, CMV promoter, NanoLuc luciferase gene, and SV40 late poly(A) signal was amplified from pNL1.1.CMV[Nluc/CMV] (19.1 kDa NanoLuc protein, Promega). The PCR product was blunt end ligated to pAD1 digested with EcoRV. Ligation insertion was confirmed by PCR and DNA sequencing. Clones in which the NanoLuc luciferase construct was inserted in the reverse orientation relative to ureA transcription were selected for transformation into WT and cagX H. pylori 26695 and WT G27 strains. Integrations into the H. pylori chromosome were confirmed by PCR. The G27 isogenic cagE mutant was generated by replacement of the cagE coding region with a kanamycin resistance cassette as previously described (Shaffer et al., 2011; Varga et al., 2021). H. pylori strains harboring monomeric LifeAct-mScarlet (Bindels et al., 2017) expression constructs were generated using the same mutagenesis strategy, with the exception that a 1872 bp region of pLifeAct_mScarlet-i_N1 (Bindels et al., 2017) (26.4 kDa LifeAct-mScarlet protein, Addgene) encompassing the CMV enhancer element, CMV promoter, LifeAct-mScarlet gene, and SV40 late poly(A) signal was amplified was cloned into pAD1 via blunt end ligation. Clones in which the LifeAct mScarlet construct was inserted in the reverse orientation relative to ureA transcription were selected for transformation into H. pylori 26695 and G27 strains. Integrations into the ureA locus were confirmed by PCR.
cGAS-STING reporter assays
293T cells seeded into 24-well plates at a density of approximately 1.5 × 105 cells per well were transfected using Lipofectamine 2000 (Life Technologies) complexed with a combination of plasmids pUNO1-hSTING-R232 (Invivogen puno1-hStingWT), pUNO1-hcGAS-HA3X (Invivogen pUNO1-HA-hcGAS), cGAS derivatives (pUNO1-hcGAS-AA, Invivogen pUNO1-hcGAS-AA; pcDNA3.1-cGASΔDBD, Addgene 102606), pRL-SV40P (Addgene 27163), IFN-Beta-pGL3 (Addgene 102597), or empty vector pcDNA3.1 (Life Technologies V79020) to ensure equivalent DNA concentrations according to the manufacturer’s protocol. At 16 h post-transfection, cells were challenged by mid-log phase WT H. pylori, corresponding cag isogenic mutants, or cag T4SS+ clinical isolates at a multiplicity of infection (MOI) of 20 bacterial cells per 293T cell. Alternatively, STING reporter cells were transfected with purified H. pylori genomic DNA (500 ng/well) at 16 h post-initial transfection of STING and IFN-β reporter plasmids. For cGAS-STING time course assays, H. pylori were added to reporter cell monolayers at a MOI of 20, and culture plates were centrifuged at 1,800 x g for 1 min to synchronize infections. At the indicated time point, gentamicin was added at a final concentration of 100 µg/ml.
At 24 h post-infection, cell supernatants were removed, and adherent cells were lysed in luciferase assay passive lysis buffer (Pierce). Luciferase luminescence generated by 20 µl cell lysate per well in a 96-well plate format was recorded with a microplate reader (BioTek Synergy H1) using the Dual-Luciferase Reporter Assay System (Pierce). Firefly luciferase luminescence (D-Luciferin, IFN-Beta-pGL3) was normalized to Renilla luciferase luminescence (coelenterzine, pRL-SV40P) for each well, and IFN-β transcriptional reporter values were expressed as the normalized fold change over mock-infected wells. A minimum of three biological replicate experiments were performed in quadruplicate for each assay.
To evaluate the requirement of direct bacteria-host cell contact for cGAS-STING signaling, 293T cells were seeded directly into the well of a 24-well plate or into a 0.4 µm pore size transwell insert at approximately 5 × 104 cells per transwell. Cells were transfected as described for cGAS-STING reporter cell assays. To physically separate bacterial cells from reporter cells, transfected cGAS-STING reporter cells challenged by the addition of WT H. pylori 26695 added to either the apical transwell chamber (293T cells seeded directly into the cell culture well), or the basolateral compartment (293T cells seeded in the transwell apparatus) at a MOI of 20. As a control, cGAS-STING assays in which H. pylori were added directly to reporter cell monolayers were performed in parallel. After 24 h of bacterial infection, monolayers were processed as described for cGAS-STING activation assays, and luminescence values were obtained. IFN-β transcriptional reporter values were expressed as the normalized fold change over mock infected wells.
STING transactivation assays
For STING transactivation assays, 293T cells were transfected with either (i) pUNO1-hcGAS-HA3X, pRL-SV40P, and pcDNA3.1 empty vector, or (ii) pUNO1-hSTING-R232, pRL-SV40P, and IFN-Beta-pGL3 using Lipofectamine 2000 as described for cGAS-STING activation assays. At 12 h post-transfection, cells were lifted using sterile phosphate buffered saline (PBS) and were re-plated in 24-well dishes at a 1:1 ratio at approximately 2 × 105 cells/well. Cells were allowed to adhere and were subsequently challenged by the indicated H. pylori strain at a MOI of 20. After 24 h of bacterial infection, IFN-β transcriptional reporter activity was assayed as described for cGAS-STING activation experiments. STING transactivation by cGAMP diffusion was expressed as the normalized fold change of IFN-β transcriptional reporter activity over mock infected wells.
TLR and RIG-I activation assays
To assess TLR9 or TLR7 activation, HEK293 stably transfected with hTLR9, hTLR7, or parental null1 cells (Invivogen) were seeded into 96-well plates at a density of approximately 2 × 104 cells per well, and were subsequently by WT H. pylori 26695 or its isogenic mutant strains at a MOI of 100 in quadruplicate as previously described (Varga et al., 2016b; Varga et al., 2021). Supernatants were collected at 24 h post-infection, and TLR9 and TLR7 activation was determined by measuring secreted embryonic alkaline phosphatase (SEAP) reporter activity in cell culture supernatants by QuantiBlue™ reagent (Invivogen) using a microplate reader (BioTek Synergy HI) to record the absorbance at 650 nm. As a positive control for TLR7 activation, HEK293-hTLR7 reporter cells were stimulated with 5 μg/ml imiquimod (R837), an imidazoquinoline amine analog to guanosine (Invivogen). TLR9 or TLR7 activation was normalized to SEAP levels produced by infected null1 parental cells and is expressed as the fold-change over mock infected controls. For ciprofloxacin inhibition of bacterial DNA replication, ciprofloxacin was added to HEK293-hTLR9 co-cultures at the same time as bacterial inoculation at multiples of the previously reported minimum inhibitory concentration (MIC), with 1X MIC equivalent to 0.125 µg/ml. Inhibition of TLR9 activation is expressed as a percent of the normalized fold change over mock treated, H. pylori-challenged wells.
For RIG-I activation studies, HEK-Lucia™ hRIG-I cells were plated in a 96-well dish at approximately 5 × 104 cells per well and were subsequently challenged by WT H. pylori 26695 or the indicated isogenic mutant strain at a MOI of 100 in quadruplicate. As a positive control, cells were transfected with 100 ng/ml of the RIG-I agonist 5’ triphosphate hairpin RNA (3p-hpRNA, Invivogen) complexed to Lipofectamine 2000 (Life Technologies) according to the manufacturer’s protocol. At 24 h post-challenge, RIG-I stimulation was assessed by analyzing Lucia luciferase reporter gene expression in 20 µl cell culture supernatants using QUANTI-Luc™ (Invivogen). Data are expressed as the normalized fold change over mock infected cells. TLR and RIG-I stimulation experiments were performed a minimum of three times with quadruplicate technical replicates per experimental condition.
CagA translocation assays
AGS or 293T cells were plated in 12-well dishes at a seeding density of 1 × 105 cells/well and were cultured overnight. Monolayers were challenged by the indicated H. pylori strain at a MOI of 100 for 6 h, as previously described (Shaffer et al., 2011; Varga et al., 2021). Wells were washed in sterile PBS to remove non-adherent bacteria, and cells were lysed in assay buffer (1% NP-40) supplemented with cOmplete™ mini EDTA-free protease inhibitor (Roche) and PhosSTOP phosphatase inhibitor (Roche). The soluble fraction was collected and prepared in 2X SDS buffer for immunoblotting. To assess CagA tyrosine phosphorylation, AGS or 293T samples were separated on 7.5% gels (Bio-Rad) for 60 min at 165V. Proteins were then transferred onto nitrocellulose using the TransBlot Turbo system (Bio-Rad) following manufacturer’s recommendations. Membranes were blocked in 3% BSA in Tris-buffered saline containing 0.1% Tween-20 (TBST), followed by incubation with anti-phosphotyrosine monoclonal antibody (α-PY99, Santa Cruz Biotechnology). Total CagA levels were assessed by subsequent incubation with an anti-CagA monoclonal antibody (α-CagA, Santa Cruz Biotechnology). Phosphorylation of CagA or total CagA was visualized using chemiluminescence (Pierce). TEM-CagA translocation was assayed as previously described (Varga et al., 2021).
Bacterial adherence and internalization assays
Adherence and internalization into gastric or kidney epithelial cells were assessed as previously described(Varga et al., 2021). Briefly, WT H. pylori or the cagX isogenic derivative were co-cultured with AGS or 293T cells at a MOI of 100. After a 4 or 6 h infection, respectively, cell culture medium was aspirated, and cell monolayers were gently washed with sterile PBS to remove non-adherent bacteria. To assess intracellular bacteria, RPMI or DMEM supplemented with gentamycin (100 μg/mL) was added to a subset of wells, and the cells were incubated for an additional hour at 37°C in 5% CO2. To assess total adherent and intracellular bacteria, fresh RPMI or DMEM was replenished into the remainder of wells following the removal of non-adherent bacteria. After the 1 h incubation, RPMI or DMEM was aspirated and all wells were washed in sterile PBS, lysed by mechanical disruption, and were serially diluted on blood agar plates for colony enumeration. Experiments were performed a minimum of three times with triplicate technical replicates per cell line and culture condition.
Detection of bacterial DNA in AGS cytosolic extracts
Digitonin extracts of AGS gastric epithelial cells were prepared essentially as previously described(West et al., 2015). Briefly, approximately 8.4 × 106 AGS cells were infected with exponentially growing WT or cagX H. pylori at a MOI of 50. As a control, an equivalent number of AGS cells were mock infected. After 6 h of infection, cells were rinsed with sterile PBS and were treated with 50 units of Turbo DNaseI (Life Technologies) in digestion buffer at 37°C for 15 min. The cells were rinsed twice with PBS, trypsinized, and collected in 2 ml of sterile PBS. Collected cells were separated into aliquots of approximately 400 µl to generate ‘total’ cell extracts, and of approximately 1600 µl to generate ‘cytosolic’ extracts. Aliquots were centrifuged at 980 x g for 3 min and cell pellets were washed once with PBS. To generate the ‘total’ cell extract, one pellet for was re-suspended in 100 µl of 50 µM NaOH and incubated for 30 min at 95°C to solubilize DNA, followed by pH neutralization by the addition of 10 µl of 1 M Tris-HCl, pH 8. These extracts served as normalization controls for the quantitation of mitochondrial DNA (mtDNA) and bacterial DNA. To generate ‘cytosolic’ extracts, cell pellets were re-suspended in cytosolic extraction buffer (150 mM NaCl, 50 mM Tris pH 8.0, and 20 µg/ml Digitonin [Sigma-Aldrich]), and homogenates were rotated end-over-end for 10 min at room temperature for selective membrane permeabilization. Cytosolic fractions were separated from intact cells and nuclear/bacterial fractions by serial centrifugations at 17,000 x g for 3 min. Recovered supernatants were incubated for 10 min at 95°C to isolate DNA. To assess the presence of nuclear DNA, mtDNA, and H. pylori DNA in cellular fractions, standard PCR and qPCR targeting fragments of the H. pylori chromosome (hp1421, 290 bp), mtDNA (coxII, 444 bp(Fernandez-Moreno et al., 2016)), and nuclear DNA (hNuc, 467 bp (Fernandez-Moreno et al., 2016)) were amplified from ‘total’ and ‘cytosolic’ fractions using Taq polymerase (standard PCR) or Fast SYBR Green chemistry (qPCR) on a Viia7 platform (Thermo). CT values obtained for cytosolic fractions were normalized to corresponding CT values obtained for total cell extracts, and cytosolic enrichment of bacterial DNA was calculated as the normalized ratio of hp1421 CT values to coxII CT values. A minimum of four biological replicate experiments were performed for each experimental condition.
Quantitation of secreted extracellular cGAMP
AGS cells seeded into T25 flasks were grown to approximately 80% confluency were mock infected or were challenged by H. pylori at a MOI of 100 for 6 h. Bacteria were inactivated by the addition of 200 µg/ml gentamicin, and monolayers were incubated overnight. At 24 h post-infection, equivalent volumes of cell culture supernatants were collected and concentrated via solvent evaporation. Samples were reconstituted in one-tenth volume assay buffer, and cGAMP levels were quantified using the DetectX® 2’,3’-cGAMP STING-Based FRET assay (Arbor Assays). For each experimental condition, cGAMP secretion assays were performed in triplicate and data represents a minimum of three biological replicates.
Confocal laser scanning microscopy
Adult primary human gastric epithelial cells were grown on collagen-coated 12 mm glass coverslips (#1.5, 170 µm thickness) overnight prior to challenge by WT or cagX H. pylori at a MOI of 50. As a control, a subset of coverslips was mock infected. After 6 h, coverslips were washed in sterile PBS three times, followed by fixation in 4% paraformaldehyde in PBS for 20 min at room temperature. Coverslips were washed in PBS and cells were permeabilized in confocal blocking buffer (3% bovine serum albumin, 0.1% TritonX-100, 1% saponin in sterile PBS) for 1 h at room temperature. For immunostaining, coverslips were stained with anti-STING monoclonal antibody (Life Technologies, 1:100) in confocal blocking buffer overnight at 4°C. Coverslips were rigorously washed three times in PBST to remove unbound antibody and were subsequently incubated in AlexaFluor 488-conjugated secondary antibody (Life Technologies, 1:1000) in blocking buffer for 1 h at room temperature. For visualization of the nuclei and actin, samples were stained with stained with 4’,6-diamidino-2-phenylindole (DAPI) and AlexaFluor 594 phalloidin for 1 h at room temperature. Coverslips were washed in PBS and were mounted with ProLong Gold antifade (Life Technologies). Fluorescence images were captured using a 60X silicon immersion objective on an Olympus FV3000 confocal laser scanning microscope and images were acquired and processed using cellSens software (Olympus). Quantification of STING particle size and number was performed using Fiji software (ImageJ) with automated thresholding and subsequent particle analysis of segmented images for mock infected (7 fields of view, n=90 cells); WT infected (11 fields of view, n=82 cells); and cagX infected (9 fields of view, n=59 cells) gastric epithelial cells. To normalize across imaging conditions, average particle sizes were calculated by multiplying the average pixel area by the pixel resolution for each field of view.
RNA isolation and gene expression analyses
Primary gastric epithelial cells were mock infected (n=3 biological replicates) or co-cultured with WT H. pylori 26695 (n=6 biological replicates) or the corresponding cagX isogenic mutant (n=3 biological replicates) for 6 h. Cells were washed three times with sterile PBS and total RNA was isolated using the Direct-Zol Miniprep Plus kit (Zymo Research) following the manufacturer’s protocol. Total RNA was stringently digested with Turbo DNase I (Invitrogen) to remove contaminant DNA. To determine inflammatory gene expression changes in response to H. pylori infection, DNA-free RNA (100 ng per sample) was analyzed on the nCounter Sprint Profiler (NanoString Technologies) using the nCounter Human Host Response Panel (NanoString Technologies), which simultaneously quantifies transcripts for 773 immune-related genes and 12 internal reference genes. Differences in gene expression between experimental groups was calculated using the ROSALIND Platform for nCounter Analysis (https://rosalind.onramp.bio/). Raw data (RCC files) were normalized to internal reference genes the nSolver 4.0 software integrated within the ROSALIND platform (ROSALIND, Inc.). Gene normalization was performed using housekeeping probes selected based on the geNorm algorithm as implemented in the Bioconductor package NormqPCR. Differentially expressed genes in H. pylori challenged cells were determined using the Benjamini– Hochberg P value adjustment method of estimating false discovery rates (FDR), with significance set at p<0.05. Read distribution percentages, violin plots, identity heatmaps, and sample multidimensional scaling (MDS) plots were generated within ROSALIND during sample QC. Read normalization, gene expression fold changes, and the associated P values were calculated using criteria provided by NanoString. Pathway enrichment analysis was performed within ROSALIND using the REACTOME database, and gene term enrichment was calculated using a hypergeometric distribution algorithm in reference to the background set of genes in the panel with significance set at p<0.01 and greater than ± 1.8-fold gene expression enrichment. Volcano plots and heat maps were generated in GraphPad Prism using normalized gene expression data exported from ROSALIND. Volcano plots were constructed by plotting the log2 of the normalized fold change versus the −log10 of the adjusted P value for each gene. The dashed Padj lines demarcate genes meeting the threshold for significance (Padj<0.01 and Padj<0.05) after correction with the Benjamini–Hochberg procedure for controlling FDR.
Gastric epithelial cell supernatant transfer assay
Primary gastric epithelial cells seeded into 24-well gelatin-coated dishes were mock infected or were co-cultured with WT H. pylori or the cagX isogenic mutant at a MOI of 100 for 6 h or 12 h. Cell culture supernatants were harvested and treated with 100 µg/ml gentamicin for 1 h at 37°C to eliminate viable bacteria. Supernatants were subsequently spun for 10 min at 10,000 x g to remove bacteria and gastric cells, and cleared supernatants were stored at −20°C. For TLR9 stimulation studies, HEK293-hTLR9 cells were seeded into 96-well plates at a density of approximately 2 × 104 cells per well in a volume of 100 µl DMEM per well. Primary gastric epithelial cell supernatants were added to hTLR9 reporter cells at an equal volume and incubated for 24 h at 37°C in 5% CO2. TLR9 activation was determined via SEAP levels determined by QuantiBlue™ (Invivogen) using a microplate reader (BioTek Synergy HI) to record the absorbance at 650 nm. The concentration of cell-free DNA in processed supernatants was determined by high sensitivity dsDNA Qubit assay (Life Technologies).
For enzyme treatments of gastric epithelial cell supernatants, pre-cleared supernatants obtained from 6 h H. pylori challenged gastric epithelial cells were either treated by (i) the addition of 10 units Turbo DNase (Life Technologies) and incubation at 37°C for 30 min, (ii) heating the supernatant to 70°C for 15 min, or (iii) heating the supernatant to 70°C for 15 min followed by cooling to room temperature and subsequent DNase treatment as described in (i). Treated supernatants were added to HEK293-hTLR9 reporter cells as described for supernatant transfer assays, and TLR9 stimulation was assessed by QuantiBlue™ after 24 h incubation. TLR9 activation is expressed as the fold change over mock treated HEK293-hTLR9 cells.
Extracellular vesicle (EV) purification
Primary gastric epithelial cells were challenged by H. pylori or were mock infected for 6 h. Supernatants were pre-cleared by serial centrifugations (10,000 x g) at 4°C. EV-containing supernatants (2 ml per biological replicate) were subsequently magnetically labeled for 1 h at room temperature by CD9, CD63, and CD81 antibodies (Human Exosome Isolation Kit, Pan, Miltenyi Biotec) followed by EV isolation via magnetic separation and elution. Immunoaffinity purified exosomes were subjected to qPCR analysis probing for H. pylori genomic DNA (hp1421 locus, 290 bp fragment) and mtDNA (coxII, 444 bp fragment) by Fast SYBR Green chemistry (Life Technologies) on a Viia7 platform (Thermo). H. pylori DNA enrichment within purified EVs was determined by quantifying the ratio of hp1421 CT values to coxII CT values for EVs obtained from WT and cagX-infected gastric cells. EVs purified from mock infected gastric epithelia contained levels of mtDNA similar to EVs derived from H. pylori-challenged primary cells.
Immunoprecipitation of cGAS-DNA complexes
293T cells were grown in 6-well plates for 24-30 hours to achieve approximately 90% confluency. Cells were transfected with pUNO1-hcGAS-HA3X (240 ng DNA/well) complexed to Lipofectamine 2000. After 12 – 16 h, approximately 7.2 × 106 transfected 293T cells were challenged by exponentially growing WT or cagX H. pylori at a MOI of 100. As a control, an equivalent number of transfected 293T cells were mock infected. After 6 h of infection, 293T cells were rinsed with PBS and DNA-protein complexes were cross-linked by 1% paraformaldehyde for 15 min at room temperature. Cross-linking reactions were quenched by the addition of 250 mM glycine, and cells were collected by mechanical detachment and centrifugation at 4000 rpm for 10 min. Cell pellets were washed once with sterile PBS, followed by re-suspension in pre-chilled lysis buffer (5 mM EDTA, 1% NP-40, and 1X protease inhibitor cocktail in PBS) and sonication at 5% amplitude (10 Sec ON, 10 Sec OFF, 4-6 cycles) to generate cleared lysates. Sonicated cell extracts were centrifuged at 14,000 rpm for 30 min at 4°C. The recovered cell-free supernatant was incubated with 4 µg of anti-HA monoclonal antibodies (clone 12CA5) overnight at 4°C with continuous end-over-end rotation. The following day, 50 µl of Protein G Dynabeads (Life Technologies) pre-blocked in PBS containing 1% BSA were incubated with immunopurification samples for 90-120 min at 4°C with continuous end-over-end rotation. Dynabeads were collected by magnetic isolation, washed twice with 1X cell lysis buffer, followed by one wash in high salt wash buffer (cell lysis buffer + 300 mM NaCl). Magnetic beads were re-suspended in 100 µl of 1% SDS + 0.1 M sodium bicarbonate buffer and de-crosslinked by incubation at 60°C overnight. Purification of cGAS-HA3x was confirmed in the eluted fractions by immunoblot analysis. To purify DNA complexed with cGAS-HA3x, de-crosslinked fractions were treated with 20 µg of Proteinase K (Sigma-Aldrich) at 60°C for 1-2 h, followed by DNA isolation via the ChIP DNA Clean and concentrator kit (Zymo Research) according to the recommended protocol. Eluted DNA was quantified using the Qubit high sensitivity dsDNA assay (Life Technologies).
cGAS ‘ChIP-seq’ library preparation and sequencing
ChIP samples were quantified using Qubit 2.0 Fluorometer (Life Technologies) the DNA integrity was analyzed with 4200 TapeStation (Agilent Technologies). cGAS ‘ChIP-seq’ library preparation and sequencing reactions were conducted at GENEWIZ, Inc. (South Plainfield, NJ, USA). NEB NextUltra DNA Library Preparation kit was used following the manufacturer’s recommendations (Illumina). Briefly, DNA eluted from cGAS immunopurifications was end repaired and adapters were ligated after adenylation of the 3’ ends. Adapter-ligated DNA was size selected, followed by clean up, and limited cycle PCR enrichment. The cGAS ‘ChIP’ library was validated using Agilent TapeStation and quantified using Qubit 2.0 Fluorometer as well as RT-PCR (Applied Biosystems). Sequencing libraries were multiplexed and clustered on two lanes of a flowcell. After clustering, the flowcell was loaded on the Illumina HiSeq instrument according to manufacturer’s protocol (Illumina). Sequencing was performed using a 2×150 paired end (PE) configuration. Image analysis and base calling were conducted by the HiSeq Control Software (HCS). Raw sequence data generated from Illumina HiSeq was converted into fastq files and de-multiplexed using Illumina’s bcl2fastq 2.17 software. One mismatch was allowed for index sequence identification. Sequence reads were processed to remove adapter sequences and nucleotides with poor quality at both 5’ and 3’ ends using CLC Genomics workbench. Sequence reads below 15 bases were discarded. Trimmed data was aligned to both human (Homo sapiens reference genome hg38) and H. pylori 26695 (reference genome NC_000915) reference genomes. Only specific alignment was allowed during mapping. To detect peaks that were differentially present in cGAS purifications obtained from WT-infected cells versus cagX-infected cells, reads were normalized to mock infected control preparations, and peak calling was performed using the Transcription Factor model within CLC Genomics workbench with p<0.01 considered significant.
Nanoluciferase (NanoLuc) bioluminescence assays
AGS or 293T cells were cultured overnight to reach approximately 70% confluence, and were challenged by WT or cagX H. pylori [NanoLuc] (26695) or WT or cagE H. pylori [NanoLuc] (G27) at a MOI of 50 for 24 h. Cell culture supernatants were removed and monolayers were washed in sterile PBS to remove non-adherent cells. Monolayers were lysed in 1% NP-40, and 50 µl of cell lysate was transferred to a white walled 96-well plate. To measure nanoluciferase bioluminescence, 20 µl Nano-Glo Luciferase substrate assay buffer containing furimazine (Promega) prepared according to the manufacturer protocol was added to each well. Luciferase activity was immediately assessed using a BioTek Synergy H1 plate reader with luminesce acquisition settings set as recommended by the manufacturer, with the exception of the gain which was adjusted to 230. To determine the level of background NanoLuc activity produced by H. pylori, 20 µl of overnight bacterial cultures were directly lysed in Nano-Glo Luciferase substrate assay buffer containing furimazine, and luciferase values were immediately obtained via plate reader using the same parameters as for eukaryotic cells. Background luminescence produced by H. pylori was determined by normalizing luciferase values by the corresponding culture OD600 and is expressed as the fold change in luminescence over values obtained for H. pylori cultures that do not harbor the nanoluciferase expression construct. Bioluminescence of H. pylori-challenged wells is expressed as the fold change over mock infected wells. For each eukaryotic cell line, a minimum of four biological replicate experiments were performed in 24-well plate technical replicate format.
Live cell fluorescence microscopy analysis of LifeAct-mScarlet
H. pylori harboring LifeAct-mScarlet constructs were grown to exponential phase in broth culture and AGS or 293T monolayers were inoculated at a MOI of 50 overnight at 37°C in 5% CO2. Monolayers were washed in sterile PBS to remove non-adherent bacteria, and monolayers were imaged via live cell, phase contrast epi-fluorescence microscopy on a Nikon Ti Eclipse equipped with a 594em filter and a Plan Apo VC 20X/0.75 NA air objective. Fluorescent images were superimposed on the corresponding phase contrast image of the same field of view. Images were processed for equivalent contrast, brightness, and magnification using the OMERO platform (Allan et al., 2012).
Flow cytometry analysis
AGS or 293T cells were cultured overnight to reach approximately 70% confluence, and were challenged by WT or cagX H. pylori [LifeAct-mScarlet] at a MOI of 50 for 24 h. Cell culture supernatants were removed and monolayers were washed in sterile PBS to remove non-adherent cells. Cells were trypsinized (AGS) or mechanically detached (293T) from tissue culture flasks, washed once in PBS via centrifugation and pelleting, and re-suspended in PBS at approximately 1 × 106 cells/ml. Samples were analyzed on an Attune™ NxT Flow Cytometer (Thermo). Forward scatter-height (FSC-H) and sideward scatter-height (SSC-H) profiles were used in gating strategies to select for single cells, and positive mScarlet fluorescence gates were determined by analyzing 293T cells that had been transfected with pLifeAct_mScarlet-i_N1 (Bindels et al., 2017).
IL-8 quantitation
Interleukin-8 (IL-8) secretion was monitored by human CXCL8 ELISA (R&D Systems) as previously described(Shaffer et al., 2011; Varga et al., 2021). Briefly, AGS cells were plated in 24-well dishes and were cultured overnight prior to infection with H. pylori at a MOI of 100 for 4.5 h. Supernatants were collected and stored at −20°C until analysis by ELISA. A minimum of three biological replicate experiments were performed in triplicate for all strains, and IL-8 secretion is expressed as a percent of IL-8 levels induced by WT H. pylori for each replicate experiment.
‘Transfer DNA’ immunoprecipitation
To assess whether processed chromosomal DNA fragments physically associate with the cag T4SS outer membrane complex, H. pylori grown for 24 h on blood agar were harvested in 2 ml PBS, and 50 µL of the cell suspension were removed to serve as the ‘input’. The remainder of the collected cells were pelleted by centrifugation and were washed once in PBS. Cell pellets were re-suspended in 500 µL PBS, and protein-DNA complexes were cross-linked by the addition of 500 µL 1% paraformaldehyde for 10 min at room temperature, followed by quenching with 1 ml 250 mM glycine. Cross-linked cells were pelleted and washed once with PBS. Cell pellets were re-suspended in 1 mL lysis buffer (5 mM EDTA, 1% NP-40 in PBS) supplemented with 2X cOmplete™ inhibitor (Roche) and were sonicated until the lysate became turbid. Cell lysates were treated with 2 units of Turbo DNase (Life Technologies) in 10 mM MgCl2 for 30 min at room temperature. To solubilize membranes, 0.2% SDS and 0.2% sodium deoxycholic acid (final concentrations) were added to cell lysates, and samples were rotated end-over-end for 1-2 h. Supernatants were separated from insoluble cell debris by centrifugation at 14,000 rpm for 30 minutes at 4°C. In a separate tube, 4 µl polyclonal anti-CagY antisera (a kind gift from Dr. Tim Cover) was added to 800 µl lysis buffer containing 10 mM MgCl2, 2 units Turbo DNase, and 25 µl Protein G Dynabeads, and was incubated for 10 min at room temperature. Cleared supernatants were added directly to CagY antibody solutions and were incubated for an additional 2-4 h with continuous end-over-end rotation. Beads were isolated by magnetic separation and were washed twice in PBS supplemented with 10 mM MgCl2 and 1 unit Turbo DNase, followed by two washes in high salt buffer (lysis buffer containing 400 mM NaCl), and a final wash in PBS. Protein-DNA complexes were eluted and de-cross-linked in 100 µl 1% SDS in 0.1 M NaHCO3 at 65°C overnight. The following day, proteins were digested using Proteinase K (10 µg) at 65°C for 30 minutes. DNA was precipitated by 100% ethanol in 0.3 M sodium acetate (1:3 v/v) at −20°C. Precipitated DNA was pelleted by centrifugation at 14,000 rpm for 30 minutes at 4°C, washed once with 70% ethanol, and re-hydrated in 20 µl ultrapure water. To serve as a control, the initial ‘input’ cell pellet was re-suspended in 100 µl of 50 µM NaOH, and DNA was liberated by incubating at 95°C 30 min, followed by pH neutralization by the addition of 10 µl of 1M Tris, pH 8. To assess chromosomal DNA association with immunopurified CagY complexes, 1 µl of ‘input’ and ‘IP’ DNA samples were used as the template in standard PCR assays targeting a 795 bp chromosomal DNA amplicon. Quantitation of DNA amplification from ‘input’ and ‘IP’ samples was conducted by densitometry analysis and amplification efficiency of ‘IP’ samples was calculated as a percent of the corresponding ‘input’ sample amplification for each biological replicate experiment. Transfer DNA immunopurification assays were performed a minimum of four times per strain.
Statistical analyses
Data are expressed as mean values ± standard error of the mean, which were calculated from a minimum of three biological replicate experiments. In all graphs, each data point represents an individual measurement, lines represent the mean, and error bars represent the standard error of the mean. Statistical analyses were performed using GraphPad Prism 9 software, with differences of p<0.05 considered statistically significant.
ACKNOWLEDGEMENTS
This work was funded by the NIH (P20 GM130456 and P30 GM110787 to CLS) and academic developments funds provided by the University of Kentucky (to CLS).
Footnotes
Figure 2 revised