Introduction

EPEC are diarrheagenic E. coli strains that cause significant morbidity and mortality in children under two years of age. While infection rates have significantly declined in industrialized nations, EPEC remains a major public health concern in low-income countries (Lozer et al., 2013). EPEC form attaching and effacing (A/E) lesions on intestinal epithelial cells and lack the ability to produce Shiga toxins or heat-labile (LT) and heat-stable (ST) enterotoxins (Croxen et al., 2013; Gomes et al., 2016; Hazen et al., 2016). The ability of EPEC to form A/E lesions is determined by the locus of Enterocyte Effacement (LEE), a large genomic pathogenicity island that encodes the essential genetic elements required for this process (Pearson et al., 2016). The LEE region of EPEC (E2348/69) encodes components of the type III secretion system (T3SS), a molecular apparatus that translocate at least 25 bacterial effector proteins into the host cell. The EPEC type III secretion system (T3SS) consists of a basal body and a needle-like structure, resembling those found in Salmonella and Shigella, but with a distinct sheath-like extension at the needle tip, primarily composed of EspA, and about ten times longer than the T3SS needles from other bacterial species. This EspA filament acts as a molecular bridge, extending from the bacterium to the host cell membrane, allowing insertion of the EspB and EspD translocon components into the host cell membrane, enabling type III effectors injection in the cell cytosol (Creasey et al., 2003; Monjarás Feria et al., 2012; Sal-Man et al., 2012). Osmoprotection assays suggest that the translocon forms a pore with an internal diameter ranging between 3 to 5 nm, allowing the passage of unfolded effector proteins into the host cell (Chatterjee et al., 2015). However, the T3SS translocon shows weak pore-forming activity during EPEC infection of epithelial cells, presumably because it forms a sealed conduct between the T3SS and host cell membranes (Guignot et al., 2016). EspC, a secreted serine protease from the autotransporter family, targets EspA and EspD and down-regulates pore formation activity associated with cytotoxicity (Guignot et al., 2015). EPEC T3SS effector proteins translocated into host cells lead are responsible for attaching and effacing (A/E) lesions and intimate bacterial adhesion to the host cells associated with the formation of an actin-rich pedestal formation (Chen & Frankel, 2005).

The detection of pathogenic bacteria by intestinal epithelial cells plays an important role in initiating pro-inflammatory responses. Recognition of bacterial surface components by pattern recognition receptors triggers pro-inflammatory signaling pathways involving the transcriptional activator NF-κB and the production of cytokines such as interleukin-8 and tumor necrosis factor-α (TNF-α) (Edwards et al., 2011). However, EPEC suppresses these signaling pathways early in infection through the coordinated action of several T3SS effector proteins. Among the first translocated T3SS effectors, Tir interacts with TNF-α receptor-associated factors (TRAF2 and TRAF6), recruiting the tyrosine phosphatases SHP-1 and SHP-2 (Mills et al., 2008; Ruchaud-Sparagano et al., 2011; Yan et al., 2013). Several non-LEE T3SS effectors, including NleE, NleB, NleH1, NleH2, NleC, and NleD, further contribute to inhibiting NF-κB and MAPK signaling. NleE and NleB stabilize the interaction between NFκB and its inhibitory subunit IκB, preventing its degradation, thereby keeping NFκB in an inactive state. The ability of NleE to inhibit NFκB signaling depends on its S-adenosyl-L-methionine (SAM)-dependent methyltransferase activity (Zhang et al., 2011). NleE-mediated methylation of TAB2/3 prevents IKK activation (Zhang et al., 2011). NleB selectively blocks NFκB activation through GlcNAcylation of the TNF-α receptor (TNFR) adaptor protein and TNFR1-associated death domain (TRADD) (S. Li et al., 2013; Pearson et al., 2013). The T3SS effectors NleH1, NleH2 and NleC also interfere with NFκB nuclear translocation (Gao et al., 2009). NleC specifically cleaves P65 RelA (Giogha et al., 2015; Ruchaud-Sparagano et al., 2011; Yen et al., 2010). Interestingly, NleF has been implicated in the activation of NF-κB, underscoring the complexity of the regulation of inflammation during bacterial infection and suggesting the timing of various T3SS effectors’ activity (Pallett et al., 2014).

A similar complexity applies to T3SS effectors regulating cell death and survival pathways during EPEC infection of epithelial cells. Tir was found to elicit a rapid Ca²⁺ influx across the host cell membrane, through the activation of a host plasma membrane Ca²⁺ channel, the mechanosensitive transient receptor potential vanilloid 2 (TRPV2) leading to pyroptosis (Zhong et al., 2022). However, the NleA effector, blocks the delivery of TRPV2 channels to the cell surface, thereby dampening Tir-induced Ca²⁺ influx. The effector NleF also directly binds caspase-4 to inhibit its activity (Zhong et al., 2020). The extrinsic apoptotic pathway is triggered by EPEC pili (Abul-Milh et al., 2001). However, the NleD and NleB effectors inhibit this pathway by cleaving JNK and GlcNAcylating the death domain adaptor proteins TRADD and FADD, respectively (Baruch, Gur-Arie, et al., 2011; Pearson et al., 2013). While EspC prevents cytotoxicity linked to pore-formation by the T3SS translocon during the early EPEC infection phases, it was shown to promote intrinsic apoptosis through increase in intracellular Ca²⁺ and calpain activation (Serapio-Palacios & Navarro-Garcia, 2016).

Central to inflammation and cell death / survival pathways induced by EPEC, bacterial-induced Ca²⁺ signals have been a matter of debate. EPEC infection is known to perturb host Ca²⁺ signaling, but the source and sequence of Ca²⁺ signals during infection remain controversial. EPEC was shown to induce Ca²⁺ influx associated with a loss of mitochondrial membranes permeability leading to cell death (Zhong et al., 2020; Ramachandran et al., 2020; Zhong et al., 2022), but was also reported to trigger IP₃-mediated Ca²⁺ release possibly involved in bacterial-induced cytoskeletal rearrangements (Baldwin et al., 1991; Baldwin et al., 1993; Foubister et al., 1994; Bain et al., 1998).

Here, we investigated the characteristics and implications of EPEC-induced Ca²⁺ responses in epithelial cells. We characterized yet undescribed Ca²⁺ signals induced by EPEC and low ATP levels, presenting the fast dynamics and small amplitude of local Ca²⁺ responses but involving large cell area. We found that these responses likely result from the coordination of elementary responses via rapid Ca²⁺-induced Ca²⁺ release over large cell area, challenging generally admitted concepts on Ca²⁺ diffusion. Importantly, we show that these newly described responses have functional implications by dampening the cell ability to respond to inflammatory signals.

Results

EPEC induces Ca^²⁺ responses that depend on Type III secretion-mediated eATP release

Despite their critical role in cellular processes key to bacterial infection, EPEC-induced Ca²⁺ responses remain to be characterized. We therefore set up to perform a detailed single cell imaging of Ca²⁺ responses elicited by cells infected by EPEC.

As shown in Fig. 1, EPEC induced Ca²⁺ transients often corresponding to a single peak of varying amplitude detected over several minutes, with a median corresponding to 6.2 ± 0.8% of the maximal histamine response (Figs. 1A, B). These Ca²⁺ responses were dependent on a functional T3SS, since they were not observed for the T3SS-deficient escN mutant (Figs. 1A, C). Only 40 ± 4.7 % of cells, however, elicited responses when challenged wild-type EPEC at a low multiplicity of infection (MOI) of 20 bacteria per cell, a value that raised to 76 ± 4.8 % when using a high MOI of 80 bacteria per cell (Fig. 1C). In contrast, even at the low MOI, more than 83 % of cells showed actin pedestals, indicating that Ca²⁺ responses were elicited only in a fraction of cells targeted by EPEC-type T3SS (Figs. 1D, E). The frequency of Ca²⁺ responses per cell increased over the incubation time with an average frequency of responses per cell raising by 6.9 to 8.3-fold from the first to the last 30 min of EPEC challenge (Fig. 1F). This increased frequency suggested the accumulation of an agonist in the extracellular medium during the course of the infection triggering IP₃-mediated Ca²⁺ release. When pooling all single cell responses, we could observe a steady increase in the average cytosolic Ca²⁺ concentration of the cell population, as previously reported (Ramachandran et al., 2020; Figs. S1B). However, when performing single cell imaging, we did not observe an increase in cytosolic Ca²⁺ basal levels even at high MOI after 2 hours incubation with wild-type bacteria (Fig. S1A).

Figure 1.

Together, these results suggest that EPEC induces isolated Ca²⁺ responses that depend on the T3SS for the release of limiting amounts of a Ca²⁺ agonist.

EPEC-mediated Ca²⁺ responses depend on ATP released in the extracellular medium via the T3SS translocon

In previous works, we showed that the EPEC T3SS translocon forms pores in host cell plasma membranes that were down-regulated by the bacterial secreted serine protease EspC (Guignot et al., 2016). We posit that low amounts of ATP released in the extracellular medium by the T3SS translocon were responsible for the isolated Ca²⁺ responses of reduced amplitude elicited by EPEC. According to this view, by removing T3SS translocons from host cell membranes, EspC would down-regulate EPEC-mediated Ca²⁺ signaling explaining the low ratio of Ca²⁺ responding cells relative to cells forming actin pedestals.

Consistent with this and as shown in Figs. 2A-C, an espC mutant induced more Ca²⁺ responses than wild-type EPEC, with 94 ± 3% responding cells and a frequency of 10.5 ± 1.2 responses per cell over the 60 min analysis, compared 40 ± 4.7% responding cells and less than 2 responses per cell for the espC mutant and wild-type EPEC, respectively. Also, the average amplitude of Ca²⁺ responses induced by the espC mutant was higher than that of wild-type EPEC, suggesting more eATP release (Fig. 1B). Accordingly, cell treatment with Suramin, an inhibitor of purinergic receptors, abolished Ca²⁺ responses induced by the wild-type and espC mutant strains (Figs. 2B, C). As expected for ATP-mediated Ca²⁺ release, sample treatment with EGTA, a cell impermeant chelator of extracellular Ca²⁺ did not decrease the percent of Ca²⁺ responding cells triggered by wild-type EPEC or the espC mutant (Fig. S2A). The frequency of responses per cell was also not inhibited and even appear to increase upon EGTA-treatment in cells challenged with wild-type EPEC and the espC mutant (Figs. S2B, C). In control experiments, Suramin treatment did not affect actin pedestal structures induced by these strains (Fig. S3).

Figure 2.

Together, these results suggest that Ca²⁺ responses induced by EPEC are mediated by ATP released in the extracellular medium via pores formed by the T3SS translocon and are down-regulated by EspC.

EPEC induces coordinated Ca²⁺ responses from single IP₃R clusters

We previously showed that Shigella induced local Ca²⁺ responses dependent on the T3SS and Ca²⁺ release (Tran Van Nhieu et al., 2013), suggesting that insertion of the Type III translocon was responsible bacterial-induced local Ca²⁺ signals. We therefore set up to investigate whether EPEC could also trigger T3SS-dependent local Ca²⁺ responses.

To explore this, we performed rapid Ca²⁺ imaging at a frequency of 57 ms acquisition per frame to sample elementary Ca²⁺ release events. As shown in Fig. 3, by performing high speed Ca²⁺ imaging, we detected fast Ca²⁺ increases associated with cell challenge with EPEC. These fast Ca²⁺ responses did not correspond to other responses previously reported since they involved the whole cell or a large cell area but show a small amplitude and fast dynamics usually associated with local Ca²⁺ responses (Fig. 3B). As observed for the ATP-dependent responses shown in Fig. 2, the espC mutant triggered a higher percent of Ca²⁺ responding cells than wild-type EPEC (Fig. 3C). Also, the response amplitude was higher for the espC mutant with an average amplitude corresponding to 7.7 ± 0.4 % (mean ± SEM) of the maximal agonist response, compared to 5.4 ± 0.4 % (mean ± SEM) for wild-type EPEC (Fig. 3E). These responses occurred repeatedly at a high frequency of up to 4.5 responses per minute during several minutes following bacterial challenge (Figs. 3D and 3F) and were dependent on Type III Secretion, as evidenced by the lack of response in cells infected with the ΔescN strain (Figs. 3B and 3C). EPEC-induced fast Ca²⁺ responses were dependent on Ca²⁺ release since they were inhibited by U73122, a PLC inhibitor (Figs. 3C and 3D). These similarities with the EPEC-induced eATP-dependent Ca²⁺ responses suggested that the atypical fast responses were triggered by low amounts of ATP released in the extracellular medium following insertion in host cell plasma membranes of discrete numbers of EPEC T3SS translocons. Consistently, these atypical EPEC-induced fast responses were abolished in the presence of the ATP receptor inhibitor Suramin (Figs. 3C, D).

Figure 3.

Together, these results show that at the onset of infection, EPEC induces an atypical pattern of fast Ca²⁺ responses involving the whole or large area of the cell, likely resulting from low ATP levels released by the insertion of a discrete number of translocons in host cell membranes.

EPEC-induced fast Ca²⁺ responses are triggered by low ATP levels

Previous studies have described local Ca²⁺ increases triggered by sub-maximal agonist concentrations and leading to limited IP₃-mediated Ca²⁺ release. These local Ca²⁺ responses are typically small, of short durations and localized to subcellular regions. Among these, the so-called “Blips” correspond to elementary events of opening of a single IP₃ receptor channel usually lasting between 50 and 100 ms, whereas “Puffs” involve the synchronized activation of multiple IP₃ receptor channels in localized clusters and last several hundreds of ms (Swillens et al., 1999). In contrast to these described local Ca²⁺ signals, EPEC-induced fast and small responses could occur throughout the cell, suggesting the coordination of elementary responses over large cell area. Since our findings suggested that these responses were elicited by minute amounts of eATP released by a discrete number of T3SS translocons, we investigated whether low ATP levels could elicit similar responses.

HeLa cells treated with 150 nM ATP showed Ca²⁺ responses that were indistinguishable from fast responses elicited by EPEC, with an average percent of Ca²⁺ responding cells of 61.2 ± 5.8 (median ± SEM) and a frequency of 3.9 responses per cell over 60 seconds (Figs. 4A, C and S4A, B). These fast Ca²⁺ responses had an amplitude that did not exceed 10 % of the maximal agonist response, and occurred over several minutes (Fig. 4A). As observed for EPEC, fast Ca²⁺ responses induced by low ATP levels involved the whole cell or large cell area encompassing the nuclear and perinuclear area and corresponding to at least 30 % of the cell area as illustrated in Fig. 4B. In this large area, all ROIs corresponding to 1 square micron showed a superimposable profile, as illustrated by traces in Fig. 4C. More detailed scrutinizing showed that in their initial mounting phase, these fast Ca²⁺ responses were created by the opening of discrete clusters involving an area of ca. 0.04 μm² that had a transient activity, or possibly were highly mobile, since they were seldom detected at a similar location for three consecutive 22 ms acquisition frames (Fig. S5). These discrete clusters showed similar Ca²⁺ kinetics suggesting the coordination of Ca²⁺ release of single IP₃R clusters throughout the area that we will hereafter termed CCRICs for “Coordinated Ca²⁺ Responses from IP₃R Clusters”. Treatment with BAPTA-AM to chelate intracellular Ca²⁺ led to a complete inhibition of CCRICs (N = 3, n > 150 cells; Fig. S4A). Ca²⁺ responses could still be detected upon cell treatment with EGTA-AM consistent with its lower k_on rate for Ca²⁺, but with a significant inhibition of the percentage of Ca²⁺ responding cells as well as of the frequency of responses per cell, suggesting that coordination could occur via Ca²⁺ diffusion and Ca²⁺-induced Ca²⁺ release (Fig. S4A-C). Local responses elicited by these ATP concentrations could also be detected often occurring at the cell periphery (Figs. 4B, red region and 4C, red arrow). While the CCRICs area and the peripheral puff area showed spatially segregated Ca²⁺ responses at the onset of the elicitation (Fig. 4C, peaks 1, 3 and 2, 4), this spatial segregation appeared to blur over the duration of the experiment with peaks integrating responses from the whole cell (Fig. 4C, green and red dashed lanes).

Fig. 4:

CCRICs are coordinated by the rapid diffusion of Ca²⁺ at low concentrations in cell area with a high density of IP₃ clusters

In Fig. 5, we used modeling to further investigate the mechanism of coordination of these fast Ca²⁺ responses. Based on our previous studies (Voorsluijs et al., 2019; Ornelas-Guevara et al., 2023), the model provides a fully stochastic spatial description of Ca²⁺ release dynamics from IP₃R clusters in a two-dimensional representation of a HeLa cell. The simulation domain extends on 10 × 10 µm² and is discretized into a 20 × 20 grid of compartments (0.5 × 0.5 µm² each), each representing a cytosolic subvolume of 10^-16 L. Each compartment contains at most one cluster of IP₃R, whose dynamics is described as a whole. Besides, cytosolic Ca²⁺ concentration can also vary because of uptake by SERCA, release by a leak or diffusion. A full description of the model can be found in the Supplementary Methods.

Fig. 5.

In Fig. 5A, low ATP levels lead to low IP₃ levels activating a limited number of IP₃ clusters, opening stochastically and releasing small amounts of Ca²⁺. In a given area of the cell, the Ca²⁺ variation integrates Ca²⁺ release from clusters within this area, as well as Ca²⁺ diffusing from or to other cell area. For a given response, the initially firing cluster is contained in an area characterized by the highest Ca²⁺ peak amplitude (Fig. 5A, blue and green arrows) that dampens in a distal area (Fig. 5A, blue and green arrowheads). If the density of IP₃R clusters is low, as expected for ER compartment at the cell periphery, the spatial segregation of the initial firing cluster and resulting Ca²⁺ diffusion to other area is clearly detected (Fig. 5A). In instances, however, the model predicts temporally coordinated responses of similar amplitude, suggesting Ca²⁺-induced Ca²⁺ release from secondary clusters (Fig. 5A, black arrow). For both type of Ca²⁺ dynamics, a large value of the Ca²⁺ diffusion coefficient of 100 μm² / s is a key parameter that needs to be taken into account in the model. While it is generally admitted that Ca²⁺ diffuses very slowly due to the Ca²⁺ buffers in the cell (∼30 μm²/s), the low levels of Ca²⁺ released by CCRICs may not be subjected to the diffusion limitations observed at higher Ca²⁺ levels, because of the relative moderate affinity of buffers for Ca²⁺.

If the IP₃R cluster density is high, as expected in the large perinuclear and nuclear area corresponding to the bulk of the ER, the coordination between individual clusters is very fast (Fig. 5B). As a result, the identification of initial firing clusters goes beyond the technical capacities of the imaging set-up, and ROI within this area show comparable profiles of fast Ca²⁺ responses (Fig. 5B). Upon prolonged incubation with increasing Ca²⁺ responses and sensitization of IP₃R clusters, the coordination of responses linked to Ca²⁺-induced Ca²⁺ release becomes predominant throughout the cell (Fig. 5C).

These findings indicate that the novel Ca²⁺ response pattern is not exclusive to EPEC infection and can be replicated by low levels of eATP, suggesting a broader physiological relevance. From an immunological perspective, CCRICs may therefore play a critical role in various signaling pathways during bacterial infection.

Low eATP levels dampen NF-κB activation

eATP is a well-characterized danger signal contributing to the elicitation of pro-inflammatory signals in various tissues in response to infections (Savio et al., 2018). Previous studies linked intracellular Ca²⁺ signaling and NF-κB activation, a key transcription factor that triggers inflammatory responses (Smedler et al., 2014). We therefore set up to investigate the effects of CCRICs triggered by low eATP levels on NF-κB activation, by performing Western blot analysis against the phosphorylated forms of IκBα (p-IκBα) and P65 (p-P65).

As shown in Figs. 6A and 6B, in control samples, TNF-α induced IκB-α phosphorylation peaking 10 minutes following challenge. In contrast, in the presence of low ATP levels, IκB-α showed a delayed phosphorylation with a 2.1 fold decrease at 10 minutes post-challenge (Figs. 6A and 6B). Consistently, the rates of IκB-α degradation were also slower in the presence of ATP relative to control (Figs. 6A and 6C). As expected from the IκB-α results, TNF-α induced the phosphorylation of the NF-κB P65 subunit and ATP led to a delay and decrease in P65 phosphorylation (Figs. 6D and 6E).

Fig. 6.

In control experiments, we did not detect differences in P65 phosphorylation in response to TNF-α stimulation when cells were treated with BAPTA-AM to chelate intracellular Ca²⁺ (Figs. 6F and 6H). However, cell treatment with BAPTA-AM prevented the dampening of P65 phosphorylation triggered by low ATP levels (Figs. 6G and 6H), suggesting that CCRICs down-regulated TNF-α -induced NF-κB activation.

Low eATP levels down-regulates NF-κB activation through Ca²⁺-dependent O-GlcAcylation

We next investigated how CCRICs could regulate NF-κB activation. O-linked β-N-acetylglucosamine (O-GlcNAc) transferase (OGT) was reported to regulate NF-κB signaling by post-translationally modifying the p65 subunit (Ruan et al., 2017). Interestingly, OGT is regulated by Ca²⁺ signaling, suggesting that CCRICs could affect NF-κB activation via O-GlcNAcylation.

As shown in Fig. S6, TNF-α in the presence of 150 nM eATP stimulated the levels of O-GlcNacylation, specifically for proteins with an apparent molecular weight superior to 100 kDa, that was not observed with TNF-α alone. To further investigate the effects of low eATP levels on NF-κB O-GlcNacylation, we performed immunoprecipitation of P65 RelA on lysates of cells stimulated for 12 minutes with TNF-α alone or co-stimulated with TNF-α and 150 nM eATP. As shown in Fig. 7, TNF-α induced increased O-GlcNacylation of P65 relative to non-stimulated cells, but this increase was inhibited by low eATP levels. Inhibition of P65 O-GlcNacylation by eATP was Ca²⁺ dependent, since it was not observed in the presence of BAPTA-AM (Fig. 7).

Fig. 7.

The results indicate that low eATP levels inhibit O-GlcNacylation of P65 induced by TNF-α in a Ca²⁺-dependent manner and suggest that differential O-GlcNacylation of P65 relative to higher molecular weight proteins.

Discussion

We report here the characterization of CCRICs, corresponding to yet undescribed Ca²⁺ responses showing the rapid kinetics and small amplitude of local Ca²⁺ responses but coordinated over large cell area. We found that CCRICs implicate the rapid coordination of IP₃R clusters in large cell area, challenging established concepts in the Ca²⁺ signaling field.

Ca²⁺ diffusion in the cytosol is regulated by mobile and immobile Ca²⁺-binding proteins acting as buffers and forming localized microdomains with steep Ca²⁺ concentration gradients. At the mouth of an open IP₃R channel, Ca²⁺ concentration can reach 100 μM, while just 1–2 μm away, it may drop below 1 μM. Diffusion is restricted by Ca²⁺ buffers with K_D’s that are generally lower than this concentration. Consequently, Ca²⁺ signaling is generally admitted to be spatially restricted, typically influencing regions within approximately 5 μm of the release site (Foskett et al., 2007). The distribution of Ca²⁺-binding proteins and the spatial arrangement of release channels allow IP₃R-mediated [Ca²⁺]_i signals to exhibit diverse spatial and temporal properties, making this system highly adaptable (Vandeput et al., 2007). High-resolution optical imaging of fluorescent Ca²⁺ indicators in intact cells indicates that IP₃-mediated [Ca²⁺]_i signals are structured levels (Foskett et al., 2007). At low IP₃ levels, individual IP₃R opens stochastically at discrete release sites, causing localized elevations in cytoplasmic [Ca²⁺]. At higher IP₃ levels, Ca²⁺ release spreads between IP₃R clusters, propagating waves that travel at tens of microns per second, coordinating intracellular signaling ultimately leading to a global Ca²⁺ response. In reference models describing intracellular Ca²⁺ dynamics, cell regions with a high IP₃R density initiate the Ca²⁺ response from which Ca²⁺ waves propagate with a diffusion coefficient of 10 – 30 μm²/s (Falcke et al., 2003).

In contrast to these described global and local Ca²⁺ responses, we found CCRICs to be highly temporally coordinated over large area, suggesting the fast diffusion of Ca²⁺ and propagation of Ca²⁺ by Ca²⁺-induced Ca²⁺ release. While challenging generally admitted concepts on the poor diffusion of Ca²⁺, this view is fully supported by our theoretical modeling implicating the fast diffusion of Ca²⁺, with a diffusion coefficient of at least 100 μm²/s that can be expected at low cytoplasmic [Ca²⁺]. Indeed, at these low Ca²⁺ concentrations not exceeding a few hundreds nM, the majority of Ca²⁺ buffers are not expected to efficiently bind to Ca²⁺ and to significantly interfere with Ca²⁺ diffusion because of their relative low affinity.

We found that CCRICs implicate a large cell area including the nuclear and perinuclar area, while local Ca²⁺ responses were detected at small area of the cell periphery. The large CCRIC area likely involves the bulk of the endoplasmic reticulum, while peripheral area may contain smaller ER compartments. In our model considering fast Ca2+ diffusion when buffers are far from saturation, the higher density of IP₃R clusters in the CCRIC area accounts for the high coordination of the responses, relative to the lack of coordination in peripheral area where lower IP₃R density is expected.

We showed that a low dose of eATP triggering CCRICs delayed and dampened NF-κB activation linked to a reduction in O-GlcNAcylation of the NF-κB p65 subunit. One major open question is the mechanism by which CCRICS down-regulate NF-κB activation. NF-κB activity can be modulated by O-GlcNAcylation, a reversible glycosylation modification catalyzed by O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA) (Liu & Ramakrishnan, 2021). Previous studies demonstrated that Ca²⁺ signals activate Ca²⁺-regulated enzymes like CaMKII, which in turn phosphorylates and activate OGT, promotes O-GlcNAcylation (Ruan et al., 2017). In other studies, OGT-mediated O-GlcNAcylation could modulate NF-κB signaling pathway (X. Dong et al., 2023). O-GlcNAcylation of P65 could also inhibit its interaction with IκB-α, promote p65 nuclear translocation and increase NF-κB transcriptional activity (Liu & Ramakrishnan, 2021). Reduced O-GlcNAcylation of P65 at residues S550 and S551 was shown to result in decreased NF-κB activation and nuclear translocation (Motolani et al., 2023). These findings are in line with our results suggesting that CCRICs elicited by low-level eATP, downregulate p65 O-GlcNAcylation and NF-κB activation, possibly by modulating OGT activity or OGT-p65 interactions. Reduced O-GlcNAcylation of p65 may affect its phosphorylation patterns indirectly, perhaps by altering the interaction of NF-κB with kinases or phosphatases involved in its activation (Özcan et al., 2010).

Our findings indicate that eATP differentially regulate inflammatory signaling pathways in epithelial cells by dampening NF-κB activation at low levels and stimulating its activation at high concentrations. These results extend the key role of eATP from a danger-associated molecular pattern (DAMP) to a fine-tuner of inflammatory responses depending on its concentration.

Materials and Methods

Cell and Bacterial culture

HeLa cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM; Gibco, Thermo Fisher Scientific) supplemented with 10% fetal bovine serum (FBS; Gibco, Thermo Fisher Scientific) at 37 °C in a humidified incubator with 10% CO₂. Wild-type Enteropathogenic Escherichia coli (EPEC WT), ΔescN, and ΔespC strains were cultured in Luria-Bertani (LB) broth at 37 °C with kanamycin at a final concentration of 15 μg/ml in a shaking incubator. All strains were transformed with the pmCherry-N1 plasmid, which encodes red fluorescent protein and carries an ampicillin resistance gene for selection. Where applicable, ampicillin (100 μg/mL) and kanamycin were included to maintain resistance markers. The red fluorescence enabled visualization of the bacteria during downstream analyses.

EPEC infection of HeLa cells

HeLa cells were seeded in 6-well plates at a density of 4.5 × 10⁵ cells per well one day before infection. EPEC WT, ΔescN, and ΔespC strains grown in the exponential phase were resuspended and primed for 5 hours before challenging HeLa cells in DMEM medium. Cells were challenged with bacteria at an OD600 = 0.2 (Low MOI) or 0.8 (High MOI).

Ca²⁺ imaging

HeLa cells were seeded onto 25 mm-diameter glass coverslips. Cells were preloaded with the fluorescent indicator dye Cal-520 (AAT #21130) for 30 minutes at room temperature, followed by two PBS washes and one time with DMEM. And placed coverslips, imaging was performed in an observation chamber in DMEM without phenol red, supplemented with 25 mM HEPES. Following a 3-minute baseline acquisition, add the required bacteria strains and OD600 into the chamber. Following a 10-minute incubation at room temperature to allow bacterial attachment. Imaging was then carried out at 35°C using a Nikon Eclipse TE200 inverted fluorescence microscope with a 60× objective for 1h of infection. Fluorescence signals were acquired 485 nm with excitation and 535 nm emission parameters. Image control and data acquisition were managed by Simple32 software (Compix Inc.), depending on the experiment requirement, imaging at 22ms, 57ms or 10 sec acquisition intervals. 3 μM of Histamine and 2 μM of ionomycin were applied to check the ability of cells to show calcium response. Images were captured using a CMOS camera (Hamamatsu) and analyzed using the same software.

Immunofluorescence analyses

Cells were washed three times with PBS and fixed with 3.7% paraformaldehyde (PFA), permeabilized with 0.1% Triton X-100 for 5 minutes and washed with PBS. Blocking was performed using 3% FBS in PBS for 30 minutes at room temperature. Cells were incubated with primary antibodies for 2 hours: mouse anti-phalloidin (1:200; Fischer Scientific, #17511176). Secondary antibodies (anti-mouse), along with DAPI were applied for 1 hour. Samples were mounted using Dako mounting medium (Agilent) and imaged using a Nikon Ti2 confocal microscope equipped with a 60× objective and Nikon acquisition software.

Modelling

We develop a fully stochastic spatial model to simulate Ca²⁺ release dynamics from IP₃R clusters in a two-dimensional representation of a HeLa cell. The simulation domain measures 10 × 10 µm² and is discretized into a 20 × 20 grid of compartments (0.5 × 0.5 µm² each), each representing a cytosolic subvolume of 10^-16 L. Ca²⁺ exchange between the Endoplasmic Reticulum (ER) and the cytosol occurs through IP₃R-mediated release, SERCA uptake, and ER Ca²⁺ leak, following the framework by Voorsluijs et al., 2019 and Ornelas-Guevara et al., 2023. Ca²⁺ diffusion is implemented stochastically as a kinetic process between adjacent compartments (Kraus et al., 1996), with a diffusion coefficient of 100 µm²/s to reflect moderate endogenous buffering. Each IP₃R cluster functions as a single unit with four possible states: Open (O), Closed (C) and two Inhibited states (i1, and i2), transitioning in response to local [Ca²⁺] and [IP₃]. This phenomenological description captures key characteristics of Ca²⁺ puffs and their transition to global signals via Ca²⁺ diffusion and Calcium Induced Calcium Release.

We perform all simulations using the Gillespie algorithm, where each event, reaction or diffusion, is selected stochastically based on its propensity. See Supplementary Methods for complete model equations and parameter values.

Western Blot analysis

To obtain total cell extracts, cells were lysed in sample buffer 1x (62.5 mM Tris pH=8, 2% SDS, 10% glycerol, 0.05% bromophenol blue, 5% β-mercaptoethanol) and boiled at 95°C for 5 minutes. Proteins from total lysates were separated by SDS PAGE and transferred to nitrocellulose membrane (0.45μM AmershamTM ProtranTM). Western Blot analysis was performed according to standard procedure using the following primary antibodies: IkappaBalpha (OZYME, 9424S), Phospho-IκB-α (OZYME, 9424S), NF-kappaB p65 (OZYME, 9246S), Phospho-NF-kappaB p65 (OZYME, 3033S), HSP90 (Santa Cruz Biotechnologies sc-13119). HRP-conjugated anti-mouse (Cytiva) and anti-rabbit (Sigma) were used as secondary antibodies.

Immunoprecipitation assays

HeLa cells were seeded in 150 cm² dishes (7.4 × 10⁶ cells/dish) and cultured in DMEM supplemented with 10% FBS. Cells were pretreated with 20 µM BAPTA for 30 minutes at room temperature where indicated, then stimulated with 10 ng/mL TNF-α alone or in combination with 150 nM ATP for an additional 12 minutes. After treatment, cells were washed with ice-cold PBS containing 1 mM NaF and 1 mM Na₃VO₄, lysed in ice-cold lysis buffer (50 mM Tris-HCl pH 7.5, 0.5% Triton X-100, 100 mM NaCl, 1 mM DTT, protease inhibitor cocktail without EDTA), and incubated on a rotating wheel at 4°C for 1 hour. Lysates were clarified by centrifugation at 13,000 rpm for 30 minutes at 4°C. Supernatants were incubated with 5 µL of anti-NF-kB p65 antibody (Abcam, ab16502) for 2 hours at 4°C with rotation, followed by overnight incubation with pre-equilibrated protein A/G beads. Immunocomplexes were washed once with lysis buffer and twice with PBS, then eluted in Laemmli buffer by boiling at 100°C for 5 minutes. Input and IP samples were analyzed by Western blotting using antibodies against O-GlcNAc (Abcam, ab2739), anti-NF-kB p65 (OZYME, 9246S), and Actin (OZYME, 4967).

Statistical Analysis

All quantitative data are presented as mean ± SEM from at least three independent experiments. Statistical significance was assessed using unpaired two-tailed Student’s t-tests with unequal variance, unless otherwise specified. GraphPad Prism 7 (GraphPad Software) was used for statistical analysis, and p-values < 0.05 were considered statistically significant.

Supplementary Methods

Following previous modeling studies of Ca²⁺ dynamics (Voorsluijs et al., 2019; Ornelas-Guevara et al. 2023), we implement a fully stochastic model of intracellular Ca²⁺ dynamics using the Gillespie algorithm. The model captures IP₃-induced Ca²⁺ release from IP₃R clusters, SERCA-mediated reuptake, ER Ca²⁺ leak, and cytosolic Ca²⁺ diffusion. The simulated HeLa cell is represented as a 10 × 10 µm² square, discretized into a 20 × 20 grid. Each compartment (0.5 × 0.5 µm²) corresponds to a cytosolic volume of 10^-16 L.

Each IP₃R cluster is represented as a single unit with four discrete states: Open (O), Closed (C), and two inhibited states (i₁, i₂). The transition from C to O depend on both [Ca²⁺] and [IP₃], while the transition from O to i1 depends on [Ca²⁺], following the model described in Ornelas-Guevara et al., 2023 (see Figure S1). Various spatial distributions of the IP₃R clusters are investigated in Figure 5A to 5C, where the locations of the clusters are indicated by white boxes. SERCA activity and ER Ca²⁺ leak are present in all compartments.

Supplementary figures

Figure S1.

Figure S2.

Figure S3.

Figure S4.

Figure S5.

Figure S6.

Acknowledgements

This work was funded by the Inserm, CNRS and ANR programs CALPLYCX and VITAL to GTVN. FG was funded by a Chinese Science Council PhD fellowship. ROG was supported by Wallonie-Bruxelles International (Excellence Grant 2025). This work was supported by a PDR FRS-FNRS project (T.0073.21). GD is Research Director at the Belgian “Fonds National pour la Recherche Scientifique” (FRS-FNRS).

Additional information

Author contribution

FR, LC and GTVN designed, performed experiments and wrote the mansucript. LO performed experiments. ROG and GD performed the mathematical modeling.