TLR-Activated Gap Junction Channels Protect Mice against Bacterial Infection through Extracellular UDP Release

Microbial-derived pathogen-associated molecular patterns (PAMPs), such as LPS and Pam₃CSK₄, are some of the strongest triggers of immune responses. However, numerous types of stimuli including mechanical trauma, ischemia, stress, and environmental cues also trigger an innate immune response through releasing pathogen-free mediators, which are collectively termed as damage-associated molecular patterns (DAMPs) (1). Meanwhile, as a kind of special stress to immune cells, the invading pathogen could also induce the release of DAMPs. In the past decades, most studies have focused on PAMP-induced secretion of cytokines, chemokines, and IFNs, whereas the internal relationship between DAMPs and PAMPs in innate immune responses has not been well investigated (2). Therefore, it is of interest to explore the function and mechanism of DAMPs release in infectious diseases.

Central among the most widely studied DAMPs are HMGB1, heat shock proteins, and extracellular nucleotides such as ATP and UDP. As the first described purinergic transmission agent, ATP has been well investigated in a series of physiological functions, including different infectious diseases (3–5). However, the study of UDP/P2Y₆ became increasingly popular only after it was described as an immune mediator of microglial phagocytosis in 2007 (6). Since then, more and more evidence has shown the important role of UDP/P2Y₆ signaling in the regulation of immune responses in bacterial infection (7–9) and even in atherosclerotic lesion development (10). Moreover, the apoptotic cells could release UDP as find-me signals through pannexin hemichannels that recruit motile phagocytes, leading to the prompt clearance of dying cells (11, 12). However, whether and how UDP was released in bacterial infection still need to be further explored.

It has been shown that injured or infected cells can release nucleotides through exocytosis, blebbing or passage via a plasma membrane channel. Among them, gap junction channels (GJCs) play important roles in cell–cell communications through the direct transfer of ions, second messengers, and other molecules including Ag peptides (13, 14). Generally, each GJC contains a serial docking of two hemichannels, which are composed of six protein subunits called connexins and pannexins (15). Although connexin 43 was first uncovered in heart muscle cells and plays important roles in heart development (16, 17), connexin 43 is expressed ubiquitously, in contrast to most connexin isoforms, which are restrictively expressed and their malfunction results in disorders such as deafness, skin diseases, fertility problems, and lens cataracts (18). Multiple immune cells have also been shown to express connexins. Among them, dendritic cells and monocytes/macrophages express connexin and can form functional gap junctions between identical as well as different cells (19, 20). Most notably, connexin can be upregulated when the immune cells become exposed to inflammatory factors, but the role of them in purinergic signaling–mediated immune regulation is poorly defined (21). Our previous study has shown that virus-infected cells release UDP through pannexin hemichannels that facilitate IFN-β production (22). Also, we have reported the protective role of UDP and P2Y₆ in bacterial infection, but whether and how UDP is released in bacterial infection are still unknown.

In this study, we demonstrated that UDP is released from macrophages mainly through connexin-mediated GJCs during bacterial infection. Then, extracellular UDP (eUDP) plays a nonredundant role in promoting host survival from Escherichia coli–induced peritonitis through activating P2Y₆ and facilitating MCP-1 production. Furthermore, we also showed that TLR-activated ERK signaling is not only involved in proinflammatory cytokine production but also can regulate innate immune responses via increasing UDP release. Taken together, our findings demonstrated a potential therapeutic significance of TLR-activated GJCs in fighting against bacterial infection through UDP release.

For peritoneal macrophages and bone marrow–derived macrophages (BMMs) isolation as well as the peritonitis mouse model, female 6- to 8-wk-old C57BL/6 mice were purchased from the Shanghai Laboratory Animal Company (Shanghai, China). The P2Y₆ knockout mice were prepared as described previously (22). All mice used in these experiments were housed under pathogen-free conditions and were maintained in accordance with institutional guidelines. All experimental protocols were approved by the Animal Investigation Committee of East China Normal University.

BMMs and peritoneal macrophages were obtained as described previously (23). The mouse macrophage cell line RAW 264.7 was obtained from the American Type Culture Collection (Manassas, VA). BMMs and RAW 264.7 were maintained in DMEM containing 10% FBS, 100 U/ml penicillin, and 100 mg/ml streptomycin, at 5% CO₂ and 95% humidity, with drugs added at the times indicated.

MTS assay was performed using MTS reagent, according to the manufacturer’s instructions. Briefly, cells were seeded at a density of 1 × 10⁴ cells/well in 96-well plates and allowed to grow overnight. Then, cells were treated as indicated. After 24-h incubation with LPS (Sigma-Aldrich, St. Louis, MO) and Atractylenolide I (AOI; Sigma-Aldrich), 20 μl MTS was added to each well and incubated for 2 h at 37°C, and then, the absorbance was measured at 490-nm wavelength using a microplate reader. Cell viability of the no treatment group was normalized to 100%.

RAW 264.7 cells and BMMs were seeded into 24-well plates (Corning Costar, Corning, NY) at 6 × 10⁴ cells/well overnight and then changed to phenol red-free DMEM. Subsequently, cells were infected with LPS for the indicated amount of time after treating with or without inhibitors. As recommended by the manufacturer of Tran screener UDP² FP Assay Kit (Bellbrook Labs), 15 μl supernatant was added into 384-well plate for the fluorescence polarization assay.

BMMs and RAW 264.7 cells were stimulated with different concentrations of LPS or inhibitors for 1 h, and total RNA was isolated by applying TRIzol reagent (Invitrogen), according to the manufacturer’s protocol. cDNA was synthesized with 500 ng RNA using a reverse transcription kit (Prime Script First Strand cDNA Synthesis kit (TAKARA, Dalian, China), according to the manufacturer’s instructions. One microliter of template from 10-fold diluted cDNA was subjected to quantification of cytokine expression using iQ SYBR Green Supermix (Bio-Rad, Hercules, CA), and the data were analyzed by the ECO Real Time PCR System (Illumina, San Diego, CA). The sequence-specific primers are shown in Tables I and II.

For ELISA assay, RAW 264.7 cells or peritoneal macrophages were seeded into 12-well plates at 1 × 10⁵/well and incubated overnight. Cells were preincubated with or without 10 μM MRS2578 for 30 min and then stimulated with 100 μM UDP or LPS for 12 h, respectively. The concentration of MCP-1 in the supernatant was measured using the mouse MCP-1 ELISA Set, as recommended by the manufacturer (BioLegend).

RAW 264.7 were seeded in 6-well plates (Corning Costar) and stimulated with LPS at the times indicated and at different dose. The concentration of protein was measured by bicinchoninic acid assay (Pierce) and equalized to the same concentration with the extraction reagent. Samples were separated by 12% SDS-PAGE and transferred to polyvinylidene fluoride membranes (Bio-Rad, Hercules, CA). After incubation with phospho-ERK1/2, ERK1/2, phospho-AKT, AKT, β-actin, phospho-p65, and p65 Ab (Cell Signaling Technology, Danvers, MA), the polyvinylidene fluoride membranes were incubated with appropriate HRP-conjugated secondary Abs (Sigma-Aldrich). Finally, the ECL (Pierce, Rockford, IL) method was applied to detect those proteins.

Transfer of a single colony from an E. coli 0111:B4 plate into 2 ml Luria–Bertani (LB) liquid medium at 220 rpm and 37°C for 16 h and the bacteria was diluted 1:100 for another 2 h vigorous shaking to prepare the E. coli in log phase. Then, the E. coli was treated with 0, 100, 500, and 1000 μM UDP, respectively, for 12 h, and then, bacteria were cultured in solid LB medium to get the single clone for counting.

Six- to eight-week-old C57BL/6 female mice were chosen to induce bacteria-mediated peritonitis. Peritoneal bacteria and survival curves were detected to reflect the protective effect of UDP. To count peritoneal fluid E. coli, mice were divided randomly into six groups and pretreated with an i.p. injection of 300 μl PBS, 100 μM UDP (Sigma-Aldrich), 10 μM FFA (Sigma-Aldrich), 10 μM MRS2578 (Sigma-Aldrich), 10 μM U0126 (GeneOperation, Ann Arbor, MI), and 10 μM Gap26 (GL Biochem, Shanghai, China). Twelve hours later, each mouse was challenged with E. coli 0111:B4 through i.p. injection. After 12 h, E. coli were lavaged with 3 ml PBS from each mouse’s abdominal cavity and then diluted 10-fold in PBS, and 20 μl bacterial suspension was cultured in solid LB medium for 12 h. Single CFU were counted to determine peritoneal fluid E. coli. Another six groups were divided and treated as described above, but instead of counting peritoneal fluid E. coli, the mice were checked every 2 h to monitor the survival.

Data are presented as mean ± SEM (n = 3–6). Statistical significance was evaluated with the Student t test or one-way ANOVA, followed by Dunnett’s multiple comparison. A p value < 0.05 was considered significant. For survival curve analysis, the log-rank test was performed, and a p value < 0.05 was considered significant.

To elucidate the potential significance of eUDP in antibacterial immune responses, we measured the UDP release using the Transcreener UDP² FP Assay kit in both E. coli–infected mice and LPS- or Pam₃CSK₄-treated macrophages. As shown in Fig. 1A, a persistent increase of eUDP was observed in the peritoneal cavity of mice challenged with E. coli 0111:B4 for 24 h. To exclude the potential influence of LPS on cell viability, RAW 264.7 cells were treated with gradient concentration of LPS. Little influence on cell viability was observed in LPS-treated cells (Fig. 1B), whereas eUDP was obviously induced by LPS in a dose (Fig. 1C)- and time (Fig. 1D)-dependent manner in RAW 264.7 cells. This kind of UDP release could also be observed in LPS-treated BMMs in a time-dependent manner (Fig. 1E). Similar data can also be seen in Pam₃CSK₄-treated cells (Fig. 1F). Furthermore, LPS-induced UDP release could be rescued obviously by AOI (a selective inhibitor to TLR4) (24), suggesting the important role of TLR signaling in UDP release (Fig. 1G).

Generally, the specific channel inhibitors such as N-ethylmaleimide (NEM, a specific inhibitor to exocytosis), carbenoxolone (CBX, a nonspecific pannexin channel inhibitor), and flufenamic acid (FFA, a specific connexin channel inhibitor) are powerful tools for investigating the mechanism of DAMPs’ release. In contrast to LPS-induced ATP release (25), LPS-induced UDP release was significantly inhibited by both CBX and FFA but not NEM, suggesting the key role of gap junctions in this process. Furthermore, FFA reduced UDP release much more than CBX, implying that LPS-induced UDP release mainly occurs through connexin channels (Fig. 2A).

A series of studies has shown the protective role of UDP in bacterial infection. Therefore, to further confirm the potential role of UDP release in host defense against invading bacteria, we set up a mouse peritonitis model by i.p. injection of E. coli 0111:B4 to monitor the clearance of bacteria and mouse survival. As shown in Fig. 2B, the total quantity of bacteria in the peritoneal cavity was reduced in UDP-injected mice, and this kind of protection could be rescued by the P2Y₆-specific inhibitor MRS2578. In addition, if we treated the mice with FFA to block UDP release, the invading bacteria were also increased. Furthermore, the total quantity of bacteria in the peritoneal cavity was increased in P2Y₆^−/− mice and UDP-induced clearance to bacteria also reduced obviously in P2Y₆^−/− mice (Fig. 2C). Although the proliferation of E. coli 0111 was little changed by high concentration of UDP (Fig. 2D), similar data were also observed in the mouse survival assay; the survival of infected mice could be increased from 30 to 60% if the mice were treated with UDP. In contrast, if we treated the mice with FFA to block UDP release and MRS2578 to inhibit P2Y₆ signaling, the mice were all dead within 31 and 38 h, respectively (Fig. 2E). To further confirm the key role of P2Y₆ in antibacterial immune responses, we infected the wild-type (WT) and P2Y₆^−/− mice with E. coli 0111:B4 to monitor the mouse survival. As shown in Fig. 2F, the survival of P2Y₆^−/− mice was decreased obviously, and UDP-mediated protection also eliminated in P2Y₆^−/− mice (Fig. 2F). These data further confirmed the protective role of endogenous UDP release and P2Y₆ in bacterial infection.

As mentioned before, GJCs are composed of different connexins proteins. Therefore, we checked the expression level of connexin family members in RAW 264.7 cells (Tables I, II). As shown in Fig. 3A, connexin 43 was most abundant in RAW 264.7 cells. Interestingly, the expression of connexin 43 was also obviously increased by LPS in RAW 264.7 (Fig. 3B) and BMMs (Fig. 3C). Furthermore, the expression of connexin 43 could also be increased by LPS in a dose (Fig. 3D)- and time-dependent (Fig. 3E) manner. Similarly, protein levels of connexin 43 were increased significantly by LPS at 12 h (Fig. 3F). These data suggested the potential role of connexin 43 in LPS-mediated immune responses.

To explore the mechanism of LPS-induced connexin 43 expression, we assessed the LPS-induced signaling by Western blotting assay. As shown in Fig. 4A and 4B, the phosphorylation of P65, AKT, and ERK was increased by LPS in a dose- and time-dependent manner. Then, we pretreated the cells with U0126 (inhibitor to MEK1/2), Bay 11-7082 (inhibitor to NF-κB), and Ly294002 (inhibitor to PI3K/Akt) to appraise the influence of those signaling pathways on LPS-induced expression of connexin 43. As shown in Fig. 4C, the LPS-induced RNA expression of connexin 43 was clearly reduced by U0126, which is a specific inhibitor to ERK signaling. In contrast, little change was seen in Bay 11-7082– and Ly294002-treated cells. Furthermore, the expression of connexin 43 at the protein level could also be reduced by U0126, suggesting the key role of ERK signaling in LPS-induced connexin 43 expression (Fig. 4D). In addition, UDP release was also measured in U0126- and Gap26-treated cells. (Fig. 4E) Consistent with the inhibition of connexin 43 expression, UDP release was clearly reduced by the ERK inhibitor U0126 and Gap26 (connexin 43 blocker peptide).

As the initial step in antibacterial immune responses, chemokine-mediated immune cell recruitment is important. As shown in Fig. 5A, the expression of MCP-1 is dramatically increased by UDP in RAW 264.7 cells. Interestingly, UDP- and LPS-induced MCP-1 expression could also be blocked by the P2Y₆-specific antagonist MRS2578, implying that UDP/P2Y₆ signaling is involved in LPS-induced immune responses. Accordingly, the protein level of MCP-1 in LPS-untreated or -treated cells was both increased obviously by UDP, and this kind of activation could be blocked by P2Y₆-specific inhibitor MRS2578 (Fig. 5B), and similar data also confirmed in P2Y₆^−/− peritoneal macrophages (Fig. 5C). Also, we pretreated cells with the plasma membrane channel inhibitors FFA, CBX, and NEM to explore the key role of nucleotide release in LPS-induced MCP-1 production. The LPS-induced MCP-1 production was blocked by FFA, which can also prevent LPS-induced UDP release. In contrast, the LPS-induced MCP-1 production was poorly influenced by CBX and NEM, which are involved in pannexin channels and exocytosis (Fig. 5D). To further confirm the role of UDP/P2Y₆ signaling in LPS-induced MCP-1 production, we treated WT and P2Y₆^−/− peritoneal macrophages with LPS for 12 and 24 h. Similar to the data in RAW 264.7 cells, the RNA (Fig. 5E) and protein (Fig. 5F) level of MCP-1 was increased by LPS in both WT and P2Y₆^−/− peritoneal macrophages, whereas both the basal level and the LPS-induced expression of MCP-1 were much lower in P2Y₆^−/− peritoneal macrophages. Thus, these data further confirmed the key role of P2Y₆ in LPS-induced immune responses.

To investigate the role of connexin 43–mediated UDP release in antibacterial immunity, we pretreated the acute peritonitis mouse model with ERK inhibitors and the connexin 43 blocking peptide. As shown in Fig. 6A, the residual bacteria in the abdominal cavity were significantly increased by pretreating the mice with U0126 and Gap26. Consequently, the survival of infected mice was also decreased dramatically by U0126 and Gap26 (Fig. 6B). Thus, our results suggest a critical role of connexin 43–mediated UDP release in the regulation of immune response and pathogen clearance in bacterial infection (Fig. 6C).

As the first defense against invaded pathogen, innate immune cells recognize PAMPs through pattern recognition receptors to initiate systemic immune responses. Therefore, communication between immune cells is a crucial part in this process. In this paper, we demonstrated a novel, to our knowledge, communication strategy between immune cells that relies on the formation of connexin-based gap junctions and eUDP release. First, we showed a persistent accumulation of eUDP in both the abdominal cavity of infected mice and LPS- or Pam₃CSK₄-treated cell supernatant. Consistent with our previous data, released UDP can protect mice from bacterial infection through increasing MCP-1 production. Accordingly, if we blocked endogenous UDP release by FFA or U0126, the survival of infected mice reduced obviously, suggesting the crucial role of UDP release in antibacterial immune responses.

Extracellular nucleotides were first termed a find-me signal from dying cells to enhance phagocyte recruitment. Then, more and more papers have published damage-induced nucleotide release and its predominant role in different pathological and physiological processes (26–29). In contrast to pathogen-induced inflammation, sterile inflammation was activated by the detection of endogenous signals. Among them, extracellular ATP has been well recognized as an endogenous signaling molecule in immunity and inflammation (30). Actually, extracellular nucleotides are tightly controlled by both cell membrane channels and ectonucleotidases, which are also involved in balancing immune responses. Thus, extracellular nucleotide signaling and metabolism is a dynamic area of research with important opportunities for novel treatments for inflammatory or infectious diseases (31).

As the initiation of inflammation during infection, endogenous nucleotides can be released through multiple channels to regulate immune responses at different times. For example, ATP can be released via exocytosis 30 min after challenged by bacterial infection, both in vivo and in vitro (25, 32). We also found that UDP could be highly released through pannexin channel in 24 h after virus infection. Therefore, it implied that the release of endogenous nucleotides is also accurately regulated by distinct infection. Interestingly, we observed a long-term and persistent release of UDP after bacterial infection, whereas the mechanism seems different from viral infection. The LPS-induced UDP release can be dramatically blocked by the connexin channel inhibitor FFA, but only slight UDP release was reduced by the pannexin channel inhibitor CBX, which was found to almost totally block virus induced UDP release (Fig. 2A). It has been well accepted that pannexin channels can be activated by caspases during apoptosis; however, the reason why connexin channels were activated in LPS-treated cells still needs to be further explored.

As the most important pattern recognition receptors in the innate immune system, TLRs sense the invasion of microbes by recognizing their PAMPs and activate intracellular signaling pathways leading to the expression of genes responsible for inflammatory and immune responses (33). About 10 human and 12 mouse TLRs were identified, each of which has distinct recognition patterns. Interestingly, although the recognition patterns of each TLR are different, the downstream signaling of each TLR is similar. TLR ligands induce formation of homodimers or heterodimers of TLRs to recruit MyD88 and dissociate IL-1R–associated kinase. Then, IL-1R–associated kinase interacts with TNFR-associated factor 6 and recruits TAK1 to phosphorylate IKKβ and MAPK (34). As shown in Fig. 1F, both LPS and Pam₃CSK₄ can induce UDP release, implying that the connexin-mediated UDP release through a classical and common TLR signaling pathway. Therefore, we pretreated the cells with U0126, Bay11, and Ly294002 to explore the key role of ERK, NF-κB, and AKT signaling in TLR-induced UDP release. As shown in Fig. 4C and 4D, the expression of connexin 43 was reduced obviously by U0126 both in mRNA and protein level, whereas LPS-induced connexin 43 was little changed by Bay11 and Ly294002. Consequently, LPS-induced UDP release, also inhibited by U0126 and connexin 43 blocking peptide Gap26, suggested the fundamental role of ERK activated connexin 43 in UDP release. More interestingly, like most other P2Y receptors, P2Y₆ is coupled with Gq to activate PLCβ and MAPKs (35). Also, protein kinase C and Erk1/2 has been found to play an important role in P2Y₆-mediated antiapoptotic functions (36). That is to say, LPS increased ERK signaling could induce UDP release and then eUDP bound to P2Y₆ further enhanced ERK activation, which constitutes a positive feedback loop in the initiation of innate immune responses. These findings provide solid evidence that eUDP could serve as a positive regulator in macrophage mediated innate immunity.

To further elucidate the important role of eUDP and P2Y₆ in TLR mediated innate immune responses, we treated RAW 264.7 cells with UDP and the P2Y₆ inhibitor MRS2578. As shown in Fig. 5, both UDP and LPS can increase MCP-1 expression obviously. However, LPS-induced MCP-1 production could be restrained by MRS2578 or P2Y₆ deficiency obviously, suggested UDP/P2Y₆ signaling also being involved in LPS mediated immune responses. Furthermore, the blocking of connexin 43–mediated UDP release by FFA can also decrease LPS-induced MCP-1 production. Accordingly, the clearance of invaded bacteria and mice survival in the acute peritonitis mouse model was depressed by U0126 and Gap26, which further confirmed the pivotal role of connexin 43–mediated GJC in the fight against bacterial infection. Taken together, our study reveals the internal relationship between danger signals and TLR signaling in innate immune responses, which suggest a potential therapeutic significance of the GJC and UDP/P2Y₆-associated signaling pathway in the prevention and control of bacterial diseases.

The authors have no financial conflicts of interest.