Activation of CD4⁺ T Cell–Derived Cannabinoid Receptor 2 Signaling Exacerbates Sepsis via Inhibiting IL-10

In the process of coevolution with the host, some microorganisms have acquired the ability to use the host’s components (such as cellular receptors) to enhance their infectivity or evade immune attacks. Unlike cannabinoid receptor 1 (CB1), which is mainly distributed in nerve cells, CB2 is predominant in immune cells, including B cells, NK cells, macrophages, neutrophils, and T cells (1). Previous studies reported the immunosuppressive role of CB2 signaling that improved the clinical condition by reducing immunopathology by restraining cytokine storms in infective models (2–12). Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection (13). Sepsis resulted in nine incidences and causes two deaths every 6 s globally, despite significant advancements in interventions in the past 30 y, causing a great economic and social burden (14). Cecal ligation and puncture (CLP) is considered the classical model in septic research because of its close similarity to human sepsis in terms of progression and characteristics (15). Cytokine storm might be one of the causes of death induced by sepsis (13, 16, 17), and the anti-inflammatory properties of CB2 draw attention to exploration of its role in this clinical situation. The effect of CB2 on sepsis remains controversial, and its mechanisms remain to be elucidated (18–20).

In the present study, we describe a novel finding that is different from previous reports that CB2 activation promotes suppressive cytokine production and inhibits inflammation. We also demonstrate that the CB2 target cells that affected survival in the CLP-induced septic model were CD4⁺ cells rather than other immune cells, such as neutrophils, macrophages, monocytes, and dendritic cells (DCs). Furthermore, IL-10 is proposed as one of the factors that may contribute to enhanced survival in septic mice.

Inbred C57BL/6 mice (CD4 Cre CNR2^fl/fl, CD11c Cre CNR2^fl/fl, Lyz2 Cre CNR2^fl/fl, CNR2^fl/fl) were kindly gifted by Dr. Cao of Jinan University. All mice were housed in specific pathogen-free conditions, and our experimental procedures were approved by the animal ethics committee of the Third People’s Hospital of Shenzhen. All these CNR2 gene conditional knockout mice were confirmed by common PCR with DNA extracted from the mouse tail for determining the profile of Cre (cyclization recombination enzyme) and loxP (locus of X-over P). Representative mice were further verified in specific cells by quantitative real-time PCR and functional assay (Supplemental Fig. 1). Mice whose sex and age were matched were randomly included in experiments.

The procedures were performed according to previous reports (21, 22). In brief, mice were anesthetized with tribromoethanol (T48402-5G, Sigma-Aldrich) with 300 mg/kg administered i.p. A 1-cm midline laparotomy was performed to expose the cecum. A total of 50% of the cecum was ligated with 4-0 silk ligature and then punctured through-and-through with a 22-gauge needle, and a small amount of fecal content extruded out of the perforation site. The cecum was then returned to the abdomen. The incision was closed with 4-0 silk ligature in two layers. Then mice received warm PBS resuscitation but without antibiotic treatment.

Activating CB2 signaling with a selective CB2 agonist HU308 (434184, Tocris Bioscience) with 4.8 mg/kg administered i.p. immediately after CLP surgery. The interruption of IL-10 signaling was achieved by IL-10 Ab (504909, BioLegend) administered i.p. twice: 1 h before and 20 h after CLP.

Blood and peritoneal lavage fluid (PLF) were harvested from mice 24 h after CLP. Splenocyte culture supernatant was collected 24 h after stimulation with LPS. IL-1β, IL-6, IL-10, and TNF-α levels were determined by ELISA following the manufacturer’s instructions (88-7013-88, 88-7064-88, 88-7105-88, 88-7324-88; Thermo Fisher Scientific).

Murine lungs, spleens, and PLF were harvested 24 h after CLP due to the inflammatory peak in the septic course (13, 15, 16). Lung were digested with 2 mg/ml collagenase D (11088866001, Roche) and 1 mg/ml DNase I (10104159001, Roche) for 0.5 h at 37°C. The single-cell suspension was obtained by the use of a gentleMACS dissociator (Miltenyi Biotec) and sifted through a 70-μm cell strainer. RBCs were lysed in ACK Lysis Buffer (RT122-02, Tiangen). Cells were stained with the following Abs: CD11b-PerCP-cyanine 5.5 (Cy5.5) (101228, BioLegend), Ly6G-PE/Cy7 (127618, BioLegend), Ly6C-PE (128008, BioLegend), CD3e-FITC (100306, BioLegend), NK1.1-PE/Cy7 (108714, BioLegend), CD314-PE (115606, BioLegend), and a dead/live dye (65-0864-14, eBioscience). Intercellular cytokine staining was performed according to a procedure described in previous reports (14, 23), and the following Abs were used in this test: CD3e-FITC (100306, BioLegend), CD4-allophycocyanin-eFluor 780 (47-0041-82, eBioscience), IL-10-allophycocyanin (65-0865-14, eBioscience). Cell staining was performed using a FACSCanto II cytometer (BD Bioscience), and data were analyzed by FlowJo software (BD Biosciences). Gating strategies are presented in Supplemental Fig. 2.

Mouse spleens were minced by sifting through a 70-μm strainer, and RBCs were lysed. Splenocytes were cultured in DMEM supplemented with 10% FCS and treated with 100 ng/ml LPS (L3012-10MG, Sigma-Aldrich) and/or 2.5 μM HU308 at 37°C and 5% CO₂. After 12–24 h, the cells were harvested for further experiments. In the intercellular cytokine staining experiment, 3 μg/ml brefeldin A (00-4506-51, Thermo Fisher Scientific) was added to block IL-10 secretion during the culture for 6 h.

Blood and PLF were harvested from mice 24 h after CLP and plated on tryptic soy agar at 37°C for 12 h, after which the colony number was counted.

Patients with sepsis in the Third People’s Hospital of Shenzhen were enrolled in the study and compared with healthy donors. All subjects provided informed consent, and this study was approved by the ethics committee of the Third People’s Hospital of Shenzhen. PBMCs from septic patients or healthy donors were collected by density gradient centrifugation with Ficoll buffer. Total RNA was obtained by using the Total RNA Kit I (R6834-02, OMEGA Bio-tek) according to the manufacturer’s instructions. The RNA was reverse transcribed to cDNA and assayed by quantitative PCR with SYBR Green on ABI7500 instruments (Thermo Fisher Scientific). All primer sequences are listed in Supplemental Table I.

The differences in survival rates were assessed by the Mantel-Cox test. Other data were presented as the mean ± SEM, and Student t tests were used to compare statistical significance between two groups. The Mann-Whitney U test was used when data were not normally distributed. A p value < 0.05 was considered significant. GraphPad Prism 6 software (GraphPad Software, La Jolla, CA) was used for the data analysis.

Previous studies reported immunosuppression of CB2 signaling in infective tissues (2–12). To investigate the role of CB2 signaling in sepsis, we first tested the toxicity of 4.8 mg/kg of a CB2 selective agonist HU308 i.p. administered into healthy CNR2^fl/fl mice. Mouse survival during the 10-d period after the treatment combined with lung H&E staining concluded that HU308 was not fatally detrimental to mice in this dosage (Fig. 1A, Supplemental Fig. 3). Then, we determined CB2’s function in septic mice modeled by CLP and observed that activating CB2 signaling with HU308 resulted in decreased survival rates in comparison with the control group (Fig. 1B). The HU308 treatment had no significant effect on the mean bacterial load in the blood and peritoneum (Fig. 1C). Representative H&E staining in the lung section revealed elevated inflammation and more severe injury in the HU308 administration group (Fig. 1D).

The above results elucidate that CB2 signaling activated by HU308 increased the incidence of death in septic mice, most likely by exacerbating pulmonary inflammation and damage.

To explore the reason behind the increased lethality activated by CB2 signaling in septic mice, the cytokine concentrations in blood and the peritoneum were determined by ELISA. Among the pro- and anti-inflammatory mediators (IL-1β, IL-6, TNF-α, IL-10), only IL-10 in PLF was found to be significantly different (reduced) in septic mice treated with HU308 (Fig. 2A). Then we tested the immune cells distribution, such as polymorphonuclear leukocytes, macrophages, monocytes, NK cells, T cells, and NKT cells, in the lung, spleen, and PLF in septic mice. No significant differences were found between the CB2 agonist group and the control, except for decreased NK percentage in the lung and spleen in septic mice that received HU308 (Fig. 2B). These results implied that IL-10 in the initial infective site might be one of the factors contributing to improved survival in sepsis.

Next, we sought to find out the target cells responsible for CB2 signaling, which are detrimental to the septic mice’s survival. Conditional knockout mice, whose CB2 genes were deleted in Lyz2⁺ cells, CD11c⁺ cells, or CD4⁺ cells, were established. All these conditional knockout mice exhibited a representative deficiency of CB2 in macrophages, alveolar macrophages, monocytes, polymorphonuclear leukocytes, DCs, and CD4⁺ T cells. We observed that lack of CB2 in CD4⁺ cells, but not in other cells, improved the septic mice’s survival, which suggested the critical role of CD4⁺ cells involved in the exacerbation of CB2 signaling on sepsis (Fig. 3A). Subsequently, we also analyzed the bacterial load between CD4 Cre CNR2^fl/fl mice and their control animals, CNR2^fl/fl mice. We found that the difference between CD4 Cre CNR2^fl/fl mice and CNR2^fl/fl mice in bacterial burden was not significant (Fig. 3B). CD4 Cre CNR2^fl/fl mice showed lower infiltration of inflammatory cells and milder lung injury in H&E stains (Fig. 3C).

To further verify the role of CB2 signaling in CD4⁺ cells, but not in other cells that exacerbated sepsis, we established Lyz2 Cre CD11c Cre CNR2^fl/fl mice (CB2 knockout in Lyz2⁺ cells and CD11c⁺ cells) and Lyz2 Cre CD4 Cre CNR2^fl/fl mice (CB2 knockout in Lyz2⁺ cells and CD4⁺ cells) and tested them in sepsis (Fig. 4). In accordance with the above results, no significant differences in survival rates were found between Lyz2 Cre CD11c Cre CNR2^fl/fl mice and the controls. However, Lyz2 Cre CD4 Cre CNR2^fl/fl mice exhibited higher survival rates than CNR2^fl/fl mice and Lyz2 Cre CNR2^fl/fl mice, but not CD4 Cre CNR2^fl/fl mice. The H&E staining of lungs in these septic mice confirmed this result. These results provide evidence that CB2 signaling in CD4⁺ cells was the reason for increased mortality rates caused by aggravating pulmonary inflammation.

IL-1β, IL-6, IL-10, and TNF-α levels in blood and PLF from CD4 Cre CNR2^fl/fl mice and CNR2^fl/fl mice were assayed 24 h after CLP. Consistent with the above results, only IL-10 in PLF was significantly different between these two groups, and CD4 Cre CNR2^fl/fl mice exhibited a higher level (Fig. 5A). The percentages of immune cells in the lung, spleen, and PLF were investigated by flow cytometry. No significant differences were found, including the NK percentages in the lung and spleen that were lower in the HU308-treated CNR2^fl/fl mice in the above results when compared with their controls. This implied that NK might not be involved in protecting mice from sepsis in the CB2 activation model (Fig. 5B). Hence, deficiency of CB2 signaling in CD4⁺ cells may result in the elevation of IL-10 in the initial infective site.

To further verify the protective role of IL-10 in the peritoneum in septic mice, we neutralized the septic mice’s IL-10 in the peritoneum by administering the IL-10 Ab i.p. We found that the blockade in CD4 Cre CNR2^fl/fl mice of peritoneal IL-10 decreased survival when compared with isogenic mice treated with an isotype Ab. Furthermore, neutralization of IL-10 also eliminated the survival advantage of CD4 Cre CNR2^fl/fl over CNR2^fl/fl mice (Fig. 6A). Lung H&E staining showed more inflammatory cell infiltration in septic mice treated with the IL-10 Ab than in the control (Fig. 6B). We address whether, consistent with the in vivo results, CB2 activated with HU308 limited IL-10 expression in vitro. Because LPS is one of the important factors resulting in death in sepsis, splenocytes derived from mice were isolated and stimulated with LPS and HU308 for 12 or 24 h, and their IL-10 levels were determined by ELISA. In accordance with the in vivo study, splenocytes whose CD4⁺ cells lacked CB2 exhibited elevated IL-10 levels after 12- and 24-h treatment with LPS and HU308 (Fig. 6C). The flow cytometric analysis uncovered that HU308 stimulation restrained CD4-derived IL-10 expression (Fig. 6D). This result also showed that IL-10 was produced mainly from CD4⁺ cells in the LPS model, which may explain why CB2 deficiency in CD4⁺ cells, but not other immune cells, enhanced survival in septic mice.

To address whether CB2 signaling involved in sepsis in the animal model also plays a critical role in humans, we compared the septic patients’ PBMCs for CB2 expression and its correlation with IL-10 expression with those in healthy donors (Table I). The results showed that septic patients expressed significantly elevated CB2 levels in comparison with healthy donors (Fig. 7A). Furthermore, CB2 expression was reverse correlated with IL-10 expression in septic patients, which was not found in healthy donors (Fig. 7B). Thus, CB2 expression is consistent with a role for lower IL-10 levels in worse outcomes in human sepsis.

With advances in research, we realize that sepsis is not only associated with the pathogen itself but also with the host’s factors, because the resulting immune response determines the prognosis of septic patients to a large extent (24–26). Mounting evidence points toward the coexistence of pro- and anti-inflammatory aspects in sepsis. Moreover, wide heterogeneity among individuals exists, making it difficult to select an appropriate therapy (13, 17, 26).

In this study, we demonstrate that activated CB2 signaling leads to increased mortality rates in septic mice. In order to clarify the CB2 gene in the immune cell population associated with sepsis, series conditional knockout mice were used. However, it is noteworthy that the CB2 conditional knockout phenotype could not be verified at the protein level due to the poor specificity of the CB2 Ab (101550, Cayman Chemical), so we addressed this issue by detecting CB2 mRNA expression and performing a functional assay through testing its downstream signaling pathway. Interestingly, the deficiency of CB2 in CD4⁺ cells, but not in other immune cells, such as neutrophils, macrophages, and DCs, enhanced animal survival during sepsis determined with CB2 conditional knockout mice. We also found that this improvement in survival might be associated with increased IL-10 production, because the interruption of IL-10 results in deteriorating conditions.

Classically, CB2 had been thought to be immunologically suppressive, which limits inflammation by inducing anti-inflammatory cytokine production, activating immunosuppressive immune cells, or inhibiting inflammatory cells (9–12, 27, 28). However, in our study, we observed that activating CB2 signaling with the CB2 selective agonist HU308 resulted in decreased IL-10 production assayed by ELISA, not only in an in vivo septic model but also in the in vitro experiment. Our flow cytometric results similarly verified CB2’s inhibition of IL-10 expression. Our explanation is that HU308, as a highly selective agonist of CB2, has more specificity for the CB2 receptor than previously employed CB2 agonists such as Δ-9-tetrahydrocannabinol that activated not only CB2 but also CB1 and other receptors (29, 30). Moreover, various models and different dosages may contribute to different outcomes for CB2 function. As two examples, one study reported that neutralizing TNF-α enhanced inflammatory cell infiltration in the lung and improved survival in influenza-infected mice (31), whereas another study found that mice deficient in TNF-α show more severe immunologic injury in the course of influenza infection (32). Furthermore, undiscovered pathways or mechanisms may be involved.

Although we found higher IL-10 levels in the in vivo experiment in the CD4 Cre CNR2^fl/fl group, a similar phenomenon was observed in the in vitro experiment performed with splenocytes. IL-10 proved to be derived mainly from CD4⁺ cells in the flow cytometry assay; CD4⁺ cells isolated from splenocytes with microbeads produced undetected IL-10 when treated with LPS or LPS plus HU308 performed as an experiment with splenocytes (data not shown). These results imply the existence of a complex interaction between CD4⁺ cells and other immune cells in terms of IL-10 production. Interestingly, we also observed in this study that CB2 signaling activating effected IL-10 levels in the initial infective site (the peritoneum) rather than in the circulation. That could be due to the stronger immune response in the peritoneum because this septic model established by CLP resulted in bacterial diffusion from the intestinal tract (15). There might exist a more powerful buffer system against IL-10 elevating in the circulation that contributes to weaker influence on the IL-10 level by CB2 signaling.

The role of IL-10 in infective disease is controversial. In the influenza virus infection model, IL-10 proved to be beneficial for the improvement of survival rates by reducing pulmonary immunopathology by restraining the cytokine storm (33, 34). A previous study on sepsis also provided evidence that IL-10 was the reason why hyperbaric oxygen therapy protected mice from sepsis, because this treatment lost its protection in IL-10^−/− mice (35). Van der Poll et al. (36) found an increase in IL-10 mRNA and protein expression during CLP-induced sepsis. Treatment with the IL-10 Ab resulted in an elevated plasma TNF level and increased mortality rates after CLP surgery. Another report revealed that endogenous IL-10 protected septic IL-10–null mice by reducing thymic apoptosis induced by sepsis (37). NK-derived IL-10 was demonstrated to be beneficial in septic mice and patients in its scope and duration by suppressing the cytokine storm (14).

The above results are consistent with our study, indicating a protective role of IL-10 on sepsis. However, some studies reported that high IL-10 levels were among the bad prognostic factors for septic patients (38). The inhibition of IL-10 production with an immunomodulator activated macrophage functions and improved survival rates in septic mice (39). The IL-10 level in serum was significantly positively correlated with the percentages of regulatory T cells in PBLs in septic patients and mice. The interruption of IL-10 and TGF-β signaling reduced the regulatory T cell percentage but enhanced CD4⁺ cells and prevented death in mice (40). Our explanations for this contradiction are as follows. (1) Sepsis is a complex syndrome induced by different pathogens and sometimes by the coinfection of two or more pathogens, leading to various outcomes. (2) Differences in the genetic background, lifestyle, and living conditions may be reasons behind the heterogeneity of septic patients. (3) As described above, immune characteristics vary during sepsis; that is, excessive inflammation and immune suppression may be involved. In the present study, we measured IL-10 levels 24 h after CLP. The IL-10 signaling block was effectuated in our test within 24 h after surgery. Arguably, the proinflammatory period during sepsis benefited from the anti-inflammatory aspect of IL-10.

Based on the complexity of sepsis and individual heterogeneity, no single therapy can be considered as most suitable for all septic patients. Hence, it is advisable to make a treatment projection while taking into account the disease course and individual patient characteristics.

The authors have no financial conflicts of interest.