Prolonged immune alteration following resolution of acute inflammation in humans

Abstract

Acute inflammation is an immediate response to infection and injury characterised by the influx of granulocytes followed by phagocytosing mononuclear phagocytes. Provided the antigen is cleared and the immune system of the host is fully functional, the acute inflammatory response will resolve. Until now it is considered that resolution then leads back to homeostasis, the physiological state tissues experienced before inflammation occurred. Using a human model of acute inflammation driven by intradermal UV killed Escherichia coli, we found that bacteria and granulocyte clearance as well as pro-inflammatory cytokine catabolism occurred by 72h. However, following a lag phase of about 4 days there was an increase in numbers of memory T cells and CD163+ macrophage at the post-resolution site up to day 17 as well as increased biosynthesis of cyclooxygenase-derived prostanoids and DHA-derived D series resolvins. Inhibiting post-resolution prostanoids using naproxen showed that numbers of tissue memory CD4 cells were under the endogenous control of PGE2, which exerts its suppressive effects on T cell proliferation via the EP4 receptor. In addition, we re-challenged the post-resolution site with a second injection of E. coli, which when compared to saline controls resulted in primarily a macrophage-driven response with comparatively fewer PMNs; the macrophage-dominated response was reversed by cyclooxygenase inhibition. Re-challenge experiments were also carried out in mice where we obtained similar results as in humans. Therefore, we report that acute inflammatory responses in both humans and rodents do not revert back to homeostasis, but trigger a hitherto unappreciated sequence of immunological events that dictate subsequent immune response to infection.

Fig 1. Acute inflammatory response to intradermal UVkEc injection.

Acute inflammation was triggered in the ventral aspect of forearm of healthy volunteers by the intradermal injection of 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline. Vascular hyperaemia at the site was assessed by laser Doppler imager and the representative flux images after a specified time point are shown here (A). A 3 mm skin punch biopsy was taken from the inflamed site under local anaesthesia at the specified interval and formalin fixed paraffin embedded (FFPE) skin sections were probed by immunohistochemistry for E. coli LPS. The representative sections are shown here (B). A suction blister was raised over the inflamed site at the specified interval to collect the inflammatory exudate. Exudate was centrifuged to separate cells from the supernatant. The immune cell subsets were identified by polychromatic flow cytometry. The total count/ml of neutrophils in the inflammatory exudate at specified interval is shown here (C). IL-6, TNFα and IL1β in the cell free exudate were measured using multiplex ELISA and their concentrations in the exudate at the specified intervals are shown here (D-F, respectively). n = 3 for each time point. Data presented as mean ± SEM.

Fig 2. Post-resolution tissue accumulation of memory T cells and lymphocyte mitogens.

Acute inflammation was triggered in the ventral aspect of forearm of healthy volunteers by the intradermal injection of 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline. A suction blister was raised over the inflamed site at the specified interval to collect the inflammatory exudate. Exudate was centrifuged to separate cells from the supernatant. The immune cell subsets were identified by polychromatic flow cytometry. The numbers of CD4⁺/CD45RO⁺/CCR7^- (A), CD8⁺/CD45RO⁺/CCR7^- memory T cells (B) and CD56⁺ NK cells (C) at the specified interval are shown here. IL-15 (D) and IL-7 (E) in the cell free exudate were measured using multiplex ELISA and their concentrations in the exudate at the specified intervals are shown here. n = 3 for each time point. Data presented as mean ± SEM.

Fig 3. Post-resolution tissue accumulation of macrophages.

Acute inflammation was triggered in the ventral aspect of forearm of healthy volunteers by the intradermal injection of 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline. A suction blister was raised over the inflamed site at the specified interval to collect the inflammatory exudate. Exudate was centrifuged to separate cells from the supernatant. The immune cell subsets were identified by polychromatic flow cytometry. The numbers of CD14⁺ mononuclear cells at the specified interval are shown here (A). A 3mm skin punch biopsy was taken from the inflamed site under local anaesthesia at the specified interval and the formalin fixed paraffin embedded (FFPE) skin sections were probed by immunohistochemistry for CD163. The representative sections are shown here (B). MCP-1, IP-10, MDC and MCP-4 (panels C-F) in the cell free exudate were measured using multiplex ELISA and their concentrations in the exudate at the specified intervals are shown here (C). n = 3 for each time point. Data presented as mean ± SEM.

Fig 4. Increased lipid mediator biosynthesis during post-resolution biology.

Acute inflammation was triggered in the ventral aspect of forearm of healthy volunteers by the intradermal injection of 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline. A suction blister was raised over the inflamed site at the specified interval to collect the inflammatory exudate. Lipid mediators in the cell free exudate were analysed by liquid chromatography tandem mass spectrophotometry (LC MS/MS). The levels of PGE₂ (A), TXB₂ (B), PGD₂ (C), PGF_2α (D), LXB₄ (E), 5,15-diHETE (F), RvD5 (G), 5,15 diHETE (G) and RvE3 (H) at specified interval are shown here. n = 3 for each time point. Data presented as mean ± SEM.

Fig 5. Inhibition of post-resolution prostanoids increases numbers of local memory CD4⁺ T cells.

Acute inflammation was triggered in the ventral aspect of forearm of healthy volunteers by the intradermal injection of 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline. Inflammation was allowed to progress for 9 days after which volunteers were administered naproxen (500 mg, BD) orally for six days to block the rise in post-resolution prostanoids. On day 14 suction blister were elicited to enumerate cell profiles focusing on CD4⁺/CD45RO⁺/CCR7^- (A) and CD8⁺/CD45RO⁺/CCR7^- memory T cells (B). To determine what effect prostanoids have on memory T cell function, peripheral blood CD3⁺ cells were isolated from chicken pox-exposed individuals and incubated with varicella zoster antigen with/without PGE₂ at concentrations found in post-resolution tissue (C) with/without EP 2/4 antagonists (D) while CD3⁺ cells from chicken pox naïve individuals were used as negative controls (E). n = 3 for each time point. Data presented as mean ± se. *p≤0.05, **p≤0.01; as determined by ANOVA followed by Bonferroni t test or by two-tailed Student's t test.

Fig 6. Post-resolution tissues mount a predominantly macrophage driven response that is cyclooxygenase mediated.

Volunteers were injected intradermally with 1.5 x 10⁷ UV-killed E. coli (UVKEc) suspended in 100 μl of sterile saline in one forearm and 100μl of sterile saline only in the contralateral forearm. Inflammation was allowed to progress for 9 days after which one group of volunteers were administered naproxen (500 mg, BD) orally for six days to block the rise in post-resolution prostanoids and the other group remained untreated. On day 14, the forearm sites on all volunteers were re-injected with UVKEc and a suction blister was raised over the inflamed site at 4hr to collect inflammatory exudate. The number of mononuclear cells (A) and neutrophils (B) in the saline injected arm, untreated group (naive); UVKEc injected arm, untreated group (post-resolution); saline injected arm, naproxen treated group (naive + naproxen) and UVKEc injected arm, naproxen treated group (post-resolution + naproxen) are shown here. For mouse experiments (panels C and D), 0.1 mg zymosan was injected in the peritoneum of male c57bl/6 mice. 14 days later either saline or live S. pneumoniae were injected i.p. and 4 h later the composition of the peritoneum was determined by polychromatic flow cytometry; naproxen was dosed from day 7. For human studies there were n = 3 volunteers for each group, while there were n = 5 mice for rodents studies/group. Data presented as mean ± se. *p≤0.05, **p≤0.01; as determined by ANOVA followed by Bonferroni t test or by two-tailed Student's t test.