TLR9 and IRF3 cooperate to induce a systemic inflammatory response in mice injected with liposome:DNA

Abstract

Liposome:DNA is a promising gene therapy vector. However, this vector can elicit a systemic inflammatory response syndrome (SIRS). Prior reports indicate that liposome:DNA vectors activate Toll-like receptor (TLR)9. We hypothesized that liposome:DNA vectors also activate the cytosolic DNA-sensing pathway, which signals via interferon (IFN) regulatory factor (IRF)3. To test this, we treated dendritic cells (DCs) with liposome:DNA in vitro and found that IRF3 was phosphorylated independent of TLR9. To test the contribution of this pathway in vivo, we injected a liposome:DNA vector into wild-type (WT), TLR9-knockout (KO), IRF3-KO, and TLR9-IRF3-double-KO (DKO) mice. WT mice exhibited a systemic inflammatory response, evidenced by elevations in serum cytokines, serum enzyme changes indicating organ damage, hypothermia, and mortality. The cytokine response was reduced in TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO mice and all three groups survived. We found that IFN-gamma-KO mice that receive liposome:DNA had a reduced cytokine response and 100% survival. CD11c(+) and NK1.1(+) cells produced IFN-gamma and depleting CD11c(+) cells reduced the cytokine response in mice injected with liposome:DNA. These findings may facilitate the development of immunologically inert gene therapy vectors and may provide general insight into the mechanisms of SIRS.

Figure 1

TLR9 and IRF3 induced serum cytokines, mortality, and hypothermia in response to liposome:DNA (LD) vector. (a) Western blot demonstrates that IRF3 serine-396 was phosphorylated (P-IRF3) in TLR9-KO and wild-type (WT) DCs, but not IRF3-KO DCs, following 1–2 hours of LD treatment in vitro. Total IRF3 and actin controls are also shown. (b) Serum IFN-γ, IL-6, IL-12, MCP-1, IFN-α, and IFN-β levels were reduced in TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO mice relative to WT mice injected with LD. Graphs show mean ± SE. (For all graphs P ≤ 0.0061 for 0–8 hours time points by two-way analysis of variance (ANOVA); Bonferroni post-test *P < 0.05 and **P < 0.01 for WT versus TLR9-KO, IRF3-KO and TLR9-IRF3-DKO; ^##P < 0.01 for WT versus TLR9-KO and TLR9-IRF3-DKO and IRF3-KO versus TLR9-KO and TLR9-IRF3-DKO). (c) There was significant mortality in WT mice following injection of LD, whereas all TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO mice survived. Graph shows survival fractions (Kaplan–Meier method), ⁺P < 0.0001 by Logrank test. (d) Surface body temperature dropped significantly in WT mice injected with LD, whereas TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO mice maintained normal temperature. Graph shows mean ± SE. P = 0.0316 by two-way ANOVA for 0- to 46-hour time points, Bonferroni post-test **P < 0.01 for WT versus TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO. DC, dendritic cells; IFN, interferon; IL, interleukin; IRF, IFN regulatory factor; KO, knockout; MCP, monocyte chemotactic protein; TLR, Toll-like receptor.

Figure 2

A trend toward reduced ALT and AST in mice lacking TLR9 or IRF3. There was a trend toward reduced serum (a) ALT and (b) AST in TLR9-KO, IRF3-KO, and TLR9-IRF3-DKO mice versus WT mice 24 hours after injection of LD. AST and ALT remained at background levels in mice injected with 5% dextrose vehicle. Graphs show mean ± SE. ALT, alanine aminotransferase; AST, aspartate aminotransferase; IRF, interferon regulatory factor; LD, liposome:DNA; TLR, Toll-like receptor; WT, wild-type.

Figure 3

IFN-γ played a key role in the innate immune response to liposome:DNA (LD) vector complex: experiments using genetic knockouts. (a) There was significant mortality in wild-type (WT) and IFN-αβR-KO mice following injection of LD, whereas all IFN-γ-KO mice survived. Graph shows survival fractions (Kaplan–Meier method), ⁺P = 0.0487 by Logrank test. (b) Serum IFN-γ and IL-6 levels were significantly reduced in IFN-γ mice versus WT mice, and there was a trend toward reduced IL-12, after injection of LD. IL-6 and IL-12 levels were similar in IFN-αβR-KO mice and WT mice, whereas IL-12 levels were significantly higher in IFN-αβR-KO mice, after injection of LD. For all graphs P ≤ 0.0387 for 0- to 6-hour time points by two-way analysis of variance (ANOVA), Bonferroni post-test **P < 0.01 for IFN-γ-KO versus WT. ^##P < 0.01 for IFN-αβR-KO versus WT. (c) Surface body temperature dropped significantly in WT mice injected with LD, whereas IFN-γ-KO mice maintained normal temperature. P = 0.0033 by two-way ANOVA for 0- to 48-hour time points, Bonferroni post-test *P < 0.05 for IFN-γ-KO versus WT. (d) There was trend toward reduced serum ALT, and AST values were significantly lower in IFN-γ-KO mice versus WT mice at 24 hours after injection of LD. ALT and AST remained at background levels in mice injected with 5% dextrose vehicle. *P < 0.05 by t-test. (b–d) Graphs show mean ± SE. ALT, alanine aminotransferase; AST, aspartate aminotransferase; IFN, interferon; IL, interleukin; KO, knockout.

Figure 4

IFN-γ played a key role in the innate immune response to liposome:DNA (LD) vector: experiments using a neutralizing antibody. (a) 100% of wild-type (WT) mice survived when administered an IFN-γ-neutralizing antibody (anti-IFN-γ) before the injection of a high-dose of LD vector (50 µg plasmid DNA), in contrast to counterparts administered rat-IgG1 anti-horseradish peroxidase (rIgG1-HRP) isotype control. Graph shows survival fractions (Kaplan–Meier method). ⁺P = 0.0043 by Logrank test. (b) Serum IFN-γ, IL-6, and IL-12 cytokine levels were lower in WT mice that received an IFN-γ-neutralizing antibody versus rIgG1-HRP isotype, before injection of LD. Graphs show mean ± SE. P ≤ 0.01 by two-way analysis of variance for 0- to 6-hour time points, Bonferroni post-test **P ≤ 0.01. IFN, interferon; IL, interleukin.

Figure 5

Liposome:DNA (LD) vector induced IFN-γ production by CD11c⁺CD11b⁺ and NK1.1⁺ spleen cells. (a) IFN-γ was produced ex vivo by liver and spleen cells from wild-type (WT) mice injected with LD. In contrast, lung cells from these mice failed to produce more IFN-γ than naive counterparts. P < 0.01 by one-way analysis of variance (ANOVA), Bonferroni post-test **P < 0.01. (b) Left panels: intracellular cytokine staining revealed an IFN-γ⁺ population in spleen cells from WT mice injected with LD (top). This IFN-γ⁺ population was not present in mice that received 5% dextrose vehicle (bottom). Center and right panels: the IFN-γ⁺ subgate predominantly contained CD11c⁺CD11b⁺ and NK1.1⁺ cells (39.8 and 62.7% respectively, bottom panels), whereas these cell types were rare in the total spleen cell gate (2.21 and 1.9% respectively, top panels). (c) Depletion of CD11c, CD11b, CD11c+CD11b, or NK1.1 populations abrogated IFN-γ production ex vivo by spleen cells from mice injected with LD. P < 0.0001 by one-way ANOVA, Bonferroni post-test **P < 0.01 unfractionated versus CD11c-depleted, CD11b-depleted, CD11c+CD11b-depleted, and NK1.1-depleted. (a and c) Graphs show mean ± SE.

Figure 6

CD11c⁺ cells played a major role in proinflammatory cytokine production in vivo. (a) Depletion of dendritic cells in CD11c-DTR mice via diphtheria toxin (DT) treatment reduced IFN-γ and IL-12 production and there was a trend toward reduced IL-6 production, in comparison to control mice [CD11c-DTR mice treated with phosphate-buffered saline (PBS) and wild-type (WT) mice treated with DT or PBS], following injection of LD. For IFN-γ and IL-12 graphs P ≤ 0.0055 by two-way analysis of variance (ANOVA) for 0- to 6-hour time points, Bonferroni post-test **P < 0.01 for CD11c-DTR+DT versus CD11c-DTR+PBS, WT+DT, and WT+PBS. (b) Depletion of NK1.1⁺ cells in WT mice via administration of NK1.1 depletion antibody (anti-NK1.1) reduced IFN-γ and IL-6 production to a small degree and did not affect IL-12 production, in comparison to control mice [administered mouse-IgG2a (mIgG2a) isotype control], following injection of LD vector. For IFN-γ and IL-6, P ≤ 0.04 by two-way ANOVA for 0- to 6-hour time points, (IL-12 = ns), Bonferroni post-test ^#P < 0.05 and ^##P < 0.001 for anti-NK1.1 versus mIgG2a. Graphs show mean ± SE. IFN, interferon; IL, interleukin; LD, liposome:DNA.