Interleukin-23 receptor expressing γδ T cells locally promote early atherosclerotic lesion formation and plaque necrosis in mice

Abstract

Aims

Atherosclerosis is a chronic inflammatory disease of the vessel wall controlled by local and systemic immune responses. The role of interleukin-23 receptor (IL-23R), expressed in adaptive immune cells (mainly T-helper 17 cells) and γδ T cells, in atherosclerosis is only incompletely understood. Here, we investigated the vascular cell types expressing IL-23R and addressed the function of IL-23R and γδ T cells in atherosclerosis.

Methods and results

IL-23R⁺ cells were frequently found in the aortic root in contrast to the aorta in low-density lipoprotein receptor deficient IL-23R reporter mice (Ldlr^−/−Il23r^gfp/+), and mostly identified as γδ T cells that express IL-17 and GM-CSF. scRNA-seq confirmed γδ T cells as the main cell type expressing Il23r and Il17a in the aorta. Ldlr^−/−Il23r^gfp/gfp mice deficient in IL-23R showed a loss of IL-23R⁺ cells in the vasculature, and had reduced atherosclerotic lesion formation in the aortic root compared to Ldlr^−/− controls after 6 weeks of high-fat diet feeding. In contrast, Ldlr^−/−Tcrδ^−/− mice lacking all γδ T cells displayed unaltered early atherosclerotic lesion formation compared to Ldlr^−/− mice. In both HFD-fed Ldlr^−/−Il23r^gfp/gfp and Ldlr^−/−Tcrδ^−/− mice a reduction in the plaque necrotic core area was noted as well as an expansion of splenic regulatory T cells. In vitro, exposure of bone marrow-derived macrophages to both IL-17A and GM-CSF induced cell necrosis, and necroptotic RIP3K and MLKL expression, as well as inflammatory mediators.

Conclusions

IL-23R⁺ γδ T cells are predominantly found in the aortic root rather than the aorta and promote early atherosclerotic lesion formation, plaque necrosis, and inflammation at this site. Targeting IL-23R may thus be explored as a therapeutic approach to mitigate atherosclerotic lesion development.

Graphical Abstract

Open in new tabDownload slide

1. Introduction

Atherosclerosis is a chronic inflammatory disease of the vessel wall driven by the progressive intimal accumulation of leucocytes. It is well recognized that immune responses play a critical role in atherogenesis.^1,2 Monocytes differentiate into macrophages and take up oxidized LDL (oxLDL) to transform into foam cells. The apoptosis of foam cells and other leucocytes eventually overwhelms the phagocytic capacities of viable macrophages, and necrosis leads to the formation of a lipid-rich, highly proinflammatory necrotic core. In addition, proinflammatory programmed cell necrosis, or necroptosis, promotes the growth of the necrotic core.^1–3 Also adaptive immune responses are activated, and a balance between proinflammatory and regulatory T-cell (Treg) responses control atherosclerosis.⁴ Among CD4⁺ T cells, interferon (IFN)-γ producing T-helper 1 (Th1) cells promote atherosclerosis, Th2 responses contribute to atheroprogression or protection, dependent on the context and model, and Foxp3⁺CD25⁺ Tregs inhibit vascular inflammation. The function of IL-17 is still controversial,⁴ likely related to different IL-17-expressing cell types, which include Th17 cells and γδ T cells. γδ T cells are found in healthy vessels and in the aortic root in mice and humans.^5,6 The absence of γδ T cells, however, has not been found to affect lesion formation in apolipoprotein E-deficient (Apoe^−/−) mice.^7,8

IL-23 receptor (IL-23R) is mainly expressed by Th17 cells and γδ T cells.⁹ The IL-23R is regarded as the only signal-transducing component of the IL-23R: IL-12Rβ1 complex and binds the α-chain (p19) of IL-23, which unlike the β-chain (p40) of IL-23 interacting with IL-12Rβ1 is not shared with IL-12.¹⁰ In Th17 cells, IL-23 promotes a pathogenic Th17 phenotype characterized by the expression of IL-17, GM-CSF, and IFN-γ,^11,12 as opposed to non-pathogenic Th17 cells expressing IL-17 and IL-10.^13,14 Similarly, pathogenic effector γδ T cells that express IL23R, IL-17, and GM-CSF have been described in experimental autoimmune encephalomyelitis.^15–17

The role of IL-23/IL-23R in atherosclerosis is controversial. Administration of recombinant IL-23 promoted plaque necrosis and macrophage apoptosis in atherosclerosis-prone low-density lipoprotein receptor knockout (Ldlr^−/−) mice,¹⁸ but Ldlr^−/− mice deficient in IL-23R did not show alterations in atherosclerotic lesion development or phenotype upon 10 weeks of high‐fat diet (HFD) feeding.¹⁹ At advanced stages of lesion formation after 16 weeks of Western diet, deficiency in IL-23 increased lesion formation in the aortic root of Ldlr^−/− mice by deterioration of the intestinal barrier and expansion of pro-atherogenic bacteria and metabolites.²⁰

Notably, IL-23R-dependent γδ T cells are found in the aortic root,⁶ and overexpression of IL-23 has been shown to cause inflammation within the aortic sinus by acting on IL23R⁺CD3⁺CD4⁻CD8⁻ cells,²¹ presumably γδ T cells. γδ T cells within tissues can respond via antigen-dependent but also antigen-independent mechanisms that allow them to rapidly activate specific effector functions.^22,23 IL23R expressing γδ T cells may thus be of particular importance for promoting early vascular inflammation and atherosclerotic lesion formation, not investigated previously, rather than lesion growth.

Therefore, we here set out to characterize IL-23R⁺ cells, and investigate the function of IL-23R and γδ T cells particularly during early atherogenesis. To this end, we used mice with a knock-in of green fluorescent protein (GFP) into the endogenous Il23r gene⁹ to investigate IL-23R expression on immune cells based on the detection of GFP in heterozygous Il23r^gfp/+ mice still fully responsive to IL-23, and its function in Il23r^gfp/gfp mice lacking functional IL-23R, crossed with atherosclerosis-prone Ldlr^−/− mice. Moreover, we investigated also δ TCR-deficient (Tcrδ^−/−) Ldlr^−/− mice that lack γδ T cells. We uncovered γδ T cells to be the predominant cell type expressing IL-23R within aortic tissue, and IL-23R⁺ γδ T cells to produce IL-17A and GM-CSF. Disruption of IL-23R signalling inhibited early lesion development and reduced necrotic core size locally in aortic root lesions, demonstrating that IL-23R⁺ γδ T cells play a role in promoting early atherosclerosis and plaque necrosis.

2. Methods

A detailed version of materials and methods appears in the Supplementary material online.

2.1 Mice and diet

IL-23R-GFP knock-in mice were generated by introducing an internal ribosome entry site-GFP-cassette into the endogenous Il23r gene.⁹ Whereas heterozygous Il23r^gfp/+ mice can be used as reporter mice to identify IL-23R-expressing GFP⁺ cells (fully responsive to IL-23 because of the intact Il23r allele), homozygous Il23r^gfp/gfp mice are unresponsive to IL-23 due to two mutated Il23r alleles. Ldlr^−/− mice (The Jackson Laboratory) were crossed with Il23r^gfp/+ mice to generate Ldlr^−/−, Ldlr^−/−Il23r^gfp/+ and Ldlr^−/−Il23r^gfp/gfp mice (C57BL7/6 background), and with Tcrδ^−/− mice (The Jackson Laboratory) to generate Ldlr^−/− and Ldlr^−/−Tcrδ^−/− mice (C57BL7/6 background). To induce atherosclerosis, male and female mice aged between 6 and 8 weeks were placed on a HFD (15% milk fat, 1.25% cholesterol, Altromin). Animals were euthanized by an overdose of isoflurane anaesthesia (5% concentration), and subsequent exsanguination and organ isolation. All animal studies conform to the Directive 2010/63/EU of the European Parliament and have been approved by the appropriate local authorities (Regierung von Unterfranken, Würzburg, Germany).

2.2 Flow cytometry

Fat was removed from the aorta, and the aorta was separated from the aortic root. They were each minced and enzymatically digested, and cells were stained with specific antibodies. Dead cells were removed from the analysis. Intracellular staining was performed in cells treated with brefeldin A, and ionomycin. To co-detect intracellular cytokines in Il23r^grp/+ mice, cells were fixed after surface staining using 2% formaldehyde to retain GFP expression before intracellular staining. Bone marrow-derived macrophages (BMDMs) were stained using the AnnexinV Apoptosis Detection Kit (Thermofisher, Scientific). Probes were analysed using a FACSCantoII or FACSCelesta and FlowJo 10.0 software (BD Biosciences).

2.3 Two-photon imaging

Aortic roots were imaged using an upright two-photon fluorescence microscope (TCS SP8 MP, Leica).

2.4 Immunohistochemistry and atherosclerotic lesion quantification

Arteries were perfusion-fixed in situ with phosphate buffered saline (PBS) followed by 4% paraformaldehyde (PFA) in PBS. The heart was removed for post-fixation in 4% PFA. The heart was embedded in Tissue Tek, frozen, and cut into 5 µm transverse sections. Aortic root sections were collected when all three valves were clearly visible. Three sections, 75 µm apart were stained by haematoxylin–eosin, and the average atherosclerotic plaque size was quantified. Quantification of plaque size in the aorta was performed after staining lipid depositions using Oil-Red-O. The area of the aorta occupied by Oil-Red-O⁺ lipids was quantified by computerized image analysis (Diskus Software), expressed as the percentage of total aortic area. Sections through the aortic root were used for immunofluorescence staining. Necrotic core formation was quantified in sections stained with haematoxylin and eosin by measuring lightly stained plaque area devoid of haematoxylin-stained cell nuclei. Apoptotic cells were detected by TUNEL. Images were recorded with a Leica DM 4000B fluorescence microscope and JVC KY-F75U camera. Plaque size and cell content were quantified by computerized image analysis (Diskus Software, Hilgers) and by investigators blinded to the group distributions.

2.5 Serum cholesterol measurement

Total cholesterol was analysed in serum (Amplex Red Cholesterol Assay Kit, Invitrogen).

2.6 Quantitative PCR analysis

Total RNA was isolated, reverse transcribed into cDNA, and quantitative real-time polymerase chain reaction (PCR) analysis was performed using SYBR-Green and specific primer pairs (QuantiStudio6 Flex Thermal Cycler, Applied Biosystems).

2.7 Bone marrow-derived antigen-presenting cells and BMDMs

Bone marrow cells were cultured in media containing 20 ng/mL GM-CSF (Peprotech) to obtain bone marrow-derived antigen-presenting cells (BM-APCs), and 15% L-929 cell-conditioned medium to obtain BMDMs.

2.8 CITE-seq analysis of total cells from mouse aortas

Circulating leucocytes were excluded from the analysis by injecting anti-CD45.2 intravenously before sacrifice. Aortas were enzymatically digested and cells were labelled with cellular indexing of transcriptomes and epitopes by sequencing (CITE-seq) and hashtag antibodies. Cells were sorted, counted in Trypan blue (>85% viability), and loaded in the 10× Genomics Chromium (Single Cell 3’ v3 reagents, 10× Genomics) and scRNA-seq, ADT, and HTO libraries were prepared.²⁴ After sequencing (Novaseq 6000 platform, Illumina), data were demultiplexed using Cell Ranger software v3.1.0. Mouse mm10-3.0.0 reference genome was used for alignment. Cell surface proteins were evaluated using--feature-ref flag of Cell Ranger software. Cell Ranger gene-barcode matrix was analysed using Seurat v3, and clustering was performed using the ‘FindClusters’ function. Mononuclear phagocytes and fibroblast subclusters were manually pooled to facilitate data visualization in Figure 1.

2.9 Western blot

Cell lysates were subjected to SDS-PAGE and blotted onto nitrocellulose. Membranes were blocked, incubated with primary/secondary antibody, and developed using enhanced chemiluminescence reagent. Images were recorded (Amersham Imager 600) and quantified (ImageQuantTL software, GE Healthcare).

2.10 Statistical analysis

Data represent mean ± SD. Comparisons were performed via unpaired Student’s t-test, one-way, two-way or three-way ANOVA followed by a Tukey’s multiple comparison test using Prism software (GraphPad). Differences were considered statistically significant at P < 0.05.

3. Results

3.1 γδ T cells are the major cell population expressing IL-23R within the aortic sinus

Previous studies have shown that mice overexpressing IL-23 develop inflammation in the aortic root, the presence of IL-23R⁺ CD3⁺CD4⁻ CD8⁻ T cells at this site,²¹ and that γδ T cells can been identified in the aortic root by immunofluorescence staining in Tcrd-H2BeGFP reporter mice.⁶ Using IL-23R reporter mice (Il23r^gfp/+) crossed with Ldlr^−/− mice, we found GFP-expressing IL-23R⁺ cells to represent ∼3% of total CD45⁺ cells in the aorta and up to ∼10% of leucocytes in the aortic root in 8 week old mice fed a standard diet (SD) (Figure 1A). Similar frequencies were observed in mice upon feeding a HFD for 6 weeks (Figure 1B). Although a significant increase in leucocytes was observed in the aorta of Ldlr^−/− mice and cell numbers tended to be increased in the aortic root of Ldlr^−/−, Il23r^gfp/+ Ldlr^−/−, and Il23r^gfp/gfp Ldlr^−/− mice after HFD-feeding, absolute counts of IL-23R⁺ cells were unaltered in the aorta and the aortic root in SD compared to HFD-fed Il23r^gfp/+ Ldlr^−/− and Il23r^gfp/gfp Ldlr^−/− mice in the aortic root (Supplementary material online, Figure S1a and b). In the aortic root and aorta, ∼90% and ∼70% of GFP⁺ cells could be identified as γδ T cells, respectively; the remaining cells corresponded to conventional αβ T cells in both SD and HFD-fed Il23r^gfp/+ Ldlr^−/− mice (Figure 1B, Supplementary material online, Figure S2). γδ T cells amounted to ∼10% of leucocytes in the aortic root but only ∼2.5% of total CD45⁺ cells in the aorta in mice fed a SD or HFD (Supplementary material online, Figure S1c). Approximately 95% of γδ T cells present in the aortic root showed expression of IL-23R in contrast to ∼75% γδ T cells in the aorta, irrespective of diet (Supplementary material online, Figure S1d). IL23R⁺ γδ T cells were also frequently detected in the aortic root of Il23r^gfp/+Ldlr^−/− mice by multiphoton microscopy (Figure 1C). Additionally, we explored the cell types expressing Il23r transcripts by scRNA-seq (10× Genomics v3 reagents) and CITE-seq²⁵ of total viable cells in atherosclerotic aortas from Ldlr^−/− mice fed a HFD for 13 weeks. Il23r transcripts were exclusively detected in a cluster of γδ T cells defined as CD3⁺CD127⁺CD4⁻CD8a⁻NK1.1⁻ cells expressing T-cell receptor delta, constant region, T-cell receptor gamma, constant 1, and C-X-C chemokine receptor type 6 (Cxcr6) (Figure 1D and E). In previously generated datasets of aortic CD45⁺ leucocytes,^24,26,27Il23r transcripts were detected in a cluster of Cd3d⁺Cd4⁻Cd8b1⁻Cxcr6⁺ innate like T cells, consistent with γδ T cells, both in non-atherosclerotic and atherosclerotic aortas (Supplementary material online, Figure S3, transcripts encoding T-cell receptor delta and gamma chain were not detected in these data generated using 10× Genomics v2 reagents).

Deficiency in IL-23R entailed a loss in relative proportions of IL-23R⁺ cells in the aortic root both in SD and HFD-fed in Ldlr^−/−Il23r^gfp/gfp mice compared to frequencies in Ldlr^−/−Il23r^gfp/+ mice (Figure 1A). Residual IL-23R⁺ cells in Ldlr^−/−Il23r^gfp/gfp mice were conventional αβ T cells (Supplementary material online, Figure S1e). Taken together, these results show that substantially higher frequencies of IL-23R⁺ cells are found in the aortic root compared to the aorta. These IL-23R⁺ cells primarily constitute γδ T cells, which do not expand in atherosclerosis and require IL-23R for their maintenance.

3.2 IL23R⁺ γδ T cells in the atherosclerotic aortic root produce IL-17A and GM-CSF

We further characterized the cytokine expression profile of these vascular IL-23R-expressing γδ T cells in aortic root digests after in vitro stimulation with PMA, brefeldin and ionomycin by flow cytometry. As expected, the majority of IL23R⁺ γδ T cells produced IL-17A and to a lesser extent also GM-CSF. IFN-γ was expressed in single γδ T cells (0–3%) and mostly confined to IL23R⁻ αβ⁺ T cells that did not express IL-17 or GM-CSF (Figure 2A–C and Supplementary material online, Figure S2). Analysis of scRNA-seq data of aortic cells was consistent with these results, with γδ T cells being enriched for Il17a and rarely presenting detectable Ifng transcripts. GM-CSF (Csf2) was mostly enriched in adventitial type 2 innate lymphoid cells (ILC2s),²⁸ but transcripts were also detected in γδ T cells (Supplementary material online, Figure S3c and d). Upon HFD-feeding for 6 weeks, IL-17A expression was unaltered in IL-23R⁺ γδ T cells in Ldlr^−/−Il23r^gfp/+ mice and was increased in the few remaining IL-23R⁺ γδ T cells in Ldlr^−/−Il23r^gfp/gfp mice (Figure 2D). We also noted a significant increase in GM-CSF production in IL-23R⁺ γδ T cells in Ldlr^−/−Il23r^gfp/+ mice with HFD-feeding, and frequencies of GM-CSF⁺ IL23R⁺ γδ T cells were lower in diet-fed Ldlr^−/−Il23r^gfp/gfp compared to Ldlr^−/−Il23r^gfp/+ mice (Figure 2E). In line with the loss of IL-23R⁺ cells in Ldlr^−/−Il23r^gfp/gfp mice, frequencies of IL-17A-producing GFP⁺ γδ T cells among total leucocytes were markedly reduced in SD and HFD-fed IL-23R-deficient Ldlr^−/−Il23r^gfp/gfp mice compared to Ldlr^−/−Il23r^gfp/+ mice (Figure 2G), and GM-CSF-producing GFP⁺ γδ T cells were lower in HFD-fed Ldlr^−/−Il23r^gfp/gfp mice (Figure 2H). When separately analysing the capacity of γδ T cells to secrete both of these cytokine, we observed an expansion of IL-17A⁺GM-CSF⁺ double positive IL-23R⁺ γδ T cells in HFD-fed Ldlr^−/−Il23r^gfp/+ mice, and again a loss in IL-17A⁺GM-CSF⁺IL-23R⁺ γδ T cells among CD45⁺ leucocytes was seen in both SD and HFD-fed Ldlr^−/−Il23r^gfp/gfp mice (Figure 2J–L). There were no major changes in IFN-γ production within αβ or γδ T cells (Figure 2F), but increased frequencies of IFN-γ⁺ αβ⁺ T cells among total leucocytes were noted in HFD compared to SD-fed Ldlr^−/−Il23r^gfp/+ mice (Figure 2I). Aortic root IL23R⁺ γδ T cells thus produce IL-17A and GM-CSF and HFD-feeding increases the frequency of IL-17A⁺GM-CSF⁺ IL-23R⁺ γδ T cells, whereas there is a loss of IL-23R⁺ γδ T cells as a source of these cytokines in Ldlr^−/−Il23r^gfp/gfp mice in the vessel wall.

3.3 Deficiency in IL-23R inhibits early atherosclerotic lesion formation

Given that IL-23R⁺ γδ T cells were present in the aortic root and aorta, we assessed atherosclerotic lesion formation in Ldlr^−/− and Ldlr^−/−Il23r^gfp/gfp mice fed an HFD. IL-23R deficient Ldlr^−/−Il23r^gfp/gfp mice of both sexes showed a decrease in plaque area in the aortic root as compared to Ldlr^−/− mice after 6 weeks of HFD (Figure 3A and Supplementary material online, Figure S4a). After 12 weeks of HFD, no differences in plaque size were noted (Figure 3E). Lesion size in the aorta was unaltered at either time point (Supplementary material online, Figure S5a and b). When characterizing the plaque cell composition, no differences in the relative macrophage and SMC content were noted between Ldlr^−/− and Ldlr^−/−Il23r^gfp/gfp mice after both 6 and 12 weeks of HFD (Figure 3B, C, F and G). No differences in Ly6G⁺ neutrophils (3.2 ± 1.1 vs. 7.8 ± 2.8 cells/mm² plaque area in Ldlr^−/− vs. Ldlr^−/−Il23r^gfp/gfp mice, non-significant) were evidenced. No change in the relative numbers of αβ⁺ T cells but a significant decrease in γδ T cells were detected in the aortic root (Supplementary material online, Figure S5c). No changes in the mRNA expression of the Th1-inducing transcription factor Txb21, the Th17-associated transcription factor Rorc, and of Foxp3 were noted in the aortic root of Ldlr^−/−Il23r^gfp/gfp compared to Ldlr^−/− mice (Supplementary material online, Figure S5d). Notably, necrotic core size was decreased in Ldlr^−/−Il23r^gfp/gfp compared to Ldlr^−/− mice after 6 and 12 weeks of HFD (Figure 3D and H). However, no significant changes in TUNEL⁺ apoptotic cells could be evidenced (Supplementary material online, Figure S5e). Likewise, the ratio of free vs. macrophage-associated TUNEL⁺ cells, indicative of the rate of efferocytosis, was unaltered between the groups (Supplementary material online, Figure S5f). No differences in body weight or plasma cholesterol levels were noted at either time point (Table 1). These results suggest that IL-23R promotes early lesion formation in the aortic root and regulates necrotic core size.

3.4 IL23R deficiency systemically alters Treg homeostasis in hypercholesterolaemic mice

Given systemic inflammatory responses in atherosclerosis, we further analysed systemic immune cell distributions in Ldlr^−/− and Ldlr^−/−Il23r^gfp/gfp mice fed a HFD for 6 weeks. White blood cell counts, absolute numbers or percentages of neutrophils or monocytes did not differ between groups in the blood or spleen (Supplementary material online, Figure S6a–e). Likewise, frequencies of CD3⁺ T cells and of γδ T cells were similar in the spleen and lymph nodes of Ldlr^−/− compared to Ldlr^−/−Il23r^gfp/gfp mice (Supplementary material online, Figure S6f and g).

IL-23R⁺ γδ T cells have been shown to restrain Treg responses during experimental autoimmune encephalomyelitis.¹⁵ In line with these findings, a small but significant increase in frequencies of CD25⁺ Foxp3⁺ CD4⁺ Tregs were noted in spleens but not peripheral lymph nodes of Ldlr^−/−Il23r^gfp/gfp mice fed a HFD for 6 weeks, associated with increased splenic Foxp3 transcript expression (Supplementary material online, Figure S7a and b). Naïve αβ T cells lack IL-23R expression¹⁵ and we could not detect IL-23R on Tregs in Il23r^gfp/+ mice (data not shown). In line, Treg frequencies did not differ after culturing isolated Il23r^+/+ or Il23r^gfp/gfp naïve CD4⁺ T cells after stimulation with anti-CD3/CD28 antibodies and TGF-β upon exposure to recombinant IL-23 (Supplementary material online, Figure S7c). In total splenocytes exposed to TGF-β, however, increased percentages of Tregs were noted among Il23r^gfp/gfp compared to Il23r^+/+ splenocytes (Supplementary material online, Figure S7d), suggesting indirect effects of IL-23R⁺ cells on Tregs, as previously reported.²⁹

Levels of IFN-γ⁺-producing CD4⁺ Th1 cells were similar in lymphoid organs between Ldlr^−/− and Ldlr^−/−Il23r^gfp/gfp mice and no major differences in frequencies of IL-17A-producing CD4⁺ Th17 cells were noted in both groups in mice fed a HFD for 6 weeks (Supplementary material online, Figure S6h and i), indicating that deficiency in IL23R does not entail systemic shifts in Th1 or Th17 cell frequencies.

3.5 The absence of the global pool of γδT cells does not affect atherosclerotic lesion formation

Given our findings suggesting that effects of IL-23R are γδ T cell-mediated, we assessed the function of γδ T cells in atherosclerotic lesion formation using δ TCR-deficient Ldlr^−/−Tcrδ^−/− mice that lack all γδ T cells. Atherosclerotic lesion size was unaltered in the aortic root after 6 weeks of HFD (Figure 4A). No differences in plaque area were observed in the aortic root after 12 weeks of HFD (Figure 4B). Blood cholesterol levels were unchanged after 6 weeks (1258 ± 178 vs. 14 527 ± 1453 ± 259 mg/dL in Ldlr^−/− vs. Ldlr^−/−Tcrδ^−/− mice, P = 0.07) or 12 weeks of diet (1014 ± 386 vs. 1083 ± 328 mg/dL in Ldlr^−/− vs. Ldlr^−/−Tcrδ^−/− mice, P = 0.62). Plaque macrophage or smooth muscle cell content was not affected in the aortic root at either time point (Figure 4C and D and Supplementary material online, Figure S8a and b). However, relative necrotic core area was reduced in the aortic root of Ldlr^−/−Tcrδ^−/− mice after 12 weeks of HFD (Figure 4E). Levels of IFNγ and IL17A-producing CD4⁺ T cells were not altered in lymph nodes and spleen (Supplementary material online, Figure S8c and d). However, an increase in the percentage of Tregs was detected in the spleen of Ldlr^−/Tcrδ^−/− mice after 6 weeks of HFD (Supplementary material online, Figure S8e). These data show that γδ T cells restrain systemic Treg responses during atherogenesis, and promote plaque necrosis.

3.6 IL-23 is produced in the aortic root

In vitro studies demonstrated IL-23 production by activated dendritic cells and macrophages.³⁰ In scRNA-seq analyses of aortic cells, Il23a transcripts were poorly detected, but rare Il23a positive cells localized to e.g. mononuclear phagocytes and SiglecF⁺ neutrophils²⁴ (Supplementary material online, Figure S3c and d). To further investigate IL-23 production in antigen-presenting cells, BM-APCs were cultured and stimulated with different atherosclerosis-relevant stimuli. LPS but not native LDL (nLDL) or oxLDL alone induced IL-23 cytokine production in cultured cells within 24 h of stimulation (Figure 5A). In BM-APCs exposed to lipopolysaccharide (LPS), however, co-incubation of LPS with oxLDL but not nLDL further increased IL-23 production in a concentration-dependent manner (Figure 5A and B), suggesting that activation of TLR-4 triggers IL-23 expression in BM-APCs that is potentiated by additional oxLDL-mediated signalling. Of note, in Ldlr^−/− mice, an increased expression of Il23a mRNA was found specifically in the aortic root, but not in the aorta or spleen after 6 weeks of HFD compared to SD (Figure 5C). We also evaluated IL-23 production in tissue digests of SD and HFD-fed Ldlr^−/− mice, stimulated with LPS and oxLDL, by flow cytometry. While some intracellular IL-23 staining was seen in macrophages and dendritic cells in the aorta and spleen in both SD and HFD-fed mice, significantly increased frequencies of IL-23⁺ macrophages were detectable in the aortic root of HFD-fed vs. SD-fed Ldlr^−/− mice and compared to macrophages in the aorta (Figure 5D, Supplementary material online, Figure S9). These data clearly show that macrophages/dendritic cells serve as a source of IL-23 in the aortic root, further corroborating local functions of IL-23/IL-23R in promoting early inflammation in atherosclerosis.

3.7 Cytokines associated with IL-23R⁺ γδT cells induce macrophage necroptosis

IL-23 signalling, induced by GM-CSF, has previously been associated with the induction of macrophage apoptosis and plaque necrosis in Ldlr^−/−Csf2^−/− mice.¹⁸ By scRNA-seq, we could not detect Il23r transcripts in Csf1r⁺Adgre1⁺ macrophages and other aortic CD45⁺ cell populations (Figure 1C and D and Supplementary material online, Figure S3a and b). Similarly, cytometric analyses of the aortic root and aorta revealed no IL-23R expression in myeloid cells, including CD11c⁻F4/80⁺ or CD11c⁺MHCII⁺ APCs, or CD19⁺ B cells of Ldlr^−/−Il23r^gfp/+ mice (Supplementary material online, Figure S10, and data not shown). As IL-23R deficiency reduces necrotic core size in Ldlr^−/− mice, we assessed whether the cytokines released from IL23R⁺ γδ T cells could directly affect macrophage cell death. Treatment of BMDMs with IL-17A, GM-CSF, or both cytokines alone did not modulate the rate of apoptosis. In contrast, frequencies of necrotic BMDMs increased upon treatment with GM-CSF but in particular when exposed to both GM-CSF and IL-17A (Figure 6A). We further evaluated expression of key signalling molecules regulating apoptosis and necroptosis.³¹ Expression of apoptosis-associated total Caspase 3 and Caspase 8 did not change, and no cleavage of Caspase 3 or Caspase 8 was detectable (Supplementary material online, Figure S11). Cell death via the necroptotic pathway is induced by activation of the protein kinases RIPK3 and RIPK1, and the pursuant RIPK3-mediated phosphorylation of mixed lineage kinase domain-like (MLKL).³¹ mRNA expression of Ripk3 and Mlkl but not Ripk1 were increased in BMDMs treated with IL-17A and GM-CSF as compared to untreated BMDMs (Figure 6B), and increased RIPK1, RIPK3, and MLKL protein levels were noted in IL-17A and GM-CSF-treated BMDMs (Figure 6C), suggesting the induction of necroptosis. Necroptosis can function as a trigger of inflammation.³¹ Treatment of BM-APCs with IL17A and GM-CSF induced inflammatory iNOS-encoding gene Nos2, and inflammatory Tnfa, Il6, Nlrp3, Il1b, and Ccl2, whereas anti-inflammatory Il10 transcripts were reduced and Il18 expression was unaltered (Figure 6D). Accordingly, decreased Ifng and increased Il10 mRNA expression was noted in atherosclerotic aortic roots of Ldlr^−/−Il23r^gfp/gfp compared to Ldlr^−/− mice (Figure 6E). This could imply that IL-23R signalling in γδ T cells, via the production of IL-17A and GM-CSF, increases plaque necrosis and inflammation in the aortic root in atherosclerosis.

4. Discussion

IL-23R expression has previously been shown to characterize pathogenic Th17 cells,⁹ and to be expressed by γδ T cells that respond to IL-23 with the secretion of cytokines and the ability to antagonize Treg functions.¹⁵ γδ T cells, however, have been reported to have no major function in experimental atherosclerosis.^7,8 We, here demonstrate that IL-23R expression was mostly confined to γδ T cells in the aortic root and aorta. Whereas IL-23R⁺ γδ T cells were frequent in the aortic root, fewer cells were identified in the aorta. Mice overexpressing IL-23 have been shown to develop inflammation in the aortic sinus, and flow cytometric analyses demonstrated the presence of IL-23R⁺ CD3⁺CD4⁻ CD8⁻ T cells at this site.²¹ These were likely γδ T cells, as more recently suggested by immunofluorescence staining in Tcrd-H2BeGFP reporter mice.⁶ Similarly, we could observe IL-23R⁺ cells in the aortic root of Ldlr^−/−Il23r^gfp/+ mice by two-photon microscopy. It was suggested that these γδ T cells constitute a cell population similar to foetal thymus-dependent resident enthesal Vγ6⁺IL-23R⁺IL-17⁺ γδ T cells that can be found specifically at sites of high biomechanical stress and trigger enthesitis in mice.⁶ The cell population of aortic root γδ T cells may indeed comprise this Vγ6⁺ subset, as suggested by PCR profiling (unpublished observations). Aortic valves share histological similarities with peripheral entheses, including the presence of cartilage and chondrocytes where the valve joins the aorta and where the valve moves and flexes like entheses.²¹ In entheses, γδ T cells have been shown to be tethered and not motile, as opposed to lymph node or dermal γδ T cells.⁶ Also in parabiotic mice, dermal and lymph node-derived IL-17-producing Vg6⁺ γδ T cells were principally, but not exclusively, shown to be tissue-resident, suggesting that Vg6⁺ T cells locally adapt to specific sites of the body as resident cells while possibly retaining the capability to circulate between tissues.³² This may suggest that γδ T cells are a mostly tissue-resident cell population in the aortic root. Given the accumulation of γδ T cells in joint entheseal regions and the aortic valve and root, Reinhardt et al.⁶ proposed that the accumulation of γδ T cells in different enthesis-related tissues reveals a common IL-23-driven mechanism. Indeed, the presence of γδ T cells in the vasculature was IL-23R-dependent, as a marked reduction in aortic root γδ T cells was noted in mice deficient in IL-23R. The exact γδ T-cell subset in the aortic root vs. the aorta or lymphoid organs, and whether IL-23R has role in the embryonic recruitment or local maintenance of γδ T cells, however, remains to be established.

In entheseal organs permanently exposed to mechanical forces, microdamage may activate enthesis-resident γδ T cells to trigger the production of proinflammatory cytokines, such as IL-17A.^6,21 The role of IL-23R⁺ γδ T cells in the vasculature under homeostatic conditions is unclear. Previous reports demonstrated that IL-23 is important for the clearance of extracellular pathogens,³³ suggesting a role in homeostatic tissue defence. We here addressed the role of IL-23R and γδ T cells in atherosclerotic lesion formation. When fed a HFD for 6 weeks, IL-23R-deficiency reduced the size of early atherosclerotic lesions in the aortic root. In contrast, mice deficient in γδ T cells did not show differences in atherosclerotic lesion size. These findings suggest an atherogenic role of IL-23R⁺ γδ T cells in early lesion formation in the aortic root. The loss of this atherogenic γδ T-cell subset but also of potentially non-pathogenic IL23R⁻ γδ T cells may have resulted in a net unaltered plaque size in the aortic root in Ldlr^−/−Tcrδ^−/− mice. At more advanced stages of lesion formation, after 12 weeks of HFD, no effects on lesion size were noted in either Ldlr^−/−Il23r^gfp/gfp or Ldlr^−/−Tcrδ^−/− mice. This is in line with previous findings showing unaltered atherosclerotic lesion size in Ldlr^−/−Il23r^gfp/gfp mice after 10 weeks of HFD feeding,¹⁹ and in Apoe^−/−Tcrδ^−/− mice with mild hypercholesterolaemia (mice fed a SD for 18 weeks)⁸ or after 10 weeks of HFD-feeding.⁷ At later stages of lesion formation, however, IL-23 has been shown to protect from diet-induced atherosclerosis in Ldlr^−/− mice by maintaining the intestinal barrier and preventing pro-atherogenic bacteria expansion and metabolite production.²⁰ The at least partial loss of such systemic protective effects of IL-23 responses in the gut in Ldlr^−/−Il23r^gfp/gfp mice may have attenuated local atheroprotective effects derived from the loss of pathogenic IL-23R signalling in the aortic root also at earlier lesion stages, and contributed to the rather mild atherosclerosis phenotype.

Mice deficient in IL-23R as well as mice lacking γδ T cells developed plaques characterized by reduced relative necrotic core area, after both 6 and 12 weeks of HFD. IL-23R-deficient mice in addition showed reduced aortic inflammation, as suggested by increased expression of IL-10 but decreased levels of IFN-γ. GM-CSF has previously been shown to induce IL-23 in myeloid cells, which in turn increases apoptosis susceptibility of macrophages. Accordingly, GM-CSF-deficient Ldlr^−/− mice displayed reduced lesional necrotic core area and a decrease in IL-23 and IFN-γ expression.¹⁸ It may thus be conceivable that IL-23R⁺ γδ T-cell activation associated with GM-CSF secretion propagated local IL-23 production, in turn inducing macrophage apoptosis, and plaque necrosis and inflammation.¹⁸ It should be noted, however, that we could not detect IL-23R expression on vascular macrophages in vivo or in BMDMs in vitro, so that it remains to be elucidated whether IL-23 can directly affect macrophage cell death. We here observed that GM-CSF alone can trigger a mild increase in BMDM cell death, but exposure of BMDMs to GM-CSF together with IL-17 markedly increased BMDM necrosis but not apoptosis in vitro, and induced signalling pathways associated with necroptosis, including the expression of RIPK1, RIPK3, and MLKL. RIPK1 activates RIPK3, which subsequently phosphorylates MLKL to induce necroptosis.³⁴ Interestingly, RIPK1 is expressed in early-stage atherosclerotic lesions in humans and mice and drives atherosclerosis and inflammation by its ability to activate NF-κB and inflammatory cytokine production in diet-fed Apoe^−/− mice.³⁵ RIPK3-mediated necroptosis was demonstrated to play a crucial role in necrotic core formation and atherosclerosis.³⁶ Also, MLKL contributes to necrotic core formation in atherosclerosis.³⁷ Activation of RIPK1, RPK3, and MLKL by IL-17/GM-CSF may thus have contributed to IL-23R-inducd necrotic core formation and inflammation in our study. Of note, MLKL also has anti-atherogenic functions by limiting lipid accumulation in foam cells within plaques.³⁷ Increased MLKL expression in BMDMs exposed to IL-17/GM-CSF may thus promote their necroptotic cell death but also reduce foam cell formation independent of cell death, and both mechanisms may have contributed during early lesion formation in Ldlr^−/−Il23r^gfp/gfp mice. In addition, direct and cytotoxic effects of γδ T cells on macrophages may be conceivable, as activated γδ T cells can acquire cytotoxic activity to eliminate macrophages.^38,39

We here show that IL-23R⁺ γδ T cells have a strong propensity towards IL-17 production, and scRNA-seq analyses corroborated that IL23R⁺ γδ T cells were the main source of IL-17 production in the aorta in atherosclerotic Ldlr^−/− mice. In the few IL-23R⁺ γδ T cells that remained in the Ldlr^−/−Il23r^gfp/gfp mice, we did not observe a reduction in IL-17A secretion in SD and HFD-fed mice compared to Ldlr^−/−Il23r^gfp/+ mice, which may suggest alternative pathways that control IL-17A production in these cells in the absence of IL-23R signalling. Some IL23R⁺ γδ T cells also produced GM-CSF, which increased upon HFD feeding, and single γδ T cells expressing this cytokine were also found by scRNA-seq, although GM-CSF expression predominated in ILC2s. GM-CSF production by other cell type, such as ILC2s may thus have complemented IL-17 secretion from IL-23R⁺ γδ T cells to induce necroptosis and inflammation in atherosclerosis, in addition to the pathogenic IL-17⁺GM-CSF⁺IL23R⁺ γδ T cells. Notably, IL-23 together with IL-1β expression is indispensable for GM-CSF expression by central nervous system-infiltrating T cells,⁴⁰ and the combination of IL-23 together with IL-1β is required for GM-CSF production by γδ T cells in the absence of TCR stimulation in experimental autoimmune encephalomyelitis.¹⁷ The cytokine IL-1β is induced early in macrophages in diet-induced atherosclerosis.⁴¹ Both IL-1β and IL-23, together with other mediators induced by HFD-feeding, may thus have contributed to GM-CSF production by γδ T cells in atherosclerosis.

We did not observe aortic CD4⁺ T cells a major source of IL-17 in our study. In a prior study investigating diet-fed Ldlr^−/−Il23r^gfp/gfp mice, pathogenic Th17 cells comprised a small population of cells (0.05% of CD4⁺ T-helper cells) in lymphoid organs, which was unaffected by deficiency of IL-23 and undetectable in the aorta. Moreover, the transfer of CD4⁺ T-helper cells from Ldlr^−/−Il23r^gfp/gfp mice into Ldlr^−/−Rag1^−/− mice did not entail a reduction in lesion formation or change in plaque phenotype,¹⁹ furthermore supporting the notion that IL-23R⁺CD4⁺ Th17 cells do not contribute substantially to the phenotype observed in Ldlr^−/−Il23r^gfp/gfp mice in early atherosclerosis.

We also found that IL-23R deficiency was accompanied by a mild increase in frequencies of Tregs in spleens but not lymph nodes of HFD-fed mice, in line with previous observations in Il23r^gfp/gfp mice.^29,42,43 We did not detect significant changes in Foxp3 mRNA expression in the aortic root, but Il10 cytokine expression was elevated at this site. Given the atheroprotective effects of this cell subset⁴⁴ and the low abundance of Tregs in the vasculature,⁴⁵ differences in few cells may have gone unnoticed by evaluating mRNA expression, and Tregs may have contributed to local atheroprotective effects in Ldlr^−/−Il23r^gfp/gfp mice. However, increased frequencies of splenic Tregs were similarly observed in diet-fed Ldlr^−/−Tcrδ^−/− mice without alterations in atherosclerotic lesion formation, arguing against a major role of these mild systemic changes in Tregs on early atherosclerotic lesion formation in our study.

Interestingly, we noted increased Il23a expression in the aortic root but not aorta upon HFD-feeding, which could be in line with a local γδ T-cell response. In vitro, oxLDL potentiated and sustained the release of IL-23 from LPS-activated BM-APCs. Increased levels of Il23a in the aortic root could also be in line with the previously documented predominant localization of APCs in the aortic valve and sinus compared to the aorta.⁴⁶ In atherosclerosis, TLR activation by oxLDL or low grade endotoxemia^47,48 in addition to uptake of oxLDL may thus trigger IL-23 production by APCs to activate atherogenic IL-23R⁺ γδ T cells in the aortic sinus. GM-CSF secretion by γδ T cells may then further amplify IL-23 production in APCs¹⁸ in a positive feed-forward mechanism. Another source of IL-23 are neutrophils.⁴⁹ Neutrophils, regarded as the first line of defence against infection, are more abundant in early atherosclerotic lesions compared to more advanced plaques⁵⁰ and via IL-23 secretion could work together with IL-23R⁺ γδ T cells to instigate inflammation in the vessel wall. After stimulation of vascular tissue digests or splenocytes with LPS/oxLDL ex vivo, however, we could not detect intracellular IL-23 in neutrophils, but this may relate to suboptimal stimulation conditions for this cell type. IL-23 production was mostly confined to macrophages and some dendritic cells in this setting. Reporter mouse models will be useful to in vivo trace cellular and temporal IL-23 expression patterns in atherosclerosis in the future.

In past years, several studies have associated IL-23 with cardiovascular diseases, such as peripheral artery disease⁵¹ and carotid atherosclerosis.⁵² Based on our findings one may speculate that IL-23 affects local plaque formation and plaque necrosis. Given that large necrotic cores are a hallmark of vulnerable plaques,⁵³ IL-23 may contribute to plaque instability and in consequence adverse cardiovascular events. Different clinical trials using antibodies against p19 (unique to IL-23) are currently being tested for the treatment of psoriasis. It will be interesting to monitor whether p19 blockade reduces the increased risk of cardiovascular events in these patients. Interestingly, p19 subunits are abundant in Giant-cell arteritis,⁵⁴ and targeting IL-12/IL-23 was shown to be effective for the treatment of this disease.⁵⁵ Given the location of IL-23R⁺ γδ T cells in the vessel wall, our findings could also be relevant to this pathology.

In summary, we demonstrate that IL-23R⁺ γδ T cells are present in the aortic root and express IL-17 and GM-CSF. IL-23R⁺ γδ T cells play an important role in locally promoting lesion formation, plaque necrosis and inflammation in the aortic root. Targeting the IL-23/IL-23R-axis could thus be further explored as a therapeutic option for the inhibition of early atherosclerotic lesion formation.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

J.G.-P., N.A., I.J., H.D.M., A.B., A.-E.S., K.L., L.B., I.P., A.W., T.K., C.C., and A.Z. designed the project and experiments; J.G.-P., N.A., I.J., H.D.M., M.R., T.B., P.C., S.S., and C.C. performed the experiments and analysed the results. J.G.-P. and A.Z. drafted the original manuscript. All authors provided critical input, critically reviewed the article, and contributed to discussion and interpretation of results.

Acknowledgements

We thank Petra Hönig-Liedl for expert technical assistance, and Nina DiFabion, Tobias Krammer, and Panagiota Arampatzi for technical assistance with single-cell RNA-seq library preparation, sequencing, and pre-processing. Illustrations of cell types in the graphical abstract are based on graphics from Servier Medical Art.

Funding

This project was supported by the Deutsche Forschungsgemeinschaft (374031971—TRR 240, 324392634—TR221, ZE827/13-1, 14-1, 15-1, and 16-1 to A.Z.; PR727/11-2 and 13-1 to I.P.; 236177352- SFB 1116, TPA09 to K.L.; TRR156, CRC1292 and Reinhart Koselleck grant to A.W.; TRR128, TRR274, SFB1054, and the Synergy excellence cluster EXC 2145, ID 390857198 to T.K.; CO1220/1-1 and CO1220/3-1 to C.C.), the Interdisciplinary Center for Clinical Research (IZKF), University Hospital Würzburg (E-352 and A-384 to A.Z., and E-353 to C.C.), the BMBF within the Comprehensive Heart Failure Centre Würzburg (BMBF 01EO1504 to C.C.). T.K. receives funding from the ERC (CoG 647215).

Data availability

Single-cell RNA sequencing data are accessible in Gene Expression Omnibus under the accession number GSE182950 (Figure 1) and GSE97310 (previously published data from Cochain et al. Circ Res 2018). Other raw data underlying this article are available upon request.

References

Author notes