Thymic stromal lymphopoietin is a key cytokine for the immunomodulation of atherogenesis with Freund's adjuvant

Abstract Adaptive immune responses regulate the development of atherosclerosis, with a detrimental effect of type 1 but a protective role of type 2 immune responses. Immunization of Apolipoprotein E‐deficient (ApoE−/−) mice with Freund's adjuvant inhibits the development of atherosclerosis. However, the underlying mechanisms are not fully understood. Thymic stromal lymphopoietin (TSLP) is an IL7‐like cytokine with essential impact on type 2 immune responses (Th2). Thymic stromal lymphopoietin is strongly expressed in epithelial cells of the skin, but also in various immune cells following appropriate stimulation. In this study, we investigated whether TSLP may be crucial for the anti‐atherogenic effect of Freund's adjuvant. Subcutaneous injection of complete Freund's adjuvant (CFA) rapidly led to the expression of TSLP and IL1β at the site of injection. In male mice, CFA‐induced TSLP occurred in immigrated monocytes—and not epithelial cells—and was dependent on NLRP3 inflammasome activation and IL1β‐signalling. In females, CFA‐induced TSLP was independent of IL1β and upon ovariectomy. CFA/OVA led to a more pronounced imbalance of the T cell response in TSLPR−/− mice, with increased INFγ/IL4 ratio compared with wild‐type controls. To test whether TSLP contributes to the anti‐atherogenic effects of Freund's adjuvant, we treated ApoE−/− and ApoE−/−/TSLPR−/− mice with either CFA/IFA or PBS. ApoE−/− mice showed less atherogenesis upon CFA/IFA compared with PBS injections. ApoE−/−/TSLPR−/− mice had no attenuation of atherogenesis upon CFA/IFA treatment. Freund's adjuvant executes significant immune‐modulating effects via TSLP induction. TSLP‐TSLPR signalling is critical for CFA/IFA‐mediated attenuation of atherosclerosis.


| INTRODUC TI ON
Atherosclerosis is a chronic inflammatory disease. 1 It is characterized by the complex immunological interaction of residential vascular cells and professional immunocytes. Monocytes enter the lesion area and differentiate into dendritic cells or macrophages that incorporate, process, present and are activated by oxidized low-density lipoproteins (oxLDL) and other antigens. Whereas classical monocytes may be responsible only for the perpetuated pro-inflammatory response, non-classical monocytes and type 2 macrophages have also been linked to repair and reconstitution at the site of inflammation. 2 T cells are activated in peripheral lymphoid organs but also adhere to and migrate into arterial lesions.
CD4-positive T cells react to antigen presentation by dendritic cells and in return, foster the activation of immune (eg macrophages) and vascular cells, mainly through cytokine production. T helper (Th) type 1 cells mostly produce interferon (IFN)-γ and TNF and promote atherogenesis, whereas Th2 cells mostly produce type 2 cytokines IL(interleukin)4, IL5 and IL13, which are mainly associated with anti-atherogenic properties. 3 Manipulation of these immunological interactions may have important therapeutic consequences as suggested by various studies 4

like recently CANTOS (Canakinumab Anti-inflammatory
Thrombosis Outcomes Study), in which the alteration of chronic inflammation with the anti-IL1β antibody Canakinumab reduced atherosclerosis independently and was introduced as an alternative therapy beyond lipid lowering. 5 Immunization of Apolipoprotein E-deficient (ApoE −/− ) mice with immuno-adjuvants in combination with low-density lipoprotein (LDL)-derived antigens reduces the progression of atherosclerotic lesions. 6 Interestingly, the repetitive administration of the water-oil-based Freund's adjuvant without additional antigens can also slow down atherogenesis, though to a lesser degree than the adjuvant/antigen regimen. 6,7 The underlying mechanisms have not been fully understood. Interestingly, however, the anti-atherogenic effect of Freund's adjuvant is lost in mice depleted of CD4 T cells. 6 We hypothesized that local processes at the site of injection may be pivotal in the induction of the atheroprotective immune response.
Thymic stromal lymphopoietin (TSLP) is an IL7-like cytokine that is crucial for the induction of type 2 immunity. 8 Its cellular sources are mainly epithelial cells in the skin (eg keratinocytes), but also murine plasmacytoid and especially myeloid dendritic cells, or human monocyte-derived dendritic cells. 9 The TSLP receptor consists of IL7Rα and a TSLP-specific subunit and is expressed on innate and adaptive immune cells. Recent studies confirm its role as a master regulator for Th2 inflammatory responses in allergic diseases. In B lymphocytes, TSLP promotes their maturation to the B220/IgM stage; in T cells, it directly acts on proliferation and survival and can directly induce IL4 production in naïve CD4 T cells. 10,11 In this study, we investigated, whether TSLP is crucial for the reduction of atherosclerosis by Freund's adjuvant.

| RNA isolation and reverse transcription
For isolation of RNA from mouse skin, explanted parts (approx.

| Histology
Skin samples were directly frozen in OCT at −20° for 1 hour and then transferred to −80°C. Transversal 10 µm sections were thawed and fixed with acetone at 4°C for 10 minutes before further staining according to standard protocols. Briefly, the samples were blocked with 10% FCS for 30 minutes and then incubated with the primary antibody (anti-TSLP, Santa Cruz or anti-MOMA2, Acris) for 120 minutes at room temperature. After washing with PBS, they were incubated with a secondary antibody for 60 minutes. Nuclei were counterstained with DAPI. Hearts were stored in 4% paraformaldehyde at 4°C. Before inclusion in a cutting medium and further storage at −80°C, hearts were treated with 10% sucrose in phosphate buffered saline for 24 hours. Successive 10 µm transversal sections of aortic sinus were obtained. Lipids were detected using Oil Red O staining, monocytes/macrophages with anti-MOMA2 (Acris).

| ELISpot and ELISA
A total of 500 000 spleen cells per well were used, and all samples were used in triplicates. Cells were cultured in RPMI 1640 supplemented with Glutamax, 10% FCS, 0.02 mmol/L β-mercaptoethanol and antibiotics Penicillin and Streptomycin (referred to as "complete" RPMI

| Statistical analysis
Data are expressed as mean ± SEM. Data were compared, and intergroup differences were analysed with one-way ANOVA and post hoc Tukey's test. Other data were analysed by a two-tailed Student's t test. Differences were considered statistically significant when the probability value was ≤.05.

| Freund's adjuvant induces the expression of TSLP at injection site in C57bl/6J wild-type and ApoE −/− mice
Atherosclerosis can be modified by the administration of selected immuno-adjuvants in combination with specific antigens, but also if only adjuvants are applied. Hence, we initially investigated whether TSLP is expressed upon the injection of adjuvants and/or immunogenic antigens in wild-type mice. Tang et al suggested that the adjuvant papain may induce TSLP mRNA in mouse ears at the site of injection peaking 12 hours after injection. 12 We did not find a significant increase of TSLP upon papain expression 12 hours after subcutaneous injection in the back skin. However, equal volumes of complete Freund's adjuvant-induced TSLP, whereas the immunogenic antigens oxidized LDL (oxLDL) or malondialdehyde-modified LDL (MDA-LDL) did not ( Figure 1A). Thus, we compared several immuno-adjuvants with known impact on atherogenesis. Freund's adjuvant (complete and incomplete) and Alum, previously shown to limit atherogenesis, 6,7 likewise induced TSLP, but CpG 1668 -oligodeoxynucleotides, used as an adjuvant for vaccination, which promotes atherogenesis, 13 had no effect on TSLP expression ( Figure 1B).
Among possible TSLP-inducing cytokines, only IL1β was significantly induced at the injection site in response to CFA ( Figures S1 and S2).
Further, a kinetic analysis revealed a peak at 12 hours for both TSLP ( Figure 1C) and IL1β ( Figure 1D). We applied the same model to ApoE −/− mice and found that TSLP was also strongly induced in their skin 12 hours upon s.c. injection of CFA ( Figure 1E).

| Inflammasome-dependent IL1β signalling induces TSLP in the skin
IL1β and TSLP were both induced upon CFA injection and peaked at 12 hours. When analysed at a much earlier time-point (4 hours after injection), IL1β was mildly, but statistically significantly increased upon CFA, whereas TSLP was not ( Figure S3). In order to understand whether those cytokines mutually influenced each other, we tested CFA in IL1β −/− -or TSLPR −/− mice. IL1β deficiency abolished CFAdependent TSLP induction (Figure 2A). Interestingly, we found that this effect was strictly gender-dependent, because female IL1β −/− mice still expressed TSLP unless they were ovariectomized ( Figure   S4A-C). TSLPR deficiency had no impact on IL1β induction by CFA ( Figure 2B).
Next, we sought to evaluate the pathway upstream of IL1β expression. CFA-induced TSLP expression was abrogated in NLRP3 −/− mice, upstream of IL1β production ( Figure 2C). Thus, NLRP3 inflammasome-dependent IL1β generation is essential for TSLP expression in response to CFA.    Figure 3C).

| Monocytes significantly contribute to TSLP expression at injection site
To confirm a significant role for monocytes in TSLP production, we depleted them with intravenous clodronate, and then subcutaneously injected CFA or PBS. CFA-induced TSLP expression was significantly reduced in the absence of both classical and non-classical monocytes ( Figure 4A-C). Interestingly, however, CFA-induced TSLP was not altered in CCR2 deficient mice ( Figure 4D), suggesting a dominant role for non-classical monocytes in TSLP production.

| TSLP signalling modulates type 2 immune responses upon immunization with Freund's adjuvant in males and females
The predominant type of inflammation in atherosclerosis features an INFγ driven, monocyte/macrophages type 1 inflammatory phenotype. Immunization models in atherosclerosis are associated with a change of this polarization and skewing towards a stronger type 2 immune response. Thus, we examined whether TSLP is relevant for the type of the immune response upon immunization with chicken Ovalbumin emulsified in CFA, and an in vitro recall 2 weeks later.
Male and female wild-type mice almost equally increase IL4 production by splenocytes upon OVA recall ( Figure 5A). Interestingly, IL4 production is still increased in male TSLPR −/− mice, whereas female knockout mice did not significantly respond to the antigen challenge in vitro ( Figure 5B). To understand, whether TSLP-TSLPR signalling is crucial for the immunization response to CFA/OVA not only in female but also in male mice, we repeated the experiment and analysed the response for IL4 and IFNγ. There we found, that-despite an abso- proportionally less Th2 and more INFγ Th1 type immunity in male TSLPR −/− mice compared with wt ( Figure 5F).  (Figure 6). We found this result consistently in male and female ApoE −/− mice, with expectedly bigger plaque burden in females ( Figure 6).

| D ISCUSS I ON
Directing inflammation in order to benefit atherosclerosis is a promising approach with high potential for translation from mice to humans. In mouse models, treatment with some immune adjuvants restrains the development of atherosclerotic plaques, 7 although the mechanisms behind this protection have not been addressed in de-

tail. Given previous indications that atheroprotection in response to
Alum and Freund's adjuvants was associated with a shift towards a type 2 immune response, 14,15 we set out to address the underlying mechanisms.
In our study, we show that Freund's and Alum, known atheroprotective adjuvants, induce the expression of TSLP at the injection The mechanisms of CFA-induced TSLP differ in males and females.
While the mechanisms responsible for TSLP induction in females are still poorly understood, CFA-induced TSLP requires NLRP3 inflammasome-dependent IL1β production in males. Although Alum particulates have been reported to activate NLRP3 inflammasome, 16 the impact of Freund's adjuvant on NLRP3 expression or activation may differ. 16,17 Importantly, the mycobacterial component in CFA is not needed for the expression of TSLP, since IFA has the same effects in our model. This is also consistent with the fact that MINCLE, the receptor for mycobacterial cord factor was not required for CFAinduced TSLP ( Figure S2). in DCs in vitro, 21 and in a recent study, myeloid cells, including neutrophils and monocytes, have been shown to produce TSLP in response to IL1α, a mechanism that was responsible for breast cancer spreading. 22 When we tested the relevance of TSLP/TSLPR signalling in immunization with CFA through antigen exposure, we found that TSLP was crucial to balance IFNγ and IL4 production. TSLPR deficiency in females abrogated IL4 production upon (re)exposition of total splenocytes to OVA, and even though IL4 was still produced to some extent in the absence of TSLPR in males, IFNγ production was dramatically boosted leading to a significant increase of IFNγ/IL4 ratio, and a likewise increase of IgG2c/IgG1 ratio. Thus, our data identify a critical role for TSLP in the modulation of the immune response to Freund's adjuvant. The role of TSLP in promoting a type 2 immune response is consistent with previous observations. TSLP production by DCs fosters a Th2 polarization when interacting with T cells, 23 and TSLP has been shown responsible for the Th2 polarizing capacity of cysteine-protease papain. 12 TSLP has been used itself as an mucosal immuno-adjuvant by van Roey and colleagues, who showed a significant on inflammatory response skewing towards a Th2 cytokine profile. 24 TSLP, together with IL25 or IL33, is a regulator of the proliferation and maturation of group 2 innate lymphoid cells pro-atherosclerotic role for TSLP, because ApoE −/− TSLPR −/− mice on a high fat diet had less atherosclerosis than ApoE −/− controls. 26 Another group described a strong expression of TSLP in human atherosclerotic plaques. 27 On the contrary, Yu and co-workers reported an attenuation of atherosclerosis following TSLP administration in mice. 28 Our data support a protective role of TSLP/TSLPR signalling in a specific setting of induction of type 2 immune responses.

ACK N OWLED G EM ENTS
Thanks to M. Houssari for his technical contributions to parts of the project.

CO N FLI C T O F I NTE R E S T
The authors have no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study included in the study or available from the corresponding author, MS, upon reasonable request.