IL-1β and TNF-α Impact Abdominal Aortic Aneurysm Formation by Differential Effects on Macrophage Polarization

Abstract

Objective

Abdominal Aortic Aneurysms (AAAs) are inflammatory in nature and are associated with some risk factors that also lead to atherosclerotic occlusive disease, most notably smoking. The purpose of our study was to identify differential cytokine expression in AAA patients and those with atherosclerotic occlusive disease. Based on this analysis, we further explored and compared the mechanism of action of IL-1β vs TNF-α in AAA formation.

Approach and Results

IL-1β was differentially expressed in human plasma with lower levels detected in AAA patients compared to matched atherosclerotic controls. We further explored its mechanism of action using a murine model and cell culture. Genetic deletion of IL-1β and IL-1R did not inhibit aneurysm formation or decrease MMP expression. The effects of IL-1β deletion on M1 macrophage polarization were compared to another pro-inflammatory cytokine, TNF-α. Bone marrow-derived macrophages from IL-1β^−/− and TNF-α^−/− mice were polarized to an M1 phenotype. TNF-α deletion, but not IL-1β deletion, inhibited M1 macrophage polarization. Infusion of M1 polarized TNF-α^−/− macrophages inhibited aortic diameter growth; no inhibitory effect was seen in mice infused with M1 polarized IL-1β^−/− macrophages.

Conclusions

While IL-1β is a pro-inflammatory cytokine, its effects on aneurysm formation and macrophage polarization differ from TNF-α. The differential effects of IL-1β and TNF-α inhibition are related to M1/M2 macrophage polarization and this may account for the differences in clinical efficacy of IL-1β and TNF-α antibody therapies in management of inflammatory diseases.

Introduction

Abdominal aortic aneurysm (AAA) is a common disease affecting 5% of males over the age of 65 and results in 15,000 deaths annually.¹ The disease is slowly progressive ultimately leading to rupture if not treated operatively. Most abdominal aortic aneurysms are asymptomatic until rupture, which leads to death in greater than 65% of patients.² Although the initiating events are not well understood, it has been established that inflammation is a major contributing factor to aneurysm progression.³ Patient studies and animal models have been used to better understand the role of inflammation in aneurysm development and progression. One widely used model is the calcium chloride model (CaCl₂), which recapitulates features of human AAA including inflammatory cell invasion, matrix metalloproteinase (MMP) upregulation, matrix degradation, and aneurysm formation.⁴ Aneurysms induced by CaCl₂ are an accepted and reliable model for chemical induction of AAAs.^5–7

In order to understand how the inflammatory process can led to aneurysm formation, we assessed the differential cytokine production by inflammatory cells from AAA patients and a matched atherosclerotic group with normal aortic diameter. Interleukin 1 beta (IL-1β) was one of the few cytokines differentially expressed. Both IL-1β and tumor necrosis factor alpha (TNF-α) have been associated with the progressive inflammatory process that promotes aneurysm progression.^8,9 The inflammasome NLRP3 has been implicated in AAA development following induction of Angiotensin II and results in IL-1β release from macrophages.^10,11 Interestingly, IL-1β blockade has been shown to have equivocal effects on atherosclerosis in a murine model.¹² Given the presumed pro-inflammatory nature, these cytokines are putative targets for preventing aneurysm progression.

IL-1β and TNF-α have long been considered among the most important factors promoting the inflammatory process in diseases such as Rheumatoid Arthritis, Crohn’s Disease and Ulcerative Colitis.^13–15 Pharmaceutical companies have devoted significant time and research into developing inhibitors of these and other inflammatory cytokines to help manage these diseases. However, relatively few studies directly compare their efficacies.¹⁶ The use of IL-1β inhibitors in more common pathologies, such as Rheumatoid Arthritis, have not been as effective as anticipated (ClinicalTrials.gov Identifier: NCT00424346). IL-1β inhibitors are effective in a small subset of specific diseases.^17,18 TNF-α blockade has emerged as a more effective clinical tool.¹⁹ The reason for these differences in clinical effectiveness of these two cytokine inhibitors is not well understood.

Johnston et al. has shown that genetic and pharmacological inhibition of IL-1β in the elastase infusion murine model protects against AAA development.⁸ This led to a recent clinical trial using an IL-1β antibody to inhibit the expansion of small AAA in humans (ClinicalTrials.gov Identifier: NCT02007252). Based on the differential IL-1β expression, we further assessed circulating levels of IL-1β. These results led us to evaluate the role of IL-1β deletion in the CaCl₂ model of AAA. Unexpectedly, IL-1β deletion had a very distinct effect compared to TNF-α, another prototypical pro-inflammatory cytokine. We further investigated the mechanism of this differential effect by studying macrophages from IL-1β^−/− and TNF-α^−/− null mice.

Materials and Methods

Materials and Methods are available in the online-only Data Supplement.

Results

IL-1β Differentiated between AAA and controls

Abdominal aortic aneurysms are inflammatory in nature with cytokines IL-1β and TNF-α believed to promote aneurysm progression. Evidence for the potential roles of IL-1β includes presence in aneurysm tissue and murine data suggesting that it promotes aneurysm development. Using high sensitivity ELISA, IL-1β was not detectable in the plasma of AAA patients (n = 9), but it was detectable in most samples (7/10) in the control group (n = 10) (Figure 1). Fisher exact test comparing the AAA vs control patients demonstrated a P value of 0.003 (Table 2). Levels of IL-1β were low in control patients, but the difference remained statistically significant (t test P = 0.015). While both atherosclerosis and abdominal aneurysms are associated with systemic inflammation, these data demonstrate IL-1β levels differentiate between patients with these two disease processes.

Figure 1. IL-1β levels in the plasma of control and AAA patients.

Blood samples were collected and compared from controls (n = 10) with normal aortic diameter to AAA patients (n = 9). The concentration of IL-1β in the plasma of control and AAA patients were determined using an ELISA assay as shown the bar graph. IL-1β was undetectable in the plasma of AAA patients. Fisher exact test was conducted (P = .003). Levels of IL-1β were low in control patients, but the difference remained statistically significant (t test P = 0.015). ND = Not Detectable

Table 2.

IL-1β levels in the plasma of control and AAA patients

	Non-Detected	Detected	p-value
Control	3	7	0.003
AAA	9	0

Blood samples were collected and compared from controls (n = 10) with normal aortic diameter to AAA patients (n = 9). The concentration of IL-1β in the plasma of control and AAA patients were determined using an ELISA assay as shown the bar graph. IL-1β was undetectable in the plasma of AAA patients. Fisher exact test was used.

AAA formation was enhanced by IL-1β inhibition

We used both IL-1R^−/− and IL-1β^−/− mice to study the effects on aneurysm formation using a murine AAA model established in our laboratory. Neither the IL-1R^−/− nor IL-1β^−/− showed inhibition of aneurysm formation. The IL-1R^−/− mice developed larger aneurysms compared to their WT controls (P < 0.05) (Figure 2). This was unexpected given previously published results in other murine AAA models.⁸ Histological imaging also demonstrated similar levels of lamellar destruction in all CaCl₂-treated mice (Figure 3A). Quantitative data (Figure 3A – bar graph) shows percentage of aortic lamellar degradation. IL-1R^−/− had significantly higher levels of lamellar fragmentation compared to the WT. Zymography demonstrated similar levels of expression of both latent and active MMP-9 and MMP-2 (Figure 3B). The analysis included 5–10 samples in each group and a representative gel is illustrated. Aortic wall thickness was also assessed, but was not different among the CaCl₂-treated mice. There was a significant difference in aortic wall thickness between the NaCl- and CaCl₂-treated mice (P < 0.05) (Supplemental Materials).

Figure 2. The effects of IL-1R^−/− and IL-1β^−/− on aortic diameter.

Relative changes of aortic diameter were compared among wild-type (WT) mice (NaCl, n = 7; CaCl₂, n = 15), IL-1R^−/− mice (NaCl, n = 9; CaCl₂, n = 11), and IL-1β ^−/− mice (NaCl, n = 17; CaCl₂, n = 15). Aneurysm induction was performed using the described protocol with CaCl₂ and NaCl. Student t-test was used for analysis. *P <0.05 compared to NaCl-treated controls; #P < 0.05 compared to CaCl₂-treated WT

Figure 3. Histologic changes and MMP expression.

(A) Aortic histologic changes after NaCl and CaCl₂ treatment in WT, IL1R^−/−, and IL-1β^−/− mice. VVG staining of aortic tissue reveals aortic lamellae disruption and elastic fiber fragmentation in the CaCl₂-treated mice. Quantitative data (right - bar graph) shows % aortic lamellar degradation. IL-1R^−/− had significantly higher levels of lamellar fragmentation compared to the WT. Student t-test was used for analysis. *P < 0.05 compared to CaCl₂-treated WT mice. (B) Representative Gelatin zymogram: Aortic samples (n = 5–10 per group) were obtained from WT, IL-1R^−/−, and IL-1β^−/− mice. MMP-2 and MMP-9 expression were quantified (right - bar graph) using ImageJ software. ANOVA was used for the analysis. There was no significant difference in MMP-2 and MMP-9 expression among the three CaCl₂-treated groups. *P < 0.05 compared to NaCl-treated WT mice

IL-1β and TNF-α exhibited different effects on macrophage activation

IL-1β and TNF-α have traditionally been considered among the most potent pro-inflammatory cytokines. Our previous work demonstrated that TNF-α inhibition prevents aneurysm formation, protects from matrix degradation, and blocks MMP expression partly by blocking macrophage activation.⁹ We realized that we could use our aneurysm model to better understand the disparate effects of these two important cytokines. We investigated the effects of each on macrophage polarization. BMDMs from IL-1β^−/− and TNF-α^−/− mice were treated with LPS and IFN-γ in order to promote polarization to the M1 phenotype. Quantitative RT-PCR analysis of macrophage gene expression demonstrated elevated M2 marker expression (IL-10, CD206) in TNF-α^−/− macrophages compared to WT macrophages (Figure 4A). IL-10 and CD206 were not increased in IL-1β^−/− macrophages when exposed to LPS and IFN-γ. M1 markers were not different among the groups. These data demonstrated important differences in the macrophage response to IL-1β and TNF-α deletion. We have previously demonstrated the importance of M1 polarization to aneurysm formation in the CaCl₂ model.²⁰

Figure 4. Macrophage markers expression.

(A) Quantitative RT-PCR. Bone marrow-derived macrophages (BMDM) were obtained from C57Bl/6, IL-1R^−/−, IL-1β^−/−, 129/SvEv and TNF-α^−/− mice. The macrophages were treated without (untreated) or with LPS and IFN-γ (LPS + IFN-γ) for 24 hours and total RNA and cell lysates were isolated. Quantitative RT-PCR was performed to measure relative expression of the M1 (iNOS, TNF-α) and M2 (IL-10, CD206) macrophage markers. (B) ELISA were performed with cell lysates from LPS and IFN-γ-treated BMDM. ANOVA was used for the analysis. *P < 0.05 compared to M0 BMDM, # P < 0.05 compared to WT M1 macrophages. § P < 0.01 compared to WT M1 macrophages. ND = not detectable.

Because of the unexpected results in the IL-1β^−/− mice, we performed an ELISA assay of macrophages derived from each of the mouse strains used for this study. IL-1β was not detectable in the cell culture media from any of the mice. Macrophage cell lysates from the IL-1β^−/− mouse confirmed lack of IL-1β expression. However, there was a compensatory increase in IL-1β levels in the IL-1R^−/− mice compared to the expression in the WT mice. There was no increase in IL-1β in the TNF-α^−/− mice (Figure 4B). The levels found in the cell lysate would reflect total, rather than secreted, IL-1β.

Activated macrophages from IL-1β^−/− and TNF-α^−/− mice demonstrate converse effects on aortic dilation

Given effects observed on ex vivo experiment described above, we next studied the effects of infusing activated macrophages into WT mice. We assessed the effects at the three-day and six-week time point. The three-day time point was chosen based on previous observations demonstrating high levels of M1 marker expression in the aorta at that time point, whereas the six-week time point was chosen to monitor long-term effect on aortic expansion. BMDMs from IL-1β^−/− and TNF- α^−/− mice were polarized to an M1 phenotype with LPS and INF-γ. Cells were given intravenously via tail vein injection prior to aneurysm induction. In order to determine if the infused macrophages reached the aortic tissue, cells were fluorescently labeled and aortic tissue homogenates were assessed on day three by flow cytometry. We found that 0.05% of all aortic cells were CFSE positive demonstrating aortic infiltration by infused macrophages. Infusion of activated macrophages from IL-1β^−/− mice resulted in aortic dilation similar to mice infused with M1-activated macrophages from WT mice (Figure 5A). In contrast, mice infused with activated macrophages from TNF-α^−/− mice developed smaller aneurysms at three days and six weeks (P < 0.05) (Supplemental Materials and Figure 5, respectively). This partial inhibition is consistent with our observations of increased M2 marker expression by activated macrophages from TNF-α^−/− mice (Figure 4A).

Figure 5. Aortic changes 6 weeks after CaCl₂ treatment in mice infused with untreated (M0 Mϕ infusion) or LPS & IFN-γ–treated (M1 Mϕ infusion) BMDM from C57Bl/6, IL-1R^−/−, IL-1β^−/−, 129/SvEv, or TNF-α^−/−.

Aortic diameters were measured before CaCl₂ incubation and at sacrifice (n = 3). (A) Aortic diameter increases are shown in the bar graph; (B) VVG staining of aortic tissue from WT mice infused with M0 Mϕ or M1 Mϕ from C57Bl/6, IL-1R^−/−. IL-1β^−/−, 129/SvEv. or TNF-α^−/− mice. Student t-test was used for analysis. *P < 0.05, compared to M0 Mϕ infused, # P < 0.05, compared to WT M1 Mϕ infused.

Discussion

Inflammation is the hallmark of aneurysm tissue with T-cells and macrophages being predominant. Macrophages exist in either a pro-inflammatory (M1) or anti-inflammatory (M2) state. We have previously shown that M1 macrophage polarization is critical to aneurysm formation in the CaCl₂ model.²⁰ Usui et al. demonstrated that angiotensin II infusion stimulated the inflammasome NLRP3 and subsequently led to the release of IL-1β, IL-6, and CCL2 from macrophages. This was mediated through mitochondria-derived reactive oxygen species resulting in AAA development.^10,11 The authors suggest that blocking NLRP3 inflammasome could be a useful strategy because it downregulates a host of inflammatory mediators. Our data suggests this effect is mediated through cytokines other than IL-1β. The differential expression we observed in cytokine expression between AAA and control patients prompted us to further investigate the role of IL-1β in the CaCl₂ murine model of AAA. Unexpectedly, we found that targeted deletion of IL-1β or its receptor, did not block aneurysm formation. Given the potential role that IL-1β can play in macrophage activation, we investigated the effects of its deletion on macrophage polarization in comparison to another pro-inflammatory cytokine, TNF-α. When macrophages were polarized to an M1 phenotype, TNF-α^−/− macrophages expressed higher levels of M2 cytokines in contrast to IL-1β^−/− macrophages. Consistent with these results, infusion of M1 polarized TNF-α^−/− macrophages inhibited aneurysm formation; M1 polarized IL-1β ^−/− macrophages exhibited similar effects to M1 polarized WT macrophages showing no inhibition in AAA progression. Taken together, these data demonstrate that IL-1β deletion does not block M1 macrophage polarization. This may explain why IL-1β inhibition has been less efficacious compared to TNF-α inhibition in many inflammatory conditions.^16–19

Preliminary studies with inflammatory cells isolated from AAA and control patients suggested differential expression of IL-1β. We analyzed plasma IL-1β levels in these two groups using high sensitivity ELISA. We found small, but significant differences in IL-1β expression. IL-1β was not detectable in any AAA patients, while low levels of expression were detected in most control patients. Finding that IL-1β was expressed in controls and not AAA patients was unexpected. We attempted to cross-validate these findings in our mouse model. Choosing control patients for studying AAA is challenging. We selected age-matched patients with peripheral or cerebrovascular disease, but with normal diameter aortas. One limitation of our study is that we did not have access to all patient medication records. Most likely, both groups of patients were on aspirin and statins, which could indirectly affect the level of inflammatory cytokines. Ideally, we would have assessed cytokine levels in aortic tissue from these patients; however, endovascular therapies have greatly reduced access to these tissues. Identifying circulating biomarkers that reflect aneurysm biology will be valuable to AAA research.

Animal models are most useful for understanding disease pathogenesis. We investigated the mechanisms of IL-1β inhibition in the CaCl₂ model of AAA. In the elastase infusion model, Johnston et al. found that aneurysm formation was attenuated by IL-1β deletion or antagonism.^8,21 We have previously shown that TNF-α inhibition or deletion prevented aneurysm formation.⁹ Given the known similarities between actions of IL-1β and TNF-α, we were surprised to find that deletion of IL-1β or its receptor did not prevent aneurysm formation. Furthermore, its deletion did not reduce MMP-9 or MMP-2 levels, which are known to correlate with AAA. In fact, we observed a trend towards increased diameter in IL-1β^−/− mice while the IL-1R^−/− mice actually showed a significant increase in aortic diameter compared to WT. These discrepancies between our observations and those of Johnston et al. can only be attributed to differences between the two models since the same mice were used in both studies. The elastase infusion model combines significant mechanical dilation with subsequent inflammation producing aneurysms within two weeks.⁸ The CaCl₂ model relies entirely on a local inflammatory response and polarization of invading macrophages to an M1 phenotype. In normal wound healing, the M2 macrophages reduce inflammation and promote fibrosis after several weeks.^22,23 Although the precise role of macrophage polarization in elastase infusion model is not known, the discrepant observations related to IL-1β inhibition are likely the results of difference in the macrophage phenotype. The effects of IL-1β manipulation on M2 polarization may not manifest in this more acute model.²⁴ Although none of the animal models can accurately reproduce all of the features of human AAA, the results obtained in CaCl₂ model accurately reflected the lack of clinical efficacy of IL-1β inhibition observed in AAA patients. The clinical trial using an IL-1β antibody was terminated early due to lack of efficacy and futility (ClinicalTrials.gov Identifier: NCT02007252). In contrast, the CANTOS trial recently demonstrated efficacy of IL-1β antibodies in reducing adverse cardiac-related events in patients with atherosclerotic coronary artery disease.²⁵ However, this small benefit was offset by an increased risk of fatal infection and no observed effect on cardiovascular mortality.²⁶

IL-1β and TNF-α have long been considered prototypical inflammatory cytokines. As such, they have been targeted as possible treatments for a number of immune mediated diseases.^13–15 Previous studies have shown profound inhibition of aneurysm formation with antagonism of TNF-α in murine models of AAA.⁹ Therefore, we expected similar results from IL-1β inhibition. Considering the inflammatory nature of both IL-1β and TFN-α, we were surprised to find decreased levels of IL-1β in AAA patients and no inhibition of aneurysm formation with IL-1β deletion in the murine AAA model. Clinical studies using TNF-α antibodies have identified anti-inflammatory effects among different white-blood cell populations including macrophages.²⁷ We have recently demonstrated the critical role of macrophage polarization toward the M1 phenotype in aneurysm formation.²⁰ In this study, we confirmed the presence of infused macrophages in the aortic tissue. This demonstrates local activity by infused macrophages; however, the relatively low proportion does not exclude systemic immune modulation. Using IL-1β^−/− or TNF-α^−/− macrophages, we identified different effects after promoting the M1 phenotype by exposure to LPS and IFN-γ. Cells lacking the expression of TNF-α in this study, expressed higher levels of M2 cytokines suggesting that TNF-α antagonism attenuates the destructive effects of M1 macrophages. Evidence of the protective effects of TNF-α antagonism were shown by the reduction in aneurysm size in the mice infused with activated macrophages from TNF-α^−/− mice.

Taken together, we demonstrated that deletion of IL-1β and TNF-α had very different effects on macrophage polarization. Human aneurysm tissue displays an abundance of infiltrating macrophages. Considering that macrophages may have significant anti-inflammatory effects, regulating the macrophage phenotype could alter the course of aneurysm disease. Using the CaCl₂ aneurysm model, we have identified a mechanism that may explain the difference in clinical efficacy of IL-1β vs TNF-α antagonism. Our studies suggest that TNF-α, rather than IL-1β, blockade would be more effective in diseases associated with M1 macrophage polarization including AAA.

Table 1.

Patient Demographics

	AAA (n = 15)	Control (n = 15)	P-value
Age
<65	1	0
>65	14	15	1.000
Gender
Male	15	15
Female	0	0	1.0000
Vascular Disease
AAA	15	0
Carotid	0	6
PVD	0	6
Carotid + PVD	0	3	<0.0001
Smoking Status
Unknown	1	1
Never	1	2
Current	1	3
Former	12	9	0.7271

AAA = Abdominal Aortic Aneurysm; Carotid = Carotid Artery Disease; PVD = Peripheral Vascular Disease

Fisher exact test was used

Highlights.

• Abdominal Aortic Aneurysms (AAAs) are inflammatory in nature with human aneurysm tissue displaying an abundance of infiltrating macrophages.

• Although IL-1β is a pro-inflammatory cytokine, its effects on aneurysm formation and macrophage polarization differ from TNF-α.

• Decreased plasma levels of IL-1β were found in AAA patients and targeted deletion of IL-1β or its receptor, did not block aneurysm in the murine calcium chloride AAA model.

• The differential effects of IL-1β and TNF-α inhibition are related to M1/M2 macrophage polarization.

• Differences in clinical efficacy of IL-1β and TNF-α antibody therapies in management of inflammatory diseases is likely related to their different effects on macrophage polarization.

Abbreviations

AAA

Abdominal Aortic Aneurysm

IL-1β

Interleukin 1 beta

TNF-α

Tumor necrosis factor alpha

WT

Wild-type

BMDMs

Bone marrow-derived macrophages

ELISA

Enzyme-linked immunosorbent assay

MMP

Matrix Metalloproteinase