IL-5 links adaptive and natural immunity specific for epitopes of oxidized LDL and protects from atherosclerosis (original) (raw)
Immunization with MDA-LDL induces a specific Th2-biased response. We first immunized normocholesterolemic C57BL/6 mice with homologous MDA-LDL in Freund’s adjuvant and examined the antigen-specific proliferation in splenic cultures. Splenocytes from immunized but not naive mice exhibited dose-dependent proliferation in response to MDA-LDL, but not to native LDL (Figure 1A). We next quantified titers of TD antibody isotypes to MDA-LDL in plasma. Measurements from three independent studies revealed more than an eight-fold greater induction of MDA-LDL-specific IgG1 titers over IgG2a titers (P < 0.01), demonstrating a strong Th2 bias of the induced response (Figure 1B), which occurred despite the use of CFA in the C57BL/6 genetic background that typically results in Th1 responses (28). Studies in which MHC class II–/– and T-cell receptor β_–/–_ mice were injected with MDA-LDL indicated that the IgG responses to MDA-LDL were dependent on MHC II class antigen presenting cells and αβT-cell receptor–expressing T cells (data not shown). In parallel studies, we also immunized C57BL/6 mice with MDA-LDL, this time without adjuvant. Although fewer than half of these animals developed an antigen-specific titer, again IgG1 was the dominant isotype in the responding mice, and even at plasma dilutions as low as 1:50, no IgG2a binding was detected (data not shown).
Immunization with MDA-LDL induces a specific Th2 response. C57BL/6 mice were immunized with homologous MDA-LDL in Freund’s adjuvant or remained naive. One week after the third injection, cellular and humoral immune responses were assessed. Three independent immunization studies were performed. (A) Splenocyte proliferation assay. Splenocytes of immunized (n = 6) or naive (n = 6) mice were cultured with titrated amounts of murine native LDL (open circles, immunized; open squares, naive) or murine MDA-LDL (filled circles, immunized; filled squares, naive), and antigen-specific proliferation was measured by 3[H]-thymidine uptake. (B) Increased IgG1 titers in plasma of immunized mice. Data represent titers from all mice studied (n = 14; P < 0.01, Student’s paired t test). (C) Increased frequency of MDA-LDL–specific IL-5 secreting cells in spleens of immunized mice as assessed by ELISpot assay. Splenocytes of immunized (n = 8; filled circles) and naive (n = 8; open circles) mice were incubated overnight with and without murine MDA-LDL, and the frequencies of MDA-LDL–specific IFN-γ or IL-5 spot-forming cells (SFCs) were assessed. Shown are the mean SFCs per 2 × 106 cells for IFN-γ and IL-5 of individual mice from two experiments (P < 0.01, Student’s paired t test; IFN-γ vs. IL-5 SFC of immunized mice). (D) MDA-LDL–specific Th2 cytokine secretion in cultures of splenocytes from four mice incubated for 72 hours with 25 μg/ml murine native LDL (white bars) or murine MDA-LDL (black bars). Control cultures stimulated with anti-CD3 and anti-CD28 produced 11.7 ± 2.1 ng/ml IFN-γ, 156 ± 39 pg/ml IL-5, 470 ± 43 pg/ml IL-10, and 1.2 ± 0.4 ng/ml IL-13. *P < 0.05; MANOVA followed by Newman-Keuls test, comparing the stimulation with increasing amounts (0.25 [not shown], 2.5 [not shown], and 25 μg/ml) of native and MDA-LDL. All values shown in Figure 1 are mean ± SEM.
To demonstrate the dominant Th2 responses, we quantified the relative frequencies of MDA-LDL–specific Th1 and Th2 cells in splenocyte cultures by enzyme-linked immunospot (ELISpot) assay using IFN-γ and IL-5 as surrogate Th1 and Th2 cytokines, respectively. After overnight incubation with MDA-LDL, an increased and significantly higher frequency of IL-5 secreting cells specific for MDA-LDL could be detected (P < 0.01) compared with an only minimal induction of specific IFN-γ–secreting cells in the spleens of the same mice (Figure 1C). Splenocytes from naive mice showed no response.
In addition, measurement of cytokines present in the supernatants of splenocytes from immunized mice cultured with MDA-LDL revealed an MDA-LDL–specific secretion of Th2 cytokines IL-5, IL-10, and IL-13, but not of the Th1 cytokine IFN-γ (Figure 1D). Supernatants from splenocytes cultured with native LDL did not secrete appreciable levels of these cytokines, with the exception of IL-10, which was not consistent in every experiment (Figure 1D). Splenocytes from naive control mice, cultured with either MDA-LDL or native LDL, did not exhibit specific cytokine secretion (data not shown).
Th2 responses persist in MDA-LDL–immunized mice with atherosclerosis. To determine whether such a Th2 bias in response to MDA-LDL immunization would also occur in mice with developing atherosclerosis, which has been associated with increased Th1 responses (17), we immunized cholesterol-fed LDLR–/– mice with MDA-LDL. Mice were initially immunized twice with either MDA-LDL (n = 10) or PBS (n = 11) emulsified in Freund’s adjuvant and were then fed a high-cholesterol diet for 13 weeks, during which time three additional boosts were given. At the end of this intervention, mice in the MDA-LDL and PBS groups gained equal weight (27.5 ± 0.9 g vs. 27.5 ± 0.7 g) and had similar total plasma cholesterol (1,119 ± 17 mg/dl vs. 1,114 ± 34 mg/dl) and triglyceride levels (224 ± 12 mg/dl vs. 251 ± 28 mg/dl). Both groups of mice had identical lipoprotein profiles, as measured by fast-performance liquid chromatography (FPLC; Figure 2A). Despite the marked hypercholesterolemia, consistent with earlier studies (13, 14), MDA-LDL immunization led to significantly reduced (P < 0.02) atherosclerosis in the aortic origin (Figure 2B), the site where atherosclerotic lesions first develop in mice. However, the percentage of surface area of the arch covered by lesions was not significantly different at this site between the two groups (10.3 ± 1.0% vs. 10.9 ± 1.0%).
Decreased atherosclerosis and dominant Th2 response in immunized LDLR–/– mice. Mice were immunized with homologous MDA-LDL (n = 10) or PBS (n = 11) in Freund’s adjuvant, and then fed a high-cholesterol diet for 13 weeks, during which they received further booster immunizations. (A) Lipoprotein profiles at time of death in pooled plasma of all mice immunized with MDA-LDL (filled circles) or PBS (open circles) as determined by FPLC. (B) Decreased atherosclerotic lesion size in cross-sections through the aortic origin in mice immunized with MDA-LDL. Values are mean ± SEM in mm2/section. (C and D) Plasma IgG1 (filled circles) and IgG2a (open circles) dilution curves of binding to MDA-LDL of mice immunized with (C) MDA-LDL or (D) PBS. Values are the mean ± SEM of all final plasma samples for each group measured in duplicate. (E and F) ELISpot assay of frequencies of MDA-LDL–specific cytokine-secreting cells in the spleens of mice immunized with (E) MDA-LDL or (F) PBS. Splenocytes were incubated overnight in the absence or presence of murine MDA-LDL with and without anti-CD28, and the frequencies of MDA-LDL–specific IFN-γ (white bars) or IL-5 (black bars) SFCs were assessed. Bars represent the mean SFCs ± SEM of 2 × 106 cells of all mice for each group. P < 0.01, Student’s paired t test.
Similarly to the normocholesterolemic C57BL/6 mice (Figure 1B), immunized cholesterol-fed LDLR–/– mice developed IgG1 titers to MDA-LDL that were more than 2 logs higher than the IgG2a titers (Figure 2C). In contrast, LDLR–/– mice injected with PBS in Freund’s adjuvant developed minimal IgG2a titers to MDA-LDL as a result of the atherogenic diet, but no specific IgG1 titers were induced (Figure 2D). This Th2 bias was further confirmed by the study of MDA-LDL–specific T cells in the spleens of these mice. ELISpot analysis for cells secreting IFN-γ or IL-5 revealed an increased frequency of IL-5–secreting cells specific for MDA-LDL in the spleens of MDA-LDL–immunized mice (Figure 2E), but not in the spleens of LDLR–/– mice injected with PBS in Freund’s adjuvant (Figure 2F). This pattern of response was further enhanced by the addition of anti-CD28 for in vitro costimulation of these MDA-LDL–specific T cells previously committed in vivo (Figure 2E). Splenocytes from LDLR–/– control mice displayed a low frequency of MDA-LDL–specific IFN-γ–secreting cells, even when cultures were costimulated with anti-CD28 (Figure 2F), consistent with the moderate IgG2a titers to MDA-LDL (Figure 2D). Thus, even during the development of atherosclerosis, which is accompanied by many proinflammatory stimuli associated with Th1 responses (17), MDA-LDL–specific Th2 responses remained dominant in immunized mice.
IL-5 is prominently secreted by MDA-LDL–specific T cells. To further define the Th2 response in the MDA-LDL–immunized LDLR–/– mice, we quantified the cytokines secreted by splenocytes from immunized mice cultured with MDA-LDL or native LDL, in the presence of anti-CD28 to provide sufficient costimulation. To identify the dominant cytokine secreted by MDA-specific T cells, we stimulated parallel cultures with anti-CD3 and anti-CD28 (anti-CD3/anti-CD28) to achieve maximal nonspecific cytokine release and compared the results with the cytokine release obtained after stimulation with MDA-LDL (Figure 3, left panels). Cytokine production in response to anti-CD3/anti-CD28 did not differ between the two groups (data not shown). In cultures with native LDL, no substantial cytokine secretion was observed. Splenocyte cultures from LDLR–/– mice that received PBS in Freund’s adjuvant exhibited no specific cytokine secretion (Figure 3, left panels). Strikingly, MDA-LDL–specific secretion of IL-5 by T cells of MDA-immunized mice was greater than 75% of the amount achieved with maximal nonspecific stimulation in parallel cultures. By comparison, MDA-LDL–specific secretion of other Th2 cytokines (IL-4, IL-10, IL-13) was only 20–40% of that induced by anti-CD3/anti-CD28. In contrast, MDA-LDL–specific IFN-γ secretion was negligible. This pattern of response was also seen in cultures where dose-dependent cytokine secretion was assessed in response to stimulation with MDA-LDL alone (i.e., without additional costimulation; Figure 3, right panels).
Antigen-specific cytokine secretion of splenocytes from cholesterol-fed LDLR–/– mice immunized with MDA-LDL (n = 10; black bars) or PBS (n = 11; white bars). Left column: Splenocytes were cultured for 72 hours in the presence of anti-CD28 and either anti-CD3, murine native LDL, or murine MDA-LDL, and supernatants were analyzed for cytokines. Data are presented as percentage of the cytokine secretion in parallel cultures maximally stimulated with anti-CD3/CD28 (=100%). Right column: splenocytes were stimulated either alone or with indicated amounts of murine MDA-LDL (without anti-CD28). Values are mean ± SEM of splenocyte cultures of all mice from each group.
Plasma levels of IL-5 are increased by MDA-LDL immunization. To assess whether IL-5 was also increased in vivo, we measured cytokine levels in the plasma of all mice at the end of the intervention study. MDA-LDL–immunized mice had significantly higher plasma levels of IL-5 when compared to PBS–immunized mice (Figure 4A). Plasma IFN-γ levels were much lower and close to the quantification limit (50 pg/ml), with no measured differences between the two groups (data not shown).
Increased levels of IL-5 and EO6 in the plasma of cholesterol-fed LDLR–/– mice immunized with MDA-LDL. At time of death, plasma was obtained from mice immunized with MDA-LDL (n = 10) and PBS (n = 11). (A) IL-5 levels in plasma of MDA-LDL immunized mice are increased. (B) EO6 antibody levels in plasma of MDA-LDL–immunized mice are increased. The amount of EO6 present was determined using an ELISA based capture assay with anti–idiotypic AB1-2 as described in Methods and was calculated based on an EO6 standard curve. (C) Comparison of T15/EO6 antibody titers in LDLR–/– mice immunized with MDA-LDL (current study) versus mice immunized with S. pneumoniae (R36a; n = 9) from a previous study (12). EO6 antibody titers were determined at a plasma dilution of 1:500 in the same assay. Purified EO6 was used as a positive control. Biot., biotinylated. (D) Circulating IgM/apoB ICs are increased in MDA-LDL–immunized mice. Results are expressed as IgM/apoB. All bars represent the mean ± SEM values of all mice from each intervention group.
MDA-LDL immunization increases T15/EO6 natural IgM antibodies to OxLDL. Consistent with a previous study (14), the MDA-LDL immunization led to significant increases in IgM titers to MDA-LDL (data not shown), but surprisingly, we also found increased titers of IgM to OxLDL, and increased titers of IgM specific to the PC of OxPLs (data not shown). As noted, such epitopes are not present in the model MDA-LDL used as an immunogen (6, 7). To determine the absolute plasma levels of T15/EO6 IgM antibodies, we used a capture assay with the T15-specific anti-idiotypic antibody AB1-2 (29). MDA-LDL–immunized mice had significantly higher T15/EO6 plasma levels compared with PBS–immunized mice (806 ± 157 vs. 277 ± 68 ng/ml, P < 0.005; Figure 4B). To put these increases in perspective, we compared in the same assay the plasma titers of T15/EO6 induced by MDA-LDL to those induced in a prior study by immunization with S. pneumoniae, a potent PC antigen that induced a robust increase in T15/EO6 sufficient to inhibit the progression of atherosclerosis (12). These data indicate that immunization with MDA-LDL led to T15/EO6 levels that were on average 35% of those achieved by immunization with pneumococci (Figure 4C).
IgMs form immune complexes in plasma of MDA-LDL–immunized mice. The increased T15/EO6 titers in MDA-LDL–immunized LDLR–/– mice demonstrated the additional engagement of innate immunity and suggested that EO6 could have contributed to the protective effect of this immunization. Therefore, we tested the possibility that there were higher levels of IgM/apoB immune complexes (ICs) in the plasma of MDA-LDL–immunized mice than in the plasma of PBS-immunized LDLR–/– mice. Indeed, mice immunized with MDA-LDL had higher plasma levels of IgM/apoB ICs than the PBS group did (Figure 4D, P < 0.01). There were only low levels of circulating IgG/apoB ICs, and these did not differ between the two groups (0.11 ± 0.02 vs. 0.10 ± 0.02 relative light units [RLU]/RLU for the MDA and PBS groups, respectively).
IL-5 stimulates B-1 cells to secrete natural antibody T15/EO6. The expansion of T15/EO6 IgM in the MDA-LDL–immunized mice was unexpected, given that MDA-LDL does not bind to T15/EO6 antibodies and thus could not have directly ligated the B-cell receptor of PC-specific B-1 cells, which would lead to the secretion of T15/EO6. Although the secretion of B-1 cell–dependent IgM antibodies is known to be T cell independent, T cells are thought to provide noncognate help for B-1 cells (30), and IL-5 has been described as an important factor in B-1 cell biology (31, 32).
To directly test the hypothesis that IL-5 derived from the MDA-LDL–specific T-cell responses provided noncognate T cell help for B-1 cells to secrete T15/EO6, we isolated B-cell subsets from the peritoneal cavity and the spleens of naive C57BL/6 mice by FACS, based on the surface expression of CD19 and CD23 (33). Peritoneal B-1 cells were separated from conventional B-2 cells, and splenic B cells were separated into marginal-zone B cells and follicular B cells. These individual subsets of B cells were cultured in growth media alone or with IFN-γ, IL-4, or IL-5. After 7 days, culture supernatants were harvested and antibody binding to OxLDL was determined by ELISA. Only IL-5 strongly stimulated B-1 cells to secrete IgM antibodies to OxLDL (Figure 5A), and addition of an IL-5 neutralizing antibody abolished this response. None of the other B cell subsets showed equivalent anti-OxLDL IgM secretion following IL-5 stimulation. There was no comparable IgG binding of these culture supernatants (data not shown), which is consistent with the fact that B-1 cells predominantly secrete IgM. In addition, using the anti-idiotypic antibody AB1-2, we definitively demonstrated that the B-1 cells secreted T15/EO6 following IL-5 stimulation (Figure 5B).
Role of IL-5 in the production of T15/EO6 natural antibodies. (A and B) IL-5 stimulates antibody secretion in vitro. Peritoneal B-1 cells (black bars) and B-2 cells (white bars), and splenic marginal zone (MZ) B cells (dark gray bars) and follicular B cells (light gray bars) were cultured for 7 days in either medium alone, or with IFN-γ, IL-4, or IL-5 with and without anti–IL-5 mAb. Polymyxin B was added to all cultures to neutralize contaminating LPS effects. (A) IgM binding to OxLDL in culture supernatants. (B) T15/EO6 antibodies in culture supernatants, measured with the anti-idiotypic antibody AB1-2. Values are mean RLU ± SEM from duplicate determinations of triplicate cultures. This experiment was repeated three times. (C) IL-5 stimulates the production of T15/EO6 antibodies in vivo. C57BL/6 mice received daily intraperitoneal injections with recombinant mouse IL-5 (n = 6) or vehicle only (BSA; n = 4) for 7 days, and the amount of T15/EO6 antibodies was determined in the plasma. Shown is the fold increase over the baseline levels at a 1:100 plasma dilution. Bars represent mean ± SEM of triplicate determinations of individual mice. P < 0.05, Student’s unpaired t test; Welch corrected. (D) Naive IL-5–/– mice have decreased T15/EO6 antibody levels. In IL-5+/+ (open circles) and IL-5–/– C57BL/6 mice (filled circles), 15–16 weeks of age (both n = 3), T15/EO6 antibody titers were determined. Shown is the binding of T15/EO6 antibodies of individual plasma samples diluted 1:100. Values are the mean RLU of triplicate determinations. (E–G) Impaired induction of T15/EO6 antibodies in IL-5–/– mice immunized with MDA-LDL. IL-5+/+ (open circles) and IL-5–/– (filled circles) C57BL/6 mice were immunized with MDA-LDL and plasma antibody titers were determined. (E) IgG1 binding to MDA-LDL (F), IgM binding to OxLDL, and (G) T15/EO6 antibodies of individual plasmas at 1:250 dilution before and after immunization. Values are the mean RLU of triplicate determinations.
We next tested the capacity of IL-5 to induce T15/EO6 antibodies in vivo by daily intraperitoneal injections of naive C57BL/6 mice with murine recombinant IL-5 for 7 days. On day 8, there was more than a four-fold increase in plasma T15/EO6 titers in mice injected with IL-5 (P < 0.05), compared to vehicle (BSA; Figure 5C).
We analyzed plasma levels of T15/EO6 antibodies in IL-5–/– mice, which have been reported to have transiently decreased numbers of B-1 cells at a young age (<8 weeks). Surprisingly, the baseline levels of T15/EO6 antibodies were strongly affected by the IL-5 deficiency even in adult mice. While 16 weeks old, naive IL-5+/+ C57BL/6 mice had measurable plasma levels of T15/EO6 antibodies at dilutions tested; age-matched IL-5–/– mice had no detectable T15/EO6 antibodies (Figure 5D). Interestingly, neither basal IgM levels to OxLDL, nor those to MDA-LDL, differed between the two groups (data not shown), suggesting the existence of other OxLDL-specific IgM antibodies that are not derived from B-1 cells.
Next, we examined whether IL-5 is essential for the induction of T15/EO6 as a consequence of MDA-LDL immunization. IL-5–/– and IL-5+/+ mice (both on a C57BL/6 background) were immunized with murine MDA-LDL. After one primary and two booster immunizations, IL-5–/– and IL-5+/+ mice displayed similar IgG1 (Figure 5E) and IgG2a titers (data not shown) to MDA-LDL, demonstrating that IL-5 deficiency did not impair TD responses. In contrast, the development of IgM titers to MDA-LDL was lower in IL-5–/– mice than in the IL-5+/+ control group (data not shown). Most striking, however, was the fact that the induction of IgMs to OxLDL (Figure 5F), and specifically the induction of T15/EO6 (Figure 5G), was greatly impaired in IL-5–/– mice, despite three injections (Figure 5, E–G), suggesting a pivotal role for IL-5 in the induction of T15/EO6 as a consequence of immunization with MDA-LDL.
IL-5 deficiency accelerates atherosclerosis. We asked whether IL-5 itself could play a functional role in atherogenesis, even in the absence of exogenous immunizations. To generate LDLR–/– mice with deficient production of IL-5, we transplanted irradiated LDLR–/– mice with bone marrow from IL-5–/– or IL-5+/+ mice. Four weeks after bone marrow transplantation (BMT), mice were switched to an atherogenic diet for the subsequent 16 weeks to induce lesion formation. At the time of sacrifice, splenocytes of the LDLR–/– mice with IL-5–/– bone marrow showed significantly less production of IL-5 following anti-CD3 stimulation compared with those that received IL-5+/+ bone marrow (Table 1). The production of the cytokines IL-4, IL-10, and IFN-γ, however, did not differ between the two groups (Table 1). Moreover, production of IL-5 by anti-CD3–stimulated peripheral blood cells was virtually absent in recipients of IL-5–/– bone marrow (Table 1). Consequently, plasma IL-5 levels were significantly decreased in these mice (Table 1). Thus, the IL-5–/– LDLR–/– bone marrow chimeras had a selective insufficiency, but not absence, of IL-5 production compared to the control mice that received IL-5+/+ bone marrow. En face analyses of the extent of atherosclerosis in the entire aorta revealed a significantly increased degree of lesion formation as a result of transplantation of IL-5–/– bone marrow, supporting the hypothesized protective role of IL-5 in atherogenesis (Table 1 and Figure 6A). In addition, atherosclerosis in the aortic origin of recipients of IL-5–/– bone marrow was also significantly greater than that in controls, despite the very advanced extent of atherosclerotic lesion formation (Table 1). To determine whether the defective IL-5 secretion resulted in impaired stimulation of B-1 cells to secrete atheroprotective antibodies, we evaluated the titers of T15/EO6 antibodies in the plasma of these mice. Indeed, at the time of sacrifice, IL-5–/– LDLR–/– bone marrow chimeras had significantly lower T15/EO6 plasma titers than the IL-5+/+ controls (Figure 6B). This was not due to an absolute deficiency of B-1 cells, as both recipient groups had similar numbers of peritoneal B-1 cells at the time of death, as determined by FACS (Table 1). Analyses of autoantibody titers to MDA-LDL and OxLDL revealed decreased levels of TI IgG3 titers but not of TD IgG1 or IgG2a titers (data not shown). T15/EO6 IgM antibodies can bind OxLDL and prevent its uptake by macrophages (7). To determine whether the decreased levels of T15/EO6 antibodies also led to an impaired formation of ICs with minimally oxidized LDL in vivo, we measured the levels of circulating LDL ICs that contained T15/EO6 antibodies. Recipients of IL-5–/– bone marrow had significantly lower levels of plasma T15/EO6 antibody–apoB ICs during the period of cholesterol feeding (Figure 6C). IgG-apoB ICs were very low and did not differ between the two recipient groups (data not shown).
Increased atherosclerosis in IL-5–deficient LDLR–/– mice. LDLR–/– mice were reconstituted with bone marrow from either IL-5–/– mice (IL-5–/–; n = 15) or IL-5+/+ mice (IL-5+/+; n = 14) and fed an atherogenic diet for 16 weeks. (A) Increased extent of atherosclerosis in aortas of recipients of IL-5–/– bone marrow (n = 15) compared to recipients of IL-5+/+ bone marrow (n = 14; P < 0.01). Horizontal bars indicate means of each group. (B) Decreased titers of T15/EO6 antibodies in plasma of LDLR–/– mice reconstituted with IL-5–/– bone marrow. Data are mean ± SEM titers of all mice of each group (P < 0.05). (C) Reduced formation of circulating T15/EO6-apoB ICs in recipients of IL-5–/– bone marrow. Results are expressed as T15/EO6 antibodies per apoB. Data are the mean ± SEM values of all mice from each intervention group at 4, 8, and 16 weeks after initiation of the atherogenic diet. P < 0.05, repeated-measures ANOVA.
Overview of experimental groups from bone marrow transplantation