Atheroprotective effect of adjuvants in apolipoprotein E knockout mice

Atheroprotective effect of adjuvants in apolipoprotein E knockout mice

Atherosclerosis 184 (2006) 330–341 Atheroprotective effect of adjuvants in apolipoprotein E knockout mice J. Khallou-Laschet a,1 , E. Tupin a,f,1 , G...

520KB Sizes 0 Downloads 10 Views

Atherosclerosis 184 (2006) 330–341

Atheroprotective effect of adjuvants in apolipoprotein E knockout mice J. Khallou-Laschet a,1 , E. Tupin a,f,1 , G. Caligiuri b , B. Poirier a , N. Thieblemont c , A.-T. Gaston a , M. Vandaele a , J. Bleton e , A. Tchapla e , S.V. Kaveri a , M. Rudling d , A. Nicoletti a,∗ a

d

INSERM U681, Institut Biom´edical des Cordeliers, 15, rue de l’Ecole de M´edecine, Paris, France b INSERM E00-16, Facult´ e de M´edecine Necker-Enfants Malades, Paris, France c CNRS FRE2444, Universit´ e Ren´e Descartes, Paris, France Molecular Nutrition Unit, Novum and CME, Department of Medicine, Karolinska University Hospital at Huddinge, Sweden e LETIAM, Groupe de Chimie Analytique de Paris Sud, EA 33-43, Orsay, France f Center for Molecular Medicine, Department of Medicine, Karolinska Institutet, Stockholm, Sweden Received 24 September 2004; received in revised form 6 April 2005; accepted 27 April 2005 Available online 26 July 2005

Abstract Strategies aimed at treating atherosclerosis by immunization protocols are emerging. Such protocols commonly use adjuvants as nonspecific stimulators of immune responses. However, adjuvants are known to modify various disease processes. The aim of this study was to determine whether adjuvants alter the development of atherosclerosis. We performed immunization protocols in apolipoprotein E knockout mice (E◦ ) following chronic administration schedules commonly employed in experimental atherosclerosis. Our results point out a dramatic effect of several adjuvants on the development of atherosclerosis; three of the four adjuvants tested reduced lesion size. The Alum adjuvant, which is the adjuvant currently used in most vaccination protocols in humans, displayed a strong atheroprotective effect. Mechanisms accounting for atheroprotective effect of Freund’s adjuvants included their capacity to increase both Th2 responses and anti-MDA-LDL IgM titers, and/or to impose atheroprotective lipoprotein profiles. The present study indicates that adjuvants have potent atheromodulating capabilities, and thus, implies that the choice of adjuvant is crucial in long-term immunization protocols in experimental atherosclerosis. © 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Immunology; Antibodies; Cytokines; Th balance; Mineral oil

1. Introduction Strategies aimed at evaluating immunization protocols in experimental atherosclerosis are emerging. Adjuvants are used as non-specific stimulators to enhance the immune responses of immunogens that otherwise would lead to a slow and inefficient process. Adjuvants prevent the catabolism of the combined antigen and attract appropriate immune cells to ∗ 1

Corresponding author. fax: +33 1 55 42 8262. E-mail address: [email protected] (A. Nicoletti). These authors contributed equally to this work.

0021-9150/$ – see front matter © 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.atherosclerosis.2005.04.021

interact with the immunogen and with each other. However, it is known that adjuvants can per se play a role in various pathological processes [1,2]. The present study aimed to evaluate whether adjuvants modulate atherogenesis. Freund’s adjuvants [3] are most commonly used because of their stronger and longer lasting immunogenic response as compared to other adjuvants. They consist of tiny droplets of water (in which immunostimulants can be incorporated) stabilized by a surfactant (such as mannide monooleate) in a continuous phase of mineral oil or other oils, such as squalene or squalane. The complete Freund’s adjuvant (CFA) contains killed cells of Mycobacterium tuberculosis while the

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

incomplete Freund’s adjuvant (IFA) does not. CFA promotes both antibody formation and delayed-type hypersensitivity reactions, while IFA is used to promote only antibody formation. Alum adjuvants are used as an alternative to Freund’s adjuvants because they have less adverse effects. Alum adjuvants consist of aluminum hydroxide or aluminum phosphatehydrated gels that have the ability to adsorb protein antigens from an aqueous solution. AlhydrogelTM (aluminum hydroxide gel, referred to as Alum) has been elected as research standard in order to minimize inconsistency due to variability between individual preparations. Alum adjuvants may present adsorbed soluble antigens to immunocompetent cells in a ‘particulate’ form, which would facilitate their uptake. Alum activates complement, attracts eosinophils to the injection site, stimulates the production of IgE but is inefficient in inducing a delayed-type hypersensitivity response [4]. Bacterial DNA, but not vertebrate DNA, has direct immunostimulatory effects on leukocytes in vitro [5,6]. This activation is due to unmethylated CpG dinucleotides [7] that are present at the expected frequency in bacterial DNA but are under-represented and methylated in vertebrate DNA [8]. CpG DNA induces proliferation of 95% of B cells and triggers Ig secretion. CpG DNA also directly activates monocytes, macrophages, and dendritic cells to secrete a variety of cytokines. In the present study, we found that Freund’s and Alum adjuvants exert potent atheroprotective effects when administered to the atherosclerosis-prone apolipoprotein E knockout (E◦ ) mice. We have identified the components in Freund’s adjuvant endowed with atheroprotective capacity and have characterized mechanisms that may explain their unexpected effect on atherosclerosis. In addition, immunization protocols in experimental atherosclerosis employ a number of adjuvant administrations for a long period of time. Because such protocols differ significantly from the conventional immunization protocols, we have determined the immune response that can be expected from repeated and long-term immunization protocols. 2. Materials and methods 2.1. Injection protocols In all the protocols, 6–7-week-old E◦ male mice from our breeding facility were used. Mice were maintained on a regular chow supplemented with 0.15% cholesterol. Adjuvants alone or associated with an antigen (50/50, v/v) were injected in a volume of 100 ␮l per mouse. Mice were kept in standard conditions and were sacrificed at 16 weeks of age (except when indicated). All experiments were approved by our institutional ethical committee. 2.1.1. Protocol #1 In the first experiment of this series, mice received a subcutaneous injection of CFA followed by intraperitoneal (i.p.)

331

injections of IFA (n = 4) every week. Control mice received injections of PBS (n = 5). Mice were sacrificed at 11 weeks of age. In the second experiment, mice received i.p. injections of CFA (n = 3), IFA (n = 3) or PBS (n = 3) once a week. In the third experiment, mice were injected i.p., every second week, with Alum (Superfos, Biosector, Denmark) (30 ␮l/mouse; n = 7), the oligonucleotide CpG 1826 (Operon Quiagen, Alameda, USA; synthesized with a nucleaseresistant phosphorothioate backbone) (500 ␮g/mouse; n = 8) or PBS (n = 9). 2.1.2. Protocol #2 Mice were injected i.p. once a week with IFA-OVA (n = 7), CFA-OVA (n = 7) (100 ␮g OVA/mouse), IFA-PBS (n = 7) or CFA-PBS (n = 7). A group of mice was left untouched (Ctrl) (n = 7). 2.1.3. Protocol #3 Mice were injected i.p. once a week with IFA-PBS (n = 7), or IFA components: mannide monooleate (MMO, ACROS; 4.5 ␮l/mouse; n =8) or mineral oil (25.5 ␮l/mouse; n = 7). The mode of preparation of mineral oils makes each lot unique. We arbitrarily chose the mineral oil ref 415080010 from ACROS. The amount of MMO and mineral oil used was proportional to their respective amount in the IFA preparation. An additional group of mice received LPS (5 ␮l/mouse; n = 8). A group of untouched animals (n = 6) was used as control. 2.2. Serum measurements and quantitation of atherosclerotic lesions Mice were bled and sacrificed under anesthesia. Total serum cholesterol (TC) level was measured using a Boehringer Mannheim kit (CHOD-PAP). The lipoprotein profiles were determined by fast performance liquid chromatography (FPLC) on the indicated number of randomized serum samples from each group as previously described [9]. The areas under the curves were used to compute the chylomicron/VLDL, LDL and HDL cholesterol levels. The aortic root was dissected and lesions were quantitated by a blinded observer as previously described [10]. Briefly, serial cryostat sections were cut from the proximal 1 mm of the aortic root. Lesion size was determined by computer-assisted morphometry on four hematoxylin/oil red-O-stained sections cut at 200, 400, 600, and 800 ␮m from the cusp origin. Lesion extension is expressed as the percentage of the surface area of the aorta occupied by lesions and is calculated with the following formula: [(Surface Area of Lesion)/(Surface Area of the Aorta)] × 100. We have previously shown that this minimizes errors caused by oblique sections that could otherwise lead to overestimation of the surface area occupied by lesions [10].

332

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

2.3. Cell culture Spleen cells were prepared as described [10]. Five hundred thousand mononuclear cells/well were cultured at 37 ◦ C, 5% CO2 into 96-well flat bottom plates (triplicates), ±concanavalin A (0.5–2.5 ␮g/ml, Sigma), ±OVA (50–200 ␮g/ml). After 48 h, 1 ␮Ci of 3 H-thymidine was added to each well. After another 24 h of incubation, incorporated 3 H-thymidine was counted in a Microbeta counter. Results are expressed as cpm. 3. ELISA 3.1. Cytokines Serum IFN␥, IL-12, IL-10, IL-6, IL-4, and IFN␥ and IL-4 from the 48 h supernatants of spleen cell cultures were titrated by ELISA using Pharmingen OptEIA sets. Serum amyloid A (SAA) was tested with a solid-phase sandwich mouse ELISA kit (PhaseTM Range, Tridelta Development Ltd.). 3.2. Antibodies Total IgG and IgM and specific antibodies to malondialdehyde (MDA)-modified LDL and to OVA were quantified using ELISA. For determination of total Ig, F(ab )2 fragments of anti-mouse IgG (1/500; Pierce) or anti-mouse IgM (Pharmingen) were coated on microtiter plates and used as capture antibodies. For determination of specific antibodies, OVA (20 ␮g/ml), native (10 ␮g/ml) or MDA-LDL (10 ␮g/ml) were coated overnight at 4 ◦ C. LDL were isolated from E◦ mouse blood by ultracentrifugation (Beckman 70.1Ti rotor, 45,000 rpm for 18 h at 4 ◦ C). Native LDL was obtained after EDTA removal with a sephadex PD10 column (Pharmacia). MDA-LDL was obtained as described [11]. Sera were plated for 2 h at room temperature, plates were washed, and the following alkaline phosphatase (AKP)-conjugated secondary antibodies were incubated for 1 h at room temperature: antimouse IgG1, anti-mouse IgG2a, anti-mouse IgM (Pharmingen), and anti-mouse IgG (Calbiochem). The AKP enzymatic activity was revealed by adding the p-nitrophenylphosphate disodic salt. Plates were read at 405 nm.

85 kPa. The injector and transfer line temperatures were set to 300 and 250 ◦ C, respectively. Samples (2 ␮l) were injected in splitless mode (30 s) and the oven programmed as follow: 50 ◦ C during 1 min, 50–300 ◦ C at 10 ◦ C/min and 10 min at 300 ◦ C. The operating conditions for EI-MS were source temperature at 150 ◦ C, filament emission current of 750 ␮A, ionising voltage of 70 eV and scan range from m/z 29 to m/z 650 with a periodicity of 1 s. 3.4. Statistical analysis Results are expressed as means ± S.E.M. Simple regression and ANOVA were performed using Statview 5.0 software (SAS Institute Inc., USA). Differences between groups were considered significant if p < 0.05.

4. Results 4.1. Atheroprotective effect of various adjuvants In order to compare the effect of various adjuvants (IFA, CFA, Alum, and CpG), three experiments were performed (Protocol #1). Animals repeatedly injected with IFA, CFA or Alum adjuvants presented a much less severe disease than their respective control littermates (Fig. 1). It became clear from the three experiments of Protocol #1 that the route of administration had no impact with regard to the atheroprotection conferred by adjuvants. Thus, they were exclusively injected i.p. in all the subsequent experiments. A statistically significant decrease in total cholesterol (TC) level was detected in mice injected with CFA and IFA (Table 1), pointing to a possible mechanism for the atheroprotective effect of the Freund’s adjuvants. However, while Alum

3.3. Gas chromatography–mass spectrometry analysis of IFA and mineral oil The two samples, a drop (12.6 mg) of mineral oil (ACROS ref 41508) and a drop (11.3 mg) of IFA, were diluted in 0.5 ml of hexane (SDS) 99% for trace analysis and injected in the chromatograph. The GC–MS system consisted of a HP5890 (Hewlett Packard) chromatograph interfaced by direct coupling to an INCOS 50 quadrupole mass spectrometer (Finnigan). The gas chromatograph was equipped with a 30 m × 0.25 mm i.d. fused-silica J&W column coated with 0.25 ␮m film of DB-5 poly (5% phenyl, 95% methyl siloxane). The carrier gas was helium with a head pressure of

Fig. 1. Effect of adjuvants on aortic lesions (Protocol #1). E◦ mice received chronic administrations of adjuvants (see Section 2) and were sacrificed at the age of 11 weeks (Exp #1) and 16 weeks (Exp #2 and #3). Extent of lesion (%) development was analyzed by computer-assisted morphometry on oil red-O-stained and hematoxylin counter-stained sections of the aortic root. Bars represent the mean ± S.E.M. per group (** p < 0.01; * p < 0.05 vs. Ctrl).

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

333

Table 1 Effect of adjuvants on body weight and total cholesterol (TC) levels TC (mg/dl)

r2 (TC vs. SAL)

nd nd

nd nd

Exp #2 Ctrl IFA CFA

658.9 ± 71.5 532.6 ± 23.5* 496.7 ± 32.1*

0.7 (p = 0.1512) 0.2 (p = 0.3182) 0.1 (p = 0.5489)

Exp #3 Ctrl Alum CpG

601.2 ± 51.2 581.3 ± 65.9 623.0 ± 62.7

0.3 (p = 0.4150) 0.4 (p = 0.2301) 0.1 (p = 0.7915)

Protocol #2 Ctrl CFA-OVA CFA-PBS IFA-OVA IFA-PBS

588.9 ± 43.3 451.3 ± 36.0** 378.7 ± 27.1** 482.9 ± 16.2** 487.4 ± 33.8**

0.3 (p = 0.2313)  0.2 (p = 0.3545) 0.4 (p = 0.1364) 0 3 (p = 0.2487) 0.0 (p = 0.9931)

Protocol #3 Ctrl IFA Mineral oil MMO LPS

503.7 ± 11.6 357.4 ± 22.9** 426.2 ± 44.3* 518.0 ± 34.3†† 526.7 ± 25.9††

0.0 (p = 0.6484)  0.7 (p = 0.0358) 0.0 (p = 0.9748) 0.1 (p = 0.4842) 0.6 (p = 0.0165)

Protocol #1 Exp #1 Ctrl IFA/CFA

Weight (g)

r2 (weight vs. SAL)

nd

25.7 ± 1.4 32.7 ± 1.0**

0.4 (p = 0.2574) 0.6 (p = 0.4132)

0.6 (p = 0.0800)

29.8 ± 1.2 28.8 ± 0.3 32.0 ± 0.6*,††

0.9 (p = 0.0520) 0.0 (p = 0.8250) 0.1 (p = 0.5812)

0.1 (p = 0.3097)

29.6 ± 1.1 28.5 ± 1.9 30.2 ± 1.6

0.7 (p = 0.0492) 0.2 (p = 0.635l) 0.5 (p = 0.3221)

0.4 (p = 0.0003)

29.3 ± 0.8 28.0 ± 1.0 29.7 ± 1.0 30.6 ± 0.8 29.4 ± 1.2

0.1 (p = 0.5677)  0.0 (p = 0.7166) 0.0 (p = 0.7826) 0.1 (p = 0.6587) 0.4 (p = 0.2052)

0.0 (p = 0.3221)

0.0 (p = 0.2374)

32.7 ± 0.9 30.6 ± 0.6 33.1 ± 0.7 30.0 ± 0.9* 31.8 ± 1.3

0.0 (p = 0.9903)  0.3 (p = 0.3035) 0.2 (p = 0.3331) 0.0 (p = 0.8788) 0.0 (p = 0.6394)

0.0 (p = 0.9869)

 



 0.5 (p = 0.0462)

 0.4 (p = 0.0535)

 0.3 (p = 0.0477)

r2 (TC vs. SAL): coefficient of regression between total cholesterol and the surface area of lesion; r2 (weight vs. SAL): coefficient of regression between body weight and the surface area of lesion. Braces: regression coefficients were also computed on merged groups within each experiment. For the correlations statistically significant or close to the statistical significance, numbers in bold indicate a positive correlation, and those in italic, a negative correlation. nd: not determined;* p < 0.05, ** p < 0.01 vs. Ctrl; † p < 0.05, †† p < 0.01 vs. IFA.

was also atheroprotective, it did not reduce TC. Furthermore, a regression analysis indicated that there was no statistically significant correlation between TC level or weight and lesion size (Table 1), suggesting that the atheroprotection observed for the Freund’s adjuvants was independent of their effects on TC. Also, in the case of CFA, mice indeed gained weight (Table 1). The CpG-injected mice displayed lesion surface area in the same order as the untreated mice (Fig. 1; Exp #3), indicating that this adjuvant, unlike all the Freund’s and Alum adjuvants, did not induce any modification in the disease process. Successive experiments were designed to clarify the atheroprotection conferred by the widely used Freund’s adjuvants. 4.2. Immune profiles imposed by adjuvants While it is classically admitted that CFA induces a Th1 bias and IFA a Th2 bias [12], the effect of repeated adjuvant injections on the Th polarization patterns is not known. A distinct Th polarization pattern can be of significant importance for atherosclerosis [13]. We, thus, assessed whether Freund’s adjuvants could exert protection through modulation of the Th balance.

New groups of mice received chronic administration of IFA or CFA, with or without ovalbumin as a model antigen not related to atherosclerosis (Protocol #2). As shown by the results of Protocol #1 (Fig. 1), both IFA and CFA injections led to a potent atheroprotective effect (Fig. 2), again associated with a reduction in TC, still not correlated with lesion size (Table 1). As expected, the presence of OVA did not modify the Freund’s adjuvant protection (Fig. 2). 4.2.1. Ig profiles Mice injected with OVA + Freund’s adjuvant showed an increase in IgG1, IgG2a, and IgM antibodies directed against OVA as compared to animals that were injected with PBS + Freund’s adjuvant or to untouched animals (Fig. 3a). All the groups exhibited similar titers of total IgG and IgM. Since the humoral immune response directed against oxidized-LDL might be atheroprotective [14,15], we assessed whether adjuvant injections modify the titer of antibodies directed against MDA-LDL. The IgG and IgM reactivities against nLDL were low and similar in all groups (Fig. 3b, insets and data not shown). The titer of anti-MDA LDL IgG was increased in the CFA-OVA and CFA-PBS-injected mice (Fig. 3b, left). Interestingly, both CFA- and IFA-injected mice had increased anti-MDA-LDL IgM as compared to control

334

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

Fig. 2. Effect of IFA and CFA ± OVA on development of lesions (Protocol #2). Mice received an i.p. injection per week of IFA or CFA, with or without ovalbumin (OVA). Extent of lesion development was analyzed as explained in the Fig. 1 legend. The micrographs are representative of the lesions observed in each group in the aortic root at 400 ␮m from the cusps origin (*** p < 0.001 vs. Ctrl).

Fig. 3. Effect of IFA and CFA ± OVA on the Th balance and Ig isotype (Protocol #2). (a) Serum titers of IgG and IgM (at dilutions indicated in parenthesis) were measured by ELISA. (b) Serum titers of IgG and IgM specific for mouse MDA-modified LDL was measured by ELISA. In the insets are shown the IgG (left) and IgM (right) reactivities to MDA-LDL (full line) and to native LDL (broken line) in control mice (*** p < 0.001; ** p < 0.01; * p < 0.05 vs. Ctrl; three randomized samples per group were used for ELISAs in (a) and (b)).

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

335

Fig. 4. In vitro proliferative response and cytokine production (Protocol #2). Spleen cells were cultured either with increasing doses of concanavalin (ConA) or ovalbumin (OVA) for 48 h. Fifty microlitres of supernatant were harvested from each well and 3 H-thymidine was added. (a) Thymidine incorporation was tested after additional 18 h of culture. (b) IFN␥ and IL-4 expression levels were evaluated by ELISA on the supernatants harvested at 48h on non-stimulated control cultures (non-stimulated) or stimulated with the highest dose of OVA (200 ␮g/ml) (** p < 0.01; * p < 0.05 vs. Ctrl; †† p < 0.01; † p < 0.05 vs. IFA; n = 7 per group).

mice (Fig. 3b, right). This increase was independent of the presence of OVA. In mice of the Exp #3 of the Protocol #1, there was no difference for the anti-MDA LDL IgG titers among the three groups (Ctrl: 0.296 ± 0.046; Alum: 0.454 ± 0.066; CpG: 0.339 ± 0.049, 405 nm absorbance ± S.E.M., at 1/80 dilution). Anti-MDA-LDL IgM were, however, higher in mice protected by the Alum injections (Ctrl: 0.233 ± 0.032; Alum: 0.404 ± 0.048* ; CpG: 0.245 ± 0.028; * p < 0.05 versus Ctrl, at 1/320 dilution). 4.2.2. Cell-mediated responses Proliferative capacities of spleen cells from the different groups were assessed in response to several stimuli. Unexpectedly, spleen cells from adjuvant-treated mice displayed a significantly reduced proliferative response to a mitogen, concanavalin A (ConA), as compared to untreated mice (Fig. 4a, left). This reduction was more pronounced in CFAthan in IFA-treated mice. However, despite their hyporesponsiveness to ConA, spleen cells from mice immunized with CFA-OVA proliferated vigorously in response to OVA while

spleen cells from IFA-OVA-injected mice did not (Fig. 4a, right), confirming that the complete but not the incomplete formulation of Freund’s adjuvant can induce delayed-type hypersensitivity reactions. The OVA immunization protocol was also designed to perform in vitro detection of T cell-specific cytokines released upon antigenic stimulation. Spleen cell culture from CFAOVA-immunized mice stimulated with OVA produced high amounts of IFN␥ and of IL-4 (Fig. 4b) indicating again that chronic CFA administration did not impose a clear Th1 polarization. Anti-OVA-specific T cells were actually biased towards the Th2 pathway by the IFA since spleen cells derived from IFA-OVA-treated mice produced IL-4 but not IFN␥ in response to OVA (Fig. 4b). 4.3. Identification of the compounds responsible for the FA effects Experiments described hitherto demonstrated that both IFA and CFA conferred the same degree of protection indicating that the M. tuberculosis cells in CFA do not play a

336

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

Fig. 5. Effect of LPS, MMO, mineral oil, and IFA on lesion development and on serum amyloid A levels (Protocol #3). Mice were injected i.p. once a week with various compounds. A group of untouched animals was used as control. (a) Extent of lesion development was analyzed as explained in the Fig. 1 legend. The micrographs are representative of the lesions observed in each group in the aortic root at 400 ␮m from the cusps origin (**** p < 0.0001 vs. Ctrl; †††† p < 0.0001; † p < 0.05 vs. IFA). (b) This graph was obtained by plotting the mean (±S.E.M.) of lesions (%) and the mean (±S.E.M.) SAA level for each group of mice.

role and that the components present in IFA are sufficient to confer atheroprotection. IFA is composed of mineral oil and MMO. The disease progression was assessed in MMO, mineral oil or IFA-injected mice. An additional group of mice was treated with LPS. Analysis of the lesion size in LPS- or MMO-injected mice did not show any difference as compared to the untouched group of mice (Fig. 5a). We reconfirmed the atheroprotective effect of IFA that reduced

lesions size by 70% (Fig. 5a). While injection of MMO did not induce any reduction in lesion size, mineral oil reduced it by 45% compared to control mice. These experiments point towards the mineral oil as the candidate component of IFA that exerts the atheroprotective effect. Both IFA and mineral oil reduced the TC (Table 1) that was again not correlated to the lesion size. LPS increased the cholesterol level, which was positively correlated to lesion size.

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

337

Table 2 Serum cytokine and serum amyloid A (SAA) levels at sacrifice in control mice and mice injected with IFA, mineral oil, MMO or LPS (Protocol #3)

IL-12 (pg/ml) IFN␥ (pg/ml) IL-4 (pg/ml) IL-10 (pg/ml) IL-6 (pg/ml) SAA (␮g/ml)

Ctrl (n = 6)

IFA (n = 7)

Mineral oil (n = 7)

MMO (n = 8)

LPS (n = 8)

126 ± 823 ± 198 ± 1316 ± 380 ± 36 ±

445 ± 712 ± 156 ± 1272 ± 367 ± 188 ±

583 ± 743 ± 178 ± 1205 ± 346 ± 210 ±

647 ± 677 ± 185 ± 1211 ± 355 ± 94 ±

1030 ± 756 ± 168 ± 1125 ± 331 ± 95 ±

30 42 12 69 14 16

145* 113 6* 150 22 54**

160* 69 11 78 8 39**

154* 34 8 84 16 40

209***,† 62 7* 48 12* 15

SAA: serum amyloid A. * p < 0.05 vs. Ctrl. ** p < 0.01 vs. Ctrl. *** p < 0.001 vs. Ctrl. † p < 0.05 vs. IFA.

Mice that received MMO had a significantly reduced body weight. In order to evaluate the inflammatory status, several cytokines were assessed in the serum (Table 2). Neither IL10 nor IFN␥ were modified by any treatment while IL-6 was reduced only in LPS-injected mice. As compared to control mice, serum IL-12 levels were increased in mice that received IFA, mineral oil, MMO or LPS. IL-4 serum concentrations were decreased in IFA- and LPS-treated mice while mineral oil and MMO did not modify it. Interestingly, the only modification that was both common and specific to IFA and to mineral oil groups concerned the SAA levels. Indeed, IFA and mineral oil increased it by 6.3- and 7-fold, respectively. Plotting the extent of lesions against the SAA levels for each group of mice (Fig. 5b) revealed that mice clustered into two groups: one with extended lesions but low SAA levels (Ctrl, MMO, LPS) and the other having smaller lesions but high SAA levels (IFA, mineral oil). 4.4. Adjuvants and serum lipoprotein patterns We then analyzed by FPLC whether the serum lipoprotein profiles were modified by IFA or its compounds. As shown in Fig. 6, MMO did not alter the lipoprotein profile as compared to control mice. On the other hand, IFA treatment led to a substantial decrease (−41%) of the CR/VLDL fraction. We observed that mineral oil and IFA had a distinct effect on the lipoprotein pattern. Indeed, mineral oil reduced LDL cholesterol by 21% and concomitantly increased HDL cholesterol by 56%. 4.5. Mineral oil composition The fact that for all the parameters tested hitherto, IFA and mineral oil had distinct effects raised the possibility that these compounds may indeed have few similarities. A gas chromatography–mass spectrometry analysis confirmed this hypothesis (Fig. 7). MMO is observed in low amount at the end of the chromatogram. The major compounds correspond to the mineral oil. Two series of saturated hydrocarbons have been clearly identified: straight-chain hydrocarbons (or “n-alkanes”) from C14 to C20, and acyclic isoprenoids pris-

tane and phytane. Compounds indicated by the symbol * are alkylcyclohexanes but the structure of the alkyl chain is not exactly determined. Other compounds are mainly methylalkanes. At variance with IFA, major compounds of the ACROS mineral oil (Fig. 7) are of lower volatility. The main constituents are compounds of the sterane series S22–S27 (characterized by the ion at m/z 217) and compounds of the terpane series T23–T31 (characterized by the ion at m/z 191). Straight-chain hydrocarbons from C17 to C21, and acyclic isoprenoids pristane and phytane are present in low amounts. Other compounds are chiefly methylalkanes. This analysis clearly showed that the mineral oil and the IFA, while displaying some similarities, have clearly distinct and complex compositions.

5. Discussion Emerging strategies of immunomodulation based on longterm immunization protocols involving repeated injections of adjuvants, which are aimed at modulating atherosclerosis are currently tested. Given that they can modulate certain immunoinflammatory disorders [1,2], we asked whether adjuvants could alter the course of atherosclerosis. Our results demonstrate that injection of E◦ mice with CFA, IFA and Alum leads to a dramatic decrease in the lesion size. The extent of protection conferred by these adjuvants does not appear to depend upon the disease stage (Procotol #1). Since CpG is the only adjuvant that did not perturb the disease progression, we propose that CpG should be the adjuvant of choice in long-term immunization protocols. This would ensure that the eventual effect observed on the atherosclerotic process is attributable to the immunointervention. Aluminum adjuvants, together with the calcium phosphate adjuvant, are the only ones licensed for human application. Aluminum-adjuvanted vaccines have been administered in human for more than 50 years. It is fascinating to speculate that Alum-based vaccines might have contributed to the reduction of cardiovascular disease these last decades by exerting an ‘atheroprotective side-effect’. The protective effect of CFA in our study was rather unexpected since other studies with CFA-based immunization

338

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

Fig. 6. Effect of MMO, mineral oil, and IFA on serum lipoprotein patterns (Protocol #3). FPLC separation was performed on randomized serum samples from untouched control mice (thick full line; Ctrl; n = 6), mice that received chronic administration of MMO (thin full line; n = 6), mineral oil (thin broken line; n = 4), or IFA (thick broken line; n = 5). Lines represent mean profiles and gray sections S.E.M. A magnification of the HDL profiles is shown in the inset. The chylomicron (CR)/VLDL, LDL, and HDL concentrations were computed from the FPLC profiles (**** p < 0.0001; *** p < 0.001; * p < 0.05 vs. Ctrl; †††† p < 0.0001; †† p < 0.01 vs. IFA).

protocols involving repeated injections in rabbits [16] and mice [17] have shown that the CFA further fortified by heatkilled preparation of M. tuberculosis induces a proatherogenic anti-heat shock protein (HSP) immune response. At present, we have no conclusive arguments to explain this discrepancy. Given that IFA had the same atheroprotective effect as CFA, the immune response directed against M. tuberculosis-derived HSPs is, however, unlikely to be implicated in adjuvant-mediated atheroprotection. Furthermore, the fact that atheroprotection was repeatedly found to be conferred by Freund’s adjuvants to a similar degree in five independent experiments in the present work, and that this effect was incidentally detected in three recent reports [18–20] consolidates our findings. Importantly, in one of these reports [19], CFA was found to be anti-atherogenic in LDLR◦ mice indicating that the atheroprotective feature of adjuvants is not restricted to the E◦ mouse model. While it is admitted that CFA induces a Th1 bias and IFA a Th2 bias [12], the effect of repeated adjuvant injections on the Th polarization patterns was not documented. A distinct Th polarization pattern can be of significance for atherosclerosis [13]. To evaluate this issue, E◦ mice were immunized with

ovalbumin. Ovalbumin was chosen as a model antigen instead of oxLDL, HSPs or other antigens related to atherosclerosis in order not to influence atherogenesis over the effect of adjuvants. First, the atheroprotective effect of Freund’s adjuvants was confirmed. Second, while CFA is described as a Th1-inducing agent in short-term experiments, we found it to promote Th2 responses as well in chronic immunization protocols. Indeed, CFA-OVA-treated animals had high levels of anti-OVA IgG1, attesting that B cells received a Th2 help. This was confirmed in in vitro spleen cell cultures from the same animals in which more than 10 pg/ml of IL-4 was detected after an OVA challenge. IFA increased the Th2 responses as expected. It also increased anti-OVA IgG2a titers indicating that Th1 responses were also stimulated, although we were not able to detect IFN␥ production in response to OVA in spleen cell cultures from IFA-OVA-treated mice. This indicates that either Th1 cells that helped B cells to produce IgG2a were eliminated or that the Th1 responses are rather modest and below our detection limit. Therefore, repeated injections of Freund’s adjuvants do not bias the Th response anymore. While adjuvants initially promote the production of pro-Th1 molecules, a sustained adjuvant-mediated

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

339

Fig. 7. Gas chromatography–mass spectrometry analysis of IFA and mineral oil. (a) Total ion chromatogram obtained from IFA and (b) from mineral oil ACROS (ref. 415080010) (Cx: straight-chain saturated hydrocarbons at x carbon atoms; Pr: pristane; Ph: phytane; Sx: steranes at x carbon atoms; Tx: terpanes at x carbon atoms; MMO: mannide monoleate; *: alkylcyclohexanes).

inflammation might rather stimulate a mixed production of pro-Th1 and pro-Th2 molecules, and therefore, invokes a mixed Th1/Th2 response. Our results also raise the possibility that atheroprotective adjuvants are those that can increase Th2 responses, regardless of their effect on the Th1 responses. Indeed, IFA, CFA, and Alum, all three being atheroprotective, increased the Th2 responses while CpG did not. While the anti-MDA-LDL IgG titers were increased only after CFA administration, both CFA and IFA significantly increased anti-MDA-LDL IgM. This latter increase could also contribute to the atheroprotective effect of Freund’s adjuvants [14,19,21]. This was corroborated by the fact that the atheroprotective Alum adjuvant also increased the anti-MDA-LDL IgM but not IgG. On the other hand, CpG injections had no effect on anti-MDA-LDL antibody titers. Interestingly, atheroprotection conferred by MDALDL-specific Th2 responses relies on the IL-5-dependent enhancement of the innate humoral response to oxLDL, which in aggregate provide protection from atherosclerosis [22]. Thus, the increase in Th2 responses and in the MDA-LDL IgM titers imposed by atheroprotective

adjuvants might be connected through an IL-5-dependent mechanism. Because they are the most commonly used adjuvants in experimental studies, we decided to determine which compound of the Freund’s adjuvants was endowed with atheroprotective properties. MMO was the most promising candidate because of its simple structure and its structural similarities with the lipid A of LPS. Indeed, LPS modulates experimental atherogenesis when administered weekly during 10 weeks at 50 ␮l/mouse/injection [23]. First, we found that at 10 times lower dose, though sufficient to dramatically increase serum IL-12 levels, LPS had no effect on lesion development. The discrepancy between our results and those reported by Ostos et al. [23] may depend upon different LPS dosage (5 ␮l versus 50 ␮l) and/or diet (0.15% cholesterol versus normal chow diet). Second, we found that MMO had no atheromodulating effect while mineral oil induced a 45% reduction in lesion size, suggesting that the mineral oil is the compound responsible for the atheroprotective effect of Freund’s adjuvants. We were, however, intrigued by several observations. We expected the effects of IFA recorded on

340

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341

various parameters to be an aggregate of the individual effects of mineral oil and MMO. This was true for some but not all parameters analyzed. For instance, IFA-reduced serum IL-4 levels while neither MMO nor mineral oil did. MMO and mineral oil led to a higher increase of serum IL-12 than IFA. Interestingly, the increase in SAA induced by IFA could be reproduced with mineral oil. While SAA could merely be a marker, these observations raised the possibility that SAA participates in the atheroprotection conferred by IFA and mineral oil. Indeed, during the acute-phase reaction, SAA may transform HDL into an acute-phase (SAA-enriched) proatherogenic HDL [24]. These observations, however, did not fit with our findings and urged us to further analyze modifications in the lipoprotein profiles eventually induced by adjuvants. We repeatedly found that IFA and CFA reduced the TC levels; a finding that was counter-balanced by the fact that TC levels never correlated with lesion size in individual groups. A significant association between TC and SAL was found in certain experiments (Protocol #1, Exp #2 and Protocol #2) when correlations were computed with the data from all groups in each experiment but emerged because of group clustering. FPLC analysis performed on serum lipoproteins from mice included in the Protocol #3 showed that IFA reduced the CR/VLDL fraction. None of its constituents tested individually could reproduce this effect. MMO had no effect on the lipoprotein profile and mineral oil decreased the LDL fraction and increased the HDL fraction. IFA and mineral oil thus imposed distinct but atheroprotective lipoprotein profiles that may contribute to their atheroprotective properties. Complex mechanisms underlying such an effect warrants further intense investigations. That mineral oil was not mimicking all effects of IFA, especially those on the lipoprotein profile, may rely on the fact that the mode of preparation of mineral oils results in lots that are unique. Indeed, the gas chromatography–mass spectrometry analysis performed on IFA and on the mineral oil from ACROS used in our experiments showed that the two mineral oils are complex mixtures and correspond to the so-called “maltene” fraction of petroleum characterized by its solubility in n-pentane [25]. The two fractions are rather different, the second being richer in polycyclic compounds. A similar analysis performed on another mineral oil (ref 124020010) from ACROS related to the one used in this study (same densities, viscosities, and refractive indexes) confirmed the high heterogeneity of these compounds (data not shown). In view of the complexity in their composition, determining the atheroprotective effect of each of the isolated components of mineral oils is beyond the scope of this study. However, the fact these atheroprotective oils were so heterogeneous and that Alum that does not contain any oil, was also atheroprotective points to a more general atheromodulating feature of adjuvants that may be related to their capacity to modulate inflammation. The current dogma proposes that an inflammatory vicious cycle maintains a continuous inflammatory cell infiltration in

the atherosclerotic lesion, which, in turn, confers to vascular cells an activated and proinflammatory deleterious phenotype. However, inflammation is in essence a protective process. Adjuvants are used for their capacity to stimulate non-specifically immunoinflammatory responses, an effect that we have observed in our experiments. However, animals with higher SAA titers developed smaller lesions. This indicates that enhancing the early inflammation can be beneficial. In support of this provocative point of view, a recent report has shown that disruption of a key regulator of inflammation, the NF-␬B transcription factor, in the macrophages leads to an increase in lesion size in LDLR◦ mice [26]. Therefore, inflammation might not be exclusively proatherogenic. Some inflammatory processes might be atheroprotective. In conclusion, we have shown that certain adjuvants have potent atheroprotective properties. The main objective of this study was not to elucidate the mechanism of action of adjuvants since their immunoregulatory effects depend on many variables that could not all be tested. However, we provide evidences showing that the atheroprotective effect of adjuvants may rely upon their capacity to increase Th2 responses, to increase anti-MDA-LDL IgM titers, and/or to impose atheroprotective serum lipoprotein profiles. Adjuvants may also instigate an atheroprotective inflammatory profile. Further studies are required to evaluate whether one of these mechanisms is predominantly involved in the atheroprotective effect of adjuvants.

Acknowledgements This work was supported by the Institut National de la Sant´e et de la Recherche M´edicale (INSERM) and the Swedish Research Council, project no. 71X-15075-01A. B Poirier was the recipient of a fellowship from the Groupe de Reflexion sur la Recherche Cardiovasculaire (GRRC) and E Tupin was the recipient of a fellowship from the ARCOL Foundation and received a research stipend from the Swedish Heart and Lung Foundation.

References [1] Zheng CL, Hossain MA, Kukita A, et al. Complete Freund’s adjuvant suppresses the development and progression of pristane-induced arthritis in rats. Clin Immunol 2002;103:204–9. [2] Zheng CL, Ohki K, Hossain A, et al. Complete Freund’s adjuvant promotes the increases of IFN-gamma and nitric oxide in suppressing chronic arthritis induced by pristane. Inflammation 2003;27:247– 55. [3] Freund J, McDermot K. Sensitization to horse serum by means of adjuvants. Proc Soc Exp Biol Med 1942;49:548. [4] Lindblad EB. Aluminium adjuvants. In: Stewart-Tull DES, editor. The theory and practical application of adjuvants. Portland: John Wiley and Sons Ltd.; 1995. p. 21–35. [5] Messina JP, Gilkeson GS, Pisetsky DS. Stimulation of in vitro murine lymphocyte proliferation by bacterial DNA. J Immunol 1991;147:1759–64.

J. Khallou-Laschet et al. / Atherosclerosis 184 (2006) 330–341 [6] Tokunaga T, Yamamoto H, Shimada S, et al. Antitumor activity of deoxyribonucleic acid fraction from Mycobacterium bovis BCG. I. Isolation, physicochemical characterization, and antitumor activity. J Natl Cancer Inst 1984;72:955–62. [7] Krieg AM, Yi AK, Matson S, et al. CpG motifs in bacterial DNA trigger direct B-cell activation. Nature 1995;374:546–9. [8] Bird AP. CpG islands as gene markers in the vertebrate nucleus. Trends Genet 1987;3:342–7. [9] Gullberg H, Rudling M, Forrest D, Angelin B, Vennstrom B. Thyroid hormone receptor beta-deficient mice show complete loss of the normal cholesterol 7alpha-hydroxylase (CYP7A) response to thyroid hormone but display enhanced resistance to dietary cholesterol. Mol Endocrinol 2000;14:1739–49. [10] Nicoletti A, Kaveri S, Caligiuri G, Bariety J, Hansson GK. Immunoglobulin treatment reduces atherosclerosis in apo E knockout mice. J Clin Invest 1998;102:910–8. [11] Nicoletti A, Caligiuri G, Tornberg I, et al. The macrophage scavenger receptor type A directs modified proteins to antigen presentation. Eur J Immunol 1999;29:512–21. [12] Yip HC, Karulin AY, Tary-Lehmann M, et al. Adjuvant-guided type1 and type-2 immunity: infectious/noninfectious dichotomy defines the class of response. J Immunol 1999;162:3942–9. [13] Laurat E, Poirier B, Tupin E, et al. In vivo downregulation of T helper cell 1 immune responses reduces atherogenesis in apolipoprotein E-knockout mice. Circulation 2001;104:197–202. [14] Zhou X, Caligiuri G, Hamsten A, Lefvert AK, Hansson GK. LDL immunization induces T-cell-dependent antibody formation and protection against atherosclerosis. Arterioscler Thromb Vasc Biol 2001;21:108–14. [15] Caligiuri G, Nicoletti A, Poirier B, Hansson GK. Protective immunity against atherosclerosis carried by B cells of hypercholesterolemic mice. J Clin Invest 2002;109:745–53. [16] Xu Q, Dietrich H, Steiner HJ, et al. Induction of arteriosclerosis in normocholesterolemic rabbits by immunization with heat shock protein 65. Arterioscler Thromb 1992;12:789–99.

341

[17] George J, Shoenfeld Y, Afek A, et al. Enhanced fatty streak formation in C57BL/6J mice by immunization with heat shock protein-65. Arterioscler Thromb Vasc Biol 1999;19:505–10. [18] Mallat Z, Gojova A, Brun V, et al. Induction of a regulatory T cell type 1 response reduces the development of atherosclerosis in apolipoprotein E-knockout mice. Circulation 2003;108:1232– 7. [19] Binder CJ, Horkko S, Dewan A, et al. Pneumococcal vaccination decreases atherosclerotic lesion formation: molecular mimicry between Streptococcus pneumoniae and oxidized LDL. Nat Med 2003;9:736–43. [20] Hansen PR, Chew M, Zhou J, et al. Freunds adjuvant alone is antiatherogenic in apoE-deficient mice and specific immunization against TNFalpha confers no additional benefit. Atherosclerosis 2001;158:87–94. [21] George J, Afek A, Gilburd B, et al. Hyperimmunization of apo-E-deficient mice with homologous malondialdehyde lowdensity lipoprotein suppresses early atherogenesis. Atherosclerosis 1998;138:147–52. [22] Binder CJ, Hartvigsen K, Chang MK, et al. IL-5 links adaptive and natural immunity specific for epitopes of oxidized LDL and protects from atherosclerosis. J Clin Invest 2004;114:427–37. [23] Ostos MA, Recalde D, Zakin MM, Scott-Algara D. Implication of natural killer T cells in atherosclerosis development during a LPSinduced chronic inflammation. FEBS Lett 2002;519:23–9. [24] Artl A, Marsche G, Lestavel S, Sattler W, Malle E. Role of serum amyloid A during metabolism of acute-phase HDL by macrophages. Arterioscler Thromb Vasc Biol 2000;20:763–72. [25] G¨urgey K. Geochemical effects of asphaltene separation procedures: changes in sterane, terpane, and methylalkane distributions in maltenes and asphaltene co-precipitates. Org Geochem 1998;29:1139–47. [26] Kanters E, Pasparakis M, Gijbels MJ, et al. Inhibition of NF-kappaB activation in macrophages increases atherosclerosis in LDL receptordeficient mice. J Clin Invest 2003;112:1176–85.