Apolipoprotein A-I gene transfer exerts immunomodulatory effects and reduces vascular inflammation and fibrosis in ob/ob mice

Obesity is associated with vascular inflammation, fibrosis and reduced high-density lipoproteins (HDL)-cholesterol. We aimed to investigate whether adenoviral gene transfer with human apolipoprotein (apo) A-I (Ad.A-I), the main apo of HDL, could exert immunomodulatory effects and counteract vascular inflammation and fibrosis in ob/ob mice. Ad.A-I transfer was performed in 8 weeks (w) old ob/ob mice, which were sacrificed 7 w later. The aorta was excised for mRNA analysis and the spleen for splenocyte isolation for subsequent flow cytometry and co-culture with murine fibroblasts. HDL was added to mononuclear cells (MNC) and fibroblasts to assess their impact on adhesion capacity and collagen deposition, respectively. Ad.A-I led to a 1.8-fold (p < 0.05) increase in HDL-cholesterol versus control ob/ob mice at the day of sacrifice, which was paralleled by a decrease in aortic TNF-α and VCAM-1 mRNA expression. Pre-culture of MNC with HDL decreased their adhesion to TNF-α-activated HAEC. Ad.A-I exerted immunomodulatory effects as evidenced by a downregulation of aortic NOD2 and NLRP3 mRNA expression and by a 12 %, 6.9 %, and 15 % decrease of the induced proliferation/activity of total splenic MNC, CD4+, and CD8+ cells in ob/ob Ad.A-I versus control ob/ob mice, respectively (p < 0.05). Ad.A-I further reduced aortic collagen I and III mRNA expression by 62 % and 66 %, respectively (p < 0.0005), and abrogated the potential of ob/ob splenocytes to induce the collagen content in murine fibroblasts upon co-culture. Finally, HDL decreased the TGF-ß1-induced collagen deposition of murine fibroblasts in vitro. Apo A-I transfer counteracts vascular inflammation and fibrosis in ob/ob mice.

Obesity is associated with low high-density lipoprotein (HDL) cholesterol levels, by which the body weight inversely correlates with HDL cholesterol [18]. HDL cholesterol levels are also decreased in several other inflammatory disorders including atherosclerosis [19], systemic lupus erythematosus [20], and rheumatoid arthritis [21], suggesting a link between HDL and the immune response. Apolipoprotein A-I (apo A-I) is the principal apolipoprotein of HDL and plasma levels of this apolipoprotein strongly correlate with HDL cholesterol levels. At the molecular and cellular level, HDL/ apo A-I are known to reduce Toll like receptor 4 expression [22], inhibit antigen presentation [23], and reduce T cell proliferation [24]. Besides these immunomodulatory characteristics, HDL/apo A-I are particularly known for their endothelial-protective properties including their capacity to facilitate vascular relaxation via activation of eNOS [25], to restore impaired NO bioavailability [26], and to decrease the expression of vascular adhesion molecules [27]. Besides these well-established endothelialprotective effects, there is accumulating evidence that HDL and apo A-I also have anti-fibrotic properties: overexpression of apo A-I in the lung abrogates fibrosis in experimental silicosis [28], low apo A-I levels predict liver fibrosis in hepatitis C patients [29], and HDL inversely correlate to the serum marker of cardiac fibrosis, procollagen type III aminopeptide [30]. Furthermore, we demonstrated that human apo A-I gene transfer decreased cardiac fibrosis in an experimental rat model of diabetic cardiomyopathy [27]. With respect to the aorta, it has been shown that infusions of apo A-I mimetic peptide reduce fibrosis in the aortic root in a mouse model of aortic valve stenosis [31]. Recently, we showed that HDL supplementation to human aortic endothelial cells (HAEC) decreased transforming growth factor (TGF)-ß1-induced EndMT [14].
Given the endothelial-protective, anti-fibrotic, and immunomodulatory effects of HDL/apo A-I on the one hand, and the occurrence of vascular inflammation and fibrosis, and innate immune activation under obese conditions on the other hand, we aimed to investigate whether human apo A-I gene transfer can decrease aortic inflammation and fibrosis in leptin-deficient ob/ob mice, which are overweight [32], insulin-resistant [33], and develop vascular inflammation and fibrosis [34,35]. We next aimed to understand whether apo A-I/HDL might also affect vascular inflammation and fibrosis via systemic immunomodulation. In this respect, in vitro studies were conducted to get insights in whether (i) HDL modulate the binding capacity of MNC to TNF-αactivated HAEC; and whether (ii) modulation of splenic activity following apo A-I transfer influences collagen production upon co-culture of those splenocytes with fibroblasts.

Animals and study design
Male B6.V-Lepod/JRj mice (ob/ob; Janvier Labs, Le Genest-Saint-Isle, France) were intravenously injected via the tail vein with saline or with 5 x 10 10 total particles of the E1E3E4-deleted adenoviral vector Ad.A-I expressing human apo A-I [36] or of Ad.Null containing no expression cassette [36], at the age of 8 weeks. As control reference, age-matched male C57BL/6 mice were injected with saline. Mice were sacrificed 7 weeks later, the aorta was excised and snap-frozen in liquid nitrogen for molecular biology and the spleen was isolated for splenocyte preparation and subsequent flow cytometry or co-culture with murine fibroblasts. Blood was withdrawn from the retro-orbital plexus at day 7, 21, 35,

Human apo A-I ELISA
Human apo A-I levels were determined by sandwich ELISA as described previously [37].

Plasma lipid analysis
Mouse lipoproteins were separated by density gradient ultracentrifugation as described by Jacobs et al. [38]. Fractions were stored at −20°C until analysis. Cholesterol in lipoprotein fractions was determined with commercially available enzymes (Roche Diagnostics, Basel, Switzerland). Precipath L (Roche Diagnostics) was used as a standard.

Glucose tolerance test
Glucose tolerance test was performed by intraperitoneal injection of glucose (2 g/kg) after 6 h (h) of fasting as described by Hofmann et al. [39]. Tail blood glucose levels were measured with an Accu-Chek® Active Glucometer (Roche Applied Science, Penzberg, Germany) before (0 min) and at 15, 30, 60 and 120 min after injection.

Splenocyte isolation
Splenocytes were isolated from the different experimental groups as described previously [40].
(Splenocyte) mononuclear cell, CD4-, and CD8-T cell proliferation Splenocytes were labeled with 10 μM of succinimidyl ester of carboxyfluorescein diacetate (CFSE Cell Tract™; Invitrogen, Carlbard, CA, USA) to be able to measure cell proliferation, indicative for MNC activation, as described previously [41]. Therefore, splenocytes were cultured in RPMI1640 medium (Invitrogen, Heidelberg, Germany), supplemented with 10 % FBS and 1 % penicillin/streptomycin for 72 h. Then, cells were stained with monoclonal anti-CD4 or anti-CD8 antibodies (BD Biosciences, Franklin Lakes, NJ, USA), followed by flow cytometry on a MACSQuant Analyzer (Miltenyi Biotec, Bergisch Gladbach, Germany) and analysis with FlowJo 8.7. software (Tree Star) for the calculation of the division index, i.e. the average number of cell divisions that the responding cells undergo (i.e., ignores peak 0).

Cell culture
Murine C4 fibroblasts, a murine fibroblast cell line derived from embryonic BALB/c mice by SV40 infection in vitro [42], were cultured in Basal Iscove medium supplemented with 10 % FBS and 1 % Penicillin/Streptomycin. Twentyfour h after plating at a cell density of 10,000 cells/96-well, cells were stimulated with/out 10 ng/ml TGF-ß1 (R&D Systems, Minneapolis, USA) under serum starvation conditions, i.e. Basal Iscove medium with 0.01 % FBS and 1 % Pencillin/Streptomycin, in the presence or absence of 50 μg/ml of HDL (MP Biomedicals, Solon, Ohio, USA) for 24 h. Next, cells were fixed in methanol overnight at −20°C for subsequent collagen deposition staining.
Human aortic endothelial cells (HAEC; Lonza, Basel, Switzerland) were cultured in Endothelial Basal Medium (Lonza) supplemented with the EGM-2 BulletKit (Lonza) on pre-coated (coating solution containing 0.02 % gelatin from bovine skin type B and 125 ng/ml fibronectin from bovine plasma; both Merck Chemicals, Darmstadt; Germany) cell culture flasks.

Co-culture of fibroblasts with splenocytes
Murine C4 fibroblasts were plated at a cell density of 10,000 cells/96-well in Basal Iscove medium supplemented with 10 % FBS and 1 % Penicillin/Streptomycin. Twenty-four h after plating, medium was removed and splenocytes isolated from control, ob/ob, ob/ob Ad.Null, and ob/ob Ad.A-I mice were added to the fibroblasts at a ratio of 10 splenic cells to 1 fibroblast in RPMI1640 medium (Invitrogen, Heidelberg, Germany) 10 % FBS, 1 % Penicillin/Streptomycin. After 24 h, splenocytes were removed and the murine fibroblasts were fixed in methanol overnight at −20°C for subsequent collagen deposition staining.

Quantification of collagen deposition
Collagen deposition was determined as described previously [14,43]. In brief, after overnight fixation in methanol at −20°C, cells were washed once with PBS and incubated in 0.1 % Direct Red 80 (Sirius red; Merck Chemicals, Darmstadt; Germany) staining solution at RT for 60 min. After second washing with PBS, the Sirius red staining of the murine fibroblasts was eluted in 0.1 N sodium hydroxide at RT for 60 min on a rocking platform. The optical density, representative for the accumulation of collagen, was measured at 540 nm. For normalization to cell amount, murine fibroblasts were stained with crystal violet (Merck Chemicals, Darmstadt, Germany) and the absorbance was measured at 495 nm. Data are represented as the ratio of the absorbance at 540 nm (Sirius Red) towards the absorbance at 495 nm (crystal violet).

Human peripheral blood mononuclear cell isolation
Blood was withdrawn from healthy donors (aged 40 to 65) with approval (EA4/115/11) from the Ethical Commission (Charité, CBF, Berlin). Human peripheral blood mononuclear cells (MNC) were isolated from the blood samples by density-gradient centrifugation (Biocoll; Merck Millipore, Darmstadt, Germany). Cells were stored in liquid nitrogen until use.

Adhesion assay
HAEC were plated in gelatin/fibronectin-coated black solid bottom 96-well plates, cultured for 24 h and treated with 10 ng/ml TNF-α (BD Pharmingen, Franklin Lakes, NJ, USA) in Endothelial Basal Medium for 4 h prior to addition of MNC. DiO-labeled human MNC (DiO, Molecular Probes, Life Technologies) were cultured with/out 50 μg/ml of HDL for 24 h in 24-well plates and next added to TNF-α-stimulated HAEC at a ratio of 1:10 for 30 min. Subsequently, MNC were removed by aspiration and HAEC were washed once with PBS. PBS was added to each well and DiO-fluorescence was detected with a Mithras LB 940 multitechnology microplate reader (Berthold technologies) at wavelengths of 485 nm for excitation and 535 nm for emission and corresponding software (MikroWin 2000, Mikrotek Laborsysteme GmbH).

Statistical analysis
Data are presented as mean ± SEM. Statistical differences between groups were assessed with Ordinary One-way ANOVA. Differences were considered to be significant when the P-value was lower than 0.05.

Metabolic modulation after human apo A-I transfer in ob/ ob mice
Human apo A-I gene transfer resulted in persistent expression of human apo A-I in ob/ob mice (Fig. 1) leading to a 79 % (p < 0.001) and 68 % (p < 0.001) increase in HDL cholesterol levels at the day of sacrifice compared to ob/ob and ob/ob Ad.Null mice, respectively (Table 1). At this age, ob/ob mice were still normoglycemic (Additional file 1: Figure S1A), but already glucose intolerant as evidenced by an intraperitoneal glucose tolerance test (Additional file 1: Figure S1B). Ad.A-I transfer did not improve glucose intolerance (Additional file 1: Figure S1B).
Human apo A-I transfer exerts anti-inflammatory and immunomodulatory effects in ob/ob mice Given the occurrence of inflammation in the vasculature under obesity on the one hand [1] and the antiinflammatory properties of HDL/apo A-I on the other hand [27,44], we first evaluated whether apo A-I gene transfer could reduce aortic inflammation in obese mice.

Anti-fibrotic effects after human apo A-I transfer in ob/ob mice
The relevance of vascular fibrosis in obesity [3] and the anti-fibrotic effects of HDL/apo A-I [14,[27][28][29][30][31] further motivated us to evaluate the impact of Ad.A-I on the aortic mRNA expression of fibrosis markers. Ob/ob mice exhibited a 459 % (p < 0.0001) and 243 % (p < 0.0005) higher collagen I and III mRNA expression, respectively, compared to control C57BL/6 mice, whereas AdA-I transfer in ob/ob mice resulted in a 62 % (p < 0.0005) decrease of collagen I mRNA and a 66 % (p < 0.0005) decrease of collagen III mRNA (Fig. 6). Given the link between immune cells and fibrosis on the one hand [13] and the importance of fibrosis in vascular remodeling on the other hand [45], we next evaluated the impact of the immunomodulatory effects of apo A-I gene transfer on collagen production in fibroblasts upon co-culture of splenocytes isolated from the different experimental groups on fibroblasts. Coculture of splenocytes from ob/ob and ob/ob Ad.Null mice with murine fibroblasts augmented collagen deposition by 54 % (p < 0.0001) and 50 % (p < 0.0005), respectively, compared to monoculture of fibroblasts. In contrast, splenocytes from ob/ob Ad.A-I mice and from control mice did not induce collagen content in fibroblasts upon co-culture (Fig. 7a). To evaluate whether HDL themselves reduce collagen deposition in murine fibroblasts, HDL were supplemented to fibroblasts stimulated with TGF-ß1. TGF-ß1 increased the collagen deposition in murine fibroblasts by 43 % (p < 0.01). In contrast, HDL reduced the TGF-ß1-induced collagen deposition to levels not different from controls (Fig. 7b).

Discussion
The salient findings of the present study are that apo A-I gene transfer exerts immunomodulatory effects and Leptin-deficient ob/ob mice are overweight [32], insulinresistant [33], and develop type 2 diabetes mellitus over time due to deficiency of the appetite regulating hormone leptin, which has immunomodulatory properties [46]. In agreement with Xu et al. [47], ob/ob mice were still normoglycemic at an age of 15 weeks. However, a glucose tolerance test 4 days before sacrifice indicated a disturbance in glucose metabolism, which could not be overcome via apo A-I gene transfer. Furthermore, ob/ob mice exhibited hypercholesterolemia and dyslipoproteinemia (non-HDL to HDL-cholesterol ratio 1:1) in the absence of raised triglyceride levels. In agreement with apo A-I gene transfer studies in certain atherosclerosis models [48,49], Ad.A-I transfer induced increased triglyceride levels in ob/ ob mice. Under these metabolic conditions, we demonstrate that apo A-I gene transfer led to a downregulation of aortic TNF-α and VCAM-1 mRNA in ob/ob mice. We and others have shown that apo A-I reduces TNF-α expression [27,50], which is an important inducer of VCAM-1 expression [44]. Downregulation of aortic VCAM-1 mRNA expression in ob/ob mice after apo A-I gene transfer is -on its turn -in agreement with the wellestablished potential of apo A-I and HDL to decrease the expression of cytokine-induced adhesion molecules [44]. This downregulated aortic VCAM-1 together with the decreased CD4 and CD8a mRNA expression in ob/ob Fig. 6 Impact of apo A-I gene transfer on aortic collagen I and III mRNA expression in ob/ob mice. Bar graphs represent the mean ± SEM of aortic a. collagen I and b. collagen III mRNA expression with n = 7 for C57BL/6, n = 9-10 for ob/ob groups; *p < 0.05, **p < 0.0005, and ***p < 0.0001 Fig. 7 High-density lipoproteins reduce collagen deposition in murine fibroblasts via immunomodulatory and direct anti-fibrotic properties. Bar graphs represent the mean ± SEM of the ratio of the absorbance at 540 nm of Sirius Red-stained murine fibroblasts towards the absorbance at 495 nm of crystal violet-stained murine fibroblasts cultured a) in the absence (basal) or presence of splenocytes isolated from C57BL/6, ob/ob, ob/ob Ad.Null and ob/ob Ad.A-I mice, as indicated, at a ratio of 10 splenocytes to 1 fibroblast for 24 h, with n = 5-6/group and *p < 0.05, **p < 0.005, ***p < 0.0005, and ****p < 0.0001 and b) depict the mean ± SEM of the ratio of the absorbance at 540 nm of Sirius Red-stained murine fibroblasts towards the absorbance at 495 nm of crystal violet-stained murine fibroblasts in the presence of TGF-ß1 with/out HDL supplementation for 24 h, with the different groups as indicated: n = 12/control, HDL, and TGF-ß1 + HDL groups and n = 9/TGF-ß1 group with *p < 0.05, **p < 0.01 and ****p < 0.0001 Ad.A-I compared to ob/ob mice suggests a lower presence of CD4 and CD8 cells in the aorta of ob/ob Ad.A-I compared to ob/ob mice. Our in vitro findings further illustrate that HDL lower the capacity of MNC to adhere to TNF-α-activated aortic endothelial cells. This implies that HDL can decrease the adhesion of immune cells to the aortic endothelium via their endothelial-protective (lowering VCAM-1 expression) as well as immunomodulating capacity (decreasing the binding capacity of MNC to activated endothelial cells). This impaired immune cell adhesion and consequently lower presence of immune cells in the vascular wall, were expected to reduce fibrosis in the aorta of ob/ob Ad.A-I versus ob/ob mice, due to the lower burden of inflammation [13]. Indeed, we found that the aortic collagen I and III mRNA expression was lower in ob/ob Ad.A-I versus ob/ob mice, which can also be explained via a direct anti-fibrotic effect of HDL. This mechanism is corroborated by our in vitro finding indicating that HDL reduce the TGF-ß1-induced collagen deposition in murine fibroblasts and is further supported by our recent observation that HDL are capable of decreasing TGF-ß1-induced EndMT in aortic endothelial cells in vitro [14]. In this context, another explanation for the reduction in aortic collagen expression following apo A-I gene transfer could be the decrease in TGF-ß1-expressing splenic MNC in ob/ob Ad.A-I versus ob/ob mice, and subsequent lower induction of EndMT in ob/ ob mice.
Evidence for the immunomodulatory capacity of apo A-I gene transfer in ob/ob mice follows at first from the reduction in aortic mRNA expression of the NLR proteins NOD2 and NLRP3 in ob/ob Ad.A-I versus ob/ob mice. NOD2 leads to NF-kB activation and subsequent expression of TNF-α and NLRP3 [51]. Its downregulation following apo A-I gene transfer can therefore partly explain the reduction in aortic TNF-α expression. Based on the involvement of NOD2 in fibrosis [11], its endothelial expression [9] and the capacity of HDL to reduce EndMT [14], we further speculate that an increase of HDL via apo A-I gene transfer can partly account for the reduction in aortic collagen I and III expression via decreasing EndMT, involving NOD2. The aortic downregulation of NLRP3, which has emerged as an important regulator of inflammation in metabolic disorders and atherosclerosis [6,7] and is recognized for its capacity to induce collagen production in fibroblasts [12], supports the link between the decrease in vascular inflammation and fibrosis after apo A-I gene transfer.
Besides its impact on the aortic expression of NLRs, Ad.A-I transfer led to systemic immunomodulatory effects as indicated by a decline in the increased proliferation/activity of splenic MNC, CD4+ and CD8+ T cells in ob/ob mice. This finding is supported by Wilhelm et al. [24] who showed that apo A-I prevents T cell activation and proliferation in peripheral lymph nodes in mice fed an atherogenic diet. In addition, we found that apo A-I gene transfer reduced the percentage of splenic MNC expressing the pro-fibrotic factor TGF-ß1. Concomitantly, co-culture of splenocytes from ob/ob Ad.A-I mice did not induce collagen production in murine fibroblasts, in contrast to splenocytes from ob/ob and ob/ob Ad.Null mice, suggesting that apo A-I-raising transfer reduces the pro-fibrotic potential of splenocytes under obese conditions and their subsequent contribution to vascular fibrosis. This hypothesis, supporting the existence of a vasculosplenic axis, i.e. the migration of immune cells from the spleen to the vasculature and subsequent involvement in vascular fibrosis builds further on the existence of the cardiosplenic axis, which importance has mainly been recognized in ischemic heart disease [52]. This postulation should be taken with caution as long as the existence of a vasculosplenic axis and its relevance in vascular fibrosis is confirmed in additional experiments in ob/ob mice.

Limitations of the study
Whereas the present study reveals systemic immunomodulatory and local (aorta) immunomodulatory/antifibrotic effects following apo A-I transfer in ob/ob mice, further characterization of the systemic immunomodulatory effects mice including cytokine profiles of circulating immune cells as well as the analysis of fibrosis in other vascular beds and/or adipose tissue is required to deepen our insights into the link between inflammation and fibrosis in ob/ob mice and the impact of apo A-I transfer on both parameters. Furthermore, experiments in high-fat diet-induced obesity, associated with hyperleptineamia (dysfunctional leptin), would be of value to confirm our findings.

Conclusions
We demonstrated that apo A-I-raising transfer exerts immunomodulatory effects in ob/ob mice, including the reduction in the aortic expression of the pattern recognition receptors NOD2 and NLRP3, and the decrease in activity of splenic MNC, which are associated with a decrease in vascular inflammation and fibrosis. These findings further corroborate the immunomodulatory and vascular-protective potential of apo A-I/HDL. Furthermore, they point out that under obese conditions, modulation of the chronic low-grade activation of the innate immune system (NOD2, NLRP3) per se, may counteract vascular inflammation and fibrosis.

Additional file
Additional file 1: Figure S1. Impact of apo A-I transfer on blood glucose levels and glucose responsiveness in ob/ob mice. (TIFF 1521 kb)