PI3K p110γ Deletion Attenuates Murine Atherosclerosis by Reducing Macrophage Proliferation but Not Polarization or Apoptosis in Lesions

Atherosclerosis is an inflammatory disease regulated by infiltrating monocytes and T cells, among other cell types. Macrophage recruitment to atherosclerotic lesions is controlled by monocyte infiltration into plaques. Once in the lesion, macrophage proliferation in situ, apoptosis, and differentiation to an inflammatory (M1) or anti-inflammatory phenotype (M2) are involved in progression to advanced atherosclerotic lesions. We studied the role of phosphoinositol-3-kinase (PI3K) p110γ in the regulation of in situ apoptosis, macrophage proliferation and polarization towards M1 or M2 phenotypes in atherosclerotic lesions. We analyzed atherosclerosis development in LDLR−/−p110γ+/− and LDLR−/−p110γ−/− mice, and performed expression and functional assays in tissues and primary cells from these and from p110γ+/− and p110γ−/− mice. Lack of p110γ in LDLR−/− mice reduces the atherosclerosis burden. Atherosclerotic lesions in fat-fed LDLR−/−p110γ−/− mice were smaller than in LDLR−/−p110γ+/− controls, which coincided with decreased macrophage proliferation in LDLR−/−p110γ−/− mouse lesions. This proliferation defect was also observed in p110γ−/− bone marrow-derived macrophages (BMM) stimulated with macrophage colony-stimulating factor (M-CSF), and was associated with higher intracellular cyclic adenosine monophosphate (cAMP) levels. In contrast, T cell proliferation was unaffected in LDLR−/−p110γ−/− mice. Moreover, p110γ deficiency did not affect macrophage polarization towards the M1 or M2 phenotypes or apoptosis in atherosclerotic plaques, or polarization in cultured BMM. Our results suggest that higher cAMP levels and the ensuing inhibition of macrophage proliferation contribute to atheroprotection in LDLR−/− mice lacking p110γ. Nonetheless, p110γ deletion does not appear to be involved in apoptosis, in macrophage polarization or in T cell proliferation.


Introduction
Atherosclerosis has traditionally been considered a disorder of cholesterol metabolism that results in lipid accumulation in the arterial wall, provoking artery wall thickening. It shares features of chronic inflammatory diseases, such as infiltration of activated immune cells into the artery wall [1], [2]. Early in the disease, oxidized low-density lipoproteins (oxLDL) that have accumulated in the intima activate endothelial cells; these secrete a number of pro-inflammatory molecules that recruit specific leukocyte types into the artery wall [3]. Monocyte/macrophages accumulate preferentially in atherosclerotic plaque, although other infiltrate components such as effector T cells, mast cells, dendritic cells and neutrophils also contribute to inflammation [3], [4], [5]. Small numbers of Foxp3 + regulatory T (Treg) cells, which mediate atheroprotection [5], are also present in plaques [6]. In early atherosclerotic lesions, most monocytes differentiate to macrophages due to the effect of macrophage colony-stimulating factor (M-CSF) and other mediators of innate and acquired immunity [7]. Neointimal macrophages internalize lipoproteins to become foam cells, which contribute to lipoprotein modification and retention, enhancing atherosclerosis progression [4], [7]. Macrophage and T lymphocyte activation lead to the release of additional mediators, including cytokines, chemokines and growth factors [1], [8]. This chronic inflammatory environment promotes progression of early lesions (or fatty streaks) to complex lesions (or advanced plaques) that protrude into the arterial lumen and can trigger atherothrombotic vascular disease [1], [3].
Macrophages are a heterogeneous cell population, able to adapt their physiology in response to a variety of microenvironmental situations. There are thought to be two main phenotypes; classically activated macrophages (M1) are pro-inflammatory, whereas alternatively-activated macrophages (M2) contribute to wound healing and regulation of inflammatory processes [9]. Granulocyte and macrophage colony-stimulating factor (GM-CSF)-stimulated bone marrow precursors generate cells of the M1 phenotype, whereas M-CSF promotes the M2 phenotype [10], [11]; studies describe both cell types in human and murine atherosclerotic lesions [12]. A recent report nonetheless showed predominance of infiltrating M2 macrophages in lesions in young apolipoprotein E (ApoE)-deficient mice, while M1 macrophages dominated in those of aged ApoE-deficient mice; further analysis suggested M2-to-M1 transition in the lesions [13].
Macrophage number in the lesions is controlled mainly by monocyte migration into plaques and, to a lesser extent, by macrophage apoptosis and by local macrophage proliferation [14], [15], [16]. Macrophage apoptosis has contrasting roles in plaque progression; in early lesions, it limits lesion cellularity, whereas in advanced lesions, it promotes development of the necrotic core, a high-risk factor for thrombosis [16]. Proliferation of infiltrating macrophages in early atherosclerotic plaque fosters lesion progression to a more advanced stage [14], [15], [17]. In lesions, modified LDL (low-density lipoproteins) induce GM-CSF release by infiltrating macrophages and by vascular endothelial and smooth muscle cells, which activates macrophage proliferation [17], [18], [19]. Although GM-CSF and phosphoinositide 3-kinase (PI3K) are implicated in macrophage proliferation in vitro [17], [20], Chang et al. did not detect GM-CSF by in situ hybridization in atherosclerotic plaque sections from ApoE-deficient mice [21]. M-CSF secreted by aortic endothelial cells also promotes macrophage proliferation in atherosclerotic lesions [14]. In murine bone marrow-derived macrophages (BMM) and in human monocytes, M-CSF induces recruitment of the PI3K p85a regulatory subunit to the M-CSF receptor, activating PI3K [22], [23], [24].
p110c is expressed mainly in hematopoietic cells. p110c 2/2 mouse neutrophils have severely impaired function and migration; these mice also show reduced mast cell degranulation [25], lower thymocyte numbers and defective T cell function in vitro and in vivo [26], [27], [28]. Germ-line deletion of p110c in ApoE 2/2 mice attenuates murine atherosclerosis [21]. In vitro and in vivo experiments showed that p110c is necessary for Akt activation in Figure 1. Macrophage and T cell infiltration in lesions of LDLR 2/2 p110c 2/2 compared to LDLR 2/2 p110c +/2 mice. Aortic sinus sections were studied in LDLR 2/2 p110c +/2 (females, n = 6) and LDLR 2/2 p110c 2/2 mice (females, n = 7) after two months on a high-fat diet. (A) Representative photomicrographs of Mac-3 + cells in aortic sinus sections after immunohistochemical staining. Bar = 200 mm. Arrows indicate Mac-3 + area. (B) Percentage of Mac-3 + -stained area relative to total lesion area, quantified with ImageJ software. (C) Representative photomicrographs of CD3 + cells in aortic sinus sections after immunohistochemical staining. Bar = 200 mm. (D) Percentage of CD3 + cells relative to total lesion area, quantified with ImageJ. (E) Representative photomicrographs of immunofluorescent staining for vascular smooth muscle cells (aSMA + ) in aortic sinus sections from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice after two months on a high-fat diet (n = 6 females/genotype). Bar = 100 mm. (F) Percentage of aSMA + area relative to total lesion area, quantified with ImageJ. Mean  macrophages in response to oxLDL, atherogenic chemokines, and angiotensin II activation [21]. Pharmacological inhibition of p110c alleviates atherosclerotic plaque development in ApoE 2/2 and LDLR 2/2 mice; its deletion in hematopoietic cells decreases macrophage and T cell numbers in plaque [29]. The in vivo mechanism underlying this reduced inflammatory cell infiltration has not been entirely clarified. Although M-CSF-stimulated p110c-deficient BMM show reduced migration rates in vitro [30], p110c deletion does not affect monocyte differentiation to macrophages [31]. p110c regulates cyclic adenosine monophosphate (cAMP) levels in cardiomyocytes through a kinase-independent pathway that involves formation of a complex that includes p110c, its p84/p87 PIKAP regulatory subunit, and the protein phosphodiesterase3B (PDE3B); this complex controls PDE3Bmediated cAMP hydrolysis [32], [33]. A similar p110c-dependent mechanism was recently shown to mediate microglial phagocytosis via lipid kinase-independent control of cAMP [34]. It is not known whether p110c regulates cAMP intracellular levels in macrophages. Macrophage proliferation is nonetheless affected by intracellular cAMP levels, as high levels are associated with cell cycle arrest [35]. In addition, cAMP response element binding protein (CREB) is linked to macrophage polarization to the M2 phenotype, thus connecting cAMP and M1/M2 macrophage polarization [36].
Here we examined the influence of p110c deletion on macrophage proliferation, apoptosis and polarization in atherosclerotic plaque, and tested whether p110c deletion contributes to lesion reduction in LDLR 2/2 mice. We identify a p110c function in macrophage proliferation within atherosclerotic lesions, a mechanism that contributes to atheroprotection in LDLR 2/2 mice lacking p110c.

Analysis of Macrophage and T cell Infiltration in Atherosclerotic Lesions
At t = 2 months of high-fat diet, mice were anesthetized (ketamine, 150 mg/kg; xylazine, 10 mg/kg; i.p.). Tail-and toepinch reflexes were tested to monitor adequacy of anesthesia and all efforts were made to minimize suffering. Whole blood was extracted by retro-orbital bleeding and hearts perfused with 4% paraformaldehyde. Hearts were extracted and paraffin-embedded. Some serial sections were stained by immunohistochemistry for T cells (CD3 + ), macrophages (Mac-3 + ) and regulatory T cells (Foxp3 + ) (see Supplement S1 for details).

In vivo Determination of Macrophage and T cell Proliferation
Macrophage and T cell proliferation was analyzed by immunofluorescence staining of the aortic valve region in paraffin-embedded sections from LDLR 2/2 p110c +/2 and LDLR 2/ 2 p110c 2/2 mice fed with a high-fat diet for two months. Markers were Mac-3 (macrophages), CD3 (T cells) and Ki67 (proliferation) (details in Supplement S1).

In vivo Determination of Lesion Apoptosis and of Vascular Smooth Muscle Cells
Lesion apoptosis was analyzed by TUNEL and cleaved caspase-3 immunofluorescence staining of the aortic valve region in paraffin-embedded sections from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice fed a high-fat diet for two months. Vascular smooth muscle cell (VSMC) staining was analyzed by anti-alpha smooth muscle actin (aSMA) immunofluorescence staining of similar sections (details in Supplement S1).

Macrophage Cell Cycle Analysis
BMM were synchronized in G0/G1 by M-CSF deprivation (36 h) and then stimulated for different times with M-CSF, collected and labeled with propidium iodide to analyze cell cycle by flow cytometry (see Supplement S1).
In a second approach, BMM from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice were differentiated in vitro and M-CSFstimulated at several times (0, 24, 48 h). Cells were washed, lysed and protein quantified. Western blot was developed to detect protein-bound cAMP, phospho-CREB (p-CREB) and total CREB with anti-cAMP antibody (clone SPM486; Abcam, Cambridge, UK; this antibody was generated using cAMP compounds as immunogen, and a chemically linked cAMP-carrier protein for antibody screening (see Supplement S1), as well as anti-pCREB (Ser133) and -CREB (both from Cell Signaling, Danvers, MA). bactin was used as loading control (clone AC-15, Sigma); band intensity was quantified using ImageJ software. As a positive control, BMM from p110c +/2 mice were differentiated in vitro and forskolin (FSK)-stimulated, and cAMP was detected in Western blot (see Supplement S1).

Statistical Analysis
Data are represented as mean 6 SD. Most statistical analyses were performed using Student's t-test to compare distinct parameters in two independent mouse groups (LDLR 2/ 2 p110c +/2 and LDLR 2/2 p110c 2/2 or p110c +/2 and p110c 2/ 2 ). Where indicated, data obtained by counting and small sample analysis were compared by the Poisson test. In all cases, differences were considered significant for p,0.05 (*p,0.05, **p,0.01).

p110c Deficiency Reduces Macrophage but not T cell Proliferation in Atherosclerotic Lesions
Macrophage proliferation in lesions enhances atherosclerosis progression to more advanced disease stages [15]. To determine whether the reduced atherosclerosis burden in LDLR 2/2 p110c 2/ 2 mice correlated with cell proliferation defects in lesions, we performed double immunofluorescence experiments in aortic cross-sections from high-fat diet-fed mice to test whether p110c deficiency affected macrophage and T cell in situ proliferation (as assessed by Ki67 expression). These studies showed a significant reduction in the number of proliferating neointimal macrophages in LDLR 2/2 p110c 2/2 compared to LDLR 2/2 p110c +/2 mice (Figure 2A, 2B). In contrast, p110c deletion did not affect T cell proliferation ( Figure 2C, 2D).
M-CSF is thought to play an important role in inducing macrophage proliferation in atherosclerotic lesions [14]. Cell cycle analysis of in vitro-differentiated BMM from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice allowed us to identify the proportion of cells in G0/G1, S and G2/M phases at various times post-stimulation with M-CSF. The proportion of S phase cells was reduced in LDLR 2/2 p110c 2/2 compared to LDLR 2/ 2 p110c +/2 macrophages at 26 h after M-CSF-stimulation ( Figure 2E), suggesting a role for p110c in macrophage cell cycle progression. In contrast, cell cycle assays to study in vitro BMM proliferation in response to GM-CSF showed no differences between p110c +/2 and p110c 2/2 BMM ( Figure S2).

Lesion Apoptosis is Unaffected by p110c Deletion
Macrophage apoptosis has been implicated in plaque progression [16]. We measured total apoptosis in lesions by TUNEL ( Figure 3A) and cleaved caspase-3 ( Figure 3B) immunofluorescent staining of aortic sinus sections from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice. Lesion area was delimited for TUNEL staining with the help of smooth muscle cells (SMC), which limit lesion area and are autofluorescent, and for cleaved caspase-3 staining by adding Mac-3 staining to the SMC guide; some lesion apoptotic cells are not Mac-3 + . We detected a tendency toward lower apoptotic rates in LDLR 2/2 p110c 2/2 compared to LDLR 2/2 p110c +/2 mice ( Figure 3A, 3B), although the differences were not significant.
In macrophages, signals that increase intracellular cAMP induce phosphorylation of cAMP response element-binding protein (CREB) [40]. As an alternative measurement of cAMP levels, we tested CREB phosphorylation status in LDLR 2/ 2 p110c +/2 and LDLR 2/2 p110c 2/2 mouse BMM. Coincident with the increased cAMP detected in the protein lysates, we found higher basal p-CREB levels in LDLR 2/2 p110c 2/2 mouse BMM ( Figure 4D). The data suggest that lack of p110c in macrophages promotes intracellular cAMP accumulation, which correlates with G0/G1 cell cycle arrest in LDLR 2/2 p110c 2/2 mouse BMM ( Figure 2E) since high cAMP levels are associated with cell cycle arrest [35].

Discussion
PI3K p110c is implicated in atherosclerosis, as its genetic deletion in ApoE 2/2 mice leads to reduced plaque size and impaired activation of the PI3K/Akt pathway in neointimal macrophages [21]. Pharmacological inhibition of p110c reduces atherosclerosis in ApoE 2/2 and LDLR 2/2 mice, and reconstitution of LDLR 2/2 mice with p110c 2/2 mouse bone marrow leads to decreased T cell and monocyte infiltration in atherosclerotic plaques [29]. Whether p110c deletion also contributes to local macrophage proliferation and apoptosis nonetheless remains unclear, as does the role of p110c in M1/M2 macrophage differentiation. In this study, we approached these questions by analyzing atherosclerosis development in LDLR 2/2 p110c 2/2 mice.
Macrophage proliferation in lesions promotes more rapid atherosclerosis progression [15]. Our studies of aortic sections showed a lower percentage of proliferating macrophages in LDLR 2/2 p110c 2/2 than in LDLR 2/2 p110c +/2 mice, although there were no differences in T cell proliferation between the two genotypes. We complemented in vivo analysis of macrophage proliferation with in vitro experiments using BMM. Whereas GM-CSF-induced proliferation was similar in BMM from p110c +/2 and p110c 2/2 mice ( Figure S2), proliferation was reduced and S phase entry delayed in M-CSF-stimulated LDLR 2/2 p110c -/2 compared with control LDLR 2/2 p110c +/2 BMM (Figure 2E), reflecting a specific p110c function in these processes after M-CSF signaling.
Macrophages undergo classical activation in response to LPS and IFNc, as part of the Th1 response (M1), or alternative activation in response to IL-4 as part of the Th2 response (M2) [50]. Advanced lesions in old ApoE-null mice show a prevalence of M1 over M2 macrophages, suggesting that the M2 phenotype is atheroprotective [13]. In activated primary macrophages, expression of M2-related genes (Arg-1, Il-10, Il13ra, Msr1) depends on CREB-induced expression of Cebpb (a gene that encodes a protein important for macrophage antibacterial activity) [36]. High cAMP levels could thus be linked to M2 macrophage polarization. Our data from LDLR 2/2 p110c +/2 and LDLR 2/2 p110c 2/2 mice showed no significant differences in the relative number of M1 and M2 macrophages in atherosclerotic lesions, although there was a tendency toward increased percentages of M2 macrophages in LDLR 2/2 p110c 2/2 compared to LDLR 2/2 p110c +/2 mice ( Figure 5B). Likewise, in vitro macrophage polarization was unaffected when we compared p110c +/2 and p110c 2/2 BMM, which showed similar M1 and M2 marker expression.
(DOC) Supplement S1 Supporting Materials and Methods, Results and References. (DOC)