|Home | About | Journals | Submit | Contact Us | Français|
Experimental models of atherosclerosis suggest that recruitment of monocytes into plaques drives the progression of this chronic inflammatory condition. Cholesterol-lowering therapy leads to plaque stabilization or regression in human atherosclerosis, characterized by reduced macrophage content, but the mechanisms that underlie this reduction are incompletely understood. Mice lacking the gene Apoe (Apoe–/– mice) have high levels of cholesterol and spontaneously develop atherosclerotic lesions. Here, we treated Apoe–/– mice with apoE-encoding adenoviral vectors that induce plaque regression, and investigated whether macrophage removal from plaques during this regression resulted from quantitative alterations in the ability of monocytes to either enter or exit plaques. Within 2 days after apoE complementation, plasma cholesterol was normalized to wild-type levels, and HDL levels were increased 4-fold. Oil red O staining and quantitative mass spectroscopy revealed that esterified cholesterol content was markedly reduced. Plaque macrophage content decreased gradually and was 72% lower than baseline 4 weeks after apoE complementation. Importantly, this reduction in macrophages did not involve migratory egress from plaques or CCR7, a mediator of leukocyte emigration. Instead, marked suppression of monocyte recruitment coupled with a stable rate of apoptosis accounted for loss of plaque macrophages. These data suggest that therapies to inhibit monocyte recruitment to plaques may constitute a more viable strategy to reduce plaque macrophage burden than attempts to promote migratory egress.
Atherosclerosis is a chronic inflammatory disease of the arterial wall that is currently the major cause of death in developed countries. The disease develops slowly and silently over decades, progressively evolving from fatty streaks characterized mainly by monocyte-derived cells filled with cholesterol esters to advanced plaques with more complex cellular composition, lipid pools filled with necrotic debris, and calcified material (1). Experiments in mouse models of atherosclerosis reveal that recruitment of monocytes into plaques drives disease progression (2). Overall, the inflammatory mechanisms of plaque formation have been extensively described, whereas mechanisms that reverse disease are less well understood. Vulnerable plaques are characterized by substantial necrotic core, reduced fibrous cap thickness, and fibrous cap inflammation wherein activated macrophages and dendritic cells along with associated T cells are invariably found to localize to sites of plaque rupture (3–7). Conversely, plaque stabilization is associated with a diminution in plaque macrophage content and a replacement of these cells with collagenous matrix. However, the cellular and molecular mechanisms underlying the process by which macrophages are removed from plaques during lipid-lowering therapy have not been determined. If these mechanisms were known, therapies to reinforce events that promote macrophage removal would likely improve clinical success in reducing further critical cardiovascular events.
Accumulation of monocyte-derived cells in atherosclerotic plaques is understood to be the result of several processes. Monocyte recruitment to plaques in mouse models of atherosclerosis occurs throughout different stages of plaque progression, and the rate of recruitment is increased when mice are maintained on a high-fat, cholesterol-enriched diet (8, 9). The potential for monocytes to be recruited into plaques from the blood is strongly affected by monocyte frequency in the circulation (10), implying that the availability of monocytes in the circulation may be rate limiting to atherosclerosis progression. Accordingly, monocytosis is an independent risk factor for atherosclerosis progression in humans (11–14).
Besides recruitment, survival and/or proliferation of monocyte-derived cells in plaques could affect overall macrophage accumulation. Recent studies indicate that intimal cell proliferation increases with hypercholesterolemia in early lesions (15) and correlates with plaque complexity in advanced lesions (16). Furthermore, the 9p21 genetic locus, which impacts proliferation of vascular cells, has been associated with susceptibility to coronary artery disease (17–19). With regard to survival, apoptosis of macrophages in the plaque is detected at all stages of the disease and is not proinflammatory in early, fatty streak lesions (20, 21). Mutations in Cx3cr1 and Csf1 accelerate loss of macrophages from plaques, at least in part through increased apoptosis (2, 22). In later stages of disease, clearance of apoptotic cells (also called efferocytosis) becomes compromised (21). As a consequence of decreased efferocytosis, apoptosis leads to secondary necrosis. In this setting, accumulation of dead cell debris amplifies proinflammatory signals and triggers further monocyte recruitment (20).
Finally, accumulation of monocyte-derived cells in the plaque may also result from their inability to leave plaque. Macrophages are normally removed from sites of resolving inflammation by migration through lymphatics to lymph nodes (23). Furthermore, in a surgical model of plaque regression in which plaque-bearing aortae are transplanted into wild-type mice with low levels of circulating cholesterol, rapid removal of plaque foam cells correlates with the onset of their emigration to lymph nodes (24). Further, this migration appears to depend upon the chemotactic ligands of the G protein–coupled receptor CCR7 (25). It has been proposed that unloading of cholesterol by plaque macrophages may restore emigration from plaques (26), consistent with the concept that cholesterol has a direct negative effect on the motility of macrophages (27).
In this study, we set out to investigate the cellular mechanism(s) that quantitatively account for macrophage removal from plaques in response to aggressive lipid lowering. Based on our earlier observations in a surgical model of plaque regression (24, 26), we predicted that emigration of monocyte-derived cells would be a major feature of macrophage removal from plaques. However, our earlier studies did not determine whether migratory egress was a minor or major aspect of macrophage removal from a quantitative perspective. Moreover, the requirement for re-anastomosis of lymphatics in the adventitia may render data from the surgical model difficult to translate to settings wherein surgical intervention and lymphatic remodeling is not a feature. Thus, the present study was designed to quantitatively determine the extent to which changes in the rate of monocyte recruitment, apoptosis, or migratory egress account for plaque macrophage removal from plaques. We focused on a model system in which surgery is not required and instead studied the mechanism of macrophage removal from plaques of Apoe–/– mice treated with a viral vector encoding apoE (28). This treatment lowers total cholesterol in plasma while raising HDL (28). In turn, plaques regress, and foam cell loss from plaques is substantial (28). Here, we show that in concert with a marked reduction in plaque cholesterol esters, plaque macrophage content in Apoe–/– mice declined by more than 70% in the course of the weeks following apoE complementation. The diminution in plaque macrophages did not result from migratory egress, as we had hypothesized, and did not require CCR7. Instead, loss of plaque macrophages resulted from greatly attenuated recruitment of new circulating monocytes. This decreased recruitment was associated with a reduced expression of inflammatory and adhesion molecules in the circulating monocytes and in the vascular wall. These findings bolster the rationale for developing therapies to target inhibition of monocyte recruitment in order to reduce macrophage burden in advanced atherosclerotic plaques.
Treatment of Apoe–/– mice with viral vectors encoding apoE (ad-hApoE3) causes lesion regression characterized by loss of oil red O (ORO) staining (to identify neutral lipids) and macrophages from plaques (28), but the kinetics by which changes occur in Apoe–/– plaques following apoE complementation have not been reported. Thus, we analyzed the course of changes observed from 2 days to 1, 2, 4, or 6 weeks after treatment of high-fat diet–fed (HFD-fed) Apoe–/– mice with ad-hApoE3, employing empty adenoviral vector (ad-Empty) as a control of plaque progression. Total cholesterol decreased from approximately 1,000 mg/dl to 100 mg/dl within 2 days following infection with ad-hApoE3 vector, and remained low for up to 6 weeks (Figure (Figure1,1, A and B, and data not shown). Within 2 days, HDL was elevated more than 4-fold and continued to rise, reaching a plateau by day 7 (Figure (Figure1B).1B). Plaque area in the aortic sinus (Figure (Figure1C)1C) and aortic arch (data not shown) increased by 4 weeks in control ad-Empty vector–treated mice, whereas plaque area was stabilized in ad-hApoE3–treated animals, showing no changes in relation to a baseline cohort euthanized at the time when vector treatment began (Figure (Figure1C1C and data not shown). In a separate experiment wherein Apoe–/– mice were euthanized 6 weeks after vector treatment, ad-hApoE3 vector reduced plaque area and stenosis relative to baseline (Supplemental Figure 1; supplemental material available online with this article; doi: 10.1172/JCI43802DS1), consistent with past evidence indicating that over many weeks, regression of plaque area occurs (28).
We next quantified macrophage area in plaques as a function of treatment with ad-hApoE3 or ad-Empty vector. CD68+ macrophages disappeared in a linear fashion from plaques over the weeks following treatment with ad-hApoE3 vector (Figure (Figure1D)1D) in contrast to the ad-Empty group, where they increased substantially over time (Figure (Figure1D).1D). However, the reduction in macrophage content only became statistically significant 4 weeks after ad-hApoE3 treatment (Figure (Figure1,1, D and E), when it was reduced by an average of 72%. The marked loss of macrophages in the ad-hApoE3–treated group was further confirmed by staining for other macrophage markers, F4/80 and MOMA-2 (Supplemental Figure 2, A and B). Accompanying the decrease in macrophage content in plaques of ad-hApoE3–infected mice, we observed increased collagen accumulation, as analyzed with Sirius red staining (Supplemental Figure 2C), and conserved smooth muscle cell content, as revealed by α-actin staining (Supplemental Figure 2D).
To document how changes in plaque cholesterol distribution and mass correlated with the decrease in circulating cholesterol or plaque macrophages observed in ad-hApoE3–treated mice, we first stained plaque sections with ORO and then quantified total, free, and esterified cholesterol using mass spectroscopy (29). ORO+ area did not change in ad-Empty vector–treated mice over time. By contrast, in ad-hApoE3 vector–treated mice, ORO+ area in the plaque was decreased by more than one-half (Figure (Figure1F1F and Supplemental Figure 3A), and also showed substantial reduction in other tissues, such as the liver, 4 weeks after infection (Supplemental Figure 3B). We verified that the adenovirus infection had no detrimental effect on liver function by assessing alanine transaminase (ALT) activity in plasma at different time points (Supplemental Figure 3C). By 4 weeks after infection, ALT levels approached baseline in all groups (Supplemental Figure 3C). Nevertheless, we noted a transient effect of ad-hApoE3 on day 7 (Supplemental Figure 3C). The specific and prominent increase in ALT on day 7 after ad-hApoE3 infection was paralleled with a peak of large cholesterol-rich HDL particles (big HDL) accumulation in the plasma (Figure (Figure1B),1B), consistent with the possibility that mobilization of cholesterol to the liver peaks around day 7 and transiently affects liver metabolism. This finding is reminiscent of a previous study on virally induced DGAT1 overexpression, where a similar outcome was observed (30).
Given the impressive reduction in ORO+ area within plaques, we set out to quantify the changes in aortic cholesterol mass. Cholesterol ester mass dropped substantially, by one-third (Figure (Figure1G),1G), whereas the clear trend downward in total cholesterol mass did not reach statistical significance (Figure (Figure1G).1G). These data corresponded to an increased ratio of free cholesterol to total cholesterol within the aorta in response to apoE complementation (49% ± 3% to 61% ± 4%, before versus after apoE complementation), indicating that the proportion of cholesterol present in esterified form was decreased (51% ± 3% to 39% ± 4%, before versus after apoE complementation). These results are consistent with our ORO staining analyses (Figure (Figure1F),1F), as ORO staining detects only esterified cholesterol (31). Putting these observations together, we observe that an initial reduction in plasma cholesterol and increase in circulating HDL were rather rapidly followed by reduced cholesterol ester–laden macrophages in plaque. The reduction in cholesterol ester–laden foam cell content occurred prior to, rather than concomitantly with, the loss of plaque macrophages.
We next set out to determine how the decrease in macrophage content within plaques was achieved. Given our past findings in the surgical model of regression, we hypothesized that monocyte-derived cell disappearance would be a consequence of their emigration out of plaque, through mobilization into adventitial lymphatics (24). Accordingly, we anticipated that deficiency in CCR7, a key mediator in leukocyte emigration through lymphatics (32–34) and a proposed modulator of regression in the surgical model (25), would prevent loss of plaque macrophages in response to ad-hApoE3 treatment. We thus crossed Apoe–/– mice with Ccr7–/– mice to generate Apoe–/–Ccr7–/– double-knockout mice and analyzed the evolution of macrophage area within plaques after ad-hApoE3 infection. The absence of CCR7 and ApoE led to increased levels of circulating leukocytes and decreased lymphocyte and dendritic cell accumulation in the lymph nodes similar to the CCR7 single knockout (32, 33) (data not shown). However, after 9 weeks of HFD feeding (14–15 weeks of age), atherosclerotic plaques of Apoe–/–Ccr7–/– mice showed no significant difference in size or in macrophage content compared with Apoe–/–Ccr7+/+ control littermates, although trends toward somewhat smaller lesions in the absence of CCR7 were observed (Figure (Figure2,2, A and B). In contrast to expectations generated from observations in the surgical model of regression (25), lipid and macrophage loss from plaques of Apoe–/–Ccr7–/– mice was massive after treatment with the ad-hApoE3 vector (Figure (Figure2,2, C and D), thus illustrating that CCR7 is not necessary for profound loss of lipid or macrophages from plaques.
To more generally address the importance of migratory egress from plaques as a means of removing macrophages during plaque regression, which may occur in a CCR7-independent manner, we employed a modified version of the phagocyte-tracking approach previously developed in our laboratory (9, 35). This technique independently labels about 10%–20% of the two endogenous subsets of monocytes in the circulation and is useful in monitoring and quantifying monocyte subset entry into (9) and egress from plaques (36). Over time, if latex bead+ monocyte-derived cells die or exocytose beads, the beads will not disappear, as they are non-degradable and are too large to cross the endothelial barrier passively, but may instead label other phagocytes within the plaque. In past studies, we have documented that latex bead transfer from phagocyte to phagocyte can occur (37) and that emigration of bead-labeled monocyte-derived cells is not impeded (9, 38, 39). Once peak accumulation of latex bead+ monocytes occurs in plaques, if thereafter some monocyte-derived cells migrate out of lesions, then the total number of bead+ cells in the lesions will decrease. On the other hand, if there is no emigration of monocyte-derived cells out of lesions, the bead frequency within plaques will remain constant.
To verify that this method can track emigration from plaques, we first tested it using the surgical model of regression, wherein migratory egress from mouse plaques was initially described (24). We labeled blood monocytes with beads and then harvested aortic arches from these animals after 3 days. A subset of aortic arches was used to quantify baseline recruitment, while the others were transplanted into wild-type C57BL/6 mice (Apoe+/+) to allow plaque regression. Mice receiving these aortic transplants were sacrificed 5 days later (Figure (Figure3A).3A). We then quantified the number of beads per 6-μm section of the lesser curvature of the aortic arch, comparing the number of beads per section at baseline prior to surgery to the number of beads per section 5 days after surgery. A 63% decline in bead content was observed at the post-transplantation end point (Figure (Figure3B).3B). Thus, in the surgical model of regression, monocyte-derived cell emigration from the intima indeed occurred, and the process appeared quantitatively significant enough to mediate loss of macrophages that accompanies regression. Furthermore, these data serve as a proof of concept that beads within plaques can be removed under conditions of phagocyte mobilization.
To monitor migratory egress from Apoe–/– plaques after apoE complementation, we labeled circulating monocytes in Apoe–/– mice with latex beads and tracked the persistence of this label in plaques as a function of ad-hApoE3 or ad-Empty vector treatment (Figure (Figure3C).3C). We set up two parallel studies wherein we independently labeled the classical (Ly-6Chi) and nonclassical (Ly-6Clo) subsets of monocytes as the populations to carry latex beads into plaques (2, 9). Mice were housed without further manipulation for the next 8 days to permit accumulation of bead+ monocytes in the plaques. Within this period, 73% of bead+ monocytes were cleared from the blood due to recruitment into tissues and normal turnover (Supplemental Figure 4). At day 8, we sacrificed one-third of the mice for baseline analysis of plaque area, lipid and macrophage content, and number of beads per plaque section. Other similarly bead-labeled mice were infected with ad-hApoE3 or ad-Empty vector 8 days after monocyte labeling and sacrificed at 4 weeks after adenoviral infection. Bead number was quantified in the aortic sinus and the lesser curvature for comparison to the number of bead+ cells in analogous sections obtained at baseline.
From the baseline data obtained in aortic sinus, we observed that, in contrast to previous studies in the aortic arch (9), labeled Ly-6Chi monocytes did not outnumber labeled Ly-6Clo monocytes in entering aortic sinus plaque (Figure (Figure3D),3D), where robust recruitment of both monocytes was apparent. To analyze migratory egress from plaques, accumulation of bead+ monocytes at the experimental end point was compared with that in the baseline group. The number of beads per plaque section in Apoe–/– mice treated with ad-Empty for 4 weeks after baseline monocyte entry was the same as observed at baseline. This result was true whether circulating classical Ly-6Chi or nonclassical Ly-6Clo monocytes were initially labeled or whether the aortic sinus or lesser curvature of the aorta was examined (Figure (Figure3D).3D). In agreement with previous findings (24), this result suggests that migratory egress of plaque phagocytes is rare or does not occur in conditions of plaque progression. Unexpectedly, we did not observe any diminution in bead labeling in Apoe–/– mice treated with ad-hApoE3 (Figure (Figure3,3, D and E), even though (as shown in Figure Figure1)1) macrophage loss from plaque was significant under these conditions. This lack of bead removal from plaques was not because the label was sequestered outside of macrophages, as even at late time points, very few beads were present in necrotic core or outside of macrophages (Figure (Figure3E3E and data not shown). Overall, our analyses failed to support the hypothesis that migratory egress is responsible for the marked removal of macrophages that characterizes ad-hApoE3–treated Apoe–/– mice.
Without experimental support for a role of migratory egress in mediating removal of macrophages from Apoe–/– plaques after apoE complementation, we initiated studies to determine whether monocyte recruitment was altered. Since hypercholesterolemia increases monocytes in the circulation, making more monocytes available for recruitment into plaques (9, 40), we first examined whether ad-hApoE3 or ad-Empty vector treatment influenced monocyte counts or subset frequency in the blood. Blood monocyte subsets were discriminated by flow cytometry, based on their expression of CD115 and Ly-6C (9). Neither ad-hApoE3 nor ad-Empty virus significantly influenced monocyte numbers of either circulating subset (Figure (Figure4A),4A), indicating that the loss of macrophages observed in plaques of ad-hApoE3–treated mice was not downstream of a notable decrease in blood monocyte counts. However, monocytes in the circulation altered their phenotype. Monocytes from ad-hApoE3 vector–treated mice had reduced surface CD11b, known to be elevated upon monocyte activation in general and following postprandial hyperlipidemia (ref. 41 and Figure Figure4B).4B). Extending evidence that monocyte activation was reduced, surface CD62L, which is expressed only on the Ly-6Chi monocyte subset (42) and which is shed upon activation, was more abundant on the plasma membrane of Ly-6Chi monocytes after apoE complementation (Figure (Figure4C).4C). Lipid loading of monocytes in the plasma may also have been modulated, because the increased side scatter associated with lipid uptake into circulating Ly-6Clo CD11c+ monocytes (43), which preferentially express scavenger receptors and lipid response pathways (44), was reversed by apoE complementation (Figure (Figure4,4, D and E). Ly-6Clo monocytes also expressed less CD115 (CSF-1 receptor), a growth factor receptor for CSF-1 known to promote atherosclerosis (45), after apoE complementation (Figure (Figure4F). 4F).
For quantification of monocyte recruitment into plaques by labeling monocyte subsets in vivo using the bead-labeling approach, monocyte subsets were labeled 1, 2, or 4 weeks after adenoviral treatment, and plaques were harvested for quantification from 1 to 5 days later, during the major period of monocyte entry (9). This experimental design was such that the bead-labeling assay could be used to monitor monocyte entry into plaques rather than their persistence, as in the experiments presented in Figure Figure3.3. Importantly, we verified that adenoviral infection did not affect the bead-labeling efficiency of blood monocytes and that the same number of monocytes was labeled in each group (data not shown). Quantification of monocyte recruitment into plaques after infection with the control adenoviral ad-Empty vector revealed that it slightly but non-significantly increased monocyte recruitment into plaques 1 week after infection, and this effect did not persist up to 2 weeks following infection (Figure (Figure5A).5A). To compile results from independent experiments, we compared the relative magnitude of monocyte subset entry into plaques between ad-hApoE3–treated mice and ad-Empty vector–treated mice. By 2 weeks after ad-hApoE3 vector treatment and thereafter, we observed a dramatic inhibition of both nonclassical (Figure (Figure5B)5B) and classical monocyte subset entry (Figure (Figure5C)5C) into the plaque. Inhibition of monocyte recruitment into plaques appeared somewhat delayed relative to the kinetics of cholesterol lowering in plasma, since apoE complementation non-significantly reduced monocyte recruitment to plaques at 1 week after infection with ad-hApoE3 vector (Figure (Figure5B). 5B).
In order for impaired monocyte recruitment to result in macrophage loss from plaques, there must also be a loss of macrophages due to turnover. We thus assessed whether apoE complementation altered apoptosis of plaque macrophages by quantifying the number of plaque cells that were CD68+, DAPI+, and TUNEL+ per section, 2 weeks after ad-Empty or ad-hApoE3 vector treatment (Figure (Figure5D).5D). The vast majority of the DAPI+ TUNEL+ cells costained for CD68+ (data not shown). Our analysis found a similar rate of macrophage apoptosis in plaques treated with ad-hApoE3 or ad-Empty (Figure (Figure5D),5D), indicating that an enhanced burst of apoptosis did not contribute to macrophage loss within Apoe–/– plaques after treatment of these mice with apoE-complementing virus. Instead, a stable rate of apoptosis coupled with a markedly reduced new supply of incoming monocytes served to reduce macrophage content in plaques over time. It is possible that macrophage content may also be influenced by changes in proliferation, but an analysis of Ki67 staining to assess this question revealed that the vast majority of proliferation in the plaques we examined was observed within α-actin+ smooth muscle cells, whose rate of proliferation nearly doubled in response to apoE complementation (data not shown).
To lend support for the importance of monocyte recruitment in maintaining macrophage content of established plaques, we compared macrophage area in Apoe–/– plaques treated for 3 weeks with biologically active pertussis toxin (PTX), an inhibitor of αi-type G protein–coupled receptors that regulate the responsiveness of most chemoattractant receptors, versus those treated with PTX denatured by boiling. While systemic treatment with PTX no doubt has multiple effects, including induction of leukocytosis, we reasoned that its use would allow us to confirm that blocking monocyte entry into plaques would be associated with a reduction in plaque macrophage accumulation, independently of aggressively lowering cholesterol. Using a model of skin contact sensitization (46), we first verified that a single i.p. injection of PTX (2.5 μg/mouse), but not boiled PTX or PBS, efficiently impaired skin dendritic cell migration to lymph nodes (data not shown). Just prior to the end of a 3-week period of PTX treatment in HFD-fed Apoe–/– mice, circulating monocytes were bead labeled. Treatment with PTX partially inhibited monocyte recruitment into plaque (Supplemental Figure 5A) and decreased CD68+ macrophage content by approximately one-half (Supplemental Figure 5B), with smaller changes in lesion area and ORO content (Supplemental Figure 5, C and D). PTX-treated mice showed a similar percentage of TUNEL+ cells per plaque section (data not shown). These data further support the concept that targeting monocyte recruitment in established advanced plaques is a viable strategy for lowering plaque macrophage burden.
Next, we carried out a series of experiments in Apoe–/– mice fed a standard chow diet to discern whether apoE complementation under conditions of less extreme hypercholesterolemia led to a similar reduction in monocyte recruitment. We studied 31-week-old Apoe–/– mice fed a standard chow diet, as these older-age chow-fed Apoe–/– mice had lesion size similar to that in the HFD-fed younger Apoe–/– mice we studied (Figure (Figure5E),5E), while baseline cholesterol levels were approximately one-half those of the HFD-fed cohort (Figure (Figure5F).5F). The rise in cholesterol following its nadir in HFD-fed Apoe–/– mice (Figure (Figure1A)1A) was less apparent in the chow-fed animals, wherein cholesterol levels still hovered near Apoe+/+ levels 2 weeks after apoE complementation (Figure (Figure5F).5F). Within 4 weeks following apoE complementation in the chow-fed Apoe–/– mice, lesion area was decreased (Figure (Figure5E),5E), in contrast to HFD-fed mice (Figure (Figure1,1, B and C), consistent with earlier work (28). Macrophage area reductions in chow-fed Apoe–/– mice were significant 2 weeks after apoE complementation (Figure (Figure5G).5G). None of these differences were sufficient to lead to migratory egress from plaques, as assessed using the bead-labeling method (data not shown), but monocyte recruitment to plaques was dramatically reduced (Figure (Figure5H)5H) to an extent similar to that observed in the HFD-fed mice (Figure (Figure5B).5B). Indeed, these studies in chow-fed animals with or without apoE complementation extended the range of plasma cholesterol variations under which we quantified monocyte recruitment. Total plasma cholesterol and the magnitude of monocyte recruitment to plaque were positively correlated (Figure (Figure5I),5I), consistent with the concept that disease reversal characterized by aggressive cholesterol lowering is pivotally regulated by modulation of the recruitment of monocytes to plaque.
Understanding the complete mechanism behind suppression of monocyte recruitment in the context of apoE complementation is beyond the scope of this study. However, we did observe reduced expression of ICAM-1, VCAM-1, and osteopontin, molecules involved in the recruitment of monocytes across the arterial endothelium, in plaque sections 2 weeks after ad-hApoE3 vector treatment (Figure (Figure6).6). Thus, apoE complementation leads to changes both within plaque (Figure (Figure6)6) and within circulating monocytes (from Figure Figure5)5) that could, especially collectively, strongly suppress monocyte recruitment.
To design therapies to reduce macrophage burden in atherosclerotic plaques, understanding of the cellular processes that regulate macrophage accumulation in atherosclerotic lesions is required, but the fundamental processes underlying removal of macrophages have not been determined. In the past years, the concept that migratory egress from plaques is a key feature of plaque regression has gained momentum (24–26, 36). However, due to lack of availability of quantitative approaches, past studies on egress have generally not been quantitative and have examined migratory egress from plaques with little comparison to other processes that might reduce macrophage burden in plaques. Methods to quantify the migration of monocytes into atherosclerotic plaques have only recently been developed (8, 9). Here we show that one of these methods, the bead-tracking method we developed previously (9), permits quantitative assessment of not only monocyte entry into, but also their exit from, plaques (24–26, 36). By using this method, we investigated the relative importance of changes in the rate of monocyte entry versus egress from plaques in reducing macrophage burden after lowering of plasma total cholesterol, increasing HDL, and restoring apoE expression in Apoe–/– mice (9, 26, 28). Following adenoviral vector–mediated apoE complementation in Apoe–/– mice, we observed that removal of macrophages from plaques was strongly correlated with robustly reduced monocyte entry into plaques, coupled with a steady rate of apoptotic turnover, whereas migratory egress from plaques was minimal.
The concept of blocking monocyte entry into plaques as a therapy is not new (20). Indeed, it has been obvious that blocking influx of monocytes would prevent the expansion of macrophage area during plaque growth. However, our findings indicate that inhibiting monocyte influx does not simply stabilize, but instead actually reduces, macrophage content in plaques. Contrasting with our conclusion, for example, is a study by Guo et al. wherein bone marrow deficient in CCR2 was transplanted into irradiated Apoe–/– mice with developed plaques (47). Loss of macrophage content was not evident in these plaques (47), even though circulating monocytes would lack CCR2, leading the authors to conclude that suppressing monocyte entry into developed plaques was inconsequential for macrophage removal. Given our findings, the work of Guo et al. requires reinterpretation. For instance, it may be necessary to inhibit receptors other than, or in addition to, CCR2 to achieve macrophage loss from plaques, consistent with findings that several chemokine receptors are involved in monocyte accumulation within atherosclerotic plaques (9, 10, 48). Mice lacking CCR2, however, also have a marked deficiency in classical Ly-6Chi monocytes in the blood (39, 49) due to a role for CCR2 in egress of monocytes from the bone marrow (49, 50), and it is this subset of monocytes that enters plaques of the aorta in a quantitatively dominant manner (9, 40). Thus, the mice studied by Guo et al. would also have had greatly reduced circulating Ly-6Chi monocytes. It is interesting to note that the analysis of lesions in that study focused on the aortic root. In the present study, we were surprised to find that entry of labeled monocytes into aortic root plaques is far more robust than in the lesser curvature of the aortic arch, and that CCR2-negative Ly-6Clo monocytes are as efficient at entering aortic root plaque as the Ly-6Chi monocytes, quite distinct from the dominance of Ly-6Chi monocytes in the lesser curvature (9, 40). That the two major subsets of monocytes appear to have different abilities to enter plaques in different regions of the arterial tree may explain why different plaques have distinct patterns of dependency on different homing molecules (51), possibly including CCR2 (47), which is not expressed on Ly-6Clo monocytes. In any case, our findings underscore the concept that blocking monocyte entry into developed plaques has the potential to promote favorable changes in plaques, including reduction in macrophage burden. Blocking more than one receptor may be necessary, and more than one subset of monocytes may be relevant, especially in certain lesion sites.
Indeed, targeting inhibition of monocyte entry into plaques may be more likely to achieve success in removing macrophages than attempting to coax macrophages out of plaques by promoting their migratory egress. As our earlier studies using the surgical model of regression were not quantitative and included analysis of the whole aorta and not just the intima (24, 25), we combined the quantitative bead tracking approach with the surgical model and revealed that migratory egress from the intima is quantitatively significant. In contrast to the surgical model, two lines of evidence failed to support involvement of cell egress in the model of regression achieved in Apoe–/– mice by apoE complementation: (a) lack of a role for CCR7, proposed to mediate migratory egress through lymphatics (25), and (b) failure to observe quantitatively measurable egress using the bead-tracking model. A major difference between the surgical and apoE complementation models is that regression is much faster in the surgical model (24) than in the apoE complementation model (28). The slower process of reducing plaque macrophage content using apoE complementation in Apoe–/– mice likely more closely resembles therapeutic scenarios in humans (52).
The explanation for the mechanistic differences between the two regression models is not completely clear. Perhaps surgical intervention itself affects cellular behavior in plaques, as we have noticed more infiltration of the media by leukocytes in the surgical model, suggesting unusual breach of medial immunoprivilege (53, 54) accompanied by overwhelming adventitial inflammation (G.J. Randolph, unpublished observations). Based on the data in the present study, it seems reasonable to suppose that the pathway involving migratory egress from plaques is operative only in the context of certain stimuli not present in the apoE complementation model (e.g., breach of immunoprivilege). By contrast, the slower pathway of macrophage removal that we describe here — involving no or minimal induction of egress but stemming from a combination of reduction in signals to recruit monocytes and ongoing apoptosis of monocyte-derived cells within plaque — acts as a default means for regulating plaque macrophage content. Even though migratory egress in the surgical model of regression appears to rely on CCR7 and its ligands (25, 55), it would be unlikely that failure to express CCR7 would govern regression even in the surgical model, because the slower process of reducing macrophage burden that we describe here would be expected to take over, rendering the CCR7 pathway ultimately, if not immediately, dispensable. Indeed, we found that CCR7 was not required to lower macrophage accumulation in plaques of Apoe–/– mice treated with viral vectors encoding apoE, and its previously reported role in other models has not been explored beyond the first several days after induction of regression (55). Even if one would raise the unlikely possibility that adenoviral vectors encoding apoE may kick off a process that inhibits migratory egress, our data clearly reveal that macrophage burden in plaques can still be robustly decreased through suppression of monocyte entry, thus revealing that migratory egress is not essential during plaque regression to relieve macrophage burden.
With this study bringing focus on monocyte recruitment rather than egress as a means to reduce macrophage burden in plaques, the important next step will be to identify the molecular mechanisms that trigger and ultimately account for reduced monocyte recruitment. It seems logical that the attenuation of monocyte recruitment would be downstream of a critical shift in lipid composition in the plaque, and indeed we observed a marked decrease in cholesterol ester content within plaques. Plasma HDL levels increased robustly after apoE complementation, and the time course suggested that plasma HDL particles were especially heavily laden with cholesterol on day 7. However, considering the mass spectroscopy data from aortae, it is unclear how much of this cholesterol originates from plaques within the first 2 weeks after apoE complementation. Nonetheless, the changes that occurred within plaques within this time frame led to a striking diminution in the inflammatory status of lesions, particularly evident in the strong blockade in recruitment of new monocytes. With regard to cholesterol mass measurements, our findings are reminiscent of previously published studies on monkeys, wherein the first period of regression, a period that included massive reduction in plaque macrophage content, was characterized by retention of cholesterol mass in plaques, with robust de-esterification of cholesterol leading to a relative increase in free cholesterol within lesions (56). As more time passed, cholesterol itself was cleared from plaques (56). In our study, we noted that the state of monocyte activation in the circulation was reduced and the expression of various adhesion receptors (ICAM-1, VCAM-1, etc.) was also decreased in lesions. These changes may account for the suppression of monocyte entry into plaques. It will be interesting in the future to determine how these changes come about in plaques and in particular whether and how they are linked to the mobilization of cholesterol from an esterified pool.
In humans, it has been shown that for every 1% reduction in LDL-C levels, relative risk for major coronary events is reduced by approximately 1% (52). The positive correlation we observed between the decrease in circulating total cholesterol, the decrease in cholesterol ester within plaque, and the reduction in monocyte recruitment to plaque is consistent with this observation in humans and the concept that foam cell content in plaque is a major independent risk factor for cardiovascular events. The data generated in this study support the idea that interfering with monocyte recruitment into plaques may be therapeutically beneficial, though it is also clear that such an approach should be carried out as ancillary therapy with aggressive lipid lowering in order to maintain and reinforce reductions in monocyte recruitment. Indeed, blocking monocyte entry into plaques as an add-on therapy to accompany aggressive lipid lowering may be a particular efficacious strategy to achieve further clinical advances in reducing rupture of vulnerable plaque that stems from activation of plaque macrophages.
Male and female C57BL/6J Apoe–/– mice expressing Ly5.1 (CD45.2) were purchased from the Jackson Laboratory, and Apoe–/– mice expressing Ly5.2 (CD45.1) were generated (24) and maintained in our colony. For each experiment, mice in all groups were matched according to their age, sex, and CD45 congenic marker (CD45.2 or CD45.1) status. Ccr7–/– (CD45.2+) mice were purchased from the Jackson Laboratory (strain B6.129P2-Ccr7tm1Dgen/J; stock 005794) and crossed to Apoe–/– (CD45.1+) mice to generate Apoe–/–Ccr7–/– mice. Littermate controls (Apoe–/–Ccr7+/+) were generated at the same time. Mice were housed in a specific pathogen–free environment and used in accordance with protocols approved by the Institutional Animal Care and Utilization Committee at Mount Sinai School of Medicine.
At the age of 5–6 weeks, mice were transitioned to a HFD (21% milk fat, 0.15% cholesterol; Harlan Teklad). Matched mice were divided into 3 groups. One group was sacrificed to establish baseline parameters in the plaque. The two other groups remained on a HFD for the duration of treatment and were infected either with apoE-encoding adenoviral vector (ad-hApoE3) or with its control empty adenoviral vector (ad-Empty) i.v. Infection was routinely verified in each animal by measuring total cholesterol levels after submandibular blood puncture. Total cholesterol levels were determined using a Wako kit (catalog 439-17501) according to the manufacturer’s protocol. Lipoprotein cholesterol distributions were determined using thawed plasma samples diluted 1:1 in cold PBS centrifuged at 10,000 g for 2 minutes at 4°C. Approximately 15 mg of cholesterol was injected onto an HPLC system with online mixing of enzymatic reagent (Cholesterol Liquid Reagent Set, Pointe Scientific Inc.) with effluent from the column. The HPLC system consisted of a Superose 6 10/300 GL column (GE Healthcare); a LaChrom Elite HPLC system (Hitachi High Technologies) consisting of an L-2200 Autosampler with Peltier cooling, an L-2420 UV-Vis Detector, and two L-2100 SMASH pumps — one for delivering 0.9% saline with 0.01% EDTA and 0.01% sodium azide at 0.4 ml/min to the column and the other for delivering cholesterol reagent (or water) at 0.125 ml/min; a 5 m (length) × 0.5 mm (internal diameter) knitted reaction coil (KRC 5-50, Aura Industries Inc.) in a water jacket (CJB-10, Aura Industries Inc.) at 37°C; an Isotemp circulating water bath (Fisher Scientific); a PEEK Tee (Upchurch part no. P-712) at the junction of column effluent and total cholesterol reagent; and a personal computer running ChromPerfect Spirit Chromatography Data System, version 5.5 (Justice Laboratory Software) for data acquisition and analysis. Infected mice were sacrificed 1–6 weeks after infection.
Nonclassical Ly-6Clo monocytes were labeled in vivo by retro-orbital i.v. injection of 1 μm Fluoresbrite green fluorescent (YG) plain microspheres (Polysciences Inc.) diluted 1:4 in sterile PBS. Classical Ly-6Chi monocytes were labeled with beads using the same protocol, but with beads administered 3 days after i.v. injection of 250 μl clodronate-loaded liposomes. Clodronate was a gift from Roche and was incorporated into liposomes as described previously (57). Labeling efficiency was verified by flow cytometry 1 day after labeling.
Surgical transfer was performed as previously described (58). Briefly, 12 female Apoe–/– (CD45.1+) mice were used at the age of 9 months. Three days before surgery, Ly-6Clo monocytes were labeled with beads, as described above. Aortic arches were transplanted into female WT retired breeders (Apoe+/+, CD45.2+) that were matched in weight.
Mice were sacrificed and perfused with 10 ml PBS containing 2% EDTA. The heart and aorta were removed and fixed in 4% paraformaldehyde for 2 hours. The heart was transferred into PBS containing 30% sucrose (wt/vol) overnight at 4°C before being embedded in OCT compound and stored at –80°C. Cryosections (8 μm thickness) of the aortic sinus were prepared. Aortic arches were either likewise frozen in OCT compound and cross-sectioned (6 μm) or were opened en face for analysis. ORO (Sigma-Aldrich O-0625) staining was used to detect neutral lipids in the plaque. Sirius red staining was used to analyze collagen content. Plaque macrophages were analyzed by staining with anti-CD68 mAb (AbD Serotec MCA1957), anti–MOMA-2 mAb (AbD Serotec MCA519G), anti-F4/80 mAb (AbD Serotec MCA497BB), anti-Ki67 mAb, and/or anti-CD11c mAb (BD Biosciences — Pharmingen 553800). Vascular wall adhesion molecules were examined by staining with anti–ICAM-1 (BD Biosciences — Pharmingen 550287), anti–VCAM-1 (BD Biosciences — Pharmingen 550547), and anti-osteopontin (R&D Systems AF808) antibodies. Anti-rat or anti-hamster Cy3- or Cy5-conjugated antibodies were used for detection (Jackson ImmunoResearch Laboratories Inc.). Smooth muscle cells were identified by staining with anti–α-actin mAb (Sigma-Aldrich C6198). Nuclei were revealed with DAPI (5 μg/ml in mounting medium). TUNEL staining was carried out using the TMR red kit (Roche 12156792910). Quantification of plaque area and area occupied by a particular stain was calculated with ImageJ software. Beads in plaques were counted manually by fluorescence microscopy (9).
Lipids were extracted from aortic tissue according to Bligh and Dyer (59), with addition of 4 μg of [3,4-14C2]cholesterol as the internal standard for cholesterol analyses. After evaporation under a stream of argon, the lipid extracts were dissolved in 1 ml chloroform/methanol (at 1:1) and stored at –80°C until analysis. An aliquot from each extract was evaporated under argon, dissolved in toluene, and then analyzed on a Thermo Fisher Trace 2000 mass spectrometer. One microliter of the reconstituted sample was injected in the splitless mode onto a 30 m × 250 μm DB-1 column (J&W Scientific) having a 0.25-μm film thickness. Helium carrier gas was set at 2 ml/min. The temperature program was as follows: 150°C for 1 minute, 150°C to 280°C at 25°C/min, ending at 280°C for 23 minutes. The samples were analyzed using positive ion–electron impact with selected ion monitoring (SIM) of m/z 386.4 for cholesterol and m/z 388.4 for [3,4-13C2]cholesterol. A second 50-μl aliquot was saponified at room temperature for 2 hours with methanolic-KOH, similar to a procedure using ethanol (60), cooled, and then extracted with hexane. After evaporation under argon, the residue was dissolved in 20 μl of toluene and analyzed for cholesterol content. Cholesteryl ester was calculated as the difference between the total cholesterol and free cholesterol values. All analyses were performed in glass vessels. A standard curve was prepared for quantitation. Aortic protein was determined by the Lowry method (61) using bovine serum albumin as the standard.
Mouse blood was collected by non-terminal submandibular or terminal cardiac puncture. Total leukocytes were quantitated by fresh blood dilution in Turk’s solution (RICCA Chemical Company 8850-16). After red blood cell lysis (BD PharmLyse), monocyte subsets were identified by staining with anti-CD115 mAb (eBiosciences clone AFS98), CD11b (eBiosciences clone M1/70), CD11c (eBiosciences clone N418), and anti-Gr1/Ly-6C antibody (BioLegend, clone RB6-8C5). Data were acquired on a BD FACSCanto Flow Cytometer or a BD LSRII Flow Cytometer and analyzed with FlowJo software (Tree Star).
C57BL/6 male mice were treated by i.p. injection of 2.5 μg PTX or 2.5 μg boiled PTX to inactivate the toxin but retain the ability to assess effects of heat-resistant endotoxin contamination. Contact hypersensitization assays (FITC painting) confirmed that this dose of PTX was sufficient to reduce dendritic cell migration to lymph nodes (46) for at least 5 days. In separate studies, 18-week HFD–fed female Apoe–/– mice were treated once weekly for 3 weeks with PTX (2.5 μg/mouse/wk) or control boiled PTX. Animals were subsequently sacrificed and arches and sinus analyzed as above.
Data are expressed as mean ± SEM, except where otherwise indicated. Statistical differences were assessed using a 2-tailed t test or ANOVA (with Tukey’s post-test analysis) with Prism software. A P value of less than 0.05 was considered statistically significant.
We thank Jianhua Liu (Mount Sinai School of Medicine), Manal Zabalawi, and Brian Fulp (Wake Forest University) for expert assistance. We thank Alan Tall, Molly Ingersoll, Andrew Platt, and Catherine Martel for critical reading of the manuscript. We thank the Department of Genetics and Genomic Science for assessment of the ALT measurement. Major support for this work includes NIH grants AI061741, AI049653, and HL096539 (subcontract to M.G. Sorci-Thomas), American Heart Association (AHA) Established Investigator Award 0740052, and a grant from the Experimental Therapeutics Institute, Mount Sinai Hospital to G.J. Randolph; NIH grants HL-49373 and HL-64163 to M.G Sorci-Thomas; and AHA grant 09GRNT2280053 and North Carolina Biotechnology Center grant 9903-IDG-1002 to M.J. Thomas. S. Potteaux was funded by Fondation pour la Recherche Medicale (France) and a 2007 Norman Alpert Visiting Scientist Award from the European Society of Cardiology/AHA. E.L. Gautier was funded by the AHA (10POST4160140).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2011;121(5):2025–2036. doi:10.1172/JCI43802.