Ligation of Macrophage Fcγ Receptors Recapitulates the Gene Expression Pattern of Vulnerable Human Carotid Plaques

Stroke is a leading cause of death in the United States. As ∼60% of strokes result from carotid plaque rupture, elucidating the mechanisms that underlie vulnerability is critical for therapeutic intervention. We tested the hypothesis that stable and vulnerable human plaques differentially express genes associated with matrix degradation. Examination established that femoral, and the distal region of carotid, plaques were histologically stable while the proximal carotid plaque regions were vulnerable. Quantitative RT-PCR was used to compare expression of 22 genes among these tissues. Distal carotid and femoral gene expression was not significantly different, permitting the distal carotid segments to be used as a paired control for their corresponding proximal regions. Analysis of the paired plaques revealed differences in 16 genes that impact plaque stability: matrix metalloproteinases (MMP, higher in vulnerable), MMP modulators (inhibitors: lower, activators: higher in vulnerable), activating Fc receptors (FcγR, higher in vulnerable) and FcγR signaling molecules (higher in vulnerable). Surprisingly, the relative expression of smooth muscle cell and macrophage markers in the three plaque types was not significantly different, suggesting that macrophage distribution and/or activation state correlates with (in)stability. Immunohistochemistry revealed that macrophages and smooth muscle cells localize to distinct and non-overlapping regions in all plaques. MMP protein localized to macrophage-rich regions. In vitro, treatment of macrophages with immune complexes, but not oxidized low density lipoprotein, C-reactive protein, or TNF-α, induced a gene expression profile similar to that of the vulnerable plaques. That ligation of FcγR recapitulates the pattern of gene expression in vulnerable plaques suggests that the FcγR → macrophage activation pathway may play a greater role in human plaque vulnerability than previously appreciated.


Introduction
The morbidity and mortality associated with stroke results from the rupture of vulnerable carotid plaques; plaque removal by carotid endarterectomy significantly decreases stroke risk [1]. In contrast, femoral plaques are histologically and clinically stable, with symptoms resulting from artery occlusion rather than plaque embolism. As stabilizing vulnerable plaques represents a potential therapeutic option, understanding the molecular differences between stable and vulnerable plaques is essential for the design of interventional therapies. Additionally, distinguishing patients with plaques more susceptible to rupture from those that could be managed medically would increase treatment choices for physicians and patients.
A major factor contributing to plaque rupture is degradation of the extracellular matrix. The matrix is predominantly collagen and elastin [2], which can be degraded by matrix metalloproteinases (MMP), specifically collagenase (MMP-1, 8), gelatinase (MMP-9) and elastase . MMPs are proteases, secreted as inactive precursors and activated by proteolytic cleavage, mainly by the plasmin system. Plasmin is generated from plasminogen by urokinase-type plasminogen activator (uPA). When bound to its receptor, uPAR, uPA can produce plasmin for focal activation of MMPs. The proteolytic activity of mature MMPs is regulated by their association with tissue inhibitors of metalloproteinases (TIMPs). Thus, the MMP:TIMP ratio determines overall MMP activity [3].
Vulnerable plaques express elevated levels of MMP-1, MMP-8, and MMP-9 that co-localize with activated macrophages [4]. Numerous plaque components activate macrophages, including cytokines [5], oxidized LDL [5], C-reactive protein [6] (CRP), and immune complexes [7] (IC). Despite the fact that IC are present in plaques and activate macrophages through Fcc receptor (FccR) cross-linking, it is only now beginning to be recognized that IC can influence plaque progression though FccR signaling. The generation of atherosclerosis prone (ApoE-/-and LDLR-/-) FccR -/-double knockout mice enabled the role for FccR in atherosclerosis to be studied in a mouse model. When fed a high fat diet, plaque area is significantly higher in single vs double knock-outs [8][9], implicating FccR in atherogenesis. However, the link between FccR activation and plaque vulnerability in humans is understudied. We postulate that plaque macrophages take up IC and secrete MMPs for matrix degradation. Herein we present evidence that vulnerable regions of human carotid plaques have a pattern of gene expression significantly different from paired proximal regions that are histologically stable. Additionally, we demonstrate that engagement of macrophage FccRs with IC induces a pattern of gene expression similar to that of the vulnerable plaques, suggesting that FccR-activated macrophages may play a greater role in plaque vulnerability than previously appreciated.

Patient selection criteria
This investigation conforms to the principles outlined in the Declaration of Helsinki. Specimens were procured and processed according to approved IRB protocols. The study is classified as exempt category 4; no informed consent was necessary. Femoral and carotid endarterectomy tissue was provided by The Vascular Group, PLLC based at Albany Medical Center. Femoral plaques (n = 14) were extracted by longitudinal arteriotomy, usually as an adjunct to distal bypass; carotids (n = 16) were removed using the eversion method from symptomatic patients (3/16 or 19%) and asymptomatic patients (13/16 or 81%) with .60% stenosis. Tissue was collected sequentially over a 6-month period with the only criteria for exclusion being absence of a morphologically distinct distal and proximal segment, making it an inherently randomized study.

Tissue Collection and Quantitative RT-PCR
Immediately upon removal, the tissue was placed in RNAlaterH (Qiagen, Inc. Valencia, CA) and transported to the lab for processing. Carotid plaques were divided into proximal (at, or near, the carotid bifurcation) and distal (up the internal carotid artery). From each carotid region, two contiguous blocks were taken, one for paraffin embedding and one preserved in OCT (optimal cutting temperature) for immunohistochemistry. RNA was extracted from the remainder of the tissue. Proximal and distal carotid and femoral tissue was weighed and RNA extracted with Tri-reagent (Molecular Research Center, Cincinnati, OH). Extraction was facilitated by homogenization with a Brinkman PolytronH. cDNA was prepared from RNA using iScript (Bio-Rad, Hercules, CA). qPCR primers (Table S1) were designed to amplify 100-200 bp fragments across an intron and were unique as determined by BLAST analysis. Relative abundance of mRNA was normalized to b-actin and calculated as 2 -(Ct gene -Ct actin) where Ct represents the threshold cycle for each transcript. The limits of detection were set at Ct $35. As the amount of RNA recovered varied with the specimen, the genes were prioritized (MMP, FccR.MMP activators, inhibitors, signaling genes). Thus, the number of samples varies among the genes (n = 8-16).

Histology
Paraffin-embedded tissue was sectioned (7 mm), stained with Tri-chrome, and classified using the American Heart Association stability criteria [2]. OCT-embedded tissue was cryosectioned (7 mm) and mounted on Superfrost slides. Fixation (acetone, 10 min) and staining was done at room temperature. Sections were blocked (45 min, PBS containing 1% Tween-20 and 10% horse serum) and primary antibodies incubated for 1 h. Biotinylated secondary antibodies (1:200, 1 h) were visualized using the Vectastain ABC system (Vector Laboratories, Burlingame, CA). Primary antibodies were mouse monoclonals (Table S2); isotype controls (25 mg/ml) had no detectable staining. Immunohistochemistry was done on 3-4 different proximal plaques. Despite variations in plaque morphology (i.e., percent stenosis, eccentricity, distribution of fibrous tissue), the localization of proteins with respect to macrophages and smooth muscle cells was reproducible.

Macrophage stimulation
Human monocytes, purified from peripheral blood by counterflow centrifugal elutriation, were purchased from The University of Nebraska Medical Center. Elutriation separates the monocytes from other blood cells based on their sedimentation in a centrifugal field. The process uses physiological media, thus maintaining cell viability and function. Monocytes were differentiated (200 ng/ml recombinant human M-CSF, Cell Sciences, Canton, MA) for 7 days. THP-1s, a human macrophage-like cell line (ATCC #TIB-202), were differentiated with PMA (8.1 mM, 18 h) before use. Monocytes-derived macrophages (5610 6 ) and differentiated THP-1 cells (1610 6 ) were stimulated (18 h) with 200 ng/ml human TNF-a (Cell Sciences, Canton MA), 50 mg protein/ml oxLDL (Invitrogen, Carlsbad, CA), 40 mg/ml Creactive protein (Calbiochem, San Diego, CA). The concentrations used reflect serum concentrations in patients with atherosclerosis [6,11]. LDL was oxidized with 5 mM CuSO 4 (20 h, 37uC); oxidation was confirmed using the TBARS assay with malondialdehyde as the standard [12]. Average level of oxidation was 18-23 nmol malondialdehyde equivalents/mg protein. Immune complexes (IC) were presented as IgG-opsonized 2 mm beads; 10 beads/cell reflecting standard phagocytosis ratios [13]. Following stimulation, RNA was extracted and qPCR performed as above.

Statistical Analysis
The characteristics of the patient population were compared using unpaired Student's T-test and Fisher's Exact Test. For the gene analysis, the most appropriate statistical analyses were performed. Specifically, as the tissue came from different patients, gene expression in femoral and stable carotid specimens was compared using an unpaired Student's T-test. In contrast, the vulnerable and stable carotid tissue came from the same plaque. Thus, gene expression in these paired samples was compared using a paired Student's T-test. Within a gene, the tests were corrected for multiple comparisons using Bonferroni's correction and significance accepted at the 0.05 level. For MMPs, uPA, and uPAR, a logarithmic transformation was applied prior to significance testing to correct for heteroscadisicity. As expected, there is variability among the specimens. Although the extreme outlyers may be due to unknown complications in the patient, none were excluded. With a minimum of 8 patients, we estimate that we would have adequate power (.80%) for detection of differences between stable to vulnerable that are as large or larger than 120% of within-group standard deviation.

Results and Discussion
Full demographic details of the 30 patients in the study are summarized in Table 1. No statistically significant differences were found between the carotid and femoral patient pools with regards to age, sex, hypertension, smoking, diabetes, and preexisting coronary artery disease; hyperlipidemia was higher in carotid vs femoral patient pools.

Stable plaques, regardless of anatomical location, have similar gene expression patterns
To mitigate against stroke, carotid endarterectomy of asymptomatic plaques with $60% stenosis is recommended [14,15,16]. Carotid endarterectomy specimens are often visually and hsitologically heterogeneous ( Figure 1A). While a necrotic/lipid core is present throughout, the core of the distal (relative to the heart) region was relatively small and protected by a thick fibrous cap while there was little or no cap overlying the proximal core ( Figure 1A). In contrast, proximal plaque regions exhibited multiple histological indicators of instability (i.e., thin cap, intraplaque hemorrhage, large necrotic core). This heterogeneity is of considerable etiological interest. Is it possible that the fibrous regions represent an earlier stage in plaque development, with expansion of the fatty/necrotic core and activation of matrix degradation a subsequent event leading to plaque vulnerability. Alternatively, plaque histology may vary as a function of blood flow, with that in the internal carotid (distal) more laminar and that in the common carotid (proximal) being more disturbed. As human specimens represent an endpoint in the disease process, such etiological questions must be answered in animal models.
In contrast to their carotid counterparts, femoral plaques rarely cause symptoms and thus are considered clinically stable, being removed or bypassed to relieve lower limb ischemia. Trichrome staining revealed that both femoral and distal carotid plaque tissue was fibrocellular, with cells (red) embedded in an extensive extracellular matrix (blue) (Figure 1, n = 14 and 16, respectively). Thus, we tested the hypothesis that their gene expression pattern was similar. Of particular interest were genes for macrophages (CD68) and smooth muscle cells (a-actin), matrix degradation (MMPs, TIMPs, uPA, and uPAR), macrophage activation (FccRs, TNF-a, MHC II, gp91 phox , PKC-d, iNOS), the inhibitory FccR IIb, and the scavenger receptor for oxLDL (SR-A). A qRT-PCR comparison between the femoral plaques and distal carotid plaque regions revealed no significant differences in expression of 20/22 genes ( Figure 2); only SR-A and uPAR were significantly different (higher in femoral specimens).
SR-A is a scavenger receptor that internalizes oxLDL and contributes to foam cell formation. That it was higher in femoral plaques is intriguing as it suggests that elevated SR-A per se may not predispose a plaque to rupture. As soluble oxLDL is taken up by     SR-A but doesn't generate a respiratory burst [17], SR-A may aid in the uptake of oxLDL by a mechanism that doesn't activate macrophages. Indeed, neither PKC-d nor gp91 phox , genes involved in the respiratory burst, were differentially expressed in femoral tissue ( Figure 2B). Thus, elevated SR-A levels may be a physiological response to the increased circulating levels of oxLDL in patients with cardiovascular disease [18].
Most carotid plaques were visually heterogeneous, raising the question as to the differences in the regions ( Figure 1A). That the distal region of carotid plaques had a gene expression pattern similar to clinically stable femoral plaques, allowed us to use the more powerful paired analysis to compare gene expression in the proximal and distal portions of the same carotid plaque. As the proximal portions of our carotid specimens had large necrotic cores, thin or no fibrous caps, and evidence of hemociterin laden foam cells (indicative of vulnerability), we predicted that they would contain more macrophages. Surprisingly, there was no significant difference in the expression of CD68 (macrophages) and a-actin (smooth muscle cells) between proximal and distal regions of carotid plaques (Figure 2). When stained for protein, CD68 and a-actin did not overlap (Figure 3). Thus, the differences in the amount of matrix (blue staining in Figure 1) could not be explained by the separation of collagen-producing smooth muscle cells from macrophages. Rather, the activation state of the macrophage may account for the decreased matrix in the proximal carotid plaques.

Gene expression in proximal regions favors matrix degradation
When gene expression was compared between proximal and distal regions of the same carotid plaque, 17 genes were differentially expressed (Figure 4).
MMPs. Proximal carotid tissue had higher MMP-1, -8, -9 and -12 mRNA expression compared to their corresponding distal regions (Figures 2,4). Immunohistochemistry revealed that MMP-8, -9 and -12 protein co-localized with macrophages in proximal regions; MMP-1 was predominantly associated with smooth muscle cells ( Figure 5). Gelatin zymography was used to determine relative levels of MMP-9 protein in stable and vulnerable tissue; Figure 6 presents the results of gelatin zymography on 3 different carotid plaques. In all cases the proximal plaque regions contained more MMP-9/unit protein when compared to the corresponding distal segments. Thus, although not all MMP protein levels were examined, the fact that 1) MMPs are constitutively secreted by plaque macrophages [19], 2) they co-localized with macrophages ( Figure 5), and that 3) mRNA for multiple MMPs is elevated in proximal carotid plaque tissue, is consistent with a model in which macrophages in proximal carotid plaques release MMPs that degrade the fibrous cap and destabilize the plaque. Although this concept is supported by published studies, this work is unique in that a comparison of gene expression has been done in stable and vulnerable regions of the same carotid plaque and has been combined with other markers of MMP regulation and macrophage activation.
uPA and uPAR. The uPA/uPAR system processes MMPs to their active forms. Accordingly, uPA and uPAR mRNA levels were significantly higher in proximal regions (Figures 2, 4). The combination of elevated uPA, uPAR, and MMPs favors MMP activation and extracellular matrix breakdown.
Tissue inhibitors of metalloproteinases (TIMPs). TIMPs bind to, and inhibit the activity of, mature MMPs. TIMP-2 and-3 mRNA were lower in proximal, compared to distal, regions (Figures 2, 4). TIMP-2 protein was distributed throughout the macrophage and smooth muscle cell regions while TIMP-3 localized to macrophages (Figure 7). Reduced TIMP-2 and-3 levels, combined with higher MMPs, would contribute to matrix degradation. TIMP-1 message was significantly higher in proximal vs distal regions (Figures 2, 4) and TIMP-1 protein localized to macrophages (Figure 7). This is consistent with reports that TIMP-1 overexpression does not alter plaque stability [20]. TIMP-4 mRNA was low and similar in both regions (Figure 2). This is not surprising as TIMP 4 is predominantly expressed in cardiac tissue.
Taken together, our results suggest that, compared with their distal counterparts, proximal carotid plaque regions contain higher levels of active MMP-9 (lower band, Figure 6) and MMP activating genes (uPA, uPAR, Figure 2). In these proximal regions, reduced TIMP-2 and-3, combined with higher MMPs, would favor matrix degradation (Figure 1), consistent with a phenotype more likely to rupture. Macrophages are activated in proximal regions of carotid plaques Several subtypes of macrophages have been described, activated in different ways and expressing unique cadres of genes. Classically activated macrophages (M1 subtype) are pro-inflammatory, expressing elevated levels of FccRI, FccRIIa, FccRIII, TNF-a, and releasing reactive oxygen species (ie, elevated NADPH oxidase) [21,22]. In contrast, the M2 macrophages are immunoregulatory, with lower expression of FccRI, FccRIIa, FccRIII (compared to M1), TNF-a, and reactive oxygen metabolites. As MMPs associate with macrophages in the proximal plaque regions ( Figure 5), we evaluated the expression of genes associated with macrophage activation, including FccRI, FccRIIa, FccRIII, TNFa, MHCII, PKC-d, and gp91 phox ; PKC-d and gp91 phox were used as indicators of elevated NADPH oxidase. With the exception of MHCII, all of the genes were higher in proximal vs distal regions ( Figure 2). This gene expression pattern is consistent with the presence of M1, classically activated macrophages [23] in the proximal plaque tissue. With their elevated expression of activating FccR, proximal plaque macrophages would be primed for proinflammatory responses to immune complexes present in plaques.
In summary, the pattern of gene expression in proximal carotid plaque regions is consistent with activated macrophages (higher FccR, TNF-a, gp91 phox , PKC-d) and increased MMP activity (i.e., higher MMP, uPA, and uPAR expression and lower TIMP 2,3) ( Figure 4). These results define a cadre of genes, elevated in proximal plaque tissue, that can cooperate to increase the level of active MMPs, contributing to matrix degradation and increased plaque vulnerability.
FccR ligation in macrophages recapitulates the gene profile of proximal carotid plaques Macrophages can be activated by many of the constituents of plaques, notably TNF-a, oxLDL, C-reactive protein (CRP), and immune complexes (IC). To assess the effects of macrophage activation on the expression of the genes associated with vulnerable Figure 8. TNF-a does not induce a vulnerable gene profile in macrophages. Differentiated THP-1 cells or monocyte-derived macrophages were treated with 200 ng/ml human TNF-a for 18 h. RNA was extracted and subjected to qPCR. Data are presented as mean 6 SEM (normalized to bactin; n indicated in control column). Statistical significance compared by paired t-test for each replicate. * p,0.05, ** p,.01, *** p,.001. Yellow boxes denote genes similarly regulated in proximal vs distal plaque tissue. Green and red numbers indicate significantly increased or decreased expression, respectively compared to controls. = Significance calculated from log values to compensate for heteroscedacity of the data. doi:10.1371/journal.pone.0021803.g008 carotid plaques, differentiated THP-1 cells and monocyte-derived macrophages were treated with TNF-a, oxLDL, CRP, or IC and the resulting changes in gene expression were quantified. Of the 16 genes differentially expressed in proximal plaques (Figure 4), TNF-a induced three in THP-1 cells and 7 in primary macrophages ( Figure 8). Of the MMPs detected in proximal plaques, only MMP-9 was induced by TNF-a; TIMP levels were unaffected (Figure 8). CRP did not alter the expression of any of the genes (data not shown). Although oxLDL induced expression of 8 genes (Figure 9), several genes involved in matrix degradation (eg, MMP-9, TIMP-1, uPAR) were decreased, arguing against SR-A/oxLDL signaling as a major contributor to matrix degradation. This finding is consistent with our results demonstrating increased expression of SR-A mRNA in femoral plaques. Although SR-A is increased, its ligation with oxLDL may not induce MMP production and the plaques remain fibrous.
Of the four compounds tested, IC induced a pattern of gene expression most closely resembling that of proximal carotid plaque tissue. Fifteen of the 16 genes were similarly regulated in ICtreated THP-1 cells, 12 of 16 in primary macrophages (Figure 9). Of particular relevance to plaque vulnerability is that IC upregulated the MMPs and down-regulated TIMPs 2 and 3. Thus, IC, but not TNF-a, oxLDL, or CRP, induced a pattern of gene expression in macrophages that recapitulates expression in vulnerable regions of carotid plaques, consistent with a model in which FccR-dependent signaling contributes to plaque vulnerability.
If FccR ligation contributes to an inflammatory environment, conditions that increase FccR expression and/or signaling would be expected to be pro-atherosclerotic. Indeed, monocytes from patients with acute coronary syndrome have significantly higher FccR expression [24]. Also, individuals expressing the R131 allele of FccR IIa have increased peripheral atherosclerosis compared to those expressing H131 [25]. Several other studies are consistent with a model in which FccR signaling contributes to atherosclerosis. First, lovastatin inhibits FccR-mediated phagocytosis [26]. Lower phagocytosis translates to less inflammation, a property of the drug that may contribute to its' atherosclerosis-reducing effects. Secondly, the double TNF-a/Apo E knockout mouse has decreased atherosclerosis when compared to the Apo E -/-single knockout [27]. Thus, TNF-a, which is rapidly produced upon FccR ligation, and is up-regulated in vulnerable carotid plaques (Figures 2, 4), is pro-atherosclerotic. Finally, and most directly relevant, Apo E-/-mice lacking FccRI and FccRIII, and LDLR-/-mice deficient in FccRIII have significantly less atherosclerosis than their Apo E/LDLR single knockout counterparts [8][9]. Conversely, deletion of the inhibitory FccR IIb in Apo E-/-or LDLR -/-mice promotes atherosclerosis in the descending Figure 9. Immune complexes recapitulate the gene expression pattern of vulnerable plaques. Differentiated THP-1 cells or monocytederived macrophages were treated with immune complexes (10 beads/cell) or oxidized LDL (50 mg/ml). RNA was extracted and subjected to qPCR. Data are presented as mean 6 SEM (normalized to b-actin; n indicated in control column). Statistical significance compared by paired t-test for each replicate. * p,0.05, ** p,.01, *** p,.001. Yellow boxes denote genes similarly regulated in proximal vs distal plaque tissue. Green and red numbers indicate significantly increased or decreased expression, respectively compared to controls. = Significance calculated from log values to compensate for heteroscedacity of the data. doi:10.1371/journal.pone.0021803.g009 artery or aortic root [28,29]. Our results provide a potential mechanism by which activating FccR may contribute to matrix degradation and plaque vulnerability.
Admittedly, as macrophages in the plaque milieu are exposed to multiple stimuli this is an assessment focusing on the FccR R MMP axis. Indeed, recent studies suggest that IC in conjunction with toll-like receptor ligation, may generate a protective response (review, [21]). However, the up-regulation of the activating FccR, TNF-a, and gp91 phox suggest that the proximal carotid plaque environment is conducive to M1 (inflammatory) macrophage polarization. Laser capture microdissection will enable macrophages to be recovered from plaques. Expression profiling of such macrophages will permit a comprehensive comparison of genes differentially expressed in macrophages from vulnerable and stable plaques and their comparison to the profile generated by activation of macrophages through the FccR. However, the results presented herein raise the intriguing possibility that FccR ligation of plaque macrophages contributes to plaque instability, a model that, to date, has received little attention.

Supporting Information
Table S1 Sequence of qRT-PCR primers used in these studies. (DOC)