GM-CSF regulates intimal cell proliferation in nascent atherosclerotic lesions

BrdU is stably incorporated into the DNA of cells during the S phase of the cell cycle. Pulse labeling is achieved because injected BrdU is cleared from the plasma within hours (Kriss and Revesz, 1962). BrdU labels proliferating monocyte precursors in the BM, and labeled monocytes appear in the circulation after 4–6 h (Goto et al., 2003; Jongstra-Bilen et al., 2006). We used flow cytometry to investigate BrdU labeling of BM and blood monocyte subsets in wild-type C57BL/6 mice (Fig. 1 a). BrdU⁺ Ly-6C^highCD115^high cells were detected in the BM at 2 h, increased, and reached a plateau at 12 h (Fig. 1 b). In the blood, BrdU⁺ Ly-6C^high monocytes were not detected within 6 h but increased rapidly thereafter and reached a plateau at 12–24 h (Fig. 1 b), at which time 35–45% of circulating blood Ly-6C^high monocytes were BrdU⁺. Ly-6C^negCD115^high BM cells and blood monocytes did not stain for BrdU. The mean fluorescence intensity (MFI) of BM BrdU-labeled Ly-6C^highCD115^high cells declined over time, consistent with ongoing rapid proliferation of monocyte precursors (Fig. 1 c). In contrast, the MFI of BrdU⁺ Ly-6C^high blood monocytes remained relatively stable over 24 h.

Wild-type mice are resistant to developing diet-induced hypercholesterolemia and atherosclerosis. In atherosclerosis-susceptible Apoe^−/− mice, prolonged and marked hypercholesterolemia as a result of feeding a high fat diet for 12–25 wk induced a substantial increase in the number of circulating Ly-6C^high monocytes, which are preferentially recruited to plaques (Swirski et al., 2007; Tacke et al., 2007). The effects of short-term hypercholesterolemia on monocyte release from the BM and circulating monocyte levels is unknown. We fed LDL receptor–null (Ldlr^−/−) mice a cholesterol-rich diet (CRD) for just 2 wk and found that the number of Ly-6C^high blood monocytes was increased 2.9-fold (Fig. 1 d). In contrast, circulating Ly-6C^neg monocytes were not significantly increased by hypercholesterolemia (Fig. 1 d). There was a trend toward higher BrdU labeling of Ly-6C^high blood monocytes at 12, 18, and 24 h in hypercholesterolemic mice, and a statistically significant difference (P < 0.03) between hypercholesterolemic and normocholesterolemic mice was found when data from these time points were combined (Fig. 1 e).

Feeding Ldlr^−/− mice a CRD initiates a marked hypercholesterolemia and atherosclerotic lesion formation throughout the aorta (Ishibashi et al., 1994a; Lichtman et al., 1999). As early as 1 wk after CRD feeding, nascent lesions were detected in the lesser curvature (LC) of the Ldlr^−/− ascending aorta, and lesion surface area increased progressively with time (Fig. 2 a). Cell proliferation in the ascending aortic intima was examined 2 h after injection of BrdU, a time point when BrdU⁺ monocytes were not detected in the circulation (Fig. 1 e). BrdU⁺ nuclei were also quantified 24 h after BrdU injection when proliferating intimal cells as well as BM-derived BrdU-labeled Ly-6C^high monocytes recruited from the blood are detected. As we reported previously (Jongstra-Bilen et al., 2006), in the ascending aorta of normal C57BL/6 mice (Ldlr^+/+), we found very low proliferation at 2 h only in the LC (0.5 BrdU⁺ nuclei; Fig. 2 b). At 24 h, BrdU⁺ nuclei increased by 20-fold to 10.3 ± 2.3 nuclei, which represents mostly recruited BrdU-labeled Ly-6C^high monocytes and is consistent with our previous findings (Jongstra-Bilen et al., 2006). In Ldlr^−/− mice fed a standard chow diet (SCD; Fig. 2 b, time point 0), the numbers of BrdU⁺ nuclei at 2 and 24 h were virtually identical to wild-type C57BL/6 mice (0.5 ± 0.3 and 10 ± 2, respectively), indicating relatively comparable baseline rates of cell proliferation and monocyte recruitment to the intima. After switching Ldlr^−/− mice to a CRD, the number of BrdU⁺ nuclei at 2 and 24 h progressively increased, which demonstrates a parallel increase in both intimal cell proliferation and monocyte recruitment from the inception of atherosclerosis. At 2 wk of CRD, intimal cell proliferation increased by 40-fold relative to C57BL/6 and Ldlr^−/− mice fed a SCD, and the abundance of BrdU⁺ nuclei detected at 24 h increased by approximately 8-fold.

All BrdU⁺ nuclei were located in the LC (Fig. 2, d and e) and many (88 ± 4% and 66 ± 12% at 2 and 24 h, respectively) were at borders of lesions (Fig. 2, c and d), suggesting that cell proliferation and monocyte recruitment occur in similar regions. Proliferating intimal cells at borders of lesions were often adjacent to cells containing lipid droplets, and their nuclei extended below the endothelial cell monolayer (Fig. S1, a–c).

BrdU labeling in lesions was initially studied within 6 h of BrdU injection, when minimal BrdU⁺ monocytes are released from the BM (Fig. 1 e). Intimal BrdU⁺ nuclei were detected at 1 h, increased at 2 and 3 h, and then remained stable (Fig. S1 d). This suggests that the BrdU labeling window is <3 h. Another possibility is that labeled intimal cells complete mitosis within 3 h. Consistent with this, BrdU⁺ nuclear doublets that likely represent daughter cells also increased within 3 h (Fig. S1, e and f).

Proliferating intimal cells may divide to produce two daughter cells or, alternatively, may undergo apoptosis. The fate of proliferating intimal cells was investigated by inhibiting blood monocyte recruitment with pertussis toxin (PTx). PTx inactivates Gαi chemokine receptors and has been used in vivo to study the contribution of chemokine receptors versus E-selectin–triggered integrin activation in neutrophil recruitment to the inflamed peritoneal cavity (Smith et al., 2004). We found that PTx markedly diminished monocyte recruitment at 24 h to thioglycollate-inflamed peritoneal cavity despite a marked increase in circulating myeloid cells (Fig. 3 a and Fig. S2). In contrast, PTx reduced but did not eliminate neutrophil recruitment, consistent with a previous report (Smith et al., 2004). These data illustrate an important difference in monocytes versus neutrophil signaling. As was shown previously, neutrophil binding to E-selectin induces a compensatory signaling pathway that bypasses Gα_i signaling irreversibly blocked by PTx, whereas our data suggest that this compensatory pathway is not operative in monocytes.

In Ldlr^−/− mice fed a CRD for 2 wk, PTx significantly diminished the number of BrdU⁺ nuclei at 24 h from 86 ± 10 to 26 ± 12 (Fig. 3 b), consistent with PTx blockade of monocyte recruitment. In a parallel experiment, 21 ± 2 BrdU⁺ nuclei representing intimal cells were detected at 2 h. Comparable BrdU⁺ nuclei (26 ± 12) were detected in the PTx experiment at 24 h, suggesting that proliferating intimal cells persist and accumulate in lesions.

Intimal cell proliferation in 2-wk lesions was determined by 2-h BrdU labeling after inhibition of monocyte recruitment for 22 or 46 h with PTx. Proliferation was not decreased by PTx at 22–24 h and, in fact, increased at 46–48 h (Fig. 3 c), which may represent a compensatory increase in cell proliferation in response to diminished blood monocyte recruitment. These data reveal that proliferating intimal cells are not newly recruited blood monocytes.

Immunostaining for various leukocyte markers was performed to determine the phenotype of BrdU⁺ cells in atherosclerotic lesions. The majority of foam cells in 2-wk lesions are CD11c⁺ and CD68⁺, whereas CD11b⁺ cells are located in clusters (Fig. S3). After 2 or 24 h of BrdU labeling, >90% of BrdU⁺ cells expressed CD45, a pan-leukocyte marker (Fig. 4 b). BrdU⁺ cells costained predominantly with CD11b 24 h after labeling (Fig. 4, a and b), consistent with recruitment to lesions of Ly-6C^high blood monocytes, which are CD11b⁺. More than 90% of proliferating cells (2 h of BrdU labeling) costained for CD11c, a DC marker, and >60% of BrdU⁺ cells expressed CD11c at 24 h, consistent with survival of proliferating CD11c⁺ cells.

The expression of additional myeloid and DC markers by proliferating cells in 2-wk lesions was evaluated. 2 h after BrdU labeling, most BrdU⁺CD11c⁺ cells costained with CD68, a marker expressed by monocytes, macrophages, and DCs (Fig. 4, c and d; and Fig. S4). Relatively low costaining with CD11b⁺ was confirmed. Many BrdU⁺CD11c⁺ cells costained with MHC class II (MHC II; Fig. 4, c and d; and Fig. S4), which is expressed by DCs, and with 33D1 (Fig. 4 f), a mAb that specifically targets mouse DCs (Nussenzweig et al., 1982; Xu et al., 2007). 33D1 stained resident intimal DCs in the normal mouse aorta (Fig. 4 e). Collectively, our data indicate that the majority of proliferating intimal cells in early atherosclerotic lesions express DC markers and, thus, may be capable of DC functions analogous to intimal DCs in the normal intima (Choi et al., 2009). Proliferating cells in early lesions may originate from intimal DCs residing in atherosclerosis-predisposed regions of the normal intima (Jongstra-Bilen et al., 2006; Choi et al., 2009) and/or recruited monocytes that differentiated in lesions to express DC markers after several days.

GM-CSF is a myeloid growth factor that induces DC differentiation from BM precursors and monocytes, and regulates the properties of mature myeloid cells in inflammation (Hamilton and Anderson, 2004). Modified LDL induces endothelial cell expression of GM-CSF (Rajavashisth et al., 1990), and GM-CSF down-regulates serum cholesterol while up-regulating VLDL receptor mRNA expression (Ishibashi et al., 1994b). GM-CSF deficiency has variable effects on mouse atherosclerosis. In Ldlr^−/− mice, lesions were decreased (Shaposhnik et al., 2007), whereas in Apoe^−/− mice they were increased (Ditiatkovski et al., 2006). Because the majority of proliferating cells in early lesions are CD11c⁺ (Fig. 3) and GM-CSF deficiency results in decreased CD11c⁺ cells in lesions (Shaposhnik et al., 2007), we investigated whether GM-CSF regulates cell proliferation in atherosclerotic lesions.

Initial experiments used real-time PCR to investigate mRNA expression of the GM-CSF receptor α chain (GMRα) in 2-wk-old lesions of Ldlr^−/− mice. GMRα is specific for GM-CSF, unlike the β chain that is shared by IL-3 and IL-5 (Rosas et al., 2007). The LC intima of the ascending aortic arch, which contains lesions, was compared with the nondiseased greater curvature (GC), a region that is protected from atherosclerotic lesion development. Expression of GMRα was markedly elevated in lesions relative to the uninvolved GC intima (Fig. 5 a). There was also enrichment of DCs and myeloid cell markers (CD11c and CD68) in lesions, but expressions of T cell (CD3), smooth muscle cell (SM22α), and endothelial cell (CD31 and intercellular adhesion molecule 2) markers were comparable. Having established that GMRα is expressed in 2-wk lesions, two complementary experiments directly evaluated the role of GM-CSF in intimal cell proliferation. Injection of mouse GM-CSF increased the abundance of BrdU⁺ nuclei by 2.7-fold in 2-wk intimal lesions (Fig. 5 b), indicating that many cells in lesions are capable of proliferating after exposure to GM-CSF. In this experiment, BrdU⁺ leukocytes were not detected in the circulation. Injection of a function-blocking antibody to GM-CSF reduced the number of proliferating cells in lesions by 4.8-fold (Fig. 5 c). Collectively, our data demonstrate that GM-CSF is critical for intimal cell proliferation in early lesions.

Germline genetic manipulations in mouse models are used to investigate molecular atherogenic mechanisms, yet analysis of atherosclerotic lesions in mice typically consists of lesion burden assessment (by staining for accumulated intimal lipid) and evaluation of cell/matrix composition using histological approaches. Leukocyte recruitment and intimal cell proliferation or death can independently result in differences in lesion burden. This study illustrates a practical, sensitive, and reproducible approach for quantifying cellular events in atherosclerosis. BrdU pulse labeling combined with en face confocal microscopy provides quantitative information on intimal cell proliferation, as well as recruitment of Ly-6C^high blood monocytes within a 24-h window. Using this approach, we showed that proliferation of intimal cells and recruitment of BM-derived Ly-6C^high monocytes increase progressively from the inception of atherosclerosis. Experiments with PTx suggested that the two processes occur independently of each other and raise the possibility that blockade of recruitment or proliferation individually may result in a compensatory response. Thus, for effective therapy it may be necessary to target both processes. Previously, GM-CSF was reported to function in the differentiation of monocytes to DCs (Chapuis et al., 1997; Daro et al., 2000) and in the proliferation of mouse DC precursors (Inaba et al., 1992). Our current data support an important role for GM-CSF in mediating intimal DC proliferation in early atherosclerotic lesions and suggest that it may be prudent to block functions of this cytokine when administering a therapy for early atherosclerosis that targets monocyte recruitment.

Wild-type C57BL/6 and Ldlr^−/− mice backcrossed >10 generations into the C57BL/6 background obtained from the Jackson Laboratory were bred. All procedures were approved by the University Health Network Animal Care Committee. At 10–12 wk of age, Ldlr^−/− males or females were switched to a 1.25% cholesterol diet (Lichtman et al., 1999). In BrdU labeling experiments, mice received a single 0.2-ml i.v. injection of 2 mg BrdU in PBS (BD). In some experiments, BrdU labeling was performed in conjunction with i.v. injection of 2 µg PTx in PBS (Sigma-Aldrich), 0.01 µg/g of body weight of mouse rGM-CSF in 1% FBS (R&D Systems), or 100 µg anti–mouse GM-CSF (functional grade MP1-22E9; eBioscience). Controls were injected with 0.2 ml PBS, 1% FBS, or isotype control rat IgG2a. Peritonitis was induced with 4% thioglycollate (1 ml i.p.; Difco).

Leukocytes obtained from blood or BM of the femur and tibia were stained with PE–anti-CD115 plus biotin–anti–Ly-6C (Biomedicals AG) or biotin–anti-CD19 (BD) mAbs and streptavidin-allophycocyanin (eBioscience). They were then fixed and permeabilized, and DNA was digested with DNase I (Invitrogen) and stained with FITC–anti-BrdU (BrdU detection kit; BD). Data were acquired with a flow cytometer (Cytomics FC500 MPL; Beckman Coulter) and analyzed using FlowJo analysis software (version 7.2.2; Tree Star, Inc.). A hemocytometer (Bright-Line; Hausser Scientific) was used to determine the concentration of blood leukocytes.

Aortas were harvested after perfusion fixation (4% paraformaldehyde, 100 mmHg), and the surrounding adipose tissue was dissected after further fixation for 1 h. For detection of BrdU-labeled intimal cells, aortas were permeabilized (0.5% Triton X-100 for 15 min) and incubated sequentially with 0.3% H₂O₂ (30 min), DNase I (1.5 h at 37°C), FITC-conjugated anti-BrdU (18 h at 4°C), horseradish peroxidase (HRP)–conjugated anti-FITC (1:300 dilution for 1 h; Abcam), and FITC-tyramide (PerkinElmer). Lipid accumulated in lesions was detected by oil red O (Sigma-Aldrich) staining (Lichtman et al., 1999), and nuclei were counterstained with 2 µg/ml Hoechst 33342 (Invitrogen).

For immunophenotyping of BrdU⁺ cells, aortas were incubated overnight with FITC–anti-BrdU and biotinylated mAb to CD11b (BD), CD11c, CD45 (eBioscience), 33D1 (BD), or CD68 (AbD Serotec). mAbs to CD11b, CD11c, and CD45 were detected with streptavidin-HRP (1:300 dilution) and cyanine 3–tyramide (PerkinElmer), and others were detected with streptavidin-PE. PE–anti–MHC II (BD) and Alexa Fluor 647–anti-CD11c (BioLegend) were also used. A confocal microscope (FV-1000; Olympus) with a 60× (NA 1.4) oil objective was used to acquire images.

Enzymatic harvesting of intimal cells, reverse transcription real-time PCR (SYBR green– or FAM-labeled GM-CSF2Rα TaqMan MGB probe; Mm00438331_g1; ABI Prism 7900HT; Applied Biosystems), and normalization of data were performed as previously described (Jongstra-Bilen et al., 2006).

Differences between two groups were determined by an unpaired two-tailed t test and analysis of variance followed by Tukey-Kramer multiple comparison tests were used to compare multiple groups.

Fig. S1 shows the location of proliferating intimal BrdU⁺ cells (a–c), and the time course of proliferating BrdU⁺ nuclei and BrdU⁺ nuclear doublets (d–f) in 2-wk lesions. Fig. S2 depicts flow cytometry analysis of the peritoneal exudates and blood of PTx-treated and control mice. Fig. S3 shows the distribution of BrdU⁺CD11c⁺ proliferating intimal cells costained with CD68 or CD11b in 2-wk lesions. Fig. S4 depicts costaining of BrdU⁺CD11c⁺ proliferating intimal cells with MHC II and CD68.

We thank the Banting and Best Diabetes Centre and the Toronto General Research Institute for contributing to the purchase of the ABI Prism 7900HT real-time PCR machine.

This research was supported by the Heart and Stroke Foundation of Ontario (HSFO; grant T-6107). M.I. Cybulsky is a Career Investigator of the HSFO and a member of the Heart and Stroke/Richard Lewar Centre of Excellence at the University of Toronto.

The authors have no conflicting financial interests.