Local macrophage colony-stimulating factor expression regulates macrophage proliferation and apoptosis in atherosclerosis

Abstract

Objective:

Previous studies have shown that deficiency of macrophage colony-stimulating factor (M-CSF or CSF1) dramatically reduces atherosclerosis in hyperlipidemic mice. We characterize the underlying mechanism and investigate the relevant sources of CSF1 in lesions.

Approach and results:

We quantitatively assessed the effects of CSF1 deficiency on macrophage proliferation and apoptosis in atherosclerotic lesions. Staining of aortic lesions with markers of proliferation, Ki-67 and BrdU, revealed around 40% reduction in CSF1 heterozygous (Csf1^+/−), as compared to wild-type (Csf1^+/+) mice. Similarly, staining with a marker of apoptosis, activated Caspase-3, revealed a 3-fold increase in apoptotic cells in Csf1^+/− mice. Next, we determined the cellular sources of CSF1 contributing to lesion development. Cell-specific deletions of Csf1 in smooth muscle cells (SMC) using SM22α-Cre reduced lesions by about 40%, and in endothelial cells (EC) deletions with Cdh5-Cre reduced lesions by about 30%. Macrophage specific deletion with LysM-Cre, on the other hand, did not significantly reduce lesions size. Transplantation of Csf1^−/− mice bone marrow into Csf1^+/+ mice reduced lesions by about 35%, suggesting that CSF1 from hematopoietic cells other than macrophages contributes to atherosclerosis. None of the cell-specific knockouts affected circulating CSF1 levels and only the SMC deletions had any effect on the percent monocytes in the circulation. Also, Csf1^+/− mice did not exhibit significant differences in Ly6C^high/Ly6C^low monocytes as compared to Csf1^+/+.

Conclusions:

CSF1 contributes to both macrophage proliferation and survival in lesions. Local CSF1 production by SMC and EC rather than circulating CSF1 is the primary driver of macrophage expansion in atherosclerosis.

Graphical Abstract

INTRODUCTION

A prominent feature of atherosclerosis is the presence of cholesterol-laden macrophages, or foam cells^1–6, derived largely from circulating monocytes. The importance of macrophage infiltration in atherosclerosis is supported by a large number of experimental studies^7–11. In particular, in mice, it has been shown that genetic deficiencies of adhesion molecules, chemokines or growth factors for monocyte/macrophages reduce the growth of atherosclerotic lesions^12–20. A recent study provided convincing evidence that macrophage proliferation within the lesion is critical in the progression of atherosclerosis²¹, but the factors responsible for such proliferation are still unclear. Of the hundreds of candidate genes examined in mouse models, deficiency of macrophage colony-stimulating factor (M-CSF or CSF1) has by far the greatest effect on atherosclerosis development^16–20. CSF1 stimulates the differentiation, growth, and survival of monocyte /macrophages through interaction with its receptor CSF1R.

CSF1 is expressed by multiple cell types, including macrophages, endothelial cells (EC) and vascular smooth muscle cells (SMC)^22,23 in 3 distinct isoforms: a soluble form, a heavily glycosylated form that associates with the extracellular matrix, and a membrane bound form that can be released through the action of proteases^24,25. It is generally believed that a soluble, circulating form of CSF1 is synthesized primarily by EC, whereas the other forms of CSF1 are synthesized locally in tissues^25,26. The dramatic effect of CSF1 on the development of atherosclerosis is likely due to effects on monocyte/macrophage function. Thus, CSF1 could influence monocyte polarization, monocyte migration into lesions, or macrophage proliferation or survival. Previous studies have suggested that CSF1 deficiency is associated with increased macrophage apoptosis in atherosclerotic lesions²⁷. Atherosclerosis is thought to be initiated as a result of the accumulation and oxidation of apolipoprotein B-containing lipoproteins trapped in the vessel wall by interaction with glycosaminoglycans. Monocytes then migrate into the the intimal region of the vessel and differentiate into macrophages²⁸. Oxidized low-density lipoproteins (LDL) are potent inducers of CSF1 as well as granulocyte macrophage-CSF (GM-CSF) and granulocyte-CSF (G-CSF) in EC¹⁹. In addition to effects on monocyte/macrophages, CSF1 deficiency augments hypercholesterolemia in mice^17,18.

We now report a detailed examination of the effects of global and cell specific CSF1 deficiency on atherosclerosis development in hypercholesterolemic mice. Our studies show that CSF1 is required for the expansion of macrophages in lesions, promoting both survival and growth. Our experiments, using cell-specific gene targeting and bone marrow transplantation, indicate a key role for local CSF1 production by SMC and EC.

MATERIALS AND METHODS

Mice and diets.

The Institutional Animal Care and Use Committee (IACUC) approved experimental protocols. All experiments were performed in accordance with relevant guidelines and regulations. For macrophage proliferation, survival and phenotype studies, mice heterozygous for CSF1 (Csf1^+/−) were purchased from the The Jackson Laboratory (Bar Harbor, ME) and interbred to generate required number of Csf1^+/− and wild type (WT, Csf1^+/+). The low-density lipoprotein receptor (Ldlr) antisense oligonucleotide (ASO, Ionis Pharmaceuticals, Carlsbad, CA) was administered intraperitoneally²⁹ to 8-10 weeks old Csf1^+/+ and Csf1^+/− female mice for three weeks. Then mice were fed high fat and high cholesterol atherogenic diet (Research Diets Inc., New Brunswick, NJ) for 11 weeks and the Ldlr ASO was administered weekly. For bone marrow transplantation experiments, we used Csf1^+/+ and null (Csf1^−/−) donor male mice on C57BL/6J background. The bone marrow recipient mice were Ldlr^−/− female mice on similar background.

Immunofluorescence staining.

To quantitate macrophage proliferation and apoptosis, immunofluorescence staining was performed using a standard protocol³⁰. Briefly, 10 micrometer (μm) thick frozen aortic sections were brought to the room temperature (RT), fixed in 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked with 5% normal goat serum and 3% bovine serum albumin. Subsequently, sections were incubated with anti-Ki-67 (1:750 dilution, abcam, Cambridge, MA) or activated caspase-3 (1:1000 dilution, abcam) together with anti-Mac-3 (1:400 dilution, BD Biosciences, San Jose, CA) overnight at 4°C. After washing with phosphate buffer saline (PBS), corresponding secondary antibodies (goat anti-rabbit (Alexa Fluor® 488, Life Technologies, Waltham, MA) and goat anti-rat (Alexa Fluor® 594, Life Technologies) were applied for 1 hour at room temperature and cover slips were mounted with fluoroshield containing DAPI (Sigma, St. Louis, MO). The images were taken using Nikon (Eclipse Ti-s) and Zeiss LSM 900 confocal microscopes.

Cell-specific deficiency studies of CSF1 using Cre-Lox recombination.

Homozygous Csf1 loxP-flanked mouse (Csf1^fl/fl) was a generous gift from Dr. Sherry Werner, UT Health Science Center, San Antonio, TX. We obtained Lysozyme M-Cre (LysM-Cre), smooth muscle protein 22-alpha-Cre (SM22α-Cre) and VE-Cadherin-Cre (Cdh5-Cre) transgenic mice from the The Jackson Laboratory. For macrophages conditional knockout (cKO), Csf1^fl/fl mice were bred to LysM-Cre mice to generate heterozygous which were then backcrossed to Csf1^fl/fl mice to produce four groups: Csf1^fl/fl, and Csf1^fl/+ on a LysM-Cre background, and Csf1^fl/fl, and Csf1^fl/+ on a WT background (no Cre). For SMC specific Csf1 cKO, SM22α-Cre transgenic mice were bred to Csf1^fl/fl mice and the progeny backcrossed to Csf1^fl/fl to generate Csf1^fl/fl and Csf1^fl/+ mice on a SM22α-Cre background. Similarly, for EC specific Csf1 cKO, mice transgenic for Cdh5-Cre were bred to Csf1^fl/fl mice and the progeny backcrossed to Csf1^fl/fl to generate Csf1^fl/fl and Csf1^fl/+ mice on a Cdh5-Cre background. The Csf1^fl/fl, and Csf1^fl/+ mice on a WT background (no Cre) were used as control (Figure I in the Data Supplement). We used 8-10 week old, female and male mice. Mice were made hyperlipidemic by overexpressing proprotein convertase subtilisin/kexin type 9 (PCSK9, Vector Biosystems Inc., Malvern, PA) using an adeno-associated viral vector (1.5 x 10¹¹ genome copies per mouse) and feeding of an atherogenic diet for 11-12 weeks. A single injection of PCSK9 in combination with a high-fat diet was sufficient to increase plasma LDL cholesterol and induce atherosclerosis in the mice³¹.

Genotyping.

Genotyping of transgenic mice was performed by RT-PCR using genomic DNA isolated from mouse tails. Three primers were used for genotyping the CSF1 floxed mice: 5’-ACAGGCAGGCCTCTGATCTA-3’; 5’-CCCAGCCAAGAATTCTCCTT-3’; and 5’-GGGCACTCTCCATCTTACCA-3’. The WT allele at 280 bp, floxed allele at 380 bp, and cKO allele at 619 bp were detected. For LysM-Cre, tail DNA was amplified using 5’-CTTGGGCTGCCAGAATTTCTC-3’, 5’-CCTCACCCCAGCATCTCTAATTC-3’ and 5’-ATCACTCGTTGCATCGACCGGTAA-3’ which yield 500 bp and 250 bp products. Animals carrying 500 bp were identified as WT control while animals with 500 and 250 bp products were identified as LysM-Cre heterozygotes. To genotype SM22α-Cre, 5’-CGCATAACCAGTGAAACAGCATTGC-3’ and 5’-CAGACACCGAAGCTACT CTCCTTCC-3’ primers were used. Mice heterozygous in SM22α-Cre showed a band of 500 bp. Similarly, Cdh5-Cre heterozygous mice were genotyped using primers: 5’-GGACCGACGATGAAGCATGT-3’ and 5’-GCAACGAGTGATGAGGTTCG-3’ which yields a 250 bp product. Genomic DNA from bone marrow derived macrophages, vascular SMCs and vascular ECs from Cre mice was used to determine Csf1 cKO (Figure II in the Data Supplement). PCR products were visualized after electrophoresis through 1.5% agarose. All cKO mice appeared healthy.

In Vivo 5-Ethynyl-2′-Deoxyuridine (EdU) and Bromo-deoxyuridine (BrdU) Incorporation and Detection.

EdU incorporation was done as described previously³⁰. Briefly, female mice (Control and Csf1^fl/fl with SM22α-Cre) were administered with PCSK9 and maintained on an atherosclerotic diet for 11-12 weeks. A total of 1.26 mg EdU (from 10 mmol/L stock solution) was injected intraperitoneally. Mice were euthanized 2 hours later, and 10 μm cryo-sections from frozen, optimal cutting temperature (OCT) compound (Tissue-Tek, Elkhart, IN)–embedded proximal aortas were prepared. EdU was detected in these tissue sections using the Click-iT EdU Alexa Fluor 488 Imaging Kit (Invitrogen, Grand Island, NY) using vendor’s protocol. CD68 (Biorad, Hercules, CA) was used to identify macrophages. The number of EdU-positive nuclei was normalized to the total number of CD68-positive macrophages in lesions. We used a Zeiss LSM 900 confocal microscope to visualize the staining. BrdU incorporation was done at a dose of 50 μg per day using osmotic mini-pumps implanted subcutaneously (Alzet, Cupertino, CA; model 1002)²¹ for 2 weeks in Csf1^+/+ and Csf1^+/− female mice. The mice were kept on atherogenic diet for 11-12 weeks. BrdU was stained using anti-BrdU antibody (abcam) and CD68 (Biorad) was used to identify macrophages. All images were taken using a Vectra Polaris (Akoya Biosciences, MA, USA) microscope.

Bone marrow isolation and transplantation.

For isolation of bone marrow cells (BMCs), from all except CSF1 null (Csf1^−/−) mice, femurs were flushed using a 25G 5/8 needle, isolated and kept in a small dish (on ice) containing the RPMI 1640 + 2% FBS + 10 units/ml heparin + penicillin and streptomycin. To isolate bone marrow from Csf1^−/− mice, we used 20-21 week old mice as hematopoietic development is delayed³². Femurs from Csf1^−/− mice were ground with 1ml culture media using a sterilized mortar and the crushed materials was passed through a sterile 40 μm nylon cell strainer (Falcon, Fischer Scientific, Waltham, MA) and centrifuged at 2000 rpm (900 x g), 10 min, 4°C. The cell pellet was washed twice with 50 ml of serum-free RPMI (RPMI 1640 + 20 mM Hepes + penicillin and streptomycin) and centrifuged at 2000 rpm, 5min, 4°C. The pellet was resuspended in media and kept on ice until used. One week before irradiation, the recipient mice were given acidified, antibiotic water (10 mg/ml neomycin and 400 μl of 25 mg/ml polymyxin B sulfate (Sigma, St. Louis, MO). Mice were irradiated with one dose of 1000 rads (we use a 137 Cesium Gamma cell source) and the next day (within 24 hours) mice were injected (via tail vein) with about 1x10⁷ BMCs. The mice were kept on acidic, antibiotic water for 2 weeks after irradiation and then switched to acidic water without antibiotics for the rest of the study. Six weeks after transplantation, the mice were placed on an atherogenic diet for 11-12 weeks. Leukocyte DNA was isolated to confirm the presence of the Csf1^−/− allele (Figure III in the Data Supplement).

Quantification of atherosclerotic lesions and leuckocyte levels.

These were as reported previously³⁰. At sacrifice, the upper portion of the heart and proximal aorta was obtained, embedded in OCT compound, and stored at −70 °C. Serial 10 μm thick cryosections of the aorta, beginning at the aortic root, were collected on poly-D-lysine–coated slides. These sections were stained with Oil Red O and hematoxylin. The images were taken using a Nikon Eclips microscope. Ten sections at 120 μM intervals were counted from each mouse. ImagePro premier software (Media Cybernetics, MD, USA) was used to count the lesion area. This study adhered to the guidelines for experimental atherosclerosis studies as described in the American Heart Association Statement. Circulating leukocyte levels were examined using a HemaTrue analyzer.

Flow cytometry for peripheral blood monocytes.

Cells were prepared as described previously³⁰. Briefly, the peripheral blood from Csf1^+/+ and Csf1^+/− female mice (n=5 per group, on Atherogenic diet for 12 weeks), was obtained from the orbital sinus using EDTA-coated capillary tubes and was then dispensed into 9ml ice cold PBS supplemented with 2 mM EDTA. Samples were centrifuged at 400 x g for 10min at 4°C. Cells were incubated, twice, with 10ml red blood cells lysis buffer (LONZA, Basel, Switzerland) and were washed with 1 ml ice-cold FACS buffer (BD Biosciences). Cells were blocked with anti-CD16/CD32 (Biolegend, San Diego, CA) for 10min and surface antigens on cells were stained for 30 min at 4° C with directly conjugated fluorescent antibodies. Antibodies to the following were purchased from Biolegend and used for flow cytometric analyses: CD45- FITC, CD3, CD19, CD335, Ly6G- PE-Cy7, CD115-BV421, CD11b- APC, and Ly6C- APC-Cy7. The antibody dilutions ranged from 1:200 to 1:400. Debris and doublets were excluded by forward- and side-scatter parameters (FSC-A and SSC-A) and live/dead discrimination was determined using Zombie Yellow dye (1:400, Biolegend). CD45 was used to identify all live immune cells. In this population, monocytes were identified as CD115+ CD11b+ Lin- (Lin defined as CD3+, CD19+, CD335+, and Ly6G+) and Ly-6C^high (Ly-6C^hi) or Ly-6C^low (Ly-6C^lo). Data were acquired on an LSRFortessa flow cytometer (BD Biosciences) and analyzed with FlowJo v8.8.2 (Tree Star, Inc., OR, USA).

Quantitative PCR of Csf1 expression.

Aortas were isolated from chow and atherogenic diet (4 weeks) fed Ldlr^−/− male mice. Adapting previously published methods, mouse aortas were perfused with 1x PBS, dissected, then flushed with Qiazol reagent (Qiagen, Germantown, MD) to lyse EC³³. The remaining aorta tissue, containing mostly SMCs, were homogenized in Qiazol with a mechanical tissue grinder. Peritoneal macrophages were isolated after peritoneal lavage with 1x PBS, spun down at 1200rpm for 5 minutes, and then lysed with 1 ml of Qiazol. Total RNA isolation was performed using the Direct-Zol RNA isolation kit with microprep columns and on-column DNase digestion, per the manufacturers protocol (Zymo Research, Irvine, CA). RNA concentrations were measured using a NanoDrop spectrophotometer (Thermo Fisher, Grand Island, NY). Reverse transcription of 30ng of RNA was performed using Applied Biosystems High-Capacity cDNA RT kit. Gene expression was measured with 45 PCR cycles on a LightCycler 480 (Roche, Basel, Switzerland) instrument with the SYBR FAST 2x qPCR Master Mix (KAPA). The Cp value was normalized with total RNA used to compare across the cell type, while comparing between the diets for each cell type, house keeping genes were used. The primer sequences for the Csf1 cDNA were: 5’-AGTATTGCCAAGGAGGTGTCAG-3’ and 5’-ATCTGGCATGAAGTCTCCATTT-3’³⁴.

Statistical analysis.

Prism 7 (San Diego, CA) was used to perform all statistical analyses. T-tests were calculated by non-parametric Mann-Whitenny test to compare two groups. Ordinary one way ANOVA (Tukey’s multiple comparisons test) was performed to compare the multiple groups. Scattered plots were constructed using Prism 7.

Data availability.

The data that support the findings of this study are available from the corresponding author upon request.

RESULTS

CSF1 deficiency results in decreased proliferation and increased apoptosis of lesional macrophages.

We investigated the hypothesis that CSF1 affects atherosclerosis by affecting macrophage growth and/or survival using 8-10 week old Csf1^+/+ and Csf1^+/− female mice. Heterozygous mice were used since homozygous null mice are unhealthy and tend to die in utero. Hyperlipidemia was induced by knockdown of the Ldlr in liver using an ASO and feeding of a high fat, high cholesterol “atherogenic diet” for 11 weeks when they were sacrificed and 10 μm frozen sections of the proximal aorta prepared.

To examine proliferation, sections were stained with antibody to the Ki-67, a marker of mitosis, as well as Mac-3 antibody to identify macrophages in the lesion. After normalizing for the number of Ki-67 positive nuclei in lesions with the number of nuclei in Mac-3 stained areas, we observed about a 40% decrease in proliferative macrophages (p<0.05) in Csf1^+/− mice as compared to Csf1^+/+ (Figure 1A, 1B). We also performed BrdU incorporation and observed a significant reduction in macrophage proliferation in Csf1^+/− mice as compared to controls(1C, 1D). Staining the sections with the apoptosis marker activated caspase-3 antibody (Figure 1E, 1F) indicated that CSF1 deficiency resulted in about a 3-fold increase in apoptosis of lesional macrophages, supporting a previous finding using a TUNEL assay²⁷. Consistent with our previous observations¹⁸, Csf1^+/− mice in this model also showed about a 65% reduction in atherosclerotic lesions compared to Csf1^+/+ mice (1G, 1H). We further observed a slightly reduced levels of non-HDL cholesterol and circulating CSF1 in Csf1^+/− mice as compared to Csf1^+/+ (Figure 1I, K, and L) while there was no difference in HDL cholesterol levels (Figure 1J) and circulating monocyte percent (Figure 1M) between these groups.

Figure 1. CSF1 deficiency reduces macrophage proliferation and promotes macrophage apoptosis.

(A-F) Macrophages in lesions from hypercholesterolemic CSF1 wild type (Csf1^+/+, top) and heterozygous deficient (Csf1^+/−, bottom) mice were examined for presence of Ki-67 (A, B), BrdU incorporation (C, D), or activated Caspase-3 (E, F). Representative images are shown at magnification X10 (A, E) and X20 (C) along with the percent positive macrophages (B, D, F). (G-M) Hypercholesterolemic Csf1^+/+ and Csf1^+/− mice were examined for lesion morphology at magnification X4 (G) and lesion area (H). total plasma cholesterol (I) HDL cholesterol (J) non-HDL cholesterol (K) plasma CSF1 (L) and circulating monocyte percent (M).

Deficiency of CSF1 in SMC reduces atherosclerosis.

Previous studies have shown that SMC are capable of abundant CSF1 expression in a steady state and in response to various stimuli²². It has also been reported that CSF1 protein expression is increased in atherosclerotic lesions of rabbits as compared to normal aorta³⁵. As described in the Methods section, we generated mice deficient in CSF1 in SMC using cell specific Cre expression in Csf1^fl/fl and Csf1^fl/+ mice with SM22α-Cre. As described above, efficient deletion of the floxed allele in the presence of SM22α-Cre was confirmed by PCR (Figure IIA in the Data Supplement). All mice were made hyperlipidemic by overexpressing PCSK9 and feeding an atherogenic diet for 11-12 weeks. The homozygous SMC null female mice showed an approximately 40% decrease in lesion size, whereas there was no difference between the WT and fl/+ mice (Figure 2A and B). In these experiments, the loss of SMC CSF1 expression had no effect on plasma lipids, but there was a small reduction in the percent of monocytes in the circulation (Figure 2G and Figure IV in the Data Supplement). We further performed the proliferation assay using Ki-67 labelling and EdU incorporation and observed about a 35% reduction in proliferating macrophages in Csf1^fl/fl mice with SM22α-Cre (Figure 2H, I, J and K).

Figure 2. Absence of SMC-derived CSF1 reduces lesion area.

(A-G) Female mice carrying floxed alleles of Csf1 (Csf1^fl/+ or Csf1^fl/fl) and with (+) or without (−) SM22α-Cre (the latter labeled ”controls”) were bred and examined for lesion morphology (A), lesion sizes (B), total cholesterol (C), HDL cholesterol (D), non-HDL cholesterol (E), circulating CSF1 levels (F) and monocytes % (G). (H-K) Ki-67 positive macrophages (H, I) and EdU positive macrophages in control and Csf1^fl/fl;SM22α-Cre mice (J, K) were quantitated. Control- Csf1^fl/fl and Csf1^fl/+ mice on a WT background (no Cre). Merged- CD68 (red), Ki-67 or EdU (green), and DAPI (blue).

Deficiency of CSF1 in EC reduces atherosclerosis.

CSF1 expression by EC is dramatically induced by oxidized lipids, suggesting that it may be particularly important in atherosclerosis. Mice deficient in EC CSF1 were produced as described under Methods using Csf1^fl/fl mice crossed with Cdh5-Cre transgenic mice. Using PCR quantitation, Cre expressing mice showed efficient deletion of the floxed Csf1 allele (Figure IIB in the Data Supplement). Following PCSK9 overexpression and feeding of an atherogenic diet for 11-12 weeks, both the Csf1^fl/+ and Csf1^fl/fl mice showed about a 30% reduction in lesion area (Figure 3A and B). However, total cholesterol, HDL cholesterol, non-HDL cholesterol, circulating levels of CSF1, and % monocytes were unchanged (Figure 3C, 3D, 3E, 3F, 3G and Figure V in the Data Supplement). While we did not observe significant differences in proliferation as judged by Ki-67 positive macrophages in Csf1^fl/fl mice with and without Cdh5-Cre, there was a trend that was similar to the decrease in lesion size (Figure 3H and I).

Figure 3. Absence of EC-derived CSF1 reduces lesion area.

(A-G) Female mice carrying floxed alleles of Csf1 (Csf1^fl/+ or Csf1^fl/fl) and with (+) or without (−) Cdh5-Cre (the latter labeled ”controls”) were bred and examined for lesion morphology (A), lesion sizes (B), total cholesterol (C), HDL cholesterol (D), non-HDL cholesterol (E), circulating CSF1 levels (F), and monocytes % (G). (H-I) Ki-67 positive macrophages in control and Csf1^fl/fl;Cdh5-Cre mice were quantitated. Control- Csf1^fl/fl and Csf1^fl/+ mice on a WT background (no Cre). Merged- CD68 (red), Ki-67 (green), and DAPI (blue).

Deficiency of CSF1 in monocyte/macrophages does not affect lesion development.

Monocyte/macrophages and neutrophils are capable of abundant CSF1 expression²³. As described above, homozygous floxed (Csf1^fl/fl) and heterozygous floxed (Csf1^fl/+) mice on either a LysM-Cre background or a WT background (no Cre, control) were generated. For females, we used both Csf1^fl/fl, and Csf1^fl/+ mice, while for male only Csf1^fl/fl mice were used with and without LysMCre. DNA was isolated from bone marrow derived macrophages from the mice and analyzed using PCR. This confirmed that the Csf1^fl/fl allele underwent nearly complete deletion in the presence of the LysM-Cre, as the 380 bp product, characteristic of the non-deleted allele, was largely replaced by a 619 bp fragment, characteristic of the deleted allele (Figure IIC in the data supplement). Following PCSK9 overexpression and feeding of an atherogenic diet for 11-12 weeks, lesion areas in the proximal aorta exhibited no significant decrease in the heterozygous or homozygous null mice as compared to WT mice (Figure 4A & B and Figure VIA in the Data Supplement), nor were there any differences in the levels of circulating CSF1 or percent circulating monocytes (Figure 4F & G and Figure VIE in the Data Supplement).

Figure 4. Absence of macrophage-derived CSF1 does not reduce lesion area.

(A-G) Female mice carrying floxed alleles of Csf1 (Csf1^fl/+ or Csf1^fl/fl) and with (+) or without (−) LysM-Cre (the latter labeled ”controls”) were bred and examined for lesion morphology (A), lesion sizes (B), total cholesterol (C), HDL cholesterol (D), non-HDL cholesterol (E), circulating CSF1 levels (F), and monocytes % (G). Control- Csf1^fl/fl and Csf1^fl/+ mice on a WT background (no Cre).

Csf1 expression in vascular SMC, EC, and peritoneal macrophages.

We quantified the expression of Csf1 in vascular SMC, EC, and peritoneal macrophages using qPCR. Ldlr^−/− male mice were fed either chow or the atherogenic diet. The atherogenic diet was given for only four weeks to avoid significant numbers of macrophages in aortic tissue. With chow diet, we observed the highest relative expression in SMC, with EC and peritoneal macrophages exhibiting about 30% and 2.5%, respectively, of the levels in SMC. Similarily, when mice were fed the atherosclerotic diet for 4 weeks, EC and peritoneal macrophages exhibit about 40% and 12%, respectively, of the levels in SMC (Figures VIIA and B in the Data Supplement). While comparing the relative Csf1 expression between the diets, there was no difference in any cell type (Figures VIIC, D and E in the Data Supplement). We note that the expression of Csf1 in vascular macrophages could be different from that in peritoneal macrophages.

Effects of CSF1-deficient BMCs on lesion development.

In a disturbed sleep model, it has been recently demonstrated that pre-neutrophil derived CSF1 in the bone marrow contributes to atherosclerosis³⁶. Recently, Tang et al²³ have shown that neutrophil and macrophage cell surface CSF1 are shed by the action of ADAM17 and that this drives macrophage proliferation in acute and chronic inflammation. We therefore explored the effect of CSF1-deficient BMCs on lesion development. We transplanted BMCs from male Csf1^−/− and Csf1^+/+ donors into lethally irradiated Ldlr^−/− female recipients and after 6 weeks placed them on an atherogenic diet for 11-12 weeks. Significant differences (~35%) were detected in aortic atherosclerosis (Figure 5A & B), plasma CSF1 levels (Figure 5D), and proliferating macrophages (Figure 5F & G). There were no differences in circulating monocyte counts between animals receiving BMCs from either Csf1^+/+ (N= 15) or Csf1^−/− (N= 10) genotypes (Figure 5E).

Figure 5: Transplantation of bone marrow (BM) from Csf1^−/− mice into wild type mice reduces atherosclerosis.

BM cells from male Csf1^−/− or Csf1^+/+ donor mice were transplanted into lethally irradiated Ldlr^−/− female recipients and after 6 weeks placed on an atherogenic diet for 11-12 weeks. The mice were then examined for lesion morphology (A), lesion area (B), plasma total cholesterol (C), circulating CSF1 levels (D), and lymphocyte (Lym), monocyte (Mono) and granulocyte (Gran) percents (N=15 for Csf1^+/+ and N=10 for Csf1^−/− BM) (E). The percentage of Ki-67 positive macrophages in mice transplanted with Csf1^+/+ BM and Csf1^−/− BM was quantitated (F, G). Merged- CD68 (red), Ki-67 (green), and DAPI (blue).

CSF1 deficiency has no effect on monocyte phenotypes.

To assess the effect of CSF1 on Ly6C^hi (pro-inflammatory) and Ly6C^lo monocytes (patrolling), flow cytometry analysis of blood monocytes in Csf1^+/+ and Csf1^+/− mice (n=5 per group) was performed. We did not observe a significant difference between Csf1^+/+ and Csf1^+/− mice in terms of the frequency of Ly6C^hi and Ly6C^lo monocytes (10.7±3.7 vs 14.9±6.6 and 9.2±0.6 vs 10.1±1.4, respectively) (Figure VIII in the Data Supplement).

DISCUSSION

Previous studies have shown that CSF1 deficiency dramatically reduces atherosclerosis in Apoe^−/− and Ldlr^−/− mouse models^16–20, and one study showed that a deficiency of CSF1 increased tunnel staining of lesional macrophages, indicating an increase in apoptosis²⁷. Recent studies have also made it clear that substantial proliferation of macrophages occurs in lesions²¹. CSF1 can promote macrophage proliferation in vitro^37–39 and some studies suggest that the extent of CSF1 dependent local macrophage proliferation is tissue-dependent⁴⁰. CSF1 acts by binding to its receptor (CSF1R), a type III receptor tyrosine kinase, resulting in the phosphorylation of several tyrosine residues of the CSF-1R cytoplasmic domain and PI3K/Akt signaling^41,42. Phosphorylation of tyrosine Y721 is thought to be necessary for macrophage survival⁴³. CSF1 is produced by a wide variety of cells, including endothelial cells, fibroblasts, bone marrow stromal cells, osteoblasts, astrocytes, myoblasts, keratinocytes, and mesothelial cells⁴⁴.

The goals of our study were to examine the role of CSF1 in local macrophage proliferation in atherosclerosis and to identify the cellular sources of the CSF1. We found that Csf1^+/− mice exhibited decreased proliferation as judged by Ki-67 staining and BrdU incorporation as well as increased apoptosis as judged by activated caspase-3 staining of Mac-3/CD68 positive cells. The Csf1^+/− mice exhibited normal levels of circulating monocytes and slight reductions in plasma CSF1 and cholesterol levels. The explanation for the latter is unclear since Csf1^−/− mice exhibit significantly increased LDL/VLDL cholesterol levels, possibly due to up-regulation of expression of the apolipoprotein E gene, scavenger receptors and LDL receptor-related protein in macrophages^45,46.

We also observed that EC- and SMC-specific deletion of the Csf1 gene resulted in significant reduction in lesion size and macrophage proliferation with little or no reduction in circulationg monocyte or CSF1 levels. This indicates that at least part of the effect of CSF1 on atherogenesis is at the local rather than systemic level. The reductions in lesion size were fairly modest, about 40% and 30%, respectively, as compared to the over 100-fold reduction in lesions in global Csf1^−/− mice. Presumably, the SMC and EC sources can complement one another.

Monocyte/macrophages express CSF1 at high levels, particularly when stimulated with LPS⁴⁷ and a recent report indicated that neutrophil and macrophage cell surface CSF1 shed by ADAM17 is an important driver of mouse macrophage proliferation in acute and chronic inflammation²³. However, in contrast to SMC and EC deficient mice, monocyte/macrophage CSF1 deficient mice did not exhibit a reduction in lesion size. Our quantitation of peritoneal macrophage CSF1 expression revealed levels substanyially lower than SMC or EC expression and, thus, the contribution of lesional macrophages to CSF1 expression could be quite modest. The observed small increase in lesions in the macrophage deficient mice could be the result of compensatory effects of other growth factors such as GM-CSF and G-CSF. GM-CSF has been shown to influence monocyte recruitment and intimal macrophage proliferation⁷.

To determine whether any bone marrow-derived cells other than macrophages contribute to the effects of CSF1 deficiency we also performed studies in which BMCs from Csf1^−/− mice was transplanted into wild type mice. Surprisingly, given the above findings, transplantation of Csf1^−/− BMCs into wild-type mice reduced lesion development by about 35%. This suggests that hematopoietic cells other than monocyte-macrophages contribute to the CSF1 production that drives atherosclerosis. One possibility is that CSF1 from neutrophils is important in lesion development²³, although relative to monocyte/macrophages the number of neutrophils in lesions is relatively small²³. In a sleep-deprived model, it was demonstrated that CSF1 produced by pre-neutrophils contributes to monocytosis and atherosclerosis³⁶. There is evidence that LysMCre is also expressed in neutrophils, but at least in certain studies the deletion of a floxed target was incomplete⁴⁸. Another possibility is mast cells, as these have been shown to contribute to atherosclerosis and express high levels of CSF1⁴⁷. Additionally, CSF1 is also produced by lymphocytes^49–51.

Since we did not observe any differences in circulating monocyte levels as well as in Ly6C^hi or Ly6C^lo monocytes in Csf1^+/− mice, our results suggest that circulating monocyte depletion or functional state is not a key factor underlying reduced atherosclerosis in CSF1 deficient mice. It had been widely assumed that the rate-limiting event in the accumulation of foam cells during atherosclerosis progression was the entry of monocytes. Several studies have shown a relationship between circulating monocyte levels and atherosclerosis^52,53. However, a study by Robbins and colleagues²¹ indicates that macrophage proliferation in the lesion rather than monocyte recruitment is key, at least within established lesions.

We conclude that macrophage growth in atherosclerotic lesions is dependent on the presence of locally produced CSF1, which stimulates proliferation and reduces apoptosis. Major sources of the CSF1 are SMC and EC but not macrophages. Our BM transplantation studies suggest that a non-macrophage hematopoietic cell, probably neutrophils, mast cells or lymphocytes, also contribute significantly to CSF1-dependent lesion growth. Our study has some limitations. Due to differential splicing and processing, CSF1 exists in three major isoforms: a cell surface species, a proteoglycan secreted form, and a soluble secreted form²⁵. These isoforms may have different functions in atherosclerosis as reported in certain inflammatory conditions⁵⁴. For example, the soluble form is likely to diffuse out of the lesion whereas the other forms are probably longer-lasting. Also, our studies are limited to a specific time point in lesion development. The roles of CSF1 may differ in early and late stages. For example, there is evidence that CSF1 contributes to monocyte migration²⁷ and could play a role in monocyte recruitment in the early stages.

HIGHLIGHTS:

• CSF1 deficiency reduces macrophage proliferation and increases macrophage apoptosis in lesions.

• The primary sources of CSF1 for lesion growth are SMC and EC but not monocyte/macrophages.

• Locally produced CSF1 rather than systemic CSF1 appears to be the main driver of atherosclerosis.

• Hematopoietic cells other than monocyte/macrophages contribute to atherosclerosis in a CSF1 dependent manner.

• None of the cell-specific deletions of Csf1 affect circulating CSF1 levels and only the SMC deletions reduced monocyte levels in the circulation.

ABBREVIATIONS:

ASO

Antisense oligonucleotide

cKO

Conditional knockout

EC

Endothelial cells

Csf1^fl/fl

Homozygous Csf1 loxP-flanked

LysM-Cre

Lysozyme M-Cre

M-CSF or CSF1

Macrophage colony stimulating factor

PCSK9

Proprotein convertase subtilisin/kexin type 9

SMC

Smooth muscle cells

SM22α-Cre

Smooth muscle protein 22-alpha-Cre

Cdh5-Cre

VE-Cadherin-Cre