|Home | About | Journals | Submit | Contact Us | Français|
Bone marrow-derived cells (BMCs) and inflammatory chemokine receptors regulate arteriogenesis and angiogenesis. Here, we tested whether arteriolar remodeling in response to an inflammatory stimulus is dependent on BMC-specific chemokine (C-C motif) receptor 2 (CCR2) expression and whether this response involves BMC transdifferentiation into smooth muscle.
Dorsal skinfold window chambers were implanted into C57Bl/6 wild-type (WT) mice, as well as the following bone marrow chimeras (donor-host): WT-WT, CCR2−/−-WT, WT-CCR2−/−, and EGFP+-WT. One day after implantation, tissue MCP-1 levels rose from “undetectable” to 463pg/mg, and the number of EGFP+ cells increased more than 4-fold, indicating marked inflammation. A 66% (28μm) increase in maximum arteriolar diameter was observed over 7 days in WT-WT mice. This arteriolar remodeling response was completely abolished in CCR2−/−-WT mice but largely rescued in WT-CCR2−/− mice. EGFP+ BMCs were numerous throughout the tissue, but we found no evidence that EGFP+ BMCs transdifferentiate into smooth muscle, based on examination of >800 arterioles and venules.
BMC-specific CCR2 expression is required for injury/inflammation-associated arteriolar remodeling, but this response is not characterized by the differentiation of BMCs into smooth muscle.
Arteriogenesis, which refers to the lumenal growth of the arterioles into larger arteries, has been primarily investigated using models of arterial insufficiency for skeletal muscle,1–5 brain,6,7 and heart.8 In these models, the contribution of bone marrow-derived cells (BMCs) has been intensely studied and is now coming into focus. Indeed, although ex-vivo enriched BMC populations are capable of transdifferentiating into endothelial and/or smooth muscle cells in the heart9,10 and skeletal muscle,11 there is also evidence that endogenous BMCs do not transdifferentiate into smooth muscle.3,12 Instead, it appears endogenous BMCs are recruited to tissue and become paracrine growth factor sources. Monocytes, in particular, appear necessary for arteriogenesis.13
BMC recruitment is regulated by proinflammatory chemokines, including monocyte chemotactic protein-1 (MCP-1). Perfusion restoration is abrogated in MCP-1 deficient mice,14,15 and the application of MCP-1 protein16,17 or plasmid DNA18 amplifies arteriogenesis. The primary receptor for MCP-1 is chemokine (C-C motif) receptor 2 (CCR2). CCR2−/− mice have normal numbers of circulating monocytes and tissue resident macrophages; however they show deficient monocyte recruitment to sites of inflammation.19,20 Interestingly, CCR2−/− mice on a C57Bl/6 background exhibit normal blood flow restoration and arteriogenesis in the ischemic hindlimb.4,5
The role of BMCs in arteriogenesis has now been studied in some detail, typically using models of arterial occlusion and ischemia1–8. However, few if any studies have examined the role of BMCs in the remodeling of arterioles in response to tissue injury or chronic inflammation, despite the critical importance of such processes in wound repair and human diseases. In addition, it is unknown whether BMCs transdifferentiate into smooth muscle/pericyte lineages in this response. Here, we tested the hypothesis that BMC-specific CCR2 expression is required for injury/inflammation induced arteriolar remodeling, but that BMCs do not transdifferentiate into smooth muscle. To this end, we first determined that dorsal skinfold window chamber implantation elicits inflammation and arteriolar remodeling by smooth muscle proliferation. Next, using EGFP+-WT (donor-host) bone marrow chimeric mice and smooth muscle α-actin staining, we determined that BMCs do not transdifferentiate into smooth muscle in this model. Finally, we determined that arteriolar remodeling was abolished in CCR2−/−-WT bone marrow chimeras and largely rescued in WT-CCR2−/− mice, indicating that this response was dependent on BMC-specific CCR2 expression.
All animal studies were approved by the Animal Research Committee at the University of Virginia and conformed to the AHA Guidelines for the Use of Animals in Research. EGFP+ [C57Bl/6-Tg(ACTB-EGFP)1Osb/J], CCR2−/− (B6.129S4-Ccr2tm1lfc/J), and C57Bl/6J mice were from Jackson Laboratory (Bar Harbor, ME). Host bone marrow was abrogated via 2 doses of irradiation (550rad), mice were reconstituted with donor marrow, and 8 weeks were allowed for engraftment. Mice were kept on sulfa drugs to boost immunity. The following bone marrow chimeras (donor-host) were created: WT-WT, CCR2−/−-WT, WT-CCR2−/−, and EGFP+-WT. “n” values for each experiment are provided in the figure legends.
Dorsal skinfold window chamber implantations were performed on WT, WT-WT, CCR2−/−-WT, WT-CCR2−/−, and EGFP+-WT mice as described in the Online Supplement at http://atvb.ahajournals.org. Six days after implantation, mice were anesthetized using inhaled 2.5% isoflurane anesthesia, the coverglass was removed, and the window was superfused with10−4M adenosine in sterile Ringer’s solution to remove vessel tone. Thus, any changes in arteriolar diameter were due to changes in vessel structure. An ~16 mm2 region of interest (ROI) containing at least one large arteriole-venule pair and smaller vessels was imaged, and arteriolar diameters were measured. Measurements were repeated at Day 13.
Window chambers were implanted on anesthetized (2.5% isoflurane) EGFP+-WT mice, and the coverglass was removed. An ~0.2 mm2 region of interest containing a bifurcating arteriole-venule pair was imaged immediately (0 hours), 1 hour, 2 hours, 4 hours, 6 hours, 8 hours, and 12 hours after surgery. Offline, the number of EGFP+ cells per unit area of tissue was determined.
MCP-1 protein levels in window chamber tissues from WT mice were measured by ELISA at Days 0, 1, 6, and 13 as described in the Online Supplement.
On Day 13, window chamber tissues from WT-WT, CCR2−/−-WT, and WT-CCR2−/− mice were harvested for F4/80 immunolabeling. Images were taken across complete cross-sections to quantify the area of F4/80+ cell coverage. F4/80+ area was then normalized to both tissue area, which could change with cell infiltration and swelling, and tissue length, which was held constant by the window chamber frame. To examine the potential transdifferentiation of BMCs into smooth muscle, EGFP+-WT and WT window chambers were harvested for SM α-actin immunolabeling at timepoints chosen to correspond with the intravital measurements (Days 0, 1, 6, and 13). Detailed immunostaining methods are provided in the Online Supplement.
Data were analyzed by either One- (Figures 1, ,2,2, ,3,3, 4C, and 4D) or Two-way (Figure 4B) ANOVA followed by pairwise comparisons with either Tukey’s (Figures 1, ,2,2, 3B, 3C, ,4)4) or Holm-Sidak (Figures 3B and 3C) t-tests. Significance was assessed at P<−0.05.
We first tested whether window chamber implantation causes inflammation. An ELISA for MCP-1 was performed at Days 0, 1, 6, and 13 (Figure 1A). MCP-1 expression was negligible in uninstrumented control tissue (n=4). Immediately after surgery, 9.4 pg/mg protein was present, increasing to 463 pg/mg protein at 24 hours and remaining elevated at Days 6 and 13. Window chambers were also implanted on EGFP+-WT chimeric mice, and tissue regions were imaged over 12 hours. Immediately after surgery, the venules contained rolling and adherent EGFP+ BMCs (Figure 1C). Sparse EGFP+ BMCs were resident in the tissue or had recently extravasated. Thereafter, EGFP+ BMC density increased rapidly with time, increasing 4-fold by 12 hours (Figures 1B and 1D).
We next tested whether structural remodeling of the arterioles occurs in this model. Figure 2A shows a vasodilated region of window chamber tissue from a WT mouse in which an arteriole of interest (arrows) underwent lumenal diameter expansion. Arteriolar diameters and diameter changes are reported in Figures 2B, 2C, and 2D. Cross-sectioned tissues were labeled for SM α-actin and nuclei with DAPI (Figure 2E) and used to determine wall area:lumen area ratio (Figure 2F) and the number of SM α-actin+ cells/wall area (Figure 2G). Wall area:lumen area ratio increased significantly from Days 1 to 13 (Figure 2F). Meanwhile, the number of smooth muscle cells per unit wall area remained constant (Figure 2G). Thus, the medial wall grew through the proportional proliferation of smooth muscle.
To then determine whether the inflammatory response included the BMC-specific CCR2-dependent recruitment of monocytes/macrophages, F4/80 staining was done on tissues from WT-WT, CCR2−/−-WT, and WT-CCR2−/− mice (Figure 3A). When compared to the WT-CCR2−/− and WT-WT groups, CCR2−/−-WT cross-sections showed an ~10-fold decrease in the percentage of tissue area covered by F4/80+ cells (Figure 3B) and an ~15-fold decrease in F4/80+ cell area per unit length of tissue (Figure 3C).
Next, we tested whether BMC-specific CCR2 expression was required for arteriolar remodeling. Arteriolar diameters within pre-selected regions of interest in WT-WT, CCR2−/−-WT, and WT-CCR2−/− mice were measured with vasodilation (Figure 4A). At Day 6, arteriolar diameters in WT-WT, CCR2−/−-WT, and WT-CCR2−/− mice were statistically similar. Between Days 6 and 13, arteriolar diameters were significantly increased in animals with WT BMCs (WT-WT and WT-CCR2−/−) but not CCR2−/−-WT mice. Similarly, implantation of the window chamber elicited 66% (28μm) and 43% (13μm) increases in, respectively, WT-WT and WT-CCR2−/− mean arteriolar diameters, whereas CCR2−/−-WT mice underwent essentially no change (0.5%, −2μm) (Figure 4C and 4D).
Finally, we determined whether this remodeling response included the transdifferentiation of BMCs into smooth muscle. Approximately 800 arterioles and venules were observed in sections from EGFP+-WT mice (36 sections × ~22 SM α-actin+ vessels per section). Representative confocal images showing the interstitial positioning of EGFP+ cells with respect to SM α-actin+ arterioles and venules are presented in Figure 5. In many cases, intralumenal EGFP+ BMCs with a rounded morphology indicative of leukocytes were seen. However, we never observed the colocalization of EGFP with SM α-actin, which would have indicated the transdifferentiation of BMCs into smooth muscle.
We report three major new findings. First, based on measurements of arteriolar diameters under conditions of maximal vasodilation, we showed that window chamber implantation elicits arteriolar remodeling as characterized by smooth muscle proliferation and proportional medial growth. Tissue MCP-1 levels and EGFP+ BMC recruitment increased immediately after surgery, indicating that arteriolar remodeling was associated with substantial injury/inflammation. Second, in this model, both monocyte/macrophage recruitment and arteriolar remodeling were abolished in CCR2−/−-WT mice, but largely rescued in WT-CCR2−/− mice. Third, a comprehensive examination of ~800 SM α-actin+ vessels from EGFP+-WT chimeric mice showed no co-localization of EGFP with SM α-actin. Overall, we conclude that BMC-specific CCR2 expression is required for arteriolar remodeling in an injury/inflammation model, but that this response does not involve the transdifferentiation of BMCs into smooth muscle.
In this study, the dorsal skinfold window chamber was used to investigate whether BMCs transdifferentiate into smooth muscle and/or whether CCR2+ BMCs, which are not required for arteriogenesis in the ischemic hindlimb of C57Bl/6J mice,4,5 are required for arteriolar remodeling. One advantage of this model was that is allowed us to track and observe individual microvessels over time. Thus, the diameter change results presented in Figures 2C, 2D, 4C, and 4D are repeated measurements and are particularly robust. As shown by the MCP-1 ELISA (Figure 1A) and EGFP+ BMC data (Figures 1B, 1C, and 1D), surgical implantation of the window chamber caused tissue injury/inflammation, and we hypothesize that this was a critical factor in eliciting arteriolar remodeling. Indeed, this hypothesis is supported by Figure 4, which shows that the BMC-specific deletion of CCR2, a molecule that is known to be critical for monocyte recruitment during inflammation, abolishes arteriolar remodeling. It is, however, likely that window chamber implantation also altered other factors that are hypothesized to influence arteriolar remodeling, including local microvascular hemodynamics21 and tissue oxygenation.17 For this reason, we have chosen to denote the dorsal skinfold window chamber as a model of “inflammation-associated” arteriolar remodeling, but not necessarily “inflammation-induced” arteriolar remodeling.
The window chamber also allowed us to observe the trafficking of EGFP+ BMCs from the circulation into subcutaneous tissue. Although the degree of BMC reconstitution was not tested in EGFP+-WT chimeras, our bone marrow transplantation protocol was identical to that used by O’Neill et al.,22 wherein 80–90% of circulating CD11b+ and CD45+ cells expressed EGFP. Immediately after window chamber implantation, EGFP+ BMCs were rolling and firmly adhered to venules (Figure 1C), consistent with the hypothesis that BMC recruitment occurred by canonical inflammatory cell trafficking. Windows were then imaged periodically over 12 hours, with more EGFP+ cells appearing in the window at each timepoint (Figures 1B and 1D). After 12 hours, individual EGFP+ cells could not be distinguished, so they were no longer counted. Consistent with our results, wounding the dorsal skin of EGFP+-WT chimeric mice elicits an increase in EGFP+ cells in the tissue over time.23
To examine the potential role of EGFP+ BMCs as smooth muscle cell precursors, we looked carefully for the colocalization of EGFP with the contractile protein SM α-actin. Because SM α-actin is expressed by both mature and immature smooth muscle,24 as well as by myofibroblasts25 and pericytes,26,27 the choice of SM α-actin afforded us perhaps the highest possible probability of observing BMC transdifferentiation into a smooth muscle or even “smooth muscle-like” lineage. In addition, as shown in Figures 1 and and5,5, the overall magnitude of BMC recruitment in this model is tremendous, and unlike skeletal muscle, wherein the physical space in which BMCs may reside is significantly limited by the presence of myofibers, BMCs in the window chamber are essentially free to move through both the subcutaneous fascia and forming granulation tissue. Nonetheless, even though we used a selective and highly sensitive SMC/pericyte marker gene and an experimental model system with conditions that would seem optimal for facilitating BMC transdifferentiation, no colocalization of EGFP with SM α-actin was ever observed.
One potential concern is that EGFP expression could have affected the ability of BMCs to transdifferentiate into smooth muscle. Unfortunately, true negative control experiments are not possible because, by definition, they would have to utilize BMCs lacking EGFP or any other marker. In turn, the absence of a marker would obviate our ability to do the experiment. Thus, the potential influence of EGFP cannot be completely ruled out. However, we note that other investigators have shown that EGFP+ BMCs are indeed capable of transdifferentiating into smooth muscle.9,28,29 Thus, we contend that it is highly unlikely that the EGFP+ BMCs in our study lost their potential for transdifferentiation.
Our results support other studies showing that endogenous BMCs do not transdifferentiate into smooth muscle during arteriogenesis. Ziegelhoeffer et al.3 used an exhaustive microscopy approach to come to this conclusion, while we observed no BMC transdifferentiation with arteriogenesis in response to ultrasonic microbubble destruction.12 Although not using an arteriogenesis model, Bentzon et al.30 also presented definitive evidence that BMCs do not become smooth muscle in artery walls. It has been shown that smooth muscle progenitor cells are present in bone marrow31,32 and that bone marrow-derived cells can transdifferentiate into smooth muscle.28,29 However, these studies used ex-vivo enriched BMCs that were re-injected into the animal after processing. In our experiments, BMCs were not enriched ex vivo, nor were they supplemented with additional cell injections, thus we argue that our experiments appropriately represent the role of endogenous BMCs in arteriolar remodeling in response to injury/inflammation.
In the context of hindlimb ischemia, the role MCP-1 in regulating monocyte recruitment and arteriogenesis has been well-documented;14–18,33 however, on the C57Bl/6J background, deletion of the primary MCP-1 receptor (CCR2) has no effect on perfusion recovery and arteriogenesis.4,5 In contrast, we demonstrated that arteriolar remodeling is abolished in response to injury/inflammation in CCR2−/−-WT mice (Figure 4) on the same C57Bl/6J genetic background. There are several potential explanations for why such differences were obsevred. For example, even though Figures 2F and 2G show that arteriolar remodeling in the window chamber is accompanied by medial thickening and proportional smooth muscle cell proliferation, which are characteristics of classical arteriogenesis in response to arterial ligation, there are still differences between the two growth responses. These differences, which include the tissue of interest, the initial vessel dimensions, and the magnitude of the growth response, could influence whether BMC-specific CCR2 expression is required for arteriolar growth. In addition, MCP-1 levels and monocyte/macrophage recruitment differ markedly between the hindlimb ischemia and window chamber models. Whereas there are 463 pg/mg of MCP-1 in the window chamber at Day 1 (Figure 1), peak MCP-1 levels are about an order of magnitude lower (10–40 pg/mg) in the collateral zone of the ischemic hindlimb.5 Moreover, we found that BMC-specific CCR2 deletion caused a 10- to 15-fold reduction in F4/80+ macrophage density in the window chamber (Figure 3). In contrast, using MOMA2 as a macrophage marker, no differences in monocyte/macrophage recruitment to the collateral zone after arterial occlusion were observed between CCR2−/− knockouts and WT mice.4
Finally, we note that hindlimb studies by other investigators were performed on complete CCR2−/− knockouts,4,5 while CCR2 deletion in our studies was BMC-specific. CCR2 is also expressed by endothelial cells34,35 and smooth muscle,36 therefore, CCR2-dependent signaling in endothelial and/or smooth muscle cells remained intact in our CCR2−/−-WT mice. While this could explain the differences between our results and those observed in ischemic hindlimb4,5 it would imply that endothelial and/or smooth muscle CCR2 expression is actually “anti-growth”. This is unlikely because MCP-1 signaling to endothelial cells34,35 and smooth muscle36 through CCR2 elicits enhanced cell migration, proliferation, and activation, all of which are essentially “pro-growth” behaviors. Furthermore, the results from the WT-CCR2−/− group (Figure 4) support the hypothesis that endothelial and/or smooth muscle CCR2 expression enhances arteriolar remodeling. Although arteriolar remodeling was completely blocked in the CCR2−/−-WT group, it was not completely rescued in the WT-CCR2−/− group, even though F4/80+ cell density was completely restored to WT-WT levels (Figure 3). This raises the possibility that, in WT-WT mice, CCR2 chemokine ligands [i.e. MCP-1 (CCL2), CCL8, CCL7, CCL13, and/or CCL16]37 are released from recruited BMCs, bind to CCR2 expressed on endothelial and/or smooth muscle cells, and amplify arteriolar remodeling. In such a scenario, the recruitment of BMCs to tissue through CCR2 would be absolutely required to initiate growth, while the endothelial and/or smooth muscle cell expression of CCR2 would be required to complete the response.
Sources of Funding
Supported by NIH HL074082 (R.J.P) and NIH HL065958 (T.C.S.).