Role of Bone Marrow–Derived Cells in the Genetic Control of Restenosis

Model of Endothelial Denudation by Wire-Mediated Arterial Injury

Specific pathogen-free male 129X1/SvJ mice and C57BL/6 mice weighing between 25 and 30 g were purchased from Charles River (Germany). All experimental procedures performed on animals met the requirements of German legislation on protection of animals and were approved by the Government of Bavaria/Germany. Surgery was carried out as described previously.¹²

Isolation of mRNA and Real-Time RT-PCR

mRNA isolation, cDNA synthesis, and PCR amplification were performed as previously described.^9,12,14

Labeling of cDNA Probes and Hybridization to cDNA Arrays

Aliquots of 25 ng of each amplified cDNA were labeled and each probe was hybridized to two arrays (Clontech Mouse 1.2 and Mouse 1.2II, Clontech Laboratories) as previously described.^9,14,15

Preparation of BMCs and BMC Transplantation

BMCs were harvested from femora and tibia as described.¹⁶ After γ-irradiation of 9.5 Gy (Cs137, Typ Ob 29/902 to 1: Buchler GmbH) of recipient mice, 1×10⁶ BMCs from 129X1/SvJ mice were suspended in 0.3 mL PBS and injected intravenously by tail vein puncture into C57BL/6 mice and vice versa, as well as from 129X1/SvJ into 129X1/SvJ mice and from C57BL/6 into C57BL/6 mice as sham controls. Ten weeks after transplantation, peripheral blood samples were collected from the retro-orbital plexus. After red blood cells (RBCs) were lysed with ammonium chloride based RBC Lysis Buffer (StemCell Technologies) white blood cells (WBCs) were washed twice with PBS containing 0.5% bovine serum albumine and 2 mmol/L EDTA. Successful transplantation was confirmed by flow cytometry to measure CD45.1 (Ly5.1, StemCell) expressed only by C57BL/6 mice and CD45.2 (Ly5.2, StemCell, USA) expressed only by 129X1/SvJ mice.

Immunohistochemistry and Immunoflourescence

Immunohistochemistry for CD45 and c-kit was performed as previously described.¹² For immunoflourescence a monoclonal antimouse CD45 (BD Bioscience) and a polyclonal antimouse c-kit (Santa Cruz) antibody were used. Counterstaining of nuclei was performed by DAPI (Sigma). Binding of the primary antibodies was detected with compatible fluorescein-conjugated secondary Alexa Fluor antibodies (Invitrogen).

Isolation and Quantification of HSPCs After Vascular Injury

Mice were anesthesized as previously described,¹² and blood was collected from the heart into 500 μL 0.5 mol/L EDTA. After lysis of RBCs and after washing of WBCs as described above, 10 μL biotin-labeled lineage antibodies (Miltenyi Biotec) was added to 1×10⁶ WBCs, resuspended in 30 μL PBS, and incubated for 10 minutes at 4°C. Blood cell populations were counted according to their FSC and SSC properties. Three-color immunofluorescence staining was performed using CXCR4 antibodies conjugated with FITC (BD Bioscience), Sca-1 antibodies conjugated wit PE (Miltenyi Biotec), and Biotin antibodies conjugated with APC (Miltenyi Biotec) according to the manufacturer’s instruction. Mononuclear cells were gated and analyzed for surface expression of lineage (lin), stem cell antigen-1 (sca-1), and CXCR4.

Statistical Analysis

Gene expression values are reported as median expression values of each group. All other values are means±SEM. Significance of differential gene expression over the indicated time courses was analyzed by Kruskas–Wallis test. A descriptive P<0.05 was regarded as significant. Significance of differential gene expression between 129X1/SvJ and C57BL/6 mice was analyzed by Mann–Whitney U test. A descriptive P<0.01 was regarded as significant. Subsequently genes with a minimum ratio of 2.5 and a minimal difference of 0.2 within their group medians were considered of differentially expressed. For real-time RT-PCR evaluation, the statistical differences in ratio were calculated using the convergence interval. For morphometric and immunofluorescence analysis, t test or nonparametric Mann—Whitney test was used to determine statistical significance between groups. Values of P<0.05 were considered statistically significant.

Vascular Remodeling After Arterial Injury Is Increased in 129X1/SvJ Mice

To clarify the question of whether or not vascular remodeling processes after wire injury of arterial vessels are genetically controlled, we applied arterial injury to the femoral artery of 129X1/SvJ mice and C57BL/6 mice.

Arterial injury of the femoral artery induced vascular remodeling leading to neointima formation and lumen narrowing in both strains. However, the extent of intimal hyperplasia was about 6 times greater in 129X1/SvJ mice compared to C57BL/6 mice 14 days after vascular injury, and about 3 times greater 28 days after injury, supporting the existence of genetically determined differences in the vasoproliferative response (Figure 1a and 1b). Furthermore, mean lumen loss was significantly less in C57BL/6 mice compared to 129X1/SvJ mice, both after 14 days and after 28 days (Figure 1a and 1c). Therefore, these data provide a reliable mouse model of arterial injury with endothelial denudation to further analyze mechanisms of genetically controlled vascular remodeling processes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. a, Representative images of femoral arteries stained with Elastica-van-Gieson at day 14 and 28 after vascular injury, scale bar=100 μm. Quantification of (b) intimal area and (c) lumen loss at day 14 (mean±SEM, n=5 for both groups at each time point, *P<0.05 between compared mouse strains).

Differences in Gene Expression Reveal Enhanced Recruitment of Inflammatory Cells and HSPCs to the Site of Arterial Injury in 129X1/SvJ Mice

To analyze the molecular mechanisms governing genetic control of vascular remodeling more systematically, we retrieved femoral arteries from uninjured and wire-injured 129X1/SvJ and C57BL/6 mice at day 3, 7, and 14 and applied differential gene expression screening. Criteria for differential expression have been applied in previous studies as described in Methods. To exclude underlying differences in the proliferative and inflammatory status of residual SMCs, we performed comparative gene expression analysis of uninjured femoral arteries of 129X1/SvJ and C57BL/6 mice showing only 41 (1.7%) of 2352 analyzed genes as differentially expressed (supplemental Figure II). However, 268 genes (11.4%) were regarded as differentially expressed between uninjured and injured arteries (supplemental Figure I). Thereof, 251 genes (10.7%) were significantly upregulated, whereas only 17 genes (0.7%) were downregulated. In 129X1/SvJ mice a total of 197 genes, and in C57BL/6 mice a total of 95 genes were significantly regulated after wire-injury. We assigned the regulated genes in both mouse lines to 6 functional subgroups, consisting of genes associated with inflammation (supplemental Figure Ia), HSPCs (supplemental Figure Ib), proliferation/apoptosis (supplemental Figure Ic), metabolism (supplemental Figure Id), cell adhesion/extracellular matrix (ECM; supplemental Figure Ie), and others (supplemental Figure If). As expected, many of these genes could be allocated to proliferation/apoptosis (n=70 in 129X1/SvJ mice; n=40 in C57BL/6 mice), metabolism (n=31 in 129X1/SvJ mice; n=7 in C57BL/6 mice), and cell adhesion/ECM (n=25 in 129X1/SvJ mice; n=7 in C57BL/6 mice; Figure 2a and supplemental Figure Ic hrough Ie). Functional clustering further revealed that more than one third of these genes were associated with inflammation (n=55 in 129X1/SvJ mice; n=35 in C57BL/6 mice) and HSPCs in both mouse lines (n=13 in 129X1/SvJ mice; n=4 in C57BL/6 mice; Figure 2a; supplemental Figure Ia and Ib). However, 129X1/SvJ mice showed a strikingly higher expression of genes that had been identified as key regulators of inflammation such as CD68 and HSPC mobilization and recruitment, such as CXCR4 or c-kit (supplemental Figure Ia and Ib).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. a, Clustering of significantly regulated genes over the time course after arterial injury (full register in supplemental Figure I) for 129X1/SvJ (white columns) and C57BL/6 mice (black columns). b, mRNA expression for CXCR4 using quantitative real-time PCR. Expression levels were normalized to GAPDH for each sample (mean±SEM, n=7, *P<0.05 vs baseline).

To further verify the gene expression data, we performed RT-PCR analysis using primers specific for CXCR4 (Figure 2b). Compared to uninjured controls, vascular injury significantly induced mRNA expression of CXCR4 after injury in 129X1/SvJ mice (P<0.05 versus baseline), whereas no changes were found in CXCR4 mRNA expression in C57BL/6 mice.

129X1/SvJ Mice Show Increased Mobilization of lin⁻sca-1⁺CXCR4⁺ HSPCs and of Granulocytes After Vascular Injury

As genetically controlled neointima formation might be not only dependent on recruitment but also on mobilization of HSPCs and inflammatory cells, we analyzed the bone marrow cell efflux of lin⁻sca-1⁺CXCR4⁺ mononuclear cells in peripheral blood 12 hours after vascular injury in our mouse model. This cell fraction is strongly predicted to contribute to vascular remodeling.^7,8 Whereas we found no difference at baseline, flow cytometry revealed that 129X1/SvJ mice respond to vascular injury with a significantly stronger mobilization of lin⁻sca-1⁺CXCR4⁺ cells (increase by 77%) 12 hours after vascular injury, whereas C57BL/6 mice showed only a slight nonsignificant mobilization (increase by 17%) (Figure 3a and 3b). Accordingly, we did not find a significant difference of granulocytes before injury, but a significant higher increase of granulocytes in 129X1/SvJ mice after vascular injury compared to C57BL/6 mice (Figure 3c).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. a, Injury-induced mobilization of lin⁻sca-1⁺CXCR4⁺ cells. Sca-1⁺ mononuclear cells were analyzed for surface expression of lin and CXCR4 (*P<0.05 vs control, n=4 for each group). Dot plots representative of 4 experiments. b, Gated lin⁻sca-1⁺CXCR4⁺ mononuclear cells were expressed as percentage of control. c, Comparison of circulating leukocytes before and 12 hours after vascular injury.

Extent of Intimal Hyperplasia Depends on the Genetic Background of BMCs

To further investigate the role of BMCs in vascular healing, we performed total BMC transplantation after irradiation from 129X1/SvJ into C57BL/6 mice and vice versa. As sham control, we transplanted BMCs from 129X1/SvJ into 129X1/SvJ mice and from C57BL/6 into C57BL/6 mice. As these 2 mouse strains feature MHC-compatibility we detected no graft-versus-host (GvH-) reaction.¹³ Successful reconstitution of transplanted bone marrow was demonstrated by a shift in the Ly-antigen expression on peripheral blood granulocytes 10 weeks after transplantation using FACS analysis (Figure 4a). The surface marker Ly5.1 was expressed on C57BL/6 granulocytes, whereas Ly5.2 was expressed on 129X1/SvJ granulocytes. According to each donor BMCs we correspondingly found exclusive Ly5.1 expression in mice with C57BL/6 total BMCs and exclusive Ly5.2 expression in mice with 129X1/SvJ total BMCs on granulocytes in peripheral blood samples 10 weeks after transplantation (Figure 4a).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4. a, Verification of BMC transplantation analyzing Ly5.1- and Ly5.2-antigen on granulocytes. b and c, Morphometric analysis of total BMC-transplanted mice. b, Representative images of injured femoral arteries stained with Elastica–van-Gieson, scale bar=100 μm. c, Quantification of intimal area, lumen loss, and adventitial area (mean±SEM, n=4 for each group, *P<0.05).

After complete reconstitution, we induced vascular injury of the femoral arteries in the transplanted mice and analyzed vascular remodeling 14 days after arterial injury. Morphometric analysis revealed that the extent of neointima formation was strictly associated with the genetic background of the BMCs. C57BL/6 mice as recipients of 129X1/SvJ total BMCs (C57BL/6 recipients) developed nearly as much neointima as 129X1/SvJ sham controls, whereas 129X1/SvJ mice as recipients of C57BL/6 total BMCs (129X1/SvJ recipients) developed even less neointima than C57BL/6 sham controls (Figure 4b and 4c). Analysis of adventitial proliferation showed that the extent of adventitial proliferation was also strictly associated with the genetic background of the BMCs (Figure 4b and 4c). Moreover, mean lumen loss was comparable between C57BL/6 recipients and 129X1/SvJ sham controls and significantly less in 129X1/SvJ recipients and in C57BL/6 sham controls (Figure 4b and 4c). To confirm that vascular injury after bone marrow transplantation in our model comes along with recruitment of specific BMCs to the site of vascular injury, we performed immunoflourescence analysis of CD45.2 (Ly5.2) expressed only by mice with 129X1/SvJ genetic background. Indeed, Ly5.2 positive cells are only recruited in C57BL/6 recipients and 129X1/SvJ sham controls representing mice with total 129X1/SvJ BMCs (supplemental Figure III).

Differences in the Cellular Composition of Intimal Tissue in Engrafted Mice

As neointima formation is strongly related to the degree of vascular inflammation,⁵ we investigated the inflammatory response to arterial injury depending on the particular BMCs. Immunofluorescence with an antibody against the pan-leukocyte marker CD45 revealed that 129X1/SvJ sham controls as well as C57BL/6 recipients recruited significantly more leukocytes as percentage of total cell number to the site of vascular injury than C57BL/6 sham controls and 129X1/SvJ recipients (Figure 5a and 5b). Moreover, adventitial cell density of CD45⁺ leukocytes was higher in mice with total 129X1/SvJ BMCs (supplemental Figure IV). Uninjured arteries of all groups showed no inflammatory activity at baseline (Figure 5a and 5b).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5. a through d, Immunofluorescence-staining with antibodies against CD45 for leukocytes (red) and c-kit for BMCs (green). Nuclei were stained with DAPI (blue). a and c, Representative sections of respective immunoflourescence stainings, scale bar=100 μm. b and d, Quantitative analysis of leukocytes and BMCs as percentage of total cell number (mean±SEM, n=8 for each group, *P<0.05).

In addition to inflammation, we followed recruitment of BMCs to the site of vascular injury by immunofluorescence detection of c-kit. 129X1/SvJ sham controls recruited significantly more BMCs as percentage of total cell number to the site of vascular injury than C57BL/6 sham controls or 129X1/SvJ recipients. On the contrary, C57BL/6 recipients tended to recruit more BMCs to the site of vascular injury than mice with C57BL/6 bone marrow (Figure 5c and 5d). Moreover, adventitial cell density of c-kit⁺ BMCs was higher in mice with total 129X1/SvJ BMCs (supplemental Figure IV). Uninjured arteries of all groups showed no c-kit⁺ cells (Figure 5c and 5d).

129X1/SvJ and C57BL/6 Mice Display Pronounced Differences in Their Reaction to Vascular Injury

Transgenic or gene-targeted mice can be engineered on different genetic backgrounds leading to phenotype differences.^17,18 Likewise, in our animal model of vascular remodeling after arterial injury, we observed considerable differences in neointima formation, adventitial proliferation, and lumen narrowing between mice with different genetic backgrounds. The rationale for the use of these 2 mouse strains on the one hand was that they significantly differ in vascular remodeling after carotid ligation¹¹ and on the other hand feature MHC-compatibility.¹³ Whereas 129X1/SvJ mice responded to endothelial denudation and vessel injury with extensive neointima formation, C57BL/6 mice generated significantly less intimal hyperplasia (Figure 1). Similarly, mouse strains with different genetic backgrounds exhibited differences in vascular remodeling after carotid ligation.¹¹ However, unlike the situation after PCI in patients, in this model, flow in the carotid artery is interrupted by ligation of the vessel resulting in a dramatic reduction in vessel diameter and neointima formation. Using wire injury of the femoral artery, we chose an experimental setting that resembles the situation after PCI in patients as it induces endothelial denudation and vessel dilatation with injury of the media. This is of special interest as it has been shown that the cellular composition of a lesion differs depending on the type of vascular injury.⁷ Hypothesizing that BMCs represent major players in the genetically determined vascular remodeling process, this mouse model comes along with the recruitment of a significant number of BMCs compared to lesion formation by ligation or perivascular cuff replacement.⁷ Using mice of same age, sex, and weight in both mouse strains we were able to minimize confounders and thereby emphasize the different genetic backgrounds as major checkpoint of different responses to vascular injury.

The Mechanisms Leading to Lumen Narrowing in the Vascular Remodeling Process After Arterial Injury and Endothelial Denudation Are Genetically Controlled

To gain a systematic insight into molecular mechanisms driving genetic control of vascular remodeling, we compared those genes which were differentially regulated over the time course between the two mouse strains. Comparative gene expression analysis of uninjured femoral arteries of 129X1/SvJ and C57BL/6 mice revealed only 41 (1.7%) of 2352 analyzed genes as differentially expressed (supplemental Figure II). Although all differentially expressed genes showed higher levels in 129X1/SvJ mice, the marginal number of differentially regulated genes suggests no major difference in inflammatory and proliferative properties in SMCs between the two mouse strains. However we cannot rule out that the basic inflammatory and proliferative degree in SMCs may differ to a small degree in 129X1/SvJ mice compared to C57BL/6 mice.

On the contrary, during the time course after wire injury, 268 genes were differentially regulated, thereof 251 genes were upregulated and 17 genes were downregulated in both mouse strains (supplemental Figure I).

We clustered these genes into functional groups according to the information existing in the literature. In both mouse strains, we found genes known to be regulated during neointima formation such as MMP-9, VEGF, and desmin.¹⁵ Whereas the involvement of proliferation and apoptosis of SMCs and the production of extracellular matrix are well known during neointima formation the impact of inflammatory processes leading to leukocyte recruitment as well as the impact of HSPC recruitment after vascular injury have been underlined more recently.^5–7 The proinflammatory gene expression pattern predominantly expressed in 129X1/SvJ-mice after arterial injury gives an impressive rationale for profound leukocyte recruitment after wire injury. Interestingly, many of these cytokines and chemokines act on cells with myeloid origin. Likewise, the activation and induced migration of monocytes, leukocytes, and T cells have been reported for small inducible cytokine A2,¹⁹ colony stimulating factor 1,²⁰ and small inducible cytokine B subfamily (Cys-X-Cys), member 9.²¹ Other cytokines such as colony-stimulating factor 3 and its receptor were also upregulated after vascular injury in 129X1/SvJ-mice and enhance the migration of granulocytes and HSPCs.²⁰ This proinflammatory gene cluster was reflected in a significantly increased inflammatory response to arterial injury in mice with total 129X1/SvJ BMCs as shown by staining against the leukocyte marker CD45 (Figure 4a and 4b). Therefore, the mechanically induced gene expression pattern favoring leukocyte recruitment in 129X1/SvJ mice may trigger increased lumen narrowing, as inflammatory cell density in neointima correlates closely with the extent of lumen loss.⁵

Moreover, we found a strong upregulation of genes related to HSPCs in 129X1/SvJ mice, associated with an increased peripheral mobilization of HSPCs. Enhanced release of HSPCs from the bone marrow has been shown to be caused by MMP-9,²² a matrix metalloproteinase known to be activated by Gro-β.²³ We found upregulation of MMP-9 as well as of the HSPC markers CXCR4²⁴ and c-kit²⁵ after arterial injury only in 129X1/SvJ mice. Accordingly, we found higher mobilization and recruitment of HSPCs/BMCs and inflammatory cells to the site of vascular injury in 129X1/SvJ mice as shown by FACS analysis of whole blood samples (Figure 3) and by immunoflourescence (Figure 5). HSPCs were defined as lin⁻sca-1⁺CXCR4⁺ cells for FACS analysis. Recently, the chemokine SDF-1α and its receptor CXCR4 were shown to participate in vascular remodeling through governing the defined recruitment of HSPCs to the site of injury.⁸ Strikingly, mobilized HSPCs and leukocytes expressed significantly higher levels of the chemokine receptor CXCR4 in 129X1/SvJ mice compared to C57BL/6 mice, indicating that differences in the responsiveness and mobilization of BMCs are involved in the genetic control of vascular remodeling.

Differences in Neointima Formation Between 129X1/SvJ and C57BL/6 Mice Are Dependent on BMCs

To further test the hypothesis that the genetically determined differences in vascular remodeling and neointima formation between 129X1/SvJ and C57BL/6 mice can be traced back to differences in BMC mobilization or recruitment, we performed total BMC transplantation after irradiation from 129X1/SvJ into C57BL/6 mice and vice versa, as well as from 129X1/SvJ into 129X1/SvJ mice and from C57BL/6 into C57BL/6 mice as sham controls and induced vascular injury. In this setting, we had to take the issue of heterologous bone marrow transplantation between different mouse strains into account. We therefore used 129X1/SvJ and C57BL/6 mice, which feature MHC-compatibility eliminating the need for immunosuppression.¹³ Interestingly, we found an increase of intimal hyperplasia and lumen loss in all BMC-engrafted mice. As this increase was detectable in sham controls as well, we ascribed this effect to stronger responsiveness of freshly reconstituted BMCs and to higher stress levels in engrafted mice, rather than to GvH reactions. However, we cannot rule out that subclinical GvH-reactions slightly affected vascular remodeling in our experiments.

Regardless of this effect, the extent of intimal hyperplasia and lumen narrowing after wire injury was strictly associated with the particular donor of the BMCs, as mice engrafted with BMCs from 129X1/SvJ mice displayed the high-risk phenotype with extensive recruitment of inflammatory cells as well as BMCs resulting in profound neointima formation, whereas BMCs from C57BL/6 mice corresponded to the low risk phenotype with moderate neointima formation (Figure 4 and 5⇑).

The role of BMCs in neointima formation after vascular injury has widely been described. BMCs give rise to substantial numbers of neointimal ECs and SMCs after endothelial denudation with controversial effects on intimal hyperplasia.^16,26,27 However, Sahara et al showed in wire-injury experiments with c-kit⁺sca-1⁺lin⁻ HSPC-engrafted mice that medial and neointimal SMCs comprised up to 60% of HSPCs, whereas no ECs originated from this cell fraction.²⁸ Likewise, in our model the enhanced recruitment of leukocytes and c-kit⁺ BMCs in mice with 129X1/SvJ total BMCs came along with significantly impaired reendothelialization capacity compared to mice with C57BL/6 total BMCs (data not shown). In congruency with our data, admission of CXCR4 antagonist AMD3100 attenuated neointima formation after vascular injury but had no significant effect on reendothelialization.²⁹ These data support the notion that reduced recruitment of BMCs in our model is an important factor in reduction of intimal hyperplasia, but has no adverse effect on reendothelialization.

Limitations of the Study

Although we report here for the first time a systematic analysis of genetically controlled vascular remodeling leading to lumen narrowing, we cannot exclude that there may be additional mechanisms underlying the genetically determined extent of neointima formation. Investigating differences in protein and gene expression in BMCs from individuals at high or low risk for restenosis might represent an excellent tool to develop new drugs and new biological markers to identify patients at high risk for restenosis.