Rationale: Endothelial Nitric Oxide (NO) synthase (eNOS)-produced NO plays a crucial role in maintaining vascular homeostasis. Bone marrow-derived circulating Endothelial Progenitor Cells (EPCs) recently emerged as an important marker of vascular health and contribute to endothelial repair after arterial injury.

Objective: We sought to identify if EPCs could participate in the vaso-protective effects of eNOS in vivo.

Methods and Results: To modulate eNOS activity in vivo, we used eNOS mutant knockin (ki) mice having the serine-1176 replaced by an aspartate (phosphomimetic mutation, Dki) or by an alanine (unphosphorylatable mutation, Aki), and eNOS Wild Type (WT) and Knockout (KO) mice. Upregulating eNOS activity by Dki mutation dramatically increased basal levels of plasma NOx and circulating EPCs, whereas Aki mutation or KO of eNOS significantly inhibited increases of circulating EPCs after arterial injury. Mechanistically, the increased basal EPCs in Dki mice were not due to increase of bone marrow EPC mobilization pathway MMP-9/sKitL, but might result from attenuated apoptosis and increased differentiation of circulating EPCs. Tracing of circulating EPCs via EGFP bone marrow transplantation demonstrated that the number of incorporated BM-derived EPCs in regenerated endothelium positively correlated with eNOS activity. Consistently, reendothelialization and vasodilation were regulated by eNOS activity as well.

Conclusions: Our data show that BM-derived circulating EPCs are closely involved in eNOS-regulated endothelial repair and vascular function after arterial injury. This finding enriches our understandings of the important role of eNOS in vascular integrity and further supports therapies targeting eNOS to prevent cardiovascular disease.

Keywords: eNOS; Endothelial progenitor cells; Arterial injury

Maintaining the integrity of endothelium after acute or chronic injury is essential for preventing endothelial dysfunction, which is closely associated with various metabolic and cardiovascular diseases [1,2]. Nitric Oxide (NO) produced by endothelial NO synthase (eNOS) plays important roles in protecting vascular integrity. Many of these roles are local effects of NO, including modulation of proliferation and migration of Endothelial Cells (ECs), growth of smooth muscle cells, platelet aggregation, and leukocyte adhesion [3]. Reduced endothelial NO bioavailability is a common feature of many cardiovascular risks.

Besides adjoining mature ECs, bone marrow-derived circulating Endothelial Progenitor Cells (EPCs) have recently been implicated in replacing injured or apoptotic endothelial cells, therefore preserving vascular function and limiting the formation of atherosclerotic lesions [4,5]. Decreased levels of circulating EPCs inversely correlate with vascular reactivity and independently predict future cardiovascular outcomes [5,6]. Circulating EPC levels can be regulated by a variety of factors, including exercise, statins, estrogen, various chemokines and cytokines upon ischemia or injury [6]. Recently, mobilization of bone marrow EPCs was reported to be impaired by eNOS deletion through blunting the activation of matrix metalloproteinase 9 (MMP-9) and soluble Kit Ligand (sKitL)-mediated pathway in bone marrow [7,8]. Moreover, eNOS has been suggested to be involved in VEGF, [8] estrogen, [9] statin, [10] exercise, [11] hyperoxia, [12] erythropoietin, [13] secretoneurin, [14] carbon monoxide (CO) [15]-mediated EPC mobilization and subsequent reendothelialization. Besides, direct administration of eNOS transcription enhancer such as AVE9488 [16] or substrate L-arginine [17] upregulated circulating EPCs. Furthermore, the plasma concentration of asymmetric dimethylarginine (ADMA), an endogenous inhibitor of eNOS, correlated inversely with the number of blood EPCs in patients with coronary artery disease [18]. Nevertheless, these studies all utilized indirect or pharmacological approaches to probe the role of eNOS. It remains unknown whether directly manipulating eNOS expression or activity alone in vivo is sufficient or not to regulate EPCs mobilization and their contribution to endothelial repair.

Phosphorylation of eNOS at serine 1176 (S1176) is a major regulatory mechanism of its enzymatic activity, enhancing NO production both at resting calcium concentration and under agonist challenges such as shear stress, VEGF, estrogen, insulin and statins [2]. To regulate eNOS activity in vivo, we made use of knockin (ki) mice that carry single amino acid mutations at the S1176 site of eNOS [19-21]. The S1176A mutation replaces the S1176 with an alanine (Aki), making the side chain unphosphorylatable. The S1176D mutation replaces the S1176 with an aspartate (Dki), which mimics the negatively charged phosphate group, rendering the mutant phosphomimetic. By comparing the responses of these mutant eNOS knockin mice with WT and eNOS KO mice, we created an in vivo activity gradient of eNOS and tested how modulating eNOS activity affects EPC mobilization, reendothelialization and subsequent endothelial function after arterial injury.

An expanded Methods section is available in the online supplemental material.

eNOS Dki and Aki mice were generated and genotyped as previously described [20]. Both mutant mice were backcrossed to the C57BL/6J genetic background by marker assisted congenic backcrossing. eNOS KO mice were also on C57BL/6J background [22]. Age-matched (8-12 weeks) C57BL/6J mice (Jackson Laboratory, Bar Harbor, ME) were used as WT controls. All animal experiments were approved by the Massachusetts General Hospital Institutional Animal Care and Use Committee. Euthanasia of mice was performed according to the recommendations of the American Veterinary Medical Association Panel on Euthanasia.

WT, Dki, Aki and eNOS KO mice (9-14 weeks) were used in this experiment. Carotid artery filament injury and assessment of endothelial denudation by Evans Blue staining were performed as previously described with minor modifications [23].

Vascular reactivities of isolated femoral arteries to acetylcholine (ACh) or sodium nitroprusside (SNP) were measured using a pressurized myograph system (Danish Myo Technology) as described previously [19].

Total mononuclear cells (MNCs) were isolated by Histopaque- 1083 density gradient centrifugation from Peripheral Blood (PB) and Bone Marrow (BM). Determination of circulating EPC number by flow cytometry and EPC culture were performed as described previously [23].

6-8 week old EGFP transgenic mice (C57BL/6-Tg(CAGEGFP) 1Osb/J, Jackson Laboratory) were used as source of bone marrow. 6-8 week old recipient mice (Dki, Aki, WT, and KO) were irradiated lethally and reconstituted with 106 EGFP mice-derived CD45⁺ bone marrow cells 24 hours later by tail vein injection.

Identification of EGFP bone marrow derived EPCs was performed as described previously with minor modifications [24]. Antibody for von Willebrand Factor (vWF; Dako, Denmark) was used to identify cells of endothelial lineage. EGFP was detected by its natural fluorescence. Cells in the endothelial layer that were doubly positive for vWF and EGFP were considered EPCs, and detected with a confocal microscope (Zeiss LSM 5 PASCAL).

These experiments were carried out all as previously described8 and detailed in Online Supplemental Methods.

All results are expressed as mean ± SEM. Statistical analysis was performed by one-way ANOVA (more than 2 groups) and unpaired Student's t test (2 groups). Probability values (P) < 0.05 were considered as statistically significant.

Serum NOx (nitrite and nitrate) levels, a reflection of vascular NO levels, were significantly increased by almost two folds as compared with WT mice (Supplemental Figure 1A), indicating upregulation of eNOS activity by Dki mutation.

Circulating EPC number was firstly determined by EPC culture assay as previously and attached cells showing both DiI-acLDL incorporation and BS-1 lectin binding were quantified, both of which are characteristics of endothelial cells. In line with NOx levels, the number of attached blood EPCs was dramatically increased in Dki mice, as compared to WT mice (Supplemental Figure 1B). Meanwhile, the numbers of circulating EPCs in Aki and eNOS KO mice were not significantly different from WT mice, in agreement with previous studies [8,25]. FACS was used as well to further determine the level of EPCs in the circulation. c-Kit⁺/Flk-1⁺ cells or Flk-1⁺ cells [7,23,26] have been widely used to represent putative EPCs in murine. Our data showed that circulating c-Kit⁺/Flk-1⁺ cells and Flk-1⁺ cells were both significantly augmented in Dki mice, in concert with EPC culture assay results. In Aki and KO mice, circulating c-Kit⁺/Flk-1⁺ cells were not significantly different from WT mice (Figure 1). Notably, circulating Flk-1⁺ cells were slightly but significantly decreased in KO mice, which has not been reported previously. These data demonstrated that increase of eNOS activity alone is sufficient to upregulate basal circulating EPCs in vivo.

We next wanted to evaluate the role of eNOS S1176 phosphorylation or eNOS in EPC mobilization during arterial injury. To induce arterial injury, we used a common carotid artery injury model as described before, [27,28] in which a rubbercoated filament is used to denude a segment of the common carotid artery. Evans blue staining and immunostaining for CD31 was performed to confirm effective endothelial denudation (Supplemental Figure 2).

Circulating c-Kit⁺/Flk-1⁺ cells and Flk-1⁺ cells were tracked with FACS after surgery for 3 days. One day after surgery, both c-Kit⁺/Flk-1⁺ cells and Flk-1⁺ cells in WT mice were rapidly mobilized from the bone marrow to circulation (Figure 1). Intriguingly, c-Kit⁺/Flk-1⁺ cells and Flk-1⁺ cells in Dki mice were not further increased after injury (Figure 1). Furthermore, Aki mice and eNOS KO mice demonstrated similarly increased levels of c-Kit⁺/Flk-1⁺ cells to WT mice after injury (Figure 1B), indicating that mobilization of these cells were not affected by A mutation of S1176 or deletion of eNOS. Of note, circulating c-Kit⁺ cells were rapidly augmented in all strains including Aki and KO mice (data not shown), implying the activation of bone marrow stem and progenitor cells mobilization. In contrast, increase of circulating Flk-1⁺ cells after injury was totally abrogated in eNOS Aki and KO mice (Figure 1C).

The matrix metalloproteinase-9 (MMP-9)/soluble Kit-ligand (sKitL) pathway has been shown to mediate activation and mobilization of bone marrow stem and progenitor cells and is impaired in eNOS KO mice after bone marrow ablation [7,8]. Given that circulating EPCs level was dramatically elevated in Dki mice, we wondered if that resulted from increase of the MMP-9/sKitL pathway. Firstly, MMP-9 expression level and activity was detected by gelatin zymography. The results displayed similar levels of bone marrow MMP-9 activity between WT mice and Dki mice (Figure 2A). We next measured levels of bone marrow sKitL. The data indicated that basal level of bone marrow sKitL was not increased in Dki mice without injury compared with WT mice (Figure 2B).

One day after injury, sKitL levels were rapidly upregulated in all four strains, in line with the circulating c-Kit⁺ cells data described above. Notably, sKitL level in Dki mice was comparable to WT mice (Figure 2B), demonstrating that constitutive phosphomimetic mutation of eNOS S1176 did not result in further increased activation of MMP-9/sKitL in bone marrow. On the other hand, neither S1176A mutation nor total deletion of eNOS blocked the increase of bone marrow sKitL after injury, although the extent of increase of sKitL level in KO mice was significantly smaller than that in WT mice (Figure 2B). This data was consistent with increased circulating c-Kit⁺/ Flk-1⁺ cells and c-Kit⁺ cells in Aki and KO mice after injury and demonstrated elevation of these cells was attributed to increased bone marrow sKitL.

Since eNOS-derived NO has been involved in regulation of cellular proliferation and apoptosis, including EPCs, we determined the proliferation and apoptosis of circulating Flk-1⁺ cells. The data showed that whereas proliferation of these cells evaluated by cell cycle analysis was not altered (Figure 3A), apoptotic circulating Flk-1⁺ cells was significantly attenuated in Dki mice by 36%, as compared to WT mice (Figure 3B). We didn't observe significant changes of EPC proliferation and apoptosis in Aki and KO mice. To explore the molecular mechanisms underlying decreased EPC apoptosis in Dki mice, activation of caspase-3 was determined by FACS. The result demonstrated that significantly less Flk-1⁺ cells in Dki mice were positive for activation of caspase-3 (Figure 3C).

Besides altered survival, enhanced differentiation of EPCs from general stem and progenitor cells could be another contributing factor to the elevated circulating EPCs in Dki mice and decreased level in eNOS KO mice. To explore this possibility, total bone marrow mononuclear cells were isolated and cultured in endothelial-promoting medium in the absence or presence of NO-donor SNP and attached DiI-acLDL/FITCBS- 1 lectin double positive EPCs were quantitated after 4 days culture. In addition, cGMP analog 8-Br-cGMP or sGC inhibitor ODQ was also included to identify the role of cGMP/sGC pathway.

The results indicated that SNP dose-dependently increased the number of attached EPCs by over 200% (Supplemental Figure 2A and 2B). This effect was totally abrogated by pretreatment of sGC inhibitor ODQ. Furthermore, cGMP analog 8-BrcGMP mimicked the effects of SNP on differentiation of EPCs (Supplement Figure 2C).

To identify the contribution of circulating EPCs to the regeneration of denuded endothelium, we transplanted bone marrow cells from age- and gender-matched EGFP-transgenic mice to Dki, WT, Aki, and KO mice, respectively. Because eNOS is mainly expressed in stromal cells in the bone marrow [8], we transplanted only CD45⁺ hematopoietic bone marrow cells, to avoid altering the bone marrow microenvironment of the recipients by donors' stromal cells. Six weeks after transplantation, the bone marrow of all four recipient mice strains was successfully reconstituted (Figure 4B) and similar to their EGFP transgenic donors (data not shown). At this time point, we subjected the recipient mice to carotid artery filament injury.

Two weeks after injury, we quantified the incorporation of BM-derived EPCs by serial cryosection and immunofluorescence staining. Nucleated cells that were EGFP and vWF doubly positive in the endothelial layer were deemed as incorporated EPCs. As shown in Figure 4C and Figure 4D, a significantly higher number of EGFP⁺/vWF⁺ EPCs were found in the regenerated endothelium in Dki mice, as compared with WT mice (Figure 4C and Figure 4D). On the contrary, significantly less EGFP⁺/vWF⁺ EPCs were found in regenerated endothelium in Aki and KO mice (Figure 4C and Figure 4D). These results indicated that recruitment and incorporation of BM-derived EPCs into neoendothelium was positively correlated with eNOS activity.

We next asked whether reendothelialization in these mice was consistent with circulating EPCs level and incorporation as well. Evans blue staining was used to evaluate reendothelialization as described before [9]. Immediately after surgery, Evans blue stained the denuded areas blue, while uninjured areas remained white (Supplemental Figure 3A). Immunostaining of cryosections with CD31 also confirmed the removal of the endothelial monolayer by surgery (Supplemental Figure 3B).

As seen in Figure 5, Dki mutation significantly promoted reendothelialization by about 38% compared with WT mice when assessed 3 days after injury (Figure 5). Aki mice showed compromised regrowth of denuded endothelium, but not significantly different from WT mice (Figure 5). Reendothelialization in eNOS KO mice was significantly delayed compared to WT mice and Dki mice.

Regenerated endothelium after arterial injury remains functionally defective, [29] which could be improved by increasing eNOS-produced NO [30,31]. In the present study, we first determined vascular reactivity in Dki and Aki mice without arterial injury. Our results indicated that the Dki mutation significantly enhanced endothelium-dependent vasodilation as compared to WT mice (Acetylcholine (ACh), (Figure 6A). Of note, these arteries for some reason displayed decreased sensitivity to NO when their vasoreactivity was determined by SNP To assess the role of eNOS in the vasoreactivity of regenerated endothelium in vivo, we repeated the myograph experiment with carotid arteries after arterial injury. Immediately after surgery, the denuded arteries were totally unresponsive to ACh, further confirming removal of the endothelium (Supplemental Figure 4A). Endothelium-independent dilation after injury was also impaired in all strains and not significantly different from each other (Supplemental Figure 4B).

Two weeks after injury, at which time the denuded arteries had been fully reendowthelialized as confirmed by Evans blue staining and CD31 IF staining (Supplemental Figure 4C and 4D) we measured the vascular reactivity again. The data showed that regenerated arteries of Dki mice responded to ACh significantly better than injured WT arteries, although the vasoreactivity was still significantly impaired than uninjured Dki arteries (Figure 6C). Endothelium-independent vasodilation to SNP in regenerated arteries of Dki mice was comparable to WT mice (Figure 6D). Meanwhile, regenerated arteries in WT mice showed impaired vascular response to ACh when compared with uninjured WT arteries (Figure 6C). Endothelium-dependent dilation of regenerated vessels of Aki mice became comparable to that in WT mice (Figure 6C), probably due to augmented sensitivity of underlying smooth muscle cells to SNP (Figure 6D).

In the present study, we created knockin mutations on the serine-1176 of eNOS, a major phosphorylation target of many kinases and vascular function-related factors, and also used eNOS WT or knockout mice, to regulate its enzymatic activity in vivo. With these mouse strains, especially the Dki mice, we demonstrated that increasing eNOS activity was associated with elevated circulating EPC number and increased incorporation of EPCs in regenerated endothelium after arterial injury. Intriguingly, bone marrow EPC mobilization pathway was not activated in Dki mice. Moreover, eNOS might increase EPC number by attenuating apoptosis and enhancing differentiation of EPC (summarized in Supplemental Figure 5).

The pleiotropic effects of NO on vascular homeostasis have underlined the central role of eNOS in preventing development of atherosclerotic cardiovascular disease. Numerous studies demonstrated that dysregulated eNOS results in reduced NO bioavailability, which leads to the loss of vascular integrity and endothelial dysfunction. Circulating EPCs as a novel marker of vascular health have attracted much attention recently. Accumulating evidence suggested their close association with endothelial function and the development of cardiovascular disease in animal models and human subjects [6,32]. In fact, circulating EPCs have become an important mechanism for endogenous endothelial repair after acute arterial injury or under chronic risk factors, such as hyperglycemia, hyperlipidemia, obesity, aging, smoking, hypertension, etc [5,33]. Given the close associations of both eNOS and EPCs with vascular integrity, it's not surprised to see that eNOS has been found to play a role in regulating the number and functions of circulating EPCs, especially EPC mobilization. However, definitive conclusion has been missing due to lack of appropriate and convincing experimental approaches, as discussed in introduction. The eNOS S1176 Dki mice used in this study provided an excellent tool, with which we could increase eNOS activity constitutively without applying any other treatment or molecules, such as exercise, estrogen, VEGF, adiponectin, or statins [2]. These treatment or factors have been reported to have multiple downstream targets therefore it has been difficult to dissect the role of eNOS in their vaso-protective effects such as on circulating EPCs. On the other hand, enhanced phosphorylation of eNOS S1176 was widely recorded in the vascular effects of these treatment or factors. Therefore, it would be of great interest to utilize the S1176 Aki mutation mice to test how much of their vascular effects are mediated by S1176 phosphorylation in future studies.

The studies probing the role of eNOS in EPC mobilization, as mentioned in introduction, reported that there was a concurrent increase of circulating EPCs and eNOS activity, and inhibition of eNOS by its inhibitor L-NAME or genetic knockout either partially or totally abrogated the increase of circulating EPCs [8-17]. However, most of these studies failed to identify the status of bone marrow MMP-9/sKitL pathway, the canonical EPC mobilization pathway. The mechanisms underlying how activation of eNOS led to increased circulating EPCs therefore remain unknown. Our study clearly showed that activation of eNOS alone did not stimulate activation of the MMP-9/sKitL in bone marrow. When challenged with arterial injury, the MMP-9/sKitL pathway was activated in all strains including eNOS KO mice, accompanied with increase of circulating c-Kit⁺ cells. In one hand, we found Dki mutation of eNOS did not synergistically act with arterial injury to further increase bone marrow sKitL levels. On the other hand, whereas increase of sKitL in KO mice was significantly impaired as compared to WT mice, the extent of increase of circulating c-Kit⁺/Flk-1⁺ cells, a widely accepted population of primitive EPCs in mice, was not significantly different from WT mice. This data actually is consistent with a previous report which showed increase of circulating EPCs during hind limb ischemia was not affected in eNOS KO mice [25].

Flk-1 (VEGFR-2, or KDR in human) is the earliest marker of angioblast precursors [4,34]. Flk-1 defines a population of cells with significant potential for vascular growth [26,34,35]. Since Flk-1 is also expressed on mature ECs and it's suggested that there is a small number of ECs in blood sloughed from vascular wall [37], another stem/progenitor cell marker is widely used together with Flk-1 to identify putative circulating EPCs, such as CD34 in human or c-Kit in mice. Of note, Flk-1⁺ cells proliferated in vivo and effectively reendothelialized injured arteries [26]. Therefore, Flk-1⁺ cells appear to be more suitable to represent the circulating EPCs in terms of in vivo reendothelialization capacity. The present study found that sKitL level after injury in Aki mice was comparable to WT mice and significantly elevated in eNOS KO mice, but increases of Flk-1⁺ cells were abrogated. Accordingly, incorporated bone marrow-derived EPCs were significantly less in Aki and KO mice. Therefore, it appears that eNOS regulated Flk-1⁺ cells mobilization and incorporation independent on MMP-9/ sKitL pathway.

Regarding the elevated basal circulating EPCs level in Dki mice, we found that Dki mutation significantly attenuated apoptosis of circulating Flk-1⁺ cells by inhibiting activation of caspase-3. This finding is consistent with the anti-apoptotic effects of eNOS on EPCs described previously [13,38,39]. In addition, eNOS-derived NO may increase circulating EPC levels via enhancing differentiation of general stem/progenitor cells into EPCs, which is mediated by a cGMP/sGC-dependent pathway.

A major effect of eNOS in angiogenesis is to promote EC functions, such as migration and tube formation. The differential reendothelialization in our eNOS strains hence could also result from the altered functionality of adjacent ECs and EPCs, which could express eNOS during maturation. Several groups have demonstrated that eNOS is expressed at intermediate levels in cultured human blood-derived EPCs and is important for their in vitro and in vivo functions [18,40-42]. Importantly, decreasing eNOS expression level in human EPCs directly reduced their in vivo reendothelialization capacity [40]. Nevertheless, it's been extremely difficult to recapitulate these findings with cultured mice EPCs with respect to expression of eNOS. Due to the low volume of peripheral blood, cultivation of mice-derived EPCs has been using bone marrow cells. So far, only one group has reported the expression of eNOS in mice EPCs [43]. In the present study, we could not detect the expression of eNOS in cultured EPCs, either from total bone marrow MNCs or purified Lineage-c-Kit⁺Flk-1⁺ cells [23]. Given previous human data, we believe modulation of eNOS activity by S1176D mutation, S1176A mutation and KO may affect the in vivo functions of circulating EPCs and endothelial repair. However, future studies using more sophisticated approaches such as monitoring EPC differentiation in vivo, or lineage tracing based eNOS expression are in need.

In summary, the current study demonstrates that modulation of eNOS activity in vivo regulates circulating EPCs levels and their direct contribution to endothelial repair and vascular function during recovery from arterial injury. Our finding provides definitive evidence for the role of bone marrowderived EPCs in eNOS-mediated vascular integrity and scientific support for many therapies which increase circulating EPCs and prevent cardiovascular disease based on activation of eNOS.

This work was supported by PHS R01 NS33335-16 to Paul L. Huang. Junlei Chang was supported by a University Postgraduate Fellowship from The University of Hong Kong.

None