Adenosine can stimulate angiogenesis, but its role in the distinct process of arteriogenesis is unknown. We have previously reported that mice lacking ecto-5′-nucleotidase (CD73−/−) show enhanced monocyte adhesion to the endothelium after ischaemia, which is considered to be an important trigger for arteriogenesis.
Hindlimb ischaemia was induced in wild-type (WT) and CD73−/− mice to study the role of extracellularly formed adenosine in arteriogenesis. Magnetic resonance angiography (MRA) was performed for serial visualization of newly developed vessels at a spatial resolution of 1 nL, and high-energy phosphates (HEP) were quantified by 31P MR spectroscopy (MRS). MRA of CD73−/− mice revealed substantially enhanced collateral artery conductance at day 7 [CD73−/−: 0.73 ± 0.11 a.u. (arbitrary units); WT: 0.44 ± 0.13 a.u.; P < 0.01, n = 6], and MRS of the affected hindlimb showed a faster restoration of HEP in correlation with enhanced functional recovery in the mutant. Additionally, histology showed no differences in capillary density between the groups but showed an increased monocyte infiltration in hindlimbs of CD73−/− mice.
Serial assessment of dynamic changes of vessel growth and metabolism in the process of arteriogenesis demonstrate that the lack of CD73-derived adenosine importantly promotes arteriogenesis but does not alter angiogenesis in our model of hindlimb ischaemia.
Modulation of arteriogenesis has the potential for new therapies of atherosclerotic obstructive diseases by inducing collateral artery growth and bridging vascular lesions by stenosis or occlusion with ‘natural bypass’ vessels.1 Haemodynamic changes induced as a consequence of vessel occlusion increase fluid shear stress (FSS) on endothelial cells in collateral vessels and transform them to an ‘activated’ state. This is thought to be an important initial trigger for arteriogenesis, followed by attraction of circulating monocytes,2 which are major players in this complex process at the cellular level.3,4 A versatile set of mediators like monocyte attracting protein (MCP-1), granulocyte–monocyte-colony-stimulating factor (GM-CSF), vascular endothelial growth factor (VEGF), fibroblast-growth factor (FGF-2), and tumour necrosis factor α (TNF-α) have been identified to promote and maintain collateral remodelling.5–8
Adenosine, a ubiquitous nucleoside formed intracellulary but also extracellularly by stepwise dephosphorylation of ATP, is known as a pro-angiogenic factor stimulating capillary growth under hypoxia.9 We tested the hypothesis whether adenosine is also a stimulant for arteriogenesis and therefore we eliminated an important source of external adenosine by targeted disruption of the CD73/ecto-5′-nucleotidase. Adenosine-signalling is mediated via four cell-surface receptors (A1, A2A, A2B, and A3) which are known to stimulate endothelial migration, proliferation, and secretion of VEGF.10 Recently, adenosine A2A receptors have been reported to be important in pulmonary vascular remodelling after pulmonary hypertension.11 Furthermore, it was shown that activation of A1 receptors could reduce leucocyte infiltration in a model of acute lung injury.12 Activation of A2A receptors on immune cells significantly attenuates cytokine and chemokine release, which is related to the anti-inflammatory action of adenosine.13 The extracellular adenine nucleotides ATP/ADP, precursors of the formation of adenosine through CD39 (ectonucleoside triphosphate diphosphohydrolase 1) and CD73 (ecto-5′-nucleotidase), signal through P2 receptors which include more than nine identified members.14 The importance of nucleotides in the regulation of platelet function is well appreciated, and nucleotides are known to modulate the secretion of various cytokines like TNF-α and interleukin (IL) 1-β and to control leucocyte adhesion and trafficking through the endothelium.14,15 It is important to recognize that it is the activity of CD39 which influences the biological half-life of extracellular ATP and together with CD73 determines whether ATP or adenosine receptors are preferentially activated.
In a previous study, we have observed that mouse mutants lacking CD73 show increased leucocyte attachment to the endothelium after ischaemia–reperfusion and ex vivo-perfused carotid arteries exhibit enhanced monocyte adhesion.16 Since monocyte adhesion is known to be a crucial initial step in arteriogenesis,3,4 the present study explored the role of CD73 in this process using a murine hindlimb ischaemia model. Arteriogenesis was monitored by serial in vivo magnetic resonance angiography (MRA) measurements. With this approach, newly formed vessels could be directly visualized, enabling the sensitive quantification of luminal areas for interindividual comparison. Additionally, we analysed energy metabolism by 31P MR spectroscopy as a measure for tissue recovery after ischaemia. The in vivo acquired data were complemented by histology and immunohistochemistry.
CD73−/− mice were generated on a C57BL/6 strain by deletion of exon 2 with a cre/lox system which leads to the loss of the active core and the GPI-anchor of CD73 as previously described.16 C57BL/6 wild-type (WT) and CD73−/− mice aged between 8 and 12 weeks were used for this study. All animal experiments were performed with the approval and permission of the local animal-care committee and in accordance with the state legislation and are conformed with the Guide for the Care and Use of Laboratory Animals published by the European Parliament (Directive 2010/63/EU).
Hindlimb ischaemia model
Mice (n = 6 for each group) were anaesthetized by intraperitoneal injection of 100 mg/kg ketamine and 10 mg/kg xylazine, and severe hindlimb ischaemia was induced by a 5 mm excision of the femoral artery after ligation proximal to the profundal femoral artery and distal to the branching of the popliteal artery.17 The femoral nerve and vein were preserved. Periodic observation of respiration, colour of mucous membranes, and pain response was used for monitoring the adequacy of anaesthesia. Formation of collateral vessels and blood flow recovery were assessed with MRA at days 3, 7, 14, and 21 after femoral artery ligation.
Anaesthesia of mice was performed with 1.5% isoflurane in a oxygen/nitrogen gas mixture (70:30). Rectal temperature was monitored throughout the entire experiment and maintained by adjustment of the gradient cooling device (Haake UWK 45) at 38 ± 1°C.
MRI experiments were carried out at a vertical Bruker DRX 9.4T wide-bore NMR spectrometer equipped with a 40 mm gradient set (capable of 1 T/m maximum gradient strength) and a linearly driven 30 mm SAW resonator.
For MRA, fast flow-compensated gradient-echo sequences were applied, resulting in a three-dimensional (3D) time-of-flight (TOF) angiography without the use of any contrast agent as previously described in detail by Jacoby et al.18 The field of view (FOV) was positioned over the hindlimb that it covered the arterial iliacal bifurcation to the distal popliteal artery at the beginning of the gastrocnemius muscle. To decrease signal intensity loss along the z-axis and to improve the sensitivity of the protocol for slow blood flow and small vessels, the FOV was partitioned into several overlapping slabs and scanned with the following parameters: TR = 23 ms, TE = 2.9, flip angle = 35°, FOV = 2.56 × 2.56 × 0.64 cm3, matrix size = 256 × 256 × 64, voxel size of 1 nL, and acquisition time (TAcq) of 6.2 min per slab. The entire length of five overlapping slabs was 1.92 cm with a total scanning time of 31 min.
Measurements of high-energy phosphates (HEP) within the operated hindlimb were performed with a 10 mm tilt resonator (1H/31P double-tuned) positioned over the thigh of the mouse. Parameters used for 31P MR spectroscopy (MRS) were TR = 352 ms, flip angle = 30°, spectral bandwidth = 6460 Hz, 2048 scans, and TAcq = 12 min. Concomitantly, we acquired T2-weighted images for the evaluation of tissue oedema and subsequent muscle atrophy.
MR data analysis
For the analysis of the acquired MRA data sets, we used the freely available software package ECCET (www.eccet.de). The 3D data were visualized and processed as described previously.18 The software module Angiotux allows us to navigate through the volume data set in real time. Basic threshold segmentation was carried out manually followed by volumetric measurements in a voxel-based manner. Luminal size of developing collateral vessels was normalized to the non-operated limb by segmentation of the intact femoral artery between the proximal end (the last branching of the iliac artery) and the branching into the popliteal artery.
Analysis of the MRS data was performed with TopSpin (Bruker, Rheinstetten). Chemical shifts were referenced to the resonance of phosphocreatine (PCr) at −2.52 p.p.m. After manual phase and baseline correction, the content of PCr and ATP was determined by integration. HEP content was measured in parallel to blood flow restoration at days 3, 7, 14, and 21 after surgery.
Functional recovery after inducing hindlimb ischaemia was assessed in a randomized and blinded fashion similar to the scoring system introduced by Heil et al.19: 0 = no active leg movement, 1 = use of the leg, 2 = active foot movement, 3 = no restrictions.
Tissue sampling and morphometry of collateral vessels
Tissue was sampled 7 days after operation—the point in time with the largest differences between the groups as assessed by in vivo measurements. Animals were sacrificed with a lethal dose of 400 mg/kg ketamine plus 20 mg/kg xylazine i.p. followed by perfusion fixation (3% paraformaldehyde) and were rinsed with buffer. The adductor muscle containing the collateral arteries and the contralateral side were harvested and embedded in TissueTek© for cryosection (CM 3000 croystat Leica). For the analysis of angiogenesis in the lower leg, the gastrocnemius muscle was harvested from both sides. Serial sections (6 µm) were stained with an FITC (fluoresceine iso-thiocyanate)-coupled smooth-muscle-actin (α-smc) antibody and with a tetramethyl rhodamine iso-thiocyanate-conjugated BS-1 lectin antibody (Sigma-Aldrich) for endothelial staining. DAPI (Mobitec) was used for nuclear staining. Transversal sections were photographed with a Leica DMLD digital camera on a Leitz DMRB fluorescence microscope. Morphometric analysis was performed using ImageJ (NIH).
Since monocyte invasion was reported to be highest in the first days of inflammation,3 accumulating monocytes were quantified in both groups at day 3 after tissue harvest. Sections (8 µm) from three different levels of the adductor muscle were prepared, and CD11b (Santa Cruz) was used for monocyte staining. Identified monocytes were counted and normalized with the total number of artery vessel segments per section.
All data are presented as mean value ± SD. Data sets were analysed by Student's t-test and a P-value of <0.05 was considered significant.
Collateral vessel growth by MR angiography
A representative MR angiogram of the mouse hindlimb 3 weeks after ligation (location indicated by the white bar) of the left femoral artery is displayed in Figure 1. Overlapping slabs covering the region of interest were acquired and subsequently merged into one single volume displayed as a maximum intensity projection (MIP). After careful vessel segmentation, surface rendering was performed, showing in great detail the entire vessel tree of the mouse hindlimb with the femoral artery and its branches into the popliteal artery (see also Supplementary material online, Movie S1). The original vascular bed distal to the ‘re-entry’ is coloured in red. One can clearly locate the origin and course as well as the re-entry zones of the newly developed collateral vessels (coloured in green). Collaterals are characterized by their corkscrew morphology, typical for the process of arteriogenesis when growth in length outweighs increase in diameter.20
For the assessment of the growth of collateral vessel volume over the course of 3 weeks, serial measurements were performed at days 3, 7, 14, and 21 after the occlusion of the femoral artery in WT and CD73−/− mice. Quantification of the acquired data sets was essentially performed as previously described by means of MIP rendering, segmentation, and voxel count using a dedicated software.18 The voxels of the newly developed collateral arteries and an anatomically defined segment as reference from the intact contralateral femoral artery (coloured in blue in Figure 2A and B) were determined for the calculation of the right–left ratio to account for differences in individual animal size and vessel geometry. Representative MR angiograms of both WT and CD73−/− acquired at day 7 after surgery are illustrated in Figure 2A and B, demonstrating the enhanced collateral vessel volume (green vessels) in the mutant mice. Figure 2C summarizes data on the time course of collateral artery development (expressed in normalized ratios) over 3 weeks in WT and CD73−/− mice after the induction of hindlimb ischaemia. As can be seen, at day 3, the CD73 mutants exhibited an improved collateralization which reached the level of significance at day 7 (0.73 ± 0.11 in CD73−/− when compared with 0.44 ± 0.13 in the WT group; P < 0.01, n = 6). The enhanced collateralization in CD73−/− mice was a transient phenomenon which subsided over the following 2 weeks. This dynamic of collateral growth is consistent with observations by Schaper's group.21 Similarly, when limb usage was assessed by functional scoring, the transgenic animals showed a faster recovery of limb function (Supplementary material online, Figure S1). Analysis of tissue oedema by MRI using anatomical cross-sections of the ischaemic hindlimb (Supplementary material online, Figure S2) did not show any significant differences between the two experimental groups (data not shown).
Muscle HEP by 31P MRS
In order to investigate the impact of the enhanced blood flow restoration on tissue metabolism, we used MRS to measure muscle HEP with a dedicated coil positioned over the affected hindlimb (Supplementary material online, Figure S2 shows the corresponding anatomy). Characteristic 31P MR spectra of the hindlimb in WT and CD73−/− mice after ischaemia and recovery are displayed in Figure 3A over a period of 3 weeks. The signals of PCr and ATP are well resolved at each point in time and thus can be used for quantification. From the data compiled in Figure 3B, it can be seen that the ischaemia-induced drop of PCr and ATP levels was significantly less pronounced in CD73−/− mice, particularly with regard to PCr at day 7. Similarly, the recovery of ATP from the ischaemic injury was faster in CD73 mutants. Thus, the metabolic changes mirrored the differences in the formation of collaterals in CD73 and WT mice as depicted in Figure 2.
Collateral vessel morphology, capillary density, and monocyte invasion
Since the in vivo MR measurements indicated that the largest differences in the arteriogenic response between WT and CD73−/− mice occurred at day 7, we carried out detailed histology of collateral arteries in the ischaemic hindlimb at this point of time in a separate experimental series. As shown in Figure 4A and B, collateral vessels of CD73−/− mice are characterized by a larger luminal diameter and a more pronounced staining of vascular smooth-muscle cells compared with the control group. From the data summarized in Figure 4C, it is evident that the inner and outer vessel diameters, the calculated wall thickness, and the surface area were significantly increased in CD73−/− mice compared with WT controls.
We also examined the angiogenic response by measuring capillary density in skeletal muscle before and after 7 days of hindlimb ischaemia in WT and CD73−/− mice. As expected, capillary density increased upon ischaemia from 46 ± 17 to 53 ± 19 per section (Figure 5A and B). However, there were no significant differences between the CD73−/− and control group (Figure 5C).
Since monocyte invasion is known to be an important prerequisite for the initiation of arteriogenesis,3,4,22 we investigated whether the enhanced collateral vessel formation observed at day 7 may be preceded by an increased monocyte infiltration. To this end, we harvested skeletal adductor muscle at day 3 and counted the collateral vessels on each section together with the surrounding monocytes identified by CD11b staining. As shown in Figure 6, the number of monocytes accumulating after ischaemia was significantly higher in CD73−/− mice compared with WT controls (17 ± 10 monocytes in WT, 31 ± 15 monocytes in CD73−/− mice).
Angiogenesis and arteriogenesis are two distinct processes by which the body responds to the obstruction of large conduit arteries. Whereas arteriogenesis denotes the adaptive outgrowth of pre-existent collateral arteries to bypass arterial stenoses in response to haemodynamic changes,23 angiogenesis describes capillary growth by sprouting from pre-existing vascular structures.24 The present study reports that in the model of hindlimb ischaemia, lack of CD73-derived adenosine promotes arteriogenesis but does not alter angiogenesis. Therefore, adenosine formed by extracellular nucleotide catabolism on endothelial cells16 and immune cells13 appears to be an important endogenous modulator of arteriogenesis.
To continuously monitor the process of arteriogenesis, we used TOF MRA at 9.4 T, which permitted the repetitive quantitative and qualitative 3D analyses of small vessel changes in vivo without any contrast agent as was previously reported by us.18 The sensitivity for detecting changes in blood flow and thereby vessel size was further improved by overlapping slab acquisition, which resulted in reduced signal loss. Assessing arteriogenesis by MR techniques has the additional advantage that MRS offers the unique possibility to measure HEP content (ATP, PCr) as a functional correlate in the affected hindlimb muscle.25,26 Combining MRI with MRS, we found that the enhanced arteriogenesis in the CD73 mutant was associated with a faster recovery of muscle HEP. This strongly suggests that through enhanced blood flow and thus oxygen supply, muscle energetics and function were significantly improved. The higher ATP tissue levels measured in the CD73−/− mutant are therefore most likely a direct consequence of better oxygen delivery through enhanced arteriogenesis and do not mirror differences in ecto-nucleotidase activities.
It should be noted that other techniques to quantify arteriogenesis such as laser Doppler perfusion imaging, which is commonly used for in vivo measurements in murine models of arteriogenesis,4,27 display tissue perfusion rather than arterial vessel diameter and this technique may be less sensitive when monitoring collateral development in deeper muscle layers. We found the maximum increase of collateral vessel conductance as measured by MR angiography at day 7 in both genotypes. This pattern corresponds well to the findings of Herzog et al.28 reporting a typically biphasic pattern of collateral vessel expansion: vessel diameter enlargement in the first week, followed by an increase in vessel length in the subsequent time period resulting in typically ‘corkscrew’ morphology.
Arteriogenesis describes the ‘outward’ remodelling of pre-existing collateral vessels, resulting in artery enlargement with increased conductance after obstruction of a main supplying stem vessel. This process is driven by haemodynamic factors such as increase in stretch and FSS on endothelial cells.23,29 FSS-activated endothelial cells, by expression of MCP-1 and ICAM-1, do attract mononuclear cells which invade the vessel wall.30 The critical role of monocytes and pro-inflammatory cytokines for adaptive arteriogenesis is well established.31,32 More recently, a role of toll-like receptors 2 and 4 in adaptive collateral artery growth has been reported,33 suggesting a link to innate immunity.34 Moreover, lack of HIF hydroxylases PHD2 prevents ischaemia by induction of arteriogenesis, most likely by controlling the differentiation state in macrophages.35
Given the central role of monocytes/macrophages in the process of arteriogenesis, the well-known ability of adenosine to suppress the pro-inflammatory responses of classically activated macrophages by Th1 cytokines36 is of particular interest. CD73−/− mice are generally characterized by a pro-inflammatory phenotype with increased leucocyte attachment to the endothelium after ischaemia–reperfusion experiments,16 elevated macrophage content after wire-induced injury of carotid arteries, and elevated endothelial NF-κB activation with the up-regulation of endothelial adhesion protein, including VCAM-1.37 CD73 was also reported as a critical mediator of vascular barrier function during hypoxia, resulting in fulminant vascular leakage when CD73 is lacking.38 In the present study, we observed enhanced monocyte invasion in CD73−/− mice 3 days after the induction of hindlimb ischaemia. Monocyte adhesion is promoted by interaction of integrin α4β1 (also very late antigen-4, VLA4) with VCAM-1, and adenosine suppresses the expression of VCAM-1 by inhibiting the release of IL-6 and IL-8.39 Stimulation of A2B receptors also reduces VLA4 expression via a cAMP/PKA-mediated pathway40 which increases endothelial barrier function, alleviates vascular permeability by strengthening of the intercellular junctions41,42 and inhibits release of TNF-α.43 Adherent leucocytes are able to reduce extracellular adenosine by direct CD73 interaction,44 thereby promoting leucocyte transmigration.45 In addition, CD73 knockdown experiments with RNAi revealed that CD73 depletion increases the expression of ICAM–1, VCAM–1, and E-selectin, promotes NF-κB translocation, and changes endothelial cell morphology with altered actin cytoskeletal organization, which resembles the phenotype observed under the treatment of TNF-α.46 Thus, the lack of CD73-derived adenosine very likely facilitated diapedesis of adherent monocytes to trigger arteriogenesis.
The anti-inflammatory potential of adenosine is well known47 and inhibition of neutrophil-activation by adenosine has already been described in 1986.48 We now know that macrophages are characterized by a vast phenotypic plasticity which led to the distinction of classically activated M1 and alternatively activated M2 populations.49 In classical activation with LPS, adenosine inhibits TNF-α, IL-6, and IL-12 release and augments IL-10 and VEGF, which is mediated through both A2A and A2B receptors.43,50,51 Recently, it was reported that adenosine also promotes alternative macrophage activation via A2A and A2B receptors.52 Whether CD73-derived adenosine modulates arteriogenesis via action on M1 and M2 macrophages remains to be elucidated. Interestingly, Takeda et al.35 reported that the PHD2 oxygen sensor regulates arteriogenesis also by modulating the macrophage phenotype through the activation of the NF-κB pathway and by promoting the M2 phenotype.
Several lines of evidence suggest that adenosine promotes angiogenesis in ischaemic tissue.53 Adenosine has also been found to drive proliferation of endothelial cells, migration, and subsequent network development in in vitro studies.9,54 Activation of A1 receptors promotes angiogenesis and the release of VEGF from monocytes.55 It was, therefore, at first sight surprising that angiogenesis as assessed by quantification of capillary density at day 7 was only small in WT mice and did not show significant differences between the two genotypes. Because hypoxia is a major stimulus for angiogenesis, it is possible that arteriogenesis attenuated tissue hypoxia and thereby reduced the extent of angiogenesis. Similarly, Deindl et al.56 reported that neither HIF-1α-mRNA nor the HIF-controlled VEGF gene expression was significantly up-regulated in a rabbit hindlimb ischaemia model. In mice lacking CD73, the enhanced arteriogenesis further attenuated tissue hypoxia, which can explain the faster recovery of HEP. Thus, it is likely that lack of effects of CD73 on angiogenesis in our model is presumably due to upstream CD73-mediated arteriogenesis, which mitigated the extent of distal ischaemia by increased blood flow.
In conclusion, our data indicate that adenosine formed extracellularly by CD73 is an important modulator of arteriogenesis and thus influences the vascular remodelling process. MRI together with MRS permitted to monitor the dynamic changes in vascular blood flow after vessel ligation together with HEP in the affected muscle. The mechanism of adenosine action most likely relates to the effect of this nucleoside on transendothelial migration of monocytes. Adenosine-dependent changes of monocyte phenotype with altered cytokine release may also have participated. As to the relevance of our findings, a recent study identified humans lacking CD7357 who exhibited extensive arterial calcifications. Whether these subjects show altered arteriogenesis is presently not known.
This study was supported by the Deutsche Forschungsgemeinschaft, subprojects B6 and Z2 of the Sonderforschungsbereich 612.
We thank Jutta Zieman, Dr Barbara Emde, and Sandra Rühl for excellent assistance and support.
Conflict of interest: none declared.