Macrophage skewing by Phd2 haplodeficiency prevents ischaemia by inducing arteriogenesis

Abstract

PHD2 serves as an oxygen sensor that rescues blood supply by regulating vessel formation and shape in case of oxygen shortage^1–5. However, it is unknown whether PHD2 can influence arteriogenesis. Here we studied the role of PHD2 in collateral artery growth by using hindlimb ischaemia as a model, a process that compensates for the lack of blood flow in case of major arterial occlusion^6–8. We show that Phd2 (also known as Egln1) haplodeficient (Phd2^+/−) mice displayed preformed collateral arteries that preserved limb perfusion and prevented tissue necrosis in ischaemia. Improved arteriogenesis in Phd2^+/− mice was due to an expansion of tissue-resident, M2-like macrophages^9,10 and their increased release of arteriogenic factors, leading to enhanced smooth muscle cell (SMC) recruitment and growth. Both chronic and acute deletion of one Phd2 allele in macrophages was sufficient to skew their polarization towards a proarteriogenic phenotype. Mechanistically, collateral vessel preconditioning relied on the activation of canonical NF-κB pathway in Phd2^+/− macrophages. These results unravel how PHD2 regulates arteriogenesis and artery homeostasis by controlling a specific differentiation state in macrophages and suggest new treatment options for ischaemic disorders.

To understand whether partial loss of PHD2 enhances perfusion of ischaemic tissues, we subjected mice to femoral artery ligation, a procedure that reduces perfusion of the lower limb, causing ischaemia in the calf (that is, the crural muscle). After ligation, Phd2^+/− mice showed a milder drop in perfusion and oxygen tension with reduced hypoxia in the crural muscle compared to wild-type (WT) mice (Fig. 1a–g). Ischaemia promotes oxidative stress early and angiogenesis as a later response^6,11. Oxidative stress (12 h post-ligation) and capillarization (14 days post-ligation) were both increased in the crural muscle of WT, but not Phd2^+/− mice (Supplementary Fig. 2a–g). As a consequence of preserved blood flow, Phd2^+/− crural muscles showed reduced ischaemic necrosis and increased viability (Fig. 1h–j and Supplementary Fig. 2h–j). WT mice showed signs of muscle regeneration that were absent in Phd2^+/− crural muscles (Supplementary Fig. 2k–m). The protection against ischaemic damage improved physical endurance of Phd2^+/− mice in ischaemia (Fig. 1k), although both genotypes had similar running capacity at baseline (Supplementary Fig. 2n).

Figure 1. Phd2^+/− mice are pre-adapted to ischaemia.

a, Laser Doppler analysis 12, 24, 48 and 72 h post-ligation. b, Magnetic resonance imaging (MRI) oxymetry in crural muscles 12 h post-ligation. c–g, Hypoxic area quantification (g) on pimonidazole-stained sections of crural muscles 12 h post-ligation (e, f); hypoxia was undetectable at baseline (c, d). Scale bars, 200 µm. g, Pimonidazole⁺ area 12 h post-ligation. h–j, Necrotic area quantification (h) on haematoxylin- and eosin-stained sections of crural muscles 72 h postligation (i, j). Scale bars, 100 µm. k, Treadmill running test 12 h post-ligation. l, m, Macroscopic view of adductors after gelatin-bismuth-based angiographies at baseline. Collateral vessels: primary (blue arrow), secondary (red arrow), and tertiary (black arrow). Femoral artery (green arrow). n, o, Haematoxylin and eosin staining of adductors at baseline after angiographies; bismuth⁺ collaterals appear black. Scale bars, 50 µm. p, q, Collateral vessel area (p) and density (q) represented in n, o. r, Quantification of large vessels (>200 µmin diameter) in the thigh at baseline after micro-computed tomography angiograms. s, t, Representative micro-computed tomography micrographs of the thigh (yellow frame). u–x, Morphological analysis on αSMA-stained sections of adductors at baseline and 72 h post-ligation, as represented in y–b′. Scale bars, 10 µm. All graphs show mean ± s.e.m. All experiments, n≥5. *, P < 0.05 towards WT. #, P < 0.05 towards baseline.

Because Phd2^+/− mice were protected against ischaemia already 12 h post-ligation, we proposed that they could tolerate ischaemic insults better due to increased collaterals at baseline¹². Macroscopic inspection of the upper limb, that is, the thigh, and histological analysis of the adductor (in the inner thigh, where collaterals grow) after gelatin-bismuth angiographies showed about twice higher numbers and area of bismuth-positive collaterals in non-ligated Phd2^+/− versus WT mice (Fig. 1l–q and Supplementary Fig. 3a, b). Also micro-computed tomography scans and X-ray radiographies showed higher numbers of large vessels (>200 µm in diameter) in Phd2^+/− than WT thighs at baseline (Fig. 1r–t and Supplementary Fig. 3c, d), whereas numbers of smaller vessels (<200 µm in diameter) and capillaries were comparable in both genotypes (Supplementary Fig. 3e–g). Both the total area and numbers of bismuth-positive collaterals were still higher in Phd2^+/− versus WT adductors 12 and 72 h post-ligation, a time-window when collateral remodelling just begins in WT mice (Fig. 1p, q and Supplementary Fig. 3a, b).

We also assessed whether Phd2^+/− mice were protected against myocardial ischaemia. Twenty-four hours after coronary artery ligation, desmin-negative area (a readout of cardiomyocyte death) was smaller in Phd2^+/− hearts (Supplementary Fig. 4a–c). Compared to WT, Phd2^+/− hearts showed higher perfusion in both infarcted and remote myocardium (Supplementary Fig. 4d–g). At baseline, density of large vessels, but not small vessels and capillaries, was higher in Phd2^+/− versus WT hearts (Supplementary Fig. 4h–l).

To increase blood flow in case of major arterial occlusion, collateral vessels undergo extensive remodelling (arteriogenesis) with thickening of the tunica media, consisting of α-smooth muscle actin (αSMA)-positive SMCs, and enlargement of vessel diameter⁸. Numbers and total area of αSMA⁺ collateral vessels were higher in Phd2^+/− adductors both at baseline and after ischaemia, whereas the mean area and thickness of the tunica media were higher only at baseline (Fig. 1u–b′). These data show that collateral vessels of Phd2^+/− mice at baseline were similar to those of WT mice after femoral artery ligation. This ‘collateral vessel preconditioning’ was protective against ischaemia.

Inflammatory cells and macrophages in particular are responsible for collateral vessel remodelling^7,8. Nevertheless, CD45⁺ leukocyte and F4/80⁺ macrophage infiltration of the adductors was similar at baseline and equally increased after ligation in both genotypes (Fig. 2a, b). We therefore analysed the phenotype of infiltrating macrophages and measured the density of M2-like, wound-healing/pro-angiogenic macrophages by their expression of the MRC1 mannose receptor^9,10. At baseline, F4/80⁺ MRC1⁺ macrophages were 75% higher in Phd2^+/− versus WT adductors (Fig. 2c–e). Seventy-two hours after ligation, their numbers were increased by 95% in WT and only 50% in Phd2^+/− mice (Fig. 2c, f, g). Gene-profiling of Phd2^+/− peritoneal macrophages (pMØ) showed higher expression of M2-type genes^9,10,13, including Tek (also known as Tie2), Arg1, Cxcr4, Ccr2, Hgf, Pdgfb, Fizz, Nrp1, Mmp2, Cxcl12 (also known as Sdf1) and Tgfb, than WT pMØ (Fig. 2h). Conversely, several proinflammatory or anti-angiogenic (that is, M1-type) molecules were downregulated in Phd2^+/− macrophages; these included Il1b, Il6, Nos2 and Il12 (Fig. 2h). Similarly, Phd2^+/− macrophages sorted from the adductor expressed higher levels of Pdgfb, Sdf1, Tie2, Mmp2 and Nrp1 at baseline (Fig. 2i). Seventy-two hours post-ligation, the expression level of these markers was similar in Phd2^+/− and WT tissue macrophages (Fig. 2i). Expression of these genes was comparable in WT and Phd2^+/− endothelial cells isolated from adductors at baseline or in ischaemia (Supplementary Table 1). Noteworthy, the basal level of Phd2 in Phd2^+/− macrophages was half of that in WT macrophages and did not change in ischaemia (Fig. 2i). Conversely, Phd2 expression in WT macrophages was reduced by ~50% in ischaemia and thus reached the same level as in Phd2^+/− macrophages (Fig. 2i). Thus, Phd2^+/− macrophages showed a unique and cell-specific gene signature, which was reminiscent, at least in part, of M2-polarized macrophages and of WT macrophages in ischaemia.

Figure 2. Phd2^+/− macrophages display an M2-like phenotype.

a, b, Quantification of CD45⁺ leukocyte and F4/80⁺ macrophage infiltration of adductors at baseline and 72 h post-ligation. c, Quantification of MRC1⁺ macrophages in adductors at baseline and 72 h post-ligation, represented in d–g. d–g, Micrographs of immunostainings for F4/80 (green), MRC1 (red) and αSMA (blue). Arrowheads (f, g) point to F4/80⁺MRC1⁺ cells. Scale bars, 20 µm. h, Gene profile by quantitative PCR of peritoneal macrophages (pMØ). i, Gene profile of F4/80⁺ macrophages sorted from adductors at baseline and 72 h post-ligation. j–m, Quantification (m) of crystal-violet-stained SMCs, migrated towards control medium (j), WT (k) or Phd2^+/− (l) macrophages. Scale bars, 50 µm. n, SMC growth in response to soluble factors released by WT or Phd2^+/− macrophages. o–s, Gene profile by quantitative PCR of SMCs exposed to control, WT macrophage-conditioned medium, or Phd2^+/− macrophage-conditioned medium. All bars show mean±s.e.m. All experiments, n≥5. *, P < 0.05 towards WT. #, P < 0.05 towards baseline. $, P < 0.05 towards control medium.

We therefore assessed whether WT and Phd2^+/− macrophages affect the behaviour of endothelial cells and SMCs, the two main cellular components of arteries. Soluble factors released by Phd2^+/− macrophages strongly increased migration and proliferation of SMCs, but not endothelial cells, probably because the latter were already highly responsive to WT macrophages (Fig. 2j–n and Supplementary Fig. 5a–e). Consistently, SMCs exposed to conditioned medium from Phd2^+/− macrophages showed reduced levels of calponin-1, Sm22, smoothelin, NMHC-B and αSMA (Fig. 2o–s), indicating enhanced proliferation^14,15. In vitro knockdown of both Sdf1 and Pdgfb, known to stimulate SMC recruitment and proliferation^16,17, abolished the enhanced response of SMCs to Phd2^+/− macrophages, although inhibition of either factor was also very effective (Supplementary Fig. 5f, g and Supplementary Note 1). Overall, these data showed that lower levels of PHD2 pre-adapt macrophages to ischaemia by skewing them towards an M2-like phenotype, which promotes SMC recruitment and growth.

We then investigated whether reduced levels of PHD2 in macrophages promoted collateral vessel preconditioning. Myeloid-cell specific Phd2 haplodeficiency (Phd2^{LysCre;lox/WT}) increased numbers and area of collateral branch arteries, thus conferring protection against ischaemic necrosis and enhancing running capacity in ischaemia (Fig. 3a–i and Supplementary Fig. 6a). In myeloid-cell-specific Phd2-null mice (Phd2^{LysCre;lox/lox}), arterialization, ischaemic necrosis and physical endurance were unchanged (Fig. 3a–i and Supplementary Fig. 6a), probably because of the compensatory activity of PHD3, another PHD family member (see below). We also transplanted WT or Phd2^+/− (hereafter HE for ‘heterozygous’) bone marrow (BM) cells into irradiated WT (referred to as WT→WT and HE→WT mice, respectively) or Phd2^+/− (WT→HE and HE→HE mice, respectively) recipients (Supplementary Note 2). Compared to WT→WT, collateral vessel density and area were higher in HE→WT and HE→HE but similar in WT→HE mice, supporting the key role of BM-derived cells in enhancing collateralization and sustaining pre-existing arteries in Phd2^+/− mice (Fig. 3j, k). In accordance, HE→HE and HE→WT, but not WT→HE mice, were protected against ischaemic necrosis (Fig. 3l). The running capacity of HE→WT mice in ischaemia was twice as high as in WT→WT mice (Fig. 3m and Supplementary Fig. 6b). Deletion of one Phd2 allele in haematopoietic cells, endothelial cells or SMCs confirmed that enhanced collateral vessel growth and maintenance were specifically conferred by Phd2 haplodeficiency in BM cells, but not endothelial cells or SMCs (Supplementary Tables 2 and 3). We also assessed the effect of acute deletion of Phd2 in macrophages by transplanting BM cells from tamoxifen-inducible Phd2-haplodeficient mice (Phd2^{Rosa26CreERT;lox/WT}) into WT recipient mice(HE^Rosa26CreERT→WT). Tamoxifen-induced deletion of one Phd2 allele in BM cells increased collateral branches and protected against ischaemic necrosis when compared to vehicle (Fig. 3n–p and Supplementary Note 3). Thus, both chronic and acute deletion of one Phd2 allele in myeloid cells was sufficient to induce pro-arteriogenic macrophages, leading to enhanced collaterogenesis and prevention of ischaemia.

Figure 3. Phd2^+/− macrophages protect against ischaemia by inducing arteriogenesis.

a, b, Quantification of secondary (a) and tertiary (b) collateral branches in WT (Phd2^LysCre;WT/WT; WT/WT), myeloid-cell specific Phd2-haplodeficient (Phd2^{LysCre;lox/WT}; lox/WT) or null mice (Phd2^{LysCre;lox/lox}; lox/lox). c, d, Bismuth⁺ collateral vessel area (c) and density (d) of adductors at baseline. e–h, Necrotic area quantification (e) on haematoxylin- and eosin-stained sections of crural muscles 72 h post-ligation (f–h). Scale bars, 100 µm. i, Treadmill running test 12 h post-ligation. j, k, Bismuth⁺ collateral vessel density (j) and area (k) of non-occluded limbs 5 weeks after bone marrow (BM) transplantation. Phd2^+/− BM in WT and Phd2^+/− recipient mice (HE→WT and HE→HE, respectively); WT BM in WT and Phd2^+/− recipient mice (WT→WT and WT→HE). l, Ischaemic necrosis 72 h post-ligation. m, Treadmill running test 12 h post-ligation in WT→WT and HE→WT mice. n–p, Quantification of secondary (n) and tertiary (o) collateral vessels at baseline and of ischaemic necrosis 72 h post-ligation (p), following tamoxifen-induced deletion of one Phd2 allele in BM cells of HE^Rosa26CreERT→WT mice. All bars show mean±s.e.m. All experiments, n≥4. *, P < 0.05 towards WT/WT and lox/lox in a–i, towards WT→WT in j–m, or towards vehicle in n–p.

PHD oxygen sensors negatively regulate HIF accumulation and NF-κB activity^1,11,18–22. Compared to WT, HIF-1α and HIF-2α protein levels were respectively four and two times higher in Phd2-null macrophages, but unchanged in Phd2-haplodeficient macrophages (Fig. 4a). Conversely, NF-κB activity was increased by 65% in Phd2-haplodeficient macrophages but unaffected in Phd2-null macrophages (Fig. 4b). We proposed that PHD3 might compensate for the complete loss of Phd2 because Phd3 transcripts were 12.2-fold higher in Phd2-null macro-phages (Supplementary Fig. 7). Indeed, Phd3 silencing resulted in a modest induction of NF-κB activity in WT and Phd2-haplodeficient macrophages in contrast to a 70% increase in Phd2-null macrophages compared to their scramble controls (Fig. 4b and Supplementary Note 4). PHD2 hydroxylase function was necessary for NF-κB regulation because ectopic expression of a wild-type PHD2 (PHD2^WT) greatly blunted the activity of NF-κB luciferase induced by Phd2 haplodeficiency, whereas a hydroxylase-deficient PHD2 (PHD2^H313A) had no effect (Fig. 4c). NF-κB activation by TNF-α was still significantly stronger in Phd2^+/− macrophages (Fig. 4d). In contrast, basal and TNF-α-induced NF-κB activity were similar in WT and Phd2^+/− endothelial cells (Supplementary Fig. 8a). When measuring the nuclear accumulation of NF-κB subunits, we found that members of the canonical pathway, that is, p65 (RelA) and p50 (NF-κB1), were more abundant in Phd2^+/− than WT macrophages (Fig. 4e). Silencing of p65 or p50 blocked NF-κB hyperactivation in Phd2^+/− macrophages and the combined knockdown of both subunits restored NF-κB function back to WT levels (Fig. 4f and Supplementary Note 5), thus highlighting the prominent role of NF-κB p65/p50 heterodimers in Phd2^+/− macrophages. To evaluate the involvement of canonical NF-κB signalling in macrophage skewing by Phd2 haplodeficiency, we generated a myeloid-cell specific double transgenic strain, haplodeficient for Phd2 and null for Ikbkb, the gene encoding IKKβ, a positive regulator of canonical NF-κB pathway. Genetic disruption (or pharmacological inhibition) of canonical NF-κB pathway prevented the upregulation of Pdgfb and Sdf1 in cultured Phd2-haplodeficient macrophages (Fig. 4g and Supplementary Fig. 8b). In vivo, gene inactivation of Ikbkb in myeloid cells abolished collateral vessel preconditioning conferred by Phd2 haplodeficiency and greatly prevented ischaemia-induced arteriogenesis in WT mice (Fig. 4h). Thus, skewing of Phd2-haplodeficient macrophages towards a pro-arteriogenic phenotype relied on activation of canonical NF-κB pathway.

Figure 4. Phd2^+/− macrophages trigger arteriogenesis via canonical NFκB pathway.

a, Western blot for HIF-1α and HIF-2α in WT(WT/WT), Phd2-haplodeficient (lox/WT) and Phd2-null (lox/lox) macrophages. Vinculin was used as loading control. Numbers represent densitometric fold change towards WT/WT. b, NF-κB activity in macrophages after Phd3 silencing (siPhd3; n.s., not significant). c, NF-κB activity in Phd2^+/− macrophages upon ectopic expression of wild-type PHD2 (PHD2^WT) or hydroxylase-deficient PHD2 (PHD2^H313A). d, NF-κB activity in macrophages at baseline and upon TNF-α stimulation. e, Western blot for nuclear p65 and p50 in WT and Phd2^+/− macrophages. Numbers represent densitometric fold change towards WT. f, NF-κB activity in macrophages after silencing of p65, p50, or combination of both. g, Pdgfb and Sdf1 expression (quantitative PCR) after genetic inactivation of Ikbkβ in Phd2-haplodeficient pMØ (Phd2^lox/WT, Ikbkb^lox/lox).

h, Quantification of αSMA⁺ collaterals in myeloid-cell-specific Phd2-haplodeficient and Ikbkb-null mice at baseline and 7 days post-ligation. All bars and values show mean ± s.e.m. All experiments, n≥4. *, P < 0.05 towards wt/ wt in a, b, g, h, or towards WT in c–f. #, P < 0.05 towards scramble in b, f, or towards baseline in d, h. $, P < 0.05 towards scramble and either short interfering RNA alone.

Specific macrophage differentiation states have been implicated in the promotion of angiogenesis during cancer and atherosclerosis progression^9,10,13,23. However, little is known of the significance of macrophage heterogeneity in arteriogenesis and its implications on ischaemic diseases. This study identifies a role of macrophage PHD2 in oxygen delivery by regulating arteriogenesis. We show that the phenotype of macrophages induced by reduced PHD2 levels not only favours the formation of new collateral branches, but is also important for collateral vessel homeostasis (Supplementary Note 6). In our model, ischaemia favours the accumulation of M2-like, pro-arteriogenic macrophages that fuel collateral vessel maturation in a NF-κB-dependent manner (Supplementary Fig. 1). Phd2 haplodeficiency unleashes constitutive NF-κB signals that pre-adapt tissue-resident macrophages to ischaemia, accounting for the enhanced arteriogenesis at baseline and thus protection against ischaemic tissue demise. In particular, we show that NF-κB activation in Phd2^+/− macrophages increases the production of SDF1 and PDGFB, which are concurrently required to prime SMC migration and growth in vitro. The relevance of SDF1 and PDGFB in vivo remains to be established, and other soluble factors may also contribute.

Repression of canonical NF-κB pathway in macrophages promotes upregulation of M1-type and downregulation of M2-type genes^24–26. PHDs negatively regulate NF-κBthrough either hydroxylase-dependent or -independent inactivation of IKKβ in different cell contexts^19–22. We show here that Phd2 haplodeficiency results in hyperactivation of canonical NF-κB pathway in macrophages and promotion of the M2-phenotype via accumulation of both p65 and p50 subunits, and that this regulation requires PHD hydroxylase function.

In summary, our study provides an insight on howthePHD2 oxygen sensor regulates arteriogenesis by modulating macrophage phenotype. The mechanism upstream to arteriogenic PHD2 downregulation in M2-likemacrophages remains to be established. Nonetheless, our findings support the rationale for therapeutic inhibition ofPHD2. Previous studies showed that unspecific inhibitors of PHD2 or silencing of PHD2 promote therapeutic revascularization against ischaemia^2–5. However, angiogenesis is a late response and organ function might be compromised before new blood vessel formation is achieved. In contrast, arteriogenesis takes place on pre-existing vascular shunts and our data suggest that eitherPHD2 inhibitors or cell-therapy-based strategies using PHD2 hypomorphic macrophages might be exploited as preventive medicine to promote collateral vascularization in patients at risk of limb or heart ischaemia, such as diabetic or hyper-cholesterolemic patients.

METHODS

Animals

129/S6 or Balb/c, female and male, WT and Phd2^+/− mice (8–12 weeks old) were obtained from our mouse facility. Phd2^+/− and Phd2 conditional knockout mice were obtained as previously described¹. VE-Cadherin:CreERT and PDGFRB:Cre transgenic mice were generated by R. Adams at the Cancer Research UK^29,30. Ikbkb conditional knockout mice were obtained from M. Karin³¹. Tie2:Cre and Rosa26:CreERT transgenic mice were purchased from the Jackson Laboratory. Housing and all experimental animal procedures were performed in accordance with Belgian law on animal care and were approved by the Institutional Animal Care and Research Advisory Committee of the K. U. Leuven (P036/2009).

Mouse model of hindlimb ischaemia and myocardial infarction

To induce hindlimb ischaemia, unilateral or bilateral ligations of the femoral artery and vein (proximal to the popliteal artery) and the cutaneous vessels branching from the caudal femoral artery side branch were performed without damaging the nervus femoralis. By using this procedure, collateral flow to adductor muscles is preserved via arterioles branching from the femoral artery, therefore 50% up to 60% of perfusion is preserved by this method. Two superficial pre-existing collateral arterioles, connecting the femoral and saphenous artery, were used for analysis. Functional perfusion measurements of the collateral region were performed using a Lisca PIM II camera (Gambro). Gelatin-bismuth-based angiography was performed as described before and analysed by photoangiographs (Nikon D1 digital camera). Collateral side branches were categorized as follows: secondary collateral arterioles directly branched from the primary collateral, and tertiary arterioles orientated perpendicularly to the secondary branches. The number of secondary and tertiary collateral arterioles was counted. After perfusion-fixation, the muscle tissue between the two superficial collateral arterioles (adductor) was post-fixed in 2% paraformaldehyde, paraffin-embedded and morphometrically analysed²⁷. An endurance treadmill-running test was performed at baseline and 12 h post-bilateral-ligation. Myocardial infarction was induced by permanent ligation of the left anterior descending coronary artery as previously described³². Briefly, the left thorax of anesthetized mice was opened in the fourth intercostal space and all the muscles overlying the intercostal region were dissected. The main left anterior descending coronary artery was ligated proximal to the main bifurcation though a small incision of the pericardium. Discoloration of the ventricle after blood flow restriction was used as readout of a successful surgical procedure. Gelatin-bismuth-based angiography was performed 24 h post-ligation and the entire heart was fixed in 2% paraformaldehyde.

Oxymetry

Oxygen tension (pO₂) in the lower limb was measured using ¹⁹F-MRI oxymetry in non-ligated and ligated legs 12 h after femoral artery ligation. The oxygen reporter probe hexafluorobenzene was injected directly into the crural muscle. Magnetic resonance imaging was performed with a 4.7 T (200MHz, ¹H), 40-cm inner diameter bore system (Bruker Biospec). A tunable ¹H/¹⁹F surface coil was used for radiofrequency transmission and reception³³.

Histology, immunostaining, and morphometry

Adductor, crural muscles and hearts were dissected, fixed in 2% paraformaldehyde, dehydrated, embedded in paraffin, and sectioned at 7-µm thickness. Area of necrotic tissues in the crural muscle was analysed by haematoxylin and eosin staining. Necrotic cells display a more glassy homogeneous appearance in the cytoplasm with increased eosinophilia, whereas the nuclear changes are reflected by karyolysis, pyknosis and karyorrhexis. Necrotic area was defined as the percentage of area which includes these necrotic myocytes, inflammatory cells and interstitial cells, compared to the total soleus area. Infarct size was measured in desmin-stained hearts 24 h after ischaemia as previously described³⁴. After deparaffinization and rehydration, sections were blocked and incubated overnight with primary antibodies: rat anti-CD31, dilution 1/500 (BD-pharmingen); mouse anti-aSMA, dilution 1/500 (Dako); rat anti-F4/80, dilution 1/100 (Serotec); dilution 1/50 (BD-pharmingen); rat anti-CD45, dilution 1/100 (BD-pharmingen); goat anti-MRC1, dilution 1/200 (R&D); rabbit anti-desmin dilution 1/150 (Cappel). To analyse capillary density and area, images of anti-CD31-stained sections of the entire soleus were taken at ×40. To measure bismuth-positive vessel density and area, haematoxylin- and eosin-stained paraffin sections were analysed and vessels filled with bismuth-gelatin (black spots)were taken in account. Images of the entire soleus were acquired at ×20 for this analysis. The values in the graph represent the averages of the mean vessel density and area per soleus muscle. The same method was used to quantify vessel capillaries and collateral branches in cardiac tissues. Density and area were measured by using a KS300 (Leica) software analysis. Hypoxic cells were analysed 2 h after injection of 60 mg kg⁻¹ pimonidazole into operated mice. Mice were killed and muscles were collected. Paraffin sections were stained with Hypoxiprobe-1-Mab-1 (Hypoxiprobe kit; Chemicon International) following the manufacturer’s instructions. Oxidative stress and proliferation rate were assessed on 7-µm thick cryosections by using the goat anti-8-OHdG antibody, dilution 1/200 (Serotec) and the rat anti-BrdU antibody, dilution 1/300 (Serotec). Sections were subsequently incubated with appropriate secondary antibodies, developed with fluorescent dies or 3,3′-disminobenzidine (DAB, Sigma). Whole-muscle viability was assessed on unfixed 2-mm thick tissue slices by staining with 2,3,5-triphenyltetrazolium chloride (TTC). Viable area, stained in red, was traced and analysed. Pictures for morphometric analysis were taken using a Retiga EXi camera (Q Imaging) connected to a Nikon E800 microscope or a Zeiss Axio Imager connected to an Axiocam MRc5 camera (Zeiss) and analysis was performed using KS300 (Leica). Angiograms were obtained by X-ray and microcomputed tomography angiographies of hearts and legs at baseline.

Macrophage preparation

To harvest peritoneal macrophages (pMØ), the peritoneal cavity was washed with 5 ml of RPMI 10% FBS. The pooled cells were then seeded in RPMI 10% FBS in 6-well plates (2 × 10⁶ cells per well), 12-well plates (1 × 10⁶ cells per well), or 24-well plates (5 × 10⁵ cells per well). After 6 h of incubation at 37 °C in a moist atmosphere of 5% CO₂ and 95% air, non-adhering cells on each plate were removed by rinsing with phosphate-buffered saline (PBS). The attached macrophages were grown in different media for 12 h or 48 h depending on the experiments performed, as described below. When high amounts of cells were needed (analysis for HIF accumulation and NF-κB activity), macrophages were derived from bone marrow (BM) precursors as described before²⁸. Briefly, BM cells (2 × 10⁶ cells per ml) were grown in a volume of 5 ml in a 10-cm Petri dish (non tissue culture treated, bacterial grade) for 7 to 10 days in DMEM supplemented with 20% FBS and 30% L929 conditioned medium as a source of M-CSF. The cells obtained in those cultures are uniformly macrophages.

Quantitative PCR analysis

In order to investigate gene expression in pMØ, quantitative real-time PCR (qPCR) was performed. After preparing pMØ, the cells were cultured in normoxic condition for 12 h and RNA was extracted. To analyse the expression levels of chemoattractants in the adductor, tissues were collected at baseline or 24 h post-ischaemia and RNA was extracted. Macrophages and endothelial cells were freshly sorted from dissected adductors as described below and RNA was extracted. Quantitative PCR was performed with commercially available or home-made primers and probes for the studied genes. The assay ID (Applied Biosystems) or the sequence of primers and probes (when custom-made) are listed in Supplementary Table 4. RNA levels of Sdf1 and Pdgfb after inhibition of NF-κB pathway were measured by qPCR on pMØ exposed for 12 h to 500nM 6-amino-4-(4-phenoxyphenylethylamino) quinazoline.

Protein extraction and immunoblot

Protein extraction was performed using 8M urea buffer (10% glycerol, 1% SDS, 5 mM DTT, 10 mM Tris-HCl, pH6.8) as previously described¹. Nuclear proteins were extracted in 1% SDS buffer upon cytoplasmic separation by using a hypotonic lysis buffer (10 mM KCl, 10 mM EDTA, 0.5% NP40, 10 mM HEPES, pH8, supplemented with phosphatase and protease inhibitors, from Roche). Signal was detected using the ECL system (Invitrogen) according to the manufacturer’s instructions. The following antibodies were used: rabbit anti-HIF-1α (Novus), rabbit anti-p105/50, rabbit anti-HIF-2α (Abcam), PM9 rabbit anti-HIF-2α (from P.M.), mouse anti-vinculin (Sigma), rabbit anti-lamin A/C, rabbit anti-p65 (Cell Signaling). Densitometric analysis was performed using ImageJ 1.44 (http://rsbweb.nih.gov/ij/).

Transduction and transfection of BM-derived macrophages and lung endothelial cells

To express an inducible NF-κB-responsive firefly luciferase reporter, commercially available lentiviral vectors (LV) were used (Cignal Lenti NF-κB Reporter; SABiosciences). BM-derived macrophages (2.5 × 10⁵) and 10⁵ primary lung endothelial cells, isolated as described before¹, were seeded in a 24-well plate in DMEM 10% FBS or M199 20% FBS for 8 h. Cells were transduced with 10⁸ transducing units per ml. Eight hours after transduction, the medium was replaced. After 48 h, cells were stimulated with TNF-α (20 ng ml⁻¹) for 8 h and the same amount of protein extract was read in a luminometer. For PHD3 silencing, siRNA oligonucleotides were designed using the Invitrogen online siRNA design tool (http://rnaidesigner.invitrogen.com). The following siRNA sequences (sense strands) were used. For Egln3/Phd3(NM_028133.2): 5′-GCCGGCTGGGCAAAT ACTATGTCA-3′; for the scramble: 5′-CACCGCTTAACCCGTATTGCCTAT-3′. In brief, one day after the transduction of macrophages with LV, cells were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. Preparation of the oligonucleotide–Lipofectamine 2000 complexes was done as follows: 25 pmol siRNA oligonucleotide (stock 20 µM) was diluted in 50 µlOpti-MEMI reduced serum medium. Lipofectamine 2000 (1.5 µl) was diluted in 50 µl Opti-MEM I reduced serum medium and incubated for 5min at room temperature. siRNA oligonucleotides were gently mixed with Lipofectamine 2000 and allowed to incubate at room temperature for 20 min to form complexes. Just before transfection, the cell culture medium was removed and cells were rinsed twice with serum-free Opti-MEM I medium. The Lipofectamine 2000-siRNA oligonucleotide complexes were added to each well in 400 µl of serum-free Opti-MEM medium for 5 h. Afterwards, cells were incubated in complete medium for 48 h at 37 °C in a CO₂ incubator and assayed for gene knockdown (qPCR) and luciferase activity. To assess if the increased NF-κB activity observed in Phd2^+/− macrophages was dependent on the hydroxylase activity of PHD2, 48 h before transduction, 4 × 10⁶ BM-derived macrophages were resuspended in 240 µl of Opti-MEM and were electroporated (250 V, 950 µF, ∞Ω) with 7µg of plasmids expressing a wild-type PHD2 (PHD2^wt) or a PHD2 containing a mutation at the catalytic site (PHD2^H313A). Silencing of the canonical pathway subunits p65 (Rela) and p50 (Nfkb1) was achieved by electroporation with specific siRNAs. Briefly, 48 h before transduction, 2.4 × 10⁶ BM-derived macrophages were resuspended in 320 µl of Opti-MEM and were electroporated (250 V, 950 µF, ∞Ω) with 60 pmol of siRNA for either scramble, p65, p50, or combination of p50 and p65. For higher efficiency of silencing, two different siRNA sequences for each respective gene were designed (http://rnaidesigner.invitrogen.com). For p65 (NM_009045.4): 5′-TGTCTGCACC TGTTCCAAATT-3′ and 5′-TGCTGATGGAGTACCCTGATT-3′; for p50 (NM_008689.2): 5′-GAATACTTCATGTGACTAATT-3′ and 5′-CAAAGGTTAT CGTTCAGTTTT-3′; for the scramble:5′-CACCGCTTAACCCGTATTGCCTAT-3′.

Cell migration and viability assays

Migration and proliferation of smooth muscle cells (SMCs) and endothelial cells were assessed by using 8-µm-pore Transwell permeable plate for migration assays and 0.4-µm-pore Transwell permeable plate for proliferation assays (Corning Life Science). To determine cell migration and proliferation in response to soluble factors secreted by pMØ, pMØ were precultured in the lower chamber for 12 h in RPMI 1% FBS or in M-199 1% FBS (migration assay), or 48 h in DMEM-F12 1% FBS or in M-199 1% FBS (proliferation assay). For migration assays, hCASMCs (human coronary artery SMCs; from Lonza) and HUVECs (human umbilical vein endothelial cells; from Lonza) were starved for 12 h in their own medium at 1% FBS and then seeded in the upper chamber (5 × 10³ cells in 200 µl of medium 1% FBS). SMCs and HUVECs were incubated for 2 days or 24 h, respectively, and migrated cells were fixed with 4% paraformaldehyde, stained with 0.25% crystal violet/50% methanol and counted under the microscope. For cell growth assays, RAOSMCs (rat aortic SMCs) and HUVECs were seeded on the upper chambers (5 × 10³ cells per transwell) and cultivated with pMØ for 24 h in DMEM-F12 1% FBS or M-199 1% FBS for RAOSMCs and HUVECs, respectively. The cell proliferative ability was then analysed using WST-1 Cell Proliferation Assay (Roche Applied Biosciences) according to the manufacturer’s instructions. Alternatively, WT and Phd2^+/− pMØ were seeded in the lower chamber of a Transwell and transduced with lentiviral vectors (10⁸ TU/ml; Sigma) carrying an shRNA against Sdf1 (NM_013655.4), Pdgfb (NM_011057.3), or a scramble control (purchased from Sigma; TRCN0000178772 for Sdf1: 5′-CCGGCTGAAGAACAACAACAGACAA CTCGAGTTGTCTGTTGTTGTTCTTCAGTTTTTTG-3′; TRCN0000042529 for Pdgfb: 5′-CCGGGAGTCGAGTTGGAAAGCTCATCTCGAGATGAGCT TTCCAACTCGACTCTTTTTG-3′; SHC002V for the scramble: 5′-CCGGCA ACAAGATGAAGAGCACCAACTCGAGTTGGTGCTCTTCATCTTGTTGTT TTT-3′). Sixty hours after macrophage transduction, SMC migration or growth assays were performed by seeding the SMCs in the upper side of the Transwell as above.

SMC differentiation assay

pMØwere seeded in a 24-well plate withDMEMF-12 5% FBS. Conditioned medium was collected after 2 days and supplemented with 25mMHEPES. RAOSMCs were seeded in a 24-well plate (80 × 10³ cells per well) and incubated for 5 h at 37 °C in a moist atmosphere of 5%CO₂ and 95% air. After 2 h of starvation in DMEM-F12 1% FBS, SMCs were stimulated with conditioned medium from WT and Phd2^+/− pMØ. After 24 h, differentiation status of the SMCs was assessed by qPCR.

Macrophage and endothelial cell sorting

For cell sorting of adductor macrophages and endothelial cells, the adductors were dissected, dissociated mechanically, and digested using collagenase I for 45min at 37 °C. For macrophage sorting, the digested cell suspension was incubated for 15min with mouse anti-CD16/CD32 mAb (Fc Block, BD-pharmingen) and stained with rat FITC-conjugated anti-F4/80 antibody (Serotec) for 20min at 4 °C. CD31⁺CD45⁻ endothelial cells were sorted from the digested adductor cell suspension after incubation with rat APC-conjugated anti-CD31 and rat FITC-conjugated anti-CD45 (BD-pharmingen) for 20 min at 4 °C.

BM transplantation and haematological analysis

Balb/c WT and Phd2^+/− recipient mice were irradiated with 7.5 Gy. Subsequently, 5×10⁶ BM cells from green fluorescent protein⁺ (GFP⁺) WT or GFP⁺ Phd2^+/− mice were injected intravenously via the tail vein. Femoral artery ligation, treadmill running test and bismuth angiography were performed 5 weeks after BM reconstitution. Red and white blood cell count was determined using a haemocytometer (Cell-Dyn 3700, Abbott) on peripheral blood collected in heparin by retro-orbital bleeding. To assess the effect on arteriogenesis of acute deletion of one Phd2 allele in macrophages, 5 × 10⁶ BM cells from Phd2^{Rosa26CreERT;lox/WT} mice were transplanted into lethally irradiated WT recipient mice. After 5 weeks, transplanted mice were injected intraperitoneally with tamoxifen (1 mg per mouse; Sigma) or vehicle for 5 days. Femoral artery ligation was performed 10 days after tamoxifen treatment as above.

Statistics

The data were represented as mean±s.e.m. of the indicated number of measurements. Statistical significance was calculated by two-tailed unpaired t-test for two data sets and ANOVA followed by Bonferroni post hoc test for multiple data sets using Prism (GraphPad Inc.), with P < 0.05 considered statistically significant.