Principle of the Device

This device uses a CO₂ gas–permeable membrane similar to the principle of an artificial lung on the extracorporeal circulatory system. The unit consists of 15 000 multilayered composite-membrane hollow fibers with porous membrane sandwiching on both sides of gas-permeable membrane (Mitsubishi-Leiyon) and is capable of instantly converting 20 L/min of water (pH 7.0) into CO₂-enriched water (free CO₂ concentration, 1000 to 1200 mg/L, pH 5.0).

Mouse Model of Unilateral Hindlimb Ischemia and CO₂ Bathing

Unilateral hindlimb ischemia was induced by resecting the right femoral arteries (including muscle branches) and veins of 8-week-old male C57BL/J mice under anesthesia with sodium pentobarbital (50 mg/kg IP).^11,13 To inhibit NOS chronically, the mice were provided water containing 1 mg/mL N^G-nitro-l-arginine methyl ester (L-NAME) for 4 weeks.¹¹ Because CO₂ bathing immediately after operation delayed the closure of this skin wound, we started the CO₂ bathing of the lower limb from 4 days after surgery. Lower limbs of mice were immersed into CO₂-enriched water for 10 minutes or freshwater (control) at 37°C once a day under anesthesia (n=10 in each group). The Institutional Animal Care and Use Committee of our university approved all animal protocols.

Immunohistochemistry

Four pieces of ischemic tissues from the adductor and semimembranous muscles were obtained 28 days after the surgery of hindlimb ischemia. Frozen sections were stained with anti-factor VIII, followed by incubation with TRIC-conjugated secondary antibody. Five fields from 2 muscle samples of each animal were randomly selected for capillary counts. To ensure that capillary densities were not overestimated as a consequence of myocyte atrophy or underestimated because of interstitial edema, the capillary/muscle fiber ratio was determined.^11,13 To examine whether cells survived in the tissues, adjacent sections were subjected to alkaline phosphatase staining by the indoxyl-tetrazolium method. Alkaline phosphatase staining turns capillary endothelial cells a dark blue color only when they are viable and when the intracellular enzyme activity remains intact.^11,13

Laser Doppler Analysis and Angiography

We measured the ratio of the ischemic (right)/normal (left) limb blood flow by use of a laser Doppler perfusion image (LDPI) analyzer (Moor Instruments). After blood flow had been scanned twice, stored images were subjected to computer-assisted quantification of blood flow, and average flows of the ischemic and nonischemic limbs were calculated. To minimize data variables caused by ambient light and temperature, the LDPI index was expressed as the ratio of ischemic (left) to nonischemic (right) limb blood flow.^11,13

Vessel density was evaluated with a microfocus x-ray television device (Hitex Co Ltd) 28 days after ischemia (n=5). Longitudinal laparotomy was performed to introduce a catheter into the abdominal aorta, followed by injection of contrast medium (lipiodol). Angiography was performed for 2 seconds after the injection. We quantitatively analyzed collateral vessel numbers as previously reported.^11,13 Briefly, numbers of vessels in the thigh area were counted by use of 5-mm² grids by 2 radiologists who were unaware of the group identity of the angiographic film. Interobserver variation was <5%.

cGMP Assay and Measurement of Blood pH Level

The assay for tissue cGMP was performed by use of the cGMP enzyme immunoassay system (Biotrak; Amersham) as previously described.¹¹ The tissues remaining after cGMP measurement were digested by use of a bicinchoninic acid protein assay kit (Pierce). Blood pH levels were measured by automated blood gas analyzer (ABL505, Radiometer A/S).

Northern and Western Blotting and Plasma VEGF Measurement

Frozen skeletal samples from hindlimbs or forelimbs were homogenized in Trizol reagent (Gibco BRL). Blots were hybridized with a random-primed ³²P-labeled cDNA probe for VEGF¹¹ and normalized by densities for GAPDH as an internal control. Hybridized signals were measured by scanning densitometry, and VEGF mRNA levels were arbitrarily normalized relative to the GAPDH mRNA levels.

Phosphorylation of eNOS (serine 1177) was analyzed by Western blotting using phospho-specific antibodies (New England Biolabs). The muscles were homogenized in lysis buffer. Lysates were immunoblotted with anti-phospho antibodies and detected with an enhanced chemiluminescence kit (Amersham).¹¹ Plasma VEGF concentration was measured by use of the ELISA kit (R&D Systems).

FACS Staining

Total nuclear cells in the peripheral blood were isolated by erythrocyte lysis with ammonium chloride solution (PharM Lyse, Becton Dickinson). Lin−/Flk+ cells were isolated by PE-labeled lineage antigens (CD11b, CD3, B220, Ter-199, Gr-1, CD4, CD8e, CD16/32), FITC-CD34, and biotin-Flk-1 and then analyzed by use of a FACScan flow cytometer.^14,15 Lin−/Flk+/endoglin+ cells were isolated by FITC-labeled lineage antigens, PE–Flk-1 and biotin-endoglin. To prove the specificity of anti-CD34 antibody, the biotin-labeled anti-mouse CD34 antibody used in this study was reacted with mouse bone marrow cells and purified with streptavidin-magnet beads, followed by fluorescence-activated cell sorter (FACS) analysis using streptavidin-PE. All anti-mouse antibodies were purchased from BD Biosciences.

Differentiation of Lin−/Flk-1+ Cells Into Endothelial Cells In Vitro

The population of Lin−/Flk-1+ cells was isolated with FACS from the peripheral blood of the mice that had undergone the limb ischemic operation and then treated with CO₂ bathing for 14 days. These cells were cultured on fibronectin-coated plastic dishes in DMEM supplemented with 100 ng/mL VEGF and 10% FBS. After 4 days, DiI-labeled acetylated LDL (Biomedical Technologies Inc) was added into medium at 2 μg/mL for 6 hours, fixed with 4% paraformaldehyde, and stained with anti–VE-cadherin antibody and FITC-labeled anti-IgG antibody.

Statistics

Statistical analyses were performed by 1-way ANOVA followed by pairwise contrasts using Dunnett’s test. Data (mean±SEM) were considered significant at a value of P<0.05.

Laser Doppler Blood Perfusion

Progressive recovery of limb perfusion was disclosed in CO₂-bathed and control freshwater-bathed mice after induction of limb ischemia. A greater degree of blood perfusion recovery was observed in the ischemic limbs of CO₂-bathed mice compared with controls (38% increase at day 28, P<0.001) (Figure 1, A and B). Inhibition of NOS activity by L-NAME administration abolished an enhancement of blood flow recovery by CO₂ bathing and reversed the recovery ratio toward the control level. Blood flow in L-NAME–treated mice tended to be lower than that in wild-type mice, but this difference was not significant (Figure 1B).

 

Figure 1. LDPI. A, Greater blood perfusion (red to yellow) was observed in CO₂-enriched water-bathed limbs, in contrast to reduced perfusion (green to blue) in freshwater-bathed ischemic limbs (control). B, Computer-assisted analyses of LDPI revealed significantly greater blood perfusion values in CO₂-enriched water-bathed group than in control group. Administration of L-NAME (1 mg/mL) in drinking water reduced increased perfusion by CO₂-enriched water bathing toward normal level. Values shown are mean±SEM (n=10) at each time point. *P<0.001 vs control mice.

 

Angiography

All animals were subjected to iliac angiography using contrast medium (lipiodol) on postoperative day 28. Collateral vessel numbers were markedly increased in ischemic limbs of CO₂-bathed mice (4.1±0.4-fold at day 28, P<0.001, n=5) compared with those in water-bathed mice (Figure 2).

 

Figure 2. Angiographic analysis. Representative angiograms were obtained on postoperative day 28. Arrows indicate ligated ends of femoral arteries. Collateral vessel numbers counted by use of 5-mm² grids were markedly increased in ischemic limbs of CO₂-bathed mice (4.1±0.4-fold at day 28, P<0.001, n=5) compared with those in water-bathed mice.

 

Analysis of Capillary Density

Immunohistochemical staining for anti-factor VIII revealed the presence of capillary endothelial cells (Figure 3A). The capillary/muscle fiber ratio in the skeletal muscle obtained 28 days after hindlimb ischemia was significantly increased in the CO₂-bathed mice (3.7-fold, P<0.001) compared with that in water-bathed mice. A similar increase (4.2-fold increase, P<0.001) was also observed in ALP staining for detection of viable endothelial cells (Figure 3B). Administration of L-NAME (1 mg/mL) in drinking water reduced the increased vessel numbers by CO₂ bathing toward the normal level (Figure 3).

 

Figure 3. Immunohistochemical analysis. A, Ischemic tissues from adductor and semimembranous muscles were obtained 28 days after surgery of hindlimb ischemia. Endothelial cells were stained with anti-factor VIII antibody, followed by incubation with TRIC-conjugated secondary antisera. B, Alkaline phosphatase staining turns viable endothelial cells blue. Five fields from 2 muscle samples of each animal (n=10) were randomly selected, and capillary density was shown as capillary/muscle fiber ratio. Administration of L-NAME (1 mg/mL) in drinking water reduced increased vessel numbers by CO₂ bathing toward control levels of freshwater-bathed ischemic limbs (n=10). *P<0.001 vs control mice. Bars=50 μm.

 

Induction of VEGF Expression, eNOS Phosphorylation, and cGMP Levels

VEGF mRNA levels were examined in hindlimb muscles dissected at days 0 (before), 1, 2, 7, 14, and 21. VEGF mRNA levels were decreased immediately after hindlimb ischemia (day 1, day 2), and then gradually reverted to the basal levels at day 7 in the control group. In the CO₂-enriched water group, a marked increase in VEGF mRNA levels was observed at day 7 (1.6-fold versus day 0 preischemic levels, P<0.01) and showed a peak level at ≈day 14 (2.7-fold versus day 0, P<0.001). Induction of the VEGF mRNA from the preischemic level was significantly higher in the CO₂ bathing group than the increase in the control group (Figure 4), whereas the increase in VEGF mRNA synthesis by CO₂ bathing was not affected by L-NAME treatment (Figure 5A).

 

Figure 4. VEGF mRNA expression in ischemic hindlimb. Hindlimb skeletal muscles were dissected after ischemia, and RNA was extracted. Densities of VEGF mRNA signals were measured by densitometry and normalized relative to those of GAPDH mRNA signals. Results (mean±SEM, n=6) were arbitrarily indicated as values relative to VEGF mRNA levels at day 0. *P<0.001 vs day 0 preischemic levels.

 

 

Figure 5. Effect of L-NAME on VEGF mRNA expression in hindlimb and forelimb skeletal muscles after hindlimb ischemia and plasma VEGF concentrations. A, Mice with hindlimb ischemia were provided water containing 1 mg/mL L-NAME, and time-dependent VEGF mRNA expression was measured as described in Figure 4. Forelimb skeletal muscles were dissected from hindlimb ischemia mice, and VEGF mRNA was analyzed. Results (mean±SEM, n=5) were arbitrarily indicated as values relative to VEGF mRNA levels at day 0. *P<0.001 vs day 0 preischemic levels, B, Plasma VEGF concentrations were measured by ELISA (control, CO₂ bathing of normal mice; ischemia, freshwater-bathed ischemic mice; ischemia+CO₂ water, CO₂-bathed ischemic mice, n=5 each). *P<0.05 vs day 0.

 

To define whether the effect of CO₂ bathing results from systemic or local VEGF synthesis, we examined the time-dependent VEGF mRNA induction in forelimb skeletal muscles after CO₂ bathing and changes in plasma VEGF levels. The mRNA levels in forelimb skeletal muscles of CO₂-bathed ischemic mice did not change significantly after CO₂ bathing compared with the preischemic levels (Figure 5A). Although only induction of hindlimb ischemia did not affect plasma VEGF levels, VEGF levels in the CO₂-bathed ischemic mice were slightly but significantly elevated at day 14 (18%, P<0.05, n=6) compared with the water-immersed ischemic mice (n=6) (Figure 5B). Furthermore, we determined plasma pH levels to study whether CO₂ bathing–mediated effects are systemic. We found that CO₂ bathing of ischemic lower limbs did not significantly affect the pH levels in the peripheral blood (control, 7.2±0.04; CO₂ bathing, 7.2±0.03 at day 14; n=6 each). These findings demonstrate that VEGF synthesis by CO₂ bathing is induced only locally and that this increase in local VEGF synthesis leads to the elevation in plasma VEGF levels.

Skeletal muscles at day 14 (in which VEGF expression is maximally increased) were dissected, and eNOS phosphorylation and cGMP levels were examined. The eNOS phosphorylation levels at day 14 (normalized with expression levels of eNOS protein) were increased significantly in both control and CO₂-enriched water groups relative to the day 0 preischemic levels (1.6- and 2.4-fold, respectively) (Figure 6A). eNOS phosphorylation levels in the CO₂-enriched water group were significantly higher than those in the control group (P<0.001 versus the control group). Consistent with eNOS phosphorylation, cGMP levels in skeletal muscles at day 14 were also significantly higher (3.4-fold, P<0.001) in the CO₂-enriched water group compared with those in the control group (Figure 6B).

 

Figure 6. cGMP accumulation in ischemic limbs. A, Skeletal muscles were homogenized and immunoblotted with anti-phospho antibodies for eNOS. Phospho signals in filters were stripped and reprobed with anti-eNOS antibody. Phospho-eNOS densities were measured by densitometry and normalized relative to those of eNOS signals. Results are arbitrarily indicated as values relative to signal densities in day 0 preischemic control. Results shown are mean±SEM (n=6), and representative data are shown. *P<0.001 vs values in day 0 preischemic control. B, Skeletal muscles were dissected at days 7, 14, and 21 after hindlimb ischemia, and tissue cGMP levels were measured. Results shown are mean±SEM (n=6 each). *P<0.05, **P<0.001 vs day 0 preischemic control.

 

Effect of CO₂ Bathing on Circulating Endothelial-Lineage Progenitor Cells

CD34+/AC133+/Flk-1+ hematopoietic stem cells circulate in the peripheral blood of humans as an endothelial precursor cell and play a critical role in neovascularization in ischemic tissue.¹⁶ Because AC133 marker is not available for mice, we isolated hematopoietic lineage-negative (Lin−) cells from the peripheral blood and then analyzed the CD34- and endothelial markers Flk-1– and endoglin-positive population to study whether endothelial-lineage precursor cells are mobilized by CO₂-enriched water bathing. FACS analysis indicated that Lin−/Flk-1+ cells are barely detected in the peripheral blood of normal mice (0.01±0.002% of total nuclear cells, n=12). Lin−/Flk-1+ cells were significantly increased after limb ischemia and showed a peak value at day 14 (≈7-fold versus the preischemic value) (from 0.01±0.002% to 0.073±0.002%, P<0.001, n=7). Interestingly, such a Lin−/Flk-1+ population was further increased by CO₂ bathing and showed a peak value at day 14 (≈24-fold increase, from 0.01±0.002% to 0.24±0.03%, P<0.001, n=7) (Figure 7A). The Lin−/Flk-1+ cells mobilized by CO₂ bathing were mostly positive for anti-endoglin antibody and in the CD34-negative fraction (Figure 7B). Considering that CD34−/Flk-1+ cells rather than CD34+/Flk-1+ are reported to be a real population of hematopoietic stem cells,¹⁷ our present data indicate that CO₂-enriched water bathing mobilizes very immature hematopoietic stem cells, including endothelial progenitor cells. To prove the specificity of anti-CD34 antibody, CD34+ cells were enriched by anti-mouse CD34 antibody from mouse bone marrow cells. As shown in Figure 7C, 83% purity of CD34-positive cells was detected by FACS, indicating that the staining for the CD34 antigen was properly performed.

 

Figure 7. FACS analysis of circulating endothelial-lineage progenitor cells. A and B, Lin−/Flk+ cells in peripheral blood nuclear cells were isolated by PE-labeled lineage antigens, FITC-CD34 and biotin–Flk-1, and then analyzed. Lin−/Flk+/endoglin+ cells were isolated by FITC-labeled lineage antigens, PE-Flk-1 and biotin-endoglin. Cell number ratio of Lin−/Flk+ cells to total nuclear cells is shown (n=6 each). *P<0.05, **P<0.001 vs day 0 preischemic control. C, To prove specificity of anti-CD34 antibody, CD34+ cells in mouse bone marrow cells were enriched by biotin-labeled anti-mouse CD34 antibody and purified with streptavidin-magnet beads, followed by FACS analysis using streptavidin-PE. Rat anti-mouse IgG was used as a negative control. D, Lin−/Flk+ cells were cultured on fibronectin-coated plastic dishes in DMEM supplemented with 100 ng/mL VEGF and 10% FCS. After 4 days, DiI-labeled acetylated LDL was added into medium at 2 μg/mL for 6 hours, fixed with 4% paraformaldehyde, and stained with anti–VE-cadherin antibody and FITC-labeled anti-IgG antibody.

 

We further examined whether Lin−/Flk-1+ cells express another endothelial marker, VE-cadherin, and also possess the characteristics of endothelial cells, such as acetylated LDL uptake. The Lin−/Flk-1+ cell population, mobilized by CO₂ bathing, was isolated and cultured with 100 nmol/L VEGF-supplemented medium for 4 days. Approximately 26±1.2% (n=12) of the Lin−/Flk-1+ cells adhered onto the fibronectin-coated plastic dishes. Approximately 74±2.3% (n=12) of the attaching cells showed the ability to incorporate the DiI-labeled acetylated LDL, and these cells expressed the VE-cadherin (Figure 7D).