A Method for the Assessment of Hypoxia in the Arterial Wall, With Potential Application In Vivo
Abstract According to the anoxemia theory of atherosclerosis, an imbalance between the demand for and supply of oxygen in the arterial wall is a key factor in the development of atherosclerotic lesions. Direct in vitro and in situ measurements have shown that Po2 is decreased in the inner part of the media, but the degree of hypoxia in vivo or the distribution of hypoxia along the arterial tree is not known. We applied a hypoxia marker, 7-(4′-(2-nitroimidazol-1-yl)-butyl)-theophylline (NITP), to develop a method for the detection of hypoxia in the arterial wall. Immunoperoxidase and immunofluorescence were used to detect the marker, and a clearly Po2-dependent staining was observed in the media of rabbit and swine aorta in vitro. The cutoff Po2 level was probably around 2 to 3 mm Hg. In experimental atherosclerotic lesions in the rabbit the marker seemed to bind to foam cells that were already at a higher surrounding Po2, which might reflect a higher local oxygen consumption. The binding of the marker to endothelial cells was not Po2 dependent. One explanation for this finding could be that the marker was metabolized via a non–oxygen-dependent pathway in these cells. We propose that this method may be used to assess arterial wall hypoxia in vivo. Furthermore, the spatial resolution allows the detection of local variations within the arterial tree.
- Received September 21, 1995.
- Accepted October 3, 1995.
The need for oxygen in the arterial wall is satisfied via diffusion from the luminal blood flow and the vasa vasora in the adventitia. The relation between the oxygen demand and the available diffusion capacity will be decisive for the local energy balance. During atherogenesis the thickness of the arterial wall increases, leading to an impaired diffusion capacity. Furthermore, experimental data indicate that the oxygen consumption of the arterial wall increases during atherogenesis.1 2 It is possible that an energy imbalance may arise that would lead to local metabolic disturbances that may endanger regression or even result in progression of the atherosclerotic process with the formation of a necrotic core. Based on this, Hueper3 formulated the anoxemia theory of atherosclerosis, which was later explored by several groups.4 5 6
The oxygen concentration in the arterial wall has been measured directly in animals in in vitro and in situ systems,7 8 9 10 and these measurements demonstrate that Po2 decreases with the distance from the lumen and adventitia, reaching a minimum of 20 to 50 mm Hg in the media. Although one may question whether such a modest reduction of Po2 may influence arterial wall metabolism in a decisive way, it should be kept in mind that these measurements were made on arteries much thinner than the human aorta. Furthermore, no measurements have been made in vivo.
Thus it seems that the validity of the presented hypothesis revolves around the crucial question: Does arterial wall hypoxia exist in vivo? And if so, is the degree of hypoxia sufficient to lead to decisive changes in arterial wall metabolism?
To address this issue, we tested a system for the demonstration of arterial wall hypoxia with a potential in vivo application. The system is based on a hypoxia marker originally developed by Hodgkiss et al11 to detect hypoxia in tumors. In this system a nitroimidazole derivative with a theophylline ligand, NITP, is used. When this compound is taken up intracellularly it undergoes nitroreduction and oxidation in a futile cycle, provided that oxygen is present. Under hypoxia, the futile cycle is broken and aggressive radicals arise, bind to cellular components, and become trapped. The theophylline residue may then be detected by immunological methods.
The main purpose of the present study was to test this method on arterial tissue of varying thicknesses incubated at varying degrees of hypoxia. Our results indicate that at least in vitro, zones of hypoxia may be detected in the media already at a surrounding O2 concentration of 21% in pig aorta and 6% in rabbit aorta. Since NITP11 and similar compounds12 13 have been used in vivo in animals and even humans,14 we propose that this technique may be used to assess arterial hypoxia in vivo.
Medium, Chemicals, and Antibodies
Eagle’s minimum essential medium with 25 mmol/L HEPES buffer, pH 7.4, (Serva Feinbiochemica) was used throughout the study.
NITP was kindly provided by Dr R.J. Hodgkiss, Gray Laboratory of the Cancer Research Campaign, Mount Vernon Hospital, Northwood, UK. To prepare the 0.25 mmol/L NITP solution that was used in the study, 0.063 mmol NITP was first dissolved in 1 mL dimethyl sulfoxide (Sigma Chemical Co) and then added to 250 mL medium. All NITP solutions were freshly prepared before use.
Sheep anti-theophylline antibody (theophylline-8–bovine serum albumin) was purchased from Biogenesis Inc. Biotinylated rabbit anti-sheep IgG antibody, fluorescein avidin D, biotinylated goat anti–avidin D, and normal rabbit serum were all obtained from Vector Laboratories. Avidin–horseradish peroxidase complex was prepared from reagents A and B as described in the Vectastain ABC kit (Vector Laboratories). DAB was purchased from Amersham International; AEC was from Sigma. Mayer’s acid hematoxylin was obtained from Apoteksbolaget, and dry milk powder was from Semper. Vectashield was used as a mounting medium (Vector Laboratories). Glass slides were coated with 0.1% poly-l-lysine (Sigma) before use.
Eight male New Zealand White rabbits were fed standard rabbit chow with (n=3) or without (n=5) the addition of 1% cholesterol to induce experimental atherosclerosis. Rabbits were anesthetized by administration of 1:1 Ketalar (50 mg/mL; Parke-Davis) and Rompun Vet (20 mg/mL; Bayer AG), 0.35 mL/kg IV, and 0.17 mL/kg IV every 15 minutes thereafter.
The thoracic aorta was removed under the aforementioned protocol anesthesia, by using a technique that has been developed to ensure maximum tissue integrity.15 The major part of the adventitia was dissected away, and the aorta was divided into 5-mm segments and put into vials with incubation medium. Approximately 200 sections from 25 aortic segments were examined.
During general anesthesia (Stresnil [Janssen Pharmaceuticals] at 40 mg/mL and 2 mL/10 kg IM as premedication; then Diprivan [ICI Pharmaceuticals] at 10 mg/mL and 2 mL/10 kg IV initially plus 2 mL/10 kg IV every 15 to 20 minutes thereafter), a 10-cm-long piece of thoracic aorta was obtained from a domestic swine that had been raised under standard conditions. The aorta was rinsed with oxygenated medium (75% O2 and 1% CO2 in nitrogen), and the major part of the adventitia was removed. The tissue was divided into 1-cm2 pieces and distributed in vials with incubation medium.
Rabbit aortic segments with or without experimental atherosclerosis were incubated for 3 hours at 37°C in 0.25 mmol/L NITP medium equilibrated with 5% CO2 and 0%, 3%, 6%, 21%, or 75% O2 in nitrogen. Pig aortic specimens were incubated under similar conditions at 3%, 10%, 21%, and 75% O2 concentrations.
As negative controls, tissue specimens were incubated at 3% O2 in medium without NITP.
After incubation, specimens were fixed in 4% buffered formaldehyde, pH 7.0, for 1 to 2 days. Paraffin sections (5 μm) were made and transferred to glass slides. On each slide samples were placed together with one positive and one negative control. Specimens incubated in medium equilibrated with 3% O2 were used as positive controls, since such incubations had given a distinct immunoreaction in preliminary experiments. Sections incubated in medium without NITP were used as negative controls.
After deparaffination and rehydration to buffer in a graded series of alcohols, sections were prepared for staining with immunoperoxidase or immunofluorescence. Endogenous peroxidase activity in sections intended for immunoperoxidase staining was blocked with 0.3% H2O2 in PBS, pH 7.4, for 20 minutes. Except for the initial step of peroxidase inactivation and the preparation of the avidin–horseradish peroxidase complex, all dilutions were made in PBS, pH 7.4, with 5% (wt/wt) dry milk powder.
The procedures to block nonspecific binding, detect the theophylline ligand of the NITP, and couple the secondary biotinylated antibody were common to both the immunoperoxidase and immunofluorescence methods, ie, 20 minutes in normal rabbit serum (1:50) followed by 60 minutes with primary sheep anti-theophylline antibody (1:2000 for peroxidase, 1:5000 for fluorescence). After this the preparations were rinsed three times for 5 minutes each in PBS with 5% dry milk powder, pH 7.4, before being treated for 30 minutes with secondary biotinylated rabbit anti-sheep IgG antibody (1:100) and rinsed three times for 5 minutes each in PBS with 5% dry milk powder.
Immunoperoxidase and Immunofluorescence Testing
Avidin–horseradish peroxidase complex was freshly prepared according to the manufacturer’s instructions, and sections were incubated for 60 minutes. Peroxidase substrate (DAB 1.3 mmol/L and 0.01% H2O2 in 50 mmol/L Tris buffer, pH 7.5) was added for 10 minutes, and sections were rinsed in tap water for 5 minutes followed by 20 minutes of counterstaining in Mayer’s acid hematoxylin. After rinsing in tap water the preparations were mounted.
In some preliminary studies AEC was used as the peroxidase substrate (60 mg AEC dissolved in 15 mL dimethylformamide to 210 mL of 0.1 mmol/L NaAc, pH 5.2).
To enhance the immunofluorescence signal, consecutive 30-minute incubations were made with fluorescein–avidin D (1:500), biotinylated goat anti–avidin D antibody (1:100 or 5 μg/mL), and fluorescein–avidin D (1:500) once again. Sections were mounted after being rinsed in tap water.
Bright-field microscopy for immunoperoxidase detection was performed on an Axiophot microscope (Carl Zeiss). Immunofluorescence was studied on the same microscope by using filter set 16 (BP 485/20, FT 510, LP 520) (Carl Zeiss).
Establishment of Experimental Conditions
In a number of preliminary experiments optimal conditions during tissue incubation and immunoperoxidase and immunofluorescence procedures were tested.
With an NITP concentration of 0.1 mmol/L the staining was weak, whereas 0.5 mmol/L did not seem to increase the observed immunoreactions compared with the 0.25 mmol/L that was selected for the further experiments. This concentration is well below toxicity levels for cells in culture. In V79 Chinese hamster cells more than 90% of them were viable under hypoxia at 0.5 mmol/L of NITP after 3 hours.11
The duration of the incubation varied between 0.5 and 6 hours. Half an hour resulted in poor staining and 1.5 hours gave varying results, but 6 hours did not seem to increase the detectability compared with the 3 hours that was judged optimal. The chosen incubation time agrees with experiments on V79 Chinese hamster cells in culture.11
To arrive at the selected incubation times, antibody concentrations, and staining procedures, an array of experiments was performed that will not be accounted for here. It suffices to mention that the alternative peroxidase substrate that was tested, AEC, which yields a crimson stain, seemed to be less easy to detect than DAB. Different filter combinations for detecting fluorescence were also tested. It seemed that using a narrower wavelength band of the activating light (BP 485/20), although it gave a less intense FITC emission, resulted in a better discrimination between the brownish yellow of the elastin autofluorescence and the greenish yellow of the FITC fluorescence.
Sections from normal rabbit aorta that had been incubated at a 3% O2 concentration for 3 hours in NITP-containing medium exhibited a zone of immunoreaction in the middle portion of the media. In immunoperoxidase sections brown staining was seen associated with the cytoplasm and the nuclei of the medial SMCs (Fig 1D⇓). The brownish yellow autofluorescence of the elastic laminae was easy to distinguish from the specific FITC immunofluorescence similarly associated with the cytoplasm and nuclei of the medial SMCs (Fig 1C⇓). The zone was 5 to 7 elastic lamellae wide (≈140 μm), and the distance from the lumen corresponded to 5 or 6 elastic lamellae (≈100 μm). In sections that had been incubated at 0% O2 the peroxidase staining and the immunofluorescence were more patchy and widespread (Fig 1E⇓ and 1F⇓). At a 6% O2 concentration a faint, narrow band with peroxidase staining and specific fluorescence could be discerned, but not in all sections. At 21% O2 no immunoreaction could be detected in the media (Fig 1A⇓ and 1B⇓). Fluorescence and peroxidase staining were also regularly seen in the endothelium (Figs 1⇓, 2A⇓, and 2B⇓) and in the fat cells in the adventitia. However, this reaction did not depend on the degree of hypoxia during the incubations, since it appeared over the whole range of O2 concentrations, ie, 0% to 75%. In the sections that had been incubated in the absence of NITP no specific fluorescence or peroxidase reaction was seen.
In sections from rabbit aorta with experimental atherosclerosis the basic pattern of immunoreaction in the media was the same, ie, a faint band of peroxidase staining or fluorescence in sections from tissue incubated at a 6% O2 concentration (Fig 3B⇓) and a wider zone in sections incubated at 3% O2 (Fig 3C⇓ and 3D⇓).
In rabbit intimal lesions weak peroxidase staining and fluorescence were seen in occasional cells, presumably foam cells, after incubation at 21% and 6% O2 concentrations (Fig 3A⇑ and 3B⇑). More cells showed immunoreaction at 3%, and at several locations intense staining and fluorescence were seen in a majority of the cells (Fig 3C⇑ and 3D⇑). Peroxidase staining and fluorescence were also observed in the endothelium and adventitial fat.
In pig aorta, again the basic pattern of immunoreaction was the same; ie, peroxidase staining and immunofluorescence were observed as a band in the inner part of the media. However, the reaction was seen at an O2 concentration of 21% in the surrounding medium (Fig 4E⇓), and the zone was much wider at 10% and 3% O2 concentrations than in rabbit aorta (Fig 4B⇓ and 4C⇓). At 75% O2 concentration no immunoreaction was observed (Fig 4A⇓ and 4D⇓). The distance from the luminal surface to the immunoreaction was about 15, 30, and 45 elastic lamellae wide (≈200, 400, and 600 μm) in sections from pig aorta incubated in 3%, 10%, and 21% O2, respectively, as judged from immunofluorescence, where the elastic laminae were easy to distinguish. It is noteworthy that the width at the 3% O2 concentration was similar to that found in rabbit aorta and that the width at the 21% O2 concentration is more than twice the thickness of the rabbit aorta. No specific fluorescence or peroxidase reaction was seen in the sections that had been incubated in the absence of NITP.
In contrast to the rabbit specimens, no staining or fluorescence was observed in the pig endothelium.
We have presented a method for the demonstration of arterial wall hypoxia in vitro by using immunohistochemical techniques. We propose that this method may be used to assess the degree of hypoxia in the arterial wall in vivo. In this way some of the crucial questions may be addressed that could validate the anoxemia theory of atherosclerosis.3
Nitroimidazole and its derivatives indicate cellular hypoxia in many different systems in vivo and in vitro, and several modes of detection have been applied. γ-Emitting16 17 and β-emitting18 19 probes have been used as well as immunoreactive ligands of different compositions.11 20 In our case nitroimidazole with a theophylline ligand was chosen, and immunoperoxidase and immunofluorescence were used for detection.
In tumor spheroids and biopsies20 21 22 a typical pattern has been described, with a clear zone of 100 to 150 μm close to the oxygen reservoir (medium/blood) followed by a zone of detectable marker. Our main findings fit well with this picture, ie, a clear zone in the media close to the surrounding medium and immunoreaction indicative of the hypoxia marker deeper in the tissue. When the oxygen concentration was increased in the medium a wider clear zone was observed and vice versa, suggesting that a detectable marker started to appear at a depth where the diffusing oxygen had been consumed and Po2 decreased below a certain level.
Direct in situ polarographic measurements on arteries from dogs and rabbits indicate that the oxygen concentration decreases from ≈90 mm Hg in the lumen to 40 mm Hg at 80 μm9 23 and to 20 mm Hg at 150 μm,24 25 ie, that Po2 decreases around 50 mm Hg at the first 80 μm from the lumen and another 20 mm Hg at the next 70 μm.
In our incubations in 3% O2 (≈23 mm Hg) immunoreaction was observed at depths of ≈100 (rabbit) and 200 (pig) μm. At 10% and 21% O2 (≈76 and 160 mm Hg) immunoreaction was observed at ≈400 and 600 μm, respectively. Judging from the direct measurements,9 23 24 25 Po2 would be close to zero at these depths. This suggests that the marker we used could be detected where conditions were close to anoxic. On the other hand, during our incubations the aortic segments were at rest, and no oxygen was consumed for mechanical work, which means that the Po2 gradient probably was less steep than in the in situ preparations. Po2 might therefore have been higher than the direct measurements would indicate. Also, by using the same probe, Hodgkiss et al11 have observed that as much as 30% of isolated cells show immunofluorescence at a 10% O2 concentration and 50% at 0.5% O2. With this in mind it is likely that the cutoff point for immunodetection of NITP in the arterial media is in the range of 2 to 3 mm Hg.
The covalent binding of nitroimidazole derivatives to intracellular structures depends on the presence of nitroreductase activity (most importantly that of cytochrome P-450 reductase26 ), which generates the reactive intermediates, and the absence of oxygen, which would neutralize these intermediates before they bind. In tissue with mainly one type of cell, such as tumor spheroids,20 21 or the arterial media the nitroreductive capacity may be expected to be rather uniform and the marker a true indicator of hypoxia. However, cell types other than SMCs may behave differently. Thus, in our incubations, groups of foam cells in the abluminal part of the atherosclerotic lesions showed a strong immunoreaction at O2 concentrations of 3% or 6% in the surrounding medium, when at the same time more deeply situated SMCs in the media did not stain. At a higher oxygen concentration only a weak reaction was observed. This suggests that foam cells may have a higher Po2 detection cutoff point.
The endothelial cells from the rabbit aorta, on the other hand, exhibited marked immunoreaction at all levels of oxygen concentration that were used (the porcine endothelium did not stain at all). The binding of nitroimidazole derivatives under presumably nonhypoxic conditions has also been observed in cells in other types of tissues, eg, stratified squamos epithelium, liver, and sebaceous glands. This may reflect a high nitroreductase activity, which would overwhelm the oxidative capacity of local oxygen.27 Another suggestion is that the nitroimidazole marker is reduced by DT-diaphorase via a pathway that bypasses the oxygen-dependent futile cycle27 ; activity of this enzyme has been demonstrated in these tissues.28 DT-diaphorase activity has also been demonstrated in both endothelial cells and macrophages in atherosclerotic plaques.29
In the rabbit endothelium the immunoreaction in our incubations was clearly independent of the oxygen concentration, and DT-diaphorase activity might be one way to explain this finding. The absence of immunoreaction in the endothelium of the incubated pig aortic segments might also be understood in this way if one considers the following: DT-diaphorase activity is often used as an indicator of nitric oxide synthase29 (in fact, it has been suggested that nitric oxide synthetase and DT-diaphorase are identical enzymes30 ), and nitric oxide synthetase activity (ie, endothelium-dependent relaxation) may be used as an indicator of endothelial integrity.31 32 It is thus possible that the lack of immunoreactivity in the endothelium of the incubated pig aorta reflects a decreased viability of these cells. Whereas the rabbit aorta was collected in a way that ensured endothelial integrity,15 the pig aorta was collected under less ideal circumstances. Of course, another possibility is that there is a true species difference.
The staining of the foam cells depended on the oxygen concentration, but immunoreaction seemed to occur at a higher local Po2 than for the SMCs of the media. One possibility is that the rate of oxygen consumption in these cells was high and competed with the oxidation of nitroradicals. In fact, a high oxygen consumption in the foam cells may be related to an energy-consuming futile cycle.2 33 Another possibility is that macrophages/foam cells are activated by hypoxic stress at a comparatively high Po2, causing the production of DT-diaphorase.34 Another possibility is that nitroreductase activity in these cells is high and that a higher oxygen concentration was needed to oxidize the reactive intermediates.
In summary, we have shown that NITP binds to the SMCs of the aortic media in a clearly oxygen concentration–dependent manner in vitro. We used immunohistochemical methods to detect the marker, and the resolution of the method allows the mapping of local variations in oxygen concentration in the tissue. Since NITP and related substances have been applied in vivo in other contexts,11 12 13 14 we propose that this method may be used to explore the presence of arterial wall hypoxia in vivo. The intriguing NITP accumulation in foam cells at higher oxygen concentrations implies that less extreme hypoxia could be detected but also partly reflect oxygen-insensitive metabolic pathways within these cells. The mechanisms of the oxygen-independent accumulation of NITP in endothelial cells remain to be explored.
Selected Abbreviations and Acronyms
|SMC||=||smooth muscle cell|
This study was supported by the Swedish Heart and Lung Foundation and the Swedish Medical Research Council.
Björnheden T, Bondjers G. Studies on oxygen consumption in aortic tissue from rabbits with dietarily induced atherosclerosis. Arteriosclerosis. 1987;7:238-247.
Hueper WC. General reviews. Arterioscler Arch Pathol. 1944;38:162-181.
Robertson AL. Oxygen requirements of the human arterial intima in atherogenesis. Prog Biochem Pharmacol. 1968;4:305-316.
Shelton ME, Dence CS, Hwang D-R, Welch MJ, Bergmann SR. Myocardial kinetics of fluorine-18 misonidazole: a marker of hypoxic myocardium. J Nucl Med. 1989;30:351-358.
Björnheden T, Bylock A, Hansson GK, Bondjers G. A system for long-term perfusion of rabbit aorta in vitro. Arteriosclerosis. 1983;3:366-382.
Mannan RH, Somayaji VV, Lee J, Mercer JR, Chapman JD, Wiebe LI. Radioiodinated 1-(5-iodo-5-deoxy-β-D-arabinofuranosyl)-2-nitroimidazole (iodoazamycin arabinoside: IAZA): a novel marker of tissue hypoxia. J Nucl Med. 1991;32:1764-1770.
Santilli SM, Fiegel VD, Knighton DR. Changes in the aortic wall oxygen tensions of hypertensive rabbits: hypertension and aortic wall oxygen. Hypertension. 1992;19:33-39.
Crawford DW, Back LH, Cole MA. In vivo oxygen transport in the normal rabbit femoral arterial wall. J Clin Invest. 1980;65:1498-1508.
Barker SG, Talbert A, Cottam S, Baskerville PA, Martin JF. Arterial intimal hyperplasia after occlusion of the adventitial vasa vasorum in the pig. Arterioscler Thromb. 1993;13:70-77.
Dawson TM, Bredt DS, Fotuhi M, Hwang PM, Snyder SH. Nitric oxide synthetase and neuronal NADPH diaphorase are identical in brain and peripheral tissues. Neurobiology. 1991;88:7797-7801.
Johns RA, Linden JM, Peach MJ. Endothelium-dependent relaxation and cyclic GMP accumulation in rabbit pulmonary artery selectively impaired by moderate hypoxia. Circ Res. 1989;65:1508-1515.
Pearce WJ, Ashwal S, Long DM, Cuevas J. Hypoxia inhibits calcium influx in rabbit basilar and carotid arteries. Am J Physiol. 1992;262:H106-H113.
Brown MS, Ho YK, Goldstein JL. The cholesterol ester cycle in macrophage foam cells. J Biol Chem. 1980;255:9344-9352.
Mitchell JA, Kohlhaas KL, Matsumoto T, Pollock JS, Förstermann U, Warner TD, Schmidt HHHW, Murad F. Induction of NADPH-dependent diaphorase and nitric oxide synthase activity in aortic smooth muscle and cultured macrophages. Mol Pharmacol. 1992;41:1163-1168.