Effect of Hyperoxia on Vascular Endothelial Growth Factor Levels in a Wound Model FREE

VASCULAR endothelial growth factor (VEGF) is the most specific known growth factor for neovascularization.^1- 2 In vitro, wound macrophages produce VEGF in response to low oxygen tensions.^3- 4 Since wound cells are characteristically hypoxic,^3,5 it is easy to assume that hypoxia is the principal inducer of VEGF expression in physiological settings such as cancer and wounds. However, further examination of VEGF in wounds challenges this conventional assumption.

Clinical and experimental experience demonstrate that increasing the oxygen concentration in severely ischemic and/or hypoxic wounds results in accelerated healing in the form of increased blood vessel growth.^6- 11 The mechanism by which hyperoxic therapy might induce blood vessel growth (in more normal circumstances), however, is not well understood. In particular, the effect of elevations of PO₂ on VEGF levels has not been documented. Recent in vitro studies suggest that hyperoxic exposure leads to upregulation of platelet-derived growth factor receptors and that oxidants such as hydrogen peroxide stimulate endothelial cells and keratinocytes to release VEGF.^12- 13 In addition, macrophages and endothelial cells that are exposed to hyperoxia up-regulate IL-8, transforming growth factor β, and VEGF messenger RNA.^14- 16 The possibility arises, therefore, that hyperoxia may actually induce VEGF production in wounds.

Lactate, a known instigator of VEGF production that is present in wounds, might also be affected by hyperoxia.^17- 18 In this study, we evaluated whether hyperoxia, as achieved by hyperbaric oxygen (HBO) treatment, influences wound fluid VEGF, PO₂, and lactate.

Sixty-three male Sprague-Dawley rats were used. Protocols were approved by the University of California, San Francisco Committee on Animal Research. Four wire mesh wound cylinders were implanted underneath the dorsal skin of the rats. These rats were randomized to 2 groups: hyperoxic and control. Wound fluid was collected from the cylinders and assayed for PO₂, lactate, and VEGF. Plasma VEGF levels were measured in 8 animals in each group.

Wound cylinders (N = 242) were manufactured with 0.25 mm stainless steel wire mesh and capped with silicone disks on each end. Cylinders measured 3.5 cm in length and 1 cm in diameter. Animals were anesthetized with 3% halothane. Two longitudinal dorsal midline incisions were made. Blunt dissection created a space for cylinder implantation. Each incision was used to place 2 cylinders, one on each side, a total of 4 cylinders per animal. Wounds were closed with 4 to 6 superficial interrupted 3-0 silk sutures. Cylinders were aspirated to complete evacuation, yielding 200 to 1200 µL of fluid depending on the day sampled. Aspiration was carried out under 3% halothane anesthesia using a 20 gauge needle and syringe on day 2, 5, or 10 following wounding. Needles were introduced through the silicon cap to avoid bleeding. Each cylinder was sampled only once.

Rats in the hyperoxic group (n = 32) were treated with 100% oxygen at 2.1 (atmospheres absolute) ATA for 90 minutes, twice a day for a total of 7 days. Treatments were administered in a hyperbaric chamber designed for small animals (Model 100; Western Hyperbaric Services, Union City, Calif). Compression and decompression was completed at a rate of 1 ATA per minute and monitored during dives and ascents. The control animals (n = 31) remained at 21% oxygen at 1 ATA for the duration of the study.

Polaragraphic oxygen electrodes (Licox; Medical Systems Corp, Greenvale, NY) were used to measure wound oxygen tension before, during, and after hyperoxia, as described previously.¹⁹ Animals (n = 3 on day 2; n = 5 on days 5 and 10) were sedated with 3% halothane, followed by pentobarbital (35 mg/kg), buprenorphine hydrochloride (0.025 mg/kg), and atropine (0.8 mg/kg). Probes were calibrated in room air prior to use and threaded up 18 gauge hubless spinal needles inserted subcutaneously into the cylinder. The spinal needles were removed prior to HBO treatment, and probes were centered in the cylinder. Oxygen tension was measured immediately before hyperoxic exposure and continuously monitored during treatment until 1 hour after oxygen exposure ceased. The monitor automatically corrected for changes in tissue temperature, which were measured continuously.

Wound fluid sampling was timed halfway between the twice-daily 90-minute treatments that began at 8:00 AM and 5:00 PM. Plasma VEGF was measured on the same schedule.

Fluid was transferred into 1.5 mL Eppendorf tubes immediately after aspiration and centrifuged at 5000g for 10 minutes at 4°C. The cell-free supernatant was then aspirated and frozen at −30°C. Thawed samples were analyzed by murine VEGF enzyme immunosorbent assay per manufacturer's instructions (R&D Systems, Minneapolis, Minn). Fifty milliliters of wound fluid (diluted 1:6 in calibrator diluent) was incubated with 50 µL of assay diluent for 2 hours at room temperature in a 96-well plate coated with polyclonal antibody to murine VEGF. After 5 washing steps, a solution of antibody against mouse VEGF conjugated to horseradish peroxidase was added. The mixture was incubated at room temperature for 2 hours. After addition of hydrogen peroxide and tetramethylbenzidine, optical density was determined using a microtiter plate reader set at 490 nm. Samples were compared with serial dilutions of a murine recombinant VEGF standard. Lactate levels were determined using a lactate analyzer (model 2700; Yellow Springs Instruments, Yellow Springs, Ohio).

Statistical significance between baseline, peak, and postexposure PO₂ was determined by 2-tailed paired t test, with significance defined as P<.01. Data are expressed as mean + SD. The VEGF and lactate levels exhibited a nonuniform distribution. Nonparametric statistical analyses were used to determine significance. The Kruskal-Wallis statistic was used to compare differences between days, and the Mann-Whitney rank sum test was used to evaluate differences on specific days.

The mean wound oxygen tensions during day 2 of hyperoxic exposure are shown in Figure 1. Table 1 gives baseline, peak, and postexposure oxygen tensions in wounds on days 2, 5, and 10 of treatment. Baseline wound PO₂ levels were hypoxic (range, [7.4 + 3.9]-[13.7 + 16.9] mm Hg) on all days, and increased significantly during hyperoxic treatment (range, [373.2 + 118.1]-[537.4 + 72.0] mm Hg). Levels decreased rapidly on return to 1 ATA but remained significantly greater than baseline 1 hour after HBO was discontinued ([45.6 + 14.2]-[150.5 + 132.0] mm Hg).

In no case was a measurable level of VEGF found in plasma in either group. On the other hand, VEGF was measurable as early as day 2 in wound fluid (HBO medians, control group: 407.0, 347.1 pg/mL, respectively) and remained measurable through day 10 (Figure 2). Peak values occurred on day 5 at which time the HBO group was significantly elevated over the controls by 42% (HBO medians, control group: 1333.5, 935.0 pg/mL, respectively; P≤.007). Values decreased somewhat on day 10, both groups becoming similar (HBO medians, control group: 658.4, 500.1 pg/mL, respectively), but both remained significantly elevated compared with day 2 (P≤.01).

Plasma lactate in rats is approximately 1 mmol/L. Wound fluid levels were significantly higher (Figure 3). Lactate increased significantly in all wounds between days 2 and 5, remaining elevated through day 10 (median control values: 4.20, 5.96, and 6.0 mmol/L, respectively) as previously reported.²⁰ However, the hyperoxic and control groups did not differ (1-β≥.90), and no distinct trends were apparent.

In clinical experience, HBO therapy often stimulates the production of red granulation tissue when administered as treatment for unhealed, ischemic wounds. Since "granulation tissue" consists mainly of new blood vessels and their requisite, supporting collagenous matrix, this is likely due to an increase in both. Increased angiogenesis in response to hyperoxia has been observed in animal models by several investigators.^6,11 Deposition of collagen in cell preparations and human and animal wounds has long been known to depend on PO₂.^21- 23 The current study explores the mechanism by demonstrating that periodic hyperoxia raises levels of VEGF in wound fluid.

Current literature shows, beyond a doubt, that hypoxia induces VEGF production by many cell types. It is easy to assume, therefore, that the characteristic hypoxic environment of wounds is the major stimulant for subsequent repair. If this were true, raising oxygen concentration would be expected to diminish VEGF and angiogenesis. However, our studies demonstrate that hypoxia is not necessary to sustain the release of VEGF in wounds. This concurs with studies by Howdeshiel et al²⁴ who demonstrate in a clever model that wound VEGF levels remain notably high despite the fact that PO₂ in the wounds continuously approached that of arterial blood.

The question of how hyperoxia influences VEGF levels and angiogenesis is not resolved by these data. Macrophages are the major source of VEGF in wounds.^3,17 Peak wound macrophage infiltration occurs by day 5 following wounding.¹⁷ This coincides with the VEGF peak observed in the control group. Previous studies have shown increased wound cell migration in response to hyperoxia, which may partially explain the increased levels of VEGF seen in the hyperoxic group.²⁵ Despite this presumptive increased cellularity, however, VEGF levels fell to control levels once the hyperoxic stimulus was removed. Hence, there are likely to be other mechanisms. Oxidants have been shown to stimulate VEGF production by some cells, including macrophages.^13,26 Present data seem to support increased production as at least 1 mechanism.

Leukocytes, plentiful in wounds, make large amounts of oxidants, and the production rate is proportional to PO₂ within the range of 0 mm Hg to as much as 500 mm Hg. Some oxidants (hydrogen peroxide and nitric oxide) are measurable in wounds.^27- 28 Brauchle et al¹³ have shown that VEGF production by keratinocytes and bronchoalveolar cells in vitro is responsive to oxidant exposure at low concentrations.¹⁵ Cho et al²⁶ report a similar finding for macrophages. Leukocytes are continuously present in wounds for many weeks. On this basis, increased oxygen would be expected to enhance oxidant production in wounds and, therefore, increase VEGF.

Oxidants in physiological concentrations also enhance collagen production by fibroblasts.²⁹ Furthermore, collagen synthesis is proportional to the concentration of molecular oxygen (PO₂) in the range of 0 mm Hg to 200 mm Hg. The mechanism involves procollagen hydroxylation, which requires molecular oxygen. Hence, collagen deposition, which serves to reinforce growing capillaries on exposure to blood pressure, is enhanced by both PO₂ and oxidant exposure.

The understanding of oxidant chemistry and related pathways in wounds is relatively young. Until recently, oxidant levels had not been measured. Some data now exists for nitric oxide and hydrogen peroxide.^27- 28,30 Addition of antioxidants and oxidant scavengers might be expected to mitigate the oxidant-induced VEGF response. However, data from introduction of such molecules into wound models have proven difficult to interpret.

Lactate was measured in these studies because it also enhances VEGF production at the concentrations found in wounds.^5,18 The sustained lactate level (Figure 3) despite HBO indicates that an important stimulus for VEGF production persists between oxygen exposures and after their cessation at 7 days. Similarly, the hypoxia that recurs between HBO exposures (Figure 3) must also provide some stimulus for VEGF production. In this light, hyperoxia must furnish an intermittent stimulus for VEGF production. Furthermore, this stimulus is either greater than that of hypoxia or, in some way, adds to residuals of the hypoxic signal. Since the properties of this stimulus are unknown, the optimal frequency and duration of exposures to hyperoxia have yet to be determined.

The data confirm again that increasing oxygenation of wounds does not lower lactate levels as one might expect.²⁰ This is because oxidant production by leukocytes is dependent on PO₂ up to levels of 600 mm Hg. The conversion of oxygen to superoxide (O₂-) is responsible for approximately 98% of oxygen consumed in activated cells.³⁰ Therefore, the energy for the conversion must be linked to substrate concentration and must come from aerobic glycolysis, as proven experimentally.³⁰ Increased oxygenation, therefore, increases net production of lactate.^17,31 The increased lactate production by activated leukocytes apparently offsets the reduction of lactate production, which occurs when oxygenation of fibroblasts increases.

The question arises as to whether the hyperoxia-induced fraction of VEGF is biologically active. In this model, such a quantification is technically difficult. However, supportive evidence of hyperoxia-induced, biologically active VEGF has been demonstrated (J.J.G., unpublished data, 2000), showing that hyperoxia-induced angiogenesis is abrogated in the presence of anti-VEGF antibody in a matrigel-murine wound model. Notably, collagenases such as matrix metalloproteinases, also a requisite for angiogenesis, are activated in oxidizing conditions.

Various VEGF-stimulating growth factors may be involved in regulating wound VEGF. Cell culture data suggest that hyperoxia may well change the concentration and receptor density of other wound-active growth factors.^12,14- 16 If VEGF concentration directly responds to hyperoxia, it is possible that other growth factors respond similarly and potentiate angiogenesis. Undoubtedly, more aspects of hyperoxia remain to be discovered.

These findings, taken with existing information, indicate the following: (1) VEGF production in wounds may be mediated by oxidants; (2) hypoxia may be a less important stimulus for neovascularization of wounds than previously thought; and (3) the benefits of HBO therapy for ischemic wound failure relate, in part, to increased VEGF concentration.

The authors wish to thank Heinz Scheuenstuhl, AB, for assistance with the described lactate assay.

Funding provided by the Genetech Foundation, South San Francisco, Calif, and National Institutes of Health, Bethesda, Md, grants GM27345 (Drs Hopf, Hussain, and Hunt) and GM08258 (Dr Gibson).

Reprints: Thomas K. Hunt, MD, Department of Surgery, University of California, San Francisco, 513 Parnassus Ave, Box 0522, HSW 1652, San Francisco, CA 94143-0522.