HIV Articles  
Antiretrovirals Induce Direct Endothelial Dysfunction In Vivo
  [Rapid Communication]
JAIDS Journal of Acquired Immune Deficiency Syndromes: Volume 42(4) 1 August 2006 pp 391-395
Jiang, Bo MD; Hebert, Valeria Y. BA; Zavecz, James H. PhD; Dugas, Tammy R. PhD From the Department of Pharmacology, Toxicology, and Neuroscience, Louisiana State University Health Sciences Center, Shreveport, LA.
" endothelial dysfunction induced by AZT and perhaps other NRTIs may contribute to the development of cardiovascular complications observed in HIV patients...."
Summary: HIV-associated cardiovascular diseases have been widely described, but clinical studies aimed at establishing cause-effect relationships between HIV-associated cardiovascular disease and either the HIV infection or antiretroviral therapy have been problematic. Endothelial dysfunction is a sensitive marker and early event in atherosclerosis, and many have suggested that protease inhibitors promote endothelial dysfunction indirectly by inducing elevations in circulating lipids.
To determine whether nucleoside reverse transcriptase inhibitors and/or protease inhibitors induce endothelial dysfunction, and to test whether this effect is dependent upon drug-mediated alteration in plasma lipid concentrations, we treated male Sprague-Dawley rats with pharmacological doses of azidothymidine (AZT), indinavir, or AZT plus indinavir through their drinking water for 1 month and assessed endothelial function in aortic rings using an isometric force measurement.
Circulating levels of plasma lipids and endothelin-1, a marker for endothelial injury and/or dysfunction, were also determined.
We found that AZT and AZT plus indinavir treatments dramatically reduced endothelium-dependent vessel relaxation. However, AZT treatment did not significantly alter plasma levels of cholesterol or triglyceride. In addition, plasma endothelin-1 levels were elevated in rats treated with AZT plus indinavir. Indinavir treatment alone increased plasma cholesterol levels but had no effect on endothelial function.
These findings suggest that in addition to modulating plasma lipid levels, antiretrovirals, particularly AZT and perhaps other nucleoside reverse transcriptase inhibitors, may have direct effects on the vascular endothelium. Together with other increased risk factors for atherosclerosis in HIV patients, AZT-induced endothelial dysfunction may contribute to the cardiovascular diseases associated with HIV antiretroviral therapy.
A variety of cardiovascular abnormalities are now widely described in even young HIV patients taking antiretroviral therapy.1 Although it is as yet inconclusive whether the HIV infection, antiretroviral therapy, or both initiate the disease, drug-related cardiovascular toxicities have been described.1-3 Protease inhibitor-associated metabolic disorders, including lipodystrophy/lipoatrophy, dyslipidemia, type 2 diabetes mellitus, and insulin resistance, are well documented,4,5 and all are known risk factors for atherosclerosis.
One of the most important factors determining of the onset or progression of atherosclerosis is the integrity of the vascular endothelium. The vascular endothelium serves a number of important roles in the maintenance of vascular homeostasis.6 For example, the endothelium regulates vascular tone by secreting vasorelaxing factors, including nitric oxide and prostacyclin, as well as vasoconstricting factors, such as angiotensin II and endothelin-1 (ET-1).6 It produces procoagulation and anticoagulation factors that control platelet activity, cytokines, and adhesion molecules that regulate the inflammatory process, and factors that either promote or retard vascular smooth muscle cell (VSMC) proliferation. In contrast, endothelial dysfunction occurs when an imbalance between vasodilatory and vasoconstrictive factors occurs.6 An impaired endothelium culminates in reduced vasodilation and an overproduction of procoagulant mediators and of mitogenic factors that promote VSMC proliferation. All of the well-known risk factors for atherosclerosis, including hypercholesterolemia, hypertension, smoking, and diabetes mellitus, are correlated with endothelial function; and endothelial dysfunction is attenuated with regression of these risk factors.7 Endothelial dysfunction can thus serve as a sensitive marker for atherosclerosis.8
Prior reports suggest that endothelial dysfunction contributes to antiretroviral-associated cardiovascular diseases.9 Studies by our laboratory and by others showed that antiretrovirals induce endothelial cell cytotoxicity in vitro,10,11 indicating that these drugs may have a direct deleterious effect on the vascular endothelium. A study conducted in FVB/n mice suggested that nucleoside reverse transcriptase inhibitors (NRTIs), such as azidothymidine (AZT) and stavudine, can impair endothelial function.12 However, this study did not elucidate whether this effect was associated with plasma lipid levels and did not compare the effects of NRTI with those of protease inhibitors or with the 2 drugs in combination. Our hypothesis was that in addition to inducing metabolic complications, antiretrovirals can have a direct deleterious effect on the vascular endothelium, inducing endothelial dysfunction in the absence of lipid abnormalities. To test our hypothesis, we treated male Sprague-Dawley rats subchronically with AZT and/or indinavir at doses similar to those administered in humans and measured both plasma lipids and indices of endothelial dysfunction. Endothelial dysfunction will be determined experimentally by measuring endothelium-dependent vasodilation, that is, the ability of the endothelium to promote vessel dilation, and by determining markers for endothelial dysfunction, for example, the mitogenic and vasoconstricting factor ET-1.13
Treatment of Animals with Antiretrovirals

Two major components of commonly prescribed antiretroviral regimens were investigated to determine which components are capable of directly damaging the vascular endothelium in vivo. In consideration of current usage, and avoiding drug-drug interactions in vivo, we chose AZT as a representative NRTI and indinavir as a representative protease inhibitor. Note that these 2 drugs are among the recommended initial therapies for HIV for the successful reduction of viral titers.14,15
Male Sprague-Dawley rats weighing 200 to 225 g (aged 49-52 days) were administered distilled water, AZT (10 mg/kg/d), indinavir sulfate (15 mg/kg/d), or AZT plus indinavir sulfate through their drinking water for 30 days. Drug concentrations in the water were adjusted according to the volume each animal consumed in 1 day (about 20-30 mL).
Measurement of Vasodilation
After 30 days treatment, the rats were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and were killed by pneumothorax. The thoracic aortas were excised, cut into rings, and connected to a force-displacement transducer. The aortic rings were then equilibrated in Kreb solution (NaCl, 118 mmol/L; KCl, 4.71 mmol/L; MgCl, 1.05 mmol/L; NaH2PO4, 1.33 mmol/L; NaHCO3, 25 mmol/L; CaCl2, 2.7 mmol/L; and glucose, 5.6 mmol/L) and were aerated with 95% O2 and 5% CO2 for 2 hours. The isometric force was recorded, as described previously.16 The rings were precontracted with phenylephrine until 70% to 80% of maximal contraction was achieved, and arterial relaxation was evaluated after addition of increasing doses of acetylcholine, for endothelium-dependent relaxation, or sodium nitroprusside, for endothelium-independent relaxation.
Note that measurement of endothelium-dependent vessel dilation after stimulation with acetylcholine is considered the "gold standard" for the measurement of endothelium-dependent vasodilation.6 Acetylcholine induces an endothelium-dependent relaxation of vascular smooth muscle via muscarinic receptor binding and a concomitant production of nitric oxide (NO).17 Endothelium-derived NO, in turn, promotes relaxation of vascular smooth muscle. On the other hand, sodium nitroprusside is an NO donor that can act directly on vascular smooth muscle to promote vasodilation. Sodium nitroprusside thus induces an endothelium-independent vasodilation. Importantly, endothelial dysfunction is considered a systemic disorder, such that measurement of endothelial dysfunction in even peripheral vessels generally parallels that of coronary arteries, etc.6 Therefore, measurement of vasodilation of aortic rings in response to acetylcholine versus sodium nitroprusside, that is, determination of endothelium-dependent versus endothelium-independent vasodilation, is an appropriate approach for accomplishing the aims of this study.
Determination of Plasma Analytes
At the time of death, blood was drawn from the vena cava into 10-mL EDTA tubes that were then centrifuged for the collection of plasma. Total cholesterol and triglyceride levels in diluted plasma samples were determined colorimetrically using kits purchased from Wako Chemicals (Richmond, VA). Plasma ET-1 was determined by enzyme-linked immunosorbent assay using a kit obtained from Cayman Chemicals (Ann Harbor, MI).
Measurement of Plasma Drug Concentrations
Plasma drug concentrations were determined as described by Notari et al.18 First, 100 μL of the plasma samples were diluted with 20 μL methanol. The samples were then vortexed vigorously and centrifuged at 13,000 rpm for 6 minutes. The supernatants were diluted with 200 μL deionized water and were loaded onto preconditioned 1-mL Sep Pak C18 column cartridges (Waters Corp, Milford, MA). Once the samples were applied, the columns were washed with 1 mL 5% methanol; and the plasma analytes were eluted first with 550 μL 10 mmol/L KH2PO4 and then with 2 mL methanol. The latter 2 fractions were combined and evaporated to dryness. The efficiency of the solid phase extraction, determined by loading similar quantities of standards, was confirmed to be greater than 90%.
The chromatographic separation was accomplished using a Waters Alliance 2695 high-performance liquid chromatography system and an Ultrasphere (Beckman Coulter, Fullerton, CA) 5 μm, 4.6 mm internal diameter X 25 cm reversed-phase C18 column. A 30-μL aliquot of plasma extract was injected. The mobile phase consisted of 10 mmol/L KH2PO4 (A) and acetonitrile (B) at a flow rate of 1 mL/min. The gradient elution program was as follows: (1) a linear gradient from 94% A/6% B to 60% A/40% B over 10 minutes; (2) isocratic elution at 60% A/40% B over 10 minutes; (3) a linear gradient to 100% B over 3 minutes; (4) isocratic elution at 100% B for 7 minutes. Using a Waters 2487 dual-wavelength UV detector, AZT was monitored at a wavelength of 260 nm, whereas indinavir was detected at 210 nm on a separate channel using Millenium 32 software (Waters). A standard curve was constructed to quantitate the antiretroviral concentrations in the plasma samples.
In this study, we treated rats for 30 days with antiretrovirals at doses similar to those administered in humans. To confirm that the plasma drug concentrations achieved in the animals were similar to efficacious steady-state concentrations in humans, we measured drug levels in the plasma using high-performance liquid chromatography with UV detection. Our data showed that the plasma concentration of AZT in the rats treated with AZT only was 0.24 ±0.02 μg/mL (Table 1), comparable with the pharmacological steady-state concentration (0.19 ±0.08 μg/mL) measured in humans treated with 300 mg AZT, bid.19 The plasma concentration of indinavir measured in rats treated with indinavir only was 2.14 ±0.67 μg/mL (Table 1), within the same range (0.32-11.1 μg/mL = C min-C max) as that measured in humans treated with 800 mg indinavir, bid, boosted with 100 mg ritonavir.20 Finally, the concentrations of AZT and indinavir were not significantly altered when the 2 drugs were administered in combination compared with individual administration of the drugs, with levels equal to 0.20 ±0.03 and 0.88 ±0.24 μg/mL, respectively (Table 1). Although there was an apparent trend toward a decrease in the concentration of indinavir in the combination treatment group, this effect was not statistically significant. Furthermore, the measured concentration remained well above the C min reportedly associated with effective reductions in viral load.20
Treatment with AZT and AZT plus indinavir dramatically impaired aortic relaxation in the presence of acetylcholine. Almost no change in vessel relaxation was observed at even high concentrations of acetylcholine (Fig. 1). In contrast, aortic relaxation elicited by sodium nitroprusside was not altered, suggesting that a specific impairment of vascular endothelial function was induced by treatment with AZT or AZT plus indinavir. Similar trends were observed in indinavir-treated animals, although there was not a significant difference compared with the control animals.
As an additional test for vascular endothelial injury, we measured plasma levels of ET-1 at the time of sacrifice. Although we previously observed antiretroviral-mediated ET-1 release in vitro,10 we did not detect an increase in ET-1 levels in animals treated with AZT only (Table 1). However, ET-1 levels were elevated in the plasma of rats treated with AZT and indinavir in combination (10.29 ±1.41 pg/mL, compared with 7.26 ±0.59 pg/mL in control rats; Table 1).
To determine whether the observed endothelial dysfunction and ET-1 release were associated with lipid modulating effects of the drugs administered, we measured total cholesterol and triglyceride levels in the plasma. Only indinavir treatment significantly elevated plasma cholesterol (Table 1). No significant effect on plasma lipid levels was observed in rats treated with AZT or AZT plus indinavir (Table 1). In addition, plasma triglycerides were not significantly elevated for any of the antiretroviral treatment groups.
Since the initial realization of the atherogenic properties of HAART, protease inhibitors were considered a major contributor to antiretroviral-associated cardiovascular complications.21 This was inferred from the significant increases in cardiovascular complications in patients administered antiretroviral combination therapies after the introduction of protease inhibitors in 1996. Protease inhibitors are indeed associated with a series of metabolic syndromes, including lipodystrophy, dyslipidemia, diabetes mellitus, and insulin resistance, and these are certainly risk factors for atherosclerosis.21 NRTIs, on the other hand, are known to inhibit mitochondrial DNA polymerase-γ; and this inhibition has been suggested to explain the toxicities observed in myocardial cells.22
An intact endothelium is important for maintaining vascular homeostasis, and injury to endothelial cells initiates pathogenesis.6 The endothelium exerts its major vascular protective effects by acting not only as a barrier but also as a regulator of vasoconstriction, vasodilation, VSMC proliferation and migration, thrombogenesis, and fibrinolysis.6 Prior in vitro experiments completed in our laboratory identified an endothelial cell toxicity induced by both AZT and indinavir,10 suggesting that these 2 drugs may have a direct deleterious effect on the vascular endothelium, perhaps further contributing to drug-associated cardiovascular diseases.
The most striking finding in this study is that a direct endothelial dysfunction was induced in rats treated with AZT or AZT plus indinavir at plasma concentrations that approximate efficacious concentrations in humans, although there was no measurable change in plasma cholesterol or triglyceride levels. These data suggest that clinical treatment with AZT may induce direct vascular endothelial damage.
ET-1, a mitogenic growth factor released by endothelial cells, promotes vascular constriction and VSMC proliferation in a paracrine manner.23 An increase in ET-1 levels indicates damage or injury to endothelial cells, and increases in ET-1 release have been shown to be correlated with atherosclerosis and other vasculopathies.23 Therefore, ET-1 can be considered an independent marker for endothelial dysfunction and cardiovascular diseases.24 In our experiment, we detected an increase in ET-1 levels after AZT and indinavir co-treatment, suggesting that endothelial function was altered. However, we did not measure an increase in plasma ET-1 levels in rats treated with AZT only (Table 1), although an impairment in endothelial function was apparent (Fig. 1). This was not entirely surprising, given that ET-1 released by the vascular endothelium is quickly removed from the circulation by a receptor-mediated pathway in the lungs.25 ET-1 levels are thus maintained at a barely detectable but constant concentration in the plasma, which may make it difficult to detect increases in ET-1 release in vivo without overt pathophysiology. Interestingly, the finding that plasma ET-1 levels were elevated in rats treated with the AZT and indinavir drug combination but not in rats treated with either drug alone is the only indication that the combination treatment was more deleterious than the individual doses of antiretrovirals. However, it is possible that the lack of additional decrease in endothelium-dependent dilation in the combination treatment group was due to an already near-complete loss of dilation in aortic rings of rats treated with AZT alone. An additional decrease in vasodilation would thus have been difficult to distinguish.
As suggested by our in vitro studies, a direct impairment of mitochondrial function and an induction of oxidative stress may be a mechanism for antiretroviral-induced endothelial dysfunction. Furthermore, as reviewed by Lund and Wallace,26 potential mitochondrial loci for the cellular effects of antiretrovirals may include any one of the mitochondrial electron transport chain complexes, as well as other enzymes that use nucleosides or nucleotides as substrates or cofactors for their activity. Additional in vitro and in vivo studies will be conducted to elucidate the mechanisms involved in AZT-induced endothelial dysfunction.
In conclusion, data presented here suggest that in addition to the lipid disorders induced by protease inhibitors and mitochondrial DNA depletion in myocardial cells, direct endothelial dysfunction induced by AZT and perhaps other NRTIs may contribute to the development of cardiovascular complications observed in HIV patients.
  icon paper stack View Older Articles   Back to Top