Early treatment with hydroxychloroquine prevents the development of endothelial dysfunction in a murine model of systemic lupus erythematosus

Introduction Accelerated atherosclerosis is one of the major causes of morbidity in patients with systemic lupus erythematosus (SLE). Endothelial dysfunction (ED) is considered an early marker of atherosclerosis. It is a reversible alteration, thus representing an attractive target for prevention strategies against cardiovascular disease. Studies have shown that ED occurs in patients with SLE even in the absence of severe, active disease. Hydroxychloroquine (HCQ) is widely used in SLE to control disease activity, but its use is also associated with an improvement in long-term prognosis. Beyond the beneficial effect in well-established disease, our hypothesis is that treatment with HCQ might have a beneficial impact on ED prevention in SLE. The aim of this study was to assess the impact of early treatment with HCQ on ED in a murine model of SLE. Methods Twelve-week-old NZB/W F1 (NZ) and C57BL/6 J mice (controls) were allocated to receive HCQ or vehicle for 6, 12, or 18 weeks. Proteinuria and anti–double-stranded DNA autoantibodies were determined. ED was assessed in mesenteric arteries (pressurized myography). Nitric oxide (NO) availability and reactive oxygen species (ROS) production were evaluated. Vascular ROS production was measured with dihydroethidium (DHE) fluorescent dye. Results Starting from 18 weeks of age, NZ mice showed a progressive reduction in NO availability, which was normalized by ascorbic acid and apocynin in the up to 24-week-old group, and partly ameliorated in older animals. HCQ administration normalized the NO availability in the up to 24-week-old group, with a partial amelioration in the 30-week-old group. DHE analysis revealed a progressive increment of vascular ROS generation among NZ groups, which was prevented by apocynin. Similarly, in the NZ HCQ-treated group, vascular ROS production was abrogated. Conclusions The ED that characterizes this mouse model of SLE is caused by the nicotinamide adenine dinucleotide phosphate oxidase–driven ROS excess. Very early treatment with HCQ is able to exert vascular protection via an antioxidant effect. Electronic supplementary material The online version of this article (doi:10.1186/s13075-015-0790-3) contains supplementary material, which is available to authorized users.


Introduction
In recent years, the survival of patients with systemic lupus erythematosus (SLE) has increased substantially. Reasons for the improved prognosis include earlier diagnosis, enhanced recognition of milder forms, and mindful use of treatments such as glucocorticoids and immunosuppressive agents. However, because of prolonged life expectancy, patients with SLE are now exposed to an increased risk of morbidity related to the sequelae of disease activity, side effects of medications, and comorbid conditions, such as premature cardiovascular (CV) morbidity [1,2].
Accelerated atherosclerosis is one of the major causes of CV morbidity and mortality in patients with SLE. Of note, over the past 3 decades, mortality has been significantly reduced for all causes among patients with SLE, with the exception of CV disease. For these reasons, the prevention of long-term CV complications in patients with SLE, while still an unsolved issue, represents a particularly attractive target for intervention. Beyond the concomitance with traditional CV risk factors, including obesity, hypertension, dyslipidemia, and diabetes mellitus, SLE is considered per se responsible for direct detrimental effects on the vasculature, which makes SLE itself a potent, independent CV risk factor [3,4].
Endothelial dysfunction (ED) is characterized by impaired nitric oxide (NO) availability and a concomitant increased reactive oxygen species (ROS) generation [5]. Such a systemic condition is shared by the majority of CV risk factors, is recognized as an early and major promoter for atherosclerosis and thrombosis, and is independently related to CV events [5]. Interestingly, SLE and atherosclerosis share mechanisms such as vascular inflammation, ROS excess, immune complexes, and complement activation, which are all able to elicit ED. Of note, ED is a reversible alteration, thus representing a potentially attractive target for preventive intervention against CV disease.
Hydroxychloroquine (HCQ) is an antimalarial compound with immunomodulatory properties widely used in SLE owing to its multiple beneficial effects, including control of disease activity, reduction of damage accrual, improvement of survival, and a relatively safe profile [6][7][8][9][10]. In recent years, HCQ therapy has been associated with an improvement of CV prognosis in SLE. A positive effect on traditional CV risk factors and on thrombotic risk has been described in SLE [11][12][13][14]; moreover, a beneficial effect on vascular reactivity indices has also been reported [15,16].
However, it is still unclear whether the protective CV effect of antimalarials is the indirect result of better control of disease activity or, alternatively, a direct effect of HCQ on the endothelial microenvironment. The availability of lupus murine models offers a great opportunity to study this relationship in a "simplified" setting to improve understanding of the role of the multiple contributing factors.
Recently, Gómez-Guzmán et al. [17] have shown that short-term treatment with HCQ started at a late disease stage is able to reverse ED at the level of large arteries in a murine model of SLE, an effect mediated by a reduction of the nicotinamide adenine dinucleotide phosphate [NAD(P)H] oxidase activity, which is a major ROS source. Beyond the beneficial effect of the antimalarial in conditions of well-established disease, early HCQ use has been also associated with delayed SLE onset in humans [18].
We hypothesized that treatment with HCQ before the onset of the full-blown disease might delay or even prevent SLE-associated ED. The present study was designed to assess the effect of HCQ on endothelial function in a murine model of SLE and its underlying mechanism.

Animal treatment
All experiments were carried out in accordance with European Union Council Directive 2010/63/EU. Female NZB/W F1 (NZ) mice and C57BL/6 J mice (controls; Ctrl) were obtained from Harlan Laboratories (Correzzana, Italy) and housed in a specific pathogen-free barrier facility. Starting from 12 weeks of age, the animals were randomly allocated to receive HCQ (3 mg/kg/day) or vehicle (sham) in the drinking water. The animals were divided into four groups: untreated control (Ctrl; n = 15), NZB/W F1 mice with untreated SLE (NZ; n = 30), control treated with HCQ (Ctrl-HCQ; n = 30), and NZB/W F1 mice with SLE treated with HCQ (NZ-HCQ, N = 30). The mice were treated for three different periods (6, 12, or 18 weeks) ( Table 1). At the end of treatment, they were killed while under avertin-induced anesthesia. The protocol was approved by the local University of Pisa ethics committee on use and care of animals.

Serum and urinary assays
At the time the mice were killed, blood samples were collected and immediately placed in heparinized tubes. Serum was separated by centrifugation and stored at −70°C. Serum titers of anti-double-stranded DNA (anti-dsDNA) autoantibodies were analyzed with a commercial enzyme-linked immunosorbent assay kit (Alpha Diagnostic International, San Antonio, TX, USA), 24-h urinary protein excretion was collected in metabolic cages and measured by dipstick analysis (Bayer, Leverkusen, Germany). For both serum and urine assays, the person who performed the measurements was blinded to the age and treatment allocation of the mice.

Histopathology
The kidneys were excised, fixed overnight in 4 % formalin, and embedded in paraffin. Sections (1-2 μm) were cut, dewaxed, and stained with hematoxylin and eosin (H&E) and periodic acid-Schiff (PAS). H&E-and PAS-stained sections were evaluated for nephritis according to a modified score described by Tao et al. as follows [19]: Glomerulonephritis: 1, focal and mild hypercellularity; 2, multifocal and moderate hypercellularity with capillary dilation and mild The nephritis score is the sum of the individual scores ranging from 0 to 12. The images were acquired using an Axio Imager Z1 microscope (Carl Zeiss MicroImaging, Jena, Germany). All evaluations were performed in a blinded manner.

Preparation of small mesenteric arteries and functional experiments
A second-order branch of the mesenteric arterial tree was dissected and mounted on two-glass microcannulae in a pressurized myograph (model 110P; Danish Myo Technology, Aarhus, Denmark) as previously described [20]. Media and lumen dimensions were measured with the intraluminal pressure maintained at 45 mmHg.
Endothelium-dependent and -independent relaxations were assessed by measuring the dilatory responses of mesenteric arteries to cumulative concentrations of acetylcholine (ACh, 0.001-100 μM; Sigma-Aldrich, St. Louis, MO, USA) and sodium nitroprusside (SNP, 0.01-100 μΜ; Sigma-Aldrich). Vessels were precontracted with norepinephrine (10 μM; Sigma-Aldrich). This concentration was determined in dose titration experiments to establish the threshold concentration able to elicit similar contractions among the experimental groups. To evaluate NO availability and ROS production, ACh concentration-response curves were constructed before and after 30-minute preincubation with the NOS inhibitor N ω -nitro-L-arginine methyl ester (L-NAME, 100 μM; Sigma-Aldrich) or the antioxidant ascorbic acid (10 mM, 30-minute preincubation; Sigma-Aldrich). To evaluate whether oxidative stress could influence NO availability, an additional ACh curve was constructed for simultaneous incubation with L-NAME and ascorbic acid.

Involvement of NAD(P)H oxidase in endothelium-dependent relaxation
To ascertain the contribution of NAD(P)H oxidase to ED in additional experiments on NZ vessels (n = 6 each group), an ACh curve was constructed after 30-minute incubation with two different NAD(P)H oxidase inhibitors, apocynin (Apo, 10 μM; Fluka-Sigma-Aldrich, St. Gallen, Switzerland) and diphenyleneiodinium (DPI, 10 μM; Sigma-Aldrich) [21]. To assess the possibility that NAD(P)H oxidase could affect NO availability, Ach was infused under simultaneous incubation with L-NAME and Apo or DPI.

Detection of vascular superoxide anion generation
The in situ production of superoxide anion was measured by means of the fluorescent dye dihydroethidium (DHE; Sigma-Aldrich) as previously described [22]. Each segment was analyzed simultaneously after incubation with Apo (100 μM) or Krebs solution. Additional incubations with the superoxide scavenger polyethylene glycol superoxide dismutase (PEG-SOD, 600 U ml −1 ; Sigma-Aldrich) were performed as technical positive controls.

Data analysis
Maximal ACh-and SNP-induced responses (E max ) were calculated as maximal percentage increments of lumen diameter. The results are presented as mean ± SEM. The statistical significance of relaxation responses was assessed, taking into consideration the time course and treatment by two-way analysis of variance (ANOVA). Other comparisons were made by repeated-measures ANOVA or by one-way ANOVA followed by a Student-Newman-Keuls test, where appropriate. A value of P < 0.05 was considered statistically significant. The variable n indicates the number of experiments. All analyses were performed using Prism version 5.0 software (GraphPad Software, La Jolla, CA, USA).

Anti-dsDNA autoantibody titers
Anti-dsDNA antibodies were detectable starting at 12 weeks of age in NZ mice (33 % positivity); however, a significant increase in titers was observed thereafter, with higher serological activity (87 % positivity) at 24 weeks of age (47 % positivity at 18 weeks, 67 % at 30 weeks). In the NZ-HCQ group, anti-dsDNA antibody positivity at 24 weeks of age was 20 %, significantly lower than that in the NZ mice (P = 0.002). At 30 weeks of age, the anti-dsDNA positivity was 66 %, with no differences from age-matched, untreated mice. Figure 1 shows the anti-dsDNA titers at different time points in NZ, NZ-HCQ, and Ctrl mice.

Nephritis score
Initial histopathological lesions suggestive of nephritis appeared on kidneys from NZ mice killed at 18 weeks of age, then, the degree of nephritis increased progressively until the last observation at age 36 weeks (mean score 2.75 ± 1.25). Similar nephritis scores were observed in kidneys from NZ-HCQ mice at 24, 30, and 36 weeks of age. These data are reported in Table 2.
No histopathological lesions suggestive of nephritis were detected in control mice.

Effects of HCQ on vascular reactivity
Among NZ-HCQ mice, in the 18-week and 24-week subgroups, HCQ administration normalized the relaxation in response to ACh (Fig. 2c) and restored the inhibition by L-NAME on ACh (Fig. 3). In older NZ animals, HCQ potentiated, without normalizing, the relaxation in response to ACh and partly restored the L-NAME effect on    (Figs. 2c and 3). HCQ was totally ineffective in modulating endothelial function in Ctrl animals (Additional file 1).

Involvement of NAD(P)H oxidase in endothelium-dependent relaxation
In the Ctrl group, the relaxation in response to ACh was not modified by Apo at any time point (Additional file 1). In contrast, in the NZ group, Apo, while ineffective at baseline, normalized the response to ACh at 18 and 24 weeks (Additional file 2) and restored the inhibitory effect of L-NAME on ACh (Fig. 3). Of note, at 30 weeks, Apo improved the relaxation in response to ACh (Additional file 2), although it did not normalize it, and partly ameliorated the inhibitory effect of L-NAME on the endothelial agonist (Fig. 3). Similar results were obtained with DPI, which normalized the vascular response to ACh in NZ mice at 18 and 24 weeks and restored the inhibitory effect of L-NAME on ACh (Additional file 3). At 30 weeks, DPI only in part potentiated the response to ACh and the inhibition by L-NAME (Additional file 3).
In the NZ-HCQ groups, Apo incubation was devoid of any effect, at any time, either on the response to ACh or on the inhibition by L-NAME (Fig. 3).

DHE analysis of vascular superoxide anion generation
At baseline, DHE red fluorescence revealed a weak and similar mesenteric superoxide anion production among the Ctrl and NZ groups (Fig. 4). In the Ctrl group, advancing age failed to elicit any significant increment of superoxide generation. In contrast, in NZ mice, a dramatic increment of superoxide anion production was already present in the 18-week group, with a progressive production at week 24 and a further generation at 30 weeks (Fig. 4). Incubation with Apo prevented the superoxide production at any time point (Fig. 4). Likewise, in the NZ-HCQ group, superoxide production was completely abrogated (Fig. 4). As technical positive controls, in vessels from NZ mice incubated with PEG-SOD, the superoxide generation was completely abolished (Fig. 4).

Discussion
In the present study, we have shown that the early administration of HCQ is able to prevent the development of ED in a murine model of SLE.
Endothelial function impairment has been described previously in mice as well as in human SLE [23][24][25][26][27][28][29][30]. In our mouse model, ED occurred at a very early stage of the disease in the absence of overt disease activity. As early as 18 weeks of age, mesenteric small arteries from the NZ group already showed a blunted endotheliumdependent relaxation, together with a reduced inhibitory effect of L-NAME on ACh, and at this stage anti-dsDNA antibodies were positive, and only initial urinary abnormalities were detectable in some mice. DHE analysis revealed a dramatic increment of vascular superoxide anion generation.
To exclude that these effects might be caused by ageing per se, a well-recognized major contributor to vascular functional changes [31], the results from each time of exposure to the disease were compared with data obtained in age-related control groups. Indeed, vessels from control mice showed a preserved inhibitory effect of L-NAME on endothelium-dependent relaxation and no evidence of ROS generation up to 30 weeks. At this age, the control group displayed a reduced NO availability, but this reduction was less pronounced than in the NZ group, without any increased ROS production. Therefore, these data indicate that SLE-associated ED is a functional vascular alteration occurring early in the disease course, regardless of disease activity and aging. These observations fully support the observation made in patients with SLE who might experience premature atherosclerosis independently of traditional risk factors for CV disease [3,32].
In our model, early and long-term HCQ administration, though ineffective in modulating endothelial function in control mice, normalized the endothelium-dependent relaxation and the inhibitory effect of L-NAME on ACh as well, in both 8-week-old and 24-week-old NZ mice. This protective effect, although still evident, was only partially effective in NZ at 30 weeks of age.
It is well known that renal involvement itself can cause ED. Thus, the observed beneficial effect of HCQ on endothelial function might be consequent to an improvement of kidney disease [33]. However, in our experimental model, this is an unlikely hypothesis, considering that the reversal of ED induced by HCQ occurred before the occurrence of overt kidney disease. In addition, even if chronic administration of HCQ has a positive effect on proteinuria and anti-dsDNA autoantibodies, histological analysis revealed a comparable activity and damage scores in the kidneys from HCQ-treated and untreated mice. In conjunction, these findings indicate that the protective vascular effects by HCQ were not exclusively related to kidney disease and suggest the ability of the drug to interfere with the vascular inflammatory process and the immunological mechanisms accounting for ED, as previously demonstrated [34][35][36][37].
Apo, an NAD(P)H oxidase inhibitor, normalized the inhibitory effect of L-NAME on ACh-induced relaxation up to 30 weeks of age. Moreover, DHE staining showed that Apo prevented the SLE-induced vascular ROS at any time point of the disease.
On the basis of these findings, it is conceivable that NAD(P)H oxidase is a major enzymatic pathway that mediates vascular ROS production in this model of SLE, a process that likely initiates at a very early stage of the disease.
Indeed, the present data extend the previous demonstration of the beneficial effect of HCQ on large arteries [17] to the peripheral microcirculation and demonstrate a property of this compound of delaying the onset of vascular functional alterations for a long period.
The main novelty of the data relies on the fact that our work offers a different perspective, given that early disease is the main focus of the study.  4 Dihydroethidium (DHE) staining for detection of superoxide production. Representative DHE staining (upper panels) and quantitative analysis of the red signal (lower panels; original magnification, ×40) in mesenteric arteries from C57BL/6 J mice control mice (Ctrl) and NZB/W F1 mice (NZ) at baseline, from NZ groups without or with apocynin (Apo) or polyethylene glycol superoxide dismutase (PEG-SOD), and from NZB/W F1 mice treated with hydroxychloroquine (NZ-HCQ) at different time points. Each column represents the mean ± SEM of six experiments. *P < 0.001; † P < 0.05 Indeed, in 2010, Thacker et al. [23] described an impaired endothelial-dependent vasorelaxation at the aortic level at 36 weeks of age in the same model, in concomitance with an overt disease expression. However, in line with our observations, they also showed an earlier impairment in phenotype and function of endothelial progenitor cells, before full disease expression [23].
More recently, Gómez-Guzmán et al. [17], in the same SLE model, described the protective effect of a short course of HCQ on vascular parameters (hypertension, heart rate, cardiac hypertrophy) as well as on disease-related variables. Similar to us, they found, even if in a large vessel bed such as the aorta, a beneficial effect of HCQ on endotheliumdependent vasodilation on vascular ROS levels and NADPH oxidase activity. However, with respect to our study, the authors started the treatment in an advanced disease stage, after a long-standing autoimmune-mediated inflammatory process. Thus, we can postulate that what we observed is not only the result of an immunological process but also a mixture of metabolism, purely CV detriment, and aging. This is supported by the fact that the administration of a mixture of antioxidants yielded similar results [17].
Some limitations of this study have to be acknowledged. First, we have histological data of the kidney from advanced disease, but no earlier evaluations (18-20 weeks of age) were available. This aspect can account for some discrepancy observed between the proteinuria trend and the kidney histology between treated and untreated mice. Indeed, whereas a delay in proteinuria occurrence has been observed in mice during HCQ treatment, this effect was not reflected by the later histology scores. Second, no serial data on anti-dsDNA antibody activity were available from each mouse, because blood samples were taken only at the time they were killed. This could reduce the accuracy in defining disease progression. Moreover, the mice were given the HCQ in their drinking water; although this is a common practice, variations in individual mice as to intake cannot be excluded.