Our previous studies have shown that broccoli sprouts high in the glucosinolate glucoraphanin decreases renal and vascular oxidative stress and inflammation as well as blood pressure in spontaneously hypertensive stroke-prone (SHRSP) rats. The objective of this study was to determine whether the metabolite of glucoraphanin, sulforaphane, was responsible for this improved blood pressure and whether this is associated with normalization of renal methylated DNA.
Sulforaphane was given by gavage to SHRSP and Sprague Dawley (SD) rats over 4 months and blood pressure measured under anesthesia just before euthanasia. Renovascular morphology was determined by histology and methylated deoxycytosine levels analyzed using high-performance liquid chromatography.
Mean arterial pressure was 20% higher in vehicle-treated SHRSP when compared to SD. Sulforaphane administration to SHRSP improved blood pressure and lowered this difference to 11%. Vehicle-treated SHRSP had significantly increased wall:lumen ratios in renal arteries, increased numbers of vascular smooth muscle cells (VSMCs), increased renal protein nitration, and decreased (11%) renal DNA methylation compared to SD. Sulforaphane administration to SHRSP significantly lowered arterial wall:lumen ratio by 35%, reduced the number of VSMCs, reduced the level of protein nitration, and increased methylated deoxycytosine levels by 14%.
Sulforaphane administration rectified pathological abnormalities in SHRSP kidneys and significantly improved blood pressure. This was associated with normalization of global kidney DNA methylation suggesting that DNA methylation could be associated with hypertension.
Essential hypertension is a polygenic disease1 with a strong environmental influence on its pathogenesis.2 It is an age-related disease3 and likely involves progressive changes in gene expression over time, due to age-related factors and environmental influences.
In view of the polygenic nature of hypertension, many candidate genes such as 11β-hydroxysteroid dehydrogenase type II (HSD11B2)4 have been investigated for their involvement in its pathogenesis. Even though clinically important data can be obtained from investigating individual genes linked with hypertension, due to the involvement of multiple genetic and environmental factors, some important causal associations may not be apparent in a reductionist approach such as this.
Therefore, some researchers have highlighted the necessity to investigate the global phenomena such as telomere shortening and subsequent genomic instability and chromosomal aneuploidy, which affect the expression of multiple genetic loci simultaneously.5 A similar phenomenon that has not drawn much attention in the field of hypertension is global DNA methylation. Although it is known that promoter methylation reduces gene expression in blood pressure related genes such as HSD11B2,4 the role of global DNA methylation in pathogenesis of hypertension has not been investigated in animal models of hypertension. It is known that genomic DNA hypomethylation increases with age.6 Since hypertension is an age-related disorder it is plausible that global DNA methylation would also play a role in its pathogenesis.
An important link in the chain of events leading to hypertension is oxidative stress.7 Oxidative stress is known to initiate and modulate different stages of pathogenesis of hypertension such as endothelial dysfunction8 and inflammatory changes in vascular smooth muscle cells (VSMCs).9 Reactive oxygen species (ROS) are also known to oxidize bases such as deoxyguanosine and 5-methyl-deoxycytosine converting them to 8-oxo-deoxyguanosine and 5-hydroxymethyl cytosine.10 These bases constitute CpG islands that are subjected to DNA methylation and, hence, oxidative changes to these bases might bring about changes in DNA methylation pattern.10 Thus there is a possibility of a link between oxidative stress, DNA methylation, and hypertension.
Our previous studies11 on spontaneously hypertensive stroke-prone (SHRSP) rats demonstrated that rats on diets supplemented with broccoli sprouts rich in the phase 2 protein inducer precursor, glucoraphanin, had significantly less oxidative stress and lower blood pressure compared to SHRSP rats on regular rat chow or chow supplemented with broccoli sprout where the glucoraphanin had been destroyed;11 furthermore, these effects were transferred to the second generation even when these animals were fed only regular rat chow,11 demonstrating heritability of these effects and, thus, epigenetic involvement. There is increasing evidence that programming during the developmental stages can play a role in hypertension in adulthood.12 Such programming likely occurs through changes in the epigenome; indeed, there is increasing evidence that the routes of chronic diseases are associated with changes in the epigenome.13
The objectives of the research presented here are threefold. First, to determine whether the blood pressure ameliorating effect of broccoli sprouts with intact glucosinolates could be attributed to sulforaphane, a potent phase 2 protein inducer that is the metabolite of glucoraphanin. Second, since the kidneys play an important role in governing blood pressure, to determine whether sulforaphane affects renopathological changes associated with hypertension. Third, to determine if such blood pressure ameliorating effects correlated with changes in global deoxycytosine methylation. For these studies, as in our previous studies11 we used SHRSP, good animal model for genetic hypertension.14
Animals, sulforaphane treatment, and blood pressure measurement. Animal experiments were carried out in accordance with the Canadian Council on Animal Care Guidelines with the research approved by the University Committee on Animal Care and Supply. Four-week-old female SHRSP as well as Sprague Dawley (SD) rats were purchased from Charles River (St Constant, Quebec, Canada) and maintained at the College of Medicine animal facility. The reasons for choosing SD rats as the blood pressure control strain is given in the Supplementary Data online. The rats were fed Purina rat chow and adapted for handling for a period of 1 week. Five-week-old animals were divided into two groups for each rat strain (n = 5–6 animals per group). To one group, 10µmol/kg body weight of sulforaphane in corn oil was administered by gavage starting at age 5 week, daily at the same time in the morning. This sulforaphane dosage level was identified as the optimal dosage in terms of blood pressure reduction in a preliminary experiment. The corn oil vehicle was administered to the other groups. The mean body weight ± s.e.m. were: SD-vehicle; 184.8 ± 3.9, SD-sulforaphane; 184.3 ± 3.9, SHRSP-vehicle; 120.1 ± 3.8, SHRSP-sulforaphane; 119.2 ± 2.4. After 4 months of treatment, the rats were anesthetized with isoflurane through inhalation using a vaporizer. The vaporizer was set at 5.0%/l O2 to induce, and 2.0%/l O2 for maintenance of anesthesia. Before euthanasia, blood pressure data were obtained using an external catheter: details are given in the Supplementary Data online. Following blood pressure measurements, animals were perfused with normal saline and visceral organs isolated. Part of the tissue was flash frozen in liquid nitrogen and stored at −80 °C to be used for DNA isolation and western blot analyses. The remaining parts were immediately stored in 4% paraformaldehyde overnight and transferred to a solution of 10% formalin.
High-performance liquid chromatography for methylated deoxycytosine. Flash frozen kidney tissue samples were ground in liquid nitrogen and pulverized in a Mikro-Dismembrator S (Sartorius, Aubagne, France) and then extracted for DNA using Proteinase K (Roche, Laval, Quebec, Canada) digestion and standard phenol-chloroform extraction. Enzymatic digestion and high-performance liquid chromatography analysis of the extracted DNA was carried out as described,15 with some minor modifications to suit the available equipment. Details can be found in the Supplementary Data online.
Histology. Five-micrometer thick sections were cut from paraffin-embedded kidney tissue and the sections were stained with periodic acid Schiff and hematoxylin stains. Some sections were stained with hematoxylin and eosin stains.
Morphometry. We assessed the percentage of sclerotic glomeruli in a blinded fashion in hematoxylin and eosin stained sections. A minimum of 62 glomeruli were assessed per section. Small intrarenal arteries (<500µm in diameter) as well as arterioles were photographed under ×400 resolution as tiff images and arterial contours were manually marked in the software and areas were automatically calculated using MacBiophotonics version (McMaster University, Canada) of Image J software (NIH, Bethesda, MD) for image analysis. We measured 2–5 intrarenal arteries per animal. The number of VSMCs in the arterial wall was counted using the above-mentioned software. Nuclei were outlined manually and the numbers were counted using the software. Further details can be found in the Supplementary Data online.
Immunoblotting. Details of protein separation and immunoblotting can be found in the Supplementary Data online. We used radioimmunoprecipitation assay buffer for extracting proteins from pulverized kidney tissue. Equal amounts of protein (75µg) were loaded onto SDS-PAGE gels and the bands were transferred after protein separation to a polyvinylidene fluoride membrane (#88520; Thermo Scientific, Rockford, IL). Primary antibodies used were rabbit anti-nitro-tyrosine (#9691; Cell Signaling, Danvers, MA) and rabbit anti-γ–glutamyl-cysteine ligase catalytic subunit (#SC-22755; Santa Cruz Biotechnology, Santa Cruz, CA). Following incubation with primary antibodies, membranes were washed and exposed to horse-radish peroxidase conjugated goat anti-rabbit secondary antibody (#SC-2004; Santa Cruz Biotechnology). After washing, proteins were visualized with Western Lightning Plus enhanced chemiluminescence reagent (#NEL104001EA; Perkin Elmer, Waltham, MA) and exposed to Kodak BioMax XAR (#150-1451; Carestream Health, Woodbridge, CT) film. The membrane was stripped of bound antibodies and incubated with β-actin monoclonal antibody (#A2228; Sigma, St Louis, MO), washed and incubated with Immun-Star Goat Anti-Mouse-HRP Conjugate (#170-5047, Bio-Rad, Hercules, CA) followed by exposure to X-ray film. The resulting β-actin bands were used as loading controls to normalize target protein levels. We did not use negative or positive controls as we have used this antibody before and it is specific.12
Statistical analysis. We used nonparametric one-tailed Mann–Whitney with P < 0.05 being considered significant. All data were compared to SHRSP vehicle.
We measured blood pressure using an external catheter under anesthesia before euthanizing the animals. As predicted, SHRSP had significantly higher systolic, diastolic blood and mean pressures (142, 100, and 114 mm Hg respectively) than SD (120, 83, and 95 mm Hg, respectively—see Table 1). Sulforaphane administration reduced all these parameters in SHRSP, without having any effect on SD (Table 1). Reductions were 9, 12, and 11% for systolic, diastolic, and mean arterial pressures respectively.
Global DNA methylation
The molar ratio of methylated deoxycytosine/deoxycytosine, which indicates the level of global DNA methylation in the genome, was 11% lower in vehicle-treated SHRSP kidneys than its SD counterpart (Table 2). Sulforaphane administration to SHRSP increased this ratio by 14% making it indistinguishable from SD. Administration of sulforaphane had no effect on ratio of methylated deoxycytosine/deoxycytosine in the kidneys of SD.
Vehicle-treated SHRSP had a significantly higher percentage of sclerotic glomeruli than its SD counterpart. Sulforaphane treatment tended to reduce the number of sclerotic glomeruli in SHRSP but had no effect on SD (Table 3). The most striking morphological changes were seen in renal small arteries (Figure 1a–d) and arterioles (Supplementary Figure S1a–d online). Vehicle-treated SHRSP small arteries (>100µm and <500µm in diameter) and arterioles (<100µm in diameter) had significantly higher wall:lumen ratio than corresponding SD group. Sulforaphane treatment reduced this ratio in SHRSP but not in SD. This reduction was due to an increase in luminal area and a decrease in wall thickness (Supplementary Table S1 online, Table 4 and Figure 1, and Supplementary Figure S1 online).
Representative images of renal small arteries. SHRSP rats were treated with (a) vehicle or (b) sulforaphane (10 μmol/kg) for 4 months and 5 μm thick sections of kidneys were stained with PAS and hematoxylin staining. Age matched SD rats were also treated the same way: with (c) vehicle or (d) sulforaphane. The bar represents 50 μm. PAS, periodic acid Schiff; SD, Sprague Dawley; SHRSP, spontaneously hypertensive stroke-prone.
Vehicle-treated SHRSP had significantly higher number of VSMCs in the tunica media. Sulforaphane treatment reduced this number in SHRSP and kept it on par with SD (Figure 2).
Vascular smooth muscle cell numbers normalized to endothelial cell numbers. Photomicrographs of PAS and hematoxylin stained renal small arteries were taken from vehicle-treated or sulforaphane-treated SHRSP or SD rat paraffin-embedded kidney sections. Nuclei were counted using Image J software (NIH). Each nucleus was assumed to represent one cell. Bar values are mean ± s.e.m. *P < 0.05. PAS, periodic acid Schiff; SD, Sprague Dawley; SHRSP, spontaneously hypertensive stroke-prone; VSMC/EC; ratio of vascular smooth muscle cell numbers to endothelial cell numbers.
In agreement with previous findings,11 the major nitrated protein was a 45 kDa band in kidney homogenates of SHRSP (Figure 3a). The mean density of this band was lower in sulforaphane-treated SHRSP than vehicle-treated SHRSP (P = 0.05, Figure 3b). A similar tendency was observed in SD rat kidneys (Figure 3).
Nitrated proteins in SHRSP and SD kidneys. (a) Nitrated protein expression in kidneys of corn oil vehicle or sulforaphane-treated SHRSP or SD rats. Bands represent a 45 kDa protein. (b) Bars represent percentage change in expression compared to the vehicle-treated group of the same strain. Band density was normalized to β-actin as loading control. Bar values are given as mean ± s.e.m. *P < 0.05. SD, Sprague Dawley; SHRSP, spontaneously hypertensive stroke-prone.
Induction of γ-glutamyl-cysteine ligase catalytic subunit
γ-Glutamyl-cysteine ligase is a phase 2 enzyme that is rate-limiting for glutathione synthesis. Glutathione plays a very central role in many oxidant-scavenging mechanisms.16 As can be seen in Figure 4, vehicle-treated SHRSP kidneys had significantly lower γ-glutamyl-cysteine ligase catalytic subunit (GCLc) protein content compared to vehicle-treated SD. Sulforaphane significantly increased GCLc in SHRSP kidneys but had no significant effect on GCLc protein in SD.
Western blot data for γ-glutamyl-cysteine ligase catalytic subunit in kidneys of SHRSP and SD rats given corn oil vehicle or sulforaphane in corn oil. The top of the figure gives a representative blot. The lower part of the figure gives quantitative information (means ± s.e.m., n = 3/group) of the amount of protein present normalized to actin. *P < 0.05. SD, Sprague Dawley; SHRSP, spontaneously hypertensive stroke-prone.
The spontaneously hypertensive rat (SHR) is the most widely used animal model for hypertension and it is a good model for primary genetic hypertension.14 SHRSP, which is the animal model we used in this study, is a substrain of SHR that develops hypertension more rapidly than SHR and is a good model for cardio and cerebrovascular diseases and hypertension.14 We have been using this strain to study the effects of broccoli sprouts on hypertension.11 Similar to those previous studies using broccoli sprouts, here we have shown that sulforaphane not only reduced blood pressure of these animals but also prevented vascular remodeling associated with hypertension, accompanied by changes in protein nitration and genomic DNA methylation.
Renal cross transplantation studies have shown that the kidney is the primary organ responsible for hypertension in SHR.17 SHR kidneys have pathophysiological18 and histopathological19 abnormalities that predispose the animal to hypertension. These abnormalities are present from a very young age18,19 and therefore precede hypertension. Glomerular sclerosis and atrophy,20 tubular atrophy,21 interstitial infiltration and fibrosis,21 and vascular tunica media thickening20,21 are the histopathological lesions that have been described in literature with regard to SHR. Even though, we did not observe any tubular atrophy or interstitial infiltration probably due to the relatively young age at euthanasia, we observed sclerosis of the glomeruli and marked thickening of arterial/arteriole walls in vehicle-treated SHRSP. Thickening of the vascular walls in vehicle-treated SHRSP is due to the proliferation of VSMC as demonstrated by increased numbers (Figure 2). Sulforaphane prevented the increase in vascular wall thickness.
Our findings confirm earlier reports that SHRSP rats have increased wall:lumen ratio compared to normotensive rats.22 There are reports indicating that this pathology was corrected by angiotensin-converting enzyme inhibitors,23 but not with hydralazine,24 even though both of these treatments reduced blood pressure. Therefore reduction of blood pressure does not necessarily correct vascular pathology and, thus, it is believed that vascular pathology is not the consequence but may be a cause for hypertension in SHR.22,23
SHR have constitutive oxidative stress especially in kidneys.18 Nicotinamide adenine dinucleotide (phosphate) oxidase is thought to be the primary source of ROS in SHRSP kidneys and this enzyme complex is elevated in SHRSP kidneys compared to normotensive rats.18 ROS react with nitric oxide, which is an essential vasodilatory agent, and reduces its availability.25 Oxidative stress can also induce vascular inflammation and VSMC growth.26 Therefore, oxidative stress can induce a chain of events that might potentially produce the pathophysiological and histopathological abnormalities needed to cause hypertension. Consequently, oxidative stress is believed to be the primary abnormality affecting the kidneys of SHRSP that eventually leads to hypertension.18
Sulforaphane is a potent inducer of phase 2 enzymes such as quinone reductase, glutathiones S-tranferase, and γ-glutamyl-cysteine ligase.16 Induction of such enzymes elevate endogenous antioxidant mechanisms and therefore leads to amelioration of oxidative stress.16 We have demonstrated that γ-GCLc is reduced in SHRSP compared to SD. Furthermore, sulforaphane administration normalizes this protein to levels comparable to that seen in SD. Induction of phase 2 protein genes decreases oxidative stress. Consequently, the resultant reduction of ROS available for reacting with nitric oxide reduces nitrative stress also.18 This is evident in our experiment too where we saw a reduction of nitrated protein content in kidneys of animals given sulforaphane. This reduction of nitrated proteins can be taken as an indirect evidence for decreased oxidative stress in these animals. Therefore, sulforaphane appears to have corrected the primary abnormality associated with hypertension in SHRSP kidneys and as a result prevented the vascular pathology and hypertension.
ROS are also known to oxidize nucleotides.10 Oxidation of deoxyguanosine to 8-oxo-deoxyguansine has been well-documented.10 5-methyl-deoxycytosine can be oxidized to 5-hydroxymethyl deoxycytosine and this conversion can affect the recognition of methylated sites in DNA.10 Oxidation of 5-methyl-deoxycytosine can also result in C→T mutations.27 In addition to having a direct effect on methylated deoxycytosines in DNA,10,27 oxidative stress-associated inflammatory changes are also thought to influence DNA methylation.28 Therefore, it is not surprising that vehicle-treated SHRSP in our experiment, which is oxidatively stressed, had lower levels of globally methylated DNA than vehicle-treated SD rats. Elevation of global methylation level in sulforaphane-treated SHRSP to a level similar to SD rats probably resulted from the amelioration of oxidative stress by sulforaphane.
Mutations,29 single nucleotide polymorphisms,30 fetal programming,31 and promoter methylation4 could result in genetic changes that lead to hypertension. The former two are permanent changes to the genetic code whereas the latter two are related to DNA methylation. Fetal stresses such as maternal protein restriction or glucocorticoid administration32 during fetal growth leads to development of hypertension later in life due to DNA methylation aberrations in angiotensin II type-1b receptor and glucocorticoid receptor gene promoters respectively. Furthermore, epigenetic regulation of HSD11B2, which is a gene causally-linked with hypertension, by specific promoter methylation has been demonstrated.4 Therefore it is reasonable to assume that specific DNA methylation plays a role in the development of hypertensive phenotype. But does nonspecific (global) DNA methylation also has a role in the development of hypertensive phenotype?
Methylation of DNA occurs in CpG dinucleotides.33 CpG dinucleotides in repetitive DNA sequences are heavily methylated but most of the CpG dinucleotides in unique sequences are not.34 Therefore, repetitive DNA sequences comprise most of the DNA methylation in the genome and thus account for most of the methylated deoxycytosines. Do these untranslated repetitive sequence methylations have anything to do with hypertension?
Molecular pathogenesis of cancer is similar to hypertension in several ways. Global methylation aberration is an independent pathological factor that is considered as an early event in carcinogenesis of many cancers. It has been shown that global DNA demethylation increases with age and precedes the development of genetic changes in gastrointestinal cancer.35 Furthermore, progressive DNA hypomethylation is observed in senescent tissue.6 Therefore, due to the age-related nature and involvement of multiple genetic defects in hypertension, global DNA methylation changes might also play a role in hypertension pathogenesis. Indeed, Lund et al.36 showed that changes in global DNA methylation pattern is an early marker of atherosclerosis in mice.
Repetitive sequences in subtelomeric regions are heavily methylated and any loss of methylation will change telomere length and increase telomere recombination and dysfunction.37 Therefore, several workers have suggested a role for telomere dysfunction in molecular pathogenesis of hypertension.38 In fact, Cao et al.39 reported an increase in telomerase activity in aorta of SHRs and related this to the increased proliferation of aorta-derived VSMC in culture. Therefore, these reports, taken together with our results, are suggestive of a possible involvement of global DNA methylation in molecular pathogenesis of hypertension. Indeed, a recent publication demonstrating lower levels of global DNA methylation in blood leukocytes of hypertensive patients gives credence to our hypothesis.40
Global DNA methylation data cannot be compared with gene specific DNA methylation data nor can it be used as an indication of promoter methylation of specific genes. As described above, most of the methylated deoxycytosines are in repetitive sequences, not related to any specific genes. Therefore, these data should be considered separately from promoter specific methylation and our intention is to point out the possible role played by global phenomena such as DNA methylation in the pathogenesis of hypertension. Sulforaphane administration reduced blood pressure and rectified other pathological abnormalities associated with hypertension such as nitrative/oxidative stress, VSMC proliferation and renal resistant artery wall:lumen ratio in SHRSP rats. Reciprocal changes of global DNA methylation levels are suggestive of an association of this parameter with pathogenesis of hypertension.
Supplementary material is linked to the online version of the paper at http://www.nature.com/ajh
Supplemental Table 1
Supplementary Figure 1
This research was supported by a postdoctoral research fellowship to the first author by Saskatchewan Health Research Foundation and by Canadian Institutes of Health Research/Regional Partnership grants.
Disclosure: The authors declared no conflict of interest.