Treatment for cerebral small vessel disease: effect of relaxin on the function and structure of cerebral parenchymal arterioles during hypertension

Abstract

We investigated the effect of hypertension on the function and structure of cerebral parenchymal arterioles (PAs), a major target of cerebral small vessel disease (SVD), and determined whether relaxin is a treatment for SVD during hypertension. PAs were isolated from 18-wk-old female normotensive Wistar-Kyoto (WKY) rats, spontaneous hypertensive rats (SHRs), and SHRs treated with human relaxin 2 for 14 d (4 μg/h; n=8/group) and studied using a pressurized arteriograph system. Hypertension reduced PA inner diameter (58±3 vs. 49±3 μm at 60 mmHg in WKY rats, P<0.05), suggesting inward remodeling that was reversed by relaxin (56±4 μm, P<0.05). Relaxin also increased PA distensibility in SHRs (34±2 vs. 10±2% in SHRs, P<0.05). Relaxin was detected in cerebrospinal fluid (110±30 pg/ml) after systemic administration, suggesting that it crosses the blood-brain barrier (BBB). Relaxin receptors (RXFP1/2) were not detected in cerebral vasculature, but relaxin increased vascular endothelial growth factor (VEGF) and matrix metalloproteinase 2 (MMP-2) expression in brain cortex. Inhibition of VEGF receptor tyrosine kinase (axitinib, 4 mg/kg/d, 14 d) had no effect on increased distensibility with relaxin, but caused outward hypertrophic remodeling of PAs from SHRs. These results suggest that relaxin crosses the BBB and activates MMP-2 in brain cortex, which may interact with PAs to increase distensibility. VEGF appears to be involved in remodeling of PAs, but not relaxin-induced increased distensibility.—Chan, S.-L., Sweet, J. G., Cipolla, M. J. Treatment for cerebral small vessel disease: effect of relaxin on the function and structure of cerebral parenchymal arterioles during hypertension.

Cerebral small vessel disease (SVD) is a leading cause of progressive cognitive decline and a major risk factor for stroke (1, 2). Cerebral SVD is clinically defined as the presence of diffuse white matter lesions, shown as hyperintensities on T2-weighted magnetic resonance imaging (3, 4). The severity of white matter lesions is strongly linked with hypertension (5–8). White matter lesions are more severe in patients with uncontrolled hypertension, and antihypertensive treatment reduces progression of white matter lesions (6–8). Moreover, increased severity of white matter lesions is associated with severity of cognitive impairment, a common consequence of cerebral SVD (9, 10). These epidemiological data strongly suggest that hypertension is an important risk factor of cerebral SVD that leads to white matter lesions and cognitive impairment.

Chronic hypertension has profound effects on cerebrovascular function and structure that may lead to cerebral SVD. Cerebral parenchymal arterioles (PAs) are a target for cerebral SVD because the white matter receives most of the blood supply from these largely unbranched, high-resistance vessels (11, 12). Histopathological studies of PAs in patients with cerebral SVD showed lumen narrowing and accumulation of fibrotic material in the vascular wall (11, 13). These vascular changes could reduce blood flow and cause ischemia to the white matter, especially during hypertension. Clinical studies have found that white matter lesions are associated with reduced white matter flow and hypoxic injury (14, 15). However, despite hypertension being a major risk factor for cerebral SVD, how hypertension affects PAs to promote cerebral SVD is largely unknown. Therefore, we studied the effect of hypertension on the function and structure of PAs.

Relaxin 2 (relaxin), a small peptide hormone, has been shown to have extensive cardiovascular effects (16). In renal and systemic small arteries from normotensive rats, relaxin causes vasodilation, reduces vascular resistance, and increases blood flow (17–19). In the cerebral circulation, our previous study showed that relaxin causes selective outward remodeling of PAs from normotensive rats, without affecting upstream arteries (20). The selective effect of relaxin on PA diameter suggests that relaxin may be a treatment of cerebral SVD by reversing inward remodeling, especially during hypertension. Thus, it is possible that hypertension-induced vascular and hemodynamic changes can be improved by relaxin treatment, leading to improved cerebral SVD. Therefore, we tested the hypothesis that relaxin can improve vascular function and structure of PAs during chronic hypertension.

The mechanism of relaxin-mediated outward remodeling in PAs remains unknown. Relaxin exerts its vascular effects through relaxin family peptide receptor 1 (RXFP1) in renal and systemic circulations (21). RXFP1 was not found to be expressed in PAs, but it is expressed in neurons of brain cortex (22). Thus, it is unclear whether relaxin acts on PA structure directly by activating RXFP1 on PAs or whether relaxin crosses the blood-brain barrier (BBB) to bind RXFP1 in neurons or other cell types of the brain cortex that are closely associated with PAs. Thus, we also investigated whether relaxin crosses the BBB and compared expression levels of relaxin receptors in PAs, middle cerebral arteries (MCAs), and brain cortex. Moreover, relaxin has been shown to up-regulate mediators that are known to promote vascular remodeling, including vascular endothelial growth factor (VEGF) and matrix metalloproteinases 2 and 9 (MMP-2 and MMP-9) (23–25). Therefore, we investigated whether relaxin up-regulates these mediators as a mechanism of relaxin-mediated remodeling of PAs.

MATERIALS AND METHODS

Animals and treatment groups

All procedures were approved by the Institutional Animal Care and Use Committee of the University of Vermont and conducted in accordance with the U.S. National Institutes of Health Guide for the Care and Use of Laboratory Animals. Female spontaneous hypertensive rats (SHRs) and female normotensive Wistar-Kyoto (WKY) rats (190–220 g, 14–16 wk, Charles River, Wilmington, MA, USA) were used for all experiments and housed in the University of Vermont Animal Care Facility. Female animals were used because our previous study showing that relaxin causes outward remodeling in PAs was performed using female animals and because females also have cerebral SVD, thus making this treatment applicable (5, 20). Animals were randomly selected and grouped as untreated female WKY rats (WKY-F group; n=8), untreated female SHRs (SHR-F group; n=8), female SHRs treated with recombinant human relaxin 2 (relaxin, 4 μg/h) for 14 d via osmotic minipump (Alzet 2ML2; Durect Corp., Cupertino, CA, USA; SHR-F-Rln group; n=8), and female SHRs cotreated with relaxin plus a nonselective VEGF receptor (VEGF-R) tyrosine kinase inhibitor axitinib (2 mg/kg, 2×/d in food, 14 d; SHR-F-Rln-Axi group; n=8). A separate SHR-F group was used to determine whether relaxin crosses the BBB by treating with relaxin (4 μg/h) for 14 d and measuring relaxin concentration in cerebrospinal fluid (CSF) and serum (SHR-F-Rln-CSF group; n=4). For the osmotic pump implant, animals were anesthetized with 3% isoflurane, and minipumps were implanted subcutaneously in the back of the neck. All animals received postsurgical analgesia (buprenorphine, 50 μg/kg, s.c.). Relaxin infusion was continuous for 14 d until animals were euthanized for isolated vessel experiments and tissue collection.

Determination of conscious blood pressure

Systemic blood pressures were determined in all groups of animals using a noninvasive tail-cuff method (20). For untreated WKY-F and SHR-F groups, blood pressure was measured on the day the animals were euthanized. For relaxin-treated groups, blood pressures were measured on d 0 (baseline, before pump implant), 2, 7, and 14. All animals were trained each day for 3 d prior to obtaining baseline blood pressure measurements. Training included placing the animal in the holder with tail cuff in place.

Determination of relaxin levels in serum and CSF

Animals were anesthetized with isoflurane (3% in oxygen) and decapitated, and trunk blood was collected for measurement of circulating relaxin. Serum was then collected and stored at −80°C for determination of relaxin levels by enzyme-linked immunosorbent assay (ELISA; human relaxin-2 quantikine ELISA kit; R&D Systems, Minneapolis, MN, USA) following manufacturer's instructions. Serum samples were diluted by 1:500 in the assay.

To determine the presence of relaxin in CSF, animals were anesthetized, and CSF was collected through a needle inserted into the cisterna magna (26). Blood from the same animals was also collected through a catheter inserted into the femoral artery to compare to the concentration of relaxin in CSF. Relaxin levels were determined in serum and CSF by ELISA, as described above, except CSF was not diluted.

Vessel preparation and pressurized arteriograph system

After collecting trunk blood, the brain was quickly removed and placed in cold physiological salt solution (PSS). PAs between the M1 and M2 region of the MCA territory and the MCAs from the same animals were carefully isolated from WKY-F, SHR-F, and SHR-F-Rln groups and mounted on glass cannulas. PAs were identified as branches off the MCA that penetrated into the brain tissue, as described previously (20). Only PAs were studied from the SHR-F-Rln-Axi group because relaxin had no effect on MCAs. The proximal cannula was connected to an inline pressure transducer and a servo-null pressure control system (Living Systems Instrumentation, Burlington, VT, USA) that allowed intravascular pressure to be maintained at a set pressure or changed at a variable rate. The distal cannula was closed during the experiment to avoid flow-mediated responses. Temperature and pH were continuously measured and maintained at 37.0 ± 0.5°C and 7.40 ± 0.05, respectively. Measurements of inner diameter (ID) and wall thickness (WT) were made via video microscopy (Living Systems Instrumentation).

Vascular reactivity and structural characteristics in PAs and MCAs

PAs and MCAs were equilibrated for 1 h, after which myogenic activity was determined by stepwise increases in pressure from 40 to 100 mmHg for PAs or 50 to 175 mmHg for MCAs. ID and WT were measured at each pressure once stable. Endothelium-dependent dilation was studied at 40 mmHg for PAs and 75 mmHg for MCAs by adding NS309 (10⁻⁸ to 10⁻⁵ M), an activator of small- and intermediate-conductance calcium-activated potassium (SK_Ca and IK_Ca) channels. NS309 was washed out of the bath, and a single concentration of the nitric oxide synthase (NOS) inhibitor l-nitro-N-arginine (l-NNA, 10⁻⁴ M) added to the bath. Constriction to NOS inhibition was used as an additional indicator of endothelial function. In the presence of l-NNA, sodium nitroprusside (SNP, 10⁻⁸ to 10⁻⁵ M) was added to the bath to assess endothelium-independent vasodilation. To obtain structural measurements, ID and WT were recorded at pressures between 5 and 200 mmHg for PAs or between 5 and 175 mmHg for MCAs in zero-calcium PSS containing ethylene glycol tetraacetic acid (EGTA), and outer diameter (OD) and distensibility were calculated.

Quantitative polymerase chain reaction (qPCR) analysis of target gene expressions

After dissecting PAs and MCAs for isolated vessel experiments, the remaining MCAs, PAs, and brain cortex from the MCA territory of 4–6 rats were collected and stored at −80°C in RNase inhibitor (1 U/μl; RiboLock; Fermentas, Glen Burnie, MD, USA) to determine mRNA expression levels of target genes using real-time qPCR methods. Target genes were assessed using Assays on Demand from Applied Biosystems (Grand Island, NY, USA). All primers were validated by the manufacturer for efficiency and did not detect homologs. All primers were designed across an exon-exon junction to avoid detecting genomic DNA, so no DNase treatment was required. Housekeeping gene β-actin was used in all qPCR experiments as a control. Standard techniques for real-time qPCR were performed by the Vermont Cancer Center DNA analysis facility at the University of Vermont, as described previously (27).

To compare expression levels of relaxin receptors in MCAs and PAs, RXFP1 and RXFP2 mRNA expression was determined. Because of no expression after 40 cycles of PCR, RNA of selected samples was amplified, and expression levels were determined again. RNA samples were amplified by Ovation Pico WTA System V2 (NuGEN Technologies, San Carlos, CA, USA), and the product cDNA was purified by an Agencourt RNAClean XP magnetic bead protocol (Beckman Coulter, Brea, CA, USA).

To investigate the mechanisms by which relaxin causes selective remodeling of PAs, mRNA expression of VEGF, MMP-2, and MMP-9 was compared between MCAs, PAs, and brain cortex. To determine whether relaxin activates peroxisome proliferator-activated receptor γ (PPARγ), expression levels of PPARγ target genes fatty acid binding protein 4 (FABP4) and plasminogen activator inhibitor 1 (PAI-1) were determined in mesenteric adipose tissue. Expression levels of PPARγ and its target genes PAI-1 and liver X receptor α (LXR-α) were also determined in brain cortex (28).

Immunohistochemistry of VEGF

Brain cortex from the MCA region from WKY-F, SHR-F, and SHR-F-Rln groups was collected and fixed in 4% formaldehyde for 48 h at 4°C and embedded in paraffin. Brain cortex was cut into 5-μm sections. After deparaffinization with ethanol, brain cortex sections were washed 3 times with 0.1 M phosphate-buffered saline (PBS) and placed in blocking buffer for 5 min (Dual Endogenous Enzyme Block; Dako, Carpinteria, CA, USA). The sections were incubated with primary antibody against VEGF (1:50; Santa Cruz Biotechnology, Santa Cruz, CA, USA) diluted in PBS with 1% bovine serum albumin for 30 min, followed by incubation with secondary antibody using Dako LSAB2 System horseradish peroxidase (HRP) for 20 min. Samples were then treated with 3,3-diaminobenzidine for 4 min, followed by hematoxylin for 30 s. After dehydration with ethanol, samples were covered with glass coverslips with xylene-based mounting medium. Imaging was performed using an Olympus BX50 upright microscope with ×20 objective (Olympus, Tokyo, Japan) and QImaging Retiga 2000R camera with QCapture Pro 6.0 (QImaging, Surrey, BC, Canada). Staining intensity was assessed using a semiquantitative scoring system from 0 (lowest staining intensity) to 5 (highest staining intensity) in a blinded fashion. Results were confirmed by grayscale histogram analysis of intensity (0: black to 255: white) using Adobe Photoshop CS4 (Adobe Systems, San Jose, CA, USA).

Drugs and solutions

All isolated vessel experiments were performed in PSS containing the following (in mM): 119.0 NaCl, 24.0 NaHCO₃, 4.7 KCl, 1.17 MgSO₄, 0.026 ethylene diamine tetraacetic acid, 5.0 CaCl₂, 1.18 KH₂PO₄, and 5.5 glucose and aerated with 5% CO₂, 10% O₂ and balanced N₂ to maintain pH at 7.40 ± 0.05. Zero-calcium PSS was made by eliminating CaCl₂, and a calcium chelator, EGTA (10 mM), was added. Relaxin was a generous gift from Corthera (San Carlos, CA, USA) and Novartis Pharmaceuticals (Basel, Switzerland). l-NNA and SNP were purchased from Sigma (St. Louis, MO, USA) and were made as stock solutions and stored at 4°C each week. NS309 was also purchased from Sigma and mixed with dimethyl sulfoxide, portioned into aliquots, and stored at −20°C until use.

Data calculations and statistical analysis

Percentage tone was calculated as percentage decrease in ID from the passive diameter at each intravascular pressure: [1 − (ID_active/ID_passive)] × 100. Percentage reactivity for vasodilators was calculated as [(ID_dose − ID_start)/(ID_passive − ID_start)] × 100. Percentage constriction for vasoconstrictor was calculated as (1 − (ID_dose/ID_start)) × 100. OD was calculated from measured ID and WT: ID + 2WT. Cross-sectional area (CSA) was calculated as π(OD/2)² − π(ID/2)². Distensibility at a given pressure was calculated as (ID_passive − ID_original)/(ID_original) × 100, where ID_original is defined as the passive ID determined at lowest pressure (5 mmHg). Data from qPCR studies were analyzed using the −2^ΔΔCt method, as described previously (29). Data were removed when the C_t values of technical replicates differed by >0.5.

All data are presented as means ± se. Differences between groups were determined with a Student's t test for 2 groups or 1-way ANOVA and a post hoc Newman-Keuls test for multiple comparisons for ≥3 groups, using Graph Pad Prism 5 (Graph Pad Software, La Jolla, CA, USA). Differences were considered significant at values of P < 0.05.

RESULTS

Effect of chronic hypertension on function and structure of PAs and MCAs

Physiological parameters of all groups of animals are shown in Table 1. Body weight and age were similar in WKY-F and SHR-F groups. At the time of the experiment, these animals were 18 wk of age. The SHR-F group has substantially higher systolic, diastolic, and mean pressures, confirming that these animals were a model of chronic hypertension.

Table 1.

Physiological parameters of all animals studied

Parameter	WKY-F	SHR-F	SHR-F-Rln	SHR-F-Rln-Axi
n	8	7	8	8
Weight (g)	213 ± 6	196 ± 5	209 ± 2	236 ± 2
Age (wk)	18.6 ± 0.2	18.4 ± 0.2	18.5 ± 0.2	18.1 ± 0.1
SBP (mmHg)	119 ± 3	165 ± 2*	169 ± 5*	173 ± 8*
DBP (mmHg)	82 ± 3	126 ± 2*	126 ± 5*	132 ± 7*
MBP (mmHg)	94 ± 3	138 ± 2*	144 ± 5*	146 ± 7*

SBP, systolic blood pressure; DBP, diastolic blood pressure; MBP, mean blood pressure.

^*P < 0.05 vs. WKY-F; 1-way ANOVA.

We determined the effect of hypertension on myogenic activity and endothelial function of PAs. PAs from the SHR-F group had smaller active IDs in response to increased pressures (Fig. 1A). Percentage tone of PAs was also increased in the SHR-F group, although this was not statistically significant (Fig. 1B). PA reactivity of NS309 was decreased in the SHR-F group, as suggested by significantly decreased vasodilation to NS309 at 10⁻⁶ M (Fig. 1C) and a higher EC₅₀ (2.02±0.85×10⁻⁶ M vs. 6.22±2.23×10⁻⁷ M in WKY-F). However, constriction to l-NNA and reactivity to the NO donor SNP were similar in PAs between the WKY-F and SHR-F groups, suggesting that NO-dependent vasodilation and smooth muscle responses to NO were not affected by hypertension (Supplemental Fig. S1A, B). These are the first results to show significant vasoconstriction and endothelial dysfunction of PAs during hypertension.

Figure 1.

Effect of hypertension on the function and structure of PAs. Graphs showing active ID (A), percentage tone (B), reactivity to NS309 (C), passive ID (D), passive OD (E), and passive distensibility of PAs (F) from WKY-F and SHR-F groups. Functionally, chronic hypertension decreased active IDs and decreased reactivity to the SK_Ca/IK_Ca activator NS309 with a higher EC₅₀ (2.02±0.85×10⁻⁶ vs. 6.22±2.23×10⁻⁷ M in WKY-F). Structurally, chronic hypertension decreased passive ID and OD, and increased distensibility. *P < 0.05 vs. WKY-F.

Hypertension-induced structural changes of cerebral penetrating vessels have been previously shown using histological methods (11, 13). A limitation of histological approaches to study vessel structure is that biomechanical properties cannot be determined. Thus, to better understand structural changes, we assessed structural and biomechanical alterations of isolated vessels in response to changes in pressures. Passive ID and OD were significantly decreased in PAs from the SHR-F group at all pressures tested (P<0.05), suggesting inward remodeling (Fig. 1D, E). However, CSA was not changed in PAs from the SHR-F group (Supplemental Fig. S1C). Together, these results demonstrate that hypertension caused inward eutrophic remodeling of PAs. Despite structural remodeling, distensibility was not significantly different in PAs from the SHR-F group (Fig. 1F), although it was decreased at all pressures compared to the WKY-F group. Inward eutrophic remodeling of PAs during hypertension was distinctly different from that of MCAs. Chronic hypertension caused decreased ID (Supplemental Fig. S2A) and increased CSA (Supplemental Fig. S2B) with decreased distensibility (Supplemental Fig. S2C) in MCAs.

Effects of relaxin treatment on MCAs and PAs during chronic hypertension

Our previous study showed that relaxin causes outward remodeling of PAs in normotensive rats (20), leading us to hypothesize that relaxin could potentially be a treatment for cerebral SVD during hypertension. The SHR-F group was treated with relaxin at 16 wk of age, and experiments were performed on the rats at 18 wk of age, which matched the age of rats in the untreated SHR-F and WKY-F groups. Relaxin treatment increased circulating relaxin to the level of mid- to late-pregnancy (Table 2 and ref. 18). Thus, circulating levels of relaxin were physiological in the treated animals. Relaxin treatment did not affect blood pressures of the SHR-F group (Table 1). Despite previous findings in small renal arteries, relaxin treatment did not affect myogenic reactivity (Fig. 2A) or myogenic tone (Fig. 2B) of PAs when compared to the untreated SHR-F group. Relaxin tended to increase reactivity to NS309 (ID increased by 67±8 vs. 39±11% in SHR-F at 10⁻⁶ M; P=0.058; Fig. 2C). Relaxin also tended to lower EC₅₀ of NS309 (7.48±1.95×10⁻⁷ vs. 2.02±0.85×10⁻⁶ M in SHR-F). Structurally, relaxin increased passive ID (Fig. 2D) and OD (Fig. 2E), demonstrating that relaxin enlarged PAs in the SHR-F group. The structural enlargement of PAs was largely due to relaxin-induced increased distensibility (Fig. 2E) rather than true remodeling (ID 42±3 vs. 44±2 μm in SHR-F at 5 mmHg; Fig. 2D). The effect of relaxin on upstream MCAs from the same animals was different from that of PAs. Relaxin had no effect on the structure of MCAs, with passive ID (Supplemental Fig. S2A), WT (Supplemental Fig. S2B), and distensibility (Supplemental Fig. S2C) of MCAs unaffected.

Table 2.

Relaxin concentration in serum and CSF from all groups of animals

Group	Serum (ng/ml)	CSF (ng/ml)
SHR-F, n = 7	−0.8 ± 0.2
SHR-F-Rln, n = 8	93 ± 11
SHR-F-Rln-CSF, n = 4	38 ± 5	0.11 ± 0.03
SHR-F-Rln-Axi, n = 8	114 ± 19

Figure 2.

Effect of relaxin on hypertension-induced vascular changes of PAs. Graphs showing active ID (A), percentage tone (B), reactivity to NS309 (C), passive ID (D), passive OD (E), and passive distensibility (F) of PAs from untreated SHR-F and SHR-F treated with relaxin (SHR-F-Rln) groups. Relaxin increased reactivity to the SK_Ca/IK_Ca activator NS309 with a lower EC₅₀ (7.48±1.95×10⁻⁷ vs. 2.02±0.85×10⁻⁶ M in SHR-F). Relaxin also increased passive ID, OD, and distensibility without affecting myogenic tone of PAs in hypertension. *P < 0.05 vs. SHR-F.

Expression of relaxin receptors in MCAs and PAs

The results above showed that relaxin selectively reverses inward remodeling by increasing distensibility of PAs, but not MCAs. We therefore investigated the underlying mechanisms of relaxin's selectivity for brain PAs. One hypothesis for the differential effect of relaxin on PAs vs. MCAs is that there is differential expression of the primary relaxin receptor, RXFP1, on these vessels. In a qPCR experiment comparing RXFP1 mRNA expression in MCAs, PAs, and brain cortex, there was no expression after 40 cycles in MCAs and PAs and low expression in brain cortex when compared to the housekeeping gene β-actin (Table 3). To confirm that RXFP1 was not expressed in PAs and MCAs, RNA was amplified to increase input of cDNA to the qPCR reaction. After amplification, we still found no expression of RXFP1 in PAs and MCAs (Table 3). We also determined expression of another relaxin receptor, RXFP2, in PAs and MCAs, because relaxin can also bind RXFP2 (21). We found that RXFP2 was also not expressed in PAs in amplified samples, and only one sample had expression of RXFP2 in MCAs. These results were not due to low cDNA input to the qPCR reaction, because cDNA input was increased by 1000-fold in amplified samples. Thus, despite a significant effect of relaxin on PAs, these results suggest that the selective effect of relaxin on PAs is not due to greater expression of relaxin receptors.

Table 3.

Threshold cycle (C_t) of qPCR for RXFP1 and RXFP 2 and β-actin (housekeeping) in MCAs, PAs, and brain cortex from untreated female SHRs

Gene	Unamplified			Amplified
	MCA	PA	Brain cortex	MCA	PA
β-Actin	27.83	28.16	27.25	16.99 ± 0.15	17.48 ± 0.55
RXFP1	ND	ND	36.00	ND	ND
RXFP2				33.84	ND

ND, not detected after 40 PCR cycles.

Determination of relaxin levels in CSF

One major difference between MCAs and PAs is that PAs are closely associated with other cell types in the brain, such as neurons that are known to express RXFP1, whereas MCAs lie on top of the brain in the subarachnoid space (22). Thus, it is possible that relaxin is acting on PAs through an interaction with other cells within the brain tissue that express RXFP1. However, relaxin must cross the BBB for this hypothesis to be correct. Therefore, we measured relaxin levels in CSF after systemic treatment for 2 wk and found that human relaxin was present in CSF, ∼0.3% of that in the circulation (Table 2). This is the first direct evidence that we are aware of showing that systemic administration of human relaxin crosses the BBB and may underlie the mechanism by which relaxin selectively affects PAs.

Role of VEGF on structural remodeling of PAs

Because relaxin is able to cross the BBB and RXFP1 is expressed in cells within the brain cortex, such as neurons that are closely associated with PAs (22), we hypothesized that relaxin acts to selectively remodel PAs through interaction with the brain tissue. In this manner, the effect of relaxin would be selective, because there is no cell type known to express RXFP1 close to MCAs that lie within the subarachnoid space. One target of relaxin that could promote remodeling of PAs is VEGF. VEGF is involved in the process of vascular remodeling, including cell proliferation and migration (30). In addition, relaxin has been shown to up-regulate VEGF expression in reproductive organs (25). Therefore, we determined expression of VEGF in brain cortex comparing the relaxin-treated SHR-F group with untreated WKY-F and SHR-F groups using qPCR and immunohistochemistry. Representative photomicrographs of VEGF staining in brain cortex from different groups of animals are shown in Fig. 3A. Relaxin treatment significantly increased VEGF staining when compared to the untreated SHR-F group (Fig. 3B). This result was confirmed by using Photoshop to analyze the grayscale histogram of the photomicrographs (data not shown). PCR analysis of mRNA expression also showed that relaxin increased VEGF mRNA expression in brain cortex (Fig. 3B).

Figure 3.

Effect of relaxin on expression of VEGF in brain cortex during hypertension. Graphs showing representative photomicrographs of VEGF immunohistochemistry (A), quantification of VEGF staining intensity (B), and qPCR analysis (C) of mRNA expression of VEGF in brain cortex from WKY-F, SHR-F, and SHR-F treated with relaxin (SHR-F-Rln) groups. Relaxin significantly increased VEGF immunoreactivity and mRNA expression in brain cortex during hypertension. *P < 0.05 vs. SHR-F.

As relaxin increased VEGF expression in the brain cortex, we determined the effect of VEGF-R inhibition and investigated the role of VEGF in relaxin-mediated remodeling of PAs in the SHR-F group. Age and body weight of the SHR-F-Rln-Axi group were similar to those of the relaxin-treated SHR-F group (Table 1). Circulating relaxin levels in the SHR-F-Rln-Axi group were similar to those of the relaxin-treated SHR-F group (Table 2). VEGF-R inhibition with axitinib did not alter blood pressure (Table 1). Axitinib cotreated with relaxin increased myogenic tone at 80 mmHg compared to the SHR-F-Rln group (Fig. 4A), but had no effect on active ID and NS309 (data not shown). Structurally, axitinib plus relaxin significantly increased ID (Fig. 4B) and CSA of PAs compared to the SHR-F-Rln group (Fig. 4C). Despite these structural changes, axitinib plus relaxin did not affect distensibility of PAs (Fig. 4D). These results suggest that inhibition of VEGF signaling in SHR with relaxin treatment caused outward remodeling. Thus, increased VEGF appears to prevent outward remodeling and hypertrophy in response to relaxin without affecting distensibility.

Figure 4.

Effect of VEGF-R tyrosine kinase inhibition with axitinib on relaxin-mediated remodeling in PAs during hypertension. Graphs showing percentage tone (A), passive ID (B), CSA (at 5 mmHg; C), and passive distensibility (D) of PAs comparing SHR-F-Rln and SHR-F cotreated with relaxin and axitinib (SHR-F-Rln-Axi) groups. Axitinib and relaxin cotreatment significantly increased tone of PAs when compared to relaxin alone. Structurally, axitinib plus relaxin increased passive ID and CSA of PAs compared to relaxin only. Despite these structure changes, axitinib plus relaxin did not affect relaxin-induced increased PA distensibility. *P < 0.05 vs. SHR-F-Rln.

Role of MMP-2 and MMP-9 on structural remodeling of PAs

Interaction between MMP-2, MMP-9, and VEGF is important in vascular remodeling. VEGF has been shown to up-regulate MMPs and enhances MMP-mediated smooth muscle migration, an important process in vascular remodeling (31). In addition, MMP-2 and MMP-9 expression are up-regulated by relaxin in small renal arteries (23, 24). Therefore, mRNA expression of MMP-2 and MMP-9 in PAs, MCAs, and brain cortex were studied as an underlying mechanism of relaxin-mediated increased distensibility of PAs. Relaxin significantly increased expression of MMP-2 in brain cortex (Fig. 5A), but not in PAs (Fig. 5B). Relaxin also tended to increase expression of MMP-9 in brain cortex (Fig. 5C). However, expression of MMP-9 was lower than that of MMP-2, and there was no comparison made in PAs because not enough PA samples expressed MMP-9. Thus, MMP-2, rather than MMP-9, is more likely involved in relaxin-mediated increase in PA distensibility.

Figure 5.

Effect of relaxin on expression of MMP-2 and MMP-9 in PAs and brain cortex during hypertension. Graphs showing mRNA expressions of MMP-2 in brain cortex (A), MMP-2 in PAs (B), and MMP-9 (C) in brain cortex comparing WKY-F, SHR-F, and SHR-F-Rln groups. Relaxin increased expression of MMP-2 in brain cortex only. *P < 0.05 vs. WKY-F.

Role of PPARγ on vascular remodeling of PAs

Relaxin causes outward remodeling in PAs through activation of PPARγ in normotensive rats (20). Therefore, we determined whether expression of PPARγ and its target genes were altered after relaxin treatment in brain cortex from hypertensive rats. We also determined PPARγ target gene expression in adipose tissue after relaxin treatment, where PPARγ is highly expressed. Expression on PPARγ target genes FABP4 and PAI-1 in adipose tissue were increased in the SHR-F group (Supplemental Fig. S3A, B). Relaxin decreased both FABP4 and PAI-1 expression in adipose tissue, although this was not statistically significant. In brain cortex, expression of PPARγ and PPARγ target genes LXR-α and PAI-1 was similar among groups (Supplemental Fig. S3C, D, E).

DISCUSSION

Chronic hypertension is one of the most important risk factors for cerebral SVD (5–8). Histological evidence from patients with cerebral SVD indicates that PAs have lumen narrowing, suggesting a major role for PAs in cerebral SVD (11, 13). White matter receives most of the blood supply from PAs, which are largely unbranched penetrating small vessels in the brain with limited collateral blood supply (11, 12). This unique vascular anatomy makes the white matter prone to hypoperfusion and ischemia, particularly when the lumen is smaller. Despite hypertension being an important risk factor for cerebral SVD, the effect of hypertension on PAs remains largely unknown. In this study, hypertension for 8 wk caused endothelium-derived hyperpolarizing factor (EDHF)-dependent endothelial dysfunction and increased myogenic tone, without affecting the NO-dependent vasodilator pathway. Moreover, chronic hypertension reduced lumen diameters of PAs, consistent with histological findings in patients with cerebral SVD (13). Inward remodeling reduces maximum vasodilation and vasodilator reserve of PAs that further promotes ischemia during reductions of blood pressure. However, hypertension did not affect WT and distensibility of PAs. Notably, the effect of hypertension on PAs appears to be unique from the upstream MCAs, which was found to be outward hypertrophic remodeling with decreased distensibility. Thus, studying PAs during hypertension in addition to pial vessels is essential for an understanding of cerebral SVD.

Our previous study showed that relaxin causes outward remodeling in PAs from normotensive rats (20). This study shows for the first time that relaxin caused outward remodeling of PAs during hypertension largely due to increased distensibility. The importance of this to cerebral hemodynamics is not clear; however, PAs with increased distensibility may have improved vasodilator reserve and better normalization to increased pulse pressure during hypertension. Increased pulse pressure is a major determinant of vascular hypertrophy, which contributes to changes in stiffness of the wall (32). In addition, relaxin tended to improve EDHF-dependent vasodilation of PAs during hypertension. Relaxin, therefore, may have two beneficial effects. First, relaxin-mediated increased distensibility may reduce the effect of increased pulse pressure and enhanced vasodilator reserve. Second, relaxin-induced enhanced EDHF-dependent vasodilation in PAs could be a compensatory mechanism for vasodilation in a disease state, such as stroke, when NO-dependent vasodilation is compromised (27). Together, these effects of relaxin could potentially decrease vascular resistance of PAs and prevent hypertension-induced hypoperfusion.

Relaxin interacts with RXFP1 to exert its effects in small renal and systemic arteries (21). However, this does not appear to be the case in PAs, because RXFP1 and RXFP2 are not expressed in PAs. Because RXFP1 is expressed in neurons (22), relaxin may activate RXFP1 in neurons (or other cell types in the brain cortex) and exerts its effects through a paracrine manner on PAs. The first evidence that we found to support this hypothesis was that relaxin crosses the BBB. Because human relaxin is not made in rat brains, only systemically administrated human relaxin was detected in CSF. Also, a human relaxin-specific ELISA was used, further suggesting that only systemically administrated human relaxin was detected in rat brain. Although a previous study indirectly suggested that relaxin may cross the BBB (33), this is the first direct evidence to show that circulating relaxin crosses the BBB (Fig. 6). It is unknown how relaxin passes through the BBB. We speculate that relaxin could not freely pass through tight junctions because of its large 6-kDa size. However, it is possible that relaxin crosses the BBB through a receptor-mediated transport similar to that of insulin, a similar protein hormone (34).

Figure 6.

Proposed mechanism for selective effects of relaxin on PAs during hypertension. RXFP1 and RXFP2 are not expressed on cerebral vasculature. However, systemic administration of human relaxin 2 crosses the BBB, thus gaining access to other cells in the brain cortex that express RXFP1, such as neurons. Relaxin activates RXFP1 in cells within the brain cortex and up-regulates VEGF and MMP-2 that are available to interact with the vasculature within the brain parenchyma. Activated MMP-2 in brain cortex may increase PA distensibility by reducing collagen in PA wall. The effect of relaxin is selective on PAs because there is no other cell type that is known to expresses RXFP1 associated with MCAs that are located within the subarachnoid space.

Relaxin may be acting in a paracrine manner on PAs through increased VEGF in neurons. VEGF is involved in processes of vascular remodeling, including cell proliferation and migration (30). Moreover, VEGF has been shown to enhance MMP-mediated smooth muscle migration, underlying the interaction between VEGF and MMP in vascular remodeling (31). The finding that VEGF expression was increased in brain cortex by relaxin led us to investigate the role of VEGF-R signaling inhibition on PA remodeling during hypertension. Interestingly, although VEGF-R inhibition did not appear to affect the increased distensibility of PAs by relaxin, relaxin plus axitinib caused hypertrophic outward remodeling of PAs. Because VEGF is known to increase production of NO by up-regulating endothelial NOS (35), it is possible that NO production is reduced because of VEGF inhibition, allowing smooth muscle growth (36). This increased smooth muscle growth could cause hypertrophic outward remodeling of PAs, as seen in this study when VEGF signaling is inhibited in the presence of relaxin.

Relaxin-induced VEGF expression appears to be selective for remodeling of PAs but not distensibility. Relaxin may increase distensibility of PAs by direct up-regulation of MMP-2 and/or MMP-9 (23, 24). MMP-2 expression was only increased by relaxin in brain cortex but not PAs. This compartmental difference in MMP-2 expression supports the concept that there is an interaction between brain cortex MMP-2 and PAs. The cell types that are involved in this process within the brain cortex are unknown, but may involve neurons, where RXFP1 and RXFP2 are known to express (22). MMP-2 is known to digest vascular wall collagen IV (37). Thus, increased MMP-2 expression in other cell types within the brain cortex may reduce PA wall collagen IV, leading to decreased collagen-to-elastin ratio, and, hence, increased distensibility (Fig. 6). A previous study supports the concept that relaxin decreases collagen IV and increases MMP-2 expression in cardiac and renal fibrosis (38). However, relaxin was also found to decrease collagen in small renal arteries, but stiffness was unaffected (39). This differential effect may be due to differences in vasculature beds. A limitation of this study is that we did not measure activity of MMP-2 or inhibition of MMP-2 with relaxin. Future studies are needed to determine increased MMP-2 expression in brain cortex in response to relaxin that may mediate the increase in PA distensibility.

The effect of relaxin in hypertensive rats was selective for PAs, similar to our previous study using normotensive rats (20). However, the effect of relaxin on PAs was distinctly different in normotensive and hypertensive conditions. Similar treatment of relaxin caused outward hypertrophic remodeling in PAs without affecting distensibility in normotensive rats, while it caused outward eutrophic remodeling of PAs in hypertensive rats largely due to increased distensibility. In addition, PPARγ is involved in relaxin-mediated remodeling of PAs in normotensive rats (20). Similar to our previous study, PPARγ target genes remained unchanged within the brain and cerebral vasculature in response to relaxin. However, in adipose tissue, where PPARγ is highly expressed, PPARγ target genes were significantly affected by relaxin only in the normotensive rats (20). The lack of significant activation of PPARγ by relaxin in hypertensive rats may be due to mutation of PPARγ coactivator-1 (PGC-1; ref. 28), and that may be one reason there is differential remodeling of PAs by relaxin between hypertensive and normotensive rats. Interestingly, relaxin-mediated remodeling of PAs in SHRs was similar to that of normotensive rats with VEGF-R signaling inhibition. This similarity may be explained by the interaction of PPARγ and VEGF. It has been reported that PPARγ agonists inhibit VEGF in endothelial cells (40). If this is the case, relaxin-induced PPARγ activation in normotensive rats could inhibit VEGF, which could be similar to relaxin and axitinib cotreatment in hypertensive rats, resulting in similar PA remodeling.

In the present study, we used young SHRs that were hypertensive for 10 wk. Despite smaller and more constricted PAs compared to normotensive animals, cerebral blood flow is similar compared to age-matched WKY rats (41). Therefore, it is likely that clinical features of cerebral SVD have not been developed at this age of SHRs. However, this study provides evidence that hypertension affects PA function and structure that can progress to hypoperfusion, and white matter damage occurs early in the disease process. Notably, early intervention to treat cerebral SVD may also be important to prevent cognitive decline later in life.

In summary, hypertension caused endothelial dysfunction and inward remodeling of PAs. Long-term effects of these vascular changes could lead to chronic hypoperfusion and white matter damage that are observed in cerebral SVD. Relaxin treatment reversed hypertension-induced inward remodeling of PAs, largely because of increased distensibility. Relaxin appears to cross the BBB and gain access to the brain cortex, an effect that may up-regulate VEGF and MMP-2 to increase distensibility of PAs (Fig. 6). Relaxin-induced increased distensibility of PAs during chronic hypertension may improve vasodilator reserve that can potentially prevent hypoperfusion.