Honokiol ameliorates angiotensin II‐induced hypertension and endothelial dysfunction by inhibiting HDAC6‐mediated cystathionine γ‐lyase degradation

Abstract Hypertension and endothelial dysfunction are associated with various cardiovascular diseases. Hydrogen sulphide (H2S) produced by cystathionine γ‐lyase (CSE) promotes vascular relaxation and lowers hypertension. Honokiol (HNK), a natural compound in the Magnolia plant, has been shown to retain multifunctional properties such as anti‐oxidative and anti‐inflammatory activities. However, a potential role of HNK in regulating CSE and hypertension remains largely unknown. Here, we aimed to demonstrate that HNK co‐treatment attenuated the vasoconstriction, hypertension and H2S reduction caused by angiotensin II (AngII), a well‐established inducer of hypertension. We previously found that histone deacetylase 6 (HDAC6) mediates AngII‐induced deacetylation of CSE, which facilitates its ubiquitination and proteasomal degradation. Our current results indicated that HNK increased endothelial CSE protein levels by enhancing its stability in a sirtuin‐3‐independent manner. Notably, HNK could increase CSE acetylation levels by inhibiting HDAC6 catalytic activity, thereby blocking the AngII‐induced degradative ubiquitination of CSE. CSE acetylation and ubiquitination occurred mainly on the lysine 73 (K73) residue. Conversely, its mutant (K73R) was resistant to both acetylation and ubiquitination, exhibiting higher protein stability than that of wild‐type CSE. Collectively, our findings suggested that HNK treatment protects CSE against HDAC6‐mediated degradation and may constitute an alternative for preventing endothelial dysfunction and hypertensive disorders.


| INTRODUC TI ON
Hypertension (high blood pressure) confers high risk for various types of diseases afflicting elderly individuals with chronic hypertension, representing a leading cause of cardiovascular disorders including stroke and heart failure. 1,2 In turn, the dilation and constriction of arterial blood vessels serves as a critical factor regulating blood pressure. 3 Notably, the gaseous transmitter hydrogen sulphide (H 2 S) is recognized as an endogenous, physiological vasodilator that contributes to lower blood pressure. 3,4 Moreover, H 2 S exhibits anti-inflammatory and anti-oxidative activities and mediates protein S-sulphydration. [4][5][6] Cystathionine γ-lyase (CSE), encoded by the CTH gene, is a major H 2 S-producing enzymes. 7 CSE (also termed γ-cystathionase) and cystathionine β-synthase, which together constitute the transsulphuration pathway, mediate successive metabolic conversions of homocysteine and cysteine via the intermediate product cystathionine. 8 Mutations in CTH cause the γ-cystathionase deficiency syndrome cystathioninuria, an autosomal recessive genetic disorder, whereas CSE deletion results in hypertension and atherosclerosis with endothelial dysfunction. 7,9,10 Alternatively, numerous studies have demonstrated that CSE-mediated H 2 S generation induces endothelium-dependent vasodilation and improves cardiovascular function and integrity. 11,12 Histone deacetylase 6 (HDAC6) is implicated in the pathophysiology of hypertension-related vascular diseases. 13,14 HDAC6 plays a significant role in the cardiac dysfunction mediated by angiotensin II (AngII), an inducer of hypertension. 15,16 An increased HDAC6 expression mediated by the atherogenic factor oxidized low-density lipoprotein impaired CSE function and vasorelaxation. 17 In similar contexts, HDAC6 has been proposed as a therapeutic target for treatment of cardiovascular diseases. [15][16][17][18][19] We also recently reported that tubastatin A, an HDAC6-specific inhibitor, could increase CSE acetylation and enhance its protein levels and H 2 S production, thereby helping to attenuate the vasoconstriction and hypertension induced by AngII. 20 Moreover, HDAC6, a member of the class IIb HDAC family, exhibits distinct characters compared to other HDAC family members in that it has unique specificity for protein deacetylation of non-histone substrates such as α-tubulin, cortactin and heat-shock protein 90. [21][22][23] Honokiol (HNK) is a natural product isolated from the phenolic extracts of the plant Magnolia officinalis. Accumulating evidence indicates that HNK potently suppresses oxidative and inflammatory responses. 24,25 Such beneficial, therapeutic capacities of HNK are mediated by its broad functionalities to modulate a wide range of signalling proteins. HNK also serves as a phytochemical to prevent cardiovascular diseases such as cardiac hypertrophy by activating mitochondrial sirtuin-3 (SIRT3), a member of the sirtuin family that belongs to class III HDACs. 24,26,27 It appears that both H 2 S and HNK retain cardioprotective activities. Currently, however, few studies have evaluated HNK effects on CSE-mediated H 2 S production. Moreover, relatively little is known about the regulatory role of HNK in AngII-induced hypertension. In this study, we examined the potential effects of HNK on hypertension, vascular function and H 2 S production using mice and aortic endothelial cells treated with AngII. We also investigated the molecular details regarding the role of HNK in regulating HDAC6 activity and CSE protein stability, as these are associated with CSE acetylation and ubiquitination.

| Blood pressure measurement
We measured blood pressure in the hypertensive mice before and 2 or 4 weeks after AngII osmotic pump insertion via a non-invasive CODA tail-cuff blood pressure monitor system (Kent Scientific, Torrington, CT, USA). 20 The volume pressure recording cuff was placed on the tail and changes in systolic, diastolic and mean blood pressure were measured on a heated platform set at 37°C. Mice were trained in a holder every day for 2 weeks and maintained in a quiet and dark location. The average of ten readings was used for the analysis.

| Force tension myography
We performed myographic measurements according to previously described procedures. 17,20 Briefly, mice had been treated with AngII and/or HNK for 4 weeks as described in Section 2.2 and aorta samples were then isolated. The isometric contractions were measured via a wire myograph system (model 620 DMT, Danish Myo Technology, Aarhus, Denmark). Aortic rings were challenged with 60 mmol/L KCl and the passive stretch was repeated after a wash with Krebs buffer, followed by equilibrium for 60 minutes. The vessels were pre-constricted with phenylephrine (1 μmol/L) for 15 minutes. We then monitored dose-dependent (0.1 nmol/L-10 μmol/L) responses to ACh followed by SNP after the washes and re-equilibrium.

| H 2 S measurement
H 2 S levels in HAECs treated with AngII and/or HNK were determined using the agar trapping method as previously described. 20,28 Briefly, H 2 S trapped as zinc sulphide was converted into methylene blue through chemical reactions. In the case of aorta samples, mouse aortas were isolated from control mice before treatments with AngII and/or HNK.

| Western blotting and immunoprecipitation (IP)
Cell lysate preparation, protein quantification, sodium dodecyl sulphate-polyacrylamide gel electrophoresis, Western blotting and IP were performed as described previously. 29 The for an additional 2 hours and washed 5 times with lysis buffer.

| Quantitative real-time PCR
Total RNA was purified using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized from the total RNA (1 μg) using ReverTra Ace qPCR RT Master Mix (Toyobo, Osaka, Japan). Then, quantita- 5′-AGGAGGCATTGCTGATGATCT-3′. All PCR samples were prepared in triplicate and the relative mRNA expression levels were normalized to GAPDH and determined using the 2 −ΔΔCt method.

| HNK-HDAC6 binding
HNK binding to HDAC6 was tested using the EpiQuik HDAC6 Assay Kit (EpiGentek #P-4046, Farmingdale, NY, USA) following manufacturer protocol with slight modification. HNK (final 10 μmol/L), rather than cell lysates, was directly added to the plate wells coated with the unique HDAC6 affinity substrate and incubated with the purified HDAC6 control protein (160 ng; 2 hours, 37°C). Alternatively, we applied a synthetic biotin-labelled HNK. 30 Following HAEC treatment with biotin-labelled HNK (5 μmol/L) for 12 hours, cell lysates (0.6 mg) were mixed with 25 μL streptavidin agarose resin (#20347, Thermo Fisher Scientific) overnight at 4°C, and then, the resulting pull-down samples were processed for immunoblot analysis.

| Statistical analysis
All experiments were performed independently at least three times with similar results. Band intensities of Western blots were measured using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Data shown in the graphs are presented as the means ± SEM Statistical significance was determined by one-way analysis of variance with Tukey's multiple comparison tests or the F I G U R E 1 Antihypertensive and vasorelaxatory effects of HNK on AngII-treated mice. Four groups of mice including untreated controls and mice infused with AngII via osmotic pump and/or HNK via peritoneal injection up to 4 wk were subjected to blood pressure measurements using a tail-cuff monitoring system. Seven mice per group were used. Changes in systolic (A), diastolic (B) and mean (C) blood pressure were measured prior to and 2 or 4 wk following the delivery of AngII and/or HNK. Values represent the means ± SEM. # AngII vs control, ## AngII vs AngII + HNK. **P < 0.01, ***P < 0.001. (D,E) Aortas were isolated from the mice treated with AngII and/or HNK in the same manner as described above and then processed for vascular reactivity measurements using wire myography. The dose-dependent responses to ACh (D) and SNP (E) were monitored in the indicated four mouse groups. Three to four aortic rings per mouse were used. The group data were analysed by logistic dose-response curves with four parameters using GraphPad Prism 7 software. The relaxation responses were prepared by nonlinear regression (curve fit) method and calculated as a percentage of tension after pre-construction using 1 μmol/L phenylephrine. Statistical significance of the indicated data sets was determined using least square fit and the extra sum-of-squares F test. Values represent the means ± SEM. # AngII vs control, ## AngII vs AngII + HNK. ***P < 0.001. Statistical significance was determined using the Holm-Sidak method indicated statistical methods using GraphPad Prism 7 software (La Jolla, CA, USA).

| HNK attenuates AngII-induced hypertension and vascular endothelial dysfunction
AngII, a potent hypertensive agent, induces vascular constriction. 15,31 To evaluate a potential effect of HNK on hypertension, we first performed tail-cuff measurements of blood pressure in in vivo AngII-and/or HNK-infused mice. AngII and HNK were chronically administered for up to 4 weeks via osmotic pump and peritoneal injection, respectively. Systolic ( Figure 1A), diastolic ( Figure 1B) and mean ( Figure 1C) blood pressure of mouse tails was increased at 2 and 4 weeks after AngII treatment, as expected. Conversely, AngIImediated blood pressure increase was blunted upon HNK co-administration ( Figure 1A-C).
We then examined whether HNK could modulate vascular reactivity by performing myographic measurements of ex vivo aorta rings that were isolated from the AngII-and/or HNK-treated mice used for the blood pressure measurements. Consistent with the ability of ACh to induce vascular relaxation by stimulating endothelial nitric oxide (NO) production, 12,17 ACh (0.1 nmol/L-10 μmol/L) dose-dependently induced the relaxation of aortic rings pre-constricted with 1 μmol/L phenylephrine in the control group whereas AngII significantly decreased the ACh-induced relaxation ( Figure 1D).
Conversely, HNK co-administration improved the AngII-attenuated vasorelaxation ( Figure 1D). We also monitored dose-dependent responses to the exogenous NO donor SNP to assess endothelium-independent relaxation of aorta rings. 17 In contrast to the results from ACh stimulation, AngII and/or HNK did not significantly affect the SNP dose responses ( Figure 1E). These results supported a role for HNK in suppressing AngII-induced endothelium-dependent vasoconstriction. NO produced by eNOS is a critical mediator of vasorelaxation. 32,33 Thus, we examined whether HNK could affect eNOS activity by detecting changes in phospho-eNOS (S1177), an active form of eNOS. 34 HNK treatment induced dose-dependent increases in phospho-eNOS levels ( Figure S1A). AngII down-regulated the phospho-eNOS levels ( Figure S1B), which was consistent with the previous results. 35 HNK was also shown to block the AngII-induced decrease in phospho-eNOS levels ( Figure S1B). These suggested a contribution of HNK to eNOS activation.

H 2 S lowers blood pressure and induces vasorelaxation whereas
AngII decreases H 2 S levels in HAECs and mouse aorta. 7,12,20 We thus tested whether HNK could affect H 2 S production using an agar trap method employing an in situ methylene blue assay that allowed sensitive detection of H 2 S generated from cultured cells. 28 In HAECs, the decreased H 2 S levels 12 hours after AngII treatment were recovered by 1-h HNK pretreatment (Figure 2A).  Figure 2D), supporting H 2 S measurement specificity. These data suggested that the antihypertensive and vasorelaxatory activities of HNK are associated with its effect on H 2 S production.

| HNK increases CSE protein levels by enhancing its stability against proteasomal degradation
We next examined the effects of AngII and HNK on CSE protein levels in HAECs. AngII treatment decreased CSE protein levels and whereas HNK abrogated this down-regulating effect, resulting in increased CSE protein levels ( Figure 3A). We also tested mouse aorta samples that were prepared 4 weeks after AngII and/or HNK administration as described in Figure 1 and observed similar changes in CSE protein levels ( Figure 3B). However, AngII and HNK minimally affected CSE mRNA levels, indicating that the CSE protein changes were not related to its transcription ( Figure 3C). Considering that HNK is a potent activator of SIRT3, 26 we examined whether SIRT3 is involved in HNK-mediated CSE regulation using cells subjected to SIRT3 knockdown by siRNA. The control and SIRT3 knockdown HAECs and HEK293 cells showed almost equal increases in CSE protein levels upon HNK treatment ( Figure 3D). Overexpression of SIRT3-FLAG in HEK293 cells did not markedly alter HNK-mediated CSE up-regulation ( Figure 3E). SIRT3 knockdown and overexpression efficiencies were confirmed by SIRT3 immunoblotting. These results indicated that HNK effects on CSE up-regulation are not mediated through SIRT3.
Both endogenous CSE ( Figure 3A) and transiently expressed CSE-Myc levels in HAECs were oppositely regulated by AngII and HNK ( Figure 3F). As the transfected CSE-Myc is constitutively expressed by the cytomegalovirus promoter activity independently of endogenous gene expression, these results further suggested that HNK regulates CSE at the protein level. HAEC pretreatment with the proteasome inhibitor MG132 blocked AngII-mediated CSE down-regulation and MG132 alone also up-regulated CSE protein levels ( Figure 3G). We thus tested HNK for its effect on CSE protein stability using standard CHX chase experiments.
Following HAEC treatment with CHX, an inhibitor of protein synthesis, CSE protein levels rapidly declined in control (DMSO only) conditions but remained relatively high in the presence of HNK during the experiment ( Figure 3H). CHX chase assay in HNK-treated HEK293 cells also showed a delay in time-dependent CSE degradation ( Figure 3I). These results indicated that CSE constitutes a protein substrate for proteasomal degradation under AngII-stimulated and steady state conditions, which can be weakened by HNK.

| HNK-mediated HDAC6 inhibition induces CSE acetylation, which contributes to CSE up-regulation
As HDAC6-mediated deacetylation of CSE promotes its proteasomal degradation, 20 we examined whether HNK might modulate CSE acetylation levels through HDAC6. We thus analysed the CSE IP products prepared from AngII-and/or HNK-treated HAEC by immunoblotting using an Ace-K antibody to detect acetylated lysine residues. Immunoblot and quantification results revealed that whereas AngII reduced CSE acetylation levels, they were significantly increased by HNK regardless of AngII presence ( Figure 4A,B). We assessed acetylation levels of α-tubulin, an HDAC6 substrate, using an acetylation-specific (K40) α-tubulin (Ace-α-tubulin) antibody. 21 Similar to the changes in CSE acetylation, the α-tubulin acetylation levels were reciprocally regulated by AngII and HNK ( Figure 4A,C). Additionally, AngII-and HNK-treated mouse aorta samples also showed a noticeable decrease and increase in α-tubulin acetylation levels, respectively ( Figure 4D).
Transfected CSE-Myc acetylation levels were much higher in cells overexpressing a deacetylase-deficient HDAC6 mutant than in cells overexpressing its wild-type HDAC6 ( Figure 4E). Moreover, the presence of coprecipitated HDAC6-FLAG in the Myc IP product suggested a possible interaction between CSE and HDAC6.
These results supported that HNK suppresses AngII-induced CSE and α-tubulin deacetylation mediated by HDAC6, leading to increased acetylation levels.  ELISA revealed that the HDAC6 binding activities were dose-dependently reduced by the presence of HNK ( Figure 5D), suggesting that HNK-HDAC6 binding interferes with the HDAC6 binding affinity for the substrate.
Alternatively, we performed a pull-down assay using a biotin-labelled HNK ( Figure 5E) recently synthesized for biological applications. 30  protein levels were not substantially affected by HNK ( Figure 5A).

F I G U R E 4
Increased acetylation levels of CSE and α-tubulin by HNK. A, HAECs were treated with AngII (100 nmol/L) for 12 h following pretreatment with HNK (5 μmol/L) for 1 h, as indicated. Cell lysates were immunoprecipitated with anti-CSE antibody, and then, the IP products and starting lysates were analysed by immunoblottings with the indicated antibodies including the anti-Ace-K and anti-Ace-αtubulin antibodies. (B,C) The ratios of Ace-K and CSE immunoreactivities in the CSE IP products in (A) and Ace-α-tubulin immunoreactivities normalized to α-tubulin immunoreactivities in cell lysates in (A), respectively, were quantified relative to the untreated control. Values represent the means ± SEM. *P < 0.05, **P < 0.01. D, Aorta samples were isolated from the AngII-and/or HNK-infused mice as shown in Figure 3B. Ace-α-tubulin immunoreactivities were quantified as in (C). Values represent the means ± SEM. **P < 0.01. E, HEK293 cells cotransfected with wild-type (WT) or inactive (DC) HDAC6-FLAG together with CSE-Myc, as indicated, were subjected to anti-Myc IP. The cell lysates and IP products were analysed by immunoblotting with the indicated antibodies. The Ace-K and Myc immunoreactivities in the Myc IP products and cell lysates, respectively, were quantified relative to those in the CSE-Myc only. Values represent the means ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001

F I G U R E 5 Inhibition of HDAC6 activity by HNK leading to CSE up-regulation. HAECs (A) or HEK293 cells (B)
were treated with the indicated concentrations of HNK for 12 h. C, HAECs were treated with 5 μmol/L HNK for the indicated times. (A-C) The resulting cell lysates were immunoblotted with the indicated antibodies. The immunoreactivities of CSE, Ace-α-tubulin and/or HDAC6 were normalized to those of α-tubulin and quantified relative to the untreated or zero time condition. Values represent the means ± SEM. D, Effects of HNK on HDAC6 binding to its unique HDAC6 affinity substrate were monitored using an ELISA-based EpiQuik HDAC6 Assay Kit as described in Materials and Methods. After 2-h incubation of purified HDAC6 protein in the absence or presence of the indicated concentrations of HNK in vitro, colorimetric changes were detected by measuring absorbance at 650 nm. Values represent the means ± SEM. *P < 0.05, **P < 0.01. E, The chemical structure of synthetic biotin-labelled HNK (molecular weight: 1060.28) having a benzophenone-containing linker. F, HAEC lysates were prepared 12 h after treatment with or without the biotin-labelled HNK (5 μmol/L) or the biotin liker alone as a control and then pulled down with streptavidin agarose resins. The starting cell lysates and resulting pull-down samples were subjected to immunoblot analysis with the indicated antibodies. G, In vitro HDAC6 activity was detected in the absence and presence of 2 μmol/L tubacin or the indicated concentrations of HNK by using a fluorometric HDAC6 activity assay kit as described in Materials and Methods. HDAC6 activity was quantified relative to the HDAC6 protein only condition. Values represent the means ± SEM. **P < 0.01, ***P < 0.001. H, Cell lysates from AngII-treated HAECs and aorta extracts from AngII-infused mice were prepared as shown in Figure 4A,D, respectively. Immunoreactivities of HDAC6 and CSE normalized to those of β-actin were quantified relative to the untreated control. Values represent the means ± SEM. **P < 0.01 In contrast, increased HDAC6 protein levels were accompanied with decreased CSE protein levels in AngII-treated HAECs and mouse aortas ( Figure 5H), which was consistent with the previous results. 20 These data indicated that AngII-mediated HDAC6 up-regulation drives CSE deacetylation resulting in its degradation, whereas these changes are blocked by HNK owing to its inhibitory effect on HDAC6 activity.

| CSE acetylation at K73 prevents its degradative ubiquitination
Based on the AngII-induced proteasomal degradation of CSE and the opposite regulation of CSE by AngII and HNK, we evaluated their potential effects on CSE ubiquitination using HEK293 cells cotransfected with CSE-Myc and HA-ubiquitin. As shown by the HA and Myc immunoblots of the anti-Myc IP products, increased polyubiquitination of CSE-Myc indicated by the multiple bands of higher molecular weight than wild-type CSE-Myc (approximately 45 kD) and its decreased levels, respectively, were clearly detectable under AngII treatment ( Figure 6A), supporting CSE degradation via the ubiquitin-proteasome system (UPS). Conversely, HNK strongly reduced CSE-Myc polyubiquitination levels regardless of AngII ( Figure 6A). As HNK increased CSE stability and its acetylation levels, these findings further suggested that the HNK-induced CSE acetylation acts to inhibit the UPS-mediated CSE degradation.
A previous proteomics study to identify ubiquitinated and acetylated protein substrates and their target lysine residues revealed that CSE undergoes both modifications and the K73 in CSE is a main target site. 36 We thus generated a CSE-Myc plasmid point mutant (K73R) and tested its acetylation and ubiquitination efficiency using transfection and anti-Myc IP as described in Figure 4E and Figure 6A, respectively. As shown by the Ace-K immunoblot of the Myc IP samples, K73R mutant acetylation levels were lower than those of its wild-type CSE irrespective of HNK ( Figure 6B).
The HA immunoblot of the Myc IP samples also showed relatively F I G U R E 6 Regulation of CSE degradation by acetylation and ubiquitination of CSE at the K73 residue. A, HEK293 cells were cotransfected with CSE-Myc and HA-ubiquitin (Ub). At 1-day post-transfection, cells were incubated in the absence or presence of AngII and/or HNK as described in Figure 3F. with those in wild-type-expressing cells ( Figure 6C). As these results indicated that K73 is a major site of both CSE acetylation and ubiquitination, we hypothesized that CSE K73R becomes more stable than the wild-type as the mutant is resistant to both acetylation and ubiquitination. To test this, we treated wild-typeor K73R-transfected cells with CHX. K73R protein levels were relatively higher than wild-type, as expected ( Figure 6D). The K73R protein levels in resting conditions (without CHX) also remained at high levels compared to the wild-type ( Figure 6D). Overall, these observations suggested that CSE is degraded through K73 ubiquitination, which can be facilitated by K73 deacetylation but prevented by its acetylation.  20 However, the precise mechanism by which HNK mediates HDAC6 inactivation remains to be elucidated.

| D ISCUSS I ON
Several lines of evidence have shown that HNK exerts a protective effect against various cardiovascular dysfunctions and that HNK-mediated SIRT3 activation contributes to its cardioprotective activities. 26,27 SIRT3, an NAD + -dependent protein deacetylase, has emerged as a critical regulator of metabolic processes in the mitochondria that contributes to maintaining cardiac function. 52,53 Conversely, the zinc-dependent deacetylase HDAC6 plays a pathologic role in heart and vessel dysfunctions such as hypertension, atherosclerosis, cardiac hypertrophy and fibrosis. [13][14][15][16][17][18] HDAC6 is structurally unrelated to and different from SIRT3 in terms of subcellular localization, activation mechanism and target deacetylation substrates. 46,54,55 Considering the differential aspects of HDAC6 and SIRT3, it might be expected that HNK ultimately leads to improvement of endothelial functions through HDAC6 inhibition in addition to SIRT3 activation. Here, our data showed that the HNKinduced CSE up-regulation was largely unaffected by SIRT3 knockdown or overexpression, supporting that the regulatory effects of HNK on CSE were SIRT3-independent.
Overall, our current results suggested that HNK contributes to restoring various physiological impairments, at least in part, by impeding the HDAC6-mediated CSE degradation. In conclusion, we demonstrated for the first time that HNK-mediated HDAC6 inhibition induces acetylation-dependent CSE stabilization, thereby attenuating AngII-induced hypertension and endothelial dysfunction.

ACK N OWLED G EM ENTS
This study was supported by research grants from the National Research Foundation of Korea funded by the Korean government to JHK (NRF-2017R1D1A1B03031468 and NRF-2020R1I1A3074045) and to SYL (NRF-2018R1A2B6004598).
This work was also supported by a grant from the Korea Health Industry Development Institute funded by the Ministry of Health & Welfare, Republic of Korea to SYL (HI18C0920).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
This manuscript does not contain sharable data.