Microbiota-governed microRNA-204 impairs endothelial function and blood pressure decline during inactivity in db/db mice

An impaired decline in blood pressure at rest is typical in people with diabetes, reflects endothelial dysfunction, and increases the risk of end-organ damage. Here we report that microRNA-204 (miR-204) promotes endothelial dysfunction and impairment in blood pressure decline during inactivity. We show that db/db mice overexpress miR-204 in the aorta, and its absence rescues endothelial dysfunction and impaired blood pressure decline during inactivity despite obesity. The vascular miR-204 is sensitive to microbiota, and microbial suppression reversibly decreases aortic miR-204 and improves endothelial function, while the endothelial function of mice lacking miR-204 remained indifferent to the microbial alterations. We also show that the circulating miR-122 regulates vascular miR-204 as miR-122 inhibition decreases miR-204 in endothelial cells and aorta. This study establishes that miR-204 impairs endothelial function, promotes impairment in blood pressure decline during rest, and opens avenues for miR-204 inhibition strategies against vascular dysfunction.

In this study, we test the hypothesis that miR-204 promotes endothelial dysfunction and impaired decline in blood pressure at rest during diabetes and investigate whether circulating miR-122 contributes to the vascular miR-204 upregulation.
www.nature.com/scientificreports www.nature.com/scientificreports/ isolated from db/db mice also had slightly impaired endothelium-independent relaxation, and that was significantly better in the aortic rings isolated from db/db-204 −/− mice (Suppl. Fig. 1e). The db/db-204 −/− mice had a better glycaemic control (Fig. 1g), and that can per se improve endothelial function 34 . Thus, to determine the vascular role of miR-204, we inhibited miR-204 ex vivo in the aortic rings isolated from the db/db mice. The ex vivo inhibition of miR-204 resulted in a robust decline in aortic miR-204 (Suppl. Fig. 1d) and a significant improvement in the endothelium-dependent vascular relaxation (Fig. 1h). Interestingly, even in the ex vivo set-up the miR-204 inhibition improved endothelium-independent relaxation (Suppl. Fig. 1f).
A downregulation of apelin-receptor (APJ) has been reported in the aorta and renal arteries of the db/db mice, and exogenous apelin activates endothelial nitric oxide synthase in the db/db mice 35,36 . Thus, we examined whether the improved endothelial function in db/db-204 −/− mice is associated with an increased endothelial level of apelin and APJ. The apelin (mRNA and protein) in the aorta was not different between db/db and db/ db-204 −/− mice (Fig. 2a,b, Suppl. Fig. 2a). Although we did not observe any difference in the APJ expression at the gene level (Suppl. Fig. 2b), the total APJ (protein) was higher in the aorta of db/db-204 −/− mice (Fig. 2d,e). As vascular endothelium constitutes a small part of the vasculature, the endothelial apelin and APJ was determined by co-immunostaining of the aortic sections with the endothelial marker von-Willebrand factor (vWF). The vascular endothelium of db/db-204 −/− mice had a significantly higher apelin and APJ compared to the db/db mice (Fig. 2c,f).

MiR-204 is crucial for microbial effect on the diabetes-associated endothelial dysfunction.
To test whether the intestinal microbiota-miR-204 axis is biologically relevant in diabetes-associated endothelial dysfunction, we provided antibiotics to the db/db and db/db-204 −/− mice, and a subgroup of antibiotics-treated db/db and db/db-204 −/− mice received fecal microbiota from the age-matched db/db mice. The microbial suppression in the antibiotics treated group was confirmed by the 16S RNA gene analysis in the freshly isolated fecal samples. Broad-spectrum antibiotics treatment led to over 95% decline in the fecal microbiota load and was restored by the recolonization (Fig. 3a). The antibiotics treated db/db mice had approximately 50% decline in the aortic miR-204 expression (Fig. 3b,c), and a significant improvement in the endothelial function (Fig. 3d). The antibiotics treated db/db mice in which microbes were recolonized had higher vascular miR-204 (Fig. 3b,c) and an impaired endothelium-dependent vascular relaxation (Fig. 3d). In contrast, the db/db-204 −/− mice had a better endothelial function (at baseline), which was neither improved by antibiotics treatment nor deteriorated by microbial recolonization (Fig. 3e) despite a comparable change in microbiota in response to antibiotics and recolonization (Fig. 3a). We did not observe any significant difference in the body weight and blood glucose level in the antibiotics-treated or recolonized db/db or db/db-204 −/− mice (Suppl. Fig. 3a,b). The aorta of antibiotics-treated db/db mice had better endothelium-independent relaxation compared to db/db mice (Suppl. Fig. 1g). The endothelium-independent relaxation of db/db-204 −/− mice was neither improved by antibiotics treatment nor deteriorated by the microbial recolonization (Suppl. Fig. 1h). As we observed that the db/db-204 −/− mice had higher expression of apelin and APJ in the endothelium, we asked if microbial suppression and recolonization also affect the apelin and APJ expression. We found that the microbial suppression led to a significant increase in the endothelial apelin and APJ expression, which reversed with the microbial recolonization ( Fig. 3f-i).
Circulating miR-122 is upregulated in db/db mice, and its systemic inhibition decreases vascular miR-204. Circulating miR-122 is upregulated during obesity/diabetes [23][24][25][26] , patients at risk of CVDs have a higher circulating miR-122 [27][28][29][30][31] , and miR-122 can regulate Stat3 activation which is a known repressor of miR-204 expression 32 . Further, the antibiotics-treated wildtype and db/db mice have lower circulating miR-122 (Suppl. Fig. 4a,b). Therefore, we investigated the potential involvement of miR-122 in the regulation of vascular miR-204. In our in-vitro studies in Human Umbilical Vein Endothelial Cells (HUVECs), the miR-122 inhibition increased Stat3 phosphorylation, decreased miR-204 expression, and led to an increase in the expression of miR-204 target gene Sirt1 (Fig. 4a,b). We inquired about the circulating level of miR-122 in the db/db mice, and it was significantly higher (Fig. 4c). The next logical step was to determine whether in-vivo inhibition of miR-122 can rescue vascular upregulation of miR-204 in the db/db mice. Therefore, we designed and synthesized the novel peptide nucleic acid (PNA)-based miR-122 inhibitor (PNA-122-I). PNAs are a synthetic DNA mimic that has charge-neutral N-(2-aminoethyl) glycine backbone instead of phosphodiester backbone that makes it efficient and resistant to the proteases/nucleases. PNAs can bind single-strand targets with high specificity/affinity, and their charge-neutral attributes make them translationally attractive. We found that the PNA-122-I with 3 lysines on both the 3′ and 5′ end had better binding with miR-122 in physiological salt solution compared to the one with single lysine at the 3′ end (Suppl. Fig. 4c) and was able to inhibit hepatic miR-122 in mice at 20 mgkg −1 day −1 (Suppl. Fig. 4d). The systemic inhibition of miR-122 at 20 mgkg −1 day −1 for seven days in the db/db mice led to a decrease in hepatic miR-122 (Fig. 4d), a tendency of decrease in the circulating miR-122 (Fig. 4e), and a significant decrease in the aortic miR-204 (Fig. 4f). We also measured the expression of other common vascular miRs such as miR-29b, miR-145, miR-148a, and miR-200b in the aorta of db/db mice receiving either saline or PNA-122-I and their level remained unaffected (Fig. 4f).
The db/db mice lacking miR-204 have an improved decline in blood pressure during inactivity. We used 20 weeks old mice to test the role of miR-204 in the blood pressure decline during inactivity.
Multiple cardiovascular parameters, including systolic arterial pressure, diastolic arterial pressure, heart rate, and mouse locomotor activity was simultaneously monitored at baseline as well as in the presence of nitric oxide synthase inhibitor (L-NAME) in the wildtype, db/+−204 −/− , db/db, and db/db-204 −/− mice. As anticipated, mean arterial pressure (MAP) was higher during the night and lowered during the day in all groups (Fig. 5a). The MAP during day or night was not different among the wildtype, db/+−204 −/− , db/db, and db/db-204 −/− mice (Suppl. Fig. 5a,b). The db/db mice had impaired circadian decline in the MAP, and the db/db-204 −/− mice www.nature.com/scientificreports www.nature.com/scientificreports/ The experimental replicate is shown as "n(N)" where 'n' represents the number of images and 'N' represents the number of mice. ns>0.05, *p < 0.05, and ***p < 0.001 vs. indicated group. The data is shown as mean, and the error bar represents s.e.m. One-way analysis of variance (ANOVA) followed by Tukey's test was performed. vWF, von-Willebrand factor. www.nature.com/scientificreports www.nature.com/scientificreports/ did not differ from db/db mice in terms of a change in MAP (Suppl. Fig. 5c). The mice were generally inactive from 4:00 AM to 3:00 PM (Fig. 5b), so the traditional 12 hour light versus dark cycle epochs (6:00 AM to 6:00 PM versus 6:00 PM to 6:00 AM) would not accurately capture the decline in the MAP during inactivity (rest). Therefore, we contrasted the MAP during activity quartiles 2 to 4 (Q ≥ 2, 25-100%) against the MAP during activity quartile 1 (Q1, 0-25%). The MAP at activity quartile Q1 or activity quartile Q ≥ 2 was not different among wildtype, db/+−204 −/− , db/db, and db/db-204 −/− mice (Fig. 5c,d). However, a decline in MAP during activity Q1 compared to MAP during activity Q ≥ 2 was significantly lower in the db/db mice compared to wildtype and db/+−204 −/− mice (Fig. 5e). The db/db-204 −/− mice had a significantly improved decline in the MAP during activity Q1 compared to MAP during activity Q ≥ 2 compared to the db/db mice (Fig. 5e). To understand the role of nitric oxide in the MAP decline, all the blood pressure parameters and mouse locomotor activity were recorded in the presence of L-NAME. The L-NAME treatment was associated with an increase in the MAP in all the groups (Suppl. Fig. 5d). The MAP during day and night was not different among the wildtype, db/+−204 −/− , db/db, and db/db-204 −/− mice (Suppl. Fig. 5f,g). However, as we observed in the absence of L-NAME, the db/db mice had a significantly lower decline in the MAP during the day (Suppl. Fig. 5h). L-NAME treatment did not affect the activity cycle, and mice were inactive from 4:00 AM to 3:00 PM (Suppl. Fig. 5e). Although the MAP during activity Q1 was lower in all groups compared to the MAP during activity Q ≥ 2, we did not observe a significant difference in the MAP during activity Q ≥ 2 or activity Q1 among wildtype, db/+−204 −/− , db/db, and db/ db-204 −/− mice (Suppl. Fig. 5i,j). However, even in the presence of L-NAME, we observed an improved decline in MAP (Q ≥ 2 vs. Q1) in the db/db-204 −/− mice compared to db/db mice (Suppl. Fig. 5k). The experimental replicate for 'g' and 'i' is shown as "n(N)" where 'n' represents the number of images and 'N' represents the number of mice. *p < 0.05, and ***p < 0.001 vs. indicated group. The data is shown as mean, and the error bar represents s.e.m. A one-way analysis of variance (ANOVA) followed by Tukey's test was performed. The significance of difference between two curves was analyzed by using global non-linear regression. Ach, acetylcholine; PE, phenylephrine; vWF, von-Willebrand factor. www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
One of the concerns in the attribution of ndHTN in the diabetes patients is increased nocturnal waking up (for reasons such as nocturia), which leads to the intermittent measurement of awake blood pressure and misclassification of patients as non-dippers. However, studies performed in patients at siesta (a short afternoon nap often not accompanied by the break) confirm the higher risk of ndHTN in T2D patients 37 . The Genome-Wide Association Studies (GWAS) have identified several genetic variants linked to hypertension, but cumulatively these variants explain less than 5% of the inter-individual variation in systolic blood pressure [38][39][40] . The GWAS studies highlight the importance of environmental factors such as diet and intestinal microbiota in the development of CVDs 9,41 . In this study, we investigated the role of miR-204 in vascular endothelial dysfunction and ndHTN during diabetes. Our results demonstrate that the absence of miR-204 rescues diabetes-associated endothelial dysfunction and impaired decline in the blood pressure at rest. www.nature.com/scientificreports www.nature.com/scientificreports/ The db/db mice overexpress miR-204 in the vasculature, and its absence rescues endothelial dysfunction despite hyperphagia and obesity. Jo et al. show that the miR-204 −/− mice at baseline does not have any improvement in the glucose disposal 42 . Similarly, we also did not observe any difference in the glycaemic control between db/+-204 +/− and db/+-204 −/− mice (Suppl. Fig. 3c,d). However, the db/db-204 −/− mice had significantly better glycaemic control compared to the db/db mice. As miR-204 has previously been reported to play a role in pancreatic β-cell functions 32,42,43 , the absence of miR-204 might have improved β-cell function and hence the glycaemic control. Previous studies in db/db mice 44 and humans 34,45 have shown that a tighter glycemic control improves endothelial function. Thus, the improved endothelial function in db/db-204 −/− mice could be a secondary effect of the slightly lower body weight and a better glycemic control rather than the vascular absence of miR-204 or a combined effect of all. However, an improvement in the endothelial function with ex vivo miR-204 inhibition in aortic rings isolated from the db/db mice (Fig. 1h) suggests that vascular miR-204 contributes to the diabetes-associated endothelial dysfunction.
The vascular miR-204 is regulated by the intestinal microbiota 19 , and to determine the role of the microbiota-miR-204 axis in diabetes-associated endothelial dysfunction, we suppressed intestinal microbiota by antibiotics and recolonized them by stopping antibiotics and providing fecal microbiota from the donor (age-matched db/db) mice. The microbial suppression decreased aortic miR-204 and improved endothelial function in the db/db mice, and this effect was reversed with microbial recolonization. In contrast, the db/db-204 −/− mice had a better endothelial function at baseline, which neither improved with microbial suppression nor was impaired by microbial recolonization. Previous studies show that abnormal gut microbiota in db/db mice contributes to the development of diabetes in the db/db mice 46 . The fasting mimetic-diet changes microbial composition and survival of the db/db mice 47 . Further, the dietary supplements such as mulberry fruit polysaccharide, the polyphenol-rich extract of Dendrobium Loddigesii, and capsaicin change the intestinal microbial composition and improve glycaemic control in the db/db mice [48][49][50] . In our previous studies, we examined the effect of antibiotics in the normal chow-fed, and high-fat diet-fed mice where endothelial dysfunction is relatively milder, and a more extended antibiotics treatment were required. However, the db/db mice develop ~60% impairment in the endothelial function by 12 weeks of age, and a two-week antibiotics treatment was enough to improve the endothelium-dependent vascular relaxation. Further, the microbial suppression for two weeks and microbial recolonization did not have a significant effect on the body weight and blood glucose level in the db/db and db/ db-204 −/− mice (Suppl. Fig. 3a,b). These data support that the protective effects of microbial suppression on endothelial function could be mediated through mechanisms other than improvement in the blood glucose level. The insensitivity of the endothelial function of db/db-204 −/− mice to microbial changes suggests that the presence of miR-204 is critical for the microbial effects.
The in-silico prediction tools at microrna.org for miR-mRNA interactions predicts that the mouse apelin (apln) is a weak putative target of miR-204. However, we did not see any significant difference in the apelin expression in the aorta at the gene as well as protein level between wildtype and miR-204 −/− mice (Fig. 2b, Suppl.  Fig. 2a). There are a couple of studies that report the downregulation of APJ in the vasculature of db/db mice. They also found that the exogenous apelin activates Akt, endothelial nitric oxide synthase, and improves the endothelial function of the aorta of db/db mice 35,36 . The apelin is also known to lower blood pressure through a nitric oxide-dependent mechanism and have angiogenic effects 51,52 . Our results demonstrate that the endothelial expression of apelin and APJ was significantly higher in db/db-204 −/− mice compared to db/db mice. Similarly, both apelin and APJ were significantly higher in the vascular endothelium of db/db mice receiving antibiotics, which was reversed by microbial recolonization. These data suggest that intestinal microbes and diabetes both upregulate vascular miR-204, decrease endothelial apelin, and impair endothelial function.
MiRs are known to regulate the expression of other miRs 27,53 . miR-122 silencing is reported to decrease SOCS3 (suppressor of cytokine signaling 3) via promoter methylation 54 , and SOCS3 is a potent inhibitor of Stat3 phosphorylation (Tyr-705), which in turn is a repressor of miR-204 32 . In addition, multiple independent studies establish that the circulating miR-122 is upregulated during obesity/diabetes [23][24][25][26] , miR-122 promotes inflammation [55][56][57][58] , and the patients at risk of CVDs have a higher circulating miR-122 [27][28][29][30][31] . Thus, anticipating that miR-122 is the potential regulator of vascular miR-204 we examined circulating miR-122 in the db/db mice, and it was significantly higher. In both in-vitro as well as in-vivo experiments we found that miR-122 regulates endothelial/ vascular miR-204 expression. The miR-122 is a conserved liver-specific miR in vertebrates and circulating miR-122 has emerged as a sensitive biomarker of hepatic injury [59][60][61][62] . Perhaps, the increased circulating miR-122 during diabetes upholds the multiple pathological pathways in the vasculature, and miR-204 upregulation is just one of them.
The endothelial dysfunction in diabetes precedes and predicts ndHTN 3,4,63,64 . However, the pathogenesis of ndHTN is affected by multiple organ systems, is tightly regulated, and develops at the later stages of diabetes. Our finding that the db/db mice do not have a higher MAP is consistent with the previous report 65,66 but differs from 67,68 . In the 12-h light/dark cycle we observed impairment in decline in MAP during the day in the db/db mice, but the db/db-204 −/− mice were not significantly different from the db/db mice. Our results showing that the db/db mice had an impaired circadian decline in MAP agree with previous reports 68,69 . The active and inactive phase of mice does not overlap with the 6:00 AM/6:00 PM light/dark cycle. The db/db mice are obese and are known to have decreased locomotor activity 68 , we also observed that the inactive phase was distributed across the 24 hrs. Therefore, we contrasted the MAP change with respect to the mouse locomotor activity and found that the db/db mice have a significant impairment in the MAP decline during inactivity, while the db/db-204 −/− mice have significantly better MAP decline during inactivity. An improved endothelial function of db/db-204 −/− mice might have contributed to a higher decline in the MAP during inactivity. Conversely, a shorter exposure of db/ db-204 −/− mice to a higher blood pressure could also improve endothelial function. Interestingly, an improved MAP decline during inactivity in the db/db-204 −/− mice, even in the presence of L-NAME suggests that the absence of miR-204 improves MAP decline at rest through nitric-oxide independent mechanisms as well. Though www.nature.com/scientificreports www.nature.com/scientificreports/ the underlying molecular mechanisms remain elusive these results demonstrate that the absence of miR-204 improves MAP decline during inactivity in the diabetic mice.
One of the limitations of this study is that whether intestinal microbes regulate circulating miR-122, and if yes how, remains unknown. Although we demonstrate that the miR-122 regulates vascular miR-204, it is not clear whether miR-122 deregulation is the primary mechanism through which intestinal microbes regulate vascular miR-204. The additional experiments determining the effects of miR-204/miR-122 inhibition and that of microbial suppression/recolonization on the MAP decline during activity and inactivity would have further ascertained the microbial role on ndHTN through miR-204/miR-122 dependent mechanisms. In the replenishment/transplant study, both db/db and db/db-miR-204 −/− mice received microbiome from the db/db mice. The idea was to determine whether the microbiome from db/db mice can impair endothelial function in both db/db and db/ db-204 −/− mice. However, we recognize that an alternative approach could be to transplant microbiota from the db/db-miR-204 −/− mice to the antibiotics-treated db/db and db/db-miR-204 −/− mice which will test whether an absence of miR-204 affects the intestinal microbial composition which in turn improves the glycaemic control and endothelial function. We also recognize that the present set of data is generated in the global miR-204 −/− mice which fail to discern whether the beneficial effects of miR-204 absence are of vascular and extravascular origin. The general downregulation of miR-204 might have a deleterious effect in a specific condition, and therefore, additional studies in the tissue-specific miR-204 knockout mice are warranted. Although multiple independent studies demonstrate miR-122 upregulation in patients with cardiovascular disorders 27,29,70,71 , the miR-122's role in blood pressure regulation remains unknown. Based on our data, we anticipate that the db/db mice lacking miR-122 will have an improved vascular function and blood pressure dipping during inactivity. However, this needs to be experimentally verified.
The miR-204's expression is regulated by intestinal microbiota and its absence rescues impaired blood pressure decline, which raises the potential effect of antibiotics on the blood pressure regulation. In experimental studies, the use of minocycline for four weeks was able to decrease the MAP in the Angiotensin-II infused rats 10 . The microbes are known to regulate the intestinal production of short-chain fatty acids (SCFAs) which are known to regulate the blood pressure through olfactory receptor Olfr78 72 . In a prospective study involving over 380,000 participants followed up for nine years the fiber intake, which is known to change the intestinal microbial demographics, was found to be associated with lower risk of cardiovascular disorders 41 . The dietary fiber is believed to act as prebiotics as they feed commensal bacterias and several epidemiological studies have shown that a high intake of fiber is associated with a reduced blood pressure 73,74 . The bacterial infection during pregnancy, as assessed by self-reported antibiotic usage, was associated with small rises in the blood pressure 75 . However, there is a paucity on the data regarding the use of antibiotics and probiotics in the context of an effect on microbial composition and blood pressure regulation. The present study does support the role of miR-204 in the diabetes-associated blood pressure deregulation, one of the limitation is that whether antibiotics mediates its effect on blood pressure through miR-204 remains unknown, and further study is warranted.
Multiple studies demonstrate the beneficial effects of microbial demographics-altering strategies on cardiovascular outcomes, and miRs are emerging as a molecular means of host-microbiota communication. The present set of data demonstrates that the vascular miR-204 and circulating miR-122 are upregulated during diabetes, absence of miR-204 rescues endothelial dysfunction and impaired decline MAP during inactivity, and circulating miR-122 at least in part regulates vascular miR-204 expression (Fig. 6). Moreover, the inhibition of miR-204 represents a therapeutically attractive approach for rescuing diabetes-associated vascular dysfunctions.

Animals.
All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Iowa and were carried out according to the National Institute of Health (NIH) guidelines. All the studies were performed in C57BL/6, miR-204 knockout (miR-204 −/− ), leptin-receptor mutant (db/db) and db/ db-miR-204 −/− (db/db-204 −/− ) mice (age 10-22 weeks) of both sexes. All the animals received a standard rodent www.nature.com/scientificreports www.nature.com/scientificreports/ diet. Intestinal microbiota was suppressed with a cocktail of broad-spectrum antibiotics (50 mg.kg −1 , vancomycin, 100 mg.kg −1 metronidazole, 100 mg.kg −1 ampicillin, and 100 mg.kg −1 neomycin) as previously described 19,76 . The antibiotic cocktail was administered to mice daily for 2 weeks by oral gavage (100 µL per 25 g mouse), using a stainless-steel feeding tube without prior sedation of the animals. For FMT, the fecal microbiota from the donor mice was transferred to recipient mice following antibiotic cocktail treatment. The fresh stool was collected every day from the db/db mice immediately upon defecation and was resuspended (1:10) in a sterile PBS solution containing 0.05% cysteine HCl. The sample was vortexed and then centrifuged at low speed (800 × g) for 2 min. A 200μl aliquot of the supernatant was orally gavaged to the recipient mice every day for two weeks. Mice were maintained in specific pathogen-free conditions at a central animal facility of the University of Iowa. The mice receiving a cocktail of antibiotics were monitored for weight loss, diarrhea and signs of discomfort. Suppression of gut microbiota was confirmed by bacterial 16S RNA gene analysis.
Cell culture. Human umbilical vein endothelial cells (HUVECs, Cat# CC-2519) were purchased from Lonza (Mapleton, IL USA) and cultured in growth factor supplemented EGM-2 (Lonza, Walkersville, MD USA). Vascular reactivity. The vascular reactivity was determined as previously described 19 . Briefly, the thoracic aorta was carefully dissected, rapidly removed and placed in ice-cold oxygenated Krebs-Ringer bicarbonate solution. The vessels were cleaned of fat and loose connective tissue and cut transversely into 5-10 rings (1.5-2.0 mm wide). The rings were placed in oxygenated organ chambers (95% O 2 /5% CO 2 ) filled with 6 mL of Krebs-Ringer buffer solution and maintained at 37 °C and pH 7.4. Each ring was suspended between two wire stirrups (150 µm) in 6 mL chambers of a myograph system (DMT Instruments, FL, USA). One stirrup was connected to a three-dimensional micromanipulator, and other to a force transducer. The mechanical force signal was amplified, digitalized and recorded (PowerLab 8/30). All concentration-effect curves were performed on arterial rings beginning at their optimum resting tone. This was determined by stretching arterial rings at 10 min intervals in an increment of 100 mg to reach optimal tone (~500 mg). One dose of KCl (60 mM) was administered to verify vascular smooth muscle viability. Endothelium-dependent and -independent vasorelaxation was determined by generating dose-response curves to acetylcholine (Ach,10 −9 -10 −4 M) and sodium nitroprusside (SNP,10 −9 -10 −4 M), respectively, on PE (10 −6 M)-induced pre-contracted vessels. Vasorelaxation evoked by acetylcholine and sodium nitroprusside was expressed as percent relaxation, determined by calculating the percentage of inhibition to the pre-contracted tension. Aortic rings not responding to the initial dose of KCl or showing auto-relaxation were excluded.
Transfection of oligonucleotides. The freshly isolated thoracic aortic rings from db/+ and db/db mice were transfected with oligonucleotides using Oligofectamine (Invitrogen). Oligos [miR-204 inhibitor and scrambled control (SC); 200 nM] were incubated with Oligofectamine (3 µL) at room temperature for 20 min before addition to the aorta in basal EGM (250 µL). The aortic rings were kept in the transfection mixture for 24 hrs (at 37 °C and 5% CO 2 ). After 24 hrs of transfection, the media was removed, and the aortic rings were used to determine the vascular reactivity and expression of miR-204. Transfection of miR-122-I into the HUVECs was performed using lipofectamine2000 (Invitrogen). The sequence of oligonucleotides used in the study is provided in Suppl. Table 1. RNA isolation and qPCR. RNA was isolated using Qiazol/Trizol as per the manufacturer's instructions.
MicroRNAs and RNAs were converted to cDNA using the qScript microRNA cDNA synthesis kit (Quantabio). Real-time qPCR for miR-204, APLN, and APJ were performed using Brilliant II SYBR Green RT-qPCR kit. As no internal control is universally accepted for the miRs, we used a battery of internal controls for the quantification of miR in the aorta comprising Gapdh/18 s/RNU6. We observed the most consistent results with Gapdh. To quantify serum miR we used a constant amount of serum (200μl), and the Ct values were not normalized using any internal control. The Gapdh was used as an internal control for the qPCR-based quantification of 'Apln' and 'APJ' expression quantification. Primer sequences are provided in Suppl. Table 1.
Total microbial load. Fecal bacterial DNA was isolated using the DNeasy PowerSoil Kit (Qiagen). The amount of DNA in each sample was quantified using nanodrop. Based on the nanodrop readings, each aliquot of the DNA sample was diluted to 1 ng/µL. To determine the microbial load, real-time qPCR for bacterial 16S RNA was performed using the Brilliant II SYBR Green RT-qPCR kit. Primer sequences are provided in Suppl. Table 1.
Histological processing and immunostaining. Sections (5 µM) of formalin-fixed paraffin-embedded tissues were heated (95 °C, 20 min) in citrate buffer (10 mM) for antigen retrieval, followed by incubation with primary antibodies. The working concentrations of anti-apelin (LSBio-LS-C149244), anti-APJ (Millipore-Sigma-ABD43), and vWF (Abcam-ab11713) antibodies for the immunofluorescence experiments were 1:100. Antigen-primary antibody complexes were probed with fluorescence-tagged secondary antibodies. Images were captured using a Zeiss confocal microscope (Model 710). To examine the change in endothelial expression of apelin and APJ the endothelial cells were cropped from the images based on the vWF staining using Photoshop CC software, and the ImageJ software was used to determine the intensity. In the intensity quantification analysis, the "n" represents the number of aortic images, while "N" represents the number of mice.
Synthesis of PNA anti-miR-122. The PNA-based antimiR was synthesized as previously described with minor modifications 77 . Boc-protected PNA monomers used for PNA synthesis were purchased from ASM Chemicals and Research (Hanover, Germany). PNAs were synthesized via solid-phase synthesis using MBHA (4-Methylbenzhydrylamine) resin and Boc-monomers (A, T, C, G). PNAs were then cleaved from the resin using a cleavage cocktail containing m-cresol: thioanisole: trifluoromethanesulfonic acid (TFMSA): trifluoroacetic www.nature.com/scientificreports www.nature.com/scientificreports/ acid (TFA) (1:1:2:6) followed by precipitation using diethyl ether. PNAs were purified and characterized using reverse-phase high-performance liquid chromatography (HPLC) and matrix-assisted laser desorption/ ionization-time of flight (MALDI-TOF) spectroscopy, respectively. The concentration of PNAs was determined using UV-vis spectroscopy, and the extinction coefficient of PNAs was calculated using the extinction coefficient of individual monomers (13,700 M −1 cm −1 (A), 6,600 M −1 cm −1 (C), 11,700 M −1 cm −1 (G), and 8,600 M −1 cm −1 (T)) of the sequence. The lysines at 3′ and 5′ end of PNAs increase their solubility. In addition to Watson Crick base pairing, the positively charged lysine amino acids of PNA improve electrostatic interactions with the negatively charged phosphate backbone of target miRs (miR-122).
Gel shift assays. Target miR-122 was incubated with PNAs in physiological ionic strength conditions (2 mM MgCl 2 , 150 mM KCl, 10 mM sodium phosphate; pH 7.4) at 37°C in a thermal cycler (T100, Bio-Rad, Hercules, CA) for 18 hrs. The samples were then separated on a pre-cast 10% nondenaturing polyacrylamide gel using 1x tris/borate/EDTA buffer (1xTBE). The gels were run at a voltage of 120 V for 35 min. After electrophoresis, gels were stained with SYBR-Gold (Invitrogen) in 1xTBE buffer for 2 min and imaged using a Gel Doc EZ imager (Bio-Rad, Hercules, CA).
Immunoblotting and In situ hybridization (ISH). Protein samples were prepared using sample buffer and resolved using 10% SDS-PAGE under reduced conditions and transferred onto nitrocellulose membranes. Membranes were blocked using 5% BSA followed by incubation with primary antibody overnight at 4oC (dilutions-1:1000 for p-Stat3, Stat3 and Sirt1 and 1:10,000 for Gapdh). Antigen-primary antibody complexes were incubated with HRP-conjugated secondary antibodies and visualized by immunoblot luminol reagent (Thermo, USA). Images were captured and quantified using the Image Lab (Biorad, USA) software. The ISH was performed as previously described with minor modifications 19 . Briefly, the deparaffinized and rehydrated aortic sections (6 µM) were treated with proteinase K (10 µg.ml −1 , 5 min, 37 °C). Sections were dehydrated using ethanol and incubated with pre-hybridization buffer, followed by a hybridization mixture containing either double digoxigenin (DIG)-tagged miR-204-5p probe or scrambled probe (at 20 nM for 72 hrs, 56 °C). Sections were washed with sodium saline citrate and incubated in blocking solution (5 mL PBS + 50 mg BSA + 100 µL sheep serum + 2.5 µL tween 20) for 15 min, followed by incubation with either Dylight 594-labelled goat anti-DIG antibody (working concentration-1:500, Vector laboratories) or alkaline phosphatase-conjugated anti-DIG Fab (fragment, antigen-binding; working concentration-1:500 in antibody dilution solution for 18 hrs at 4 °C). Following alkaline phosphatase-conjugated FAB binding, slides were incubated in a solution containing BCIP/NBT (5-bromo-4-chloro-3′-indolyl phosphate/nitro-blue tetrazolium) at 30 °C for 48 hrs. Slides were counterstained with 4,6-diamidino-2-phenylindole (DAPI) and dehydrated with alcohol and mounted in DPX. Images were captured by a charge-coupled device camera attached to the Zeiss microscope (Model BX61). Six images originating from at least two aortic rings were used for the quantification. The blue-purple color range was selected using Photoshop CS6, converted to grey and quantified using Image J software. were implanted during isoflurane-induced general anesthesia (Phoenix Scientific, St. Louis, MO), as previously described 78 . Flunixin meglumine (2.5 mg/kg, Phoenix Scientific) was administered subcutaneously at the time of anesthetic induction, and 0.5% bupivacaine (Pfizer, New York, NY) was applied to the wound margin. After a 7-day recovery period, arterial pressures, heart rate, and relative locomotor activity were recorded for 10 sec every 5 min for 60 hrs (encompassing 3 dark cycles and 2 light cycles). Following those baseline recordings, L-NAME (1 mg/ml, Sigma St. Louis, MO) was administered in the drinking water, and the previous experimental protocol was repeated.

Radio-telemetry.
Statistical analysis. Statistical analysis was performed using GraphPad Prism (Version 7.0). One-way analysis of variance (ANOVA) was used for multiple comparisons and posthoc analysis was performed with Tukey's test. Independent sample t-test was used to determine the significance of difference between the two groups. The significance of the difference between two curves was analysed by using global non-linear regression. Results were expressed as mean ± s.e.m and considered significant if p values were <0.05.