Page 1 of 32 Diabetes

Hyaluronidase 1 deficiency preserves endothelial function and glycocalyx

integrity in early streptozotocin-induced diabetes

Sophie Dogné,1 Géraldine Rath,2 François Jouret,3 Nathalie Caron,1 Chantal Dessy,2* and

Bruno Flamion,1*

1 Molecular Physiology Research Unit, NARILIS, University of Namur, Namur,

Belgium;

2 Pole of Pharmacology and Therapeutics, IREC, Université catholique de Louvain,

Brussels, Belgium;

3 Groupe Interdisciplinaire de Génoprotéomique Appliquée (GIGA), Cardiovascular

Sciences, University of Liège, Liège, Belgium.

* These authors contributed equally to the study

Corresponding author: Sophie Dogné, Laboratory of Physiology & Pharmacology, University

of Namur, 61 rue de Bruxelles, 5000 Namur, Belgium; Phone: +3281725660; Fax:

+3281724329; [email protected]

Short running title: HYAL1 deficiency protects the endothelium

Word count of main text (max 4000 words): 4052

Number of Tables: 1

Number of Figures: 7

Diabetes Publish Ahead of Print, published online May 31, 2016 Diabetes Page 2 of 32

ABSTRACT (max. 200 words): 193

Hyaluronic acid (HA) is a major component of the glycocalyx involved in vascular wall and endothelial glomerular permeability barrier. Endocytosed HYAL1 is known to degrade HA into small fragments in different cell types including endothelial cells. In diabetes, the size and permeability of the glycocalyx are altered. In addition, type 1 diabetic patients present increased plasma levels of both HA and HYAL1.

To investigate the potential implication of HYAL1 in the development of diabetesinduced endothelium dysfunction, we measured endothelial markers, endotheliumdependent vasodilation, arteriolar glycocalyx size, and glomerular barrier properties in wildtype and

HYAL1 knockout (KO) mice with or without streptozotocininduced diabetes. We observed that, 4 weeks after streptozotocin injections, the lack of HYAL1: 1) prevents diabetesinduced increases in soluble Pselectin concentrations and limits the impact of the disease on endotheliumdependent hyperpolarization (EDH)mediated vasorelaxation; 2) increases glycocalyx thickness and maintains glycocalyx structure and HA content during diabetes; 3) prevents diabetesinduced glomerular barrier dysfunction assessed using urinary albumin/creatinine ratio and urinary 70/40kDa dextran ratio. Our findings suggest that

HYAL1 contributes to endothelial and glycocalyx dysfunction induced by diabetes. HYAL1 inhibitors could be explored as a new therapeutic approach to prevent vascular complications in diabetes.

INTRODUCTION

The glycosaminoglycan (HA), or hyaluronan, is a major component of the extracellular matrix. HA mediates cellcell and cellmatrix interactions and plays key roles in cell migration, tumor growth and progression, inflammation, and wound healing (1). HA is synthesized at the plasma membrane by different HA synthases and degraded by a family of Page 3 of 32 Diabetes

endoglucosaminidases named , mainly HYAL1 and HYAL2 in somatic tissues

(2). HYAL1 is the only hyaluronidase present in human and mouse plasma (3). In all cell

types, its enzymatic activity occurs at pH levels <4.0, which requires the enzyme to undergo

endocytosis (4).

HYAL1 deficiency in humans is a rare disease; it is associated with bone erosions, synovitis,

and polyarthritis together with high plasma HA levels (5). A mouse model of HYAL1

deficiency showed HA accumulation in serum without gross abnormalities except for a loss of

proteoglycans in knee joints (6).

In the vascular network, HA is a major component of the endothelial glycocalyx alongside

heparan sulfate and chondroitin sulfatecontaining proteoglycans (7). In the glycocalyx, HA

binds to its receptor CD44 but has no covalent linkage and may freely exchange with the

bloodstream. The glycocalyx is recognized as a major factor in vascular physiology and

pathology; it contributes to shear force sensing and transduces these forces into intracellular

responses, such as NO release (7). The glycocalyx also acts as a regulator of vascular

permeability, a reservoir for various antithrombotic factors, and an antiadhesive barrier for

leukocytes (8).

Through in vivo perfusion of hyaluronidase, which removes all HA in the endothelial surface

layer, HA has been found to be essential to maintain glycocalyx integrity and functional

barrier (9). Hyaluronidase infusion also abolishes the NOdependent response to increased

shear stress in segments of pig iliac artery or dog coronary arteries but not the acetylcholine

induced NO production (10,11).

In patients with Type 1 and Type 2 diabetes, endothelial dysfunction appears to be a

consistent finding underlying the pathophysiology of macro and microvascular

complications, and therefore contributes to the increased mortality rates observed in the

diabetic population. Glycocalyx defects may play a central role in diabetes pathogenesis by Diabetes Page 4 of 32

contributing to the proinflammatory state implicated in impaired skin wound healing and atherosclerosis (12). Indeed, the glycocalyx itself is disturbed during both acute (13) and chronic hyperglycemia in man (14,15). In addition, both Type 1 and Type 2 diabetic patients have increased plasma HA levels (15,16) and hyaluronidase activity (14,15).

To date, the implication of elevated plasma HYAL1 and/or HA levels in the pathogenesis of diabetes remains unexplored. As plasma HYAL1 is endocytosed into endothelial cells and could therefore modulate their function possibly through glycocalyx regulation, we decided to investigate the potential role of HYAL1 in the development of diabetesinduced endothelial dysfunction. To this aim, diabetes was induced in wildtype and HYAL1 KO mice using streptozotocin (STZ) injections, and endothelialdependent vasorelaxation, circulating endothelial markers, and the size and HA content of the glycocalyx were measured.

RESEARCH DESIGN AND METHODS

Animals. All experiments were performed on 7 to 9weekold male C57Bl/6 (WT) mice and

B6.129X1Hyal1tm1Stn/Mmcd (Hyal1-/- or KO) mice obtained from MMRRC (Mutant Mouse

Regional Resource Centers, USA) backcrossed onto a C57Bl/6 genetic background for 9 generations. The animals were fed regular chow and tap water ad libitum. All experiments were approved by the local animal ethics committees of the University of Namur and the

Université catholique de Louvain (2012/UCL/MD/004).

Type 1 diabetes was induced by 5 daily intraperitoneal injections of 55 mg/kg STZ in 10mM citrate buffer, pH 4.5. Control mice received buffer alone. Four weeks after treatment, glycemia was measured using One Touch Vita test strips (LifeScan Europe, Zug, Switzerland, limited to an upper value of 600 mg/dl). Animals with glycemia ≥300 mg/dl were assigned to the diabetic groups for the experiments. Mean arterial blood pressure was measured using a Page 5 of 32 Diabetes

noninvasive computerized tailcuff method (CODA, Kent Scientific, Torrington, CT) in non

anesthetized mice after acclimation (17).

Chemical assays. Blood was collected through cardiac puncture into 0.2% EDTA tubes.

Soluble intercellular cell adhesion molecule1 (sICAM1), vascular cell adhesion molecule1

(sVCAM1), and Pselectin (sPselectin) were quantified using ELISA kits, and HA using an

ELISAlike assay that allows detection of HA molecules ≥15 kDa (18), all obtained from

R&D Systems, Minneapolis, MN. Syndecan1 was measured using an ELISA kit from

Diaclone (Besançon, France). Albumin and creatinine concentrations were measured in urine

samples using Albuwell (Exocell, Philadelphia, PA) and Creatinine (Enzo Life Sciences,

Lausen, Switzerland) kits, respectively.

Preparation of aortic samples. Aortas were isolated, cleaned of fat on ice, frozen in liquid

nitrogen, and stored at 80°C. They were then lyophilized during 16h and treated with Pronase

(at 3mg/ml in 100 mM ammonia/formic acid buffer, pH 78) for 24h at 55°C. After

relyophilization, the samples were resuspended in water to allow HA measurement. In some

experiments, aortas were first flushed on ice with a strong injection of cold PBS (2ml) by

holding their extremity on the needle using pliers.

Hyaluronidase activity. Plasma HYAL1 activity was measured using two different

approaches: a) zymography in renatured and native conditions, as described previously (4); b)

gel electrophoresis (19) followed by quantification of oligosaccharide bands using imageJ

(public domain, NIH).

Glycocalyx staining in myocardial arterioles. The method followed a procedure previously

described (20). Briefly, the aorta of anesthetized mice was retrogradely cannulated and the

vena cava transsected. The following solutions were infused at a flow rate of 0.4 ml/min and a

pressure of 33±5 mmHg: a cardioplegic solution during 3 min; a phosphate buffered 4%

paraformaldehyde – 1% glutaraldehyde (pH 7.4) fixative solution for 2 min, and finally, the Diabetes Page 6 of 32

same solution containing 0.05% Alcian Blue 8GX (SigmaAldrich) during 30 min. The left ventricular wall was cut in 2mm segments, fixed for 1h in the fixative solution, and postfixed in 1% osmium tetroxide and 1% lanthanum nitrate for 1h, then processed for transmission electronic microscopy using a standard procedure. Sections were visualized with a FEI Tecnai microscope and photomicrographs analyzed using imageJ. The glycocalyx thickness of cardiac arterioles was calculated by dividing the surface area by the underlying endothelium length.

Dextran excretion. Anesthetized mice were intravenously injected with a mixture of 10 mg/ml Texas Red40kDa neutral dextran and 2.5 mg/ml FITC70kDa anionic dextran

(Molecular Probes, Eugene, OR). Urine was collected during 30 min and fluorescence was measured to determine glomerular permselectivity based on the 70kDa to 40 kDa dextran ratio. The urinary albumin/creatinine (A/C) concentration ratio was also measured.

Endothelium-dependent vasodilation. Secondorder mesenteric arteries were isolated from animals under terminal anesthesia and placed in icecold Tyrode solution. Arteries were cleared of fat and connective tissue, then cut into <2mm rings and mounted in a wire myograph (model 610MDMT, Danish Myo Technology A/S, Aarhus, Denmark) as previously described (21). After 45min stabilization in Tyrode solution containing 105M indomethacin, tension normalization, and 60min equilibration, vessels were contracted using

100mM KCl. Then, cumulative concentrations of ACh (108M to 3.105M) were added to induce endotheliumdependent relaxation. After washout and stabilization, vessels were again contracted using 3.106M phenylephrine in the absence or the presence of 104M NnitroL arginine methyl ester (LNAME). Cumulative amounts of ACh were again added and the percentage of residual contraction was calculated. To test small conductance potassium channel3 (SK3) activity, an SK3 opener, cyclohexyl[2(3,5dimethylpyrazol1yl)6 methylpyrimidin4yl]amine (CYPPA, SigmaAldrich), was used in the presence of Page 7 of 32 Diabetes

indomethacin and LNAME. Vessels were precontracted with 107M U46,619 (Sigma

Aldrich), a thromboxane A2 agonist, instead of phenylephrine to obtain a stable contractile

state. Cumulative amounts of CYPPA (3.107M – 6.105M) were then added.

mRNA expression. Total RNA was isolated from microdissected mesenteric arteries using

RNeasy Micro Kit (Qiagen, Hilden, Germany) and treated with DNAse. Reverse transcription

was performed using random hexamers and Superscript II MMLVreverse transcriptase

(InVitrogen, Carlsbad, CA). The levels of expression of several were determined using

realtime PCR (Applied Biosystems 7300 Real Time PCR System, Warrington, Cheshire,

UK) with SYBRgreen detection. mRNA levels were calculated using the 2ddCT method.

Immunohistology. Mesenteric arteries were fixed in alcoholic Bouin’s solution for 72h and

embedded in paraffin. Detection of SK3 and von Willebrand factor (vWF) in 6m sections

were performed with an antiSK3 rabbit polyclonal antibody (Sc28621, Santa Cruz

Biotechnology, Santa Cruz, CA) and an antivWF rabbit polyclonal antibody (A0082, DAKO

A/S, Denmark) followed by biotinylated secondary antibodies and streptavidinperoxidase.

Quantification of the SK3 immunostaining was carried out using imageJ.

Statistical methods. Twoway Analysis of Variance (ANOVA) followed by Bonferroni post

hoc tests was performed to compare the four groups of mice in each experiment.

RESULTS

Effects of diabetes on health conditions of WT and Hyal1-/- mice

Baseline fasting glycemia was similar in WT and KO mice and significantly increased,

comparably in both genotypes, 28 days post STZ treatment (Fig. 1A). Baseline body weights

of 8week old KO mice were not significantly lower than those of WT mice. Furthermore, the

growtharresting impact of diabetes was similar in both genotypes (Fig. 1B). There was no

effect of either genotype or diabetes on mean blood pressure (Fig. 1C). Diabetes Page 8 of 32

Effect of diabetes on HA and hyaluronidase activity in WT and Hyal1-/- mice

A twofold increase in circulating HA levels was observed in diabetic vs nondiabetic WT mice (Fig. 2A). In KO mice, baseline HA concentrations were 4 times higher than in WT mice, and did not further increase after induction of diabetes.

Zymography of serum samples revealed a single band of hyaluronidase activity around 80 kDa in both denaturing and native conditions, corresponding to the activity of the HYAL1 precursor, which did not increase following diabetes induction (Fig. 2B&C). In aortic wall homogenates, zymography in native conditions showed two bands, corresponding to the precursor and cleaved forms of HYAL1 (4), again with no increase in hyaluronidase activity during diabetes (Fig. 2D). Gel electrophoresis of HA solutions incubated with sera of non diabetic and diabetic WT mice allowed a more accurate measurement of serum hyaluronidase activity, and revealed a slight but significant increase in the amount of smaller oligosaccharides produced using hyperglycemic sera (Fig. 2EH), suggesting an increased

HYAL1 activity (3) in diabetic WT mice. The same analysis performed on HYAL1 KO mouse serum (Fig. 2E) confirmed the assay specificity.

HA on the luminal side of vessels

To determine whether increased plasma HA concentrations are accompanied by changes in glycocalyx HA content, we measured the amount of HA in flushed segments of the aortic wall, which almost completely removes the endothelium while leaving the underlying aortic wall intact, and compared it with that of unflushed segments. There was not difference in the amount of HA in the unflushed aortic segments between any of the groups (Fig. 3A).

However, while the flushable/glycocalyx HA accounted for approximately onequarter of the total aortic HA in healthy WT and KO mice, it was nearly absent in diabetic WT mice but was preserved in diabetic KO mice.

Glycocalyx thickness and integrity Page 9 of 32 Diabetes

Since HA is not the only component of glycocalyx, we examined its structure in small vessels

(ventricular arterioles) using a wellestablished electron microscopic method as described in

the Materials and Methods section (Fig. 3BC). As summarized in Fig. 3D, glycocalyx

thickness was >3fold higher in KO than WT mice, both in diabetic and nondiabetic mice.

Diabetes did not induce any significant change in glycocalyx thickness in either genotype.

Plasma syndecan1 concentration (Fig. 3E), measured to investigate glycocalyx shedding, was

not impacted by either genotype or diabetes.

Glycocalyx barrier function

The relative permeability of the glomerular endothelial glycocalyx was then evaluated by

measuring the concentration ratio of 70kDa to 40kDa fluorescent dextran recovered in the

urine following intravenous injection. As shown in Fig. 4, the 70/40kDa excretion ratio

increased during diabetes in WT but not in KO mice, suggesting that glomerular

permselectivity to high molecular weight dextran is altered in diabetic WT mice but not in

diabetic KO mice. Similarly, the urinary albumin/creatinine ratio increased in diabetic WT but

not KO mice. This suggests that the absence of HYAL1 protects the glomerular endothelial

glycocalyx against diabetesinduced functional damage.

Markers of endothelium dysfunction

In order to detect early signs of endothelial dysfunction, we measured the levels of circulating

adhesion molecules. The level of sICAM1 was significantly upregulated during diabetes in

both genotypes (Fig. 5B). The level of sPselectin was also upregulated during diabetes but

only in WT mice (Fig. 5A). The concentration of sVCAM1 was not affected by diabetes in

any genotype (Fig. 5C). The baseline plasma concentrations of sICAM1 and sVCAM1were

lower in KO than WT mice, suggesting a healthier endothelial status in the absence of

HYAL1.

Endothelial-dependent vasorelaxation Diabetes Page 10 of 32

Endothelial function was investigated ex vivo in smalldiameter, secondbranch mesenteric arteries, allowing assessment of both nitric oxide (NO) and endotheliumdependent hyperpolarization (EDH)mediated relaxation. NOdependent AChinduced vasodilation was similar in all groups of mice (data not shown). EDHdependent AChinduced vasodilation, on the other hand, was severely altered in diabetic WT mice, with only a remaining 34% EDH dependent vasodilation, while partially preserved in diabetic KO mice, with a remaining 56%

EDHdependent vasodilation (Fig. 6). The difference between diabetic WT and KO mice was highly significant (p<0.0001). There was no difference between healthy WT and KO mice.

Exploration of the EDH pathway components

The mRNA expression level of several components of the EDH pathway (22) in mesenteric arteries was screened using realtime RTPCR. No difference amongst experimental groups was detected for connexins 37, 40, 43, and 45; small conductance potassium (SK) channels

SK1 and SK2; intermediate conductance potassium (IK) channel IK1; transient receptor potential (TRP) channels TRPV4 and TRPC1; and caveolin-1 (data not shown). Conversely,

SK3 mRNA expression was significantly upregulated in KO mice independently of diabetes

(Fig. 7A). Immunohistochemistry allowed the detection of SK3 along the endothelium in a pattern similar to that of vWF (Fig. 7B). Quantification of the staining confirmed increased expression of SK3 in KO than in WT vessels independently of diabetes (Fig. 7C).

To determine whether the activity of SK3 was increased in KO vs WT mice, mesenteric artery segments were mounted on wire myographs, contracted with U46,619, and exposed to increasing concentrations of the CYPPA SK3opener (23). SK3dependent relaxation was more efficient in KO than WT mice, whether the animals were healthy or diabetic (Fig. 7D).

SK3 overexpression may thus explain the preservation of the EDHdependent vasodilation observed in diabetic KO mice.

Page 11 of 32 Diabetes

DISCUSSION

Increased serum hyaluronidase activity in diabetes has been reported in man (15) and rats (24)

but the implication of these observations has been poorly studied. The present study confirms

a slight increase in serum hyaluronidase activity, which is likely due to HYAL1, in early

diabetic WT mice (Fig. 1H). Furthermore, experiments using HYAL1 deficient mice at a

relatively early stage of Type 1 diabetes (4 weeks after STZ injections) show, for the first

time, that HYAL1 may have a pathogenic role in diabetesinduced endothelial dysfunction.

Lack of HYAL1 prevents endothelial dysfunction

Endothelial dysfunction in murine diabetes can be evaluated by measuring plasma levels of

sPselectin, sICAM1, and sVCAM1, as well as AChdependent vasorelaxation. We showed

that 4 weeks of STZtreatment in WT mice are sufficient to increase sPselectin and sICAM1

but not sVCAM1, suggesting that the latter may associate with a later stage in diabetes

development, as suggested previously (25).

Alteration in AChdependent vasorelaxation is another reliable marker of endothelial

dysfunction, at least in rat models of Type 1 and Type 2 diabetes (26). In mice, loss of

endotheliumdependent relaxation appears only after 10 weeks of hyperglycemia in STZ

induced diabetes (27). Concordant with these data and previous observations in diabetic rats

(28), we found no defect in the overall AChdependent vasorelaxation after 4 weeks of STZ

injections. Still, we demonstrated a clear reduction in EDHdependent relaxation.

Furthermore, HYAL1 deficiency may prevent diabetesinduced increase in sPselectin, but

not in sICAM1, and limit the impact of the disease on EDHdependent vasorelaxation. The

data on sPselectin and EDHdependent vasorelaxation suggest functionally important

cardiovascular protective effects of HYAL1 deficiency during diabetes. Diabetes Page 12 of 32

Interestingly, the absence of HYAL1 also modified the baseline levels of circulating markers of endothelial damage, two of which (sICAM1 and sVCAM1) were lower in KO than WT mice. This suggests that HYAL1deficient endothelia may be less attractant to leukocytes.

This hypothesis could be tested by measuring endothelial chemoattraction and diapedesis. Of note, HYAL1 deficiency did not prevent renal neutrophil and macrophage infiltration in a murine model of severe ischemia reperfusion injury (29), but this is a complex acute lesion model in which multiple factors could modulate the phenotype.

Through a deeper analysis of several key endothelial proteins, we observed increased expression of SK3 in HYAL1deficient endothelium. This endothelial potassium channel is reportedly a fundamental determinant of vascular tone and blood pressure (30) and a mainspring of the EDH pathway (31). In the absence of diabetes, however, the EDHmediated vasodilation measured in mesenteric arteries was not enhanced by lack of HYAL1. This suggests baseline endothelial SK3 levels are not ratelimiting for EDHinduced vasodilation in normal physiological conditions but could become so when the EDH pathway needs to be activated, e.g. during early diabetes. SK3 downregulation was previously observed in STZ induced diabetic ApoEdeficient mice (22) or in the cavernous tissues of diabetic rats (32) but not in C57Bl/6 diabetic mice (27). In line with the latter results, SK3 was not downregulated by diabetes in our study. The longterm benefits of HYAL1 deficiencyassociated increase in

SK3 expression in diabetes remain to be demonstrated. The absence of impact of diabetes and genotype on other EDH components at the RNA level does not preclude an effect at the protein level. The small size of vascular samples has prevented us from thoroughly examining this hypothesis in the current study.

Lack of HYAL1 maintains glycocalyx structure and prevents HA shedding Page 13 of 32 Diabetes

The HA content of the glycocalyx is reportedly crucial for endothelial barrier function (9). A

detrimental effect of diabetes or acute hyperglycemia on endothelial glycocalyx has been

demonstrated in several studies in man (13,15) and mice (33). We postulated that the

mechanism of endothelial protection in HYAL1 deficient mice is linked to a stronger

glycocalyx. In our study, 4week STZinduced diabetes did not reduce the size of the

endothelial glycocalyx in WT mice, as measured with a sensitive electron microscopic

technique. Nevertheless, diabetesexposed glycocalyx became HAdepleted and thus

potentially more vulnerable. HA seems to be incorporated within the glycocalyx in a shear

stressdependent way (34). It was therefore highly relevant to observe that HYAL1 deficient

endothelial surfaces, contrary to WT endothelia, maintained their HA content during diabetes

(demonstrated in Fig. 3).

Furthermore, the absence of HYAL1 dramatically increased the thickness of the glycocalyx

(in both healthy and diabetic glycocalyx). However, aortic flushes failed to demonstrate a

higher HA content as the reason for the increased size of glycocalyx. This may be due to

insufficient sensitivity of the methods and/or additional factors. One possibility is a better

anchoring of HA into the glycocalyx and a longer stretching of the HA chains in the absence

of HYAL1. This could lead to an apparent increased thickness of the glycocalyx without HA

accumulation.

Glycocalyx shedding under severe inflammatory conditions, such as postischemic

reperfusion, can be prevented using various treatments, e.g. hydrocortisone, antithrombin, or

heparin (35, 36). Sulodexide, a mix of heparan sulfate and dermatan sulfate, increases

glycocalyx thickness in Type 2 diabetes (14). However, although sulodexide had global

beneficial effects on renal manifestations of experimental diabetes in C57Bl/6 mice (37), it

failed to demonstrate renoprotection in overt Type 2 diabetic nephropathy in man (38). Diabetes Page 14 of 32

To our knowledge, such an increase in glycocalyx size and resistance as observed in our study using HYAL1 KO mice has never been described as a result of therapeutic intervention or genetic manipulation (39). A thicker glycocalyx may thus correspond to reduced access of circulating inflammatory cells to the endothelium (8) as well as more efficient shear stress induced signals. The glycocalyx has indeed been suggested to mediate shear stress or flow induced NO production and vascular remodelling (40,41), although, perhaps surprisingly, its role in EDHmediated vasorelaxation has not been explored to date. A thicker glycocalyx could also explain the higher baseline expression of SK3 observed in our study.

Finally, although a thicker glycocalyx correlated with beneficial effects on diabetesinduced endothelial dysfunction in the current study, this was only observed in the absence of

HYAL1. Thus, caution is warranted before the cardiovascular consequences of a thicker baseline glycocalyx can be described as indisputably favorable.

Lack of HYAL1 prevents glomerular barrier dysfunction

Diabetes is a wellknown aggressor of the glomerular glycocalyx and endothelial barrier resulting in loss of permselectivity and ultrafiltration of albumin (42). Glycocalyx damage coincides with microalbuminuria in Type 1 diabetes (15). Obese diabetic db/db mice have an altered glycocalyx with higher access of a 70kDa dextran tracer to the vessel wall and higher clearance of this tracer (43). In the current study, 4week STZinduced diabetes significantly increased the urinary A/C ratio and 70/40kDa dextran ratio, confirming altered glomerular permselectivity. These effects were completely prevented in diabetic HYAL1 KO mice, bringing further arguments for a functionally preserved glycocalyx and endothelial barrier when diabetes develops in the absence of HYAL1. Baseline A/C and 70/40kDa dextran ratios were similar in WT and HYAL1 deficient mice.

Page 15 of 32 Diabetes

Mechanisms of endothelial protection in the absence of HYAL1

Table 1 summarizes the beneficial effects observed in HYAL1 KO mice exposed to STZ

induced diabetes as compared with WT mice. The main protective effects include lower

release of Pselectin into the circulation, lower shedding of HA from the endothelial surface,

preserved integrity of the glomerular endothelial glycocalyx, and conservation of EDH

mediated vasorelaxation.

As for the mechanisms involved in HYAL1mediated endothelial protection, our main

hypothesis is that the absence of HYAL1 prevents glycocalyx HA shedding during diabetes

and, from there, affords protection against diabetesinduced vascular damage and maintain a

higher level of SK3 expression. In turn, this would preserve EDH relaxation during diabetes,

when SK3 expression becomes rate limiting.

Another hypothesis is that elevated plasma HA levels in HYAL1 KO mice facilitate its

reincorporation into the glycocalyx during diabetic injury. Previous studies have shown that

the glycocalyx is quickly (30 min) restored after hyaluronidase treatment if HA and

chondroitin sulfate are infused in sufficiently high amounts, i.e. >100 times the baseline

plasma HA concentration (9). However, the difference in plasma HA concentrations between

diabetic WT mice and diabetic HYAL1 KO mice in our study was only moderate (+58% in

the latter) and unlikely to explain the large difference in glycocalyx HA content during

exposure of WT vs KO mice to hyperglycemia. Additionally, we have shown similar

molecular size profiles of circulating HA in HYAL1 KO and WT mice (44). Finally, the

hypothesis that higher baseline expression of SK3 would explain a thicker glycocalyx seems

unlikely based on the known function of SK3.

Conclusion and perspectives Diabetes Page 16 of 32

HYAL1 deficiency had several protective effects against early diabetesinduced endothelial and glycocalyx damage. Most, if not all, of these effects may result from a thicker and sturdier glycocalyx (with, e.g., a higher level of endothelial SK3 channels). This stronger glycocalyx may, in the long run, protect against macro and microvascular complications of diabetes, both of which have been linked to glycocalyx impairment. In ApoEdeficient atheromatosis prone mice, for instance, HA synthesis inhibition using 4methylumbelliferone led to a loss of glycocalyx and enhanced atherosclerosis (45). HYAL1 inhibition may have an opposite, protective effect. In addition, the main diabetesassociated microvascular complications, i.e. retinopathy, neuropathy, and glomerular capillary injury, have been linked to a loss of glycocalyx. For instance, early diabetes impact glomerular permeability and glycocalyx alterations are reported in diabetic mice before the development of albuminuria (46). In early diabetic retinopathy in rats and mice, both glycocalyx and EDHmediated vasodilation of retinal vessels are altered (47, 48). Diabetic neuropathy in mice is accompanied by glycocalyx alteration in brain microvessels (49). Finally, diabetes significantly reduces EDHmediated vasorelaxation in human penile resistance arteries and downregulates SK3 and IK1 channels in rat corpus cavernosum tissues (32).

HYAL1 inhibition could thus be a first step to prevent loss of glycocalyx during the development of diabetic nephropathy, retinal microangiopathy, blood brain barrier damage, and erectile dysfunction.

Interestingly, even though HA has been shown to accumulate in the aortic tunica media of

Type 2 diabetic patients (50) where it may have potential deleterious effects, our study fails to demonstrate any HA accumulation in the aorta of congenital HYAL1 deficient mice, neither before nor after induction of diabetes. Page 17 of 32 Diabetes

Overall, this report suggests that HYAL1 inhibitors could potentially be explored as a new

therapeutic approach to prevent endothelial dysfunction in diabetes. As a caveat, the

beneficial effects of HYAL1 deficiency were observed in early diabetes and should be

confirmed over the course of the disease.

ACKNOWLEDGMENTS

Author contributions: SD designed and performed the experiments, contributed to discussion

and wrote the manuscript; GR and FJ designed and performed experiments; NC contributed to

the discussion; CD and BF designed and supervised the experiments, contributed to the

discussion and reviewed/edited the manuscript. Our sincere thanks to Laurence Jadin for

reviewing the manuscript.

SD is the guarantor of this work and, as such, had full access to all the data in the study and

takes responsibility for the integrity of the data and the accuracy of the data analysis. No

potential conflicts of interest relevant to this article were reported.

Parts of the study were presented in poster form at the Experimental Biology scientific

meeting, San Diego, CA, USA, 2630 April 2014 and at the 10th International Conference on

Hyaluronan, Florence, Italy, 711 June 2015.

This research used resources of the Electron Microscopy Service located at the University of

Namur. This Service is member of the “Plateforme Technologique Morphologie – Imagerie”.

CD is a senior research associate of the Fonds National de la recherche Scientifique (FNRS).

FJ is an MD postdoctoral fellow of the FNRS.

REFERENCES

1. Hascall V, Esko JD. Hyaluronan. In: Essentials of Glycobiology. 2nd ed. Edited by Varki A, Cummings RD, Esko JD, et al.. Cold Spring Harbor (NY), 2009. 2. Csoka AB, Frost GI, Stern R. The six hyaluronidaselike genes in the human and mouse genomes. Matrix Biol 2001;20:499508. 3. FiszerSzafarz B, Litynska A, Zou L. Human hyaluronidases: electrophoretic multiple forms in somatic tissues and body fluids. Evidence for conserved hyaluronidase Diabetes Page 18 of 32

potential N sites in different mammalian species. J Biochem Biophys Methods 2000;45:103116. 4. Puissant E, Gilis F, Dogne S, Flamion B, Jadot M, Boonen M. Subcellular trafficking and activity of Hyal1 and its processed forms in murine macrophages. Traffic 2014;15:500515. 5. Natowicz MR, Short MP, Wang Y, Dickersin GR, Gebhardt MC, Rosenthal DI, et al. Clinical and biochemical manifestations of hyaluronidase deficiency. N Engl J Med 1996;335:10291033. 6. Martin DC, Atmuri V, Hemming RJ, Farley J, Mort JS, Byers S, et al. A mouse model of human IX exhibits osteoarthritis. Hum Mol Genet 2008;17:19041915. 7. Tarbell JM, Pahakis MY. Mechanotransduction and the glycocalyx. J Intern Med 2006;259:339350. 8. Mulivor AW, Lipowsky HH. Role of glycocalyx in leukocyteendothelial cell adhesion. Am J Physiol Heart Circ Physiol 2002;283:H12821291. 9. Henry CB, Duling BR. Permeation of the luminal capillary glycocalyx is determined by hyaluronan. Am J Physiol 1999;277:H508514. 10. Kelly R, RuaneO'Hora T, Noble MI, DrakeHolland AJ, Snow HM. Differential inhibition by hyperglycaemia of shear stress but not acetylcholinemediated dilatation in the iliac artery of the anaesthetized pig. J Physiol 2006;573:133145. 11. Mochizuki S, Vink H, Hiramatsu O, Kajita T, Shigeto F, Spaan JA, et al. Role of hyaluronic acid glycosaminoglycans in shearinduced endotheliumderived nitric oxide release. Am J Physiol Heart Circ Physiol 2003;285:H722726. 12. Shakya S, Wang Y, Mack JA, Maytin EV. Hyperglycemiainduced changes in hyaluronan contribute to impaired skin wound healing in diabetes: Review and perspective. Int J Cell Biol 2015;2015:701738. 13. Nieuwdorp M, van Haeften TW, Gouverneur MC, Mooij HL, van Lieshout MH, Levi M, et al. Loss of endothelial glycocalyx during acute hyperglycemia coincides with endothelial dysfunction and coagulation activation in vivo. Diabetes 2006;55:480486. 14. Broekhuizen LN, Lemkes BA, Mooij HL, Meuwese MC, Verberne H, Holleman F, et al. Effect of sulodexide on endothelial glycocalyx and vascular permeability in patients with type 2 diabetes mellitus. Diabetologia 2010;53:26462655. 15. Nieuwdorp M, Mooij HL, Kroon J, Atasever B, Spaan JA, Ince C, et al. Endothelial glycocalyx damage coincides with microalbuminuria in type 1 diabetes. Diabetes 2006;55:11271132. 16. Ceriello A, Giugliano D, Dello Russo P, Passariello N, Saccomanno F, Sgambato S. Glycosaminoglycans in human diabetes. Diabete Metab 1983;9:3234. 17. Krege JH, Hodgin JB, Hagaman JR, Smithies O. A noninvasive computerized tailcuff system for measuring blood pressure in mice. Hypertension 1995;25:11111115. 18. Haserodt S, Aytekin M, Dweik RA. A comparison of the sensitivity, specificity, and molecular weight accuracy of three different commercially available Hyaluronan ELISAlike assays. Glycobiology 2011;21:175183. 19. IkegamiKawai M, Takahashi T. Microanalysis of hyaluronan oligosaccharides by polyacrylamide gel electrophoresis and its application to assay of hyaluronidase activity. Anal Biochem 2002;311:157165. 20. van den Berg BM, Vink H, Spaan JA. The endothelial glycocalyx protects against myocardial edema. Circ Res 2003;92:592594. 21. Ghisdal P, Godfraind T, Morel N. Effect of nitroLarginine on electrical and mechanical responses to acetylcholine in the superior mesenteric artery from stroke prone hypertensive rat. Br J Pharmacol 1999;128:15131523 Page 19 of 32 Diabetes

22. Ding H, Hashem M, Wiehler WB, Lau W, Martin J, Reid J, et al. Endothelial dysfunction in the streptozotocininduced diabetic apoEdeficient mouse. Br J Pharmacol 2005;146:11101118. 23. Hougaard C, Eriksen BL, Jorgensen S, Johansen TH, Dyhring T, Madsen LS, et al. Selective positive modulation of the SK3 and SK2 subtypes of small conductance Ca2+activated K+ channels. Br J Pharmacol 2007;151:655665. 24. IkegamiKawai M, Suzuki A, Karita I, Takahashi T. Increased hyaluronidase activity in the kidney of streptozotocininduced diabetic rats. J Biochem 2003;134:875880. 25. Gustavsson C, Agardh CD, Zetterqvist AV, Nilsson J, Agardh E, Gomez MF. Vascular cellular adhesion molecule1 (VCAM1) expression in mice retinal vessels is affected by both hyperglycemia and hyperlipidemia. PLoS One 2010;5:e12699. 26. De Vriese AS, Verbeuren TJ, Van de Voorde J, Lameire NH, Vanhoutte PM. Endothelial dysfunction in diabetes. Br J Pharmacol 2000;130:963974. 27. Matsumoto T, Miyamori K, Kobayashi T, Kamata K. Specific impairment of endotheliumderived hyperpolarizing factortype relaxation in mesenteric arteries from streptozotocininduced diabetic mice. Vascul Pharmacol 2006;44:450460. 28. De Vriese AS, Van de Voorde J, Blom HJ, Vanhoutte PM, Verbeke M, Lameire NH. The impaired renal vasodilator response attributed to endotheliumderived hyperpolarizing factor in streptozotocininduced diabetic rats is restored by 5 methyltetrahydrofolate. Diabetologia 2000;43:11161125. 29. Colombaro V, Jadot I, Decleves AE, Voisin V, Giordano L, Habsch I, et al. Lack of hyaluronidases exacerbates renal postischemic injury, inflammation, and fibrosis. Kidney Int 2015. 30. Taylor MS, Bonev AD, Gross TP, Eckman DM, Brayden JE, Bond CT, et al. Altered expression of smallconductance Ca2+activated K+ (SK3) channels modulates arterial tone and blood pressure. Circ Res 2003;93:124131. 31. Brahler S, Kaistha A, Schmidt VJ, Wolfle SE, Busch C, Kaistha BP, et al. Genetic deficit of SK3 and IK1 channels disrupts the endotheliumderived hyperpolarizing factor vasodilator pathway and causes hypertension. Circulation 2009;119:23232332. 32. Zhu JH, Jia RP, Xu LW, Wu JP, Wang ZZ, Wang SK, et al. Reduced expression of SK3 and IK1 channel proteins in the cavernous tissue of diabetic rats. Asian J Androl 2010;12:599604. 33. Zuurbier CJ, Demirci C, Koeman A, Vink H, Ince C. Shortterm hyperglycemia increases endothelial glycocalyx permeability and acutely decreases lineal density of capillaries with flowing red blood cells. J Appl Physiol (1985) 2005;99:14711476. 34. Gouverneur M, Spaan JA, Pannekoek H, Fontijn RD, Vink H. Fluid shear stress stimulates incorporation of hyaluronan into endothelial cell glycocalyx. Am J Physiol Heart Circ Physiol 2006;290:H458452. 35. RubioGayosso I, Platts SH, Duling BR. Reactive oxygen species mediate modification of glycocalyx during ischemiareperfusion injury. Am J Physiol Heart Circ Physiol 2006;290:H22472256. 36. Chappell D, HofmannKiefer K, Jacob M, Rehm M, Briegel J, Welsch U, et al. TNF alpha induced shedding of the endothelial glycocalyx is prevented by hydrocortisone and antithrombin. Basic Res Cardiol 2009;104:7889. 37. Yung S, Chau MK, Zhang Q, Zhang CZ, Chan TM. Sulodexide decreases albuminuria and regulates matrix protein accumulation in C57BL/6 mice with streptozotocin induced type I diabetic nephropathy. PLoS One 2013;8:e54501. 38. Packham DK, Wolfe R, Reutens AT, Berl T, Heerspink HL, Rohde R, et al. Sulodexide fails to demonstrate renoprotection in overt type 2 diabetic nephropathy. J Am Soc Nephrol 2012;23:123130. Diabetes Page 20 of 32

39. Becker BF, Jacob M, Leipert S, Salmon AH, Chappell D. Degradation of the endothelial glycocalyx in clinical settings: searching for the sheddases. Br J Clin Pharmacol 2015;80:389402. 40. Pohl U, Herlan K, Huang A, Bassenge E. EDRFmediated shearinduced dilation opposes myogenic vasoconstriction in small rabbit arteries. Am J Physiol 1991;261:H20162023. 41. Florian JA, Kosky JR, Ainslie K, Pang Z, Dull RO, Tarbell JM. Heparan sulfate proteoglycan is a mechanosensor on endothelial cells. Circ Res 2003;93:e136142. 42. Salmon AH, Satchell SC. Endothelial glycocalyx dysfunction in disease: albuminuria and increased microvascular permeability. J Pathol 2012;226:562574. 43. Eskens BJ, Zuurbier CJ, van Haare J, Vink H, van Teeffelen JW. Effects of two weeks of metformin treatment on wholebody glycocalyx barrier properties in db/db mice. Cardiovasc Diabetol 2013;12:175. 44. Bourguignon V, Flamion B. Respective roles of hyaluronidases 1 and 2 in endogenous hyaluronan turnover. FASEB J 2016 (in press). 45. Nagy N, Freudenberger T, MelchiorBecker A, Rock K, Ter Braak M, Jastrow H, et al. Inhibition of hyaluronan synthesis accelerates murine atherosclerosis: novel insights into the role of hyaluronan synthesis. Circulation 2010;122:23132322. 46. Nagasu H, Satoh M, Kiyokage E, Kidokoro K, Toida K, Channon KM, et al. Activation of endothelial NAD(P)H oxidase accelerates early glomerular injury in diabetic mice. Lab Invest 2016;96:2536. 47. Kumase F, Morizane Y, Mohri S, Takasu I, Ohtsuka A, Ohtsuki H. Glycocalyx degradation in retinal and choroidal capillary endothelium in rats with diabetes and hypertension. Acta Med Okayama 2010;64:277283. 48. Nakazawa T, Kaneko Y, Mori A, Saito M, Sakamoto K, Nakahara T, et al. Attenuation of nitric oxide and prostaglandinindependent vasodilation of retinal arterioles induced by acetylcholine in streptozotocintreated rats. Vascul Pharmacol 2007;46:153159. 49. Liao YJ, Ueno M, Nakagawa T, Huang C, Kanenishi K, Onodera M, et al. Oxidative damage in cerebral vessels of diabetic db/db mice. Diabetes Metab Res Rev 2005;21:554559. 50. Heickendorff L, Ledet T, Rasmussen LM. Glycosaminoglycans in the human aorta in diabetes mellitus: a study of tunica media from areas with and without atherosclerotic plaque. Diabetologia 1994;37:286292.

TABLES

Table 1. Beneficial effects observed in HYAL1 KO mice exposed to streptozotocininduced diabetes as compared with WT mice.

Diabetic Diabetic Likely mechanism of protection in absence of

WT KO HYAL1 Effect of diabetes on: a) endothelium

Plasma Pselectin (↑) Endothelium is less activated Page 21 of 32 Diabetes

Plasma ICAM1 (↑) No protection *

Plasma VCAM1 () No effect of diabetes *

EDH vasorelaxation (↓) Higher baseline SK3 or thicker glycocalyx

SK3 () No effect of diabetes *

b) glycocalyx

Glycocalyx HA (↓) Glycocalyx HA is more adherent

Glycocalyx size () No effect of diabetes *

Syndecan1 shedding () No effect of diabetes *

c) Fonctionnal glomerular barrier

Urinary A/C ratio (↑) Preserved integrity of glomerular glycocalyx

Urinary 70/40kDa dextran ratio (↑) Preserved integrity of glomerular glycocalyx

WT, wildtype; KO, knockout; ICAM1, intercellular cell adhesion molecule1; VCAM1, vascular cell adhesion

molecule1; EDH, endotheliumderived hyperpolarization; SK3, small conductance potassium channel3; A/C,

albumin/creatinine.

* Soluble ICAM1, VCAM1 were lower, while SK3 expression was higher and glycocalyx was thicker, in HYAL1 KO

mice than in WT mice.

(↑) indicates that a significant increase in this parameter was measured 4 weeks after streptozotocininduced diabetes;

(↓), a significant decrease; and (), the absence of change. Diabetesinduced injuries in WT mice are indicated by a

black rectangle, and the absences of change are indicated by a hatched rectangle. A black rectangle staying black in

the « KO » column means no benefit of HYAL1 deficiency; a black rectangle becoming white in the « KO » column

means a significant protection afforded by HYAL1 deficiency.

FIGURE LEGENDS

Figure 1. Effects of diabetes on glycemia, body weight and arterial pressure. A. Fasting

glycemia (box and whisker plots) during diabetes induction (streptozotocin [STZ] injections)

vs control conditions (citrate buffer injections). N= 7 to 36 in each group of wildtype (WT)

and HYAL1 knockout (KO) mice without or with diabetes. Diabetic mice include only those

with fasting glycemia ≥300 mg/dL. B. Body weight tracking before, during, and 28 days after Diabetes Page 22 of 32

STZ or citrate buffer injections (n= 16 to 36 in each group; means ± SEM). C. Mean arterial pressure (means ± SEM) 28 days after STZ or citrate buffer treatment (n= 3 to 5 in each group). In all data sets, statistical significance was calculated using 2way ANOVA and

Bonferroni posttests (# p<0.05, ### p<0.001, #### p<0.0001 within genotype).

Figure 2. Plasma hyaluronan (HA) concentration and hyaluronidase activity. A. Plasma

HA concentrations of WT and KO mice without or with STZinduced diabetes (n= 4 or 5 in each group). BD. Serum (BC) and aortic wall (D) hyaluronidase activity measured using zymography under native (B & D) or denaturing (C) conditions in each experimental group.

Contrary to denaturing conditions, the native procedure does not allow to determine the molecular weight of the HAdegrading . EH. Measurement of HA oligosaccharide production by serum hyaluronidase activity using polyacrylamide gel electrophoresis. E.

Rooster comb HA (5 µg) was incubated for 5 h with 1 µl of serum obtained from 2 animals of each group (lanes 18). The serum of KO mice (lanes 58) and undigested HA (lane 10) served as negative controls. HA incubated with 5U bovine testes hyaluronidase during 2 h

(lane 9) served as positive control. F. A gradient of bovine testes hyaluronidase at concentrations ranging of from 0 (lane 1) to 5 U (lane 5), incubated for 2 h with the same amount of HA, was also performed for comparison. GH. Quantification of serum hyaluronidase activity: Polyacrylamide gel electrophoresis of HA samples incubated with serum of WT mice without (lanes 14) and with (lanes 58) diabetes and used for quantification of hyaluronidase activity. The negative and positive controls used in G are similar to those used in E. (H) Quantification of hyaluronidase activity: signal intensity measurements for successive and sufficiently distinct oligosaccharide bands (i.e. below the dashed line in G). Statistical analysis uses 2way ANOVA (# and §, p<0.05; **, p<0.01; §§§§ and ****, p<0.0001). Page 23 of 32 Diabetes

Figure 3. Flushable HA in aortas of diabetic and non-diabetic WT and Hyal1 KO mice

and structural evaluation of the endothelial glycocalyx. A. HA content of aortic

homogenates (n= 5 to 7 in each group) standardized to dry weight before (white columns) and

after (black columns) saline flushing. Statistical analysis was performed using 2way

ANOVA and Bonferroni posttests to compare HA content before vs after flushing in each

group (####, p<0.0001). BC. Representative transmission electronic microscopy (TEM)

images of myocardial arterioles of WT(B) and KO (C) mice without or with STZinduced

diabetes following perfusion of the heart with Alcian Blue 8GX, in which the thickness of the

glycocalyx can be appreciated. Bars correspond to 1µm. D. Evaluation of glycocalyx

thickness obtained by dividing the glycocalyx surface area by the endothelium length

measured using ImageJ. N= 3 to 4 mice in each group, corresponding to approximately 50

TEM images of microvessels (≥8µm in diameter) in each group. Data are means ± SEM of

each experimental group. Statistical analysis was performed using 2way ANOVA (**,

p<0.01). E. Plasma syndecan1 levels in each experimental group (mean ± SEM, N = 710 in

each group). Statistical analysis was performed using 2way ANOVA (N.S).

Figure 4. Functional status of endothelial glycocalyx. A: Ratio of 70/40 kDa dextran in

urine 30 min after intrajugular injection of a 1:4 mixture of 40 and 70kDa dextrans in WT

(n=17) and KO (n=8) mice without or with diabetes. Statistical analysis was performed using

2way ANOVA (*, p<0.05) and Bonferroni posttests within genotypes (##, p<0.01). B:

Albumin/creatinine ratio measured in the same urine samples. Statistical analysis was

performed using 2way ANOVA (*, p<0.05) and Bonferroni posttests within genotypes (#,

p<0.05) or across genotypes (§, p<0.05).

Diabetes Page 24 of 32

Figure 5. Endothelial injury markers. Evaluation of three markers of endothelial damage in

WT and KO mice before and after STZinduced diabetes (n=5 to 8 in each group). A. Serum soluble Pselectin (sPselectin) levels. B. Plasma soluble ICAM1 (sICAM1) levels. C. Plasma soluble VCAM1 (sVCAM1) levels. Data are means ± SEM. Statistical analysis was performed using 2way ANOVA (*, p<0.05; **, p<0.01; ****, p<0.0001) and Bonferroni posttests inside genotype (#, p<0.05; ##, p<0.01; ###, p<0.001) or across genotype (§, p<0.05).

Figure 6. Endothelium-derived hyperpolarization (EDH)-mediated vasodilation.

Mesenteric arteriolar rings of WT and KO mice without or with STZ induced diabetes (n=24 to 26 rings in each group) were preconstricted with 3.106 M phenylephrine in the presence of

104 M LNAME and 105 M indomethacin. Vasodilatation was then induced using increasing concentrations (108 M to 3.105 M) of acetylcholine. Data are means ± SEM. Statistical analysis, 2way ANOVA (****, p<0.0001).

Figure 7. Expression and activity of SK3 channels in mesenteric arteries. A: SK3 relative mRNA expression in mesenteric arteries of WT and KO mice, without or with STZinduced diabetes (normalized to βactin, n=8 to 10 in each group). Data are means ± SEM. Statistical analysis was performed on ∆CT values using 2way ANOVA (***, p<0.001) and Bonferroni posttests across genotypes (§, p<0.05; §§, p<0.01). B: SK3 and von Willebrand factor (vWF) immunostaining in mesenteric arteries of WT and KO mice without and with STZinduced diabetes. C: Quantification of SK3 immunostaining using ImageJ (n=9 to 12 vessel sections in each group). Data are means ± SEM. Statistical analysis was performed using 2way ANOVA

(***, p<0.01) and Bonferroni posttests across genotypes (§, p<0.05; §§, p<0.01). D: Residual contraction after CYPPA activation of SK3 channels. Mesenteric rings of WT and KO mice Page 25 of 32 Diabetes

without or with STZinduced diabetes (n=10 to 12 in each group) were preconstricted with

7 4 10 M U46,619 (PGH2 analog, vasocontractant) in the presence of 10 M LNAME (NO

synthase inhibitor) and 105 M indomethacin (prostaglandin synthesis inhibitor).

Vasodilatation was then induced using increasing concentrations (3.107 M to 6.105 M) of

CYPPA, an SK3 opener. Data are means ± SEM. Statistical analysis was performed using 2

way ANOVA across genotype (§§§, p<0.001; §§§§, p<0.0001). Diabetes Page 26 of 32

A

B

C

Figure 1

Page 27 of 32 Diabetes

WT WT STZ KO KO STZ 1 2 3 4 5 6 7 8 A B

C

80 KDa

D

E WT WT STZ KO KO STZ F 1 2 3 4 5 6 7 8 9 10 0 0.6 1.2 2.5 5

G WT WT STZ H 1 2 3 4 5 6 7 8 9 10

1 5

10

15

Figure 2

Diabetes Page 28 of 32 New Figure 3

A

B C

D E

Page 29 of 32 Diabetes

A

B

Figure 4

Diabetes Page 30 of 32

A

B

C

Figure 5

Page 31 of 32 Diabetes

Figure 6

Diabetes Page 32 of 32

A B WT SK3 VWF

CT

STZ

KO C SK3 VWF

CT

STZ

D