Endothelial Arginine Resynthesis Contributes to the Maintenance of Vasomotor Function in Male Diabetic Mice

Aim Argininosuccinate synthetase (ASS) is essential for recycling L-citrulline, the by-product of NO synthase (NOS), to the NOS substrate L-arginine. Here, we assessed whether disturbed arginine resynthesis modulates endothelium-dependent vasodilatation in normal and diabetic male mice. Methods and Results Endothelium-selective Ass-deficient mice (Assfl/fl/Tie2Cretg/− = Ass-KOTie2) were generated by crossing Assfl/fl mice ( = control) with Tie2Cre mice. Gene ablation in endothelial cells was confirmed by immunohistochemistry. Blood pressure (MAP) was recorded in 34-week-old male mice. Vasomotor responses were studied in isolated saphenous arteries of 12- and 34-week-old Ass-KOTie2 and control animals. At the age of 10 weeks, diabetes was induced in control and Ass-KOTie2 mice by streptozotocin injections. Vasomotor responses of diabetic animals were studied 10 weeks later. MAP was similar in control and Ass-KOTie2 mice. Depletion of circulating L-arginine by arginase 1 infusion or inhibition of NOS activity with L-NAME resulted in an increased MAP (10 and 30 mmHg, respectively) in control and Ass-KOTie2 mice. Optimal arterial diameter, contractile responses to phenylephrine, and relaxing responses to acetylcholine and sodium nitroprusside were similar in healthy control and Ass-KOTie2 mice. However, in diabetic Ass-KOTie2 mice, relaxation responses to acetylcholine and endothelium-derived NO (EDNO) were significantly reduced when compared to diabetic control mice. Conclusions Absence of endothelial citrulline recycling to arginine did not affect blood pressure and systemic arterial vasomotor responses in healthy mice. EDNO-mediated vasodilatation was significantly more impaired in diabetic Ass-KOTie2 than in control mice demonstrating that endothelial arginine recycling becomes a limiting endothelial function in diabetes.


Introduction
The endothelium regulates vasomotor tone by releasing several relaxing (endothelium-derived relaxing factors, EDRF) and contractile factors (EDCF). The major relaxing factors are nitric oxide (NO), prostacyclin (PGI 2 ) and endothelium-dependent hyperpolarization (EDH). NO is not only an important vasodilator, but also inhibits atherogenic processes, such as smooth musclecell proliferation, platelet adhesion and aggregation and oxidation of low-density lipoproteins (LDL) [1][2][3][4]. Several studies demonstrated an impaired production of endothelial NO in patients with hypertension, heart failure, hypercholesteremia, atherosclerosis, and diabetes [5][6][7][8][9]. Nitric-oxide synthases (NOS) produce NO from the substrate arginine. Reported intracellular concentrations of arginine vary between 300 [10] and 800 mM [11], which is much higher than the K m (3 mM) for endothelial NOS (NOS3). Despite this high intracellular arginine concentration, improved NO production [11] or improved endothelial function of small coronary vessels [12] have been reported after arginine supplementation. This phenomenon, which is known as the arginine paradox [13,14], shows that the intracellular arginine concentration can become limiting under some conditions. Intracellular availability of arginine depends on transport, recycling, metabolism and catabolism [15].
Arginine can be resynthesized from citrulline, the by-product of NO production, via argininosuccinate synthetase (ASS) and argininosuccinate lyase (ASL). Both enzymes are expressed in many cell types [16]. Arginine is catabolized by arginases to ornithine and urea. The two isoforms, arginase 1 (cytoplasmic, also known as liver-type) and arginase 2 (mitochondrial, also known as kidney-type) are both reported to be expressed in endothelial cells [17,18]. An increased activity of both arginase 1 and arginase 2 was demonstrated in diabetes and aging [19,20], two conditions, which are associated with decreased NO production.
Although intracellular arginine sources for NOS3 are controversial, previous in-vitro studies have shown that arginine recycling is important for NO production [21]. It has, however, not yet been demonstrated whether this system is also relevant in endothelial cells in vivo. We hypothesize that deficient arginine resynthesis from citrulline in the endothelium predisposes to endothelial dysfunction (ED), which will be aggravated in diabetes. We tested this hypothesis in mice with a genetically impaired capacity to recycle arginine in their endothelium and investigated their saphenous arteries. We have previously shown that upon aging endothelium-dependent relaxing responses to acetylcholine become predominantly mediated by endothelium-derived NO in these muscular resistance arteries [22].

Animals
All procedures were performed in accordance with the guidelines of the Committee for Animal Care and Use of Maastricht University and have been approved by this Committee. Approval numbers for the protocols used in this study were: DEC 2008-182 and DEC 2012-027. Animals were killed by CO2/O2 inhalation to isolate blood vessels for vascular reactivity measurements.
For invasive hemodynamics, indwelling catheters were placed under isoflurane anesthesia. Analgesia was obtained by perioperative subcutaneous injections of buprenorphine (0.03 mg/kg). Two days after introduction of the catheters, blood pressure was measured in conscious animals. After the experiments, animals received 250 mg/kg pentobarbital through the catheter for euthanasia.
Endothelial Ass-deficient (Ass fl/fl /Tie2Cre tg/2 ) mice were generated by crossing animals carrying the floxed allele Ass fl/fl [23] with Tie2Cre mice. The endothelial knockout animals will be designated as Ass-KO Tie2 , and the Ass fl/fl mice as controls. We have previously shown that Ass fl/fl mice are indistinguishable from their wild type littermates [23]. 12-and 34-week-old male and female mice were used for the experiments. Animals were housed in standard cages (constant room temperature and humidity, 12 hr light/dark cycles) and had free access to standard chow (pellets) and tap water. Diabetes was induced at the age of 10 weeks by intraperitoneal (IP) injections of streptozotocin (STZ, 50 mg/kg) for 5 consecutive days (AMDCC protocols; https://www. diacomp.org). Fasting blood glucose was measured after 1, 4, and 10 weeks following STZ injections, and male mice with stable blood glucose levels of $20 mmol/L were used for the experiments (mean 6 SEM: 2260.7 mmol/L, n = 8). Female mice were excluded from these experiments due to low fasting blood glucose levels (mean 6 SEM: 7.760.3 mmol/L, n = 11; Table S1) 10 weeks after the streptozotocin treatment.

Immunohistochemistry
Saphenous arteries were fixed in 4% phosphate-buffered formalin at room temperature (RT) for 4 hrs and embedded in paraffin. Sections (4 mm thick) were rehydrated and boiled in sodium citrate buffer (10 mM, pH 6.0) for 15 min for epitope retrieval. Subsequently, sections were incubated overnight at 4uC in a humidified chamber with rabbit antibodies directed against ASS (1:10,000 in normal goat serum (NGS); AMC, Amsterdam [24]). Excess antibody was washed off with PBS before sections were incubated with horseradish peroxidase-coupled rabbit antibodies against sheep IgG (1:400 in NGS, DAKO, Glostrup, DK). The localization of HRP was visualized with 3, 3,diaminobenzidine (Sigma Aldrich). The presence of ASS was visualized with an Axioscope (Carl Zeiss, Jena, Germany) and a standard charge-coupled digital camera (model DFC 280; Leica, Wetzlar, Germany).

Plasma amino-acid analysis
For the determination of plasma amino acids, 50 mL of plasma was added to 4 mg sulfosalicylic acid, vortexed, snap-frozen in liquid nitrogen and stored at 280uC until use. The acid plasma supernatant was used for amino-acid analysis on a gradient reversed-phase HPLC system as described [25]. Prior to separation on a BDS Hypersil C18 column (Thermo Scientific, Breda, The Netherlands), the amino acids were labeled with ophtalaldehyde (Pierce)/3-mercaptopropionic acid (Sigma). The amino acids were eluted with an acetonitrile gradient (0.7 mL/ min) and fluorescence detected at 450 nm (excitation: 340 nm).

In vivo hemodynamics
We determined hemodynamic parameters in conscious, unrestrained control and Ass-KO Tie2 mice. Heparinized indwelling polyethylene catheters were introduced under isoflurane anesthesia into the femoral artery and jugular vein, two days before the experiments. Analgesia was obtained by perioperative subcutaneous injections of buprenorphine (0.03 mg/kg). On the day of the experiment, the arterial line was connected to a pressure transducer (Micro Switch 150 PC) and the signal was sampled at 2.5 kHz. MAP and heart rate (HR) were derived from this signal using the IDEEQ data acquisition system (Instrument Services, Maastricht University, NL). The venous line was extended outside the cage and filled with 0.9% NaCl solution. Hemodynamic parameters were allowed to stabilize before pharmacological interventions. The following compounds were tested: bovine arginase 1 (200 U (Cell Sciences, Canton, MA, USA), dissolved in HEPES buffer (pH 7.4) and administered as an intravenous (IV) bolus), L-NAME (10 mg/kg, dissolved in 0.9% saline, IV), and ACh (7 mg/kg.min, dissolved in 0.9% NaCl, IV infusion). After the experiments, animals were euthanized with 250 mg/kg pentobarbital applied through the catheter. We were not able to cannulate the femoral arteries of diabetic mice because of the poor general condition of these animals and the reduction of vessel size.

In vitro studies
Tissue preparation. Saphenous arteries, muscular resistance arteries with a diameter of ,250 mm that are well suited to assess vasomotor responses ex vivo [22,26], were used to study endothelium-dependent vasodilatation. Animals were euthanized by CO 2 / O 2 inhalation. The arteries were dissected free from surrounding fat and connective tissue, and were mounted in a wire myograph (Danish MyoTechnology, Aarhus, DK). Arterial segments (2 mm long) were distended to the diameter at which maximal contractile responses to 10 mM NA (noradrenaline) were obtained [22,27]. Optimal diameters (D opt ) and maximal contractile responses to NA for male mice are summarized in Table S2.
Contribution of NO, EDH and cyclooxygenase products to endothelium-dependent relaxation. A concentration-response curve (CRC) for PHE (0.01-10 mM) was recorded. During the contraction induced by 10 mM PHE, a CRC for ACh (0.01-10 mM) was generated. Thirty min later, arteries were contracted using K + (40 mM) and again a CRC for ACh (0.01-10 mM) was recorded. These experiments were repeated in the presence of the cyclooxygenase inhibitor indomethacin (INDO, 10 mM) and in the presence of both INDO and the NOS inhibitor L-NAME (100 mM).
Sensitivity of vascular smooth muscle to NO. Arteries were contracted with PHE (10 mM) in the presence of INDO (10 mM) and L-NAME (100 mM), and the relaxing effects of the NO donor SNP (0.01-10 mM) were recorded.

Statistical analysis
All CRCs for contractile stimuli were expressed as percentage of the maximal response to 10 mM NA prior to the administration of any pharmacological inhibitor. Relaxing responses were expressed as percentage of the level of pre-contraction. Individual CRCs were fitted to a sigmoid regression curve (Graphpad Prism 5.0). Sensitivity (pEC 50 ) and maximal effect (E max ) are shown as means 6 SEM. Unpaired Student's t-tests were performed to determine the significance of differences in pEC 50 and E max .

Ablation of Ass gene from endothelial cells
To verify that there was indeed no ASS expression in the endothelium of Ass-KO Tie2 mice, paraffin-embedded saphenous arteries were sectioned and stained immunohistochemically for ASS. Specific staining was observed in the arteries of control mice ( Figure 1A), but was absent in arteries from Ass-KO Tie2 mice ( Figure 1C). The presence of intact endothelium was confirmed by H&E staining (Figures 1B, D).

Plasma arginine concentrations in male control and Ass-KO Tie2 mice
ASS catalyzes the conversion of citrulline to argininosuccinate, the rate-determining step in arginine resynthesis from citrulline. To evaluate the effect of endothelial Ass gene deletion on plasma concentrations of arginine and citrulline, plasma amino-acid profiles were determined in healthy and diabetic control and Ass-KO Tie2 mice. Arginine and citrulline concentrations were similar in healthy control (117610 and 5664 mmol/L, resp.) and Ass-KO Tie2 (12766 and 6263 mmol/L, resp.). Similar results were also observed in diabetic control (128612 and 6865 mmol/ L, resp.) and Ass-KO Tie2 (129621 and 73611 mmol/L, resp.) (Table S2). This suggests that the gene ablation did not result in a major disturbance of circulating arginine and citrulline concentrations.

Hemodynamics in control and Ass-KO Tie2 mice
To evaluate the effect of endothelial Ass deletion on hemodynamics, mean arterial pressure was recorded in conscious male mice as described in Materials and Methods. MAP did not differ between male control (10262 mmHg) and Ass-KO Tie2 (10763 mmHg) mice (Figure 2A). To assess the role of circulating arginine in blood pressure maintenance, 34-week-old control mice received an intravenous bolus of 200 U arginase 1, which resulted in a rapid decrease of the circulating arginine concentration to ,13% of the original plasma concentration (84 mmol/L). The lowest arginine concentration was achieved within 10 minutes and the concentration remained at this low level for at least 20 minutes (as determined beforehand in 12-week-old male control mice; Figure S1). This led to a similar (P = 0.66), significant increase of MAP in control (+1063 mmHg) and Ass-KO Tie2 (+1263 mmHg) mice ( Figure 2A). A comparable increase was found in a single female Ass-KO Tie2 mouse (MAP 98 and 116 mm Hg under basal conditions and after arginase 1 treatment, respectively). In comparison, a bolus injection of the NOS inhibitor L-NAME (10 mg/kg) resulted in a threefold larger increase of MAP in both control (+3763 mmHg) and Ass-KO Tie2 (+3461 mmHg) male mice (difference between genotypes: P = 0.42; Figure 2B). These data show that in healthy mice, circulating arginine is essential for blood pressure maintenance.

Contractile reactivity of control and Ass-KO Tie2 arteries in vitro
To assess the effects of Ass gene ablation on vasomotor responses in vitro, we characterized the contractile responses of muscular resistance arteries. Saphenous arteries of male control and Ass-KO Tie2 mice at 12 and 34 weeks of age were isolated and subjected to wire myography. The maximal contractile response to 10 mM NA was comparable in control and Ass-KO Tie2 mice (Table S2) in both age groups. Furthermore, the sensitivity (pEC 50 (2log M), Table S2) and maximal contraction (E max ) to PHE (0.01-10 mM) or K + (40 mM) in the absence or presence of NOSand cyclooxygenase inhibitors were similar in all groups (Table  S2). The lack of arginine resynthesis did not affect contractile responses. Another group of mice was then rendered diabetic by streptozotocin injections to assess the role of arginine resynthesis under pathological conditions. Contractile responses again did not differ between male diabetic control and diabetic Ass-KO Tie2 mice (Table S2).

Relaxing responses to ACh during PHE-induced contraction
We then characterized relaxation responses to ACh in healthy and diabetic Ass-KO Tie2 and age-matched control mice to assess whether resynthesis of arginine from citrulline may become more important in compromised vessels. E max and pEC 50 to ACh were similar in healthy control and Ass-KO Tie2 (9262% and 6.660.1 versus 9064% and 6.660.1) male mice at 12-weeks of age in the absence ( Figure 3A; Table 1) and presence of INDO ( Figure 3B, Table 1). Similar results were also observed in 34-week-old male and female mice of both age groups (Figures 3D-E and Figure S2; Table 1; Table S3). In male diabetic Ass-KO Tie2 mice, however, the E max to ACh was significantly lower (6168%) than in diabetic control mice (8666%) in the absence ( Figure 3G; Table 1) and presence of INDO (5169 and 8164, resp.; Figure 3H; Table 1). In the presence of both INDO and L-NAME, however, relaxation was suppressed to a similar degree in both healthy and diabetic control and Ass-KO Tie2 mice (Figures 3C, F and I; Table 1). Female mice subjected to the same STZ protocol as male mice only temporarily developed elevated blood glucose concentrations, but by 10 weeks after the last STZ treatment, blood glucose was back to normal concentrations (see Table S2). We, nevertheless, measured vascular relaxation in 3 control and 3 Ass-KO Tie2 female mice ( Figure S2, G-I) and observed no difference between control and STZ-treated mice (P = 0.294 for diabetic control versus diabetic Ass-KO Tie2 mice without inhibitors and P = 0.233 in the presence of INDO). We conclude from these data that impaired endothelial arginine resynthesis is responsible for the diminished endothelium-dependent relaxation in male diabetic Ass-KO Tie2 mice.

Endothelium-derived NO
To evaluate the contribution of endothelium-derived NO in vascular relaxation, we inhibited EDH-mediated relaxations by depolarizing the vessels with high potassium buffer ([K + ] = 40 mM) and inhibited cyclooxygenases with INDO [22]. Maximal relaxations to ACh were comparable in healthy control and Ass-KO Tie2 mice of both age groups (Figures 4A, B; Table 1). In diabetic mice, however, E max to ACh was significantly lower in Ass-KO Tie2 (3564%) than in control mice (4962%) (P = 0.008; Figure 4C; Table 1). This shows that EDNO-dependent relaxation does not require arginine resynthesis in vessels of healthy mice, whereas NO production relies at least partially on arginine resynthesis in vessels of diabetic mice.

Relaxing responses to SNP
To confirm that the responses of the vascular smooth muscle cells were not affected by the genetic manipulation, we blocked endothelial NO production and measured endothelium-independent relaxation in response to an NO donor. PHE-contracted arteries were treated with L-NAME (100 mM) and INDO to block the production of NO and prostaglandins, respectively. Subsequently, the relaxing response to the NO-donor SNP (0.01-10 mM) was measured. pEC 50 and E max to SNP were comparable in vessels of healthy ( Figure 5A, B and Figure S3; Table 1) and diabetic ( Figures 5C; Table 1) control and Ass-KO Tie2 mice. Relaxing responses to the endothelium-independent NO donor SNP were not affected by genotype, age, or diabetes, indicating that the sensitivity of the vascular smooth muscle cells to NO was unchanged.

Discussion
In the present study, we evaluated whether deficient arginine resynthesis via endothelial ASS predisposes to endothelial dysfunction. In addition, we addressed the question whether deficient arginine resynthesis aggravates endothelial dysfunction in diabetes. The major finding of this study is that endotheliumdependent relaxations were clearly diminished by endothelial ASS deficiency in diabetic mice, indicating that arginine resynthesis is required to maintain NO production in such compromised vessels. In healthy mice, however, elimination of the Ass gene did not influence vasomotor responses or hemodynamic parameters. Apparently, arginine resynthesis is not rate-limiting for NO production in the endothelium of healthy arteries.
We used Tie2 as promoter for the Cre gene to delete the floxed Ass allele in endothelial cells. It is well established that the Tie2 promoter-enhancer is active in endothelial cells and early hematopoietic precursors [28], resulting in the ablation of the floxed allele in erythrocytes, macrophages, B-cells and T-cells. We, however, never observed ASS protein expression in erythrocytes or lymphocytes of control mice, which makes an effect of deletion of the Ass gene in these cells in our experiments unlikely. Expression of Ass in macrophages has been reported [29], but saphenous arteries of diabetic mice did not show inflammatory changes or ASS-positive cells in their vascular walls ( Figure S4 G, H). Based on these findings, it is unlikely that the presence or absence of ASS protein in macrophages or other hematopoietic cells affected our data.
Blood pressure was recorded in unrestrained mice to assess the effect of ASS deficiency on hemodynamics. Baseline blood pressure values did not differ between control and knockout mice. In addition, L-NAME-induced blood-pressure increases were similar in both groups, suggesting that the contribution of NO to hemodynamics was not affected by ASS deficiency. We used intra-arterial arginase 1 infusion to address the question to what extent plasma arginine contributes to blood pressure regulation. As expected, arginase 1 infusion drastically reduced the plasma arginine concentration and led to a small, but significant increase of MAP. This finding, which appears to reflect the essence of the ''arginine paradox'' [13], implies that endothelial NO production declines under this condition, because endothelial arginine consumption exceeds its supply or because NOS3 activity is rapidly inactivated in an [arginine]-dependent way. However, the observed increase in MAP after arginine depletion was much smaller than that induced by inhibition of NOS by L-NAME infusion. These findings show that plasma arginine concentration is a determinant of blood pressure, but also that endothelial cells have alternative arginine sources for NO generation.
We used wire myography to study the role of endothelial arginine resynthesis in NO-mediated endothelium-dependent vasodilatation in saphenous arteries. In our previous work, we showed that the relaxation responses in these arteries depend on NO and EDH [22]. In addition, we showed that the contribution of these relaxing factors changed with age. In the present study, we compared the contribution of relaxing factors in 12-and 34-weekold Ass-KO Tie2 and control mice and did not find differences in the relaxation responses of healthy mice of both genotypes. Interestingly and consistent with other studies [30], the relaxation responses mediated by EDH were reduced in diabetic mice compared to healthy mice. We used the classical KRB buffer that does not contain arginine to focus on the contribution of resynthesized arginine to NO production. NO-mediated relaxations were significantly reduced in diabetic Ass-KO Tie2 mice when compared to diabetic control mice. Since all relaxation differences between control and Ass-KO Tie2 mice were abolished by the presence of L-NAME, they were not due to the effects of ASS deficiency on EDH-mediated relaxations. In addition, SNPinduced relaxations displayed similar pEC 50 and E max in both genotypes. We also did not find quantitative differences in the response to SNP between diabetic control and diabetic Ass-KO Tie2 mice. The difference between control and Ass-KO Tie2 mice was, therefore, not due to an altered sensitivity of smooth muscle cells to NO. We have considered carrying out experiments on diabetic mice supplemented with arginine and myograph experiments with isolated arteries from Ass-KO Tie2 mice in the presence of arginine. In principle, both interventions should reverse the effect of deficient arginine recycling. However, because our recent studies showed that intravascular arginine supplementation does not increase intracellular arginine availability and that, instead, intravascular citrulline is the supplementation of choice [31], we did not carry out such experiments. Further support comes from a recent publication in Hypertension that demonstrated that supplementation with L-citrulline was more effective in preventing glomerular hyperfiltration and proteinuria in diabetic rats than L-arginine supplementation, even though both increased plasma L-arginine concentrations [32].
The difference in the magnitude of the effect of intravenous arginase 1 infusion and L-NAME administration shows that arterial endothelium can use arginine from several different arginine pools. These pools, which include intracellular arginine that is resynthesized from citrulline or released from protein breakdown, and extracellular arginine imported via specific transporters, may be available to NOS3 under different circumstances [33,34]. In healthy arteries, ASS deficiency apparently does not lead to profound endothelial dysfunction because of extensive functional redundancy of the arginine pools. In arteries of diabetic mice, however, we observed that ASS deficiency reduced NO-mediated endothelium-dependent relaxations. It was reported earlier that diabetes attenuates the endothelium-dependent relaxation responses and increases expression and activity of arginases in the aortic wall [33]. However, we did not observe arginase 1 or arginase 2 expression in diabetic saphenous arteries ( Figure S4 A-F). One should keep in mind that since arginases have a very high catalytic activity, amounts that are not detectable by immunohistochemistry could still represent physiologically relevant activities. STZ-induced diabetes was shown to decrease expression of the arginine transporter CAT1 in the kidney [35]. Although a similar effect of diabetes on CAT1 in saphenous artery  Table 1). *P,0.05 vs. the control, **P,0.01 vs. the control (unpaired t-test). doi:10.1371/journal.pone.0102264.g003 endothelium has not been reported thus far, downregulation of arginine transporter(s) may contribute to the observed dependence on arginine resynthesis in diabetes to maintain adequate intracellular arginine availability for NOS3. Whether or not endothelial protein degradation is enhanced in diabetic mice remains to be sorted out [36][37][38], but even if it is increased, it can probably not affect arginine availability under the long-term steady state circumstances that we used in the current experiments.
An aspect that requires attention in future studies is that endothelial cells in intact resistance arteries are coupled to smooth muscle cells via gap junctions [39]. These proteins allow for diffusion of small molecules (,1000 Da), including free amino acids, from one cell to another [40]. It is therefore conceivable that the smooth muscle cells in arteries from healthy mice represent an arginine reservoir for endothelial cells. In endothelial cells, gap junctions are primarily formed of connexins proteins CX37, CX40 Table 1. Effect of endothelium-specific Ass deletion on relaxation responses in male mice.

Control
Ass-KO Tie2 pEC 50 E max % n pEC 50 E max % n 12-week-old mice and CX43. Interestingly, their expression is reduced in vascular walls of diabetic mice [41,42]. Unfortunately, it is technically challenging to establish whether a gap junction-dependent arginine flux contributes to the maintenance of intra-endothelial arginine concentration. Firstly, Cx43 deficiency is neonatally lethal [43] and secondly, both Cx40 [24] and Cx37 [44] have a direct interaction with NOS3, with Cx37 deficiency even increasing NO production in vitro [44]. Pharmacological tools, such as carbenoxolone and heptanol, are notoriously non-selective [45], while the applicability of the ''GAP'' peptides cocktail in vivo and their specificity with respect to the homo-and hetero-cellular communication still need to be explored [46]. Although the aforementioned issues complicate the firm establishment of a role for gap junctions in arginine bioavailability in the endothelium, we speculate that diabetic Ass-KO Tie2 mice display endothelial dysfunction due to a decreased gap junction-dependent arginine flux.
The concentration of intra-endothelial arginine may also indirectly affect the production of NO. Previous studies showed that arginine supplementation increases the transcription of GTP cyclohydrolase 1 in diabetic rats [47]. GTP cyclohydrolase 1, the first enzyme in the de novo synthesis of BH4, elevates the intracellular concentration of BH4 which is a necessary cofactor for NOS3 activity [47]. In our diabetic Ass-KO Tie2 mice, impaired resynthesis of arginine could be responsible for the uncoupling of NOS3 due to reduced BH4 production, but this notion needs to be investigated further.
In summary, the present study shows that deletion of the floxed Ass gene with Cre recombinase under the control of Tie2-cre promoter does not affect MAP or heart rate in healthy mice. In addition, in vitro studies of isolated saphenous arteries showed that, in healthy mice, relaxation responses were unaffected by the ablation of the Ass gene. In diabetic mice, however, ablation of Ass resulted in diminished endothelium-derived NO-mediated vascular relaxation responses. These results are exciting, since they suggest that diabetic patients suffering from endothelial dysfunction may benefit from therapies focusing on either increasing ASS activity or boosting intracellular arginine levels. In this respect it is interesting to note that Ass gene expression is diminished in STZtreated rats and that insulin treatment upregulates ASS transcription in these animals [48].  Figure S4 Immunohistochemical staining for the presence of arginase 1, -2 and ASS in the walls of saphenous arteries of diabetic mice. Panels A and D represent staining for arginase 1 and 2, respectively. Note the absence of arginase 1 and -2 positive cells both in the endothelium and the media/ adventitia. Panels B and E represent the negative controls for arginase 1 and -2, respectively. Panels C and F show positive controls for arginase 1 (liver) and arginase 2 (kidney cortex). Note that plasma proteins do cause background staining for arginase 1. Panel G shows ASS staining of the endothelium, but no ASSpositive cells in the tunica media. Panel H shows an H&E staining  Table 1). doi:10.1371/journal.pone.0102264.g005 of the vessel shown in panel G to demonstrate absence of inflammatory changes. Bar = 10 mm for all panels. (PPT) Table S1 Fasting plasma blood glucose concentrations in male and female control and Ass-KO Tie2 mice before and after streptozotocin treatment. Mice were fasted for 4 hours before blood glucose was measured before or 1, 4, or 10 weeks after the last STZ injection. Data are shown as mean 6 SEM (n = 5 for STZ-treated mice). Note that basal blood glucose values for male and female control mice were taken from 12-to 15-week-old C57BL/6J wild type mice in another experiment. Basal values for Ass-KO Tie2 mice (12-week-old) are from this series of experiments. (DOCX)