Asymmetric dimethylarginine positively modulates Calcium-Sensing Receptor signalling to promote lipid accumulation and adiposity

Irreversible methylation of arginine residues generates asymmetric dimethylarginine (ADMA). ADMA is a competitive inhibitor of nitric oxide (NO) synthase and an independent risk factor for cardiovascular disease. Plasma ADMA concentrations increase with obesity and fall following weight loss. Here, we demonstrate that ADMA drives lipid accumulation through a newly identified NO-independent pathway via the amino-acid sensitive calcium-sensing receptor (CaSR). ADMA treatment of 3T3-L1 and HepG2 cells activates mTOR signalling and upregulates a suite of lipogenic genes with an associated increase in triglyceride content. Pharmacological blockade of CaSR inhibits ADMA driven lipid accumulation and ADMA treatment potentiates CaSR signalling via both Gq and Gi/o pathways. Impairment of ADMA metabolism in adipocytes in vivo, by dimethylamine dimethylaminohydrolase-1 (DDAH1) deletion, increases visceral adiposity and adipocyte hypertrophy. This study identifies a signalling mechanism for ADMA as an endogenous ligand of the G protein-coupled receptor CaSR that potentially contributes to the impact of ADMA in cardiometabolic disease.


Introduction
Methylarginines are formed via the methylation of arginine residues by protein arginine methyltransferases (PRMT) (Bedford and Clarke, 2009). The PRMT family consists of nine enzymes each encoded by a separate gene. These methylate histone proteins as a key step in transcriptional regulation and a wider range of non-histone proteins involved in protein synthesis. All PRMT enzymes are capable of forming L-N G monomethylated arginine (L-NMMA); while type I PRMTs, predominantly PRMT 1, form asymmetric dimethylarginine (ADMA) and type II PRMTs, predominantly PRMT 5, form symmetric dimethylarginine (SDMA). Unlike other post-translational modifications, arginine methylation is irreversible and therefore, proteolysis generates a constant flow of free methylarginines. These are initially released into the cytosol from where they may exit the cell to the circulation and ultimately be cleared by the kidney (Vallance et al., 1992a). Previous studies, from our group and others, have demonstrated that asymmetrically methylated methylarginines are competitive inhibitors of all three isoforms of nitric oxide synthase (Vallance et al., 1992b;Cardounel and Zweier, 2002;Leiper et al., 2007). In contrast, SDMA has no inhibitory action on NOS. As ADMA concentrations are ~10-fold higher than that of L-NMMA it is thought ADMA is the most significant endogenously produced inhibitor of NO synthesis.
Relatively modest increases in ADMA in both humans and experimental models are associated with profound disruption of cardiovascular homeostasis and have been linked to an increased risk of hypertension, atherosclerosis and stroke (Boger et al., 1998;Leiper et al., 2007;Lambden et al., 2015). Further clinical studies have associated elevated plasma ADMA concentrations with obesity and the metabolic syndrome (Eid et al., 2004;Kocak et al., 2011;McLaughlin et al., 2006;Palomo et al., 2011). Concentrations vary widely between studies but on average increase by 0.3 μM from ~0.95 μM to ~1.3 μM. ADMA accumulates in the visceral adipose tissue of obese subjects (Assar et al., 2016) and circulating ADMA levels decrease upon weight loss although causal mechanisms have not been identified (McLaughlin et al., 2006;Krzyzanowska et al., 2004). Interestingly however, NO levels are also increased in obesity and correlate with both BMI and adiposity (Choi et al., 2001) with higher expression of eNOS and iNOS in the adipose tissue of obese individuals (Elizalde et al., 2000). These observations might suggest additional roles of ADMA in adipose tissue that are independent of NOS. Therefore, in this study we set out to test the hypothesis that ADMA regulates adipocyte function and physiology via a NO-independent mechanism. Upon establishing that NO-independent pathways exist in both adipocytes and hepatocytes we then went on to test the hypothesis that ADMA may act via the amino-acid sensitive calciumsensing receptor.

Contact for reagents and resource sharing
Further information and requests for resources and reagents can be directed to, James Leiper (james.leiper@glasgow.ac.uk) or Laura Dowsett (laura.dowsett@glasgow.ac.uk).

Mice
All studies were conducted in accordance with UK Home Office Animals (Scientific Procedures) Act 1986 and institutional guidelines with the Imperial College ethics review board approval.
Cre-LoxP site were inserted into the DDAH1 gene on either side of exon one . The adipocyte specific DDAH1 knockout mouse (DDAH1 Ad-/-) was developed by crossing the DDAH1 fl/fl with the Adiponectin driven cre mouse strain (FVB-Tg(Adipoqcre)1Evdr/J) developed by Evan Rosen et al. (Eguchi et al., 2011) DDAH1 fl/fl mice were used as controls. All mice were derived on a C57/bl6 background. Mice were maintained in house on a fixed light/dark cycle and fed ad libitum. Male DDAH1 Ad-/and littermate controls were weighed every two weeks from six weeks of age. Fat and lean mass was assessed using a body composition analysis Scanner (EchoMRI, Houston, USA). Whole body calorimetry was measured using the CLAMS: Comprehensive lab Animal Monitoring System (Columbus Instruments, Columbus, USA). Mice were acclimatised in the system for 8 hours before data collection for the following 72 hours.

Immortalized Cell Lines
The murine pre-adipocyte cell line 3T3-L1 was purchased from American Type Culture Collection (ATCC CL-173) and subcultured according to manufacturer's instruction at 37 °C in a humidified atmosphere with 5% CO 2 . To stimulate differentiation to mature adipocytes, 2 days post-confluent cells were exposed to high glucose (4.5 g/L) Dulbecco's modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 1x penicillinstreptomycin and 2mM L-glutamine and containing 10 µg/ml insulin (Sigma-Aldrich, I9278), 1 µM dexamethasone (Sigma-Aldrich, D1756) and 0.5 mM iso-butyl-methylxanthine (Sigma-Aldrich, I5879). Induction media was replaced two days later with media containing 10 µg/ml insulin only. Cells were then cultured a further 4 days in high glucose DMEM. Adipocytes were therefore considered mature at 10 days post induction.
The human hepatoma cell line HepG2 was purchased from American type Culture Collection (ATCC HB-8065) Cells were maintained at 37 °C in a humidified atmosphere with 5% CO 2 in DMEM supplemented with 10% FBS, 2 mM glutamine and 1% penicillin-streptomycin.

Primary Murine Adipocytes
Primary adipocytes were isolated as developed by Rodbell (Rodbell, 1964). Epididymal fat pads were removed from 22 week old mice. Fat pads were washed in PBS and dissected to remove larger blood vessels. The remaining tissue was digested in low-glucose (1 mg/L) DMEM supplemented with 100 mM HEPES, 1.5% BSA and 0.2% collagenase II, with gentle shaking at 37°C for 1 hour. Cells were then centrifuged for 2 minutes at 400 x g; the floating fat cells were collected, washed in PBS and processed for future analysis.

Method Details
Cell Culture Treatments

HepG2 Treatments
HepG2 cells treatments were performed as above with the 3T3-L1 cells but over an 18 hour time course for studies exploring mRNA expression.

Triglycerides and cholesterol quantification
Triglyceride content was measured using Triglycerides Quantification Assay kit from Abcam (ab65336) and total cholesterol content was assessed with Amplex Red Cholesterol Assay Kit from Invitrogen (A12216) according to manufacturer's instructions. Free fatty acids were measured using an enzyme-based method from Abcam (ab65341).

SREBP -1 Transcription Factor Assay
Fully differentiated 3T3-L1 cells were treated for 6 hours with 10µM ADMA. Nuclear extracts were made using the nuclear extraction kit from Abcam (ab113474). Nuclear SREBP-1 was detected using the SREBP-1 transcription factor assay kit from Abcam (ab133125) which uses an ELISA based approach to detect the change in nuclear SREBP-1 in a 96-well format.

Insulin Sensitivity
Glucose uptake was measured by the uptake of 2-[ 3 H]-deoxy-D-glucose as described previously (Boyle et al., 2011). Briefly, 3T3-L1 adipocytes were stimulated with 10 μ M ADMA for 48 hours, the medium removed and replaced with Krebs-Ringer-phosphate (KRP (128 mM NaCl, 4.7 mM KCl, 5 mM NaH 2 PO 4 (pH 7.4), 1.2 mM MgSO 4 , 2.5 mM CaCl 2 , 1% (w/v) BSA, 3 mM glucose)) buffer containing the same concentration of ADMA. After incubation for 1 h at 37°C, cells were incubated in the presence or absence of 0-100 nM insulin for 12 min and transport was initiated by adding [ 3 H]-2-deoxyglucose (50 μ M, 2 μ Ci/ml). Cells were incubated for 3 min, after which cells were rapidly washed in ice-cold PBS, air-dried, and then solubilized in 1% (v/v) Triton X-100. The radioactivity associated with the cells was determined by liquid scintillation spectrophotometry. Non-specific cellassociated radioactivity was determined in parallel incubations performed in the presence of 10 μ M cytochalasin B.

RNA isolation and Rt-qPCR analysis
RNA was extracted and isolated from both cell cultures and mouse tissues using Qiagen's Universal RNeasy Kit. Mouse tissues were homogenised using a tissue-lyser (Qiagen). cDNA was synthesised from total RNA with iScript cDNA Synthesis kit (Bio-rad). Real-time PCR's were run on the 7900 Fast Real-time PCR system (Applied Biosciences) using SYBR green Supermix (Bio-rad). Fold changes in mRNA expression were calculated from at least three independent experiments using the standard curve method and normalised to 18S gene expression. Primer sequences are provided in Supplementary Table 1.

Western blotting
Following protein electrophoresis and protein transfer, the membranes were probed with primary antibodies from the following sources: mTOR antibody (Cell Signalling, #2972), total AKT (Cell Signalling, #9272), phospho-AKT, Ser 473 (Cell Signalling, #4060), goat monoclonal DDAH1 and DDAH2 antibodies both developed in house 24 , CaSR (Abcam, ab137408)). Loading controls used were pan-actin (Cell Signalling #8456), and vinculin (Invitrogen #700062). Primary antibodies were used at dilution 1:1000 and secondary antibodies at dilution 1:5000. The relative intensity of the immune-reactive bands was determined by densitometry using Li-COR Image Studio Software. The results were normalised to the loading protein and expressed as fold change over control. Untreated control samples were included in every experiment and all experiments were repeated at least 3 times.

Immunohistochemistry
Cells were plated on coverslips, cultured and then fixed with 4% formaldehyde solution in PBS for 10 minutes at room temperature, washed in PBS for 5 minutes and incubated in 3% BSA, 0.1% Triton-X100 blocking buffer in PBS for 1 hour at room temperature. Rabbit anti-CaSR antibody (Abcam #137408) was incubate 1:200 overnight in 3% BSA at 4°C before washing with 0.1% Triton -X100 in PBS. Secondary antibodies (Alexafluor 488, ThermoFisher) were incubated for 2 hours at room temperature before washing and mounting with Prolong Diamond Antifade with DAPI (ThemoFisher).

Bodipy
Cells were cultured on coverslips, stimulated with various treatments and then fixed with 4% formaldehyde solution in PBS for 10 minutes at room temperature, washed in PBS for 5 minutes and incubated in PBS containing 1 ug/ml Bodipy (ThermoFisher, D3921) for 15 minutes upon agitation and in the dark. Coverslips were washed three times in PBS and

Epididymal fat pad staining
Epididymal fat pads were fixed in 4% PFA overnight and then stored in 50% ethanol until embedding in paraffin and sectioning. To assess adipocyte size the tissue was stained with Elastin Van Geisen Stain and all complete cells within each image was measured.
Macrophage infiltration was measured by counting the number of Galectin-3 positive cells (Cedarlane). Analysis was performed using Image J software.

Flp-In T-REX Hek293-CaSR Cells
Inducible CaSR overexpressing cells were generated using the Flp-In T-REx 293 cell line using the protocol developed by Ward et. al 19 . The N-terminal Myc-tagged human CaSR was inserted by Genescript into the pcDNA5-FRT-TO plasmid. The pcDNA5-FRT-TO and the pOG44 plasmids were transfected using Lipofectamine 2000 (ThermoFisher). Positive cells were selected using hygromycin B. CaSR expression was induced by 0.5 µg/ml doxycycline treatment for 24 hours.

Intracellular Ca 2+ Imaging
Hek293-CaSR cells were incubated at 37˚C with 1 µM Cal520 (Abcam, ab171868) in Optimem Reduced Serum I media (HEPES, 2.4 g/l Sodium Bicarbonate, L-glutamine) containing relevant treatments for 1 hour before imaging. Cells were then washed and then incubated in Ca 2+ -free HEPES buffer (130 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 20 mM HEPES, 10 mM Glucose, pH7.4) containing relevant treatments. Gd 3+ was used as a CaSR agonist (0.01 mM-5 mM), NPS-2143 was used as a CaSR antagonist (10 µM). Changes in cytosolic Ca 2+ were imaged using a Zeiss ZenPro inverted microscope; and FITC fluorescence determined using the physiology definition programme as part of the Zen Software (Zeiss Microscopy GmbH, Germany). Data are represented as the change in fluorescence (FΔ) over the fluorescence at baseline (F0) normalised to the maximum control response to each experiment.

Cyclic AMP Analysis
HEK293-CaSR were plated into 96 well plates and treated with doxycycline 24 hours prior to the experiment to induce CaSR expression. Cells were treated with 0.5 mM IBMX for 30 minutes to block phosphodiesterase activity. Cells were then incubated in Optimem reduced serum media with relevant treatments for 15 minutes; 1 µM forskolin was used to stimulate cAMP production. cAMP concentration was assessed using an ELISA kit (Cell Signalling, #4339) as per manufacturers protocol. Data are presented normalised to the maximal forskolin induced response.

Quantification and Statistical analysis
Statistical analysis was performed using GraphPad Prism 6 Software. Comparisons were carried out with unpaired two-tailed Student's t-test, One-way and Two-way ANOVA with Bonferroni post-hoc tests as appropriate. Dose response curves were assessed by nonlinear regression. Statistical significance was accepted for P<0.05. Data are expressed as ± 1 0 SEM. All in vitro experiments are a mean of at least three independent experiments, numbers for each experiment are confirmed in the figure legends.

ADMA induces lipid accumulation in 3T3-L1 cells through a NO-independent pathway
To evaluate the consequences of chronic ADMA exposure on adipocyte function we utilised the mouse 3T3-L1 cell line as a model for mature adipocytes. 3T3-L1 fibroblasts were differentiated to lipid laden cells through the addition of insulin, dexamethasone and isobutylmethylxanthine. 3T3-L1 cells were considered fully differentiated 10 days post-induction, at which point they were treated with ADMA for an additional 72h. ADMA concentrations were chosen to reflect those occurring in disease. 3T3-L1 cells treated with 1 and 3 µM ADMA showed significant cellular hypertrophy compared to those cultured in control media ( Fig. 1a and b); whereas SDMA (10 μM) had no effect. Given the currently understood mechanism of action of ADMA is as a competitive inhibitor of NOS we assessed the effect of two structurally distinct synthetic NOS inhibitors N(g)-nitro-L-arginine methyl ester (L-NAME, 1 mM) and 1,3-PBI-TU, Dihydrobromide (20 μM) at concentrations to maximally block NO production. Interestingly, these did not cause adipocyte hypertrophy suggesting that the effect of ADMA on cell size may be independent of NOS inhibition. Taken together these data suggest that a NO independent mechanism specific for ADMA drives adipocytes hypertrophy in 3T3-L1 cells.
In order to identify the potential mechanism through which ADMA is acting we first established whether cellular hypertrophy was due to increased lipid content. Adipocytes treated with 3 µM and 10 µM ADMA have increased lipid area per a cell ( Fig. 1c) with a rise in intracellular cholesterol and triglyceride (Fig. 1d). Given that an increase in the proportion of differentiated cells could also elevate the total triglyceride content we established whether ADMA alters 3T3-L1 differentiation. Over the 72 hours of treatment from post-induction day 10 to 13 ADMA had no effect on the percentage of lipid laden cells ( Supplementary Fig. 1a).
Furthermore, ADMA treatment of 3T3-L1 cells throughout the entire differentiation period had no effect on the differentiation marker perilipin ( Supplementary Fig. 1b). Sensitivity of fully differentiated 3T3-L1 cells to insulin was unaltered by ADMA treatment ( Supplementary   Fig. 1c). These observations are consistent with ADMA causing hypertrophy in 3T3-L1 cells predominantly via de-novo lipid synthesis.

ADMA induces lipogenesis in HepG2 cells
As the liver is a significant site for ectopic fat deposition, we assessed whether ADMA treatment alters lipid accumulation in the human lipogenic hepatocyte-derived HepG2 cell line. ADMA increased the lipid content of HepG2 cells, as evidenced by increased BODIPY staining ( Fig. 1e and f). As with 3T3-L1 cells SDMA and NOS inhibitors had no effect on lipid accumulation. HepG2 cell triglyceride content was increased by ADMA treatment (Fig. 1g).
These data indicate that the effects of ADMA are seen in two of the most highly lipogenic cell types and in both mouse and human cells.
As mTOR (mammalian target of rapamycin) is important for SREBP-1 cleavage (Porstmann et al., 2009)  We again utilised HepG2 cells to assess whether ADMA-mTOR signalling occurs in multiple cells types. ADMA again increased the transcription of SREBP-1 target genes Acaca and Ldlr (Fig. 2l) as well as mTOR (Fig. 2m). In HepG2 cells rapamycin also blocked the ADMA driven upregulation of Acaca (Fig. 2n). These data indicate that ADMA drives neolipogenesis through the upregulation of mTOR signalling and SREBP-1 activation in both adipocytes and hepatocytes.

Adipocyte-specific DDAH1 deletion increases body weight and fat mass in mice.
In vivo, dimethylarginine dimethylaminohydrolase (DDAH) enzymes metabolise 80% of asymmetric methylarginines with the remaining 20% excreted by the kidneys (Leiper et al., 2007). In 3T3-L1 cells we established DDAH1 was strongly upregulated through differentiation to mature adipocytes while DDAH2 was significantly down-regulated before returning to basal levels ( Supplementary Figure 2a and b). Therefore, to elevate ADMA concentrations in mature adipocytes, we choose to develop a mouse in which DDAH1 was specifically deleted in mature adipocytes (termed DDAH1 Ad-/-) using the adiponectin-BAC cre driver (Eguchi et al., 2011). DDAH1 floxed mice (DDAH1 fl/fl ) were used as controls . DDAH1 protein expression was reduced by approximately 75% in primary adipocytes (Supplementary Fig. 3a and b). DDAH2 expression was unaffected by DDAH1 deletion (Supplementary Fig. 3a and c). The ADMA content in primary epididymal adipocyte lysates was towards the limit of detection by mass spectrometry. ADMA in DDAH1 Ad-/tended to be higher, but this did not reach statistical significance (Fig. 3a). However, adipocyte NOx content were significantly lower in DDAH1 Ad-/mice (Fig. 3b) suggesting that there is a physiologically significant increase in ADMA. On the other hand, plasma concentrations of both ADMA and NOx were unaffected by DDAH1 adipocyte deletion ( Fig.   3c and d). This is in keeping with other tissue specific DDAH1 knockout mice where systemic ADMA remains unchanged Tomlinson et al., 2015). Male DDAH1 Ad-/mice maintained on a normal chow diet weighed significantly more than littermate controls over a period of 6 to 22 weeks of age (Fig. 3e). This was due to increased fat mass ( Fig. 3f) with no difference in lean mass (Fig. 3g). At 22 weeks of age DDAH1 Ad-/mice had larger epididymal fat depots (Fig. 3h), and elevated plasma free fatty acids (Fig. 3i). Staining of epididymal fat pads demonstrated adipocyte hypertrophy in DDAH1 Ad-/- (Fig. 3j and k) which was accompanied by an increase in macrophage infiltration as detected by galectin-3 staining ( Fig. 3j and l). Analysis of primary adipocytes isolated from DDAH1 Ad-/mice indicated that both mTOR and ACC mRNA expression was significantly elevated compared to littermate controls (Fig. 3m) suggesting ADMA-induced mTOR signalling and adipocyte lipogenesis also occurs in vivo.
To confirm that changes in adipocyte and adipose size in DDAH1 Ad-/mice were due to changes in adipocyte signalling rather than a shift in whole body metabolism we utilised the CLAMS (Comprehensive Lab Animal Monitoring) system. VCO 2 and VO 2 were unaltered in DDAH1 Ad-/-(Supplementary Fig. 4a and b) and over the time period studied both groups consumed the same amount of food ( Supplementary Fig. 4c). Surprisingly, DDAH1 Ad-/mice showed a small but statistically significant increase in physical activity (Supplementary Fig.   4d) with higher total ambulatory counts. The mechanism underlying this change in activity is unclear and will require further investigation. These data are consistent with ADMA driving adipocyte hypertrophy and adipose expansion through a cell autonomous manner rather than through altered metabolic rate.

Adipose ADMA is significantly raised by high fat feeding
In human subjects it has previously been reported that plasma ADMA concentrations are elevated in obese individuals (Eid et al., 2004;Kocak et al., 2011). To discover whether adipocyte DDAH1 plays a role in this increase we placed male DDAH1 Ad-/and control littermates on a high fat diet (HFD) consisting of 60% calories from fat from 6 weeks old for a period of 16 weeks. Following high fat feeding both DDAH1 Ad-/and DDAH1 fl/fl mice gained weight, due to increased fat mass with no change in lean mass compared to control mice on suggesting that ADMA formation is of greater physiological importance than ADMA metabolism in obesity.

Adipocyte DDAH1 expression is reduced in eNOS -/mice
To determine whether the effects of DDAH1 Ad-/deletion in vivo is NO independent we investigated eNOS knockout mice over the same time period. As has previously been

ADMA induced lipogenesis via CaSR activation in 3T3-L1 cells
In comparison to the concentrations of ADMA required for NOS inhibition in cell culture systems (100 μM) the concentration of ADMA capable of driving adipocyte hypertrophy is low (1 µM). At this concentration ADMA entry into cells via the cationic amino acid transporter is likely to be limited due to competition from the high concentration of cationic amino acids in the culture media (Strobel et al., 2012). Therefore, we hypothesised that ADMA may act via an extracellular receptor. The calcium-sensing receptor (CaSR) is a member of the class C family of amino acid sensitive GPCRs (Conigrave and Hampson, 2010;Chun, Zhang and Liu, 2012). CaSR is expressed in both adipocytes and hepatocytes (Cifuentes et al., 2010) and has been linked to adipose dysfunction and fat accumulation (Bravo-Sagua et al., 2016;Villarroel et al., 2016). We confirmed CaSR expression within our 3T3-L1 and HepG2 cultures (Supplementary Figure 7) and that CaSR expression was unaffected by 3T3-L1 differentiation (Fig. 4a). Activation of CaSR with the positive modulator cinacalcet achieved a similar level of adipocyte hypertrophy and lipid accumulation as 10 µM ADMA (Fig. 4b-d). We then investigated whether inhibition of CaSR blocked the lipogenic effects of ADMA on 3T3-L1 cells. The CaSR inhibitors Calhex-231 (10µM) and NPS-2143 (10µM) inhibited the lipogenic effect of ADMA on both adipocyte cell size and lipid area.
Interestingly, NPS-2143 alone but not Calhex-231 caused an increase in 3T3-L1 cell size ( Fig. 4e-g); but both compounds antagonised the effect of ADMA. NPS-2143 inhibited ADMA induced lipid accumulation in HepG2 cells confirming CaSR signalling was not restricted to the adipocyte model ( Fig.4h and i).

ADMA increases CaSR sensitivity
CaSR is a promiscuous GPCR signalling through G q -resulting in increased cytosolic Ca 2+ , G i/o inhibiting cAMP signalling and finally G 12/13 Rho signalling (Conigrave and Ward (2013)).
To determine if ADMA directly modifies CaSR activity we developed a HEK293-CaSR cell line using the Flp-In T-REX transfection system (Ward et al., 2011) which overexpresses CaSR following doxycycline treatment ( Fig. 5a and (Fig. 5e). As L-arginine competes with ADMA for NOS binding we investigated whether the same is true for CaSR. Increasing L-arginine to physiological (80 µM) or supra-physiological (1 mM) concentrations had no effect on CaSR signalling either basally or in the presence of ADMA (Fig. 5f). CaSR is also known to regulate cAMP concentrations via G i/o signalling. Basally neither Gd 3+ nor ADMA altered cAMP concentrations; however, following the addition of forskolin to induce adenylyl cyclase activity Gd 3+ inhibited cAMP production, an effect that was enhanced in the presence of ADMA (Fig. 5g). As Gd 3+ is not an endogenous agonist for CaSR we confirmed that ADMA also had an effect on extracellular Ca 2+ stimulated CaSR activation. Using intracellular Ca 2+ mobilisation to visualise CaSR activation ADMA (10 μM) significantly left-shifted the doseresponse curve to a similar degree as the known CaSR modulator phenylalanine (100 μM) compared to the PBS control (Figure 5h; EC 50 -Control 1.7 mM Ca 2+ ± 0.19, EC 50 -Phenylalanine 0.76 mM Ca 2+ ± 0.11, EC 50 -ADMA 1.1 mM Ca 2+ ± 0.10, P<0.05).

SDMA may act as a competitive antagonist at the amino acid site on CaSR
As a range of amino acid species are known to bind CaSR and modulate its signalling, it is interesting that ADMA causes adipocyte hypertrophy whereas the methylarginine analogue SDMA does not (Figure 1a and b). Therefore, we modelled both ADMA and SDMA within the CaSR amino acid binging site. L-Tryptophan has previously been shown to bind CaSR (Geng et al., 2016;Zhang et al., 2016), the superimposition of ADMA within this site (Fig. 6a) shows both the amino acid moiety and guanidine are well accommodated. The alkyl sidechain makes close interactions with hydrophobic residues on the opposite side of the "hinge" region, particularly ILE416. The overlay of SDMA with ADMA ( Fig. 6b) show that again the amino acid and guanidine are accommodated, however, the differing methylation state of SDMA impacts how the alkyl portion of the side chain can be accommodated. The close interactions between ADMA and ILE416 are not observed in docked SDMA due to the alternative positioning required to accommodate the symmetrical N ω -methylation. These results provide supporting evidence for L-ADMA binding at the L-Trp site and are consistent with our finding that ADMA can serve as a CaSR activator by making interactions that reinforce the closed conformation of CaSR, and that SDMA does not. Given that SDMA seems to be able to bind to CaSR but perhaps not actively modulate it we hypothesised that SDMA may act to antagonise positive allosteric binding. Incubation of HEK-CaSR cells with 100 μM phenylalanine increased intracellular Ca 2+ mobilisation in response to stimulation with 1.6 mM Ca 2+ (Figure 6c) as previously demonstrated in Figure 5h. Co-incubation with increasing SDMA concentrations (30-1000 μM) showed a dose-dependent trend to suppress the effect of phenylalanine. Finally, to determine whether SDMA has a physiological effect on CaSR activity we incubated differentiated 3T3-L1 cells with 10 µM ADMA in the absence or presence of excess (100 µM) SDMA ( Fig. 6d and e). The presence of SDMA blocked ADMA driven adipocyte hypertrophy suggesting that the ADMA/SDMA ratio may be an important modulator of ADMA-CaSR signalling.

Discussion
ADMA is an independent risk factor for cardiovascular disease with elevated plasma concentrations being associated with obesity in clinical studies (Eid et al., 2004;Kocak et al., 2011). However, an increase in the plasma concentration of many amino acids has been observed in obesity (Katsanos and Mandarino, 2011); therefore, we set out to establish whether ADMA is simply a marker of metabolic disease or a mediator of disease pathology and progression. Here we show for the first time that ADMA has a direct effect on adipocyte physiology and propose a new mechanism through which excess ADMA, or DDAH dysregulation, leads to adipocyte hypertrophy via stimulation of the calcium-sensing receptor ( Fig. 7).
To examine what role ADMA plays in adipose physiology we employed the well characterised 3T3-L1 model, in which we established that prolonged (72 hour) exposure to ADMA increased cellular lipid content. In keeping with this, ADMA upregulated the mTOR-SREBP1 pathway leading to increased transcription of the lipogenesis genes Fasn, Acaca and Hmgcr. Similarly, HepG2 cells accumulated lipid when treated with ADMA suggesting this pathway is not exclusively restricted to adipocytes and may be a common feature among lipogenic cells.
To explore this pathway in vivo we developed a novel mouse deficient in DDAH1 specifically in mature adipocytes. In these mice reduced adipocyte methylarginine metabolism was sufficient to increase visceral fat mass on a normal chow diet with a corresponding increase in mTOR and lipogenesis genes as seen in 3T3-L1 cells. The AdipoQ BAC cre utilised in this mouse model (Eguchi et al., 2011) removed the potential confounding effect of DDAH1 deletion in pre-adipocytes; although our experiments in 3T3-L1 cells would suggest that ADMA has no effect on adipocyte differentiation. In contrast when placed on a HFD differences between DDAH1 Ad-/and DDAH1 fl/fl mice were lost ( Supplementary Fig. 5); with the predominant effect a very significant increase in adipose tissue ADMA concentrations in both strains. This would suggest that elevated ADMA concentrations in obesity are a result of increased production rather than dysfunctional ADMA metabolism. Our data presented here contrasts with the observations of Li et al. (2017) who reported that mice globally deficient in DDAH1 gain more fat mass than wild type controls when both are fed a high fat diet. This may suggest that further increases in ADMA concentration including elevated plasma ADMA due to global loss of DDAH1 have a greater effect on adipocyte and hepatocyte function, not seen here due to the tissue specificity.
Although the DDAH1 Ad-/mouse reflects the pathways identified in 3T3-L1 and HepG2 cells following ADMA treatment, DDAH1 deletion also resulted in a significant decrease in adipocyte NOx concentrations. Therefore, we cannot rule out that a reduction in NO production plays a role in fat accumulation in this mouse model. eNOS knockout mice have previously been shown to have increased visceral fat tissue and undergo adipocyte hypertrophy (Nakata et al. 2008) with decreased mitochondrial activity and impaired βoxidation resulting in increased plasma free fatty acid levels (Gouill et al., 2007). However, here we show that eNOS knockout mice ( Supplementary Fig. 6) also have decreased adipocyte DDAH1 expression suggesting dysfunctional ADMA metabolism occurs in this mouse model as well as reduced NO production with both pathways possibly contributing to adipocyte hypertrophy and visceral obesity.
ADMA is a competitive inhibitor of all three NOS isoforms (Vallance et al., 1992a;Cardounel and Zweier, 2002). However, unexpectedly the synthetic NOS inhibitors L-NAME and PBI-TU did not replicate ADMA driven adipocyte hypertrophy or upregulation of mTOR expression in cell culture models. Our data indicates that ADMA has significant NOindependent biological effects at pathophysiological concentrations (1-3 µM). Cardounel et al. (2007) have calculated that these low concentrations would lead to at most a 10% inhibition of NOS. In contrast concentrations of ADMA that are capable of producing significant NOS blockade are only reached during the later stages of chronic kidney disease (Vallance et al., 1992a). Taken  Previous studies have suggested that ADMA may have actions in addition to NOS inhibition.
In our laboratory we examined the effect of physiological and pathophysiological concentrations of ADMA on cultured endothelial cells. We identified ~50 genes that were regulated in response to ADMA and demonstrated that for some of these genes the effect was not replicated by a synthetic NOS inhibitor (L-NIO) (Smith et al., 2005). The pathways regulated by ADMA included BMP and Osteocalcin signalling both of which have been shown to be regulated by CaSR. Our identification of a novel ADMA/CaSR signalling pathway now provides a plausible mechanism that might mediate these effects. Further studies will be required to elucidate the full range of ADMA/CaSR signalling in different cells and tissues. In addition to our own observations, Juretic and co-workers (Juretic et al., 1996) have reported that the induction of IL-2 in L-NMMA treated cultured PBMC's is not replicated 1 by treatment with synthetic inhibitors of NOS. Further work will be needed to establish whether L-NMMA can interact with and modulation CaSR activity.
Our data for the first time identifies a NO-independent receptor for ADMA, the amino acid sensitive GPCR CaSR. CaSR has been previously implicated in lipid homeostasis and its expression is upregulated in fully differentiated adipocytes (He et al., 2011). The CaSR agonist GdCl 3 increases lipid accumulation and pro-adipogenic gene expression in the SW872 pre-adipocyte cell line (He et al., 2011); while cinacalcet increases HepG2 triglyceride content in culture (Villarroel et al., 2016). Furthermore, Rybchyn et al. (2019) have recently demonstrated CaSR-mediated activation of mTOR complex 2 signalling leading to increased phosphorylation of AKT. Interestingly, this study highlights the importance of the scaffolding protein homer-1 as a key player in linking CaSR to mTORC2 signalling. Currently, our hypothesis built on our modelling studies is that ADMA and SDMA bind directly to the amino acid site of CaSR and alter activity. However, further studies will be necessary to fully understand whether ADMA alters CaSR activity by altering CaSR protein complexes.
Whilst ADMA competes with L-arginine for binding to the active site of NOS our data presented here indicates that arginine is unable to compete with ADMA for CaSR. The literature relating to L-arginine binding to CaSR is somewhat contradictory with some studies reporting no effect of basic amino acids on the receptor (Conigrave and Ward, 2013) while others report a strong stimulation of CaSR by L-arginine in the gut (Mace, Schindler and Patel, 2012). In our hands, using stably transfected HEK293 cells, L-arginine was unable to directly activate CaSR or modulate the effect of ADMA on the receptor. However, these experiments were performed in a setting without calcium present in the buffer, and previous studies have shown that L-amino acids in general have a greater effect on CaSR activation at higher (2.5 mM) Ca 2+ concentrations (Conigrave et al., 2000). L-arginine in particular, seems sensitive to the prevailing Ca 2+ concentration only showing a low potency to activate CaSR at around 2mM Ca 2+ and therefore, the absence of calcium here may mask the full effect of L-arginine on CaSR activity (Conigrave et al., 2004). A full series of experiments exploring L-arginine and ADMA interacting at the CaSR will need to be performed at a range of Ca 2+ , L-arginine and ADMA concentrations. ADMA increased CaSR sensitivity to Gd 3+ and Ca 2+ in its physiological range (low µM) suggesting greater affinity for CaSR than the known CaSR modulators L-phenylalanine and L-tryptophan both of which have been shown to act in the mM range (Zhang et al., 2016). CaSR is a member of the family C amino acid-sensing GPCRs which are conserved in many species. The most closely related family C member to CaSR is GPRC6a which is activated directly by amino acids (particularly L-ornithine and L-arginine) and is modulated by Ca 2+ .
This receptor has also been suggested to interact with methylarginines in the absence of Ca 2+ albeit at supra-physiological concentrations (Christiansen et al., 2006) perhaps suggesting that ADMA-receptor signalling may be more widespread than CaSR alone. This proposition is supported by studies that have demonstrated that asymmetrically monomethylated arginine (L-NMMA) can inhibit canavanine sensing by the Drosophila DmXR, a family C receptor related to mGluR (Mitri et al., 2009), suggesting that methylated arginine signalling via GPCRs may be an ancient and conserved mechanism.
Our identification of an ADMA-CaSR signalling pathway that is sensitive to small changes in ADMA concentration in the micromolar range and is independent of prevailing arginine concentrations provides a potential mechanistic explanation of the association between ADMA and cardiovascular and metabolic risk, therefore offering novel therapeutic opportunities to mitigate the effects of elevated ADMA. Further studies will be required to fully understand the significance of ADMA signalling via GPCRs and the potential for this pathway to link catabolic and metabolic cellular processes via regulation of mTOR expression and activity.
JL is a founder, director and shareholder in Critical Pressure Ltd. Critical Pressure owns patents relating to small molecule inhibitors of enzymes that metabolise ADMA (dimethylarginine dimethylaminohydrolases, DDAH).

Supplementary Materials
This report includes additional supplementary figures 1-7.  Analysis was by one-way ANOVA followed by multiple comparisons test (Bonferroni) in c, e, h, j, k and n; a two-tailed students t-test in a, b, l and m, and a two-way ANOVA in f. * P<0.05, **P<0.01, ***P<0.001. Data presented are mean ± S.E.M.