Overactive cannabinoid 1 receptor in podocytes drivestype 2 diabetic nephropathyTony Jourdana,1, Gerg}o Szandaa, Avi Z. Rosenbergb,c, Joseph Tama, Brian James Earleya, Grzegorz Godlewskia,Resat Cinara, Ziyi Liua, Jie Liua, Cynthia Jud, Pál Pachere, and George Kunosa,1

aLaboratory of Physiologic Studies, Section on Neuroendocrinology, and eSection on Oxidative Stress and Tissue Injury, Laboratory of Physiologic Studies,National Institute on Alcohol Abuse and Alcoholism, bKidney Diseases Section, National Institute on Diabetes, Digestive, and Kidney Diseases, National Institutesof Health, Bethesda, MD 20892; cThe Johns Hopkins University School of Medicine, Baltimore, MD 21205; and dSkaggs School of Pharmacy and PharmaceuticalSciences, Department of Pharmaceutical Sciences, University of Colorado Anschutz Medical Campus, Aurora, CO 80045

Edited by Lutz Birnbaumer, National Institute of Environmental Health Sciences, Research Triangle Park, NC, and approved October 30, 2014 (received forreview October 17, 2014)

Diabetic nephropathy is a major cause of end-stage kidney disease,and overactivity of the endocannabinoid/cannabinoid 1 receptor(CB1R) system contributes to diabetes and its complications. Zuckerdiabetic fatty (ZDF) rats develop type 2 diabetic nephropathy withalbuminuria, reduced glomerular filtration, activation of the renin-angiotensin system (RAS), oxidative/nitrative stress, podocyte loss,and increased CB1R expression in glomeruli. Peripheral CB1R block-ade initiated in the prediabetic stage prevented these changes orreversed them when animals with fully developed diabetic ne-phropathy were treated. Treatment of diabetic ZDF rats with los-artan, an angiotensin II receptor-1 (Agtr1) antagonist, attenuatedthe development of nephropathy and down-regulated renal corti-cal CB1R expression, without affecting the marked hyperglycemia.In cultured human podocytes, CB1R and desmin gene expressionwere increased and podocin and nephrin content were decreasedby either the CB1R agonist arachydonoyl-2′-chloroethylamide, an-giotensin II, or high glucose, and the effects of all three wereantagonized by CB1R blockade or siRNA-mediated knockdown ofCNR1 (the cannabinoid type 1 receptor gene). We conclude that in-creased CB1R signaling in podocytes contributes to the develop-ment of diabetic nephropathy and represents a common pathwaythrough which both hyperglycemia and increased RAS activity ex-ert their deleterious effects, highlighting the therapeutic potentialof peripheral CB1R blockade.

endocannabinoid | podocyte | angiotensin II | nephropathy |hyperglycemia

Diabetic nephropathy, a highly prevalent and serious com-plication of both type 1 and type 2 diabetes mellitus and

a leading cause of renal failure, is characterized by albuminuria,decreased glomerular filtration rate (GFR), mesangial expan-sion, thickening of the glomerular basement membrane, andglomerular sclerosis (1). Multiple mechanisms have been impli-cated in the development of diabetic nephropathy, includingactivation of the renin-angiotensin system (RAS) (2), increase inoxidative (3) and nitrosative/nitrative stress (4), as well as anincrease in local inflammation (5).The endocannabinoid system plays a well-documented role in

obesity and its metabolic complications, including insulin re-sistance and type 2 diabetes (T2DM). Globally acting cannabi-noid 1 receptor (CB1R) antagonists/inverse agonists improveobesity-related insulin resistance, dyslipidemia, fatty liver, andβ-cell loss, and attenuate obesity-related inflammatory changesboth in preclinical models of diet-induced or genetic obesity andin clinical trials in overweight subjects with metabolic syndrome(reviewed in refs. 6 and 7). Global CB1R blockade also hasbeneficial effects in mouse models of type 1 and type 2 diabeticnephropathy (8–11). However, the therapeutic development ofthis class of compounds has been halted because of adverseneuropsychiatric side effects in a small proportion of treatedsubjects (12). Recent studies in rodent models have demon-strated that peripherally restricted CB1R antagonists are as

effective as globally acting ones in reversing obesity and itsmetabolic sequelae (13, 14), and also delay the onset of overtT2DM (15) without inducing any behaviors that are predictive ofadverse neuropsychiatric effects in humans (13, 14). Conse-quently, there is considerable interest in translating these find-ings to clinical practice by developing peripherally restrictedCB1R antagonists/inverse agonists for clinical use. In this studywe used the Zucker diabetic fatty (ZDF) rat, a well-establishedmodel of type 2 diabetic nephropathy (16, 17), to explore therole of peripheral CB1R in diabetic kidney disease, and thetherapeutic potential of its inhibition by a nonbrain-penetrantCB1R inverse agonist. We report that selective antagonism ofperipheral CB1R can both prevent the deterioration of kidneyfunction when treatment starts in the prediabetic stage and re-verse it when treatment is started after the full development ofdiabetic nephropathy. The results also indicate that increasedendocannabinoid/CB1R signaling in podocytes plays a key role inthe development of nephropathy, which can be dissociated fromthe hyperglycemia and is likely induced via the RAS.

ResultsCB1R Antagonism Prevents the Development of Nephropathy inPrediabetic ZDF Rats. Six-week-old male, prediabetic ZDF ratswere started on a daily oral dose (3 mg/kg) of the nonbrainpenetrant CB1R inverse agonist JD5037 or vehicle for 90 d. Asillustrated in Fig. 1A, vehicle-treated ZDF rats developed extreme

Significance

Diabetic nephropathy is the leading cause of chronic kidneydisease in the United States, and one of the most significantlong-term complications of both type 1 and type 2 diabetes,which currently lack fully effective therapy. Hyperglycemia andactivation of the renin-angiotensin system (RAS) are thought tobe the two main drivers of this pathology. We have recentlyshown that selective blockade of peripheral cannabinoidreceptor-1 (CB1R) delayed and attenuated the development oftype 2 diabetes in a rat model. Here we show that the ne-phropathy-inducing effects of both hyperglycemia and activa-tion of the RAS involve CB1R activation in glomerular podocytes,and that antagonism of peripheral CB1R could represent a novel,effective, and rational approach to both prevent and reversediabetic nephropathy.

Author contributions: T.J., P.P., and G.K. designed research; T.J., G.S., J.T., B.J.E., G.G., R.C.,Z.L., and J.L. performed research; C.J. contributed new reagents/analytic tools; T.J., A.Z.R.,and P.P. analyzed data; and T.J. and G.K. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence may be addressed. Email: jourdant@mail.nih.gov or george.kunos@nih.gov.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419901111/-/DCSupplemental.

E5420–E5428 | PNAS | Published online November 24, 2014 www.pnas.org/cgi/doi/10.1073/pnas.1419901111

Dow

nloa

ded

by g

uest

on

June

8, 2

020

hyperglycemia by the end of the treatment period, accompanied byincreased kidney weight, greatly elevated plasma creatinine andblood urea nitrogen levels, polydipsia, polyuria, a marked re-duction in GFR, as well as increased excretion of albumin, glucose,and uric acid, compared with lean controls. Chronic treatment ofZDF rats with JD5037 completely or nearly completely normalizedall these parameters (Fig. 1A). The same difference was evidentfor Cnr1 (the cannabinoid type 1 receptor gene) expression inthe glomerulus, its diabetes-related increase being fully re-versed by treatment with JD5037 (Fig. 1B). Body weight gainwas similar in vehicle-treated and JD5037-treated ZDF ratsduring the first 4 wk of treatment, as reported earlier (15), butas the vehicle-treated rats got sicker with extreme hyperglyce-mia, their body weight plateaued, whereas JD5037-treated ratscontinued to grow, resulting in significantly greater weight inthe latter by the end of the treatment period (Fig. 1). CB1Rimmunoreactivity was also detectable in proximal tubular cells(Fig. S1), in agreement with earlier findings in the mouse (14),and the specificity of immunostaining was evident in its absencein preparations from CB1R

−/− mice (Fig. S1).Vehicle-treated ZDF rats also displayed a marked reduction

in the number of podocytes compared with control lean ratsbased on podocin and Wilms’ tumor (WT-1) immunostaining aswell as podocalyxin/WT-1 double immunostaining, which waslargely prevented by JD5037 treatment (Fig. 2A). CB1R proteinwas tightly colocalized with podocalyxin in the perisomal regionof podocytes, indicating its expression primarily in podocytes(Fig. 2B). Consistent with podocyte loss, renal cortices of ZDFrats had much lower podocin mRNA (Nphs2) and protein (Fig.2A) and lower nephrin (Nphs1) and zonula occludens-1 (Tjp1)mRNA, but higher desmin (Des) mRNA than cortices of controlrats, and these changes were prevented by peripheral CB1Rantagonism (Fig. 2C).Histopathological findings in glomeruli and tubulointerstitium of

vehicle-treated ZDF rats were subtle and no vascular changes werenoted. Occasional glomeruli contained podocytes with prominentprotein resorption droplets (Fig. S2), although podocyte enlarge-ment was more commonly observed. No evidence of chronicity(glomerulosclerosis, interstitial fibrosis or tubular atrophy) wasobserved. Focal tubular injury in association with dilated tubularlumens containing proteinaceous material was noted. Many tubularepithelial cells contained prominent resorption droplets. In ratsreceiving JD5037, podocyte and tubular resorption droplets tendedto be smaller, with a lesser degree of coalescence and more limitedtubular epithelial simplification (Fig. S2). These latter findings areconsistent with the observed decrease in proteinuria with JD5037treatment (Fig. 1A). The limited histopathologic findings are notsurprising in light of the relatively early sampling point (19 wkof age) because the more prominent diabetic histomorphologicchanges were observed in older cohorts of ZDF rats (18).

Peripheral CB1R Antagonism Normalizes Increased Xanthine OxidaseActivity and Uric Acid Levels in ZDF Rat Kidney.Xanthine oxidase (orxanthine oxidoreductase, XO/XOR) is a potential source of re-active oxygen species (ROS) in diabetes (19), and is also involvedin the metabolism of uric acid, a product of excessive purine ca-tabolism that can lead to the development of gout or nephropa-thy. As shown in Fig. 3A, renal cortex XO activity and expressionwere markedly increased and accompanied by elevated serum andkidney uric acid levels in ZDF compared with controls rats, and allof these changes were prevented by JD5037 treatment (Fig. 3A).A potential off-target, direct effect of JD5037 on XO activitycould be excluded by the failure of 10–100 nM JD5037 to in-fluence the activity of recombinant XO, which was completely

Fig. 1. Peripheral CB1R antagonism prevents the development of diabetesand the associated nephropathy in ZDF rats. Blood and urine parameters (A)and CB1R protein (scale bar, 20 μm) and mRNA expression in the renal cortex(B) were analyzed in lean controls (open columns, body weight: 356.4 ±6.7 g), ZDF rats treated with vehicle (gray, BW: 370.3 ± 29.6 g) or 3 mg·kg·dJD5037 from age 6–18 wk (black, BW: 427.3 ± 24.2 g, P < 0.05 compared toboth lean control and vehicle-treated ZDF groups). Columns and bars repre-sent mean ± SEM (n = 20 rats per group). Significant difference from corre-

sponding value in lean (L) (*P < 0.05, **P > 0.01, ***P < 0.001) or vehicle-treated ZDF rats (V) (##P < 0.01, ###P < 0.001).

Jourdan et al. PNAS | Published online November 24, 2014 | E5421

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

June

8, 2

020

blocked by the known XO inhibitor allopurinol (Fig. 3B). Theconcentrations of JD5037 used were relevant to the in vivocondition, as peak plasma concentration of JD5037 following a3 mg/kg oral dose was 175 ± 40 nM (n = 3), of which >99% isprotein bound.

Adiponectin has anti-inflammatory effects and recent studiessuggest an inverse correlation between adiponectinemia, uric acidlevels, and renal function (20, 21). Accordingly, plasma adipo-nectin levels as well as the mRNA for adiponectin receptors 1 and2 in the kidneys were reduced in ZDF rats and normalized byJD5037 treatment (Fig. 3C).

Proinflammatory Cytokine Profile in the Diabetic Kidney Is Modulatedby Peripheral CB1R Blockade.As a further indication of inflammationin the diabetic kidney, the renal cortical expression of the proin-flammatory cytokines Tnf, Il-18, and Il-6, as well as the activity ofthe apoptotic enzyme caspase 3/7 were increased, and JD5037treatment attenuated all these changes (Fig. S3A). However,contrary to our earlier findings of robust proinflammatory

Fig. 2. Markers of podocyte health and injury in kidneys of lean and ZDF ratsand in ZDF rats treated with JD5037. (A) Podocytes identified in glomeruli oflean and vehicle- or JD5037-treated ZDF rats by WT-1 (scale bars, 50 μm) andpodocin immunohistochemistry and podocalyxin/WT-1 double immunohisto-chemistry. Analysis of podocin protein by Western blotting in renal corticesfrom the same animals. (B) Colocalization of podocalyxin with CB1R, as ana-lyzed by double immunohistochemistry. Specificity of CB1R labeling is in-dicated by its absence in CB1R

−/− mice. (C) Nphs1, Tjp1, and DesmRNA levels inrenal cortices from the same three groups of rats. Symbols as described in thelegend of Fig. 1, n = 20 rats per group. Significance of differences indicated asdescribed for Fig. 1 (#P < 0.05).

Fig. 3. Peripheral CB1R antagonism in ZDF rats prevents the increase in uricacid production and decrease in adiponectin and adiponectin receptor ex-pression. (A) XOR activity in kidney tissue and plasma and kidney concen-tration of uric acid in lean control rats (open columns) and ZDF rats treatedwith vehicle (gray columns) or JD5037 from age 6–19 (black columns). Im-munoreactive XOR protein expression and renal Xdh mRNA levels are alsoshown. (Scale bars, 100 μm.) (B) JD5037 does not directly inhibit XO activity.The activity of recombinant XO was measured in vitro in the presence ofvehicle, the XO inhibitor allopurinol (100 μM) or JD5037 (10 and 100 nM). (C)Plasma adiponectin and renal levels of Adipor1 and Adipor2 mRNA in thesame three groups of rats as in A. Columns represent means ± SEM from 20animals per group; Significant difference from corresponding value in lean(*P < 0.05, **P < 0.01, ***P < 0.001), or vehicle-treated ZDF rats (#P < 0.05,##P < 0.01, ###P < 0.001).

E5422 | www.pnas.org/cgi/doi/10.1073/pnas.1419901111 Jourdan et al.

Dow

nloa

ded

by g

uest

on

June

8, 2

020

macrophage infiltration in pancreatic islets of ZDF rats (15), therewas no evidence for macrophage infiltration or activation of theNlrp3 inflammasome in the kidney, as reflected in unchangedexpression of Cd68 and Nlrp3, unchanged caspase 1 activity, andIL-1b protein levels (Fig. S3B), and in the similar number ofCD68+ macrophages in the three treatment groups (Fig. S3C).

Peripheral CB1R Antagonism Prevents Activation of the Renin-Angiotensin System and Oxidative Stress in ZDF Rat Kidney. Com-pared with lean rats, vehicle-treated ZDF rats had increasedplasma aldosterone and angiotensin II levels, as well as increasedphosphorylation of the p65 subunit of NF-κB and increased an-giotensin II receptor-1 (Agtr1) mRNA in the renal cortex(Fig. 4A), suggesting up-regulation of the RAS, which is known

Fig. 4. Chronic JD5037 treatment prevents the up-regulation of the RAS in ZDFrats (A) and the parallel increase in the renal expression of ROS-generating NADPHisoforms as well as 4-hydroxynonenal (4HNE) and 3-nitrotyrosine (3NT) levels inkidney tissue (B). RAS activity is reflected by plasma aldosterone and angiotensin IIlevels, Agtr1 mRNA, and phospho–NF-κB (Ser536) levels in kidney. Treatmentgroups, symbols, and statistical analyses as described in the legend of Fig. 1 (#P <0.05). Columns and bars represent means ± SEM from 20 animals per group.

Fig. 5. Effects of JD5037 treatment on renal function and hyperglycemiain diabetic ZDF rats. ZDF rats were treated with daily oral doses of 3 mg/kgJD5037 or vehicle from age 15–19 wk. Blood glucose and plasma and urinemarkers of renal function were measured at the beginning and endof treatment in lean (open columns), vehicle-treated ZDF (gray columns),and JD5037-treated ZDF rats (black columns). Significant differencesfrom corresponding values in lean rats (***P < 0.005), from correspond-ing values in vehicle-treated ZDF rats (#P < 0.05, ###P < 0.005), orfrom corresponding pretreatment value is indicated as (&P < 0.05, &&P <0.01, &&&P < 0.005).

Jourdan et al. PNAS | Published online November 24, 2014 | E5423

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

June

8, 2

020

to increase oxidative stress through activation of NADPH oxi-dases (22). Indeed, renal cortices from vehicle-treated ZDF rats

displayed increased expression of ROS-generating NADPH oxi-dase isoforms, such as NADPH oxidase4 (Nox4), GP91Phox(Nox2), p47phox, and Nox1 (Fig. 4B). Moreover, ZDF kidneyscontained elevated 4-hydroxynonenal and 3-nitrotyrosine levelscompared with lean rats (Fig. 4B), suggesting increased oxidativeand nitrative stress. The development of all these changes wasmarkedly attenuated by peripheral CB1R blockade (Fig. 4).

Peripheral CB1R Antagonism Reverses Fully Developed DiabeticNephropathy Without Affecting the Associated Hyperglycemia. Totest whether peripheral CB1R blockade can reverse T2DM andthe associated nephropathy once it has developed, 15-wk-oldZDF rats were treated with 3 mg·kg·d JD5037 for 4 wk to matchthe age of ZDF rats undergoing preventive treatment at thetime of killing. The pronounced hyperglycemia of these ani-mals, which is a result of β-cell loss (15), remained unaffectedby JD5037 treatment, whereas their polyuria, albuminuria,uricosuria, increased plasma and urinary creatinine, and in-creased plasma levels of angiotensin II and aldosterone weredecreased and their reduced GFR was increased significantlybeyond pretreatment levels (Fig. 5). Additionally, plasma levelsof two hormones known to affect water reabsorption in the distaltubules were also altered: vasopressin levels were decreased andlevels of atrial natriuretic peptide increased in ZDF vs. controlrats and both changes were reversed by JD5037 treatment (Fig.S4). Similarly, increases in relative kidney weight and the renalexpression of Cnr1, Agtr1, Nox4, and Des and decreases in theexpression of Nphs1 and Nphs2 were attenuated by JD5037treatment of these animals (Fig. 6A). Interestingly, the proximaltubular marker megalin was also decreased in ZDF rats in aCB1R-dependent manner (Fig. 6A), which was similar to earlierfindings in mice with diet-induced obesity/insulin resistance (14),and suggested that CB1R-mediated changes in proximal tubularfunctions may contribute to the nephropathy of ZDF rats. Finally,podocyte loss was also reversed, as indicated by WT-1 immu-nostaining (Fig. 6B), a widely used marker of podocytes inrodents (23). Thus, 4-wk treatment of diabetic ZDF rats witha peripheral CB1R antagonist reversed the nephropathy, butnot the β-cell loss.Modest histopathological changes in sections from five ZDF

rats treated with vehicle from 15 to 19 wk of age (Fig. S5) weresimilar to those described above for the vehicle-treated controlsin the prevention paradigm. In five age-matched ZDF rats re-ceiving JD5037 during the same period, there were very milddecreases in protein droplet size and tubular injury with luminaldilation, and intraluminal protein was similarly very mildly de-creased (Fig. S5). Of note, this treatment effect is not surprisingbecause few chronic features (i.e., synechia, glomerulosclerosis,or significant interstitial fibrosis and tubular atrophy) havedeveloped at the point at which the cohort is sampled forhistopathological examination.

Angiotensin II Receptor Blockade Improves the Nephropathy of ZDFRats. Twelve-week-old male ZDF rats were started on daily oraltreatment (20 mg/kg) with the Agtr1 antagonist losartan or ve-hicle for 28 d. Relative to lean controls, both vehicle-treated andlosartan-treated ZDF rats developed extreme hyperglycemia bythe end of the treatment period (Fig. 7, Top Left). However,losartan treatment improved renal function as reflected in a re-duction in polyuria, albuminuria, blood urea nitrogen, plasmaand urine creatinine and preservation of a higher GFR (Fig. 7).As expected, the elevated circulating levels of angiotensin II inZDF rats were unaffected, whereas the elevated plasma aldo-sterone levels were normalized by losartan (Fig. 7, Bottom Left).Although losartan treatment did not reverse the increase inkidney weight, the increased expression of Agtr1 and Nox4, andthe decreased expression of Nphs2 and Npsh1 in the renal cortex

Fig. 6. Effect of JD5037 treatment on relative kidney weight and the renalexpression of genes implicated in podocyte loss (Cnr1, Agtr1, Nox4, Des) orsurvival (Nphs1, Nphs2) (A); on the presenceWT-1+ podocytes in renal glomeruli(scale bars, 20 μm) (B); on caspase 3/7 activity in kidney tissue (C); or in culturedhuman podocytes incubated with 30 mM glucose (D). Treatment groups andstatistical significance are indicated as described in the legend of Fig. 1 (#P <0.05), BW: 297 ± 6.7 g (lean); 401.9 ± 15.2 g (ZDF + V); 413.2 ± 15.0 g (ZDF + JD).

E5424 | www.pnas.org/cgi/doi/10.1073/pnas.1419901111 Jourdan et al.

Dow

nloa

ded

by g

uest

on

June

8, 2

020

were normalized, and the elevated level of Cnr1 mRNA wasreduced to levels below that in lean control rats (Fig. S6).

Effects of CB1R Activation in Cultured Human Podocytes. Podocytesplay a central role in maintaining normal glomerular functionand podocyte damage contributes to diabetic nephropathy (24).CB1R are present in podocytes (Fig. 2B), where their activationhas been shown to induce endoplasmic reticulum stress andapoptosis (25). Hyperglycemia and increased RAS activity arethought to be the two main pathogenic factors in diabetic ne-phropathy (24). First, we analyzed the potential interaction be-tween angiotensin II and the endocannabinoid/CB1R system inhuman podocytes. Exposure of the cells to 100 nM angiotensin IIresulted in a significant increase in the cellular content of bothanandamide and 2-arachidonoylglycerol, the two major endo-cannabinoids, and these effects were abrogated either by losartan(10 μM) or by JD5037 (100 nM) (Fig. 8A). Similar angiotensin IItreatment of podocytes resulted in up-regulation of the expres-sion of both AGTR1 and CNR1, and both losartan and JD5037prevented these changes. Furthermore, NPHS2 and NPHS1 ex-pression were significantly decreased, whereas DES expressionwas increased by angiotensin II, and these effects were alsoblocked by either losartan or JD5037, except for NPHS1, whereJD5037 was ineffective (Fig. 8B).We also examined the relationship between the effects of high

glucose and CB1R activation in cultured human podocytes. Ex-posure of the cells to either 30 mM glucose or 5 μM of the CB1Ragonist arachidonyl-2’-chloroethylamide (ACEA) increased theexpression of CNR1. Both stimuli decreased the expression ofNPHS2 and NPHS1 and increased the expression of DES andNOX4, and similar changes were observed at the protein ex-pression level for podocin and nephrin. Interestingly, all of theseeffects induced either by glucose or ACEA were completelyprevented by CB1R blockade by 10 nM JD5037 (Fig. S7A). Therole of CB1R in these changes was further indicated by the abilityof siRNA-mediated knockdown of podocyte CNR1 to attenuateor abolish the effects of both glucose and ACEA on the aboveparameters (Fig. S7B). ACEA activated the CB1R/Gi/o/cAMPpathway as indicated by its ability to inhibit forskolin-stimulatedcAMP accumulation in a CB1R-dependent manner (Fig. S8A),and this signaling pathway is involved in the cellular effects ofboth ACEA and high glucose, as indicated by the inhibition ofboth by pertussis toxin (Fig. S8B).

DiscussionThe development of nephropathy is a common and ominouscomplication of diabetes that reduces life expectancy and currentlylacks fully effective therapy. Hyperglycemia is thought to be one ofthe major drivers of the glomerulopathy, and treatments aimed toachieve tight glycemic control can slow the progression and at-tenuate renal impairment (24). However, the presence of end-stage kidney disease in diabetic patients with good glycemiccontrol suggests the role of additional pathogenic factors (26).An important conclusion of the present study is that renal

CB1R play a critical role in diabetic nephropathy, as indicated byboth in vivo and in vitro observations. Chronic in vivo treatmentof ZDF rats with a novel, nonbrain-penetrant CB1R inverseagonist for 3 mo completely prevented pathologies related toglomerular dysfunction, including albuminuria, reduced GFR,and elevated plasma creatinine. The parallel prevention of uri-cosuria and polyuria suggest additional effects of CB1R signalingon proximal and distal tubular functions, respectively, which is alsocompatible with the presence of CB1R on both the proximal anddistal convoluted tubules (27). Nephropathy in ZDF rats was alsoassociated with a marked up-regulation of CB1R in the diabetickidney, which was similarly reversed by chronic treatment with theCB1R inverse agonist. This finding suggests that increased sig-naling by renal CB1R plays a key role in the development of

Fig. 7. Chronic losartan treatment of ZDF rats improves renal functionwithout affecting hyperglycemia. ZDF rats were treated from age 12–16 wkwith daily oral doses of 20 mg/kg losartan (hatched columns) or vehicle (graycolumns). Age-matched lean rats were used as controls (open columns). Notethat losartan treatment did not affect blood glucose or plasma angiotensin IIlevels, but normalized the other parameters which are markers of ne-phropathy. Statistical significance as indicated for Fig. 1 (#P < 0.05).

Jourdan et al. PNAS | Published online November 24, 2014 | E5425

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

June

8, 2

020

diabetic nephropathy, and also indicates autoinduction of CB1Rexpression, as documented earlier in other tissues (28–30).

Earlier studies have documented the nephroprotective effectof globally acting CB1R inverse agonists (8–11), but therapeuticdevelopment of this class of compounds has been halted becauseof neuropsychiatric side effects (12). Peripherally restrictedCB1R antagonists, which in rodents are devoid of behavioraleffects secondary to blockade of CB1R in the CNS (13, 14), areunlikely to produce such side effects. Furthermore, JD5037 treat-ment started after the full development of diabetic nephropathyand continued for 4 wk improved the indicators of renal functionbeyond their level at the start of treatment. By reversing thehyperglycemia-induced increase in apoptotic mechanisms such ascaspase 3/7 activity, CB1R blockade could preserve the functionand prevent further loss of podocytes. Thus, peripheral CB1Rblockade can not only prevent but also reverse established ne-phropathy, further highlighting the therapeutic potential of suchcompounds in diabetic nephropathy, which is notoriously diffi-cult to treat (31).Another important finding in the present study is the disso-

ciation of nephropathy from the hyperglycemia, the former beingreversed and the latter remaining unaffected by peripheral CB1Rblockade (Fig. 6). The improvement of established glomerulop-athy, including albuminuria, reduced GFR, and elevated plasmacreatinine may be due to the observed reversal of podocyte loss bychronic CB1R blockade (Fig. 7B), whereas the reduction inpolyuria in the face of unchanged hyperglycemia is likely related toreversal of some distal tubular dysfunction affecting water reab-sorption. Indeed, both increased plasma atrial natriuretic peptide,which is similar to that found in another rat model of T2DM (32),and reduced vasopressin can contribute to water loss, and botheffects were reversed by JD5037 treatment (Fig. S4).The resistance of the hyperglycemia to CB1R blockade in the

same animals has interesting implications. First, it suggests thatdifferent pathologies drive these two manifestations of T2DM.Indeed, pancreatic islets of ZDF rats are heavily infiltrated by

Fig. 8. The effects of losartan or JD5037 treatment on angiotensin II-induced changes in cellular endocannabinoid levels and (A) CNR1, AGTR1,DES, NPHS2, and NPHS1-expession in cultured human podocytes (B). Cells wereincubated with vehicle, losartan (10 μM), or JD5037 (100 nM) for 1 h and thenexposed to 100 nM angiotensin II for 24 h. Columns represent means± SEM from12 independent replicates. Significant difference from corresponding value invehicle (*P < 0.05, **P < 0.01, ***P < 0.001), or angiotensin II-treated cells (#P <0.05, ##P < 0.01, ###P < 0.001).

Fig. 9. Schematic representation of the proposed CB1R-mediated podocyteloss in diabetic nephropathy. This figure was prepared using a template fromthe Servier Medical Art website.

E5426 | www.pnas.org/cgi/doi/10.1073/pnas.1419901111 Jourdan et al.

Dow

nloa

ded

by g

uest

on

June

8, 2

020

CD68+, proinflammatory macrophages that express the Nlrp3inflammasome and release active IL-1β and IL-18 (15), whereasthere was no evidence for macrophage infiltration in the kidney ofthe same rats, which displayed a proinflammatory cytokine profilevery different from that in β-cells (Fig. S3). Although others havedocumented macrophage infiltration of the kidneys of ZDF rats,this phenomenon and the associated nephrosclerosis were detectedin much older (>36 wk) animals (18), whereas ZDF rats of similarage as used here were reported to be devoid of either renalmacrophage infiltration or significant nephrosclerosis (33).A second implication of the dissociation of hyperglycemia

from diabetic nephropathy is that podocyte injury, which playsa key role in diabetic glomerulopathy (34), can be driven bymultiple factors. Indeed, our in vitro findings in primary culturedhuman podocytes suggest that overactivation of CB1R mayrepresent a common pathway mediating podocyte injury trig-gered by different stimuli. Activation of CB1R by ACEA in-creased the expression of genes linked to podocyte injury, suchas DES and NOX4, and decreased NPHS2 and NPHS1 expression,which are required for normal podocyte function (Fig. S7A).These changes, as well as the increased expression of CNR1 (Fig.S7A), are similar to those we have observed in the diabeticcompared with normal kidney. Exposure of podocytes to highglucose caused similar effects, which were similarly preventedeither by CB1R blockade or by siRNA-mediated knockdown ofCNR1 (Fig. S7B). Knockdown of podocyte CNR1 has beenreported to prevent high glucose-induced Akt and NF-kB acti-vation (25) and up-regulation of collagen and plasminogenactivator inhibitor-1 (PAI1) (9). Together, these observationsindicate the obligatory role of CB1R in mediating podocyte in-jury induced by high glucose. A role for CB1R in high glucose-induced apoptosis has also been documented in retinal pigmentepithelial cells (35). Also, a link between CB1R and diabeticnephropathy is suggested by the reported association betweena polymorphism in the CNR1 gene and nephropathy in patientswith T2DM (36).Because podocyte injury also develops under normoglycemic

conditions, factors other than glucose may also activate podocyteCB1R. A likely candidate is the RAS, the other major driver ofdiabetic nephropathy (37, 38), activation of which in ZDF rats isindicated by the elevated plasma angiotensin II and aldosteronelevels and the increased renal expression of Agtr1 (Fig. 4A). Theimportant role of the RAS in diabetic nephropathy is furthersupported by the existence of a positive feedback loop, wherebypodocyte injury promotes the filtration of liver-derived angio-tensinogen and the generation of angiotensin II in the kidney (39).Regardless of its renal or extrarenal origin, angiotensin II actingvia Agtr1 has been implicated in pronephrotic changes similar tothose detected in the kidney of diabetic ZDF rats. Thus, in ad-dition to the well-established link between angiotensin II and theNF-kB–dependent NOX activation and ROS production (40, 41),Agtr1 activation has been shown to reduce plasma adiponectin(42), down-regulate adipoR1 (43), increase the plasma uric acid(44) and IL-18 (45), increase the gene expression of Il-6 (46) andTNF-α (47), and increase caspase 3/7 activity in kidney tissue (48).The finding that all of the similar changes observed in ZDF

rats were prevented by peripheral CB1R blockade suggests thatangiotensin II acting via AGTR1 may be the glucose-independentsignal that induces the increased CB1R signaling in podocytes, asa cause of podocyte injury and diabetic glomerulopathy. Indeed,losartan treatment of ZDF rats resulted in much improved renalfunctions without any improvement in hyperglycemia. Re-markably, Cnr1 expression in the renal cortex of ZDF rats

was drastically reduced by losartan, supporting a direct linkbetween Agtr1 activation and Cnr1 expression. Our findings inhuman podocytes provide additional evidence for this: angio-tensin II treatment increased endocannabinoid levels andCNR1 expression in podocytes, and also promoted a “nephro-pathic” phenotype by decreasing NPHS2 and NPHS1 and in-creasing DES expression. More importantly, the obligatory roleof CB1R signaling in the effects of angiotensin II is indicated bytheir absence in cells preincubated with JD5037 (Fig. 8). Similarto the present findings, activation of Agtr1 induced CB1R-mediated effects in other tissues via increasing the synthesis ofthe endocannabinoid 2-arachidonoyl glycerol (49, 50), whereasCB1R blockade was found to down-regulate AGT1R (51). Thereis also evidence that CB1R involvement in angiotensin II-induced pathologies may reflect Agtr1/CB1R heteromeriza-tion, a process which amplifies Agtr1 signaling (52).In summary, the data presented indicate that enhanced CB1R

signaling in podocytes plays a key role in the pathogenesis ofdiabetic nephropathy, which likely represents a final commonpathway through which both hyperglycemia and increased RASactivity promote the multiple pathological changes, includingincreased oxidative/nitrosative stress, that culminate in thedevelopment of glomerulopathy, as schematically illustrated inFig. 9. Peripherally restricted CB1R antagonists could representa novel and rational therapeutic approach to treat this life-threatening complication of diabetes.

Materials and MethodsFor a detailed description of materials and methods used, see SI Materials andMethods. See Table S1 for a list of primers.

Animal Treatment Protocols. Animal protocols were approved by the In-stitutional Animal Care and Use Committee of the National Institute onAlcohol Abuse and Alcoholism, NIH. Male ZDF rats and their lean controlswere obtained from Charles River Laboratories. Rats were individually housedand maintained under a 12-h light/dark cycle and fed ad libitum a standardlaboratory diet (STD, NIH-31 rodent diet).

As a preventive treatment, 6-wk-old prediabetic ZDF rats were placed ona regimen of JD5037 (3 mg·kg·d) or vehicle [4% (vol/vol) DMSO + 1% Tween80 in saline] by oral gavage for 90 d (20 rats per group). To test whetherJD5037 treatment can reverse the pathology, 15-wk-old diabetic ZDF ratswere treated daily for 4 wk with the same dose of JD5037 or vehicle (10 ratsper group). Finally, 12-wk-old diabetic ZDF were treated orally with losartan(TCI America, L0232) for 28 d (20 mg·kg·d).

Peripheral CB1R Antagonist. JD5037 was synthesized and its pharmacologicalproperties analyzed as described previously (53).

Endocannabinoid Content. The endocannabinoid content of podocytes wasquantified by liquid-chromatography/tandem mass spectrometry, as describedpreviously (54).

Statistics. Values are expressed as means ± SEM. Data were analyzed byANOVA followed by the Tukey–Kramer post hoc test. Time-dependent varia-bles were analyzed and results in multiple groups were compared by ANOVAfollowed by Bonferroni test. (GraphPad Prism v6 for Windows). Significancewas set at P < 0.05.

ACKNOWLEDGMENTS. We thank Drs. J. F. McElroy and R. J. Chorvat (JenrinDiscovery) for providing the cannabinoid 1 receptor antagonist JD5037,Dr. J. Kopp (National Institute on Diabetes, Digestive, and Kidney Diseases,National Institutes of Health) for providing the human podocyte cell lineand for helpful comments on the manuscript, Ms. J. Harvey-White fortechnical assistance, Dr. R. Kechrid for assistance with the animal studies, andMr. Kris Ylaya for his help on histological preparation. This study wassupported by intramural funds from the National Institute on Alcohol Abuseand Alcoholism, National Institutes of Health.

1. Eknoyan G (2007) Obesity, diabetes, and chronic kidney disease. Curr Diab Rep 7(6):

449–453.2. Leehey DJ, Singh AK, Alavi N, Singh R (2000) Role of angiotensin II in diabetic ne-

phropathy. Kidney Int Suppl 77:S93–S98.

3. Gorin Y, Block K (2013) Nox as a target for diabetic complications. Clin Sci (Lond)

125(8):361–382.4. Szabo C (2009) Role of nitrosative stress in the pathogenesis of diabetic vascular

dysfunction. Br J Pharmacol 156(5):713–727.

Jourdan et al. PNAS | Published online November 24, 2014 | E5427

MED

ICALSC

IENCE

SPN

ASPL

US

Dow

nloa

ded

by g

uest

on

June

8, 2

020

5. Horváth B, Mukhopadhyay P, Haskó G, Pacher P (2012) The endocannabinoid systemand plant-derived cannabinoids in diabetes and diabetic complications. Am J Pathol180(2):432–442.

6. Pacher P, Kunos G (2013) Modulating the endocannabinoid system in human healthand disease—Successes and failures. FEBS J 280(9):1918–1943.

7. Engeli S (2012) Central and peripheral cannabinoid receptors as therapeutic targets inthe control of food intake and body weight. Handb Exp Pharmacol (209):357–381.

8. Barutta F, et al. (2010) Cannabinoid receptor 1 blockade ameliorates albuminuria inexperimental diabetic nephropathy. Diabetes 59(4):1046–1054.

9. Nam DH, et al. (2012) Blockade of cannabinoid receptor 1 improves insulin resistance,lipid metabolism, and diabetic nephropathy in db/db mice. Endocrinology 153(3):1387–1396.

10. Janiak P, et al. (2007) Blockade of cannabinoid CB1 receptors improves renal function,metabolic profile, and increased survival of obese Zucker rats. Kidney Int 72(11):1345–1357.

11. Rohrbach K, et al. (2012) Ibipinabant attenuates β-cell loss in male Zucker diabeticfatty rats independently of its effects on body weight. Diabetes Obes Metab 14(6):555–564.

12. Le Foll B, Gorelick DA, Goldberg SR (2009) The future of endocannabinoid-orientedclinical research after CB1 antagonists. Psychopharmacology (Berl) 205(1):171–174.

13. Tam J, et al. (2010) Peripheral CB1 cannabinoid receptor blockade improves car-diometabolic risk in mouse models of obesity. J Clin Invest 120(8):2953–2966.

14. Tam J, et al. (2012) Peripheral cannabinoid-1 receptor inverse agonism reduces obe-sity by reversing leptin resistance. Cell Metab 16(2):167–179.

15. Jourdan T, et al. (2013) Activation of the Nlrp3 inflammasome in infiltrating macro-phages by endocannabinoids mediates beta cell loss in type 2 diabetes. Nat Med19(9):1132–1140.

16. Hoshi S, et al. (2002) Podocyte injury promotes progressive nephropathy in zuckerdiabetic fatty rats. Lab Invest 82(1):25–35.

17. Coimbra TM, et al. (2000) Early events leading to renal injury in obese Zucker (fatty)rats with type II diabetes. Kidney Int 57(1):167–182.

18. Mizuno M, et al. (2006) The effect of angiotensin II receptor blockade on an end-stage renal failure model of type 2 diabetes. J Cardiovasc Pharmacol 48(4):135–142.

19. Pacher P, Nivorozhkin A, Szabó C (2006) Therapeutic effects of xanthine oxidase in-hibitors: Renaissance half a century after the discovery of allopurinol. Pharmacol Rev58(1):87–114.

20. Mangge H, et al. (2013) Uric acid best predicts metabolically unhealthy obesity withincreased cardiovascular risk in youth and adults. Obesity (Silver Spring) 21(1):E71–E77.

21. Abdullah AR, Hasan HA, Raigangar VL (2009) Analysis of the relationship of leptin,high-sensitivity C-reactive protein, adiponectin, insulin, and uric acid to metabolicsyndrome in lean, overweight, and obese young females. Metab Syndr Relat Disord7(1):17–22.

22. Lassègue B, Clempus RE (2003) Vascular NAD(P)H oxidases: Specific features, expres-sion, and regulation. Am J Physiol Regul Integr Comp Physiol 285(2):R277–R297.

23. Venkatareddy M, et al. (2014) Estimating podocyte number and density using a singlehistologic section. J Am Soc Nephrol 25(5):1118–1129.

24. Gruden G, Perin PC, Camussi G (2005) Insight on the pathogenesis of diabetic ne-phropathy from the study of podocyte and mesangial cell biology. Curr Diabetes Rev1(1):27–40.

25. Lim SK, Park SH (2012) The high glucose-induced stimulation of B1R and B2R ex-pression via CB(1)R activation is involved in rat podocyte apoptosis. Life Sci 91(19-20):895–906.

26. Krolewski M, Eggers PW, Warram JH (1996) Magnitude of end-stage renal disease inIDDM: A 35 year follow-up study. Kidney Int 50(6):2041–2046.

27. Larrinaga G, et al. (2010) Expression of cannabinoid receptors in human kidney. HistolHistopathol 25(9):1133–1138.

28. Mukhopadhyay B, et al. (2010) Transcriptional regulation of cannabinoid receptor-1expression in the liver by retinoic acid acting via retinoic acid receptor-gamma. J BiolChem 285(25):19002–19011.

29. Jourdan T, et al. (2010) CB1 antagonism exerts specific molecular effects on visceraland subcutaneous fat and reverses liver steatosis in diet-induced obese mice. Diabetes59(4):926–934.

30. Börner C, Höllt V, Sebald W, Kraus J (2007) Transcriptional regulation of the canna-binoid receptor type 1 gene in T cells by cannabinoids. J Leukoc Biol 81(1):336–343.

31. de Zeeuw D, et al.; BEACON Trial Investigators (2013) Bardoxolone methyl in type 2diabetes and stage 4 chronic kidney disease. N Engl J Med 369(26):2492–2503.

32. Vesely DL, et al. (1999) Elevated atrial natriuretic peptides and early renal failure intype 2 diabetic Goto-Kakizaki rats. Metabolism 48(6):771–778.

33. Suzaki Y, Ozawa Y, Kobori H (2007) Intrarenal oxidative stress and augmented an-giotensinogen are precedent to renal injury in Zucker diabetic fatty rats. Int J Biol Sci3(1):40–46.

34. Reiser J, Sever S (2013) Podocyte biology and pathogenesis of kidney disease. AnnuRev Med 64:357–366.

35. Lim SK, et al. (2012) Hyperglycemia induces apoptosis via CB1 activation through thedecrease of FAAH 1 in retinal pigment epithelial cells. J Cell Physiol 227(2):569–577.

36. Buraczynska M, Wacinski P, Zukowski P, Dragan M, Ksiazek A (2014) Common poly-morphism in the cannabinoid type 1 receptor gene (CNR1) is associated with micro-vascular complications in type 2 diabetes. J Diabetes Complications 28(1):35–39.

37. Komers R (2013) Renin inhibition in the treatment of diabetic kidney disease. Clin Sci(Lond) 124(9):553–566.

38. Kobori H, Kamiyama M, Harrison-Bernard LM, Navar LG (2013) Cardinal role of theintrarenal renin-angiotensin system in the pathogenesis of diabetic nephropathy.J Investig Med 61(2):256–264.

39. Matsusaka T, et al. (2014) Podocyte injury enhances filtration of liver-derived an-giotensinogen and renal angiotensin II generation. Kidney Int 85(5):1068–1077.

40. Gill PS, Wilcox CS (2006) NADPH oxidases in the kidney. Antioxid Redox Signal 8(9-10):1597–1607.

41. Campbell KN, Raij L, Mundel P (2011) Role of angiotensin II in the development ofnephropathy and podocytopathy of diabetes. Curr Diabetes Rev 7(1):3–7.

42. Ran J, et al. (2006) Angiotensin II infusion decreases plasma adiponectin level via itstype 1 receptor in rats: An implication for hypertension-related insulin resistance.Metabolism 55(4):478–488.

43. Li L, et al. (2013) Angiotensin II reduces cardiac AdipoR1 expression through AT1receptor/ROS/ERK1/2/c-Myc pathway. PLoS ONE 8(1):e49915.

44. Burnier M, Brunner HR (1996) Renal effects of angiotensin II receptor blockade andangiotensin-converting enzyme inhibition in healthy subjects. Exp Nephrol 4(Suppl 1):41–46.

45. Touyz RM, et al. (2007) Increased inflammatory biomarkers in hypertensive type 2diabetic patients: Improvement after angiotensin II type 1 receptor blockade. J AmSoc Hypertens 1(3):189–199.

46. Zhang W, et al. (2012) Interleukin 6 underlies angiotensin II-induced hypertensionand chronic renal damage. Hypertension 59(1):136–144.

47. Rosa AC, Rattazzi L, Miglio G, Collino M, Fantozzi R (2012) Angiotensin II inducestumor necrosis factor-α expression and release from cultured human podocytes. In-flamm Res 61(4):311–317.

48. Kagawa T, et al. (2008) Angiotensin II receptor blocker inhibits tumour necrosis fac-tor-alpha-induced cell damage in human renal proximal tubular epithelial cells. Ne-phrology (Carlton) 13(4):309–315.

49. Turu G, et al. (2009) Paracrine transactivation of the CB1 cannabinoid receptor by AT1angiotensin and other Gq/11 protein-coupled receptors. J Biol Chem 284(25):16914–16921.

50. Szekeres M, et al. (2012) Angiotensin II induces vascular endocannabinoid release,which attenuates its vasoconstrictor effect via CB1 cannabinoid receptors. J Biol Chem287(37):31540–31550.

51. Tiyerili V, et al. (2010) CB1 receptor inhibition leads to decreased vascular AT1 re-ceptor expression, inhibition of oxidative stress and improved endothelial function.Basic Res Cardiol 105(4):465–477.

52. Rozenfeld R, et al. (2011) AT1R-CB1R heteromerization reveals a new mechanism forthe pathogenic properties of angiotensin II. EMBO J 30(12):2350–2363.

53. Chorvat RJ, Berbaum J, Seriacki K, McElroy JF (2012) JD-5006 and JD-5037: Peripherallyrestricted (PR) cannabinoid-1 receptor blockers related to SLV-319 (Ibipinabant) asmetabolic disorder therapeutics devoid of CNS liabilities. Bioorg Med Chem Lett22(19):6173–6180.

54. Mukhopadhyay B, et al. (2011) Hyperactivation of anandamide synthesis and regu-lation of cell-cycle progression via cannabinoid type 1 (CB1) receptors in the re-generating liver. Proc Natl Acad Sci USA 108(15):6323–6328.

E5428 | www.pnas.org/cgi/doi/10.1073/pnas.1419901111 Jourdan et al.

Dow

nloa

ded

by g

uest

on

June

8, 2

020