Renal tubular arginase‐2 participates in the formation of the corticomedullary urea gradient and attenuates kidney damage in ischemia‐reperfusion injury in mice

Arginase 2 (ARG2) is a mitochondrial enzyme that catalyses hydrolysis of l‐arginine into urea and l‐ornithine. In the kidney, ARG2 is localized to the S3 segment of the proximal tubule. It has been shown that expression and activity of this enzyme are upregulated in a variety of renal pathologies, including ischemia‐reperfusion (IR) injury. However, the (patho)physiological role of ARG2 in the renal tubule remains largely unknown.

V2 type vasopressin receptor (Avpr2) mRNAs were not different between control and cKO mice that received water ad libitum, but were lower in kidneys of water-deprived cKO mice as compared to water-deprived control mice (Supplementary Figure 1A). However, AQP2 protein expression was similar under water deprivation in both genotypes (Supplementary Figure 1B).
These results differ significantly from those published for Arg2-null mice 20 .
Role of ARG2 in the early phase of renal injury triggered by IR: 24-hour reperfusion. The role of ARG2 in the early events in IR-induced renal injury was studied 24 hours after unilateral IR injury (UIRI) or sham surgery. As shown in Figure 2A, Arg2 mRNA expression was significantly increased in the left ischemic kidney of control mice as compared to the contralateral (right) kidney and to kidneys from sham-operated animals. Arg2 expression was also increased in the left ischemic kidney from cKO mice ( Figure 2B). However, quantitatively, Arg2 levels in kidneys of cKO mice remained negligible compared to those of control mice.
Immunohistochemical analysis revealed dramatically increased ARG2 expression in the medulla of the left ischemic kidney in control mice, as compared to the right kidney ( Figure   2C). Histological scoring of Periodic Acid-Schiff (PAS)-stained kidney sections from control and cKO mice ( Figures 3A and 3B, respectively) demonstrated significantly greater tubular damage in the medulla of the left ischemic kidney from cKO mice ( Figure 3C). TUNEL assay showed a diffuse staining of damaged tubules in kidneys of UIRI-subjected control and cKO mice ( Figures 3D and 3E, respectively, and Supplementary Figure 2), characteristic of necrotic cell death. Quantitation of TUNEL staining revealed a similar trend towards higher cellular damage in UIRI-subjected cKO mice (p=0.058, Figure 3F). Expression levels of most tested early tissue AKI biomarkers (Kim-1, Il-6 and Tnf-a) were increased in left ischemic kidneys of both genotypes but not different between control and cKO mice (Supplementary Figure 3). However, expression of fibrinogen gamma chain (Fgg) was significantly higher in left ischemic kidneys of cKO mice, suggesting enhanced wound repair 23,24 (Supplementary Figure 3).
Analysis of urine samples collected in metabolic cages over 24 hours following UIRI revealed significantly greater functional impairments in cKO mice. As shown in Table 1, cKO mice exhibited albuminuria, kaliuria, calciuria, magnesuria, polyuria, increased urinary urea excretion and generalized aminoaciduria (Supplementary Figure 4). Analysis of blood collected 24 hours after sham or UIRI surgery showed no significant difference in plasma creatinine levels ( Table 1). Urea concentration was slightly but significantly higher in UIRI-subjected control mice, thus correlating with increased ARG2 expression (Table 1 and, see below).
Sham-operated mice were also used to test the possible role of renal ARG2 in the formation of corticomedullary urea gradient. To this aim, we measured urea content in the cortex, outer medulla and inner medulla of right and left kidneys from sham-operated control and cKO mice. As shown in Figure 4A, urea content was significantly lower in the outer and inner medulla of both right and left kidneys from sham-operated cKO mice as compared to control mice (n.b. -p-value for comparison of urea content in the outer medulla of the right kidney of sham-operated control and cKO mice = 0.054). Thus, these data support the idea that the S3-located ARG2 participates to the formation of urea concentration gradient along the corticomedullary axis. The lower medullary urea content in cKO mice was associated with statistically significant but quantitatively modest decrease in the total osmolyte content of the outer medulla in both right and left kidneys from sham-operated cKO mice as compared to control mice ( Figure 4B; n.b. -the total osmolyte content in the inner medulla could not be measured due to the small sample size).
UIRI caused a dramatic decrease in urea content in left ischemic kidneys from both genotypes ( Figure 4C). However, urea and the total osmolyte contents in the outer and inner medulla were not different in left kidneys of UIRI-subjected control and cKO mice ( Figure 4D).
These results suggest that polyuria observed in cKO mice subjected to UIRI (Table 1) is caused by osmotic diuresis and/or results from a dysfunction in the tubular mechanism of water reabsorption rather than from a difference in corticomedullary osmotic gradients in kidneys of control and cKO mice.
Role of ARG2 in long-term consequences of UIRI: 14 days reperfusion. It has been shown that in the model of UIRI with an intact contralateral kidney the ischemic kidney exhibit longterm progression towards chronic kidney disease (CKD) 25,26 . Fourteen days after UIRI the cKO mice had significantly lower body weight compared to UIRI-subjected controls ( Figure 5A).  Figure 5D, in the left ischemic kidney of control mice there was a strong positive correlation between Arg2 mRNA expression and expression levels of several markers of interstitial fibrosis (Vim, Fn1, Col3a1 and Col1a1). However, expression levels of these fibrosis markers and fibrosis score assessed by Masson's trichrome staining were not different between the two genotypes (Supplementary Figure 6A and 6B, respectively).
Analysis of urine samples showed that UIRI-subjected mice of both genotypes did not develop proteinuria and did not display major differences in urine volume and solute excretion rates (Supplementary Table 3). As shown is Supplementary Table 3, there was no difference in inulin clearance between UIRI-subjected cKO and control mice. Analysis of plasma revealed that UIRI-subjected cKO mice exhibit hypernatremia and significantly increased levels of symmetric dimethylarginine (SDMA) and asymmetric dimethylarginine (ADMA), two early markers of kidney disease progression (Figure 5E and Supplementary Table 4) 27,28 .
Compared to UIRI-subjected cKO mice, UIRI-subjected control mice had both higher plasma urea levels and lower arginine levels (Supplementary Table 3). These findings together with results showing no difference in inulin clearance between mice of the two genotypes (see above) suggest that higher plasma urea levels in control mice may result from increased ARG2 activity in the left ischemic kidney. Because ornithine generated through arginase activity can be further metabolized to polyamines (putrescine, spermidine and spermine) and proline or, used for energy production by replenishing TCA cycle intermediates 11,12 , we examined kidney tissue levels of these compounds. As shown is Supplementary Figure 7, kidney tissue putrescine and spermine contents were significantly decreased in left ischemic kidneys of both genotypes.
However, there was no difference in tissue levels of putrescine, spermidine, spermine and proline in left ischemic kidneys of control and cKO mice. These results prompted us to investigate possible alterations in kidney energy metabolism induced by UIRI. First, we evaluated the effect of UIRI on mitochondrial DNA content. As shown is Supplementary Figure   8, the ratio between the DNA levels of mitochondrial gene NADH dehydrogenase (mt-ND1) and nuclear cyclophilin A gene (Ppia) was markedly reduced in ischemic kidneys, most probably reflecting fibrosis. However, this reduction was similar in both genotypes. By using Seahorse analysis, the degree of coupling between the electron transport chain (ETC) and the oxidative phosphorylation machinery (OXPHOS) was examined on mitochondria isolated from kidneys of control and cKO mice (two way repeated measures ANOVA (Genotype X Left vs   6). However, a significant interaction effect in state III respiration suggested a greater impairment in the OXPHOS machinery in the ischemic kidney of cKO mice (interaction effect p=0.015, Figure 6C).

DISCUSSION
Arginase-2 is an extensively studied enzyme involved in a wide variety of biological processes and pathophysiological conditions. Historically, arginase was among the first enzymes that were (partially) purified and biochemically characterized in the kidney 29,30 . Since then, several strategies have been developed to study the role of this enzyme in the kidney, mainly based on the use of mice with whole-body inactivation of the Arg2 gene (Arg2-null mice) or, on the pharmacological inhibition of arginase activity. Here, we addressed this issue in a new mouse model in which Arg2 was deleted specifically in the renal tubule. This new model allowed us to directly test the hypothesis raised in several studies [9][10][11]19 that urea formed by ARG2 might contribute to the formation of the corticomedullary urea and osmolality gradients. The present study supports this hypothesis, thereby providing new insight into the mechanism of urine formation. Our results, however, differ from those of Huang and colleagues who found in Arg2null mice that ARG2 negatively regulates AQP2 expression and water reabsorption in the kidney. In our model, Aqp2 mRNA expression was reduced in kidney of cKO mice subjected to water deprivation and, AQP2 protein expression and urine conservation ability remained unchanged. We hypothesize that this discrepancy may be due to activity of ARG2 in extra-renal tissues.
Under normal conditions, no changes in plasma levels of L-arginine and urea were observed in cKO mice, suggesting that ARG2 in the kidney accounts for only a small fraction of L-arginine ureahydrolase activity in the body. However, 24 hours after UIRI plasma urea levels were significantly increased in control mice without concomitant elevation in plasma creatinine levels. Fourteen days after UIRI, higher plasma urea levels in control mice were inversely correlated with lower arginine concentration compared to cKO mice, with no difference observed in plasma creatinine levels between the two genotypes. These results suggest that UIRI-induced upregulation of ARG2 activity in the injured kidney leads to a dramatic increase in plasma urea levels (~80% in the present study) that are not due to changes in GFR. Interestingly, similar findings have been recently described by Nikoaleva et al. in another model of renal stress, i.e. in the kidney lacking the circadian clock activity 21 . The authors demonstrated that mice devoid of the circadian clock specifically in the renal tubule exhibited a significant increase in ARG2 activity in the kidney, which reached ~25% of arginase activity in the liver. In parallel these mice showed a ~20% increase in plasma urea levels but no difference in GFR. These results may have important clinical implications because plasma urea concentration is one of the most commonly used endogenous markers of glomerular filtration. Our study demonstrates that ARG2, which is upregulated in virtually all kidney stress conditions tested so far, can significantly influence plasma urea levels, and that these changes may not reflect changes in GFR. A challenging task for future studies is to elucidate the mechanism(s) regulating Arg2 transcription. Since (i) the S3 segment of the proximal tubule is located in the renal medulla, a low oxygen environment and, (ii) the medullary oxygen tension further decreases in many types of AKI or CKD previously noted, it is plausible to hypothesize that the primary factor inducing Arg2 levels is hypoxia. However, renal Arg2 expression is also increased upon furosemide treatment (CA, GC and DF, unpublished results), a condition that increases medullary oxygen tension 31 . This suggests that a complex multifactorial mechanism is involved in Arg2 regulation.
Twenty-four hours after UIRI both histological and functional outcomes were significantly worse in cKO mice. The cKO mice exhibited significantly greater tubular damage in the ischemic kidney as compared to the ischemic kidney from control mice. Surprisingly, however, among tested mRNA markers of early injury 24 (Kim-1, Il-6, Tnf-a and Fgg), only Fgg mRNA expression was significantly higher in the ischemic kidney of cKO mice. This can be explained by the relatively poor correlation between the degree of AKI severity and expression levels of these mRNA markers, at least in the UIRI model 25 . Urine analysis revealed that cKO mice developed polyuria, albuminuria, aminoaciduria, kaliuria, calciuria, phosphaturia and magnesuria, a phenotype reminiscent of Fanconi syndrome, or generalized dysfunction of the proximal tubule. Fanconi syndrome is a complex renal disease characterized by impaired capacity of the proximal tubule to reabsorb and/or secrete water and solutes.
Importantly, while there are many different causes of Fanconi syndrome, they all give rise to a similar phenotype (see above). Thus, it has been proposed that impaired energy production could be a common mechanism responsible for abnormal tubular transport 32 . Accordingly, in addition to extensive tubular loss, altered metabolism in the proximal tubule may explain the Fanconi syndrome phenotype in the early phase of UIRI in cKO mice.
Two weeks after UIRI, the cKO mice had lower body weight but the overall renal function was not different between the two genotypes. However, significantly greater adaptive growth of the contralateral (right) kidney in cKO mice suggested more severe functional damage in the ischemic (left) kidney. In parallel, the UIRI-subjected cKO mice exhibited increased plasma levels of SDMA and ADMA, two early biomarkers of CKD progression 28,33 .
SDMA is a dimethylated form of arginine, which is released into the circulation upon protein degradation. It is thought that this molecule indirectly inhibits the nitric oxide synthase (NOS) by decreasing L-arginine availability in endothelial cells. ADMA has been reports to directly inhibit NOS isoforms 34 . It has been shown that ADMA accumulates in IR injured kidneys 35 and that plasma ADMA is a strong prognostic indicator of CKD. Importantly, cKO mice exhibited significantly impaired mitochondrial function in the ischemic kidney compared to kidneys from sham-operated mice and to UIRI-subjected controls. It is well established that mitochondria play a central role in IR-induced kidney damage 36 . A possible explanation for the worsening of mitochondrial function in the ischemic kidney of cKO mice may be provided by the metabolic role of ARG2-generated ornithine in the proximal tubule. Indeed, it has been shown that ARG2 in the kidney co-localizes with ornithine aminotransferase (OAT), an enzyme catalyzing the

Study approval
All experiments with animals were performed in accordance with the Swiss guidelines for animal care (authorisation #29111 (to D.F.)), which conform to the National Institutes of Health animal care guidelines. Histokit and stored at room temperature until image acquisition.

Polyamines levels in kidney tissue
Kidney tissue levels of putrescine, spermidine and spermine were determined by Metabolon® (Potsdam, Germany) using UPLC-MS/MS.
GFR GFR was measured on anesthetized animals with inulin-FITC as previously described 39 .
Inulin clearance was calculated using a two phase exponential decay curve model 40 .

Statistics
All data are expressed as mean ± SEM. Normality of distribution has been tested using Shapiro-Wilk test to ensure that data meet the requirements for parametric tests.