Design and discovery of a highly potent ultralong- acting GLP-1 and glucagon co-agonist for attenuating renal brosis by modulating mitochondrial function

Xianxing Jiang (  jiangxx5@mail.sysu.edu.cn ) Sun Yat-sen University https://orcid.org/0000-0002-7508-2368 Qian Zhao Sun Yat-Sen University Jiale Dong Sun Yat-Sen University Hui Chen Sun Yat-Sen University Huan Yu Sun Yat-Sen University Shuyin Ye Shenzhen Turier Biotech. Co., Ltd. Shuangjin Yu The First A liated Hospital of Sun Yat-Sen University Nazi Song Sun Yat-Sen University Hongjiao Xu Sun Yat-sen University Zhiteng Luo Sun Yat-Sen University Guodong Chen The First A liated Hospital of Sun Yat-Sen University Rui Wang Lanzhou University https://orcid.org/0000-0002-4719-9921


Introduction
Chronic kidney disease (CKD) characterized by persisting renal damage and /or loss of renal function is a severe disease with potentially high morbidity 1,2 . Chronic renal failure, as the end-stage renal disease, is manifested by glomerulosclerosis, tubulointerstitial brosis, and vascular sclerosis 3,4 . Unfortunately, the molecular mechanisms linking pathogenetic factors in kidney brosis are complicated and incompletely understood 5,6 . It is most likely a combination of oxidative stress 7-10 , in ammation [11][12][13] , epithelial to mesenchymal transition, apoptosis 14 , and extracellular matrix deposition 15 . Thus, studies aimed at exploring attractive therapeutic targets and developing corresponding approaches for the treatment of CKD are urgently needed.
The mitochondrion is a central organelle that is responsible for the cellular redox, energy homeostasis, and therefore a major source of intracellular oxidative stress 16 . Recognition is growing that mitochondrial dysfunction and damages are recognized as a primary factor to various CKDs through impaired ATP generation, increased mitochondrial DNA (mtDNA) damage, and accumulated ROS production [17][18][19][20] . More recently, it has been demonstrated that ameliorating mitochondrial dysfunction prevents the progression of renal diseases [21][22][23] . Hence, mitochondria have been identi ed as an attractive platform for drug targeting in restraining CKD progression.
The protective role of GLP-1R signaling in diabetic nephropathy attributes to its hypoglycemic effect and anti-oxidative protective effect on various tissues [24][25][26][27] . On the other hand, recent evidence indicates that glucagon has lots of effects via a corresponding GCGR like improving mitochondrial turnover and function 28 . At present, the application of GLP-1R/GCGR dual-target agonists in NASH 29,30 , obesity [31][32][33][34][35] and diabetes 36,37 has attracted extensive attention, such as Cotadutide (MEDI0382) 38 , HM-12525A 39 , Oxyntomodulin (OPK-88003, TT401), BI456906 in phase II clinical trials. However, the therapeutic effects of these co-agonists on renal brosis remain unknown. In the context of this study, the emerging evidence indicated that levels of GLP-1R and GCGR in the kidney of human and experimental mice with CKD were lower than in healthy controls. Furthermore, GLP-1R and GCGR knockdown mice exhibited enhanced hyperglycemia-induced mitochondrial dysfunction, in ammation and renal brosis. Different from previous studies that only focused on the role of GLP-1R in CKD, we rstly found that GCGR also played a key role in the progression of CKD. The purpose of the current study was to follow up on this nding and develop a highly potent GLP-1 and glucagon co-agonist and further investigate its role in the CKD treatment.
Current GLP-1R/GCGR dual-target agonists regimens are potent and well-tolerated, enabling sustained suppression of many chronic and life-threatening disease. However, chronic disease control is linked to regimen adherence affected by pain and erythema at the site of injection. Such limitations were the impetus towards the development of ultralong-acting peptide drugs available [40][41][42][43][44][45] . To such ends, semaglutide once weekly is a successful attempt [45][46][47] . Despite such improvements, there remains no de ned pathways to optimize prodrug hydrolysis and physicochemical features 48 . To this end, we rstly proposed the concept of "double-site recognition" via octadecanedioic (C18) and "GGSGSG" binding to albumin to optimize drug release. Through in vitro or in vivo screening, a high-potency, balanced coagonist 1907 for GLP-1R and GCGR was identi ed. Next, 1907-B with ultralong pharmacokinetic pro les was obtained by attaching an albumin-binding side chain derived from C18 and "GGSGSG" to the Lys10 of 1907 (~2-3 fold to long-acting semaglutide in SD rats or cynomolgus monkeys).
Finally, we get a GLP-1R/GCGR dual-target agonist with ultralong-acting half-life, 1907-B, which prevents renal injury and brosis via anti-oxidative effect targeting GLP-1R and promoting mitochondrial turnover targeting GCGR. Furthermore, 1907-B notably relieves in ammation via NFκB pathway and improves renal brosis via TGF-β-SMAD pathway. In addition, cAMP-PKA pathway, the common downstream of GLP-1R and GCGR, plays a mechanistic role during 1907-B treatment by promoting mitochondrial biogenesis and bioenergetics. To our knowledge, no precedent has yet revealed the role of GLP-1R/GCGR dual targets, especially of GCGR target on attenuating renal brosis, and the current study provides an effective and promising alternative treatment strategy against renal brosis and CKD.

Results
GLP-1R and GCGR expression is correlated with progressive renal disease. We rst detected the GLP-1R and GCGR levels of human kidney tissue in normal human or patients with diabetic nephropathy and other chronic kidney disease. The immuno uorescence staining results showed that GLP-1R was mainly expressed in the glomeruli and GCGR was expressed in whole kidney including renal tubules and glomeruli ( Figure 1a). Moreover, we also found GLP-1R and GCGR levels were markedly lower in the kidneys of individuals with kidney disease than healthy individuals. Generally, GLP-1R and GCGR are closely correlated with metabolic disease. Consistently, less GLP-1R and GCGR levels were detected in kidney from patients with DN, even if the Remuzzi score (assessed the degree of chronic changes in glomeruli, tubules, interstitium, and blood vessels) of kidneys from patients with DN was lower than other patients with CKD ( Figure 1b). The absence of GLP-1R and GCGR protein expression, as revealed by immuno uorescence analysis, was also found in diabetic nephropathy (DN) model in db/db mice and renal brosis model in modelunilateral ureteral obstruction (UUO) mice ( Figure 1c). Furthermore, qPCR analysis also con rmed that Glp1r and Gcgr mRNA expression were also lower in the kidney of db/db mice and UUO mice than in those of lean control group ( Figure 1d). Collectively, the marked downregulation of GLP-1R and GCGR levels observed in renal disease indicates that they may have roles in the development of CKD.
To de ne the importance of GLP-1R and GCGR in progressive renal disease, we chose AAV9, which is used to transduce GLP-1R or GCGR shRNA to kidney, to down-regulate the expression of GLP-1R and GCGR in db/db mice 49 . qPCR analysis indicated that a signi cant reduction of Glp1r or Gcgr expression in the kidney was seen in the db/db mice treated with AAV9-GLP-1R or GCGR shRNA ( Supplementary  Fig. 6d), together with much enhanced renal brosis ( Figure 1e) and signi cant increase in mRNA expression of brotic factors (Figure 1f). Increased in ammation and impaired mitochondrial function were also observed in AAV9-GLP-1R or GCGR shRNA-infected mice but not control mice ( Figure 1f). Thus, our ndings demonstrated that GLP-1R and GCGR are key regulators in the progress of renal disease and can be used as the potential targets for the treatment of CKD.
GLP-1R and GCGR dual-agonist design. GLP-1 and glucagon bind to related members of the same receptor superfamily and exert important effects on glucose homeostasis, insulin secretion, and energy regulation. Therefore, combining the activity of two regulatory hormones with complementary biological effects, maybe offer a favorable strategy to treat diabetic nephropathy and related chronic kidney disease 29 .
To obtain just such a compound, we have designed seventeen novel GLP-1/glucagon chimeric peptides 1901-1917. These chimeric peptides have been created through fusion of the key amino acid sequences in maintaining biological activity of GLP-1 and glucagon ( Supplementary Fig. 1a) 50 . Comparing to the sequence information between GLP-1 and glucagon, positions 2, 3, 10, 12, 16, 20 are noteworthy in maintaining glucagon activity and positions 15, 18 are important in improving GLP-1R potency ( Figure   2a). Considering that GLP-1 and glucagon commonly recognize their individual receptors through an αhelical conformation, we performed the amino acid substitutions in the middle of the peptides to stabilize the secondary structure and enhance binding a nities to related receptors. It has been identi ed that the salt bridges formed between 16-20 or 17-21 with speci c helix-favoring amino acids in pairs could enhance helix stability through constrained conformation (Figure 2b). Moreover, glucagon achieves selectivity to GCGR due to its C-terminal acid. To reduce speci city for GCGR and activate GLP-1R more effectively, these chimeric peptides are designed to end with C-terminal amide. Besides, the C-terminal decapeptide from exendin-4 was assembled in the C-terminus of chimeric peptides to further enhance GLP-1R e cacy.
Considering that N-terminal region of peptides is highly conserved, only modi cation occurs in the Ser2, which replaces with unnatural amino acid, D-Ser or 2-aminoisobutyric acid (Aib), to protect these peptides against the hydrolysis from DPPIV. Other kinds of alterations mainly occur in the middle and C-terminus of these peptides. In previous study, the fatty acid side chain has been reported to help these compounds to interact with the albumin and decrease their renal clearance. Based on the three-dimensional model of 1907/GLP-1R complex by homologous modeling (Figure 2c), we speculate that the introduction of fatty acid side chain at residue 10 or 14 of these chimeric peptides would probably not perturb receptor binding interaction 50 . Therefore, these 2 residues were used as modi cation sites for side chain to improve drug activity. In this study, high water solubility PEG spacer C18 was introduced to improve water solubility and prolong the half-lives of peptide analogues 51 .
Structural assessment and agonistic activity test of GLP-1R and GCGR dual-agonists. To study the structural conformation of these chimeric peptides, the circular dichroism (CD) spectra for each peptide was measured in an aqueous environment (PBS) and a membrane-mimicking environment (50% TFE). As revealed in Supplementary Fig. 1b, these peptides showed random coil structures in aqueous buffer while displaying the typical α-helical spectra in 50% TFE solution with two minimum peaks at 208 and 222 nm.
According to the Supplementary Table 1, GLP-1 analogues (liraglutide and semaglutide) and glucagon differ in primary as well as secondary structure and the differences in biological properties are likely a function of these secondary structural differences. The other hybrid peptides also differed appreciably in secondary structure in tri uoroethanol (TFE) concentrations of 50% (v/v). The formation of salt bridges in these peptides and the Aib or Ser in position 16 are very important to the helical content of these peptides.
The agonistic activities among these peptide analogs were determined in HEK293 cells stably expressing human GLP-1R or GCGR using a cAMP response element (CRE)-driven luciferase reporter (Supplementary  Table 2 GLP-1R and GCGR dual-agonist screening in vitro and in vivo. With these highly potent GLP-1 and glucagon co-agonist analogs in hand, we next explored the e cacy of combined GLP-1 and glucagon agonism in a single peptide. Hyperglycemia is a main performance of diabetes and affects many intracellular processes. Diabetic nephropathy characterized by glomerular mesangial expansion and tubular epithelial-mesenchymal transition (EMT) induced by hyperglycemia, could result in increased cell extracellular matrix synthesis and further renal brosis formation 52,53 . To evaluate the therapeutic effects of these hybrid peptides on progressive renal disease, high glucose-induced GMC (glomerular mesangial cells) model and mTEC (tubular epithelial cells) model were used in vitro studies (Supplementary Fig. 1c and 1d). Fibronectin (FN) is an extracellular matrix component and is present along glomerular basement membranes. During glomerular injury, the deposition of FN would increase. Thus, FN is used as a marker of diabetic glomerular injury. Western blotting assay con rmed that these analogues could signi cantly decrease the expression level of FN in high glucose induced GMC cells. Moreover, alpha smooth muscle actin (α-SMA) level from the kidney areas increases with ECM deposition, and this is a hint that indicates alteration in myo broblasts. Western blotting assay showed that part of analogues, especially peptides with balanced GLP-1R and GCGR agonism could obviously reduce the α-SMA level in the high glucose induced mTEC cells. These results were consistent with the previous ndings, suggesting that the impact of these peptides on glomeruli outcomes was mediated through GLP-1R and GCGR signaling in glomeruli, while other impact on tubules directly mediated through GCGR signaling in tubules.
Next, we performed glucose tolerance test in vivo at different time points after peptides treatment to assess peptides response to glucose and their long-acting potency. Liraglutide and semaglutide are used as the positive control groups because of their long half-life. As shown in Supplementary Fig. 2, blood glucose levels of these peptides treated mice were signi cantly lower than those of PBS treated control mice after administration 8 h and 12 h (Supplementary Fig. 2a and 2b). Among them, the glucose tolerance improvement effects of 1907 and 1917 were more obvious than other peptides analogues. Even after 1907 and 1917 treatment for 24 h, the blood glucose level still could be maintained in a proper range as same as semaglutide treatment, and their effects on glucose control were better than that of liraglutide ( Supplementary Fig. 2c). It was stated clearly that 1907 and 1917 had relatively great half-time and could be used for the further evaluation of therapeutic effect on progressive renal disease in vivo.
1907 and 1917 generate synergistic metabolic bene ts and ameliorate renal injury in db/db mice. The db/db mice were chosen in the present study because they could spontaneously develop detectable DN 54 . Before evaluating the therapeutic effects of 1907 and 1917 on DN, the optimal doses of 1907 and 1917 were explored between 80, 100 and 120 µg/kg, and 120 µg/kg performed best (Supplementary Fig. 3b-3d). Semaglutide was included in the assay with 120 µg/kg as an optimal dose (Supplementary Fig. 3a). 12-week-old male db/db mice suffered from 1907, 1917 or semaglutide treatment for 8 weeks ( Figure 2d), while db/m mice served as the nondiabetic controls. At 12 weeks of age, db/db mice became more obese than db/m mice. After 8 weeks of treatment, compared to those of the control db/db mice, 1907, 1917 and semaglutide-treated db/db mice showed obviously lower serum lipid (TC and TG) levels ( Supplementary Fig. 4e). H&E staining exhibited that hepatocyte injury due to lipotoxicity was observed in db/db mice; whereas, 1907, 1917 and semaglutide treatment markedly reduced the ballooning and hepatic steatosis in the mice liver ( Supplementary Fig. 4a). Moreover, 1907 signi cantly decreased the body weight, liver weight and epididymis fat weight via suppressing food intake, water intake and promoting lipolysis in liver compared to vehicle control (Figure 2e and Supplementary Fig. 4). In contrast, 1917 and semaglutide showed no obvious decline in obesity levels. In addition, levels of serum ALT and AST were remarkably reduced in the 1907 treated db/db mice, as compared to the other group, which indicated the excellent liver protection function ( Supplementary Fig. 4e). Collectively, these data suggested that 1907 treatment promoted lipid metabolism bene ts and protected mice against lipid dysfunction.
The db/db mice exhibited higher fasting blood glucose levels when compared to db/m mice during the 8week experiment (as shown in Figure 2e), whereas the 1907, 1917 and semaglutide-treated groups showed lower fasting glucose levels than vehicle control. In addition, 1907, 1917 and semaglutide treatment signi cantly reduced hyperglycemia-induced glucose intolerance and insulin resistance, as assessed by OGTT and ITT, and among them, 1907 improved the insulin resistance best (Figure 2f and 2g). Furthermore, compared with db/m mice, db/db mice obviously affected the islet topography and the number of insulins + cells. However, 1907, 1917 and semaglutide treatment could signi cantly improve islet and promote the secretion of insulin ( Supplementary Fig. 5c). These data suggested that 1907, 1917 and semaglutide promoted glucose metabolism bene ts and protected mice against the damage from hyperglycemia.
The db/db mice showed signi cantly increased serum creatinine (SCr) and blood urea nitrogen (BUN) compared with the control mice ( Figure 2h). Additions of 1907, 1917 and semaglutide signi cantly reduce BUN and SCr in the db/db mice. Microalbuminuria is considered the earliest clinical indicator in the early stage of DN 55 . As shown in Figure 2h, albuminuria was signi cantly decreased in the 1907, 1917 and semaglutide-treated diabetic db/db mice compared with db/db control mice. Meanwhile, urine glucose was also obviously decreased under the 1907, 1917 and semaglutide treatment. Furthermore, the renal histopathology analysis was performed and the results indicated that the db/db mice displayed kidney damage and severe in ammatory in ltration ( Supplementary Fig. 5a), while administration of 1907, 1917 and semaglutide resulted in a marked decrease in renal injury by changes in kidney sections stained with H&E ( Figure 2i). Compared with db/m mice, the db/db mice progressed with glomerular hypertrophy and mesangial matrix expansion as PAS results. 1907, 1917 and semaglutide treatment maintained the renal glomerular structure and reduced the glomerular mesangial matrix accumulation compared to the db/db mice. Moreover, accumulation of collagen bers was also reduced in the 1907, 1917 and semaglutide treatment groups compared to the db/db mice (Figure 2i and Supplementary   Fig. 5b). TGF-β1, as one of crucial mediators for excessive ECM deposition in DN, could give rise to glomerulosclerosis 56 . In this study, immunohistochemistry was performed to assess the level of TGF-β1 in glomeruli of db/db mice ( Supplementary Fig. 5b). A signi cant increase for the positive staining of TGF-β1 in the glomeruli was observed in the db/db mice compared to the normal control mice. Whereas, glomerular staining of TGF-β1 was e ciently abrogated in db/db mice treated with 1907, 1917 and semaglutide. Combined, these observations indicated that 1907, 1917 and semaglutide could decrease the collagen deposition, thereby ameliorating renal brosis.
A rationally designed 1907 analogue, 1907-B, exhibits an extended half-life in vivo. Based on above data, the analogue 1907 was nally picked up for further modi cation to obtain a highly potent ultralongacting GLP-1R and GCGR co-agonist with su cient bioavailability. In preliminary study, we found that GCGR played a vital role in the progression of CKD and appropriately increasing the a nity of peptides to GCGR may be more conducive to the treatment of CKD. Therefore, the C-terminal amide bond of 1907 was replaced with a C-terminal acid (Figure 3a). Moreover, our study revealed the fatty acid side chain, C18 helped the 1907 to interact with the albumin and prolonged its half-time to 24 h or even longer.
Considering that "GGSGSG" also could bind to the albumin, we hypothesized that strong and reversible binding to albumin by the combination of C18 and "GGSGSG" in the side chain of Lys at position 10 of 1907(as 1907-B), would mediate increased stability toward proteolysis and lead to 1907-B with desired properties (Figure 3b). As our strategy about enhancing albumin a nity to obtain protraction, 1907 or 1907-B binding to the human serum albumin (HSA) was evaluated in biosensor experiments with surface plasmon resonance (SPR). The sensorgrams showed that 1907-B exhibited >10-fold higher a nity binding to HSA (KD=2.54 µM) than 1907 (KD=26.5 µM) (Figure 3c). Consistent with above result, isothermal titration calorimetry (ITC) assay also identi ed that 1907-B could increase the binding a nity with ~4-fold compared with 1907 ( Figure 3d). Furthermore, to con rm the role of "GGSGSG" in increasing the binding a nity with HSA, tetraethylene glycol or "GGSGSG" was tested by ITC. The results exhibited that "GGSGSG" had high binding a nities with KD=866 µM, whereas tetraethylene glycol showed no HSA binding. Next, the changes of structural conformation and agonistic activities on GLP-1R or GCGR between 1907 and 1907-B were examined. The results indicated that the additional modi cation on 1907 could raise α-helical content, decrease the GLP-1R activity (~44-fold) and increase GCGR activity (~10fold) to promote more balanced GLP-1R/GCGR activity ratio (Figure 3e).
On the basis of the enhanced binding a nity between 1907-B and HSA, we need to identify if the halftime of 1907-B could be prolonged. After 1907-B treatment for 60 h, the OGTT results showed that 1907-B still obviously improved the glucose tolerance compared with saline group. In comparison, the 1907 and semaglutide robustly lowered glucose excursion after treatment for 48 h (Figure 3f). Next, we evaluated the PK properties of 1907-B and semaglutide by intravenous injection or subcutaneous injection in SD rats. According to the Figure 3g, following i.v. treatment, the half-lives (t 1/2 ) of 1907-B and semagoutide were ~16.8 h and ~8 h, accordingly. The observed AUCs from 0 to in nity (AUCinf) of 1907-B and semaglutide were ~8588 and ~6194 ng·h/mL, respectively. Following s.c. treatment, t 1/2 of 1907-B and semaglutide were ~20.5 h and ~7.95 h, accordingly. Meanwhile, we also assessed the PK properties of 1907-B in cynomolgus monkeys (Figure 3h). Following s.c. treatment, the t 1/2 of 1907-B was 84 h and the t 1/2 of reported semaglutide was 53 h. These data indicated that the t 1/2 of 1907-B was prolonged to ~2-3-fold than that of semaglutide, which has a dosing frequency of once weekly in humans. These results pointed toward 1907-B, a ultralong-acting peptide, having a potential possibility that can be administered once every two weeks or even less frequency.
Toxicological evaluation of 1907-B in SD rats. The toxicity level of 1907-B was also evaluated in SD rats with low dose (0.25 mg/kg), medium dose (0.75 mg/kg) and high dose (1.5 mg/kg). The results indicated that 1907-B reduced the body weight in a dose-dependent manner because of the GCG-signaling component of 1907-B (Supplementary Table 3-4). Besides, there was no signi cant abnormalities in general behavior, appetite, physical appearance and various hematology (Supplementary Table 5-8).
Meanwhile, pathological analysis showed that 1907-B did not cause obvious pathological changes on the organs (liver, spleen, lung, kidney, stomach, duodenum, uterus, ovary, testis and epididymis) of rats under high dose (Supplementary Fig. 8). Taken together, 1907-B has great safety and can be used for further research.
1907-B improves renal injury depend on GLP-1R and GCGR signaling in db/db mice. To directly examine the in vivo e cacy of ultralong-acting 1907-B, we performed a series of dose and administration frequency of 1907-B in db/db mice with diabetic nephropathy (60, 120, 240 µg/kg/day or 120 µg/kg/2 days). The blood glucose changes showed 1907-B dramatically decreased the blood glucose level in a dose dependent manner and even revealed better e cacy than semaglutide in db/db mice (120 and 240 µg/kg/day) (Figure 4a). Meanwhile, we also measured the HbA1c levels of each group after 1907-B treatment for 12 weeks. Consist with above results, 1907-B also could signi cantly decrease the HbA1c with dose dependency and 120 and 240 µg/kg/day groups showed better effect in reducing the HbA1c than semaglutide. Based on the principle that the lowest dose and the lowest frequency were given priority under ensuring the e cacy, we chose 1907-B with 120 µg/kg/2d, which is equivalent to the effect of semaglutide for the further study.
Metabolic and kidney outcomes of 1907-B administration were evaluated in vivo using db/db mice accompanied by diabetic nephropathy. Following 8 weeks of equivalent dosing (120 µg/kg) with 1907-B (dual GLP-1R/GCGR agonist) once every 2 days, semaglutide (GLP-1R agonist) daily or glucagon (GCGR agonist) daily, there was no difference for weight-lowering effects between the treatment groups and the vehicle group (Supplementary Fig. 6a). Fasting glucose level was decreased in db/db mice with 1907-B and semaglutide treatment compared with vehicle, in contrast, glucagon alone induced higher fasting glucose level (Supplementary Fig. 6b). Meanwhile, the HbA1c level was obviously decrease in 1907-B treatment, but not in semaglutide and glucagon treated db/db mice. Interestingly, although treatment with the glucagon resulted in much higher blood glucose, it still exhibited the similar effect on reducing BUN and SCr as 1907-B and semaglutide (Figure 4b). This effect was also identi ed in pathological results. As shown in Figure 4c, H&E, Masson and Sirius Red staining showed that db/db mice exhibited more obvious glomerulosclerosis and tubulointerstitial brosis than the control group. However, 1907-B, semaglutide and glucagon treatment could signi cantly decrease glomerulosclerosis and tubulointerstitial brosis in the db/db mice. To systematically explore the effects of 1907-B, semaglutide and glucagon on DN, the RNA-seq analysis were performed. Gene Set Enrichment Analysis (GSEA) pathway enrichment results indicated that enriched pathways involved in brosis and in ammation were downregulated and enriched pathway involved in mitochondrial function was upregulated by 1907-B, semaglutide and glucagon treatment compared with the control db/db mice (Figure 4e and Supplementary Fig. 6c). A heatmap based on the GSEA results indicated that genes involved in brosis and in ammation were downregulated and genes involved in mitochondrial function were upregulated by 1907-B, semaglutide and glucagon treatment compared with the control db/db mice (Figure 4d) 1907-B attenuates renal brosis mainly depend on GCGR signaling in UUO mice. The UUO method, most commonly of the left one, has been widely applied to establish animal models with obstructive nephropathy via the ureter ligation ( Supplementary Fig. 6f) 57 . H&E, Masson and Sirius Red staining were performed to assess kidney injury and brosis. Histologically, compared with sham group, UUO group is characterized by tubular dilation and atrophy, renal parenchyma loss, in ammatory cells in ltration, and extracellular matrix (ECM) accumulation. In order to obtain the optimal e ciency, we performed a series of dose and administration frequency of 1907-B in UUO mice (60, 120, 240 µg/kg/day or 120 µg/kg/2 days). Similarly, the pathological results indicated that 1907-B dramatically reduced renal in ammation and brosis in UUO mice in a dose dependent manner and even exhibited better e cacy than semaglutide (Figure 5a). Based on these results, we chose the 120 µg/kg/2days for the further study.
The effect of 1907-B treatment on kidney brosis improvement were assessed in vivo using UUO mice accompanied by renal brosis. Following 14 days of equivalent dosing (120 µg/kg) with 1907-B once every 2 days, semaglutide daily or glucagon daily, H&E, Masson and Sirius Red staining revealed that glucagon treatment exhibited markedly fewer tubular injuries and ECM accumulation than 1907-B and semaglutide treatment (Figure 5b). To further systematically investigate how 1907-B, semaglutide and glucagon suppress UUO-induced renal brosis, we performed RNA-seq analysis in the kidney from sham and UUO-treated mice. GSEA systematically clari ed that cellular signaling pathways involved in brosis, in ammation and mitochondrial function were enriched ( Supplementary Fig. 6h). Similarly, a heatmap based on GSEA revealed that the expression of renal genes involved in brosis and in ammation was signi cantly downregulated and the expression of renal genes involved in mitochondrial function was upregulated by 1907-B and glucagon in UUO mice, compared with semaglutide treatment (Figure 5c). Subsequently, the data in combination with the transcriptomes of DN and UUO mice were analyzed, which demonstrated that nine signaling pathways were consistently in uenced (Figure 5d), and the in ammation and mitochondrial function pathways were overlapped (Figure 5e).
To further con rm the role of GLP-1R and GCGR during the 1907-B treatment for renal brosis, we generated AAV9-GLP-1R or GCGR shRNA treated UUO mice (Supplementary Fig. 6g) Next, we used oxygen-consumption rate (OCR) assay to evaluate mitochondrial respiratory function. GMC and mTEC cells exposing in medium contain high glucose or TGF-β displayed notably reduced basal and maximal respiration rates compared with that in low glucoses medium (Figure 8c-f). Treatment with 1907-B restored basal and maximal respiratory rates in high glucose or TGF-β induced GMC and mTEC cells. Moreover, maximal respiratory capacity, and ATP turnover were found to be signi cantly reduced in high glucose or TGF-β-treated cells compared with the control group. However, 1907-B application improved the mitochondrial respiratory parameters in high glucose or TGF-β-treated cells. When PKA was inhibited, 1907-B was unable to increase OCR in GMC and mTEC cells, which indicated that the improved mitochondrial function was mediated by PKA signaling. These data demonstrated that 1907-B improved mitochondrial oxidative capacity directly via PKA signalling.

Discussion
The pathogenesis of CKD is complex. Many studies indicate that mitochondrial dysfunction, oxidative stress, in ammation, apoptosis and ECM deposition are implicated in the progression of renal diseases 60 . Therefore, strategies for CKD treatment should target a key factor(s) that is involved in prominent pathogenic pathways in this condition to ultimately inhibit multiple pathological features in a simultaneous manner 5,61 .
The present study, we identi ed GLP-1R and GCGR as key regulators of CKD, which were correlated with disease severity during progressive kidney disease in patients with CKD, as well as db/db mice or UUO mice. Notably, downregulation of GLP-1R and GCGR in the kidney of db/db mice signi cantly promoted more advanced diabetic nephropathy than wild-type mice, accompanied by increased renal brosis, in ammation and decreased mitochondrial function. Furthermore, immuno uorescence staining analyses demonstrated that GLP-1R was mainly expressed in the glomeruli and GCGR was expressed in whole kidney including renal tubules and glomeruli. Considering that the hypoglycemic effect and antioxidant effect of GLP-1 and the effect of glucagon on improving mitochondrial function, we speculate that targeting GLP-1R and GCGR could be considered as promising avenues for CKD treatment by preventing hyperglycemia and attenuating mitochondrial dysfunction.
On the basis of our ndings, pharmacological approaches targeting GLP-1R and GCGR should ideally ameliorate mitochondrial function and block progression of renal injury. At present, the very large investment is put into the development of GLP-1R/GCGR co-agonists, some of them has been applied for the treatment of NASH, obesity and diabetes successfully. However, the effect of GLP-1R/GCGR coagonists on CKD treatment is still unknown 62 . In the present study, we have engineered GLP-1R/GCGR coagonist analogues that ful ll all the design criteria. Among them, we have set up to engineer a safe, highe ciency and ultralong-acting basal candidate, 1907-B (proteolytic stability, strong and reversible albumin binding, strong GLP-1/GCG receptor a nity). Stability toward proteolytic degradation was achieved by connecting C18 and "GGSGSG" with Lys in position 10 of 1907-B. Pharmacokinetic results in vivo showed that the half-life of 1907-B was prolonged ~2-3-fold than semaglutide, which makes it available to administer twice a week or even less frequency.
Next, the improvement effect of 1907-B in DN was assessed in db/db mice. Indeed, 1907-B could signi cantly decrease the high blood glucose and ameliorate the renal injury. Interestingly, we found the GLP-1R-speci c agonist semaglutide and GCGR-speci c agonist glucagon also effectively inhibit renal damage, respectively. According to the study from Yamada et al. 24 Figure 9. From a clinical perspective, our study highly implicates the future development of a highly potent ultra-long acting and safe GLP-1R/GCGR co-agonist, 1907-B that disrupt the process of progressive renal disease as promising avenues for CKD therapy. Excitingly, the accredited institutions are conducting the preclinical studies of 1907-B in diabetic nephropathy, and the relevant results will be disclosed in the future.  Figure 1 GLP-1R and GCGR expression is correlated with progressive renal disease. a, GLP-1R and GCGR immuno uorescence staining of kidneys from patients with diabetic nephropathy or other chronic kidney disease. b, H&E, PAS staining and Remuzzi score of kidneys from patients with diabetic nephropathy or other chronic kidney disease. c, GLP-1R and GCGR immuno uorescence staining of kidneys from db/db or UUO mice. d, The mRNA levels of Glp1r and Gcgr expression in kidneys from db/db or UUO mice. e, GSEA of pathways related to brosis, in ammation and mitochondrial function in db/db mice treated with AAV9-CON, AAV9-GLP1R-RNAi or AAV9-GCGR-RNAi. f, Heatmap of brosis-, in ammation-and mitochondrial function-related gene expression pro les based on the RNA-Seq dataset. Here and later, unless otherwise indicated, for each assay, three separated experiments were performed. And for each experiment (n = 3 cell samples/group or n = 3-6 mice/group). All the above data are presented as mean values ± SEM using unpaired Student's t test. Source data are provided as a Source Data le.     Here and later, unless otherwise indicated, the quanti cations of protein expression level were performed using three independent western blotting experiments. c, the mRNA level of Serpine1, Desmin, Mmp3, Mmp8, and Mmp9 were measured in the kidney samples of the indicated groups. d, The protein expression level of TGF-β1 in GMC and mTEC cells of the indicated groups. e, GSEA pathway enrichment analysis of pathways related to in ammation. f, the mRNA level of Nos2, F4/80, Il1b , Il6, Ccl2 and Ccl4 were measured in the kidney samples of the indicated groups. g, Representative images of immunohistochemistry staining of CD68 on kidneys from db/db mice at indicated groups and statistical positive area, respectively. h, The expression level of key protein CD68, TNFα, IL-1β, p-IKBα/IKBα and p-P65/P65 in the kidney of the indicated groups. i, The protein expression level of p-IKBα/IKBα and p-P65/P65 in GMC and mTEC cells of the indicated groups.   Summary of 1907-B's ultralong-acting design and its action mechanism at targets organs. Compared with semaglutide, 1907-B exhibits more prolonged half-time due to introduction of C18 and "GGSGSG" in the side chain of Lys at position 10 of 1907. The dual-agonist activity of 1907-B acts at the GLP-1R and GCGR in the kidney to ameliorate the diabetic nephropathy in db/db mice. Meanwhile, the GCGR agonist