LRG1 Promotes Diabetic Kidney Disease Progression by Enhancing TGF-B–Induced Angiogenesis

BASIC RESEARCH www.jasn.org

LRG1 Promotes Diabetic Kidney Disease Progression by Enhancing TGF-b–Induced Angiogenesis

Quan Hong,1,2 Lu Zhang,3,1 Jia Fu,1 Divya A. Verghese,1 Kinsuk Chauhan,1 Girish N. Nadkarni,1 Zhengzhe Li,1 Wenjun Ju,4 Matthias Kretzler ,4 Guang-Yan Cai,2 Xiang-Mei Chen,2 Vivette D. D’Agati,5 Steven G. Coca ,1 Detlef Schlondorff,1 John C. He,1,6 and Kyung Lee1

Due to the number of contributing authors, the afﬁliations are listed at the end of this article.

ABSTRACT Background Glomerular endothelial dysfunction and neoangiogenesis have long been implicated in the pathogenesis of diabetic kidney disease (DKD). However, the specific molecular pathways contributing to these processes in the early stages of DKD are not well understood. Our recent transcriptomic profiling of glomerular endothelial cells identified a number of proangiogenic genes that were upregulated in diabetic mice, including leucine-rich a-2-glycoprotein 1 (LRG1). LRG1 was previously shown to promote neovascularization in mouse models of ocular disease by potentiating endothelial TGF-b/activin receptor- like kinase 1 (ALK1) signaling. However, LRG1’s role in the kidney, particularly in the setting of DKD, has been unclear. Methods We analyzed expression of LRG1 mRNA in glomeruli of diabetic kidneys and assessed its localization by RNA in situ hybridization. We examined the effects of genetic ablation of Lrg1 on DKD progression in unilaterally nephrectomized, streptozotocin-induced diabetic mice at 12 and 20 weeks after diabetes induction. We also assessed whether plasma LRG1 was associated with renal outcome in patients with type 2 diabetes. Results LRG1 localized predominantly to glomerular endothelial cells, and its expression was elevated in the diabetic kidneys. LRG1 ablation markedly attenuated diabetes-induced glomerular angiogenesis, podocyte loss, and the development of diabetic glomerulopathy. These improvements were associated with reduced ALK1-Smad1/5/8 activation in glomeruli of diabetic mice. Moreover, increased plasma LRG1 was associated with worse renal outcome in patients with type 2 diabetes. Conclusions These findings identify LRG1 as a potential novel pathogenic mediator of diabetic glomerular neoangiogenesis and a risk factor in DKD progression.

J Am Soc Nephrol 30: ccc–ccc, 2019. doi: https://doi.org/10.1681/ASN.2018060599

Diabetic kidney disease (DKD) is a frequent compli- early stages of DKD for the design of novel preventive cation of diabetes mellitus (DM) and the most com- therapeutic strategies. mon cause of ESRD in the United States.1 Intensive glycemic control, BP control, and blockade of the renin-angiotensin-aldosterone system are the gold Received June 8, 2018. Accepted January 28, 2019. standard for current treatment of patients with Q.H. and L.Z. contributed equally to this work. 2–4 DKD. However, these established regimens pro- Published online ahead of print. Publication date available at 5 vide only partial therapeutic effects, indicating that www.jasn.org. pathogenic mechanisms driving DKD progression Correspondence: Dr. Kyung Lee or Dr. John C. He, Division of may not be targeted by current treatments. In addi- Nephrology, Mount Sinai School of Medicine, One Gustave L. tion, several recent clinical trials in DKD treatments Levy Place, New York, NY 10029. E-mail: [email protected] or were unsuccessful,5–8 highlighting an unmet need for [email protected] better understanding of mechanisms mediating the Copyright © 2019 by the American Society of Nephrology

J Am Soc Nephrol 30: ccc–ccc, 2019 ISSN : 1046-6673/3004-ccc 1 BASIC RESEARCH www.jasn.org

Glomerular neoangiogenesis has long been implicated as Significance Statement contributing to the morphology and pathophysiology of DKD,9–11 and dysregulated expression of various endothelial Although glomerular endothelial dysfunction and neoangiogenesis growth factors are implicated in DKD pathogenesis.12 Vascu- have long been implicated as factors contributing to diabetic kidney lar endothelial growth factor (VEGF) A is thought to contrib- disease (DKD) pathophysiology, the molecular basis of these processes is not well understood. The authors previously found fi 13 ute to the initial hyper ltration and microalbuminuria, such that a proangiogenic gene encoding leucine-rich a-2-glycoprotein 1 that blockade of VEGF signaling by pan-VEGF receptor in- (LRG1) was upregulated in isolated glomerular endothelial cells hibitor ameliorates diabetic albuminuria in mice.14 Imbalance from diabetic mice. In this work, they demonstrate in a diabetic of angiopoietin 1 (Ang-1) and angiopoietin 2 (Ang-2), an- mouse model that LRG1 is a novel angiogenic factor that drives DKD pathogenesis through potentiation of endothelial TGF- other family of vascular growth factors necessary for the nor- b fi /activin receptor-like kinase 1 (ALK1) signaling. They also show mal functioning of the glomerular ltration barrier, is also that plasma LRG1 is associated with renal outcome in a cohort of – implicated in altered permeability in DKD.15 17 In addition patients with type 2 diabetes. These findings indicate that LRG1 to vascular growth factors, several other mechanisms of early has a pivotal role in DKD pathogenesis through TGF-b/ALK1 glomerular endothelial cell (GEC) injury are implicated in signaling and is a risk factor for disease progression. DKD pathogenesis, such as decreased nitric oxide availabil- 18–20 21,22 ity and disturbances in endothelial barrier. These of Bambi significantly worsened diabetic glomerulopathy studies suggest that endothelial dysfunction is among the ear- in relatively DKD-resistant C57BL/6 mice, which was as- liest functional changes in diabetic kidneys. Indeed, a study by sociated with increased ALK1-mediated TGF-b signaling. 23 Weil et al. indicated that GEC injury precedes podocyte foot In an opposite manner to BAMBI, LRG1 was shown to process effacement and microalbuminuria in Pima Indians potentiate the proangiogenic ALK1 pathway by recruit- with type 2 diabetes. ment of TGF-b accessory receptor endoglin (ENG).30 Toassess the molecular changes associated with GEC injury This led us to speculate that the function of LRG1 would fi in early DKD, we recently performed transcriptomic pro ling promote diabetes-induced glomerular angiogenesis via of isolated GECs from streptozotocin (STZ)-induced diabetic ALK1-induced signal transduction in GECs. Therefore, in fi mice with endothelial nitric oxide synthase (eNOS) de - this study, we examined the role of LRG1 in diabetic glo- 24 ciency. Analysis of differentially expressed genes indeed merular injury. indicated that angiogenesis and endothelial proliferation pathways were significantly upregulated in GECs of diabetic mice,24 and that leucine-rich a-2-glycoprotein 1 (LRG1) was METHODS among the top upregulated genes. LRG1 is a secreted glycoprotein belonging to the leucine-rich repeat protein family,25 Study Approval whose expression is increased in plasma and urine of patients All mouse studies were approved by the Institutional Animal with a variety of cancer malignancies and is thought to con- Care and Use Committee at the Icahn School of Medicine at tribute to tumor-growth by promoting angiogenesis.26–29 Mount Sinai (ISMMS) and were performed in accordance with Notably, a recent finding by Wang et al.30 demonstrated that its guidelines. LRG1 mediates the proangiogenic effects through enhance- ment of TGF-b1/ALK1-mediated signaling in murine models Mouse Models of ocular disease. Mice were housed in a specific pathogen-free facility with TGF-b regulates many aspects of endothelial functions in- free access to chow and water and a 12-hour day/night cycle. 2 2 cluding cell proliferation in early diabetic kidneys resulting in Lrg1 / mouse strain in C57BL/6J background was obtained glomerular hypertrophy, but also the induction of apoptosis in from the Knock-Out Mouse Project Repository (#VG10067; microvascular endothelial cells.31 These contrasting actions of www.komp.org). Flk1-H2B-EYFP (EYFP) mice in the C57BL/6J TGF-b on endothelial cells (ECs) are regulated by differential background38 were a generous gift from Dr. Margaret Baron activation of two type 1 receptors: the ubiquitously expressed (ISMMS). Diabetes was induced with intraperitoneal injec- ALK5 and predominantly EC-restricted ALK1 receptors. tions of STZ (Sigma-Aldrich, St. Louis, MO) for five consecu- ALK5 activation induces the Smad2/3 activation to block EC tive days (50 mg/g body wt per day). To accelerate diabetic proliferation, migration, and angiogenesis; in contrast, ALK1 nephropathy, unilateral right nephrectomy was performed, as activation induces the Smad1/5/8 activation to promote described,39 3 weeks before STZ injection. Body weight and EC proliferation, migration, and tube formation, resulting fasting blood glucose levels were monitored biweekly. Diabetes in neoangiogenesis.32–36 However, the role of ALK1 in glomer- was confirmed by fasting blood glucose level .300 mg/dl. The ular angiogenesis has not been fully explored. The importance age- and sex-matched littermates injected with citrate vehicle of the TGF-b–induced ALK1 pathway in driving the DKD served as nondiabetic controls. BP was measured using the pathogenesis was shown in our recent report on BMP noninvasive tail-cuff BP system (Kent Scientific, Torrington, and Activin Membrane-Bound Inhibitor (BAMBI), a TGF-b CT). All mice were euthanized at either 12 or 20 weeks post- pseudoreceptor that represses TGF-b signaling.37 Genetic loss STZ or -vehicle injection.

2 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

AB 30 4 P=0.004 P<0.001 non-diabetic P<0.05 diabetic

20 2 P<0.001

10 0 Lrg1 mRNA (RPKM) Median-centered Log2 glomerular Lrg1 expression 0 -2 Vehicle STZ DBA/2J db/db db/db;eNOS-/-

C FVB WT OVE26 Neg control D In situ 4 hybridization P<0.01 3 Probes: CD31 2 LRG1 (Scores 0-4) 1 LRG1/CD31 expression 0

OVE26 FVB-WT

E F HLD DKD Neg control In situ P<0.01 hybridization 4

Probe: LRG1 3

2 IF: (Scores 0-4) anti-CD31 1 DAPI LRG1/CD31 expression 0 HLD DKD

Figure 1. LRG1 expression is increased in glomeruli of mouse and human diabetic kidneys. (A) RNA-sequencing data of Lrg1 ex- 2 2 pression in isolated GECs of vehicle- or STZ-injected eNOS / mice (reproduced from Fu et al.24). Each sample represents isolated GECs pooled from three to four mice. RPKM, reads per kilobase million. (B) Levels of Lrg1 mRNA in isolated glomeruli of STZ-induced 2 2 diabetic DBA/2J, and db/db and in eNOS / db/db mice in C57BLKS background compared with their respective nondiabetic controls from Nephroseq database (nephroseq.org). (C) In situ hybridization detecting Lrg1 mRNA (blue) and Pecam (CD31) mRNA (pink) in kidneys of diabetic OVE26 mice and wild-type control mice (FVB WT). Magniﬁed view of the outlined glomeruli in the upper panel is shown in the lower panels. Arrows highlight examples of Lrg1 expression in renal tubules and arrowhead highlight examples of

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 3 BASIC RESEARCH www.jasn.org

Statistical Analyses time during the follow-up period was counted as an event, Data are expressed as mean6SD. For comparisons of means and otherwise counted as a nonevent. We computed the time between two groups, two-tailed, unpaired t tests were performed. to event as the time to first dialysis/transplant for ESRD or the For comparison of means between three or more groups, time to the first episode of 40% sustained eGFR decline. For ANOVAwith Tukey post-test was applied. Statistical significance participants without an event, the follow-up time was com- was considered as P,0.05. Prism software v.6 (GraphPad, puted until the end of follow-up. A sequential adjustment was La Jolla, CA) was used for statistical analyses. used to assess the independent association of the biomarkers with the composite endpoint. The area under the receiver Clinical Cohort and Analyses operative curves was constructed to assess the discrimination The BioMe Biobank Program of The Charles Bronfman In- for clinical variables versus clinical variables plus LRG1 at a stitute for Personalized Medicine started in 2007 at the timepoint of 4.5 years (chosen to coincide with the median ISMMS.40,41 We randomly selected 997 BioMe participants follow-up of 4.5 years). The base clinical model consisted of with type 2 diabetes, a baseline eGFR between 45 and 90 ml/min age, sex, baseline eGFR, mean arterial pressure, hemoglobin per 1.73 m2 at the time of BioMe enrollment, and at least 3 A1C, hypertension, heart failure, coronary artery disease, years of follow-up data in the electronic health record. We body mass index, and angiotensin-converting enzyme in- defined the baseline period as 1 year before the BioMe en- hibitor/angiotensin receptor blocker use. All analyses rollment date. We determined baseline and all follow-up were performed using Stata version 13 (StataCorp., College eGFR values using the CKD Epidemiology Collaboration Station, TX). creatinine equation,42 calculated median values per Additional methods used for the study are included in the 3-month period of follow-up and utilized these for covariate Supplemental Methods. and outcome ascertainment. We calculated follow-up time from BioMe enrollment date to the latest visit in the electronic health record. RESULTS

LRG1 Measurements Expression of LRG1 Is Increased in Mouse and Human Plasma samples were taken at the time of BioMe enrollment Diabetic Kidneys andstoredat280°C were used to derive the baseline bio- Our recent transcriptomic profiling of isolated GECs24 marker measures. Plasma concentrations of LRG1 were mea- indicated that Lrg1 was one of the highly expressed genes in 2 2 sured via ELISA (#27769; IBL America, Minneapolis, MN). STZ-induced diabetic eNOS-deficient (eNOS / ) mice when 2 2 The lower limit of detection was 0.17 ng/ml, and the intra- compared with control eNOS / mice (Figure 1A). Our assay coefficient of variation for the quality control samples previous transcriptomic analysis of isolated glomeruli and was 6.03%, and the interassay coefficient of variation on 5% of podocytes from the same diabetic mouse model43 indicated samples run in duplicate was 6.3%. The laboratory personnel that although Lrg1 is increased in the glomeruli, no differ- performing the biomarker assays were blinded to clinical in- ence was detected in isolated podocytes (Supplemental formation about the participants. Figure 1A). Notably, consistent with the EC-restricted expression of ALK1 receptor, the transcriptomic data further Outcomes confirmed high expression of Alk1 mRNA in specifically in We defined the primary outcome as a composite of ESRD via GECs, whereas much lower Alk5 expression was found in linkage to the US Renal Data System or a sustained 40% decline podocytes and GECs (Supplemental Figure 1B). Increased in eGFR (confirmed decline on two or more separate 3-month Lrg1 mRNA expression was similarly observed in isolated intervals) over the follow-up period. glomeruli of three other diabetic mouse models, namely 2 2 STZ-induced diabetic DBA/2J, and db/db and eNOS / ; Statistical Analyses db/db mice in C57BLKS background44 in the Nephroseq da- The association of LRG1 with the composite renal end point tabase (nephroseq.org) (Figure 1B). was evaluated both continuously (log base 2–transformed) and We next determined the expression pattern of LRG1 mRNA by tertile. We utilized Cox regression with a time-to-event in mouse and human control and diabetic kidneys by in situ analysis where the composite renal outcome occurring any RNA hybridization.45 For Lrg1 expression in diabetic mouse

Lrg1 expression that colocalize with CD31 outside the tubules. Scale bars, 20 mm. (D) Semiquantitative scoring of Lrg1 mRNA colocalized with Pecam mRNA (n=3 mice per group, 20 gloms scored per mouse). (E) In situ hybridization of LRG1 mRNA (red) combined with CD31 immunoﬂuorescence (green) on human kidney sections with nuclear counterstain. Glomeruli are outlined with dotted lines on upper panels, and magniﬁed view of the box is shown in lower panels. Scale bar, 50 mm. (F) Semiquantitative scoring of LRG1 mRNA colocalized with CD31 (n=3 kidneys per group, 15 gloms scored per kidney).

4 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

A UNx +STZ Euthanize

8-week old Lrg1+/+ -/- Lrg1 -4 -2 0 2 4 6 8 10 12 weeks

B 800 C 18 Lrg1+/+ -/- -STZ *** # Lrg1 600 +/+ +/+ 12 Lrg1 Lrg1 -/- +STZ -/- +STZ Lrg1 400 Lrg1

+/+ 6 200 Lrg1 -STZ Lrg1-/- Kidney/BW (mg/g) Blood glucose (mg/dL) 0 0 0246810 12 Weeks

D -STZ +STZ Lrg1+/+ Lrg1-/- Lrg1+/+ Lrg1-/-

E 500 0.4 Lrg1+/+ *** # -STZ *** Lrg1-/- 400 *** ### 0.3 +/+ ) * Lrg1 3 +STZ 300 Lrg1-/- μ m

3 0.2 200 (×10 0.1

Glomerular volume 100 Mesangial matrix fraction 0 0.0

FG 500 300 Lrg1+/+ +/+ -STZ Lrg1 -/- +STZ *** Lrg1 400 Lrg1-/- +/+ 200 Lrg1 -/- +STZ 300 *** Lrg1 ## 200 100 UAE ( μ g/24h) UACR ( μ g/mg) 100 Lrg1+/+ -STZ Lrg1-/- 0 0 0246810 12 Weeks

Figure 2. LRG1 ablation protects against diabetic glomerulopathy. (A) Schematics of the experimental design. UNx was performed in 2 2 8-week-old Lrg1+/+ and Lrg1 / mice 3 weeks before injection of STZ or citrate buffer vehicle. Mice were euthanized at 12 weeks postinjection for analysis (n=6–8 mice per group). (B) Biweekly blood glucose measurements of control and diabetic mice. (C) Kidney- to-body weight (BW) ratio at 12 weeks post-DM induction. (D) Representative images of periodic acid-Schiff (PAS)-stained kidneys. Scale bar, 20 mm. (E) Quantiﬁcation of glomerular volume and mesangial matrix fraction per mouse (20–30 gloms counted per mouse,

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 5 BASIC RESEARCH www.jasn.org

ABLrg1+/+ Lrg1-/- 1500 +/+ *** Lrg1 -STZ Lrg1-/- ## +/+ 1000 ** Lrg1 -/- +STZ -STZ Lrg1

500

Podocyte FP widths (nm) 0

+STZ

CDLrg1+/+ Lrg1-/- 25 +/+ Lrg1 -STZ ### -/- 20 Lrg1 p57 *** Lrg1+/+ DAPI 15 +STZ Lrg1-/- -STZ 10 5

p57+ cells/glom section 0

+STZ

Figure 3. LRG1 ablation reduces podocyte foot process effacement. (A) Representative transmission electron microscopy images of mouse glomeruli. Red arrowheads highlight effaced podocyte foot processes. Scale bar, 5 mm. (B) Quantification of average podocyte foot process width per mouse (n=5–6 mice per group, 10–20 fields analyzed per mouse). (C) Representative images of p57 immunofluorescence. Glomeruli are outlined with dotted white lines. Scale bar, 20 mm. (D) Quantification of number of p57-positive podocytes per mouse (n=5 mice per group, 20–30 gloms scored per mouse). ***P,0.001 versus nondiabetic control; ##P,0.01 and ###P,0.001 versus diabetic Lrg1+/+ mice. kidneys, we used kidney sections from another experimental detected in the tubulointerstitium in normal kidneys, which model of type 1 DM, namely the OVE26 mouse model,46 further increased in the diabetic kidneys (Figure 1C, arrows to further corroborate the above transcriptomic findings. and arrowheads). However, overall Lrg1 expression was more Mouse Lrg1 mRNA probe (blue) colocalized predominantly prominent in the glomeruli than in the tubulointerstitium. with CD31 mRNA probe (pink) in the glomeruli of control In situ hybridization for LRG1 in human kidneys combined mouse kidneys, and its expression markedly increased in the with immunostaining for CD31 showed colocalization of glomeruli of OVE26 (Figure 1, C and D). Lrg1 signal was also LRG1 mRNA (red) with CD31 immunostained cells (green)

n=6 mice). (F) Urinary albumin-to-creatinine ratio (UACR) over time (n=6 per group). (G) Twenty four-hour urinary albumin excretion (UAE) at 12 weeks postinjection (n=6 mice per group). *P,0.05; **P,0.01; and ***P,0.001 versus respective nondiabetic controls; #P,0.05; ##P,0.01; and ###P,0.001 versus diabetic Lrg1+/+ mice.

6 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

A UNx +STZ DIVAA Euthanize 8-week old Lrg1+/+ Lrg1-/- (n=6 per group) -4 -2 0 2 4 6 8 10 12 weeks

Neg. Pos. B Lrg1+/+ Lrg1-/- control control

1800 +/+ ### Lrg1 -STZ *** -/- -STZ *** Lrg1

1200 Lrg1+/+ +STZ Lrg1-/-

600 +STZ

FITC-lectin intensity (AU) 0

C Lrg1+/+ Lrg1-/- CD31 DAPI 30 +/+ Lrg1 -STZ Lrg1-/- -STZ *** 20 ### Lrg1+/+ +STZ *** Lrg1-/-

+STZ % CD31+ glom area 0

D Lrg1+/+;EYFP Lrg1-/-;EYFP EYFP

75 Lrg1+/+:EYFP -STZ -/- -STZ *** ## Lrg1 :EYFP * 50 Lrg1+/+:EYFP +STZ Lrg1-/-:EYFP

+STZ # EYFP cells per glom 0

E Lrg1+/+ Lrg1-/- CD31 8-oxo-dG DAPI 20 Lrg1+/+ -STZ *** Lrg1-/- -STZ 15 ### Lrg1+/+ *** -/- +STZ 10 Lrg1

5 +STZ % 8-oxo-dG+ glom area 0

Figure 4. LRG1 ablation attenuates angiogenesis in vivo. (A) Schematics of the experimental design. UNx was performed in 8-week-old 2 2 Lrg1+/+ and Lrg1 / mice 3 weeks before injection of either low doses of STZ or citrate buffer vehicle. Directed in vivo angiogenesis

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 7 BASIC RESEARCH www.jasn.org in the glomeruli (Figure 1, F and G), consistent with its high ablation would alter the DM-induced angiogenesis in expression in GECs and its upregulation in DKD. vivo. Directed in vivo angiogenesis assay50,51 was performed by implanting the angioreactors filled with basement mem- Genetic Deletion of LRG1 Attenuates Diabetes- brane extracts subcutaneously in mice at 7 weeks post-DM. Induced Albuminuria and Glomerulopathy The vascularization within the angioreactors was analyzed af- To determine the role of LRG1 in DKD, we examined the ter 14 days (Figure 4A). As anticipated, there was a marked 2 2 effectsofLRG1lossusingtheglobalLrg1-null (Lrg1 / ) increase in vascularization into the angioreactors in diabetic 2 2 mice. Lrg1 / mice were grossly normal, as reported previ- mice compared with nondiabetic controls (Figure 4B). However, 2 2 ously.30 Ablation of LRG1 had no effect on kidney function this response was attenuated in diabetic Lrg1 / mice when when examined up to 9 months of age (data not shown). compared with diabetic Lrg1+/+ mice. Consistent with this 2 2 Because Lrg1 / mice were in the relatively DKD-resistant finding, there was an increased CD31-positive immunofluo- C57BL/6J background, unilateral nephrectomy (UNx) was rescence area in glomeruli of diabetic mice, which was also 2 2 performed before diabetes induction to aggravate the ensuing attenuated in diabetic Lrg1 / kidneys at 12 weeks post-DM diabetic kidney injury.47 As shown in Figure 2A, diabetes was (Figure 4C). 2 2 induced by low-dose STZ injections in Lrg1 / mice and in To quantitate the change in the number of GECs in the Lrg1+/+ littermate controls (+STZ) three weeks post-UNx. context of LRG1 ablation in early diabetic kidneys, Lrg1+/+ 2 2 Citrate buffer vehicle-injected UNx mice served as nondia- and Lrg1 / mice were crossed with Flk1-H2B-EYFP trans- betic controls (2STZ),andallmicewereeuthanized genic mice,24,38 which expresses a nuclear-enhanced yellow at 12 weeks post-DM induction. The extent of hyperglyce- fluorescent protein (EYFP) driven by Flk1 promoter. Dia- mia, body weight loss, and BP were similar between diabetic betes was similarly induced with low-dose STZ injections 3 2 2 Lrg1+/+ and Lrg1 / mice compared with nondiabetic mice weeks after UNx in Lrg1+/+;Flk1-H2B-EYFP (Lrg1+/+; 2 2 2 2 (Figure 2B, Supplemental Figure 2A, and Supplemental EYFP) and in Lrg1 / ;Flk1-H2B-EYFP (Lrg1 / ;EYFP) Table 1). However, DM-induced kidney hypertrophy was mice. Because the above directed in vivo angiogenesis assay observed only in diabetic Lrg1+/+ mice, as measured by results showed differential angiogenic potential when ex- kidney-to-body weight ratio (Figure 2C). Histologic anal- amined between 7–9 weeks post-STZ injection, Lrg1+/+; 2 2 ysis showed that the development of glomerular hypertro- EYFP and Lrg1 / ;EYFPmiceweresacrificed at 8 weeks phy and mesangial expansion was significantly blunted in post-DM, and the total number of EYFP+ cells per glomer- 2 2 diabetic Lrg1 / mice compared with diabetic Lrg1+/+ uli were quantified in 500 mm optically cleared kidney sec- mice(Figure2,DandE).Theextentofalbuminuria,as tions49 by two-photon microscopy. Indeed, the number of assessed by the albumin-to-creatinine ratio of spot-collected EYFP+ cells in the diabetic Lrg1+/+;EYFP mouse glomeruli urine samples over 12 weeks duration and by 24-hour urine was found to be increased at 8 weeks post-DM, but this albumin excretion at 12 weeks post-DM, was markedly attenu- increase was mitigated in the diabetic Lrg1+/+;EYFP mouse 2 2 ated in diabetic Lrg1 / mice compared with diabetic Lrg1+/+ glomeruli (Figure 4D, Supplemental Files 1–4). An in- mice (Figure 2, F and G). DM-induced podocyte foot creased number of Ki-67–positive cells was also observed process effacement and loss (as ascertained by podocyte in the glomeruli of diabetic Lrg1+/+mice at 12 weeks post- 2 2 marker p5748,49) were also reduced in diabetic Lrg1 / mice DM compared with those of diabetic Lrg1-/- mice (Supple- (Figure 3, A–D). Together, these data indicate that the genetic mental Figure 2B). deletion of LRG1 results in marked attenuation of diabetic As oxidative stress is a key component in DKD develop- glomerulopathy. ment,52 we next examined the extent of oxidative damage in by the detection of 8-oxo-29deoxyguanosine (8-oxo-dG). LRG1 Deletion Attenuates Diabetes-Induced The presence of 8-oxo-dG was markedly elevated in the glo- Angiogenesis meruli of diabetic mice compared with those of nondiabetic As LRG1 was previously shown to be proangiogenic in mice, which largely overlapped with CD31+ cells. How- the retinal vasculature,30 we next examined whether LRG1 ever much less 8-oxo-dG were detected in the glomeruli of

assay (DIVAA) was performed at 7 weeks postinjection, and implanted angioreactors were taken out after 2 weeks (n=6 mice per group). (B) Images of angioreactors 14 days postimplantation. Positive control (Pos.) angioreactor was supplemented with FGF and VEGF, negative control (Neg.) was supplemented with no factors, and all others were supplemented with FGF alone. Quantification of FITC-lectin bound dissociated cells from angioreactors is shown on the right. (C) Representative CD31 immunofluorescence images and quantification of CD31-positive glomerular area per mouse (n=5 mice per group, 20–30 glomeruli per mouse). (D) Two-photon microscopy imaging 2 2 and quantification of glomerular EYFP-positive cells in Lrg1+/+;EYFP and Lrg1 / ;EYFP mice (n=3 mouse, ten sections per mouse). (E) Representative 8-oxo-dG/CD31 immunofluorescence images and quantification of 8-oxo-dG glomerular area per mouse (n=5 mice per group, 20–30 glomeruli per mouse). Scale bar, 20 mm. *P,0.05 and ***P,0.001 versus nondiabetic controls; ##P,0.01 and ###P,0.001 versus diabetic Lrg1+/+.

8 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

AB+HM +HG +HM +HG + - + - Control vector Control - shScr shLrg1 --+ + LRG1-V5 LRG1 V5 IP: ENG p-Smad1/5/8 TGFβRII t-Smad1/5/8

p-Smad3 Total Input: V5 t-Smad3

ENG CD31

TGFβRII GAPDH

4 *** *** C HM *** * HG 3 ** * HG+shScr HG+shLrg1 2 n.s.

(normalized) 1

Fold change in intensity 0 LRG1 pSmad1/5/8 pSmad3 CD31

D Lrg1+/+ Lrg1-/- p-Smad1/5/8 CD31 1.0 DAPI +/+ Lrg1 -STZ 0.8 *** Lrg1-/- -STZ +/+ 0.6 Lrg1 +STZ Lrg1-/- 0.4 ### 0.2 +STZ

per total glomerular cells 0.0 Ratio of p-Smad1/5/8+ cells

E Lrg1+/+ Lrg1-/- p-Smad3 CD31 DAPI 1.5 +/+ Lrg1 -STZ Lrg1-/- -STZ *** ### 1.0 +/+ *** Lrg1 +STZ Lrg1-/- 0.5 Ratio of p-Smad3+ cells +STZ per total glomerular cells 0.0

Figure 5. LRG1 potentiates the ALK1-Smad1/5/8 pathway in GECs in diabetic conditions. (A) Immunoprecipitation with anti-V5 antibody using lysates of conditionally immortalized mGECs transduced with pHR lentiviral vector (Control vector) or pHR expressing recombinant V5-tagged LRG1 (LRG1-V5) cultured under high mannitol (HM) or high glucose (HG) conditions for 48 hours. Recombinant LRG1 coimmunoprecipitates with endogenous ENG and TGF-b receptor type 2 (TbRII) in mGECs in both conditions. (B and C) Effects of Lrg1 knockdown was examined in mGECs transduced with pLKO lentiviral vector (2), pLKO vector expressing scrambled shRNA

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 9 BASIC RESEARCH www.jasn.org

2 2 diabetic Lrg1 / mice compared with diabetic Lrg1+/+ mice. glomeruli in the diabetic kidneys, thereby promoting DKD Together, our data strongly indicate that LRG1 is a critical progression. mediator of angiogenesis and endothelial injury in diabetic kidneys. Renoprotection Conferred by LRG1 Deletion Is Maintained at a Later Stage of DKD LRG1 Potentiates ALK1-Mediated Smad1/5/8 As diabetes-induced glomerular neoangiogenesis is thought to Activation in GECs occur in the early stages of DKD, we next examined the efficacy LRG1 was previously shown to promote TGF-b–induced of long-term loss of LRG1 in a later stage of DKD and whether ALK1-Smad1/5/8 signal transduction by recruitment of its ablation may have potentially detrimental consequences in accessory TGF-b receptor ENG and TGF-b receptor type 2 the disease progression. Diabetes was induced similarly as 2 2 (TbR-II) in brain endothelial cells and in HUVECs.30 To above in UNx Lrg1+/+ and Lrg1 / mice(aswellasinLrg1+/+; 2 2 examine whether LRG1 similarly interacted with ENG and EYFP and Lrg1 / ;EYFP mice) and the effects of LRG1 loss TbR-II in GECs, we transduced the immortalized murine were examined after 20 weeks post-DM (Figure 6A). The extent GECs (mGECs)24 with the recombinant lentivirus expressing of hyperglycemia, body weight loss, and kidney-to-body the control V5-tag or V5-tagged LRG1 (Supplemental Figure weight ratio compared with nondiabetic mice was similar be- 2 2 3A). mGECs were then exposed to high-glucose (HG, tween the diabetic Lrg1+/+ and Lrg1 / mice at 20 weeks post- 30 mM) or high-mannitol (HM, 25 mM mannitol and DM (Supplemental Figure 4). Although a modest reduction in 5 mM glucose) culture conditions for 48 hours. Indeed, glomerular hypertrophy and mesangial expansion still persis- 2 2 both ENG and TbR-II coimmunoprecipitated with V5- ted in diabetic Lrg1 / mice as compared with diabetic Lrg1+/+ tagged LRG1 and these interactions appeared to increase in mice (Figure 6C), proteinuria and overall renal function, HG conditions (Figure 5A). assessed by BUN measurement, showed a significant improve- 2 2 We recently showed that HG conditions induced the ex- ment in diabetic Lrg1 / mice (Figure 6D). As expected, pression of LRG1 in mGECs and that the shRNA-mediated podocyte loss and effacement had worsened in the diabetic knockdown of Lrg1 attenuated the HG-induced increase in mice of both genotypes at 20 weeks post-DM compared endothelial tube formation in vitro.24 Therefore, we next ex- with 12 weeks post-DM, but the ablation of LRG1 signifi- amined whether the suppression of LRG1 could impede the cantly reduced podocyte loss and injury (Figures 3 and 7, A ALK1 signaling in GECs. mGECs stably transduced with re- and B). Similarly, DM-induced glomerular basement 2 2 combinant lentivirus expressing shRNA against Lrg1 or membrane thickening was reduced in diabetic Lrg1 / scrambled control sequence (Supplemental Figure 3B) were mice (Figure 7A). Moreover, LRG1 ablation conferred pro- exposed to HG or HM for 48 hours. HG-induced elevation of tection against DM-induced oxidative damage, as assessed LRG1 expression in mGECs was associated with increased by detection of 8-oxo-dG (Figure 7C). Interestingly, when phospho-Smad1/5/8, but not phospho-Smad3 (Figure 5, B the total number of EYFP-positive GECs per glomeruli was and C). Correspondingly, knockdown of Lrg1 markedly re- calculated at 20 weeks post-DM, we found that the initial duced phosphorylation of Smad1/5/8 under HG conditions, increase in GECs observed at 8 weeks post-DM (Figure 4D) without affecting the phosphorylation of Smad3. Consistent was no longer observed at 20 weeks in diabetic Lrg1+/+; with the in vitro findings, Smad1/5/8 activation was markedly EYFP mice (Figures 4D and 7D, Supplemental Files 5 and enhanced in the glomeruli of diabetic Lrg1+/+ mice, which 6), whereas this number appeared to be unaffected in di- 2 2 largely colocalized with CD31 immunostaining (Figure 5D). abetic Lrg1 / ;EYFP mice. Taken together, these data in- However, very little phospho-Smad1/5/8 was detected in the dicate that the ablation of LRG1 does not contribute to any 2 2 glomeruli of diabetic Lrg1 / mice (Figure 5D), indicating unfavorable outcome, but confers long-term renoprotec- that the loss of LRG1 nearly abrogated the ALK1 signal trans- tion against DKD. ductioninGECsintheglomeruliofdiabetickidneys.Incon- trast, the activation of Smad3 detected in the glomeruli of Plasma LRG1 as a Marker of DKD Progression in diabetic mice was only modestly affected by the loss of Humans LRG1 (Figure 5E). Taken together, our results indicate As LRG1 is a secreted glycoprotein elevated in human and that LRG1 potentiates ALK1-mediated angiogenesis in the mouse diabetic kidneys, we next examined whether plasma

(shScr), or shRNA against LRG1 (shLrg1). Lysates were subjected to Western blot analysis using primary antibodies as indicated. Repre- sentative western blot image (of triplicate experiments) is shown in (B) and densitometric analysis is shown in (C). *P,0.05; **P,0.01; and ***P,0.001 when compared between indicated groups by ANOVA with Tukey corrections. (D and E) Immunostaining of (D) phospho- 2 2 Smad1/5/8 (p-Smad1/5/8) and CD31 or (E) phospho-Smad3 (p-Smad3) and CD31 on frozen kidney sections of Lrg1+/+ and Lrg1 / mice. The ratio of p-Smad1/5/8–positive or p-Smad3–positive cells per total glomerular cells per kidney section is shown on the right (n=3 mice, 15–20 sections scored per mouse). Scale bar, 20 mm. White arrowheads show examples of nuclear-localized p-Smad1/5/8 or p-Smad3. ***P,0.001 versus nondiabetic control; ###P,0.001 versus diabetic Lrg1+/+.

10 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

A Sacrifice and UNX +STZ harvest tissues

8-week old Lrg1+/+ -/- Lrg1 -4 -2 0 2 4 6 8 10 12 14 16 20 weeks

B -STZ +STZ Lrg1+/+ Lrg1-/- Lrg1+/+ Lrg1-/-

C 800 0.8 Lrg1+/+ *** *** -STZ ## ### Lrg1-/- 600 *** 0.6 *** ) 3 Lrg1+/+ -/- +STZ μ m Lrg1

3 400 0.4 (×10 200 0.2 Glomerular volume Mesangial matrix fraction 0 0.0

D 600 180 *** Lrg1+/+ *** -STZ Lrg1-/- 400 120 ## Lrg1+/+ +STZ Lrg1-/-

200 60

## BUN (mg/dL) UAE (ug/24h)

0 0

Figure 6. Renoprotection by LRG1 loss persists in the later stage of DKD. (A) Schematics of the experimental design. UNx was 2 2 performed in 8-week-old Lrg1+/+ and Lrg1 / mice 3 weeks before injection of low doses of either STZ or citrate buffer vehicle. Mice were euthanized at 20 weeks postinjection for analysis (n=5–8 mice per group). (B) Representative image of kidneys stained with periodic acid–Schiff at low magnification (top panels; scale bar, 50 mm) and at high magnification (bottom panels; scale bar, 20 mm). (C) Quantification of glomerular volume and mesangial fraction at 20 weeks post-DM induction per mouse (n=6 mice per group, 20–30

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 11 BASIC RESEARCH www.jasn.org

LRG1 levels in patients with DKD are associated with renal well as infiltrating immune cells in the diseased kidneys. Fu- outcome. We measured LRG1 on plasma samples from a ture studies utilizing tissue-specific Lrg1 knockout or over- subcohort of 871 participants with type 2 diabetes in expression mouse models are required to delineate the role the BioMe Biobank Cohort at the ISMMS (Supplemental of LRG1 in diabetic kidneys in a cell-specificmanner. Figure 5). Supplemental Table 2 shows the baseline char- Our previous24 and current work demonstrate that the acteristics of patients. Median age was 60 years, 507 expression of LRG1 is elevated in GECs in response to HG (58.2%) were women, and median eGFR was 68.4 conditions in vitro, suggesting that hyperglycemia is one (interquartile range, 55.3–80.0) ml/min per 1.73 m2.Par- of the factors leading to its upregulation in diabetic kidneys. ticipants with higher LRG1 were older, more likely to be A recent study also showed that TNF-a induces LRG1 ex- women, and more likely to have a history of congestive pression to promote angiogenesis and mesenchymal stem heart failure. Plasma LRG1 was not significantly correlated cell migration in the subchondral bone of osteoarthritis with eGFR (r=20.02; P=NS) or albuminuria (r=0.06; joints.53 In silico analysis of LRG1 promoter indeed indi- P=NS) at baseline at the time of enrollment into the BioMe cated multiple potential binding sites for p65 NF-kB Biobank Program. Of the 871 patients analyzed, 121 and STAT3 transcription factors. Thus, it is plausible that (13.8%) experienced the composite end point, as defined the inflammatory cytokines known to be involved in the pro- by 40% sustained eGFR decline or ESRD during a median gression of DKD, such as TNF-a and IL-6,54 may also follow-up of 4.5 (interquartile range, 3.3–6.1) years. Notably, participate in LRG1 upregulation in diabetic kidneys. In participants with the renal end point had higher levels of addition, miR-335 has been shown to target multiple genes LRG1 (80.74 versus 53.79 ng/ml). The event rates for the in the noncanonical TGF-b pathway, including LRG1,55,56 composite renal end points of 40% sustained decline in and to inhibit HG-mediated apoptosis in osteoblasts.57 The eGFR or ESRD were 7.5% for the lowest tertile, 12.7% for suppression of miR-355 in diabetic conditions may be an- themiddletertile,and21.3%forthetoptertileofLRG1 other factor contributing to elevated LRG1 expression in (Figure 8), which yielded an adjusted hazard ratio of 1.6 GECs. Future studies are required to elucidate the detailed (95% confidence interval [95% CI], 0.9 to 2.7) and 2.7 mechanism of LRG1 upregulation in GECs of diabetic (95% CI, 1.6 to 4.4) for the middle and top tertiles, respec- kidneys. tively, versus the bottom tertile (Table 1). For each doubling In this study, we have demonstrated that the genetic de- in LRG1, the adjusted hazard ratio was 1.3 (95% CI, 1.1 to letion of LRG1 significantly attenuated diabetic kidney injury 1.4). The area under the receiver operative curves for the and TGF-b/ALK1-induced angiogenesis in the experimental composite kidney end point was 0.740 (95% CI, 0.69 to model of early DKD. The importance of ALK1-mediated 0.79) for the clinical model and increased to 0.765 (95% TGF-b signaling in endothelial cells in diabetic kidneys is CI, 0.72 to 0.82) with the addition of LRG1. Together, these also supported by our previous study demonstrating that results indicate that angiogenic molecule LRG1 is a risk factor the genetic loss of BAMBI, a negative regulator of TGF-b for DKD progression. signaling, resulted in significant augmentation of glomerular injury and proteinuria, which was associated with enhanced activation of ALK1 pathway in the glomeruli of diabetic DISCUSSION mice.37 Together, our prior and current work demonstrate the key pathogenic role of glomerular TGF-b/ALK1 signaling Our study highlights the role of LRG1 as a potential pathogenic in diabetic kidneys and suggest that the targeting of LRG1 mediator of glomerular neoangiogenesis through could be a unique approach for inhibition of the specific the activation of ALK1 signaling pathway. LRG1 expression TGF-b/ALK1 pathway to attenuate DKD progression. Indeed, is enriched in endothelial cells, particularly in GECs, and up- total blockade of TGF-b signaling may not be effective and regulated in diabetic kidneys. Although Lrg1 mRNA does not could be potentially hazardous, as it affects multiple pathways appear to be significantly expressed in podocytes, whether it is and systems through different receptors.58,59 Indeed, a recent expressed in other glomerular cells to potentiate TGF-b sig- clinical study of anti–TGF-b1 antibody blockade in patients naling remains to be determined. In situ hybridization anal- with DKD failed to show beneficial effects when added to RAS ysis showed that LRG1 expression was not limited to GECs in blockade.60 Although this lack of treatment efficacy could be diabetic kidneys, but expressed in tubulointerstitial compart- multifactorial, it further underscores the need for better un- ments. Thus, LRG1 is likely expressed in tubular epithelial and derstanding of TGF-b functions and their downstream sig- peritubular capillary endothelial cells in normal kidneys, as naling pathways in specifickidneycelltypesforthedesignof

gloms scored per mouse). (D) Renal function assessment at 20 weeks post-DM of 24-hour urinary albumin excretion (UAE) levels and BUN levels (n=5–8 mouse per group). ***P,0.001 versus respective nondiabetic controls; ##P,0.01 and ###P,0.001 versus diabetic Lrg1+/+ mice.

12 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

A Lrg1+/+ Lrg1-/- Lrg1+/+ 10000 -STZ Lrg1-/- *** 8000 Lrg1+/+ -/- +STZ 6000 Lrg1 ### +Vehicle * 4000

2000 * Podocyte FPW (nm) FPW Podocyte

800 Lrg1+/+ -STZ Lrg1-/- *** ### 600 Lrg1+/+ ** +STZ Lrg1-/- +STZ 400

GBM width (nm) 200

0 B Lrg1+/+ Lrg1-/- 25 Lrg1+/+ p57 -/- -STZ ### Lrg1 DAPI 20 Lrg1+/+ +STZ 15 *** Lrg1-/- +STZ 10

5 p57+ cells/p57+ glom section 0

C Lrg1+/+ Lrg1-/- Lrg1+/+ 40 -STZ CD31 *** Lrg1-/- 8-oxo-dG 30 Lrg1+/+ DAPI -/- +STZ ### Lrg1 20 *** +STZ 10 8-oxo-dG+ area(%)

+/+ -/- D Lrg1 ;EYFP Lrg1 ;EYFP +/+ 60 Lrg1 -/- -STZ EYFP n.s. Lrg1 Lrg1+/+ 40 +STZ Lrg1-/- +STZ

20 # EYFP cells/glom

Figure 7. LRG1 loss confers protection against podocyte and endothelial cell injury at the later stage of DKD. (A) Transmission electron microscopy 2 2 images of Lrg1+/+ and Lrg1 / mouse kidneys at 310,000 magnification (scale bar, 5 mm). Red arrowheads highlight the severe effacement of podocyte foot processes in the diabetic Lrg1+/+ mice. Quantification of average podocyte foot process widths (PFWs) and glomerular basement membrane (GBM) widths are shown on the right (n=3 mice per group, ten fields per mouse). (B) Representative images of p57 immunofluorescence and quantification per mouse (n=5 mice per group, 20–30 gloms scored per mouse). (C) Representative 8-oxo-dG and CD31 immunofluorescence images and quantification of 8-oxo-dG glomerular area per mouse. (D) Two-photon microscopy imaging and quantification of glomerular EYFP- 2 2 positive cells in Lrg1+/+;EYFP and Lrg1 / ;EYFP mice (n=3 mouse, ten sections per mouse; Supplemental Files 5 and 6 show examples of the z-stacks). Scale bar, 20 mm. *P,0.05; **P,0.01; ***P,0.001 versus respective nondiabetic controls; ###P,0.001 versus diabetic Lrg1+/+ mice.

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 13 BASIC RESEARCH www.jasn.org

1.0

0.8

0.6

0.4

Cumulative survival LRG1 (ng/ml) 0.2 ≤ 40.0 40.1-78.0 ≥ 78.1 0.0 0 20 40 60 80 Follow-up time in months

≤ 40.0 ng/ml 290 263 214 111 30 40.1-78.0 ng/ml 291 265 197 97 29 ≥ 78.1 ng/ml 290 256 185 116 51

Figure 8. Plasma LRG1 is associated with renal outcome in patients with type 2 DM. Kaplan–Meier curves stratified by tertiles of plasma LRG1 for the composite renal outcome of sustained 40% decline in eGFR or ESRD. more targeted treatment strategies. Moreover, in contrast to cancer malignancies26–29 and inflammatory diseases.62–65 Our ALK1 or ENG, whose haploinsufficiency is associated with findings in the BioMe cohort, consisting of mixed races of hereditary hemorrhagic telangiectasia,61 LRG1 deficiency in Western patients with diabetes, indicate that plasma LRG1 is mice does not appear to result in major physiologic defi- associated with renal outcome. Of note, during the prepara- cits.30 We did not observe any loss of renal function in non- tion of this report, similar findings regarding plasma LRG1 diabetic mice when examined up to 9 months of age, and were reported in a T2DM patient cohort in Singapore by Liu LRG1 deficiency persisted in conferring renoprotection in et al.66 Thus, the prognostic role of LRG1 in DKD progression later-stage DKD at 20 weeks post-DM, making LRG1 an has been confirmed in two separate patient cohorts, further attractive target for anti-angiogenic therapy in DKD. How- strengthening our observations and confirming the human ever, the limitation of our study is that the mouse model relevance of LRG1 in the pathogenesis of DKD. Future studies utilized in this study is a global LRG1 knockout model, are required to confirm the clinical implications of LRG1 and and as discussed above, LRG1 expression is not restricted other angiogenesis factors in DKD, and to assess the efficacy of to the endothelial cells. Thus, future studies using the endo- urinary LRG1 detection as a prognostic marker of DKD pro- thelial-specific Lrg1 knockout or overexpression model gression. In conclusion, our study demonstrates a pivotal role would be required to confirm the endothelial phenotypes of proangiogenic molecule LRG1 in the pathogenesis of DKD, observed in this study. and suggests that LRG1 is not only a promising therapeutic As secreted glycoprotein, LRG1 levels have been shown to target against DKD, but also a risk factor for disease increase in the plasma and urine of patients with a variety of progression.

Table 1. Association of plasma LRG1 with the renal outcome Model 1 Model 2 Model 3 LRG1, ng/ml NN(%) with outcome HR (95% CI) HR (95% CI) HR (95% CI) Continuous (per doubling) 871 121 (13.9) 1.3 (1.2 to 1.5) 1.3 (1.1 to 1.5) 1.3 (1.1 to 1.4) Tertiles #40.0 ng/ml 290 22 (7.5) 1.0 (Ref) 1.0 (Ref) 1.0 (Ref) 40.1–78.0 ng/ml 291 37 (12.7) 1.8 (1.1 to 3.1) 1.7 (1.0 to 2.9) 1.6 (0.9 to 2.7) $78.1 ng/ml 290 62 (21.3) 3.0 (1.8 to 4.8) 2.9 (1.7 to 4.7) 2.7 (1.6 to 4.4) Model 1: unadjusted analysis; model 2: adjusted for age, sex, and baseline eGFR; model 3: model 2 plus hemoglobin A1C, hypertension, coronary disease,heart failure, baseline mean arterial pressure, angiotensin-converting enzyme inhibitor/angiotensin receptor blocker use, and urine albumin to creatinine ratio (available as a covariate in 608 of the 871 participants). HR, hazard ratio.

14 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

ACKNOWLEDGMENTS 4. Brenner BM, Cooper ME, de Zeeuw D, Keane WF, Mitch WE, Parving HH, et al.: RENAAL Study Investigators: Effects of losartan on renal and cardiovascular outcomes in patients with type 2 diabetes and ne- Q.H., J.C.H., and K.L. designed the study. Q.H., L.Z., J.F., D.A.V., and phropathy. NEnglJMed345: 861–869, 2001 Z.L. conducted the experiments. K.C., W.J., G.N.N., M.K., V.D.D., and 5. de Zeeuw D, Heerspink HJL: Unmet need in diabetic nephropathy: S.G.C. acquired the data. Q.H., L.Z., D.A.V., G.-Y.C., X.-M.C., V.D.D., Failed drugs or trials? Lancet Diabetes Endocrinol 4: 638–640, S.G.C., J.C.H., and K.L. analyzed the data. Q.H., L.Z., S.G.C., J.C.H., 2016 and K.L. drafted the manuscript. Q.H., L.Z., V.D.D., D.S., J.C.H., and 6. Fried LF, Emanuele N, Zhang JH, Brophy M, Conner TA, Duckworth W, et al.: VA NEPHRON-D Investigators: Combined angiotensin inhibition K.L. revised the manuscript, and all authors approved the final version for the treatment of diabetic nephropathy. NEnglJMed369: 1892– of the manuscript. 1903, 2013 Q.H. is supported by the China Scholarship Council (grant no. 7. Parving HH, Brenner BM, McMurray JJ, de Zeeuw D, Haffner SM, 201503170082) and National Natural Science Foundation of China Solomon SD, et al.: ALTITUDE Investigators: Cardiorenal end points (grant no. 81870491), S.G.C. is supported by National Institutes of inatrialofaliskirenfortype2diabetes.NEnglJMed367: 2204–2213, 2012 Health (NIH) grants R01DK096549, U01DK106962, U01DK082185, 8. de Zeeuw D, Akizawa T, Audhya P, Bakris GL, Chin M, Christ-Schmidt H, andR01HL085757;G.N.N.issupported byNIH grantK23DK107908; et al.: BEACON Trial Investigators: Bardoxolone methyl in type 2 J.C.H. is supported by NIH grants R01DK078897, R01DK088541, diabetes and stage 4 chronic kidney disease. NEnglJMed369: R01DK109683, and P01DK56492, and VA merit award IBX000345C; 2492–2503, 2013 and K.L. is supported by NIH grant R01DK117913-01. 9. Osterby R, Nyberg G: New vessel formation in the renal corpuscles in advanced diabetic glomerulopathy. JDiabetComplications1: 122–127, 1987 10. Nyengaard JR, Rasch R: The impact of experimental diabetes mellitus DISCLOSURES in rats on glomerular capillary number and sizes. Diabetologia 36: None. 189–194, 1993 11. Guo M, Ricardo SD, Deane JA, Shi M, Cullen-McEwen L, Bertram JF: A stereological study of the renal glomerular vasculature in the db/db SUPPLEMENTAL MATERIAL mouse model of diabetic nephropathy. JAnat207: 813–821, 2005 12. Karalliedde J, Gnudi L: Endothelial factors and diabetic nephropathy. This article contains the following supplemental material Diabetes Care 34[Suppl 2]: S291–S296, 2011 online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ 13. Dei Cas A, Gnudi L: VEGF and angiopoietins in diabetic glomerulopathy: How far for a new treatment? Metabolism 61: 1666–1673, 2012 ASN.2018060599/-/DCSupplemental. 14. Sung SH, Ziyadeh FN, Wang A, Pyagay PE, Kanwar YS, Chen S: Supplemental Figure 1. RNA-sequencing expressionof mouse Lrg1 Blockade of vascular endothelial growth factor signaling ameliorates mRNA in isolated glomeruli, podocytes, or GECs. diabetic albuminuria in mice. JAmSocNephrol17: 3093–3104, Supplemental Figure 2. Additional characterization of diabetic and 2006 2 2 control Lrg1+/+ and Lrg1 / mice at 12 weeks post-DM. 15. Davis B, Dei Cas A, Long DA, White KE, Hayward A, Ku CH, et al.: Podocyte-specific expression of angiopoietin-2 causes proteinuria and Supplemental Figure 3. Overexpression and knockdown of LRG1 apoptosis of glomerular endothelia. JAmSocNephrol18: 2320–2329, in mGECs. 2007 Supplemental Figure 4. Additional characterization of diabetic and 16. Jeansson M, Gawlik A, Anderson G, Li C, Kerjaschki D, Henkelman M, 2 2 control Lrg1+/+ and Lrg1 / mice at 20 weeks post-DM. et al.: Angiopoietin-1 is essential in mouse vasculature during development and in response to injury. JClinInvest121: 2278–2289, 2011 Supplemental Figure 5. Flow diagram of patient selection into 17. Dessapt-Baradez C, Woolf AS, White KE, Pan J, Huang JL, Hayward AA, study cohort. et al.: Targeted glomerular angiopoietin-1 therapy for early diabetic Supplemental Table 1. BP measurements of diabetic and control kidney disease. JAmSocNephrol25: 33–42, 2014 mice. 18. Yuen DA, Stead BE, Zhang Y, White KE, Kabir MG, Thai K, et al.: eNOS fi Supplemental Table 2. Baseline characteristics of patients with de ciency predisposes podocytes to injury in diabetes. JAmSoc Nephrol 23: 1810–1823, 2012 DKD. 19. Zhao HJ, Wang S, Cheng H, Zhang MZ, Takahashi T, Fogo AB, et al.: Supplemental Methods. Endothelial nitric oxide synthase deficiency produces accelerated ne- Supplemental Files 1-6. phropathy in diabetic mice. JAmSocNephrol17: 2664–2669, 2006 20. Nakagawa T, Sato W, Glushakova O, Heinig M, Clarke T, Campbell- Thompson M, et al.: Diabetic endothelial nitric oxide synthase knock- REFERENCES out mice develop advanced diabetic nephropathy. J Am Soc Nephrol 18: 539–550, 2007 1. US Renal Data System: Annual Data Report: Atlas of End-Stage-Renal- 21. Satchell SC, Tooke JE: What is the mechanism of microalbuminuria in Disease in the United States, 2011. Available at : https://www.usrds. diabetes: A role for the glomerular endothelium? Diabetologia 51: org/atlas11.aspx. Accessed February 24, 2019 714–725, 2008 2. Lewis EJ, Hunsicker LG, Bain RP, Rohde RD; The Collaborative Study 22. Haraldsson B, Nyström J: The glomerular endothelium: New insights Group: The effect of angiotensin-converting-enzyme inhibition on di- on function and structure. Curr Opin Nephrol Hypertens 21: 258–263, abetic nephropathy. NEnglJMed329: 1456–1462, 1993 2012 3. Lewis EJ, Hunsicker LG, Clarke WR, Berl T, Pohl MA, Lewis JB, et al.: 23. Weil EJ, Lemley KV, Mason CC, Yee B, Jones LI, Blouch K, et al.: Collaborative Study Group: Renoprotective effect of the angiotensin- Podocyte detachment and reduced glomerular capillary endothelial receptor antagonist irbesartan in patients with nephropathy due to type fenestration promote kidney disease in type 2 diabetic nephropathy. 2 diabetes. NEnglJMed345: 851–860, 2001 Kidney Int 82: 1010–1017, 2012

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 15 BASIC RESEARCH www.jasn.org

24. Fu J, Wei C, Zhang W, Schlondorff D, Wu J, Cai M, et al.: Gene 43. Fu J, Wei C, Lee K, Zhang W, He W, Chuang P, et al.: Comparison of expression profiles of glomerular endothelial cells support glomerular and podocyte mRNA profiles in streptozotocin-induced their role in the glomerulopathy of diabetic mice. Kidney Int 94: diabetes. J Am Soc Nephrol 27: 1006–1014, 2016 326–345, 2018 44. Hodgin JB, Nair V, Zhang H, Randolph A, Harris RC, Nelson RG, et al.: 25. Takahashi N, Takahashi Y, Putnam FW: Periodicity of leucine and tan- Identification of cross-species shared transcriptional networks of dem repetition of a 24-amino acid segment in the primary structure of diabetic nephropathy in human and mouse glomeruli. Diabetes 62: leucine-rich alpha 2-glycoprotein of human serum. Proc Natl Acad Sci 299–308, 2013 USA82: 1906–1910, 1985 45. Wang F, Flanagan J, Su N, Wang LC, Bui S, Nielson A, et al.: RNAscope: 26. Li Y, Zhang Y, Qiu F, Qiu Z: Proteomic identification of exosomal LRG1: A novel in situ RNA analysis platform for formalin-fixed, paraffin- A potential urinary biomarker for detecting NSCLC. Electrophoresis 32: embedded tissues. J Mol Diagn 14: 22–29, 2012 1976–1983, 2011 46. Zheng S, Noonan WT, Metreveli NS, Coventry S, Kralik PM, Carlson EC, 27. Andersen JD, Boylan KL, Jemmerson R, Geller MA, Misemer B, et al.: Development of late-stage diabetic nephropathy in OVE26 di- Harrington KM, et al.: Leucine-rich alpha-2-glycoprotein-1 is upre- abetic mice. Diabetes 53: 3248–3257, 2004 gulatedinseraandtumorsofovariancancerpatients.JOvarianRes3: 47. Steffes MW, Brown DM, Mauer SM: Diabetic glomerulopathy following 21, 2010 unilateral nephrectomy in the rat. Diabetes 27: 35–41, 1978 28. Lindén M, Lind SB, Mayrhofer C, Segersten U, Wester K, Lyutvinskiy 48. Andeen NK, Nguyen TQ, Steegh F, Hudkins KL, Najafian B, Alpers CE: Y, et al.: Proteomic analysis of urinary biomarker candidates The phenotypes of podocytes and parietal epithelial cells may overlap for nonmuscle invasive bladder cancer. Proteomics 12: 135–144, in diabetic nephropathy. Kidney Int 88: 1099–1107, 2015 2012 49. Puelles VG, van der Wolde JW, Schulze KE, Short KM, Wong MN, 29. Furukawa K, Kawamoto K, Eguchi H, Tanemura M, Tanida T, Tomimaru Bensley JG, et al.: Validation of a three-dimensional method for Y, et al.: Clinicopathological significance of leucine-rich a2-Glyco- counting and sizing podocytes in whole glomeruli. J Am Soc Nephrol protein-1 in sera of patients with pancreatic cancer. Pancreas 44: 27: 3093–3104, 2016 93–98, 2015 50. Guedez L, Rivera AM, Salloum R, Miller ML, Diegmueller JJ, Bungay 30. Wang X, Abraham S, McKenzie JAG, Jeffs N, Swire M, Tripathi VB, PM, et al.: Quantitative assessment of angiogenic responses by the et al.: LRG1 promotes angiogenesis by modulating endothelial TGF-b directed in vivo angiogenesis assay. Am J Pathol 162: 1431–1439, signalling. Nature 499: 306–311, 2013 2003 31. Böttinger EP, Bitzer M: TGF-beta signaling in renal disease. JAmSoc 51. Guillot N, Kollins D, Badimon JJ, Schlondorff D, Hutter R: Accelerated Nephrol 13: 2600–2610, 2002 reendothelialization, increased neovascularization and erythrocyte 32. Goumans MJ, Mummery C: Functional analysis of the TGFbeta receptor/ extravasation after arterial injury in BAMBI-/- mice. PLoS One 8: Smad pathway through gene ablation in mice. Int J Dev Biol 44: 253–265, e58550, 2013 2000 52. Jha JC, Banal C, Chow BS, Cooper ME, Jandeleit-Dahm K: Diabetes 33. Pardali E, ten Dijke P: Transforming growth factor-beta signaling and and kidney disease: Role of oxidative stress. Antioxid Redox Signal 25: tumor angiogenesis. Front Biosci 14: 4848–4861, 2009 657–684, 2016 34. Seki T, Yun J, Oh SP: Arterial endothelium-specific activin receptor-like 53. Wang Y, Xu J, Zhang X, Wang C, Huang Y, Dai K, et al.: TNF-a-induced kinase 1 expression suggests its role in arterialization and vascular LRG1 promotes angiogenesis and mesenchymal stem cell migration in remodeling. Circ Res 93: 682–689, 2003 the subchondral bone during osteoarthritis. Cell Death Dis 8: e2715, 35. Oh SP, Seki T, Goss KA, Imamura T, Yi Y, Donahoe PK, et al.: Activin 2017 receptor-like kinase 1 modulates transforming growth factor-beta 1 54. Navarro-González JF, Mora-Fernández C: The role of inflammatory signaling in the regulation of angiogenesis. Proc Natl Acad Sci U S A 97: cytokines in diabetic nephropathy. JAmSocNephrol19: 433–442, 2626–2631, 2000 2008 36. Ren B, Deng Y, Mukhopadhyay A, Lanahan AA, Zhuang ZW, Moodie 55. Lynch J, Fay J, Meehan M, Bryan K, Watters KM, Murphy DM, et al.: KL, et al.: ERK1/2-Akt1 crosstalk regulates arteriogenesis in mice and MiRNA-335 suppresses neuroblastoma cell invasiveness by direct zebrafish. JClinInvest120: 1217–1228, 2010 targeting of multiple genes from the non-canonical TGF-b signalling 37. Fan Y, Li X, Xiao W, Fu J, Harris RC, Lindenmeyer M, et al.: BAMBI pathway. Carcinogenesis 33: 976–985, 2012 elimination enhances alternative TGF-b signaling and glomerular 56. Yan Z, Xiong Y, Xu W, Gao J, Cheng Y, Wang Z, et al.: Identification of dysfunction in diabetic mice. Diabetes 64: 2220–2233, 2015 hsa-miR-335 as a prognostic signature in gastric cancer. PLoS One 7: 38. Fraser ST, Hadjantonakis AK, Sahr KE, Willey S, Kelly OG, Jones EA, e40037, 2012 et al.: Using a histone yellow fluorescent protein fusion for tagging and 57. Li J, Feng Z, Chen L, Wang X, Deng H: MicroRNA-335-5p inhibits tracking endothelial cells in ES cells and mice. Genesis 42: 162–171, osteoblast apoptosis induced by high glucose. Mol Med Rep 13: 2005 4108–4112, 2016 39. Zhong F, Mallipattu SK, Estrada C, Menon M, Salem F, Jain MK, et al.: 58. Varga J, Pasche B: Antitransforming growth factor-beta therapy in Reduced krüppel-like factor 2 aggravates glomerular endothelial cell fibrosis: Recent progress and implications for systemic sclerosis. Curr injury and kidney disease in mice with unilateral nephrectomy. Am J Opin Rheumatol 20: 720–728, 2008 Pathol 186: 2021–2031, 2016 59. Sureshbabu A, Muhsin SA, Choi ME: TGF-b signaling in the kidney: 40. Tayo BO, Teil M, Tong L, Qin H, Khitrov G, Zhang W, et al.: Genetic Profibrotic and protective effects. Am J Physiol Renal Physiol 310: background of patients from a university medical center in Manhattan: F596–F606, 2016 Implications for personalized medicine. PLoS One 6: e19166, 60. Voelker J, Berg PH, Sheetz M, Duffin K, Shen T, Moser B, et al.: Anti- 2011 TGF-b1 antibody therapy in patients with diabetic nephropathy. JAm 41. Nadkarni GN, Galarneau G, Ellis SB, Nadukuru R, Zhang J, Scott SA, Soc Nephrol 28: 953–962, 2017 et al.: Apolipoprotein L1 Variants and blood pressure traits in African 61. Lebrin F, Deckers M, Bertolino P, Ten Dijke P: TGF-beta receptor Americans. JAmCollCardiol69: 1564–1574, 2017 function in the endothelium. Cardiovasc Res 65: 599–608, 2005 42. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, Feldman 62. Serada S, Fujimoto M, Ogata A, Terabe F, Hirano T, Iijima H, et al.: HI, et al.: CKD-EPI (Chronic Kidney Disease Epidemiology Collabora- iTRAQ-based proteomic identification of leucine-rich alpha-2 glycoprotein tion): A new equation to estimate glomerular filtration rate. Ann Intern as a novel inflammatory biomarker in autoimmune diseases. Ann Rheum Dis Med 150: 604–612, 2009 69: 770–774, 2010

16 Journal of the American Society of Nephrology J Am Soc Nephrol 30: ccc–ccc,2019 www.jasn.org BASIC RESEARCH

63. Ha YJ, Kang EJ, Lee SW, Lee SK, Park YB, Song JS, et al.: Usefulness of 65. Shinzaki S, Matsuoka K, Iijima H, Mizuno S, Serada S, Fujimoto M, serum leucine-rich alpha-2 glycoprotein as a disease activity biomarker et al.: Leucine-rich alpha-2 glycoprotein is a serum biomarker of in patients with rheumatoid arthritis. J Korean Med Sci 29: 1199–1204, mucosal healing in ulcerative colitis. J Crohn’s Colitis 11: 84–91, 2014 2017 64. Kentsis A, Ahmed S, Kurek K, Brennan E, Bradwin G, Steen H, Bachur R: 66. Liu JJ, Pek SLT, Ang K, Tavintharan S, Lim SC; SMART2D study: Plasma Detection and diagnostic value of urine leucine-rich a-2-glycoprotein leucine-rich a-2-Glycoprotein 1 predicts rapid eGFR decline and al- in children with suspected acute appendicitis. Ann Emerg Med 60: buminuria progression in type 2 diabetes mellitus. J Clin Endocrinol 78–83 e71, 2012. Metab 102: 3683–3691, 2017

AFFILIATIONS

1Division of Nephrology, Department of Medicine, Icahn School of Medicine at Mount Sinai, New York, New York; 2Department of Nephrology, State Key Laboratory of Kidney Diseases, National Clinical Research Center of Kidney Diseases, Chinese PLA General Hospital, Chinese PLA Institute of Nephrology, Beijing, China; 3Department of Nephrology, The First Afﬁliated Hospital of Xiamen University, Xiamen, China; 4Division of Nephrology, University of Michigan, Ann Arbor, Michigan; 5Department of Pathology, Columbia University Medical Center, New York, New York; and 6Renal Section, James J. Peters Veterans Affair Medical Center, Bronx, New York

J Am Soc Nephrol 30: ccc–ccc, 2019 LRG1 and Diabetic Kidney Disease Progression 17 Supplemental Material

LRG1 promotes DKD progression by enhancing TGF-β-induced angiogenesis

Quan Hong, Lu Zhang, Jia Fu, Divya A. Verghese, Kinsuk Chauhan, Girish N. Nadkarni, Zhengzhe Li, Wenjun Ju, Matthias Kretzler, Guang-Yan Cai, Xiang-Mei Chen, Vivette D. D’Agati, Steven G. Coca, Detlef Schlondorff, John C. He, and Kyung Lee

Supplemental Table of Contents

Supplemental Figures 1-5 Suppl. Fig. 1: RNA-seq expression of mouse Lrg1 mRNA in isolated glomeruli, podocytes, or GECs. Suppl. Fig. 2: Additional characterization of diabetic and control Lrg1+/+ and Lrg1-/- mice at 12 weeks post-DM. Suppl. Fig. 3: Overexpression and knockdown of LRG1 in mGECs. Suppl. Fig. 4: Additional characterization of diabetic and control Lrg1+/+ and Lrg1-/- mice at 20 weeks post-DM. Suppl. Fig. 5: Flow diagram of patient selection into study cohort.

Supplemental Tables 1-2 Suppl. Table 1: Blood pressure measurements of diabetic and control mice. Suppl. Table 2: Baseline characteristics of DKD patients.

Supplemental Methods and References

Supplemental Video Files 1-6 Suppl. File 1: Two-photon image of EYFP+ cells in control Lrg1+/+ mouse glomerulus (12w) Suppl. File 2: Two-photon image of EYFP+ cells in control Lrg1-/- mouse glomerulus (12w) Suppl. File 3: Two-photon image of EYFP+ cells in diabetic Lrg1+/+ mouse glomerulus (12w) Suppl. File 4: Two-photon image of EYFP+ cells in diabetic Lrg1-/- mouse glomerulus (12w) Suppl. File 5: Two-photon image of EYFP+ cells in diabetic Lrg1+/+ mouse glomerulus (20w) Suppl. File 6: Two-photon image of EYFP+ cells in diabetic Lrg1-/- mouse glomerulus (20w)

Supplemental Figures

30 Lrg1

RPKM 10

Glom_con Glom_STZ Podo_con Podo_STZ Endo_con Endo_STZ

B 120 Acvrl1 (Alk1) Tgfbr1 (Alk5) 80

RPKM 40

Glom_con Glom_STZ Podo_con Podo_STZ Endo_con Endo_STZ

Suppl. Fig. 1: RNA-seq expression of mouse Lrg1 mRNA in isolated glomeruli, podocytes, or GECs. mRNA

sequencing data (combined data from Fu et al. 2016 and Fu et al. 2018) shows the normalized expression of (A)

Lrg1 and (B) Alk1 and Alk5 mRNA in isolated glomeruli (Glom), sorted podocytes, and sorted glomerular

endothelial cells (GECs) from STZ-induced diabetic eNOS-/- (STZ) and nondiabetic eNOS-/- (con) mice. n=3-4 samples per group, where each sample is pooled from 3-4 mice. RPKM, reads per kilobase million. A

30 +/+ Lrg1 -STZ -/- 25 Lrg1

+/+ 20 Lrg1 +STZ Lrg1-/- 15

Body weight (g) 10

5 0 2 4 6 8 10 12 Weeks

B Lrg1+/+ Lrg1-/-

CD31/Ki-67 DAPI 0.15 +Vehicle *** 0.10

### ** 0.05

+STZ Ki67+ cells/glom cells 0.00

Suppl. Fig. 2: Additional characterization of diabetic and control Lrg1+/+ and Lrg1-/- mice at 12 weeks post- DM. (A) Body weight measurements in control and diabetic mice. (B) Ki-67/CD31 immunofluorescence on kidney sections of control and diabetic mice. Scale bar, 20μm. Quantification of Ki-67+ cell per total glomerular cells per mouse is shown on the right (n=5 mice per group, 20-30 glomeruli per mouse). **P<0.01 and ***P<0.001 vs. nondiabetic controls; ###P<0.001 vs. diabetic Lrg1+/+ mice. A pHR pHR-LRG1-V5

4 p<0.01 pHR pHR-LRG1-V5 LRG1 3

V5 2

Fold change in 1 GAPDH LRG1 expression 0

p<0.01 shScr 1.5 shLrg1#202 B shScr shLrg1#202

1.0 LRG1 0.5 Fold change in

LRG1 expression GAPDH 0.0

Suppl. Fig. 3: Overexpression and knockdown of LRG1 in mGECs. (A) Immortalized mouse GECs (mGECs) was transduced with either empty pHR lentivector or with pHR vector expressing V5 tagged LRG1 (pHR-LRG1-V5). Lysates from stably tranduced cells were subjected to western blot analysis using LRG1 and V5 antibodies. GAPDH was used as loading control. Representative blot is shown on the left and quantification of the western blots are shown on the right. (B) mGECs were transduced with pLKO lentivector expressing scrambled shRNA (shScr) or shRNA against LRG1 (shLrg1#202). Representative blot is shown on the left and quantification of the western blots are shown on the right.

A 500 800 Lrg1+/+ -/- +STZ 400500 Lrg1 600 Lrg1+/+ -/- +STZ g/mg) 300400 Lrg1

µ ***

400 g/mg) 200300 µ ***

200 UACR ( 100200 +/+ Lrg1 -STZ 500 Lrg1-/-

UACR ( +/+ Blood glucose (mg/dL) 1000 Lrg1+/+ -/- +STZ-STZ 0 4000 2 4 6 8 10 12 Lrg1-/- 0 5 10 15 20 0 Weeks g/mg) 3000 2 Weeks4 6 8 10 12

µ *** Weeks B 200 35 UACR ( 100 +/+ 30 Lrg1 -STZ Lrg1-/- 5000 25 0 2 4 6 8 10 12 +/+ Lrg1 +STZ 400 Weeks Lrg1-/- 20

g/mg) 300

µ *** Body weight (g) 15 200 10

UACR ( 0 100 5 10 15 20 +/+ Lrg1 -STZ Weeks Lrg1-/- 0 C 0 2 4 6 8 10 12 1825 *** WeeksLrg1+/+ -/- -STZ *** **# Lrg1 20 Lrg1+/+ 12 +STZ 15 Lrg1-/-

10 6 5 Kidney/BW (mg/g) Kidney/BW (mg/g) 0

+/+ -/- Suppl. Fig. 4: Additional characterization of diabetic and control Lrg1 and Lrg1 mice at 20 weeks post- DM. (A-C) Blood glucose (A), body weight (B), and kidney-to-body weight (C) measurements in control and diabetic mice. **P<0.01 and ***P<0.001 when compared with respective nondiabetic control mice. Random Selection of Type 2 Diabetes participants in BioMe [N=996]

- Missing Baseline eGFR (Mean of +/- 1 year eGFRs from Biobank enrollment date) value [N= 23] - Missing Follow-up eGFR values after Biobank enrollment date [N= 28] - Baseline eGFR < 30 [N= 12] - First Dialysis date before Biobank enrollment date (First dialysis date were taken from USRDS data) [N= 17] - Follow-up eGFR values are not available between enrollment date and last date [N= 2] - Only one eGFR value after biobank enrollment date (no follow up values to compare decline in eGFR) [N= 43]

N= 871

40% sustained decline in eGFR No decline in eGFR or ESRD [N=121] [N=750]

Suppl. Fig. 5: Flow diagram of patient selection into study cohort. Supplemental Tables

12 weeks post-injection 20 weeks post-injection

Vehicle-Lrg1+/+ Vehicle-Lrg1-/- STZ-Lrg1+/+ STZ-Lrg1-/- STZ-Lrg1+/+ STZ-Lrg1-/-

SBP 103.3±14.6 95.73±6.4 117.1±15.4 107.5±12.3 100.0±7.4 98.4±4.9 (mean±SEM) DBP 56.6±8.9 72.87±5.5 62.93±12.2 83.8±12.7 70.5±6.5 77.6±4.2 (mean±SEM)

Suppl. Table 1: Blood pressure measurements of diabetic and control mice. Blood pressure was measured by tail-cuff blood pressure monitor at 12 or 20 weeks post vehicle or STZ injections.

Overall Without Renal Endpoint With Renal Endpoint p value (n= 871) (n= 750) (n= 121)

Clinical Characteristics Age in years, Median (IQR) 60 [53 - 66] 60 [53 - 66] 61 [51 - 66] 0.64 Female, n (%) 507 (58.2) 434 (57.9) 73 (60.3) 0.62 Body Mass Index in kg/m2, Median [IQR] 30.9 [26.6 - 36.1] 31.0 [26.9 - 36.2] 29.7 [24.7 - 35.5] 0.28 Race, n (%) 0.09 White 62 (71.2) 59 (7.9) 3 (2.5) African-American 346 (39.7) 290 (38.7) 56 (46.3) Hispanic/Latino 412 (47.3) 355 (47.3) 57 (47.1) Other 51 (58.6) 46 (6.1) 5 (4.1) Hypertension, n (%) 813 (93.3) 694 (92.5) 119 (98.4) 0.02 Coronary Artery Disease, n (%) 432 (49.6) 392 (52.3) 40 (33.1) <0.001 Heart Failure, n (%) 192 (22.0) 144 (19.2) 48 (39.7) <0.001 Systolic Blood Pressure in mm Hg, Median 131.9 [123.7 - 143.4] 131.2 [123.5 - 141.5] 140.3 [125.1 - 150.1] 0.01 [IQR] Diastolic Blood Pressure in mm Hg, Median 73.4 [68.7 - 79.3] 73.2 [68.5 - 78.9] 75.2 [70.6 - 81.0] 0.04 [IQR] Follow up Time in years, Median [IQR] 4.5 [3.3 - 6.1] 4.4 [3.2 - 5.9] 5.2 [3.8 - 6.8] 0.01 Laboratory Characteristics Baseline eGFR, Median [IQR} 68.4 [55.3 - 80.0] 69.0 [56.3 - 80.3] 62.6 [50.5 - 78.0] 0.02 Baseline Hemoglobin A1C, Median [IQR] 7.0 [6.2 - 8.7] 6.9 [6.2 - 8.5] 7.6 [6.3 - 9.7] 0.01 Baseline urine albumin/creatinine, Median 13.0 [4.0 - 66.3] 11.0 [4.0 - 38.0] 204.0 [25.0 - 1240.0] <0.001 [IQR] Medications ACE/ARB, n (%) 675 (77.5) 575 (76.7) 100 (82.6) 0.15 Plasma LRG1 Concentrations LRG1, in ng/ml, Median [IQR] 57.0 [33.8 – 105.5] 53.8 [32.8 – 96.9] 80.7 [53.2 – 172.1] <0.0001

Suppl. Table 2: Baseline characteristics. Values are presented as mean (SD) for normally distributed continuous values, median (interquartile range) for skewed continuous values, and N (%) for categorical values. Renal Endpoint is defined as ESRD (defined by the initiation of maintenance dialysis or receipt of kidney transplant) or a sustained (on two or more time intervals ≥3 months apart) decline in eGFR of ≥40% from baseline eGFR.

Supplemental Methods

Human archival kidney samples Human kidney biopsy samples were obtained from archival biopsy specimens collected at Icahn School of Medicine at Mount Sinai (ISMMS) under a protocol approved by the Institutional Review Board at ISMMS.

RNA in situ hybridization Mouse kidneys (dual ISH): Lrg1 mRNA expression was assessed by RNAscope 2.5 HD Duplex Reagent Kit (Advanced Cell Diagnostics, Hayward, CA; #322430) according to the manufacturer’s instructions. Briefly, deparaffinized 4-µm FFPE kidney sections were pre-treated with RNAscope 2.5 Universal Pretreatment reagent at 100°C for 15 minutes, rinsed in deionized water, and treated with Protease Plus reagent at 40°C for 30 minutes in a HybEZ Hybridization System. mLrg1 (#423381) and mPecam1-C2 (#316721-C2) probes were then hybridized for 2 hours at 40 °C in a hybridization oven. After amplification, chromogenic substrate was added for the visualization of the target mRNA. Freshly cut sections of same batch of FFPE specimens were used in parallel with positive (mPolr2a probe, #321651) and negative (bacterial DapB probe, #320751) controls. The specimens which exhibited negative staining with a negative control probe (DapB) and grade 4 + or higher staining with the positive control (mPolr2a) probe, were considered adequate for mLrg1 analysis. Two independent observers blinded to mouse genotypes evaluated the FFPE slides for mLrg1 expression using five graded categories (0-4), according to the manufacturer’s guidelines.

Human kidneys (dual ISH-IF): Deparaffinized FFPE kidney sections were processed similarly as above using hLRG1 probe (#439311) with RNAscope 2.5 HD Assay-RED (#439311). Sections were then subjected to immunofluorescence with CD31 antibody (Abbiotec, 250590). hLRG1 probe was detected with a filter set at Ex554nm/Em576nm, and CD31 was detected using the Alexa Fluor 488-conjugated secondary antibody.

Measurements of urinary albumin and BUN Urine albumin was quantified using the ELISA kit from Bethyl Laboratories, Inc. (Houston, TX). Urine creatinine levels were measured in the same samples using the Creatinine Colorimetric Assay Kit (Cayman, MI, USA) according to the manufacturer’s instructions. 24-hour urine collections in the metabolic cages were also used for determination of urinary albumin excretion. Blood urea nitrogen (BUN) levels were measured using a commercially available kit according to manufacturer’s protocol (Bioassay Systems, Hayward, CA).

Kidney histologic and morphometric analyses Kidney samples were fixed in 10% formalin, embedded in paraffin and sectioned to 4-µm thickness. Sections were stained with periodic acid–Schiff (PAS) for analysis of glomerular area and mesangial matrix expansion. The mean glomerular tuft volume was determined from the mean glomerular cross-sectional area by light microscopy. The glomerular tuft volume was calculated using the following equation: GV=(β⁄κ) × GA3/2 , where β=1.38, the shape coefficient of spheres (the idealized shape of glomeruli), and κ=1.1, the size distribution coefficient. Mesangial expansion was defined as a periodic acid–Schiff-positive and nuclei-free area in the mesangium. Quantification of mesangial expansion was based on a minimum of 10 glomeruli per section in a blinded fashion, under 400x magnification (Zeiss AX10 microscope, Carl Zeiss Canada Ltd, Toronto, ON, Canada). For immunofluorescence staining, kidney samples were frozen in OCT embedding compound. For transmission electron microscopy, kidney cortex samples fixed in 2.5% glutaraldehyde were sectioned and mounted on a copper grid; then images were photographed using a Hitachi H7650 microscope (Tokyo, Japan). Briefly, digitized images were scanned and profile areas were traced using ImageJ. The quantification of podocyte foot process width was determined as previously described 1.

Immunofluorescence Frozen kidney sections were used for immunofluorescence staining as previously described 2 3 using CD31 (BD Biosciences Inc., #BDB550274) and phospho-Smad1/5/8 (Cell Signaling Technologies, #9516) primary antibodies. Fluorescent images were taken using the Zeiss Axioplan 2 IE microscope.

Directed in vivo angiogenesis angioreactor (DIVAA) In vivo angiogenesis was assessed using the Directed In Vivo Angiogenesis Assay (DIVAA) kit from Trevigen (Gaithersburg, MD; #3450-048-K) according to manufacturer’s protocol. Briefly, at 7 weeks post diabetes induction mice were anesthetized with isoflurane, and the angioreactors were subcutaneously implanted into the dorsal flank. Angiogenesis was measured as the amount of FITC-Lectin detected 14 days postimplantation. Amount of fluorescence in each sample was collected using the SpectraMax M3 plate reader (Molecular Devices, Sunnyvale CA).

Quantification of EFYP+ cells with two-photon microscopy 4% formaldehyde-perfused kidneys were cut in 500µm sections with the Vibratome (Leica VT1200, Germany), and processed as described in Puelles et al.4 Images from each sections were obtained with Leica TCP-SP5 Two-Photon Microscope (Leica Microsystems) equipped with a 20x and 0.95 numerical aperture oil–immersion objective and Chameleon Laser (Coherent, Santa Clara, CA). Fluorescence emission was guided directly to supersensitive external hybrid photodetectors (Leica/Hammamatsu). Images were reconstructed to 3D view, and EYFP+ cells were quantified using the Imaris software (Bitplane AG, CH-8048 Zurich, Switzerland).

Overexexpression and knockdown of LRG1 in mGECs Based on mouse Lrg1 NCBI reference sequence (NM_029796.2), full length Lrg1 cDNA tagged with V5 tag was cloned into pGEM-Teasy vector with the following primers: 5’-TTA AGA TCT GGA TCC GCC ACC ATG GTC TCT TGG CAG CAT CAA GGA AG-3’, 5’-AAC CGG TGG ATC CTT ATG TAG AAT CAA GTC CCA GAA GAG GAT TTG GAA TAG GTT TTC CCA GGG ACC CCA GCT CTG CCA C-3’, then subcloned into BamH I and Sma I restriction enzyme site of pHR-CMV-MCS-IRES-GFP-delta B lentivirus vector 5, named pHR-LRG1- V5. Together with psPAX2 packaging and VSVG envelope plasmids, pHR-LRG1-V5 plasmid or empty pHR-V5 control plasmid was transfected in human embryonic kidney (HEK)-293T cells using PolyJet transfection reagent (SignaGen Laboratories, Rockville, MD) according to the manufacturer’s manual. Collection of supernatant 48 hours post-transfection was harvested and concentrated. Lentivectors expressing short hairpin RNA (shRNA) for mouse Lrg1 (pLKO.1-shLRG1; 4 independent clones) and scrambled shRNA (pLKO.1- shScr) were obtained from Sigma-Aldrich (St. Louis, MO), as described previously 6.

Western Blot Cells were homogenized in NP-40 lysis buffer containing protease and phosphatase inhibitor cocktail. Equal amounts of protein samples were separated on SDS polyacrylamide gel, transferred to PVDF membranes (Millipore) and probed with primary antibodies (~1:1000). Membranes were then washed with PBST and incubated with a secondary antibody (horseradish peroxidase conjugated antibodies to mouse IgG or to rabbit IgG, ~1;10,000). Blots were developed with the enhanced chemi-luminescence system and imaged with Odyssey FC Imager (LI-COR Biosciences, NE).

Supplemental References

1. Hong, Q, Zhang, L, Das, B, Li, Z, Liu, B, Cai, G, Chen, X, Chuang, PY, He, JC, Lee, K: Increased podocyte Sirtuin-1 function attenuates diabetic kidney injury. Kidney Int, 93: 1330-1343, 2018.

2. Fu, J, Wei, C, Lee, K, Zhang, W, He, W, Chuang, P, Liu, Z, He, JC: Comparison of Glomerular and Podocyte mRNA Profiles in Streptozotocin-Induced Diabetes. J Am Soc Nephrol, 27: 1006-1014, 2016.

3. Fan, Y, Li, X, Xiao, W, Fu, J, Harris, RC, Lindenmeyer, M, Cohen, CD, Guillot, N, Baron, MH, Wang, N, Lee, K, He, JC, Schlondorff, D, Chuang, PY: BAMBI elimination enhances alternative TGF-beta signaling and glomerular dysfunction in diabetic mice. Diabetes, 64: 2220-2233, 2015.

4. Puelles, VG, van der Wolde, JW, Schulze, KE, Short, KM, Wong, MN, Bensley, JG, Cullen-McEwen, LA, Caruana, G, Hokke, SN, Li, J, Firth, SD, Harper, IS, Nikolic-Paterson, DJ, Bertram, JF: Validation of a Three-Dimensional Method for Counting and Sizing Podocytes in Whole Glomeruli. J Am Soc Nephrol, 27: 3093-3104, 2016.

5. Liu, R, Zhong, Y, Li, X, Chen, H, Jim, B, Zhou, MM, Chuang, PY, He, JC: Role of transcription factor acetylation in diabetic kidney disease. Diabetes, 63: 2440-2453, 2014.

6. Fu, J, Wei, C, Zhang, W, Schlondorff, D, Wu, J, Cai, M, He, W, Baron, MH, Chuang, PY, Liu, Z, He, JC, Lee, K: Gene expression profiles of glomerular endothelial cells support their role in the glomerulopathy of diabetic mice. Kidney Int, 2018.