The sequelae of diabetes include microvascular complications such as diabetic kidney disease (DKD), which involves glucose-mediated renal injury associated with a disruption in mitochondrial metabolic agility, inflammation, and fibrosis. We explored the role of the innate immune complement component C5a, a potent mediator of inflammation, in the pathogenesis of DKD in clinical and experimental diabetes. Marked systemic elevation in C5a activity was demonstrated in patients with diabetes; conventional renoprotective agents did not therapeutically target this elevation. C5a and its receptor (C5aR1) were upregulated early in the disease process and prior to manifest kidney injury in several diverse rodent models of diabetes. Genetic deletion of C5aR1 in mice conferred protection against diabetes-induced renal injury. Transcriptomic profiling of kidney revealed diabetes-induced downregulation of pathways involved in mitochondrial fatty acid metabolism. Interrogation of the lipidomics signature revealed abnormal cardiolipin remodeling in diabetic kidneys, a cardinal sign of disrupted mitochondrial architecture and bioenergetics. In vivo delivery of an orally active inhibitor of C5aR1 (PMX53) reversed the phenotypic changes and normalized the renal mitochondrial fatty acid profile, cardiolipin remodeling, and citric acid cycle intermediates. In vitro exposure of human renal proximal tubular epithelial cells to C5a led to altered mitochondrial respiratory function and reactive oxygen species generation. These experiments provide evidence for a pivotal role of the C5a/C5aR1 axis in propagating renal injury in the development of DKD by disrupting mitochondrial agility, thereby establishing a new immunometabolic signaling pathway in DKD.

Diabetic kidney disease (DKD), affecting up to 30% of both patients with type 1 diabetes (T1D) and patients with type 2 diabetes (T2D), is the leading cause of end-stage renal disease in Western societies (1,2). Despite optimal conventional management by pharmacologically inhibiting the renin angiotensin system (RAS) and controlling glycemia and blood pressure, a significant proportion of patients with DKD still progress over time to end-stage renal failure. Thus, identifying new molecular pathways with an opportunity for therapeutic targeting in order to slow, halt, or potentially reverse DKD progression would address a main unmet need.

The diabetic kidney is exposed to persistent metabolic and hemodynamic stressors; this exposure results in cellular injury and activation of the innate immune response, including the complement system (3). The complement system is a highly sophisticated network of proteins that are activated in response to invading pathogens or tissue injury. Complement homeostasis is finely balanced, and when subject to states of dysregulation or hyperactivation, it can propagate a severe inflammatory response (4). Complement comprises three mechanistic pathways: classical, lectin, and alternative. Activation of any one of these commonly converges to the production of the complement C3 and C5 convertases. These enzymes cleave complement substrates C3 and C5, respectively, resulting in the generation of the opsonic C3b and anaphylatoxins C3a and C5a, and subsequently the formation of the membrane attack complex (MAC; comprising C5b-9). As the terminal effector component of complement cascade, the MAC lyses, damages, or activates target cells to drive inflammation (5). All complement activation pathways result in the formation of C5a, a main potent effector molecule of complement, which, via ligation with the receptor C5aR1, initiates and propagates pathology in inflammatory disease states (6). The role of the C5a-C5aR1 signaling cascade has been implicated in a number of kidney diseases in both animal models (710) and humans (11); however, its role in DKD-associated renal injury is poorly defined.

Here, we report that complement is hyperactivated in both patients with T1D and patients with T2D, and is not targeted by conventional therapies that modify the trajectory of DKD, such as RAS inhibitors. C5a was generated and C5aR1 upregulated in the kidney in several experimental models of diabetes, before the onset of albuminuria. Inhibition of C5aR1 via genetic deletion or pharmacological targeting with PMX53 improved albuminuria and renal injury in streptozotocin (STZ)-induced diabetic mice. Interrogation of the lipidomics signature revealed abnormal cardiolipin remodeling in the diabetic kidney, a cardinal sign of disrupted mitochondrial architecture and bioenergetics. Inhibition of the C5a/C5aR axis restored the renal mitochondrial fatty acid profile, most notably cardiolipin, and targeted metabolomics showed normalized citric acid cycle intermediates. Proof-of-concept studies in human primary proximal tubule cells showed that C5aR1 signaling disrupts mitochondrial respiratory function and induces reactive oxygen species generation. Taken together, our results show that the C5a/C5aR1 axis propagates renal injury in DKD via mitochondrial reprogramming and establish complement C5a as a new immunometabolic signaling pathway in DKD.

Patient Recruitment and Blood Sample Collection

The study was approved by the Human Research Ethics Committee at Austin Health and Alfred Health and was conducted in accordance with the principles of the Declaration of Helsinki. Participating in this study were individuals with T1D or T2D who were drawn from a population of patients attending the diabetes clinic at the Austin and Repatriation Medical Centre, Melbourne, Australia. Because the primary referral base (80%) is from general practitioners, with only 20% referred from within hospitals, the cohort is representative of patients with diabetes in the wider community. Patient recruitment methods have been described previously (12). A control group without diabetes was also recruited from Austin Health and the Baker Heart and Diabetes Institute (Alfred Health). Written informed consent was obtained from all participants who donated blood. EDTA plasma was collected by centrifuging blood at 3,500 rpm and 4°C for 15 min and storing it at −80°C.

Assessment of Complement Components in Human Plasma

Commercially available ELISA kits were used to determine plasma C5a (BD OptEIA), C3a (BD OptEIA), and C5b-9 (MicroVue Quidel) according to the manufacturer’s instructions.

Animal Experiments

All animal experiments were performed in accordance with guidelines from the Alfred Medical Research and Education Precinct (AMREP) Animal Ethics Committee and the National Health and Medical Research Council of Australia. All rodents were housed in a temperature-controlled environment, with a 12-h light/12-h dark cycle, and had access to chow (Specialty Feeds, Perth, Western Australia, Australia) and water ad libitum. C57BL/6J and C5aR1−/− mice backcrossed onto a C57BL/6 background (gift from Professor Rick Wetsel, University of Texas) were bred at AMREP Animal Services (13). Ins2-Akita mice (C57BL/6-Ins2Akita/J) were purchased from The Jackson Laboratory (Bar Harbor, ME). Six-week-old heterozygous Ins2-Akita mice and their wild-type (WT) littermates were followed for 20 weeks. db/db mice (lepr(+/+)C57 BL/KsJ) and db/h controls were purchased from The Jackson Laboratory, and a colony was maintained at AMREP Animal Services. Six-week-old db/db and db/h mice were followed for 14 weeks. Sprague-Dawley rats were sourced from AMREP Animal Services. Experimental diabetes was induced in 6-week-old male Sprague-Dawley rats (weight, 200–250 g) by injection of STZ (50 mg/kg i.v.; sodium citrate buffer [pH 4.5]) after they had been deprived of food overnight. A rat time course model has been previously described (14).

Diabetes was induced in 6-week-old mice by five daily intraperitoneal injections of low-dose STZ (55 mg/kg) (Sigma-Aldrich, St. Louis, MO). Mice in the nondiabetic group were given 0.5 M sodium citrate. For the knockout study, diabetic and nondiabetic WT mice (n = 7–10) and C5aR1−/− mice (n = 4–12) were followed for 24 weeks after the onset of diabetes. For the PMX53 study, diabetic and nondiabetic mice were randomized to receive either 1) the C5aR1 peptide inhibitor PMX53 (Ac-Phe-[Orn-ProdCha-Trp-Arg]; synthesized as previously described [15]) at 2 mg/kg body weight, administered in the drinking water, or 2) drinking water alone (n = 6–12 mice). All mice were followed for 24 weeks. At the end of the study, plasma and kidneys were collected for analysis.

Assessment of Renal Function and Metabolic Parameters

Plasma glucose was measured using a glucose colorimetric assay kit (Cayman Chemical). We determined glycated hemoglobin using a Cobas Integra 400 Autoanalyzer (Roche Diagnostics Corp.). Urinary albumin was measured using a mouse albumin ELISA kit (Bethyl Laboratories, Montgomery, TX). Plasma cystatin C was determined using a commercially available ELISA kit from R&D Systems. Urinary C5a was measured using a mouse Complement Component C5a DuoSet ELISA kit (R&D Systems), urinary 8-isoprostane using an 8-isoprostane ELISA kit (Cayman Chemical), and urine and plasma creatinine using the Creatinine plus ver.2 (CREP2) on a Cobas Integra 400 plus analyzer (Roche Diagnostics Corp.). Plasma interleukin (IL)-18 was measured using a mouse IL-18 ELISA kit (Invitrogen, Carlsbad, CA).

Renal Histology

Kidney sections (3 µm thick) were stained with periodic-acid Schiff (PAS) and picrosirius red. For PAS-stained sections, we subjectively graded the degree of sclerosis in each glomerulus on a scale of 0–4: 0 = normal, 1 = sclerotic area up to 25% (minimal), 2 = sclerotic area 25–50% (moderate), 3 = sclerotic area 50–70% (moderate to severe), and 4 = sclerotic area 75–100% (severe). We then calculated the glomerulosclerotic index (GSI) using the following equation:

formula

where nx is the number of glomeruli in each grade of glomerulosclerosis. We analyzed the mesangial index from digital images of glomeruli using Image-Pro Plus software version 6.0 (Media Cybernetics, Bethesda, MD); this index is expressed as the percentage of PAS-stained area per glomerular cross-sectional area. For picrosirius red–stained sections, positive collagen staining (red) was examined under an Olympus BX-50 light microscope (Olympus Optical) and digitized with a high-resolution camera. All digital quantitation (Image-Pro Plus software version 6.0) and assessments were performed with the assessors blinded to experimental groups.

Immunohistochemistry

Paraffin sections of mouse kidney (4 µm thick) were immunostained for Forkhead box P3 (FoxP3; clone, FJK-16s; Affymetrix eBioscience), F4/80 (clone CI:A3–1; Abcam), or collagen IV (Southern Biotech). Briefly, endogenous peroxidases were blocked with 3% hydrogen peroxide for 15 min and incubated in 0.5% skim milk/Tris-buffered saline for 1 h at room temperature. The primary antibody was left on overnight at 4°C. This was followed by incubation with biotinylated secondary antibody at room temperature for 10 min. Sections were then incubated with Vectastain ABC reagent (Vector Laboratories). Peroxidase activity was identified by reaction with 3,3′-diaminobenzidine tetrahydrochloride (Sigma-Aldrich Pty. Ltd., New South Wales, Australia). Sections were counterstained with hematoxylin. All sections were examined under an Olympus BX-50 light microscope (Olympus Optical) and digitized with a high-resolution camera. All digital quantitation (Image-Pro Plus software version 6.0) and assessments were performed in a blinded manner.

Quantitative Real-time RT-PCR

RNA from renal cortex and liver was extracted using TRIzol Reagent, and cDNA was synthesized as described previously (16). Gene expression was determined using a 7500 Fast Real-Time PCR System (Applied Biosystems, Victoria, Australia). Gene expression was normalized relative to 18S rRNA, and the relative fold difference in expression was calculated using the comparative 2−ΔΔCt method.

RNA Sequencing and Analysis

Approximately 200 ng of total RNA underwent RNA depletion using the NEBNext rRNA Depletion Kit followed by library construction using an NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (both from New England BioLabs Inc.). Barcoded libraries underwent Illumina 100 cycle single-read sequencing at the Australian Genome Research Facility, Melbourne, using HiSeq version 4 reagents. Reads underwent 3′ trimming in Skewer in order to remove bases with quality less than Phred 20 (17), and the reads were mapped to the Ensembl mouse genome (GRCm38) by using STAR (18). The resulting count matrix underwent differential analysis with the EdgeR package (19). Genes were ranked from most upregulated to most downregulated by multiplying the sign of the log2 fold change by the inverse of the P value. This preranked list was used for pathway analysis using gene set enrichment analysis (20) with Reactome gene sets (21). Multicontrast enrichment was undertaken as described previously (22).

Lipidomics

Renal cortical tissues were homogenized and sonicated in PBS (pH 7.4). Protein concentration was determined using a bicinchoninic acid assay (Thermo Scientific). Lipid extraction was performed as previously described (23). Lipidomic analyses of the lipid extract were performed by liquid chromatography electrospray ionization tandem mass spectrometry as described by Huynh et al. (24) with modifications, using an AB Sciex Qtrap_4000 coupled to an Agilent 1200 HPLC system. Free fatty acids were measured under the same chromatographic conditions but in negative ionization mode. Data were analyzed using MultiQuant software 2.1.1. Some lipid species were also normalized to total phosphatidylcholine levels. Some lipid species were also normalized to total protein as an alternate normalization strategy.

Metabolomics

Tissue metabolites from the renal cortex were extracted using a cryogenically cooled bead mill (25). Metabolite extracts were derivatized and analyzed by gas chromatography–mass spectrometry (26). The resultant data matrix of integrated metabolite areas was log-transformed and median normalized prior to statistical analysis by using MetaboAnalyst software version 4.0 or GraphPad Prism software version 7.0 (GraphPad Software, La Jolla, CA).

Proximal Tubule Cell Culture and Mitochondrial Respiratory Function

Human primary proximal tubule endothelial cells (PTECs) obtained from ATCC were maintained in Renal Epithelial Cell Basal Medium (ATCC) supplemented with the Renal Epithelial Cell Growth Kit (ATCC). PTECs, at 70–90% confluence, were seeded into 96-well XF96 cell culture microplates (Seahorse Bioscience, North Billerica, MA) at 22,000 cells per well and then left to recover in Renal Epithelial Cell Basal Medium for 24 h. Cells were washed and incubated at 37°C in DMEM containing either 500 ng/mL recombinant human C5a (R&D Systems) or 500 ng/mL C5a with 2.5 μmol/L PMX53 (pH 7.4) for 24 h. The mitochondrial bioenergetic profile of human primary PTECs was assessed using a Seahorse XF96 Flux Analyzer (Seahorse Bioscience), applying the mitochondrial stress test. Three basal oxygen consumption rate measurements were performed, and these were then repeated after sequential exposure of the PTECs to the ATP synthase inhibitor oligomycin (1 μmol/L), the proton ionophore carbonyl cyanide-4-(trifluoromethoxy) phenylhydrazone (FCCP; 0.75 μmol/L), and the complex III inhibitor antimycin A (2 μmol/L) plus rotenone (0.5 μmol/L). Calculation of mitochondrial respiratory function was then completed (27). At the end of the assay, cells were lysed and protein concentration was determined in each well using a bicinchoninic acid protein assay kit (Thermo Scientific) according to the manufacturer’s instructions.

Reactive Oxygen Species Generation

Human primary PTECs, described above, were cultured in T25 cell culture flasks (1 × 106 cells) and were exposed to 20 nmol/L human C5a (Complement Technology) or 20 nmol/L C5a plus 2.5 μmol/L PMX53 for 24 h. After harvesting and resuspending in Hanks’ balanced salt solution (with sodium bicarbonate, calcium, magnesium, and 10 mmol/L HEPES, but without phenol red [pH 7.4]), cells were incubated with 10 μmol/L 5′,6′-chloromethyl-2′,7′dichlorodihydro-fluorescein diacetate (Molecular Probes, Eugene, OR) at 37°C in the dark for 30 min. The cells were washed twice with Hanks’ balanced salt solution and analyzed on a FACSCalibur flow cytometer (BD Biosciences). We analyzed a minimum of 15,000 cells per sample.

Statistical Analyses

The data are expressed either as scatterplots of the data showing the mean, or as mean ± SEM. Statistical analyses were performed using GraphPad Prism software version 7.0 (GraphPad Software). For the clinical studies, one-way ANOVA with Tukey posttest analysis was used to determine statistical significance. Data not normally distributed were analyzed after logarithmic transformation. For the C5aR1 knockout study, all data were analyzed with two-way ANOVA with the Tukey post hoc test, unless otherwise stated. For the PMX53 study, all data were analyzed with one-way ANOVA with the Tukey post hoc test, unless otherwise stated. We performed between-group comparisons using the two-tailed Student t test. If data were nonparametric, we used a Mann-Whitney U test. A P value <0.05 was considered statistically significant. We used Pearson correlation to determine relationships between variables. Lipidomics data were analyzed using the R package (3.4.0), and all P values were corrected for multiple comparisons using the Benjamini-Hochberg method, controlling for the false discovery rate (FDR) (28).

Data and Resource Availability

Sequence data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus with accession number GSE118089.

Both Type 1 and Type 2 Diabetes Are Associated With Activation of the Distal Pathway of Complement, Which Is Not Targeted by Conventional DKD Trajectory–Modifying Therapies

We explored whether the terminal complement components, including the main effector molecules C3a and C5a, and the terminal complement component C5b-9 (MAC), were altered in T1D and in T2D. We found markedly elevated levels of C5a, C3a, and C5b-9 in plasma from normoalbuminuric patients with T1D or T2D (Fig. 1A–C and Supplementary Table 1). These data clearly indicate that the C5aR1 axis is activated in human diabetes prior to DKD development. Since the first-line therapy for patients with DKD is RAS inhibition, we determined whether this commonly used therapy could affect complement activation. The elevated levels of plasma C5a, C3a, and C5b-9 in patients with diabetes was not normalized in subjects in whom RAS was inhibited (Fig. 1A–C and Supplementary Table 1). In patients with diabetes, we found a correlation between plasma C5a and urine albumin (r = 0.25); however, this correlation did not reach significance (P = 0.09) (Supplementary Fig. 5A). Further, we found a negative correlation between plasma C5a and eGFR (r = −0.32), which also did not reach significance (P = 0.06) (Supplementary Fig. 5B).

Figure 1

Classical clinical therapy does not target complement activation in patients with T1D or T2D. Plasma C5a (A), C3a (B), and C5b-9 (C) were measured in control individuals without diabetes (Con; n = 38), in patients with T1D (n = 18) or T2D (n = 20), and in patients with diabetes treated with RAS inhibitors (T1D+RAS, n = 19; T2D+RAS, n = 20). The line within each scatterplot represents the mean. Comparison between the groups was performed using one-way ANOVA followed by the Tukey test. ***P < 0.001 vs. Con.

Figure 1

Classical clinical therapy does not target complement activation in patients with T1D or T2D. Plasma C5a (A), C3a (B), and C5b-9 (C) were measured in control individuals without diabetes (Con; n = 38), in patients with T1D (n = 18) or T2D (n = 20), and in patients with diabetes treated with RAS inhibitors (T1D+RAS, n = 19; T2D+RAS, n = 20). The line within each scatterplot represents the mean. Comparison between the groups was performed using one-way ANOVA followed by the Tukey test. ***P < 0.001 vs. Con.

Close modal

C5a/C5aR1 Is Upregulated in Multiple Animal Models of Diabetes

To investigate whether complement is also activated in experimental diabetes, we evaluated the status of this axis in two disparate models: an insulin-deficient model (the Ins2-Akita mouse model, a genetic model of T1D) (Fig. 2A) and an insulin-resistant model (the db/db mouse model, a genetic model of T2D and obesity) (Fig. 2D). Indeed, C3, C5, and C5aR1 transcripts were upregulated in the kidney in both mouse models of diabetes (Fig. 2A and D). Because the liver is the main site of complement biosynthesis, hepatic complement levels largely reflect systemic complement (29). Within the liver, C3 and C5aR1 transcripts were upregulated in both the Ins2-Akita and db/db mice (Fig. 2B and E). In addition, urinary excretion of C5a was markedly increased (Fig. 2C and F), indicating that the C5a/C5aR1 axis is upregulated in experimental diabetes and DKD, consistent with the human condition.

Figure 2

The C5a/C5aR1 axis is upregulated in diverse mouse models of diabetes. C3, C5, and C5aR1 expression in the renal cortex (A) and liver (B) of Ins2-Akita mice (n = 8–20 mice per group). *P < 0.05, **P < 0.01, ****P < 0.0001 vs. WT. C: Urinary excretion of C5a in Ins2-Akita mice (n = 8–13 mice per group). **P < 0.01 vs. WT. C3, C5, and C5aR1 expression in the renal cortex (D) and liver (E) of db/db mice (n = 3–9 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs. db/h mice. F: Urinary excretion of C5a in db/db mice (n = 9–18 mice per group). ***P < 0.001 vs. db/h mice. STZ-induced diabetic rats were followed for 4, 8, 16, and 32 weeks; C5aR1 mRNA expression (G) and urinary C5a excretion (H) were determined (n = 6–10 rats per group). *P < 0.05, **P < 0.01 vs. Con at the same time point. The line within each scatterplot represents the mean.

Figure 2

The C5a/C5aR1 axis is upregulated in diverse mouse models of diabetes. C3, C5, and C5aR1 expression in the renal cortex (A) and liver (B) of Ins2-Akita mice (n = 8–20 mice per group). *P < 0.05, **P < 0.01, ****P < 0.0001 vs. WT. C: Urinary excretion of C5a in Ins2-Akita mice (n = 8–13 mice per group). **P < 0.01 vs. WT. C3, C5, and C5aR1 expression in the renal cortex (D) and liver (E) of db/db mice (n = 3–9 mice per group). *P < 0.05, **P < 0.01, ***P < 0.001 vs. db/h mice. F: Urinary excretion of C5a in db/db mice (n = 9–18 mice per group). ***P < 0.001 vs. db/h mice. STZ-induced diabetic rats were followed for 4, 8, 16, and 32 weeks; C5aR1 mRNA expression (G) and urinary C5a excretion (H) were determined (n = 6–10 rats per group). *P < 0.05, **P < 0.01 vs. Con at the same time point. The line within each scatterplot represents the mean.

Close modal

C5a/C5aR1 Upregulation in Diabetes Occurs Prior to the Development of Renal Injury

To determine the time course of C5a activation in the development of DKD, we mapped urinary C5a excretion and renal C5aR1 expression in a rat model of STZ-induced T1D. We previously characterized the onset of glomerular injury in STZ-induced diabetic rats (14). In diabetic rats, 4 weeks after the establishment of diabetes—preceding the onset of albuminuria (evident at 16 weeks [14])—C5aR1 expression significantly increased in the renal cortex (Fig. 2G), and we observed a trend toward an increase in urinary excretion of C5a, which reached significance at 16 weeks and persisted to 32 weeks (Fig. 2H). Taken together, these results suggest that the upregulation of C5a/C5aR1 occurs prior to development of detectable renal injury, consistent with complement activation playing an early key injurious role in DKD pathogenesis.

Genetic Deletion of C5aR1 Attenuates Diabetes-Induced Renal Injury

To assess the effect of genetic deletion of C5aR1 on the development of DKD, C5aR1−/− mice on a C57BL/6 background and their WT littermates were treated with STZ to induce diabetes. The metabolic characteristics of the mice are shown in Table 1. Plasma cystatin C was lower in diabetic WT mice and in diabetic C5aR1−/− mice than in control WT mice (Table 1). Although we found no statistically significant difference between control WT and control C5aR1−/− mice, control C5aR1−/− mice tended to exhibit hyperfiltration. Diabetic WT mice had 20-fold higher urinary excretion of albumin than did control mice (Fig. 3A), whereas deletion of C5aR1 reduced diabetes-induced albuminuria by >75% (Fig. 3A). Deletion of C5aR1 also inhibited oxidative stress in the setting of diabetes, as reflected by a decrease in the amount of urinary 8-isoprostanes (Fig. 3B).

Table 1

Basic characteristics of diabetic and nondiabetic mice after 24 weeks of diabetes

Wild-type miceC5aR1−/− mice
Nondiabetic controlDiabeticNondiabetic controlDiabetic
BW, g 33.5 ± 0.7 23.1 ± 1.3**** 38.0 ± 1.3* 23.3 ± 0.8†††† 
Plasma glucose, mg/dL 284 ± 19 629 ± 56** 329 ± 27 578 ± 73** 
HbA1c, % (mmol/mol) 4.30 ± 0.04 (23 ± 1) 13.30 ± 0.30**** (120 ± 3) 4.40 ± 0.20 (21 ± 1) 13.20 ± 0.60†††† (120 ± 7) 
CrCl, mL/min/m2 18.9 ± 2.5 30.6 ± 7.8 27.0 ± 8.6 26.3 ± 3.4 
Cystatin C, ng/mL 450.3 ± 23.4 284.9 ± 31.3*** 374.8 ± 24.1 306.0 ± 25.5** 
L.Kid:BW, mg/g 4.8 ± 0.2 11.6 ± 0.8**** 5.1 ± 0.3 10.8 ± 0.6†††† 
Wild-type miceC5aR1−/− mice
Nondiabetic controlDiabeticNondiabetic controlDiabetic
BW, g 33.5 ± 0.7 23.1 ± 1.3**** 38.0 ± 1.3* 23.3 ± 0.8†††† 
Plasma glucose, mg/dL 284 ± 19 629 ± 56** 329 ± 27 578 ± 73** 
HbA1c, % (mmol/mol) 4.30 ± 0.04 (23 ± 1) 13.30 ± 0.30**** (120 ± 3) 4.40 ± 0.20 (21 ± 1) 13.20 ± 0.60†††† (120 ± 7) 
CrCl, mL/min/m2 18.9 ± 2.5 30.6 ± 7.8 27.0 ± 8.6 26.3 ± 3.4 
Cystatin C, ng/mL 450.3 ± 23.4 284.9 ± 31.3*** 374.8 ± 24.1 306.0 ± 25.5** 
L.Kid:BW, mg/g 4.8 ± 0.2 11.6 ± 0.8**** 5.1 ± 0.3 10.8 ± 0.6†††† 

Data are mean ± SEM. BW, body weight; CrCl, creatinine clearance; L.Kid:BW, left kidney weight–to–body weight ratio.

*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 vs. control WT mice; ††††P < 0.0001 vs. C5aR1−/− nondiabetic controls.

Figure 3

Genetic deletion of C5aR1 protects against diabetes-induced inflammation and renal injury. Urinary albumin (A) and 8-isoprostane (B) were determined in control (Con) and diabetic (Diab) WT mice and in control and diabetic C5aR1−/− mice (n = 4–12 mice per group). C and E, top row: F4/80-positive cells in the renal cortex. D and E, bottom row: FoxP3-positive cells the renal cortex. Original magnification ×200. Scale bar = 50 μm. *P < 0.05, ****P < 0.0001 vs. Con WT mice; #P < 0.05, ##P < 0.01, ###P < 0.001 vs. Diab WT mice; ^^^P < 0.001 vs. Con C5aR1−/− mice. The line within each scatterplot represents the mean. Arrows represent positively stained cells.

Figure 3

Genetic deletion of C5aR1 protects against diabetes-induced inflammation and renal injury. Urinary albumin (A) and 8-isoprostane (B) were determined in control (Con) and diabetic (Diab) WT mice and in control and diabetic C5aR1−/− mice (n = 4–12 mice per group). C and E, top row: F4/80-positive cells in the renal cortex. D and E, bottom row: FoxP3-positive cells the renal cortex. Original magnification ×200. Scale bar = 50 μm. *P < 0.05, ****P < 0.0001 vs. Con WT mice; #P < 0.05, ##P < 0.01, ###P < 0.001 vs. Diab WT mice; ^^^P < 0.001 vs. Con C5aR1−/− mice. The line within each scatterplot represents the mean. Arrows represent positively stained cells.

Close modal

Infiltrating macrophages in the tubulointerstitium as assessed by F4/80 immunohistochemistry were increased in diabetic WT mice (Fig. 3C and E). Deletion of C5aR1 abrogated the increase in F4/80-positive cells in the tubulointerstitium in diabetic mice (Fig. 3C and E). FoxP3 regulatory T cells (Tregs) are thought to play a role in ameliorating inflammation in the diabetic kidney (30). We found a reduced number of FoxP3-positive cells in the tubulointerstitium of diabetic WT mice (Fig. 3D and E). C5aR1 deletion protected against FoxP3-positive Treg depletion in the setting of diabetes (Fig. 3D and E).

Pharmacological Inhibition of C5a/C5aR1 With PMX53 Is Efficacious in DKD

PMX53 is an orally active cyclic peptide antagonist of C5aR1 (6). Prior to administering PMX53 to mice with diabetes, we performed pharmacokinetic experiments with PMX53 in C57BL/6J mice (Supplementary Data). Plasma concentrations versus the time profile of PMX53 via i.v. and oral routes demonstrated a curvilinear pattern, with rapid clearance (Supplementary Fig. 1); the pharmacokinetic parameters reflected fast absorption and distribution of the drug (Supplementary Table 2). Oral bioavailability of ∼6% was determined on the basis of the area under the curve.

Treatment of STZ-induced diabetic mice with PMX53 (2 mg/kg/day) as a prophylactic regimen for 24 weeks inhibited diabetes-induced albuminuria (Fig. 4A), without affecting blood glucose (Table 2). PMX53 reduced urinary 8-isoprostanes in diabetic mice (Fig. 4B). PMX53 significantly decreased the extent of renal structural injury, as reflected by both the GSI (Fig. 4D and E) and mesangial matrix expansion (Fig. 4F). Similarly, PMX53 treatment decreased renal fibrosis in diabetic mice (determined on the basis of collagen IV immunostaining [Fig. 4D and G] and picrosirius red staining [Fig. 4D and H]).

Figure 4

C5aR1 inhibition by PMX53 is efficacious in DKD. Control WT mice (Con) and STZ-induced diabetic mice (Diab) (n = 6–12 mice per group) were treated with PMX (2 mg/kg/day) for 24 weeks. Urinary albumin (A), 8-isoprostane (B), and plasma IL-18 (C) were determined. Renal histology was examined by staining paraffin-embedded sections with PAS, and the GSI (D and E) and mesangial matrix expansion (D and F) were quantitated. Glomerular collagen IV (Col IV) deposition was determined (D and G). Kidney fibrotic area was determined on the basis of picrorsirius red staining (D and H). Inflammatory F4/80-positive cells (D and I) and anti-inflammatory FoxP3-positive cells (D and J) in the kidney were examined by immunohistochemistry. For PAS and Col IV, original magnification ×400, scale bar = 25 μm. For picrosirius red, F4/80, and FoxP3, original magnification ×200, scale bar = 50 μm. *P < 0.05, ***P < 0.001, ****P < 0.0001 vs. Con; #P < 0.05, ##P < 0.01, ####P < 0.0001 vs. Diab. The line within each scatterplot represents the mean. Arrows represent positively stained cells.

Figure 4

C5aR1 inhibition by PMX53 is efficacious in DKD. Control WT mice (Con) and STZ-induced diabetic mice (Diab) (n = 6–12 mice per group) were treated with PMX (2 mg/kg/day) for 24 weeks. Urinary albumin (A), 8-isoprostane (B), and plasma IL-18 (C) were determined. Renal histology was examined by staining paraffin-embedded sections with PAS, and the GSI (D and E) and mesangial matrix expansion (D and F) were quantitated. Glomerular collagen IV (Col IV) deposition was determined (D and G). Kidney fibrotic area was determined on the basis of picrorsirius red staining (D and H). Inflammatory F4/80-positive cells (D and I) and anti-inflammatory FoxP3-positive cells (D and J) in the kidney were examined by immunohistochemistry. For PAS and Col IV, original magnification ×400, scale bar = 25 μm. For picrosirius red, F4/80, and FoxP3, original magnification ×200, scale bar = 50 μm. *P < 0.05, ***P < 0.001, ****P < 0.0001 vs. Con; #P < 0.05, ##P < 0.01, ####P < 0.0001 vs. Diab. The line within each scatterplot represents the mean. Arrows represent positively stained cells.

Close modal
Table 2

Basic characteristics of nondiabetic control and STZ-induced diabetic mice treated with vehicle or the C5aR1 antagonist PMX53 for 24 weeks

Control miceDiabetic mice
Receiving vehicleReceiving PMX53Receiving vehicleReceiving PMX53
BW, g 34 ± 1 35 ± 1 23 ± 1**** 29 ± 1*### 
Plasma glucose, mg/dL 258 ± 27 201 ± 20 639 ± 38**** 559 ± 80*** 
HbA1c, % (mmol/mol) 5.0 ± 0.2 (31 ± 4) 5.4 ± 0.1 (36 ± 4) 12.2 ± 1.4** (123 ± 3) 9.4 ± 1.1 (79 ± 12) 
CrCl, mL/min/m2 17.7 ± 4.5 15.7 ± 1.8 28.4 ± 5.2 28.7 ± 3.4 
Cystatin C, ng/mL 464.8 ± 25.8 467.8 ± 49.5 287.5 ± 25.6** 373.4 ± 32.2 
L.Kid:BW, mg/g 5.368 ± 0.218 5.355 ± 0.190 10.670 ± 0.541**** 7.923 ± 0.245**### 
Control miceDiabetic mice
Receiving vehicleReceiving PMX53Receiving vehicleReceiving PMX53
BW, g 34 ± 1 35 ± 1 23 ± 1**** 29 ± 1*### 
Plasma glucose, mg/dL 258 ± 27 201 ± 20 639 ± 38**** 559 ± 80*** 
HbA1c, % (mmol/mol) 5.0 ± 0.2 (31 ± 4) 5.4 ± 0.1 (36 ± 4) 12.2 ± 1.4** (123 ± 3) 9.4 ± 1.1 (79 ± 12) 
CrCl, mL/min/m2 17.7 ± 4.5 15.7 ± 1.8 28.4 ± 5.2 28.7 ± 3.4 
Cystatin C, ng/mL 464.8 ± 25.8 467.8 ± 49.5 287.5 ± 25.6** 373.4 ± 32.2 
L.Kid:BW, mg/g 5.368 ± 0.218 5.355 ± 0.190 10.670 ± 0.541**** 7.923 ± 0.245**### 

Data are mean ± SEM. BW, body weight; CrCl, creatinine clearance; L.Kid:BW, left kidney weight–to–body weight ratio.

****P < 0.0001, ***P < 0.001, **P < 0.01, *P < 0.05 vs. control mice; ###P < 0.001 vs. diabetic mice.

Treatment with PMX53 was associated with a trend toward fewer F4/80-positive macrophages infiltrating the tubulointerstitium in diabetic kidneys, although this reduction was not statistically significant (Fig. 4D and I). Signaling through C5aR1 induces type 1 T helper cell responses and inhibits induction of CD4+ T cells into FoxP3-positive Tregs (3133). PMX53 has been shown to induce Treg activation (34). Indeed, PMX53 treatment in the diabetic setting led to an increase in the number of FoxP3-positive Tregs within the kidney (Fig. 4D and J). Furthermore, levels of the proinflammatory cytokine IL-18 were increased in mice with diabetes and were significantly attenuated with PMX53 (Fig. 4C).

Inhibition of C5aR1 Restores the Mitochondrial Fatty Acid Profile and Cardiolipin Remodeling in DKD

We next performed transcriptomic analysis on renal cortex using RNA sequencing. PMX53 significantly modulated the diabetes gene signature: almost half (49%) of the genes altered by diabetes were restored by PMX53 (Fig. 5A). Interestingly, Reactome gene set analysis identified that 5 of the top 10 pathways that were downregulated by PMX53 in diabetes were involved in cellular metabolism, including glycolysis, mitochondrial fatty acid β-oxidation, and fatty acyl-CoA biosynthesis (Fig. 5B). Of interest, the top gene upregulated by PMX53 treatment in diabetes was the acyl-CoA dehydrogenase family (ACAD) member 10 (Acad10; FDR = 1.18E-135, log2 fold change), which was also the top downregulated gene in diabetes (FDR = 3.19E-132, log2 fold change). Acad10 is a member of the ACAD family of enzymes, which participate in fatty acid oxidation (FAO) within mitochondria (35). Intriguingly, PMX53 treatment restored the majority of genes associated with mitochondrial fatty acid β-oxidation that were altered in diabetes (Fig. 5D and E and Supplementary Table 3). Furthermore, PMX53 significantly altered only 2 of 11 genes examined in the fatty acid metabolism pathway in the diabetic kidneys (Supplementary Table 3). Validation by quantitative RT-PCR also confirmed the Acad10 findings (Fig. 5F). Furthermore, Acad10 expression, which was downregulated in the renal cortex in diabetic WT mice, was restored in diabetic C5aR1−/− mice (Fig. 5G).

Figure 5

RNA sequencing analysis reveals a role for C5aR1 in pathways involved in mitochondrial fatty acid β-oxidation. A: Rank-rank density plot of differential gene expression due to diabetes and PMX53 treatment. B: Gene set enrichment analysis of diabetes (Diab) and diabetes + PMX53 (Diab+PMX53), showing the top Reactome pathways; the mitochondrial fatty acid β-oxidation pathway is red. C: MA plot of top differential genes in the Diab and Diab+PMX53 groups; the top three downregulated and top three upregulated genes are noted. D: Rank-rank density plot of differential genes in the mitochondrial fatty acid β-oxidation pathway. E: Gene set enrichment analysis highlighted genes that were altered by PMX53 in the mitochondrial fatty acid β-oxidation pathway in diabetic mice. Acad10 expression in PMX53-treated mice (F) and C5aR1−/− mice (G) (n = 6–12 mice per group). **P < 0.01 vs. Con or Con WT; ##P < 0.01 vs. Diab or Diab WT. The line within each scatterplot represents the mean.

Figure 5

RNA sequencing analysis reveals a role for C5aR1 in pathways involved in mitochondrial fatty acid β-oxidation. A: Rank-rank density plot of differential gene expression due to diabetes and PMX53 treatment. B: Gene set enrichment analysis of diabetes (Diab) and diabetes + PMX53 (Diab+PMX53), showing the top Reactome pathways; the mitochondrial fatty acid β-oxidation pathway is red. C: MA plot of top differential genes in the Diab and Diab+PMX53 groups; the top three downregulated and top three upregulated genes are noted. D: Rank-rank density plot of differential genes in the mitochondrial fatty acid β-oxidation pathway. E: Gene set enrichment analysis highlighted genes that were altered by PMX53 in the mitochondrial fatty acid β-oxidation pathway in diabetic mice. Acad10 expression in PMX53-treated mice (F) and C5aR1−/− mice (G) (n = 6–12 mice per group). **P < 0.01 vs. Con or Con WT; ##P < 0.01 vs. Diab or Diab WT. The line within each scatterplot represents the mean.

Close modal

The two key transcription factors responsible for inducing lipogenic genes are SREBP and carbohydrate-responsive element-binding protein (ChREBP) (36,37). To determine whether regulation of fatty acid synthesis was altered, we used quantitative RT-PCR to determine the gene expression of ChREBP and SREBP. Total ChREBP was decreased in diabetes and was not altered with PMX53 (Supplementary Fig. 4A) or C5aR1−/− (Supplementary Fig. 4B). Similarly, ChREBP-α was downregulated in diabetes and was not altered with PMX53 (Supplementary Fig. 4C) or C5aR1−/− (Supplementary Fig. 4D). In contrast, ChREBP-β mRNA expression was highly upregulated in the kidney in the diabetic setting (an ∼15-fold increase) (Supplementary Fig. 4E) and decreased in diabetic mice treated with PMX53. Interestingly, C5aR1−/− mice did not show a reduction in diabetes-induced ChREBP-β gene expression (Supplementary Fig. 4F). We found no change in SREBP-1a (Supplementary Fig. 4G and H) or SREBP-2 (Supplementary Fig. 4K and L). SREBP-1c, although unchanged in kidneys from diabetic mice in the PMX53 study cohort (Supplementary Fig. 4I), was decreased in the kidneys of diabetic WT mice in the C5aR1−/− study cohort, and mRNA expression was restored in the diabetic C5aR1−/− mice (Supplementary Fig. 4J). Because it is thought that ChREBP-β expression best reflects total ChREBP lipogenic activity (38), these results suggest that fatty acid synthesis is increased in the diabetic kidney and normalized by PMX53.

To gain insight into the fatty acid profile of the kidney, lipidomics was undertaken. Acylcarnitines are involved in transporting fatty acids into the mitochondria for oxidation and are elevated when fatty acid oxidation is suppressed. Alteration in the plasma acylcarnitine profile has been observed in DKD (39). However, few studies have reported acylcarnitine levels in the kidney in T1D. Lipidomics of renal cortex showed that levels of total (Supplementary Fig. 2A), short-chain (Supplementary Fig. 2D), and long-chain (Fig. 6A and Supplementary Fig. 2D) acylcarnitines were downregulated in the diabetic kidney. Total acylcarnitines were also standardized to protein (Supplementary Fig. 2E) and showed the same trend as total acylcarnitines standardized to phosphatidylcholine (Supplementary Fig. 2A). Inhibition of C5aR1 by PMX53 led to a restoration of total (Supplementary Fig. 2A), short-chain (Supplementary Fig. 2D), and long-chain (Fig. 6A and Supplementary Fig. 2D) acylcarnitine species. Additional lipid species showed significant changes in the kidneys of mice with diabetes (Supplementary Fig. 2G and Supplementary Table 4), including a decrease in ubiquinone (Supplementary Table 4), a member of the mitochondrial electron transport chain.

Figure 6

Diabetes induces changes to mitochondrial agility, including kidney cardiolipin remodeling and enhanced TCA cycle intermediate generation, which are normalized by C5aR1 inhibition. Lipidomics of renal cortex was performed in order to determine acylcarnitine species (A) and cardiolipin species (B) that were downregulated in diabetes and restored by PMX53 treatment, and cardiolipin species that were upregulated in diabetes and attenuated by PMX53 treatment (C). Data are mean ± SEM (n = 6–12 mice per group). D: TCA cycle intermediates in renal cortex were determined by metabolomics (n = 6–12 mice/group). *P < 0.05, ***P < 0.001 vs. control (Con) mice; #P < 0.05, ##P < 0.01, ###P < 0.001 vs. diabetic (Diab) mice. E: Human primary PTECs were exposed to C5a or C5a plus PMX53 for 24 h, and mitochondrial oxygen consumption rate (OCR), including basal, FCCP-uncoupled (UCR), and ATP-linked respiration, was determined by using a Seahorse flux analyzer. Values are the mean ± SEM (n ≥ 9 replicates per group). *P < 0.05 vs. Con; #P < 0.05 vs. C5a-treated cells. F: PTECs were exposed to C5a or C5a plus PMX53 for 24 h, and reactive oxygen species production was determined by using a dichlorodihydro-fluorescein diacetate (DCFDA) probe detected with flow cytometry. Values are the mean ± SEM (n = 3 replicates per group). *P < 0.05 vs. Con; #P < 0.05 vs. C5a-treated cells. PC, phosphatidylcholine.

Figure 6

Diabetes induces changes to mitochondrial agility, including kidney cardiolipin remodeling and enhanced TCA cycle intermediate generation, which are normalized by C5aR1 inhibition. Lipidomics of renal cortex was performed in order to determine acylcarnitine species (A) and cardiolipin species (B) that were downregulated in diabetes and restored by PMX53 treatment, and cardiolipin species that were upregulated in diabetes and attenuated by PMX53 treatment (C). Data are mean ± SEM (n = 6–12 mice per group). D: TCA cycle intermediates in renal cortex were determined by metabolomics (n = 6–12 mice/group). *P < 0.05, ***P < 0.001 vs. control (Con) mice; #P < 0.05, ##P < 0.01, ###P < 0.001 vs. diabetic (Diab) mice. E: Human primary PTECs were exposed to C5a or C5a plus PMX53 for 24 h, and mitochondrial oxygen consumption rate (OCR), including basal, FCCP-uncoupled (UCR), and ATP-linked respiration, was determined by using a Seahorse flux analyzer. Values are the mean ± SEM (n ≥ 9 replicates per group). *P < 0.05 vs. Con; #P < 0.05 vs. C5a-treated cells. F: PTECs were exposed to C5a or C5a plus PMX53 for 24 h, and reactive oxygen species production was determined by using a dichlorodihydro-fluorescein diacetate (DCFDA) probe detected with flow cytometry. Values are the mean ± SEM (n = 3 replicates per group). *P < 0.05 vs. Con; #P < 0.05 vs. C5a-treated cells. PC, phosphatidylcholine.

Close modal

Cardiolipin, which is a signature phospholipid of the inner mitochondrial membrane, is essential for optimal oxidative phosphorylation, mitochondrial architecture, and mitophagy. Although the total cardiolipin content in the kidney was unchanged (Supplementary Fig. 2B), we discovered striking changes to the composition of particular cardiolipin species in the diabetic kidney, indicating cardiolipin remodeling (Fig. 6B and C). Lipidomics identified a decrease in 9 of 38 cardiolipin species in the diabetic kidney—notably, those containing omega-3 polyunsaturated fatty acyl chains (22:6) in addition to the typical linoleoyl acyl chains (18:2) (Fig. 6B). Conversely, 6 of the 38 cardiolipin species analyzed increased (Fig. 6C), predominantly those containing monounsaturated fatty acids (18:1) in addition to the linoleoyl chains. This indicates the loss of omega-3 polyunsaturated fatty acids from cardiolipin within the inner mitochondrial membrane and replacement with monounsaturated fatty acids at this site. PMX53 treatment restored the composition of cardiolipin in diabetes (Fig. 6B and C), recapitulating a nondiabetic profile, suggesting that C5a may regulate mitochondrial homeostasis and play a key role in metabolism.

Diabetes Induces Changes to Mitochondrial Agility That Are Normalized by C5aR1 Inhibition

To explore the concept that the C5a/C5aR1 axis can promote changes in mitochondrial agility, we performed targeted metabolomics on renal cortex, focusing on citric acid cycle metabolites. The citric acid cycle intermediates cis-aconitate and isocitrate were increased (Fig. 6D), indicating increased flux through the citric acid cycle. Lending further weight to the role of C5a/C5aR1 in metabolism, PMX53 treatment was able to restore levels of citric acid cycle intermediates to control levels.

We then performed proof-of-concept studies to determine a direct effect of C5a on mitochondrial bioenergetics. We assessed mitochondrial respiratory function in human primary PTECs using a Seahorse flux analyzer. Although PTECs exposed to human C5a for 24 h did not display significantly altered basal mitochondrial oxygen consumption, C5a treatment increased mitochondrial respiration following FCCP-stimulated mitochondrial uncoupling and ATP-linked respiration (Fig. 6E), both of which were normalized by PMX53. Finally, reactive oxygen species generation, as measured by dichlorodihydro-fluorescein diacetate fluorescence, was heightened with C5a treatment and attenuated by PMX53 (Fig. 6F). These studies indicate that the C5a/C5aR1 axis can indeed promote changes in mitochondrial metabolic agility.

This report, describing both clinical and preclinical studies, emphasizes a potential role for C5a in the progression of DKD and as an attractive therapeutic target for renoprotection in the setting of diabetes. Collectively, these studies show that C5a/C5aR1 signaling is involved in the initiation and progression of renal disease via changes in mitochondrial agility, specifically through changes in cardiolipin remodeling, mitochondrial metabolite flux, and mitochondrial respiratory function.

Although the liver is the primary site of complement synthesis (40), local production of complement in the kidney, particularly within renal tubules, could contribute to tissue injury in a variety of renal diseases (41). C5aR1 has been shown to be expressed on proximal tubular cells (42), podocytes (43), fibroblasts (44), mesangial cells, and vascular endothelial and smooth muscle cells (11). Genome-wide transcriptome analysis showed upregulation of the complement pathway in microdissected human renal glomerular and tubule samples from patients with DKD (45). Indeed, renal cortical expression of C5aR1 increased early after the onset of diabetes in STZ-induced diabetic rats, before the development of albuminuria, and urinary excretion of C5a was evident later in diabetes, which persisted over the diabetes duration. C5aR1 was also upregulated in the livers of db/db mice and STZ-induced diabetic mice, also reflecting systemic complement activation. Thus, it seems that diabetes is a state of C5a activation at both a systemic and a local tissue level. A finding pertinent for translation is that plasma C5a was not reduced in patients with T1D or T2D who were treated with conventional RAS inhibitors, indicating that the C5a/C5aR1 signaling pathway is a pathogenic pathway that conventional renoprotective therapies do not effectively target or suppress.

In human diabetes, plasma C5a was positively associated with urinary albumin (r = 0.25), though this correlation was weak, as it did not reach statistical significance (P = 0.09). Plasma C5a was negatively correlated with estimated glomerular filtration rate (r = −0.32); however, this association did not reach significance (P = 0.06). In the STZ-induced diabetic mouse model, we found a highly significant correlation between urinary C5a and urinary albumin (r = 0.80; P < 0.0001). Although the upstream events that lead to complement activation in diabetes are unknown, complement activation in this setting is likely to be mediated by glucose. HbA1c was not correlated with plasma C5a in individuals with diabetes, but the increase in plasma C5a may be related to the ambient glucose concentration or to other proinflammatory stimuli that exist in the diabetic milieu.

Increasing evidence indicates that defective FAO contributes to renal fibrosis in chronic kidney diseases such as DKD (46,47). In this study, transcriptomic analyses showed that PMX53 restored genes associated with FAO that were altered in the diabetic kidney, specifically Acad10. Acad10 has only recently been identified as a long-chain acyl-CoA dehydrogenase (35), and its expression in the kidney is largely restricted to mitochondria (48). Intriguingly, Acad10 was ranked among the top 1% for association with early onset of T2D in a genome-wide association study in Pima Indians, a population with an extremely high prevalence of renal disease in the context of diabetes (49). However, its specific association with DKD has not been reported. In this study, inhibition of C5aR1 by PMX53 or genetic deletion of C5aR1 restored the expression of Acad10 in the diabetic kidney, indicating that C5a/C5aR1 signaling is involved in fatty acid homeostasis by suppressing the transcription of FAO enzymes such as Acad10.

l-carnitine (levocarnitine; 3-hydroxy-4-N-trimethylaminobutyrate) plays a pivotal role in transporting fatty acids into the mitochondria for subsequent β-oxidation. Disturbances to the relative composition of the endogenous carnitine pool, which comprises l-carnitine with short-, medium-, and long-chain acylcarnitines, is associated with impaired FAO and mitochondrial dysfunction (50). Literature describing the levels of acylcarnitines in DKD has primarily focused on plasma, urine, or both, and the results are conflicting (39). Although a recent study showed that circulating long-chain acylcarnitines are associated with incident renal functional decline in adults (51), it is well appreciated that plasma lipid profiles may not correlate with lipid levels in kidney tissue (52). Several studies have shown changes in renal cortical acylcarnitine levels in T2D mouse models (53,54). However, our study is, to our knowledge, the first to determine the acylcarnitine profile in the kidney in T1D. In our mouse model of STZ-induced diabetes, lipidomics showed acylcarnitines to be decreased in the renal cortex. Interestingly, these findings contrast with findings regarding acylcarnitine levels from db/db mice with T2D, in which levels of acylcarnitines were increased in the kidney cortex at 24 weeks of age, indicating increased FAO flux (53). This result is not totally surprising, as that model is driven by insulin resistance. Several other studies have also shown that more acylcarnitines but fewer tricarboxylic acid (TCA) cycle intermediates in the skeletal muscle in T2D may suggest that FAO flux does not match TCA cycle flux, leading to incomplete FAO (55). Nonetheless, very little is known of the relationship between acylcarnitine content, FAO, and TCA cycle flux in the kidney in T1D. We show here that TCA cycle intermediates such as cis-aconitate and isocitrate are increased in the diabetic kidney, concomitant with reduced levels of long-chain acylcarnitine, which may suggest a dysregulation of FAO leading to a compensatory increase in these TCA metabolites. However, metabolic flux assays would be required to confirm specific changes in FAO. Most importantly, the role of C5a/C5aR1 in disturbing the mitochondrial fatty acid signature in DKD is further substantiated by PMX53 restoring the acylcarnitine profile and TCA cycle intermediates in the diabetic renal cortex. Indeed, Kang et al. (46) showed that decreased tubule epithelial FAO induces metabolic reprogramming of tubular epithelial cells into a profibrotic phenotype. Thus, in our study, the restoration of the mitochondrial fatty acid signature that occurred by inhibiting C5aR1 may be driving the attenuation of renal injury in STZ-induced diabetic mice.

Cardiolipin, the “signature” phospholipid of the inner mitochondrial membrane, is crucial for optimal mitochondrial bioenergetics and dynamics (56). The appropriate fatty acid composition of cardiolipin is crucial for maintaining normal mitochondrial structure and function, including the curvature formation of the inner mitochondrial membrane (57) and stabilization of respiratory chain complexes (58). In this study, the kidneys from mice with diabetes exhibited a markedly distinct profile of cardiolipin species, with few species containing omega-3 (22:6) fatty acids and the incorporation of species containing monounsaturated fatty acids (oleic acid). Cardiolipin is highly sensitive to oxidative damage induced by reactive oxygen species because of its large amount of polyunsaturated fatty acids (59). Indeed, we found evidence of increased oxidative stress with increased lipid peroxidation (8-isoprostanes) in the diabetic kidney, which may have precipitated cardiolipin remodeling. In this study, inhibition of C5aR1 reversed the aberrant cardiolipin profile in the diabetic renal cortex, indicating that targeting the C5a/C5aR1 axis could prevent detrimental cardiolipin remodeling and restore mitochondrial homeostasis in DKD. A recent study implicated a role for C5a/C5aR1 signaling in the regulation of lipid metabolism in diabetes by using an aptamer to C5aR1 in a model of T2D (db/db mice). That study showed a reduction in serum triglycerides and a decrease in the expression of genes involved in regulating fatty acid synthesis, such as DGAT1 and SREBP1 (60). Our study is, to our knowledge, the first to demonstrate that the C5a/C5aR1 axis may be involved in mitochondrial cardiolipin remodeling, potentially resulting in a mitochondrial defect.

Results from our study have important clinical implications. The upstream positioning of complement in inflammatory propagation renders it an attractive system for targeted modulation (61). A suite of complement inhibitors have been developed and target complement at various stages of the activation pathway, including at the level of C3 (e.g., compstatin), C5 (e.g., eculizumab), and C5aR1 (e.g., PMX53 and avacopan) (61). However, selective targeting of C5a/C5aR1 is considered to be more amenable for managing chronic diseases because the immune defense functions of the proximal complement system, such as opsonization and formation of the MAC, are preserved, while the properties of C5a that promote inflammation and metabolic reprogramming are impeded. PMX53, an orally active peptide antagonist of C5a (6), is one of the most widely studied and widely used inhibitors of C5aR1 in models of inflammatory diseases, albeit not previously in diabetes. Although PMX53 was determined to have rapid clearance/distribution and low oral bioavailability in mice, it is a noncompetitive antagonist with effectively irreversible action and extended pharmacodynamics after oral administration (62). In the clinical context, early phase Ib/IIa trials for the treatment of rheumatoid arthritis have found PMX53 to be safe and well tolerated when administered orally (63). PMX53 and related agents have the potential to advance into clinical development for DKD.

We note that this study has limitations: 1) We performed RNA sequencing and lipidomics on renal cortical tissue, so we cannot determine which cell types are most affected by diabetes or C5aR1 inhibition. 2) We did not examine the role of C3a, an alternate anaphylatoxin of the complement system, which was also increased in the plasma of patients with diabetes. However, dual function of C3a and its receptor C3aR in controlling inflammation (64), as well as a lack of specific antagonists in clinical development (65), are main impediments to the translational potential of C3a. 3) We could not analyze stable isotope flux to assess FAO.

In summary, this series of clinical, preclinical, and in vivo mechanistic studies indicate that the C5a/C5aR1 signaling axis 1) plays a crucial role in the pathogenesis of DKD via mitochondrial reprogramming, 2) is not yet touched by conventional clinical therapies for DKD, and 3) may be directly targeted in order to potentiate meaningful renoprotection in diabetes. Targeting the C5a/C5aR1 pathway with orally bioavailable antagonists such as PMX53 could potentially fulfill the unmet clinical need of treating and preventing the development of DKD, the main cause of end-stage renal disease worldwide.

Acknowledgments. The authors thank the following people for providing technical assistance: Maryann Arnstein, Runa S. Lindblom, Karly C. Sourris, Phillip Kantharidis, Carlos Rosado, Antony Kaspi, and Gavin C. Higgins (Department of Diabetes, Monash University) and Natalie Mellet, Sally A. Penfold, and Brooke E. Harcourt (Baker Heart and Diabetes Institute). The authors acknowledge the use of Illumina sequencing at the Australian Genome Research Facility (and the support it receives from the Commonwealth of Australia). The authors thank Antony Kaspi for providing bioinformatics support related to multidimensional pathway analysis, and the authors acknowledge the Baker IDI Biobank, Melbourne, Australia, for providing human plasma samples. The authors also acknowledge their use of facilities at Monash Histology Platform and Metabolomics Australia, Bio21.

Funding. This study was supported by a JDRF Innovative grant (1-SRA-2014-261-Q-R) and a Diabetes Australia Research Program grant. S.M.T. is supported by a JDRF Advanced Postdoctoral Fellowship. E.I.E. was supported by a Viertel Clinical Investigatorship, a Royal Australian College of Physicians–JDRF fellowship, and Sir Edward Weary Dunlop Medical Research Foundation and Diabetes Australia Research Program research grants. T.M.W. was supported by an Australian National Health and Medical Research Council Career Development Fellowship (APP1105420). M.T.C. has received a Career Development Award from JDRF Australia and is the recipient of the Australian Research Council Special Research Initiative in Type 1 Juvenile Diabetes.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.M.T. researched the data, performed the experiments, and wrote the manuscript. M.Z. and A.E.-O. performed the RNA sequencing and bioinformatics and wrote the manuscript. V.T.-B., M.S., A.L., T.-V.N., and M.V.C. performed laboratory experiments. V.K. generated PMX53 and evaluated the pharmacokinetics. K.H. and P.J.M. performed the lipidomics. R.L., S.T.B., A.S., D.A.P., and R.J.M. provided clinical samples. D.C.H., S.G.W., J.M.F., and M.E.C. provided input into writing the manuscript. R.A.W. provided the C5aR1−/− mice. E.I.E. oversaw the clinical studies and wrote the manuscript. T.M.W. designed the experiments and provided intellectual input into the study. M.T.C. conceived, designed, and oversaw the study and wrote the manuscript. M.T.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the American Society of Nephrology Kidney Week 2018, San Diego, CA, 23–28 October 2018, and the 11th International Conference on Complement Therapeutics, Chania, Crete, Greece, 23–28 June 2018.

1.
Thomas
MC
,
Weekes
AJ
,
Broadley
OJ
,
Cooper
ME
,
Mathew
TH
.
The burden of chronic kidney disease in Australian patients with type 2 diabetes (the NEFRON study)
.
Med J Aust
2006
;
185
:
140
144
2.
Nathan
DM
,
Zinman
B
,
Cleary
PA
, et al.;
Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Research Group
.
Modern-day clinical course of type 1 diabetes mellitus after 30 years’ duration: the diabetes control and complications trial/epidemiology of diabetes interventions and complications and Pittsburgh epidemiology of diabetes complications experience (1983-2005)
.
Arch Intern Med
2009
;
169
:
1307
1316
3.
Tesch
GH
.
Diabetic nephropathy - is this an immune disorder
?
Clin Sci (Lond)
2017
;
131
:
2183
2199
4.
Walport
MJ
.
Complement. First of two parts
.
N Engl J Med
2001
;
344
:
1058
1066
5.
Flyvbjerg
A
.
The role of the complement system in diabetic nephropathy
.
Nat Rev Nephrol
2017
;
13
:
311
318
6.
Woodruff
TM
,
Nandakumar
KS
,
Tedesco
F
.
Inhibiting the C5-C5a receptor axis
.
Mol Immunol
2011
;
48
:
1631
1642
7.
Liu
L
,
Zhang
Y
,
Duan
X
, et al
.
C3a, C5a renal expression and their receptors are correlated to severity of IgA nephropathy
.
J Clin Immunol
2014
;
34
:
224
232
8.
Peng
Q
,
Li
K
,
Smyth
LA
, et al
.
C3a and C5a promote renal ischemia-reperfusion injury
.
J Am Soc Nephrol
2012
;
23
:
1474
1485
9.
Williams
AL
,
Gullipalli
D
,
Ueda
Y
, et al
.
C5 inhibition prevents renal failure in a mouse model of lethal C3 glomerulopathy
.
Kidney Int
2017
;
91
:
1386
1397
10.
Bao
L
,
Osawe
I
,
Puri
T
,
Lambris
JD
,
Haas
M
,
Quigg
RJ
.
C5a promotes development of experimental lupus nephritis which can be blocked with a specific receptor antagonist
.
Eur J Immunol
2005
;
35
:
2496
2506
11.
Abe
K
,
Miyazaki
M
,
Koji
T
, et al
.
Enhanced expression of complement C5a receptor mRNA in human diseased kidney assessed by in situ hybridization
.
Kidney Int
2001
;
60
:
137
146
12.
Libianto
R
,
Jerums
G
,
Lam
Q
, et al
.
Relationship between urinary sodium excretion and serum aldosterone in patients with diabetes in the presence and absence of modifiers of the renin-angiotensin-aldosterone system
.
Clin Sci (Lond)
2014
;
126
:
147
154
13.
Hollmann
TJ
,
Mueller-Ortiz
SL
,
Braun
MC
,
Wetsel
RA
.
Disruption of the C5a receptor gene increases resistance to acute Gram-negative bacteremia and endotoxic shock: opposing roles of C3a and C5a
.
Mol Immunol
2008
;
45
:
1907
1915
14.
Coughlan
MT
,
Nguyen
TV
,
Penfold
SA
, et al
.
Mapping time-course mitochondrial adaptations in the kidney in experimental diabetes
.
Clin Sci (Lond)
2016
;
130
:
711
720
15.
March
DR
,
Proctor
LM
,
Stoermer
MJ
, et al
.
Potent cyclic antagonists of the complement C5a receptor on human polymorphonuclear leukocytes. Relationships between structures and activity
.
Mol Pharmacol
2004
;
65
:
868
879
16.
Tan
SM
,
Sharma
A
,
Yuen
DYC
, et al
.
The modified selenenyl amide, M-hydroxy ebselen, attenuates diabetic nephropathy and diabetes-associated atherosclerosis in ApoE/GPx1 double knockout mice
.
PLoS One
2013
;
8
:
e69193
17.
Jiang
H
,
Lei
R
,
Ding
S-W
,
Zhu
S
.
Skewer: a fast and accurate adapter trimmer for next-generation sequencing paired-end reads
.
BMC Bioinformatics
2014
;
15
:
182
18.
Dobin
A
,
Davis
CA
,
Schlesinger
F
, et al
.
STAR: ultrafast universal RNA-seq aligner
.
Bioinformatics
2013
;
29
:
15
21
19.
Robinson
MD
,
McCarthy
DJ
,
Smyth
GK
.
edgeR: a Bioconductor package for differential expression analysis of digital gene expression data
.
Bioinformatics
2010
;
26
:
139
140
20.
Subramanian
A
,
Kuehn
H
,
Gould
J
,
Tamayo
P
,
Mesirov
JP
.
GSEA-P: a desktop application for Gene Set Enrichment Analysis
.
Bioinformatics
2007
;
23
:
3251
3253
21.
Fabregat
A
,
Sidiropoulos
K
,
Garapati
P
, et al
.
The reactome pathway knowledgebase
.
Nucleic Acids Res
2016
;
44
:
D481
D487
22.
Quaife-Ryan
GA
,
Sim
CB
,
Ziemann
M
, et al
.
Multicellular transcriptional analysis of mammalian heart regeneration
.
Circulation
2017
;
136
:
1123
1139
23.
Tham
YK
,
Huynh
K
,
Mellett
NA
, et al
.
Distinct lipidomic profiles in models of physiological and pathological cardiac remodeling, and potential therapeutic strategies
.
Biochim Biophys Acta Mol Cell Biol Lipids
2018
;
1863
:
219
234
24.
Huynh
K
,
Barlow
CK
,
Jayawardana
KS
, et al
.
High-throughput plasma lipidomics: detailed mapping of the associations with cardiometabolic risk factors
.
Cell Chem Biol
2019
;
26
:
71
84.e4
25.
Kowalski
GM
,
De Souza
DP
,
Burch
ML
, et al
.
Application of dynamic metabolomics to examine in vivo skeletal muscle glucose metabolism in the chronically high-fat fed mouse
.
Biochem Biophys Res Commun
2015
;
462
:
27
32
26.
Overgaard
AJ
,
Weir
JM
,
De Souza
DP
, et al
.
Lipidomic and metabolomic characterization of a genetically modified mouse model of the early stages of human type 1 diabetes pathogenesis
.
Metabolomics
2016
;
12
:
13
27.
McGee
SL
,
Sadli
N
,
Morrison
S
,
Swinton
C
,
Suphioglu
C
.
DHA protects against zinc mediated alterations in neuronal cellular bioenergetics
.
Cell Physiol Biochem
2011
;
28
:
157
162
28.
Benjamini
Y
,
Hochberg
Y
.
Controlling the false discovery rate - a practical and powerful approach to multiple testing
.
J R Stat Soc B
1995
;
57
:
289
300
29.
Qin
X
,
Gao
B
.
The complement system in liver diseases
.
Cell Mol Immunol
2006
;
3
:
333
340
30.
Lim
AKH
,
Tesch
GH
.
Inflammation in diabetic nephropathy
.
Mediators Inflamm
2012
;
2012
:
146154
31.
Strainic
MG
,
Shevach
EM
,
An
F
,
Lin
F
,
Medof
ME
.
Absence of signaling into CD4⁺ cells via C3aR and C5aR enables autoinductive TGF-β1 signaling and induction of Foxp3⁺ regulatory T cells
.
Nat Immunol
2013
;
14
:
162
171
32.
Weaver
DJ
 Jr
.,
Reis
ES
,
Pandey
MK
, et al
.
C5a receptor-deficient dendritic cells promote induction of Treg and Th17 cells
.
Eur J Immunol
2010
;
40
:
710
721
33.
van der Touw
W
,
Cravedi
P
,
Kwan
WH
,
Paz-Artal
E
,
Merad
M
,
Heeger
PS
.
Cutting edge: receptors for C3a and C5a modulate stability of alloantigen-reactive induced regulatory T cells
.
J Immunol
2013
;
190
:
5921
5925
34.
Dick
J
,
Gan
PY
,
Ford
SL
, et al
.
C5a receptor 1 promotes autoimmunity, neutrophil dysfunction and injury in experimental anti-myeloperoxidase glomerulonephritis
.
Kidney Int
2018
;
93
:
615
625
35.
He
M
,
Pei
Z
,
Mohsen
A-W
, et al
.
Identification and characterization of new long chain acyl-CoA dehydrogenases
.
Mol Genet Metab
2011
;
102
:
418
429
36.
Abdul-Wahed
A
,
Guilmeau
S
,
Postic
C
.
Sweet sixteenth for ChREBP: established roles and future goals
.
Cell Metab
2017
;
26
:
324
341
37.
Wang
Y
,
Viscarra
J
,
Kim
SJ
,
Sul
HS
.
Transcriptional regulation of hepatic lipogenesis
.
Nat Rev Mol Cell Biol
2015
;
16
:
678
689
38.
Herman
MA
,
Peroni
OD
,
Villoria
J
, et al
.
A novel ChREBP isoform in adipose tissue regulates systemic glucose metabolism
.
Nature
2012
;
484
:
333
338
39.
Chuang
W-H
,
Arundhathi
A
,
Lu
C
, et al
.
Altered plasma acylcarnitine and amino acid profiles in type 2 diabetic kidney disease
.
Metabolomics
2016
;
12
:
108
40.
Morgan
BP
,
Gasque
P
.
Extrahepatic complement biosynthesis: where, when and why
?
Clin Exp Immunol
1997
;
107
:
1
7
41.
Zhou
W
,
Marsh
JE
,
Sacks
SH
.
Intrarenal synthesis of complement
.
Kidney Int
2001
;
59
:
1227
1235
42.
Zahedi
R
,
Braun
M
,
Wetsel
RA
, et al
.
The C5a receptor is expressed by human renal proximal tubular epithelial cells
.
Clin Exp Immunol
2000
;
121
:
226
233
43.
Weiss
S
,
Rosendahl
A
,
Czesla
D
, et al
.
The complement receptor C5aR1 contributes to renal damage but protects the heart in angiotensin II-induced hypertension
.
Am J Physiol Renal Physiol
2016
;
310
:
F1356
F1365
44.
Mathern
DR
,
Heeger
PS
.
Molecules great and small: the complement system
.
Clin J Am Soc Nephrol
2015
;
10
:
1636
1650
45.
Woroniecka
KI
,
Park
AS
,
Mohtat
D
,
Thomas
DB
,
Pullman
JM
,
Susztak
K
.
Transcriptome analysis of human diabetic kidney disease
.
Diabetes
2011
;
60
:
2354
2369
46.
Kang
HM
,
Ahn
SH
,
Choi
P
, et al
.
Defective fatty acid oxidation in renal tubular epithelial cells has a key role in kidney fibrosis development
.
Nat Med
2015
;
21
:
37
46
47.
Herman-Edelstein
M
,
Scherzer
P
,
Tobar
A
,
Levi
M
,
Gafter
U
.
Altered renal lipid metabolism and renal lipid accumulation in human diabetic nephropathy
.
J Lipid Res
2014
;
55
:
561
572
48.
Bloom
K
,
Mohsen
A-W
,
Karunanidhi
A
, et al
.
Investigating the link of ACAD10 deficiency to type 2 diabetes mellitus
.
J Inherit Metab Dis
2018
;
41
:
49
57
49.
Bian
L
,
Hanson
RL
,
Muller
YL
, et al.;
MAGIC Investigators
.
Variants in ACAD10 are associated with type 2 diabetes, insulin resistance and lipid oxidation in Pima Indians
.
Diabetologia
2010
;
53
:
1349
1353
50.
Reuter
SE
,
Evans
AM
.
Carnitine and acylcarnitines: pharmacokinetic, pharmacological and clinical aspects
.
Clin Pharmacokinet
2012
;
51
:
553
572
51.
Wang
F
,
Sun
L
,
Sun
Q
, et al
.
Associations of plasma amino acid and acylcarnitine profiles with incident reduced glomerular filtration rate
.
Clin J Am Soc Nephrol
2018
;
13
:
560
568
52.
Sas
KM
,
Lin
J
,
Rajendiran
TM
, et al
.
Shared and distinct lipid-lipid interactions in plasma and affected tissues in a diabetic mouse model
.
J Lipid Res
2018
;
59
:
173
183
53.
Sas
KM
,
Kayampilly
P
,
Byun
J
, et al
.
Tissue-specific metabolic reprogramming drives nutrient flux in diabetic complications
.
JCI Insight
2016
;
1
:
e86976
54.
Weiser
A
,
Giesbertz
P
,
Daniel
H
,
Spanier
B
.
Acylcarnitine profiles in plasma and tissues of hyperglycemic NZO mice correlate with metabolite changes of human diabetes
.
J Diabetes Res
2018
;
2018
:
1864865
55.
Schooneman
MG
,
Vaz
FM
,
Houten
SM
,
Soeters
MR
.
Acylcarnitines: reflecting or inflicting insulin resistance
?
Diabetes
2013
;
62
:
1
8
56.
Claypool
SM
,
Koehler
CM
.
The complexity of cardiolipin in health and disease
.
Trends Biochem Sci
2012
;
37
:
32
41
57.
Renner
LD
,
Weibel
DB
.
Cardiolipin microdomains localize to negatively curved regions of Escherichia coli membranes
.
Proc Natl Acad Sci U S A
2011
;
108
:
6264
6269
58.
Claypool
SM
,
Oktay
Y
,
Boontheung
P
,
Loo
JA
,
Koehler
CM
.
Cardiolipin defines the interactome of the major ADP/ATP carrier protein of the mitochondrial inner membrane
.
J Cell Biol
2008
;
182
:
937
950
59.
Toda
N
,
Imamura
T
,
Okamura
T
.
Alteration of nitric oxide-mediated blood flow regulation in diabetes mellitus
.
Pharmacol Ther
2010
;
127
:
189
209
60.
Yiu
WH
,
Li
RX
,
Wong
DWL
, et al
.
Complement C5a inhibition moderates lipid metabolism and reduces tubulointerstitial fibrosis in diabetic nephropathy
.
Nephrol Dial Transplant
2018
;
33
:
1323
1332
61.
Ricklin
D
,
Lambris
JD
.
Progress and trends in complement therapeutics
.
Adv Exp Med Biol
2013
;
735
:
1
22
62.
Seow
V
,
Lim
J
,
Cotterell
AJ
, et al
.
Receptor residence time trumps drug-likeness and oral bioavailability in determining efficacy of complement C5a antagonists
.
Sci Rep
2016
;
6
:
24575
63.
Vergunst
CE
,
Gerlag
DM
,
Dinant
H
, et al
.
Blocking the receptor for C5a in patients with rheumatoid arthritis does not reduce synovial inflammation
.
Rheumatology (Oxford)
2007
;
46
:
1773
1778
64.
Coulthard
LG
,
Woodruff
TM
.
Is the complement activation product C3a a proinflammatory molecule? Re-evaluating the evidence and the myth
.
J Immunol
2015
;
194
:
3542
3548
65.
Hawksworth
OA
,
Li
XX
,
Coulthard
LG
,
Wolvetang
EJ
,
Woodruff
TM
.
New concepts on the therapeutic control of complement anaphylatoxin receptors
.
Mol Immunol
2017
;
89
:
36
43
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data