CLINICAL RESEARCH www.jasn.org Complement Activation in Peritoneal Dialysis–Induced Arteriolopathy Maria Bartosova,1 Betti Schaefer,1 Justo Lorenzo Bermejo,2 Silvia Tarantino,3 Felix Lasitschka,4 Stephan Macher-Goeppinger,5 Peter Sinn,4 Bradley A. Warady,6 Ariane Zaloszyc,7 Katja Parapatics,8 Peter Májek,8 Keiryn L. Bennett,8 Jun Oh,9 Christoph Aufricht,3 Franz Schaefer,1 Klaus Kratochwill,3,10 and Claus Peter Schmitt1 1Division of Pediatric Nephrology, Center for Pediatric and Adolescent Medicine, 2Department of Medical Biometry, Institute of Medical Biometry and Informatics, and 4Department of General Pathology, Institute of Pathology, University of Heidelberg, Heidelberg, Germany; 3Department of Pediatrics and Adolescent Medicine and 10Christian Doppler Laboratory for Molecular Stress Research in Peritoneal Dialysis, Medical University of Vienna, Vienna, Austria; 5Department of Pathology, University Medical Center Mainz, Mainz, Germany; 6Division of Pediatric Nephrology, Children’s Mercy Hospital, University of Missouri- Kansas City School of Medicine, Kansas City, Missouri; 7Department of Pediatrics 1, University Hospital of Strasbourg, Strasbourg, France; 8CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences, Vienna, Austria; and 9Department of Pediatric Nephrology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany ABSTRACT Cardiovascular disease (CVD) is the leading cause of increased mortality in patients with CKD and is further aggravated by peritoneal dialysis (PD). Children are devoid of preexisting CVD and provide unique insight into specificuremia-andPD-inducedpathomechanismsofCVD.We obtained peritoneal specimens from children with stage 5 CKD at time of PD catheter insertion (CKD5 group), children with established PD (PD group), and age-matched nonuremic controls (n=6/group). We microdissected omental arterioles from tissue layers not directly exposed to PD fluid and used adjacent sections of four arterioles per patient for transcriptomic and proteomic analyses. Findings were validated in omental and parietal arterioles from independent pediatric control (n=5), CKD5 (n=15), and PD (n=15) cohorts. Transcriptomic analysis revealed differential gene expression in control versus CKD5 arterioles and in CKD5 versus PD arterioles. Gene ontology analyses revealed activation of metabolic processes in CKD5 arterioles and of inflammatory, immunologic, and stress-response cascades in PD arterioles. PD arterioles exhibited particular upregulation of the complement system and respective reg- ulatory pathways, with concordant findings at the proteomic level. In the validation cohorts, PD specimens had the highest abundance of omental and parietal arteriolar C1q, C3d, terminal complement complex, and phosphorylated SMAD2/3, a downstream effector of TGF-b. Furthermore, in the PD parietal arterioles, C1q and terminal complement complex abundance correlated with the level of dialytic glucose exposure, abundance of phosphorylated SMAD2/3, and degree of vasculopathy. We conclude that PD fluids activate arteriolar com- plement and TGF-b signaling, which quantitatively correlate with the severity of arteriolar vasculopathy. J Am Soc Nephrol 29: 268–282, 2018. doi: https://doi.org/10.1681/ASN.2017040436 In children on dialysis, the risk of mortality is in- Received April 20, 2017. Accepted August 2, 2017. creased 40-fold compared with healthy popula- K.K. and C.P.S. contributed equally to this work. tions.1,2 A third of the patients with juvenile onset Published online ahead of print. Publication date available at of ESRD die before the fourth decade of life, the www.jasn.org. majority from cardio- and cerebrovascular dis- Correspondence: Dr. Claus Peter Schmitt, Division of Pediatric 1,3,4 ease. Endothelial dysfunction develops early in Nephrology, Center for Pediatric and Adolescent Medicine, Im children with CKD5 and is followed by structural Neuenheimer Feld 430, 69120 Heidelberg, Germany. Email: and functional vascular changes including in- [email protected] creased arterial stiffness and wall thickening,6,7 Copyright © 2018 by the American Society of Nephrology 268 ISSN : 1046-6673/2901-268 J Am Soc Nephrol 29: 268–282, 2018 www.jasn.org CLINICAL RESEARCH arteriosclerotic lesions,8 and vascular calcifications,9 the pro- Significance Statement gression of which is markedly accelerated after initiation of dialysis.6–8 The risk for progressive arterial disease is enhanced Mortality due to cardiovascular disease is markedly increased in by both classic (hypertension, hyperlipidemia) and nonclassic, chronic kidney disease (CKD); cardiovascular risk may be increased CKD-associated risk factors (hyperphosphatemia, hyperpara- further by peritoneal dialysis (PD). We isolated omental arterioles from children at time of initial PD catheter insertion, with established fl 3,9,10 thyroidism, subclinical in ammation). PD or without CKD for transcriptomic and proteomic analysis. We Peritoneal dialysis (PD) is the preferred dialysis modality in identified activation of metabolic processes at initiation of PD and children11 and is increasingly applied in adults,12 due to its inflammatory, immunological and stress-response cascades during independence of a vascular access, cost effectiveness, and PD, with highest upregulation of the complement system and re- advantages of a home-based therapy. Patients on PD, however, spective regulatory pathways. We also demonstrated close corre- lations ofdialytic glucose exposure to activation of complement and face additional cardiovascular risks related to the resorption of TGF-ß pathways and to the severity of vasculopathy. This study glucose and vasculotoxic glucose degradation products (GDP) provides evidence for an essential role of complement and TGF-ß from the dialysis fluid. Observational and interventional evi- signaling in PD induced vasculopathy. dence have suggested a role of peritoneal glucose and/or GDP exposure in patient survival on dialysis.13,14 The relative effect biologic relevance at the level of effector proteins and subsequent of dialytic glucose and GDP content remains controversial. quantitative validation. GDP exposure can be substantially reduced by use of multi- chamber PD fluids; the effect of such solutions on long-term patient outcome, however, is still unknown.15 RESULTS In this work, we sought to obtain a comprehensive under- standing of the pathomechanisms of CKD-associated arterio- Transcriptomics and Gene Ontology Analysis pathy and to explore the specific contribution of PD therapy. Clinical and biochemical findings were similar in healthy chil- The analysis of small arteries and precapillary arterioles is of dren, in children with CKD5, and in children on dialysis (Table particular interest because they control peripheral resistance 1). Median lumen to vessel (L/V) ratio of arterioles, a marker and microcirculation. Vasculopathy in this part of the arterial of vasculopathy, used for omics analysis was 0.76 tree predicts left ventricular hypertrophy and cardiovascular (interquartile range [IQR], 0.14) in healthy controls, 0.58 events in hypertensive patients.16–19 Peritoneal arterioles were (0.08) for the CKD5 group, and 0.59 (0.10) for the low- obtained from children with CKD5, children undergoing GDP PD group (P,0.001, P=0.52 CKD5 versus PD). Com- chronic PD with double-chamber PD fluids, and healthy control pared with the samples from age- and sex-matched healthy children. This population is unique due to the absence of second- children, transcriptome analysis of arterioles from CKD5 chil- ary pathologies typically present in adults, such as changes related dren showed upregulation of 173 and downregulation of 117 to long-standing hypertension, diabetes, smoking, and ageing. We genes (P,0.01) (Supplemental Table 1). Over-represented analyzed microdissected arteriolar tissue samples by means of pathways among these differentially expressed genes mainly transcriptomic and proteomic profiling and in independent co- included protein targeting to membranes and translational ini- horts by automated immunohistochemistry of intact tissue. The tiation of various metabolic processes (Supplemental Table combination of these techniques allowed sensitive detection of 2).InchildrenonPD,88geneswereup-and11genes molecular changes on the transcript level while controlling for downregulated compared with children with CKD5. Gene Table 1. Characteristics of the omic-screening cohort Characteristic Controls, Median (IQR) CKD5, Median (IQR) PD, Median (IQR) P (ANOVA) Age, yr 7.6 (7.2) 6.7 (6.1) 3.7 (9.3) 0.88 BMI SDS 20.4 (1.6) 20.25 (0.92) 20.3 (2.7) 0.56 eGFR 108 (22.1) 5.6 (3.3) 6.9 (5.9) ,0.001a PD duration, mo n.a. n.a. 21.5 (31) n.a. BP, mmHg 102/60 (26/11) 108/65 (28/18) 91/57 (30/19) 0.71/0.37 CRP, mg/L ,2 (0) 2.0 (1.2) 2.0 (1.4) 0.38 Hemoglobin, g/dl 12.1 (2.5) 9.7 (3.5) 11.4 (1.5) 0.04a Calcium, mg/dl 9.6 (4.6) 9.6 (1.9) 9.7 (0.8) 0.27 Phosphate, mg/dl n.a. 5.6 (3.6) 4.9 (2.2) 0.21 Parathyroid hormone, pg/ml n.a. 134 (633) 156 (197) 0.23 Creatinine, mg/dl 0.5 (0.1) 5.6 (3.2) 6.3 (6) 0.01a Albumin, g/dl n.a. 4.2 (1.5) 3.8 (0.7) 0.79 BUN, mg/dl n.a. 79 (54) 43 (37) 0.22 n=6 per group. BMI SDS, body mass index standard deviation score; n.a., not applicable; CRP, C-reactive protein. aNo significant difference CKD5 versus PD. J Am Soc Nephrol 29: 268–282, 2018 Complement in Peritoneal Dialysis Arteriolopathy 269 CLINICAL RESEARCH www.jasn.org Figure 1. Cross-omics analysis of the complement pathway activation. Transcriptomic and proteomic analyses from adjacent sections of microdissected omental arterioles demonstrate (A) concordant activation of the IPA complement pathway and (B) correlation of complement-associated transcripts and proteins. Red shapes containing the gene symbols indicate upregulation, green symbols downregulation. In the correlation plot, black triangles represent direct members of the complement pathway; black round symbols represent gene symbols associated to the complement system in Panther, GO, and UniProt. The x axis gives mRNA expression ratios PD/CKD5, y axis respective protein ratios. The association between gene and protein ratios is highly significant when the transcriptomics outlier (CFB) is removed (P,0.01).