Carbamylated sortilin associates with cardiovascular calcification in patients with chronic kidney disease

kidney disease 55 56 57 58 59 60 61 62 63 64 65 66 67 68 Vera Jankowski, Turgay Saritas, Mads Kjolby, Juliane Hermann, Thimoteus Speer, Anika Himmelsbach, Kerstin Mahr, Marina Augusto Heuschkel, Stefan J. Schunk, Soren Thirup, Simon Winther, Morten Bottcher, Mette Nyegard, Anders Nykjaer, Rafael Kramann, Nadine Kaesler, Joachim Jankowski, Juergen Floege, Nikolaus Marx and Claudia Goettsch IMCAR, Medical Faculty, RWTH Aachen University, Aachen, Germany; Division of Nephrology and Clinical Immunology, University Hospital, RWTH Aachen, Aachen, Germany; Institute of Experimental Medicine and Systems Biology, University Hospital RWTH Aachen, Aachen, Germany; PROMEMO and DANDRITE, Department of Biomedicine, Aarhus University, Aarhus, Denmark; Danish Diabetes Academy, Novo Nordisk Foundation, Denmark; Department of Clinical Pharmacology, Aarhus University Hospital, Aarhus, Denmark; Department of Internal Medicine 4, Translational Cardio-Renal Medicine, Saarland University, Saarland University Hospital, Homburg/ Saar, Germany; Department of Internal Medicine I, Cardiology, University Hospital, Medical Faculty, RWTH Aachen, Aachen, Germany; Department of Molecular Biology and Genetics, Aarhus University, Aarhus, Denmark; Department of Cardiology, Gødstrup Hospital, NIDO, Herning, Denmark; and Department of Health Science and Technology, Aalborg University, Aalborg, Denmark

Sortilin, an intracellular sorting receptor, has been identified as a cardiovascular risk factor in the general population. Patients with chronic kidney disease (CKD) are highly susceptible to develop cardiovascular complications such as calcification. However, specific CKD-induced posttranslational protein modifications of sortilin and their link to cardiovascular calcification remain unknown. To investigate this, we examined two independent CKD cohorts for carbamylation of circulating sortilin and detected increased carbamylated sortilin lysine residues in the extracellular domain of sortilin with kidney function decline using targeted mass spectrometry. Structure analysis predicted altered ligand binding by carbamylated sortilin, which was verified by binding studies using surface plasmon resonance measurement, showing an increased affinity of interleukin 6 to in vitro carbamylated sortilin. Further, carbamylated sortilin increased vascular calcification in vitro and ex vivo that was accelerated by interleukin 6. Imaging by mass spectrometry of human calcified arteries revealed in situ carbamylated sortilin. In patients with CKD, sortilin carbamylation was associated with coronary artery calcification, independent of age and kidney function. Moreover, patients with carbamylated sortilin displayed significantly faster progression of coronary artery calcification than patients without sortilin carbamylation. Thus, carbamylated sortilin may be a risk factor for cardiovascular calcification and may contribute to elevated cardiovascular complications in patients with CKD.
specific CKD-associated protein modifications linked to CV calcification are unknown.
Sortilin is a ubiquitously expressed member of the vacuolar protein sorting 10 protein family of intracellular sorting receptors. 7 It is a single-pass type I transmembrane protein with various roles in protein sorting, trafficking, and cell signaling. As an endocytosis receptor, sortilin can trigger the internalization of ligands from the cell surface via endocytosis and sort ligands between intracellular compartments, such as trans-Golgi network, endosome, lysosome, and secretory pathway. 8 Preclinical in vivo evidence suggests an important role of sortilin in the pathogenesis of vascular and metabolic disorders through contributions to arterial wall inflammation and calcification, dysregulated lipoprotein metabolism, and type 2 diabetes mellitus, all CV risk factors. 7 In human vascular smooth muscle cells (hSMCs), intracellular sortilin regulates the loading of the procalcific protein tissue nonspecific alkaline phosphatase (TNAP) into extracellular vesicles, thereby conferring the calcification potential that contributes to microcalcification formation. 9 The ectodomain of plasma membrane-bound sortilin can be shed and secreted into the circulation. 7 In a community-dwelling cohort of men aged >50 years, we reported an association of high sortilin serum levels with aortic calcification and CV events, suggesting sortilin as a CV risk factor in the general population. 10 A role of sortilin in CKD, a patient population with a marked increase in CV calcification, continues to defy elucidation. Most studies have focused on the function of cellular sortilin rather than exploring the biological function of the circulating soluble form. 7 Therefore, achieving a better understanding of the mechanistic relationship between circulating sortilin and the regulatory impact of PTM in CKD will broaden the knowledge of sortilin in CV calcification.
This study interrogates the hypothesis that PTM of circulating sortilin is involved in the development of CV calcification in patients with CKD.

Cardiovascular and Renal Outcome in CKD 2-4 Patients-The Fourth Homburg evaluation (CARE FOR HOMe) cohort
The CARE FOR HOMe study has been previously described in detail. 11 A subset of 97 patients was used for the analysis.

Dan-NICAD 1 cohort
The study design of the Danish study of noninvasive testing in coronary artery disease (Dan-NICAD 1) trial has been described previously. 12 For the analysis, we identified a subset of 97 enrolled patients with estimated glomerular filtration rate (eGFR) >60 ml/ min and performed frequency matching based on age, sex, body mass index, smoking status, diabetes mellitus, and CV disease. Identification was performed blinded for sortilin levels.

Cardiovascular In Depth Assessment (CARVIDA) cohort
The CARVIDA is a substudy of the German Chronic Kidney Disease study. 13 Only samples of CARVIDA patients included in the trial in Aachen, Germany, were used (n ¼ 78). Computed tomographic imaging was performed on a Dual Source CT scanner (SOMATOM Definition Flash or Force; Siemens), as previously described. 14

2020.
Human tissue Femoral arteries were obtained during autopsies from patients with and without CKD from RWTH Q16 Aachen University, Germany. The study was approved by the ethical committee of the RWTH Aachen University (ethical votes EK180/14 and EK239/11) and performed according to the Declaration of Helsinki.

Ex vivo carbamylation
Recombinant proteins were carbamylated in vitro by O-methylisourea bisulfate solution (pH 11.0) at 25 C for 3 hours, as previously described. 15 Matrix-assisted laser desorption/ionization (MALDI)-time-offlight mass spectrometry PTM was identified using MALDI-time-of-flight (TOF) mass spectrometry (MS; Ultraflex III; Bruker-D Q17 altonic), as previously described. 16 MS imaging of human vessel sections Tissue sections were analyzed with Rapiflex (Bruker-Daltonic) in positive reflector mode, in a 600-to 3000-dalton mass range and a grid size of 30 mm.

Statistical analysis
Experimental study data are presented as mean AE SD; n indicates the number of independent experiments or number of patients. Normality was tested using the Shapiro-Wilk test, and quantilequantile plot and variance heterogeneity were tested using the Brown-Forsythe test. A paired or unpaired 2-tailed Student t test with equal or unequal variances was performed to compare 2 groups. For comparison among $3 treatment groups, 1-or 2-way analysis of variance followed by Tukey post hoc was performed for data with normal distribution and equal variance. Data with skewed distribution were assessed by the Kruskal-Wallis test followed by Dunn post hoc test. Data with unequal variances were tested by Welch analysis of variance followed by Dunnett T3 post hoc test.
In the clinical studies, continuous data are presented as mean AE SD when normally distributed or as median and interquartile range for variables with skewed distribution. Categorical data are presented as percentage. Pearson/Fisher c 2 test was used to study the association between categorical variables and unpaired Student t test or Mann-Whitney U test for continuous variables. Differences between 3 groups were compared using 1-way analysis of variance followed by Sidak post hoc test. Least-square means multivariate-adjusted numbers of carbamylated sortilin residues were calculated using generalized linear models, as described previously. 11,17,18 Coronary artery calcification (CAC) volume was log-transformed (i.e., natural logarithm, ln; ln[CAC þ 1]) to reduce skewness. Bivariate correlation was assessed using Eta (1 nominal variable and 1 metric variable) or Pearson correlation coefficients (if both variables were metric). Change in CAC volume per year was calculated as follows: [(CAC follow-up -CAC baseline) / follow-up time in months] * 12. Analysis of variance with change in CAC volume as the dependent variable and a fixed-effect term for carbamylation (yes vs. no) was used for the analysis of CAC progression. In addition, this model was adjusted by the use of analysis of covariance for the covariates CAC at baseline and the variables listed in Supplementary Table S7 Q18 . P < 0.05 was considered statistically significant. Statistical analyses were performed using GraphPad Prism (Prism Software Inc., version 9) or SPSS (version 26.0).

Carbamylated sortilin increases with kidney function decline
Initially, we assessed sortilin serum levels in patients with CKD from the CARE FOR HOMe study 11 and found increased sortilin levels compared with a matched control group with normal kidney function from the Dan-NICAD 1 trial (Table 1). Circulating proteins are prone to PTM in patients with CKD Q20 . 5 Therefore, we performed a detailed mapping of PTM residues of circulating sortilin in participants of the CARE FOR HOMe study and healthy control subjects (Supplementary Table S1) using MALDI-TOF/TOF-MS. Compared with control subjects, CKD patients had 8 of the 30 lysine residues that were predominately carbamylated in the extracellular domain of sortilin (Supplementary  Table S3 Q21 and Supplementary Figure S1). Representative MS spectra from a control subject and a patient with CKD are illustrated in Figure 1a and b. The specificity of the signal was supported by MALDI-TOF/TOF-MS/MS spectra (Supplementary Figure S2A). Quantification revealed a CKD stage-dependent increase of carbamylated residue number ( Figure 1c) and intensity of carbamylated sortilin peptides ( Figure 1d). Lysine residue 205 was equally modified in all CKD stages, whereas residues 95, 260, and 294 were more often modified in advanced CKD stages ( Figure 1e).
Next, we assessed associations between sortilin carbamylation and baseline characteristics (Supplementary Table S2) in the CARE FOR HOMe study. We observed an agedependent increase of sortilin carbamylation residues ( Table 2). Lower kidney function, based on eGFR based on serum cystatin c and creatinine (eGFR cys-crea ), and higher urea and N-terminal pro-brain natriuretic peptide (NT-proBNP) levels were associated with higher carbamylated sortilin residues ( Table 2). The associations remained significant after adjustment for age and gender Q22 ( Table 2). CKD patients under aldosterone antagonist medication displayed reduced sortilin carbamylation (Table 2), whereas there was no difference in eGFR cys-crea (no, 50.6 AE 23.2 ml/min per 1.73 m 2 ; yes, 43.5 AE 22.9 ml/min per 1.73 m 2 ; P ¼ 0.236) and urea (no, 68.6 AE 40.1 mg/dl; yes, 72.3 AE 40.9 mg/dl; P ¼ 0.729) between patients with or without aldosterone antagonists. Besides urea, myeloperoxidase may mediate protein carbamylation in CV disease. 19 However, we found no association between total levels and activity of myeloperoxidase and sortilin carbamylation (Supplementary Figure S3A and B).
Furthermore, we assessed the presence of carbamylated sortilin in human femoral arteries using MS imaging. Carbamylated peptide SEDYGK*NFK* (m/z 1172) was highly present in calcified femoral arteries from CKD patients and absent in noncalcified femoral arteries ( Figure 1f). MS/MS spectra supported the identification of SEDYGK*NFK* (Supplementary Figure S2B). SEDYGK*NFK* is located close to calcified areas in the tunica media ( Figure 1g). In contrast, the mass-signal intensity of non-post-translationally modified peptide SEDYGKNFK (m/z 1086) was higher in control arteries ( Figure 1f).
Taken together, compared with controls with normal kidney function, patients with CKD have higher sortilin serum levels and exhibit post-translational carbamylated sortilin in the circulation, which can also be detected in the vasculature.

Carbamylated sortilin promotes smooth muscle cell calcification
Given our finding that carbamylated sortilin localized to calcified areas, we next assessed the effect of sortilin carbamylation on vascular calcification in vitro and ex vivo. To determine the functional relevance of sortilin carbamylation, we induced in vitro carbamylation of recombinant sortilin (Sort Carb ) by urea and detected a similar carbamyl-lysine residue pattern as detected in humans in vivo (Supplementary Figure S4A and Supplementary Table S4). In vitro, carbamylation did not alter the protein integrity of Sort Carb compared with control mock-modification (Sort Co ), as assessed by gel electrophoresis and Western blot (Supplementary Figure S4B and C).
Our data indicate that carbamylated soluble sortilin directly affects vascular cells and promotes vascular calcification in vitro and ex vivo.  Figure S1). (f,g) Matrix-assisted laser desorption ionization-imaging mass spectrometry heat map image shows the distribution/presence of the indicated sortilin peptides (m/z 1086, SEDYGKNFK; m/z 1172, SEDYGK*NFK*; asterisk indicates lysine carbamylation) in a calcified femoral artery from CKD patients and non-CKD controls. Intensity relative within one image in arbitrary units (AUs). Representative images from 3 cases per group. Bar c l i n i c a l i n v e s t i g a t i o n

Carbamylated sortilin alters ligand binding
Sortilin is a coreceptor for several ligands. 8 Carbamylation of lysine significantly alters the properties of the lysine sidechain, whereby the positively charged amino group is replaced by a bulky polar group, potentially affecting ligand binding. We examined the interactions of the most frequently modified sortilin lysine sidechains found in CKD (amino acid sequence [crystal structure sequence]), Lys95(62), Lys205(172), Lys260(227), and Lys294(261), by in silico modeling to predict possible structure-function consequences of lysine carbamylation. We used the existing sortilin ectodomain crystal structure from both the neurotensin-bound form at neutral pH (PDB Q24 file: 4PO7; Figure 3a) and the ligand-free form at acidic pH (PDB file: 6EHO; Supplementary Table S5).
At neutral pH, Lys95(62) forms a salt-bridge with Glu609(576) and a hydrogen bond with the main-chain carbonyl group of Arg90(57). At acidic pH, the interaction with Glu609(576) is tighter as the distance between the amino group of Lys95(62) and the carboxyl group of Glu609(576) now allows the formation of a hydrogen bond, whereas the interaction with Arg90(57) is not present. Therefore, carbamylation at Lys95(62) may destabilize both sortilin conformations.
Lys205 (172) is forming a salt bridge with Glu148(115) at neutral pH, whereas no interactions are found at acidic pH.
Thus, carbamylation at Lys205 would possibly favor the acidic pH conformation of sortilin. Lys294(261) has no interactions at both neutral and acidic pH.
Finally, at neutral pH, Lys260(227) interacts with neurotensin-Tyr11 in the ligand-binding site in the b-propeller tunnel of sortilin. As the acidic pH sortilin structure has no ligand affinity, we find no interactions of the Lys260(227) sidechain in this structure. The interaction with neurotensin-Tyr11 will be affected by carbamylation and might lead to a lower affinity for neurotensin. Interleukin 6 (IL-6), a known sortilin ligand, 20 binds sortilin through its C-terminal tail, which contains an arginine at the equivalent position of neurotensin-Tyr11. Thus, carbamylation of Lys260(227) might probably increase the affinity for IL-6, a potential contributor to vascular calcification. 21 Subsequent binding studies using surface plasmon resonance spectroscopy and Sort Co or Sort Carb revealed more efficient binding of IL-6 to Sort Carb (KD Q25 ¼ 23 nM) compared with nonmodified Sort Co (KD ¼ 141 nM), suggesting carbamylated sortilin as a potential IL-6 binding partner (Figure 3b and c). Progranulin, another known sortilin ligand, 22 bound to Sort Co , but a binding to Sort Carb could not be detected (Supplementary Figure S7A and B). Both Sort Co and Sort Carb are bound to human receptor-associated protein fused to a GST Q26 tag, 23 but not to GST alone, supporting the structural integrity of sortilin regardless of its carbamylation status and demonstrating the specificity of our experimental setting (Supplementary Figure S7C-F).
Next, we assessed whether the altered affinity to IL-6 might affect signaling pathways involved in vascular calcification. In calcifying hSMCs, the addition of IL-6 to Sort Carb promoted ALPL and RUNX2 mRNA expression (Figure 3d and e) as well as TNAP activity (Figure 3f), whereas it had no effect in combination with Sort Co .
These data indicate that sortilin carbamylation increases the ability of ligand binding to IL-6, which causes increased smooth muscle cell calcification.

Sortilin carbamylation is associated with vascular calcification in CKD patients
On the basis of our data that carbamylated sortilin residues were found to associate with CKD and calcification, we next studied CKD participants of the CARVIDA study, in whom computed tomographic scans quantified CAC. Like the CARE FOR HOMe study participants, most CARVIDA participants had 2 and 3 PTMs, with lysine residues 95 and 260 frequently affected (Supplementary Figure S8A). Compared with patients without carbamylated residues, patients with $1 sortilin carbamyl-lysine residues were older (P ¼ 0.002), had lower eGFR (P ¼ 0.008), were more frequently (former) smokers (P ¼ 0.023), and had significantly higher CAC volume (P < 0.001; Figure 4a and Supplementary Table S6).
Patients with carbamyl-lysine residue 260 exhibited the highest increase in CAC volume than patients with no carbamylated residue or carbamyl-lysine residues other than 260 (Supplementary Figure S8B).
Overall, the data indicate that post-translational carbamylated sortilin associates with CAC volume and its progression in patients with CKD.

DISCUSSION
In various experimental in vitro and in vivo studies and 2 independent prospective cohorts, we demonstrate that patients with CKD exhibit a specific pattern of posttranslational carbamylated sortilin lysine residues in the circulation, which can also be detected in the vascular wall. Furthermore, sortilin carbamylation was associated with CAC in CKD patients, independent of age, kidney function, and other risk factors for calcification. Mechanistically, we could show that sortilin carbamylation increases its ability to bind IL-6 and acts directly on vascular cells to promote vascular calcification (Figure 4d). Taken together, this study revealed a novel nontraditional risk factor for calcification, which potentially contributes to the high burden of CV diseases in CKD.
Protein carbamylation can be mediated by cyanate, which is developed during the spontaneous decomposition of urea or is generated by myeloperoxidase-catalyzed thiocyanate oxidation at sites of inflammation. 24 We showed that CKD patients with higher urea levels due to reduced kidney function exhibited more sortilin carbamylation residues than healthy controls. However, we found no association between total levels and activity of myeloperoxidase and sortilin carbamylation, suggesting uremia as a primary driver for sortilin carbamylation in CKD. In addition, our data demonstrate that age, NT-proBNP, and a lower intake of aldosterone antagonists are associated with more carbamylated residues. Our data are consistent with previous reports that identified aging as a significant mediator of protein carbamylation. 25 It has been shown that the antihypertensive drugs hydrochlorothiazide and amlodipine affect homocitrulline levels, a urea cycle-related amino acid, and carbamylation-derived product. 26 Whether intake of aldosterone antagonists lowers homocitrulline, and thus reduces carbamylated residues in CKD patients, is unknown. Drechsler et al. demonstrated a correlation between serum carbamylated albumin and NT-proBNP, which was associated with heart failure in dialysis- dependent CKD patients. 27 Whether carbamylation of sortilin may also serve as a risk factor for heart failure is not known, but increased coronary calcification with potentially subsequent harmful effects on heart function could explain the association between carbamylated sortilin residues and NT-proBNP levels in our study.
Recent studies identified the role of sortilin in the pathogenesis of vascular and metabolic disorders but mainly focused on tissue expression of sortilin rather than exploring the function of the soluble form. 7 Circulating sortilin may originate from cellular shedding of their luminal domain, as demonstrated in vitro, [28][29][30] and from secreted sortilin-packed extracellular vesicles. 9,31 Carbamylation leads to alterations in charge, structural, and functional properties of proteins, mediating loss of function and potentially pathophysiological cellular and molecular responses. 5 We established for the first time a specific lysine residue carbamylation pattern of circulating sortilin in CKD, using a targeted MS approach, and predicted functional and structural alterations by in silico modeling. We propose that lysine carbamylation will favor the acidic pH conformation of sortilin, affecting the pHdependent ligand affinity of sortilin. Previous crystal structure studies demonstrated that at pH 5.5, which represents an environment similar to that of late endosomes, sortilin undergoes conformational changes and dimer formation, making known binding sites unavailable for ligand binding. 32,33 Lysine 260 forms a hydrogen bond to a tyrosine of neurotensin inside the b-propeller tunnel, intimating that other ligands that bind similarly to neurotensin would also be affected. Our investigation found increased affinity of IL-6 to carbamylated sortilin, suggesting participation in IL-6 signaling pathways. Several studies suggested that IL-6 may contribute to vascular calcification in CKD. 22,34,35 Previously, a study on cellular sortilin found that phosphorylation of the intracellular domain of sortilin promoted vascular calcification. 9 Herein, we found that posttranslational modification of soluble sortilin plays a functional role in calcific vascular pathology. Soluble carbamylated sortilin promoted vascular calcification in vitro and ex vivo by increasing osteogenic target genes, TNAP activity, and matrix mineralization. Moreover, the addition of IL-6 to carbamylated sortilin further increased osteogenic target genes and TNAP activity.
Carbamylation has been associated with extracellular matrix alteration, 36 oxidative stress, 25 endothelial dysfunction, 37 and atherosclerosis, 38 all involved in calcification initiation and progression. Although Mori et al. reported that protein carbamylation promotes vascular calcification through carbamylation of mitochondrial proteins, 6 no specific target proteins were identified. A recent study demonstrated that carbamylation of uromodulin resulted in a loss of its anticalcific properties in vitro. 39 Both studies used Western blot to detect carbamylation in general. We demonstrated that in vitro carbamylated sortilin mimics the specific in vivo lysine modification pattern found in CKD patients using MS. Thus, our study provides clues to specific mechanisms connecting CKD, protein carbamylation, IL-6, and calcification.
Previously, high serum sortilin levels were associated with both abdominal aortic calcification and CV events independent of traditional Framingham risk factors in a community-dwelling cohort of men aged >50 years. 10 In low-to intermediate-risk chest pain patients, sortilin levels did not associate with the severity of coronary artery disease. 40 In this study, we could show that patients with CKD have higher sortilin levels than controls with normal kidney function. Sortilin levels did not increase with decreased kidney function once a patient had a CKD with eGFR <60 ml/min per 1.73 m 2 . However, we observed an increase of carbamylated sortilin residues with decreased kidney function. Similarly, Kalim et al. reported no association between total and carbamylated albumin levels in CKD patients. 41 Sortilin carbamylation, resulting from impaired kidney function, might further augment vascular calcification related to CKD-associated mineral disturbances. Our study has several strengths and limitations. Strengths include analyzing data obtained in vitro, ex vivo, and in patients using different experimental approaches. The careful characterization of patients with CKD, including standardized questionnaires to assess participants' characteristics, in-person study visits conducted by trained study nurses, and measurements of many laboratory values and calcification, is a particular strength of the study. The small cohort may appear to limit our study; however, we used independent CKD cohorts with follow-up data for calcification to validate the carbamylation status of sortilin by laborious targeted MS. The clinical findings were made in German CKD patients and may thus restrict the generalizability of the findings to other countries or ethnicities. Finally, although the study design allowed adjustment for many important confounders, residual confounding, as in any observational study, cannot entirely be ruled out.