Urinary Exosomal Cystatin C and Lipopolysaccharide Binding Protein as Biomarkers for Antibody−Mediated Rejection after Kidney Transplantation

We aimed to discover and validate urinary exosomal proteins as biomarkers for antibody−mediated rejection (ABMR) after kidney transplantation. Urine and for-cause biopsy samples from kidney transplant recipients were collected and categorized into the discovery cohort (n = 36) and a validation cohort (n = 65). Exosomes were isolated by stepwise ultra-centrifugation for proteomic analysis to discover biomarker candidates for ABMR (n = 12). Of 1820 exosomal proteins in the discovery cohort, four proteins were specifically associated with ABMR: cystatin C (CST3), serum paraoxonase/arylesterase 1, retinol-binding protein 4, and lipopolysaccharide−binding protein (LBP). In the validation cohort, the level of urinary exosomal LBP was significantly higher in the ABMR group (n = 25) compared with the T-cell-mediated rejection (TCMR) group and the no major abnormality (NOMOA) group. Urinary exosomal CST3 level was significantly higher in the ABMR group compared with the control and NOMOA groups. Immunohistochemical staining showed that LBP and CST3 in the glomerulus were more abundant in the ABMR group compared with other groups. The combined prediction probability of urinary exosomal LBP and CST3 was significantly correlated with summed LBP and CST3 intensity scores in the glomerulus and peritubular capillary as well as Banff g + ptc scores. Urinary exosomal CST3 and LBP could be potent biomarkers for ABMR after kidney transplantation.


Introduction
Kidney transplantation is a cornerstone treatment for end-stage renal diseases, regardless of their etiology. Advances in immunosuppressive regimen and our understanding of the biological processes in transplantation across immunologic barriers have improved the success of kidney transplantation in the past decades; however, injury to the graft remains a critical hurdle to long-term graft survival [1][2][3][4]. Early detection of graft pathology and early intervention and treatment are crucial to a long-lasting functional graft. Currently, the gold standard in diagnosing kidney allograft pathology is graft biopsy, which is impractical for routine and repetitive monitoring due to its invasiveness, risk of complications, and limitation in histological interpretation [5].
Antibody-mediated rejection (ABMR) is the leading cause of late allograft failure after solid organ transplantation and contributes to two-thirds of kidney transplant failure [6,7].

Patients
All consenting kidney transplant recipients who had a post-transplant for-cause biopsy at Asan Medical Center between January 2017 and April 2020 for clinical reasons such as proteinuria or deterioration in function were recruited for this observational case-control study. Multiple solid organ transplant recipients and pediatric renal transplant recipients were excluded from this study. Urine samples were collected at 2-3 h before the indication biopsy. In addition, living kidney donors were enrolled as healthy non-transplant controls in this study, and urine specimens were collected just before donor nephrectomy.
Patients in the discovery cohort were classified into 5 groups according to their allograft histopathology: ABMR group (n = 12), T-cell mediated rejection (TCMR) group (n = 8), BK virus nephropathy (BKVN) group (n = 5), no major abnormality (NOMOA) group (n = 11), and donor group (n = 24). In BKVN, tubular cell injury and interstitial inflammation are often observed, potentially leading to the misdiagnosis of TCMR. Therefore, specimens with BKVN were included to find biomarker candidates to distinguish BKVN from TCMR. In the validation cohort, the ABMR, TCMR, NOMOA, and control groups had 25, 10, 19, and 11 recipients, respectively, of which the control group was composed of those who had stable renal function without a history of a rejection episode. This study received approval from the institutional review board of Asan Medical Center and all patients provided written informed consent. No organs/tissues were procured from prisoners. All living donor nephrectomies were performed at Asan Medical Center, while kidneys from deceased donors were procured at several centers in South Korea. All deceased donors were strictly managed by a government organization (KONOS, Korean Network for Organ Sharing) without any organ trade or illegal distribution.

Histopathology
For-cause biopsies were performed by an ultrasound-guided percutaneous puncture. Each biopsy was stained with hematoxylin and eosin, Masson trichrome, periodic acid-Schiff, and Jones-methenamine silver for interpretation. C4d immunohistochemistry (1:100, rabbit polyclonal; Cell Marque, Rocklin, CA, USA) was performed on paraffin-embedded, formalin-fixed specimens using the Ventana Medical Systems (Tucson, AZ, USA) according to the manufacturer's protocol [19]. Every allograft biopsy specimen was evaluated for histologic characteristics according to the Banff 2015 criteria [20] by two renal pathologists (Y. Park and H. Go).

Urine Processing and Exosome Extraction
Exosomes were isolated by step-wise ultracentrifugation as previously reported [11,21] with a few modifications. Briefly, urine samples (≥30 mL) were collected by simple voiding and protease inhibitor (combination of 4-(2-aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF-HCl, Sigma-Aldrich, St. Louis, MO, USA), leupeptin-hemisulfate (Sigma-Aldrich, St. Louis, MO, USA), and NaN3 (Sigma-Aldrich, St. Louis, MO, USA)) were immediately added. To remove urinary sediments, the samples were centrifuged at 4000 rpm for 15 min at 4 • C and the resulting supernatants were collected and stored at −80 • C until exosome extraction. Frozen urine samples (15 mL) were thawed and vortexed for 1 min, centrifuged at 17,000 g for 15 min at room temperature, and the resulting supernatant was collected and ultracentrifuged at 200,000 g for 70 min at room temperature using a Beckman Coulter Optima L-80xp ultracentrifuge, rotor SW40Ti (Beckman Coulter; Brea, CA, USA). The resulting supernatant was discarded and the pellet was dissolved in 11 mL DPBS for washing, which was ultracentrifuged again at 200,000× g for 70 min at room temperature. The supernatant was discarded and the resulting pellet of exosomes was used for protein isolation or quantification of exosome particles.

Sample Preparation for Proteome Analysis
Exosome fraction was lysed using lysis buffer (50 mM Tris-HCl (pH 8.5), 5% SDS) and quantified using BCA assay. A measure of 100 ug of exosome proteins was digested into peptides using the amicon-adapted enhanced FASP method [22], and salt was removed by the C18 desalting cartridge (Sep-Pak C18 1 cc, Waters, Milford, MA, USA). For details, refer to previously published papers [23,24]. Peptide mixture was dried and stored at −20 • C until proteome analysis. Prior LC-MS analysis, the dried peptide mixture was reconstituted with 0.1% formic acid.

Nano-LC-ESI-MS/MS Analysis
Exosome peptide mixture was separated using Dionex UltiMate 3000 RSLCnano system (Thermo Fisher Scientific, Waltham, MA, USA). The reconstituted peptide sample was injected into a C18 Pepmap trap column (20 mm × 100 µm i.d., 5 µm, 100 Å; Thermo Fisher Scientific) and separated by an Acclaim™ Pepmap 100 C18 column (500 mm × 75 µm i.d., 3 µm, 100 Å; Thermo Fisher Scientific) over 200 min (350 nL/min) using a 0-48% acetonitrile gradient in 0.1% formic acid and 5% dimethyl sulfoxide (DMSO) for 150 min at 50 • C. The LC was coupled to a Q Exactive mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) with a nano-ESI source. Mass spectra were acquired in a data-dependent mode with 20 data-dependent MS/MS scans. The target value for the full scan MS spectra was 3,000,000 with a maximum injection time of 100 ms and a resolution of 70,000 at m/z 400. The ion target value for MS/MS was set to 1,000,000 with a maximum injection time of 50 ms and a resolution of 17,500 at m/z 400. Dynamic exclusion of repeated peptides was set to 20 s.

Database Searching and Label Free Quantitation
The MS/MS spectra from exosome peptide mixture were processed using the Se-questHT on Proteome Discoverer (version 2.2, Thermo Fisher Scientific, Waltham, MA, USA) with the SwissProt human protein sequence database (May 2020). Precursor mass tolerance and MS/MS tolerance was ±10 ppm and 0.02 Da, respectively. A fixed modification was set as default including cysteine carbamidomethylation, and variable modifications were set to be n-terminal acetylation and methionine oxidation with 2 miscleavages. The false-discovery rates (FDR) were set at 1% for the peptides using "Percolator" [25]. Peptide filters which included the peptide confidence, peptide rank, score versus charge state and search engine rank were set at the default values. Label-free quantitation (LFQ) was performed using the peak intensity for unique and razor peptide of each protein and excluded peptides including methionine oxidation.

Biomarker Candidate Selection
The selection process of candidate biomarkers in the discovery cohort is summarized in Figure 1. To select specific biomarkers representative of each pathologic group, we selected proteins with significantly different mean scaled abundance compared with the NOMOA and DONOR groups using Student's t-test. We also selected proteins that showed more than a 2.0-fold difference compared with the NOMOA group using the Volcano plot. Thereafter, the proteins that overlapped in the two analyses were selected. Among them, we selected up-regulated proteins with a higher abundance ratio (fold increase >2.5) and adjusted p value, and the results were visualized in a heatmap format using MeV software, version 4.9.0. We also selected overlapping proteins in the other two pathologic groups with a significant difference in abundance ratio (fold increase >2.5).

Evaluation of Exosomes
For immuno-gold labeling, we used a modified whole mount immuno-gold labeling method [26]. The specimens were viewed under a transmission electron microscope at 80 kV (Hitachi H-7600, Hitachi, Chiyoda City, Tokyo). The distribution of the size of exosomes was measured by dynamic light scattering using the Zeta Sizer Nano ZS apparatus (Malvern Instruments, Malvern, UK). The measurement runs were performed using standard settings. PE-conjugated anti-CD63 antibody (BD Biosciences San Jose, CA, USA) was used for exosome surface staining, and the stained samples were analyzed by flow cytometry (Beckman Coulter, Brea, CA, USA) and the FlowJo program.

Western Blot and Immunohistochemistry
Detailed methods for Western blot analysis (WB) and immunohistochemistry (IHC) are provided in the Supplementary Material. WB and IHC were performed using appropriate antibodies (Table S4).

Statistical Analysis
Categorical variables are summarized as absolute and relative frequencies and the differences among values were analyzed using the Chi-squared test or Fisher's exact test. Quantitative variables in each group are presented as the mean and standard error of the mean (SEM). Differences among means were compared using Student's t-test, analysis of variance (ANOVA), or Kruskal-Wallis test. Bonferroni post hoc tests were performed when the differences between groups were statistically significant. To assess the ability of each biomarker for discriminating between NOMOA and ABMR, receiver operating characteristic (ROC) curves were generated and the area under the curve (AUC) was obtained for each biomarker. The predictive probability of combined biomarkers was assessed using binary logistic regression. Spearman correlation was used to calculate the correlation between two parameters and was reported with r values. Statistical significance was set at p value of 0.05. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 21 (IBM Corp., Armonk, NY, USA), R software version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria), and GraphPad Prism 5 (GraphPad Software Inc. San Diego, CA, USA)

Patient Characteristics
Baseline characteristics of patients in the discovery and validation cohorts are summarized in Table 1. The interval between kidney transplantation and for-cause biopsy was significantly longer in the ABMR group compared with other groups in the discovery cohort (p = 0.003) as well as in the validation cohort (p = 0.002). In addition, the proportion of recipients without an induction regimen was significantly higher in the ABMR group than in the other groups in both cohorts. The histologic characteristics of the groups were assessed according to the Banff classification (Table S1).

Discovery of Four Biomarkers for Discriminating Antibody-Mediated Rejection
The study design for identifying urinary exosomal proteins in the discovery and validation cohort is presented in Figure 2. Exosomes in urine were derived from 5 groups and detected for label-free quantitative LC-MS/MS proteomic analysis. A total of 4193 urinary proteins were identified, of which 1820 exosomal proteins were identified in the five groups. We quantitatively compared the abundance of urinary exosomal proteins in each group with those in the NOMOA and DONOR groups. In total, 63, 108, and 53 exosomal proteins in TCMR, ABMR, and BKVN compared to NOMOA and DONOR groups, respectively, were significantly different from the others ( Figure 3A). We selected 108 proteins with significantly different scaled abundance in the ABMR group compared with the NOMOA and DONOR groups and 834 proteins with more than a 2-fold difference compared with the NOMOA group ( Figures 2B and 3B). As a result, 46 proteins were shown to overlap between the two samples ( Figure 3B). The fold change (FC) ratios of abundance in these 46 proteins were transformed by log 2 . The FC ratios were separated and visualized by Volcano plots according to log 2 FC 1.5 ( Figure 3C). Based on the abundance ratio, excluding proteins with missing values, we selected 18 significantly upregulated proteins with a log 2 > 1.5 and adjusted p < 0.001 compared to the NOMOA group, and their abundance ratios were visualized by a heatmap ( Figure 3D and Table S2). Among the 18 up-regulated proteins, cystatin-C (CST3), serum paraoxonase/arylesterase 1 (PON1), retinol-binding protein 4 (RBP4), and lipopolysaccharide-binding protein (LBP) were shown to have significantly higher abundance ratios in the ABMR group compared with the TCMR and BKVN groups ( Figure 3E). The ROC curves for each exosomal protein in differentiating ABMR from other allograft pathologies in the discovery cohort are shown in Figure 3F,G, and the AUC, 95% confidence interval (CI), cut-off value, specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) for each exosomal protein are described in Table S3. LBP and CST3 had higher AUC values than PON1 and RBP4 did. LBP and CST3 were also present in the published common proteins of reported exosomal proteins in the Exocarta, Vesiclepedia, and Urinary exosomal protein database. Urinary exosomal LBP and CST3 were the most potent biomarker for differentiating ABMR from other pathologies.

Validation of the Four Biomarkers for Discriminating Antibody−Mediated Rejection
The abundance of the four urinary exosomal proteins in the validation cohort was quantified by Western blot. The fold change of LBP was significantly greater in the ABMR group than in all other groups (control; 3.24 ± 0.83, NOMOA;13.08 ± 8.23, and TCMR; 7.15 ± 5.08) ( Figure 4A,C). The fold change of CST3 in the ABMR group (43.65 ± 13.19) was also significantly greater than in the control (9.08 ± 2.42) and NOMOA (9.87 ± 3.05) groups, but not significantly greater than the TCMR group (39.45 ± 31.18) ( Figure 4A,C) (Figure 4B,C). The ROC curves for each exosomal protein in differentiating ABMR from NOMOA in the validation cohort are shown in Figure 4D and Table S3. When compared with CST3, LBP was a potent biomarker for differentiating ABMR from TCMR as well as from the control and NOMOA groups ( Figure 4E).   (A,B). The results of Western blot analysis were assessed after normalization of means for the control group. Statistical significance was determined using two−tailed Mann−Whitney U test. * p < 0.05 ** p < 0.01 *** p < 0.001. (D,E) AUC of biomarker candidates in the independent validation cohort. D. ROC curves of single and two combined proteins between NOMOA and ABMR groups. (E) ROC curve of LBP for discriminating between ABMR and TCMR. ABMR, antibody−mediated rejection; NOMOA, no major abnormality. TCMR, T cell−mediated rejection; AUC, area under the ROC curve. CST3, Cystatin C; PON1, paraoxonase/arylesterase 1; RBP4, retinol−binding protein 4; LBP, lipopolysaccharide−binding protein; AUC, area under the ROC curve; 95% CI, 95% confidence interval.

Protein-Protein Interaction Analysis of LBP and CST3
The protein-protein interaction network for LBP and CST3 showed a total of 20 interacting genes. The common interactors between LBP and CST3 were shown to have functional relationships with each other, such as leukocyte activation, cell activation involved in immune response, vesicle-mediated transport, and myeloid leukocyte activation ( Figure 7A). LBP and CST3 were responsible for immune response and cell activation by forming a network with the related factors such as CD14 (monocyte differentiation antigen), TLR4 (Toll-like receptor 4), and IL6 (interleukin-6). KEGG analysis for interaction partners showed that differentially expressed genes were abundant in pathways which are known to be associated with rejection such as Toll-like receptor [27], NF-kB [28], HIF-1 [29], apoptosis [30], and PI3K-AKT signals [31] (Figure 7B).

Combined Use of LBP and CST3 as Biomarkers for ABMR
Considering that LBP and CST3 have the highest AUC among four biomarkers, we combined these two biomarkers to improve the efficacy for discriminating ABMR from NOMOA. Estimating predicted probability from logistic regression was used to establish a diagnostic model that could evaluate whether the combination of the two biomarkers could improve the diagnostic value. The combination of two biomarkers has the highest AUC than each biomarker alone for discriminating ABMR from NOMOA ( Figure 4D and Table S3). However, it was not shown that combined biomarkers are superior to LBP alone for discrimination between ABMR and TCMR. The prediction probability of urinary exosomal LBP and CST3 was significantly correlated with summed LBP (r = 0.677, p = 0.003) and CST3 (r = 0.801, p = 0.0001) ( Figure S2A,B) scores in the glomerulus and peritubular capillary as well as Banff g + ptc scores (r = 0.712, p = 0.001) in the histologic lesion of ABMR in the validation cohort ( Figure S2C). (A) Protein−protein interaction network of the interactors using the STRING online database. The confidence score was set to 0.4 and the interactors were set to no more than 20 with first shell of interaction for human. Red, green, blue, purple, yellow, light blue, and black lines indicate the presence of fusion evidence, neighborhood evidence, co−occurrence evidence, experimental evidence, text mining evidence, database evidence, and co-expression evidence, respectively. (B) Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of differentially expressed genes. A total of 28 pathways were significantly enriched. Asterisk indicates that the pathways are associated with allograft rejection according to previous studies. * pathways which are known to be associated with rejection.

Discussion
This study shows that urinary exosomal proteins in kidney transplant recipients present a rich source of personalized biomarkers for monitoring kidney allografts. Particularly, we discovered and validated urinary exosomal proteins CST3 and LBP, which could be combined to serve as a potent composite biomarker of ABMR in kidney transplant recipients. Additionally, we found that urinary exosomal CST3 and LBP are strongly correlated with expression in the glomerulus and peritubular capillary of kidney allografts, especially in recipients with ABMR. To our knowledge, this study is one of the few that discovered and validated urinary exosomal biomarkers for ABMR in kidney allografts.
As in Fekih et al.'s study, biopsy specimens with borderline TCMR were excluded from this study because of the ongoing controversies regarding the clinical and immunological importance of borderline TCMR [14,32]. In addition, those with mixed borderline TCMR and ABMR were classified into the ABMR group considering the rarity of cases of pure ABMR without TCMR. Different from the study by Fekih et al., however, biopsy samples with mixed TCMR and ABMR were excluded from this study to discover ABMR-specific biomarkers.
In our study, LBP was strongly expressed in the glomerulus and peritubular capillary as well as tubules and interstitium in the kidney allograft specimens from patients with ABMR. In addition, urinary exosomal LBP was significantly increased in patients with ABMR. LBP is mainly produced in hepatocytes and plays a crucial role in the innate immune response [33]. The LPS-LBP complex binds to CD14 and the MD-4/MD-2 complex, thereby activating signal transduction pathways and releasing cytokines and pro-inflammatory molecules [34,35]. Single-nucleotide polymorphisms in the LBP gene are associated with infectious diseases, inflammatory disease, metabolic disorders, and malignancy [33]. However, only a few studies have reported the association between LBP and rejection after solid organ transplantation. Gerlach et al. showed that plasma LBP levels increased during acute and chronic rejections after intestinal transplantation in rats [36], and Freue et al. reported that seven plasma proteins including LBP were increased in recipients with biopsy-proven acute rejection [37]. A case-control discovery study by the Genome Canada Biomarkers in Transplantation Group showed that LBP was one of the proteomic signatures in the plasma during early acute kidney allograft rejection. Our study is thus meaningful in that it demonstrated the possible use of LBP as an ABMR-specific biomarker. CST3 was shown to be effective for estimating the glomerular filtration rate and is thus considered a potent biomarker for acute kidney injury [38,39]. Recently, it was suggested that urinary exosomal CST3 mRNA expression levels are representative of the changes in renal mRNA and protein expression and may thus be a more specific biomarker of renal damage than glomerular-filtered free CST3 [40]. Similar to LBP, our results showed that CST3 and its interaction with LBP may be useful as a combined biomarker for ABMR. Future studies are needed to decipher the biological mechanism by which CST3 and LBP are involved in acute rejection after kidney transplantation. This study is limited by its single-center design and the small number of samples, especially for TCMR. Due to rare BKVN cases during the study period, the BKVN group was not included in the validation cohort. Owing to the small number of ABMR samples, it is not feasible to verify a difference between acute ABMR and chronic active ABMR in terms of urinary exosomal biomarkers. Moreover, this was a cross-sectional study and did not include serial urine samples collected from various time points following kidney transplantation. Therefore, we are currently conducting a prospective study by collecting serial urine samples to assess how early these biomarkers can predict ABMR before pathologic diagnosis. For clinical application of exosomal urinary biomarkers, a modular microfluidic platform can be applied to isolate and enrich exosomes from urine samples of kidney transplant recipients. For quantification of specific exosomal proteins related to pathologic status, the exosomes on microbeads can be directly subjected to the molecular analysis [41].

Conclusions
In conclusion, we discovered and validated urinary exosomal proteins LBP and CST3 as potent non-invasive biomarkers for ABMR in kidney transplant recipients. Moreover, the 20 molecules and 5 complex molecular pathways that were significantly associated with the interactions between LBP and CST3 might play important roles in the biological progression of ABMR and become new treatment targets.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/biomedicines10102346/s1, Supplementary Methods; Figure S1: Validation of proteomic biomarkers in urine and serum; Figure S2: Association of prediction probabilities of combined urinary biomarkers with expression in allografts; Table S1: Comparison of histologic findings of for-cause biopsies of each group; Table S2: List of eighteen urinary exosomal biomarker candidates; Table S3: Area under the curve of the receiver operating characteristics curves of the four candidate biomarkers in differentiating AMBR from NOMOA in the validation cohort; Table S4: List of antibodies used for Western blot analysis and Immunohistochemistry.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Written informed consent was obtained from the patients to publish this paper.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy of the participants.