Omics data integration analysis identified new biological insights into chronic antibody-mediated rejection (CAMR)

Abstract

Background

In the last two decades, many studies based on omics technologies have contributed to defining the clinical, immunological, and histological fingerprints of chronic antibody-mediated rejection (CAMR), the leading cause of long-term kidney allograft failure. However, the full biological machinery underlying CAMR has only been partially defined, likely due to the fact thatsingle-omics technologies capture only specific aspects of the biological system and fail to provide a comprehensive understanding of this clinical complication.

Methods

This study integrated mass spectrometry-based proteomic profiling of serum samples from 19 patients with clinical and histological evidence of CAMR and 26 kidney transplant recipients with normal graft function and histology (CTR) with transcriptomic analysis of peripheral blood mononuclear cells (PBMCs) from an independent cohort of 10 CAMR and 8 CTR patients. Data analysis was conducted using unsupervised hierarchical clustering (multidimensional scaling with k-means) and Spearman’s correlation test. Partial least squares discriminant analysis (PLS-DA) with the importance in projection (VIP) score identified key proteins differentiating CAMR from CTR. ELISA was used to validate the omics results.

Results

Proteomic analysis identified 18 proteins that significantly differentiated CAMR from CTR (p < 0.01): five were more abundant (CHI3L1, LYZ, PRSS2, CPQ, IGLV3-32), while 13 were less abundant (SERPINA5, SERPING1, KNG1, CAMP, VNN1, BTD, WDR1, PON3, AHNAK2, MELTF, CA1, CD44, CUL1). Transcriptomic profiling revealed 6 downregulated and 33 upregulated genes in CAMR versus CTR (p < 0.01). Notably, only 2 biological elements were significantly deregulated in both omics analyses: chitinase-3-like protein 1 (CHI3L1) and plasma protease inhibitor C1 (SERPING1). CHI3L1, previously associated with the severity of tissue damage in kidney diseases, was up-regulated in CAMR in both transcriptomics and proteomics, while SERPING1, a serine esterase inhibitor that blocks the classical and lectin pathway of complement, was up-regulated in CAMR in transcriptomics but down-regulated in proteomics. ELISA validated the omics results, and the ROC curve showed that CHI3L1 has good discrimination power between CAMR and CTR (AUC of ROC curve of 0.81).

Conclusions

Our multi-omics data, although performed in a relatively small cohort of patients, revealed new systemic biological elements involved in the pathogenesis of CAMR and identified CHI3L1 as a new potential biomarker and/or therapeutic target for this important clinical complication. Future validation of these findings in larger patient cohorts should be conducted to better evaluate their clinical utility.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12967-025-06203-0.

Background

In the last decade, many studies have been performed to define the main clinical characteristics and histological features of chronic antibody-mediated rejection (CAMR), a widespread long-term complication with an incidence range from 4.6 to 20.2% over 1 to 10 years [1–3], and to identify the leading causes and triggers of this condition. Several risk factors may contribute to the development of CAMR, including HLA class II mismatch (particularly HLA-DR/DQ) [4], younger recipient age [5], a history of T cell-mediated rejection [6–8], and poor adherence to maintenance immunosuppressive medications [4, 9].

However, the complete biological machinery associated with this clinical complication has only been partially defined.

To provide new molecular insights into the pathogenesis of CAMR, several studies using innovative technologies (mainly omics) along with bioinformatics tools have been recently conducted. Omics technologies, including transcriptomics, proteomics, and metabolomics, which respectively quantify the abundance of mRNA, proteins, and metabolites in cells, tissue extracts, or biological fluids, allow for the simultaneous generation of large datasets. These data can be valuable for better understanding the mechanisms underlying CAMR and for identifying potential diagnostic biomarkers and therapeutic targets. Most of these studies have revealed significant deregulation of circulating immune-inflammatory cells.

As reported by our group, deregulation of several type I interferon genes in both peripheral blood mononuclear cells (PBMCs) and CD4 T cells, and a reduction in circulating BDCA2(+) dendritic cells (type I interferon-producing cells) together with their increment in kidney tissue have been associated with CAMR [10].

In 2020, Rocchetti et al. performed a phosphoproteomic analysis in PBMCs of patients with biopsy-proven CAMR and revealed an increase in lactotransferrin, actin-related protein 2 (ARPC2), and calgranulin B in these patients compared with kidney transplant recipients with normal renal function, demonstrating a significant alteration of the cytoskeleton organization in circulating immune cells [11].

Recently, using microarray technology and CellCODE analysis to quantify the strength of relationships between the canonical gene sets of distinct immune cell types and a gene of interest, we revealed that the complement-related transcripts C1QA and C1QB were enhanced in B cells of patients with CAMR compared to those with normal kidney function [12].

Single-cell transcriptomic analysis in patients with CAMR revealed, for the first time, significant upregulation of genes encoding key regulators of FoxO signaling in naive B cells and for endoplasmic reticulum stress in CD8 T cells [13].

Additionally, the miRNA can contribute to this type of rejection [14]. A miRNome analysis identified miR-142-5p, which regulates the expression of genes involved in leukocyte activation, upregulated in PBMCs and kidney tissue of patients with CAMR but not in patients with acute rejection or chronic non-immunological renal failure [15].

Moreover, although these interesting findings have improved our understanding of CAMR, the molecular signature of this allograft complication remains incomplete, and none of the already identified biological elements have been introduced into clinical practice. Studies relying on single-omics technologies typically capture only one aspect of the biological system, limiting their ability to provide a comprehensive understanding of the molecular landscape of this clinical complication. Therefore, in this study, we combined transcriptomic and proteomic data from patients with CAMR to identify previously unrecognized key factors and novel diagnostic and therapeutic biomarkers for this complex condition. This approach can also assist clinicians in identifying patients at high risk for developing this complication, with no need to perform allograft biopsy, and provides a foundation for personalized medicine in kidney transplantation.

Materials and methods

Patients

We enrolled 19 kidney transplant recipients with clinical and histological evidence of chronic antibody-mediated rejection (CAMR) according to our standard clinical procedures, 26 kidney transplant recipients with normal graft function and histology (Control, CTR), and 19 healthy subjects matched for age and gender. The inclusion criteria were age > 18 years, follow-up time ≥ 6 months, and CAMR diagnosis according to the most recent Banff classification [16], characterized by chronic transplant glomerulopathy (cg score > 0), severe peritubular capillary (ptc score) changes, basement membrane multilayering (with or without C4d deposition in peritubular capillaries), and the presence of anti-HLA DSA. Patients were on calcineurin inhibitor (CNI)-based immunosuppressive therapy at the time of enrollment and had no systemic infections, systemic inflammatory diseases, or known or clinically suspected malignancy. The control group included kidney transplant recipients who underwent protocol graft biopsies, with normal renal function and histology, and no circulating anti-HLA antibodies.

The main demographic and clinical characteristics of the patients are summarized in Table 1.

Table 1.

Demographic and clinical characteristics of the patients

	CTR	CAMR	p-Value
Number (patients)	26	19
Age (years)	59.7 ± 13.1	55 ± 13.2	NS
Gender (M/F)	14/12	10/9	NS
Time since transplantation (years)	7.6 ± 7.7	9.6 ± 6.1	NS
Serum creatinine (mg/dl)	1.16 ± 0.18	1.71 ± 0.62	< 0.0001
Proteinuria (g/24 h)	0.12 ± 0.1	2.8 ± 2.9	< 0.0001

CTR: control, kidney transplant recipients with normal graft function and histology; CAMR: chronic antibody-mediated rejection

The sample sizes for controls and patients (n) were determined a priori based on data variability (d) and a predefined fold change requirement (f) of 0.5. The experimental design accounted for a beta error corresponding to 80% statistical power and an alpha error threshold of ≤ 0.05, adjusted for multiple comparisons using the Benjamini–Hochberg correction. The calculation was performed using the equation n = 1 + 2c(d/f)2, as proposed by Forshed [17], where c is a constant derived from the specified alpha and beta error values.

The study was conducted according to the latest version of the Declaration of Helsinki and was approved by the local ethics committee (Prot. N. 670/C.E.). Informed consent was obtained from all enrolled patients.

Mass spectrometry (MS) profile

Serum samples (2 µl) were denatured, reduced, and alkylated in 50 µl LYSE buffer (Preomics, Planegg, Germany, PO00032) for 10 min at 95 °C and 1000 rpm mixing. Then 50 µg of each sample was digested using the protein aggregation capture (PAC) method automated on a KingFisher™ Apex robot (ThermoFischer Scientific Instrument, Waltham, MA, USA) in the 96-well format as previously described [18]. Digestion was performed by adding trypsin and LysC (at a 1:50 w/w and 1:100 w/w ratio of enzyme to sample protein, respectively), mixing, and incubating at 37 °C overnight (ThermoFischer Scientific, Waltham, MA, USA, A41007). The resulting peptides were analyzed using a nano-UHPLC-MS/MS system with an Ultimate 3000 RSLC coupled to an Orbitrap Fusion Tribrid mass spectrometer (ThermoFischer Scientific Instrument, Waltham, MA, USA). Elution was performed using an EASY spray column 150 μm x 15 cm, 2 μm particle size (ThermoFischer Scientific, Waltham, MA, USA, ES906) at a flow rate of 1200 ul/min with a 45 min gradient consisting of 1 min of 2% buffer B (80% v/v acetonitrile, 5% v/v dimethyl sulfoxide, and 0.1% v/v formic acid), then increasing to 30% B over 34 min, with a further increase to 80% B in 3 min, followed by a 2 min wash at 80% B and a 5 min re-equilibration at 2% B. Mass spectrometry (MS) analysis was performed in data-independent acquisition (DIA) mode. Orbitrap detection was used for MS1 measurements at a resolution power of 120 K in the range between 375 and 1500 m/z, with a 300% normalized AGC target and 50 ms maximum injection time. Advanced Peak Determination was enabled for MS1 measurements. The FAIMS CV was set to 50 at the standard resolution. Precursors were selected for data-independent fragmentation in 40 windows of 15 m/z, with 2 m/z overlap. The HCD collision energy was set to 30%, and MS2 scans were acquired at a resolution of 30 k, 54 ms max. IT, and 1000% normalized AGC target. The data were acquired in the profile mode using positive polarity.

All raw DIA files were processed with Spectronaut version 17 [19] using a library-free approach (directDIA) under the default settings. The library was generated against the UniProt Human database (release UP000005640_9606 November 2022, 102572 entries). Carbamidomethylation was selected as a fixed modification, whereas methionine oxidation, N-terminal acetylation, and Deamidation (NQ) were selected as variable modifications. The False Discovery Rate (FDR) of the peptide-spectrum match (PSM) and peptide/protein groups were set to 0.01. For quantification, Precursor Filtering was set to Identified (Qvalue), and MS2 was chosen as the quantity MS-level. These parameters, which were obtained from the literature and internal testing, ensured optimal protein identification and quantification, and the reproducibility of data [20–22].

All chemicals used for proteomic analysis were purchased from ThermoFisher Scientific (Waltham, MA, USA), unless otherwise stated. All other chemical reagents were of analytical grade and were obtained from Merck (Darmstadt, Germany). All solutions were prepared using deionized water with a resistivity of not less than 18.2 MΩ cm–1.

Transcriptomic profile

We analyzed previously obtained gene expression microarray data (GeneChip Human Genome U133 Array; Affymetrix, Santa Clara, CA) from PBMCs isolated from 10 patients with CAMR (aged > 18 years) and 8 kidney transplant recipients with normal graft function and histology (CTR) [10]. At the time of enrollment, all patients with CAMR met the following inclusion criteria: clinical and histological evidence of CAMR according to the Banff 2011 criteria [23], including diffuse peritubular capillary C4d deposition (C4d3 according to Banff 2011) and circulating donor-specific antibodies.

The control group included kidney transplant recipients who underwent protocol graft biopsies, with normal renal function and histology, and no circulating anti-HLA antibodies.

Briefly, total RNA was extracted from a minimum of 10⁷ PBMCs with a commercially available kit (miRNeasy Mini Kit; Qiagen, Hilden, Germany). The total RNA concentration and integrity were assessed using a Nanodrop spectrophotometer ND-1000 (Thermo Scientific, Waltham, MA, USA) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA), respectively.

Results of the microarray experiment are available in the Gene Expression Omnibus (accession numbers: GSE51675).

ELISA

Chitinase-3 like-protein-1 (CHI3L1) levels were measured using a commercially available ELISA kit (AssayGenie, Dublin, Ireland, HUFI00083) following the manufacturer’s instructions in the entire enrolled patients’ population and controls.

Briefly, standards and samples were diluted in the appropriate buffer and incubated for 90 min. Next, standards and samples were removed, and the wells were washed three times in washing buffer, followed by incubation with 100 µL of the detection solution. The wells were washed again three times with washing buffer, and TMB substrate (Bio-Rad, Hercules, CA, USA, 1721066) was added. The reaction was quenched by adding a stop solution, and absorbance was measured at 450 nm using an iMark plate reader (Bio-Rad, Hercules, CA, USA). CHI3L1 content was measured using a standard curve run in triplicate. A box plot was created to visualize the protein concentration. The lower detection limit was defined as the lowest protein concentration that could be differentiated from the blank wells.

Statistical analysis

Briefly, after log2 conversion, identified proteins were filtered for 70% presence in at least one group. Then, missing values were imputed using a normal distribution, and the entire dataset was normalized using the quantile method and analyzed by unsupervised hierarchical clustering (multidimensional scaling with k-means) and Spearman’s correlation to identify outliers and dissimilarity between samples. The t-test was used for unpaired samples to identify proteins with significant changes in abundance. For the t-test, proteins were considered significantly differentially expressed between two groups with a power of 80% and an adjusted p-value ≤ 0.05 after correction for multiple interactions (Benjamini–Hochberg method). Volcano plots were used to visualize the t-test analyses, and the cut-off line was established using the function y =|c/(x − x0)|. In addition, partial least squares discriminant analysis (PLS-DA) with the variable importance in projection (VIP) score was used to identify a prioritized list of statistically significant proteins in the discrimination between CAMR and CTR. Furthermore, a receiver operating characteristic (ROC) curve analysis was performed to determine the power of discrimination of CHI3L1 between CAMR and CTR. The same analysis was performed for the PBMC transcriptomic profile.

Gene set enrichment analysis was performed to build a functional protein network based on their Gene Ontology (GO) annotation terms extracted from the Gene Ontology Consortium (http://www.geneontology.org/, accessed on 17 January 2024) using all statistically significant proteins. Biochemical pathways were visualized using PathVisio 4 version software. For ELISA, the Kruskal-Wallis test was used to assess differences in serum CHI3L1 concentration among the different sample groups. Results were expressed as median and interquartile range (IQR). A ROC curve was generated to assess the diagnostic accuracy of ELISA in the discrimination between the CAMR and CTR. The AUC values were classified as follows: 0.5, not discriminant; 0.5–0.6, fail; 0.6–0.7, poor; 0.7–0.8, fair; 0.8–0.9, good, and 0.9–1, excellent. The likelihood ratio was used to identify the diagnostic performance of ELISA. Two-sided p-values ≤ 0.05 were considered statistically significant. All statistical tests were performed using OriginPro (version 2024b) and R (version 4.3.3) [24].

Results

Proteome profile

A total of 1037 proteins were identified (Supplementary Table S1), among which 822 (79.3%) were common in both groups. Only 208 (20.1%) and 7 (0.7%) proteins were exclusive to CTR and CAMR, respectively. Most differentially abundant proteins can be categorized into three main molecular function annotation terms: binding, catalytic activity, and structural constituent.

The t-test was used to identify the proteins that discriminated CAMR from CTR. Of the 1037 proteins identified, 68 and 18 were statistically significant in CAMR versus CTR, with p-value <0.05 and <0.01, respectively (Supplementary Table S1 and Fig. 1A). The expression profiles of these 18 highly statistically significant proteins after Z-score normalization are visualized in the heatmap in Fig. 1B. Five proteins were more abundant (CHI3L1, LYZ, PRSS2, CPQ, IGLV3-32) whereas 13 were less abundant (SERPINA5, SERPING1, KNG1, CAMP, VNN1, BTD, WDR1, PON3, AHNAK2, MELTF, CA1, CD44, CUL1) in CAMR compared with CTR patients.

Fig. 1.

Differentially abundant proteins in CAMR versus CTR. (A) Volcano plot showing differentially abundant proteins in CAMR and CTR. Each dot represents a protein. The x-axis represents log2 fold change and the y-axis represents -log10 (p-value). Black lines represent the threshold of statistical significance after adjusting for multiple testing (p-value < 0.05 or <0.01). The red and blue dots indicate proteins that are more or less abundant in CAMR compared to CTR, respectively.. (B) Heatmap of highly significant proteins that discriminated CAMR from CTR. In the heatmap, red, white, and blue indicate positive, equal, and negative expression for each protein value. The dendrogram positioned above and to the left of the heatmap shows the results of the unsupervised hierarchical clustering analysis, which placed similar protein expression profile values close to each other

Furthermore, partial least squares-discriminant analysis (PLS-DA) was used to rank proteins according to their variable importance in the projection (VIP) score. In particular, for the VIP score, a higher value corresponds to the maximum discriminating power. The k-means analysis of the PLS-DA showed two distinct clusters corresponding to CAMR and CTR (Supplementary Figure S1).

Transcriptomic profile

We re-interrogated our previous gene expression microarray data obtained from 10 patients with biopsy-proven CAMR and 8 CTR.

Of the 41,000 genes, 84 and 39 were differentially expressed with p-value cut-off values of 0.05 and 0.01 (Supplementary Table S2), respectively (Fig. 2A). After Z-score normalization, the expression profile of these 39 statistically significant genes was visualized in the heatmap presented in Fig. 2B. Six genes were downregulated and 33 genes were upregulated in CAMR compared with CTR patients.

Fig. 2.

Differentially expressed genes in PBMCs of CAMR versus CTR. (A) Volcano plot showing differentially expressed genes in PBMCs from CAMR and CTR. Each dot represents a gene. The x-axis represents log2 fold change, and the y-axis represents -log10 (p-value). Black lines represent the threshold of statistical significance after adjusting for multiple testing (p-value < 0.05 or <0.01). Red dots represent genes that are upregulated in CAMR compared with CTR, whereas genes downregulated in CAMR are shown in blue. (B) Heatmap of highly significant genes separating CAMR and CTR. In the heatmap, red, white, and blue indicate positive, equal, and negative expression for each gene. The dendrogram positioned above and to the left of the heatmap shows the results of the unsupervised hierarchical clustering analysis, which placed similar gene expression profile values close to each other

In addition, the 39 genes could be categorized into 3 main groups: signaling, immune system, and complement cascade.

Notably, only 2 biological elements were significantly deregulated in both omics analyses: chitinase-3-like protein 1 (CHI3L1) and plasma protease inhibitor C1 (SERPING1). CHI3L1 was upregulated in CAMR in both transcriptomics and proteomics, whereas SERPING1 was upregulated in CAMR in transcriptomics but downregulated in proteomics (Figs. 1A and 2A).

Gene ontology enrichment analysis

To understand the biological roles of the identified proteins, we performed Gene Ontology (GO) based enrichment analysis. The comparison of CAMR versus CTR identified 64 significantly enriched GO annotation terms (Supplementary Table S3 and Fig. 3) that can be categorized into three terms: Immune System, Response to Stimuli, and Binding.

Fig. 3.

Gene ontology enrichment analysis. Gene ontology enrichment analysis of statistically significant transcriptome/proteome data identified 64 dysregulated biological processes in CAMR compared with CTR. The color map, ranging from red to blue, illustrates the order of statistically significant strength of the dysregulated pathways. The x-axis represents the ratio of the intersection sizes of each pathway (number of genes identified for each biological process)

Furthermore, we explored the interaction network between CHI3L1 and immune response pathways involved in allograft rejection in WikiPathways database. As shown in Fig. 4, CHI3L1 participates in the T-helper cell type 2 (Th2) inflammatory response and regulates inflammatory cell apoptosis, dendritic cell accumulation, and M2 macrophage differentiation. It interacts with CD55 by regulating the inhibition of C3 and C5 convertase.

Fig. 4.

Role of CHI3L1 in allograft rejection. This pathway. (adapted from https://www.wikipathways.org/instance/WP2328) illustrates the molecular interactions between CHI3L1 and the immune response for allograft rejection. CHI3L1 plays a role in T-helper cell type 2 (Th2) inflammatory response and IL-13-induced inflammation, regulating inflammatory cell apoptosis, dendritic cell accumulation, and M2 macrophage differentiation. It also plays a role in tissue remodeling by interacting with cell adhesion molecules (CAMs) and in the ability of cells to respond and cope with changes in their environment. CHI3L1 has been reported to promote renal fibrosis after kidney injury via activation of myofibroblasts.CHI3L1 binds to IL-13 and this complex interacts with IL-13Rα2 to activate the MAPK/Erk, Akt, and Wnt/β-catenin cell signaling pathways to regulate apoptosis, oxidant injury-induced cell death, pyroptosis and inflammasome activation. It interacts with CD55 by regulating the inhibition of C3 and C5 convertase. The proteins selected by proteomic analysis are shown by colored rectangles, and their expression are depicted with a pseudocolor scale (based on normalized Z-scores of protein abundance), red denoting high expression and blue denoting low expression level in CAMR versus CTR patients

ELISA of CHI3L1

To confirm the proteomic and transcriptomic data, serum CHI3L1 levels were measured by ELISA in all enrolled patients and 19 healthy subjects.

As shown in Fig. 5, CHI3L1 was more abundant in CAMR compared with CTR and healthy subjects (p < 0.0001), with values of 2.65 ng/ml (1.92–4.55), 1.69 ng/ml (1.27–2.09), and 1.12 ng/ml (0.87–1.24), respectively.

Fig. 5.

CHI3L1 protein levels by ELISA. (A) Box plot showing the median and interquartile range values of CHI3L1 serum content in CAMR, CTR and healthy subjects. CHI3L1 protein in serum from CAMR was statistically more abundant (p < 0.0001) than in CTR and healthy subjects. (B) Received operating characteristic (ROC) curve analysis confirmed the good discrimination power of CHI3L1 between CAMR and CTR

The assay sensitivity (confidence interval), specificity (confidence interval), and likelihood ratio for discrimination between CAMR and CTR were 68% (46–84), 85% (66–93), and 4.45, respectively. The areas under the curve (AUC), confidence interval, and p-value were 0.81, 0.69–0.94, and 0.0003, respectively. The ROC curve analysis indicated that CHI3L1 had good discrimination power between CAMR and CTR.

Discussion

At present, traditional clinical biomarkers (such as creatinine, daily proteinuria and donor-specific antibodies) and percutaneous allograft biopsy are largely employed to diagnose CAMR, but none of them can be considered early diagnostic tools. Their use may often result in delayed CAMR diagnosis (occurring when the graft damage is irreversible and advanced), with a consequent significant impact on allograft survival. Kidney biopsy remains invasive, costly, and requires specific technical skills that are not available in all nephrology/transplant centers. Moreover, the biopsy interpretation can vary between pathologists, and the actual Banff antibody-mediated rejection classification may be vulnerable to misinterpretation, which potentially has patient management implications [25].

Several potential noninvasive biomarkers of graft rejection have been developed [26]. Among these, donor-derived cell-free DNA (dd-cfDNA) and urinary exosome mRNA signatures have been tested for both active and chronic antibody-mediated rejection and are currently commercially available although not yet extensively introduced into clinical practice. Dd-cfDNA is fragmented extracellular DNA released into the bloodstream from cells undergoing apoptosis or necrosis. In kidney transplant recipients, levels > 1% of dd-cfDNA are associated with the development of allograft rejection [27, 28]. In a study of 280 kidney biopsies, dd-cfDNA levels were elevated in both DSA-positive and DSA-negative antibody-mediated rejection, suggesting that combining DSA and dd-cfDNA could enhance patient management [29]. In a recent randomized trial, 40 kidney transplant recipients with de novo DSA were assigned to either dd-cfDNA-guided biopsy (intervention group) or clinician-guided biopsy (control group) for 12 months. Patients with dd-cfDNA > 50 copies/mL underwent biopsy. Monitoring of dd-cfDNA reduced the time to diagnosis of active or chronic active antibody-mediated rejection [30].

Another promising noninvasive tool for detecting graft rejection is the analysis of urinary exosome mRNA signatures. A clinically ready-to-use biomarker for antibody-mediated rejection was identified by El Fekih et al., who selected a 5-gene exosome signature (CD74, C3, CXCL11, CD44, and IFNAR2) that could distinguish antibody-mediated rejection from T cell-mediated rejection with an AUC of 0.87, sensitivity of 87.5%, and specificity of 82.9% [31].

Moreover, there is a lack of high-quality evidence based on large randomized controlled trials to guide the optimal therapy for patients with this clinical complication, and most therapies for CAMR are opinion-based, and no US Food and Drug Administration (FDA) or European Medicines Agency (EMA) approved [32, 33].

Therefore, to identify early biomarkers and new therapeutic targets for more personalized medicine, we employed high-throughput omics technologies to simultaneously screen thousands of genes and proteins.

In particular, in our study, after a highly conservative statistical analysis, we identified 18 top significant proteins that were able to discriminate CAMR from CTR (5 more- and 13 less-abundant in CAMR). In transcriptomic analysis, 6 genes resulted up- and 33 were down-regulated in CAMR versus CTR. Interestingly, 2 biological elements were significantly deregulated in both omics analysis: chitinase-3-like protein 1 (CHI3L1) and plasma protease inhibitor C1 (SERPING1). CHI3L1 was upregulated in CAMR in both transcriptomics and proteomics, whereas SERPING1 was upregulated in CAMR in transcriptomics but downregulated in proteomics.

CHI3L1, also known as YKL-40, is a secreted glycoprotein involved in inflammation, tissue remodeling, and repair [34]. It is produced by various cells such as macrophages, neutrophils, fibroblast-like cells, and endothelial cells [34]. Synthesis and secretion of this protein are regulated by extracellular matrix (ECM) changes, miRNAs, growth factors, cytokines, stress, and drugs [34–36]. CHI3L1 overexpression has been observed in patients with inflammatory conditions, including asthma, sepsis, diabetes, hepatic fibrosis, and cardiovascular disease [37–41].

In kidney diseases, this protein is related to the severity of tissue damage. In patients with nephrotic syndrome, the blood level of CHI3L1 is associated with proteinuria [42], and in kidney transplant recipients, the level of CHI3L1 is higher in patients with diabetes, coronary artery disease, and proteinuria than in patients without these clinical complications [43, 44]. Moreover, urinary content of this glycoprotein could predict the occurrence of acute kidney injury (AKI) in patients admitted to an intensive care unit and was associated with chronic kidney disease (CKD) progression [45] and/or death in hospitalized patients [46, 47]. High serum CHI3L1 levels in patients undergoing hemodialysis have been associated with all-cause mortality [48]. Nevertheless, CHI3L1 production/secretion may be a mechanism by which apoptosis/necrosis is controlled via the activation of the AKT pathway [49].

However, this is the first study to report the potential role of CHI3L1 in kidney transplant rejection. One research letter described a correlation between CHI3L1 levels and heart graft rejection episodes [50]. In our opinion, upregulation of CHI3L1 in PBMCs and serum of patients with CAMR may represent a potential hallmark of allograft injury.

SERPING1 (also known as C1 inhibitor) is a serine esterase inhibitor that blocks the C1s and C1r proteases of the classical and MASP2 of the lectin pathway of complement [51]. Increased expression of its transcript was previously identified in kidney biopsies of patients with antibody-mediated rejection [52, 53]. Our results confirmed the pathogenetic role of the complement pathways in antibody-mediated rejections [12] induced by DSA, which resulted in membrane attack complex (MAC) formation and endothelial cell lysis [54] with propagation of the immune response.

Our study also has several limitations worth considering. First, we compared proteomic and transcriptomic profiles derived from two different patient cohorts; second, the number of patients included was too low for biomarker definition. However, stringent statistical analysis ensures the validity of the results.

Conclusions

Our study provides new insights into the complex and multifactorial systemic pathogenetic machinery associated with the onset/development of CAMR and reveals new potential early disease biomarkers and therapeutic targets. Among them, CHI3L1, which reached good discrimination power for CAMR identification, may be useful in clinical practice. Several recently identified molecules targeting CHI3L1 (such as K284 and G721-0282) are currently under investigation for anticancer therapies [55, 56] and could also be proposed for the treatment of rejection.

In addition, SERPING1 upregulation in patients with CAMR confirmed the potential clinical impact of complement pathway inhibition. Data from a phase 1/2 study on 20 patients assessing the use of a C1 inhibitor (Berinert®) in the prevention of antibody-mediated rejection in kidney transplantation showed a favorable safety profile and a potential benefit by reducing the capacity of anti-HLA antibodies to bind C1q [57]. Then, in a pilot study evaluating the use of Berinert® for the treatment of antibody-mediated rejection nonresponsive to standard treatment for 3 months, the authors found improvement in the mean estimated glomerular filtration rate (eGFR) after 6 months of treatment, no worsening of histological features, and a reduction of the DSA C1q status post-treatment [58]. A phase 2b, multicenter, double-blind, randomized, placebo-controlled pilot study reported good results with the use of Cinryze®, a human plasma-derived C1 inhibitor, as an add-on therapy for antibody-mediated rejection occurring within the first year after transplantation [59]. Although encouraging, this clinical evidence and our results require further validation in larger patient cohorts, including prospective evaluation of CHI3L1 levels after transplantation and in different clinical settings of CAMR. This approach may early identify patients at high risk for developing this complication, with no need to perform allograft biopsy, and provides a foundation for personalized medicine in kidney transplantation.

Nonetheless, our translational study underlined the need to incorporate new technical figures/specialists (such as bioinformaticians and molecular biologists) that could support transplant clinicians in both research and decision-making processes, developing the new paradigm of the “translational medicine”.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

CAMR

Chronic antibody-mediated rejection

PAC

Protein aggregation capture

DIA

Data-independent acquisition

MS

Mass spectrometry

FDR

False Discovery Rate

VIP

Variable importance in projection

PSM

Peptide-spectrum match

ROC

Receiver operating characteristic

AUC

Area under the curve

DSA

Donor-specific antibodies

CHI3L1

Chitinase-3 like-protein-1

SERPING1

Plasma protease inhibitor C1

AKI

Acute kidney injury

eGFR

Estimated glomerular filtration rate

CKD

Chronic kidney disease

MAC

Membrane attack complex

Author contributions

MB: Investigation, Formal analysis, Visualization, Writing—original draft; SG: Investigation, Writing—original draft, Formal analysis; FL, LB, GC, PP, AP, MB, SS: Investigation and data acquisition; LG: Writing—review and editing; GZ: Study design, interpretation of data, Supervision, Writing—review and editing.

Funding

This work was supported by Ministero della Salute “Ricerca Corrente” e “Cinque per mille” (ID 23680420), Fondazione Malattie Renali del Bambino ETS (ID FMRB–2024), and by the European Union– Next Generation EU- NRRP M6C2- Investment 2.1 Enhancement and strengthening of biomedical research in the NHS (PNRR-MR1-2022-12375880).

Data availability

The mass spectrometry data have been deposited in the ProteomeXchange Consortium via the PRIDE partner repository with the following dataset identifier: PXD057746

Results of the microarray experiment are available in the Gene Expression Omnibus (accession numbers: GSE51675).

Declarations

Ethics approval and consent to participate

The study was conducted according to the latest version of the Declaration of Helsinki and was approved by the local ethics committee of Policlinico di Bari (Prot. N. 670/C.E.).

Consent for publication

Not applicable.

Competing interests

The authors declared no competing interests.