CLINICAL RESEARCH www.jasn.org

Non-HLA Antibodies to Immunogenic Epitopes Predict the Evolution of Chronic Renal Allograft Injury

Tara K. Sigdel, Li Li, Tim Q. Tran, Purvesh Khatri, Maarten Naesens, Poonam Sansanwal, Hong Dai, Szu-chuan Hsieh, and Minnie M. Sarwal

Department of Pediatrics, Stanford University School of Medicine, Stanford, California

ABSTRACT Chronic allograft injury (CAI) results from a humoral response to mismatches in immunogenic epitopes between the donor and recipient. Although alloantibodies against HLA antigens contribute to the pathogenesis of CAI, alloantibodies against non-HLA antigens likely contribute as well. Here, we used high- density protein arrays to identify non-HLA antibodies in CAI and subsequently validated a subset in a cohort of 172 serum samples collected serially post-transplantation. There were 38 de novo non-HLA antibodies that significantly associated with the development of CAI (P,0.01) on protocol post-transplant biopsies, with enrichment of their corresponding antigens in the renal cortex. Baseline levels of preformed antibodies to MIG (also called CXCL9), ITAC (also called CXCL11), IFN-g, and glial-derived neurotrophic factor posi- tively correlated with histologic injury at 24 months. Measuring levels of these four antibodies could help clinicians predict the development of CAI with .80% sensitivity and 100% specificity. In conclusion, pre- transplant serum levels of a defined panel of alloantibodies targeting non-HLA immunogenic antigens associate with histologic CAI in the post-transplant period. Validation in a larger, prospective transplant cohort may lead to a noninvasive method to predict and monitor for CAI.

J Am Soc Nephrol 23: 750–763, 2012. doi: 10.1681/ASN.2011060596

Despite improvements in short-term graft survival HLA mismatched grafts without demonstrable anti- over the past decades,1 chronic allograft injury bodies to donor-specific HLA antigens.7 These find- (CAI) remains a major challenge in renal and other ings strongly suggest that additional pathogenic solid organ transplants. In addition, the rate of pro- antibodies drive the humoral axis of injury in CAI,9,11 gression of CAI remained relatively unabated over and may be the final common outcome. The acti- the last decade, with limited improvements in ex- vation or transition of these non-HLA antibodies tending graft survival.2,3 CAI in the graft is the pri- toward pathogenicity is likely through the underlying mary reason for accelerated graft loss,4 categorized triggers of acute rejection, hypoperfusion, ischemia by progressive interstitial fibrosis (IF) and tubular reperfusion, calcineurin toxicity, infection, and re- atrophy (TA) of the parenchyma,5 that is also asso- current diseases.12 Currently, there is no means to ciated with glomerulopathy, fibrointimal hyperpla- predict which transplant patients will develop accel- sia of arteries, and arteriolar hyalinosis.6 This injury erated CAI in the allograft. Histologic diagnosis of is believed to be mostly a humoral injury response CAI is the current gold standard for diagnosing to mismatched immunogenic epitopes between the donor and recipient,7 with current understanding mostly focused on HLA antigens.8 Despite improved Received June 21, 2011. Accepted October 28, 2011. understanding of the assessment and monitoring for Published online ahead of print. Publication date available at HLA mismatched epitopes,8 the pathogenicity of www.jasn.org. fi donor-speci c anti-HLA antibodies on CAI remains Correspondence: Dr. Minnie M. Sarwal, Department of Pediat- elusive.9 This is shown with complete HLA matched rics, Stanford University, G306, 300 Pasteur Drive, Stanford, CA transplants exhibiting finite survival due to progres- 94305-5208. Email: [email protected] sive CAI,10 and not infrequently, CAI observed in Copyright © 2012 by the American Society of Nephrology

750 ISSN : 1046-6673/2304-750 J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

CAI; however, this only detects advanced, established, and often analysis of the most statistically significant (P#0.05) non-HLA irreversible injury. The renal biopsy is also an invasive procedure antibodies, the CAI-specific non-HLA antibody could be categor- that suffers from sampling heterogeneity, is associated with var- ically divided into an non-HLA showing significant detection ei- ious complications,13 and provides a limited prognostic value. ther early (,6 months) or late (.6 months) post-transplantation. Identification of informative, minimally invasive biomarkers is The early response was observed to be against non-HLA proteins critically needed to monitor and predict CAI, and remains a involved in pathways such as cell-mediated immune response, critically important unmet need in solid organ transplantation. connective tissue development and function, and EGF signaling. The antibody axis has been implicated in different disease Some of these selected antigens are shown in Table 1. The late conditions, and profiling and measuring the level of IgG an- responding non-HLA proteins were noted to immunologically tibodies against thousands of defined human proteins to iden- reactive proteins involved in pathways such as cell death, cell-to- tify antibodies against non-HLA antigens have been previously cell signaling and interaction, and cellular movement (Figure 2B attempted in autoimmune diseases14 and CKD.15 In an at- and Supplemental Figure 1). tempt to evaluate clinically relevant de novo, renal-specific, We analyzed the antibody response for 12 HLA antigens, non-HLA antibody responses, we utilized protein arrays and including 2 HLA class I molecules (HLA-B and HLA-C) and 10 customized informatics for studying non-HLA antibody re- HLA class II molecules (HLA-DOA, HLA-DOB, HLA-DPA1, sponses in stable grafts16,17 and acute rejection11,18 to inter- HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DRA, HLA-DRB1, rogate previously unidentified non-HLA antibody CAI. To HLA-DRB3, and HLA-DRB5) in the CAI and non-CAI groups. investigate the non-HLA pathogenicity in CAI, we excluded The signal intensity of antibody detection to these targets is very patients with signs of acute rejection, purely focusing on low (average ,500 RFU signal intensity). Thus, it seems that non-HLA antibody responses in the absence of any interval HLA antibody levels may be less influenced in the course of CAI, acute rejection episodes. From this cohort, we then excluded in the absence of interval acute rejection. pediatric patients with interval episodes of infection, delayed graft function, or body surface area ,0.75 m2,19 thus Cross-Mapping CAI-Specific Non-HLA Antibodies to creating a homogenous set of patients without any major con- Expression of the Target Proteins in Subcellular founders to CAI. This resulted in the removal of 47 patients Compartments of the Kidney with interval acute rejection episodes, 3 patients in interval To triage the selection of non-HLA antibodies to further pursue, infectious episodes, 3 patients with delayed graft function, as we chose to assign physiologic and potentially pathologic well as 4 recipients with body surface area ,0.75 m2.From relevance to the significant CAI-specific antibodies, by selecting these remaining patients, we carefully selected a subset of 20 antibodies with reactivities to kidney-specific antigens. The patients (n=10 CAI versus n=10 without CAI) as the learning assumption is that these de novo antibody responses after kidney set and collected surveillance samples of sera matched to di- transplantation are more likely against the antigens in the newly agnostic biopsy at 0, 6, and 24 months post-transplantation transplanted kidney. Toenable this analysis, we performed cross- from each patient, thus analyzing 60 serum samples for diag- mapping of kidney and kidney-compartment–specific genes ob- nosis and prediction of CAI. The increased presence of signif- tained from microdissected compartments of normal kidney by icant non-HLA antibodies was analyzed for their correlation profiling these tissues on cDNA microarrays with protoarray with injury progression post-transplant. Finally, from the protein targets, using our published approach of integrated anti- larger confounder-controlled cohort, we validated the most biomics.18,21,22 This analysis revealed that there was an enrich- significant and relevant antibodies (from discovery) by using ment for antigens expressed in the renal cortex (P=0.029). As ELISAs on 112 unique and independent sera samples from 68 previously shown, the renal pelvis antigens are highly immuno- patients (Figure 1). genic, and enrichment for pelvis-specificantigenswasalso noted. These antigen lists are shown in Table 2.

RESULTS CAI-Specific Antibodies Track Injury Progression We took the antibodies that were increased with CAI and Identification of CAI-Specific Novel Non-HLA performed Spearman correlation analyses with the Chronic Antibodies Allograft Damage Index (CADI) and IF/TA scores of each Compared with the non-HLA antibody levels before trans- corresponding renal biopsy. A total of 34 antibodies and 41 plantation, there was a significant immune responsepost- antibodies demonstrated a positive correlation with CADI transplant withthedevelopment ofCAI.Byusing M-statistics of score (overall P,0.04) and IF/TA (overall P,0.04), respec- the Prospector Analyzer with the robust linear normalization tively. The top 20 antibodies are listed in Table 3. Pathway method published elsewhere,20 we identified a total of 231 analysis suggested that these antigens are involved in cellular non-HLA antibodies with statistically significant P values #0.05. movement, antigen presentation, cell-to-cell signaling and in- Among the 231 changed antibodies, 111 non-HLA antibodies teraction, and cell death (P,0.002). Scatter plots for the six increased significantly in the CAI group (P#0.05) (Figure 2A most significant antibodies (IFNG [IFN-g], MIG, CSNK2A2, and Supplemental Table 1). On the basis of a time course CCL21, glial cell-derived neurotrophic factor [GDNF], and

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 751 CLINICAL RESEARCH www.jasn.org

(preformed) and the post-transplant period (de novo) (Figure 4, A–D) (Table 4). Univariate and multivariate logistic re- gressions were performed to examine the relationship between the baseline signal intensities of each of the preformed four antibodies (MIG, ITAC, GDNF, and IFNG) and CADI scores from the protocol biopsies performedinthesesamepatientsat6and24 months post-transplantation. A CADI score of $3 was used as a cutoff to define either severe ($3 CADI score) or mild (,3 CADI score) histologic CAI. The receiver operating characteristic (ROC) curves for each of these antibodies to predict the de- velopment of CAI are shown for the 6-month protocol biopsy (Figure 4E) and the 24-month protocol biopsy (Figure 4F). Figure 1. Study schematic. A novel approach is used to identify novel non-HLA When a regression model was built by uni- antibodies through an integrative approach of analyzing sera samples with matched biopsies on protein microarray for the discovery step. The validation step is per- variate analysis, detection of baseline levels formed after filtering the data for highly correlated antibodies for renal graft injury. of preformed antibody to MIG at levels The highly correlated antibodies are then validated by indirect ELISAs on an in- .200 RFU had a significant association dependent set of patients for cross-sectional and longitudinal analyses. nCAI, non- with the patient developing CAI on the CAI; STA, stable. protocol biopsy at both 6 months (odds ratio, 1.04; 95% confidence limit, 1.003– 1.078; P=0.034), and at 24 months post- ITAC) and their correlation with the CADI score are presented transplantation (odds ratio, 1.023; 95% confidence limit, in Figure 3. 1.001–1.045; P=0.0375). Patients who never developed CAI had baseline antibody levels to MIG of ,50 RFU. Predicting Injury Progression after Transplantation As shown by our group in a previous publication,15 chronic Validation of Selected Potential Antibody Biomarkers renal injury and end stage renal failure result in uncovering of ELISAs were developed and optimized to demonstrate that kidney-specific epitopes, mostly in the renal cortex and the discovery of target non-HLA antibodies by protoarray could be renal pelvis, and the immunologic recognition of these newly validated by ELISA. ELISAs were set up, customized, and discovered non-HLA antigens with humoral responses in the performed for some selected (and significant) targets in which form of non-HLA antibodies, specific to chronic renal injury. the full-length proteins could be commercially obtained from To evaluate if this repertoire of preformed non-HLA antibod- stock supplies for setting up the reverse ELISAs for antibody ies specific to chronic renal failure could predict the subse- measurements. These assays were done for MIG, ITAC, quent development of CAI after kidney transplantation, the CSNK2A2, and PDGFRA. A significant increase in the antibody level of these non-HLA antibodies was evaluated at the baseline levels were confirmed in CAI by reverse ELISA for all four targets: (day 0 sera) of each patient. Spearman correlation analysis was MIG (P,0.02), ITAC (P,0.014), CSNK2A2 (P,0.0002), and performed to identify the specificities or levels with the pro- PDGFRA (P,0.0001) (Figure 5, A–D). For further validation, gression of CAI on the biopsy performed at 6 months and 24 seven independent patients were selected, who were not used for months after transplantation. Interestingly, baseline levels of discovery by protoarrays, in whom serial sera and matched pro- these preformed antibodies to MIG, ITAC, IFNG, GABPA, and tocol biopsies were available and who had a CAI grade of at least GDNF positively correlated (P,0.05) with CADI score at .3 on both the 6-month and the 24-month protocol biopsies. 6 months, and baseline levels of four of these five antibodies Three of these targets were confirmed by our customized reverse (MIG, ITAC, IFNG, GDNF), as well as two additional antibod- ELISA to also show significant increases in this longitudinal ies to IL-8 and CCL21, all positively correlated (P=0.04) with analysis. Antibody levels for patients with CAI at 24 months CADI score at 24 months (Table 4). As expected for this selec- were significantly higher in 24-month sera versus 6-month tion of antibodies, antibody levels at 6 months post-transplant sera for CSNK2A2 (P=0.05), ITAC (P=0.05), and MIG for IL-8, MIG, IL21, CCL19, LRRK2, CCL21, GDNF, and IFNG (P=0.005) (Figure 5, E and F). In addition, there was a strong also correlated with CAI at 24 months post-transplantation, correlation of two of these antibody levels with CADI scores on suggesting that these antibody levels could also be tracked for the 24-month protocol biopsies (MIG: r=0.62, P=0.05; ITAC: progression of CAI in both the pretransplant period r=0.62, P=0.05).

752 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

Figure 2. A cross-sectional analysis of sera samples taken from patients with matched biopsies showing either the presence (n=20) or absence of CAI (n=20). (A) A volcano plot demonstrates a significant shift in antibody responses CAI. A total of 111 antibodies (Ab) are significantly increased (in red dots) in sera collected from renal transplant patients with CAI (n=20) compared with patients without CAI (n=20) (P#0.05), and 40 antibodies are increased in CAI (P#0.01). (B) The dynamics of change of antibody levels of most significant antibodies over the period of 24 months are shown. Ab, antibody.

DISCUSSION targeting for various proteins/antigens in the renal allograft. These altered antigens mount specific antibody responses. CAI is related to a humoral response driven by alloantibodies The ability to recognize these relevant antigens, to measure against HLA antigens, with recent data also supporting a role these specific antibody levels in the circulation, and to correlate for immunogenic non-HLA renal antigens. An understanding these levels with histologic CAI in the allograft itself is a power- of the specificities and correlative levels of these antibodies can ful approach to identify noninvasive, clinically relevant sero- provide an understanding of molecular injury triggers in CAI logical biomarkers for detecting and predicting CAI. Our initial and a means to noninvasively monitor the development of this studies strongly suggested a pathogenic role for non-HLA anti- injury. Despite the contribution of newly available immuno- bodies after transplantation.11,16,18 From gene expression anal- suppressive drugs in the short-term survival of the trans- yses, we reported an increased level of Ig gene transcripts as a planted kidney,3 CAI progresses relentlessly post-transplant, function graft injury.24 We thus hypothesized that accumulated accelerating the risk for graft loss. There is a lack of reliable organ injury in the form of CAI could be associated with a and robust noninvasive biomarkers to track the health status of specific set of circulating non-HLA antibodies that can be de- the transplanted organ, without the requirement of serial pro- tected, quantified, and used to follow chronic graft injury. tocol biopsies. Elaborating injury mechanisms and developing We undertook a carefully designed study of highly selected predictive, noninvasive monitoring methods for CAI is an im- patients with established histologic CAI, with available serial portant unmet clinical need.2,23 The process of biologic tissue sera and matched protocol biopsies, and with no confounding injury from events such as exposure to calcineurin inhibitor influences of delayed graft function, acute rejection, or in- agents, from graft ischemia, or from recipient hypertension, fection on the evolution of chronic injury. Unbiased discovery likely results in the alteration of the levels, exposure, or of a panel of correlative non-HLA antibodies with CAI was

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 753 CLINICAL RESEARCH www.jasn.org

Table 1. Early and late responding non-HLA antibodies in CAI Serial No. Gene Symbol Protein Name P Value Early response (,6 mo post-transplant) 1 CSNK2A2 Casein kinase 2, a2 0.01 2 CSNK2A1 Casein kinase 2, a1 0.02 3 KLK6 Kallikrein-related peptidase 6 0.01 4 PPID Peptidylprolyl isomerase D 0.00 5 CCDC55 Coiled-coil domain containing 55 0.03 6 NEK3 NIMA (never in mitosis gene a)–related kinase 3 0.02 7 CSNK1G3 Casein kinase 1, g3 0.002 8 CCL21/6CKINE Chemokine (C-C motif) ligand 21 0.00 9 BIN1 Bridging integrator 1 0.05 10 MAPRE2 Microtubule-associated protein, RP/EB family, member 2 0.00 11 SGK2 Serum/glucocorticoid regulated kinase 2, transcript variant 1 0.03 13 GYG2 Glycogenin 2 0.01 14 MEF2D Myocyte enhancer factor 2D 0.0006 15 EGFR L861Q EGF receptor (erythroblastic leukemia viral [v-erb-b] 0.002 oncogene homolog, avian), transcript variant 1 16 EGFR EGF receptor (erythroblastic leukemia 0.00 viral [v-erb-b] oncogene homolog, avian); see catalog number for detailed information on wild-type or point mutant status 17 Jo-1/HARS Histidyl-rRNA synthetase (Jo-1) 0.01 Late response (.6 mo post-transplant) 18 MIG Chemokine (C-X-C motif) ligand 9 19 GDNF Glial cell-derived neurotrophic factor 0.0001 20 CSNK1G1 Casein kinase 1, g1 0.05 21 BHMT2 Betaine-homocysteine methyltransferase 2 0.00 22 PKN1 Serine/threonine-protein kinase N1 0.00 23 ATXN3 Ataxin-3 0.01 24 MARK4 MAP/microtubule affinity-regulating kinase 4 0.02

Table 2. Compartment-specific enrichment of CAI-specific antibodies cross-mapping of kidney and kidney compartment–specific genes in between cDNA microarraya Kidney/Compartment Kidney-Specific Targets CAI Specific (20 np versus 20 P) (P,0.05) P Value Kidney 261 3 (ACY1, BIN1, CSNK2A1) 0.10 Glomerulus 245 7 (FLT1, FLT4, MIG, PDGFRB, PRKCE, KLK6, BHMT2) 0.11 Inner cortex 267 3 (ACY1, BHMT2, CLK4) 0.03 Outer cortex 433 8 (ACY1, PRKCE, PRKCZ, KLK6, VDR, 0.13 SNF1LK2, BHMT2, CLK4) Outer medulla 42 0 0.41 Inner medulla 11 0 0.79 Papillary tip 222 4 (PDGFRA, PRKCB1, RARB, AFAP1l2) Pelvis 635 6 (IFNG, JAK3, CCL19, CCL21, WEE1, NDE1) 0.001 aThe cross-mapping was performed using previously published data and methods (18,21,22). evaluated from pretransplant and post-transplant sera in these CAI and, most importantly, can predict the future de- patients. Selection of antibody targets was biased for kidney velopment of CAI by pretransplant sampling for the same expressed antigens, and customized reverse ELISAs were gen- antibody panel. erated to validate the findings in independent sera samples Irrespective of the initial trigger, it is believed that the path- from an independent group of patients with CAI. These results ophysiology of CAI involves endothelial injury leading to an demonstrated, verified, and validated a small panel of newly increased expression of cytokines and several other immune- identified non-HLA antibodies that correlate not only with related genes, resulting in proliferative processes, remodeling, histologic CAI in the matched sera sample, but that also have and scarring of the graft. The increased IgG antibody level of correlative levels post-transplantation with progressive cytokines IFNG, MIG, and ITAC strengthens the belief that the

754 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

Figure 3. CAI-specific antibodies correlate with the severity of CAI at the time of sampling. (Upper panel) IgG antibody level of six antigens: (A) IFNG, (B) MIG (CXCL9), (C) CSNK2A2, (D) CCL21, (E) GDNF, and (F) ITAC (CXCL11) are shown to be correlated with chronic injury in terms of CADI score. (Lower panel) Evidence of the presence of four of the corresponding antigens of CAI-specificantibodies: (A) IFNG, (B) ITAC (CXCL11), (C) GDNF, and (D) PDGFRA. transplanted kidney is under continuous immunologic attack. mononuclear cells to the site of tissue damage.35 Upregula- Similar results were also shown by gene expression analysis on tion of these chemokine proteins has been demonstrated to CAI protocol biopsies from the same patient group.24 IFNG is a be correlated with T lymphocyte recruitment during acute type II IFN that is encoded by the IFNG gene25 and is exten- and chronic rejection events. In addition, pretransplant se- sively produced by a wide range of immune-related cells such rum levels of MIG and CXCL10 have been reported to be as CD4+ T helper cell type 1 lymphocytes, CD8+ cytotoxic associated with graft failure,36,37 andrecentlyourgrouphas lymphocytes, natural killer cells, B cells, natural killer T cells, reported that there are high levels of circulating MIG protein and antigen presenting cells.26–30 IFN-g gene polymorphism in the serum of kidney and heart transplant patients at the +874 T/A (rs2430561) has been suggested to be associated time of acute rejection.38 with AR31,32 and activated T lymphocytes in alloimmune in- We also observed an increase in the IgG antibodies of a jury are a major source of IFNG expression, regulated by IL-12 number of novel antigens that have not been previously reported and IL-18.29,33 IFNG is critical in trafficking of specific to be relevant in organ transplantation. Some of these continue immune cells to sites of inflammation by upregulating ex- to support the activation of the immune axis as a critical trigger pression of adhesion molecules and chemokines.34 In this for CAI, such as antibodies to the Jo-1 antigen, also known as context, our observation of a significant increase in the anti- histidyl-tRNA synthetase, and are responsible for the synthesis of body level against MIG and ITAC is important. Both MIG and histidyl-transfer RNA and the migration of activated monocytes, ITAC are chemokine proteins that also serve as chemoattrac- immature dendritic cells, and activated lymphocytes.39 An in- tants for leukocytes, monocytes, neutrophils, and other teresting antibody in CAI is the GDNF, which is involved in

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 755 CLINICAL RESEARCH www.jasn.org

Table 3. CAI-specific antibodies correlate with CADI score and IF-TA scores Serial No. Gene Symbol Protein Name CADI Score (r, P) IF-TA Score (r, P) 1 IFNG IFN-g 0.68, ,0.0001 0.61, ,0.0001 2 MIG C-X-C motif chemokine 9 0.61, ,0.0001 0.55, 0.0002 3 ITAC C-X-C motif chemokine 11 0.51, 0.0009 0.42, 0.007 4 CSNK2A2 Casein kinase 2, a prime polypeptide 0.51, 0.0008 0.52, 0.0006 5 GDNF Glial-derived neurotrophic factor 0.63, ,0.0001 0.58, ,0.0001 6 BHMT2 Betaine-homocysteine methyltransferase 2 0.47, 0.002 0.54, 0.0003 7 6CKINE C-C motif chemokine 21 0.56, 0.0002 0.54, 0.0003 8 CSNK2A1 Casein kinase 2, a1 polypeptide, 0.50, 0.001 0.54, 0.0004 transcript variant 2 9 J0-1(HARS) Histidyl-rRNA synthetase 0.63, ,0.0001 0.63, ,0.0001 10 CSNK1G1 Casein kinase 1, g1 0.49, 0.001 0.51, 0.0008 11 IL21 IL-21 0.57, 0.0001 0.51, 0.0008 12 CSNK1G3 Casein kinase 1, g3 0.35, 0.026 0.43, 0.006 13 IL-8 IL-8 0.43, 0.006 0.48, 0.002 14 PRKCE Protein kinase C, « 0.41, 0.009 0.48, 0.002 15 FLJ21908 RNA polymerase II-associated protein 3 0.48, 0.002 0.47, 0.002 16 WIBG Within bgcn homolog (Drosophila) 0.39, 0.01 0.46, 0.003 17 ATXN3 Ataxin-3 0.46, 0.003 0.45, 0.003 18 RNAPOL RNA polymerase 0.39, 0.01 0.45, 0.004 19 MAPRE2 Microtubule-associated protein, 0.34, 0.03 0.45, 0.004 RP/EB family, member 2 20 CCL19 chemokine (C-C motif) ligand 19 0.40, 0.009 0.43, 0.006

kidney development. Signaling by the secreted protein GDNF also associated with AR in our previous reports.38,43 This ob- through the RETreceptor tyrosine kinase and the GDNF family servation suggests a common activation of cytokines not only receptor a1, a GDNF co-receptor, are involved in kidney devel- as immediate trigger of inflammation and injury in AR but opment,40 specifically in metanephros41 and ureteric bud devel- also as activated cytokines, in the case of CAI (Figure 6). The opment.42 GDNF antibodies are increased in CAI and correlate future direction of this work requires a large-scale validation with histologic CAI (Figure 3), and baseline/pretransplant antibody of these data in a prospective trial to assess the specificity and levels are associated with a greater risk of CAI post-transplantation sensitivity of this antibody panel before transplant to assess the (Figure 4). risk of CAI, and perhaps to subsequently titrate the immuno- In conclusion, we identified and validated a panel of cir- suppression induction and maintenance strategy for each culating non-HLA antibodies in renal transplant patients that transplant patient on the basis of the risk of post-transplant correlate with established CAI, can be serially measured post- CAI. transplant to identify which patients will develop accelerated CAI over time, and, most importantly, to stratify risk for de- velopment of CAI by measuring this antibody panel even be- CONCISE METHODS fore organ engraftment. The repertoire of the panel suggests immunologic “preconditioning” of the recipient as an impor- Patients and Samples tant risk factor for CAI progression after engraftment because We analyzed serological response on serum samples that were this panel of antibodies is enriched against cell-to-cell signal- collected from renal transplant patients from Lucile Packard Child- ing and interaction as well as cell-mediated immune response, ren’s Hospital, Stanford University, Stanford, California. After exclu- antigen presentation, and cell death (P,0.02). The altered sion of patients with incidences of acute rejection (antibody mediated expression of these factors may predispose the recipient to or cellular), infection, delayed graft function, donor pathology, and continued alloimmune injury, which then drives the turnover C4d-positive peritubular capillary staining on biopsies, 172 sera sam- of chronic interstitial injury, fibrosis, and TA, with loss of graft ples from 98 patients were selected and split into training and verification function over time. Importantly, very similar results are noted groups with a one-third (n=60 sera samples for discovery) to two-thirds by microarray profiling of the very same biopsy samples, from split (n=112 sera samples for validation). The transplant patients in- the same patient cohort, in which very coordinated expression cluded in the study were unsensitized patients, with a peak plasma renin of immune-specific genes in the allograft in the early protocol activity ,20% and mean recipient and donor HLA match of 3.561.29. biopsy can predict the progression of CAI in the later protocol The demographic information of the transplant patients in the learning biopsy sample.24 It is intriguing that the reactive cytokines and the verification groups is summarized in Table 5. For the discovery such as MIG, ITAC, and IFNG observed in this study are group, a total of 60 sera samples were used. We selected a total of 30

756 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

cutoff CADI score ,2 was used to score nonsig- nificant histologic injury progression; a CADI score $5.0 was used to score significant histo- logic injury progression. These CAI samples were compared with 30 sera samples from 10 de- mographically matched transplant recipients with stable graft function, with histologically clean pro- tocol biopsies and minimal to no injury in each of the 6- and 24-month protocol biopsies. For vali- dation of discovered antibodies by cross-sectional analysis, we identified sera samples collected from demographically matched 31 renal transplant pa- tients with biopsy CAI and 30 without CAI. The details are summarized in Table 6. This study was approved by the Institutional Review Board of Stanford University and the other participating centers of the clinical trial. All 172 study biopsies were blindly analyzed by a Stanford University pathologist and were graded by the Banff classification5,46,47 for acute rejection, and intragraft C4d stains were per- formed48,49 to assess for acute humoral rejec- tion.50,51 The histologic lesions of CAI were extensively identified and a semi-quantitative score for CAI applied to each biopsy, based on standardized definitions from the Banff (2), CADI, (3), and chronic calcineurin inhibitor toxicity (19) scores. For the semi-quantitative scoring criteria, chronic lesions are preceded by “c”; interstitial fibrosis is denoted as “ci” and scored as ci0–ci3; TA is denoted as “ct” and scored as ct0–ct3 based on the area of cortical tubular atrophy; glomerulopathy is denoted as “cg: and scored as cg0–cg3 based on the extent of double contours in glomerular capillary loops; and arteriolar hyalinosis is denoted as “ah” andscoredasah0–ah3 based on the extent of Figure 4. CAI-specific antibodies at the time of transplant correlate with the injury PAS-positive hyalinosis. On the basis of the com- progression post-transplant. IgG antibody level of two MIG antibodies at the time of bination of these features, a diagnosis of CAI was transplantation is positively correlated with CADI score post-transplantation at (A) 6 and graded on the basis of severity as grade I (6%– (B) 24 months. IgG antibody level of two ITAC antibodies at the time of transplantation 25% of cortex), grade II (26%–50% of cortex), is positively correlated with CADI score post-transplantation at (A) 6 and (B) 24 months. and grade III (.50% of cortex). If the biopsy of ROC curves for the MIG, GDNF, ITAC, and IFNG antibodies predict development of the patient showed a CAI score ,6% at the time CAI. The predictive ability of baseline antibody level to predict injury at (E) 6 months (F) of serum collection, the sample was considered and (B) 24 months post-transplantation is shown in terms of ROC curves and area under as stable. Samples with biopsy scores for CAI the curve for different antibodies as well as their combined contribution in the pre- diction. Ab, antibody. between 6% and 25% were deliberately excluded to allow for clean separation of CAI samples in this discovery process. sera collected at time 0, 6, and 24 months post-transplantation from 10 patients with confirmed CAI by 24 months post-transplantation. Serum Sample Collection and Storage The quality of the graft at the time of transplantation was excellent in Blood samples (4.5 ml) were collected in a 5-ml cryotube and incu- this selected group of patients, with a mean Remuzzi score of 0.560.76 bated at room temperature for 30minutes until the clot was formed. The at the time of transplantation and a median score of 0. In the overall sample was then centrifuged at 20003g for 5 minutes using a swinging group, the mean CADI score was used as a semi-quantitative measure bucket rotor. The serum was transferred to another cryotube and was of the chronic injury grade.44,45 For analytical purposes, an arbitrary stored at 280°C until use.

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 757 CLINICAL RESEARCH www.jasn.org

Table 4. List of antibodies the level (from protein arrays) of which at the time of transplant and 6 months post-transplant correlates with CAI progression in the future P Gene Symbol Antigen Name Spearman Correlation (r) Value Antibody level measured at the time of transplantation correlates with 6-mo chronic injury GABPA GA binding protein transcription 0.46 4.31E202 factor, a subunit MIG Chemokine (C-X-C motif) ligand 9 0.74 2.00E204 ITAC Chemokine (C-X-C motif) ligand 11 0.49 2.90E202 IFNG IFN-g 0.60 5.50E203 GDNF Glial cell-derived neurotrophic 0.55 1.12E202 factor Antibody level measured at the time of transplantation correlates with 24-mo chronic injury IL-8 IL-8 0.46 3.98E202 CCL21 Chemokine (C-C motif) ligand 21 0.49 2.94E202 MIG Chemokine (C-X-C motif) ligand 0.71 5.00E204 ITAC Chemokine (C-X-C motif) ligand 11 0.54 1.36E202 IFNG IFN-g 0.73 3.00E204 GDNF Glial cell-derived neurotrophic factor 0.62 3.20E203 Antibody level measured at 6 mo post-transplantation correlates with 24-mo chronic injury IL-8 IL-8 0.47 3.87E202 MIG Chemokine (C-X-C motif) ligand 9 0.46 4.25E202 IL21 IL-21 0.51 2.16E202 CCL19 Chemokine (C-C motif) ligand 19 0.48 3.30E202 LRRK2 Leucine-rich repeat kinase 0.46 4.25E202 CCL21 Chemokine (C-C motif) ligand 21 0.47 3.77E202 GDNF Glial cell-derived neurotrophic factor 0.60 5.30E203 IFNG IFN-g 0.59 6.30E203

Immune Response Biomarker Profiling by Protein Sunnyvale, CA). The data were acquired by using GenePix Pro 6.0 Microarrays microarray analysis software (Molecular Devices). The slides were We used the ProtoArray Human Protein Microarray (v4.0 and v4.1; scanned at 635 nm with a photomultiplier gain of 600, a laser power Invitrogen, Carlsbad, CA) for this study. The slides contained a total of of 100%, and a focus point of 0 mm. The .gal files were obtained approximately 8200 antigens printed on each slide. The proteins from a ProtoArray central portal on the Invitrogen website (www. included in the arrays belong to different cellular locations with a wide invitrogen.com/ProtoArray) by submitting the barcode of each pro- variety of functions. For the detection of immune response, we used tein microarray. manufacturer’s protocol with slight changes. Briefly, the microarray slides were blocked with 5.0 ml of blocking buffer (100 mM sodium Statistical Analyses phosphate pH 7.4, 200 mM NaCl, 0.08% Triton X-100, 25% glycerol, Signal intensities per spot for each individual array were obtained 20 mM reduced glutathione, 1.0 mM DTT, 1% Hammarsten-grade using GenePix Pro 6.0 software. We used Prospector Analyzer 5.2 casein) with gentle agitation in four-well trays for 1 hour at 4°C. After (Invitrogen Inc) to subtract the background and to normalize and to the blocking step, the blocking buffer was removed by aspiration and analyze the GenePix results files for each array. We used a standard 5.0 ml diluted serum (1:150) in PBST buffer (13 PBS, 1% Hammarsten- cutoff Z score of 3.0. Individual antigen reactivity was ranked based grade casein, 0.1% Tween 20) was added onto the plate to incubate for on Z score above or below the mean signal for each array. Arrays from 90 minutes with gentle agitation at 4°C. In the next step, the serum patients of a distinct clinical phenotype were analyzed as a group. sample was removed and the plates were washed five times with 5.0 Group analyses (.5 per group as suggested by the manufacturer) ml of fresh PBST buffer with 5-minute incubations per wash with were made by comparing two sets of individual antibody level for gentle agitation. A 5.0-ml portion of secondary antibody (Alexa every antigen present on the array using M-statistics of the Prospector Fluor 647 conjugate anti-human antibody) diluted in PBST buffer Analyzer with the robust linear normalization method.20 Differences was added and the plates were incubated for 90 minutes with gentle in significance were displayed as P values, with #0.05 considered agitation at 4°C. The secondary antibody solution was removed by significant. We performed two-factor ANOVA on the protein array aspiration. The plates were then washed five times with 5.0 ml of data using the CADI score and sample time as factors. The ANOVA P fresh PBST buffer, with 5-minute incubations per wash with gentle values were corrected for multiple hypotheses using the Benjamini– agitation. At the end, the slides were dried by centrifugation and Hochberg correction. We found 72 antibodies that were significantly scanned using an Axon GenePix 4000B scanner (Molecular Devices, affected by CADI score but not by time at false discovery rate ,0.2

758 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

Figure 5. CAI-specific antibodies are validated with immunoassays using indirect ELISA. Levels of four of the CAI-specific antibodies are assayed in CAI sera and their increased presence in biopsy-proven CAI (n=31) is analyzed by comparing their level in renal transplant patients with biopsy-proven stable graft function (n=30). Levels of (A) CSNK2A2 (P,0.002), (B) MIG (P,0.02), (C) ITAC (P,0.014), and (D) PDGFRA (P,0.0001) are increased. The increased level of antibody level of (E) CSNK2A2 and (F) MIG, along with increased CAI, is observed from ELISA on independent samples (n=17). Increases in antibody level measured by ELISA, as well as CADI score for both the antibodies are statistically significant (P,0.05). STA, stable. for CADI score, of which 53 were increased in the CAI group. Hyper- CA) was used to map target antigens to canonical pathways within the geometric enrichment, univariate and multivariate analyses, and logis- Ingenuity Knowledge Base. Tissue expression of antigens against sig- tic regression modeling were performed using customized algorithms. nificant antibodies correlated with CAI was examined using the hu- Ingenuity software (version 7.5; Ingenuity Systems Inc, Redwood City, man protein atlas.52–54

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 759 CLINICAL RESEARCH www.jasn.org

Figure 6. A developing picture of involvement of non-HLA antibodies against cytokines and other novel antigens in CAI. Abs, anti- bodies.

Table 5. Demographic data Discovery Set (60 Sera) Validation Set (112 Sera)

Characteristic Protein Array Analysis Cross-Sectional ELISA Longitudinal ELISA Non-CAI CAI P Value Non-CAI CAI P Value Number of patients 10 10 30 31 17 Steroid free/steroid 4/6 6/4 0.65 12/18 12/19 1.00 6/11 based Donor sex (M/F) 8/2 4/6 0.17 18/12 22/9 0.13 9/8 Donor age (yr) 24610 35.0610 0.03 30610 30610 0.97 29610 (21; 14–47) (36; 17–54) (31; 14–47) (31; 14–54) Recipient sex (M/F) 5/5 5/5 1.00 13/17 10/21 0.43 10/7 Recipient age (yr) 1464 1066 0.17 1464 1265 0.08 1066 (13; 8–19) (9; 2–19) (15; 3–19) (12; 2–20) Living/deceased 4/6 6/4 0.66 14/16 17/14 0.61 7/10

Cross-Mapping of Gene IDs of the Compartment- coated onto the immunosorbent 96-well plate. The 96-well microwell Specific Gene on cDNA Microarray Platform ELISA plates were coated with corresponding protein in 50 ml of coating fi Cross-mapping of kidney and kidney compartment-speci cgenesin buffer (15 mM Na2CO3, 30 mM NaHCO3,0.02%NaN3, pH 9.6). The between cDNA microarray was performed using the data published21,22 subsequent washing and antibody incubation followed the method pre- and the Protein Microarray V.1 was conducted using the method pub- viously published.11,15 The color was developed by using AP-pNPP lished earlier by our group18 using AILUN software (http://ailun. Liquid Substrate System for ELISA (Sigma-Aldrich, St. Louis, MO). stanford.edu/) to re-annotate probes to the most recent NCBI Entrez Absorption was measured at 405 nm with a SpectraMax 190 microplate gene identifiers. reader (Molecular Devices). To control for nonspecificbinding,wells with no proteins coated were served as negative controls. Development and Optimization of ELISA Assay for Validation of CAI-Based Non-HLA Antibody Validation of CAI-Specific Antibody by ELISA For validation purposes, we developed antibody ELISA to detect serum After the optimization step, we validated the discovery made by the Ig binding to MIG (also called CXCL9), ITAC (also called CXCL11), protein arrays on 112 sera collected from 78 renal transplant patients. CSNK2A2,andPDGFRAfollowingthemethodpublishedpreviously.11,15 There were 61 cross-sectional sera samples collected at 1 year post- In brief, purified proteins, CSNK2A2 (cat# PV3624), PDGFRA (cat# transplantation from 31 unique patients with biopsy-confirmed CAI PV3811), ITAC (cat# PHC1694), and MIG (cat# OHC1604) were ac- and 30 unique patients with stable biopsies (non-CAI). In addition, quired from Invitrogen (Carlsbad, CA). A titration with various coated there were 51 serial (longitudinal) samples from 17 patients with CAI amounts starting at 30.00, 15.00, 7.5, 3.75, 1.87, 0.94, 0.47, and 0 ng, confirmed on their 6-month and 24-month biopsies, with sera respectively, was performed to determine the optimal amount to be samples at 0, 6, and 24 months post-transplantation.

760 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

Table 6. Histologic data for the discovery set for protein array analysis Non-CAI CAI P Value Histology at implantation (n=10) IF/TA grade .0 0/10 1/10 1.00 arteriolar hyalinosis (present/absent) 1/10 1/10 1.00 intimal vascular thickening 1/10 0/10 1.00 (present/absent) glomerulosclerosis (present/absent) 1/10 2/10 1.00 Remuzzi score 0.260.42 (0; 0–1) 0.860.92 (0.5; 0–2) 0.08 Histology at 6 months (n=10) tubulitis score .0 0/10 0/10 1.00 interstitial inflammation score .0 0/10 1/10 1.00 vasculitis score .0 0/10 0/10 1.00 IF/TA grade .1 0/10 2/10 0.47 arteriolar hyalinosis (present/absent) 0/10 0/10 1.00 intimal vascular thickening 1/10 4/10 0.30 (present/absent) glomerulosclerosis (present/absent) 0/10 2/10 0.47 CADI score 0.660.70 (10.5;0–2) 3.1761.65 (3.5; 1–6) ,0.0001 Histology at 24 months (n=10) tubulitis score .0 0/10 0/10 1.00 interstitial inflammation score .0 0/10 0/10 1.00 vasculitis score .0 0/10 0/10 1.00 IF/TA grade .1 1/10 9/10 0.001 arteriolar hyalinosis (present/absent) 0/10 1/10 1.00 intimal vascular thickening 3/10 5/10 0.65 (present/absent) glomerulosclerosis (present/absent) 0/10 6/10 0.02 CADI score 1.661.6 (1.5; 0–5) 7.161.5 (7;4–9) ,0.0001 CADI score slope in first 2 years 0.760.9 (0.6; 20.3 to 2.5) 2.960.6 (3.1; 1.9–3.8) ,0.0001 Schwartz GFR at 6 months 99.2629.5 (93.6; 52–154) 117.0630.5 (102.9; 69.12–164.8) 0.42 at 24 months 93.9630.2 (98.6; 62–121) 110.3619.5 (100.1; 97.15–144) 0.38 Absolute GFR at 6 months 73.7621.9 (70.6; 41–109) 66.0613.7 (67.2; 39–81) 0.36 at 24 months 78.3649.1 (102; 22–111) 71.7629.9 (87.8; 34–101) 0.82

We coated 15.6 ng of the purified proteins (MIG, ITAC, CSNK2A2, microplate reader (Molecular Devices). Statistical calculation (t test andPDGFRA)ontoanimmunosorbent96-wellplate(NUNCbrandcat# and Spearman correlation) was performed using GraphPad Prism 446612). We followed a previously published protocol developed by our software. P,0.05 was considered significant. laboratory.11,15 Briefly, the 96-well microwell ELISA plates were coated with corresponding protein in 50 ml of coating buffer (15 mM Na2CO3,

30 mM NaHCO3, 0.02% NaN3, pH 9.6) and incubated overnight at 4°C. Standard curves were generated using anti-GST tag (mouse monoclonal ACKNOWLEDGMENTS IgG) (Millipore, Temecula, CA) and AP-conjugated AffiniPure goat anti- mouse IgG (Jackson ImmunoResearch, West Grove, PA). After washing We thank Dr. Matthew Vitalone for critically reading this manuscript. the plates with TBST buffer five times, the nonspecific protein binding Weappreciate the support from the Sarwal Laboratory members during was blocked by 100 ml of 5% dry milk in TBST buffer for 1 hour at room the study period and the patients and their families who participated in temperature. After the blocking step, 50-ml serum samples (40-fold di- this research study. We also thank the Stanford Functional Genomics luted with 2% milk in TBST buffer) were incubated in the wells for 1 Facility at Stanford University for providing scanning facilities for the hour at room temperature. The plates were washed fivetimeswithTBST protein arrays. buffer and incubated in 50 mlofAP-conjugatedAffiniPure mouse anti- human IgG (Jackson ImmunoResearch). The color was developed by using AP-pNPP Liquid Substrate System for ELISA (Sigma-Aldrich). DISCLOSURES Absorption was measured at 405 nm with a SpectraMax 190 None.

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 761 CLINICAL RESEARCH www.jasn.org

REFERENCES targets after renal transplantation by integrating transcriptome and “antibodyome” measures. Proc Natl Acad Sci USA 106: 4148–4153, 1. Hariharan S, Alexander JW, Schroeder TJ, First MR: Impact of first acute 2009 rejection episode and severity of rejection on cadaveric renal allograft 19. Naesens M, Lee S, Sigdel T, Kambham N, Li L, Sarwal M: Renal survival. Clin Transplant 10: 538–541, 1996 stanniocalcin-1 is involved in functional adaptation of adult-sized 2. Weir MR, Wali RK: Minimizing the risk of chronic allograft nephropathy. kidneys transplanted into pediatric recipients [Abstract 1035]. Am – Transplantation 87[Suppl]: S14–S18, 2009 JTransplant10[Suppl S4]: 340 341, 2010 3. Nankivell BJ, Alexander SI: Rejection of the kidney allograft. NEngl 20. Sboner A, Karpikov A, Chen G, Smith M, Mattoon D, Freeman-Cook L, JMed363: 1451–1462, 2010 Schweitzer B, Gerstein MB: Robust-linear-model normalization to re- 4. Nankivell BJ, Borrows RJ, Fung CL, O’Connell PJ, Allen RD, Chapman duce technical variability in functional protein microarrays. JProteome – JR: The natural history of chronic allograft nephropathy. NEnglJMed Res 8: 5451 5464, 2009 349: 2326–2333, 2003 21. Higgins JP, Wang L, Kambham N, Montgomery K, Mason V, 5. Solez K, Colvin RB, Racusen LC, Haas M, Sis B, Mengel M, Halloran PF, Vogelmann SU, Lemley KV, Brown PO, Brooks JD, van de Rijn M: Gene Baldwin W, Banfi G, Collins AB, Cosio F, David DS, Drachenberg C, expression in the normal adult human kidney assessed by comple- – Einecke G, Fogo AB, Gibson IW, Glotz D, Iskandar SS, Kraus E, Lerut E, mentary DNA microarray. Mol Biol Cell 15: 649 656, 2004 Mannon RB, Mihatsch M, Nankivell BJ, Nickeleit V, Papadimitriou JC, 22. Su AI, Cooke MP, Ching KA, Hakak Y, Walker JR, Wiltshire T, Orth AP, Randhawa P, Regele H, Renaudin K, Roberts I, Seron D, Smith RN, Vega RG, Sapinoso LM, Moqrich A, Patapoutian A, Hampton GM, Valente M: Banff 07 classification of renal allograft pathology: Updates Schultz PG, Hogenesch JB: Large-scale analysis of the human and – and future directions. Am J Transplant 8: 753–760, 2008 mouse transcriptomes. Proc Natl Acad Sci USA 99: 4465 4470, 2002 fi 6. Fletcher JT, Nankivell BJ, Alexander SI: Chronic allograft nephropathy. 23. Sigdel TK, Lee S, Sarwal MM: Pro ling the proteome in renal trans- – Pediatr Nephrol 24: 1465–1471, 2009 plantation. Proteomics Clin Appl 5: 269 280, 2011 7. Joosten SA, Sijpkens YW, van Kooten C, Paul LC: Chronic renal allograft 24. Naesens M, Khatri P, Li L, Sigdel TK, Vitalone MJ, Chen R, Butte AJ, fi rejection: Pathophysiologic considerations. Kidney Int 68: 1–13, 2005 Salvatierra O, Sarwal MM: Microarray expression pro ling associates 8. Mao Q, Terasaki PI, Cai J, Briley K, Catrou P, Haisch C, Rebellato L: progressive histological damage of renal allografts with innate and – Extremely high association between appearance of HLA antibodies and adaptive immunity. Kidney Int 80: 1364 1376, 2011 failure of kidney grafts in a five-year longitudinal study. Am J Transplant 25. Naylor SL, Sakaguchi AY, Shows TB, Law ML, Goeddel DV, Gray PW: 7: 864–871, 2007 Human immune interferon gene is located on chromosome 12. JExp 9. Opelz G Collaborative Transplant Study: Non-HLA transplantation Med 157: 1020–1027, 1983 immunity revealed by lymphocytotoxic antibodies. Lancet 365: 1570– 26. Bach EA, Aguet M, Schreiber RD: The IFN gamma receptor: a paradigm 1576, 2005 for cytokine receptor signaling. Annu Rev Immunol 15: 563–591, 1997 10. Sumitran-Holgersson S, Wilczek HE, Holgersson J, Söderström K: 27. Young HA: Regulation of interferon-gamma gene expression. Identification of the nonclassical HLA molecules, mica, as targets for J Interferon Cytokine Res 16: 563–568, 1996 humoral immunity associated with irreversible rejection of kidney al- 28. Carnaud C, Lee D, Donnars O, Park SH, Beavis A, Koezuka Y, Bendelac lografts. Transplantation 74: 268–277, 2002 A: Cutting edge: Cross-talk between cells of the innate immune system: 11. Sutherland SM, Li L, Sigdel TK, Wadia PP, Miklos DB, Butte AJ, Sarwal NKT cells rapidly activate NK cells. J Immunol 163: 4647–4650, 1999 MM: Protein microarrays identify antibodies to protein kinase Czeta 29. Frucht DM, Fukao T, Bogdan C, Schindler H, O’SheaJJ,KoyasuS:IFN- that are associated with a greater risk of allograft loss in pediatric renal gamma production by antigen-presenting cells: Mechanisms emerge. transplant recipients. Kidney Int 76: 1277–1283, 2009 Trends Immunol 22: 556–560, 2001 12. Sarwal MM, Yorgin PD, Alexander S, Millan MT, Belson A, Belanger N, 30. Harris DP, Haynes L, Sayles PC, Duso DK, Eaton SM, Lepak NM, Johnson Granucci L, Major C, Costaglio C, Sanchez J, Orlandi P, Salvatierra O Jr: LL, Swain SL, Lund FE: Reciprocal regulation of polarized cytokine pro- Promising early outcomes with a novel, complete steroid avoidance duction by effector B and T cells. Nat Immunol 1: 475–482, 2000 immunosuppression protocol in pediatric renal transplantation. 31. Zibar L, Wagner J, Pavlinić D, Galić J, Pasini J, Juras K, Barbić J: The Transplantation 72: 13–21, 2001 relationship between interferon-gamma gene polymorphism and 13. Reichelt O, Müller J, von Eggeling F, Driesch D, Wunderlich H, acute kidney allograft rejection. Scand J Immunol 73: 319–324, Schubert J, Gröne HJ, Stein G, Ott U, Junker K: Prediction of renal al- 2011 lograft rejection by urinary protein analysis using ProteinChip Arrays 32. Crispim JC, Wastowski IJ, Rassi DM, Mendes-Junior Silva CT, Bassi C, (surface-enhanced laser desorption/ionization time-of-flight mass Castelli EC, Costa RS, Saber LT, Silva TG, Donadi EA: Interferon-g +874 spectrometry). Urology 67: 472–475, 2006 polymorphism in the first intron of the human interferon-g gene and 14. Roche S, Dauvilliers Y, Tiers L, Couderc C, Piva MT, Provansal M, kidney allograft outcome. Transplant Proc 42: 4505–4508, 2010 Gabelle A, Lehmann S: Autoantibody profiling on high-density protein 33. Gołab J, Zagozdzon, Stokłosal T, Kaminski R, Kozar K, Jakóbisiak M: microarrays for biomarker discovery in the cerebrospinal fluid. Direct stimulation of macrophages by IL-12 and IL-18—abridgetoo J Immunol Methods 338: 75–78, 2008 far? Immunol Lett 72: 153–157, 2000 15. Butte AJ, Sigdel TK, Wadia PP, Miklos DB, Sarwal MM: Protein micro- 34. Schroder K, Hertzog PJ, Ravasi T, Hume DA: Interferon-gamma: An over- arrays discover angiotensinogen and PRKRIP1 as novel targets for au- view of signals, mechanisms and functions. J Leukoc Biol 75: 163–189, 2004 toantibodies in chronic renal disease. Mol Cell Proteomics 10: 35. Fischereder M, Schroppel B: The role of chemokines in acute renal al- M110.000497, 2011 lograft rejection and chronic allograft injury. Front Biosci 14: 1807– 16. Li L, Chen A, Chaudhuri A, Kambham N, Sigdel T, Chen R, Sarwal MM: 1814, 2009 Compartmental localization and clinical relevance of MICA antibodies 36. Rotondi M, Netti GS, Lazzeri E, Stallone G, Bertoni E, Chiovato L, after renal transplantation. Transplantation 89: 312–319, 2010 Grandaliano G, Gesualdo L, Salvadori M, Schena FP, Romagnani P, 17. Li L, Sigdel T, Vitalone M, Lee SH, Sarwal M: Differential immuno- Serio M: High pretransplant serum levels of CXCL9 are associated with genicity and clinical relevance of kidney compartment specifican- increased risk of acute rejection and graft failure in kidney graft recip- tigens after renal transplantation. J Proteome Res 9: 6715–6721, ients. Transpl Int 23: 465–475, 2010 2010 37. Lazzeri E, Rotondi M, Mazzinghi B, Lasagni L, Buonamano A, Rosati A, 18. Li L, Wadia P, Chen R, Kambham N, Naesens M, Sigdel TK, Miklos DB, Pradella F, Fossombroni V, La Villa G, Gacci M, Bertoni E, Serio M, Sarwal MM, Butte AJ: Identifying compartment-specificnon-HLA Salvadori M, Romagnani P: High CXCL10 expression in rejected kidneys

762 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 750–763, 2012 www.jasn.org CLINICAL RESEARCH

and predictive role of pretransplant serum CXCL10 for acute rejection and 47. Racusen LC, Halloran PF, Solez K: Banff 2003 meeting report: New chronic allograft nephropathy. Transplantation 79: 1215–1220, 2005 diagnostic insights and standards. Am J Transplant 4: 1562–1566, 2004 38. Chen R, Sigdel TK, Li L, Kambham N, Dudley JT, Hsieh SC, Klassen RB, 48. Jianghua C, Wenqing X, Huiping W, Juan J, Jianyong W, Qiang H: C4d Chen A, Caohuu T, Morgan AA, Valantine HA, Khush KK, Sarwal MM, as a significant predictor for humoral rejection in renal allografts. Clin Butte AJ: Differentially expressed RNA from public microarray data Transplant 19: 785–791, 2005 identifies serum protein biomarkers for cross-organ transplant rejection 49. Vargha R, Mueller T, Arbeiter K, Regele H, Exner M, Csaicsich D, and other conditions. PLOS Comput Biol 6: 6, 2010 Aufricht C: C4d in pediatric renal allograft biopsies: A marker for neg- 39. Howard OM, Dong HF, Yang D, Raben N, Nagaraju K, Rosen A, ative outcome in steroid-resistant rejection. Pediatr Transplant 10: Casciola-Rosen L, Härtlein M, Kron M, Yang D, Yiadom K, Dwivedi S, 449–453, 2006 Plotz PH, Oppenheim JJ: Histidyl-tRNA synthetase and asparaginyl- 50. Crespo M, Pascual M, Tolkoff-Rubin N, Mauiyyedi S, Collins AB, tRNA synthetase, autoantigens in myositis, activate chemokine re- Fitzpatrick D, Farrell ML, Williams WW, Delmonico FL, Cosimi AB, ceptors on T lymphocytes and immature dendritic cells. JExpMed196: Colvin RB, Saidman SL: Acute humoral rejection in renal allograft re- 781–791, 2002 cipients: I. Incidence, serology and clinical characteristics. Trans- 40. Moore MW, Klein RD, Fariñas I, Sauer H, Armanini M, Phillips H, plantation 71: 652–658, 2001 Reichardt LF, Ryan AM, Carver-Moore K, Rosenthal A: Renal and neu- 51. Mauiyyedi S, Colvin RB: Humoral rejection in kidney transplantation: ronal abnormalities in mice lacking GDNF. Nature 382: 76–79, 1996 new concepts in diagnosis and treatment. Curr Opin Nephrol Hyper- 41. Basson MA, Akbulut S, Watson-Johnson J, Simon R, Carroll TJ, Shakya tens 11: 609–618, 2002 R, Gross I, Martin GR, Lufkin T, McMahon AP, Wilson PD, Costantini FD, 52. Pontén F, Jirström K, Uhlen M: The Human Protein Atlas—a tool for Mason IJ, Licht JD: Sprouty1 is a critical regulator of GDNF/RET- pathology. J Pathol 216: 387–393, 2008 mediated kidney induction. Dev Cell 8: 229–239, 2005 53. Berglund L, Björling E, Oksvold P, Fagerberg L, Asplund A, Szigyarto 42. Sainio K, Suvanto P, Davies J, Wartiovaara J, Wartiovaara K, Saarma M, CA, Persson A, Ottosson J, Wernérus H, Nilsson P, Lundberg E, Arumäe U, Meng X, Lindahl M, Pachnis V, Sariola H: Glial-cell-line-derived Sivertsson A, Navani S, Wester K, Kampf C, Hober S, Pontén F, Uhlén M: neurotrophic factor is required for bud initiation from ureteric epithelium. A genecentric Human Protein Atlas for expression profiles based on Development 124: 4077–4087, 1997 antibodies. Mol Cell Proteomics 7: 2019–2027, 2008 43. Sarwal M, Chua MS, Kambham N, Hsieh SC, Satterwhite T, Masek M, 54. Uhlén M, Björling E, Agaton C, Szigyarto CA, Amini B, Andersen E, Salvatierra O Jr: Molecular heterogeneity in acute renal allograft re- Andersson AC, Angelidou P, Asplund A, Asplund C, Berglund L, jection identified by DNA microarray profiling. N Engl J Med 349: 125– Bergström K, Brumer H, Cerjan D, Ekström M, Elobeid A, Eriksson C, 138, 2003 Fagerberg L, Falk R, Fall J, Forsberg M, Björklund MG, Gumbel K, 44. Yilmaz S, Tomlanovich S, Mathew T, Taskinen E, Paavonen T, Navarro Halimi A, Hallin I, Hamsten C, Hansson M, Hedhammar M, Hercules G, M, Ramos E, Hooftman L, Häyry P: Protocol core needle biopsy and Kampf C, Larsson K, Lindskog M, Lodewyckx W, Lund J, Lundeberg J, histologic Chronic Allograft Damage Index (CADI) as surrogate end Magnusson K, Malm E, Nilsson P, Odling J, Oksvold P, Olsson I, Oster point for long-term graft survival in multicenter studies. JAmSoc E, Ottosson J, Paavilainen L, Persson A, Rimini R, Rockberg J, Runeson Nephrol 14: 773–779, 2003 M, Sivertsson A, Sköllermo A, Steen J, Stenvall M, Sterky F, Strömberg 45. Isoniemi H, Taskinen E, Häyry P: Histological chronic allograft damage S, Sundberg M, Tegel H, Tourle S, Wahlund E, Waldén A, Wan J, index accurately predicts chronic renal allograft rejection. Trans- Wernérus H, Westberg J, Wester K, Wrethagen U, Xu LL, Hober S, plantation 58: 1195–1198, 1994 Pontén F: A human protein atlas for normal and cancer tissues based on 46. Racusen LC, Solez K, Colvin RB, Bonsib SM, Castro MC, Cavallo T, antibody proteomics. Mol Cell Proteomics 4: 1920–1932, 2005 Croker BP, Demetris AJ, Drachenberg CB, Fogo AB, Furness P, Gaber LW, Gibson IW, Glotz D, Goldberg JC, Grande J, Halloran PF, Hansen HE, Hartley B, Hayry PJ, Hill CM, Hoffman EO, Hunsicker LG, Lindblad AS, Yamaguchi Y: The Banff 97 working classification of renal allograft This article contains supplemental material online at http://jasn.asnjournals. pathology. Kidney Int 55: 713–723, 1999 org/lookup/suppl/doi:10.1681/ASN.2011060596/-/DCSupplemental.

J Am Soc Nephrol 23: 750–763, 2012 Antibodies for CAI 763 Supplemental material for JASN 2011060596

Supplemental Table 1: A list of non-HLA antibodies that was significantly increased in the sera of chronic allograft imnjury phenotype.

S.No P- . Gene Symbol Protein Name value 0.000 1 BHMT2 betaine-homocysteine methyltransferase 2 (BHMT2) 1 0.000 2 GDNF Glial-derived Neurotrophic factor (GDNF) 1 3 MEF2D myocyte enhancer factor 2D (MEF2D) 0.001 4 TWF1 Twinfilin-1 0.002 5 PPID peptidylprolyl isomerase D (cyclophilin D) (PPID) 0.002 epidermal growth factor receptor (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian) 6 EGFR L861Q (EGFR), transcript variant 1 0.002 7 MAPRE2 microtubule-associated protein, RP/EB family, member 2 (MAPRE2) 0.002 8 CSNK1G3 casein kinase 1, gamma 3 (CSNK1G3) 0.002 9 6CKINE C-C motif chemokine 21 0.003 10 FLJ21908 RNA polymerase II-associated protein 3 0.003 11 MEF2D myocyte enhancer factor 2D (MEF2D) 0.001 12 LOH11CR2A loss of heterozygosity, 11, chromosomal region 2, gene A (LOH11CR2A) 0.004 13 PRKCE protein kinase C, epsilon (PRKCE) 0.004 14 DMPK dystrophia myotonica-protein kinase (DMPK), transcript variant 2 0.004 15 PKN1 Serine/threonine-protein kinase N1 0.004 16 ACY1 aminoacylase 1 (ACY1) 0.004 17 CUEDC1 CUE domain containing 1 (CUEDC1) 0.004 fms-related tyrosine kinase 1 (vascular endothelial growth factor/vascular permeability factor receptor) 18 FLT1 (FLT1) 0.004 19 SLAIN2 SLAIN motif family, member 2 (SLAIN2) 0.004 20 EIF5A2 eukaryotic translation initiation factor 5A2 (EIF5A2) 0.004 21 CSNK1A1 casein kinase 1, alpha 1 (CSNK1A1), transcript variant 2 0.004 epidermal growth factor receptor (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian) 22 EGFR (EGFR); see catalog number for detailed information on wild-type or point mutant status 0.004 23 CCL19 chemokine (C-C motif) ligand 19 (CCL19) 0.007 leucine-rich repeat kinase 2 (LRRK2); see catalog number for detailed information on wild-type or point 24 LRRK2 mutant status 0.007 25 KLK6 kallikrein-related peptidase 6 (KLK6), transcript variant A 0.007 epidermal growth factor receptor (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian) 26 EGFR (EGFR); see catalog number for detailed information on wild-type or point mutant status 0.004 27 ACBD6 acyl-Coenzyme A binding domain containing 6 (ACBD6) 0.007 28 GYG2 glycogenin 2 (GYG2) 0.009 29 RET Proto-oncogene tyrosine-protein kinase receptor ret 0.010 30 SRPK2 Serine/threonine-protein kinase SRPK2 0.010 31 ATXN3 Ataxin-3 0.010 v-abl Abelson murine leukemia viral oncogene homolog 1 (ABL1), transcript variant a; see catalog number 32 ABL1 for detailed information on wild-type or point mutant status 0.010 33 IFN-GAMMA Interferon gamma 0.010 34 C20orf18 chromosome 20 open reading frame 18 (C20orf18), transcript variant 4 0.010 35 CTTN cortactin (CTTN), transcript variant 2 0.010 cDNA clone IMAGE:482924 36 5 cDNA clone IMAGE:4829245 0.010 37 VDR vitamin D (1,25- dihydroxyvitamin D3) receptor (VDR), transcript variant 1 0.010 38 SMTNL2 Smoothelin-like protein 2 0.010 39 CSNK2A2 casein kinase 2, alpha prime polypeptide (CSNK2A2) 0.011 40 Jo-1/HARS histidyl-rRNA synthetase (Jo-1) 0.013 Janus kinase 2 (a protein tyrosine kinase) (JAK2) ; see catalog number for detailed information on wild- 41 JAK2 type or point mutant status 0.015 42 RNA-POL RNA Polymerase 0.015 43 HMGN1 high-mobility group nucleosome binding domain 1 (HMGN1) 0.015 44 PDGFRB platelet-derived growth factor receptor, beta polypeptide (PDGFRB) 0.015 45 WIBG within bgcn homolog (Drosophila) (WIBG) 0.015 46 CSNK2A1 casein kinase 2, alpha 1 polypeptide (CSNK2A1), transcript variant 2 0.020 47 LUC7L LUC7-like (S. cerevisiae) (LUC7L), transcript variant 1 0.020 48 CAMKK2 Calcium/calmodulin-dependent protein kinase kinase 2 0.020 49 RAR-beta LBD Retinoic Acid Receptor beta, Ligand Binding Domain (RAR beta-LBD) 0.020 50 MIG C-X-C motif chemokine 9 0.022 51 TRIM69 tripartite motif-containing 69 (TRIM69), transcript variant b 0.022 52 LZTFL1 leucine zipper transcription factor-like 1 (LZTFL1) 0.022 53 PRKCI protein kinase C, iota (PRKCI) 0.022 54 ROS1 v-ros UR2 sarcoma virus oncogene homolog 1 (avian) (ROS1) 0.022 55 WEE1 WEE1 homolog (S. pombe) (WEE1) 0.022 56 FLT4 fms-related tyrosine kinase 4 (FLT4), transcript variant 2 0.022 57 EPHA1 Ephrin receptor A1 (EPHA1) 0.022 58 TCP10 t-complex 10 (mouse) (TCP10) 0.024 59 NEK3 NIMA (never in mitosis gene a)-related kinase 3 (NEK3), transcript variant 1 0.024 60 CLK4 CDC-like kinase 4 (CLK4) 0.024 61 CALCOCO1 calcium binding and coiled-coil domain 1 (CALCOCO1) 0.024 62 MARK4 MAP/microtubule affinity-regulating kinase 4 0.024 63 GYS1 glycogen synthase 1 (muscle) (GYS1) 0.024 64 MAP3K11 mitogen-activated protein kinase kinase kinase 11 (MAP3K11) 0.024 65 DYRK1A Dual specificity tyrosine-phosphorylation-regulated kinase 1A 0.024 66 ALK ALK tyrosine kinase receptor 0.024 67 IL-8 Interleukin-8 0.024 68 MSN moesin (MSN) 0.024 69 IRS1 insulin receptor substrate 1 (IRS1) 0.024 70 ITAC C-X-C motif chemokine 11 0.024 71 G3BP1 GTPase activating protein (SH3 domain) binding protein 1 (G3BP1), transcript variant 2 0.024 72 VIM Vimentin 0.024 73 RIPK5 serine?threonine kinase 0.027 74 SGK2 serum/glucocorticoid regulated kinase 2 (SGK2), transcript variant 1 0.027 75 PPCDC phosphopantothenoylcysteine decarboxylase (PPCDC) 0.027 76 TUBG1 tubulin, gamma 1 (TUBG1) 0.027 77 CSNK2A1 casein kinase 2, alpha 1 polypeptide (CSNK2A1), transcript variant 2 0.020 78 UBE2S Ubiquitin-conjugating enzyme E2 S 0.032 79 CCDC55 coiled-coil domain containing 55 (CCDC55), transcript variant 1 0.032 80 NMT2 N-myristoyltransferase 2 (NMT2) 0.032 81 LBD Thyroid hormone receptor beta 0.032 82 ERBB2 Receptor tyrosine-protein kinase erbB-2 0.032 83 B2GP1 beta-2 Glycoprotein I (Apolipoprotein H) 0.032 84 SGK3 Serine/threonine-protein kinase Sgk3 0.032 85 RPS6KB1 ribosomal protein S6 kinase, 70kDa, polypeptide 1 (RPS6KB1) 0.032 86 JAK3 Tyrosine-protein kinase JAK3 0.032 epidermal growth factor receptor (erythroblastic leukemia viral (v-erb-b) oncogene homolog, avian) 87 EGFR L858R (EGFR), transcript variant 1 0.041 88 COL23A1 collagen, type XXIII, alpha 1 (COL23A1) 0.041 89 BIN1 bridging integrator 1 (BIN1), transcript variant 6 0.046 90 CSNK1G1 casein kinase 1, gamma 1 (CSNK1G1) 0.046 91 FLJ11184 hypothetical protein FLJ11184 (FLJ11184) 0.046 92 BAG5 BCL2-associated athanogene 5 (BAG5), transcript variant 2 0.046 93 PRKCB1 protein kinase C, beta 1 (PRKCB1), transcript variant 2 0.046 94 IL21 interleukin 21 (IL21) 0.046 95 BIN1 bridging integrator 1 (BIN1), transcript variant 6 0.046 96 MUSK muscle, skeletal, receptor tyrosine kinase (MUSK) 0.046 97 AFAP1L2 actin filament associated protein 1-like 2 (AFAP1L2), transcript variant 1 0.046 98 NDE1 nudE nuclear distribution gene E homolog 1 (A. nidulans) (NDE1) 0.046 99 LUC7L2 LUC7-like 2 (S. cerevisiae) (LUC7L2) 0.046 100 SIVA1 Apoptosis regulatory protein Siva 0.046 101 PRKCZ protein kinase C, zeta (PRKCZ) 0.046 102 SNF1LK2 SNF1-like kinase 2 (SNF1LK2) 0.046 103 PDGFRA platelet-derived growth factor receptor, alpha polypeptide (PDGFRA) 0.046 104 ACVR1 (ALK2) Activin receptor type-1 0.046 105 MAP3K2 Mitogen-activated protein kinase kinase kinase 2 0.046 106 CSNK2A1 casein kinase 2, alpha 1 polypeptide (CSNK2A1), transcript variant 2 0.020 107 MYLK2 myosin light chain kinase 2, skeletal muscle (MYLK2) 0.048 108 PDAP1 PDGFA associated protein 1 (PDAP1) 0.048 109 INTS3 integrator complex subunit 3 (INTS3) 0.048 110 GABPA GA binding protein transcription factor, alpha subunit 60kDa (GABPA) 0.048 111 TPRXL tetra-peptide repeat homeobox-like (TPRXL) 0.048

Supplemental Figure 1: [LEGEND] The late responding non-HLA proteins were in pathways such as cell death, cell-to-cell signaling and interaction, and cellular movement