Expression of Surfactant Protein-C, S100A8, S100A9, and B Cell Markers in Renal Allografts: Investigation of the Prognostic Value

Home , S100A8, S100A9, Surfactant protein C

Michael Eikmans,* Marian C. Roos-van Groningen,* Yvo W.J. Sijpkens,† Jan Ehrchen,‡ Johannes Roth,‡§ Hans J. Baelde,* Ingeborg M. Bajema,* Johan W. de Fijter,† Emile de Heer,* and Jan A. Bruijn* *Department of Pathology, Leiden University Medical Center, Leiden, The Netherlands; †Department of Nephrology, Leiden University Medical Center, Leiden, The Netherlands; and ‡Institute of Experimental Dermatology and §Department of Pediatrics, University of Mu¨nster, Munich, Germany

The intent of this study was to identify genes of which expression during acute rejection is associated with progression to chronic allograft nephropathy using gene expression profiling. Ten patients who had graft loss through chronic allograft nephropathy (progression [PR] group) and 18 patients who had stable graft function over time (nonprogression [NP] group) were studied. Rejection severity and extent of infiltrating leukocytes in acute rejection biopsies were similar for both groups. Microarray analysis and real-time PCR validation showed that surfactant protein-C (SP-C), S100 calcium-binding protein A8 (S100A8), S100A9, and ␤-globin levels distinguished the two groups. Relationship between expression of B cell markers and prognosis was also examined. Location in the graft of the protein and mRNA expression of candidate genes was investigated. The prognostic value of mRNA transcripts was tested in an independent cohort of 43 rejection biopsies. mRNA and protein expression of S100A8 and S100A9 in infiltrating cells was significantly higher in the NP group compared with the PR group. Expression of SP-C was four-fold higher in the PR group and was detected in glomeruli. No association between B cell clusters and outcome was found. In the second group of acute rejection biopsies, SP-C mRNA levels predicted renal function course beyond 6 mo in multivariate analysis. Relatively high expression of S100A8 and S100A9 during acute rejection is associated with a favorable prognosis, and high SP-C expression is associated with an unfavorable prognosis. Messenger RNA transcripts complement the biopsy in the prediction of graft function deterioration. J Am Soc Nephrol 16: 3771–3786, 2005. doi: 10.1681/ASN.2005040412

hanges in mRNA levels of inflammatory components occurrence of acute interstitial rejection (15). Second, the nature and extracellular matrix (ECM)-regulating molecules of the inflammatory infiltrate and the restorative effect of the C within the first 12 mo after transplantation are associ- renal tissue upon acute rejection may partly account for pa- ated with progressive allograft dysfunction (1–4). Various re- tients’ predisposition to develop CAN (16). Recently, Sarwal et ports have emphasized the capacity of microarray technology al. (4) identified three distinct subtypes of acute rejection with to find genes and elucidate molecular pathways that are in- cDNA microarray profiling and showed that the composition volved in the progression of renal allograft damage (4–10). of the infiltrate is related to the prognosis: A high expression of Identification of prognostic factors in the transplantation set- B cell–specific genes during acute rejection predicts an adverse ting may lead to the development of improved intervention outcome. strategies and may contribute to a better understanding of the We intended to identify genes whose early expression after pathophysiology of chronic allograft nephropathy (CAN). transplantation is related to the occurrence of CAN. Such genes In kidney transplantation, CAN is the major cause of late may be involved in the pathogenesis of CAN. Unlike Sarwal’s graft failure (11). The number of acute rejection episodes and approach, we selected two groups of patients who expressed their timing and severity are the strongest predictors of CAN acute rejection within the setting of a case-control study: Those and graft failure (12–14). However, not all acute rejection epi- who had an unfavorable prognosis and lost their grafts as a sodes lead to CAN. First, the occurrence of acute vascular result of CAN within 5 yr after transplantation and those who rejection is associated with a more adverse prognosis than the retained stable graft function after acute rejection for at least 6 yr. In a previous study, TGF-␤ mRNA levels in early acute Received April 19, 2005. Accepted September 15, 2005. rejection biopsies were associated with graft prognosis (2). In this study, we performed gene expression profiling to identify Published online ahead of print. Publication date available at www.jasn.org. molecules for which the expression during acute rejection is Address correspondence to: Dr. Michael Eikmans, Leiden University Medical Center, Department of Pathology, Albinusdreef 2, P.O. Box 9600, 2300 RC, Leiden, The Nether- associated with graft outcome. Several candidate genes were lands. Phone: ϩ31-71-526-6574; Fax: ϩ31-71-524-8158; E-mail: [email protected] studied to investigate location of the mRNA and the protein in

Copyright © 2005 by the American Society of Nephrology ISSN: 1046-6673/1612-3771 3772 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005 the graft tissue and to test whether protein levels also could Control Groups discriminate patient groups with different outcomes. The pre- In all molecular analyses, two control groups were analyzed. One dictive value of the candidate genes was investigated further in group consisted of 6-mo protocol transplant biopsies. The patients Ϯ an independent cohort of acute rejection biopsies. (45.0 6.0 yr) showed stable graft function at time of biopsy, and they received cyclosporine as maintenance therapy (initial dosage of 4 mg/kg twice daily). The biopsies showed no evidence of CAN, acute Materials and Methods rejection, or drug toxicity. The other control group consisted of three Selection of Patient Groups cadaveric donor kidneys that initially were intended for transplanta- To establish two patient groups that represent extremes with respect tion purposes: One sample from the unaffected part of a tumor ne- to outcome, we constructed from 654 consecutive kidney transplants phrectomy kidney and two autopsy kidneys. Mean age was 55.7 Ϯ 18.7 regression lines through reciprocal serum creatinine values beyond 6 yr. The kidneys showed normal histology, and the patients had normal mo after transplantation. These had been performed between 1986 and renal function and no proteinuria. For the immunohistochemical stain- 1995 at the Leiden University Medical Center (17,18). From this cohort, ings, follow-up biopsies from the PR group (patients 1, 3 through 5, and patients who met all of the following criteria were selected: (1) Recip- 7 through 9; see Table 4), showing signs of CAN, were analyzed. ient of a cadaveric donor kidney, (2) treated with cyclosporine (Sand- immune formula; initial dosage of 4 mg/kg twice daily) as maintenance Assessment of Morphologic Alterations immunosuppressive regimen, (3) a graft survival of at least 6 mo, and The extent of infiltrate, chronic damage, and C4d staining during (4) at least one biopsy-supported rejection episode within the first 6 mo acute rejection in individual patients was determined. The extent of after transplantation (n ϭ 200). Two patient groups were defined: One interstitial fibrosis was assessed in Sirius red–stained paraffin sections, group consisted of patients who lost their graft as a result of biopsy- according to methods that have been described elsewhere (19). The supported CAN within 5 yr (progression [PR] group; n ϭ 29). From this percentage of the Sirius red–positive area of the total tubulointerstitial group, 10 patients (11 samples) were included. The other group re- surface was quantified by digital image analysis (20). Histology of all tained stable graft function for at least 6 yr (nonprogression [NP] group; biopsies was evaluated according to Banff 1997 criteria (21). Differences n ϭ 160). From this group, 18 patients (20 samples) were included. in renal gene expression levels between patient groups may be the Slope values of the regression lines and graft survival plots are shown result of differences in the number of inflammatory cells. Therefore, the in Figure 1. total extent of infiltrate, the amount of granulocytes, and the amount of From the 28 patients who were included in the selection, either the plasma cells between groups were compared. These were scored semi- only or the last biopsy with rejection occurring within the first 6 mo quantitatively on a scale from 0 to 3 in silver stainings. The total amount after transplantation was used for molecular analyses. Of three pa- of leukocytes and monocytes was assessed by stainings on paraffin tients, two biopsy samples were included. Table 4 shows the rejection sections for CD45 (Dakocytomation, Leuven, Belgium) and CD68 episodes during which the biopsies were obtained. Of the 31 biopsies (Dako, Glostrup, Denmark), respectively. The extent of staining in the studied, 15 were taken during a first-rejection episode and 16 were renal cortices was quantified by image analysis. Immunofluorescence taken during a follow-on rejection episode. Treatment of rejections staining for C4d was performed on frozen slides using mouse anti-C4d consisted of methylprednisolone, anti-thymocyte globulin, and meth- antibody (Quidel, San Diego, CA). The extent of staining in the peritu- ylprednisolone for first, second, and third rejection episodes, respec- bular capillaries was scored as negative, weakly positive, or strongly tively. positive.

Figure 1. Outcome parameters in patient groups. (A) Slope values of regression lines through reciprocal serum creatinine values beyond 6 mo after transplantation in the progression group (PR) and in the nonprogression (NP) group. (B) Kaplan-Meier curves of graft loss in the PR group (dotted line) and in the NP group (solid line). Patients in the PR group lost their graft within 5 yr after transplantation. Follow-up in the NP group was at least 6 yr. Crosses indicate end of follow-up. J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3773

RNA Extraction and cDNA Synthesis transcripts that were differentially expressed between groups showed a For microarray and real-time analyses, RNA was extracted from the significance level of the change call of lower than 0.01. renal cortex of frozen biopsies (22). For removing as accurately as possible only renal cortex from the biopsies, cryostat sections were cut Real-Time PCR and the cortex was pinpointed on the basis of light-microscopic local- To validate microarray results, we studied 10 known genes in the ization of glomeruli in the sections (2). Regions that did not contain microarray list with relative hybridization intensity of at least 100 in glomeruli, which may represent medulla, were removed from the one of the patient groups by real-time PCR in cDNA samples from biopsy cores before sections were cut for RNA extraction. Such regions individual patients. Primer sets (Table 1) were obtained from Isogen were detected only in the minority of the biopsies. Half of the RNA (Maarssen, The Netherlands). Message for TGF-␤ and IL-10 was also from each individual sample was used for cDNA synthesis (2) and quantified. Sequences for TGF-␤ have been described previously (2). real-time PCR. The other half was used for microarray. RNA was Sequences for IL-10 were a gift of Dr. Franka Hartgers (Department of purified further (RNeasy mini kit; Qiagen, Chatsworth, CA) and tested Parasitology, Leiden, The Netherlands). The real-time PCR protocol for integrity by gel electrophoresis. For supplying the microchips with and reagents were identical to those described in another publication sufficient cRNA for hybridization, pools of RNA were produced for (24). The mean signal of four stably expressed household genes (25), both groups. After RNA extraction and cleanup, biopsies from the PR namely hypoxanthine guanine phosphoribosyltransferase-1 (HPRT-1), ␮ group (nine samples) rendered 11 g of total RNA, and biopsies from porphobilinogen deaminase (PBGD), glyceraldehyde-3-phosphate de- ␮ the NP group (13 samples) rendered 25 g of total RNA. Integrity of the hydrogenase (GAPDH), and TATA box-binding protein (TBP), served RNA samples proved to be intact on an agarose gel. as a normalization factor for the reverse transcriptase–PCR data (24). Raw data of each of the household genes in the microarray and real- Microarray Analysis time PCR analyses between the two acute rejection groups with different outcomes and the control group of stable grafts were not signifi- To detect differences at the mRNA level between patient groups, we cantly different. For real-time PCR, the mean Ct value in the three examined gene expression profiles with Human Genome U95Av2 ar- groups was as follows: HPRT-1 30.8 Ϯ 2.4, 30.4 Ϯ 2.4, and 30.8 Ϯ 1.6; rays (Affymetrix, Santa Clara, CA) that contain Ͼ12,000 cDNA se- PBGD 33.2 Ϯ 2.4, 35.4 Ϯ 3.5, and 36.8 Ϯ 4.6; GAPDH 24.1 Ϯ 2.8, 25.3 Ϯ quences. Procedures were carried out according to the manufacturer’s 3.0, and 26.7 Ϯ 3.3; and TBP 30.1 Ϯ 2.6, 32.3 Ϯ 2.7, and 32.8 Ϯ 2.4. In the instructions. Shortly thereafter, 10 ␮g of total RNA was used in a microarray experiments, mean hybridization intensities in the two re- first-strand cDNA synthesis. After second-strand cDNA synthesis and jection groups were as follows: HPRT-1 55 and 48 Ϯ 8, PBGD 12 and in vitro transcription, 12 ␮g of biotin-labeled cRNA was available for the 19 Ϯ 20, GAPDH 1021 and 1037 Ϯ 74, and TBP 20 and 31 Ϯ 33. PR group. After cleanup and fragmentation, all of this cRNA was applied to a microchip. For the NP group, 40 ␮g of biotin-labeled cRNA was available, allowing a duplicate hybridization experiment. Each of Immunohistochemistry these duplicates was compared with results from the PR group. Nor- Immunohistochemical analyses were performed on biopsies from the malization of expression data and statistical evaluation were performed two rejection groups to investigate the location of S100A8, S100A9, according to methods that have previously been described (23) and that CD79B, and SP-C protein in the grafts and to test whether protein can be found in the Microsuite User’s Guide, version 5.0 (http:// expression could discriminate between patient groups. Antibodies for www.affymetrix.com/support/technical/manuals.affx). In brief, Wil- S100A8 and S100A9 monomers and S100A8/A9 heterodimers (Ab coxon’s signed rank tests were performed by the Microarray Suite 27E10) have been described previously (26). Antibodies against CD79b software to determine P values when expression levels of transcripts were a gift of Dr. Jackie Cordell (Headington, England). Antibodies from one array were compared with those from another. All mRNA against CD79a and CD20 were obtained from Dako. Antibodies (SN-

Table 1. Sequences for real-time PCR and characteristics of antibodies

Gene Forward Primer Reverse Primer Probe

Surfactant tctccacatgagccagaaacac cgttgctgggcttccg FAM-caatgctcatctccagaaccatctccg protein C Collagen cctcaagggctccaacgag tcaatcactgtcttgcccca TET-atggctgcacgagtcacaccgga ␣1(I) Collagen catgccgtgacttgagactca gtatactttgatagcatccatagtgcatc FAM-cagtagtaaccactgctccactctgggtgg ␣2(I) Dead/ tggaagacttagataaacggtcatttc tttttggtctccacaaacactaaagta TET-aatcactccctgttgcacctaaaatgtccagt H box Y PPAP2A aggccactcttcgttttcca cccagtctcccttcatcctg FAM-tactgcatgctgtttgtggcactttatcttca ␤-globin gctgcactgtgacaagctgc gcacacagaccagcacgttg TET-cgtggatcctgagaacttcaggctcct ␣-globin gcgcacaagcttcgggt agggtcaccagcaggcag TET-cccggtcaacttcaagctcctaagcc S100A8 gggaatttccatgccgtct cctttttcctgatatactgaggacact TET-ggtctctagcaatttcttcaggtcatccctg S100A9 ctgtgtggctcctcggct gcgttccagctgcgacat FAM-caagtcatcgtcttgcactctgtccca CD79b catcatcgtgcctatcttcctg gtaggtgtgatcttcctccatgc TET-gccttgctgtcatccttgtccagca CD45 ggtcgtcaaacaaaaacttccc tgagatccatccctgcagtg FAM-tgaagggaacaagcatcacaagagtacacctct 3774 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

Table 2. Characteristics of antibodies

Protein Pretreatment Type Dilution Incubation Time Secondary Antibody Detection

CD45 — Monoclonal 1:400 1 h Mouse-Envision DAB CD68 Trypsin Monoclonal 1:100 1 h Mouse-Envision NovaRed C4d — Monoclonal 1:200 1 h FITC–anti-mouse S100A8 Prot K Polyclonal 1:300 Overnight Rabbit/peroxidase AEC S100A9 Prot K Polyclonal 1:500 Overnight Rabbit/peroxidase AEC 27E10 Prot K Monoclonal 1:10 Overnight Mouse/peroxidase AEC CD79a Citrate Monoclonal 1:200 1 h Mouse-Envision NovaRed CD79b Tris/EDTA Monoclonal 1:75 Overnight Envision NovaRed CD20 — Monoclonal 1:400 1 h Mouse-Envision NovaRed Pro SP-C — Polyclonal 1:2000 1 h Rabbit NovaRed CD31 Citrate Monoclonal 1:400 1 h Envision NovaRed Desmin — Monoclonal 1:2000 1 h Envision NovaRed Vimentin — Polyclonal 1:1000 Overnight Envision NovaRed

524) that recognize the first 15 N-terminal amino acids of proSP-C were positive cells in the single high-power field (hpf) with the highest a gift of Dr. Koos Batenburg (University of Utrecht, Utrecht, The positive cell count identified, as described in another report (4). Spec- Netherlands). To gain insight into the site of proSP-C protein expres- imens were considered positive in case of a cell count Ͼ275/hpf (4). To sion in the glomerulus, we performed proSP-C stainings on sequential quantify the amount of glomerular proSP-C protein, we calculated the sections with the help of the endothelial marker CD31 (Dako) and the percentage of staining in the total glomerular capillary tuft area. Data mesenchymal markers desmin and vimentin (Euro-Diagnostica, Arn- are represented as the percentage of glomeruli showing proSP-C stain- hem, The Netherlands). ing in Ͼ1% of the capillary tuft area. Immunohistochemistry was performed according to a standardized protocol (27). Details concerning antigen retrieval and staining procedures are given in Table 2. For all stainings, computerized image RNA In Situ Hybridization analysis was used to quantify the overall core percentage of staining in To determine the location of mRNA transcripts in the grafts, we the renal cortical tissue (20). For B cell markers CD20 and CD79, we also performed RNA in situ hybridization (RISH). The 282-nucleotide frag- focused on cell clusters and depicted for each specimen the number of ments that correspond to the coding region of human S100A8 (nucle-

Table 3. Clinical characteristics of patient groupsa

PR Group NP Group

Donor age (yr) 39.2 Ϯ 17.5 41.7 Ϯ 14.3 Donor gender (% male) 70 39 Underlying renal disease (% hereditary/ 0/30/30/40 22/44/17/17 glomerular/systemic/other) Transplantation era (% 1983–1987, 1988–1992, 40/40/20 28/61/11 1993–1996) Recipient age (yr) 40.5 Ϯ 16.9 41.9 Ϯ 16.3 Recipient gender (% male) 70 78 Panel reactive antibodies 28.0 Ϯ 30.2 26.1 Ϯ 27.2 Body weight (kg) 62.9 Ϯ 15.9 70.1 Ϯ 10.9 Cold ischemia time (h) 26.5 Ϯ 5.9 29.1 Ϯ 6.5 Delayed graft function (%) 40 44 CMV infection (%) 30 28 Antihypertensive therapy (%) 10 17 SBP at 6 mo (mmHg) 132 Ϯ 22 141 Ϯ 19 DBP at 6 mo (mmHg) 80 Ϯ 13 86 Ϯ 10 Creat clearance at 6 mo (ml/min) 48.9 Ϯ 24.3 61.2 Ϯ 19.0 Proteinuria at biopsy (% Ͼ trace) 50 33 Proteinuria at 6 mo (% Ͼ trace) 67 33b

aPR, progression; NP, nonprogression; CMV, cytomegalovirus; SBP, systolic BP; DBP, diastolic BP. bP Ͻ 0.05 versus PR group. J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3775

Table 4. Biopsy samples included in the studya

No. Banff Rejection Type ARb Timec Graft Loss Arrayd

PR group 1 Ia Interstitial 2 154 19 ϩ 2 Ib Interstitial 1 6 119 3 III Vascular 3 117 12 ϩ 4a IIb Vascular 1 9 58 ϩ 4b III Vascular 2 30 58 ϩ 5 Ib Interstitial 2 92 23 ϩ 6 Ib Interstitial 2 37 23 ϩ 7 Ib Interstitial 3 64 7 ϩ 8 Ib Interstitial 1 39 47 ϩ 9 Bord Interstitial 1 10 34 ϩ 10 Bord Interstitial 2 94 11 27% vascular 1.8 59 Ϯ 49 37 Ϯ 33 NP group 1 Ib Interstitial 1 29 147 2 III Vascular 1 36 141 ϩ 3 Ib Interstitial 2 35 127 ϩ 4 Ib Interstitial 2 16 122 ϩ 5 Ia Interstitial 1 8 106 6 Ib Interstitial 1 7 96 7 Ib Interstitial 1 22 94 ϩ 8 Ia Interstitial 1 67 88 ϩ 9a Ia Interstitial 1 7 82 ϩ 9b III Vascular 3 33 82 ϩ 10 Ia Interstitial 2 171 82 ϩ 11a Ib Interstitial 1 19 82 11b IIb Vascular 2 29 82 ϩ 12 Ib Interstitial 1 9 81 ϩ 13 Ia Interstitial 2 34 79 ϩ 14 Ia Interstitial 1 40 76 ϩ 15 Bord Interstitial 2 64 72 ϩ 16 III Vascular 2 24 70 17 Bord Interstitial 2 68 70 18 Ib Interstitial 1 133 67 20% vascular 1.5 43 Ϯ 42 92 Ϯ 24

aSamples were analyzed individually with real-time PCR. bRejection episode at which biopsy was obtained. First, second, and third rejection episodes were treated with methylprednisolone, anti-thymocyte globulins, and methylprednisolone, respectively. cNumber of days after transplantation that had passed before biopsy. dϩ, sample included in microarray analysis.

otide 56 to 338 of sequence X06234) were cloned into pBluescript predictive value in a second set of rejection biopsies. Messenger RNA (Stratagene, La Jolla, CA) and used for RISH. Plasmid pHSP-C6 that levels of SP-C, S100A8, and ␤-globin were measured in 43 biopsies with contained the full-length human SP-C cDNA (28) was a gift of Dr. acute rejection from 43 patients. CD79b mRNA levels were also stud- Susan Wert (Cincinnati, OH). For antisense and sense probes, 730- ied. The majority of the biopsies had been collected between 1998 and nucleotide and 880-nucleotide fragments were used, respectively. The 2002 in a consecutive manner. Established risk factors, including donor RISH was performed according to earlier described protocols (29,30) age, recipient age, delayed graft function, time of rejection, and rejec- using proteinase K (20 ␮g/ml; 30 min at 37°C) for tissue permeabiliza- tion severity, were also assessed and were related to outcome. Mean tion. donor age was 45 Ϯ 12 yr, and mean recipient age was 46 Ϯ 12 yr. A total of 35% of the patients had delayed graft function. Biopsies had Validation of Prognostic Value of mRNA Levels in a been taken 29 Ϯ 35 d after transplantation. A total of 88% of the biopsies Second Group of Rejection Biopsies had been taken during a first rejection episode, and 16% expressed Several molecules of which mRNA levels distinguished the two vascular rejection. patient groups with different outcomes were tested further for their The outcome parameter was the slope of the regression line through 3776 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005 serum creatinine values between 6 mo and 2 yr. The predictive value of corresponding P values. Predicting variables in univariate analysis the different parameters was evaluated with linear regression. For were tested further in a multivariate analysis. Correlation coefficients obtaining binary outcome data, patients who showed a regression line were calculated with Pearson correlation tests. Data are represented as that was higher than 0.04 were regarded as progressors. This cutoff means Ϯ SD. value of 0.04 is based on observations from the two patient groups representing progressors and nonprogressors in the first experiment. Levels were compared between progressors (slope regression line Results Ն0.04) and nonprogressors (slope regression line Ͻ0.04) to evaluate the Clinical Variables and Histology consistency of upregulation of the 4 mRNA transcripts in the rejection Demographic and clinical characteristics, except urine pro- biopsies. An age-matched control group of 15 transplant biopsies with tein levels at 6 mo, were not significantly different for the PR no rejection was also studied. and the NP groups (Table 3). Characteristics of the studied biopsies are summarized in Table 4. The distribution of the rejection severity, the percentage of vascular rejections, and the Statistical Analyses timing of the biopsies were similar for the two groups. One Statistical analyses were performed using SPSS version 10.0. For biopsy in the PR group stained strongly positive for C4d, and nonparametric variables, differences were tested with ␹2 tests. For three biopsies in the NP group were weakly positive for C4d. continuous variables, differences were tested with independent sample The extent of infiltrate and chronic damage in different pa- t test. Linear regression analyses were used to calculate r values and tient groups was compared. Semiquantitative scores for the

Figure 2. Analysis of infiltrate and chronic damage in patient groups. (A) Amount of total infiltrate, eosinophil granulocytes, and plasma cells in patient groups during acute rejection. Number of cells for each variable was scored in silver-stained paraffin sections on a scale from 0 to 3. (B and C) Expression of pan-leukocyte marker CD45 in patient groups at the mRNA level and at the protein level. (D) The extent of staining for the monocyte marker CD68. (E) As a measure of the amount of collagen deposition in the biopsies, the extent of Sirius red staining was quantified by image analysis. For all measurements, the PR group (AR-progr.) did not significantly differ from the NP group (AR-nonprogr.). *P Ͻ 0.05; #P Ͻ 0.01. J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3777

Table 5. Genes for which mRNA expression is at least two-fold different between PR group and NP groupa

Fold PR NP Access no. Gene Function

Genes at least two-fold increased in PR group 6.6 Ϯ 4.7 19 3 Ϯ 1 M90696 Cathepsin S (CTSS) Immune response, macrophages 5.7 Ϯ 0.1 432 79 Ϯ 13 J03553 Surfactant protein C Respiration 4.9 Ϯ 4.3 268 80 Ϯ 74 Y15915 Collagen type I, ␣ 1 ECM related 3.7 Ϯ 0.9 291 66 Ϯ 19 J03464 Collagen type I, ␣ 2 ECM related 3.6 Ϯ 2.9 69 19 Ϯ 18 AB014511 ATPase, class II, type 9A ATP biosynthesis 3.4 Ϯ 0.8 59 18 Ϯ 5 AF000984 DEAD/H box polypeptide, Y RNA biosynthesis chromosome 3.0 Ϯ 0.1 58 19 Ϯ 0.1 J03464 Collagen type I, ␣ 2 ECM related 2.8 Ϯ 1.7 304 139 Ϯ 83 AL050141 Phosphatidic acid phosphatase type 2A Intracellular signaling, cell growth 2.7 Ϯ 0.5 368 113 Ϯ 25 M58459 Ribosomal protein S4 (RPS4), Y-linked Protein biosynthesis 2.1 Ϯ 0.1 66 26 Ϯ 5 AF070648 Caveolin 1, caveolae protein (CAV1) Intracellular signaling, cell cycle 2.1 Ϯ 0.5 146 54 Ϯ 4 AF013570 Myosin, heavy polypeptide 11, smooth Cytoskeleton related muscle 2.1 Ϯ 0.7 47 23 Ϯ 3 W26634 Unknown

Genes at least two-fold decreased in PR group 0.14 Ϯ 0.02 9 58 Ϯ 7 AA128249 Fatty acid binding protein 4, adipocyte Lipid metabolism 0.19 Ϯ 0.06 341 2966 Ϯ 867 L48215 ␤-globin (HBB) Oxygen transport 0.19 Ϯ 0.05 170 1113 Ϯ 456 J00153 ␣-globin (HBA) Oxygen transport 0.19 Ϯ 0.06 283 1478 Ϯ 504 M25079 ␤-globin (HBB) Oxygen transport 0.35 Ϯ 0.03 42 110 Ϯ 4 U80114 Unknown 0.35 Ϯ 0.03 1596 4626 Ϯ 425 X67301 CD79B antigen (Ig-␤) Immune response, B cell 0.36 Ϯ 0.15 28 102 Ϯ 37 AL049250 BANP homolog, SMAR1 homolog 0.37 Ϯ 0.14 53 131 Ϯ 11 AI126134 S100 Calcium-binding protein A8 Immune response, signal transduction (Calgranulin A) 0.37 Ϯ 0.02 12 43 Ϯ 20 AF043586 Unknown 0.38 Ϯ 0.07 86 212 Ϯ 21 Pim-2 protooncogene homolog Growth, differentiation 0.39 Ϯ 0.12 138 250 Ϯ 16 U80114 Unknown 0.41 Ϯ 0.12 39 115 Ϯ 41 W72424 S100 Calcium-binding protein A9 Immune response, signal transduction (Calgranulin B) 0.42 Ϯ 0.02 26 55 Ϯ 8 M64595 Ras-related C3 botulinin toxin substrate 2 Intracellular signaling 0.44 Ϯ 0.09 15 48 Ϯ 26 Y09392 TNF receptor superfamily, member 12 Immune response, cytokine signaling 0.44 Ϯ 0.04 48 108 Ϯ 9 AA846749 Apolipoprotein M Lipid metabolism 0.46 Ϯ 0.16 69 179 Ϯ 99 X53416 Filamin A, ␣ (actin-binding protein 280) Cytoskeleton related 0.46 Ϯ 0.11 241 567 Ϯ 37 AF015128 Unknown 0.47 Ϯ 0.09 3512 8936 Ϯ 2545 X67301 CD79B antigen (Ig-␤) Immune response, B cell 0.48 Ϯ 0.14 34 87 Ϯ 27 M25809 ATPase, H ϩ transport, lysosomal, ATP biosynthesis ␤ polypeptide 0.48 Ϯ 0.25 50 139 Ϯ 64 AJ245434 Apolipoprotein M Lipid metabolism Ϯ Ϯ 0.49 0.07 598 1340 --192 X55954 Ribosomal protein L23 Protein biosynthesis 0.49 Ϯ 0.07 1151 2370 Ϯ 424 Z26876 Ribosomal protein L38 Protein biosynthesis 0.49 Ϯ 0.07 554 906 Ϯ 70 AI147237 Ig heavy constant ␥ 3 (G3m marker) Immune response, Ig related 0.50 Ϯ 0.05 393 950 Ϯ 251 X92997 Unknown

aECM, extracellular matrix. amount of total infiltrate, eosinophil granulocytes, and plasma rejections and biopsies that were taken during second or third cells were identical for the patient groups (Figure 2A). The rejections revealed no significant difference in the mRNA levels groups also showed similar cortical mRNA levels of CD45 and protein stainings presented in the study in either group. (Figure 2B) and a similar extent of staining for CD45 and CD68 Humoral involvement, which was reflected by C4d staining, (Figure 2, C and D). The extent of Sirius red staining did not staining of B cell markers, and panel reactive antibodies, did differ significantly for the two groups (Figure 2E). not differ between first rejection episodes and follow-on rejec- Comparisons between biopsies that were taken during first tion episodes either. 3778 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

Figure 3. Validation of microarray results by real-time PCR. Message for 10 potential candidate genes was quantified in the same biopsies that had been used for microarray analysis and corrected to the mean signal of four household genes (hypoxanthine guanine phosphoribosyltransferase-1 [HPRT-1], porphobilinogen deaminase [PBGD], glyceraldehyde-3-phosphate dehydroge- nase [GAPDH], and TATA box-binding protein [TBP]). As control groups, 6-mo transplant protocol biopsies and native kidneys with no histologic abnormalities were analyzed. *P Ͻ 0.05; $P Ͻ 0.1.

Microarray Analysis Reverse Transcriptase–PCR for Candidate Genes Gene chip analyses were performed to detect differences in gene expression profiles between patient groups. In the PR To validate microarray results, we analyzed 10 genes from group, 65 genes were significantly increased and 77 genes were the gene expression profiling experiment with real-time PCR in significantly decreased in comparison with the NP group. Table the individual patient biopsies (Figure 3). The NP group 5 represents the 36 gene sequences for which mRNA expression showed mRNA levels for S100 calcium binding protein A8 was at least two-fold different between groups. (S100A8), S100A9, ␤-globin, ␣-globin, and CD79b that were J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3779

Figure 4. Pattern of staining for S100A8 and S100A9 in patient biopsies with acute rejection. Staining for the S100A8/A9 heterodimeric complex (antibody 27E10; A and B) and S100A9 (D) was seen in focal infiltrates and in infiltrating cells between tubules. (C) Enlarged recording of tubulointerstitium showing positive signal for S100A8/A9 heterodimers. Arrow indicates a neutrophil granulocyte that is positive in the staining. S100A8/A9-dimeric protein complexes are presumably also deposited extracellularly (arrowheads). (E) Typical result for RNA in situ hybridization for S100A8 mRNA in a biopsy with acute rejection. Yellow arrows indicate positivity for S100A8 mRNA. higher than those of the PR group by a factor 6.8 (P Ͻ 0.05), 1.9 Renal cortical steady-state levels of TGF-␤ differed significantly (P Ͻ 0.1), 8.4 (P Ͻ 0.05), 3.9 (P Ͻ 0.1), and 2.2 (P Ͻ 0.1), between patient groups (NP 1.8 Ϯ 1.6; PR 1.0 Ϯ 0.8; P Ͻ 0.05; respectively. In contrast, the PR group showed SP-C message data not shown), which is in accordance with findings from an that was higher than the NP group by a factor 4.0 (P Ͻ 0.05). earlier study (2). The NP group showed 2.6-fold higher IL-10 mRNA levels than the PR group, but this difference was NS Expression of TGF-␤ and IL-10 (data not shown). In all rejection biopsies, correlations were The expression of genes that were identified in the experi- found for TGF-␤ mRNA with S100A8 mRNA and S100A8 stain- ments described above may be regulated by TGF-␤ and IL-10. ing (r ϭ 0.40, P Ͻ 0.05), S100A9 mRNA (r ϭ 0.54, P Ͻ 0.005), 3780 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

Figure 5. The extent of deposition of S100A8 and S100A9 protein in patient groups. Antibodies that detect S100A8 (A), S100A9 (B), or the S100A8/A9 heterodimeric complex (C) were applied on the acute rejection biopsies. The NP group showed significantly more S100A9 and 27E10 staining than the PR group. The extent of S100A9 protein in follow-up biopsies with chronic allograft nephropathy (CAN) in the PR group was lower than that in the acute rejection biopsies. (D and E) mRNA expression of S100A8 and S100A9 in biopsies with acute rejection. S100A8 mRNA levels were significantly higher in the NP group compared with the PR group. *P Ͻ 0.05; #P Ͻ 0.01. and S100A9 staining (r ϭ 0.43, P Ͻ 0.05). IL-10 mRNA levels tial infiltrate fields (Figure 4A) and in infiltrating cells between showed significant correlations with S100A9 mRNA (r ϭ 0.38) tubules (Figure 4, B through D). Glomeruli occasionally showed and with S100A8/S100A9 staining (r ϭ 0.46; P Ͻ 0.05). These positive cells (Figure 4A). Neutrophil granulocytes, which results may suggest that TGF-␤ and IL-10 are involved in could be recognized through their multigranular appearance, regulation of S100A8 and S100A9 expression. stained positive in the 27E10 stainings (Figure 4C, arrow). The Investigation of S100A8 and S100A9 heterodimeric S100A8/A9 protein is probably also deposited Protein deposition of S100A8 was examined together with extracellularly (Figure 4C, arrowheads). S100A8 mRNA tran- that of S100A9, because the two form a heterodimer. Staining script was detected in the renal cortex with RISH in a similar for the S100A8/S100A9 heterodimer (mAb 27E10) and for pattern as that seen with immunohistochemistry. Occasional S100A8 and S100A9 monomers was observed in tubulointersti- regions in the tubulointerstitium were positive, and glomeruli

Figure 6. Results of immunohistochemical staining for B cell markers CD79 and CD20. The extent of CD79a staining (A) and CD20 staining (B) was quantified in biopsies with acute rejection by computerized image analysis. During acute rejection, no differences in the extent of staining were found between the PR group and the NP group. Follow-up biopsies with CAN in the PR group showed a significantly higher proportion of CD79a-positive B cells than the acute rejection groups. The percentage of CD20-positive B cells in the CAN biopsies was significantly higher than that in acute rejection samples from NP group. *P Ͻ 0.05; #P Ͻ 0.01. J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3781

Table 6. Results of B cell stainings in individual analysis. We performed stainings for CD79a, which forms an samples Ig-associating complex with CD79b on the surface of B cells. Absolute amounts of CD79a did not differ between patient a b Image Analysis Cell Cluster Score groups (Figure 6A; Table 6). In addition, staining for the B cell Sample CD79 CD20 CD79 CD20 marker CD20 was examined. Absolute amounts of CD20 did not significantly differ between groups (Figure 6B; Table 6). The PR group extent of staining of both CD79a and CD20 was higher in ϩϩ 3 1.47 0.99 follow-up biopsies with CAN (Figure 6, A and B). In the acute ϪϪ 4a 0.20 0.04 rejection biopsies, we additionally counted the number of pos- ϩϩ 4b 0.55 0.37 itively stained cells per hpf, while regarding biopsies with a cell ϪϪ 5 0.26 0.60 count of Ͼ275 as positive. The percentage of biopsies that were ϪϪ 7 0.80 0.28 positive for CD79a in the PR group (37.5%) was not signifi- ϪϪ 8 0.47 0.39 cantly different from that in the NP group (35.3%; Table 6). ϩϩ 9 1.31 0.76 These percentages were the same for CD20 (Table 6). In con- ϪϪ 10 0.17 0.02 clusion, the presence of CD79-positive and CD20-positive B Ϯ Ϯ 0.65 0.50 0.43 0.34 38% 38% cells, scored either as overall core staining percentages or as NP group number of cells per hpf, is not associated with an adverse ϪϪ 1 0.34 0.31 outcome. 2 0.37 0.40 ϪϪ 3 0.16 0.21 ϪϪ 4 0.22 0.17 ϪϪInvestigation of SP-C 5 0.30 0.53 ϩϩ Four rejection biopsies that showed the highest SP-C mRNA 7 1.01 1.20 ϩϩlevels through real-time PCR did not show significant signals 8 0.91 0.60 ϩϩabove background by RISH (data not shown), whereas SP-C 9a 0.52 0.50 ϪϪmRNA was detected in peripheral lung tissue. Antibodies 9b 0.69 0.29 ϪϪagainst the proform of SP-C rendered positive staining in the 11a 2.19 0.80 ϩϩglomeruli of rejection biopsies (Figure 7, B and C). Staining 11b 0.37 0.29 ϪϪcould not be detected in control tissue that had been obtained 12 0.74 0.75 ϩϩfrom native kidneys (Figure 8B). Immunohistochemistry for 13 0.79 0.39 ϪϪmature SP-C did not produce positive signals in any biopsies 15 0.35 0.63 ϪϪtested. To determine which glomerular compartment expressed 16 0.26 0.22 ϪϪthe protein, we performed staining for proSP-C (Figure 7C) on 17 0.44 0.29 ϪϪsequential sections together with stainings for the endothelial 18 0.95 0.68 ϩϩmarker CD31 (Figure 7D) and the mesenchymal markers vi- 0.62 Ϯ 0.49 0.49 Ϯ 0.27 35% 35% mentin (Figure 7E) and desmin (Figure 7F). The proSP-C staining pattern primarily resembled that of desmin. The difference a Core staining density was scored with computerized in the number of glomeruli that were positive for proSP-C image analysis as percentage of staining per total renal between acute rejection groups was 1.1-fold (NS; Figure 8B). cortical surface assessed. bClusters of B cells were scored under light microscopy, The results show that SP-C mRNA levels during acute rejec- depicting for each specimen the number of positive cells in tion are associated with adverse outcome. Glomerular SP-C the single high-power field (hpf) with the highest positive protein, which is present probably in the mesangium, was Ͼ cell count identified. Specimens with a cell count 275/hpf detected in acute rejection biopsies and CAN biopsies but not in were considered positive. normal renal tissue. were negative in the RISH (Figure 4E). Expression patterns were comparable between patient groups. Validation of Prognostic Value of Candidate Genes in a Quantitative image analysis of stained sections showed that Second Group of Rejection Biopsies the extent of staining for 27E10 and the S100A9 monomer was A number of accepted risk factors and several molecules of 3.2- and 1.8-fold (P Ͻ 0.05) higher, respectively, in the NP which the mRNA level distinguished the two patient groups group than in the PR group (Figure 5, B and C). The extent of with different outcomes were tested for their prognostic value S100A8 staining was not statistically different between groups in an independent set of 43 rejection biopsies. In univariate (Figure 5A). In the PR group, follow-up biopsies with CAN linear regression, time of rejection, SP-C mRNA, and ␤-globin contained less S100A9 protein than the acute rejection samples mRNA were significantly associated with the course of renal (Figure 5B). function beyond 6 mo (Table 7). In multivariate analysis, SP-C mRNA predicted outcome (Table 7). Figure 9 depicts the data Investigation of B Cell Markers with the outcome parameter in a binary manner. Results are Staining for CD79b rendered high background signals in the shown for the four mRNA transcripts measured in the patients tubules, which infringed on the reliability of digital image with acute rejection, who had been divided into progressors 3782 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

Figure 7. Localization of pro–surfactant protein-C (proSP-C) protein in acute rejection biopsies. (A) A typical beads-on-a-string staining pattern for proSP-C was observed in peripheral lung sections, which were used as positive controls. (B and C) ProSP-C protein was detected in glomeruli of sections from renal transplants with acute rejection. Sequential sections were stained for proSP-C protein (C), the endothelial marker CD31 (D), and the mesenchymal markers vimentin and desmin (E and F). The proSP-C staining pattern resembled the most that of desmin.

Figure 8. Extent of protein deposition of proSP-C in patient groups. (A) mRNA expression of SP-C was quantified by real-time PCR in biopsies with acute rejection. SP-C mRNA levels in the PR group were significantly higher than those in the NP group. (B) Immunohistochemistry was performed using antibodies that detect proSP-C protein. The percentage of surface stained by the antibody of the total glomerular capillary tuft area in each glomerulus was quantified by image analysis. All patient groups showed significantly more SP-C message and proSP-C protein than control subjects. *P Ͻ 0.05; #P Ͻ 0.01. and nonprogressors. SP-C mRNA levels were significantly dif- indicator of progression to CAN. By gene expression profiling, ferent for progressors and nonprogressors. we identified gene sequences of which the mRNA levels during acute rejection were significantly different between two patient Discussion groups with different outcome. One patient group retained Changes in gene expression levels underlying differences in stable graft function over a prolonged period after transplan- composition of the inflammatory infiltrate and regeneration tation, whereas the patients from the other group lost their capacity of the graft tissue upon acute rejection may be an grafts as a result of CAN. Morphologic parameters at the time J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3783

Table 7. Evaluation of predictive value of molecular, clinical, and morphologic parameters in second set of rejection biopsiesa

Univariate Correlation Coefficient P

Donor age 0.04 Ͼ0.05 Recipient age Ϫ0.08 Ͼ0.05 Delayed graft function 0.06 Ͼ0.05 Rejection severity 0.17 Ͼ0.05 Time of rejection 0.41 0.006 Surfactant protein-C mRNA 0.42 0.006 S100A8 mRNA 0.23 Ͼ0.05 ␤-globin mRNA 0.40 0.009 CD79b mRNA Ϫ0.19 Ͼ0.05

Multivariate Unstandardized Coefficient P Time of rejection 0.0008 (CI, 0.000 to 0.002) 0.110 Surfactant protein-C 0.073 (CI, 0.021 to 0.126) 0.007 ␤-globin 0.024 (CI, Ϫ0.027 to 0.076) 0.343

aCI, 95% confidence interval. of acute rejection could not discriminate between the patient these factors. We hypothesize that a disturbance in S100A9 groups. Real-time PCR confirmed that mRNA expression for induction or S100A8/A9-dimer formation may be one mecha- several molecules during acute rejection was associated with nism accounting for fibrogenesis and progression to CAN. This graft outcome. Discriminating molecular factors were studied concept would need further testing in a model for renal fibrosis further with immunohistochemistry and RISH. Messenger with S100A9 knockout mice (32). RNA transcripts were tested for their prognostic value in a CD79b antigen, a protein that is crucial for functional activity second group of biopsies with rejection. of the antigen receptor complex on B lymphocytes (40), ap- Results from a study in renal allografts with acute rejection peared several times in the microarray data list. Furthermore, indicate that S100A8 and S100A9 are produced mainly by in a recent microarray study of rejection biopsies, Sarwal et al. CD68-positive monocytes and that the S100A8/A9 heterodimer (4) found an association between expression of the B cell is the predominant form in which these proteins appear (31). marker CD20 and late graft loss. These observations led us to The heterodimer is virtually absent during CAN (31). We found analyze further B cell markers in the patient cohort. A signifi- that the expression of S100A8 and S100A9 was significantly cant difference in CD79b mRNA levels between patient groups higher during rejection in the NP group than during rejection in could not be confirmed with real-time PCR. Moreover, a sig- the PR group. The extent of CD68-positive cell infiltrate, how- nificant difference between patient groups for CD79 and CD20 ever, was similar between the two groups. Overall, the results in either overall core staining or number of positive cells per suggest that the groups differ in the extent of S100A8/A9- hpf was not found. With respect to association between B cell dimer formation and excretion from monocytes. staining and outcome, the discrepancy between findings from S100A8 and S100A9 have proinflammatory effects by pro- our study and those from Sarwal’s study may be explained by moting transendothelial migration of phagocytes (32,33). In- different inclusion criteria of patients and rejection biopsies. deed, S100A9Ϫ/Ϫ mice show a diminished recruitment of Message for SP-C was significantly increased during acute granulocytes into excisional wounds of the skin (33). In contrast, S100A8 and S100A9 play a role in wound repair (34) and rejection in the PR group in comparison with the NP group. can have antiproliferative effects on macrophages and lympho- SP-C provides alveolar stability in the lungs, which is necessary cytes, probably in cooperation with TGF-␤ and IL-10 (35,36). for normal breathing (41). We are the first to show SP-C mRNA The S100A8/A9 heterodimeric complex has the capacity to and protein expression in human renal allografts. The question exert growth-inhibitory effects on fibroblasts (37). Treatment of remains what the function of SP-C is in the kidney. SP-C is able IL-10 enhances expression and secretion of S100A8 by cultured to give rise to amyloid fibril formation by conversion of the ␣ ␤ macrophages (38) and raises S100A8 and S100A9 levels in den- protein from an -helical state to a -sheet fibril (42). Such dritic cells (39). The functional relationship of S100A8 and structural conversions may lead to formation of insoluble SP-C S100A9 with TGF-␤ and IL-10 is supported by the correlations aggregates, a phenomenon that is possibly associated with that we observed between expression of the S100 molecules and pulmonary disease (43). Further studies are necessary to dis- that of TGF-␤ and IL-10. The finding that S100A8 and S100A9 cover whether similar mechanisms are involved in renal trans- protein expression is increased during acute rejection in the NP plants. SP-C was ectopically expressed in ureteric bud tissue group may reflect the predominance of the beneficial actions of (44,45), reflecting cell differentiation processes. In parallel, SP-C 3784 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

Figure 9. Consistency of upregulation of four molecular parameters in a second set of rejection biopsies. Messenger RNA levels of SP-C, S100A8, ␤-globin, and CD79b were measured with real-time PCR in 43 biopsies with acute rejection and, for reasons of comparison, in 15 transplant biopsies with no rejection. The group of 43 patients with acute rejection was divided into progressors (slope regression line through renal function values between 6 mo and 2 yr Ն0.04) and nonprogressors (slope Ͻ0.04). SP-C message was present in 47% of the patients and absent in the remaining 53% and therefore has been depicted as a categorical variable. *P Ͻ 0.05; #P Ͻ 0.01.

expression during acute rejection may represent tissue remod- demonstrated expression of SP-C in human renal allografts eling in renal allografts of adult kidneys. with rejection, which is associated with worse prognosis. We have identified several genes for which mRNA and protein expression during acute rejection differed between two References patient groups with different outcome, whereas morphologic 1. Kirk AD, Jacobson LM, Heisey DM, Radke NF, Pirsch JD, parameters did not distinguish between these groups. The pre- Sollinger HW: Clinically stable human renal allografts con- dictive value of mRNA expression levels was validated in a tain histological and RNA-based findings that correlate second population of rejection biopsies. The results support the with deteriorating graft function. Transplantation 68: 1578– hypothesis that expression levels may function as additive 1582, 1999 markers for outcome in kidney transplantation. Molecular fac- 2. Eikmans M, Sijpkens YW, Baelde HJ, De Heer E, Paul LC, tors that are present during acute rejection, which are associ- Bruijn JA: High transforming growth factor-beta and ex- ated with the presence or absence of CAN occurring later, may tracellular matrix mRNA response in renal allografts dur- have promoting or inhibitory effects on the pathophysiologic ing early acute rejection is associated with absence of chronic rejection. Transplantation 73: 573–579, 2002 process of CAN. Relatively high levels of S100A8 and S100A9 in 3. Baboolal K, Jones GA, Janezic A, Griffiths DR, Jurewicz infiltrating cells during acute rejection, possibly affected by ␤ WA: Molecular and structural consequences of early renal TGF- and IL-10, are associated with a favorable outcome. allograft injury. Kidney Int 61: 686–696, 2002 S100A8 and S100A9 may enhance tissue repair mechanisms or 4. Sarwal M, Chua MS, Kambham N, Hsieh SC, Satterwhite inhibit fibroblast proliferation. Evidence for association be- T, Masek M, Salvatierra O Jr: Molecular heterogeneity in tween expression of B cell markers during acute rejection and acute renal allograft rejection identified by DNA microar- outcome was not found in this study. For the first time, we ray profiling. N Engl J Med 349: 125–138, 2003 J Am Soc Nephrol 16: 3771–3786, 2005 Prognostic Factors in Transplantation 3785

5. Akalin E, Hendrix RC, Polavarapu RG, Pearson TC, Ney- 21. Racusen LC, Solez K, Colvin RB, Bonsib SM, Castro MC, lan JF, Larsen CP, Lakkis FG: Gene expression analysis in Cavallo T, Croker BP, Demetris AJ, Drachenberg CB, Fogo human renal allograft biopsy samples using high-density AB, Furness P, Gaber LW, Gibson IW, Glotz D, Goldberg oligoarray technology. Transplantation 72: 948–953, 2001 JC, Grande J, Halloran PF, Hansen HE, Hartley B, Hayry 6. Kretzler M, Cohen CD, Doran P, Henger A, Madden S, PJ, Hill CM, Hoffman EO, Hunsicker LG, Lindblad AS, Grone EF, Nelson PJ, Schlondorff D, Grone HJ: Repunctur- Yamaguchi Y: The Banff 97 working classification of renal ing the renal biopsy: Strategies for molecular diagnosis in allograft pathology. Kidney Int 55: 713–723, 1999 nephrology. J Am Soc Nephrol 13: 1961–1972, 2002 22. Eikmans M, Baelde HJ, De Heer E, Bruijn JA: Processing 7. Eikmans M, Baelde HJ, De Heer E, Bruijn JA: RNA expres- renal biopsies for diagnostic mRNA quantification: Im- sion profiling as prognostic tool in renal patients: Toward provement of RNA extraction and storage conditions. JAm nephrogenomics. Kidney Int 62: 1125–1135, 2002 Soc Nephrol 11: 868–873, 2000 8. Chua MS, Barry C, Chen X, Salvatierra O, Sarwal MM: 23. Baelde JJ, Eikmans M, Doran PP, Lappin DWP, De Heer E, Molecular profiling of anemia in acute renal allograft re- Bruijn JA: Gene expression profiling in glomeruli from jection using DNA microarrays. Am J Transplant 3: 17–22, human kidneys with diabetic nephropathy. Am J Kidney 2003 Dis 43: 636–650, 2004 9. Scherer A, Krause A, Walker JR, Korn A, Niese D, Raulf F: 24. Koop K, Eikmans M, Baelde HJ, Kawachi H, De Heer E, Early prognosis of the development of renal chronic allo- Paul LC, Bruijn JA: Expression of podocyte-associated mol- graft rejection by gene expression profiling of human pro- ecules in acquired human kidney diseases. J Am Soc Neph- tocol biopsies. Transplantation 75: 1323–1330, 2003 rol 14: 2063–2071, 2003 10. Mansfield ES, Sarwal MM: Arraying the orchestration of 25. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy allograft pathology. Am J Transplant 4: 853–862, 2004 N, De Paepe A, Speleman F: Accurate normalization of real- 11. Hayry P, Isoniemi H, Yilmaz S, Mennander A, Lemstrom time quantitative RT-PCR data by geometric averaging of K, Raisanen-Sokolowski A, Koskinen P, Ustinov J, Lauten- multiple control genes. Genome Biol 3: 0034.1–0034.12, 2003 schlager I, Taskinen E: Chronic allograft rejection. Immunol 26. Frosch M, Vogl T, Waldherr R, Sorg C, Sunderko¨tter C, Rev 134: 33–81, 1993 Roth J: Expression of MRP8 and MRP14 by macrophages is 12. Hariharan S, Johnson CP, Bresnahan BA, Taranto SE, a marker for severe forms of glomerulonephritis. J Leukoc McIntosh MJ, Stablein D: Improved graft survival after Biol 75: 198–206, 2004 renal transplantation in the United States, 1988 to 1996. 27. Bakker RC, Koop K, Sijpkens YW, Eikmans M, Bajema IM, N Engl J Med 342: 605–612, 2000 De Heer E, Bruijn JA, Paul LC: Early interstitial accumu- 13. Massy ZA, Guijarro C, Wiederkehr MR, Ma JZ, Kasiske BL: lation of collagen type I discriminates chronic rejection Chronic renal allograft rejection: Immunologic and nonim- from chronic cyclosporine nephrotoxicity. J Am Soc Nephrol munologic risk factors. Kidney Int 49: 518–524, 1996 14: 2142–2149, 2003 14. Sijpkens YW, Doxiadis II, De Fijter JW, Mallat MJ, Van Es 28. Glasser SW, Korfhagen TR, Perme CM, Pilot-Matias TJ, LA, De Lange P, Zwinderman AH, Westendorp RG, van Kister SE, Whitsett JA: Two SP-C genes encoding human Kemenade FJ, Bruijn JA, Claas FH, Paul LC: Sharing cross- pulmonary surfactant proteolipid. J Biol Chem 263: 10326– reactive groups of MHC class I improves long-term graft 10331, 1988 survival. Kidney Int 56: 1920–1927, 1999 29. Ehrchen J, Heuer H, Sigmund R, Schafer MKH, Bauer K: 15. van Saase JL, Van der Woude FJ, Thorogood J, Hollander Expression and regulation of osteopontin and connective AA, Van Es LA, Weening JJ, van Bockel JH, Bruijn JA: The tissue growth factor transcripts in rat anterior pituitary. J relation between acute vascular and interstitial renal allo- Endocrinol 169: 87–96, 2001 graft rejection and subsequent chronic rejection. Transplan- 30. de Boer WI, Sont JK, van Schadewijk A, Stolk J, van tation 59: 1280–1285, 1995 Krieken JH, Hiemstra PS: Monocyte chemoattractant pro- 16. Bonsib SM, Abul-Ezz SR, Ahmad I, Young SM, Ellis EN, tein 1, interleukin 8, and chronic airways inflammation in Schneider DL, Walker PD: Acute rejection-associated tubu- COPD. J Pathol 190: 619–626, 2000 lar basement membrane defects and chronic allograft ne- 31. Goebeler M, Roth J, Burwinkel F, Vollmer E, Bocker W, phropathy. Kidney Int 58: 2206–2214, 2000 Sorg C: Expression and complex formation of S100-like 17. Sijpkens YWJ, Zwinderman AH, Mallat MJK, Boom H, De proteins MRP8 and MRP14 by macrophages during renal Fijter JW, Paul LC: Intercept and slope analysis of risk allograft rejection. Transplantation 58: 355–361, 1994 factors in chronic allograft nephropathy. Graft 5: 109–114, 32. Manitz MP, Horst B, Seeliger S, Strey A, Skryabin BV, 2002 Gunzer M, Frings W, Schonlau F, Roth J, Sorg C, Nacken 18. Sijpkens YW, Bruijn JA, Paul LC: Chronic allograft ne- W: Loss of S100A9 (MRP14) results in reduced interleukin- phropathy categorised in chronic interstitial and vascular 8-induced CD11b surface expression, a polarized microfila- rejection. Transplant Proc 33: 1153, 2001 ment system, and diminished responsiveness to chemoat- 19. Masseroli M, O’Valle F, Andujar M, Ramirez C, Gomez- tractants in vitro. Mol Cell Biol 23: 1034–1043, 2003 Morales M, de Dios LJ, Aguilar M, Aguilar D, Rodriguez- 33. Vogl T, Ludwig S, Goebeler M, Thorey I, Strey A, Reichelt Puyol M, Del Moral RG: Design and validation of a new R, Foell D, Gerke V, Nacken W, Werner S, Sorg C, Roth J: image analysis method for automatic quantification of in- The S100 proteins MRP8 (S100A8) and MRP14 (S100A9) terstitial fibrosis and glomerular morphometry. Lab Invest control microtubule reorganization during transendothe- 78: 511–522, 1998 lial migration of phagocytes. Blood 104: 4260–4268, 2004 20. Eikmans M, Baelde HJ, Hagen EC, Paul LC, Eilers PH, De 34. Thorey IS, Roth J, Regenbogen J, Halle JP, Bittner M, Vogl Heer E, Bruijn JA: Renal mRNA levels as prognostic tools T, Kaesler S, Bugnon P, Reitmaier B, Durka S, Graf A, in kidney diseases. J Am Soc Nephrol 14: 899–907, 2003 Wockner M, Rieger N, Konstantinow A, Wolf E, Goppelt 3786 Journal of the American Society of Nephrology J Am Soc Nephrol 16: 3771–3786, 2005

A, Werner S: The Ca2ϩ-binding proteins S100A8 and Rawlings DJ: Aberrant B cell receptor signaling from B29 S100A9 are encoded by novel injury-regulated genes. J Biol (Igbeta, CD79b) gene mutations of chronic lymphocytic Chem 276: 35818–35825, 2001 leukemia B cells. Proc Natl Acad Sci U S A 97: 5504–5509, 35. Aguiar-Passeti T, Postol E, Sorg C, Mariano M: Epithelioid 2000 cells from foreign-body granuloma selectively express the 41. Whitsett JA, Weaver TE: Hydrophobic surfactant proteins calcium-binding protein MRP-14, a novel down-regulatory in lung function and disease. N Engl J Med 347: 2141–2148, molecule of macrophage activation. J Leukoc Biol 62: 852– 2002 858, 1997 42. Johansson J: Molecular determinants for amyloid fibril for- 36. Yui S, Nakatani Y, Mikami M: Calprotectin (S100A8/ mation: Lessons from lung surfactant protein C. Swiss Med S100A9), an inflammatory protein complex from neutro- Wkly 133: 275–282, 2003 phils with a broad apoptosis-inducing activity. Biol Pharm 43. Gustafsson M, Thyberg J, Naslund J, Eliasson E, Johansson Bull 26: 753–760, 2003 J: Amyloid fibril formation by pulmonary surfactant pro- 37. Yui S, Mikami M, Tsurumaki K, Yamazaki M: Growth- tein C. FEBS Lett 464: 138–142, 1999 inhibitory and apoptosis-inducing activities of calprotectin 44. Lin YF, Zhang SB, Rehn M, Itaranta P, Tuukkanen J, Hel- derived from inflammatory exudate cells on normal fibro- jasvaara R, Peltoketo H, Pihlajaniemi T, Vainio S: Induced blasts: Regulation by metal ions. J Leukoc Biol 61: 50–57, repatterning of type XVIII collagen expression in ureter 1997 bud from kidney to lung type: Association with sonic 38. Xu K, Yen T, Geczy CL: IL-10 up-regulates macrophage hedgehog and ectopic surfactant protein C. Development expression of the S100 protein S100A8. J Immunol 166: 128: 1573–1585, 2001 6358–6366, 2001 45. Vainio S, Lin YF, Pihlajaniemi T: Induced repatterning of 39. Kumar A, Steinkasserer A, Berchtold S: Interleukin-10 in- type XVIII collagen associates with ectopic sonic hedgehog fluences the expression of MRP8 and MRP14 on human and lung surfactant C gene expression and changes in dendritic cells. Int Arch Allergy Immunol 132: 40–47, 2003 epithelial epigenesis in the ureteric bud. J Am Soc Nephrol 40. Gordon MS, Kato RM, Lansigan F, Thompson AA, Wall R, 14[Suppl 1]: S3–S8, 2003