SEARCH

SEARCH BY CITATION

Keywords:

  • Bioinformatics;
  • cardiac transplantation;
  • gene expression;
  • immune response;
  • microarray

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References

High-throughput microarrays promise a comprehensive analysis of complex biological processes, yet their applicability is hampered by problems of reproducibility and data management. The current study examines some of the major questions of microarray use in a well-described model of allograft rejection. Using the Brown Norway to Lewis heterotopic heart transplant model, highly purified RNA was isolated from cardiac tissue at postoperative days (POD) 3, 5 and 7 and hybridized onto Affymetrix U34A microarrays. Using the log average ratio (LAR), changes in gene expression were monitored at each timepoint and p-values generated through statistical analysis. Microarray data were verified for 13 significant transcripts using RT-PCR. Of the 8800 transcripts studied, 2864 were increased on POD 3, 1418 on POD 5 and 2745 on POD 7. Verifying previous studies, many up-regulated genes appeared to be associated with the inflammatory process and graft infiltrating cells. Down-regulated transcripts included many novel molecules such as SC1 and decorin. LAR analysis provides a useful approach to analyze microarray data. Results were reproducible and correlated well with both RT-PCR and prior studies. Most importantly, these results provide new insights into the pathogenesis of acute rejection and suggest new molecules for future studies.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References

Despite improvements in immunosuppression, acute allograft rejection still occurs in 20–40% of organ allograft recipients (1). In addition to increasing patient cost and morbidity, acute rejection has been associated with the development of chronic allograft rejection, the major cause of late graft loss in heart, lung and kidney transplant recipients (2–4).

Previous studies have demonstrated that acute allograft rejection is a progressive inflammatory process involving the ischemia/reperfusion injury, presentation of antigen in regional lymph nodes, infiltration of the graft by a variety of cell types and terminal graft damage (5). The rodent heterotopic cardiac allograft model has been instrumental in defining critical events of allograft rejection (6–9). Despite intensive research, the molecular mechanisms and cellular pathways responsible for organ allograft rejection still have been only partially defined.

High-throughput microarrays allow for the simultaneous assessment of mRNA from thousands of transcripts and promise a comprehensive analysis of the pathogenesis of complex biological processes such as acute rejection (10–14). However, the study of thousands of transcripts at one time raises concerns regarding the accuracy of the data when each transcript is not verified by other methods of RNA analysis such as RT-PCR or Northern blot. To address this issue we investigated several critical variables specific to the use of microarray technology. First, we investigated the most appropriate means of analyzing the raw expression data to determine the most useful algorithm while allowing for the application of thorough statistical analyses. Next, we confirmed several significant changes with other methods (RT-PCR, etc.). Finally, we addressed other areas of concern including the run-to-run variability of the technique and the validity of the results from somewhat heterogeneous mRNA. The current studies address these fundamental questions of microarray technology in the investigation of the pathophysiology of acute cellular rejection in a standard rat heterotopic heart transplant model.

Materials and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References

Animals and procedures

Rats were purchased from Harlan Sprague-Dawley (Indianapolis, IN) and maintained in accordance with guidelines from the American Association for Laboratory Animal Care and the Institutional Animal Care and Use Committee of the Mayo Foundation. Using the standard heterotopic abdominal technique (15), donor hearts from male Brown Norway rats were transplanted into either the MHC incompatible male Lewis rats (RT1l, allografts) or inbred male Brown Norway rats (RT1n, isografts). No immunosuppression was used. Transplanted hearts were examined by daily palpation and rejection was determined by cessation of heartbeats.

Cardiac isografts and allografts were explanted on postoperative day (POD) 3, 5 and 7 (n = 5 at each timepoint). Five test allografts rejected on POD 7.6 ± 0.5. Five native Brown Norway hearts served as normal controls. Tissue was either fixed in 10% neutral buffered formalin and embedded in paraffin for histopathologic examination or divided into 5 mm pieces, placed in RNAlater® (Ambion Inc., Austin, TX) and stored in − 80 °C for RNA extraction and GeneChip® analyses.

RNA isolation and microarray hybridization

Total RNA was extracted with TRIzol® Reagent (Invitrogen Corp, Carlsbad, CA) and further purified using the RNeasy Mini Kit® (Qiagen Inc., Valencia, CA). Sample quality was assessed with an Agilent 2100 Bioanalyzer® (Agilent Technologies, Palo Alto, CA). All samples possessed 18S and 28S rRNA peaks with no signs of RNA degradation. The minimum total RNA quantity used for labeling was 8.0 μg. RNA from 5 native, 15 isograft (n = 5 at POD 3, 5 and 7) and 15 allograft (n = 5 at POD 3, 5 and 7) samples were processed for microarray hybridization.

Using protocols described in the Affymetrix GeneChip® Expression Analysis Manual (Affymetrix, Inc., Santa Clara, CA) double-stranded cDNA was synthesized from each total RNA sample via oligo-dT-mediated reverse transcription. Double-stranded cDNA was cleaned using phase lock gel/phenol-chloroform precipitation. Biotin-labeled cRNA was synthesized using the Enzo BioArray High Yield Kit® (Enzo Diagnostics, Inc., Farmingdale, NY), further purified using Qiagen RNeasy Kit® (Qiagen, Inc), and then quantified by spectrophotometer. The purified cRNA was fragmented into 100–150 nt fragments. Fifteen micrograms of fragmented, biotin-labeled cRNA was hybridized to a U34A (Affymetrix, Inc.) microarray. Each array was washed with buffers of varying stringencies, stained with streptavidin/phycoerythrin and scanned with the Hewlett-Packard GeneArray® Scanner (Hewlett-Packard, Santa Clara, CA). The GeneChip® Microarray Suite v4.01 (Affymetrix, Inc) was used to generate the subsequent data used for statistical evaluation.

Reproducibility of the microarray data

Using this protocol we performed a blinded reproducibility study with two of the POD 5 allograft samples. Total RNA from each POD 5 sample was aliquoted into four separate reactions and prepared as illustrated in Table 1. All eight total RNA aliquots were subjected to separate fragmentation, labeling and hybridization to Affymetrix U34A GeneChips®. The RNA aliquots in Assay II were processed for hybridization approximately 3 weeks after the Assay I samples. Variance components analyses (16) were performed using the log average ratio (LAR) values for the entire U34A dataset of each sample. Based upon the small number of samples, these analyses showed the hybridization measures across probe sets to be measurably different for the separate samples, although it was found that the intra-assay components (chip, sample processing, etc.) contributed about half or more of the total variability to the system.

Table 1. Design of reproducibility
 Allograft POD5–1Allograft POD5–2
Assay I:A (8 μg)A (8 μg)
B (8 μg)B (8 μg)
Assay II:C (8 μg)C (8 μg)
D (8 μg)D (8 μg)

Statistical analysis of microarray data

The basic design of the GeneChip® is a comparison of perfect match (PM) oligonucleotides (25 mer sequences believed to represent portions of the actual genomic DNA sequence) to that of mismatch (MM) oligonucleotides containing a single base mismatch at the center position. Each microarray contains a probe set of approximately 20 PM/MM pairs, for a particular transcript. The manufacturer has selected the probe pairs to maximize sensitivity and specificity when considered as a set. A probe pair with PM minus MM values greater than three standard deviations from the mean of the probe set are considered outliers and are eliminated from the analysis. After taking into account background and variability, the values from the remaining probe pairs are reduced to a single numerical value by the GeneChip® Microarray Suite. This value, or measure of expression, can be represented as either the average difference (AD) or the LAR.

Rationale for statistical approach

Since the statistical approach to microarray data is one of the major thrusts of the paper, a discussion of our rationale is in order. The AD is commonly used as the measure of transcript expression and is calculated by the mean fluorescence of the PM – MM for a probe set. By using the AD values from two microarrays, the GeneChip® Microarray Suite can calculate a fold-change (FC) or semi-quantitative measure of the change in expression for each transcript. Transcripts usually are compared by FC differences with a 3-fold-change increase or decrease considered significant. While this approach simplifies data analysis, it has several limitations. We believe that this approach is likely to underestimate the number of genes altered. For example, a transcript is only deemed significantly altered when it shows a 3-fold increase (or decrease) in every pairwise comparison performed. Given the combined variability of biological systems, RNA preparations and microarray technology, it is likely that the number of transcripts deemed altered would decrease to near zero as the sample size (number of pairwise comparisons) increases, thus discarding many significant alterations in gene expression. In addition, because of a discontinuity between −1 and 1 and the high influence of outliers, FC is not well suited to analysis with standard statistical methodology. Our approach has been to use the second measure of expression, the LAR, to develop a hybridization index as described below. This measure of relative fluorescence allows for the use of more conventional statistical analysis and generation of p-values.

Detailed statistical methods

The LAR is also calculated by the GeneChip® Microarray Suite and as illustrated in the Affymetrix Expression Analysis Manual is based upon the ratio of PM/MM, specifically

  • LAR = 10 × (( log10 PM/MM)/# pairs tested)

The LAR, based upon the log transformation, by its very nature, transforms multiplicative effects to additive effects. Thus the LAR also changes multiplicative errors, which are commonly observed in bioassays, to additive errors, rendering data amenable to standard statistical analyses. Importantly, LAR is not intended to be an absolute quantitative measure of the expression level for a transcript. Instead it is a relative measure used when comparing two groups.

The average expression of a probe set for a group is represented as the average of the LAR ± standard error, hereafter termed the Hybridization Index (HI). The HI for a transcript can be compared to the HI for the same transcript in another group using the two-sided Student's t-test for the difference of means where p < 0.05 is considered statistically significant. We have termed this the change in HI (ΔHI) such that: ΔHI = HIallografts − HIisografts. For each POD the p-value and ΔHI were calculated for each transcript. A ΔHI of 1.0 or more suggests that there is about a ∼ 26% difference or more in the mean PM/MM hybridization of allografts vs. isografts (this measure does not account for the standard errors of the groups). A major advantage of this method is its justification based upon fundamental statistical principles.

There are two situations where we have excluded significant ΔHI data. First, because a HI less than − 1 indicates that the MM probe sets had a much higher fluorescence than the PM, we omitted transcripts from this group. Second, since the lower limits of detection of the Affymetrix system are not well known, we had concerns about data in which the HI of both groups is low. Therefore, transcripts with HI between − 1 and + 1 for both allografts and isografts were considered to be below the level of detection.

All statistical analyses were performed using the statistical software package SAS (SAS Institute Inc, Cary, NC).

Gene expression by RT-PCR

Semi-quantitative RT-PCR was performed on 13 significant transcripts for validation of the microarray data. Probe sequences from the U34A were used to create primers for two-step RT-PCR. The rat transcripts investigated were: Perforin, VEGF, Decorin, Fas-ligand, Allograft Inflammatory Factor (AIF), IL-10, Fibronectin 1, Heme Oxygenase 1 (HO1), Homeobox 1, Transforming Growth Factor-β1 (TGF-β1), SC1, Fc-γ receptor and EST197895. Additionally, the starting amount of cDNA was normalized using primers for GAPDH and the quantity of DNA contamination assessed with primers for β-actin. Reverse transcription was performed using 1 μg of total RNA with the ProSTARTM First-Strand RT-PCR kit (Stratagene, La Jolla, CA) to a final reaction volume of 50 μL. PCR was performed with 2.5 μL of cDNA, 1.5 units of Taq DNA polymerase (PE Biosystems, Foster City, CA), 2.5 μL PCR Buffer II (PE Biosystems), 2.0 mm MgCl2, 0.2 mm each dNTP (Roche, Indianapolis, IN) and 0.4 μm of each primer. PCR products were electrophoresed through 2% agarose, stained with ethidium bromide and visualized by Gel Doc 2000® (Bio-Rad, Hercules, CA). For each primer pair the linear range of amplification was determined by densitometry.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References

Histology

Minimal lymphocytic infiltration was noted on POD 3 in the allografts and no infiltration was seen on any POD in the isografts by H&E stain (Figure 1). A progressive, dense lymphocytic infiltrate was noted by POD 5 in all allografts with microvascular thrombosis by POD 7, verifying that severe acute cellular rejection occurred in all of the allograft hearts.

image

Figure 1. Histology of cardiac grafts. Hematoxylin and Eosin staining ( × 100) of cardiac grafts on POD 3, 5 and 7. Isografts (panels A–C) showed normal staining patterns, with no evidence of interstitial edema or inflammation. Allografts (panels D–F) showed increasing amounts of interstitial infiltrate and myocyte necrosis.

Download figure to PowerPoint

Comparing LAR and AD to measure expression

In the current studies we used the LAR rather than the AD to measure transcript expression. We inspected data from 18 microarrays, selected 35 probe sets hypothesized to be related to rejection using probability plots and observed that the LAR was more nearly normally distributed than AD. This is in accord with what would be expected for a process where effects and errors tend to be multiplicative rather than additive. Therefore we feel LAR provides a more representative estimate of the level of transcript expression. Also, the use of LAR as a measure of expression provides the opportunity to determine the level of significance of a particular transcript through the use of traditional statistical analyses and allows for the comparison between multiple samples in each group.

Criteria for determining increased and decreased expression

As previously described, we excluded some of the microarray data if they did not meet certain criteria. For example, transcripts were excluded if the HI of one or both groups was less than − 1, suggesting that the MM hybridization was greater than the PM hybridization. Taking POD 3 as an example, the HI for both the isograft and the allograft groups were < − 1 in 80 transcripts (0.9%) and 11 were found to be significantly decreased using the Student's t-test (Table 2). In addition, the HI of one group, but not the other was found to be < − 1 in 46 transcripts (1.9%) and 46 of these were found significantly altered (45 decreased and 1 increased). Finally, we also excluded data when transcripts in both groups showed low expression (HI between −1 and 1) because we believed that this low level of expression might often be below the level of detection for the microarray system. On POD 3, 3577 (40.7%) transcripts were in this category, suggesting that over half of the transcripts tested by the microarray had little or no expression in both groups at this timepoint. Of this group, 408 transcripts (320 increased and 88 decreased) were significantly altered but excluded due to their overall low expression.

Table 2. : Summary of the significant changes identified at each timepoint
 POD 3POD 5POD 7
CategoryTran- scripts (no.) – ΔHI (p < 0.05) + ΔHI (p < 0.05)Tran- scripts (no.) – ΔHI (p < 0.05) + ΔHI (p < 0.05)Tran- scripts (no.) – ΔHI (p < 0.05) + ΔHI (p < 0.05) Definition ofcategory
A  80 11   0 124  6   4  66 14   0HI is less than −1 for both groups
B 167 45   1 184 30  19 173 69   7HI less than −1 in one group
C3577 88 3203146168 1973505186 478HI between −1 and +1 in both groups
Total Excluded: 3824 144 321 3454 204 220 3744 269 485 
D1617  4 9421302211 4511839 501188HI in one group between −1 and+1
E3358  019224043361 9673216 551557HI > 1 in bothgroups
Total Included: 4975    4 2864 5345 572 1418 5055 105 2745  

Several observations support the use of these exclusion criteria. First, the ΔHI between isografts and allografts using these criteria was generally small – it must be less than 2 by definition. Second, the p-values of these significant but excluded transcripts generally were higher than those that were included for further analysis. No p-values were < 0.001 in the excluded group, while p-values < 0.001 were common in the included group. Finally, a relatively small percentage of those excluded on POD 3 were significant (12.2%, 465 of 3824), while a large percentage of those included were significant (57.6%, 2868 of 4975). Similarly, we excluded significant transcripts using the same criteria on POD 5 (n = 424) and POD 7 (n = 754) (Table 2).

After the above exclusions, 2864 transcripts were found to be significantly increased on POD 3 in allografts compared to isografts. In 942 of these transcripts, the HI of the isograft was < 1 while the HI in allograft rejection was > 1. This represents transcripts that were only minimally expressed in isografts but significantly increased during allograft rejection. In another 1922 significantly increased transcripts, both the HI of the isograft and the allograft groups were > 1. This represents transcripts that were expressed at detectable levels in the isografts, but were increased during allograft rejection. By POD 5, 1418 transcripts were significantly increased (451 with minimal expression in the isografts and 967 with moderate expression in the isografts). By POD 7, 2745 transcripts were significantly increased (1188 with minimal expression in isografts and 1557 with moderate expression in isografts).

A slightly different picture emerges when transcripts with decreased expression are studied. Only 4 transcripts were significantly decreased on POD 3 and 3 of these were expressed significantly in isografts and then only minimally in allografts. By POD 5, 572 transcripts were significantly decreased in the acute rejection group, including 211 transcripts decreased to minimally detectable levels and 361 transcripts decreased from high levels in isografts to lower levels in allografts (HI of isograft and allografts both > 1). By POD 7, 105 transcripts were significantly decreased – 50 decreased to minimally detectable levels and 55 decreased from high levels to lower levels in allografts.

Transcripts most increased during acute allograft rejection

Table 3 shows the most increased transcripts at all three timepoints. The transcripts are sorted by ΔHI and classified by the minimum HI criteria. Many of the most up-regulated transcripts on POD 3 were those commonly associated with inflammatory processes including interferon regulatory factor 1 (IRF-1), MHC Class I and II antigens and AIF. All of these genes remained up-regulated at POD 5 and 7 and were among the most significantly increased at these later timepoints. Some of the most significantly up-regulated genes were those that have never been associated with allograft rejection such as isoprenylated 67 kDa protein.

Table 3. : Most activated transcripts at each timepoint
 Probe setTranscript descriptionΔHIA,Bp-value
  • A

    Transcripts with isograft HI values less than 1 and allograft HI values greater than 1 are identified by plain text,

  • B

    B Transcripts in bold have HI values for the isografts and allografts greater than one.

POD3
 1.M80367_atIsoprenylated 67 kDa protein5.94 ± 0.94< 0.001
 2. X17053mRNA_s_at Immediate-early serum-responsive JE 5.68 ± 0.73 < 0.001
 3.rc_AA892553_atEST196356 Rat cDNA clone5.66 ± 0.48< 0.001
 4.X53054_atRT1.D beta chain5.32 ± 0.47< 0.001
 5. M34253_g_at Interferon regulatory factor 1 (IRF-1) 5.11 ± 0.69 < 0.001
 6. U17035_s_at Mob-1 mRNA 5.09 ± 0.71 < 0.001
 7. rc_AA891944_at EST195747 Rat cDNA clone 5.07 ± 0.78 < 0.001
 8. U17919_s_at Allograft inflammatory factor-1 4.99 ± 0.69 < 0.001
 9. K03039mRNA_s_at Thymocyte L-CA(leukocyte common antigen) 4.77 ± 1.14 0.003
10. M32062_g_at Fc-gamma receptor 4.70 ± 1.02 0.002
11. X17053cds_s_at Immediate-early serum-responsive JE 4.51 ± 0.75 < 0.001
12.rc_AA892506_atEST196309 Rat cDNA clone4.47 ± 0.50< 0.001
13.M34253_atInterferon regulatory factor 1 (IRF-1)4.36 ± 0.820.001
14. M36151cds_s_at MHC class II A-beta RT1.B-b-beta gene 4.34 ± 0.46 < 0.001
15. X56596_at MHC class II antigen RT1.B-1 beta-chain 4.22 ± 0.43 < 0.001
16.X57523_g_atMtp1 mRNA4.22 ± 0.73< 0.001
17. K02815_s_at MHC RT1-B region class II A-α glycoprotein 4.21 ± 0.35 < 0.001
18. X53054_g_at RT1.D beta chain 4.10 ± 0.42 < 0.001
19.rc_AA892259_atEST196062 Rat cDNA clone4.08 ± 0.56< 0.001
20. Y12009_at Chemokine coreceptor CKR5 4.08 ± 0.77 0.001
POD5
 1.M80367_atIsoprenylated 67 kDa protein7.16 ± 0.24< 0.001
 2.U17035_s_atMob-1 mRNA5.47 ± 0.48< 0.001
 3.rc_AA892553_atEST196356 Rat cDNA clone5.46 ± 0.61< 0.001
 4.J02722cds_atHeme oxygenase gene5.37 ± 0.42< 0.001
 5. U17919_s_at Allograft inflammatory factor-1 5.32 ± 0.49 < 0.001
 6.AF036537_g_atHomocysteine respondent protein HCYP25.28 ± 0.51< 0.001
 7. X53054_at Rat mRNA for RT1.D beta chain 5.21 ± 0.45 < 0.001
 8. D10757_g_at Proteasome subunit R-RING12 5.20 ± 0.25 < 0.001
 9. M34253_g_at Interferon regulatory factor 1 (IRF-1) 5.05 ± 0.42 < 0.001
10. X17053mRNA_s_at Immediate-early serum-responsive JE 5.00 ± 0.69 < 0.001
11.X57523_g_atMtp1 mRNA4.98 ± 0.34< 0.001
12. M32062_g_at Fc-gamma receptor 4.94 ± 0.36 < 0.001
13. U77777_s_at IFN-gamma inducing factor isoform α precursor 4.89 ± 0.54 < 0.001
14.X59012mRNA_s_atRat mRNA for trypsin V a-form4.87 ± 0.49< 0.001
15. K03039mRNA_s_at Thymocyte L-CA(leukocyte common antigen) 4.85 ± 0.63 < 0.001
16. U10894_s_at Rat mRNA expressed in carotid artery tissue 4.76 ± 0.45 < 0.001
17.X14319cds_g_atT-cell receptor beta chain4.76 ± 0.26< 0.001
18. D29646_at ADP-ribosyl cyclase (CD38) 4.69 ± 0.43 < 0.001
19.X71127_g_atComplement protein C1q beta chain4.66 ± 0.42< 0.001
20.M34253_atInterferon regulatory factor 1 (IRF-1)4.59 ± 0.31< 0.001
POD7
 1.X53054_atRT1.D beta chain7.44 ± 0.24< 0.001
 2.J02720_atLiver arginase mRNA6.79 ± 0.21< 0.001
 3. U17919_s_at Allograft inflammatory factor-1 6.74 ± 0.34 < 0.001
 4. M32062_g_at Fc-gamma receptor 6.59 ± 0.27 < 0.001
 5.M80367_atIsoprenylated 67 kDa protein6.58 ± 0.40< 0.001
 6.K03039mRNA_s_atThymocyte L-CA(leukocyte common antigen)6.34 ± 0.44< 0.001
 7.rc_AA892553_atEST196356 Rat cDNA clone6.21 ± 0.37< 0.001
 8.U77777_s_atIFN-gamma inducing factor isoform α precursor6.20 ± 0.28< 0.001
 9. rc_AA891944_at EST195747 Rat cDNA clone 6.19 ± 0.61 < 0.001
10.U10894_s_atRat mRNA expressed in carotid artery tissue6.05 ± 0.30< 0.001
11. M14656_at Osteopontin mRNA 5.96 ± 0.65 < 0.001
12. M34253_g_at Interferon regulatory factor 1 (IRF-1) 5.95 ± 0.42 < 0.001
13. J02722cds_at Heme oxygenase 5.92 ± 0.38 < 0.001
14.X71127_g_atComplement protein C1q beta chain5.89 ± 0.33< 0.001
15. D10757_g_at Proteasome subunit R-RING12 5.88 ± 0.41 < 0.001
16. S68135_s_at Glucose transporter (GLUT1) 5.83 ± 0.57 < 0.001
17.rc_AA800908_atEST190405 Rat cDNA clone5.82 ± 0.48< 0.001
18. L25387_g_at Phosphofructokinase C (PFK-C) 5.77 ± 0.63 < 0.001
19. M57276_at Leukocyte antigen MRC-OX44 5.75 ± 0.32 < 0.001
20.M36151cds_s_atMHC class II A-beta RT1.B-b-beta5.73 ± 0.34< 0.001

The probe sets increased on POD5 and POD7 appear to continue the association with inflammation. HO1, for example, was up-regulated more by POD 5 and 7 than on POD 3. Other probe sets were generally up-regulated at all three timepoints (AIF, TGF-β1, Fc-γ receptor, thymocyte L-CA and IRF-1 (Figure 2).

image

Figure 2. Kinetics of the most significantly increased transcripts. AIF, HO1, Fc-gamma, TGF-β1, Thymocyte L-CA and IRF-1 mRNA expression for isografts and allografts at POD3, 5 and 7. For each transcript, the differences in expression were statistically significant (p < 0.05) at all three timepoints. To assess variability, error bars representing the min/max hybridization index value for each transcript have been added.

Download figure to PowerPoint

Transcripts most decreased during acute allograft rejection

Table 4 shows the most decreased transcripts at all three timepoints sorted by ΔHI and classified by the minimum HI criteria. The p-values of the 4 transcripts decreased on POD 3 ranged from 0.006 to 0.048, while the p-values of the other 20 most-decreased transcripts on POD5 and 7 were generally smaller with nearly all < 0.001.

Table 4. : Most down-regulated transcripts at each timepoint
 Probe setTranscript descriptionΔHIA,Bp-value
  1. A Transcripts with allograft HI values less than 1 and isograft HI values greater than 1 are identified by plain text, BTranscripts in bold have HI values for the isografts and allografts greater than one.

POD3
 1.U07610_atZinc finger protein (HF-1b)− 1.32 ± 0.560.048
 2.D63886_s_atRat MT3-MMP-del− 0.94 ± 0.330.022
 3.rc_AI639417_atRat cDNA clone rx02173 3− 0.89 ± 0.370.040
 4.rc_AA926193_atR Rat cDNA clone UIA1evh020UI.s1− 0.79 ± 0.210.006
POD5
 1.U27562_atSC1 protein mRNA− 4.34 ± 0.27< 0.001
 2. rc_AA893242_g_at EST197045 Rat cDNA clone − 4.09 ± 0.79 0.001
 3.AB005743_g_atRat mRNA for fatty acid transporter− 3.95 ± 0.54< 0.001
 4. AB005743_at Rat mRNA for fatty acid transporter − 3.81 ± 0.61 < 0.001
 5. rc_AI169612_at EST215498 Rat cDNA clone − 3.79 ± 0.55 < 0.001
 6. X67948_at Channel integral membrane protein 28 − 3.62 ± 0.34 < 0.001
 7. rc_AA799666_g_at EST189163 Rat cDNA clone − 3.62 ± 1.16 0.014
 8. U08976_at Wistar peroxisomal enoyl hydratase-like protein − 3.29 ± 0.58 < 0.001
 9.rc_AI639162_atRat cDNA clone rx01122 3− 3.26 ± 0.44< 0.001
10. AF072411_at Fatty acid translocase/CD36 − 3.12 ± 0.63 0.001
11.D89730_g_atRat T16 mRNA− 3.07 ± 0.550.001
12.L19998_atMinoxidil sulfotransferase− 3.02 ± 0.52< 0.001
13.M27726_atPhosphorylase (B-GP1) mRNA− 2.99 ± 0.43< 0.001
14.Y13275_atD6.1 A protein− 2.98 ± 0.42< 0.001
15. D00680_at Plasma glutathione peroxidase (EC 1.11.1.9) − 2.95 ± 0.25 < 0.001
16. S49491_s_at Proenkephalin − 2.90 ± 0.70 0.003
17. rc_AA946368_at EST201867 Rat cDNA clone − 2.88 ± 0.33 < 0.001
18. D28561_s_at Glucose transporter, GLUT4 − 2.85 ± 0.35 < 0.001
19. L15556_at Phospholipase C (BETA4) mRNA − 2.83 ± 0.60 0.002
20.rc_AA851223_atEST193991 Rat cDNA clone− 2.79 ± 0.41< 0.001
POD7
 1.Z12298cds_s_atDermatan sulfate proteoglycan-II (decorin)− 5.73 ± 0.38< 0.001
 2. rc_AI639233_s_at Rat cDNA clone rx05007 3 − 5.62 ± 0.86 < 0.001
 3. X59859_i_at Decorin (DCN) − 4.88 ± 0.63 < 0.001
 4.U27562_atSC1 protein− 4.51 ± 0.46< 0.001
 5. rc_AA894092_at EST197895 Rat cDNA clone − 4.00 ± 0.57 < 0.001
 6.U65656_atGelatinase A− 3.55 ± 0.28< 0.001
 7.X59859_r_at Decorin (DCN) − 3.34 ± 0.76 0.002
 8.X67948_atChannel integral membrane protein 28− 3.26 ± 0.640.001
 9. rc_AI172411_at EST218418 Rat cDNA clone − 3.04 ± 0.29 < 0.001
10. M21354_s_at Collagen type III alpha-1 − 3.03 ± 0.51 < 0.001
11. X70369_s_at Pro alpha 1 collagen type III − 3.03 ± 0.83 0.006
12. D00680_at Plasma glutathione peroxidase (EC 1.11.1.9) − 2.96 ± 0.44 < 0.001
13. U08976_at Wistar peroxisomal enoyl hydratase-like protein − 2.88 ± 0.91 0.013
14.rc_AA851223_atEST193991 Rat cDNA clone− 2.66 ± 0.520.001
15. rc_AI012030_at EST206481 Rat cDNA clone − 2.60 ± 0.50 0.001
16. X84039_at Lumican mRNA − 2.60 ± 0.70 0.006
17.D89730_g_atRat T16 mRNA− 2.54 ± 0.620.003
18.Z24721_atRat SOD gene− 2.34 ± 0.430.001
19.rc_AA800190_atEST189687 Rat cDNA clone− 2.34 ± 0.540.002
20.X64827cds_s_atCytochrome c oxidase (subunit VIII-h)− 2.29 ± 0.40< 0.001

The kinetics of selected decreased genes are shown in Figure 3. On POD 3, a group of genes not commonly associated with rejection were significantly down-regulated including mRNA up-regulated during prostatic apoptosis, phosphacan mRNA, prolactin receptor and follicle stimulating hormone receptor. On POD 5, the SC1 protein was the most down-regulated along with three genes involved in fatty acid transport. By POD7 the SC1 was still decreased along with decorin, an antagonist of TGFβ1.

image

Figure 3. Kinetics of the most significantly decreased transcripts. Lumican, Collagen IIIα1, Cytochrome C oxidase, Decorin, SC1 and Gelatinase mRNA expression for isografts and allografts at POD 3, 5 and 7. Although all transcripts showed significant differences by POD 7, only Lumican, Collagen IIIα1 and Cytochrome C oxidase expression were significant (p < 0.05) at all three timepoints. Error bars illustrate the min/max hybridization index value for each transcript.

Download figure to PowerPoint

In contrast to many of the up-regulated genes that were increased at all timepoints, the genes most down-regulated by POD 7 were not down-regulated on POD 3. These genes include: lumican, collagen IIIα1, cytochrome C oxidase, decorin, SC1 and gelatinase (Figure 3).

Verifying microarray gene expression data with RT-PCR

To confirm that microarray data actually represented mRNA expression, we performed RT-PCR on 13 transcripts. Transcripts with high induction (n = 3), mild induction (n = 4), minimal changes (n = 3) and high down-regulation (n = 3) were tested. First, the linear range assessment was performed using semi-quantitative RT-PCR with primers designed in our laboratory. A sample of these linear range assessments is shown in Figure 4. Figure 5 shows that the RT-PCR results were visually correlative with what was observed in the microarray. For 10 of the transcripts, the microarray data correctly predicted the magnitude and direction of the change identified by RT-PCR. In three up-regulated transcripts (Homeobox 1, VEGF and Fas Ligand), the direction (up/down) of expression was consistent, but the magnitude of the difference was errant. Further investigation of the microarray data showed the HI values for these transcripts were negative for both the allograft and isograft samples. This further supports that transcripts with negative HI values may be below the level of sensitivity of the system and caution should be taken when interpreting data from these transcripts.

image

Figure 4. Linear range assessment of Fc-gamma (A) and Decorin (B). Semi-quantitative RT-PCR was performed to confirm trends identified by the GeneChips®. For each transcript, allograft, isograft and normal RNA was used to determine the linear range of PCR amplification. During PCR amplification, samples were removed at the indicated PCR cycle, products separated via electrophoresis, stained with ethidium bromide and analyzed via densitometry. The center of the linear range of PCR amplification for fc-gamma was cycle 27 and cycle 24 for decorin.

Download figure to PowerPoint

image

Figure 5. Analysis of normal, isograft and allograft mRNA expression by RT-PCR. After determining the linear range of PCR amplification for significant transcripts, a comparison between normal, isograft (POD3, 5 and 7) and allograft (POD3, 5 and 7) samples was made for decorin, IL-10, HO1, TGF-β1 and SC1. PCR products were electrophoresed through 2% agarose and visualized with ethidium bromide. In general, trends in mRNA expression identified with the GeneChip® correlated well with RT-PCR. GAPDH primers were used as housekeeping control to assess the amount of cDNA in each reaction.

Download figure to PowerPoint

Expression of known ‘transplant-related’ genes

We selected prospectively a subset of 400 transcripts identified as known to be associated with acute cellular rejection. These transcripts were primarily MHC molecules, cytokines, chemokines, adhesion molecules and ECM proteins. Significant changes in expression were identified on POD 3 (157), 5 (131) and 7 (166). Table 55 shows that 95 of these 400 transcripts were significant throughout the study of acute rejection.

Table 5. : 95 Transplant related transcripts (significant at all timepoints)
 Probe setTranscript descriptionPOD3 (ΔHI)POD5 (ΔHI)POD7 (ΔHI)
 1.U17919_s_atAllograft inflammatory factor-14.99 ± 0.695.32 ± 0.496.74 ± 0.34
 2.S76511_s_atApoptosis inducer (BAX)1.21 ± 0.422.00 ± 0.301.84 ± 0.22
 3.L14680_g_atBcl-2 mRNA0.75 ± 0.290.46 ± 0.191.07 ± 0.18
 4.M77246_atBeta-chain clathrin protein complex AP-21.11 ± 0.480.73 ± 0.261.34 ± 0.43
 5.M17069_atCalmodulin (RCM3) mRNA1.42 ± 0.400.97 ± 0.171.48 ± 0.44
 6.X13933_s_atCalmodulin gene2.02 ± 0.331.74 ± 0.252.07 ± 0.24
 7.E02315cds_f_atCalmodulin gene2.95 ± 0.741.83 ± 0.302.49 ± 0.41
 8.Y12009_atChemokine coreceptor CKR54.08 ± 0.773.73 ± 0.495.62 ± 0.36
 9.AF030358_atChemokine CX3C1.81 ± 0.381.05 ± 0.251.69 ± 0.21
10.AF030358_g_atChemokine CX3C2.26 ± 0.711.33 ± 0.293.15 ± 0.47
11.U77349cds_atChemokine receptor CCR2 gene2.84 ± 0.452.82 ± 0.981.24 ± 0.35
12.U16025_atClass Ib RT1 mRNA1.85 ± 0.341.61 ± 0.242.29 ± 0.32
13.U90610_g_atCXC chemokine receptor (CXCR4)1.18 ± 0.360.92 ± 0.353.02 ± 0.37
14.C07012_f_atCyclophilin C mRNA1.19 ± 0.201.44 ± 0.192.12 ± 0.17
15.X52815cds_f_atCytoplasmic-gamma isoform of actin2.07 ± 0.381.17 ± 0.302.56 ± 0.19
16.X05834_atFibronectin gene2.31 ± 0.501.40 ± 0.583.86 ± 0.31
17.D86641_atFK506-binding protein 121.25 ± 0.360.92 ± 0.241.65 ± 0.33
18.L01624_atGlucocorticoid-regulated kinase (sgk)1.72 ± 0.562.34 ± 0.264.11 ± 0.49
19.X66693_f_atGranzyme-like protein I1.39 ± 0.603.56 ± 0.372.41 ± 0.26
20.J02722cds_atHeme oxygenase2.45 ± 0.485.37 ± 0.425.92 ± 0.38
21.X14323cds_atIgG receptor FcRn large subunit p510.67 ± 0.18− 0.87 ± 0.19− 0.66 ± 0.27
22.E01884cds_s_atIL-1-beta2.22 ± 0.552.64 ± 0.321.44 ± 0.19
23.M14050_s_atImmunoglobulin heavy chain binding protein3.15 ± 0.801.18 ± 0.482.93 ± 0.60
24.U59801_atIntegrin alpha-M (Itgam)1.14 ± 0.270.97 ± 0.422.64 ± 0.43
25.AF003598_atIntegrin beta-7 subunit0.97 ± 0.290.96 ± 0.221.00 ± 0.17
26.AF017437_g_atIntegrin-associated protein form 4 (IAP)2.68 ± 0.441.81 ± 0.512.11 ± 0.53
27.M34253_atInterferon regulatory factor 1 (IRF-1)4.36 ± 0.824.59 ± 0.314.06 ± 0.30
28.M34253_g_atInterferon regulatory factor 1 (IRF-1)5.11 ± 0.695.05 ± 0.425.95 ± 0.42
29.U77777_s_atIFN-gamma inducing factor isoform α3.60 ± 0.884.89 ± 0.546.20 ± 0.28
30.M98820_atIL-1-beta3.38 ± 0.943.21 ± 0.342.70 ± 0.41
31.M98820_g_atIL-1-beta2.15 ± 0.602.19 ± 0.551.27 ± 0.44
32.X69903_atIL-4 receptor1.27 ± 0.270.68 ± 0.141.72 ± 0.26
33.M58587_atIL-6 receptor ligand binding chain1.67 ± 0.481.38 ± 0.522.22 ± 0.33
34.U14647_atIL-1 beta converting enzyme2.65 ± 0.573.39 ± 0.523.48 ± 0.35
35.U14647_g_atIL-1 beta converting enzyme1.96 ± 0.712.74 ± 0.573.33 ± 0.25
36.S79676_s_atIL-1 beta-converting enzyme3.03 ± 0.563.18 ± 0.363.82 ± 0.33
37.M55050_atIL-2 receptor beta chain2.46 ± 0.543.03 ± 0.261.73 ± 0.28
38.S79263_s_atIL-3 receptor beta-subunit1.16 ± 0.430.81 ± 0.291.50 ± 0.11
39.M82826_i_atLeucopus neurofibromatosis protein type I0.75 ± 0.170.49 ± 0.130.43 ± 0.11
40.M57276_atLeukocyte antigen MRC-OX443.97 ± 0.444.23 ± 0.545.75 ± 0.32
41.M25823_s_atLeukocyte-common antigen (L-CA or CD45)1.23 ± 0.270.94 ± 0.231.32 ± 0.13
42.S79523_atLymphocyte membrane protein A.112.29 ± 0.723.09 ± 0.432.00 ± 0.39
43.U22414_atMacrophage inflammatory protein-1alpha1.66 ± 0.543.30 ± 0.562.41 ± 0.29
44.U06434_atMacrophage inflammatory protein-1beta1.98 ± 0.442.90 ± 0.302.65 ± 0.25
45.S73424_s_atMacrophage migration inhibitory factor1.53 ± 0.440.96 ± 0.192.22 ± 0.42
46.AF074608mRNA_f_atMHC class I antigen (RT1.EC2)2.27 ± 0.292.13 ± 0.193.09 ± 0.22
47.AF074609mRNA_f_atMHC class I antigen (RT1.EC3)2.31 ± 0.371.65 ± 0.142.57 ± 0.29
48.M64795_f_atMHC class I antigen gene (RT1u haplotype)2.63 ± 0.362.44 ± 0.243.24 ± 0.22
49.M11071_f_atMHC class I cell surface antigen1.52 ± 0.400.68 ± 0.231.72 ± 0.46
50.L23128_g_atMHC class I mRNA1.57 ± 0.361.92 ± 0.311.63 ± 0.28
51.L23128_atMHC class I mRNA1.72 ± 0.391.97 ± 0.231.71 ± 0.30
52.M31038_atMHC class I non-RT1.A alpha-1-chain3.32 ± 0.604.20 ± 0.364.99 ± 0.60
53.M24026_f_atMHC class I RT1 (RT44)2.01 ± 0.311.38 ± 0.362.39 ± 0.32
54.M24324_f_atMHC class I RT1 (RTS)2.15 ± 0.241.96 ± 0.262.46 ± 0.27
55.M31018_f_atMHC class I RT1.Aa alpha-chain1.92 ± 0.341.35 ± 0.282.03 ± 0.35
56.L40362_f_atMHC class I RT1.C-type protein1.72 ± 0.461.85 ± 0.232.33 ± 0.28
57.L40364_f_atMHC class I RT1.O type − 149 processed2.02 ± 0.371.62 ± 0.182.22 ± 0.37
58.M10094_atMHC class I truncated cell surface antigen1.26 ± 0.351.23 ± 0.401.19 ± 0.34
59.M10094_g_atMHC class I truncated cell surface antigen2.79 ± 0.382.67 ± 0.323.45 ± 0.42
60.AF025308_f_atMHC class Ib antigen (RT1.Cl)2.56 ± 0.312.50 ± 0.203.90 ± 0.31
Table 5. : Continued
 Probe setTranscript descriptionPOD3 (ΔHI)POD5 (ΔHI)POD7 (ΔHI)
61.AF029240_g_atMHC class Ib RT1.S3 (RT1.S3)2.61 ± 0.512.45 ± 0.242.40 ± 0.33
62.AF029240_atMHC class Ib RT1.S3 (RT1.S3)3.85 ± 0.794.49 ± 0.444.35 ± 0.34
63.M36151cds_s_atMHC class II A-beta RT1.B-b-beta4.34 ± 0.463.93 ± 0.345.73 ± 0.34
64.X56596_atMHC class II antigen RT1.B-1 beta-chain4.22 ± 0.433.32 ± 0.395.15 ± 0.36
65.M15562_g_atMHC class II RT1.u-d-alpha chain2.00 ± 0.441.01 ± 0.192.43 ± 0.42
66.M15562_atMHC class II RT1.u-d-alpha chain3.42 ± 0.372.87 ± 0.263.87 ± 0.52
67.X14254cds_atMHC class II-associated invariant chain3.14 ± 0.442.52 ± 0.333.36 ± 0.35
68.X14254cds_g_atMHC class II-associated invariant chain3.09 ± 0.322.28 ± 0.283.60 ± 0.32
69.U31598_s_atMHC class II-like alpha chain (RT1.DMa)2.29 ± 0.312.66 ± 0.294.24 ± 0.32
70.U31599_atMHC class II-like beta chain (RT1.DMb)2.58 ± 0.272.94 ± 0.394.18 ± 0.21
71.U31599_g_atMHC class II-like beta chain (RT1.DMb)2.22 ± 0.392.80 ± 0.214.31 ± 0.29
72.X07551cds_s_atMHC RT1.B-alpha gene for class II antigen3.27 ± 0.383.01 ± 0.304.20 ± 0.31
73.K02815_s_atMHC RT1-B region class II (Ia antigen) A-α4.21 ± 0.354.16 ± 0.495.02 ± 0.43
74.X55986mRNA_s_atMulticatalytic proteionase subunit L ingensin1.73 ± 0.511.00 ± 0.272.01 ± 0.35
75.Z14120cds_s_atPlatelet-derived growth factor A chain1.20 ± 0.281.14 ± 0.381.02 ± 0.22
76.AJ222813_s_atPrecursor IL-182.53 ± 0.544.12 ± 0.414.47 ± 0.23
77.C06598_atRapamycin-binding protein FKBp-130.78 ± 0.180.86 ± 0.170.95 ± 0.18
78.E13732cds_atRat CC chemokine receptor protein1.61 ± 0.691.43 ± 0.352.58 ± 0.36
79.M22366_atRat MHC class II RT1.B-alpha chain0.77 ± 0.170.70 ± 0.221.21 ± 0.20
80.AF075383_atSuppressor of cytokine signaling-31.21 ± 0.431.47 ± 0.231.12 ± 0.16
81.U76836_g_atT cell receptor V alpha 2 chain subunit1.33 ± 0.282.03 ± 0.270.74 ± 0.15
82.D13555_atT cell receptor zeta chain1.94 ± 0.502.27 ± 0.461.63 ± 0.27
83.X14319cds_g_atT-cell receptor beta chain3.35 ± 0.494.76 ± 0.263.32 ± 0.18
84.S75435_i_atT-cell receptor gamma chain0.87 ± 0.171.21 ± 0.232.07 ± 0.17
85.M10072mRNA_s_atThymocyte leukocyte common antigen2.90 ± 0.824.07 ± 0.584.29 ± 0.28
86.M18349cds#1_s_atThymocyte leukocyte common antigen2.54 ± 0.823.08 ± 0.693.27 ± 0.26
87.K03039mRNA_s_atThymocyte leukocyte common antigen4.77 ± 1.144.85 ± 0.636.34 ± 0.44
88.X04310_atThymocyte mRNA for CD8 antigen1.14 ± 0.312.22 ± 0.232.19 ± 0.24
89.X03015_atThymocyte mRNA for OX-8 antigen2.31 ± 0.624.00 ± 0.394.19 ± 0.31
90.AJ012603UTR#1_g_atTNF-alpha converting enzyme (TACE)1.67 ± 0.332.27 ± 0.412.79 ± 0.31
91.AJ012603UTR#1_atTNF-alpha converting enzyme (TACE)1.54 ± 0.491.44 ± 0.302.41 ± 0.23
92.X52498cds_atTransforming growth factor-beta 11.93 ± 0.291.48 ± 0.342.36 ± 0.29
93.M63122_atTumor necrosis factor receptor1.97 ± 0.301.47 ± 0.301.58 ± 0.38
94.X63722cds_s_atVCAM-12.57 ± 0.511.51 ± 0.381.04 ± 0.37
95.M15768_atW3/25 antigen (homologue of human CD4)1.28 ± 0.350.69 ± 0.272.47 ± 0.36

The entire dataset and RT-PCR sequences may be obtained electronically from park.walter@mayo.edu.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References

The use of high-throughput microarrays promises to provide detailed measurements of gene expression on an unprecedented global scale. However, skeptics rightly criticize the interpretation of microarray data obtained from complex biological systems because of sampling variability, assay variability and difficulties in managing the enormous amount of data. The current study was an initial attempt to address some of the fundamental questions of microarray technology using a well-described rat heterotopic cardiac allograft rejection model.

Given the homogeneity of the model system and the purity of mRNA isolated, run-to-run variability was found to be quite low. Our statistical approach using the HI based on the LAR allowed for the evaluation of the large amount of microarray data. The HI also allowed for the censoring of some significant results when hybridization was very low or when the hybridization of the MM primers was greater than that of the PM ones. In addition, the HI could generally classify transcript expression as very low (HI < 1) or detectable to high (HI > 1).

Using these methods, thousands of transcripts were found to be significantly altered in cardiac allografts undergoing acute rejection at various timepoints after transplantation. For example, 1418 transcripts were determined to be significantly increased by POD 5 during acute rejection – 451 increasing from minimal or no expression and 967 increasing from a lower but detectable level in isografts. Also at POD 5, 572 transcripts were significantly decreased – 211 decreased to undetectable levels and 361 decreased from higher levels in isografts to lower levels in the allograft group. Of note, comparisons between HI and mean FC were at most only weakly correlated. Using FC would have incorrectly predicted the magnitude and direction of the change for 38% of the POD 7 transcripts we verified by RT-PCR (data not shown).

Our approach resulted in the identification of many more significantly altered transcripts compared to two previous reports which used FC to evaluate microarray data from acute rejection models (13,14). In a mouse cardiac allograft model, only 84 of 10 000 transcripts were reported as increased on POD 5. Transcripts that showed a 3-fold increase in every one of the three rejecting grafts when compared individually to the three isografts were considered ‘reliable’ for consideration (13). Only 10 of 7129 probe sets were reported as increased during acute rejection of human renal allografts. Transcripts from seven renal allograft biopsies were sequentially compared to three normal renal allograft biopsies, yielding only four transcripts that were consistently up-regulated (14). We believe that given the striking pathologic changes of acute rejection, these results seem unlikely. Further, given the variability of biological systems and the microarray technology, it is unlikely that any significant alterations in transcript expression would be found using the FC method as the sample size increases.

The LAR, as used in the current studies, has several additional advantages over the FC. Using this measure we can determine relative changes in expression, ΔHI, between two groups. We can also determine which transcripts are significantly changing between the two groups by using the Student's t-test to calculate the p-value. Further, we can easily select and remove outliers which fall outside the criteria of a predetermined minimum hybridization ratio. These calculations allow for expedient analysis of microarray data.

A detailed presentation and analysis of such a large amount of gene expression data is beyond the scope of this initial manuscript. Using microarrays we identified a large cohort of significantly altered genes at each timepoint. Most of these have not been previously reported and therefore may or may not be related to acute rejection. Instead, these changes may be due to downstream events and ‘side-effects’ unrelated to the rejection process. However, a general assessment of the data lends further support to our approach. Many of the transcripts found to be significantly up-regulated by the current microarray analysis were those demonstrated to be increased in previous studies of gene expression in the same rat cardiac allograft model using RT-PCR and Northern blot analyses. This includes molecules such as the MHC molecules, Perforin, Fas Ligand, IFN-γ, RANTES, ICAM, E-selectin and TNF-α(6–9,17,18). In addition, we found many other ‘transplant-related genes’ increased at all 3 timepoints. As expected, most of the up-regulated genes were associated with inflammation. We found many seemingly non-inflammatory genes with unknown function in transplantation were up-regulated. In addition we found many seemingly non-related non-inflammatory ‘transplant-related genes’.

Interestingly, not all molecules believed to be associated with inflammation were significantly altered in our study. For example, we did not identify significant changes in gene expression for the cytokines IL-2, IL-4 or IL-10, which have been shown previously to be involved in various models of transplant rejection (19–21). The HI values for these transcripts consistently placed them into Category C and we therefore concluded these molecules were expressed below the level of detection of the GeneChip® in this model. Although not within the scope of our study, it would be possible to monitor changes in these molecules on an individual basis with a similar RT-PCR design as the one presented here.

This is the first study of acute rejection either using microarrays or other techniques that describe genes that are down-regulated in this process. This down-regulation may represent changes secondary to cytokine release by the infiltrating cells or to graft injury by ischemia or other mechanisms. Of the down-regulated genes, SC1 and decorin appear to be especially attractive for further study. Decorin is a small chondroitin-dermatan sulphate proteoglycan and was recently identified as a potential antagonist of TGF-β1 (22,23). The absence of decorin might allow the unimpeded activity of TGF-β1 in its role of stimulating inflammation and fibrogenesis. SC1 is an ECM glycoprotein detected in both the developing CNS and adult brain tissue (24). SC1 shares sequence homology with SPARC, considered to be anti-adhesive because it can selectively disrupt cellular contacts with the matrix (25). Thus, the absence of SC1 might allow for improved adhesion with the ECM of the graft. No previous studies of either decorin or SC1 exist in the transplantation literature.

The finding that a large number of the transcripts were either up-regulated or down-regulated Expressed Sequence Tags (EST) is especially interesting. ESTs are genes that show evidence of mRNA and protein production for which there is no known function. Without the microarray technology, it is unlikely that these transcripts would be associated with acute rejection, yet they may be found to be crucial for its pathogenesis. Similarly, genes that are associated with a known biological process such as mRNA up-regulated during prostatic apoptosis (most down-regulated on POD 3 in acute rejection) or osteopontin (significantly up-regulated on POD 7) will likely lead to a greater understanding of their role in fundamental biological pathways.

One potential limitation of the microarray that we were unable to address in this analysis is the issue of sequence variants. Given the small length of the probes, it is possible that at least some of the mRNA labeled as SC1 (or other seemingly unrelated genes) might actually be SC1-like molecules that share sequence homology but actually arise from separate genes or at least represent spliced variants. Further, more detailed analysis with actual sequencing of both DNA and mRNA will be necessary to answer this question.

We believe that microarray technology as is used in the current studies of acute rejection is a powerful screening test identifying numerous targets for further study. An exhaustive, detailed analysis of all the potential targets will take much more time and is beyond the scope of this article. Elucidation of the mechanisms of acute rejection will involve a more thorough analysis of transcripts that are involved in the fibrogenic process. Similarly, the localization of transcripts within the allograft will require other techniques such as in-situ PCR. Finally, verification of these data also will require the demonstration that protein corresponding to the altered gene is altered in the graft.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. References
  • 1
    Ginns L, Cosimi A, Morris P, eds. Transplantation. Malden: Blackwell Science; 1999.
  • 2
    Pascual M, Swinford RD, Ingelfinger JR, Williams WW, Cosimi AB, Tolkoff-Rubin N. Chronic rejection and chronic cyclosporin toxicity in renal allografts. Immunol Today 1998; 19: 514519.
  • 3
    Gao SZ, Alderman EL, Schroeder JS, Silverman JF, Hunt SA. Accelerated coronary vascular disease in the heart transplant patient: coronary angiographic findings. J Am Coll Cardiol 1988; 12: 334340.
  • 4
    Chaparro C, Scavuzzo M, Winton T, Keshavjee S, Kesten S. Status of lung transplant recipients surviving beyond five years. J Heart Lung Transplant 1997; 16: 511516.
  • 5
    Dallman M. Immunobiology of graft rejection. In: Ginns L, Cosimi A, Morris P, eds. Transplantation. Malden: Blackwell Science; 1999. pp. 2342.
  • 6
    Kageyama Y, Li XK, Suzuki S et al. Apoptosis is involved in acute cardiac allograft rejection in rats. Ann Thorac Surg 1998; 65: 16041609.
  • 7
    Alexander DZ, Pearson TC, Hendrix R, Ritchie SC, Larsen CP. Analysis of effector mechanisms in murine cardiac allograft rejection. Transplant Immunol 1996; 4: 4648.
  • 8
    Yamazaki S, Isobe M, Suzuki J et al. Role of selectin-dependent adhesion in cardiac allograft rejection. J Heart Lung Transplant 1998; 17: 10071016.
  • 9
    Mulligan MS, McDuffie JE, Shanley TP et al. Role of RANTES in experimental cardiac allograft rejection. Exp Mol Pathol 2000; 69: 167174.
  • 10
    Schulze A, Downward J. Navigating gene expression using microarrays – a technology review. Nat Cell Biol 2001; 3: E190E195.
  • 11
    Harkin DP. Uncovering functionally relevant signalling pathways using microarray-based expression profiling. Oncologist 2000; 5: 501507.
  • 12
    Notterman DA, Alon U, Sierk AJ, Levine AJ. Transcriptional gene expression profiles of colorectal adenoma, adenocarcinoma and normal tissue examined by oligonucleotide arrays. Cancer Res 2001; 61: 31243130.
  • 13
    Saiura A, Mataki C, Murakami T et al. A comparison of gene expression in murine cardiac allografts and isografts by means DNA microarray analysis. Transplantation 2001; 72: 320329.
  • 14
    Akalin E, Hendrix RC, Polavarapu G et al. Gene expression analysis in human renal allograft biopsy samples using high-density oligoarray technology. Transplantation 2001; 72: 948953.
  • 15
    Ono K, Lindsey E. Improved technique of heart transplantation in rats. J Thorac Cardiovasc Surg 1969; 57: 225229.
  • 16
    Searle S, Casella G, McCulloch C, eds. Variance Components. New York: John Wiley & Sons, 1992.
  • 17
    Pascoe MD, Marshall SE, Welsh KI, Fulton LM, Hughes DA. Increased accuracy of renal allograft rejection diagnosis using combined perforin, granzyme B and Fas ligand fine-needle aspiration immunocytology. Transplantation 2000; 69: 25472553.
  • 18
    Segerer S, Cui Y, Eitner F et al. Expression of chemokines and chemokine receptors during human renal transplant rejection. Am J Kidney Dis 2001; 37: 518531.
  • 19
    Xu XY, Honjo K, Devore-Carter D, Bucy RP. Immunosuppression by inhibition of cellular adhesion mediated by leukocyte function-associated antigen-1/intercellular adhesion molecule-1 in murine cardiac transplantation. Transplantation 1997; 63: 876885.
  • 20
    Coito AJ, Korom S, Graser E, Volk HD, Van De Water L, Kupiec-Weglinski JW. Blockage of very late antigen-4 integrin binding to fibronectin in allograft recipients: I. Treatment with connecting segment-1 peptides prevents acute rejection by suppressing intragraft mononuclear cell accumulation, endothelial activation, and cytokine expression. Transplantation 1998; 65: 699706.
  • 21
    Shirwan H, Barwari L, Khan NS. Predominant expression of T helper 2 cytokines and altered expression of T helper 1 cytokines in long-term allograft survival induced by intrathymic immune modulation with donor class I major histocompatibility complex peptides. Transplantation 1998; 66: 18021809.
  • 22
    Yamaguchi Y, Mann DM, Ruoslahti E. Negative regulation of transforming growth factor-β by the proteoglycan decorin. Nature 1990; 346: 281284.
  • 23
    Fischer JW, Kinsella MG, Levkau B, Clowes AW, Wight TN. Retroviral overexpression of decorin differentially affects the response of arterial smooth muscle cells to growth factors. Arterioscler Thromb Vasc Biol 2000; 21: 777784.
  • 24
    Johnston IG, Paladino T, Gurd JW, Brown IR. Molecular cloning of SC1: a putative brain ECM glycoprotein show in partial similarity to osteonectin/BM40/SPARC. Neuron 1990; 4: 165176.
  • 25
    Mothe AJ, Brown IR. Selective transport of SC1 mRNA, encoding a putative ECM glycoprotein, during postnatal development of the rat cerebellum and retina. Mol Brain Res 2000; 76: 7384.