Molecular mechanisms of chronic kidney transplant rejection via large-scale proteogenomic analysis of tissue biopsies

Abstract

The most common cause of kidney transplant failure is the poorly characterized histopathologic entity interstitial fibrosis and tubular atrophy (IFTA). There are no known unifying mechanisms, no effective therapy, and no proven preventive strategies. Possible mechanisms include chronic immune rejection, inflammation, drug toxicity, and chronic kidney injury from secondary factors. To gain further mechanistic insight, we conducted a large-scale proteogenomic study of kidney transplant biopsies with IFTA of varying severity. We acquired proteomic data using tandem mass spectrometry with subsequent quantification, analysis of differential protein expression, validation, and functional annotations to known molecular networks. We performed genome-wide expression profiling in parallel. More than 1400 proteins with unique expression profiles traced the progression from normal transplant biopsies to biopsies with mild to moderate and severe disease. Multiple sets of proteins were mapped to different functional pathways, many increasing with histologic severity, including immune responses, inflammatory cell activation, and apoptosis consistent with the chronic rejection hypothesis. Two examples include the extensive population of the alternative rather than the classical complement pathway, previously not appreciated for IFTA, and a comprehensive control network for the actin cytoskeleton and cell signaling of the acute-phase response. In summary, this proteomic effort using kidney tissue contributes mechanistic insight into several biologic processes associated with IFTA.

Figures

Figure 1.

(A) overlap between independently collected data sets, each composed of multiple technical replicates (data 1, green; data 2, blue) for the b0, b1, and b23 classes represents the consensus proteomic data. (B) Venn diagram showing the overlap between consensus b0, b1, and b23 protein identifications. (C) Relative abundance distribution of differentially expressed proteins. The binary overlaps (b0-b1, b0-b23, and b1-b23) contain proteins that are differentially expressed at P = 0.05 level of significance (two-tailed independent t test), whereas the b0-b1-b23 overlap represents differentially expressed proteins at P = 0.05 level of significance as determined with one-way ANOVA. The areas marked with arrows represent total pair-wise differentially expressed proteins between Banff categories.

Figure 2.

(A and B) Hierarchical clustering outcome for the biologic function (A) and canonical pathway (B) annotations of differentially expressed proteins is shown. Protein identifications were split into 15 groups (data set key) that represent expression programs in different stages of CAN (b0 upregulated, green; b1 upregulated, yellow; b23 upregulated, red). Both functional categories and differential groups of proteins were clustered using Spearman rank average linkage metric. Each data point is colored according to the probability score (expressed as P value) of a functional category in a particular differentially expressed group of proteins.

Figure 3.

(A and B) Top biologic functions (A) and canonical pathways (B) of differentially expressed and unique proteins among b0, b1, and b23 stages of CAN are shown. Each data series consists of proteins upregulated during b0 (black), b1 (red), and b23 (green) stages of CAN. Each enumerated annotation is assigned a probabilistic significance score represented as P value. The table lists enumerated annotations as well as the fractions of identified proteins that belong to that particular biologic function category (A) and the number of identified proteins that belong to a particular canonical pathway (B).

Figure 4.

Complement activation pathway as represented by Ingenuity Pathway Analysis. Differentially expressed proteins are color coded as unique to b23 (gray), upregulated in b23 (shades of red), and upregulated in b1 (yellow). The upregulated proteins are labeled with the abundance fold change relative to the CAN stage with the lowest abundance. Colored rectangular boxes next to identified proteins represent correlation coefficient (from −1 [green] to +1 [red]) between the protein and gene expression across the three stages of CAN.

Figure 5.

Actin cytoskeleton signaling pathway as represented by Ingenuity Pathway Analysis. Differentially expressed proteins are color coded as unique to b23 (gray), upregulated in b23 (shades of red), and upregulated in b1 (yellow). The upregulated proteins are labeled with the abundance fold change relative to the CAN stage with the lowest abundance. Colored rectangular boxes next to identified proteins represent correlation coefficient (from −1 [green] to +1 [red]) between the protein and gene expression across the three stages of CAN.