Figures

Click on the images for enlarged versions

Figure 1. Age-regulated genes. A. Shown are expression levels for gene CDO1. White and black circles represent expression from cortex and medulla, respectively. y-axis indicates log2(expression level) and x-axis indicates age of patient (years). Dotted and solid lines are least squares fits for cortex and medulla samples respectively, constrained to have a common slope.

Figure 1B. Histogram of age-regulated genes. For every gene, we calculated a one sided p-value that its expression changes with age. Shown is a histogram representing all of the genes represented by the Affymetrix DNA chip. Genes that decrease with age have p-values near 0, and genes that increase with age have p-values near 1. If there were no age-regulated genes (i.e. the true Bkj=0 for every gene j), then the histogram of one sided p-values would be flat (i.e. have a uniform distribution on the interval from 0 to 1). x-axis shows the p-value and the y-axis shows the number of genes with that p-value. There are 985 genes with a two sided p-value less than .001.

Figure 2. Similar age-regulation in cortex and medulla. A. For every gene, we calculated a one sided p-value that there is a Tissue(i)xAge(i) effect, and plotted the results in a histogram. Genes that show different age-regulation in the cortex or the medulla would be contained in peaks on the left and right parts of the histogram. The figure shows that the number of genes that have different expression levels in the cortex and medulla is about the same or less as would be expected by chance. x-axis shows one sided p-values for Tissue(i)xAge(i), and the y-axis shows the number of genes with that value. There is a systematic under-representation of the edge regions compared to a random sample of uniform random variables due to correlations among the 44928 values computed from 133 samples.
Figure 2B. To show whether aging in the cortex and the medulla are similar, we selected age-regulated genes in the cortex and calculated the one-sided p value for age-regulation in the medulla samples. The histogram shows these selected p-values. The peak at the right contains genes that increase with age in both tissue types.
Figure 2C. Shown is a scatterplot of all 684 genes that are age regulated in either the medulla or the cortex (p<0.001). y-axis is the slope of the expression change in the medulla with respect to age and the x-axis is the slope for the cortex. The solid line has a slope of one and passes through the origin. The dotted line is the least squares fit, with a slope of .58.
Figure 2D. Same as C but for 22 genes that are age-regulated in both the cortex and the medulla (p<0.001).
Figure 3. Expression of the 447 genes as a function of age. Rows correspond to age-regulated genes, ordered from most highly induced to the most highly repressed. Columns correspond to individual patients, ordered from youngest to oldest. Age of certain patients is shown. Left panel refers to data from cortex samples, and right depicts data from medulla samples. The first row shows the chronicity index, from blue (best) to yellow (worst) as indicated in the scale bar. Key genes discussed in the text are marked. Scale shows log2 of the expression level. Navigable version of this figure can be found at http://cmgm.stanford.edu/~kimlab/aging_kidney/explorer.html.
Figure 4. Differential expression in the cortex and the medulla. For each gene, we calculated a one sided p-value for expression differences in the cortex versus the medulla. Shown is a histogram of these p-values. Genes enriched in the cortex are in a peak on the left, and genes enriched in the medulla are in a peak on the right. x-axis indicates p-value and y-axis indicates number of genes.
Figure 5. Developmental profile of the age-regulated genes. Shown are the log2 of the expression levels for 227 age-regulated genes in 26 human tissues, using data form (Su et al., 2002). Rows correspond to genes, columns correspond to human tissues.a, kidney; b, cerebellum; c, whole brain; d, cerebral cortex; e, caudate nucleus; f, amygdala; g, thalamus; h, corpus callosum; i, spinal cord; j, whole blood; k, testis; l, pancreas; m, placenta; n, pituitary gland; o, thyroid gland; p, prostate; q, ovary; r, uterus; s, salivary gland; t, trachea; u, lung; v, thymus; w, spleen; x, adrenal gland; y, liver; z, heart. Scale shows log2 of the expression level. Navigable version of this figure can be found at http://cmgm.stanford.edu/~kimlab/aging_kidney/explorer.html.
Figure 6. Chronicity index of kidney samples. Histology from patient 40 is shown in the left-hand panel, demonstrating a normal glomerulus (G), tubules and interstitial space (T), and ateriole (A), respectively (CHI score of 0). Histology from patient 62 is shown in the right hand panel, demonstrating glomerulosclerosis (g), tubular atrophy and interstitial fibrosis (t), and arterial intimal hyalinosis (a), respectively (CHI score of 10). Hematoxylin and eosin staining of paraffin-embedded sections.
Figure 7. Chronicity index increases with age. Shown is the chronicity index versus age for most of the kidney samples used in this study. The line shows the least squared fit through the data points.
Supplemental Figure 1. Age distribution of medical factors. Each row shows the presence of a medical factor. Age of patients is shown on the y-axis. Only transitional cell carcinoma showed a strong age-bias.
Supplemental Figure 2. (A) Effect of RCC on age. Scatterplot showing age-related slopes using a regression model that includes a term for renal cell carcinoma compared to slopes from a regression model that does not include renal cell carcinoma. The scatterplot shows that RCC does not affect the age-related slopes very much. (B) Effect of TCC on age. Scatterplot showing age-related slopes with and without a term for transitional cell carcinoma. The scatterplot shows that TCC does not affect age-related slopes very much. (C) Effect of tumor size on age. Scatterplot showing age-related slopes with and without a term for size of the tumor. The scatterplot shows that tumor size does not affect age-related slopes very much. (D) Effect of hypertension on age. Scatterplot showing age-related slopes with and without a term for hypertension. The scatterplot shows that hypertension does not affect age-related slopes very much. (E) Effect of systolic blood pressure on age. Scatterplot showing age-related slopes with and without a term for SBP. The scatterplot shows that SBP does not affect age-related slopes very much. (F) Effect of diastolic blood pressure on age. Scatterplot showing age-related slopes with and without a term for DBP. The scatterplot shows that DBP does not affect age-related slopes very much. (G) Effect of diabetes melilitus on age. Scatterplot showing age-related slopes with and without a term for diabetes melilitus. The scatterplot shows that diabetes does not affect age-related slopes very much.
Supplemental Figure 3. Weak correlation between aging in kidney and muscle. Regression slopes of the 447 age-regulated genes from the human kidney and human muscle (Welle et al. 2003) aging datasets are plotted against one another. Co-regulation between these tissues with respect to age would result in correlation between regression slopes. A weak correlation is observed (r = 0.085).
Supplemental Figure 4. No correlations between aging in human and either flies or worms. (A) Regression slopes of age-regulated genes from human kidney and D. melanogaster. Open triangles denote age-regulated genes in human and their orthologs in flies. Open circles denote age-regulated genes in flies and their orthologs in humans. The scatterplot shows the regression slopes from the human kidney and the fly aging datasets (Pletcher et al. 2002)
Supplemental Figure 4(B). Regression slopes of age-regulated genes from human kidney and C. elegans. Open circles denote age-regulated genes in human and their orthologs in worms. Open triangles denote age-regulated genes in worms and their orthologs in humans. The scatterplot shows the regression slopes from the human kidney and C. elegans aging datasets (Lund et al. 2002).