Figure 1.
Flow chart of filtering process.
After data normalization, SLR values were calculated as described in the methods, and replicate pools were randomly paired to obtain 3 complete time courses—referred to as Rep 1, Rep 2 and Rep 3. The first 2 filtering steps were done for each replicate independently, and identified probe sets that were significantly detected on the chip (detection p-value filter; all p-values < = 0.06) as well as significantly differentially expressed (SLR filter; SLR was < = −0.5 or > = 0.5 for any one time point). The overlap filter intersected all three filtered lists to identify the set of probe sets that passed the first 2 filtering steps independently in all replicates. Finally, probe sets were filtered for consistency in pattern (PCC score) and magnitude (ED score) of expression over time across all replicate time courses (consistency filter), where 2 of the 3 PCC scores had to be > = 0.70, and 2 of the 3 ED scores had to be < = 0.60. Those probe sets passing all filtering criteria were considered consistently and significantly differentially expressed across the time course in all 3 biological replicates, and were used for further analysis. Results of this process on each of the data sets described in the text can be found in Tables S1 and S2.
Figure 2.
Histology results demonstrating osteophyte formation over the time course.
Sections from DMM joints obtained 2 weeks (A), 4 weeks (B), 8 weeks (C), and 16 weeks (D) after DMM surgery were stained with safranin-O. Bar = 100 µm. Arrows point to the original medial abaxial joint margin and abaxial osteophytes that formed at this site. These start as chondrophytes at 2 weeks (A) and progressively mature. Histomorphometric measures included the size of the abaxial osteophytes on the medial plateau (E) and axial osteophytes on the lateral plateau (F) for the DMM, sham, and contralateral control groups as indicated. **p<0.0001 vs sham and control; *p = 0.02 vs sham and control by ANOVA.
Figure 3.
Histology results demonstrating articular cartilage lesions over the time course.
Sections from DMM joints obtained 2 weeks (A), 4 weeks (B), 8 weeks (C), and 16 weeks (D) after DMM surgery were stained with safranin-O. Bar = 100 µm. Arrows indicate an early cartilage lesion in the medial tibial plateau at 4 weeks (B) and a more severe lesion at 8 weeks (C) on the medial tibial plateau (lower tissue in image) and the femoral condyle (upper tissue in image). Note advanced articular cartilage loss at 16 weeks. Articular cartilage structure scores (E) and safranin-O scores (F) for the medial tibial plateaus and articular cartilage structure scores for the lateral tibial plateau (G) are shown for the DMM, sham, and contralateral control groups as indicated. *p<0.05 vs sham and control; ??p = 0.01 vs sham; **p<0.01 vs control; #p<0.05 vs control by ANOVA.
Figure 4.
Histologic analysis of chondrocyte death over the time course.
Dead chondrocytes were counted on H&E sections from the medial tibial plateau (A) and lateral tibial plateau (B) from DMM, sham, and contralateral control joints at the various time points after surgery. *p<0.05 vs sham and control joints; **p<0.001 vs sham and control joints by ANOVA.
Table 1.
Histomorphometric measures of subchondral bone area and thickness in DMM and sham control joints.
Figure 5.
Global view of the differentially expressed genes in the sham time course.
A) A heat map illustrates that the majority of differentially expressed genes (406 probe sets for 347 genes), in the sham time course are down-regulated, with few being up-regulated over time. Heat map rows represent genes, and columns represent replicates of each time points. Red indicates up-regulation, blue down-regulation and white indicates no change in gene expression from the respective control (see Methods). B) Gene expression distributions of the sham time course exhibit a uni-modal distribution throughout the experiment; however, the distributions center becomes shifted toward the negative side of zero at 4 weeks, indicating a global down-regulation of these genes over time due to aging. An SLR of zero is marked by a vertical line on the graph, and replicate time courses are colored as follows: replicate 1 = black, replicate 2 = blue, replicate 3 = red.
Figure 6.
Global view of the differentially expressed genes in the DMM time course.
A) The heat map illustrates that the majority of differentially expressed genes (427 probe sets for 371 genes) were up-regulated throughout the DMM time course, except for a small group of genes, which were down-regulated 4 and 8 weeks after surgery (see Figure 5A legend for heat map formatting details). B) The DMM time course exhibits a bi-modal distribution at 4 and 8 weeks after surgery, with the majority of genes being up-regulated at 4 weeks and not different from control at 8 weeks (see Figure 5B legend for formatting details).
Figure 7.
The group of genes differentially expressed in both sham and DMM data sets.
Intersection of the 406 differentially expressed sham time course probe sets and the 427 DMM time course probe sets revealed 97 probe sets (94 genes) found in both data sets. A) A Venn diagram illustrates the number of probe sets in each data set as well as the overlap, and indicates that the majority of genes with OA induced expression changes are not altered in sham joints. B) A heat map of gene expression changes for the 97 intersecting probe sets was created using hierarchical agglomerative clustering and Euclidean distance, with sham expression on the left and DMM expression on the right. The heat map is formatted as in Figures 5A and 6A with the addition of the hierarchical dendrogram on the left and the gene names on the right.
Figure 8.
Heat maps of the largest 16 consensus clusters in the DMM time course.
Rows represent genes and columns are time point replicates in the following order: 2 weeks rep2, rep3; 4-, 8- and 16 weeks rep1, rep2, rep3. Red indicates up-regulation, blue down-regulation and white indicates no change in expression from the sham time-matched control. All heat maps in this figure are colored on the same scale for easy comparison with maximum up-regulation at 1.7 SLR and maximum down-regulation at −1.3 SLR. A variety of expression patterns are present with most including significant up-regulation at 4 weeks. Only four clusters contain down-regulated genes (clusters 5, 7, 10 and 12), and cluster 15 is the only one with significant up-regulation 16 weeks after surgery.
Table 2.
Significant DAVID annotations for the top 16 Gene Clusters.
Figure 9.
Immunolocalization of type III collagen, fibromodulin, and Prelp.
Sections from mice in the 16 week-old sham and DMM groups were immunostained as detailed in the Methods using antibodies to type III collagen (A,B,C), fibromodulin (D,E,F) and Prelp (G,H,I). A, D, and G are 4x views of sham control joints to demonstrate overall tissue distribution. B, E, and H are 20x views of DMM joints and C, F, and I are 20x views of sham control joints. Bar = 1 mm for 4X, 200um for 20X.
Table 3.
Immunostaining results for Prelp, Type III collagen, and Fibromodulin.
Table 4.
Real time PCR results for selected genes showing fold change in DMM joints relative to sham controls.