Figure 3. Biomechanical response profile of human primary osteoblast from IS patients with elongated cilia vs. control.

We used RT-qPCR to examine the effect of oscillatory fluid flow on gene expression after 4, 8, and 16 hours of stimulation. Gene expression in every sample has been normalized to two endogenous controls (GAPDH and HPRT). The 0 h of every sample has been defined as its calibrator. The graphs represent the fold changes at each time point, compared to the calibrator. For each gene, the two groups (control and IS) were compared at each time point using a pairwise t-test. In addition, the expression at 0 h for IS was compared to each of the other time points (4 h, 8 h and 16 h) using separate pairwise t-tests. For a post hoc Bonferroni analysis, the maximum number of comparisons per gene is 6, three comparisons per question (i.e. three comparisons per family of test). Even if we consider each gene as a family (i.e. six comparisons), using this formula (FWER = 1 − (1 − α)M⁸⁸) the family-wise error rate (FWER) would be: 1−(1–0.05) 6 ≈ 1 − 0.73 ≈ 0, 26. Solving the Bonferroni (0.26/6) new α would be 0.043 which does not affect our findings. For example CTNNB1 results at 4 (p = 0.03) and 8 hours (p = 0.008) will still be significant. It is the same case for POC5 4 h (p = 0.01) but ITGB1 with a p = 0.047 will not pass the test. Overall, the multiple test error is not significant in our analysis and it will not change the results. Genes were chosen based on the following characteristics: Biomechanically responsive genes in bone tissue: BMP2, PTGS2 (COX2), RUNX2, SPP1 (OPN); role in mechanotransduction through cilia: ITGB1, ITGB3; indicator of Wnt pathway activity: CTNNB1; GSK3B, AXIN2, INVS; indicator of Hedgehog signaling: PTCH1; or implicated in an IS study: POC5. Each error bar is constructed using 1 standard error from the mean. n = 8, (4 IS vs. 4 controls) for all genes except CTNNB1, where n = 4 (2 IS vs. 2 controls).