Figure Figure 5.. Overexpression of Pim-1 and Pim-3 by means of episomal supertransfection. (A): Embryonic day 14 (E14)/T (E14/T) embryonic stem (ES) cells were electroporated with pPHPGK (control), pPHPGK-pim1, and pPHPGK-pim3 episomal vectors. ES cell transfectants were subsequently plated at 1 × 103 cells per 100-mm tissue culture dish and further cultured in complete ES cell medium for 7 days. Histograms represent the means and SEs, measured in three replicates, of the percentages of undifferentiated, mixed, and differentiated colonies. (B): E14/T ES cells were electroporated with pPHCAG, pPHCAG-pim1, pPCAGIZ, and pPCAGIZ-pim3 episomal vectors and propagated in culture medium supplemented with hygromycin B (pPHCAG and pPHCAG-pim1) or with zeocine (pPCAGIZ and pPCAGIZ-pim3) for 15 days. The four ES cell transfectants were subsequently plated at 1 × 103 cells per 100-mm tissue culture dish. Cells were exposed to medium without LIF for 0, 12, 24, 36, or 48 hours and were subsequently cultured in complete ES cell medium with 1,000 U/ml LIF for 5–7 days (LIF rescue assay). Curves are based on the variations in the proportion of undifferentiated colonies as a function of the duration of LIF withdrawal in pPHCAG (control) and pPHCAG-pim1 transfectants (left panel), or in pPCAGIZ (control) and pPCAGIZ-pim3 transfectants (right panel) (hereafter termed the four conditions). The proportion of undifferentiated colonies as a function of the duration of LIF removal was modeled using logistic regression (generalized linear model with a binomial family). According to this model, the proportions, p, transformed as logit(p) = ln(p/[1 − p]) are related linearly to the explanatory variables, gene and time. However, we noted a systematic oscillation in the residuals as a function of time for each condition on the transformed scale due to the upper and lower asymptotes differing from 1 and 0, respectively. This deviation from linearity was well described by a fixed sine wave across the four conditions, with frequency f estimated to be 0.025 cycles per hour (a period of approximately 40 hours) over the four conditions. This value was used in the subsequent analysis. The final model required four estimates of intercept and slope (a, b), one for each condition, and one estimate of the sine and cosine coefficients (c, d) for the four conditions (logit(p) = a + b × time + c × sin(2 × π × f × time) + d × cos[2 × π × f × time)]. After taking into account differences in intercept and the residual oscillatory component, the differences between slopes within each groups were significant, indicating that overexpression of pim-1 and Pim-3 decreased the rate of decrease in the proportion of undifferentiated colonies as a function of the duration of LIF starvation (Pim-1 vs. control: diff = 0.02, SE (standard error) = 0.003, z = 5.75, p < .001; Pim-3 vs. control: diff = 0.009, SE = 0.004, z = 2.08, p < .05). The figure shows the data on the logit scale after normalization by the intercepts and removal of the oscillatory component with the best-fitting straight lines for each condition. Abbreviations: hrs, hours; LIF, leukemia inhibitory factor.
Download figure to PowerPoint