Distinct Localization of E2f3 Isoforms
As noted above, E2f3 and Rb staining in SACs was both nuclear and cytoplasmic (Figure 5A and 5B). The antibody that worked in immunostaining recognizes a C-terminal region and thus, does not distinguish a/b isoforms. To our knowledge, the subcellular location of E2f3 isoforms has not been determined in any cell type. To verify the dual locations of E2f3 and to determine which isoforms were present in retina, we analyzed nuclear and cytoplasmic fractions by Western blot at different times during development. Analysis with the pan-E2f3 antibody (sc-878, Santa Cruz Biotechnology) detected a 55-kD E2f3a species and a 40-kD E2f3b polypeptide (Figure 6). To confirm that the upper species in our retinal lysates was E2f3a, we exploited novel mice that lack E2f3 exon 1a and thus express E2f3b exclusively (R. O. and G. L., unpublished data). The genotyping strategy is discussed in detail later and is outlined in Figure 7A. Western analysis confirmed that the upper band was absent in E2f3a−/− mice (Figures 6 and S8). Consistent with the drop in E2f3-expressing cells during WT retinal maturation (Figure 5A), the total amount of E2f3a was less at P18 compared to P0 (Figure 6). E2f3b was present in similar amounts at both time points. At P0 and P18, E2f3a was present in both nuclear and cytoplasmic fractions, but in marked contrast, E2f3b was exclusively nuclear at both times (Figure 6). Two closely migrating E2f3a bands were detected, more clearly evident at P18, of which the faster migrating species was dominant in nuclear and the slower species was dominant in cytoplasm (Figure 6). The identity of both as E2f3a species was confirmed by their absence in the P18 E2f3a KO retina (Figure S8). Analysis of Pou4f2, a nuclear transcription factor expressed in ganglion cells, showed that nuclear proteins had not contaminated the cytoplasmic fraction, and analysis of Slc18a3, a cytoplasmic SAC marker, confirmed that the reverse had also not occurred (Figure 6). These data show, to our knowledge for the first time, that E2f3a and E2f3b exhibit distinct patterns of subcellular distribution, and raise the possibility that E2f3a localization may be regulated by as yet unknown post-translational modifications.
Figure 6  Subcellular Distribution of E2f3 Isoforms and Other Cell Cycle Proteins in the Developing Retina
Nuclear and cytoplasmic extracts from an equivalent number of retinal cells from mice of the indicated genotypes and ages were analyzed by Western blotting to detect the indicated proteins. Lysate from E2f3a−/− mice was used as a control to confirm the location of E2f3a protein. C, cytoplasmic extracts; N, nuclear extracts.
Figure 7  The E2f3a Isoform Drives the Differentiation Defect in Rb KO SACs
(A) Schematic diagrams of the mouse WT, E2f3a−/−, and the Cre-recombined floxed E2f3 loci (indicated here as E2f3−/− for simplicity). E2f3a−/− mice lack most of E2f3 exon 1a and part of intron 1a (red dotted box). Arrows indicate PCR primers. Genotyping of an E2f3a+/− mouse is shown on the right.
(B) RT-PCR detection of E2f3a and E2f3b mRNA in the retina. The sequences of primers are 1aF (5′-GCCTCTACACCACGCCACAAG-3′), 1bF (5′-CGGAAATGCCCTTACAGC-3′), and 4R (5′-CTCAGTCACTTCTTTGGACAG-3′). WT retina expresses both E2f3a and E2f3b mRNA. As expected, E2f3a−/− retina lacks E2f3a mRNA and still expresses E2f3b mRNA. E2f3−/− retina lacks full-length E2f3a and E2f3b mRNAs, and instead expresses a truncated mRNA lacking exon 3.
(C) Real-time RT-PCR analysis of E2f genes in P8 retinas of the indicated genotypes. Error bars represent SD of measurements from three animals, and asterisks indicate a significant difference between WT and the indicated genotypes (*, p <0.05; **; p <0.01; ANOVA and Tukey HSD test).
(D) Rescue of Rb KO SACs by E2f3a deletion. Horizontal retinal sections of the indicated genotypes and ages were stained for nuclei (DAPI, blue), M-phase (PH3, green), and the SAC marker Slc18a3 (red). E2f3a deletion does not suppress ectopic division, but rescues the SAC defect. Scale bars are 50 μm.
M, molecular size marker. We also examined the distribution of other cell cycle regulators during retinal development. Like E2f3a, Rb was present in both the WT cytoplasm and nucleus at P0, but at P18, when the levels of Rb had increased, it was primarily nuclear (Figure 6). A very faint cytoplasmic Rb signal was evident at P18, which is consistent with Rb staining of SAC processes (Figure 5B), and with the very small proportion of SACs in the retina [38]. E2f1 was also detected in both nuclear and cytoplasmic fractions, although unlike E2f3a it was predominantly nuclear both at P0 and P18 (Figure 6). The E2f dimerization partner, Tfdp1, which lacks a nuclear localization signal [54], was primarily cytoplasmic at both P0 and P18, and the Cdk inhibitors Cdkn1a and Cdkn1b showed a similar pattern of distribution (Figure 6). Thus, among the cell cycle regulators we examined, most showed bivalent distribution, and E2f3b was unusual in its solely nuclear compartmentalization.