Structural features and comparison of CPEB-like proteins. (A) All CPEB-like proteins have an amino-terminal region and a carboxy-terminal region containing two RRMs and two zinc-fingers (Zif). CPEB T171, which is not conserved in CPEB2–4, must be phosphorylated for polyadenylation to occur. Among the amino-terminal regions of the CPEB proteins, there is little identity (NS, not significant). Among the RBDs, there is considerable identity. However, mouse CPEB (designated CPEB1 for convenience) is closer to Drosophila CPEB (Orb) than it is to mouse CPEB2; in addition, Drosophila CPEB2 (Orb2) is more similar to mouse CPEB2 than it is to Orb. Mouse CPEB2–4 are nearly identical in the RBDs. (B) Xenopus oocytes were injected with water or RNA encoding myc-CPEB or CPEB3, or myc fused with the amino domain of CPEB3 and the RBD of CPEB. The oocytes were incubated with progesterone and scored for oocyte maturation as assessed by germinal vesicle breakdown (GVBD). A Western blot probed with myc antibody shows the level of the myc fusion proteins in oocytes.

The CPEB4 RBD recognizes RNA secondary structure. (A) Nucleotide changes were introduced into the 1904 sequence and the resulting RNAs were used for gel shifts with the CPEB4 RBD; the relative amount of binding is indicated. (B) Compensatory mutagenesis of M14 RNA. The mfold program generated a predicted secondary structure of M14 RNA. The blue line denotes the nucleotides derived from the constant sequence, the black box indicates the part of the stem used for the mutagenesis, and the red nucleotides indicate the mutated sequences. Gel shifts of these RNAs using the CPEB4 RBD is shown at right. (C) Structure mapping and RNA footprinting. 5′ end-labeled RNA was untreated (lane 1), or subjected to partial alkaline hydrolysis (lane 2), digested with RNase T1 (lane 3), subjected to lead acetate cleavage in the absence or presence of MgCl₂ (lanes 4–6, left panel), or digested with the double-strand-specific nuclease RNase V1 (lane 7, left panel). For RNA footprinting, RNA samples either lacked added protein (lane 4, right), or were 0.25, 0.5, 1, 2, or 4 μM CPEB4 RBD and were treated with the hydroxyl radical (lanes 5–9, right panel). The products were then analyzed on a sequencing gel. The protected region is denoted by a black bar. (D) Predicted RNA secondary structure and nucleotides protected from hydroxyl radical cleavage by CPEB4 RBD; the red box indicates the protected nucleotides. (For colour figure see online version.)

Localization of CPEB3 and CPEB4 in neurons. (A) Co-staining of CPEB3 or CPEB4 with synaptophysin in 21-day-old cultures of hippocampal neurons. CPEB4 is also co-stained with the RNA marker Syto-RNAselect after UV crosslinking and fixation. (B) Detection of CPEB3 and CPEB4 in the PSD fraction. P-CPEB4 refers to putative phosphorylated CPEB4, B refers to the brain extract, S refers to the synaptoneurosome fraction, and syn refers to synaptophysin. (C) Detection of transfected GFP-CPEB3 and GFP-CPEB4 proteins in dendrites of hippocampal neurons co-stained with MAP2.

CPEB binds the GluR2 3′ UTR. (A) Maltose-binding fusion proteins containing the RBD of CPEB or CPEB3 were expressed in bacteria and used for in vitro UV crosslinking with ³²P-labeled 3′ UTRs of Arc and GluR2 mRNAs. (B) Deletions of the GluR2 3′ UTR were used for UV crosslinking with His-tagged CPEB3 RBD. (C) Additional deletions of GluR2 3′ UTR were used for RNA gel shift reactions with His-tagged CPEB3 RBD. (D) Some of the gel shift mixtures also contained unlabeled 1904 or 1904-M1 RNAs as competitors for CPEB3 RBD binding. (E) Structure mapping of GluR2 3′ UTR fragment, S4. 5′ end-labeled S4 RNA was either untreated (lane 1), subjected to partial alkaline hydrolysis (lane 2), RNase T1 digested (lane 3), treated with 100 mM lead acetate in the absence or presence of 10 mM or 100 mM MgCl₂ (lanes 4–6, left panel), or digested with nuclease V1 (lane 7, left panel) and analyzed on a sequencing gel. The sequence corresponding to the uridine-rich region is shown. (F) RNA footprinting of S4. 5′ end-labeled S4 RNA was cleaved by hydroxyl radical in the absence or presence of 0.25, 0.5, 1.0, 2.0, and 4.0 μM CPEB3 RBD (lanes 4–8). Lanes 1–3 is the same as in panel E.

Analysis of CPEB and CPEB4 interaction with RNA. (A) The CPEB4 RBD was mixed with RNA synthesized in vitro from oligonucleotides containing a randomized 25-mer central domain flanked by constant regions used for transcription and PCR. The RNA–protein complexes were collected on filters, eluted, and the cycle repeated for eight rounds before cloning. Representatives of 50 clones are shown. (B) Several SELEX RNAs were synthesized in vitro with ³²P-UTP and used for gel shifts with 500 nM CPEB4RBD. (C) The CPEB4 RBD, containing or lacking the zinc-finger (ZF), was used in gel shifts with SELEX sequence 1904. (D) RNA gel shifts were performed with CPEB, CPEB3, or CPEB4 with CPE-containing RNA or the 1904 sequence. (E) A kinetic analysis of CPEB3 and CPEB4 RBD interactions with the 1904 sequence or the CPE was used to calculate equilibrium dissociation constants (K_d). Similar experiments were performed with CPEB binding to the 1904; the binding of CPEB to the CPE was taken from Hake et al (1998). (F) Various regions of the RBDs of CPEB and CPEB3 were fused to MBP, expressed in bacteria, and used for RNA gel shifts and UV crosslinks with the CPE or the 1904 sequence. The bottom panels show that equal amounts of the MBP fusion proteins were used on all experiments.

CPEB3 and CPEB4 in the brain. (A) Western blots of several mouse tissues probed for CPEB3 and CPEB4. The blot also shows that the antibodies for CPEB3 and CPEB4 do not crossreact and do not recognize CPEB. The calculated molecular sizes of CPEB3 and CPEB4 are ∼74 and 78 kDa, respectively. However, on SDS gels, their apparent molecular sizes are ∼100 and ∼98 kDa, respectively. (B) Immunocytochemistry for CPEB3 and CPEB4 in rat hippocampus, cerebellum, and olfactory bulb. The arrows point to specific regions of immunoreactivity in interneurons (IN) and Purkinje cells (PC) of the cerebellum and Mitral cells (MC) of the olfactory bulb.

Translational repression in neurons. (A) Hippocampal neurons were transfected with mRNAs encoding (1) dimeric MS2CP fused to CPEB3 or amino- or carboxy-terminal truncations of this protein, or MS2CP fused to GFP, and (2) firefly luciferase whose 3′ UTR contained or lacked the MS2 stem–loop, and (3) Renilla luciferase. (B) Western blot showing the expression of the fusion proteins in transfected neurons. The right panel shows an RNA gel shift of transfected proteins binding to RNA containing the MS2 stem–loops. (C) The ratio of firefly/Renilla luciferase activity in neurons (normalized to the MS2CP-GFP control) transfected with the RNAs noted in panel A. Neurons were transfected with mRNA encoding myc-CPEB3 in place of the MS2CP fusions. All the firefly luciferase RNAs contained the MS2 stem–loop in the 3′ UTR.(D) Luciferase values calculated as those in panel C from neurons transfected with RNA encoding firefly luciferase that contained or lacked the MS2 stem–loop and that was treated with NMDA. (E) Semiquantitative RT–PCR of luciferase RNA containing or lacking the MS2 stem–loops following RNA transfection into neurons; some of the neurons were treated with NMDA. (F) Xenopus oocytes were injected with mRNAs encoding fusions between myc and CPEB or CPEB2–4. The CPEB proteins were then immunoprecipitated with myc antibody and probed for myc, to note the level of expression, and the 100 kDa subunit of CPSF. (G) Hippocampal neurons were transfected with firefly luciferase mRNA appended with a 3′ UTR containing the 1904 SELEX sequence, Renilla luciferase mRNA, and either myc-CPEB3 or β-galactosidase. Some of the neurons were also treated with NMDA. Firefly luciferase, normalized to Renilla luciferase, was then determined. Probability (p) values were derived from the Student's t-test.

CPEB3 controls GluR2 mRNA translation. (A) Living cultures of hippocampal neurons were irradiated with UV light, homogenized, and subjected to immunoprecipitation with CPEB3 antibody or IgG in the presence of SDS-containing buffer. The precipitated RNA was extracted after proteinase K digestion and subjected to RT–PCR for Arc, Map2, NF, or GluR2 mRNAs. (B) Endogenous CPEB3 knockdown by RNAi. Cultured hippocampal neurons were infected with a control lentivirus or one expressing a short hairpin RNA against CPEB3 under the control of the U6 promoter. Extracts were then prepared from the cells and analyzed for levels of CPEB3, GluR2, αCaMKII, and synaptophysin. From other cultures, the RNA was extracted and analyzed for GluR2 mRNA by real-time PCR. The bottom panel shows the level of GluR2 in wild-type and CPEB KO hippocampus. Tubulin served as a loading control.