Coilin can form a complex with the U7 small nuclear ribonucleoprotein

Abstract

Coiled bodies (CBs) in the amphibian oocyte nucleus are spherical structures up to 10 microm or more in diameter, much larger than their somatic counterparts, which rarely exceed 1 microm. Oocyte CBs may have smaller granules attached to their surface or embedded within them, which are identical in structure and composition to the many hundreds of B-snurposomes found free in the nucleoplasm. The matrix of the CBs contains the diagnostic protein p80-coilin, which is colocalized with the U7 small nuclear ribonucleoprotein (snRNP), whereas the attached and embedded B-snurposomes contain splicing snRNPs. A few of the 50-100 CBs in the oocyte nucleus are attached to lampbrush chromosomes at the histone gene loci. By coimmunoprecipitation we show that coilin and the U7 snRNP can form a weak but specific complex in the nucleoplasm, which is dependent on the special U7 Sm-binding site. Under the same conditions coilin does not associate with the U1 and U2 snRNPs. Coilin is a nucleic acid-binding protein, as shown by its interaction with single-stranded DNA and with poly r(U) and poly r(G). We suggest that an important function of coilin is to form a transient complex with the U7 snRNP and accompany it to the CBs. In the case of CBs attached to chromosomes at the histone gene loci, the U7 snRNP is thus brought close to the actual site of histone pre-mRNA transcription.

Figure 1

Structure and composition of CBs from Xenopus oocytes (stages IV–VI). (a and b) DIC and immunofluorescence image of a CB stained for coilin (serum C236, fluorescein). Stain is limited to the matrix of the CB and is absent from the attached B-snurposome and the B-like inclusion. (c and d) DIC and fluorescence image of a CB from an oocyte injected 1 d previously with capped, fluorescein-labeled U7 snRNA, showing that U7, like coilin, is strictly limited to the matrix. (e–h) CB stained for coilin (serum C236, fluorescein) and Sm proteins (mAb Y12, Cy3). Overlap shows colocalization of coilin and Sm proteins in the matrix, only Sm proteins in the B-snurposomes and the inclusion. (i–l) CB stained for coilin (serum C236, fluorescein) and the trimethylguanosine (TMG) cap of snRNAs (mAb K121, Cy3). TMG occurs at highest concentration in the matrix but is also strong in the B-snurposomes and the inclusion. Upper arrow in the DIC image points to the inclusion in the CB; lower arrow points to a vacuole in a nucleolus. Note that the shadowing is on opposite sides, demonstrating that the inclusion in the CB is denser than the matrix in which it is embedded, whereas the vacuole is less dense than the body of the nucleolus. (m–p) CB stained exactly as in panels i–l, except that oocyte was injected 1 d previously with an antisense oligodeoxynucleotide against bases 1–16 of U7 snRNA. This treatment results in complete loss of U7 from the GV (Figure 2). Concomitantly, TMG stain in the matrix of the CB is reduced by ∼90% (compare o with k), but stain in the B-snurposome on the surface and the two inclusions is unaffected.

Figure 2

Truncation of U7 snRNA by an antisense oligodeoxynucleotide. Xenopus oocytes were injected with an antisense U7 or a control oligonucleotide. After 2 h or 18 h incubation at 19°C, RNA was purified from single GVs, separated on a polyacrylamide gel, and hybridized with anti-U7 and anti-U2 snRNA probes (ratio 1000:1). Truncated U7 is detected at 2 h, corresponding to loss of 16 nucleotides from the 5′-end. The truncated product was unstable and was not detected at 18 h. U2 snRNA was unaffected by either the anti-U7 or the control oligonucleotide.

Figure 3

(A) Western blot of Xenopus oocyte proteins probed with mAb H1 to determine the intracellular distribution of coilin. Lane 1 contains total proteins from three cytoplasms (Cyt). Lanes 2 and 3 contain soluble nucleoplasmic (Np) and insoluble pellet (Pel) proteins from 10 GVs. Coilin is detected exclusively in the GV, where most is in the soluble nucleoplasmic fraction. (B) Western blot probed with mAb H1 against coilin. Lanes 1 and 2 contain total soluble proteins from 10 GVs and 3 cytoplasms, respectively. Lanes 3 and 4 contain proteins immunoprecipitated by mAb H1 from similar numbers of GVs and cytoplasms. Coilin is efficiently immunoprecipitated from the soluble nucleoplasmic fraction. (C) Western blot similar to that in panel B, except immunoprecipitated with mAb Y12 against the Sm proteins and probed with mAb Y12. Sm proteins are efficiently immunoprecipitated from both cytoplasmic and nucleoplasmic fractions.

Figure 4

Capped ³²P-labeled snRNA transcripts were injected into the cytoplasm of stage VI Xenopus oocytes. After 18 h incubation, GVs and cytoplasm were isolated and immunoprecipitated with either mAb H1 (anti-coilin) or mAb Y12 (anti-Sm). RNA was isolated from the immunoprecipitates and from untreated control fractions and separated on a polyacrylamide gel. (A) Autoradiograph of entire gel from experiment in which wild-type U7 snRNA was injected. Lane 1 contains a sample of the injected snRNA. Lane 2 contains a sample of total cytoplasmic RNA before immunoprecipitation. Lanes 3 and 4 contain RNA immunoprecipitated from cytoplasm by mAbs H1 and Y12, respectively. Lanes 5, 6, and 7 contain a sample of total GV RNA and the RNA immunoprecipitated by mAbs H1 and Y12. (B) Autoradiograph of relevant portion of gels from experiments in which various constructs were injected; lanes as in panel A. U1 and U2 are wild-type U1 and U2 snRNA. U7, U7(U2), and U7(mut) are wild-type U7, U7 with its Sm site replaced by that of U2, and U7 with an unrelated sequence at the Sm site (wild-type U7 lane is the same as in panel A). Note that all constructs except U7(mut) are immunoprecipitated from the GV by mAb Y12, demonstrating that they exist as Sm complexes. However, only wild-type U7 and, to a lesser extent, U7(U2) are immunoprecipitated from the GV by mAb H1, indicating an association with coilin. (C) The lanes in panel B were quantitated with a phosphorimager, and lanes 3, 4, 6, and 7 were plotted as % immunoprecipitated.

Figure 5

Association of endogenous U7 snRNA in the GV with endogenous coilin. Soluble nucleoplasm from 200 GVs was prepared without detergent by centrifuging and discarding the pellet. The discarded pellet contained ∼5–10% of the nuclear coilin and 90% of the U7 snRNA. Soluble coilin in the nucleoplasm was immunoprecipitated with mAb H1 (anti-coilin). A Northern blot of one fifth of the supernate (S) and all of the immunoprecipitate (P) was probed with antisense U7 snRNA (upper panel). Quantitation showed that ∼20% of the soluble U7 snRNA was immunoprecipitated by the anti-coilin antibody. When the blot was reprobed with antisense U6 snRNA, no significant signal was seen in the precipitate (lower panel). BSA-coated beads served as control.

Figure 6

(A) Western blot of proteins purified from the nucleoplasm of Xenopus GVs by affinity chromatography on single-stranded DNA agarose. Coilin was detected by mAb H1. Upon addition of 1 M heparin (*), coilin remained on the column and could be eluted only at high-salt concentrations (0.3–0.7 M NaCl), suggesting that it is a single-stranded nucleic acid-binding protein. RT, Run-through. (B) Western blot of nucleoplasmic proteins purified by affinity chromatography on RNA homopolymers. Coilin bound to poly r(G) and poly r(U) but had no affinity for poly r(C) or poly r(A). Coilin was detected by mAb H1. Mobility of molecular weight markers in kilodaltons.

Figure 7

Soluble nucleoplasm from Xenopus GVs was incubated with poly r(G)-coated agarose beads. After washing with increasing salt concentration (A) or tRNA (B), bound coilin was monitored on a Western blot with mAb H1. The binding of coilin to poly r(G) is resistant to high salt and to an excess of tRNA competitor, as is its binding to single-stranded DNA. Mobility of molecular weight markers in kilodaltons.

Figure 8

Bacterially expressed Xenopus coilin binds to RNA homopolymers. (A) E. coli expressing 6 histidine-tagged Xenopus coilin were lysed and centrifuged to yield supernatant and pellet fractions. Proteins from these fractions were electrophoresed and stained with Coomassie blue (lanes S and P). The pellet fraction was solubilized with 8 M urea and bound to a Ni²⁺ column. Nearly pure 6 his-coilin was recovered from the column (6 his-coilin). (B) Purified 6 his-coilin was incubated with beads coated with RNA homopolymers. A Western blot of proteins that bound to the beads was probed with mAb H1. Purified 6 his-coilin bound to poly r(G) and poly r(U) but not to poly r(C) or poly r(A).

Figure 9

(A) Diagram of full-length myc-tagged human coilin and various deletion constructs. To the right is a summary of poly r(G) binding data for all constructs. Four of these constructs were examined earlier for their ability to target to CBs (Wu et al., 1994). The first 116 amino acids are necessary and sufficient for both poly r(G) binding and for targeting to the CBs. (B) Each of these proteins was expressed in stage VI oocytes by injecting in vitro transcripts from appropriate plasmids into the cytoplasm. After overnight incubation, soluble nuclear proteins were bound to a poly r(G) column, eluted, and electrophoresed. The bound proteins were detected on Western blots with mAb 9E10 against the myc tag. A Western blot for three proteins is shown: full-length coilin, C484 that lacks the first 116 residues at the amino terminus, and A116 that contains only these 116 residues. In each case the first lane (G) contains only material that bound to poly r(G), whereas the second lane (Np) contains total soluble GV proteins. Note that A116 bound to poly r(G) but C484 did not. (C) The same blot reprobed with mAb H1 to demonstrate binding of endogenous coilin.

Figure 10

The poly r(G)-binding region of human coilin compared with the corresponding region of Xenopus coilin (45% identity) and with the RNP consensus sequence from a selection of other proteins. Boxed amino acids at a given residue have similar properties: a, aromatic, h, hydrophobic, z, polar or charged, and b, basic. The RNP1 octamer and RNP2 hexamer are shown in bold. Putative α helices (α1, α2) and β sheets (β1, β2, β3) are based on structure determinations by Nagai et al. (1990) and Wittekind et al. (1992). The RNA-binding region of coilin is at best loosely related to the previously defined RNP consensus sequence.