Tissue-Specific TAFs Counteract Polycomb to Turn on Terminal Differentiation

See allHide authors and affiliations

Science  04 Nov 2005:
Vol. 310, Issue 5749, pp. 869-872
DOI: 10.1126/science.1118101


Polycomb transcriptional silencing machinery is implicated in the maintenance of precursor fates, but how this repression is reversed to allow cell differentiation is unknown. Here we show that testis-specific TAF (TBP-associated factor) homologs required for terminal differentiation of male germ cells may activate target gene expression in part by counteracting repression by Polycomb. Chromatin immunoprecipitation revealed that testis TAFs bind to target promoters, reduce Polycomb binding, and promote local accumulation of H3K4me3, a mark of Trithorax action. Testis TAFs also promoted relocalization of Polycomb Repression Complex 1 components to the nucleolus in spermatocytes, implicating subnuclear architecture in the regulation of terminal differentiation.

Male germ cells differentiate from adult stem cell precursors, first proliferating as spermatogonia, then converting to spermatocytes, which initiate a dramatic, cell type–specific transcription program. In Drosophila, five testis-specific TAF homologs (tTAFs) encoded by the can, sa, mia, nht, and rye genes are required for meiotic cell cycle progression (1, 2) and normal levels of expression in spermatocytes of target genes involved in postmeiotic spermatid differentiation (3). Requirement for the tTAFs is gene selective: Many genes are transcribed normally in tTAF mutant spermatocytes. Tissue-specific TAFs have also been implicated in gametogenesis and differentiation of specific cell types in mammals (4, 5). In addition to action with TBP (TATA box–binding protein) in TFIID, certain TAFs associate with HAT (histone acetyltransferase) or Polycomb group (PcG) transcriptional regulatory complexes (6, 7). To elucidate how tissue-specific TAFs can regulate gene-selective transcription programs during development, we investigated the mechanism of action of the Drosophila tTAFs in vivo.

The tTAF proteins were concentrated in a particular subcompartment of the nucleolus in primary spermatocytes (Fig. 1). Expression of a functional green fluorescence protein (GFP)–tagged genomic sa rescuing transgene revealed that expression of Sa-GFP turned on specifically in male germ cells soon after initiation of spermatocyte differentiation and persisted throughout the remainder of the primary spermatocyte stage, disappearing as cells entered the first meiotic division (Fig. 1A). Some Sa-GFP was detected associated with condensing chromatin (arrowheads in Fig. 1, D and E). However, most Sa-GFP localized to the nucleolus (Fig. 1, C to E), in a pattern complementary with Fibrillarin, which marks a fibrillar nucleolar subcompartment (Fig. 1, J and K). Staining with antibodies against endogenous Sa, Can, Nht, or Mia proteins showed similar temporal expression and nucleolar localization in primary spermatocytes, consistent with collaborative function of the tTAFs (Fig. 1, F to K) (8). In contrast, the generally expressed sa homolog TAF8 and its binding partner TAF10b were excluded from the nucleolus (8).

Fig. 1.

Testis TAFs are expressed only in spermatocytes and concentrate in a subcompartment of the nucleolus. (A and B) Apical region of wild-type testis: (green) Sa-GFP; (red) anti-Fibrillarin, nucleolar marker in all cells. (Sg) spermatogonia, (eSc) early spermatocytes, (Sc) spermatocytes. (Arrowhead) Onset of Sc differentiation; (arrow) onset of Sa-GFP expression; (bracket) cells entering division for meiosis I. (C to E) Live spermatocytes from sa-GFP testis squash (C) phase contrast; (D) DNA stained with Hoechst; (E) Sa-GFP. (Arrowheads) Partially condensed autosomes; (arrow) nucleolus. (F to H) Identical field of fixed spermatocytes stained with (F) anti-Sa, (G) anti-Mia, and (H) anti-Myc (detecting expression of a can-6myc genomic rescue transgene). (I) Single spermatocyte nucleus immunostained with (green) anti-Can, (red) anti-Fibrillarin, and (blue) DAPI (4′,6-diamidino-2-phenylindole). Dotted outline: nucleolus. (J and K) Enlarged spermatocyte nucleoli: (red) anti-Fibrillarin; (green) (J) anti-Can, (K) Sa-GFP. Bar: 4 μm.

Several components of the Polycomb Repression Complex 1 (PRC1) transcriptional regulator appear in the nucleolus in spermatocytes, coincident with tTAF expression and dependent on tTAF function. Polycomb (Pc) protein expressed from a Pc-GFP genomic transgene localized on chromatin, but in addition became concentrated in the nucleolus in primary spermatocytes (Fig. 2, A to C) (9). Both Pc-GFP and staining of endogenous protein with antibody against Pc (anti-Pc) revealed localization to the same nucleolar subcompartment as the one containing tTAFs (Fig. 2, A to F). Recruitment of Pc to the nucleolus exactly coincided with onset of expression of the tTAFs after early G2 phase in spermatocytes (Fig. 2, G to I; fig. S1). Relocalization of Pc depended on wild-type tTAF activity: Pc localized to chromatin but was not concentrated in the nucleolus in tTAF mutant spermatocytes (Fig. 2, J to L; fig. S2) (8). Two other components of the PRC1 core complex, Polyhomeotic (Ph) and Drosophila Ring protein (dRing) (10), also became concentrated in the nucleolus in primary spermatocytes dependent on tTAF function (fig. S2). Failure of PRC1 components to localize to the nucleolus in tTAF mutants was not caused by nucleolar loss because Fibrillarin staining appeared normal in the mutants (Fig. 2J). H3K27me3 laid down by action of the PRC2 complex acts as a docking site for the Pc chromodomain to recruit PRC1 and block transcription initiation (11, 12). H3K27me3 localized on chromatin in spermatocytes, along with Pc. However, no H3K27me3 was detected in the nucleolus in spermatocytes (Fig. 2, M to O), suggesting that PRC1 components may be recruited to the nucleolus by a different mechanism independent of chromatin.

Fig. 2.

Recruitment of Pc to the nucleolus in primary spermatocytes requires tTAFs. Spermatocyte nuclei showing localization of Pc-GFP (green). (Arrow) Nucleolus; (arrowheads) partially condensed autosomes. (A to F) Wild-type: (A) anti-Fibrillarin; (C) merge: complementary staining; (D) anti-Sa; (F) merge: overlapping staining. (G to I) Wild-type spermatocytes at the transition to tTAF expression; spermatocytes in early G2 are to the left of the dotted line: (G) anti-Sa; (I) merge. (J to L) sa mutant spermatocytes: (J) anti-Fibrillarin; (L) merge: Fibrillarin present, but Pc-GFP absent from the nucleolus. (M to O) Wild-type: (M) anti-H3K27me3; (N) Pc-GFP; (O) merge: H3K27me3 on chromatin, but not in the nucleolus. Bar: 4 μm.

The tTAFs are required for activation of robust transcription of several spermatid differentiation genes, whereas the PcG proteins are known to mediate transcriptional repression. Chromatin immunoprecipitation (ChIP) suggested that the tTAFs might allow robust transcription of spermatid differentiation genes in part by counteracting repression by Pc, perhaps causing dissociation of PRC1 from cisacting control sequences at target genes.

ChIP from wild-type testes using anti-Sa revealed enrichment of tTAF binding at three different known target genes (fzo, Mst87F, and dj), compared with binding at intergenic regions 10 to 20 kb away or at a tTAF-independent gene expressed in the same cell type (cyclin A or sa itself, Fig. 3A), suggesting that the tTAFs are in occupancy at target genes. Real-time polymerase chain reaction (PCR) analysis revealed ∼10-fold enrichment of Sa at a target (mst87F) compared with a non-target gene (sa) (fig. S3).

Fig. 3.

Testis TAFs reduce binding of Pc to tTAF-dependent target genes. Immunoprecipitates from ChIP using anti-Sa, anti-Pc, or anti-H3K4me3 were tested for enrichment of three tTAF target genes (fzo, mst87F, and dj) and two nontarget genes (cycA and sa). (Pr) PCR product amplified by primers flanking promoter. (U) PCR product from primers flanking an intergenic sequence 10 to 20 kb away from the promoter. Numbers below indicate PCR cycle multiple of 5 at which the product first became visible. (Numbers in italics: No band detected after the indicated number of PCR rounds.) (A) Anti-Sa ChIP on wild-type testes (cycle 35). (B) Anti-Pc ChIP on can mutant testes (cycle 40). (C) Anti-Pc ChIP on wild-type testes (cycle 50). (D) Anti-Sa, anti-Pc, and anti-H3K4me3 ChIP on wild-type and can mutant testes (cycle 35), with PCR primers tiling across the fzo region. (E) Anti-H3K4me3 ChIP on wild-type and can mutant testes (cycle 35). Products from the U-region of mst87F and Pr-region of cycA in wild-type remained in can, independent of tTAF activity, suggesting recruitment by an alternative mechanism. In all cases, no obvious band was visible at the same PCR cycle in mock immunoprecipitation experiments performed in parallel.

ChIP analysis also revealed that Pc protein bound to tTAF-dependent target genes in tTAF mutant testes, and that wild-type function of the tTAFs reduced Pc binding (Fig. 3, B and C). ChIP with anti-Pc from can mutant testes preferentially precipitated the three tTAF target promoters, compared with intergenic regions or promoters from two different nontarget controls (Fig. 3B). Quantification by real-time PCR showed more than 50-fold enrichment of Pc at the target gene mst87F compared with the tTAF-independent control sa (fig. S3). In contrast, relative occupancy of Pc at the tTAF targets was not significantly different from that at the nontargets in wild-type testes (Fig. 3C, fig. S3).

The tTAFs may act near the promoter of target genes (fzo, Fig. 3D) to allow expression by directly or indirectly reducing nearby binding of PRC1. ChIP using primer pairs across the promoter region of fzo revealed that the tTAF enriched most strongly for sequences just upstream of the transcription start site. In contrast, Pc-containing protein complexes (in tTAF mutant testes) enriched for a broader distribution, including sequences near and downstream of the transcription start site, consistent with localization of Pc at Ultrabithorax (Ubx) locus in wing discs and on the hsp26 promoter in vivo (12, 13).

Binding of the tTAFs at target promoters may allow expression through recruitment or activation of the Trithorax group (TrxG) transcriptional activation complex, which often acts in opposition to repression by PcG proteins (14). Trx, like its mammalian homolog MLL, creates an H3K4me3 epigenetic mark (15). ChIP from wild-type testes revealed H3K4me3 at or near the promoter regions of the three tTAF targets tested, as well as at nontargets (Fig. 3E). Analysis using primer pairs across the tTAF target fzo region revealed that H3K4me3 associated most strongly with sequences spanning the promoter (Fig. 3D). In contrast, ChIP with anti-H3K4me3 from can mutant testes did not enrich for the tTAF target promoters (Fig. 3, D and E). Quantitative PCR revealed 36-fold enrichment of the promoter region of the tTAF-dependent mst87F gene by ChIP for H3K4me3 in wild-type compared with can mutant testes (fig. S3).

Consistent with the presence of H3K4me3 at target promoters in wild-type testes, trx function appeared to be required for continued expression of two different kinds of tTAF-dependent targets. Boule triggers the G2/M transition in meiosis I by allowing translation of twine (16) and requires tTAFs for protein accumulation (Fig. 4A), setting up a cross-regulatory mechanism so that meiotic cell cycle progression awaits expression of terminal differentiation genes (3, 17). When temperature-sensitive trx1 flies grown at permissive temperature were shifted to nonpermissive temperature as adults, the Boule protein level in mutant testes substantially decreased over time at nonpermissive temperature compared with the level in wild-type flies shifted in parallel or trx1 flies held at permissive temperature (Fig. 4, B and C) (8). Likewise, analysis of mRNA levels by semiquantitative PCR revealed a ∼40% decrease in transcript level for the tTAF target gene fzo, but not for the tTAF-independent gene cyclin A, in testes from trx1 mutant flies shifted to non-permissive temperature compared with the level in testes from similarly treated wild-type flies (Fig. 4, D and E).

Fig. 4.

Expression of tTAF-dependent target genes requires trx function. (A) Accumulation of Boule protein depends on tTAF function. Western blot of wild-type and tTAF mutant testes probed with anti-Boule. Loading control in (A) and (B): α-Tubulin (α-Tub). (B) Western blot of Boule levels over time after shift to 29°C in trx1 or wild-type (WT) testes. (C) Quantitation of (B) from three independent experiments. Boule/α-Tub were normalized to 1.0 for Day 1. (D) Semiquantitative reverse transcription (RT)-PCR for transcript level of the tTAF target gene fzo and the nontarget gene cycA in wild-type and trx1 testes 5 days after temperature shift. Loading control: b1-Tub (cycle 25, within the linear range for input mRNA). (E) Quantitation of (D) from three independent experiments; fzo/b1-Tub and cycA/b1-Tub were normalized to 1.0 for wild-type testes.

In summary, occupancy of tTAFs and Pc at target promoters appeared to be mutually exclusive in wild-type and tTAF mutant spermatocytes, suggesting that the tTAFs may turn on target gene expression by counteracting repression by Polycomb, either directly or indirectly reducing Pc binding and allowing local action of Trx (fig. S4). Loss of function of Pc in marked clones of homozygous mutant cells did not restore terminal differentiation in a tTAF mutant background (8), suggesting that in addition to counteracting repression by Pc, tTAFs may also be required at the promoter region independent of Pc, possibly to recruit Trx or other cofactors for transcription activation. Transcriptional derepression by sequestration of PcG proteins has been observed during HIV-1 infection, when the viral Nef protein recruits the PRC2 component Eed to the plasma membrane (18). Likewise, the tTAFs may sequester Pc to the nucleolus. The tTAFs Nht, Can, and Mia are homologs of the generally expressed TAF4, TAF5, and TAF6, which were previously found as stoichiometric components of the PRC1 complex purified from fly embryos (7), raising the possibility that the tTAFs might associate with a population of Pc-, Ph-, and dRing-containing complexes in the nucleolus. If so, interactions in the nucleolus are likely to differ from interactions at the promoters of target genes, because the ChIP results indicate immunoprecipitation of tTAFs without Pc (Fig. 3).

The PcG and TrxG proteins act to maintain cell fates set during embryogenesis throughout development (19). Emerging evidence indicates that PcG and TrxG complexes also play critical roles in decisions between proliferating precursor cell fates and terminal differentiation, for example, in the blood cell lineages. In particular, the mammalian PcG protein Bmi-1 promotes proliferation and blocks differentiation of normal and leukemic stem cells (20), and is required for establishment or maintenance of adult hematopoietic stem cells in mouse (21). Transcriptional silencing by PcG action may allow self-renewal and continued proliferation of precursor cells by blocking expression of terminal differentiation genes. This repression must be reversed to allow production of terminally differentiated cells, whereas failure may allow overproliferation of precursors and eventually cancer. Although central for both normal development and understanding the genesis of cancer, little is known about the mechanisms that reverse such epigenetic silencing to allow expression of the terminal differentiation program. Our findings in the male germ line provide an example of how cell type– and stage–specific transcriptional regulatory machinery, turned on as part of the developmental program, might allow onset of terminal differentiation by counteracting repression by the PcG and highlight the importance of subnuclear localization in regulation of transcriptional regulation.

Supporting Online Material

Materials and Methods

Figs. S1 to S4


References and Notes

Stay Connected to Science

Navigate This Article