Determinants of coactivator LXXLL motif specificity in nuclear receptor transcriptional activation

Abstract

Ligand-dependent activation of gene transcription by nuclear receptors is dependent on the recruitment of coactivators, including a family of related NCoA/SRC factors, via a region containing three helical domains sharing an LXXLL core consensus sequence, referred to as LXDs. In this manuscript, we report receptor-specific differential utilization of LXXLL-containing motifs of the NCoA-1/SRC-1 coactivator. Whereas a single LXD is sufficient for activation by the estrogen receptor, different combinations of two, appropriately spaced, LXDs are required for actions of the thyroid hormone, retinoic acid, peroxisome proliferator-activated, or progesterone receptors. The specificity of LXD usage in the cell appears to be dictated, at least in part, by specific amino acids carboxy-terminal to the core LXXLL motif that may make differential contacts with helices 1 and 3 (or 3') in receptor ligand-binding domains. Intriguingly, distinct carboxy-terminal amino acids are required for PPARgamma activation in response to different ligands. Related LXXLL-containing motifs in NCoA-1/SRC-1 are also required for a functional interaction with CBP, potentially interacting with a hydrophobic binding pocket. Together, these data suggest that the LXXLL-containing motifs have evolved to serve overlapping roles that are likely to permit both receptor-specific and ligand-specific assembly of a coactivator complex, and that these recognition motifs underlie the recruitment of coactivator complexes required for nuclear receptor function.

Figure 1

LXXLL helical motif requirements for actions of ER. (A) Map of NCoA-1/SRC-1 with LXXLL helical motif (LXDs) indicated, and of CBP, with a portion of a LXXLL motif in the NCoA-1/p/CIP interaction domain. (Arrows) LXXLL motifs; (bottom) sequences of these motifs are compared. (B) Effects of mutation of LXD1, LXD2, or LXD3 (LXXLL → LAAAA) on ability to rescue 17β estradiol (E₂)-dependent ER activation abolished by nuclear microinjection of anti-NCoA-1/SRC-1 affinity-purified IgG. (C) Western blot analysis of expression of NCoA-1/SRC-1, wild type (wt), or the indicated mutations, in extracts of cells transfected with specific IgG. No significant differences in protein levels were observed for the mutations studied.

Figure 2

Differential role of LXD motifs in actions of ER (A), PR (B), RAR (C), or TR (D) receptors. In each case, ligands were added at 10⁻⁶ m, and reporters were under control of the appropriate response element (Torchia et al. 1997). (ERE) Estrogen-responds element; (PRE) progesterone-response element; (RARE) retinoic acid-response element; (TRE) thyroid hormone-response element. Rat-1 cells were microinjected with anti-NCoA-1/SRC-1 IgG and the CMV-expression vectors encoding the indicated proteins (Torchia et al. 1997). In addition to the mutations in leucine residues +4 and +5 of each motif, a deletion of 30 amino acids between LXD2 and LXD3 (LXD2S3mut) were created, leaving the 10 amino acids immediately flanking the LXD intact. Ligands were retinoic acid (RA), triiodothyronine (T3), 17β-estradiol (E₂), or progesterone (Prog). Where indicated, receptors were also expressed; similar results were obtained in at least three independent experiments with >300 cells microinjected for each data point.

Figure 3

Requirements of LXD domains of NCoA-1/SRC-1 for activation of PPARγ on a response element (PPARγRE)-dependent promoter (AOX/LacZ; Korzus et al. 1998), repeated with different ligands. Nuclear microinjection of at least 300 Rat-1 cells was performed for each data point with αNCoA-1/SRC-1 IgG, and CMV-expression plasmids encoding wild-type NCoA-1/SRC-1 (NCoA1_wt) or NCoA-1/SRC-1 with point mutations in LXXLL motifs 1, 2 or 3 (LXD1mut, LXD2mut, and LXD3mut). Ligands used were TGZ (10⁻⁶ m), 15-deoxyΔ^12,14- prostaglandin J2 (PGJ2, 10⁻⁶ m), or indomethacin (Ind, 10⁻³ m). Results were repeated in three separate experiments; mean ± s.e.m.

Figure 4

Carboxy-terminal flanking regions dictate specificity of LXXLL domain function. Nuclear microinjection studies in Rat-1 cells were performed with NCoA-1/SRC-1 proteins in which LXD2-flanking residues (−1 → −8) or (+6 → 13) were mutated to alanine and were evaluated RAR, TR, and ER, as shown in A, B, and C, respectively. Results of the average ± s.e.m. of two sets of nuclear microinjected cells; three independent experiments gave similar results. (D) Avidin–biotin DNA complex assay with thrombin-cleaved, bacterially expressed RARβ and RXRα proteins were bound to biotinylated direct repeat core sequence spaced by 5 bp (DR+5) oligonucleotide, and the NRID of NCoA-1/SRC-1 (amino acids 700–763) was prepared as a ³²P-labeled bacterial protein. Competition was assessed with wild-type LXD2 (21-mer) synthetic peptides (LXD2wt), or peptides containing alanine substitutions in the indicated amino acids used in excess (1 μm) to compete for binding. Binding of the NCoA-1 NRID indicates the efficiency of the peptide competition; with more binding indicating loss of function, as observed with mutation of L4 and L5. (E) Similar analysis performed with a bacterially expressed GST–ER carboxy-terminal protein, to evaluate the effects of residue substitution. Mutation of +8 through +12 caused considerable loss of function (i.e., less competition).

Figure 5

Identification of critical carboxy-terminal residues in the NCoA-1/SRC-1 LXD2 motif for function in transcriptional activation by ER from an expression plasmid (A), or endogenous RAR (B). In each case, the ability of NCoA-1/SRC-1 wild-type, LXD2 mutant (LAAAA), or NCoA-1 proteins with single amino acid alanine substitutions [LXD2(+6)mut through LXD2(+13)mut] were evaluated by use of the single cell microinjection assay. The critical amino acids differed in the case of ER and RAR; mean ± s.e.m.

Figure 6

Mapping of critical carboxy-terminal residues in the NCoA-1/SRC-1 LXD2 required for function in transcriptional activation by PPARγ with either TGZ (10⁻⁶ m) (A), or BRL49653 (10⁻⁶ m) (B) as ligand. The entire series of alanine substitutions was evaluated; (*) critical amino acid residues identified.

Figure 7

Role of LXDs in the CBP/p300-interaction domain of NCoA-1/SRC-1 in interaction and receptor activation function. (A) The LXD motifs (4, 5; Fig. 1A) were mutated to place alanines in positions 2, 3, 4, 5 alone, or together, and evaluated for function on RAR-dependent gene activation. Mutations of LXD4 or LXD4 and LXD5 abolished the ability of SRC-1/NCoA-1 to function in retinoic acid-dependent activation events in Rat-1 single cell nuclear microinjection studies. (B,C) The requirement for the LXD4 and LXD5 in the CBP/p300-interaction domain of NCoA-1/SRC-1 is demonstrated for coactivation of the TR and PPARγ. (D) Role of LXDs in the interactions between ³⁵S-labeled NCoA-1/SRC-1 CBP-interaction domain and CBP (wild type) or CBP in which the CBP LXD is mutated (LXXLL → LAAAA; CBP LXDmut). The mutation of LXD4 and LXD5 virtually abolished interactions by GST pull-downs, but mutation of the CBP LXXLL motif in the interaction domain did not affect interactions. (E) Gal4–NCoA-1 (896–1200) fusion protein activates transcription from the UAS p36 promoter (Torchia et al. 1997) and was blocked by addition of anti-CBP IgG (Kamei et al. 1996). (F) Effect of deletion of the NCoA-1 interaction domain of CBP on function of CBP in RAR activation. Anti-CBP IgG (Kamei et al. 1996) was used to block RAR activation, and CMV-expression vectors encoding wild-type CBP, CBPΔN (Δ 2–468), or CBP ΔNCoA-1 (Δ 2098–2163) were evaluated for their ability to rescue. The three CBP protein variants were expressed at comparable levels in transcripted cells as detected by anti-Flag IgG (top). (G) Predicted structure of the CBP region interacting with SRC-1/NCoA-1, with helix 3 as a predicted hydrophobic helix. (H) Requirements of each helix were tested by mutation to break the helix (helix 1, QDLL → PDPG; helix 2, QQQV → PQPG; helix 3, FIKQ → PIPG; helix 4, NLNA → PLPG), or by helix removal (ΔH4, amino acids 2058–2133; ΔH1, amino acids 2078–2163). Regions were tested in vitro for their ability to interact with amino acids 900–970 of NCoA-1/SRC-1, ³⁵S-labeled by in vitro transcription and translation.

Figure 8

Model of SRC-1 LXD2 (−6) through (+14) bound to BRL49653-liganded PPARγ LBD based on the co-crystal structure with SRC-1 amino acids 623–710 (Nolte et al. 1998). A ribbon drawing of the LXD2 motif of human SRC-1 is shown in yellow with the human PPARγ LBD shown in green. When the electron density maps of the co-crystal structure of the SRC-1 heterodimer with liganded PPARγ (Nolte et al. 1998) are examined and modeled, the +12, +13 amino acids form weak interactions at the amino terminus of helix 1; the +6 side chains contact the carboxyl terminus of helix 3; amino acids +9 and +10 form interactions with the small helix (H3′) between helices 3 and 4.