A rapid cytoplasmic mechanism for PI3 kinase regulation by the nuclear thyroid hormone receptor, TRβ, and genetic evidence for its role in the maturation of mouse hippocampal synapses in vivo

Abstract

Several rapid physiological effects of thyroid hormone on mammalian cells in vitro have been shown to be mediated by the phosphatidylinositol 3-kinase (PI3K), but the molecular mechanism of PI3K regulation by nuclear zinc finger receptor proteins for thyroid hormone and its relevance to brain development in vivo have not been elucidated. Here we show that, in the absence of hormone, the thyroid hormone receptor TRβ forms a cytoplasmic complex with the p85 subunit of PI3K and the Src family tyrosine kinase, Lyn, which depends on two canonical phosphotyrosine motifs in the second zinc finger of TRβ that are not conserved in TRα. When hormone is added, TRβ dissociates and moves to the nucleus, and phosphatidylinositol (3, 4, 5)-trisphosphate production goes up rapidly. Mutating either tyrosine to a phenylalanine prevents rapid signaling through PI3K but does not prevent the hormone-dependent transcription of genes with a thyroid hormone response element. When the rapid signaling mechanism was blocked chronically throughout development in mice by a targeted point mutation in both alleles of Thrb, circulating hormone levels, TRβ expression, and direct gene regulation by TRβ in the pituitary and liver were all unaffected. However, the mutation significantly impaired maturation and plasticity of the Schaffer collateral synapses on CA1 pyramidal neurons in the postnatal hippocampus. Thus, phosphotyrosine-dependent association of TRβ with PI3K provides a potential mechanism for integrating regulation of development and metabolism by thyroid hormone and receptor tyrosine kinases.

Figure 1.

Thyroid hormone (T₃) stimulates PIP₃ production in rat pituitary GH₄C₁ cells. A, Cells that had been serum starved overnight were treated with serum-free medium minus or plus 100 nM T₃ or with serum from thyroidectomized animals without or with 100 nM T₃ (n = 8). Only the latter treatment produces a significant increase in PIP₃ levels, which was measured as the percentage change in the ratio of YFP to CFP emission (FRET) from the indicator after the CFP excitation. B, The increase in FRET was prevented by preincubating cells for 10 minutes in either 3 μM 1–850 to block hormone binding to the receptor (n = 4) or 50 nM wortmannin to block PI3K (n = 4). C, The increase in PIP₃ produced by 100 nM T₃ was sufficient to increase Akt phosphorylation on Ser 473 and on Thr 308 within 5 minutes. A representative blot is shown from one of four separate experiments. IB, immunoblotting.

Figure 2.

Thyroid hormone stimulates PIP₃ production in CHO cells expressing human TRβ1. A, CHO cells expressing TRβ1 show reproducible increases in FRET in response to 0.1 nM T₃ (n = 4), but cells expressing TRα did not respond significantly to 100 nM T₃ over 20 minutes (n = 4). B, The increase in FRET produced by 100 nM T₃ on cells expressing TRβ1 was prevented by preincubating cells for 10 minutes in either 3 μM 1–850 or 50 nM wortmannin (n = 4). C, Immunoprecipitation of FLAG-tagged TRα from CHO cell lysates did not bring down any p85. D, In contrast, immunoprecipitation of the FLAG-tagged TRβ1 pulled down native p85 in the lysate. Subsequent addition of the hormone (far right lane) eliminates the association of p85 with TRβ1. Samples in panels C and D were run on the same gel for ease of comparison. A representative blot is shown from one of three separate experiments. IP, immunoprecipitation; wt, wild type.

Figure 3.

The cellular distribution of TRβ1 under different culture conditions. A, Representative images of fixed CHO cells show the range of distribution patterns of FLAG-tagged TRβ1 proteins, which were stained using a FITC-conjugated antibody against FLAG. B–D, Proportion of cells with FLAG-TRβ1 distribution corresponding to the four distribution patterns in panel A. Each treatment was repeated four times, and each time two to four cells in 16 different areas of each plate (30–60 cells/plate) were counted in each condition. B, In serum-starved cells, TRβ1 is predominantly in the nucleus. C, In cells treated with thyroidectomized serum for 30 minutes, TRβ1 is predominantly in the cytoplasm. D, In cells treated with thyroidectomized serum plus 100 nM T₃ for 30 minutes, TRβ1 is predominantly in the nucleus.

Figure 4.

Tyrosine-dependent binding of TRβ1 to p85. A, Schematic representation of the second TRβ1 zinc finger and the locations of Y147 and Y171. B, Time course of PIP₃ production in response to 100 nM T₃ in CHO cells expressing TRβ1 and in cells that were preincubated with 200 nM SU6656, a Src kinase inhibitor (n = 4). C–E, Mutant and wt TRβ1 proteins were immunoprecipitated from cell lysates with anti-FLAG antibodies as described in Materials and Methods. Levels of coimmunoprecipitated p85 or Lyn were visualized in samples. Samples in panels C–E were run on the same gel for ease of comparison. A representative blot is shown. IP, immunoprecipitation; wt, wild type.

Figure 5.

Both Y147 and Y171 are required for TRβ regulation of PI3K but not for binding to thyroid hormone response elements. A, Time course of PIP₃ production in response to 100 nM T₃ in CHO cells expressing TRβ1 or Y147F or Y171F mutants. B and C, Thyroid hormone-dependent transcription of luciferase reporter constructs containing a minimal promoter downstream of tandem TREs coexpressed in CV-1 cells with wild-type TRα, TRβ1, or the mutants in response to 10 or 100 nM T₃ for 18 hours. The values were normalized to the T₃-independent expression of Renilla luciferase and plotted as bar graphs showing the average and SEM of six measurements from three separate experiments run in duplicate with independent transfections. In all cases thyroid hormone produced a significant increase in transcription, but wild-type and mutant receptors were equally effective. B, The TRE was two contiguous palindromic half-sites, AGGTCATGACCT. The differences between wild type and mutant were not significant (TRβ1 vs Y147F, P = .69; TRβ1 vs Y171F, P = .25). C, The TRE was two direct repeats separated by a four-nucleotide spacer (DR4), AGGTCAnnnnAGGTCA. The differences between wild type and mutant were not significant at either concentration of T₃ (10 nM T₃, TRβ1 vs Y147F, P = .95; 100 nM T₃, TRβ1 vs Y147F, P = .89).

Figure 6.

Mutant and wild-type animals show no differences in thyroid hormone homeostasis or gene regulation. A, Schematic of part of the Thrb locus with relevant restriction sites. Two homology arms were generated by PCR and the A-to-T mutation in exon 3 was introduced. The 5′ arm and 3′ arm were cloned upstream and downstream of a floxed neo gene, respectively. The final ES cell targeting vector was linearized with NotI before electroporation and contained a diphtheria toxin gene (DTA) for negative selection of random recombination. B–D, Data were presented in box (25%–75%) and whisker (5%–95%) plots showing the median value as a horizontal line. The n values are given in the Figure. B, Total T₄ and T₃ levels in serum from wild-type and mutant mice as determined by mass spectrometry are not significantly different (P = .13 for T₄, and P = .87 for T₃. C, Expression levels of Spot14 and Dio1 in liver from wild-type and mutant as measured by qPCR using Rplp0 (36B4) as a standard are not significantly different (P = .13 for Spot14, and P = .21 for Dio1). D, Expression levels of Thrb in the cortex and Tshb in the pituitary from wild-type and mutant mice as measured by qPCR using GAPDH as a standard are not significantly different (P = .19 for Thrb, and P = .88 for Tshb). mut, mutant; wt, wild type.

Figure 7.

Y147F mice have decreased synaptic strength and do not show LTP in response to high-frequency stimulation. A, Input-output curve shows a decrease in the synaptic strength in the mutant animal vs wild type. Insert shows the measured response is blocked almost completely by 10 μM 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), an antagonist of 2-amino-3-hydroxy-5-methyl-4-isoxazol propionic acid glutamate receptors. B, Mutant animals also fail to produce LTP in response to high frequency stimulation (two 1 sec periods of 100 Hz stimuli separated by 20 sec). Insert shows representative traces of the recordings at two time points.