HESX1-interacting proteins in yeast. Transformation of PJ69-4A yeast strain (b), and yeast strains expressing either Gal4DBD–HESX1 (aa 1–185) (a), Gal4DBD–HESX1 (aa 1–107) (d), Gal4DBD alone (c), with plasmids encoding HESX1-interacting proteins fused to VP16 activation domain. Specific interactions are observed with DNMT1, SRFBP1, LONP2, SAFB1 and ZFP592, as shown by the growth of yeasts transformed only with plasmids containing HESX1 sequences (a, d). However, RNF2 is a false positive, as it is able to bind Gal4DBD and to activate the expression of the selecting genes in the absence of HESX1 sequences.

Expression analyses of the HESX1 partners in the forebrain and Rathke's pouch of wild-type mouse embryos. (A–L) Whole-mount in situ hybridisation on wild-type mouse embryos. Probes and developmental stages are indicated on the panels. Top row, lateral view, anterior to the left. Bottom row, rostral view. All interactors are co-expressed with Hesx1. (M–O) In situ hybridisation on sections of Rathke's pouch of 12.5 dpc mouse embryos. Note the high-dorsal to low-ventral gradient of expression of Hesx1, Dnmt1 and Safb1. (P) RT-PCR from RNA samples purified from mouse pituitaries at 12.5 dpc and mouse heads at 8.5 dpc for Hesx1 and the five interactors. Amplification of the RNA samples prior to retrotranscription (c-) demonstrates the absence of DNA contamination.

Co-immunoprecipitation of HESX1 and DNMT1 tagged proteins in 293T cells. Cells were transfected with plasmids expressing full-length HA–HESX1 and/or full-length Flag–DNMT1 (indicated by + or −) and immunoprecipitated with anti-flag and anti-HA antibodies. Immunoprecipitates were blotted and detected with anti-HA or anti-Flag HRP-conjugated antibodies. Specific immunoreactive bands are indicated with arrowheads.

Localisation of HESX1 and DNMT1 in CHO cells. Immunofluorescence was performed on cells transfected with HA–HESX1 and Flag–DNMT1 tagged proteins. (A, D) HA–HESX1 staining (green) is mostly nuclear, but sporadic cells also show cytoplasmic staining. (B, E) Flag–DNMT1 staining (red) is predominantly nuclear. (C) Merged photographs of panels A and B. (F) Merged photographs of panels D and E. Co-localisation of HA–HESX1 and Flag–DNMT1is observed in the nucleus (yellow staining in panels C and F). Scale bar: 100 μm.

Diagram of HESX1 and DNMT1 proteins showing their functional domains and interacting partners. (A) Two domains in the N-terminal half of HESX1 (eh1 and HRPW) interact with the repressor TLE1, whilst the homeodomain mediates interaction with the N-CoR. (B) DNMT1 contains multiple domains including (1) the PCNA binding motif; (2) the nuclear localisation signal (NLS); (3) the targeting sequence (TS); (4) the CxxC domain; (5) the polybromo domains 1 and 2 (BAH1 and 2); and (6) the catalytic domain. The regions involved in protein–protein interactions with previously reported DNMT1 partners are indicated. This image has been adapted from Spada et al. [21].

GST pull-down experiments with DNMT1, LONP2, SRFBP1, Zfp592 and SAFB1. HESX1-interacting proteins were in vitro translated in the presence of ³⁵S-methionine and incubated with 2.0–3.0 μg of either, GST alone, GST–HESX1 (aa 1–185) or GST–HESX1 (aa 1–107). One tenth of the labelled protein used for the pull-down experiment (arrowheads) was loaded as a control (input). (A–E) Results of GST pull-down experiments with DNMT1 (A), LONP2 (B), SRFBP1 (3), ZFP592 (D) and SAFB1 (E) show specific binding to HESX1 sequences, but only very weakly or not at all to GST. (F) Representative Coomassie stain of the proteins used in the GST pull-down experiments. Molecular weight markers are indicated on the left of each panel. Predicted molecular weights are: 178 kDa (DNMT1); 94 kDa (LONP2); 49 kDa (SRFBP1); 121 kDa (ZFP592); 103 kDa (SAFB1).

Mapping of the regions responsible for the interaction between HESX1 and DNMT1. (A) Schematic representations of HESX1 fragments expressed as GST fusion proteins. (B) GST pull-down experiments. DNMT1 (aa 1–1620) (arrowhead) strongly interacts with GST–HESX1 (aa 1–185), whilst binding to GST–HESX1 (aa 1–107) is weak. (C) Representative Coomassie stain of the GST–HESX1 fusion proteins used in panel B. (D) Schematic representation of DNMT1 fragments expressed as GST fusion proteins. (E) GST pull-down experiments. Strong interaction with HESX1 (aa 1–185) (arrowhead) is observed with DNMT1 fragments containing the first 748 aa (aa 1–343, aa 350–609, aa 613–7480 and aa 653–730) and the catalytic domain (aa 1124–1620). No binding was detected with a DNMT1 fragment containing the BAH1 domain (aa 686–812). A DNMT1 fragment including BAH1 and BAH2 domains (aa 746–1110) exhibited weak binding. (F) Representative Coomassie stain of the GST–DNMT1 fusion proteins used in panel B. As input, 1/10th of the total amount of ³⁵S-Met-labelled protein was used in panels B and E. (G) Approximately 1 mg of protein extracts obtained from untransfected ES cells was incubated with either an anti-DNMT1 antibody (IP: anti-DNMT1), GST or GST–HESX1 (aa 1–185), and bound proteins analysed by Western blot using an anti-DNMT1 antibody. Only GST–HESX1 (aa 1–185), but not GST, interacts with endogenous DNMT1. Molecular weight markers are indicated on the left of each panel.

(A) Schematic representation of full-length DNMT1 (aa 1–1620) and fragments used in these experiments. Cells were transfected with full-length HA–HESX1 and/or several Flag–DNMT1 deletion proteins as indicated (+ or −) and immunoprecipitated with anti-Flag or anti-HA antibodies. The presence of HESX1–DNMT1 complexes was analysed by Western blotting the anti-Flag immunoprecipitates with an HRP-conjugated anti-HA antibody. Specific immunoreactive bands are indicated with arrowheads. Fragments containing part of the N-terminus and the catalytic domain of DNMT1 (aa 158–1620 and aa 296–1620) strongly interact with HA–HESX1. Immunoprecipitation with anti-HA antibody was used to verify the presence of HA–HESX1 in the lysates.

HESX1 repressor activity is not enhanced by DNMT1 in a mammalian one-hybrid system. (A) The basal activation of a luciferase reporter vector containing the SV40 promoter and Gal4 DNA binding sites (p-Gal4BS-SV40-Luc) can be repressed by expression of Gal4 DNA binding domain (DBD)–HESX1 fusion protein from the p-Gal4DBD–Hesx1 effector plasmid in a dose-dependent manner. (B) Co-transfection of DNMT1 cannot increase the repressor activity of the Gal4DBD–Hesx1 fusion protein.