The non-methylated DNA-binding function of Kaiso is not required in early Xenopus laevis development

Abstract

Mammalian forms of the transcription repressor, Kaiso, can reportedly bind methylated DNA and non-methylated CTGCNA motifs. Here we compare the DNA-binding properties of Kaiso from frog, fish and chicken and demonstrate that only the methyl-CpG-binding function of Kaiso is evolutionarily conserved. We present several independent experimental lines of evidence that the phenotypic abnormalities associated with xKaiso-depleted Xenopus laevis embryos are independent of the putative CTGCNA-dependent DNA-binding function of xKaiso. Our analysis suggests that xKaiso does not play a role in the regulation of either xWnt11 or Siamois, key signalling molecules in the Wnt pathway during X. laevis gastrulation. The major phenotypic defects associated with xKaiso depletion are premature transcription activation before the mid-blastula transition and concomitant activation of a p53-dependent cell-death pathway.

Fig. 1.

The methyl-CpG binding activity of Kaiso is confined to ZF1 and 2. (A) Alignment of the zinc-finger domains of Gallus, Xenopus, mouse and Danio Kaiso proteins. The regions required for Hmat-dependent binding (solid line) and methyl-CpG-specific binding (dotted line) according to our results are compared with the previously published DNA-binding motif (dashed line) (Daniel et al., 2002). (B) SDS-PAGE showing the indicated purified 6xHis-tag ZF1-3 fusion proteins (arrow) used in the EMSAs. Size markers are on the left. (C) EMSA experiment with the indicated purified ZF1-3 proteins of xKaiso, dKaiso and gKaiso with methylated Sm, non-methylated S or human matrilysin (Hmat) probes in the presence of 2 μg pdIC competitor. Arrow indicates the Kaiso ZF-specific band shift. (D) EMSA using GST-ZF12 (xKZF12, gKZF12) and GST-ZF23 (xKZF23, gKZF23) deletion constructs from xKaiso and gKaiso, respectively, with Sm and Hmat oligos. xKaiso and gKaiso ZF domains (KZF123) were used with Sm probe as positive controls. (E) EMSA using eukaryotically expressed and affinity-purified full-length Xenopus (xKaisoFL) and Danio (dKaisoFL) Kaiso proteins with Sm (1), S (2), Hmat (3) and CTGCNA-containing probes from the promoter regions of Siamois (4) and xWnt11 (5).

Fig. 2.

Danio rerio Kaiso is a methyl-CpG-dependent repressor that is necessary for zebrafish development. (A) Methyl-CpG-dependent repression by dKaiso in a transient transfection assay. Kaiso expression constructs were co-transfected with a methylated SV40-luciferase reporter into mouse cells that are compromised in methyl-CpG-dependent transcriptional repression (Kaiso/Mecp2/Mbd2^-/-). The methylated SV40-luciferase reporter is repressed in the presence of dKaiso. The relative percentage (methylated reporter expression/nonmethylated reporter expression) is the average of at least three experiments. Human Kaiso (hKaiso) and xKaiso expression constructs were used as positive controls for methyl-CpG-dependent transcriptional repression. (B) The phenoptypes of KMO-injected zebrafish embryos compared to control embryos 24 hours after fertilisation. Lower panel is an FITC image of the upper panel and shows that the severity of phenotypical defects correlates with the amount of injected fluorescein-labelled morpholino. (C) The percentages of normal embryos, embryos dead at 24 hours of development and embryos with strong (non-viable) and medium developmental abnormalities at 48 hours after fertilisation are shown for non-injected (n=259), standard control MO (Std Ctrl MO, n=62) and dKMO (n=182) embryos. (D) The microcephaly phenotype of dKMO morphants at 24 hours post-fertilisation (left panel) can be rescued by xKaiso mRNA (middle panel). A control (CMO-injected) embryo (right panel) at the same stage is shown for comparison.

Fig. 3.

xKaiso has no specific affinity for CTGCNA-binding sites in the Siamois and xWnt 11 promoters. (A) EMSA experiment with the indicated purified 6xHis-ZF1-3 proteins dKaiso and gKaiso with methylated Sm probe and CTGCNA-containing probes derived from the promoter regions of Siamois (Siam) and xWnt11 (Wnt) in the presence of 2 μg pdIC competitor (B) Same experiment as in A but with xKaiso. (C) Competition experiment with 6xHis-ZF1-3 xKaiso protein under standard EMSA conditions with the Sm probe but with 200-fold excess of the following cold competitors: S, Sm, Siamois, xWnt11 and Hmat. Notice only the Sm and Hmat probes compete effectively. The last lane is a super-shift experiment in which an anti-His-tag antibody is included that shifts the 6xHis-Kaiso-specific complex. (D) Competition experiment as in C, with increasing amounts of cold competitors (2×, 20× and 200×). The signal quantification using AIDA software is shown below. Note that the Hmat oligo competes at least 10 times less efficiently than Sm. (E) EMSA with purified GST fusions of ZF1-3 domains of xKaiso with labelled Hmat, Siamois, xWnt11 and (non-CTGCNA) TCF3-binding site (TCFbs) probes. A fixed amount of protein and the indicated decreasing amount of pdIC competitor was used. Arrows indicate the shifted complex for each probe.

Fig. 4.

Kaiso preferentially interacts with methylated CpGs but not with CTGCNA sequences in vivo. (A-D) The results of genome-wide ChIP/sequencing experiments in HEK293 transiently transfected with mouse Kaiso (mKaiso) and HA-tagged Danio Kaiso (dKaiso-HA). The ChIPs were performed using anti-mKaisoZF or anti-HA-tag antibodies with additional controls using preimmune serum (control for mKaiso ChIP) or anti-HA-tag antibody on non-transfected cells (control for dKaiso-HA experiment). The DNA obtained in the ChIP was amplified and 454 sequenced. After initial data filtering all the remaining sequences were mapped on to the human genome, susbsequently 1 kb regions in the vicinity of the ChIP sequences were analysed for the presence of either CpG-rich regions or CTGCNA sites. The ChIP sequences for both the mKaiso and dKaiso-HA experiments were enriched in CpG-rich regions in comparison to either to preimmune serum or anti-HA-tag antibody on non-transfected cells, respectively (A,C), but not in CTGCNA sites (B,D). The data were normalised with respect to the genome-wide distribution of CpG rich regions and CTGCNA sites as shown. (E) Diagram indicating the DNA methylation status of the Oct 91 distal promoter fragment used for ChIP in (F) in A6 cells. Filled circles represent methylated, and empty circles non-methylated, CpGs. (F) ChIP experiment using transiently transfected Xenopus HA-tagged Kaiso (xKaiso) and T7 tagged Danio Kaiso (dKaiso) on A6 cells. Both xKaiso and dKaiso bind to the heavily methylated distal region of the Oct91 gene, but do not show any detectable binding to the Siamois promoter under the same experimental conditions. IgG was used as an antibody control. 1/10 and 1/50 of inputs are loaded for the Siamois and Oct91 experiments, respectively. (G) The xKaiso ZF domain VP16 fusion (xKZF) does not activate transcription of a Siamois-driven luciferase reporter (S01234) but does activate transcription from a methylated Tex19 promoter (Tex19Me). The xTcf3 HMG domain fusion (xTcf3) activates transcription from the Siamois reporter 5.5 times. A Siamois luciferase reporter containing mutated xTcf3-binding sites (S0) and an unmethylated Tex19 promoter reporter (Tex19Un) were used as controls.

Fig. 5.

dKaiso can rescue Kaiso-depleted Xenopus laevis embryos to the same extent as its human counterpart. (A) The phenotypes of uninjected control (n=150), KMO (n=53) and xKMO co-injected with dKaiso RNA (n=84) embryos (KMO+dKaiso). Development stages are indicated. FITC image of two pictures are presented as well as an injection control; arrow indicates neural fold. Notice that even at the later stage (St. 39), when there are reduced numbers of survivors, the xKMO morphants are arrested whereas the rescued embryos can form complete tadpoles or attenuated tadpoles that differ in appearance from the xKMO morphant. (B) Bar graphs showing the percentages of normal embryos and embryos with developmental defects in the rescue experiments using xKMO co-injected with dKaiso or human Kaiso (hKaiso) RNA. Dead embryos are not included. The stages of development are indicated.

Fig. 6.

Inhibition of apoptosis in KMO embryos results in their successful gastrulation. (A) The presence of a caspase inhibitor, Z-DEVD-FMK, prevents apoptosis in xKMO morphants and allows development to proceed. The rescued embryos can complete gastrulation but neurulation is impaired and they exhibit development delay compared to control embryos. Arrows indicate a poorly developed neural fold in the rescued xKMO morphant at stage 19. The rescued xKMO morphants do not show evidence of axis duplication (indicating no hyper β-catenin activation during gastrulation) at stage 38. In addition they exhibit developmental delay and axis defects that result from poor neurulation. Control embryos incubated with Z-DEVD-FMK are phenotypically normal. (B,C) Phenotypes of embryos co-injected with xKMO together with an xp53 morpholino (p53MO) are presented at stages 15 and 26. Uninjected control embryos are also shown (C in figures). Note the completion of delayed gastrulation in KMO/p53MO embryos (blue arrow points to the blastopores).

Fig. 7.

The Wnt signalling pathway is not activated in xKMO morphants. (A) The xKMO morphants exhibit a delay in closing the blastopore (short arrows) compared with wild-type embryos. This phenotype appears in 85-90% of the embryos and is identical to that presented by McCrea and colleagues (Kim et al., 2004; Park et al., 2005). By stage 15 (neurulation), the xKMO morphants (downward arrow) cannot form a neural fold; they are apoptotic and are shedding cells through the open blastopore. Control embryos are shown with a proper neural fold (long arrows). (B,C) Neither Siamois nor xWnt11 are ectopically activated in xKMO morphants when assayed by semi-quantitative RT-PCR according to Park et al. (Park et al., 2005) or real-time PCR relative to a histone H4 control at stage 10 (Siamois) or 12 (xWnt11). Caspase7 and Caspase9 expression is activated in 2.5-3.5 times compared to control in the same sets of KMO embryos at stages 10 and 12, respectively. (D) Whole-mount RNA in situ analysis demonstrates that xWnt11 is not prematurely activated in pre-MBT xKMO morphants in comparison to a control transcript, xID2, that is activated prematurely (Ruzov et al., 2004).