Two-element transcriptional regulation in the canonical Wnt pathway

Two-element transcriptional regulation in the canonical Wnt

pathway

Kibeom Kim

1,†

Jaehyoung Cho

1,†

Thomas S. Hilzinger

1,†,‡

Harry Nunns

Andrew Liu

Bryan E. Ryba

, and

Lea Goentoro

1,*,§

Division of Biology and Biological Engineering, California Institute of Technology, Pasadena, CA

91125

Department of Physics and Department of Molecular Biology, University of California, San Diego,

CA 92093

Summary

The canonical Wnt pathway regulates numerous fundamental processes throughout development

and adult physiology, and is often disrupted in diseases [

]. Signal in the pathway is

transduced by

-catenin, which in complex with Tcf/Lef, regulates transcription. Despite the many

processes that the Wnt pathway governs,

-catenin acts primarily on a single cis-element in the

DNA, the Wnt-Responsive Element (WRE), at times potentiated by a nearby Helper site. In this

study, working with

Xenopus

, mouse, and human systems, we identified a cis-element, distinct

from WRE, that

-catenin and Tcf act on. The element is 11-bp long, hundreds of bases apart from

WRE, and exhibits a suppressive effect. In

Xenopus

patterning, loss of the 11-bp negative

regulatory elements (11-bp NREs) broadened dorsal expression of

siamois

. In mouse embryonic

stem cells, genomic deletion of the 11-bp NREs in the promoter elevated

Brachyury

expression.

This reveals a previously unappreciated mechanism within the Wnt pathway, where gene response

is not only driven by WREs, but also tuned by 11-bp NREs. Using EMSA and Chip, we found

evidence for the NREs binding to

-catenin and Tcf – suggesting a dual action by

-catenin as a

signal and a feedforward sensor. Analyzing

-catenin Chip-Seq in human cells, we found the 11-

bp NREs co-localizing with WRE in 45–71% of the peaks, suggesting a widespread role for the

Corresponding author: goentoro@caltech.edu.

†

These authors contributed equally to the work.

‡

Current address: PricewaterhouseCooper LLP, Chicago, IL 60654

Lead Contact

Publisher's Disclaimer:

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our

customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of

the resulting proof before it is published in its final citable form. Please note that during the production process errors may be

discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Author Contributions

J.C., T.S.H., K.K., A.L., B.E.R. and L.G. designed experiments, performed experiments, and performed data analysis. H.N. performed

modeling analysis, contributed to data analysis, and wrote the model supplement. K.K. and L.G. wrote the manuscript.

DATA AND SOFTWARE AVAILABILITY.

The full sequence of the 3kb

siamois

promoter will be deposited to Pubmed. Constructs generated in the study are available upon

request.

KEY RESOURCE TABLE.

Table provided in a separate file.

Methods S1

. Modeling analysis of incoherent feedforward loop.

This analysis describes how IFFL can produce different gene

response dynamics, and how inclusion of an IFFL downstream the Wnt pathway can recapitulate endogenous gene regulation.

HHS Public Access

Author manuscript

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Published in final edited form as:

Curr Biol

. 2017 August 07; 27(15): 2357–2364.e5. doi:10.1016/j.cub.2017.06.037.

Author Manuscript

mechanism. This study presents an example of a more complex cis-regulation by a signaling

pathway, where a signal is processed through two distinct cis-elements in a gene circuitry.

Keywords

Canonical Wnt pathway;

-catenin; Tcf; cis-regulation; gene transcription;

siamois

;

Brachyury

;

Xenopus

; mouse embryonic stem cells

Results

We started with a previously reported discrepancy between the endogenous gene and

TopFlash reporter expression (Figure 1B–D) [

]. Wnt signaling in early

Xenopus

blastulas

activates dorsal regulators, including

siamois

and

Xnr3

. Treating the embryos for 5–10

minutes with 300 mM lithium is known to inhibit GSK3

[

], stabilize

-catenin [

], and

dorsalize the embryos (Figure 1B–C) [

]. We observed, however, that embryos treated

with moderate doses of lithium (150 and 200 mM) largely retained a wild-type level of

siamois

and

Xnr3

expression (Figure 1C, see red arrows), and developed into wild-type

tailbuds (Figure 1B). More strikingly, despite absence of marked phenotypic effects, the

embryos showed increased

-catenin level (Figure 1D, see red arrows). Similar lack of

embryo phenotypes despite increased

-catenin level was observed with other perturbations

to the Wnt pathway, including injection of

Axin1

and

GBP

mRNA. This led us to assay the

TopFlash reporter, a commonly used reporter of Wnt signaling driven by a tandem repeat of

WREs, to test whether the increased

-catenin was transcriptionally active. Indeed, the

TopFlash reporter also showed increased activity at moderate doses of lithium (Figure 1D,

red arrows). Therefore, even though the TopFlash reporter faithfully tracked the rise of

catenin level, the contrasting wild-type expressions of the endogenous genes suggest a

missing mechanism beyond WRE. It is notable that

-catenin itself, as a central regulator in

the pathway, did not correlate with the target gene expression but tracked the extent of the

perturbations, hinting at a possible role for

-catenin in the missing regulation.

An 11-bp negative regulatory element is necessary for dorsal specification of

siamois

To locate the missing regulation, we focused on

siamois

[

]. Three WREs are found

within 500 bp upstream of

siamois

, and are necessary for

siamois

activation (Figure 2A) [

]. We built a luciferase reporter using a 3kb fragment of

siamois

promoter (pSia). We

confirmed that 3kb pSia-luc mimics the temporal expression of

siamois

, beginning at mid-

stage 8 and reaching steady state by stage 10. Reported here is the luciferase expression at

early stage 10, reliably identified by the onset of dorsal lip formation.

We perturbed the GSK3

activity in a dose-response manner as described earlier. We found

that the 3kb promoter recapitulated the endogenous

siamois

response, preserving wild-type

expression at 150 mM LiCl (Figure 2B). By contrast, an 848bp fragment of

siamois

promoter that still contains the three WREs responded readily to perturbations (Figure 2C).

To examine the spatial regulation of these promoter fragments, we tested lacZ reporters.

Indeed, even though 3kb and 848bp-lacZ retained dorsal expression, the 848bp fragment

showed expanded expression (Figure 2C–D inset). The missing regulation in the 848bp

Kim et al.

Page 2

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Author Manuscript

fragment was especially revealed when a moderate lithium dose was applied: the 3kb pSia-

lacZ preserved dorsal expression at 150 mM LiCl, whereas the 848bp pSia-lacZ readily

expanded.

Therefore, a suppressive, distal element located between 848bp and 3kb is necessary for the

dorsal specification of pSia expression. Using the differential response at 150 mM lithium as

a readout, we performed a promoter bashing screen to locate the element. Halving to 1.3kb

preserved wild-type reporter expression, as did further truncations up until 963bp (Figure

2D, F). Loss of suppression at 150 mM LiCl was finally observed with truncation to 952bp,

and with subsequent shorter constructs (Figure 2E, F). This analysis identified a suppressive

element at 1kb upstream from

siamois

, with most activity centered on an 11-bp between 963

and 952bp (Figure 2G).

The 11-bp negative regulatory element (11-bp NRE) is AT-rich, 5’- CTG TTA TTT AA -3’.

To further characterize the 11-bp NRE, we performed a mutagenesis analysis. With the

1.3kb promoter fragment, we mutated the 11-bp NRE one base at a time (

i.e.

, purine into

pyrimidine, or vice versa). We found that the majority of single-base alterations affected the

response, with the largest effect produced by a single-base mutation on the 11th nucleotide,

which recapitulated the effect of deletion of the 11-bp NRE (Figure 2H). Finally, the 11-bp

NRE is sufficient to recapitulate the suppression. Inserting the 11-bp NRE to 848bp

promoter (Figure 2F) rescued the wild-type expression. The same effect was observed when

the 11-bp NRE was pasted to even shorter promoters (Figure S1). These results suggest that

regulation of

siamois

not only requires WREs, but also a suppressive 11-bp NRE.

The 11-bp NRE interacts with Tcf

Next, we set out to find the factors that bind to the 11-bp NRE using EMSA (Figure 3A, B).

As the protein source, we used the high-speed supernatant of

Xenopus

egg extracts. We

reasoned that the binding factor(s) must be maternal, as Wnt signaling begins in early

blastula before zygotic transcription [

]. As the DNA probe, we synthesized a 30-bp

double stranded oligomer containing the 11-bp NRE, flanked by the endogenous sequence.

As a control for the probe, we used the strongest mutant with an A–>C switch at the 11th

nucleotide (m11 mutant; Figure 2H).

We observed specific binding to the 11-bp NRE in the EMSA (Figure 3C). The band was

competed away when excess, unlabeled DNA probe was added (Figure 3C, lanes 3–4). The

band was competed away much less effectively when excess, unlabeled m11 mutant probe

was used (Figure 3C, compare lanes indicated by blue arrows). The specific band was

reproducible across batches of extracts, despite variation in nonspecific binding patterns and

the amount of competitors needed (Figure 3C).

Having the specific EMSA activity, we proceeded to identify the binding factor(s) by testing

some known transcription factors. Lacking good antibodies to these factors in

Xenopus

, we

tested if the DNA binding sites of the factors would compete with the EMSA band. Binding

sites of various transcription factors,

e.g.

, CREB, p53, NF-

B, AP-1, failed to compete

significantly (Figure S2A). Unexpectedly, the one DNA fragment that competed away the

EMSA band was the WRE itself (Figure 3D). The specific band was competed away when

Kim et al.

Page 3

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Author Manuscript

excess, unlabeled WRE probe was added (Figure 3D, lanes 5–6) – to a similar extent as it

was competed away by excess, unlabeled wild-type probe (Figure 3C, lanes 3–4). By

contrast, a negative control probe, carrying two mutations that destroy binding of the Tcf

protein (

i.e.

, FopFlash construct) [

], competed much less effectively (Figure 3D, lanes 7–

8, see blue arrows).

The competition by WRE suggests that Tcf proteins bind to the 11-bp NRE. Indeed, XTcf3

antibodies, raised against an N-terminal (XTcf3n) and C-terminal fragment (XTcf3c) [

competed away the specific band (Figure 3E). These results suggest that a Tcf/Lef protein

binds the 11-bp NRE, and that XTcf3 is the predominant binder in our

in vitro

assay.

The 11-bp NRE interacts with

-catenin

The binding of Tcf protein to the 11-bp NRE raises questions on the binding partner, as Tcf

does not usually act alone [

]. The binding of Tcf to the 11-bp NRE is also interesting in

light of our earlier findings (Figure 1) that

-catenin intriguingly tracks the extent of

perturbations, motivating us to test the roles of

-catenin.

We found that adding polyclonal

-catenin antibody to the EMSA reaction produced a

strong supershift, and a weak downshift smear (Figure 3F). To further confirm the binding of

-catenin, we performed binding assay using purified recombinant

Xenopus

-catenin and

Tcf3 (Figure S2B). As expected, Tcf3 alone produces a smear signal, suggesting an unstable

binding. A strong, sharp band was observed in the presence of both Tcf3 and

-catenin.

To test if

-catenin acts on the 11-bp NRE in the

siamois

promoter

in vivo

, we performed

chromatin immunoprecipitation in the embryos (Figure 3G). We collected embryos at stage

10, when

siamois

expression is at peak. We observed a strong signal indicating

-catenin

binding in the 11-bp region. In the same experiment, as expected,

-catenin binding was also

detected from the WRE region in the

siamois

promoter. As a negative control, negligible

signal was detected with IgG antibody and from

siamois

ODC

coding regions. These

results suggest that the 11-bp NRE, required for

siamois

regulation, interacts with

-catenin

and Tcf.

The 11-bp NRE is present in the promoter of more target genes

Beyond

siamois

, might the 11-bp NRE regulate other Wnt target genes? At first, we found

no results looking for the exact 11-bp sequence in other

Xenopus

targets. However,

examining the

siamois

promoter closely, we identified two sites with a sequence pattern

similar to that of the 11-bp NRE (Figure 4A–B), a G/C cap, followed by 8 A/T’s – with an

occasional C/G in the 5th position. All 11bp-like elements competed with the WRE (Figure

S3A). Hence, three 11-bp NREs are present in the

siamois

promoter, although losing the

distal one is sufficient to disrupt the suppression. Following this lead, we searched for a

similar sequence pattern in other known direct Wnt targets in

Xenopus

. We found candidate

11-bp NREs in the promoter of

Xnr3

and

engrailed

(Figure 4B). The predicted 11-bp NREs

Xnr3

and

engrailed

produced a specific EMSA band, which was shifted by polyclonal

catenin antibody (Figure S3B).

Kim et al.

Page 4

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Author Manuscript

The 11-bp NRE regulates

Brachyury

expression in mouse embryonic stem cells

With the predictive sequence pattern (Figure 4C), we investigated if 11-bp NRE is present in

mammals. We examined promoters of known direct Wnt targets in mouse embryonic stem

cells (mESC). We identified several candidates, including some in the promoters of well-

characterized Wnt targets, such as

Brachyury

(

Axin2

, and

Cdx4

(Figure 4B). EMSA

assay confirmed that these 11-bp NREs produced a specific band that was shifted by

catenin antibody (Figure S3C).

To test whether the 11-bp NREs function in regulation of mammalian Wnt targets, we

examined

regulation.

promoter has two WRE sites, at 191 and 273 bp upstream of the

transcription start site (Figure 4D) [

is an early marker of mesoderm differentiation.

Basal expression of

, present in a fraction of stem cell population [

], is activated by

endogenous secretion of Wnt proteins [

] and marks the early mesoderm-committed (EM)

progenitors [

We identified two candidate 11-bp NREs, at 999 and 1613 bp upstream from the

transcription start site (Figure 4D). To test whether the 11-bp NREs interact with

-catenin

in vivo

, we performed chromatin immunoprecipitation using

-catenin antibody. We

observed signal from the proximal 11-bp NRE (Figure 4E). As positive control, strong signal

was also observed from the −273bp WRE (Figure 4E). As negative controls, almost tenfold

lower signal was observed using mouse IgG antibody, and negligible signal from the exon 2

region of

We observed much lower signal from the distal NRE, suggesting the proximal

one as the dominant NRE in the

promoter.

To test the roles of the 11-bp NREs in

regulation, we used Crispr/Cas9 to delete the 11-bp

NREs from the endogenous promoter of

. Four sequence-verified

Δ11bp clones were then

assayed for

expression by qRT-PCR. We observed that

Δ11bp clones showed

significantly higher expression levels of

than the wild type cells (Figure 4F), suggesting

increased EM progenitors. As a control, four wild-type clones that underwent blank

transfection and clonal selection showed no significant change of

expression. These results

suggest that balanced regulation of

in mESC require 11-bp NREs as well as the WREs.

The 11-bp NRE is prevalent in

-catenin Chip-Seq in human cells

Finally, equipped with the criterion of 11-bp NRE gathered across several Wnt direct targets

(Figure 4C), and one confirmed with binding and functional assays (Figure 2,3, 4E–F), we

asked whether 11-bp NREs are present more widely across Wnt target genes. To address this

question, we analyzed

-catenin Chip-Seq datasets from human cells [

]. We found

significant enrichment of 11-bp NREs in

-catenin peaks from HEK293T (p-value: 2.58 e–

31) and HCT116 cells (p-value: 4.24e–12). Multiple 11-bp NREs were often predicted

within the

-catenin-bound fragments. We found significant co-localization of 11-bp NREs

with WRE: of the 2818

-catenin peaks that contain WRE, 71% also contain 11-bp NREs in

in HEK293T cells (Figure 4G), and 45% do so in HCT116 cells (Figure S4). Our analysis

also indicates a significant fraction of

-catenin-bound fragments that contains 11-bp NREs

only (Figure 4G), suggesting that the 11-bp NRE may have more functions beyond what we

found here.

Kim et al.

Page 5

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Author Manuscript

Discussion

We found an 11-bp NRE that

-catenin and Tcf act on in the promoters of many Wnt target

genes. The 11-bp NRE is necessary for

siamois

regulation in

Xenopus

embryos and

Brachyury

regulation in mouse embryonic stem cells. Our study suggests a new model of

gene regulation in the Wnt pathway, where signal in the pathway not only activates target

genes through WRE, but also in some contexts, tunes expression of the genes through a

distinct 11-bp NRE. (Figure 4I). Interestingly, although multiple NREs are present in

siamois

and

promoter, there seems to be a dominant one that mediates suppression. The

Chip data in

promoter suggests that the other NRE binds less strongly to

-catenin and

therefore may play subtler roles. In

Drosophila

, it was found that Wnt signaling represses

Ugt36Bc and TiG genes through a WGAWAW site that binds to

-catenin [

]. Our finding

identifies a distinct repressive site in vertebrates, and one that implicates a different role for

-catenin, where it acts in opposite manner in the same promoter, as an activator through

WRE and a suppressor through the 11-bp NRE. With regard to Tcf, our finding can be

consistent with models in the field, where the same or distinct Tcf proteins may specialize as

an activator or a repressor [

A circuit where the input activates the output, and at the same time suppressively tunes the

output, is known in engineering as the incoherent feedforward loop (IFFL). Such a circuit

has been found in transcriptional networks of bacteria, yeast, and human cells [

–

While these analysis focused on protein-protein interactions, an interesting aspect of the

IFFL we found in the Wnt pathway is its implementation through multiple cis-elements. We

know of one other instance where a regulator works through multiple cis-elements in an

IFFL mode, in the regulation of porin OmpF in

E. coli

[

More generally, IFFL belongs to a class of recurring strategy in biological systems, where a

biological molecule is used in a paradoxical manner (

). Paradoxical circuits, and the

incoherent feedforward circuit specifically, moreover, are versatile circuits. By tuning the

relative strengths and time scales of the activation and repression arm, an incoherent

feedforward circuit can generate a sustained, net activation – but beyond that, also a net

repression, a temporal pulse, response acceleration, band-pass filtering, and fold-change

detection (see Methods S1, as well as refs. [

–

]). Moreover, inclusion of an IFFL

downstream the Wnt pathway can explain how the endogenous gene response remains wild

type despite perturbations, either by acting as an amplitude filter (if the timescales of

activation and repression are similar) or fold-change detector (if the repression is slow and

strong) (see Methods S1). Although it is difficult presently to estimate the relative affinities

of the two sites, an extract system combined with sophisticated methods such as SPR could

be a next approach. It would also be interesting to investigate whether different co-factors or

chromatin modifiers are recruited by

-catenin to the 11-bp NRE. Thus, the finding of the

11-bp NRE not only suggests a different model of gene regulation in the Wnt pathway, and

one that implicates a function for

-catenin as a feedforward sensor, but also provides a

plausible mechanism by which gene regulation by

-catenin and Tcf can generate more

versatile dynamics across contexts.

Kim et al.

Page 6

Curr Biol

. Author manuscript; available in PMC 2018 August 07.

Author Manuscript