Non-acylated Wnts can promote signaling

SUMMARY

Wnts are a family of 19 extracellular ligands that regulate cell fate, proliferation and migration during metazoan embryogenesis and throughout adulthood. Wnts are acylated post-translationally at a conserved serine and bind the extracellular cysteine-rich domain (CRD) of Frizzled (FZD) seven-pass transmembrane receptors. Although crystal structures suggest that acylation is essential for Wnt binding to FZDs, we show here that several Wnts can promote signaling in Xenopus laevis and Danio rerio embryos – as well as in an in vitro cell culture model – without acylation. The non-acylated Wnts are expressed at similar levels to wild-type counterparts and retain CRD binding. By contrast, we find that certain other Wnts do require acylation for biological activity in Xenopus embryos – although not necessarily for FZD binding. Our data argue that acylation-dependence of Wnt activity is context-specific. They further suggest that acylation may underlie aspects of ligand/receptor selectivity and/or control other aspects of Wnt function.

INTRODUCTION

Wnts are extracellular ligands with diverse roles in embryonic patterning, maintenance of adult stem cells, and cancer (Nusse and Clevers, 2017). Mammalian genomes contain 19 unique Wnt genes, encoding fatty acylated extracellular proteins that regulate numerous cell surface receptors (Nile and Hannoush, 2016). Certain Wnts (e.g. Wnts 1, 3a, and 8) recruit LRP5/6 co-receptors to Frizzleds (FZDs) to promote stabilization of a cytoplasmic pool of β-catenin and consequent (β-catenin-dependent) activation of TCF-dependent transcription (Janda et al., 2017; Nusse and Clevers, 2017). Others (e.g. Wnts 5a and 11) engage distinct co-receptors such as Ror2, Ryk and PTK7, in β-catenin independent signaling (Niehrs, 2012). How Wnts recognize their specific receptor complements and promote transmembrane signaling remains incompletely understood.

A breakthrough crystal structure (Janda et al., 2012) of Xenopus Wnt8 (xWnt8) bound to the extracellular ligand-binding cysteine-rich domain (CRD) of murine FZD8 (mFZD8) provided the first visualization of Wnt/receptor interactions (Figure 1A). This structure confirmed that a conserved serine (S187 in xWnt8) is the only acylation site, and suggested that the S187-linked palmitoleoyl moiety plays a crucial role in FZD binding by occupying a hydrophobic channel on the CRD. This hydrophobic channel also binds free fatty acids in a manner thought to promote FZD oligomerization (DeBruine et al., 2017; Nile et al., 2017). Since all Wnts except WntD are predicted to be acylated at this conserved serine (Nile and Hannoush, 2016; Takada et al., 2006), it is thought that Wnts all engage and activate FZDs through such acylation-dependent interactions (Figure 1A). The quantitative importance of acyl chain docking for Wnt signaling has not been directly investigated, however.

FIGURE 1. Effect of site1 and site 2 mutations on xWnt8 activity.

(A) Crystal structure of the xWnt8/mFZD8 CRD complex (PDB: 4F0A), with ‘thumb’ and ‘index finger’ projections on xWnt8 binding to the CRD at sites 1 and 2 respectively (Janda et al., 2012). Residues mutated in site 1 (green) and site 2 (blue) are marked. The palmitoleoyl chain and S187 are red.

(B) Representative dorsalization phenotypes observed upon ectopic xWnt8 expression in ventral cells of Xenopus laevis embryos. The top row shows tailbud-stage embryos with corresponding phenotype scores. Example phenotypes are shown in the bottom two rows. Yellow arrow = partial axis duplication; black = full axis duplication; red = radial dorsalization.

(C) Quantitation of dorsalization phenotypes in Xenopus embryos for site 1 and site 2 mutations. Total number of embryos scored (across 3 biological replicates) is listed for each bar. Dorsal Scores for xWnt8^WT and xWnt8^S187A are from the dataset in Figure 2A, represented here for comparison.

(D) Initial RT-PCR quantitation of Siamois and Xnr3 induction for each variant, represented as mean ± SEM (n = 3). Significance denoted as ‘ns’ (p ≥ 0.05), * (p ≤ 0.05), ** (p ≤ 0.01), or *** (p ≤ 0.001).

(E) Expression of injected xWnt8 variants assessed by Western blotting of mid gastrula stage embryos. Representative of at least three repeats.

(F) Dorsalization phenotypes observed in zebrafish embryos upon ectopic expression of xWnt8^WT or xWnt8^S187A mRNA. Pictures (top row) show representative embryos at 1 day post fertilization displaying normal (left), moderately dorsalized (“twisted”, center), or highly dorsalized (“bustled”, right) phenotypes. Quantitation of observed phenotypes is shown below, with number of embryos scored across at least two biological replicates listed for each bar. See also Figure S1.

Although acylation is stated to be essential for Wnt function (Langton et al., 2016; Nile and Hannoush, 2016; Nusse and Clevers, 2017), it is known that Wnt receptors can nonetheless be activated by non-acylated ligands such as Norrin (Chang et al., 2015) and artificial ‘Wnt surrogates’ that simply cross-link FZDs and LRP5/6 (Janda et al., 2017). Moreover, CRDs in some Wnt-responsive proteins – such as Ror2 – are predicted to lack a hydrophobic channel (Janda and Garcia, 2015). We therefore asked whether Wnt acylation is absolutely required for signaling activity and receptor engagement, or whether – as with EGFR ligands in Drosophila (Miura et al., 2006) – it might play some other important, but modulatory, role. While investigating whether acylation is necessary for Wnt for function, we found that xWnt8 lacking its acylation site retains some ability to bind the CRD of FZD8 and to activate Wnt signaling in both Xenopus laevis and Danio rerio (zebrafish) embryos. We also found that Wnt3a is capable of acylation-independent CRD binding and signaling in Xenopus embryos, whereas Wnts 1 and 5a appeared to require acylation in these contexts. Parallel in vitro studies also revealed acylation-independent signaling for several Wnts. Our findings argue that Wnts can engage CRDs without acylation of their conserved serine, and also that individual Wnts differ in their acylation requirements. These findings may reflect unappreciated differences in Wnt/receptor specificity – possibly defined by acylation – and/or roles for modulating acylation in controlling Wnt signaling outcomes.

RESULTS

Assessing contributions of Wnt/FZD interaction sites to xWnt8 signaling in vivo

The xWnt8/mFZD8 CRD crystal structure (Figure 1A) shows two distinct Wnt/FZD interfaces (Janda et al., 2012) formed by long projections of the crescent-shaped xWnt8 molecule. Site 1 (which includes the acyl chain attached to S187) is formed by the ‘thumb’ of the Wnt molecule, and site 2 by the ‘index finger’. Specific mFZD8 binding by a ‘mini-xWnt8’ comprising only the index finger confirmed the importance of site 2 interactions (Janda et al., 2012). Site 1 is dominated by the palmitoleoyl chain, which fills a hydrophobic channel on the CRD. By burying ~580 Ų of total surface area, this ‘lipid-in-groove’ interface should contribute ~14 kcal/mol to binding (Sharp et al., 1991), which should increase affinity by >10¹⁰ fold. Mutating the acylation site would therefore be expected to abolish xWnt8 activity completely.

To assess the relative importance of sites 1 and 2 for xWnt8 signaling, we made mutations designed to disrupt key interactions at each site and analyzed their effects in a well-established Wnt signaling assay: dorsal development in Xenopus laevis embryos. Injecting mRNA encoding wild-type xWnt8 (xWnt8^WT) into a ventrally-fated cell of a four-cell stage Xenopus embryo activates β-catenin dependent signaling, induces transcription of dorsal markers such as Siamois and Xnr3 (Lemaire et al., 1995; Smith et al., 1995), and expands dorsal development (McMahon and Moon, 1989; Sokol et al., 1991). Low doses of xWnt8^WT cause bifurcation of the dorsal axis, resulting in partial (incomplete) or full (complete) duplication of dorsoanterior structures (Figures 1B and S1A). High doses of xWnt8^WT mRNA cause radial dorsalization of the embryo, as seen when β-catenin dependent signaling is activated by lithium (Kao and Elinson, 1988). We scored mutants across different treatment conditions by assigning each phenotype a numerical value (Figure 1B): 0 = normal; 1 = partial axis duplication; 2 = full axis duplication; and 3 = radial dorsalization. For each treatment group, we summed phenotype scores for tailbud-stage embryos and divided the sum by the number of individuals inspected to give a mean dorsal score between 0 and 3 (Figure 1C).

Mutations were chosen to disrupt key interactions in the xWnt8/mFZD8-CRD complex (Figure 1A). Three xWnt8 variants were designed to progressively disrupt the site 2 interface (W319A, W319A/F317A, and W319A/F317A/V323A) – the triple mutant altering all site 2 contact residues in xWnt8 except C321. K182S, K182S/W196A, and K182S/W196A/I186A variants were also generated to interrogate side chain-mediated interactions at site 1. Separately, we replaced the acylation site serine (S187: Figure S1C) with alanine to prevent acylation and investigate its importance for FZD binding.

Disrupting site 2 interactions abolishes xWnt8 signaling

Site 2 mutations prevented xWnt8 from dorsalizing embryos, as anticipated. As shown in Figures 1B and C, xWnt8^W319A dorsalized embryos very weakly compared to xWnt8^WT – inducing infrequent and incomplete secondary axes. No dorsalizing activity was seen for xWnt8^W319A/F317A and xWnt8^{W319A/F317A/V323A} variants. This loss of function for mutations at site 2 was also evident in studies of transcription of Wnt-specific dorsal gene targets (Figure 1D). Whereas xWnt8^WT strongly induced transcription of both Siamois and Xnr3, this was abolished by site 2 mutations. Importantly, immunoblots of embryo lysates (Figure 1E) showed that each site 2 variant was expressed at least as well as xWnt8^WT. Thus, site 2 interactions play crucial roles in xWnt8 biological activity.

Site 1 mutations have smaller effects on xWnt8 signaling

Consistent with the suggestion from the crystal structure (Janda et al., 2012) that side chain-mediated interactions are less important than ‘lipid-in-groove’ acyl chain binding at site 1, xWnt8 activity was not affected by mutating K182 – which forms a predicted salt bridge with E64 in mFZD8 (not shown). xWnt8^K182S/W196A and xWnt8^{K182S/W196A/I186A} variants failed to express (not shown), making their lack of activity uninterpretable. Contrary to the expectation that removing the acyl chain would completely abolish FZD binding, however, xWnt8 with S187 mutated to alanine (xWnt8^S187A) showed significant residual activity in assays for embryo dorsalization (Figures 1B, C, and S1D) and promoted strong Siamois and Xnr3 transcription (Figure 1D) when injected at just ~10-fold higher levels than xWnt8^WT. Thus, unlike site 2 – where a single mutation (W319A) can abolish xWnt8 activity – site 1 mutations had smaller effects, whether at protein-protein or acyl-protein interfaces.

To determine whether xWnt8^S187A retains function in other in vivo contexts, we compared activities of xWnt8^WT and xWnt8^S187A in zebrafish embryos. Injecting xWnt8^WT or xWnt8^S187A mRNA into one-cell stage zebrafish embryos induced either a “twisted” or a more severe “bustled” phenotype (Figure 1F), as described previously for activation of Wnt signaling (Kelly et al., 1995; Stachel et al., 1993). Over 50% of embryos injected with xWnt8^WT (n = 67), and all embryos injected with xWnt8^S187A (n = 38) were dorsalized at the highest mRNA level used (900 pg). We also observed loss of anterior structures (e.g. eyes) with lower doses of xWnt8^WT mRNA (< 20 pg/embryo), as described previously (Kelly et al., 1995), but not with xWnt8^S187A (not shown); raising the intriguing possibility that non-acylated xWnt8 has distinct activities at different stages of development. Dorsalization was confirmed by in situ hybridization (Figure S1E) showing loss of the ventral marker sizzled (Salic et al., 1997) and expansion of the dorsal neural marker otx2b (Li et al., 1994). These data suggest that differences in potency between xWnt8^S187A and xWnt8^WT may be smaller in zebrafish than in Xenopus embryos.

Non-acylated xWnt8 retains signaling activity

To compare activities of xWnt8^WT and xWnt8^S187A more quantitatively, we performed dose-response analyses of induced dorsalization and gene expression (Figure 2A, B). xWnt8^WT and xWnt8^S187A both demonstrated statistically significant dorsalizing activity (both p < 0.001) as determined by a mixed models regression analysis, and reached equivalent maxima in mean dorsal score (Figure 2A). Moreover, when 500 pg of mRNA were injected, xWnt8^S187A induced Siamois and Xnr3 expression levels that equaled (or exceeded) those seen with xWnt8^WT (Figure 2B). At equivalent intermediate mean dorsal score values (e.g. 5 pg xWnt8^WT and 50 pg xWnt8^S187A), the two xWnt8 variants induced nearly identical ratios of the three dorsalization phenotypes (Figure 2A, right panel). Thus, although capable of the same effects, xWnt8^S187A has a reduced in vivo potency – manifest by the requirement for ~10-fold higher mRNA doses to achieve half-maximum effects than for xWnt8^WT. This difference does not reflect reduced xWnt8^S187A production, as demonstrated by Western blotting of whole embryos; xWnt8^WT and xWnt8^S187A are expressed at similar levels (Figures 1E, 2D ‘input’, S2A), consistent with some other reports (Tang et al., 2012). Our data therefore argue that – although acylation must play an important role in maximizing the biological activity of xWnt8 – it is not essential for in vivo signaling activity.

FIGURE 2. xWnt8 retains biological activity without acylation.

(A) xWnt8^S187A causes dorsalization of Xenopus embryos. Data are represented both as mean dorsal score (left – with number of embryos injected above each bar) and phenotype frequency (right), scored across 4 biological replicates.

(B) Dose-response curve for Siamois and Xnr3 expression (RT-PCR) induced by xWnt8^WT and xWnt8^S187A (n = 3), as in Figure 1D. Although the difference in maximal Xnr3 expression at 500 pg mRNA for xWnt8^WT and xWnt8^S187A appears statistically significant (p ≤ 0.05), biological significance is unclear.

(C) Inhibition of xWnt8^WT and xWnt8^S187A signaling by xFZD8 CRD. xWnt8 and xFZD8 CRD mRNA were co-injected and RT-PCR performed at gastrula stage. Siamois and Xnr3 induction is represented as in Figure 1D (n = 4).

(D) xWnt8^WT and xWnt8^S187A proteins co-immunoprecipitate with myc-tagged xFZD8 CRD in embryos co-injected with both mRNAs. The band in lanes 2 and 6 of the α-myc IP blot corresponds to mouse α-myc antibody heavy chain, which co-migrates with the CRD (~50kD). Representative of at least three biological repeats. See also Figure S2.

Acylation-impaired xWnt8 retains FZD8 CRD-binding capacity

We next asked whether xWnt8^S187A retains its ability to bind FZDs. Biochemical (Hsieh et al., 1999; Janda et al., 2012) and in vivo (Deardorff et al., 1998) studies argue that FZD8 is a xWnt8 receptor, and xWnt8^WT activity in Xenopus embryos can be inhibited by co-expressing the FZD8 CRD (Deardorff et al., 1998). If xWnt8^S187A still binds FZDs, co-expressing the Xenopus (x)FZD8 CRD should likewise inhibit its signaling. We tested this hypothesis by injecting 5 pg of xWnt8^WT mRNA or 50 pg of xWnt8^S187A mRNA (to match signaling levels) – either alone or together with xFZD8 CRD mRNA – into the ventral marginal zone (VMZ) of four-cell stage Xenopus embryos. We collected embryos at the gastrula stage and analyzed Siamois and Xnr3 expression. As shown in Figure 2C, xFZD8 CRD co-expression substantially inhibited Siamois and Xnr3 induction by both xWnt8^S187A and xWnt8^WT, and to similar degrees. Thus, xWnt8^S187A signaling (like xWnt8^WT) is sensitive to inhibition by xFZD8 CRD.

To determine directly whether xWnt8^S187A can interact with FZDs, we next asked whether it co-immunoprecipitates with the xFZD8 CRD in vivo. We co-expressed a myc-tagged xFZD8 CRD in Xenopus embryos with xWnt8^WT or xWnt8^S187A and blotted anti-myc immunoprecipitates with xWnt8 antibodies. Both xWnt8 variants co-immunoprecipitated with the myc-tagged xFZD8 CRD in a dose-dependent manner (Figure 2D). xWnt8^WT did so more efficiently (by ~10-fold) – suggesting either that it binds FZD8 more tightly than xWnt8^S187A or that acylation increases local Wnt concentration to promote receptor binding. These data strongly suggest that xWnt8 can bind FZDs without S187 acylation – although we cannot formally exclude the possibility that co-overexpressed xWnt8 and xFZD8 CRD form disulfides with one another. Excepting this caveat, Figure 2D provides further evidence that S187 acylation is not absolutely required for xWnt8 to bind FZD and activate β-catenin dependent signaling.

Preventing acylation does not abolish xWnt8^S187A secretion

Mutating the acylation site serine in xWnt8 appears to affect its activity in a manner reminiscent of effects seen when the palmitoylation site in the Drosophila EGFR ligand Spitz is mutated (Miura et al., 2006) – reducing in vivo potency but neither blocking Spitz production nor preventing it from binding and activating EGFR. Mutation of the acylation site does not impair xWnt8 production (Figures 1E and S2A), and does not affect xWnt8 secretion into the culture medium when expressed in vitro in HEK293 cells (Figure S2B). Immunofluorescence microscopy studies in animal hemisphere cells of blastula stage Xenopus embryos (Figure S2C) also show qualitatively similar intracellular distributions of xWnt8^WT and xWnt8^S187A. The S187A mutation does appear to block cell-surface xWnt8 accumulation, however, just as mutation of the palmitoylation site in Spitz prevents its cell-surface accumulation (Miura et al., 2006) – which is thought to play a role in restricting extracellular diffusion of the ligand and thus influencing local signaling events.

Wnt1 activity in vivo requires acylation

Given previous literature supporting an absolute requirement for acylation in Wnt function, particularly for Wingless (Wg) in Drosophila (Franch-Marro et al., 2008), we next asked whether murine (m)Wnt1 could dorsalize Xenopus embryos when its acylation site (S224) is mutated to alanine (Figure S1C). As with xWnt8, we injected mRNA encoding wild-type or S224A-mutated mWnt1 into the VMZ of four-cell stage Xenopus embryos. Dorsalization of embryos by mWnt1^WT was seen at mRNA doses as low as 1 pg (p < 0.001), whereas mWnt1^S224A showed no significant dorsalizing activity at doses as high as 500 pg of mRNA (p = 0.877; Figures 3A and S3A) – and only rarely induced partial axis duplication when the mRNA dose was increased to 1000 pg (not shown). Similarly, mWnt1^S224A failed to induce Siamois or Xnr3 at any mRNA dose (Figure 3B). We nonetheless found by Western blotting that mWnt1^S224A protein expresses at least as well as mWnt1^WT (Figures 3C and S3C), and that it co-immunoprecipitates to a small degree with the xFZD8 CRD (Figure S3C). Moreover, mWnt1^S224A protein is secreted by Expi293 cells at similar levels to mWnt1^WT (Figure S3D). These data argue that Wnt1, in contrast to Wnt8, is highly dependent upon acylation for biological activity (but not production) in Xenopus embryos, revealing an important difference between Wnts 1 and 8.

FIGURE 3. mWnt1 requires acylation for biological activity.

(A) mWnt1^S224A does not dorsalize Xenopus embryos. Mean dorsal scores (left) were calculated for tailbud-stage embryos, with injected embryo number listed above each bar (scored across 3 biological replicates). Right: photographs of representative embryos.

(B) mWnt1^S224A fails to induce Siamois or Xnr3. Data presented as in Figure 1D (n = 3).

(C) mWnt1^WT and mWnt1^S224A proteins are expressed at similar levels as assessed by Western blotting of injected mid gastrula stage embryos. Representative of at least two biological repeats. See also Figure S3.

Wnt3a resembles xWnt8 in acylation-independent signaling ability

The distinct acylation dependencies of Wnts 1 and 8 prompted us to examine additional Wnts to ask whether xWnt8 might be unique. In equivalent experiments, we found that hWnt3a^S209A, lacking the acylation site (Figure S1C), can also signal – causing statistically significant dorsalization of Xenopus embryos (p < 0.001; Figures 4A and S4A) and robustly inducing Siamois and Xnr3 (Figures 4B and S4B). Like xWnt8^S187A, hWnt3a^S209A was ~10-fold less potent than its wild-type counterpart (compare 10 pg WT and 100 pg S209A in Figure 4A). This shift in potency likely explains why hWnt3a^S209A activity went undetected in previous work using the same experimental design but a smaller range of mRNA doses (0.8 – 5 pg, Takada et al., 2006), and in zebrafish (Kumar et al., 2014). Interestingly, however, hWnt3a^S209A induced significantly fewer full axis duplications than hWnt3a^WT (Figure S4C) – even at the highest doses. Western blotting of whole embryo lysates confirmed that mutating the acylation site does not greatly impair hWnt3a expression (Figures 4C ‘input’ and S4D). Expression and secretion of hWnt3a^S209A from HEK293 cells without acylation was previously reported (Gao and Hannoush, 2014). As with xWnt8^S187A, hWnt3a^S209A signaling is inhibited by xFZD8 CRD (Figure 4B), and hWnt3a^S209A co-immunoprecipitates with the xFZD8 CRD (Figure 4C), albeit weakly. Thus, like xWnt8, Wnt3a appears to retain residual ability to activate signaling in the absence of acylation.

FIGURE 4. Effects of acylation site mutations in hWnt3a and mWnt5a.

(A) hWnt3a^S209A dorsalizes Xenopus embryos. Mean dorsal score data (left) are represented as in Figure 2A, for 3 biological replicates, with representative embryos shown at right (arrows as in Figure 1B).

(B) Signaling by hWnt3a^WT and hWnt3a^S209A is inhibited by xFZD8 CRD (n = 3), as described for xWnt8 in Figure 2C.

(C) Co-immunoprecipitation, as in Figure 2D, shows that hWnt3a^WT and hWnt3a^S209A both interact with xFZD8 CRD. The faint band in lanes 2 and 6 of the α-myc IP blot is α-myc heavy chain. Representative of at least three biological repeats.

(D) mWnt5a^S244A does not affect CE in Xenopus embryos. Data represent the percent of individuals injected with mWnt5a^WT or mWnt5a^S244A mRNA displaying CE defects at neurula stages, with the number of scored individuals (across 3 biological replicates) listed above each bar. Photographs show representative tailbud stage embryos injected with 0 pg and 100 pg mRNA.

(E) mWnt5a^WT and mWnt5a^S244A proteins are expressed at similar levels in mid gastrula stage embryos. See also Figure S4.

Acylation-dependence of Wnt5a

Wnt5a commonly regulates vertebrate morphogenesis through the planar cell polarity (PCP) pathway, a β-catenin independent signaling cascade (Nishita et al., 2010). During gastrulation, PCP signaling controls a series of coordinated cellular rearrangements, called convergent extension (CE). Overexpressing or depleting PCP components, including Wnt5a, causes well-established CE defects in Xenopus embryos (Deardorff et al., 1998; Djiane et al., 2000; Wallingford et al., 2000; Wallingford et al., 2001) as shown in Figure S1B. To ask whether acylation is required for Wnt5a to disrupt CE movements in Xenopus embryos, we expressed wild-type murine Wnt5a (mWnt5a^WT) or an acylation site mutant (mWnt5a^S244A: Figure S1C) from the upper dorsal marginal zone (DMZ) of four-cell stage embryos. Monitoring morphology at the neurula stage (post-gastrulation), we determined that mWnt5a^WT disrupted CE cell movements in a statistically-significant, dose-dependent manner between 10 and 100 pg of injected mRNA (p < 0.001; Figures 4D and S4E). By contrast, mWnt5a^S244A induced no significant CE defects (p = 0.189). Severe morphological defects beginning at the gastrula stage were observed when injecting 500 pg or 1000 pg of mWnt5a^S244A mRNA, making it difficult to assess CE defects at these mRNA doses (not shown). Western blotting of whole embryo lysates indicated that mWnt5a^S244A protein is expressed at least as well as mWnt5a^WT (Figure 4E). Thus, by contrast with Wnt8 and Wnt3a, Wnt5a appears highly dependent on acylation for signaling in Xenopus embryos.

In vitro analysis of acylation-independent Wnt signaling

Although less representative of in vivo effects, we also analyzed the ability of non-acylated Wnts secreted from Expi293 cells (Figures S2B and S3D) to activate TCF-dependent transcription in a commonly used luciferase reporter cell-line (Xu et al., 2004). Neither xWnt8^WT nor xWnt8^S187A from Epi293 cells reliably induced robust luciferase expression, consistent with previous reports that 293 cells produce inactive xWnt8 (Hsieh et al., 1999). By contrast, Expi293 cell-secreted wild-type mWnt1 or hWnt3a induced significant β-catenin-dependent Wnt signaling in the luciferase reporter line (Figure S4F). Introducing the S209A acylation site mutation in hWnt3a substantially blunted this response but, as for our in vivo studies, hWnt3a^S209A retained significant activity (p = 0.0015; Figure S4F). Interestingly, conditioned medium from Expi293 cells expressing mWnt1^S224A gave luciferase reporter responses similar to those seen with mWnt1^WT. These results further confirm the signaling capacity of non-acylated Wnts. Moreover, the contrast between the ability of mWnt1^S224A to promote signaling in 293 cells and its failure to do so in Xenopus embryos argues that requirements for Wnt acylation are highly context dependent.

DISCUSSION

Our studies of mutated xWnt8 confirm the importance of binding site 2 in xWnt8 binding to FZD8 (Figure 1A) but leave the contribution of site 1 less clear. In particular, we conclude that the acyl chain in site 1 is not universally essential for Wnt activity, biosynthesis or secretion as typically argued.

Considering Wnt biosynthesis first, the substantial responses seen when hWnt3a^S209A or xWnt8^S187A mRNAs are injected into Xenopus embryos demonstrate that acylation is not absolutely required for production of these proteins. Moreover, xWnt8^S187A activity is similar to that of xWnt8^WT in zebrafish embryos, and acylation-site mutations do not prevent xWnt8 secretion from mammalian cells. These findings, in addition to results from other studies, show that mutating the conserved serine-acylation site in Wnts does not abolish expression or secretion (Gao and Hannoush, 2014; Miranda et al., 2014; Rios-Esteves et al., 2014; Tang et al., 2012). It is important to note that the earliest reports arguing for an absolute requirement for acylation in Wnt processing employed cysteine mutations now known to disrupt disulfide bonds that are likely to be crucial for Wnt folding (MacDonald et al., 2014).

Arguably our most important finding is that the acyl chain is not universally required for FZD-CRD binding. Contrary to the suggested ‘linchpin’-like function for the acyl chain (Bienz and He, 2012), we find that its loss reduces signaling potency and apparent CRD binding strength by only ~10-fold for hWnt3a or xWnt8 – rather than the <10¹⁰ fold predicted based on burial of its hydrophobic surface. Our data suggest either that the proposed linchpin does not form – or (if it does) that it is not as energetically favorable as presumed from the crystal structure. One possibility is that the palmitoleoyl chain location in Figure 1A represents an adventitious docking mode for the Wnt thumb that occurs only: i). when the Wnt is acylated, and ii). when the membrane (or other docking site) is absent. Certainly, other ligands such as Norrin (Chang et al., 2015) and artificial ‘Wnt surrogates’ that cross-link FZDs with LRP5/6 (Janda et al., 2017) can activate FZD-mediated signaling without engaging the FZD-CRD hydrophobic channel with an acyl chain. Collectively, these data argue that the binding mode shown in Figure 1A is not absolutely required for FZD activation.

One possibility is that the acyl chain of a Wnt preferentially partitions into the adjacent lipid bilayer when the CRD is presented as part of the membrane-embedded intact FZD. It is clear that the hydrophobic channel on the FZD CRD has a strong preference for binding fatty acids or other lipid-like molecules. Indeed, a recent crystal structure of the hFZD7 CRD showed the hydrophobic channel occupied by a fatty acid that originated in the cells used for protein expression – which do not overexpress a Wnt (Nile et al., 2017). In this study and another (DeBruine et al., 2017), it was further shown that binding of free fatty acids induces FZD-CRD oligomerization. Along analogous lines, the crystal structure of the related Smoothened protein (Byrne et al., 2016) revealed an unexpected cholesterol molecule bound to the CRD. This cholesterol is clearly unrelated to that present in the cholesteroylated ligand Hedgehog, which binds Patched to regulate Smoothened only indirectly. It is possible that the acyl chains that bind FZD CRDs are similarly distinct from those attached to Wnt molecules – themselves bound primarily through site 2 interactions (Figure 1). Free fatty acid binding to FZDs may reflecting an as-yet-unappreciated regulatory role for lipid-like molecules in Wnt signaling, paralleling recent discoveries with Smoothened (Huang et al., 2018).

If, on the other hand, the Wnt-attached acyl chain is engaged directly by the FZD CRD, this interaction might play a specificity-determining role rather than being the central driver of all Wnt/FZD interactions. As shown for mini-xWnt8 by Janda et al. (2012), CRD engagement by the Wnt index finger at site 2 can define specificity to a significant extent – and our mutational data support a key role for site 2 in defining interaction strength. The FZD CRD hydrophobic channel is unlikely to be capable of distinguishing between Wnts to any appreciable extent (since they all bear similar acyl chains), but the balance between sites 1 and 2 in driving FZD binding could provide selectivity across the 19 mammalian Wnts and 10 FZDs. For the Wnts studied here, for example, site 2 interactions alone might be sufficient for Wnt3a and Wnt8 to fully engage the Xenopus FZDs expressed during the developmental windows of our experiments (which include FZD4) – making acylation dispensable. By contrast, site 2-mediated binding of Wnts 1 and 5a may not be sufficient in Xenopus embryos, and might require acyl chain-mediated contacts for productive binding. The fact that we see signaling by mWnt1^S224A in 293 cells further suggests that this can vary by FZD-expression context.

Although our data suggest that Wnts can signal without acylation, they demonstrate that acylation is nonetheless very important for Wnt potency. Beyond simply being required for FZD binding and activation, the Wnt acyl chain could control extracellular Wnt diffusion/range and local concentration. Wnts bind tightly to cell membranes in many contexts (Franch-Marro et al., 2008; Galli and Burrus, 2011; Ng et al., 2016; Pfeiffer et al., 2002), which can reflect partitioning of the acyl chain into membranes, Wnt interaction with cell-surface proteoglycans, and/or FZD binding (Farin et al., 2016) – possibly acyl chain-mediated in part. Association of xWnt8 with the periphery of Xenopus cells is lost when its acylation site is mutated (Figure S2C). Intriguingly, in vivo studies of acylation site-mutated Wingless (Wg^S239A) suggested that this defective ligand variant retains its ability to promote cell proliferation, and does so over a longer range than Wg^WT (Baena-Lopez et al., 2009). These observations are reminiscent of findings with the Drosophila EGFR ligand Spitz, where palmitoylation increases signaling activity by limiting the ligand’s spatial range in vivo. Recent studies investigating Wnt gradients suggest that Wnts must act over different ranges at different stages of development. Normal patterning in Drosophila embryos can be achieved by a form of Wg that is fused to an integral membrane protein (Alexandre et al., 2014), but later signals that may depend on longer-range Wg signaling seem to be defective. This longer-range signaling may be retained by Wg^S239A (Baena-Lopez et al., 2009), suggesting in turn that – as with Spitz – acylation may control range rather than receptor engagement.

The view of Wnt acylation that emerges from our findings raises an interesting potential axis for control of Wnt signaling that has not been discussed. Several recent studies have demonstrated that the secreted deacylase, Notum, enzymatically removes the acyl chain from Wnts extracellularly, yielding deacylated Wnts that may form inactive oligomers (Kakugawa et al., 2015; Zhang et al., 2015). Our finding that several Wnts retain some FZD binding and signaling capacity without acylation raises questions about whether Notum might also uncover – or promote – unique signaling properties of deacylated Wnts and/or modulate Wnt signaling range at different stages of development.

STAR×METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Mark Lemmon (mark.lemmon@yale.edu).

EXPERIMENTAL MODEL DETAILS

Xenopus laevis studies

Experiments were performed on developing Xenopus laevis embryos. Adult Xenopus were used only for egg procurement and in vitro fertilization. All protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania and conform to relevant regulatory standards. In vitro fertilization, microinjection and culture of Xenopus laevis embryos were performed as described (Sive et al., 2000). A non-isogenic adult frog colony was housed in an aquatic facility at the University of Pennsylvania overseen by University Laboratory Animal Resources (ULAR) under an IACUC approved protocol. Adult Xenopus were maintained in a recirculating aquatic system (Pentair) using deionized water supplemented with marine salts at 20˚C in fiberglass tanks at a density of no more than one adult per 4 l. Temperature, conductivity (1500 µS/cm), and pH (7–8) were monitored continuously and other water quality parameters were monitored weekly. Frogs were maintained in a 12 h light/12 h dark cycle and were fed exclusively Nasco adult frog pellets. Frog health was monitored daily and was overseen by ULAR veterinarian technicians and veterinarians. Egg laying was induced in wild-type adult females between 2 and 20 years of age via the sequential subcutaneous injection of 75 UI of pregnant mare serum gonadotropin (3–6 days pre-experiment) and 600 UI human chorionic gonadotropin (12–18 h pre-experiment) according to standard methods (Sive et al., 2000). Females were subsequently returned to the colony and rested for a minimum of 12 weeks before egg laying was re-induced. Wild-type adult male frogs were euthanized by immersion in 1.5 g/l MS-222 and testes were harvested for in vitro fertilization. Fertilized embryos were defolliculated using a solution of 3% w/v cysteine (pH 7.0). Healthy embryos from a single parental cross were randomly assigned to experimental groups and each embryo was considered an individual experimental unit. Embryos were maintained in 0.1 X MMR (10 mM NaCl, 0.2 mM KCl, 0.1 mM MgSO₄, 0.2 mM CaCl₂, 0.5 mM HEPES, pH 7.9) within 6-well plastic dishes at temperatures between 12–23^˚C. All experiments were carried out between the 2-cell and tailbud stages of development (Nieuwkoop and Faber stages 2–35). Specific age ranges are noted for each unique experiment in the Method Details. As these developmental windows precede the formation of sex-specific organs, no sex determination was performed. The number of animals used for each experimental condition (n) is noted within the figures included in the main text.

Danio rerio studies

Experiments in this study were performed on developing zebrafish embryos. Adult zebrafish of the TU wild-type strain were crossed in pairs to produce eggs that were used in experiments described here. All protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Pennsylvania and conform to relevant regulatory standards. Fish were housed in an aquatic facility at the University of Pennsylvania overseen by University Laboratory Animal Resources (ULAR) under an IACUC approved protocol, under standard husbandry conditions (13 h light, 11 h dark cycle). Eggs were collected for microinjection from male/female pairings at 15-minute intervals to ensure that their stages were consistent. Microinjection was performed on one-cell stage embryos as described (Westerfield, 2007).

Cell culture

Mycoplasma-free HEK293STF (ATCC Cat# CRL-3249, RRID:CVCL_AQ26) and Expi293 (RRID:CVCL_D615) cells (human female) were cultured at 37˚C in 5% and 8% CO₂ respectively in a humidified atmosphere. HEK293STF cells were grown adherent in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS) and 100 U/ml penicillin/streptomycin. Expi293 cells were grown in suspension in Expi293 Expression medium supplemented with 100 U/ml penicillin/streptomycin.

METHOD DETAILS

Plasmids and in vitro transcription

Wnt open reading frames and YFP were subcloned into pCS2 as described in Deardorff et al. (1998). The pCS2-memGFP plasmid was a gift from the laboratory of John Wallingford (Wallingford et al., 2000). pCS2-Venus (YFP) was constructed by cloning an NcoI/EcoRI fragment of pVenus-C1 (Clontech) into pCS2. pCS2-xWnt8 is described in Deardorff et al. (1998). Site-directed mutagenesis was used to introduce point mutations in xWnt8 site 1 and site 2. pcDNA-hWnt3A (Najdi et al., 2012) was a gift from Marian Waterman (Addgene plasmid # 35908). The open reading frame (start to stop codon only) of hWnt3A was amplified by PCR and subcloned into pCS2 using the restriction enzymes BamHI and XbaI. pCS2-mWnt1 was a gift from the laboratory of Jean-Pierre Saint-Jeannet (Saint-Jeannet et al., 1997). The mWnt1 open reading frame was PCR amplified and subcloned into pCS2 using restriction enzymes EcoRI and XbaI. pcDNA-mWnt5a was a gift from the laboratory of Ed Morrisey (University of Pennsylvania). The mWnt5a open reading frame (start to stop codon only) was PCR amplified and subcloned into pCS2 using the restriction enzymes BamHI and XhoI. Site-directed mutagenesis was used to create acylation site mutants of all Wnts. Capped mRNA was synthesized using the mMessage mMachine™ SP6 Transcription Kit (ThermoFisher AM1340) and purified using an RNeasy Mini Kit (QIAGEN).

Analysis of Wnt phenotypes in Xenopus laevis embryos

Xenopus embryos were obtained by in vitro fertilization as previously described (Sive et al., 2000) and cultured at room temperature in 0.1 x Marc’s Modified Ringer’s (MMR) buffer unless otherwise stated. All buffer recipes are described in the handbook by Sive et al. (Sive et al., 2000). Ectopic expression of Wnt constructs was achieved by microinjection of mRNAs. Groups of 20–30 individual embryos were placed in 0.5 X MMR, 3% Ficoll at 12˚C, and microinjected at the four-cell stage. All mRNAs were co-injected with 300 pg of Venus YFP mRNA, and negative control groups were injected with YFP mRNA only. Biological replicates were performed using embryos from unique parental crosses. At the gastrula stage, embryos were sorted for YFP expression (as a positive control for successful mRNA injection) under a fluorescence dissecting microscope. Discrepancies between sample sizes at the beginning and end of each analysis are due to attrition caused by either a failure of embryos to demonstrate YFP fluorescence (resulting in embryos being discarded from the analysis) or the death of embryos before phenotypes were analyzed. All YFP negative individuals were discarded from downstream analyses. For Wnts 1, 3a and 8, 10 nl of mRNA was injected into the lower marginal zone (4 o’clock) of one ventrally-fated blastomere (ventral marginal zone, VMZ). At tailbud stages, a numerical score was assigned to individual embryo phenotypes (normal phenotype = 0, partial axis duplication = 1, full axis duplication = 2, radial duplication = 3). For each treatment, phenotype scores from individuals across three (or more) biological replicates were summed. Dividing by the total number of individuals analyzed yielded a mean dorsal score, which falls on a 0 – 3 scale. For mWnt5a, 5 nl of mRNA was injected in the upper marginal zone (2 o’clock) of both dorsally-fated blastomeres. Embryo phenotypes were sorted for YFP expression and analyzed for convergent extension defects at the mid-neurula stage. For each treatment, the number of embryos displaying convergent extension defects across three biological replicates was summed and represented as percent of the total individuals analyzed.

Analysis of Wnt phenotypes in zebrafish embryos

To evaluate dorsalizing activity of xWnt8^WT and xWnt8^S187A in D. rerio embryos, mRNAs were microinjected into embryos at the one-cell stage. Control embryos were uninjected. Dorsalization phenotypes were assessed at 24 h post fertilization using the dorsalization scale as described (Mullins et al., 1996). Embryos were collected and fixed in 4% paraformaldehyde in PBS. Embryos were unsexed, as all assays were performed prior to sex determination.

RT-PCR

Embryos used for the analysis of Wnt target gene transcription were cultured and microinjected using the methods described above. Ten YFP-positive embryos were flash-frozen at the early gastrula stage (Nieuwkoop and Faber stage 10). mRNA was isolated from embryos using an RNeasy Mini Kit (QIAGEN). Reverse transcription was performed using M-MLV Reverse Transcriptase (ThermoFisher) with 2 μg template mRNA for 1 h at 42˚C. Real-time PCR was performed with Power SYBR™ Green PCR Master Mix (ThermoFisher) for 30 cycles using the primers listed below. Ornithine decarboxylase (ODC) served as a reference gene. Data were analyzed using the Comparative C_T Method (ΔΔC_T) of qPCR analysis. Due to variability in the magnitude of fold changes across clutches from different parental crosses, ΔΔC_T values were calculated by normalizing test samples to a positive control. For dose-response experiments comparing activity of wild-type and acylation-site-mutated Wnt (Figures 1D, 2B, 3B and S4B), the ΔC_T value from the sample injected with the highest dose of WT mRNA was used as the calibrating sample for each biological replicate (500 pg of xWnt8^WT and mWnt1^WT, 100 pg of hWnt3a^WT). For co-injection experiments with Wnt and xFZD8 CRD mRNA (Figures 2C, 4B and S3B), the ΔC_T value of the sample injected with 5 pg Wnt^WT mRNA was used as the calibrating sample.

Primers used were (Blythe et al., 2010; Iwasaki and Thomsen, 2014; Skirkanich et al., 2011):

ODC_forward: 5’ – GATCATGCACATGTCAAGCC – 3’

ODC_reverse: 5’ – TCTACGATACGATCCAGCCC – 3’

Siamois_forward: 5’ – CTGTCCTACAAGAGACTCTG – 3’

Siamois_reverse: 5’ – TGTTGACTGCAGACTGTTGA – 3’

Xnr3_forward: 5’ – CTGGAGTCACCACAAATCTACCCAGA – 3’

Xnr3_reverse: 5’ – AGGCATCGCCATCAGTGGGG – 3’

xWnt8 antibody development

Peptides A (aa21 – 119) and B (aa197 – 308) from xWnt8 were expressed in E. coli as maltose binding protein (MBP) fusion proteins from the pMAL-c2X vector (New England BioLabs). Fusion proteins were produced using standard methods. Immunization of rabbits with both peptides A and B was performed by Cocalico Biologicals (Stevens, PA). The resulting antisera reacted with peptide B only. The antibody was purified and concentrated from antisera using the EpiMAX Affinity Purification Kit (Abcam) with immobilized Peptide B-fused glutathione-S-transferase (GST) fusion protein as bait. The antibody recognizes Xenopus Wnt8 but not murine Wnt8.

Western blotting of Xenopus embryo lysates

Embryos used for the analysis of Wnt protein expression were cultured and microinjected using the methods described above. Five YFP-positive embryos were flash-frozen at the mid gastrula stage (Nieuwkoop and Faber stage 10.5). Embryos were homogenized through a P200 pipet tip in 10 μl/embryo lysis buffer (20 mM Tris, pH 7.5, 1% Triton X-100, 140 mM NaCl, 10% glycerol, 10 mM EDTA, 1 mM DTT) supplemented with protease inhibitor cocktail (Sigma P8340, 1:100 dilution). Lysates were centrifuged at 1000 x g for 5 min to remove yolk protein, and subsequently at 14,000 rpm for 10 minutes to remove cellular debris. The lipid-free layer of supernatant was recovered and 1–2 embryo equivalents were loaded on to a 10% SDS-PAGE gel. Protein gels were run at 100 V and transferred to a nitrocellulose membrane for 2 h at 350 mA. Membranes were blocked for 1 h in bløk™-FL Fluorescent Blocker (Millipore), and probed with antibodies against β-tubulin (BD Pharmingen 556321, 1:1000), xWnt8 (see above, 0.14 μg/ml), Wnt1 (Abcam 15251, 1:1000), Wnt3a (Abcam 28472, 20 μg/ml) and Wnt5a (Cell Signaling Technology C27E8, 1:1000). Blots were incubated with fluorescently-conjugated secondary antibodies (LI-COR, 1:50,000) for 1 h, washed for at least 2 h (optimally overnight) in TBS-T, and visualized on a LI-COR Odyssey scanner.

Immunoprecipitation from Xenopus embryo lysates

5 nl of a mixture of mRNAs encoding the relevant Wnt variant and xFZD8 CRD-myc was injected into both animal blastomeres of two-cell stage embryos with YFP mRNA as a control. For studies with xWnt8, 10 YFP-positive embryos were frozen for processing from each treatment group at the mid gastrula stage (stage 10.5). Embryos were resuspended in 100 μl immunoprecipitation (IP) buffer (25 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100 and 2 mM EDTA) supplemented with protease inhibitor cocktail (Sigma, 1:100 dilution) and homogenized through a P200 pipet tip. Lysates were centrifuged at 1000 x g for 5 min to remove yolk protein, transferred to a new Eppendorf tube, and subsequently centrifuged at 14,000 rpm for 10 min to remove cellular debris. One embryo equivalent of whole cell lysate was collected as an “Input” sample and 7 embryo equivalents were used for IP. Each IP sample was pre-cleared at 4˚C for 1 h with 10 μl settled protein-G agarose (ThermoFisher Scientific) and incubated, rotating, overnight at 4˚C with 1 μg 9E10 α-myc antibody (UPenn Cell Center). Lysates were subsequently incubated for 2 h at 4˚C with 10 μl settled protein-G agarose. Beads were washed for 10 min in 700 μl cold IP buffer three times. Beads were subsequently resuspended in 15 μl of standard 2 X SDS loading buffer, and incubated for 5 min on a 95˚C heat block. 12 μl of IP eluate was then run on a 10% SDS-PAGE gel alongside “Input” samples. Gels were run and transferred as described above. For hWnt3a IPs, twice the volume of Input and IP samples were loaded on gels. For Wnt1 IPs, 2 μg of 9E10 α-myc antibody was used for the overnight immunoprecipitation. Both protocols were otherwise identical to that described for xWnt8. Membranes were blocked for 1 h in bløk™-FL Fluorescent Blocker (Millipore), and probed with antibodies against xWnt8 (see above, 0.14 μg/ml), Wnt1 (Abcam 15251, 1:250), Wnt3a (Abcam 28472, 40 μg/ml). The mouse 9E10 α-myc antibody (UPenn Cell Center, 1:1000) was used to detect xFZD8 CRD-myc expression in Input samples. Rabbit α-myc (Cell Signaling 2272, 1:1000) was used to detect xFZD8 CRD-myc in immunoprecipitates, as this protein runs at the same molecular weight as the 9E10 heavy chain. Blots were incubated with fluorescently-conjugated secondary antibodies (LI-COR, 1:20,000) for 1 h, washed for at least 2 h in TBS-T, and visualized using a LI-COR Odyssey scanner.

Immunofluorescence

For the immunofluorescence studies shown in Figure S2C, two-cell stage embryos were injected with 5 nl of 100 ng/μl memGFP mRNA in both blastomeres. At the 32-cell stage, embryos were again injected with 5 nl of 200 ng/μl mRNA encoding xWnt8^WT or xWnt8^S187A in one cell closest to the animal pole. At the late blastula stage (stage 9), animal hemisphere tissues (animal caps) were surgically removed, fixed, bleached and stained (Lee et al., 2008). Antibodies were used to detect GFP (Abcam 13970, 1:2000) and xWnt8 (see above, 6.9 μg/ml), with secondary antibodies Alexa Fluor 488 goat α-chicken for GFP (ThermoFisher Scientific A-11039) and Alexa Fluor 594 goat α-rabbit for xWnt8 (ThermoFisher Scientific A-11012). Imaging was performed using a Leica TCS SP8 Confocal microscope. Scale bars were added to images using Fiji. The ‘levels’ function in Adobe Photoshop was used to perform linear contrast stretching equally in all images for clarity of viewing.

In situ hybridization in D. rerio embryos

One-cell stage zebrafish embryos were microinjected with either xWnt8^WT or xWnt8^S187A mRNA and were allowed to develop until the bud stage. Control embryos remained uninjected. Whole-mount in situ hybridization was performed as described (Hashiguchi and Mullins, 2013), with probes for otx2b (Li et al., 1994) and sizzled. Embryos were mounted in 30% glycerol and illuminated with a gooseneck light source for photography on a Leica MZ12.5 dissecting microscope.

Wnt production in vitro

Wnt-containing conditioned medium was prepared by expressing wild-type or acylation-site-mutated Wnt proteins using the Expi293™ Expression System (ThermoFisher) according to procedures recommended by the manufacturer. Expi293 cells were transfected with plasmids encoding wild-type or mutated Wnt under control of a CMV promoter. Culture medium was harvested 96 h post-transfection and cleared by centrifugation. For luciferase assays, the supernatant was then mixed 1:1 with DMEM/HamsF12 1:1 growth media containing 10% FBS. Successful production and secretion of Wnts was determined by Western blot. Conditioned supernatant was analyzed by immunoblotting with the above-mentioned antibodies for xWnt8 and Wnt1 (1:1000) and peroxidase-conjugated goat anti-rabbit IgG secondary antibody (MP Biomedicals 55689), with detection by chemiluminescence using SuperSignal West Pico substrate (ThermoFisher), detected with a LI-COR Odyssey Fc Imaging System.

Luciferase assays for in vitro Wnt activity

For analyzing β-catenin dependent signaling in vitro we used a previously described luciferase reporter cell-line (Xu et al., 2004) comprising HEK 293 cells stably transfected with a plasmid containing 7 x LEF/TCF sites and a minimal thymidine kinase promoter that drives expression of firefly luciferase (HEK 293 STF). The reporter cells were obtained from ATCC (CRL-3249), and sub-cultured as recommended. For luciferase assays, cells were seeded at a density of 2.5 × 10⁴ cells per well in 96-well plates and allowed to settle for 24 h before the culture medium was replaced with Wnt-containing conditioned medium prepared as described above. After 20 h of subsequent Wnt stimulation, cells were lysed and luminescence determined using the One-Glo™ luciferase assay system (Promega) according to the manufacturer’s instructions with a BioTek Synergy 2 plate reader.

QUANTIFICATION AND STATISTICAL ANALYSIS

Gene expression and phenotype data were analyzed with STATA statistical software and graphical representations were generated using GraphPad Prism. The phenotype data in Figures 2A, 3A, and 4A were analyzed using a mixed models regression with mean dorsal score as a continuous variable, clustering by genotype. Phenotype data in Figure 4D were analyzed using the same model with mean scores calculated using a modified scale: 0 = no phenotype, 1 = convergent extension defect. For each data set, the statistical significance of the interaction between predictor variables Wnt type (wild-type or acylation mutant Wnt) and dose (in picograms) was analyzed and two-sided p values are reported in the main text. All RT-PCR data were analyzed using a fixed effects regression model treating the predictor variable treatment (dose of wild-type or mutated Wnt mRNA) as categorical in order to account for clutch-to-clutch variation in gene expression maxima. Activation of Wnt-specific gene expression by the site 1 and site 2 xWnt8 mutants (Figure 1D) was analyzed using control samples (YFP mRNA-injected embryos) as the reference group. Wnt 8, 1 and 3a dose curves (Figures 2B, 3B and S4B) were also fit using the control-injected samples as a reference group and subsequent pair-wise comparisons were performed using a Wald chi-squared test with (1, 24) degrees of freedom. The effect of xFz8 CRD co-expression on Wnt-dependent transcription (Figures 2C, 4B and S3B) was analyzed with the WT alone treatment group as a reference group. Each data set was first analyzed for interaction between the two predictor variables – Wnt type (WT or acylation mutant Wnt) and inhibitor (CRD) presence – but no significant interaction was found. Final analysis was performed using a regression model without interaction. Two-sided p values were used to determine statistical significance, which was set as follows: ‘ns’ denotes p ≥ 0.05, *p ≤ 0.05, **p ≤ 0.01 and ***p ≤ 0.001. For luciferase reporter assays, significance of differences between means was determined using a two-tailed paired sample t-test, with p values reported as above.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
β-tubulin	BD Pharmingen	Cat#: 556321
xWnt8	Cocalico Biologicals	Custom antibody, this study
Wnt1	Abcam	Cat#: 15251
Wnt3a	Abcam	Cat#: 28472
Wnt5a	Cell Signaling Technology	Cat#: C27E8
Myc 9E10	UPenn Cell Center	Cat#: 3207
Rabbit α-Myc	Cell Signaling	Cat#: 2272
GFP	Abcam	Cat#: 13970
Donkey α-Mouse 680 secondary	LI-COR	Cat#: 925–68072
Donkey α-Rabbit 800CW secondary	LI-COR	Cat#: 925–32213
Alexa Fluor 488 Goat α-Chicken IgY (H+L) secondary	ThermoFisher Scientific	Cat#: A-11039
Alexa Fluor 594 Goat α-Rabbit IgG (H+L) secondary	ThermoFisher Scientific	Cat#: A-11012
Goat-anti-rabbit IgG, HRP, secondary	MP Biomedicals	Cat#: 55689
Chemicals, Peptides, and Recombinant Proteins
Protease Inhibitor Cocktail	Sigma	Cat#: P8340
bløk™-FL Fluorescent Blocker	Millipore	Cat#: WBAVDFL01
Recombinant Protein G Agarose	ThermoFisher Scientific	Cat#: 15920010
One-Glo™ luciferase assay system	Promega	Cat#: E6120
SuperSignal West Pico substrate	ThermoFisher Scientific	Cat#: 34077
Critical Commercial Assays
mMessage mMachine™ SP6 Transcription Kit	ThermoFisher Scientific	Cat#: AM1340
RNeasy Mini Kit	QIAGEN	Cat#: 74106
M-MLV Reverse Transcriptase	ThermoFisher Scientific	Cat#: 28025013
Power SYBR™ Green PCR Master Mix	ThermoFisher Scientific	Cat#: 4367659
EpiMAX Affinity Purification Kit	Abcam	Cat#: ab138915
ONE-Glo™ EX Luciferase Assay System	Promega	Cat#: E8130
Experimental Models: Cell Lines
HEK293 luciferase reporter cell-line (HEK293STF)	Xu et al., 2004	ATCC: CRL-3249
Expi293™ cells and expression system	ThermoFisher Scientific	Cat#: A14635
Experimental Models: Organisms/Strains
Xenopus laevis adult males and females	Nasco	LM00531MX
Zebrafish: TU wild-type strain, embryos	Zebrafish International Resource Center	RRID:ZIRC_ZL57
Oligonucleotides
ODC_forward: 5’-GATCATGCACATGTCAAGCC-3’	Skirkanich et al., 2011	ODC forward
ODC_reverse: 5’-TCTACGATACGATCCAGCCC-3’	Skirkanich et al., 2011	ODC reverse
Siamois_forward: 5’-CTGTCCTACAAGAGACTCTG-3’	Iwasaki and Thomsen, 2014	siamois U
Siamois_reverse: 5’-TGTTGACTGCAGACTGTTGA-3’	Iwasaki and Thomsen, 2014	siamois D
Xnr3_for: 5’-CTGGAGTCACCACAAATCTACCCAGA-3’	Blythe et al., 2010	Xnr3 5’: Forward
Xnr3_rev: 5’-AGGCATCGCCATCAGTGGGG-3’	Blythe et al., 2010	Xnr3 5’: Reverse
Recombinant DNA
pCS2-memGFP	Wallingford et al., 2000	memEGFP
pVenus-C1	Clontech	N/A
pCS2-xWnt8	Deardorff et al., 1998	xWnt8
pcDNA-hWnt3a	Najdi et al., 2012	Addgene Cat#: 35908
pCS2-Wnt1	Saint-Jeannet et al., 1997	mWnt1
pcDNA-mWnt5a	Gift from Ed Morrisey, U. Penn	N/A
pMAL-c2X	New England Biolabs	Cat#: #N8076S
Software and Algorithms
STATA	https://www.stata.com/	N/A
Fiji	https://fiji.sc/	N/A
Adobe Photoshop	https://adobe.com	N/A
GraphPad Prism	https://www.graphpad.com/scientific-software/prism/	N/A