The CK1α Activator Pyrvinium Enhances the Catalytic Efficiency (k_cat/K_m) of CK1α

Abstract

The serine/threonine protein kinase casein kinase 1α (CK1α) functions as a negative regulator of Wnt signaling, phosphorylating β-catenin at serine 45 (P–S45) to initiate its eventual ubiquitin-mediated degradation. We previously showed that the repurposed, FDA-approved anthelminthic drug pyrvinium potently inhibits Wnt signaling in vitro and in vivo. Moreover, we proposed that pyrvinium’s Wnt inhibitory activity was the result of its function as an activator of CK1α. An understanding of the mechanism by which pyrvinium activates CK1α is important because pyrvinium was given an orphan drug designation by the FDA to treat familial adenomatous polyposis, a precancerous condition driven by constitutive Wnt signaling. In the current study, we show that pyrvinium stimulates the phosphorylation of S45 β-catenin, a known CK1α substrate, in a cell-based assay, and does so in a dose- and time-dependent manner. Alternative splicing of CK1α results in four forms of the protein with distinct biological properties. We evaluated these splice products and identified the CK1α splice variant, CK1αS, as the form that exhibits the most robust response to pyrvinium in cells. Kinetic studies indicate that pyrvinium also stimulates the kinase activity of purified, recombinant CK1αS in vitro, increasing its catalytic efficiency (k_cat/K_m) toward substrates. These studies provide strong and clear mechanistic evidence that pyrvinium enhances CK1α kinase activity.

Casein kinase 1α (CK1α) negatively regulates the Wnt signaling pathway to suppress the initiation and progression of a subset of cancers.¹ In the absence of Wnt, CK1α forms a “destruction complex” with other Wnt signaling components, including adenomatous polyposis coli (APC), axin, and glycogen synthase kinase 3 (GSK3), which recruits cytoplasmic β-catenin. Subsequent to recruitment, β-catenin is phosphorylated by CK1α on serine 45 (P-S45), which creates a phospho-substrate recognition site for GSK3.^2–7 This dual phosphorylation of β-catenin results in its ubiquitin-mediated proteasomal degradation.^8–10 Upon Wnt activation, the destruction complex is inhibited, and β-catenin accumulates and enters the nucleus.¹¹ β-catenin associates with the T cell factor (TCF) family of transcription factors and lymphoid enhancer-binding factor 1 (LEF1) in the nucleus to drive a Wnt transcriptional program.^12,13 A variety of other transcriptional cofactors, including B cell CLL/lymphoma 9 (BCL9) and pygopus, modulate this Wnt transcriptional response.^14–16

The anthelmintic drug pyrvinium was initially identified as a small-molecule Wnt inhibitor in a large-scale screen of FDA-approved drugs capable of attenuating Wnt activity in Xenopus laevis egg extracts.¹⁷ Pyrvinium was shown to specifically activate CK1α to inhibit Wnt-driven activity in multiple colorectal cancer cell lines.¹⁸ This work suggested a model in which pyrvinium inhibited Wnt signaling by promoting the activity of CK1α, which resulted in the reduced stability of β-catenin and pygopus.^17,18 Pyrvinium also attenuated Wnt activity in vivo, inhibiting the growth of intestinal adenomas driven by constitutive Wnt activity in a mouse model of familial adenomatous polyposis (FAP).¹⁸

Although numerous publications have now utilized pyrvinium as a Wnt inhibitor,^19–22 the mechanism by which it does so has remained controversial. It was also suggested that pyrvinium functions as a Wnt inhibitor via a mechanism that does not involve activation of CK1α.²³ This work was unable to show that pyrvinium could activate CK1α, using either a cell-based assay or an in vitro protein kinase assay. Thus, we determined to perform a more detailed biochemical analysis of the activation of CK1α by pyrvinium in order to try and resolve this controversy.

To examine the potential activation of cellular CK1α by pyrvinium, we first treated HEK293T cells with different concentrations of pyrvinium, or the chemically similar, inactive pyrvinium analogue VU-WS211 (Figure 1A),¹⁷ and evaluated the phosphorylation of a known CK1α substrate, S45 β-catenin, using a phospho-specific antibody. Pyrvinium increased P–S45 β-catenin levels relative to total β-catenin and did so in a dose-dependent manner (Figure 1B). Consistent with pyrvinium acting in a specific manner, VU-WS211 treatment did not increase P–S45 β-catenin levels. Following treatment with pyrvinium, the levels of P–S45 β-catenin plateaued at approximately 500 nM, with an EC₅₀ of ~20 nM (Figure 1C). We next treated HEK293T cells with two different concentrations of pyrvinium or VU-WS211 for different periods of time and analyzed cellular lysates for P–S45 β-catenin. Pyrvinium increased P–S45 β-catenin levels in a time-dependent manner, which peaked at different times (30 or 60 min) depending on the dose of pyrvinium used, approximately 3-fold higher than the vehicle or VU-WS211 treated cells (Figure 1D–G). However, in the presence of the higher dose (500 nM) of pyrvinium, this increase in phosphorylated β-catenin was transient, returning to basal levels after 1 h.

Figure 1.

CK1α is activated by pyrvinium in HEK293T cells. (A) Chemical structures of pyrvinium and VU-WS211. (B) HEK293T cells were treated with increasing concentrations of pyrvinium or VU-WS211 for 30 min. (D, F) HEK293T cells were treated with pyrvinium or the inactive pyrvinium analogue VU-WS211 at 100 nM (D) or 500 nM (F) for the indicated time. Cell lysates were resolved by SDS-PAGE and subjected to immunoblotting for the indicated proteins. Representative immunoblots are shown (n = 3). (C, E, G) Quantification (mean ± S.E.M., n = 3) of panels B, D, and F, respectively, was performed using Li-Cor Image Studio software. CK1α activity indicates the level of P–S45 β-catenin normalized to that of total β-catenin. An asterisk indicates statistical significance (Student’s t test, p value ≤ 0.05).

CSNK1α1 (the gene encoding CK1α) encodes two alternative exons, a long insert of 28 amino acids (L) within the kinase domain and a short insert of 12 amino acids (S) within the C-terminus. Thus, alternative splicing can result in four proteins with distinct biological properties: a form that lacks either insert (CK1αNI), a form expressing both inserts (CK1αLS), and two forms expressing only one of the two inserts (CK1αS or CK1αL) (Figure 2A).^24—26 It was reported that CK1α containing the L insert is less sensitive to CKI-7, a small-molecule CK1 inhibitor, compared to CK1αNI and CK1αS.²⁶ Thus, we also investigated whether the CK1α splice variants exhibit a differential response to pyrvinium. Plasmids encoding the four HA-tagged CK1α splice variants were transfected into HEK293T cells. We observed that CK1αS, CK1αL, and CK1αLS exhibited comparable abilities to phosphorylate β-catenin in cells, while CK1αNI was approximately 70% more active (Figure 2B,C). This finding is consistent with previous work showing that CK1αNI is the most active splice variant.²⁵ These CK1α splice variant expressing cells were subsequently treated with pyrvinium for 30 min. Interestingly, only two of the four CK1α splice variants appeared to be pyrvinium-responsive in this assay, CK1αNI and CK1αS, with CK1αS exhibiting the most robust activation by pyrvinium (Figure 2B,D). However, the differential basal activity of CK1αNI and CK1αS may also contribute to this difference in pyrvinium fold activation.

Figure 2.

Pyrvinium activates the CK1α splice variant CK1αS. (A) Schematic of CK1α splice variants. (B) Plasmids expressing HA-tagged CK1α splice variants were transfected into HEK293T cells for 48 h, followed by treatment with DMSO or 200 nM pyrvinium for 30 min. Cell lysates were resolved by SDS-PAGE and subjected to immunoblotting for the indicated proteins. Representative immunoblots are shown (n = 3). (C, D) Quantification (mean ± S.E.M., n = 3) of CK1α activity in response to each indicated splice variant from cells treated with a vehicle (DMSO) or pyrvinium is shown. CK1α activity indicates the level of P–S45 β-catenin normalized to that of total β-catenin and then normalized to that of HA-CK1α. An asterisk indicates statistical significance (Student’s t test, p ≤ 0.05), and NS indicates no statistical significance (Student’s t test, p > 0.05).

Since CK1αS was most responsive to pyrvinium in cell cultures, we used a commercially available recombinant, GST-tagged form of CK1αS to determine its mechanism of activation in vitro. We first determined the inherent enzymatic properties of this CK1αS protein and showed that the reactions proceeded in a linear fashion for at least 90 min (Figure 3A) when using up to 40 nM CK1αS (Figure 3B). We then determined the K_m values for α-casein and ATP, which were 2.0 μM and 5.1 μM, respectively (Figure 3C,D), consistent with previous studies.²⁷

Figure 3.

Enzymatic characteristics of recombinant CK1αS. The protein kinase activity of recombinant GST-CK1αS (CK1αS) was determined as indicated in each panel. Kinase reactions were performed with 40 nM CK1αS, 50 μM α-casein, and 50 μM ATP at 30 °C unless otherwise indicated. (A) Time dependence. (B) CK1αS concentration dependence at 30 min. (C) α-Casein dependence at 30 min. (D) ATP dependence at 30 min. Representative figures are shown (n = 3). Error bars indicate the range of kinase activity in duplicate reactions. An asterisk indicates statistical significance (Two-way Anova, p ≤ 0.05).

We next examined the kinetic mechanism of the action of pyrvinium. Pyrvinium increased CK1αS kinase activity in a time-dependent manner. In the presence of the inactive control analogue, VU-WS211, CK1αS showed activity similar to the vehicle control (Figure 4A). We also noted that the DMSO (5%) required to solubilize pyrvinium exhibited a small inhibition of CK1αS kinase activity (compare Figure 3A to 4A). We next performed the CK1αS kinase reaction with varying concentrations of α-casein or ATP, in the presence of DMSO, pyrvinium, or VU-WS211. Pyrvinium increased CK1αS kinase activity in a manner dependent on the concentration of α-casein or ATP, saturating at around 60 μM of α-casein and 100 μM of ATP, respectively (Figure 4B,D). To more accurately determine the effect of pyrvinium on the K_m and V_max of CK1α, we modeled the data using an Eadie-Hofstee analysis (Figure 4C,E). In the presence of pyrvinium, the V_max of the CK1α reactions was increased by 50% with α-casein and 15% ATP, the K_m of both substrates was not changed. We then summarized CK1α enzymatic parameters from three independent experiments (Table 1). These results suggest that pyrvinium enhances CK1α’s catalytic efficiency (k_cat/K_m) by increasing the V_max and k_cat of the reaction, without altering the K_m. VU-WS211 exhibited no significant effect on kinase activity.

Figure 4.

Pyrvinium increases the enzymatic efficiency of CK1αS. The protein kinase activity of recombinant CK1αS (40 nM) was determined in the presence of DMSO, 200 nM pyrvinium, or 200 nM VU-WS211 at 30 °C (A) at the indicated time points. (B, D) At the indicated concentrations of α-casein (B) or ATP (D) at 60 min. (C, E) The data generated in panels B and D was transformed into an Eadie-Hofstee plot, respectively. Representative figures are shown (n = 3). Error bars indicate the range of kinase activity in duplicate reactions. An asterisk indicates statistical significance (Two-way Anova, p ≤ 0.05).

Table 1.

Summary of CK1αS Enzymatic Parameters

Enzymatic parameters	α-Caaein			ATP
	DMSO	Pyrvinium	VU-WS211	DMSO	Pyrvinium	VU-WS211
Vmax (pmol·min−1)	9.39 ± 0.49	13.62 ± 0.51 *	9.91 ±0.22	11.35 ± 0.14 NS	12.87 ±0.25 *	11.47 ± 0.20
Km (μM)	5.26 ± 0.84	6.51 ± 0.95	8.83 ± 1.18	14.02 ± 0.86	15.50 ± 0.18	14.12 ± 0.52
kcat(min−1)	4.69 ± 0.24	6.81 ± 0.26 *	4.95 ± 0.11	5.68 ± 0.07 NS	6.44 ±0.13 *	5.73 ± 0.10
kcat/Km (μM−1·min−1)	0.88 ± 0.12	1.15 ± 0.13 *	0.66 ± 0.18	0.38 ± 0.01	0.43 ± 0.01 *	0.40 ± 0.01

^*:p value ≤ 0.05, compared to Group “α-Casein, DMSO”,

^NS:p value >0.05, compared to Group “α-Casein, DMSO”,

^*:p values ≤ 0.05, compared to Group “ATP, DMSO”, Student’s t-test.

Summarized enzymatic parameters of CK1αS (mean ± S.E.M.) were calculated from the mean ± SD (n = 2) of three independent biological replicates. Representative data of such experiments are shown in Figure 4. An asterisk indicates statistical significance compared to the relevant DMSO control (Student’s t test, p ≤ 0.05), and NS indicates no statistical significance (Student’s t test, p > 0.05).

During our studies of CK1α activation by pyrvinium, we identified several potential reasons for the discrepancies between our findings and the contrary findings of Venerando et al.²³ (1) We found that the activation of CK1α by pyrvinium in cells is a rapid event, and Venerando et al. only examined pyrvinium’s ability to activate CK1α after 16 h.²³ (2) We observed that pyrvinium’s ability to induce β-catenin phosphorylation varied depending on the confluence of HEK293T cells, consistent with previous work examining β-catenin activation,²⁸ and note that Venerando et al. did not mention the confluence status of their cells. (3) Our work indicated that CK1α splice variants exhibit differential responses to pyrvinium in cellular assays. Although Venerando et al. did not specify the CK1α spliced variant used in their protein kinase assays, it remains possible that the spliced variant of CK1α they used is not activated by pyrvinium (see Figure 2).

While the number of small-molecule protein kinase activators described is limited, their ability to activate key regulatory enzymes, whose activity is down-regulated under pathological circumstances, highlights their therapeutic potential.^18,29–32 Several of these protein kinase activators are already in clinical use, including metformin, a 5′ AMP-activated protein kinase (AMPK) activator, and bryostatin, a protein kinase C (PKC) activator.^33,34 As a class, these small-molecule protein kinase activators typically function by improving the catalytic efficiency of their target kinase, changing its V_max, K_m, or both.³⁵ For example, the synthetic AMPK activator A-769662 enhances its catalytic efficiency by increasing its V_max and decreasing its K_m for its peptide substrate. Alternatively, the endogenous AMPK activator AMP increases the V_max of AMPK without altering its K_m.³⁶ Similarly, we show that pyrvinium increases the V_max of CK1α but does not modulate its K_m for either α-casein or ATP. This suggests that pyrvinium is activating CK1α by inducing a conformational change. In our previous work, we showed that pyrvinium-bound CK1α exhibits a distinct pattern upon a limited trypsin digest, consistent with pyrvinium inducing a conformational change in CK1α upon binding.¹⁷

We show here that CK1α splice variants without the L insert exhibit greater sensitivity to pyrvinium than those with the L insert. The L sequence inserts after amino acid 152 of CK1α, between β-strands 7 and 8, which form the back of the active site, and very close to the activation loop (DFG—SIN, aa 156–190).^25,37 The reduced effect of pyrvinium on forms with the L insert suggests that the L insert blocks access of the small molecule to the active site. The β-strands and activation loop appear pivotal to the function of pyrvinium through either a direct interaction or via an allosteric mechanism.

Taken together, our work shows that pyrvinium potently activates CK1α in cells and does so by enhancing the catalytic efficiency (k_cat/K_m) of CK1α. The mechanism of action of pyrvinium will provide the rationale for the development of other CK1α activators that target Wnt signaling in various human diseases. On the basis of the findings presented here, we suggest that Wnt-dependent tumors or cell lines expressing the CK1αS variant may be the most vulnerable to treatment with pyrvinium.