Wntless degradation mediated by CaMKII inhibition drives endoplasmic reticulum stress and apoptosis

SUMMARY

Noncanonical Wnt signaling stimulates calcium release and subsequent activation of calcium/calmodulin-dependent protein kinase 2 (CaMKII). We find that CaMKII inhibition decreases Wnt synthesis and identify the downregulation of Wntless (WLS) as the basis for this decrease. WLS transports palmitoylated Wnts to the cell surface, and in the absence of Wnts, it undergoes endoplasmic reticulum (ER)-associated degradation (ERAD). CaMKII inhibition increases proteasome-dependent WLS degradation, which can be suppressed by Wnt overexpression, indicating that CaMKII contributes to ER handling of unliganded WLS and prevention of ERAD. CaMKII inhibition also causes a loss of calnexin, an ER stress response through the activation of PERK, and apoptosis in cells expressing high levels of WLS. Moreover, these effects can be prevented by WLS knockdown. The findings reveal a previously unidentified role for CaMKII in supporting ER chaperone functions. They further show that high levels of unliganded WLS could be a potent driver of ER stress in the presence of CaMKII inhibitors, which may have efficacy in tumors expressing high levels of WLS.

In brief

Gu et al. identified a CaMKII function in ER homeostasis, with CaMKII inhibition causing degradation of unliganded Wntless (WLS), which was associated with a loss of calnexin, PERK-driven unfolded protein response, and apoptosis. Tumor cells expressing high levels of WLS may be vulnerable to CaMKII-targeted therapies.

Graphical abstract

INTRODUCTION

Dysregulation of Wnt signaling is a hallmark of many cancers, with frequent genetic alterations in various Wnt pathway components that result in ligand-dependent or independent hyperactivation of the pathway.^1–3 Canonical Wnts signal through a receptor complex consisting of a seven-pass transmembrane frizzled (FZD) protein and the single-pass LRP5/6 receptor (lipoprotein-related protein 5 or 6). Wnt binding to these receptors recruits the scaffold protein disheveled (DVL), which leads to LRP5/6 phosphorylation and recruitment of AXIN and GSK3β away from the β-catenin destruction complex. The stabilized β-catenin moves to the nucleus, coactivates the TCF family of transcription factors and regulates the expression of their downstream target genes.⁴ Wnts that signal noncanonically are independent of β-catenin stabilization. Instead, they signal through binding to FZD and co-receptors, including ROR1 and ROR2, to regulate distinct pathways.^5–7 These pathways include the planar cell polarity (PCP) and the Wnt/calcium pathways. The PCP pathway, mediated by FZD and DVL, activates c-Jun N-terminal or Rho-associated kinases.⁷ In the Wnt/calcium pathway, Wnt binding increases intracellular calcium levels, activating calcium-sensitive proteins, such as calcium/calmodulin-dependent protein kinase 2 (CaMKII) and protein kinase C (PKC).⁸

Although these canonical and noncanonical Wnt pathways are distinct, some Wnts can signal through both pathways, with opposing outcomes depending on the receptor context.⁹ The best-studied example is Wnt5a, a ligand family member traditionally considered to activate noncanonical signaling. However, Wnt5a can also signal through the co-receptors FZD4 and LRP5 to activate the canonical β-catenin-dependent pathway.^10,11 Wnt5a can also heterodimerize with another noncanonical Wnt (Wnt11) to upregulate canonical Wnt signaling.¹² We recently reported that canonical Wnt/β-catenin signaling in prostate cancer (PCa) cells increases the expression of co-receptors to enhance noncanonical responses to Wnt5a.¹³ Together, these findings indicate that there may be substantial cross-talk between canonical and noncanonical Wnt pathways.

Multiple studies have implicated Wnt signaling in PCa development and found evidence of increased Wnt signaling in tumors that progress after androgen deprivation therapies (castration-resistant prostate cancer [CRPC]).^14,15 Genomic alterations that constitutively activate canonical Wnt signaling, including loss or truncating mutations in APC and stabilizing mutations in β-catenin, occur in a subset of primary PCa and at a somewhat higher frequency in metastatic PCa.^13,16–20 We previously characterized the transcriptional response to canonical Wnt signaling in PCa cells and confirmed that a Wnt activity signature was greatest in tumors with genomic alterations in the pathway.¹³ In tumors without these genomic alterations, higher Wnt signaling was associated with increased expression of Wntless (WLS), which binds Wnts in the endoplasmic reticulum (ER) after they undergo palmitoylation by the enzyme porcupine (PORCN) and is required for their transport to the plasma membrane.^21–23 These results support a role for increased autocrine Wnt production in PCa development and progression to CRPC.

WLS is also overexpressed in cancers of the brain, breast, colorectum, stomach, and liver,^24–27 and its increased level has been associated with worsening clinical stage, metastasis, and poor survival. Notably, WLS-mediated Wnt secretion may also contribute to Wnt/β-catenin pathway activity in cells with genomic alterations that increase β-catenin stability.²⁸ Wnt synthesis and secretion are gaining increasing importance as potential drug targets in cancer, and a better understanding of Wnt biosynthesis in cancers may identify new therapeutic targets.^1,29,30

To further study the mechanisms contributing to autocrine Wnt synthesis and signaling in PCa, we initially identified PCa cell lines with high levels of canonical Wnt signaling driven by autocrine Wnt synthesis. Notably, we found that canonical Wnt signaling in these cells was driven by both the higher expression of a canonical Wnt (Wnt3a) and the noncanonical Wnt5a, with subsequent signaling through ROR2 and downstream activation of CaMKII. Further studies showed that CaMKII inhibition resulted in a dramatic loss of WLS through increased proteasome degradation, which was associated with an ER stress response and apoptosis in cells expressing high levels of WLS. These and additional findings reveal that CaMKII contributes to the handling of ER proteins, including WLS, and that cancer cells carrying high levels of WLS may be vulnerable to CaMKII inhibition.

RESULTS

WLS-dependent autocrine Wnt signaling in PCa

To characterize autocrine Wnt/β-catenin activities in PCa, we initially screened a series of commonly studied PCa cell lines with a lentiviral transduced Wnt/β-catenin activity reporter, expressing firefly luciferase (7TFP) or GFP (7TGC) driven by multimerized TCF promoters.³¹ Among the eight lines screened, CWR22R1 (R1) cells demonstrated the highest endogenous Wnt activity, followed by PC3 cells (Figure 1A). Notably, CWR22Rv1 (Rv1) cells had markedly lower activity. Rv1 and R1 cells were derived from the same patient-derived xenografts (CWR22), except that the R1 line was adapted to in vitro culture from a castration-resistant CWR22 xenograft.^32,33 The activity difference was not limited to the lentiviral vectors, as transient transfection with a TOP-FLASH Wnt/β-catenin reporter also showed greater activity in the R1 cells (Figure S1A). Our R1 cells showed a comparable Wnt/β-catenin activity to cells obtained from a commercial source (CWR22R1-ca cells) (Figure S1A).³⁴ The activity in the R1 cells could be markedly decreased by treatment with the PORCN inhibitors ETC-159 or LGK974, indicating it was being driven by endogenous Wnt ligands (Figure 1B). Conversely, activity in the Rv1 cells (as well as the R1 cells) could be stimulated by exogenous Wnt3a (Figure S1B). Moreover, activity in the Rv1 cells was also stimulated by transfection with vectors encoding active Wnt3a or Wnt1 (Figure S1C), indicating that these cells could process and respond to endogenous and exogenously introduced Wnt. Together, these data show that the low basal activity in Rv1 was due to low endogenous Wnt production.

Figure 1. WLS-mediated endogenous Wnt/β-catenin is driven by both canonical and noncanonical Wnts.

(A) PCa cell lines with stable expression of the Wnt/β-catenin activity reporter (7×Tcf-FFluc [7TFP]) were measured for endogenous Wnt activity per 10 μg of protein (luciferase activity, light unit [Lu]) (n = 3; **p < 0.01).

(B) R1–7TFP cells were treated with porcupine inhibitor (LGK 974 or ETC-159) for 24 h and assessed for luciferase activity (n = 3; *p < 0.05 and **p < 0.01).

(C) R1–7TFP cells were transfected with three distinct WLS siRNA, a pool of all three, or a non-target control siRNA, and the luciferase activities and WLS protein were measured at 72 h post-transfection (n = 3; **p < 0.01).

(D) Relative mRNA expression of Wnt ligands in PCa cells measured by TaqMan RT-PCR.

(E–G) R1–7TFP cells were transfected with siRNAs targeting Wnt3A, Wnt5A, ROR1, or ROR2 and examined at 72 h post-transfection (n = 3; **p < 0.01 and ns, not significant). Vinculin was included in the immunoblot for protein loading control. Single values are indicated as circles.

Data represent mean ± SEM of the indicated number of biological repeats (n). A representative result was presented from at least two independent experiments. The p values were calculated by Student’s t test (A–C and E–G).

R1 cells expressing the 7TGC (GFP) reporter were further cloned by limiting dilution. Most of the single-cell-derived clones were found to have sustained GFP expression, further supporting autocrine Wnt signaling among these PCa cells. GFP expression in one clone, D11, is shown in Figure S1D. As expected, the 7TGC and 7TFP reporter activities in R1 cells could be suppressed by expression of a dominant-negative TCF4, lacking the β-catenin binding domain (Figures S1E and S1F).³¹ Consistent with the effects of the PORCN inhibitors, the 7TFP activity in R1 cells was suppressed by small interfering RNA (siRNA) targeting PORCN (Figure S1G). Moreover, it was also suppressed by siRNA targeting WLS, indicating that proper transport of Wnt, mediated by WLS from the ER and Golgi to the cell surface, was required for the activity (Figure 1C).

We next used quantitative reverse-transcription PCR (RT-qPCR) to profile Wnt ligands in Rv1, R1, and D11 cells and found that Wnt3a expression was markedly higher in the R1 and D11 cells compared to Rv1 cells (Figure 1D). Although the expression of Wnt5a, 7b, 10B, and 11 was similarly higher in R1 and D11 cells, these Wnts generally drive noncanonical Wnt signaling, suggesting that Wnt3a was the major driver of the canonical Wnt/β-catenin activity in the cells. To test this hypothesis, we used siRNA to deplete Wnt3a in R1 cells. This resulted in a strong suppression of Wnt/β-catenin activity, confirming that endogenous Wnt3a production and secretion is the primary driver of canonical Wnt signaling in these cells (Figure 1E).

Wnt5a/ROR2 signaling supports Wnt/β-catenin activity

As noted above, several Wnts associated with noncanonical Wnt signaling were also increased in the R1 and D11 cells relative to the Rv1 cells. As Wnt5a is the predominant noncanonical Wnt in PCa,^35,36 we next assessed the effects of Wnt5a siRNA in R1 cells. Surprisingly, although we obtained only partial knockdown, the decrease in Wnt5a significantly suppressed the activity of 7TFP reporter (Figure 1F). The siWnt5a-mediated reductions of canonical Wnt reporter activities in R1 or PC3 cells could be partially restored by adding exogenous Wnt5a (Figures S2A and S2B), supporting the specific role of Wnt5a in sustaining the canonical Wnt activities. We also examined siRNA targeting Wnt11, which decreased the reporter gene activity (Figure S2C). Noncanonical signaling through Wnt5a is mediated primarily by the co-receptors ROR1 or ROR2. Therefore, we used siRNA to determine whether signaling through ROR1 or ROR2 was contributing to basal Wnt/β-catenin activity. Notably, the depletion of ROR2, but not ROR1, significantly decreased 7TFP reporter activity (Figure 1G). Moreover, ROR2 depletion was more suppressive than the individual Wnt5a and Wnt11 siRNAs or the combination, and Wnt11 siRNA did not further enhance the effect of the ROR2 silencing (Figure S2D). Together, these data indicated that canonical Wnt/β-catenin activity was enhanced by noncanonical Wnt signaling, primarily mediated through ROR2.

CaMKII inhibition suppresses Wnt/β-catenin activity

Noncanonical Wnt pathways include Wnt/Ca⁺⁺ signaling, with Wnt5a-stimulated intracellular Ca⁺⁺ release leading to calmodulin-mediated activation of CaMKII.⁸ We have also reported that Wnt5a can stimulate CaMKII activation in PCa cells.¹³ Therefore, we next assessed the effects of CaMKII inhibition on Wnt/β-catenin signaling in R1 cells. CaMKII inhibitor KN93 acts by preventing calmodulin binding to CaMKII.³⁷ Treatment with KN93, but not an inactive derivative of KN93 (KN92), suppressed Wnt/β-catenin reporter activity in the R1 cells (Figure 2A). To determine whether CaMKII inhibition suppressed Wnt/β-catenin signaling by decreasing Wnt production versus by impairing the response to Wnt, we treated the cells with exogenous Wnt3a simultaneously with CaMKII inhibition. Notably, CaMKII inhibition did not prevent the response to exogenous Wnt3a, indicating that it impaired Wnt synthesis (Figure 2B).

Figure 2. CaMKII supports WLS expression and Wnt/β-catenin activity.

(A and B) R1–7TFP cells were treated with a CaMKII inhibitor (KN93) or its inactive analog (KN92) for 48 h, and luciferase activities were measured. LU (light unit; n = 3; ns, not significant and **p < 0.01) (B) R1–7TFP cells were treated with DMSO or KN93 (10 or 15 μM) for 48 h, and Wnt3A was added during the last 8 h. The relative fold changes of luciferase activities are plotted in Wnt3a treatment over PBS treatment (n = 3; **p < 0.01).

(C) R1–7TFP cells were treated as in (A) and blotted for p-CaMKII (T286) and WLS. β-Actin was blotted for loading control.

(D) R1–7TFP cells were transfected with non-targeted siRNA (NTC) or pooled siRNA against the indicated genes (ROR1, ROR2, or Wnt5A). Samples were collected at 72 h post-transfection and immunoblotted.

(E) HCT116 or SW480 colon cancer cells were treated at the indicated dose for 48 h. Cell lysates were immunoblotted.

(F) HCT116 cells were transfected with pooled siRNA against NTC or the CaMKII isotypes or with an individual siRNA against WLS (siWLS #5). Samples were collected at 72 h post-transfection and assessed for CaMKII activity (p-CaMKII) and WLS.

(G) HCT116 cells were pre-treated with DMSO or AS100397 (2.5 μM) for 24 h. The cells were then treated with added DMSO or KN93 at the indicated doses for an additional 48 h. The lysates were assessed for CaMKII activity (p-CaMKII) and WLS. Single values are indicated as circles.

Data represent mean ± SEM of the indicated number of biological repeats (n). A representative result was presented from at least two independent experiments. The p values were calculated by Student’s t test (A and B).

Consistent with this conclusion, CaMKII inhibition blocked the secretion of Wnt3a from L-Wnt3a cells, a common source for bioactive Wnt3a (Figure S3). As expected, the PORCN inhibitor ETC-159 also markedly decreased Wnt3a secretion. CaMKII inhibition had less effect on the secretion of an unrelated protein Munc 18-1. We then assessed the effects of the CaMKII inhibitor on WLS protein levels. At concentrations that suppress Wnt/β-catenin signaling, KN93, but not KN92, caused a marked decrease in WLS (Figure 2C). This was also associated with a decrease in CaMKII phosphorylation at T286, a marker of CaMKII activity that is mediated by autophosphorylation in response to calmodulin binding and that drives kinase activation.³⁷ Finally, the siRNA-mediated downregulation of ROR2 or Wnt5a also reduced WLS, consistent with Wnt5a/ROR2 signaling through CaMKII to modulate WLS (Figure 2D).

CaMKII inhibition suppresses Wnt/β-catenin activity in CRC cells

We next examined colorectal carcinoma (CRC) cells with constitutive activation of the Wnt/β-catenin pathway due to APC loss (SW480) or a stabilizing mutation in β-catenin (HCT116). Notably, although β-catenin has increased stability in these cells, they have been shown to still have dependence on WLS-mediated Wnt production for optimal Wnt/β-catenin signaling and growth.²⁸ In HCT116 cells, PORCN inhibition with ETC-159 decreased WLS (which is stabilized by palmitoylated-Wnt binding) and LRP6 phosphorylation (a marker of Wnt-mediated cell membrane signalosome activity), consistent with these cells producing endogenous Wnt to stabilize WLS and having an intact FZD/LRP complex (Figure 2E, left). Notably, ETC-159 also decreased AXIN2 (a marker of Wnt/β-catenin activity), indicating that endogenous Wnt is contributing to Wnt/β-catenin signaling despite the stabilizing mutation in β-catenin.²⁸ CaMKII inhibition with KN93 similarly decreased WLS and phosphorylation of LRP6 and markedly decreased AXIN2, in parallel to the decrease in CaMKII phosphorylation. In SW480 cells, ETC-159 also decreased WLS, AXIN2, and phosphorylation of LRP6 (Figure 2E, right). However, the effects on AXIN2 were less marked than those in the HCT116 cells. KN93 generated similar responses as ETC-159. Notably, the decrease in AXIN2 with KN93 was greater than with ETC-159, possibly reflecting greater suppression of the WLS-mediated Wnt production.

Pooled siRNA against each CaMKII isoform was also tested in HCT116 cells to measure the contribution of each to WLS protein levels. The siRNA silencing of the CaMKIIβ and CaMKIIδ isoforms had the most effect on overall CaMKII activity (as assessed by T286 phosphorylation) and WLS protein levels (Figure 2F). To further confirm that the KN93 effects were specifically through CaMKII inhibition, we treated cells with a selective ATP-competitive CaMKII inhibitor AS100397.³⁸ AS100397 suppressed CaMKII activity (based on decreased phosphorylation at T286) and decreased WLS to a similar extent as KN93 (Figure 2G). The addition of KN93 to AS100397 did not further decrease WLS, consistent with KN93 effects on WLS being mediated through CaMKII. Together these findings indicate that CaMKII inhibition suppresses Wnt/β-catenin signaling by decreasing WLS and subsequently decreasing Wnt production.

CaMKII inhibition increases WLS protein degradation

We generated a tetracycline-inducible WLS line in LNCaP PCa cells that have low levels of endogenous WLS³⁹ (Figure S4A). Treatment of the uninduced and doxycycline-induced cells with KN93 caused a progressive decrease in WLS (Figure 3A). In the uninduced cells, KN93 at 10 μM caused a decrease in endogenous WLS at 6 h, with further loss at 24 h. Notably, the effects on the exogenous WLS were comparable, with a decrease at 6 h and further decreases at 24 or 48 h. The exogenous WLS protein was similarly greatly decreased by another CaMKII inhibitor (XII)⁴⁰ (Figure S4B). CaMKII inhibition with KN93 caused only a small decrease in endogenous WLS mRNA, which was not seen with a second CaMKII inhibitor (XII) (Figure 3B). In combination with the above effects on exogenously expressed WLS, this indicated that CaMKII inhibition decreased WLS by a posttranscriptional mechanism.

Figure 3. CaMKII inhibition decreases the exogenously expressed WLS, which is independent of lysosome.

(A) LNCaP cells with tet-inducible WLS expression were treated with vehicle (uninduced) or doxycycline (250 ng/mL, induced) for 48 h. KN93 (10 μM) was added during last 6 or 24 h or the entire 48 h, as indicated.

(B) R1–7TFP cells were treated with KN93 (10 μM) or XII (20 μM) for 24 h. The mRNA levels of WLS were then measured by real-time RT-PCR with two independent WLS probes (#1 and #2) (n = 3; *p < 0.05).

(C) LNCaP cells with tet-inducible WLS were treated with doxycycline (250 ng/mL) for 48 h in combination with bafilomycin at indicated doses. KN93 (10 μM) was added during the last 8 h.

(D) LNCaP with tet-inducible WLS was similarly treated with vacuolin 1 at the indicated doses as in (C).

Single values are indicated as circles. Data represent mean ± SEM of the indicated number of biological repeats (n). A representative result was presented from at least two independent experiments. The p values were calculated by Student’s t test (B).

WLS at the cell surface can be recycled back to the ER or trafficked to lysosomes for degradation.^41,42 To determine whether CaMKII inhibition increased lysosomal degradation, we treated LNCaP cells expressing the exogenous doxycycline-induced WLS with bafilomycin to block lysosomal proteolysis. Bafilomycin treatment led to an on-target effect of blocking lysosomal degradation as reflected by the marked accumulation of autophagosomal LC3B protein⁴³ (Figure 3C). Notably, bafilomycin also increased WLS expression in the absence of CaMKII inhibition, consistent with basal lysosomal degradation. However, bafilomycin did not prevent the marked decrease in WLS in response to KN93, indicating that CaMKII inhibition does not increase lysosomal degradation of WLS. We similarly found that another lysosome inhibitor (vacuolin 1) did not prevent the decrease in WLS in response to KN93 (Figure 3D).

In contrast, KN93-induced WLS degradation was blocked by the treatment with proteasome inhibitors (MG115/132), similarly to the block of p53 degradation (Figure 4A). Previous studies have found that WLS can undergo proteasome-mediated degradation through the ER-associated degradation (ERAD) pathway, and that this ERAD-mediated degradation targets the unliganded WLS.⁴⁴ Consistent with this mechanism, treatment with ETC-159 to block basal Wnt synthesis in mock transfected cells caused a decrease in WLS (Figure 4B, pcDNA). Moreover, the overexpression of transfected Wnt2a, 3a, or 5a markedly increased WLS, and this increase was abrogated by ETC-159. Notably, CaMKII inhibition with KN93 decreased WLS to a greater extent than ETC-159 in the mock transfected cells, but Wnt overexpression greatly overcame the WLS reduction caused by CaMKII inhibition. These results indicate that CaMKII inhibition drives the ERAD-mediated degradation of unliganded WLS protein within ER.

Figure 4. CaMKII inhibition causes protea- some-mediated WLS protein degradation and reduction in calnexin.

(A) LNCaP with tet-induced WLS were treated with doxycycline (250 ng/mL) together with DMSO or KN93 for 24 h in the absence or presence of MG132 and MG115 (MG, 5 μM each). p53 was blotted as a positive control.

(B) 293T cells were transfected with various V5 epitope-tagged Wnt expression vectors. At 48 h post-transfection, the cells were treated with DMSO, ETC-159 (1 μM) or KN93 (10 μM) for an additional 8 h.

(C) HCT116 cells were treated with increasing doses of KN93 for 48 h and blotted.

CaMKII inhibition leads to decreased calnexin

As noted above, WLS has been shown to undergo ERAD when not binding to endogenous Wnt.⁴⁴ Being an ER transmembrane glycoprotein, newly synthesized WLS is likely being stabilized by the ER resident chaperones calnexin and/or calreticulin for their proper folding and stability.⁴⁵ Indeed, calreticulin has recently been reported to be in the Wnt/WLS complex,⁴⁶ and WLS has been shown to co-localize with calnexin during the antegrade transport of Wnt.⁴⁷ Notably, while little change could be detected in calreticulin in response to KN93 treatment, there was a marked decrease in calnexin (Figure 4C). Consistent with this loss of calnexin, another calnexin chaperone client glycoprotein, MHC class I,⁴⁸ was also reduced by the KN93 treatment (Figure 4C).

CaMKII inhibition causes WLS-dependent and PERK-directed ER stress response

The ER unfolded protein response (UPR) is one of the mechanisms for sustaining homeostasis of cells.⁴⁹ We hypothesized that CaMKII inhibitor treatment would be associated with the accumulation of misfolded WLS in the ER and initiation of the UPR. Indeed, KN93 treatment caused phosphorylation of eIF2, a marker of PERK-mediated ER stress,⁵⁰ in parallel to the loss of Axin2 and WLS (Figure 5A). The UPR is also mediated by two other sensors, IRE1 and ATF6.⁵¹ We next compared the response of KN93 with that of thapsigargin (TG), which is a potent and well accepted UPR inducer.⁵² As expected, TG induced robust UPR responses as reflected by the marked increase in BiP (Figure 5B). This was associated with the activation of all three arms of the UPR, with IRE1 activation being reflected by the increased levels of XBP1s, and ATF6 response reflected by the cleavage of ATF6 and the accumulation of the smaller fragment. The PERK response was reflected by increased eIF2a phosphorylation (without changes in total eIF2α levels) and by downstream increased levels of ATF4 and CHOP.

Figure 5. CaMKII inhibition causes WLS-dependent and PERK-driven ER stress responses.

(A) HCT116 cells were treated with the indicated doses of KN93 for 48 h, followed by immunoblotting.

(B) HCT116 cells were treated with KN93 (48 h) or thapsigargin (TG, 8 h) at the indicated doses, followed by immunoblotting.

(C) HCT116 cells were treated with DMSO or GSK2606414 (1 μM) for 2 h, followed by the addition of DMSO or KN93 treatment at the indicated dose for an additional 48 h.

(D) HCT116 cells were transfected with non-target control (NTC) or WLS siRNA (siWLS) for 72 h, and cells were treated with KN93 during the last 24 h.

(E) HCT116 cells were similarly transfected and treated with TG as in (G).

While KN93 treatment caused an increase in eIF2α phosphorylation that was comparable to that from TG, there was less of an increase in ATF4 or CHOP (Figure 5B). Moreover, there were no clear increases in XBP1s or ATF6 cleavage, raising the possibility that the eIF2α phosphorylation may not reflect activation of PERK in UPR. Indeed, in addition to PERK, eIF2α phosphorylation can be mediated by PKR, GCN2, or HRI as part of the integrated stress response (ISR).⁵³ Therefore, we assessed the effects of a PERK inhibitor (GSK2606414) on the response to KN93. Notably, GSK2606414 caused an increase in basal p-eIF2α, consistent with a previous study showing that it can directly activate GCN2⁵⁴ (Figure 5C). However, KN93 did not further increase p-eIF2α in the GSK2606414-treated cells, and the increases in ATF4, CHOP, and BiP were also mitigated. To further assess PERK activation, we determined whether its phosphorylation could be detected based on decreased mobility on immunoblotting.⁵⁵ Both KN93 and TG treatments led to an upper shift of PERK at doses that increased BiP (Figure S5). Together, these findings support the conclusion that CaMKII inhibition activates the PERK arm of the UPR and that this mechanism is distinct from the dissociation of BiP that can mediate the activation of IRE1 and ATF6 in addition to PERK.

Notably, the decreases in AXIN2 and WLS were observed at lower KN93 concentrations than those that caused an increase in eIF2α phosphorylation (Figure 5A), suggesting that misfolding and degradation of WLS may be a driver of the UPR. To test this hypothesis, we used siRNA to decrease WLS and examined its effects on the response to KN93. Significantly, the KN93-induced eIF2α phosphorylation was markedly reduced after WLS silencing (Figure 5D). The loss of calnexin was similarly prevented. Interestingly, KN93 induced a decrease in XBP1s (by a longer exposure of the film than Figure 5A to detect the basal levels of XBP1s), which was blunted by the siWLS. In contrast, UPR induction by TG was not affected by the silencing of WLS, as it caused similar increases in eIF2α phosphorylation and XBP1s in the control and siWLS-treated cells (Figure 5E).

Finally, we asked whether the KN93-mediated UPR might be driven by the loss of calnexin. To address this, we overexpressed GFP-tagged calnexin in 293T cells and assessed the responses to KN93. Notably, KN93 treatment depleted endogenous calnexin (-En) and decreased exogenous calnexin (-GFP), but substantial levels of exogenous calnexin persisted at the highest KN93 concentrations (Figure S6). This was associated with a blunting of the increase in eIF2α phosphorylation, suggesting that the loss of calnexin may contribute to UPR. Together, our results reveal distinct patterns of ER stress induced by CaMKII inhibition compared with canonical ER stress inducers and indicate that WLS contributes to ER stress in response to CaMKII inhibition.

CaMKII inhibition leads to enhanced apoptosis in cells with high WLS expression

We hypothesized that the WLS-dependent ER stress in response to CaMKII inhibition may lead to apoptosis in cells expressing high levels of WLS. Indeed, KN93 caused apoptosis in D11 cells (Figure 6A), and this was mitigated by WLS silencing (Figure 6B). Previous studies have found that WLS expression increases with the progression to CRPC and with resistance to enzalutamide, an androgen receptor inhibitor.^39,40,56 Consistent with these previous reports, we found a progressive increase in WLS protein levels, in parallel to the recovered expression of PSA (androgen receptor regulated gene) during the progression to enzalutamide resistance in VCaP prostate cancer cells⁵⁷ (Figure 6C). Notably, this increase in WLS was associated with elevated apoptosis and decrease in viable cells in response to KN93 (Figures 6D and 6E). As PERK activation appeared to be the key mediator of the KN93-induced UPR, we also examined if PERK suppression would affect KN93-induced apoptosis. Pretreatment with the PERK inhibitor (GSK2606414) was able to reduce KN93-mediated apoptosis in enzalutamide-resistant VCaP prostate cancer cells (Figure S7).

Figure 6. CaMKII inhibition induces WLS-dependent apoptosis.

(A) D11 cells were treated with KN93 (20 μM) for the indicated durations.

(B) D11 cells were transfected with control or WLS siRNA for 72 h and treated with KN93 for an additional 24 h.

(C) VCaP cells were cultured for long-term in high doses of enzalutamide (16 μM). Samples were collected at the indicated time points and blotted. β-Actin was blotted for protein loading control.

(D) Enzalutamide-resistant VCaP cells (EnzR) and their control cells (DMSO) were treated with KN93 for 24 h (FL, full length and C-, cleaved).

(E) EnzR or DMSO VCaP cells were treated with KN93 at the indicated doses for 24 h. The live and dead cells were counted based on trypan blue exclusion (n = 3; ***p < 0.001 and ns, not significant).

(F) HCT116 cells were treated with ruxolitinib at the indicated doses for 24 h.

Single values are indicated as circles. Data represent mean ± SEM of the indicated number of biological repeats (n). A representative result was presented from at least two independent experiments. The p values were calculated by Student’s t test (E).

A recent report showed that an FDA-approved JAK1/2 inhibitor (ruxolitinib) can strongly suppress CaMKII activity and reverse the effects on myocardial arrhythmia and ischemia.³⁸ Therefore, we tested ruxolitinib for its CaMKII suppression and associated WLS degradation. Ruxolitinib reduced CaMKII activity in HCT116 cells, which was associated with a reduction in WLS protein levels (Figure 6E). Moreover, there was a corresponding increase in apoptosis, as indicated by the increased cleaved caspase-3. These findings suggest that ruxolitinib might be repurposed for the treatment of WLS-high tumors.

DISCUSSION

We initially found that canonical Wnt signaling was driven not only by Wnt3a but also by noncanonical Wnts, including Wnt5a, signaling through ROR2. By examining the pathways activated by noncanonical Wnt signaling, we found that CaMKII inhibition markedly suppressed canonical Wnt signaling. This suppression could be bypassed by exogenous Wnt3a, indicating that CaMKII inhibition was interfering with Wnt production. Indeed, we found that CaMKII inhibition markedly increased WLS protein degradation. Wnt-bound WLS traffics to the plasma membrane and can then be recycled to the ER or targeted for lysosomal degradation.^41,42 In the absence of Wnt binding, WLS has been shown to undergo proteasome degradation through ERAD.⁴⁴ The effects of CaMKII inhibition were not prevented by inhibitors of lysosomal proteases but were mitigated by proteasome inhibitors, indicating that the KN93-induced WLS degradation was ERAD mediated. WLS in the ER is stabilized by binding to newly synthesized Wnts after they have undergone palmitoylation by PORCN.⁴⁴ Notably, we found that Wnt overexpression could substantially prevent the loss of WLS protein in response to CaMKII inhibition, indicating that CaMKII plays a role in regulating the turnover of unliganded WLS in the ER.

A recent study showed that calreticulin forms a complex with PORCN and WLS to facilitate the transfer of Wnt to WLS.⁴⁶ Calreticulin resides in the ER lumen and, in conjunction with calnexin, a type I integral ER membrane protein, acts as a chaperone for N-linked glycoproteins.⁵⁸ Notably, while we did not observe an alteration in calreticulin in response to CaMKII inhibition, we found a marked decrease in calnexin. Moreover, this decrease was WLS-dependent, as it was not seen in cells after siRNA-mediated WLS depletion, indicating that it may be a downstream consequence of CaMKII inhibitor-mediated WLS dysregulation. Interestingly, calnexin may play a direct role in targeting misfolded ER proteins to lysosomes for degradation (ER-phagy).⁵⁹ However, the possible role of this pathway in driving the CaMKII inhibitor-mediated loss of calnexin and WLS is unclear as we found that WLS degradation was lysosome independent. Another recent study suggested a Wnt-independent role for WLS as an ER chaperone working in conjunction with calreticulin and calnexin, as WLS knockout in dendritic cells caused an ER stress response and decreases in calreticulin and calnexin.⁶⁰ Further studies are needed to determine the precise basis for loss of calnexin and its relationship to the loss of WLS.

CaMKII inhibition also resulted in increased eIF2α phosphorylation. The marked effects of CaMKII inhibition on ER proteins, including WLS, calnexin, and MHC class I, suggested that this reflected ER stress and activation of PERK. Consistent with this conclusion, treatment with a PERK inhibitor mitigated the increase in eIF2α phosphorylation. Nonetheless, we cannot rule out a role for other eIF2α kinases (PKR, GCN2, or HRI) mediating the ISR. Notably, while ER stress and activation of the UPR generally lead to the activation of PERK, IRE1, and ATF6, we did not observe activation of IRE1 or ATF6 in response to CaMKII inhibition. The basis for this apparent selective activation of PERK is not clear, but it is presumably not dependent on the redistribution of BiP that would activate all three arms of the UPR. One possibility is that CaMKII inhibition-mediated PERK activation is driven by the depletion of calnexin, as it could be partially relieved by the overexpression of calnexin, although it would still be anticipated that calnexin depletion would cause redistribution of BiP. Alternatively, we may have failed to detect IRE1 or ATF6 activation due to distinct kinetics or feedback, with one previous study showing that PERK activation could feed back to suppress IRE1 endoribonuclease activity and enhance apoptotic signaling.⁶¹ Consistent with this hypothesis, CaMKII inhibition caused a decrease in the levels of XBP1s (see Figure 5D).

Notably, the CaMKII inhibitor-mediated increase in eIF2α phosphorylation was prevented by WLS knockdown. Similarly, the apoptotic response to CaMKII inhibition was also mitigated by WLS knockdown. Interestingly, the dose response for the CaMKII inhibitor-mediated eIF2α phosphorylation was shifted to the right relative to the reduction in WLS. Together, these observations suggest that while CaMKII inhibition may have global effects on ER protein folding, the dysregulation of WLS in response to CaMKII inhibition presents a particular challenge to ER protein homeostasis and may be a driver of the stress response and apoptosis. The stress may be even more pronounced in cells with low levels of Wnt where the ER becomes dependent on CaMKII to handle the constitutively made unliganded WLS.

CaMKII is a cytosolic serine/threonine kinase that has four isoforms (α, β, γ, and δ), encoded by distinct genes.^62,63 Structurally, it includes an N-terminal catalytic domain, a regulatory domain containing an auto-inhibitory and calmodulin-binding region, and a C-terminal domain mediating multimerization. CaMKII is activated in response to increased calcium bound calmodulin, which displaces the autoinhibitory domain, with subsequent autophosphorylation of T286 leading to sustained activity.⁶⁴ The α and β isoforms are expressed primarily in the brain and play important roles in learning and memory, while the γ and δ isoforms are broadly expressed.⁶⁵ CaMKII has been found to phosphorylate a large number of proteins and has been implicated in diverse cellular processes, including synaptic plasticity, learning and memory, vascular smooth muscle polarization, cardiac arrhythmia, myocardial survival, cell migration, and proliferation.^66,67 However, while it can be a downstream effector of ER stress and subsequent Ca⁺⁺ flux, it has not been previously implicated as an upstream regulator of ER homeostasis.

While our results indicate a protective role of CaMKII in WLS-high cells against ER stress or apoptosis, increased CaMKII activity has conversely been associated with ER stress and apoptosis in other settings, including myocardial infarction and neurodegenerative diseases.^66,68,69 In these cases, prolonged and elevated CaMKII activity can be detrimental to cell survival as it can cause a sustained increase in cytoplasmic calcium and activation of pro-apoptotic caspases. Interestingly, one study found that CaMKII inhibition could mitigate doxorubicin-mediated cardiac toxicity by reducing ER stress through increased IRE1α and XBP1s in doxorubicin-treated hearts.⁷⁰ There have been limited studies on CaMKII in PCa. Inhibition of CaMKII activity using KN93 significantly repressed cell proliferation and invasion of C4-2B PCa cells, which was at least in part mediated through decreased Notch1 signaling.⁷¹ In another study, KN93 induced dose-dependent apoptosis in LNCaP PCa cells, which was linked to decreased CaMKII-mediated AKT activation.⁷² A further study found that CaMKII activity was increased in metastatic PCa, and that CaMKII inhibition could suppress metastasis in PC3 cells expressing high levels of active CaMKII.⁷³ It remains to be determined if our newly defined role of CaMKII in ER homeostasis plays any role in the above effects in PCa cells.

Notably, the cytoplasmic tail of calnexin can undergo phosphorylation. A previous study showed that the C-terminal cytoplasmic domain of calnexin can be phosphorylated by CK2 and ERK1, which enhances calnexin binding to ribosomes and may thus enhance co-translational glycoprotein folding.⁷⁴ Moreover, a subsequent study found that phosphorylation of this site by ERK1 (Ser563) was increased in response to protein misfolding, and that blocking its phosphorylation could enhance protein secretion.⁷⁵ Phosphorylation of this site may also modulate calnexin interaction with SERCA2b, with subsequent effects on Ca⁺⁺ flux.⁷⁶ Calnexin phosphorylation at additional cytoplasmic sites has also been reported, but the functional significance is unclear. Moreover, no data address whether the Ser563 site or other sites on calnexin are phosphorylated by CaMKII, and no data indicate that preventing this phosphorylation would decrease calnexin or increase the degradation of ER proteins. Therefore, specific substrates through which CaMKII may directly or indirectly regulate ER functions remain to be identified.

We also examined the effects of CaMKII inhibition in CRC cells with genomic alterations in APC or β-catenin. Notably, while these alterations increase β-catenin stability independently of Wnt, previous studies have shown that these cells still have some dependence on autocrine or paracrine Wnt.²⁸ Indeed, in both HCT116 and SW480 cells, we found that treatment with a PORCN inhibitor decreased canonical Wnt signaling as assessed by AXIN2. Interestingly, CaMKII inhibition and subsequent loss of WLS resulted in an even greater suppression of AXIN2. It is not clear whether this reflects a more robust disruption of Wnt synthesis or other mechanisms, but it suggests that CaMKII inhibitors may have efficacy in these tumors through both the suppression of Wnt signaling and induction of ER stress.

There has been substantial interest in targeting CaMKII to prevent cardiac diseases and arrhythmias.⁶⁷ However, clinical trials have been limited due to challenges in the development of specific antagonists for clinical use and concerns related to drug toxicities. Notably, a recent study identified ruxolitinib as a potent CaMKII inhibitor.³⁸ Ruxolitinib is an orally bioavailable JAK2 inhibitor and is an FDA-approved medication for myelopro-liferative diseases. We confirmed that ruxolitinib decreased CaMKII activity and WLS, suggesting that this agent may have efficacy in some Wnt-driven tumors, especially those expressing high levels of WLS that may be more vulnerable to CaMKII inhibition and its induced apoptosis. Recently, direct WLS inhibitors have been identified by virtual screening, on-demand synthesis, and testing in cell-based Wnt reporter and other functional assays.⁷⁷ A prototype ETC-451 blocked WLS-Wnt3A interaction and decreased Wnt-addicted pancreatic cancer cell line proliferation. It is conceivable that more potent and specific WLS inhibitors will be available soon and can be tested for efficacy in combination with CaMKII inhibitors.

In summary, our studies have identified a pathway through which noncanonical Wnt signaling can form a feedback loop to enhance Wnt synthesis and canonical Wnt signaling. Moreover, we have revealed that CaMKII plays a major role in supporting Wnt production by regulating the ER handling of WLS, although the relevant and direct CaMKII substrates for this activity remain to be determined. The data further show that the CaMKII inhibitor-mediated dysregulation of WLS drives an ER stress response and apoptosis in cells expressing higher levels of WLS. Notably, this ER stress response appears to be mediated by PERK, but some contributions from other eIF2α kinases cannot be excluded. Together, these findings indicate that CaMKII inhibitors may have enhanced therapeutic efficacy in Wnt-driven tumors, particularly in those with high levels of WLS.

Limitations of the study

There are several limitations to our study. One is that the specific CaMKII substrates that participate in the handling of WLS and other ER proteins remain to be identified. A second limitation is that the precise basis for the relatively selective activation of PERK in response to CaMKII inhibition, and whether other pathways outside the ER may also contribute to eIF2a phosphorylation in response to CaMKII inhibition, is not clear. A third limitation is that the WLS-linked pathway remains to be determined for the loss of calnexin in response to CaMKII inhibition, whether it is through ERAD or through a distinct pathway. Finally, the extent to which apoptosis in response to CaMKII inhibition is driven by ER stress versus other CaMKII-regulated pathways, and whether it can be exploited therapeutically, is unclear. Studies are underway to address these and related questions.

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Xin Yuan (xyuan@bidmc.harvard.edu).

Materials availability

Any newly generated materials associated with the paper, including stable expression cell lines and expression vectors generated in this study, will be made available on request, but we may require a payment and/or a completed materials transfer agreement if there is potential for commercial application.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Cell lines were authenticated by short tandem repeat analysis (ATCC), used within 20 passages from the source, and were routinely tested for Mycoplasma (R&D Systems).

METHOD DETAILS

Cell lines

LNCaP, PC3, VCaP, CWR22Rv1, HCT116, SW480, L-Wnt3A and 293T cells were from ATCC and were maintained under conditions recommended by the provider. LAPC4 cells were from the Charles Sawyers lab (Memorial Sloan Kettering) and maintained in RPMI1640 medium with 10% FBS. CWR22R1 (R1) is a PCa cell line derived from CWR22 xenografts and a gift from Yun Qiu (University of Maryland). R1 cells were maintained in DMEM with 10% FBS.

LNCaP cells with doxycycline-regulated WLS or Luc expression were generated using the Lenti-X Tet-One inducible expression system (Puro) (Takeda). The WLS cDNA sequence was PCR amplified from an hGPR177 VersaClone cDNA template (R&D system). Lentiviruses were generated by transfecting 293FT cells (ATCC) with the expression vectors, and the supernatants were used to transduce LNCaP cells. The infected cells were selected with puromycin for 72 h to generate stable cell lines. Lentiviruses and stable R1-7TFP or -7TGP lines were similarly generated using the 7TFP or 7TGP vectors (Addgene).

Live cell count

Cells were trypsinized with 0.25% trypsin-EDTA (Invitrogen; 5min, 37°C), neutralized in DMEM/10% FBS, and pelleted. The cell suspension was mixed with an equal volume of 0.4% trypan blue stain (Thermofisher). The fractions of live and dead cells were counted on a Countess 3 automated cell counter (Invitrogen).

DNA and siRNA transfection

DNA expression vectors were transfected into cells using the Lipofectamine 2000 transfection reagent (ThermoFisher #11668500). siRNA (60 μM) were transfected using the Lipofectamine RNAiMAX transfection reagent (ThermoFisher #13778150).

Immunoblotting

Proteins were extracted using SDS buffer (2% SDS, 50 mM Tris-HCl, pH 7.5, and 2 mM EDTA). The extracted samples were heated at 100°C for 20 min and cleared by microfuge centrifugation (21k x g) for 20 min. The samples were mixed with 4x reducing Laemmli buffer and subjected to gel electrophoresis. The proteins were transferred to a nitrocellulose membrane using a Trans-Blot Turbo Transfer System (Bio-Rad, #1704150). After blocking with TBST containing 5% milk, the primary antibodies were incubated overnight at 4⁰C. After washing, the filers were incubated with HRP-conjugated anti-mouse/rabbit secondary antibody for 45 min at room temperature. The filter was washed again and treated with SuperSigna Western Blot Enhancer for ECL (Thermofisher, #46640) and exposed to X-ray film.

QUANTIFICATION AND STATISTICAL ANALYSIS

GraphPad Prism 9 Software (GraphPad Software Inc., RRID: SCR_002798) was used for all statistical analyses. The significance of the difference between the two groups was determined as mean ± SEM by the two-tailed Student t test. Statistical significance was accepted at p < 0.05. The statistical details of experiments can be found in the corresponding figure legends.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2026.117256.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse mAb anti-β-Actin	Santa Cruz Biotechnology	Cat# sc-69879; RRID: AB_1119529
Rabbit polyclonal anti-Axin2 (clone DM1A)	Cell Signaling Technology	Cat# 2151,; RRID: AB_2062432
Rabbit mAb anti- Phospho-CaMKII (Thr286) (D21E4)	Cell Signaling Technology	Cat# 12716,; RRID: AB_2713889
Rabbit mAb anti-Cleaved Caspase-3 (Asp175) (5A1E)	Cell Signaling Technology	Cat# 9664,; RRID: AB_2070042
Rabbit mAb anti-Calnexin (C5C9)	Cell Signaling Technology	Cat# 2679,; RRID: AB_2228381
Rabbit mAb anti-Calreticulin (D3E6)	Cell Signaling Technology	Cat# 12238,; RRID: AB_2713889
Rabbit mAb anti-LC3B (D11)	Cell Signaling Technology	Cat# 3868,; RRID: AB_2137707
Rabbit mAb anti-Phospho-eIF2α (Ser51) (D9G8)	Cell Signaling Technology	Cat# 3398,; RRID: AB_2096481
Rabbit anti-eIF2α	Cell Signaling Technology	Cat# 9722,; RRID: AB_2230924
Mouse mAb anti-MHC Class I (EMR8-5)	Cell Signaling Technology	Cat# 88274, RRID: AB_3714874
Rabbit polyclonal anti Phospho-LRP6 (Ser1490)	Cell Signaling Technology	Cat# 2568,; RRID: AB_2139327
Mouse mAb anti-p53 (DO-1)	Santa Cruz Biotechnology	Cat# sc-126,; RRID: AB_628082
Rabbit mAb anti-ROR1 (D6T8C) #	Cell Signaling Technology	Cat# 16540,; RRID: AB_2798764
Rabbit mAb anti-ROR2 (D3B6F)	Cell Signaling Technology	Cat# 88639,; RRID: AB_2800126
Mouse mAb anti-vinculin (7F9)	Santa Cruz Biotechnology	Cat# sc-73614,; RRID: AB_1131294
Rabbit mAb anti-V5-Tag (D3H8Q)	Cell Signaling Technology	Cat# 13202,; RRID: AB_2687461
Mouse mAb anti-Wntless (WLS, GPR177) (YJ5)	Biolegend	Cat# 655902,; RRID: AB_2562228
Rabbit mAb anti-Wnt3a (C64F2)	Cell Signaling Technology	Cat# 2721,; RRID: AB_2215411
Rabbit mAb anti-Wnt5a/b (C27E8)	Cell Signaling Technology	Cat# 2530; RRID: AB_2215595
Rabbit mAb anti-BiP (C50B12)	Cell Signaling Technology	Cat# 3177,; RRID: AB_2119845
Rabbit mAb anti-XBP-1s (E9V3E)	Cell Signaling Technology	Cat# 40435,; RRID: AB_2891025
Rabbit mAb anti-ATF6 (D4Z8V)	Cell Signaling Technology	Cat# 65880,; RRID: AB_2799696
Rabbit anti-PARP Antibody	Cell Signaling Technology	Cat# 9542;RRID: AB_2160739
Rabbit mAb anti-PERK (C33E10)	Cell Signaling Technology	Cat# 3192; RRID: AB_2095847
Experimental models: Cell lines
R1-7TFP, Rv1-7TFP, R1-7TGP	This paper	N/A
LNCaP-tet-Luc, LNCaP-tet-WLS	This paper	N/A
293T	ATCC	CRL-3216
LNCaP	ATCC	CRL-1740
CWR22R1	A gift from Qiu lab32	N/A
22Rv1	ATCC	CRL-2505
CWR-R1Ca	Millipore	SCC118
PC3	ATCC	CRL-3471
Du145	ATCC	HTB-81
LAPC4	A gift from Sawyers lab78	N/A
HCT116	ATCC	CCL-247
SW480	ATCC	CCL-228
L Cells	ATCC	CRL-2648
L Wnt-5A	ATCC	CRL-2814
L-WNT-3A	ATCC	CRL-2647
Wnt3a siRNA #5, #6, #7 and #8 and pool	Horizon Discovery	J-008338-05, –6, –07,-08 and L-008338-00
Wnt5a siRNA #9, #11 and pool	Horizon Discovery	J-003939-09, –11 and L-003939-00
Wnt11 siRNA pool	Horizon Discovery	L-009474-00
WLS siRNA #5, #6, #7 and pool	Horizon Discovery	J-018728-05, –6, –7 and L-018728-00
ROR1 siRNA #9 and pool	Horizon Discovery	J-003171-09 and L-003171-00
ROR1 siRNA #755	Ambion	#4427038, S9755
ROR2 siRNA #10, #12 and pool	Horizon Discovery	J-003172-10, –12 and L-003172-00
Porcn siRNA #6 and #7	Horizon Discovery	J-009613-06 and –7
CaMK2A siRNA pool	Horizon Discovery	L-004942-00
CaMK2B siRNA pool	Horizon Discovery	L-004943-00
CaMK2G siRNA pool	Horizon Discovery	L-004536-00-0005
CaMK2D siRNA pool	Horizon Discovery	L-004042-00-0005
Chemicals, peptides and recombinant proteins
KN93	Selleck Chemicals	S7423; CAS: 1188890-41-6
KN92	Selleck Chemicals	S6507; CAS: 1135280-28-2
AS100397	A gift from Dr. ED Luczak38	N/A
GSK2606414	Selleck Chemicals	S6507; CAS: 1337531-36-8
Ruxolitinib	Selleck Chemicals	S1378; CAS: 941678-49-5
MG132	Millipore Sigma	M8699; CAS: 1211877-36-9
MG115	Millipore Sigma	474780; CAS: 133407-86-0
ETC-159	A*STAR (Singapore)	N/A
LGK974	MediChemExpress	HY-17545 CAS: 1243244-14-5
Bafilomycin A1	Selleck Chemicals	S1413; CAS: 88899-55-2
Vacuolin-1	Selleck Chemicals	S6912; CAS: 351986-85-1
CaMKII Inhibitor XII	Millipore Sigma	208923 UNSPSC Code: 12352200
Recombinant mouse Wnt-3a protein	R&D System	1324-WN-002/CF
Critical commercial assays
Dual-Luciferase Reporter Assay System	Promega	E1910
TaqMan™ Array Human WNT Pathway	ThermoFisher	4414100

Highlights.

• Noncanonical Wnt supports WLS-mediated Wnt/β-catenin activity through CaMKII

• CaMKII inhibition drives degradation of WLS and calnexin, UPR, ER stress, and apoptosis

• Effects of CaMKII inhibition are driven by high levels of unliganded WLS

Data and code availability

Any data not provided in the manuscript of supplement can be obtained from the lead contact. There was no unique code generated.