Differential Roles of AXIN1 and AXIN2 in Tankyrase Inhibitor-Induced Formation of Degradasomes and β-Catenin Degradation

Inhibition of the tankyrase enzymes (TNKS1 and TNKS2) has recently been shown to induce highly dynamic assemblies of β-catenin destruction complex components known as degradasomes, which promote degradation of β-catenin and reduced Wnt signaling activity in colorectal cancer cells. AXIN1 and AXIN2/Conductin, the rate-limiting factors for the stability and function of endogenous destruction complexes, are stabilized upon TNKS inhibition due to abrogated degradation of AXIN by the proteasome. Since the role of AXIN1 versus AXIN2 as scaffolding proteins in the Wnt signaling pathway still remains incompletely understood, we sought to elucidate their relative contribution in the formation of degradasomes, as these protein assemblies most likely represent the morphological and functional correlates of endogenous β-catenin destruction complexes. In SW480 colorectal cancer cells treated with the tankyrase inhibitor (TNKSi) G007-LK we found that AXIN1 was not required for degradasome formation. In contrast, the formation of degradasomes as well as their capacity to degrade β-catenin were considerably impaired in G007-LK-treated cells depleted of AXIN2. These findings give novel insights into differential functional roles of AXIN1 versus AXIN2 in the β-catenin destruction complex.


Introduction
The Wnt signaling pathway orchestrates multiple developmental and adult homeostatic processes, whereas aberrant activation of the pathway underlies numerous human diseases such as cancer [1]. β-catenin, the key mediator of Wnt signaling output [2], is earmarked for proteasomal degradation by the so-called β-catenin destruction complex, which consists of the structural proteins adenomatous polyposis coli (APC) and axis inhibition protein 1 and 2 (AXIN1/ 2), and the kinases casein kinase 1α (CK1α) and glycogen synthase kinase 3 (GSK3) [3]. This signal-limiting complex is compromised in the majority of colorectal cancers due to mutations in the APC gene [4]. Recently, the poly-ADP-ribosyltransferases tankyrase 1 (TNKS1) and tankyrase 2 (TNKS2) were implicated as positive regulators of Wnt signaling by transferring a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 ADP-ribose moieties onto AXIN, the rate-limiting factor for destruction complex stability, thereby marking it for degradation by the ubiquitin-proteasome system [5,6]. Consequently, tankyrase inhibitors (TNKSi) have emerged as promising new cancer therapeutics that stabilize AXIN and reduce Wnt signaling output [7]. Intriguingly, several studies have reported the formation of distinct cytoplasmic puncta upon treatment with TNKSi [8][9][10]. These protein assemblies, referred to as degradasomes, contain destruction complex components and most likely represent the morphological and functional correlates of endogenous destruction complexes.
In the current study, we have compared the role of AXIN1 with AXIN2/Conductin in the formation of degradasomes and degradation of β-catenin induced by the TNKSi G007-LK in colorectal cancer cells (SW480), as their relative contribution in the Wnt signaling pathway still remains incompletely understood. Surprisingly, we find that AXIN1 is neither required for G007-LK-induced degradasome formation nor for G007-LK-induced degradation of βcatenin, despite its significant upregulation after prolonged TNKS inhibition (24 h). In contrast, the formation and function of degradasomes were considerably impaired in G007-LKtreated SW480 cells depleted of AXIN2. In addition, new synthesis of AXIN2 was required for degradasome formation upon G007-LK incubation. Taken together, our results imply that AXIN2 is more important than AXIN1 in both initiation of degradasomes and turn-over of βcatenin upon TNKS inhibition.

Cell-based assays
The SW480 cell line was purchased from American Type Culture Collection (ATCC). Upon receipt, cells were frozen and individual aliquots were taken into cell culture, typically for analysis within 15 passages. Cells were grown in RPMI medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin. Testing for mycoplasma contamination was performed every sixth week. For inhibition of TNKS activity, cells were treated with 0.5 μM G007-LK for indicated time points. Dimethyl sulphoxide (DMSO) was used as a control. For inhibition of proteasomal activity, cells were treated with 10 μM MG132 for 6 h, either alone or in combination with G007-LK. mRNA translation was inhibited by adding 25 μg/ml cycloheximide (CHX) for up to 6 h, either alone or in combination with G007-LK. The SW480 cell line stably expressing GFP-TNKS1 has been described previously [10].
Time-lapse live-cell imaging SW480 cells stably expressing GFP-TNKS1 were treated with G007-LK or a combination of G007-LK and MG132 and then imaged on a Deltavision live cell microscope (Applied Precision, GE Healthcare). To cover the whole cellular volume, stacks were acquired. One stack was acquired before addition of the inhibitors (timepoint "0"), afterwards images were acquired every 10 minutes for 6 h. After deconvolution, sum intensity projections were made. Image analysis was done in ImageJ/Fiji [14] to count the number of GFP-TNKS1 puncta.

ScanR high-throughput microscopy
Cells were grown on coverslips and fixed in paraformaldehyde. Images were automatically taken using the Olympus ScanR system with an UPLSAPO 40×/0.95 objective. All images were taken with the same settings and below pixel saturation. The Olympus ScanR analysis program was used to measure the average number of G007-LK-induced GFP-TNKS1 puncta. Typically between 2000 and 10000 cells were analyzed per condition in each experiment.

Statistics
Student's t-test was done on ScanR and Western blot quantifications to examine statistical significance.

Proteasome inhibition reduces stabilization of AXIN2 and formation of G007-LK-induced degradasomes
The colorectal cancer cell line SW480 is commonly used as a model for Wnt-dependent cancers due to a mutation in the APC tumor suppressor gene [15]. TNKS inhibition antagonizes Wnt signaling in SW480 cells by inducing degradasomes that phosphorylate and earmark βcatenin for proteasomal degradation [8][9][10]. Degradasomes are rapidly induced upon treatment with TNKSi and contain all components of the endogenous β-catenin destruction complex [10]. Surprisingly, we recently discovered that combining G007-LK with various proteasome inhibitors for 6 h reduced the formation of TNKSi-induced degradasomes [16]. Concurrently, a similar observation was reported upon combination of the TNKSi XAV939 with proteasome inhibitors [17]. To investigate the dynamics of this phenomenon in more detail, we performed live-cell imaging of SW480 cells stably expressing GFP-TNKS1, as TNKS1 localizes to degradasomes and can thus serve as a readout for degradasome formation (Fig 1A, [9,10]). Time-lapse imaging revealed that degradasomes were rapidly induced (<0.5 h) in cells treated with G007-LK ( Fig 1B' and 1C). Furthermore, we observed a significant reduction (~40%) in degradasome formation when combining G007-LK and MG132 for 6 h compared with G007-LK treatment alone ( Fig 1B" and 1C), which is in accordance with previous results from high-throughput microscopy analysis [16]. Interestingly, quantitative image analysis demonstrated that, during the first 30 min of incubation, combined treatment with G007-LK and MG132 increased the number of degradasomes almost to the same extent as treatment with G007-LK alone ( Fig 1C). However, in contrast to G007-LK treatment alone, the number of degradasomes was not further increased over the last 5.5 h when combining G007-LK and MG132. Thus, live-cell imaging confirmed the inhibitory role of proteasome inhibitors on G007-LK-induced degradasome formation and added further insight into the rapid-onset kinetics of this process.
AXIN protein levels are regarded as the rate limiting step for the stability and function of endogenous β-catenin destruction complexes [18]. The formation of degradasomes also appears to be AXIN-dependent, although recent studies indicate a role for additional complex components in this process (e.g. TNKS and APC2) [10,17,19]. However, the role of AXIN1 versus AXIN2 in TNKSi-induced degradasome formation remains elusive. Although they share the key sequence elements [20,21], the transcriptional regulation of AXIN1 and AXIN2 differs significantly: in contrast to AXIN1, which is constitutively expressed at very low levels, transcription of AXIN2 is regulated by Wnt signaling activity [22]. To address their relative contribution in the initial formation of degradasomes, we explored the protein levels of AXIN1 and AXIN2 after 6 h G007-LK treatment, either alone or in combination with MG132 ( Fig 1D). In line with previous results [10,16], Western blot analysis revealed a substantial G007-LK-mediated increase in AXIN2 protein levels. However, the protein levels of AXIN1 did not change significantly within the first 6 h of TNKS inhibition. Strikingly, the G007-LKinduced increase in AXIN2 levels was counteracted in the presence of MG132, while AXIN1 protein levels were not affected (Fig 1D, [16]).
Taken together, our findings indicate that formation of degradasomes during the initial 6 h of TNKSi treatment requires a stabilization of AXIN2 protein levels. We further propose that the immediate formation of degradasomes (<30 min) observed in live-cell imaging upon combined G007-LK and MG132 treatment is due to a temporary stabilization of AXIN2 protein levels when inhibiting the proteasome and/or a delay in the MG132-mediated repression of AXIN2 transcription [16]. However, the formation of degradasomes is abrogated as AXIN2 protein levels fail to be stabilized upon incubation of G007-LK in combination with proteasomal inhibition.
G007-LK-induced degradasome formation requires sustained protein translation β-catenin, the key mediator of Wnt signaling, initiates transcription of target genes including AXIN2 by complexing predominantly with the TCF/LEF family of transcription factors [23]. Importantly, transcription of AXIN2 initiates a negative regulatory feedback loop upon activation of the Wnt pathway [22]. Thus, AXIN2 protein levels are positively regulated via β-catenin-mediated transcription and negatively regulated via TNKS-mediated PARsylation cells are shown. Scale bar: 5 μm. (C) Quantification of the number of GFP-TNKS1 puncta per cell. Shown are values +/-SEM of two independent experiments with 24 (G007-LK) and 36 (G007-LK + MG132) cells in total. (D) SW480 cells expressing GFP-TNKS1 were incubated with G007-LK and MG132 for 6 h, either alone or in combination. Cells were then lysed and whole cell lysate was applied for Western blotting. Membranes were incubated with antibodies against AXIN1, AXIN2 and Actin (loading control). One representative blot is shown. doi:10.1371/journal.pone.0170508.g001 Differential Roles of AXIN1 and AXIN2 in beta-Catenin Destruction PLOS ONE | DOI:10.1371/journal.pone.0170508 January 20, 2017 followed by proteasomal degradation. As TNKS inhibition on one hand leads to the degradation of β-catenin and thereby reduced AXIN2 mRNA transcription, but on the other hand mediates decreased PARsylation and degradation of AXIN2 protein, we sought to further investigate this delicate regulation of AXIN proteins stability upon TNKS inhibition. To elucidate the contribution of AXIN2 new synthesis, we investigated to which extent degradasomes are formed in G007-LK-treated cells after inhibiting mRNA translation with cycloheximide (CHX). SW480 cells expressing GFP-TNKS1 were treated with CHX and G007-LK for 6 h, either alone or in combination, and then investigated by high-throughput image acquisition using the Olympus ScanR microscope (Fig 2A). The number of GFP-TNKS1 puncta was quantified using the ScanR analysis software (Fig 2B). Strikingly, degradasome formation was completely abolished when combining G007-LK with CHX, indicating that G007-LK-induced degradasome formation may require continuous synthesis of AXIN2. Indeed, Western blot analysis revealed that AXIN2 protein levels were not stabilized or increased at any timepoint during 6 h combination treatment (Fig 2C and 2D). In conclusion, our results indicate that the formation of degradasomes requires sustained new synthesis of AXIN2. However, at this point we could not exclude that other, still unknown, initiating factors for degradasome formation may be affected upon inhibition of mRNA translation. Therefore we proceeded with knockdown experiments of AXIN1 and AXIN2.

AXIN1 is not required for sustained degradasome formation after prolonged TNKSi treatment
The levels of AXIN1 remained essentially unaltered upon 6 h treatment with G007-LK ( Fig  1D). However, after prolonged G007-LK treatment (up to 24 h) we observed that both AXIN1 and AXIN2 protein levels were significantly increased (Fig 3A), which is consistent with the effects of the TNKSi XAV939 on AXIN levels after 24 h [9]. To investigate whether this latephase upregulation of AXIN1 is crucial for degradasome stability and function, SW480 cells stably expressing GFP-TNKS1 were treated with G007-LK for 24 h after short interfering RNA (siRNA)-mediated depletion of either AXIN1 or AXIN2, or both proteins (Fig 3B). Importantly, depletion of AXIN1 did not lead to compensatory up-regulation of AXIN2 or vice versa. For each condition, the capacity of cells to induce degradasomes was investigated with the Olympus ScanR microscope (Fig 3C). The number of GFP-TNKS1 puncta was quantified using the ScanR analysis software and revealed a~50% reduction in degradasome formation in cells depleted of AXIN2 compared to control cells (Fig 3D). Surprisingly, depleting cells of AXIN1 did not influence the number or size of G007-LK-induced degradasomes and did not significantly further reduce the number of degradasomes in cells depleted for both AXIN1 and AXIN2 (Fig 3D and 3E). We therefore suggest that the 2-3-fold increase in AXIN1 protein level observed after 24 h G007-LK treatment is not required for degradasome formation whereas the 15-20-fold increase in AXIN2 protein level is both necessary and sufficient. Interestingly, AXIN1 is not up-regulated and does not seem to rescue the depletion of AXIN2, despite its higher abundance at steady-state [24] and its moderate increase after 24 h of G007-LK incubation (Fig 3A).
Of note, it was recently demonstrated that a three-to four-fold increase in Drosophila Axin levels is insufficient to inhibit signaling for nearly all Wingless-driven developmental processes, thereby establishing a physiological range within which AXIN levels may fluctuate and still remain compatible with the activation of the pathway following Wnt stimulation in all cells [25]. We suggest that this threshold is not exceeded without the significant increase in AXIN2 levels upon G007-LK treatment. However, additional functional differences between AXIN1 and AXIN2 in degradasomes cannot be excluded.

AXIN2 depletion counteracts TNKSi-induced reduction in β-catenin protein levels
We recently demonstrated that TNKSi-induced degradasomes in SW480 cells contain phosphorylated β-catenin, ubiquitin and the E3 ubiquitin ligase component β-TrCP, indicating   functional degradasomes. Moreover, fluorescence recovery after photobleaching (FRAP) experiments demonstrated that β-catenin-mCherry was rapidly turned over in the G007-LKinduced degradasomes. Taken together, these findings provided a direct mechanistic link between degradasome formation and reduced Wnt signaling in colorectal cancer cells [10]. To further elucidate the individual contribution of the AXIN proteins to the functionality of degradasomes, we investigated whether siRNA-mediated depletion of either AXIN1 or AXIN2 counteracts G007-LK-induced degradation of β-catenin. SW480 cells stably expressing GFP-TNKS1 were treated with G007-LK for 24 h after depletion of either AXIN1 or AXIN2, or both proteins (Fig 3B), and the remaining total and active β-catenin protein levels were measured by Western blot analysis (Fig 4A and 4B). Strikingly, quantification of the protein levels revealed that the G007-LK-mediated reduction in β-catenin protein levels was nearly completely reversed in AXIN2-depleted cells. In contrast, depleting AXIN1 did neither compromise the reduction in β-catenin protein levels induced by tankyrase inhibition nor did it further aggravate the effect of depleting AXIN2 (Fig 4A and 4B). Similar to our results, combined siRNA-mediated depletion of AXIN1 and AXIN2 in SW480 cells reversed the effect of XAV939 on β-catenin degradation and diminished the inhibitory activity of XAV939 in a Super TopFlash reporter assay [6]. However, AXIN1 and AXIN2 were not individually depleted in the study by Huang and colleagues, therefore concealing their individual contributions. Our functional analysis of β-catenin degradation in cells depleted for AXIN1 or AXIN2 indicates a predominant role of AXIN2 in the TNKSi-induced degradation of β-catenin in SW480 cells. As AXIN2 was initially shown to compensate for AXIN1 in a mouse knock-in approach, both proteins were thought to be functionally redundant in vivo [26]. However, recent in vitro findings from Bernkopf and colleagues report a surprisingly predominant role of AXIN1 in cells stimulated with Wnt ligand [24], which is due to a reduced ability of AXIN2 to polymerize with Dvl2 when compared to AXIN1. This explains how AXIN2 can serve as a negative-feedback regulator of Wnt signaling despite its low abundance. Our study investigates the role of AXIN1 versus AXIN2 in degradasomes by using TNKS inhibition as a means to stimulate β-catenin degradation in APC-mutant cells. Our results indicate a predominant role of AXIN2 compared to AXIN1. We suggest that this can be explained by the severely increased amount of AXIN2 after G007-LK treatment, which induces the polymerization of degradasome components. Polymerization of AXIN1/2, APC and TNKS1/2 are thought to increase avidity for downstream signaling effectors and thus being a prerequisite for efficient β-catenin degradation [27][28][29][30][31]. Thus it seems that AXIN2 may be more important for switching off Wnt signaling, while AXIN1 may predominate in mediating the transition from Wnt off to Wnt on state by being recruited to signalosomes.
In conclusion, we have studied the individual contributions of AXIN1 and AXIN2 in the formation of TNKSi-induced degradasomes and degradation of β-catenin. Taken together our results reveal a surprisingly differential role of AXIN1 and AXIN2. AXIN2 stands out to be the most important player in TNKSi-induced degradasome formation and β-catenin degradation in SW480 colorectal cancer cells. However, the functional role and relative contributions of AXIN1 versus AXIN2 during steady-state turnover of β-catenin in APC wild-type cells remain to be investigated.